Author’s Accepted Manuscript

Rapid Detection of Single E. coli Bacteria Using a

Graphene-based Field-Effect Transistor Device

Bhawana Thakur, Guihua Zhou, Jingbo Chang,

Haihui Pu, Bing Jin, Xiaoyu Sui, Xiaochen Yuan,
Ching-Hong Yang, Matthew Magruder, Junhong
Chen
www.elsevier.com/locate/bios

PII: S0956-5663(18)30178-7
DOI: https://doi.org/10.1016/j.bios.2018.03.014
Reference: BIOS10340
To appear in: Biosensors and Bioelectronic
Received date: 1 January 2018
Revised date: 16 February 2018
Accepted date: 6 March 2018
Cite this article as: Bhawana Thakur, Guihua Zhou, Jingbo Chang, Haihui Pu,
Bing Jin, Xiaoyu Sui, Xiaochen Yuan, Ching-Hong Yang, Matthew Magruder
and Junhong Chen, Rapid Detection of Single E. coli Bacteria Using a Graphene-
based Field-Effect Transistor Device, Biosensors and Bioelectronic,
https://doi.org/10.1016/j.bios.2018.03.014
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
 Rapid Detection of Single E. coli Bacteria Using a Graphene-based Field-Effect

Transistor Device

Bhawana Thakura, Guihua Zhoua, Jingbo Changa, Haihui Pu a, Bing Jina, Xiaoyu Suia,

Xiaochen Yuanb, Ching-Hong Yangb, Matthew Magruderc, and Junhong Chen a*

a
Department of Mechanical Engineering, University of Wisconsin-Milwaukee, 3200 North

Cramer Street, Milwaukee, Wisconsin 53211, United States

b
Department of Biological Sciences, University of Wisconsin-Milwaukee, Wisconsin 53211,

United States
c
Milwaukee Metropolitan Sewerage District, Milwaukee, Wisconsin 53211, United States

ABSTRACT

Contamination of surface and drinking water due to the presence of Escherichia coli bacteria

is a major cause of water-borne disease outbreak. To address unmet challenges for practical

pathogen detection in contaminated samples, we report fabrication of thermally reduced

graphene oxide-based field-effect transistor (rGO FET) passivated with an ultrathin layer of

Al2O3 for real-time detection of E. coli bacteria. The sensor could detect a single E. coli cell

within 50 s in a 1µL sample volume. The ultrathin layer of Al2O3 acted as a barrier between

rGO and potential interferents present in the sample. E. coli specific antibodies anchored on

gold nanoparticles acted as probes for selective capture of E. coli. The high density of

negative charge on the surface of E. coli cells strongly modulates the concentration of

majority charge carriers in the rGO monolayer, thereby allowing real-time monitoring of E.

coli concentration in a given sample. With a low detection limit of single cell, the FET sensor

had a linear range of 1-100 CFU in 1 µL volume of sample (i.e., 103 to 105 CFU/ mL). The

biosensor with good selectivity and rapid detection was further successfully demonstrated for

E. coli sensing in river water. The rGO-based FET sensor provides a low cost and label-free

1
approach, and can be mass produced for detection of a broad spectrum of pathogens in water

or other liquid media.

KEYWORDS

Graphene oxide, field-effect transistor, E. coli, biosensor, bacteria detection

1. INTRODUCTION

Contamination of drinking and surface water by Escherichia coli (E. coli) has a profound

effect on public health. The acute toxicity associated with Shiga toxins released by the

enterohemorrhagic strains of E. coli bacteria can cause life-threatening illness amongst the

general masses (Edberg et al. 2000; Pandey et al. 2017). Delay in the treatment after the onset

of symptoms causes rapid deterioration of patients’ health. Hence, real-time monitoring of E.

coli is essential for disease control and ultimately to safeguard public health. In spite of the

fact that gastro intestinal infection is caused by both O157:H7 and non-O157:H7 species of

E. coli, majority of literature reports focus on detection of O157:H7 strain (Lim et al. 2010).

The current methods for detection of E. coli, e.g., colony-counting method, consist of

multiple steps, require trained personnel and give confirmatory results only after 24-48 h

(Bulard et al. 2015). Recent amplification-based molecular diagnosis methods like

polymerase chain reaction (PCR) can reduce the assay time to hours. However, even the

combination of PCR and colony-counting methods is not sensitive enough to detect bacteria

at low concentrations (≪1–100 colony-forming units per mL (CFU/ mL) of sample) (Kang et

al. 2014; Lee et al. 2007). Moreover, to achieve a low detection limit, PCR requires an

additional enrichment step. For practical applications, the detection platform should be

ultrasensitive because even a single cell of bacteria can cause serious illness. The

conventional methods also suffer from poor specificity and high background signals due to

the turbidity and presence of non-target species in real samples. Therefore, the existing

2
methods for E. coli detection not only require expensive equipment and special training but

are also time consuming and less sensitive. To address these limitations, it is important to

develop a sensitive, rapid, and cost-effective sensor platform to detect E. coli in drinking and

surface water.

There are several literature reports based on the combination of various techniques and smart

materials for the detection of E. coli contamination in both food and water. Majority of these

detection techniques are based on optical, electronic and electrochemical methods which

offer rapid response and ease of operation (Acharya et al. 2006; Chen et al. 2017; Labib et al.

2012; Pandey et al. 2017). However, optical methods often use chromogenic or fluorescent

probes and the turbidity of sample matrix affects both sensitivity and detection limit (Miranda

et al. 2011; Wang et al. 2012; Yoo and Lee). In the case of electrochemical detection

methods, requirement of electro-active substrates or electro-active labels in addition to the

background signal associated with various interfering species limit the application for E. coli

detection in real samples (Zhu et al. 2015). In addition, E. coli detection is also reported

based on various methods like real-time mass spectrometry and piezoelectric techniques

(Campbell and Mutharasan 2007; Li et al. 2010). Over the years, field-effect transistors

(FETs) have become a very attractive choice for sensing applications due to their rapid

response, high sensitivity, good reproducibility and real-time monitoring compared with

several other detection techniques (Chang et al. 2013a; Chang et al. 2013b; Chang et al. 2015;

He et al. 2012; Zang et al. 2016; Zhou et al. 2014; Zhu et al. 2015). Usually materials like

carbon nanotubes, graphene, and silicon nanowires are integrated as active channel along

with suitable probe molecules into an FET device for sensing applications (Ahn et al. 2010;

Allen et al. 2007; Liu and Guo 2012). Amongst various channel materials, graphene holds

great promise as transducer in FET-based sensors. Graphene is a low electronic noise 2D

system with low contact resistance, wherein the whole volume is available for interactions

3
with the analyte (Geim and Novoselov 2007; Schedin et al. 2007; Zhang et al. 2005). Along

with desirable properties like high electron mobility, excellent thermal conductivity, and

large surface-to-volume ratio, graphene exhibits high capacitance, and tunable ambipolar

field-effect characteristics (Zhou et al. 2014). This has prompted the application of graphene-

based FET for sensing of plethora of analytes like heavy metal ions, pathogens, proteins, and

nucleic acids (Cai et al. 2014; Chang et al. 2013b; Chang et al. 2015; Martínez et al. 2009;

Ohno et al. 2010; Zhou et al. 2014).

Herein we have developed a thermally-reduced graphene oxide (rGO)-based FET platform

modified with E. coli antibodies (anti- E. coli) for detection of single E. coli bacteria. To

detect more than one strain of E. coli, generic anti- E. coli were used as a probe. The active

channel is thermally-reduced graphene oxide (GO) sheet which is deposited on Au/ SiO2

based electrode surface via chemical modification. Presence of oxygen-containing functional

groups like epoxy, hydroxyl, and carboxyl enables chemical modification of substrates with

GO. GO is a less expensive derivative of graphene that can be easily reduced either by a

chemical method or thermal annealing. The sensing mechanism of graphene-based FETs

relies on change in the electrical conductivity of the rGO channel upon binding with the E.

coli because E. coli has net negative charge around its membrane, which modulates the

conductivity of rGO in FET. To passivate the rGO channel, the FET is modified with few

nanometre thick Al2O3 layer using an Atomic Layer Deposition (ALD) technique. The Al2O3

layer inhibits any direct contact of water or unwanted species with the rGO surface without

affecting its FET characteristics, which improves the reproducibility and stability of the

sensor. This sensor platform is capable of detecting a single E. coli cell. The sensor could

also detect E. coli in river water sample, which proves its potential for practical applications

in point-of-use sensing.

4
2. EXPERIMENTAL METHODS

2.1. Materials

KMnO4, NaNO3, H2SO4, glutathione (GSH), α-ethyl-tryptamine (AET), 1-Ethyl-3-(3-

dimethylaminopropyl)-carbodiimide (EDC), N-Hydroxysuccinimide (NHS),and the blocking

buffer (0.1% Tween 20) were purchased from Sigma-Aldrich. Purified natural graphite was

purchased from SP-1 BayCarbon, MI. Thiol Green Indicator Assay kit was procured from

abcam. Micro-BCA Protein Assay Kit was purchased from ThermoFisher Scientific. Heat

killed Salmonella typhimurium and Streptococcus pneumonia were procured from InvivoGen.

E. coli serotype O/K polyclonal antibody (anti-E. coli) was purchased from ThermoFisher

Scientific. The E. coli (ATCC 25922) strains were grown in LB medium (1% tryptone, 0.5%

yeast extract, and 1% NaCl) at 37°C for 12h.The concentration of saturated culture was

estimated by a plate counting method and UV- vis spectrophotometer. Prior to sensing, the

culture was washed with DI water and centrifuged to remove growth medium. The culture

was resuspended in DI water and diluted to obtain desired concentrations. River water

samples supplied by Milwaukee Metropolitan Sewerage District (MMSD) were used without

any pre-treatment.

2.2. GO Synthesis

The modified Hummer’s method was used for the synthesis of GO (Marcano et al. 2010).

The purified natural graphite was subjected to oxidization in the presence of KMnO4 and

NaNO3 in concentrated H2SO4. The GO dispersion was centrifuged to remove possible

agglomeration materials followed by subsequent washing in DI water. The stable suspension

of GO sheets of required concentration was obtained after ultasonication.

2.3. Device Fabrication

The step wise fabrication is shown in schematic (Figure 1). The Au interdigitated electrodes

with 2 mm length and 50 nm thickness were fabricated using a photolithographic process on

5
a highly doped Si wafer with an upper layer of dry-formed SiO2(thickness of 200 nm).The

finger-width and inter-finger spacing of electrodes was 2 µm. The self-assembly of thiol was

used to anchor GO on the interdigitated Au electrodes. The electrode was modified with AET

(1 mg/mL for 10 min) to give amine-terminated functionality. The amine-terminated

electrodes were immersed in the GO solution for 10 minutes followed by washing with

copious amount of deionised water. The electrodes with self-assembly of GO were then

subjected to thermal annealing in a tube furnace (Lindberg Blue, TF55035A-1) by heating the

device for 10 min at 400 °C in Ar flow of 1 litre/min. A thin layer of Al2O3 (3 nm) was

coated over the rGO-modified electrodes using ALD. To anchor the anti- E. coli, Au

nanoparticles (NPs) were sputter coated on the FET for 3 s using an RF (60 Hz) Emitech

K550x sputter coater apparatus with an Au target (99.999% purity), at an Ar pressure of 0.03

mbar and a working current of 20 mA. 5 µL of GSH solution (10 mM) was dropcasted on

electrodes for 1 h. The carboxylic acid functionalized electrodes were then linked with anti-

E. coli with the help of EDC- NHS chemistry. For this purpose, 5 µL of 1:1 mixture of EDC

and NHS solution (20 mM) was dropcasted on electrodes for 30 min. The electrode with the

reactive ester group was kept for overnight modification with anti-E. coli at 4 ᵒC. The total

loading of anti- E. coli on the sensor was determined using bicinchoninic acid (BCA) protein

assay (Figure S1). Such type of analysis is important to obtain batch-wise quality control for

reproducibility. To avoid non- specific binding, the sensors were finally incubated in a

blocking buffer for 1 h at room temperature followed by washing with the DI water. The

fabricated sensors were stored at 4 ᵒC in dry airtight containers.

6
 rGO Al2O3
(1) (2)

(3)
E.coli
Anti-E.coli AuNPs
(5) (4)

(1) Self assembly of rGO followed by thermal reduction , (2) Deposition of atomic
layer of Al2O3, (3) Deposition of AuNPs, (4) Immobilization of Anti- E. coli, and
(5) E. coli sensing

Figure 1: Schematic depicting various steps involved in the fabrication of anti- E.coli/

AuNPs/ Al2O3/ rGO FET sensor.

2.4. Measurement and Characterization:

Atomic force microscopy (AFM) was recorded by Agilent Technology 5420 AFM with a

cantilever (Nanosensors PPP-NCH). Electrical and transport measurements were performed

on the sensors using a Keithley 4200 semiconductor characterization system. X-ray

photoelectron spectroscopy (XPS) spectra of GO samples were obtained using an HP5950A

ESCA spectrometer with monochromatic Al Kα radiation as X-ray source. With a back-gate

applied to the FET device at room temperature, the source- drain current (ISD) was measured

as a function of the gate voltage (Vgs) and the source-drain voltage (Vds). The gate bias was

varied from −40 to +40 V. The electrical conductivity of the sensor was recorded by

monitoring the change in the drain current (ISD) with a fixed Vds for various concentrations of

E. coli solutions. Hitachi S4800 field-emission scanning electron microscope (SEM) at a 2

kV acceleration voltage was used to characterize the sensors after every fabrication step. The

Infinite 200 PRO (Tecan) plate reader was used for the bicinchoninic acid (BCA) assay and

to determine amount of GSH on sensors.

7
3. RESULTS and DISCUSSIONS

3.1. Sensor Characterization

Due to the facile method of fabrication we could easily control the quality of rGO on our

sensors. Prior to the immobilization of GO on Au electrodes, GO dispersion was subjected to

sonication to obtain monolayers. The weak binding between layers of GO allows easy

separation of GO into monolayers during sonication. Sonication time and power was

optimized to obtain monolayer dispersion with uniform size of GO sheets. GO was self

assembled on Au interdigited electrode with AET as linker molecule. The strong affinity of

thiol group in AET towards Au aids formation of self assembled monolayers (SAM) of AET

on interdigited electrode. The amine end of AET the binds with various functional groups

present in GO monolayers (carboxylic acid, and epoxide) via electrostatic adsorption.

Assembly time of GO solution on AET modified Au electrodes was optimized to control the

coverage area of sensor. Followed by self assembly of GO, the devices were subjected to

thermal annealing under argon atmosphere (400 °C with argon flow of 1 litre/min). Thermal

reduction under argon atmosphere improves the semiconducting property of GO by reducing

the oxygen-containing groups in GO monolayer. Moreover, this process also decreases the

junction barriers between the Au electrodes and rGO. Followed by the thermal reduction, to

passivate rGO, Al2O3 layer (3nm) is deposited over the FET with the help of ALD technique.

To immobilize anti- E. coli (probe molecule) on the FET, we used AuNPs as the anchoring

sites. Owing to the chemical inertness, and facile methods of functionalization, AuNPs are

ideal candidates for immobilization of bio-recognition elements during fabrication of

biosensors. Functionalised AuNPs conjugated with antibodies have been reported for sensing

and imaging (Jazayeri et al. 2016). In our work, we have functionalised AuNPs with GSH

followed by EDC/ NHS binding. The thiol group on GSH binds to AuNPs through

electrostatic interaction and the terminal carboxylic acid group at the other end of GSH

8
molecule is converted to reactive ester by EDC/ NHS reaction. The reactive ester group then

combines with amine groups on anti- E. coli, thereby immobilizing it on FET sensors.

The thickness of deposited GO sheet on electrode (∼1.2 nm) was measured using AFM

(Figure S2). Since the thickness of a single-layer graphene oxide sheet is∼ 1.1 nm, the AFM

measurement suggests that the GO sheet on our sensor is monolayer (Zhou et al.

2014). Figure S2 shows FTIR spectra of GO at 25 ᵒC and post thermal annealing at 400ᵒC.

GO contains various oxygen functionalities which appear in the FTIR spectra (3400 cm-1 for

O-H stretching vibrations, 1720 cm-1 for C=O stretching vibrations, 1600 cm-1for C=C

stretching vibrations, 1220 cm-1 for C-OH stretching vibrations, and 1060 cm-1 for C-O

stretching vibrations) (Choi et al. 2010; Kellici et al. 2014). The spectra of GO after thermal

annealing showed significant decrease in the absorption peak of C=O (1720 cm−1), while

band for C–OH stretching vibrations at 1220cm−1 disappeared completely. This decrease in

oxygen containing functionalities confirmed reduction of GO due to thermal annealing at

400ᵒC under argon atmosphere. The XPS spectra of GO and rGO, which shows increase in

C/O ratio after thermal annealing also confirmed the reduction of GO (Figure S3).

The SEM image of monolayer rGO sheets bridging the gap between the Au interdigitated

electrodes is shown in Figure 2a. The inset of Figure 2a shows the uniform distribution of

Au NPs on the surface of the rGO sheet protected with a 3 nm thick layer of Al2O3. The

electrical properties of rGO FETs were measured at room temperature using the back-gated

FET devices (Figure 2b). For this purpose, the source-drain current (ISD) was measured while

the gate bias (Vgs) was varied from -40 to +40 V. The p-type semiconducting property of rGO

FET is evident from the decrease in conductivity with increasing voltage.

9
 0.48

(a) AuNPs
0.46
(b)
0.44
rGO 100
100nm
nm

ISD(A)
0.42

0.40
5µm
0.38

0.36

10 µm 0.34

0.32
-40 -20 0 20 40

0.76 0.53

0.75 (c) (d)

0.52
0.74
ISD (A)

ISD(A)
0.51
0.73
0.50
0.72

0.71 0.49

0.70 0.48
-40 -20 0 20 40 -40 -20 0 20 40
Vgs (V) Vgs (V)

Figure 2:(a) SEM images of AuNPs/ Al2O3/ rGO FET, (b) ISD−Vgs characteristics of rGO

FET, (c) Al2O3/ rGO FET, and (d) anti- E. coli/ AuNPs/ Al2O3/ rGO FET (Vds = 0.01 V, Vgs

= −40 to 40 V, step = 0.2 V).

The abundance of defects in the rGO has also been demonstrated with the help of Raman

spectrum in the previous reports, which showed that the D band is higher than the G band in

rGO (Ghosh et al. 2012; Zhou et al. 2014). Further, the Al2O3-protected rGO FET showed

bipolar FET characteristics (Figure 2c). This can be attributed to the presence of electrons

(introduced by Al2O3), which now in addition to holes act as charge carriers. A similar type of

phenomenon was reported by Schmid et al (Schmid et al. 2006). Ultrathin layers of Al2O3

possess oxygen (O) vacancies. Due to these vacancies, two Al 3p electrons fail to find the

corresponding empty O 2p conduction band states. In the present case, the rGO layer beneath

it acts as potential electron reservoir. Thus in addition to holes, rGO sheets have added

electrons which contribute to the FET characteristic of the sensor. The FET characteristic of

10
the sensor reverts back to p- type after modification with antibodies. This could be because of

positively charged antibodies which pull the electrons from Al2O3 layer, thereby reducing

their concentration in the rGO sheet (Figure 2d).

We also investigated the effect of size and number of AuNPs on adsorption of linker

molecule on the surface of FET sensor. AuNPs provide binding site for anti- E. coli via linker

molecule i.e., GSH in our study. Thus it is important to immobilize maximum amount of

GSH on the FET sensor. Additional studies were carried out to improve the adsorption

capacity of the AuNPs and the areal density of GSH. A group of control experiments were

conducted for six AuNPs samples with different sputtering parameters (Table S1). The

morphology of AuNPs in each sample was characterized with SEM to measure the size and

number of particle on the sensing surface. Increasing the sputtering time increases the number

of AuNPs and higher sputtering current gives larger particle size. Since the sputtering current

of 25 mA resulted in congregation of particles (sample No.6), we did not extend our study

above this current range. The areal density (=N×πD2) was defined as the indicator for

effective adsorption surface. Each group of sample has 1.5 cm×1.5 cm silicon wafers (n=4

for each group), which was submersed in 1 mL of GSH solution (50 µM). After 1 h, the

silicon wafers were taken out, and the remaining concentration of GSH in the solution was

measured with fluorescence plate reader by using Thiol Green Indicator test (SI). Figure S4

shows the correlation between GSH adsorption and AuNPs size, count and areal density.

From this study it can be concluded that compared to particle size or particle number,

adsorption of GSH has higher dependency on areal density of AuNPs. Therefore, balancing

the size and number of AuNPs to achieve a uniform sensing surface with a maximum areal

density is the key to maximize the probe adsorption. For this purpose, both current and

sputter coating time were optimized to get the desired areal density of AuNPs on the surface

of sensor. Further, the concentration of linkers (GSH, EDC/ NHS) was also optimized to

11
immobilize maximum amount of anti- E. coli on the surface of sensor. The amount of anti- E.

coli was estimated by BCA protein assay (SI). The concentration of GSH was varied from 2

mM to 15 mM with corresponding two fold higher concentrations of EDC/ NHS. The amount

of immobilized anti- E. coli increased sharply for 2 mM to 10 mM of GSH. Increase in

concentration of GSH beyond 10 mM did not have any further effect on the concentration of

immobilized anti- E. coli (Table S2).

3.2. Sensing Performance

The E. coli FET sensor is fabricated on Au electrodes with rGO as the conducting channel. In

the FET, rGO is bridging the source and drain electrodes. Under normal conditions GO is

electrically insulating due to the presence of saturated sp3 bonds, the high density of

electronegative oxygen atoms bonded to carbon, and other defects give that rise to an energy

gap in the electron density of states. After thermal reduction, due to larger sp2 domains rGO

becomes electrically conducting and the FET sensor shows resistance between 50 kΩ to 150

kΩ (Jung et al. 2008; Kumar et al. 2016). The sensing of E .coli is dependent on modulation

of electrical conductivity of rGO in FET sensor. In the absence of rGO or prior to reduction

of GO, the FET behaves like an open circuit and does not respond to addition of the analyte.

We also studied the effect different thickness of Al2O3 layer (1nm, 2nm, 3nm, 4nm) on the

performance of FET sensor. For Al2O3 layer below 3nm, the sensors were quite unstable and

susceptible to oxidation. Owing to the Debye length limitation, for thickness above 3nm the

devices became inert to external environment and do not respond to the addition of analyte

(Wu et al. 2016).

For sensing experiments, the sensor was subjected to various concentrations of E. coli in DI

water (103 CFU/ mL to 109 CFU/ mL).1 µL of sample volume was used for analysis. The

actual number of E. coli detected on the sensor depends on both the concentration of E. coli

and the sample volume used for analysis. Calculation to estimate the number of E. coli cells

12
dropcasted on the sensor surface for various concentrations of sample solutions is shown in

Table S3. Although the lowest concentration of E. coli sample used for sensing is 103 CFU/

mL, based on the calculations the sensor is detecting a single E. coli cell.

3
1.10 10 4
10
5
10
1.05
6
10
ISD (A)

1.00

0.95
9
10
7
10
0.90
8
10
0.85 9
2 X 10

50 100 150 200 250 300 350 400 450

Time(s)
Figure 3: Real-time detection (Vds = 0.1 V) of E.coli (CFU/ mL) in water with the anti- E.

coli/ AuNPs/ Al2O3/ rGObiosensor.

The anti- E. coli anchored on the surface of AuNPs selectively captures E. coli cells. The

capture of E. coli cells by anti- E. coli from the sample is reflected by the change in the

electrical conductivity of the rGO channel, i.e., ISD. Figure 3 shows the real-time monitoring

of E. coli by the sensor. There is a linear decrease in ISD after the addition of E. coli aliquot

from 103 CFU/ mL to 106 CFU/ mL. Alternatively, the ISD increases after addition of 107

CFU/ mL of E. coli solution. The sensor has a rapid response time for the E. coli detection. A

short response time is one of the important advantages of FET sensors. One of the previous

works reported by our group based on rGO FET for detection of Pb2+ ions also had a very

quick response. This was attributed to the rapid diffusion of ions from the liquid drop on the

13
surface of the device to the contact area (Zhou et al. 2014). Although E. coli is very much

larger than ions or molecules, the low volume of sample reduces the diffusion length and

enables rapid analysis of samples. The reported methods for E. coli detection usually

involved minimum 30 minutes of incubation time after multiple sensor fabrication steps

(Hassan et al. 2015; Laczka et al. 2008). Moreover, such methods of analysis are unsuitable

for real-time analysis of samples like flowing water or other samples where continuous

monitoring of water purity is essential. We also carried out control experiments with the

sensor which did not have any probes, i.e., anti- E. coli. The sensor for control experiments

was AuNPs/ Al2O3/ rGO FET. Addition of E. coli did not generate any signal, which proves

that the anti- E. coli FET sensor responds only after specific binding with E. coli (Figure S5).

The sensing mechanism in the present case can be attributed to the synergistic effect between

the Al2O3 layer and the rGO channel, the conductivity of which is modulated upon coming in

contact with negatively-charged E. coli cells. The rGO layer is the potential electron reservoir

for the free electrons since the oxygen vacancies in Al2O3 fail to find the corresponding

empty O 2p conduction band states. In other words, the oxygen vacancies lead to the

dangling bonds in aluminium atoms, which can act as electron donors. The 1-2 µm sized E.

coli, which has rich density of negative charge on its surface, is in close vicinity with the

surface of electron-rich Al2O3 layer. This cloud of negative charge further enhances the

electron donating capability of Al2O3. To restore its insulating property, the oxygen-deficient

domain boundary of Al2O3 behaves like a native electron donor and rGO sheets acts as

recipient (Schmid et al. 2006). The majority carriers (i.e., holes) in the rGO sheet combine

with these electrons. This reduction in the concentration of majority charge carriers decreases

the current flow between source and drain. In addition, the sensing mechanism can be

associated with the gating effect of the E. coli antibodies present on the sensor surface. The

antibodies lose positive charge which is induced by adsorption of negatively-charged E. coli

14
cells. The magnitude of the positive charge on antibodies decreases with the increasing

concentration of E. coli. This decreased positive charge of the antibodies indicates that the

AuNPs are less negatively charged since the antibody itself can work as a dielectric medium.

As a result, the concentration of holes in the rGO sheet from the gate-insulator/semiconductor

interface, i.e., the current flow between source and drain decreases upon adsorption of E. coli.

Above a threshold concentration of 107 CFU/ mL, there are excess of E. coli cells in the

sample solution. Rather than coming in direct contact with the Al2O3 layer, E. coli cells stack

over each other. An additional experiment has been carried out to monitor the effect of E. coli

concentration on conductivity of DI water (Table S4). It was observed that the conductivity

of DI water increases exponentially with increasing concentration of E. coli. The effect of

concentration on the conductivity of DI water is more pronounced after 106 CFU/ mL

concentration of E. coli. The observed trend is due to the accumulation of E coli cells in the

sample solution, which increases the ionic conductivity of the DI water leading to increase in

ISD. This increase in conductivity can be attributed to the surface charge present on the

membrane of E. coli cells. The high density of these negatively charged groups directly

affects the rGO in FET by acting as gate electrolyte that exerts negative potential. This in turn

generates more holes in the rGO channel thereby increasing the ISD of FET sensor. Thus there

are two competing effects wherein one leads to decrease in current (103 CFU/ mL to 106

CFU/ mL) and other causes increase in current (106 CFU/ mL to 109 CFU/ mL). By looking

at the graphs for real time detection of E. coli (Figure 3 and 4) and the values for

conductivity (Table S4), it can be inferred that effect of solution conductivity on FET signal

due to increasing concentration of E. coli is more pronounced after addition of 106 CFU/ mL

of cells. Although, Figure 3 shows significant decrease in ISD after addition of 106 CFU/ mL,

instead of 103 CFU/ mL to 106 CFU/ mL the graph has good linear fitting from 103 CFU/ mL

to 105 CFU/ mL. Thus, effect of conductivity below 105 CFU/ mL on FET sensor, even if any

15
can be neglected. The calibration plot showing linear relation between % response (%R) and

concentration of E. coli is shown in Figure 4. The response is linear from 103 CFU/ mL to

105 CFU/ mL of E. coli concentration. This work has focused on a concentration range of E.

coli below 106 CFU/ mL. Besides, there are some commercial kits which can detect presence

of bacteria, especially of coliforms with 105 CFU/ mL as lowest possible detection limit. It is

more challenging to detect concentration of bacteria below 105 CFU/mL, which is important

for practical applications.

40
30

Y= 2.166 + (2.144E-4)X
35 25

20
%R

30 15

10
25
5

20
%R

3 4 5
10 10 10
[E. coli ]/ (CFU/ mL)
15

10

0
3 4 5 6
10 10 10 10
[E. coli ]/ (CFU/ mL)

Figure 4: The response curve of anti- E. coli/ AuNPs/ Al2O3/ rGO biosensor to E. coli

addition, where % R= [(Resistancefinal-Resistanceinitial)/Resistanceinitial]*100. Inset: Linear

fitting from 103 CFU/ mL to 106 CFU/ mL of E. coli concentration.

The distribution of E. coli cells on the surface of sensors at various concentrations is shown

in Figure 5. The detailed calculation showing number of E. coli dropcasted on the sensor is

16
shown in Table S3. The SEM image of E. coli cells on sensor surface is recorded over a very

small portion of electrode surface. The total area of electrode is 0.24 mm (0.8mm  0.3 mm)

on which E. coli cells are distributed. To make E. coli visible in the image we have focused

on very small area of the electrode and hence the visible number of cells in the images is far

less than the total number of cells dropcasted on sensor. As seen from the SEM images, from

106 CFU/mL onwards the sensor has a dense coverage of E. coli which causes saturation in

the sensor response. Although the lowest concentration of E. coli to which the sensor is

responding is 103 CFU/mL, 1 µL of the sample droplet has a single CFU of E. coli or a single

E. coli cell. It may be noted that since the sample volume used for analysis in our work is

1µL, the sensor is thus capable of detecting a single E. coli cell. Coupling of this platform

with a filtration setup can easily help in detection of single or few E. coli cells in a larger

volume of samples. It is strictly required in drinking water that E. coli must not be detectable

in any 100 mL sample. Thus, for such type of practical applications it is very important for a

sensing platform to detect as low as a single bacterial cell. The rGO-based FET sensor

reported in this work has shown a tremendous potential for this application as it can precisely

detect a single E. coli cell in a given sample. There are very few literature reports where the

sensor was able to detect as low as a single cell (Hahn et al. 2005; Kang et al. 2014; Sepunaru

et al. 2015; Zhao et al. 2004). The performance of these sensors with respect to the linear

range, detection time, and detection limit are summarised in Table S5 in supporting

information.

17
 103 CFU/ mL 104 CFU/ mL 105 CFU/ mL

50 µm 50 µm 50 µm

106 CFU/ mL 107 CFU/ mL 108 CFU/ mL

50 µm 50 µm 50 µm

Figure 5: SEM images showing distribution of E. coli dropcasted on the anti- E. coli/

AuNPs/ Al2O3/ rGO biosensor.

Further, the electrostatic self-assembly of GO on amino-terminated Au electrode allowed

precise control of uniformity of the GO film which in turn contributed to the stability of

device. When preserved at a storage temperature of 4 C, the E. coli FET sensors did not

show any significant deterioration in performance after 14 days of fabrication. These sensors

are designed for one time use. As far as stability of sensor with respect to reusability is

concerned, the very small sensing area of FET sensor is not suitable for regeneration.

However, suitable reagents (for e.g. buffers) can aid the complete regeneration of sensor

surface without affecting the FET performance. The stability of the E. coli FET sensors in

terms of reusability needs further investigation. The sensing experiments were carried out at

25 ᵒC. Increase in temperature will positively affect the conductivity of rGO. Since our sensor

is targeted for room temperature analysis we did not extend our studies for range of

temperature.

18
3.3. Interference Test and Real Sample Analysis

The sensor uses anti- E. coli which makes its highly specific to target i.e., E. coli. River water

can have various species of bacteria. Instead of testing whole array of these species we have

performed interference test with two commonly occurring bacteria- Salmonella typhimurium

and Streptococcus pneumonia, each from gram negative and gram positive species of

bacteria. The commercially available antibodies like anti- E. coli are raised against surface

protein of bacteria. These surface proteins which have site- specific binding sites for

antibodies are present on the outer membrane of bacteria, and are retained even in heat killed-

bacteria (Vanaja et al. 2016). Hence, although we have used dead bacteria for interference

tests, similar results can be expected with live cells. To evaluate the reliability of the sensing

platform, interference tests were performed by monitoring the response of E. coli in the

presence of heat killed Salmonella typhimurium and Streptococcus pneumonia. For this

purpose, the dynamic response of the sensor to a mixture of E. coli (104 CFU/ mL) +heat

killed Salmonella typhimurium (106 cells/ mL) and E. coli (104 CFU/ mL) + heat killed

Streptococcus pneumonia (106 cells/ mL) was monitored. The presence of the bacteria apart

from E. coli did not influence the signal. This proves that the sensor is specific to E. coli

which can be attributed to the E. coli antibodies that selectively capture E. coli cells from the

sample to generate the response (Figure 6).

19
 10

% Response
6

0
ST+ E.coli SP+ E.coli E.coli

Figure 6: Bar chart depicting selectivity of anti- E. coli/ AuNPs/ Al2O3/ rGO biosensor for E.

coli in the mixture of E. coli (104 CFU/ mL) + heat killed Salmonella typhimurium (106 cells/

mL) and E. coli (104 CFU/ mL) + heat killed Streptococcus pneumonia (106 cells/ mL).

Presence of the bacteria apart from E. coli did not have any significant impact on the %

response. *ST: Salmonella typhimurium, SP: Streptococcus pneumonia

The proposed analytical protocol was also applied to determine the E. coli concentration in

river water samples spiked with increasing concentrations of E. coli. Since river water has lot

of potential interfering species (like metal ions, organic matter, etc.), contrary to deionised

water wherein the detection limit was 103 CFU/ mL the sensor was able to detect 104 CFU/

mL of E. coli in real time in river water (Figure S6).

4. CONCLUSION

This study has reported the fabrication and application of anti - E. coli/AuNPs/Al2O3/rGO/

FET sensor for the detection of E. coli. The sensing relies on modulation in the conductivity

of rGO channel of the FET after coming in contact with E. coli. The sensor had a rapid

response and is very selective to E. coli with potential of detecting single E. coli cell in a

large volume of sample when coupled with a suitable filtration setup. These sensors are

20
designed for one time use and the results indicate very good stability of the sensor during the

entire course of measurement of electrical conductivity. The sensor showed good

reproducibility for n=3 measurements. Moreover, the sensor could detect E. coli in a complex

matrix of river water sample, promising its practical applications. The sensor is designed for

one-time use and has the possibility of regeneration by using a suitable regeneration buffer.

These disposable, low cost, and robust sensors can be readily mass produced. This approach

can be utilized for detection of other bacteria by incorporating appropriate probe molecules,

i.e., antibodies during the sensor fabrication. The performance of this new genre of bacteria

sensors is either superior or comparable to most of the highly sensitive bacteria sensors

reported in literature. We are further exploring the potential of the sensor for its ability to

differentiate between dead and viable E. coli cells.

Acknowledgements

This work was supported by the U.S. National Science Foundation (NSF) through an NSF

Industry/University Cooperative Research Centre Grand Challenge Grant and MMSD. River

water samples and lab results were provided by MMSD.

REFERENCES

Acharya, G., Chang, C.-L., Savran, C., 2006. An Optical Biosensor for Rapid and Label-Free Detection
of Cells. Journal of the American Chemical Society 128(12), 3862-3863.
Ahn, J.-H., Choi, S.-J., Han, J.-W., Park, T.J., Lee, S.Y., Choi, Y.-K., 2010. Double-Gate Nanowire Field
Effect Transistor for a Biosensor. Nano Letters 10(8), 2934-2938.
Allen, B.L., Kichambare, P.D., Star, A., 2007. Carbon Nanotube Field-Effect-Transistor-Based
Biosensors. Advanced Materials 19(11), 1439-1451.
Bulard, E., Bouchet-Spinelli, A., Chaud, P., Roget, A., Calemczuk, R., Fort, S., Livache, T., 2015.
Carbohydrates as New Probes for the Identification of Closely Related Escherichia coli Strains Using
Surface Plasmon Resonance Imaging. Analytical Chemistry 87(3), 1804-1811.
Cai, B., Wang, S., Huang, L., Ning, Y., Zhang, Z., Zhang, G.-J., 2014. Ultrasensitive Label-Free Detection
of PNA–DNA Hybridization by Reduced Graphene Oxide Field-Effect Transistor Biosensor. ACS Nano
8(3), 2632-2638.
Campbell, G.A., Mutharasan, R., 2007. A Method of Measuring Escherichia Coli O157:H7 at 1 Cell/mL
in 1 Liter Sample Using Antibody Functionalized Piezoelectric-Excited Millimeter-Sized Cantilever
Sensor. Environmental Science & Technology 41(5), 1668-1674.

21
Chang, J., Mao, S., Zhang, Y., Cui, S., Steeber, D.A., Chen, J., 2013a. Single-walled carbon nanotube
field-effect transistors with graphene oxide passivation for fast, sensitive, and selective
proteindetection. Biosensors and Bioelectronics 42, 186-192.
Chang, J., Mao, S., Zhang, Y., Cui, S., Zhou, G., Wu, X., Yang, C.-H., Chen, J., 2013b. Ultrasonic-assisted
self-assembly of monolayer graphene oxide for rapid detection of Escherichia coli bacteria.
Nanoscale 5(9), 3620-3626.
Chang, J., Zhou, G., Gao, X., Mao, S., Cui, S., Ocola, L.E., Yuan, C., Chen, J., 2015. Real-time detection
of mercury ions in water using a reduced graphene oxide/DNA field-effect transistor with assistance
of a passivation layer. Sensing and Bio-Sensing Research 5, 97-104.
Chen, J., Andler, S.M., Goddard, J.M., Nugen, S.R., Rotello, V.M., 2017. Integrating recognition
elements with nanomaterials for bacteria sensing. Chemical Society Reviews 46(5), 1272-1283.
Choi, E.-Y., Han, T.H., Hong, J., Kim, J.E., Lee, S.H., Kim, H.W., Kim, S.O., 2010. Noncovalent
functionalization of graphene with end-functional polymers. Journal of Materials Chemistry 20(10),
1907-1912.
Edberg, S.C., Rice, E.W., Karlin, R.J., Allen, M.J., 2000. Escherichia coli: the best biological drinking
water indicator for public health protection. Journal of Applied Microbiology 88(S1), 106S-116S.
Geim, A.K., Novoselov, K.S., 2007. The rise of graphene. Nat Mater 6(3), 183-191.
Ghosh, T., Biswas, C., Oh, J., Arabale, G., Hwang, T., Luong, N.D., Jin, M., Lee, Y.H., Nam, J.-D., 2012.
Solution-Processed Graphite Membrane from Reassembled Graphene Oxide. Chemistry of Materials
24(3), 594-599.
Hahn, M.A., Tabb, J.S., Krauss, T.D., 2005. Detection of Single Bacterial Pathogens with
Semiconductor Quantum Dots. Analytical Chemistry 77(15), 4861-4869.
Hassan, A.-R.H.A.-A., de la Escosura-Muñiz, A., Merkoçi, A., 2015. Highly sensitive and rapid
determination of Escherichia coli O157:H7 in minced beef and water using electrocatalytic gold
nanoparticle tags. Biosensors and Bioelectronics 67, 511-515.
He, Q., Wu, S., Yin, Z., Zhang, H., 2012. Graphene-based electronic sensors. Chemical Science 3(6),
1764-1772.
Jazayeri, M.H., Amani, H., Pourfatollah, A.A., Pazoki-Toroudi, H., Sedighimoghaddam, B., 2016.
Various methods of gold nanoparticles (GNPs) conjugation to antibodies. Sensing and Bio-Sensing
Research 9(Supplement C), 17-22.
Jung, I., Dikin, D.A., Piner, R.D., Ruoff, R.S., 2008. Tunable Electrical Conductivity of Individual
Graphene Oxide Sheets Reduced at “Low” Temperatures. Nano Letters 8(12), 4283-4287.
Kang, D.-K., Ali, M.M., Zhang, K., Huang, S.S., Peterson, E., Digman, M.A., Gratton, E., Zhao, W., 2014.
Rapid detection of single bacteria in unprocessed blood using Integrated Comprehensive Droplet
Digital Detection 5, 5427.
Kellici, S., Acord, J., Ball, J., Reehal, H.S., Morgan, D., Saha, B., 2014. A single rapid route for the
synthesis of reduced graphene oxide with antibacterial activities. RSC Advances 4(29), 14858-14861.
Kumar, P.V., Bardhan, N.M., Chen, G.-Y., Li, Z., Belcher, A.M., Grossman, J.C., 2016. New insights into
the thermal reduction of graphene oxide: Impact of oxygen clustering. Carbon 100(Supplement C),
90-98.
Labib, M., Zamay, A.S., Kolovskaya, O.S., Reshetneva, I.T., Zamay, G.S., Kibbee, R.J., Sattar, S.A.,
Zamay, T.N., Berezovski, M.V., 2012. Aptamer-Based Viability Impedimetric Sensor for Bacteria.
Analytical Chemistry 84(21), 8966-8969.
Laczka, O., Baldrich, E., Muñoz, F.X., del Campo, F.J., 2008. Detection of Escherichia coli and
Salmonella typhimurium Using Interdigitated Microelectrode Capacitive Immunosensors: The
Importance of Transducer Geometry. Analytical Chemistry 80(19), 7239-7247.
Lee, A., Mirrett, S., Reller, L.B., Weinstein, M.P., 2007. Detection of Bloodstream Infections in Adults:
How Many Blood Cultures Are Needed? Journal of Clinical Microbiology 45(11), 3546-3548.
Li, F., Zhao, Q., Wang, C., Lu, X., Li, X.-F., Le, X.C., 2010. Detection of Escherichia coli O157:H7 Using
Gold Nanoparticle Labeling and Inductively Coupled Plasma Mass Spectrometry. Analytical Chemistry
82(8), 3399-3403.

22
Lim, J.Y., Yoon, J.W., Hovde, C.J., 2010. A Brief Overview of Escherichia coli O157:H7 and Its Plasmid
O157. Journal of microbiology and biotechnology 20(1), 5-14.
Liu, S., Guo, X., 2012. Carbon nanomaterials field-effect-transistor-based biosensors. NPG Asia Mater
4, e23.
Marcano, D.C., Kosynkin, D.V., Berlin, J.M., Sinitskii, A., Sun, Z., Slesarev, A., Alemany, L.B., Lu, W.,
Tour, J.M., 2010. Improved Synthesis of Graphene Oxide. ACS Nano 4(8), 4806-4814.
Martínez, M.T., Tseng, Y.-C., Ormategui, N., Loinaz, I., Eritja, R., Bokor, J., 2009. Label-Free DNA
Biosensors Based on Functionalized Carbon Nanotube Field Effect Transistors. Nano Letters 9(2),
530-536.
Miranda, O.R., Li, X., Garcia-Gonzalez, L., Zhu, Z.-J., Yan, B., Bunz, U.H.F., Rotello, V.M., 2011.
Colorimetric Bacteria Sensing Using a Supramolecular Enzyme–Nanoparticle Biosensor. Journal of
the American Chemical Society 133(25), 9650-9653.
Ohno, Y., Maehashi, K., Matsumoto, K., 2010. Label-Free Biosensors Based on Aptamer-Modified
Graphene Field-Effect Transistors. Journal of the American Chemical Society 132(51), 18012-18013.
Pandey, A., Gurbuz, Y., Ozguz, V., Niazi, J.H., Qureshi, A., 2017. Graphene-interfaced electrical
biosensor for label-free and sensitive detection of foodborne pathogenic E. coli O157:H7. Biosensors
and Bioelectronics 91, 225-231.
Schedin, F., Geim, A.K., Morozov, S.V., Hill, E.W., Blake, P., Katsnelson, M.I., Novoselov, K.S., 2007.
Detection of individual gas molecules adsorbed on graphene. Nat Mater 6(9), 652-655.
Schmid, M., Shishkin, M., Kresse, G., Napetschnig, E., Varga, P., Kulawik, M., Nilius, N., Rust, H.P.,
Freund, H.J., 2006. Oxygen-Deficient Line Defects in an Ultrathin Aluminum Oxide Film. Physical
Review Letters 97(4), 046101.
Sepunaru, L., Tschulik, K., Batchelor-McAuley, C., Gavish, R., Compton, R.G., 2015. Electrochemical
detection of single E. coli bacteria labeled with silver nanoparticles. Biomaterials Science 3(6), 816-
820.
Vanaja, Sivapriya K., Russo, Ashley J., Behl, B., Banerjee, I., Yankova, M., Deshmukh, Sachin D.,
Rathinam, Vijay A.K., 2016. Bacterial Outer Membrane Vesicles Mediate Cytosolic Localization of LPS
and Caspase-11 Activation. Cell 165(5), 1106-1119.
Wang, Y., Knoll, W., Dostalek, J., 2012. Bacterial Pathogen Surface Plasmon Resonance Biosensor
Advanced by Long Range Surface Plasmons and Magnetic Nanoparticle Assays. Analytical Chemistry
84(19), 8345-8350.
Wu, G., Meyyappan, M., Lai, K.W.C., 2016. Graphene field-effect transistors-based biosensors for
Escherichia coli detection. 2016 IEEE 16th International Conference on Nanotechnology (IEEE-
NANO), pp. 22-25.
Yoo, S.M., Lee, S.Y., Optical Biosensors for the Detection of Pathogenic Microorganisms. Trends in
Biotechnology 34(1), 7-25.
Zang, Y., Huang, D., Di, C.-a., Zhu, D., 2016. Device Engineered Organic Transistors for Flexible
Sensing Applications. Advanced Materials, n/a-n/a.
Zhang, Y., Tan, Y.-W., Stormer, H.L., Kim, P., 2005. Experimental observation of the quantum Hall
effect and Berry's phase in graphene. Nature 438(7065), 201-204.
Zhao, X., Hilliard, L.R., Mechery, S.J., Wang, Y., Bagwe, R.P., Jin, S., Tan, W., 2004. A rapid bioassay for
single bacterial cell quantitation using bioconjugated nanoparticles. Proceedings of the National
Academy of Sciences of the United States of America 101(42), 15027-15032.
Zhou, G., Chang, J., Cui, S., Pu, H., Wen, Z., Chen, J., 2014. Real-Time, Selective Detection of Pb2+ in
Water Using a Reduced Graphene Oxide/Gold Nanoparticle Field-Effect Transistor Device. ACS
Applied Materials & Interfaces 6(21), 19235-19241.
Zhu, C., Yang, G., Li, H., Du, D., Lin, Y., 2015. Electrochemical Sensors and Biosensors Based on
Nanomaterials and Nanostructures. Analytical Chemistry 87(1), 230-249.

23
Highlights

 A graphene-based FET is demonstrated for real-time detection of E. coli.

 The device is passivated with ultrathin Al2O3 for improved performance.
 The negative charge on E. coli surface modulates the FET channel conductivity.
 The biosensor could detect a single E. coli within 50 s in 1 µL sample.
 The biosensor was successfully demonstrated for E. coli detection in river water.

24