UNIVERSITY OF OULU P .O. Box 8000 F I -90014 UNIVERSITY OF OULU FINLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

University Lecturer Tuomo Glumoff

University Lecturer Santeri Palviainen

Postdoctoral researcher Jani Peräntie

University Lecturer Anne Tuomisto

University Lecturer Veli-Matti Ulvinen

Planning Director Pertti Tikkanen

Professor Jari Juga

Associate Professor (tenure) Anu Soikkeli

University Lecturer Santeri Palviainen

Publications Editor Kirsti Nurkkala

ISBN 978-952-62-3033-7 (Paperback)ISBN 978-952-62-3034-4 (PDF)ISSN 0355-3213 (Print)ISSN 1796-2226 (Online)

U N I V E R S I TAT I S O U L U E N S I SACTAC

TECHNICA

U N I V E R S I TAT I S O U L U E N S I SACTAC

TECHNICA

OULU 2021

C 797

Joni Kilpijärvi

RF-MICROWAVE SENSOR DEVELOPMENT FOR CELL AND HUMAN IN VITRO AND EX VIVO MONITORING

UNIVERSITY OF OULU GRADUATE SCHOOL;UNIVERSITY OF OULU,FACULTY OF INFORMATION TECHNOLOGY AND ELECTRICAL ENGINEERING;LINKÖPING UNIVERSITY,DEPARTMENT OF PHYSICS, CHEMISTRY AND BIOLOGY (IFM);INFOTECH

C 797

AC

TAJoni K

ilpijärvi

C797etukansi.fm Page 1 Tuesday, August 17, 2021 1:51 PM

ACTA UNIVERS ITAT I S OULUENS I SC Te c h n i c a 7 9 7

JONI KILPIJÄRVI

RF-MICROWAVE SENSOR DEVELOPMENT FOR CELL AND HUMAN IN VITRO AND EX VIVO MONITORING

Academic dissertation to be presented with the assent ofthe Doctoral Training Committee of InformationTechnology and Electrical Engineering of the University ofOulu for public defence in Martti Ahtisaari auditorium(L2), Linnanmaa, on 22 October 2021, at 11 a.m.

UNIVERSITY OF OULU, OULU 2021

Copyright © 2021Acta Univ. Oul. C 797, 2021

Supervised byDocent Jari JuutiDoctor Niina HalonenDoctor Maciej SobocińskiProfessor Kajsa UvdalProfessor Anita Lloyd Spetz

Reviewed byProfessor Hong WangProfessor Mailadil Thomas Sebastian

ISBN 978-952-62-3033-7 (Paperback)ISBN 978-952-62-3034-4 (PDF)

ISSN 0355-3213 (Printed)ISSN 1796-2226 (Online)

Cover DesignRaimo Ahonen

PUNAMUSTATAMPERE 2021

OpponentDoctor Karol Malecha

Kilpijärvi, Joni, RF-microwave sensor development for cell and human in vitroand ex vivo monitoring. University of Oulu Graduate School; University of Oulu, Faculty of Information Technologyand Electrical Engineering; Linköping University, Department of Physics, Chemistry andBiology (IFM); InfotechActa Univ. Oul. C 797, 2021University of Oulu, P.O. Box 8000, FI-90014 University of Oulu, Finland

Abstract

In this research new RF/microwave-based sensor solutions were developed for the monitoring ofbiological cells and human beings to obtain a better understanding of their activity or state in aquick, cheap, easy and continuous way. The effect of different substances on cell behaviour canbe monitored by measuring the electrical environment where changes are observed as cells reactto a stimulus. The starting point of the study was a microchip with a capacitance measurementsystem integrated into the culturing chamber, enabling the monitoring of cell proliferation ordeath. The main challenge of the study was the correct interpretation of the received signals andthe combination of “dry” electronics and “wet” biology, which is a difficult issue in terms ofreliability and durability of the system. For this purpose, a low temperature co-fired ceramicpackage was developed which could withstand cell culture conditions and which did not interferewith the cell activity. A 1.1 MHz shift in resonance frequency of the system could clearly bemeasured, where the shift depended on the number of cells. Another topic of the researchconcentrated on a microwave sensor that can be utilized in the examination and analysis of fluidsamples collected from the body which provide information about a person’s health status. Amicrowave sensor was developed, which was tested with liquid samples. Microfluidics were alsointegrated into the system which allowed the use of very small sample volumes and improved theusability of the device. The challenge of the work was to build the system so that the parts wereintegrated seamlessly without interfering with each other. The sensor concept was testedsuccessfully using typical concentrations of NaCl found in human blood plasma i.e. 125 to 155mmol/mol of water. The third topic of the thesis was aiming for a microwave sensor that enablesreal-time measurement of body fluid balance directly from the skin. The operation of thedeveloped microwave sensor was based on a resonator whose resonance frequency reacted to theelectrical properties of materials in its proximity, in this case the water content of the skin and itschanges. The function of the sensor was tested with artificial skin, made in the laboratory, whichcorresponded to the properties of real skin. The observed changes in resonance frequency was+370 MHz and -220 MHz for dehydrated and hydrated skin compared to normal skin, thusproviding a wide frequency range for detection of the status of the skin.

Keywords: cell culture monitoring, CSRR, dielectric properties, health monitoring, IDE,lab-on-a-chip, low temperature co-fired ceramic (LTCC), microwave sensor

Kilpijärvi, Joni, RF-mikroaaltoanturien kehitys solujen ja ihmisen in vitro ja exvivo monitorointiin. Oulun yliopiston tutkijakoulu; Oulun yliopisto, Tieto- ja sähkötekniikan tiedekunta; Linköpinginyliopisto, Fysiikan, kemian ja biologian laitos (IFM); InfotechActa Univ. Oul. C 797, 2021Oulun yliopisto, PL 8000, 90014 Oulun yliopisto

Tiivistelmä

Tutkimuksessa kehitettiin uusia RF-/mikroaaltoihin perustuvia anturiratkaisuja solujen ja ihmi-sen mittaukseen, jotta niiden toiminnasta tai tilasta saataisiin parempi kuva nopeasti, edullisesti,helposti ja jatkuvatoimisesti. Solujen toimintaa voidaan seurata mittaamalla niiden sähköistäympäristöä missä havaitaan muutoksia, kun solut reagoivat erilaisiin aineisiin. Tutkimuksen läh-tökohtana käytettiin soluanturiksi suunniteltua mikrosirua, jossa oli viljelyalustaan integroitukapasitanssin mittausjärjestelmä, jonka avulla voitiin monitoroida solujen jakaantumista tai kuo-lemista. Tutkimuksen haasteena oli saatujen signaalien oikeanlainen tulkinta sekä ”kuivan”elektroniikan ja ”märän” biologian yhdistäminen järjestelmän luotettavuuden ja kestävyydenkannalta. Tähän tarkoitukseen työssä kehitettiin matalan lämpötilan yhteissintrattavaan keraa-miin perustuva pakkaus, joka kestää soluviljelyn olosuhteita eikä häiritse solujen toimintaa. Tes-teissä voitiin havaita selvä 1.1 MHz muutos resonanssitaajuudessa, jonka suuruus riippui solu-jen lukumäärästä. Tutkimuksen toinen alue oli mikroaaltoanturi, jolla voidaan tutkia ja analysoi-da kehosta saatavia nestemäisiä näytteitä ja saada tietoa henkilön terveydentilasta. Työssä kehi-tettiin mikroaaltoanturi, jota testattiin nestemäisillä näytteillä. Nestenäytteiden käsittelemiseksisysteemiin integroitiin myös mikrofluidistiikka mikä mahdollistaa hyvin pienten näytemäärienkäyttämisen ja parantaa laitteen käytettävyyttä. Työn haasteena oli järjestelmän rakentaminensiten, että osat integroituvat toisiinsa saumattomasti toisiaan häiritsemättä. Anturikonsepti testat-tiin onnistuneesti käyttämällä tyypillistä ihmisen veriplasmasta löytyvää NaCl-pitoisuutta vaih-teluvälillä 125–155 mmol/mol vedessä. Väitöstyön kolmas aihealue oli mikroaaltoanturin hyö-dyntäminen kehon nestetasapainon mittauksessa reaaliaikaisesti suoraan iholta. Kehitetyn mik-roaaltoanturin toiminta perustui resonaattoriin, jonka resonanssitaajuus reagoi sen lähiympäris-tön sähköisiin ominaisuuksiin eli tässä tapauksessa ihon vesipitoisuuteen ja siinä tapahtuviinmuutoksiin. Anturin toimintaa testattiin laboratoriossa valmistettujen keinoihojen avulla, jotkavastasivat ominaisuuksiltaan oikeata ihoa kuvastaen eri tilannetta kehon nestetasapainossa.Mitattu resonanssitaajuus muuttui +370 MHz ja -220 MHz kuivan ja kostean ihon välillä verrat-tuna normaaliin ihoon, tarjoten laajan taajuusalueen ihon tilanteen havainnointiin.

Asiasanat: CSRR, dielektriset ominaisuudet, IDE, lab-on-a-chip, matalan lämpötilanyhteissintrautuva keraami (LTCC), mikroaaltoanturi, soluviljelmä monitorointi, terveysmonitorointi

Kilpijärvi, Joni, RF-mikrovågssensorutveckling för cell- och människa in vitro ochex vivo-övervakning. Forskarsskolan vid Uleåborgs universitet; Uleåborgs universitet, Fakulteten för information ochelektroteknik; Linköpings universitet, Institutionen för fysik, kemi och biologi (IFM); InfotechActa Univ. Oul. C 797, 2021Uleåborgs universitet, PB 8000, FI-90014 Uleåborgs universitet, Finland

Abstract

I denna forskning utvecklades nya RF / mikrovågsbaserade sensorlösningar för övervakning avceller och människor för att få en bättre förståelse för deras aktivitet eller tillstånd snabbt, billigt,enkelt och kontinuerligt. Effekten av olika ämnen på beteendet hos celler kan övervakas genomatt mäta deras elektriska miljö där förändringar observeras när celler reagerar på stimulanser.Utgångspunkten för studien var ett mikrochip med ett kapacitansmätsystem integrerat iodlingskammaren, vilket möjliggör övervakning av cellproliferation eller död. Utmaningen medstudien var den korrekta tolkningen av de mottagna signalerna och kombinationen av "torr"elektronik och "våt" biologi, vilket är utmanande problemställningar som måste lösas försystemets tillförlitlighet och hållbarhet. För detta ändamål utvecklades en keramisk förpackning“tillverkad vid låg temperature” som tål cellodlingsförhållanden och inte stör cellaktiviteten. Etttydligt skift på 1,1 MHz i systemets resonansfrekvens kunde mätas, där storlek på skiftet beroddepå antalet celler. En annan del av forskningen i avhandlingen koncentrerade sig påmikrovågssensorn som kan användas vid undersökning och analys av vätskeprover, som kansamlas in från kroppen och ge information om en persons hälsotillstånd. En mikrovågssensorutvecklades som testades på flytande prover. Mikrofluidik integrerades i systemet för hanteringenav flytande prover, vilket möjliggör användning av mycket små provvolymer och förbättrarenhetens användbarhet. Utmaningen med arbetet var att bygga systemet med full integration avdelarna och utan att dessa störde varandra. Sensorkonceptet testades framgångsrikt medanvändning av en typisk NaCl-koncentration som finns i human blodplasma, dvs. 125 till 155mmol / mol vatten. Den tredje delen av avhandlingen syftade till en mikrovågssensor sommöjliggör realtidsmätning av kroppsvätskebalansen direkt på huden. Funktionen för denutvecklade mikrovågssensorn baserades på en resonator, resonansfrekvensen beror på deelektriska egenskaperna hos material i dess närhet, i detta fall hudens vätskeinnehåll och dessförändringar. Sensorns funktion testades med konstgjord hud, som tillverkades i laboratoriet, medmotsvarade egenskaper som hos riktig hud. Observerade förändringar i resonansfrekvensen var+370 MHz och -220 MHz för uttorkad och hydratiserad hud jämfört med normal hud, vilket gerett brett frekvensområde för detektion av hudens status

Ämnesord: CSRR, dielektrisk mätning, hälsoövervakning, IDE, lab-on-a-chip, lågtemperatur keramik (LTCC), mikrovågssensor, mikrovågssensor, övervakning avcellodling

”Tietä käyden tien on vanki.Vapaa on vain umpihanki.” -Aaro Hellaakoski

10

11

Acknowledgements

The majority of the work was done in the Microelectronics Research Unit at the

University of Oulu and part of the work was done in the Department of Physics,

Chemistry and Biology (IFM) at the Linköping University. Greatly appreciated

personal grants and scholarships were granted to me from Tauno Tönning

Foundation, Nokia Foundation, Finnish Technology Promotion, Walter Ahlströmin

and Riitta and Jorma J. Takanen foundation. The following funding allowed me to

do the necessary work; Proof of Concept funding of Oulu University, short-term

scientific mission (STSM) COST funding EuNetAir TD1105 to visit Linköping

University, Academy of Finland (The ClintoxNP #268944) and TEKES (The

Chempack # 1427/31/2010) projects. Linköping University funded travelling and

living at Linköping and a ITEE-DP funded short-term doctoral student for finishing

the doctoral thesis.

Special thanks go to Prof. Anita Lloyd Spetz, Dr. Niina Halonen and Dr. Maciej

Sobocinski for taking me through the academic working of the master’s thesis and

finally the doctoral thesis. Then massive thanks to Docent Jari Juuti for being my

principal supervisor and giving technological, emotional, and motivational support.

Many thanks to Dr. Jari Hannu for kiking me forward in different stages of thesis.

Without him I would still be making an endless loop of experiments in the

laboratory. Thanks for Prof. Heli Jantunen controlling the ship (research unit)

where I was a passenger. Thanks to Prof. Kajsa Uvdal for inviting me warmly to

the Linköping University and guiding me together with Prof. Anita Lloyd Spetz to

handle practical things and finish the thesis work. Dr. Antti Hassinen is greatly

thanked for performing the cell culture experiments in Oulu.

Thanks to my steering group Assoc. Prof. Ilkka Nissinen, Docent Merja

Teirikangas and Dr. Mikko Leinonen for your support and effort for this work.

Many thanks for Prof. Arthur Hill for doing language checks to nearly all

publications and to this thesis. Co-authors I would like also to thank because this

would not be possible without you. I wish to thank the Centre of Microscopy and

Nanotechnology for their help and sharing knowledge, especially Veli-Pekka

Moilanen, Kai Metsäkoivu and Hannu Moilanen.

Thanks to all colleagues for helping work related things and moments in the

coffee room. Thanks for Msc. Kari Nurmi and Dr. Jarkko Tolvanen for friendship

and support starting from the first year in the University. Timo Vahera thanks for

many intresting conversations during lunches and workdays. Extra special thanks

go to our expedition team Dr. Jussi Putaala, Dr. Mikko Nelo and Prof. Henrikki

12

Liimatainen for going through darkness (physical and mental), mud, rocky rivers

and snow blizzards just to hang out in a tent or remote cottage without electricity.

I am already waiting for the next adventure!

I would like to thank my family for supporting and loving me through life in

uphills and downhills. Thank you my new and old friends. Thank you my lovely

partner Suvi Sirviö for being my loved and watching over me for many years and

we will continue our journey together. Now I have time to take the trash out and

prepare nice dinners due the thesis is ready!

Oulu, July. 2021 Joni Kilpijärvi

13

List of abbreviations and symbols

µTAS Micro Total Chemical Analysis System

BGA Ball Grid Array

CSF Cerebro Spinal Fluid

CBCM Charge-Based Capacitance Measurement

CMOS Complementary Metal Oxide Semiconductor

CSRR Complementary Split Ring Resonator

DAK Dielectric Assesment Kit

ECIS Electric Cell-substrate Impedance Sensing

HTCC High Temperature Co-fired Ceramic

IEEE Institute of Electrical and Electronics Engineers

IDE Interdigitated Electrode

ICA Isotropic Conductive Adhesive

IT’IS The Foundation for Research on Information Technologies in

Society

LOC Lab-On-a-Chip

LTCC Low Temperature Co-fired Ceramic

MEMS Micromechanical Systems

PCB Printed Circuit Board

PCR Polymerase Chain Reaction

PECVD Plasma-enhanced Chemical Vapour Deposition

PDMS Polydimethylsiloxane

POC Proof Of Concept

RIE Reactive Ion Etching

SMA SubMiniature version A

SIW Substrate Integrated Waveguide

VNA Vector Network Analyzer

VCO Voltage Controlled Oscillator

14

15

List of original publications

This thesis is based on the following publications, which are referred to throughout

the text by their Roman numerals:

I Halonen, N., Kilpijärvi, J., Sobocinski, M., Datta-Chaudhuri, T., Hassinen, A., Prakash, S. B., Möller, P., Abshire, P., Kellokumpu, S., & Lloyd Spetz, A. (2016). Low temperature co-fired ceramic packaging of CMOS capacitive sensor chip towards cell viability monitoring. Beilstein Journal of Nanotechnology, 7, 1871–1877. https://doi.org/10.3762/bjnano.7.179

II Kilpijärvi, J., Halonen, N., Sobocinski, M., Hassinen, A., Senevirathna, B., Uvdal, K., Abshire, P., Smela, E., Kellokumpu, S., Juuti, J., & Lloyd Spetz, A. (2018). LTCC packaged ring oscillator based sensor for evaluation of cell proliferation. Sensors, 18(10), 3346. https://doi.org/10.3390/s18103346

III Kilpijärvi, J., Halonen, N., Juuti, J., & Hannu, J. (2019). Microfluidic microwave sensor for detecting saline in biological range. Sensors, 19(4), 819. https://doi.org/10.3390/s19040819

IV Kilpijärvi, J., Tolvanen, J., Juuti, J., Halonen, N., & Hannu J. (2020). A non-invasive method for hydration status measurement with a microwave sensor using skin phantoms. IEEE Sensors Journal, 20(2), 1095–1104. https://doi.org/10.1109/JSEN.2019.2945817

In Paper I a sketch of an LTCC package for a small sized CMOS biosensor was

designed by Sobocinski M and the packaging method was developed and optimized

by Kilpjärvi J. In Paper II, Kilpijärvi J, Halonen N, and Sobocinski M. jointly made

alterations to the chip packaging design. Sensors were used for monitoring cell

culture in real-time, and cell experiments were performed by A. Hassinen. Data

provided by the sensor was analyzed by Senevirathana B. andKilpijärvi J. The

package was manufactured with low temperature co-fired ceramic technology by

Kilpijärvi J.

In Paper III a sensor based on microwaves was developed by Kilpijärvi J. The

sensor was built on printed circuit board and microfluidics were seamlessly

integrated to the structure using polydimethylsiloxane (PDMS) by Kilpijärvi J. A

pumping system for small µl sized samples was designed, built and tested by

Kilpijärvi J to form a full measurement system. The concept was utilized for

measuring salinity of water in the biological range, samples for which were prepared

by Halonen N.

In Paper IV a second micromachined microwave sensor for measuring the

hydration state of a human in real time was developed. Modelling, manufacturing

and testing using different skin phantoms were made by Kilpijärvi J. Skin phantoms

were manufactured by Tolvanen J.

16

Papers I–IV were jointly written by the author and co-authors.

Papers not included in the thesis

Bunnfors, K., Abrikossova, N., Kilpijärvi, J., Eriksson, P., Juuti, J., Halonen, N.,

Brommesson, C., Lloyd Spez, A., & Uvdal, K. (2020). Nanoparticle activated

neutrophils-on-a-chip: A label-free capacitive sensor to monitor cells at work.

Sensors and Actuators B: Chemical, 313,128020.

https://doi.org/10.1016/j.snb.2020.128020

17

Contents

Abstract

Tiivistelmä

Abstrakt

Acknowledgements 11 

List of abbreviations and symbols 13 

List of original publications 15 

Contents 17 

1  Introduction 19 

1.1  History and state-of-the-art of biosensors ............................................... 19 

1.2  Biosensors and materials ......................................................................... 20 

1.3  The packaging process for small sized biosensor chips .......................... 22 

1.4  Microwave biosensors ............................................................................. 25 

1.4.1  Detection techniques and methods ............................................... 27 

1.5  Objectives and outline of the thesis ........................................................ 29 

2  Electromagnetic waves and biological materials 31 

2.1  Microwave basics .................................................................................... 31 

2.2  Dielectric properties of biological materials ........................................... 35 

3  Experimental methods 37 

3.1  The LTCC process ................................................................................... 37 

3.2  The packaging process of a CMOS biosensor ........................................ 39 

3.3  PCB-PDMS microfluidic integration for salinity sensor ........................ 43 

3.4  Integrating split-ring resonator to SMA connector for

dehydration sensing ................................................................................. 47 

3.4.1  The skin phantoms ........................................................................ 48 

4  Results 51 

4.1  The LTCC package and cell culture measurement .................................. 51 

4.1.1  Chip generation I .......................................................................... 51 

4.1.2  Chip generation II ......................................................................... 52 

4.2  Microfluidic microwave sensor for blood salinity monitoring................ 55 

4.2.1  Modelling of the microfluidic sensor ........................................... 56 

4.2.2  Measurements with the proposed sensor ...................................... 58 

4.3  Non-Invasive dehydration sensor based on split ring resonator .............. 64 

4.3.1  Modelling of the dehydration sensor ............................................ 64 

4.3.2  Measurements and simulations with the developed

dehydration sensor ........................................................................ 66 

18

5  Discussion 69 

6  Summary and outlook 71 

List of references 73 

Original publications 79 

19

1 Introduction

Transforming biological quantities and qualities into easy-to-interpret numbers is

of great interest for faster, cost cutting diagnostics in today’s healthcare industry.

Another big, interested industry field is sport and wellness monitoring. However,

combining the electronic world with the biological world is a great challenge. The

biological world consists of cells, tissues, organs and organisms, which

communicate through complicated chemical, physical and electrical signals, not to

mention the diversity of whole systems i.e. between individual humans or animals.

In comparison, the electrical world is composed of conductors, semiconductors,

insulators, capacitors, coils, resistors, integrated circuits and systems, which

operate by electrical or electromagnetic signals. If we compare these two, it is

obvious that the overall complexity of biology is overwhelming. Biology has

developed through evolution over several billions of years, whereas electrical

applications have been around for only about 200 years. However, the development

of electronics has been extremely rapid in recent decades due to the creativity of

mankind. Today’s modern sensors can convert biological signals or phenomena into

electrical signals. Therefore, it is possible to produce sensing devices by combining

the knowledge of both worlds; these sensing devices are called biosensors.

1.1 History and state-of-the-art of biosensors

The number of publications on biosensors has increased rapidly in the last 20 years.

A current search with the topic “biosensor” produced 59,677 hits in the web of

science. The first biosensor, depending on definition, was developed in 1953 and

applied to oxygen detection in blood which led to an invention in 1962, later

referred to as the “Clark electrode” [1], [2]. This invention was further developed

by a company called YSI Inc. and led to a commercial whole blood glucose analyzer

in 1975. At the end of the 1960s another of the first biosensors, for urea monitoring,

was presented [3]. By the end of the 1970s, the term biosensor, which was first

mentioned in 1977 but officially accepted by the International Union of Pure and

Applied Chemistry (IUPAC) in 1997, was connected to a lab-on-a-chip (LOC). This

was also referred as a micro Total Chemical Analysis system (µTAS) starting from

the1990s [4]. One of the first LOCs, published in 1979 was a gas sensor made on a

silicon wafer [5]. The aim was to miniaturize and combine large laboratory

instruments as in the case of a miniaturized polymerase chain reaction (PCR) device.

This provided several benefits: lower sample volume and faster and more precise

20

measurements which could be achieved by small sensors, microfluidics,

micromachining, and nanotechnology. Micromechanical systems (MEMS) is one

of the key technologies of this regime and has been used for several decades to

produce miniaturized components, including microfluidics [6], and the first LOC

sensor [5].

Satirically, the current status of lab-on-chip technology is known as the chip-

on-lab stage. Despite the frustration over the slow development of LOCs, light has

been seen at the end of the tunnel when new commercial applications have appeared

based on novel research. For example, pregnancy home tests have been available

now for quite some time and the glucose meter is another good example of an LOC

enabling the measurement of glucose values from a single drop of blood. Even

noninvasive continuous meters, which are the holy grail of the biosensor industry,

have been developed based on millimeter waves (> 300 GHz) for people suffering

from diabetes (Glucowise). There is a real need for breakthrough research to get

these miniaturized systems on the market. It is anticipated that the lab-on-chip

market will grow substantially by 2025 up to 9.06 billion dollars [7]. The biggest

challenges for LOCs include the micro-nano interface (e.g. how to introduce a tiny

volume sample into the system), wet biology damaging the unprotected electronics,

material contaminations and mass production feasibility.

This work focuses especially on a novel packaging technology for small sized

sensor chips for cell culture monitoring. Many sophisticated sensors have been

implemented on chips based on complementary metal oxide semiconductor

(CMOS) technology, including charge-based capacitance measurement (CBCM)

[8], charge sharing [9], [10], electric cell-substrate impedance sensing (ECIS) [11],

dielectric spectroscopy [12]–[15] and magnetic sensing [16]. However, the

packaging of these small chips is one reason slowing the breakthrough of

commercial devices. Another focus of this thesis is biosensors based on microwave

technology which are still at an early stage, compared to the microwave sensors

used in industrial settings (e.g. the paper and wood industries).

1.2 Biosensors and materials

Material selection is an essential part of developing biosensors. Typically, LOC

systems are manufactured from several materials yielding a hybrid structure. Some

of the most common materials are listed in Table 1. Polydimethylsiloxane (PDMS)

has been used extensively to build fluidic systems. It is soft, transparent to visible

light and is self-adhering to many smooth surfaces (e.g. itself, glass or a silicon

21

wafer). Permanent bonding can be achieved with the aid of oxygen plasma treatment

and then joining the PDMS parts together. It is a good material for prototyping since

it can be easily processed and the cost is low. The material is hydrophobic and it

has good chemical stability although some solvents can cause unwanted effects such

as swelling [17].

Glass is a versatile material from which very delicate systems can be built.

Conductors can be manufactured directly on glass (e.g. by sputter deposition).

However, the manufacturing costs of glass devices are high due to challenges with

the necessary bonding and machining of glass. Several special processes have been

developed including etching with supercritical water [18].

Epoxy or polymer-based materials have been used extensively and many

specific processes have been developed to manufacture biosensors. One example

is click chemistry [19], where epoxy laminates can be bonded together. One of the

most promising techniques of dealing with polymer parts is 3D printing which has

evolved (in accuracy, materials and speed) considerably in recent years and which

could be a viable option to produce plastic structures [20].

The high temperature co-fired ceramics (e.g. Al2O3) used in electronics are inert

and robust materials. However, their drawback is that their high sintering

temperature (> 1000 ºC) limit cofiring with low melting temperature metals and

thus expensive precious metals needs to be used (such as gold and palladium).

A more advanced ceramic technology, the low temperature co-fired ceramic

(LTCC) technology, allows more flexibility compared to HTCC in terms of more

material choices, reflecting into the design of the system and thus enabling the

manufacture of complex multilayer monolithic structures [21]. Integration of

electronics with LTCC technology is well established [22] and it is possible to

manufacture cavities, microfluidic channels and even fluidic components [23], [24],

such as valves [25], biosensors [26] or bioreactors [27]. For example, Dupont 951

LTCC material has been tested to be biocompatible with cell cultures, as shown by

our tests (Paper I and II) and others [28]. Moreover, some thick film metals used in

the system are also biocompatible [29]. The downside of LTCC ceramics is their

opacity, but the overall benefits are clear.

22

Table 1. Qualitative properties of biosensor material systems.

Material Resolution Electronics

integration

Trans-

parent

Mass

production

Process

time

Chemical

stability

Mechanical

stability

Price

PDMS Nano No Yes No, fast

prototyping

Fast Solvent

absorption,

nonhermetic

Flexible Low

Polymer Nano Yes Some

are

Yes, fast

prototyping

Fast Excellent or

bad1

Rigid or

flexible

Low

Glass Micro2 Yes Yes Yes Slow Good,

hermetic

Rigid, high

temperature

High

Silicon Nano Yes No Yes Slow Good Rigid,

medium

temperature

Low3

Paper Micro No No Yes Fast Bad4 Flexible Very

low

HTCC Micro Yes No Yes Medium Good,

hermetic

Rigid, high

temperature

High

LTCC Micro Yes No Yes Medium Good,

hermetic

Rigid, high

temperature

Medi-

um

1Depending on the type, 2Special micromachining, 3High in prototyping, 4Typically needs coating

1.3 The packaging process for small sized biosensor chips

In this chapter the packaging process for a lab-on-a-chip (LOC) device based on

CMOS technology is described. Ideal LOC devices are miniaturized laboratory

instruments which can easily perform analyses even with samples of nanolitre size.

This results in high demands on the packaging [30]. Metals used as conductors can

easily corrode in contact with biological liquids.

Typically, silicon chips are bonded to a carrier substrate (printed circuit board

[PCB] or ceramic) using a wire bonding process or flip chip technology followed

by over moulding. Then, this more convenient carrier with higher mechanical

robustness and more convenient size bonding pads, is bonded to a motherboard

using solder. Soldering is typically performed with a ball grid array (BGA).

However, in LOC applications where an active area of the chip needs to be in

contact with cells and cell media, conventional packaging methods are not suitable

due to the active area of the chip being buried in the package material.

Suitable packaging methods are listed in Table 2. CMOS based small chips are

typically packaged using a method where the chip is mounted to a board and then

wire bonds are made. Then epoxy is poured around the chip to cover the bond wires

23

and protect them [13]. This packaging method is effective when used in prototyping,

but the active area can easily be contaminated with the used protective layer e.g.

epoxy. Another method is to cover the active area (masking) of the chip with some

other material such as a cut down tube or PDMS and then pour on the epoxy, silicone

or Parylene to cover the bond wires [31], [32]. There is also a method called “chip

in hole” where the surface of the chip is plugged using PDMS and moulded to

epoxy. Then conductors are made by metal sputtering on the flat surface [30], [33].

A more sophisticated solution is to use the flip chip technique accompanied by a

ceramic alumina substrate with a hole for the active area [34]. A similar solution

but with a flexible substrate (Kapton) and electrical connections made using solder

balls is described in [35]. There is also one flexible packaging presented which

utilizes electrical connections made from liquid metal [36].

Table 2. Packaging solutions of CMOS based biosensors.

Packaging

style

Materials Sensor Chip size

(mm2)

Microfluidics Complexity Refs.

Wirebond PCB, epoxy Electrical - No Low [13],

[31]

Wirebond Standard ceramic

package, parylene

Electrochemical 3x3 No Medium [32]

Flip-chip,

thermocom-

pression

Alumina, glass, Ag/Pt

contacts

Electrical 8x8 No Low [34]

Flip-chip

solder

bumps

Flexible PCB

(Kapton), epoxy,

glass

- 2x2 Yes Medium [35]

Liquid metal

interconnec-

tions

PDMS (flexible),

liquid metal

MAGFET 1.5x1.5 Yes Low [36]

Chip in hole Epoxy, parylene, Au

contacts

Electrical 3x3 Yes Medium [30],

[33]

Flip-chip,

ICA

LTCC, Ag contacts,

epoxy

Electrical 3x3 Yes Medium [37],

Paper I

Flip-chip,

Thermocom-

pression

LTCC, Au contacts,

epoxy

Capacitance

(resonance

frequency)

3x3 No Low feasibility

for mass

production

Paper II

A bare chip can be bonded using the flip chip technique but in such cases the active

area is upside down against the substrate. This can be solved by making a hole in

the substrate. The electrical contacts can be made using the BGA process if the chip

24

is large in size but there are also problems because the solder flux can contaminate

the active surface and careful cleaning is needed before the underfill process. After

this, the gap between the chip and substrate needs to be filled, while still leaving

the active area open. This process is called underfilling where the interconnects are

protected from the corrosive liquids. The underfill is a non-conductive epoxy with

a suitable viscosity to allow it to flow but yet not cover the active area of the chip.

The capillary effect is important since the epoxy “automatically” fills the narrow

space between the chip and substrate. Another important aspect is long-term

stability when the underfill is in contact with bio liquids. Figure 1 a)–b) shows the

non-suitable packaging methods and in c) the suitable one.

Re-use of the sensor increases demands on the robustness of the package.

Autoclaving, ethanol and other cleaners (acid and enzyme cleaners e.g. Tergazyme)

for sterilization should be possible to use. In addition, there are typically other

electronic components beside the sensor itself, so the sensor should be able to be

removed from the motherboard for cleaning separately. This can be done with

easily removable connector types such as zero insertion force sockets.

Fig. 1. a) Typical BGA package, b) wire bonded package, and c) designed package with

hole in the substrate exposing the active area.

Due to the small size of the chips (3×3 mm2) used in this study (Papers I and II),

soldering with typical equipment was not possible due to the contact pads being

small (< 100 µm). In Paper I an isotropic conductive adhesive (ICA) suitable for

small sized contact pads was used to form the electrical connections. A special

process involving stamping the ICA was applied due to the small size of the pads.

25

In Paper II, a further improvement to the packaging process was made by

replacing the ICA with thermocompression bonding. Gold bumps in the chip were

pressed against gold conductors on the substrate to form a bond with the aid of heat

and pressure.

1.4 Microwave biosensors

Microwave sensors have been in use for a long time, beginning in the 1950s, where

the application was measuring the dielectric properties of materials [38]. At that

time the microwave sensors were expensive and rare. Nowadays these sensors are

used frequently especially in industrial settings [39], for example in a paper factory

[40].

Designing a microwave biosensor starts by studying the properties of the

sensed material and related materials at different frequencies. Due to the dipolar

molecular structure and resulting net polarisation of water (explained in detail in

Chapter 2) and its high abundance in organisms, it is a good material to measure

using a microwave sensor because of its strong interaction with electromagnetic

waves. Microwave sensors are typically tailored for each application where shape,

operation frequency and principle all need to be taken into account, so designing is

an important process. Nowadays, complex designing features and an improved

understanding of the operation requires the designer of the sensors to utilise

commercially available computer aided simulation software (e.g. COMSOL, CST

Studio).

Development of a microwave sensor starts with a definition of what needs to

be measured, how accurate is the measurement and does it need to be placed in a

particular medium? These considerations give hints about the suitable sensor type

(coaxial, antenna, resonator, etc.) and the required performance (bandwidth

accuracy, material handling in measurements, contact/non-contact measurements,

required measurement system). The measured materials should be characterized in

terms of their electrical and dielectric properties. Often a literature search provides

the data about the properties (papers, datasheets, etc.), however the values are not

typically presented in broadband format, causing problems in electromagnetic

simulation and later in sensor characterization. Therefore, if laboratory analysis of

the properties is possible it is preferrable to use this exact data. Furthermore,

although some materials have good dielectric properties for building the sensor,

mechanical and chemical stability also need to be taken in account. When all the

materials’ properties used in the sensor are known, a model of the system can be

26

build using simulation software and its use in the sensor system can be optimized.

After the simulation, a prototype can be manufactured and tested with samples with

known dielectric properties. It is important to think one-step ahead when

developing a sensor, i.e. keeping in mind the limitations of the manufacturing

processes in the modelling stage. Meaning not to create structures that are too

cumbersome to make or require tremendous accuracy. For example, the conductor

line can have a minimum width of 100 µm depending on the manufacturing

technique, which could be in this case a milled PCB trace. Usually (always) the

development process needs several iterations before it works properly. It should be

also noted that modelling is not perfect either since the system in the real world

differs from that in the simulation environment where many approximations have

been made. If design is critical in this aspect it can be adjusted by changing

dimensions or tuning, for example, the dielectric properties of the substrate. Also,

compromises are needed in simulations since too “realistic” a model can increase

the simulation time extensively, making it impractical. Simulation softwares have

optimization algorithms which can save time by automating parametric sweeps of

certain parameters or dimensions in the model. Figure 2 shows a scheme of

microwave sensor development.

Fig. 2. Workflow of designing a microwave sensor.

27

1.4.1 Detection techniques and methods

Typically, microwave sensors are characterized in the laboratory using a high-end

vector network analyser (VNA). Sensors are connected directly to ports of the VNA

with SMA connectors and with a coaxial cable. Several measurement methods can

be applied with either one or two ports. In one port measurements, only a single

signal path is used for transmitting and receiving. The signal is reflected from the

measured object and, by observing the magnitude and phase of the reflected signal,

sample properties can be determined. The reflected signal is presented in S11 form

(real and imaginary parts). In two port measurements the transmitter and receiver

are separated and the transmission of the signal is inspected. Monitoring of the

signal transmitted through the sample is used to determine the sample properties. S11,

S12, S21 and S22 signals can be monitored simultaneously. Since the sensors in this

thesis (Papers III and IV) are based on one port measurement this work focuses

particularly on that technique.

The most used type of one port sensor is the open-ended coaxial probe. The

probe is put in contact with the sample and properties are measured over a wide

frequency range. The dielectric properties of the sample, which can be solid or liquid,

can be calculated from the signal response. Several commercial probes are

available, e.g. Speag DAK-3.5, which is used in this work. Another example is the

Keysight N1501A Dielectric Probe Kit. These are useful tools for material

characterization but cannot be directly utilised in certain applications. There are a

vast number of microwave sensors described in the literature; Table 3 presents some

of them with key metrics. The most typical is a resonator or oscillator type of sensor

due to its high sensitivity.

In this thesis, two types of sensors based on one port design were applied. The

magnitude of the signal was measured in a sufficiently broad frequency band to

reveal the changes of resonance frequency and the corresponding amplitude. This

simplified the interpretation of results due to there being less data (S11 vs. S11 & S12)

since the fundamental principle is that when the permittivity is lower the resonance

frequency is higher and vice versa. Fundamentals for this phenomenon are

presented in Chapter 2. Compared to two port measurements, the used one port

sensing principle is much simpler and easier to manufacture due to there being fewer

conductors and connectors/inputs/outputs. Integrating these sensors, for example

in smart watch applications, would be more feasible. Also, the amount of data is

reduced making the data handling algorithms simpler, although nowadays the

microcontrollers are so powerful that data handling is not normally a problem. The

28

main benefit is lower power consumption since fewer calculations are needed

yielding a longer battery life.

Table 3. Biosensors applying microwave technology based on one-port measurement.

Sensing

technique

Data

handling

Materials Sample Sample

volume

MF1 Frequency

(GHz)

Sensitivity

Δ2

Ref.

Cavity

resonator

Attenuation/

shift of

resonance

frequency

Aluminum,

silicon tube

Glucose,

pig blood

800 µl Yes 4.7–5 6.26 dB/

11.25 MHz3

[41]

Oscillator,

open stub,

shunt stub

Shift of

resonance

SiGe:C

BiCMOS

chip,

wirebond+

epoxy

Methanol-

ethanol

Probe

immersed >

1000 µl

No 26–28 1.4 GHz4 [42]

SIW cavity

resonator

Attenuation/

shift of

resonance

PTFE,

Ceramic/

glass,

PDMS

Fibroblast

cells

3 µl No 12.8–14 0.056 (lin)/

20 MHz5

[43]

Oscillator,

open stub

differential

Colpitts

oscillator

Shif of

resonance

SiGe:C

BiCMOS

chip, Si3N4

passivation

Water fat

and calcium

Probe

immersed

>1000 µl

No 25–28.5 700 MHz6 [44]

Coplanar

waveguide

Calculated

permittivity

Quartz/

PDMS

Methanol calculated 6

nL tested?

Yes 0.045–40 - [45]

Interdigital

capacitor

Calculated

permittivity

Quartz/SU-

8

Yeast cells calculated

0.054 µl

tested ?

No 0.0005–1 - [46]

Wave-

quide

resonator

QMSIW

Attenuation/

shift of

resonance

frequency

Ceramic/

glass, 2-

side tape,

PDMS

Water/etha-

nol dual

detection

Dual;

18 and 13

µl

Yes 5.4–6 31 MHz7 [47]

Ring

resonator

Attenuation/

Shif of

resonance

PTFE,

Ceramic/

glass

Broiler meat 30×35×

15 mm

No 0.2–1.1 345 MHz8 [48]

29

Sensing

technique

Data

handling

Materials Sample Sample

volume

MF1 Frequency

(GHz)

Sensitivity

Δ2

Ref.

Quarter-

mode

spoof

plasmonic

resonator

Attenuation/

shift of

resonance

epoxy/glass

laminate

and PDMS

Water/

ethanol

3.9 µl Yes 3–7 1.55 GHz9 [49]

Ring

oscillator

Integrated

microwave

circuit -

Attenuation

Glass and

PDMS

Droplets

-

Yes 2.3–2.4 2–4 dB10 [50]

Interdigital

electrode

Attenuation/

Shift of

resonance

frequency 3

modes

FR-4 and

PDMS

Water/

isopropanol

and salinity

1 µl Yes 0.7–6 300 MHz

(@3.1

GHz) 400

MHz (@5.5

GHz)11 0.55

dB12

Paper I

Split ring

resonator

Shift of

resonance

SMA and

Cu tape

Dehydration

from skin

-

No 4.6–6.8 302 MHz13 Paper II

1MF=Microfluidic, 2Attenuation / resonance frequency change, 3150→550mg/dl, 4εr′: 4.1→6.5, 5PBS→

fibroblast cells, 610–100% Glucose, 7Ethanol→DI water, 8Meat aged 6 hours, 9Ethanol 10%→90%, 10Detects droplets attenuation changes, 11DI water→IPA, 12Salinity 0→155 mmol, 13Dehydrated→Normal

1.5 Objectives and outline of the thesis

The main purposes of biosensors is to improve the overall quality of life of humans

and animals, to boost performance and to enhance medical diagnostics. The best

way to improve life quality is to tackle the problems before they develop into more

serious ones. For example, people may suffer from dehydration without knowing

it. In the long run this causes health problems and occasionally reduced physical

and cognitive performance [51]. Moreover, the harmful effects of new drugs can

be tested with cell culture sensors at an early phase before moving to animal testing

and human trials, thus saving money and avoiding ethical issues. Many types of

biosensors exist already but the field is constantly evolving and new types of

parameters are found which need to be measured as the understanding of the

physiological complexity of human beings progresses. For example, in recent years

the cerebrospinal fluid (brain water) has been discovered to be part of the brain

cleaning system, “the sewer of the brain”. If the “sewer system” is not functioning

correctly it might cause dementia and Alzheimer’s disease [52]. This system could

30

be monitored by using a biosensor and maybe the incipient condition could be cured

or slowed down. Microwave sensors are a potential candidate for monitoring such

effects [53].

This thesis focuses on developing and researching biosensor systems to make

measurements for human beings in vivo or tissues ex vivo. The biosensor

development included problem solving, applying microwave theories into practice,

testing of several types of materials, and the development of new processes together

with the analysis of the results. Processes used in the electronic industry have been

preferred and if necessary modified to fit the measurement of humans.

Chapter 2 focuses on the electromagnetic theory on which all the sensors of

this thesis are based. The theory is explained through simple formulas and

simulation is used to illustrate theories and mechanisms occurring in the materials

and components.

Chapter 3 depicts the experimental methods used in this work. The processes

and materials used are described in detail to enable the manufacture of microwave

biosensors. The biosensor materials used in this work are commercial products

(ceramic, metal, epoxy, silicone and printed circuit board), but the samples used to

characterize microwave sensors were tailor-made for this purpose since those were

not commercially available.

Finally, Chapter 4 concerns the results achieved through the thesis work. Firsly

the packaging of sensors for cell culture monitoring is presented. CMOS

technology allows integration of the measurement electronics and sensors on one

chip, however traditional packaging technologies are not suited for this purpose. In

Paper I and II a reliable packaging system for small sized silicon chips and the

integration of microfluidics was developed. The packages survived in the corrosive

cell media environment. The behaviour of the system was demonstrated by several

tests with cancer cells. Then the results of Papers III and IV are presented where

two types of microwave sensors are described for applications involving salinity

and hydration measurement.

31

2 Electromagnetic waves and biological materials

All different biological materials have individual dielectric properties and in that

respect they do not differ from “manmade” materials. Electromagnetism is well

defined through Maxwells’s equations but here the equations are simplified for the

purpose of designing and understanding of a microwave sensor. To simplify the

properties, typical biological materials are “nonmagnetic” (relative magnetic

permeability µr =1 and without magnetic remanence).

2.1 Microwave basics

The capacitance of a parallel-plate capacitor with dielectric material between the

plates can be described by the formula

𝐶 𝜀 𝜀 , (1)

where C is the capacitance, ε0 is the permittivity of vacuum (8.854×10−12 F⋅m−1), εr

is the relative permittivity, A is the area of the plate and d is the distance between

them. The relative permittivity εr term can be expressed as

𝜀 𝜀 𝑗𝜀 , (2)

where the permittivity is presented as complex value with real part εr′ and imaginary

part εr′′. The real part explains how much energy can be stored in the capacitor and

the imaginary part tells how much energy is lost. Relative permittivity is caused by

polarization of the material and is a unitless value relative to the permittivity of

vacuum. Polarization is caused by the alignment of the polar molecule, ion, or

particle in an external electric field. Moreover, there can be atomic and electronic

polarizations which dominate the permittivity at higher frequencies. The loss

tangent is caused by the movement/vibration of these elements causing friction,

which is transformed mostly into heat.

If we think of electromagnetic energy as a wave travelling in vacuum its speed

is the speed of light (c0 =2.998×108 m/s) as defined by equation

𝑐 . (3)

The constants ε0 (8.854×10−12 F/m) and µ0 (1.257×10−6 H/m) have defined values.

The permeability term is needed in this formula as there cannot be an

electromagnetic wave without coexistence of alternating electric and magnetic

32

fields. If we think of a case where there is a non-magnetic dielectric in the path of

the wave the real part of the permittivity explains the decrease in speed of the

electromagnetic wave and the imaginary part the damping of the wave amplitude

due to more interactions with the material compared to vacuum. The decrease in

speed of the electromagnetic wave is given by equation

𝑐 , (4)

where c is the speed of the wave in the material, c0 is the speed of light in vacuum.

The dielectric losses are typically presented as the loss tangent of

tan𝛿 , (5)

which gives the relation between energy stored and energy lost due the damping.

The behaviour of an electromagnetic wave can be visualized using simulation

software. In the first case, a waveguide where the electromagnetic wave is travelling

in vacuum, which has relative permittivity 1 and no losses, is shown in Figure 3. In

the second case, a slab of dielectric material is placed along the waveguide which

has a real part of relative permittivity of 78 and a loss tangent is 0.00366. With the

dielectric in the path (signal propagates from left to right), the wave is delayed due

to the higher relative permittivity and is attenuated (loss tangent). Moreover, the

phase of the wave, speed and wavelength are all changed but the final frequency

stays the same in the waveguide.

Fig. 3. Characteristics of electromagnetic wave propagation in vacuum (top) and

through dielectric medium (bottom) as a function of time.

33

Electromagnetic waves can also bend, similarly as light. The bending can be

described by Snell’s law. In nonmagnetic and lossless material the reflection of the

electromagnetic waves is described by equation

, (6)

where θ1 is the incident angle and θ2 is the angle after going through the surface,

εr′2 is the relative permittivity of the material into which the wave is entering, εr′1 is

relative permittivity of the material where the wave is coming from, and refractive

indexes are expressed as n1 and n2. Depending on the angle of the incident wave

and the relative permittivity of the materials, the wave is reflected, but some of its

energy is transmitted (and absorbed) into the material depending on the dielectric

properties of the materials at the interface. This phenomenon is shown in Figure 4,

where a waveguide has a 90 degree bend and there is a dielectric slab similar to that

in Figure 3. Most of the wave is going through the bend but some of the wave’s

energy is transmitted into the dielectric creating waves inside the slab.

Fig. 4. Electromagnetic wave pulse travelling in vacuum and its reflection from a lossy

dielectric at 45° angle. Time instant of reflection (t = 0) shown on the left and time instant

after reflection (t > 0) shown on the right.

Resonance occurs when the wave forms a standing wave pattern and the

interference reinforces the waves depending on the phase of the interference

relative to the incoming wave. The occurrence of resonance is related to the wave’s

original wavelength and the physical length of the structure. The permittivity of the

34

material changes the speed of the signal and therefore the wavelength and the

resonance frequency. As an example, a simple coaxial open-ended resonator can be

described by equation

𝑓µ

, (7)

where fr is the resonance frequency, n is the number of fundamental mode

(1,2,3,4,5…), l is the length of the resonator structure and c is the actual velocity in

the dielectric. There are a vast number of different types of resonators where the

structure causes resonance. The resonance angular frequency ω0 or resonance

frequency fr can be calculated with lumped elements by using equation

𝜔 √

, (8)

when the structure exhibits capacitance C and inductance L. Different shapes can

have interesting interactions with microwaves. This can be explained by a lumped

element model. For example, a ring resonator can be thought of as a loop with a

gap as shown in Figure 5. The loop causes inductance L and the gap capacitance C.

Moreover, there are parasitic effects such as the capacitance caused by the substrate

material and in the sensor applications there is also changing capactitance close to

the ring according to the dielectric properties of the material. In this way the

structure can be tuned to a specific frequency if the dielectric properties of the used

materials are known. Electromagnetic simulation can be applied for more

complicated structures, which is the usual case since the sensors need a number of

different elements, feed points, connectors etc. But by using simple equations

estimations can also be performed to get a rough estimate of oscillator operation.

Fig. 5. Lumped element model of a split ring oscillator.

35

2.2 Dielectric properties of biological materials

Water is an important compound as all life on Earth is based on water and it also

has interesting dielectric and physical properties. A water molecule has a permanent

dipole moment due to the asymmetric structure of the molecule and due also to the

difference in electronegativity between the oxygen atom and the two hydrogen

atoms, where the oxygen atom occupies a larger space than the hydrogen atoms. If

there is an external electric field present, the molecules will be oriented according

to the field direction causing a change in net polarization (Figure 6). This yields a

high real part of relative permittivity, close to 80 up to a frequency of 10 GHz,

compared to many solid materials which have permittivity values less than 10.

Dielectric loss (loss tangent) increases linearly with frequency (at frequencies

below 10 GHz) due to the friction caused by the rotating water molecules. A good

example of loss is the heating effect of a microwave oven, where a high power

electric field heats the water in foodstuff.

Fig. 6. The asymmetric shape of the water molecule causes it to rotate in an oscillating

external electric field from left to right.

Water is present in all living organisms. In human beings the water mass may vary

from 55 wt.% to 75 wt.% depending on age [51]. This body water is not pure water

as it also contains minerals such as sodium chloride (NaCl), which must be

maintained within certain limits to preserve a healthy metabolism. This is called

salinity, which for a healthy person is typically about 0.9 g/litre. It has a pronounced

impact on the dielectric properties because the Na+ and Cl− are charged ions able to

move freely when diluted in water. In such cases, a higher relative permittivity can

be observed at frequencies below 200 MHz because these ions can create polarized

regions and travel in liquid and concentrate at surfaces causing interfacial

polarization. This phenomenon is referred to as the Maxwell-Wagner-Sillars effect.

36

At higher frequencies this effect decreases and finally diminishes as the rapidly

changing electric field does not give the ions time to move to the interface (inertia

of mass) but only oscillate back and forth. The ions have a large impact on the

conductivity of the liquid due to the fact that ions carry charges enabling a flow of

electric current. Moreover, they also cause increased dielectric loss due to small

vibration of ions inflicting collisions with adjacent ions which is transforming to

heat [54].

Dielectric properties, at 3 GHz, of different tissues and body fluids are listed

in Table 4, data is from The Foundation for Research on Information Technologies

in Society (IT’IS) database [55]. Pure water is also shown as a reference. Tissues

with lower water content have lower relative permittivity. Different dielectric

properties give a unique fingerprint for each tissue.

Table 4. Dielectric properties of different biological samples at 3 GHz [55].

Specimen εr′ σ (S/m)

Water 82.7 2.02

Blood 57.4 3.05

Muscle 52.1 2.14

Urine 50.0 1.75

Skin 37.5 1.74

Bone 11.1 0.51

Fat 10.7 0.34

37

3 Experimental methods

In this work, a variety of materials, including ceramics, different types of polymers

and metals, were used to fulfil the requirements of different applications. In addition,

many different manufacturing methods and techniques were applied including

LTCC and PCB for substrate manufacturing, laser processing for micromachining

and reactive ion etching (RIE) for cleaning.

3.1 The LTCC process

LTCC tapes are the basic building material for LTCC technology. The tapes are

flexible and easy to process due to the additives (mainly polymers) in the ceramic

glass matrix. Moreover, additional carrier tape improves mechanical integrity and

ruggedness during handling. The LTCC process starts with preconditioning the

pristine tapes (also referred as green tapes) in an oven for 45–60 minutes at around

100 °C to relieve internal stresses and reduce deformation during the next

processing steps. Tapes, vias and registration holes are cut to shape with a punch

(mechanical cutting) or laser. If electrical vias exist, these are filled by pushing

paste into the vias through a mask. The mask is typically the carrier tape which

simplifies the process as in this case there is no need to make a separate mask for

the via filling process. Then thick film conductor lines (gold, silver paste, etc.) are

screen printed on the tape to create the required wiring for the circuitry. Resistive,

dielectric and special pastes are also commercially available. Registration holes

made in the corners are used for optical or mechanical alignment. After levelling

the paste by keeping the tape on a flat surface for about 5 minutes, the tape is dried

in an oven (80–100 °C for 5 min). After drying, stacking of the tapes with the aid

of optical cameras or mechanical jigs makes the product more rigid and allows

multilayer 3D design of circuitry, closed cavities and channels for microfluidics,

embedded elements etc. Aligning of different LTCC layers can be made by using

an optical or mechanical setup. Subsequently, the stacked layers are laminated

together using uniaxial (press) or isostatic (water tank) pressure (3000 psi) with the

aid of an elevated temperature (70 °C for 10 min). Isostatic pressure is usually

preferred because multiple pieces can be laminated in one step and the pressure is

more homogenous. If necessary, stacked piece can be cut again after lamination

using a laser. Cutting in general is easier at this stage due to the material still being

“soft” before co-firing. Co-firing of the laminated piece is done in an oven with a

peak temperature between 800–900 °C with temperature ramp of 5 °C/min. During

38

this sintering process organic materials are decomposed (400 °C) and around

800 °C the glass melts, fills the voids between the ceramic particles and acting as a

sintering aid. As the LTCC and paste material sinters the structure forms one solid

seamless ceramic piece. Shrinkage during sintering of the tapes depends on the type

of tape and its direction, e.g. DuPont 951 has lateral shrinkage of 12.7% and 15%

in the thickness direction. In another tape, for example Ferro A6M-E, the

specifications are slightly different; 15.6% and 28% respectively. Post-fire

processes can be done after the sintering phase. These can include printing thick

film resistors, mounting solderable components, brazing or gluing of lids. If the

piece consists of multiple components these can be separated using a dicing saw,

ultrasonic machining or laser cutting. Figure 7 shows the manufacturing process for

a LTCC package (Papers I and II) where a CMOS biosensor can be mounted.

Fig. 7. LTCC packaging process utilized for biosensor package development (Adapted

under CC BY 4.0 license from Paper I © 2016 Authors).

39

Examples of fired LTCC substrates developed for biosensor packages in this work

are shown in Figure 8.

Fig. 8. Fresh batch of LTCC substrates (1.6×1.9 cm) for CMOS biosensor packages after

sintering. Conductors (100 µm width) are silver and the holes are made using laser

cutting before co-firing.

3.2 The packaging process of a CMOS biosensor

The pristine CMOS chip was prepared for flip-chip bonding by adding gold bumps

on the pads using a wedge type wire bonder (Kulicke & Soffa 4523). The pads of

the chip are made of aluminium and have a naturally forming oxide layer on top.

This was noticed in the failed first attempt at flip chip bonding the chip on the LTCC

substrate. The gold bumps remove this insulating oxide layer by mechanically

punching through the oxide making the electrical connection of the chip more

reliable. The gold bumps enable the use of conductive epoxy or thermocompression

bonding in the flip-chip process. An example of making of the gold bumps is shown

in Figure 9.

Packaging of the CMOS chip was implemented with an LTCC substrate

manufactured using the instructions provided by the material supplier (DuPont 951).

The key factors were a planar substrate and choosing packaging-compatible metals

for each packaging technique. This is explained in the next chapter. Also, the

tolerances around the hole, which is cut for the chip, are important (Figure 1). The

accuracy and quality of the optimized laser cutting process with the LPKF

ProtoLaser U3 was sufficient (absolute error < 50 µm).

40

Fig. 9. Wedge type wire bonds were made followed by removal of the excess wire.

During the bumping process, the chip was glued temporarily to the standard ceramic

package with silver paint (Electrolube). The chip was subsequently easily removed from

the carrier with a drop of acetone.

In Paper I electrical connections between the chip and the LTCC substrate were

made with conductive silver epoxy (H20E-PFC, EPO-TEK, USA) which was

suitable for small dot sizes (minimum diameter of 100 µm) and had a rapid curing

profile (e.g. 175 °C for 45 s or 80 °C for 3 h). The ICA was patterned on the LTCC

substrate by using a miniature stamp made of aluminium oxide by laser processing.

The stamp contours were optimized by changing the dimensions; the goal was to

have a good electrical contact and sufficient mechanical adhesion without short

circuits to neighbouring pads. The optimized stamp with 80×120 µm2 contours is

shown in Figure 10 a). The stamp was dipped in a 100 µm deep cavity, where ICA

was deposited, and then aligned and pressed against the LTCC substrate to form the

dots. The stamping process and curing were done with a Finetech Electronics

FINEPLACER Pico 145 with 3-axis adjustment. ICA dots with a uniform diameter

of 100 µm on the substrate are shown in Figure 10 b). Alignment of the chip with a

split vision microscope is shown in Figure 10 c). In Figure 10 d) the chip is gently

mounted onto the substrate and the ICA is cured on an integrated hotplate (175 °C

for 5 minutes).

41

Fig. 10. a) Laser processed alumina stamp (3×3 mm2) for making ICA dots on the LTCC

substrate. b) ICA dots stamped on the LTCC substrate with 100 μm wide conductor lines.

c) Alignment of the CMOS chip (3×3 mm2) by using split vision camera before mounting.

d) The mounted chip on the LTCC substrate after ICA curing on a hot plate.

Underfilling seals the gap between the chip and LTCC substrate and protects the

interconnects. This was performed by applying a small amount of nonconducting

epoxy, which could withstand water and chemicals, to the side of the chip. Capillary

forces move the epoxy across the gap between chip and the substrate. In this work

EPO-TEK 302-3M (Paper I) and 377 (Paper II) was used. In Figure 11 a) the

packaged chip is shown after the underfilling process. In Figure 11 b) it can be seen

that the underfill has been spread correctly by observing the edge of the underfill

with an optical microscope. The size of the hole through the LTCC which exposed

42

the active area of the chip was optimized so that the glue would not spread over the

sensor array seen in the middle.

In Paper I the package structure with integrated microfluidics was developed

and demonstrated. Sacrificial layers/inserts of carbon tape (C12 Advanced

Technologies, LLC, USA) were used to fill the laser-cut fluid channels in the LTCC

during the lamination process. In the sintering phase the carbon tape decomposes

to gaseous species at a temperature of 600 °C leaving the channel intact. Coupling

of the fluid-fluid outlet and inlet channels were made from metal tubes which were

attached by glue to the open ends of the channels. Moreover, the pins for the

external connections were attached to the sides of the package with ICA and sealed

using the underfill epoxy. Finally, the cell well, made of polypropylene in Paper I

and glass in Paper II, was attached on top using the same epoxy to allow cell

culturing.

In Paper II the chip bonding was performed using thermocompression bonding

i.e. without ICA. The chip was prepared as earlier but the conductors on the LTCC

were made from gold (DuPont 5742) allowing gold to gold adhesion. The chip was

aligned to the gold conductors and, with the aid of a force (40 N for 20 s) and an

elevated temperature (380 °C for 25 s), the chip was bonded to the substrate.

Electrical and mechanical contacts were formed between the gold conductors and

the gold bumps on the chip. Extra care was taken to ensure the chip and substrate

were parallel using a FINEPLACER Pico 145 with 3-axis adjustment system. The

underfill epoxy was also changed to EPO-TEK 377 which showed better

performance when the chip was used multiple times.

43

Fig. 11. Topviews of a) underfilled CMOS sensor chip (orange rectangles highlight the

underfill edge) and b) the underfill edge seen as a black frame covering the edges

(sensors are shown in the middle).

The external connector pins on the edge of the LTCC substrate were glued to the

LTCC with ICA and then sealed with underfill. The package’s outer surfaces were

covered using urethane (Urethane 71, CRC Industries, USA) to make the package

more robust to humidity and cleaning procedures.

After the processing, the whole chip was cleaned with oxygen plasma in an

inductively coupled plasma deep reactive ion etching machine ICP-DRIE (Oxford

Instruments, Plasmalab 80 Plus), which performed cleaning of all the processing

scrap and also sterilized the chip surface. Moreover, it made it more hydrophilic

which helped the cells to attach to the surface.

Lung epithelial BEAS cells were used in all cell experiments done with the

CMOS chips (Paper I and II). The cells were imaged with fluorescence microscopy

in biocompatibility tests (Paper I and II) and in Paper II also on the chip for

reference. A scanning electron microscope (Zeiss ULTRA Plus) was used to

confirm what was left on the chip after the cell experiment (Paper II).

3.3 PCB-PDMS microfluidic integration for salinity sensor

In Paper III a sensor for measuring salinity was presented. After optimizing the

sensor with electromagnetic simulation software (CST Studio), manufacture started

with the PCB part (29.5×29.5 mm2) which was first cut out from a two-sided FR-4

billet (1 mm thick with 18 µm copper metallization) using a computerized

numerical control (CNC) router (Roland mill, SRM-20, Roland DGA Corporation).

In the same process the fluid inlet/outlet holes and alignment holes were drilled in

44

the corners. A laser (LPKF) was used, with the aid of the alignment holes, to shape

the signal feedline and interdigitated electrode (250 µm / 250 µm, line / space

according to the simulations in Chapter 3) on the copper layer. On the bottom side

the copper was removed leaving a small area around the fluid inlet and outlet for

soldering the metal fittings.

An edge type of SubMiniature version A (SMA) connector was soldered to the

feedline and metal plugs for the fluidic tubing connections were soldered on the

bottom side. The whole structure was dipped vertically in PDMS (Sylgard 184,

Down Corning, USA) to form a smooth base layer for the fluidic components. After

dipping, the structure was left at room temperature for a few hours to remove excess

PDMS with the aid of gravity followed by curing in a box oven (60 °C for 3 h). A

needle was used to punch the fluid vias open. The thickness of the PDMS layer was

20 µm, measured with a micrometre calliper.

Modified soft lithography was used to manufacture the fluid layer with

channels and cavities. The master mould for this part was made on a microscope

glass slide using Kapton tape (thickness 60 µm). The Kapton tape was attached to

the glass and shaped with a laser to a serpentine pattern (width 250 µm) which was

selected for maximum mechanical stability. The cavity roof in this design does not

collapse compared to the situation where there is one big open cavity. With the

serpentine shape the area of the interdigitated electrode was covered. The mould

was placed on a petri dish and a specified amount of PDMS, measured with a scale,

was poured on top to form a 2.3 mm thick layer. Reduced-pressure treatment was

used to remove bubbles after which curing was done in a box oven (60 ºC for 3 h).

Finally, the PDMS layer was cut to specific size using a scalpel and then peeled off.

The fluid layer was placed at the top of the PCB part with a thin PDMS base

layer; the fluid vias were utilized for the alignment. PDMS layers are self-adherent

but this is not permanent and so to make the piece more robust it was again dipped

in PDMS to form a single monolithic structure. The fluidic connections were sealed

using a loop of silicone tubing during the last dipping process. The whole

manufacturing scheme of a microfluidic microwave sensor device is shown in

Figure 12.

45

Fig. 12. Manufacturing process for microfluidic microwave sensor.

Liquid flow through the microfluidic channel (width and height 250×60 µm2) was

generated by a miniature vacuum pump (DC 3V Micro Vacuum Air Pump, China).

A glass flask was used as a vacuum reservoir to make the pressure more stable and

to prevent liquid from going inside the pump, which was installed on the plastic cap.

Moreover, a manometer was also installed in the same part to observe the pressure.

The flask also served as a waste reservoir for the samples. A pipette tip was used as

the sample holder. The pump was controlled by a voltage source and an on-off

switch was installed for easy control. A USB camera (Mighty Scope, Aven Inc,

USA) was placed on top of the sensor to give a real time optical feedback of the

behaviour of the channel. The whole measurement system for measuring salinity is

shown in Figure 13 a)–b).

46

Fig. 13. Sample pumping system: a) schematic and b) fabricated system (Under CC BY

4.0 license from Paper III © 2019 Authors).

Bubbles are a major issue microfluidic systems, especially in the case of a

microwave sensor, because the air bubbles generate a lower effective permittivity

of the system, thus changing the interpretation of results. This issue was resolved

by using an isopropanol flush prior to sample loading which removed the bubbles.

The bubbles were generated when a new sample was passed through the system.

The USB camera was used to observe that the channel was bubble-free when

performing a measurement. An example of effective bubble removal is shown in

Figure 14 a)–b).

Fig. 14. Microfluidic system sample a) after sample loading with several bubbles in the

channel highlighted by black circles and b) after preceding isopropanol flush with no

visible bubbles (Under CC BY 4.0 license from Paper III © 2019 Authors).

47

3.4 Integrating split-ring resonator to SMA connector for

dehydration sensing

In Paper IV the prototype of the microwave sensor for dehydration monitoring was

manufactured directly on a SMA connector. First the pins of the SMA were

grounded to form a flat surface. Copper tape (Copper foil tape 1181, 3M, USA) was

attached to the flat part and then laser machined to form the complementary split

ring resonator shape which was designed by the electromagnetic simulation

presented in the next Chapter. The tape was soldered to the SMA connector’s body

to attach it robustly and a thin (< 20 µm) urethane coating (Urethane 71, CRC

Industries, USA) was used to protect the copper from oxidising. After curing of the

urethane in an oven at 60 ºC for 24 h the sensor was ready for measurements. The

manufacturing process is shown in Figure 15. The sensor was mounted to a wrist

strap made from PCB and flexible zip ties, shown in Figure 16. The size of the

sensor was 9.1×4.4×6.4 mm3.

Fig. 15. Manufacturing of CSRR-based dehydration sensor (Under CC BY 4.0 license

from Paper IV © 2019 Authors).

48

Fig. 16. The sensor mounted to a wrist strap (Under CC BY 4.0 license from Paper IV ©

2019 Authors).

3.4.1 The skin phantoms

Normal values for skin (acquired from literature [55]) were a relative permittivity

of 34.88 and a loss tangent of 0.34. The dehydrated skin model had less water than

normal which caused a lower relative permittivity and loss due to the increased

salinity. Thus, the dielectric values for hydrated and dehydrated skin were estimated

by changing the values as follows: in hydrated skin relative permittivity was

increased 5% and loss tangent −5%, dehydrated skin was −5% and +5%

respectively. These values were selected by studying earlier work [56], but more

conservative values were used. Moreover, the range was extended so that, in the

simulation, 10% changes were also shown in order to give a better insight into how

the sensor performed under larger changes. Dielectric values are presented in Table

4.

Table 4. Properties of the skin models used in simulation with different hydration levels,

value for normal skin from reference [55].

Model Relative permittivity tan

Dry skin 31.40 0.374

Dehydrated skin 33.14 0.357

Normal skin 34.88 0.340

Moist skin 36.63 0.323

Saturated skin 38.37 0.306

49

Skin phantoms used for real life testing in the laboratory were based on reference

[57], polyurethane rubber (Vytaflex 20, Smooth-On Inc, Texas) was selected as the

base material. Polyurethane rubbers are flexible, robust and resemble skin in their

mechanical properties, but the electrical properties are not the same. The goal was

to have a similar relative permittivity as in the modelling stage at 6 GHz with the

skin models (dehydrated, normal and hydrated). Pure urethane rubber has a low

relative permittivity and losses because it has virtually no free charges or dipoles

and losses arise mostly from electronic and atomic polarizations. However, when

conductive particles are added to the composite, the relative permittivity increases,

notably as the surface of the conductive particles becomes positively and negatively

charged by external field become polarized acting as dipoles [58]. By this means,

polymer and conductive particles are forming the matrix of small charged

capacitors. In this case, effective conductivity and dielectric losses of the composite

material increase due to the freely moving electrons in the conductive regions.

A loading of two types of carbon was used: graphite powder (Timcal, TIMREX

SFG75, Imerys, France) and carbon black powder (VULCAN VX72, FuelCellStore,

USA). These loadings were increased step by step to reach the targeted values, but

the mixing of the composite became difficult and degassing was impossible due to

the mixture becoming too thick to allow the gas to escape. The maximum loading

was 26 wt.% of graphite powder with 10 wt.% of carbon black powder with

dielectric properties of εr = 29.2 and tan δ = 0.154. To boost the relative permittivity,

barium strontium titanate ceramic powder (Ba0.65Sr0.35TiO3) with a high relative

permittivity was added to the composite. The ceramic powder was easier to mix

into the composite compared to the carbon materials. Then, a test set of a constant

graphite and carbon black loading of 24.3 wt.% and 8.2 wt.%, respectively, was

used with ceramic powder loadings of 10, 12.5 and 15 wt.% to achieve a composite

material exhibiting dielectric properties similar to those of natural skin with different

hydration levels. At 6 GHz, relative permittivity/tan δ were the following:

27.78/0.135, 33.53/0.179, 39.44/0.2. These values are also shown in Table 5. The

relative permittivity was close to the goals but the losses were not. However, the

test set was suitable for testing the sensor as the losses did not change the resonance

frequency but only the attenuation. This was confirmed by simulation with different

loss values while keeping the relative permittivity constant. Mechanical

characteristics were checked qualitatively by observing the sample after

measurement. Sensor contours were seen on the sample surface shown in Figure 17

and they were similar to those seen on sport watches used in real life.

50

Table 5. Dielectric properties of manufactured skin phantoms with same carbon black

(26 wt. %) and graphite powder (10 wt. %) loadings (Paper IV).

Skin Phantom Ba0.65Sr0.35TiO3 loading

(wt.%) εr′ tan δ

Dry 10.0 27.78 0.135

Normal 12.5 33.53 0.179

Moistured 15.0 39.44 0.200

Fig. 17. Sensor marks on the skin phantom after test. Inset: author’s wrist with clearly

visible marks after wearing a sports watch (Under CC BY 4.0 license from Paper IV ©

2019 Authors).

51

4 Results

4.1 The LTCC package and cell culture measurement

4.1.1 Chip generation I

In Paper I a complementary metal-oxide-semiconductor (CMOS) chip was

fabricated in a commercial foundry. The chip was based on the capacitive

measurement of cells by monitoring the overall change of capacitance when cells

attach themselves or die on the surface of the chip. The size of the chip was 3×3

mm2 and it had 40 contact pads (Al/TiN/Cu) at the edges, each one 85×85 μm2 in

size and with 150 μm pitch. More details about the chip can be found in reference

[59].

The LTCC substrate for the chip was manufactured from commercial DuPont

951 material and the conductors were screen-printed silver (DuPont 6142D).

Moreover, a package with a microfluidic channel was fabricated using a sacrificial

carbon tape (LTCC Carbon Tape, C12 Advanced Technologies, USA). The package

size was 5.3×1.9 cm2 and its thickness varied according to the package version. The

thickness with and without the microfluidic channel was 500 μm and 700 μm

respectively. Different packages are shown in Figure 18.

Fig. 18. Packaged sensors: (a) package bottom side showing silver conductors and the

chip in the middle, (b) package top side with cell well attached to the substrate as

sample holder, and (c) package top side with integrated microfluidic channel inlet and

outlet shown on top (Under CC BY 4.0 license from Paper I © 2016 Authors).

52

Fig. 19. Biocompatibility testing with BEAS2B cells: (a)–(c) the cells grow normally on

a silicon chip, and (d)–(f) cells on a DuPont 951 substrate showing normal cell

proliferation (Under CC BY 4.0 license from Paper I © 2016 Authors).

Biocompatibility of the Dupont 951 LTCC material was tested using BEAS2B cells

in which case the cells showed normal proliferation, shown in Figure 19. Electrical

connections were also performing as expected. Preliminary testing was performed

by measuring the DC resistance of a dummy chip with U shaped electrodes in an

incubator with cell media on top. Testing showed that the contacts were shielded

from the cell media and the resistivity of the dummy pattern remained constant

(average 26 ± 0.6 Ω). Then short-term measurement of the cell culture was

performed and shown in Figure 20. The shown voltage is output of the interconnect

CBCM, where the corresponding capacitance can be calculated.

4.1.2 Chip generation II

In Paper II, the chip was fabricated in a commercial foundry. The chip had an

integrated digital output (I2C) enabling the use of a smaller amount of cabling for

the electrical connections. Two sets of eight identical pads were positioned on the

opposite edges of the chip. Measurement of the cells was based on interdigitated

53

electrodes (IDE) which were excited by a three-stage ring oscillator circuit. Each

ring oscillator caused a delay yielding an oscillating signal and the IDE was

connected to one of the stages modulating the delay causing the change in the

frequency of oscillation. Each IDE had its own oscillating circuit. There were 16

sensors in total on each chip in a 4×4 array and data as a frequency from each sensor

was captured sequentially.

Fig. 20. Readout of the functioning chip as a function of time under different conditions

(Under CC BY 4.0 license from Paper I © 2016 Authors).

LTCC substrates were made using the same DuPont 951 material as in earlier

packages but due to the different packaging techniques the conductors were made

from gold paste (Dupont 5742). The new packaging technique was based on

thermocompression and there was no need for ICA in the bonding process.

Moreover, the underfill was changed to EPO-TEK 377, which was more reliable in

the case when the sensor was used multiple times. In addition, the cell well was

changed to glass and gluing was made with EPO-TEK 377 cured at 100 ºC. The

packages were cleaned using an oxygen plasma treatment (Oxford Plasmalab 80,

Oxford Instruments, UK) after the packaging process. Plasma treatment enabled

better cell adhesion. The sizes of the packages were 1.9×1.6 cm2 or 3.8×4.5 cm2.

The larger size package enabled the use of microscope imaging of the cells. The

packages are shown in Figure 21.

54

Biocompatibility of the new underfill was rechecked as in Paper I and it

performed similarly to the earlier version. The cell experiment was made using

human lung epithelial cells showing an electrical signal change when cells attached,

multiplied and detached. The number of cells could be counted on each sensor with

the aid of optical microscopy. The detachment of cells was induced by a trypsin

treatment. The results were confirmed using microscopic imaging and cells were

stained using Hoechst DNA dye. Results are shown in Figure 22, where the

resonance frequency corresponded to the number of cells on each IDE. The

frequency change was from 471 kHz up to 958 kHz when there were only cells on

the edges of the electrode, and covered with cells, respectively.

Fig. 21. Smaller (top row) and larger (bottom row) chip packages suitable for

microscope imaging, are shown in bottom (Under CC BY 4.0 license from Paper II ©

2018 Authors).

55

Fig. 22. The sensor frequency shift response to cell media (top) and to cell proliferation

and trypsin addition (middle), and pictures of the grown cells on sensor elements

(bottom). Higher frequency shift on IDE sensor 11 is due to larger amount of cells in

comparison to other IDEs (Under CC BY 4.0 license from Paper II © 2018 Authors).

4.2 Microfluidic microwave sensor for blood salinity monitoring

In Paper III, the design of the sensor for measuring salinity was performed with

electromagnetic simulation software (CST studio) and based on an interdigitated

electrode (IDE) microwave resonator. The IDE generated three resonances on the

used bandwidth (0.2–6 GHz) and was selected to give several resonances in the

limited physical area. The sensor was manufactured using printed circuit board

(PCB) as the electronic parts and polydimethylsiloxane (PDMS) for the

microfluidic channel. Both materials are widely used in their fields and integration

of PCB and PDMS enabled realization of a hybrid device with an electrical sensor

and microfluidic channels. The sensor is shown in Figure 23.

The total volume of the serpentine type fluid channel was 1 μl (one drop of

blood is enough for about 50 samples). The pumping system was based on a small

56

vacuum pump sucking the sample through the fluid channel which was in contact

with the sensor. Resonances were monitored using a one-port vector network

analyser (Anritsu MS46121AB). The sensor was characterized using different

isopropanol and water mixtures (0, 20, 40, 60, 80 and 100 wt.%). Then a test set

from 125 to 155 mmol of saline water was prepared resembling the salinity of blood.

Fig. 23. Schematic presentation of the microwave sensor with microfluidics (Under CC

BY 4.0 license from Paper III © 2019 Authors).

4.2.1 Modelling of the microfluidic sensor

Simulation (CST Microwave Studio) was performed in the time domain and the

port was placed on the feedline and an automatic adaptive meshing setup was used.

FR-4 was used as the substrate, which had a relative permittivity of 4.5 and tan δ

from 0.016 to 0.003 from 1 to 5 GHz, respectively. Copper was simulated as a lossy

metal and values were from the CST material library. The PDMS layer was

measured using a Speag dielectric assessment kit and the dielectric properties were

imported to the simulation (εr = 2.6–2.675, tan δ = 0.02–0.0002 from 1 to 5 GHz)

showing a steady relative permittivity and low losses at high frequency. The sensor

was simulated with air and water samples. Distilled water was measured with a

57

Speag DAK and the results were imported to the model (εr = 78.2–73.6, tan δ =

0.0002–0.02 at 1–5 GHz).

Surface currents of the system were investigated to observe their distributions

at different resonance frequencies. The sample in the liquid channel was set as

vacuum. At the first resonance frequency (0.9 GHz) the corners of the finger

structure had the most intensive current. At the second resonance frequency (3.2

GHz) the current was spread also on the centre area of the structure and sides. The

third resonance frequency (5.5 GHz) was at its most uniform around the

interdigitated electrode. In Figure 24 the current distributions are shown for each

resonance. Current distributions was used in designing to optimize the current

density in the location of the microfluidic channel on top of the IDE to maximize

sensitivity. The serpentine liquid channel shown in Figure 25 was made to conform

to the IDE structure. As can be seen from the current distributions, the channel was

on top of the high intensity sites seen as red color in the Figure 24.

Fig. 24. Simulation of surface current distributions at different resonance frequencies

of a) 0.9 GHz, b) 3.2 GHz, and c) 5.5 GHz.

The sensor was then simulated with a water sample. The S11 curve as a function of

frequency is shown in Figure 26. All the resonances were shifted to lower

frequencies due to the higher relative permittivity of water. The shape was not

changed significantly with the 1st and 2nd resonance but the 3rd spike was wider. The

three resonance frequencies were 1.024, 3.106 and 5.189 GHz for air, and 1.006,

3.077 and 5.113 GHz for water. Attenuation for air was –0.785, –7.019 and –14.777

dB and for water –0.901, –9.285 and –6.40 dB respectively. It can be seen that the

resonance frequencies were reduced as expected when water was present.

58

Fig. 25. The serpentine liquid channel with channel width of 250 μm, channel height of

60 μm, and total volume of 1 μl.

Fig. 26. Simulated sensor response to air and water. Resonances are marked with

numbering (Under CC BY 4.0 license from Paper III © 2019 Authors).

4.2.2 Measurements with the proposed sensor

A test set of isopropanol-water mixtures was prepared to characterize the sensor.

Volume fractions of isopropanol were 0, 20, 40, 60, 80 and 100 wt.%. Samples were

characterized with a Speag DAK used as “standard” and results are shown in Figure

27 a)–b). The relative permittivity decreased as the isopropanol ratio increased.

This can be easily seen from the data due to the pure isopropanol yield in relative

permittivity below 15 and pure water around 80, thus permittivities between those

values can be achieved by mixing. Losses increased around 1 GHz as the

isopropanol volume increased. At 3 GHz the pure isopropanol sample showed

59

lower or similar losses compared to lower volume fractions of isopropanol. At 5

GHz the 80 wt.% solution also showed similar behaviour.

Fig. 27. Dielectric properties of isopropanol-water mixtures: a) relative permittivity and

b) dielectric losses. Regions of resonance frequencies are marked with mesh pattern

(Under CC BY 4.0 license from Paper III © 2019 Authors).

Measurements with the developed sensor are shown in Figure 28 and 29. The first

resonance (~0.9 GHz, Figure 28 a)) showed some shift to higher frequencies as the

relative permittivity of isopropanol was lower. The attenuation in sensor increased

according to the losses measured with the Speag DAK. The second resonance (~3

GHz, Figure 28 b)) showed a greater change of resonance frequency, however the

attenuation does not behave according to the DAK measured values. The third

resonance (~5.4 GHz, Figure 28 c)) behaved similarly to the 2nd resonance.

Fig. 28. Measurement results with the microwave sensor: a) first, b) second, and c) third

resonance (Under CC BY 4.0 license from Paper III © 2019 Authors).

60

Fig. 29. Measured values by developed microwave sensor vs the DAK measured

dielectric properties. a) The first resonance frequency attenuation vs. the dielectric

losses, b) second resonance frequency vs. the relative permittivity and c) third

resonance vs. the relative permittivity (Under CC BY 4.0 license from Paper III © 2019

Authors).

The sensor measurement results (attenuation and resonance frequency) are plotted

as a function of dielectric properties measured with the Speag DAK in Figure 29.

The plots show that all resonance frequencies are useful due to the relatively easy

linear calibration when measuring volume fractions of isopropanol in water,

because of the shift in resonance frequency. However, when concerning about the

attenuation only the 1st resonance was useful due to correlation with the dielectric

losses of the samples. To compare the three resonance frequencies a linear fitting

was used to give an easy to interpret qualitative comparison. The most accurate

linear correlation (R2 = 0.97, coefficient of determination) between attenuation and

dielectric loss was found for the first resonance frequency (Figure 29 a)). For the

2nd resonance a similar correlation (R2 = 0.95, coefficient of determination) could

be seen with the resonance frequency and measured relative permittivity (Figure

29 b)) while the correlation was lower for 3rd resonance (R2 = 0.85, coefficient of

determination, Figure 29 c)). The simulations showed that with a higher frequecy,

especially accompanied with a sample having low permittivity, the IDE structure

61

started to behave as an antenna and unwanted radiation was present. Simulation

results are shown in Figure 30. This was one reason for the unlinear behaviour.

Lower resonance frequencies did not show increased radiation when comparing air

and water sample radiation patterns as can be seen from Figure 30 a) and b). Data

was analyzed using an AutoCAD area calculator. A polygon with 400 V/m electric

field treshold was drawn around the radiation pattern and a 30% larger radiation

area was seen at 3rd resonance with the air sample compared to water.

Fig. 30. Radiation patterns as electric field strengths were inspected. a) With 1st

resonance the radiation patterns are close to the structure and similar for air and water

sample. b) More radiation compared to 1st resonance, but there is no relative change

between air and water. c) There is more radiation with air compared to water (Under CC

BY 4.0 license from Paper III © 2019 Authors).

62

To demonstrate the measurement of salinity, a set of natrium chloride (NaCl)

solutions was prepared. A test set contained 0, 125, 135, 145 and 155 mmol of NaCl

mixed with water, as a weight this corresponds to 0 to 9 g/l, respectively. Saline

samples were first measured with the Speag DAK and results are shown in Figure

31 a) and b). When salinity increases the relative permittivity decreases, except at

frequencies below 200 MHz where the interfacial polarization, the Maxwell-

Wagner effect, causes higher permittivity. For example, positive ions (Na+) move

towards the interface (metal-dielectric) of a negatively charged surface causing net

polarization. The same also happens for negative ions (Cl–) and the opposite

interface accumulates a positive charge. When the electric field is switched the Na+

ions are pushed away towards the negative ions (Cl–) causing increased loss due to

the friction. At higher frequencies this polarization mechanism is no longer

significant and NaCl decrease the relative permittivity due to the ions not being able

to move with the pace of the exciting electric field. However, water molecules

(dipoles) rotate according to the electric field and cause a net polarization. Losses

are significantly higher at lower frequencies where the first and second resonance

are present. Na+ and Cl– ions make the liquid more conductive and the movement

of these ions causes increased absorption of the electromagnetic wave. At higher

frequencies, this movement diminished due to the inertia of mass, and the loss

became closer to that of pure water. Measurement results with the microfluidic

sensor are shown in Figure 32 a)–c). In the first resonance (Figure 32 a)) the

attenuation shows that the losses increased, as indicated by the Speag DAK

measured loss. Moreover, the frequency shifted to lower values as the NaCl

concentration increased. Figure 32 b) shows the increased attenuation compared to

pure water with the resonance also shifting to lower frequencies. At third resonance

(Figure 32 c)) all samples showed similar curves and therefore this was not usable

in this application. The sensor indicated that the samples had a higher relative

permittivity in all three resonances as the resonance frequency became lower, in

contrast to the Speag DAK measured values which showed a reduced relative

permittivity (Figure 31 a)). The reason could be the serpentine shape of the channel

filled with the conductive liquid. It provides a coil type structure causing inductance

in the system and lowering the resonance frequency (Eq. 8). However, the

attenuation of the first resonance shows that the sensor could be applied to measure

salinity, as can be seen in Figure 33 where the attenuation is plotted against the loss

values measured with the Speag DAK demonstrating a highly linear correlation.

63

Fig. 31. Dielectric properties of saline water: a) relative permittivity and b) dielectric

losses (Under CC BY 4.0 license from Paper III © 2019 Authors).

Fig. 32. Sensor response to the different saline concentrations: a) first resonance, b)

second resonance, and c) third resonance (Under CC BY 4.0 license from Paper III ©

2019 Authors).

64

Fig. 33. Attenuation of the developed microwave sensor vs. dielectric losses measured

with DAK at 0.9 GHz with different salinity concentrations (Under CC BY 4.0 license from

Paper III © 2019 Authors).

4.3 Non-Invasive dehydration sensor based on split ring resonator

Sensing of dehydration was performed using a microwave resonator designed and

modelled (CST Studio) for sensing the skin in real time. The sensor was tested by

using tailored skin phantoms. Moreover, the sensor was tested in an extended

timeframe using water-saturated polyester tissue, which during drying generated a

wide range of effective dielectric properties. This gave long term timespan (48 h)

information of how the sensor behaved.

4.3.1 Modelling of the dehydration sensor

A complementary split ring resonator (CSRR) was selected as the sensing element

since promising results have been reported in earlier work by other researchers in

different applications such as gas and ethanol sensing [60]−[62]. Due to the

intended application of measuring dehydration from skin, the CSRR was already

included in the simulation in the first step.

The dimensions of the CSRR were adjusted so that the resonance frequency

was around 6 GHz, where the Maxwell-Wagner polarization is not dominant. Large

errors may have occurred due to sweating because precipitating salt on the skin may

interfere with the skin-sensor interface, especially at low frequencies. Choosing

higher frequencies can minimise this effect. On the other hand, the relative

65

permittivity of water is still high around 6 GHz (εr = 72). The penetration depth of

the electromagnetic wave was also taken into account, since the skin upper layer

was the target of the measurement. If the signal went deeper, it could cause

measurement errors due to the sensing of the underlying tissues (fat, muscle and

bone). The dimensions of the probe and the resonance frequency were adjusted so

that most of the power was concentrated at a depth of 1 mm.

The selected dimensions of the probe are shown in Figure 34. Simulation

results with an extended frequency range (±10%) are shown in Figure 35 a). The

resonance frequency shifted to a higher frequency in the dehydration state as

expected. As designed, the resonance frequency shift correlated with the hydration

state. The penetration of the signal is shown in Figure 35 b), where the attenuation

was high (> –35 dB) when the depth was more than 1 mm. Resonance frequencies

are shown in Table 6.

Table 6. Simulated resonance frequencies with different skin models (Paper IV).

Model Resonance frequency (GHz) Attenuation (dB)

Dry skin 5.70 −27.5

Dehydrated skin 5.53 −26.4

Normal skin 5.45 −25.9

Moist skin 5.29 −25.7

Saturated skin 5.19 −26.0

Fig. 34. The dimensions of the designed CSRR resonator (Under CC BY 4.0 license from

Paper IV © 2019 Authors).

66

Fig. 35. The simulation results from the dehydration sensor: a) results with different

states of hydration and b) penetration depth of the electric field (Under CC BY 4.0

license from Paper IV © 2019 Authors).

4.3.2 Measurements and simulations with the developed dehydration

sensor

First, when the sensor was measured without a sample, no resonance was observed.

Then a dehydrated skin phantom sample was measured and the resonance

frequency of 5.89 GHz was obtained. The measurement setup is shown in Figure

36. Normal skin and hydrated skin phantoms had resonance frequencies of 5.52

and 5.3 GHz, where attenuation decreased from −9 dB to −13 dB, repectively. The

results are shown in Figure 37 a) with corresponding additional simulation results.

Resonances from additional simulations with the same dielectric values as the

manufactured skin phantoms (6.09, 5.53 and 5.11 GHz) correlated well with the

measured values shown in same Figure 37 a). In Figure 37 b) the resonance

frequency results of the microwave sensor are plotted as a function of relative

permittivity measured with the Speag DAK, yielding a linear fitting of r2 = 0.99

(coefficient of determination). Results showed that samples imitating different

hydration states were easy to recognize.

As the sensor was planned for real time monitoring, a longer-term

measurement was also performed in which a naturally drying polyester tissue was

measured for 48 h and measurement results are shown in Figure 38. The sensor

showed a response over a wide relative permittivity range, the limiting factor was

67

the low relative permittivity; when the permittivity was close to that of air the

resonance was lost. However, in the range of permittivity of skin (εr from 20 to 45)

the sensor performed as designed.

Fig. 36. Measurement setup for skin phantoms. Sample is sandwiched between a plastic

cylinder and the sensor (Under CC BY 4.0 license from Paper IV © 2019 Authors).

Fig. 37. The measurement results from the dehydration sensor: a) measurement results

with skin phantoms and corresponding simulation results and b) resonance frequency

vs. relative permittivity of the skin phantoms plotted (Under CC BY 4.0 license from

Paper IV © 2019 Authors).

68

Fig. 38. Sensor response to the drying polyester tissue as function of time. The

resonance is lost after approximately 22 h (Under CC BY 4.0 license from Paper IV ©

2019 Authors).

69

5 Discussion

The first package presented in Paper I was based on a flip-chip technique having

40 pads, all of which needed to be connected to the conductors on the substrate, thus

posing difficulties. This problem was resolved by using conductive and non-

conductive epoxies to provide a barrier between the liquid and the electrical

connections to the chip. Conductive epoxy was placed on the substrate with a

tailored alumina stamp. Moreover, a version where the microfluidic channels were

integrated into the package was demonstrated. The 2nd generation chip in Paper II

had fewer contact pads and the packaging process was changed to take advantage

of thermocompression bonding. This also enabled mass production and fewer

processing steps were needed. Comparing this to wire bond style packages [31],

[32], [63]. The benefits of flip-chip bonding are clear as there is no need for special

techniques to protect the active area of the chip during covering of the wires. The

same feature exists in chip-in-hole style packages [30], [33]. Chip-in-hole derives

some benefits from the relatively flat surface where the microfluidics can be

integrated. However, in the LTCC based package, presented in this work, the

microfluidic channels can be directly manufactured on the package.

In Paper III a resonator with 3 resonating modes at frequencies between 0.7–6

GHz was simulated, manufactured and tested. The system was designed for easy

use with an integrated pumping system. This type of sensor could be suitable to

measure different types of body liquids such as blood, sweat, urine or cerebrospinal

fluid (CSF), which can give indications of certain diseases or problems in humans.

For example, if blood is detected in the CSF it suggests the possibility of a major

health issue. The device was manufactured from printed circuit board, which has

renowned mass production capabilities, and PDMS which is known as a good

material in prototyping or microfluidics and sensor demonstrations. In final tests

with the saline water the sensor showed correspondence of the attenuation of the

1st resonance and amount of saline.

The size of the dehydration sensor, presented in Paper IV, was small (area in

contact with skin 6.4×6.4 mm2) compared to the work presented by other

researchers [56], [64]–[67]. In spite of this small size it yielded clear results from

which distinctions between hydration states in skin phantoms could easily be made.

In the future, the structure could be integrated to a sports watch type of application.

However, before moving to human experiments the vector network analyzer (VNA)

needs to be miniaturized and integrated in conjunction with the sensor. This could

be engineered to generate a signal in a suitable frequency band measuring reflected

70

power. This could be achieved by a voltage-controlled oscillator (VCO) and the

reflected power may be determined using off-the-shelf components.

71

6 Summary and outlook

In this thesis biosensors have been the focus, especially microwave sensors. Two

types of packages for CMOS-based cell culture biosensors were developed where

packages based on LTCC material were proven to be well suited for small sized

CMOS biosensors. The materials and electromagnetic theory behind the response

of the sensors were discussed. Secondly, a sensor for salinity measurement,

including a sample pumping system, was designed and manufactured. Thirdly, a

microwave sensor for real-time dehydration measurement of the skin was

developed.

CMOS based sensor chips are well suited for mass production and enable the

integration of a microprocessor and electronics to drive the sensor elements.

However, these sensor elements need to be in contact with the biological materials

under measurement while the chip size is small. Thus, the packaging is challenging

due to the need for the electrical contacts to be covered, and the materials should

also be biocompatible. LTCC technology was proven to be suitable for the

packaging of these CMOS based biosensors.

Electrical connections to the substrate was made by conductive adhesive which

was stamped onto the substrate. Since the contact pads of the chip were oxidized,

the pads on the chip were gold bumped to achieve a lower resistance and better

reliability of the connection. The chip was bonded to the substrate and the adhesive

was cured. An epoxy underfill was used to seal the gap between the chip and the

substrate. A well for the cell culture made from polypropylene was glued on top of

the package.

Due to the flexibility of the LTCC process these ceramic packages can be easily

customised for different purposes. For example, a small sized package for normal

usage and a larger one, enabling simultaneous optical measurements and

observations by microscope, was demonstrated. Packages proposed in this thesis

have been used on several occasions in the USA and Sweden [68], [69] for cell

experiments, hence proving their durability. In the future, the LTCC technique could

be used to realise a standalone system for cell culture measurement with heaters,

temperature and pH sensors to control the environment for incubation in the

package. Such a package would allow sterilization of the sensor with chemicals

(ethanol) or a heat/steam-based technique. This would enable sensor reusability,

thus reducing the overall cost. The greatest disadvantage of the LTCC package is

the opaque nature of its material but this can be solved by integrating transparent

materials to the package (e.g. plastic or glass lids [70]). However, if the sensor is

72

well characterized, optical inspection is not needed, and the system could be used

for example for drug screening. Data about cell viability would be gathered

automatically without human intervention, thus decreasing the cost of the screening

and enabling continuous monitoring instead of only end-point inspection.

All biological materials have different dielectric properties making microwave

sensors a powerful instrument for monitoring biological activity. By measuring

NaCl (ion) concentration from blood, hydration balance could be determined.

Microfluidics allow the use of samples with a small volume and a microwave sensor

with multiple resonances for monitoring the sample. The resonator developed in

this work was based on interdigitated finger electrodes and sufficient electrical

loading of the medium under measurement. The NaCl concentration (osmolality)

in the medium (water) could be determined. The loss tangent measured by the

microwave-microfluidic sensor was shown to correlate with the concentration of

NaCl in water around the 1 GHz measurement frequency. Groundwork for

developing this type of sensor was established including a manufacturing process

and sample handling with an integrated pumping system. In the future, further tests

with blood should be carried out in order to take other elements into account and to

determine if osmolality can be measured selectively.

To monitor dehydration in vivo, a micromachined split ring resonator was

manufactured and tested in the laboratory using tailored skin phantoms. The design

was based on CSRR and it was optimized using electromagnetic simulation. The

sensor itself worked and showed clear resonance frequency shift when the hydration

state was changing. The results show the feasibility of this sensor for detection of

the moisture level in the skin, but practical use requires further studies and

development. If successful the sensor could provide great advantages in the

monitoring of human activity and dehydration, similar to the heart rate monitoring

which is nowadays a standard procedure.

73

List of references

[1] L. C. Clark, R. Wolf, D. Granger, and Z. Taylor, “Continuous recording of blood oxygen tensions by polarography.,” J. Appl. Physiol., vol. 6, no. 3, pp. 189–193, 1953.

[2] L. C. Clark and C. Lyons, “Electrode systems for continuous monitoring in cardiovascular surgery,” Ann. N. Y. Acad. Sci., vol. 102, no. 1, pp. 29–45, 1962.

[3] G. G. Guilbault and J. G. Montalvo, “A Urea-Specific Enzyme Electrode,” J. Am. Chem. Soc., vol. 91, no. 8, pp. 2164–2165, 1969.

[4] A. Manz, N. Graber, and H. M. Widmer, “Miniaturized total chemical analysis systems: A novel concept for chemical sensing,” Sensors Actuators B. Chem., vol. 1, pp. 244–248, 1990.

[5] S. C. Terry, J. H. Herman, and J. B. Angell, “A Gas Chromatographic Air Analyzer Fabricated on a Silicon Wafer,” IEEE Trans. Electron Devices, vol. 26, no. 12, pp. 1880–1886, 1979.

[6] E. Verpoorte and N. F. De Rooij, “Microfluidics meets MEMS,” Proceedings of the IEEE, vol. 91, no. 6. pp. 930–953, 2003.

[7] “Global Lab On a Chip Market Size, Share, Trends, COVID-19 Impact & Growth Analysis Report – Segmented By Product Type, Technology & Region (North America, Europe, Asia Pacific, Latin America, Middle East, and Africa) – Industry Forecast (2021 to 2026)," Market Data Forecast, Inc., Hyderabad, India. Accessed: Jun., 5, 2021. [Online]. Available: https://www.marketdataforecast.com/market-reports/lab-on-a-chip-market.

[8] G. Nabovati, E. Ghafar-Zadeh, M. Mirzaei, G. Ayala-Charca, F. Awwad, and M. Sawan, “A new fully differential CMOS capacitance to digital converter for lab-on-chip applications,” IEEE Trans. Biomed. Circuits Syst., vol. 9, no. 3, pp. 353–361, 2015.

[9] S. B. Prakash and P. Abshire, “Tracking cancer cell proliferation on a CMOS capacitance sensor chip,” Biosens. Bioelectron., vol. 23, no. 10, pp. 1449–1457, 2008.

[10] N. Couniot, L. A. Francis, and D. Flandre, “A 16 ?? 16 CMOS Capacitive Biosensor Array Towards Detection of Single Bacterial Cell,” IEEE Trans. Biomed. Circuits Syst., vol. 10, no. 2, pp. 364–374, 2016.

[11] C. Xiao and J. H. T. Luong, “On-line monitoring of cell growth and cytotoxicity using electric cell-substrate impedance sensing (ECIS),” Biotechnol. Prog., vol. 19, no. 3, pp. 1000–1005, 2003.

[12] J. C. Chien and A. M. Niknejad, “Oscillator-Based Reactance Sensors With Injection Locking for High-Throughput Flow Cytometry Using Microwave Dielectric Spectroscopy,” IEEE J. Solid-State Circuits, vol. 51, no. 2, pp. 457–472, 2016.

[13] N. Couniot, D. Bol, O. Poncelet, L. A. Francis, and D. Flandre, “A capacitance-to-frequency converter with on-chip passivated

74

microelectrodes for bacteria detection in saline buffers up to 575 MHz,” IEEE Trans. Circuits Syst. II Express Briefs, vol. 62, no. 2, pp. 159–163, 2015.

[14] J. C. Chien, A. Ameri, E. C. Yeh, A. N. Killilea, M. Anwar, and A. M. Niknejad, “A high-throughput flow cytometry-on-a-CMOS platform for single-cell dielectric spectroscopy at microwave frequencies,” Lab Chip, vol. 18, no. 14, pp. 2065–2076, 2018.

[15] T. Mitsunaka et al., “CMOS Biosensor IC Focusing on Dielectric Relaxations of Biological Water with 120 and 60 GHz Oscillator Arrays,” IEEE J. Solid-State Circuits, vol. 51, no. 11, pp. 2534–2543, 2016.

[16] H. Wang, A. Mahdavi, D. A. Tirrell, and A. Hajimiri, “A magnetic cell-based sensor,” Lab Chip, vol. 12, no. 21, p. 4465, 2012.

[17] J. N. Lee, C. Park, and G. M. Whitesides, “Solvent Compatibility of Poly(dimethylsiloxane)-Based Microfluidic Devices,” Anal. Chem., vol. 75, no. 23, pp. 6544–6554, 2003.

[18] P. Karásek, J. Grym, M. Roth, J. Planeta, and F. Foret, “Etching of glass microchips with supercritical water,” Lab Chip, vol. 15, no. 1, pp. 311–318, 2015.

[19] C. E. Hoyle and C. N. Bowman, “Thiol-ene click chemistry,” Angewandte Chemie - International Edition, vol. 49, no. 9, pp.1540-1573, 2010.

[20] B. K. Gale et al., “A review of current methods in microfluidic device fabrication and future commercialization prospects,” Inventions, no. 3, p. 60, 2018.

[21] Y. Imanaka, Multilayered low temperature cofired ceramics (LTCC) technology. New York, NY, USA: Springer, 2005.

[22] M. Sobocinski, J. Putaala, and H. Jantunen, “Multilayer low-temperature co-fired ceramic systems incorporating a thick-film printing process,” in Printed Films: Materials Science and Applications in Sensors, Electronics and Photonics. Cambridge, UK: Woodhead Publishing Ltd., 2012, pp. 134–164.

[23] M. R. Gongora-Rubio, P. Espinoza-Vallejos, L. Sola-Laguna, and J. J. Santiago-Avilés, “Overview of low temperature co-fired ceramics tape technology for meso-system technology (MsST),” Sensors Actuators A Phys., vol. 89, no. 3, pp. 222–241, Apr. 2001.

[24] K. A. Peterson et al., “Novel Microsystem Applications with New Techniques in Low-Temperature Co-Fired Ceramics,” Int. J. Appl. Ceram. Technol., vol. 2, no. 5, pp. 345–363, Sep. 2005.

[25] M. Sobocinski, J. Juuti, H. Jantunen, and L. Golonka, “Piezoelectric unimorph valve assembled on an LTCC substrate,” Sensors Actuators, A Phys., vol. 149, no. 2, pp. 315–319, Feb. 2009.

[26] T. Reinecke et al., “Continuous noninvasive monitoring of cell growth in disposable bioreactors,” Sensors Actuators, B Chem., vol. 251, pp. 1009–1017, 2017.

75

[27] H. Bartsch, T. Welker, K. Welker, H. Witte, and J. Müller, “LTCC based bioreactors for cell cultivation,” IOP Conf. Ser. Mater. Sci. Eng., vol. 104, p. 012001, 2016.

[28] H. B. De Torres, C. Rensch, M. Fischer, A. Schober, M. Hoffmann, and J. Müller, “Thick film flow sensor for biological microsystems,” Sensors Actuators, A Phys., vol. 160, no. 1–2, pp. 109–115, 2010.

[29] H. Bartsch et al., “Surface properties and biocompatibility of thick film materials used in ceramic bioreactors,” Materialia, vol. 5, no. 100213, 2019.

[30] T. Datta-Chaudhuri, P. Abshire, and E. Smela, “Packaging commercial CMOS chips for lab on a chip integration.,” Lab Chip, vol. 14, no. 10, pp. 1753–66, 2014.

[31] H. Wang, A. Mahdavi, D. A. Tirrell, and A. Hajimiri, “A magnetic cell-based sensor,” Lab Chip, vol. 12, no. 21, pp. 4465–4471, 2012.

[32] Lin Li and A. J. Mason, “Post-CMOS parylene packaging for on-chip biosensor arrays,” in 2010 IEEE Sensors, 2010, pp. 1613–1616.

[33] S. Lu, B. Senevirathna, M. Dandin, E. Smela, and P. Abshire, “System Integration of IC chips for Lab-on-CMOS Applications,” in 2018 IEEE International Symposium on Circuits and Systems (ISCAS), 2018, vol. 2018-May, pp. 1–5.

[34] R. Huys et al., “Single-cell recording and stimulation with a 16k micro-nail electrode array integrated on a 0.18 μm CMOS chip,” Lab Chip, vol. 12, no. 7, p. 1274, 2012.

[35] A. Wu, L. Wang, E. Jensen, R. Mathies, and B. Boser, “Modular integration of electronics and microfluidic systems using flexible printed circuit boards.,” Lab Chip, vol. 10, no. 4, pp. 519–521, 2010.

[36] B. Zhang, Q. Dong, C. E. Korman, Z. Li, and M. E. Zaghloul, “Flexible packaging of solid-state integrated circuit chips with elastomeric microfluidics,” Sci. Rep., vol. 3, no. 1, p. 1098, Dec. 2013.

[37] N. Halonen et al., “Low temperature co-fired ceramic package for lab-on-CMOS applied in cell viability monitoring,” Procedia Eng., vol. 120, pp. 1079–1082, 2015.

[38] A. R. von Hippel, Ed., Dielectric materials and applications. Cambridge, MA, USA: The MIT Press, 1952.

[39] E. Nyfors, “Industrial Microwave Sensors---A Review,” Subsurf. Sens. Technol. Appl., vol. 1, no. 1, pp. 23–43, 2000.

[40] P. Jakkula and E. Tahkola, “Microwave Sensor System for the Consistency Measurement in the Pulp and Paper Industry,” Sensors Updat., vol. 7, no. 1, pp. 211–232, Jan. 2000.

[41] S. Kim et al., “Noninvasive in vitro measurement of pig-blood d-glucose by using a microwave cavity sensor,” Diabetes Res. Clin. Pract., vol. 96, no. 3, pp. 379–384, 2012.

[42] F. I. Jamal et al., “Low-Power Miniature K -Band Sensors for Dielectric Characterization of Biomaterials,” IEEE Trans. Microw. Theory Tech., vol.

76

65, no. 3, pp. 1012–1023, 2017. [43] A. Salim, S. H. Kim, J. Y. Park, and S. Lim, “Microfluidic Biosensor Based

on Microwave Substrate-Integrated Waveguide Cavity Resonator,” J. Sensors, vol. 2018, p. 1324145, 2018..

[44] F. I. Jamal, S. Guha, M. H. Eissa, S. Vehring, D. Kissinger, and C. Meliani, “K-Band BiCMOS based near-field biomedical dielectric sensor for Detection of fat and calcium in blood,” in 2015 European Microwave Conference (EuMC), 2015, pp. 821–824..

[45] J. C. Booth, N. D. Orloff, J. Mateu, M. Janezic, M. Rinehart, and J. A. Beall, “Quantitative permittivity measurements of nanoliter Liquid volumes in microfluidic channels to 40 GHz,” IEEE Trans. Instrum. Meas., vol. 59, no. 12, pp. 3279–3288, 2010.

[46] X. Bao et al., “A Planar One-Port Microwave Microfluidic Sensor for Microliter Liquids Characterization,” IEEE J. Electromagn. RF Microw. Med. Biol., vol. 2, no. 1, pp. 10–17, Mar. 2018.

[47] A. Salim and S. Lim, “TM02quarter-mode substrate-integrated waveguide resonator for dual detection of chemicals,” Sensors (Switzerland), no. 6, pp. 1964, 2018.

[48] M. T. Jilnai, W. P. Wen, L. Y. Cheong, and M. Z. U. Rehman, “A microwave ring-resonator sensor for non-invasive assessment of meat aging,” Sensors (Switzerland), vol. 16, no. 1, pp. 52 2016.

[49] R. L. Shao, Y. J. Zhou, and L. Yang, “Quarter-mode spoof plasmonic resonator for a microfluidic chemical sensor,” Appl. Opt., vol. 57, no. 28, p. 8472, Oct. 2018.

[50] G. Yesiloz, M. S. Boybay, and C. L. Ren, “Label-free high-throughput detection and content sensing of individual droplets in microfluidic systems,” Lab Chip, vol. 15, no. 20, pp. 4008–4019, 2015.

[51] B. M. Popkin, K. E. D’Anci, and I. H. Rosenberg, “Water, hydration, and health,” Nutr. Rev., vol. 68, no. 8, pp. 439–458, Aug. 2010.

[52] J. H. Ahn et al., “Meningeal lymphatic vessels at the skull base drain cerebrospinal fluid,” Nature, vol. 572, no. 7767, pp. 62–66, 2019.

[53] J. Hakala, J. Kilpijarvi, M. Sarestoniemi, M. Hamalainen, S. Myllymaki, and T. Myllyla, “Microwave Sensing of Brain Water – a Simulation and Experimental Study Using Human Brain Models,” IEEE Access, vol. 8, pp. 111303–111315, 2020.

[54] M. Tanaka and M. Sato, “Microwave heating of water, ice, and saline solution: Molecular dynamics study,” J. Chem. Phys., vol. 126, no. 3, p. 034509, 2007.

[55] P. Hasgall et al., “IT’IS Database for thermal and electromagnetic parameters of biological tissues, Version 4.0,” IT’IS, 2018. [Online]. Available: https://www.itis.ethz.ch/virtual-population/tissue-properties/database/database-summary/.

[56] F. Trenz, V. Kalpen, R. Weigel and D. Kissinger, "Modeling and

77

measurement of a tissue-equivalent liquid for noninvasive dehydration sensing in the 2.45 GHz ISM-band," 2015 IEEE MTT-S International Microwave Workshop Series on RF and Wireless Technologies for Biomedical and Healthcare Applications (IMWS-BIO), pp. 32-33, 2015.

[57] J. Garrett and E. Fear, "Stable and Flexible Materials to Mimic the Dielectric Properties of Human Soft Tissues," IEEE Antennas and Wireless Propagation Letters, vol. 13, pp. 599-602, 2014.

[58] E. Nyfors and P. Vainikainen, Industrial microwave sensors. Norwood, MA, USA: Artech House, 1989.

[59] S. B. Prakash and P. Abshire, “A fully differential rail-to-rail CMOS capacitance sensor with floating-gate trimming for mismatch compensation,” IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 56, no. 5, pp. 975–986, May 2009.

[60] A. Bogner, C. Steiner, S. Walter, J. Kita, G. Hagen, and R. Moos, “Planar microstrip ring resonators for microwave-based gas sensing: Design aspects and initial transducers for humidity and ammonia sensing,” Sensors (Switzerland), vol. 17, no. 10, p. 2422, 2017.

[61] A. Salim and S. Lim, “Complementary split-ring resonator-loaded microfluidic ethanol chemical sensor,” Sensors (Switzerland), vol. 16, no. 11, pp. 1802, 2016.

[62] S. R. M. Shah et al., “Penetration Depth Evaluation of Split Ring Resonator sensor using In-Vivo Microwave Reflectivity and Ultrasound Measurements,” in 12th European Conference on Antennas and Propagation (EuCAP 2018), 2018.

[63] N. Couniot, D. Bol, O. Poncelet, L. A. Francis, and D. Flandre, “A Capacitance-to-Frequency Converter With On-Chip Passivated Microelectrodes for Bacteria Detection in Saline Buffers Up to 575 MHz,” IEEE Trans. Circuits Syst. II Express Briefs, vol. 62, no. 2, pp. 159–163, Feb. 2015.

[64] I. Butterworth, J. Serallés, C. S. Mendoza, L. Giancardo, and L. Daniel, “A wearable physiological hydration monitoring wristband through multi-path non-contact dielectric spectroscopy in the microwave range,” IEEE MTT-S Int. Microw. Symp. Dig., Sep. pp. 60—61, 2015.

[65] J. Wang, Z. Zilic, and Y. Shu, “Evaluation of an RF wearable device for non-invasive real-time hydration monitoring,” in Proc. IEEE 14th Int. Conf. Wearable Implant. Body Sensor Netw. (BSN), pp. 91—94, May 2017.

[66] M. A. Ionescu and H. Guerin, “System for the remote monitoring of the hydration status of a living being,” U.S. Patent 2016/0058364 A1, Mar. 3, 2016.

[67] A. R. Brendtke, M. Wiehl, F. Groeber, T. Schwarz, H. Walles, and J. Hansmann, “Feasibility study on a microwave-based sensor for mea- suring hydration level using human skin models,” PLoS ONE, vol. 11, no. 4, Apr. 2016, Art. no. e0153145.

78

[68] B. Senevirathna, S. Lu, E. Smela, and P. Abshire, “An imaging platform for real-Time in vitro microscopic imaging for lab-on-cmos systems,” in BioCAS 2019 - Biomedical Circuits and Systems Conference, Proceedings, pp. 1–4, 2019.

[69] K. Bunnfors et al., “Nanoparticle activated neutrophils-on-a-chip: A label-free capacitive sensor to monitor cells at work,” Sensors Actuators B Chem., vol. 313, p. 128020, Jun. 2020.

[70] P. Bembnowicz and L. J. Golonka, “Integration of Transparent Glass Window with LTCC Technology for μTAS Application,” J. Eur. Ceram. Soc. vol. 30, p. 743–749, Feb. 2010.

79

Original publications

I Halonen, N., Kilpijärvi, J., Sobocinski, M., Datta-Chaudhuri, T., Hassinen, A., Prakash, S. B., Möller, P., Abshire, P., Kellokumpu, S., & Lloyd Spetz, A. (2016). Low temperature co-fired ceramic packaging of CMOS capacitive sensor chip towards cell viability monitoring. Beilstein Journal of Nanotechnology, 7, 1871–1877. https://doi.org/10.3762/bjnano.7.179

II Kilpijärvi, J., Halonen, N., Sobocinski, M., Hassinen, A., Senevirathna, B., Uvdal, K., Abshire, P., Smela, E., Kellokumpu, S., Juuti, J., & Lloyd Spetz, A. (2018). LTCC packaged ring oscillator based sensor for evaluation of cell proliferation. Sensors, 18(10), 3346. https://doi.org/10.3390/s18103346

III Kilpijärvi, J., Halonen, N., Juuti, J., & Hannu, J. (2019). Microfluidic microwave sensor for detecting saline in biological range. Sensors, 19(4), 819. https://doi.org/10.3390/s19040819

IV Kilpijärvi, J., Tolvanen, J., Juuti, J., Halonen, N., & Hannu J. (2020). A non-invasive method for hydration status measurement with a microwave sensor using skin phantoms. IEEE Sensors Journal, 20(2), 1095–1104. https://doi.org/10.1109/JSEN.2019.2945817

Reprinted under Creative Commons CC BY 4.0 license1(Papers I–IV © 2016, 2018,

2019 Authors).

Original publications are not included in the electronic version of the dissertation.

1 https://creativecommons.org/licenses/by/4.0/

80

A C T A U N I V E R S I T A T I S O U L U E N S I S

Book orders:Virtual book store

http://verkkokauppa.juvenesprint.fi

S E R I E S C T E C H N I C A

781. Asgharimoghaddam, Hossein (2021) Resource management in large-scale wirelessnetworks via random matrix methods

782. Kekkonen, Jere (2021) Performance of a 16×256 time-resolved CMOS single-photon avalanche diode line sensor in Raman spectroscopy applications

783. Borovkova, Mariia (2021) Polarization and terahertz imaging for functionalcharacterization of biological tissues

784. de Castro Tomé, Mauricio (2021) Advanced electricity metering based on event-driven approaches

785. Järvenpää, Esko (2021) Yläpuolisen siltakaaren optimaalinen muoto

786. Lindholm, Maria (2021) Insights into undesired load factors at work now andtomorrow : findings from different professions and working conditions

787. Miettinen, Jyrki & Visuri, Ville-Valtteri & Fabritius, Timo (2021) Carbon-containingthermodynamic descriptions of the Fe–Cr–Cu–Mo–Ni–C system for modelingthe solidification of steels

788. Tavakolian, Mohammad (2021) Efficient spatiotemporal representation learningfor pain intensity estimation from facial expressions

789. Longi, Henna (2021) Regional innovation systems and company engagement in theArctic context

790. Moltafet, Mohammad (2021) Information freshness in wireless networks

791. Nevanperä, Tuomas (2021) Catalytic oxidation of harmful chlorine- and sulphur-containing VOC emissions : a study of supported Au, Pt and Cu catalysts

792. Baharmast, Aram (2021) Wide dynamic range CMOS receiver techniques for apulsed Time-of-Flight laser rangefinder

793. Haiko, Oskari (2021) Effect of microstructural characteristics and mechanicalproperties on the impact-abrasive and abrasive wear resistance of ultra-highstrength steels

794. Oppenlaender, Jonas (2021) Crowdsourcing creative work

795. Kimbi Yaah, Velma Beri (2021) Development of novel biomass-based materialsfor the treatment of emerging contaminants in water

796. Liu, Chen-Feng (2021) Resource allocation for ultra-reliable and low-latency 5Gcommunication

C797etukansi.fm Page 2 Tuesday, August 17, 2021 1:51 PM

UNIVERSITY OF OULU P .O. Box 8000 F I -90014 UNIVERSITY OF OULU FINLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

University Lecturer Tuomo Glumoff

University Lecturer Santeri Palviainen

Postdoctoral researcher Jani Peräntie

University Lecturer Anne Tuomisto

University Lecturer Veli-Matti Ulvinen

Planning Director Pertti Tikkanen

Professor Jari Juga

Associate Professor (tenure) Anu Soikkeli

University Lecturer Santeri Palviainen

Publications Editor Kirsti Nurkkala

ISBN 978-952-62-3033-7 (Paperback)ISBN 978-952-62-3034-4 (PDF)ISSN 0355-3213 (Print)ISSN 1796-2226 (Online)

U N I V E R S I TAT I S O U L U E N S I SACTAC

TECHNICA

U N I V E R S I TAT I S O U L U E N S I SACTAC

TECHNICA

OULU 2021

C 797

Joni Kilpijärvi

RF-MICROWAVE SENSOR DEVELOPMENT FOR CELL AND HUMAN IN VITRO AND EX VIVO MONITORING

UNIVERSITY OF OULU GRADUATE SCHOOL;UNIVERSITY OF OULU,FACULTY OF INFORMATION TECHNOLOGY AND ELECTRICAL ENGINEERING;LINKÖPING UNIVERSITY,DEPARTMENT OF PHYSICS, CHEMISTRY AND BIOLOGY (IFM);INFOTECH

C 797

AC

TAJoni K

ilpijärvi

C797etukansi.fm Page 1 Tuesday, August 17, 2021 1:51 PM