biosensors
Article
Electronic Biosensing with Functionalized rGO FETs
Ciril Reiner-Rozman 1,2,† , Caroline Kotlowski 1,2,† and Wolfgang Knoll 1,2, *
1
2
*
†
Center for Electrochemical Surface Technology (CEST), Wiener Neustadt 2700, Austria;
rozman.ciril.fl@ait.ac.at (C.R.-R.); caroline.kotlowski@boku.ac.at (C.K.)
AIT Austrian Institute of Technology, Wien 1220, Austria
Correspondence: wolfgang.knoll@ait.ac.at; Tel.: +43-50-550-4002
These authors contributed equally to this work.
Academic Editors: Mark A. Reed and Mathias Wipf
Received: 28 February 2016; Accepted: 31 March 2016; Published: 22 April 2016
Abstract: In the following we give a short summary of examples for biosensor concepts in areas
in which reduced graphene oxide-based electronic devices can be developed into new classes of
biosensors, which are highly sensitive, label-free, disposable and cheap, with electronic signals
that are easy to analyze and interpret, suitable for multiplexed operation and for remote control,
compatible with NFC technology, etc., and in many cases a clear and promising alternative to optical
sensors. The presented areas concern sensing challenges in medical diagnostics with an example for
detecting general antibody-antigen interactions, for the monitoring of toxins and pathogens in food
and feed stuff, exemplified by the detection of aflatoxins, and the area of smell sensors, which are
certainly the most exciting development as there are very few existing examples in which the typically
small and hydrophobic odorant molecules can be detected by other means. The example given here
concerns the recording of a honey flavor (and a cancer marker for neuroblastoma), homovanillic acid,
by the odorant binding protein OBP 14 from the honey bee, immobilized on the reduced graphene
oxide gate of an FET sensor.
Keywords: reduced graphene oxide (rGO graphene); FET; liquid-gate; biosensing; receptor
immobilization; antigen-antibody interaction; food toxins; aflatoxins; odorant-binding proteins;
olfaction; smell sensor
1. Introduction
The detection and quantitative monitoring of biomolecules in air, in water, in complex liquids
like bodily fluids or similar is still a challenge, both from a fundamental point of view as well as in
the context of practical applications. Whether it is the need for sensors in general air management
scenarios—e.g., to detect air pollutants (including explosives), to sense crop disease markers, in food
quality management for the identification of toxins, or in medical applications like breath analysis—Or
for the detection of certain markers in blood, plasma, saliva, urine, wound liquids, etc., that report a
patient’s health and/or disease status, in all cases we are dealing with three major problems; (i) the
sensitivity of the technical device for the quantification of a particular molecule of interest; (ii) the
selectivity needed to differentiate between similar molecules; and (iii) the suppression of non-specific
interactions with and binding to the sensor surface by other components in the sample volume,
typically in excess to the actually targeted molecule.
Numerous concepts and formats for sensors based on optical transduction principles have been
reported in the literature, some of which even made it to the market place and are commercialized [1].
Not so widespread are sensors based on electrical, electrochemical or electronic transduction
principles [2]. The latter category, in particular, is the youngest member of the huge family of
current-based bio-sensors; typical examples are CMOS-transistor-based devices that have been used to
Biosensors 2016, 6, 17; doi:10.3390/bios6020017
www.mdpi.com/journal/biosensors
Biosensors 2016, 6, 17
2 of 12
monitor action potentials of excited neurons [3] or cardiomyocytes [4] grown on the gate surface of the
Biosensors 2016, 6, 17
2 of 11
device, Si-nanowire based transistor read-out of protein and DNA binding [5], or the promising use of
organic
(“plastic”)
electronics
the development
and, hence,
sensors
[6,7]. The
to monitor
action
potentials for
of excited
neurons [3] of
or cheap
cardiomyocytes
[4] disposable
grown on the
gate surface
mostofrecent
development
in based
this area,
the use
of theof
novel
carbon
materials,
i.e.,
nanotubes or
the device,
Si-nanowire
transistor
read-out
protein
and DNA
binding
[5],carbon
or the promising
graphene,
opens a(“plastic”)
totally new
but very
range of
ofcheap
options
for
biosensors
based
on electronic
use of organic
electronics
forpromising
the development
and,
hence,
disposable
sensors
[6,7].
The most
this area, the
use of the
novel carbon materials, i.e., carbon nanotubes
devices
maderecent
fromdevelopment
these highlyininteresting
materials
[8].
or
a totallywe
new
butreport
very promising
range
of options
on
Ingraphene,
this shortopens
summary,
will
some of our
own
efforts for
in biosensors
designing,based
fabricating,
electronic
devices
made
from
these
highly
interesting
materials
[8].
bio-functionalizing and characterizing biosensors based on reduced graphene oxide as the gate
thiswill
short
summary,
we will
report of
some
our preparation
own efforts protocols,
in designing,
material.InWe
give
a very brief
overview
the of
basic
andfabricating,
describe the
bio-functionalizing and characterizing biosensors based on reduced graphene oxide as the gate
electronic performance of the devices as transistors [9]. The practical examples given are (i) the
material. We will give a very brief overview of the basic preparation protocols, and describe the
detection of a standard biomolecule, bovine serum albumin (BSA), binding from solution to its
electronic performance of the devices as transistors [9]. The practical examples given are (i) the
antibody, immobilized on the sensor surface; (ii) the detection and quantification of food toxins
detection of a standard biomolecule, bovine serum albumin (BSA), binding from solution to its
(mycotoxins)
binding to their
antibodies;
and (iii)
of an
odorant molecule
antibody, immobilized
on the
sensor surface;
(ii)the
themonitoring
detection and
quantification
of food(dissolved
toxins
in an(mycotoxins)
aqueous solution)
to
the
surface-bound
odorant
binding
protein
from
an
insect,
the
honey
binding to their antibodies; and (iii) the monitoring of an odorant molecule (dissolved bee
Apisinmellifera
[10].
an aqueous solution) to the surface-bound odorant binding protein from an insect, the honey bee
Apis mellifera [10].
2. Preparation and General Electronic Performance of rGO-FET Biosensors
2. Preparation and General Electronic Performance of rGO-FET Biosensors
Reduced graphene oxide field-effect transistor (rGO-FET) devices were fabricated with a
Reducedwidth
graphene
field-effect
(rGO-FET)
devices
fabricated
with a typical
typical channel
of oxide
40 µm.
Figure transistor
1A briefly
summarizes
thewere
individual
preparation
steps
channel
width
of
40
µm.
Figure
1A
briefly
summarizes
the
individual
preparation
steps
for
the
for the devices [10]. Silicon wafers with a 300 nm oxide layer were chosen as base substrates.
devices
[10]. Silicon
wafers awith
a 300 nm
oxide
layer were
chosenand
as base
They
They
were cleaned
following
standard
RCA
cleaning
procedure
thensubstrates.
submerged
in awere
1%–2%
cleaned following a standard RCA solution
cleaning in
procedure
and
in aa monolayer
1%–2%
3-amino-propyl-triethoxy-silane-(APTES)
ethanol for
1 h,then
with submerged
APTES forming
3-amino-propyl-triethoxy-silane-(APTES)
in ethanol
forgraphene
1 h, with
APTES forming a
on the
substrate that increases the adhesion solution
of graphene
(reduced
oxide).
monolayer on the substrate that increases the adhesion of graphene (reduced graphene oxide).
Graphene oxide sheets were prepared using a variation of Hummers method [11,12]. The obtained
Graphene oxide sheets were prepared using a variation of Hummers method [11,12]. The
graphene oxide flakes were deposited onto Si-wafers via spin coating. A scanning electron microscopic
obtained graphene oxide flakes were deposited onto Si-wafers via spin coating. A scanning electron
image
of the GO flakes on the substrate before reduction of rGO and before the evaporation of the
microscopic image of the GO flakes on the substrate before reduction of rGO and before the
Au electrodes
for
are for
shown
in Figureare
1B.shown
After in
rinsing
ethanol,
thewith
substrates
evaporation
ofthe
the transistor
Au electrodes
the transistor
Figurewith
1B. After
rinsing
ethanol,were
˝
heated
120 C for
two
hourstoand
cooled
to room
then treated
the to
substrates
were
heated
120then
°C for
two hours
andtemperature.
then cooled toThey
roomwere
temperature.
Theyovernight
were
in hermetically
glass
dishes with
hydrazine
at dishes
70 ˝ C in
order
to reduce
the°Cgraphene
then treatedsealed
overnight
inpetri
hermetically
sealed
glass petri
with
hydrazine
at 70
in order oxide,
to
2 -hybrid bonds.
2
forming
the
graphene
structure
consisting
of
sp
reduce the graphene oxide, forming the graphene structure consisting of sp -hybrid bonds.
Figure 1. (A) Schematic illustration of the individual fabrication steps of the graphene biosensor
Figure
1. (A) Schematic illustration of the individual fabrication steps of the graphene biosensor
device: After the reduction of the GO flakes to rGO, source and drain Au electrodes were evaporated
device: After the reduction of the GO flakes to rGO, source and drain Au electrodes were evaporated
(with a thin layer of Cr as an adhesion promotor), then coated (via self-assembly) by a linker, PBSE,
(with a thin layer of Cr as an adhesion promotor), then coated (via self-assembly) by a linker, PBSE,
and finally functionalized by the attachment of antibodies or by odorant binding proteins, here OBP
and finally functionalized by the attachment of antibodies or by odorant binding proteins, here OBP 14
14 from the honey bee. (B) Scanning electron microscopic image of the GO flakes on the chip substrate
from(before
the honey
bee; (B) Scanning electron microscopic image of the GO flakes on the chip substrate
reduction to rGO and before being coated with the electrodes).
(before reduction to rGO and before being coated with the electrodes).
For the attachment of the antibodies (here, antibodies against bovine serum albumin, BSA, and
For the attachment of the antibodies (here, antibodies against bovine serum albumin, BSA, and
aflatoxin B1, respectively) to the sensing area, the graphene surface was chemically modified by a
aflatoxin
B1, respectively)
to the sensing
area,
the graphene
modified
bi-functional
linker, 1-pyrenebutanoic
acid
succinimidyl
ester surface
(PBSE). was
One chemically
end the linker
firmly by
a bi-functional
linker,
1-pyrenebutanoic
succinimidyl
(PBSE).
Onewhile
end the
theother
linker
firmly
attaches to the
graphene
surface through acid
π-π interactions
withester
the pyrene
group
hand
attaches
to
the
graphene
surface
through
π-π
interactions
with
the
pyrene
group
while
the
other
hand
covalently reacts with one of the amino group of the protein to form an amide bond. All reagents
covalently
reacts
with
one
of
the
amino
group
of
the
protein
to
form
an
amide
bond.
All
reagents
were purchased from Sigma-Aldrich unless otherwise indicated and used without further purifications.were
purchased from Sigma-Aldrich unless otherwise indicated and used without further purifications.
Biosensors 2016, 6, 17
Biosensors 2016, 6, 17
3 of 11
3 of 12
For the detection of different odorants, the biosensors were functionalized with an odorant
Biosensors
2016, 14
6, 17(OBP 14) from the honey Bee, Apis mellifera. The protein was expressed in
3 of 11
binding
For protein
the detection
of different odorants, the biosensors were functionalized with an bacterial
odorant
systems
using
established
protocols
[13,14].
Purification
of
the
proteins
was
accomplished
using a
binding protein
14
(OBP
14)
from
the
honey
Bee,
Apis
mellifera.
The
protein
was
expressed
in
bacterial
For the detection of different odorants, the biosensors were functionalized with an odorant
combination
ofestablished
conventional
chromatographic
techniquesoffollowed
by awas
finalaccomplished
gel filtration using
step on
systems
using
Purification
a
binding
protein
14 (OBP protocols
14) from the[13,14].
honey Bee,
Apis mellifera.the
Theproteins
protein was
expressed in bacterial
Superose-12
(GE-Healthcare)
as
previously
described
in
standard
protocols
[14].
combination
of conventional
chromatographic
techniques
followed
bywas
a final
gel filtration
systems using
established protocols
[13,14]. Purification
of the
proteins
accomplished
usingstep
a on
The chip
was
mounted
into a flow
celltechniques
for ininsitu
real-time
measurements
combination
of then
conventional
followed
bykinetic
a final[14].
gel filtration step(in
onorder
Superose-12
(GE-Healthcare)
as chromatographic
previously
described
standard
protocols
to quantify
the
association,
k
on
,
and
dissociation
rate
constants,
k
off, respectively)
as
well
for
Superose-12
(GE-Healthcare)
as
previously
described
in
standard
protocols
[14].
The chip was then mounted into a flow cell for in situ real-time kinetic measurements (in as
order
titration
experiments
for
the
determination
of
affinity,
K
A, and
dissociation
constants,
K
d,
The
chip
was
then
mounted
into
a
flow
cell
for
in
situ
real-time
kinetic
measurements
(in
order
to quantify the association, kon , and dissociation rate constants, koff , respectively) as well as for
to
quantify
the
association,
k
on
,
and
dissociation
rate
constants,
k
off
,
respectively)
as
well
as
for
respectively.
An artist’s
sketch
of the FET
OBP interacting
with the
titration
experiments
for the
determination
of setup
affinity,and
KA ,the
andimmobilized
dissociation constants,
Kd , respectively.
titration
experiments
for the
determination
of whole
affinity,
KAcell,
, and
dissociation
constants,
Kd,
odorant
ligand,
together
with
a
photograph
of
the
flow
are
given
in
Figure
2.
An artist’s sketch of the FET setup and the immobilized OBP interacting with the odorant ligand,
respectively. An artist’s sketch of the FET setup and the immobilized OBP interacting with the
odorant ligand, together with a photograph of the whole flow cell, are given in Figure 2.
together with a photograph of the whole flow cell, are given in Figure 2.
Figure 2. Schematics of the set-up and the device configuration consisting of a liquid gate electrode,
Figure
2. Schematics
Schematics of
of the
the set-up
set-up and
and the
the device
device configuration
configuration consisting of a liquid gate electrode,
Figure
2.
source and drain electrodes, evaporated onto the reduced graphene oxide flakes on a solid Si/SiO2
source
and
drain
electrodes,
evaporated
onto
the
reducedgraphene
grapheneoxide
oxideflakes
flakeson
onaasolid
solidSi/SiO
Si/SiO22
source and drain electrodes, evaporated onto the reduced
substrate, functionalized by OBPs that are covalently coupled via a linker molecule. In the lower left
substrate, functionalized
functionalized by OBPs
OBPs that
that are
are covalently
covalently coupled
coupled via
via aa linker
linker molecule.
molecule. In
In the
the lower
lower left
left
substrate,
corner is a photographby
of the mounted
flow cell.
corner is
is aa photograph
photograph of
of the
the mounted
mounted flow
flowcell.
cell.
corner
Figure 3. Source-drain current ISD-vs-gate voltage VG characteristics of the OBP14 functionalized rGO
FET-biosensor device at different Eugenol concentrations (as indicated) flowing through the cell.
Figure
-vs-gate voltage
characteristics of
of the
the OBP14
OBP14 functionalized
functionalized rGO
Figure 3.
3. Source-drain
Source-drain current
currentIISD
SD-vs-gate
voltage V
VG characteristics
rGO
The electrical properties of the FET devices were tested as described before [15]. Electrical
FET-biosensor
device
at
different
Eugenol
concentrations
(as
indicated)
flowing
through
the
cell.
FET-biosensor
device
at
different
Eugenol
concentrations
(as
indicated)
flowing
through
the
cell.
measurements were performed using a Keithley 4200 semiconductor characterization system. An
Ag/AgCl reference electrode (Flexref, World Precision Instruments) was used to operate the FET
The
The electrical
electrical properties
properties of
of the
the FET
FET devices
devices were
were tested
tested as
asdescribed
describedbefore
before[15].
[15]. Electrical
Electrical
measurements
were
performed
using
a
Keithley
4200
semiconductor
characterization
system.
measurements were performed using a Keithley 4200 semiconductor characterization system.
An
Ag/AgCl reference electrode (Flexref, World Precision Instruments) was used to operate the FET
Biosensors 2016, 6, 17
Biosensors 2016, 6, 17
4 of 12
4 of 11
An Ag/AgCl reference electrode (Flexref, World Precision Instruments) was used to operate the
FET
device
a liquid
gate configuration
with a source-drain
constant source-drain
of The
VSDcathodic
= 50 mV.
device
in a in
liquid
gate configuration
with a constant
bias of VSD =bias
50 mV.
The
cathodic
ofG the
ISD
-vs-V
Figure 3)
the
hole
mobility
branches
of branches
the ISD-vs-V
scans
(cf.
Figure
3) are (cf.
dominated
byare
the dominated
hole mobilityby
[9].
The
Dirac
voltage[9].
G scans
The
voltage
at Recording
ca. VG = 0.4
V. Recording
-vs-V
scans
at
different
bulk
solution
is Dirac
seen at
ca. VGis=seen
0.4 V.
these
ISD-vs-VG these
scans Iat
different
bulk
solution
conditions,
in
SD
G
particularin
inparticular
aqueous solutions
of different
concentrations,
results in a distinct
the
conditions,
in aqueous
solutionsligand
of different
ligand concentrations,
resultsshift
in aof
distinct
cathodic
which
we attribute
to attribute
a slight modification
of the dipolarof
layer
the OBP
shift
of the branch
cathodic
branch
which we
to a slight modification
thewith
dipolar
layerprotein
with the
monolayer
immobilized
on the graphene
surfacegate
uponsurface
partialupon
binding
of the
ligandsoftothe
theligands
free
OBP
protein monolayer
immobilized
on thegate
graphene
partial
binding
binding
sites
in
the
OBP
acting
as
receptors.
to the free binding sites in the OBP acting as receptors.
Antigen–AntibodyInteraction
Interaction and the
3. 3.Antigen–Antibody
the Limit
Limitof
ofDetection
Detection
The
first
of these
these rGO-FETs
rGO-FETsasasbiosensors
biosensorsconcerns
concerns
The
firstexample
examplethat
thatwe
wedescribe
describe for
for the
the use
use of
thethe
“classical”
system,
bovine serum
serum albumin,
albumin,BSA,
BSA,asasantigen
antigen
“classical”
system,i.e.,
i.e.,the
thebinding
binding of
of the
the protein
protein bovine
to to
its its
FET-immobilized
antibody.
The
global
analysis
protocol,
i.e.,
the
stepwise
increase
of
the
bulk
analyte
FET-immobilized antibody. The global analysis protocol, i.e., the stepwise increase of the bulk analyte
concentration
while
time-dependentchange
changeofofthe
the
source-drain
current,
concentration
whilesimultaneously
simultaneouslyrecoding
recoding the time-dependent
source-drain
current,
∆ISD
, is, is
given
inin
Figure
4. 4.
One
can
seesee
that
thethe
current
decreased
each
time
thethe
bulk
concentration
was
ΔISD
given
Figure
One
can
that
current
decreased
each
time
bulk
concentration
100 nM,
it gradually
reached
a saturation
level
was increased
from
an initial
concentration
c0 =nM,
increased
from an
initial
concentration
c0 = 100
untiluntil
it gradually
reached
a saturation
level
for a
forconcentration
a bulk concentration
c0 = 25
µM.
The current
decrease
following
each concentration
bulk
near c0 near
= 25 µM.
The
current
decrease
following
each concentration
changechange
contains
contains
kinetic information
the red
fit the
current
traces)
byexponential
a single
kinetic
information
(cf. the red(cf.curves
thatcurves
fit thethat
current
traces)
occurs
by occurs
a single
exponential
a time
constant,
k, that is dependent
concentration
dependent
(increases
with bulk
giving
a time giving
constant,
k, that
is concentration
(increases
with
bulk concentration).
concentration).
This
is
in
agreement
with
the
Langmuir
model
for
this
1:1
complex
between
the
This is in agreement with the Langmuir model for this 1:1 complex between the analyte (the antigen)
analyte
(the
antigen)
from
solution
and
the
surface–immobilized
receptor
(the
antibody).
Upon
from solution and the surface–immobilized receptor (the antibody). Upon rinsing with pure buffer,
with
pure buffer,
current
returns level
to its original
level with a single
exponential
(cf.
therinsing
current
returns
to its the
original
baseline
with a baseline
single exponential
(cf. the
blue fit curve)
the blue fit curve) indicating the full reversibility of the binding event (a prerequisite for any analysis
indicating
the full reversibility of the binding event (a prerequisite for any analysis according
according the Langmuir model). The fit to this dissociation process results in a quantitative measure
the Langmuir model). The fit to this dissociation process results in a quantitative measure of the
of the dissociation rate constant, koff. From the slope of the plot of all the rate constants determined
dissociation rate constant, koff . From the slope of the plot of all the rate constants determined as a
as a function of the bulk concentration, one obtains the association rate constant, kon, which together
function of the bulk concentration, one obtains the association rate constant, kon , which together with
with koff gives the affinity constant, KA. This has been documented recently and reported to result in
koff gives the affinity constant, KA . This has been documented recently and
reported to result in an
an affinity constant for the binding of BSA to this antibody of KA = 1.6 × 105 M−1 [9], corresponding to
5
´
1
affinity
constantconstant
for the binding
of BSA
a dissociation
of Kd = 6.2
µM. to this antibody of KA = 1.6 ˆ 10 M [9], corresponding to a
dissociation constant of Kd = 6.2 µM.
Figure
4. 4.
Global
to surface-immobilized
surface-immobilizedanti-BSA
anti-BSA
antibody;
Figure
Globalanalysis
analysisofofBSA
BSAbinding
binding from
from solution
solution to
antibody;
solutions
with
concentrations
from
100
nM
to
(near)
saturation
at
25
µM
(as
indicated
in
red)
were
solutions with concentrations from 100 nM to (near) saturation at 25 µM (as indicated in red) were
rinsed
through
the
flow
cell.
The
first
few
minutes
show
a
drift
due
to
the
graphene
charging
behavior,
rinsed through the flow cell. The first few minutes show a drift due to the graphene charging
which
stabilizes
after
7 min.after 7 min.
behavior,
which
stabilizes
Figure
Figure 4,4, focusing
focusinghere
hereononthe
thetitration
titration
Figure5 5summarizes
summarizesthe
the results
results taken
taken from Figure
of of
thethe
equilibrium
surface
coverage
(i.e.,(i.e.,
the new
currentcurrent
that one
reads
the after
new equilibrium
equilibrium
surface
coverage
the source-drain
new source-drain
that
oneafter
reads
the new
has beenas
established)
function
of the bulk
analyte concentration
in solution,
i.e., the
hasequilibrium
been established)
a functionasofathe
bulk analyte
concentration
in solution,
i.e., the Langmuir
Langmuir
By surface
plottingcoverage,
the surface
Θ, i.e., the source-drain
currentconcentration,
at a given
isotherm.
Byisotherm.
plotting the
Θ,coverage,
i.e., the source-drain
current at a given
Biosensors 2016, 6, 17
5 of 12
Biosensors 2016, 6, 17
5 of 11
Biosensors 2016, 6, 17
5 of 11
concentration,
by drain
ISD (c∞current
), the source
drain
infinite bulk one
analyte
ISD (c
ISD(c(c0),8 ),divided
the source
at infinite
bulkcurrent
analyteatconcentration,
obtains
0 ), divided byISD
concentration,
oneKobtains
the
affinity
constant,
K
A:
the affinity
constant,
:
concentration, ISD (cA0), divided by ISD (c∞), the source drain current at infinite bulk analyte
concentration, one obtains Θ
the
constant,
KA: = KKA c0c/(1{p1
Θ =pcISD
(cSD
0)/Ipc
SD (c
A cK
0) c q
“affinity
ISD
8 q∞)“
0 q{I
A 0 + K`
A 0
Θ = ISD (c0)/ISD (c∞) = KA c0/(1 + KA c0)
(1) (1)
(1)
Figure
5. Langmuir
adsorption
isothermobtained
obtained by
by plotting
plotting the
change
in in
thethe
source-drain
Figure
5. Langmuir
adsorption
isotherm
the(negative)
(negative)
change
source-drain
current, ΔISD, as obtained from Figure 4 from the new stationary current after changing the
current,
, as obtained
from
Figure
4 from
newthe
stationary
currentinafter
changing the
Figure∆I
5.SD
Langmuir
adsorption
isotherm
obtained
bythe
plotting
(negative) change
the source-drain
concentration in the bulk solution to a new value. The red curve is a fit to the data according to the
concentration
in, the
bulk solution
to a new
value.theThe
curve is current
a fit to after
the data
according
as obtained
from Figure
4 from
newred
stationary
changing
the to
current, ΔISD
Langmuir model, resulting in a dissociation constant (half saturation concentration) Kd = 3 µM.
the Langmuir
model,
in a dissociation
constant
(half
saturation
Kd =
µM.
concentration
in theresulting
bulk solution
to a new value.
The red
curve
is a fit to concentration)
the data according
to 3the
Langmuir model, resulting in a dissociation constant (half saturation concentration) Kd = 3 µM.
In Figure 5, we have plotted the surface coverage as a function of the (logarithm of the) bulk
In Figure 5, we have plotted the surface coverage as a function of the (logarithm of the) bulk
analyte concentration, and obtain the expected S-shaped curve. The corresponding fit to the data (full
In Figure 5, we have
plotted the
the expected
surface coverage
as a function
of corresponding
the (logarithm offitthe)
bulkdata
analyte
concentration,
obtain
S-shaped
to the
red curve
in Figure 5)and
results
in an affinity
constant
KA = 3.3 curve.
× 105 M−1The
, corresponding
to a dissociation
analyte
concentration,
and
obtain
the
expected
S-shaped
curve.
The
corresponding
fit
to
the
data
(full
5
´
1
(full constant,
red curve
5) results
in an K
affinity
constant KAto= the
3.3value
ˆ 10 obtained
M , corresponding
i.e.,ina Figure
half-saturation
constant,
d = 3 µM. Compared
from the kineticto a
red curve in Figure 5) results in an affinity constant KA = 3.3 × 105 M−1, corresponding to a dissociation
dissociation
constant,
half-saturation
constant, Kas
µM. Compared
to theofvalue
obtained from
measurements
(Kd i.e.,
= 6.2a µM)
this can be considered
an3excellent
confirmation
the applicability
d=
constant, i.e., a half-saturation constant, Kd = 3 µM. Compared to the value obtained from the kinetic
the kinetic
(Kto=the
6.2 quantitative
µM) this candescription
be considered
as an
excellent
of the
of the measurements
Langmuir model
of the
binding
assayconfirmation
of BSA to its
measurements (Kd = 6.2 µM)d this can be considered as an excellent confirmation of the applicability
surface-immobilized
antibody.
Furthermore,
it confirms description
that the electronic
this of
model
applicability
of the Langmuir
model
to the quantitative
of the read-out
binding of
assay
BSA to
of the Langmuir model to the quantitative description of the binding assay of BSA to its
binding reaction can be
considered
as a quantitative
methodthat
for general
biosensing
purposes.
its surface-immobilized
antibody.
Furthermore,
confirms
electronic
read-out
of this
model
surface-immobilized antibody.
Furthermore,
it it
confirms
that thethe
electronic
read-out
of this
model
The
next
question
that
naturally
comes
up
is
that
of
the
limit
of
detection
of
this
electronic
binding
reaction
cancan
bebe
considered
methodfor
forgeneral
generalbiosensing
biosensing
purposes.
binding
reaction
consideredasasaaquantitative
quantitative method
purposes.
monitoring
of antigen-antibody
interactions.
To
this
end, of
wethe
show
in Figure
6 the kinetic
recording
TheThe
next
question
that
naturally
comes
up
is
that
limit
of
detection
of
electronic
next question that naturally comes up is that of the limit of detection of thisthis
electronic
of the current change upon switching the solution that was running through the flow cell from pure
monitoring
of antigen-antibody
To this
thisend,
end,we
weshow
showinin
Figure
6 the
kinetic
recording
monitoring
of antigen-antibodyinteractions.
interactions. To
Figure
6 the
kinetic
recording
PBS buffer to a 1 nM BSA solution and then again back to pure buffer. As one can see, ΔISD can still
of the
current
change
upon
switchingthe
the solution
solution that
through
thethe
flow
cellcell
from
purepure
of the
current
change
upon
switching
thatwas
wasrunning
running
through
flow
from
be monitored with a superb signal-to-noise ratio. Based on this measurement, the LOD could be
buffer
1 nMBSA
BSAsolution
solution and
back
to pure
buffer.
As one
ΔIsee,
SD can
PBSPBS
buffer
to to
a 1a nM
andthen
thenagain
again
back
to pure
buffer.
Ascan
onesee,
can
∆Istill
SD can
estimated to be in the 100 pM analyte concentration range.
be
monitored
with
a
superb
signal-to-noise
ratio.
Based
on
this
measurement,
the
LOD
could
be be
still be monitored with a superb signal-to-noise ratio. Based on this measurement, the LOD could
estimated
to
be
in
the
100
pM
analyte
concentration
range.
estimated to be in the 100 pM analyte concentration range.
Figure 6. Single kinetic run, i.e., recording of the change in the source-drain current, ΔISD, after
switching in the flow cell from pure PBS buffer (blue arrow) to a 1 nM BSA solution (brown arrow)
Figure 6. Single kinetic run, i.e., recording of the change in the source-drain current, ΔISD, after
Figure
6. Single
kinetic
i.e., recording
of the
thebuffer
source-drain
current,
∆ISD
, after
in order
to monitor
therun,
association
rate constant,
andchange
back to in
pure
again (blue
arrow), in
order
switching in the flow cell from pure PBS buffer (blue arrow) to a 1 nM BSA solution (brown arrow)
switching
in
the
flow
cell
from
pure
PBS
buffer
(blue
arrow)
to
a
1
nM
BSA
solution
(brown
arrow)
to monitor the dissociation rate constant.
in order to monitor the association rate constant, and back to pure buffer again (blue arrow), in order
in order to monitor the association rate constant, and back to pure buffer again (blue arrow), in order
to monitor the dissociation rate constant.
to monitor the dissociation rate constant.
Biosensors 2016, 6, 17
6 of 12
Within the Langmuir model this means that, according to Equation (1) with c0 << 1 / KA ,
2016,
17 low concentration is given by
6 of 11
the Biosensors
coverage
at 6,
this
Within the Langmuir model this means that, according to Equation (1) with c0 << 1 / KA, the
Θ “ KA c0
(2)
coverage at this low concentration is given by
and with KA = 3.3 ˆ 105 M´1 , and c0 = 100 pM,
corresponds to a coverage of bound analyte
Θ = this
KA cthen
0
(2)
of Θ = 3.3 ˆ 10´5 . In other words, the minute change of the interfacial surface potential that is induced
and1with
KA30,000
= 3.3 ×antibodies
105 M−1, andonc0the
= 100
pM, this
thenbinds
corresponds
to a molecule
coverage of
of
if only
out of
sensor
surface
an analyte
is bound
enoughanalyte
to generate
−5
Θ = 3.3 × 10 . In other words, the minute change of the interfacial surface potential that is induced if
a current signal that can be quantified.
only 1 out of 30,000 antibodies on the sensor surface binds an analyte molecule is enough to generate
a current
signal
that can Comparison
be quantified.with Optical Sensing
4. Food
Toxin
Detection,
example
thatComparison
we discusswith
concerns
detection of food pathogens, aflatoxins, in
4.The
Foodnext
Toxin
Detection,
Opticalthe
Sensing
particular. They constitute a class of mycotoxins produced mainly by Aspergillus flavus and Aspergillus
The next example that we discuss concerns the detection of food pathogens, aflatoxins, in
parasiticus
which
grow
in a number
ofmycotoxins
agriculturalproduced
products.
Aflatoxin
M1 (AFM1)
is and
the hydroxylated
particular.
They
constitute
a class of
mainly
by Aspergillus
flavus
Aspergillus
metabolite
of
aflatoxin
B1
(AFB1)
and
can
be
found
in
urine,
blood,
milk,
and
internal
organs
parasiticus which grow in a number of agricultural products. Aflatoxin M1 (AFM1) ofisanimals
the
thathydroxylated
have ingested
AFB1-contaminated
feed
[16].and
Duecan
to be
itsfound
hepatotoxic
and
carcinogenic
metabolite
of aflatoxin B1
(AFB1)
in urine,
blood,
milk, and effects
internal[17]
andorgans
the relative
stability
pasteurization
or other thermal
measurements
of animals
thatduring
have ingested
AFB1-contaminated
feed treatments,
[16]. Due tocontrol
its hepatotoxic
and
were
established.
For instance,
therelative
European
Commission
stipulates a maximum
level of
50 pg¨ ml´1
carcinogenic
effects
[17] and the
stability
during pasteurization
or other thermal
treatments,
for AFM1
milk [18].
controlinmeasurements
were established. For instance, the European Commission stipulates a
−1 for AFM1 in milk [18].
maximum
level ofand
50 pg·ml
As a reference
benchmark,
we start with the presentation of an optical assay based on
As a reference
and benchmark,
we start
with the
presentation
of an
optical assay
based
on
surface-plasmon
fluorescence
spectroscopy,
developed
in our
own group,
combined
with an
inhibition
surface-plasmon
fluorescence
spectroscopy,
developed
in
our
own
group,
combined
with
an
immunoassay in which a derivative of AFM1 was immobilized on the sensor surface and monoclonal
inhibition
immunoassay
in
which
a
derivative
of
AFM1
was
immobilized
on
the
sensor
surface
and
anti-AFM1 antibodies from the rat were used as recognition elements [19]. In this protocol the analyte
monoclonal
anti-AFM1with
antibodies
from
the rat wereantibodies
used as recognition
[19]. InSome
this of
sample
is first incubated
a solution
of anti-AFM1
of a knownelements
concentration.
protocol the analyte sample is first incubated with a solution of anti-AFM1 antibodies of a known
the antibodies bind to the free AFM1 molecules; upon rinsing this cocktail through the sensor flow cell,
concentration. Some of the antibodies bind to the free AFM1 molecules; upon rinsing this cocktail
the remaining unoccupied antibodies can then bind to the surface-immobilized AFM1 antigens, and
through the sensor flow cell, the remaining unoccupied antibodies can then bind to the
are detected by decoration with a secondary, fluorescently labeled goat anti-rat antibody (Cy5-GaR,
surface-immobilized AFM1 antigens, and are detected by decoration with a secondary, fluorescently
approximately
dyes
per antibody).
labeled goat 10.2
anti-rat
antibody
(Cy5-GaR, approximately 10.2 dyes per antibody).
TheThe
result
of
a
series
of of
measurements
with
concentrations is
result of a series
measurements
withsample
samplesolutions
solutionsof
ofdifferent
different AFM1
AFM1 concentrations
presented
in Figure
7. The7.lower
the bulk
aflatoxin
concentration
the the
higher
is the
sensor
signal
from
is presented
in Figure
The lower
the bulk
aflatoxin
concentration
higher
is the
sensor
signal
the from
unoccupied
antibodies,
now being
immobilized
on the sensor
The half-saturation
value as
the unoccupied
antibodies,
now
being immobilized
on the surface.
sensor surface.
The half-saturation
a measure
sensitivity
the opticalofassay
is at c0assay
= 100ispM,
a LOD
of about
0.6 pg¨
ml´1 at
value asofathe
measure
of theofsensitivity
the optical
at cwith
0 = 100
pM, with
a LOD
of about
0.6 pg·ml−1time
at a processing
time of 1 h.
a processing
of 1 h.
Figure
7. Normalizedcalibration
calibration curves
curves for
for the
the detection
and
milk
Figure
7. Normalized
detection of
ofAFM1
AFM1ininbuffer
buffer(squares)
(squares)
and
milk
samples
(circles).
samples (circles).
Biosensors 2016, 6, 17
Biosensors 2016, 6, 17
7 of 12
7 of 11
In contrast to the optical reference assay, the electronic read-out has a number of advantages:
(i) the sensor monitors the analyte binding in real time; hence is only diffusion limited and therefore
faster; (ii) it gives direct
direct kinetic
kinetic information,
information, i.e., one also
also obtains
obtains association
association and
and dissociation
dissociation rate
constants; (iii) the assay does not use any secondary antibody, and hence requires fewer
fewer processing
processing
steps;
(iv)
the
electronic
sensing
device
based
on
“plastic
electronics”
is
a
disposable
and
does
not need
steps; (iv) the electronic sensing device based on “plastic electronics” is a disposable and does
not
any
optical
detection
instrumentation,
and hence
is cheaper.
needsophisticated
any sophisticated
optical
detection
instrumentation,
and hence
is cheaper.
We
addressed the
thequestion
questionofofnon-specific
non-specific
binding
of the
analyte
aflatoxin
to bare
the bare
We first addressed
binding
of the
analyte
aflatoxin
to the
rGO
rGO
Figure
8A demonstrates
that—Not
totally
unexpected
giventhe
themolecular
molecular structures
structures of
gate. gate.
Figure
8A demonstrates
that—Not
totally
unexpected
given
the analyte molecule and graphene, respectively, suggesting a significant non-specific binding by
π-π-interactions—A
π-π-interactions—Achange
changein
inthe
thesource-drain
source-draincurrent,
current,ISD
ISD, upon the injection of a 64 nM solution of
AFB1 into
into the
the flow
flow cell,
cell, can
can indeed
indeed be
be monitored.
monitored.
However, this non-specific binding can be totally suppressed for a test surface that is first
functionalized by
byan
anantibody
antibodyspecific
specific
a totally
different
analyte
this case).
Asseen
can
forfor
a totally
different
analyte
(BSA(BSA
in thisincase).
As can be
be
seen in8B,
Figure
even significant
concentrations
AFB1 through
rinsed through
the
flow
not
in Figure
even8B,
significant
concentrations
of AFB1ofrinsed
the flow
cell
do cell
not do
lead
tolead
any
to
any detectable
in the source-drain
current,
detectable
changechange
in the source-drain
current,
ΔISD. ∆ISD .
Figure 8. (A) Analysis of the binding of aflatoxin B1 (structure formula given in the inset of Figure 9)
on a bare rGO gate surfaces using a blank FET. The signal strength for nanomolar concentrations was
compared to the
the measurements
measurements were the target
target antibody
antibody was used
found to be only around 10% compared
(cf. Figure
Figure 9);
9). (B) Unspecific responses for aflatoxin B1 were measured
measured on
on graphene
graphene FET’s with
immobilized
observed,
immobilized PBSE-linker
PBSE-linker and
and BSA antibodies. For all tested devices no response signal was observed,
indicating that no binding of AFB1 to
to the
the linker
linker or
or to
to aa non-target
non-target antibody
antibody isis occurring.
occurring.
The specific
itsits
antibody
on on
thethe
sensor
gategate
surface
and the
issue
The
specific binding
bindingofofAFB1
AFB1toto
antibody
sensor
surface
andsensitivity
the sensitivity
can
be
best
judged
by
referring
to
the
data
displayed
in
Figure
9.
Here,
the
antibody
against
AFB1,
issue can be best judged by referring to the data displayed in Figure 9. Here, the antibody against
was directly
immobilized
on the rGO
gate
of gate
the transistor.
Rinsing analyte
of different
AFB1,
was directly
immobilized
on the
rGO
of the transistor.
Rinsingsolutions
analyte solutions
of
concentrations
(as
indicated
by
the
green
arrows
in
Figure
9,
alternating
with
pure
PBS
buffer,
blue
different concentrations (as indicated by the green arrows in Figure 9, alternating with pure PBS buffer,
arrows)
through
the attached
flow
cellcell
ledled
to atodirect
sensor
signal,
blue
arrows)
through
the attached
flow
a direct
sensor
signal,i.e.,
i.e.,aachange
changein
inthe
the source-drain
source-drain
current,
ΔI
SD, of the FET. As one can see the sensor responds with a change of its source-drain current,
current, ∆ISD , of the FET. As one can see the sensor responds with a change of its source-drain current,
with aa good
goodsignal-to-noise
signal-to-noiseratio,
ratio,already
alreadyatatanalyte
analyteconcentrations
concentrations
the
several
range.
This
with
inin
the
several
1010
pMpM
range.
This
is
is
quite
comparable
to
the
much
more
demanding
optical
approach
described
in
Figure
7.
quite comparable to the much more demanding optical approach described in Figure 7.
Biosensors 2016, 6, 17
Biosensors 2016, 6, 17
8 of 12
8 of 11
Figure9.9.Real-time
Real-timefood
foodpathogen
pathogensensor
sensoroutput,
output,i.e.,
i.e.,∆I
ΔISD
SD,, as
Figure
asaafunction
functionof
oftime,
time,while
whileAFB1
AFB1solutions
solutions
of
different
concentrations
(green
arrows),
alternating
with
pure
buffer
(blue
arrows)
of different concentrations (green arrows), alternating with pure buffer (blue arrows)were
wererinsed
rinsed
throughthe
thesample
samplecell.
cell.
through
morecomprehensive
comprehensive and
and quantitative
quantitative evaluation
evaluation of
AAmore
of the
the kinetic
kinetic and
and titration
titrationdata
dataofofthis
this
electronicfood
foodtoxin
toxinsensor
sensorisiscurrently
currentlybeing
beingperformed
performed by
by our
our group.
electronic
SmellSensing
Sensing
5.5.Smell
Thefinal
final example
give
for the
of the rGO
FET-based
biosensorsbiosensors
concerns
The
example that
thatwewe
give
for performance
the performance
of the
rGO FET-based
the development
of a smell of
detector.
Despite
the enormous
importance
of chemical
communication
concerns
the development
a smell
detector.
Despite the
enormous
importance
of chemical
in nature we have
essentially
no technical
device
or sensordevice
that could
detect that
smells
withdetect
the
communication
in nature
we have
essentially
no technical
or sensor
could
sensitivity
for mostrequired
applications
in food
quality control,
for the
detection
of
smells
with and
the selectivity
sensitivityrequired
and selectivity
for most
applications
in food
quality
control,
crop
diseases,
for
medical
applications
like
breath
analysis,
etc.
Despite
the
fact
that
monitoring
for the detection of crop diseases, for medical applications like breath analysis, etc. Despite the fact that
chemicals in
chemotaxis,
i.e., in the search
of many
organisms
ororganisms
the exchange
of chemicals
monitoring
chemicals
in chemotaxis,
i.e., infor
thefood
search
for food
of many
or the
exchange
between
species
as
a
way
to
communicate
with
each
other
is
the
oldest
of
our
sensory
repertoire,
we
of chemicals between species as a way to communicate with each other is the oldest of our sensory
have no sensor
thatnooffers
thethat
sensitivity
and
the bandwidth
needed
to sense
and to
repertoire,
we have
sensor
offers the
sensitivity
and the
bandwidth
needed
to differentiate
sense and to
many
different
odors.
differentiate many different odors.
Theconcept
conceptfor
foraasmell
smellsensor
sensorthat
that we
we are
are currently
currently developing
developing in
The
in our
our group
groupisisbased
basedon
onthe
the
functionalization of a thin film transistor with a grapheme gate by the immobilization of odorant
functionalization of a thin film transistor with a grapheme gate by the immobilization of odorant
binding proteins (OBPs) from insects as a functional element, as a bio-mimetic building block. This
binding proteins (OBPs) from insects as a functional element, as a bio-mimetic building block.
approach is aiming at reducing the complexity of nature to just the use of these proteins as a
This approach is aiming at reducing the complexity of nature to just the use of these proteins as
particularly robust element for the build-up of a first device in a bio-inspired sensor concept. OBPs
a particularly robust element for the build-up of a first device in a bio-inspired sensor concept. OBPs
from insects like those from mammals (and humans) are at the beginning of a complex amplification
from insects like those from mammals (and humans) are at the beginning of a complex amplification
scheme in odorant perception that translates a first event, the binding of an odorant molecule to such
scheme in odorant perception that translates a first event, the binding of an odorant molecule to such
an odorant binding protein, and eventually ends with the trigger of action potentials that run down
an odorant binding protein, and eventually ends with the trigger of action potentials that run down
the smell nerve directly into the brain.
the smell
nerve directly into the brain.
Figure 10 gives a complete set of experimental data taken with a reduced graphene oxide (rGO)
Figure
10 gives a complete
set of
experimental
with a reduced
graphene
oxide (rGO)
FET,
functionalized
with OBP14
from
the honeydata
beetaken
(A. mellifera).
The example
concerns
the
FET,
functionalized
with
OBP14
from
the
honey
bee
(A.
mellifera).
The
example
concerns
the
monitoring
monitoring of Homovanillic acid (structure formula given in Figure 10D), an odorant for bees
which
ofisHomovanillic
acid
(structure
formulaforgiven
in Figure 10D),
an odorant
for bees which is also
also considered
to be
a tumor marker
neuroblastoma
and malignant
pheochromocytoma.
considered
to
be
a
tumor
marker
for
neuroblastoma
and
malignant
pheochromocytoma.
We start with the recordings of the change of the source-drain current from a transistor with an
We
start
with
recordings
of linker
the change
of the
source-drain
from a transistor
with an
rGO
gate
that
wasthe
coated
with the
molecule
PBSE
(cf. Sectioncurrent
2) but without
an OBP coupled
rGO
gate
wassee
coated
the10A,
linker
PBSE (cf. Section
2) of
but
without
OBPtocoupled
to it.
As that
one can
formwith
Figure
thismolecule
test for non-specific
binding
the
analytean
leads
only a
tonegligible
it. As onesignal
can seewhich
form may
Figure
10A,
this
test
for
non-specific
binding
of
the
analyte
leads
to only
result from residual π-stacking interactions between the odorant
a molecule
negligibleand
signal
which
may
result
from
residual
π-stacking
interactions
between
the
odorant
free sites on the reduced graphene oxide gate surface.
molecule
and freeafter
sitesthe
on covalent
the reduced
graphene oxide
gate
surface.OBP14, the global analysis with
However,
immobilization
of the
receptor
both kinetic and titration information in one run results in a clear sensor signal with an excellent
signal-to-noise ratio (Figure 10B). The different concentrations of the analyte solutions that were
Biosensors 2016, 6, 17
9 of 12
However, after the covalent immobilization of the receptor OBP14, the global analysis with
both kinetic and titration information in one run results in a clear sensor signal with an excellent
signal-to-noise ratio (Figure 10B). The different concentrations of the analyte solutions that were rinsed
through
Biosensorsthe
2016,flow
6, 17 cell in this experiment are indicated by red arrows, and the flow of pure9buffer,
of 11
indicating the full reversibility of the ligand binding to the sensor-immobilized receptor is marked by
rinsed
through the flow cell in this experiment are indicated by red arrows, and the flow of pure
a blue
arrow.
buffer,
reversibility
the ligand
binding
sensor-immobilized
receptor
is
The indicating
analysis ofthe
thefull
kinetic
data is of
given
in Figure
10C. to
Asthe
predicted
by the Langmuir
model,
marked
by a blue
arrow.
the
association
process
becomes faster with increasing bulk concentration, cHomovanillic acid . By plotting
The
analysis
of
kinetic data
given in Figure
10C. As concentration,
predicted by the
model,
the
the corresponding ratethe
constant,
k, asisafunction
of the analyte
c0Langmuir
, one obtains
a straight
association
process
becomes
faster
with
increasing
bulk
concentration,
c
Homovanillic acid. By
plotting
the
3
´
1
´
1
line. Also, according to k = kon c0 – koff , the slope of this line gives kon = 1.1 ˆ 10 M s , and
corresponding rate constant, k, as afunction of the analyte concentration, c0, one obtains a straight
the intersection with the ordinate results in koff = 8 ˆ 10´3 s´1 . This then leads to the kinetically
line. Also, according to k = konc0 – koff, the slope of this line gives kon = 1.1 × 103 M−1s−1, and the
determined affinity constant KA = kon /koff = 1.4 ˆ −3105−1 M´1 , corresponding to a dissociation constant
intersection with the ordinate results in koff = 8 × 10 s . This then leads to the kinetically determined
Kd = 7.1 µM.
affinity constant KA = kon/koff = 1.4 × 105 M−1, corresponding to a dissociation constant Kd = 7.1 µM.
This value can be compared with the data obtained from the titration experiment displayed in
This value can be compared with the data obtained from the titration experiment displayed in
Figure 10D. The fit here yields KA = 2.5 ˆ 1055 M−1´1 in excellent agreement with the kinetic determination.
Figure 10D. The fit here yields KA = 2.5 × 10 M in excellent agreement with the kinetic determination.
This
confirms the applicability of the Langmuir model to this affinity binding reaction. It should be
This confirms the applicability of the Langmuir model to this affinity binding reaction. It should be
pointed
out
dissociationconstants
constantsranging
rangingfrom
froma few
a few
µM
pointed
outthat
thatby
bytesting
testingother
otherligands
ligands (odorants),
(odorants), dissociation
µM
to to
several
mM
were
found,
indicating
the
selectivity
of
this
sensor
concept
based
on
an
OBP
receptor/FET
several mM were found, indicating the selectivity of this sensor concept based on an OBP
bio-hybrid
device
[20]. device [20].
receptor/FET
bio-hybrid
Figure10.
10.Full
Fullanalysis
analysisof
ofthe
the recognition
recognition and
and binding
the
odorant
Figure
binding of
ofthe
theodorant
odoranthomovanillic
homovanillicacid
acidtoto
the
odorant
binding protein OBP14. (A) Taken with a sensor that was functionalized by the linker PBSE, however,
binding protein OBP14. (A) Taken with a sensor that was functionalized by the linker PBSE, however,
had no protein coupled to it; (B) shows the global analysis for homovanillic acid binding to OPB 14
had no protein coupled to it; (B) shows the global analysis for homovanillic acid binding to OPB 14
on the FET gate surface; (C) rate constants taken from the fits of (B), plotted as a function of the bilk
on the FET gate surface; (C) rate constants taken from the fits of (B), plotted as a function of the bilk
concentration c0; (D) Langmuir isotherm of the titration data taken from (B).
concentration c0 ; (D) Langmuir isotherm of the titration data taken from (B).
6. Outlook
Electronic biosensing shows great promise for complementing electrochemical detection
schemes on the one hand and optical concepts for the quantitative monitoring of bioanalytes on the
other. In particular, the use of graphene as the conductive gate material in the preparation of thin
film transistors as the sensing device offers a tremendous advantage compared to the use of organic
semiconducting materials or compared to Si-based transistors, which both require far more
demanding preparation protocols.
Biosensors 2016, 6, 17
10 of 12
6. Outlook
Electronic biosensing shows great promise for complementing electrochemical detection schemes
on the one hand and optical concepts for the quantitative monitoring of bioanalytes on the other.
In particular, the use of graphene as the conductive gate material in the preparation of thin film
transistors as the sensing device offers a tremendous advantage compared to the use of organic
semiconducting materials or compared to Si-based transistors, which both require far more demanding
preparation protocols.
The few examples that we presented here demonstrate the versatility of the graphene-based
transistor concept for biosensing in different fields of application. It should be noted that the use of
monomolecular graphene oxide flakes obtained by exfoliating graphite and their reduction to reduced
graphene oxide (rGO) eventually may be replaced by higher quality graphene gate materials prepared
by chemical vapor deposition (which needs to be seen, though!). The basic concept, however, of
functionalizing this gate material by different types of biorecognition elements, receptors, antibodies,
etc., able to bind the analyte molecule of interest, has proven already to yield the sensitivity and
selectivity needed for practical applications of these devices.
Some of the fundamental issues associated with rGO FET-based biosensing still need further work
and clarification; e.g., the role of local pH changes, direct but also indirect modifications of the surface
potential at the gate-electrolyte interface by variations of the local ionic milieu, the concentration, profile
and chemical nature of ions and counter-ions associated with the biorecognition reaction, the role of
dipole potentials, etc., are far from being completely understood. Other than in optical biosensing, this
also gives the concept of anti-fouling coatings a different meaning: it is absolutely not clear how to
best minimize non-specific binding. What determines eventually the limit of detection in rGO-based
electronic biosensing? What is the relationship between the gate architecture and the minimal surface
coverage that is needed in order to sense an electronic signal? Would eventually a single occupied site
on the biofunctionalized gate surface be enough to modify the source-drain current?
A final comment refers to the last application, i.e., the sensing of smells. The example given
was based on monitoring of changes induced in an odorant binding protein immobilized on the
gate surface of the transistor by the binding of a ligand, an odorant, from the aqueous phase of the
respective analyte solution in contact with the sensor. This format mimics to some extend the situation
in the nose of vertebrates where the sensory neurons and the associated odorant binding proteins are
covered and, hence, protected and hydrated by the mucosa. It also reflects properly the situation in the
sensilla of the insect antennae, where the OBPs are in the lymph, i.e., also in an aqueous environment.
In that sense, the presented results are quite relevant also for the development of a smell sensor. In fact,
a direct comparison between the sensitivity of a whole insect antenna from an insect, the red flour
beetle Tribolium castaneum, responding to a certain partial pressure of an odorant in a carrier gas,
with that of a rGO transistor functionalized by one of the key OBPs from this species, responding to an
aqueous solution of the same odorant, gave very comparable results [21]. Of course, our final goal
is to operate these devices also in air, which seems to be absolutely possible by using hydrogels as a
protective coating of the device, mimicking the mucosa that keeps all the biocomponents of the sensor
in a hydrated and therefore functional environment.
Acknowledgments: We would like to thank Barbara Cvak, Alois Schiessl, and Romer Labs Division Holding
GmbH for their continued support. We are grateful to a number of colleagues that contributed with their
advice: Jakub Dostalek, Christoph Kleber, Melanie Larisika, Rosa Mastrogiacomo, Christoph Nowak, Paolo Pelosi,
and Serban F. Peteu. Partial support for this work was provided by the European Science Foundation (ESF),
Grant Number: 10-EuroBioSAS-FP-005, by the European Union’s Horizon 2020 program under the
Sklodowska-Curie Grant Agreement No. 645686, the Austrian Science Fund (FWF) (I681-N24), the Austrian
Federal Ministry for Transport, Innovation and Technology (GZ BMVIT-612.166/0001-III/I1/2010) and the
Austrian Research Promotion Agency (FFG) within the COMET framework, by the Province of Lower Austria.
Author Contributions: Ciril Reiner-Rozman, Caroline Kotlowski, Wolfgang Knoll conceived and designed
the experiments; Ciril Reiner-Rozman, Caroline Kotlowski performed the experiments; Ciril Reiner-Rozman,
Biosensors 2016, 6, 17
11 of 12
Caroline Kotlowski, Wolfgang Knoll analyzed the data; Ciril Reiner-Rozman, Caroline Kotlowski, Wolfgang Knoll
wrote the paper.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
Jones, R.; Meixner, H. Sensors, A. Comprehensive Survey, Volume 8, Micro- and Nanosensor Technology: Trends in
Sensor Markets; Jones, R., Meixner, H., Eds.; Wiley: New York, NY, USA, 2008.
Zhang, X.J.; Ju, H.X.; Wang, J. Electrochemical Sensors, Biosensors and Their Biomedical Applications; Elsevier:
Amsterdam, The Netherlands, 2008.
Offenhäusser, A.; Maelicke, C.; Matsuzawa, M.; Knoll, W. Field-Effect Transistor Array for Monitoring
Electrical Activity from Mammalian Neurons in Culture. Biosens. Bioelectron. 1997, 12, 819–826. [CrossRef]
Yeung, C.K.; Ingebrandt, S.; Krause, M.; Sprössler, C.; Denyer, M.; Britland, S.T.; Offenhäusser, A.; Knoll, W.
The use of Field Effect Transistors (FETs) in pharmacological research. J. Anat. 2002, 200, 210.
Ingebrandt, S. Bioelectronics: Sensing Beyond the Limit. Nat. Nanotechnol. 2015, 10, 734–735. [CrossRef]
[PubMed]
Khan, H.U.; Jang, J.; Kim, J.J.; Knoll, W. In Situ Antibody Detection and Charge Discrimination Using
Aqueous Stable Pentacene Transistor Biosensors. J. Am. Chem. Soc. 2011, 133, 2170–2176. [CrossRef]
[PubMed]
Khan, H.U.; Roberts, M.E.; Johnson, O.; Förch, R.; Knoll, W.; Bao, Z. In Situ, Label-Free DNA Detection Using
Organic Transistor Sensors. Adv. Mater. 2010, 22, 4452–4456. [CrossRef] [PubMed]
Kuila, T.; Bose, S.; Khanra, P.; Mishra, A.K.; Kim, N.H.; Lee, J.H. Recent Advances in Graphene-Based
Biosensors. Biosens. Bioelectron. 2011, 26, 4637–4648. [CrossRef] [PubMed]
Reiner-Rozman, C.; Larisika, M.; Nowak, C.; Knoll, W. Graphene-based liquid-gated field effect transistor
for biosensing: Theory and experiments. Biosens. Bioelectron. 2015, 70, 21–27. [CrossRef] [PubMed]
Larisika, M.; Kotlowski, C.; Steininger, C.; Mastrogiacomo, R.; Pelosi, P.; Schütz, S.; Peteu, S.F.; Kleber, C.;
Reiner-Rozman, C.; Nowak, C.; et al. Electronic Olfactory Sensor Based on A. mellifera Odorant Binding
Protein 14 on a Reduced Graphene Oxide Field-Effect Transistor. Angew. Chem. Int. Ed. Engl. 2015, 54,
13245–13248. [CrossRef] [PubMed]
Larisika, M.; Huang, J.F.; Tok, A.; Knoll, W.; Nowak, C. An Improved Synthesis Route to Graphene for
Molecular Sensor Applications. Mater. Chem. Phys. 2012, 136, 304–308. [CrossRef]
Stankovich, S.; Dikin, D.A.; Piner, R.D.; Kohlhaas, K.A.; Kleinhammes, A.; Jia, Y.; Wu, Y.; Nguyen, S.B.T.;
Ruoff, R.S. Synthesis of Graphene-Based Nanosheets via Chemical Reduction of Exfoliated Graphite Oxide.
Carbon 2007, 45, 1558–1565. [CrossRef]
Dani, F.R.; Iovinella, I.; Felicioli, A.; Niccolini, A.; Calvello, M.A.; Carucci, M.G.; Qiao, H.; Pieraccini, G.;
Turillazzi, S.; Moneti, G.; et al. Mapping the Expression of Soluble Olfactory Proteins in the Honeybee.
J. Proteome Res. 2010, 9, 1822–1833. [CrossRef] [PubMed]
Iovinella, I.; Dani, F.R.; Niccolini, A.; Sagona, S.; Michelucci, E.; Gazzano, A.; Turillazzi, S.; Felicioli, A.;
Pelosi, P. Differential Expression of Odorant-Binding Proteins in the Mandibular Glands of the Honey Bee
According to Caste and Age. J. Proteome Res. 2011, 10, 3439–3449. [CrossRef] [PubMed]
Ban, L.P.; Zhang, L.; Yan, Y.H.; Pelosi, P. Binding Properties of a Locust’s Chemosensory Protein.
Biochem. Biophys. Res. Commun. 2002, 293, 50–54. [CrossRef]
Shreeve, B.J.; Patterson, D.S.P.; Roberts, B.A. Mycotoxins and Their Metabolites in Humans and Animals.
Food Cosmet. Toxicol. 1979, 17, 151–152. [CrossRef]
Paniel, N.; Radoi, A.; Marty, J.L. Development of an electrochemical biosensor for the detection of aflatoxin
M1 in milk. Sensors 2010, 10, 9439–9448. [CrossRef] [PubMed]
Commission Regulation (EC) No. 466/2001. Available online: http://ec.europa.eu/food/fs/sfp/fcr/
fcr02_en.pdf (accessed on 12 April 2016).
Wang, Y.; Dostalek, J.; Knoll, W. Long Range Surface Plasmon-enhanced Fluorescence Spectroscopy for
the Detection of Aflatoxin M-1 in milk. Biosens. Bioelectron. 2009, 24, 2264–2267. [CrossRef] [PubMed]
Biosensors 2016, 6, 17
20.
21.
12 of 12
Kotlowski, C.; Larisika, M.; Guerin, P.; Kleber, C.; Kröber, T.; Mastrogiacomo, R.; Nowak, C.; Pelosi, P.;
Schütz, S.; Schwaighofer, A.; et al. Electronic Olfactory Biosensor Based on Honeybee Odorant Binding
Protein 14 on a rGO FET. In preparation.
Kotlowski, C.; Ramach, U.; Balakrishnan, K.; Breu, J.; Gabriel, M.; Kleber, C.; Mastrogiacomo, R.;
Pelosi, P.; Schütz, S.; Knoll, W. Monitoring Crop Disease Markers by Odorant Binding Proteins of
Tribolium castaneum—Electroantennograms versus Reduced Graphene-Oxide Based Electronic Biosensor.
In preparation.
© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).