Modifying Surface Energy of Graphene via Plasma-Based ChemicalFunctionalization to Tune Thermal and Electrical Transport at MetalInterfacesBrian M. Foley,*,† Sandra C. Hernandez,‡ John C. Duda,† Jeremy T. Robinson,§ Scott G. Walton,*,‡

and Patrick E. Hopkins*,†

†Department of Mechanical and Aerospace Engineering, University of Virginia, Charlottesville, Virginia 22904, United States‡Plasma Physics Division and §Electronics Science and Technology Division, Naval Research Laboratory, Washington, DC 20375,United States

*S Supporting Information

ABSTRACT: The high mobility exhibited by both supportedand suspended graphene, as well as its large in-plane thermalconductivity, has generated much excitement across a variety ofapplications. As exciting as these properties are, one of theprincipal issues inhibiting the development of graphenetechnologies pertains to difficulties in engineering high-qualitymetal contacts on graphene. As device dimensions decrease, thethermal and electrical resistance at the metal/graphene interfaceplays a dominant role in degrading overall performance. Herewe demonstrate the use of a low energy, electron-beam plasmato functionalize graphene with oxygen, fluorine, and nitrogengroups, as a method to tune the thermal and electrical transportproperties across gold-single layer graphene (Au/SLG)interfaces. We find that while oxygen and nitrogen groups improve the thermal boundary conductance (hK) at the interface,their presence impairs electrical transport leading to increased contact resistance (ρC). Conversely, functionalization with fluorinehas no impact on hK, yet ρC decreases with increasing coverage densities. These findings indicate exciting possibilities usingplasma-based chemical functionalization to tailor the thermal and electrical transport properties of metal/2D material contacts.

KEYWORDS: Graphene, contacts, functionalization, thermal boundary conductance, contact resistivity

The remarkable electrical1−3 and thermal4−6 properties ofsingle-layer graphene (SLG) have had a staggering impact

on the research landscape over the past decade. Groups acrossthe world in engineering, basic sciences, and medicine havecharacterized and attempted to exploit these enhancedproperties with the goal of revolutionizing their respectivefields. One example is the field of graphene electronics wherean incredible amount of research has focused on transistors,sensors, and optoelectronics based on this material. While greatprogress has been made toward integrating graphene into manyof these applications, there remain several practical issues thathave hindered the large-scale development of graphenetechnologies and led to a migration away from this materialto other two-dimensional (2D) structures such as MoS2.One such issue relates to difficulties associated with

engineering high-quality metal contacts on graphene.7−9 Forfield effect transistors (FETs), there is a critical channel lengthbelow which the metal/graphene interface is the dominantresistance in the device, thereby limiting the on-current of thetransistor.7,9 In order to scale-down graphene FETs for highdevice-densities, further minimization of the contact resistivity(ρC) is required. While a great deal of work has been carried

out in refs 8−16 to reduce ρC via the choice of metal andprocessing conditions (i.e., resist removal and thermalannealing postpatterning), it is estimated that ρC must decreaseby at least another order of magnitude before graphene FETswith channel lengths ⩽1 μm can achieve the requiredperformance characteristics.7,8

Additionally, there are a multitude of device applicationsrelated to high-power radio frequency (RF) and microwave(MW) electronics where larger device geometries (length scalesof 1−10 μm) dictate that larger values of ρC may be acceptable.Several groups17−20 have demonstrated graphene transistorswith cutoff frequencies greater than 10 GHz, indicating theirpotential as a possible alternative to GaAs, InP, and GaN inMW amplifiers, mixers, detectors, etc. While these results arepromising, it has also been shown that Joule heating can have aprofound effect on the saturation current and transientperformance of graphene devices, potentially limiting theirapplicability for high-power applications.21−24 Because of this,

Received: January 29, 2015Revised: June 29, 2015Published: June 30, 2015

Letter

pubs.acs.org/NanoLett

© 2015 American Chemical Society 4876 DOI: 10.1021/acs.nanolett.5b00381Nano Lett. 2015, 15, 4876−4882

the thermal transport between graphene and adjacent layers(i.e., substrate, gate dielectric, contacts, and so forth) are criticalparameters that must be well understood to ensure reliable andrepeatable performance of future devices.Without addressing these issues through development of

novel, useful methods to tune thermal and electrical transportat these interfaces, the immense potential of graphene andother 2D materials may never be fully realized. This workreports measurements of the thermal and electrical transportproperties of chemically functionalized gold-single layergraphene (Au/SLG) interfaces. The cross-plane thermaltransport is measured via time domain thermoreflectance(TDTR) and the in-plane electrical transport is measured fortwo-terminal devices using a Keithley source-measure unit(SMU). Both properties are found to vary as a function of thepercent coverage of functional groups on the graphene,providing an opportunity to tune the interfacial transportproperties and subsequently offer increased flexibility for futuredevice design.Single layer graphene was grown via low-pressure chemical

vapor deposition (CVD) on copper foil and transferred to 300nm SiO2/Si substrates as outlined in refs 25−27 and detailed inthe Supporting Information. Following transfer, X-ray photo-

electron spectroscopy (K-Alpha XPS) and micro-Ramanspectroscopy were performed to confirm minimal residualand structural quality. Physical masks were affixed to thesamples to define localized processing regions. These maskswere made via chemical etching of 25 μm thick molybdenumstock by Towne Engineering, Inc. and consisted of 16 TLMpatterns per mask (see Figure 1 for pattern dimensions). Thesamples were then individually exposed to electron beamgenerated plasmas to introduce chemical moieties to theexposed areas.27−29 While various plasma sources have beenused in the synthesis and modification of graphene,30,31

electron beam generated plasmas are well-suited for chemicalfunctionalization as they are capable of delivering a flux ofreactive species while limiting the ion kinetic energies to a fewelectronvolts.32 Thus, they provide the ability to chemicallymodify the graphene without etching or introducing unwantedphysical changes, which is an attribute that makes these plasmasystems particularly attractive for the large scale processing ofvery thin or 2D materials. In particular, they have been used tochemically functionalize CVD-derived graphene on copper,28

CVD-derived graphene that has been transferred to Si/SiO2substrates,27−29 and epitaxial graphene.33 Fine control overcoverage is achieved by increasing the operating pressure,

Figure 1. Images (a−d) outline the patterning process used to fabricate functionalized contacts. (a) CVD-grown graphene previously transferredonto 300 nm of SiO2 on p-type silicon is covered with a shadow mask (b) consisting of 16 identical TLM patterns, chemically etched into 25 μmthick molybdenum stock (Towne Engineering, Inc.). The mask is affixed to the sample with kapton tape and loaded into the chamber for plasmafunctionalization (c) with a variety of adsorbates in the areas (red) that remain exposed. Following functionalization, the sample is loaded into ametal evaporator where metals are deposited (d) onto the same areas that were functionalized previously. (e) Optical plan-view image of the TLMgeometry used in this study after fabrication, including the measured distances between pads. (f) Schematic of the plan-view image in (e) providingan in-plane illustration of the final structure showing the functional groups underneath the gold pads.

Figure 2. (a) X-ray photoelectron spectra of the core level C1s of (from bottom to top) as-transferred, nitrogen, fluorine, and oxygen functionalizedgraphene films at comparable surface coverages. The envelope line (red) is dashed and the raw data (black) is solid. (b) Raman spectra of (frombottom to top) as-transferred, nitrogen, fluorine, and oxygen functionalized graphene films at comparable surface coverages.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00381Nano Lett. 2015, 15, 4876−4882

4877

which serves to increase the flux of reactive species delivered tothe graphene surface.33,34 Patterning via physical masks ratherthan by lithographic processes was chosen to avoid theintroduction of unwanted edge defects and minimize the affectof any residual photoresist on the measurements made in thestudy.10

Witness samples of pristine graphene from the same batch ofCVD grown material were functionalized along with themasked samples to enable representative characterization of thegraphene postfunctionalization. Figure 2 depicts the X-rayphotoelectron spectroscopy (XPS) and Raman signaturesacquired from graphene films that were functionalized with avariety of chemical functional groups at similar coverages.Figure 2a depicts the XPS spectra of as grown and chemicallyfunctionalized graphene films with different functional groupsat similar surface coverages. The high-resolution C1s spectraconfirms the introduction of oxygen, nitrogen, and fluorinefunctionalities. Deconvolution of the C1s attributes thesubcomponents of pristine graphene to be sp2 C−C (284.4eV) bonding with a slight asymmetry, sp3 C−C (285.3 eV), anda small degree of oxygen at 286.3 eV associated with thepresence of native oxygen functionalities likely present on smallartifacts or native surface defects. After functionalization, theXPS spectra show a variety of different chemical moieties foreach functional group type. The nitrogen functionalizedgraphene film contained 13.2 at % total nitrogen, mostly inthe form of N−sp2−C and N−sp3−C bonds, located at 285.5and 287.6 eV respectively. An additional peak is observed at288.8 eV, which can be a combined contribution of CN andCO functionalities. Tracking of the C1s oxygen function-alities is challenging since C−O and CO have overlappingbinding energy assignments to those of nitrogen functionalities.Because of the overwhelming oxygen contribution from thesilicon oxide layer on the O1s spectrum, the exact contributionof oxygen functionalities is difficult to determine. However, itshould be noted that complementary to the C1s, the N1s wasalso used to cross check peak identifications. The N1s spectrumshows “pyridinic” (N with two carbon bonds) component at399.5 eV and “pyrrolic” bonding configuration (N incorporatedin five-membered heterocyclic ring) at 400.4 eV. Takentogether, we can deduce that the bonding structures of theC−N bonds most likely correspond to N−sp2−C and N−sp3−C bonds, respectively.29 The fluorinated sample contained atotal of 16.9 at % total fluorine, which is present in the form ofC−CF (286.1 eV), C−F (288.6 eV), C−F2 (290.8 eV), and C−F3 (292.82 eV). Finally the oxygenated graphene samplecontained 14.6 at % oxygen in the forms of C−O (285.8 eV),CO (287.3 eV), and O−CO (288.7 eV) species.Figure 2b depicts the characteristic Raman spectra of the as-

grown graphene (black line) with the G (1570 cm−1) and 2D(2700 cm−1) graphene signature peaks. Conjugation of the sixmember ring structure of graphene can become disrupted whenfunctional groups are introduced to the carbon structure due toelectron sharing or sp3-bond formation leading to sp3

hybridization evident by the presence of a D (1350 cm−1)peak while reducing the G and 2D peak intensities. Althoughthe Raman spectra is for samples with comparable surfacecoverages, there are clear differences in the Raman signaturesattributed to the differences in bonding types and theirinfluence over the resulting structural configuration ofgraphene. Oxygen, for example, forms a covalent doublebond at a faster rate and more readily than all the otherfunctional groups at the same surface coverage. A covalent

double bond of such sort would severely impact the structuralintegrity of the graphene lattice changing the C−C−C anglefrom 120° to close to 115°. The changes of the structuralintegrity of the graphene lattice to accommodate the newcovalent double bond to an oxygen causes a “puckering” effectof the neighboring carbons, making that geometrical structuresp3 in nature. These structural changes can be tracked byRaman, and their essence captured by the peak intensities ofthe Raman signatures.35,36 Fluorine functionalities, however,can maintain sp2 hybridization at low F coverage, while sp3

configuration is expected at larger F coverages. Ramanspectroscopy is very sensitive to structural changes, thereforeat comparable surface coverage, oxygen functionalities willshow more pronounced D peaks because they perturb the sp2

hybridization more drastically than nitrogen or fluorinefunctionalities.36 The domain size La can be estimated by theD/G ratio as described by Tuinstra and Koenig37 that showsthat initial pristine graphene domain size was 100 nm and afterfuntionalization with ≈6 atom % adatoms the domain sizesdecreased to 1.95, 2.3, and 7.2 nm for oxygen, nitrogen, andfluorine functionalities, respectively. At even higher surfacecoverage ≈10 atom % adatoms, the calculated domain sizeddipped further to 1.9, 2.2, and 2.9 nm for oxygen, nitrogen, andfluorine, respectively. Combined, the Raman and XPS spectraof functionalized graphenes demonstrate that even atcomparable surface coverages, the resultant atomically thinfunctionalized surface differs dramatically both structurally andchemically from each other, demonstrating the impact of a fewadsorbates on the surface of graphene.Two separate sets of samples were fabricated due to certain

requirements regarding the thickness of the Au layer for thethermal and electrical parts of the study. These sets of sampleswere functionalized at different times and the depositionconditions were replicated as closely as possible. For thethermal studies, 90 nm of Au was deposited on thefunctionalized SLG samples for measurement via TDTR. Forthe electrical studies, 500 nm of Au was deposited on thefunctionalized samples to facilitate probing of the contact padswith Kelvin (dual-tip) probes. Such a thick Au layer wasrequired because of difficulties encountered while trying toprobe the 90 nm pads in the thermal study. Because the weakadhesion between the Au/SLG and SLG/SiO2 interfaces, it isdifficult to make electrical contact with the pads without deeplypenetrating the Au film and perturbing the interface. While itwas nearly impossible to consistently probe the 90 nm pads, wewere able to make contact to the 500 nm Au pads much moreeffectively.The samples fabricated for the thermal part of the study were

measured using time-domain thermoreflectance (TDTR).38

TDTR is a noncontact, optical pump−probe technique utilizingultrafast subpicosecond laser pulses to characterize the thermaltransport properties of nanostructured materials and systems.The details of our measurement system and the analysis used toextract the thermal properties of interest from the data areprovided in the Supporting Information. Figure 3 shows thethermal boundary conductance, hK, across the metal/SLGinterface as a function of surface coverage of the variousfunctional groups on graphene (see Table 1 for exact values). Inaddition, hK was measured across the Au/SLG and Au/Ti/SLGinterfaces with no added functional groups between the metaland graphene. The measured value of 23 ± 0.7 MW m−2 K−1

for the Au/SLG and Au/Ti/SLG interfaces agrees well withprevious work,39 and these results indicate that the inclusion of

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00381Nano Lett. 2015, 15, 4876−4882

4878

the Ti layer appears to have little to no effect on the thermalboundary conductance. Typically, Ti is used as an adhesionlayer as it is more reactive than Au with a wide array ofmaterials. Therefore, following other experimental works on theeffect of interfacial bonding on thermal boundary conduc-tance,40−42 one would expect a larger thermal boundaryconductance with an increase in bond strength. The resultshere likely indicate that the graphene surface reactivity isdominating the bonding environment and not the reactivity ofthe metal film, which we further assert below.The type of functional group present at the Au/SLG

interface clearly impacts the measured thermal boundaryconductance in different ways, and the observed differencesbetween the chemical moieties can be explained by consideringthe individual effects of the adsorbates on the surface energy ofthe SLG. In the case of fluorinated graphene, the formation ofionic C−F bonds with the out-of-plane π-orbital of thegraphene have been shown to decrease the surface energy ofthe functionalized-SLG sheet.27 This decrease in surface energywill negatively impact the adhesion between the metal andgraphene. While the strong C−F bond should be beneficial

from a phononic point of view on a local scale, the adhesionbetween the fluorine-functionalized SLG and the Au layerabove remains weak, resulting in no net enhancement of hKacross the interface. By comparison, plasma functionalizationwith oxygen27 and nitrogen29 have both been shown to increasethe surface energy, thereby making the functionalized SLGsheet more reactive. This is further corroborated by workshowing that the surface energy of graphene oxide is larger thanthat of graphene alone,43,44 resulting in increased reactivity.Here, the addition of oxygen or nitrogen functional groupsincreases the surface reactivity and thus improves adhesion ofthe metal contact to the functionalized graphene, resulting in anincreased hK.We offered a similar explanation in our previous work42 to

describe the difference between oxygen and hydrogenfunctionalization at an aluminum/SLG interface, where weargued that oxygen provides a stronger coupling betweengraphene and Al via the formation of Al−O bonds, whilehydrogen leaves the graphene surface inert. In that work, a100% increase in hK over the value for Al/SLG was observed atan Al/O/SLG interface with 25 atom % coverage, while an Al/H/SLG interface of similar coverage exhibited a slight decreasein hK. While the increase associated with oxygen in the presentwork is more modest, the enhancement in hK suggests that thetransport across the metal/graphene interface can be heavilyinfluenced by the reactivity of the graphene surface, particularlyin the case of contacts made with less reactive metals such asAu. This suggests that when designing a metal/grapheneinterface for optimal thermal transport, it is important toconsider the surface reactivity of the graphene along with thechoice of metal.To complement the thermal measurements, the width-

normalized electrical contact resistance (ρC) at these function-alized metal/graphene interfaces were measured via the transferlength method (TLM). TLM is a widely used technique formeasuring ρC at a metal/semiconductor interface and detailsregarding our measurement equipment and the analysis of thedata are provided in the Supporting Information. It should benoted that 10 TLM structures like that depicted in Figure 1ewere measured for each sample in the study and our reportedvalues of ρC are determined from the average resistance over all10 structures. Figure 4 plots ρC across the metal/SLG interfaceas a function of surface coverage for a similar set of samples tothose investigated in the thermal part of the study. The value ofρC for both Au/SLG and Au/Ti/SLG interfaces is essentiallythe same (see Table 2 for exact values). This is consistent withother measurements8,13 of the contact resistance betweengraphene and these metals, and highlights an auxiliaryconclusion from this work; the inclusion of a titanium wettinglayer between Au and SLG improves neither the electrical, northe thermal transport across the metal/SLG interface. In thecase of oxygenated graphene, ρC increases by more than anorder of magnitude with increasing coverage. For fluorinatedsamples, following an initial increase over the nonfunctionalizedcase at the lowest coverage, ρC actually decreases withincreasing coverage and approaches the values observed inthe nonfunctionalized contact case (Au/SLG and Au/Ti/SLG).The differing trends in ρC between functionalization with

oxygen and fluorine are indicative of the potential to engineerelectrical contacts on graphene via the introduction of thesechemical moieties at the interfaces. Central works, bothcomputational45−47 and experimental,9,11,48,49 on metal/gra-phene contacts have shown that the choice of metal used for

Figure 3. Metal/SLG thermal boundary conductance (hK) versusatomic percent coverage of various adsorbates on single layergraphene. Coverages of approximately 13−15 atom % oxygen ornitrogen result in a 25−40% enhancement in hK, while functionaliza-tion with fluorine at coverages ranging from 11.4−22.1% yields noenhancement of hK compared to the Au/SLG baseline case.

Table 1. Atomic Percent Coverage of Functional Groups (ifApplicable) and the Resulting Thermal BoundaryConductance (hK) for the Samples Investigated in theThermal Part of the Study

interfacial layer coverage (atom %) hk (MW m−2 K−1)

none (Au/SLG) 22.91titanium (Au/Ti/SLG) 100.00 23.29oxygen (Au/O/SLG) 4.40 24.33

14.60 32.41fluorine (Au/F/SLG) 11.40 23.61

16.90 22.1022.10 24.33

nitrogen (Au/N/SLG) 1.00 21.4113.20 28.98

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00381Nano Lett. 2015, 15, 4876−4882

4879

the contact has a significant impact on their transportcharacteristics. The doping of graphene directly underneaththe metal,45,46 as well as in regions adjacent to the contacts47−49

due to 2D confinement can impede charge flow and result inundesirable behavior, such as contact resistance that can vary asa function of applied gate bias.7,9,16 Reference 9 suggests thatthe intentional doping of the graphene underneath the metalcontacts through nonmetal adsorbates may lead to an increasednumber of conduction modes, thereby enhancing both thecross-plane and in-plane conduction pathways and lowering thecontact resistance. To this point, it is instructive to furtherexplore the effects of adsorbates on the electronic structure ofgraphene to gain insight into their potential impact on ρC.Calculations of the density of states (DOS) for oxy-

genated50,51 and fluorinated36 graphene highlight an importantdifference between the use of these two adsorbates.Functionalization with either adsorbate induces a band gap inthe electronic structure that widens with increasing coverage.

However, in the oxygen case the Fermi level remains positionedwithin the center of the band gap, and the resulting graphene isbest characterized as being in a semiconducting state. Anincrease in the sheet resistance (RS) is observed in theoxygenated samples, which is likely due to the doping of thegraphene adjacent to the metal contacts that occurred whilefunctionalizing the graphene via physical masks (see inset ofFigure 4 and the discussion in the Supporting Information).The positioning of the Fermi Level and reduction inmobility52−54 due to adsorbates are two of the factors thatcan explain the 1−2 order of magnitude increase in RS.Functionalization with fluorine is a bit different in that ascoverage increases and the band gap widens, the Fermi Levelmoves into the valence band and the graphene remainssemimetallic.36 This downshift in the Fermi Level is attributedto the interaction of the π-orbitals of the carbon atoms with thep-orbitals of the fluorine adsorbates and the electron local-ization that occurs in the C−F bond. While the fluorineadsorbates do lead to increased scattering and a correspondingreduction in mobility,36 the increase in RS for fluorinatedgraphene is not as dramatic as the oxygenated case due to theposition of the Fermi level in the valence band.The impact of these adsorbates on the transport properties of

functionalized Au/SLG contacts is two-fold. First, thedistortion of the electronic structure and the shift in positionof the Fermi level of the functionalized graphene will have asignificant impact on the nature of the metal/grapheneinteraction outlined in refs 45 and 46, as well as thetransmission probability across the dipole created at theinterface.9 The resulting work function of a functionalizedmetal/graphene contact will be different from that of anonfunctionalized contact comprised of the same metal witha difference of up to ≈1 eV possible, depending upon thedegree to which the adsorbates screen the chemical interactionbetween the metal and graphene.45,46 Second, the difference inwork function between the metal-covered graphene and thegraphene channel adjacent to the contact will affect the in-planetransport of carriers between the contact and channel regions.Some of the factors that contribute to such affects includecharge density profiles that exhibit pn junctions within thetransport path47 and screening potentials that can extend forhundreds of nanometers into the channel.11,47,49

On the basis of these two effects, it would be reasonable toassume that the inclusion of adsorbates would serve to furtherincrease ρC as we observe with oxygenated contacts. On theother hand, the fluorinated contacts exhibit a decreasing trendin ρC with increasing surface coverage, possibly due to theaforementioned doping of the graphene adjacent to the metalcontacts. Reference 11 makes an important point that ρC canchange by a factor of 10 depending on the carrier type andconcentration present in the region adjacent to the metalcontact. In that work, they state that such an effect is normallyobserved for high-quality contacts and is attributed to thestrong in-plane doping that extends well into the adjacentgraphene. In our fluorinated contacts, the adsorbates at theedges of the contacts that were not covered in the metallizationstep have strongly p-doped the graphene adjacent to thecontacts, effectively increasing the number of conductionmodes available to carriers between the contact and thechannel. Therefore, if the decrease that we observe in ρC withincreasing fluorine coverage can be attributed to an affectsimilar to that presented in ref 11 it would suggest that it maybe possible to further decrease ρC at a fluorinated Au/SLG

Figure 4. Width-normalized contact resistivity (ρC) versus atomicpercent coverage of various adsorbates on single layer graphene. Theintroduction of oxygen or nitrogen at coverages greater than 5 atom %causes ρC to increase dramatically. Conversely, while low coverages of≈5 atom % fluorine causes an increase in ρC, coverages greater than 5atom % cause ρC to decrease with increasing coverage. The dashedlines represent contact resistivity values reported for Au/Ti/SLG8 andpalladium/SLG,9 respectively. Inset: Sheet resistance (RS) of thegraphene between the TLM pads due to doping of the grapheneadjacent to the metal contacts that occurred while functionalizing thegraphene via physical masks.

Table 2. Atomic Percent Coverage of Functional Groups (ifApplicable), Sheet Resistance (RS) of the Graphene betweenthe Contacts, and Contact Resistivity (ρC) for the SamplesInvestigated in the Electrical Part of the Study

interfacial layer coverage (atom %) RS (Ω/sq) ρC (kΩ μm)

none (Au/SLG) 110.75 7.04titanium (Au/Ti/SLG) 100.00 117.15 6.11oxygen (Au/O/SLG) 4.40 148.35 5.44

14.60 7945.50 166.0019.00 21926.50 235.00

fluorine (Au/F/SLG) 5.50 422.65 24.4010.91 200.55 12.5012.26 454.55 8.15

nitrogen (Au/N/SLG) 9.60 256.60 20.74

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00381Nano Lett. 2015, 15, 4876−4882

4880

contact to values below that of the Au/SLG control case.Clearly, additional experiments are necessary to validate theseassertions, but the observed results have significant implicationsin the engineering of graphene and other two-dimensionalmaterial-based devices.The presence of chemical moieties between graphene and

gold contacts is shown to affect the thermal and electricaltransport properties compared to unfunctionalized Au/graphene contacts. Concerning thermal transport across anAu/SLG interface, we have shown that functionalization withoxygen and nitrogen enhances the conductance by as much as40% at coverages close to 15 atom %, which we attribute to theincreased surface energy and subsequently enhanced reactivityof graphene functionalized with these adsorbates. Conversely,we find that functionalization with fluorine results in little to nochange in the thermal conductance, potentially due to adecrease in surface energy following functionalization. Regard-ing electrical transport at similar interfaces, we have shown thatfunctionalization with oxygen leads to a significant increase inρC, while functionalization with fluorine can actually decreaseρC with increasing percent coverage. While the exactmechanism behind this observation is quite complex, theexperimental results here suggest that it is due in large part tothe presence of fluorinated graphene adjacent to the function-alized contacts that aids the injection and extraction of carriersbetween the contact and channel regions. This workdemonstrates the ability to tune both the thermal and electricaltransport properties in graphene/metal contacts, an importantaspect in the development of graphene devices.

■ ASSOCIATED CONTENT

*S Supporting InformationGraphene growth and transfer methods, as well as TDTR andTLM experimental methods and data analyses. The SupportingInformation is available free of charge on the ACS Publicationswebsite at DOI: 10.1021/acs.nanolett.5b00381.

■ AUTHOR INFORMATION

Corresponding Authors*E-mail: bmf4su@virginia.edu.*E-mail: scott.walton@nrl.navy.mil.*E-mail: peh4v@virginia.edu.

Author ContributionsP.E.H. and S.G.W. conceived the idea for this experiment andsupervised all aspects. J.T.R. synthesized and transferred thegraphene onto the SiO2/Si substrates. S.C.H. functionalizedand chemically patterned the graphene, metalized the samples,and performed XPS and Raman characterization and theirrespective data analysis. B.M.F., J.C.D., and P.E.H performedthe thermal and electrical measurements and analysis of thedata.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

P.E.H. appreciates funding from the Office of Naval ResearchYoung Investigator Program (N00014-13-4-0528). B.M.F isgrateful for support from the Army Research Office (W911NF-13-1-0378) and the ARCS Foundation Metro WashingtonChapter. This work was partially supported by the NavalResearch Laboratory Base Program. This work was performed

in part at the Center for Atomic, Molecular, and OpticalScience (CAMOS) at the University of Virginia.

■ 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 inatomically thin carbon films. Science 2004, 306, 666−669.(2) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.;Katsnelson, M. I.; Grigorieva, I. V.; Dubonos, S. V.; Firsov, A. A. Two-dimensional gas of massless dirac fermions in graphene. Nature 2005,438, 197−200.(3) Bolotin, K.; Sikes, K.; Jiang, Z.; Klima, M.; Fudenberg, G.; Hone,J.; Kim, P.; Stormer, H. Ultrahigh electron mobility in suspendedgraphene. Solid State Commun. 2008, 146, 351−355.(4) Balandin, A. A.; Ghosh, S.; Bao, W.; Calizo, I.; Teweldebrhan, D.;Miao, F.; Lau, C. N. Superior thermal conductivity of single-layergraphene. Nano Lett. 2008, 8, 902−907.(5) Cai, W.; Moore, A. L.; Zhu, Y.; Li, X.; Chen, S.; Shi, L.; Ruoff, R.S. Thermal transport in suspended and supported monolayer graphenegrown by chemical vapor deposition. Nano Lett. 2010, 10, 1645−1651.(6) Seol, J. H.; Jo, I.; Moore, A. L.; Lindsay, L.; Aitken, Z. H.; Pettes,M. T.; Li, X.; Yao, Z.; Huang, R.; Broido, D.; Mingo, N.; Ruoff, R. S.;Shi, L. Two-dimensional phonon transport in supported graphene.Science 2010, 328, 213−216.(7) Nagashio, K.; Nishimura, T.; Kita, K.; Toriumi, A. Metal/graphene contact as a performance killer of ultra-high mobilitygraphene analysis of intrinsic mobility and contact resistance. InElectron Devices Meeting (IEDM) 2009 IEEE International,Baltimore, MD, Dec 7−9, 2009; IEEE: Bellingham, WA, 2009.(8) Nagashio, K.; Nishimura, T.; Kita, K.; Toriumi, A. Contactresistivity and current flow path at metal/graphene contact. Appl. Phys.Lett. 2010, 97, 143514.(9) Xia, F.; Perebeinos, V.; Lin, Y.-M.; Wu, Y.; Avouris, P. The originsand limits of metal-graphene junction resistance. Nat. Nanotechnol.2011, 6, 179−184.(10) Robinson, J. A.; LaBella, M.; Zhu, M.; Hollander, M.; Kasarda,R.; Hughes, Z.; Trumbull, K.; Cavalero, R.; Snyder, D. Contactinggraphene. Appl. Phys. Lett. 2011, 98, 053103.(11) Blake, P.; Yang, R.; Morozov, S.; Schedin, F.; Ponomarenko, L.;Zhukov, A.; Nair, R.; Grigorieva, I.; Novoselov, K.; Geim, A. Influenceof metal contacts and charge inhomogeneity on transport properties ofgraphene near the neutrality point. Solid State Commun. 2009, 149,1068−1071.(12) Moon, J. S.; Antcliffe, M.; Seo, H. C.; Curtis, D.; Lin, S.;Schmitz, A.; Milosavljevic, I.; Kiselev, A. A.; Ross, R. S.; Gaskill, D. K.;Campbell, P. M.; Fitch, R. C.; Lee, K.-M.; Asbeck, P. Ultra-lowresistance ohmic contacts in graphene field effect transistors. Appl.Phys. Lett. 2012, 100, 203512.(13) Watanabe, E.; Conwill, A.; Tsuya, D.; Koide, Y. Low contactresistance metals for graphene based devices. Diamond Relat. Mater.2012, 24, 171−174.(14) Lin, Y.-F.; Wang, S.-T.; Pao, C.-C.; Li, Y.-C.; Lai, C.-C.; Lin, C.-K.; Hsu, S.-Y.; Jian, W.-B. Probing into the metal-graphene interface byelectron transport measurements. Appl. Phys. Lett. 2013, 102, 033107.(15) Malec, C. E.; Davidovic, D. Electronic properties of Au-graphene contacts. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 84,033407.(16) Balci, O.; Kocabas, C. Rapid thermal annealing of graphene-metal contact. Appl. Phys. Lett. 2012, 101, 243105.(17) Lin, Y.-M.; Jenkins, K. A.; Valdes-Garcia, A.; Small, J. P.; Farmer,D. B.; Avouris, P. Operation of graphene transistors at gigahertzfrequencies. Nano Lett. 2009, 9, 422−426.(18) Lin, Y.-M.; Dimitrakopoulos, C.; Jenkins, K. A.; Farmer, D. B.;Chiu, H.-Y.; Grill, A.; Avouris, P. 100-GHz transistors from wafer-scaleepitaxial graphene. Science 2010, 327, 662.(19) Lee, J.; Ha, T.-J.; Li, H.; Parrish, K. N.; Holt, M.; Dodabalapur,A.; Ruoff, R. S.; Akinwande, D. 25 GHz embedded-gate graphenetransistors with high-k dielectrics on extremely flexible plastic sheets.ACS Nano 2013, 7, 7744−7750.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00381Nano Lett. 2015, 15, 4876−4882

4881

(20) Cheng, R.; Bai, J.; Liao, L.; Zhou, H.; Chen, Y.; Liu, L.; Lin, Y.-C.; Jiang, S.; Huang, Y.; Duan, X. High-frequency self-alignedgraphene transistors with transferred gate stacks. Proc. Natl. Acad.Sci. U. S. A. 2012, 109, 11588−11592.(21) Bae, M.-H.; Islam, S.; Dorgan, V. E.; Pop, E. Scaling of high-fieldtransport and localized heating in graphene transistors. ACS Nano2011, 5, 7936−7944.(22) Liao, A. D.; Wu, J. Z.; Wang, X.; Tahy, K.; Jena, D.; Dai, H.;Pop, E. Thermally limited current carrying ability of graphenenanoribbons. Phys. Rev. Lett. 2011, 106, 256801.(23) Islam, S.; Li, Z.; Dorgan, V.; Bae, M.-H.; Pop, E. Role of jouleheating on current saturation and transient behavior of graphenetransistors. IEEE Electron Device Lett 2013, 34, 166−168.(24) Grosse, K. L.; Bae, M.-H.; Lian, F.; Pop, E.; King, W. P.Nanoscale joule heating, peltier cooling and current crowding atgraphene-metal contacts. Nat. Nanotechnol. 2011, 6, 287−290.(25) Li, X.; Cai, W.; An, J.; Kim, S.; Nah, J.; Yang, D.; Piner, R.;Velamakanni, A.; Jung, I.; Tutuc, E.; Banerjee, S. K.; Colombo, L.;Ruoff, R. S. Large-area synthesis of high-quality and uniform graphenefilms on copper foils. Science 2009, 324, 1312−1314.(26) Kim, K. S.; Zhao, Y.; Jang, H.; Lee, S. Y.; Kim, J. M.; Kim, K. S.;Ahn, J.-H.; Kim, P.; Choi, J.-Y.; Hong, B. H. Large-scale patterngrowth of graphene films for stretchable transparent electrodes. Nature2009, 457, 706−710.(27) Hernandez, S. C.; Bennett, C. J. C.; Junkermeier, C. E.; Tsoi, S.D.; Bezares, F. J.; Stine, R.; Robinson, J. T.; Lock, E. H.; Boris, D. R.;Pate, B. D.; Caldwell, J. D.; Reinecke, T. L.; Sheehan, P. E.; Walton, S.G. Chemical gradients on graphene to drive droplet motion. ACSNano 2013, 7, 4746−4755.(28) Baraket, M.; Walton, S. G.; Lock, E. H.; Robinson, J. T.; Perkins,F. K. The functionalization of graphene using electron-beam generatedplasmas. Appl. Phys. Lett. 2010, 96, 231501.(29) Hernandez, S. C.; Bezares, F. J.; Robinson, J. T.; Caldwell, J. D.;Walton, S. G. Controlling the local chemical reactivity of graphenethrough spatial functionalization. Carbon 2013, 60, 84−93.(30) Ostrikov, K.; Neyts, E. C.; Meyyappan, M. Plasma nanoscience:from nano-solids in plasmas to nano-plasmas in solids. Adv. Phys.2013, 62, 113−224.(31) Lee, S. W.; Mattevi, C.; Chhowalla, M.; Sankaran, R. M. Plasma-assisted reduction of graphene oxide at low temperature andatmospheric pressure for flexible conductor applications. J. Phys.Chem. Lett. 2012, 3, 772−777.(32) Walton, S. G.; Boris, D. R.; Hernandez, S. C.; Lock, E. H.;Petrova, T. B.; Petrov, G. M.; Fernsler, R. F. Electron beam generatedplasmas for ultra low Te processing. ECS J. Solid State Sci. Technol.2015, 4, N5033−N5040.(33) Hernandez, S.; Wheeler, V.; Osofsky, M.; Jernigan, G.;Nagareddy, V.; Nath, A.; Lock, E.; Nyakiti, L.; Myers-Ward, R.;Sridhara, K.; Horsfall, A.; Eddy, C. R.Jr.; Gaskill, D.; Walton, S.Plasma-based chemical modification of epitaxial graphene with oxygenfunctionalities. Surface and Coatings Technology, Selected Papers fromthe Society of Vacuum Coaters 56th Annual Technical Conference,Providence, RI, April 20−25, 2013; Elsevier: New York, 2014.(34) Baraket, M.; Stine, R.; Lee, W. K.; Robinson, J. T.; Tamanaha,C. R.; Sheehan, P. E.; Walton, S. G. Aminated graphene for DNAattachment produced via plasma functionalization. Appl. Phys. Lett.2012, 100, 233123.(35) Felts, J. R.; Oyer, A. J.; Hernandez, S. C.; Whitener, K. E., Jr;Robinson, J. T.; Walton, S. G.; Sheehan, P. E. Direct mechanochemicalcleavage of functional groups from graphene. Nat. Commun. 2015, 6,1−7.(36) Robinson, J. T.; Burgess, J. S.; Junkermeier, C. E.; Badescu, S.C.; Reinecke, T. L.; Perkins, F. K.; Zalalutdniov, M. K.; Baldwin, J. W.;Culbertson, J. C.; Sheehan, P. E.; Snow, E. S. Properties of fluorinatedgraphene films. Nano Lett. 2010, 10, 3001−3005.(37) Tuinstra, F.; Koenig, J. L. Raman spectrum of graphite. J. Chem.Phys. 1970, 53, 1126−1130.

(38) Cahill, D. G.; Goodson, K.; Majumdar, A. Thermometry andthermal transport in micro/nanoscale solid-state devices andstructures. J. Heat Transfer 2002, 124, 223−241.(39) Koh, Y. K.; Bae, M.-H.; Cahill, D. G.; Pop, E. Heat conductionacross monolayer and few-layer graphenes. Nano Lett. 2010, 10, 4363−4368.(40) Losego, M. D.; Grady, M. E.; Sottos, N. R.; Cahill, D. G.; Braun,P. V. Effects of chemical bonding on heat transport across interfaces.Nat. Mater. 2012, 11, 502−506.(41) Duda, J. C.; Yang, C.-Y. P.; Foley, B. M.; Cheaito, R.; Medlin, D.L.; Jones, R. E.; Hopkins, P. E. Influence of interfacial properties onthermal transport at gold:silicon contacts. Appl. Phys. Lett. 2013, 102,081902.(42) Hopkins, P. E.; Baraket, M.; Barnat, E. V.; Beechem, T. E.;Kearney, S. P.; Duda, J. C.; Robinson, J. T.; Walton, S. G.Manipulating thermal conductance at metal-graphene contacts viachemical functionalization. Nano Lett. 2012, 12, 590−595.(43) Wang, S.; Zhang, Y.; Abidi, N.; Cabrales, L. Wettability andsurface free energy of graphene films. Langmuir 2009, 25, 11078−11081.(44) Shin, Y. J.; Wang, Y.; Huang, H.; Kalon, G.; Wee, A. T. S.; Shen,Z.; Bhatia, C. S.; Yang, H. Surface-energy engineering of graphene.Langmuir 2010, 26, 3798−3802.(45) Giovannetti, G.; Khomyakov, P. A.; Brocks, G.; Karpan, V. M.;van den Brink, J.; Kelly, P. J. Doping graphene with metal contacts.Phys. Rev. Lett. 2008, 101, 026803.(46) Khomyakov, P. A.; Giovannetti, G.; Rusu, P. C.; Brocks, G.; vanden Brink, J.; Kelly, P. J. First-principles study of the interaction andcharge transfer between graphene and metals. Phys. Rev. B: Condens.Matter Mater. Phys. 2009, 79, 195425.(47) Khomyakov, P. A.; Starikov, A. A.; Brocks, G.; Kelly, P. J.Nonlinear screening of charges induced in graphene by metal contacts.Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 82, 115437.(48) Lee, E. J. H.; Balasubramanian, K.; Weitz, R. T.; Burghard, M.;Kern, K. Contact and edge effects in graphene devices. Nat.Nanotechnol. 2008, 3, 486−490.(49) Mueller, T.; Xia, F.; Freitag, M.; Tsang, J.; Avouris, P. Role ofcontacts in graphene transistors: A scanning photocurrent study. Phys.Rev. B: Condens. Matter Mater. Phys. 2009, 79, 245430.(50) Nourbakhsh, A.; Cantoro, M.; Vosch, T.; Pourtois, G.;Clemente, F.; van der Veen, M. H.; Hofkens, J.; Heyns, M. M.; DeGendt, S.; Sels, B. F. Bandgap opening in oxygen plasma-treatedgraphene. Nanotechnology 2010, 21, 435203.(51) Nourbakhsh, A.; Cantoro, M.; Klekachev, A. V.; Pourtois, G.;Vosch, T.; Hofkens, J.; van der Veen, M. H.; Heyns, M. M.; Gendt, S.De; Sels, B. F. Single layer vs bilayer graphene: A comparative study ofthe effects of oxygen plasma treatment on their electronic and opticalproperties. J. Phys. Chem. C 2011, 115, 16619−16624.(52) Leconte, N.; Moser, J.; Ordejon, P.; Tao, H.; Lherbier, A.;Bachtold, A.; Alsina, F.; Sotomayor Torres, C. M.; Charlier, J.-C.;Roche, S. Damaging graphene with ozone treatment: A chemicallytunable metal-insulator transition. ACS Nano 2010, 4, 4033−4038.(53) Childres, I.; Jauregui, L. A.; Tian, J.; Chen, Y. P. Effect of oxygenplasma etching on graphene studied using raman spectroscopy andelectronic transport measurements. New J. Phys. 2011, 13, 025008.(54) Mathieu, C.; Lalmi, B.; Mentes, T. O.; Pallecchi, E.; Locatelli, A.;Latil, S.; Belkhou, R.; Ouerghi, A. Effect of oxygen adsorption on thelocal properties of epitaxial graphene on SiC (0001). Phys. Rev. B:Condens. Matter Mater. Phys. 2012, 86, 035435.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00381Nano Lett. 2015, 15, 4876−4882

4882