Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Biosensing and biomimetic electronics
(USC Thesis Other)
Biosensing and biomimetic electronics
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
I
BIOSENSING AND BIOMIMETIC ELECTRONICS
By
Qingzhou Liu
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(MATERIALS SCIENCE)
August 2019
Copyright 2019 Qingzhou Liu
II
Acknowledgement
First of all, I would like to express my sincere gratitude to my advisor, Dr.
Chongwu Zhou, for offering me such a great opportunity to pursue my interest
in nanoelectronics. Professor Zhou’s visions, advises, and resources have
broadened my view and made many difficult goals achievable. The capability
of problem solving and knowledge accumulation that I’ve gained under his
guidance will always be invaluable to me. Without his academic and financial
support, this dissertation would not be possible.
I would like to thank my dissertation committee, Professor Steve Nutt and
Professor Aiichiro Nakano for helping me with the dissertation defense
process. I am also grateful to Prof. Wei Wu and Prof. Jongseung Yoon for
serving as the committee members of my qualifying exam.
I would also like to thank Prof. Mark Thompson, Porf. Paul Weiss, Prof. Anne
Andrews, Prof. Jeffery Draper, Prof. Moh Amer, Prof. Ivan Sanchez Esqueda,
Prof. Ram Datar and Prof. Richard Cote for their collaboration.
I would like to thank Dr. Noppadol Aroonyadet, Dr. Xiaoli Wang, Dr. Yan
Song, and Dr. Xuan Cao for passing down their experience and technique
which makes my research life easier.
III
In addition, I would like to thank my colleagues: Dr. Bilu Liu, Dr. Yihang Liu,
Chuanzhen Zhao, Dr. Haitian Chen, Dr. Xin Fang, Dr. Jiepeng Rong, Dr.
Mingyuan Ge, Dr. Luyao Zhang, Dr. Hui Gui, Dr. Liang Chen, Fanqi Wu, Dr.
Yuqiang Ma, Dr. Anyi Zhang, Christian Lau, Dr. Lang Shen, Haotian Shi, Dr.
Sen Cong, Dr. Yu Cao, Dr. Chenfei Shen, and Jenny Lin, for all of your help
and goodwill you have shown me.
Finally, I would like to thank my family for the unconditional support, love,
and encouragement of my Ph.D study, emotionally helping me get out of the
dilemmas I met over these years.
IV
Table of contents
Acknowledgement .......................................................................................................................... II
List of figures ................................................................................................................................ VI
Abstract ...................................................................................................................................... XIII
Chapter 1 Introduction to biosensing and biomimetic electronics ............................................... 1
1.1 Overview .......................................................................................................................... 1
1.2 Mechanism of biosensing electronics using field-effect transistors .............................. 2
1.3 Mechanism of biomimetic synapse using organic electronics ....................................... 4
1.4 References ........................................................................................................................ 5
Chapter 2 Highly sensitive and quick detection of acute myocardial infarction biomarkers
using In 2O 3 nanoribbon biosensors fabricated using shadow masks ........................................... 7
2.1 Introduction ..................................................................................................................... 7
2.2 Indium oxide nanoribbon field-effect transistor device fabrication and
characterization ........................................................................................................................ 12
2.3 Quick detection of acute myocardial infarction biomarkers using electronic enzyme-
linked immunosorbent assay .................................................................................................... 18
2.4 Summary ........................................................................................................................ 35
2.5 References ...................................................................................................................... 36
Chapter 3 Highly sensitive and wearable In 2O 3 nanoribbon transistor biosensors with
integrated on-chip gate for glucose monitoring in body fluids ................................................... 41
3.1 Introduction ................................................................................................................... 41
3.2 Preparation process of flexible indium oxide field-effect transistor array ................. 45
3.3 Development of on-chip gold side gate for flexible indium oxide field-effect
transistors .................................................................................................................................. 49
3.4 Flexibility of indium oxide TFTs .................................................................................. 53
3.5 Glucose sensing in human body fluids ......................................................................... 56
3.6 Supporting information ................................................................................................. 68
3.7 Summary ........................................................................................................................ 75
3.8 References ...................................................................................................................... 76
Chapter 4 Multiplexed sensing of serotonin and dopamine using ultra-flexible In 2O 3
nanoribbon aptamer-field-effect transistors ................................................................................ 82
4.1 Introduction ................................................................................................................... 82
4.2 Development of ultra-flexible In 2O 3 nanoribbon biosensor array .............................. 86
V
4.3 Electrical performance and stability test of In 2O 3 nanoribbon biosensor ................... 89
4.4 Multiplexed sensing of serotonin and dopamine ......................................................... 94
4.5 Summary ...................................................................................................................... 100
4.6 References ......................................................................................................................... 100
Chapter 5 Fully-printed all-solid-state organic flexible artificial synapse for neuromorphic
computing ................................................................................................................................... 106
5.1 Introduction ................................................................................................................. 106
5.2 Fabrication of organic flexible artificial synapse ...................................................... 109
5.3 Long-term potentiation/depression (LTD/LTP) of fully-printed organic artificial
synapses................................................................................................................................... 113
5.4 Paired-pulse facilitation (PPF) of fully-printed organic artificial synapses ............. 118
5.5 Applications of fully-printed organic artificial synapses ........................................... 122
5.6 Summary ...................................................................................................................... 129
5.7 References .................................................................................................................... 130
Chapter 6 Conclusions and future work .................................................................................... 135
6.1 Conclusions ................................................................................................................. 135
6.2 Future direction of biosensing and biomimetic electronics ....................................... 137
6.3 References .................................................................................................................... 140
Bibliography ............................................................................................................................... 141
VI
List of figures
Figure 1.1 (a) Schematic showing the field-effect transistor sensors with nanostructure channel.
With bonded receptors, the current does not show any changes (inset of (a)). (b) As
the targeted molecules approaching the sensor surface, the changes in the enviroment
will change the current in the sensor. .......................................................................... 2
Figure 1.2 Simulation of drain current (I ds) against gate voltage (V lg) curves for before (black)
and after (red) protein attachment on an ambipolar carbon nanotube FET sensor due to
(a) mobility change, (b) dielectric change, and (c) gate bias. ...................................... 4
Figure 1.3 Schematic showing switching mechanism in “read” and “write” operations in the
organic neuromorphic devices. During the “read” operation (a), the external switch is
open to forbid electron flow, leading to a stable channel conductance. During the
“write” operation (b), the external switch is closed, permitting electrons to flow in and
out of the gate, resulting in changes of channel conductance. ..................................... 5
Figure 2. 1 In 2O 3 FET-based biosensors fabricated by shadow mask. (a) Schematic illustration
of the lithography-free biosensor fabrication procedure. First, 1
st
layer shadow mask
was attached onto the SiO 2/Si substrate. In 2O 3 thin films were deposited by RF
sputtering. After detaching the 1
st
layer shadow mask and replacing with 2
nd
layer, 1/50
nm Ti/Au was deposited to the surface by e-beam evaporation. By removing the 2
nd
layer shadow mask, we got FET-based biosensor with pristine surface. (b) Optical
image of a 3-inch wafer of In 2O 3 nanoribbon biosensor. (c) Magnified image of a
biosensor chip with 5 nanoribbon devices in one group. (d) A SEM image of two In 2O 3
nanoribbon devices (L=500 μm, W=25 μm). (e) An AFM image of a ~16 nm-thick
In 2O 3 nanoribbon. ...................................................................................................... 14
Figure 2. 2 Electrical characterization of In 2O 3 nanoribbon biosensor. (a) Family of I DS-V DS
curves measured in ambient environment. Drain current as a function of drain voltage
with the back gate voltage varying from 0 to 50 V in steps of 10 V. (b) Drain current
versus back gate voltage with drain voltage fixed at 1 V. Current was plotted in
logarithmic scale in left axis and in linear scale in right axis. (c) Family curves of I DS-
V DS measured in 0.01 x PBS with liquid gate varying from 1 V to 0.5 V in steps of 0.1
V. (d) Drain current versus liquid gate voltage with drain voltage fixed at 1 V, also
plotted in linear and logarithmic scale. (e) Change in threshold voltage with pH range
from 5 to 10, and obtained a pH sensitivity of ~ 60.5 mV/pH. (f) Real-time responses
obtained from an In 2O 3 nanoribbon device exposed to commercial buffer solutions
with pH 5 to 10. ......................................................................................................... 16
Figure 2. 3 Histogram of the mobilities of 50 devices randomly selected from a wafer, showing
the mobility = 13.09 ± 1.39 cm
2
V
-1
S
-1
....................................................................... 17
Figure 2. 4 In 2O 3 nanoribbon biosensor and electronic ELISA for cardiac Troponin I (cTnI),
Creatine kinase-MB (CK-MB) and B-type natriuretic peptide (BNP) detection. (a)
Schematic diagram of the antibodies capture specific antigens in buffer solutions. (b)
Schematic illustrate the electronic ELISA sensing setup. (c) Real-time sensing results
of 1 pg/ml, 10 pg/ml and 300 pg/ml of cTnI antigens in 1 x PBS buffer. (d) Average
VII
sensing results of 3 devices from 3 concentrations of cTnI proteins in 1 x PBS buffer
marked as black square and 1 concentration of troponin I in diluted human whole blood
marked as red dot. Error bars represent standard deviations of 3 devices. (e) Real-time
sensing results of 0.1 ng/ml, 1ng/ml and 3 ng/ml of CK-MB proteins in 1 x PBS buffer.
(f) Average sensing responses from different concentrations of CK-MB. (g) Real-time
responses of 10 pg/ml, 50 pg/ml and 90 pg/ml of BNP proteins in 1 x PBS buffer. (h)
BNP biomarker concentrations versus signal for 3 concentrations of BNP in 1×PBS
and 1 concentration in diluted human whole blood. .................................................. 22
Figure 2. 5 Real-time sensing results of cardiac biomarkers in diluted human whole blood. (a)
Schematic diagram of antibodies that were anchored on the biosensor surfaces capture
specific biomarkers in diluted human whole blood. (b) Real-time sensing results of 10
pg/ml of cTnI in 10×diluted human whole blood, and the averaged results were plotted
as red dot in Figure 2.4d. (c) Real-time sensing results of 300 pg/ml of CK-MB in 10
x diluted human whole blood, and the results were plotted as red dot in Figure 2.4f. (d)
Real-time sensing results of 30 pg/ml of BNP in 10 x diluted human whole blood, and
the results were plotted as red dot in Figure 2.4h. ..................................................... 29
Figure 2. 6 Reusability of In 2O 3 nanoribbon biosensors. (a) Schematic diagram of regeneration
process by applying 50 mM NaOH to the used sensor surface. After antibodies release
antigens, the sensors are ready to repeat sensing. (b) Real-time responses of 100 pg/ml
cTnI proteins in 1×PBS buffer. (c) Real-time sensing responses from the same
concentration of cTnI and the same devices after regeneration. (d) Real-time response
of the same sensors after 3 more cycles of regeneration and sensing process. .......... 31
Figure 2. 7 Real-time sensing results from sensor chips functionalized with Troponin I antibody.
(a) Incubated in Troponin I antigen solution with concentration of 10 pg/ml. (b)
Incubated in CK-MB antigen solution with concentration of 1 ng/ml. (c) Incubated in
BNP antigen solution with concentration of 50 pg/ml. ............................................. 32
Figure 2. 8 Real-time sensing results from sensor chips functionalized with CK-MB antibody.
(a) Incubated in CK-MB antigen solution with concentration of 1 ng/ml. (b) Incubated
in Troponin I antigen solution with concentration of 10 pg/ml. (c) Incubated in BNP
antigen solution with concentration of 50 pg/ml. ...................................................... 32
Figure 2. 9 Real-time sensing results from sensor chips functionalized with BNP antibody. (a)
Incubated in BNP antigen solution with concentration of 50 pg/ml. (b) Incubated in
Troponin I antigen solution with concentration of 10 pg/ml. (c) Incubated in CK-MB
antigen solution with concentration of 1 ng/ml. ........................................................ 33
Figure 3. 1 (a) Schematic diagrams showing the fabrication procedure of In 2O 3 FETs on a PET
substrate using 2-step shadow masks. (b) Schematic diagrams of functionalization on
the surface of the electrodes using ink-jet printing. (c) Photograph of as-fabricated
In 2O 3 FETs. Scale bar is 1 cm. (d) Optical image shows a group of In 2O 3 biosensors
with two gold side gate electrodes. Scale bar is 500 µm. (e) SEM image of an In 2O 3
nanoribbon devices (L=500 μm, W=25 μm) and a gold side gate electrode (W=150
μm). (f) Photograph of In 2O 3 FET foil laminated on an artificial human hand. Scale
bar is 3 cm. (g) In 2O 3 biosensor foil in a rolled-up state. Scale bar is 3 cm. (h)
Photograph of an In 2O 3 FET chip attached onto the back casing of a watch. Scale bar
is 1 cm. ...................................................................................................................... 47
VIII
Figure 3. 2 Electrical characterization of In 2O 3 nanoribbon biosensors using gold side gate
electrodes. (a) Drain current versus Ag/AgCl gate voltage with drain voltage changing
from 0.4 V to 0 V in steps of 0.1 V. Inset shows the schematic diagram of the
measurement setup. (b) Family of I DS-V DS curves measured with a Ag/AgCl gate
electrode. (c) Drain current versus gold side gate voltage with drain voltage varying
from 0.4 V to 0 V in steps of 0.1V. (d) Family curves of I DS-V DS measured with gold
side gate voltage varying from 1 V to 0 V in steps of 0.2V. (e) Reference voltage
measured with a gold electrode versus the gold side gate voltage. (f) Transfer
characteristics of a representative FET with different gate-to-channel distances under
V DS = 0.2 V. ............................................................................................................... 52
Figure 3. 3 Flexibility of In 2O 3 FETs on a PET substrate. (a) Photograph of a biosensor foil
wrapping around a glass cylinder. Scale bar is 5 mm. (b) Transfer characteristics of a
representative In 2O 3 FET under relaxed state, bent with a radius of ~ 3 mm, and after
bending 100 times. (c) Mobility, (e) threshold voltage, and (g) on-off ratio of In 2O 3
FETs bent with different radius. (d) Mobility, (f) threshold voltage, and (h) on-off ratio
of In 2O 3 FETs bent with a radius of ~ 3 mm after different bending cycles. ............ 55
Figure 3. 4 PH sensing and glucose sensing. (a) Schematic diagram showing a PDMS microwell
is attached to the PET substrate with In 2O 3 FETs. (b) Family curves of I DS-V DS
measured with the channel area submerged in the PDMS well. (c) Family curves of
I DS-V GS measured with the channel area submerged in the PDMS well. (d) PH sensing
when the gate bias was applied with a Ag/AgCl electrode or a gold side gate electrode.
(e) Schematic diagram showing the working principle of glucose sensor. (f) Glucose
sensing results in 0.1×PBS with a gold side gate. ..................................................... 57
Figure 3. 5 Glucose sensing in human body fluids. (a) Real-time glucose sensing results in
artificial tears. (b) Real-time glucose sensing resulting in artificial sweat. (c) Real-time
glucose sensing in human saliva. (d) Comparison of sensing responses verses different
glucose concentrations in 0.1×PBS, artificial tears, artificial sweat, and saliva. ....... 61
Figure 3. 6 Off-body glucose sensing. (a) Photographs of the In 2O 3 biosensors attached onto an
eyeball replica and an artificial arm. (b) Real-time glucose sensing results on an
artificial eyeball. (c) Real-time glucose sensing results on an artificial arm. (d) Real-
time glucose sensing with real sweat collected from human subjects. (e) Glucose
sensing results of real sweat collected before and after glucose beverage intake. .... 65
Figure 3. 7 (a) AFM image with height profile of a ~ 20 nm thick In 2O 3 nanoribbon. (b) Zoom-
in AFM scan of a 1 µm × 1 µm square on the In 2O 3 film. (c) XRD of RF sputtered
In 2O 3 film deposited on top of PET substrate. ........................................................... 68
Figure 3. 8 A representative transfer curve of an In 2O 3 nanoribbon FET with V DS = 0.2 V and its
g m. .............................................................................................................................. 69
Figure 3. 9 Electrical performance of 50 In 2O 3 nanoribbon transistors (a) Mobilities (μ), (b)
Threshold voltage (V TH), (c) On/off current ratios at V DS = 0.2 V, and (d) On-state
current (I ON) at V GS = 0.6 V and V DS = 0.2 V. ........................................................... 70
Figure 3. 10 Transfer characteristics of (a) an unfunctionalized In 2O 3 FET under relaxed state,
bent with a radius of ~3, 10, and 15 mm, (b) an unfunctionalized In 2O 3 FET after
IX
bending with 5, 10, 50, and 100 cycles, (c) a functionalized In 2O 3 FET under relaxed
state, bent with a radius of ~3, 10, and 15 mm, and (d) a functionalized In 2O 3 FET after
bending with 5, 10, 50, and 100 cycles. .................................................................... 71
Figure 3. 11 The mobilities of In 2O 3 FETs as a function of tensile strain. ................................ 72
Figure 3. 12 Real-time sensing responses of an In 2O 3 FET to standard pH calibration solutions.
Gate voltage is applied with (a) a Ag/AgCl gate electrode, and (b) a gold side gate
electrode. ................................................................................................................... 73
Figure 3. 13 Glucose sensing results of an In 2O 3 nanoribbon biosensor functionalized with
chitosan and SWCNT only. ....................................................................................... 74
Figure 3. 14 Glucose sensing results with a functionalized sensor after 1, 2, 4, 7, and 14 days.
................................................................................................................................... 75
Figure 4. 1 (a) Schematic illustration of the fabrication process flow of ultra-flexible In 2O 3 field-
effect transistor (FET) based multiplexed sensors. We began with a Si/SiO 2 substrate
coated with polydimethylsiloxane (PDMS) sacrificial layer. Next, a 1.4-μm-thick
polyethylene terephthalate (PET) film was laminated on the wafer. In 2O 3 nanoribbons
were sputtered onto the PET substrate through a shadow mask. Ti/Au (1/50 nm thick)
were deposited through another shadow mask and patterned as source, drain, side gate
and temperature sensor. As-fabricated biosensor foil was then peeled off from the rigid
carrier wafer. (b) A photograph of as-fabricated multiplexed sensors on a 1.4-μm-
thick PET film. Scale bar is 1 cm. (c) Optical microscope image of a gold resistive
temperature sensor, four In2O3 nanoribbon FETs, and a gold side gate (from top to
bottom). Scale bar is 0.5 mm. (d) Ultra-flexible biosensor array conformally attached
to human skin. Scale bar is 2 cm. (e) A prototype biosensor foil compressed as human
body movement. Scale bar is 1 cm. ........................................................................... 88
Figure 4. 2 Electrical performance of ultra-flexible In 2O 3 nanoribbon biosensor. (a) Typical
transfer characteristics of an In 2O 3 transistor with L = 500 μm and W = 25 μm using
gold liquid gate. Black line represents I D (drain current) in logarithmic scale and blue
line represents I D in linear scale; V GS, the gate−source voltage. The applied drain–
source voltage, V DS, is 0.2 V. (b) Corresponding output curves in saturation regime. In
this plot, I DS is a function of V DS with V GS from 0 V to 1 V in 0.2 V steps. (c) Map of
charge-carrier mobility of 56 transistors in the array. (d) Histograms showing on-
currents and threshold voltages from the 56 randomly selected transistors in the
array. Average mobility is 17.84 ± 1.79 cm
2
V
−1
s
−1
, average on-current is 0.86 ± 0.13
μA, and average threshold voltage is 0.268 ± 0.024 V. ............................................. 90
Figure 4. 3 Stability of ultra-flexible In 2O 3 nanoribbon biosensors. (a) Photograph of a group of
ultra-flexible In 2O 3 biosensors on 1.4-μm-thick PET films wrapped around a copper
wire with a radius of 100 μm. Scale bar, 5 mm. (b) Crumpled In 2O 3 biosensor foil
(original size: 5 cm × 5 cm square). Scale bar is 0.5 cm. (c) Transfer characteristics of
a representative In 2O 3 transistor under relaxed state, bent with a radius of ∼0.2 mm
and after crumpling, respectively. Mobilities and threshold voltages obtained in a
released state during (d) 100 bending cycles and (e) 100 crumpling cycles. (f) Transfer
characteristics of an In 2O 3 transistor measured immediately after fabrication (Day 0)
and after immersed in 1×PBS for 1, 2, 3, 5, and 7 day(s). ......................................... 92
X
Figure 4. 4 Experimental characterizations of serotonin and dopamine sensors. (a) Aptamer stem-
loops reorient away from semiconductor channels, thereby increasing
transconductance. (b) Real-time sensing results from three In 2O 3 nanoribbon
biosensors functionalized with serotonin aptamers. The devices showed responses to
serotonin (in 1×artificial cerebrospinal fluid) concentrations ranging from 10 fM to 1
μM. (c) The relationship between the serotonin concentration and the saturated current
response from total 9 different devices. Negative control experiment results from
unfunctionalized devices were also plotted in this figure. (d) Aptamers reorient closer
to FETs to deplete channels electrostatically. (e) The results of real-time sensing of
dopamine. (f) A summary of dopamine concentration and the corresponding responses
from 9 different devices. All the devices were operated under V DS = 0.2 V and V GS =
0.25 V. ....................................................................................................................... 96
Figure 4. 5 Control experiments for serotonin and dopamine sensing. The same sensing
procedure on unfunctionalized devices (a) without serotonin aptamers and (b) without
dopamine aptamers as a negative control experiment. .............................................. 97
Figure 4. 6 (a) The resistance of the temperature sensor in buffer solution with temperature
ranging from 20 °C to 50 °C. (b) Real-time pH sensing with three unfunctionalized
In 2O 3 nanoribbon devices exposed to commercial buffer solutions with pH value from
10 to 4. (c) Biosensor foil conformally attached on an artificial PDMS brain. The
devices were connected out using Indium wire for further electrical measurements.
Scale bar is 1 cm. (d) Multiplexed sensing with temperature, pH, serotonin and
dopamine. 1×artificial cerebrospinal fluid (pH = 7.4) spiked with 1 pM serotonin, 1
pM dopamine, and 1 nM serotonin, and 1 nM dopamine were sequentially dropped on
the sensors, and only devices functionalized with target aptamers showed responses to
corresponding target molecules. All sensors responded to pH changes (pH value from
7.4 to 7.3). The devices were operated under V DS = 0.2 V and V GS = 0.25 V. .......... 99
Figure 5. 1 Fully-printed organic neuromorphic devices. a) Schematic illustration of the key
fabrication procedures for organic neuromorphic devices with printing technology. b),
c) Schematic showing switching mechanism in “read” and “write” operations in the
organic neuromorphic devices. During the “read” operation (left), the external switch
is open to forbid electron flow, leading to a stable channel conductance. During the
“write” operation (right), the external switch is closed, permitting electrons to flow in
and out of the gate, resulting in changes of channel conductance. d) Image of a device
array with 45 organic neuromorphic devices. Scale bar is 1 cm. e), f) Magnified image
of one device in the array, clearly showing the PEDOT:PSS layer before electrolyte
layer patterning (e) and the PDADMAC film after patterning (f). Both scale bars
represent 2 mm. g) Photograph of an array of organic neuromorphic devices on flexible
substrate while being bent. Scale bar is 1 cm. ......................................................... 110
Figure 5. 2 Surface modification with oxygen plasma. Contact angles of water on unmodified
PET substrate (a), on oxygen plasma-modified PET (b), on unmodified silver
conductive film (c), and on oxygen plasma-modified silver conductive film (d). After
being treated with oxygen plasma (100 W, 150 mTorr) for 30 seconds, the contact
angle of water changes from 71 to 30 on PET substrate, and from 138 to 89 on
silver film, which indicates the surface of both PET and silver film become more
hydrophilic. e).......................................................................................................... 112
XI
Figure 5. 3 Oxidation and reduction reaction of an PEDOT:PSS based all-solid organic
neuromorphic devices. The molecular structures of PEDOT:PSS and PDADMAC are
illustrated in this figure. Upon applying a negative Vpre to the PEDOT:PSS electrode,
protons flow from the postsynaptic electrode into the presynaptic electrode through
PDADMAC electrolyte, resulting in deprotonation of the PEI, and further cause the
oxidation of PEDOT due to charge neutrality. This causes holes to be generated on the
PEDOT backbone, thereby reducing the electronic resistivity of the postsynaptic
electrode. The reaction is reversed when applying a positive presynaptic potential. The
charge transfer is marked in red in the figure. ......................................................... 112
Figure 5. 4 Long-term neuromorphic behavior. a) Long-term potentiation and depression
exhibiting 100 discrete states when the device is programmed with presynaptic pulses.
The two insets are zoom-in plots showing the individual states. b) LTP/LTD cycling
stress tests when the organic neuromorphic device is in a relaxed state (upper panel)
and after 500 bending cycles (bottom panel). c) State retention for organic
neuromorphic devices. The conductance is monitored for 10 seconds after a 1 s pulse
is applied to change the states. The pulse amplitudes are as labelled. Different pulse
amplitude can switch the conductance into different states, and after a pulse with
amplitude equal to the initial one, the conductance switched back to the state similar
to the initial one. d) Retention of the HRS and LRS currents at V post = 100 mV and
V pre = 0 V in an eight-hour period. e) Schematic showing the electrical implementation
for STDP measurement. The organic neuromorphic device is connected between a pre-
synaptic spike generator and a post-synaptic spike generator. f) STDP behavior of the
device stimulated with a pair of spikes with different values of t. ........................ 116
Figure 5. 5 Paired-pulse facilitation. a) Schematic showing the electrical setup for PPF
measurement. Two pre-synaptic spike generators are probed on the pre-synaptic
electrode. The inset shows the recorded waveform of pulses applied to the devices.
The pulse amplitude is 50 mV, the pulse width is 25 ms, and the spiking timing t is
ranging from 1 ms to 500 ms. b) Post-synaptic current with different spike timing. c)
Post-synaptic weight changes trigged by a paired-pulse with time interval of 25 ms. G 1
and G 2 represent the conductance change of the first pulse and the second pulse,
respectively. G equals the difference between G 2 and G 1. d) Paired pulse facilitation
with different time intervals. An exponential fit is applied to obtain two characteristic
time scales. e) Switching energy measured as a function of device area. The slope of
linear fit is 20.5 nJ/mm
2
. .......................................................................................... 119
Figure 5. 6 Change in postsynaptic conductance as a function of presynaptic pulse duration (a)
and amplitude (b). The measured excitatory post-synaptic currents (EPSC) are
converted into conductance change (G) of the post-synaptic electrode. With the
presynaptic voltage fixed at 20 mV, the G values increases from 0.5 S (inset) to 371
S for spike duration ranges from 10 ms to 8 s, respectively. The spike voltage-
dependent EPSCs are also studied. With the presynaptic pulse duration fixed at 2 s, the
G increases from 48 S to 251 S for spike voltage ranges from 10 mV to 50 mV.
As the linear fitting shown in both figures (in red), the conductance change is a linear
function of presynaptic pulse duration and voltage. ................................................ 121
XII
Figure 5. 7 Logic circuits based on neuromorphic devices. a) b) Schematic diagram showing the
circuits used for logic gates, AND gate (series connection) and OR gate (parallel
connection), respectively. c) The change of the AND gate conductance depends on the
presynaptic inputs. When the presynaptic signals only came from synapse 1 or synapse
2, the conductance change did not reach the threshold line. When both synapses fired,
the change of conductance passed the threshold line. d) For OR gate, even when a
single synapse fired, the change of conductance slightly passed the threshold line. 123
Figure 5. 8 Simulation of organic neuromorphic device-based neural networks. a) Schematic
illustration of the implementation learning for pattern recognition. b) Schematic
illustration of the architectural neural network with fabricated three-terminal devices.
c) Conductance variation (G) as a function of the conductance states showing the
switch statistics of neuromorphic devices during long-term potentiation (red squares)
and depression (blue squares). d), e) Backpropagation training results using Optical
Recognition of Handwritten Digits and MNSIT database handwritten digits data-sets
in the format of 8 × 8 pixel digit image (d) and 28 × 28 pixel digit image (e). f)
Backpropagation training results for face recognition using the AT&T Laboratories
Cambridge ORL database of faces. ......................................................................... 126
Figure 5. 9 Dynamic image processes and retention. (a) Functionality test on an organic synaptic
transistor. As the presynaptic input arrived at 0 s, the conductance (synaptic weight)
changes from low conductance state (~ 190 S) to high conductance state (~ 230 S).
(b) Dynamic image processes on a 3 × 3 synaptic array with local presynaptic input.
The diagram shows that the capital letter “U”, “S”, and “C” were memorized by the
synaptic array. (c) Retention (forgetting) evaluation of the capital letter “S” after 1, 2,
8 and 24 hours after the image was memorized by the synaptic array. ................... 128
Figure 6. 1 (a), Schematics showing the system that is fabricated layer by layer. VIAs are used
for interlayer electrical connections. Key components in each layer are labelled. BLE,
Bluetooth; EP, electrophysiological potential. (b), Optical micrographs of the system
when freestanding (top), twisted at 90° (middle) and poked with a dome height of
~8 mm (bottom), highlighting its superb mechanical compliance and robustness. . 138
Figure 6. 2 Reaction scheme for biofunctionalization of PEDOT:PSS by incorporation of PVA.
................................................................................................................................. 139
XIII
Abstract
In this dissertation, I present my work on the development of indium oxide
nanoribbon field-effect transistors for biosensing applications and the
development of artificial synapses using organic semiconductors. Sensors
built from nanostructure metal oxide field-effect transistors (FET) have the
combined advantages of high sensitivity, good flexibility, being equipped with
an electronic read-out that can be fast-responding, wearable, and accessible.
Non-volatile, flexible artificial synapses that can be used for brain-inspired
computing are highly desirable for emerging applications such as human-
machine interfaces, soft robotics, medical implants and biological studies.
Chapter 1 is an introduction of biosensing and biomimetic electronics. It
mainly focuses on the mechanism biological sensing of FET sensors based
on metal oxide, and the mechanism of artificial synapses based on organic
materials.
In chapter 2, a scalable and facile lithography-free method for fabricating
highly uniform and sensitive In
2
O
3
nanoribbon biosensor arrays are
XIV
demonstrated. Combining with electronic enzyme-linked immunosorbent
assay (ELISA) for signal amplification, the In
2
O
3
nanoribbon biosensor arrays
are optimized for early, quick and quantitative detection of cardiac biomarkers
in diagnosis of acute myocardial infarction. With the demonstrated sensitivity,
quick turnaround time, and reusability, the In
2
O
3
nanoribbon biosensors have
shown great potential toward clinical test for early and quick diagnosis of
acute myocardial infarction (AMI).
In chapter 3, highly sensitive and conformal In
2
O
3
nanoribbon FET biosensors
with fully integrated on-chip gold side gate are demonstraded. The devices
can be laminated onto various surfaces, such as artificial arms and watches,
and have enabled glucose detection in various body fluids, such as sweat and
saliva. With the electrodes functionalized with glucose oxidase, chitosan, and
single-walled carbon nanotubes, the glucose sensors show very wide detection
range spanning at least 5 orders of magnitude and detection limit down to 10
nM. Therefore, our high-performance In
2
O
3
nanoribbon sensing platform has
great potential to work as indispensable components for wearable healthcare
electronics.
XV
In chapter 4, ultra-flexible and highly sensitive aptamer-field-effect-transistor
In
2
O
3
nanoribbon biosensors for real-time multiplexed neurotransmitters
sensing are demonstrated. Arrays of In
2
O
3
nanoribbon field-effect transistors
were fabricated on 1.4-μm-thick plastic substrate using shadow mask
techniques, showing excellent electrical performance and device uniformity.
The conformal sensor array exhibited multiplexed sensing of temperature, pH,
serotonin, and dopamine, and can function properly in artificial cerebrospinal
fluid and on an artificial brain. These results represent significant progress in
the fabrication of ultra-flexible aptamer-field-effect-transistors for brain
mapping and physiological monitoring applications.
In chapter 5, the experimental realization of a non-volatile artificial synapse
using organic polymers in a scalable fabrication process are demonstrated.
The three-terminal electrochemical neuromorphic device successfully
emulates the key features of biological synapses: long-term
potentiation/depression, spike-timing-dependent plasticity learning rule,
paired-pulse facilitation, and ultralow energy consumption. The artificial
synapse network exhibits excellent endurance against bending tests and
enables a direct emulation of logic gates, which shows the feasibility of using
them in futuristic hierarchical neural networks. Based on our demonstration
XVI
of 100 distinct, non-volatile conductance states, we achieved high accuracy in
pattern recognition and face classification neural network simulations.
The last chapter, chapter 6, is the summary and future direction of biosensing
and biomimetic electronics.
1
Chapter 1 Introduction to biosensing
and biomimetic electronics
1.1 Overview
Sensor technology has been an important part of many sectors of society
ranging from agricultural and energy to transportation security and healthcare.
The explosion of nanotechnology within the decades has pushed the boundary
of response times, detection limits, sensitivity, portability, etc. for sensor
technology, particularly for biological sensors. The sensors based on
nanotechnology have at least one dimension in the range of 1 to 100 nm,
which gives the sensor good mechanical response, high sensitivity due to their
large surface-to-volume ratio, portability for wearable sensing applications
One obstacle in producing nanosensors for practical, everyday use is the
difficulty in assembling nanostructures in a controllable, repeatable, and
scalable fashion to allow the development of a systematic technology. This
thesis will explore the sensing applications from point-of-care platform to
wearable and implantable electronics. In addition, this thesis presents artificial
synapses based on organic materials, as such electronics are also flexible and
have ultralow power consumption, which will benefit the portable/wearable
sensing applications.
2
1.2 Mechanism of biosensing electronics using
field-effect transistors
The operation principles of a field-effect transistor (FET) nanobiosensor are
illustrated in Figure 1.1 and explained in literature.
1
The metal oxide
nanostructure is used as the channel of the FET, bridging source-drain (S-D)
electrodes. The surface of the channel is chemically bonded to receptor
molecules (Figure 1.1(a)). These receptors have a high binding affinity and
selectivity toward a particular target analyte. When the target molecule
approaches the sensor surface, it is captured by the receptor molecule (Figure
1.1(b)). This binding interaction changes the environment at the channel
surface—the analyte is usually electrically charged—causing a change in the
conductance of the channel (1.1(b) inset) at binding. This electrical signal read
out enables for label-free detection of analytes, without complex optical
equipment.
Figure 1.1 (a) Schematic showing the field-effect transistor sensors with nanostructure
channel. With bonded receptors, the current does not show any changes (inset of (a)). (b)
S
D
Receptors
Nanostructure
Channel
(a)
Current
Time
(b)
Current
Time
Captured
Target
3
As the targeted molecules approaching the sensor surface, the changes in the enviroment
will change the current in the sensor.
2
The mechanism of the conduction changing during molecular binding is also
a debated topic.
3-6
According to the ideal transistor linear (region often used
for biosensing) current equation
𝐼 𝑑𝑠
=𝑒𝜇𝜀 𝜀 𝑟 𝐴 𝑑 𝑉 𝑑𝑠
𝐿 (𝑉 𝑔 −𝑉 𝑇 ) (10)
While the transistor dimenions (A, d, and L) and the drain voltage (V
ds
) are
constant, a change in conduction current (I
ds
) can be caused by either a change
in mobility (μ), a change in capacitance due to the difference in the dielectric
constant (ε
r
) of the sensing environment versus the binding molecule, or a
gating effect (V
g
) caused by charges from the binding molecule. These three
situations are illustrated in Figure 1.1 by comparing the I
ds
–V
gs
curves of an
ambipolar FET device before and after protein binding. Figure 1.1(a) shows
that a decrease in the slope of the I
ds
–V
gs
curve after protein binding also
decrease the I
ds
at fixed V
gs
. A change in I
ds
due to the slope indicates a
reduction in mobility and transconductance inside the channel, possibly due
to an uneven electrostatic field distribution caused by random binding with
charged biomolecules. In Figure 1.2 (b) the gate bias is shown to be less
effective at inducing I
ds
. The current reduction in this case can be attributed
to a reduced gate capacitance caused by the low permittivity of the bound
4
biomolecule. Finally, Figure 1.2(c) shows a I
ds
change due to electrostatic
gating of the FET channel by charged target biomolecules. This type of
change causes a threshold voltage (V
T
) shift seeing in the figure.
Figure 1.2 Simulation of drain current (Ids) against gate voltage (V lg) curves for before
(black) and after (red) protein attachment on an ambipolar carbon nanotube FET sensor
due to (a) mobility change, (b) dielectric change, and (c) gate bias.
3
1.3 Mechanism of biomimetic synapse using
organic electronics
The programming (“read” and “write”) of the neuromorphic devices based on
redox reaction is similar to charging and discharging a battery.
7-10
During a
“read” operation, the external switch is open and there is no electric signal
flow, and therefore the proton concentration remains unaltered in each layer,
as exhibited in Figure 1a. To achieve “write” operation, the switch is closed
and a signal from the gate electrode is regarded as the presynaptic stimulus,
as exhibited in Figure 1b. When applying a positive presynaptic pulse Vpre,
cations are injected into the postsynaptic electrode through the presynaptic
(a) Mobility
(b) Capacitance
(c) Gating
5
electrode and the electrolyte. Thus, the organic device is switched to a low
conductance stage since the number of holes is reduced in the postsynaptic
electrode due to protonation of the PEDOT film. After this “write” step, the
device is disconnected, and the energy barrier between the channel and
electrolyte forbids electronic charge transport, keeping the electrode
conductance state in a non-volatile way.
Figure 1.3 Schematic showing switching mechanism in “read” and “write” operations in
the organic neuromorphic devices. During the “read” operation (a), the external switch is
open to forbid electron flow, leading to a stable channel conductance. During the “write”
operation (b), the external switch is closed, permitting electrons to flow in and out of the
gate, resulting in changes of channel conductance.
1.4 References
1. Curreli, M.; Zhang, R.; Ishikawa, F.N., et al. Real-Time, Label-Free Detection of
Biological Entities Using Nanowire-Based FETs. Ieee T Nanotechnol 2008, 7 (6), 651-
667.
6
2. Ishikawa, F.N. Applications of one-dimensional structured nanomaterials as
biosensors and transparent electronics. University of Southern California, Los Angeles,
CA, 2009.
3. Heller, I.; Janssens, A.M.; Mannik, J., et al. Identifying the mechanism of
biosensing with carbon nanotube transistors. Nano Lett 2008, 8 (2), 591-595.
4. Nair, P.R.; Alam, M.A. Screening-limited response of nanobiosensors. Nano Lett
2008, 8 (5), 1281-1285.
5. Stern, E.; Wagner, R.; Sigworth, F.J., et al. Importance of the debye screening
length on nanowire field effect transistor sensors. Nano Lett 2007, 7 (11), 3405-3409.
6. Tang, X.; Bansaruntip, S.; Nakayama, N., et al. Carbon nanotube DNA sensor and
sensing mechanism. Nano Lett 2006, 6 (8), 1632-6.
7. van de Burgt, Y.; Melianas, A.; Keene, S. T.; Malliaras, G.; Salleo, A., Organic
electronics for neuromorphic computing. Nat. Electron. 2018, 1.
8. Gkoupidenis, P.; Schaefer, N.; Garlan, B.; Malliaras, G. G., Neuromorphic
Functions in PEDOT:PSS Organic Electrochemical Transistors. Adv. Mater. 2015, 27,
7176-80.
9. Gkoupidenis, P.; Koutsouras, D. A.; Malliaras, G. G., Neuromorphic device
architectures with global connectivity through electrolyte gating. Nat. Commun. 2017, 8,
15448.
10. Wu, C.; Kim, T. W.; Choi, H. Y.; Strukov, D. B.; Yang, J. J., Flexible three-
dimensional artificial synapse networks with correlated learning and trainable memory
capability. Nat. Commun. 2017, 8, 752.
7
Chapter 2 Highly sensitive and quick
detection of acute myocardial infarction
biomarkers using In
2
O
3
nanoribbon
biosensors fabricated using shadow
masks
2.1 Introduction
Every year about 5 million patients visit the emergency department because
of chest pain symptoms, but only 10% of these patients experience acute
myocardial infarction (AMI).
1
If an initial electrocardiogram (ECG)
assessment at the emergency department reveals a ST-segment elevation, the
patient is placed at high risk for acute myocardial infarction (AMI), or heart
attack, and the established medical procedures are administered to the patient.
However, the ECG sensitivity may be as low as 50%,
2-5
and patients who
show no ST elevation can still be at high risk for unstable angina or non-ST
segment elevation AMI. For this reason, cardiac biomarkers have become
increasingly important for swift risk stratifying and diagnosing patients who
may still need immediate treatment.
8
The effectiveness of the biomarkers to properly diagnose and triage chest pain
patients is based on several factors. First, the test turnaround time should be
short because early treatment of myocardial infarction is crucial to recovery.
The American Heart Association has stated a recommended turn-around time
of 60 minutes and a preferred turnaround time of 30 minutes from sample
collection to result reporting.
6
Second, obtaining the trend in the cardiac
biomarker concentration in the hours after a patient’s arrival is a crucial
addition to the initial cardiac biomarker reading for accurate diagnosis.
Current biomarker trends are collected through serial biomarker readings,
such as testing at 0, 30, 60, and 90 minutes after patient arrival at the
emergency department.
7
Such fast turnaround times are difficult to achieve in
a central laboratory setting and is often aided by a point-of-care (POC)
device.
8
Additionally, multiple cardiac biomarkers testing may improve the
diagnosis process of heart attack over single biomarker testing.
9
The National
Academy of Clinical Biochemistry has recommended testing for an early
biomarker that elevates within the first 6 hours of chest pain in conjuncture
with an AMI-specific biomarker that is increased in the blood even after 6 to
9 hours.
10
Point-of-care platforms are ideal for multiple cardiac biomarker
testing with fast turnaround times, but current POC devices lack the good
sensitivity and high specificity of central laboratory biomarker testing.
11
For
9
POC devices to more effectively aid rapid decision making in both the
emergency department and in the field, there is a need for further investigation
of emerging sensor technology in order to bridge the performance gap
between POC device and central laboratory testing for cardiac biomarkers.
Great progress was made with silicon nanoribbons fabricated using top-down
methods which have been shown to be highly sensitive, scalable, and
uniform.
12, 13
However, the fabrication process is rather complicated; for
example, it needs oxidation, photolithography and wet etching. Moreover, Si
nanoribbon devices usually require silicon-on-insulator (SOI) wafers, which
are expensive. Semiconducting-metal-oxide-based biosensors have several
advantages, such as the use of low-cost Si/SiO
2
wafers and the ease of
fabrication when compared to silicon based biosensors. By using the shadow
mask fabrication method, we can choose many kinds of substrates, such as
Si/SiO
2
wafers, glass, or even plastic substrates; however, some of the
commercially available plastic substrates may not be compatible with the
photoresist baking step used for photolithography, which would make the
shadow mask approach advantageous. Furthermore, in comparison with
multiple-step cleanroom fabrication, fabrication with shadow masks is cost-
efficient, highly reliable, photolithography-free, room-temperature processing,
10
and can be performed without using a cleanroom. In addition, nowadays
shadow masks can be obtained rather easily by submitting the mask design to
commercial service providers (such as Photo Sciences Inc.), and the size of
the shadow masks can be as large as 6”, making wafer-scale fabrication very
easy.
Indium oxide (In
2
O
3
) filed-effect transistors (FETs) have been shown to be
real-time and label-free detectors with superb signal-to-noise ratio and the
potential for integrated multiplexing.
14-18
The quick response time makes the
In
2
O
3
nanoribbon sensors especially advantageous for analyzing the first
blood-drawn sample, from which rapid decisions can be made for the patients’
treatment. The small device-to-device variation demonstrated previously
14
can provide good statistical confidence for calibrating cardiac biomarker
concentrations. Furthermore, In
2
O
3
nanoribbon sensors can provide
quantitative analysis for a large detectable concentration range spanning at
least 4 orders of magnitude and a detection limit in the picogram per milliliter
range.
14
This sensitivity can help to differentiate biomarker changes at each
serial reading. Due to the electronic sensing, the final product enjoys facile
interface and compactness while having the capability to integrate with other
microfluidic and electronic functional groups, such as wireless data output.
19
11
These properties make In
2
O
3
nanoribbon sensors well suited for analyzing
medical conditions such as heart attack that require urgent, point-of-care
(POC) medical attention. However, in our previous work, two steps of
photolithography were performed to produce In
2
O
3
nanoribbon sensors,
which may increase the cost and the fabrication time significantly. It is
therefore highly important to develop a low-cost, time-efficient and scalable
lithography-free process to produce In
2
O
3
nanoribbon field effect transistors,
which may generate broad impact to applications such as chemical sensing,
20-
22
protein detection,
23, 24
cancer diagnosis and prognosis,
25
infectious disease
diagnosis, biomedical research,
12, 26
and even thin film transistors for displays
and macro electronics.
Here, we present a lithography-free process for the fabrication of highly
sensitive and scalable FET-based In
2
O
3
nanoribbon biosensors. The
nanoribbons are prepared by sputter-coating In
2
O
3
through a shadow mask
onto a substrate and have ribbon-like cross-section of ~ 16 nm in thickness
and 25 μm in width and 500 μm in length, followed by metal electrode
deposition through another shadow mask. The devices fabricated by shadow
masks show good electrical performance in both ambient and aqueous
environment, with the surfaces never exposed to undesirable chemicals like
photoresist or e-beam resist. In addition, In
2
O
3
nanoribbon devices also show
12
good performance in pH sensing experiments. Through all the sensing
experiments, we have demonstrated that In
2
O
3
nanoribbon biosensors
fabricated using shadow masks can be used to quantitatively detect 3 cardiac
biomarkers within the concentrations relevant to clinical diagnosis with the
turnaround time ~ 45 minutes. We further demonstrated tests using spiked
cardiac biomarkers in diluted human blood. Lastly, by first applying
regeneration buffer to the used sensor surface to anti-bond the antigen-
antibody conjugation and then repeating the sensing experiments, we
demonstrated the reusability of the In
2
O
3
nanoribbon biosensors with very
small variation of each sensing results.
2.2 Indium oxide nanoribbon field-effect transistor
device fabrication and characterization
Indium oxide as an active channel material has been shown to function
effectively in the biosensing platform. In this work, we fabricate the
nanoribbon devices using a shadow mask technique on 3-inch silicon
substrates covered with 500 nm thick silicon oxide. Figure 2.1a shows the
schematic diagram depicting the fabrication process. The first shadow mask
step defines the dimension and position of the nanoribbons by attaching the
13
shadow mask onto the SiO
2
/Si wafer. Then the In
2
O
3
ribbons were deposited
using radio frequency (RF) sputtering with a thickness ~16 nm. We got well-
defined nanoribbons by simply removing the shadow mask instead of lift-off
as in photolithography. The source and drain electrodes were defined using
the second shadow mask. After the alignment and attachment, we deposited 1
nm Ti and 50 nm Au using electron beam evaporation. The whole process is
photoresist-free, so it not only simplifies the fabrication process, but can also
avoid the effect of any photoresist residue in the later sensing experiments.
Figure 2.1b shows a photograph of 28 groups of In
2
O
3
nanoribbon FETs
patterned over a 3-inch wafer using shadow masks, and each group contains
five FET devices (Figure 2.1c). Here each chip has an array of sensors, so that
we can use multiple devices in each round of biosensing to study the
uniformity and consistency. In addition, biosensor arrays can be used for
multiplex sensing to detect multiplexed antigens in one chip in the future. The
scanning electron microscope (SEM) image of the channel regions (Figure
2.1d) shows that the nanoribbons are identical and have very clear edges. In
this structure, the channel width and length were 25 μm and 500 μm,
respectively. Furthermore, In
2
O
3
nanoribbons are smooth with 16 nm
thickness (Figure 2.1e). Thickness is the key factor for sensitivity. In our
previous work,
14
we have demonstrated that the optimum thickness of In
2
O
3
14
nanoribbon should be within 23 nm (Debye length). Here, we obtained
nanoribbon thickness of 16 nm, which is very close to the optimum thickness.
The width and length of the channel are also believed to affect the sensitivity
when they are comparable to the Debye length; however, such channel width
and length are beyond what we can pattern with shadow masks.
Figure 2. 1 In2O3 FET-based biosensors fabricated by shadow mask. (a) Schematic
illustration of the lithography-free biosensor fabrication procedure. First, 1
st
layer shadow
mask was attached onto the SiO2/Si substrate. In2O3 thin films were deposited by RF
sputtering. After detaching the 1
st
layer shadow mask and replacing with 2
nd
layer, 1/50 nm
Ti/Au was deposited to the surface by e-beam evaporation. By removing the 2
nd
layer
shadow mask, we got FET-based biosensor with pristine surface. (b) Optical image of a 3-
inch wafer of In2O3 nanoribbon biosensor. (c) Magnified image of a biosensor chip with 5
nanoribbon devices in one group. (d) A SEM image of two In2O3 nanoribbon devices
(L=500 μm, W=25 μm). (e) An AFM image of a ~16 nm-thick In2O3 nanoribbon.
15
The electrical characterization of the devices was first carried out in ambient
environment by measuring the output and transfer characteristics as a function
of drain and back gate voltages. Figure 2.2a and 2.2b show family curves of
drain current-drain voltage (I
DS
-V
DS
) and drain current-gate voltage (I
DS
-V
GS
)
with drain voltage fixed at 1V, respectively. High back gate voltage is required
to turn on the device due to the presence of very thick back gate oxide. The
output characteristics of the FET devices illustrate n-type transistor behavior
with good saturation, and the In
2
O
3
FETs show high field-effect mobilities
(μ
sat
) of 13.09 ± 1.39 cm
2
V
-1
S
-1
(averaged over 50 devices) and on/off ratios
(I
on
/I
off
) above 10
7
.
The edge of devices fabricated using shadow masks is not
as sharp as the edge of devices fabricated using photolithography. However,
the shadow masks we used have small thickness of 125 μm and can be
attached to the Si/SiO
2
wafer with no visible gap in between. In addition, the
sputtering system we use has a rather long working distance ~ 15 cm between
the sputtering target and the substrate, make the “shadow” due to the edges of
the shadow mask almost negligible. We statistically analyzed 50 devices
randomly selected from a wafer, and as the result shown in Figure 2.3, our
devices showed very good uniformity, which is almost comparable to the
uniformity of previous devices made using photolithography.
16
Figure 2. 2 Electrical characterization of In2O3 nanoribbon biosensor. (a) Family of IDS-VDS
curves measured in ambient environment. Drain current as a function of drain voltage with
the back gate voltage varying from 0 to 50 V in steps of 10 V. (b) Drain current versus
back gate voltage with drain voltage fixed at 1 V. Current was plotted in logarithmic scale
in left axis and in linear scale in right axis. (c) Family curves of IDS-VDS measured in 0.01
x PBS with liquid gate varying from 1 V to 0.5 V in steps of 0.1 V. (d) Drain current versus
liquid gate voltage with drain voltage fixed at 1 V, also plotted in linear and logarithmic
scale. (e) Change in threshold voltage with pH range from 5 to 10, and obtained a pH
sensitivity of ~ 60.5 mV/pH. (f) Real-time responses obtained from an In2O3 nanoribbon
device exposed to commercial buffer solutions with pH 5 to 10.
17
For biosensing applications, it is necessary that these devices can be operated
in a wet environment. Hence, the devices were measured with the active
channel materials immersed in a micro well filled with electrolyte solution
(0.01 x Phosphate Buffered Saline (PBS)). A Ag/AgCl reference electrode is
used to apply bias to the electrolyte to stably operate the biosensor, which is
referred to as a liquid gate. The performance of liquid-gated In
2
O
3
FETs is
shown in Figure 2.2c (I
DS
-V
DS
) and 2.2d (I
DS
-V
GS
). It illustrates that the
biosensor device is efficiently controlled in the wet environment, and the
In
2
O
3
FETs have good FET behavior with saturation and low driving voltage.
Figure 2. 3 Histogram of the mobilities of 50 devices randomly selected from a wafer,
showing the mobility = 13.09 ± 1.39 cm
2
V
-1
S
-1
18
In order to determine the pH sensitivity of the In
2
O
3
FETs, we selected six
devices randomly from the wafer, and recorded their response to pH solutions.
The pH sensing is based on the protonation/deprotonation of the OH groups
on the surface due to the pH of the electrolyte, and resultant changes in local
FET electric fields, which cause changes in the conductance and current. The
shift in threshold voltage was calculated using the extrapolation in the
saturation region, and found to be 60.5 ± 2.44 mV/pH at room temperature
(Figure 2.2e), close to the ideal result of 59.1 mV/pH at 25 °C.
27
Figure 2.2f
shows the real-time sensing response of an unfunctionalized In
2
O
3
FET to
standard pH calibration solutions. The initial current I
o
was obtained by using
PBS to stabilize the device, and then the PBS buffer was sequentially changed
to commercial pH buffer solutions ranging from pH 10 to pH 5. The drain
current responded quickly and log-linearly to each pH buffer.
2.3 Quick detection of acute myocardial infarction
biomarkers using electronic enzyme-linked
immunosorbent assay
Direct electrical detection of biomolecules in their physiological environment
is often impeded by the Debye screening from the high salt concentration in
the sample solutions.
28
Sandwich enzyme-linked immunosorbent assay
19
(ELISA), on the other hand, detects signals associated with the reactions
between a test solution and the conjugated enzymes on secondary antibodies
instead of the biomarker. The sandwiched structure not only overcomes the
Debye screening from salts in the fluid, but also incorporates an amplification
scheme to improve the signal-to-noise ratio (SNR), which can be much higher
than direct analyte detection without amplification, especially when the
amount of analytes is small.
In the following In
2
O
3
nanoribbon sensing experiments, we applied an
electronic ELISA technique that uses pH change due to urease enzyme
activities as the amplification signal. Figure 3a and 3b show the schematic
diagram depicting the electronic ELISA process. Prior to using In
2
O
3
FET
biosensors for biomarker detection, the surfaces were treated with phosphonic
acid to confer phosphonic linker molecules to the indium oxide surface.
Subsequently, we functionalized our devices with N-(3-
Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride/ N-
Hydroxysuccinimide (EDC/NHS) chemistry to immobilize the capture
antibodies on the surface of In
2
O
3
FETs, as mentioned in the Method section
and in our previous papers.
15, 29, 30
This was followed by a washing step that
removed unbound capture antibodies (all binding steps described below were
20
followed by three times washes). A bovine serum albumin (BSA) solution was
used to prevent nonspecific protein adsorption to the chip and reservoir
sidewalls, which is a typical blocking step used in conventional colorimetric
ELISA protocols to minimize nonspecific binding.
31
This was followed by
introducing known concentrations of the antigen-containing samples to the
sensor for antigen-antibody binding (Figure 2.4a). The biomarkers were
contained either within the physiological fluid sample of the patient or in a
solution of buffer for experimental purposes. The biomarkers were
subsequently captured by the antibodies, and any unbound ones were washed
off. Next a solution of biotinylated secondary antibodies, which are also
specific to the cardiac biomarker, was introduced to the sensors using
incubation, and the secondary antibodies bound to the biomarkers. After
rinsing out unbound biotinylated antibody, streptavidin solution in PBS was
introduced. The biotin end of the secondary antibody group was used to bind
to a streptavidin, which in turn was bound to a biotinylated urease, the last
solution to incubate the sensor.
When a solution of urea is introduced to the nanoribbon sensor surface with
this sandwich structure (Figure 2.4b), the urea causes an increase in the pH of
the solution due to consumption of protons according to the following reaction.
21
𝑈𝑟𝑒𝑎 +2𝐻 2
𝑂 +𝐻 +
𝑈𝑟𝑒𝑎𝑠𝑒 → 2𝑁𝐻
4
+
+𝐻𝐶𝑂 3
−
The urease reaction raises the pH of the solution, leading to deprotonation the
surface hydroxyl groups on the In
2
O
3
nanoribbon, ultimately lowering the
surface potential. The increase in negative surface charges is responsible for
the decrease in conduction of the n-type In
2
O
3
nanoribbon FETs. The pH
change is easily detected by the In
2
O
3
nanoribbon sensors because the
catalytic reaction promoted by urease amplifies the charge generated by the
binding of the analyte by orders of magnitude over the direct binding between
the antigens and the capture antibodies.
14, 30
This amplifies the detection signal,
allowing the sensor to detect very low concentrations of the antigen.
Furthermore, the solution for the pH detection step is independent of the fluid
containing the biomarker, since the solutions are rinsed out after each step.
This allows cardiac biomarkers to be collected in physiological samples such
as whole blood without the limitation of the Debye screening effect, which
complicates detection schemes based solely on direct binding between the
antigens and the capture antibodies.
15
22
Figure 2. 4 In2O3 nanoribbon biosensor and electronic ELISA for cardiac Troponin I (cTnI),
Creatine kinase-MB (CK-MB) and B-type natriuretic peptide (BNP) detection. (a)
Schematic diagram of the antibodies capture specific antigens in buffer solutions. (b)
Schematic illustrate the electronic ELISA sensing setup. (c) Real-time sensing results of 1
pg/ml, 10 pg/ml and 300 pg/ml of cTnI antigens in 1 x PBS buffer. (d) Average sensing
results of 3 devices from 3 concentrations of cTnI proteins in 1 x PBS buffer marked as
black square and 1 concentration of troponin I in diluted human whole blood marked as
red dot. Error bars represent standard deviations of 3 devices. (e) Real-time sensing results
of 0.1 ng/ml, 1ng/ml and 3 ng/ml of CK -MB proteins in 1 x PBS buffer. (f) Average
sensing responses from different concentrations of CK-MB. (g) Real-time responses of 10
pg/ml, 50 pg/ml and 90 pg/ml of BNP proteins in 1 x PBS buffer. (h) BNP biomarker
23
concentrations versus signal for 3 concentrations of BNP in 1 x PBS and 1 concentration
in diluted human whole blood.
Troponin, a Food and Drug Administrative (FDA) approved biomarker for
AMI, is the biomarker of choice for evaluating chest pain patients for possible
heart attack. Troponin I and T are released to blood streams due to the death
of cardiac muscle cells; therefore, troponin I and T are not present in the blood
of healthy people. Elevated blood troponin levels have a positive correlation
to the risk of death in the heart disease patients, and the biomarker is a good
guide for identifying patients for certain types of treatment.
32, 33
The 99th
percentile of a reference decision limit (medical decision cutoff) for cardiac
troponin (cTn) assays is over 40 pg/ml.
34
In the first biomarker detection
experiment, we used troponin I as the model cardiac biomarker to demonstrate
that the In
2
O
3
nanoribbon biosensors can be used to optimize the electronic
ELISA assay turnaround time by shortening the incubation of the cardiac
biomarkers to 30 minutes in total. The incubation time of target analytes,
biotinylated secondary antibodies, streptavidin and biotinylated urease
enzymes were 10, 10, 5, and 5 minutes, respectively.
We performed experiments with known concentration of cardiac Troponin I
(cTnI) in 1 x PBS, namely 1, 10 and 300 pg/ml, to build a standard curve
covering the beginning of the second quartile for non-AMI patients to the
median of AMI patients.
34
At time t = 0 in Figure 2.4c, the devices were rinsed
24
with and submerged in 0.01 x PBS buffer when the baseline current was taken.
The buffer was then replaced with 10 mM urea in 0.01 x PBS around 200s as
indicated by the arrow. It shows the real-time responses from 3 sensors when
the urea solution was introduced into the sensing chamber that was previously
incubated in 1 pg/ml, 10 pg/ml and 300 pg/ml of cardiac troponin I (cTnI) in
100 μl of 1 x PBS buffer. The urease-urea interaction drastically reduces the
device conductance by 25.3%, 42.5% and 69.5% of the baseline signal,
respectively. Figure 2.4d shows the average normalized responses for each
cTnI concentration mentioned earlier. Each data point was calculated from
three sensors monitored simultaneously during the experiment. The sensing
response decreases exponentially upon the decrease in the concentration of
cTnI target molecules. The current of the In
2
O
3
nanoribbon sensor drops to
about 42% of the baseline at a troponin concentration of 10 pg/ml and 25% at
a concentration of 1 pg/ml. The sensitivity corresponds to about 17%
conduction change per decade of biomarker concentration change. This is
beneficial for covering a large range of concentrations for biomarkers like
cTnI whose elevation in AMI patients is high. The pH changes between the
buffer solutions used for the baseline and the final solutions in the sensing
chamber was measured to be 0.17, 0.87 and 2.17, respectively, by a
commercial Mettler Toledo pH meter. These increases in pH are consistent
25
with the decreases in conduction of the In
2
O
3
nanoribbon devices. Moreover,
the total sample-collection-to-result time is around 45 mins, meeting the
expectation of 1 hour for practical use in diagnosis of myocardial infarction,
In addition to cTnI, the blood biomarker Creatine kinase-MB (CK-MB) has
long been used for AMI detection. Including the detection of CK-MB can
improve early diagnosis of AMI, since the level of CK-MB increases within
2 to 4 hours after cardiac muscle injury.
35
The CK-MB level in the blood is
relatively high compared to other biomarkers, with an interquartile range level
of non-AMI patients at 0.6 ng/ml to 1.7 ng/ml, and that of AMI patients from
1.5 ng/ml to 10.5 ng/ml.
35
Thus, the detection of CK-MB must be able to
distinguish the concentration change less than one order of magnitude for
effective diagnosis. We repeated the sensing for 0.1, 1 and 3 ng/ml, and the
results were shown in Figure 2.4e. Heart failure is strongly indicated when the
blood sample has 30 ng/ml CK-MB before dilution or 3 ng/ml with 10 x
dilution. The average of data from 3 sensors for each concentration is plotted
in Figure 2.4f with standard deviations plotted as the error bars. At 0.1 ng/ml,
the current is ~ 33% of the baseline, and at 1 ng/ml, the current is ~ 60% of
the baseline, yielding a difference equivalent to 27% of the baseline for a
concentration difference of a decade. This large sensing response enable
26
detection of minute changes in concentration, such as from 250 pg/ml to 300
pg/ml or 2.5 ng/ml to 3 ng/ml before 10 x dilution. In addition, the small
device-to-device signal standard deviation makes readout at this precision
possible using the In
2
O
3
nanoribbon sensor platform.
Beside cTnI and CK-MB, B-type natriuretic peptide (BNP) is also associated
with heart failure and has been shown to substantially improve AMI diagnosis
when included in a multiple cardiac biomarker panel.
36
More importantly, for
blood samples taken when chest pain patients first arrive at the emergency
department, BNP is shown to have quicker response for AMI diagnosis than
other cardiac biomarkers such as CK-MB and troponin, which do not elevate
until at least 2 hours after the onset of AMI symptoms.
37
In fact, even when a
patient’s troponin level is normal, a BNP concentration greater than 100 pg/ml
is a good indicator of AMI. BNP higher than 900 pg/ml is considered severe
heart failure.
38
In order to simulate quantitative detection of BNP in patients’
blood, we first obtained a standard calibration curve with known
concentrations of BNP in buffer. As shown in Figure 2.4g, we have targeted
BNP concentrations of 10 pg/ml, 50 pg/ml, and 90 pg/ml. For each of the 3
concentrations, 3 In
2
O
3
nanoribbon sensors were used in the electronic ELISA
assay as described in the previous sections. For the smallest concentration of
27
10 pg/ml, the current drops to ~ 60% of the baseline after the introduction of
10mM Urea solution. For the 50 pg/ml and the 90 pg/ml detection, the current
drops to 72% and 77% of the baseline, respectively. The average and the
standard deviation for each of the three concentrations are plotted in Figure
2.4h. We also performed the orthogonal tests for the three biomarkers to verify
that our technology can distinguish these different biomarkers, and the results
can be found in the Supporting Information.
Detection of cardiac biomarkers in whole blood is essential to POC sensor
platforms used for situations where complicated patient blood processing is
not possible and defeats the purpose of fast, cheap, and convenient disease
testing. The main problems for FET sensor detection caused by whole blood
are the nonspecific binding of non-target proteins and the Debye length
screening from salts. Recent efforts to process whole blood for FET sensors
have been demonstrated using a microfluidic chip,
39
desalting columns
40
, and
filtration.
15
In this work, we have made some improvements and demonstrated
that by applying electronic ELISA assay on In
2
O
3
nanoribbon sensors, we can
detect cardiac biomarkers such as cTnI, CK-MB, BNP in whole blood without
any sample processing at all, as described below.
28
Cardiac biomarkers, cTnI, CK-MB and BNP, are spiked with healthy human
whole blood (purchased from Innovative Research) to simulate an AMI
patient sample with a cTnI concentration 100 pg/ml, a CK-MB concentration
3ng/ml and a BNP concentration of 300 pg/ml, all indicating mild heart failure.
Due to the fact that the human whole blood is very viscous, which may affect
the sensing results, a real patient sample would be first diluted 10 times to be
detected by the nanoribbon sensor. We note that this sample dilution is not
due to the difficulties in ionic screening and does not filter out any non-
specific proteins or blood cells. To simulate this sample preparation, 100 μl
of healthy whole blood was first diluted with 1 x PBS to 930 μl. Then 10 μl
of 1 ng/ml cTnI, 50 μl of 6 ng/ml CK-MB and 10 μl of 3 ng/ml BNP in 1 x
PBS was added to the diluted whole blood to simulate 10 pg/ml cTnI, 0.3
ng/ml CK-MB, 30 pg/ml BNP in 10 x diluted whole blood. This sample was
used to incubate the nanoribbon sensors prepared with capture antibodies, as
shown in Figure 2.5a. The remaining steps of the electronic ELISA assay
followed those described in previous sections. Figure 2.5b, 2.5c and 2.5d
show the real-time signal when 10mM of urea in 0.01 x PBS is introduced to
each sensor. The current drop is 40.6% for 10 pg/ml cTnI, 43.3% for 0.3 ng/ml
CK-MB and 67.5% for 30 pg/ml BNP. In Figure 2.4d, 2.4f and 2.4h, the
average responses of the 3 In
2
O
3
nanoribbon sensors for this detection are
29
placed on the calibration curve as red dots. The graph shows that the deviation
of the detection signal from the calibration curve are all below 5%. This falls
within the device-to-device variation and is expected for the experiment.
Figure 2. 5 Real-time sensing results of cardiac biomarkers in diluted human whole blood.
(a) Schematic diagram of antibodies that were anchored on the biosensor surfaces capture
specific biomarkers in diluted human whole blood. (b) Real-time sensing results of 10
pg/ml of cTnI in 10 x diluted human whole blood, and the averaged results were plotted as
red dot in Figure 2.4d. (c) Real-time sensing results of 300 pg/ml of CK-MB in 10 x diluted
human whole blood, and the results were plotted as red dot in Figure 2.4f. (d) Real-time
sensing results of 30 pg/ml of BNP in 10 x diluted human whole blood, and the results
were plotted as red dot in Figure 2.4h.
Since the cardiac biomarkers concentration elevate in AMI patient, it is
important for diagnosis and treatment to obtain the trend in the cardiac
30
biomarker concentration in the hours after patient’s arrival. The reusability of
the In
2
O
3
nano-biosensor can give results every hour when the sensing is
repeated. What’s more, the reusability of the biosensors is also cost-effective.
Following the regeneration process of antibodies and antigens
41
, we applied
50 mM NaOH as regeneration buffer (from GE healthcare) to a sensor that
had already been used for cTnI biomarker sensing. As shown in Figure 2.6a,
when rinsing the sensors with washing buffer, the proteins will anti-bind the
capture antibody. After rinsing with 1 x PBS buffer, the proteins will all be
washed away, leaving the antibody still active and bond to the surface of the
sensor. To demonstrate the sensors can still work well after washing, we start
from incubation with samples containing cTnI biomarkers and repeated the
electronic ELISA process as before. Figure 2.6b, 2.6c and 2.6d show the
sensing results for the first time, the second time and the fifth time,
respectively. They all fall around 60%, indicating the regeneration process can
efficiently wash proteins away and leaves sufficient capture antibodies for
reusability.
31
Figure 2. 6 Reusability of In2O3 nanoribbon biosensors. (a) Schematic diagram of
regeneration process by applying 50 mM NaOH to the used sensor surface. After antibodies
release antigens, the sensors are ready to repeat sensing. (b) Real-time responses of 100
pg/ml cTnI proteins in 1 x PBS buffer. (c) Real-time sensing responses from the same
concentration of cTnI and the same devices after regeneration. (d) Real-time response of
the same sensors after 3 more cycles of regeneration and sensing process.
We also performed the orthogonal tests for the three biomarkers to verify that
our technology can distinguish these different biomarkers. The sensor chips
were functionalized with cTnI antibodies on the surface as described before,
and then each chip was incubated in 100 μL of 1 x PBS buffer spiked with 10
pg/ml of cTnI protein solution, 1 ng/ml CK-MB protein solution and 50 pg/ml
BNP protein solution, respectively. After that, we incubated the sensor chip
sequentially with biotinylated cTnI antibodies, streptavidin, and biotinylated
32
urease. The results (Figure 2.7) showed that the sensor chips functionalized
with cTnI antibodies only responded to cTnI proteins and there was no current
drop for CK-MB and BNP proteins. We also did similar experiments for
sensor chips functionalized with CK-MB antibodies and BNP antibodies. The
results (Figure 2.8 and 2.9) showed that the functionalized sensor chips only
responded to the specific proteins. This is consistent with the fact that the
antibodies and antigens have specific binding, and our design can effectively
minimize the false positive results caused by unspecific binding.
Figure 2. 7 Real-time sensing results from sensor chips functionalized with Troponin I
antibody. (a) Incubated in Troponin I antigen solution with concentration of 10 pg/ml. (b)
Incubated in CK-MB antigen solution with concentration of 1 ng/ml. (c) Incubated in BNP
antigen solution with concentration of 50 pg/ml.
Figure 2. 8 Real-time sensing results from sensor chips functionalized with CK-MB
antibody. (a) Incubated in CK-MB antigen solution with concentration of 1 ng/ml. (b)
33
Incubated in Troponin I antigen solution with concentration of 10 pg/ml. (c) Incubated in
BNP antigen solution with concentration of 50 pg/ml.
Figure 2. 9 Real-time sensing results from sensor chips functionalized with BNP antibody.
(a) Incubated in BNP antigen solution with concentration of 50 pg/ml. (b) Incubated in
Troponin I antigen solution with concentration of 10 pg/ml. (c) Incubated in CK-MB
antigen solution with concentration of 1 ng/ml.
We note that while this paper focuses on In
2
O
3
nanoribbon sensors, our
approach is not limited to In
2
O
3
and can be applied to many other materials,
such as other kinds of semiconducting metal oxides, as long as the
semiconducting material can be deposited through a shadow mask, including
InGaZnO, ZnO, InN,
42, 43
and SnO
2
. In our previous work,
14
we studied several
possible materials, like In
2
O
3
, InGaZnO, SnO
2
, ZnO, and tin-doped indium-
oxide (ITO). In the study,
14
InGaZnO and SnO
2
showed higher resistance,
higher threshold voltage and lower on/off ratio than the In
2
O
3
samples, ITO
showed poor gate dependence, and ZnO showed poor stability in aqueous
environment. CdO, NiO and other possible materials may also be suitable to
work as channel materials, which need further investigation.
34
In a production setting, further improvements can be made for even better
uniformity by monitoring the nanoribbon film thickness after sputtering and
chemical modification to reduce the device-to-device variation down a
fraction of a percentage. Such highly uniform batches of sensors can give
good statistical confidence for their reported biomarker concentrations. This
confidence level combined with a turnaround time of 45 minutes is a good
basis for improving current POC devices for cardiac marker detection in an
emergency situation. Moreover, the platform can be integrated with other
electronic components for better data analysis.
35
2.4 Summary
In conclusion, we have demonstrated the fabrication of highly uniform and
scalable In
2
O
3
nanoribbon biosensor chips using two simple shadow masks to
define the position and dimension of metal electrodes and nanoribbons, and
the devices have shown outstanding performance. Furthermore, In
2
O
3
nanoribbon devices show good electrical performance in the aqueous
condition when gate voltage is applied through the liquid gate electrode. In
addition, the In
2
O
3
nanoribbon devices show good performance in pH sensing
experiment with change in conduction by a factor of 12 when pH is reduced
from 10 to 5. Through all the sensing experiments, we have demonstrated that
In
2
O
3
nanoribbon biosensors fabricated by shadow masks can be used to
quantitatively detect 3 cardiac biomarkers within the concentrations relevant
to clinical diagnosis with the turnaround time ~ 45 minutes. We further
demonstrated tests using spiked cardiac biomarkers in diluted human whole
blood, with results consistent with the calibration curve established using PBS
buffer. Lastly, by applying a regeneration buffer to the used sensor surfaces
to anti-bond the antigen-antibody conjugation and repeating the sensing
experiments, we demonstrated the reusability of the In
2
O
3
nanoribbon
biosensors with very small variation of each sensing results.
36
2.5 References
1. Nourjah, P., National Hospital Ambulatory Medical Care Survey: 1997 Emergency
Department Summary. Advance Data from Vital and Health Statistics 1999, 304, 1-24.
2. Gibler, W. B.; Young, G. P.; Hedges, J. R.; Lewis, L. M.; Smith, M. S.; Carleton, S.
C.; Aghababian, R. V .; Jorden, R. O.; Allison, E. J.; Often, E. J., Acute Myocardial
Infarction in Chest Pain Patients with Nondiagnostic Ecgs: Serial CK-MB Sampling in the
Emergency Department. Annals of Emergency Medicine 1992, 21, 504-512.
3. Brush Jr, J. E.; Brand, D. A.; Acampora, D.; Chalmer, B.; Wackers, F. J., Use of the
Initial Electrocardiogram to Predict in-Hospital Complications of Acute Myocardial
Infarction. N. Engl. J. Med. 1985, 312, 1137-1141.
4. Lee, T. H.; Rouan, G. W.; Weisberg, M. C.; Brand, D. A.; Cook, E. F.; Acampora,
D.; Goldman, L., Sensitivity of Routine Clinical Criteria for Diagnosing Myocardial
Infarction within 24 Hours of Hospitalization. Ann. Intern. Med. 1987, 106, 181-6.
5. Rude, R. E.; Poole, W. K.; Muller, J. E.; Turi, Z.; Rutherford, J.; Parker, C.; Roberts,
R.; Raabe, D. S.; Gold, H. K.; Stone, P. H.; Willerson, J. T.; Braunwald, E.,
Electrocardiographic and Clinical-Criteria for Recognition of Acute Myocardial-Infarction
Based on Analysis of 3,697 Patients. Am. J. Cardiol. 1983, 52, 936-941.
6. Braunwald, E.; Antman, E. M.; Beasley, J. W.; Califf, R. M.; Cheitlin, M. D.;
Hochman, J. S.; Jones, R. H.; Kereiakes, D.; Kupersmith, J.; Levin, T. N.; Pepine, C. J.;
Schaeffer, J. W.; Smith, E. E., 3rd; Steward, D. E.; Theroux, P.; Gibbons, R. J.; Alpert, J.
S.; Eagle, K. A.; Faxon, D. P.; Fuster, V .; el al., Acc/Aha Guidelines for the Management
of Patients with Unstable Angina and Non-St-Segment Elevation Myocardial Infarction:
Executive Summary and Recommendations. A Report of the American College of
Cardiology/American Heart Association Task Force on Practice Guidelines (Committee on
the Management of Patients with Unstable Angina). Circulation 2000, 102, 1193-209.
7. Ng, S. M.; Krishnaswamy, P.; Morissey, R.; Clopton, P.; Fitzgerald, R.; Maisel, A.
S., Ninety-Minute Accelerated Critical Pathway for Chest Pain Evaluation. Am. J. Cardiol.
2001, 88, 611-617.
37
8. Lewandrowski, K.; Chen, A.; Januzzi, J., Cardiac Markers for Myocardial
Infarction. A Brief Review. Am. J. Clin. Pathol. 2 002, 118 Suppl, S93-9.
9. Han, J. H.; Gibler, W. B., Biomarkers in the Emergency Department: Rapid
Diagnosis and Triage. In Cardiovascular Biomarkers: Pathophysiology and Disease
Management, Morrow, D. A., Ed. Humana Press: 2006, pp 61-77.
10. Wu, A. H. B.; Apple, F. S.; Gibler, W. B.; Jesse, R. L.; Warshaw, M. M.; Valdes, R.,
National Academy of Clinical Biochemistry Standards of Laboratory Practice:
Recommendations for the Use of Cardiac Markers in Coronary Artery Diseases. Clin.
Chem. 1999, 45, 1104-1121.
11. Lee-Lewandrowski, E.; Januzzi, J. L., Jr.; Grisson, R.; Mohammed, A. A.;
Lewandrowski, G.; Lewandrowski, K., Evaluation of First-Draw Whole Blood, Point-of-
Care Cardiac Markers in the Context of the Universal Definition of Myocardial Infarction:
A Comparison of a Multimarker Panel to Troponin Alone and to Testing in the Central
Laboratory. Arch. Pathol. Lab. Med. 201 1, 135, 459-63.
12. Mu, L.; Droujinine, I. A.; Rajan, N. K.; Sawtelle, S. D.; Reed, M. A., Direct, Rapid,
and Label-Free Detection of Enzyme–Substrate Interactions in Physiological Buffers
Using Cmos-Compatible Nanoribbon Sensors. Nano lett. 2014, 14, 5315-5322.
13. Kong, T.; Su, R.; Zhang, B.; Zhang, Q.; Cheng, G., Cmos-Compatible, Label-Free
Silicon-Nanowire Biosensors to Detect Cardiac Troponin I for Acute Myocardial Infarction
Diagnosis. Biosens. Bioelectron. 2012, 34, 267-272.
14. Aroonyadet, N.; Wang, X.; Song, Y .; Chen, H.; Cote, R. J.; Thompson, M. E.; Datar,
R. H.; Zhou, C., Highly Scalable, Uniform, and Sensitive Biosensors Based on Top-Down
Indium Oxide Nanoribbons and Electronic Enzyme-Linked Immunosorbent Assay. Nano
lett. 2015, 15, 1943-1951.
15. Chang, H.-K.; Ishikawa, F. N.; Zhang, R.; Datar, R.; Cote, R. J.; Thompson, M. E.;
Zhou, C., Rapid, Label-Free, Electrical Whole Blood Bioassay Based on Nanobiosensor
Systems. ACS nano 201 1, 5, 9883-9891.
16. Kim, J.; Rim, Y . S.; Chen, H.; Cao, H. H.; Nakatsuka, N.; Hinton, H. L.; Zhao, C.;
38
Andrews, A. M.; Yang, Y.; Weiss, P. S., Fabrication of High-Performance Ultrathin In2O3
Film Field-Effect Transistors and Biosensors Using Chemical Lift-Off Lithography. ACS
nano 2015, 9, 4572-4582.
17. Li, C.; Lei, B.; Zhang, D.; Liu, X.; Han, S.; Tang, T.; Rouhanizadeh, M.; Hsiai, T.;
Zhou, C., Chemical Gating of In2O3 Nanowires by Organic and Biomolecules. Appl. Phys.
Lett. 2003, 83, 4014-4016.
18. Lei, B.; Li, C.; Zhang, D.; Zhou, Q.; Shung, K.; Zhou, C., Nanowire Transistors
with Ferroelectric Gate Dielectrics: Enhanced Performance and Memory Effects. Appl.
Phys. Lett. 2004, 84, 4553-4555.
19. Gao, W.; Emaminejad, S.; Nyein, H. Y . Y .; Challa, S.; Chen, K.; Peck, A.; Fahad,
H. M.; Ota, H.; Shiraki, H.; Kiriya, D., Fully Integrated Wearable Sensor Arrays for
Multiplexed in Situ Perspiration Analysis. Nature 2016, 529, 509-514.
20. Cui, Y .; Wei, Q.; Park, H.; Lieber, C. M., Nanowire Nanosensors for Highly
Sensitive and Selective Detection of Biological and Chemical Species. Science 2001, 293,
1289-1292.
21. Goldsmith, B. R.; Mitala, J. J.; Josue, J.; Castro, A.; Lerner, M. B.; Bayburt, T. H.;
Khamis, S. M.; Jones, R. A.; Brand, J. G.; Sligar, S. G.; Luetje, C. W.; Gelperin, A.; Rhodes,
P. A.; Discher, B. M.; Johnson, A. T. C., Biomimetic Chemical Sensors Using
Nanoelectronic Readout of Olfactory Receptor Proteins. ACS nano 201 1, 5, 5408-5416.
22. Chen, P. C.; Sukcharoenchoke, S.; Ryu, K.; Gomez de Arco, L.; Badmaev, A.; Wang,
C.; Zhou, C., 2, 4, 6‐Trinitrotoluene (Tnt) Chemical Sensing Based on Aligned Single‐
Walled Carbon Nanotubes and Zno Nanowires. Adv. Mater. 2010, 22, 1900-1904.
23. Naylor, C. H.; Kybert, N. J.; Schneier, C.; Xi, J.; Romero, G.; Saven, J. G.; Liu, R.;
Johnson, A. T. C., Scalable Production of Molybdenum Disulfide-Based Biosensors. ACS
nano 2016, 10, 6173-6179.
24. Star, A.; Gabriel, J.-C. P.; Bradley, K.; Grüner, G., Electronic Detection of Specific
Protein Binding Using Nanotube Fet Devices. Nano lett. 2003, 3, 459-463.
25. Zheng, G.; Patolsky, F.; Cui, Y .; Wang, W. U.; Lieber, C. M., Multiplexed Electrical
39
Detection of Cancer Markers with Nanowire Sensor Arrays. Nat. Biotechnol. 2005, 23,
1294-1301.
26. Chen, R. J.; Bangsaruntip, S.; Drouvalakis, K. A.; Kam, N. W. S.; Shim, M.; Li, Y .;
Kim, W.; Utz, P. J.; Dai, H., Noncovalent Functionalization of Carbon Nanotubes for
Highly Specific Electronic Biosensors. Proc. Natl. Acad. Sci. 2003, 100, 4984-4989.
27. Sarkar, D.; Liu, W.; Xie, X.; Anselmo, A. C.; Mitragotri, S.; Banerjee, K., Mos 2
Field-Effect Transistor for Next-Generation Label-Free Biosensors. ACS nano 2014, 8,
3992-4003.
28. Butler, J. E., Enzyme-Linked Immunosorbent Assay. J. Immunoassay 20 00, 21,
165-209.
29. Li, C.; Curreli, M.; Lin, H.; Lei, B.; Ishikawa, F. N.; Datar, R.; Cote, R. J.;
Thompson, M. E.; Zhou, C., Complementary Detection of Prostate-Specific Antigen Using
In2O3 Nanowires and Carbon Nanotubes. J. Am. Chem. Soc. 2005, 127, 12484-5.
30. Stern, E.; Vacic, A.; Li, C.; Ishikawa, F. N.; Zhou, C.; Reed, M. A.; Fahmy, T. M.,
A Nanoelectronic Enzyme-Linked Immunosorbent Assay for Detection of Proteins in
Physiological Solutions. Small 2010, 6, 232-8.
31. Hermanson, G. T., Bioconjugate Techniques. Academic Press: 2013.
32. Newby, L. K.; Christenson, R. H.; Ohman, E. M.; Armstrong, P. W.; Thompson, T.
D.; Lee, K. L.; Hamm, C. W.; Katus, H. A.; Cianciolo, C.; Granger, C. B.; Topol, E. J.;
Califf, R. M.; Investigators, G.-I., Value of Serial Troponin T Measures for Early and Late
Risk Stratification in Patients with Acute Coronary Syndromes. Circulation 1998, 98,
1853-1859.
33. Ohman, E. M.; Armstrong, P. W.; Christenson, R. H.; Granger, C. B.; Katus, H. A.;
Hamm, C. W.; OHanesian, M. A.; Wagner, G. S.; Kleiman, N. S.; Harrell, F. E.; Califf, R.
M.; Topol, E. J., Cardiac Troponin T Levels for Risk Stratification in Acute Myocardial
Ischemia. N. Engl. J. Med. 1996, 335, 1333-1341.
34. Mahajan, V . S.; Jarolim, P., How to Interpret Elevated Cardiac Troponin Levels.
Circulation 201 1, 124, 2350-2354.
40
35. Saenger, A. K.; Jaffe, A. S., Requiem for a Heavyweight the Demise of Creatine
Kinase-MB. Circulation 2008, 118, 2200-2206.
36. Christenson, R. H.; Azzazy H. M. E., Biomarkers of Myocardial Necrosis. In
Cardiovascular Biomarkers: Pathophysiology and Disease Management, Morrow, D. A.,
Ed. Humana Press Inc.: Totowa, New Jersy, 2006, pp 3-25.
37. Bassan, R.; Potsch, A.; Maisel, A.; Tura, B.; Villacorta, H.; Nogueira, M. V .;
Campos, A.; Gamarski, R.; Masetto, A. C.; Moutinho, M. A., B-Type Natriuretic Peptide:
A Novel Early Blood Marker of Acute Myocardial Infarction in Patients with Chest Pain
and No St Segment Elevation. Eur. Heart J. 2005, 26, 234-240.
38. Morrison, L. K.; Harrison, A.; Krishnaswamy, P.; Kazanegra, R.; Clopton, P.;
Maisel, A., Utility of a Rapid B-Natriuretic Peptide Assay in Differentiating Congestive
Heart Failure from Lung Disease in Patients Presenting with Dyspnea. J. Am. Coll. Cardiol.
2002, 39, 202-209.
39. Stern, E.; Vacic, A.; Rajan, N. K.; Criscione, J. M.; Park, J.; Ilic, B. R.; Mooney, D.
J.; Reed, M. A.; Fahmy, T. M., Label-Free Biomarker Detection from Whole Blood. Nat.
Nanotechnol. 2010, 5, 138-42.
40. Zheng, G. F.; Patolsky, F.; Cui, Y .; Wang, W. U.; Lieber, C. M., Multiplexed
Electrical Detection of Cancer Markers with Nanowire Sensor Arrays. Nat. Biotechnol.
2005, 23, 1294-1301.
41. Duan, X.; Li, Y .; Rajan, N. K.; Routenberg, D. A.; Modis, Y .; Reed, M. A.,
Quantification of the Affinities and Kinetics of Protein Interactions Using Silicon
Nanowire Biosensors. Nat. Nanotechnol. 2012, 7, 401-407.
42. Cai, X.-M.; Hao, Y .-Q.; Zhang, D.-P.; Fan, P., Direct Current Magnetron Sputtering
Deposition of Inn Thin Films. Appl. Surf. Sci. 200 9, 256, 43-45.
43. Tang, T.; Han, S.; Jin, W.; Liu, X.; Li, C.; Zhang, D.; Zhou, C.; Chen, B.; Han, J.;
Meyyapan, M., Synthesis and Characterization of Single-Crystal Indium Nitride
Nanowires. J. Mater. Res. 2004, 19, 423-426.
41
Chapter 3 Highly sensitive and
wearable In
2
O
3
nanoribbon transistor
biosensors with integrated on-chip gate
for glucose monitoring in body fluids
3.1 Introduction
Wearable biosensors are smart electronic devices that can be worn on the body
as implant or accessories. Recent advances in microelectronics,
telecommunications and sensor manufacturing have opened up possibilities
for using wearable biosensors to continuously monitor an individual’s body
status without interrupting or limiting the user’s motions.
1-8
However, while
many commercially available wearable electronics can track users’ physical
activities, devices which can provide insightful view of user’s health status at
molecular level need more development. On the other hand, although some
commercial hand-held analyzers enable glucose or lactate detection, most of
these devices rely on blood samples.
9
Neither finger-prick nor invasive
sensors (such as a needle embedded under skin) is desired for wearable
biomedical applications. Continuous analyte monitoring, a key advantage
offered by wearable biosensors, has great potential in many cases. For
42
example, optimum diabetes management needs regular glucose monitoring,
and a trend of glucose level is more meaningful than an accurate data point.
10
Besides glucose monitoring, real-time detection of some pathogens in body
fluids can alert about the possible onset of certain diseases.
11
Although blood is by far the most understood sample for diagnosis, other
biological fluids such as sweat, tears, and saliva also contain tremendous
biochemical analytes which can provide valuable information, and are more
readily accessible compared to blood.
10, 12
Recent studies suggest a diagnosis
system based on the glucose concentration in body fluids to estimate blood
glucose levels.
13-15
However, many challenges still exist for the accurate
glucose sensing in body fluids.
16, 17
For example, the glucose levels in body
fluids are much lower than that in blood.
18
The sensing results can be affected
by ambient temperature changes, mechanical deformation caused by body
motion, and the sample collection procedure.
Among various types of sensors (optical, piezoelectric, and electrochemical
sensors, etc.), electrochemical sensors are the most promising candidate for
wearable technology owing to their high performance, portability, simplicity,
43
and low cost.
19-24
Considering the demands of wearable biosensors, the
selection of the sensing platform is critical to high sensitivity and
reproducibility, real-time detection, and simple integration with wearable
environments (e.g., human skin, tooth and eye).
11, 25
Nanobiosensors based on
indium oxide (In
2
O
3
) field-effect transistor (FET) are well suited for wearable
biosensor applications because of the quick response time for real-time and
continuous monitoring, large detectable concentration range, high sensitivity,
high uniformity for reliable sensing, and the capability to integrate with other
microfluidic and electronic functional groups.
26-30
Furthermore, the exposed
semiconductor channel regions can be modified with various functional
groups or receptors easily, and thus enable the In
2
O
3
nanobiosensors for
multiplexed sensing.
The current FET-based biosensor platform is usually comprised of individual
sensors with an external Ag/AgCl solution gate electrode to set the operational
point of the sensors to the optimal detection mode. In order to build a wearable
biosensor platform, a stand-alone sensor array is desired. The Ag/AgCl
electrode is commonly used as the reference electrode in the electrochemical
measurements and biosensing applications because it can provide stable
potential and can read signals precisely. However, the integration of the
44
Ag/AgCl electrode into a biosensor chip remains challenging. For FET-based
biosensors, the gate electrode only needs to supply stable gate bias to the
devices, which can be achieved by an on-chip metal side electrode. The
source-drain electrodes and the on-chip gate electrodes can be incorporated
into the straightforward 2-step shadow mask fabrication process so that no
additional fabrication steps are required. The integration of In
2
O
3
glucose
sensors with wearable electronics can generate high impact for diabetes
monitoring, since diabetes is one of the most prevalent and pressing diseases
in the world. Moreover, the development of wearable sensors for in-situ, real-
time, and low-cost detection of biologically and medically important targets
will generate broad impact in many applications involving electronic skin,
31
thermal regulation,
32
chemical sensing,
33
and the detection of pathogens in
body fluids.
11
Here, we demonstrate highly sensitive and conformal In
2
O
3
nanoribbon FET
biosensors with fully integrated on-chip gold side gate, which have been
laminated onto various surfaces, such as artificial arms and watches, and have
enabled glucose detection in various body fluids, such as sweat and saliva.
The devices are fabricated through two shadow masks. First, a shadow mask
is used to define the sputter-coating of In
2
O
3
nanoribbons, and the second
45
shadow mask is used for metal deposition of the source, drain and side gate.
The source and drain electrodes are modified with the enzyme glucose oxidase
(GOx), biocompatible polymer chitosan, and single-walled carbon nanotubes
(SWCNT) using ink-jet printing. The gold side gated In
2
O
3
FETs show good
electrical performance on highly flexible substrates. The optimized glucose
sensors show very wide detection range spanning at least 5 orders of
magnitude and detection limits down to 10 nM. The non-invasive glucose
detections in human body fluids, such as tears and sweat, and the sensing on
artificial skin and eye replicas are demonstrated. Therefore, this type of
devices are a highly sensitive platform for not only glucose detections but also
many other types of sensing applications.
3.2 Preparation process of flexible indium oxide
field-effect transistor array
The In
2
O
3
nanoribbon devices were fabricated following our previous
reported shadow mask fabrication technique,
26
however, this time we have
added side gate patterns to the source/drain shadow mask and have used 5 µm
ultraflexible PET substrate. Figure 3.1a illustrates the scheme for fabricating
flexible In
2
O
3
macroelectronics on PET substrates. First, a PET substrate was
46
attached to the first shadow mask using antistatic tape. Then we used radio
frequency (RF) sputtering to deposit 16-nm-thick In
2
O
3
nanoribbons through
the openings on the shadow mask. The second shadow mask was then
laminated onto the PET substrate for the following metal deposition. After
using a single mask to define the source, drain and side gate electrodes, the
as-made biosensor foil was peeled off from the shadow mask for further
electrical characterization. While most previous glucose sensing studies used
electrochemical sensors with large working electrodes by functionalization
the electrodes with drop casting.
2, 8
Here, we have developed an ink-jet
printing techniques to functionalize our FET In
2
O
3
glucose biosensors as
shown in Figure 3.1b. Due to the small dimension (~ 25µm × 500 µm) of our
nanoribbon biosensors, the traditional drop-cast deposition method leads the
whole active sensing area to be covered by the chitosan film. In order to keep
the channel area exposed, we employed a SonoPlot printer with a 50 µm glass
nozzle to print the chitosan ink only on the source and drain pads. The ink was
made of chitosan, single-walled carbon nanotube and glucose oxidase, and the
preparation of ink was described in detail in the Experimental Methods section.
Figure 3.1c shows a photograph of an as-fabricated In
2
O
3
biosensor foil with
a size of 5 cm × 5 cm. An Optical image of a group of In
2
O
3
biosensors and
two gold side gate electrodes are shown in Figure 3.1d.
47
Figure 3. 1 (a) Schematic diagrams showing the fabrication procedure of In 2O3 FETs on a
PET substrate using 2-step shadow masks. (b) Schematic diagrams of functionalization on
the surface of the electrodes using ink-jet printing. (c) Photograph of as-fabricated In2O3
FETs. Scale bar is 1 cm. (d) Optical image shows a group of In2O3 biosensors with two
gold side gate electrodes. Scale bar is 500 µm. (e) SEM image of an In 2O3 nanoribbon
devices (L=500 μm, W=25 μm) and a gold side gate electrode (W=150 μm). (f) Photograph
of In2O3 FET foil laminated on an artificial human hand. Scale bar is 3 cm. (g) In 2O3
Ink-jet nozzle
a
Sputtering of In
2
O
3
Metal evaporation
Chitosan/Carbon nanotubes/Glucose oxidase
b
c d
500 μm
25 μm In
2
O
3
Gold side gate
e
f
First shadow mask
Second shadow mask
h g
Gate
Gate
48
biosensor foil in a rolled-up state. Scale bar is 3 cm. (h) Photograph of an In2O3 FET chip
attached onto the back casing of a watch. Scale bar is 1 cm.
Figure 3.1e shows a scanning electron microscope (SEM) image of the
channel region and the gold side gate of a biosensor. To further characterize
the In
2
O
3
nanoribbons, we used atomic force microscopy (AFM) and X-ray
diffraction (XRD) on samples deposited on PET substrate (Figure S1). The
AFM images show that the nanoribbons are solid and have clear edges. The
height profile shows the thickness of In
2
O
3
nanoribbons is ~ 20 nm. The XRD
pattern shown presents only PET peaks, indicating the In
2
O
3
is amorphous.
Figure 1f shows the fabricated In
2
O
3
nanoribbon FET foil was conformably
laminated onto an artificial human hand, indicating the bendability and
wearability of the In
2
O
3
nanoribbon biosensors. Figure 1g exhibits the
biosensor foil rolled up with a radius of curvature of ~ 1 mm. The flexible
biosensor can be further attached onto the back casing of a watch (Figure 1h),
showing the concept that such In
2
O
3
transistor biosensors can be integrated
with smart watches in the future. We believe that flexible, lab-on-a-chip, and
conformal In
2
O
3
nanoribbon electronics are highly advantageous for wearable
biosensor applications.
49
3.3 Development of on-chip gold side gate for
flexible indium oxide field-effect transistors
In many previous studies, a Ag/AgCl electrode is commonly used as the
reference electrode in electrochemical measurements and biosensing
applications because it can provide stable potential and can also read voltage
precisely. However, the integration of the Ag/AgCl electrode onto a biosensor
chip increases the steps and difficulties of fabrication. Herein, we have
replaced Ag/AgCl external electrode with gold side gate to supply gate bias
to the devices. There are two gold side gate electrodes in a group of four In
2
O
3
FET devices. The one in the middle will replace the Ag/AgCl external liquid
gate to supply gate voltage, and the other one at the rear can be used to monitor
changes in potential on the devices. We first compared the device performance
with gate voltage applied by the external Ag/AgCl electrode and the on-chip
gold side electrode. Here, all the measurements were done when the device
active area was immersed into a microwell filled with 300 µL electrolyte
solution (0.1 x Phosphate Buffered Saline (PBS)). Figure 3.2a and 3.2b show
family curves of drain current-gate voltage (I
DS
-V
GS
) and drain current-drain
voltage (I
DS
-V
DS
) when the gate voltage was biased through a Ag/AgCl
electrode. The schematic diagram of the measurement setup is shown in the
inset of Figure 3.2a. The performance of gold side gate controlled In
2
O
3
FET
50
is presented in Figure 3.2c (I
DS
-V
GS
) and 3.2d (I
DS
-V
DS
). The output and
transfer curves of the FET devices demonstrate that the In
2
O
3
nanoribbon
devices can work properly under gate bias supplied by the gold side gate. The
output characteristics of the FET devices demonstrated Ohmic behavior with
a good linear regime in the “on” state, and the drain current got saturated when
the bias increased further. All the curves in Figure 3.2b and 3.2d passing
through the origin point indicate the minimal contribution of the gate leakage
current to the drain current. The field-effect mobility of the In
2
O
3
FET is
extracted to be 22.34 ± 1.44 cm
2
V
−1
s
−1
using the following equation:
g
m
=
dI
D
dV
GS
=
W
L
C
DL
𝜇 FE
V
D
(1)
Where W is the channel width, L is the channel length, and C
DL
is the electrical
double layer capacitance per unit area in 0.1 M ionic strength aqueous solution
reported previously (25.52 μF cm
−2
).
34
The maximum transconductance 5.69
μS was obtain at a drain voltage of 0.2 V and a gate voltage of 0.527 V (Figure
S2 in the Supporting Information). To further confirm the gate control of the
on-chip side gate electrode, we used one electrode as the gate bias supplier
and another one as a reference electrode to monitor the actual change of
potential on the devices, as the scheme shows in the inset of Figure 3.2e. In
Figure 3.2e, we plotted the reference voltage (V
REF
) against the gold side gate
51
voltage (V
GS
) with different distances between those two electrodes, 150 μm,
750 μm, and 1350 μm, respectively. It shows that V
REF
is almost identical to
V
GS
regardless of the distance. We further plotted drain current versus gate
bias applied through the gold side gate at difference distances (Figure 2f), and
the curves show negligible differences. A statistical study of key electrical
properties for 50 In
2
O
3
nanoribbon devices comparing gate biased through the
Ag/AgCl electrode and the gold side gate was conducted. Figure S3 in the
Supporting Information exhibits the device performance including mobility
(μ), threshold voltage (V
th
), on/off ration, and on-state current, is very similar
to each other, which implies that the gold side gate and the Ag/AgCl gate can
give analogous gating effect. From all these figures of merit, we can conclude
that the on-chip gate electrode has a great control over the nanoribbon
transistors in the aqueous environment.
52
Figure 3. 2 Electrical characterization of In2O3 nanoribbon biosensors using gold side gate
electrodes. (a) Drain current versus Ag/AgCl gate voltage with drain voltage changing from
0.4 V to 0 V in steps of 0.1 V. Inset shows the schematic diagram of the measurement setup.
(b) Family of IDS-VDS curves measured with a Ag/AgCl gate electrode. (c) Drain current
versus gold side gate voltage with drain voltage varying from 0.4 V to 0 V in steps of 0.1V.
(d) Family curves of IDS-VDS measured with gold side gate voltage varying from 1 V to 0
V in steps of 0.2V. (e) Reference voltage measured with a gold electrode versus the gold
side gate voltage. (f) Transfer characteristics of a representative FET with different gate-
to-channel distances under VDS = 0.2 V.
-0.6 -0.3 0.0 0.3 0.6
0
3
6
9
Drain Current (A)
Ag/AgCl Gate Voltage (V)
-0.6 -0.3 0.0 0.3 0.6
0
3
6
9
Drain Current (A)
Gold Side Gate Voltage (V)
-0.6 -0.3 0.0 0.3 0.6
-0.6
-0.3
0.0
0.3
0.6
150 m
750 m
1350 m
Reference Voltage (V)
Gold Side Gate Voltage (V)
-0.6 -0.3 0.0 0.3 0.6
0
1
2
250 m
550 m
1100 m
1400 m
Drain Current (A)
Gold Side Gate Voltage (V)
0.0 0.2 0.4 0.6
0
2
4
6
Drain Current (A)
Drain Voltage (V)
V
G (Gold Side Gate)
= 1V, step = 0.2 V
V
DS
= 0.4 V, step = 0.1 V
V
DS
= 0.2V
a
c d
e f
0.0 0.2 0.4 0.6
0
1
2
3
4
5
Drain Current (A)
Drain Voltage (V)
b
V
G (Ag/AgCl gate)
= 1V, step = 0.2 V
V
DS
= 0.4 V, step = 0.1 V
Distance between side
gate and channel:
V DS
V GS
Source
0.1 x PBS
Ag/AgCl gate
electrode
PET
In 2 O 3 Drain
V DS
V GS
Source
0.1 x PBS
Gold side gate
electrode
PET
In 2 O 3 Drain
V ref
V GS
0.1 x PBS
Gold side gate
electrode
PET
Gold electrode
V
53
3.4 Flexibility of indium oxide TFTs
In order to characterize the flexibility of the wearable In
2
O
3
FETs, bending
tests were carried out. As shown in Figure 3.3a, we tightly wrapped our
fabricated In
2
O
3
foil around a cylinder. The electrical performance of the
devices under tensile strain was measured. Figure 3.3b compares the transfer
characteristics of a representative In
2
O
3
nanoribbon FET in three conditions:
relaxed status, bent with a radius of curvature of ~ 3 mm, and after 100
bending cycles. The devices exhibited n-type behavior in all three conditions
without any perceptible change of its performance. In order to verify the
reliability of our platform when deformed, the flexibility tests were performed
on In
2
O
3
FETs functionalization with a gel film containing chitosan, SWCNT,
and glucose oxidase. Figure 3.3c, 3.3e, and 3.3g plot the mobility, the on-off
ratio, and the threshold voltage averaged over 9 devices bent with a radius of
curvature of infinity (relaxed), 3, 10, and 15 mm, respectively. The typical
transfer curves of the devices under different bending conditions are plotted
in Figure S4, in the Supporting Information. When the foil was bent with a
radius of curvature of ~ 3 mm, a tensile strain of ~ 0.25%, parallel to the drain-
to-source current direction, was applied to the In
2
O
3
FETs. We have further
plotted out the mobility as a function of strain. There was no significant
change of the electrical performance of the In
2
O
3
FETs when the devices were
54
in different bending conditions, as the mobility only showed small variation
between 22.15 ± 1.68 cm
2
V
−1
s
−1
and 22.70 ± 1.65 cm
2
V
−1
s
−1
, the threshold
voltage only showed variation between 0.273 ± 0.028 V and 0.280 ± 0.027 V,
and the logarithm on-off ratio showed variation between 4.71 ± 0.13 and 4.84
± 0.12. Figure 3.3d, 3.3f, and 3.3h plot the mobility, the threshold voltage, and
the on-off ratio of the devices after 0 (before bending), 5, 10, 50, and 100
bending cycles, respectively, and the changes in device performance were
negligible as well. The mobility varied in the range of 22.98 ± 1.34 cm
2
V
−1
s
−1
and 23.78 ± 1.87 cm
2
V
−1
s
−1
, the threshold voltage varied between 0.273
± 0.005 V and 0.266 ± 0.016 V, and the logarithm on-off ratio varied between
4.98 ± 0.17 to 4.96 ± 0.14. On the basis of the test results, all the In
2
O
3
nanoribbon FETs after bending tests still maintained excellent performance,
confirming that our platform is reliable under mechanical deformation.
55
Figure 3. 3 Flexibility of In2O3 FETs on a PET substrate. (a) Photograph of a biosensor foil
wrapping around a glass cylinder. Scale bar is 5 mm. (b) Transfer characteristics of a
representative In2O3 FET under relaxed state, bent with a radius of ~ 3 mm, and after
bending 100 times. (c) Mobility, (e) threshold voltage, and (g) on-off ratio of In2O3 FETs
bent with different radius. (d) Mobility, (f) threshold voltage, and (h) on-off ratio of In2O3
FETs bent with a radius of ~ 3 mm after different bending cycles.
-0.6 -0.3 0.0 0.3 0.6
0.0
0.5
1.0
1.5
2.0 R = 3 mm
After 100 Bending Cycles
Relaxed
Drain Current (A)
Gold Side Gate Voltage (V)
0 20 40 60 80 100
10
15
20
25
30
Bending Cycles
Mobility (cm
2
V
-1
s
-1
)
0 20 40 60 80 100
0.0
0.1
0.2
0.3
0.4
0.5
Bending Cycles
Threshold Voltage (V)
0 20 40 60 80 100
10
3
10
4
10
5
10
6
Bending Cycles
On/off Ratio
20 15 10 5
10
15
20
25
30
Bending Radius (mm)
Mobility (cm
2
V
-1
s
-1
)
20 15 10 5
10
2
10
3
10
4
10
5
10
6
Bending Radius (mm)
On/off Ratio
20 15 10 5
0.0
0.1
0.2
0.3
0.4
0.5
Bending Radius (mm)
Threshold Voltage (V)
a b
c d
g
e f
h
∞
∞
∞
56
3.5 Glucose sensing in human body fluids
The ability of sensing in a small amount of liquid is crucial to wearable sensors,
because of the limited amount of body fluids at regular intervals. A
polydimethylsiloxane (PDMS) stamp was adopted as a microwell to
accumulate body fluids (Figure 3.4a). It can also serve as a passivation layer
to ensure reliable sensing without disturbance introduced by electrical contact
of the metal lines with body and body fluids. A mixture of curing agent and
PDMS at a ratio of 1:10 was first spin-coated onto a silicon wafer before
thermally cured at 80 °C for 1 h. After punching a hole with a diameter of 3
mm, the PDMS stamp was laminated onto the biosensor substrate by van der
Waals force. To guarantee the biosensor can work properly in a limited
amount of liquid, we filled the PDMS microwell with 10 µL solution and
performed electrical measurements using a gold side gate electrode. Figure 4b
and 4c show the transfer curves and output curves of the In
2
O
3
FETs measured
with a gold side gate in the electrolyte of ~ 10 µL 0.1 x PBS. The electrical
performance measured in a small amount of liquid is comparable to the results
shown in Figure 3.2c and 3.2d (measured in 300 µL 0.1 x PBS). It illustrates
that our biosensing platform can efficiently work in the liquid with amount as
small as 10 µL, which is a 30-fold decrease from that we previously
reported.
26
57
Figure 3. 4 PH sensing and glucose sensing. (a) Schematic diagram showing a PDMS
microwell is attached to the PET substrate with In 2O3 FETs. (b) Family curves of IDS-VDS
measured with the channel area submerged in the PDMS well. (c) Family curves of IDS-
VGS measured with the channel area submerged in the PDMS well. (d) PH sensing when
the gate bias was applied with a Ag/AgCl electrode or a gold side gate electrode. (e)
Schematic diagram showing the working principle of glucose sensor. (f) Glucose sensing
results in 0.1 x PBS with a gold side gate.
0 500 1000
1
2
3
I/I
0
Time (s)
-0.6 -0.3 0.0 0.3 0.6
0
1
2
3
4
5
6
Drain Current (A)
Gold Side Gate Voltage (V)
1.00
1.05
1.10
10 nM
50 nM
100 nM
0.1 x PBS
Chitosan/Carbon nanotubes/Glucose oxidase
0.0 0.2 0.4 0.6
0
1
2
3
4
5
Drain Current (A)
Drain Voltage (V)
V
G (Side Gate)
= 1 V, step = 0.2 V
V
DS
= 0.4 V, step = 0.1 V
a
c d
b
e f
Gold electrode
Chitosan film
GOx
Glucose in PBS
G
SWCNT
10 9 8 7 6 5
0.1
1
10
I/I
(%)
PH
Ag/AgCl external gate
Gold side gate
0.5 µM
1 µM
10 µM
100 µM
250 µM
1 mM
58
To further confirm the sensing ability of our biosensor platform, we conducted
pH sensing experiments to test the ionic sensitivity of the biosensor chip in
responses to commercial pH solutions. Figure 3.4d shows the comparison of
the pH sensing responses (ΔI/I
0
) with gate bias supplied using either a gold
side electrode or a Ag/AgCl electrode. The responses are plotted into black
up-triangles and red down-triangles for devices gated with a Ag/AgCl external
liquid electrode and a gold side electrode, respectively. The baseline current
I
0
was obtained when using 0.1 x PBS (pH=7.4) to stabilize the device, and
then the PBS was sequentially changed to commercial pH buffer solutions
ranging from pH 10 to pH 5. Devices showed increase in conduction when
the pH value of the solution decreased, because hydroxyl groups on the
nanoribbon surface were protonated due to more H
+
ions in the solution,
resulting in the positive gating effect on the channel area of the n-type In
2
O
3
nanoribbon transistor. As observed, the sensing results from two different gate
electrodes are almost identical to each other. They both are exponentially
dependent on pH changes, and the drain current increased ~ 2.4 times when
the pH value increased by 1.
After ensuring that our devices has good electrical performance and ionic
sensitivity with the gold side gate, the In
2
O
3
nanoribbon biosensors were used
to detect D-glucose. Figure 3.4e shows the schematic diagram depicting the
59
working principle of the glucose determination using In
2
O
3
nanoribbon
biosensors. The surfaces of source and drain electrodes were functionalized
with chitosan/carbon nanotube/glucose oxidase using ink-jet printing (see
Experimental Methods for details). Chitosan is chosen to work as the
immobilization layer since it is a biocompatible polymeric matrix with good
film-forming ability and high water permeability.
24
Carbon nanotubes have
been reported as efficient routes for increasing the sensitivity for many types
of sensors, owing to their good electrocatalytic property and capacity for
biomolecule immobilization.
24, 35, 36
After being immobilized by the chitosan
film and carbon nanotubes, the glucose oxidase enzymes accept electrons
when they interact with glucose in the solution, and thereafter transfer
electrons to molecular oxygen, and consequently produce hydrogen peroxide
(H
2
O
2
). The enzymatically produced H
2
O
2
will be oxidized under a bias
voltage. The reactions are as follows:
𝐷 glucose+O
2
Glucose oxidase
→ Gluconic acid+H
2
O
2
(2)
H
2
O
2
→ 𝑂 2
+2𝐻 +
+2𝑒 −
(3)
The generation of H
+
depends on the concentration of glucose. Decreasing of
the pH leads to protonation the OH groups on the In
2
O
3
surface and results in
changes in the local FET electric field, and ultimately causes changes in the
60
conductance and current. Figure 3.4f shows the continuous monitoring of the
sensing signal in responses to different glucose concentrations. The channel
current increases with the additions of glucose and shows the detection limit
of about 10 nM (~ 2.2% of the baseline current). Our glucose sensor can detect
glucose in the concentration range between 10 nM to 1 mM, covering typical
glucose concentrations in human body fluids. These glucose concentrations
correspond to typical sweat glucose concentrations of both diabetes patients
and healthy people.
10
The detection limit we obtained here is much lower than
a typical electrochemical amperometric glucose sensor.
2, 8
We also performed
control experiments on sensors without glucose oxidase, and as observed,
those sensors did not respond to glucose.
61
Figure 3. 5 Glucose sensing in human body fluids. (a) Real-time glucose sensing results in
artificial tears. (b) Real-time glucose sensing resulting in artificial sweat. (c) Real-time
glucose sensing in human saliva. (d) Comparison of sensing responses verses different
glucose concentrations in 0.1 x PBS, artificial tears, artificial sweat, and saliva.
Wearable In
2
O
3
nanoribbon glucose sensors are further used for human body
fluid analysis. The glucose concentration is much lower in tears, sweat, and
saliva than in blood. While normal blood glucose levels range between 70
mg/dL (3.9 mM) and 140 mg/dL (7.8 mM) or higher, by contrast, tear glucose
levels are on the order of 0.1 -0.6 mM,
37-39
sweat glucose has been reported of
5 to 20 mg/dL (0.277 mM – 1.11 mM),
18
and saliva glucose concentrations
0 500 1000 1500
1
2
3
I/I
0
Time (s)
0 500 1000 1500
1
2
3
Time (s)
I/I
0
1.00
1.05
1.10
1.15
1.20
Artificial tears
0 500 1000 1500 2000
1
2
3
I/I
0
Time (s)
Artificial sweat
Saliva
0.01 0.1 1 10 100 1000
0
1
2
0.1 x PBS
Artifiical tears
Artificial sweat
Saliva
I/I
(%)
Glucose Concentration (M)
a b
c d
0.5 µM
1 µM
10 µM
100 µM
250 µM
1 mM
1 µM
10 µM
100 µM
1 mM
1 µM
10 µM
100 µM
1 mM
1.00
1.05
1.10
1.15
Artificial sweat
10 nM
100 nM
1.00
1.05
1.10
1.15
Saliva
10 nM
100 nM
Artificial tears
10 nM
100 nM
50 nM
62
are around 0.51 – 2.32 mg/dL (28.3 μM – 0.129 mM).
40-42
Figure 3.5a, 3.5b,
and 3.5c show the representative current responses of the glucose in artificial
human tears, artificial human sweat, and saliva, respectively. The details
about the preparation of the body fluids are described in the Experimental
Methods section. Initially, in Figure 3.5a, the devices were submerged in 0.1
x PBS to obtain the baseline current. After changing the electrolyte from 0.1
x PBS to artificial tears at 150 s, the sensing signal bumped up a little bit,
which is due to the pH difference between artificial tears and 0.1 x PBS. Noise
levels of the glucose sensing in artificial tears were higher than the results in
PBS when comparing the inset figures in Figure 3.4f and Figure 3.5a. This
high noise level comes from the weaker buffer capability of the artificial tears,
and results in lower signal-to-noise ratio, which consequently affects the
detection limit. We extracted the relationship between the glucose
concentration and the saturated current response from the real-time sensing
data in PBS solution, and plotted into Figure 3.5d. For comparison, the
sensing results in artificial tears, sweat, and saliva are also plotted. The high
agreement between the data with PBS and the data with artificial tears is a
good indicator that the signals from both media are attributed to mainly
glucose instead of other nonspecific proteins. In the cases of artificial sweat
and saliva, even though the sensing signals are slightly lower than the
63
responses from PBS, which may be due to their different ionic strengths and
complex ingredients, our sensors can differentiate the glucose concentration
as low as 0.1 µM. This sensitivity is sufficient to detect glucose in both sweat
and saliva.
The In
2
O
3
biosensors can be comfortably attached onto an artificial eyeball
and an artificial arm (Figure 3.6a). To ensure the on-body sensing ability, we
imitated the data collection on an artificial eyeball with the biosensor facing
out. Figure 6b shows the ex-situ glucose sensing results using artificial tears.
After using indium wires to connect the bonding pads to our measurement
unit, we constantly flowed artificial tears through the sensing area, as shown
in the inset of Figure 3.6b. After obtaining a stable baseline current, we
sequentially flowed artificial tears spiked with 0.01, 0.1, 1, 10, 100, and 1000
µM glucose, respectively. Overall, it demonstrates that our wearable glucose
sensing platform has the potential to work as contact lenses with embedded
sensors for monitoring the tear glucose level. Similarly, we performed glucose
sensing on an artificial arm, but with the sensor facing down to skin. As the
sensing results shown in Figure 3.6c, our In
2
O
3
biosensor can work as sweat
patch for glucose monitoring. To further confirm that our sensing platform
can be utilized as wearable sweat analyzer, we collect sweat samples from
64
human subject’s forehead during exercise. After we spiked the real sweat with
different concentrations of glucose, glucose sensing was performed with the
prepared samples. Figure 3.6d shows the sensing results with real sweat. After
we added real sweat to replace the PBS solution, the sensing signal shows a
large increase, which is because the original real sweat sample has different
pH and contains glucose before spiking. Good sensitivity was observed
ranging from 0.1 µM to 1 mM, indicating that our sensing platform is a great
candidate for wearable sweat analysis. We also measured sweat glucose level
before and after meal for a healthy person. The before- and after- meal sweat
sample were collected 30 min before and 30 min after a glucose beverage
intake. The sensing results are shown in Figure 6e, the inset figure shows the
device transfer curve measured using before-meal and after-meal sweat
samples as electrolyte. The subject’s blood sugar level before and after meal
is also recorded using a commercial glucose meter, giving the reading of 79
mg/dL and 118 mg/dL, respectively.
65
Figure 3. 6 Off-body glucose sensing. (a) Photographs of the In2O3 biosensors attached
onto an eyeball replica and an artificial arm. (b) Real-time glucose sensing results on an
artificial eyeball. (c) Real-time glucose sensing results on an artificial arm. (d) Real-time
glucose sensing with real sweat collected from human subjects. (e) Glucose sensing results
of real sweat collected before and after glucose beverage intake.
0 500 1000 1500 2000
1.0
1.5
2.0
2.5
I/I
0
Time (s)
0 500 1000
0.8
1.0
1.2
1.4
1.6
Device 1
Device 2
Device 3
I/I
0
Time (s)
-0.6 -0.3 0.0 0.3 0.6
0.0
0.5
1.0
1.5
2.0
2.5
After meal
Before meal
Drain Current (A)
Gold Side Gate Voltage (V)
0 500 1000 1500
1
2
3
Device 1
Device 2
Device 3
I/I
0
Time (s)
1.00
1.05
Sensing on an eye ball
0 500 1000 1500 2000
1.0
1.5
2.0
2.5
I/I
0
Time (s)
Sensing on an artificial hand
a
b c
d e
0.1 µM
1 µM
100 µM
1 mM
Artificial sweat
10 µM
Artificial sweat
0.1 µM
1 µM
100 µM
1 mM
10 µM
10 nM
0.1 µM
1 µM
100 µM
1 mM
Real sweat
10 µM
66
We previously demonstrated that our In
2
O
3
biosensors showed very stable
performance when it was kept in aqueous solutions.
27
However, the proteins
such as glucose oxidase may not be as robust as the sensors after a long time.
43
We characterized an In
2
O
3
biosensor with functionalization
(chitosan/CNT/GOx) for 2 weeks. The device was measured every day and
was stored at 4 ℃ after each measurement. The sensing signals showed only
very small differences in the first 4 days and the responses to 10 µM and 100
µM glucose in PBS decreased only about 25% and 30% after 2 weeks,
respectively (Figure S8 in the Supporting Information). The decrease of the
sensing responses can be attributed to the deactivation of the enzyme glucose
oxidase, and the loss of the enzymes during washing steps. The degradation
of the devices would not be a problem in consideration of our low-cost
disposable biosensors.
Wearable biosensors have great potential to be widely used in healthcare and
fitness applications. Many groups have experimentally demonstrated
functional prototypes. For example, Gao et al. made great progress and
developed wearable sensor platform for multiplexed in-situ perspiration
analysis.
2
The reported potentiometric biosensor arrays can detect a wide
variety of sweat metabolites and electrolytes, including glucose. Lee et al.
67
have introduced a “patch-like” electrochemical device for diabetes monitoring
and therapy.
44
In this work, we demonstrated FET-based prototype glucose
biosensor using In
2
O
3
nanoribbons which can be complementary to
electrochemical sensors. The facile and low temperature shadow-mask
fabrication process can produce wearable biosensors with high sensitivity (~
10 nM detection limit) and wide detection range (~ 5 orders of magnitude).
Moreover, our biosensor platform can be easily integrated with digital
wristbands, smart watches, and implantable electronics.
Recently, great progress has been made using liquid metal-based reaction to
produce 2D semiconductors.
45-47
Because of their good charge-carrier
mobility and outstanding mechanical properties, 2D materials are very
promising for next-generation wearable electronics. Wearable biosensors
using single-crystalline, sub-nanometer layers of 2D materials, such as In
2
O
3
,
Ga
2
O
3
, SnO
2
, ZnO, and InGaZnO, can be explored in the future.
68
3.6 Supporting information
To characterize the In
2
O
3
nanoribbons, we used atomic force microscopy
(AFM) and X-ray diffraction (XRD) on samples deposited on PET substrate
(Figure 3.7). The AFM images (Figure 3.7a and Figure 3.7b) show that the
nanoribbons are solid and have clear edges. The height profile (Figure 3.7a
inset) shows the thickness of In
2
O
3
nanoribbons is ~ 20 nm. The XRD pattern
shown in Figure 3.7c presents only PET peaks, indicating the In
2
O
3
is
amorphous.
Figure 3. 7 (a) AFM image with height profile of a ~ 20 nm thick In2O3 nanoribbon. (b)
Zoom-in AFM scan of a 1 µm × 1 µm square on the In2O3 film. (c) XRD of RF sputtered
In2O3 film deposited on top of PET substrate.
The field-effect mobility of the In
2
O
3
FET is calculated using the following
equation:
g
m
=
dI
D
dV
GS
=
W
L
C
DL
𝜇 FE
V
D
(1)
0 5 10 15
-15
-10
-5
0
5
10
15
20
Height (nm)
Distance (m)
Thickness ~ 20 nm
a b
20 40 60 80
Intensity (a.u.)
2 (degree)
PET
c 50 nm
25 nm
0 nm
8 nm
0 nm
5 µm
69
Where W is the channel width, L is the channel length, and C
DL
is the
electrical double layer capacitance per unit area in 0.1 M ionic strength
aqueous solution reported previously (25.52 μF cm
−2
). The maximum
transconductance 5.69 μS was obtain at a drain voltage of 0.2 V and gate
voltage of 0.527 V as shown in Figure 3.8.
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
0.0
0.5
1.0
1.5
2.0
Gate Voltage (V)
Drain Current (A)
-1
0
1
2
3
4
5
6
7
g
m
(S)
Figure 3. 8 A representative transfer curve of an In2O3 nanoribbon FET with VDS = 0.2 V
and its gm.
We also conducted a statistical study of key electrical properties for 50 In
2
O
3
nanoribbon devices on a biosensor chip measured in 0.1 x PBS and biased
through the Ag/AgCl electrode and the gold side gate. Figure 3.9 shows that
the device performance including mobility (μ), threshold voltage (V
th
),
on/off ration and on-state current, is very similar when using gold side gate
electrodes and Ag/AgCl gate electrodes.
V
DS
=0.2V
70
Figure 3. 9 Electrical performance of 50 In2O3 nanoribbon transistors (a) Mobilities (μ),
(b) Threshold voltage (V TH), (c) On/off current ratios at VDS = 0.2 V, and (d) On-state
current (ION) at V GS = 0.6 V and VDS = 0.2 V.
In order to characterize the flexibility of the wearable In
2
O
3
FETs, bending
tests were carried out on both functionalized and unfunctionalized devices.
We tightly wrapped our fabricated In
2
O
3
foil around cylinder with different
radius of curvatures. The electrical performance of the devices under tensile
strain was measured in aqueous environment. In Figure 3.10, we compared
0 10 20 30 40 50
10
0
10
1
10
2
10
3
10
4
10
5
10
6
10
7
Gold Side Gate
Ag/AgCl gate
On/off ratio
Device Index
0 10 20 30 40 50
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Gold Side Gate
Ag/AgCl gate
On-state current ()
Device Index
0 10 20 30 40 50
0.0
0.1
0.2
0.3
0.4
0.5
Ag/AgCl gate
Gold Side Gate
V
th
(V)
Device Index
0 10 20 30 40 50
0
5
10
15
20
25
30
35
40
Gold Side Gate
Ag/AgCl gate
(cm
2
V
-1
s
-1
)
Device Index
a
b
c
d
Gold Side Gate: 22.34 ± 1.44 cm
2
V
-1
s
-1
Ag/AgCl Gate: 22.02 ± 1.52 cm
2
V
-1
s
-1
71
transfer curves of a representative In
2
O
3
biosensor FET under different
bending radius and with different bending cycles.
Figure 3. 10 Transfer characteristics of (a) an unfunctionalized In2O3 FET under relaxed
state, bent with a radius of ~3, 10, and 15 mm, (b) an unfunctionalized In2O3 FET after
bending with 5, 10, 50, and 100 cycles, (c) a functionalized In2O3 FET under relaxed state,
bent with a radius of ~3, 10, and 15 mm, and (d) a functionalized In2O3 FET after bending
with 5, 10, 50, and 100 cycles.
To calculate the tensile strain when the wearable In
2
O
3
nanoribbon FET foil
was wrapped tightly around a cylinder with radius ~ 3 mm, we used the
following formula
1
:
-0.6 -0.3 0.0 0.3 0.6
0.0
0.5
1.0
1.5
2.0
3 mm
10 mm
15 mm
Relaxed
Drain Current (A)
Gold Side Gate Voltage (V)
-0.6 -0.3 0.0 0.3 0.6
0.0
0.5
1.0
1.5
2.0
5 cycle
10 cycle
50 cycle
100 cycle
Drain Current (A)
Gold Side Gate Voltage (V)
Unfunctionalized
V
DS
=0.2V
Unfunctionalized
V
DS
=0.2V
-0.6 -0.3 0.0 0.3 0.6
0.0
0.5
1.0
1.5
2.0
2.5
3 mm
10 mm
15 mm
Relaxed
Drain Current (A)
Gold Side Gate Voltage (V)
-0.6 -0.3 0.0 0.3 0.6
0.0
0.5
1.0
1.5
2.0 5 cycle
10 cycle
50 cycle
100 cycle
Drain Current (A)
Gold Side Gate Voltage (V)
Functionalized
V
DS
=0.2V
Functionalized
V
DS
=0.2V
a
b
c
d
72
𝜀 =
1
𝑅 ×
𝑑 𝑠 +𝑑 𝑓 2
×
𝜒 ·𝛾 2
+2 ·𝜒 ·𝛾 +1
𝜒 ·𝛾 2
+ 𝜒 ·𝛾 + 𝛾 +1
Here, 𝑅 is the bending radius, 𝑑 𝑠 is the thickness of the substrate, and 𝑑 𝑓 is
the thickness of In
2
O
3
nanoribbon transistor (TFT). 𝛾 =𝑑 𝑓 / 𝑑 𝑠 and 𝜒 =
𝑌 𝑓 /𝑌 𝑠 , where 𝑌 𝑓 and 𝑌 𝑠 are the Young’s modulus of In
2
O
3
FET and the
substrate, respectively. We assume 𝑌 𝑓 = 𝑌 𝑠 and the above equation can be
further simplified:
𝜀 =
1
𝑅 ×
𝑑 𝑠 +𝑑 𝑓 2
The thickness of the substrate is 15 μm and the total thickness of the TFT is
less than 100 nm. With the bending radius of 3 mm, the tensile strain is
calculated to be ~ 0.25%. We have further plotted out the mobility as a
function of tensile strain (Figure 3.11).
0.0 0.1 0.2 0.3
10
15
20
25
30
Tensile strain (%)
Mobility (cm
2
V
-1
s
-1
)
Figure 3. 11 The mobilities of In2O3 FETs as a function of tensile strain.
73
To further confirm the sensing ability of our biosensor platform, we
conducted pH sensing experiments to test the ionic sensitivity of the
biosensor chip in responses to commercial pH solutions. Figure 3.12 shows
the real-time sensing response of an In
2
O
3
FET to standard pH calibration
solutions. The initial current I
o
was obtained by using PBS to stabilize the
device, and then the PBS buffer was sequentially changed to commercial pH
buffer solutions ranging from pH 10 to pH 5. The drain current responded
quickly and log-linearly to each pH buffer.
Figure 3. 12 Real-time sensing responses of an In2O3 FET to standard pH calibration
solutions. Gate voltage is applied with (a) a Ag/AgCl gate electrode, and (b) a gold side
gate electrode.
To further confirm that the environment changes are introduced by the
reaction of glucose and glucose oxidase, we performed a control experiment
0 400 800 1200
0
1
2
3
4
PH=5
PH=6
PH=7
PH=8
PH=9
I/I
0
Time (s)
PH=10
0 200 400 600 800 1000 1200
0
1
2
3
4
PH=5
PH=6
PH=7
PH=8
PH=9
I/I
0
Time (s)
PH=10
V
DS
= 200 mV
V
G (gold side gate)
= 300 mV
V
DS
= 200 mV
V
G (Ag/AgCl gate)
= 300 mV
a
b
74
on sensors without glucose oxidase. In this control experiment, we
functionalized the surface of the gold electrodes with chitosan and carbon
nanotube only. We sequentially added glucose in 0.1 x PBS with
concentrations of 1 µM, 10 µM, 100 µM, and 1 mM. As expected, sensors
did not respond to glucose as shown in Figure 3.13.
0 500 1000 1500
1.00
1.05
1.10
1 mM
100 M
1 M
I/I
0
Time (s)
Figure 3. 13 Glucose sensing results of an In2O3 nanoribbon biosensor functionalized with
chitosan and SWCNT only.
We characterized an In
2
O
3
biosensor with functionalization
(chitosan/CNT/GOx) for 2 weeks. The device was measured every day and
store at 4 ℃ after each measurement. The sensing signals shows very small
differences in first 4 days and the responses to 10 µM and 100 µM glucose
in PBS decreased for about 25% and 30% after 2 weeks, respectively (Figure
3.14). The decrease of the sensing responses can be attributed to the
deactivation of the enzyme glucose oxidase over a long time, and the loss of
75
the enzymes during washing steps. The degradation of the devices will not
be a problem in consideration of our low-cost disposable biosensors.
0 500 1000 1500
1.0
1.5
2.0
100 M
10 M
1 M
100 nM
10 nM
Day 1
Day 2
Day 4
Day 7
Day 14
I/I
0
Time (s)
Figure 3. 14 Glucose sensing results with a functionalized sensor after 1, 2, 4, 7, and 14
days.
3.7 Summary
In summary, the In
2
O
3
FET-based wearable biosensors with on-chip gold side
gate electrodes can be used for highly sensitive detection of glucose with the
detection limit down to 10 nM. The all-on-a-chip device structure can be
incorporated into the straightforward two-step shadow mask fabrication. The
gold side gate electrodes show stable and efficient gating effect on In
2
O
3
FETs
on flexible substrates. Mobilities in 0.1 x PBS of ~ 22 cm
2
V
-1
s
-1
and on-off
ratio over 10
5
were achieved. The non-invasive glucose detections in human
body fluids, such as tears and sweat, was also demonstrated. We further
investigated glucose sensing on an eyeball replica and on an artificial hand.
76
Lastly, we demonstrated that our glucose sensor can work in real human sweat
and distinguish glucose level before- and after- meal. Given the facile and
highly scalable fabrication process, low driving voltage, and reliable sensing
behavior even when deformed, such sensing platform is promising for
continuous personalized health monitoring, food industry and environmental
monitoring.
3.8 References
1. Takei, K.; Takahashi, T.; Ho, J. C.; Ko, H.; Gillies, A. G.; Leu, P. W.; Fearing, R.
S.; Javey, A., Nanowire Active-Matrix Circuitry for Low-V oltage Macroscale Artificial
Skin. Nat. Mater. 2010, 9, 821-826.
2. Gao, W.; Emaminejad, S.; Nyein, H. Y . Y .; Challa, S.; Chen, K.; Peck, A.; Fahad,
H. M.; Ota, H.; Shiraki, H.; Kiriya, D.; Lien, D; Brooks, G. A.; Davis, R. W.; Javey, A.,
Fully Integrated Wearable Sensor Arrays for Multiplexed in Situ Perspiration Analysis.
Nature 2016, 529, 509-514.
3. Kaltenbrunner, M.; Sekitani, T.; Reeder, J.; Yokota, T.; Kuribara, K.; Tokuhara, T.;
Drack, M.; Schwödiauer, R.; Graz, I.; Bauer-Gogonea, S.; Bauer, S.; Someya, T., An Ultra-
Lightweight Design for Imperceptible Plastic Electronics. Nature 2013, 499, 458-463.
4. Xu, S.; Zhang, Y .; Jia, L.; Mathewson, K. E.; Jang, K.-I.; Kim, J.; Fu, H.; Huang,
X.; Chava, P.; Wang, R.; Bhole, S.; Wang, L.; Na, Y . J.; Guan, Y .; Flavin, M.; Han, Z.;
Huang, Y .; Rogers, J. A.,Soft Microfluidic Assemblies of Sensors, Circuits, and Radios for
the Skin. Science 2014, 344, 70-74.
5. McAlpine, M. C.; Ahmad, H.; Wang, D.; Heath, J. R., Highly Ordered Nanowire
Arrays on Plastic Substrates for Ultrasensitive Flexible Chemical Sensors. Nat. Mater.
77
2007, 6, 379-384.
6. Kim, D. H.; Lu, N.; Ma, R.; Kim, Y . S.; Kim, R. H.; Wang, S.; Wu, J.; Won, S. M.;
Tao, H.; Islam, A.; Yu, K. J.; Kim, T. I.; Chowdhury, R.; Ying, M.; Xu, L.; Li, M.; Chung
H. J.; Keum, H.; McCormick, M.; Liu, P., et al. Epidermal Electronics. Science 201 1, 333,
838-843.
7. Lipomi, D. J.; V osgueritchian, M.; Tee, B. C.; Hellstrom, S. L.; Lee, J. A.; Fox, C.
H.; Bao, Z., Skin-Like Pressure and Strain Sensors Based on Transparent Elastic Films of
Carbon Nanotubes. Nat. Nanotechnol. 201 1, 6, 788-792.
8. Lee, H.; Song, C.; Hong, Y . S.; Kim, M. S.; Cho, H. R.; Kang, T.; Shin, K.; Choi,
S. H.; Hyeon, T.; Kim, D. H., Wearable/Disposable Sweat-Based Glucose Monitoring
Device with Multistage Transdermal Drug Delivery Module. Sci. Adv. 2017, 3.
9. Bandodkar, A. J.; Wang, J., Non-Invasive Wearable Electrochemical Sensors: A
Review. Trends Biotechnol. 2014, 32, 363-371.
10. Makaram, P.; Owens, D.; Aceros, J., Trends in Nanomaterial-Based Non-Invasive
Diabetes Sensing Technologies. Diagnostics 2014, 4, 27-46.
11. Mannoor, M. S.; Tao, H.; Clayton, J. D.; Sengupta, A.; Kaplan, D. L.; Naik, R. R.;
Verma, N.; Omenetto, F. G.; McAlpine, M. C., Graphene-Based Wireless Bacteria
Detection on Tooth Enamel. Nat. Commun. 2012, 3, 763.
12. Liao, C.; Mak, C.; Zhang, M.; Chan, H. L.; Yan, F., Flexible Organic
Electrochemical Transistors for Highly Selective Enzyme Biosensors and Used for Saliva
Testing. Adv. Mater. 2015, 27, 676-681.
13. Veiseh, O.; Tang, B. C.; Whitehead, K. A.; Anderson, D. G.; Langer, R., Managing
Diabetes with Nanomedicine: Challenges and Opportunities. Nat. Rev. Drug Discovery
2015, 14, 45-57.
14. Olarte, O.; Chilo, J.; Pelegri-Sebastia, J.; Barbe, K.; Van Moer, W. Glucose
Detection in Human Sweat Using An Electronic Nose, Conf. Proc. IEEE Eng. Med. Biol.
Soc. 2013; 1462-1465.
78
15. Yan, Q.; Peng, B.; Su, G.; Cohan, B. E.; Major, T. C.; Meyerhoff, M. E.,
Measurement of Tear Glucose Levels with Amperometric Glucose Biosensor/Capillary
Tube Configuration. Anal. Chem. 201 1, 83, 8341-8346.
16. Vashist, S. K., Non-Invasive Glucose Monitoring Technology in Diabetes
Management: A Review. Anal. Chim. Acta 2012, 750, 16-27.
17. Tierney, M.; Kim, H.; Burns, M.; Tamada, J.; Potts, R., Electroanalysis of Glucose
in Transcutaneously Extracted Samples. Electroanalysis 2000, 12, 666-671.
18. Moyer, J.; Wilson, D.; Finkelshtein, I.; Wong, B.; Potts, R., Correlation between
Sweat Glucose and Blood Glucose in Subjects with Diabetes. Diabetes Technol. Ther. 2012,
14, 398-402.
19. Ronkainen, N. J.; Halsall, H. B.; Heineman, W. R., Electrochemical Biosensors.
Chem. Soc. Rev. 2010, 39, 1747-1763.
20. Zhang, M.; Liao, C.; Mak, C. H.; You, P.; Mak, C. L.; Yan, F., Highly Sensitive
Glucose Sensors Based on Enzyme-Modified Whole-Graphene Solution-Gated Transistors.
Sci. Rep. 2015, 5, 8311.
21. Zhai, D.; Liu, B.; Shi, Y.; Pan, L.; Wang, Y .; Li, W.; Zhang, R.; Yu, G., Highly
Sensitive Glucose Sensor Based on Pt Nanoparticle/Polyaniline Hydrogel Heterostructures.
ACS Nano 2013, 7, 3540-3546.
22. Lin, P.; Yan, F.; Chan, H. L., Ion-Sensitive Properties of Organic Electrochemical
Transistors. ACS Appl. Mater. Interfaces 2010, 2, 1637-1641.
23. Claussen, J. C.; Kumar, A.; Jaroch, D. B.; Khawaja, M. H.; Hibbard, A. B.;
Porterfield, D. M.; Fisher, T. S., Nanostructuring Platinum Nanoparticles on Multilayered
Graphene Petal Nanosheets for Electrochemical Biosensing. Adv. Funct. Mater. 2012, 22,
3399-3405.
24. Tang, H.; Yan, F.; Lin, P.; Xu, J.; Chan, H. L., Highly Sensitive Glucose Biosensors
Based on Organic Electrochemical Transistors Using Platinum Gate Electrodes Modified
with Enzyme and Nanomaterials. Adv. Funct. Mater. 201 1, 21, 2264-2272.
79
25. Heller, A., Implanted Electrochemical Glucose Sensors for the Management of
Diabetes. Annu. Rev. Biomed. Eng. 1999, 1, 153-175.
26. Liu, Q.; Aroonyadet, N.; Song, Y .; Wang, X.; Cao, X.; Liu, Y .; Cong, S.; Wu, F.;
Thompson, M. E.; Zhou, C., Highly Sensitive and Quick Detection of Acute Myocardial
Infarction Biomarkers Using In2O3 Nanoribbon Biosensors Fabricated Using Shadow
Masks. ACS Nano 2016, 10, 10117-10125.
27. Aroonyadet, N.; Wang, X.; Song, Y .; Chen, H.; Cote, R. J.; Thompson, M. E.; Datar,
R. H.; Zhou, C., Highly Scalable, Uniform, and Sensitive Biosensors Based on Top-Down
Indium Oxide Nanoribbons and Electronic Enzyme-Linked Immunosorbent Assay. Nano
Lett. 2015, 15, 1943-1951.
28. Rim, Y . S.; Bae, S. H.; Chen, H.; Yang, J. L.; Kim, J.; Andrews, A. M.; Weiss, P. S.;
Yang, Y .; Tseng, H.-R., Printable Ultrathin Metal Oxide Semiconductor-Based Conformal
Biosensors. ACS Nano 2015, 9, 12174-12181.
29. Kim, J.; Rim, Y . S.; Chen, H.; Cao, H. H.; Nakatsuka, N.; Hinton, H. L.; Zhao, C.;
Andrews, A. M.; Yang, Y.; Weiss, P. S., Fabrication of High-Performance Ultrathin In2O3
Film Field-Effect Transistors and Biosensors Using Chemical Lift-Off Lithography. ACS
Nano 2015, 9, 4572-4582.
30. Tang, T.; Han, S.; Jin, W.; Liu, X.; Li, C.; Zhang, D.; Zhou, C.; Chen, B.; Han, J.;
Meyyapan, M., Synthesis and Characterization of Single-Crystal Indium Nitride
Nanowires. J. Mater. Res. 2004, 19, 423-426.
31. Cao, X.; Cao, Y .; Zhou, C., Imperceptible and Ultraflexible p‑Type Transistors and
Macroelectronics Based on Carbon Nanotubes. ACS Nano. 2016, 10, 199-206.
32. Someya, T.; Kato, Y .; Sekitani, T.; Iba, S.; Noguchi, Y .; Murase, Y .; Kawaguchi, H.;
Sakurai, T., Conformable, Flexible, Large-Area Networks of Pressure and Thermal Sensors
with Organic Transistor Active Matrixes. Proc. Natl. Acad. Sci. U. S. A. 20 05, 102, 12321-
12325.
33. Abbas, A. N.; Liu, G.; Liu, B.; Zhang, L.; Liu, H.; Ohlberg, D.; Wu, W.; Zhou, C.,
Patterning, Characterization, and Chemical Sensing Applications of Graphene Nanoribbon
80
Arrays Down to 5 nm Using Helium Ion Beam Lithography. ACS Nano 2014, 8, 1538-
1546.
34. Park, S.; Lee, S.; Kim, C.-H.; Lee, I.; Lee, W.-J.; Kim, S.; Lee, B.-G.; Jang, J.-H.;
Yoon, M.-H., Sub-0.5 V Highly Stable Aqueous Salt Gated Metal Oxide Electronics. Sci.
Rep. 2015, 5, 13088.
35. Hrapovic, S.; Liu, Y .; Male, K. B.; Luong, J. H., Electrochemical Biosensing
Platforms Using Platinum Nanoparticles and Carbon Nanotubes. Anal. Chem. 2004, 76,
1083-1088.
36. Wang, J.; Musameh, M.; Lin, Y ., Solubilization of Carbon Nanotubes by Nafion
toward the Preparation of Amperometric Biosensors. J. Am. Chem. Soc. 2 003, 125, 2408-
2409.
37. Berman, E. R., Biochemistry of the Eye. Plenum Press, New York.
38. Yao, H.; Shum, A. J.; Cowan, M.; Lähdesmäki, I.; Parviz, B. A., A Contact Lens
with Embedded Sensor for Monitoring Tear Glucose Level. Biosens. Bioelectron. 201 1, 26,
3290-3296.
39. Yao, H.; Liao, Y .; Lingley, A.; Afanasiev, A.; Lähdesmäki, I.; Otis, B.; Parviz, B., A
Contact Lens with Integrated Telecommunication Circuit and Sensors for Wireless and
Continuous Tear Glucose Monitoring. J. Micromech. Microeng. 20 12, 22, 075007.
40. Abikshyeet, P.; Ramesh, V .; Oza, N., Glucose Estimation in the Salivary Secretion
of Diabetes Mellitus Patients. Diabetes, Metab. Syndr. Obes.: Targets Ther. 2012, 5, 149.
41. Zhang, W.; Du, Y .; Wang, M. L., Noninvasive Glucose Monitoring Using Saliva
Nano-Biosensor. Sens. Bio-Sensing Res. 2015, 4, 23-29.
42. Zhang, W.; Du, Y .; Wang, M. L., On-Chip Highly Sensitive Saliva Glucose Sensing
Using Multilayer Films Composed of Single-Walled Carbon Nanotubes, Gold
Nanoparticles, and Glucose Oxidase. Sens. Bio-Sensing Res. 2015, 4, 96-102.
43. Valdes, T. I.; Moussy, F., In Vitro and in Vivo Degradation of Glucose Oxidase
Enzyme Used for an Implantable Glucose Biosensor. Diabetes Technol. Ther. 2000, 2, 367-
81
376.
44. Lee, H.; Choi, T. K.; Lee, Y . B.; Cho, H. R.; Ghaffari, R.; Wang, L.; Choi, H. J.;
Chung, T. D.; Lu, N.; Hyeon, T., A Graphene-Based Electrochemical Device with
Thermoresponsive Microneedles for Diabetes Monitoring and Therapy. Nature
Nanotechnol. 2016, 11, 566-572.
45. Zavabeti, A.; Ou, J. Z.; Carey, B. J.; Syed, N.; Orrell-Trigg, R.; Mayes, E. L.; Xu,
C.; Kavehei, O.; O’mullane, A. P.; Kaner, R. B., A Liquid Metal Reaction Environment for
the Room-Temperature Synthesis of Atomically Thin Metal Oxides. Science 2017, 358,
332-335.
46. Kim, Y . D.; Hone, J., Materials Science: Screen Printing of 2D Semiconductors.
Nature 2017, 544, 167-168.
47. Carey, B. J.; Ou, J. Z.; Clark, R. M.; Berean, K. J.; Zavabeti, A.; Chesman, A. S.;
Russo, S. P.; Lau, D. W.; Xu, Z.-Q.; Bao, Q., Wafer-scale two-dimensional semiconductors
from printed oxide skin of liquid metals. Nat. Commun. 2017, 8, 14482.
82
Chapter 4 Multiplexed sensing of
serotonin and dopamine using ultra-
flexible In
2
O
3
nanoribbon aptamer-
field-effect transistors
4.1 Introduction
There are tremendous interests in developing electronic devices that can
extend or offer new capabilities to understand the complex system of a brain.
Recently, extensive contributions have been made in soft brain-machine
interface electronics
1-3
and minimally invasive bioelectronics.
4-6
Detection of
neurotransmitters is of great importance in studying the brain activities, as
neurotransmitters are essential chemicals in the synaptic transmission process.
Development of neural sensors will also benefit diagnosis and treatment of
neurological disorders and diseases by real-time monitoring of
neurotransmitter concentration. Conventional neurotransmitter monitoring
approaches, such as microdialysis with off-line analysis
7, 8
and
electrochemical sensors with fast-scan cyclic voltammetry,
9, 10
have been
extensively studied. However, limitations remain with these methods: long
turnaround time of a single measurement for microdialysis,
11
poor selectivity
83
for electrochemical sensors when chemical species have similar
oxidation/reduction potentials,
12, 13
and device fouling and degradation over
time when using enzyme-based electrochemical sensors.
14, 15
Therefore, an
implantable sensor, with capabilities to simultaneous detect multiple
neurotransmitters in real time with high specificity, high temporal resolution,
and long-term stability, is of great importance in the neuroscience field.
Recent studies on small molecules detection in high-ionic strength condition
using aptamer-field-effect transistors
16, 17
suggest that such sensors with ultra-
flexibility could help enable real-time, in vivo multiplexed neurotransmitter
sensing. An aptamer is a short DNA or RNA oligomer whose base sequences
are carefully designed so that it binds to a specific target molecule of interest
with high affinity. Aptamers have several advantages over antibodies, such as
better selectivity against small, non-immunogenic molecules, simple scale-up
and purification with less cost, and longer shelf life.
18-20
More importantly, the
size of aptamers are 5 – 10 times smaller than antibodies, which helps
overcoming the Debye length limitations when applied on field-effect-
transistor (FET) based sensors.
21
Pre-selected aptamers with negative
backbones, can be covalently functionalized on the semiconductor channel
(e.g., organic,
22
carbon-based,
23
metal oxide,
17
etc.) of a FET. The binding
84
activity of small target molecules can trigger conformational rearrangement
of aptamers. Changes introduced from aptamer backbones are within or near
the Debye length, which is 0.74 nm in 1 × Phosphate-buffered saline (PBS),
resulting in surface potential changes via electrostatic interaction. Sensing
with FETs is inherently nonlinear,
24
which enables wide detection range and
low detection limit. Therefore, aptamer-field-effect-transistors are of great
interest to the biosensor community.
To work as channel materials of FET-based sensors, metal oxides, such as
indium oxide (In
2
O
3
), have several advantages over other materials, such as
organic materials, carbon-based materials, two-dimensional (2D) materials,
and silicon. In comparison to organic materials, In
2
O
3
is stable in both ambient
condition and aqueous solution,
25
and they have better electrical performance
than organic materials due to their higher electron mobility.
26
Carbon-based
27-
29
and 2D-material-based
30, 31
electronics have attracted a lot of attention in the
application of sensors; however, the preparation and fabrication techniques of
these novel materials are still premature, and the stability and uniformity of
such sensors need further research. Similar to silicon, which can be fabricated
by foundries with high scalability and uniformity, metal-oxide thin-film
transistors have been widely used in industry (e.g. touch screen, displays, and
85
solar cells) due to their excellent electrical performance and large-area
uniformity. As silicon fabrication process usually contains multiple steps of
etching and oxidation, the main advantages of metal oxides over silicon in the
application of wearable sensors are the ease of fabrication and the
compatibility with ultra-flexible plastic substrates.
17, 32-35
Recently, many
ultra-flexible electronics, including displays,
36, 37
photovoltaics,
38
and
wearable biosensors
17, 39
have been reported using metal oxides in their
structures. As we demonstrated previously,
25, 39
In
2
O
3
field-effect transistor
are well suited for wearable sensing applications among varieties of metal
oxides, such as indium-gallium-zinc oxide and zinc oxide. Our In
2
O
3
nanoribbon biosensors have short response time, wide detectable
concentration range, low limit of detection, high uniformity, and the capability
to be integrated with microfluidics and microprocessors.
25
In addition, the
surface of In
2
O
3
can be easily functionalized with different chemicals and
receptors, which would enable the capability of detection of unconventional
target and multiplexed sensing.
In this work, we fabricated arrays of ultra-flexible In
2
O
3
nanoribbon field-
effect transistors using shadow mask techniques. The devices exhibited
excellent electrical performance and uniformity on 1.4-μm-thick polyethylene
86
terephthalate (PET) film. The as-fabricated devices showed good mechanical
flexibility and long-term robustness in high-ionic solutions over a week.
Direct and real-time sensing of serotonin and dopamine were demonstrated
using aptamer In
2
O
3
nanoribbon FETs in high ionic concentrations (1 × PBS),
overcoming the Debye length limitations. The aptamer-field-effect-transistor
sensors showed wide detection range (at least 8 orders of magnitude) and
achieved detection of serotonin and dopamine at ultra-low concentration (~10
fM). The conformal and highly sensitive aptamer-functionalized In
2
O
3
nanobiosensor arrays can function successfully in artificial cerebrospinal fluid,
and multiplexed sensing of temperature, pH, serotonin, and dopamine on an
artificial brain have also been demonstrated.
4.2 Development of ultra-flexible In
2
O
3
nanoribbon
biosensor array
The fabrication process flow of an In
2
O
3
nanoribbon biosensor array on a 1.4-
µm-thick ultra-flexible PET substrate is described in Figure 4.1a. First, a 3-
inch Si/SiO
2
substrate was spin-coated with a 20-μm-thick
polydimethylsiloxane (PDMS) as a sacrificial layer. A 1.4-µm-thick PET film
was then laminated on the carrier wafer via van der Waals force. After
87
attaching the first shadow mask to the wafer, 16-nm-thick In
2
O
3
nanoribbons
were deposited onto the PET foil using radio frequency (RF) sputtering and
defined by the openings of the shadow mask. The second shadow mask was
then aligned and attached to the PET film and carrier wafer. Source, drain,
side gate and temperature sensors with 1 nm titanium and 50 nm gold film
were deposited using a metal evaporator with the second shadow mask for
patterning. After detaching the shadow mask, the biosensor foil was peeled
off from the sacrificial layer, and the multiplexed nanobiosensor array was
ready for further measurement. As shown in Figure 4.1b, the In
2
O
3
biosensor
foil with a size of 5 cm × 5 cm were fabricated. The devices are extremely
thin and ultra-flexible, as the total thickness is only ~1.5 µm and the total
weight is only ~5 mg. Figure 4.1c presents a sensor chip containing one
temperature sensor, one gold side gate, and four In
2
O
3
nanoribbon FET
sensors, indicating that the sensor chip has potential to work as multiplexed
sensing platform after surface functionalization. The as-deposited In
2
O
3
nanoribbons have thicknesses of 16 nm, widths of 25 μm, and lengths of 500
μm. As an example, our sensor foil can be conformally laminated onto human
skin without any discomfort (Figure 4.1d), illustrating the capability for
wearable electronics and healthcare monitoring applications. As the devices
can be bent and wrinkled (Figure 4.1e), they can work on uneven surfaces or
88
even conform to the movements of human body. We believe that ultra-flexible,
lab-on-a-chip, and conformal In
2
O
3
nanoribbon electronics developed here are
promising for multiplexed wearable biosensor applications.
Figure 4. 1 (a) Schematic illustration of the fabrication process flow of ultra-flexible In2O3
field-effect transistor (FET) based multiplexed sensors. We began with a Si/SiO2 substrate
coated with polydimethylsiloxane (PDMS) sacrificial layer. Next, a 1.4-μm-thick
polyethylene terephthalate (PET) film was laminated on the wafer. In2O3 nanoribbons were
sputtered onto the PET substrate through a shadow mask. Ti/Au (1/50 nm thick) were
deposited through another shadow mask and patterned as source, drain, side gate and
temperature sensor. As-fabricated biosensor foil was then peeled off from the rigid carrier
wafer. (b) A photograph of as-fabricated multiplexed sensors on a 1.4-μm-thick PET film.
Scale bar is 1 cm. (c) Optical microscope image of a gold resistive temperature sensor, four
In2O3 nanoribbon FETs, and a gold side gate (from top to bottom). Scale bar is 0.5 mm.
(d) Ultra-flexible biosensor array conformally attached to human skin. Scale bar is 2 cm.
(e) A prototype biosensor foil compressed as human body movement. Scale bar is 1 cm.
89
4.3 Electrical performance and stability test of
In
2
O
3
nanoribbon biosensor
Our FETs exhibited excellent uniformity and stability in aqueous conditions,
and all the following measurements in Figure 4.2 were performed in 1 × PBS.
Figure 4.2a shows the transfer characteristics of an In
2
O
3
nanoribbon
transistor controlled using a gold side gate. The black and blue curve represent
drain current-gate voltage (I
DS
-V
GS
) in logarithmic scale and linear scale,
respectively. The corresponding output characteristics in Figure 2b
demonstrated that the devices have excellent FET behavior. The gate leakage
was negligible, and as an example, was measured to be smaller than 2 nA at
V
DS
= 0.2 V. After measuring the transistor array, a yield of nearly 100 % was
achieved. The charge-carrier mobilities of 56 transistors were measured and
calculated (details shown in the Supporting Information) and they showed
good uniformity (Figure 4.2c), with an average of 17.84 ± 1.79 cm
2
V
−1
s
−1
.
Figure 4.2d exhibits a statistical study of the on-state current and threshold
voltage of 56 devices, and both showed very narrow distributions. It is worth
to note that our In
2
O
3
nanoribbon devices showed excellent electrical
performance and uniformity in the aqueous environment at relatively low
voltages (< 1 V), which is favorable for on-skin healthcare electronics.
90
Figure 4. 2 Electrical performance of ultra-flexible In2O3 nanoribbon biosensor. (a) Typical
transfer characteristics of an In2O3 transistor with L = 500 μm and W = 25 μm using gold
liquid gate. Black line represents ID (drain current) in logarithmic scale and blue line
represents ID in linear scale; V GS, the gate−source voltage. The applied drain–source
voltage, VDS, is 0.2 V. (b) Corresponding output curves in saturation regime. In this plot,
IDS is a function of VDS with V GS from 0 V to 1 V in 0.2 V steps. (c) Map of charge-carrier
mobility of 56 transistors in the array. (d) Histograms showing on-currents and threshold
voltages from the 56 randomly selected transistors in the array. Average mobility is
17.84 ± 1.79 cm
2
V
−1
s
−1
, average on-current is 0.86 ± 0.13 μA, and average threshold
voltage is 0.268 ± 0.024 V.
To demonstrate the robustness of our In
2
O
3
FETs, we conducted several
stability tests. The mechanical flexibility of the devices was evaluated using
experiments that consisted of extreme bending and crumpling. Figure 4.3a
shows a photograph of a group of In
2
O
3
FETs tightly wrapped around a copper
91
wire with a bending radius of 100 µm. At this condition, tensile strain ~0.75 %
was applied to the In
2
O
3
nanoribbons, and the direction of the strain was
parallel to the current flow. In another test, a 5 cm × 5 cm as-fabricated In
2
O
3
sensor foil was severely crumpled (Figure 4.3b) and then flattened for
electrical measurements. The electrical performance of In
2
O
3
FETs in bending
conditions and after crumpling were also measured in 1 × PBS. The typical
transfer curves of a device in the relaxed state, bent with a radius curvature of
~0.1 mm, and after 100 crumpling cycles were shown in Figure 4.3c,
respectively. There was no noticeable change in the electrical characteristics
during the bending and after crumpling. Figure 4.3d and 4.3e show the
mobility and the threshold voltage averaged over 15 devices after 100 bending
cycles and 100 crumpling cycles, respectively. After 100 bending cycles, the
change in mobility was 5.6 % (from 18.35 cm
2
V
−1
s
−1
to 17.31 cm
2
V
−1
s
−1
),
and the threshold voltage only showed small variation between 0.23 V and
0.27 V. Similar results were observed in repeated crumpling tests. The change
in mobility was 6.1 % after 100 crumpling cycles and the threshold voltage
remained around 0.26 V. With the test results shown above, we believe that
all our In
2
O
3
nanoribbon FETs are highly reliable under such small degree of
strain. Figure 4.3f shows the transfer curves of In
2
O
3
devices before and after
being immersed in 1 × PBS buffer solution for 1, 2, 3, 5, and 7 day(s). Based
92
on the results in both linear and logarithmic scales, there is no significant
change in the device performance, which demonstrates the long-term stability
of the In
2
O
3
transistors in high ionic strength solution and indicates our sensors
are at least robust for one week in human body fluids.
Figure 4. 3 Stability of ultra-flexible In2O3 nanoribbon biosensors. (a) Photograph of a
group of ultra-flexible In2O3 biosensors on 1.4-μm-thick PET films wrapped around a
copper wire with a radius of 100 μm. Scale bar, 5 mm. (b) Crumpled In2O3 biosensor foil
(original size: 5 cm × 5 cm square). Scale bar is 0.5 cm. (c) Transfer characteristics of a
representative In2O3 transistor under relaxed state, bent with a radius of ∼0.2 mm and after
crumpling, respectively. Mobilities and threshold voltages obtained in a released state
during (d) 100 bending cycles and (e) 100 crumpling cycles. (f) Transfer characteristics of
an In2O3 transistor measured immediately after fabrication (Day 0) and after immersed in
1 × PBS for 1, 2, 3, 5, and 7 day(s).
Serotonin and dopamine are important neurotransmitters for brain information
processing.
40
Imbalances in these neurotransmitters can cause serious mental
93
disorders such as Parkinson’s disease, Alzheimer’s disease, and drug
addiction.
13
Therefore, monitoring the concentrations of neurotransmitters is
of great importance in diagnosing such mental illnesses and in-depth brain
study. However, real-time monitoring of neurotransmitters is challenging
because of the fast response time in the neurotransmissions
41
and typically
low concentrations of such molecules.
42
Recent emerging researches typically
used microdialysis and electrochemical sensors to measure the concentrations
of neurotransmitters.
43
However, drawbacks remain these two techniques. For
microdialysis, the body fluids are collected through a semi-permeable probe
and analyzed by off-body instruments such as high-performance liquid
chromatography (HPLC) with mass spectrometry (MS), so the turn-around
time (in the order of minutes) is rather long for real-time monitoring. For
electrochemical sensors with fast-scan cyclic voltammetry, the selectivity is
poor when the oxidation/reduction potentials of targeted molecules are similar.
Enzyme-based electrochemical sensors provide an alternative approach to
address those issues, but suffer from instability and degradation of enzymatic
activities.
44
94
4.4 Multiplexed sensing of serotonin and dopamine
In
2
O
3
field-effect-transistors functionalized with aptamers are promising for
continuous monitoring of various neurotransmitters in real time, as they
benefit from highly selective nature of aptamers with short response time and
ultra-low detection limits. As serotonin and dopamine molecules both have
only one positive charge at physiological pH, the impact of those charges on
the chemical gate effect on the In
2
O
3
nanoribbon should be very weak at very
low target concentrations, e.g. 10 fM. By using the charges induced by
reorientations of aptamers on the surface of In
2
O
3
nanoribbons, the sensing
signals can be amplified.
16
Moreover, this method can overcome the Debye
length limitations of FET-based sensors by the amplification. As shown in
Figure 4.4a, the negatively charged backbones of the serotonin aptamers are
hypothesized to move away from the In
2
O
3
surface upon target capture. In this
way, the electrostatic repulsion between the electrons in n-type semiconductor
channels and negatively charged aptamers would decreased, and thus the
conductance in the channel would increase. The results of serotonin detection
using three In
2
O
3
nanoribbon biosensors functionalized with serotonin
aptamers are shown in Figure 4.4b. To evaluate sensing reliability in an
undiluted biological matrix, serotonin was dissolved in 1 × artificial
cerebrospinal fluid (aCSF). Initially, in Figure 4.4b, the devices were operated
95
under 0.25 V gate bias applied using the gold side gate, and the channels were
submerged in 1 × aCSF to obtain the baseline current. After changing the
electrolyte to 10 fM serotonin in 1 x aCSF at 150 s, the sensing signal bumped
up by ~1 % of the starting current. Then we changed buffer with different
serotonin concentration ranging from 0.1 pM to 1 μM sequentially, and
stepped current changes were observed. To confirm the reproducibility of the
detection, we conducted two more rounds of sensing with six different devices.
The relationship between the serotonin concentration and the saturated current
response from total 9 different devices is concluded in Figure 4.4c. We also
performed the same sensing procedure on unfunctionalized devices (i.e.,
without serotonin aptamers) as a negative control experiment (Figure 4.5), and
the sensing results are also plotted in Figure 4.4c for comparison, showing
minimal signals compared with functionalized biosensors. Our aptamer-
functionalized sensors can differentiate the serotonin concentration as low as
10 fM. In contrast, the negatively charged backbones of the dopamine aptamer
are hypothesized to move closer to the n-type semiconductor channels, and
thus it would increase the electrostatic repulsion and decrease the conductance
in the In
2
O
3
nanoribbon (Figure 4.4d). The results of real-time sensing of
dopamine are shown in Figure 4.4e, and a summary of dopamine
concentrations and the corresponding responses from 9 different devices was
96
plotted in Figure 4.4f. The limit of detection on real-time sensing of dopamine
was also demonstrated to be 10 fM. Those real-time I
DS
-t curves supported
our hypothesis regarding the conformational changes of serotonin and
dopamine aptamers, providing a complementary and more straightforward
evidence in addition to the direction changes in I
DS
-V
GS
curves, as shown in
our previous work.
16
Figure 4. 4 Experimental characterizations of serotonin and dopamine sensors. (a) Aptamer
stem-loops reorient away from semiconductor channels, thereby increasing
transconductance. (b) Real-time sensing results from three In2O3 nanoribbon biosensors
functionalized with serotonin aptamers. The devices showed responses to serotonin (in 1 ×
artificial cerebrospinal fluid) concentrations ranging from 10 fM to 1 μM. (c) The
relationship between the serotonin concentration and the saturated current response from
total 9 different devices. Negative control experiment results from unfunctionalized
devices were also plotted in this figure. (d) Aptamers reorient closer to FETs to deplete
channels electrostatically. (e) The results of real-time sensing of dopamine. (f) A summary
of dopamine concentration and the corresponding responses from 9 different devices. All
the devices were operated under VDS = 0.2 V and V GS = 0.25 V.
97
Figure 4. 5 Control experiments for serotonin and dopamine sensing. The same sensing
procedure on unfunctionalized devices (a) without serotonin aptamers and (b) without
dopamine aptamers as a negative control experiment.
Figure 4.6a shows the resistance of the gold temperature sensor in response to
the temperature of the physiological solution ranging from 20 to 50 °C. The
results showed good linearity with a sensitivity of ~ 4 Ω/°C. Figure 4.6b
displays the sensing results of unfunctionalized In
2
O
3
nanoribbon devices in
response to commercial pH solutions ranging from pH 10 to pH 4. Devices
showed increase in conduction when the pH value of the solution decreased,
because hydroxyl groups on the nanoribbon surface were protonated due to
more H
+
ions in the solution, resulting in a positive gating effect on the
channel area of the n-type In
2
O
3
nanoribbon transistors. The multiplexed
sensors can be comfortably attached onto an artificial brain as shown in Figure
4.6c. To demonstrate the on-body sensing capability, we imitated the data
collection on an artificial brain with the biosensor facing up. During the
98
multiplexed sensing procedure, a PDMS stamp with a microwell was
laminated onto the devices to collect fluids and passivate devices. Figure 4.6d
shows the ex-situ multiplexed sensing results in 1 × aCSF. Two different
groups In
2
O
3
nanoribbon transistors on the same sensor chip were
functionalized with serotonin and dopamine aptamers, respectively, to
achieve the multiplexed sensing. After using indium wires to connect the
bonding pads to our measurement unit, we constantly flowed artificial
cerebrospinal fluids through the sensing area, covering pH sensors, serotonin
sensors, dopamine sensors, and temperatures sensors at the same time. After
obtaining a stable baseline current in 1 × aCSF with pH value equivalent to
7.4, we flowed artificial cerebrospinal fluids spiked with 1 pM serotonin, 1
pM dopamine, 1 nM serotonin, and 1 nM dopamine, sequentially. Based on
the results, our sensors were able to distinguish serotonin from dopamine in
measurements of interneuronal signaling. In addition, the incorporation of the
pH sensor into our multiplexed sensor array can help us distinguish response
due to pH fluctuation from response due to neurotransmitter sensing. For
example, when we introduced 1 pM and 1 nM serotonin, only the serotonin
sensor showed response. In contrast, when we introduced 1 × aCSF with pH
7.3 at t = 1100 s, all three In
2
O
3
sensors showed response and we can thus
attribute such response to pH change. Overall, it demonstrates that our ultra-
99
flexible sensing platform has the potential for real-time multiplexed
monitoring of the key information in the cerebrospinal fluids.
Figure 4. 6 (a) The resistance of the temperature sensor in buffer solution with temperature
ranging from 20 °C to 50 °C. (b) Real-time pH sensing with three unfunctionalized In2O3
nanoribbon devices exposed to commercial buffer solutions with pH value from 10 to 4.
(c) Biosensor foil conformally attached on an artificial PDMS brain. The devices were
connected out using Indium wire for further electrical measurements. Scale bar is 1 cm. (d)
Multiplexed sensing with temperature, pH, serotonin and dopamine. 1 × artificial
cerebrospinal fluid (pH = 7.4) spiked with 1 pM serotonin, 1 pM dopamine, and 1 nM
serotonin, and 1 nM dopamine were sequentially dropped on the sensors, and only devices
functionalized with target aptamers showed responses to corresponding target molecules.
All sensors responded to pH changes (pH value from 7.4 to 7.3). The devices were operated
under VDS = 0.2 V and V GS = 0.25 V.
100
4.5 Summary
In summary, we have demonstrated ultra-flexible conformal plastic foil with
an array of four different sensors for multiplexed and selective sensing. The
sensors displayed excellent flexibility against mechanical deformation and
long-term stability in high ionic solutions. Our approaches achieved real-time
sensing of neurotransmitters at ultra-low concentrations of target molecules
with good selectivity. Importantly, our prototype sensing platform can
function properly on uneven surfaces, and it has the potential to be combined
with integrated circuits that can be used for devices operation and signal
processing. Thus, this technology can be delivered to practical sensors for
wearable or implantable applications. The above results showcase the
potential application that our multiplexed neurotransmitter sensors, which
could be used to advance the understanding of the brain and benefit diagnosis
and treatment of neurological disorders and diseases.
4.6 References
1. Kim, D.-H.; Viventi, J.; Amsden, J. J.; Xiao, J.; Vigeland, L.; Kim, Y.-S.; Blanco,
J. A.; Panilaitis, B.; Frechette, E. S.; Contreras, D., Dissolvable Films of Silk Fibroin for
Ultrathin Conformal Bio-Integrated Electronics. Nat. Mater. 2010, 9, 511−517.
101
2. Minev, I. R.; Musienko, P.; Hirsch, A.; Barraud, Q.; Wenger, N.; Moraud, E. M.;
Gandar, J.; Capogrosso, M.; Milekovic, T.; Asboth, L., Electronic Dura Mater for Long-
Term Multimodal Neural Interfaces. Science 2015, 347, 159−163.
3. Wang, S.; Xu, J.; Wang, W.; Wang, G.-J. N.; Rastak, R.; Molina-Lopez, F.;
Chung, J. W.; Niu, S.; Feig, V. R.; Lopez, J., Skin Electronics from Scalable Fabrication
of an Intrinsically Stretchable Transistor Array. Nature 2018, 555, 83−88.
4. Xie, C.; Liu, J.; Fu, T.-M.; Dai, X.; Zhou, W.; Lieber, C. M., Three-Dimensional
Macroporous Nanoelectronic Networks as Minimally Invasive Brain Probes. Nat. Mater.
2015, 14, 1286−1292.
5. Zhou, T.; Hong, G.; Fu, T.-M.; Yang, X.; Schuhmann, T. G.; Viveros, R. D.;
Lieber, C. M., Syringe-Injectable Mesh Electronics Integrate Seamlessly with Minimal
Chronic Immune Response in the Brain. Proc. Natl. Acad. Sci. 2017, 114, 5894−5899.
6. Luan, L.; Wei, X.; Zhao, Z.; Siegel, J. J.; Potnis, O.; Tuppen, C. A.; Lin, S.;
Kazmi, S.; Fowler, R. A.; Holloway, S., Ultraflexible Nanoelectronic Probes Form
Reliable, Glial Scar–Free Neural Integration. Sci. Adv. 2017, 3, e1601966.
7. Yang, H.; Sampson, M. M.; Senturk, D.; Andrews, A. M., Sex-and SERT-
Mediated Differences in Stimulated Serotonin Revealed by Fast Microdialysis. ACS
Chem. Neurosci. 2015, 6, 1487−1501.
8. Yang, H.; Thompson, A. B.; McIntosh, B. J.; Altieri, S. C.; Andrews, A. M.,
Physiologically Relevant Changes in Serotonin Resolved by Fast Microdialysis. ACS
Chem. Neurosci. 2013, 4, 790−798.
9. Robinson, D. L.; Venton, B. J.; Heien, M. L.; Wightman, R. M., Detecting
Subsecond Dopamine Release with Fast-Scan Cyclic Voltammetry in vivo. Clin. Chem.
2003, 49, 1763−1773.
10. Rodeberg, N. T.; Sandberg, S. G.; Johnson, J. A.; Phillips, P. E.; Wightman, R.
M., Hitchhiker’s Guide to Voltammetry: Acute and Chronic Electrodes for in vivo Fast-
Scan Cyclic Voltammetry. ACS Chem. Neurosci. 2017, 8, 221−234.
11. Van Schoors, J.; Viaene, J.; Van Wanseele, Y.; Smolders, I.; Dejaegher, B.;
Vander Heyden, Y.; Van Eeckhaut, A., An Improved Microbore UHPLC Method with
Electrochemical Detection for the Simultaneous Determination of Low Monoamine
102
Levels in in vivo Brain Microdialysis Samples. J. Pharm. Biomed. Anal. 2016, 127,
136−146.
12. Shao, Y.; Wang, J.; Wu, H.; Liu, J.; Aksay, I. A.; Lin, Y., Graphene Based
Electrochemical Sensors and Biosensors: a Review. Electroanalysis 2010, 22,
1027−1036.
13. Si, B.; Song, E., Recent Advances in the Detection of Neurotransmitters.
Chemosensors 2018, 6, 1.
14. Lin, Y.; Yu, P.; Mao, L., A Multi-Enzyme Microreactor-Based Online
Electrochemical System for Selective and Continuous Monitoring of Acetylcholine.
Analyst 2015, 140, 3781−3787.
15. Ispas, C. R.; Crivat, G.; Andreescu, S., Recent Developments in Enzyme-Based
Biosensors for Biomedical Analysis. Anal. Lett. 2012, 45, 168−186.
16. Nakatsuka, N.; Yang, K.-A.; Abendroth, J. M.; Cheung, K. M.; Xu, X.; Yang, H.;
Zhao, C.; Zhu, B.; Rim, Y. S.; Yang, Y., Aptamer–Field-Effect Transistors Overcome
Debye Length Limitations for Small-Molecule Sensing. Science 2018, 362, 319−324.
17. Kim, J.; Rim, Y. S.; Chen, H.; Cao, H. H.; Nakatsuka, N.; Hinton, H. L.; Zhao,
C.; Andrews, A. M.; Yang, Y.; Weiss, P. S., Fabrication of High-Performance Ultrathin
In2O3 Film Field-Effect Transistors and Biosensors Using Chemical Lift-Off
Lithography. ACS Nano 2015, 9, 4572−4582.
18. Willner, I.; Zayats, M., Electronic Aptamer-Based Sensors. Angew. Chem., Int.
Ed. 2007, 46, 6408−6418.
19. Hamaguchi, N.; Ellington, A.; Stanton, M., Aptamer Beacons for the Direct
Detection of Proteins. Anal. Biochem. 2001, 294, 126−131.
20. Nakatsuka, N.; Cao, H. H.; Deshayes, S.; Melkonian, A. L.; Kasko, A. M.; Weiss,
P. S.; Andrews, A. M., Aptamer Recognition of Multiplexed Small-Molecule-
Functionalized Substrates. ACS Appl. Mater. Interfaces 2018, 10, 23490−23500.
21. Vacic, A.; Criscione, J. M.; Rajan, N. K.; Stern, E.; Fahmy, T. M.; Reed, M. A.,
Determination of Molecular Configuration by Debye Length Modulation. J. Am. Chem.
Soc. 2011, 133, 13886−13889.
103
22. Hammock, M. L.; Knopfmacher, O.; Naab, B. D.; Tok, J. B.-H.; Bao, Z.,
Investigation of Protein Detection Parameters Using Nanofunctionalized Organic Field-
Effect Transistors. ACS Nano 2013, 7, 3970−3980.
23. So, H. M.; Park, D. W.; Jeon, E. K.; Kim, Y. H.; Kim, B. S.; Lee, C. K.; Choi, S.
Y.; Kim, S. C.; Chang, H.; Lee, J. O., Detection and Titer Estimation of Escherichia Coli
Using Aptamer-Functionalized Single−Walled Carbon−Nanotube Field-effect
Transistors. Small 2008, 4, 197−201.
24. Weiss, P.; Trevor, P.; Cardillo, M., Gas–Surface Interactions on InP Monitored by
Changes in Substrate Electronic Properties. J. Chem. Phys. 1989, 90, 5146−5153.
25. Liu, Q.; Aroonyadet, N.; Song, Y.; Wang, X.; Cao, X.; Liu, Y.; Cong, S.; Wu, F.;
Thompson, M. E.; Zhou, C., Highly Sensitive and Quick Detection of Acute Myocardial
Infarction Biomarkers Using In2O3 Nanoribbon Biosensors Fabricated Using Shadow
Masks. ACS Nano 2016, 10, 10117−10125.
26. Tang, H.; Lin, P.; Chan, H. L.; Yan, F., Highly Sensitive Dopamine Biosensors
Based on Organic Electrochemical Transistors. Biosens. Bioelectron. 2011, 26,
4559−4563.
27. Ishikawa, F. N.; Curreli, M.; Olson, C. A.; Liao, H.-I.; Sun, R.; Roberts, R. W.;
Cote, R. J.; Thompson, M. E.; Zhou, C., Importance of Controlling Nanotube Density for
Highly Sensitive and Reliable Biosensors Functional in Physiological Conditions. ACS
Nano 2010, 4, 6914−6922.
28. Allen, B. L.; Kichambare, P. D.; Star, A., Carbon Nanotube Field-Effect-
Transistor-Based Biosensors. Adv. Mater. 2007, 19, 1439−1451.
29. Wu, L.; Chu, H.; Koh, W.; Li, E., Highly Sensitive Graphene Biosensors Based
on Surface Plasmon Resonance. Opt. express 2010, 18, 14395−14400.
30. Wang, L.; Xiong, Q.; Xiao, F.; Duan, H., 2D Nanomaterials Based
Electrochemical Biosensors for Cancer Diagnosis. Biosens. Bioelectron. 2017, 89,
136−151.
31. Sarkar, D.; Liu, W.; Xie, X.; Anselmo, A. C.; Mitragotri, S.; Banerjee, K., MoS2
Field-Effect Transistor for Next-Generation Label-Free Biosensors. ACS Nano 2014, 8,
3992−4003.
104
32. Duan, X.; Li, Y.; Rajan, N. K.; Routenberg, D. A.; Modis, Y.; Reed, M. A.,
Quantification of the Affinities and Kinetics of Protein Interactions Using Silicon
Nanowire Biosensors. Nat. Nanotechnol. 2012, 7, 401−407.
33. Stern, E.; Vacic, A.; Li, C.; Ishikawa, F. N.; Zhou, C.; Reed, M. A.; Fahmy, T.
M., A Nanoelectronic Enzyme-Linked Immunosorbent Assay for Detection of Proteins in
Physiological Solutions. Small 2010, 6, 232−238.
34. Zhao, C.; Xu, X.; Yang, Q.; Man, T.; Jonas, S. J.; Schwartz, J. J.; Andrews, A.
M.; Weiss, P. S., Self-Collapse Lithography. Nano Lett. 2017, 17, 5035−5042.
35. Zhao, C.; Xu, X.; Bae, S.-H.; Yang, Q.; Liu, W.; Belling, J. N.; Cheung, K. M.;
Rim, Y. S.; Yang, Y.; Andrews, A. M., Large-Area, Ultrathin Metal-Oxide
Semiconductor Nanoribbon Arrays Fabricated by Chemical Lift-Off Lithography. Nano
Lett. 2018, 18, 5590−5595.
36. Lewis, J.; Grego, S.; Chalamala, B.; Vick, E.; Temple, D., Highly Flexible
Transparent Electrodes for Organic Light-Emitting Diode-Based Displays. Appl. Phys.
Lett. 2004, 85, 3450−3452.
37. Park, J.-S.; Kim, T.-W.; Stryakhilev, D.; Lee, J.-S.; An, S.-G.; Pyo, Y.-S.; Lee,
D.-B.; Mo, Y. G.; Jin, D.-U.; Chung, H. K., Flexible Full Color Organic Light-Emitting
Diode Display on Polyimide Plastic Substrate Driven by Amorphous Indium Gallium
Zinc Oxide Thin-Film Transistors. Appl. Phys. Lett. 2009, 95, 013503.
38. Fan, Z.; Razavi, H.; Do, J.-w.; Moriwaki, A.; Ergen, O.; Chueh, Y.-L.; Leu, P.
W.; Ho, J. C.; Takahashi, T.; Reichertz, L. A., Three-Dimensional Nanopillar-Array
Photovoltaics on Low-Cost and Flexible Substrates. Nat. Mater. 2009, 8, 648−653.
39. Liu, Q.; Liu, Y.; Wu, F.; Cao, X.; Li, Z.; Alharbi, M.; Abbas, A. N.; Amer, M. R.;
Zhou, C., Highly Sensitive and Wearable In2O3 Nanoribbon Transistor Biosensors with
Integrated On-Chip Gate for Glucose Monitoring in Body Fluids. ACS Nano 2018, 12,
1170−1178.
40. Muller, P.; Seeman, P., Brain Neurotransmitter Receptors After Long-Term
Haloperidol: Dopamine, Acetylcholine, Serotonin, α-Noradrenergic and Naloxone
Receptors. Life Sci. 1977, 21, 1751−1758.
41. Venton, B. J.; Wightman, R. M., Psychoanalytical Electrochemistry: Dopamine
and Behavior. Anal. Chem. 2003, 75, 414−421.
105
42. Perry, M.; Li, Q.; Kennedy, R. T., Review of Recent Advances in Analytical
Techniques for the Determination of Neurotransmitters. Anal. Chim. Acta 2009, 653,
1−22.
43. Kennedy, R. T.; Watson, C. J.; Haskins, W. E.; Powell, D. H.; Strecker, R. E., In
vivo Neurochemical Monitoring by Microdialysis and Capillary Separations. Curr. Opin.
Chem. Biol. 2002, 6, 659−665.
44. Hanssen, B. L.; Siraj, S.; Wong, D. K., Recent Strategies to Minimise Fouling in
Electrochemical Detection Systems. Rev. Anal. Chem. 2016, 35, 1−28.
106
Chapter 5 Fully-printed all-solid-state
organic flexible artificial synapse for
neuromorphic computing
5.1 Introduction
The human brain can manage efficient information processing, learning and
memory with extremely low energy-consumption. Artificial intelligence (e.g.
AlphaGo) based on multi-core chips with traditional CMOS (complementary-
metal-oxide-semiconductor) has exhibited the revolutionary computing
power of neural networks.
1, 2
However, due to the physical separation of
computing and memory units (von Neumann bottleneck), traditional CMOS
devices and circuits are not ideal for neuromorphic computing in regard to
energy consumption and design complexity.
3
Inspired by synaptic activity in
biological processes, electronic devices with tunable resistance, such as
memristors,
4-13
phase-change memory,
14-16
field-effect transistors,
17, 18
spintronic,
19-21
and ferroelectric devices,
22, 23
have been widely demonstrated
to emulate synaptic operations, including long-term potentiation/depression
(LTP/LTD), short-term potentiation/depression (STP/STD), and low power
consumption, with the device conductance representing the synaptic weight.
107
Although memristors have been developed as non-volatile resistive random-
access memory with high endurance and fast read/write abilities,
24, 25
these
devices cannot achieve long retention time and low-power switching at the
same time.
Organic electrochemical devices can overcome the above dilemma with a
unique switching mechanism.
26-28
A recently developed conductance-tuning
mechanism
26, 29, 30
can make organic electronic devices work as a battery: upon
applying an electric voltage pulse to the device, the proton concentration in
the channel material changes due to redox reaction, thus, changing the film
conductance; a counter-redox reaction in the gate can keep electrical
neutrality through the device. As a result, the proton concentration in the
organic film changes, thus, film conductance changes. Due to their
biocompatibility, low-power consumption, and flexibility, organic
memristive devices have great potential to act as memory and perform
analogue information processing in wearable electronics and brain-machine
interface applications.
26, 29-31
Recently, there have been great progress on
artificial synapses based on organic field-effect transistors.
32, 33
Although
various synaptic behaviour and flexibility have been demonstrated, switching
between multi-stage has not been proved yet. In spite of the great potential of
108
using multi-stage organic electrochemical devices for neuromorphic
computing, the progress has been hindered by the difficulty in making such
device. Previous demonstrations were often limited to single or a few devices
and those devices were usually bulky in size or may even involve the use of
liquid electrolyte, making integration of arrays of devices nearly impossible.
Here, we demonstrate the use of screen printing to produce an array of all-
solid, three-terminal neuromorphic devices
with a layer of
polydiallyldimethylammonium chloride (PDADMAC) electrolyte on top of
two poly(3,4-ethylene dioxythiophene):polystyrene sulfonate (PEDOT:PSS)
films. Screen printing is a cost-effective and scalable technology compatible
with organics with high throughput and low-temperature processing.
34, 35
The
fabricated devices can successfully emulate the basic characteristics of
biological synapses, including long-term potentiation/depression, spike-
timing-dependent plasticity (STDP), paired-pulse facilitation (PPF), and
ultralow energy consumption. Our all-solid-state devices pave the way to low-
cost and highly scalable fabrication of flexible neuromorphic device arrays,
which would allow the integration of electronics with on-board computing
and learning capability in implantable prosthetics or other wearable electronic
109
systems. Furthermore, the demonstrated flexibility of the devices shows the
potential for three-dimensional integrated system.
5.2 Fabrication of organic flexible artificial synapse
The fabrication process of our organic-based flexible neuromorphic devices
is illustrated in Figure 5.1a. We developed a 3-step printing process with high
yield and high uniformity, using silver conductive ink as the metal contact,
using PEDOT:PSS as the postsynaptic and presynaptic electrodes, and using
PDADMAC film as the solid electrolyte. Silver nanoparticle ink was first
printed through a screen mesh on a flexible polyethylene terephthalate (PET)
substrate. To enable aqueous ink deposition, we modified the surface energy
of the patterned silver layer and PET surface with oxygen plasma treatment
(Figure 5.2). A thin-film of water-based PEDOT:PSS dispersions was then
screen printed as the active layer, followed by a double-layer printing of a
thick layer of PDADMAC electrolyte. PEDOT is an electronic semiconductor
degenerately doped by the ion-conducting electrical insulator PSS. The
electrical conductivity of PEDOT:PSS based polymer can be tuned by
localization or delocalization of electrons along the polymer backbone (Figure
5.3) , and the conductivity is considered as postsynaptic weight of the
110
connection between two neuros, a key feature of a synapse. The device was
laterally gated with a solid cationic polyelectrolyte layer, PDADMAC.
Figure 5. 1 Fully-printed organic neuromorphic devices. a) Schematic illustration of the
key fabrication procedures for organic neuromorphic devices with printing technology. b),
c) Schematic showing switching mechanism in “read” and “write” operations in the organic
neuromorphic devices. During the “read” operation (left), the external switch is open to
forbid electron flow, leading to a stable channel conductance. During the “write” operation
(right), the external switch is closed, permitting electrons to flow in and out of the gate,
PEDOT:PSS
PDADMAC
a
b
g
c d
e f
Screen Printing of PDADMAC
Electrolyte Layer
Screen Printing of PEDOT:PSS
Pre/Post Synapse Layer
Screen Printing of Silver Contact
Layer on PET Substrate
111
resulting in changes of channel conductance. d) Image of a device array with 45 organic
neuromorphic devices. Scale bar is 1 cm. e), f) Magnified image of one device in the array,
clearly showing the PEDOT:PSS layer before electrolyte layer patterning (e) and the
PDADMAC film after patterning (f). Both scale bars represent 2 mm. g) Photograph of an
array of organic neuromorphic devices on flexible substrate while being bent. Scale bar is
1 cm.
The programming (“read” and “write”) of the neuromorphic devices is similar
to charging and discharging a battery. During a “read” operation, the external
switch is open and there is no electric signal flow, and therefore the proton
concentration remains unaltered in each layer, as exhibited in Figure 5.1b. To
achieve “write” operation, the switch is closed and a signal from the gate
electrode is regarded as the presynaptic stimulus, as exhibited in Figure 1c.
When applying a positive presynaptic pulse V
pre
, cations are injected into the
postsynaptic electrode through the presynaptic electrode and the electrolyte.
Thus, the organic device is switched to a low conductance stage since the
number of holes is reduced in the postsynaptic electrode due to protonation of
the PEDOT film. After this “write” step, the device is disconnected, and the
energy barrier between the channel and electrolyte forbids electronic charge
transport, keeping the electrode conductance state in a non-volatile way.
112
Figure 5. 2 Surface modification with oxygen plasma. Contact angles of water on
unmodified PET substrate (a), on oxygen plasma-modified PET (b), on unmodified silver
conductive film (c), and on oxygen plasma-modified silver conductive film (d). After being
treated with oxygen plasma (100 W, 150 mTorr) for 30 seconds, the contact angle of water
changes from 71 to 30 on PET substrate, and from 138 to 89 on silver film, which
indicates the surface of both PET and silver film become more hydrophilic. e)
Figure 5. 3 Oxidation and reduction reaction of an PEDOT:PSS based all-solid organic
neuromorphic devices. The molecular structures of PEDOT:PSS and PDADMAC are
illustrated in this figure. Upon applying a negative Vpre to the PEDOT:PSS electrode,
protons flow from the postsynaptic electrode into the presynaptic electrode through
PDADMAC electrolyte, resulting in deprotonation of the PEI, and further cause the
oxidation of PEDOT due to charge neutrality. This causes holes to be generated on the
PEDOT backbone, thereby reducing the electronic resistivity of the postsynaptic
PEDOT
PSS
PEDOT+
PSS-
PDADMAC
cationic polyelectrolyte
Oxidation
Reduction
113
electrode. The reaction is reversed when applying a positive presynaptic potential. The
charge transfer is marked in red in the figure.
Using the screen printing process, we fabricated a sheet comprising a flexible
array of 45 three-terminal neuromorphic devices (Figure 5.1d), with channel
width of 100 µm and channel length ranging from 200 µm to 3 mm. In a
magnified optical image of one device in the array with and without the
PDADMAC layer (Figure 5.1e and 5.1f), all components are fully patterned
and well aligned. Owing to the intrinsic flexibility of organic materials, the
all-solid-state neuromorphic devices are fully compatible with flexible
substrates (Figure 5.11g), opening up opportunities to work as memory and
analogue information processor for wearable electronics.
5.3 Long-term potentiation/depression (LTD/LTP)
of fully-printed organic artificial synapses
The long-term potentiation/depression (LTD/LTP) has been considered as one
of the most important forms of plasticity that is closely related to the synaptic
activity and signal transmission between two neurons.
36
To mimic excitatory
and inhibitory synapses in organisms, the LTD/LTP behaviors of our organic
neuromorphic devices are experimentally analyzed. Figure 5.4a shows the
114
result for a neuromorphic device (channel length = 250 µm, channel width =
100 µm) measured with a series of 100 identical positive pulses (10 mV, 1s),
followed by a series of 100 negative voltage pulses (-10mV, 1s). As the
number of positive pulses increases, the device becomes more conductive,
representing continuous tunability of 100 distinct conductance states. The
inset images exhibit the discrete conductance states under LTD/LTP process
with mean step ~ 1.1 S. After applying 100 negative pulses, the
neuromorphic device restored to its initial low-conductance state. We cycled
the device in the LTD/LTP process more than 10 times using the above
method, as exhibited in Figure 5.4b. The cycling data reveal good
electrochemical stability, repeatability of the synaptic characteristics, as well
as fine synaptic resolution (analogue programmability). Because of the good
flexibility and uniformity of the materials, the organic neuromorphic devices
exhibit stable and reproducible potentiation-depression cyclic behavior,
regardless of mechanical bending.
It is worth of noting that the changes in the synaptic weight is non-volatile, as
shown in Figure 5.4c, we programmed 10 pulses ranging from 10 mV to 100
mV, followed with ~ 10 s relaxation time after each pulse. The current curve
shows ten distinct conductance states, and during the “read” state, after the
pulse is applied, the conductance did not show a significant drop. The study
115
continues with a signal identical to the first pulse. At ~150 s, the external
switch was closed and a - 10 mV pulse was applied simultaneously, protons
can migrate between pre-synapse, electrolyte, and post-synapse freely during
the “write” operation, and thus the proton concentration can return to the
initial state, which indicating the good stability and reproducibility of our
devices. To further demonstrate the non-volatility of our devices, we
monitored 8-hour retention of both the low resistance state (LRS) and high
resistance state (HRS) (Figure 5.4d). Only ~ 3 S (~ 1 %) change in the
synaptic weight was observed in the retention test, which demonstrates that
our devices have excellent non-volatile long-term memory.
116
Figure 5. 4 Long-term neuromorphic behavior. a) Long-term potentiation and depression
exhibiting 100 discrete states when the device is programmed with presynaptic pulses. The
two insets are zoom-in plots showing the individual states. b) LTP/LTD cycling stress tests
when the organic neuromorphic device is in a relaxed state (upper panel) and after 500
bending cycles (bottom panel). c) State retention for organic neuromorphic devices. The
conductance is monitored for 10 seconds after a 1 s pulse is applied to change the states.
The pulse amplitudes are as labelled. Different pulse amplitude can switch the conductance
200
300
Conductance (S)
0 200 400
150
200
250
Conductance (S)
Time (s)
0 50 100 150
16
17
18
19
20
Current (A)
Time (s)
0 500 1000 1500 2000
200
300
Pulse number
-10 mV
-20 mV
-30 mV
-40 mV
-50 mV
-60 mV
-70 mV
-80 mV
-90 mV
-100 mV
-10 mV
Step ~ 1.1 uS
V
post
= 100 mV
-100 0 100
-5
0
5
G (S)
t (ms)
1
= 18.8 ms
2
= -15.4 ms
Step ~ 1.15 uS
0 1 2 3 4 5 6 7 8
138
141
144
237
240
243
Conductance (S)
Time (hr)
Low Resistance State
High Resistance State
After 100 bending cycles
Relaxed state
a b
c d
e f
Pre-synapse
Post-synapse
t
Presynaptic spike
Postsynaptic spike
t = t
pre
- t
post
50 mV
- 25 mV
25 mV
10 mV, 1 s
- 10 mV, 1 s
LTP
LTD
117
into different states, and after a pulse with amplitude equal to the initial one, the
conductance switched back to the state similar to the initial one. d) Retention of the HRS
and LRS currents at Vpost = 100 mV and Vpre = 0 V in an eight-hour period. e) Schematic
showing the electrical implementation for STDP measurement. The organic neuromorphic
device is connected between a pre-synaptic spike generator and a post-synaptic spike
generator. f) STDP behavior of the device stimulated with a pair of spikes with different
values of t.
The synaptic strength in biological systems can be regulated by the timing and
causality of pre- and post-synaptic spikes with the STDP rules, which is one
of the fundamental rules for emulating synapses. For our devices, the observed
STDP characteristics are similar to those in biological synapses. A pair of
pulses were applied to the pre-synaptic electrode and post-synaptic electrode
to work as pre-synaptic and post-synaptic spikes, respectively. A rectangular
wave with 25 ms duration and 50 mV amplitude was applied to the pre-
synapse, and a triangular wave of period of 200 ms and oscillating between -
25 mV and 25 mV was applied to post-synapse (ramping from 0 V to 25 mV
in 95 ms, dropping from 25 mV to -25 mV in the following 10 ms, and
ramping from -25 mV to 0 V in the last 95 ms) (Figure 5.4e). The two pulses
were separated by a time difference (spike timing), Δt = t
pre
-t
post
. A summary
of the conductance change (G) of the devices with different value of spike
timing (ranging from -100 ms to 100 ms) is shown in Figure 5.4f. When the
presynaptic spike triggered before the postsynaptic peak, the Δt was negative
and the resulted spike was positive, enabling LTP of the synapse. When the
118
presynaptic spike triggered after the postsynaptic peak, the Δt was positive
and the resulted spike was negative, enabling LTD of the synapse. In both
cases, when the pulses have high degree of relevance (absolute value of Δt is
small), the synaptic weight changes dramatically. As the time difference
increases, the change of synaptic weight gradually decreases. The time
constants extrapolated from the data are ~ 18.8 ms and 15.4 ms, which are
comparable to the response times of biological synapses.
37
5.4 Paired-pulse facilitation (PPF) of fully-printed
organic artificial synapses
Paired-pulse facilitation (PPF) is an important form of short-term synaptic
plasticity, which describes the phenomenon that the amplitude of a
postsynaptic spike evoked by a pulse increased when that pulse closely
follows a prior pulse. We realized PPF functions in our artificial synapse using
two sequential pre-synaptic spikes (50 mV, 25 ms) with a time interval (t
ranging from 1 to 500 ms) to emulate signals from different pre-synaptic
neurons, as shown in Figure 5.5a. It can be seen clearly from Figure 5.5b that
the amplitude of the second pulse is larger than that of the first one when t <
100 ms, because such short time interval is insufficient for the cations injected
119
from the first pulse to return to the electrolyte layer before the second pulse
arrives. For the time interval larger than 100 ms, there is enough time for the
injected cations move back to the electrolyte, which means the paired pulses
have low degree of relevancy. Therefore, we can consider the spikes as two
independent stimulations to the device, and no significant change in the
synaptic strength occurs.
Figure 5. 5 Paired-pulse facilitation. a) Schematic showing the electrical setup for PPF
measurement. Two pre-synaptic spike generators are probed on the pre-synaptic electrode.
The inset shows the recorded waveform of pulses applied to the devices. The pulse
amplitude is 50 mV, the pulse width is 25 ms, and the spiking timing t is ranging from 1
ms to 500 ms. b) Post-synaptic current with different spike timing. c) Post-synaptic weight
changes trigged by a paired-pulse with time interval of 25 ms. G1 and G2 represent the
conductance change of the first pulse and the second pulse, respectively. G equals the
0 200 400 600
20.5
21.0
Postsynaptic current (A)
Time (ms)
0.0 0.1 0.2 0.3
0
2
4
6
Energy (nJ)
Area (mm
2
)
0 250 500 750 1000
0
2
4
G (S)
t (ms)
PPF = C
1
C
2
τ
1
= 10 ms
τ
2
= 132 ms
0 250 500
200
210
Synaptic Weight (S)
Time (ms)
ΔG = G
2
- G
1
= 3.2 μS
G
2
G
1
2
nd
Pulse
Slope = 20.5 nJ/mm
2
Δt
Presynaptic
pulse(mV)
Pulse width = 25 ms
V
read
Pre-synapse
Post-synapse
Presynaptic spikes
a
c
b
e
d
120
difference between G2 and G1. d) Paired pulse facilitation with different time intervals. An
exponential fit is applied to obtain two characteristic time scales. e) Switching energy
measured as a function of device area. The slope of linear fit is 20.5 nJ/mm
2
.
The PPF effect can be better illustrated by calculating the difference between
the two conductance peaks generated by the first pulse (G
1
) and the second
pulse (G
2
). A typical PPF curve of the organic device is shown in Figure 5.5c.
When the time interval between the paired pulses is 25 ms, the amplitude of
the second postsynaptic peak was ~ 3.2 S higher than the amplitude triggered
by the first spike. The dependence of the postsynaptic weight enhancement on
the pulse interval (Figure 5.5d) exhibits a similar trend to that observed in
biological systems.
36
The two-phase behavior can be fitted well by a double
exponential function: where Δt is the pulse interval time, C
1
and C
2
are the
initial facilitation magnitudes of the respective phases, and
1
and
2
are the
characteristic relaxation times of the respective phases. In the fitting,
1
= 10
ms and
2
= 240 ms, which are comparable to those of a biological synapse.
38
The conductance changes are proportional to the presynaptic pulse amplitude
and duration (Figure 5.6), indicating a good analogue programmability of the
short-term synaptic plasticity. The energy consumption for a single operation
can be found by calculating the power dissipation at each time point (dE = V
I dt) and taking an integral over the operation time. The switching energy
121
is proportional to the channel area with a slope of 20.5 nJ/mm
2
, as displayed
in Figure 5.5e. The power of the smallest device was measured to be ~ 200 pJ,
demonstrating the extraordinary low switching energy of our devices. In
comparison, the energy consumption of 45 nm silicon CMOS devices is 931
J/mm
2
with supply voltage at 1.1 V. The hardware performance of silicon
implementations is obtained by synthesizing with the 45 nm Nangate Open
Cell Library
39
using Synopsys Design Compiler. The proposed flexible
artificial synapses are suitable to be operated in ultra-low voltage regimes,
indicating these devices have great potential for portable and low-power
applications.
Figure 5. 6 Change in postsynaptic conductance as a function of presynaptic pulse duration
(a) and amplitude (b). The measured excitatory post-synaptic currents (EPSC) are
converted into conductance change (G) of the post-synaptic electrode. With the
presynaptic voltage fixed at 20 mV, the G values increases from 0.5 S (inset) to 371 S
for spike duration ranges from 10 ms to 8 s, respectively. The spike voltage-dependent
Presynaptic spike duration –dependent behavior Presynaptic spike voltage –dependent behavior
2 s presynaptic pulse
0 100 200
0
2
4
6
8
10
G (S)
Presynaptic pulse length (ms)
10 20 30 40 50
0
100
200
300
G (S)
Presynaptic pulse amplitude (mV)
0 2 4 6 8
0
100
200
300
400
G (S)
Presynaptic pulse length (s)
a
b
122
EPSCs are also studied. With the presynaptic pulse duration fixed at 2 s, the G increases
from 48 S to 251 S for spike voltage ranges from 10 mV to 50 mV. As the linear fitting
shown in both figures (in red), the conductance change is a linear function of presynaptic
pulse duration and voltage.
5.5 Applications of fully-printed organic artificial
synapses
We can build logic gates by integration of two or more neuromorphic
devices.
40
Figure 5.7a and 5.7b show the schematic diagram of two artificial
synapses connected in series and in parallel, respectively. The spiking signals
(V
pre
) from two pre-synapses are applied to the gates of the devices, then the
signals summed in the dendrite of a post-synaptic neuron. As shown in Figure
5.7c, with a pulse (50 mV, 25 ms) representing “1” and no pulse representing
“0”, the binary inputs of “00”, “01”, “10”, and “11” were applied on synapse
1 and synapse 2, respectively. Only when input signals are “11”, the change
of synaptic weight in the series connected devices is larger than the threshold
value (24%), indicating the “AND” logic. For parallel connected devices, as
long as one input voltage is “1”, the signal amplitude is larger than the
threshold, indicating the “OR” logic (Figure 5.7d). The realized AND and OR
logic gates can have important implications in capturing the computing power
123
of neural system where the nonlinear and analogue mechanisms are
predominant.
Figure 5. 7 Logic circuits based on neuromorphic devices. a) b) Schematic diagram
showing the circuits used for logic gates, AND gate (series connection) and OR gate
(parallel connection), respectively. c) The change of the AND gate conductance depends
on the presynaptic inputs. When the presynaptic signals only came from synapse 1 or
synapse 2, the conductance change did not reach the threshold line. When both synapses
fired, the change of conductance passed the threshold line. d) For OR gate, even when a
single synapse fired, the change of conductance slightly passed the threshold line.
Furthermore, to fully illustrate the capability of the low noise and linearly
programmable conductance states of our neuromorphic devices, we simulated
a neural network based upon its experimentally measured properties, as
2 4 6 8 10 12 14
0
10
20
30
40
50
Synaptic weight change (%)
Time (s)
2 4 6 8 10 12 14
0
10
20
30
40
50
Synaptic weight change (%)
Time (s)
Synapse 1
Synapse 2
Synapse 1
Synapse 2
50 mV, 25 ms
Threshold line Threshold line
a b
c d
Electrolyte
PEDOT:PSS
PEDOT:PSS
Source
Gate
Drain
V post
Electrolyte
PEDOT:PSS
PEDOT:PSS
Source
Gate
Drain
V pre
V pre
Synapse 1 Synapse 2
Electrolyte
PEDOT:PSS
PEDOT:PSS
Source
Gate
Drain
V post
Electrolyte
PEDOT:PSS
PEDOT:PSS
Source
Gate
Drain
V pre
V pre
Synapse 1 Synapse 2
124
illustrated in Figure 5.8a: pixel signals of the training image were employed
as the input for the simulation. The architecture of the neural network was
described schematically in Figure 5.8b: the devices in a row are arranged by
connecting the transistor source to the same input line and connecting the
transistor gate to the same gate line, while the devices in a column are
arranged by connecting the drain to the same output line. The range of
numerical weight values was linearly scaled to the range of conductance states
of devices. To accurately account for the effects of device variations, we
extracted experimental device conductance states from 10 linear potentiation
and depotentiation cycles through the complete dynamic range based on more
than 2000 experimentally measured states (Figure 5.4b). The statistics of
device variations were concluded in Figure 5.8c. The measured non-linearity
and write noise from Figure 5.8c were fed into the software simulator such
that the application evaluation considers device-induced numerical weight
variations.
Using the designs presented so far, we performed holistic evaluation of neural
networks based on our neuromorphic devices on three popular datasets for
image recognition and face detection. Firstly, two databases of handwritten
digits (Optical Recognition of Handwritten Digits
41
and MNIST
42
) were
125
evaluated. The Optical Recognition of Handwritten Digits database contains
normalized bitmaps of handwritten digits from a total of 43 people, where 30
contributed to the training set and different 13 to the test set. 32 × 32 bitmaps
are divided into non-overlapping blocks of 4 × 4 and the number of on pixels
are counted in each block. A 64 × 50 × 10 network was configured for
evaluation. Both ideal numeric and experimentally derived results were
exhibited in Figure 5.8d. The detail of the working principle of image
recognition and face detection is described in the Experimental section. The
ideal numeric data presented an initial accuracy of 67.1 %, which quickly
raised to 87.0 % at the third training epoch and stabilized at ~ 90 % from the
5
th
to the 40
th
epoch. With the similar trend, the experimentally derived curve
presented an initial accuracy of 51.7 %, and the accuracy was stabilized ~ 88 %
after 40 epochs. On the other hand, MNSIT dataset consists of 60,000 training
data and 10,000 testing data, and each entry is a 28 × 28 grayscale image. A
deep neural network with 784 × 300 × 10 configuration was used to evaluate
the network performance. Backpropagation and gradient descent optimizer
were utilized during evaluation. As exhibited in Figure 5e, accuracy of ~ 95 %
and ~ 90 % was obtained for the ideal numeric and experimentally derived
data after 40 training epochs. Owing to the exceptional linearity and low noise
126
of our neuromorphic devices, the experimentally derived data is very close to
the accuracy limitation presented by the simulated ideal data.
Figure 5. 8 Simulation of organic neuromorphic device-based neural networks. a)
Schematic illustration of the implementation learning for pattern recognition. b) Schematic
illustration of the architectural neural network with fabricated three-terminal devices. c)
Conductance variation (G) as a function of the conductance states showing the switch
statistics of neuromorphic devices during long-term potentiation (red squares) and
depression (blue squares). d), e) Backpropagation training results using Optical
Recognition of Handwritten Digits and MNSIT database handwritten digits data-sets in the
format of 8 × 8 pixel digit image (d) and 28 × 28 pixel digit image (e). f) Backpropagation
training results for face recognition using the AT&T Laboratories Cambridge ORL
database of faces.
In addition, another dataset we used is the AT&T Laboratories Cambridge
ORL database of faces,
43
which provides typical experimental setups for face
recognition.
44
The ORL database is comprised of 400 grayscale face images
of size 92 × 112 pixels from 40 persons of different gender, ethnic background
127
and age. For some subjects, the images were taken at different times, varying
the lighting, facial expressions (open / closed eyes, smiling / not smiling) and
facial details (glasses / no glasses). We first scaled the original image from 92
× 112 to 23 × 28 to significantly reduce the number of devices required in the
first layer. The hidden layer contained 200 neurons and the output layer had
40 neurons. The face recognition result based on the ORL database is
exhibited in Figure 5.8f, and the experimentally derived data is almost
identical to the simulated ideal data, the accuracy quickly raised to ~ 80 %
within 3 training epochs and stabilized ~ 85 % after that, indicating our
architectural network is a feasible and favorable route to the practical pattern
recognition applications.
In addition to building a simulated neural network for pattern recognition and
face recognition, a dynamic process of image was realized on a 3 3
postsynaptic array with local presynaptic inputs (Figure 5.9). As shown in
Figure 5.9a, we first built a baseline with presynaptic voltage (V
pre
) = 0 V and
postsynaptic voltage (V
post
) = 100 mV, and the postsynaptic weight was 190
S. When a presynaptic signal with 100 mV amplitude and 1 s duration
arrived the corresponding presynaptic electrode, the post synapse was
switched to a higher conductance state (~ 230 S). After opening the external
128
switch, the synaptic weight got maintained in the post synapse. Based on this
result, we further addressed the pixels individually and have successfully
enabled the image processing with the capital letters “U”, “S”, and “C”, as
exhibited in Figure 5.9b. The image of a letter was divided into 9 pixels and
the information of each pixel was represented by the conductance of the
corresponding device. Notably, the memory effect of this printed devices was
non-volatile, and the information can be retained for a long time (longer than
24 hours) after removing the applied voltage (Figure 5.9c). The results that
our postsynaptic array can successfully simulate the uppercase letter “U”, “S”,
and “C”, indicate that the fully-printed all-solid-state artificial synapse array
can be used for relatively complex neuromorphic application.
Figure 5. 9 Dynamic image processes and retention. (a) Functionality test on an organic
synaptic transistor. As the presynaptic input arrived at 0 s, the conductance (synaptic
weight) changes from low conductance state (~ 190 S) to high conductance state (~ 230
S). (b) Dynamic image processes on a 3 × 3 synaptic array with local presynaptic input.
129
The diagram shows that the capital letter “U”, “S”, and “C” were memorized by the
synaptic array. (c) Retention (forgetting) evaluation of the capital letter “S” after 1, 2, 8
and 24 hours after the image was memorized by the synaptic array.
5.6 Summary
In conclusion, we demonstrate fully screen-printed, flexible, all-solid, three-
terminal organic neuromorphic devices, which can act as non-volatile
memory units and neuromorphic computing. The fabricated devices can
behave like biological synapses and exhibit the characteristics of LTP/LTD,
the STDP learning rule, PPF, and ultralow energy consumption. The
demonstrated 100 almost linear and stable conductance states suit well with
the analogue world, with no need of power- and time-inefficient analogue-to
digital converters. The all-solid-state devices pave the way to low-cost
fabrication of flexible neuromorphic device arrays, which enables correlated
learning, multi-stage trainable memory, and integration of three-dimensional
neural network. The results here provide an encouraging pathway toward
biological synaptic emulation using printed organic devices for neuromorphic
computing in application such as wearable sensors.
45
130
5.7 References
1. Merolla, P. A.; Arthur, J. V .; Alvarez-Icaza, R.; Cassidy, A. S.; Sawada, J.; Akopyan, F.;
Jackson, B. L.; Imam, N.; Guo, C.; Nakamura, Y .; Brezzo, B.; Vo, I.; Esser, S. K.; Appuswamy, R.;
Taba, B.; Amir, A.; Flickner, M. D.; Risk, W. P.; Manohar, R.; Modha, D. S. A Million Spiking-
Neuron Integrated Circuit with a Scalable Communication Network and Interface. Science 2014,
345, 668– 673.
2. Silver, D.; Huang, A.; Maddison, C. J.; Guez, A.; Sifre, L.; van den Driessche, G.;
Schrittwieser, J.; Antonoglou, I.; Panneershelvam, V.; Lanctot, M.; Dieleman, S.; Grewe, D.; Nham,
J.; Kalchbrenner, N.; Sutskever, I.; Lillicrap, T.; Leach, M.; Kavukcuoglu, K.; Graepel, T.; Hassabis,
D. Mastering the Game of Go with Deep Neural Networks and Tree Search. Nature 2016, 529,
484– 489.
3. Backus, J. Can Programming Be Liberated from the von Neumann Style? A Functional
Style and Its Algebra of Programs. Commun. ACM 1978, 21, 613– 641.
4. Strukov, D. B.; Snider, G. S.; Stewart, D. R.; Williams, R. S. The Missing Memristor Found.
Nature 2008, 453, 80– 83.
5. Yang, J. J.; Pickett, M. D.; Li, X.; Ohlberg, D. A.; Stewart, D. R.; Williams, R. S.,
Memristive Switching Mechanism for Metal/Oxide/Metal Nanodevices. Nat. Nanotechnol. 20 08,
3, 429– 433.
6. Kuzum, D.; Yu, S.; Wong, H. S. Synaptic Electronics: Materials, Devices and Applications.
Nanotechnology 2 013, 24, 382001.
7. Yang, J. J.; Strukov, D. B.; Stewart, D. R. Memristive Devices for Computing. Nat.
Nanotechnol. 2013, 8, 13– 24.
8. Shulaker, M. M.; Hills, G.; Park, R. S.; Howe, R. T.; Saraswat, K.; Wong, H.-S. P.; Mitra,
S. Three-Dimensional Integration of Nanotechnologies for Computing and Data Storage on a
Single Chip. Nature 2017, 547, 74– 78.
9. Sheridan, P. M.; Cai, F.; Du, C.; Ma, W.; Zhang, Z.; Lu, W. D. Sparse Coding with
Memristor Networks. Nat. Nanotechnol. 2017, 12, 784– 789.
10. Kim, S.; Choi, S.; Lu, W. D. Comprehensive Physical Model of Dynamic Resistive
Switching in an Oxide Memristor. ACS Nano 201 4, 8, 2369– 2376.
131
11. Zhu, X.; Lu, W. D. Optogenetics-Inspired Tunable Synaptic Functions in Memristors. ACS
Nano 2018, 12, 1242– 1249
12. Wang, Z.; Joshi, S.; Savel’ev, S.; Song, W.; Midya, R.; Li, Y .; Rao, M.; Yan, P.; Asapu, S.;
Zhuo, Y .; Jiang, H.; Lin, P.; Li, C.; Yoon, J. H.; Upadhyay, N. K.; Zhang, J.; Hu, M.; Strachan, J. P.;
Barnell, M.; Wu, Q.; Wu, H.; Williams, R. S.; Xia, Q.; Yang, J. J. Fully Memristive Neural
Networks for Pattern Classification with Unsupervised Learning. Nat. Electron. 20 18, 1, 137– 145.
13. Li, C.; Hu, M.; Li, Y .; Jiang, H.; Ge, N.; Montgomery, E.; Zhang, J.; Song, W.; Dávila, N.;
Graves, C. E.; Li, Z.; Strachan, J. P.; Lin, P.; Wang, Z.; Barnell, M.; Wu, Q.; Williams, R. S.; Yang,
J. J.; Xia, Q. Analogue signal and image processing with large memristor crossbars. Nat. Electron.
2018, 1, 52– 59.
14. Eryilmaz, S. B.; Kuzum, D.; Jeyasingh, R.; Kim, S.; BrightSky, M.; Lam, C.; Wong, H.-S.
P. Brain-Like Associative Learning Using a Nanoscale Non-Volatile Phase Change Synaptic Device
Array. Front. Neurosci. 201 4, 8, 205.
15. Wong, H.-S. P.; Raoux, S.; Kim, S.; Liang, J.; Reifenberg, J. P.; Rajendran, B.; Asheghi,
M.; Goodson, K. E. Phase Change Memory. Proc. IEEE 2010, 98, 2201– 2227.
16. Tuma, T.; Pantazi, A.; Le Gallo, M.; Sebastian, A.; Eleftheriou, E., Stochastic Phase-
Change Neurons. Nat. Nanotechnol. 2016, 11, 693– 699.
17. Zhu, L. Q.; Wan, C. J.; Guo, L. Q.; Shi, Y .; Wan, Q. Artificial Synapse Network on
Inorganic Proton Conductor for Neuromorphic Systems. Nat. Commun. 2 014, 5, 3158.
18. Sangwan, V. K.; Lee, H.-S.; Bergeron, H.; Balla, I.; Beck, M. E.; Chen, K.-S.; Hersam, M.
C. Multi-Terminal Memtransistors from Polycrystalline Monolayer Molybdenum Disulfide. Nature
2018, 554, 500– 504.
19. Locatelli, N.; Cros, V.; Grollier, J. Spin-Torque Building Blocks. Nat. Mater. 2014, 13, 11–
20.
20. Yu, G.; Upadhyaya, P.; Shao, Q.; Wu, H.; Yin, G.; Li, X.; He, C.; Jiang, W.; Han, X.; Amiri,
P. K.; Wang, K. L. Room-Temperature Skyrmion Shift Device for Memory Application. Nano Lett.
2017, 17, 261– 268.
21. Shao, Q.; Yu, G.; Lan, Y .-W.; Shi, Y .; Li, M.-Y .; Zheng, C.; Zhu, X.; Li, L.-J.; Amiri, P. K.;
Wang, K. L. Strong Rashba-Edelstein Effect-Induced Spin–Orbit Torques in Monolayer Transition
Metal Dichalcogenide/Ferromagnet Bilayers. Nano Lett. 2 016, 16, 7514– 7520.
132
22. Chanthbouala, A.; Garcia, V .; Cherifi, R. O.; Bouzehouane, K.; Fusil, S.; Moya, X.; Xavier,
S.; Yamada, H.; Deranlot, C.; Mathur, N. D. A Ferroelectric Memristor. Nat. Mater. 2012, 11, 860–
864.
23. Boyn, S.; Grollier, J.; Lecerf, G.; Xu, B.; Locatelli, N.; Fusil, S.; Girod, S.; Carrétéro, C.;
Garcia, K.; Xavier, S. Learning Through Ferroelectric Domain Dynamics in Solid-State Synapses.
Nat. Commun. 2017, 8, 14736.
24. Pershin, Y . V.; Di Ventra, M. Memory Effects in Complex Materials and Nanoscale
Systems. Adv. Phys. 201 1, 60, 145– 227.
25. Wong, H.-S. P.; Lee, H.-Y .; Yu, S.; Chen, Y .-S.; Wu, Y .; Chen, P.-S.; Lee, B.; Chen, F. T.;
Tsai, M.-J. Metal–Oxide RRAM. Proc. IEEE 2 012, 100, 1951– 1970.
26. van de Burgt, Y .; Lubberman, E.; Fuller, E. J.; Keene, S. T.; Faria, G. C.; Agarwal, S.;
Marinella, M. J.; Talin, A. A.; Salleo, A. A Non-Volatile Organic Electrochemical Device as a Low-
Voltage Artificial Synapse for Neuromorphic Computing. Nat. Mater. 2017, 16, 414– 418.
27. Xuan, Y .; Sandberg, M.; Berggren, M.; Crispin, X. An All-Polymer-Air PEDOT Battery.
Org. Electron. 2012, 13, 632– 637.
28. van de Burgt, Y .; Melianas, A.; Keene, S. T.; Malliaras, G.; Salleo, A. Organic Electronics
for Neuromorphic Computing. Nat. Electron. 20 18, 1, 386– 397.
29. Gkoupidenis, P.; Schaefer, N.; Garlan, B.; Malliaras, G. G. Neuromorphic Functions in
PEDOT:PSS Organic Electrochemical Transistors. Adv. Mater. 2015, 27, 7176– 7180.
30. Gkoupidenis, P.; Koutsouras, D. A.; Malliaras, G. G. Neuromorphic Device Architectures
with Global Connectivity Through Electrolyte Gating. Nat. Commun. 2017, 8, 15448.
31. Wu, C.; Kim, T. W.; Choi, H. Y .; Strukov, D. B.; Yang, J. J. Flexible Three-Dimensional
Artificial Synapse Networks with Correlated Learning and Trainable Memory Capability. Nat.
Commun. 2017, 8, 752.
32. Jang, S.; Jang, S.; Lee, E.-H.; Kang, M.; Wang, G.; Kim, T.-W. Ultrathin Conformable
Organic Artificial Synapse for Wearable Intelligent Device Applications. ACS Appl. Mater.
Interfaces 2019, 11, 1071– 1080.
33. Fu, Y .; Kong, L.-a.; Chen, Y .; Wang, J.; Qian, C.; Yuan, Y .; Sun, J.; Gao, Y .; Wan, Q.
Flexible Neuromorphic Architectures Based on Self-Supported Multiterminal Organic Transistors.
133
ACS Appl. Mater. Interfaces 2018, 10, 26443– 26450.
34. Cao, X.; Lau, C.; Liu, Y .; Wu, F.; Gui, H.; Liu, Q.; Ma, Y .; Wan, H.; Amer, M. R.; Zhou, C.
Fully Screen-Printed, Large-Area, and Flexible Active-Matrix Electrochromic Displays Using
Carbon Nanotube Thin-Film Transistors. ACS Nano 2 016, 10, 9816– 9822.
35. Fukuda, K.; Takeda, Y .; Yoshimura, Y .; Shiwaku, R.; Tran, L. T.; Sekine, T.; Mizukami, M.;
Kumaki, D.; Tokito, S. Fully-Printed High-Performance Organic Thin-Film Transistors and
Circuitry on One-Micron-Thick Polymer Films. Nat. Commun. 2014, 5, 4147.
36. Bliss, T. V. P.; Collingridge, G. L. A Synaptic Model of Memory: Long-Term Potentiation
in the Hippocampus. Nature 1993, 361, 31– 39.
37. Abbott, L. F.; Nelson, S. B. Synaptic Plasticity: Taming the Beast. Nat. Neurosci. 2000, 3,
1178– 1183.
38. Zucker, R. S.; Regehr, W. G. Short-Term Synaptic Plasticity. Annu. Rev. Physiol. 2 002, 64,
355– 405.
39. NanGate, S. 45nm Open Cell Library, 2008; URL: http://www.nangate.com/.
40. Liu, Y . H.; Zhu, L. Q.; Feng, P.; Shi, Y .; Wan, Q. Freestanding Artificial Synapses Based
on Laterally Proton‐Coupled Transistors on Chitosan Membranes. Adv. Mater. 2015, 27, 5599–
5604.
41. Bache, K.; Lichman, M. UCI machine learning repository. University of California,
Irvine, School of Information and Computer Sciences, 2013; URL: http://archive. ics. uci.
edu/ml, 2013.
42. Deng, L. The MNIST Database of Handwritten Digit Images for Machine Learning
Research. IEEE Signal Processing Magazine 2012, 29, 141– 142.
43. AT&T Laboratories, Cambridge ORL Database of Faces. 1992 - 1994.
44. Rubiolo, M.; Stegmayer, G.; Milone, D. Compressing Arrays of Classifiers Using Volterra-
Neural Network: Application to Face Recognition. Neural Computing and Applications 20 13, 23,
1687– 1701.
134
45. Liu, Q.; Liu, Y.; Wu, F.; Cao, X.; Li, Z.; Alharbi, M.; Abbas, A. N.; Amer, M. R.; Zhou,
C. Highly Sensitive and Wearable In2O3 Nanoribbon Transistor Biosensors with Integrated On-
Chip Gate for Glucose Monitoring in Body Fluids. ACS Nano 2018, 12, 1170– 1178.
135
Chapter 6 Conclusions and future work
6.1 Conclusions
In this thesis, nanostructure indium oxides as the active channel material in
field-effect transistors (FET) have been investigated for various sensing
applications ranging from portable biosensor for point-of-care to wearable
sensor for skin and implantable sensor. Organic non-volatile artificial
synapses for neuromorphic computing have also been investigated in this
thesis. The overarching objective of the thesis is to apply indium oxide FET
sensing technology in portable and wearable platform. To this end, scalable
processing using shadow masks was explored for both rigid and flexible
substrate. In Chapter 2, fabrication of highly uniform and scalable In
2
O
3
nanoribbon biosensor chips using two simple shadow masks to define the
position and dimension of metal electrodes and nanoribbons. The In
2
O
3
nanoribbon devices show good electrical performance and can be used to
quantitatively detect 3 cardiac biomarkers within the concentrations relevant
to clinical diagnosis with the turnaround time ~ 45 minutes using eELISA. In
Chapter 3, the In
2
O
3
FET-based wearable biosensors with on-chip gold side
gate electrodes can be used for highly sensitive detection of glucose with the
detection limit down to 10 nM. The non-invasive glucose detections in human
136
body fluids, such as tears and sweat, was also demonstrated. We further
investigated glucose sensing on an eyeball replica and on an artificial hand. In
chapter 4, ultraflexible conformal plastic foil with an array of four different
sensors for multiplexed and selective sensing was demonstrated. The sensors
displayed excellent flexibility against mechanical deformation and long-term
stability in high ionic solutions. Our approaches achieved real-time sensing of
neurotransmitters at ultra-low concentrations of target molecules with good
selectivity. In chapter 5, we demonstrate fully screen-printed, flexible, all-
solid, three-terminal organic neuromorphic devices, which can act as non-
volatile memory units and neuromorphic computing. The fabricated devices
can behave like biological synapses and exhibit the characteristics of
LTP/LTD, the STDP learning rule, PPF, and ultralow energy consumption.
The demonstrated 100 almost linear and stable conductance states suit well
with the analogue world, with no need of power- and time-inefficient
analogue-to digital converters.
137
6.2 Future direction of biosensing and biomimetic
electronics
For biosensors based on In
2
O
3
nanoribbons, the platform is currently robust
and ready for more practical relevant functionalities. One goal of the
nanoribbon sensor platform is a compact point-of-care (POC) and wearable
device that can be used by patients and emergency personal outside of the
central laboratory facilities. Toward that goal, focus will be on fully
integerated three-dimensional electronics. A second goal of the biosensor is
to be used for intrinsic stretchable skin electroncis.
Firstly, a framework for engineering three-dimensional integrated biosensor
electronics by combining strategies in material design and advanced
microfabrication is needed. Previously, three-dimensional devices-built layer
by layer through transfer printing pre-designed circuits is reported.
1
Here, I
propose an approach by combining our biosensor platform with this high
integration density. Using this engineering framework, we can create a
biosensor platform on human–machine interface testbed that can offer
Bluetooth data communication capabilities and sensing capabilities such as,
glucose, lactose, pH, temperature, etc.
138
Figure 6. 1 (a), Schematics showing the system that is fabricated layer by layer. VIAs are
used for interlayer electrical connections. Key components in each layer are labelled. BLE,
Bluetooth; EP, electrophysiological potential. (b), Optical micrographs of the system when
freestanding (top), twisted at 90° (middle) and poked with a dome height of ~8 mm
(bottom), highlighting its superb mechanical compliance and robustness.
Secondly, stretchable electronics is an emerging technology that creates devices
with the ability to conform to nonplanar and dynamic surfaces such as the human
body. However, metal oxide such as indium oxide cannot exceed strain over 10%,
which may not meet the requirements for stretchable skin electronics.
Investigation of intrinsic stretchable materials for sensor application is needed.
As reported in the literature, the authors show a versatile biofunctionalization
technique for the solution processable conducting polymer poly(3,4-
139
ethylenedioxythiophene) doped with poly(styrenesulfonate) PEDOT:PSS, which
is a commercially available material, and has a record high conductivity. Addition
of poly(vinyl alcohol) (PVA) into the solution with PEDOT:PSS provides a
handle for subsequent silanization with a well-characterised silane reagent,
allowing for covalent linkage of biological moieties onto PEDOT:PSS films.
After apply to stretchable substrate, such PEDOT:PSS/PVA can provide
homogenous and large-scale biofunctionalization with polypeptides and proteins
maintenance of the biological functionalities of the proteins.
Figure 6. 2 Reaction scheme for biofunctionalization of PEDOT:PSS by incorporation of
PVA.
140
6.3 References
1. Huang, Z.; Hao, Y .; Li, Y .; Hu, H.; Wang, C.; Nomoto, A.; Pan, T.; Gu, Y .; Chen, Y .;
Zhang, T.; Li, W.; Lei, Y .; Kim, N.; Wang, C.; Zhang, L.; Ward, J. W.; Maralani, A.;
Li, X.; Durstock, M. F.; Pisano, A.; Lin, Y .; Xu, S. Three-dimensional integrated
stretchable electronics. Nature Electronics 2018, 1, 473-480.
2. Strakosas, X.; Sessolo, M.; Hama, A.; Rivnay, J.; Stavrinidou, E.; Malliaras, G. G.;
Owens, R. M., A facile biofunctionalisation route for solution processable
conducting polymer devices. Journal of Materials Chemistry B 2014, 2, 2537-2545.
141
Bibliography
1. Abbas, A. N.; Liu, G.; Liu, B.; Zhang, L.; Liu, H.; Ohlberg, D.; Wu, W.; Zhou, C.,
Patterning, Characterization, and Chemical Sensing Applications of Graphene
Nanoribbon Arrays Down to 5 nm Using Helium Ion Beam Lithography. ACS Nano
2014, 8, 1538-1546.
2. Abbott, L. F.; Nelson, S. B. Synaptic Plasticity: Taming the Beast. Nat. Neurosci. 2000,
3, 1178– 1183.
3. Abikshyeet, P.; Ramesh, V .; Oza, N., Glucose Estimation in the Salivary Secretion of
Diabetes Mellitus Patients. Diabetes, Metab. Syndr. Obes.: Targets Ther. 20 12, 5, 149.
4. Allen, B. L.; Kichambare, P. D.; Star, A., Carbon Nanotube Field-Effect-Transistor-
Based Biosensors. Adv. Mater. 2007, 19, 1439−1451.
5. Aroonyadet, N.; Wang, X.; Song, Y .; Chen, H.; Cote, R. J.; Thompson, M. E.; Datar,
R. H.; Zhou, C., Highly Scalable, Uniform, and Sensitive Biosensors Based on Top-
Down Indium Oxide Nanoribbons and Electronic Enzyme-Linked Immunosorbent
Assay. Nano lett. 2015, 15, 1943-1951.
6. AT&T Laboratories, Cambridge ORL Database of Faces. 1992- 1994.
7. Bache, K.; Lichman, M. UCI machine learning repository. University of California,
8. Backus, J. Can Programming Be Liberated from the von Neumann Style? A Functional
Style and Its Algebra of Programs. Commun. ACM 1978, 21, 613– 641.
9. Bandodkar, A. J.; Wang, J., Non-Invasive Wearable Electrochemical Sensors: A
Review. Trends Biotechnol. 2014, 32, 363-371.
10. Bassan, R.; Potsch, A.; Maisel, A.; Tura, B.; Villacorta, H.; Nogueira, M. V .; Campos,
A.; Gamarski, R.; Masetto, A. C.; Moutinho, M. A., B-Type Natriuretic Peptide: A
Novel Early Blood Marker of Acute Myocardial Infarction in Patients with Chest Pain
and No St Segment Elevation. Eur. Heart J. 2005, 26, 234-240.
11. Berman, E. R., Biochemistry of the Eye. Plenum Press, New York.
142
12. Bliss, T. V . P.; Collingridge, G. L. A Synaptic Model of Memory: Long-Term
Potentiation in the Hippocampus. Nature 1993, 361, 31– 39.
13. Boyn, S.; Grollier, J.; Lecerf, G.; Xu, B.; Locatelli, N.; Fusil, S.; Girod, S.; Carrétéro,
C.; Garcia, K.; Xavier, S. Learning Through Ferroelectric Domain Dynamics in Solid-
State Synapses. Nat. Commun. 2017, 8, 14736.
14. Braunwald, E.; Antman, E. M.; Beasley, J. W.; Califf, R. M.; Cheitlin, M. D.; Hochman,
J. S.; Jones, R. H.; Kereiakes, D.; Kupersmith, J.; Levin, T. N.; Pepine, C. J.; Schaeffer,
J. W.; Smith, E. E., 3rd; Steward, D. E.; Theroux, P.; Gibbons, R. J.; Alpert, J. S.; Eagle,
K. A.; Faxon, D. P.; Fuster, V .; el al., Acc/Aha Guidelines for the Management of
Patients with Unstable Angina and Non-St-Segment Elevation Myocardial Infarction:
Executive Summary and Recommendations. A Report of the American College of
Cardiology/American Heart Association Task Force on Practice Guidelines
(Committee on the Management of Patients with Unstable Angina). Circulation 2000,
102, 1193-209.
15. Brush Jr, J. E.; Brand, D. A.; Acampora, D.; Chalmer, B.; Wackers, F. J., Use of the
Initial Electrocardiogram to Predict in-Hospital Complications of Acute Myocardial
Infarction. N. Engl. J. Med. 1985, 312, 1137-1141.
16. Butler, J. E., Enzyme-Linked Immunosorbent Assay. J. Immunoassay 200 0, 21, 165-
209.
17. Cai, X.-M.; Hao, Y .-Q.; Zhang, D.-P.; Fan, P., Direct Current Magnetron Sputtering
Deposition of Inn Thin Films. Appl. Surf. Sci. 200 9, 256, 43-45.
18. Cao, X.; Cao, Y .; Zhou, C., Imperceptible and Ultraflexible p‑Type Transistors and
Macroelectronics Based on Carbon Nanotubes. ACS Nano. 2016, 10, 199-206.
19. Cao, X.; Lau, C.; Liu, Y .; Wu, F.; Gui, H.; Liu, Q.; Ma, Y .; Wan, H.; Amer, M. R.; Zhou,
C. Fully Screen-Printed, Large-Area, and Flexible Active-Matrix Electrochromic
Displays Using Carbon Nanotube Thin-Film Transistors. ACS Nano 2016, 10, 9816–
9822.
20. Carey, B. J.; Ou, J. Z.; Clark, R. M.; Berean, K. J.; Zavabeti, A.; Chesman, A. S.; Russo,
143
S. P.; Lau, D. W.; Xu, Z.-Q.; Bao, Q., Wafer-scale two-dimensional semiconductors
from printed oxide skin of liquid metals. Nat. Commun. 2017, 8, 14482.
21. Chang, H.-K.; Ishikawa, F. N.; Zhang, R.; Datar, R.; Cote, R. J.; Thompson, M. E.;
Zhou, C., Rapid, Label-Free, Electrical Whole Blood Bioassay Based on
Nanobiosensor Systems. ACS nano 201 1, 5, 9883-9891.
22. Chanthbouala, A.; Garcia, V .; Cherifi, R. O.; Bouzehouane, K.; Fusil, S.; Moya, X.;
Xavier, S.; Yamada, H.; Deranlot, C.; Mathur, N. D. A Ferroelectric Memristor. Nat.
Mater. 2012, 11, 860– 864.
23. Chen, P. C.; Sukcharoenchoke, S.; Ryu, K.; Gomez de Arco, L.; Badmaev, A.; Wang,
C.; Zhou, C., 2, 4, 6‐Trinitrotoluene (Tnt) Chemical Sensing Based on Aligned Single‐
Walled Carbon Nanotubes and Zno Nanowires. Adv. Mater. 2010, 22, 1900-1904.
24. Chen, R. J.; Bangsaruntip, S.; Drouvalakis, K. A.; Kam, N. W. S.; Shim, M.; Li, Y .;
Kim, W.; Utz, P. J.; Dai, H., Noncovalent Functionalization of Carbon Nanotubes for
Highly Specific Electronic Biosensors. Proc. Natl. Acad. Sci. 2003, 100, 4984-4989.
25. Christenson, R. H.; Azzazy H. M. E., Biomarkers of Myocardial Necrosis. In
Cardiovascular Biomarkers: Pathophysiology and Disease Management, Morrow, D.
A., Ed. Humana Press Inc.: Totowa, New Jersy, 2006, pp 3-25.
26. Claussen, J. C.; Kumar, A.; Jaroch, D. B.; Khawaja, M. H.; Hibbard, A. B.; Porterfield,
D. M.; Fisher, T. S., Nanostructuring Platinum Nanoparticles on Multilayered
Graphene Petal Nanosheets for Electrochemical Biosensing. Adv. Funct. Mater. 2012,
22, 3399-3405.
27. Cui, Y .; Wei, Q.; Park, H.; Lieber, C. M., Nanowire Nanosensors for Highly Sensitive
and Selective Detection of Biological and Chemical Species. Science 2001, 293, 1289-
1292.
28. Deng, L. The MNIST Database of Handwritten Digit Images for Machine Learning
Research. IEEE Signal Processing Magazine 201 2, 29, 141– 142.
29. Duan, X.; Li, Y .; Rajan, N. K.; Routenberg, D. A.; Modis, Y .; Reed, M. A.,
Quantification of the Affinities and Kinetics of Protein Interactions Using Silicon
144
Nanowire Biosensors. Nat. Nanotechnol. 2012, 7, 401-407.
30. Eryilmaz, S. B.; Kuzum, D.; Jeyasingh, R.; Kim, S.; BrightSky, M.; Lam, C.; Wong,
H.-S. P. Brain-Like Associative Learning Using a Nanoscale Non-V olatile Phase
Change Synaptic Device Array. Front. Neurosci. 2 014, 8, 205.
31. Fan, Z.; Razavi, H.; Do, J.-w.; Moriwaki, A.; Ergen, O.; Chueh, Y.-L.; Leu, P. W.;
Ho, J. C.; Takahashi, T.; Reichertz, L. A., Three-Dimensional Nanopillar-Array
Photovoltaics on Low-Cost and Flexible Substrates. Nat. Mater. 2009, 8, 648−653.
32. Fu, Y .; Kong, L.-a.; Chen, Y .; Wang, J.; Qian, C.; Yuan, Y .; Sun, J.; Gao, Y .; Wan, Q.
Flexible Neuromorphic Architectures Based on Self-Supported Multiterminal Organic
Transistors. ACS Appl. Mater. Interfaces 2018, 10, 26443– 26450.
33. Fukuda, K.; Takeda, Y .; Yoshimura, Y .; Shiwaku, R.; Tran, L. T.; Sekine, T.; Mizukami,
M.; Kumaki, D.; Tokito, S. Fully-Printed High-Performance Organic Thin-Film
Transistors and Circuitry on One-Micron-Thick Polymer Films. Nat. Commun. 2014,
5, 4147.
34. Gao, W.; Emaminejad, S.; Nyein, H. Y . Y .; Challa, S.; Chen, K.; Peck, A.; Fahad, H.
M.; Ota, H.; Shiraki, H.; Kiriya, D.; Lien, D; Brooks, G. A.; Davis, R. W.; Javey, A.,
Fully Integrated Wearable Sensor Arrays for Multiplexed in Situ Perspiration Analysis.
Nature 2016, 529, 509-514.
35. Gibler, W. B.; Young, G. P.; Hedges, J. R.; Lewis, L. M.; Smith, M. S.; Carleton, S. C.;
Aghababian, R. V .; Jorden, R. O.; Allison, E. J.; Often, E. J., Acute Myocardial
Infarction in Chest Pain Patients with Nondiagnostic Ecgs: Serial CK-MB Sampling
in the Emergency Department. Annals of Emergency Medicine 1992, 21, 504-512.
36. Gkoupidenis, P.; Koutsouras, D. A.; Malliaras, G. G. Neuromorphic Device
Architectures with Global Connectivity Through Electrolyte Gating. Nat. Commun.
2017, 8, 15448.
37. Gkoupidenis, P.; Schaefer, N.; Garlan, B.; Malliaras, G. G. Neuromorphic Functions
in PEDOT:PSS Organic Electrochemical Transistors. Adv. Mater. 2015, 27, 7176–
7180.
145
38. Goldsmith, B. R.; Mitala, J. J.; Josue, J.; Castro, A.; Lerner, M. B.; Bayburt, T. H.;
Khamis, S. M.; Jones, R. A.; Brand, J. G.; Sligar, S. G.; Luetje, C. W.; Gelperin, A.;
Rhodes, P. A.; Discher, B. M.; Johnson, A. T. C., Biomimetic Chemical Sensors Using
Nanoelectronic Readout of Olfactory Receptor Proteins. ACS nano 201 1, 5, 5408-5416.
39. Hamaguchi, N.; Ellington, A.; Stanton, M., Aptamer Beacons for the Direct
Detection of Proteins. Anal. Biochem. 2001, 294, 126−131.
40. Hammock, M. L.; Knopfmacher, O.; Naab, B. D.; Tok, J. B.-H.; Bao, Z.,
Investigation of Protein Detection Parameters Using Nanofunctionalized Organic
Field-Effect Transistors. ACS Nano 2013, 7, 3970−3980.
41. Han, J. H.; Gibler, W. B., Biomarkers in the Emergency Department: Rapid Diagnosis
and Triage. In Cardiovascular Biomarkers: Pathophysiology and Disease
Management, Morrow, D. A., Ed. Humana Press: 2006, pp 61-77.
42. Hanssen, B. L.; Siraj, S.; Wong, D. K., Recent Strategies to Minimise Fouling in
Electrochemical Detection Systems. Rev. Anal. Chem. 2016, 35, 1−28.
43. Heller, A., Implanted Electrochemical Glucose Sensors for the Management of
Diabetes. Annu. Rev. Biomed. Eng. 1999, 1, 153-175.
44. Hermanson, G. T., Bioconjugate Techniques. Academic Press: 2013.
45. Hrapovic, S.; Liu, Y .; Male, K. B.; Luong, J. H., Electrochemical Biosensing Platforms
Using Platinum Nanoparticles and Carbon Nanotubes. Anal. Chem. 2004, 76, 1083-
1088.
46. Irvine, School of Information and Computer Sciences, 2013; URL: http://archive. ics.
uci.
47. Ishikawa, F. N.; Curreli, M.; Olson, C. A.; Liao, H.-I.; Sun, R.; Roberts, R. W.; Cote,
R. J.; Thompson, M. E.; Zhou, C., Importance of Controlling Nanotube Density for
Highly Sensitive and Reliable Biosensors Functional in Physiological Conditions.
ACS Nano 2010, 4, 6914−6922.
48. Ispas, C. R.; Crivat, G.; Andreescu, S., Recent Developments in Enzyme-Based
Biosensors for Biomedical Analysis. Anal. Lett. 2012, 45, 168−186.
146
49. Jang, S.; Jang, S.; Lee, E.-H.; Kang, M.; Wang, G.; Kim, T.-W. Ultrathin Conformable
Organic Artificial Synapse for Wearable Intelligent Device Applications. ACS Appl.
Mater. Interfaces 2019, 11, 1071– 1080.
50. Kaltenbrunner, M.; Sekitani, T.; Reeder, J.; Yokota, T.; Kuribara, K.; Tokuhara, T.;
Drack, M.; Schwödiauer, R.; Graz, I.; Bauer-Gogonea, S.; Bauer, S.; Someya, T., An
Ultra-Lightweight Design for Imperceptible Plastic Electronics. Nature 2013, 499,
458-463.
51. Kennedy, R. T.; Watson, C. J.; Haskins, W. E.; Powell, D. H.; Strecker, R. E., In
vivo Neurochemical Monitoring by Microdialysis and Capillary Separations. Curr.
Opin. Chem. Biol. 2002, 6, 659−665.
52. Kim, D. H.; Lu, N.; Ma, R.; Kim, Y . S.; Kim, R. H.; Wang, S.; Wu, J.; Won, S. M.; Tao,
H.; Islam, A.; Yu, K. J.; Kim, T. I.; Chowdhury, R.; Ying, M.; Xu, L.; Li, M.; Chung
H. J.; Keum, H.; McCormick, M.; Liu, P., et al. Epidermal Electronics. Science 201 1,
333, 838-843.
53. Kim, D.-H.; Viventi, J.; Amsden, J. J.; Xiao, J.; Vigeland, L.; Kim, Y .-S.; Blanco, J. A.;
Panilaitis, B.; Frechette, E. S.; Contreras, D., Dissolvable Films of Silk Fibroin for
Ultrathin Conformal Bio-Integrated Electronics. Nat. Mater. 2010, 9, 511−517.
54. Kim, J.; Rim, Y . S.; Chen, H.; Cao, H. H.; Nakatsuka, N.; Hinton, H. L.; Zhao, C.;
Andrews, A. M.; Yang, Y .; Weiss, P. S., Fabrication of High-Performance Ultrathin
In2O3 Film Field-Effect Transistors and Biosensors Using Chemical Lift-Off
Lithography. ACS nano 2 015, 9, 4572-4582.
55. Kim, S.; Choi, S.; Lu, W. D. Comprehensive Physical Model of Dynamic Resistive
Switching in an Oxide Memristor. ACS Nano 2014, 8, 2369– 2376.
56. Kim, Y . D.; Hone, J., Materials Science: Screen Printing of 2D Semiconductors. Nature
2017, 544, 167-168.
57. Kong, T.; Su, R.; Zhang, B.; Zhang, Q.; Cheng, G., Cmos-Compatible, Label-Free
Silicon-Nanowire Biosensors to Detect Cardiac Troponin I for Acute Myocardial
Infarction Diagnosis. Biosens. Bioelectron. 2012, 34, 267-272.
147
58. Kuzum, D.; Yu, S.; Wong, H. S. Synaptic Electronics: Materials, Devices and
Applications. Nanotechnology 2013, 24, 382001.
59. Lee, H.; Choi, T. K.; Lee, Y . B.; Cho, H. R.; Ghaffari, R.; Wang, L.; Choi, H. J.; Chung,
T. D.; Lu, N.; Hyeon, T., A Graphene-Based Electrochemical Device with
Thermoresponsive Microneedles for Diabetes Monitoring and Therapy. Nature
Nanotechnol. 2016, 11, 566-572.
60. Lee, H.; Song, C.; Hong, Y . S.; Kim, M. S.; Cho, H. R.; Kang, T.; Shin, K.; Choi, S.
H.; Hyeon, T.; Kim, D. H., Wearable/Disposable Sweat-Based Glucose Monitoring
Device with Multistage Transdermal Drug Delivery Module. Sci. Adv. 2017, 3.
61. Lee, T. H.; Rouan, G. W.; Weisberg, M. C.; Brand, D. A.; Cook, E. F.; Acampora, D.;
Goldman, L., Sensitivity of Routine Clinical Criteria for Diagnosing Myocardial
Infarction within 24 Hours of Hospitalization. Ann. Intern. Med. 1987, 106, 181-6.
62. Lee-Lewandrowski, E.; Januzzi, J. L., Jr.; Grisson, R.; Mohammed, A. A.;
Lewandrowski, G.; Lewandrowski, K., Evaluation of First-Draw Whole Blood, Point-
of-Care Cardiac Markers in the Context of the Universal Definition of Myocardial
Infarction: A Comparison of a Multimarker Panel to Troponin Alone and to Testing in
the Central Laboratory. Arch. Pathol. Lab. Med. 2 01 1, 135, 459-63.
63. Lei, B.; Li, C.; Zhang, D.; Zhou, Q.; Shung, K.; Zhou, C., Nanowire Transistors with
Ferroelectric Gate Dielectrics: Enhanced Performance and Memory Effects. Appl.
Phys. Lett. 2004, 84, 4553-4555.
64. Lewandrowski, K.; Chen, A.; Januzzi, J., Cardiac Markers for Myocardial Infarction.
A Brief Review. Am. J. Clin. Pathol. 2002, 118 Suppl, S93-9.
65. Lewis, J.; Grego, S.; Chalamala, B.; Vick, E.; Temple, D., Highly Flexible
Transparent Electrodes for Organic Light-Emitting Diode-Based Displays. Appl.
Phys. Lett. 2004, 85, 3450−3452.
66. Li, C.; Curreli, M.; Lin, H.; Lei, B.; Ishikawa, F. N.; Datar, R.; Cote, R. J.; Thompson,
M. E.; Zhou, C., Complementary Detection of Prostate-Specific Antigen Using In2O3
Nanowires and Carbon Nanotubes. J. Am. Chem. Soc. 2005, 127, 12484-5.
148
67. Li, C.; Hu, M.; Li, Y .; Jiang, H.; Ge, N.; Montgomery, E.; Zhang, J.; Song, W.; Dávila,
N.; Graves, C. E.; Li, Z.; Strachan, J. P.; Lin, P.; Wang, Z.; Barnell, M.; Wu, Q.;
Williams, R. S.; Yang, J. J.; Xia, Q. Analogue signal and image processing with large
memristor crossbars. Nat. Electron. 2018, 1, 52– 59.
68. Li, C.; Lei, B.; Zhang, D.; Liu, X.; Han, S.; Tang, T.; Rouhanizadeh, M.; Hsiai, T.;
Zhou, C., Chemical Gating of In2O3 Nanowires by Organic and Biomolecules. Appl.
Phys. Lett. 2003, 83, 4014-4016.
69. Liao, C.; Mak, C.; Zhang, M.; Chan, H. L.; Yan, F., Flexible Organic Electrochemical
Transistors for Highly Selective Enzyme Biosensors and Used for Saliva Testing. Adv.
Mater. 2015, 27, 676-681.
70. Lin, P.; Yan, F.; Chan, H. L., Ion-Sensitive Properties of Organic Electrochemical
Transistors. ACS Appl. Mater. Interfaces 2010, 2, 1637-1641.
71. Lin, Y.; Yu, P.; Mao, L., A Multi-Enzyme Microreactor-Based Online
Electrochemical System for Selective and Continuous Monitoring of Acetylcholine.
Analyst 2015, 140, 3781−3787.
72. Lipomi, D. J.; V osgueritchian, M.; Tee, B. C.; Hellstrom, S. L.; Lee, J. A.; Fox, C. H.;
Bao, Z., Skin-Like Pressure and Strain Sensors Based on Transparent Elastic Films of
Carbon Nanotubes. Nat. Nanotechnol. 201 1, 6, 788-792.
73. Liu, Q.; Aroonyadet, N.; Song, Y .; Wang, X.; Cao, X.; Liu, Y .; Cong, S.; Wu, F.;
Thompson, M. E.; Zhou, C., Highly Sensitive and Quick Detection of Acute
Myocardial Infarction Biomarkers Using In2O3 Nanoribbon Biosensors Fabricated
Using Shadow Masks. ACS Nano 2016, 10, 10117-10125.
74. Liu, Q.; Liu, Y.; Wu, F.; Cao, X.; Li, Z.; Alharbi, M.; Abbas, A. N.; Amer, M. R.;
Zhou, C., Highly Sensitive and Wearable In2O3 Nanoribbon Transistor Biosensors
with Integrated On-Chip Gate for Glucose Monitoring in Body Fluids. ACS Nano
2018, 12, 1170−1178.
75. Liu, Y . H.; Zhu, L. Q.; Feng, P.; Shi, Y .; Wan, Q. Freestanding Artificial Synapses
Based on Laterally Proton‐Coupled Transistors on Chitosan Membranes. Adv. Mater.
149
2015, 27, 5599– 5604.
76. Locatelli, N.; Cros, V .; Grollier, J. Spin-Torque Building Blocks. Nat. Mater. 2014, 13,
11– 20.
77. Luan, L.; Wei, X.; Zhao, Z.; Siegel, J. J.; Potnis, O.; Tuppen, C. A.; Lin, S.; Kazmi,
S.; Fowler, R. A.; Holloway, S., Ultraflexible Nanoelectronic Probes Form Reliable,
Glial Scar–Free Neural Integration. Sci. Adv. 2017, 3, e1601966.
78. Mahajan, V . S.; Jarolim, P., How to Interpret Elevated Cardiac Troponin Levels.
Circulation 201 1, 124, 2350-2354.
79. Makaram, P.; Owens, D.; Aceros, J., Trends in Nanomaterial-Based Non-Invasive
Diabetes Sensing Technologies. Diagnostics 2014, 4, 27-46.
80. Mannoor, M. S.; Tao, H.; Clayton, J. D.; Sengupta, A.; Kaplan, D. L.; Naik, R. R.;
Verma, N.; Omenetto, F. G.; McAlpine, M. C., Graphene-Based Wireless Bacteria
Detection on Tooth Enamel. Nat. Commun. 2012, 3, 763.
81. McAlpine, M. C.; Ahmad, H.; Wang, D.; Heath, J. R., Highly Ordered Nanowire
Arrays on Plastic Substrates for Ultrasensitive Flexible Chemical Sensors. Nat. Mater.
2007, 6, 379-384.
82. Merolla, P. A.; Arthur, J. V.; Alvarez-Icaza, R.; Cassidy, A. S.; Sawada, J.; Akopyan,
F.; Jackson, B. L.; Imam, N.; Guo, C.; Nakamura, Y.; Brezzo, B.; Vo, I.; Esser, S.
K.; Appuswamy, R.; Taba, B.; Amir, A.; Flickner, M. D.; Risk, W. P.; Manohar, R.;
Modha, D. S. A Million Spiking-Neuron Integrated Circuit with a Scalable
Communication Network and Interface. Science 2014, 345, 668– 673.
83. Minev, I. R.; Musienko, P.; Hirsch, A.; Barraud, Q.; Wenger, N.; Moraud, E. M.;
Gandar, J.; Capogrosso, M.; Milekovic, T.; Asboth, L., Electronic Dura Mater for
Long-Term Multimodal Neural Interfaces. Science 2015, 347, 159−163.
84. Morrison, L. K.; Harrison, A.; Krishnaswamy, P.; Kazanegra, R.; Clopton, P.; Maisel,
A., Utility of a Rapid B-Natriuretic Peptide Assay in Differentiating Congestive Heart
Failure from Lung Disease in Patients Presenting with Dyspnea. J. Am. Coll. Cardiol.
2002, 39, 202-209.
150
85. Moyer, J.; Wilson, D.; Finkelshtein, I.; Wong, B.; Potts, R., Correlation between Sweat
Glucose and Blood Glucose in Subjects with Diabetes. Diabetes Technol. Ther. 2012,
14, 398-402.
86. Mu, L.; Droujinine, I. A.; Rajan, N. K.; Sawtelle, S. D.; Reed, M. A., Direct, Rapid,
and Label-Free Detection of Enzyme–Substrate Interactions in Physiological Buffers
Using Cmos-Compatible Nanoribbon Sensors. Nano lett. 2014, 14, 5315-5322.
87. Muller, P.; Seeman, P., Brain Neurotransmitter Receptors After Long-Term
Haloperidol: Dopamine, Acetylcholine, Serotonin, α-Noradrenergic and Naloxone
Receptors. Life Sci. 1977, 21, 1751−1758.
88. Nakatsuka, N.; Cao, H. H.; Deshayes, S.; Melkonian, A. L.; Kasko, A. M.; Weiss, P.
S.; Andrews, A. M., Aptamer Recognition of Multiplexed Small-Molecule-
Functionalized Substrates. ACS Appl. Mater. Interfaces 2018, 10, 23490−23500.
89. Nakatsuka, N.; Yang, K.-A.; Abendroth, J. M.; Cheung, K. M.; Xu, X.; Yang, H.;
Zhao, C.; Zhu, B.; Rim, Y. S.; Yang, Y., Aptamer–Field-Effect Transistors
Overcome Debye Length Limitations for Small-Molecule Sensing. Science 2018,
362, 319−324.
90. NanGate, S. 45nm Open Cell Library, 2008; URL: http://www.nangate.com/.
91. Naylor, C. H.; Kybert, N. J.; Schneier, C.; Xi, J.; Romero, G.; Saven, J. G.; Liu, R.;
Johnson, A. T. C., Scalable Production of Molybdenum Disulfide-Based Biosensors.
ACS nano 2016, 10, 6173-6179.
92. Newby, L. K.; Christenson, R. H.; Ohman, E. M.; Armstrong, P. W.; Thompson, T. D.;
Lee, K. L.; Hamm, C. W.; Katus, H. A.; Cianciolo, C.; Granger, C. B.; Topol, E. J.;
Califf, R. M.; Investigators, G.-I., Value of Serial Troponin T Measures for Early and
Late Risk Stratification in Patients with Acute Coronary Syndromes. Circulation 1998,
98, 1853-1859.
93. Ng, S. M.; Krishnaswamy, P.; Morissey, R.; Clopton, P.; Fitzgerald, R.; Maisel, A. S.,
Ninety-Minute Accelerated Critical Pathway for Chest Pain Evaluation. Am. J. Cardiol.
2001, 88, 611-617.
151
94. Nourjah, P., National Hospital Ambulatory Medical Care Survey: 1997 Emergency
Department Summary. Advance Data from Vital and Health Statistics 1999, 304, 1-24.
95. Ohman, E. M.; Armstrong, P. W.; Christenson, R. H.; Granger, C. B.; Katus, H. A.;
Hamm, C. W.; OHanesian, M. A.; Wagner, G. S.; Kleiman, N. S.; Harrell, F. E.; Califf,
R. M.; Topol, E. J., Cardiac Troponin T Levels for Risk Stratification in Acute
Myocardial Ischemia. N. Engl. J. Med. 1996, 335, 1333-1341.
96. Olarte, O.; Chilo, J.; Pelegri-Sebastia, J.; Barbe, K.; Van Moer, W. Glucose Detection
in Human Sweat Using An Electronic Nose, Conf. Proc. IEEE Eng. Med. Biol. Soc.
2013; 1462-1465.
97. Park, J.-S.; Kim, T.-W.; Stryakhilev, D.; Lee, J.-S.; An, S.-G.; Pyo, Y.-S.; Lee, D.-
B.; Mo, Y. G.; Jin, D.-U.; Chung, H. K., Flexible Full Color Organic Light-Emitting
Diode Display on Polyimide Plastic Substrate Driven by Amorphous Indium Gallium
Zinc Oxide Thin-Film Transistors. Appl. Phys. Lett. 2009, 95, 013503.
98. Park, S.; Lee, S.; Kim, C.-H.; Lee, I.; Lee, W.-J.; Kim, S.; Lee, B.-G.; Jang, J.-H.; Yoon,
M.-H., Sub-0.5 V Highly Stable Aqueous Salt Gated Metal Oxide Electronics. Sci. Rep.
2015, 5, 13088.
99. Perry, M.; Li, Q.; Kennedy, R. T., Review of Recent Advances in Analytical
Techniques for the Determination of Neurotransmitters. Anal. Chim. Acta 2009, 653,
1−22.
100.Pershin, Y . V .; Di Ventra, M. Memory Effects in Complex Materials and Nanoscale
Systems. Adv. Phys. 201 1, 60, 145– 227.
101.Rim, Y . S.; Bae, S. H.; Chen, H.; Yang, J. L.; Kim, J.; Andrews, A. M.; Weiss, P. S.;
Yang, Y .; Tseng, H.-R., Printable Ultrathin Metal Oxide Semiconductor-Based
Conformal Biosensors. ACS Nano 2015, 9, 12174-12181.
102.Robinson, D. L.; Venton, B. J.; Heien, M. L.; Wightman, R. M., Detecting
Subsecond Dopamine Release with Fast-Scan Cyclic Voltammetry in vivo. Clin.
Chem. 2003, 49, 1763−1773.
152
103.Rodeberg, N. T.; Sandberg, S. G.; Johnson, J. A.; Phillips, P. E.; Wightman, R. M.,
Hitchhiker’s Guide to Voltammetry: Acute and Chronic Electrodes for in vivo Fast-
Scan Cyclic Voltammetry. ACS Chem. Neurosci. 2017, 8, 221−234.
104.Ronkainen, N. J.; Halsall, H. B.; Heineman, W. R., Electrochemical Biosensors. Chem.
Soc. Rev. 2010, 39, 1747-1763.
105.Rubiolo, M.; Stegmayer, G.; Milone, D. Compressing Arrays of Classifiers Using
V olterra-Neural Network: Application to Face Recognition. Neural Computing and
Applications 2013, 23, 1687– 1701.
106.Rude, R. E.; Poole, W. K.; Muller, J. E.; Turi, Z.; Rutherford, J.; Parker, C.; Roberts,
R.; Raabe, D. S.; Gold, H. K.; Stone, P. H.; Willerson, J. T.; Braunwald, E.,
Electrocardiographic and Clinical-Criteria for Recognition of Acute Myocardial-
Infarction Based on Analysis of 3,697 Patients. Am. J. Cardiol. 1983, 52, 936-941.
107.Saenger, A. K.; Jaffe, A. S., Requiem for a Heavyweight the Demise of Creatine
Kinase-MB. Circulation 2008, 118, 2200-2206.
108.Sangwan, V . K.; Lee, H.-S.; Bergeron, H.; Balla, I.; Beck, M. E.; Chen, K.-S.; Hersam,
M. C. Multi-Terminal Memtransistors from Polycrystalline Monolayer Molybdenum
Disulfide. Nature 2018, 554, 500– 504.
109.Sarkar, D.; Liu, W.; Xie, X.; Anselmo, A. C.; Mitragotri, S.; Banerjee, K., Mos2 Field-
Effect Transistor for Next-Generation Label-Free Biosensors. ACS nano 2014, 8,
3992-4003.
110.Shao, Q.; Yu, G.; Lan, Y .-W.; Shi, Y .; Li, M.-Y .; Zheng, C.; Zhu, X.; Li, L.-J.; Amiri,
P. K.; Wang, K. L. Strong Rashba-Edelstein Effect-Induced Spin–Orbit Torques in
Monolayer Transition Metal Dichalcogenide/Ferromagnet Bilayers. Nano Lett. 2016,
16, 7514– 7520.
111.Shao, Y.; Wang, J.; Wu, H.; Liu, J.; Aksay, I. A.; Lin, Y., Graphene Based
Electrochemical Sensors and Biosensors: a Review. Electroanalysis 2010, 22,
1027−1036.
112.Sheridan, P. M.; Cai, F.; Du, C.; Ma, W.; Zhang, Z.; Lu, W. D. Sparse Coding with
153
Memristor Networks. Nat. Nanotechnol. 2017, 12, 784– 789.
113.Shulaker, M. M.; Hills, G.; Park, R. S.; Howe, R. T.; Saraswat, K.; Wong, H.-S. P.;
Mitra, S. Three-Dimensional Integration of Nanotechnologies for Computing and Data
Storage on a Single Chip. Nature 2017, 547, 74– 78.
114.Si, B.; Song, E., Recent Advances in the Detection of Neurotransmitters.
Chemosensors 2018, 6, 1.
115.Silver, D.; Huang, A.; Maddison, C. J.; Guez, A.; Sifre, L.; van den Driessche, G.;
Schrittwieser, J.; Antonoglou, I.; Panneershelvam, V .; Lanctot, M.; Dieleman, S.;
Grewe, D.; Nham, J.; Kalchbrenner, N.; Sutskever, I.; Lillicrap, T.; Leach, M.;
Kavukcuoglu, K.; Graepel, T.; Hassabis, D. Mastering the Game of Go with Deep
Neural Networks and Tree Search. Nature 2016, 529, 484– 489.
116.So, H. M.; Park, D. W.; Jeon, E. K.; Kim, Y. H.; Kim, B. S.; Lee, C. K.; Choi, S. Y.;
Kim, S. C.; Chang, H.; Lee, J. O., Detection and Titer Estimation of Escherichia Coli
Using Aptamer-Functionalized Single−Walled Carbon−Nanotube Field-effect
Transistors. Small 2008, 4, 197−201.
117.Someya, T.; Kato, Y .; Sekitani, T.; Iba, S.; Noguchi, Y .; Murase, Y .; Kawaguchi, H.;
Sakurai, T., Conformable, Flexible, Large-Area Networks of Pressure and Thermal
Sensors with Organic Transistor Active Matrixes. Proc. Natl. Acad. Sci. U. S. A. 2005,
102, 12321-12325.
118.Star, A.; Gabriel, J.-C. P.; Bradley, K.; Grüner, G., Electronic Detection of Specific
Protein Binding Using Nanotube Fet Devices. Nano lett. 2003, 3, 459-463.
119.Stern, E.; Vacic, A.; Li, C.; Ishikawa, F. N.; Zhou, C.; Reed, M. A.; Fahmy, T. M., A
Nanoelectronic Enzyme-Linked Immunosorbent Assay for Detection of Proteins in
Physiological Solutions. Small 2010, 6, 232-8.
120.Stern, E.; Vacic, A.; Rajan, N. K.; Criscione, J. M.; Park, J.; Ilic, B. R.; Mooney, D. J.;
Reed, M. A.; Fahmy, T. M., Label-Free Biomarker Detection from Whole Blood. Nat.
Nanotechnol. 2010, 5, 138-42.
121.Strukov, D. B.; Snider, G. S.; Stewart, D. R.; Williams, R. S. The Missing Memristor
154
Found. Nature 2008, 453, 80– 83.
122.Takei, K.; Takahashi, T.; Ho, J. C.; Ko, H.; Gillies, A. G.; Leu, P. W.; Fearing, R. S.;
Javey, A., Nanowire Active-Matrix Circuitry for Low-V oltage Macroscale Artificial
Skin. Nat. Mater. 2010, 9, 821-826.
123.Tang, H.; Lin, P.; Chan, H. L.; Yan, F., Highly Sensitive Dopamine Biosensors
Based on Organic Electrochemical Transistors. Biosens. Bioelectron. 2011, 26,
4559−4563.
124.Tang, H.; Yan, F.; Lin, P.; Xu, J.; Chan, H. L., Highly Sensitive Glucose Biosensors
Based on Organic Electrochemical Transistors Using Platinum Gate Electrodes
Modified with Enzyme and Nanomaterials. Adv. Funct. Mater. 201 1, 21, 2264-2272.
125.Tang, T.; Han, S.; Jin, W.; Liu, X.; Li, C.; Zhang, D.; Zhou, C.; Chen, B.; Han, J.;
Meyyapan, M., Synthesis and Characterization of Single-Crystal Indium Nitride
Nanowires. J. Mater. Res. 2004, 19, 423-426.
126.Tierney, M.; Kim, H.; Burns, M.; Tamada, J.; Potts, R., Electroanalysis of Glucose in
Transcutaneously Extracted Samples. Electroanalysis 2000, 12, 666-671.
127.Tuma, T.; Pantazi, A.; Le Gallo, M.; Sebastian, A.; Eleftheriou, E., Stochastic Phase-
Change Neurons. Nat. Nanotechnol. 2016, 11, 693– 699.
128.Vacic, A.; Criscione, J. M.; Rajan, N. K.; Stern, E.; Fahmy, T. M.; Reed, M. A.,
Determination of Molecular Configuration by Debye Length Modulation. J. Am.
Chem. Soc. 2011, 133, 13886−13889.
129.Valdes, T. I.; Moussy, F., In Vitro and in Vivo Degradation of Glucose Oxidase Enzyme
Used for an Implantable Glucose Biosensor. Diabetes Technol. Ther. 2000, 2, 367-376.
130.van de Burgt, Y .; Lubberman, E.; Fuller, E. J.; Keene, S. T.; Faria, G. C.; Agarwal, S.;
Marinella, M. J.; Talin, A. A.; Salleo, A. A Non-V olatile Organic Electrochemical
Device as a Low-V oltage Artificial Synapse for Neuromorphic Computing. Nat. Mater.
2017, 16, 414– 418.
131.van de Burgt, Y .; Melianas, A.; Keene, S. T.; Malliaras, G.; Salleo, A. Organic
155
Electronics for Neuromorphic Computing. Nat. Electron. 2018, 1, 386– 397.
132.Van Schoors, J.; Viaene, J.; Van Wanseele, Y.; Smolders, I.; Dejaegher, B.; Vander
Heyden, Y.; Van Eeckhaut, A., An Improved Microbore UHPLC Method with
Electrochemical Detection for the Simultaneous Determination of Low Monoamine
Levels in in vivo Brain Microdialysis Samples. J. Pharm. Biomed. Anal. 2016, 127,
136−146.
133.Vashist, S. K., Non-Invasive Glucose Monitoring Technology in Diabetes
Management: A Review. Anal. Chim. Acta 2012, 750, 16-27.
134.Veiseh, O.; Tang, B. C.; Whitehead, K. A.; Anderson, D. G.; Langer, R., Managing
Diabetes with Nanomedicine: Challenges and Opportunities. Nat. Rev. Drug Discovery
2015, 14, 45-57.
135.Venton, B. J.; Wightman, R. M., Psychoanalytical Electrochemistry: Dopamine and
Behavior. Anal. Chem. 2003, 75, 414−421.
136.Wang, J.; Musameh, M.; Lin, Y ., Solubilization of Carbon Nanotubes by Nafion
toward the Preparation of Amperometric Biosensors. J. Am. Chem. Soc. 2003, 125,
2408-2409.
137.Wang, L.; Xiong, Q.; Xiao, F.; Duan, H., 2D Nanomaterials Based Electrochemical
Biosensors for Cancer Diagnosis. Biosens. Bioelectron. 2017, 89, 136−151.
138.Wang, S.; Xu, J.; Wang, W.; Wang, G.-J. N.; Rastak, R.; Molina-Lopez, F.; Chung,
J. W.; Niu, S.; Feig, V. R.; Lopez, J., Skin Electronics from Scalable Fabrication of
an Intrinsically Stretchable Transistor Array. Nature 2018, 555, 83−88.
139.Wang, Z.; Joshi, S.; Savel’ev, S.; Song, W.; Midya, R.; Li, Y .; Rao, M.; Yan, P.; Asapu,
S.; Zhuo, Y .; Jiang, H.; Lin, P.; Li, C.; Yoon, J. H.; Upadhyay, N. K.; Zhang, J.; Hu,
M.; Strachan, J. P.; Barnell, M.; Wu, Q.; Wu, H.; Williams, R. S.; Xia, Q.; Yang, J. J.
Fully Memristive Neural Networks for Pattern Classification with Unsupervised
Learning. Nat. Electron. 2018, 1, 137– 145.
140.Weiss, P.; Trevor, P.; Cardillo, M., Gas–Surface Interactions on InP Monitored by
Changes in Substrate Electronic Properties. J. Chem. Phys. 1989, 90, 5146−5153.
156
141.Willner, I.; Zayats, M., Electronic Aptamer-Based Sensors. Angew. Chem., Int. Ed.
2007, 46, 6408−6418.
142.Wong, H.-S. P.; Lee, H.-Y .; Yu, S.; Chen, Y .-S.; Wu, Y .; Chen, P.-S.; Lee, B.; Chen, F.
T.; Tsai, M.-J. Metal–Oxide RRAM. Proc. IEEE 2 012, 100, 1951– 1970.
143.Wong, H.-S. P.; Raoux, S.; Kim, S.; Liang, J.; Reifenberg, J. P.; Rajendran, B.; Asheghi,
M.; Goodson, K. E. Phase Change Memory. Proc. IEEE 2010, 98, 2201– 2227.
144.Wu, A. H. B.; Apple, F. S.; Gibler, W. B.; Jesse, R. L.; Warshaw, M. M.; Valdes, R.,
National Academy of Clinical Biochemistry Standards of Laboratory Practice:
Recommendations for the Use of Cardiac Markers in Coronary Artery Diseases. Clin.
Chem. 1999, 45, 1104-1121.
145.Wu, C.; Kim, T. W.; Choi, H. Y .; Strukov, D. B.; Yang, J. J. Flexible Three-Dimensional
Artificial Synapse Networks with Correlated Learning and Trainable Memory
Capability. Nat. Commun. 2017, 8, 752.
146.Wu, L.; Chu, H.; Koh, W.; Li, E., Highly Sensitive Graphene Biosensors Based on
Surface Plasmon Resonance. Opt. Express 2010, 18, 14395−14400.
147.Xie, C.; Liu, J.; Fu, T.-M.; Dai, X.; Zhou, W.; Lieber, C. M., Three-Dimensional
Macroporous Nanoelectronic Networks as Minimally Invasive Brain Probes. Nat.
Mater. 2015, 14, 1286−1292.
148.Xu, S.; Zhang, Y .; Jia, L.; Mathewson, K. E.; Jang, K.-I.; Kim, J.; Fu, H.; Huang, X.;
Chava, P.; Wang, R.; Bhole, S.; Wang, L.; Na, Y . J.; Guan, Y .; Flavin, M.; Han, Z.;
Huang, Y .; Rogers, J. A.,Soft Microfluidic Assemblies of Sensors, Circuits, and Radios
for the Skin. Science 201 4, 344, 70-74.
149.Xuan, Y .; Sandberg, M.; Berggren, M.; Crispin, X. An All-Polymer-Air PEDOT
Battery. Org. Electron. 2 012, 13, 632– 637.
150.Yan, Q.; Peng, B.; Su, G.; Cohan, B. E.; Major, T. C.; Meyerhoff, M. E., Measurement
of Tear Glucose Levels with Amperometric Glucose Biosensor/Capillary Tube
Configuration. Anal. Chem. 201 1, 83, 8341-8346.
157
151.Yang, H.; Sampson, M. M.; Senturk, D.; Andrews, A. M., Sex-and SERT-Mediated
Differences in Stimulated Serotonin Revealed by Fast Microdialysis. ACS Chem.
Neurosci. 2015, 6, 1487−1501.
152.Yang, H.; Thompson, A. B.; McIntosh, B. J.; Altieri, S. C.; Andrews, A. M.,
Physiologically Relevant Changes in Serotonin Resolved by Fast Microdialysis. ACS
Chem. Neurosci. 2013, 4, 790−798.
153.Yang, J. J.; Pickett, M. D.; Li, X.; Ohlberg, D. A.; Stewart, D. R.; Williams, R. S.,
Memristive Switching Mechanism for Metal/Oxide/Metal Nanodevices. Nat.
Nanotechnol. 2008, 3, 429– 433.
154.Yang, J. J.; Strukov, D. B.; Stewart, D. R. Memristive Devices for Computing. Nat.
Nanotechnol. 2013, 8, 13– 24.
155.Yao, H.; Liao, Y .; Lingley, A.; Afanasiev, A.; Lähdesmäki, I.; Otis, B.; Parviz, B., A
Contact Lens with Integrated Telecommunication Circuit and Sensors for Wireless and
Continuous Tear Glucose Monitoring. J. Micromech. Microeng. 20 12, 22, 075007.
156.Yao, H.; Shum, A. J.; Cowan, M.; Lähdesmäki, I.; Parviz, B. A., A Contact Lens with
Embedded Sensor for Monitoring Tear Glucose Level. Biosens. Bioelectron. 201 1, 26,
3290-3296.
157.Yu, G.; Upadhyaya, P.; Shao, Q.; Wu, H.; Yin, G.; Li, X.; He, C.; Jiang, W.; Han, X.;
Amiri, P. K.; Wang, K. L. Room-Temperature Skyrmion Shift Device for Memory
Application. Nano Lett. 2 017, 17, 261– 268.
158.Zavabeti, A.; Ou, J. Z.; Carey, B. J.; Syed, N.; Orrell-Trigg, R.; Mayes, E. L.; Xu, C.;
Kavehei, O.; O’mullane, A. P.; Kaner, R. B., A Liquid Metal Reaction Environment
for the Room-Temperature Synthesis of Atomically Thin Metal Oxides. Science 2017,
358, 332-335.
159.Zhai, D.; Liu, B.; Shi, Y .; Pan, L.; Wang, Y .; Li, W.; Zhang, R.; Yu, G., Highly Sensitive
Glucose Sensor Based on Pt Nanoparticle/Polyaniline Hydrogel Heterostructures. ACS
Nano 2013, 7, 3540-3546.
160.Zhang, M.; Liao, C.; Mak, C. H.; You, P.; Mak, C. L.; Yan, F., Highly Sensitive Glucose
158
Sensors Based on Enzyme-Modified Whole-Graphene Solution-Gated Transistors. Sci.
Rep. 2015, 5, 8311.
161.Zhang, W.; Du, Y .; Wang, M. L., Noninvasive Glucose Monitoring Using Saliva Nano-
Biosensor. Sens. Bio-Sensing Res. 2015, 4, 23-29.
162.Zhang, W.; Du, Y .; Wang, M. L., On-Chip Highly Sensitive Saliva Glucose Sensing
Using Multilayer Films Composed of Single-Walled Carbon Nanotubes, Gold
Nanoparticles, and Glucose Oxidase. Sens. Bio-Sensing Res. 2015, 4, 96-102.
163.Zhao, C.; Xu, X.; Bae, S.-H.; Yang, Q.; Liu, W.; Belling, J. N.; Cheung, K. M.; Rim,
Y. S.; Yang, Y.; Andrews, A. M., Large-Area, Ultrathin Metal-Oxide Semiconductor
Nanoribbon Arrays Fabricated by Chemical Lift-Off Lithography. Nano Lett. 2018,
18, 5590−5595.
164.Zhao, C.; Xu, X.; Yang, Q.; Man, T.; Jonas, S. J.; Schwartz, J. J.; Andrews, A. M.;
Weiss, P. S., Self-Collapse Lithography. Nano Lett. 2017, 17, 5035−5042.
165.Zheng, G. F.; Patolsky, F.; Cui, Y .; Wang, W. U.; Lieber, C. M., Multiplexed Electrical
Detection of Cancer Markers with Nanowire Sensor Arrays. Nat. Biotechnol. 2005, 23,
1294-1301.
166.Zheng, G.; Patolsky, F.; Cui, Y .; Wang, W. U.; Lieber, C. M., Multiplexed Electrical
Detection of Cancer Markers with Nanowire Sensor Arrays. Nat. Biotechnol. 2005, 23,
1294-1301.
167.Zhou, T.; Hong, G.; Fu, T.-M.; Yang, X.; Schuhmann, T. G.; Viveros, R. D.; Lieber,
C. M., Syringe-Injectable Mesh Electronics Integrate Seamlessly with Minimal
Chronic Immune Response in the Brain. Proc. Natl. Acad. Sci. 2017, 114,
5894−5899.
168.Zhu, L. Q.; Wan, C. J.; Guo, L. Q.; Shi, Y .; Wan, Q. Artificial Synapse Network on
Inorganic Proton Conductor for Neuromorphic Systems. Nat. Commun. 201 4, 5, 3158.
169.Zhu, X.; Lu, W. D. Optogenetics-Inspired Tunable Synaptic Functions in Memristors.
ACS Nano 2018, 12, 1242– 1249
170.Zucker, R. S.; Regehr, W. G. Short-Term Synaptic Plasticity. Annu. Rev. Physiol. 2002,
159
64, 355– 405.
Abstract (if available)
Abstract
In this dissertation, I present my work on the development of indium oxide nanoribbon field-effect transistors for biosensing applications and the development of artificial synapses using organic semiconductors. Sensors built from nanostructure metal oxide field-effect transistors (FET) have the combined advantages of high sensitivity, good flexibility, being equipped with an electronic read-out that can be fast-responding, wearable, and accessible. Non-volatile, flexible artificial synapses that can be used for brain-inspired computing are highly desirable for emerging applications such as human-machine interfaces, soft robotics, medical implants and biological studies. ❧ Chapter 1 is an introduction of biosensing and biomimetic electronics. It mainly focuses on the mechanism biological sensing of FET sensors based on metal oxide, and the mechanism of artificial synapses based on organic materials. ❧ In chapter 2, a scalable and facile lithography-free method for fabricating highly uniform and sensitive In₂O₃ nanoribbon biosensor arrays are demonstrated. Combining with electronic enzyme-linked immunosorbent assay (ELISA) for signal amplification, the In₂O₃ nanoribbon biosensor arrays are optimized for early, quick and quantitative detection of cardiac biomarkers in diagnosis of acute myocardial infarction. With the demonstrated sensitivity, quick turnaround time, and reusability, the In₂O₃ nanoribbon biosensors have shown great potential toward clinical test for early and quick diagnosis of acute myocardial infarction (AMI). ❧ In chapter 3, highly sensitive and conformal In₂O₃ nanoribbon FET biosensors with fully integrated on-chip gold side gate are demonstraded. The devices can be laminated onto various surfaces, such as artificial arms and watches, and have enabled glucose detection in various body fluids, such as sweat and saliva. With the electrodes functionalized with glucose oxidase, chitosan, and single-walled carbon nanotubes, the glucose sensors show very wide detection range spanning at least 5 orders of magnitude and detection limit down to 10 nM. Therefore, our high-performance In₂O₃ nanoribbon sensing platform has great potential to work as indispensable components for wearable healthcare electronics. ❧ In chapter 4, ultra-flexible and highly sensitive aptamer-field-effect-transistor In₂O₃ nanoribbon biosensors for real-time multiplexed neurotransmitters sensing are demonstrated. Arrays of In₂O₃ nanoribbon field-effect transistors were fabricated on 1.4-μm-thick plastic substrate using shadow mask techniques, showing excellent electrical performance and device uniformity. The conformal sensor array exhibited multiplexed sensing of temperature, pH, serotonin, and dopamine, and can function properly in artificial cerebrospinal fluid and on an artificial brain. These results represent significant progress in the fabrication of ultra-flexible aptamer-field-effect-transistors for brain mapping and physiological monitoring applications. ❧ In chapter 5, the experimental realization of a non-volatile artificial synapse using organic polymers in a scalable fabrication process are demonstrated. The three-terminal electrochemical neuromorphic device successfully emulates the key features of biological synapses: long-term potentiation/depression, spike-timing-dependent plasticity learning rule, paired-pulse facilitation, and ultralow energy consumption. The artificial synapse network exhibits excellent endurance against bending tests and enables a direct emulation of logic gates, which shows the feasibility of using them in futuristic hierarchical neural networks. Based on our demonstration of 100 distinct, non-volatile conductance states, we achieved high accuracy in pattern recognition and face classification neural network simulations. ❧ The last chapter, chapter 6, is the summary and future direction of biosensing and biomimetic electronics.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
One-dimensional nanomaterials for electronic and sensing applications
PDF
Surface functionalization of nanomaterials and development of field effect transistor nanobiosensors
PDF
In2O3 COVID-19 biosensors and two-dimensional materials: synthesis and applications
PDF
Semiconducting metal oxide nanostructures for scalable sensing applications
PDF
Surface functionalization of nanomaterials and the development of nanobiosensors
PDF
Hot carrier enhanced photocatalysis in plasmon resonant metal grating systems
PDF
Printed electronics based on carbon nanotubes and two-dimensional transition metal dichalcogenides
PDF
GaAs nanowire optoelectronic and carbon nanotube electronic device applications
PDF
Printed and flexible carbon nanotube macroelectronics
PDF
Synthesis, characterization, and device application of two-dimensional materials beyond graphene
PDF
Carbon nanotube macroelectronics
PDF
Integrated photonics assisted electron emission devices
PDF
Circuit design with nano electronic devices for biomimetic neuromorphic systems
PDF
Growth, characterization of gallium arsenide based nanowires and application in photovoltaic cells
PDF
Multi-scale biomimetic structure fabrication based on immersed surface accumulation
PDF
Carbon nanotube nanoelectronics and macroelectronics
PDF
Carbon material-based nanoelectronics
PDF
Hot carriers in bare metals and photocatalytically active defect sites in dielectric/metal structures
PDF
Optoelectronic, thermoelectric, and photocatalytic properties of low dimensional materials
PDF
Power-efficient biomimetic neural circuits
Asset Metadata
Creator
Liu, Qingzhou
(author)
Core Title
Biosensing and biomimetic electronics
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Materials Science
Publication Date
01/24/2020
Defense Date
06/11/2019
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
biomimetic electronics,biosensor,OAI-PMH Harvest,semiconductors,wearable electronics
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Zhou, Chongwu (
committee chair
), Nakano, Aiichiro (
committee member
), Nutt, Steve R. (
committee member
)
Creator Email
liuqingzhou88@gmail.com,qingzhol@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-197151
Unique identifier
UC11662946
Identifier
etd-LiuQingzho-7653.pdf (filename),usctheses-c89-197151 (legacy record id)
Legacy Identifier
etd-LiuQingzho-7653.pdf
Dmrecord
197151
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Liu, Qingzhou
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
biomimetic electronics
biosensor
semiconductors
wearable electronics