INTEGRATION OF ENZYME-BASED BIOSENSORS AND CMOS
ELECTROCHEMICAL INSTRUMENTATION THROUGH A LAB-ON-CMOS
PLATFORM
By
Yue Huang

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILSOPHY

Electrical Engineering
2011

ABSTRACT
INTEGRATION OF ENZYME-BASED BIOSENSORS AND CMOS
ELECTROCHEMICAL INSTRUMENTATION THROUGH A LAB-ON-CMOS
PLATFORM
By
Yue Huang
Redox enzyme based electrochemical biosensors provide label-free continuous
monitoring of biomolecules. This thesis work aims to solve the key challenges in
constructing a microsystem that integrates enzyme-based electrochemical sensors,
electrode arrays, CMOS instrumentation circuits, and microfluidics. A CMOS compatible
enzyme immobilization technique based on conductive polymers is introduced and
demonstrated through a biosensor based on an alcohol dehydrogenase enzyme. Utilizing
a thorough study of the cross-disciplinary compatibility requirements for on-CMOS
electrochemical sensors, a microfabricated electrode array scheme is identified and
further optimized through a concentric ring working electrode design that minimizes
electrode area and processing complexity. A new CMOS bipotentiostat architecture is
introduced which, when used in conjunction with the concentric ring electrodes,
implements an electrochemical interrogation scheme that enables signal amplification
through the redox recycling with enzyme modified electrodes. Finally, a novel lab-onCMOS integration platform is introduced that unites the capabilities of lab-on-chip
microfluidic systems with the performance advantages of CMOS microsystems to
integrated bios. This work establishes a miniaturized platform for integrating a variety of
enzymes as biosensing elements that can be utilized to analyze biological samples using
powerful electrochemical techniques. By integrating significant developments in enzyme

immobilization, CMOS compatible microelectrode arrays, and CMOS instrumentation
for redox recycling, this research advances the fields of point-of-care medical
diagnostics, high-throughput screening, and a wide range of additional sensing
applications.

ACKNOWLEDGEMENT

I would like to thank my advisors, Prof. Andrew Mason, for his invaluable
mentorship, encouragement and inspiration. I feel so grateful and lucky to have the
opportunities of free research, and the experience of grant applications and teaching
under his supervision. Prof. Mason’s support is not only as a mentor, but also as a good
friend in life.
I would like to thank my committee members, Prof. Mark Worden, Prof.
Shantanu Chakrabartty and Prof. Tim Hogan for their valuable advice and unconditional
support during my study. I would also like to thank Prof. Xiangqun Zeng, Prof. Wen Li
and Prof. Clare Vieille. Your help introduced me into a fascinating world of electronics,
microfabrication, analytical chemistry and biochemistry that ultimately linked together.
I would like to thank my colleagues, Jichun Zhang, Junwei Zhou, Chao Yang,
Daniel Rairigh, Xiaowen Liu, Lin Li, and many more whose name are not listed here,
whom I developed friendships with that I will never forget. It remains one of my most
valuable possessions that will prevail.
I would not thank my parents and family members, because words are not enough
to express my appreciation and gratitude for their support, tolerance, encouragement and
sacrifice.

iv

TABLE OF CONTENTS
LIST OF TABLES ............................................................................................................ vii
LIST OF FIGURES ......................................................................................................... viii
NOMENCLATURE .......................................................................................................... xi
ABBREVIATIONS .......................................................................................................... xii
1

Introduction ................................................................................................................. 1
1.1 Overview of Integrating Electrochemical Enzyme Biosensor on CMOS
Potentiostat...................................................................................................................... 1
1.2 Enzyme-based Electrochemical Biosensors ......................................................... 4
1.3 Microfabricated Planar Microelectrode ............................................................... 8
1.4 CMOS Electrochemical Instrumentation ............................................................. 9
1.5 Challenges in Integration of Electrochemical Enzyme Biosensors and CMOS
Potentiostats .................................................................................................................. 10
1.5.1
Overview ..................................................................................................... 10
1.5.2
Enzyme Immobilization on CMOS ............................................................ 11
1.5.3
Microelectrode and Post-CMOS Microfabrication ..................................... 12
1.5.4
CMOS Potentiostat for Enzyme-based Biosensor Array ............................ 13
1.5.5
Lab-on-CMOS Packaging ........................................................................... 14
1.6 Research Goal .................................................................................................... 15

2

Electrochemical Enzymatic Bioelectronic Interface ................................................. 16
2.1 Overview ............................................................................................................ 16
2.2 Background ........................................................................................................ 16
2.3 Immobilization of Enzyme by Molecular Self-assembly................................... 18
2.3.1
Background ................................................................................................. 18
2.3.2
Experiment .................................................................................................. 19
2.3.3
Results and Discussion ............................................................................... 19
2.4 Immobilization of Enzyme by Conductive Polymer .......................................... 21
2.4.1
Background ................................................................................................. 21
2.4.2
Design ......................................................................................................... 23
2.4.3
Experiment .................................................................................................. 25
2.4.4
Results and Discussion ............................................................................... 28
2.5 Conclusion.......................................................................................................... 29

3

Planar Electrode System and Microfabrication ........................................................ 31
3.1 Overview ............................................................................................................ 31
3.2 Background ........................................................................................................ 31
3.3 Analysis of Design Requirements ...................................................................... 33
3.3.1
Electrochemical Electrode System ............................................................. 33
3.3.2
Geometry of Planar Microelectrode............................................................ 34
3.3.3
Electrode Materials ..................................................................................... 36
3.4 Planar Microelectrode Array .............................................................................. 37
v

3.4.1
Design ......................................................................................................... 37
3.4.2
Fabrication .................................................................................................. 38
3.4.3
Results and Discussion ............................................................................... 40
3.5 Reference Electrode ........................................................................................... 43
3.5.1
Introduction ................................................................................................. 43
3.5.2
Design ......................................................................................................... 45
3.5.3
Fabrication .................................................................................................. 47
3.5.4
Results and Discussion ............................................................................... 48
3.6 Conclusion.......................................................................................................... 50
4

CMOS Bipotentiostat ................................................................................................ 52
4.1 Overview ............................................................................................................ 52
4.2 Background ........................................................................................................ 52
4.3 Potentiostat Architecture .................................................................................... 53
4.3.1
Electrochemical Cell and Potentiostat ........................................................ 53
4.3.2
Bipotentiostat Architecture ......................................................................... 59
4.4 Redox Recycling on Concentric Ring Electrode................................................ 62
4.4.1
Background ................................................................................................. 62
4.4.2
Fabrication of Concentric Ring Electrode .................................................. 65
4.4.3
Results and Discussion ............................................................................... 66
4.5 Result and Discussion ........................................................................................ 68
4.6 Conclusion.......................................................................................................... 74

5

Lab-on-CMOS System Integration ........................................................................... 75
5.1 Overview ............................................................................................................ 75
5.2 Packaging a CMOS Die for Microfluidics ......................................................... 76
5.3 Lab-on-CMOS Integration Procedure ................................................................ 78
5.3.1
CMOS Die Preparation ............................................................................... 79
5.3.2
Silicon Carrier Micromachining ................................................................. 81
5.3.3
Surface Leveling and Fixing ....................................................................... 82
5.3.4
Planar Wiring .............................................................................................. 83
5.3.5
Passivation and Planarization ..................................................................... 85
5.3.6
Microfluidic Structure Integration .............................................................. 85
5.4 Result and Discussion ........................................................................................ 86
5.5 Conclusion.......................................................................................................... 89

6

Summary, Contributions and Future Work ............................................................... 90
6.1 Summary ............................................................................................................ 90
6.2 List of Contributions .......................................................................................... 91
6.3 Future Work ....................................................................................................... 93

APPENDIX ....................................................................................................................... 95
Modeling of Redox Recycling on Concentric Ring Electrode ..................................... 96
REFERENCES ............................................................................................................... 102

vi

LIST OF TABLES
Table 1.1: Comparison of bio-recognition elements........................................................... 5
Table 1.2: Substrates that can be detected by electrochemical enzyme biosensors............ 5
Table 2.1 Comparison of sensitivity, CMOS compatibity and process time of MSA and
CP-MSA methods applied to form an alcohol sensor. ...................................................... 30
Table 5.1: Resistance from terminal to terminal (units: Ω). ............................................. 87

vii

LIST OF FIGURES
Figure 1.1: The conceptual lab-on-CMOS device consists of an electrochemical
biosensor, an electrode array, a microfluidic sample handling device and a CMOS
integrated electrochemical instrumentation circuit. (For interpretation of the references to
color in this and all other figures, the reader is referred to the electronic version of this
dissertation.) ........................................................................................................................ 4
Figure 1.2: Diagrams to show the “lock-and-key” hypothesis of enzyme action. 1.
Substrates entering enzyme active sites; 2. Substrates-enzyme complex; 3. Substratesproducts complex; 4. Products leaving complex. ............................................................... 6
Figure 1.3: Immobilized enzyme along with cofactor and mediator. Recycling of cofactor
and mediator maintains continuous enzyme activity. ......................................................... 8
Figure 2.1: Enzyme immobilized by molecular self-assembly on gold electrode by using
a hetero-functional linker molecule. ................................................................................. 19
Figure 2.2: CV plot and the calibration curve of various concentration of ethanol on
sADH I86A mutant enzyme immobilized electrode. ........................................................ 20
Figure 2.3: Molecular structure of aminophenylboronic acid and its polymerized form. 24
Figure 2.4: Enzyme immobilized onto conductive polymer through self-assembled
cofactor. ............................................................................................................................ 24
Figure 2.5: A microfabricated 8-electrode array for enzyme immobilization experiments.
........................................................................................................................................... 26
Figure 2.6: Raman spectrum of PPy and PPy-pAPBA formed on roughened gold
electrode. ........................................................................................................................... 27
Figure 2.7: Cyclicvoltamogram (top) and calibration curve (bottom) of PPy-pAPBANAD-sADH I86A modified electrode tested in isopropyl alcohol concentration of
ranging from 0 to 1250 mM. ............................................................................................. 29
Figure 3.1: Electrochemical electrodes with 2×2 WE array and embedded Ag/AgCl RE.
........................................................................................................................................... 38
Figure 3.2: Fabrication sequence for integrated electrochemical electrode arrays with
embedded Ag/AgCl reference electrodes. ........................................................................ 40
Figure 3.3: Potential difference between microfabricated planar on-chip Ag/AgCl
reference electrode and commercial Ag/AgCl electrode (N=8). ...................................... 42
Figure 3.4: CV plots of 4 working electrodes in 10mM K3Fe(CN)6, 3M NaCl, scanning
rate 100 mV/S using on-chip Ag/AgCl reference electrode. ............................................ 43
viii

Figure 3.5: The molecular structure of Nafion. ................................................................ 46
Figure 3.6: Microfabricated Ag/AgCl/PPy(Cl-)/Nafion quasi-reference electrode
structure............................................................................................................................. 46
Figure 3.7: The photograph of the microfabricated Ag/AgCl/PPy(Cl-)/Nafion quasireference electrode. ........................................................................................................... 48
Figure 3.8: Potential difference of microfabricated Ag/AgCl/PPy(Cl-)/Nafion quasireference electrode vs. commercial Ag/AgCl reference electrode in various concentration
of KCl. In the plot, trace a and b are the Ag/AgCl/PPy(Cl-)/Nafion quasi-reference
electrode; c is a bare Ag/AgCl electrode. ......................................................................... 49
Figure 3.9: Potential difference of microfabricated Ag/AgCl/PPy(Cl-)/Nafion quasireference electrode vs. commercial Ag/AgCl reference electrode in standard pH buffers.
In the plot, trace a and b are the Ag/AgCl/PPy(Cl-)/Nafion quasi-reference electrodes; c is
a bare Ag/AgCl electrode.................................................................................................. 50
Figure 4.1: Electrochemical cell and its equivalent impedance network. (a) A
electrochemical cell symbol. (b) The equivalent impedance network. (c) The impedance
network with the built-in potential and impedance of RE, (d) A simplified impedance
network with the built-in potential and of RE. ................................................................. 54
Figure 4.2: A simplified potentiostat control over an electrochemical cell. ..................... 55
Figure 4.3: Potentiostats with the electrochemical cells being replaced by the equivalent
circuit models. (a) Scheme A: The excitation signal Ex feeds through an opamp. (b)
Scheme B: The exiciation signal Ex feeds through the working electrode. ...................... 56
Figure 4.4: The equivalent circuit models of the two schemes of potential control and
excitation signal insertion. . (a) Scheme A: The excitation signal Ex feeds through an
opamp. (b) Scheme B: The excitation signal Ex feeds through the working electrode. ... 57
Figure 4.5: The equivalent circuit model of expanded Scheme A (top) and Scheme B
(bottom) to multiple channels; .......................................................................................... 59
Figure 4.6: A bi-potentiostat based on traditional adder potentiostat architecture. .......... 60
Figure 4.7: (a) A bipotentiostat by expanding Scheme A, the traditional adderpotentiostat architecture. (b) A bi-potentiostat for electrode array based on Scheme B
with two parallel channels. ............................................................................................... 61
Figure 4.8: An exampleconcentric ring electrode fabricated on a bipotentiostat chip by
CMOS 0.5 µm technology. The diameter of the outer ring is 90 µm. .............................. 65
Figure 4.9: A concentric ring electrode fabricated on glass substrate, including two
neighboring working electrodes, WE1 and WE2, a surrounding ring counter electrode
and CE a disk reference electrode RE............................................................................... 66
ix

Figure 4.10: CVs of bmiBF4 exposed in air and in ENB saturated air. ........................... 67
Figure 4.11: DPV plots of bmiBF4 exposed in air and in ENB saturated air. .................. 68
Figure 4.12: A CMOS bipotentiostat with concentric ring electrode and the circuit
architecture. The chip size is 1.56×1.69mm..................................................................... 70
Figure 4.13: Readout opamp I-V conversion output at 100 KHz, 50 KHz and 10 kHz
clock frequencies. ............................................................................................................. 71
Figure 4.14: D-Fructose sensor response on MDH modified IDA. (a) Cyclic voltammetry
of 200 mM D-furctose with (solid) and without (dotted) redox recycling (top). (b)
Chronoamperometery of 400 mM, 200 mM and 50 mM D-fructose with redox recycling
(bottom)............................................................................................................................. 72
Figure 4.15: Fructose concentration vs. current measured by a commercial potentiostat
and the CMOS bipotentiostat. Diamond marked data is from CHI 760 bipotentiostat.
Triangle marked data are from the CMOS bipotentiostat. ................................................ 73
Figure 5.1: Lab-on-CMOS integration process flow. ....................................................... 79
Figure 5.2: A cross-section view of a CMOS pad after electroless plating of zinc, nickel
and gold. ............................................................................................................................ 80
Figure 5.3: Spin-on Apiezon wax thickness as a function of spin speed and wax diluted in
xylenes at 1:8, 1:4 and 1:2. ............................................................................................... 84
Figure 5.4: Surface profile from the carrier to the CMOS chip, measured before
metallization, passivation and planarization. Bottom picture is a SEM image of
interconnects over the bridge after metalization, to scale profilometer plot, confirming the
continuity of metal traces routing the chip pads to the carrier perimeter over the gap via
the bridge. Note: the horizontal axis scale is 10x the vertical scale. The zero position on
the vertical axis is the elevation of the carrier. ................................................................. 86
Figure 5.5: Current load test of planar wires. ................................................................... 88
Figure 5.6: Surface profile from the carrier to the CMOS chip, measured before
metallization, passivation and planarization. Inset is a SEM image of interconnects over
the bridge after metalization, to scale profilometer plot, confirming the continuity of
metal traces routing the chip pads to the carrier perimeter over the gap via the bridge.
Note: the horizontal axis scale is 10x the vertical scale. The zero position on the vertical
axis is the elevation of the carrier. .................................................................................... 89

x

NOMENCLATURE
Cdl

Double layer capacitance

D

Diffusion coefficient

F

Faraday constant

n

Number of electrons transferred

J

Current density

kp

Partition coefficient

kcat

Catalytic turnover number

Km

Michaelis-menten constant

me

Rate of mass transfer in the bulk

Pm

Permeability inside the enzyme layer

Rct

Charge transfer resistance

Rs

Resistance of solution

Zw

Warburg-diffusion impedance

xi

ABBREVIATIONS
AFM

Atomic force microscopy

BLM

Bilayer lipid membrane

CE

Counter electrode

CMOS Complementary metal–oxide–semiconductor
CP

Conductive polymer

CV

Cyclic voltammetry

CVD

Chemical vapor deposition

CYS

Cystein

DTT

Dithiothreitol

EIS

Electrochemical impedance spectroscopy

EDS

Energy dispersive spectroscopy

FAD

Flavin adenine dinucleotide

HPLC

High performance (pressure) liquid chromatography

IC

Integrated circuit

IDA

Interdigited array electrode

IR

Current resistor product

MOSIS An Multi-project wafer (MPW) of integrated circuit fabrication service provider
MSA

Molecular self-assembly

NAD

β-nicotinamide adenine dinucleotide (NAD )

+

xii

NADP

+

β-nicotinamide adenine dinucleotide phosphate (NADP )

pAPBA Poly-aminophenylboronic acid
PBS

Phosphate buffer solution

PDMS

Poly(dimethylsiloxane)

PMMA Polymethyl-methacrylate
PPy

Poly-pyrrole

PVD

Physical vapor deposition

QCM

Quartz crystal microbalance

RE

Reference electrode

SAM

Self-assembled monolayer

SEM

Scanning electron microscopy

sADH

Secondary alcohol dehydrogenase
th

sADH I86A Secondary alcohol dehydrogenase with isoleucine at 86 codon replaced by
alanine
tBLM

Tethered bilayer lipid membrane

xiii

1
1.1

Introduction

Overview of Integrating Electrochemical Enzyme Biosensor on CMOS
Potentiostat
The ability to identify and measure analyte concentrations is indispensable in

chemical, biological and medical application areas, such as environmental monitoring,
bio-product manufacturing, biohazard detection, clinical diagnosis, and drug discovery
[1-4]. By utilizing antibody, antigen, aptamer, DNA and peptide etc., biosensors offer
great diversity, sensitivity and specificity in their ability to hybridize, bind, transfer and
interact with target biological analytes [5]. Biosensors based on soluble proteins and
membrane proteins carry the same merits. Furthermore, some of proteins can provide
continuous and label-free measurements which are critical but most of other sensor
technologies cannot. Enzyme-based biosensors represent a sub-class having been applied
in monitoring glucose, lactate, etc [4, 6, 7]. For example, glucose sensors, the first and the
most successful biosensors, are typically constructed of electrochemical electrodes
conditioned with glucose enzymes [6]. The industry has been seeking miniaturized
enzyme-based systems that allow continuous monitoring of bio-molecule. Realization of
such systems inevitably requires synergistic integration of enzyme, transducer,
instrumentation and packaging.
Reagentless continuous operation of an enzyme-based electrochemical biosensor
suggests enzymes, cofactors and mediators to be immobilized on an electrode [8].
Immobilization prevents enzymes from being consumed or washed away and maintains
the interface. By immobilization an electron transfer pathway between the enzyme and
1

the electrode is built to supply the enzyme with electrons or protons for it to function
continuously. In order to be immobilized, enzyme could be physically entrapped in
polymer matrices or chemically bounded onto an electrode surface. By coupling enzymes
to an electrode, molecular interactions between the enzymes and the target analyte can be
directly converted into electrical signals that reflects the analyte concentration [9, 10].
The transducer of an enzyme-based electrochemical biosensor, the electrode,
converts the collective response of enzymes into electrical signals. To be more specific,
the enzyme-modified electrode is used as the working electrode of an electrochemical
cell which often consists of a working electrode, a counter electrode, a reference
electrode and electrolyte. An electrochemical transducer requires no labeling molecules
and is relatively simple and easy to miniaturize than an optical transducer consisting of
light source, lens, filter and light sensor [11, 12]. In addition miniaturization of
electrochemical

transducer

can

create

microelectrodes

that

are

favored

in

electrochemistry for its fast response, higher signal to noise ratio and other advantages
[13].
The electrochemical instrumentation circuits can be implemented by the popular
complementary metal–oxide–semiconductor (CMOS) microelectronics technology. As
proper electrochemical technique is determined, CMOS application specific integrated
circuits (ASIC) can be customized and optimized to interrogate an enzyme-based
biosensor. A commonly used potentiostat is the instrument for performing amperometry
to control the potential over a working electrode and records the current flowing through
it. Microfabrication technology for making electrodes including thin film deposition and
photolithography is a subset of the tools of the semiconductor industry for fabricating
2

CMOS integrated circuits (IC) [14]. A foundry is capable of producing the CMOS
electrochemical instrumentation ICs and electrodes at low cost in mass production. Nonstandard processes can be performed post CMOS for electrodes, enzyme immobilization
and packaging.
Many fluidic handling lab-on-chip (LOC) devices have been realized with
CMOS-compatible processes [15]. However, a distinct lack of integration of CMOS ICs
and LOC remains as a major obstacle toward miniaturization of the entire instrumentation
system, which is appealing for many applications including high throughput drug
screening, medical point-of-care, and biological research. Bring together microsystems
and LOCs onto a CMOS IC to be a lab-on-CMOS combines the power of all,
significantly reduces footprint of the system, reduces the consumption and expenses of
analyte and consumables and enhances the detection performance. A microelectrode
array with incorporating microfluidic components can provide parallelism to improve
throughput or redundancy to improve reliability in case of failures for fast simultaneous
high-throughput measurements. Very few microsystems with electrochemical sensors and
corresponding circuits have been realized [16-21]. Figure 1.1 illustrates a conceptual
electrochemical biosensor array lab-on-CMOS based on a CMOS IC, with fluid delivery
to the electrode array on top. To the best of our knowledge, an enzyme-based lab-onCMOS has not been reported.

3

Figure 1.1: The conceptual lab-on-CMOS device consists of an electrochemical
biosensor, an electrode array, a microfluidic sample handling device and a CMOS
integrated electrochemical instrumentation circuit. (For interpretation of the references
to color in this and all other figures, the reader is referred to the electronic version of
this dissertation.)
1.2

Enzyme-based Electrochemical Biosensors
Electrochemical biosensors are critical reviewed by Wang [3], Grieshaber [22],

Wang [23] and Ronkainen [24]. Among them, enzyme-based electrochemical biosensors
are more suitable for implementing on a CMOS chip than affinity based biosensors
because it functions continuously that merit the cost of CMOS. Enzymes are excellent
molecular recognition elements for biosensors, offering great diversity, sensitivity and
specificity. Enzymes are known to catalyze about 4,000 biochemical reactions. Compared
to DNA, aptamer and antibody-antigen that rely on binding that are normally one-shot,
some enzymes, Oxidoreductase, provide label-free, continuous detection when used as
bio-recognition element. Comparison of bio-recognition elements with respect to their
recognition mode, transducer types, labeling requirements and operation mode is shown
in Table 1.1. An example glucose, lactate, and pyruvate enzyme biosensor array on a
microfabricated gold electrode exhibited an opportunity of applying to intensive blood
4

glucose monitoring [4]. Other bio-molecules such as amino acid, steroid, alcohols can
also be detected by enzyme biosensors. Table 1.2 lists compounds that can be detected by
amperometric enzyme biosensors.

Bio-recognition
element

Recognition
mode

Transducer

Label
Free

Continuous

Hydrogen bonds

Optical
Electrochemical

N

N

Aptamer

Affinity

Optical

N

N

Antibody-Antigen

Affinity

Optical,
Electrochemical

N

N

Reductionoxidization

Optical
Electrochemical

Y

Y

DNA probe

Enzyme

Table 1.1: Comparison of bio-recognition elements.

Class
Carbohydrates
Amino acids
Carboxylic acids
Alcohols and
Phenols
Amines and
Heterocycles
Aldehyde
Inorganic ions
Oxide and Peroxide
Quinone
Steroids
Azide
Hemoprotein

Compounds
Glucose, Lactose, Maltose, Fructose, Xylose, Galactose,
Sucrose, Cellobiose, Mannose, Arabinose
Aspartate, Sarcosine, N-Benzoyl-L-tyrosine, Glutamate, LPhenylalanine
Ascorbate, Pyruvate, Lactate, Malate, Oxalate, Glycolate,
Tartarate, Fumarate
Ethanol, Methanol, Cholesterol esters, Cholesterol,
Bilirubin, Phenol, p-Cresol, Catechol, o-Aminophenol, oCresol, Dopamine
Acetylcholine, Uric acid, Hypoxanthine, Choline, Xanthine,
Histamine, Putrescine
Formaldehyde
Nitrate, Sulfite, Oxygen, Potassium ion
Nitric oxide, Hydrogen peroxide
Vitamin K
Testosterone, Estradiol, Bile acid
Sodium azide
Cytochrome c

Table 1.2: Substrates that can be detected by electrochemical enzyme biosensors.
5

To understand the mechanism of enzyme operation, a “lock-and
he mechanisms
and-key” hypothesis
model is described in Figure 1.2 The active site of an enzyme is shaped like a pocket
1.2.
with geometry suitable for containing a specific group of molecules with the chemical
group inside the pocket to attract, lock, convert and release the specific molecule.
According to their functionality, enzymes are categ
categorized into Oxidoreductase,
to
Transferases, Hydrolase Lyases, Isomerases, Ligases. Oxidoreductase, including
Hydrolases,
Oxidoreductase
oxidase, dehydrogenase and peroxidase associates with direct electron transfer, is suitable
dase,
for constructing electrochemical biosensor [25].
substrates

1

products

2

3

4

Figure 1.2: Diagrams to show the “lock
“lock-and-key” hypothesis of enzyme action. 1.
key”
action
Substrates entering enzyme active sites; 2. Substrates enzyme complex; 3. SubstratesSubstrates-enzyme
Substrates
products complex; 4. Products leaving complex.
wer
Enzymes lower the activation energy that a reaction requires. In free solution, a
group of molecules, called cofactor or co-enzymes, help recover the enzymatic activity.
ed cofactors
activity
There are three nature enzyme cofactors, NAD, NADP and FAD, each serving
FAD
corresponding enzymes. In an organism, cofactors carry electrons to or from enzymes to
recover the activity of enzymes, are recycled as a part of metabolism. In an enzyme-based
cover
biosensor the activation energy can be tuned to favor a reaction. Immediately, the
enzymatic activity must be recovered for continuous operation.
To construct self
self-contained (autonomous) label-free biosensors with enzymes,
s
enzymes and their cofactors are immobilized on an electrode surface. The activation
6

energy of individual enzyme and its catalytic reaction that involves electron transfer can
be controlled by an excitation potential. As a collective of electron transfer activity, the
resulting current is a function of the substrate concentration. For example, alcohol
dehydrogenase (ADH, EC 1.1.1.1) is a group of seven dehydrogenase enzymes that
occurs in many organisms and facilitates the interconversion between alcohols and
aldehydes or ketones. In the human body, those enzymes break down alcohols which
could otherwise be toxic. As a model for this research, secondary alcohol dehydrogenase
(sADH), an Oxidoreductase, once drained electrons, catalyzes the reaction for oxidizing
ethanol (۱۶૜ ۱۶૛ ‫ ,۶۽‬substrate) to produce acetaldehyde (۱۶૜ ۱‫ ,۶۽‬product) is:

࢙࡭ࡰࡴ

࡯ࡴ૜ ࡯ࡴ૛ ࡻࡴ ርۛۛሮ ࡯ࡴ૜ ࡯ࡻࡴ + ૛ࡴ
ࡺ࡭ࡰ + ࡴ ↔ ࡺ࡭ࡰࡴ

where nicotinamide adenine dinucleotide (NAD), a cofactor of ADH, is reduced as
ethanol is oxidized. When a biosensor with NAD immobilized is positively biased to the
point of oxidizing NADH, NAD is continuously recovered. Enzyme activity is also
maintained by recycling process. The recycling process of immobilized ADH enzyme,
along with cofactor and mediator, is illustrated in Figure 1.4. The mediator facilitates
electron transfer, thus lowering the overpotential to drive the reaction and protecting the
enzyme from being denatured. An alcohol sensor based on sADH can be realized by
measuring the current of transferred electrons.

7

Figure 1.3: Immobilized enzyme along with cofactor and mediator. Recycling of cofactor
and mediator maintains continuous enzyme activity.

1.3

Microfabricated Planar Microelectrode
As the transducer of an electrochemical biosensor, the electrode transforms

chemical processes at the electrode surface into electrical signals for equipment to
interpret. Such an electrode where interested reaction takes place is called a working
electrode. A working electrode gains its sensing specificity through surface modification.
Coupling an enzyme such as Oxidoreductase onto the electrode, molecular interactions
between the enzyme and the target molecules can be converted directly into electrical
signals, as explained in Section 1.2.
A typical electrode system consists of a working electrode, a counter electrode
and a reference electrode. The performance of an electrode system can be greatly affected
by its geometry [26]. A planar electrode system is adopted to integrate electrode system
on the surface of a CMOS potentiostat, microfabrication technology, the same technology
for fabricating CMOS chips, is used for making microelectrodes. The electrode
geometrical factors of a planar electrode system involve considerations of
microfabrication capability, electrochemical effects, enzyme activity, circuit layout, chip

8

size and surface profile, electrode materials, enzyme immobilization method, packaging
and microfluidics.
Reference electrodes play a crucial role in an electrochemical system. It
establishes a constant potential standard in measuring and controlling the working
electrode potential, without being interfered by electrolyte solution. To match the
geometrical requirements of planar microelectrode system, traditional reference electrode
is not suitable. Miniaturized reference electrodes can be constructed by microfabrication.
Microfabricated reference electrode in biosensing applications is reviewed by Sinwari,
et.al. [27].
1.4

CMOS Electrochemical Instrumentation
An electrochemical instrument is a control and measurement device that keeps the

potential of the working electrode at the desired level with respect to the reference
electrode. At the same time, it records the current flowing between the working electrode
and the counter electrode. Such an instrument is normally referred to as a potentiostat. A
modern bench-top potentiostat is built with general purpose discrete electronic devices,
including a microprocessor to communicate with computer-based software.
The electrochemical techniques that a potentiostat can perform include
potentiometry, coulometry, voltammetry and impedance spectroscopy. Potentiometry
measures the potential between two electrodes in a solution, which is related to the
concentration of the analyte. A selective indicator electrode and a stable reference
electrode are required for potentiometry. Coulometry uses applied current or potential to
completely convert an analyte. By integrating the passing current, the number of
9

electrons indicates the amount of analyte. Coulometry is good for a small known volume
of analyte or titration. Voltammetry measures the potential of an indicator electrode
versus a reference electrode. Amperometry applies a constant and/or varying potential at
an electrode's surface and measures the resulting current. This method can reveal the
oxidation and reduction potential of an analyte and its reactivity that relates to the analyte
concentration. Basics of electrochemical instrumentation and techniques are introduced in
reference [26] and [28].
Electrochemical techniques that require complex data analysis and interpretation
challenge the data processing capability, battery life and cost of a portable device. For a
lab-on-CMOS, simple potential control and sensor data output is highly desired. For a
particular electrochemical sensor, a suitable technique to extract the sensory response
may be found. Once a sensor is characterized, redundant potentiostat functions of a
general potentiostat can be removed to optimize the CMOS potentiostat.
1.5

Challenges in Integration of Electrochemical Enzyme Biosensors and CMOS
Potentiostats

1.5.1

Overview
Figure 1.1 illustrates an example lab-on-CMOS platform integrating an enzyme-

based biosensor onto a CMOS potentiostat. The device consists of electrochemical
biosensors, electrode array, microfluidic sample handling structure and a CMOS
integrated electrochemical instrumentation circuit. Some challenges are critical yet not
stressed in literature. These includes the CMOS-bioelectronics interface, the
microelectrode

array

and

the

post-CMOS
10

microfabrication

on

CMOS,

the

implementation of a CMOS electrochemical instrument and the packaging of an
integrated circuit die for CMOS-microfluidics integration.
1.5.2

Enzyme Immobilization on CMOS
Basic requirements for enzyme immobilization are not only to physically attach

enzyme to the electrode surface but also to create an electron transfer pathway and that
maintains enzymatic activity. Furthermore, enzyme coverage should be uniform across
the electrode, and the process should be suitable for batch fabrication. Two
immobilization methods, molecular self-assembly (MSA) and conductive polymer coimmobilization are evaluated in this thesis research.
Many MSA heavily depends on thiol-metal bonding to bind the initial molecule
layers onto an electrode. Noble metals are used because of the strong affinity of sulfur
head groups for these metals. With self-assembly method, a mono-layer of enzyme is
formed on metal electrode surface, providing uniform enzyme coverage from batch to
batch. One challenge for this method is to facilitate electron transfer between enzymes
and the electrode. Since the bonding molecules are not conductive, electron transfer is
sluggish unless the electrons can acquire extra energy to hop across molecule layers or
facilitated by an electron transfer mediator. The challenge in using MSA on a CMOS
electrode array is to avoid fouling of counter electrode and reference electrode by MSA
molecules, which reduces the CE and RE performance.
An alternative immobilization approach is to use electropolymerized conductive
polymers. A challenge is to control the enzyme coverage uniformity that affects time
response and sensitivity of the sensor. The conductive polymer and the embedded
11

enzyme could be selectively immobilized on the working electrode only. The tradeoff is
that electropolymerization requires applying electric potential on the working electrode
for the polymerization to happen; however it is not difficult if the working electrode is
tied to a potentiostat as it would be in the lab-on-CMOS platform. This thesis work
evaluates the MSA method and identifies an alternative method suitable for immobilizing
enzyme on CMOS.
1.5.3

Microelectrode and Post-CMOS Microfabrication
Before fabricating microelectrodes on a CMOS chip, the electrode geometrical

factors of planar electrodes must be determined. These geometrical factors include shape,
size, spacing, and location. The design limitations factors of a planar electrode system
involve microfabrication etc. as introduced in Section 1.3. Designing the microelectrode
is a system level decision making problem that requires all of the above factors be taken
into account.
Post-CMOS microfabrication of the electrodes on CMOS dice introduces
practical challenges in photolithography and interconnection between electrodes and
CMOS. Firstly, during photolithography, when photoresist is spin-coated on the chip
during photolithography, the non-uniform buildup at chip edges or conglomeration in the
chip center of photoresist greatly affects the quality of photolithography. Secondly, the
sidewall of the contact openings in the passivation dielectric is steep which can break
down the interconnection between the chip and the electrode. Thirdly, since the chip’s
final passivation dielectric layer is not planarized, the CMOS layout and the electrode
location needs to be carefully planned to maintain smooth electrode surface.

12

Reference electrode plays an important role in voltammetry and amperometry.
However, for a lab-on-CMOS platform, construction of a solid-state planar reference or
pseudo reference electrode is limited by CMOS or CMOS compatible processes. These
electrodes should be bio-compatible, not affected by bio-interface formation, and show
low drift in the test environment. Finally, the reference electrode fabrication formation
process must be scalable to fabricate on a CMOS die.
1.5.4

CMOS Potentiostat for Enzyme-based Biosensor Array
The first challenge is to determine a proper electrochemical technique that can

extract the enzymatic sensory response in direct relation with substrate concentration and
could easily be implemented to a CMOS potentiostat. For example, cyclic voltammetry
(CV) is a good electrochemical technique for analytical purposes that could complicate
the system. The species and concentration can be obtained from CV by the peak currents
and the potentials at the peaks. To implement CV on a self-contained CMOS IC, triangle
wave signal generator circuit should be designed for output the excitation signal, data
digitization and storage should be available for storing data for using a microprocessor to
extract the peak information. An open challenge is to identify more suitable
electrochemical techniques for enzyme biosensor that could be effectively implemented
in a simple CMOS.
The following challenge is to architect a potentiostat to support the determined
electrochemical technique. For example, to take the advantage of redox recycling by
applying amperometry on a pair of working electrodes could be manipulated by
bipotentiostat architecture. Further, the ideal potentiostat architecture should support

13

electrode array for simultaneously readout. An architecture that is expandable to an array
is appealing to replace the traditional adder potentiostat structure.
1.5.5

Lab-on-CMOS Packaging
The major challenge is to enable a CMOS chip to work in a liquid environment

and to allow a high density microfluidic structure to access the on-chip electrode array.
CMOS chip packaging process should free the chip from risk of ion contamination.
Microfluidic structures are relatively larger than CMOS chips thus demand surface
expansion for functional fluidic structures and interconnections. Microfluidic structures
need a flat surface to bond to and reside on. Surface leveling is required for fluidic and
electrical connections.
The packaging material for Lab-on-CMOS should be biocompatible, chemically
inert and electrically insulating. Biocompatible materials are identified by ISO 10993, the
international standard set for evaluating the biocompatibility of a medical device. The
device should be reliable and endurable for working in harsh chemical environment
during enzyme immobilization and testing, without adversely affecting the enzyme
activity and test environment. Inter-compatibility of materials used in processes and
bonding of interfacial materials are major fabrication challenges.
The packaging materials and process should also be determined by CMOS
compatibility. Process of higher than 400 ˚C may cause failure of aluminum traces. The
packaging material should not require processing above 400 ˚C. Certain metal ions
diffuse quickly in silicon. The bulk silicon should be properly protected from contacting

14

aqueous solution during processes to avoid detrimental effects on threshold voltage
chronically.
1.6

Research Goal
The goal of this research is to establish a Lab-on-CMOS methodology to highlight

the power of integrating microelectronics and biosensors by demonstrating an
incorporation of enzyme biosensors onto electrode arrays formed CMOS instrumentation
and combines with microfluidics. This goal will be achieved through analysis, judicious
selection and optimization of appropriate technologies. To achieve this goal, efforts in
enzyme immobilization, planar electrode system and fabrication, and CMOS potentiostat
instrumentation are reported in Chapter 2, 3 and 4, respectively. Chapter 5 introduces a
Lab-on-CMOS scheme that allows integration of CMOS with microfluidic structure.

15

2
2.1

Electrochemical Enzymatic Bioelectronic Interface

Overview
This research seeks solutions of integrating enzyme-based biosensors on CMOS.

Enzyme-based biosensors were introduced in Section 1.2. The challenges for building up
enzymatic bioelectronic interface on CMOS are discussed in Section 1.5.2. In this
chapter, two enzyme immobilization schemes are introduced. One scheme uses MSA to
bind mono layers of linking molecules, cofactors, electron mediators and enzymes onto to
a gold electrode. The other scheme we invented combines conductive polymer and
molecular self-assembly, referred to hereafter CP-MSA. The CP-MSA method uses
conductive polymers as initial layers, eliminating the dependence of gold, and then binds
the cofactors and enzymes by molecular self-assembly as the MSA scheme.
2.2

Background
For enzyme biosensors based on oxidase enzymes, the substrate concentration is

indirectly deduced by detecting the byproduct, hydrogen peroxide, produced during the
catalyzed oxidation reaction and diffused onto the detector electrode. One of the inherent
drawbacks is the oxygen concentration dependency. Its non-oxygen-dependant
substituent, dehydrogenases normally require specific cofactors. To improve the poor
electron transfer efficiency from cofactors to electrode surface, electron mediators are
added to the free solution as electron-carrying vehicles to improve mass transport. By
adding mediators, the destructive high over-potential applied to the protein interface is
mitigated. Instead of contained in solution, electron mediators have been co-immobilized
along with the enzymes and the corresponding cofactors [7]. By constructing a direct
16

electron transfer pathway from enzyme to electrode, no mediator in free solution is
involved, thus increasing the sensory current and minimizing the side-effects of
mediators. As an extension of this work, researchers are currently trying to simplify the
process by engineering mutant enzymes that embed the functions of both cofactors and
mediators.
Enzyme immobilization methodology for electrochemical sensors follows some
criteria. First, the active sites of enzyme should be accessible by substrate after
immobilization. Second, the enzyme has to maintain functionality without being
denatured. Third, a facilitated electron pathway is crucial for the sensitivity and stability
of the sensor.
There are many immobilization techniques available, such as adsorption,
encapsulation, entrapment, cross-linking, and covalent bonding. MSA method by
covalent bonding and conductive polymer entrapment method are suitable methods for
application on a CMOS microelectrode because they allow non-manual immobilization
on small electrode footprints and are individually addressable. The immobilization
methods using self-assembled molecules and polymer matrices were critically reviewed
[2,8][29].
The target enzyme, Oxidoreductase, including oxidase and dehydrogenase,
requires cofactors or electron mediators to restore its activity. In nature, there are only a
few types of cofactors for the above enzymes. It allows researchers to develop a sensor
platform by immobilizing those cofactors and thus corresponding enzymes. Numerous
biosensors can be built based on such platform by exchanging only the enzymes to detect
17

correspondent biomolecues. This platform is promising in building an array of biosensors
for detecting several bio-molecules in biological relevance.
The objective of research in this chapter is to determine suitable technologies for
immobilizing enzyme on post-CMOS microelectrodes. With the microsystem constraint
in mind, immobilization of dehydrogenase enzymes using MSA is evaluated and a new
immobilization scheme using conductive polymers and MSA to overcome fabrication
limitations of MSA method is introduced.
2.3
2.3.1

Immobilization of Enzyme by Molecular Self-assembly
Background
+

Enzymes that are dependent on cofactors (NAD, NAD(P) or FAD) have been
immobilized on gold electrode by several MSA schemes [9]. The collaborators in Prof.
Worden’s lab have developed a novel immobilization method using a hetero-functional
linker molecule [10]. The structure is shown in Figure 2.1. It uses cysteine, a branched
amino acid having sulfhydryl, amino, and carboxyl functional groups, as a linker. The
sulfhydryl group of cysteine provides thiol-Au bonding to anchor itself on gold electrode
surface. Cysteine also links to the electron mediator toluidine blue O (TBO) through the
+

carboxyl group, and to the cofactor (e.g. NAD(P) ) through the amino group. This
structure is comparable with previously reported approaches that bind the enzyme,
cofactor, and mediator in a linear chain.

18

Figure 2.1: Enzyme immobilized by molecular self-assembly on gold electrode by
using a hetero-functional linker molecule.

2.3.2

Experiment
A 2mm diameter gold disk electrode was fabricated using CMOS compatible

processes by physical vapor deposition of 5nm Ti/100nm Au on dielectric substrate
followed by photolithography and lifting off. The processes are discussed in details in
Chapter 3. An alcohol sensor was developed on the gold electrode by immobilizing sitemutated secondary alcohol dehydrogenase (sADH I86A) using the MSA method
described in Section 2.3.1. The modified electrode was connected to a CHI 660
Electrochemical Workstation as the working electrode. Cyclic voltammetry (CV) was
performed to examine the sensor’s response. The potential swept from -200 mV to 400
-1

mV at a scanning rate of 100 mV s . A series concentration of several representative aryl
and chiral alcohols in 10X Phosphate Buffer Solution (PBS) was tested.
2.3.3

Results and Discussion
The cyclic voltammogram in Figure 2.2 showed reduction peaks and the oxidation

peaks at 160mV (vs. Ag/AgCl reference electrode) and 200 mV respectively. The

19

magnitude of peaks was found to increase with ethanol concentration in the tested range
from 5mM to 25mM.

Figure 2.2: CV plot and the calibration curve of various concentration of ethanol on
sADH I86A mutant enzyme immobilized electrode.

CV experiments were also performed on the sensor in isopropanol, butanol and
acetophenone substrates. Table 2.1 compares the saturation current and the sensitivity of
both wild-type and mutant enzyme in those substrates respectively.
Experimental results show a functioning enzyme-based alcohol biosensor. The
experiment suggests MSA method can be utilized to create an electrochemical enzymatic

20

sensor interface on microfabricated gold electrode. This method relies on gold or other
noble metals for the initial molecular bindings that require post-CMOS fabrication
process on a foundry CMOS chip.

Substrate

Sensitivity (µAmM-1cm-2)

Ethanol

0.71

Isopropanol

1.16

Butanol

0.29

Acetophenone

1.89

Table 2.1: The sensitivity results from the cyclic voltammograms for the
+
CYS-TBO-NADP -sADH I86A functionalized gold electrode in the
presence of different concentrations of alcohols or ketones. Acetophenone is
25% w:v acetonitrile in PBS.

2.4
2.4.1

Immobilization of Enzyme by Conductive Polymer
Background
An alternative enzyme immobilization technique for CMOS electrode is using

conductive polymer (CP), that is compatible with silicon technology [30]. Since MSA
method requires gold electrode surface, to form such an interface onto a CMOS chip,
fabrication steps in addition to standard CMOS processes are required to deposit and
pattern gold working electrodes. Furthermore, to avoid the counter electrode being fouled
by MSA thiol linking molecules, additional masking processes or surface pretreatment of

21

the counter electrode is needed. Otherwise, the effective area of counter electrode would
be greatly reduced by fouling. In contrast, CP can be selectively synthesized on working
electrode without fouling other electrodes. Besides, CP can be formed on CMOS
aluminum [31]. If it is true that enzyme immobilization can be achieved on CMOS
aluminum, compared to the MSA method, CP method will have the advantage in
processing time, materials cost and electrode immobilization selectivity. The trade-off for
these advantages is the added complexity of applying an external potential on the target
electrode during polymerization.
Immobilization of biomolecules, including enzymes using CPs was reviewed by
Schuhmann [30], Wallace [32], Gerard [33], Ahuja [34], Cosnier [29], et. al. The
existing methods are co-entrapment of enzymes in conductive polymer matrices and
adsorption of amphiphilic monomers and biomolecules [29]. Co-entrapment method
produces a conductive polymer doped with multiple layers of enzymes. Such as a
multilayer sensor based on the sequential electrochemical polymerization of pyrrole or
pyrrole derivatives [35]. It generates a larger sensory current than the sensor with a
monolayer of enzyme using MSA method, partly because of larger enzyme coverage per
unit area. However, with the co-entrapment method, a greater portion of immobilized
enzymes are buried in the conductive polymer matrices, retarding the sensor response.
Response time becomes more critical as the scale of sensor array increases with fixed
instrumentation circuits. Another disadvantage of co-entrapment is the lack of control in
enzyme coverage uniformity, which is necessary to reduce sensor-to-sensor difference.
Immobilization with enzyme monolayer avoids those drawbacks. A biotinylated glucose
oxidase monolayer was immobilized through an avidin bridge [36]. Conductive polymer
22

polypyrrole was used for covalent binding to glucose oxidase to improve sensor response
time [37]. Direct electron transfer was shown possible between vapor phase polymerized
CP and GOx [38].
2.4.2

Design
+

+

In MSA method, phenylboronic acid is used to link FAD, NAD or NAD(P) by
the double hydroxyl groups of the boronic acid. The challenge is the linking molecules
are not conducting electrons, so electron mediator is needed. If a conductive media were
bind to electrodes and enzymes, electron mediators may be saved. A candidate is a
phenylboronic acid with amino group that can be electropolymerized, the precursor of
poly-aminophenylboronic acid, both shown in Figure 2.3. The double hydroxyl group is
not evolved in polymerization, suggesting that they are available to link enzyme
cofactors.
Instead of entrapping multiple layers of enzymes in conductive polymer, a
monolayer of enzymes can be immobilized through its corresponding cofactor as by the
MSA method. This hybrid CP-MSA method can overcome the limitations in their
applications to CMOS. This enzyme immobilization approach has not been reported by
other researchers.

23

Figure 2.3: Molecular structure of aminophenylboronic acid and its polymerized form.
:

However, the poly-aminophenylboronic acid polymer shows poor adhesion to metal
aminophenylboronic
electrode surface. To promote the adhesion, an initial layer of conductive polymer, polypyrrole (PPy), was applied [39]. CMOS top metal layer, normally aluminum, is an active
,
i
metal. By adding tartaric acid or saccharine, PPy was successfully formed on aluminum
without corrosion [40, 41]
41].
This new design features a PPy-pAPBA conductive polymer bi-layer, selfhis
assembled enzyme cofactor by the double hydroxyl groups on pAPBA surface and
roups
attached cofactor dependant enzymes, as shown in Figure 2.4.

Figure 2.4: Enzyme immobilized onto conductive polymer through self assembled
:
self-assemb
cofactor.

24

2.4.3

Experiment
To validate the CP-MSA enzyme immobilization idea, two conductive polymer

layers were electropolymerized on gold electrode surface, followed by bonding NAD
cofactor and sADH enzyme. Each formation step was characterized. The formed alcohol
sensor was tested using cyclic voltammetry.
Gold electrodes were prepared by photolithography and physical vapor deposition
on glass substrate, as shown in Figure 2.5. Before electropolymerization, to remove
absorbed organic contaminants the electrodes were cleaned using isopropyl and rinsed by
DI water. To remove the native gold oxide on the electrodes, they were soaked in 0.1M
HCl for 10 minutes, rinsed with DI water and dried with nitrogen. To electropolymerize
pyrrole, a 100mM pyrrole solution was prepared in 10X PBS pH 7.4. The solution was
homogenized and degassed before using. The electrode, along with an Ag/AgCl reference
electrode and a gold counter electrode was then connected to an Omni-101 potentiostat.
-1

Two cycles of potential sweep from 0mV to 600mV at 100mV s deposited a thin layer
of PPy on the electrode. The cyclic voltammogram showed a sharp increasing oxidation
current at over 400 mV, indicating the formation of polypyrrole. The electrode was rinsed
and transferred to 1 mM 3-aminophenyleboronic acid solution. To deposited a thin layer
of pAPBA on top of PPy, two cycles of potential sweep from 0 mV to 800 mV at 100 mV
-1

s were performed.

25

Figure 2.5: A microfabricated 8-electrode array for enzyme immobilization experiments.

Raman spectroscopy was used to examine the formation of conductive polymer
layers by the signature spectrum of their chemical bonds. To confirm the formation of
PPy, over-oxidized PPy on roughened gold electrode was examined before and after the
formation of pAPBA. Figure 2.6 shows both of the labeled spectrums. The peaks of the
PPy spectrum are the signature peaks of benzyl ring, amino group, etc., confirming the
formation of PPy. The peak at wave number 1590 represents C-C in the benzyl ring,
while the peak at wave number 1049 represents C=C. The ratio of the height of the two
indicates the degree of oxidation. Because PPy and pAPBA are sharing similar molecular
structures, the spectrum of pAPBA and PPy are alike, albeit the intensity at wave number
1590 and wave number 1049 differs. The higher ratio at wave number 1590 over wave
number 1049 is due to the higher maximum potential, 800 mV, applied for polymerizing
APBA. At this potential, the already polymerized PPy was further oxidized to a higher
oxidation state. The reflected PPy light from underneath the pAPBA adds to that of
pAPBA. The subtle differences are the small peaks between wave number from 600 to

26

800, which fall into the spectrum range of B-OH bonding that clearly differentiate PPy
and pAPBA. These results confirm the formation of a layer of a pAPBA on PPy.
This hybrid method takes advantage of both conductive polymer and selfassembly. Without relying on gold surface, it provides greater flexibility for choosing
electrode material including CMOS aluminum and cheap carbon ink printed electrodes.
For immobilizing enzyme on the CMOS chip electrode surface, gold deposition and
electrode patterning steps are saved. Compared to the self-assembly method, it spends

1590

less steps, processing time and chemicals.

500

932
975
1049

Intensity

400

300

PPy + pAPBA

1257
1337
1413

600

200

PPy

100
0
600

1000

1400
Wave number

1800

Figure 2.6: Raman spectrum of PPy and PPy-pAPBA formed on roughened gold
electrode.

27

After electropolymerization of pAPBA, the NAD cofactor and sADH were selfassembled following published protocol for the MSA methods [10]. Briefly, the PPypAPBA electrode w soaked in 1 mM NAD solution for one hour and rinsed thoroughly to
bind NAD. The PPy-pAPBA-NAD electrode was incubated in sADH I86A solution (6.8
mg/ml) for one hour and rinsed in running DI water to remove the loosely attached
enzymes.
2.4.4

Results and Discussion

To verify that the enzymes were immobilized on the electrode and remained active, the
PPy-pAPBA-NAD-sADH I86A modified electrode was tested using cyclic voltammetry
in substrate solutions for the sensory response. It was tested in isopropyl solutions of
various concentrations. Figure 2.7 shows the cyclic voltammogram. By recording the
-1

oxidation current at 300 mV, the sensitivity 1.30 nAmM cm

-2

was observed from the

calibration curve. Unlike the MSA method, there is no distinctive peak present in the
cyclic voltammogram. It matches Zayats’ experimental results. In Zayats’ model on
electron transfer, the exponential curves were explained as a result of electrons hopping
over a band gap. Thus it is possible that because of the good wiring provided by CP, no
electron hopping presented in the electron transport path, so there were no peaks in the
cyclic voltammogram. However, comparing with the MSA method, the current is
smaller, which may be caused by lack of immobilized enzyme being immobilized, or
because the substrate was continuously consumed by the enzyme without preconcentrating by electric force. Moreover, the maximum current is not limited by the
hopping of the electron over a driving potential. Without having to drive the electrons to
hop, the potential to drive the sensor can be reduced. Lowering working potential is
28

beneficial for reducing interference by other substances that are transformed at higher
potential and contribute to the background current.

Figure 2.7: Cyclicvoltamogram (top) and calibration curve (bottom) of PPy-pAPBANAD-sADH I86A modified electrode tested in isopropyl alcohol concentration of
ranging from 0 to 1250 mM.

2.5

Conclusion
A monolayer of enzyme was formed on microfabricated electrodes by both MSA

and CP-MSA methods. Experimental results suggest both methods can be utilized to
create an electrochemical enzymatic bioelectronic sensor interface. The CP-MSA method
29

developed by this thesis work does not require a gold electrode surface as its counterpart
does and therefore it is directly compatible with CMOS process. CP-MSA is selective
and does not foul the counter electrode. In addition, it saves not only expensive gold
material, but also at least 50% of the processing time. Table 2.2 summarizes the
performance of these two methods. The CP-MSA biosensor is favored from fabrication
point of view, but could benefit from sensitivity improvement.

Immoblization
Methods

Alcohol Sensor
Sensitivity
-1
-2
nAmM cm

CMOS
Compatibility

Process
Time

MSA

1160

post-CMOS

<12hrs

CP-MSA

1.30

Direct CMOS

<6hrs

Table 2.1 Comparison of sensitivity, CMOS compatibity and process time of MSA and
CP-MSA methods applied to form an alcohol sensor.

30

3
3.1

Planar Electrode System and Microfabrication

Overview
To integrate an electrochemical electrode system on surface of a CMOS chip

using foundry compatible microfabrication techniques, three electrodes system is
designed in a planar fashion. Section 1.3 introduces the role of working electrode, planar
electrode system performance and its geometry factors and design considerations. The
challenges for realizing a planar electrode system on CMOS exist in design and
fabrication, discussed in Section 1.5.3. In this research, issues involved electrode design
and fabrication is discussed in detail. For electrode on chip, chip surface profile, circuit
layout and microfabrication are also considered. The minimum size of the microelectrode
is also limited by the enzyme activity after immobilization. In respect of material, noble
metals such as gold are required for seeding the self-assembled bio-interface on the
working electrode. The noble metal counter electrode could be formed the same time.
But, the more complicated reference electrode requires additional processes, and remains
stable in test environment.
In this chapter, planar electrode system and microfabrication background and
analysis are given in Section 3.2 and Section 3.3. Prototyping planar microelectrode array
and improvement on reference electrode are discussed in Section 3.4 and Section 3.5.
3.2

Background
Microfabrication has been used for many applications in medicine and biology to

enhance the functions of conventional devices, such as tools for molecular biology and
31

biochemistry, tools for cell biology, medical devices, and biosensors [42].
Electrochemical sensors and microsystems were thoroughly reviewed in terms of
techniques and performance along with the issues by Suzuki [43]. Microfabrication of a
post-CMOS microelectrode system is an interdisciplinary problem spanning materials,
microfabrication and electrochemistry. To date, fabrication of microelectrodes on a
CMOS chip for enzyme-based electrochemical biosensors has not been reported. The
electrode geometrical factors of planar electrodes have to be determined before
fabrication. The design of an electrode system is complicated with electrochemical
effects, electrode materials, immobilization of enzymes, enzyme activities, CMOS
circuits and the microfabrication capabilities within the limitation of the small chip
surface area.
The geometries on microelectrode size, shape, gap and positioning were studied
by Bard [26], Whiteman [13] et. al. Few researches have constructed on-CMOS
microelectrodes. Kovacs et.al. has designed an electrode array for mercury anodic
stripping voltammetry [21]. Levine et.al. created a stepped-electrode process to simplify
fabrication [17]. Thewes’s group designed an CMOS gold interdigitated electrode array
for redox recycling of probe molecules on immobilized DNA probe [16]. A
microelectrode array was fabricated using CMOS process and electrolessplating [44]. To
date, the geometry of on-CMOS electrode design has not been fully discussed. Research
on planar microelectrodes for redox enzyme-based biosensors on CMOS has not been
reported yet.
A crucial part of an electrochemical cell is reference electrode. Microfabricated
planar reference electrode is a limitation of electrode performance. Research on
32

miniaturization of reference electrode and analysis of the parameters affecting its
performance, stability and lifetime, is sparse. The requirements for reference electrodes
are summarized in [27].
3.3

Analysis of Design Requirements
Well-established analytical electrochemical methods, such as cyclic voltammetry

and electrochemical impedance spectroscopy can be used to analyze the analyte
concentration. Although these methods are typically performed by using bench-top
instruments, the biosensor microsystem introduced in the previous section is
advantageous in many aspects. As the transducer for the integrated biosensing interfaces,
the electrodes are firstly to be built onto the silicon chip surface, targeting on building a
planar ultra-microelectrode array on the top of a CMOS chip. The involved issues include
electrode configuration, geometry, material, and reliability etc. In this chapter, above
issues are addressed by reviews, theoretical discussions and experiments.
3.3.1

Electrochemical Electrode System
A typical electrochemical cell consists of three electrodes: working, counter and

reference electrodes. We are only interested in potential and excited reactions on the halfcell of the working electrode. In the three-electrode system, no current passes through the
reference electrode. In a two-electrode system, the auxiliary electrode functions as both
the reference electrode and the counter electrode. The potential on the auxiliary electrode
is easily affected by the current passing through it, i.e. the electrode is polarized. Though
the polarization effect can be reduced by increasing the size of auxiliary electrode to at
least 100 times bigger than the working electrode, the drift of potential can hardly be
33

ignored, in terms of noise and potential control. Increasing the size of the auxiliary
electrode is often not practical. In applications when high density array is required, and
bulk silicon cannot be used as the auxiliary electrode, the usable chip area is limited to
within dimension of millimeters. Planar electrodes are more feasible to be fabricated onto
the CMOS chip surface than 3D electrodes. The chip surface is prioritized for building
the high density working electrode array. In all, a three-electrode transducer provides
more accurate and reliable potentiostat control to keep the protein from being
accidentally denatured, as well as a less noisy current readout. Low noise is especially
important to enhance the limit of detection, when the working electrode is miniaturized
and the proteins do not carry out a significant current in response of excitation.
3.3.2

Geometry of Planar Microelectrode
A working electrode plays a major role in an electrode system whose performance

is not only affected by its materials or modification, but also its geometric factors. A
microelectrode is an electrode having its critical dimension in micrometers. As the
dimension of the electrode became less than the thickness of the diffusion layer, a
microelectrode or an ultra-microelectrode presents many properties that are favored in
aspect of electrochemistry. A microelectrode not only has great spatial and temporal
resolution, but also provides the properties of high current density, fast to achieve steady
state current, low capacitance and immunity to high solution resistance. Thus, a
microelectrode or UME is more suitable for electrochemical sensors on-CMOS than a
regular electrode. By taking advantages of the microelectrode, the consumption of
analyte can be minimized. In addition, because a microelectrode can work in a high
resistance solution, a supporting electrolyte is not required, which simplifies the fluidic
34

delivery system and saves the precious space and weight of a portable device. On account
of the low capacitance and rapid achievement of steady state current, measurement of
chemical concentration can be made on a microsecond time scale [13]. Fast response of
individual sensor is desired for a large sensor array. Geometric factors of other electrodes
also affect the performance of the electrode system.
Planar (2-dimensional) electrode design is chosen over 3-dimensional electrode
arrangement to permit the fabrication on CMOS, although 3-dimensional electrode
arrangement has some advantages. Current microfabrication techniques are designed to
work on a planar surface. The electrochemical lab-on-CMOS is based on an integrated
circuit with its surface area in the millimeter scale, which is suitable for microelectrodes
but limits the fabrication area to the small chip surface. Once built, a microelectrode can
feed signals directly to the underlying CMOS circuits. CMOS and compatible
microfabrication techniques, including thin film deposition, photolithography, lift-off, dry
etching etc., can be employed to fabricate microelectrode system. Limiting the design to
use CMOS-compatible only techniques allows the developed process to be transferred to
a foundry for scaling up the production. The same microfabrication process as CMOS can
construct a high density microelectrode array that improves reliability by redundancy. A
microelectrode array can also accommodate several different bio-interfaces to detect
multiple bio-molecules.
For a planar electrode system, the geometrical layout of the three electrodes must
be optimized to performance. The working electrode and counter electrode should be
placed adjacent to (1) prevent loss of energy through joule heating across the solution
resistant, (2) prevent the possible change of solution micro-environment, eg. pH and (3)
35

reduce resistance across the solution reduces the charging time of the double layer
capacitance and speeds up amperometry readout. By putting the working and the counter
electrode close to each other, interference of convection along the flux between them is
reduced too. Theoretically, the distance from the reference electrode to the working
electrode should also be minimized to reduce the R contribution of IR loss for accurate
potential control. Since the flux between the working and the counter electrode may not
be uniformly distributed for a given geometry, the route from the reference electrode to
the working electrode is better chosen to be at a place with less dense flux, i.e. less
passing-by current, to minimize the I contribution of the IR loss. In practice, because a
reference electrode is not as easy to be miniaturized, it is hard to put the reference
electrode close to the working electrode. Exceptions are: the above IR loss can be
experimentally determined and then electrically compensated by the potentiostat; the
solution resistance is low; or enough supporting electrolyte can be added. Under these
circumstances, the reference electrode could be placed away from the working electrode.
3.3.3

Electrode Materials
Materials are chosen in compatible with microfabrication, chemical environment,

etc. For MSA method, the working electrode, gold is chosen because it is chemically
inert. It has a wide enough electrochemical potential window for biosensor. Also, gold
planar electrode is easy to be microfabricated. Furthermore, the gold-thiol bonding, a
strong covalent bonding, is widely used for attaching biosensing elements. Besides, gold
is also used for counter electrode material. An issue for using gold electrode is that it is
not a standard foundry fabrication process. Either a costly request on change of foundry
process or a post-CMOS process is required. Post-CMOS gold process can be chosen
36

from thin film deposition or wet chemical plating can be utilized, by weighing the pros
and cons. A pro for thin film gold is its surface quality; the cons are bonding, alignment
and more steps of processing compared to plating. Electroplated gold has good quality
but hard to apply on aluminum and requires wirings to a power supply. Electroless
plating of gold on aluminum can be done within a few steps, but the gold quality is
inferior to other gilding methods.
For reference material, the Nafion-coated Ag/AgCl reference electrode is chosen
over calomel reference electrode because it is biocompatible and easier to fabricate .
Nafion, provided by DuPond, is a sulfonated tetrafluorethylene copolymer with both
hydrophobic and hydrophilic regimes. The molecular structure of Nafion is shown in
Figure 3.1. A conventional Ag/AgCl reference electrode is a surface chlorinated silver
wire immersed in a 3M or 4M KCl solution to achieve constant electrochemical potential
and enclosed in a cylinder with a glass frit to provide liquid junction as well as preserving
-

Cl ions. As the system is miniaturized into an integrated electrochemical interface,
fabrication of a planar reference electrode becomes challenging. Screenprinted silver,
Nafion and other coatings over AgCl have been utilized to fabricate planar reference
electrode by providing diffusion barrier, ion-selective path, and hydrolyzation-protection,
at same time maintaining the inner chloride ions.
3.4
3.4.1

Planar Microelectrode Array
Design
In order to prove a post-CMOS fabrication process of a planar 3-electrode system,

a prototype is designed for immobilizing enzymes.
37

Figure 3.1: Electrochemical electrodes with 2×2 WE array and embedded Ag/AgCl RE.

The working electrodes of Figure 3.1 are 2mm in diameter, in a size that can
generate enough current for a normal bench-top potentiostat to detect after enzyme
immobilization. The surrounding counter electrode is sufficient larger than working
electrodes. The traces leading to the working electrodes and reference electrodes are
passivated with SU-8. The passivation leavings a well confined working electrode surface
2

area of 0.0314 cm . The 2×2 working electrode array were grouped in two, each group
shared one Ag/AgCl reference electrode.
3.4.2

Fabrication
The electrode array was fabricated on SiO2 substrate, mimicking surface of

integrated circuit for verifying process flow. As shown in Figure 3.2, the process flow for
fabricating system consisted of 5 steps. The first step was a thin film deposition of 20 nm
Ti/150 nm Au/200 nm Ag (Kurt J. Lesker Co., Clairton, PA) onto the SiO2 substrate
using an Edward Auto306 thermal evaporator. The second step was photolithography and

38

wet etching to form the electrode pattern. The positive film masks used for
photolithography were generated through a 3556 dpi Scitex Dolev 450 immagesetter
(Infinity Graphics, Okemos, MI). After spin coating, exposure, developing and post
development baking of the photoresist, the sample was dipped in to silver etchant
(NH4OH : H2O2 1:1), gold etchant (I2 : KI : H2O 4:1:4), and titanium etchant (HF : H2O
1 : 9) sequentially, leaving silver on all electrode sites. Step 3 was removing the
unwanted silver on the working and counter electrodes with another mask, covering the
reference sites for protection before dipping into silver etchant. Step 4 was to chlorinate a
thin layer of Ag to AgCl by oxidizing Ag by Fe3+. Researchers found out the reliability of
AgCl reference electrode were proportional to the thickness of AgCl [45]. Since only
200nm Ag was deposited, Ag need to be chlorinated, while preserving a layer of Ag on
Au contact without exposing AgCl to Au. 10mM FeCl3 solution was used. Because AgCl
10

has extreme low solubility product Ksp, which equals 1.8×10- , and the concentration of
-

Cl of 10 mM FeCl3 is 30 mM, AgCl can only go as high as 6×10-9 M in solution, from
which the amount of AgCl loss can be calculated. The amount of AgCl loss was further
minimized by limiting the volume of FeCl3 solution being used. After chlorination, the
3+

silver chloride surface was rinsed with saturated KCl to remove absorbed Fe

2+

and Fe ,

and then dried with nitrogen. Generally, anode chlorination was an alternative iron-free
method, with coulometry to determine the progress of chlorination, wherever a
galvanostat is available. For our special purpose of integrated reference electrode on a
circuit chip, anodic chlorination is not convenient. Gas phase chlorination of Ag is
promising for the same purpose, resulting uniform and denser packed AgCl [46]. Step 5
39

was to cover AgCl with Nafion (Nafion 117 5% solution, DuPond) layer on the reference
electrode. Nafion allows only protons to pass the membrane, thus to keep the junction
-

+

conductive, while detaining the Cl and Ag ions. With Nafion, the potential of the
reference electrode was kept stable from interfering Cl-, Br- and S2-. The Nafion was
coated on the electrode and cured at 120°C for 1 h according to Moussy’s method [47].
An optional polydimethylsiloxan (PDMS) mold was prepared to facilitate modification of
individual working electrode as well as protecting all the other electrodes from being
contaminated.

Figure 3.2: Fabrication sequence for integrated electrochemical electrode arrays with
embedded Ag/AgCl reference electrodes.

3.4.3

Results and Discussion
After the electrode array was fabricated, it was characterized using cyclic

voltametry. The reference electrode was first tested. The microfabricated planar Ag/AgCl
40

reference electrode was characterized by comparing the open potential versus a
commercial Ag/AgCl reference electrode over time in 3M KCl solution, shown in Figure
3.3. The Nafion membrane was a dry film initially. When dipped in to solution, the
sulfonic acid sites were hydrated for protons to tunnel through. Region A displays the
membrane potential fluctuated during the progress of the hydration of the hydrophilic
sulfonic acid site of the polymer. It is not preferable to use this RE in Region A. In region
B, the potential is relatively stable, can be used for amperometry. The potential
fluctuation was largely due to the hydration process and destruction of the Nafion
membrane. After about 5h, in Region C, the Nafion membrane was not holding
membrane potential, indicating the corruption of the membrane. The Nafion thin film
finally failed because of hydrolysis. Since the glass frit used in commercial electrodes,
there is potential difference even when equilibrium is reached. The reference electrode
thus acted as it would without a Nafion layer. The reference electrode will perform like
pseudo reference electrode that the potential is easily affected. It is a complex process
that greatly affected the stability of reference electrode. Diffusion barrier and other
membrane can be added in attempt to fortify the structure. More sophisticated coatings
can improve long-term stability of microfabricated on-chip reference electrode, and
applied when application requires. In next section, a coating method is developed to
provide diffusion barrier and a protective layer between Ag/AgCl and Nafion. Based on
the experiments, Ag/AgCl reference electrode stability needs to be improved. In Section
-

3.5, a Ag/AgCl/PPy(Cl )/Nafion quasi-reference electrodes is developed to solve the
stability issue.

41

Figure 3.3: Potential difference between microfabricated planar on-chip Ag/AgCl
reference electrode and commercial Ag/AgCl electrode (N=8).
Individual working electrodes are examined with cyclic voltammetry using CHI
660B potentiostat (CH Instrument, Austin, Texas) in potassium ferricynide. Figure 3.4
shows the experiment with microfabricated on-chip planar Ag/AgCl reference electrode.
The peak shift due to potential difference versus commercial reference electrode can be
observed as expected. In addition to being affected by the microfabricated reference
electrode, the difference and variety may also cause by electrode surface conditions or
convection by comparing the after-peak diffusion-limited regime approaching 0V. By
using the on-chip reference electrode, the formal potential shift is 16 mV and is stable
over experiments. All the working electrodes are within small variations, and can be
regard as identical.

42

Figure 3.4: CV plots of 4 working electrodes in 10mM K3Fe(CN)6, 3M NaCl, scanning
rate 100 mV/S using on-chip Ag/AgCl reference electrode.

3.5
3.5.1

Reference Electrode
Introduction
Reference electrode is an electrode which has a stable and known electrode

potential. The potential represents the energy required to take a charge from one side of
the electrode/electrolyte interface to the other, relates to the chemical activities of the
constituents of the electrochemical reaction by the Nernst equation. By employing a
redox system with constant concentrations of each participants of the redox reaction, a
high stability of the electrode potential can be reached. Reference electrodes are used to
determine and control an electrode of interest in an electrochemical cell. Quasi-reference
or pseudo-reference electrode is so named because it does not maintain a constant
potential; therefore, by definition, it is not a true reference electrode. However, its
potential depends on conditions in a well-defined manner; if the conditions are known,
the potential can be calculated and the electrode can be used as for reference potential.

43

Fundamentals of reference electrodes and review of current research can be found in [27,
48].
Common reference electrodes are standard and normal hydrogen electrode,
saturated calomel electrode, copper/copper(II) sulfate electrode, silver/silver chloride
electrode, palladium-hydrogen electrode, iridium/iridium oxide electrode etc. The
silver/silver-chloride (Ag/AgCl) reference electrode is by far the most common type of
electrode used in biosensing applications due to its simple construction and
biocompatibility. According to Nernst equation, the chlorine ion acts as the chemical
species in the operation of the electrode, and its activity, which is related to its
concentration, has a significant impact on the electrode potential [49]. Other types of
reference electrode were also developed. Yttria-stabilized zirconia membrane electrodes
were developed with a variety of redox couples, e.g., Ni/NiO. Their potential depends on
pH. When the pH value is known, these electrodes can be employed as a reference with
notable applications at elevated temperatures. A recent advance is a metal/PPy reference
electrode, which is solid-state and works in both aqueous and non-aqueous environment
[50]. A solid state reference electrode is fabricated for in situ voltammetric analysis in
solutions containing little or no added supporting electrolytes, coated with an electrolyte
immobilized and protected with Nafion or polyurethane [51].
Microfabrication procedure of planar reference electrode includes base metal
deposition and patterning, activation to create the reference material such as AgCl, and
chamber interface formation. Thin film deposition, screen printing and electroplating are
often used for creating a patterned metal base [45, 52-54]. Then the reference material is

44

applied as a form of mixture, chemically or electrochemically transformed [46, 54, 55].
The last step is coating with porous materials or gels [14, 47, 56-61].
In our integrated biosensor application, a bio-compatible solid-state quasireference electrode that works in buffed solution is desired. The quasi-reference electrode
should be produced by microfabrication processes and can be sized to match the
microelectrodes. In last section, an electrode array with Nafion coated reference electrode
was fabricated. The reference electrode provides anti-interference capability.
3.5.2

Design
To miniaturize electrochemical microsystems, microfabricated Ag/AgCl reference

electrodes have been investigated by Moussy et. al . The potential of the bare Ag/AgCl
without ion-selective and diffusion barrier membranes has been observed to be affected
by chloride concentration and other interferences. Meanwhile, PPy has been used for
reference electrode. It can be formed on Ag/AgCl without additional patterning, and it
+

-

retains Ag and Cl ions and provides ion-exchange capabilities.

The reference electrode is a crucial element in electrochemical biosensors. The
size and reliability of the reference electrode are limiting factors for miniaturizing
electrochemical systems. A microfabrication procedure for planar Ag/AgCl reference
electrodes utilizing polymer coatings is developed. The reference electrode consists of an
evaporated Ti/Au/Ag thin film with a portion of the silver chlorinated to Ag/AgCl and bilayer polymer coatings. Electropolymerized polypyrrole seamlessly covers the AgCl to
serve as a diffusion barrier and an anion exchanger. A cation exchanging Nafion coating
layer compliments the ion exchange properties of the polypyrrole while blocking other
45

interference ions. The structure of Nafion molecule is displayed in Figure 3.5. It has both
hydrophobic fluorine carbon sites and hydrophilic thiol sites.

Figure 3.5: The molecular structure of Nafion.
The structure of the quasi-reference electrode is illustrated in Figure 3.6. The
Ag/AgCl disk electrode is embraced by an un-closed gold ring, because Nafion was
-

discovered having good bonding to gold because of its thiol group. PPy(Cl ) is
electropolymerized on Ag/AgCl to provide a diffusion barrier and bonding agent to
Nafion.

1mm

Nafion
PPy

AgCl
Ag

Au
Ti

Glass
-

Figure 3.6: Microfabricated Ag/AgCl/PPy(Cl )/Nafion quasi-reference electrode
structure.

46

3.5.3

Fabrication
-

Ag/AgCl/PPy(Cl )/Nafion quasi-reference electrodes were fabricated on a SiO2
substrate, presented in Figure 3.3. The size was 1 mm in diameter, and can further be
scaled down to microns. To achieve this, thin films of 5 nm Ti and 100nm Au were
evaporated on the substrate and patterned using photolithography and wet etching to form
electrode bases and traces to an edge connector. 400nm Ag was deposited over the Ti/Au
adhesion layers and patterned using lift off over the reference electrode sites. The
electrodes were dipped in a 10 mM ferric chloride (FeCl3) solution to chlorinate ~200 nm
of the silver layer to form an Ag/AgCl interface. The electrode was then dipped into 100
mM pyrrole and 1M KCl solution. By running cyclic voltammetry from 0 to 650 mV at
100 mV s

-1

for one cycle, a layer of partially oxidized polypyrrole doped with Cl

-

-

(PPy(Cl )) was formed over the AgCl. Finally, a 5% Nafion 117 solution was spun
coated, cured at 120˚C for 1hr, and then soaked in 1M KCl for 60 min so the hydrophilic
clusters of Nafion could be hydrated to conduct protons. To characterize the electrode,
the potential difference of the quasi-reference electrode, relative to a commercial
Ag/AgCl reference electrode was recorded using a data acquisition card and LabView.
-

The microfabricated Ag/AgCl/PPy(Cl )/Nafion quasi-reference electrode is shown in
Figure 3.7:

47

1mm

-

Figure 3.7: The photograph of the microfabricated Ag/AgCl/PPy(Cl )/Nafion quasireference electrode.

3.5.4

Results and Discussion
-

The Ag/AgCl/PPy(Cl )/Nafion quasi-reference electrode was tested in KCl
solutions. The results show that the polymer bi-layer stabilized the potential of the
-

Ag/AgCl system, comparing with a bare Ag/AgCl electrode in Figure 3.8. The PPy(Cl )
-

provides effective chloride diffusion barrier to maintain constant Cl concentration at
Ag/AgCl interface. The potential changes with KCl concentration because KCl affects
the hydration state of the Nafion layer. Since the hydrated group passes proton, the
membrane potential is affected as a result. The built-in potential of the two polymer
interface with the PPy as electron donor and the Nafion as proton donor is secondly being
affected by the reduced carrier concentration as a consequence of the process of
dehydration due to the higher concentration of salt. The slope of potential indicates the
progress of the process. The result shows that the chloride dependence is significantly
reduced with the new electrode structure. Data were taken once a steady potential was
reached; ramping before steady state is due to the penetration of ions across the

48

membrane and redox states of the PPy layer when concentration changes. Further tests in
multiple interfering ions can be performed to examine its rejection to interferences.

Figure 3.8: Potential difference of microfabricated Ag/AgCl/PPy(Cl-)/Nafion quasireference electrode vs. commercial Ag/AgCl reference electrode in various concentration
of KCl. In the plot, trace a and b are the Ag/AgCl/PPy(Cl-)/Nafion quasi-reference
electrode; c is a bare Ag/AgCl electrode.
The membrane potential is a function of the proton concentration in the solution.
The redox state of PPy is also affected by protons crossing the Nafion layer. Figure 3.9
-

shows the potential difference of microfabricated Ag/AgCl/PPy(Cl )/Nafion quasireference electrodes vs. a commercial Ag/AgCl reference electrode in standard pH
-

buffers. In the plot, trace a and b are the Ag/AgCl/PPy(Cl )/Nafion quasi-reference
electrodes; c is a bare Ag/AgCl electrode.

49

-

Figure 3.9: Potential difference of microfabricated Ag/AgCl/PPy(Cl )/Nafion quasireference electrode vs. commercial Ag/AgCl reference electrode in standard pH buffers.
In the plot, trace a and b are the Ag/AgCl/PPy(Cl )/Nafion quasi-reference electrodes; c is
a bare Ag/AgCl electrode.
For biosensors operating in a pH-buffered solution, the potential can be
predetermined by using its potential value in standard pH buffer and Nernst equation, so
the shift is tolerable. The Electrode fails after prolonged use, due in part to cracks that
develop in the Nafion coating, which limits its long-term application.
3.6

Conclusion
A fabrication process for microfabricating microelectrode system on CMOS chip

was developed based on the understanding of electrochemistry, within limitations of
microfabrication capability. A prototype system was fabricated and tested. The working
electrodes were identical and the reference electrodes were characterized. Geometry of
planar electrode was discussed in order to be implemented on CMOS. To verify
microfabricated electrode with enzyme immobilization, a gold electrode array was
-

designed and fabricated. To overcome the limitation of Nafion, Ag/AgCl/PPy(Cl ) and ph
50

buffered The method was identified for fabricating a reference electrode that can be
miniaturized to fit on a CMOS die.

51

4
4.1

CMOS Bipotentiostat

Overview
A potentiostat is the instrument to control a three electrode cell and provides

current readout. For enzyme biosensor array and CMOS work synergistically, redox
recycling mechanism can be utilized to increase sensitivity and reduce the overall
electrode area. It requires a pair of interdigitated working electrode under support of a
bipotentiostat, a potentiostat capable of controlling two working electrodes respectively.
In this chapter, the traditional adder potentiostat and a new architecture of a bipotentiostat
are discussed in section 4.3. Redox recycling on concentric ring electrode is analyzed
experimentally and mathematically in section 4.4. A CMOS bi-potentiostat is reported in
section 4.5.
4.2

Background
Redox enzymes catalyze oxidation or reduction reactions with direct electron

transfer, making it highly suitable for electrochemical sensors that provide continuous
real-time monitoring of the target biomolecules. However, some redox enzymes have
very small turnover rates, resulting in low biosensor sensitivity and low current levels
when miniaturized. The redox recycling effect can be utilized to electrochemically
magnify the faradic current using, for example, an interdigitated electrode array with
enzyme immobilized on both working electrodes. To support redox recycling, a
potentiostat must have simultaneous control of both WEs.

52

In the past two decades, significant research has been devoted to CMOS
integrated potentiostats, which have been applied to many applications including anodic
stripping voltammetry to detect heavy metal ions [21], high density electrode arrays with
low current readout [62], wide potential window [63], examining DNA match [20], for
reversible redox species [64], for neuron transmitters [65], with wide dynamic range [66],
for detector array [54] and others. In addition to CMOS, a potentiostat was constructed
using thin film transistors [49]. However, potentiostat architecture is not fully studied
[67]. A CMOS potentiostat supporting redox recycling with redox-enzyme-based
biosensors has not been reported.
4.3
4.3.1

Potentiostat Architecture
Electrochemical Cell and Potentiostat
In electrochemistry, a potentiostat is a piece of equipment that is used to control the

potential of working electrode with respect to the reference electrode, reading out the
current flow through the working electrode while the excitation wave form is applied on
it. Before designing a potentiostat, the basic electrical model of the electrochemical cell is
worth a discussion beyond textbook. Electrochemical cell is two or three electrode in
electrolyte. It can be modeled as an impedance network tapped by connections to
electrodes as symbolized by Figure 4.1(a). The key to potential control is setting a
potential at the surface of working electrode, where the interested reaction takes place at
this potential. In an often seen circuit equivalent model displayed in Figure 4.1(b), the
surface of electrode is labeled as point A, between the working electrode impedance Zw
and uncompensated solution resistance Ru. B is the tapping point of the reference
53

electrode into the cell, the point of which electrochemical potential is certain. Point C is a
reference point in electrical circuit. A reference electrode relates electrochemical
potential to the electrical world. Also in this model, ZC and RΩ represent the impedance
and compensated solution resistance on the counter electrode side. Notice that the total
solution resistance between WE and CE is divided into RΩ and Ru, There values depend
on the tapping position of the reference electrode on the current path between counter and
working electrode. A reference electrode has its impedance ZR and solution resistance
RR, as shown Figure 4.1(c). Besides, the thermal dynamics at the solid/liquid interface
established a potential, represented by ER. In fact, there is virtually no current flow
through the reference electrode. Therefore, its internal resistance, junction resistance and
capacitance can be omitted. The simplified equivalent impedance network with the
reference potential ER and is shown in Figure 4.1(d).

Figure 4.1: Electrochemical cell and its equivalent impedance network. (a) A
electrochemical cell symbol. (b) The equivalent impedance network. (c) The
impedance network with the built-in potential and impedance of RE, (d) A simplified
impedance network with the built-in potential and of RE.

54

In an electrochemical cell the potential at the working electrode is concerned. By
analyzing current responses under excitation waveforms, the reaction on the working
electrode surface can be analyzed. Since the potential of reference electrode in the
electrolyte is relatively stable, it is used as the reference point to control the potential of
the working electrode by a potentiostat. A simple potentiostat is shown in Figure 4.2,
where an op-amp is used for potential control. A stimulus EX is inserted between RE and
WE. Further, by replacing the electrochemical cell with equivalent circuit, the potential
control circuit model is shown in Figure 4.3 (a). It appears more clear that the RE tapping
point is biased to EX plus ER versus ground, and that the working electrode is grounded.

Figure 4.2: A simplified potentiostat control over an electrochemical cell.

Because of the portion of solution resistance Ru, the true potential on the working
electrode is less than the desired value by an uncompensated voltage drop iRu, called IR
loss. This loss could not be eliminated, but can be reduced by minimizing Ru or be
ignored if the inaccurate potential control is not significantly affecting the experimental
results. Mounting the reference electrode to the vicinity of working electrode reduces Ru.

55

The potential control circuit applies potential through counter electrode so that the
potential on the current path of current at the point of reference electrode follows the
excitation signal.
An alternative scheme of potential control circuit is by inserting an excitation
signal from the working electrode, as shown in Figure 4.3 (b) Scheme B. To differentiate,
the scheme in Figure 4.3 (a) is referred to as scheme A. In Scheme B, the electrochemical
potential of the point of tapping is controlled to approximate the same potential as
reference electrode. The excitation Ex is applied from the working electrode versus
ground, as shown in Figure 4.3(b).

Figure 4.3: Potentiostats with the electrochemical cells being replaced by the
equivalent circuit models. (a) Scheme A: The excitation signal Ex feeds through an
opamp. (b) Scheme B: The exiciation signal Ex feeds through the working electrode.

To further study these two schemes, equivalent circuits of scheme A and scheme
B are derived by converting the opamp negative inputs to grounds and outputs to voltage
sources Av. The two schemes of potential control circuit models are simplified for better

56

comparison, as shown in Figure 4.4. From Kirchhoff's voltage law, a loop voltage for the
circuit in Figure 4.4 (a) is:

Figure 4.4: The equivalent circuit models of the two schemes of potential control and
excitation signal insertion. . (a) Scheme A: The excitation signal Ex feeds through an
opamp. (b) Scheme B: The excitation signal Ex feeds through the working electrode.

−‫ܧ‬௑ – ‫ܧ‬ோ + ܴ݅௨ + ܼ݅ௐ = 0
The loop voltage for Figure 4.4 (b) is:

– ‫ܧ‬ோ + ܴ݅௨ + ܼ݅ௐ − ‫ܧ‬௑ = 0

From above circuits and equations, we can observe that the voltages across the
working electrode are identical, if internal resistance of excitation is ignorable:

ܼ݅ௐ = ‫ܧ‬௑ + ‫ܧ‬ோ − ܴ݅௨

57

From above equation, it is observed that the true working electrode potential
versus the potential reference EX, iZW- ER equals to EX - iRu which is desired excitation
EX offsets by the uncompensated iRu drop. From above analysis, the two schemes of
potential control have identical equivalent working electrode potential (ignore internal
resistance of excitation source EX).

For an electrode array, it is desired that each electrode can be individually
controlled by an excitation signal. To determine the suitable potentiostat architecture, the
models of Scheme A and Scheme B are expanded to multiple channels, as illustrated in
Figure 4.5. Scheme A requires subtracting Ex from each additional channel by additional
subtractors. The control scheme B in is more hardware efficient for applying excitation
signals on the multiple electrodes. As of electrode geometry for both control schemes, it
is equally important to position the working electrode as close as possible to reduce Ru
whenever possible.

58

Scheme A

Scheme B
Figure 4.5: The equivalent circuit model of expanded Scheme A (top) and Scheme
B (bottom) to multiple channels;

4.3.2

Bipotentiostat Architecture
In some electrochemical experiments, two working electrodes are required to

individually controlled and work simultaneously. The electrochemical instrument is
called bipotentiostat. In Section 4.3.1, potentiostat models were expanded to support
array. Those models can be simplified to support two channels. The potentiostat of
Scheme A in Figure 4.5 is called “adder potentiostat”, because the excitation signal is
59

added to the reference potential at the inputs of an analog adder. A circuit to implement
the scheme is comprised of buffers and current to voltage converter.
In the adder potentiostat, the potential at the reference electrode tip approximates
the true potential on the working electrode in vicinity, equals to the excitation signal for
this first working electrode. To adapt an adder potentiostat to simultaneously control a
second working electrode by this approach, the excitation potential is subtracted from the
excitation signal of the second working electrode through an analog subtractor, as shown
in Figure 4.6. The resulting signal is the difference of the two signals, used to bias the
second working electrode. At the interface of 2
st

versus the 1 excitation signal recovers the 2

nd

nd

electrode, the potential difference

excitation signal.

Figure 4.6: A bi-potentiostat based on traditional adder potentiostat architecture.

60

The above adder potentiostat approach is complicated in potential control because
it employs the potential of the first working electrode. Moreover, the error introduced by
the uncompensated potential loss as shown in the equation below is propagated to the
second working electrode.

‫ܧ‬௑ = ܼ݅ௐ − ‫ܧ‬ோ + ܴ݅௨
An alternative approach is as suggested by the above analysis in section 4.1,
removing the excitation signal from the adder and biasing the working electrode with it.
In this approach, the reference potential is disassociated with excitation signal, so each
working electrode is biased with the original excitation signal without additional
arithmetic processing, as illustrated in Figure 4.7.

Figure 4.7: (a) A bipotentiostat by expanding Scheme A, the traditional adderpotentiostat architecture. (b) A bi-potentiostat for electrode array based on Scheme B
with two parallel channels.

61

As a result, Scheme B saved two opamps for performing inversion of e2 and
subtraction of e1 out of e2, compared to Scheme A potentiostat circuit illustrated in
Figure 15.46 in Bard’s book [26]. Because of these benefits, the Scheme B architecture is
adopted. The impact of saving two opamps on CMOS layout is discussed in Section 4.5.
4.4
4.4.1

Redox Recycling on Concentric Ring Electrode
Background
The current of enzyme biosensor is a function of enzyme activity, electrode

geometry and substrate concentration. For on-chip enzyme modified electrode, the chip
surface area is limited and the enzyme activity is fixed. Increasing the sensitivity of the
sensor can reduce the sensor footprints to allow more sensors on the chip surface.
4.4.1.1 Redox Recycling
We assume the reversible reactions below take place on a pair of electrodes at
reduction potential and oxidation potential respectively:

ܴ − ݊݁ ↔ ܱ
ܱ + ݊݁ ↔ ܴ
where R and O represent the reduction and oxidation state, n is the number of electrons
involve in the reaction, and e is an electron.
In traditional electrochemical system, the species including the concerned
analytes migrate toward the working electrode and leave the electrode surface after being
transformed. The reverse transformation will not happen until a favoring potential being
applied. The current is limited by mass transfer including diffusion and migration when
62

the species are transformed quicker than replenished. The migration is proportional to the
electric field strength. According to Fick’s law of diffusion, the current density is
proportional to the bulk concentration of the species being transformed. For a given size
of electrode, the detection limit is thus limited.

‫ܦ− = ܬ‬

߲‫ܥ‬
߲‫ݔ‬

To improve the limit of detection on an electrode of determined size, an additional
working electrode can be situated close by, preferably within the distance of the diffusion
layer ݈, determined by equation below:

݈ = √2‫ݐܦ‬

where ‫ ܦ‬is diffusion coefficient, ‫ ݐ‬is the time span of experiment. The limit of detection
can be improved in several aspects by this additional working electrode. By setting it to a
potential of the reverse reaction, the transformed species consumed at one electrode are
reproduced at the other, locally increasing the concentration of the reacting species, thus
to increase the concentration gradient for diffusion and current density in consequences.
In addition it increases the electric field by a larger potential difference and a shorter
distance, promoting the migration of charged species thus increases the background
current as well.

Since the species being reduced or oxidized at one electrode and

oxidized or reduced on the other without diffusing into the bulk solution, it reduces the
invasion of the environment being tested. This is known as “redox recycling”.

63

To realize redox recycling, the two electrodes of an interdigitated band array are
often set to reduction potential and oxidation potential respectively. It also requires a
bipotentiostat to control the potential on each electrode independently. The geometry of
the interdigitated array electrode (IDA) and bipotentiostat are discussed further in the
next sections.
4.4.1.2 Concentric Ring Electrode
From the discussion in previous section, an IDA has some advantages over a
traditional disk electrode. From the analysis, the steady state current of an IDA, unlike
disk electrode, is a function of time, though ignorable when the length to width ratio is
sufficiently large [26]. The disk electrode is stable at steady state, but the interior of the
disk supports very little electrochemistry so can be eliminated. The resulting ring
electrode resembles the behavior of a disk electrode and is more area-efficient. A carbon
film-based interdigitated ring array electrodes was designed as detectors in thin-layer
radial flow cells [68]. A e-beam lithography microfabricated gold ring electrode array
was constructed for redox chemicals [64]. Not only in electrochemistry, were ring
electrodes also applied for detecting surface electromyographic signals with high spatial
selectivity [69].
The concentric ring structure combines disk and IDA, and can benefit from both
of the advantages and avoid the disadvantages. As an example shown in Figure 4.8 shows
a 90 µm concentric ring electrode fabricated on a 0.5µm AMI C5N CMOS process
through MOSIS.

64

Figure 4.8: An exampleconcentric ring electrode fabricated on a bipotentiostat chip by
CMOS 0.5 µm technology. The diameter of the outer ring is 90 µm.
A mathematical model was developed based on diffusion equation, and its
boundary conditions were defined in Appendix. This model can be used for finite element
analysis using software tools such as COMSOL MultiPhysiscs to estimate the diffusion
current at redox recycling to evaluate the electrode geometry factors.
4.4.2

Fabrication of Concentric Ring Electrode
To investigate the redox behaviors on a concentric ring electrode, a concentric

rings-disk electrode was fabricated. It comprised of two neighboring ring working
electrodes, an outer ring counter electrode and an inner disk reference electrode. The disk
electrode in the center is in a diameter of 2.54 mm. The disk electrode is surrounded by
three concentric ring electrodes each has an opening at the same direction for wiring. The
first, the second, and the third rings are 0.7 mm, 0.5 mm, and 0.75 mm width,
respectively, noted as WE1, WE2 and CE respectively. The gap between the first and the
second ring electrodes and the second and the third electrodes are 50 µm and 250 µm,
respectively, shown in Figure 4.9.

65

Figure 4.9: A concentric ring electrode fabricated on glass substrate, including two
neighboring working electrodes, WE1 and WE2, a surrounding ring counter electrode
and CE a disk reference electrode RE.

The electrode in Figure 4.10 was prepared by physical vapor deposition of 5nm Ti
followed by 100nm Au on glass slides. Then photolithography and wet etching were
applied.
4.4.3

Results and Discussion
To explore the redox recycling capability of the electrode in Figure 4.10 was

coated with ionic liquid BMIBF to study the redox behaviors of nitro aromatic
4

compounds. BMIBF is able to absorb nitro compounds vapor such as ethylnitrobenzene
4

(ENB) in atmosphere. The electrochemical measurements were done with an EG&G 273
potentiostat. The scan rate of cyclic voltammetry was 100 mV/s. Differential pulse
voltammetry (DPV) was done with a pulse height of 25 mV, a scan rate of 20 mV/s, and
a pulse width of 50 ms.
The experiment was conducted in air and in 1 mM ENB in BMIBF . The results
4

of CV showed the two reduction processes of the ENB shown in Figure 4.10. Typically, a
reversible reduction and a consequent irreversible reduction were observed. The first

66

reduction peak at -1.7 V (vs. Au) corresponds to the reduction of the nitro (-NO ) group
2

-.

to a nitro anion radical (-NO ). In most electrolyte solutions this reduction is
2

electrochemically and chemically reversible. The second reduction of ENB at -2 V (vs.
Au), mostly irreversible, is related to the protons in the electrolyte solutions. In this step,
the nitro radical anion is further reduced to nitro dianion and eventually hydroxylamine.
The oxygen peak at -1.25 V (vs. Au) in air can be observed in ENB saturated air, and can
be used as a reference.

Figure 4.10: CVs of bmiBF4 exposed in air and in ENB saturated air.

DPV was conducted to help identify the redox recycling. In the experiment, one
working electrode was fixed beyond the oxidation potential of nitro anion radical at -1.6
V (vs. Au). Te differential pulse was applied on the other working electrode and the
current was plotted. A magnified reduction current can be reserved, which is due to the
67

regenerated nitro group. On the contrary, the other peak was diminished because the
oxidation reaction was irreversible. It’s unable to be replenished once consumed, even
though the potential on the other working electrode was set to a potential beyond its
oxidation point.

Figure 4.11: DPV plots of bmiBF4 exposed in air and in ENB saturated air.

The above results experimentally proved the current magnification effect of a
redox reversible species on a microfabricated concentric ring electrode. From these
result, we can predict that it is promising that a similar current magnification effect will
work on an enzyme-modified concentric ring electrode.
4.5

Result and Discussion
A CMOS bipotentiostats were fabricated using AMI 0.5µm process through

MOSIS [14]. The feedback resistors were replaced by switching capacitors for better
68

matching in the trade-off of control complexity. The bipotentiostat is constructed by
using the op-amps from the earlier adder potentiostat. In this CMOS bipotentiostat, the
potential control pin Vcomp is tied to ground. The two readout circuits are in parallel. The
two excitation signals bias the two working electrodes. The functional structure of the
circuit is as shown in Figure 4.8. The chip also carries an on-chip three by three electrode
array formed on top of CMOS metal 3, including a set of concentric ring working
electrodes locate in chip center.
Figure 4.12 below shows the layout of the CMOS bipotentiostat with concentric
ring electrode on the chip. The chip size is 1.56×1.69mm. Each readout channel occupies
2

600µm×280µm, or 0.168 mm . The two opamps in the potential control block used two
2

identical class A-B opamps, 390µm×173µm each, or 0.067 mm . If an adder potentiostats
were to adopt those opams in their readout arithmetic circuits of additional channels, each
2

channel uses two additional opamps, the extra area would be 0.134 mm . Based on above
assumption, each additional channel of a potentiostat using the new architecture saves
44.37% area.

The offsets were tested and adjusted through the Vcomp input. By connecting
WE1 or WE2 to a Keithley 2400 source meter, the current to voltage (-V) converter
readout channel was characterized for subsequent measurements. Figure 4.13 shows the
I-V converter calibration curves ranging from 10 µA to 10 nA, adjusted by changing the
clock frequency.

69

Figure 4.12: A CMOS bipotentiostat with concentric ring electrode and the circuit
architecture. The chip size is 1.56
1.56ൈ1.69mm.

The final test steps include circuit test, sensor test and final system test. T
The
integrated potentiostat works in either single channel mode or dual channel bipotentiostat
dual-channel
mode. The test of the potentiostat follows the published method as the previous version
of potentiostat. Once configured in bipotentiostat mode, it is to adapt co
concentric ring
electrode to readout the current on both generator and regenerator electrode. The packed
integrated potentiostat die is to be tested before enzyme is being immobilized on.
The example fructose biosensor was assembled on a 5 mm by 5 mm IDA
fabricated on oxidized silicon using photolithography, deposition of Ti/Au and lift
ricated
lift-off.
The IDA had 5 µm wide fingers and 5µm gaps. The MDH enzyme was immobilized on
5µm
the IDAs following published procedures [10]. It was then fixed in glutaric acid for 20
.
minutes. The functionality of the MDH modified IDA was tested using a CHI 760
commercial bipotentiostat in the presence of 200 mM D fructose in pH 6.0 phosphate
D-fructose
buffer solution. A platinum counter electrode and an Ag/AgCl reference electrode were
used for subsequent experiments.
70

Clk=100 KHz

Vout (V)

1.3
y = -0.039x + 1.277
1.1

0.9
0

10

Iin ( μA)

Clk=50 KHz

Vout (V)

1.3
y = -0.129x +1.305
1.25

1.2
0

Iin ( µA)

1

Clk=10 KHz
1.3
Vout (V)

y = -0.423x +1.304
1.28

1.26
0

Iin ( µA)

0.1

Figure 4.13: Readout opamp I-V conversion output at 100 KHz, 50 KHz and 10 kHz
clock frequencies.

71

A cyclic voltammetry experiment of redox recycling was conducted by sweeping
the potential on WE1 while WE2 was set to 0.2V. Figure 4.14 (a) shows the D-fructose
reduction peak at 0.03V and the D-mannitol oxidation peak at about 0.2V vs. Ag/AgCl.
This data established the potential range for subsequent chronoamperometry experiments.
As expected, the cyclic voltammogram shows a reduction peak current under redox
recycling (3.3nA) that is 27% higher than without redox recycling (2.6nA),
demonstrating

the

enhanced

sensitivity

of

redox

recycling.

(a)

(b)
Figure 4.14: D-Fructose sensor response on MDH modified IDA. (a) Cyclic
voltammetry of 200 mM D-furctose with (solid) and without (dotted) redox recycling
(top). (b) Chronoamperometery of 400 mM, 200 mM and 50 mM D-fructose with
redox recycling (bottom).
72

In a chronoamperometry experiment, in order to measure the fructose
concerntration, the commercial bipotentiostat was configured by setting to perform a
potential step from E1 at -0.1V on WE1 to E2 at 0.2V on WE2 versus an Ag/AgCl
reference electrode. The steady-state currents were recorded at 250ms after applying the
potential steps. Then the CMOS bipotentiostat was connected to the fructose sensor to
perform above chronoamperometry in 100mM, 200mM and 300mM fructose solutions.
The output voltage was converted to concentration using the I-V calibration curves. For
comparison, chronoamperomtetry experiments were repeated using a commercial
bipotentiostat in 400mM, 200mM and 50mM fructose solution. The fructose
concentration measured by both CMOS bipotentiostat and commercial bipotentiostat are
plotted in Figure 4.15, demonstrating that our CMOS bipotentiostat shows good
agreement with standard commercial instrumentation. This research was published in

Concentration (mM)

500
Triangle:
CMOS bipotentiostat
Diamond: Comercial bipotentiostat

400
300
200
100
0
0

50

100

Current (µA)
Figure 4.15: Fructose concentration vs. current measured by a commercial
potentiostat and the CMOS bipotentiostat. Diamond marked data is from CHI 760
bipotentiostat. Triangle marked data are from the CMOS bipotentiostat.
[14].
73

4.6

Conclusion
A new bipotentiostat architecture was constructed to replace bipotentiostat based

on traditional adder potentiostat. A CMOS bipotentiostat was structured to support redoxenzyme-based biosensors, providing simultaneous potential control over two WEs,
sufficient potential window and current readout range, flexibility for offset adjustment.
The new architecture saves 44.37% area for each additional channel implemented using
AMI 0.5μ݉ CMOS process and previously published opamp. A redox-enzyme-based
biosensor was shown to work seamlessly with the CMOS bipotentiostat in applying
chronoamperometry to determine biomolecule concentration. A fructose sensor was
created by molecular self-assembled MDH on a microfabricated gold IDA. Interrogated
in the redox recycling mode, the sensor exhibited an enhanced sensitivity of 27%.
Chronoamperometry experiments using the CMOS bipotentiostat showed good
consistency in accordance with a commercial bipotentiostat. Success with the fructose
sensor suggests that, by exchanging the redox enzyme, the CMOS bipotentiostat can be
used to enhance sensitivity of other redox-enzyme-based biosensors for detecting
glucose, alcohol and other important biomolecules.

74

5
5.1

Lab-on-CMOS System Integration

Overview
The advance of IC technology allows integrating an electrochemical instrument on

a single chip, such as potentiostat and impedance spectroscopy circuits. Using the
microfabrication technique employed by IC industry, microelectrodes can be directly
formed at the time of ICs fabrication by foundry, or after foundry by post-CMOS process,
allowing on-chip chemical sensing. Biosensing capability can be added to the electrode
by immobilizing biosensing materials. The challenges of integrating biosensors on
CMOS die exist on their compatibilities during processes and tests. The sensors must be
in contact with the solution, but the circuits must be protected from liquid [63]. CMOS
chip should avoid being contaminated of alkali, copper or other ions; otherwise, those
ions will diffuse into silicon and cause drifting of threshold [70]. Processes in high
temperature and would cause CMOS material degeneration, eg., aluminum alloy. High
electric field will cause migration of electrons, accumulation and breakdown [71].
Passivation materials should be biocompatible and able to survive harsh chemical
environment. All materials should be compatible with microfabrication.
Electrochemical biosensor sensors, corresponding readout circuits, and sample
handling microfluidic devices have been realized with CMOS and CMOS-compatible
processes. However, a distinct lack of integration of these components remains as a major
obstacle toward miniaturization of the entire system. A few researchers have been
attempting to integrating biosensors directly on CMOS dies. The few known CMOSmicrofluidic integration approaches have been built upon wirebonded die in industry
75

standard packages. Hindered by bonding wires, incorporating complex microfluidic
structures has not been achieved. CMOS die level integration provides opportunities to
improve packing density [18]. The reliability of microfabricated planar interconnects on
pre-fabricated CMOS has been demonstrated [23]. So far, CMOS die level packaging
with microfluidic using microfabricated interconnects has not been reported.
This chapter describes a die level CMOS-microfluidic integration approach using
microfabricated interconnects routing signals from a CMOS die padframe to edge of a
carrier chip to create an extended planarized surface for complex microfluidic structures.
Key features of this new “lab-on-CMOS” process include complex microfluidic
structures and batch-fabrication capabilities. This thesis work reports the lab-on-CMOS
integration scheme and its application on enzyme-based biosensors. Major processes
include electroless plating, microfabricated planar wiring, silicon chip carrier micromachining, surface leveling, metallization and planarization. The CMOS potentiostat
reported in Chapter 4 is prepared and packed into a silicon carrier by those processes and
then integrated with a PDMS microfluidic mixer.
5.2

Packaging a CMOS Die for Microfluidics
Biosensor formation and test in ion containing liquid environment is detrimental for

CMOS. Ion solution is conductive, can cause CMOS short circuits if exposed. Some ions,
such as alkali and copper diffuse quickly in silicon will cause transistor threshold drifting,
and unrecoverable circuit malfunction. A general rule is that only the electrodes should
be exposed to the solution. The rest area of a CMOS die must be properly passivated to
block the circuit from contacting with liquid.
76

Another challenge is incorporating high density microfluidic structures with
CMOS. CMOS package should provide real estate that accommodates complex
microfluidic structures with micro channels running across the chip. A planar surface is
desired for leakage-free bonding and continuous fluid flow. Microfluidic channel
crossing the chip edge over to chip should be even and continuous to eliminate turbulent
flow. The packaged CMOS should survive harsh chemical environment during enzymebased biosensor formation, and the packaging process should not denature the enzyme.
CMOS-microfluidics integration works were done on top of a factory packed
chip. In this approach, a CMOS die is seated on bottom of the open cavity of an industrial
standard dual inline package (DIP) or quad flat package (QFP), with the CMOS pads
wirebonded to contacts around the perimeter of the cavity. The exposed metal bonding
wires, the bonding ball and the pads should be insulated. Obviously, the bonding wires
are obstacles for incorporating microfluidic structures. The bonding of microfluidic
structure to the chip should be enduring and leakage free. Researchers have been
unsuccessful in dealing with above issues. Other industrial standard IC packaging
technologies, such as chip size package (CSP) and ball grid array (BGA), use solder
bumps instead of bonding wires and have higher packing density. However, the solder
bumps elevate the chip surface level, and are still obstacles for microfluidic channels.
CMOS-microfluidic integration on those types of package have not been reported so far.
This research uses microfabricated planar interconnects and passivation to
provide planar surface for microfluidics for enzyme-based biosensor integration.
Microfluidic structures can be added on after enzyme immobilization. Key processes

77

such as leveling, planarization, low temperature PECVD oxide for permanent PDMS
bonding were used to resolve the challenges.
5.3

Lab-on-CMOS Integration Procedure
An even and continuous flat surface is the key to successful lab-on-CMOS

integration introduced in Section 1.1. High quality photolithography is fundamental for
microfabrication, requires a flat surface to dispense, expose and develop photoresist
evenly. A flat surface is essential to planar interconnection and microfluidic sealing and
continuity. A carrier is placed on a silicon wafer which serves as a flat bed. A CMOS die
is flipped, facing down to the flat bed, fitted into the cavity and fixed. The chip and
carrier assembly is released from the flat bed.
The lab-on-CMOS integration process flow is illustrated in Figure 5.1. After the
chip is prepared by gold plating, a silicon carrier is prepared using DRIE to create a
cavity in size of the chip, with some tolerances, as shown in Step 1. The second step is to
mount the chip and carrier on a wax covered glass wafer with their face down, followed
by a fixing step, Step 3, to fill the trench between chip and carrier with filling materials
such as epoxy and to cap the opening to improve mechanical strength. Then the CMOS
chip and silicon carrier assembly is released from the glass wafer, cleaned and coated
with a thin layer of polyimide around the gap between the CMOS chip and carrier, shown
in Step 4. In Step 5 and Step 6, the assembly is metalized with thin film gold and
passivated with thick polyimide and oxide to further planarize the surface. Step 7 opens
the polyimide above the electrodes, exposing the electrodes to fluid. Then the electrodes

78

are ready to be modified by enzyme. In Step 8, a PDMS optional microfluidic structure
can be added onto the assembly.

Figure 5.1: Lab-on-CMOS integration process flow.

5.3.1

CMOS Die Preparation
Aluminum pads and electrodes on the CMOS die were plated with gold over to

avoid corrosion due to hydroxide in common photoresist developer. Electroless gold
plating over CMOS aluminum has been proven to be successful [23, 72, 73]. In this
process, the CMOS die was cleaned to remove organic contaminants, then was etched
slightly with aluminum etchant to remove aluminum oxide and expose fresh aluminum.
79

An intermediate zincation step replaced a thin layer of aluminum with zinc to permit
subsequent deposition of a nickel film. After cleaning the nickel in dilute hydrochloride
acid (10%) briefly, gold plating was performed by dipping the CMOS chip into
immersion gold solution. Zinc, nickel and gold plating solutions were acquired from
Transene Company (Danvers, MA). To examine the plated gold, a pad was crosssectioned using focus ion beam. As shown in Figure 5.2, about 2µm plated metals fills
the contact opening without noticeable metallization on the overglass surrounding the
contact. To verify if the metals would survive chemicals used in photolithography, the
chip was then soaked in 5% sodium hydroxide for 30 minutes. No sign of attack on the
metals was observed under a microscope, indicating the chip pads are safe for
photolithography using hydroxide based developer.

Figure 5.2: A cross-section view of a CMOS pad after electroless plating of zinc,
nickel and gold.

80

5.3.2

Silicon Carrier Micromachining
A silicon chip carrier was used to host the CMOS chip and broad the surface to

match a microfluidic device. A silicon chip carrier s prepared by etching a cavity through
a silicon piece that would hold the CMOS chip. To form the cavity, potassium hydroxide
(KOH) wet etching and dry etching were evaluated. KOH wet etching through a (100)
silicon wafer creates sidewall slopes. Mask design for wet etching should compensate
undercuts and sidewall slopes. Double side polished wafer should be used to allow
photolithography on both side. Dimension control over the opening is challenging partly
due to undercuts and the inconsistent etch rate across the cavity caused by bubbletrapping especially when etching goes deep. The resulting wide opening on one side of
the silicon carrier facilitates chip placement and provides self-adjusting by gravity, but
the chip tends to lean and rotate in the cavity. Compared to KOH wet etching, deep
reactive ion etching (DRIE) process allows precise dimensional control with negligible
undercuts and thus is preferred. DRIE also creates round corners that help evenly
distribute the gaps between the die and the carrier. Based on experience, a cavity 10µm
lager than the CMOS die on each side is ideal. After DRIE, 1µm thermal oxide is grown
on the chip carrier to insulate the silicon substrate before subsequent metallization.
Silicon carriers can be produced in batch using a whole wafer. The described lab-onCMOS integration processes can also perform in wafer level.
A silicon chip carrier was prepared by etching a cavity through a 10mm square of
silicon wafer, where the size of the cavity is 10µm lager than the CMOS die on each side.
A DRIE process created straight side walls for precise dimension. DRIE also creates

81

rounded corners that allow leveling of the gaps between the die and the carrier. The etch
rate using STS Pegsus DRIE tool was 0.3024 µm/s.
Dry-wet-dry thermal oxidation was performed to generate a silicon oxide coating
over the carrier. The oxidation layer provided good electrical insulation in addition to
polyimide. It is easy for water-based polyimide to be evenly distributed on hydrophilic
silicon oxide instead of bare silicon.
5.3.3

Surface Leveling and Fixing
Surface leveling of the chip and carrier is critical for continuity of planar

interconnects and microfluidic channels. To achieve micron-scale leveling, a unique
process using wax was developed for the lab-on-CMOS platform. The chip carrier was
place with its face down onto an Apiezon wax (M&I Materials, Manchester, UK) coated
glass wafer. A CMOS chip was then placed faced down into the cavity of the carrier as
shown as illustrated in Step 2 of Figure 5.1. The chip carrier and the chip were initially
self-leveled by gravity. Before fixing the chip and carrier together, the wax was heated
under vacuum to 150 ºC to reflow and to expel air bubbles. The reflowed wax further
leveled the chip and carrier. The front of the chip and carrier were attached to wax at
cooling down.
The thickness of the wax coating is critical to the success of integration. In order
to control wax coating thickness, the Apiezon wax was diluted in xylene at 1:2, 1:4 and
1:8 and spun coated at 1000, 2000 and 3000rpm and. After the wax dried, its thickness
was characterized using surface profiler. Plots in Figure 5.3 give the wax thickness as a
function of spin speed and diluting ratio. In experiment trials, thinner coatings did not
82

provide adequate sealing; while thicker coating excessively extruded into the gap
between chip and carrier. 2µm Apiezon wax coating was found appropriate. It was
achieved by spin-on 1:2 Apiezon wax at 1000rpm.
After wax reflowing, epoxy was applied to fill the gap and to fix the CMOS chip
to the carrier as shown in Figure 5.1 (Step 3). A thin glass cap was attached to the
backside of the chip-carrier assembly to provide mechanical support to prevent carrier
breakage along the sides of the cavity during subsequent handling. Although several
fixing materials were explored, EPO-TEK 302-3M epoxy (Epoxy Technology, Billerica,
MA) was chosen because Apiezon wax has low solubility in the solvent of the epoxy, so
it does not mix the wax. Further, the expoxy resists to xylene and cholorobenzene used to
remove wax. Low viscosity, low shrinkage and chemical are other desirable properties
exhibited by this epoxy.
The chip-carrier assembly was released from the glass wafer by dissolving wax in
xylene, and then cleaned using xylene, chlorobenzene, isopropyl alcohol and deionized
water in sequence.
5.3.4

Planar Wiring
As shown in Fig. 5.3, because of wax flowing into the gap, a shallow trench was

formed unfilled by the fixing material in the front side of the chip-carrier assembly. The
depth of this trench is critical for successful electrical and fluidic interconnection across
the gap and was controlled by controlling the wax thickness discussed previously.

83

Figure 5.3: Spin-on Apiezon wax thickness as a function of spin speed and wax diluted in
on
xylenes at 1:8, 1:4 and 1:2.
Although the trench depth can be minimized by careful selection of wax conditions, it
cannot be eliminated. To fill this trench to create an even and continuous surface before
metalization, a polyimide layer was applied, with the added advantage of smoothing the
sharp edges of the silicon carrier and chip. PI 2554 polyimide precursor was obtained
from HD MicroSystems (Parlin, NJ). A polyimide precursor diluted by its thinner at 1:1
ms
was prepared, spun on the surface of the assembly and soft baked. S1813 photoresist was
spun on, soft baked, exposed and developed. The base developer also etches off
unmasked polyimide precursor. As shown in Step 4 of Fig. 5.1, the polyimide ring acts as
recursor.
,
a bridge over the trench providing an even and continuous surface for metallization.
Characterization of the bridge structure is discussed in the Results section below.
Planar interconnects leading the CMOS chip pads to finger contacts at the edges
interconnects
of the carrier were formed using liftoff thin film metals. Ti/Au (5nm/100nm) was
thin-film
deposited by thermal evaporation after photolithography. Tilted rotation during
evaporation improves metal coverage on the slopes of the trenches and pad wells. Gold
coverage
84

electrodes (for electrochemical sensors) on the CMOS chip were formed at the same time
as the interconnects, as shown in Step 5 of Figure 5.1.
5.3.5

Passivation and Planarization
Following metallization, a second polyimide layer was applied to insulate

interconnects and form the bottom plane of the microfluidic structure, as illustrated in
Step 6 of Figure 5.1. A thick polyimide layer was used to further planarize the surface.
The polyimide was patterned as described in Section III.B to expose the on-CMOS
electrodes and contact fingers at the edge of the carrier, as illustrated in Step 7 of Figure
5.1. A 500nm silicon dioxide layer was then deposited over polyimide by plasma
enhanced chemical vapor deposition (PECVD) at 200 ºC and patterned using
photolithography and HF wet etching. The oxide provides an adhesion layer for the
PDMS fluidic structure and insulates the polyimide from the fluid channel. For a higher
quality oxide, PECVD could be done at a higher temperature (~300 ºC); however, this
would cure the polyimide and require alternative patterning tools such as O2/CF4 plasma
dry etching or laser ablation.
5.3.6

Microfluidic Structure Integration
The final step in the lab-on-CMOS integration process was to attach the PDMS

microfluidic structure to the chip-carrier assembly, as illustrated in Step 8 of Figure 5.1.
The PDMS and the oxide surface were treated in oxygen plasma for enhanced bonding
[6]. PDMS was chosen for the microfluidic structure because it permits electrode
modification with sensitive materials, such as enzyme immobilization, before attachment
of the microfluidic structure.

85

A fluidic structure of four mixers was designed in a 6mm by 10mm footprint,
with the detection channels designed to run across the electrochemical sensors on the
CMOS chip. An SU 8 master was created by photolithography using a transparency file
photomask of the design. The channel height was 20µm, defined by the thickness of SU
8. Silane was vapor-primed on the master for easy peeling. Sylgard PDMS (Dow
Corning, Midland, MI) was mixed and cured. The structure was cut out and released for
use.
5.4

Result and Discussion

Figure 5.4: Surface profile from the carrier to the CMOS chip, measured before
metallization, passivation and planarization. Bottom picture is a SEM image of
interconnects over the bridge after metalization, to scale profilometer plot, confirming
the continuity of metal traces routing the chip pads to the carrier perimeter over the gap
via the bridge. Note: the horizontal axis scale is 10x the vertical scale. The zero
position on the vertical axis is the elevation of the carrier.

86

A previously reported CMOS potentiostat chip [8] was prepared by gold plating bonding
pads. A silicon chip carrier with a cavity matching the size of the potentiostat chip was
fabricated using an STS Pegasus DRIE tool. The chip-carrier assembly was constructed
using the integration process described by Figure 5.1.
The surface profile of the polyimide bridge between the CMOS chip and the carrier
was characterized by a Dektak 3 surface profilometer, and is plotted in Figure 5.4. The
plot shows shallow slopes were created by polyimide filling the gap and covering the
sharp silicon edges. Notice that the x-axis is 10x scale of the y-axis and that the
maximum slope along the surface is only 10º. The inset SEM image in Figure 5.4
confirms the metal traces were continuous over the trench. Using probe station, the
interconnect resistance was measured from pads to finger contact to be 8.4±0.4Ω, as
shown in Table 5.1. The current load capacity was over 500mA, sufficient for the CMOS
Measured

Wire

Net (Ω)

9.5

0.6

8.9

8.4

0.6

7.8

8.9

0.6

8.3

9

0.6

8.4

9.2

0.6

8.6

Average

8.4

Std

0.4

Table 5.1: Resistance from terminal to terminal (units: Ω).
potentiostat, as shown in Figure 5.5. Furthermore, the carrier and the CMOS chip were
properly leveled, leaving the chip-carrier height difference of only 1µm as shown in the
plot of Figure 5.4.
87

A 4-channel PDMS microfluidic mixer was constructed. The PDMS structure was
treated under oxygen plasma and attached to the surface of the chip-carrier assembly, as
shown in Figure 5.6. A toluidine blue-O solution was passed through the structure as
shown without leakage. Electrical and fluidic characterization indicates successful
integration of the lab-on-CMOS platform. The lab-on-CMOS device connected to a PCB
interface board that allows on-CMOS electrochemical sensors to be tested in a
microfluidic environment.

Current (mA)

800
600
400
200
0
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

Applied Voltage (Volts)

Figure 5.5: Current load test of planar wires.

88

Figure 5.6: Surface profile from the carrier to the CMOS chip, measured before
metallization, passivation and planarization. Inset is a SEM image of interconnects over
the bridge after metalization, to scale profilometer plot, confirming the continuity of
metal traces routing the chip pads to the carrier perimeter over the gap via the bridge.
Note: the horizontal axis scale is 10x the vertical scale. The zero position on the vertical
axis is the elevation of the carrier.

5.5

Conclusion
A new lab-on-CMOS platform for integrating microfluidics, on-CMOS electrode

arrays, and CMOS electrochemical instrumentation chip was introduced. The fabrication
process involving a CMOS die and a silicon chip carrier was described, and a prototype
assembly was demonstrated. Electrical and fluidic characterizations verify successful
integration of the lab-on-CMOS platform.

89

6
6.1

Summary, Contributions and Future Work

Summary
The mysteries of the vast unexplored world of biology have been motivating

scientists to leverage micro/nanotechnology using lab-on-chip devices and microsystems.
The power of microelectronics has not been adequately explored to enhance the signaling
between biology and silicon. This thesis work introduces a lab-on-CMOS platform
concept that enables integration of biosensors directly on top of a CMOS chip to establish
continuous bio-to-silicon interaction pathways.
This research developed methodologies of optimizing the integration of redox
enzyme-based electrochemical biosensor, microfabricated electrode and CMOS
instrumentation with breakthroughs in those areas. The molecular self-assembly enzyme
immobilization method was evaluated, leading to the development of a fabricationfavored method combining conductive polymers and molecular self-assembly. An
-

miniaturized CMOS-compatible Ag/AgCl/PPy(Cl )/Nafion quasi-reference electrode was
developed. A parallel bi-potentiostat architecture was developed replacing the traditional
adder-potentiostat architecture to support redox recycling. An example fructose sensor
was developed and the chronoamperometry technique was implemented through the
CMOS bipotentiostat to extract analyte concentration without needs of further data
processing. Finally, a unique CMOS die packaging scheme featuring a silicon chip carrier
and microfabricated planar interconnections was developed to incorporate a microfluidic
structure. The individual areas of contribution are summarized as below.

90

6.2

List of Contributions
a. Introduced a methodology for integration of redox-enzyme-based biosensor
arrays with a CMOS instrumentation chip
This is the first known research into integration of an enzyme-based biosensor

onto a CMOS electrochemical instrument chip using CMOS compatible electrode
system. In order to minimize the footprint of an Oxidoreductase enzyme-based biosensor
to fit on CMOS, redox recycling, a signal amplification mechanism was utilized to
increase sensitivity, and a concentric ring electrode on CMOS chip surface. A CMOS
bipotentiostat was developed for placement directly underneath the concentric ring
electrode, enabling measurement using the chronoamperometry technique to extract the
analyte concentration directly.
b. Developed a novel CP-MSA technique to immobilize enzyme on aluminum
A

conductive

polymer

molecular

self-assembly

(CP-MSA)

enzyme

immobilization method was developed as an alternative to the MSA method for direct
immobilization on CMOS aluminum without post-CMOS photolithography. With CPMSA, the enzyme can be selectively immobilized on multiple working electrodes and
does not foul the counter electrode like the alternative MSA process does. This method
eliminates the dependence on gold and saves at least 50% processing time compared to
MSA.
c. Developed electrochemically optimized, CMOS compatible, planar electrode
systems and fabrication processes

91

Planar electrode systems were designed and fabricated to simultaneously establish
optimized for 1) post-CMOS microfabrication, 2) biointerface quality and 3)
electrochemical performance. For the first time, redox recycling on a concentric ring
electrode was introduced to increase sensitivity and to reduce footprint of a redoxenzyme-based biosensor so that it can be implemented on CMOS. A CMOS-compatible
-

Ag/AgCl/PPy(Cl )/Nafion quasi-reference electrode process was developed to enable
down scaling and placement within a short distance from the working electrode, uniquely
improving on-chip electrode density.
d. Introduced a new bi-potentiostat architecture for effectively implementing onto
CMOS
A new bipotentiostat architecture was designed to support enzyme redox
recycling through interdigitated electrodes. By replacing traditional adder-potentiostat
architecture, the new architecture reduces circuit complexity and thus silicon area by 44%
for each additional channel. The new architecture also completely avoids propagated
potential control errors.
e. Introduced a novel lab-on-CMOS package platform that enables high
performance CMOS biosensor arrays to utilize complex microfluidics
A lab-on-CMOS platform concept was introduced that provides an expanded area
beyond the CMOS chip surface to allow integration of complex microfluidics. Through
an innovative leveling process, a surface height difference between the CMOS chip and a
silicon carrier of only 1µm was achieved. Microfabricated planar interconnection was

92

shown to successfully replace traditional bonding wires or solder bumps. Integrity of the
microfluidic structures was verified through experimental measurement.
6.3

Future Work
This research has established the foundation for a lab-on-CMOS platform. Future

full integration of enzyme-based biosensors on a CMOS bi-potentiostat would further
demonstrate the power of combining microelectronics, microfabrication and biosensor
technologies. To achieve this goal, the following works are crucial:
a. Enzyme biosensors on concentric-ring electrode
Using the mathematic model developed by this research, finite element analysis of
concentric ring electrode can help predict current generated by the sensor. Electroless
gold plating on the concentric ring electrode is required for using MSA method.
Immobilization using CP-MSA method needs improvement on sensitivity and validation
on CMOS aluminum concentric ring electrodes.
b. A self-contained CMOS electrochemical instrumentation
This research uses some off-chip auxiliary circuits and a DAQ. A self-contained
CMOS bipotentiostat with digital control logic to interface with a wireless device will
greatly enhance the optimization of biosensors, electrochemical instrumentation and
electrochemical techniques as a system.
c. Characterization of on-chip planar electrodes in microfluidic channels

93

Off-chip biosensors were used for characterizing CMOS bipotentiostat in this
research. On-chip biosensors using the concentric ring electrode can be characterized
show its advantages over a disk electrode of the same area. Once packed in microfluidic
channels, the performance of the on-chip electrodes can be characterized, with
-

commercial potentiostat and then with the CMOS chip. An Ag/AgCl/PPy(Cl )/Nafion
quasi-reference electrode can be fabricated on a CMOS chip.
d. Microfluidic integration
A suitable microfluidic structure such as mixing and separation could be designed
and characterized to demonstrate the sample handling capability provided by the die-level
lab-on-CMOS packaging scheme. More sophisticated passive or active microfluidic
devices could be built on the chip carrier using silicon micro machining.

94

APPENDIX

95

Modeling of Redox Recycling on Concentric Ring Electrode
To characterize the concentric ring electrode, we consider that it is in a solution of
speciesO. One working electrode, defined as generator electrode, is initially set at a
potential where O is not reduced. The other electrode, defined as regenerator, is initially
set to a potential where the reduction product ܴ is oxidized. A step waveform is applied
at t = 0, so that O is to reduced to ܴ at the generator electrode; at the same time, another
step waveform is applied on the regenerator electrode so that a portion of the reduced ܴ
transported onto generator can be oxidized to O. The concentration of O is increased by
the regenerated O locally on the regenerator electrode. As a consequence, the regenerated
O that diffused onto generator electrode increases the reduction current.
The diffusion of a planar electrode occurs in dimensions both normal to the
electrode plane and radical with respect to the symmetrical axis. Consider the generator
electrode in a cylindrical coordinate system, the diffusion equation is written as follows:

1 ߲‫ܥ‬௢ (‫ ߲ )ݐ ,ݖ ,ݎ‬ଶ ‫ܥ‬௢ (‫ܥ߲ 1 )ݐ ,ݖ ,ݎ‬௢ (‫ ߲ )ݐ ,ݖ ,ݎ‬ଶ ‫ܥ‬௢ (‫)ݐ ,ݖ ,ݎ‬
=
+
+
‫ܦ‬௢
߲‫ݐ‬
߲‫ ݎ‬ଶ
‫ݎ‬
߲‫ݎ‬
߲‫ ݖ‬ଶ
where r describes radial position normal to the axis of symmetry at r = 0, and ‫ ݖ‬describes
linear displacement normal to the plane of the electrode at ‫ ,0 = ݖ‬with D0 for diffusion
coefficient, C0 for concentration at the electrode surface andt for time. Assuming the
initial solution is homogenous, then concentration of O equals to the bulk concentration
∗
‫ܥ‬௢ all over the solution, including the electrode surface.

96

∗
‫ܥ‬௢ ሺ‫0 ,ݖ ,ݎ‬ሻ = ‫ܥ‬௢

After the perturbation, the bulk concentration deep into the solution remains
unchanged both radically and normal to the electrode plane.

∗
lim ‫ܥ‬௢ ሺ‫ݐ ,ݖ ,ݎ‬ሻ = ‫ܥ‬௢

௥→ஶ

∗
lim ‫ܥ‬௢ ሺ‫ݐ ,ݖ ,ݎ‬ሻ = ‫ܥ‬௢

௭→ஶ

If the kth ring electrode is a generator electrode, then the adjacent (k-1)th and
(k+1) th ring electrode is a generator electrode. In between the rings, there is no flux of

‫ܥ‬௢ into or out of the mantle, since there is no reaction in the gap.

߲‫ܥ‬௢ ሺ‫ݐ ,ݖ ,ݎ‬ሻ
=0
߲‫ݖ‬
( ‫ݎ‬௞ + ݀ൗ2 ≤ ‫ݎ ≤ ݎ‬௞ାଵ − ݀ൗ2 ܽ݊݀ ‫ݎ‬௞ିଵ + ݀ൗ2 ≤ ‫ݎ ≤ ݎ‬௞ − ݀ൗ2)
where ‫ݎ‬௞ is the central radius of the kth ring electrode, d is the width of the ring. The ring
radius of the outer ring is larger than the inner ring by ݀, ‫ݎ‬௞ାଵ > ‫ݎ‬௞ + ݀ , making sure
the rings are not overlapped. Above conditions applies to the situation where the solution
is homogenous before the perturbations and where the electrolyte extends beyond the

97

limit of the diffusion layer. After the experiment begins, we assume the concentration of
ܱ is driven to zero at the electrode surface:

‫ܥ‬௢ ሺ‫ݐ ,0 ,ݎ‬ሻ = 0,

(‫ݎ‬௞ − ݀ൗ2 ≤ ‫ݎ ≤ ݎ‬௞ + ݀ൗ2 , ‫)0 > ݐ‬

Similarly, the diffusion equation of the regenerator electrode is:

1 ߲‫ܥ‬ோ (‫ ߲ )ݐ ,ݖ ,ݎ‬ଶ ‫ܥ‬ோ (‫ܥ߲ 1 )ݐ ,ݖ ,ݎ‬ோ (‫ ߲ )ݐ ,ݖ ,ݎ‬ଶ ‫ܥ‬ோ (‫)ݐ ,ݖ ,ݎ‬
=
+
+
‫ܦ‬ோ
߲‫ݐ‬
߲‫ ݎ‬ଶ
‫ݎ‬
߲‫ݎ‬
߲‫ ݖ‬ଶ
Initial conditions of the regenerator electrodes are considered as follows. There is
no ܴ in presence at the beginning of the process all over the solution. The concentration
of ܴ equals to zero on the regenerator electrode surface as well as in the bulk solution.

‫ܥ‬ோ ሺ‫0 ,ݖ ,ݎ‬ሻ = 0
After the perturbation begins, the bulk concentration of ܴ deep into the solution
remains zero both radically and normal to the electrode plane.

lim ‫ܥ‬ோ ሺ‫ݐ ,ݖ ,ݎ‬ሻ = 0

௥→ஶ

lim ‫ܥ‬ோ ሺ‫ݐ ,ݖ ,ݎ‬ሻ = 0

௭→ஶ

98

As the nearby generator electrode, there is no flux of ‫ܥ‬ோ into or out of the mantle
in the gap between the (k-1)th regenerator and the kth or (k-2)th generator electrode.

߲‫ܥ‬ோ ሺ‫ݐ ,ݖ ,ݎ‬ሻ
=0
߲‫ݖ‬
(‫ݎ‬௞ିଶ + ݀ൗ2 ≤ ‫ݎ ≤ ݎ‬௞ିଵ − ݀ൗ2 ܽ݊݀ ‫ݎ‬௞ିଵ + ݀ൗ2 ≤ ‫ݎ‬
≤ ‫ݎ‬௞ − ݀ൗ2)
where ‫ݎ‬௞ିଵ is the radius at the center of the kth-1 ring, a regenerator electrode, d is the
width of the ring. After the experiment begins, we assume the concentration of ܴ is
driven to zero at the regenerator electrode surface:

‫ܥ‬ோ ሺ‫ݐ ,0 ,ݎ‬ሻ = 0,

(‫ݎ‬௞ିଵ − ݀ൗ2 ≤ ‫ݎ ≤ ݎ‬௞ିଵ + ݀ൗ2 , ‫)0 > ݐ‬

After the experiment begins, ܱ is reduced to ܴ at the kth ring generator electrode
and transports to the adjacent (k-1)th and (k+1)th regenerator electrodes. The reduction
∗
reaction starts with ‫ܥ‬௢ , after the diffusion layer formed, is limited by mass transport. As

a result, a wave of ܴ is produced and then followed by a steady replenishing. If ܱ is fully
converted to ܴ without any delay, the product concentration ‫ܥ‬ோ approximately traces the

99

concentration of ܱ right next to the electrode surface ‫ܥ ,)0 → ߜ( ߜ = ݖ‬ை (‫ .)ݐ ,ߜ ,ݎ‬This

‫ܥ‬ோ adds to the boundary conditions of the diffusion equation of the regenerator electrode.

‫ܥ‬ோ ሺ‫ݐ ,0 ,ݎ‬ሻ ≅ ‫ܥ‬ை ሺ‫ݐ ,ߜ ,ݎ‬ሻ
(‫ݎ‬௞ − ݀ൗ2 ≤ ‫ݎ ≤ ݎ‬௞ + ݀ൗ2 , ‫)0 → ߜ ,0 > ݐ‬

The generated ܴ transports from kth ring generator electrode to the adjacent (k1)th and (k+1)th regenerator electrodes. Considering only diffusion portion of the mass
transport, after a time delay‫ݐ‬ௗ , it is re-oxidized on the regenerator electrodes.

‫ݐ‬ௗ = (‫ݎ‬௞ − ‫ݎ‬௞ିଵ )ଶ /(2‫ܦ‬ோ )

The re-oxidized ܱ carries the same temporal and spacial concentration profile as
‫݃݁ݎ‬

the ܴ produced on the generator electrode, noted ‫ܥ‬ை . On the regenerator electrode, the
௥௘௚

total concentration of ܱ is ‫ܥ‬ை

adding to the existing ܱ in situ.

௥௘௚
்௢௧௔௟ ሺ
‫ܥ‬ை
‫ݐ ,0 ,ݎ‬ሻ = ‫ܥ‬ை ሺ‫ݐ ,݋ ,ݎ‬ሻ + ‫ܥ‬ை ሺ‫ݐ ,݋ ,ݎ‬ሻ

(‫ݎ‬௞ିଵ − ݀ൗ2 ≤ ‫ݎ ≤ ݎ‬௞ିଵ + ݀ൗ2 , ‫)0 → ߜ ,0 > ݐ‬

100

்௢௧௔௟
The increased ‫ܥ‬ை
updates the boundary condition of the diffusion equation of

the generator electrodes, increasing the reduction current in a dynamic fashion due to the
interaction of the generation and regeneration. The boundary condition of the outmost
ring and innermost ring is a special case of the inner rings.
Due to the complexity of the problem, the analytical solution of the partial
differential equation is not tempted. However, it is feasible to model the concentric ring
electrode using finite element analysis program, and to derive the current-time
relationship graphically by applying the boundary conditions discussed above.

101

REFERENCES

102

REFERENCES
[1]

K. Rogers, "Biosensors for environmental monitoring: a regulatory
perspective," TrAC Trends in Analytical Chemistry, vol. 14, pp. 289-294, 1995.

[2]

M. Keusgen, "Biosensors: new approaches
Naturwissenschaften, vol. 89, pp. 433-444, 2002.

[3]

J. Wang, "Amperometric biosensors for clinical and therapeutic drug
monitoring: a review," Journal of Pharmaceutical and Biomedical Analysis, vol.
19, pp. 47-53, 1999.

[4]

I. Moser, G. Jobst, and G. A. Urban, "Biosensor arrays for simultaneous
measurement of glucose, lactate, glutamate, and glutamine," Biosensors and
Bioelectronics, vol. 17, pp. 297-302, 2002.

[5]

S. P. Mohanty and E. Kougianos, "Biosensors: a tutorial review," Potentials,
IEEE, vol. 25, pp. 35-40, 2006.

[6]

L. Clark and C. Lyons, "Electrode Systems for Continuous Monitoring in
Cardiovascular Surgery," Annals of the New York Academy of Sciences, vol.
102, pp. 29-45, 1962.

[7]

E.-H. Yoo and S.-Y. Lee, "Glucose Biosensors: An Overview of Use in Clinical
Practice," Sensors, vol. 10, pp. 4558-4576, 2010.

[8]

W. Schuhmann, C. Kranz, J. Huber, and H. Wohlschlager, "Conducting
polymer-based amperometric enzyme electrodes. Towards the development
of miniaturized reagentless biosensors," Synthetic Metals, vol. 61, pp. 31-35,
1993.

[9]

M. Zayats, E. Katz, and I. Willner, "Electrical contacting of flavoenzymes and
NAD(P)(+)-dependent enzymes by reconstitution and affinity interactions on
phenylboronic acid monolayers associated with Au-electrodes," Journal of the
American Chemical Society, vol. 124, pp. 14724-14735, 2002.

[10]

B. L. Hassler, N. Kohli, J. G. Zeikus, I. Lee, and R. M. Worden, "Renewable
dehydrogenase-based interfaces for bioelectronic applications," Langmuir,
vol. 23, pp. 7127-7133, 2007.

[11]

N. Ravindra, C. Prodan, S. Fnu, I. Padronl, and S. Sikha, "Advances in the
manufacturing, types, and applications of biosensors," JOM Journal of the
Minerals, Metals and Materials Society, vol. 59, pp. 37-43, 2007.

[12]

M. Koudelka-Hep, "Microelectrode sensors for biomedical and environmental
applications," Electrochimica Acta, vol. 45, pp. 2437-2441, 2000.
103

in

drug

discovery,"

[13]

R. M. Wightman, "Voltammetry with Microscopic Electrodes in New
Domains," Science, vol. 240, pp. 415-420, 1988.

[14]

Y. Huang, A. Mason, A. J. Greiner, R. Y. Ofoli, and R. M. Worden, "Post-CMOS
Compatible Microfabrication of a Multi-Analyte Bioelectrochemical Sensor
Array Microsystem," presented at 5th IEEE Int. Sensors Conf., Dageu, Korea,
2006.

[15]

A. C. R. Grayson, R. S. Shawgo, A. M. Johnson, N. T. Flynn, Y. Li, M. J. Cima, and
R. Langer, "A BioMEMS Review: MEMS Technology for Physiologically
Integrated Devices," Proceedings of the IEEE, vol. 92, pp. 6-21, 2004.

[16]

A. Frey, M. Schienle, C. Paulus, Z. Jun, F. Hofmann, P. Schindler-Bauer, B.
Holzapfl, M. Atzesberger, G. Beer, M. Fritz, T. Haneder, H. C. Hanke, and R.
Thewes, "A digital CMOS DNA chip," presented at Circuits and Systems, 2005.
ISCAS 2005. IEEE International Symposium on, 2005.

[17]

P. Levine, P. Gong, K. Shepard, and R. Levicky, "Active CMOS Array for
Electrochemical Sensing of Biomolecules," San Jose, CA, USA, 2007.

[18]

A. Manickam, A. Chevalier, M. McDermott, A. Ellington, and A. Hassibi, "A
CMOS electrochemical impedance spectroscopy biosensor array for labelfree biomolecular detection," presented at ISSCC, San Francisco, CA, USA,
2010.

[19]

A. Hassibi and T. H. Lee, "A programmable electrochemical biosensor array in
0.18 µm standard CMOS," presented at Solid-State Circuits Conference, 2005.
Digest of Technical Papers. ISSCC. 2005 IEEE International, San Francisco, CA,
USA, 2005.

[20]

P. Kruppa, A. Frey, I. Kuehne, M. Schienle, N. Persike, T. Kratzmueller, G.
Hartwich, and D. Schmitt-Landsiedel, "A digital CMOS-based 24x16 sensor
array platform for fully automatic electrochemical DNA detection,"
Biosensors and Bioelectronics, vol. 26, pp. 1414-1419, 2010.

[21]

R. Reay, "Microfabricated electrochemical analysis system for heavy metal
detection," Sensors and Actuators B: Chemical, vol. 34, pp. 450-455, 1996.

[22]

D. Grieshaber, R. MacKenzie, J. V 枚 r 枚 s, and E. Reimhult, "Electrochemical
Biosensors - Sensor Principles and Architectures," Sensors, vol. 8, pp. 14001458, 2008.

[23]

Y. Wang, H. Xu, J. Zhang, and G. Li, "Electrochemical Sensors for Clinic
Analysis," Sensors, vol. 8, pp. 2043-2081, 2008.

[24]

N. Ronkainen, B. Halsall, and W. Heineman, "Electrochemical biosensors,"
Chem. Soc. Rev., vol. 39, pp. 1747-1763, 2010.
104

[25]

Y. Wu and S. Hu, "Biosensors based on direct electron transfer in redox
proteins," Microchimica Acta, vol. 159, pp. 1-17, 2007.

[26]

A. Bard and L. Faulkner, Electrochemical Methods: Fundamentals and
Applications. New York: John Wiley & Sons, 2001.

[27]

W. Shinwari, D. Zhitomirsky, I. Deen, P. R. Selvaganapathy, J. Deen, and D.
Landheer, "Microfabricated Reference Electrodes and their Biosensing
Applications," Sensors, vol. 10, pp. 1679-1715, 2010.

[28]

I. Rubinstein, Physical Electrochemistry: Principles, Methods and Applications.
New York: Marcel Dekker, 1995.

[29]

S. Cosnier, "Biomolecule immobilization on electrode surfaces by entrapment
or attachment to electrochemically polymerized films. A review," Biosensors
and Bioelectronics, vol. 14, pp. 443-456, 1999.

[30]

W. Schuhmann, "Conducting polymer based amperometric enzyme
electrodes," Microchimica Acta, vol. 121, pp. 1-29, 1995.

[31]

F. Kuechenmeister and E. Meusel, "Polypyrrole as an interlayer for bonding
conductive adhesives to activated aluminum bond pads," IEEE Transactions
on Components, Packaging, and Manufacturing Technology: Part A, vol. 20, pp.
9-14, 1997.

[32]

G. G. Wallace, M. Smyth, and H. Zhao, "Conducting electroactive polymerbased biosensors," TrAC Trends in Analytical Chemistry, vol. 18, pp. 245-251,
1999.

[33]

M. Gerard, A. Chaubey, and B. D. Malhotra, "Application of conducting
polymers to biosensors," Biosensors and Bioelectronics, vol. 17, pp. 345-359,
2002.

[34]

T. Ahuja, I. A. Mir, D. Kumar, and Rajesh, "Biomolecular immobilization on
conducting polymers for biosensing applications," Biomaterials, vol. 28, pp.
791-805, 2007.

[35]

C. Kranz, H. Wohlschlager, H.-L. Schmidt, and W. Schuhmann, "Controlled
Electrochemical Preparation of Amperometric Biosensors Based on
Conducting Polymer Multilayers," Electroanalysis, vol. 10, pp. 546-552, 1998.

[36]

S. Cosnier, B. Galland, C. Gondran, and A. Le Pellec, "Electrogeneration of
Biotinylated Functionalized Polypyrroles for the Simple Immobilization of
Enzymes," Electroanalysis, vol. 10, pp. 808-813, 1998.

[37]

W. Schuhmann, R. Lammert, B. Uhe, and H. L. Schmidt, "Polypyrrole, a new
possibility for covalent binding of oxidoreductases to electrode surfaces as a
105

base for stable biosensors," Sensors and Actuators B: Chemical, vol. 1, pp. 537541, 1990.
[38]

B. Thompson, O. Winther-Jensen, J. Vongsvivut, B. Winther-Jensen, and D.
MacFarlane, "Conducting Polymer Enzyme Alloys: Electromaterials
Exhibiting Direct Electron Transfer," Macromol. Rapid Commun., vol. 31, pp.
1293-1297.

[39]

J. Rick and T.-C. Chou, "Using protein templates to direct the formation of
thin-film polymer surfaces," Biosensors and Bioelectronics, vol. 22, pp. 544549, 2006.

[40]

J. Martins, S. Costa, M. Bazzaoui, G. Goncalves, E. Fortunato, and R. Martins,
"Electrodeposition of polypyrrole on aluminium in aqueous tartaric
solution," Electrochimica Acta, vol. 51, pp. 5802-5810, 2006.

[41]

M. Bazzaoui, J. Martins, T. Reis, E. Bazzaoui, M. Nunes, and L. Martins,
"Electrochemical synthesis of polypyrrole on ferrous and non-ferrous metals
from sweet aqueous electrolytic medium," Thin Solid Films, vol. 485, pp. 155159, 2005.

[42]

J. Voldman, M. Gray, and M. Schmidt, "Microfabrication in Biology and
Medicine," Annual Review of Biomedical Engineering, vol. 1, pp. 401-425,
1999.

[43]

H. Suzuki, "Advances in the Microfabrication of Electrochemical Sensors and
Systems," Electroanalysis, vol. 12, pp. 703-715, 2000.

[44]

S. Hwang, C. LaFratta, V. Agarwal, X. Yu, D. Walt, and S. Sonkusale, "CMOS
Microelectrode Array for Electrochemical Lab-on-a-Chip Applications,"
Sensors Journal, IEEE, vol. 9, pp. 609-615, 2009.

[45]

B. Polk, A. Stelzenmuller, G. Mijares, W. MacCrehan, and M. Gaitan, "Ag/AgCl
microelectrodes with improved stability for microfluidics," Sensors and
Actuators B: Chemical, vol. 114, pp. 239-247, 2006.

[46]

S.-I. Park, S. Jun, S. Park, H. Kim, and S. Kim, "Application of a new Cl-plasmatreated Ag/AgCl reference electrode to micromachined glucose sensor," IEEE
Sensors Journal, vol. 3, pp. 267-273, 2003.

[47]

G. Valdesramirez, G. Alvarezromero, C. Galanvidal, P. Hernandezrodriguez,
and M. Ramirezsilva, "Composites: A novel alternative to construct solid state
Ag/AgCl reference electrodes," Sensors and Actuators B: Chemical, vol. 110,
pp. 264-270, 2005.

[48]

D. J. G. Ives and G. J. Janz, Reference Electrodes, Theory and Practice (1st
edition): Academic Press, 1961.
106

[49]

M. Kimura, H. Fukushima, Y. Sagawa, K. Setsu, H. Hara, and S. Inoue, "An
Integrated Potentiostat With an Electrochemical Cell Using Thin-Film
Transistors," IEEE Transactions on Electron Devices, vol. 56, pp. 2114-2119,
2009.

[50]

J. Ghilane, P. Hapiot, and A. J. Bard, "Metal/Polypyrrole Quasi-Reference
Electrode for Voltammetry in Nonaqueous and Aqueous Solutions," Anal.
Chem., vol. 78, pp. 6868-6872, 2006.

[51]

M. Nolan, S. Tan, and S. Kounaves, "Fabrication and Characterization of a
Solid State Reference Electrode for Electroanalysis of Natural Waters with
Ultramicroelectrodes," Analytical Chemistry, vol. 69, pp. 1244-1247, 1997.

[52]

L. Tymecki, E. Zwierkowska, and R. Koncki, "Screen-printed reference
electrodes for potentiometric measurements," Analytica Chimica Acta, vol.
526, pp. 3-11, 2004.

[53]

H. Suzuki, A. Hiratsuka, S. Sasaki, and I. Karube, "Problems associated with
the thin-film Ag/AgCl reference electrode and a novel structure with
improved durability," Sensors and Actuators B: Chemical, vol. 46, pp. 104-113,
1998.

[54]

S. Ayers, K. Gillis, M. Lindau, and B. Minch, "Design of a CMOS Potentiostat
Circuit for Electrochemical Detector Arrays," IEEE transactions on circuits
and systems. I, Regular papers : a publication of the IEEE Circuits and Systems
Society, vol. 54, pp. 736-744, 2007.

[55]

F. Yalcinkaya, "Ag/AgCl/Cl− coated silver-stripe reference electrode," Medical
Engineering & Physics, vol. 19, pp. 299-301, 1997.

[56]

R. Zielinska, "All-solid-state planar miniature ion-selective chloride
electrode," Analytica Chimica Acta, vol. 451, pp. 243-249, 2002.

[57]

W.-Y. Liao and T.-C. Chou, "Fabrication of a Planar-Form Screen-Printed Solid
Electrolyte Modified Ag/AgCl Reference Electrode for Application in a
Potentiometric Biosensor," Analytical Chemistry, vol. 78, pp. 4219-4223,
2006.

[58]

K.-S. Yun, H.-J. Kim, S. Joo, J. Kwak, and E. Yoon, "Analysis of heavy-metal-ions
using mercury microelectrodes and a solid-state reference electrode
fabricated on a Si wafer," Tokyo, Japan, 2000.

[59]

K. Nagy, K. Eine, K. Syverud, and O. Aune, "Promising New Solid-State
Reference Electrode," Journal of The Electrochemical Society, vol. 144, pp. L1L2, 1997.

107

[60]

F. Moussy and J. Harrison, "Prevention of the Rapid Degradation of
Subcutaneously Implanted Ag/AgCl Reference Electrodes Using Polymer
Coatings," Analytical Chemistry, vol. 66, pp. 674-679, 1994.

[61]

N.-H. Kwon, K.-S. Lee, M.-S. Won, and Y.-B. Shim, "An all-solid-state reference
electrode based on the layer-by-layer polymer coating," Analyst, vol. 132, pp.
906-912, 2007.

[62]

A. Gore, S. Chakrabartty, S. Pal, and E. C. Alocilja, "A Multichannel
Femtoampere-Sensitivity Potentiostat Array for Biosensing Applications,"
Circuits and Systems I: Regular Papers, IEEE Transactions on [Circuits and
Systems I: Fundamental Theory and Applications, IEEE Transactions on], vol.
53, pp. 2357-2363, 2006.

[63]

S. M. Martin, F. H. Gebara, T. D. Strong, and R. B. Brown, "A low-voltage,
chemical sensor interface for systems-on-chip: the fully-differential
potentiostat," presented at Circuits and Systems, 2004. ISCAS '04.
Proceedings of the 2004 International Symposium on, 2004.

[64]

X. Zhu and C. H. Ahn, "On-Chip Electrochemical Analysis System Using
Nanoelectrodes and Bioelectronic CMOS Chip," Sensors Journal, IEEE, vol. 6,
pp. 1280-1286, 2006.

[65]

K. Murari, M. Stanacevic, G. Cauwenberghs, and N. V. Thakor, "Integrated
potentiostat for neurotransmitter sensing," IEEE Engineering in Medicine and
Biology Magazine, vol. 24, pp. 23-29, 2005.

[66]

W.-S. Wang, W.-T. Kuo, H.-Y. Huang, and C.-H. Luo, "Wide Dynamic Range
CMOS Potentiostat for Amperometric Chemical Sensor," Sensors, vol. 10, pp.
1782-1797, 2010.

[67]

L. Busoni, M. Carlà, and L. Lanzi, "A comparison between potentiostatic
circuits with grounded work or auxiliary electrode," Review of Scientific
Instruments, vol. 73, pp. 1921, 2002.

[68]

O. Niwa and M. Morita, "Carbon Film-Based Interdigitated Ring Array
Electrodes as Detectors in Radial Flow Cells," Analytical Chemistry, vol. 68,
pp. 355-359, 1996.

[69]

D. Farina and C. Cescon, "Concentric-ring electrode systems for noninvasive
detection of single motor unit activity," IEEE Transactions on Biomedical
Engineering, vol. 48, pp. 1326-1334, 2001.

[70]

S. Sze, Semiconductor Devices: Physics and Technology, 2nd Edition. New York:
John Wiley & Sons, 2002.

[71]

R. F. Pierret, Semiconductor Device Fundamentals: Addison-Wiley, 1995.
108

[72]

K. C. Tang, E. Liao, W. L. Ong, J. D. S. Wong, A. Agarwal, R. Nagarajan, and L.
Yobas, "Evaluation of bonding between oxygen plasma treated polydimethyl
siloxane and passivated silicon," Journal of Physics: Conference Series, vol. 34,
pp. 155, 2006.

[73]

A. Strandjord, S. Popelar, and C. Jauernig, "Interconnecting to aluminum- and
copper-based semiconductors (electroless-nickel/gold for solder bumping
and wire bonding)," Microelectronics Reliability, vol. 42, pp. 265-283, 2002.

109