NNIN REU Program Summer 2011

• table

of co

nte

nts

• table

of co

nte

nts

• table

of co

nte

nts

• table

of co

nte

nts

•2011 NNIN REU Research Accomplishments Page i

The National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program

2011 NNIN REU Research Accomplishments

Table of Contents

Biological Applications, pages 2-39

Development of a Fluorescence-Based Quantification Method to Determine the Amount of Glycans Immobilized on a Surface ... ... ... ... ... ... ...2

Andrew Acevedo, Biomedical Engineering, Washington University in St. Louis

Microfluidic Protein Dialysis Device for X-Ray Scattering .. ...4Amani Alkayyali, Biomedical Engr., Wayne State University

Gradient Surface Wettability Induced by Nanofilms on Titanium Surfaces and Osteoblastic Cell Morphology.. ...6

Noelia Almodovar, Chemical Engineering, University of Puerto Rico, Mayaguez Campus

Fabrication of a Novel Microfilter for Circulating Tumor Cell Enrichment and Culture.. ... ... ... ... ... ... ...8

Julie Chang, Chemistry and Physics, Harvard University

Characterization of Embryonic Rat Cortical Cells Grown on Microcontact Printed Protein Patterns ... ... . 10

Steven Chase, Biomedical Engr. and Biochem. and Molecular Biology, Rose-Hulman Institute of Technology (Graduated)

2011 NNIN REU Reports by Research Focus; i-x

Welcome, Prof. Roger Howe, NNIN Director; xi

The NNIN Sites, REU Sponsors, & Photograph Credits; xii

2011 NNIN Interns, by Site:

University of Colorado, page xviiUniversity of Michigan, page xviiUniversity of Minnesota, page xviiiUniversity of Texas, page xviiiUniversity of Washington, page xixWashington University, page xixNNIN iREG Program, page xxNNIN iREU Program, page xx

Arizona State, page xiiiCornell, page xiiiGeorgia Tech, page xivHarvard, page xivHoward, page xvPenn State, page xvStanford, page xviUCSB, page xvi

Commonly Used Abbreviations & Their Meaning; xxi-xxviii

• t

able

of

conte

nts

•

table

of

conte

nts

•

table

of

conte

nts

•

table

of

conte

nts

•

Page ii 2011 NNIN REU Research Accomplishments

Nanomechanical Properties of Structured Biopolymer Networks ... ... ... ... ... ... . 12

Vinh Diep, Nanoengineering, UC San Diego

A Microchannel-Scaffold Electrode Array for Peripheral Nerve Interfacing . ... ... ... ... ... ... ... . 14

Jaideep S. Dudani, Bioengineering, UCLA

Multiplexed Silicon Nanophotonic Biosensing via Immobilized Protein Glycoconjugates . ... ... ... ... . 16

Kevin Huang, Engineering, Trinity College

Patterning Antigens with Near-Field Optics ... ... ... ... ... . 18Matthew Kiok, Chemistry, Tulane University

Dry Electrodes for Electroencephalography Headsets ... ... . 20Sibu Kuruvilla, Materials Science and Engineering, UIUC

Synthesis of Poly(Amino Ether) Capped Gold Nanoparticles for Transgene Delivery ... ... ... ... . 22

Olivia Lambdin, Biological Systems Engineering, UNL

Neutrophil Chemotactic Response to Chemokine Gradients in a Microfluidic Device ... ... . 24

Leah Laux, Biomedical Engineering, WUSTL

Gold Nanostructures with Tunable Photothermal Properties for Cancer Treatment . ... ... . 26

Max Li, Biomedical Engineering, and Mathematics, Duke

Chromatin Remodeling by Brahma Motor Unit on Mono-Nucleosome DNA .. ... ... . 28

Evan Mirts, Biology/Physics, Truman State University

Metallic Nanostructure for Surface Plasmon Resonance Biosensing ... ... ... ... ... . 30

Lauren Otto, Physics and Mathematics, Bethel University

Three-Dimensional Label-Free Photoacoustic Microscopy of the Tumor Microenvironment in vivo ... ... ... ... ... . 32

Ernest Puckett, Biomedical Engineering, UTX

Biosensing Based on Surface-Enhanced Raman Scattering. . 34Laurel Rognstad, Chemical Engineering, TTU

Measurement of Platelet Clot Volume in a Microscale Thrombosis Screening Device. ... ... ... ... ... ... ... ... . 36

Laura Seaman, Biological Engineering, MIT

Measuring the Thermodynamic Properties of Water at Negative Pressures in Synthetic Trees . ... ... ... ... . 38

Zachary Sonner, Microelectronic Engineering, RIT

• table

of co

nte

nts

• table

of co

nte

nts

• table

of co

nte

nts

• table

of co

nte

nts

•2011 NNIN REU Research Accomplishments Page iii

Chemistry, pages 40-55

Defect Analysis of Molecular Monolayers with Electrochemistry ... ... ... ... ... ... ... ... ... ... ... . 40

Clara Chow, Biomedical Engineering and Chemistry, University of Wisconsin - Madison

Novel pH-Sensitive Dendrimer Nanoparticles for Targeted Imaging ... ... ... ... ... ... ... ... ... ... ... . 42

Audrey Dang, Chemical Engineering, Vanderbilt University

Time Resolve Study of Anisotropic Nanostructure Growth Using Integrated Droplet-Based Microfluidics and X-Ray Absorption Spectroscopy ... ... ... ... ... ... . 44

Giovanni Esteves, Chemical Engineering, ASU

Synthesis of Self-Assembling Silver Nanoparticles for Surface Enhanced Raman Spectroscopy ... ... ... ... . 46

Jennifer Gilbertson, Chemistry, Beloit College

Measuring Height Mismatch and Miscibility Temperatures of Model Cell Membranes .. ... ... ... ... . 48

Morgan McGuinness, Physics, Lafayette College

Film Making in Digital 3D: Selective Area Atomic Layer Deposition ... ... ... ... ... . 50

Jade M. Noble, Chemical Physics, Columbia University

Synthesis of Silicon and Germanium Nanowires .. ... ... ... . 52Yoichi Ogata, School of Materials Science,

Japan Advanced Institute of Science and Technology

Stability of Zwitterionic-Modified Gold Nanoparticles in Complex Media: Effect of Surface Packing Densities . 54

Julia Podmayer, Biochemistry, Seattle Pacific University

• t

able

of

conte

nts

•

table

of

conte

nts

•

table

of

conte

nts

•

table

of

conte

nts

•

Page iv 2011 NNIN REU Research Accomplishments

Electronics, pages 56-89

Inkjet Printing of Zinc Oxide Based-Semiconductors for Thin Film Transistors ... ... ... ... ... ... ... ... ... ... . 56

Carlos Koladele Biaou, EE, Prince George’s Community College

Conformal Copper Seed Layers for Through-Silicon Vias Using Chemical Vapor Deposition . ... ... ... ... ... . 58

Parker Clark, EE and MS, Stanford University

Optimization of Switching Layer for Retention in Tungsten Oxide Memristive Devices . ... ... ... ... ... . 60

Emily Griffin, Physics, Case Western Reserve University

Using Electron Beam Lithography for Nanowire Transistor Fabrication ... ... ... ... ... ... . 62

Andreas M. Haggerty, Mechanical Engineering, Harvard

Shape-Specific Iron Platinum Nanocrystals for Spin-Transfer Torque in Magnetic Tunnel Junctions ... . 64

Kai He, Electrical and Computer Engineering, Rice UniversityFabrication of GaAsBi Heterojunction Bipolar Transistors 66Hilary Hurst, Engineering Physics, Colorado School of Mines

The Micro-Fabrication of a Composite Thermal Capacitor.. . 68Israel Ilufoye, Mechanical Engineering, UMBC

Deterministic Assembly of Alternative Materials onto Silicon Substrates ... ... ... . 70

Jia Kuang, Electrical Engr., The City College of New York

Development of Carbon Nanotube Field-Effect Transistors for Use in Next Generation Electronics. ... . 72

Hongliang Liang, Engineering, Swarthmore College

Characterization of Floating-Gate Graphene ... ... ... ... ... . 74Maria Veronica Mateus, Mechanical Engineering,

The University of Texas Permian Basin

Fabrication of Organic Transistors Using Inkjet Printing ... . 76Michelle Pillers, Chemistry, Southern Methodist University

Characterization of Ion Sensitive Field Effect Transistors for Cellular Scale pH Measurement.. ... ... . 78

Kendall Pletcher, Bioengineering, Olin College of Engineering

Growth and Characterization of Graphene for Use in Nanoelectronics . ... ... ... ... ... ... ... ... ... . 80

Bethany Robinson, Electrical Engineering, Howard University

Fabrication and Characterization of ZnO Nanowire Field-Effect Transistors and ZnTe Nanosheet Field-Effect Transistors.. ... ... ... ... . 82

William Scheideler, EE and Biomedical Engr., Duke University

Synthesis of Few-Layer Graphene Films of Large Lateral Dimensions ... ... ... . 84

Claire Spradling, Biological Engr., UMC

Fabrication of Flexible Organic Thin-Film Transistors... ... . 86Kevin Tien, Electrical Engineering, Cooper Union

Electronic Graphene Devices through Tip-Based Nanotechnology ... ... ... ... ... ... . 88

Cassandra Todd, Electrical Engineering, UCF

• table

of co

nte

nts

• table

of co

nte

nts

• table

of co

nte

nts

• table

of co

nte

nts

•2011 NNIN REU Research Accomplishments Page v

Materials, pages 90-123

Characterization of 2-300 nm Alumina Thin Films Deposited via ALD . ... ... ... ... ... ... ... ... . 90

Malena Agyemang, Chemistry, Norfolk State University

Acceptor State Activation Energy of Tin Monosulfide ... ... . 92Kyle Arean-Raines, Chemical Engr., University of RochesterAdhesion and Cohesion Testing in Square Solar Cells ....94Karl Bayer, Physics, Pacific Lutheran University

Heteroepitaxial Growth of Diamond for Device Applications. ... ... ... ... ... ... ... . 96

Antanica Boneparte, Chemical Engr., Howard University

The Dynamics and Control of Bubbles Residing under Graphene Films .. ... ... ... ... ... ... ... . 98

Lauren Cantley, Physics, Grinnell College

Utilizing Solution-Grown Silicon Nanowires in Photovoltaic Devices . ... ... ... ... ... ... ... ... ... ... 100

Elizabeth Fullerton, Chemical Engr., University of Arkansas

Patterning of the Metal Induced Crystallization of Amorphous Silicon for Silicon Wire Array Photovoltaics .. ... ... ... 102

Julie A. Georgiev, Electrical Engineering, Alfred University

PZT Films with Reduced & Exaggerated Zr:Ti Gradients ... 104Michael Gerhardt, Materials Science and Engineering, MIT

Self-Assembled Gold Nanoparticles for Biosensing Applications ... ... ... ... ... ... ... ... ... 106

Abigail Halim, Materials Science and Engr., Georgia Tech

Fabrication of Free-Standing Graphene Films for Probing the Ultrafast Electron Dynamics . ... ... ... 108

Jeffrey Hart, Electrical and Computer Engineering, Franklin W. Olin College of Engineering

Growth and Characterization of Aluminum Nitride Nanowires.. ... ... ... ... ... ... ... 110

Alicia Herro, Physics, University of North Texas

Deposition and Characterization of Ruthenium Films for Neural Electrodes ... ... ... ... 112

Diana Wu, Chemical Engineering and Biology, MITEmily Hoffman, MSE, Northwestern University

Flexible and Stretchable Networks of Metals .. ... ... ... ... 116Gawain Lau, Chemical and Biomolecular

Engineering, University of Pennsylvania

Overcoming Cellular Breakdown in Hyperdoped Silicon Alloys.. ... ... ... ... ... ... ... ... 118

Brandon Piercy, Materials Science and Engineering, CWRU

Surface Characterization of Etched Micro- and Nano-Structures in Silicon for Phonon Heat Transport ... ... ... ... ... ... 120

Victoria Savikhin, Electrical and Computer Engr., Purdue

Confinement Assisted Self-Organization of Photonic Templates . ... ... ... ... 122

Daryl I. Vulis, Electrical Engineering, SUNY Stony Brook

• t

able

of

conte

nts

•

table

of

conte

nts

•

table

of

conte

nts

•

table

of

conte

nts

•

Page vi 2011 NNIN REU Research Accomplishments

Mechanical Devices, pages 124-137

Fabrication of Microfluidic Devices for Synthesis of Janus Particles.. ... ... ... ... ... ... ... 124

Brittany Alphonse, Biomedical Engineering, URI

Development of Paper Accelerometers for Cheap Applications .. ... ... ... ... ... ... ... ... ... ... 126

Brendon Lee Gobert, Math and Science, Blackfeet Community College

Dimensional Analysis of Microlitre-Sized Microbial Fuel Cells ... ... ... ... ... 128

Nicole Hams, Biochem., University of Missouri - Columbia

Graphene Resonators for Mass and Charge Sensing.. ... ... 130Reyu Sakakibara, Chemical Biology, UC Berkeley

Using MEMS Sensor Array to Map the Temperature of Hot Springs in Yellowstone National Park. ... ... ... 132

Karl Schliep, Chemistry and Mathematics, University of Minnesota Morris

Measuring van der Waals Forces in Graphene .. ... ... ... ... 134Mariah Szpunar, Mechanical Engineering, University of Miami

Calibration of Optical Particle Sizer by Wafer Surface Scanner... ... ... ... ... ... ... ... ... ... 136

Laura Windmuller, Biomedical Engr., Boston University

• table

of co

nte

nts

• table

of co

nte

nts

• table

of co

nte

nts

• table

of co

nte

nts

•2011 NNIN REU Research Accomplishments Page vii

Optics & Opto-Electronics, pages 138-163

Sol-Gel Route for Ultra-High-Quality Optical Resonators ... 138Michael Akenhead, Biomedical Engr., Vanderbilt University

Design and Characterization of Multiple Quantum-Well Lasers ... ... ... ... ... ... ... 140

Seyedshahin Ashrafzadeh, Electrical Engineering, WVC

Distributed Bragg Reflectors in Ultra Low Loss Silicon Nitride Waveguides ... ... ... 142

Issa Beekun, EECS, University of Nevada, Reno

Integrated Silicon Nitride Waveguides: Optimization of Fabrication ... ... ... ... ... ... ... ... ... 144

Alex Bryant, Materials Science and Engr., UC Berkeley

Fabrication and Testing of Voltage-Tunable Plasmonic Metamaterials in Mid-Infrared . ... ... ... ... 146

Ting Chia Chang, EECS, University of California at Berkeley

Characterization of Optical Devices using a Pigtailed Fiber.. ... ... ... ... ... ... ... ... ... ... 148

Mark Dong, Applied Physics, Cornell University

Aluminum Nanowire Fabrication for use in Polarization Filters. ... ... ... ... ... ... ... ... 150

Nicholas Heugel, Biomedical Engr., Saint Louis University

Electrical Properties of the Germanium-on-Silicon Interface ... ... ... ... ... 152

Travis Lloyd, Engineering Physics, Brown University

Flexible Membrane Liquid Lens. ... ... ... ... ... ... ... ... ... 154David Mallin, Physics, University of San Diego

Soft Lithographic Fabrication of Bar Chart Phantoms for Axial Resolution Measurements in Optical Coherence Tomography ... ... ... ... ... ... ... 156

Meagan Pipes, Biomedical Engineering, NCSU

Analog Lithography of Complex Phase Plates for Sub-Diffraction Lithography.. ... ... ... ... ... ... ... 158

Drew D. Schiltz, Physics, Winona State University

Graphene-Based Ultrafast Electro-Optical Modulators .. ... 160Seiya Suzuki, Graduate School of Engineering,

Toyota Technological Institute, Tempaku-ku, Nagoya, Japan

Zero-Mode Waveguides for Single-Molecular Imaging .. ... 162Jin Zhang, Chemical Engineering, University of Arizona

• t

able

of

conte

nts

•

table

of

conte

nts

•

table

of

conte

nts

•

table

of

conte

nts

•

Page viii 2011 NNIN REU Research Accomplishments

Physics & Nanostructure Physics, pages 164-185

Electrical Single Molecule Investigations by Means of Mechanical Break Junctions .. ... ... ... ... 164

Brian Benton, Physics, University of Minnesota – Twin Cities

Investigation of Electron Transport in Functionalized Carbon Nanotubes ... ... ... ... ... ... 166

Mark Brunson, Mechanical Engineering, SFSU

Capacitance Measurements of Single Indium Arsenide Nanowires.. ... ... ... ... ... 168

Kevin Chen, Electrical Engr., Arizona State University

Resistive Switching of Iron-Doped SrTiO3 ... ... ... ... ... ... 170Zachary Connell, Mechanical Engineering, UNL

Growth of Silicon, Silicon Carbide, and Boron Nitride Nanowires for Electronic Applications . ... ... ... ... ... 172

Won Jun Kuk, Chemistry, Williams College

Nanoscale Diamond Lenses for Atomic-Scale Sensing ... ... 174Dominic E. Labanowski, Electrical Engineering, OSU

Local and Global Effects on the Growth of Carbon Nanotube Micropillar Arrays ... ... ... ... ... 176

Yuki Matsuoka, Future Industry-Oriented Basic Science and Materials, Toyota Technological Institute, Japan

Devices for Investigating Electrical Transport in Topological Insulators ... ... ... ... ... ... ... ... ... ... 178

Joshua Mendez, Physics, Louisiana State University

DNA in Nanochannels .. ... ... ... ... ... ... ... ... ... ... ... ... 180Francisco Pelaez III, Chemical Engineering, UTX

Characterization of Optoelectronic Properties of Colloidal Quantum Dots in a Nanogap.. ... ... ... ... 182

Margeaux Wallace, MSE, Cornell University

Microfabricated Silicon Carbide Thermionic Energy Converters for Solar Energy Generation. ... ... 184

Leah Weiss, Physics, Harvard University

• table

of co

nte

nts

• table

of co

nte

nts

• table

of co

nte

nts

• table

of co

nte

nts

•2011 NNIN REU Research Accomplishments Page ix

Process & Characterization, pages 186-205

Characterization of YBCO Superconducting Thin-Films for Fluxonic Applications ... ... ... ... ... ... 186

Brian T. Chung, Engineering Physics, UMI Ann Arbor

Engineering Multifunctional Nanoparticles with Dual Modality Imaging Capabilities .. ... ... ... ... 188

Courtney Crouch, Biology, Emory University

Electrodeposition of Metals onto Aligned Carbon Nanotube Microstructures. ... ... 190

Matthew Diasio, Physics, Rice University

Morphological Characterizations of Collagen-Modified Alumina Membranes ... ... ... ... 192

Tiffany Dunston, Chemistry, Syracuse University

Characterization of AIN Thin Films for Applications in Bulk Acoustic Filters ... ... ... ... ... 194

Lisa Anne Hendricks, Electrical Engineering, Rice University

Novel Process to Fabricate Raised Polymer Electrodes for Electroencephalography ... ... ... ... ... 196

Fiona O’Connell, Materials Engr., Loyola University Maryland

Design, Fabrication, and Testing of Hg/Au Microelectrodes for Voltammetric Sensing of Trace Metals for Environmental Monitoring .. ... ... 198

Andrew Raebig, EE and Chemistry, Butler University

Optical and Electron Beam Patterning for Graphene Nanoribbon Devices ... ... ... ... ... ... ... 200

Nathanial Sheehan, Electrical Engineering, UCSB

Systematic Investigation of Morphology of Polymer:Bis-Fullerene Blends for Bulk Heterojunction Organic Photovoltaics .. ... ... 202

Joseph Smalley, Engineering Science, PSU (graduated), Electrical and Computer Engineering, UC San Diego

Tribology of Atomic Layer Deposition Films ... ... ... ... ... 204Kelly Suralik, Chemistry, Middlebury College

• t

able

of

conte

nts

•

table

of

conte

nts

•

table

of

conte

nts

•

table

of

conte

nts

•

Page x 2011 NNIN REU Research Accomplishments

Societal & Ethical Issues of Nanotechnology, pages 206-209

Researcher Views on the Perceived Influence of Funding Sources in Nanotechnology Research ... ... 206

Rachel Brockhage, Biology and Communication, GCC

The Ethical, Legal and Societal Implications of Nanotechnology . ... ... ... ... ... ... ... 208

Nina Hwang, Chemistry, Rice University

Index, pages 210-212

2011 NNIN REU Research Accomplishments Page xi

• intro

ductio

n • in

troductio

n • in

troductio

n • in

troductio

n • in

troductio

n •

Welcome to the 2011 edition of the NNIN REU Research Accomplishments! This publication reflects the hard work of the undergraduate researchers, as well as the dedication and commitment of their mentors, the site staff, REU coordinators, and the principal investigators.

Our summer program brings undergraduates from colleges and universities across the U.S. into some of the nation’s leading academic nanofabrication laboratories for an intensive ten-week research experience. The participants are trained in safe laboratory practices, learn the essential scientific background for their project, and then perform independent research in nanotechnology, under the guidance of their mentor. For many of the students, the NNIN REU is their first experience when the answer, or even the existence of an answer, is not yet known. The summer experience provides the students with a sample of what life for them could be like as a graduate student in applied science and engineering, both in and outside the clean room. For many, the NNIN REU helps them to decide to pursue a career in research and development.

During the summer of 2011; 85 students participated in our Research Experience for Undergraduates (REU) Program. From the previous summer’s NNIN REU Program, 16 students were selected to participate in NNIN’s International Research Experience for Undergraduates (iREU) Program in Belguim, France, Germany and the Netherlands. Four graduate students from Japan participated in our exchange (iREG) program.

Welcome from Prof. Roger T. Howe, NNIN Director

The 2011 NNIN REU Interns at the network-wide convocation at Georgia Tech in August. Photograph by Christopher Julian.

NNIN is committed to making all three of these programs a substantial experience for the participants, by focusing on advanced research and knowledge, seeking strong mentors and staff support, exposing the students to a professional research environment, and having high expectations for the research projects, as well as for the presentations at the final convocation. The students’ exposure to a wider variety of research conducted by their peers and the other nanofab users across diverse disciplines of science and engineering provides a significant complementary experience.

I would like to thank the NNIN REU staff, graduate student mentors, and faculty for their contributions to the success of this year’s programs. Particular thanks are due to Ms. Melanie-Claire Mallison and Dr. Lynn Rathbun at Cornell, and Dr. Nancy Healy at Georgia Institute of Technology, for their contributions to organizing the logistics of these programs. In addition, I am grateful to Dr. Healy and Ms. Katie Hutchison for organizing the network-wide convo-cation at Georgia Tech.

I wish all our program participants the best for their future careers, whether in science, engineering, or other disciplines. I hope you will build on this summer’s experience and I look forward to hearing from you on your future successes!

Roger T. HoweDirector, [email protected]

Page xii 2011 NNIN REU Research Accomplishments• i

ntr

oduct

ion •

intr

oduct

ion •

intr

oduct

ion •

intr

oduct

ion •

intr

oduct

ion •

ASU NanoFab, Arizona State UniversityIra A. Fulton Schools of Engineering • PO Box 876206

Tempe, AZ 85287-6206(480) 965-3808 • http://www.fulton.asu.edu/nanofab/

Cornell NanoScale Science & Technology Facility, Cornell University250 Duffield Hall • Ithaca, NY 14853-2700

(607) 255-2329 • http://www.cnf.cornell.edu

Nanotechnology Research Center, Georgia Institute of Technology791 Atlantic Dr NW • Atlanta, GA 30332-0269(404) 385-4307 • http://www.mirc.gatech.edu/

Center for Nanoscale Systems, Harvard University11 Oxford Street, LISE 306 • Cambridge, MA 02138-2901

(617) 384-7411 • http://www.cns.fas.harvard.edu

Howard Nanoscale Science & Engineering Facility, Howard UniversityDowning Hall Room 1124 • 2300 Sixth St NW • Washington, DC 20059-1015

(202) 806-6618 • http://www.msrce.howard.edu/

Penn State Nanofabrication Laboratory, The Pennsylvania State University188 Materials Research Institute • 230 Innovation Blvd • University Park, PA 16802-6300

(814) 865-7443 • http://www.mri.psu.edu/facilities/nnin/

Stanford Nanofabrication Facility, Stanford UniversityP.G. Allen Bldg, Rm 303x • 420 Via Palou • Stanford, CA 94305-4070

(650) 725-3607 • http://snf.stanford.edu/

Colorado Nanofabrication Laboratory, University of Colorado at BoulderECEE Campus Box 425 • Boulder, CO 80309-0425

(303) 492-5324 • http://cnl.colorado.edu/

Nanotech, University of California, Santa BarbaraECE Dept • Engineering Science Bldg 1109F • Santa Barbara, CA 93106-9560

(805) 893-5999 • http://www.nanotech.ucsb.edu/

Lurie Nanofabrication Facility, The University of Michigan, Ann Arbor1301 Beal Ave 1241 EECS • Ann Arbor, MI 48109-2122

(734) 763-0231 • http://www.lnf.umich.edu

Nanofabrication Center, University of Minnesota-Twin Cities200 Union St SE • 1-165 Keller Hall • Minneapolis, MN 55455-0171

(612) 625-3069 • http://www.nfc.umn.edu/

Microelectronics Research Center, The University of Texas at AustinJ.J. Pickle Research Campus • 10100 Burnet Rd, Bldg 160 • Austin, TX 78758-4445

(512) 471-4493 • http://www.mrc.utexas.edu/

Center for Nanotechnology, University of WashingtonBox 352140 • Fluke Hall Rm 215 • Seattle, WA 98195-1700

(206) 616-9320 • http://www.nano.washington.edu/

Nano Research Facility, Washington University in St. LouisEarth & Planetary Sciences Bldg Rm 189

Campus Box 1169, One Brookings Drive • St. Louis, MO 63130-4899(314) 935-7264 • http://www.nano.wustl.edu/

The National Nanotechnology Infrastructure Networkhttp://www.nnin.org/

is comprised of the following fourteen sites, and is supported byThe National Science Foundation, the NNIN sites, our corporate sponsors and research users.

This book was designed and formatted by the NNIN REU Program Assistant, Melanie-Claire Mallison.

The book was printed using soy-based ink and 10% post consumer paper. We encourage you to reduce, reuse, and

recycle. This publication is also available online at http://www.nnin.org/

The photographs on pages i-x were taken by NNIN site staff and iREU interns. Photographs on pages xi-xx were taken by Christopher Julian (Georgia Institute of Technology).

The photographs on pages xxi-xxviii were taken by Christopher Julian and iREU interns.

2011 NNIN REU Corporate Sponsors

Agilent TechnologiesAnalog Devices

Applied MaterialsCambridge Nanotech

CanonDaihen Corporation

EricssonIBM Corporation

InfineonIntel Corporation

InvensenseNational Semiconductor Corporation

NEC CorporationNXP SemiconductorsOlympus Corporation

PanasonicQualcomm

Renesas Electronics CorporationRobert Bosch Corporation

STMicroelectronicsTexas Instruments, Incorporated

Toshiba

2011 NNIN REU Research Accomplishments Page xiii

• nnin

reu in

tern

s b

y s

ite • n

nin

reu in

tern

s b

y s

ite • n

nin

reu in

tern

s b

y s

ite •

The 2011 NNIN REU Interns at Arizona State University

Ms. Olivia Lambdin . ... ... ... ... ... ... ... ... ... page 22Mr. William Scheideler . ... ... ... ... ... ... ... ... page 82Ms. Kendall Pletcher ... ... ... ... ... ... ... ... ... page 78Mr. Karl Schliep .. ... ... ... ... ... ... ... ... ... .. page 132Ms. Nicole Hams . ... ... ... ... ... ... ... ... ... .. page 128

The 2011 NNIN REU Internsat Cornell University

Front Row:Ms. Maria Veronica Mateus ... ... ... ... ... ... ... page 74Ms. Reyu Sakakibara ... ... ... ... ... ... ... ... .. page 130Ms. Daryl Vulis ... ... ... ... ... ... ... ... ... ... .. page 122

Middle:Ms. Amani Alkayyali ... ... ... ... ... ... ... ... ... .. page 4Ms. Jade Noble ... ... ... ... ... ... ... ... ... ... ... page 50Ms. Cassandra Todd ... ... ... ... ... ... ... ... ... page 88Mr. Matthew Kiok ... ... ... ... ... ... ... ... ... ... page 18Ms. Rachel Brockhage.. ... ... ... ... ... ... ... .. page 206

Back:Ms. Victoria Savikhin (in hat) ... ... ... ... ... .. page 120Mr. Zachary Sonner . ... ... ... ... ... ... ... ... ... page 38Mr. Alex Bryant .. ... ... ... ... ... ... ... ... ... .. page 144Mr. Joshua Mendez.. ... ... ... ... ... ... ... ... .. page 178

Page xiv 2011 NNIN REU Research Accomplishments• n

nin

reu i

nte

rns b

y s

ite •

nnin

reu i

nte

rns b

y s

ite

• n

nin

reu i

nte

rns b

y s

ite •

The 2011 NNIN REU Interns atGeorgia Institute of Technology

Front Row:Ms. Malena Agyemang. ... ... ... ... ... ... ... ... page 90Ms. Laura Seaman ... ... ... ... ... ... ... ... ... ... page 36Ms. Noelia Almodovar . ... ... ... ... ... ... ... ... .. page 6Ms. Brittany Alphonse . ... ... ... ... ... ... ... .. page 124

Back Row:Mr. Andrew Acevedo ... ... ... ... ... ... ... ... ... .. page 2Mr. Jaideep Dudani.. ... ... ... ... ... ... ... ... ... page 14Mr. Israel Ilufoye ... ... ... ... ... ... ... ... ... ... page 68

The 2011 NNIN REU Internsat Harvard University

Ms. Antanica Boneparte ... ... ... ... ... ... ... ... page 96Mr. Brandon Piercy.. ... ... ... ... ... ... ... ... .. page 118Mr. Mark Dong ... ... ... ... ... ... ... ... ... ... .. page 148Mr. Parker Clark.. ... ... ... ... ... ... ... ... ... ... page 58

Not Pictured:Mr. Kyle Arean-Raines . ... ... ... ... ... ... ... ... page 92

2011 NNIN REU Research Accomplishments Page xv

• nnin

reu in

tern

s b

y s

ite • n

nin

reu in

tern

s b

y s

ite • n

nin

reu in

tern

s b

y s

ite •

The 2011 NNIN REU Internsat Howard University

Mr. Won Jun Kuk . ... ... ... ... ... ... ... ... ... .. page 172Ms. Tiffany Dunston ... ... ... ... ... ... ... ... .. page 192Ms. Alicia Herro.. ... ... ... ... ... ... ... ... ... .. page 110Mr. Andreas Haggerty.. ... ... ... ... ... ... ... ... page 62Mr. Brendon Gobert. ... ... ... ... ... ... ... ... .. page 126

The 2011 NNIN REU Interns atThe Pennsylvania State University

Mr. Jia Kuang . ... ... ... ... ... ... ... ... ... ... ... page 70Ms. Julie Georgiev .. ... ... ... ... ... ... ... ... .. page 102Mr. Michael Gerhardt ... ... ... ... ... ... ... ... .. page 104Ms. Julie Chang .. ... ... ... ... ... ... ... ... ... ... .. page 8Mr. Giovanni Esteves ... ... ... ... ... ... ... ... ... page 44

Page xvi 2011 NNIN REU Research Accomplishments• n

nin

reu i

nte

rns b

y s

ite •

nnin

reu i

nte

rns b

y s

ite

• n

nin

reu i

nte

rns b

y s

ite •

The 2011 NNIN REU Internsat Stanford University

Front Row:Mr. Kevin Tien ... ... ... ... ... ... ... ... ... ... ... page 86Ms. Leah Weiss ... ... ... ... ... ... ... ... ... ... .. page 184Ms. Meagan Pipes ... ... ... ... ... ... ... ... ... .. page 156

Back Row:Mr. Karl Bayer ... ... ... ... ... ... ... ... ... ... ... page 94Mr. Hongliang Liang ... ... ... ... ... ... ... ... ... page 72Ms. Kelly Suralik. ... ... ... ... ... ... ... ... ... .. page 204

The 2011 NNIN REU Interns at theUniversity of California, Santa Barbara

Mr. Seyedshahin Ashrafzadeh ... ... ... ... ... .. page 140Mr. Issa Beekun.. ... ... ... ... ... ... ... ... ... .. page 142Mr. Travis Lloyd .. ... ... ... ... ... ... ... ... ... .. page 152Mr. Dominic Labanowski .. ... ... ... ... ... ... .. page 174Mr. Vinh Diep . ... ... ... ... ... ... ... ... ... ... ... page 12

2011 NNIN REU Research Accomplishments Page xvii

• nnin

reu in

tern

s b

y s

ite • n

nin

reu in

tern

s b

y s

ite • n

nin

reu in

tern

s b

y s

ite •

The 2011 NNIN REU Interns at theUniversity of Colorado, Boulder

Front Row:Mr. David Mallin.. ... ... ... ... ... ... ... ... ... .. page 154Mr. Seiya Suzuki . ... ... ... ... ... ... ... ... ... .. page 160Mr. Gawain Lau ... ... ... ... ... ... ... ... ... ... .. page 116

Middle Row:Ms. Hilary Hurst . ... ... ... ... ... ... ... ... ... ... page 66Ms. Mariah Szpunar. ... ... ... ... ... ... ... ... .. page 134

Back Row:Mr. Jeffrey Hart .. ... ... ... ... ... ... ... ... ... .. page 108Ms. Nina Hwang . ... ... ... ... ... ... ... ... ... .. page 208Mr. Drew Schiltz . ... ... ... ... ... ... ... ... ... .. page 158

The 2011 NNIN REU Interns at theUniversity of Michigan, Ann Arbor

Front Row:Ms. Emily Griffin. ... ... ... ... ... ... ... ... ... ... page 60Ms. Lisa Anne Hendricks .. ... ... ... ... ... ... .. page 194Ms. Jin Zhang. ... ... ... ... ... ... ... ... ... ... .. page 162

Middle Row:Mr. Carlos Biaou . ... ... ... ... ... ... ... ... ... ... page 56Mr. Yuki Matsuoka ... ... ... ... ... ... ... ... ... .. page 176Mr. Matthew Diasio . ... ... ... ... ... ... ... ... .. page 190

Back Row:Mr. Andrew Raebig.. ... ... ... ... ... ... ... ... .. page 198

Page xviii 2011 NNIN REU Research Accomplishments• n

nin

reu i

nte

rns b

y s

ite •

nnin

reu i

nte

rns b

y s

ite

• n

nin

reu i

nte

rns b

y s

ite •

The 2011 NNIN REU Interns at theUniversity of Minnesota-Twin Cities

Ms. Leah Laux ... ... ... ... ... ... ... ... ... ... ... page 24Mr. Francisco Pelaez, III ... ... ... ... ... ... ... .. page 180Ms. Lauren Otto.. ... ... ... ... ... ... ... ... ... ... page 30Mr. Nathanial Sheehan. ... ... ... ... ... ... ... .. page 200Ms. Laura Windmuller.. ... ... ... ... ... ... ... .. page 136

The 2011 NNIN REU Interns atThe University of Texas at Austin

Ms. Claire Spradling ... ... ... ... ... ... ... ... ... page 84Mr. Ting Chia Chang. ... ... ... ... ... ... ... ... .. page 146Ms. Elizabeth Fullerton ... ... ... ... ... ... ... .. page 100Mr. Kai He.. ... ... ... ... ... ... ... ... ... ... ... ... page 64Ms. Bethany Robinson . ... ... ... ... ... ... ... ... page 80

2011 NNIN REU Research Accomplishments Page xix

• nnin

reu in

tern

s b

y s

ite • n

nin

reu in

tern

s b

y s

ite • n

nin

reu in

tern

s b

y s

ite •

The 2011 NNIN REU Interns at theUniversity of Washington

Front Row:Ms. Morgan McGuinness ... ... ... ... ... ... ... ... page 48Ms. Courtney Crouch ... ... ... ... ... ... ... ... .. page 188Ms. Jennifer Gilbertson ... ... ... ... ... ... ... ... page 46

Back Row:Ms. Laurel Rognstad ... ... ... ... ... ... ... ... ... page 34Mr. Kevin Huang . ... ... ... ... ... ... ... ... ... ... page 16Ms. Julia Podmayer . ... ... ... ... ... ... ... ... ... page 54

The 2011 NNIN REU Interns atWashington University in St. Louis

Front Row:Ms. Abigail Halim ... ... ... ... ... ... ... ... ... .. page 106

Middle Row:Mr. Max Li.. ... ... ... ... ... ... ... ... ... ... ... ... page 26Ms. Audrey Dang ... ... ... ... ... ... ... ... ... ... page 42

Back Row:Mr. Ernest Puckett .. ... ... ... ... ... ... ... ... ... page 32Mr. Nicholas Heugel ... ... ... ... ... ... ... ... .. page 150Mr. Michael Akenhead.. ... ... ... ... ... ... ... .. page 138

Page xx 2011 NNIN REU Research Accomplishments• n

nin

reu i

nte

rns b

y s

ite •

nnin

reu i

nte

rns b

y s

ite

• n

nin

reu i

nte

rns b

y s

ite •

The 2011 NNIN iREU Program

Mr. Brian Chung.. ... ... ... ... ... ... ... ... ... .. page 186Ms. Clara Chiting Chow ... ... ... ... ... ... ... ... page 40Mr. Kevin Chen ... ... ... ... ... ... ... ... ... ... .. page 168Mr. Mark Brunson ... ... ... ... ... ... ... ... ... .. page 166Mr. Zachary Connell. ... ... ... ... ... ... ... ... .. page 170Mr. Brian Benton ... ... ... ... ... ... ... ... ... .. page 164

Not Pictured:Ms. Lauren Cantley.. ... ... ... ... ... ... ... ... ... page 98Mr. Steven Chase ... ... ... ... ... ... ... ... ... ... page 10Ms. Emily Hoffman.. ... ... ... ... ... ... ... ... .. page 112Mr. Sibu Kuruvilla ... ... ... ... ... ... ... ... ... ... page 20Mr. Evan Mirts ... ... ... ... ... ... ... ... ... ... ... page 28Ms. Fiona O’Connell ... ... ... ... ... ... ... ... .. page 196Ms. Michelle Pillers. ... ... ... ... ... ... ... ... ... page 76Mr. Joseph Smalley . ... ... ... ... ... ... ... ... .. page 202Ms. Margeaux Wallace . ... ... ... ... ... ... ... .. page 182Ms. Diana Wu . ... ... ... ... ... ... ... ... ... ... .. page 112

The 2011 NNIN iREG Program

Mr. Yuki Matsuoka ... ... ... ... ... ... ... ... ... .. page 176Mr. Seiya Suzuki . ... ... ... ... ... ... ... ... ... .. page 160

Not Pictured:Mr. Yoshihiro Nakano .. ... ... ... ... ... ... ... (no report)Mr. Yoichi Ogata . ... ... ... ... ... ... ... ... ... ... page 52

2011 NNIN REU Research Accomplishments Page xxi

• Com

monly

Used A

bbre

via

tions a

nd T

heir M

eanin

g •

µl �� microliterµm � �� micron, micrometerµN � �� micro-Newtons~ �� approximately< �� is less than> �� is greater than1D�� one dimensional2D�� two dimensional2DEG �� two dimensional electron gas3D�� three dimensional3DOM carbon �� 3D-ordered macroporous carbon4He� �� helium-4a-Si �� amorphous siliconA&M� �� Agricultural & MechanicalAC� �� alternating currentAFM�� atomic force microscopy/microscopeAFOSR�� Air Force Office of Scientific ResearchAg�� silveragLDL �� aggregated low-density lipoproteinsAgNO3 �� silver nitrateAgSR� �� silver-alkanethiolateAIC �� aluminum-induced crystallizationAl�� aluminumAl2O3 � �� aluminum oxideALD �� atomic layer depositionAlGaN �� aluminum gallium nitrideAM �� amplitude modulationAPD�� avalanche photodiodeAPS �� advanced photon sourceAr �� argonARC �� anti-reflective coatingArF �� argon fluorideAs �� arsenideAST �� aspartate transaminaseatm� �� standard atmosphere (as a unit of pressure)

ATRP� �� atom transfer radical polymerizationAu�� goldAuNPs �� gold nanoparticlesBAM � �� bisphenol aminomethylBCL3�� boron trichlorideBDM � �� 2,3-butanedione monoximeBES �� bioelectrochemical systemBHJ �� bulk heterojunctionBi �� bismuthBioSAXS�� biological small angle x-ray scatteringBN� �� boron nitrideBOE�� buffered oxide etchBOX �� buried oxide layerBPB �� bisphenol baseBPF �� bisphenol FBSA �� bovine serum albuminBST �� barium strontium titanateBTO �� barium titanateC �� carbonC �� centigradeC-V �� capacitance-voltageC3N4 �� carbon nitrideCAD �� computer-aided designCCI �� Centers for Chemical InnovationCCS �� continuous compositional spreadsCdS �� cadmium sulfideCdSe�� cadmium selenideCDW � �� charge-density-waveCe �� ceriumCF4 �� carbon tetrafluoride or tetrafluoromethaneCFD �� computational fluid dynamicsCFMA �� carbon-fiber microelectrode amperometryCH4 �� methane

Commonly Used Abbreviations and Their Meaning

Page xxii 2011 NNIN REU Research Accomplishments

• C

om

monly

Used A

bbre

via

tions a

nd T

heir

Meanin

g •

CHF3�� trifluoromethaneCIGS � �� copper indium gallium diselenideCION� �� colloidal iron oxide nanoparticlesCl �� chlorineCl2�� chlorine gasCl2/SF6 �� chlorine sulfur hexafluoridecm�� centimeterCMOS �� complementary metal oxide semiconductorCMOSFET �� complimentary metal oxide

field effect transistorCMP �� chemical mechanical polishingCNL �� charge neutrality levelCNTFET �� carbon nanotube field-effect transistorCo�� cobaltCO2 �� carbon dioxideCOF �� covalent organic frameworkCoFeAl�� cobalt iron aluminumCoP �� cobalt porphyrinCPC �� colloidal photonic crystalCPD �� contact potential differenceCr�� chromiumCRDS �� cavity ring-down spectrometerCTC �� circulating tumor cellCTC �� composite thermal capacitorsCTL �� confinement tuning layerCu�� copperCu2ZnSnS4� �� copper zinc tin sulfideCVD �� cardiovascular diseaseCVD �� chemical vapor depositionCW �� continuous waveCXRF �� confocal x-ray fluorescence microscopyDC� �� direct currentDCB �� double cantilever beamDCE�� 1,2-dichloroethaneDCM � �� dichloromethaneDEP �� dielectrophoresisDFT �� density functional theoryDFT �� discrete Fourier transformDH-PSF� �� double helix point-spread function

DI �� de-ionizedDIC �� differential interference contrastDMF�� dimethyl formamideDNA�� deoxyribonucleic acidDNP �� dynamic nuclear polarizationDODAB �� dimethyl dioctadecyl ammonium bromideDPPC� �� 1,2-dipalmatoyl-sn-glycero-3-phosphocholineDPPG �� 1,2-dimyristoyl-sn-glycero-

[phospho-rac-(1-glycerol)]DRAM�� dynamic random access memoryDRIE �� dry reactive ion etchDUV�� deep ultraviolete-beam �� electron beam lithographyE. coli �� Escherichia coliEB � �� exchange biasEBID � �� electron beam induced depositionEBL �� electron beam lithographyECD�� electrochemical detectorsECM�� extracellular matrixEDS �� energy dispersive spectroscopyEDTA �� ethylenediaminetetraacetic acidEELS� �� electron energy loss spectroscopyEG � �� ethylene glycolEIS� �� electrochemical impedance spectroscopyEMCCD �� electron multiplying charge coupled deviceEO � �� electro-opticEOT �� equivalent oxide thicknessEPICs �� electronic photonic integrated circuitsEPR �� enhanced permeability and retentionEr �� erbiumErAs �� erbium arsenideESM �� effective screening mediumEUV �� extreme ultravioletex vivo �� Latin for “out of the living” -- that which

takes place outside an organismFcCOOH �� ferrocenecarboxylic acidFDMA �� fluorinated perfluorodecyl methacrylateFDMNES �� finite-difference method approach to

predicting spectroscopic transitions

2011 NNIN REU Research Accomplishments Page xxiii

• Com

monly

Used A

bbre

via

tions a

nd T

heir M

eanin

g •

Fe�� ironFeDRAM�� ferroelectric dynamic random access memoryFEM �� finite element methodFES �� functional electrical stimulationFESEM�� field-emission scanning electron

microscopy/microscopeFET �� field-effect transistorFFTs �� fast Fourier transformsfg �� femto gramFIB �� focused ion beamFIR �� far infraredfJ �� femto JoulesFLT �� field-like torqueFM� �� frequency modulationFMR �� ferromagnetic resonanceFOTS� �� fluorosilane, tridecafluoro-

1,1,2,2-tetrahydrooctyltrichlorosilaneFRAP� �� fluorescence recovery after photobleachingFRET� �� fluorescence resonance energy transferFTIR�� Fourier transform infrared spectroscopyFWM� �� four-wave mixingGa�� galliumGaAs � �� gallium arsenideGaAsN �� gallium arsenide nitrideGaInNAs �� gallium indium nitride arsenideGaN �� gallium nitrideGaP �� gallium phosphideGASP �� growth advantage in stationary phaseGB� �� glass beadGBLMA �� a-gamma butyrolactone methacrylateGC� �� gas chromatographGC-C-IRMS�� gas chromatography combustion

isotope ratio mass spectrometryGe�� germanium

GEDI µdevices�� geometrically enhanced differential immunocapture microdevices

GFET� �� graphene field effect transistorGHz �� gigahertzGI �� gastrointestinalGMFI� �� gross mean fluorescence intensityGMR � �� giant magnetoresistanceGNR �� gold nanorodGNR �� graphene nanoribbonsGPa �� gigapascalGPC �� gel permeation chromatographyGPS �� global positioning systemGRIN� �� gradient refractive indexGUI �� graphical user interfaceGVD�� group-velocity dispersionh� �� hoursH-NMR � �� hydrogen-1 nuclear magnetic

resonance spectroscopyH2O2 �� hydrogen peroxideHAMA�� hydroxyl adamantyl methacrylateHAuCl4�� chloroauric acidhBN �� hexagonal boron nitridehcp� �� hexagonal close packingHCP1� �� Heme Carrier Protein 1HEMTs�� high electron mobility transistorsHF � �� hydroflouric acidHFEs � �� hydrofluoroethersHfO2 �� hafnium dioxideHg�� mercuryhigh-k �� high dielectric constantHMDS �� hexamethyldisilazaneHOMO-LUMO � �� highest occupied molecular orbital

& lowest unoccupied molecular orbitalHOPG �� highly oriented pyrolytic graphiteHRS �� high resistance stateHRTEM� �� high-resolution transmission

electron microscopyHS-ssDNA� �� thiol terminated single stranded

deoxyribonucleic acidHSQ�� hydrogen silsesquioxane

Page xxiv 2011 NNIN REU Research Accomplishments

• C

om

monly

Used A

bbre

via

tions a

nd T

heir

Meanin

g •

HSQ/FOx �� negative electron beam resist hydrogen silsesquioxane

Hz�� hertzI-V� �� current-voltageI/O � �� input/outputIC�� integrated circuitICP �� inductively coupled plasmaICP-MS � �� inductively coupled plasma massICP-RIE� �� inductively coupled plasma

reactive ion etcherIFVD � �� impurity free vacancy diffusionIID� �� impurity induced disorderingIIEI �� ion implant enhanced interdiffusionIJCMSSE�� International Journal of Computational

Materials Science & Surface EngineeringIn �� indiumin situ� �� Latin phrase which translated literally as

‘in position’ -- to examine the phenomenon exactly in place where it occurs

in vitro �� Latin for “within glass” -- refers to studies in experimental biology that are conducted using components of an organism that have been isolated from their usual biological context in order to permit a more detailed or more convenient analysis than can be done with whole organisms�

in vivo �� Latin for “within the living” -- experimen-tation using a whole, living organism

InAlN �� indium aluminum nitrideInAs �� indium arsenideInAs NWs�� indium arsenide nanowiresInGaAsN �� indium gallium arsenide nitrideInP� �� indium phosphideIPA� �� isopropyl alcohol

IPE� �� ion & plasma equipmentIPT� �� in-plane torqueIR�� infraredIRMS� �� isotope ratio mass spectrometryIrO2 �� iridium oxideIrOx �� iridium oxideISFET �� ion-sensitive field effect transistorITO �� indium tin oxideJP-8 �� Jet Propellant 8k �� dielectric constantK �� Kelvin (a unit of measurement

for temperature)kDa �� kilodaltonsKFM�� Kelvin force microscopykg�� kilogramkHz �� kilohertzKOH�� potassium hydroxideKPFM �� Kelvin probe force microscopyL/D �� length-to-diameterLaLuO3�� lanthanum aluminateLAO �� lanthanum aluminum oxideLED �� light-emitting diodeLER �� line edge roughnessLO � �� local oscillatorlow-k � �� low dielectric constantLPCVD � �� low pressure chemical vapor depositionlpm �� liter per minuteLRS �� low resistance stateLSPR� �� localized surface plasmon resonanceLWGs �� liquid-core/liquid-cladding waveguidesLWR�� line width roughnessM-OPTG �� microring-based optical pulse-train generatorMACE �� metal-assisted chemical etchingMAMA�� methyl adamantyl methacrylateMBE�� molecular beam epitaxyMCBJ �� mechanically controllable break junctionMD �� molecular dynamics

2011 NNIN REU Research Accomplishments Page xxv

• Com

monly

Used A

bbre

via

tions a

nd T

heir M

eanin

g •

ME� �� magnetoelectricMEG�� maleimide-ethylene glycol disulfideMEMs �� microelectromechanical systemsMFC �� microbial fuelMFMR�� microfabricated micro-reactorsMgO �� magnesium oxideMGs�� molecular glassesMHz �� megahertzmicron �� micrometer, aka µmMIFIS �� metal-insulator-ferroelectric-

insulator-semiconductormin �� minutesml �� millilitermm �� millimetermM �� millimolarMMA-MAA�� methyl-methacrylate-co-methacrylic acidMnO2 NPs � �� manganese oxide nanoparticlesMo� �� molybdenumMOCVD �� metal oxide chemical vapor depositionMONOS �� metal/oxide/nitride/oxide/semiconductorMOS�� metal oxide semiconductorMOSFET �� metal oxide semiconductor

field effect transistorMOVPE� �� metal organic vapor phase epitaxyMPM � �� multiphoton microscopyMQCA �� magnetic quantum-dot cellular automataMQW �� multiple quantum wellMRA�� multifunction reconfigurable antennaMRAM�� magnetic random access memoryMRFM�� magnetic resonance force microscopyMRI �� magnetic resonance imagingms�� microsecondMSM � �� metal-semiconductor-metalMTJ �� magnetic tunneling junctionmTorr� �� millitorrmV� �� millivolt

MVD � �� molecular vapor depositionMWNT�� multiwalled carbon nanotubeMΩ �� megohmsN �� nitrogenn-type �� negative semiconductorN2 �� nitrous oxideNaCl �� sodium chlorideNb�� niobiumNEMs �� nanoelectromechanical systemsNEXAFS �� near edge x-ray absorption fine structureNH4F � �� ammonium florideNi�� nickelNIR �� near-infrarednL �� nanoliternm � �� nanometerNMP�� n-methyl-2-pyrrolidoneNMR � �� nuclear magnetic resonance microscopy /

spectroscopyNNIN iREG �� National Nanotechnology Infrastructure

Network International Research Experience for Graduates Program

NNIN iREU �� National Nanotechnology Infrastructure Network International Research Experience for Undergraduates Program

NNIN REU �� National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program

NORIS�� nanometrology optical ruler imaging systemNPR �� nonlinear polarization rotationNPs �� nanoparticlesNPs �� nanoporesns �� nanosecondNSF �� National Science FoundationNSOM �� near-field scanning optical microscopyNSSP� �� nanostructured semipolarNVM � �� non-volatile memoryNW FETs �� nanowire field-effect transistors

Page xxvi 2011 NNIN REU Research Accomplishments

• C

om

monly

Used A

bbre

via

tions a

nd T

heir

Meanin

g •

O �� oxygenOFET� �� organic field effect transistorOh number �� Ohnesorge numberOLED �� organic light-emitting diodeONO�� oxide/nitride/oxideOPS �� optical particle sizerOPV �� organic photovoltaic cellsOST-MRAM � �� orthogonal spin-transfer

magnetic random access memoryOTFT� �� organic thin-film transistorp-n, p/n �� p-type & n-type semiconductors

joined togetherp-type �� positive semiconductorP/E� �� program/erasePa�� pascalsPAB �� post-apply bakePAE �� power-added efficiencyPAG �� photoacid generatorPAMAM �� polyamidoaminePANOMs �� planarized aperatures for

near-field optical microscopyPb �� leadPBG �� photonic bandgapPBPK� �� physiologically-based pharmacokineticPBS �� phosphate-buffered salinePC�� persistent currentPC�� photocurrentPCB �� printed circuit boardPCBM �� fullerene derivative [6,6]-phenyl-C61-

butyric acid methyl esterPCM �� phase change materialPCN �� photonic crystal nanocavityPd �� palladiumPD � �� photodetectorPDMS �� polydimethylsiloxanePE-GNR �� polyelectrolyte gold nanorod

PEB �� post-exposure bakePEC �� photoelectrochemicalPECVD � �� plasma enhanced chemical vapor depositionPEDOT:PSS �� poly(3,4-ethylenedioxythiophene):

poly(styrenesulfonate)PEG �� polyethylene glycolPEI� �� poly(ethyl imine)PFM�� piezo-response force microscopypH�� potential of hydrogenPh�D� � �� doctorate of philosophyPID �� proportional-integral-derivativepL �� picoliterPL�� photoluminescencePLD �� pulsed laser depositionPLGA �� poly(lactic-co-glycolic) acidPMGI� �� poly(methyl glutarimide)PMMA�� poly(methyl methacrylate)PmPV �� poly(m-phenylenevinylene-co-2,5-

dioctoxy-p-phenylenevinylene)poly-Si �� polycrystalline siliconPOP �� polyolefin plastomerPPM�� photolithographic phase masksPS �� polystyrenePSL �� polysterene latexPSMO �� praseodymium strontium manganitePSµM �� phase separation micro-moldingPt �� platinumPt/Ir �� platinum/iridiumPTX �� paclitaxelPV�� photovoltaicPVA �� poly-vinyl alcoholPVC �� polyvinyl chloridePVDF �� polyvinylidene fluoridePVP �� polyvinylpyrrolidonePy �� Ni81Fe19

PZT �� Lead zirconate titanate (PbZr0�52Ti0�48O3)Q �� high quality factorQD� �� quantum dotsQW �� quantum wellQWI �� quantum well intermixing

2011 NNIN REU Research Accomplishments Page xxvii

• Com

monly

Used A

bbre

via

tions a

nd T

heir M

eanin

g •

Re number� �� Reynolds numberRF�� radio frequencyRFID � �� radio frequency identificationRIE �� reactive ion etchRMS or rms �� root mean squareRNA �� ribonucleic acidROS �� reactive oxygen speciesRPEVCD �� remote plasma-enhanced

chemical vapor depositionRRAM �� resistive random access memoryRTA �� rapid thermal annealRTD �� resistance temperature devices � �� secondsS �� sulfurSA-MOVPE �� selective area metal organic

vapor phase epitaxySABC �� surface active block copolymersSAED �� selected area electron diffractionSAMs �� self-assembled monolayersSb �� metallic antimonySBH �� Schottky barrier heightSc�� scandiumSCAN �� single-chromatin analysis at the nanoscalesccm �� standard cubic centimeters per minutescCO2 �� supercritical carbon dioxideSCOFET �� single crystal organic field effect transistorSCORE�� SNARE Complex ReporterSDS �� sodium dodecyl sulfatesec � �� secondsSECM �� scanning electrochemical microscopySEM �� scanning electron microscopy/microscopeSERS� �� surface enhanced Raman spectroscopySF6� �� sulfur hexafluorideSFLS � �� supercritical fluid-liquid-solidSH � �� second harmonicSi �� siliconSi3N4 �� silicon nitrideSiAlON � �� silicon aluminum oxynitride

SiC� �� silicon carbideSiN �� silicon nitrideSiNWs �� silicon nanowiresSiO2 �� silicon dioxideSIROF �� sputtered iridium oxide filmSLBs�� supported lipid bilayersSLG �� single-layer grapheneSLM �� spatial light modulatorSLUG �� superconducting low-inductance

undulatory galvanometerSMS�� single molecule spectroscopySn �� tinSNARE � �� soluble n-ethylmaleimide-sensitive factor

attachment protein receptor complexSnO2 �� tin oxideSNPs�� silver nanoparticlesSOFC� �� solid oxide fuel cellsSOI �� silicon-on-insulatorSPCM �� scanning photocurrent microscopySPD �� switching phase diagramSPR �� surface plasmon resonanceSr2RuO4 � �� strontium ruthenateSrTiO3 �� strontium titanatessDNA �� single-stranded deoxyribonucleic acidSTEM �� scanning transmission electron microscopy/

microscopeSTJ �� superconducting tunnel junctionSTM �� scanning tunneling microscopy/microscopeSTO �� strontium titanateSTT �� spin-transfer torquesSTT-MRAM�� spin-transfer torque random access memorySVA �� solvent vapor annealingt-BOC �� tert-butoxycarbonylTa �� tantalumTa2O5 � �� tantalum pentoxideTaN �� tantalum nitrideTCO�� transparent conducting oxideTe �� crystalline tellurium

Page xxviii 2011 NNIN REU Research Accomplishments

• C

om

monly

Used A

bbre

via

tions a

nd T

heir

Meanin

g •

TE�� transverse electricTEC �� thermionic energy converterTEM �� tunneling electron microscopy/microscopeTER �� transepithelial resistanceTFM �� traction force microscopyTFT �� thin-film transistorTg �� glass transition temperatureTH � �� third harmonicTHz �� terahertzTi �� titaniumTiN �� titanium nitrideTiO2 �� titanium dioxideTIR-FRET� �� total internal reflection - fluorescence

resonance energy transferTLM �� transfer length measurementTM� �� transverse magneticTMAH �� tetramethylammonium hydroxideTMOS �� tetramethyl orthosilicateTMR�� tunneling magnetoresistanceTO � �� thermo-opticTO � �� torsional oscillatorTO � �� transformation opticsTPoS�� thin-film piezoelectric-on-substrateTRT �� thermal release tapeTSVs � �� through silicon viasTTD �� transverse translational diversityTTV �� total thickness variationTXM�� transmission x-ray microscopyUHV�� ultra-high vacuumUV� �� ultravioletUV-Vis �� ultraviolet-visibleV �� vanadiumV �� voltageVA-CNT �� vertically aligned carbon nanotubevdW �� van der WaalsVLS �� vapor-liquid-solidVRMs �� voltage regulator modulesVSM�� vibrating sample

magneto metryW �� tungstenWDM �� wave length-division

multi plexingWe number �� Weber numberWGM �� whispering gallery

modeXMCD�� x-ray magnetic circular

dichroismXPM�� cross-phase modulationXPS �� x-ray photoelectron

spectroscopyXRD �� x-ray diffractionXRR �� x-ray reflectivityYBCO �� yttrium-barium-

copper-oxideYBS �� y-branch switch

ZMW� �� zero-mode waveguideZnO �� zinc oxideZnO:Al �� zinc aluminum oxideZnS �� zinc sulfide or zinc-blendeZr �� zirconiumZTO �� zinc tin oxide

The National Nanotechnology Infrastructure Network

Research Experience for Undergraduates (NNIN REU) Program

2011 Research Accomplishments

• 20

11

NN

IN R

EU

Research

Acco

mplis

hm

ents

• 2011 N

NIN

REU

Research

Acco

mplis

hm

ents

•

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •

Page 2 2011 NNIN REU Research Accomplishments

Development of a Fluorescence-Based Quantification Method to Determine the Amount of Glycans Immobilized on a Surface

Andrew AcevedoBiomedical Engineering, Washington University in St. Louis

NNIN REU Site: Nanotechnology Research Center, Georgia Institute of Technology, Atlanta, GANNIN REU Principal Investigator: Dr. Julia Babensee, Biomedical Engineering, Georgia Institute of TechnologyNNIN REU Mentor: Nathan Hotaling, Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of TechnologyContact: [email protected], [email protected], [email protected]

Abstract:

The current best method for quantifying the amount of glycan immobilized on a surface involves connecting a large fluorescent linker to a carbohydrate and measuring the fluorescent intensity after surface modification to validate immobilization. However, it has been shown that the chemical moiety to which the carbohydrate is attached drastically alters the binding affinity of carbohydrate binding proteins. The purpose of this project is to address this issue. The approach was to attach a smaller, less intrusive azide linker to a glycan and then use this linker to attach the glycan to the surface. Microbead surfaces were chosen for modification because they allow for easy control of surface area and also because they allow carbohydrates to be presented on a scale that is either phagocytosable or non-phagocytosable to cells. The surface reaction is a substitution reaction in which a fluorescent moiety (dansyl chloride) is substituted for an azide-modified glycan. A dansyl-modified surface was first modified with Alexa Fluor 594, an azide-modified fluorophore. Green fluorescence, from the dansyl group, was shown to decrease, and red fluorescence, from the Alexa Fluor 594, increased over the course of the reaction. The final step used an azide-modified glycan and used the change in gross mean fluorescence intensity (GMFI) to quantify the amount of carbohydrate immobilized on the surface. Fluorescent microscopy and thermo x-ray photoelectron spectroscopy (XPS) were used to observe the surface modifications.

Methods:

Dansyl Modification. Amine functionalized silica microbeads, 1 µm and 45 µm diameter, were modified with dansyl chloride (DsCl) according to the procedure in [1]� The 45 µm beads were reacted at 5000:1 mol DsCl: mol amine and the 1 µm beads at 100:1 mol DsCl: mol amine to achieve optimal fluorescence.

Azide Fluorophore Substitution (See Figure 1A)� We added 300 µM Alexa Fluor 594 in dimethylformamide (DMF) to 106 1 µm dansyl modified beads in a 384-well glass-coated polypropylene plate� The plate was heated to 70°C� Using a Tecan Infinite F500 microplate reader, GMFI was measured at excitation/emission wavelengths of 340/535 nm and 585/617 nm over the course of three hours� The beads were washed between readings�

Azide Glycan Substitution (See Figure 1B). We added 5 mM 1-azido-1-deoxy-b-D-lactopyranoside in DMF to 4�35 × 106 1 µm dansyl modified beads in a 384-well glass-coated polypropylene plate� The plate was heated to 70°C� Gross mean fluorescence intensity was measured at excitation/emission wavelength of 340/535 nm over a span of 26 hours� The beads were washed between readings�

Figure 1: Schematic of the chemistry involved in (A) the azide fluorophore substitution reaction, and (B) the azide linked glycan substitution reaction.

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •

2011 NNIN REU Research Accomplishments Page 3

384-well and 1536-well glass-bottom microplates and 45 µm silica beads� These surfaces would then be used in high throughput quantification assays. The ultimate goal of this project is to use these modified surfaces to present glycans to dendritic cells to invoke controlled immune responses�

Acknowledgements:I would like to thank the National Science Foundation and the National Nanotechnology Infrastructure Network for supporting this Research Experience for Undergraduates program� I would also like to thank my mentor, Nathan Hotaling, and my Principal Investigator, Dr� Julia Babensee, for hosting me�

References:[1] “A real time monitoring using fluorescent dansyl group as a

solid phase leaving group”: Suenaga, T; Schutz, C; Nataka, T; Tetrahedran Letters, 2003, 44, 5799�

Figure 4: Nitrogen spectra of (A) dansyl modified, and (B) Alexa Fluor 594 modified 1 µm beads.

Fluorescent Microscopy and XPS. Using a Nikon TI fluorescent microscope, 45 µm beads, before and after dansyl modification, were observed at excitation/emission wavelengths of 340/535 nm� XPS analysis of unmodified, dansyl modified, and Alexa Fluor 594 modified 1 µm beads was conducted using a Thermo K-alpha XPS�

Results:To determine the lower threshold of number of beads that could be detected by the microplate reader, dansyl modified beads were serially diluted in a 2:1 ratio� It was found that at least 2000 45 µm beads and 325,000 1 µm beads gave a fluorescent intensity signal above background� Fluorescent microscopy was used as validation that dansyl modification had taken place, as seen in Figure 2, A and B�

The azide fluorophore substitution reaction was performed as verification that our azide linker was capable of substituting off the dansyl group on the beads. Red fluorescence, from the Alexa Fluor 594, and green fluorescence, from the dansyl chloride, were measured over three hours� A representative trial is displayed in Figure 3A. Red fluorescence is shown to reach a maximum and green fluorescence a minimum around 30 minutes after the reaction begins. Red fluorescence then begins to decrease slightly after that; this is suspected to be due to bleaching of the fluorophore after prolonged heat exposure� Figure 4 illustrates the XPS analysis of the nitrogen spectra of dansyl modified 1 µm beads and Alexa Fluor modified 1 µm beads. The Alexa Fluor 594 adds nitrogen with a higher binding energy as seen by the peak at 407eV (A), as opposed the nitrogen peak at 401eV in the spectra of the dansyl modified beads (B).

When the substitution reaction with the lactopyranoside was performed, a decrease in fluorescence greater than that of background was observed over the course of 26 hours� However, this reaction is still being optimized� This can be seen in a representative graph in Figure 3B�

Conclusion and Future Directions:These experiments have shown that fluorescent modification of a surface is a viable alternative to current quantification methods of surface immobilization� Dansylation of a surface is reproducible and can be used to monitor reaction progress� Azide-linked glycan substitution was shown to occur; however, that reaction is still being optimized�

The next step in this project would be to focus on the dansyl modification of other surfaces. Surfaces of interest include

Figure 2: Fluorescent microscopy photos of (A) unmodified, and (B) dansyl modified 45 µm beads. 20x objective.

Figure 3: Representative trials of (A) azide fluorophore substitution, and (B) lactopyranoside substitution reactions. 1 µm beads.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Microfluidic Protein Dialysis Device for X-Ray Scattering

Amani AlkayyaliBiomedical Engineering, Wayne State University

NNIN REU Site: Cornell NanoScale Science and Technology Facility (CNF), Cornell University, Ithaca, NYNNIN REU Principal Investigator: Richard Gillilan, Macromolecular Diffraction Facility

at Cornell High Energy Synchrotron Source (MacCHESS), Cornell UniversityNNIN REU Mentor: Soeren Nielsen, MacCHESS, Cornell UniversityContact: [email protected], [email protected], [email protected]

Abstract and Introduction:

The purpose of the microfluidic protein dialysis device is to create an in situ protein enrichment device for small-angle x-ray scattering (BioSAXS) on a synchrotron beamline, allowing a protein’s structure to be determined� The device allows users to create concentration series necessary for BioSAXS starting with a diluted sample, rather than diluting a pre-concentrated sample� This avoids potential irreversible aggregation of protein molecules in solution�

Most of the research was conducted using a five-layer device, consisting of two layers of polymethyl methacrylate (PMMA), two layers of double-sided medical adhesive tape, and a semi-permeable cellulose dialysis membrane� The CNF’s VersaLaser 3�50 was used to carve channels into each tape layer, to allow a lysozyme protein solution, a commonly used standard for testing structural biology, to flow through one side, while a polyethylene glycol (PEG) solution flowed through the opposite channel.

By the process of osmosis, water moved from the protein solution across the semipermeable membrane into the PEG solution, resulting in a concentrated protein solution� The five-layer device was tested using a spectrophotometer to determine the absorbance of protein solutions at various protein solution and PEG flow rates. The results obtained thus far indicate the protein was being concentrated�

Preliminary flow tests for a seven layer chip containing two dialysis membranes, and results of a Polydimethylsiloxane (PDMS) device, inspired by Chijung et al’s article [1], are also presented�

Experimental Procedure:The construction of the five-layer protein dialysis device, shown in Figure 1, began by using the VersaLaser to carve channels 1�00 mm wide and 0�05 mm deep into the tape, and 1�6 mm wide holes into the PMMA� The pieces were prepared, placed on top of each other, and aligned� Air bubbles were removed by tapping the device with a scissor handle. Teflon® tubes, used to allow fluid in and out of the device, were placed in the PMMA holes and adhered with epoxy and super glue� The device was tested with different

Figure 1: Side-view schematic of five-layer device.

Figure 2: PDMS device creation: (a) Resist on wafer, (b) Etch, (c) Resist Removal, (d) PDMS on wafer, (e) Final device.

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


covering the entire tape area with dialysis membrane, rather than just the channel area, and by placing epoxy around the sides of the device� To allow for more concentration, new devices can be designed to have longer channels, with a port for the x-ray� Also, the device should be tested in the synchrotron with an x-ray beam�

Acknowledgments:I would like to thank the National Science Foundation (NSF), NNIN REU Program, and Cornell NanoScale Facility; Richard Gillilan, Soeren Nielsen, and Magda Moeller; Rob Ilic and Melanie-Claire Mallison, the CNF REU Program Coordinators; all of the CNF staff, but particularly Beth Rhoades for her help with the Versalaser and PDMS; and CNF user Stephen Jones�

CHESS is supported by the NSF and NIH/NIGMS award DMR-0936384� MacCHESS is supported by NIH/NCRR award RR-01646�

References:[1] C. Kim, et al., “Microfluidic Dialysis Device Fabrication for Protein

Solution Enrichment and Its Enrichment Enhancement by Plasma Surface Treatment of a Membrane,” Journal of the Korean Physical Society, September 2007�

protein and PEG flow rates, controlled by pumps, for 0.1 mg/ml and 1�0 mg/ml starting protein solutions� The absorbance and concentration values were obtained by UV absorption spectroscopy at 280 nm�

The seven-layer device had one tape, and dialysis membrane, layer in addition to the five-layer device.

The construction of the PDMS device, shown in Figure 2, began by spinning negative resist, nLOF2020, onto a silicon wafer, undergoing a soft bake, exposing, undergoing a post exposure bake, developing, etching with the Unaxis 770, and removing resist with a hot bath� PDMS was then poured over the wafer, with protection to easily remove the PDMS� The chip was then assembled by placing the PDMS and a dialysis membrane in a Harrick Plasma Cleaner and layering as seen in Figure 2�

Results and Conclusions:As seen in Graphs 1 and 2, the five-layer device was able to concentrate a protein sample, sometimes by two-fold, and provide promising results� Graph 1 results were collected using UV absorption spectroscopy� The absorbances were converted to concentrations using the Beer-Lambert Law, A = ECl. Graph 2 shows much fluctuation, which could be the result of human or machine error� The data was collected using a NanoVue spectrometer to minimize sample evaporation� Although there was concentration, there is no clear relation between flow rates and concentration.

While long-term adhesion of the dialysis membrane to the tape and PDMS remains an issue, preliminary tests with 0�1 mg/ml and 1�0 mg/ml lysozyme solutions indicate that concentration is occurring on a reasonable timescale for BioSAXS measurements� The use of thin tape, carved with the VersaLaser, is a very effective fast-prototyping method for testing microfluidic designs. Regardless of the larger than expected fluctuations in concentration over time and the formation of leaks at high PEG flow rates, an enrichment factor of approximately 50% was achieved at protein flow rates of 1 µl/min into the device�

Based on these tests, a seven-layer chip was expected to provide a significantly increased protein concentration. The problem with the seven-layer device was placing tubes to allow fluid to travel to the middle tape layer without leaking. While further design refinements will be necessary to achieve an ideal 10-fold enrichment target, the existing five-layer device should serve as a good proof-of-concept test on the beamline� Also, the PDMS device had proper fluid flow, but it was difficult to correctly place the syringe in the channel, between the membrane and PDMS�

Future Work:Future work should consist of increasing the shelf life of the device� The current device design is able to last for four to twelve hours of continuous fluid flow through the chip, and should be able to last for 36 to 40 hours of continuous fluid flow. Future device designs should have reduced leakage, by

Graph 1, top: Changes in PEG flow rates, from 0.25 µl/min to 3 µl/min, with 0.1 mg/ml protein solution.

Graph 2, bottom: Changes in PEG flow rates, from 0.25 µl/min to 3 µl/min, with 1.0 mg/ml protein solution.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Gradient Surface Wettability Induced by Nanofilms on Titanium Surfaces and Osteoblastic Cell Morphology

Noelia AlmodovarChemical Engineering, University of Puerto Rico, Mayaguez Campus

NNIN REU Site: Nanotechnology Research Center, Georgia Institute of Technology, Atlanta, GANNIN REU Principal Investigator: Dr. Barbara Boyan, Biomedical Engineering, Georgia Institute of TechnologyNNIN REU Mentor: Jung Hwa Park, Materials Science and Engineering, Georgia Institute of TechnologyContact: [email protected], [email protected], [email protected]

Introduction:

Figure 1: Experimental process for modifying the samples’ surface.

Figure 2: Gradient in contact angle measurements.Values are given in degree measurements.

To reduce healing time and enhance osseointegration of bone implants, it is important to engineer and optimize ideal surface properties [1]� Surface properties of implanted materials can directly influence cellular response at the bio-interface between the living tissue and the outermost surface layer of orthopaedic and dental implants [2]� The surface properties that we focused on for this project were: surface roughness and wettability�

The aims of this project were to develop a coating method to control surface wettability without altering the complicated micro- and submicro-scale surface roughness, and to examine how surface wettability influences cell morphology in a time-dependent manner�

Experimental Procedure:Sand-blasted/acid-etched titanium (SLA) disks were oxygen plasma (OP) treated for two minutes on each side� Right after oxygen plasma treatment, the surface energy is increased and unless the surfaces are coated, they continuously lose surface energy until they are stable again� Using this fact, each disk was coated with a polyelectrolyte called chitosan (CHI) for 30 minutes after oxygen plasma treatment (0, 2, 10, and 24h)� This process is shown in Figure 1�

Results and Discussion:Surface roughness was measured using confocal laser microscopy (CLM)� Oxygen plasma-treated and then CHI-coated SLA surface roughness was not significantly different compared to control SLA surfaces�

Contact angle measurements were made between a drop of water and the surface of the disks to find the wettability of

the samples’ surfaces� We were able to develop a gradient in surface wettability, increasing the hydrophilicity of the material between each sample, as shown in Figure 2�

Surface chemical composition was analyzed using x-ray photoelectron spectroscopy (XPS)� Nitrogen was detected on CHI-coated SLA surfaces except on the control SLA

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


Figure 3: Chemical composition percentage of the samples.

Figure 4: Scanning electron microscopy images of the cell morphology.

surfaces� The treated and coated groups had a gradient in nitrogen composition. The first group to be coated had the least nitrogen in its composition and the last group that was coated had more nitrogen in its composition� The change in wettability between each sample could be explained as a function of the nitrogen content� The Table of the chemical composition of the surfaces is shown in Figure 3� A high-resolution analysis was done to determine if the wettability gradient was due to the quantity of NH3

+ in the samples� The results showed that the NH3

+ content increased as the surface was more hydrophilic�

MG63 osteoblast-like cells were plated on surfaces with different wettability to see the cell morphology on top of each sample’s surface� After 1�5, 3, 6, 12, and 24h incubations, cells were fixed, dehydrated, and dried with a critical point dryer� Cell morphology was examined using scanning electron microscopy (SEM)�

SEM imaging demonstrated that during the initial cell-material interactions, cell shapes, whether they were elongated or rounded, depended on surface roughness and surface wettability� When the cell morphology was evaluated, it was found that cells in the control group

elongated through the surface, while cells in the surfaces that were coated kept a rounder shape in the first 12 hours of cell growth and elongated at 24 hours� This indicates that surface wettability is important for cell morphology in the first 12 hours of cell growth.

At 24 hours of cell growth, all of the groups had elongated cell morphology, demonstrating that as time passes roughness guides cell morphology� The SEM images for the control and one of the groups are shown in Figure 4�

Conclusions:Our results showed that: we successfully developed a method to control surface wettability with the same polyelectrolyte without modifying the micro-scale surface roughness� MG63 cell shapes are sensitive to surface wettability at an early time point� As time passes, the surface’s roughness guides cell growth�

Acknowledgements:I would like to thank my principal investigator, Dr� Barbara Boyan and my mentor, Jung Hwa Park, for all of their guidance and support throughout this research project� I would also like to thank Dr� Boyan’s Research Group for allowing me to learn so much during their meetings� I would like to acknowledge the staff at the Nanotechnology Research Center, the National Nanotechnology Infra-structure Network International Research Experience for Undergraduates (NNIN REU) Program and the National Science Foundation for funding�

References:[1] Park J�H�, Schwartz Z�, Olivares-Navarrete R�, Boyan B�D�,

Tannenbaum R�; Langmuir� 2011�[2] Boyan, B�D�; Hummert, T�W�; Dean, D�D�; Schwartz, Z�;

Biomaterials� 1996�

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Fabrication of a Novel Microfilter for Circulating Tumor Cell Enrichment and Culture

Julie ChangChemistry and Physics, Harvard University

NNIN REU Site: Penn State Nanofabrication Laboratory, The Pennsylvania State University, State College, PANNIN REU Principal Investigator: Siyang Zheng, Ph.D., Department of Bioengineering, The Pennsylvania State UniversityNNIN REU Mentor: Mingda Zhou, Ph.D. candidate, Department of Bioengineering, The Pennsylvania State UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Detection of circulating tumor cells (CTCs) provides valuable information in diagnosis and prognosis of metastatic cancers, but the rarity of CTCs in blood makes it challenging to develop a method with sufficiently high sensitivity. Filtration of CTCs based on size has shown potential as a quick and inexpensive option. This project presents a novel three-dimensional slot-patterned microfilter efficient in viable capture of CTCs. This design better supports the captured cells, eliminates stress concentration on the cell surface during filtration, and prevents cell rupture. The filter design is also optimized to simplify the fabrication process. Device characterization confirms the structure, and preliminary filtration testing with cultured tumor cells exhibits effective viable capture. Thus the process achieved successful fabrication of a microfilter that can be further evaluated for performance in enrichment of CTCs.

Introduction:

Cancer metastasis, the spread of cancer throughout the body, is the leading cause of death for cancer patients� Detection and monitoring of circulating tumor cells (CTCs) for those with metastatic cancer can provide valuable information into treatment response, patient survival, and risk of relapse [1]� However, the rarity of CTCs in blood, on the order of 1 in 1010 blood cells, and the necessarily large sample volume, usually 7�5 ml blood, make it challenging to develop a method with sufficiently high sensitivity and selectivity.

The two most common methods of CTC isolation, immunomagnetic selection and density centrifugation, are expensive, labor-intensive, and do not have consistently high recovery rates [2]� Isolation based on size is potentially a cheaper and quicker procedure� Tumor cells are usually larger than blood cells, so a filter can be designed to capture CTCs while allowing blood cells to pass� With commercially available polycarbonate track-etched filters, lack of control over pore locations and clogging during filtration lead to poor performance [2]� In contrast, with microfabrication, both size and location of the pores can be precisely defined by lithography and etching techniques, enabling higher capture efficiency. Viable capture is also a goal of the current filter design. By controlling the filter’s microenvironment to capture CTCs without harm, post-filtration live cell studies can be made possible�

We present a novel “3D” microfilter efficient in viable capture of CTCs and relatively simple to fabricate� The device is a 10 µm-thick parylene membrane sectioned into

rectangular basket-like filtration units with 15 µm high walls and two parallel 4�5 µm wide slots on the bottom adjacent to each sidewall. Compared to single-layer designs, this filter has the advantage that both the bottom and the walls of the filter will support the captured cells to prevent rupture (see Figure 1)� It also reduces the number of lithography steps and overall fabrication complexity as compared to previous three dimensional (3D) filter designs.

Experimental Procedure:Fabrication started with 4-inch <100>-oriented p-type silicon wafers� As shown in Figure 2, a 15 µm-thick layer of sacrificial photoresist SPR 220-7 was spin-coated and patterned with photolithography to define the height of the basket walls� The trenches were sealed with 10 µm of Parylene-C to yield an even surface, a result of the conformal coating feature of parylene deposition� Parylene-C is very suitable for this application due to its biocompatibility, ease of processing, optical transparency, and low cost�

Figure 1: Forces on a trapped cell.

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


A thin layer of 100 nm aluminum (Al) was then thermally evaporated and patterned by a second lithography with SPR 955 photoresist� Oxygen plasma etching with this layer of Al as the mask defined the slots in the parylene. Finally, the sacrificial photoresist was dissolved, and the parylene membrane was released from the silicon substrate and cut into individual filter devices. Dimensions of the filter design were based on previous optimization tests�

Results and Conclusions:After careful adjustment of procedure specifics, the desired device structure was obtained, with clear-cut edges and precise patterning (Figure 3). The filters were tested by passing through solutions that contained GFP-labeled breast carcinoma UACC cells. The tests were run on several filters, all of which showed high capture efficiency (approximately > 90%)� It was evident that captured cells were caught near the slots by the wall (Figure 4), supporting the design

concept that both the bottom layer and wall would provide support to reduce stress on the cell� More rigorous testing of cell viability will be conducted in the future�

The filter needs to be further evaluated, but this project demonstrates that research is headed in the right direction� Ultimately, this device could aid significantly in the study of cancer metastasis in both clinical and research settings�

Future Work:Cell culture on the device will be observed to evaluate cell viability. A functional assay can confirm ongoing metabolic activity in captured cells� Scanning electron microscopy observations of the filter can also provide more information on cell membrane integrity post-capture. Further filtration testing will be conducted with blood samples containing or spiked with tumor cells to verify that only the tumor cells are captured while blood cells filter through. Optimization of the device will increase sample volume and/or filtration speed�

Acknowledgements:I would like to thank my PI, Dr� Siyang Zheng, for allowing me the opportunity to work in his lab and mentor, Mingda Zhou, for guiding me along the way� I would also like to thank Dr� Cheng Dong and his student, Ph� D� candidate Pu Zhang, for their help in testing� Much appreciation goes to Kathy Gehoski for coordinating the program� Funding for this project was provided by the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program and the National Science Foundation�

References:[1] M� Cristofanilli, et al� “Circulating tumor cells, disease progression,

and survival in metastatic breast cancer�” N Engl J� Med� 351(8), 781–791 (2004)�

[2] O� Lara, et al� “Enrichment of rare cancer cells through depletion of normal cells using density and flow-through, immunomagnetic cell separation�” Exp� Hematol� 32(10), 891–904 (2004)�

Figure 2: Micro-

fabrication process flow.

Figure 3: Fabricated 3D microfilter. Figure 4: Device post-filtration (arrow points to a captured tumor cell).

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Characterization of Embryonic Rat Cortical Cells Grown on Microcontact Printed Protein Patterns

Steven ChaseBiomedical Engineering and Biochemistry and Molecular Biology, Rose-Hulman Institute of Technology (Graduated)

NNIN iREU Site: Institut Für Bio- Und Nanosysteme (IBN), Forschungszentrum, Jülich, Germany NNIN iREU Principal Investigator and Mentor: Dr. Simone Meffert, Institute of Complex Systems

and Peter Grünberg Institute, Forshungzentrum JülichContact: [email protected], [email protected]

Abstract:

Microcontact printing (µCP) provides a simple, reproducible method of creating a defined pattern on various substrates for the purpose of studying neuronal cell growth and polarity (direction of axon and dendrite growth relative to the cell body). A protein mixture was stamped onto substrates (generally silanized glass) in a specific pattern. Cells are plated and allowed to grow for one, two, or three days in vitro (DIV). The growth of the axon and dendrites are controlled by the pattern and were visualized by immunostaining. Single-cell polymerase chain reaction (PCR) was used to analyze changes in gene regulation caused by the protein pattern.

Methods:

Microcontact Printing. Stamp master molds were created using standard photolithographic methods described elsewhere [1]. Polyolefin plastomer (POP) stamps were created by heating a hot embossing method� POP was placed on the master mold, heated to 90°C, and weighted with 900 grams�

Stamps were washed with ethanol for 10 minutes and soaked in a protein solution of 1:100 (v/v) FITC-labeled poly-L-lysine (FITC-PLL, green flourescent) and 1:200 laminin (v/v) in Gey’s Balanced Salt Solution (GBSS) for 20 minutes� Stamps were dried by nitrogen stream, and manually held against the substrate for two minutes� Figure 1 shows a diagram of the pattern used�

Substrates were homogeneously coated by covering the entire surface with the above solution for one hour and washed three times with GBSS�

Silanization. Glass coverslips (25 mm diameter) or microelectronic arrays (MEAs) were silanized using (3-glycidoxypropyl) trimethoxysilane (3-GPS)� Substrates were cleaned in a plasma generator at 180W for three minutes using oxygen plasma� Substrates were then placed in a desicator with 70 µl of 3-GPS at 45 mbar for one hour� Silanized substrates were kept in a dry, oxygen-free environment until use�

Figure 2: Representative micrograph of a neuron growing on a microgradient pattern. A) MAP 2 proteins, B) Tau-1 protiens, and C) unoccupied pattern.

Figure 1: Schematic representation of the microgradient pattern. Units are in micrometers. The label stands for slope of 0.04 (0.03/7.5), width of 2 µm, and length of 200 µm. Positive direction is towards larger squares.

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


Cell Culture. Embryonic rat cortical neurons were plated on patterned or homogeneously coated substrates at 16,000 or 50,000 cells per substrate, respectively� Cells were cultured in supplemented neurobasal media (InvitrogenTM), which was changed 3-5 hours after initial plating and every third day� Cells were allowed to grow for one to three (DIV) for PCR and three DIV for immunostaining�

Immunostaining. Cells at three DIV were fixed with paraformaldehyde and then blocked in a bovine serum albumin and goat serum solution� Neurons were stained with Anti-Tau-1 (axons) and anti-MAP-2 (dendrites) antibodies that were visualized by secondary antibodies bound to Alexa-488 (green) and Alexa-568 (orange), respectively�

Polymerase Chain Reaction (PCR). For bulk PCR, messenger RNA (mRNA) was isolated from cells grown for one, two, or three DIV and reverse transcribed into complementary DNA (cDNA)� This was done off of homogeneously coated substrates� For single-cell PCR, cells grown for one, two, or three DIV were placed in a patch clamp solution� Individual cells were then sucked off the substrate (homogeneously coated or microgradient patterns), cells were lysed, and mRNA was reverse transcribed to cDNA� cDNA was then cleaned from other cellular debris by centrifugation� PCR was then performed, on both methods of collection, using a Roche LightCycler� Melting point data was used to determine specificity of products and absolute quantification was used to determine the crossing point� The relative expression of desired product to the housekeeping gene can then be determined�

Results and Discussion:Microcontact Printing, Cell Culture, and Immuno-staining. Microcontact printing allowed for creation of reproducible protein patterns on various substrates� Figure 2 shows a representative micrograph of a gradient on silanized glass with plated neurons� Silanized glass was chosen over silicon oxide, hydrophilized glass, or polystyrene� Under qualitative analysis, silanized glass guided the cell on the gradients most effectively� The neuron grows on the node while the axon grows in the positive direction and the dendrites in the negative direction� This matches our predictions and results of others [1]�

PCR. Using the bulk PCR method, GAPDH was chosen as the housekeeping gene due to constant expression at DIV one, two, or three� Establishing this housekeeping gene

then allows for comparison of relative amounts of gene expression� To determine changes in gene expression, single-cell PCR was used as it, in principle, prevents contamination from the solution or other cell types� Unfortunately, a protocol for single-cell reverse transcription and PCR could not be established which eliminated contamination (GAPDH specific products were present in negative controls) and could quantify APOE expression�

Conclusions and Future Work:Microcontact printed patterns allows for guidance of neuronal cells and for analysis of the mechanisms behind neuronal cell polarity� Single-cell PCR can then be used to measure the genetic changes that occur because of pattern geometry or chemistry� In the immediate future, single-cell PCR optimization, additional controls (such as with a solid pattern) and additional data to determine statistical significance are needed to confirm the effects of these genes and show statistical significance. Long term, gene silencing experiments are planned to determine which genes have the greatest impact on neuronal polarity� This analysis adds to the understanding of the mechanisms of early neuronal cell growth and will allow for better control of neuron growth on bioelectronic devices�

Acknowledgements:I would like to thank the National Nanotechnology Infrastructure Network International Research Experience for Undergraduates (NNIN iREU) Program, the National Science Foundation, and Forshungzentrum Jülich, especially everyone in PGI-8, for making this opportunity possible�

References:[1] Fricke, Rita, Peter Zentis, Lionel Rajappa, Boris Hofmann, Marko

Banzet, Andreas Offenhäusser, and Simone Meffert� “Axon Guidance of Rat Cortical Neurons by Microcontact Printed Gradients�” Biomaterials (2010): 2070-2076� Web� 26 July 2011�

[2] Holtzman, D� M� “Low Density Lipoprotein Receptor-Related Protein Mediates Apolipoprotein E-Dependent Neurite Outgrowth in a Central Nervous System-Derived Neuronal Cell Line�” Proceedings of the National Academy of Sciences, 92�21 (1995): 9480-484� Web� 9 Aug� 2011�

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Nanomechanical Properties of Structured Biopolymer Networks

Vinh DiepNanoengineering, University of California, San Diego

NNIN REU Site: Nanotech, University of California, Santa Barbara, CANNIN REU Principal Investigator: Dr. Megan T. Valentine, Mechanical Engineering, University of California, Santa BarbaraNNIN REU Mentor: Bugra Kaytanli, Department of Mechanical Engineering, University of California, Santa BarbaraContact: [email protected], [email protected], [email protected]

Abstract:

The process by which nanoscale motor proteins operate in time and space to generate the force needed to complete cell division is not well understood. The current study aimed to better understand the forces that are generated during cell division on the length scale of whole cells. This top-down approach can give insight into the forces that are generated by the nanoscale motor proteins during cell division. Sea urchin cells were encapsulated in a hydrogel to be characterized using three dimensional traction force microscopy (3D TFM). The forces that a cell exerts can be quantified based on the deformations in the hydrogel. To identify the optimal matrix for 3D TFM, several hydrogels were tested with cell viability studies and magnetic tweezers-based microrheology. Of three potential hydrogels, a collagen-based one was found to be the most promising.

Figure 1: Model of how 3D TFM is used to relate displacements to forces

Introduction:The interactions between cells and their environment have become recognized as important in many biological processes, including cell proliferation and differentiation [1, 2]. This study aims to apply a modified form of traction force microscopy (TFM) in order to understand the effect of mechanical confinement on cell division. By tracking the displacements of fluorescent beads in the gel, forces that are generated during cell division can be determined (Figure 1) [3]� Sea urchin embryos were used as a model system due to the predictable timing of first division, large size (~ 50 µm), and availability at the laboratory� Hydrogels, the primary transducer of force in TFM, were characterized to determine compatibility for use in TFM�

Experimental Procedure:Cell Viability Tests. Peptide-based, hyaluronan-based, and collagen-based hydrogels were selected as potential candidates for use in TFM� A Live/Dead sperm viability kit (InvitrogenTM) was used to test for sea urchin embryo viability in each gel� To initiate spawning, KCl was injected into the sea urchins. Release of gold fluid signifies eggs, while a white fluid signifies sperm. Each fluid is collected and combined in artificial seawater (ASW) to initiate fertilization� The fertilization envelope is removed by passing embryos through a 53 µm mesh� Approximately 50 µL of the embryos are added to each sample of gel, prepared at concentrations of 25% gel in ASW� Propidium iodide from the Live/Dead kit was added to the cells as directed� Each sample was imaged under a scanning laser confocal microscope and cells with a permeabilized membrane (and thus dead) fluoresced red. Cells were also visually inspected for division within the expected period of 90 minutes�

Mechanical Tests. Gels were prepared at concentrations of 25%, 33%, 50% and 75% gel in ASW� Approximately 0�1 µL of magnetic beads was mixed into each gel� The gels were then flowed into glass cover slide flow-cells� Samples were tested under magnetic tweezers to characterize viscoelasticity� Known forces were applied to the gels by a permanent magnet in the vertical direction� The displacements of the magnetic beads were tracked based on the changing diffraction patterns around the magnetic beads�

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


Results and Conclusions:Confocal images of the propidium iodide-stained sea urchin cells were taken and compared to a control population of cells in ASW� The peptide gel showed large amounts of red staining, indicating no cells remained viable within the gel� This was likely due to the very low pH of this gel (~ 2�0)� The hyaluronan gel showed no red staining, but images showed no cell division, even after 120 minutes, indicating that the cells, while alive, were not healthy in this gel� The collagen gel performed best: no red staining was observed and cells also divided normally (Figure 2)� Thus, mechanical testing was performed on the collagen gels only�

Data from the magnetic tweezers were used to characterize the collagen gel elasticity� A plot of the displacement versus the force indicated that at short times, the displacements of the magnetic beads are directly proportional to the applied force, as expected for a linear elastic material� This linearity also holds over a range of gel concentrations (Figure 3)� By taking the inverse slope of this plot, the stiffness of each gel concentration can be determined in pN/nm (Figure 4)� The stiffness increases with increased gel concentration, indicating that the gel stiffness is tunable� This is expected, because as the concentration of gel increases, the polymer content increases, allowing more cross-linking to occur� By varying the concentrations from 25% to 75% gel in ASW, the stiffness increased by a factor of ~ seven�

Future Work:The next step in this project is to implement the collagen gel in TFM� Fluorescent beads will be embedded in the gel along with the cells� Stacks of images will be taken with a confocal microscope and assembled into 3D images as the cells divide� A tracking algorithm will monitor the displacements of thousands of beads in the gel� In addition, further work can be performed on other cells such as neural and cancer cells�

Acknowledgements:I would like to thank Dr� Valentine, Bugra Kaytanli, the Valentine Group, Kathy Foltz, and Angela Berenstein� Also, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program and the National Science Foundation for funding this research�

References:[1] Marklein R, Burdick J� Spatially controlled hydrogel mechanics to

modulate stem cell interactions� Soft Matter, 2010, 6, 136-143�[2] Saha K, et al� Substrate Modulus Directs Neural Stem Cell Behavior�

Biophysical Journal, 2008� Vol� 95� 4426-4438�[3] Legant W, et al� Measurement of mechanical tractions exerted by cells in

3D matrices� Nature Methods, 2010, 7, 969-971�

Figure 2: Cell viability data for three different hydrogels.

Figure 3: Results of magnetic tweezer studies for different collagen hydrogel concentrations.

Figure 4: Tunable stiffness of collagen hydrogel over a range of gel concentration.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


A Microchannel-Scaffold Electrode Array for Peripheral Nerve Interfacing

Jaideep S. DudaniBioengineering, University of California, Los Angeles

NNIN REU Site: Nanotechnology Research Center (NRC), Georgia Institute of Technology, Atlanta, GANNIN REU Principal Investigator: Ravi Bellamkonda, Biomedical Engineering, Georgia Institute of TechnologyNNIN REU Mentors: Yoonsu Choi, Biomedical Engineering, Georgia Institute of Technology;

Akhil Srinivasan, Biomedical Engineering, Georgia Institute of TechnologyContact: [email protected], [email protected], [email protected], [email protected]

Introduction:Injuries to the nerve can severally impair individuals, but peripheral nerves have limited regeneration ability� This means that there is potential for functional recovery� Implementing technologies for nerve interfacing can be used to guide nerve growth and introduce user control to advanced prosthetics� To this effect, studies have demonstrated that scaffolds, with appropriate topographical cues such as microchannels, enhance nerve regeneration through extended gaps [1, 2]� Current interfacing technologies fall far short of ideal� For example, cuff electrodes cannot directly interact with axons for directional control whereas other penetrating electrodes are too invasive [3]�

Implementing this technology allows for directed guidance of axonal growth such that it passes over the electrodes to facilitate electrical recording and stimulation� Despite great interest in pursuing this technology, much more quantitative and qualitative information is needed to understand how nerves grow and respond to electric signals in microchannels�

Here, we present a novel device to further study the above-described phenomena� Additionally, we present work being done in developing the functional scaffold� This work will be crucial in attaining direct control of advanced prosthetic limbs through closed-loop mechanisms�

Experimental Procedure:Multielectrode Array (MEA) Fabrication. We fabricated a MEA device for comparison purposes prior to starting on the microchannel MEA (mMEA). Briefly, we patterned NR4-8000P photoresist for liftoff� Gold was deposited by an e-beam evaporator to a height 2000Å with titanium adhesion of 500Å� Following liftoff, the gold electrodes were insulated using SU-8 2007 leaving exposed areas for cell culture and wiring� Polydimethylsiloxane (PDMS) was mixed in a 10:1 base to curing agent ratio to make a cell culture chamber� Conductive epoxy was mixed in a 1:1 ratio of Part A to Part B to glue on microwires (Figure 1)�

Microchannel MEA (mMEA) Fabrication. All steps for the standard MEA device were followed with the exception of patterning for microchannel walls� SU-8 2035 was used to pattern the walls before adding the PDMS chamber (Figure 1)� Wall alignment was such that there is a gold electrode in each microchannel�

Electrical Evaluation of Devices. Impedance measurements at f = 1000 Hz were conducted using a Neurocraft ICM� Connectors were soldered onto the microwires to allow setup with a stimulation and recording system� One microwire was connected to a function generator and the other to a signal recorder� These were connected to MC Rack software, which reported the input signal and the recorded signal�

Cell Culture on Devices. After sterilization with 70% ethanol and UV exposure, the cell plating area was coated with poly-L-lysine and placed in an incubator overnight� Following this, cortical neuronal cells were seeded on the surface� Images of cell culture were taken three days after seeding�

Microchannel Scaffold. Fabrication of the microchannel scaffold was accomplished� Additionally, preliminary patterning of electrodes on PDMS was accomplished using methods similar to those described previously�

Figure 1: a) Schematic cartoon of MEA device. b) Schematic cartoon of mMEA array device.

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


Results and Discussion:In vitro Device Fabrication. Successful fabrication of devices was accomplished for both the MEA and mMEA devices. Important fabrication concerns included sufficient undercut for electrode patterning, strong gold adhesion, and a smooth photoresist spin for microchannels� Additionally, microchannel walls were not extremely stable causing some to bend and fall over� This posed problems in preparation for cell culture as this caused stress to the walls�

this time limitation, cell stimulation and recording was not achieved� Despite this, cell culture was established� Interestingly, the cells preferentially adhered to SU-8 substrate (Figure 3)� Utilizing this, we can choose materials for better directional growth�

Microchannel Scaffold. Preliminary gold electrodes were successfully patterned on PDMS substrate (Figure 4)� After gold patterning, following microchannel wall fabrication, the final scaffold can be achieved. The final scaffold will be similar to the scaffold presented in Figure 4 with the addition of electrodes�

Conclusions:Here we have presented significant progress towards developing a system for peripheral nerve interfacing� We have developed a novel platform for better understanding nerve growth and interfacing in microchannels, as well as a system to serve as a control� We further demonstrated electrode viability and functionality in stimulation and recording� We further demonstrated proof of concept by seeding cortical neuronal cells� We hope that this device will serve as a platform for amassing quantitative information necessary to develop a system for use in vivo�

Additionally, work towards attaining a fully operational microchannel-scaffold electrode array shows promising results� This scaffold will need to be integrated with the necessary wiring and implantable chips for future use�

Acknowledgements:The authors would like to acknowledge the NSF, National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, NRC Staff, and members of the Bellamkonda group�

References:[1] Lacour S�P� IEEE 17, volume 5 (2009)�[2] Clements I�P� Biomaterials 30, 3834 – 3846 (2009)�[3] Srinivasan A� IEEE EMBSs 978-1-4244-4141-9/11 (2011)�

Figure 2: a) Impedance testing. b) Feasibility testing of stimulation and recording.

Impedance Testing. Impedance measurements showed that all recording sites had impedance to the cell culture chamber of less than 10 kΩ, with most under 2 kΩ (Figure 2). This illustrates that action potentials given off by nerve cells can be recorded� Additionally, impedance measurements on the device with microchannels showed no difference compared to the device without microchannels, illustrating that addition of walls did not damage electrode viability�

Signal Processing. We were able to illustrate the feasibility in simultaneously stimulating and recording of the device� Nerve cells need a stimulus to fire an action potential. Previously, in action potential studies, external stimuli have been used� Here we can stimulate and record simultaneously� As a control, we tested the system in saline and saw that there was good signal transduction over the devices (Fig�2)�

Cell Culture on Devices. Cortical neuronal cells take approximately 20 days to form a neural network� Due to

Figure 4: a) Preliminary patterning of electrodes (darker lines) on PDMS. b) Rolled scaffold without electrodes.

Figure 3: a) Nerve cells on MEA device. b) Nerve cells on mMEA device.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Multiplexed Silicon Nanophotonic Biosensing via Immobilized Protein Glycoconjugates

Kevin HuangEngineering, Trinity College

NNIN REU Site: Center for Nanotechnology, University of Washington, Seattle, WANNIN REU Principal Investigator: Prof. Daniel Ratner, Department of Bioengineering, University of WashingtonNNIN REU Mentor: James T. Kirk, Department of Bioengineering, University of WashingtonContact: [email protected], [email protected], [email protected]


Cell surface interactions are responsible for a myriad of biological processes, including cell signaling, cell growth, morphology and host-pathogen interactions� Many of these cellular functions are mediated by cell surface carbohydrates and glycoconjugates—including glycolipids, glycoproteins and extracellular polysaccharides� In particular, carbohydrates frequently act as receptors in the biochemical mechanisms involved in intercellular communication� Because glycans play an integral role in cellular biology, our interests lie in developing methods to better understand the biological functionality of glycans� One technology that can potentially offer such insight is silicon nanophotonic biosensors�

This project focuses on the integration of glycan receptors with silicon nanophotonic biosensors through the development of an optimized (i�e�, low cost and time) protein-carbohydrate conjugation and immobilization scheme� Protein-carbohydrate conjugates were used to present glycans of biological interest on silicon photonic platforms for the purpose of developing functional biosensors for glycomics applications� Both disaccharide-protein and tetrasaccharide-protein conjugates were employed to validate the bioconjugation strategy, and bioactivity was confirmed by the binding of relevant plant lectins (carbohydrate-binding proteins) ricin (RCA) and concanavalin A (ConA)� Finally, we demonstrated facile immobilization of oligosaccharide-protein conjugates for the specific and multiplexed detection of lectin analytes in solution using an integrated silicon nanophotonics biosensing platform�

Methods:General Procedure for Reductive Amination. Reductive aminations were carried out as described by Gildersleeve, et al� [1]� Lactose and maltotetrose were conjugated to bovine

Figure 1: Reductive amination reaction scheme (lactose, BSA).

serum albumin (BSA) to form BSA-galactose and BSA-maltotriose (BSA-M3) respectively�

Figure 2: Microring SEM image (Courtesy of M. Hochberg). As light travels within the linear waveguide, photons of frequency matching a resonant mode of the micro-ring are coupled into the ring. Moreover, the rings are constructed such that the coupled light experiences a phase shift of π. Any light coupled into the ring thus destructively interferes with incoming light in the linear waveguide, resulting in a sharp negative peak in the optical power spectrum density of the linear waveguide.

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


BSA (150 mg/ml), sodium borate (200 mM solution, pH 8�5), sodium sulfate (3M solution), saccharide (20 mM solution), water, and sodium cyanoborohydride (3M solution) were mixed then incubated at 56°C for 96 hours� Samples were then dialyzed against water for two days� The reaction scheme is shown in Figure 1�

Mass Spectrometry. Matrix assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF MS) was used to characterize conjugation efficiency. More specifically, the molecular masses of conjugates were compared to that of bare BSA protein (66�5 kDa)� Mass spectrometry was carried out on a Bruker Autoflex II MALDI-TOF instrument with 25% laser power in linear mode� We used a-cyano-4-hydroxycinnamic acid (CHCA) as the matrix for all the MALDI analysis� Samples were prepared by spotting 1 µl BSA/BSA-glycoconjugate and 1 µl CHCA onto MALDI target�

Silicon Microring Biosensing. Each microring structure is fabricated adjacent to a linear silicon waveguide� Figure 2 shows a scanning electron microscopy (SEM) image of the microring device� The resonant modes of a microring are directly proportional to the effective refractive index of the nearby media [3]� Any change in the refractive index (e�g� binding of a biomolecule) results in a change in the resonant frequency of the microring, and a corresponding shift in the negative peak of the linear waveguide optical power spectrum density�

To assess bioactivity, 1 µl BSA-glycoconjugate was first spotted onto the silicon surface via silicone masks, functionalizing the biosensor surface� The model lectins, RCA and ConA were used to test for bioactivity� RCA specifically binds to galactose residues (BSA-galactose) whereas ConA to mannose or glucose residues (BSA-M3)� As buffer was passed over the chip via integrated fluidics, a laser diode was rastered across discrete grating couplers in order to interrogate microrings� Photo-detectors were used to determine the output band of least power and thus the resonant frequency of the microring�

Results and Discussion:Mass spectrometry determined our conjugation ratio to be approximately seven sugars per BSA molecule (Table

1)� Multiplexed analysis with both BSA-galactose and BSA-M3 functionalized on the sensor platform was performed successfully (Figure 3). Specific binding to RCA was observed only with BSA-galactose functionalized microrings, while only BSA-M3 microrings responded to ConA� We have demonstrated both a low-cost conjugation scheme and the immobilization of conjugates for the specific and multiplexed detection of lectin analytes in solution on integrated nanophotonic platforms� In addition to the utility of our method, the incorporation of masking and spotting furthers the cost effectiveness of this conjugation scheme (only 1 µl of reagent is required per spot, and effective conjugation is achieved within 96 hours)�

Acknowledgements:This work was sponsored by NSF CBET (Award #0930411) and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program� The Center for Nanotechnology Facility was instrumental in image acquisition, and the Mass Spectrometry Center at the University of Washington’s School of Pharmacy provided instrumentation for MALDI analysis� Finally, the Ratner Lab in the Bioengineering Department at the University of Washington provided the means necessary for experimental procedure and analysis�

References:[1] Gildersleeve, J�, O� Oyelaran, J� Simpson, and B� Allred� “Im proved

Procedure for Direct Coupling of Carbohydrates to Proteins via Reductive Amination�” Bioconjugate Chem 19�7: 1485-490, 2008�

[2] Kirk, J�T�, G�E� Fridley, J�W� Chamberlain, E�D� Christensen, M� Hochberg, and D�M� Ratner� “Multiplexed Inkjet Functionalization of Silicon Photonic Biosensors�” Lab on a Chip (2011)�

[3] Washburn, A�L�, L�C� Gunn, and R�C� Bailey� “Label-Free Quanti tation of a Cancer Biomarker in Complex Media using Si Photonic Microring Resonators�” Analytical Chem (2009): 091022153841077�

Table 1: MALDI-MS Analysis of BSA, BSA-galactose, BSA-M3 and calculated ratio. [a = Calculated number of sugars bound per BSA.]

Figure 3: Microring association and dissociation curves for multiplexed testing of BSA-galactose and BSA-M3 functionalized rings with RCA and ConA. Double-hash marks at 50 minutes indicate renormalization from PBS buffer to HEPES2 + buffer.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Patterning Antigens with Near-Field Optics

Matthew KiokChemistry, Tulane University

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Professor Harold Craighead, Applied and Engineering Physics, Cornell UniversityNNIN REU Mentor: Dr. Christopher Kelly, Applied and Engineering Physics, Cornell UniversityContact: [email protected], [email protected], [email protected]

Abstract:

The membrane of a cell is a constantly flowing sea of lipids stacked together in a bilayer. Gaining a better understanding of the diffusion of lipids across the surface can be achieved by modeling with supported lipid bilayers (SLBs) that have incorporated ultraviolet (UV) polymerizable and fluorescent lipids. The goal of this project was to fabricate planar, microscale patterned substrates for creation of SLBs that would be observed with optical fluorescence techniques. Fused silica wafers were patterned by contact lithography, etched by reactive ion etching, coated with a thin film of aluminum by a metal evaporator, and processed by wet chemical liftoff, resulting in a smooth surface with aluminum lying flush with the glass surface. An SLB was constructed on the surface and the diffusion coefficient calculated by fluorescence recovery after photobleaching (FRAP). Future developments include improving the smoothness of the glass substrate, promoting SLB formation with polymerizable lipids, and working with patterns at the nanoscale. After optimization, these techniques can be applied to observe fluorescently tagged antigen-antibody-receptor complexes on the membrane surface and to better understand the cell membrane signaling cascade process.

Introduction:

The interaction between antigens, antibodies, and cellular receptors is a phenomenon not easily examined at high resolution through conventional optics� The use of fluorescent molecules to tag structures of interest permits the observation of their bulk diffusion with a fluorescence microscope� Supported lipid bilayers are used to model cell membranes and can easily be customized to simulate a variety of membrane types� Fluorescence recovery after photobleaching (FRAP) is employed to determine the diffusion coefficient for a particular lipid composition. The calculated diffusion coefficient is related to the affinity between a particular antigen-receptor complex and yields useful information about their relationship�

Experimental Procedure:The objective of wafer fabrication was to create a planar substrate with aluminum filled patterns that lay flat with the surrounding surface� The process began with a two-layer resist stack being spun on fused silica wafers� The bottom layer was comprised of LOR3A liftoff resist, approximately 150 nm thick, and the top layer consisted of S1813 photoresist, approximately 1 µm thick� Micron-size features were patterned by contact lithography, using an ABM contact aligner with a seven second exposure� After development in MF321 for 30 seconds, the patterns were

etched 150 nm deep by fluoroform/oxygen gas chemistry in an Oxford 80 etcher� Finally, 150 nm of aluminum was evaporated into the etched channels with a CHA evaporator� After wafer fabrication was complete, it was diced up into 14 mm × 14 mm squares. Pattern fidelity was verified by atomic force microscopy (AFM) after fabrication�

A supported lipid bilayer was constructed on top of each die via vesicle fusion, aided by calcium chloride� Shortwave UV radiation was shone through the bottom of the wafer and polymerized lipids not protected by the aluminum pattern� Afterwards, the diffusion coefficient of the hindered lipids in various regions of the die was determined by FRAP�

FRAP is a technique used to determine a diffusion constant for mobile, fluorescent substrates. A small region of the substrate is bleached by brief, intense exposure to radiation� The spot is then observed over a period of time and the diffusion of the photobleached lipids out of the spot and unbleached lipids into the region is noted� The rate of this diffusion is thus indicative of the overall diffusion coefficient for the membrane. The FRAP process employed was performed on an Olympus IX71 inverted microscope�

SLBs prepared on glass bottom dishes were examined on the microscope with 1% light intensity and photobleached with 100% intensity� Spot size was controlled by partially opening and closing the aperture on the microscope� This

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


process was recorded on video and processed in MATLAB®

to produce spot intensity versus time plots� The recovery time of the photobleached spot to half original intensity is proportional to the ability of the particular lipid composition to diffuse freely�

Results and Conclusions:This technique has been established to the extent that it can be applied to more complex membrane models� Lipid mobility is, however, affected by surface roughness at the wafer-aluminum interface. The initial diffusion coefficients for SLBs of the same composition determined by this method have ranged from 180 seconds to 1200 seconds� Currently there is significant variation in the results for what should be equivalent samples. Increasing the proportion of fluorescent lipid in the membrane composition may improve contrast and offer more consistent results� This technique has much potential to profile lipid interactions in great detail after it has been honed and perfected�

Future Work:Future work includes improvement of surface smoothness for more consistent lipid diffusion, the use of more complex lipid models that include receptors and fluorescently tagged antigens, and scaling the patterns down to the nanoscale�

Acknowledgements:This work was made possible by funding from the NSF and by the REU program by the NNIN� Thanks to Dr� Christopher Kelly for continued support during the project, as well as Professor Harold Craighead and the Craighead Research Group� Thanks also go to the staff of the Cornell NanoScale Facility who assisted with the research and helped organize the REU program�

Figure 1: Micron-scale patterns coated in aluminum with edge roughness due to angled deposition.

Figure 2: Profile scan of previous pattern generated from AFM.

Figure 3: Image of photobleached region of lipids and resulting intensity curve during FRAP.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Dry Electrodes for Electroencephalography Headsets

Sibu KuruvillaMaterials Science and Engineering, University of Illinois at Urbana-Champaign

NNIN iREU Site: Interuniversity Microelectronics Center (imec), Leuven, BelgiumNNIN iREU Principal Investigator: Dr. Pereira Hercules Neves, Heterogeneous Integrated Microsystems Group, imecNNIN iREU Mentor: Filip Vanlerberghe, Heterogeneous Integrated Microsystems Group, imecContact: [email protected], [email protected], [email protected]

Abstract and Introduction:Electroencephalography, or EEG, is the field of measuring and recording the brain’s electrical activity [1]� Although it is already widely used to study the brain and its disorders, the EEG measuring protocol can still be improved� Currently, the conventional EEG is read using “wet” electrodes (containing wet gel) placed on the scalp� The use of this electrolytic gel, though effective, brings about significant room for improvement in the EEG� The gel tends to cause itching and occasionally allergic reactions, and the frequent renewal of the gel is labor-intensive, hence expensive� While also being messy, the gel dries out within a few hours, restricting any long-term measurement from being taken� For uses such as in nascent wireless EEG headsets [2], being able to measure for long periods of time are very important� For that reason, the current study presents the design of a different type of electrode, a “dry” electrode, in which no gel is necessary� These dry electrodes instead employ needles to create contact with the conductive regions of the skin� Once designed and fabricated, the electrodes are tested in commercial neuroheadsets, and then compared to the standard wet ones�

Experimental Methods:In order to realize functional dry electrodes, a three step fabrication process was used� First, the electrodes were designed through computer modeling (SolidEdge) with their backsides designed to fit the neuro headset electrode holders. Next, the designs were made into prototypes through a 3D printing process, using the polymer FullCure720� These proto types were then coated with four metals: TiW (10 nm), copper (50 nm), TiW (10 nm), gold (50 nm), as seen in Figure 1� This deposition would allow for the ionic current within the body to pass through to the metal coating of the electrode� Gold was chosen as the outer layer due to its high electrical conductivity and biocompatibility [3]�

Once the dry electrodes were fabricated, they were placed into the neuroheadsets and tested for functionality, and then compared to the wet electrodes� Three tests were performed to compare the two types of electrodes. The first test involved reading the signal quality as displayed by software provided with the headsets� The second test measured the accuracy of facial expressions being read by the software� Finally, the third test measured the EEG of a subject repeating three different actions ten times: blinking, looking right, and looking left�

Results and Discussion:The first test sought to compare the signal quality of the dry electrodes to

that of the standard wet electrodes� When the wet electrodes were tested, the contact quality displayed a good signal — or at least 80% of the signal quality that would be measured in medical grade biosensors — for each of the electrodes, as expected� When tested with the dry electrodes, the sensors showed the same response — good signals for each location on the headset� Therefore, the dry electrodes performed comparably well to the wet electrodes in terms of signal quality�

The next test evaluated the neuroheadsets’ ability to read facial expressions� Each of the ten sensible expressions (from blinking to smiling to raising eyebrows) were tested 25 times and recorded for every successful response� This was done for both dry and wet electrodes in order to get an accurate comparison� Although the results are varied, they showed that the dry electrodes allowed the facial expressions to be read at 64% accuracy, while the wet electrodes were only at 61% (Figure 2)� Although 3% is not a significant difference, it can be seen that the dry electrodes allow reading of facial expressions at least as well, if not better, than the wet electrodes�

The third and final test involved recording the test subject’s EEG while performing a 3-action test� The three actions —

Figure 1: A dry electrode.

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


blinking, looking right, and looking left — were repeated ten times for both the wet and dry electrodes� The wet results (Figure 3) showed three distinct responses based on the action. For blinking, there are significant peaks in each of the electrode channels, while in both directional tests, one half of the brain responded more prominently� When the subject looked right, the right half of the brain showed more significant response, and vice versa for when the subject looked left� When this data was compared to that of the dry electrodes (Figure 4), mixed results were noted� In the case of blinking, the channels behaved relatively similarly in both cases� For the directional tests, however, the dry electrodes did not respond the same way� When looking to the right, some of the channels (AF4, F8, FC6) still performed similarly, while others (F4, T8) were stagnant� Comparable behavior occurred when looking to the left�

From these results, it is evident that dry electrodes have a high potential to replace wet ones, at least in the application of neuroheadsets� They eliminate the need for any electrolytic gel, thereby providing long term measurements�

Future Work:Despite the promising results, there is still work to be done� The amount of noise recorded in the EEG from the dry electrodes should be addressed, perhaps by designing different dimensions of electrodes based on their location in the headset� This way, increased stability on the scalp may promote less motion artifact� Once this issue is addressed, dry electrodes in EEG headsets can be realized�

Acknowledgements:The support from the NSF and the NNIN iREU Program, as well as the guidance from Filip

Figure 3, top: Wet electrode EEG results with headset map.Figure 4, bottom: Dry electrode EEG results.

Figure 2: Facial expression test results.

Vanlerberghe, Dr� Hercules Neves and colleagues at imec has made this work possible�

References:[1] Webster, J� G�, and Clark, J� W� (1992)� Medical Instrumentation:

Application and Design. Boston, MA: Houghton Mifflin Company.[2] EPOC Neuroheadset� (2010)� Retrieved 08 02, 2011, from Emotiv:

http://emotiv�com/store/hardware/epoc-bci/epoc-neuroheadset/[3] Voskericiana, G�

et al� “Biocom-patibility and biofouling of MEMS drug delivery devices�” Biomaterials (2003): 1959-1967�

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Synthesis of Poly(Amino Ether) Capped Gold Nanoparticles for Transgene Delivery

Olivia LambdinBiological Systems Engineering, University of Nebraska – Lincoln

NNIN REU Site: ASU NanoFab, Arizona State University, Tempe, AZNNIN REU Principal Investigator: Dr. Kaushal Rege, Chemical Engineering, Arizona State UniversityNNIN REU Mentor: James Ramos, Bioengineering, Arizona State UniversityContact: [email protected], [email protected], [email protected]


The search for safe and effective gene delivery agents is at the forefront of biomedical research� Delivering plasmid deoxyribonucleic acid (DNA) to live cells for gene therapy may one day cure diseases from autoimmune disorders to cancer� Polymers, liposomes, viruses, and gene guns have been investigated as potential vehicles for gene delivery [1]� Although these methods show promise, concerns regarding safety and low transfection efficiency remain. Thus, literature reflects a search for competitive, alternative gene therapy agents�

Our lab has created a promising polymer library for use in polymer-mediated gene delivery [2]� We chose lead polymers with high transfection efficacies to determine whether they could be employed for a one pot synthesis and capping of gold nanoparticles (AuNPs) aimed at gene delivery� By acting as a reducing and capping/protecting agent, the poly(amino ether) polymers encapsulate the colloidal nanoparticles, enhance the dispersion’s stability, and prevent their aggregation in biologically relevant media� We used leads from our polymer library, 1,4C-1,4Bis and EGDE-3,3’ and used 25-kDa poly(ethyl imine) (PEI) as a control�

Our research focused on synthesizing these poly(amino ether) gold nanoparticles (PAE-AuNPs) and characterizing their potential for biomedical applications� The character-ization techniques used included: ultraviolet-visible (UV-Vis) and infrared spectroscopy, zeta potential and dynamic light scattering (DLS) measurements, and MTT assay� Our research concluded with a preliminary qualitative analysis of in vitro transgene expression�

Experimental Procedure:Poly(amino ethers) were synthesized as described by Barua, et al� [2]�

Generation of Polymer Capped AuNPs. PAE-AuNPs were synthesized

at different polymer:gold weight ratios by mixing polymer and 0�001 M HAuCl4� Following formation, excess polymer was removed via centrifugation� The nanoparticles were centrifuged at 800 rpm for 10 min, and the supernatant was discarded� AuNPs were suspended in nanopure water and adjusted to an optical density of 0�25 a�u�

Cytotoxicity Studies with Prostate Cancer Cells. Human PC3 prostate cancer cells (PC3) were seeded in 24-well plates at a density of 50,000 cells/well� Following overnight incubation (37°C, 5% CO2), the cells were treated with different amounts of PAE-AuNPs and the well volumes were brought to 500 µL using serum free media� The MTT Assay was conducted to determine the cell viability�

Results:Kinetics of Formation Revealed Unique Qualities of Our Polymers. After mixing gold salt and polymer at various polymer to gold weight ratios, UV-Vis spectroscopy was employed to observe the rate of PAE-AuNP formation� The emergence of a peak at 520 nm indicated formation of spherical AuNPs (Figure 1)� It was observed that both 1,4C-1,4Bis and EGDE-3,3’ reduced and capped the gold salt faster than PEI� Also, at high polymer to gold ratios, such as 250:1 and 500:1, our lab’s polymers successfully made AuNPs, while PEI was unsuccessful�

PAE-AuNPs Exhibit Short-Term Stability in Biologically Relevant Media. Following PAE-AuNP formation, we investigated short-term optical stability of our nanoparticles when dispersed in serum-free media (SFM), and their

Figure 1: Spectra readings illustrate 50:1 PAE-AuNPs kinetics of formation.

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


UV-Vis spectrum was monitored over a 48-hour period� Short-term optical stability was determined by monitoring any changes in the peak at 520 nm� It was observed that all three polymer capped AuNPs exhibited good stability (Figure 2), indicating a lack of aggregation and potential for biological applications�

1,4C-1,4Bis-AuNPs Demonstrate Lower Cytotoxicity than PEI-AuNPs� Though polymers stabilize gold in biologically relevant media, they are potentially toxic to cells� Utilizing the MTT assay, we investigated the toxicity limits for 50:1 polymer capped AuNPs in the PC3 cell line� Our results revealed that 1,4C-1,4Bis capped AuNPs are less toxic than those capped by PEI at lower optical densities (Figure 3)� EGDE-3,3’ AuNPs were inconclusive� Our study also showed that optical densities below 0�005 a�u� are desirable for biological applications�

Preliminary Zeta Potential and DLS Measurements Reveal Colloidal Stability and Nanometric Scale of 1,4C-1,4Bis Capped AuNPs� Zeta potential and size measurements were performed for 50:1 1,4C-1,4Bis capped AuNPs with and without EGFP plasmid DNA� Our preliminary findings verified the stability of both polymer coated colloidal systems (Figure 4)� The positive charge indicates successful polymer coating on the AuNPs, while the measurement of 20 mV suggest colloidal stability of the complex� DLS showed naked 1,4C-1,4Bis capped AuNPs were approximately 104 nm in diameter and 126 nm after DNA complexing�

Preliminary Fluorescence Microscopy Images Show Successful in vitro Transfection of Enhanced Green Fluorescent Protein (EGFP) Plasmid DNA. PC3 cells were treated with different doses of polymer capped AuNPs loaded with various amounts of EGFP plasmid DNA� Preliminary fluorescent microscopy images exhibited cells had produced the green fluorescence protein, indicating successful transfection� The resulting images also showed that PEI and 1,4C-1,4Bis appeared equally effective, while EGDE-3,3’ AuNPs did not transfect�

Conclusions and Future Work:Our results revealed a potential non-viral vector for gene delivery� PEA-AuNPs demonstrated promising char-acteristics: stability in SFM, colloidal stability, low cytotoxicity, and successful transfection of EGFP plasmid

DNA in PC3 cells� We also discovered that our p o l y m e r , 1 , 4 C - 1 , 4 B i s , exhibited more favorable characteristics than the polymer standard, PEI� Future work will de-termine whether this class of nanomaterials is a com-petitive gene delivery vector

and will focus on producing a quantitative analysis of transfection efficacies, taking TEM images, and repeating the above experiments to ensure accuracy and precision of results�

Acknowledgements:I would like to thank Dr� Kaushal Rege for giving me the opportunity to work in his lab and for his continuous support� I would also like to thank James Ramos for dedicating countless hours to training/guiding me� I would like to acknowledge Dr� Trevor Thornton, NNIN REU Program, NSF, ASU, and the Rege Lab group for their support�

References:[1] Mohammed, S�, et al� Nonviral Gene Delivery: Principle,

Limitations, and Recent Progress� The AAPS Journal� 11 (4) 671-681 (2009)�

[2] Barua, S� et al� Parallel Synthesis and Screening of Polymers for Nonviral Gene Delivery� Molecular Pharmaceutics� 6 (1) 86-97 (2009)�

Figure 4: Preliminary zeta potential and DLS measurements for 50:1 1,4C-1,4Bis and PEI capped AuNPs.

Figure 2: 50:1 PAE-AuNPs dispersed in SFM maintain absorption peak at 520 nm.

Figure 3: Cytotoxicity comparison between 50:1 1,4C-1,4Bis and PEI capped AuNPs at different optical densities.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Neutrophil Chemotactic Response to Chemokine Gradients in a Microfluidic Device

Leah LauxBiomedical Engineering, Washington University in St. Louis

NNIN REU Site: Nanofabrication Center, University of Minnesota-Twin Cities, Minneapolis, MNNNIN REU Principal Investigator: Dr. Christy L. Haynes, Department of Chemistry, University of Minnesota-Twin CitiesNNIN REU Mentor: Donghyuk Kim, University of Minnesota-Twin CitiesContact: [email protected], [email protected], [email protected]

Abstract:

Neutrophils play an important role in the immune system by both degrading foreign bacteria and releasing mediators that contribute to the inflammatory response. Chemokines released from an injury site create a gradient of signaling molecules that guides neutrophil movement, or chemotaxis, towards the site. The human immune system employs a variety of different chemokines to achieve this task. In an effort to better understand how each of these chemokines affects chemotaxis, our group has fabricated a microfluidic device that allows us to expose neutrophils to a time-constant, controllable gradient of chemokines and monitor chemotactic responses of neutrophils to the gradient. Preliminary results suggest that a gradient of n-formyl-methionine-leucine-phenylalanine (fMLP) affects neutrophil chemotaxis.

Background:During neutrophil chemotaxis, chemokine molecules are released from the tissue near an injury� These chemokines form a concentration gradient; chemokines are most concentrated at the injury site and less concentrated far from the injury� Neutrophils are able to detect this gradient, orient themselves along it, and move in the direction of increasing chemokine concentration� The goal of this research was to fabricate a device capable of exposing neutrophils to a steady, controllable chemokine gradient and to determine neutrophil response to a range of different chemokines at various concentrations�

Experimental Procedure:Previous research in this area was limited by the inability to create a chemokine gradient that did not diffuse and change with time� The device design, based on work by the Jeon lab at the University of California at Irvine, overcomes this hurdle using microfluidics. Traditional photolitho-graphy techniques were used to pattern a polydimethylsiloxane (PDMS) layer with the desired device channels, and the layer was plasma-bonded channel-side-down to a glass slide to create the final device (Figure 1). A syringe pump flushed a chemokine solution and a buffer solution into the device through two separate inlet ports, which were systematically separated and mixed in a series of serpentine channels� After several rounds of this mixing, the device blended the original solutions into eight separate channels, each containing slightly different proportions of the chemokine and buffer (pictured in Figure 2). These channels filtered into

Figure 1: The completed device.

Figure 2: Device schematic.

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


on average, away from the IL-8 source� The motility index and velocity for IL-8 were comparable to that of the fMLP trials, indicating that this unusual result was not due to random, non-directional neutrophil movement� As chemotaxis toward IL-8 is a well-documented phenomenon, however, we expect that these results are due to either a small sample size or contamination of the IL-8 solution� In addition, all cells for a given trial are drawn from the same donor, increasing the chance of atypical results due to individual variation�

Future Work:Although our research to date includes only one chemokine in any given trial, biological chemokine gradients often involve multiple chemokines at once� To fully understand the role of chemotaxis in disease states, we must investigate how chemotaxis changes in multi-chemokine systems� The Haynes Lab aims to develop a hierarchy of chemokines, describing which are the most powerful mediators of chemotaxis and which provide more subtle, weaker signaling in multi-chemokine situations�

This research will be conducted by introducing a second chemokine to the microfluidic device through the inlet port previously containing buffer solution� Doing so will create a smooth gradient from one chemokine to another, exposing neutrophils to a set of competing signals� By observing which of the two chemokines the neutrophils chemotax towards, we will determine which chemokine dominates movement� Single-chemokine research conducted previously will serve as a control and comparison point for multi-chemokine trials�

Acknowledgments:I would like to thank the National Nanotechnology Infrastructure Network and the National Science Foundation for funding my research, as well as the University of Minnesota for the use of their facilities� In addition, I would like to thank my advisor, Dr� Christy L� Haynes, and my mentor, Donghyuk Kim, for their direction and support�

Figure 4: Data collected.

a cell collection chamber, producing an approximately smooth concentration gradient from the pure chemokine solution to pure buffer� Neutrophils were introduced into the cell chamber through a separate inlet port� Images of the cell chamber were captured every thirty seconds for one hour and combined to form a time lapse video of cell movement in response to the gradient in the device�

Cell trajectories were analyzed using several key parameters (Figure 3)� The chemotaxis index, which describes the efficiency of the cell in orienting and traveling toward a higher chemokine region, and the motility index, which indicates whether or not a high chemotaxis index is due to non-chemotactic random motion, form the basis of our analysis� Multiplying these parameters yields the effective chemotaxis index, which acts as a rough indicator of the degree to which the cell exhibited chemotaxis� Velocity data was also collected�

Results and Conclusions:Preliminary trials were completed with two well-known chemokines� IL-8, a chemokine involved in the inflammatory response, was chosen for its prominence in previous neutrophil chemotaxis research� A bacterially-derived chemokine, fMLP, was chosen as an example of chemokines derived from alternate sources� One trial was run with each of the two chemokines, and trajectories gathered in each trial were analyzed�

The data, pictured in Figure 4, displays some unexpected results� The effective chemotaxis index for the fMLP trial was positive, indicating that neutrophils oriented in the appropriate angle towards increasing chemokine concentration and traveled along a fairly straight path in that direction� The effective chemotaxis index for IL-8, however, was negative, indicating that neutrophils moved,

Figure 3: Chemotaxis parameters.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Gold Nanostructures with Tunable Photothermal Properties for Cancer Treatment

Max LiBiomedical Engineering, and Mathematics, Duke University

NNIN REU Site: Nano Research Facility, Washington University in St. Louis, St. Louis, MONNIN REU Principal Investigator: Professor Younan Xia, Biomedical Engineering, Washington University in St. LouisNNIN REU Mentor: Dr. Yucai Wang, Department of Biomedical Engineering, Washington University in St. LouisContact: [email protected], [email protected], [email protected]

Introduction:

Gold nanoparticles (AuNPs) have a variety of properties that make them favorable for biomedical applications, including the low reactivity and low cytotoxicity intrinsic to gold [1]� Photothermal therapy is based on the tunable electromagnetic absorption characteristics of AuNPs in a phenomenon known as localized surface plasmon resonance (LSPR)� The absorption peak can easily be tuned into the near-infrared (NIR) region by controlling the shape of the NP� The NIR region, covering wavelengths from 650-900 nm, is a region of low absorption for both water and hemo globin in soft tissues, allowing for photothermal therapy to be useful in biological settings [1]� We have synthesized and compared the photothermal properties of three different types of AuNPs: nanohexapods [2], nanocages, and nanorods� By studying the photothermal properties and experimental in vitro and in vivo capabilities of these three classes of nanostructures, we gain a sense for their potential as agents for cancer treatment�

Experimental Procedure:Nanoparticle Synthesis, Characterization, and Use in Photothermal Solution Studies. In order to take advantage of the NIR region, gold nanohexapods, nanocages, and nanorods were synthesized using the reported methods with their absorption peaks tuned to 800 nm as measured with ultraviolet-visible (UV-Vis) spectroscopy (Figure 1)� Solutions of nanohexapods, nanocages, and nanorods at various concentrations were irradiated with laser at a wavelength of 808 nm and data was captured with a thermal imaging camera�

Influx and Viability Studies in vitro. Nanostructures were also prepared for application in vitro and in vivo by process of PEGylation� Melanoma cancer cells were cultured and then incubated with our three types of gold nanostructures. Experiments determined the rate of influx of gold nanoparticles into melanoma cells, as influx was halted and the gold concentration within cells determined by inductively coupled plasma mass spectrometry (ICP-MS) after various incubational time periods� In addition, experiments of viability were conducted examining the cytotoxicity of various concentrations of nanoparticles� After a 24-hour incubation period, estimates of viability were made using an MTT assay�

Photothermal Studies in vivo. Preliminary photothermal studies in vivo were conducted by directly injecting a suspension of gold nanostructures into a tumor-bearing mouse� Tests compared an injection of nanohexapods with an injection of a saline control� In addition, the tissue distribution of

Figure 1: UV-Vis spectra, TEM images, and models of gold nanohexapods (A), nanocages (B), and nanorods (C).

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


Our preliminary studies in vivo showed that only five minutes of laser irradiation raised the tumor temperature to over 60°C in our nanohexapod trial, while the temperature change recorded during our injection of saline was insignificant. Also, tissue distribution studies indicated that 48 h post injection, nanohexapods had the highest level of accumulation in tumor among the three types of nanoparticles at a rate of nearly 15%� While the largest proportions of nanohexapods were found in the liver and the spleen, nanoparticle accumulation in the tumor was significant and is based on the enhanced permeability and retention (EPR) effect�

Conclusions and Future Work:Gold nanohexapods, when compared with nanorods and nanocages, were shown to have comparable or higher levels of photothermal efficiency, cellular uptake, viability, and accumulation in tumor� As such, they have great potential for use in cancer therapy� However, a large-scale method for nanohexapod synthesis still needs to be developed� In addition, a detailed investigation of the in vivo toxicity of gold nanohexapods and the photothermal efficiency of tail-vein injections would contribute greatly to our understanding of nanohexapods as photothermal agents�

Acknowledgements:I would like to thank Professor Younan Xia for his guidance in my project and Dr� Yucai Wang for serving as my mentor throughout my research� I would also like to thank the National Science Foundation, the NNIN REU Program, the staff of Nano Research Facility (NRF) at Washington University in St� Louis, and all of the members of the Xia Research Group�

References:[1] Chen, J� Y�; Yang, M� X�; Zhang, Q� A�; Cho, E� C�; Cobley, C� M�; Kim, C�;

Glaus, C�; Wang, L� H� V�; Welch, M� J�; Xia, Y� N�, Gold Nanocages: A Novel Class of Multifunctional Nanomaterials for Theranostic Applications� Adv� Func� Mater� 2010, 20, 3684-3694�

[2] Xia, Y� N�; Kim, D� Y�; Yu, T�; Cho, E� C�; Ma, Y� Y�; Park, O� O�, Synthesis of Gold Nano-hexapods with Controllable Arm Lengths and Their Tunable Optical Properties� Angew� Chem� Int� Ed� 2011, 50, 6328-6331�

AuNPs was measured 12 hours and 48 hours after tail-vein injection into a mouse with ICP-MS�

Results and Discussion:Photothermal studies of gold nanohexapods showed that even a low concentration at 12.5 pM was sufficient to cause a rapid temperature increase of a few degrees� At a higher concentration of 0�4 nM, the temperature increased to 60°C after only 3 min of laser irradiation (Figure 2)� In addition, photothermal characterization on all three types of gold nanostructures indicated comparable photothermal efficiency for nanohexapods, nanocages, and nanorods.

Our cellular uptake results demonstrated that gold nano-hexapods had significantly higher levels of uptake than both nanocages and nanorods after an incubation time of 20 h (Figure 3). While the exact mechanism of influx is unknown, the shape of a nanoparticle largely determined the rate at which it was picked up by the cell� Also, our results indicated that gold nanohexapods displayed no apparent cytotoxicity even up to concen trations of 1 nM, while nanocages and nanorods affected a cell viability of approximately 20% lower than nano hexapods at a comparable concentration� We believe this was because the silver present in the gold-silver alloyed nanocages was toxic to living cells, while one of the reagents, hexadecyltrimethylammonium bromide (CTAB), involved in nanorod synthesis was particularly difficult to completely remove from the nanorod product and was also toxic to cells�

Figure 3: Influx results for PEGylated gold nanohexapods, nanocages, and nanorods.

Figure 2: Thermal images of gold nanohexapods (0.4 nM in concentration) after 0 (A), 45 (B), 90 (C), 135 (D), 180 (E), and 225 s (F) of laser irradiation. The temperature is in degrees celsius.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Chromatin Remodeling by Brahma Motor Unit on Mono-Nucleosome DNA

Evan MirtsBiology/Physics, Truman State University

NNIN iREU Site: Delft University of Technology (TU Delft), NetherlandsNNIN iREU Principal Investigator: Cees Dekker, Bionanoscience, TU DelftNNIN iREU Mentors: Gautam Soni and Rifka Hoogeboom-Vlijm, Bionanoscience, TU DelftContact: [email protected], [email protected], [email protected], [email protected]

Abstract:

Cellular deoxyribonucleic acid (DNA) exists as chromatin, a carefully regulated complex of DNA and proteins. Nucleosomes are the basic units of chromatin organization, and a suite of remodeling protein complexes constantly reorganizes them to allow or restrict access to genetic information. One such complex, Brahma, is abundant in Drosophila, but little is understood of its interactions with DNA and nucleosomes. Through gel assays and observation under atomic force microscopy (AFM), we have investigated and characterized binding functionality of the Brahma core motor protein (BRM) with mono-nucleosomes and naked DNA.

Introduction:

Access to genetic information and its expression is restricted in the form of chromatin and the action of regulatory complexes� Several large chromatin remodeling complexes are highly conserved among eukaryotes, including the Brahma complexes belonging to the trithorax group (trxG), which are abundant at sites of actively transcribed DNA in Drosophila [1]� The remodeling and recruitment mechanisms of Brahma are still poorly understood; although, the complex has been found to be an important activator to trxG proteins [2] and to be closely associated with the nucleosome-assembly protein ASF1 [3]� In Drosophila, Brahma complexes contain at least ten subunits, including BRM, a large ATPase motor-unit of the SWI/SNF family. We combined purified BRM with short strands of mono-nucleosome chromatin to investigate its activity directly associated with DNA and the potential for BRM participation in nucleosome remodeling�

Methods:Yeast DNA of 344 bp (3D), 394 bp (9D), 529 bp (5D), and 606 bp (6D) was used as specified. Nucleosomes were formed by mixing DNA and histone octamers (H8) in a salt dialysis (2 M to 250 mM KCl) at 4°C in 20 mM Tris-HCl, pH 7�5; 1 mM EDTA, pH 8�0; 10 mM DTT, 0�5 mM benzamidine; and subsequently exchanged to a storage buffer (20 mM Tris-HCl, pH 7�5; 1 mM EDTA, pH 8�0; 1 mM DTT)� DNA and H8 were combined in molar ratios of 1:1 for 3D, 9D, 5D, and 6D� The presence of mono-nucleosomes was confirmed in 6%PAGE/0.8%agarose gels. BRM (185kDa) was isolated from Drosophila embryo

extract and purified by column chromatography. Purified BRM was stored at -80°C in 0�1 M KCL, 0�01% NP40, 0�2 mg/ml Flag peptide + protease inhibitors, and 0�5 mM DTT� DNA and BRM were combined in 50:1 and 4:1 ratios [2] at room temperature for 60 minutes in a remodeling buffer (10 mM Tris-HCl, pH 7�5; 50 mM KCl, 4 mM MgCl2, 2 mM ATP)� BRM was added to the DNA constructs with mono-nucleosomes, to naked DNA, and to naked DNA with 1:1 H8 in solution�

Results:A bright band of material did not migrate through the gels in samples for which BRM was added to DNA with mono-nucleosomes (Figure 1); this effect was also observed for lanes with BRM and naked DNA in 6% PAGE� The intensity of the band increased with the concentration of BRM (50:1 vs 4:1), and ratios of 4:1 BRM always produced a highly visible band in the loading well� Ratios of 50:1 occasionally yielded no clear difference between the 50:1 lane and the BRM-absent control� Progressively higher concentrations of BRM also increased the brightness of the nucleosome band versus the BRM-absent lane and lanes with lower concentrations� Conversely, the intensity of the naked DNA band was observed to decrease in higher BRM concentrations due to binding of BRM to naked DNA (Figures 2 and 3)� In samples for which BRM, naked DNA, and H8 were combined, the sample behaved as though it did not contain H8, and no nucleosome band appeared in the gel�

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


Conclusions:We observed BRM to interact with DNA containing nucleosomes and naked DNA in our remodeling conditions� It appears likely from gel results that BRM binds readily to both DNA configurations, and this behavior was observed under AFM with proteins bound to naked DNA and DNA which had at least one nucleosome� Though there have been suggestions that Brahma complexes may assist in nucleosome assembly activity [3], BRM did not facilitate the formation of nucleosomes in mixtures of DNA and histone octamers� The intensity of mono-nucleosomes was lower, however, in samples without BRM than in samples which had the protein� By binding to DNA, BRM may exert a stabilizing effect on nucleosomes, which we observed to disintegrate after long periods at temperatures over 20°C�

Acknowledgements:Thanks to Rifka Hoogeboom-Vlijm, Gautam Soni, and Cees Dekker for opportunity for this research and their invaluable guidance, and for funding the NSF and National Nanotechnology Infrastructure Network iREU Program�

References:[1] Mohrmann, Lisette, K� Langenberg, J� Krijgsveld, et al� 2004�

Differential Targeting of Two Distinct SWI/SNF-Related Drosophila Chromatin-Remodeling Complexes� Mol� Cel� Biol� 24:3077-88�

[2] Kal, Arnoud J�, T� Mahmoudi, N� B� Zak, et al� 2000� The Drosophila Brahma complex is an essential coactivator for the trithorax group protein Zeste� Genes Dev� 14:1058-71�

[3] Moshkin, Yuri M�, J� A� Armstrong, R� K� Maeda, et al� 2002� Histone chaperone ASF1 cooperates with the Brahma chromatin-remodeling machinery� Genes Dev� 16:2621-26�

Figure 3: BRM height (nm) distributions measured by AFM. A: Unbound BRM with distribution curve and mean height (11.48 ± 0.703). B: BRM attached to DNA with distribution curve, mean height (15.49 ± 0.585), and predicted height for DNA (1.92 ± 0.0925) added to unbound BRM without conformational effects from binding. Conformational changes from binding were observed (t0 = 0, t = -4.49, P < 0.001).

Figure 1: 6% PAGE of BRM with naked DNA and mono-nucleosome samples. A1, D1: Ratios of 50:1 DNA:BRM are the lower limit of observable effects on naked DNA. A2, D2: Large BRM concentrations (4:1) in nucleosome samples showed a striking increase in the loading well mass and reduction of free DNA, and a moderate increase of nucleosome intensity for 3D and 6D compared to 50:1 and 1:0. B, C: 9D & 5D, respectively, (4:1).

Figure 2: AFM images of BRM and 606 bp DNA (6D). A: Free BRM (lower) and BRM bound to naked DNA (upper). B: BRM bound to naked DNA. C: BRM bound to naked DNA (left) and to DNA with a nucleosome (right). D: BRM on DNA with a nucleosome.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Metallic Nanostructure for Surface Plasmon Resonance Biosensing

Lauren OttoPhysics and Mathematics, Bethel University

NNIN REU Site: Nanofabrication Center, University of Minnesota-Twin Cities, Minneapolis, MNNNIN REU Principal Investigator: Prof. Sang-Hyun Oh, Electrical and Computer Engr., University of Minnesota-Twin CitiesNNIN REU Mentor: Dr. Nathan Lindquist, Electrical and Computer Engineering, University of Minnesota-Twin CitiesContact: [email protected], [email protected], [email protected]

Abstract:

Metallic nanostructures enable the creation of surface plasmons and offer many applications in nanotechnology and biotechnology. Utilizing these surface plasmon resonances, label-free optical sensing techniques with high sensitivity at the surface of the metal are possible [1]. With a multiplex imaging technique, the plasmons can provide detection specific to the area in which bio-molecules are located [2, 3]. This leads to more efficient and less costly biosensing. Further more, metal nanostructures are easy and cheap to produce using template stripping with reusable silicon molds [4]. This can reduce current experimental costs, leading to a new generation of sensors based on plasmonics. This project covers fabrication of the metallic nanostructure sensing chips, realization of a surface plasmon resonance imaging system, and biosensing.

Background:

Figure 1: SPR imaging schematic.

dry, reactive ion etching or wet etching with potassium hydroxide to pattern the silicon wafer� Alternatively, focused ion beam lithography was also used to pattern the silicon� A method called “template stripping” produced metallic nanostructures and enabled the template to be used repeatedly� A thin layer of metal was deposited onto the patterned silicon� Epoxy was applied and cured between the metal and a glass slide� Poor adhesion between the metal and the Si allowed the template to be stripped, which yielded a nanostructure pattern attached to the epoxy and glass�

Detection and Imaging:An imaging technique was needed to collect data during a SPR biosensing experiment� In the past, it was common to use a white light microscope to excite SPs� A new setup, as seen in Figure 1, was constructed involving an inverted microscope for both reflection and transmission experiments using a helium neon (HeNe) laser to excite SPs� In both cases, a camera collected an image that could be analyzed with MATLAB® or LabVIEW�

Surface plasmons (SPs) are oscillations of conduction-band electrons induced by incoming photons in an electromagnetic wave� These electron oscillations form an electron density wave that is confined to travel along the metal-dielectric interface� Through multiplexing and real-time detection, biosensing with SP resonance (SPR) can be performed� It has the capability of being more efficient and less costly than other types of biosensing based on fluorescence imaging, potentially reducing experimental costs. As analytes in fluid flow over receptors on metallic nanostructures, binding occurs. The analyte will bind to specific receptor partners. This binding changes the resonant wavelength of the SPs, resulting in an observable change that can be analyzed� This allows biophysical properties of the analyte or receptor in question to be determined, which is important for new drug discovery�

Fabrication:Silicon templates were needed for efficient and in-expensive metallic nanostructure fabrication� Electron-beam lithography and photo lithography were used with

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


Results:When using white light, colors appear on the surface of the nanostructure based on the resonant wavelength of the SPs� This is represented with the varying shades in the black and white image in Figure 2a� During a biosensing experiment, binding would occur in certain areas, and this color would change as the resonant wavelength of the SPs changed� This shift in wavelength was observed by the camera and analyzed� For biosensing experiments with the HeNe laser setup, the camera image showed varying intensities based on the resonant wavelength of the SPs� SPs were being created in areas of sudden intensity changes� Figure 2b shows dark gratings at the top, where SPs were being created, and brighter gratings at the bottom, where there were no SPs� When binding occurred, the intensities throughout the image would change, resulting in observable and analyzable data�

The developed metallic nanostructures were tested with fluids differing in index of refraction, beginning with water (1�33) and continuing with additional amounts of glycerol (1�47)� Figure 3a shows how the resonant wavelength of the SPs changed based on the sudden increase of index of refraction� The system was washed and bovine serum albumin (BSA) was also used� Another washing followed� This can be seen in Figure 3b�

These results show that our devices can be used for multiplexing and real-time detection, which is valuable for biosensing experiments�

Future Work:In the future, finalization and optimization of the SPR imaging system will be completed� This will allow for a new method of data collection usable for numerous experiments� Additionally, further biosensing experiments will be conducted to determine the capability of the devices newly fabricated�

Acknowledgements:I would like to thank my principal investigator, Professor Sang-Hyun Oh, my mentor, Dr� Nathan Lindquist, Timothy Johnson, Si Hoon Lee, the Nanofabrication Center staff, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, and the National Science Foundation�

References:[1] Homola, J� and Yee, S� S� and Gauglitz, G�, “Surface plasmon

resonance sensors: review,” Sensors and Actuators B: Chemical 54, 3-15 (1999)�

[2] Smith, E� A� and Corn, R� M�, “Surface plasmon resonance imaging as a tool to monitor biomolecular interactions in an array based format,” Applied Spectroscopy 57, 320-332 (2003)�

[3] Lindquist, N. C. and Lesuffleur, A. and Im, H. and Oh, S. H., “Sub-micron resolution surface plasmon resonance imaging enabled by nanohole arrays with surrounding Bragg mirrors for enhanced sensitivity and isolation,” Lab Chip 9, 382-387 (2009)�

[4] Nagpal, P� and Lindquist, N� C� and Oh, S� H� and Norris, D� J�, “Ultrasmooth patterned metals for plasmonics and metamaterials,” Science 325, 594-597 (2009)�

Figure 3: Biosensing data. Resonant plasmon wavelength for differing indices of refraction given by varying ratios of water (1.33) and glycerol (1.47). (b) BSA binding to the gold surface and washing with water.

Figure 2: Gratings for SPR biosensing. (a) Image taken with white light microscope. (b) Image taken with SPR imaging setup. The outer areas of the Gaussian beam show the greatest differences.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Three-Dimensional Label-Free Photoacoustic Microscopy of the Tumor Microenvironment in vivo

Ernest PuckettBiomedical Engineering, The University of Texas at Austin

NNIN REU Site: Nano Research Facility, Washington University in St. Louis, St. Louis, MONNIN REU Principal Investigator: Dr. Lihong V. Wang, Department of Biomedical Engineering, Washington University in St. LouisNNIN REU Mentor: Junjie Yao, Department of Biomedical Engineering, Washington University in St. LouisContact: [email protected], [email protected], [email protected]

Introduction:

Angiogenesis is a hallmark of cancer� Tumor angiogenesis produces a unique tissue microenvironment with features such as vessel dilation, perivascular detachment, and irregular vasculature� Monitoring tumor angiogenesis is important to understanding the tumor microenvironment and developing new means to treat cancer� Photoacoustic microscopy has been proven powerful for three-dimensional imaging of blood vessels with capillary resolution�

The photoacoustic microscope (PAM) relies on the photoacoustic effect and uses hemoglobin as the endogenous contrast agent to produce high-resolution images [1]� Here, we propose that photoacoustic microscopy can be used for imaging of tumor microenvironment in vivo and quantification of key parameters of tumor growth�

Materials and Methods:PAM System Calibration. The imaging system was first calibrated for the measurement of absolute total hemoglobin concentration� Five solutions of lysed bovine blood were measured by PAM� The calibration factor between the hemoglobin concentration and PA signal amplitude was calculated via linear fitting. Then, the system was calibrated for the measurement of absolute absorption coefficient. Five different absorbers: lysed bovine blood, red ink, blue ink, a 1:1 mixture of red and blue ink, and DQOCI were measured. The absorption coefficient of each dye at 570 nm was first measured using a spectrophotometer. Then the calibration factor between the absorption coefficient and PA signal amplitude was calculated for each individual dye�

After calibrating for absorption coefficient, a graphite phantom was made for daily system calibration� The PA signal from the graphite phantom was measured under normal system alignment, and the absorption coefficient

Figure 2: 2D back projections (left) were used to make the 3D region of interest (right).

Figure 1: White light image of the tumor with the box signifying the tumor region (left), and the corresponding PA image (right).

calibration factor was calculated using DQOCI dye� This procedure was repeated under three additional system alignments�

Longitudinal Imaging of the Tumor Microenvironment. As a demonstration, a U87 glioblastoma tumor was longitudinally imaged for three weeks with white light imaging and the PAM as shown in Figure 1�

Tumor Volume and Tumor Vasculature Volume. Tumor volume and tumor vasculature volume were both quantified by back-projecting the three dimensional data onto planes at different angles to create a three dimensional region of interest as shown in Figure 2� This three dimensional region of interest was used to calculate the tumor volume and determine the tumor boundaries� The tumor vasculature

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


volume was calculated by computing the volume of all the blood vessels within the tumor boundaries�

Total Hemoglobin Concentration. Total hemoglobin concentration was extracted at each pixel using the hemoglobin calibration factor that was calculated during the hemoglobin calibration experiment, and then averaged across the tumor region�

Vessel Tortuosity. Vessel tortuosity was quantified as the ratio of the vessel length to the linear distance between its two endpoints as shown in Figure 3� The vessels that were visible in all images were identified using a cross-section-tracing based vessel segmentation algorithm� Tortuosity for each of these vessels was calculated and then averaged among the vessels�

Results:System Calibration. The PAM was successfully calibrated to measure absolute hemoglobin concentration. Linear fitting revealed that the ratio of PA signal (A�U�) to hemoglobin concentrations (g/dL) was 0�027 with an offset of -0�009� The absorption coefficient calibration experiment yielded unexpected results� Each dye showed a linear relationship between the absorption coefficient and PA signal. The ratio was 0�0018 for lysed blood, 0�00021 for blue dye, 0�0042 for DQOCI, 0�00099 for the blue and red dye mixture, and 0�0018 for the red dye� Testing of the graphite phantom revealed that the relationship between absorption coefficient calibration factor and graphite PA signal is linear regardless of system alignment�

Longitudinal Imaging of the Tumor Microenvironment. The results show that, within three weeks, the tumor had grown to a size of 0�72 cm3� Vascular volume grew to 0�03 cm3, and vessel tortuosity increased by 25% compared with the baseline�

Figure 3: Tortuosity is defined as the ratio of the length of the vessel (L) to the distance between the end points of the vessel (D).

Conclusions and Discussion:PAM was calibrated to measure total hemoglobin concen-tration and absorption coefficient. However, the absorption coefficient calibration results were different than what was expected� Assuming that the dye-water solutions all had roughly the same physical properties besides absorption coefficient, it was expected that all of the data for all of the dyes should be collinear�

The data points for each dye solution were linear, but for a given absorption coefficient each dye had a different PA signal� The explanation for these results is not yet know, but further absorption coefficient calibration experiments are being conducted� The graphite phantom proved to be a reliable way to calibrate the PAM to measure absorption coefficient. Given that there is a different relationship between absorption coefficient and PA signal for each absorber, more experiments need to be performed� In addition to the calibration experiments, PAM was used to longitudinally image a tumor for three weeks and extract the physiological parameters including hemoglobin concentration, tumor vasculature tortuosity, tumor volume, and tumor vasculature volume� These results demonstrate that PAM is a good tool for imaging the tumor microenvironment�

Acknowledgements:I would like to thank my PI Dr� Lihong V Wang and my mentor Junjie Yao for their guidance throughout this project� I would also like to thank all of the members of the optical imaging lab at Washington University in St� Louis along with all of the NRF staff� In addition, I would like to thank the NNIN REU Program and the NSF for providing me with funding and resources�

References:[1] Wang, L�V�; “Multiscale photoacoustic microscopy and computed

tomography”; Nature Photonics, 3, 3-9 (2009)�

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Biosensing Based on Surface-Enhanced Raman Scattering

Laurel RognstadChemical Engineering, Tennessee Technological University

NNIN REU Site: Center for Nanotechnology, University of Washington, Seattle, WANNIN REU Principal Investigator: Qiuming Yu, Chemical Engineering, University of WashingtonNNIN REU Mentor: Jiajie Xu, Chemical Engineering, University of WashingtonContact: [email protected], [email protected], [email protected]

Abstract and Introduction:Raman scattering involves the inelastic scattering of a photon from a sample in which the incident photon transfers energy to a molecule and is emitted at a different frequency� This change in energy allows for a spectroscopic technique that provides a molecular fingerprint of a sample. However, Raman scattering is only a small fraction of the light that is scattered from a molecule� Noble metal nanoparticles or nanostructures can greatly enhance Raman scattering to the orders of 106-1014-fold due to the localized surface plasmon resonance, i�e�, the so-called surface-enhanced Raman scattering (SERS)� SERS is an effective technique for detection and distinction of biological samples�

Using specially tailored substrates with quasi-three-dimensional (3D) gold nanostructures, developed by Dr� Yu’s group at the University of Washington, the Raman spectra of numerous biological species were enhanced greatly� SERS was first used to confirm that two different strains of the bacteria Vibrio parahaemolyticus could be identified and distinguished in a mixed sample by comparison with the SERS barcoding of each strain� Additionally, the optimal concentration of bacteria to obtain high reproducibility and intensity was determined� Furthermore, SERS was employed to detect chemicals spiked in whole milk to explore its capability of being used for direct detection of chemical contaminates in complex media�

Figure 1: 3D illustration of the quasi-3D gold nanostructure composed of

a separated gold thin film.

Methods:Specially tailored substrates with quasi-3D gold nano-structures, developed by Dr� Yu’s group at the University of Washington, were used for biosensing [1]� Four 50 µm × 50 µm nanohole arrays with a 400 nm hole diameter and 100 nm edge-to-edge distance were generated on each chip� For bacterial samples, the arrays were fabricated on indium tin oxide (ITO) coated glass with a 300 nm hole depth� For milk samples, 390 nm hole depths were produced on a silicon chip� Each chip was coated with 50 nm gold as illustrated in Figure 1�

Bacterial Samples. A 1:1 ratio of V. parahaemolyticus strains 551 and 3256 was prepared in situ by placing a SERS-active chip, cleaned in UV ozone for 20 minutes and rinsed with deionized water, in a custom-made Teflon® holder� After vortexing, 500 µl of the V. parahaemolyticus solution was deposited onto the quasi-3D nanohole patterns, and a microscope cover slide was carefully positioned on top� To determine the optimal concentration of bacteria for SERS detection, five samples of V. parahaemolyticus 551 of different concentration, ranging from 104 to 108 cfu/ml, were prepared in situ in the same fashion�

Milk Samples. Three samples were prepared: whole milk, whole milk spiked with 0�1% 4-mercaptopyridine (4-MP), and whole milk spiked with 0�0001% rhodamine 6G (R6G)� These samples were prepared using a dip-and-dry method in which a clean SERS-active chip was placed in a vial containing 2 ml of the sample� The chips were carefully removed and dried with air after the vials were refrigerated for three hours�

Using a Renishaw inVia Raman microscope, SERS spectra were gathered from 400 to 2000 cm-1 with a 785 nm laser and 5 mW power�

Results and Conclusions:As seen in Figure 2, the barcode of the mixed sample of bacteria clearly displays the characteristic peaks of each pure strain, especially the most intense peak of V. parahaemolyticus 551 at 1532 cm-1 and the most intense peak of V. parahaemolyticus 3256 at 525 cm-1�

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


Ergo, it was shown that SERS can be used to detect separate strains of the same bacteria in a mixed sample�

Figure 3 illustrates the peak intensities and error bars of different bacterial concentrations� The optimal concentration of bacteria to obtain high reproducibility and intensity to prepare bacterial samples was found to be 108 cfu/ml� At this concentration, the greatest peak intensity was found, which is necessary for efficient detection of bacteria. Additionally, at concentrations below 108 cfu/ml, the normalized intensity of peak 1539 cm-1 drops significantly, and considerable error is introduced�

When 4-mercaptopyridine (4-MP) was added to whole milk, it was found that the spectra of 4-MP predominated over the spectra of the milk� This is most likely due to the aromatic thiol adsorbing quickly and easily to the gold, forming a self-assembled monolayer before fats, proteins, carbohydrates, minerals, and other larger components of milk could reach the gold surface� However, as illustrated in Figure 4, R6G was successfully detected, proving as a basis for SERS detection of chemical contaminates in milk�

Future Applications:Future work includes testing strains of V. parahaemolyticus from different sources, along with collecting Raman spectra of live toxic marine phytoplankton using an optical tweezer method� Additionally, future work involves spiking milk with pesticides to imitate a more natural contamination, and testing unpasteurized milk for contaminants using SERS�

Acknowledgements:I would like to thank the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and the National Science Foundation for funding this research� Additionally, I’d like to thank my PI, Qiuming Yu, my mentor, Jiajie Xu, and fellow undergraduate researcher, Stephanie Hare, for their guidance and support� The Nanotech User Facility at the University of Washington is gratefully acknowledged for use of their research tools� Bacterial samples were collected by the National Oceanic and Atmospheric Administration Northwest Fisheries Science Center laboratory�

References:[1] Xu, J�, Zhang, L�, Gong, H�, Homola, J� and Yu, Q� (2011), Tailoring Plasmonic

Nanostructures for Optimal SERS Sensing of Small Molecules and Large Microorganisms� Small, 7: 371–376� doi: 10�1002/smll�201001673�

Figure 2, top: Comparison of the normalized SERS barcoding of V� parahaemolyticus 551, a 1:1 ratio of V� parahaemolyticus 551 to

V� parahaemolyticus 3256, and V� parahaemolyticus 3256.

Figure 3, middle: The normalized intensities of peak 1539 cm-1 for bacterial concentrations of V� parahaemolyticus 551 ranging from 104 to 108 cfu/ml.

Figure 4, bottom: Comparison of SERS spectra of milk, R6G, and milk spiked with R6G.

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Measurement of Platelet Clot Volume in a Microscale Thrombosis Screening Device

Laura SeamanBiological Engineering, Massachusetts Institute of Technology

NNIN REU Site: Nanotechnology Research Center, Georgia Institute of Technology, Atlanta, GANNIN REU Principal Investigator: Dr. Craig Forest, Mechanical Engineering, Georgia Institute of TechnologyNNIN REU Mentor: Melissa Li, Biomedical Engineering, Georgia Institute of TechnologyContact: [email protected], [email protected], [email protected]

Introduction:Cardiovascular disease (CVD) is the leading cause of death in the United States� In patients with CVD, plaque builds up in arteries creating local constrictions called stenoses� These stenoses increase shear rates within arteries causing platelets to form clots� This process, known as thrombosis, can result in the occlusion of the artery causing stroke or heart attack�

We have developed a microfluidic device to measure clot formation under restricted flow. Clot formation is monitored using simultaneous readings of laser light transmission through the stenosis region and mass flow. During each trial, the light transmission increases due to the increased scattering absorption of light in red blood cells compared to platelets� To calibrate this changing light transmission to a clot volume, an estimate of the final clot volume in the device was needed�

The volume was initially estimated to be around 30 nL� This presented a challenge because it was too small for weighing or displacement measurements� Additionally, the clot was too fragile to remove from the device� Previous methods have used only light microscopy or confocal microscopy� Light microscopy has excellent edge detection in single-plane images, but poor z-direction depth resolution [1]� Conversely, confocal microscopy has excellent z-direction resolution, but poor edge finding abilities [2]. Thus we sought to create a more accurate volumetric calculation through the combination of these methods�

Figure 1: A) Brightfield image of the top of the clot. B) Selected clot region.

Experimental Procedure:In our procedures, we first estimated the dimensions of the platelet clot using light microscopy� Next, we created estimates of density within the clot using confocal measurements at multiple planes�

First, the Keyence digital microscope was used as a 250x brightfield microscope to obtain single plane images of the top and bottom of the clot as shown in Figure 1� We were able to select for the clot areas using ImageJ color thresholding, and subsequently calculate the clot’s area� These measurements were then averaged and multiplied by the channel depth in order to obtain measurements of clot volume�

Next, we conducted confocal studies to obtain density measurements� To contrast between the clot and surrounding red blood cells, we used Mepacrine, a fluorescent platelet stain with excitation and emission at 488 and 520 nm respectively at a final concentration of 0.24 M. Images of the fluorescently stained clot were acquired using a Zeiss laser scanning confocal microscope with a 20x objective at 1 µm z-axis resolution through a 71 µm pinhole and a dwell time of 1�61 µs�

In confocal microscopy, the deeper into the tissue the images are taken, the less intense the image is� This decay causes a maximum depth penetration of 60 µm, preventing direct volume calculation� The decay also must be corrected before a threshold can be applied since the same intensity naturally selects a much larger area in the deeper images because they are naturally darker�

To correct the depth-related decrease in intensity, ImageJ was utilized to threshold the images using Shangabg’s algorithm� The threshold was applied to each picture and the number of bright, or clot, pixels were counted� The density was calculated by standardizing the areas of the clot in each picture by dividing by the area in the top picture� Then these standardized areas were averaged within each stack, resulting in a density. To verify the efficacy of the method, controls were created� They went through the process as the

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


real clots and the error between the known density and that measured by the process was calculated, as shown in Figure 2� The error of 18�5% indicates that the process failed to accurately select the clot regions�

Instead, a custom MATLAB® script was used to apply a noise reduction filter, wiener2, and adjust the images using imadjust, so that in each one 1% of the pixels were at the maximum and minimum value� It then calculated a single threshold and applied it to all the pictures in a stack�

Figure 3 shows the clot before any changes, after the intensity is adjusted, and after the threshold is applied� Controls were run then through the same script as the real clots� Results, shown in Figure 4, showed errors under 1�4%� The average density of the clots was calculated to be 76�86%�

As shown in Figure 4, these measurements were used to calculate three different volumes for comparison using the known height of the stenosis region. The first volume estimate used just the brightfield image of the bottom of the clot� The second version involved averaging the areas found from the top and bottom brightfield images. The averaged the top and bottom areas then scaled it by the density�

Results and Conclusions:The thrombus platelet density was 76�86%, which compares with measurements from a previous study of 80%� The overall platelet thrombus volume was estimated as 30 million mircometers [2], full results in Figure 4� This corresponds with the experimental recorded increase of 24% light transmission and the theoretical increase of 18�8% calculated in previous work [3]�

Acknowledgments:I would like to thank Dr� Craig Forest, Melissa Li, the Precision Biosystems Laboratory members, and the site coordinators for their support throughout the summer, as well as the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and the National Science Foundation for funding�

Figure 3: Images of a real clot before processing, after adjusting the intensity, and after thresholding.

Figure 2: A) Control images before and after thresholding. B) Results using ImageJ.

Figure 4: A) Clot density control results. B) Density of real clots. C) Clot volumes. D) Clot volume average.

References:[1] Ku, D�, Para, A�, et al; “Rapid Platelet

Accumulation Leading to Thrombotic Occlusion”; Annals of Biomedical Engineering, Vol� 39, 1961-1971 (2011)�

[2] Ruggeri, Z., Savage, B., et al; “Specific Synergy of Multiple Substrate-Receptor Interaction in Platelet Thrombus Formation under Flow”; Cell, Vol� 94, 657-666 (1998)�

[3] Li, M., Ackerman, J., et al; “Microfluidic System for Multichannel Optical Measurement of Shear Induced Platelet Thrombosis in Unfractionated Blood”; (2011)�

• b

iolo

gic

al

appli

cati

ons •

bio

logic

al

appli

cati

ons

• b

iolo

gic

al

appli

cati

ons •


Measuring the Thermodynamic Properties of Water at Negative Pressures in Synthetic Trees

Zachary SonnerMicroelectronic Engineering, Rochester Institute of Technology

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Abraham D. Stroock, Chemical and Biomolecular Engineering, Cornell UniversityNNIN REU Mentor: I-Tzu Chen, Chemical and Biomolecular Engineering, Cornell UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Liquids at negative pressures are observed in nature as a means for liquid to travel through the xylem of a plant or tree. The mechanism by which this occurs is similar to an osmotic process whereby liquid within the xylem is placed in tension (negative pressure) through diffusion of water through an external membrane in the presence of sub-saturated air or soil. The focus of this project was the fabrication of a microfluidic device in silicon capable of holding water at these metastable, negative pressures. Bulk silicon bonded to glass was used to form holding vessels for the liquid water with a semi-permeable porous silicon membrane to couple to the outside environment. When subjected to reduced relative humidity, water vapor diffused through the porous silicon membrane placing the contained water in tension. Quantitatively, the amount of negative pressure presented from a change in relative humidity across the membrane is given by Equation 1 in Figure 1 where R is the gas constant (8.3145 J/mol•K), T is the temperature (K), Vliq is the molar volume of the liquid (m3/mol) and av is the vapor activity, essentially the relative humidity divided by one hundred. In conjunction, the maximum amount of negative pressures able to be sustained across the membrane is described by the Young-Laplace equation as shown in Equation 2 from Figure 1 where γ is the surface tension of the water (J/m2), θ is the contact angle made with the pore wall (°) and rmax is the largest pore size (m).

Experimental Procedure:

To create these devices 4-inch, 300 µm thick, 1-10 Ω-cm p-type silicon wafers were used and the corresponding process flow is shown in Figure 2. The first step involved the generation of masking layers of furnace grown silicon dioxide (SiO2) and LPCVD silicon nitride (Si3N4) of 650 nm and 150 nm thicknesses, respectively� Contact photolithography was then used to generate two hundred 400 µm × 400 µm squares in two locations on the Si wafer� Using 33�3% KOH at 80°C as an Si etchant, the devices were then etched to form tetragonal pits of depth ~ 250 µm, and the nitride masking layer then removed� The backside of the wafer was coated with 350 nm of aluminum and then annealed as shown in Figure 2, Step 4�

Electrochemical (EC) porousification of the frontside then took place, which generated pores less than 10 nm in diameter for a depth of roughly 30 µm. The final step involved removal of remaining masking layers, silicon etching to open the pits to the porous silicon and anodic

bonding of a borosilicate wafer to the backside� The stability limit provided by the membranes was tested by filling the cavities with liquid water and allowing it to come to equilibrium with sub-saturated vapors�

Results and Conclusions:Figure 3 shows a scanning electron microscopy (SEM) image of the surface of the porous silicon membrane, noting all pores appeared to be less than 10 nm in diameter� This is an important factor within the membrane as this dictates the amount of negative pressure sustained as described by Equation 2 from Figure 1� Additionally, the development of these pores occurred at the propagation tip� This means that the pores will only grow to a certain size, which is dependent upon the silicon doping concentration and applied current density, and then become a self-limiting process fixing the diameters of the pores at a certain size�

Figure 1: Pressure (1), and Young-Laplace (2) equations [1].

• bio

logica

l applica

tions • b

iolo

gica

l applica

tions • b

iolo

gica

l applica

tions •


To simulate the capillary action as seen in nature, testing of the devices involved exposure to varying humidity levels� Figure 4 shows frames from a one-hundred minute timeline of one such device tested at 95% relative humidity; the backside of the device region with its many cavities is shown. Initially, all of the cavities were filled with water as explained by the ordered, lighter features highlighting the KOH cavities� When the holding vessels cavitated, the regions became dark as shown in Figure 4, where all vessels have cavitated� Cavitation at 95% relative humidity would suggest that the device withheld water just below -50 bars� However, in this short timescale, the devices did not equilibrate and were not stable at this humidity level�

Figure 4: Device testing at 95% relative humidity.

Figure 2: Process flow.

Figure 3: SEM image of pore size.

Upon further inspection, the porous silicon membrane had disintegrated after testing due to instabilities related to micro-cracks within the surface of the membrane, which were created during the anodic bonding process when the devices are exposed to elevated temperatures (400°C)� The fragility of the porous silicon to mechanical and thermal stresses evidently played a significant role in the quality of the membrane�

Future Work:In addition to the brittle behavior of the porous silicon membrane, there was one main issue that arose during the development and fabrication of these devices� This problem was with leaking of hydrofluoric acid (HF) during the EC etch� The leakage occurred at the peak of the anisotropic cavities due to a stronger electric field at this location. The HF that leaked, reached the backside of the wafer; removing masking layers for the later silicon etch� With these masking layers removed by the HF, the silicon etch then roughened the surrounding surface, reducing bonding quality� A method currently being pursued is creating a deeper KOH cavity and carrying out a long (4-5 hour) EC etch� Using this method, one would perform a KOH etch for roughly 280 µm in depth and then execute a longer EC etch, which would ensure that the porous silicon reaches every cavity; also eliminating the afterward silicon etch step�

Acknowledgements:National Science Foundation (NSF), National Nano-technology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, Cornell NanoScale and Technology Facility (CNF), Abraham D� Stroock, I-Tzu Chen, the Stroock Research Group, Rob Ilic, Melanie-Claire Mallison and all CNF staff�

References:[1] Stroock, Abraham D�; “The transpiration of water at negative

pressures in a synthetic tree”; Nature, 455, 208-12 (2008)�

• c

hem

istr

y •

chem

istr

y

• c

hem

istr

y •

chem

istr

y •

chem

istr

y •

chem

istr

y •


Defect Analysis of Molecular Monolayers with Electrochemistry

Clara ChowBiomedical Engineering and Chemistry, University of Wisconsin - Madison

NNIN iREU Site: Institut Für Bio- Und Nanosysteme (IBN), Forschungszentrum, Jülich, GermanyNNIN iREU Principal Investigator: Dr. Dirk Mayer, Molecular Bioelectronics, Peter Grünberg Institute 8, ForschungszentrumNNIN iREU Mentor: Nils Sanetra, Molecular Bioelectronics, Peter Grünberg Institute 8, Forschungszentrum JülichContact: [email protected], [email protected], [email protected]

Introduction:

Self-assembled monolayers (SAMs) can be formed on metal surfaces to modify the surface and control the properties of interfaces [1]� Although the electronic properties of alkanethiols have been extensively studied over the years and have been shown to form well ordered and closely packed SAMs [2], oligopeptides have not been as prevalent in SAM studies� Therefore, we electrically characterized oligo-peptides of varying lengths with drop electrode experiments to study the electron transfer through the monolayer in fluid gallium indium eutectic and gold electrode junctions�

The oligopeptides showed tunneling current-voltage (I-V) curves, but histograms displayed a broad range of current densities per oligopeptide, indicating that the molecular layer was not defect free� Even though oligopeptides may have better order in comparison to other molecules, their ordering is not sufficient for these junctions. This finding motivated further research into monolayer formation and defects�

Electrochemistry and cyclic voltammetry are powerful methods to explore the processes in metal electrode interfaces� The reduction and oxidation reactions that occur at the surface produce electrical currents, which provide information about the order and density of the SAM� If the monolayer were ordered and closely packed, redox molecules would be blocked by the monolayer and could not exchange electrons with the metal surface� Redox behavior can give insight to molecular defects and pinholes�

Experimental Procedure:Materials. The oligopeptides used in this study had the following sequences: Cys-(Ala)n-Ala and Cys-(Asp)n-Asp, where n ranged from 2-5 (Figure 1)� The peptides were

Figure 1: Structures of investigated oligopeptides.

purchased from Caslo Laboratory (Lyngby, Denmark) with purity greater than 95%� One millimolar (mM) peptide solutions in Milli-Q grade water were prepared and stored at 4°C� Lithium per chlorate (LiClO4), potassium hexa-cyanoferrate(II) (K4Fe(CN)6), sodium hydroxide (NaOH), and perchloric acid (HClO4) solutions were prepared with Milli-Q grade water and stored at room temperature�

SAM Preparation. 100 nm gold (Au) substrates were prepared via electron beam physical vapor deposition on a 400 nm silicin oxide (SiO2) insulating layer and a 10 nm titanium (Ti) adhesion layer on Si� The Au substrates

were cleaned with isopropanol and then O2 plasma for three minutes to remove organics on the surface� A 50 µl peptide solution was placed on top of the clean substrate for at least 72 hours. The modified substrates were finally rinsed with water and dried with nitrogen gas before testing�

Electrochemistry Measurements. A 3-electrode com-pression cell was used, with a modified Au substrate working electrode, standard calomel reference electrode (SCE), and platinum wire counter electrode� We then exposed 0�32 cm2 of the modified Au substrate to the electrolyte solution. Cyclic voltammograms started at 0�0 V, swept to -0�1 V, +0�45 V, and then back to 0�0 V� Scans were repeated twice before the third scan was recorded, at a scan rate of 25 mV/s� For each modified Au substrate, measurements were first performed with increasing electrolyte concentration (5, 25, 50, 75, and 100 mM LiClO4), then with 5 mM HClO4, and finally with 100 mM NaOH, all using 2.5 mM K4Fe(CN)6 as the redox molecule� The electrochemical cell system was rinsed with water three times and allowed to cycle with pure electrolyte before each recording�

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

•


Results and Discussion:In electrochemistry measurements, the Fe(CN)6 redox response of Au substrates coated with oligopeptide SAMs was investigated� At 5 mM LiClO4 electrolyte concen-tration, no redox behavior was observed� However, increasing the electrolyte concentration revealed increasing redox behavior (Figure 2)� This trend, seen with all peptide monolayers, is due to the decrease of the effective diameter of the redox molecules� As the electrolyte concentration increases, the redox molecules are less hydrated and have a smaller solvent cage surrounding them� Since the size of the pinholes and defects remain constant for each sample, more redox molecules can access the gold layer when their effective diameters are smaller� This effect can change the current density up to one order of magnitude between the lowest and highest concentrations of LiClO4�

The increase in current density with electrolyte concen-tration for the Cys-Ala-Ala-Ala SAM can be seen in Figure 3� Although not all peptide SAMs display the same threshold

shape curve, they all have a positive correlation between current density and electrolyte concentration�

Potential induced desorption peaks are another indication of monolayer defects� These cyclic voltammograms are recorded with 100 mM NaOH, and swept from 0�0 V, -1�3V, +0.6V, to 0.0 V. Only the first scan was recorded.

The Cys-Ala-Ala-Ala SAM desorption is shown in Figure 4� The sharpest peak, the desorption peak of the monolayer, occurs at -0�81V� If the monolayer contained no defects, this desorption peak would be the only peak between -1�3V and 0�0 V� However, a broader peak at -1�06V is also present, which indicates that there are domains that desorbs at more negative potentials� This behavior is seen with all peptide SAMs�

Conclusion:Defects of peptide SAMs on gold were investigated with electrochemistry� Pinholes were discovered by changing the electrolyte strength, thereby changing the effective diameter of the redox molecules� Redox molecules larger than the pinhole diameter were blocked by the monolayer and could not exchange electrons with the underlying gold substrate� The effects of gold substrate roughness, peptide length, and potential of hydrogen (pH) were also analyzed for the peptide monolayers (data not shown)�

Although these results give insight to monolayer defects using electrochemistry, further experimentation is necessary to fully understand whether these peptide monolayers can truly form defect-free and closely packed SAMs�

Acknowledgements:I would like to thank host Dr� Dirk Mayer, my mentor Nils Sanetra, Dr� Lynn Rathbun, Dr� Nancy Healy, the PGI 8 staff at FZJ, the National Nanotechnology Infrastructure Network International Research Experience for Undergraduates, and the National Science Foundation�

References:[1] A� Ulman; “Formation and Structure of Self Assembled

Monolayers”; Chem� Rev�, Vol� 96, p� 1533 1554 (1996)�[2] M� D� Porter, T� B� Bright, D� L� Allara, and C� Chidsey; “Structural

Characterization of nAlkyl Thiol Monolayers on Gold by Optical Ellipsometry, Infrared Spectroscopy, and Electrochemistry”; J� Am� Chem� Soc�, Vol� 109, p� 3559 3568 (1987)�

Figure 3, left: Cys-Ala-Ala-Ala SAM on gold, plotted at 0.35 V.

Figure 4, right: Desorption of Cys-Ala-Ala-Ala SAM on gold.

Figure 2: Cyclic voltammograms of Cys-Ala-Ala-Ala SAM on Au.

• c

hem

istr

y •

chem

istr

y

• c

hem

istr

y •

chem

istr

y •

chem

istr

y •

chem

istr

y •


Novel pH-Sensitive Dendrimer Nanoparticles for Targeted Imaging

Audrey DangChemical Engineering, Vanderbilt University

NNIN REU Site: Nano Research Facility, Washington University in St. Louis, St. Louis, MONNIN REU Principal Investigator: Professor Samuel Achilefu, Radiology, Washington University School of Medicine,

Washington University in St. Louis, St. Louis, MONNIN REU Mentor: Dr. Rui Tang, Radiology, Washington University School of Medicine,

Washington University in St. Louis, St. Louis, MOContact: [email protected], [email protected], [email protected]

Abstract:

The high detection sensitivity of optical contrast agents and the low autofluorescence of biological tissue in the near-infrared range enable noninvasive imaging of molecular conditions and processes. In particular, optical probes can be functionalized to target the molecular signatures of cancer cells in order to image tumors. Reducing background fluorescence would improve the contrast between normal and pathogenic tissue. We report progress on a novel dendrimer nanoparticle with potential of hydrogen (pH)-sensitive fluorescence activation in the acidic lysosomes of cancer cells for improved tumor imaging with reduced background fluorescence in normal tissue.

Introduction:Polyamidoamine (PAMAM) dendrimers are tree-like polymer nanoparticles with tunable pharmacokinetics and multifunctionality� Many recent studies report the design and synthesis of novel dendrimer carriers with imaging, targeting, and therapeutic moieties� It has been demonstrated that PAMAM dendrimers are delivered to acidic lysosomes after cell internalization [1]� Functionalization of the dendrimer with a pH-sensitive fluorescent dye yields an imaging probe

Figure 1: Structure of dye with pH-sensitive fluorescence (Absorbance maximum at 520 nm and 760 nm; Emission maximum of protonated fluorescent species at 780 nm).

that fluoresces after cell internalization and trafficking to the acidic lysosome (Figure 1)� If further functionalized with a tumor-targeting peptide, the nanoparticle would fluoresce only when delivered to and internalized by a cancer cell [2]� Thus, fluorescence background signal would be reduced in normal tissue resulting in greater contrast�

Methods:Synthesis of Dye-Dendrimer Conjugates. Generation 4 PAMAM dendrimers, which each possess 64 terminal amine groups, were reacted with 100 equivalents of acid-functionalized pH-sensitive fluorescent dye in the presence of activating reagents commonly used to couple amino acids. Specifically, hydroxybenzotriazole (HOBT) and O-benzotriazole-N,N,N’,N’-tetramethyl-uronium-hexa fluoro-phosphate (HBTU) were used to activate the dendrimer and dye in the presence of the organic base N,N-diisopropylethylamine in dimethyl sulfoxide� The reaction was carried out overnight with vigorous mixing� The crude product was purified by size exclusion chromatography using a PD-10 column� The same method was performed with a pH-insensitive dye to synthesize a control nanoparticle�

Results and Discussion:We have demonstrated pH-sensitive fluorescence on a dendrimer scaffold. The fluorescence of the pH-sensitive dye-dendrimer conjugate achieved greatest intensity in acidic buffer as previously observed with the free dye

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

•


(Figure 2)� The wavelengths at which the dye-dendrimer conjugate had maximum absorbance and emission were shifted compared to those of the free dye (Figure 3)� This red shift, or increase in wavelength, is due to changes in the electron system from conjugation to the dendrimer�

The large excess of dye was sufficient to saturate the reactive terminal amines of the dendrimer� Thus, no amine groups remained for conjugation to a targeting moiety� However, we found that using a smaller dye-to-dendrimer ratio resulted in an unexpected transformation of the spectral properties of the dye� A transformation was also observed when other organic dyes were subjected to the same reaction conditions� In particular, pH-insensitive dyes were transformed into pH-sensitive dyes (Figure 4)�

We found that the pH-sensitive dye transformation was dependent on the presence of excess amine groups� To avoid the dye transformation, the 100:1 dye-to-dendrimer ratio was used to decrease the amine-to-dye ratio� In addition, we made progress on synthesizing peptide-dye conjugates for coupling to modified dendrimers with greater control. Specifically, a portion of the amine groups of the dendrimer could be blocked with acetic anhydride to avoid the dye transformation by eliminating excess amine groups�

Conclusions and Future Work:Molecular probes that are pH-sensitive promise improved contrast and decreased background� A novel dendrimer nanoparticle with pH-sensitive fluorescence in the near-infrared range was synthesized and characterized� In vitro studies are in progress� In addition, an unexpected pH-sensitive dye transformation was observed and characterized�

Characterization of the products of the unexpected pH-sensitive dye transformation by nuclear magnetic resonance (NMR) and high resolution mass spectroscopy is ongoing� Future work includes functionalizing the nanoparticle with tumor-targeting peptides, particularly continuing to synthesize a peptide-dye conjugate for coupling to the dendrimer�

Acknowledgments:Many thanks to Professor Samuel Achilefu, Dr� Rui Tang, and the members of the Optical Radiology Lab at Washington University School of Medicine� I am also grateful to the staff of the Nano Research Facility especially Dee Stewart, Kate Nelson, and Nathan Reed� This research was generously supported by the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and the National Science Foundation�

References:[1] Albertazzi, L� et al� Mol� Pharmaceutics, Vol� 7 (3), 680–688 (2010)�[2] Le, H� et al� Bioconjugate Chem�, Vol� 22 (4), pp 777–784 (2011)�

Figure 4: Fluorescence emission of pH-insensitive control dye after unexpected pH-sensitive transformation.

Figure 2: Fluorescence emission of pH-sensitive dye-dendrimer conjugates (Maximum emission at 800 nm).

Figure 3: Absorption of pH-sensitive dye-dendrimer conjugates (Maximum absorbance at 510 nm and 780 nm).

• c

hem

istr

y •

chem

istr

y

• c

hem

istr

y •

chem

istr

y •

chem

istr

y •

chem

istr

y •


nanoliters to picoliters with high frequencies (10 Hz to 10 kHz) [1]� Droplets generated within the device act as well-mixed batch reactors where nanoparticle-forming reactions occur. High-frequency droplet generation enables sufficient statistics to determine which system variables affect the nanoparticle structure. The application of microfluidic devices eliminates many of the inherent deficiencies of traditional reactors. The characteristic length of microfluidic devices is on the micron scale which encourages mixing times on the order of 1-10 µs [2]� The surface area-to-volume ratio of the channels coupled with the small thermal mass of the reactants (in comparison to the entire device) leads to rapid thermal equilibration� A well-mixed system with spatially defined temperature precincts facilitates precise process control. These well defined heating zones can be integrated directly on the devices�

Experimental Procedures:The microfluidic device was constructed using photolithography coupled with replica molding by soft lithography� Transparency masks provided resolution of features as small as 10 µm, which was adequate for this application� SU-8-50 photoresist was spin-coated onto a

Time Resolve Study of Anisotropic Nanostructure Growth Using Integrated Droplet-Based Microfluidics and X-Ray Absorption Spectroscopy

Giovanni EstevesChemical Engineering, Arizona State University

NNIN REU Site: Penn State Nanofabrication Laboratory, The Pennsylvania State University, State College, PANNIN REU Principal Investigator: Robert Rioux, Department of Chemical Engineering, The Pennsylvania State UniversityNNIN REU Mentor: Nick Sturgis, Department of Chemical Engineering, The Pennsylvania State UniversityContact: [email protected], [email protected], [email protected]

Abstract:

The objective of this project was to develop an integrated droplet-based microfluidic device, and study the structural evolution of nanomaterials during their synthesis within droplets via x-ray absorption spectroscopy and high-energy x-ray scattering. Flow-focusing microfluidic devices enable the formation of droplets with volumes ranging from nanoliters to picoliters at high frequencies (10 Hz to 10 kHz). These droplets behave as well-mixed batch reactors in which nanoparticle-forming reactions can occur. The high-frequency of droplet generation allows for large sampling sizes to determine the influence of system variables on the final nanostructure product. Our prototype microfluidic device was constructed out of polydimethylsiloxane (PDMS) using contact lithography, soft lithography, and plasma bonding to fabricate a PDMS microfluidic device bonded to a glass slide. Our first-generation device was unstable when generating droplets due to pressure fluctuations in the channels caused by debris removal filters. A re-design of the device enabled consistent droplet formation at variable flow rates of the continuous- and reactant-phase fluids at frequencies over 100 hertz (Hz). Currently we are incorporating resistive heaters and thermocouples onto the glass portion of the device for on-chip heating and temperature measurement. For future applications, the integrated device will be constructed entirely out of glass and x-ray transparent material for in situ x-ray analysis.

Introduction:

The mechanism by which nano-materials assemble through processes, such as nucleation and growth, are presently un-define. Conditions for nanoparticle n u c l e a t i o n , c rys ta l l iza t ion , and growth normally require high temperatures and occur in a solution phase; thus they are difficult to study by in situ methods� This leads to the motivation of this project in which an integrated droplet-based microfluidic device compatible with x-ray analyses will serve as a platform to study the mechanism of nanoparticle formation�

The use of a flow-focusing junction (Figure 1) induces droplet formation ranging with volumes on the order of

Figure 1: Flow-focusing junction allows for continuous generation

of monodisperse droplets.

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

•


silicon <100> wafer followed by cross-linking under ultraviolet light exposure through the mask� The silicon substrate with the epoxy positive relief or “master” had channels with profilometry determined height of 69 ± 2 µm� A negative replica of the device was produced in PDMS by replica molding (Figure 2)� The negative replica was separated from the master, and bonded to a glass slide using oxygen plasma (Figure 3)� After exposure to oxygen plasma, the channels are hydrophilic due to hydroxyl groups formed on the PDMS and glass surfaces� This forms a covalent bond between glass and PDMS via condensation of hydroxyl groups� Since our continuous phase is perfluoromethyldecalin, we chemically modified the walls of the reactor with trichloroperfluorooctylsilane to ensure preferential wetting by the continuous phase� Small diameter polyethylene tubing was then inserted into the complete device to allow for independent syringe pump control of the continuous and reactant phases. By varying the relative flow rates of these two phases we can control the droplet size and generation frequency, allowing control over the nanoreactor characteristics�

Results:The use of our first generation device proved to be unsuccessful. The first generation device was created with integrated filters to capture any particles (i�e�, dust) that may have been pumped through the channels. These filters caused significant pressure fluctuations within the device which led to flow instabilities. These flow instabilities were significant enough that our microfluidic device was not capable of generating consistent droplets� As a result we re-designed the device, excluding filters which led to consistent droplet formation (Figure 4)� Droplet volumes ranged from 1�4-4�0 nL and frequency generation from 14-116 Hz� The size of the droplets are determined by the ratio of the carrier phase to the reactant phase. A high carrier phase flowrate yields smaller droplets with higher frequency while a higher reactant phase flowrate yields larger droplets with lower frequency.

Future Work:The final microfluidic device will be fabricated entirely out of glass with fully integrated on-chip heaters and resistance temperature detectors once a channel design is decided upon� Subsequently, metallic nanoparticle forming reactions will be studied in the newly constructed device to determine the effect reaction parameters have on size and shape� An all glass system will allow high temperature reactions and compatibility with in situ x-ray spectroscopy analysis�

Acknowledgements:Special thanks to my mentor Nick Sturgis, advisor Prof� Robert Rioux, site coordinator Kathy Gehoski, and the staff at Materials Research Institute at the Pennsylvania State University� The National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and the National Science Foundation are acknowledged for funding this experience�

References:[1] Joanicot, M� et al� Science 2005, 309, 887�[2] Solvas, X�C�I� et al� Anal� Chem� 2010, 82, 3950�[3] Figure 2, Courtesy of the Whitesides Group�

Figure 4: Consistent droplet formation in flow-focusing microfluidic device.

Figure 2: Droplet-generating microfluidic device fabricated by soft lithography [3].

Figure 3: Photograph of a completed PDMS-glass device. The channels are filled with a black dye in order to contrast them against the PDMS.

• c

hem

istr

y •

chem

istr

y

• c

hem

istr

y •

chem

istr

y •

chem

istr

y •

chem

istr

y •


Synthesis of Self-Assembling Silver Nanoparticles for Surface Enhanced Raman Spectroscopy

Jennifer GilbertsonChemistry, Beloit College

NNIN REU Site: Center for Nanotechnology, University of Washington, Seattle, WANNIN REU Principal Investigator: Danilo C. Pozzo, Chemical Engineering, University of WashingtonNNIN REU Mentor: Kjersta Larson-Smith, Chemical Engineering, University of Washington, Seattle WAContact: [email protected], [email protected], [email protected]

Introduction:Research in metallic nanoparticles has become a topic of great interest due to potential applications across a diverse range of fields including photonics, nano-electronics and catalysis� Silver nanoparticles in particular have a special interest due to their unique optical properties� There were several possible methods to synthesize silver nanoparticles� However, silver nanoparticles have a tendency to aggregate and to oxidize. Moreover, synthesis routes are difficult to control due to their strong dependence on reaction parameters� On the other hand, these particles show stronger plasmonic effects when compared to other metallic nanoparticles due to increased absorption and light scattering� The focus of this project was to create a standardized and reproducible synthesis route for monodisperse amphiphilic silver nanoparticles that would self-assemble into stable clusters with controllable geometry� Previous research done by Larson-Smith and Pozzo [1] demonstrated the clustering of gold nanoparticles with controlled geometries using specific dose concentrations of 10 kilodaltons (kDa) thiolated polyethylene glycol chains (PEG)� The steric interactions of the nanoparticles were controlled by different concentrations of polyethylene glycol chains� These silver nanoparticle clusters will be used in surface enhanced Raman spectroscopy (SERS) to enhance Raman scattering of specific analytes.

Methods:Monodisperse 20 nm silver nanoparticles were synthesized with sodium citrate as a stabilizing agent and sodium borohydride as a reducing agent in a protocol based off of the Turkevich and Lok methods [2, 3]� Silver nanoparticles

Figure 1: Schematic representation of silver nanoparticle clustering.

were then sterically stabilized by functionalizing with varying amount of 10 kDa thiolated PEG ranging in dose concentration from 0�01 to 20 PEG/nm2 Ag� The silver nanoparticles were then treated with octane-thiol, rendering them amphiphilic� This initiated particle self-assembly due to hydrophobic attraction� The initial PEG concentration controlled the cluster size and geometry� Ultraviolet-visible spectroscopy was performed using a Thermo Scientific Evolution 300 systemin the range of 200-800 nm� Dynamic light scattering (DLS) was carried out using a Malvern Zetasizer Nano ZS with a 633 nm wavelength� Clustered silver nanoparticles were coated on a copper transmission electron microscope (TEM) grid and looked at under a FEI Technai G2 F20 TEM� An Anton Paar SAX Sess small angle x-ray scattering (SAXS) instrument with a wavelength of 1�54Å was used to examine the structures of the silver clusters in water�

Results:Bare 20 nm silver nanoparticles had an absorption peak at 396 nm due to the surface plasmon resonance� The peak was shifted for different sized particles and clustered particles� The width and location of the absorption peak increased once the silver nanoparticles were treated with

Figure 2: UV-Vis absorption spectra of bare, unclustered, and octane-thiol clustered silvernanoparticles in water.

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

•


PEG and then again, once they were clustered� Samples with a higher concentration of PEG chains/nm2 Ag turned from yellow to red� Once the particles were clustered, the lower concentrations of PEG-treated samples aggregated into larger clusters� Figure 2 demonstrates that the larger the size of the clusters, the wider the absorbance peak reading on the UV-Vis� As the concentration of PEG chains/nm2 Ag reached saturation, the silver nanoparticles remained as singlets�

Conclusions:Our synthesis route reproducibly produced amphiphilic 20 nm spherical silver nanoparticles� While some polydispersity exists, the silver nanoparticles were still able to cluster� While we recognize that the trend of cluster size to PEG concentration was not perfect with the silver nanoparticles, our research with silver nanoparticles has demonstrated the application of this controlled assembly onto other metallic nanoparticles�

The controllable geometry of these nanoparticle clusters has important future applications for selective SERS detection based on steric effects or hydrophobicity� The center of the silver clusters will be able to selectively separate analytes based on size and chemical functionality�

Acknowledgements:I would like to thank Dr� Danilo Pozzo and Kjersta Larson-Smith for their guidance and patience� I would like to acknowledge the National Science Foundation and the NNIN Research Experience for Undergraduates Program for funding this research� I would also like to thank the NanoTech User Facility at the University of Washington for the use of their equipment�

References:[1] Larson-Smith, K� and Pozzo, D�C� Soft Matter� 7, 5339 (2011)�[2] Turkevich, J� et al� Discuss� Faraday Soc� 11, 55� (1951)�[3] Lok, C� et al� J Biol Inorg Chem� 12, 527� (2007)�

Figure 4: Slit-smeared SAXS scattering data of clustered silver at varying PEG concentrations.The TEM images (a-c) show several of the clustering geometries.

Figure 3: Hydrodynamic radii for octane-thiol clustered and unclustered silver nanoparticles inwater.

DLS was used to determine the hydrodynamic radius (RH) of the entire silver-PEG complex� The hydrodynamic size increased with PEG concentration as the chains were more tightly packed� Figure 3 shows that the hydrodynamic size of octane-thiol functionalized clusters increased at low PEG concentration because more particles were in each cluster�

TEM was used to confirm the radii and size distribution of the bare silver particles and to examine the clustering of the PEG and octane-thiol treated samples� Figure 4 a-c shows examples of the nanoparticles clustering� The TEM images demonstrate that although we had achieved control over cluster geometry, there was still some polydispersity�

SAXS was used to analyze the radius and size distribution of the bare silver nanoparticles� It was also used to analyze the clustering of the PEG and octane-thiol treated silver nanoparticles� PEG has a low scattering length density compared to silver, therefore the scattering curve was primarily from the silver nanoparticles� Thus, samples that did not cluster scatter similar to bare silver particles�

The scattering plots demonstrated that the scattering was changing as a function of PEG concentration� The size of the silver nanoparticle clusters decreased as the concentration of PEG chains/nm2 Ag increased and approaches bare silver� This data illustrated that we can control the steric interactions of the silver nanoparticles on the nanoscale�

• c

hem

istr

y •

chem

istr

y

• c

hem

istr

y •

chem

istr

y •

chem

istr

y •

chem

istr

y •


Measuring Height Mismatch and Miscibility Temperatures of Model Cell Membranes

Morgan McGuinnessPhysics, Lafayette College

NNIN REU Site: Center for Nanotechnology, University of Washington, Seattle, WANNIN REU Principal Investigator: Sarah Keller, Chemistry, University of WashingtonNNIN REU Mentor: Joan Bleecker, Chemistry, University of WashingtonContact: [email protected], [email protected], [email protected]

Introduction:

Hundreds of different lipids exist in the cell membrane, and little is understood regarding their physical and organizational properties [1]� Model lipid bilayers serve as a simplified means of studying many of the inter-molecular interactions that occur within cell membranes� For example, interactions between two or more lipids and cholesterol can result in demixing of lipid bilayers into regions with different lipid compositions� This membrane phase separation can be observed in model lipid bilayers composed of a high-melting temperature (Tm) lipid, a low-Tm lipid, and cholesterol� At temperatures above the phase separation temperature (miscibility temperature) of a composition, lipids mix together into one uniform liquid phase� At lower temperatures, two distinct phases coexist: a liquid ordered (Lo) and liquid disordered (Ld) phase� The Lo phase is primarily composed of high Tm lipids, whereas the Ld phase is composed mainly of low Tm lipids� After phase separation occurs, like phases coalesce, forming larger single-phase domains [2]�

Currently, the factors determining the miscibility temperature of a lipid composition are unknown� Recent research proposes that miscibility temperature increases with the height mismatch of the Lo and Ld phases [3, 4, 5]� When the membrane separates into macroscopic phases of different heights, the total area of the hydrophobic lipid

Figure 1: Schematic of height mismatch; the top line represents the surface exposed to water, thecircles are the lipid head groups, and the straight and bent lines are the lipid tails.

tail group exposed to the solvent (water) is thought to be reduced, making this configuration energetically favorable [3] (Figure 1)�

The goal of our research was to elucidate the contribution of height mismatch vs� other lipid structural parameters in determining miscibility temperature� The purpose of this project was to discover whether two bilayers with similar miscibility temperatures could have different height mismatches between the demixed phases� In order to accomplish this, we had to perfect our methods of measuring miscibility temperatures and height mismatch in lipid bilayers� In this study, miscibility temperatures were determined by observing phase separation of giant unilamellar vesicles (GUVs) using fluorescence micro-scopy, and height mismatch was measured by atomic force microscopy (AFM) of supported lipid bilayers (Figure 2)�

Methods:Giant Unilamellar Vesicles (GUVs). Dipalmitoyl-sn-glycero-3-phosphotidylcholine (DPPC, 16:0) and dioleoyl-sn-glycero-3-phosphotidylcholine (DOPC, 18:1) were purchased from Avanti Polar Lipids, Inc� Cholesterol

Figure 2: The top part of the figure shows a micron-scale, phase separated vesicle. The bottompicture shows a supported bilayer.

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

•


was purchased from Sigma, and Texas Red DHPE, a fluorescently labeled lipid, was purchased from Invitrogen®� All were purchased dissolved in chloroform�

Solutions of DOPC, DPPC, cholesterol (2:2:1), and 0�8% Texas Red were vortexed, spread on the conductive surface of indium tin oxide coated glass slides, and vacuum dried for 1/2 h� GUVs were formed by electroformation [6] at 60°C, 1�5 V, and10 Hz for 1 h� Vesicles were imaged using fluorescence microscopy. The transition temperature was determined to be the temperature at which 50% of vesicles showed phase separation�

Supported Bilayers. Solutions of DOPC, DPPC, and cholesterol (2:2:1) were dried under nitrogen and vortexed with purified water, making the final concentration of lipids 1 mM. In cases where phase separation was confirmed by fluorescence microscopy, the lipid solution contained 0�8% Texas Red DHPE� The solution was bath-sonicated for 1 h to form small vesicles and centrifuged for 1/2 h� 20 µL of solution was added to 180 µL of 5 mM CaCl2 and immediately deposited as a thick droplet on a 4�91 cm2 freshly cleaved mica substrate� An additional 5 ml of CaCl2 was added for 1/2 h to allow a bilayer to form on the mica�

The preceding steps were carried out above the transition temperature of the composition (~34°C)� The mica-supported bilayer was thoroughly rinsed with purified water� Supported bilayers containing dye were imaged by fluorescence microscopy. Supported bilayers without dye were imaged by AFM in constant-amplitude tapping mode in a liquid cell� The liquid cell kept bilayers submerged in purified water. We used a SNL-10 tip (spring constant: 0�32 N/m), a scan area of 100 µm2, and a scan frequency of 1 Hz� The amplitude set point was varied to reduce artifacts�

Results:Using fluorescence microscopy, we determined the misc-ibility temperature of a vesicle containing 2:2:1 DOPC, DPPC, and cholesterol to be 34�6 ± 0�6°C (Figure 3)� The uncertainty represents the full range of miscibility temperatures� Additionally, we observed phase separation in supported bilayers with fluorescence microscopy. Using AFM, we observed nanometer-scale height mismatch in supported lipid bilayers�

Discussion:In our research we observed phase separation on both supported bilayers and GUVs� Because the miscibility temperature we measured for DOPC, DPPC, and cholesterol (2:2:1) agrees with previous measurements by our lab [7], we have confidence in our lipid composition. In the future, our lab will study compositions of lipids with greater height mismatches and similar miscibility temperatures to continue this project�

Acknowledgements:This research was supported by the National Nanotechnology Infrastructure Network Research Experience for Under-graduates (NNIN REU) Program and the National Science Foundation (NSF)� I would also like to thank the Keller Lab for their valuable guidance and assistance, and the NanoTech User Facility at the University of Washington for providing the equipment and training necessary for this project�

References:[1] Mouritsen, O� Life- As a Matter of Fat� Springer, 2005�[2] Veatch, S�L�, Keller, S�L� Biophys� J� 2006, 90, 4428-4436�[3] Garcia-Saez, A�J�, Chiantia, S�, Schwille, P� J Biol�Chem� 2007, 282,

33537-33544�[4] Perlmutter, J�D�, Sachs, J�N� J� Am� Chem� Soc� 2011, 133, 6563-

6577�[5] Kuzmin, P�I�, Akimov, S�A�, Chizmadhev, Y�A�, Zimmerberg, C�,

Frederic, S� Biophys� J� 2005, 88, 1120-1133�[6] Angelova, M� I�, Soleu, S�, Meleand, P�, Faucon, J�F�, Bothorel, P�

Prog� Colloid� Polym� Sci� 1992, 89, 127-131�[7] Veatch, S�L�, Keller, S�L� Biophys� J� 2003, 85, 3074-3083�

Figure 4: AFM image; the black is the mica substrate, and the two lighter shades show a height difference in the sample.

Figure 3: These pictures show the same GUV at different temperatures. Its miscibility temperature is -34°C.

• c

hem

istr

y •

chem

istr

y

• c

hem

istr

y •

chem

istr

y •

chem

istr

y •

chem

istr

y •


Film Making in Digital 3D: Selective Area Atomic Layer Deposition

Jade M. NobleChemical Physics, Columbia University

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Professor James Engstrom, Department of Chemical Engineering, Cornell UniversityNNIN REU Mentors: Wenyu Zhang and Rambert Nahm, Department of Chemical Engineering, Cornell UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Atomic layer deposition (ALD) is commonly used to deposit thin films in integrated circuits and other micro- and nano-devices. However, these films still require patterning via photolithography. Selective area ALD can be achieved on substrates with areas of differing surface chemistries to grow thin films on the desired surface but not the other. This can overcome the limitations of traditional ALD and eliminate the need for patterning steps directly proceeding thin film deposition by ALD. In this experiment, tantalum nitride (TaN) was deposited on both silicon dioxide (SiO2) and copper (Cu) using ALD to determine if selective growth could be achieved. Specifically, growth of TaN was desired on SiO2 but not on Cu. The resulting TaN thin films were analyzed using spectroscopic ellipsometry, atomic force microscopy (AFM), and x-ray photoelectron spectroscopy (XPS). A growth rate of 0.431Å/cycle was determined for the deposition of TaN on SiO2. An analysis of the roughness of the Cu samples after various cycles of TaN deposition suggests that the roughness of the thin film increases with increasing process temperature. Although at 255 and 300°C there was a significant amount of TaN deposited on the surface of the copper samples after 10 cycles, at 225°C there was very little deposition, suggesting the possibility for selective area ALD for less than 10 cycles.

Introduction:When making electronic devices, thin films can be created with precise control over thickness using ALD� Although the resulting thin films are highly uniform and conformal, if they could be selectively deposited according to an inherent pattern on the substrate, additional patterning steps, used in making devices such as transistors, may not be necessary� In integrated circuits, copper (Cu) is commonly used as the interconnect material between transistors� However, Cu diffuses easily into SiO2 and other dielectric materials that surround the Cu� In Cu metallization schemes, TaN is deposited onto the dielectric before the Cu interconnects are placed to prevent this diffusion�

If possible, it would be ideal for TaN to grow on the dielectric but not on the Cu so as not to increase the electrical resistance of the Cu line� Selectivity may be achieved by manipulating the differences in the chemistries (such as the number of nucleation sites) of two different materials� This may allow for growth of a thin film on one material and not on the other�

Experimental Procedures:The goal for this experiment was to examine growth on two materials (SiO2 and Cu) and thus determine if selective area ALD is possible for these two materials�

Deposition on Chemical Oxide. Bare Si samples were treated with BOE for total of four minutes and Nanostrip for 30 minutes to produce a 3 ± 0�1287Å thick SiO2 layer, which will be called chemical dioxide� TaN was deposited onto chemical oxide at 255°C via ALD, varying the ALD cycle count, using the Oxford ALD FlexAL tool in Cornell NanoScale Science and Technology Facility (CNF)� The thin film thicknesses were then analyzed using a Woollam spectroscopic ellipsometer�

Deposition on Copper. TaN was deposited onto as-received Cu thin films with 0, 10, 20, 30, 40, and 50 cycles of deposited TaN at 225, 255, and 300°C� The surface roughness of the Cu samples was determined using AFM in tapping mode� The surface composition was analyzed using XPS with a fixed photoelectron takeoff angle of 38.5°.

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

•


Results:Deposition on Chemical Oxide. Using spectroscopic ellipsometry, a TaN growth rate of 0�431Å/cycle was determined� This value corresponded to a TaN growth rate of 0�41-0�46Å/cycle found in the ALD tool data sheets at CNF�

Deposition on Cu Samples. AFM showed that after 10 cycles there were virtually no change in roughness for all temperatures examined� However, there was a noticeable decrease in roughness between 0 cycles and 10 cycles� This suggests that an interesting phenomenon occurs between 0 and 10 cycles� Furthermore, there was an increase in the roughness of the thin film with

Figure 4: Integrated intensity of the Ta(4d) XPS feature vs. number of ALD cycles.

Figure 1, left: TaN film thickness with varying ALD cycle counts on chemical oxide at 255°C.

Figure 2, right: RMS roughness of TaN thin films on Cu vs. number of TaN ALD cycles.

increasing temperature� Further experimentation would be necessary to propose a cause for these observations�

XPS revealed an increased amount of deposited TaN on the surface of the copper samples with increasing temperature� At 50 cycles, the Ta 4d peak of the film deposited at 225°C was smaller than at 255 and at 300°C� Figure 4 compares the integrated intensities of the Ta (4d) feature from XPS at different cycle counts and temperatures and shows that there is comparatively very little deposition of TaN on Cu at 225°C� Therefore, it may be possible to achieve selective area ALD of TaN on SiO2 and Cu at 225°C at cycle counts less than 10� However, this selectivity does not occur for a cycle count greater than 10 at any of the examined temperatures�

Conclusions:Thin film deposition on SiO2 occurred after 10 cycles at 255°C� For TaN deposition on Cu, AFM suggests that the roughness of the thin film increases with increasing temperature� At 255 and 300°C, there was a substantial amount of TaN on the Cu surface after 10 cycles� At 225°C there was very little deposition� This may allow for selective ALD for less than 10 cycles� In the future, the growth rate of TaN on Cu at lower cycle counts on both SiO2 and Cu can be analyzed to determine the lack or presence of such selectivity�

Acknowledgements:I would like to thank the National Science Foundation, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, the Cornell NanoScale Science and Technology Facility, Professor James R� Engstrom, Wenyu Zhang, Rambert Nahm, CNF staff, Rob Ilic, and Melanie-Claire Mallison�

Figure 3: AFM images of surface roughness of Cu samples after 0, 10, and 50 cycles (left-right) of TaN deposition at 300°C.

• c

hem

istr

y •

chem

istr

y

• c

hem

istr

y •

chem

istr

y •

chem

istr

y •

chem

istr

y •


Synthesis of Silicon and Germanium Nanowires

Yoichi Ogata School of Materials Science, Japan Advanced Institute of Science and Technology

NNIN iREG Site: Microelectronics Research Center, The University of Texas, Austin, TX NNIN iREG Principal Investigator: Prof. Brian A. Korgel, Chemical Engineering, The University of Texas at Austin NNIN iREG Mentor: Vincent C. Holmberg, Chemical Engineering, The University of Texas at AustinContact: [email protected], [email protected]

Abstract:

Crystalline silicon (Si) and germanium (Ge) nanowires were grown in toluene heated and pressurized above its critical point. Colloidal gold nanocrystals were used to seed nanowire growth by the supercritical fluid-liquid-solid (SFLS) mechanism with monophenylsilane (MPS) and diphenylgermane (DPG) as reactants. Using a growth temperature of 490°C, the simultaneous addition of MPS and DPG to the reactor led to the formation of distinct Si and Ge nanowires, as opposed to a Si(1- x)Gex alloy. The nanowire product was characterized using a range of methods, including x-ray diffraction (XRD), transmission electron microscopy (TEM), scanning electron microscopy (SEM) and nanobeam energy dispersive spectroscopy (EDS) mapping. The optical properties of the nanowires, including the reflectivity and absorptance, were also measured using an integrating sphere.

Experimental Procedure: The SFLS synthesis procedures for silicon (Si) and germanium (Ge) nanowire growth are very similar� In an attempt to synthesize Si1-xGex alloy nanowires, the reaction was run with the simultaneous thermal decomposition of monophenylsilane (MPS) and diphenylgermane (DPG) precursors� Colloidal Au nanocrystals were used to seed nanowire growth� In an example synthesis, a solution composed of Au nanocrystals and silicon and germanium precursors is injected into pressurized toluene at 490°C and 10 MPa at a flow rate of 0.5 mL/min for 40 min. For the example shown here, the atomic ratio of silicon to germanium precursor was adjusted to 3:1� After synthesis, the nanowires were characterized using XRD, SEM, and TEM�

In addition, we fabricated composite Si/Ge nanowire fabric by combining separate batches of Si and Ge nanowires and measured their optical properties using an integrating sphere� The fabric was made by mixing Si nanowires (75 vol%) with Ge nanowires (25 vol%), drop casting the mixture onto a Teflon® substrate, and removing the resulting film. We then measured the optical absorptance and reflectance of the material.

Figure 1: Reitveld analysis of XRD pattern for 25% Ge injection sample.

Figure 2: (a) TEM images of Ge nanowires coated with poly(phenylsilane); (b) with Si nanoparticles; and (c) an SEM image of the nanowires.

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

•


Results and Conclusions: Figure 1 displays the XRD pattern of the 25% Ge injection sample� By analyzing this XRD pattern, the lattice constant was determined, consistent with that of a pure Ge single crystal� Also, Figure 2 shows SEM and TEM images of the injection sample� According to the SEM image, the maximum diameter of crystalline Si and Ge nanowires was around 1 µm� Furthermore, in the TEM image, we can see a Ge nanowire coated with a poly(phenylsilane) shell and Si nanoparticles, as opposed to a Si(1-x)Gex alloy� These compounds were confirmed by FTIR and EDS spectra.

For the optical properties of the composite Si/Ge nanowire fabric, we used a UV-Vis-NIR spectrophotometer in conjunction with an integrating sphere (see Figure 3)� We measured absorptance and reflectance of the material as a function of wavelength as seen in Figure 4� First, compared with pure Si and Ge nanowire fabric, the color of the composite sample is quite different� The absorption increases sharply above the optical gap for Si (1�1 eV) and Ge (0�7 eV), respectively� In addition, peaks in the Si spectrum from 2000 nm to 2500 nm are due to absorption of the poly(phenylsilane) shell, and the peak at 600 nm in the Ge spectrum is related to light trapping�

Figure 4: Absorptance and reflectance of composite Si/Ge nanowire fabric

Figure 3: Equipment for measuring optical properties.

However, for both of spectra, we can say that only a small (25 vol%) addition of Ge nanowires to Si nanowire fabric results in optical properties very similar to pure Ge nanowire fabric�

Future Work: Synthesis of the Si(1-x)Gex alloy nanowire requires further investigation of optical properties for composite Si/Ge nanowire fabrics�

Acknowledgments: National Nanotechnology Infrastructure Network International Research Experience for Graduates (NNIN iREG) Program, National Institute for Materials Science (NIMS), and the National Science Foundation (NSF) are acknowledged for supporting and funding this research�

• c

hem

istr

y •

chem

istr

y

• c

hem

istr

y •

chem

istr

y •

chem

istr

y •

chem

istr

y •


Stability of Zwitterionic-Modified Gold Nanoparticles in Complex Media: Effect of Surface Packing Densities

Julia PodmayerBiochemistry, Seattle Pacific University

NNIN REU Site: Center for Nanotechnology, University of Washington, Seattle, WANNIN REU Principal Investigator: Shaoyi Jiang, Chemical Engineering, University of WashingtonNNIN REU Mentor: Wei Yang, Chemical Engineering, University of WashingtonContact: [email protected], [email protected], [email protected]

Introduction:

Nanoparticles have shown potential to be tools for advancement in the medical world, such as in drug delivery, diagnosis and other areas� Despite their promise, one of the largest obstacles to these applications is nonspecific protein adsorption, which can result in cellular uptake, nanoparticle aggregation, immune system response, and other disastrous problems for in vivo applications� The current solution for combating this protein adsorption is to coat the nanoparticles in a non-fouling poly(ethylene glycol) (PEG) or oligo(ethylene glycol) (OEG) shell� Recent studies have shown zwitterionic materials such as poly(carboxybetaine) (pCB) have ultra-low fouling properties� To achieve high surface resistance to nonspecific protein adsorption, the surface packing density and film thickness are important.

In this study, an atom transfer radical polymerization (ATRP) reaction was employed to coat nanoparticles with pCB� Different sized particles and surface packing densities were obtained by adjusting the ratio of reactants and the reaction time in the ATRP process�

Experimental Procedure:Synthesis of Bare and PEG-Coated Gold Nanoparticles (GNPs). Bare GNPs and PEG-5000-SH-coated GNPs were prepared by using a previous method [1]�

Synthesis of Initiator-Modified GNPs. Initiator-modified GNPs were synthesized according to literature [2]� After reaction, the system was dried with a Rotary evaporator and washed with ethanol to precipitate the GNPs� The precipitate was collected and re-dispersed in acetone�

ATRP Reaction. In a typical reaction, 200 mg CBMA, 61�707 mg 2,2 bipyridiyl, 28�533 mg copper(I) bromide, and 4�4 mg copper(II) bromide were placed in a reaction tube� The tube was thoroughly purged by vacuum and flushed with nitrogen before N2-purged methanol was added to the flask via

syringe� The reaction mixture was homogenized by agitation using a vortex mixer and ultrasound for 5 min and 1 min, respectively� The deoxygenated initiator-coated GNPs (0�4 mL) in toluene were mixed with the above solution under N2 protection. The final mixture was stirred (13, 200 rpm) at room temperature� In this reaction 1:3, 1:1, and 2:1 volume ratio of methanol to acetone was applied�

Analysis. After polymerization, the pCB-GNPs were washed through centrifuging and re-dissolved in water� Using diffraction light scattering (DLS) and a syringe filter the size of the nanoparticles in solution was measured� Confocal laser scanning microscopy (CLSM) was further applied to gain some insight into the interaction between cells and nanoparticles� In this experiment, COS-7 cells were exposed to different ATRP product ratios and their interactions observed under microscope�

Figure 1: Scheme of the preparation of initiator-coated GNPs and pCB-coated GNPs.

Table 1: Hydrodynamic sizes of GNPs in different media.

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

• chem

istry

•


Results:

Stability of Polymer-Coated GNPs in Complex Media. As observed from Table 1, the hydrodynamic size of PEG-GNPs in water was 53�4 nm� However, after the addition of lysozyme solution, nanoparticles showed an increase of 20 nm in size compared to those in water� In the case of pCB-GNPs, their diameters did not change, indicating the high in vitro stability of pCB-GNPs in high ionic strength or in the presence of proteins under physiological conditions� Results also show that pCB-GNPs with different packing densities exhibit different stability when exposed to complex media and those with 1:3 ratio exhibits the best performance�

Cell Morphology. Cell density and morphology was observed by microscopy� In the bare-GNP case there were few to no living cells on the surface, whereas all other samples did not inhibit growth� PEG-GNPs and pCB-GNPs (1:1) exhibited similar morphology, widely dispersed and fewer cells� PCB-GNPs (1:3) and pCB-GNPs (2:1) allowed greater cell growth, and the highest density of cells was observed in the pCB-GNPs (2:1) ratio�

Conclusions and Future Work:Conclusions. In this work, we investigated the stability and cell interactions of pCB-coated GNPs coated with different surface packing densities� It was found that the surface resistance to nonspecific protein adsorption highly depends on the surface packing density� We also observed GNPs made with different ATRP ratios caused dissimilar cell

morphology in their respective cultures� In addition, when mixed with cells, the bare GNPs proved to be toxic�

Future Work. We will continue to focus on the ratios that have shown the most promising experimental results� This focus will include repeating experiments and working to understand why a particular ratio works, or does not work� Future research into the best performing ratios will yield further insights as to the effects of SPDs on coated GNPs� Ideally, results will help prove that pCB is a more effective and longer circulating nonfouling coating�

Acknowledgements:I would like to pay special thanks to my graduate mentor Wei Yang and my principal investigator Dr� Shaoyi Jiang for all of their support during this project� This work would not be possible without them, the support of the National Science Foundation and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program�

References:[1] Bergen, J�, H� von Recum, T� Goodman, A� Massey, and S� Pun�

Macromolecular Bioscience� (2006)�[2] Dong, H� Zhu, M� Yoon, J� A� Gao, H� Jin, R� Matyjaszewski, K� J�

Am� Chem� Soc� Vol� 130, Pg� 12852-12853 (2008)�

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Inkjet Printing of Zinc Oxide Based-Semiconductors for Thin Film Transistors

Carlos Koladele BiaouElectrical Engineering, Prince George’s Community College

NNIN REU Site: Lurie Nanofabrication Facility, University of Michigan, Ann Arbor, MINNIN REU Principal Investigator: Dr. Rebecca L. Peterson, Electrical Engineering and Computer Science, University of MichiganNNIN REU Mentor: Wenbing Hu, Department of Electrical Engineering and Computer Science, University of Michigan, Ann ArborContact: [email protected], [email protected], [email protected]

Abstract:

Ink-jet printing is widely used in arts and our daily lives. It consists of creating digital images by propelling droplets of ink onto a substrate. This technique can also be used to produce thin film electronics at a fairly low cost. The objectives of this project were to develop ink-jet printed thin film transistors from zinc oxide-based solutions, and explore their morphological and electrical properties. Zinc oxide precursors were directly printed onto a silicon/silicon dioxide (Si/SiO2) substrate by controlling the substrate temperature and drop spacing. These variables had a strong effect on the film thickness and its electrical performance. After printing, the samples were annealed to evaporate the solvent and react the chemical precursors to form the semiconductor layer. Transistor I-V testing was done by landing source and drain probes directly onto the printed structure and using the Si substrate as a bottom gate. The devices showed field-effect electron mobility between 0.001 and 0.024 cm2V-1s-1. Further work is needed to optimize the processing so that zinc oxide-based transistors can be applied in roll-to-roll processing of circuits with moderately good performance for large-area displays.

Introduction:

The transistor became the heart of modern electronics after its invention in the 1940’s thanks to its unique property of generating a greater output current than the one inputed, as well as its low production cost, wide range of applications, and reliability� Transistors need a semiconductor layer to perform the critical switching or amplifying function� Zinc oxide (ZnO), an inorganic material, has shown promising semiconductor properties, and is non-toxic and inexpensive [1]. Therefore, making thin film transistors (TFTs) out of ZnO is an important research topic� Ink-jet printing is a very efficient solution-based process in that the active layer of the transistor is only formed in the channel region, and no additional process is needed for patterning [2]� Consequently, material wastage, production costs and possibly fabrication time are all reduced� Here, we report our investigations on methods of optimizing the electrical performance of ink-jet printed ZnO transistors�

Experimental Procedures:A ZnO solution was made of zinc acetate dihydrate (ZnAc2•2H2O) mixed with 2-methoxyethanol to form a 0�5 M solution� Ethanolamine was added as a stabilizing agent in the solution� The TFT bottom gate was made out a doped silicon wafer of thickness 550 µm with a 200 nm thick thermally grown SiO2 layer� The Si/SiO2 substrate was cleaned with acetone and isopropanol and treated with an oxygen plasma reactive ion etch with a Technics West

PEII-A Plasma System� The ZnO precursor was inkjet printed with a Fujifilm Dimatix-2800 printer onto the prepared substrates that were maintained at temperatures of 50ºC or 60ºC with drop spacings of 5 µm, 10 µm or 15 µm� The printed liquid was then pre-baked for 10 min at 100ºC before being annealed for one hour at 480ºC� Morphological characterization was done with an Olympus LEXT OLS4000 interferometer, and three-point probe measurements were performed to electrically characterize the devices, shown in Figure 1�

Figure 1: Cross-section of a ZnO ink-jet printed transistor.

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


Results and Discussion:Increasing the drop spacing strongly in-fluenced the morph-ological and electrical characteristics of the printed shapes as shown in Figure 2� At a drop spacing of 5 µm, the droplets overlapped form ing a very thick film (819 nm), which exhibited no transistor-like electrical properties� At a drop spacing of 10 µm, the film thickness was 563 nm and transistors with channel lengths and widths of 704 µm and 555 µm, respectively, exhibited electron mobility of 0�004 cm2V-1s-1 and an on-to-off current ratio of about 40� Figure 3 shows the output characteristics,

Figure 3, left: A transistor made with 10 µm drop spacing exhibits only the Ohmic region.

Figure 4, right: Optimum printing conditions yield maximum drain current of about 0.45 µA.

Figure 2: ZnO films printed at (a) 5, (b) 10, and (c) 15 µm. Notice the decrease of solvent (dark regions) from (a) to (c).

which show their Ohmic properties� At a drop spacing of 15 µm, the print thickness was 133 nm� They were better transistors with a channel length and width of 602 µm and 442 µm, an electron mobility of 0�024 cm2V-1s-11, and on-to-off ratios of 1�1 × 103, with a clear contrast between the Ohmic and saturation regions� Therefore, increasing the drop spacing decreases the film thickness, which improves the electrical performance of the films.

Substrate temperature is also an important variable� ZnO films printed at a substrate temperature of 60ºC showed higher electron mobility than those printed at 50ºC� Since the flash point of the solvent used was 40ºC, maintaining the substrate at a higher temperature during the print accelerated solvent evaporation and, thereby, improved the electrical properties of the films. The drain current at a given voltage almost doubles when substrate temperature is increased from 50ºC to 60ºC� Figure 4 shows the output characteristics of prints at 60ºC substrate temperature�

Conclusion and Future Work:We have fabricated ink-jet printed ZnO TFTs using a precursor at a concentration of 0.5 M. The printed film was then subjected to a pre-bake at 100ºC for 10 min before

annealing at 480ºC for 1 h� It was found that the performance of the ZnO TFTs improved as the substrate temperature and drop spacing increased during printing� Both factors affect how fast the solvent is evaporated for better film formation. Further investigation must be done to confirm the trends observed and optimize the performance of the transistors� Also, we would like to print on other substrates besides Si/SiO2 as well as use nanoparticles or metal solutions for printing electrodes�

Acknowledgments:This project was supported by the NNIN REU Program funded by the National Science Foundation and NSF Bridge Award # ECCS 1032538� Thanks go to the Lurie Nanofabrication Facility and its staff for the logistical support�

References:[1] J�J� Schneider et al�, Adv� Mater� 2008, 20, 2283-3387�[2] C� Avis and J� Jang, Electrochem� Solid St�, 2011, 14 (2), J9-J11�

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Conformal Copper Seed Layers for Through-Silicon Vias Using Chemical Vapor Deposition

Parker ClarkElectrical Engineering and Materials Science, Stanford University

NNIN REU Site: Center for Nanoscale Systems, Harvard University, Cambridge, MANNIN REU Principal Investigator: Professor Roy Gordon, Department of Chemistry, Harvard UniversityNNIN REU Mentor: Yeung (Billy) Au, Department of Chemistry, Harvard UniversityContact: [email protected], [email protected], [email protected]

Abstract and Introduction:In recent years, leading semiconductor manufacturers have acknowledged that the time-tested technique for increasing processor performance—shrinking transistor size—cannot be sustained much longer [1]� Soon, performance will be enhanced by expanding today’s two dimensional integrated circuits to the third dimension [1]. One significant issue that arises from this radical rethinking of design is interconnectivity between the circuit’s stacked device layers� Recent advancements in through-silicon via (TSV) technology solve this problem by allowing for cheap internal interconnects with high-current capabilities that can be easily incorporated into modern circuit processing techniques (Figure 1) [2]� Because these vias must be filled with copper before functioning, we have explored an efficient two step filling procedure. We use chemical vapor deposition to deposit a thin conformal copper seed layer on the TSV structures, which can subsequently be filled completely with electroplated copper. In this paper we highlight our procedure’s successes on current TSV technology and discuss its ability to satisfy the needs of future TSV developments�

Experimental Procedure:After receiving blank TSV substrates from one of our industrial collaborators, we first exposed the substrates to five minutes of UV ozone treatment to remove surface

Figure 1: A typical technique for semiconductor integration of TSV features.

contamination and increase the surface density of reactive sites� The wafer coupons were then placed in a custom-built chemical vapor deposition (CVD) reactor that employed a bubbler system to transport both copper and manganese precursor molecules to the growth chamber (Figure 2)� Both the manganese precursor, bis(N,N’-diisopropylpentylamidinato) manganese (II), and the copper precursor, copper (N, N0-di-sec-butylacetamidinate) dimer, were synthesized by previously-described methods [3, 4]�

Growth of the seed layer began with a five minute deposition of manganese nitride� Manganese (Mg) was evaporated from the precursor liquid at 90°C into a 60 standard cubic centimeters per minute (sccm) flow of highly purified nitrogen (N) and then mixed with 60 sccm of N and 60 sccm of purified ammonia before entering the growth chamber, which we maintained at 130°C and 5 Torr�

The reactor was then cooled to approximately 30°C under a flow of 60 sccm purified N before exposure to an ethyl iodide (CH3CH2I) source for 30 seconds� This exposure occurred without a carrier gas and was controlled by a needle valve to a pressure of 0�05 Torr� We completed growth of the copper seed layer by simultaneous deposition of both copper and Mg� Manganese vapor was evaporated at 90°C with 100 sccm N carrier gas� Copper was evaporated from liquid at 130°C with 100 sccm N carrier gas�

Figure 2: A basic schematic of our custom CVD system.

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


An additional flow of 100 sccm hydrogen gas mixed with the precursor gases upstream from the growth chamber, which was maintained at 180°C and 10 Torr� Deposition duration varied from 60 to 120 minutes� In some trials the substrates were then annealed at 350°C for one hour in a purified nitrogen atmosphere.

Results and Conclusions:Under the conditions described above, we conformally coated up to 20:1 TSV features. SEM and x-ray fluorescence revealed the seed layer to be approximately 90 nm thick with 1% manganese content (Figure 3)� Upon inspection, we found that the narrowing sidewalls of our TSV structures significantly increased the features’ effective length to diameter aspect ratios� As such, we hypothesize that our procedure could conformally coat uniform features with aspect ratios of 25:1 or even 30:1 (Figure 4)�

Most substrates passed the Scotch® tape adhesion test without annealing, indicating sufficient adhesion for subsequent via fill by electroplating. We believe that this enhanced adhesion stemmed from the manganese atoms in our copper film, which diffused, even at deposition temperatures, to the silicon-manganese nitride interface� At this substrate interface, the manganese formed manganese silicate compounds and improved film adhesion as indicated by previous research [5]� An additional post-deposition anneal has been shown to further increase copper film adhesion [5]� Such strong bonding energies, combined with the diffusion barrier properties provided by the film’s manganese nitride underlayer, ensure that our CVD films can survive both semiconductor processing and operation�

Without our procedure’s brief ethyl iodide exposure, we would not be able to achieve such high aspect ratio conformal coatings in a reasonable timescale, if at all� Previous research has indicated that the iodine atoms form a weakly-adsorbed monolayer and catalyze copper nucleation before detaching from the MnNx underlayer [5]� Iodine then pools at the bottom of each via and augments the local catalytic effect, resulting in a bottom-up superfill that maintains a similar film thickness along the entire structure.

Although we were able to conformally coat most 20:1 vias, we discovered that 20:1 TSV features patterned in very high densities could not be adequately coated� We hypothesize that the large effective surface area created by the high feature density resulted in a local depletion of reactive gases before all vias were coated� Fortunately, this loading effect only became problematic for feature densities that were much too high for practical uses�

Future Work:Although recent semiconductor outlooks call for 50:1 TSV features by 2015, we currently cannot test our procedure on such structures because are not yet mass producible� As such, we are currently refining our process for 20:1 features in order to optimize both growth efficiency and film characteristics�

Acknowledgements:I would like to thank my mentor, Billy Au, and PI, Professor Roy Gordon, for motivating and directing this research� I would also like to thank the NNIN REU Program for support and funding� Finally, we would like to thank DOW Chemical for supplying CVD precursors�

References:[1] Chau, R�; Doyle, B�; Datta, S�; Kavalieros, J�; Zhang, K� Nat� Mater�

6, 810-812� [2] Lassig, Steve� ElectroIQ Magazine� 50 (12)� 2007� [3] B� S� Lim, A� Rahtu, J� S� Park, and R� G� Gordon, Inorg� Chem�, 42,

7951 (2003)� [4] Z� Li, A� Rahtu, and R� G� Gordon, J� Electrochem� Soc�, 153, C787

(2006)� [5] Au, Yeung; Lin, Youbo; Roy G� Gordon� J� Electrochem� Soc�, 158

(5)� 248-253�

Figure 4: Conformally coated 25:1 aspect ratio TSV features.

Figure 3: Conformally coated 10:1 aspect ratio TSV features.

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Optimization of Switching Layer for Retention in Tungsten Oxide Memristive Devices

Emily GriffinPhysics, Case Western Reserve University

NNIN REU Site: Lurie Nanofabrication Facility, University of Michigan, Ann Arbor, MINNIN REU Principal Investigator: Dr. Wei Lu, Electrical Engineering and Computer Science, University of Michigan, Ann ArborNNIN REU Mentor: Ting Chang, Electrical Engineering and Computer Science, University of Michigan, Ann ArborContact: [email protected], [email protected], [email protected]

Figure 1: Schematic of the device structure.

Abstract and Introduction:The memristor, essentially a resistor with memory, has generated much interest as a newly accessible circuit component that is dependent on nanotechnology� A switching layer between two electrodes can change the resistance of the device as oxygen vacancies, acting as a dopant in the oxide, redistribute when exposed to high electric fields. The film composition, specifically the oxygen concentration gradient, therefore has a significant impact on device performance� Reactive sputtering was employed for the tungsten oxide switching layer, and the process was characterized using differing oxygen flow rates and substrate temperatures during sputtering, which can be fine-tuned to change the retention, or the ability to remember a resistance when voltage is removed, based on device application� One application used by our group is as “synapses” in neuromorphic circuits: memristors’ ability to connect in large networks can be used to emulate neuron/synapse connections, and the memory component of the memristors simulates learning in the brain [1, 2]� For neuromorphic circuits, retention should be high, so this project aimed to fabricate memristors that demonstrate good resistance retention at a photolithography-compatible scale�

Experimental Procedure:All tungsten oxide thin films in this study were reactively sputtered using differing gas flow rates and substrate temperatures. Thin films deposited on silicon substrates with a silicon dioxide insulating layer were characterized using x-ray photoelectron spectroscopy (XPS) to determine the oxygen and tungsten concentration� The memristive devices themselves were prepared using photolithography for patterning and liftoff for both electrodes and gold probing pads�

Fabrication began with a 40 nm layer of palladium evaporated onto the patterned wafer and lifted off to form the bottom electrode� The switching layer, 30 nm of tungsten oxide, was sputtered onto the device, and then after patterning again, a 65 nm layer of either palladium, or titanium and platinum for an asymmetric device, was evaporated onto and lifted off of the wafer� The top electrode was then used as an etch mask and the tungsten oxide was etched through to reveal the bottom electrode via plasma etching� Finally, gold probing pads 100 nm thick were patterned and evaporated, along with an adhesion layer of 10 nm of nickel-chrome� After a final liftoff, characterization using a probe station was carried out to find the I-V characteristics of the devices. Figure 1 shows a schematic of the device structure�

Results and Conclusions:It has been documented in the literature that below 30% O2, flow rate in the sputtering chamber tungsten oxide becomes less stoichiometric and increasingly conductive [3]� This was verified by XPS measurements on samples with 35%, 25%, 15%, and 5% O2 flow rates. As the flow rate drops, the oxygen concentration in the sample falls from 42% to 31% and the tungsten concentration increases from 37% to 53%� The remaining percentage of the sample surface is carbon, since every sample had been exposed to air� Two XPS survey scans for films deposited with 35% and 5% O2

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


flow are shown in Figure 2. A flow rate around 25% was ultimately deemed best for controlled memristive effects, as a flow rate close to 15% resulted in too many oxygen vacancies and the devices simply showed a linear I-V characteristic, which is just a resistor with no memory; while flow rates above 30% would make the film too stoichiometric to allow resistance changes, having high oxygen concentrations and therefore very few oxygen vacancies�

Figures 3 and 4 show I-V characteristics of devices fabricated with 25% O2 flow. Figure 3 exhibits the well-defined pinched-hysteresis loops expected from a memristor� Repeated positive voltage sweeps result in a continued increase in device conductance, demonstrating good retention� Figure 4, however, while exhibiting pinched-hysteresis loops, shows poorer retention than the device in Figure 3; repeated positive voltage sweeps do not result in much change in device conductance� The different behaviors between devices may be related to differences in tungsten oxide film quality because of non-uniformity during film fabrication. Asymmetric devices were also fashioned, because the symmetry of earlier devices made their polarity unpredictable�

Future Work:Since only limited flow rates and substrate temperatures were tested, it could behoove the lab to continue characterization of the films at temperatures near 350°C with added flow rates, especially if finer control of retention is needed� There are also plans to test devices using two layers of tungsten oxide sputtered with different flow rates, or to sputter films with gradients of oxygen concentration, to see if these arrangements can make the polarity still more predictable, and the devices useable as fabricated without a forming sweep�

Acknowledgements:I would like to thank my research partner, Joshua Holt, for his work alongside me on these devices� I would also like to thank my mentor, Ting Chang, and my principal investigator, Dr� Wei Lu, for their willing explanations and suggestions, and the rest of Dr� Lu’s research group for their assistance and camaraderie� I would also like to acknowledge the staff at the University of Michigan who coordinated this program: Brandon Lucas, Sandrine Martin, and Trasa Burkhardt� Finally, I thank the Lurie Nanofabrication Facility at the University of Michigan and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program, funded by the National Science Foundation, for allowing me to have this opportunity�

References:[1] “Nanoscale Memristor Device as Synapse in Neuromorphic Systems,” Sung

Hyun Jo, Ting Chang, Idongesit Ebong, Bhavi Bhavitavya, Pinaki Mazumder and Wei Lu, Nano Lett�, 10, 1297-1301 (2010)�

[2] “Synaptic Behaviors and Modeling of a Metal Oxide Memristive Device”, T� Chang, S� H� Jo, K�-H� Kim, P� Sheridan, S� Gaba, and W� Lu, Appl� Phys� A� 102, 851-855 (2011)�

[3] “Stoichiometry and microstructure effects on tungsten oxide chemiresistive films”, Moulzolf, S.C., Ding, S. and Lad, R.J., Sensors and Actuators B: Chemical� 77, 375-382 (2001)�

Figure 2, top: Two XPS survey scans for films deposited with 35% and 5% O2 flow.

Figure 3, middle: Well-defined pinched-hysteresis loops expected from a memristor.

Figure 4, bottom: Repeated positive voltage sweeps do not result in much change in device conductance, showing poorer retention than the device in Fig. 3.

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Using Electron Beam Lithography for Nanowire Transistor Fabrication

Andreas M. HaggertyMechanical Engineering, Harvard University Class of 2013

NNIN REU Site: Howard Nanoscale Science and Engineering Facility (HNF), Howard University, Washington, DCNNIN REU Principal Investigator: Dr. Gary L. Harris, Electrical Engineering, Howard UniversityNNIN REU Mentor: James A. Griffin, HNF Lab Manager, Howard UniversityContact: [email protected], [email protected], [email protected]

Abstract:

The objective of this project was to fabricate a nanowire transistor using electron beam (e-beam) lithography. Nanowire transistors are devices made to solve the scaling problem that contemporary silicon transistors have begun to face. By placing nanowires on the surface of silicon chips, as opposed to using conventional photolithographic processes to create gates, the gate length restriction imposed on conventional methods can be overcome. To start this process, a 60-100 nm silicon dioxide (SiO2) layer was grown by dry thermal oxidation on a boron-doped, 0.005 Ω-cm resistivity, <100> silicon wafer. A 10 nm layer of chrome (Cr) and a 100 nm layer of gold (Au) were then evaporated onto the backside of the Si wafer and tested for ohmic behavior. Photolithography was then performed to create Cr/Au (10nm/100nm) fingers and pads on the top oxide layer. Nanowires were then deposited on the top surface by brushing with cloth followed by e-beam resist deposition. Scanning electron microscopy (SEM) was performed to locate the nanowires relative to the fingers, and holes were developed in the resist and coated with Cr/Au (10nm/20nm) to attach the nanowires to the fingers. Devices were then tested.

Introduction:Since the second half of the twentieth century, transistor technology has been the driving force behind the evolution of modern electronic systems� There are semiconductor devices that function as on/off switches or amplifiers in logic circuits. As such, the more transistors you can fit on a chip, the greater computational power you can achieve� Therefore, modern electronics are dependent on the steady miniaturization of transistors� Currently, the predominant transistor fabrication method is the complimentary metal oxide field effect transistor (CMOSFET). However, the CMOSFET method is reaching its limits in reducing the size of transistors�

Within the last few years, the National Institute of Standards and Technology (NIST) proposed a nanowire transistor design to overcome the restrictions conventional photolithographic methods place on device scale� Additionally, NIST claimed that nanowire transistors reduce current leakage, switch more effectively, and would allow industry to continue utilizing its silicon infrastructure� For this project, a manual stage scanning electron microscope and electron beam writing setup was used specifically to test whether nanowire transistors could easily be fabricated�

Experimental Procedure:To start the process, a 60-100 nm layer of SiO2 was grown on a low resistivity p-type <100> silicon wafer� The backside oxide layer was subsequently removed with hydrofluoric acid� A layer of Cr/Au (10nm/100nm) was then deposited onto the backside by e-beam evaporation in order to create the backgate contact� This metal-semiconductor contact was then tested for ohmic behavior� The current-voltage behavior of the contact is shown in Figure 1�

Figure 1: IV characteristics of backgate contact. Exhibits ohmic characteristics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


Next, Microposit S1818 resist was spun on the topside and exposed through a mask to form fingers and pads. A layer of Cr/Au (10nm/100nm) was evaporated onto the sample and the excess metal was lifted off to create metalized fingers and pads on the oxide surface, which served as transistor sources and drains�

In order to place the nanowires in close proximity to the fingers, a KIMTECH wipe was gently brushed over a sample of gallium nitride (GaN) nanowires (approximately 200 nm in diameter) and then brushed onto the sample� MicroChem 950PMMA resist was then deposited onto the sample and loaded in the SEM� Once the nanowires were located, patterns were written into the resist using Raith Elphy e-beam writing software to connect the nanowires to the source and drain fingers. This part of the work was critical because the e-beam used for imaging (not etching) also partially exposed the sample� Much effort was undertaken to optimize imaging conditions to avoid unwanted pre-exposure of the resist� Optimization parameters included accelerator voltage, spot size and contrast� After e-beam writing, the samples were developed using a 1:3 mixture of methyl isobutyl ketone and isopropanol� A layer of Cr/Au (10nm/20nm) was evaporated onto the sample and lifted off� An image of the completed device is shown in Figure 2�

Results and Conclusions: Near-ohmic contacts were established between the source/drain contacts and the gallium nitride nanowires; see Figure 3� Unfortunately, when the gate-to-source bias was adjusted between -20V and +20V, no change was observe in the voltage-current curve for the transistor; see Figure 4� It is speculated that the GaN nanowires were heavily doped and therefore higher bias voltages were needed to modulate the current across the device�

In conclusion, SiO2 was grown at desired thickness on wafer� Ohmic and near-ohmic contacts were established on the device� Nanowires were found and connected to pads with SEM and electron beam lithography�

Nanowires conducted current, however, bias modulation did not exhibit proper transistor characteristics�

Acknowledgments:Dr. Gary L. Harris, Principal Investigator; Mr. James A. Griffin, Mentor; National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program; Howard Nanoscale Science and Engineering Facility (HNF); HNF Lab Faculty�

References:[1] S�M� Koo, M�D� Edelstein, et al� 2005� Silicon nanowires as

enhancement-mode Schottky barrier field-effect transistors. Nanotechnology 16�

Figure 4: Bias modulation of device from -20V to 20V. Characteristics change in slope does not occur as with usual transistors.

Figure 2: Nanowires connected to source and drain contacts with Au. Represents completed device.

Figure 3: IV characteristics of topside contact (source to drain). Exhibits near-ohmic characteristics.

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Shape-Specific Iron Platinum Nanocrystals for Spin-Transfer Torque in Magnetic Tunnel Junctions

Kai HeElectrical and Computer Engineering, Rice University

NNIN REU Site: Microelectronics Research Center, The University of Texas, Austin, TXNNIN REU Principal Investigator: Prof. Sanjay Banerjee, Electrical and Computer Engineering, University of Texas at AustinNNIN REU Mentor: Dr. Domingo Ferrer, Electrical and Computer Engineering, University of Texas at AustinContact: [email protected], [email protected], [email protected]

Abstract and Introduction:Using magnetic states for data storage has the advantages of nonvolatile storage bits, nondestructive read/write operations, and no wear-out mechanism [1]� We fabricated spin transfer torque magnetic tunnel junctions (STT-MTJ’s); bit level devices of spin transfer torque random access memory (STTRAM)� These devices have a ferromagnet / dielectric / ferromagnet top to bottom construction and operate on the principle of tunneling magnetoresistance� Resistance of the junction varied depending on the relative orientation of the spins in the magnetization directions of the ferromagnetic layers� Parallel alignment displayed low resistance, while antiparallel ordering portrayed high resistance [2]�

Figure 1: (A) Thin film device stack. (B) Nanoparticle device stack.

High anisotropy, chemical and thermal stability, and tunable morphology of iron platinum (FePt) nanocrystals make them attractive for STTRAM applications� This project sought to assess the viability of replacing the top ferromagnet layer with ferromagnetic FePt nanocrystals (Figure 1b)�Bias dependence of differential resistance and tunneling magnetoresistance were also investigated on magnetic tunneling junctions (MTJs) with an ultrathin magnesium oxide (MgO) tunnel barrier (~20Å) fabricated via atomic layer deposition (ALD) (Figure 1a)�

Nanocrystal Synthesis and Device Fabrication:Nanocrystals were synthesized by reduction of platinum acetylacetonate and thermal decomposition of iron pentacarbonyl� Oleylamine and oleic acid were used as surfactants� Shape was controlled by varying solvents, sequence of surfactant addition, and reaction temperatures� We synthesized obloids and nanowires for the MTJ’s�To create obloids (Figure 2a) 0�192g platinum acetyl-acetonate, 10 ml benzyl ether, and 5 ml octadecene were added to a three-necked flask and stirred at 45°C for one hour� Next, the reaction was heated to 120°C followed by the immediate addition of 0�15 ml iron pentacarbonyl and 1�6 ml oleic acid� Five minutes later, 1�65 ml oleylamine was added� The reaction was then heated to 200°C and held for two hours [3]� To yield nanowires (Figure 2b) 0�098g platinum acetylacetonate, 100 mg 1,2-hexadecanediol, and 10 ml of oleylamine were added to a three-necked flask and stirred at room temperature for 20 minutes� The reaction was heated to 100°C and 10 µL iron pentacarbonyl was added� After holding the temperature at 100°C for the next 20 minutes, 2 ml oleic acid was injected� The reaction was heated to 300°C and held there for 30 minutes [3]� Both reactions ran under a nitrogen atmosphere�The devices were fabricated on a silicon wafer by depositing silicon dioxide (SiO2) through evaporation, tantalum nitrite (TaN) using sputtering, and SiO2� Wells of 70 × 70 nm2, 100 × 100 nm2, and 200 × 200 nm2 for the stacks were patterned in the top SiO2 layer using electron-beam

Figure 2: Transmission electron microscope images of (A) obloids and (B) nanowires.

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


(e-beam) lithography and dry etched afterwards� The stack was sputtered, with the exception of the MgO layer where ALD was used�In the future, the nanoparticle device-free layer (nano-particles) will be attached by first immersing the wafer in a 20 mg/ml poly(ethyleneimine) (PEI) solution in chloroform for five minutes followed by immersion in a 1 mg/ml nanoparticle solution for 20 minutes [4]�Top and bottom electrodes were patterned using photo lithography� The top electrode was created through evaporation of titanium followed by lift-off. RIE was used to define the underlying TaN bottom electrode�We have only finished fabricating the proposed thin film devices (Figure 3) by the time of submission of this report� The nanoparticle device wafer has been patterned by e-beam and awaits stack deposition�

Results and Future Work:FePt nanoparticle magnetic properties were measured by sweeping an applied magnetic field and then acquiring magnetic moment data via a superconducting quantum interference device (SQUID) magnetometer� Both unannealed and annealed obloidal samples showed a discernable hysteresis� Annealing put the FePt into a more ordered L10 state and as Figure 4c shows, the annealed sample had higher magnetic moment when exposed to applied fields of comparable magnitude as the unannealed samples� Nanoparticles with an L10 structure were required for room temperature device operation�We have preliminary thin film device data made through a two point probe� Resistance measurements were made after biasing the device in the range of -1V to 1V (Figure 4b), as well as after applying varying pulse amplitudes (Figure 4d)� Biasing the device showed a higher resistance on the voltage up-sweep than the down-sweep� Pulsing currents of increasing amplitudes decreased the resistance of the junction� Then pulsing currents of decreasing amplitude switched the device back into a higher resistance state� This data shows a clear hysteresis in the switching characteristic of the device� Both tests suggest switching� However, in further pulse testing, the device resistance proved difficult to switch back from low resistance (parallel state) into the high resistance state (antiparallel state)� This was perhaps due to

the oxidation of the top titanium electrode�In the future, we would like to confirm this is indeed the case by changing the top electrode to copper or gold� Additional work includes implementation of nanoparticle devices and verification of their functionality. Comparisons of switching critical current and switching speed in these thin-film and nanoparticle devices also stand as future goals�

Acknowledgements:I thank Prof� Sanjay Banerjee and Dr� Domingo Ferrer for their guidance and opportunity to work with them this summer� I also want to thank the Banerjee group, in particular Samaresh Guchhait, Urmimala Roy, Soheil Gharahi, Archana Rai, and Prateek Gadkari for their efforts� I also thank the NNIN REU Program and the NSF for organizing and funding this experience�

References:[1] Ferrer, D� A�; “Origin of shape anisotropy effects in solution-phase

synthesized FePt nanomagnets”; J� Appl� Phys� 110, 014316 (2011)�

[2] Slaughter, J�; “Materials for Magnetoresistive Random Access Memory”; Annual Review of Materials Research, Vol� 39, 277-296 (2009)�

[3] Narayan P�; “Shape control of FePt nanocrystal”; J� Appl� Phys� 105, 07A749 (2009)�

[4] Yildirim, O�; “Monolayer-directed Assembly and Magnetic Properties of FePt Nanoparticles on Patterned Aluminum Oxide”; International Journal of Molecular Sciences, 11� pp� 1162-1179 (2010)�

Figure 4: SQUID data for (A) unannealed and (C) annealed obloid nanoparticles show ferromagnetism. Resistance measurements made after (B) voltage sweeps and (D) current pulses of increasing amplitude show device switching.

Figure 3: Finished thin film device.

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Fabrication of GaAsBi Heterojunction Bipolar Transistors

Hilary HurstEngineering Physics, Colorado School of Mines

NNIN REU Site: Colorado Nanofabrication Laboratory, University of Colorado, Boulder, CONNIN REU Principal Investigator: Prof. Bart Van Zeghbroeck, Electrical Engineering, University of Colorado BoulderNNIN REU Mentor: Ian Haygood, Electrical Engineering, University of Colorado BoulderContact: [email protected], [email protected], [email protected]

Abstract and Introduction:III-V semiconductor based bipolar junction transistors are favored in high performance circuits because of their high transconductance and low output conductance� Heterojunction bipolar transistors (HBTs), which utilize a junction of lattice-matched semiconductors with different bandgap energies, excel in high frequency applications� The gain of a HBT depends exponentially on the discontinuity between the base and emitter band junction [1]� This allows for higher doping in the base, lowering base resistance and improving RF performance�

Gallium arsenide bismuthide (GaAs(1-x)Bix) is a novel material for HBTs� GaAs(1-x)Bix is lattice matched to GaAs and narrows the bandgap by ~ 90 meV/%Bi [2]� As shown in Figure 1, the band discontinuity in GaAs(1-x)Bix is entirely in the valence band, which preserves the electron mobility but negatively affects the hole mobility� The smaller bandgap in GaAs(1-x)Bix allows for a lower turn-on voltage, therefore HBTs fabricated using GaAs(1-x)Bix have the potential to be more efficient than homojunction GaAs transistors. The focus of the project is to characterize GaAs(1-x)Bix pn junctions and to create a new GaAs(1-x)Bix HBT device for RF power amplifiers.

Methodology:An epitaxially grown wafer with a bismuth concentration of x ~ 2�3% was used, and a schematic of the HBT design is shown in Figure 2� A double-mesa structure was designed to contact the base and subcollector, with ohmic contacts for the emitter, base, and collector� The base p-type material was doped higher than the n-type collector for lower capacitance and maximum current; a higher-doped subcollector was used for ohmic contacts� Gold germanium (AuGe) was used as the emitter/collector metal; gold zinc (AuZn) for the base contacts� Gold (Au) with a titanium (Ti) adhesion layer was used as a common metal contact to eliminate variation in contact resistance�

Fabrication was completed in six optical lithography steps, and samples were placed in a 30s reactive ion etch (RIE) to remove excess photoresist (PR) after each lithography� The emitter and base isolation etches were completed using AZP4210 positive PR with 8s exposure and 35s AZ400K spinner development� Samples were etched in a 1:8:160 H2SO4:H2O2:H2O solution at a rate of 260 nm/min [3], verified by etch tests.

Figure 1: HBT forward bias band-energy diagram.

Figure 2: GaAsBi HBT cross-section and SEM image of the actual device.

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


For the emitter, collector, and base metals, the samples were patterned with NR71-1500PY negative PR with 80s exposure and 10s RD6 spinner development� 50 nm of AuGe, 20 nm nickel (Ni) and 50 nm Au were evaporated to form the emitter and collector contacts� Capping AuGe with nickel and a layer of gold smoothed the contact surface� Excess metal was removed using lift-off, and the wafers were again patterned� 100 nm AuZn was evaporated onto the wafer and lifted off to form the base contact� Finally the samples were placed in a 45s rapid thermal anneal (RTA)�

In order to isolate the common metal, a 250 nm layer of nitride (SiNx) was deposited using a 20 min plasma-enhanced chemical vapor deposition (PECVD)� After deposition, the samples were patterned with NR9-1000P negative PR, exposed for 8s and spinner developed for 10s with RD6� The nitride layer was etched in a PlasmaTherm RIE for 5 min at a rate of 50 nm/min to create via holes with which to contact the sample� Finally, the common metal lithography was completed in the same manner as the emitter and base metals using NR71-1500PY� 20 nm of Ti and 300 nm of Au were evaporated and excess metal was removed using lift-off�

doping slightly below 1017 atoms/cm3 as expected� The pn junction exhibited a linear relationship between applied voltage and the inverse square of capacitance�

HBT devices were fabricated with a minimum emitter mesa size of 12 µm × 100 µm� Poor contacts made devices difficult to test because the common metal deposition did not cover the large step from the subcollector to the emitter� Transistors were not achieved and diode characteristics were measured rather than true transistor curves� A thicker nitride layer or metal deposition could mitigate the issue� As shown in Figure 4 the base-collector junction exhibited ideal behavior, but the base-emitter junction showed poor rectification, possibly due to traps at the heterojunction. Future experiments are needed to improve the HBT fabrication process, particularly the contacts, and to further understand the behavior of GaAs(1-x)Bix in the HBT base-emitter junction�

Acknowledgements:National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program and the National Science Foundation for this opportunity, principal investigator Prof� Bart Van Zeghbroeck, mentor Ian Haygood, and Ph�D� student Zefram Marks for their knowledge and assistance� Finally thank you to CNL staff, specifically Mark Leonas.

References:[1] Muller R�, Kamins T�, Chan M�; Device Electronics for Integrated

Circuits; 3rd ed�, New York: John Wiley & Sons, 2003, pp� 313-320�[2] Alberi, K� ;“Electronic Structure of GaAs(1-x)Bix”; National

Renewable Energy Laboratory� Golden, Colorado� 7 June 2011� Private communication�

[3] Williams, R�; Modern GaAs Processing Methods; 1st ed�, Boston: Arctech House, 1990�

Figure 4: Base-emitter and base-collector characteristics (see Figure 2 for material structure).

Figure 3: PN junction characteristics. PN junctions were charac-terized using material with a doped p-type layer of Bismuthide or GaAs on top of a doped n-type layer of GaAs and a GaAs substrate.

Results and Conclusions:Electrical tests of GaAs:GaAs(1-x)Bix pn junctions were conducted as shown in Figure 3� GaAs pn diodes exhibited an ideality factor of n = 2 or lower, and GaAs(1-x)Bix diodes exhibited ideal behavior at a lower turn-on voltage than the GaAs diodes corresponding to a smaller bandgap in the transistor base material as expected� The ideality factor of the GaAs(1-x)Bix diodes rose rapidly above n = 2�

In material characterization tests, GaAs(1-x)Bix was found to have a sheet resistance of 2,010 Ω/, resistivity ρ = 6 × 10-2 Ω-cm and a hole mobility of 13 cm2/V·s� Hole mobility was significantly higher in GaAs (93 cm2/V·s) while sheet resistance and resistivity were lower (450 Ω/, ρ = 1�35 × 10-2 Ω-cm respectively). This agrees with theory and previous material characterizations [2]� Capacitance measurements showed the material to exhibit uniform

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


The Micro-Fabrication of a Composite Thermal Capacitor

Israel IlufoyeMechanical Engineering, University of Maryland Baltimore County

NNIN REU Site: Nanotechnology Research Center, Georgia Institute of Technology, Atlanta, GANNIN REU Principal Investigator: Dr. Andrei Fedorov, Mechanical Engineering Georgia Institute of TechnologyNNIN REU Mentor: Craig Green, Mechanical Engineering, Georgia Institute of TechnologyContact: [email protected], [email protected], [email protected]

Abstract:

In order to maintain the rate of advancement achieved in traditional microprocessor technology over the past few decades, recent studies have focused on three dimensional stacked microprocessor architectures because of their potential computational advantages as well as a decrease in noise, power consumption, and reduction in parasitic capacitance. The implementation of this solution has been limited by the generation of transient hot-spots. To manage the thermal impact of these transient hotspots, we fabricated composite thermal capacitors (CTC), which locally increase the thermal capacitance of the chips in the vicinity of the hotspots, allowing for longer device operating times. The fabricated and tested prototype devices consist of phase change materials (PCM), and heat spreader matrices monolithically integrated into silicon chips also containing micro-heaters, which simulate microprocessors. The CTC’s novelty arises from its efficient combination of high thermal conductivity materials with PCMs to yield enhanced effective thermal conductivity while retaining desirable thermal capacitance in order to increase the micro-processor operating time. By using microfabrication techniques (Figure 1) such as chemical vapor deposition, photolithography, metal deposition via electron beam evaporation and sputtering, and reactive ion etching, we constructed a fabrication outline for a prototype CTC device, while a heater was monolithically fabricated with the CTC to simulate the microprocessor.

Fabrication Outline:

Figure 1: Fabrication outline for positive and negative photoresist.

An oxide passivation layer was deposited on the front and back side of a blank silicon wafer in order to control current flow through the heater coils and also to act as a mask for the CTC patterns� The heater was fabricated on the chosen front side�

Futurrex NR71 3000P positive photoresist was spun on the front side for the heater device lithography� It was essential to run a descum process after the lithography process in order to strip remnants of resist left on the pattern� Using a metal evaporator, 250 A of titanium (Ti, adhesion layer), and 2500 A platinum (Pt, resistive property) was deposited� After lift-off, the leads, which channel current to the heater device, were patterned using NR71 3000P photoresist�

For the lead metallization, 500 A of Ti, 3000 A copper (Cu, electrical properties), and 2000 A of gold (Au, wire bonding purposes) were deposited using a metal evaporator�

On the back side, Microposit 1827A positive photoresist was spun for the CTC lithography� In order to attain the monolithic integration of the heater with the CTC device, a back side alignment (BSA) procedure was used� After the lithography, the wafer was mounted CTC side up using SPR 220 7�0 photoresist, done to prep the wafer for reactive ion etching� A mask etch was done to remove the oxide on the CTC pattern� Since the surface of the patterns was now exposed, a 200-250 µm vertical depth trench etch of silicon was performed to produce the PCM trench, using

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


an inductively coupled plasma� In order to reduce the temperature gradient between PCM and the hotspot, the wafer was thinned by stripping the surface oxide, thus exposing the surface to the etching plasma (Figure 2); ideally the surface and trench etch rate should now be equal, therefore thinning the wafer� Once the change in distance from the PCM trench to the heater was between 75-100 µm, 500 A of Ti and 1�5 µm of Pt were added for wettability with the PCMs�

Fabrication Challenges:For the heater fabrication, some issues regarding pattern adhesion arose after development due to an anisotropic post exposure bake� Since the patterns were not well baked, after development some features were detached from the device, thus leaving the device useless� By using a vacuum contact post bake, the wafer was brought closer to the hot plate, therefore increasing the level of uniformity and reducing the amount of device defects (Figure 3)�

The etching equipment for the CTC device operates by cooling the backside of the wafer while the plasma is

Figure 2: CTC etch process.

Figure 3: Before and after vacuum post bake.

etching through the patterns to increase the etching quality� So the ideal adhesion material was thermal grease because of its high thermal conductivity� But since a low thermal conductive material (SPR 220) was used, the cooling was anisotropic, thus producing a low quality etch� Another issue that arose during the CTC etch was the trench depth decreased during the thinning step, meaning that the etch surface etch rate was not equal to the

trench etch rate� In order to increase the directionality of the etching plasma, the etching pressure was reduced by 20%; as a result the etch rate was relatively equal, therefore maintaining a desirable trench depth�

Conclusion:Using thermal grease as the adhesion material increased the difficulty of un-mounting the wafer after final CTC processing, which increased the chances of damaging the devices; so by using instead a thermal conductive material with adhesive properties similar to SPR 220, we will be able to increase the quality of the CTC device�

In conclusion, we were able to construct a fabrication outline for a CTC device which includes procedures that can be followed to reduce device defects�

Future Work:Tin-, lead-, and indium-based alloys, like alloy 134 and 174, will be deposited into etched trenches to prep the CTC device for testing� The CTC device will be tested by being assembled in a microprocessor-like structure combined with a solid state cooler (SSC) (Figure 4)� The goal of the SSC is to regenerate the phase change material during the idle state of the core in order to allow the PCM to absorb heat effectively during the active mode of the core�

Acknowledgements:I would like to thank Craig Green, Dr� Andrei Fedorov, Dr� Nancy Healy, Mrs� Katie Hutchinson, NNIN REU Program, National Science Foundation (NSF) and also Georgia Institute of Technology�

Figure 4: CTC SSC assembly.

References:[1] Green, C�E�, A� G� Fedorov, and Y�K� Joshi (2011);

“Thermal Capacitance Matching in 3D Many-Core Architectures�” In IEEE SEMI-THERM 27�

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Deterministic Assembly of Alternative Materials onto Silicon Substrates

Jia KuangElectrical Engineering, The City College of New York

NNIN REU Site: Penn State Nanofabrication Laboratory, The Pennsylvania State University, State College, PANNIN REU Principal Investigator: Dr. Theresa S. Mayer, Electrical Engineering, The Pennsylvania State UniversityNNIN REU Mentor: Scott M. Levin, Electrical Engineering, The Pennsylvania State UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Deterministic assembly of III-V compound semiconductor devices offers the promise of enabling innovative functions that go beyond the conventional use of digital electronics. Recent studies have achieved controlled placement of metal and semiconductor nanowires via electric-field-assisted assembly techniques. However, many III-V optoelectronic and electronic devices that are interesting are considerably larger in size. In this project, we studied the electric-field assisted assembly of 100 nm thick and 10 µm long chromium (Cr) microtiles having widths of 1.2 µm, 2.4 µm, and 3.0 µm. The non-uniform electric field used to manipulate the microtiles was induced in the isopropanol (IPA) solution that suspended the microtiles by applying a 100 kHz, 20 Vp-p bias voltage to pairs of interdigitated metal electrodes patterned on a silicon nitride (Si3N4) coated silicon (Si) substrate. The electrodes pairs are separated by 3 µm wide gap, and were coated with a 1 µm thick photoresist layer. The results of the study showed that uniformly spaced arrays of all of the microtiles could be assembled using this technique. The space between adjacent tiles in the array depended on the width of the tile and the gap. These promising results provide a proof-of-concept for the assembly of III-V epitaxial device layers on Si substrates.

Introduction:

Silicon complementary metal oxide semiconductor integrated circuit (CMOS IC) technology is reaching the physical limits of transistor scaling� Consequently, there is an increasing interest in adding value to the circuits through functional diversification [1]� Integrating diverse devices onto Si IC’s has the potential to enable new functions ranging from light detection and emission to highly sensitive chemical and biological detection� However, this requires the monolithic integration of many different kinds of materials, which are generally not compatible with conventional top-down fabrication methods� Electric-field assisted assembly has been used to deterministically position arrays of single nanowires on a Si substrate with high accuracy between each nanowire and a lithographically defined feature on a Si substrate [2].

The goal of this project was to assemble and study the deterministic assembly of much larger micron-size tiles using the same technique� In this study, chromium tiles were used as a model system to understand how changing

the dimensions of the starting materials impact the deterministic assembly process� In the future, alternative materials such as III-V compound semiconductors will be assembled using this bottom-up approach�

Fabrication and Experimental Procedure:The interdigitated electrode structure used for the assembly experiments, which is shown in

Figure 1, was fabricated using standard i-line projection lithography� The electrode structures were fabricated on 3-inch diameter thermally oxidized Si substrates by exposing the patterns printed on a quartz reticle in a 3012 photoresist layer and developing the exposed photoresist in SF-11. Each substrate contained 92 separate reticle fields that were separated from one another prior to the deterministic assembly experiments� Following lithography, thermal evaporation was used to deposit 10 nm of titanium (Ti) and 40 nm of gold (Au) on the patterned Si substrate, and the metal on top of the unexposed photoresist was lifted off

Figure 1: Cross section of the electrode.

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


using Microposit 1165 remover� A polydimethyglutarimide (PMGI) dielectric layer was coated onto the wafer and baked to form the planarized insulating spacer that separated the metal electrodes from the fluid used to suspend the microtiles during assembly� In parallel, the i-line stepper was also used to fabricate dense arrays of Cr microtiles having a length of 10 µm and varying widths of 1�2 µm, 2�4 µm and 3�0 µm using a similar liftoff process�

The microtiles were flowed across the assembly electrode structure in a fluidic channel that was formed by between a glass cover slip and the Si substrate� A manual probe station was used to electrically contact the assembly structure� A dielectrophoretic (DEP) force was induced on the microtiles by applying a sinusoidal bias voltage using a function generator with an amplitude of 20 Vp-p and a frequency of 100 KHz� Immediately prior to assembly, the IPA solution containing the Cr microtiles was sonicated for 20 seconds to ensure that aggregated tiles were separated and uniformly suspended� A 3 µl of Cr microtile solution was injected into the fluidic channel using a micropipette for assembly.

Microtiles are currently being fabricated with indium gallium arsenide (InGaAs) to be integrated onto Si circuits�

Acknowledgements:I would like to thank my mentor Scott M� Levin, my advisor Dr� Theresa M� Mayer, Kathy Gehoski, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates, and the National Science Foundation�

References:[1] More Than Moore White Paper� International Technology Roadmap

for Semiconductors� (August 04, 2011)� Retrieved from http://www�itrs�net/Links/2010ITRS/IRC-ITRS-MtM-v2%203�pdf

[2] Li, M�, Bhiladvala, R� B�, Morrow, T� J�, Sioss, J� A�, Lew, K�, Redwing, J� M�, Keating, C� D� and Mayer, T� S�; “Bottom-up Assembly of Large-Area Nanowires Resonator Arrays”; Nature Nanotechnology, 3, 88 (2008)�

Figure 4: Microtiles of three different widths compared.

Figure 2: A) long-range dielectrophoretic force; B) Shorter-range electrostatic force.

Results and Conclusions:The electric-field profile that defines the deterministic assembly process was simulated using COMSOL Multiphysics finite element modeling software. During the initial phase of the assembly process, long-range dielectrophoretic forces attracted and aligned the Cr microtiles across the interdigitated electrode gaps as shown in Figure 2a� Next, the shorter-range electrostatic interaction between adjacent Cr microtiles caused the tiles to space relatively uniformly in an array as shown in Figure 2b�

As shown in Figures 3a and 3b, the Cr microtiles were assembled according to the result predicted by our simulations� After analyzing the three different widths of Cr microtiles, we concluded that there is a dependency between the width of the Cr microtiles and the separation between the adjacent Cr microtiles� The numerical values of the spacing are shown in Figure 4�

This experiment demonstrated that the deterministic assembly technique can be applied to microtiles, and is viable for future experiments on a vast array of materials�

Figure 3: a) 1.2 µm Cr-microtile from FESEM; b) 3 µm Cr-microtile from optical microscope; dark field.

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Development of Carbon Nanotube Field-Effect Transistors for Use in Next Generation Electronics

Hongliang LiangEngineering, Swarthmore College

NNIN REU Site: Stanford Nanofabrication Facility (SNF), Stanford University, Stanford, CANNIN REU Principal Investigator: Hongjie Dai, Department of Chemistry, Stanford UniversityNNIN REU Mentor: Justin Wu, Electrical Engineering, Stanford UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Due to its exceptional electronic properties, the carbon nanotube field-effect transistor (CNTFET) is a promising alternative to the traditional metal-oxide-semiconductor field-effect transistor (MOSFET) used in current devices, replacing the channel material in MOSFETs with thin carbon nanotubes with diameters of 1-2 nm. Because various difficulties have arisen in the mass production of reliable CNTFETs, this project aimed to fabricate and analyze the efficiency of a number of these devices, examining the viability of their optimization for industrial applications. Spinning solutions that were specifically sorted for a high concentration of semiconducting CNTs onto silicon substrates, we then located these tubes on an atomic force microscope (AFM). The devices were manufactured through electron beam (e-beam) lithography and deposition. We applied test voltages between the source and drain of these devices and observed the amount of current flowing through the channels with respect to the backgate voltage. Insights gained as a result of these experiments will have a significant impact on the performance and power of future electronic devices.

Introduction:

For more than 40 years, Moore’s Law has held true in that the number of transistors on a chip doubles approximately every two years� However, an increase in the transistor count must lead to a decrease in transistor size, and serious limitations of current fabrication technology are faced at the sub-22 nm range� Since Intel, a world leader in the semiconductor industry, has already begun processing at the 32 nm scale in 2010, the success of predictions made by Moore’s Law may soon reach an end� CNTFETs provide a possible alternative in that they have near ballistic transport (minimal electron scattering) and much higher electron mobility than silicon (up to 70 times) [1]� However, challenges in their fabrication still exist, which includes the separation of metallic and semiconducting nanotubes in the same mixture and having a wide range of diameter in CNTs found in the mixture, which affects their electronic properties�

The Shinohara group from Japan has been working on a separation method known as multicolumn gel chromatography� This method is based on the structure dependent interaction strength of CNTs with an allyl dextran based gel� There are multiple columns of gel and the CNT solution is dispersed (using a surfactant called SDS) on the top of the first column. The absorption site at this column is fully occupied by nanotubes that exhibit the strongest interaction with the gel and the rest of the solution moves on

Figure 1: 1 µm × 1 µm image of a carbon nanotube as seen under an AFM.

Figure 2: Image of actual device as seen under a probe station.

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


to the next column� Metallic nanotubes are collected at the end because they exhibit the lowest interaction with the gel [2]� Since transistors need to constantly switch between the on and off state, semiconducting CNTs are desired�

In our project, we examined the viability of multicolumn gel chromatography through CNTFET fabrication using a pre-sorted solution prepared by the Shinohara group using the above method�

Procedure:We began the fabrication process by placing cross-shaped markers on silicon wafers through optical lithography� A sorted solution of CNTs was then spun onto the silicon substrate� Individual CNTs were located using an AFM, and Figure 1 shows a zoomed-in view of one such tube� The actual devices were manufactured through e-beam lithography and deposition, and one such device is shown in Figure 2� Afterwards, the devices were tested under a probe station to observe on/off ratio and on-current through the CNT channel�

Results and Discussion:Due to time constraint, only devices fabricated from eight CNTs were fully tested, and of those, two were found to be nonfunctioning� The on/off ratio we observed was comparable to previous experimental data, but the on-current was not� Figure 3 shows a plot of the CNT diameter versus on-current�

Theoretically, a larger diameter should allow for higher current flow, but our results indicate the highest on-current was measured from a device with the smallest CNT diameter� Also, we expected the on-current to be at least 5 µA, but the highest we achieved with our devices was only 0�9 µA [3]� A data summary of the working devices can be found in Figure 4�

Conclusion and Future Work:The on/off ratio that devices typically demonstrated suggests that the solution was well sorted and the tubes were semi-conducting (numerical accuracy was limited by probe station)� However, our experimental results for on-current do not match with previously reported data, so we will need to test more devices to confirm the viability of our process flow (including gel chromatography) and improve the on-current data� For future work, we aim to achieve comparable on-current from CNTs of similar diameters� Other than testing more devices, we will try to improve the contact resistance between the CNT and electrode� This could be done by better cleaning of SDS (a surfactant used in dispersing CNTs), using a metal other than palladium such as rhodium, and doping via plasma treatment prior to deposition�

Acknowledgements:I would like to thank the NNIN REU Program for an amazing research experience, and the NSF and Stanford’s Center for Integrated Systems for financial support. I also thank SNF for letting me using its equipment and appreciate the invaluable assistance provided by my PI Hongjie Dai, mentor Justin Wu, and site coordinator Mike Deal�

References:[1] Dürkop, T�; Getty, S� A�; Cobas, Enrique; Fuhrer, M� S� (2004)�

“Extraordinary Mobility in Semiconducting Carbon Nanotubes”� Nano Letters 4, p� 35�

[2] Huaping Liu, Daisuke Nishide, Takeshi Tanaka, and Hiromichi Kataura, 2011� “Large-scale single-chirality separation of single-wall carbon nanotubes by simple gel chromatography”� Nature Communications, v�2, Article Number 309�

[3] Wong Kim, Ali Javey, Ryan Tu, Jien Cao, Qian Wang, and Hongjie Dai, 2005� “Electrical contacts to carbon nanotubes down to 1 nm in diameter”� Applied Physics Letters, v�87, p�3�

Figure 3: Our experimental data for sorted CNTs.

Figure 4: Summary of working devices (6/8 tested).

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Characterization of Floating-Gate Graphene

Maria Veronica MateusMechanical Engineering, The University of Texas Permian Basin

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Dr. Edwin Kan, Electrical and Computer Engineering, Cornell UniversityNNIN REU Mentor: Lieh-Ting (Adrian) Tung, Electrical Engineering, Cornell UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Graphene is known for its high electron mobility (over 5,000 cm2V-1s-1). However, because of its zero bandgap, graphene is a non-ideal semiconductor for switches. Theoretical and experimental results indicate that graphene nanoribbon (GNR) opens up a bandgap because of the quantum confinement effect (inversely proportional to the width). Therefore, it is crucial to control the width of the growth down to sub-10 nm with a reliable method. However, the line edge roughness (LER) from electron beam (e-beam) lithography or photolithography degenerate the quantum confinement, which hinders the further application of GNR. Thus, combining the controllable width and sharp edge was one of our technical goals. The spacer lithography, a state of art patterning technology, has been widely used in current industry to create sub-pitch smooth patterns such as FinFET (a non-planar, double-gate field effect transistor). We combined e-beam lithography with well-calibrated spacer lithography to seek a better electrical performance on the metal-oxide semiconductor field-effect transistor (MOSFET) structure using graphene as a channel layer.

Calibration Procedure:

The calibration process to obtain the GNR needed for our transistor started with finding the optimal etching selectivity between silicon (Si), silicon dioxide (SiO2) and silicon nitrate (Si3N4) for the spacers that would reduce the LER and scale the width down to 10 nm� We started with a 4-inch Si wafer and deposited 180 nm of SiO2 to act as our spacer sidewalls, followed by a deposition of 20 nm of chrome (Cr) as to act our mask, obtain more anisotropic features, and avoid a facet� We then spun 3 µm of positive photoresist, exposed and developed vertical lines across the wafer, and followed by etching the Cr mask and removing the photoresist�

We etched the spacer sidewalls and used a scanning electron microscope (SEM) to check the sidewalls, as seen in Figure 1� After we found the right SiO2 deposition and correct etching times and rates, we moved on to depositing the Si3N4� We found that the Si3N4 deposited and the spacer widths obtained were directly proportional� We started by depositing 50 nm of Si3N4 and obtained about 53 nm spacer sidewalls�

For our calibration process, we deposited down to 40 nm Si3N4 and obtained about 39 nm spacers� We checked with the SEM to assure that the spacers were smooth, and we also checked the thickness, as seen in Figure 2� After we had the right spacer sidewall as well as the spacers, the next part of our calibration was finding the right chemistry to etch the Si3N4�

We changed the power as well as the plasma density in the Oxford 100 etcher, because with higher power, we obtained better anisotropic

Figure 1, above: SEM of our sidewalls.

Figure 2, below: SEM of our spacers, checking for smoothness and thickness.

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


features, and with higher plasma density, we obtained a cleaner and smoother edge� Although we assumed that high plasma density and high power would give us our best etching result, we found that instead, it resulted in over-etching the Si3N4; but medium power and high plasma density did not�

By checking with the SEM and with optical microscopy for the color reference, we found our final chemistry as seen in Figure 3� With the right chemistry, and the right deposition times and rates, our calibration process was done� The selectivity we obtained was about 3:1, and although ideally we were looking for a higher selectivity, our resulting selectivity was acceptable�

Experimental Procedure:To begin our experiment, we grew about 90 nm of SiO2, which was for the graphene to reflect from our wafer. We carved a coordinate system to help us relocate the graphene� Using the “tape method,” we exfoliated graphene and applied it on our wafer, being very careful not to apply too much pressure nor too little — to avoid dense graphite and too small graphene� The next steps were to use optical microscopy to locate graphene that was greater than 20 µm wide and record its location�

We then created a layout of lines that were 50 nm thick by

20-30 µm long, depending on the graphene thickness, and we send it to the e-beam tool to expose those lines across the possible graphene� But before this step, we had to protect our graphene and add spacer sidewalls�

We wanted to deposit a protection layer for the graphene, because since our wafers were going to be exposed to oxygen plasma, we didn’t want the graphene to go away� We deposited the SiO2 and the Cr mask using the deposition rates and times found in the calibration process, and then we wrote our layout with 200 nm of negative e-beam resist using

e-beam lithography� We followed with another calibration process, and then finally deposited about 30 nm of Si3N4 to be sure to obtain spacers� Ideally we would have wanted 80 nm total, 20 nm spacers and 40 nm of SiO2�

Results and Conclusion:Unfortunately we were unable to finish this project, and so were not able to see what spacers we finally obtained, but we assume that we obtained about 100-110 nm total� Although we were unable to finish, we did obtain the graphene nanoribbon� The next step will be to write our source/drain with positive resist, then wet etch the SiO2 and protection layer, and finally deposit 1 nm of Cr as an adhesive layer for the 100 nm gold� We’ll use the lift-off method to remove excess metal as well as other excess material� Finally we’ll clean the oxide off the back of our wafer and deposit 1 nm of Cr and 100 nm of Au as our back gate� Electrical measurements will be implemented to investigate the device characteristics�

Acknowledgements:National Science Foundation; NNIN REU Program; Cornell NanoScale Science and Technology Facility; Dr� Edwin Kan; Adrian Tung; Rob Ilic and Melanie-Claire Mallison, CNF Program Coordinators; Dr� Lynn Rathbun; CNF staff�

Figure 4: Here we see our calibration process flow as well as our future work process flow.

Figure 3: We used SEM and optical microscopy for color reference and final chemistry.

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Fabrication of Organic Transistors Using Inkjet Printing

Michelle PillersChemistry, Southern Methodist University

NNIN iREU Site: Centre Microélectronique de Provence, Ecole Nationale Supérieure des Mines de Saint Etienne, FranceNNIN iREU Principal Investigator: Professor George Malliaras, Biological Electronics Lab, Centre Microélectronique de ProvenceNNIN iREU Mentor: Dr. Sébastien Sanaur, Biological Electronics Laboratory, Centre Microélectronique de ProvenceContact: [email protected], [email protected], [email protected]


Inkjet printing is being explored as an alternative technique for device fabrication� This cost effective method is a direct-write, non-contact process that avoids the use of masks, excessive chemical treatments, and cumbersome fabrication steps� In this project, a Dimatix Materials Printer was used; this printer utilizes user-fillable piezocrystal-based jetting cartridges to jet specifically formulated inks to print a predesigned pattern� The jetting of the inks is achieved by the application of a voltage� The voltage pulses cause mechanical deformation (contraction or expansion) of the piezocrystals, resulting in jetting of ink for printing�

This project focused on the use of the Dimatix Materals Printer for fabrication of biocompatible organic electro-chemical transistors (OECTs) for use as biosensors� The project was separated into two phases�

The first phase of the project focused on the optimization of printing parameters for successful printing of three specifically formulated inks: silver nanoparticle ink, polyimide (PI) ink, and poly(3,4-ethylenedioxythiophene)- poly(4-styrenesulfonate) (PEDOT:PSS) ink� The parameters included drop spacing (the distance between the center of each printed drop), waveform (the pattern of electrical stimulation of the piezocrystal that causes ink jetting), and substrate temperature�

The second phase of the experiment consisted of the design and fabrication of two types of OECTs: a multilayered OECT incorporating the three optimized inks and a planar all-PEDOT:PSS OECT� All designs and devices were printed on Parylene-C-coated glass slides� Parylene-C is a biocompatible polymer that can be peeled off of the slide to make the device flexible and more easily used as a biosensor.

PEDOT:PSS was the active material for both OECT designs� PEDOT:PSS is a degenerately doped p-type semiconductor [1]� The design for the OECTs, seen in Figure 1, utilized a source and drain electrode and a channel where the electrolyte being tested was placed� The gate voltage was applied through the electrolyte� When a positive gate voltage was applied, positive ions from the electrolyte entered the conducting polymer film and changed its hole concentration�

The decrease of holes in the PEDOT:PSS is called de-doping and, when it occurs, the source-drain current decreases� Then, as the gate voltage is increased, the source-drain current is decreased� If a small change in gate voltage causes a large change in source-drain current, then the device is a good ion-to-electron convertor and a suitable transducer for biosensors�

Experimental Procedure:To identify the optimal printing parameters for each ink, a test pattern consisting of horizontal lines, vertical lines, and dots of various sizes was printed on Parylene-C glass slides with varying parameters� Patterns were printed with drop spacing between 12 to 35 µm, and substrate temperatures between 28°C and 40°C�

When the optimal printing parameters were identified, the OECTs were printed� The multilayered OECT was printed in three steps� First, two layers of silver nanoparticle ink were printed; the ink was annealed in an oven for 2�5 hours at 130°C� Second, three layers of polyimide ink were printed; the ink was dried on a hotplate for five minutes at 80°C and annealed at 150°C in an oven for an hour� Finally, several

Figure 1: A. Pattern for multilayered OECTs using source (S) and drain (D) silver nanoparticle electrodes, polyimide passive layer, and a PEDOT active site. B. Pattern for all-PEDOT OECT.

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


layers of PEDOT:PSS ink were printed and annealed for five minutes on a hotplate at 120°C.

The planar all-PEDOT:PSS OECT was printed in one step with four layers� The PEDOT:PSS was then annealed using a hot plate at 150°C for two hours�

Electrical testing was performed on each device using probe station and phosphate buffered saline (PBS) as the electrolyte�

Results and Conclusions:It was found that the PEDOT:PSS ink would not spread adequately on the Parylene-C surface to print a successful pattern; the Parylene-C was too hydrophobic to allow for the ink to spread� To make the Parylene-C more hydro philic, each slide was treated with a 30 second O2 plasma treatment� With this, the surface energy of the Parylene-C was adequately changed to allow for printing of PEDOT:PSS�

The optimized parameters for polyimide and silver ink could be varied significantly and still print successfully; however, a drop spacing of 23 µm and a substrate temperature of 35°C gave the most accurate printed pattern for both inks� PEDOT:PSS, when printed on plasma treated Parylene-C, was most effectively printed at a drop spacing of 23 µm and a substrate temperature of 28°C�

Both OECT designs were successful transistors� However, due to inadequate annealing of the silver nanoparticle ink in

the multilayered device, the OECT failed in electrolyte and detailed electrical characterization could not be performed� The picture in Figure 2 shows the degradation of the silver nanoparticle electrode after exposure to electrolyte� Nevertheless, the all-PEDOT:PSS OECT was successfully tested using PBS electrolyte and a probe station with a Ag/AgCl gate electrode� According to the data in Figure 3, as the gate voltage was increased, the drain current was decreased�

In conclusion, all-PEDOT:PSS OECT were fabricated using inkjet printing techniques on a biocompatible substrate (Parylene-C)� It showed regular ion-to-electron converter behavior and is suitable for biosensing applications�

Acknowledgements:NSF, National Nanotechnology Infrastructure Network International Research Experience for Undergraduates (NNIN iREU) Program, EMSE, Professor George Malliaras, Associate Professor Sébastien Sanaur, my research partner Jaqueline Mandelli, my travel partner Fiona O’Connell, and all of the lab members of the fantastic BEL in Gardanne�

References:[1] Daniel A� Bernards and George Malliaras� “Steady State and

Transient Behavior of Organic Electrochemical Transistors�” Advanced Functional Materials� 2007�

Figure 3: Current vs. voltage graph for all-PEDOT OECT using an Ag-AgCl gate electrode. The gate voltage (Vg) is applied between 0.0V and 0.4 V at 0.1V intervals.

Figure 2: The silver nanoparticle electrode on the multilayered OECT after exposure to electrolyte and degrading. The silver nanoparticle ink has been removed between the PEDOT:PSS layer (left) and the polyimide layer (right).

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Characterization of Ion Sensitive Field Effect Transistors for Cellular Scale pH Measurement

Kendall PletcherBioengineering, Franklin W. Olin College of Engineering

NNIN REU Site: ASU NanoFab, Arizona State University, Tempe, AZNNIN REU Principal Investigator: Jennifer Blain Christen, Electrical, Computer and Energy Engineering, Arizona State UniversityNNIN REU Mentor: David Welch, School of Biological and Health Systems Engineering, Arizona State University

(2005 NNIN REU at Pennsylvania State University)Contact: [email protected], [email protected], [email protected]

Abstract:

Commercially-available ion-sensitive field effect transistors (ISFETs) are ideal for use as potential of hydrogen (pH) sensors in space- and material-limited applications. However, drift continues to be a major obstacle to precision and accuracy in ISFET measurements. Though it may not be possible to entirely eliminate drift in currently manufactured ISFETs, techniques have been developed that may aid in the prediction and minimization of its effects. In this work, a unique approach was used where relays were integrated into the ISFET control circuit in order to determine the effect of electric field change on drift. Drift characterization was performed by varying the pH of the solution to which the five identical ISFETs were exposed and the frequency at which the relays were switched on and off. A correlation was found between the pH of the solution and the rate of drift. Switching off was found to decrease drift in the subsequent on period.

Introduction:Good cell culture practice is essential to the fabrication of many of the products of biotechnology� A variety of factors influence the growth and health of cells in culture including temperature, nutrient concentration, oxygen levels, and pH� Consequently, pH is a parameter which should be optimized in culture for favorable proliferation [1]� Additionally, the measurement of pH can communicate valuable information about growth and cell death in culture [2]�

ISFETs are a type of field effect transistor whose conduct-ivity is regulated, not by a gate electrode, but by the ion con centration of the solution in direct contact with the ion-sensitive substrate� As a result, ISFETs are sensitive to hydrogen ion concentration, and the resulting current through the transistor can be used to measure pH levels in micro-scale devices�

A major drawback to ISFET use in practice is the incidence of drift, or change in current despite a constant pH, which lowers the accuracy and precision of the measurements taken by the device� However, drift is not random noise but is in fact a result of the movement of mobile ions in solution which neutralize dangling charges along the oxide layer or diffuse into the oxide layer itself� Consequently, drift may be minimized and predicted� The purpose of this body of work is to answer two questions: “Can we characterize drift?” and “What effect does modulating electric field have on drift?”

Figure 1: The experimental circuit featuring SPST Reed Relays with inset showing the switching cycles for the five ISFETs.

Methods:Data was taken from multiple ISFETs on the commercially-available Bionas metabolic chip SC1000� The chip was integrated into a circuit incorporating SPST Reed Relays, shown in Figure 1� To determine the current through each of the five ISFETs, a data acquisition module was employed, taking voltage measurements over every ISFET’s corresponding resistor� The reference electrode in the solution was set to -2V while the switches were set to “on�”

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


Custom MATLAB® code controlled data collection, switch-ing each ISFET at a different frequency over three hours (see Figure 1)� In order to answer the question of whether or not a change in electric field could affect drift, the ISFET switching experiment was repeated, employing pH buffers with values 4, 7, 10 and 12�

Analysis of the collected data was performed via curve-fitting to the linear region of each “on” region, neglecting any non-linear behavior at the start of the run, as is accepted practice when dealing with ISFETs� The slope of the line of best fit gives an approximation of drift over time in amps per second�

indicating that multiple ISFETs and a switching pattern could be employed to create a system with measurably lower drift�

Acknowledgements:I would like to thank my PI, Dr� Jennifer Blain Christen, and my mentor, David Welch, for their help and guidance throughout the summer� Thanks also to the National Nanotechnology Infrastructure Network Research Experience for Undergraduates, the National Science Foundation, and the Center for Solid State Electronics Research at Arizona State University for research support and funding�

References:[1] Hanson, M�A�, X� Ge, Y� Kostov, K� A� Brorson, A� R� Moreira, and

G� Rao, “Comparisons of optical pH and dissolved oxygen sensors with traditional electrochemical probes during mammalian cell culture,” Biotechnol Bioeng, vol� 9, pp� 833-841, 2007�

[2] Perani, A�, R� P� Singh, R� Chauhan, and M� Al-Rubeai, “Variable functions of bcl-2 in mediating bioreactor stress-induced apoptosis in hybridoma cells,” Cytotechnology, vol� 28, pp� 177-188, 1998�

Figure 3, top: Average difference in drift after the first switch. The diagram illustrates the process by which drift was determined for ISFET C.

Figure 4, bottom: Average difference in drift after each switch. Note the decreasing slopes in each successive set of data.

Figure 2: Correlation between pH and drift with average values in the table inset.

Results:A correlation was found between average drift as measured by the slope of the line of best fit and the pH of solution, as shown in Figure 2� Additionally, drift was found to decrease, on average, following the first switch in ISFET C, as shown in Figure 3� For ISFET D, the same trend was found to be true of the first “on” period following a switch; however, the results following the second switch were inconclusive� For trials with pH 4 and 12, the average drift of the third “on” period was less than the second while, for trials with pH 7 and 10, the average drift of the third “on” period was greater than the second, as illustrated in Figure 4�

Conclusions:Further work is needed in this area; in order to conclusively answer the question of whether a drift-decreasing trend continues past the first switch in ISFET D, a larger data set is needed to decrease the possibility of experimental error being responsible for inconsistencies in results� However, a correlation between pH and drift was found, indicating that prediction and subtraction of drift is possible with further work� In addition, switching was found to decrease drift,

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Growth and Characterization of Graphene for Use in Nanoelectronics

Bethany RobinsonElectrical Engineering, Howard University

NNIN REU Site: Microelectronics Research Center, The University of Texas, Austin, TXNNIN REU Principal Investigator: Deji Akinwande, Electrical and Computer Engineering, University of Texas at AustinNNIN REU Mentor: Li Tao, Electrical Engineering, University of Texas at AustinContact: [email protected], [email protected], [email protected]

Introduction:

Graphene, an atomic layer allotrope of sp2-bonded carbon, has gained the attention of the researchers in electronics and materials science� This is due to graphene’s unique potential to enable high speed in future nanoelectronics, such as graphene field effect transistors (GFETs).

The current challenge with moving forward in the creation of these devices is the inability to create large-area graphene with low-defect density� The focus of this work was to investigate a reliable protocol to produce large area high-quality graphene for device application� Chemical vapor deposition (CVD) was employed to synthesize graphene on copper substrates� Certain parameters of the CVD were varied in an effort to find an optimized, more efficient protocol for graphene synthesis�

Methods:The synthesis of graphene is performed in an Aixtron® Black Magic, a commercially produced CVD apparatus� Our state-of-the-art Black Magic CVD allows computer-based control (a standard set of commands referred to as recipe) of key parameters, including temperature, pressure, gas flow and time, for the synthesis of carbon nanomaterials in a vertical cold-wall chamber� The carbon source for CVD graphene is hydrocarbon gas which is decomposed on a 25 µm thick copper foil substrate that was cleaned by ultra-sonication in an acetone bath for 10 minutes�

The samples were loaded into the CVD chamber on a graphite stage, which served as the substrate heater to heat the samples together with a showerhead heater about 2-3 inches above� Using a heat ramp up to 1000°C, annealing was done for five minutes in hydrogen (H2) at a flow rate of 1000 sccm� After annealing, graphene synthesis occurred� Pure methane (CH4) was used as the processing gas for

graphene synthesis and kept on at a flow rate of 10 standard cubic centimeters per minute (sccm) for five minutes. After cooling down to 180°C, the sample was pulled from the chamber and stored in a dessicator, until characterized using Raman spectroscopy�

For this project, certain variables of the CVD parameters were manipulated to see if the recipe for obtaining graphene film could be optimized. Specifically, the amount time in which the substrate was exposed to methane

was varied to 3, 4, 6, 7, and 9 minutes� The amount of annealing time in conjunction with the flow rate of the annealing gas, hydrogen, was also varied to times of 3, 5, 7, and 10 minutes and the flow rate was varied to 10 sccm, 500 sccm, and 1000 sccm�

Results and Conclusions:Raman spectroscopy was used to analyze the synthesized graphene� There are several characteristics in a Raman spectrum that help determine the quality and thickness of the graphene [1]: full width at half maximum (FWHM) of 2D peak (2D Peak occurs between 2720 and 2750 cm-1), ratio of 2D peak to G peak (G Peak occurs around 1600 cm-1), and D peak (represents defect in sample, D Peak occurs around 1400 cm-1)�

Our goal was to get monolayer graphene with the least or no defects, which required a 2D width of 25-30 cm-1, 2D/G > 3 and the absence of a D peak� According to the data obtained from the Raman spectroscopy, there appeared to be a trend in the graphs showing the variation of growth time in methane� There was a direct correlation between time exposed to the growth gas and the quality of the graphene that was synthesized� The longer the copper substrate was exposed to the methane, the lower the defects in the graphene� In the runs where the annealing time and

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


the flow rate of the annealing gas were varied, it appeared that the higher the flow rate of the annealing gas, the better the quality of the graphene� This could be attributed to the adsorption of hydrogen on the copper catalyst, which could help decompose methane for the growth of graphene�

In a brief summary, we found that high annealing flow rates and longer times for both annealing and growth would yield the best results. The underlying scientific explanation of this optimized recipe needs further investigation, which in future studies will be in the scope of this project�

Using the results obtained from these syntheses, it can be concluded that each of the variables that were changed and tested in this project had a significant impact on the outcome of the graphene� The ability to create good graphene is a crucial step in the fabrication of graphene field effect transistors (GFETs) in a large scale�

Acknowledgements:I would like to thank the National Science Foundation, the NNIN REU Program, and the Microelectronics Research Center at the University of Texas at Austin for the opportunity to participate in this program� I would also like to thank my Principal Investigator, Dr� Deji Akinwande, for the opportunity to work with his group, and my mentor, Dr� Li Tao, for his assistance throughout my project�

References:[1] Ferrari, A� C�, J� C� Meyer, V� Scardaci, C� Casiraghi, M� Lazzeri,

F� Mauri, S� Piscanec, D� Jiang, K� S� Novoselov, S� Roth, and A� K� Geim� “Raman Spectrum of Graphene and Graphene Layers�” Physical Review Letters 97�18 (2006)�

[2] Geim, A� K�, and K� S� Novoselov� “The Rise of Graphene�” Nature Materials 6�3 (2007): 183-91�

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Fabrication and Characterization of ZnO Nanowire Field-Effect Transistors and ZnTe Nanosheet Field-Effect Transistors

William ScheidelerElectrical Engineering, Biomedical Engineering, Duke University

NNIN REU Site: ASU NanoFab, Arizona State University, Tempe, AZNNIN REU Principal Investigator: Prof. Hongbin Yu, Electrical, Computer, and Energy Engineering, Arizona State UniversityNNIN REU Mentor: Ebraheem Ali Azhar, School of Electrical, Computer, and Energy Engineering, Arizona State UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Zinc oxide (ZnO) is a promising material for nanodevices due to its high mobility and robust mechanical properties. The transparent nature of ZnO makes it useful for transparent and flexible electronics. Zinc telluride (ZnTe) is an easily doped direct bandgap material for optoelectronic devices. Together, ZnO and ZnTe may form a p-n junction, allowing the fabrication of essential electrical components such as LED’s. The first focus of this project was to fabricate ZnO transistors to study the effects of low temperature fabrication process. Low temperature oxides were grown by remote plasma-enhanced chemical vapor deposition (RPEVCD) and plasma enhanced chemical vapor deposition (PECVD) for comparison with devices on thermally grown oxides. The second focus was fabricating ZnTe field-effect transistors to understand the electrical properties of nanosheets.

Introduction:

Figure 1: FESEM image of a zinc oxide nanowire transistor.

Transparent and flexible electronics have a number of exciting applications in a new generation of displays, mobile devices, and sensors� Current transparent technology uses organic materials with low mobility� ZnO is an attractive material to improve transparent and flexible technology as it is transparent, physically robust, and has high mobility to support high performance electronic applications� ZnO nanowire transistors are attractive for their high mobility and sensitivity to gas adsorption and the ultraviolet (UV) spectrum� ZnTe is an easily doped II-VI material that may potentially form a p-n junction with ZnO� However, realizing

a II-VI p-n junction device requires greater knowledge of ZnTe electrical properties� This work investigated ZnTe electrical properties and low temperature ZnO transistor fabrication�

Experimental Procedure:ZnO back-gated nanowire field-effect transistors (Figure 1) were first fabricated on 200 nm thermally grown silicon oxide (SiO2)� ZnO nanowires were grown by chemical vapor deposition (CVD) in a tube furnace with source material composed of a 1:1 ZnO and graphite mixture� Growth substrates were placed at approximately 800°C and source material was heated to 1100°C� Argon carrier gas and O2 gas were flowed to enable carbothermal reduction of ZnO.

Growth substrates were characterized by field-emission scanning electron microscopy (FESEM), showing nanowires 5-10 µm in length with a characteristic wurtzite structure� Nanowires were transferred by mechanical slide transfer to a Si substrate with 200 nm thermally grown SiO2. Photolithography was used to define contacts on the nanowires, and 250 nm of gold over 10 nm of chromium was deposited by thermal evaporation� ZnO nanowire current voltage characteristics were obtained with an electrical probe system� A UV lamp and meter were used to characterize the UV response of the ZnO field-effect transistors�

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


ZnTe nanosheet transistors were fabricated by similar methods, but sheets were grown at lower temperatures (approximately 600°C) with ZnTe source material� Sheets were characterized by FESEM and atomic force microscopy (AFM) showing nanosheet side lengths of 20-40 µm and a thickness of approximately 500 nm� Transfer and photolithography were performed in the same fashion as with ZnO� ZnTe nanosheet devices were characterized with an electrical probe system to determine mobility and On/Off ratios�

ZnO nanowire transistors were also fabricated on transparent glass and polyethylene napthalate (PEN) substrates� Indium tin oxide (ITO) was deposited by electron beam evaporation on silicon substrates and annealed at 135°C for eight hours in air� Low temperature oxides were grown by PECVD at 100°C and by RPECVD at 150°C� Oxides were characterized by a mercury probe system prior to transistor fabrication and nanowires were then transferred by mechanical slide transfer. Contacts were defined by photolithography and exposures were calibrated for substrate transparency� Cr/Au contacts were deposited by thermal evaporation, but in the future we hope to make the devices fully transparent using sputtered ITO contacts�

Results and Conclusions:ZnO transistors on thermally-grown oxides were used to study the transport behavior of ZnO nanowires (Figure 2)� Current-voltage characteristics showed strong n-type behavior� The cylinder-on-plate model was used to calculate gate capacitance, and a polynomial fit with 50 points was used to derive the transconductance of devices� A carrier mobility of 11 cm2/V·s was calculated for the ZnO nanowire devices on thermally-grown oxides and On/Off ratios were approximately 105� Both of these parameters are comparable to other reported NWFET’s [1]� The UV photoresponse of ZnO was studied by comparing transfer characteristics at variable intensities of UV radiation (Figure 3)� ZnO showed a strong UV response due to its wide bandgap, as incident photons caused electron-hole pairs to form� Nanowire

surface states caused the trapping of holes, providing a large photoconductive gain of 1�1 × 106�

ZnTe nanosheet transistors on thermally-grown oxide substrates also showed good performance as transistors� ZnTe nanosheet transfer characteristics showed predominately p-type behavior (Figure 4)� Similar methods were used to calculate transconductance and gate capacitance of nanosheet devices, yielding mobility of 246 cm2/V·s and a lower On/Off ratio of 103�

Successful transistor behavior was not achieved by low temperature fabrication methods� While low temperature oxides on Si substrates yielded tolerable C-V characteristics, transparent devices with 200 nm and 300 nm SiO2 layers were found to short circuit through the gate insulator� We expect that either the inherent surface roughness of the PEN substrates or the roughness of the electron beam deposited ITO conductive back gate is the reason for this persistent problem�

Future Work:Future work will seek to improve low temperature deposited oxide layers with low temperature annealing processes� We hope to characterize the electron beam deposited ITO conductive layer with AFM and learn how to mitigate pin-hole causing surface roughness� We hope that answering these questions will lead to flexible, transparent devices by more conventional methods than have currently been achieved�

Acknowledgments:Prof� Hongbin Yu, Ali Azhar, Dr� Trevor Thornton, NNIN REU Program, ASU Center for Solid State Electronics Research, NSF�

References:[1] Chang, Pai-Chun, and Jia Grace Lu� “ZnO Nanowire Field-Effect

Transistors�” IEEE Transactions on Electron Devices 55, no� 11: 2977-2987 (November 2008)�

Figure 2: Zinc oxide source-drain current vs. gate bias.

Figure 3: ZnO UV response.

Figure 4: ZnTe source-drain current vs. gate voltage.

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Synthesis of Few-Layer Graphene Films of Large Lateral Dimensions

Claire SpradlingBiological Engineering, University of Missouri-Columbia

NNIN REU Site: Microelectronics Research Center, The University of Texas, Austin, TXNNIN REU Principal Investigator: Rodney S. Ruoff, Mechanical Engineering, University of Texas at AustinNNIN REU Mentor: Carl W. Magnuson, Materials Science and Engineering, University of Texas at AustinContact: [email protected], [email protected], [email protected]

Abstract:

Graphene is of great interest to many in the scientific and technological communities due to its fundamental properties and possible applications. While methods to synthesize graphene are being discovered and improved, the challenge of making graphene in large lateral dimensions of the same high quality as small mechanically exfoliated flakes is ongoing. Although graphene can currently be produced in various commercial chemical vapor deposition (CVD) machines, such as those used for synthesis from silicon carbide, these are costly. Less expensive machines are an attractive alternative. Members of our group assembled a thermal CVD system to grow graphene, hoping for high quality and large lateral dimensions for fundamental science and with a new design. This paper briefly presents this new system.

strength), extremely high thermal conductivity even at room temperature, and the high electron and hole carrier mobilities also at room temperature [2, 3]� Figure 1 shows an image of graphene and a cartoon showing how other materials such as fullerenes or nanotubes could be conceptually “cut out” from it; in reality nanotubes and fullerenes would never be made in quantity from graphene precursors [4]�

Many researchers have and are still trying to create and improve the methods by which graphene is synthesized� One technique is thermal CVD. Gas precursors flow into a reactor where they can collide with a substrate and some fraction decompose yielding surface diffusing species that can then incorporate into solid films, such as graphene.

Experimental Procedure:While the Ruoff lab already has small thermal CVD systems, they are more routine and are in fact not home-built with optimizing graphene growth in mind, other than through relatively simple modifications. Figure 2 shows a schematic diagram of the CVD system that our group worked on this summer� The system has a manifold for use of six different gases to grow graphene on a copper substrate�

To ensure that the external environment could not penetrate our system, we used flanges with copper gaskets to connect each piece of equipment. The flanges had a knife edge on each side that, when tightened, cut into the copper gasket and created a high vacuum seal� We also used gate valves

Figure 1: This image shows graphene and cartoons of how it conceptually can be used to “cookie cutter” out nanotubes and fullerenes. However, it cannot be a “base structure” for cookie-cutting out diamond, or negative curvature carbon, or the myriad amorphous carbons that are a mixture of sp2- and sp3-bonded carbon [5].

Introduction:Graphene, a two-dimensional (2D) material, consists of carbon atoms that are sp2-bonded to create a single layer film [1]. It is of interest due to its mechanical properties (high elastic modulus and, in defect free areas, high

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


that, when open, allowed both the most gas flow and large area samples to pass through� In some CVD systems, turbo pumps are installed at an angle to the main system, and, therefore, molecules must bounce off the system wall at the correct angle in order to enter the pump� We installed our pump with line-of-sight of the growth chamber�

The system was computer-controlled using LabView programs in order to give higher accuracy and repeatability in controlling growth conditions; we programmed control gas flow, system pressure, valve positions, and the temperature at which the system was being operated� The programs were designed so that a user could specify times to change the growth conditions�

The system had a four-inch quartz tube instead of a one-inch quartz tube (as our other thermal CVD systems have) so that larger samples could be inserted� By using a larger four-inch tube, we would be able to insert three-inch wafers�

By assembling this CVD system on our own, we avoided the high cost of ordering a commercial system�

Future Work:We have almost finished assembly, but have not yet grown a graphene sample� I look forward to hearing from the team about whether or not they get high quality graphene over large areas in the future�

Figure 2: A schematic of the thermal CVD system that members of our lab are assembling.

Acknowledgements:Thanks to Rodney S� Ruoff, Carl W� Magnuson, the Ruoff Group, Jean Toll, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, and the National Science Foundation� Special thanks to Rodney S� Ruoff for extensive editing�

References:[1] Xuesong Li, Weiwei Cai, Jinho An, Seyoung Kim, Junghyo Nah,

Dongxing Yang, Richard Piner, Aruna Velamakanni, Inhwa Jung, Emanuel Tutuc, Sanjay K� Banerjee, Luigi Colombo, Rodney S� Ruoff; “Large-Area Synthesis of High Quality and Uniform Graphene Films on Copper Foils”; Science, 324, 1312-1314 (2009)�

[2] Dreyer, Daniel R�; Ruoff, Rodney S�; Bielawski, Christopher W� From Conception to Realization: An Historical Account of Graphene and Some Perspectives for Its Future� Angewandte Chemie International Edition, (2010), 49, 9336-9344�

[3] Zhu, Yanwu; Murali, Shanthi; Cai, Weiwei; Li, Xuesong; Suk, Ji Won; Potts, Jeffrey R�; Ruoff, Rodney S� Graphene and Graphene Oxide: Synthesis, Properties, and Applications� Advanced Materials (2010), 22, 3906-3924�

[4] Geim, A�K�; “Graphene Status and Prospects”; Science, 324, 1530-1524 (2009)�

[5] Geim, A�K�, Novoselov, K�S�; “The Rise of Graphene”; Nature Materials, 6, 183-191 (2007)�

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Fabrication of Flexible Organic Thin-Film Transistors

Kevin TienElectrical Engineering, The Cooper Union for the Advancement of Science and Art

NNIN REU Site: Stanford Nanofabrication Facility, Stanford University, Stanford, CANNIN REU Principal Investigator: Zhenan Bao, Chemical Engineering, Stanford UniversityNNIN REU Mentor: Benjamin C.K. Tee, Electrical Engineering, Stanford UniversityContact: [email protected], [email protected], [email protected]

Abstract:

In this work, a process for manufacturing thin-film transistors on a polyimide substrate was explored, and a novel, fast method is presented for the micro-machining of stencil masks for material deposition, with minimum opening size of 20 µm. Both n- and p-type transistors were fabricated, with the p-type transistors demonstrating saturation mobilities on the order of 1 cm2/V·s, and the n-type transistors demonstrating saturation mobilities on the order of 0.1 cm2/V·s. These devices can therefore be readily integrated into flexible circuits, further enabling advances in relevant technologies.

Introduction:Several flexible materials are well-established as viable active materials for thin-film devices, such as amorphous silicon (a-Si), polycrystalline silicon, and organic materials� Organic materials are especially attractive for flexible electronics, as they have low processing temperatures that allow for compatibility with a wider range of substrates relative to other flexible materials. Organic electronics also have a demonstrated potential for low-cost mass manufacture, for instance in the form of ink-jet printing using solution-processed semiconductor materials [1]�

The transistor and capacitor are two devices central to the operation of many electronic circuits. In flexible circuits specifically, the classic thin-film transistor (TFT) architecture is employed� Both capacitors and TFTs require a high-quality dielectric for successful fabrication: pinhole shorts in the dielectric render the device useless, and non-uniformity of the dielectric layer negatively impacts performance� Thus, the process for dielectric fabrication must be chosen with care�

Experimental Procedure:Process. The thin-film architecture employed was the traditional bottom-gate top-contact type� The fabrication process roughly followed the following procedure:

1� Spin-coat polyimide on silicon wafer as substrate2� Deposit aluminium (Al) using stencil masked

evaporation as gate layer3� Deposit Al oxide using atomic layer deposition

(ALD) as dielectric layer4� Deposit organic semiconductor materials using

stencil masked evaporation as active layer5� Deposit gold using stencil masked evaporation as

interconnection layer

Stencil Masks. The minimum opening size available in the stencil masks used determined the minimum channel length available for the transistor devices fabricated, which directly impacted the maximum speed at which the device could operate� A metric for the upper limit of operation was the transition frequency fT, given in Figure 1� Its inverse-square dependence on length implies that small channel lengths improve the operational frequency of the device�

Mid-cost laser-cut stencil masks can achieve minimum openings around 50 µm� Micro-machined Si wafer masks can achieve opening sizes smaller than those of laser-cut masks, and thus we explored the use of deep reactive ion etching using the Bosch process to etch through Si wafers for a target minimum feature size of 20 µm� This process had a significant time advantage, as fabrication could be done in-house within a day, given the availability of a transparency mask for the photolithographic step� Externally

Figure 1: Definition of transition frequency.

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


manufactured transparency masks have a turnaround time of 1-2 days, compared to externally manufactured laser-cut stencil masks, which may take upwards of three or four weeks to obtain�

Fabricated Devices. Using the above fabrication process, n- and p-channel TFTs were fabricated, with minimum channel lengths of 20 µm� The fabricated devices are pictured in Figure 2, and the measured Vgs-Ids data for the devices are presented in Figure 3� The p-channel devices were fabricated using pentacene as the active material, and demonstrated saturation mobilities on the order of 1 cm2/V·s� The n-channel devices were fabricated using a perylene tetra-carboxylic di-imide (PTCDI) derivative and demonstrated saturation mobilities on the order of 0�1 cm2/V·s� These values agreed with accepted values found in literature [2]�

The attained channel lengths were measured to be around 20 µm, thus validating the use of the micro-machined stencil masks� It was found that from photolithography to completed mask, the fabrication of the stencil masks took roughly five hours given the equipment available to us.

SPICE Modeling. Level 62 SPICE models for the TFTs were generated using an HSPICE optimization deck� The model Ids-Vgs curves are presented in Figure 3� The close agreement of the models to the data support their use as models for complex circuits and allows for an initial design step before fabrication. As an example, a five-stage ring oscillator was simulated using the generated models, and the effect of channel length on frequency is demonstrated in Figure 4� A frequency of 1 kHz can be achieved at low channel lengths, highlighting the importance of small device length�

Conclusions:We have developed a process for making both n- and p-channel thin-film transistors on a flexible substrate, polyimide� These devices exhibit performances comparable to that reported in literature� The minimum device size of these devices was approximately 20 µm, enabled by micro-machined silicon masks� The simplicity of the process as well as the fast turnaround time of the masks allowed

for reasonably fast fabrication in an academic setting� Because both n- and p-channel devices were available, this process can be used to fabricate digital circuits using a complementary design� Future work will involve fabrication of digital integrated circuits using this process for mounting on curved surfaces�

Acknowledgements:We would like to thank Zhenan Bao for her role as principal investigator, Benjamin C�K� Tee for his role as mentor, and the Bao research group for their overall support� We thank the Goldhaber-Gordon group for their assistance with ALD, and J� Provine for his assistance with micro-machining� We thank the staff of the Stanford Nanofabrication Facility for all of their assistance during fabrication, and Michael Deal for his role as site coordinator� Finally, we thank NNIN REU Program, NSF, and the Stanford Nanofabrication Facility for their support�

References:[1] Minemawari, H�, Yamada, T�, et al� “Inkjet printing of single-crystal

films”; Nature 475, 365-367 (2011).[2] Bao, Z� and Locklin, J�� Organic FETs� CRC Press, 2007�

Figure 3, above: Vgs-Ids curves for fabricated devices and for constructed device models.

Figure 4, below: Simulated frequency of five-stage ring oscillator as function of device channel length.

Figure 2: Fabricated devices on flexible substrate.

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

• e

lect

ronic

s

•


Electronic Graphene Devices through Tip-Based Nanotechnology

Cassandra ToddElectrical Engineering, University of Central Florida

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Prof. Amit Lal, Electrical Engineering, Cornell UniversityNNIN REU Mentor: Hadi Hosseinzadegan, Electrical Engineering, Cornell UniversityContact: [email protected], [email protected], [email protected]

Abstract and Introduction:In order to test the electromechanical properties of graphene for this project, it was necessary to build a device that would actuate a graphene film mounted on a clamped-clamped thin-film plate of silicon nitride (SiN), which was then mechanically actuated using a piezoelectric actuator (PZT)� Our device was fabricated by coating Si with a 400 nm layer of low stress nitride; the backside of the nitride was patterned and etched so that the Si layer could be completely etched by potassium hydroxide (KOH) to form nitride membranes on the front of the device� A 150 nm bi-layer of gold/chromium (Au/Cr) electrodes was deposited onto the front of the device to form a four-point probe measurement system� Bond wires were attached to electrodes to interface the device to a DIP package� Graphene was transferred on top of the device, so that its electrical conductivity and piezoresistivity could be measured�

The piezoresistivity tensor elements of graphene could be extracted by putting the 4-point probes on each side of the membrane� The elements were tensor because the piezoresistivity might be different, in different directions, due to a graphene samples’ lack of directional uniformity� Using an optical interferometer, the vibration profile of the graphene/SixNy films, mode shapes and amplitude could be measured as well� With this knowledge, graphene could be applied to microelectromechanical systems (MEMS) for smaller and more sensitive devices�

Despite the increase in publications on graphene, there is still not much known about its electromechanical properties� By having a better understanding of the tensor elements of graphene it could be applied to various MEMS such as resonators, transistors, and cantilevers�

Figure 1: An illustration of the devices structure with example values for thicknesses.

Experimental Procedure:Device fabrication consisted of nine steps�

1� About 400 nm of low stress nitride was deposited onto our Si wafers with low-pressure chemical vapor deposition (CVD)�

2� Contact lithography was used to pattern the topside of the wafer in order to deposit a 150 nm bi-layer of Au/Cr electrodes�

3� Once the resist and excess metal were lifted off with 1165, the backside of the wafer was patterned and the nitride was etched to form windows in the bare Si�

4� Si was then completely etched with KOH to create nitride membranes ranging in size from 1500 µm to 50 µm squares� At this stage, the graphene was ready to be transferred onto our wafers that we diced into chips to make the process easier�

5� We used CVD-grown graphene that was deposited onto a copper (Cu) film and had poly(methyl methacrylate) (PMMA) spun onto it to make the graphene visible for transfer� A wet etch was used to remove the Cu from the graphene, and the graphene was then moved to de-ionized (DI) water to be repeatedly cleaned before it was transferred to our devices� The graphene placement was relatively easy as it was simply scooped out of the water onto a device and then left to air dry�

6� After the PMMA was removed with acetone, resist was spun onto the graphene and, using contact lithography, the graphene was patterned and then,

7� oxygen plasma etched so that only a 30 µm by 10 µm rectangle was left at the ends of our probes for our measurements�

8� A PZT was then attached to the backside of the devices followed by:

9� metal wire bonding leads, which were connected to the contact pads�

See Figure 1 for the complete layout of the design�

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

• ele

ctronics

•


Our metal electrodes were designed specifically for a four-point probe measurement as seen in Figure 2� The pads were 200 µm squares� The actual probes extended to the center of each membrane at a final width and gap spacing of 1 µm� The PZT was connected to a function generator� We generated sine waves normally around 100 mV-500mV peak-to-peak voltage� The two outermost wire-bonded probes were connected to a current source of ~ 1 nA and the inner probes were connected to the voltmeter� To measure the amplitude at each point of a cell, the devices were placed in a vacuum under an interferometer�

Results and Conclusions:By sweeping the frequency of the function generator attached to the PZT, the graph in Figure 3 could be obtained� The higher voltage resulted in a higher conductivity in the material� At certain frequencies, the nitride membrane was bended more, causing a peak in the conductivity� This was caused by the stress in the graphene layer, which decreased its resistivity�

Furthermore, the shape of the nitride membrane was also mapped� In Figure 4, one can see how the nitride membrane and graphene reacted when actuated at a frequency of 403 kHz�

Future Work:At this time, not enough measurements have been completed to come up with a conclusive answer on the piezoresistivity, vibration modes and amplitude of graphene� Once this has been accomplished, the next step of the project will be to create a transistor using graphene by opening up its band gap�

Acknowledgments:Special thanks to Prof� Amit Lal, Hadi Hosseinzadegan, and the SonicMEMS group; Stephen Jones; Mark Levendorf; the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program; the National Science Foundation (NSF); the Cornell NanoScale Facility (CNF); and the CNF staff, especially Rob Ilic and Melanie-Claire Mallison�

References:[1] Norimasa Yoshimizu, “Scanning probe nanoscale patterning of

highly ordered pyrolytic graphite,” February 2010, Nanotechnology V21N9, 21 095306 doi:10�1088/0957-4484/21/9/095306�

[2] K.S. Novoselov, “Electric field effect in atomically thin carbon films,” Science 22 October 2004: Vol. 306 no. 5696 pp. 666-669, DOI: 10�1126/science�1102896�

[3] Xuesong Li, “Large-area synthesis of high-quality and uniform graphene films on copper foils,” Science 5 June 2009: Vol. 324 no. 5932 pp� 1312-1314, DOI: 10�1126/science�1171245�

Figure 2: An illustration of the metal electrode design.

Figure 4: An illustration of the shape of a nitride membrane at 403 kHz. (See the cover for a full color version.)

Figure 3: An analysis of the voltage when the frequency is swept and its relation to the conductivity of graphene.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Characterization of 2-300 nm Alumina Thin Films Deposited via ALD

Malena AgyemangChemistry, Norfolk State University

NNIN REU Site: Nanotechnology Research Center, Georgia Institute of Technology, Atlanta, GANNIN REU Principal Investigator and Mentor: Dr. Rosario A. Gerhardt,

School of Materials Science and Engineering, Georgia Institute of TechnologyContact: [email protected], [email protected]

Abstract:

Thin films are used widely across the engineering field, as electrical semiconductors, optical coatings, in drug delivery systems and in solar cells. Alumina (Al2O3) thin films are the most ideal type of films due to their properties. They are hard and stiff, have good chemical and thermal stability, firm adhesion to many surfaces, a high dielectric constant as well as excellent insulating properties. However, even though these thin films are used frequently, the properties of the thinnest films are assumed to behave the same way as thicker films. The objective of this research is to characterize alumina thin films to give accurate measurements of the electrical properties at the nano-scale. Al2O3 thin films were deposited on silicon wafers via atomic layer deposition at thicknesses ranging from 2-300 nm and ellipsometry was used to determine the films thickness. Then, electrodes of various diameters were deposited on the films and impedance measurements were carried out at different frequencies. From these measurements, resistance and capacitance values were derived as a function of film thickness and electrode size and compared to expected properties. Results are found to depend heavily on the geometric parameters.

Introduction:Thin films are made from layers of materials, ranging from fractions of a nanometer to several micrometers in thickness� There are many ways to deposit thin films. The most efficient method is atomic layer deposition (ALD). ALD is a chemical gas phase thin film deposition method based on alternating surface reactions� As the steps are repeated, one layer is deposited one at a time� This self-limiting growth nature is one of the most unique features of ALD, making it an exclusive thin film growth method differing from other deposition techniques� Compared to other deposition techniques, ALD offers ideal film characteristics like thick-ness uniformity, film density and high interface quality [1]. In addition, it is a low temperature deposition method that has industrial applicability�

Alumina (Al2O3) thin films are ideal films used due to their excellent properties� They are sturdy, have good chemical and thermal stability, firm adhesion to many surfaces, a high dielectric constant and excellent insulating properties [2]� These properties are excellent in microelectromechanical systems (MEMS) devices to decrease surface wear� How-ever, even though these thin films are used frequently, the properties of the thinnest films are assumed to behave the same way as the thicker films. In this context, our goal was to investigate the electrical properties of Al2O3 thin films ranging from 2-300 nm in thickness deposited via ALD�

Experimental Procedure:Silicon (Si) wafers were cleaned and then Al2O3 was deposited on them using a Cambridge NanoTech Fiji 200 ALD at 250°C� Trimethyl aluminum (TMA) and water were used as the precursors with pulse times of 0�06 seconds� The process was repeated for 20, 50, 100, 200, 500, 1000 and 3000 cycles to calculate deposition rate per cycle; first on wafer pieces and then on 4-inch wafers� Film thickness was then measured using the Woollam M-2000 ellipsometer� Silver (Ag) electrodes were deposited on the wafer pieces using the DESK II TSC sputter coater at a pressure of 50 mTorr and 30% power for 999 seconds� Electrical properties were measured using a Solartron 1260 impedance analyzer at frequencies ranging from 107-10 Hz and then compared to expected properties�

Results and Discussion:Figure 1 shows the relationship between the measured im-pedance to the log of frequency for various film thicknesses measured with 3 mm electrodes� It is clear that the thicker films had lower resistances while the thinner films had higher resistances. It is noteworthy that for the thinner films, some frequencies were un-measurable�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

•2011 NNIN REU Research Accomplishments Page 91

Figure 2 summarizes the measured response with the calculated resistance response at a frequency of 0�1 Hz� It is clear that large discrepancies occurred in all cases�

Figure 3 displays the frequency dependence for the capacitance of films with different thicknesses, which were measured with 1 mm size electrodes� It can be seen that the thinner the film is, the higher the capacitance as would be expected based on the C % 1/t response� In Figure 4, it can be seen that the capacitance of films thicker than 50 nm followed the expected values, but for the films with thinner thicknesses, the capacitance values were much smaller than expected�

Conclusions:Al2O3 thin films were successfully deposited onto silicon wafers and wafer pieces, and film thicknesses were accurately determined� Ag electrodes were deposited on the Al2O3 thin films using shadow masks, and impedance measurements were taken� Results suggested discrepancies between the expected and the measured properties, especially for the thinnest films. Discrepancies could be due to possible film defects such as voids in the thinnest films or the electrodes not being good enough or small enough�

Future Work:A more extensive evaluation will be done in the future by acquiring more data on films of various thicknesses and depositing more electrodes of different sizes� Also, it is important to characterize the microstructure of the films using atomic force microscopy to find out if they have any voids or if the trends observed are related to possible quantum effects�

Acknowledgements:This work was supported by the National Nanotechnology Infrastructure Network Research Experience for Under-graduate (NNIN REU) Program� We also acknowledge the Georgia Institute of Technology (GIT) Marcus Nanotechnology Center, and the GIT Department of Materials Science and Engineering for technical support� Special thanks to Dr� Rosario Gerhardt and her lab�

References:[1] D� Callister, William� Materials Science and Engineering: An

Introduction� York, PA: 2007�[2] R� Katamreddy, R� Inman, G� Jursich, A� Soulet, and C� Takoudisb�

“ALD and Characterization of Al2O3 Deposited on Si <100> using Tris(diethylamino) Aluminum and Water Vapor�” Journal of The Electrochemical Society, 153 (10) C701-C706 (2006)�

Figure 1: Impedance measurements of films with 3 mm electrodes.

Figure 2: Comparison of experimental with expected resistance.

Figure 3: Real Capacitance vs. Log of Frequency with 1 mm electrodes.

Figure 4: Comparison of measured capacitance with expected value.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Acceptor State Activation Energy of Tin Monosulfide

Kyle Arean-RainesChemical Engineering, University of Rochester

NNIN REU Site: Center for Nanoscale Systems, Harvard University, Cambridge, MANNIN REU Principal Investigator: Prof. Roy Gordon, Department of Chemistry and Chemical Biology, Harvard UniversityNNIN REU Mentor: Prasert Sinsermsuksakul, Department of Chemistry and Chemical Biology, Harvard University Contact: [email protected],[email protected], [email protected]

Abstract:

The positive (p)-type semiconductor tin monosulfide (SnS) was studied as a potential solar cell absorber material. Temperature-dependent Hall measurements were taken on tin sulfide thin films to determine the film’s resistivity as a function of temperature over a range of 200°K, from which the carrier concentrations at each temperature were obtained. By fitting the carrier concentration data to a model and assuming a single activation energy, an acceptor state activation energy of 0.109 eV was calculated for tin sulfide.

Introduction:Current research in the development of solar cells for clean energy production seeks to mitigate the cost of high-efficiency solar cells resulting from material scarcity. To this end, earth-abundant, non-toxic materials are being studied as an alternative for use in solar cells�

Tin (II) sulfide has attracted considerable attention due to its desirable optical and electrical properties� It is a p-type semiconductor with a direct band gap of 1�30 eV� In order to evaluate the viability of using tin sulfide as an absorber material, it is necessary to study the characteristics of its charge carriers—in this case, holes in the lattice� An important property of a semiconductor material is its donor or acceptor activation energy, or the amount of energy required to dislodge an electron or hole�

Experimental Procedure:Tin sulfide thin films were created by depositing 300-600 nm of SnS on a glass substrate using atomic layer deposition (ALD)� The resistivity data were obtained by performing temperature-dependent Hall measurements over a 200°K temperature range� Starting at 125°K, and in 25°K increments up to 325°K, resistivity, mobility, and carrier density were obtained at each temperature for each of the films.

Results and Conclusions:Since SnS is p-type, charge is carried primarily by holes in the lattice� Hall measurements yielded the density of holes as a function of temperature� Mark Thomas Winkler proposed the following model to describe non-degenerate n-type semiconductors [1]:

where n is the total carrier concentration, Nd is the density of donor states, Na is the density of acceptor states, b is the spin degeneracy, εd is the reduced binding energy (given by Eq� 3), and Nc is the band edge density of states, given by the following:

where me* is the reduced electron mass, kb is the Boltzmann

constant, h is Planck’s constant, and T is the absolute temperature�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


Since SnS is a p-type material and the model describes negative (n)-type materials, the quantities Nd and Na were switched when performing the necessary calculations (i�e� Nd was used to represent the number of acceptor states and Na the number of donor states), and the binding energy in this case represented the acceptor state activation energy�

Fitting the data to this model by varying two parameters (Na and Nd, the number of acceptor and donor states, respectively), a best fit was achieved that yielded an acceptor state activation energy of 0�109 eV� The resulting curves are plotted in Figure 1�

References:[1] Winkler, M�T� “Non-Equilbrium Chalcogen Concentrations in

Silicon: Physical Structure, Electronic Transport, and Photovoltaic Potential�” Harvard University Press� October 2009�

Figure 1

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Adhesion and Cohesion Testing in Square Solar Cells

Karl BayerPhysics, Pacific Lutheran University

NNIN REU Site: Stanford Nanofabrication Facility, Stanford University, Stanford, CANNIN REU Principal Investigator: Professor Reinhold H. Dauskardt, Materials Science and Engineering, Stanford UniversityNNIN REU Mentor: Fernando Novoa, Department of Materials Science and Engineering, Stanford UniversityContact: [email protected], [email protected], [email protected]

Abstract:

A thorough understanding of environmental effects on adhesion and cohesion in solar cells is necessary for continuing improvements in their efficiency and longevity. A novel technique for adhesion and cohesion testing of square solar cells is developed using a modified double cantilever approach with load pins at the corner of square samples. The resulting method has greater sample versatility, control in testing and ease in sample preparation than other configurations. A closed form solution to obtain the energy release rate, G, for this configuration is derived. Using this approach, we can easily prepare and test samples in glove box environments. Validation of methodology for testing of encapsulated solar cells in controlled environments is obtained using a proportional–integral–derivative (PID) feedback control system.

Introduction:

Increasing demands for reliable solar cell technology place ever increasing expectations on their durability� Solar cells are exposed to moisture, ultraviolet (UV) light, and contaminants which can slowly break down the cell’s layers, leading to the growth of internal cracks, flaws and consequent cell failure� The study of adhesive strength within solar cells is necessary for an understanding of their durability. Adhesion is quantified as the energy cost of extending a crack, equivalent to the rate at which energy is released as a crack propagates through, and is called the energy release rate, or G� Environmental conditions such as relative humidity and oxide (O2) concentration may influence the strength of a sample and edit its energy release rate� It is desirable to understand how G is dependent on those environmental factors�

Test Structure:The energy release rate of a sample was measured using a double cantilever beam (DCB) test in which a sample was sandwiched between two beams and placed in a delaminator� The delaminator consisted of an actuator which applied a displacement to one end of the sample and a load cell which measured the load experienced by the sample (see Figure 1)� The displacement forced a crack to propagate down the sample� The changes in load, in combination with the displacement data from the actuator, were used to determine the energy release rate�

Figure 1: Delaminator test structure for DCB testing.

The conventional DCB sample preparation is both time consuming and low yield� Typically, each solar cell is fabricated on a glass square which must be sawn into strips for the DCB test configuration since beams form the basic structure of a DCB test� The dicing required to cut the square samples into strips is destructive because it exposed the sample to water and oxygen, and produced an uneven edge� In order to test square solar cell samples in a controlled environment, a different method for adhesion testing was developed�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


The standard DCB method was modified so that the samples were left in their square form to avoid dicing� Displacement through the delaminator was applied through load pins attached to the corner of the square� Since the solar cells were fabricated in squares as required for a square-DCB test, no dicing was necessary, damage and contamination to the sample was minimized, and glove-box preparation was made possible�

A consequence of the square-DCB configuration was that G did not depend on the crack length. For a specific test with constant conditions and displacement rate, a plateau formed in the load by time graph� It then followed that the energy release rate was directly proportional to the load in each test� As a result, each test did not need to be closely monitored, and data analysis and the calculation of G was simplified. Figure 2 illustrates a plateau obtained from a square-DCB sample test. The load fluctuated as the sample’s pre-crack developed a meniscus, resulting in the variations seen before 1700 µm� The plateau shows a constant load as expected and results in an energy release rate falling within the margin of error for standard DCB test results of the same film.

determined by the user, the program would increase the displacement velocity, thereby increasing the load toward the set-point and the desired energy release rate�

Figure 3 shows the delamination of a square-DCB sample which used the PID control system to set the load� The lower curve shows the minor fluctuations in load as the control system maintained the set-point� The decreasing stair-step occurred as the user lowered the set-point� The top curve shows the actuator’s displacement developing into a quadratic relationship with time� The decreasing quadratic slope of the displacement curve correlates with the three set-points as expected�

Conclusions and Future Work:The square DCB method for sub-critical analysis was confirmed by the PID load-control experiment. The method may be used to perform sub-critical and critical tests in a glove-box (0% relative humidity, 0�01 ppm O2) or an environmental test chamber for varying humidity and temperature�

The complete method with full results of organic solar cell tests will be published by Fernando Novoa�

Acknowledgments:Special thanks to Fernando Novoa for his guidance, Michael Deal for his confidence, and Professor Dauskardt for opening his lab to me, making this experience so rewarding� Additional thanks to the National Science Foundation, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, and Stanford’s Center for Integrated Systems for making the program possible�

Figure 3: Delamination of a square barrier film using the PID computer controller.

Figure 2: Delamination of a square barrier film at a constant rate (1 micron per second).

When samples were tested at extremely slow delamination speeds, or sub-critical speeds, only critical crack propagation was observed� Samples would crack suddenly rather than slowly and continuously� This created a saw tooth pattern in the load-by-time graph rather than the desired plateau� Delaminator testing by displacement-control was deemed insufficient for sub-critical speeds and a load-control system to control the displacement rate was required�

Load control was achieved by writing a PID computer controller using Labview� The PID controller is a loop-feedback computer controller which determines the displacement velocity of the actuator in real time based on the load� For example, if the load was below the set-point

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Heteroepitaxial Growth of Diamond for Device Applications

Antanica BoneparteChemical Engineering, Howard University

NNIN REU Site: Center for Nanoscale Systems, Harvard University, Cambridge, MANNIN REU Principal Investigator and Mentor: Dr. Gary Harris, Electrical Engineering, Howard University

(on sabbatical at Harvard University)Contact: [email protected], [email protected]

Abstract:

Diamonds grown by chemical vapor deposition have many characteristics that enable their use in medical, electronic and electro-optic applications. The objective of this research was to demonstrate the growth of diamond, through microwave plasma enhanced chemical vapor deposition (PECVD), on various non-diamond silicon (Si) substrates (i.e. Si <100>, Si <111>, 6H-silicon-carbide (SiC), Si coated with nickel, etc.). The substrates were coated with a diamond solution in an ultrasonic bath that created nucleation sites for diamond growth. With the specified parameters—1100 watts, 50 torr, 400 standard cubic centimeters per minute (sccm) hydrogen (H2), 1.6 sccm methane (CH4)—layers grew at about 0.4 µm/hr. Diamond growth was confirmed using energy dispersive spectroscopy (EDS) due to the high levels of carbon on the substrates. Scanning electron microscope (SEM) images proved that diamond layers grew in various orientations based on the substrate it grew on. Si <100> grew high faceted and high quality layers with pyramidal, hexagonal, and cuboidal structures, while Si <100> coated with nickel formed spherical shaped diamonds.

Introduction:

CVD diamonds are an important family of materials used in microelectronic and optoelectronic packaging and for laser and detector windows [1]� Diamond is the hardest known natural material� On the Mohs scale of mineral hardness (a scaling system that characterizes the scratch resistance of various minerals from 1-10) diamond is a 10� Its ultra-high thermal enhances high frequency in optoelectronic systems [2]� Diamond is chemically inert and if doped can be a strong semiconductor� Diamond can be extremely useful in devices that have extremely high power densities, high mechanical loads, and severe abrasive conditions�

Experimental Procedure:There are essentially two methods to synthetically grow diamond: a high pressure-high temperature (HPHT) method and the chemical vapor deposition (CVD) method� In the HPHT process, graphite and a metallic catalyst are placed in a hydraulic press under high temperatures and pressure� HPHT diamonds are usually only a few millimeters across and are too flawed to apply to devices. In the CVD method, the substrate is exposed to gaseous hydrogen and methane that is heated and activated, and there is then a reaction on the surface of the substrate to produce diamond� Microwave plasma enhanced CVD (PECVD) is the method we used, because it can yield better quality material and has the potential to morph the diamond into different shapes�

The various Si wafer substrates were first ultrasonically cleaned with acetone, followed by methanol for 15 minutes each, and then dried with nitrogen (N) gas� This process removed any surface contaminations� The substrates were then seeded with a diamond solution, also referred to as “diamond dust,” in an ultrasonic bath for 15 minutes� The “diamond dust” contained nanoparticles of diamonds that scratched the surface of the substrates while in the ultrasonic bath to create nucleation sites from which the diamonds would grow� After selecting a substrate, it was then put into the PECVD reactor where the growth process took place� In the PECVD reactor, H2 and CH4 were the two gases used for the procedure. There was a significantly larger concentration of H2 to CH4, making CH4 only 0�5 - 1% of the gas concentration mixture� This was to ensure a slow and even growth rate� The gaseous mixture was heated and activated with microwaves initially generated at 350 watts� Under 10 torr of pressure, the gaseous mixture formed a plasma ball� The wattage was increased to 1100 watts and the pressure to 50 torr� These were the general growing conditions that were initially tested�

The plasma ball heated the substrate and activated hydrogen to react with methane to form hydrocarbons that absorbed and reacted with the nano diamond particles on the surface of the substrate. This process continued until the flow of methane was stopped� After the substrate was cooled, tests and data were taken�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


Results and Conclusions:Each substrate created a differently-oriented diamond film because of the substrate surface orientation�

Figure 1 shows an SEM image of Si <100>. This film yielded the highest quality diamond layers compared to all of the substrates� The Si <100> was very faceted and had highly defined shapes which ranged from octahedron and cubes, to tetrahedrons and hexagons� The diamond sizes ranged from 0�5 µm to approximately 1�5 µm in size� Likewise, 3C-SiC layered on Si grew defined facets, although it displayed only one central pyramidal structure throughout the film. The size of the individual diamond structures were approximately 0�5 µm in size� Both substrates were grown under the same parameters of; 1250 watts, 64 torr, 400 sccm of hydrogen gas to 1�6 sccm of methane gas�

Figure 2 shows Si <111>� Both Si <111> and 6H-SiC, however, grew low quality diamond films. The diamond structures were not defined and were very rigid. The sizes of the individual diamonds on both films were relatively small compared to the substrates yielding higher quality diamond, with a size range from 50-100 nm� These substrates were grown under the same parameters previously stated for the Si <100> and 3C-SiC�

Figure 3 shows Si <100> layered with 30Å of nickel� This substrate yielded the highest quality and most defined shapes. The films grew in spherical diamond balls, a very rare structure to form� We believe that the cause of the phenomenon was because the substrate was coated with nickel, which caused the diamond to grow in a different orientation (wet surface)�

Hall measurements of diamond on Si <100> substrates were performed� The electron mobility was 403 cm2/V·s, which is comparable to natural diamond� This indicates that the synthetic growth was successful and these substrates can be applied to devices�

Acknowledgements:I would like to thank Dr� Gary Harris, the National Science Found ation and the National Nanotechnology Infrastructure Net work Research Experience for Undergraduates (NNIN REU) Program�

References:[1] K� Fabisiak, E� Staryga, CVD diamond: from growth to application,

Journal of Achievements in Materials and Manufacturing Engineering 37/2 (2009) 264-269�

Figure 3: Silicon <100> layered with nickel.

Figure 1: Silicon <100> substrate.

Figure 2: Silicon <111> substrate.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


The Dynamics and Control of Bubbles Residing under Graphene Films

Lauren CantleyPhysics, Grinnell College

NNIN iREU Site: Delft University of Technology (TU Delft), NetherlandsNNIN iREU Principal Investigator: Prof.dr. ir. Lieven Vandersypen, Quantum Transport Group, TU DelftNNIN iREU Mentor: Drs. Stijn Goosens, Quantum Transport Group, TU DelftContact: [email protected], [email protected], [email protected]

Introduction:

Graphene, a single atomic sheet of carbon atoms, is known for its superior electrical, thermal and mechanical properties� Because of these advantageous qualities, graphene is currently being used in a variety of applications and must undergo various fabrication processes� In one such process, mechanically exfoliated graphene is transferred from poly(methyl methacrylate) (PMMA) to an hexagonal boron nitride (hBN) substrate� After the transfer, the formation of bubbles between the graphene and substrate are observed� Previous studies have also observed the formation of these bubbles and have detected mass transport beneath the graphene films when annealed, observing the bubbles coalescing into larger units [1]� It is not known with certainty what is trapped beneath these bubbles or why they occur, however understanding this could lead to improved fabrication methods or even new, innovative applications of the material� In this project the properties of the observed bubbles are further investigated by manipulation via an applied back gate voltage and obtaining force-distance curves� Additionally, the interaction between graphene and substrate is of interest and can be probed with these methods�

Methods:Graphene was acquired by mechanical exfoliation and transferred onto a silicon (Si)-wafer with a top layer of Aquasave and PMMA. Monolayer graphene flakes were found by optical contrast and confirmed via Raman

Figure 1: Schematic (left) and AFM image (right) of the device.

spectroscopy. The hBN flakes were transferred to the Si substrate by exfoliation and measured to be 20-30 nm in height using atomic force microscopy (AFM)� Graphene was then transferred from PMMA to the hBN substrate in ambient conditions via the ‘dry transfer method’ [2]� At this point, small bubbles were observed with the optical microscope� To enlarge the bubbles, the sample was annealed in hydrogen (H2) and silver (Ag) at 400°C for three hours� Gold contacts were patterned via electron beam lithography and the sample was glued onto a printed circuit board (PCB) where wires were bonded for electrical measurements�

Figure 1 depicts a schematic of the final device. Two devices were fabricated, each with several bubbles on which measurements could be conducted� Bubble size ranged from 0�4-1�0 µm in diameter and between 30-150 nm in height�

Measurements were carried out using a Nanoscope Multi-mode atomic force microscope (AFM) in conjunction with an electrical set up� The contacts of the sample were connected to IVVI rack via a matrix box, allowing a bubble to be imaged by AFM while a back gate voltage was applied�

Using WxSM image software, AFM images were used to obtain quantitative data on the volume of the bubble at each applied back gate voltage� AFM images were taken using both contact mode and tapping mode� The following measurements were carried out using this set up: volume changes in response to applied backgate voltage swept from -30 to 30V in 3V intervals, volume changes in response to a constant voltage sustained over a period of 48 hours, and IV curves�

Additionally, force distance curves were obtained using AFM for regions both on the graphene bubble and on graphene flush with the substrate. The probes used were MicroCantilever contact mode tips, model OMCL-AC160TS-R3 with a spring constant of ~ 26�1 N/m�

Results:The graphene bubbles presented a diverse array of responses when subject to the AFM measurements� Irreversible changes in volume were observed over the sweep from

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


-30V to 39V both qualitatively in the AFM images as well as quantitatively in the flooded volume measurements. As the applied voltage increased, the bubble pulled into contact with the substrate along its edges, as shown in Figure 2� Modeling the bubble as a parallel plate capacitor, the electrostatic force initially required to pull graphene into contact with the substrate was approximately 19�6 pN, corresponding to 3V�

In response to the constant backgate voltage applied over an extended time (48 hrs), there was evidence of the graphene bubble deflating. Before and after images reveal the occurrence of mass transport beneath the graphene film (Figure 3). The proximity of the deflated graphene bubble to the edge of the graphene film is an indication that the contents of the bubble may have escaped out the edge� Additionally, several graphene bubbles ruptured under the force of the AFM tip due to shear forces present during contact mode imaging�

Force-distance measurements taken on the bubble show that the bubble is deformed by the AFM tip ~ 12 nm before tip deflection occurs (Figure 4). Upon deflection, the slope of the curve is consistent with the slope of an AFM tip interacting with graphene on the substrate, indicating that after initial deformation the tip is deflected as if it were on a hard substrate� The applied force necessary to deform the bubble was calculated to be 300 nN�

Conclusions:The measurements and observations gathered do much in helping us to characterize these graphene bubbles� The deformation of the bubbles reveals that the contents are a compressible substance� While further study is necessary to fully to understand these bubbles, the ability to rupture, move, deflate and control the shape of the bubbles by applying shear,

normal and electrostatic forces make graphene bubbles an exciting prospect for future applications of the material�

Future Work:Measurements should be repeated to confirm the above observations� Additionally, puncturing a graphene bubble via focused ion beam will give insight as to the contents of bubble�

Acknowledgements:I would like to thank the NNIN iREU Program and the NSF for their support of this research, and Delft University of Technology for hosting me, especially Prof�dr� ir� Lieven Vandersypen, Drs� Stijn Goosens, Ir� Victor Calado MSc�, and Prof�dr�ir� Hans Mooij�

References: [1] E� Stolyarova, et al� Observation of Graphene Bubbles and Effective

Mass Transport under Graphene Films� Nano Lett�, 2009, 9(1) 332-337�

[2] Baart, T�A� February 2011� Quantum Dots on Bilayer Graphene made on a Substrate of Boron Nitride using Split Gates� Master Thesis, Delft University of Technology�

Figure 2: Changes in volume in response to applied backgate voltage.

Figure 3: Deflated bubble after constant voltage applied for 48 hours. Figure 4: Force-distance curve

produced on graphene bubble.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Utilizing Solution-Grown Silicon Nanowires in Photovoltaic Devices

Elizabeth FullertonChemical Engineering, University of Arkansas

NNIN REU Site: Microelectronics Research Center, The University of Texas, Austin, TXNNIN REU Principal Investigator: Dr. Brian Korgel, Chemical Engineering, The University of Texas at AustinNNIN REU Mentor: Chet Steinhagan, Chemical Engineering, The University of Texas at AustinContact: [email protected], [email protected], [email protected]

Abstract and Introduction:Photovoltaic (PV) devices were fabricated using solution-grown silicon nanowires and phenyl-C61-butyric acid methyl ester (PCBM)� Silicon nanowires (SiNWs) were synthesized in relatively large quantities using the supercritical fluid-liquid-solid (SFLS) approach. Gold nanocrystals were used to seed nanowire growth and monophenylsilane (MPS) was used as the Si precursor� Devices were fabricated on glass substrates using a device architecture typical for bulk heterojunction PV device with a back contact of aluminum, an active layer of SiNWs and PCBM, poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate) (PEDOT:PSS) as the hole-transport layer, and a top contact of indium tin oxide (ITO)�

Relatively low photovoltaic power conversion efficiency (PCE) under AM 1�5 illumination of 0�002% has been achieved to date� Several approaches were explored to improve device performance, including varying the weight ratio of SiNWs and PCBM, etching and redispersing the SiNWs prior to deposition, alternative spin coating methods, as well as chemical etching of the nanowire surface prior to device fabrication�

Figure 1: System used for SiNW synthesis.

Methods:SiNW synthesis was carried out using the system shown in Figure 1� The small piston and reactor were brought into the glove box and closed off� Reactant solutions of 0�5 ml MPS and 1�2 mg gold nanocrystals were loaded into the small piston followed by toluene to eliminate the remaining void space� The reactor was placed in a heating block with a Variac temperature controller and heated to 490°C� The inlet and outlet line to the reactor were purged while the six-way valve was under load mode� The reactor was then pressurized to 9 MPa and the six-way valve was changed to deliver mode� The remaining lines were purged of air and the inlet line to the small piston was attached� The reactor was then brought to 10 MPa and the flow dropped to 0.5 ml/min� SFLS growth within the reactor: MPS degrades into liquid silicon, in the supercritical region, which then bombards the gold nanocrystals, once completely saturated any additional silicon will generate the nanowire (Figure 2) from the gold seed [1]�

An aqua regia etch was performed on some SiNWs prior to device fabrication� A 3:1 mixture of hydrochloric (10 ml) and nitric acid (3 ml) was achieved and placed on a stirrer�

10 mg of SiNWs were then dispersed in 20 ml of chloroform and added to the mixture� The stir speed was increased to obtain emulsion and the mixture was left covered for 4 h� The wires were cleaned several times using a centrifuge

Figure 2: SEM image of SiNWs.

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


and redispersed in 20 ml of chloroform� Immediately following cleanup, the HF etch was completed� The wires were added to a 1:1:1 mixture of water, ethanol, and HF on the stirrer� The stir speed was again increased to obtain emulsion and the mixture was left for 15 minutes� The etch was followed by cleaning the wires and redispersing them in either chloroform, isopropanol, or 1,2-dichlorobenzene and quickly moved into an oxygen free environment�

Devices were fabricated on glass substrates covered with ITO and patterned using Kapton® tape and an Oxford plasma etch. The substrates were cleaned in consecutive five minute baths of acetone, isopropanol and water followed by 10 minutes in the UV ozone� PEDOT:PSS was then spin-coated onto the substrate at 5000 rpm� A mixture of PCBM and MPS SiNWs was then made to a desired weight ratio� The mixture was then injected onto the substrate and spun up to 600 rpm� After drying, the aluminum back contact was then deposited onto the substrate using a shadow mask and a Denton thermal evaporator� Silver paint was then painted on top of the aluminum contacts to help the card reader on the solar simulator�

Discussion and Results:Figure 3 shows scanning electron microscopy (SEM) images of the average coverage of three devices� The gray area is PEDOT:PSS, while the darker areas in the images are composed of PCBM and SiNWs� The active layer of the

devices was spin-coated in one of two methods: 1) injecting the PCBM/SiNW mixture once the substrate was spinning at 500 rpm, or 2) injecting the mixture while the substrate was stationary and then spinning it up to the desired speed� Figures 3a and 3b show SEM images for these two methods using non-etched SiNWs redispersed in chloroform� The second method shows an increased coverage of the active layer� Figure 3c shows the coverage of the active layer using the second method for spin coating and etched SiNWs redispersed in isopropanol� The coverage is greatly reduced from the non-etched SiNW devices however a PV response was seen after testing with a solar simulator� Figure 4 shows the current-voltage characteristics for this PV� The PCE of the device is 0�002%� The low PCE is most likely due to the non-uniform coverage of the nanowires�

Future Work:In the future, we hope to prove hydrogen termination of the SiNWs post HF etch to ensure that oxygen is not interfering with current transfer� Although the etched SiNWs dispersed relatively well post etch in 1,2-dichlorobenzene, another option to enhance dispersibility could include functionalizing the wires with something that would increase dispersibility in dichlorobenzene� Device coverage might be improved by new spray coating methods or drop casting�

Acknowledgements:Dr� Brian Korgel, Chet Steinhagen, Catherine Shipman and the remainder of the Korgel Group as well as the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, the National Science Foundation, the Robert A� Welch Foundation, Texas Materials Institute and the Center for Nano- and Molecular Science�

Figure 3, left: A. 6 weight percent (wt%) SiNW in PCBM spin coat method 1. B. 6 wt% SiNW in PCBM spin coat method 2. C. 6.5 wt% etched SiNWs spin coat method 2.

Figure 4, above: PV response from etched SiNW device.

References:[1] Hanrath, T� and Korgel,

B.A., Supercritical fluid-liquid-solid synthesis of Si and Ge nanowires seeded by colloidal metal nanocrystals� AdvMats (2003), 15(5), 437-440�

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Patterning of the Metal Induced Crystallization of Amorphous Silicon for Silicon Wire Array Photovoltaics

Julie A. GeorgievDepartment of Electrical Engineering, Alfred University

NNIN REU Site: Penn State Nanofabrication Laboratory, The Pennsylvania State University, State College, PANNIN REU Principal Investigator: Dr. Joan M. Redwing, Materials Science and Engineering, The Pennsylvania State UniversityNNIN REU Mentor: Dr. Chito E. Kendrick, Department of Materials Science and Engineering, The Pennsylvania State UniversityContact: [email protected], [email protected], [email protected]


Photovoltaics (PVs) are a promising clean energy alternative to fossil fuels but they are not currently cost-competitive� Sixty percent of the cost of monocrystalline silicon PV modules comes from the silicon (Si) wafer� Si PVs fabricated with Si wire arrays could eliminate the bulk Si substrate� Not only would Si wires reduce the amount of Si used, but the geometry would also allow for the decoupling of the two main processes required for a PV to work: photon absorption and carrier collection [1]�

For Si wire PVs, it is preferred that the wires are vertically aligned, have controllable diameters and placement, and can be grown on a cheaper substrate like glass� Chemical vapor deposition (CVD) combined with the vapor-liquid-solid (VLS) growth mechanism is one way to meet these criteria� However, this requires a Si <111> orientated layer to grow off [2]� The aluminum induced crystallization (AIC) process is ideal for the fabrication of the Si <111> layer, as it is a low temperature process and Si <111> can be achieved with the help of an aluminum oxide (Al2O3) layer [3]�

Since AIC also produces crystal orientations other than <111>, the objective of this project is to maximize the percentage of Si <111> islands to promote vertical wire growth� To maximize the Si <111> islands, patterned AIC substrates were investigated to confine the island growth� We were interested in determining whether it was possible to maximize the percentage of Si <111> islands by reducing the pattern size to the same magnitude as the individual silicon islands� Oxygen plasma exposure was also investigated as an alternative to air exposure to oxidize the aluminum layer, as ambient environments only produce a monolayer of Al2O3�

Experimental Procedure:Samples were prepared by depositing 100 nm of Al on a quartz substrate either through e-beam evaporation or sputtering� Air exposure (30 minutes, 45 minutes or 16 hours) or oxygen plasma exposure (1, 2�5, 5 or 7�5 minutes) was then used to oxidize the surface of the Al, followed by the deposition of 100 nm of amorphous silicon (a-Si)�

Figure 1a represents the sample before AIC�

Patterned samples were prepared through pre-patterning, by lift-off, and the top layer of Al was oxidized by air exposure for either 45 minutes or 16 hours� The mask pattern contained close-packed hexagons with diameters of 12, 25, 50 and 100 µm� Optical microscopy and scanning electron microscopy (SEM) were used to image the samples after the annealing and etching of the Al layer�

These samples were then annealed at either 450 or 550°C for a period of 1 to 72 hours� Figure 1b shows the AIC process: during the annealing, Si diffuses into the Al layer and becomes supersaturated (1), leading to Si nucleation

Figure 1: A) Sample before AIC, B) During AIC, C) After AIC, and D) After etching the Al.

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


sites throughout the Al layer (2)� The polycrystalline silicon is confined to the thickness of the Al layer, therefore it grows laterally and there is a complete layer exchange between the Al and Si layers (3, 4, 5)� After annealing the Al layer was etched using RCA2, leaving a layer of polycrystalline Si on quartz, shown in Figure 1d� These layers were then coated with 1 nm of gold to act as nucleation sites for the nanowire growth and we subsequently evaluated the amount of vertical wire growth�

Results and Conclusions:The patterned samples prepared with 45 minutes of air exposure had smooth surfaces with no evidence of a double layer of a-Si and polycrystalline Si, which suggests continuous crystallization� Growth on these patterned samples resulted in continuous growth contained to the patterned areas, but at this time it is unclear whether there was improved vertical wire growth� Figure 2 shows an SEM image of a hexagonal pattern after three minutes of growth�

For samples where the top Al layer was oxidized using oxygen plasma exposure, greater times of oxygen plasma exposure resulted in increased amounts of vertical wire growth, with 5 and 7�5 minutes having similar amounts of vertical wire growth to samples prepared with 24 hours of air oxidation� Figure 3 is an SEM image of growth on a sample exposed to an oxygen plasma for 7�5 minutes� The amount of vertical wire growth is similar to the results in Figure 4, a sample exposed to air for 24 hours�

Future Work:At this stage, it is not definitive whether these processes improved the formation of Si <111> grains� It is critical that orientation imaging microscopy be used to verify the orientation of the Si grains to further evaluate our results� The effect of patterning on microwire growth is also currently being explored�

Acknowledgments:I’d like to thank my principal investigator Joan Redwing and my mentor Chito Kendrick for their guidance and for providing me with this opportunity; the Materials Research Institute staff for training and technical advice, particularly Kathy Gehoski for her hard work coordinating this year’s program; also the NNIN REU Program and the National Science Foundation for funding�

References:[1] B� M� Kayes; “Radial P N Junction - Wire Array Solar Cells”; Ph�D�

dissertation, CalTech, Pasadena, California, 2009�[2] R� S� Wagner and W� C� Ellis; “Vapor-Liquid-Solid Mechanism of

Single Crystal Growth”; Appl� Phys� Lett�, 4(5), 1964�[3] M� Kurosawa, K� Toko, N� Kawabata, T� Sadoh, and M� Miyao;

“Al-induced oriented-crystallization of Si films on quartz and its application to epitaxial template for Ge growth”; Sol� Stat� Elec�, 60(1), 2011�

Figure 2, top: 25 µm hexagon, 24 hr anneal, 3 min growth with CVD combined with the VLS growth mechanism.

Figure 3, middle: 7.5 min oxygen plasma exposure, 24 hr anneal, 3 min growth with CVD combined with VLS growth mechanism.

Figure 4, bottom: 24 hour air oxidation exposure, the amount of vertical growth is comparable to that of the 7.5 minutes oxygen plasma exposure sample.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


PZT Films with Reduced and Exaggerated Zr:Ti Gradients

Michael GerhardtDepartment of Materials Science and Engineering, Massachusetts Institute of Technology

NNIN REU Site: Penn State Nanofabrication Laboratory, The Pennsylvania State University, State College, PANNIN REU Principal Investigator: Professor Susan Trolier-McKinstry,

Department of Materials Science and Engineering, The Pennsylvania State UniversityNNIN REU Mentors: Charles Yeager, Department of Materials Science and Engineering,

The Pennsylvania State University; Nicolas Neiman, École Centrale Paris (co-author)Contact: [email protected], [email protected], [email protected]

Abstract:

Lead zirconate titanate (PbZr0.52Ti0.48O3, abbreviated PZT) thin films have shown great promise in many applications, including energy harvesting devices, nonvolatile memory, and miniaturized sensors and actuators, due to their piezoelectric and ferroelectric properties. Efforts to prepare PZT thin films via standard sol-gel deposition result in a compositional gradient, as a Ti-rich phase nucleates first [1]. This compositional gradient adversely affects the piezoelectric properties of the film. In this study, several PZT films were prepared with varying degrees of this gradient, and the effect of the gradient on the electrical properties of the material was examined. The films were prepared using a standard sol-gel deposition process, a “gradient-free” sol-gel process introduced by Calame and Muralt [1], and a “gradient-enhanced” process. The relative permittivity and piezoelectric coefficient e31,f were characterized for each of these films. The gradient-free films show the best piezoelectric coefficient (e31,f = -14 C/m2 on silicon (Si) substrates), while the gradient-enhanced films show poor piezoelectric properties (e31,f = -7.5 C/m2 on Si). These results confirm that the gradient-free process produces PZT films with superior piezoelectric properties.

Introduction:A piezoelectric material is a material which produces a charge when it is strained and will deform under an applied electric field. Applications for these materials include mobile communications and microscale sensors and actuators [2]� Lead zirconate titanate (PZT) is the piezoelectric material of choice due to its high effective transverse piezoelectric coefficient (e31,f), which relates the strain in the plane of the film to the polarization developed perpendicular to the film [2]. This coefficient is dependent on composition and reaches a maximum value at the composition PbZr0�52Ti0�48O3 [3]�

Unfortunately, standard sol-gel processing produces a compositional gradient throughout the film thickness, as a Ti-rich phase will nucleate first during annealing [3]. This gradient can be reduced by using precursor solutions of varying Zr:Ti compositional ratio to construct a gradient in the opposite direction to counteract the one which will form during annealing [1]� In this project, we deposited PZT films on silicon and magnesium oxide substrates using the standard sol-gel process, the reduced-gradient sol-gel process, and an exaggerated-gradient sol-gel process, and compared the piezoelectric properties of each film.

Experimental Procedure:PZT films of approximately 1 µm in thickness were deposited on platinized Si and magnesium oxide (MgO) substrates using a sol-gel process� Four PZT precursor solutions of varying Zr:Ti ratio were prepared based on the 2-methoxyethanol route [4]�

Films with a reduced and exaggerated Zr:Ti gradient were produced following the procedure described by Muralt [1]� First, a lead titanate (PbTiO3) seed layer was used to promote <100> texturing on Si substrates� Then, the four PZT precursor solutions were spun onto the substrate, starting with the Zr-rich solution for a reduced gradient film and the Ti-rich solution for an exaggerated gradient� Each layer was dried and pyrolyzed on hot plates. The films were annealed every four layers. The “standard-process” films were deposited using one 52:48 Zr:Ti solution, dried, pyrolyzed, and annealed every layer under the same conditions�

The crystalline texture of each PZT film was determined using x-ray diffraction. Polarization-electric field hysteresis loops were also obtained� The transverse piezoelectric coefficient e31,f of each film was measured using the wafer flexure technique [5].

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


Strain gauges were glued to the film, and then the film was glued to a Si wafer to which a pressure wave was applied� Samples were poled at three to five times their coercive voltages for 15-20 minutes. A lock-in amplifier was used to measure the charge produced and strain in the film in phase with the pressure wave.

Results and Conclusions:Figure 1 shows the x-ray diffraction patterns of a sample (a) with a PbTiO3 seed layer and (b) without� The intensity of the <110> peak is diminished in the sample with a seed layer, indicating that the seed layer enhanced the <100> orientation of the film.

Polarization-electric field hysteresis curves of select films are shown in Figure 2. The coercive field, or the field at which the polarization switches direction, increases as the concentration gradient is increased� This could be due to impaired domain wall movement caused by non-uniformity in the film.

The measured e31,f and relative permittivity are reported in Table 1� The e31,f coefficient has been shown to increase as the Zr:Ti gradient is reduced� Therefore, reducing the gradient is crucial for applications which take advantage of the piezoelectric properties of PZT, such as energy harvesting�

Acknowledgements:I thank Dr� Susan Trolier-McKinstry, Charley Yeager, and the rest of the STM research group, and the staff at the Penn State Materials Research Laboratory� Many thanks to Kathy Gehoski and Kathy Gummo for their work in coordinating the REU program, and the NNIN REU Program and the NSF for funding�

References:[1] Calame, F�, and Muralt, P� “Growth and properties of gradient free sol-gel

lead zirconate titanate thin films.” App.Physics Lett., 90, 062907 (2007).[2] Muralt, P�, Polcawich, R�G�, and Trolier-McKinstry, S� “Piezoelectric

Thin Films for Sensors, Actuators, and Energy Harvesting�” MRS Bulletin, 34, 658-664 (2009)�

[3] Ledermann, N� et al� “{1 0 0}-Textured, piezoelectric Pb(Zrx,Ti1-x)O3 thin films for MEMS: integration, deposition, and properties.” Sensors and Actuators A, 105, 12-170 (2003)�

[4] Budd, K�D�, Dey, S�K�, and Payne, D�A� “Sol-Gel Processing of PbTiO3, PbZrO3, PZT, and PLZT Thin Films�” British Ceramic Proceedings, 36, 107-121 (1985)�

[5] Wilke, R�H�T� et al� “Wafer Mapping of the Transverse Piezoelectric Coefficient, e31,f, Using the Wafer Flexure Technique with Sputter Deposited Pt Strain Gauges�” to be submitted to Sensors and Actuators A�

Table 1: e31,f and permittivity values for all three film types on Si and MgO substrates.

Figure 1: XRD pattern of PZT films on Si substrates (a) with and (b) without a PbTiO3 seed layer.

Figure 2: Polarization-electric field hysteresis curves for reduced and exaggerated gradient films on (a) Si and (b) MgO substrates.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Self-Assembled Gold Nanoparticles for Biosensing Applications

Abigail HalimMaterials Science and Engineering, Georgia Institute of Technology

NNIN REU Site: Nano Research Facility, Washington University in St. Louis, St. Louis, MONNIN REU Principal Investigator: Prof. Srikanth Singamaneni, Department of Mechanical Engineering

and Materials Science, Washington University in St. LouisNNIN REU Mentors: Dr. Abdennour Abbas, Mechanical Engineering and Materials Science, Washington University in St. Louis;

Saide Zeynep Nergiz, Department of Mechanical Engineering and Materials Science, Washington University in St. LouisContact: [email protected],[email protected], [email protected], [email protected]

Introduction:

Localized surface plasmon resonance (LSPR) is a collective oscillation of conductive electrons of metal nanoparticles, produced by light excitation under certain conditions [1]� The LSPR wavelength at which the resonance occurs is extremely sensitive to changes in the refractive index and, consequently, to the composition of the nanoparticle environment� The LSPR wavelength shift can then be measured and used to detect molecules adsorbed on the surface of the nanoparticle, which is known as LSPR spectroscopy [2]� The assembly of these nanoparticles generates a plasmonic coupling between adjacent particles, leading to significant sensitivity enhancement [3]. Such a property makes nanoparticle assemblies an attractive

Figure 1: The mechanism for self-assembly of AuNP with pATP.

platform for biosensing applications� However, the use of these assemblies for biosensing applications requires the generation of stable nanoparticle clusters along with their efficient transfer to solid substrates.

This study presents a controlled, rapid process for self-assembly of gold nanoparticles (AuNPs) using aminothiols and a method to transfer AuNP assemblies onto substrates, while preserving the assembly structure and optical properties� AuNPs were self-assembled using p-aminothiophenol (pATP) as a crosslinker� The assemblies were then transferred to glass and silicon substrates by immersion and examined using ultraviolet-visible spectroscopy (UV-Vis), Raman spectroscopy, scanning electron microscopy (SEM), and transmission electron microscopy (TEM)�

Experimental Procedures:AuNPs were synthesized using the standard method of reducing chlorauric acid with sodium citrate [4]� For self-assembly, the AuNP solution was diluted to a desired concentration, and pATP was added� As shown in Figure 1, the mechanism for self-assembly involves the amino group on one end of the pATP molecule binding with a AuNP, and the thiol group on the other end binding with another AuNP� By varying the concentrations of pATP, the self-

assembly could be “frozen” at different stages, ranging from dimers to branched networks� The distinct optical properties of the solutions resulting from different concentrations of PATP are shown in Figure 2�

To enable the transfer of AuNP assemblies onto solid substrates, glass and silicon substrates were functionalized using two different methods: polyelectrolyte multilayers (comprised of poly(allylamine hydrochloride) (PAH) and poly(sodium 4-styrenesulfonate) (PSS)) using Figure 2: AuNP assemblies with increasing pATP concentrations from left to right.

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


layer by layer (LbL) assembly or a single layer of 3-aminopropyltriethoxysilane (APTES)� A spin coater was used to deposit 3�5 bilayers, ending on PAH, on substrates for the LbL method� For the APTES method, substrates were immersed in APTES solution for 20 minutes and sonicated in water for one hour� The negatively charged citrate-capped AuNPs adsorbed to the positively charged top layer of PAH or APTES through electrostatic interactions�

The self-assembled AuNPs were deposited onto the substrates by immersing the substrates in the assembled AuNP solution� LBL substrates were immersed for 2 h and 40 min, while APTES substrates were immersed for three days� After immersion, the substrates were rinsed and stored in water�

Results:Figure 3 shows the UV-Vis extinction spectra of AuNP solution without and with assembly, demonstrating a clear change in AuNP optical properties with the addition of pATP, evidenced by the appearance of a second peak� Extinction spectra of the glass substrates immersed in these solutions, also shown in Figure 3, exhibited peaks at the same wavelengths and with the same shapes, but with lower relative intensities, indicating the successful deposition of AuNP assemblies onto the substrates while retaining their properties as compared to the assemblies in solution�

Figure 4 depicts the results of AuNP transfer with varying pATP concentrations� The SEM images of AuNP assemblies clearly indicate an increase in the extent of assembly (i�e� individual particles to dimers to finally a network-like structure) with increasing pATP concentration�

Conclusions:The use of pATP as a crosslinker for AuNP self-assembly is a very efficient and versatile technique. By varying the molar ratio of pATP/AuNP, the extent of AuNP self-assembly can be finely controlled. Furthermore, the immersion of APTES or polyelectrolyte functionalized solid substrates in AuNP solutions during the assembly process leads to the transfer of AuNP assemblies with a high degree of fidelity by retaining their structure and optical properties� The sensitivity and analytical parameters of assembled AuNP on solid substrates and their use for biological sensing are under investigation�

Acknowledgements:I would like to thank my principal investigator, Professor Srikanth Singamaneni, my mentors, Dr�Abdennour Abbas and Zeynep Nergiz, and other members of the Soft Nanomaterials Laboratory in the Washington University

Figure 4: SEM images of assembled AuNP on silicon substrates with increasing pATP concentration from left to right.

Figure 3: UV-visible extinction spectra of AuNP with (gray) and without (black) assembly. Left: Spectra for AuNP in solution and on substrates, with AuNP in solution having a higher relative intensity. Right: Spectra of AuNP on substrates.

Department of Mechanical Engineering and Materials Science, including Dr� Ramesh Kattumenu and Limei Tian, for the opportunity to work with them and for their guidance throughout this project� I would also like to thank Dee Stewart, Kate Nelson, and Nathan Reed at the Nano Research Facility at Washington University for coordinating my stay during the program and for valuable instrument trainings and general advice throughout the research process� Lastly, I would like to thank Dr� Dong Qin, NNIN WUSTL Site Director, the National Nanotechnology Infrastructure Net-work Research Experience for Undergraduates Program and the National Science Foundation for funding my stay and work�

References:[1] Ko, H�, Singamaneni, S�, and Tsukruk, V�; Small 2008, 4, 10, 1576-

1599� [2] Willets, K� A� and Van Duyne, R� P�; Annu� Rev� Phys� Chem� 2007,

58, 267-297� [3] Abbas, A�, Kattumenu, R�, Tian, L�, Nergiz, S� Z�, and Singamaneni,

S�; J� Nanoscience Lett� 2011� Advanced online publication� [4] Grabar, K� C�, Freeman, R� G�, Hommer, M� B�, and Natan, M� J�;

Anal� Chem� 1995, 67, 735-743�

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Fabrication of Free-Standing Graphene Films for Probing the Ultrafast Electron Dynamics

Jeffrey HartElectrical and Computer Engineering, Franklin W. Olin College of Engineering

NNIN REU Site: Colorado Nanofabrication Laboratory, University of Colorado, Boulder, CONNIN REU Principal Investigator: Professor Thomas Schibli, Department of Physics, University of Colorado at BoulderNNIN REU Mentor: Chien-Chung Lee, Department of Physics, University of Colorado at BoulderContact: [email protected], [email protected], [email protected]

Introduction:

Graphene has generated much interest due in part to very high carrier mobility (200,000 cm2V-1s-1) and ultrafast carrier relaxation time (<100fs) [1, 2]� However, trapped charges in the substrate or at the substrate/graphene interface couple with the graphene, causing doping effects and lowering the carrier mobility [1, 3]� In order to accurately study the ultrafast electron dynamics in graphene, these negative substrate effects must be mitigated� Therefore, this project aims to fabricate free-standing graphene grown by chemical vapor deposition (CVD)�

The substrates for free-standing graphene were fabricated from silicon wafers by etching holes (2-50 µm) with potassium hydroxide (KOH)� CVD graphene with domain sizes from 20-50 µm was grown and transferred to the patterned silicon wafer� High-quality, free-standing graphene with low p-doping was observed and confirmed with Raman spectroscopy�

Procedure:Fabrication of free-standing graphene was accomplished in two distinct processes (see Figure 1). In the first process, holes were etched into double-side polished 100 µm thick 2-inch <100> silicon wafers� Low-stress silicon nitride was deposited via plasma enhanced CVD to create a ~ 330 nm layer to serve as a mask for etching� Negative photoresist was applied to both sides of the wafers and holes were patterned and etched into the nitride on the back side with a buffered oxide etch� A 250-minute KOH etch (which etches the <100>-plane ~ 100 times faster than the <111>-plane) at 70°C was then used to create holes through the wafer, 2-50 µm wide on the front side�

In the second process, graphene was grown via CVD on a 25 µm copper foil catalyst [4]� More than 90% of this graphene was single layer since graphene can only grow where the copper catalyst is exposed� The CVD growth process was conducted in a 10 mTorr vacuum at 1000°C for four hours while methane (0�2 sccm) and hydrogen (10 sccm) were injected into the system� Ultra-low-pressure

Figure 1: Illustration of the wafer processing and graphene growth and transfer methods used to fabricate free-standing graphene.

growth was chosen to minimize the number of graphene nucleation sites on the catalyst and to allow for each domain to grow to its maximum size� Following growth, the graphene was transferred to the patterned silicon wafer [5]� For this process, the copper was spin-coated with a layer of poly(methyl methacrylate) (PMMA) following graphene growth� After allowing the PMMA to cure, a 0�5M ferric chloride hexahydrate solution was used to etch the copper and leave graphene stabilized with PMMA� The samples were rinsed with deionized water before being transferred to the silicon wafer� After drying, the PMMA was removed in a furnace (400°C, one hour) in an H2/N2 atmosphere�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


Results:The graphene was characterized via scanning electron microscopy (SEM)� This analysis revealed single-crystal graphene domains ranging in width from 20-50 µm, with an average size of approximately 30 µm� Graphene was found to cover between 80% and 90% of the copper surface (see Figure 2)� Large domain sizes (> 20 µm) and high coverage was needed to maximize the probability that a single-crystal graphene domain would cover a hole in the silicon wafer�

The finished samples were examined by optical microscope. The presence of small particles in the middle of a hole in both light and dark field modes (see Figure 3) was found to be a clear indicator that free-standing graphene was present�

Raman spectroscopy was used to verify the presence of single-layer graphene and determine its quality and doping� Figure 4 compares the Raman spectrum for a sample of free-standing graphene with a sample of graphene on silicon� The Raman spectrum clearly shows the characteristic G and 2D peaks of graphene� The insubstantial D peak in the spectrum of the suspended graphene indicates that it is largely defect free and that the hole is covered by a single-crystal graphene domain (domain edges contribute to the D peak)� Analyzing peak shifts and the 2D/G peak ratio provides information about the doping characteristics of the sample� Both the G and the 2D peaks are down-shifted in comparison to the graphene on silicon with native oxide, indicating lower p-doping� The 2D/G peak ratio of 3�6 further indicates that this graphene is minimally doped and high-quality [6, 7]� Moreover, there is no PMMA signature in the Raman spectrum for the suspended graphene, indicating that the chosen removal method was effective�

Conclusions:In conclusion, free-standing graphene was fabricated via CVD growth and transfer to etched silicon� CVD growth reliably produces high-quality single-layer graphene ideal for this study. Graphene was identified by optical microscope on 15 holes in four wafers and verified by Raman spectroscopy� High-quality free-standing graphene was observed over holes as large as ~ 20 µm� The p-doping level in the free-standing graphene was observed to be lower than in the graphene on silicon with native oxide� With the development of this free-standing graphene, further studies in fundamental physics are now possible�

Acknowledgments:We greatly appreciate the generous funding provided by the National Science Foundation through the NNIN REU Program� Facilities and support were provided by the Colorado Nanofabrication Laboratory (CNL), University of Colorado at Boulder�

Special thanks to Thomas Schibli (PI), Chien-Chung Lee (mentor), Jonah Miller, and the CNL staff�

References:[1] Du, X� et� al� Nature Nanotechnology� 3, 491 (2008)�[2] Dawlaty, J� M� et� al� Applied Physics Letters� 92, 042116 (2008)�[3] Joshi, P� et� al� J� of Physics� Condensed Matter 2010, 22, 334214�[4] Li, X� et al� Science� 324, 1312-1314 (2009)�[5] Li, X� et al� Nano Letters 9, 4359-4363 (2009)�[6] Ferrari, A� C� et al� Physical Review Letters� 97, 187401 (2006)�[7] Ni, Z� H� et� al� ACS Nano 2009, 3, 569-574�

Figure 2, top: SEM image showing 20-50 µm graphene domains covering 80-90% of the copper.

Figure 3, middle: Light Field (left) and Dark Field (right) optical microscope images of free-standing

graphene, as indicated by suspended particles.

Figure 4, bottom: Comparison of Raman spectra for graphene on silicon and free-standing graphene.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Growth and Characterization of Aluminum Nitride Nanowires

Alicia HerroPhysics, University of North Texas

NNIN REU Site: Howard Nanoscale Science and Engineering Facility, Howard University, Washington, DCNNIN REU Principal Investigator: Dr. Gary L. Harris, Electrical Engineering, Howard UniversityNNIN REU Mentor: Mr. Crawford Taylor, Research Associate, Howard UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Nanowires are at the forefront of the advancement in electronics due to their versatility and cost effectiveness when compared to bulk materials. Aluminum nitride (AlN) nanowires have many interesting properties in comparison to other wide bandgap group III-V nitride semiconductors. They have a comparatively high thermal conductivity at low temperatures and a high resistivity along with the largest piezoelectric coefficient of the group. AlN nanowires are therefore very attractive for use in electromechanical and optoelectronic devices. They have shown promise in surface acoustic wave devices, such as those found in cell phones, to keep unwanted frequencies out, and in the touch screen displays that have become commonplace in our electronics. AlN nanomaterials, including wires, have been found to trap hydrogen very well because of their structural makeup. The growth of AlN nanowires however can be difficult to achieve because of the difference in the free energy of formation of aluminum oxide (Al2O3) versus that of AlN, making Al2O3 the favored product. At a growth temperature of 773°K, the Gibbs free energy of formation for Al2O3 is -1432.6, while that for AlN is -219.2 [1]. Any oxygen in the system would therefore bond to the aluminum, preventing the formation of AlN.

Introduction:

Here we attempted to grow AlN nanowires without the aid of a catalyst, though for the sake of experiment, we also attempted grow AlN on samples that had been catalyzed as well� We used a low temperature and low pressure system to allow for a simple and easy to reproduce experiment� In order to test our results, we used energy dispersive spectroscopy (EDS), which shows the elements that are present on a sample and in what percentage� These methods have allowed us to gain valuable data on the affects of growth parameters on manufacturing of nanowires and also on the precision that is at times needed�

Experimental Procedure:In this work, a horizontal chemical vapor deposition (CVD) system, employing ammonia (NH3) and aluminum chloride (AlCl3) as the sources of non-diatomic nitrogen (N) and aluminum (Al) respectively, was used� Silicon (Si) substrates were used as seed material and some were coated with thin layers of various metals as catalysts� Also some samples were partially coated with aluminum and kept upstream to promote growth on the Si� Growth temperatures ranged between 1000-1100°C with pressures of 100 Torr and flow rates between 100-500 sccm� In order to keep oxygen in the system to a minimum, the system was flushed during

both heat up and cool down with a flow of nitrogen gas at a rate of 100 sccm� Growth times, measured only when the machine was fully heated up, varied but most runs lasted 120 minutes�

Results and Conclusions:A film of AlN and some micro-scale deposits of AlN were found on silicon- and nickel-coated silicon samples by EDS, which can be seen in Figure 1� However no AlN nanowires wires were found on any samples, even those with catalysts� Chlorine deposits were also found on the surface of some samples, and ammonium chloride deposits were found in the reaction tube� This reaction was expected since AlCl3 was the source of aluminum, and it had little effect on experimental outcomes� However, the main reason for the lack of AlN growth was aluminum oxidation�

The source of the oxygen was possibly due to the AlCl3 and system leaks� A very large deposit of aluminum oxide was formed in the quartz tube used for heating and on the substrates, indicating the extent of oxygen in the system� Figures 2 and 3 show deposits of Al2O3 on some samples and the almost artistic way in which they grew due to the changing growth conditions�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


Future Work:In the future, we hope to successfully grow AlN wires by starting with gallium nitride (GaN) wires, which have already been proven to be capable of growth� The source of Ga would slowly be replaced by Al, adjusting variables as necessary to keep the growth of wires stable, until there is no longer any Ga in the wires at all� The hope is that by growing the wires in this fashion the formation of aluminum oxide and ammonium chloride could be kept under control so that AlN would have a chance to form� There is also the possibility of using powdered aluminum as the Al source rather than AlCl3, to stop the chlorine from interacting and to cut down on the amount of oxygen being brought into the system, as there is a chance that oxygen was being trapped in the powder and introduced into the system that way�

Acknowledgments:This project would not have been possible without the funding of the National Science Foundation and the wonderful people behind the National Nanotechnology Infrastructure Network Research Experience for Under-graduates (NNIN REU) Program� Helping to guide me along, and solve problems as they arose, were my PI, Dr� Gary Harris, and my mentor, Mr� Crawford Taylor� James Griffin, Karina Moore, and the rest of the staff at Howard were all a great aid, and good teachers on the workings of the lab and the machines it contains�

References:[1] Sharafat, S� and Ghoniem, N�; “Thermodynamic Stability

Assessment of Oxides, Nitrides, and Carbides in Liquid Sn-25Li”; APEX Study, UCLA (August 2000)�

Figure 1: Large growth of AlN on Ni-coated Si.

Figure 2: Al2O3 spheres and nanowires.

Figure 3: Another example of Al2O3 spheres growing off a large Al2O3 mass.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Deposition and Characterization of Ruthenium Films for Neural Electrodes

Diana WuChemical Engineering and Biology, Massachusetts Institute of Technology

Emily HoffmanMaterials Science and Engineering, Northwestern University

NNIN iREU Site: Interuniversity Microelectronics Center (imec), Leuven, BelgiumNNIN iREU Principal Investigator: Dr. Hercules P. Neves, imec, Leuven, BelgiumNNIN iREU Mentor: Dr. Aleksander Radisic, imec, Leuven, BelgiumContact: [email protected], [email protected], [email protected], [email protected]

Abstract:

Neural probes are used to stimulate neurons or record electrical signals, which can be instrumental in understanding the neural network and treating disease. Platinum and iridium are currently used as the electrode material, but ruthenium has promising properties. It is important that materials have high charge storage capacity and low impedance. We investigated the deposition of ruthenium oxide films on gold, titanium nitride, platinum, and atomic layer deposited ruthenium. The samples were characterized using cyclic voltammetry and impedance spectroscopy to predict their performance for neural probe applications.

Introduction:

Neural probes interface electrical stimulation with biological tissue such as neurons� These probes can be used to stimulate or to record the activity of the neurons� Applications of the probes include deep brain stimulation for Parkinson’s disease, epilepsy, and depression [1, 2]�

Neurons can be stimulated by creating a functional response by depolarizing membranes of excitable cells through the injection of current from the probe. This creates a flow of ionic current between two or more electrodes, one which is near tissue� Neural activity can also be recorded by micro-probes by measuring the potential created by neuron membrane depolarization� Current probes are made of platinum (Pt), iridium oxide (IrO2), and titanium nitride (TiN) [1, 3, 4]�

For stimulation, probes must be able to send a charge-balanced waveform to avoid damage to the electrode and surrounding tissue [1]� Chemically, probes must be bio-compatible� Any reactions that occur at the electrode surface must not release harmful molecules into the body or degrade the electrode so that performance is affected� Neural probes should be small in size to be less intrusive in the brain tissue [1]�

The goal of this project is to improve the electrical performance of neural probes by plating ruthenium (Ru) metal onto possible probe substrates� It is worthwhile to investigate the use of Ru because it has more reduction-

oxidation states and could provide a better interface with tissue� An increased number of states may correspond to a greater charge injection, which would be useful for brain stimulation� Ruthenium also shares similar biocompatibility and corrosion resistance properties to other currently successful materials [5]�

Deposition Methods:Electrochemical deposition is selective and a very small area can be plated� In this report, electrochemical deposition was used to deposit Ru on multiple substrates to investigate its usefulness for improving neural probes�

To deposit Ru onto different substrates, a three probe setup was used with a potentiostat (Autolab Booster 20 A)� A Ag|AgCl reference electrode and platinum counter electrode were used� Ruthenium was deposited using constant voltage deposition, and constant current deposition at varying magnitudes and for various times� Two electrolytes were investigated: (Ru(NO2))2(SO4)3 and RuCl3� In some samples, the electrolyte was heated to approximately 60°C to increase the kinetics of the solution�

Several substrates were used for deposition: gold (Au), platinum (Pt), titanium nitride (TiN) and titanium nitride treated with hydrofluoric acid (TiN HF). Atomic layer deposited ruthenium (ALD Ru) was used as a control�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


Deposition Results and Discussion:The electrochemical plating process was used to deposit a film of Ru on the various substrates. After deposition, the samples were visually inspected to determine the continuity of the film and the overall quality. Visually, the continuous samples appeared shiny, while non-continuous films were milky and spotted because of the clustered of nucleated Ru� The other continuous samples were silver to purple in color and appeared smooth�

To test for initial adherence to the substrate, a tape test was performed� All samples remained bonded to the substrate after tape was adhered and pulled from the ruthenium film, showing electrochemically plated Ru has sufficient adhesion properties�

Promising samples were imaged using scanning electron microscopy (SEM) to evaluate the grain size and continuity of the film, as seen in Figure 1. Pt and ALD Ru had films with the smaller grains (> 100 nm), Au films had grains around 100 nm, and the various TiN substrates all produced the largest grains (< 200 nm)� All samples showed a continuous film, except for the non-continuous TiN (TiN NC). TiN NC film had islands of Ru and a large variance in grain size� By using HF to remove the oxide layer on TiN before deposition, more nucleation sites occurred and the grain size became more uniform. TiN can grow a continuous film without HF (TiN C), but non-planar growth of the Ru can sometimes leave gaps and varying grain size�

Nucleation plays a part in the type of films produced. The ALD Ru surface produced a shiny film that had high performance� Because the Ru electrolyte was nucleating on a Ru substrate, the growth was most likely layer growth, which is smooth and continuous with fewer film defects. On the other substrates, Ru grew through a nucleation-

coalescence mechanism, leading to a variety of grains and more defects in the film. ALD Ru pretreatment on the electrodes substrate could be a successful mechanism to improve the Ru electro chemical deposition� Ru electro-chemical deposition is required in addition to ALD because it increases the thickness of the film to the desired level [6].

For the mechanical robust ness of a probe film, a continuous film with small grains would most likely be best, but the CSC and impedance will further be tested to determine the film properties.

Characterization Methods:Cyclic voltammetry (CV) can be used to identify the presence of reactions such as electron transfer, reduction and oxidation� Voltammograms, the graphs resulting from CV, provide information on the reversibility of reactions, quantity of electro-active material on the electrode and the stability of the electrode [7, 8]� CV was conducted with a three electrode set up using a potentiostat (CompactStat, Ivium Technologies), a Pt counter electrode, and a non-current carrying reference electrode Ag|AgCl [7]� It is standard to conduct CV scans at potentials between -0�6 and 0�8V for electrodes meant for biological applications, since it is important to stay within the water electrolysis window [1]� In this report, CV scans were performed for 3 or 10 cycles [7]�

After initial CV scans it was seen that the shape and area of the curve significantly change after the first cycle. This possibly indicates that the surface is undergoing a sensitive irreversible reaction� To reduce this possibility, the potential window was decreased to -0�6 to 0�4 V�

Figure 1: SEM images of the primary samples investigated.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Figure 3: Cycle ranges and resulting

voltammograms.

Figure 4: Voltammograms of samples compared to blank substrates.

Figure 5: The calculated CV with the percent increase from blank substrate.

Phosphate buffered saline (PBS) was chosen as the electrolyte for the CV scan and impedance testing because it is representative to the ion concentrations found in the body�

The cathodic charge storage capacity (CSC) is a quantitative assessment of a neural probe as a recording or stimulating device� The CSC is calculated using the results of a voltammogram using the following equation in Figure 2:

where v is the scan rate (V/s), A is the exposed surface area (cm2), E is the electrode potential (V), Ea and Ec are the anodic and cathodic potential limits (V) and i is the measured current (A) [9]�

Impedance is the quantification of a substrate’s opposition to alternating current and is often used as another characterization method for neural probes� Probes require low impedance so charge can easily be transferred to or from the probe [10]� Most literature measures the impedance for frequencies between 10-1 to 105 Hz, but report impedances at

1 kHz for comparison� Impedance investigation is performed with a potentiostat (CompactStat, Ivium Technologies)�

Characterization Results and Discussion:Samples were initially CV scanned from -0�6 to 0�8 V but the scan window was adjusted to -0�6 to 0�4V after a peak indicating oxidation was seen as in Figure 3� The peak is not seen in scans from -0�6 to 0�4V� CV scans of Ru plated substrates show significantly more internal area than their blank substrate counter parts as seen in Figure 4� Additionally, between, each cycle, the voltammogram shape remains the same, but shifts vertically� Initial changes between the first and second cycles can be attributed to a sensitive reaction [7]�

Charge storage capacity was calculated for all samples created and the CSC’s of the best performing samples for each substrate and their blank standards are displayed in Figure 5� Au, TiN, TiN treated with HF, and ALD Ru all saw a significant increase in CSC after the deposition of Ru. Interestingly, Pt saw a decrease in CSC after the deposition of Ru, most likely due to strong gas evolution during plating�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


Figure 6: Impedance measurements as compared to the blank substrate.

References:[1] Cogan, Annu� Rev� Biomed� Eng� 10 (2008) 275-309�[2] Cogan, Troyk, Ehrlich, Plante, IEEE Trans� Biomed� Eng� 52 (2005)

1612-1614�[3] Frölich, Rzany, Riedmüller, Bolz, Schaldach, J� Mat� Sci�: Materials

in Medicine 7 (1996)393-397� [4] Lu, Wang, Cai, Cao, Yang, Duan, Sensors and Actuators 137 (2009)

334-339�[5] Galizzioli, Tantardini, Trasatti, J� of App� Electrochem� 4 (1974)

57-67�[6] Paunovic, Milan, Mordechay Schlesinger, and Electrochemical

Society� Fundamentals of electrochemical deposition� Wiley-Interscience, 1998� Print�

[7] Compton, Banks� Understanding Voltammetry� Singapore: World Scientific, 2007. Print.

[8] Pletcher, and Southampton Electrochemistry� Instrumental methods in electrochemistry� Chichester: Horwood Publishing, 2001� Print�

[9] Negi, Bhandari, Rieth, Solzbacher, Biomed� Mater� 5 (2010) 15007�[10] Robinson, Davidson, Wright, Pomfrett, McCann, IEEE Eng� in

Med� And Biol (2008) 1171-1174�[11] Wang, Liu, Durand, IEEE Trans� Biomed� Eng� 56 (2009) 6-14�[12] Mann, Freyland, Raz, Ein-Eli, Chem� Phy� Letts 460 (2008) 178-

181�

Figure 6 indicates that Ru film on a substrate lowers the impedance at frequencies below 1 kHz� Above 1 kHz, Au, Pt, TiN treated with HF, TiN and ALD Ru all have very similar impedances of approximately 100 ohms�

All samples were characterized using these techniques, however, only the best performing samples for each substrate are shown in the figures.

Conclusions:Electrochemically plating ruthenium increased the CSC values for all substrates except for Pt and decreased the impedance for all substrates except for TiN� Iridium oxide is the current high performance standard, but CSC values ranging from 2�8-45 mC/cm2 [1, 2, 4, 12]� These values were calculated using various solutions and voltage windows, but provide a range to compare Ru films. The Ru films produced have a CSC ranging from 0.81 ± 0.07 to 2.0 ± 0�1 mC/cm2� These results show that Ru might be useful for improving the electrical properties of electrode coatings, and would be best suited for Au, TiN, or ALD Ru probes� Further investigation to improve Ru coatings on Pt would allow for a broader application of Ru electrodes�

Acknowledgements:We would like to thank the National Science Foundation, the National Nanotechnology Infrastructure Network International Research Experience for Undergraduates (NNIN iREU) Program, Lynn Rathbun, imec, Herc Neves, Alex Radisic, and John O’Callaghan�

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Flexible and Stretchable Networks of Metals

Gawain LauChemical and Biomolecular Engineering, University of Pennsylvania

NNIN REU Site: Colorado Nanofabrication Laboratory, University of Colorado, Boulder, CONNIN REU Principal Investigator: Prof. Mark Stoykovich, Chemical and Biological Engineering, University of Colorado BoulderNNIN REU Mentor: Ian Campbell, Chemical and Biological Engineering, University of Colorado BoulderContact: [email protected], [email protected], [email protected]

Abstract:

Flexible and stretchable networks of metals were pursued, with a vision for their application in flexible electronics. This project aimed to develop: 1) methods for the nanofabrication of these networks with ~ 25 nm dimensions, and 2) processes for the transfer of these structures to elastomeric substrates. Block copolymers were used as a unique method of lithography for patterning the networks [1]. Lamellar nanostructures of unique connectivity were fabricated using polymers of varying compositions of polystyrene (PS) and poly(methyl methacrylate) (PMMA). The selective removal of the PMMA domains followed by metal deposition and liftoff processes enabled the production of interconnected metal networks that could subsequently be isolated and transferred to a flexible substrate.

Introduction:

Figure 1: (a) Molecular structure of PS-b-PMMA. b) FESEM image of the PS-b-PMMA assembly, where the lighter regions are the PS domains.

Electronics have traditionally been limited to rigid and planar surfaces� However, many applications—from computing and energy to sensing and biomedical technology—would be vastly enhanced if electronic materials could be fabricated upon flexible substrates that can undergo mechanical deformation [2]� The fabrication of interconnected metal networks consisting of thin nanowires is thought to provide the flexibility needed while maintaining conductive properties�

Thermodynamically controlled, self-assembling morphologies of block copolymer systems provide a reproducible method of lithography for the production of these networks� P o l y ( s t y r e n e - b l o c k - m e t h y l me thac ry l a t e )—PS-b -PMMA—assembly of specific molecular weight

On top, a blend of copolymer (53 kg/mol PS and 54 kg/mol PMMA) with PMMA homopolymer (20 kg/mol) was spin-coated on the substrates and thermally annealed in vacuum at 180°C for 1200 minutes� The PMMA domain was removed by exposure to 254 nm ultraviolet radiation for 60 seconds, followed by development in acetic acid for 120 seconds� Subsequently, samples were treated with oxygen plasma (40 W, 150 mtorr, 5 sccm O2) for 20 seconds� Via thermal evaporation, 2 nm of titanium was deposited as an adhesion layer followed by 5 nm of gold� After cooling, the samples were sonicated in toluene for 90 minutes at about 50°C� The progress of each step in the network fabrication process was confirmed by a JEOL JSM-7401F field-emission scanning electron microscope�

A patterned support layer can aid the lift off of the metal network from the oxide surface [3]� A layer of PMMA (950 kg/mol, 5�5 wt % solids in anisole) was spin-cast at 1500 RPM and baked at 180°C to yield an approximately 500 nm thick polymer layer� On top, a layer of positive photoresist, AZ P4210, was applied and patterned with 2 µm diameter holes� The samples were overexposed under 320 nm UV radiation and overdeveloped with AZ 400K solution to ensure complete development through the resist to the

ratios result in lamellar, or “fingerprint,” patterns, shown in Figures 1a and b, respectively� These patterns are ideal as masks�

Experimental Procedure:Clean silicon wafers with 289 nm thick, thermally grown silicon dioxide layers were treated by a polymer brush layer with ~ 59 mol % of PS and ~ 41 mol % of PMMA�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


underlying PMMA layer� Oxygen plasma etching (10 sccm O2, 150 W, 200 mtorr) for 170 seconds selectively removed the exposed PMMA regions all the way to the oxide surface� After removal of the photoresist layer by exposing and developing entire surfaces, the samples were immersed in 49% hydroflouric acid (HF) solution. After 200 minutes, the patterned PMMA layer delaminated from the surface� Samples were lightly dried with nitrogen after being rinsed in water� A PDMS (Dow Corning Sylgard 184, 10:1 silicone elastomer base: curing agent) stamp was laminated onto the PMMA layer and slowly peeled away�

To confirm the interconnectedness of the metal network, gold contacts were deposited on the metal network on silicon oxide surfaces using a shadow mask� A two point probe was used to measure conductivity between the gold contacts�

Results and Discussion:The addition of the 20 kg/mol PMMA homopolymer successfully shifted polymer connectivity to the PMMA domain over the PS domain, which leads to better connected metal networks after deposition and liftoff� The UV/acetic acid development process effectively removes the PMMA domain, leaving the PS domain intact, as shown in Figure 2a� The short oxygen plasma etch step shows some promise, but some brush layer or polymer residue may remain, indicated by disconnected islands of metal after deposition� Figure 2b shows that toluene sonication proved to be effective at removing the PS domain without affecting the quality of the metal on the oxide surface�

Two point probe testing provided enough data to confirm the lateral connectivity of networks greater than 50 µm�

IV curves in Figure 3 reveal obvious variance among the recorded resistances of a continuous network, a continuous sheet of gold of the same thickness, and a discontinuous network� Immersion in HF successfully etched away the sacrificial oxide layer through the patterned holes� Figure 4 illustrates the successful transfer of the support layer to a PDMS stamp via preferential van der Waals forces [4] and the deformation of the metal network�

Acknowledgments:I would like to express my deep gratitude to Professor Mark Stoykovich and Ian Campbell for their invaluable guidance and support throughout this project� I would also like to thank the

National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program and the National Science Foundation for the opportunity to participate in this research internship and the staff at the Colorado Nanofabrication Laboratory for their assistance�

References:[1] Stoykovich, M�P� et al� Materials Today� 9, 9-29 (2006)�[2] Rogers, J�A� et al� Proc� Natl� Acad� Sci� USA� 106, 10875–10876

(2009)�[3] Yang, Y� et al� Small� 7, 484-491 (2010)�[4] Meitl, M�A� et al� Nature Materials� 5, 33-38 (2006)�

Figure 4: (a) Polymer support layer laminated on PDMS, (b) Optical micrograph, and (c) FESEM image show the patterned layer and network.

Figure 2: (a) PS mask after PMMA removal, and the b) gold network after PS removal.

Figure 3: IV curve showing unique resistance values for three different conditions.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Overcoming Cellular Breakdown in Hyperdoped Silicon Alloys

Brandon PiercyMaterials Science and Engineering, Case Western Reserve University

NNIN REU Site: Center for Nanoscale Systems, Harvard University, Cambridge, MANNIN REU Principal Investigator: Prof. Michael Aziz, School of Engineering and Applied Sciences, Harvard UniversityNNIN REU Mentor: Daniel Recht, School of Engineering and Applied Sciences, Harvard University Contact: [email protected], [email protected], [email protected]

Abstract and Introduction:Hyperdoped semiconductors, i�e� semiconductors that have been doped far above the solid solubility limit, represent a route towards high-efficiency, scalable, and low-cost solar cells and photodetectors [1]� Ion implantation followed by pulsed laser melting (PLM) has been demonstrated to be an effective method for hyperdoping semiconductors [2]� PLM-created alloys are currently limited in doping concentration by a phenomenon known as cellular breakdown: solidification front instabilities force the solute to come out of solution [3, 4]� This effect is primarily governed by the speed of solidification in the material as described by Hoglund, et al. (1998) [5, 6]. An increased solidification speed should be able to trap more impurities without inducing cellular breakdown, as shown in Figure 1� A faster, lower energy laser pulse will induce a steeper thermal gradient across the melted region, increasing the solidification rate. This work demonstrates that by increasing the solidification velocity from 3-5 m/s to 8 m/s, single-crystal sulfur-silicon (S-Si) alloys of a world-record 2 at% can be created�

Figure 1: Minimum solute concentration of iron and sulfur in Si that can be achieved as a function of solidification velocity, cal-culated using parameters from [6], and accessible velocity regimes for different lasers. Vd = 10 m/s was used as an estimate for iron.

Experimental Procedure:Positive (p)-type silicon (Si) wafers were ion implanted with: 32S+ at 45 keV to a concentration of 1 × 1016 cm-2 (shallow implant), 32S+ at 90 keV to a concentration of 1 × 1016 and 3 × 1016 cm-2 (deep implant), and 56Fe+ at 80 keV to a concentration of 1 × 1016 cm-2� The implantation depth was calculated, using the standard Stopping and Range of Ions In Matter (SRIM) code, to be 100 nm for the 45 keV S sample and 300 nm for the 90 keV S and the iron (Fe) samples� Samples of each type were melted in air using a Continuum Surelite pulsed Nd:YAG laser (355 nm, 4 ns FWHM, and 8 ns total duration) with a single shot� Additionally, the deep-implant S and the Fe samples were irradiated four times with a XeCl excimer laser (308 nm, 25 ns FWHM, and 50 ns total duration)� The deeply implanted S and Fe samples were irradiated using the Nd:YAG at 0�9-1�1 J/cm2, while the shallow-implant S sample was irradiated using the Nd:YAGat 0�5-0�7 J/cm2� Fluence calibration was achieved by irradiating a single-crystalline Si wafer, recording the melt duration through time-resolved reflectivity of a low-power Ar+ laser (488 nm, continuous wave), and comparing the data to predictions from a one-dimensional heat flow simulation [4, 5]� Imaging was performed using a Zeiss Supra field emission scanning electron microscope (SEM).

Results and Discussion:The 1 × 1016 cm-2, 90 keV, S-implanted Si is known to not undergo cellular breakdown when melted with the XeCl laser, and therefore acts as a reference for determining whether breakdown has occurred� Electron backscattering diffraction has previously been used to verify that XeCl-shot Si implanted with 3 × 1016 cm-2 S and 1 × 1016 cm-2 Fe both undergo cellular breakdown� The surface structure of the Nd:YAG-shot Fe-implanted Si (Figure 2) closely resembles the Fe-implanted Si shot with the XeCl laser, implying that cellular breakdown still occurs with iron ions at 1 at%�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


However, in the 3 × 1016 cm-2 S-implanted Si, there was significant variation between the microstructures created from the XeCl laser and the Nd:YAG laser� The pattern visible from the XeCl laser was periodic and on the order of 200-250 nm – unlike the surface structure created by the Nd:YAG laser, which was either a regular pattern on the order of 100-150 nm (Figure 3), or had no surface patterning� This indicates that this sample was at the point where minor variations in fluence across the laser beam were able to cause or prevent cellular breakdown�

Heat flow simulations indicated that the peak solidification velocity, and therefore the maximum solute concentration, could be achieved with a combination of a thin implant layer of 100-150 nm and laser fluence around 0.4 J/cm2� The results from the shallow S-implanted, 1 × 1016 cm-2 Si sample, indicated no regular surface patterning, and were qualitatively identical to samples in which breakdown did not occur (Figure 4)�

Pending verification from XTEM, this represents the first single-crystalline sulfur-silicon alloy above 2 at%� Future work will explore the creation of previously impossible hyperdoped alloys across a range of alloying elements�

Acknowledgements:I would like to acknowledge the exceptional guidance and help of my mentor, Daniel Recht, my P� I�, Michael Aziz, and the rest of the Aziz group, and offer my thanks to the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program as well as the National Science Foundation for supporting this project�

References:[1] M�T� Winkler, D� Recht, M�J� Sher, A�J� Said, E� Mazur, and M�J� Aziz,

Physical Review Letters, 106, 178701 (2011)�[2] B� Bob, A� Kohno, S� Charnvanichborikarn, J� Warrender, I� Umezu, M�

Tabbal, et al�, Journal of Applied Physics, 107, 123506-123506 (2010)�[3] J� Narayan, Journal of Crystal Growth, 59, 583-598, (1982)�[4] M� J� Aziz, C� W� White, J� Narayan, and B� Stritzker, in Energy Beam-

Solid Interactions and Transient Thermal Processing, edited by V� T� Nguyen and A� G� Cullis (Editions de Physique, Paris, 1985), p� 231–236�

[5] R� Reitano, P� Smith, M�J� Aziz, Journal of Applied Physics, 76, 1518-1529 (1994)�

[6] D� Hoglund, M� Thompson, M� J� Aziz, Physical Review B, 58, 189 (1998)�

Figure 4: SEM image of Nd:YAG-irradiated shallowly-implanted S in Si. The white clusters and dark lines are not associated with cellular breakdown.

Figure 2: SEM image of Nd:YAG-irradiated Fe-implanted Si, showing clear signs of cellular breakdown.

Figure 3: SEM image of Nd:YAG-irradiated, deep-implant S in Si at 3 × 1016 cm-2, showing what may be cellular breakdown but with little periodicity.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Surface Characterization of Etched Micro- and Nano-Structures in Silicon for Phonon Heat Transport

Victoria SavikhinElectrical and Computer Engineering, Purdue University

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Dr. Richard Robinson, Materials Science and Engineering, Cornell UniversityNNIN REU Mentors: Dr. Jared Hertzberg, Materials Science and Engineering, Cornell University;

Obafemi Otelaja, Materials Science and Engineering, Cornell UniversityContact: [email protected], [email protected], [email protected], [email protected]

Introduction:

In insulators, heat is carried by excited vibrational states (phonons) with wavelengths from a few nanometers to several hundred nanometers (nm)� Phonons travel through solids ballistically until they are scattered by phonon-phonon interactions, imperfections in the crystal, or boundaries [1]� At low temperatures and device dimensions comparable to phonon wavelengths, surface reflection becomes the primary factor in scattering� Phonons scatter from surfaces with roughness larger than their wavelengths [2]� Because the efficiency of thermoelectric devices is inversely proportional to thermal conductivity [3], surface roughness of devices must be quantified for a better understanding of heat flow.

While the scanning electron microscope (SEM) can be used to visualize topographical outlines and the atomic force microscope (AFM) can be used to measure roughness of near-horizontal surfaces, current microscopy tools lack the capability to measure roughness of vertical or near-vertical sidewalls� The focus of this project was to develop and test a method to characterize the vertical surfaces of nanowalls� Three different techniques were investigated� The most effective method was used to characterize sidewalls created by varying etch processes�

Etched silicon substrates were used�

Experimental Procedure:Method 1: Break and Tip Walls. Zyvex S-100 tungsten nanoprobes with a tip radius greater than 2 µm installed on the Zeiss Ultra-55 SEM and operating under vacuum during imaging were used� Each probe was placed with 5 nm precision adjacent to a studied silicon nanowall roughly 200 nm thick and 500 nm tall� Piezoelements pushed the probe against the wall with the goal of cracking it at the base and laying it flat for consequent AFM spatial profiling.

Method 2: Tilt Sample in AFM. An aluminum wedge was fabricated to mount samples at 30° to the horizontal in the AFM� The 90° sidewalls therefore became tilted

approximately 60° from the horizontal, requiring a tip with an inner half-cone angle less than 30° for successful visualization� NanoScience Aspire CT300 conical tips were used for this purpose� Samples used included 1�8 × 2�2 µm squares etched 200 nm deep into silicon using the focused ion beam (FIB) and silicon mesas etched using the Unaxis 770 deep Si etcher for 5 and 10 minutes generating 1�2 µm and 2�4 µm deep structures respectively�

Method 3: Break and Tilt Substrate. A diamond cutter was used to cleave samples along crystal planes with a precision of approximately 2 mm. Walls sufficiently close to edges of the samples were found using an optical micro scope, and samples were mounted vertically using an aluminum block and observed in the AFM using non-contact tips� Samples used had silicon structures etched in a continuous passivation process using the Unaxis 770 deep Si etcher for 5 and 10 min� Walls parallel and at 45° to the cleave planes were imaged�

Results:Method 1: Break and Tip Walls. The probes were successfully used for cracking pieces of the nanowalls� However, only small sections of walls that had come into direct contact with the probes were knocked over, possibly affecting the observable roughness of the exposed surfaces� Moreover, the pieces were not stationary on the substrate and this method was found to be unsuitable for profiling the nanowall surfaces�

Method 2: Tilt Sample in AFM. Most scans showed unexpected convex curving toward the top of walls, inaccurate wall slants, and other AFM interferences indicating untrustworthy roughness measurements� Only a few accurate scans could be used for sidewall roughness calculations, showing a 68 nm z-range and 10 nm RMS value on a 3 × 3 µm scan� The method was considered too unreliable for use without further improvements�

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


Method 3: Break and Tip Substrate. All samples tested yielded AFM scans that corresponded well to structures observed under SEM (see Figure 1)� Walls parallel to cleave planes were most easily observable� The method was used to measure surface roughness of samples retaining a layer of polymer deposited during continuous passivation etch and after the polymer was stripped using the Gasonics Aura 1000, with results shown in Figure 2� The data was inconclusive as to the effects of the fabrication step but demonstrated that this method can be used reliably and quantifiably to analyze the effects of fabrication conditions.

Conclusions:Method 1 was not viable for future use� Method 2 was the least destructive and most applicable of all methods, but was too unreliable for immediate use, requiring further improvement� Method 3 was the most destructive method in that an entire device would be cleaved in half for measurement, but at the same time was the most reliable and therefore the best method for immediate use�

Future Work:The phonon spectrometer shown in Figure 3 will be used to find thermal conductivity through silicon nanowalls experimentally, and the relationship between conductivity and nanowall length will be established� The schematic form of the expected relationship and determination of mean free path (MFP) is shown in Figure 4� Fabrication processes will be varied to alter surface roughness and obtain the relationship between roughness and MFP, which will aid researchers in modeling thermal conductivity in devices�

Acknowledgements:I would like to thank the Cornell NanoScale Science and Technology Facility for hosting my research and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program for funding; Rob Ilic and Melanie-Claire Mallison for coordinating the CNF REU Program; Daron Westly, Lynn Rathbun, and other CNF staff for their constant guidance; and above all my mentors and PI for enabling my research and helping me every step of the way�

References:[1] Kittel� Introduction to Solid State Physics� 2004, Wiley�[2] Heron et al, O� Nano Letters 2009, 9, 1861-65�[3] Hochbaum et al� Nature Publishing Group 2008, 451, 163-7�[4] Veeco Nanoscope Analysis Software�

Figure 1: A comparison of SEM (left) and AFM (right) scans of the same structure using Method 3.

Figure 2: Average roughness of nanowall sides over 2-15 µm2 area with horizontal resolution 3.5-47 nm [4].

Figure 3: Model of the phonon spectrometer.

Figure 4: Schematic form of the expected relationship of phonon transmission versus nanowall length.

• m

ate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•

mate

rials

•


Photonic templates of self-assembled spherical colloidal particles have generated interest for their applications in integrated optical devices. However, stronger light-matter interactions have been predicted for colloids with complex geometries. Although simulations have shown that large and robust photonic band gaps may be achieved by lowering crystal symmetry using asymmetric scattering units [1], empirically verifying the bandgap properties of nonspherical two-dimensional (2D) monolayer and quasi-2D transition colloidal crystals has been limited by the lack of large area samples.

This project used colloidal self-assembly via gravitational sedimentation in height-confined cells to grow large photonic templates with controllable phase. Wedge-geometry cells [2] constructed using photoresist promoted the growth of colloidal crystals from mushroom cap-shaped particle building blocks (Figure 1) that measured an order of magnitude greater than previously achieved. The hexagonal and unconventional rotator and buckled phases were observed by confocal microscope and characterized using positional and orientational correlation functions as well as order parameters. A parallel plate-geometry cell [3] is being developed as the next cell refinement to produce larger crystals.

Figure 1: SEM image of mushroom cap particles at 25000X magnification.

Confinement Assisted Self-Organization of Photonic Templates

Daryl I. VulisElectrical Engineering, SUNY Stony Brook

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Professor Chekesha Liddell Watson, Materials Science and Engineering, Cornell UniversityNNIN REU Mentor: Erin Riley, Materials Science and Engineering, Cornell UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Approach:By creating a wedge cell with a smaller spacer, and thus lower minimum angle, a significantly larger cell area restricted to two or less particle diameter height encouraged the growth of monolayers and bilayers (Figure 2)� The next design iteration employed parallel plate geometry with a fixed height and a filter at the bottom edge of the cell to increase colloidal concentration� The controllable height ranged from 900 nm to 1�6 µm� During the course of this project, we completed and tested the wedge cell, and began fabrication of the parallel plate cell�

Methods:Patterning the Wedge Cell Coverslip. Glass microscope coverslips were scrubbed sequentially with acetone, isopropyl alcohol, and deionized (DI) water� Primer (P-20) and photoresist (Shipley 1800 series S1827) were spun onto the coverslips and baked after each spin step to obtain a target height of ~ 2�7 µm� Due to their unconventional rectangular shape, coverslips were exposed using the ABM contact aligner and hand developed in 726MIF developer�

Figure 2: Schematic of wedge cell highlights phases as a function of height.

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials

• mate

rials


Assembling the Wedge Cell. Wedge cell construction consisted of two coverslips and a 1 × 3-inch support glass slide� A blank microscope coverslip and the unpatterned portion of a developed coverslip were washed in a NaOH, ethanol and DI water solution of pH 14 to remove contaminants� The blank coverslip was bonded to the support slide using UV adhesive� The patterned coverslip was placed on the blank coverslip with the photoresist between the glass surfaces� Pressure applied by hand to the top coverslip created a zero-height region at the base of the cell� The sides and bottom of the cell were sealed with adhesive� To reduce colloid absorption to the cell walls, the cell was filled with 15 µL of polyvinylpyrrolidone (PVP) and DI water solution via pipette through the unsealed edge, and placed in a low pressure chamber to boil off excess. The cell was filled with approximately 15 µL of the colloid suspension in PVP, sealed, and tilted at ~ 80º to promote gravitational sedimentation of the particles into the monolayer and bilayer regions�

Patterning the Parallel Plate Cell Wafer. Fused silica wafers were coated with ~ 200 nm of polycrystalline silicon (polysilicon)� ARC AR3 primer and UV210 photoresist were consecutively spun and baked onto the wafers for a total height of ~ 900 nm of resist� The wafers were exposed using the ASML 300C DUV stepper� Plasma etching was required to achieve the target maximum height of 1�6 µm� The Oxford 80 etcher was used to remove the primer, and additional etching with the Oxford 100 etcher relied on the polysilicon as a mask to encourage increased fused silica etching�

Results:Wedge Cell Results. Confocal data was collected using a Zeiss LSM 5 LIVE confocal microscope. The refined wedge cell produced an unconventional rotator phase colloidal crystal monolayer (Figure 3), an order of magnitude greater than achieved by the previous wedge cell — specifically 7�4 mm by 300-500 µm compared to former crystals of a couple hundred micro-meters� The asymmetric properties of the rotator phase suggest interesting optical properties� Additionally, square bi-layer phase crystals had begun to form at larger confinement heights (Figure 4); this phase has been shown in simulations to have multiple bandgaps�

Crystal Analysis. Representative confocal micro graphs of the rotator monolayer and square bi-layer phases were char-

acterized using a Fast Fourier Transform (FFT) and a radial distribution function to determine rotational and translational order, respectively� A Voronoi diagram was generated to identify crystal defects and grain boundaries� However, these sedimentation samples required additional time to organize high-quality crystals prompting the analysis of more mature crystals in previously fabricated wedge cells to better illustrate the phases observed�

Conclusions and Future Work:The refined wedge cell enhanced the growth of single phase monolayers by an order of magnitude� The parallel plate cell is expected to increase the crystal size to centimeters enabling an empirical verification of photonic bandgaps via transmission and reflection spectra.

Acknowledgements:I thank Professor Chekesha Liddell Watson, Erin Riley, and the Liddell Group for their guidance and insight, as well as the Cornell NanoScale Science and Technology Facility and its staff, especially Noah Clay, John Treichler, and Meredith Metzler, for their support throughout fabrication� Lastly, I thank the National Science Foundation and National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program for providing this amazing opportunity, and Melanie-Claire Mallison and Rob Ilic for organizing throughout�

References:[1] Xia Y, Gates B, and Li Z -Y; “Self-assembly approaches to 3D

photonic crystals”; Advanced Materials 2001, 13, 409-413�[2] Riley E and Liddell C; “Confinement-controlled self assembly of

colloids with simultaneous isotropic and anisotropic cross-section”; Langmuir 2010, 26, 11648-11656�

[3] Park S, Qin D, and Xia Y; “Crystallization of mesoscale particles over large areas”; Advanced Materials 1998, 10, 1028-1032�

Figure 3: Confocal image of rotator phase in refined wedge cell.

Figure 4: Confocal image of square bilayer phase in refined wedge cell.

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

•


Fabrication of Microfluidic Devices for Synthesis of Janus Particles

Brittany AlphonseBiomedical Engineering, University of Rhode Island

NNIN REU Site: Nanotechnology Research Center, Georgia Institute of Technology, Atlanta, GANNIN REU Principal Investigator: Dr. Todd Sulchek, Assistant Professor,

George W. Woodruff School of Mechanical Engineering Georgia Institute of TechnologyNNIN REU Mentor: Kipp Schoenwald, Doctoral Student in Department of Mechanical Engineering,

George W. Woodruff School of Mechanical Engineering Georgia Institute of TechnologyContact: [email protected], [email protected], [email protected]

Abstract:

Microfluidic flow focusing devices (MFFD) are an easier, more accurate and repeatable way to create monodispersed Janus particles. Size, shape, and uniformity of droplets determine particle application, which is dependent on channel width. Decreases in channel width allows for more biological applications. The goal of this study was to fabricate a MFFD to synthesize Janus particles using a T-junction design using three immiscible liquids. To do this, a mask was designed with the negative image of five of the same MFFDs to reduce cost, material waste, and increase repeatability. Quartz wafers were used due to lower etch rate, smoothness, and purity compared to other transparent materials. An adhesion layer was spun prior to photoresist to help with developing. Chromium was deposited and etched to create the channels to be fusion bonded to Pyrex® wafers. Ongoing studies will apply this design process to recreate MFFDs with 15 µm and 3 µm channel widths.

for more biological applications including drug delivery, diagnostic testing, and biosensing [1]�

A MFFD was created due to ease of use, design, repeatability, and accuracy� A T-junction design uses cross flowing streams to create the forces for pinchoff with the shear force and interfacial tension, shown in Figure 1� This geometry can form droplets with nontraditional shapes and morphologies with precise control [2]�

Background:Here, T-junction configurations with two immiscible co-flowing liquids were tested as a means of creating droplets ranging from 1 to 3 µm� The popularly tested PDMS-on-silicon was not able to be tested for the smaller channel widths; due to the size of the channel versus the O2 layer, the corresponding wettability for the smaller channel sizes with the hydrophilic polydimethylsiloxane (PDMS) method prevents the correct morphology of the Janus droplet—hence, the quartz-on-Pyrex method [1]�

T-junctions create droplets of equal size with steady flow from the upstream pressure due to the junction shape� Droplet formation occurs when viscous stresses overcome interfacial tension [3]� The size of the droplets can be altered by channel width and the forces that change the fluid flow rates [4]�

Figure 1: MFFD with three immiscible liquids.

Introduction:Microfluidic techniques for Janus emulsions composed of two immiscible liquids have the continuous phase coming from either side of the device with the dispersed phase through the center channel in stokes flow [1]. A MFFD creates Janus emulsions through flow-focusing, where the liquids are forced through a narrow channel causing symmetric shearing [1]� Decreases in droplet size allows

• mech

anica

l device

s • m

ech

anica

l device

s • m

ech

anica

l device

s •


Methods:Masks were designed using AutoCAD software for the three different channel thicknesses 50 µm, 13 µm, and 3 µm. On each mask is the negative image of five of the same MFFD, single MFFD shown in Figure 2� An adhesion layer, Microprime Primer P-20, was spun on the 4-inch quartz wafer prior to the photoresist to help the thin long lines to adhere� Photoresist SC1827 was spun to a thickness of 2�7 µm and then patterned� Chrome was deposited over the pattern in a uniform 1 µm layer� For the quartz to be etched, the 4-inch quartz wafer was mounted on a 6-inch silicon carrier using crystalbond or fomblin� Schematic of channel formation shown in Figure 3�

Results and Conclusions:Significant progress towards fabricating a MFFD with a T-junction design was accomplished as well as creating protocols for lithography, deposition, etching, and bonding� During early stages of lithography the resist puckered, the thin long lines washed away during developing and there was not a uniform development� This was corrected using an oven and glass slides to create a uniform heating surface� During chrome deposition there was unequal deposition and peeling� By lowering the deposition rate, increasing the deposition time and using a CVC DC Sputter, this was corrected� During etching, the wafer cracked due to the different thermal properties of the quartz and silicon� By changing the etch process, it is possible to prevent cracking�

Figure 4: Final design of microfluidic flow focusing device.

Figure 2: T-junction design with specified features.

Figure 3: (A) Adhesion layer spun on quartz wafer; (B) Photoresist spun 2.7 µm thick; (C) Lithography of resist; (D) Resist developed; (E) Cr deposited 1µm thick; (F) Resist removal; (G) Quartz etched pattern; (H) Cr removed; (I) Fusion bond to Pyrex.

Future Work:Future studies will focus on finishing the fabrication of the 50 µm MFFD, shown in Figure 4� After etching, the two wafers will be separated with heat and the chrome removed� This segment of quartz will be fusion bonded to the Pyrex slide� Pyrex was chosen to insure a transparent top for real time imaging of droplet formation� Ferrule connectors will be placed into the channel, top down, to help thread the tubing into place and UV epoxy will be used to secure the tubing� The tubes will be attached to syringes to control various flow rates and a UV light source will be at the other end of the channel to continuously photopolymerize the droplets [5]� Work will also be done in creating a 15 µm and a 3 µm MFFD with the protocols that were created� Tests will be done with the immiscible liquids and determine how flow rates affect droplet formation.

Acknowledgments:I would like to thank Dr� Todd Sulchek and Kipp Schoenwald for their guidance; Mikkel Thomas, Janet Cobb-Sullivan, and the cleanroom staff; GT NNIN Coordinators Katie Hutchinson, Joyce Palmer, and Nancy Healy; and Baret Kilbacak and Venkat Goli for their parallel research� Also, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and the National Science Foundation for funding�

References:[1] Teh, S�;Lin, R�; Hung, L�; Lee, A� J� P� R� Soc� Chem� 2008�[2] Nei, Z�; Xu, S�; Seo, M�; Lewis, P� C�; Kumacheva, E� J� Am� Chem�

Soc� 2005�[3] Christopher, G� F�; Anna, S� L� J� Phys� D: Appl� Phys� 2007�[4] Xu, S�; Nei, Z�; Seo, M�; Lewis, P�; Kumacheva, E�; Stone, H� A�;

Garstecki, P�; Wiebel, D� B�; Gitlin, I�; Whitesides, G� M� Angew� Chem� Int� Ed� 2005�

[5] Nisisako, T�; Hatsuzawa, T� Springer� 2010�

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

•


Development of Paper Accelerometers for Cheap Applications

Brendon Lee GobertMath and Science, Blackfeet Community College

NNIN REU Site: Howard Nanoscale Science and Engineering Facility (HNF), Howard University, Washington, DCNNIN REU Principal Investigator: Dr. Gary Harris, Electrical and Computer Engineering,

Director of HNF, Howard University, Washington, DCNNIN REU Mentor: Dr. William Rose, HNF, Howard University, Washington, DCContact: [email protected], [email protected], [email protected]

Abstract:

The purpose of this project was to explore the use of paper in the fabrication of a micro electro-mechanical sensing device. To fabricate the sensors, we constructed a cantilever from paper with piezoresistive carbon ink painted on. Conductive silver ink was used to form the contacts. Both bamboo and cellulose paper were used for this process. The overall sensitivity of these micro electro-mechanical (MEMS) paper devices is approximately 120 µN/Ω compared to a similar silicon device, which is typically 80 µN/Ω. These paper MEMS sensors are cheap and easy to fabricate, often in less than one hour. A comparative silicon device would be far more time-consuming and would require the use of a clean room. Our paper MEMS devices would therefore find applications in less developed countries.

Introduction:

The implementation and use of MEMs has substantially grown in the last three decades [1]� MEMs are used in many important applications of today from engineering to medicine� These applications include; accelerometers, toys, airbags, analog devices, digital devices, and instruments� Silicon MEMs devices are the primary devices being used today� Fabrication of these silicon devices require many hours in both the lab and clean room� Decreasing the time of fabrication and eliminating the need for clean room usage would substantially reduce the cost of production� New MEMs technology using paper can lower this cost [2] without substantially lowering performance�

Experimental Procedure:A cantilever was made from cellulose or bamboo paper using a laser cutter� The piezoresistive material graphite ink was painted on� Figure 1 shows the device� After the graphite ink dried, contact pads were fabricated using silver ink� The total fabrication time for each device was approximately one hour�

Figure 2 shows the setup of the Wheatstone bridge we used in measuring the device resistance� A series of resistance measurements was taken with different forces applied to the cantilever, for both the cellulite and the bamboo paper devices� The corresponding change in resistance was calculated and tabulated� Graphs of the change in resistance

Figure 1: Paper cantilever device.

Figure 2: Wheatstone bridge used in measuring the device resistance.

• mech

anica

l device

s • m

ech

anica

l device

s • m

ech

anica

l device

s •


versus force were made for the different devices� Additionally we calculated Young’s modulus for both the cellulose and bamboo devices, and achieved readings of 5�5 gigapascals (GPa) for the cellulose and 2�5 GPa for the bamboo�

Results and Conclusions:Figure 3 shows that the cellulose paper has a greater linear change in resistance� Figure 4 shows that the bamboo paper has a wider range in force in micro-Newtons (µN)�

The cellulose paper is excellent in measuring the sensitivity of the slightest change in force� The bamboo is twice as thick as the cellulose and can measure 2 µN more in force compared to the cellulose�

Future Work:In the future, we will use trichlorosilane vapor to make the paper hydrophobic, which would aid when working in moist or humid conditions� We also could use more layers of paper glued together, thereby increasing the overall stiffness and sturdiness of the cantilever� Finally, testing the resistance on various lengths of paper cantilever would be advantageous�

Acknowledgments:I would like to thank Dr� William Rose, Dr� Gary Harris, and all other members of the Howard Research Group� I am also grateful to the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program and the National Science Foundation�

References:[1] Liu, Mwangi, et al� “Paper-Based Piezoresistive MEMS Sensors�” (2011):

In Print�[2] Patel, Prachi� “Paper Accelerometer Could Mean Disposable Devices�”

IEEE spectrum� (2011): In Print�

Figure 3: Linear change in resistance for cellulose paper.

Figure 4: Range in force for bamboo paper, in µN.

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

•


Dimensional Analysis of Microlitre-Sized Microbial Fuel Cells

Nicole HamsBiochemistry, University of Missouri - Columbia

NNIN REU Site: ASU NanoFab, Arizona State University, Tempe, AZNNIN REU Principal Investigator: Dr. Junseok Chae, Department of Electrical Engineering, Arizona State UniversityNNIN REU Mentor: Dr. Seokheun Choi, Department of Electrical Engineering, Arizona State UniversityContact: [email protected], [email protected], [email protected]

Abstract:

A microbial fuel (MFC) is a bioelectrical device that uses a microbe to oxidize a substrate and captures the electrons that would normally continue through the microbes’ electron transport chain. The captured electrons at the anode are conducted through load (resistor bank) and are reduced at the cathode to complete the bioelectrical circuit; this is how MFC converts chemical energy into electricity. With today’s emerging drive towards a more sustainable energy source, macro-sized MFCs use biomass, such as sludge, to generate electricity from any bacteria present in the media. However, applicability of such technology remains low due to only generating power densities of 10 W/m2. Such low power density cannot compete with power sources in macro-size. In this research we aim to miniaturize macro-sized MFCs by taking advantage of microelectromechanical systems (MEMS) technology. As a result, research interest is to studying a micro-sized MFC for portable power sources and to better understand variables in order to optimize power density. Our study monitors how dimensional parameters effects power density as well as directly comparing power density and columbic efficiency of a micro-sized and macro-sized fuel cell.

Introduction:A microbial fuel (MFC) is a closed-system bioelectrical device that uses a microbe to oxidize a substrate and captures the electrons that would normally continue through the microbes’ electron transport chain� The captured electrons are conducted through an anode, resistor, and cathode and membrane is placed between the anode and cathode plates to maintain electro-neutrality� In essence, microbial fuel cells use bacteria to directly convert chemical energy into electricity�

With today’s emerging drive towards a more self-sustainable energy source, microbial fuel cells are considered a viable candidate� Conventional macro-sized MFC’s use biomass, such as sludge, to generate electricity from any bacteria present in the media� Although such macro-size MFC’s are capable of producing power densities of up to 10,000 mW/m2 [1], the applicability of such technology remains low due to practicality; macro-sized fuel cells are often large, and therefore portability is limited� As a result, research interest has shifted to studying micro-sized MFC’s in order to better optimize power output and increase portability�

The objective of this research is to monitor how dimensional parameters effect power density and columbic efficiency. We hope to develop this technology to eventually power small electrical devices such as nanosensors�

Materials and Methods:Device Fabrication. Four different anode devices were made, 50 mm2, 100 mm2, 200 mm2 and 400 mm2� The micro-MFC device was constructed by mechanically drilling six 0�9 mm diameter holes into glass slides using a Craftsman drill press: one inlet, one outlet and four screws� The glass was then coated with 10 nm chrome and 100 nm gold using a physical vapor deposition technique. Nafion (Sigma-Aldrich, St� Louis, MO) proton exchange membrane was cut to the dimensions of the glass and sandwiched between rubber silicone and a gold-surfaced glass slide, defining the anode and cathode chambers� Electrical contacts were established by attaching copper tape to the anode and cathode sides of the glass� The nanoports were then attached to the inlet and outlet holes with an epoxy adhesive and the device was bolted together with four 0�9 mm nuts and bolts�

Medium and Microorganism. Geobacter sulferreducens inncoluted in 25 mM acetate media was used as the anolyte, electron donor, and 50 mM ferricyanide was used as the catholyte, electron acceptor�

Experimental Procedures and Calculations:Nanotubes were attached to the nanoports on both the anode and cathode side� The anolyte and catholyte were pumped

• mech

anica

l device

s • m

ech

anica

l device

s • m

ech

anica

l device

s •


through the nanotubes at a flow rate of 1 µl/minute for one hour to determine the open circuit voltage� Data was collected every one minute via LabView and analyzed in Microsoft Excel� Once the baseline of 500 mV was established, the 100 mm2 and 400 mm2 devices were closed via a 150Ω resistor� The remaining devices, 50 mm2 and 200 mm2 were not used because they did not achieve the high enough open circuit voltages� The remaining two cells were left to run for 17 days, with data points being collected every one minute� Current through the resistor was calculated using I = V/R and power using P = IV, where I is current (amps), V is voltage (volts), R is resistance (ohms) and P is power (watts)�

Results:During data collection, the current was monitored on a daily basis by plotting the data points in Microsoft Excel� After 120 hours, fresh anolyte was added to both systems and a drastic increase in current from the 400 mm2 device (Figure 1) could be seen as a result�

After 160 hours of collecting data, we compared the open circuit voltage (voltage without a resistor) to the voltage across different values of resistors� Ohms law, I = V/R, was applied to that data to generate a polarization curve (Figure 2). From the polarization curve, P = IR was used to find the total power output of each system (Figure 3)� The power density and current density were calculated by dividing power output and current output by area in meters (Figure 4)� Power densities of up to 23 µW were achieved�

References:[1] S� Choi, H-S� Lee, Y� Yang, P� Parameswaran, C� I� Torres, B� E�

Rittmann, and J� Chae, “A µl-scale Micromachined Microbial Fuel Cell Having High Power Density,” Lab-on-a-Chip, v� 11, pp� 1110-1117, 2011�

Figure 4: Power density and current density calculated by dividing power output and current output by area in meters.

Figure 1: Results of adding fresh anolyte to the 400 mm2 device.

Figure 2: Generated polarization curve.

Figure 3: Total power output of each system.

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

•


Nanoelectromechanical systems (NEMS) mechanical resonators are capable of detecting single atoms or molecules [1-3] and can potentially be used as compact, sensitive, and minimally destructive alternatives to traditional analytical chemistry techniques. Graphene, with its low mass, large area, and electrical conductivity, is an even better sensor of mass per unit area that can detect both mass and charge. The purpose of this project was to fabricate an electrically contacted graphene resonator from graphene grown by chemical vapor deposition (CVD). The device consisted of graphene suspended over a circular trench and contacted with platinum electrodes. Graphene electromechanical resonators have potential applications in biological sensing, gas chromatography, and mass spectrometry.

Introduction:

Graphene Resonators for Mass and Charge Sensing

Reyu SakakibaraChemical Biology, University of California, Berkeley

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Professor Jeevak Parpia, Department of Physics, Cornell UniversityNNIN REU Mentor: Robert A. Barton, Jr., Department of Applied and Engineering Physics, Cornell UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Figure 1: Device layout, consists of trench over which to suspend graphene and electrodes with which to contact the graphene.

The graphene resonator senses mass through its resonant frequency: adsorption of particles onto a resonator changes its mass and resonant frequency� The graphene is actuated in two separate ways� One is to apply an AC voltage across the back gate, which drives graphene resonance by applying a periodic attractive force� The other method is to heat the device at its resonant frequency with a laser� Light absorption and the associated temperature change result in a change in tension in the graphene that mechanically actuates the device� The motion can be detected using optical inteferometry, in which a laser illuminates the graphene and the interference pattern of the light reflected from the graphene and the light reflected from the back gate is measured�

Measurements of conductivity give information about the effective doping and the charge on the graphene� Applying a voltage across the electrodes induces a current through the graphene, which can be measured to infer conductance� Additionally, applying a voltage on the back gate varies the Fermi level of the graphene and thus the concentration of charge carriers of the graphene� Variation in the Fermi level can be detected as a change in conductivity, which is proportional to the product of mobility and number of charge carriers�

Experimental Procedure:Fabrication was based on that described by a previous paper [4]� Fabrication began by growing about 240 nm of oxide on a 10 kΩ•cm silicon (Si) wafer. The first level of features, trenches over which to suspend the graphene, was exposed using a deep ultraviolet stepper, which required spinning both photoresist and anti-reflective coating, or ARC� Following the lithography, trenches were etched into the wafer using deep reactive ion etching� An ARC descum

Figure 2: SEM image of desired undercut profile in the silicon, taken after the silicon etch.

• mech

anica

l device

s • m

ech

anica

l device

s • m

ech

anica

l device

s •


was performed with oxygen and argon plasma� The silicon oxide was etched through with either CHF3/O2 or CHF3/O2 followed by CF4� Silicon was etched with SF6/O2, first anisotropically and then isotropically to create an undercut profile. An undercut profile was desired to prevent metal from evaporating onto the sidewalls, which would short the device� The desired trench depth was about a micron�

After stripping the photoresist and ARC, an additional 220 nm of oxide was grown� Next, the gate, source, and drain electrodes were patterned using photolithography, followed by an evaporation of metal (5 nm titanium and 25 nm platinum)� The resist and ARC were removed in a liftoff process using sonicating n-methyl-pyrrolidone�

Graphene was grown via CVD on copper foil and transferred using a method cited previously [5]� First, graphene on copper foil was patterned with a checkerboard pattern of 50 µm by 50 µm squares and etched using oxygen plasma� Then poly(methyl methacrylate) (PMMA) was spun onto the surface of the copper and the copper was etched in ferric chloride� The graphene was rinsed in water and transferred

onto the wafer� The graphene squares landed randomly when transferred� The PMMA was removed and the devices were critical point dried to prevent stiction�

Results and Conclusions:Profilometry and interferometry were used to characterize the etches and the oxidation� The initial oxidation yielded 243 nm of oxide� The SF6/O2 etched 1170 nm of Si with less than 0�1% nonuniformity across the wafer� Scanning electron microscopy (SEM) confirmed an undercut profile in the Si� Roughly 210 nm of oxide grew within the trenches, while outside, the oxide grew to 372 nm� Optical microscopy confirmed correct alignment of second layer exposure. SEM was used to find working devices: devices that were not shorted and with graphene that was not ruptured or stuck to the bottom of the trench. Raman spectroscopy confirmed that graphene was successfully grown and transferred�

The largest devices had a diameter of 12 µm� The resonant frequencies of the devices were on the order of megahertz (MHz); for example, the 6 µm devices had a frequency range of 30 to 40 MHz�

Future Work:Future experiments will include using both the mechanical resonance and the conductance of these devices to study biological and chemical analytes adsorbed to graphene�

Acknowledgments:This work was supported via the National Nanotechnology Infrastructure Network Research Experience for Under-graduates (NNIN REU) Program by the National Science Foundation Grant No� ECS-0335765� This work was performed at the Cornell NanoScale Science and Technology Facility (CNF)� Special thanks to Professors Jeevak Parpia and Harold Craighead, mentor Robert Barton, Program Coordinators Melanie-Claire Mallison and Rob Ilic, the CNF staff, and the Craighead Group�

References:[1] Jensen, K�, et al� An atomic-resolution nanomechanical mass sensor�

Nat� Nanotechnol� 3, 533-537 (2008)�[2] Schedin, F�, et al� Detection of individual gas molecules adsorbed on

graphene� Nat� Mater� 6, 652-655 (2007)�[3] Naik, A�K�, et al� Towards single-molecule nanomechanical mass

spectrometry� Nat� Nanotechnol� 4, 445-450 (2009)�[4] van der Zande, A�M�, et al� Large-Scale Arrays of Single-Layer

Graphene Resonators� Nano Lett� 10, 4869-4873 (2010)�[5] Li, X�, et al� Large-Area Synthesis of High-Quality and Uniform

Graphene Films on Copper Foils� Science 324, 1312-1314 (2009)�[6] Reina, A�, et al� Large Area, Few-Layer Graphene Films on

Arbitrary Substrates by CVD� Nano Lett� 9, 30-35 (2009)�

Figure 3, above: SEM image after graphene transfer; dark squares are graphene.

Figure 4, below: SEM image of a completed device with suspended graphene.

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

•


Using MEMS Sensor Array to Map the Temperature of Hot Springs in Yellowstone National Park

Karl SchliepChemistry and Mathematics, University of Minnesota Morris

NNIN REU Site: ASU NanoFab, Arizona State University, Tempe, AZNNIN REU Principal Investigator: Hongyu Yu, School of Electrical, Computer, and Energy Engineering,

and School of Earth and Space Exploration, Arizona State UniversityNNIN REU Mentor: Jonathon Oiler, School of Earth and Space Exploration, Arizona State UniversityContact: [email protected], [email protected], [email protected]

Abstract and Introduction:During its exploration of Mars, the Spirit exploration rover uncovered areas of high concentration of silica, which was proposed by scientists to be the evidence for possible ancient hot springs that may have been similar to those that are currently present on Earth� One such analog environment on earth may be the hot springs located at Yellowstone National Park (YNP), which are teeming with thermophilic organisms, both chemosynthetic and photosynthetic� This is an important revelation because if the hot springs on Mars are at all similar to those at YNP, then they too might have been home to similar kinds of life� At the same time, there has been a goal of many researchers to understand the conditions under which life can exist in the hot springs of YNP�

Yellowstone National Park hot springs’ ecological environ-ment consists of areas where microbial communities thrive juxtaposed to areas where no life is present� The abrupt transitions between these regions happen on a millimeter-scale, rendering commercial sensors impractical for measuring the changes in the potential of hydrogen (pH), temperature, flow, and conductivity (factors which may influence the type of life found at any particular location) across these transitions� Microelectromechanical systems (MEMS) technology, enabled by the development of chemically and thermally tolerant materials, can be employed in the exploration of the microbial boundaries in the hot springs�

The focus of our research was on the fabrication and utilization of MEMS thermistor arrays to measure the temperature of the microbial boundaries in the hot springs of YNP�

Experimental Procedure:Silicon dioxide (SiO2) was deposited onto one side of Si wafers using plasma-enhanced chemical vapor deposition (PECVD)� On top of the SiO2, temperature-sensitive resistors (thermistors), made of titanium/platinum bi-layers, were patterned in a linear array at spatial intervals of 5 mm using photolithographic methods� Several wafers were made with varying thicknesses of the bi-layer metals to optimize the temperature coefficient of resistance (TCR).

Thermistor resistances were measured at various locations on the wafer at different temperatures in order to determine the TCR via the equation:

where a is the TCR, ΔR is the change in resistance, ΔT is the change in temperature from the initial set point of 30°C, and R0 is the initial resistance at 30°C�

Photolithography and the metal deposition were employed again to produce leads of copper on chromium to connect the bond pads to the thermistors� The sensor arrays were then coated with a biocompatible, chemically inert and water-resistant polymer, Parylene-C� To enable the thermistors to be wired, the Parylene-C was etched to provide access to the bond pads� The wafers were diced producing six arrays per four-inch wafer, each with 12-15 resistors in an array configuration. The array allowed for simultaneous temperature measurements and the data could be utilized in mapping the temperature gradient� Wires were then soldered to the bond pads on the arrays and later connected

• mech

anica

l device

s • m

ech

anica

l device

s • m

ech

anica

l device

s •


to a data acquisition card (DAQ)� The wire connections were coated with a commercial epoxy for increased mechanical strength�

Using the DAQ, each array was calibrated in a hot water bath between 40 and 100°C� The DAQ was programmed to measure the voltage drop across a known resistor in series with a thermistor to determine the current through the circuit following Ohm’s law I = V/R�

This current and the resulting voltage drop across the thermistor were used to calculate the resistance of the thermistor, whose value could be paired to a specific temperature.

Results:The results of how the varied thicknesses of metal affect the TCR are compiled in Table 1� The sensors with the best TCR were used in the field at YNP. The measurements from one of the hot springs are detailed in Figure 1� The data was taken by measuring at one spot, averaging the temperature readings across the array, and then sliding the sensor back in a straight line before taking the next measurement� At location A no life existed� At location B life existed as a green-black microbial community� At location C the life transitioned from a green microbial community to an orange/brown microbial community�

Figure 1: Surface gradient across microbial boundary of pH 6 hot spring using our MEMS temperature sensor.

(See cover for full cover version.)

Table 1: Resistor Metal Composition and TCR data. High Avg. TCR is better for thermistors.

Conclusions:Our MEMS temperature array was able to accurately measure the temperature across the microbial boundaries in the hot springs of YNP� Our results are shedding light on the interactions between the water temperature and the location of the different microbial communities living in the hot spring environment� This is evident by the drastic change of life with a change of temperature� Our success opens the door for new MEMS sensors arrays (such as pH, conductivity, and flow rate) to be used in small scale gradient measurements and demonstrates their versatility and practicality in research�

Acknowledgements:I’d like to thank my PI Dr� Hongyu Yu, my mentor Jonathon Oiler, Trevor Thornton, CSSER staff, my editor Sara Butterfass, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, and the National Science Foundation for making this possible�

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

•


Measuring van der Waals Forces in Graphene

Mariah SzpunarMechanical Engineering, University of Miami

NNIN REU Site: Colorado Nanofabrication Laboratory, University of Colorado, Boulder, CONNIN REU Principal Investigator: Prof. J. Scott Bunch, Mechanical Engineering, University of Colorado BoulderNNIN REU Mentor: Xinghui Liu, Mechanical Engineering, University of Colorado BoulderContact: [email protected], [email protected], [email protected]

Abstract:

The focus of this project was to suspend thin layers of graphene over circular wells to test how the graphene interacted with substrates in connection with a nanomechanical switch. Van der Waals forces adhered the graphene to the substrate. Through pressurization, graphene was disconnected from the center post of the annular ring, and attractive van der Waals forces could then be observed as the graphene was pulled back to the center post. This relates to a switch as graphene can be released and reattached to the annular ring post multiple times. Rings with suspended graphene both attached and not attached to the annular ring post were observed with a tapping atomic force microscope (AFM).

Introduction:

Graphene, a single atomic layer of graphite, is one of science’s newest materials� Graphene layers are bonded together by van der Waals forces, weak dipole-dipole bonds that also bond graphene layers to substrates� The long range van der Waals force can be expressed by the Casimir equation:

(1)where is Planck’s constant, c is the speed of light, A is the area of contact, and d is the distance between the two materials� Since graphene is impermeable, a gas must diffuse through the

Figure 1: Cross-section view of SiO2 wells with exfoliated graphene covering the wells.

substrate and not the graphene� Therefore, creating pressurized graphene balloons with a permeable substrate such as silicon dioxide aids in uncovering properties of graphene, such as adhesion energies�

Experimental Procedure:A silicon wafer with a 90 nm silicon oxide (SiO2) thickness was used throughout this process� Chip geometry consisted of 15 µm outer diameter and 5 µm inner diameter wells, which created hollow wells with a post in the center� The pattern was etched using a reactive ion etcher to obtain a well depth of 500 nm� A cross section of this geometry can be seen in Figure 1�

A second set of chips was made with a slightly different geometry: a 3 µm outer diameter and 1 µm inner diameter with a 100 nm depth� These chips were covered with 3 nm of chromium and then 3 nm of gold via vacuum evaporation�

Graphene was applied using the scotch tape method� In this method, graphite is separated into thinner layers by peeling using scotch tape� Applying pressure to the tape against the wafer transferred graphene flakes that fully covered some wells� Placing the chip in a chamber at a pressure greater atmosphere allowed

Figure 2: AFM of a pressurized annular ring at 500 kilopascals.

• mech

anica

l device

s • m

ech

anica

l device

s • m

ech

anica

l device

s •


high-pressure helium gas to diffuse into the well through the substrate� Once the chamber had reached equilibrium and the internal pressure was greater than atmospheric, the chip could be removed back to atmospheric pressure and the graphene bulged upward from the pressure difference, creating a graphene balloon as is demonstrated by Figure 2� A tapping-mode AFM was used before pressurization and at various pressure steps to check the height of the graphene balloon and to see if the graphene was peeling from the substrate�

Results and Conclusions:We observed delamination of the graphene from the substrate during various stages of pressurization� Figure 3 shows the cross-section profile of a graphene balloon as it became pressurized. The flat line represents the graphene at atmospheric pressure. The line with two modes is the profile after the cavity has come to 500 kPa equilibrium�

As the internal cavity was pressurized, but not enough to delaminate from the post, the graphene bulged over the cavity� The cavity was pressurized further until it reached the critical pressure, at which point the graphene snapped off of the center post and became one large graphene balloon� This is represented by the one tall curve�

We also observed deflation of the fully pressurized balloon. Figure 4 shows the cross-section profile of a graphene balloon on a gold-covered well� The top curve represents the balloon at the highest internal pressure, immediately after it was taken out of the pressure chamber� Each subsequent curve is the profile of the balloon 11 minutes after the previous curve� Notice that despite the noise to the left and right of the post on the final profile, the graphene snapped back to the post due to the van der Waals forces that become more significant at shorter distances. The gold-

graphene adhesion energy was calculated to be ~ 0�01 J/m2, while the SiO2-multilayer graphene adhesion is 0�31 J/m2� We believe this significant difference is due largely to the surface roughness of the gold�

The graphene behaved as we expected� It detached from the center post of the substrate once it reached the critical pressure� It was also drawn to the post once the distance between the two was short enough to allow van der Waals attraction� The switch-like connection was achieved�

Acknowledgements:Funding was provided by the National Science Foundation, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, and the Colorado Nanofabrication Laboratory� Professor Scott Bunch, Xinghui Liu, Narasimha Boddeti and the members of the Bunch lab were instrumental in making this project a success� The administration of the Nanofabrication Laboratory was very helpful in training and fabrication support�

References:[1] J� S� Bunch et al� “Electromechnical Resonators from Graphene

Sheets�” Science, 315, 490-493� 2007�[2] J� S� Bunch et al� “Impermeable Atomic Membranes from Graphene

Sheets�” Nano Letters, 8�8, 2458-2462� 2008�[3] H�B�G� Casimir, “On the Attraction Between Two Perfectly Con-

ducting Plates,” Proc�K� Ned� Akad� Wet�, vol� 60, pp� 793-795, 1948�[4] A� K� Geim and K� S� Novoselov� “The Rise of Graphene�” Nature, 6,

183-191, 2007�[5] Koenig, S� P�, Boddeti, N� G�, Dunn, M� L�, and Bunch, J� S�

“Ultrastrong adhesion of graphene membranes�” Nat Nano, advance online publication� Retrieved from http://dx�doi�org/10�1038/nnano�2011�123, (2011)

Figure 4: Center cross-section of the gold well graphene balloon. The top curve is the balloon profile almost immediately after leaving the pressure chamber. Each subsequent line is 11 minutes after the previous.

Figure 3: Center cross-section profiles of the graphene at each stage of pressurization.

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

• m

ech

anic

al

devic

es

•


Calibration of Optical Particle Sizer by Wafer Surface Scanner

Laura WindmullerBiomedical Engineering, Boston University

NNIN REU Site: Nanofabrication Center, University of Minnesota-Twin Cities, Minneapolis, MNNNIN REU Principal Investigator: Prof. David Pui, Mechanical Engineering, University of Minnesota-Twin CitiesNNIN REU Mentor: Dr. Lin Li, Mechanical Engineering, University of Minnesota-Twin CitiesContact: [email protected], [email protected], [email protected]

Abstract:

An optical particle sizer (OPS) is typically calibrated using polysterene latex (PSL) spheres and an electrometer, or similar instruments. However, these current calibration methods require high particle concentrations. Our project designed a low-concentration calibration method with micrometer-sized PSL particles for the OPS. By depositing a controlled number of particles on the wafer surface, we could calibrate the OPS based on the wafer surface scanner’s (WSS) analysis of the deposited particles. In this experiment, we used a settling chamber for deposition. Residue particles were a primary problem. A long differential mobility analyzer (DMA) and a virtual impactor were included to decrease residue particles and increase the 3 µm particle concentration into the targeted range of 10-100 particles/L. We also tested for background residue particles and the application of an electric field in the deposition chamber. Our best trial obtained 70%-80% 3 µm particle deposition. Future work will explore residue particle sources, WSS accuracy, DMA effectiveness, and flow rate control.

Experimental Procedure:

Figure 1: Experimental setup used for OPS and WSS comparison.

In Figure 1, our general deposition and calibration process is outlined� Stemming from a compressed air source, we used a nebulizer without an impactor plate that was effective with 3 µm PSL spheres as our aerosol source� Two driers evaporated droplets and combated residue particles created by empty droplet deposition�

The virtual impactor with a 0�185 cm nozzle increased the 3 µm concentration. The total-minor flow ratio controlled the impactor’s cutoff diameter� The cutoff diameter was 2.9 µm for a 10:1 flow ratio [1]. We sought a slightly smaller ratio, 3 liter per minute (lpm) total flow and 0.35 lpm minor flow, so that the majority of the 3 µm particles would be collected from the minor flow exit.

The long DMA helped create a monodispere aerosol of single-polarity doubly-charged 3 µm particles at a 0�3 lpm aerosol flow, 1.4 lpm sheath flow, and 9 kV potential [2]. We decided to use the higher sheath flow and pass through doubly-charged particles in order to decrease the likelihood of residue particles diffusing through the DMA�

Using a flow splitter and laminar flow meters (LFM), we ensured equal particle concentration and controlled flow rates entered the OPS and deposition chamber for accurate particle count comparison� Inside the self-enclosed settling deposition chamber, the wafer had a 5 mm separation from

• mech

anica

l device

s • m

ech

anica

l device

s • m

ech

anica

l device

s •


the aerosol entrance� For 3 µm particles, the deposition radius calculated to fit within a wafer diameter of 4 inches (10.16 cm) was solved as 1.69 inches (4.3 cm) for a flow rate of 0�1 lpm [2]� Because the settling velocity was greatest for gravitation when compared to electrostatic precipitation and thermophoresis, it was deemed the primary method for particle deposition. However, an electric field in the settling chamber was employed to increase particle settling velocity and enhance deposition results� From the chamber, the wafer was physically transferred to the WSS to have particles sized and counted�

70% to 80% of the 3 µm particles identified by the OPS were captured� In Figure 3, the deposition pattern appears even and few particles were identified near the edge. This result corroborated our calculations for the settling radius and validating our settling technique� Wafer 6304 was set at a potential of -6 kV/cm to increase the deposition velocity of 3 µm particles increase deposition� The wafer particle deposition significantly exceeded OPS measurements. Due to this result and the burn marks found inside the chamber, we concluded that electrical arcing within the deposition chamber may be creating additional small particles to settle on the wafer�

Future Work:We seek a 3% to 10% uncertainty in our combined deposition-WSS method. A refined method of measuring the 0.1 lpm flow to the OPS and deposition chamber is needed. Also, the effectiveness of DMAs at low sheath flow rates should be investigated� To decrease residue particle size and concentration, an impactor plate may be added to the aerosol nebulizer to minimize the production of large droplets. While the electric field increased total particle deposition, the effects of arcing should be explored� Finally, the WSS’s size channel allocation accuracy should be evaluated further with a calibration wafer or by other trials with various PSL sphere sizes�

Acknowledgements:I would like to thank Lin Li, David Pui, George Mulholland, and Miles Owen for their guidance, the University of Minnesota for hosting me, the National Nanotechnology Infrastructure Network Research Experience for Under-graduate Program and the National Science Foundation for this opportunity, the NSF and Army for funding this project, and the entire Pui Group for welcoming me into their ranks�

References:[1] Bernard Olson (private communication)�[2] Hinds, W�C�, Aerosol Technology, John Wiley & Sons, Inc�, New

York 1999�

Figure 3: Wafer 6303 (best trial) deposition and size distribution.

Figure 2: WSS and OPS test conditions and particle count comparison.

Results:Figure 1 illustrates our deposition setup� Figure 2 describes the test conditions for each wafer deposition trial and the comparison of particle counts between the OPS and WSS� It should be noted that the OPS had a sharp peak at its 3 µm channel while the WSS had a flat and broad peak from its 0.6 and 0.7 µm channel to the 4 µm channel. The significant number of particles smaller than 0.6 µm identified by the WSS were probably from corona discharge inside the DMA� However, there was also concern that the WSS may not be assigning particles to the correct sizing bin� Preliminary checks of WSS sizing accuracy with a scanning electron microscope confirmed our suspicions and indicated that particles from the 0�6 µm channel through the 4 µm channel were actually 3 µm PSL spheres� For this reason, we included counts in this range from the WSS for our particle deposition totals�

From wafer 6301, we estimated cleanroom contamination as 20 particles per channel from 0�4 µm to 7 µm channels� Figure 3 shows wafer 6303, our best trial� Approximately

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Sol-Gel Route for Ultra-High-Quality Optical Resonators

Michael AkenheadBiomedical Engineering, Vanderbilt University

NNIN REU Site: Nano Research Facility, Washington University in St. Louis, St. Louis, MONNIN REU Principal Investigator: Dr. Lan Yang, Electrical and Systems Engineering, Washington University in St. LouisNNIN REU Mentor: Dr. Sahin Kaya Ozdemir, Electrical and Systems Engineering, Washington University in St. LouisContact: [email protected], [email protected], [email protected]

Introduction:Microlasers are of great interest in various fields of science and technology� They are used in optical communications, and are being used for single nanoparticle detection� In the future, microlasers could even reduce the size of silicon computer chips� To generate a microlaser, three important components are required: a high-quality microresonator, a gain medium, and a pump source� Lasing in microresonators can be achieved by optically pumping the gain medium incorporated in the microresonator structure� While many different options for incorporating a gain medium exist, this study examines the sol-gel method�

The sol-gel method is an easy, cost effective, and flexible way of incorporating a gain medium into the microresonator structure� Our group uses silica microtoroids as resonators� To include gain medium, silica is prepared using the sol-gel process, during which the gain medium is introduced� Three steps are needed to produce sol-gel silica films. The first step involves the hydrolysis of silicon alkoxide by water molecules in order to produce a colloidal suspension (the sol)� The second step involves condensation reactions forming a gel network� The third step is the annealing step� By heating the sol-gel film at high temperatures, the porous silica coating is annealed into dense glass� Upon cooling back to room temperature, the desired silica film is obtained.

This study is concerned with optimizing the sol-gel process to produce high-quality microresonators and use them for microlasing� The quality factor of the microtoroid depends on the sol-gel film quality. By producing a film that is optically smooth and has a homogeneous structure, a high quality factor microtoroid may be fabricated� This will result in more light being confined with the microtoroid, reducing the threshold for lasing. The quality of the sol-gel film depends on many factors, such as annealing temperature, solution aging time, acidity, ingredient ratios, and the speed and time of spin coating� Annealing temperature and solution aging time are examined in this study�

Methods:The initial goal during the sol-gel optimization process was to determine the proper annealing temperature to produce an optically smooth sol-gel film. To test the best annealing temperature, sol-gel layers were annealed at 800°C, 900°C, 1000°C, and 1100°C for three hours� After annealing, layers were allowed to cool to room temperature� The resulting sol-gel layers were then analyzed under a scanning electron microscope (SEM)�

After analyzing the films using the SEM, it became clear that small pores were produced in each sol-gel film. These pores varied in size, shape, and concentration, depending on the annealing temperature� In order to characterize the formation of pores at different annealing temperatures, new wafers were coated with sol-gel layers and annealed at the previously used temperatures� After annealing, microdisks were patterned on the wafers using photolithography and etching techniques in order to observe the pore distribution�

Additionally, tests were performed to determine the extent of sol-gel solution aging time� Two types of aged solution were tested – solutions that had less than 24 hours of aging (“fresh” solutions), and solutions that had greater than 24 hours of aging� Each sol-gel solution was coated on a silicon wafer and annealed at the same temperature for three hours�

Results:The initial films analyzed using the SEM revealed some trends regarding sol-gel porosity� The largest pore diameters were found at lower temperatures of 800°C and 900°C, as seen in Figure 1� These pores ranged in size from 150 nm to 200 nm� The smallest pore diameters were found at higher temperatures of 1000°C and 1100°C, as seen in Figure 2� Pore diameters ranged from 100 nm to 150 nm for these temperatures� In addition, pore density was much greater at lower annealing temperatures than at higher annealing temperatures�

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


Testing the aging time of sol-gel solutions on porosity revealed that aging has a significant impact on formation of pores� Solutions that were aged for less than 24 hours produced little to no pores, while solutions aged longer resulted in porosity� Annealing temperature has very little effect on solutions aged for less than 24 hours� Once the solutions have aged past 24 hours, however, annealing temperature has a much more significant effect.

When microdisks were etched into the silicon wafers using standard wet and dry etching process and analyzed under the SEM, the same trends were observed as with the initial sol-gel layers� Lower temperatures produced greater porosity than higher temperatures� Porosity did not affect the etching of the microdisk� The microdisks underwent thermal reflow to finish microresonator development. After reflow, it became apparent that porosity further diminished, as seen in Figure 3�

Conclusion:Porosity has a negative impact on sol-gel film quality; as a result, minimal porosity is desired� For best results, a “fresh” solution that has not aged for more than 24 hours should be used to eliminate the majority of pores� The resulting sol-gel layer should be annealed at higher temperatures, to further minimize porosity� Since creating “fresh” solutions for every sol-gel layer is expensive, new solutions should be made after a week� One sol-gel coating is not enough to prepare a wafer for etching; the wafer should be coated with multiple layers for required thickness, as seen in Figure 4� By minimizing porosity and coating with multiple layers, the highest quality microresonators may be created� More

Figure 1: SEM image revealing porosity at a low annealing temperature (800°C).

Figure 2: SEM image revealing porosity at a high annealing temperature (1000°C).

Figure 3: Surface of a microdisk after thermal reflow.

Figure 4: SEM image of a microtoroid coated with three sol-gel layers.

work is needed to determine the effects of other factors on sol-gel quality, such as acidity�

Acknowledgements:I would like to thank Dr� Lan Yang, Dr� Sahin Ozdemir, and the graduate students of the Micro/Nano Photonics lab for their help and guidance� Funding for this project was provided by National Science Foundation (NSF) and National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program� Additional thanks to Melanie-Claire Mallison, Dolores Stewart, Kate Nelson, and Nathan Reed�

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


With the exponential growth of internet users and data-hungry devices, the need for faster and more accessible communication has been on the rise. Unlike copper wires, the integration of light amplification by stimulated emission of radiation (LASER) components in optical fibers enables us to transmit data through photons with the speed of light without heating up. In this regard, vertical cavity surface emitting lasers (VCSELs) could be good candidates due to their easy manufacturing in optical applications. However, resistive losses slow down such lasers. By studying the structure of the lasers and comprehending the losses, lower resistive VCSELs can reach higher modulation speeds. In this process, different components of VCSELs, distributed Bragg reflectors, and separate confinement heterostructures were tested to characterize their properties and propose desirable designs. The new structures show great potential for lowering the resistance.

Introduction:

Design and Characterization of Multiple Quantum-Well Lasers

Seyedshahin AshrafzadehElectrical Engineering, West Valley College

NNIN REU Site: Nanotech, University of California, Santa Barbara, CANNIN REU Principal Investigator: Prof. Larry Coldren, Electrical and Computer Engineering Department

and Material Department, University of California, Santa BarbaraNNIN REU Mentor: Yan Zheng, Electrical and Computer Engineering, University of California, Santa Barbara

(2004 NNIN REU Intern at Georgia Institute of Technology)Contact: [email protected], [email protected], [email protected]

Abstract:

Figure 1: Different components of a VCSEL relative to each other, but not to scale.Similar to other lasers, VCSELs consist of an active

medium, which creates photons, and two mirrors, known as distributed Bragg reflectors (DBRs). Electrical current is supplied to the laser through metal contacts� The injected carriers, electrons and holes, pass through the DBRs into the active region, but these mirror structures are resistive�

In the active medium, separate confinement heterostructures (SCH) funnel the carriers toward multiple quantum-wells (MQWs). In MQWs, electrons and holes become confined in energy wells, get stimulated by a photon, recombine with each other, and then create another photon coherent to the first photon. Afterward, these two photons get reflected between the two DBRs and stimulate other electron-hole pairs, amplifying the total photon density� At some point, a small portion of the light passes through the bottom DBR and leads to lasing phenomenon�

Distributed Bragg Reflectors (DBRs). DBRs are periodic layers of aluminum gallium arsenide (AlGaAs) with a transition layer from the same material between each two consecutive sheets to lower their resistance� Each layer has

different aluminum (Al) concentration, which corresponds to different index of refraction; because of this difference in refractive indexes, photons become constrained between the two DBRs to increase stimulated recombination�

As mentioned, the resistivity of DBRs can be reduced by having an effective transition layer with the right aluminum concentration� This was achieved through a comparison between two different DBR structures by applying transmission line measurements (TLM)� By this method, the resistivity of DBRs can be evaluated and compared with each other�

Two DBRs were grown on gallium arsenide (GaAs) substrates by molecular beam epitaxy with the following design rules. The first sample consisted of four periods of alternating layers of 0�050 µm GaAs with 8 × 1018 n-type doping and 0�0448 µm AlGaAs with 90% Al concentration, and 8 × 1018 n-type doping separated by 0�016 µm AlGaAs

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


with 30% Al concentration, and 8 × 1018 n-type doping in the transition layer� For the second DBR, only the transition layer was changed to a linear grading of Al concentration from 0% to 90%� After the growth, the DBRs were fab-ricated, and metal contacts were deposited in circular patterns with different gap spacings�

By passing a current through the sample, shown in Figure 2(a), the total resistance was measured� With the use of known values for the specific contact resistivity between the metal and the semiconductor, 615.1396 Ω-µm2, the resistance of each DBR was extracted� According to the data, the second DBR with linear Al had a lower average resistance than the first DBR—2.93Ω compared to 5.62Ω. As a result, the linear grading of Al in the transition layer of our structure reduced the resistance of the n-type DBR�

Separate Confinement Heterostructure (SCH). The need for a structure to direct the carriers toward small layers of MQWs is significant in VCSELs. Aside from guiding the electrons and holes to MQWs, a SCH is required to have low resistivity and capacitance� Hence, the new design of SCH had to consider both factors, resistance and capacitance� Simulating different structures in SimWindows [1] was the approach taken to reach this goal�

A SCH was made from AlGaAs, and Figure 3 shows the Al concentration and the structure of three SCHs� As is evident in Figure 3, the new SCH, ParbL2, consists of a parabolic Al grading from 0�85 to 0�5, then a linear Al concentration to 0�3, and after MQWs, another linear Al from 0�3 to 1�

Using a simulation tool called SimWindows, all SCHs were simulated by applying voltages from 0�0 V to 3 V, and their currents were extracted� From the slope of the current vs� voltage graphs, the resistance of each SCH was determined� Based on the results, ParbL2 had the lowest resistance— 6.52 Ω-cm2—than either ParbL3, with 6.95 Ω-cm2, or ParbL2Parb, with 6.73 Ω-cm2� Furthermore, Figure 4 demonstrates that ParbL2 fairly had the lowest carrier

concentration compared to other designs� This low accumulation of carriers in SCH corresponds to lower capacitance� Therefore, the new model lowered the resistance and capacitance over the older versions�

Conclusion:This paper presented a comparison between two DBRs and a simulation analysis for different SCHs� According to the results, the proposed structures lowered the resistance in n-type DBR and reduced the resistivity and capacitance in the new SCH� Future work remains to be done to grow the simulated SCHs and determine their resistance and capacitance�

Acknowledgements:I would like to show my deepest gratitude to my mentor, Yan Zheng, my principal investigator, Prof� Larry Coldren, and my coordinator, Angela Berenstein, whose guidance directed me to better understanding of the research� It is also a pleasure to thank the National Nanotechnology Infrastructure Network Research Experience for Under-graduates Program, the National Science Foundation, and University of California, Santa Barbara for their great support and funding�

References:[1] Winston, D; “Optoelectronic Device Simulation of Bragg Reflectors

and Their Influence on Surface-Emitting Laser Characteristics”; IEEE Journal of Quantum Electronics, Vol� 34, No� 4, (1998)�

Figure 4: Carrier concentration in log scale.

Figure 2: Transmission line measurement.

Figure 3: Al concentration in 3 different SCHs.

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Distributed Bragg Reflectors in Ultra Low Loss Silicon Nitride Waveguides

Issa BeekunElectrical Engineering and Computer Science, University of Nevada, Reno

NNIN REU Site: Nanotech, University of California, Santa Barbara, CANNIN REU Principal Investigator: John Bowers, Electrical and Computer Engineering, University of California, Santa BarbaraNNIN REU Mentor: Jock Bovington, Electrical and Computer Engineering, University of California, Santa Barbara

(2005 NNIN REU Intern at the University of Washington)Contact: [email protected], [email protected], [email protected]

Abstract:

The focus of this project was to fabricate distributed Bragg reflector (DBR) structures into a silicon dioxide (SiO2)-clad silicon nitride (Si3N4) waveguide. DBRs are used extensively in the standard operation of semi-conductor lasers. Such gratings have the ability to reflect select wavelengths of light as a means of producing laser feedback. An efficient grating requires precise fabrication techniques as the grating features need to be very small with periods ~ 140-520 nm depending on the effective refractive index of the mode. After fabrication, it is then useful to characterize the gratings to determine their respective scattering loss, strengths, and Bragg wavelengths. Characterization was accomplished using a superluminescent light emitting diode (SLED) with an optical spectrum analyzer (OSA).

Introduction:

Though not widely known, semiconductor lasers have been one of the hallmark inventions driving the information age. Their integration as sources enabling the fiber optic networks which span the globe should not be understated� DBRs are a fundamental element for these lasers providing the wavelength selective mirror to enable an array of tightly spaced single mode lasers, often able to tune wavelengths to adjust for changes in the network� This element can also be found in other non-telecommunication applications with equally stringent requirements for wavelength selectivity�

The first objective of the project was to select reasonable grating parameters by simulating various etch depths and modal properties� The next part was to successfully fabricate the DBRs using electron beam lithography (EBL) and a dry etch process, yielding preliminary test chips (Figure 1) and a grating chip with actual waveguides�

Figure 1: A chip used to test the parameters of the fabrication process.

The final part of this project was to test the fabricated DBRs by coupling light into the waveguide containing a single DBR and detecting the spectral response of the light emitted at the output� By this method, fundamental grating parameters were inferred to aid in future laser design�

Simulation:In order to simulate the relevant grating parameters, a literature search was conducted to discover more about gratings and the equations which govern their operation [1]� A simple equation was examined which allowed for initial designs around the specifications of a 1.045 µm test laser. Further calculations were made with the aid of a MATLAB® finite difference method (FDM) mode solver program and by implementing a script representing consecutive grating periods as the product of multiple sets of 2 × 2 scattering matrixes� From these results and by selecting ideal values for the length and strength of the gratings, fabrication of the structures could begin�

Fabrication:Electron beam lithography (EBL) is a technique with very high resolution� It uses electrons to expose the resist, as opposed to standard photolithography processes, which use light� EBL also uses a raster scan of the beam to expose select areas rather than the standard masked single exposure of a large area� EBL can require a large allotment of time as compared to standard photolithography, and therefore it is not scalable for production of large numbers of devices;

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


however, it enables the flexibility needed to consistently create gratings with tight space wavelengths�

After patterning gratings into the resist, atomic force microscopy (AFM) images were taken to confirm the patterns as they were not visible in an optical microscope (Figure 2)� Finally, a series of dry etches in inductively coupled plasma etcher using an oxide and chrome double hard-mask translated the pattern to the oxide cladding of our Si3N4 waveguide. The resulting profile was imaged using a scanning electron microscope (Figure 2)�

Testing:Testing the DBR gratings was prefaced by polishing the rough facets of the chip until they were mirror-like� Spectral transmission measurements were done by coupling on and off the chip, with cleave fibers using index-matching gel, a broad band SLED source with OSA� An infrared camera lens column and objective lens setup as seen in Figure 3 served to confirm that light was coupled to the waveguide and not any slab modes and to confirm a TE polarization.

Figure 4: Transmitted power through gratings of length 55, 111, 163, 230, 380, 2000 µm from top to bottom. Each spectrum has been normalized to a straight waveguide and shifted by 2.5 dB from the adjacent spectra for ease of viewing.

Figure 2: A 3D AFM of the fabricated gratings and corresponding SEM of the grating cross section.

Figure 3: Test setup for measuring spectral response from the gratings.

After confirmation, the objective lens was removed and replaced with a fiber connected to the OSA allowing the spectral response to be measured� Such a response was obtained by sweeping the output wavelength of the test laser from 960-1160 nm� Outputs from grating structures were normalized to a straight waveguide without gratings and the spectra� Fitting to the simulation model failed likely due to an over etch through the cladding into the core� This is currently under investigation�

Results and Conclusion:Complications to the project occurred during the etching portion of fabrication and resulted in structures with varied functionality� While a new chip was being fabricated, testing proceeded and spectral responses from the gratings were obtained for comparison with simulations as seen in Figure 4� At this time, new measurements are being recorded to compare against the simulation generated expectations�

This project demonstrates a comprehensive approach to understanding the principles, fabrication, and testing of distributed Bragg reflectors.

Acknowledgements:I would like to acknowledge the Bowers Optoelectronics Research Group, particularly Daryl Spencer, and the staff of the Electrical and Computer Engineering Department at the University of California at Santa Barbara, for their help and guidance during this summer internship� I would also like to thank Angela Berenstein for her help in organizing the program, and the NNIN REU Program and the NSF for their roles in making my project possible�

References:[1] Coldren, L�, and S� Corzine� Diode lasers and photonic integrated

circuits� Wiley-Interscience, 1995� Print�

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Integrated Silicon Nitride Waveguides: Optimization of Fabrication

Alex BryantMaterials Science and Engineering, University of California, Berkeley

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Professor Michal Lipson, Electrical and Computer Engineering, Cornell UniversityNNIN REU Mentors: Daniel Lee and Jaime Cardenas, Department of Electrical and Computer Engineering, Cornell UniversityContact: [email protected], [email protected], [email protected], [email protected]

Figure 1: Scanning electron microscope (SEM) image showing the trapezoidal shape of previous waveguides that have been produced from Si3N4 [1].

Abstract and Introduction:Integrated waveguides enable the use of photons for transfer of data in a manner similar to that of electrons in integrated circuits� Silicon nitride (Si3N4) is an ideal material for these devices as it is transparent in the wavelengths between 300 nm and 5 µm, has a high refractive index, and does not have non-linear power losses� However, as seen in Figure 1, previous etch processes with silicon oxide (SiO2) masks have produced sub-optimal waveguides [1]� An etch process with high selectivity and 90° sidewalls will enable enhanced optical performance and lower propagation loss� This will allow fabrication of high performance integrated photonic devices�

This project utilized the existing SiO2 mask Si3N4 plasma etch process with trifluoromethane (CHF3) and O2� It was seen in the Blain, et al�, paper [2] that the addition of nitrogen increased the Si3N4 selectivity, so N2 was added into the etch chemistry� Varying the parameters of CHF3, O2, and N2 flow rates, bias voltage, and pressure in multiple design of experiments (DOEs) and then measuring the resultant

etch rates, selectivities, and sidewall angles allowed for the creation of an etching process optimized for the Oxford 100� The results enable the fabrication of 90° sidewalls with a selectivity of roughly 1�8:1�

Experimental Procedure:For the plasma etching, we chose to use the Oxford 100 with the configuration shown in Figure 2 because, as an inductively coupled plasma (ICP) system, the plasma density and ion energies are decoupled� This allowed for high aspect ratios and small feature etching [3]�

Then, as it was seen in the literature that a high selectivity of 100:1 was achieved for the chemistry NF3/O2/NH3 in an isotropic downstream process with high pressure, RF power, and flow rates as well as very low temperature [4], nitrogen was added into the usual etch chemistries CHF3/O2 and CF4/O2 of our Oxford 100� This addition was meant to

Figure 2: Typical ICP configuration showing antenna coil for plasma generation and electrode for ion acceleration [3].

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


result in the same constituent elements being present as was in the 100:1 selectivity etch chemistry, with the addition of carbon. The Blain, et al., paper [2] further verified this as a reasonable method of increasing the selectivity, showing that an addition of nitrogen increases the etch rate of Si3N4 relative to SiO2 due to the formation of NO gas in the plasma, which subsequently reacts with the surface in one or more of the mechanisms shown in Figure 3�

As such, multiple 2-level Fractional Factorial DOEs were run with both of these chemistries, where the factors concerned were: CF4 or CHF3 flow rate, O2 flow rate, N2 flow rate, pressure, RIE power, and ICP power. From these DOEs, general trends were determined for each of these factors as well as their interactions� This enabled further DOEs which focused further on the parameter spaces which resulted in the most favorable results�

During each DOE, the fabrication process involved the following steps: First, with low pressure chemical vapor deposition, about 50 nm of high temperature silicon oxide (HTO), and then 350 nm of Si3N4, and then another 250 nm of HTO were deposited� Then, about 750 nm of SPR 955-CM photoresist was spun on top� This resist was then patterned into the shape of the waveguides with the Autostepper� The top SiO2 layer was then etched in the Oxford 100, and the resist was stripped in an oxygen plasma in a coupled reactive ion etcher� Finally, the Si3N4 was etched with the SiO2 layer as a mask, leaving the SiO2 mask as cladding�

Results and Conclusion:The etch rate and selectivity were determined using a FilMetrics (an interferometer) and the changes in thickness of the SiO2 mask and Si3N4 layers after the Si3N4 etch� To determine the sidewall angles, the wafers were cleaved perpendicular to the waveguides, and the resulting pieces were mounted in the SEM on a 90° mount after Au/Pd were sputtered on top to reduce charging of the dielectric material�

Six DOEs were performed with the intention of finding the highest selectivities possible� Then, SEM imaging was performed to determine if the highest selective parameters also resulted in 90° sidewalls� The CF4 chemistry produced the highest selectivities, but their sidewalls were not

perfectly 90°� However, the CHF3 chemistry resulted in a slightly lower selectivity of about 1�8 and had perfect 90° sidewalls� Further, the waveguides which were coupled (less than a micron apart) were also resolved to a high degree of precision� The CHF3 etch devised here will now be used by the Lipson group for the fabrication of their Si3N4 waveguides�

Acknowledgements:I would like to thank my principal investigator, Professor Michal Lipson� I also wish to acknowledge the guidance and assistance of my mentors Daniel Lee and Jaime Cardenas� Further, I appreciate the help of Meredith Metzler, Vince Genova, and the CNF staff� In addition, this project could not have happened without funding from the National Science Foundation and the National Nanotechnology Infra-structure Network Research Experience for Undergraduates (NNIN REU) Program, the Cornell NanoScale Science and Technology Facility, and Rob Ilic and Melanie-Claire Mallison, the CNF REU Program Coordinators�

References:[1] Levy, J� et al�; “CMOS-compatible multiple-wavelength oscillator

for on-chip interconnects”; Nature Photonics, Vol 4, January 2010�[2] Blain, M� G� et al�; “Role of nitrogen in the downstream etching

of silicon nitride”; J� Vac� Sci� Technol� A, Vol� 14, No� 4, Jul/Aug 1996�

[3] Oxford Instruments; “ICP Chamber Diagram”; <http://www�oxford-instruments�com/products/etching-deposition-growth/processes-techniques/plasma-etch/icp/Pages/icp�aspx>�

[4] Wang, Y� et al�; “Ultrahigh-selectivity silicon nitride etch process using an inductively coupled plasma source”; J� Vac� Sci� Technol� A, Vol� 16, No� 3, May/Jun 1998�

Figure 4: SEM image showing 90° straight sidewalls from developed etch process.

Figure 3: Mechanisms by which nitrogen in the surface is removed due to the addition of nitrogen into the plasma chemistry [2].

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Fabrication and Testing of Voltage-Tunable Plasmonic Metamaterials in Mid-Infrared

Ting Chia ChangElectrical Engineering and Computer Science, University of California at Berkeley

NNIN REU Site: Microelectronics Research Center, The University of Texas, Austin, TXNNIN REU Principal Investigator: Prof. Mikhail Belkin, Electrical and Computer Engineering, Microelectronics Research CenterNNIN REU Mentor: Dr. David Austin, Department of Electrical and Computer Engineering, Microelectronics Research CenterContact: [email protected], [email protected], [email protected]

Abstract:

The device introduced in this project has the potential to further the scope of voltage-tunable frequency-selective surfaces in the mid-infrared spectral range. Our device consists of an array of sub-wavelength plasmonic elements, resonant in the mid-infrared, fabricated on top of GaAs/Al0.8Ga0.2As coupled-quantum-well heterostructure. The plasmonic structures were fabricated using electron beam lithography (EBL), and the devices were characterized using Fourier-transform-infrared-spectrometer-based (FTIR) reflection measurements. The fabricated plasmonic structures were resonant at 4.5 µm and intersubband absorption was measured at 6.8 µm. This resonance mismatch was too high to observe tuning.

Introduction:Metamaterials are artificial materials constructed on the sub-wavelength scale to provide electromagnetic properties such as negative refractive index that usually cannot be obtained by naturally occurring materials. Owing to the narrow fixed frequency response of plasmonic metamaterials, they may be used as frequency selective surfaces [1, 2]� Our device introduces the possibility of electrically controlling the absorption wavelength of these surfaces with a potential 10 percent of wavelength tuning range in the mid-infrared spectral range� To accomplish this, we mated the plasmonic metamaterials with quantum-well structures in which the refractive index depended strongly on applied bias voltage�

The plasmonic metamaterial consisted of an array of sub-wavelength elements patterned in a 40 nm thick gold (Au) layer� This was fabricated on the top surface of a gallium

Figure 1: Device structure designed in CST Studio composed of an array of plasmonic sub-wavelength complementary (a) cross and (b) split ring resonators.

arsenide / aluminum gallium arsenide (GaAs/Al0�8Ga0�2As) coupled-quantum-well heterostructure� The simulated device structure is shown in Figure 1, and was designed in CST Microwave Studio®� The resonance wavelength of this material was determined by the geometry of the elements and the refractive index of the surrounding environment, which in this case was the quantum-well heterostructure�

The quantum-well layer absorbed in the mid-infrared through intersubband transitions in the conduction band� These had an atomic-like absorption profile with a sharp, narrow linewidth� This allowed for a large change in the refractive index through the Kramers-Kronig relation, for transverse

Figure 2: Intersubband transition simulation and dielectric constant of quantum wells. The device is biased at; (a) 50 kV/cm, (b) -50 kV/cm, (c) real part, and (d) imaginary part of dielectric constant at different bias voltage.

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


magnetic (TM) polarized electromagnetic radiation� Band structure engineering using a one dimensional Schrodinger-Poisson simulation was performed to determine the transition energies and hence the operating wavelength� This was subject to the Stark shift, so could be tuned by application of a bias voltage� This wavelength is shown in Figure 2�

Since the refractive index of the surrounding environment of the plasmonic metamaterial was altered, there was a resultant shift in the resonance frequency� The resonance frequencies of plasmonic metamaterial and quantum-well structure should be coincident in order to achieve maximal tuning of the plasmonic resonance�

a solution of 1:4:45 ratio of phosphoric acid:hydrogen peroxide:water (H3PO4:H2O2:H2O), the surrounding area was etched down by ~ 500 nm� The samples were then thinned down to 200 µm and mounted with indium solder to a copper carrier block� Multiple wire bonds to the array were made, to accommodate a potentially large current�

The finished device was characterized by FTIR, and spectral measurements were taken� A broad band mid-infrared source (globar) was focused onto the plasmonic array and the reflected signal measured with a liquid helium cooled mercury cadmium telluride (MCT) detector�

Results and Conclusions:The resonance frequencies of the cross and split ring resonators were 4�5 µm and 3�1 µm respectively, while the resonant frequency of quantum well structure was 6�8 µm, as shown in Figure 4� Voltage biasing of the underlying semiconductor layer was attempted, but no tuning was observed� This was due to the large resonance mismatch between the fabricated plasmonic structures and the quantum wells�

Future Work:The frequency-selective-surface device presented relies on matching the resonance wavelength of the plasmonic metamaterial with the absorption wavelength of the semiconductor layer� We expect future structures in which the plasmon and quantum well resonances are matched should allow for widely tunable plasmonic devices�

Acknowledgements:Appreciations and thanks are directed toward the NNIN REU and NSF� I would like to thank Prof� Mikhail Belkin, Dr� David Austin and Karun Vijaraghavan as well as the rest of the research group for helping me carry out the project�

References:[1] X� Liu, T� Starr, A� F� Starr, and W� J� Padilla, Phys� Rev� Lett� 104,

207403 (2010)�[2] X� Liu, T Taylor, T� Starr, A� F� Starr, N� M� Jokerst, and W� J�

Padilla, Phys� Rev� Lett� 107, 045901 (2011)�

Figure 4: Experimental data of absorption intensity versus wavelength; (a) quantum wells, (b) cross resonator, and (c) split ring resonator.

Figure 3: SEM images of; (a) cross, (b) split ring pattern covered by photoresist, and (c) cross, (d) split ring pattern after gold evaporation.

Experiment Procedure:We fabricated two devices with different plasmonic patterns: complementary crosses and split rings� The quantum well structure was grown by molecular beam epitaxy (MBE) on the n-doped gallium arsenide (GaAs) substrate� EBL was used to fabricate the plasmonic structures� These had minimum feature sizes on the order of a few hundred nanometers� Several tests were performed to obtain a close approximation to the designed structure, including adjusting the dose factor of the electron beam and changing the dimension of the pattern file. 40 nm of Au was evaporated onto the samples’ surface, and a lift-off performed� Figure 3 shows scanning electron microscopy (SEM) images of the array of cross and split ring patterns both in photoresist and after metallization� Mesa structures were fabricated to prevent current spreading and instead, confine the current through the plasmonic metamaterial� Using

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Characterization of Optical Devices using a Pigtailed Fiber

Mark DongApplied Physics, Cornell University

NNIN REU Site: Center for Nanoscale Systems, Harvard University, Cambridge, MANNIN REU Principal Investigator: Prof. Marko Loncar, School of Engineering and Applied Sciences, Harvard UniversityNNIN REU Mentor: Parag Deotare, School of Engineering and Applied Sciences, Harvard UniversityContact: [email protected], [email protected], [email protected]

Abstract:

The hope of achieving integrated optical devices on-chip deals with how to couple light from off-chip sources into the optical devices. The coupling efficiency between off-chip fibers and on-chip waveguides was investigated in this project. We designed a small acrylic plastic base-holder to permanently hold the on-chip waveguides and two optical fibers with room curing epoxy, enabling a strong fiber-waveguide coupling. To minimize loss, the facets of the SU-8 waveguides were polished using aluminum oxide polishing paper. The waveguide and fibers were aligned and the coupling efficiency was measured using a cutback method. We found, for a wafer without the base holder, a coupling loss of 1.26 dB for a polished facet, 3.0 dB for an unpolished facet. Using these losses as a reference, we successfully pigtailed the wafer and fiber using the base holder with a loss of more than 20 dB.

Introduction:

Much of optical devices research focuses on the area of using on-chip waveguides to transport signals and information� The success of such devices relies on coupling these waveguides with minimal losses to off-chip fibers [1]. While low coupling losses using various tapering methods between the waveguide and fiber have been achieved [2, 3], we desire a device that can permanently couple the fibers onto the waveguide without inducing significant losses.

Losses due to coupling depend on the amount of overlap between the two propagating modes in the fiber and on-chip waveguide� By using a Gaussian description for the propagating modes, Joyce [4] found analytical approximations for misalignment� The misalignment tolerances (losses up to 1 dB) were calculated to be on the order of less than a few microns [5]� This important result drove much of the design and preparation of our fiber pigtailing device�

Methodology:In order to measure the effectiveness of pigtailing, we fabricated on-chip waveguides� We used plasma enhanced chemical vapor deposition (PECVD) for silicon oxide deposition, and photolithography to define the shapes. The waveguides contained four layers: 1) a silicon substrate, 2) 2 µm silicon oxide layer, 3) 2�5 µm height SU-8 photoresist waveguide, and 4) a 2 µm top layer of protective silicon oxide� The chip was cleaved to reveal the edges of the

Figure 1: Polished waveguide facet.

Figure 2: Fiber pigtailing device.

waveguide for coupling and then the facets were polished down to produce smoothness on the micrometer-scale� Silicon carbide polishing paper of particle sizes 25�5 µm, 6�5 µm, 2�5 µm, and aluminum oxide paper of particle sizes

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


1 µm, 0�3 µm, and 0�05 µm, were used, from large particles to small� The resulting edge is seen in Figure 1�

The coupler itself is a simple plastic holder that secures in place the waveguide and the fibers on both ends. As shown in Figure 2, the chip and fibers rest on top of the plastic base with all components held down by a room temperature curing epoxy�

We glued down the silicon substrate, aligned the optical fibers for minimum transmission loss, applied epoxy under the fibers and realigned during the curing process. The measuring apparatus is described in Figure 3 and monitored in real-time� Once all the epoxy cured, the device can be detached from the laser and transported as a mobile device�

In order to determine how well the plastic holder held the silicon sample and fibers, we first determined the coupling losses of the on-chip waveguide to the optical fiber without the plastic holder� A cut-back method was employed to find the coupling losses for both a polished and unpolished sample�

Results:We used a 100 µW mid-IR laser at 1540 nm wavelength to determine the coupling loss of the fiber and waveguide. Four different lengths of polished waveguides were measured and linearly fitted to find the propagation loss and coupling efficiency. A single length of unpolished waveguide was also measured and extrapolated to find the unpolished coupling efficiency using the same value for the propagation loss as found in the polished waveguides�

The results, seen in Figure 4, show a propagation loss of 0�62 dB/mm and total coupling losses of 2�53 dB and 5�99 dB, or 1�26 dB and 3�0 dB per each of the two coupling sides, for the polished and unpolished samples respectively� We then successfully pigtailed our device using the plastic holder with a coupling loss faring two times worse than that of an unpigtailed device�

The misalignment during the epoxy curing is still significant, as we have found the epoxy dries to a point where alignment maneuverability was limited, yet still displaced the fibers during curing shrinkage�

Conclusions:We successfully demonstrated the effectiveness of polishing waveguide facets in reducing coupling loss and a pigtailing device� The loss for a permanently pigtailed device is worse than a bare sample, but mobility and utility are invaluably gained, allowing for more complex experiments� Future work can address the misalignment problem in greater detail�

Acknowledgements:I would like to thank my mentor and principal investigator, Parag Deotare and Prof� Marko Loncar, for their valuable insight and guidance throughout the project, Kathryn Hollar and John Free for providing helpful resources, Center for Nanoscale Systems and staff at Harvard, and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and the National Science Foundation for funding the research�

References:[1] S�J� McNab, N� Moll, Y�A� Vlasov, “Ultra-low loss photonic

integrated circuit with membrane-type photonic crystal waveguides”, Optics Express, Vol� 11, Iss� 22, p� 2927-2939, November 2003�

[2] K�K� Lee, D� R� Lim, D� Pan, C� Hoepfner,W� Oh, “Mode transformer for miniaturized optical circuits”, Optics Letters� Vol� 30, No� 5, p� 498, March 2005�

[3] V� R� Almeida, R� Panepucci, M� Lipson, “Nanotaper for Compact Mode Conversion”, Optics Letters, V28, #15, pp� 1302-1304, 8/2003�

[4] W� Joyce, B� Deloach, “Alignment of gaussian beams”, Applied Optics, Vol� 23, p� 4187-4196, 1984�

[5] R� Orobtchouk, L� Pasevi, G� Guillot, ed� Optical Interconnects� Berlin: Springer, 2006� Pg� 264-266�

Figure 3, above: Measurement setup.

Figure 4, below: Cut-back data results.

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Aluminum Nanowire Fabrication for use in Polarization Filters

Nicholas HeugelBiomedical Engineering, Saint Louis University

NNIN REU Site: Nano Research Facility, Washington University in St. Louis, St. Louis, MONNIN REU Principal Investigator and Mentor: Dr. Viktor Gruev, Electrical Engineering, Washington University in St. LouisContact: [email protected], [email protected]

Abstract:

Polarization is a property of light that humans are incapable of seeing, but which contains a plethora of information. Material properties, shape, and surface properties can be determined from polarization, even if the image is hidden in a shadow. To take polarization images with a camera, a polarization filter on the nanoscale must be used. One type of filter is composed of aluminum nanowires. Aluminum and silicon dioxide are thermally evaporated onto a glass slide and then coated with poly(methyl methacrylate) (PMMA). Photolithography and reactive ion etching are used to create the nano features. Then a scanning electron microscope is used to check the features of the wires.

Introduction:

Figure 1: Polarized light interacting with a polarization filter.

Intensity, wavelength, and polarization are the three main properties of light, however, humans are only able to see the first two. The polarization of light contains a plethora of useful information� Shape, surface characteristics, and material characteristics can all be determined from polarization information� Polarization is the angle at which the wave is oscillating as it propagates forward� In Figure 1, there are several orientations to the light approaching the filter even though it is all still moving in the same direction with the same wavelength� Each one of those different orientations represents a degree of polarization from 0 to 180 degrees�

Figure 1 illustrates how a polarization filter works. If the wires are aligned perpendicular to the plane of oscillation, the light is completely quenched� If they are parallel, then the light completely passes through� If nanowires are placed over the pixels of a camera, then different pixels

in the camera will see different pieces of the polarization information�

There are many uses a polarization camera can have� For example, cancer cells have a different cell membrane composition than healthy cells� To the human eye, they both appear the same, but by viewing the polarization image, the cancer cells stand in sharp contrast to the healthy cells, making identifying cancerous tissue during surgery an easier process� Figure 2 shows two different images of the same horse. The first image shows the intensity information, while the second shows the polarization information� In the polarization image, the finer features of the horse are much more apparent compared to the intensity photograph� This is because the polarization information only relies on the angle and degree of polarization while the intensity depends on the amount of light present�

Figure 2: Intensity image of a horse (left) and a polarization image of a horse (right).

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


Methods:The production of the aluminum (Al) nanowires involved several steps� First, glass slides were coated in 140 nm of Al and 20 nm of silicon dioxide (SiO2)� The slides were then coated with poly(methyl methacrylate) (PMMA) to act as a soft mask� Electron beam lithography was used to imprint the nanowire patterns into the PMMA (Figure 3)� The PMMA was developed by exposing the SiO2 below�

An Oxford Plasmalab 100 reactive ion etch system was used to perform the etching� Hydrogen bromide was used to etch through the Al oxide layer and chlorine was used to etch the Al� This step was followed by an oxygen ashing step to burn off any remaining photoresist and remove the SiO2�

The process was first optimized on a micro-sized wire (Figure 4) for speed and then moved to a nanoscale wire pattern�

Results and Discussion:There were several difficulties in optimizing the fabrication of the Al nanowires� First, the photoresist had to be perfectly smooth for features to be entirely transferred to the SiO2� This was most prominently an issue with regard to the edge bead� The increasing thickness of the photoresist, as it approached the edge, caused the lines to grow shallow closer to the edge. This created difficulties in viewing sidewall profiles. The addition of an edge bead remover helped to eliminate this problem, but also made uniformity hard to control�

The second difficulty arose when switching to the nanoscale. The etch recipe that had worked for the micro-wires removed everything when trying to etch the nanowires� Several new recipes using several gases were used, but none were selective enough to prevent over etching of the SiO2 mask� In future attempts, trifluoromethane (CHF3) will be utilized due to its higher selectivity to SiO2 and its sidewall polymer formation, which will help to produce straighter sidewalls�

Once the optimization is completed nanowires of varying sizes should be able to be produced consistently for use in polarization cameras�

Acknowledgements:I would like to thank Professor Viktor Gruev, Raphael Njuguna, and Shengkui Gao in the Electrical Engineering department, Nathan Reed and Kate Nelson with the clean room facilities, Dolores Stewart, and Melanie-Claire Mallison with the NNIN REU Program, and the Nano Research Facilities, NSF, and NNIN for funding�

References:[1] Viktor Gruev, Rob Perkins, and Timothy York, “CCD polarization

imaging sensor with aluminum nanowire optical filters,” Opt. Express 18, 19087-19094 (2010)�

[2] Wang, Jian Jim; Walters, Frank; Liu, Xiaoming; Sciortino, Paul; Deng, Xuegong; “High-performance, large area, deep ultraviolet to infrared polarizers based on 40 nm line/78 nm space nanowire grids,” Applied Physics Letters, vol�90, no�6, pp�061104-061104-3, Feb 2007 doi: 10�1063/1�2437731�

[3] Gruev, V�; Perkins, R�; “A 1 MPixel CCD image sensor with aluminum nanowire polarization filter,” Circuits and Systems (ISCAS), Proceedings of 2010 IEEE International Symposium on, vol�, no�, pp�629-632, May 30 2010-June 2 2010 doi: 10�1109/ISCAS�2010�5537513�

[4] Gruev, V�; Van der Spiegel, J�; Engheta, N�; “Nano-wire dual layer polarization filter,” Circuits and Systems, 2009. ISCAS 2009. IEEE International Symposium on, vol�, no�, pp�561-564, 24-27 May 2009 doi: 10�1109/ISCAS�2009�5117810�

[5] http://www�olympusmicro�com/primer/lightandcolor/polarization�html

Figure 3, above: Nanowire pattern imprinted in PMMA.

Figure 4, below: Aluminum microwires after reactive ion etching.

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Electrical Properties of the Germanium-on-Silicon Interface

Travis LloydEngineering Physics, Brown University

NNIN REU Site: Nanotech, University of California, Santa Barbara, CANNIN REU Principal Investigator: John Bowers, Electrical and Computer Engineering, University of California, Santa BarbaraNNIN REU Mentor: Molly Piels, Department of Electrical and Computer Engineering, University of California, Santa BarbaraContact: [email protected], [email protected], [email protected]


Germanium-on-silicon (Ge/ Si) photodetectors show promise for use in photonics applications due to their high bandwidth, low cost, and compatibility with silicon-based electronics� The performance of such devices is limited by defects at the interface between the two materials� Si and Ge crystals have a 4% lattice mismatch, which results in traps, or energy states between the valence and conduction bands, which carriers can inhabit� Information about how these traps affect performance is useful when designing devices�

This project tested Ge/Si photodetectors in order to characterize their electrical properties� Two separate types of measurements were taken� The open-circuit voltage (Voc) of the detectors was measured as a function of temperature to determine the activation energy of the transport mechanism across the detector’s p-n junction, and the transient decay of an electrical signal under open and short circuit conditions was measured in order to determine the minority carrier lifetime in the device�

Device Specifications:This work tested two different devices� Device A was a Ge waveguide detector with a negative (n)-doped Ge layer, an intrinsic Ge layer, and a positive (p)-doped Si layer� Device B was a top-illuminated uni-traveling carrier (UTC) photodetector with a p-doped Ge layer, an intrinsic Si layer, and an n-doped Si layer�

Experimental Procedure:Open-circuit voltage measurements were performed on devices A and B� Light from a 1310 nm laser was directed by a lensed fiber to the photosensitive region, which was reached via a Si waveguide for Device A and top-down illumination for Device B� The devices were placed on a stage that allowed device temperature to be controlled, and

Figure 1: Block diagram of experimental setup used for transient measurements.

measurements were taken at 5°C increments from 15-65°C� At each temperature, the probe applied a range of voltages to the device and the current at each voltage was recorded� The results were analyzed in order to find Voc, the voltage corresponding to a current of 0A�

Transient measurements were performed on Device B using the experimental set-up in Figure 1� The pulse generator sent a square-wave into the modulator in order to change the light emitted by the 1310 nm laser into an optical pulse of the same shape, and traces of the produced electrical signal were recorded� The “relevant circuit” indicated in Figure 1 is the circuit used to produce either short or open circuit conditions� To create short-circuit conditions, the device was attached directly to a digital communication analyzer (DCA), which provided a 50Ω parallel resistance.

Because the measured current was small, this only resulted in a 12�5 mV voltage drop across the device� To create open-circuit conditions and to amplify the signal to a measureable level, an op-amp circuit with a 2.2 MΩ input impedance was used� To ensure the measured decay was due to carrier recombination, an LCR meter was used to measure the RC limit of the device, and the fall time of the experimental setup was measured with the DCA�

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


Result and Conclusions:The Voc measurements were fit to Equation 1:

[1]

to determine Ea, the activation energy of the transport mechanism across the device’s p-n junction, as seen in Figure 2� This value was calculated to be 0�09 eV less than the ideal value for Device B and 0�29 eV less than the ideal value for Device A� These lowered activation energies demonstrate that traps present in the material contribute a portion of the device’s total current�

The traces from the transient measurements were fit to exponential decays as seen in Figure 3� The open-circuit

measurement yielded a decay constant (τV) of 15�2 ± 1�7 ns and the short-circuit measurement yielded a decay constant (τJ) of 0�543 ± 0�008 ns� Mathematical methods presented in [2] were used to relate τV and τJ to the minority carrier lifetime, τ�

The analysis yielded no self-consistent solution, likely due to assumptions made in the mathematical model that did not apply to these devices� Nonetheless, trends in values of τV, τJ, and τ presented in [3] imply a τ value in the 10-100 ns range�

Reported values for Ge grown on lattice matched substrates are between 1 and 10 µs, demonstrating that defects reduce the minority carrier lifetime in Si/Ge devices�

Future Work:Future work could further characterize the types of traps present in the devices� Defect density in the material could be measured directly through etch pit studies or with the use of transmission electron microscopy (TEM)� Also, simulations could be conducted using the experimental data to determine if a single type, capture rate and recombination velocity describe the traps� Measurements of the temperature dependence of the minority carrier lifetime could yield further information about the types of traps present�

Acknowledgements:I would like to thank Molly Piels for her guidance and assistance, as well as John Bowers and all members of the Optoelectronics Research Group� I would also like to thank Angela Berenstein and Courtney Cechini for their valuable feedback, and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program and the National Science Foundation for funding� Device B was fabricated at UCSB as part of the DARPA CIPhER program� Device A was fabricated by Intel�

References:[1] Kirchartz, T�, Ding, K� and Rau, U� (2011) Fundamental

Electrical Characterization of Thin-Film Solar Cells, in Advanced Characterization Techniques for Thin Film Solar Cells (eds D� Abou-Ras, T� Kirchartz and U� Rau), Wiley-VCH Verlag GmbH & Co� KGaA, Weinheim, Germany� doi: 10�1002/9783527636280�ch2�

[2] Grummt, G., Tousek, J. and Tryzna, B. (1988), “Transients in p+πn+ photodiodes�” physica status solidi (a), 110: 687–695� doi: 10�1002/pssa�2211100242�

[3] Rose, B�H�, “Minority-carrier lifetime measurements on silicon solar cells using Isc and Voc transient decay,” Electron Devices, IEEE Transactions on, vol�31, no�5, pp� 559 565, May 1984� doi: 10�1109/T ED�1984�21569�

Figure 3: Experimental data and theoretical fit for open- and short-circuit transient measurements with DC offset.

Figure 2: Flat band diagram of p-n junction showing mechanism for Device B.

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Flexible Membrane Liquid Lens

David MallinPhysics, University of San Diego

NNIN REU Site: Colorado Nanofabrication Laboratory, University of Colorado, Boulder, CONNIN REU Principal Investigator: Dr. Juliet Gopinath, Department of Electrical, Computer,

and Energy Engineering, University of Colorado, BoulderNNIN REU Mentor: Robert Niederriter, Department of Physics, University of Colorado, BoulderContact: [email protected], [email protected], [email protected]

Abstract and Introduction:Reconfigurable optical devices are important for optical system characterization, materials measurements, communications, and imaging� In particular, variable-focus lenses have potential in many adaptive optical systems� In this project, a lens whose focus could be varied by fluid pressure was designed, fabricated, and characterized� The lens consisted of an aluminum, liquid-filled cavity that is sealed with a flexible polydimethylsiloxane (PDMS) membrane on one side and glass on the other� Pressure changes in the liquid-filled cavity cause the membrane to curve inwards or outwards, resulting in lensing [1, 2]� The goal of this project was the fabrication of a lens that maximized focal length variability while minimizing aberrations� Several liquid lenses were fabricated and demonstrated focal length variability from 20 mm to 125 mm and an average spherical aberration of 5 mm� Current work focuses on optimization of design to increase possible focal lengths and minimize aberrations�

Experimental Procedure:The lens design is shown in Figure 1� Two cavities were drilled into an aluminum (Al) substrate� The Al substrate thickness was 1�5 ± 0�1 mm or 6�5 ± 0�1 mm� One cavity was for the lens and the other acted as a liquid reservoir� The lens aperture diameter was either 12�7 ± 0�1 mm or 19�1 ± 0�1 mm� The two cavities were connected by a shallow channel on one side of the Al� A Fisher Brand 25x75 glass microscope slide sealed the lens cavity and reservoir on the side containing the channel� The slide was attached using a UV curing adhesive to avoid outgassing� The other side was sealed using a thin, flexible polydimethylsiloxane (PDMS) membrane� The membrane was clamped onto the Al substrate with a similar Al top plate. The PDMS films were spun onto Corning 50x75 mm glass slides�

The process began by mixing the two part kit in a 1:10 ratio of curing agent to silicone elastomer� The mixture was put in a sonic dessicator for 30 minutes to remove bubbles� Cleaned slides were coated in a solution, 10% Dawn dish detergent and 90% DI water, that acted as a release layer for the PDMS [3]� The slides were spun with approximately 1�5 ml of PDMS mixture for 40 seconds� Spin speed was varied from 500 to 1900 rpm to produce a range of film thicknesses. The film thickness was measured using a DekTak contact profilometer on a low force setting of 1 mN. The resulting spin curve can be seen in Figure 2�

The PDMS was cured on a hot plate or at room temperature� The baked films were placed on a 150°C hot plate for

Figure 1: A top and side view of the lens designFigure 2: A spin curve for PDMS shows

film thickness as a function of spin speed.

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


35 minutes. The films cured at room temperature took 36 hours. The surface roughness of the films was characterized for both the hot plate and room temperature cure using differential interference contrast (DIC) microscopy� The resulting films were removed from the glass slides and cut to the desired dimensions using a razor blade�

Results and Conclusions:Several flexible membrane liquid lenses were designed, fabricated, and characterized� The lenses were tested for their focal length variability and spherical aberration� In addition, the PDMS films were characterized in thickness and surface roughness�

Focal length variability was tested for the different aperture diameters� The smaller aperture diameter (12�7 mm) had an effective focal range from 20 to 90 mm while that of the larger diameter (19�1 mm) was 30 to 125 mm�

Figure 4: DIC microscope image of 100 µm PDMS film: a) left to cure at room temperature for 36 hours, and b) baked at 150°C for 35 minutes. Long cure times reduced the bubbles and nonuniformities in the PDMS.

Figure 3: A plot of aberration as a function of film thickness. There was no apparent correlation

Spherical aberration was measured as the distance between the marginal focus and paraxial focus using a knife edge test [2]� The spherical aberration varied from 4�36 to 6�55 mm with an average of 5�6 mm� As can be seen in Figure 3, no correlation was found between spherical aberration and film thickness, indicating the large observed aberration was not due to the thickness of the PDMS film.

The PDMS film thickness was measured using a contact profilometer and varied from 30 to 200 µm. Surface

roughness was measured using DIC microscopy� Films cured at room temperature for 36 hours had less surface roughness than those cured at 150°C for 35 minutes (Fig� 4)�

Acknowledgments:I would like to thank Professor Juliet Gopinath and mentor Robert Niederriter for their incredible support and guidance� I would also like to thank Professor Mark Siemens of the University of Denver, Professor Carol Cogswell of the University of Colorado - Boulder, and the entire Colorado Nanofabrication Laboratory staff for their generous support� Funding and support was provided by the National Science Foundation, National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, and the Colorado Nanofabrication Laboratory�

References:[1] Ren, Hongwen, et� Al�;”Tunable-focus Liquid Lens Controlled

Using a Servo Motor”; Opt� Express 14, 8031-8036 (2006)�[2] Werber, Armin, and Hans Zappe. “Tunable microfluidic

microlenses�” Applied Optics 44�16 (2005) : 3238-45�[3] DJ.C. Chang, G.J. Brewer and B.C. Wheeler, A modified

microstamping technique enhances polylysine transfer and neuronal cell patterning, Biomaterials 24 (2003), pp� 2863-2870�

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Soft Lithographic Fabrication of Bar Chart Phantoms for Axial Resolution Measurements in Optical Coherence Tomography

Meagan PipesBiomedical Engineering, North Carolina State University

NNIN REU Site: Stanford Nanofabrication Facility, Stanford University, Stanford, CANNIN REU Principal Investigator: Professor Audrey Ellerbee, Department of Electrical Engineering, Stanford UniversityNNIN REU Mentors: Kristen Lurie, Department of Electrical Engineering, Stanford University;

Tom Gwinn, Department of Electrical Engineering, Stanford UniversityContact: [email protected], [email protected], [email protected], [email protected]

Abstract:

Optical coherence tomography (OCT) enables non-invasive, three dimensional imaging of biological materials based upon reflection of light within a sample. As a relatively new technology, convenient methods for characterizing OCT systems, such as measuring resolution, are needed. Methods for determining lateral resolution must be adapted for use in axial resolution. In this project, we created a phantom with features of known characteristics in the axial rather than lateral plane. We used photolithography to create a master template, which was then used to create polydimethylsiloxane (PDMS) phantoms with soft lithography. PDMS was used for this phantom because it is less reflective than the standard chrome and allows light to penetrate to multiple features at various depths. Using the OCT systems in lab, we determined the advantages and limitations of the phantom through comparison with other accepted characterization methods. A variation of this phantom may provide a convenient method for measuring axial resolution in OCT systems.

Introduction:

Optical coherence tomography (OCT) is a growing technology that enables non-invasive, three dimensional imaging of biological structures� OCT systems have been used to image the sub-layers of the back of the eye (retina) without injury or discomfort to the patient� Unlike traditional microscopy, which images in the lateral plane, OCT images in the axial plane� Characterizing the resolution of an OCT system is important for determining what sizes of features can be detected� In this research project, we created an adaptation of a standard lateral resolution bar chart to measure axial resolution�

OCT works by propagating polychromatic light into a biological sample [1]� When a change in the index of refraction occurs at a boundary between two different sub-layers, some of the light reflects back out of the sample into a detector. The detected signal from the light is mathematically analyzed and plotted as Gaussian-shaped intensity peaks that correspond to the location of the interfaces in depth (Figure 1)� When features are close together, the peaks overlap and appear as one peak instead of two� The distance between two features determines the amount of overlap between the peaks, as illustrated in Figure 2� The current trend in OCT is to define the axial resolution as the full width at half maximum of a single peak, which represents the closest distance at which two peaks are still resolvable� This number, however, does not fully characterize the resolution� In Figure 2, the height from the peak to the valley between the peaks

Figure 1, top: OCT images the layers of a sample by sending in light and detecting the reflected light at the boundaries between sub-layers.

Figure 2, bottom: Current axial resolution estimate, com-pared with contrast information available with a test chart.

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


correlates to the amount of contrast between features in the final OCT image. A more complete assessment of resolution is a function relating the contrast and spatial frequency of features, called the modular transfer function (MTF)� By creating an object with features of known size and a range of spatial frequencies, called a phantom, the MTF for the axial resolution can be determined from the resulting cross-sectional OCT image [2]� The goal of this project was to create a phantom that can be used to fully characterize the resolution of an OCT system in the axial plane�

Experimental Procedure:Using a mask with features of a known range of spatial frequencies in the axial plane, we performed traditional photolithography to create a silicon wafer mold� The wafers were etched using deep reactive ion etching� After applying a silane coat to the etched wafer, a clear polymer called polydimethylsiloxane (PDMS) was mixed and poured into the mold [3]� Cured PDMS was removed from the mold, cut, and placed into the OCT system for imaging� Prior to imaging the phantom in the OCT system, we also wrote code in MATLAB® to simulate how the OCT image should appear�

the PDMS features can be clearly seen, corresponding to the simulation� The OCT image is repeated several times due to additional reflection between features, called multiple interference� Our simulation did not model this�

To create a better PDMS phantom, the process for removing the PDMS stamp must be improved to ensure smooth, rectangular features� Next, the design should be altered to minimize multiple interference in the final image. Once these design problems are solved, this phantom has potential to become a standard for characterizing the axial resolution of OCT systems�

Acknowledgments:NNIN REU Program; NSF; Principal Investigator Professor Audrey Ellerbee; Mentors Kristen Lurie and Tom Gwinn; SNF REU Program Director Michael Deal, Ph�D�; Stanford Nanofabrication Facility Staff; Stanford Biomedical Optics Group�

References:[1] Wojtkowski, M� (2010)� High-speed optical coherence tomography

basics and applications� Applied Optics, Volume 49, Issue 16�[2] Woolliams, P� D�, and Tomlins, P� H� (2011)� Estimating the

resolution of a commercial optical coherence tomography system with limited spatial sampling� Measurement Science and Technology� 22 065502 doi:10�1088/0957-0233/22/6/065502

[3] Qin, D�, Xia, Y�, and Whitesides, G� M� (2010)� Soft Lithography for micro- and nanoscale patterning� Nature 5, 491 - 502 (2010) doi:10�1038/nprot�2009�234

Figure 4: Above, single bars in PDMS imaged in a microscope; below, the corresponding OCT scan in plane of the bars.

Figure 3: MATLAB simulation written to determine how the expected OCT image would appear. On the left is the object to be imaged (i.e., the mask), and the image on the right is the ideal OCT image we expect to see.

Results and Conclusions:The results of the simulation (Figure 3) illustrate the true object being imaged and the OCT image; only the top and bottom boundaries of features can be detected� Once the PDMS features were aligned in the plane of the moving OCT light, the features were successfully imaged in the OCT system (Figure 4)� We observed that the PDMS did not perfectly fill the rectangular wells of the silicon mold, likely due to the ripping of smaller PDMS features upon removal from the mold� The upper and lower boundaries of

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Analog Lithography of Complex Phase Plates for Sub-Diffraction Lithography

Drew D. SchiltzPhysics, Winona State University

NNIN REU Site: Colorado Nanofabrication Laboratory, University of Colorado, Boulder, CONNIN REU Principal Investigator: Dr. Robert McLeod, Electrical Engineering, University of Colorado, BoulderNNIN REU Mentors: Benjamin Kowalski and Darren Forman, Electrical Engineering, University of Colorado, BoulderContact: [email protected], [email protected], [email protected], [email protected]

Abstract:

Analog phase plates enable one to arbitrarily tailor the shape of a laser focus with lower loss than alternative methods such as computer-generated holograms. Applications of these shaped foci include sub-diffraction microscopy and lithography. A half-wave step and a full-wave spiral phase plate are fabricated using analog, maskless photolithography. The phase plates are characterized with a Shack-Hartmann wavefront sensor, a profilometer, a scanning electron microscope (SEM) and other microscopic analysis techniques.

Introduction:

The fabrication of the phase plates was performed using the IMP SF-100 Xpress maskless lithography system� The system used 1024 × 768 grayscale bitmap images, which can vary the intensity of the UV dosage for a given pixel corresponding to a 3�5 × 3�5 µm feature size� This process utilized reflective microoptoelectromechanical (MOEM) elements to vary the intensity of each pixel�

Experimental Procedure:Samples were prepared on 1 × 1 inch glass microscope slides by spinning on a hexamethyldisilazane surface modifier followed by a 1.8 µm layer of AZ-P4210 positive photoresist� The samples were exposed to 365 nm radiation using the IMP SF-100 Xpress maskless lithography system� There was no closed-loop intensity control of the mercury lamp used to expose the samples, causing results to vary on a daily basis� Therefore, each time the lamp was illuminated, recalibration was required�

Samples were tested using a Dektak profilometer to quantify the depth of features� Further examination of the phase plates was provided with optical microscopy and interference microscopy, as well as SEM for sub-micron surface topography information� A Shack-Hartmann wavefront sensor was used to observe the functionality of the phase plates when incident with Gaussian laser radiation�

Results and Conclusions:Two distinct types of phase plates were fabricated: a half-wave step plate and a full-wave spiral plate� Each phase plate

Figure 1: An SEM image of a full-wave spiral phase plate.

was confirmed to have the correct step height to within 10 nm� The spiral phase plate was created with 20 consecutive gray values� One such spiral phase plate can be observed in Figure 1, where lighter regions indicate a deeper feature�

Profilometer measurements and SEM images indicated that the surface of the exposed region was extremely rough compared to the unexposed region� The intensity of this surface roughness increased with the depth of the feature� When the phase plates were tested with a Shack-Hartmann wavefront sensor, it became apparent that a significant fraction of the light was diffracting off the rough surface and didn’t convert into the desired mode�

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


Observing Figure 2, there is a very sharp, vertical step transitioning from the exposed region to the unexposed surface of the spun-on photo resist� This SEM image also demonstrates that the regions of the rough surface were approximately 5 to 10 µm� There was no surface roughness on the vertical step, suggesting that the developing process was not the cause of the surface roughness� Therefore, in an attempt to compare exposure methods, new samples were underexposed using a mask aligner, with every other processing step identical to the samples prepared with grayscale lithography�

Comparing profilometer measurements from a half-wave step plate fabricated with maskless lithography (Figure 3) to a sample underexposed with a mask-aligner (Figure 4), it is evident that the mask aligner resulted in a much smoother surface, independent of the depth of the feature� Not only was the surface smoother, but the surface topography was completely different altogether, as there was no trace of the peaks and valleys that occurred in the maskless lithography samples�

This difference in surface roughness between two samples with different exposure methods reveals that the surface roughness results from the method in which the exposure is controlled in the maskless lithography system� The reflective MOEM elements that control the exposure intensity are believed to be causing this roughness, but it is unclear as to why the regions created on the exposed surface are on the order of 5-10 µm when the spot size for each pixel is 3�5 × 3�5 µm� The roughness is also very randomly oriented, which wouldn’t be expected if it was being generated from

a regular array of reflective MOEM elements. Any non-uniformities resulting from the focusing optics would be expected to produce large regions of distortion as opposed to micron-scale imperfections�

It has been demonstrated that complex phase plates can be fabricated using grayscale lithography, although there is a very rough surface in the exposed region� If the source of the surface roughness can be located and corrected, fully functional, cheap and efficient phase plates will be able to be fabricated�

Future Work:Work will be directed towards locating the source of the surface roughness. The first step in this process involves placing a camera in the maskless lithography system in an attempt to detect variations in intensity� If the source of error can be corrected, complex phase plates will be able to be produced for the desired wavelength�

Acknowledgments:I would like to thank Dr� Robert McLeod, my principal investigator; Benjamin Kowalski and Darren Forman, my mentors; Carol Higgins, my site coordinator; and all the people at the Colorado Nanotechnology Laboratory for all their help� Additional thanks to the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program and the National Science Foundation for funding�

References:[1] Scott, T� F�, B� A� Kowalski, A� C� Sullivan, C� N� Bowman, and

R� R� McLeod� “Two-Color Single-Photon Photoinitiation and Photoinhibition for Subdiffraction Photolithography�” Science 324�5929 (2009): 913-17� Print�

[2] Saleh, Bahaa E� A�, and Malvin Carl� Teich� Fundamentals of Photonics� Chichester: John Wiley and Sons, 2007� Print�

Figure 3, left: Profilometer measurement of a 255 nm step fabricated with the maskless lithography system.

Figure 4, right: Profilometer measurement of a 700 nm step fabricated with a mask aligner.

Figure 2: An SEM image of a step transitioning from the exposed to unexposed region, observed from a shallow angle.

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Graphene-Based Ultrafast Electro-Optical Modulators

Seiya SuzukiGraduate School of Engineering, Toyota Technological Institute, Tempaku-ku, Nagoya, Japan

NNIN iREG Site: Colorado Nanofabrication Laboratory (CNL), University of Colorado, Boulder, CONNIN iREG Principal Investigator: Prof. Thomas Schibli, Physics, University of Colorado at BoulderNNIN iREG Mentor: Chien-Chung Lee, Physics, University of Colorado, at BoulderContact: [email protected], [email protected], [email protected]

Abstract:

Monolayer graphene exhibits many incredible physical properties such its ultrahigh electron mobility and ultrafast relaxation time for photo-excited carriers. This fast relaxation, on the order of eight femtoseconds [1], enables ultrashort pulse generation from mode-locked lasers. Because of graphene’s unique band structure, the optical absorption in graphene is controllable by applying an electric field that changes its carrier density, thus graphene can be exploited as an electro-optic modulator that actively controls optical loss in mode-locked lasers to avoid gain instabilities. In this work, we demonstrated graphene-based electro-optical modulators in a reflective geometry, which consisted of a thin-film structure consisting of a Cu/Al top-electrode on transferred graphene sheet prepared by chemical vapor deposition, and a 185 nm thick Ta2O5 dielectric layer on a reflective Al bottom electrode. The Ta2O5 film thickness was optimized to enhance the electric field of the 1.5 µm light at the location of the graphene. The optical loss of graphene was controlled by an electric field between top and bottom electrodes. The modulator showed a modulation depth close to the theoretical maximum, excellent high-frequency response, and a large active area.

Figure 1: Schematic diagram of the modulator fabrication.

Experimental Procedure:Figure 1 shows the fabrication procedures for the graphene-based electro-optical modulator� A 100 nm thick aluminum (Al) film, as mirror and bottom electrode, was deposited by vacuum evaporator on a microscope glass slide� A tantalum pentoxide (Ta2O5) film was then deposited by DC reactive magnetron sputtering�

The deposited Ta2O5 showed no electrical breakdown for electric fields as high as 1 MV/cm, leading to a large tunable range in the optical absorption of graphene� The optical thickness of Ta2O5 (185 nm) was adjusted to be quarter wavelength of the incident laser (1�5 µm) light, such that the graphene film on top interacted with the

maximum electric field. Monolayer graphene sheets were then prepared by chemical vapor deposition (CVD) and transferred onto the Ta2O5 film. The growth and transferring methods are described in [2]� Patterned Al and Copper (Cu) top electrodes, 100 nm thick each, were then deposited by vacuum evaporator using a lift-off method�

Figure 2: Schematic diagram of the measurement setup for the modulation depth.

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


The modulation depth was measured in a reflective geometry with a lock-in amplifier as shown in Figure 2. The control voltage was applied between top and bottom electrodes� A continuous-wave laser (1�5 µm) illuminated the device under normal incidence and a computer-controlled x-y translation stage was used for obtaining two-dimensional images of the modulation depth�

Results and Discussion:Figure 3(a) shows the dependence of the modulation depth (%) on the root mean square AC voltage (Vrms) driven at 1 kHz� Modulation depths as high as 3�7% at 5 Vrms, corresponding to approximately 75% of the insertion loss of graphene, were observed� This surprisingly high modulation was achieved due to the high-k value of Ta2O5, and the local and enhancement of the electric field of the 1.5 µm light. Figure 3(b) shows the dependence of the modulation depth (%) on driving frequency (kHz) with a driving voltage of 1 Vrms� The modulation depth was nearly independent from the driving frequency at least up to 100 kHz which was our instruments limit�

Figures 4(a) and (b) show an optical microscope and a two-dimensional modulation depth image of the same region� The inner diameter of the ring electrode is 120 µm� The modulator was driven by 1 Vrms at 1 kHz� The measured

region was entirely covered by graphene except for the cracks pointed out in Figure 4(a)� As shown in Figure 4(b), the electrode and cracks did not show modulation but the inside of the ring showed large modulation depths with high uniformity corresponding to approximately 10,000 µm2, which were clearly coincident with Figure 4(a)� The regions of no modulation outside of the ring were surface impurities or defects in the graphene induced by the fabrication process� On the other hand, modulation was observed over the entire area of the graphene sheet, even 1000 µm away from the electrodes� This indicates that the transferred CVD graphene was continuous and our fabrication processes of the modulator were reproducible�

Conclusions and Future Work:We demonstrated graphene-based electro-optical mod-ulators in a reflective geometry. The modulators showed surprisingly high modulation depth of 3�7% at 5 Vrms owing to the high-k value of Ta2O5, and the local enhancement of the electric field of the 1.5 µm wavelength light. The frequency response was higher than 100 kHz� The entire area of graphene, even 1,000 µm away from electrodes, showed uniform modulation, which indicates the high uniformity of the transferred CVD graphene�

With further improvements in the mirror design and patterning of graphene, and by combining this graphene-based modulator with the inherent saturable absorption of graphene, an active/passive hybrid ultra-fast pulse generator for mode-locked lasers can be achieved�

Acknowledgements:Special thanks to Prof� Thomas Schibli and Chien-Chung Lee for guiding me throughout the project� Also, thanks to the CNL staff, including Prof� Bart Van Zeghbroeck and Tomoko Borsa for their efforts and assistance� Finally, I would like to acknowledge the generous support from the NSF through the NNIN iREG Program, and from the National Institute for Materials Science in Japan�

Figure 4: (a) An optical microscope image of the graphene electro-optic modulator. The inner diameter of the ring electrode is 120 µm. (b) A two-dimensional modulation depth (%) image of the same regions as in (a). The modulator was driven by 1 Vrms at 1 kHz.

Figure 3: The dependence of the modulation depth (%) on (a) the AC voltage (in Vrms) and (b) driving frequency (kHz) at driving frequency of 1 kHz and voltage of 1 Vrms, respectively.

References:[1] Guo, X�G� et al�; “The Physics of

Ultrafast Saturable Absorption in Graphene”; Optics Express, 18, 4564 (2010)�

[2] Li, X� et al�; “Graphene Films with Large Domain Size by a Two-Step Chemical Vapor Deposition Process”; Nano Lett�, 10, 4328 (2010)�

• o

pti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •

opti

cs &

opto

-ele

ctro

nic

s •


Zero-Mode Waveguides for Single-Molecular Imaging

Jin ZhangChemical Engineering, University of Arizona

NNIN REU Site: Lurie Nanofabrication Facility, University of Michigan, Ann Arbor, MINNIN REU Principal Investigator: Pei-Cheng Ku, Electrical and Computer Engineering, University of MichiganNNIN REU Mentor: Chu-hsiang Teng, Electrical and Computer Engineering, University of MichiganContact: [email protected], [email protected], [email protected]


Zero-mode waveguides (ZMW) are arrays of sub-wavelength apertures in a metal film that allow for the observation of single-molecule interactions at micromolar concentrations through the excitation of fluorescently tagged molecules. When light is shown through a zero-mode waveguide, photons with longer wavelengths are unable to propagate through the waveguide and only evanescent modes will exist. In a fluorescent microscope, the evanescent waves will exponentially decay at the glass/water interface, leading to a detection volume on the scale of zeptoliters (10-21 L) [1]� Zero-mode waveguides have many biological applications, such as allowing real-time deoxyribonucleic acid (DNA) sequencing [2], and observing protein-protein interactions [3]� To date, fabrication of ZMWs requires expensive and/or low-throughput nanolithography such as focused ion beam milling, e-beam lithography and dry etching, e-beam lithography and lift-off, deep UV lithography and lift-off, and nanoimprint and lift-off�

The aim of this project was to utilize an alternative method of fabrication involving conventional photolithography, lift-off, and electrodeposition to achieve affordable and efficient single-molecular imaging that could be adopted in most cleanroom facilities�

Experimental Procedure:Conventional photolithography (AutoStep200) and negative photoresist was used to create a square pattern of waveguides surrounded by four square trench patterns� Lithography doses were adjusted, ranging from 0�15s to 0�22 s, and tested to determine the best parameters� Following photolithography, samples were baked in a large oven for 30 min for image reversal properties of the negative resist to take effect� Baking temperatures of 102ºC, 107ºC, and 112ºC were tested to determine the temperature that would yield optimum results� After baking, all samples were flood-exposed (40s, MA-6/BA-6 mask/bond aligner) and developed in solution� Development time was adjusted once again to determine the optimum time for each exposure dose�

Figure 1: Image of waveguides after lift-off (left). Image of final waveguide after plating (right).

After the photolithography patterning, 100Å of chrome (Cr) and 1000Å of gold (Au) were deposited on the surface using electron-beam evaporation physical-vapor deposition (EnerJet Evaporator)� Lift-off of photoresist was performed using a heated solution of PRS-2000 (Baker)�

In order to shrink the size of waveguides produced after lift-off, samples were further electroplated with gold� Waveguide diameters were measured using a scanning electron microscope (SEM) and variation in diameter locally and globally were calculated to quantify uniformity and consistency of fabrication procedures�

Results and Discussion:Waveguides 50 nm to 100 nm in diameter were successfully fabricated, and molecular diffusion testing showed single-molecule detection at a concentration of 1 µM� The optimum conditions for fabrication included lithography doses from 0�16s to 0�19s, a baking temperature of 102ºC, and lithography dose dependent development times from 27s to 36s� To determine the uniformity and controllability of the fabrication process, the global and local uniformity in diameter size was calculated for 3×3-inch samples� Local uniformity refers to waveguides within the same cell whereas global uniformity refers to diameter measurements throughout the entire sample�

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

• optics

& o

pto

-ele

ctronics

•


Under optimum conditions, the best local uniformity achieved was in a cell in the center of the sample, with a mean diameter of 54�4 nm, standard deviation 9�6 nm, and a range of ± 20�2 nm� However, the samples were not as uniform globally as they were locally� While many cells achieved high uniformity, cells around the outer edges of the sample generally had a lower uniformity, indicating a problem in the fabrication process� This may be due to lack to uniformity in photoresist thickness, which then creates irregularities in pillar size and waveguide diameter� Furthermore, random fluid movements during development may cause sections around the edge to be less uniform�

Figure 3 shows the relationship between fluorescence intensity and waveguide diameter for diameters ranging from 50 nm to 1000 nm� Waveguides below and above 350 nm in diameter showed 3�01 and 2�35 power depend-ences of fluorescence intensity versus diameter, respectively.

In small waveguides, fluorescence intensity is a function of excitation intensity in the waveguide and the fluorescence out-coupling efficiency. In larger waveguides, light is able to propagate through the aperture, and fluorescence intensity is mainly determined by the area of the aperture, which is theoretically second-power dependent on the radius or diameter�

Conclusions:Waveguides of 50-100 nm were fabricated and tested to yield single-molecule detection at a concentration of 1 µM� The signal to noise ratio of detection was approximately 2:1 and can be further improved� Fluorescence intensity measurements taken using waveguides of different diameters showed an approximately third power dependence in smaller waveguides and an approximately second power dependence for larger waveguides� While local uniformity in diameter is fairly good, future studies should aim to improve global uniformity�

Acknowledgements:I would like to thank my PI Pei-cheng Ku, my mentor Chu-hsiang Teng, and LNF site coordinators and staff for their continued help and support� This project is funded by NSF, partially via the NNIN REU Program and partially NSF CBET Grant #0966723�

References:[1] Levene, M� J� et al�, Zero-mode waveguides for single-molecule

analysis at high concentrations� Science 2003, 299, 682�[2] Flusberg, B� A� et al�, Direct detection of DNA methylation during

single-molecule, real-time sequencing� Nature Methods 2010, 7, 461�

[3] Miyake, T. et al., Real-time imaging of single-molecule fluorescence with a zero-mode waveguide analysis of protein-protein interaction� Anal� Chem� 2008, 80, 6018�

Figure 3: Fluorescence intensity at waveguides of various diameters ranging from 50-1000 nm. Experiments are conducted with green light.

Figure 2: Single-molecule detection of 1 µM TMR-streptavidin in a 70 nm waveguide. Peaks of higher fluorescence indicate excitation of single molecules in the waveguide.

Using a waveguide 70 nm in diameter, diffusion experiments were conducted to test the effectiveness in single molecule detection� Figure 2 shows that single molecule detection was achieved in a solution 1 µM in concentration with a signal to noise ratio of about two�

Background noise was still fairly significant and could be further reduced to improve detection. Background fluor-escence noise may be caused by auto-fluorescence of contaminants or surface reflection of Au or Cr. This may be reduced through better cleaning methods prior to Au electroplating�

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


Electrical Single Molecule Investigations by Means of Mechanical Break Junctions

Brian BentonPhysics, University of Minnesota – Twin Cities

NNIN iREU Site: Institut Für Bio- Und Nanosysteme (IBN), Forschungszentrum, Jülich, GermanyNNIN iREU Principal Investigator: Dr. Dirk Mayer, Peter Grünberg Institute 8, Forschungszentrum JülichNNIN iREU Mentor: Feliks Pyatkov, Peter Grünberg Institute 8, Forschungszentrum JülichContact: [email protected], [email protected], [email protected]

Abstract:

The mechanically controllable break junction (MCBJ) technique was employed to investigate the conductance of single molecules. Octanedithiols have been extensively investigated in previous work and were measured as a check of our techniques and equipment. Dendrimer molecules with a ferrocene core (CSA-FcGluOH) were measured using these techniques for the first time. Artifacts in the measurements resulting from our lithographically prepared samples hindered our ability to make accurate and reliable measurements. However, extensive studies of the MCBJ and measurement tool behaviors were carried out to improve future work.

Introduction:

As the size of silicone-based electronic components reaches its limit, investigations of alternative electronics technologies have become increasingly vital� One promising area is the field of molecular electronics. A crucial step towards realizing these goals is the characterization of a variety of molecules and their typical conductance values, reported in terms of the conductance quanta, G0, of a single atom� Our chosen method for this process was the mechanically controllable break junction (MCBJ) technique� This employs a piezo to control the position of a pushrod, which then bends the flexible substrate upon which the junction lies� The contacts are separated with an aspect ratio in relation to the pushrod position on the order of 1:200000� This creates an incredibly stable junction because small vertical variations will not greatly influence the gap size. The MCBJ method was employed to investigate bare gold electrodes as well as 1,8-octanedithiol test molecules, which have been extensively investigated in previous work� The ultimate objective was to carry out the first investigations of a set of dendrimer molecules containing a ferrocene core�

Experimental Procedure:Because of the variations in the single molecule contact geometries in the measurements, a statistical investigation was required [1]� This necessitated the taking of a large number of measurements� Such measurements involve opening the junction slowly with an applied voltage bias while recording current and voltage readings� The

conductance values for single atom conductance and single molecule conductance are seen in these measurements as plateaus in the conductance vs� time plots as the junction is separated� An example can be seen in Figure 1, which is a plot from a nanowire sample with 1,8-octanedithiol molecules� The single atom conductance plateau can be seen at 10-6 amps, while the single molecule conductance plateau can be seen at 10-10 amps�

After at least 100 such measurements are carried out with a given sample, the conductance values can be plotted in a histogram� The plateaus appear as peaks in the histogram corresponding to the various stable values observed through out the breaking of the junction� A strong single atom peak should always be observed (along with less pronounced two and three atom peaks)� Lower conductance

Figure 1: Current (amps) vs. time (seconds) curve for junction opening with 1,8-octanedithiol on nanowire. Voltage is 13 mV, pushrod velocity is 3 µm/sec.

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


peaks should then correspond to the typical conductance values of the molecules under investigation� The variation in these values corresponds to geometric and contact fluctuations as well as different possible molecular conformations across the junction�

Results and Conclusions:Measurements were carried out with both macrowire (0�1 mm diameter gold wire glued to a spring steel substrate) and nanowire (lithographically fabricated gold on a spring steel substrate) samples� A histogram of the results from the macrowire samples with no molecules is shown in Figure 2 along with the histogram for the nanowire samples with no molecules� The artifacts observed in the nanowire histogram are not present in the macrowire data� The inset shows the high conductance region with single and multi-atom peaks, providing confirmation of the quality of our equipment and methods� Note how the peaks are more pronounced in the macrowire histogram� These results indicate that, while the nanowire samples should be more stable, macrowire samples can still provide more reliable results�

Figure 3 shows the histogram for the nanowire samples with 1,8-octanedithiol molecules� The artifacts in the histogram make the results less obvious, but a clear peak can still be seen around 10-4G0, a similar result to previous work� Results in the literature can vary over a wide range, however, due to different electrode and apparatus designs� One previously published paper indicates a conductance between 10-4G0 and 3*10-4G0 [2]�

Figure 4 shows histograms from five different nanowire samples with CSA-FcGluOH molecules. Note that there are no significant peaks found in all five histograms. This may be partially due to the large effect of the artifacts obscuring the results� However, since these molecules have not been measured before, it is possible that the conductance lies below our working range or, due to the increased length and complexity of these molecules compared to others commonly investigated, there may not be such well defined peaks to be found�

Acknowledgements:Special thanks to Dr� Dirk Mayer and Feliks Pyatkov for their support and guidance throughout the summer� Also, thanks to Forschungszentrum Jülich and the faculty and staff, including Dr� Andreas Offenhäuser, for providing an excellent working environment and access to a wide range of tools� Funding was provided by the National Nanotechnology Infrastructure Network International Research Experience for Undergraduates (NNIN iREU) Program and the National Science Foundation�

References:[1] Lörtscher, E� et al; “Statistical Approach to Investigating Transport Through

Single Molecules”; Physical Review Letters, 98, 176807 (2007)� [2] Kim, Y� et al�; “Characteristics of Amine-Ended and Thiol-Ended Alkane

Single-Molecule Junctions Revealed by Inelastic Electron Tunneling Spectroscopy”; ACS Nano, 5 (5), 4104-4111 (2011)�

Figure 4: Histogram of data from four different samples of CSA-FcGluOH molecules measured with nanowire samples. 290 curves total (50-60 each). The histograms are shifted vertically to separate them.

Figure 2: Histograms for both macrowire (solid) and nanowire (dashed) data with no molecules. 132 and 180 curves, respectively. Macrowire histogram scaled to match nanowire. Inset: High conductance region showing the atomic conductance peaks.

Figure 3: Histogram of data for 1,8-octanedithiol molecules measured with nanowire samples. 90 curves.

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


Investigation of Electron Transport in Functionalized Carbon Nanotubes

Mark BrunsonMechanical Engineering, San Francisco State University

NNIN iREU Site: Institut Für Bio- Und Nanosysteme (IBN), Forschungszentrum, Jülich, GermanyNNIN iREU Principal Investigator: Dr. Carola Meyer, Peter Grünberg Institute 6, Forschungszentrum JülichNNIN iREU Mentor: Robert Frielinghaus, Peter Grünberg Institute 6, Forschungszentrum JülichContact: [email protected], [email protected], [email protected]

Introduction:

Since the discovery of carbon nanotubes (CNTs), they have shown potential for use in applications that benefit from their mechanical or electrical properties� CNTs are hollow cylinders of graphene sheets, which can consist of one or more walls depending on the growth parameters� They can be metallic or semiconducting�

One advantage of CNTs is the ability to tune properties by functionalizing them with various molecules� In this work, characterization of CNTs functionalized with tailor-made Mn4 molecules was performed� These manganese molecules were single molecular magnets, which allowed for control of the spintronic properties of CNT devices� Oxidizing the CNTs introduced binding sites to the surface for the carboxylate groups of Mn4 clusters to bond�

Experimental Procedure:CNTs were synthesized on silicon (Si) substrates using a chemical vapor deposition (CVD) process� Ferritin was

Figure 1: Image of a typical device obtained by scanning electron microscopy (SEM). Each nanotube is contacted by four isolated contacts.

used as a catalyst; it was deposited onto the Si/SiO2 substrate and heated in air to 450°C in order to remove the protein surrounding the iron core� The sample was then placed in a quartz tube of a tube furnace, purged in argon, and heated to 850°C� Upon reduction in 1�0 slpm hydrogen, CNTs were synthesized for 10 minutes with 0�52 slpm methane and 0.7 slpm hydrogen. The sample was finally cooled to room temperature in argon�

In order to fabricate the device, a marker structure and bonding pads were written using electron beam (e-beam) lithography with 5 nm titanium and 60 nm gold for the structures. Sufficiently long and isolated CNTs were located using atomic force microscopy (AFM) and four-point contacts were designed� The contacts were written using the same lithography process: this time with 3 nm chromium and 60 nm palladium contacts to reduce contact resistance� An example of a typical device is shown in Figure 1� Since this sample was destroyed during the oxidation process for functionalization, a similar sample, which was instead made with titanium and gold contacts, was measured�

A three-point probe station was used to identify working devices by measuring both the bias sweeps and gate sweeps� These measurements also indicate the type of nanotube: semiconducting or metallic� A dilution refrigerator was used to observe the Coulomb blockade effect in a single nanotube, which behaves as a quantum dot� Once cooled to 70 mK, a dual sweep was performed such that there was a bias sweep of ± 4 mV for each gate voltage measured� Since a new matrix box was fabricated to reduce the cable length connecting the sample and the transducer, the new measurement setup required testing�

Results and Conclusions:Though the CNTs were spread out with a slightly higher than optimal density, the sample was nevertheless processed� However, due to poor liftoff, gate leakage was evident in

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


all but two regions� Transport measurements were then performed so that influence from the chemical treatment could be quantified. In the electrical measurements at room temperature, the CNTs were all observed to be metallic, though mechanical strain suppressed the current in some cases. Resistances of ohmic nanotubes ranged from 80 kΩ to 1.4 MΩ.

In order to prepare the sample for functionalization, oxidation was performed by heating up the sample in air to 450°C� After oxidation, none of the previously working devices were still functional� However, the resistivity of the palladium was unchanged� It is possible that the contact resistance has greatly increased due to the difference of the coefficients of thermal expansion of the three present materials: gold, chromium, and palladium� Attempts to reduce the sample at 450°C in hydrogen were unsuccessful�

the dilution refrigerator has in fact been optimized and is prepared for the necessary experiments with functionalized CNTs�

Future Work:The effects of functionalizing CNTs with the manganese molecule clusters still need to be observed� After these measurements have been successfully performed, the manganese can be substituted in the cluster with other metals� Compounds with various magnetic properties will be bonded to the CNTs making it possible to observe the coupling of the CNT electronic system to the molecular magnets�

Acknowledgments:The guidance of Dr� Carola Meyer and Robert Frielinghaus throughout this project is greatly appreciated� I would also like to thank Karin Goss, Caitlin Morgan, and Dr� Claire Besson for their input and involvement� My part in this project was made possible by the National Nanotechnology Infrastructure Network International Research Experience for Undergraduates (NNIN iREU) under a grant from the National Science Foundation (NSF)� This research was conducted in the Peter Grünberg Institute (PGI-6) at Forschungszentrum Jülich�

References:[1] Fuhrer, A�, Fasth, C�; “Coulomb Blockade in Quantum Dots”;

Lecture Notes, 23 April 2007�

Figure 3: Coulomb peaks are fitted using the values obtained from the measured Coulomb blockade to determine the electron temperature.

Figure 2: Measurement of the Coulomb blockade structure shows excited states as extra lines parallel to the topsides of the diamonds. Regions of negative differential conductance are also observed.

In order to test the optimized setup for the dilution refrig-erator, a dual sweep was performed to obtain the Coulomb diamond structure of a CNT device� The diamonds, shown in Figure 2, are well defined, showing multiple excited states and regions of negative differential conductance� Coulomb peaks can be seen where the diamond points meet� The lever arm, aG, was determined to be 0�0014� Using this information, the Coulomb peaks can be fit to determine the electron temperature [1]� From the peaks with the least disorder, the electron temperature is found to be 157 mK, as shown in Figure 3� This result was expected since the lattice temperature is 70 mK� This measurement shows that

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


Capacitance Measurements of Single Indium Arsenide Nanowires

Kevin ChenElectrical Engineering, Arizona State University

NNIN iREU Site: Institut Für Bio- Und Nanosysteme (IBN), Forschungszentrum, Jülich, GermanyNNIN iREU Principal Investigator and Mentor: Mihail Lepsa, PGI-9 Semiconductor Nanoelectronics,

Peter Grünberg Institute, Forschungszentrum, Jülich, GermanyContact: [email protected], [email protected]


Recently, semiconductor nanowires have been given much attention as potential candidates for future high-end electronic devices due to their one-dimensional transport and favorable geometry for processing all-around gates with improved electrostatics� As for the material for the nanowire, of particular interest is indium arsenide (InAs), which has a very high bulk mobility at room temperature� However to date, very little experimental evidence is available on the mobility of single InAs nanowires due to the difficulty of obtaining accurate gate capacitance measurements over parasitics which can be orders of magnitudes higher� In this project, InAs nanowire field effect transistors (FETs) were fabricated and capacitance-voltage (C-V) measurements were carried out using a calibration setup involving open and short circuits to calibrate out any parasitic elements�

Experimental Procedure:Electron beam lithography (EBL) was used to define contact pads and scanning electron microscopy (SEM) markers (Figure 1) on p-type silicon wafers with 200 nm of thermally grown silicon dioxide (SiO2)� The contact pads were designed as 60 × 60 µm squares with 40 µm spacing

Figure 1: Optical image of a device for capacitance measurements.

to be compatible with RF measurement probes which were used to conduct the capacitance measurements�

InAs nanowires grown by molecular beam epitaxy (MBE) were transferred onto the samples using a mechanical transfer method� The wires, characterized beforehand, had an average length and diameter of 1�9 µm and 93 nm respectively. SEM images of each field with markers were taken to locate the position of the nanowires� Contacts to the wires were subsequently designed in Autocad using the SEM images�

Three different types of devices were designed and processed so that different types of measurements could be carried� Four terminal devices were fabricated for four point probe measurements to extract the resistivity of the wires� Three terminal FET devices with a top gate were fabricated so that transistor measurements could be carried out� Also, two terminal devices with a top gate and shorted source and drain contacts were designed for capacitance measurements� Regarding the devices fabricated for capacitance measurements, two more fields with identical contact structures—one with a short between the source, drain and gate contacts in place of the nanowire and one with an open circuit—were also fabricated for calibration purposes� The capacitance of the open circuit is measured to calibrate out the parasitic capacitances while the short circuit is used to account for any parasitic inductances�

The source-drain contacts were defined using EBL and after metal deposition, 50 nm of lanthanum lutetium oxide (LaLuO3) with a dielectric constant of 20 was deposited as a high-k gate dielectric onto the samples using pulsed laser deposition (PLD)� The gate contacts were then processed using EBL� Finally, using EBL again, the LaLuO3, on top of the contact pads, was etched by a buffered hydrochloric acid/ammonia solution so measurement probes could contact the pad metal�

Results and Conclusions:Four-point-probe resistivity measurements were carried out on nanowire devices with inner contact spacings of approximately 600 nm� Resistances ranging between

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


Figure 2: Drain current vs. drain voltage graph for different gate voltages.

Figure 4: Capacitance-voltage characteristics obtained with an AC signal of 500 mV at 45 MHz.

Figure 3: Drain current vs. gate voltage graph for different drain voltages.

4-20 kΩ were measured, resulting in resistivities between 4.5-23 mΩ-cm. Two point probe resistance measurements were also carried out to determine the contact resistances of the devices� Measuring resistance values between contacts of different spacings, and extrapolating down to a spacing of 0.0 nm, yielded contact resistances in the 1 kΩ range, indicating good ohmic contacts�

For the transistor measurements, sweeps of both source-drain and gate voltages were conducted to obtain characteristic curves (Figure 2) and the transfer characteristics of the nanowire FETs (Figure 3)� The measured device had a gate length of 160 nm and source-drain distance of 2�25 µm� From the transfer characteristics graph, we found the transconductance of the device, calculated as gm = dId/dVg, to be 1�14 µS at a source-drain bias of 25 mV�

Capacitance measurements were carried out with an Agilent 4294A Precision Impedance Analyzer using RF probe tips

at a frequency of 45 MHz with a 500 mV signal [1]� From Figure 4, it can be seen that the measured capacitance with no applied bias is 620 aF. The field effect mobility is calculated as [2]

where LG is the gate length, C is the gate capacitance and VDS is the applied source-drain voltage� Using the data from the transistor measurements, this yields a field effect mobility of 188 cm2V-1s-1� We were unable to obtain a full CV curve because as can be seen in the figure, the device breaks down at around 10 V bias�

The theoretical gate capacitance of the wires is calculated as [2]

where tox is the oxide thickness and r is the nanowire radius� This leads to a theoretical capacitance of 131 aF, which is much lower than the capacitance measured� This large difference can be partly attributed to a fabrication error for the shorted calibration devices which required manually fabricating a short with a bonded wire, decreasing the accuracy of the measurement�

Future Work:Through this project, we have successfully fabricated nanowire FETs using MBE grown InAs nanowires and shown the possibility of obtaining the capacitances of individual nanowires� However, more work must be done to ensure that the measurements are more accurate and future work will include fabricating devices with working short calibrations and possibly doing capacitance measurements in a low temperature environment�

Acknowledgements:Special thanks goes to the NNIN iREU Program and the NSF for the funding that made this experience possible along with the Forschungszentrum Jülich for hosting me� Also, I would like to thank my principal investigator, Dr� Mihail Ion Lepsa, for his guidance and the staff at the Peter Grünberg Institute for assisting me with my project�

References:[1] “The Parametric Measurement Handbook”; Agilent Tech� (2010)�[2] Dayeh, S�, et al�; “High Electron Mobility InAs Nanowire Field-

Effect Transistors”; Small, 3, 326-332 (2007)�

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


Resistive Switching of Iron-Doped SrTiO3

Zachary ConnellMechanical Engineering, University of Nebraska - Lincoln

NNIN iREU Site: Institut Für Bio- Und Nanosysteme (IBN), Forschungszentrum, Jülich, GermanyNNIN iREU Principal Investigator: Dr. Regina Dittmann, Electronic Materials Research Lab (EMRL),

Peter Grünberg Institut, Forschungszentrum JülichNNIN iREU Mentor: Christian Lenser, EMRL, Peter Grünberg Institut, Forschungszentrum JülichContact: [email protected], [email protected], [email protected]

Abstract:

We report on the resistive switching behavior of iron-doped strontium titanate (SrTiO3 or STO), particularly related to electroforming. Epitaxial iron:STO (Fe:STO) thin films were deposited on a single crystalline nio-bium:STO (Nb:STO) substrate with pulsed laser deposition (PLD). Large defect structures were located on the film using scanning electron microscopy (SEM). Platinum electrodes were placed over areas both with and without these defects; electroforming and switching behavior were characterized. Results indicated that defect structures have significantly different electroforming behavior than other areas of the film.

Introduction:

As the scalability limits of transistor-based memory approach, it is becoming increasingly necessary to find alter natives to modern flash memory. One such alternative is to utilize resistive switching effects to create resistive random access memory (RRAM)� Resistive switching is the phenomena in which certain materials can be switched between high and low resistance states, which at as binary “ON” and “OFF” states in memory� In resistive switching devices, a typically insulating layer, generally an oxide, is deposited between two conducting layers� The application of voltage switches the device between high and low resistance states� In some cases, an electroforming step is needed to activate the switching�

In this case, the one-time application of voltage changes the switching device from its initial virgin state and enables resistive switching� STO is often used as a model material for resistive switching� However, there lacks a clear consensus and understanding for the underlying physicochemical processes involved in resistive switching, particularly those for electroformation [1]�

Experimental Procedure:To begin the fabrication process, 0�5 wt�% Nb-doped STO substrates were annealed for three hours to ensure a terrace step topography, which was then verified with atomic force microscopy� These substrates also act as the bottom electrode of the switching device� Epitaxial 5 at�% Fe-doped STO thin films were deposited using PLD, of thickness either 20 nm or 100 nm. On some samples, a 30 nm platinum thin film

Figure 1: A defect structure located on the surface of an Fe:STO thin film.

was then sputtered, patterned with photolithography, and etched using reactive ion etching (RIE) to create arrays of 200 µm by 200 µm top electrodes�

On a 100 nm Fe:STO thin film sample, a layer of poly(methyl methacrylate) (PMMA) was spun on top of the Fe:STO thin film, marker structures were patterned with e-beam lithography and etched with RIE prior to the sputtering of the 30 nm platinum film. The PMMA was then removed, leaving behind platinum marker structures on the Fe:STO thin film. Then, using SEM, the film was examined and large defect structures were located relative to the marker

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


structures� Figure 1 is an SEM image of one such defect structure located on a different Fe:STO film. Another layer of PMMA was spun, patterned, and etched using the same process to create wells for the top electrode pads� Subsequently, 30 nm of platinum was sputtered on top� The PMMA was then removed, leaving behind square platinum electrodes of side length 3 µm, 5 µm, or 10 µm� These electrodes were placed over areas that defect structures as well as areas that were free of noticeable surface defects�

Future Work:To generalize these results, a greater range of iron con-centrations should be examined in the STO thin films. Furthermore, other defect structures should be tested, including surface defects smaller than the defect structures tested in this experiment� Finally, more tests are required to increase confidence bounds on electroforming statistics.

Acknowledgements:Many thanks to Dr� Regina Dittmann, Christian Lenser, Prof� Rainer Waser, my group in PGI-7, Forschungzentrum Jülich staff, and the National Nanotechnology Infrastructure Network International Research Experience for Under-graduates (NNIN iREU) Program staff� I would also like to thank the NNIN iREU and the National Science Foundation for funding�

References:[1] Waser, R�; “Redox-Based Resistive Switching Memories –

Nanoionic Mechanisms, Prospects, and Challenges”; Advanced Materials, 21, 2632-2663 (2009)�

[2] Muenstermann, R�; “Coexistence of Filamentary and Homogeneous Resistive Switching in Fe-Doped SrTiO3 Thin-Film Memristive Devices”; Advanced Materials, 22, 4819-4822 (2010)�

Figure 3, above: A sample area free of large defects (left) and its current versus time curve at 12 V electroforming (right).

Figure 4, below: Current-voltage plot after electroforming (right) for an area with a large defect structure (left).

Figure 2: Current plotted as a function of applied voltage for 20 nm thin film Fe:STO sample (a) and 100 nm thin film sample (b).

Results and Conclusions:For the 20 nm and 100 nm thin film samples, the switching behavior of the 200 µm side length pads was characterized by measuring current as a function of voltage� Figure 2 shows the plots of absolute value of current versus applied voltage for an electrode pad on the 20 nm thin film sample (left) and on the 100 nm thin film sample (right). The 20 nm sample displays an eightwise switching behavior whereas the 100 nm sample shows countereightwise switching� All measurements on the 20 nm sample displayed this switching pattern, whereas some 100 nm pads displayed eightwise switching� Literature suggests that this indicates that both filamentary and homogenous switching mechanisms are present in these samples [2]�

In one 100 nm thin film sample, a significant difference was observed between areas free of large defects and areas that had such defects� For areas without defects, it was not possible to electroform the devices even for large voltages, 10 or 12 V, held for over an hour� Figure 3 shows the voltage versus time for a 5 µm pad free of large defects� However, the samples with large defects formed rapidly� Figure 4 shows the current-voltage plot for one such pad that formed in ~ 1�8 s at 10V� It displayed counter-eightwise switching, but other large defect structures had eightwise switching behavior� The extreme difference in formation time between areas with and without large defect structures suggests that these structures play a significant role in the electroforming process�

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


Growth of Silicon, Silicon Carbide, and Boron Nitride Nanowires for Electronic Applications

Won Jun KukChemistry, Williams College

NNIN REU Site: Howard Nanoscale Science and Engineering Facility, Howard University, Washington, DCNNIN REU Principal Investigator: Dr. Gary L. Harris, Electrical Engineering, Howard UniversityNNIN REU Mentor: Crawford Taylor, Research Associate, Howard UniversityContact: [email protected], [email protected], [email protected]

Abstract:

The objective of this project was to explore the growth of silicon, silicon carbide, and boron nitride nanowires for electronic device applications. These nanowires were chosen because they all have excellent mechanical, optical and electrical properties. In this project, we grew Si and silicon carbide wires with diameters less than a hundred nanometers (nm). The wires were grown in a horizontal chemical vapor deposition system using precursors of silane, propane, ammonia and diborane at 200 torr. The deposition temperatures were varied from 800 to 1100°C. The wires were grown on silicon substrates using catalysts of 5-10 nm nickel, aluminum, gold, iron and cobalt. Experiments with the boron nitride nanowires did not produced any significant results.

Introduction:

Nanowires hold a lot of promise for applications in the next generation of electronics because they can be used to create extremely powerful and versatile circuits� They can be mass-produced by chemical vapor deposition (CVD)� This project focused on the growth of silicon, silicon carbide, and boron nitride nanowires by low pressure CVD�

Bulk silicon (Si) is the base material for most modern electronics in use today� Si, on the nanoscale, carries many of the same mechanical, electrical, and optical properties of bulk Si that make it viable for electronic applications� Furthermore, Si nanowires have been shown to display a thermoelectric effect, which allows them to convert excess heat back into electrical energy� With this effect, Si nanowires can be applied to electronics ranging from car engines to computer chips to increase their efficiency and reduce thermal waste�

Silicon carbide (SiC) nanowires display high thermal conductivity and mechanical strength� While Si is used for general semiconductor construction, SiC, due to its high bandgap, can be applied in high power or high temperature applications� Furthermore, SiC nanowires are chemically inert, allowing them to be used in many different biological and medicinal applications�

Like SiC, boron nitride (BN) is also a wide band gap semiconductor, allowing it to be applied in similar harsh environments� BN is shown to be compatible with comple-

mentary metal oxide semiconductor (CMOS) chips, allow-ing it to be easily incorporated into currently existing tech-nology�

Procedure: Nanowires were grown on Si <100> wafers using a low pressure CVD reactor� The Si wafer was cut in sample pieces 0�5 inch by 0�5 inch and then ultrasonically cleaned with tricholoroethylene, acetone, and methanol� Some of the samples were then coated with 5 to 10 nm of nickel (Ni), aluminum (Al), gold (Au), cobalt (Co), or iron (Fe) using an electron beam evaporator� Also, some of the wafers were sand-blasted or left untouched as control samples�

The Si samples were placed on a graphite susceptor inside a quartz inner liner tube of the CVD reactor� Control parameters included the flow rates of the various gases, the temperature within the reactor, as well as the length of each run. A scanning electron microscope (SEM), fitted with an energy dispersive x-ray spectroscopy (EDS) attachment, was then used to characterize the nanowires�

Results and Conclusions: Figure 1 shows an SEM image of the Si nanowires grown on Au catalyst� Even on the Au catalyst, there were few nanowires with a diameter of a 100 nm or less� Although some wires were grown on the other metal catalysts, these

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


wires were not at the nanoscale� The image in Figure 1 zooms in on a long Si nanowire with a diameter of around 80 nm� The EDS results confirm that the wire was made up of silicon, with it being 93�75% silicon and 6�25% oxygen� The growth parameters for this run can be seen on Table 1�

Figure 2 shows an SEM image of SiC nanowires on Ni catalyst, while Figure 3 shows an SEM image of SiC nanowires on Fe catalyst� There was an abundance of long wires with diameters of 100 nm or less on both catalysts, with Ni promoting a little more growth than Fe. The EDS results for Figure 2 confirm that the wires are SiC with 50�74% carbon and 49�26% silicon� They also confirm the wires in Figure 3 to be SiC as well, with 41.36% carbon, 52�02% silicon, and the other 6�62% being iron and oxygen� The growth parameters for this run can also be found on Table 1�

Unfortunately, the experiments on growing boron nitride nanowires did not yield any wires� The main problem may have revolved around the diborane and ammonia reaction, which required specific conditions to form boron nitride. The diborane source used was 880 parts per million, and the maximum diborane flow was restricted to 100 sccm, therefore restricting the concentration of diborane into the chamber to very dilute concentrations�

Acknowledgements:I would like to thank Dr� Gary Harris, Mr� Crawford Taylor, Mr. James Griffin, Ms. Karina Moore, and other members of the Howard Nanoscale Facility� I would also like to thank the National Science Foundation and NNIN REU Program for the funding to make this project possible�

References:[1] Lu, Wei, and Charles M� Lieber� “Semiconductor Nanowires�” Journal

of Physics D: Applied Physics 39�21 (2006)�[2] Shen, G�, Y� Bando, C� Ye, B� Liu, and D� Golberg� “Synthesis, Character-

ization and Field-emission Properties of Bamboo-like a-SiC Nanowires�” Nanotechnology, 17�14 (2006)�

[3] Sun, C�, H� Yu, L� Xu, Q� Ma, and Y� Qian� “Recent Development of the Synthesis and Engineering Applications of One-Dimensional Boron Nitride Nanomaterials�” Journal of Nanomaterials, 1-16 (2010)�

[4] Zhu, Y�, F� Xu, Q� Qin, W�Y� Fung, and W� Lu� “Mechanical Properties of Vapor-Liquid-Solid Synthesized Si Nanowires�” Nano Letters (2009)�

Table 1: Growth parameters for Si and SiC nanowires.

Figure 1: SEM of Si nanowire on Au catalyst.

Figure 2: SEM of SiC nanowire on Ni catalyst.

Figure 3: SEM of SiC nanowire on Fe catalyst.

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


Nanoscale Diamond Lenses for Atomic-Scale Sensing

Dominic E. LabanowskiElectrical Engineering, The Ohio State University, Columbus, OH

NNIN REU Site: Nanotech, University of California, Santa Barbara, CANNIN REU Principal Investigator: Prof. David Awschalom, Physics, University of California, Santa BarbaraNNIN REU Mentor: David Christle, Physics, University of California, Santa BarbaraContact: [email protected], [email protected], [email protected]

Abstract:

Nitrogen-vacancy centers in diamond have demonstrated a number of useful room-temperature properties for applications in quantum information processing and magnetic sensing. The spin state of these defects can be optically initialized, manipulated using on-chip microwave fields, and optically read-out due to a spin-dependent contrast in photoluminescence. Because of diamond’s high index of refraction, conventional optics collect only a fraction of the emitted photons. Solid immersion lenses (SIL) can drastically increase this collection efficiency, allowing for faster measurements and opportunities for new experiments such as “single shot” read-out. In this work, we discuss progress made in the fabrication of solid immersion lenses on the surface of diamond substrates using two techniques: focused ion beam milling and photoresist reflow. The shapes of the lenses made with both techniques were characterized and optimized, and their effect on increasing the collection efficiency of nitrogen-vacancy center photoluminescence was studied.

Experimental Procedure:The initial stage of the project used a focused ion beam (FIB) system to mill hemispherical lenses directly on the surface of single-crystal diamond substrates using 30 kV accelerated Ga+ ions to sputter material from the surface� Initially, a perfectly hemispherical etch pattern was programmed into the FIB� The SIL fabricated using this method was then cross-sectioned and measured, shown in Figure 1(a)� This cross-sectioning revealed non-linearities in the FIB etch, which lead to a non-hemispherical lens� SILs enhance the collection efficiency of nitrogen-vacancy (NV) center photoluminescence by centering a hemisphere on an NV center� This hemisphere allows for light leaving the NV center to exit the diamond normal to the surface at every point, lowering losses due to a mismatch in refractive indices between diamond and air (2�4 and 1, respectively)�

Thus, fabricating perfectly hemispherical SILs is critical to observing this enhancement� In order to correct for these non-linearities, MATLAB® was used to apply a first-order correction to the etch pattern� A comparison of the initial FIB etch profile and the corrected etch profile can be seen in Figure 1(b). This new etch profile resulted in substantially more hemispherical etch profiles than previously obtained, shown in the SEM micrographs in Figures 1(c) and 1(d)�

Figure 1: (a) SEM image of an uncorrected SIL cross-section. (b) Comparison of corrected and uncorrected SIL etch profiles. (c) and (d) SEM images of corrected SILs.

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


To characterize the new SILs, they were placed into a home-built confocal fluorescence microscopy setup for spatial imaging of photoluminescence� The measured surface photoluminescence around the SILs, shown in Figure 2(a), was extremely high compared to the background elsewhere, making it impossible to image the NV centers within the SILs� It is likely that this background signal was due to Ga+ ion damage from the FIB etch, as is common in FIB etched structures� Chemically processing the sample using common solvents and boiling perchloric acid showed only a small reduction in the background surface photoluminescence�

Finally the surface of the samples was etched with an inductively coupled plasma (ICP) etcher using simultaneous Ar/Cl2 gas flows to etch approximately 80 nm of diamond, removing the damaged layer entirely� This technique was effective in reducing surface photoluminescence around the SILs, as can be seen in Figure 2(b)�

Figure 4: (a) Photoresist pillars before heating. (b) Photoresist domes after heating.

Figure 2: (a) Surface photoluminescence intensity before Ar/Cl2 ICP etch. (b) Surface photoluminescence intensity after Ar/Cl2 ICP etch.

reducing the amount of averaging necessary in what are essentially shot-noise limited measurements�

The final stage of the project focused on producing SILs using standard semiconductor processing techniques� While the FIB was effective in fabricating SILs, it was rather slow and expensive, producing one SIL for every hour of FIB time. If SIL production using photoresist reflow could be perfected, it would allow for the creation of thousands of SILs in just a few hours, at a very low cost per SIL�

In photoresist reflow etching, a thick layer (7 µm) of SPR220-7 photoresist was deposited on the diamond surface and patterned into pillars, shown in Figure 4(a)� These pillars were then heated above the glass transition temperature of the photoresist in order to melt the pillars into more hemispherical shapes�

Figure 3: (a) NV center in bulk diamond. (b) NV center in SIL.

After the surface photoluminescence was reduced using the argon / chlorine etch, data were collected on the photoluminescence intensity of NV centers (both inside and outside the SILs), as a function of incident laser power� These plots can be found in Figure 3� The photoluminescence intensity increased by approximately a factor of three within the SILs, while background photoluminescence dropped by approximately a factor of two� This led to an increase in signal to noise ratio of approximately five for NV centers within SILs�

These increases, though still short of the theoretical en-hance ment of ~ 10x, allow for experiments to be done in substantially less time than was previously possible by

After several rounds of trial and error, a temperature and heating time was found that led to relatively hemispherical resist patterns, shown in Figure 4(b)� These photoresist domes were put into the ICP etcher and subjected to the same Ar/Cl2 etch�

Unfortunately, low selectivity of this etch resulted in very shallow features (on the order of 200 nm) that deviated too much from the ideal hemispherical SIL profile desired for this research�

Conclusions:We have demonstrated a method for the fabrication of SILs on the surface of diamond using a FIB� The SILs etched using this method have displayed the ability to significantly increase the photoluminescence from NV centers located within them, while reducing background�

In addition, a method was explored to produce SILs using semiconductor processing techniques, though no suitable SILs were fabricated using this method�

Acknowledgements:I would like to gratefully acknowledge my mentor David Christle and all members of the Awschalom group, as well as my principal investigator Dr� David Awschalom for their support� This work was supported by the NSF, the AFOSR, and the NNIN REU Program�

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


Local and Global Effects on the Growth of Carbon Nanotube Micropillar Arrays

Yuki MatsuokaDepartment of Future Industry-Oriented Basic Science and Materials, Toyota Technological Institute, Japan

NNIN iREG Site: Lurie Nanofabrication Facility, University of Michigan, Ann Arbor, MINNIN iREG Principal Investigator: Professor Anastasios John Hart, Mechanical Engineering, University of MichiganNNIN iREG Mentor: Sameh Tawfick, Mechanical Engineering, University of MichiganContact: [email protected], [email protected], [email protected]


The mechanism of carbon nanotubes (CNTs) isn’t unveiled in detail despite the efforts of many researchers� Therefore, non-uniform CNT films are synthesized frequently. Generally speaking, a CNT film thickness was increased or decreased around the substrate edge� Jeong, et al�, reported whole CNTs film thicknesses changed depending on the total catalyst area [1]� Here, these effects affecting whole CNTs film structures are called “global effect.” Moreover, Hart has observed partial CNT growth enhancing [2]� Here, this effect is called “local effect�” It is vital to understand these effects to control precise CNTs structure—for example, homogeneous film thickness, straight pillar with small diameter�

In this study, we investigated the effect of pattern density on the growth rate and straightness of high aspect ratio cylindrical CNT micropillars� To this end, we designed catalyst circle patterns having different diameters and spacings keeping a constant total catalyst area per substrate to study the global effect� We also designed patterns having sinusoidal spacings between circles to investigate local and non-symmetric pattern density effects�

Our experiments show that the pillars can deflect towards (attractive) or away from (repulsive) one another and these two deflection regimes can occur on the same substrate on identical patterns� We hypothesize that the non-uniform growth rates among the patterns can be responsible for the pillars deflection. A simple theoretical model was formulated to quantitatively capture the effect of non-uniform growth rate on bending of individual pillars�

Experimental:We employed photolithography methods to prepare for patterns having diameters of 10, 30 and 90 µm, and different spacings—keeping a constant total catalyst area, 0�16, 0�77 and 1�84 mm2 per substrate for global effect—and having sinusoidal spacings between circles under a total catalyst area of 0�77 mm2 for local effect� Iron (Fe) and Alumina (Al2O3) were deposited on the pattern with thicknesses of 1 and 10 nm, respectively, followed by development using sonication in acetone solution� After reduction of patterned

substrate by heating (775°C) in H2 atmosphere for 10 min, CNT micropillars were synthesized by a hot wall chemical vapor deposition (CVD) tube furnace with C2H4 precursor gas under constant temperature (775°C)� Table 1 shows the experimental conditions used in this study� The length and deflection of grown CNT micropillars were measured by scanning electron microscopy (SEM)�

Figure 1: Relationship between deflection of CNT micropillars and pillar diameter.

Table 1: Experimental conditions.

Results:Figure 1 shows the result of CNT micropillars grown from standard condition A. In Figure 1, deflection was inversely decreased with increasing pillar diameter. Deflection was

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


larger than higher pillar density� This indicates the grown pillars were supporting one another at high pillar density� To explain these results, we made a simple theoretical model based on the theory of thermal bending of bimetal [3]�

In the model, the pillar was divided into two equal parts in a longitudinal direction� Then, we hypothesized the misfit strain generated from the difference of growth rate in catalyst area, Δε = Δgt. Using the equation of misfit strain and the moment, deflection was deduced (Equation 1),

where d is pillar diameter and L is the length of the CNT micropillars� The theoretical curve was well accorded with the trend of deflection. The result of short CNT micropillars grown from condition B also showed the same trend and fitted well Equation 1.

However, one of the long CNT micropillars grown from condition C showed differently. The deflection of a diameter of 10 µm was decreased, compared to 30 µm of each total catalyst area� From a cross-sectional SEM image (not shown here), it was evident the difference was also caused by high pillar density� This indicates Equation 1 is only effective at low pillar density� We could not see the effect of total catalyst area to pillar length in every condition�

Figure 2: SEM image of the observed attractive bending effects.

In this study, two kinds of local bending effects were observed, i.e. deflection away from (Figure 2) and towards (not shown here) one another on the density-modified pattern� To explain the repulsive effect, we hypothesize a CNT-growth enhancing field around the catalyst circle.

When the space between catalyst circles is close, the fields are overlapped, and the growth rates are more enhanced�

From cross-sectional SEM images, the growth rate enhancing was verified, because about 7% of the longer pillars were grown in a dense catalyst area than a sparse area� Therefore, the CNT growth on the side near other catalyst circles increased, resulting in repulsive deflection. This repulsive effect was not influenced by gas flow direction.

The attractive effect seemed to be affected by the substrate size or the distance from the substrate edge to the patterned catalyst� However, this is still under investigation�

Conclusions:The relationship between deflections of CNT micropillars and pillar diameter was clearly identified. The simple bending theory based on the growth rate difference was in good accordance with the obtained results� It suggests the origin of bending is the growth rate difference�

When the pillar length and catalyst density increased, bending theory departed from the results, because of the pillars supporting one another�

From the density modified pattern, two types of local effects were observed, i�e� attractive and repulsive effects� The repulsive effect was explained by the CNT-growth enhancing field.

Future Works:We need to improve the theoretical model using the misfit strain generated from feasible growth rate distribution�

Acknowledgements:I would like to thank my mentor, Sameh Tawfick, and principal investigator, Dr� Anastasios John Hart, for their kind support of my experimental work, and their valuable discussions� I would also like to acknowledge the staff at the University of Michigan: Brandon Lucas, Sandrine Martin, and Trasa Burkhardt, and the Japanese staff: Keijiro Hirahara and Kayoko Tomisawa� Finally, I thank the Lurie Nanofabrication Facility and the National Nanotechnology Infrastructure Network International Research Experience for Graduates (NNIN iREG) Program�

References:[1] G� H� Jeong et al�, Carbon 47 (2009) 696�[2] A� J� Hart et al�, J� Phys� Chem� B,110 (2006) 8250�[3] Y� Clyne, Key Engineering Materials, 116 (1996) 307�

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


Devices for Investigating Electrical Transport in Topological Insulators

Joshua MendezPhysics, Louisiana State University

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Dr. Daniel C. Ralph, Physics Department, Cornell UniversityNNIN REU Mentors: Alex Mellnik and Jennifer Grab, Physics Department, Cornell UniversityContact: [email protected], [email protected], [email protected], [email protected]

Abstract:

Topological insulators are a recently discovered class of materials that have a unique set of properties. These materials are electronically insulating in the bulk, but have conducting states that exist only on their surface. Bismuth (III) telluride (Bi2Te3) and bismuth (III) selenide (Bi2Se3) are two examples of topological insulators that we investigated here. We attached leads to bismuth telluride nanostructures, grown by solvothermal synthesis, using aligned electron beam lithography. We also fabricated gated Hall bar structures from bismuth selenide thin films, which were grown using molecular beam epitaxy by collaborators.

Figure 1: The spin of the carrier is locked to its momentum. The result is a strong protection of the carrier from back-scattering.

Introduction:Topological insulators are a new class of materials that have unique properties� These materials are electronically insulating in the bulk, but have conducting states at the surface� This can better appreciated in Figure 1� The carriers have their spins locked perpendicular to their momentum� There are many examples of these materials, but only two were of our interest�

The first one was bismuth telluride (Bi2Te3), which was synthesized in the lab in the configuration of nanorods with the dimension of few micrometers long and several hundred nanometers wide� The other one was bismuth selenide (Bi2Se3), which was grown on a gallium arsenide substrate

on top of zinc selenide with molecular beam epitaxy with a thickness of 10 nm and protected by a thin layer of selenium�

Our goal was to fabricate devices that could be use to measure electronic properties of these materials�

Experimental Procedure:The fabrication process for the Bi2Te3 devices was divided in two steps. The first was the synthesis of the nanorods and the second was the fabrication itself� The nanorods were grown by solvothermal synthesis�

Figure 2: The nanorod is the long wire that is seen in the middle. The circular feature is a gold alignment mark. The source and drain are at the ends of the wire with the four leads in between.

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


In a vessel a suspension, composed of ethylene glycol, high purity polyvinylpyrrolidone (PVP), ethylenediamine tetraacetic acid (EDTA), tellurium and bismuth oxide, was mixed until the suspension was homogeneous� The vessel was sealed, heated up to 220°C, and sustained for at least four hours� The product was collected using a high speed centrifuge and washed with ethanol and water� This process was repeated until all the ethylene glycol was replaced� The now-clear suspension was cast onto a silicon wafer previously prepared with gold alignment marks� Resist was spun on top of the wafer and an electrode pattern, consisting of a gate, a source, a drain, and four additional contacts, was exposed using electron beam lithography� Following development of the resist, gold was evaporated onto the wafer surface and the resist layer was lifted off to leave the desired configuration of electrodes. An example of the device can be seen in Figure 2�

The process for fabricating a Hall bar structure on Bi2Se3 started by spinning, patterning and developing a photoresist layer� This resist protected the covered area of the Bi2Se3, while the rest of the surface was etched away down to the gallium arsenide substrate using an ion mill� After removal

of the resist, the chip was heated to 220°C to evaporate the Se layer, exposing the Bi2Se3� Silicon nitride was then deposited on the entire chip to protect the Bi2Se3 surface and was spun with photoresist, patterned and developed once more� To make electrical contacts, silicon nitride was etched way at the desired locations and gold was deposited using a liftoff process� The device cross section is shown in Figure 3� The last step is to deposit larger gold leads to connect to the contacts. The final device can be seen in Figure 4.

Results and Conclusions:The Bi2Te3 synthesis proved to be difficult, but with a well establish method, it was possible to fabricate several devices from a single cast� Initially the selenium vaporization step caused trouble for the Bi2Se3 devices, but we solved this and made many of the Bi2Se3 devices out of a single chip�

Future Work:For future work, the Bi2Te3 synthesis can be refined with the goal of having fewer contaminant particulates at the moment of casting the nanorods unto the wafer� Also the selenium layer can be thinner with the intention of having a more robust device fabrication process� The back gate and top gate can be laid and impedance measurements can be done on the devices� The goal would be to produce ultra-low power transistors using a small electric field to open a gap in the surface states�

Acknowledgments:I would like to thank the National Science Foundation and National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program for this wonderful opportunity� This would not have been possible without the Cornell NanoScale Science and Technology Facility and the staff� I would like to thank Rob Ilic and Melanie-Claire Mallison, the CNF REU Program Coordinators, which with much effort and patience they coordinated the program� My special gratitude goes to my principal investigator Dr� Daniel C� Ralph, and mentors Alex Mellnik and Jennifer Grab for their guidance and support�

References:[1] G. Brumfield; “Topological insulators: Star material”; Nature, 466,

310-311 (2010)�[2] M� Franz; “Solid-state physics: U-turns strictly prohibited”; Nature,

466, 323-324 (2010)�[3] F� Xiu et al�; “Manipulating surface states in topological insulator

nanoribbons”; Nature Nanotechnology 6, 216-221, (2011)�[4] M� Liu et al�; “Electron interaction-driven insulating ground state in

Bi2Se3 topological insulators in the two-dimensional limit”; Phys� Rev� B 83, 165440 (2011)�

Figure 4: Microscope picture of a typical device geometry. Gold leads are in yellow (light gray) and silicon nitride is the light green (medium gray).

Figure 3: The different layers of the bismuth selenide chip. The gold leads are insulated from each other by the silicon nitride.

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


DNA in Nanochannels

Francisco Pelaez, IIIChemical Engineering, University of Texas at Austin

NNIN REU Site: Nanofabrication Center, University of Minnesota-Twin Cities, Minneapolis, MNNNIN REU Principal Investigator: Dr. Kevin Dorfman, Chemical Engineering and Materials Science, University of MinnesotaNNIN REU Mentor: Daniel Olson, Chemical Engineering and Materials Science, University of MinnesotaContact: [email protected], [email protected], [email protected]

Figure 1: Illustration of Odjik and de Gennes regimes and possible other extensions as a function of channel size. Reprinted with permission from [3]. Copyright August 1, 2011, ACS.

Abstract and Introduction:Deoxyribonucleic acid (DNA) is an important part of biological studies to understand diseases and evolution� By placing DNA in nanochannels, we have a top down approach to study DNA that can lead to the fabrication of chip-based devices that can detect and separate single DNA molecules by length� By placing DNA in nanochannels, however, the Brownian dynamics is different than that of DNA in a bulk solution [1]� Further understanding of DNA in confined areas is fundamental for chip design. By making nanochannels of varying channel sizes, we can show how DNA’s extension varies with channel size and how that changes the Brownian dynamics of DNA� Figure 1 shows how DNA’s extension is affected by channels size� There are two well-described regimes in Figure 1, Odjik and de Gennes�

In Odjik’s regime, the channel size is smaller than DNA’s persistence length and behaves as a stiff chain� In de Gennes’s regime, the channel size is smaller than the radius of gyration, and DNA behaves like linked blobs� Based on

evidence from computer simulations on different channel sizes, there are two transition regimes between the Odjik and de Gennes regimes [2]� Since DNA is a large polymer, any fundamental property, we find on DNA can also be extrapolated down to smaller polymers�

Methods:The goal of this project was to fabricate nanochannels of varying widths to experimentally observe the behavior of DNA in confined spaces, and relate them to the theoretical data�

Electron beam lithography is a great way to make nanochannels of varying widths, but is very costly� By implementing optical photolithography, we can achieve the same goal at a cheaper price� Figure 2 shows how the whole process was carried out� The problem with optical p h o t o l i t h o g r a p h y though was that we could only achieve nanochannels widths of about 500 nm after reactive ion etching� We wanted to reach nanochannels of about 100 nm to 700 nm� In order to achieve that, we used plasma enhanced chemical vapor deposition (PECVD) to deposit oxide on the channels to make them thinner� By placing the channels in the PECVD for 16 minutes in three 5 min 20 s intervals, Figure 2: Process to

make nanochannels.

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


we shrank the 500 nm channels down to 150 nm, and the 750 nm channels down to 450 nm�

In order to get DNA inside the channels, reservoirs were placed in the silicon wafer� A potassium hydroxide bath etched rectangles through the silicon wafer� To complete the prototype, we used a substrate bonder to bond glass on top of the silicon wafer� The glass provided a transparent material to observe the DNA under a microscope�

To see the DNA under the microscope, we dyed the DNA with YOYO, an intercalating dye that fluoresces green when exposed to blue light� To make the DNA travel across to the channels, we used an electric field to drive the DNA toward the channels from the reservoir� This worked because we placed two electrodes and applied a voltage potential between the two� The oxide provided an insulating layer and the current could only travel through the salts of the buffer solution� Since DNA is negatively charged due to the phosphate groups along the backbone, DNA travels toward the positively charged electrode�

Results and Discussion:After making the prototype, it was really important to see if DNA would travel into the channels by the methods described above. In the first attempt, the DNA kept sticking to the oxide walls of the microchannels leading up to the nanochannels� Because of the sticking, the DNA never made it to the nanochannels� To counteract this problem, we let the prototype sit in a polymer solution of polyvinylpyrrolidone (PVP) over a time period of over 48 hours� PVP coated the oxide walls of the channels and prevented the DNA from sticking to the sidewalls�

Figure 3 shows the nanochannels under a brightfield microscope� Figure 4 shows the same channels, but under a blue light that shows the DNA fluorescing in the channels. This is significant because this shows that DNA can be placed inside the channels and with that, we can experimentally study the behavior of DNA in confined spaces.

By making different channel sizes, we can observe their extension and diffusivity coefficients based on a variety of channel size�

Acknowledgements:I would like to thank National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, National Science Foundation, Dr� Kevin Dorfman and his group, Dan Olson, Sung Gyu Park, Zhicheng Long, Jia Ou Kat Völzing, and Kathryn Johnson for making this project possible�

References:[1] W� Reisner, K�J� Morton, R�Riehn, Y�M� Wang, Z� Yu, M� Rosen,

J�C� Sturm, S�Y� Chou, E� Frey, and R�H� Austin� Statics and Dynamics of Single DNA Molecules Confined in Nanochannels. Phys� Rev� Lett� 94, 196101 (2005)�

[2] T. Odijk. Scaling Theory of DNA confined in Nanochannels and Nanoslits� Phys� Rev� Lett� 77, 060901 (2008)�

[3] Y� Wang, D�R� Tree and K�D� Dorfman, Simulating DNA Extension in Nanochannels� Macromolecules 44, 6594 6604 (2011)�

Figure 4: Channels under blue light, showing the DNA fluorescing in the channels.

Figure 3: Channels under brightfield microscope.

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


Characterization of Optoelectronic Properties of Colloidal Quantum Dots in a Nanogap

Margeaux WallaceMaterials Science and Engineering, Cornell University

NNIN iREU Site: Delft University of Technology (TU Delft), NetherlandsNNIN iREU Principal Investigator: Prof. Herre van der Zant, Molecular Electronics and Devices, TU DelftNNIN iREU Mentor: Michele Buscema, Molecular Electronics and Devices, TU DelftContact: [email protected], [email protected], [email protected]

Abstract:

This work focused on a fundamental exploration of charge transport through a self-aligned nanogap which had been deposited with colloidal cadmium selenide (CdSe) quantum dots via solution phase processing. Char acter ization of charge transport through the gap, which was only a few nanometers wide, centered around measurements made at room and cryogenic temp eratures, and the response of the device as a result of interaction with light.

Figure 1: (a) No bias energy level structure of the device. (b) Applied bias is non zero. The source is now within tunneling distance of the unoccupied level in the island. (c) An exciton. Probabilities the charge carriers will tunnel in each direction are represented with the size of the arrows. (d) Light with applied bias. The probabilities of tunneling change and make it more likely excited electrons will contribute to the current.

Introduction:Single electron transistors are made up of a quantum dot (or island in the following text) that sits between two metallic electrodes (source and drain)� In the case under investigation, the island was weakly coupled to the two leads meaning that the level of the island did not mix with the bands of the leads and, therefore, they had a definite energy and degeneracy. This caused incoherent transport, the electron not retaining its phase as it moved between the different elements� The electrons would only be able to tunnel to the island when the bias was high enough that accessible island energy levels were within range (conduction)� This was ultimately decided by the probabilities of the charge carriers to tunnel between either source or drain [1]�

Charge transport was also affected by light interaction� When shining light on the devices, excitons were created and could separate, the electron and hole tunneling in either direction� But without the applied bias, there would be no net conduction through the devices as the probabilities that determined the direction of charge carrier tunneling would all be equal� However, under biased conditions, tunneling would shift in one direction and conduction would increase [1]�

Methods:Planar electrodes of different widths were fabricated using a self-alignment fabrication scheme [2]� The nanoparticles were deposited by dip-coating with ligand exchange� The sample was pretreated with short ligands and dipped into

the solution where nanoparticles attached and long ligands exchanged with short ligands� This closer contact allowed for better coupling of the particles to the leads� The process was completed by dipping the chip into clean solvent to remove excess particles�

The devices were characterized empty and post deposition in a vacuum probe station� Post deposition the devices were measured in the dark and then exposed to a 545 nm laser� Analysis was done to find the best devices, which were tested at a single voltage with a chopper at varied frequencies�

In order to eradicate effects due to thermal broadening, the current-voltage responses of the devices were characterized at low temperature� The probe station was cooled down to about 10°K using liquid helium�

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


Luminescence experiments were done using a CCD camera in a fully dark probe station� Six volts were applied to the device while the shutter of the camera was open to absorb any light emissions from the particles�

Results:Figures 2 and 3 show the best results from the experiments with CdSe particles, showing a marked difference when the device interacted with light� The total average difference in the light response compared to the dark can be seen in the chopper graphs� For Device 1 at 5V, the average difference was about 0�015 nA and for Device 2, it was 0�014 nA at 4V� At higher voltages there was more of a response� The chopper measurements for Device 1 were taken at room temperature and Device 2 at 10°K�

The photon to electron efficiency was calculated at 4V using the definitions in equations 0.1 and 0�2 to determine the relation in 0.3. At 10°K the efficiency of devices one and two were 2�17E-8 and 1�97E-7, respectively�

The response of the devices was similar to the theory developed based on rate equations, but it is still under investigation and thus there are some discrepancies�

Luminescence experiments were performed on all working devices, but only one device generated light, seen in Figure

Figure 4: False-color picture of the luminescence of the device [in greyscale]. The circled area is the only light not due to noise of the camera. (See cover for full color version.)

Figure 2: (a) Empty, dark and light IV curves from +/-6V at room temperature. Empty means before particle deposition while dark is after deposition without any light interaction. (b) Light and dark curves from +/-6V at 10K. (c) Chopper measurements at 5V room temperature.

Figure 3: (a) Empty, dark and light IV curves from +/-4V at room temperature where empty is pre-deposition and dark is post-deposition without any light interaction. (b) Light and dark curves from +/-6V at 10K. (c) Chopper measurements of intensity at 5V.

4� The false color red indicates areas where light was emitted� Unfortunately, this device broke before the required further tests; this also suggests more deeply investigation into the device stability at high biases�

Conclusions:The addition of CdSe particles in this manner results in increased conductance of the devices tested� When the interaction with light is included, the conductance is increased even more, due to the intrinsic nature of the particles� More work should still be done regarding the particles to have a fuller picture of the level structure, varying power, wavelength polarization and gate voltage�

Acknowledgements:Thank you to everyone at TU Delft who helped to make my summer a success, especially Michele Buscema, Herre van der Zant, Ferry Prins, Hans Mooij, and Lynn Rathbun� Also, thanks to the National Nanotechnology Infrastructure Network International Research Experience for Undergraduates (NNIN iREU) Program and the National Science Foundation for funding�

References:[1] F� Prins, M� Monrabal-Capilla, E� O� Osorio, E� Coronado and H� S�

J� van der Zant, Adv� Mater�, 2011, 23, 1545�[2] Thijssen, J� M�; van der Zant, H� S� J� phys� stat� sol� (b) 2008�

• p

hysic

s &

nanostr

uct

ure

physic

s •

physic

s &

nanostr

uct

ure

physic

s

• p

hysic

s &

nanostr

uct

ure

physic

s

•


Microfabricated Silicon Carbide Thermionic Energy Converters for Solar Energy Generation

Leah WeissPhysics, Harvard University

NNIN REU Site: Stanford Nanofabrication Facility, Stanford University, Stanford, CANNIN REU Principal Investigator: Prof. Roger Howe, Electrical Engineering, Stanford UniversityNNIN REU Mentors: Dr. Igor Bargatin and Jae Lee, Electrical Engineering, Stanford UniversityContact: [email protected], [email protected], [email protected], [email protected]

Abstract:

Microfabricated thermionic energy converters (TECs) could be a crucial candidate for concentrated solar thermal power plants. TECs convert heat directly to electricity. They can therefore be thought of as heat engines in which the working fluid is electrons themselves. Electrons are “boiled” off the hot cathode and then condensed and absorbed at the cooler anode: the temperature difference produces a voltage between the cathode and anode. This study explored the fabrication of silicon carbide (SiC) cathodes (emitters) with integrated SiC anodes, i.e., complete converter structures. The fabrication process involved wafer bonding, reactive ion etching to pattern the structures, and vapor HF releasing to suspend the cathodes (emitters). We fabricated the devices with a 1.7 µm gap between the cathode and anode, with a yield of over 50%. Using resistive heating, we measured the thermionic current. Later, we will use laser heating to demonstrate the first microfabricated TECs and measure the conversion efficiency to explore future use in clean energy production.

had only 10-15% efficiencies [1]. Yet, microfabricated TECs are potentially a more cost-effective and efficient means of producing clean energy as a component of concentrated solar thermal power generators� Microfabricated TECs have the potential for 30-40% efficiencies [1].

One way of increasing efficiency is to reduce the size of the vacuum gap to diminish space charge� In practice, however, it is challenging to create a structure stable and effective under the necessary temperatures� Potential problems can arise due to the stress of thermal expansion� Radiative heat transfer from the cathode to the anode can also decrease efficiencies.

Figure 1: A simplified diagram of a TEC.

Introduction:Thermionic energy converters (TECs) have three main components: cathode (emitter), vacuum gap, and anode (collector). Figure 1 shows a simplified diagram of a TEC. The cathode is typically at a temperature greater than 1500°K and the anode is typically less than 1000°K� In the hot cathode, electrons with energy greater than the work function escape into vacuum and are collected at the cooler anode [1]� In the past, TECs developed on the macro-scale

Figure 2: TEC structure with a SiC cathode center pad surrounded by legs that connect to contact pads.

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

• physics

& n

anostru

cture

physics

•


Results and Conclusions:Figure 4 is a graph of the anode current as the anode voltage was varied: first no current flowed until ~ 8V, where-upon current began to increase with increased anode voltage� The current continued to increase as the voltage was increased and did not reach the expected plateau� This enhancement was because the device was biased, and so the electric field produced field-enhanced thermionic emission called “Schottky emission” in which the electric field lowered the energy barrier for electrons to escape into vacuum� Although the devices were effective, there was vertical bowing of the cathode under thermal stress, and so the true gap size is uncertain�

Future Work:Future experiments will include testing with laser heating instead of resistive heating in order to obtain a conversion efficiency without biasing the device. To overcome vertical buckling, future structures will be fabricated using a vertical wall pattern to enhance stability under high temperatures� Methods of producing thicker cathodes will further stabilize the structure� Continued fabrication efforts will be focused not only on SiC, but also on different materials� Finally, in the future devices will be fabricated to include vacuum encapsulation in a quartz or sapphire layer to facilitate mass production and use in energy generation�

Acknowledgments:I would like to acknowledge the guidance and support of my mentors Jae Lee and Dr� Igor Bargatin, my P�I� Prof� Roger Howe, SNF site coordinator Michael Deal, Maureen Baran, and the SNF staff� I appreciate the support of my fellow REUs� I would further like to thank the NNIN REU Program, NSF, and CIS for funding�

References: [1] National Research Council Committee on Thermionic Research and

Technology� Thermionics Quo Vadis? An Assessment of the DTRA’s Advanced Thermionics Research and Development Program (National Academy 2001)�

[2] J�-H� Lee et al�, “Effect of illumination on thermionic emission of microfabricated silicon carbide structures,” 16th Int� Conf� on Solid-State Sensors, Actuators, and Microsystems (Transducers ‘11), Beijing, China, June 5-9, 2011, pp� 2658-2661�

Here we report the fabrication and detection of thermionic current from a micron-scale TEC with an integrated silicon carbide cathode and anode� The TEC structure is a SiC cathode center pad surrounded by legs that connect to contact pads and are suspended 1�7 µm above a SiC layer that forms the anode (see Figure 2)�

This study used n-type SiC and so will enable comparison with future p-type TECs� Test structures were varied: legs ranged from 25-100 µm and center pads ranged from 500 to 800 µm� Legs serve to thermally isolate the center pad and provide stability under thermal stress� Etch holes in the center pad allow vapor HF release of the cathode�

Experimental Procedure:The fabrication process began with deposition of the following layers on a Si substrate: thermally grown oxide (1�7 µm), SiC (2 µm), low temperature oxide (LTO, 1�7 µm), SiC (2 µm), LTO (1�8 µm), and photoresist (1�7 µm)�

First, we patterned the photoresist and LTO layers with reactive ion etching (RIE)� After removing the photoresist to reduce surface roughness, the first SiC layer (cathode) was etched using RIE� The LTO layer was then patterned using a BOE 6:1 isotropic etching� Gold contact pads were created by first spinning photoresist, patterning photoresist, evaporating gold, and using an acetone lift-off to leave gold contacts to the cathode and anode� Finally, the cathode was released from the anode using vapor HF etching. The final yield of fully released structures was over 50%� The structure of the final devices included an ~ 1.5 µm SiC center pad and legs suspended above a SiC anode layer connected to the silicon substrate via a thermally grown oxide isolation layer�

Figure 4: Graph of the anode current as the anode voltage was varied.

Figure 3: Vacuum chamber diagram.

Testing of devices was done in a vacuum chamber (see Figure 3 for vacuum chamber diagram [2])� The cathode was biased on one side to -16V and connected to ground at the opposite side, leaving the center pad approximately 8 V� The anode voltage was then varied from -20V to16V�

• p

roce

ss &

chara

cteri

zati

on

• p

roce

ss &

chara

cteri

zati

on •

pro

cess &

chara

cteri

zati

on •


Characterization of YBCO Superconducting Thin-Films for Fluxonic Applications

Brian T. ChungEngineering Physics, University of Michigan – Ann Arbor

NNIN iREU Site: Institut Für Bio- Und Nanosysteme (IBN), Forschungszentrum, Jülich, GermanyNNIN iREU Principal Investigator: Prof. Dr. Roger Wördenweber, Peter Grünberg Institut (PGI-8), Forschungszentrum JülichNNIN iREU Mentors: Dr. Eugen Hollmann, Peter Grünberg Institut (PGI-8), Forschungszentrum Jülich;

Mr. Rolf Kutzner, Peter Grünberg Institut (PGI-8), Forschungszentrum JülichContact: [email protected], [email protected], [email protected], [email protected]

Abstract:

The use of superconducting materials for thin film oxides has recently sparked interest in a new field of engineering research called “Fluxonics,” named after the physical quantization of magnetic flux lines that occurs in such materials when a threshold current has been injected. Structures consisting of patterned indentations, termed “antidots,” have been fabricated using standard ultraviolet (UV) lithography methods on high-temperature Type II superconductors such as yttrium-barium-copper-oxide (YBCO) to structurally guide and support magnetic flux vortices. Various growing and processing conditions concerning film qualities, topological constructions, and antidot geometry arrangements have been investigated for their effectiveness in meeting the criteria for several electronic and computing applications, one of which is the development of a super-fast method for storing and processing data [1].

The focus of this project was to characterize critical parameters of a superconducting thin film to be used for antidot patterning. Novel samples of YBCO with a cerium oxide (CeO2) buffer layer were previously grown on an R-cut sapphire substrate. For characterization, a liquid helium cryostat and an inductive coil system with a built-in heater were implemented. Automatic testing algorithms were written and administered through LabVIEW. Extracted critical temperature and voltage values were indicative of the sample’s successful onset of superconductive behavior.

Figure 1: SEM image of the surface topology for a completed antidot structure (2010).

layer on an appropriately buffered substrate, are useful for their potential in understanding vortex dynamics� It has recently been determined that for microwave experiments, YBCO may be optimally grown on a suitable substrate such as aluminum oxide (Al2O3, sapphire) with an introduced CeO2 buffer layer, which acts as both a lattice-matched interface between dissimilar unit cells and an inhibitor to the possible diffusion of aluminum [2]� Figure 1 depicts a scanning electron microscopy (SEM) image of the surface topology of a fully completed antidot structure�

Prior to its utilization as a finished structure, a sample wafer must first be characterized to ensure its functionality as a superconductive material [3]� In this experiment, we have successfully characterized the critical temperature and voltage of a previously grown thin-film sample (R2Z5-47�c), consisting of 100 nm of YBCO grown on a 30 nm thick CeO2 buffer layer on an R-cut sapphire substrate�

Experimental Procedure:For characterization, a sample of R2Z5-47�c was structurally supported between two inductive coil pads and suspended

Introduction:Recent developments in the field of complex oxide thin films have led to the study of magnetic flux vortex movements within patterned indented structures� These antidot structures, comprised of a Type II superconducting

• pro

cess &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n •


within a pressure sealed cryostat chamber. A constant flow of liquid helium and a built-in heater were then conjunctively introduced to maintain a controlled temperature, while a lock-in amplifier was employed for the generation and recording of input and output voltages�

An automatic algorithm written in LabVIEW was used to administer the experiment and record the critical temperature of the sample without heater incorporation; modifications were later made to the algorithm to account for the systematic alteration of an enclosed heater and to measure the critical voltage of the sample� Figure 2 depicts a sample screenshot of the updated program’s block diagram from LabVIEW’s graphic user interface�

To characterize critical temperature, a constant input voltage was injected while temperature was slowly decreased from 105°K to 35°K� Induced output voltages measured from the other coil were recorded� For critical voltage, temperatures were steadily maintained at 85, 86, and 87°K, while input voltage was swept from 0 to 2�5 volts� Induced output voltages were similarly measured�

Results:Figures 3 and 4 depict the results of the characterization tests for critical temperature and voltage respectively� The critical temp-erature is seen to fall within the onset-offset range of 85 to 95°K, with the 50% critical temperature value falling at 90�8°K� The critical voltage graphs similarly delineate the sample’s increase in nonlinear response with increasing temperature, indicating the existence of a critical current and the onset of superconductive behavior�

Possible explanations for the measurement fluctuations seen in Figures 3 and 4 include inaccuracies associated with the cryostat ap-paratus and failures regarding sample orient-ation and stability within the sample holder�

Conclusions and Future Work:Critical temperature and voltage values of a YBCO thin-film were successfully characterized, and the results obtained are indicative of the sample’s successful superconducting behavior� Further charac-terization of the sample may include measuring the sample’s field dependence and complex susceptibility with varying degrees of applied external magnetic fields. In addition, improvements regarding the sensitivity of the equipment, including the coil and stability of the sample holder, may be further improved�

Acknowledgments:Thank you to the National Nanotechnology Infrastructure Network International Research Experience for Under-graduates (NNIN iREU) Program and the National Science Foundation (NSF) for funding this wonderful opportunity to travel abroad and conduct research� Special thanks to Prof� Dr� Roger Wördenweber, Dr� Eugen Hollmann, and Mr� Rolf Kutzner at PGI-8 for their excellent support and instruction in this project�

References:[1] R� Wördenweber, T� Grellmann, K� Greben, J� Schubert, R� Kutzner,

and E. Hollmann, “Stress generated modifications of structural, morphologic and ferroelectric properties of epitaxial SrTiO3 films on sapphire,” EMF 2011, June 2011�

[2] V� Moshchalkov, R� Wördenweber, W� Lang, Nanoscience and Engineering in Superconductivity, Springer, pp�46, 2010�

[3] P� Bhattacharya, R� Fornari, and H� Kamimura, (eds�), Comprehensive Semiconductor Science and Technology, Volume 4, pp� 177-205, Amsterdam: Elsevier, 2011�

Figure 3, left: Critical temperature measurements of R2Z5-47.c.

Figure 4, right: Critical voltage measurements of R2Z5-47.c.

Figure 2: Sample screenshot of LabVIEW program used for characterization.

• p

roce

ss &

chara

cteri

zati

on

• p

roce

ss &

chara

cteri

zati

on •

pro

cess &

chara

cteri

zati

on •


Engineering Multifunctional Nanoparticles with Dual Modality Imaging Capabilities

Courtney CrouchBiology, Emory University

NNIN REU Site: Center for Nanotechnology, University of Washington, Seattle, WANNIN REU Principal Investigator: Dr. Xiaohu Gao, Department of Bioengineering, University of WashingtonNNIN REU Mentor: Dr. Xiaoge Hu, Department of Bioengineering, University of WashingtonContact: [email protected], [email protected], [email protected]

Abstract:

Multifunctional nanoparticles with dual modality imaging capabilities offer the ability to deliver multiple optical imaging agents to targeted locations for use with various instruments. Here we describe the size distribution, NIR dye and MNP incorporation rate, and stability of nanoparticles formed with microemulsions using various concentrations of their respective components.

Introduction:Multifunctional nanoparticles may be both synthesized and targeted to specific locations by a wide variety of processes [1, 2, 3] using a broad array of materials� Microemulsions, which consist of a thermodynamically stable emulsion of water and oil that forms a transparent dispersion in the presence of an amphiphilic surfactant [4, 5], have garnered particular interest as drug delivery systems due to their capacity to hold both hydrophilic and hydrophobic drugs, allowing simultaneous delivery of multiple drugs with little risk of causing an immune response [4, 6]� Typical imaging agents include near infra-red (NIR) dyes, whose emission and absorption spectra are not subject to background fluor-escence overlap with tissues, and magnetic nano particles (MNP) used for imaging within magnetic fields [2, 3].

Within this study, we characterized several formulations used to engineer a dual modality nanoparticle carrier system encapsulating 10 nm MNP and the widely used NIR dye indocyanine green (ICG) [7]� The size of the nanoparticles and dye incorporation were primary areas of interest in examining each method�

Methods:Nanoparticle Synthesis. Oil in water (O/W) micro-emulsions encapsulating ICG and MNP were prepared with varying concentrations of ICG, DOTAP, and purified MNP in 0�75 ml poly(sterene-b-arylic acid) (PSPAA) dissolved in 1 ml CHCl3) with 0�25 ml surfactant (either polyvinyl alcohol -PVA, bovine albumin [BSA], or poly(maleic anhydride- alt-1-octadecane) [PMAO]). PVA was modified for targeting using a carbodiimide EDC crosslinking reaction� Emulsions were formed by sonication for two minutes and evaporated while stirring overnight� Excess ICG and MNP were purified from the nanoparticles via three rounds of centrifugation (dispersed in H2O, 13K, 20 min)�

Determining Maximum ICG Encapsulation. Purified particles were measured via bright field transmission electron microscopy (CM100 TEM)� ICG encapsulation rates were determined using a Fluoromax4 fluorometer to compare ICG concentration in purified particles with the initial ICG in the preparation�

Figure 1: Nanoparticles formed with a) 1% PVA, b) 2% BSA, and c) 1% PMAO.

• pro

cess &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n •


Results and Discussion:We expected that ICG encapsulation rates would increase with greater concentrations of polymer (PS-PAA), lipid (DOTAP, N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethyl ammonium methyl-sulfate), and ICG, and lower concentrations of surfactant (PVA, BSA, PMAO) and MNP� ICG has a dual hydrophobic and hydrophilic nature, causing it to preferentially partition into the aqueous phase, while the relatively hydrophobic MNPs might further push ICG out of the organic phase� Conversely, the addition of a lipid (DOTAP) might help to pull it into the organic phase� Likewise, it has previously been suggested that the fixed capacity of nanoparticles causes the concentration of polymer to play a role in determining dye loading [8]�

In fact, only a small negative correlation between high MNP presence and ICG encapsulation was observed, although TEM images show that virtually all MNPs were taken into the organic phase (Figure 1)� There was no certain relationship between surfactant concentration and ICG encapsulation rate� Adding increasing amounts of ICG increased the encapsulation rate up to 2 mg, at which point adding more ICG decreased fluorescence intensity (Figure 2)� These results are consistent with earlier studies observing the self-quenching properties of ICG [7]� It was generally observed that higher percentage concentrations of PS-PAA resulted in greater ICG encapsulation efficiency, most likely since higher amounts of carrier polymer resulted in formation of more nanoparticles�

Although we expected that increasing the concentration of surfactant would result in small and more uniform particle size, we observed little relationship between surfactant concentration and particle size� Increased DOTAP concentrations resulted in higher dye loading, but combined with low surfactant concentrations, might have contributed to nanoparticle aggregation� Aggregation was also observed in particles formed using surface modified PVA, indicating that BSA, which formed relatively stable nanoparticles with a high ICG encapsulation rate and a 200-400 nm size distribution, might be better suited to future research with targeting molecules�

Future Work:Future work could include the application of ICG / MNP encapsulated microemulsions to targeting studies in live cell lines�

Acknowledgements:Special thanks to Dr� Xiaohu Gao and Dr� Xiaoge Hu for their guidance and support� This material is based upon work supported by the National Science Foundation (NSF) and the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program� Part of this work was conducted at the University of Washington NanoTech User Facility, and facilities and materials were provided by the University of Washington Department of Bioengineering�

References:[1] Kim J� S�, Rieter W� J�, Taylor K� M� L�, An H�, Lin W�, Lin W�

2007� J� AM� Chem� Soci� 129: 8962-8963�[2] Kim J�, Lee J� E�, Lee S� H�, Yu J� H�, Lee J� H�, Park T� G�, Hyeon

T� 2008� Adv� Mater� 20: 478-483�[3] Raut S� L�, Kirthivasan B�, Bommana M� M�, Squillante E�, Sadoqi

M� 2010� Nanotechnology� 21: 10pp�

[4] Bagwe R� P�, Kanicky J� R�, Palla B� J�, Patanjali P� K�, Shah D� O� 2001� Therapeutic Drug Carrier Systems� 18(1): 77-140�

[5] Kahlweit M� 1988� Science� 240(4852): 617-621�[6] Lee P� J�, Langer R�, Shastri V� P� 2003� Pharmaceutical Research�

20(2): 264-269�[7] Maarek J� I�, Holschneider D� P�, Harimoto J� 2001� Journal of

photochemistry and Photobiology B: Biology� 65: 157-164�[8] Saxena V�, Sadoqi M�, Shao J� 2004� International Journal of

Pharmaceutics� 278, 293-301�

Figure 2: Standard was 1 mg MNP, 0.20 ml 6% PS-PAA, 0.70 ml 2% PVA, and 0.5 mg ICG. D = DOTAP (1 mg unless otherwise stated).

• p

roce

ss &

chara

cteri

zati

on

• p

roce

ss &

chara

cteri

zati

on •

pro

cess &

chara

cteri

zati

on •


Electrodeposition of Metals onto Aligned Carbon Nanotube Microstructures

Matthew DiasioPhysics, Rice University

NNIN REU Site: Lurie Nanofabrication Facility, University of Michigan, Ann Arbor, MINNIN REU Principal Investigator: Prof. Anastasios John Hart, Mechanical Engr., and Art and Design, University of MichiganNNIN REU Mentor: Davor Copic, Mechanical Engineering, University of MichiganContact: [email protected], [email protected], [email protected]

Abstract and Introduction:Nanocomposites offer a way to create materials with specific properties by ordering components on the nanometer scale� Carbon nanotubes (CNTs) are a particularly attractive choice for composites due to their small diameter and outstanding mechanical and electrical properties� Previous studies on CNT-metal composites codeposited the components, resulting in materials with a low concentration of randomly oriented CNTs�

We investigated electrodeposition directly onto vertically aligned CNTs as a means to realize nickel-CNT and copper-CNT composites, and attempted to optimize the procedure to produce uniform composite microstructures� The aligned nature of our structures could greatly enhance the composite’s properties compared to randomly oriented CNT composites� Electrodeposition was performed at current densities ranging from approximately 2�5 to 500 mA/cm2, with deposition occurring consistently at densities above 25 mA/cm2 and composites becoming uniform above 100 mA/cm2� Morphologies ranged from isolated nanoparticles on CNTs at low current density to thick coatings at high current densities. Infiltration of metal to the interior of the CNT forest was observed under some conditions, and is still under investigation� Future studies will characterize properties of the nickel (Ni) composites for mechanical applications and the copper (Cu) composites for electrical and thermal applications�

Figure 1: a) Wafer preparation. b) Crowding of CNTs results in c) organization and d) vertical growth until e) termination.

Methods:The CNT structures were produced by thermal chemical vapor deposition on iron catalyst over an aluminum oxide layer (Al2O3)� Figure 1 shows the process of growing CNTs� Unpatterned substrates resulted in thin, aligned CNT films called “forests.” Patterned CNT structures would be produced by photolithographic patterning of the aluminum oxide and iron layers. Patterned structures would be densified by condensation and evaporation of acetone before metal deposition [2]� Some CNTs were plasma etched in argon before deposition� Electrochemical deposition of Ni and Cu occurred in proprietary electroplating solutions�

Results:Ni-CNT composites were studied more extensively due to greater ability to control the deposition� Ni nanoparticles were found on CNTs at deposition current densities as low as 12�5 mA/cm2, but consistent coating was only noticed at 25 mA/cm2 or greater� Below 100 mA/cm2, Ni nanoparticles could be found regularly on the surface of non-etched CNT structures, but the deposition was not a uniform thin film coating, as seen in Figure 2a�

The density of Ni nanoparticles was greater near the top of a bundle of CNTs than the bottom on non-etched structures� This is believed to be due to plating on the unorganized

“crust” of horizontal CNTs at the top that remains from the crowding phase of growth� At equivalent low current densities and deposition run times, plasma etched CNTs showed less deposition or as Figure 2b shows, none at all, compared to non-etched structures� At current densities of 100 mA/cm2 and greater, the external surfaces of etched and non-etched structures varied less� However, nickel was only found to penetrate

• pro

cess &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n •


the exterior surface of the structure and coat internal sur-faces on the plasma etched CNTs� Running a 500 mA/cm2 deposition current for 40 minutes resulted in a uniform Ni film on the CNTs, shown in Figure 3a, and also coated the interior, shown in Figure 3b� This interior nickel coating is less uniform than that on the exterior�

Copper was able to successfully plate the CNT structures, but yielded different morphologies� On non-etched “forest” samples, copper only deposited on the top of the structure, as shown in Figure 4a� No nanoparticles were observed on the side of the structures or the interior� On an etched sample, non-uniform depositions of Cu nanoparticles could be found on the side of CNT structures, as seen in Figure 4b, but not in the interior�

The growth of patterned structures did not occur as the use of a titanium nitride thin film instead of a silicon dioxide thin film underneath the aluminum oxide resulted in horizontal CNT carpets�

Conclusions and Future Work:Our work has shown that electrodeposition is a feasible method of producing uniform metal CNT composites� Plasma etched CNTs were plated more uniformly and showed penetration of the plating metal, though only at higher current densities and deposition times� We suspect the uniformity may be related to the etching process destroying the top crust and also creating defect sites over the entirety of the surface, making it easier for ions to deposit uniformly over the surface� Similarly, the higher required current and deposition time might be a result of the defects and attachment of hydroxide groups from etching reducing the conductivity of the CNTs� The lack of interior Cu deposition might be from the Cu plating on the exterior occurring so fast as to prevent interior penetration�

The Cu electrodeposition needs further refinement. Future work will also look at depositing both metals onto patterned structures� Of particular interest is how metal penetration might be affected by the size of the structures� Characterization of the composites is also necessary to determine appropriate uses of these materials� The Ni-CNT composites are theorized to be useful in mechanical applications and the Cu-CNT composites could be useful for electronic or thermal applications�

Acknowledgements:This project was supported by the National Science Foundation (NSF) and the National Nanotechnology Infrastructure Network

Research Experience for Undergraduates (NNIN REU) Program� Guidance from Davor Copic, Dr� John Hart, and other members of the Mechanosynthesis Group of the University of Michigan was invaluable in the completion of this project�

References:[1] Y� Sun, et al� “Mechanical strength of carbon nanotube–nickel

nanocomposites�” Nanotechnology (2007)�[2] M� De Volder, et al� “Diverse 3D Microarchitectures Made by

Capillary Forming of Carbon Nanotubes�” Advanced Materials (2010): 4384-4389�

Figure 4: Cu plating of a) non-etched and b) etched structures.

Figure 2: a) Non-etched and b) etched CNTs show different plating at low current density.

Figure 3: a) Exterior of etched CNT after 40 min deposition shows uniform coating. Arrows point out regular ridges, suggesting conformal coating. b) Interior of same structure.

• p

roce

ss &

chara

cteri

zati

on

• p

roce

ss &

chara

cteri

zati

on •

pro

cess &

chara

cteri

zati

on •


Morphological Characterizations of Collagen-Modified Alumina Membranes

Tiffany DunstonDepartment of Chemistry, Syracuse University

NNIN REU Site: Howard Nanoscale Science and Engineering Facility, Howard University, Washington, DCNNIN REU Principal Investigator: Dr. Kimberly L. Jones, Civil and Environmental Engineering, Howard UniversityNNIN REU Mentor: Dr. Malaisamy Ramamoorthy, Civil and Environmental Engineering, Howard UniversityContact: [email protected], [email protected], [email protected]

Introduction:

Membranes can be engineered to separate materials based on size, concentration, charge, temperature or other characteristic, which makes them versatile enough to be employed in multiple industries such as environmental and chemical engineering, waste-water applications, food processing, biotechnology, desalination, and medical� In biopharmaceutical and biomedical applications, membrane techniques replace other conventional separation methods such as ELISA, western blotting, and polymerase chain reaction (PCR)� Recently the protein collagen has been identified for coatings, films and membranes and found to be controllable in growth and morphology [1]� Those surface modified membranes need to be well characterized in order to be applied for a specific application [2].

The molecular weight cut off (MWCO) of a membrane gives insight about the pore size, the function of the membrane, selectivity, and its ability to separate a particular molecule� The aim of this project was to evaluate large pore modified alumina membranes coated with layers of collagen with the goal of developing a membrane with 90% solute rejection for biomolecules of different size fractions� The pore size of the modified membranes was determined by measuring the MWCO, which is the smallest possible molecular weight of a solute, in this case dextran, with at least 90% rejection [3]� Surface morphology was examined by scanning electron microscopy (SEM) imaging to further confirm the magnitude of modification in addition to its effect on membrane pore size�

Experimental Procedure:The membranes used for the study were 20 nm pore size anodized alumina membranes etched with concentrated sulfuric acid, and spin-coated with 1, 3, 6 and 9 layers of dermal collagen�

Morphology. Freeze-fractured modified membranes were coated with a thin layer (5 nm) of copper metal and scanned under a JEOL SEM at an applied voltage of 5 kV� The membrane surfaces were viewed with a magnification of 30K-50K times and the images captured�

MWCO. Dextran (a polysaccharide) with molecular weights such as 25, 50, 80, 150, 270 kDa, was prepared at a concentration of 20 ppmC as the feed for the filtration experiments in an aqueous medium� The membrane was loaded in a stirred batch filtration cell and connected to a feed tank that was pressurized with ultrapure nitrogen gas at 10 psi� Filtration of each of the molecular weight dextran for each modified membrane (1, 3, 6 and 9 layered) was carried out and the permeate sample collected� An UV-Persulfate TOC analyzer (Teledyne-Tekhmar, Phoenix 8000) equipped with a NDIR detector was used to determine the unknown concentration of the dextran in the permeate in terms of the total organic carbon content [4]� Initial standard calibration standards were run for each molecular weight dextran� From the concentration values obtained from the analysis, the percentage rejection of each dextran for each modified membrane was calculated using the formula, % rejection = (1 – (Cp/Cf)*100, where Cp is the concentration of the permeate and Cf the concentration of feed�

Results and Discussion:Figure 1 confirms that the 9-layer spun-collagen was in the form of long extended fibrils completely covering the

Figure 1: Nine-layer collagen modified membrane.

• pro

cess &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n •


surface of the porous sulfuric-acid-etched alumina (refer to insert)� Small monomeric collagen molecules were seen on three layers, and a mixture of monomers and short fibrils on six layers were observed (images not shown)�

Figure 2 shows the percent rejections of various dextran by various modified-membranes� Generally, for each collagen-layered membrane, a trend of increasing percent rejections with increasing molecular weight of dextran was observed� For an unmodified membrane, the rejection for even the highest molecular weight (270 kDa) was negligibly small� However with three layers, the rejection significantly increased for all the dextrans, with 270 kDa nearing 90%� With six layers, except for the 25 kDa, all the other dextrans reached the 90% mark, where the MWCO of a membrane was determined� With further increase to nine layers, the rejection of 25 kDa also reached 90%� From the rejection values, the MWCO of the membranes were found to be > 270, 270, 50-80 and 25-50 kDa respectively for 0, 3, 6 and 9 layer modified membranes� Knowing the MWCO of these membranes, molecules of a given size can be specifically removed from a mixture of other molecules of varying size fraction�

In future work, these membranes can be potentially applied for separating biomolecules such as DNA, Heme or proteins of a given size�

Conclusion:Increasing the number of collagen layers significantly decreased the molecular weight cut-off of the membranes� The number of layers can be tailored for a specific separation� Reproducibility of the MWCO of the membrane confirms the stability of the collagen layers on the surface.

Figure 2: Dextran rejections of each membrane.

References:[1] H� R� Baker, E� F� Merschrods, and K� M� Poduska,

Electrochemically controlled growth and positioning of suspended collagen membranes, Langmuir 24 (2008) 2970-2972�

[2] S�-I� Nakao, Determination of pore size and pore size distribution: 3� Filtration membranes; Journal of Membrane Science 96(1-2) 1994 131-165�

[3] S� Mochizuki and A�L� Zydney, Dextran transport through asymmetric ultrafiltration membranes: Comparison with hydrodynamic models; Journal of Membrane Science 68(1-2) (1992) 21-41�

[4] J� S� Taurozzi, H� Arul, V�Z� Bosak, A�F� Burban, T�C� Voice, M�L� Bruening and Volodymyr V. Tarabara. Effect of filler incorporation route on the properties of polysulfone–silver nanocomposite membranes of different porosities; Journal of Membrane Science 325 (1), (2008), 58-68�

• p

roce

ss &

chara

cteri

zati

on

• p

roce

ss &

chara

cteri

zati

on •

pro

cess &

chara

cteri

zati

on •


Characterization of AIN Thin Films for Applications in Bulk Acoustic Filters

Lisa Anne HendricksElectrical Engineering, Rice University

NNIN REU Site: Lurie Nanofabrication Facility, University of Michigan, Ann Arbor, MINNIN REU Principal Investigator: Mina Rais-Zadeh, Electrical and Computer Engineering, University of Michigan, Ann ArborNNIN REU Mentor: Vikram Thakar, Mechanical Engineering, University of Michigan, Ann ArborContact: [email protected], [email protected], [email protected]

Abstract:

This paper describes the characterization of aluminum nitride (AlN) thin films, deposited using reactive magnetron sputtering, for applications in bulk acoustic filters. Argon (Ar) and nitrous oxide (N2, reactive gas) flow and direct current (DC) bias were controlled to optimize film properties. Increasing gas flow rates resulted in reduced compressive stress. Increasing the Ar flow improved crystal orientation quality while increasing the N2 flow degraded crystal orientation. A higher DC bias led to a more uniform film. Acoustically coupled monolithic filters were fabricated after the conditions for the best quality films were determined.

Introduction:

The objective of this study was to optimize AlN films for use in filter devices such as thickness-mode thin-film piezoelectric-on-substrate (TPoS) filters [1]. The AlN layer influences device properties such as center frequency, bandwidth and insertion loss� AlN is a material of choice for use in complementary metal oxide semiconductor (CMOS)-compatible electro-acoustic devices as it can be deposited at low temperatures (below 400°C)� For low loss filters, AlN was chosen as the transduction material because it has low dielectric loss, high resistivity, and a reasonably large piezoelectric coupling [1]� Reactive sputtering of AlN was characterized on molybdenum (Mo) electrodes with a titanium (Ti) seed layer on a silicon (Si) wafer since previous studies have shown that Mo and Ti are optimal materials for deposition of highly c-axis oriented AlN films [2].

Deposition parameters of AlN were optimized for film stress, crystal orientation, and uniformity. Sputtered AlN films are polycrystalline in nature and the effective piezoelectric coupling is a vector summation of grain orientation� Thus, good c-axis crystalline orientation is important to achieve low loss devices� The piezoelectric properties of the film have a strong correlation with the full-width at half maximum (FWHM) value of the x-ray diffraction (XRD) rocking curve [3], and they can be controlled by adjusting gas flow rates. Large film stress leads to complications in releasing suspended structures�

A released device with high stress could have large device bow or film cracking. While reducing device size can help alleviate these problems, the motional impedance increases with reducing device area. As a result, control of film stress

is important in order to achieve good device yield� Changing gas flow rates causes thickness uniformity degradation, which can then be independently tuned by applying a DC bias to the substrate�

Experimental Methods:The 1 µm thick AlN characterization films were deposited on patterned bottom electrodes (100 nm Mo on a 20 nm Ti seed layer) on a clean 4-inch <100> Si wafer. AlN films were deposited using the Tegal AMS 2004 dual cathode reactive magnetron sputtering system. Post film characterization, TPoS filters were fabricated on silicon-on-insulator (SOI) wafers using published fabrication methods [1]�

In order to characterize the effect of gas flow variation, N2 flow was held constant at 28 standard cubic centimeters per minute (sccm) while varying Ar flow. Ar flow was held constant at 35 and 45 sccm in two separate experiments while varying the N2 flow (Figures 1-3). DC bias was varied from 0�1 to 0�7 kW to tailor the uniformity (Figure 4)� AlN film crystal orientation was characterized by taking XRD measurements with a Rigaku Ultima IV system�

Results and Conclusions:Increasing the gas flow rate reduced compressive stress (Figures 1 and 2). At higher flow rates, further increasing the flow rates had a smaller impact on the stress of the films. When the Ar flow rate was constant at 45 sccm, the

• pro

cess &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n •


total gas flow appeared high enough that increasing N2 flow had little effect on stress. At lower flow rates, species in the sputtering chamber transfer more energy to the substrate due to fewer collisions in the sputtering chamber� This allows atoms in the substrate to relocate more easily, enabling the removal of vacant sites and an increase in compressive stress [4]� The ion bombardment energy is reduced with increased flow rates and may not exceed the activation energy, thereby causing further increases in flow rate to have a smaller impact on the stress.

A peak at 35�8° on a θ-2θ scan verified that the AlN films had a c-axis orientation� Rocking curves indicated FWHM measurements from 1�25° to 1�54°� Increasing N2 flow led to an increase in FWHM, implying worse film orientation, while increasing Ar flow led to a decrease in FWHM (Figure 3). At lower flow rates, we expected a better FWHM because species colliding with the substrate transfer more energy to the ad-atoms of the substrate, allowing for better crystal formation� Ar ions have been shown to affect crystal orientation more than N species in the chamber due to a larger energy transfer to the ad-atoms of the substrate [5]. Despite higher flow rates, it is possible that FWHM improved with increased Ar because of the greater role of Ar ions in energy transfer�

Film uniformity varied linearly between 2�4% to 5�2% with DC bias (Figure 4). At a lower Ar flow of 35 sccm, film uniformities better than 1�8% were achieved�

Acknowledgements:I would like to thank my principal investigator, Professor Mina Rais-Zadeh, my mentor Vikram A� Thakar, the staff at the Lurie Nanofabrication Facility, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, and the National Science Foundation�

References:[1] W� Pan, R� Abdolvand and F� Ayazi, “A low-loss 1�8 GHz monolithic thin-

film piezoelectric-on-substrate filter,” MEMS, 2008.[2] V� Felmetsger, P� Laptev, and S� Tanner, “Crystal Orientation and Stress

in AC Reactively Sputtered AlN Films on Mo Electrodes for Electro-Acoustic Devices,” Ultrasonics Symposium, pp� 2146-2149, 2008�

[3] M� Dubois and P� Muralt, J� Applied Physics, Vol� 89, pp� 6389-6395, 1999�

[4] G�F� Iriarte, J�G� Rodriguez, F� Calle, Microsyst� Technol�, Vol�17, pp� 381-386, 2011�

[5] M. Clement, et. al., “Influence of sputtering mechanisms on the preferred orientation of aluminum nitride thin films,” J. Applied Physics, Vol. 94, pp� 1495-1500, 2003�

Figure 4: Uniformity variation with DC bias.

Figure 1: Stress variation with Ar flow rate.

Figure 2: Stress variation with N2 flow rate.

Figure 3: FWHM variation with gas flow rate.

• p

roce

ss &

chara

cteri

zati

on

• p

roce

ss &

chara

cteri

zati

on •

pro

cess &

chara

cteri

zati

on •


Novel Process to Fabricate Raised Polymer Electrodes for Electroencephalography

Fiona O’ConnellMaterials Engineering, Loyola University Maryland

NNIN iREU Site: Centre Microélectronique de Provence, Ecole Nationale Supérieure des Mines de Saint Etienne, FranceNNIN iREU Principal Investigator: Dr. George G. Malliaras, Department of Bioelectronics,

Centre Microélectronique de Provence, Ecole Nationale Supérieure des Mines de Saint EtienneNNIN iREU Mentor: Pierre Leleux, Department of Bioelectronics, Centre Microélectronique de Provence,

Ecole Nationale Supérieure des Mines de Saint EtienneContact: [email protected], [email protected], [email protected]


Twenty percent of epileptic patients are unable to receive drugs and must undergo surgery, requiring an invasive procedure to determine the epileptogenic zone� Electrodes are implanted into the brain and neural activity is monitored for one week while electroencephalography (EEG) is used concurrently to create multi-layered recordings� EEG systems use conducting gels to adhere electrodes to the scalp� However, current EEG systems suffer from poor long-term skin adherence and are difficult to apply around the implanted electrodes�

The purpose of this work was to develop a novel process for the fabrication of flexible conducting polymer electrodes for EEG, which would record long-term signals� For the electrode insulator, Parylene-C (PaC) was deposited into a flexible, biocompatible, thin film. The electrodes were patterned via photolithography and consisted of metal deposited on a conducting polymer layer� To provide high conductivity and biocompatibility, poly(3,4-thylenedioxythiophene) doped with poly(styrene sulfonate) (PEDOT:PSS) was used as the conducting polymer� The electrodes were designed to protrude from the PaC matrix through the use of a sacrificial polyvinyl alcohol (PVA) layer. The sacrificial layer increased the surface contact area at the skin to electrode interface while decreasing impedance� Once fabrication was completed, the device

Figure 1: Process flow diagram for the fabrication of electrodes. a.) electrode patterning, b.) channel patterning and PEDOT and Au deposition, c.) lift-off of resist and excess PEDOT andAu, and d.) PaC deposition and peel off.

was released from the glass substrate by dissolving the PVA in water, resulting in conformable and flexible electrodes.

Fabrication Process:The development of the following process for flexible polymer electrodes was largely based on reversing methods described in the literature [1]� The fabrication process, shown in Figure 1, began with the deposition of a 2�5 µm sacrificial layer of 10 wt.% PVA onto a cleaned glass substrate� The solution was spin-coated onto the substrate at 500 rpm for five seconds, followed by 1500 rpm for 30 seconds and finally the substrate was baked for five minutes at 95°C� After, a 2�5 µm layer of PaC was chemically vapor deposited onto the PVA layer�

Subsequently AZ 9260 resist was spin-coated onto the PaC layer and the substrates were patterned using conventional photolithography and oxygen plasma etching techniques A second layer of resist was spin-coated and the samples underwent lithography and etching� The excess resist was not removed after the second etching, as it was a necessary scaffold�

A PEDOT:PSS (Clevios PH 500, HC Starck) and ethylene glycol mixture (4:1 by volume) was then spin-coated onto

the twice patterned PaC at 1700 rpm for 30 seconds and baked at 65°C for 60 seconds, and again spin-coated at 1000 rpm for 30 seconds followed by a five minute annealing at 110°C to produce a 200 nm layer� Next a 100 nm layer of gold (Au) was deposited using thermal joule source evaporation� The excess resist, Au and PEDOT were removed by soaking the substrate in acetone for approximately one hour. A final layer of PaC was deposited using the same parameters

• pro

cess &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n •


as above� Finally, the device was peeled off of the glass substrate after a 6-hour soak in DI water to dissolve the PVA and release the PaC�

Results and Discussion:During process development, the Au layer originally lacked continuity along the sidewalls between the site and channel patterns due to inadequate step coverage of metal evaporation� The necessary slopes were eventually attained by overexposure during the first lithography step. The duration of exposure that provided the optimal slope was ten seconds using SI 1813 Shipley Microposit photoresist� The effects of exposure time were analyzed using an Ambios XP2 profilometer. [Figure 2]

Through the optimization of both etching steps, electrodes were developed that protruded from the PaC matrix and provided stress relief during peel off� The PVA layer not

only served as a sacrificial layer, it also allowed for the actual protrusion� By etching several microns into the PVA layer, when the device peeled off the glass substrate, the sites raised above the PaC matrix� The raised PEDOT electrode sites increase the surface contact between the skin and electrodes, decreasing impedance. The optimal first etching time for these electrodes was eight minutes and the second etching time was three minutes� After fabrication, the electrodes were tested for conductivity and displayed acceptable conductivity for their use�

Conclusions and Prospects:A novel process has been developed to produce conformable, biocompatible electrodes� The procedure reverses the standard fabrication steps in order to have the desired electronics mounted on a flexible PaC platform. The electrodes exhibited adequate conductivity�

Future work will focus on further optimizing the above process in order to fabricate electrodes of different and smaller dimensions� Once optimization of the process is completed, both stretchable electrodes as well as stretchable substrates will need to be explored in order to formulate a complete EEG system� Organic electronics have become an exciting new possibility in the area of research over the past twenty years and definitely have the potential to revolutionize current methods for epilepsy diagnosis�

Acknowledgements:I would like to thank the National Science Foundation (NSF), the National Nanotechnology Infrastructure Net-work International Research Experience for Undergraduates (NNIN iREU) Program, and the Ecole Nationale Superieure des Mines Saint Etienne for funding and providing facilities for conducting this research� Furthermore I would like to express my appreciation to Dr� George Malliaras, Pierre Leleux, and the Bioelectronics Lab for including me in their research and guiding me through this work�

References:[1] Ismailova, E�, Doublet, T�, Khodagholy D�, Quilichini P�, Ghestem

A�, Yang S�Y�, Bernard C�, Malliaras G� (2011) “Plastic neuronal probes for implantation in cortical and subcortical areas of the rat brain”, Int� J� Nanotechnology, Vol� 10�

Figure 2: Effect of overexposure of site pattern on sidewall slope.

• p

roce

ss &

chara

cteri

zati

on

• p

roce

ss &

chara

cteri

zati

on •

pro

cess &

chara

cteri

zati

on •


Design, Fabrication, and Testing of Hg/Au Microelectrodes for Voltammetric Sensing of Trace Metals for Environmental Monitoring

Andrew RaebigElectrical Engineering and Chemistry, Butler University

NNIN REU Site: Lurie Nanofabrication Facility, University of Michigan, Ann Arbor, MINNIN REU Principal Investigators: Dr. Hélène Craigg, Electrical and Computer Engineering, University of Michigan;

Dr. Becky Peterson, Electrical and Computer Engineering, University of MichiganNNIN REU Mentor: Dr. Hélène Craigg, Electrical and Computer Engineering, University of MichiganContact: [email protected], [email protected], [email protected]

Abstract:

In electroanalytical chemistry, electrodes are used in voltammetry to detect trace amounts of an analyte in an aqueous system. Microelectrode arrays can be made using nanofabrication techniques. A series of mercury/gold (Hg/Au) microelectrode arrays was designed as a working electrode in a voltammetric system for trace metal sensing; these arrays test the influence of electrode diameter, array size, and electrode spacing on signal detection. After device fabrication, all electrode arrays were taken to the Georgia Institute of Technology for mercuric electroplating and testing by our geochemistry collaborator.

Introduction:

Microelectrode (ME) arrays have come into highly increased use since their first invention in the 1970s due to several unique characteristics� They are very advantageous compared to individual microelectrodes, exhibiting increased sensitivity and high spatial resolution� Their small size and workable area allow them to be used for measurement without overly disturbing the system� Since they are often made from noble or heavy metals, they can be used to detect a virtually unlimited range of compounds and elements� In this study, ME arrays with varying characteristics were created on a gold (Au) film covered with silicon nitride (Si3N4) and electroplated with mercury (Hg) similar to the procedure done by Belmont, et al� [1]� The electrode array characteristics were varied both individually and collectively to determine their impact on performance�

Mask Development. For the lithographic process, two masks were designed using AutoCAD. The first mask was used to pattern the Au pads� The second mask was used to pattern the electrode arrays on the top Si3N4 layer� Multiple arrays were created to encompass variations of electrode diameter (5-30 µm), electrode spacing [2] (25-900 µm), and array size (30-1800 electrodes)� Figure 1 shows an AutoCAD drawing of a device�

Electrode Fabrication. Figure 2 displays a cross-sectional sketch of the finished device. On standard 100 mm Si wafers, a 200 nm layer of Si3N4 was deposited via low pressure chem ical vapor deposition (LPCVD)� Then a 200 nm

Figure 1, top: Mask layout of a basic array. The yellow lines indicate the Au mask; the red lines the Si3N4 mask.

Figure 2, bottom: Cross-section sketch of the finished device.

• pro

cess &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n •


layer of Au was evaporated; a 5 nm layer of chromium (Cr) was evaporated above and below as an adhesion layer and protection layer, respectively� Pre-spun wafers were exposed for 6 s under the light-field Au pattern mask. Once development was complete, each wafer was wet-etched to pattern Au and Cr� Photoresist was stripped in a hot acetone bath, quenched in isopropanol, and rinsed in deionized water before a 10-minute O2 plasma etch�

A 200 nm layer of Si3N4 was then deposited using plasma-evaporated CVD (PECVD)� Compared to LPCVD-deposited films, PECVD films have many inherent flaws; the most prolific of which is the presence of pinholes. Pinhole testing was done by submerging wafers in buffered hydroflouric acid (HF) for short periods of time (~ 5 s)� The electrode arrays were patterned using photolithography and reactive-ion etching (RIE) for an average of 6 min� Following photoresist strip, a final Cr etch was done to reveal the Au electrodes. The finished array is shown in Figure 3.

Each wafer was diced to separate the devices� On each device, a solid copper wire was bonded to the Au contact pad using silver colloidal conductive epoxy� The completed working electrode is shown in Figure 4�

Results and Discussion:This study had three primary goals: mask development for the photolithographic production of microelectrode arrays;

design and characterization of a microfabrication process; and fabrication and testing of the physical electrodes themselves� Creation of a mask that tested many different scenarios of microelectrodes required detailed planning� All controllable characteristics of the device were manipulated in as many ways as possible in order to narrow the search for an optimal device� In order to create these electrodes, the microfabrication process had to be characterized� In order to optimize Si3N4 layer thickness and quality, vapor deposition parameters had to be adjusted� The RIE etch rate also required careful characterization and timing in order to avoid over-etching�

After the devices were fitted with wires, each one was electrically tested for shorts or poor connections� Using electrochemical apparatus provided by the Georgia Institute of Technology, each electrode was placed in mercuric solution and plated at -0�1 V for ~ 30 s� The plated electrodes were subjected to a cyclic voltammetry test to measure the stability of the Hg/Au amalgam, and then tested overnight in 0�5 M NaCl (seawater) doped with 10 mM manganese to determine long-term stability�

Conclusions:In this study, it was determined that usable microelectrode arrays of differing dimensions could be created through microfabrication in order to design a voltammetric working electrode. The voltammetry profiles, while successful, indicated that more optimization of electroplating was required. Several modifications to this process can be made in future studies: fabrication with different mercury-compatible metals, deposition of conductors and insulators via different methods, and adjustment of electrode specifications.

Acknowledgements:I would like to thank the National Nanotechnology Infra-structure Network Research Experience for Under graduates Program for the opportunity to conduct this project; the Lurie Nanofabrication Facility (LNF) for providing the instrumentation; my PI, Dr� Hélène Craigg; Dr� Becky Peterson for her assistance in mask design and electrical testing; Dr� Katharine Beach for her suggestions regarding fabrication; the Georgia Institute of Technology for their assistance plating and testing the devices; and all the LNF staff for providing aid whenever needed�

References:[1] Belmont, C., Tercier, M.L., Buffle, J., Fiaccabrino, G.C., Koudelka-

Hep, M� “Mercury-plated iridium-based microelectrode arrays for trace metals detection by voltammetry: optimum conditions and reliability�” Anal� Chim� Acta, 329, 203 (1996)�

[2] Wang, J�, Lu, J�, Tian, B�, Yarnitzky, C� “Screen-printed ultramicroelectrodes arrays for on-site stripping measurements of trace metals�” J� Electroanal� Chem�, 361, 77 (1993)�

Figure 3, above: Zoom of electrode array.

Figure 4, left: Diced and bonded electrode.

• p

roce

ss &

chara

cteri

zati

on

• p

roce

ss &

chara

cteri

zati

on •

pro

cess &

chara

cteri

zati

on •


Optical and Electron Beam Patterning for Graphene Nanoribbon Devices

Nathanial SheehanElectrical Engineering, University of California, Santa Barbara

NNIN REU Site: Nanofabrication Center, University of Minnesota-Twin Cities, Minneapolis, MNNNIN REU Principal Investigator: Dr. Steven J. Koester, Electrical Engineering, University of Minnesota-Twin CitiesNNIN REU Mentor: Yoska Anugrah, Electrical Engineering, University of Minnesota-Twin CitiesContact: [email protected], [email protected], [email protected]

Abstract and Introduction:Graphene is a monolayer material that could allow for smaller, high-performance field effect transistors. Transistor scaling has allowed the microelectronics industry to advance over the last forty years, leading to denser, higher performance systems that consume much less power� However, silicon scaling has reached fundamental limits and a replacement material that can be scaled more is needed, leading to further performance improvements� The carrier mobility of graphene has been shown to be roughly an order of magnitude higher than silicon, making it a possible replacement [1]� Graphene must be sliced into nanoribbons less than 5 nm wide to induce a band gap (due to quantum confinement) suitable for room temperature transistor operation [2]� This width is beyond the patterning ability of traditional optical lithography, thus novel lithographic techniques that will work with graphene processing techniques are needed�

This research explored using block copolymers (BCPs) as a mask for etching graphene nanoribbons (GNRs)� BCPs are composite materials containing two or more homopolymers� They can be induced to phase separate and align to existing lithographic guiding structures� The research performed in this REU project focused on fabrication of guiding structures for the alignment of straight, parallel BCP patterns� The effect of the depth of optically patterned guiding structures on BCP alignment was investigated� Electron-beam lithography (EBL) was also employed to pattern narrow-width guiding structures to determine how the width of narrow guiding structures affects alignment�

Method:Figure 1 outlines the process used to fabricate optically patterned guiding structures� The result was patterned channels in the silicon nitride (Si3N4) with different initial nitride layer thicknesses� We chose to make Si3N4 depths of 20, 40, 60, and 80 nm� The width and length of the channels also varied in a large array pattern (Figure 2), with a minimum width of approximately 300 nm� The wafer was then cleaved, BCP spun on it, and it was solvent-annealed to phase separate and align the BCPs. The alignment efficacy

Figure 1: Optically patterned wafer process outline. Different shades of gray represent different layers, and different processes in steps 2 through 5.

Figure 2: Close up microscope image of completed 60 nm deep guiding structures. Nitride is dark and the raised part.

• pro

cess &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n •


of the variously dimensioned channels was determined by atomic force microscopy (AFM)�

EBL was explored to make guiding structures of narrower widths, in the range 50 to 500 nm, with the depth varying as above� The process was similar to the optical process� We determined the proper EBL exposure strength by performing scanning electron microscopy on patterned test chips�

Results and Discussion:The 60 nm deep guiding channels were shown to align some of the BCPs in large width (2 µm) channels, which is promising because this allows for optical lithographic patterning of the guiding structures (Figure 3)� It was observed that the deposition and annealing conditions for the BCP film play a large role in the quality of the alignment to the guiding structures. These conditions affected film uniformity, which is required for high quality alignment� The height of the film was non-uniform in some channels, leading to misalignment of the BCP� This may be due to dewetting and flow of BCP on the regions between channels, i.e. the BCP “balled up” in raised regions and later flowed back into the channels, creating non-uniform thickness in some areas�

Preliminary AFM scans of the 20 nm deep channels showed some unexpected results (Figure 4). The BCP film was continuous over the mesas, and was still aligned in areas of high film uniformity. However, the direction of BCP alignment was perpendicular to the alignment channels instead of the expected parallel alignment� This suggests the topography of the guiding structures, and not just the conditions under which the BCPs are aligned, plays a large role in alignment direction [3]�

Analysis was performed on how EBL exposure conditions affected the EBL-defined guiding structure dimensions. We found that channels of the same target width for doses of 600 µC/cm2 and 800 µC/cm2 had similar dimensions, but the smallest channel widths (50 nm) were not resolved at the higher dose� All of the channels were too wide, so further refinement of the initial pattern is needed.

Future Work:We would like to investigate the effect of the other optically patterned nitride channel depths (40 and 80 nm) to determine the best conditions for optimal BCP alignment� The EBL guiding structure fabrication process must also be refined to allow narrow-width EBL written channels to be fabricated, so that they may guide the BCPs� Once the optimum guiding dimensions are found, a process for transferring the BCP-defined patterns into graphene must be developed. Finally, the guiding structures must be integrated with a graphene transistor process flow to enable functional devices.

Acknowledgments:I would like to thank: Yoska Anugrah and Dr� Steven J� Koester for hosting me this summer, teaching me fabrication, and guiding me through the research process; Steve Brown and Dr� Brian Olmsted for their help with the BCP process, and gathering and analyzing data; and the NFC Staff for training� Additional thanks to the University of Minnesota, the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program, and the National Science Foundation for hosting and funding this opportunity�

References:[1] Novoselov, K� S�, Geim, A� K�, Morozov, S� V�, Jiang, D�, Zhang, Y�,

Dubonos, S. V., Grigorieva, I. V., Firsov, A. A.; “Electric field effect in atomically thin carbon films”; Science, 306, 666–669 (2004).

[2] Schwierz, F; “Graphene transistors”; Nature Nanotechnology, 5, 487-496 (2010)�

[3] Jung, Y�S�, Ross, C�A; “Orientation-controlled self-assembled nanolithography using a polystyrene-polydimethylsiloxane block copolymer”; Nano Letters, 7, 2046-2050 (2007)�

Figure 4: Preliminary AFM image showing BCP alignment perpendicular to guiding channels; 20 nm depth, 0.95 µm width.

Figure 3: AFM showing parallel BCP alignment; 60 nm depth, 1 µm width guiding channel. Figure is typical of imaged results.

• p

roce

ss &

chara

cteri

zati

on

• p

roce

ss &

chara

cteri

zati

on •

pro

cess &

chara

cteri

zati

on •


Systematic Investigation of Morphology of Polymer:Bis-Fullerene Blends for Bulk Heterojunction Organic Photovoltaics

Joseph SmalleyEngineering Science, The Pennsylvania State University (graduated), Electrical and Computer Engineering, University of California, San Diego

NNIN iREU Site: Interuniversity Microelectronics Center (imec), Leuven, BelgiumNNIN iREU Principal Investigator: Dr. Barry Rand, Polymer and Molecular Electronics, imecNNIN iREU Mentor: Eszter Voroshazi, Polymer and Molecular Electronics, imec, Katholieke Universiteit, Leuven, BelgiumContact: [email protected] [email protected], [email protected]

Abstract:

Active layer morphology plays an essential role in determining the performance of polymer:fullerene bulk heterojunction (BHJ) organic photovoltaic (OPV) cells. Upon processing with several amorphous polymers, a promising new bis-fullerene—bis-oquino dimethane C60 (bis-oQDMC60)—was recently found to exhibit undesired formation of holes at the surface with diameter on the order of 10’s of nanometers at a density of approximately 25 µm-2 (~ 0.1-1.0% of film surface area). This poor film formation significantly hinders device performance. We prepared and characterized BHJ thin films to systematically study the processing conditions under which these holes arise. Spin-coated blends of poly(3-hexylthiophene) (P3HT):bis-oQDMC60 and P3HT:bis-phenyl-C61-butyric acid methyl ester (PCBM) were investigated by comparison to the well-known P3HT:PCBM system. The blends were characterized by atomic force microscopy (AFM), ultraviolet-visible (UV-Vis) spectrophotometry, and contact angle measurements. Throughout the annealing temperature range of 27°C-130°C, we found P3HT:bis-oQDMC60 (1:4 ratio) always exhibited hole formation. Furthermore, at room temperature, films of P3HT:bis-PCBM (1:4) also exhibited holes, indicating hole formation may be a general property of bis-fullerene derivatives at high loadings within polymer films. Measurement of the absorption spectra and contact angle of our samples revealed the likelihood that the nanoholes form a porous network in the bulk.

Introduction:

Because they can be fabricated inexpensively, OPV cells need have neither the efficiency nor lifetime of inorganic solar cells. To date, researchers have reported a record power conversion efficiency of 8.3% [1], and outdoor operating lifetimes of more than one year [2], with hopes of reaching an efficiency of over 10%. One route by which efficiency can be improved is through the use of novel acceptor materials that enable a high open circuit voltage, Voc� Whether this increased Voc can be realized without a related reduction in short circuit current density, Jsc, is an essential question for determining the utility of these new materials�

With a higher Voc than the reference P3HT:PCBM, and a comparable Jsc, the P3HT:bis-oQDMC60 (1:1) blend demon-strates that bis-oQDMC60 is a promising acceptor [3]� In contrast, Figure 1 shows the performance of an OPV cell made of poly(2-methoxy-5-(3’,7’-dimethyl-octyloxy))-p-phenylene vinylene (MDMO-PPV):bis-oQDMC60(1:3)� In this unfortunate case, we see that Voc is improved at the expense of a much lower Jsc� AFM scans revealed that the unannealed blend of MDMO-PPV:bis-oQDMC60 exhibited nanoholes whose properties are explained in the abstract� Whether the loading factor or annealing conditions

Figure 1: Current density vs. voltage characteristics of MDMO-PPV:PCBM and MDMO-PPV:bis-oQDMC60, measured under AM1.5G simulated solar illumination.

• pro

cess &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n •


were responsible for formation of theses holes, and hence the reduced current density, was an open question, as was the extent of the holes in the bulk�

Methods:Blends of P3HT:bis-oQDMC60 and P3HT:bis-PCBM were investigated by comparison to the well-known P3HT:PCBM system� Substrates consisted of glass coated with indium tin oxide (ITO) that were cleaned with sequential ultrasonic baths in soapy water, deionized water, acetone, and isopropyl alcohol, followed by an ultraviolet radiation cleaning� Afterwards, a 25 nm thick poly(3,3-ethylenedioxythiophene) poly(styrenesulfonate) (PEDOT:PSS) film was spin-coated and the samples were annealed in nitrogen for 10 min at 130°C� Solutions of P3HT and bis-oQDMC60, each dissolved in odichlorobenzene, were prepared and then combined in respective ratios� These blends were deposited by spin-coating on the PEDOT:PSS layer and followed by an optional 10 min annealing in nitrogen at 130°C, or a variable temperature�

The blends were characterized by AFM, UV-Vis spectro-photometry, thickness, and contact angle measurements�

expected, because bis-oQDMC60 also absorbs more strongly at lower wavelengths� However, the opposite trend was exhibited; viz, higher loading ratios have lower attenuation coefficients. We attributed this unexpected trend to the presence of a porous network of nanoholes at the 1:4 and 1:3 ratios, which was supported by contact angle measurements�

The contact angles of water on the active layer of the 1:4 and 1:3 ratios of P3HT:bis-oQDMC60 were measured to be much lower than any of the other samples� During measurement, we observed that the water dissolved the PEDOT:PSS layer underneath the active layer upon which water was dropped� The dissolved PEDOT:PSS then effectively washed away the active layer, causing the measurement to record values similar to water in contact with the glass substrate� Based on this observation and our absorption measurements, we concluded that the holes do indeed form a porous network throughout the bulk, allowing water to pass through and limiting the absorption considerably�

Conclusion:We have prepared and characterized samples of P3HT:bis-oQDMC60 and compared them to P3HT:PCBM and P3HT:bis-PCBM� Our results indicate that hole formation may be a general occurrence for bis-fullerenes blended with polymers� Additionally, we have strong evidence from UV-Vis and contact angle measurements suggesting that the nanoholes form a porous network� These results are valuable for continued study of polymer:bis-fullerene adducts, and ultimately for high-performance OPV�

Acknowledgments:Foremost, I thank my mentor, Eszter Voroshazi, for her guidance throughout all of this work� I also thank Dr� Barry Rand, for overseeing the progress of this research� Finally, I thank the National Nanotechnology Infrastructure Network International Research Experience for Undergraduates (NNIN iREU) Program and the National Science Foundation for financial support.

References:[1] M�A� Green et al� Progress in Photovoltaics: Research and

Applications, 19, 562-572 (2011)�[2] J� Hauch et al�, Solar Energy Materials and Solar Cells, 92, 7, 727-

731 (2008)�[3] E� Voroshazi et al� J� of Materials Chemistry, submitted May 2011�

Figure 3: Absorption of P3HT:fullerene for various ratios: (a) P3HT:PCBM, (b) P3HT:bis-oQDMC60.

Figure 2: AFM scan of unannealed P3HT:bis-oQDMC60 (1:4).

Results and Discussion:Throughout the annealing temperature range of 27°C-130°C, we found P3HT:bis-oQDMC60 (1:4) to always exhibit hole formation� Figure 2 shows the surface of an unannealed sample of P3HT:bis-oQDMC60 (1:4)� At room temperature, films of P3HT:bis-PCBM (1:4) also exhibited holes, indicating hole formation may be a general property of bis-fullerene derivatives at high loadings within polymer films.

Absorption measurements suggested that the holes affect absorption in the bulk of the film. As seen in Figure 3a, higher loading ratios of P3HT:PCBM had higher attenuation coefficients, resulting from the fact that absorption of PCBM was strongest between 300 and 400 nm� For P3HT:bis-oQDMC60, a similar trend was

• p

roce

ss &

chara

cteri

zati

on

• p

roce

ss &

chara

cteri

zati

on •

pro

cess &

chara

cteri

zati

on •


Tribology of Atomic Layer Deposition Films

Kelly SuralikChemistry, Middlebury College

NNIN REU Site: Stanford Nanofabrication Facility, Stanford University, Stanford, CANNIN REU Principal Investigator: Prof. Roger T. Howe, Electrical Engineering, Stanford UniversityNNIN REU Mentors: Dr. J. Provine, Electrical Engineering, Stanford University; W. Scott Lee, Electrical Engineering,

Stanford University; and Dr. Wes Smith, Mechanical Engineering, Hewlett-Packard Research LabsContact: [email protected], [email protected], [email protected],[email protected], [email protected]

Abstract:

Atomic layer deposition (ALD) is revolutionizing the fabrication of nanoscale devices. ALD employs sequential, self-limiting vapor surface reactions and presents the ability to coat concavities and convexities of a surface uniformly with thin inorganic films. Since ALD films can deposit uniformly on a variety of materials with precise thickness control, they can tune surface properties independent of the substrate. The electrical properties of ALD films have been investigated extensively, yet the films’ mechanical traits have not been well characterized. In this work, we investigated mechanical properties such as wear, adhesion and friction of the interface between ALD coatings. Custom micromachined silicon tips on compliant cantilevers were coated with various films using thermal and plasma ALD processes. A scanning electron microscope (SEM) was used to observe tips before and after deposition. The coated tips were tested through laser Doppler vibrometry to monitor friction and adhesion. The knowledge of the mechanical effects of ALD films will improve our ability to fabricate nanoscale electromechanical systems.

Figure 1: SEM of compliant cantilever.

Introduction:Tribology is the study of interacting surfaces in relative motion, which includes properties of wear, friction and adhesion� The tribological test structure in this project examined mesoscopic forces that cause nanoscale friction and were comprised of compliant cantilevers arrays on silicon chips� Figure 1 shows a SEM of a compliant cantilever� Square pieces of silicon, called sliders, rested on sharp tips at the end of the cantilevers� The cantilevers were compliant to ensure all cantilever tips engaged when the slider rested on the array� The wear and friction testing was completed through laser Doppler vibrometry at Hewlett-Packard in Palo Alto, California�

The slider rests unconstrained on the array, and the array moves on the horizontal shaker relative to the slider, causing wear to the tips. Laser light shines on and reflects off the shaker and slider into the laser Doppler vibrometer (LDV)� The LDV uses the laser light reflection to calculate the relative motion of the shaker and slider� A LDV schematic is seen in Figure 2�

By coating the cantilever tips and the sliders with ALD films, the mechanical properties of the coatings could be examined� ALD utilizes sequential, self-limiting surface reactions to deposit a thin film of inorganic material on the

Figure 2: Schematic of laser Doppler vibrometer [1].

• pro

cess &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n • p

roce

ss &

chara

cteriz

atio

n •


substrate� The vapor-phase reactants and the self-limiting nature of the reactions cause a uniform deposition on the surface� During deposition, two gaseous reactants, called precursors, are pumped into the reaction chamber in succession� Each precursor reacts with all available reactive sites, and the excess is pumped out of the chamber� The two half-reactions constitute one cycle, and a cycle deposits a thin layer of the new substance on the substrate� Cycles are repeated to obtain the desired film thickness. The mechanical characteristics are dominated by surface qualities at the nanoscale due to the high surface-area-to-volume ratio� Thus, ALD can change the exhibited properties of a material greatly by depositing thin films.

Experimental Procedure:Custom micromachined silicon tips on silicon wafers were processed to produce the compliant cantilevers arrays� We employed traditional lithography methods to pattern the wafers with cantilevers� Silicon dioxide deposited by low pressure chemical vapor deposition was etched from the wafer’s surface through reactive ion etching with CHF3� The cantilevers were defined from the bulk of the wafer through an anisotropic silicon etch with sequential introduction of SF6 and C4F8 gases� The wafers were diced with a wafer saw� The cantilevers of each individual chip were released from the surface of the wafer through a XeF2 isotropic silicon etch� The protective oxide on the cantilever tip was removed with a silicon dioxide wet etch of 6:1buffered oxide etchant� The sliders were prepared by removing the protective oxide through a silicon dioxide wet etch�

A carrier wafer was fabricated to secure the chips and sliders within the ALD reaction chamber� After manufacturing

several carrier wafer designs, we created a carrier wafer with slots for the chips and sliders etched completely through the 500 µm wafer� The chips and the sliders sat directly on the reaction chamber surface� The chips and sliders were coated with their respective film through ALD. SEM images were taken after each step in the fabrication procedure�

Results and Conclusions:Forty nanometers of aluminum oxide and hafnium oxide were deposited on compliant cantilever test structures� Figure 3 shows a cross-section of a hafnium oxide tip obtained with a focused ion beam� The arrays were initially characterized� To begin wear and friction testing, sliders were placed on the cantilever arrays� The initial slider placement damaged the sharp cantilever tip upon impact, as seen in previous tribological studies of silicon-silicon interactions� The coated cantilever arrays were proved as a viable method to examine the mechanical properties of ALD films.

Future Work:Future work will compile a comprehensive analysis of the mechanical properties of ALD films. The LDV testing of the aluminum oxide and hafnium oxide arrays will be completed, and testing will continue for other ALD films such as titanium oxide, titanium nitride, tungsten nitride, platinum and ruthenium� The differences in mechanical characteristics of plasma ALD films versus thermal ALD films will also be investigated.

Acknowledgements:I extend sincere gratitude to my mentors Dr� J� Provine, Scott Lee and Dr� Wes Smith� I thank Prof� Roger T� Howe and the Howe Research Group for their constant support� I also thank Dr� Michael Deal, Maureen Baran and the Stanford Nanofabrication Facility staff, lab members and REUs for their aid� Finally, I acknowledge the National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program, the National Science Foundation and the Center for Integrated Systems for financial support.

References:[1] Smith, W� (2010)�Shear Adhesion, Friction, and Wear of Multi-Point

Micro- and Nano-Scale Contacts� Ph�D� Thesis� Stanford University�Figure 3: Cross-section of ALD HfO2 tip. The bright film on the silicon is HfO2. The two black lines were added to the image to indicate the film thickness clearly.

• s

oci

eta

l &

eth

ical

issues

• s

oci

eta

l &

eth

ical

issues •

soci

eta

l &

eth

ical

issues •


Researcher Views on the Perceived Influence of Funding Sources in Nanotechnology Research

Rachel BrockhageBiology and Communication, Grove City College

NNIN REU Site: Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NYNNIN REU Principal Investigator: Dr. Katherine McComas, Department of Communication, Cornell UniversityNNIN REU Mentor: Christopher Clarke, Department of Communication, Cornell UniversityContact: [email protected], [email protected], [email protected]

Abstract:

Scholars have increasingly focused on the influence of funding sources on research directions and potential conflicts of interest (COI) that arise in scientific research today. Conflicts of interest occur when an individual has a stake in the outcome of a behavior along with the means to influence the particular outcome. COI may be directly influenced by funding arrangements. Understanding researcher perceptions and identifying opportunities for managing COI that may arise are essential to this inquiry. Building on a study of nanotechnology industry and academic researcher views of funding sources and COI, the present study explores the extent to which graduate students, as the next generation of researchers helping shape the direction of the field, (1) believe funding arrangements influence research directions, and (2) recognize and evaluate COI that arise in their work. The study included the implementation of a web survey of users of the fourteen National Nanotechnology Infrastructure Network (NNIN) sites who have recently received their terminal degree.

Introduction:

Funding for nanotechnology research and development has increased both in the U�S�A� and internationally in recent years� Cumulative investment by the U�S� government from 2001 to 2012, as part of the National Nanotechnology Initiative (NNI), totaled $16�5 billion, including $2�1 billion for fiscal year 2012 (NNI, 2011). The NNI (2010) estimated that large corporations and industry support approximately half of the research and development in nanotechnology� The prominence of industry funding and academic-industry collaborations speak to the role of funding sources and their relationship to issues of research direction, research conduct, questions researchers ask, and the extent to which results are shared with other scholars and the larger public� It also speaks to potential conflicts of interest (COI), which are said to occur “when an individual or entity has a stake in a decision and also the means to influence it.” (McComas, forthcoming, p� 2�)

Financial COI are generally the most visible type researchers may encounter in their work� The COI can be relatively straightforward, such as when a researcher reviews his/her own article or grant proposal, or less well-defined, as when a researcher exaggerates the significance of results to attract additional funding, or downplays potential risks to protect a funding source (McComas, forthcoming)� Merely having a potential COI is not necessarily synonymous with improper

behavior; rather, the conditions may favor the development of COI�

There has been increased attention to managing COI via the peer review process and the disclosure of financial interests, on understanding researcher perceptions of such conflicts, on understanding the influence of funding sources on research and professional norms of conduct, and on identifying opportunities for discussion among and training of researchers (McComas, forthcoming)� These ethical issues are not unique to nanotechnology� However, the emergence of nanotechnology research warrants additional consideration of the views of its researchers and the implications for addressing the ethical issues raised by the content of the survey�

Survey Implementation:A web survey was sent to NNIN users on July 6, 2011, by Cornell University’s Survey Research Institute� The survey took respondents, on average, eleven minutes to complete� Reminder emails were sent on July 10, 14, and 17� Overall, 656 complete responses were recorded out of a total of 2963 email addresses, yielding a response rate of 22�14%�

• socie

tal &

eth

ical is

sues • s

ocie

tal &

eth

ical is

sues • s

ocie

tal &

eth

ical is

sues •2011 NNIN REU Research Accomplishments Page 207

Results and Conclusions:We began by first examining the profile of survey respondents� A majority (54�2%) completed their program of study (Table 1)� A large majority (85�4%) depended on external funding for their research (Table 2), and 60�4% had not received advice on how to manage financial COI (Table 3).

Second, the belief that people important to them would want them to pay attention to the influence of funding (i.e., injunctive norms) and the perceived responsibility to do so emerged, after running a hierarchical regression, as significant predictors of researchers’ behavioral intentions to recognize COI and take action� The prominence of injunctive norms reveals that perceptions of actions of which other researchers may approve or disprove can have a significant effect on researcher behavior. Other variables did not achieve the same level of statistical significance.

In terms of perceived responsibility, there was variability among respondents in assigning responsibility for considering the influence of funding on research directions to themselves or their supervisors� Respondents were more likely to consider their supervisor as more responsible than themselves for considering the influence of funding sources and for identifying COI�

We also measured the acceptability of several scenarios researchers could encounter related to funding influences. Statistically significant differences in acceptability emerged when respondents answered what they would do, versus what they thought other researchers would do� In essence, they personally would be much more cautious in approaching issues of funding arrangements and COI compared to other researchers�

The data are subject to limitations� Despite being randomly drawn from the larger pool of NNIN users, a low survey response rate means that respondents may not necessarily represent all users� There may also be differences in respondent beliefs depending on the type of nanotechnology research conducted�

Future Work:The present study provides insight into researcher views on the perceived influence of funding sources and COI in nanotechnology research� Future implications include further exploration of the views measured in the survey, while also providing opportunities for further training for early career professionals on understanding and managing COI and similar considerations�

Acknowledgments:I would like to acknowledge the National Nanotechnology Infrastructure Network Research Experience for Under-graduates (NNIN REU) Program and the National Science Foundation for their support� I also extend a special thanks to Katherine McComas for her guidance and insight, and to Chris Clarke for mentoring me throughout this project� I would also like to thank Rob Ilic and Melanie-Claire Mallison for their support this summer, along with the CNF Staff and the other CNF REUs�

References:[1] McComas, K� (forthcoming)� Researcher views about funding

sources and conflicts of interest in nanotechnology. Science and Engineering Ethics� Published online February 2011�

[2] National Nanotechnology Initiative� (2010)� Funding opportunities� National nanotechnology coordination office. http://www.nano.gov/html/funding/home_funding�html� Accessed 14 May, 2010�

• s

oci

eta

l &

eth

ical

issues

• s

oci

eta

l &

eth

ical

issues •

soci

eta

l &

eth

ical

issues •


The Ethical, Legal and Societal Implications of Nanotechnology

Nina HwangChemistry, Rice University

NNIN REU Site: Colorado Nanofabrication Laboratory, University of Colorado, Boulder, CONNIN REU Principal Investigator: Professor Carl Mitcham, Liberal Arts and International Studies, Colorado School of MinesNNIN REU Mentor: Professor Lupita Montoya, Civil, Environmental and Architectural Engr., University of Colorado BoulderContact: [email protected], [email protected], [email protected]

Abstract:

Globally, billions of dollars [1] are being invested nanotechnology, which is expected to contribute to society in significant ways through advances in the fields of medicine [2], environment [3], and even world hunger [4]. The study presented in this report compared how the United States, the Netherlands and China are approaching the research and development (R&D) of nanotechnology, as well as the legislation, workplace safety and public education related to nanotechnology. These international comparisons are meant to increase understanding of the ethical, legal and societal implications (ELSI) of nanotechnology in the increasingly globalized society of today.

Introduction:

Since Richard Feynman first posited the concept of mole-cular manipulation in 1959 [5], interest in nanotechnology has exploded� In 1994, the National Science Foundation (NSF) established the National Nanofabrication User Net work (NNUN) [6], the precursor of the National Nanotechnology Infrastructure Network (NNIN)�

The National Nanotechnology Initiative [6] (NNI) was established in 2000 to coordinate the research efforts of 25 federal U�S� agencies, and in 2003, the NSF created NNIN [7] as part of NNI� Following the precedence of the Human Genomes project, the 21st Century Nanotechnology Act [8] (2003) was passed to ensure that “ethical, legal… and appropriate societal concerns” were addressed during federally funded projects�

In 2005, the Netherlands was already conducting advanced microtechnology research [4] and existing collaborations between research groups and industry easily transitioned to nanotechnology with the creation of NanoNed [9]� This government-funded consortium between universities and industry separated research projects into different flagships and included a mandatory risk analysis component� NanoNed and two microtechnology programs were combined in 2009 to create NanoNextNL [10], which encompasses over 100 research institutions and companies�

China became one of the only developing countries with aggressive nanotechnology R&D with the establishment of the National Center for Nanoscience and Technology [11] (NCNST) in 2001� ELSI components are not compulsory

for individual projects, but in 2006, NCNST established the Nanosafety Lab [12] to exclusively study the economic, environmental and social aspects of nanotechnology�

Methodology:The addressing of the ELSI of nanotechnology in the United States, the Netherlands, and China were compared across three categories: occupational safety, community outreach efforts, and historical context. Peer-reviewed scientific literature as well as published reports by federal agencies in each country were reviewed� As an initial step, this study looked at regulations for consumer and occupational safety� The next comparison reviewed community outreach and education efforts� This study also examined how technology adoption has been historically addressed by each nation using the regulation of the electronic industry’s waste as an example�

Results and Discussion:Occupational Safety. Nanoparticle regulation in all three countries was found to be similarly limited. Significant insufficiencies exist within the Toxic Substances Control Act (TSCA) in the U�S� as well as in the similar Registration, Evaluation, Authorization and Restriction of Chemicals (REACH) acts of the Netherlands and China� Under these acts, new nanoparticles are exempt from registration if the bulk (material) equivalents have already been approved�

• socie

tal &

eth

ical is

sues • s

ocie

tal &

eth

ical is

sues • s

ocie

tal &

eth

ical is

sues •2011 NNIN REU Research Accomplishments Page 209

Officials from the Nanotechnology Task Force [13], created by the US Food and Drug Administration, and from the Netherlands delegation to the European Commission [14] have released documents expressing concern over treating nanoparticles and bulk chemicals equally�

Each country has additional information for ensuring worker safety� The US National Research Council, for example, updated “Prudent Practices in the Lab” in 2011 [15] to include safe nanoparticle handling practices� The Netherlands released a similar document in 2008 [16]� In China, the National Nanotechnology Standardization Technical Committee [11] has been charged with establishing nanotechnology standards and research methodologies�

Community Outreach. Various educational programs have been launched and reflect the importance each country attributes to nanotechnology� NNIN alone reached over 20,000 people in 2010 through events like summer camps and lab tours [6]� In the Netherlands, where public dialogues are popular, the Committee for Societal Dialogue on Nanotechnology launched Nanopodium [17] in 2009 to act as a platform for citizens to share ideas and opinions about nanotechnology� In China, the Ministry of Science and Technology holds an annual “Science Week” with lectures introducing the basics of nanotechnology [13]�

Historical Context. A good parallel to nanotechnology regulation can be found in the electronics industry� The role the Precautionary Principle played in regulating the waste produced was a reflection of cultural differences, as only the Netherlands seemed to promote it� This principle states that a “lack of full scientific certainty” is not an excuse to neglect taking preventative safety measures [18]� This initial comparison may indicate the direction nanotechnology regulation will take�

In 1980, it was discovered that leaking industrial waste tanks owned by electronic manufacturers had heavily contaminated the groundwater in Silicon Valley, CA [19]� Because the chemicals had been grandfathered into the TSCA, the lack of regulation and information lead to the creation of 25 Superfund Sites in the immediate area� The Netherlands also had a vibrant electronics industry, including Royal Philips Electronics, but managed to avoid similar disasters through carefully planned pollution taxes and incentives for processing end-of-life products� The Restriction of Hazardous Substances (ROHs) directive was adopted in 2003 to restrict six toxic chemicals used in the industry [20], something the United States has yet to do� China modeled its own ROHs [20] after the Netherlands’, but weak local implementation created serious consequences� For example, the city of Guiyu is home to one of the largest electronic-waste dumpsites in the world and its residents are exposed to lethal levels of chemicals like cadmium [21]�

Conclusions:This study compared regulation of nanotechnology in the United States, the Netherlands, and China� Each country has predominantly approached nanotechnology according to their unique cultures, as exemplified by the educational programs and the credence given to the Precautionary Principle� Some actions, such as existing regulations and the push for nano-specific legislation, have been internationally influenced. With the advent of this globalized technology, understanding of policies in place must also be viewed from an international standpoint�

Acknowledgements:I would like to thank my mentor, Dr� Montoya, and Principal Investigator, Dr� Mitcham, for their guidance and insight� I would also like to thank Dr� Bart Van Zeghbroeck, Kendra Krueger, and the rest of the Colorado Nanofabrication Laboratory� Funding was provided by the National Nanotechnology Infrastructure Network Research Experience for Undergraduates (NNIN REU) Program and the National Science Foundation�

References:[1] Strategic Research Agenda-Nanotechnology (2008)�[2] Roco, M� Current Opinion in Biotechnology (2003)�[3] Theron, J� Critical Reviews in Microbiology (2008)�[4] Sastry, R� Food Policy (2011)�[5] Feynman� Engineering and Science (1960)�[6] NSF� Nanotechnology Research Directions (1999)�[7] NNIN� Annual Report-Abridged (2010)�[8] Pub� L� 108-153 (2003)�[9] NanoNed, Annual Report (2005)�[10] De Winter, J�; Overview of the Dutch Nanotechnology Landscape

(2011)�[11] Bai, Chunli� Science (2005)�[12] Zhao, Feng� Journal of Cleaner Production (2007)�[13] FDA� Nanotechnology Task Force Report (2007)�[14] Note 11626/11� Council of the European Union (2011)�[15] Prudent Practices in the Laboratory: Handling and Management of

Chemical Hazards (2011)�[16] Guidance Working Safely with Nanomaterials and Nanoproducts

(2011)�[17] CieMDN� Verantwoord Verder Met Nanotechnologie (2007)�[18] Communication from the Commission on the Precautionary

Principle (2000)�[19] SVTC� Regulating Emerging Technologies in Silicon Valley and

Beyond (2008)�[20] Hicks, C� Environmental Impact Assessment Review (2005)�[21] Leung, Anna� Environmental Science and Technology (2008)�

• i

ndex

• i

ndex •

index

• i

ndex •

index •

index

• i

ndex •

index •

index •




INDEX of reports by SITE

NNIN REU SITESASU NanoFab, Arizona State University, Tempe, AZ ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... . 22, 78, 82, 128, 132

Cornell NanoScale Science and Technology Facility, Cornell University, Ithaca, NY.. ... ... .. 4, 18, 38, 50, 74, 88, 120, 122, 130, 144, 178, 206

Nanotechnology Research Center, Georgia Institute of Technology, Atlanta, GA . ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...2, 6, 14, 36, 68, 90, 124

Center for Nanoscale Systems, Harvard University, Cambridge, MA ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... . 58, 92, 96, 118, 148

Howard Nanoscale Science and Engineering Facility, Howard University, Washington, DC ... ... ... ... ... ... ... ... ... ... ... ... .62, 110, 126, 172, 192

Penn State Nanofabrication Laboratory, The Pennsylvania State University, State College, PA ... ... ... ... ... ... ... ... ... ... ... ... 8, 44, 70, 102, 104

Stanford Nanofabrication Facility, Stanford University, Stanford, CA ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 72, 86, 156, 184, 204

Nanotech, University of California, Santa Barbara, CA ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... .12, 140, 142, 152, 174

Colorado Nanofabrication Laboratory, University of Colorado, Boulder, CO ... ... ... ... ... ... ... ... ... ... ... ... ... 66, 108, 116, 134, 154, 158, 208

Lurie Nanofabrication Facility, University of Michigan, Ann Arbor, MI .. ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...56, 60, 162, 190, 194, 198

Nanofabrication Center, University of Minnesota-Twin Cities, Minneapolis, MN ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 24, 30, 136, 180, 200

Microelectronics Research Center, The University of Texas, Austin, TX ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... . 64, 80, 84, 100, 146

Center for Nanotechnology, University of Washington, Seattle, WA. ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... .16, 34, 46, 48, 54, 188

Nano Research Facility, Washington University in St. Louis, St. Louis, MO .. ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... .26, 32, 42, 106, 138, 150

NNIN iREG SITESMicroelectronics Research Center, The University of Texas, Austin, TX ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...52

Colorado Nanofabrication Laboratory, University of Colorado, Boulder, CO ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... .160

Lurie Nanofabrication Facility, University of Michigan, Ann Arbor, MI .. ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... .176

NNIN iREU SITESNNIN iREU Site: Interuniversity Microelectronics Center (imec), Leuven, Belgium .. ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ..20, 112, 202

NNIN iREU Site: Centre Microélectronique de Provence, Ecole Nationale Supérieure des Mines de Saint Etienne, France . ... ... ... ... ... .. 76, 196

NNIN iREU Site: Institut Für Bio- Und Nanosysteme (IBN), Forschungszentrum, Jülich, Germany .. ... ... ... ... .. 10, 40, 164, 166, 168, 170, 186

NNIN iREU Site: Delft University of Technology (TU Delft), Netherlands . ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 28, 98, 182

The National Nanotechnology Infrastructure Network Research Experience for Undergraduates Program2011 NNIN REU Research Accomplishments are also online in secure PDF, at http://www.nnin.org/

• index • in

dex • in

dex • in

dex • in

dex • in

dex • in

dex • in

dex • in

dex •




INDEX of INTERNS, MENTORS, and PRINCIPAL INVESTIGATORS

AAbbas, Abdennour .................................................106Acevedo, Andrew.......................................2Achilefu, Samuel .....................................................42Agyemang, Malena .................................. 90Akenhead, Michael ................................. 138Akinwande, Deji ......................................................80Alkayyali, Amani .......................................4Almodovar, Noelia .....................................6Alphonse, Brittany ................................. 124Anugrah, Yoska .....................................................200Arean-Raines, Kyle .................................. 92Ashrafzadeh, Seyedshahin ...................... 140Austin, David ........................................................146Au, Yeung (Billy) .....................................................58Awschalom, David ................................................174Azhar, Ebraheem Ali ...............................................82Aziz, Michael ......................................................... 118

BBabensee, Julia .........................................................2Banerjee, Sanjay .....................................................64Bao, Zhenan ............................................................86Bargatin, Igor .......................................................184Barton, Robert A. ..................................................130Bayer, Karl ............................................. 94Beekun, Issa ......................................... 142Belkin, Mikhail ......................................................146Bellamkonda, Ravi ..................................................14Benton, Brian ....................................... 164Biaou, Carlos Koladele ............................. 56Blain Christen, Jennifer ..........................................78Bleecker, Joan .........................................................48Boneparte, Antanica ................................ 96Bovington, Jock .....................................................142Bowers, John .................................................142, 152Boyan, Barbara .........................................................6Brockhage, Rachel ................................. 206Brunson, Mark ...................................... 166Bryant, Alex ......................................... 144Bunch, J. Scott ......................................................134Buscema, Michele .................................................182

CCampbell, Ian........................................................ 116Cantley, Lauren ....................................... 98Cardenas, Jaime ...................................................144Chae, Junseok .......................................................128Chang, Julie..............................................8Chang, Ting .............................................................60Chang, Ting Chia ................................... 146Chase, Steven ......................................... 10Chen, I-Tzu..............................................................38Chen, Kevin .......................................... 168Choi, Seokheun .....................................................128Choi, Yoonsu ...........................................................14Chow, Clara ............................................ 40Christen, Jennifer Blain ..........................................78

Christle, David ......................................................174Chung, Brian T. ..................................... 186Clarke, Christopher ..............................................206Clark, Parker .......................................... 58Coldren, Larry ......................................................140Connell, Zachary ................................... 170Copic, Davor .........................................................190Craigg, Hélène ......................................................198Craighead, Harold ..................................................18Crouch, Courtney ................................... 188

DDai, Hongjie............................................................72Dang, Audrey .......................................... 42Dauskardt, Reinhold H. ..........................................94Dekker, Cees ............................................................28Deotare, Parag .....................................................148Diasio, Matthew .................................... 190Diep, Vinh .............................................. 12Dittmann, Regina ..................................................170Dong, Mark .......................................... 148Dorfman, Kevin .....................................................180Dudani, Jaideep S. ................................... 14Dunston, Tiffany ................................... 192

EEllerbee, Audrey ....................................................156Engstrom, James .....................................................50Esteves, Giovanni .................................... 44

FFedorov, Andrei .......................................................68Ferrer, Domingo .....................................................64Forest, Craig ...........................................................36Forman, Darren ....................................................158Frielinghaus, Robert .............................................166Fullerton, Elizabeth ............................... 100

GGao, Xiaohu ..........................................................188Georgiev, Julie A. .................................. 102Gerhardt, Michael .................................. 104Gerhardt, Rosario A. ...............................................90Gilbertson, Jennifer ................................. 46Gillilan, Richard .......................................................4Gobert, Brendon Lee .............................. 126Goosens, Stijn .........................................................98Gopinath, Juliet ....................................................154Gordon, Roy ......................................................58, 92Grab, Jennifer .......................................................178Green, Craig ...........................................................68Griffin, Emily .......................................... 60Griffin, James A. .....................................................62Gruev, Viktor .........................................................150Gwinn, Tom ...........................................................156

HHaggerty, Andreas M. ............................... 62Halim, Abigail ....................................... 106Hams, Nicole ........................................ 128Harris, Gary L. ......................... 62, 96, 110, 126, 172Hart, Anastasios John ...................................176, 190Hart, Jeffrey ......................................... 108Haygood, Ian ..........................................................66Haynes, Christy L. ..................................................24He, Kai .................................................. 64Hendricks, Lisa Anne .............................. 194Herro, Alicia ......................................... 110Hertzberg, Jared ...................................................120Heugel, Nicholas ................................... 150Hoffman, Emily ..................................... 112Hollmann, Eugen ..................................................186Holmberg, Vincent C. ..............................................52Hoogeboom-Vlijm, Rifka ........................................28Hosseinzadegan, Hadi ............................................88Hotaling, Nathan ......................................................2Howe, Roger T. .............................................184, 204Huang, Kevin .......................................... 16Hurst, Hilary........................................... 66Hu, Wenbing ............................................................56Hu, Xiaoge ............................................................188Hwang, Nina ......................................... 208

IIlufoye, Israel ......................................... 68

JJiang, Shaoyi ...........................................................54Jones, Kimberly L. ................................................192

KKan, Edwin..............................................................74Kaytanli, Bugra .......................................................12Keller, Sarah ...........................................................48Kelly, Christopher ...................................................18Kendrick, Chito E. ................................................102Kim, Donghyuk .......................................................24Kiok, Matthew......................................... 18Kirk, James T. .........................................................16Koester, Steven J. ..................................................200Korgel, Brian A. ..............................................52, 100Kowalski, Benjamin ..............................................158Kuang, Jia .............................................. 70Kuk, Won Jun ........................................ 172Ku, Pei-Cheng .......................................................162Kuruvilla, Sibu ........................................ 20Kutzner, Rolf..........................................................186

LLabanowski, Dominic E. .......................... 174Lal, Amit ..................................................................88Lambdin, Olivia ....................................... 22Larson-Smith, Kjersta .............................................46

• i

ndex

• i

ndex •

index

• i

ndex •

index •

index

• i

ndex •

index •

index •


Lau, Gawain ......................................... 116Laux, Leah ............................................. 24Lee, Chien-Chung .........................................108, 160Lee, Daniel ............................................................144Lee, Jae .................................................................184Lee, W. Scott ..........................................................204Leleux, Pierre ........................................................196Lenser, Christian ...................................................170Lepsa, Mihail ........................................................168Levin, Scott M. ........................................................70Li, Lin ....................................................................136Li, Max .................................................. 26Li, Melissa ...............................................................36Liang, Hongliang ..................................... 72Liddell Watson, Chekesha .....................................122Lindquist, Nathan....................................................30Lipson, Michal ......................................................144Liu, Xinghui ..........................................................134Lloyd, Travis ......................................... 152Loncar, Marko .......................................................148Lurie, Kristen ........................................................156Lu, Wei ....................................................................60

MMagnuson, Carl W. .................................................84Malliaras, George G. ......................................76, 196Mallin, David ........................................ 154Mateus, Maria Veronica ............................. 74Matsuoka, Yuki ..................................... 176Mayer, Dirk .....................................................40, 164Mayer, Theresa S. ....................................................70McComas, Katherine ............................................206McGuinness, Morgan ................................ 48McLeod, Robert ....................................................158Meffert, Simone .......................................................10Mellnik, Alex .........................................................178Mendez, Joshua ..................................... 178Meyer, Carola .......................................................166Mirts, Evan............................................. 28Mitcham, Carl .......................................................208Montoya, Lupita ....................................................208

NNahm, Rambert .......................................................50Neiman, Nicolas ....................................................104Nergiz, Saide Zeynep ............................................106Neves, Hercules P. ........................................... 20, 112Niederriter, Robert ................................................154Nielsen, Soeren .........................................................4Noble, Jade M. ........................................ 50Novoa, Fernando ....................................................94

OO’Connell, Fiona .................................... 196Ogata, Yoichi .......................................... 52Oh, Sang-Hyun ........................................................30Oiler, Jonathon......................................................132Olson, Daniel ........................................................180Otelaja, Obafemi ...................................................120Otto, Lauren ........................................... 30Ozdemir, Sahin Kaya.............................................138

PPark, Jung Hwa ........................................................6Parpia, Jeevak ......................................................130Pelaez, III, Francisco .............................. 180Peterson, Becky .....................................................198Peterson, Rebecca L. ..............................................56Piels, Molly ...........................................................152Piercy, Brandon ..................................... 118Pillers, Michelle ...................................... 76Pipes, Meagan ...................................... 156Pletcher, Kendall ..................................... 78Podmayer, Julia ....................................... 54Pozzo, Danilo C. .....................................................46Provine, J. .............................................................204Puckett, Ernest ....................................... 32Pui, David .............................................................136Pyatkov, Feliks ......................................................164

RRadisic, Aleksander............................................... 112Raebig, Andrew .................................... 198Rais-Zadeh, Mina..................................................194Ralph, Daniel C. ...................................................178Ramamoorthy, Malaisamy ....................................192Ramos, James ..........................................................22Rand, Barry ...........................................................202Ratner, Daniel .........................................................16Recht, Daniel ........................................................ 118Redwing, Joan M. .................................................102Rege, Kaushal .........................................................22Riley, Erin .............................................................122Rioux, Robert ..........................................................44Robinson, Bethany................................... 80Robinson, Richard .................................................120Rognstad, Laurel ..................................... 34Rose, William ........................................................126Ruoff, Rodney S. ......................................................84

SSakakibara, Reyu ................................... 130Sanaur, Sébastien ....................................................76Sanetra, Nils............................................................40Savikhin, Victoria .................................. 120Scheideler, William .................................. 82Schibli, Thomas .............................................108, 160Schiltz, Drew D...................................... 158Schliep, Karl ......................................... 132Schoenwald, Kipp .................................................124Seaman, Laura ........................................ 36Sheehan, Nathanial ................................ 200Singamaneni, Srikanth ..........................................106Sinsermsuksakul, Prasert ........................................92Smalley, Joseph .................................... 202Smith, Wes .............................................................204Soni, Gautam ..........................................................28Sonner, Zachary ...................................... 38Spradling, Claire ..................................... 84Srinivasan, Akhil .....................................................14Steinhagan, Chet ...................................................100Stoykovich, Mark .................................................. 116Stroock, Abraham D. ...............................................38Sturgis, Nick ............................................................44Sulchek, Todd ........................................................124Suralik, Kelly ........................................ 204Suzuki, Seiya ........................................ 160Szpunar, Mariah .................................... 134

TTang, Rui .................................................................42Tao, Li .....................................................................80Tawfick, Sameh ......................................................176Taylor, Crawford ........................................... 110, 172Tee, Benjamin C.K. .................................................86Teng, Chu-hsiang ..................................................162Thakar, Vikram ......................................................194Tien, Kevin ............................................. 86Todd, Cassandra ...................................... 88Trolier-McKinstry, Susan ......................................104Tung, Lieh-Ting (Adrian) ........................................74

VValentine, Megan T..................................................12van der Zant, Herre ...............................................182Van Zeghbroeck, Bart ..............................................66Vandersypen, Lieven ...............................................98Vanlerberghe, Filip .................................................20Voroshazi, Eszter ...................................................202Vulis, Daryl I. ....................................... 122

WWallace, Margeaux ................................. 182Wang, Lihong V. ......................................................32Wang, Yucai .............................................................26Weiss, Leah .......................................... 184Welch, David ...........................................................78Windmuller, Laura ................................. 136Wördenweber, Roger .............................................186Wu, Diana ............................................ 112Wu, Justin ................................................................72

XXia, Younan .............................................................26Xu, Jiajie .................................................................34

YYang, Lan ..............................................................138Yang, Wei .................................................................54Yao, Junjie ...............................................................32Yeager, Charles .....................................................104Yu, Hongbin .............................................................82Yu, Hongyu ............................................................132Yu, Qiuming .............................................................34

ZZhang, Jin ............................................ 162Zhang, Wenyu ..........................................................50Zheng, Siyang............................................................8Zheng, Yan .............................................................140Zhou, Mingda ............................................................8

NNIN REU Program Summer 2011

Documents

Transcript of NNIN REU Program Summer 2011