Sub-ppm detection of nerve agents using chemically functionalized silicon nanoribbon field-effect transistors

Article abstract of DOI:10.1002/anie.201000122

(Figure Presented) A chemical receptor specific to traces of organophosphorus agents (OPs) has been synthesized and grafted to a silicon nanoribbon field-effect transistor (SiNRFET). X-ray structures illustrate the structural modifications of the receptor upon exposure to nerve agent simulants. A highly sensitive and selective detector of OPs can be obtained by monitoring the Drain-Source current of the SiNR-FET at an optimum back-gate voltage as a function of time.

Full text of DOI:10.1002/anie.201000122

Angewandte
Chemie
DOI: 10.1002/anie.201000122
Nerve Agent Sensors
Sub-ppm Detection of Nerve Agents Using Chemically Functionalized
Silicon Nanoribbon Field-Effect Transistors**
Simon Clavaguera, Alexandre Carella,* Laurent Caillier, Caroline Celle, Jacques Pꢀcaut,
Stꢀphane Lenfant, Dominique Vuillaume, and Jean-Pierre Simonato*
Organophosphorus compounds (OPs) represent one of the
most important and lethal classes of chemical warfare agents
(e.g. sarin, tabun, soman). Highly active volatile OPs are
powerful inhibitors of acetylcholinesterase, which is a critical
enzyme of the nervous system.[1] The ease of manufacturing
OPs based on inexpensive starting materials makes these
agents a weapon of choice for terrorist attacks.[2] Thus, the
rapid sensing of these nerve agents has recently become an
increasingly important research goal. Various approaches
have been reported for the detection of these chemical
warfare agents including colorimetric and fluorimetric spec-
troscopies,[3] enzymatic assays,[4] piezoelectric devices,[5]
single-walled carbon nanotube resistors[6] and capacitors.[7]
However, these systems are plagued by limitations such as
slow response time, moderate selectivity, operational com-
plexity, or limited portability.
we report the development of an OPs chemical sensor based
on highly sensitive silicon nanoribbon field-effect transistors
(SiNR-FETs) functionalized with compound 1 (Scheme 1).
Scheme 1. Sensitive receptor towards OPs. Compound 1 converts into
compound 2 upon exposure to OPs simulant (DPCP).
Field-effect transistors (FET) based on nanomaterials
such as semiconducting nanowires, nanoribbons, or carbon
nanotubes have been recently explored for chemical and
biological detection.[8] Their high effectiveness is mainly
ascribed to an extreme sensitivity to electrostatic changes at
the surface of the semiconductor and/or modifications of the
Schottky barrier at the semiconductor/metal interface. A
charge generation in the vicinity of the semiconductor of a
FET is known to alter the electrical properties of the device.[9]
Several research groups have independently developed a
series of small-molecule fluorescent sensors for OPs detec-
tion.[10] They investigated organic moieties reactive towards
OPs by formation of a phosphate ester intermediate and
subsequent intramolecular nucleophilic substitution, which
led to an ammonium salt and thus charge formation. We
thought monitoring this charge generation with a functional-
ized FET could be a particularly promising approach. Herein,
Compound 1, which is based on the scaffold developed by
Dale and Rebek,[10a] was synthesized in four steps starting
from the Kempꢀs triacid with an overall yield of 40%. The
ethynyl substituent was chosen to covalently graft the
receptor to the semiconductor. Diphenylchlorophosphate
(DPCP) was used as a simulant of nerve agents because of
its similar structure and chemical reactivity, but much lower
toxicity. Compound 1 reacted cleanly with DPCP and
produced the azaadamantane quaternary ammonium salt 2
as confirmed by 1H and 31P NMR spectroscopy. The DPCP 31
P
signal (d = ꢀ5.8 ppm) decreased while the signal of the
phosphate generated after reaction with 1 increased (d =
ꢀ26.2 ppm). Monocrystals of 1 and 2 were analyzed by
single-crystal X-ray analysis (Figure 1).[11] In 1, a strong
intramolecular hydrogen bond between the tertiary nitrogen
atom and the hydroxy hydrogen atom (N···H distance is
1.950 ꢁ, N···O distance is 2.805 ꢁ and N···H-O angle is 178.18)
bends the adjacent cycle where five carbon atoms are nearly
coplanar (average out-of-plane deviation = 0.01 ꢁ). The
nucleophilic power of the oxygen atom is probably enhanced
because of the participation of the hydrogen atom in the
intramolecular hydrogen bond.[12] The azaadamatanium
moiety in 2 is highly symmetrical (C3 axis) which corroborates
the simplification of NMR signals from 1 to 2.
[*] Dr. S. Clavaguera, Dr. A. Carella, Dr. L. Caillier, Dr. C. Celle,
Dr. J.-P. Simonato
CEA Grenoble/LITEN/DTNM
17 rue des Martyrs, 38054 Grenoble Cedex 9 (France)
E-mail: alexandre.carella@cea.fr
jean-pierre.simonato@cea.fr
Dr. J. Pꢀcaut
CEA Grenoble/INAC
17 rue des Martyrs, 38054 Grenoble Cedex 9 (France)
The SiNR-FETs were fabricated from p-doped (1015
B
atomcmꢀ3) silicon on insulator (SOI) wafers (Figure 2).
SiNR-FETs of 70 nm thickness with different lengths and
widths (4 ꢂ 4 mm; 4 ꢂ 1 mm; 2 ꢂ 1 mm, and 2 ꢂ 0.2 mm) were
obtained by using e-beam lithography and dry reactive-ion
etching steps. The thickness of the SiO2 gate dielectric was
140 nm. The Ti/Au (10/100 nm) source and drain contacts
were achieved by using e-beam lithography and lift-off
Dr. S. Lenfant, Dr. D. Vuillaume
IEMN—CNRS, BP 60069
Avenue Poincarꢀ, 59652 Villeneuve d’Ascq Cedex (France)
[**] We thank Nicolas Durrafourg for access to NMR instrumentation
and Florian Nachon for helpful discussions.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201000122.
Angew. Chem. Int. Ed. 2010, 49, 4063 –4066
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4063

Communications
A statistical study on 30 nanoribbons was carried out and
shows a clear trend upon exposure to vapors of DPCP. The
IDS–VGS curves are shifted to more positive gate-voltage bias.
To simplify data handling, a particular gate voltage called V0,
which corresponds to off-current minimum intensity (IDS
min), was used. Figure 4 illustrates the shift of that particular
Figure 1. X-ray structures of 1 (left) and 2 (right). Hydrogen atoms and
counter ions are omitted for clarity. The thermal ellipsoids are drawn
at the 50% probability level.
Figure 4. Functionalized SiNR-FET, (V0; IDS min) couples picked from
I
DS–VGS curves at VDS =ꢀ1 V before (diamonds) and after (squares)
DPCP exposure.
Figure 2. Left: AFM image of a SiNR-FET (70 nm thick, 2 mm long and
1 mm wide). Right: schematic view of FET device with source (S), drain
(D), and gate (G) electrodes.
gate voltage V0 before and after exposure to vapors of DPCP.
One might observe that the scatter plots of the couples (V0;
IDS min) before and after DPCP exposure are not covered.
The V0 is consequently clearly shifted to more positive gate
voltages after exposure to DPCP with an average variation
DV0 of (7.3 ꢁ 3.5) V.
process. The Si degenerated substrate was used as the back-
gate electrode. The semiconducting part of the sensor was
functionalized by covalent grafting through thermal hydro-
silylation of 1 onto the HF-pretreated substrate in mesitylene
at reflux for 2 hours.
In Figure 3, the drain-source current (IDS) is plotted versus
the back-gate voltages (VGS) and shows that the device is
ambipolar (electron(hole) conduction for positive(negative)
back-gate bias, respectively). The exposure of functionalized
SiNR-FETs to vapors of DPCP induced a strong modification
of their transfer curves.
Modification of the drain current in a molecularly
functionalized SiNR-FET can be a result of several factors:
1) creation of a net charge in the molecules which are acting
as a virtual top gate;[13] 2) change in the density of the silicon
surface states;[9,14] 3) charge transfer between Si and mole-
cules, which leads to interface dipoles and change in the
silicon surface potential.[15]
In our case, positive charges are created on top of the
monolayer of compound 1 after treatment with DPCP (see
Scheme 1), which can eventually act as a virtual top gate for
the SiNR-FET. However, a negatively charged counter-ion is
also created and the monolayer is likely to remain neutral.
Thus, factor 1) may be excluded here. As the reaction of
DPCP with compound 1 does not change the nature and
density of grafting links between 1 and silicon, factor 2) can be
also excluded. Charge transfer at the molecule/substrate
interface is a well-known effect[15] and any change in the
chemical structure of the molecule (i.e. upon reaction with
DPCP) is likely to modify this charge transfer and the
resulting interface dipole. Here, we assume that upon reaction
with DPCP, and probably because of the presence of a
positive charge on the N atom, some electrons are transferred
from the Si to the molecules. As a consequence, this leads to
more holes in the Si channel when the SiNR-FET is working
in accumulation (p-type behavior at VGS < V0), and less
electrons at VGS > V0 (n-type behavior). This observation is
consistent with the current increase(decrease) experimentally
Figure 3. 4ꢁ1 mm2 functionalized SiNR-FET, IDS–VGS curves at
VDS =ꢀ1 V before (diamonds) and after DPCP exposure (squares).
4064 www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 4063 –4066

Angewandte
Chemie
observed at negative(positive) back-gate voltages, respec-
tively. The amplitude of this charge transfer can be estimated
as follows; the created interface dipole induces a modification
of the surface potential Df according to:
semiconductor resistor (MOCSER) has already been
observed upon exposure to gas.[17]
The design of FET sensors allows us to modulate the
output by applying the optimum gate voltage and thus
enhancing the response. Hence, DI/I0 is maximized at V0. In
the case of Figure 3, V0 is at VGS = ꢀ2 V. At this gate voltage,
IDS increases by a factor of two to four orders of magnitude
upon exposure to vapors of DPCP.[18]
end
ð1Þ
Df ¼
eSAM
A
To avoid charge-trap effects or Joule effect[19] during the
IDS versus time measurement, we developed a pulsed
acquisition mode. Figure 5 shows a stable IDS before exposure
with e the elementary electron charge, n the number of
transferred charge per molecule, d the size of the interface
dipole (between its two sides), eSAM the dielectric constant of
the self-assembled monolayer (SAM), and A the area per
molecule. We assume that d is of the same order of magnitude
as the distance between the positive charges localized on the
N atom of compound 1 and the silicon surface and can be
estimated to d = 0.7–0.8 nm. We take eSAM ꢂ 2.0–2.5, a usual
value for organic monolayers. Without a precise knowledge of
the molecular organization and packing inside the monolayer
of 1 (or 2), and accordingly to the size of the molecule
elucidated by X-ray crystallographic data (we assume here
that the tilt angle of the molecule with respect to the surface
normal is q = 08), we can calculate an average molecular
density of approximately Ncalc ꢂ 2.8 ꢂ 1014 molecules.cmꢀ2,
that is, A ꢂ 35 ꢁ2. We have not a direct measurement of Df,
but we can estimate it from the shift of the back-gate voltage.
Considering the classical scaling rules of the FET device, a
potential change of Df at a distance d (through a material
with a dielectric constant eSAM) from the Si channel (virtual
top gate) is roughly equivalent to a change DVGS of the back-
gate voltage through a gate oxide tox (dielectric constant eox) if
Figure 5. 4ꢁ1 mm2 functionalized SiNR-FET, IDS measured in pulsed
mode as a function of time (VDS =ꢀ1 V, VGS =V0 =ꢀ2 V). DPCP
vapors were introduced at t=240 s. The device is the same as in
Figure 3.
DVGS = (tox/d)(eSAM/eox)Df. With
tox = 140 nm, eox = 3.9, a
back-gate shift of approximately 7.3 V corresponds to Df
ꢂ 70 mV. From Equation (1), we estimated that approxi-
mately 4 ꢂ 10ꢀ3 electron/molecule are transferred from Si to
the molecules (or equivalently, ca. 1.1 ꢂ 1012 electron.cmꢀ2).
While deduced using crude approximations, this value is in
agreement with other results on functionalized silicon surfa-
ces and devices.[15,16] A more detailed analysis of SiNR-FET
would require 2D device simulations, which is beyond the
scope of this Communication.
The transfer characteristic of pristine SiNR-FETs pres-
ents no substantial deviation upon exposure to vapors of
DPCP. The sensitivity of the SiNR-FET sensor is thus
inherent to the chemical functionalization of the semicon-
ducting channel.
An increase of IDS min can also be observed upon
exposure to DPCP vapors which could be ascribed to gate
leakage. The IGS–VGS measurements were carried out before
and after exposure to vapors of DPCP, they show no
significant variation (see Figure S6 in the Supporting Infor-
mation). However, as after exposure V0 is shifted towards
more positive values and because IGS increases with the gate
voltage, a tiny part of the increase in the IDS min can result
from this gate leakage. We also surmise that the charge
transfer at the silicon–molecule interface involved to explain
the IDS–VDS shift can also affect the off current, because
charge density in the nanoribbon are modified by this
interface dipole. For example, change in the conductance/
resistance value of a molecular-controlled functionalized
to DPCP vapors. After exposure, IDS rose immediately with a
steep current increase and a plateau was reached within few
tens of seconds. The signal remained stable, thus suggesting
that all the chemical receptors had reacted. The small
differences (i.e. a factor 2 to 4) between the currents IDS
before and after DPCP exposure measured by the pulsed
method (Figure 5) and the ramp method (Figure 3) are
usual[20] and have been ascribed to phenomena such as
hysteresis, slow current drift, or local heating which affect the
static method and not the pulsed one. It must be noted that
the vapor pressure of DPCP was measured in the 500–800 ppb
range, which reveals how sensitive these sensors are. More-
over, the sensor is cumulative, which is appropriate to OPs
toxicity which poisons cumulatively the nervous system.[21]
Besides very good sensitivity, a crucial point in the
realization of chemical sensors is to achieve high selectivity.
The response of our sensor was evaluated for various common
organic compounds and gases (Figure 6). No significant
variation was observed for these molecules. It appears that
the response of the sensor is much higher for DPCP compared
to the other compounds. Moreover, it should be emphasized
that the response under DPCP vapors was obtained at a sub-
ppm concentration (see above) while the sensor exhibited
extremely weak responses for exposures to other compounds,
at concentrations close to their vapor pressure (> 20000 ppm,
see Table S6 in the Supporting Information). As a conse-
Angew. Chem. Int. Ed. 2010, 49, 4063 –4066
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 4065

Communications
[1] C. P. Holstege, M. Kirk, F. R. Sidell, Crit. Care Clin. 1997, 13,
923 – 942.
[2] R. J. Brennan, J. F. Waeckerle, T. W. Sharp, S. R. Lillibridge,
Ann. Emerg. Med. 1999, 34, 191 – 204.
[3] a) K. J. Wallace, J. Morey, V. M. Lynch, E. V. Anslyn, New J.
Chem. 2005, 29, 1469 – 1474; b) S. Royo, R. Martꢃnez-Mꢄꢅez, F.
Sancenꢆn, A. M. Costero, M. Parra, S. Gil, Chem. Commun.
2007, 4839 – 4847; c) T. J. Dale, J. Rebek, Jr., Angew. Chem.
2009, 121, 7990 – 7992; Angew. Chem. Int. Ed. 2009, 48, 7850 –
7852.
[4] X. Ji, J. Zheng, J. Xu, V. K. Rastogi, T.-C. Cheng, J. J. DeFrank,
R. M. Leblanc, J. Phys. Chem. B 2005, 109, 3793 – 3799.
[5] a) C. Hartmann-Thompson, J. Hu, S. N. Kaganove, S. E. Kei-
nath, D. L. Keeley, P. R. Dvornic, Chem. Mater. 2004, 16, 5357 –
5364; b) Q. Zhao, Q. Zhu, W. Y. Shih, W.-H. Shih, Sens.
Actuators B 2006, 117, 74 – 79.
[6] a) J. P. Novak, E. S. Snow, E. J. Houser, D. Park, J. L. Stepnow-
ski, R. A. McGill, Appl. Phys. Lett. 2003, 83, 4026 – 4028; b) F.
Wang, H. Gu, T. M. Swager, J. Am. Chem. Soc. 2008, 130, 5392 –
5393.
[7] E. S. Snow, F. K. Perkins, E. J. Houser, S. C. Badescu, T. L.
Reinecke, Science 2005, 307, 1942 – 1945.
[8] a) N. Elfstrꢇm, A. E. Karlstrꢇm, J. Linnros, Nano Lett. 2008, 8,
945 – 949; b) K. S. Chang, C. C. Chen, J. T. Sheu, Y.-K. Li, Sens.
Actuators B 2009, 138, 148 – 153; c) J. Du, D. Liang, H. Tang,
X. P. A. Gao, Nano Lett. 2009, 9, 4348 – 4351; d) D. R. Kauffman,
A. Star, Angew. Chem. 2008, 120, 6652 – 6673; Angew. Chem. Int.
Ed. 2008, 47, 6550 – 6570.
[9] N. Elfstrꢇm, R. Juhasz, I. Sychugov, T. Engfeldt, A. E. Karl-
strꢇm, J. Linnros, Nano Lett. 2007, 7, 2608 – 2612.
[10] a) T. J. Dale, J. Rebek, Jr., J. Am. Chem. Soc. 2006, 128, 4500 –
4501; b) S.-W. Zhang, T. M. Swager, J. Am. Chem. Soc. 2003, 125,
3420 – 3421; c) A. M. Costero, S. Gil, M. Parra, P. M. E. Mancini,
R. Martꢃnez-Mꢄꢅez, F. Sancenꢆn, S. Royo, Chem. Commun.
2008, 6002 – 6004.
Figure 6. IDS changes for a 3 min exposure to vapors of common
organic compounds, gases, and DPCP at 258C in air (VDS =ꢀ1 V,
GS =V0). All vapor pressures and gas concentrations are listed in
V
Table S6 in the Supporting Information.
quence, we demonstrate that high selectivity can be obtained
in combination with excellent sensitivity.
In summary, we have developed a highly sensitive and
selective nerve agent sensor based on electrical transduction
of a chemical reaction occurring on the surface of a
functionalized SiNR-FET. The tiny semiconductor device,
which operated at sub-ppm level of OPs simulant, showed a
very fast and marked response which could be advantageously
used for detection of Sarin-like agents. We hope this
technology of hybrid sensors compatible with complemen-
tary–metal–oxide semiconductor (CMOS) technology will
help to develop sensitive, compact, low-cost, and low-
consumption portable devices for widespread applications in
the fields of defense and homeland security.
[11] CCDC 760718 (1) and 760719 (2) contain the supplementary
crystallographic data for this paper. These data can be obtained
free of charge from The Cambridge Crystallographic Data
Centre via www.ccdc.cam.ac.uk/data_request/cif.
[12] V. E. Anderson, M. W. Ruszczycky, M. E. Harris, Chem. Rev.
2006, 106, 3236 – 3251.
[13] O. Shaya, M. Shaked, A. Doron, A. Cohen, I. Levy, Y.
Rosenwaks, Appl. Phys. Lett. 2008, 93, 043509.
[14] R. Cohen, L. Kronik, A. Shanzer, D. Cahen, A. Liu, Y.
Rosenwaks, J. K. Lorenz, A. B. Ellis, J. Am. Chem. Soc. 1999,
121, 10545 – 10553.
[15] D. Cahen, R. Naaman, Z. Vager, Adv. Funct. Mater. 2005, 15,
1571 – 1578.
[16] T. He, J. He, M. Lu, B. Chen, H. Pang, W. F. Reus, W. M. Nolte,
D. P. Nackashi, P. D. Franzon, J. M. Tour, J. Am. Chem. Soc.
2006, 128, 14537 – 14541.
[17] D. G. Wu, D. Cahen, P. Graf, R. Naaman, A. Nitzan, D. Shvarts,
Chem. Eur. J. 2001, 7, 1743 – 1749.
Experimental Section
Experimental details are given in the Supporting Information,
including the synthetic procedures, device details, electrical charac-
terizations, and crystallographic data for compounds 1 and 2.
Functionalized SiNR-FETs were prepared from the SOI chip bearing
silicon nanoribbons of various dimensions. The chip was sonicated in
acetone then in isopropanol for 5 min prior to deoxidization with an
immersion in 1% HF aqueous solution for 20 seconds. The chip was
thoroughly rinsed three times with deionized water and dried under
argon flow. The treated chip was introduced in a mesitylene solution
of 1 (1.5 mg in 10 mL, 0.5 mm) and heated at reflux for 2 h. The chip
was washed successively with toluene, acetone, isopropanol, and dried
under argon flow and stored in air prior to use.
[18] A simpler detection output is the source-drain resistance of the
device (Figure 3 at VGS = 0 V). Indeed the RSD value decreases
by one order of magnitude upon exposure to vapors of DPCP
which is already a significant value.
[19] I. Park, Z. Li, A. P. Pisano, R. S. Williams, Nano Lett. 2007, 7,
3106 – 3111.
Received: January 8, 2010
Revised: March 20, 2010
[20] a) D. Estrada, S. Dutta, A. Liao, E. Pop, Nanotechnology 2010,
21, 085702; b) C. Petit, D. Zander, K. Lmimouni, M. Ternisien,
D. Tondelier, S. Lenfant, D. Vuillaume, Org. Electron. 2008, 9,
979 – 984.
[21] D. E. Lenz, D. Yeung, J. R. Smith, R. E. Sweeney, L. A. Lumley,
D. M. Cerasoli, Toxicology 2007, 233, 31 – 39.
Published online: April 30, 2010
Keywords: field-effect transistors · nanotechnology ·
.
nerve agents · sensors · silicon nanoribbons
4066 www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 4063 –4066