Carbon nanotube field effect transistor aptasensors for estrogen detection in liquids

Han Yue Zheng, Omar A. Alsager, Cameron S. Wood, Justin M. Hodgkiss, and Natalie O. V. Plank

Citation: J. Vac. Sci. Technol. B 33, 06F904 (2015); doi: 10.1116/1.4935246

View online: http://dx.doi.org/10.1116/1.4935246
View Table of Contents: http://avs.scitation.org/toc/jvb/33/6
Published by the American Vacuum Society
Carbon nanotube field effect transistor aptasensors for estrogen
detection in liquids
Han Yue Zheng, Omar A. Alsager, Cameron S. Wood, Justin M. Hodgkiss, and
Natalie O. V. Planka)
School of Chemical and Physical Sciences, Victoria University of Wellington, Wellington 6021, New Zealand and
The MacDiarmid Institute for Advanced Materials and Nanotechnology, Victoria University of Wellington,
Wellington 6021, New Zealand
(Received 22 June 2015; accepted 19 October 2015; published 9 November 2015)
The authors demonstrate a small molecule 17 b-estradiol (E2) sensor based on aptamer functional-
ized carbon nanotube network film field effect transistors (CNT FETs). The real time current
response for the 35-mer E2 aptamer functionalized CNT FET shows a clear increase in current
over the range of 50 nM to 1.6 lM of E2. The E2 response using a longer 75-mer version of the
aptamer functionalized CNT FETs, where the aptamer/E2 binding occurs beyond the Debye length,
shows no obvious evidence of sensing. The CNT FET sensing platform has been fabricated via a
simple surfactant free solution processing route, compatible with further carbon nanotube functional-
ization to develop a versatile sensing platform. The CNT FET aptasensors are able to perform real
time monitoring of E2 levels for selective and quantitative detection of E2 in liquids. V C 2015

American Vacuum Society. [http://dx.doi.org/10.1116/1.4935246]

I. INTRODUCTION imaging,21,22 microgravimetric analysis,23 and by electronic

Hormone contamination in waterways could cause serious
problems for the environment, as well as being detrimental
to human health.1–8 There is demand for simple devices
capable of detecting low levels of hormones such as 17
b-estradiol (E2); however, the small size of the molecule
makes accurate sensitive real-time detection challenging.
We are investigating a route to detect E2 molecules sensi-
tively and selectively in real time using liquid gated carbon
nanotube field effect transistors (CNT FETs) functionalized
with aptamers.
Aptamers are short strands of DNA or RNA and can be
selected in vitro for a specific target using systematic evolution
of ligands by exponential enrichment techniques.9,10
Compared to other recognition receptors, for example, anti-
bodies,11,12 using aptamers as the recognition elements in bio-
sensors, offer several advantages.13,14 The artificially
synthesized aptamers allow a very broad detection of analytes,
as aptamers can be selectively synthesized for specific targets,
including cells, proteins, viruses, and small molecules, which
is a more versatile target range than for antibody based sen-
sors.1,13,14 Another major advantage of aptamers is that the 50
or 30 end sites of the aptamers can be readily functionalized
with a range of functional groups, such as amines, thiols, and
carboxylic acids, which is beneficial for immobilization onto
the sensing platforms.13,14 Furthermore, the handling of
aptamers is significantly easier and more robust than for anti-
bodies. Usually, antibodies must be maintained in buffer under
specific conditions13,14 whereas aptamers are found to be sta-
ble in a variety of buffers and even in water resulting in simple
functionalization procedures for biosensor fabrication.2,3,14–16
Previous biosensor research based on aptamers as the rec-
ognition elements has used a range of analytical techni-
ques,14,17 including electrochemical sensing,5,18–20,26 optical
a)
Electronic mail: Natalie.Plank@vuw.ac.nz
signal readout via functionalized FETs.24,25 Electrochemical
methods for detection of E2 molecules have included the use
of aptamer functionalized conducting electrodes as the sensing
platform for cyclic voltammetry measurements,20,26 square
wave voltammetry,20 or impedance measurements.5,19 Optical
techniques for aptasensors have utilized aptamer functional-
ized nanoparticles or fibers to determine the E2 concentration
by colorimetric analysis2,27 or fluorescence spectroscopy.21
Using FET sensor platforms, it has been possible to oper-
ate the FETs at low bias (<1 V) and to measure real time
quantitative electronic responses via a current meter, paving
the way for sensors that are integrated with handheld elec-
tronics. The current examples of aptamer based CNT FET
biosensors have all been utilized to detect large molecules in
real time, for example, thrombin28 or IgE.25,29 The detection
of small molecules such as adenosine triphosphate (ATP)
has been achieved using a CNT FET aptasensor device;
however, the detection measurements were from the carbon
nanotube (CNT) resistance between two electrodes, e.g., the
FET device was operated as a simple chemiresistor.30 The
chemiresistor measurements based on FETs do successfully
detect small molecules; however, they do not realize real
time monitoring sensing response. The real time detection of
using CNT FET aptasensors is still in its infancy with only
one report till date, for bisphenol A (BPA) detection.31 Here,
we show successful real time detection of the small molecule
E2 by thin film CNT FET aptasensors. To the best of our
knowledge, this is the first example of real time response for
E2 detection using CNT FET aptasensors.
The CNT FET aptasensors were fabricated using the 35-
mer E2 aptamer and 75-mer E2 aptamer.2 We are able to
show that successful detection of E2 depends on the length
of the aptamer, as the conformational change in the aptamer
structure upon exposure to the E2 analyte must occur within
the Debye screening length to affect the CNT FET

06F904-1 J. Vac. Sci. Technol. B 33(6), Nov/Dec 2015 2166-2746/2015/33(6)/06F904/8/$30.00 C 2015 American Vacuum Society
V 06F904-1
06F904-2 Zheng et al.: CNT FET aptasensors for estrogen detection in liquids
current.12,25,32 We have observed a measurable current
change on exposure to E2 for the 35-mer E2 CNT FET apta-
sensor, but no E2 response is observed for the 75-mer E2
aptamer. We also tested a control device with a random 35-
mer DNA sequence, which shows no E2 specific response.
II. EXPERIMENT
A. Carbon nanotube thin film transistor fabrication
Carbon nanotube thin film transistors are fabricated
through a surfactant free solution processing route. First of
all, a CNT suspension was produced in anhydrous 1,2-
dichlorobenzene (DCB) (Sigma Aldrich). A tweezer tip
amount of 99% IsoNanotube-S CNTs bucky paper
(NanoIntegris) was dispersed into a 5 ml DCB solvent and
was sonicated using the Sonorex ultrasonic water bath for 1
h to create the suspension to the point where there are no
obvious particles visible through the naked eye. The SiO2/Si
substrates (Silicon Quest International, Inc.) with highly
doped p-type Si and an oxide thickness of 100 nm were
cleaned by sonication in acetone for 3 min and followed by
thoroughly rinsing in Isopropyl alcohol and then dried in
clean N2. In order to functionalize the SiO2 surface, a polydi-
methylsiloxane (PDMS) (Sylgard 184) stamping method
was used to transfer a thin layer of 2-thiolpyridine onto SiO2
surface.33 The 2-thiolpyridine functionalized SiO2 substrates
were then submerged into the premade CNT DCB suspen-
sion for half an hour. The SiO2/Si substrates were then taken
from CNT suspension and cleaned in ethanol for 5 min to
remove excess solvents before being dried in clean N2.
After submersion, the entire SiO2 surface was coated with
a uniform CNT thin film. In our experiment, 100 lm (width)
 300 lm (length) areas of CNT thin films are left at controlled
locations defined by photolithography using O2 plasma etching
using an Oxford Plasmalab 80 Plus etcher at 600 mTorr, 200
W, 40 sccm O2 flow, and an etch time of 3 min. These CNT
regions then form the active channel of FETs. We use the Karl
Suss MJB3 UV lamp mask aligner for photolithography.
Metallization is achieved using the Angstrom Engineering
Nexdep Evaporator with contacts of 5 nm Cr and 35 nm Au de-
posited directly onto the CNT thin films as the source and drain
electrodes. After a lift-off process, the channel length of CNT
FETs is 40 lm, and the channel width is 100 lm.
To encapsulate electrodes to avoid current leakage and
electrode damage in the aqueous sensing environment, pho-
toresist SU8 2150 from Microchemicals was patterned by
photolithography. The SU8 2150 photoresist was diluted
using cyclopentane (Simga Aldrich, reagent 98%) at a
weight ratio of 1.8:1 before use. The active exposed area of
CNT is 10 lm length by 100 lm width. SU8 encapsulated
TABLE I. DNA sequences.
DNA
35-mer E2 5 -NH2-AAGGGATGCCGTTTGGGCCCAAGTTCGGCATAGTG-30
0
35-mer random 50 -NH2-ACGGGTGGCCGCCAGGTCTTGAAGTGGCAGTATTA-30
75-mer E2 5 -NH2-ATACGAGCTTGTTCAATACGAAGGGATGCCGTTTGGGCCCAAGTTCGGCATAGTGTGGTGATAGTAAGAGCAATC-30
0
J. Vac. Sci. Technol. B, Vol. 33, No. 6, Nov/Dec 2015
06F904-2
CNT FET was then baked at a 200  C on a hot plate in air.
Scanning electron microcopy (SEM) via the Jeol JSM 6500F
was used to determine the morphology of the CNTs in the
device platform. The CNT FETs are then ready for chemical
functionalization to fabricate aptasensors.
B. CNT aptasensor fabrication
To immobilize aptamers on the surface of the CNTs,
1-pyrenebutanoic acid succinimidyl ester (PBASE) (95%
purity) (Sigma Aldrich) was exposed to the CNTs by sub-
merging the entire device in a 1 mM PBASE methanol solu-
tion for an hour. The PBASE methanol solution of 1 mM
concentration was prepared by sonication for 1 min before
use. After PBASE functionalization, any excess PBASE was
removed by rinsing three times in pure methanol before
finally dipping in deionized (DI) water for 5 s. The samples
are then dried in clean N2.
All aptamers were purchased from AlphaDNA. The
sequence of 35-mer E2 aptamer, 35-mer random sequence,
and 75-mer E2 aptamer is listed in Table I. The aptamers
were hydrated in DI water and stored at 4  C when not in
use. To prepare the aptamer solution, the aptamers were
initially denatured at 75  C in an oven for 5 min before
cooling down to room temperature. The aptamers were
then diluted using 10 mM phosphate-buffered saline
(PBS) buffer (1 PBS, pH ¼ 7.4, 0.0027 M potassium chlo-
ride, and 0.137 M sodium chloride) (Sigma Aldrich) at a
concentration of 2 lM mixed together with 1 mg/ml N-
(3-dimethylaminopropyl)N-ethylcarbodiimidehydrochloride/
N-hydroxysuccinimide (EDC/NHS). CNT FETs were
exposed to the aptamer/EDC/NHS solution for 3 h. After
functionalization, excess aptamers were then washed off by
rinsing in PBS buffer three times followed by a rinse in
DI water before drying in clean N2. The remaining exposed
CNT FETs were passivated by submerging the devices in
0.05 wt. % Tween20 (Sigma Aldrich) in PBS for 2 h. The
Tween20 solution was prepared by stirring Tween20 in 1
PBS at a weight ratio of 0.05% until well mixed. Excess
Tween20 was removed by rinsing the CNT FETs in PBS
solution three times before a final clean in DI water and
drying in clean N2.
C. Electrical measurement
The analyte 17 b-estradiol (E2) (Sigma Aldrich) was dis-
solved in ethanol at a concentration of 0.03 M. The E2 etha-
nol solution is stored at 4  C when not in use. The 0.03 M
E2 ethanol solution was then diluted to 20 lM concentration
using diluted 0.05 PBS (5% ethanol). More diluted concen-
trations were also prepared in 0.05 PBS (5% ethanol)
Sequence
06F904-3 Zheng et al.: CNT FET aptasensors for estrogen detection in liquids
FIG. 1. (Color online) (a) SEM image shows uniform CNT film on SiO2/Si substrates. (b) A schematic diagram of CNT FET measured in liquid using Ag/
AgCl as reference electrodes. (c) An overview optical image of CNT FET post SU8 electrode encapsulation: the areas of CNTs exposed for aptamer function-
alization are indicated by the rectangular solid lines.
solution if required. To test the devices in a liquid environ-
ment, a handmade PDMS well was placed on top of the
CNT FET to avoid liquid leakage to the probe electrodes.
Buffer solution of 150 ll was added into PDMS well before
measurement to ensure that the exposed active area of the
reference electrode, which is enclosed in plastic up to the
end pad, was completely covered in buffer at all times to pre-
vent the issue of an altered I-V response coming from a
change in the active area of the gate electrode. All electrical
measurements were taken with the Agilent 4156C parameter
analyzer connected to the sample via micromanipulators and
a Rucker and Kolls probes station. During real time sensing
measurements, the applied voltage and time are set from pa-
rameter analyzer. The gate voltage was set to 0, the source–-
drain voltage was set to 50 or 100 mV, and the time step was
1 s. Throughout the sensing measurements, 10 ll of the E2
analyte was added. The concentration of E2 in PDMS varies
from 50 nM, 147 nM, 640 nM, and 1.6 lM.
III. RESULTS AND DISCUSSION
A. Device platform
Uniform films of CNTs have been successfully realized on
the 2-thiolpyridine functionalized SiO2/Si substrates [Fig. 1(a)]
via the CNT DCB suspension method detailed in Sec. II. In order
to use CNT FETs as sensor platforms, the CNT FET electrodes
are encapsulated with SU8 to prevent leakage currents from the
reference electrode and the source and drain, as schematically
FIG. 2. (Color online) Transfer characteristics of the same CNT FET using
back gating (Vds ¼ 100 mV) and liquid gating (Vds ¼ 50 mV) after SU8 elec-
trode encapsulation.
JVST B - Nanotechnology and Microelectronics: Materials, Processing, Measurement, and Phenomena
06F904-3
shown in Fig. 1(b). Figure 1(c) shows an optical microscope
image of the electrodes post SU8 encapsulation. The exposed
CNT area marked by the rectangle is 100  10 lm where those
CNTs are the active area for aptamer functionalization.
The transfer characteristics of a CNT FET are shown in
Fig. 2 for both the case of a back gated CNT FET through the
SiO2 and a liquid gated CNT FET. As expected, the CNT
FETs exhibit p-type semiconductor behavior. In back gating,
the voltage Vbg is applied through the highly doped Si, which
acts as the electrode, and the 100 nm thermal SiO2 is an effec-
tive dielectric. The voltage Vbg is swept from þ10 to 10 V
and a fixed Vds ¼ 100 mV. The back gated device shows poor
field dependency with undefined Vth. In liquid gating, the volt-
age Vlg is applied through the Ag/AgCl reference electrode.
Charges within the PBS buffer redistribute due to the applied
voltage, and an electric double layer (EDL) is formed, which
acts as the dielectric region.32,34 The transfer characteristics
achieved from the same CNT FET show much stronger field
dependency properties under liquid gating compared to back
gating, even at a shorter sweeping voltage range from þ0.5 to
0.5 V and a fixed Vds ¼ 50 mV. Moreover, for the liquid
gated device, there is a clearly defined Vth ¼ þ0.1 V. To oper-
ate the CNT FET as a biosensor for E2, it is advantageous to
operate the device under a liquid gate environment. Figure 2
clearly shows that these network CNT FETs exhibit suitable
device performance.
B. Aptamer immobilization
In order to immobilize aptamers on the CNT FET surfaces,
it is necessary to functionalize the pristine CNTs, as described
in Sec. II. PBASE is chosen to act as the linker between the
CNT and the aptamer because it provides nondestructive func-
tionalization of the CNT surface, meaning that the structure
and hence the electronic properties of the CNTs are not dam-
aged. The aptamer used in this research is functionalized by
amine groups (-NH2) at the 50 end. In our experiments, the
CNT FETs are first immersed in PBASE methanol solu-
tion.12,35 PBASE stacks onto the CNT sidewalls due to the
p–p interaction between the pyrene on PBASE and the ben-
zene rings on CNTs, as shown schematically in Fig. 3(a).
Once the PBASE functionalized CNT FET is immersed in the
aptamer/EDC/NHS solution for 3 h, the amine groups on the
50 end of the aptamer are then linked onto PBASE by nucleo-
philic substitution and formation of an amide bond (OC-NH)
between the aptamer and PBASE.35 The NH2 terminated
aptamers are the most widely available form used for
06F904-4 Zheng et al.: CNT FET aptasensors for estrogen detection in liquids
FIG. 3. (Color online) Schematic representation of the CNT FET functionalization process: (a) PBASE stacks onto the CNT surfaces via p–p interaction. (b)
Aptamers are tethered on to the PBASE coated CNTs, via nucleophilic substitution. (c) When the aptamer senses the E2, the DNA strand is wrapped around
the analyte, bringing charges closer to the CNT surface.
conjugation and aptasensor research.36 Primary amines on
base groups of DNA as part of aromatic rings are not suffi-
ciently reactive for controlled binding, unlike the alkyl amine
added to the end of the aptamer.37 Therefore, the alkyl amine
terminated aptamer on the 50 end is necessary to link onto the
N-hydroxysuccinimide group of PBASE to form the amide
tether.35 Without the NH2 termination, the aptamers will
adsorb nonselectively onto the CNTs due to the p–p interac-
tion between the benzene ring on CNT sidewall and the purine
and pyrimidine bases on aptamer.38 However, with the NH2
termination, the aptamer linking is controlled, and it takes
place on the PBASE, as schematically shown in Fig. 3(b). To
avoid hydrolysis of the PBASE during the aptamer functional-
ization in liquid, the aptamer solution is prepared in a buffer
containing the EDC/NHS coupling solution. This ensures that
EDC/NHS activates the carboxyl groups if hydrolysis does
happen.1,15 After the aptamer immobilization, the entire CNT
FET device chips are submerged in 0.05 wt. % Tween20 to
passivate any remaining unfunctionalized CNTs and the
exposed SiO2 substrate to avoid nonspecific adsorption of the
target analyte during the sensing measurements.28,30
Aptamers are negatively charged due to the phosphate
groups of the DNA backbone. After recognition between the
E2 aptamer and the E2 target analyte, the conformation of
the aptamer will change and it may appear as though the
aptamer “latches on” to the E2 molecule [Fig. 3(c)]. The
change in conformation of the aptamer from random coil
motion in the solution to the aptamer wrapping onto the tar-
get brings the negative charges from the DNA closer to the
CNT surface. Various degrees of folding are expected for
the unbound aptamer; however, the more compact conforma-
tion associated with E2 binding means that the negative
charges from the DNA will, on average, be closer to the sur-
face of the CNTs than in the unbound state. The binding
induced more compact conformation has been demonstrated
by previous studies using this aptamer tethered to nanopar-
ticles1 or electrode surfaces.5 The conformational change of
the aptamer structure also alters the charge distribution in
the EDL.1,15,25,32 The subsequent screening effect due to
charge redistribution has a large impact on the current
response of the CNT FET, and it is this effect that we aim to
observe in real time for the devices.
C. Electrical measurement
1. Transfer characteristics
The transfer characteristics for a bare CNT FET, the same
CNT FET functionalized with the 35-mer E2 aptamer (with
J. Vac. Sci. Technol. B, Vol. 33, No. 6, Nov/Dec 2015
06F904-4
Tween20 passivation), and the same CNT FET post E2 response
are all compared under liquid gating in 0.05 PBS (Fig. 4).
It is clear that after the CNT FET has been functionalized
with the E2 aptamer and Tween20 passivation, the magni-
tude of the current is reduced in comparison to the pristine
CNT FET. The reduction in current is mainly due to the
Tween20 passivation, which can act as scattering centers on
the CNT surface and has the effect of both hindering the
hole transportation along the CNTs and influencing the tube
to tube junction sites in the percolating network of CNTs.
After exposure to 1.6 lM of the E2 analyte, the functional-
ized CNT FET device shows an increase in current (Fig. 4),
which can be explained by the negatively charged molecules
of the DNA aptamer moving closer to CNT surface and
changing the effective gating of the device.34 During the liq-
uid gate measurements of the CNT FETs, an EDL is formed
close to the CNT surface. The thickness of the EDL is the
Debye length.25,32,34 In CNT FET aptasensors, the confor-
mational change of negatively charged aptamers after bind-
ing alters the EDL at the CNT surface. For the binding
events that take place within the Debye length, a change in
the current response will be observed for the CNT FET due
to the effective change to the gate potential. Thus, the length
of recognition element prior to and upon detection is critical.
Using the transfer characteristics of the CNT FETs (Fig.
4), we have been able to determine that the E2 molecule has
altered the current of the device. Previously, Pacios et al.15
have observed changes in the CNT FET transfer characteris-
tics for a thrombin aptamer functionalized CNT FET to
FIG. 4. (Color online) Liquid gate performance of a pristine CNT FET, the
same CNT FET functionalized with the 35-mer E2 aptamer and 0.05 wt. %
Tween20 passivation, and the same CNT FET postexposure to 1.6 lM con-
centration of E2. The measurement is swept from þ1 V to 500 mV at fixed
Vds ¼ 100 mV. The device is measured in 0.05 PBS using an Ag/AgCl ref-
erence electrode as the gate.
06F904-5 Zheng et al.: CNT FET aptasensors for estrogen detection in liquids 06F904-5

detect thrombin molecules. They observed that the current

increases when the thrombin concentration increases from
0.1 to 1 nM, but the current then starts decreasing when the
concentration increases from 1 to 5 nM.15 Our results are
consistent with their low thrombin concentration detection
results. They also conclude that the reduced current is due to
the aptamer/target binding, which causes the negatively
charged aptamers to move closer to the CNT surface and
alters the effective gating. However, the thrombin protein
molecule is slightly negatively charged; therefore, the altera-
tion to the ionic strength in the buffer results in a decreased
current with a large quantity of thrombin. In our system,
there are some differences. First, the E2 molecule is neutral,
and so, the ionic strength of the buffer plus analyte liquid
remains the same even at high concentrations; therefore, we
do not observe a reduced current for the CNT FET across
our E2 concentration range. Second, the E2 molecule is
physically small, whereas thrombin is large. It is possible
that the thrombin molecule itself disrupts the charge transfer
along the CNT networks, whereas this is unlikely to happen
for the E2 case. Detection of small molecules via CNT FET
aptasensors is still in its infancy and up till date used only
for ATP (Ref. 30) and BPA detection.31 Although transfer
characteristics are able to determine recognition (comparing
the shift prior and post E2 exposure), this is not a true real
time detection as had been previously observed for the apta-
mer–thrombin system.15

2. Influence of aptamer length on real time response

to E2
The real time current response to E2 was tested for CNT
FETs that were functionalized with the 35-mer E2 aptamer
and the 75-mer E2 aptamer [Figs. 5(a) and 5(b)], respec-
tively. The concentration of E2 added into the PDMS well is
increased from 50 nM to 147 nM, 640 nM, and 1.6 lM
(where the total concentration takes into account the content
in the well prior to each addition). To do so, 10 ll of E2 solu-
tion was added every 10 min, and the current output was
monitored in real time [Figs. 5(a) and 5(b)]. For the 35-mer
E2 aptamer functionalized CNT FET in Fig. 5(a), the current
after adding 1.6 lM E2 to the well increases by 25% with
respect to the initial current I0. However, over identical mea-
surement conditions, there is almost no response to E2 from
the 75-mer E2 aptamer functionalized CNT FET [Fig. 5(b)].
As a further control, the response to PBS for the 35-mer E2
aptamer functionalized CNT FET was tested under similar
conditions [Fig. 5(c)], and although a very slight increase in
current over the 3000 s measurement is observed, there is no
clear detection signal. With incorrect passivation or elec-
trode preparation, a CNT FET device can show changes due
to PBS addition that can be confused with a real sensing
event. With this in mind, we reliably conclude that the 35-
mer E2 aptamer functionalized CNT FET is showing a real
time response to the E2 present in the PDMS well.
As discussed in Sec. III C 1, during measurements in liq-
uid, an EDL forms nearby the CNT surface. The sensitivity
of the biosensor strongly depends on the EDL
FIG. 5. (Color online) Real time response to E2 for (a) CNT FET þ 35-mer
E2 aptamer; (b) CNT FET þ 75-mer E2 aptamer; and (c) a control measure-
ment from a CNT FET þ 35-mer E2 aptamer response PBS buffer.
thickness.12,15,32 The thickness of EDL and the subsequent
Debye length is inversely related to the ionic strength of the
liquid,14,15,32 e.g., by reducing the ionic strength of the
buffer, the Debye length is increased. The buffer during the
real-time current measurements shown here was diluted
0.05 PBS. The Debye length in 0.05 PBS buffer is
4.6 nm (compared to 1 nm for 1 PBS buffer).15 The
length of the aptamer also has an influence on the successful
operation of the device as a sensor. For the E2 aptamers, the
fully extended length is about 11 nm for the 35-mer and
25 nm for the 75-mer. Various degrees of folding are
expected for the unbound aptamers, and the length of the
random coil structures is shorter than the fully extended
length of the aptamers. However, an alteration in the current
response of the device will be only possible if upon binding
of E2 the conformational change to the aptamer takes place
within the Debye length. The conformational change upon
binding to the target molecule brings the negative charges of

JVST B - Nanotechnology and Microelectronics: Materials, Processing, Measurement, and Phenomena

06F904-6 Zheng et al.: CNT FET aptasensors for estrogen detection in liquids 06F904-6

aptamer/E2 sensors in the literature. The linear relationship

FIG. 6. (Color online) Relative current response vs E2 concentration in loga-
rithmic scale from the 35-mer E2 aptamer functionalized CNT FET.
the aptamer close to the CNT surface and fixes them into
position. We can conclude from the response to E2 shown in
Figs. 5(a) and 5(b) that the E2 binding occurs within the
Debye length for CNT FETs functionalized with the 35-mer
aptamer; however, for the CNT FETs with the 75-mer
aptamer, the change in conformation of the aptamer due to
the presence of E2 happens beyond the Debye length 4.6 nm.
There is no clear current response for the CNT FET with the
75-mer aptamer even after exposure to 1.6 lM concentration.
The detection limit associated with EDL thickness has also
been observed by Kim et al.12 using a fragment antibody
instead of a whole antibody to ensure that binding occurs
within EDL for immunoglobulin (IgE) detection to improve
the sensitivity from 1000 ng/ml to 1 pg/ml by comparing the
current at each concentration.12 Figure 6 shows the stabilized
current response of the CNT FET þ 35-mer E2 aptamer post
E2 exposure at 50 nM, 147 nM, 640 nM, and 1.6 lM with
reference to the initial current I0. The current DI shows a lin-
ear increase as the concentration of the analyte is increased
logarithmically, with the linear coefficient, R2 ¼ 0.9987.
Although our results are the only real time detection of
E2 from a CNT FET aptasensor, there are other examples of
between the concentration of E2 and the detection signal is
also observed in aptasensors fabricated on other sensing plat-
forms such as aptamer coated Au nanoparticle sensing plat-
forms.2,26 Huang et al. also observed a linear relationship for
detection of E2 against the logarithmic concentration range
from 10 pM to 10 nM using pulse voltammetry measure-
ments. Alsager et al.2 found a linear relationship for detection
of E2 from 200 to 800 pM (R2 ¼ 0.94) using colorimetric
analysis of nanoparticle aggregates in solution. Lin et al.19
reported an E2 detection limit of 10 pM from electrochemical
impedance measurements based on aptamer functionalized
Au electrode sensing platform. We can therefore trust that the
aptamer/E2 binding is known to be reliable for sensing appli-
cations in a variety of different sensor architectures. We com-
pared our E2 detection results to presently known E2
detection based on aptasensors, as shown in Table II.
Although our experimental detection limit at 50 nM based on
our CNT FET aptasensors is higher than previously published
E2 detection based on voltammetry,20,26,39 colorimetric,2,27
impedance spectra,5,8,19 or fluorescence characteriza-
tions,1,21,40 the E2 sensing based on our sensing platform is a
fast and direct method with a simple device platform and able
to real time monitor the aptamer/E2 recognition that fulfills
the demand for environmental screening.
To further ensure that our sensing signal is truly a selec-
tive response, we have also compared the response to E2 for
the CNT FETs functionalized with the 35-mer E2 aptamer to
a 35-mer randomly sequenced aptamer (consists of the same
base composition but in a random order, see Table I). Figure
7 shows the DI/I0 vs E2 concentration plot for the CNT FET
functionalized with the 35-mer E2 aptamer (squares) and the
35-mer random aptamer (circles) along with the current
response of a 35-mer E2 aptamer functionalized CNT FET
to PBS buffer (triangles).
The results clearly show a much weaker response to E2
from the 35-mer random DNA sequence CNT FET device in

TABLE II. E2 molecule detection based on aptasensors.

Sensor platform Analytical techniques Detection limit (M) Dynamic range (M) Real time Year and references

Au electrode Electrochemical voltammetry 1  1010 1  1011 to 1  109 no 2007 (Ref. 20)

Poly(3,4-ethylenedioxylthiopene)-gold Electrochemical voltammetry 2  1011 1  1010 to 1  107 no 2010 (Ref. 39)
nanocomposite
Au nanoparticles and vanadium disulfide Electrochemical voltammetry 1  1012 1  1011 to 1  108 no 2014 (Ref. 26)
nanoflowers
Au electrode Electrochemical impedance spectra 2  1012 1  1011 to 1  108 no 2012 (Ref. 19)
Dendritic gold modified boron-doped Electrochemical impedance spectra 5  1015 1  1014 to 1  109 no 2014 (Ref. 8)
diamond electrode
Nanoporous conducting polymer Electrochemical impedance spectra 1  1015 1  1015 to 1  106 no 2015 (Ref. 5)
Evanescent wave all-fiber Fluorescence 2.1  109 5  109 to 7.5  108 Time-course 2012 (Ref. 21)
Au nanoparticles Fluorescence 5  109 5  109 to 1.5  107 no 2014 (Ref. 1)
Quantum dot Fluorescence 2.2  1010 8.2  1010 to 2.05  108 no 2014 (Ref. 40)
CdSe nanoparticles modified Photocurrent 3.3  1014 5  1014 to 1.5  1011 Time-course 2014 (Ref. 3)
TiO2 nanotube
Au electrode Electrochemiluminescence 1.1  1012 1  1011 to 1  108 no 2014 (Ref. 41)
Au nanoparticles Colorimetric 2  1010 2  1010 to 8  1010 no 2015 (Ref. 2)
Carbon nanotube Transistor device output current 5  108 5  108 to 1.6  106 yes This work

J. Vac. Sci. Technol. B, Vol. 33, No. 6, Nov/Dec 2015

06F904-7 Zheng et al.: CNT FET aptasensors for estrogen detection in liquids 06F904-7

increase in current over the range of 50 nM to 1.6 lM of E2.

For the same device structure functionalized with the 75-mer
E2 aptamer, no sensing response is observed over the same
concentration range of the E2 analyte due to the detection
taking place beyond the Debye screening length. Our 35-mer
aptamer functionalized CNT FETs sensors are the first
example of electronic real-time detection of E2, paving the
way for mobile sensors for selective and quantitative detec-
tion of E2 in liquids.

ACKNOWLEDGMENTS
H.Z. thanks Victoria University of Wellington and the
FIG. 7. (Color online) Normalized current change (DI/I0) in response to the Chinese Government joint Scholarship. N.P. thanks the
added concentration of the E2 analyte for the 35-mer E2 aptamer CNT FET, Foundation for Research Science and Technology Project
the random DNA sequence CNT FET, and the 35-mer E2 aptamer CNT
FET response to PBS buffer.
No. VICX0911 and the Marsden fund, Project No. E1728.
O.A. acknowledges a fellowship from King Abdulaziz City
comparison to the 35-mer E2 aptamer CNT FET. Although for Science and Technology.
there is a small increase in current at 50 nM, there is no real 1
O. A. Alsager, S. Kumar, G. R. Willmott, K. P. McNatty, and J. M.
further increase over the range up to 1.6 lM (15% of the Hodgkiss, Biosens. Bioelectron. 57, 262 (2014).
2
signal observed with E2 recognition using the 35-mer O. A. Alsager, S. Kumar, B. Zhu, J. Travas-Sejdic, K. P. McNatty, and
aptamer), whereas the 35-mer E2 device shows a linear J. M. Hodgkiss, Anal. Chem. 87, 4201 (2015).
3
L. Fan, G. Zhao, H. Shi, M. Liu, Y. Wang, and H. Ke, Environ. Sci.
increase in current versus the logarithm of the concentration Technol. 48, 5754 (2014).
throughout the measurement range between 50 nM and 4
C. G. Campbell, S. E. Borglin, F. Bailey Green, A. Grayson, E. Wozei,
1.6 lM. The sensitivity of our CNT FET aptasensor is esti- and W. T. Stringfellow, Chemosphere 65, 1265 (2006).
5
mated to be 7%/decade, which is comparable to other elec- B. Zhu, O. A. Alsager, S. Kumar, J. M. Hodgkiss, and J. Travas-Sejdic,
Biosens. Bioelectron. 70, 398 (2015).
trical aptasensors in the literature.20,39 Kim et al.20 6
M. Auriol, Y. Filali-Meknassi, R. D. Tyagi, C. D. Adams, and R. Y.
fabricated E2 aptasensors on Au electrode platforms and Surampalli, Process Biochem. 41, 525 (2006).
7
observed a linear increase in current, as determined by vol- H.-S. Chang, K.-H. Choo, B. Lee, and S.-J. Choi, J. Hazard. Mater. 172, 1
tammetry measurements; with increased E2 concentration (2009).
8
H. Ke, M. Liu, L. Zhuang, Z. Li, L. Fan, and G. Zhao, Electrochim. Acta
over the range of 0.01–1 nM, the sensitivity is estimated to 137, 146 (2014).
be 10.5%/decade. Olowu et al.39 also used voltammetry 9
A. R. Oliphant, C. J. Brandl, and K. Struhl, Mol. Cell. Biol. 9, 2944 (1989).
10
measurements for E2 detection, by fabricating aptasensors A. D. Ellington and J. W. Szostak, Lett. Nat. 346, 818 (1990).
11
J. Oh, S. Yoo, Y. W. Chang, K. Lim, and K.-H. Yoo, Curr. Appl. Phys. 9,
on poly (3,4-ethylenedioxylthiopene)-Au nanocomposite
e229 (2009).
platforms. A linear increase in current versus the logarithm 12
J. P. Kim, B. Y. Lee, S. Hong, and S. J. Sim, Anal. Biochem. 381, 193
of the E2 concentration over 0.1–100 nM was observed, and (2008).
13
the estimated sensitivity is given as 7.3%/decade. Previous S. D. Jayasena, Clin. Chem. 45, 1628 (1999), available at http://
www.clinchem.org/content/45/9/1628.short.
work by Lee et al.31 has shown real time detection of BPA 14
J.-O. Lee, H.-M. So, E.-K. Jeon, H. Chang, K. Won, and Y. H. Kim, Anal.
using a CNT FET device where the Au electrodes are func- Bioanal. Chem. 390, 1023 (2008).
15
tionalized with BPA aptamer. Upon exposure to the analyte M. Pacios, I. Martin-Fernandez, X. Borrise, M. del Valle, J. Bartrolı, E.
mixed with a second BPA aptamer, the Schottky barrier of Lora-Tamayo, P. Godignon, F. Perez-Murano, and M. J. Esplandiu,
Nanoscale 4, 5917 (2012).
the device is altered, resulting in a sensitive detection signal. 16
T. Hianik, V. Ostatna, M. Sonlajtnerova, and I. Grman,
Although that method is effective, the possibility of short Bioelectrochemistry 70, 127 (2007).
17
circuits via the nonencapsulated gold electrodes and the liq- I. Willner and M. Zayats, Angew. Chem. Int. Ed. Engl. 46, 6408 (2007).
18
A.-E. Radi and C. K. O’Sullivan, Chem. Commun. (Cambridge) 2006,
uid gate electrode means that the further utilization of the
3432.
devices may be problematic, particularly for the detection of 19
Z. Lin, L. Chen, G. Zhang, Q. Liu, B. Qiu, Z. Cai, and G. Chen, Analyst
polar molecules. Our device structure avoids the possibility 137, 819 (2012).
20
of short circuits due to the SU-8 encapsulation of the electro- Y. S. Kim, H. S. Jung, T. Matsuura, H. Y. Lee, T. Kawai, and M. B. Gu,
Biosens. Bioelectron. 22, 2525 (2007).
des. The versatility of the functionalization of the CNTs 21
N. Yildirim, F. Long, C. Gao, M. He, H. Shi, and A. Z. Gu, Environ. Sci.
means that a wide variety of aptamer-analyte systems could Technol. 46, 3288 (2012).
22
be effectively utilized with this device architecture. K. Nam et al., J. Mater. Chem. 22, 23348 (2012).
23
S. Tombelli, M. Minunni, E. Luzi, and M. Mascini, Bioelectrochemistry
67, 135 (2005).
IV. SUMMARY AND CONCLUSIONS 24
Y. Ohno, K. Maehashi, and K. Matsumoto, J. Am. Chem. Soc. 132, 18012
We have successfully fabricated aptamer sensors based (2010).
25
K. Maehashi, T. Katsura, K. Kerman, Y. Takamura, K. Matsumoto, and E.
on functionalized CNT FETs for selective detection of the
Tamiya, Anal. Chem. 79, 782 (2007).
small molecule E2. The real time current response for the 26
K. Huang, Y.-J. Liu, G.-W. Shi, X.-R. Yang, and Y.-M. Liu, Sens.
35-mer aptamer functionalized CNT FET shows a clear Actuators B 201, 579 (2014).

JVST B - Nanotechnology and Microelectronics: Materials, Processing, Measurement, and Phenomena

06F904-8 Zheng et al.: CNT FET aptasensors for estrogen detection in liquids 06F904-8

27 34
J. Liu, W. Bai, S. Niu, C. Zhu, S. Yang, and A. Chen, Sci. Rep. 4, 7571 I. Heller, A. M. Janssens, J. M€annik, E. D. Minot, S. G. Lemay, and C.
(2014). Dekker, Nano Lett. 8, 591 (2008).
28 35
H.-M. So, K. Won, Y. H. Kim, B.-K. Kim, B. H. Ryu, P. S. Na, H. Kim, R. J. Chen, Y. Zhang, D. Wang, and H. Dai, J. Am. Chem. Soc. 123, 3838
and J.-O. Lee, J. Am. Chem. Soc. 127, 11906 (2005). (2001).
29 36
S. Okuda, S. Okamoto, Y. Ohno, K. Maehashi, K. Inoue, and K. H. Peng, L. Zhang, C. Soeller, and J. Travas-Sejdic, Biomaterials 30, 2132
Matsumoto, J. Phys. Chem. C 116, 19490 (2012). (2009).
30 37
B. K. Das, C. Tlili, S. Badhulika, L. N. Cella, W. Chen, and A. F. Luderer and U. Walschus, Top. Curr. Chem. 260, 37 (2005).
38
Mulchandani, Chem. Commun. (Cambridge) 47, 3793 (2011). G. A. Zelada-Guillen, J. Riu, A. D€ uzg€un, and F. X. Rius, Angew. Chem.
31
J. Lee, M. Jo, T. H. Kim, J.-Y. Ahn, D. Lee, S. Kim, and S. Hong, Lab Int. Ed. Engl. 48, 7334 (2009).
39
Chip 11, 52 (2011). R. A. Olowu, O. Arotiba, S. N. Mailu, T. T. Waryo, P. Baker, and E.
32
E. Stern, R. Wagner, F. J. Sigworth, R. Breaker, T. M. Fahmy, and M. A. Iwuoha, Sensors (Switzerland) 10, 9872 (2010).
40
Reed, Nano Lett. 7, 3405 (2007). F. Long, H. Shi, and H. Wang, RSC Adv. 4, 2935 (2014).
33 41
N. O. V. Plank, M. Ishida, and R. Cheung, J. Vac. Sci. Technol. B: J.-J. Zhang, J.-T. Cao, G.-F. Shi, K.-J. Huang, Y.-M. Liu, and Y.-H. Chen,
Microelectron. Nanometer Struct. 23, 3178 (2005). Anal. Methods 6, 6796 (2014).

J. Vac. Sci. Technol. B, Vol. 33, No. 6, Nov/Dec 2015