Hydroxy-terminated organic semiconductor-based field-effect transistors for phosphonate vapor detection

Article abstract of DOI:10.1021/ja068964z

Organic field-effect transistors (OFETs) with a hydroxy-functionalized semiconductor incorporated into a receptor layer were fabricated and shown to respond strongly to the analyte dimethyl methylphosphonate (DMMP) that simulates phosphonate nerve agents.

Full text of DOI:10.1021/ja068964z

Published on Web 07/11/2007
Hydroxy-Terminated Organic Semiconductor-Based
Field-Effect Transistors for Phosphonate Vapor Detection
Jia Huang,† Joseph Miragliotta,‡ Alan Becknell,‡ and Howard E. Katz*,†
Contribution from the Department of Materials Science and Engineering, Johns Hopkins
UniVersity, Baltimore, Maryland 21218, and Sensor system group, Research and Technology
DeVelopment Center, Applied Physics Laboratory, The Johns Hopkins UniVersity,
Laurel, Maryland 20723
Received December 14, 2006; E-mail: hekatz@jhu.edu
Abstract: Organic field-effect transistors (OFETs) with a hydroxy-functionalized semiconductor incorporated
into a receptor layer were fabricated and shown to respond strongly to the analyte dimethyl methylphos-
phonate (DMMP) that simulates phosphonate nerve agents. Large and reproducible source-drain current
changes were observed upon exposure to DMMP vapor. Compared to single component transistors, OFETs
with a mixed hydroxylated and nonhydroxylated semiconductor upper layer exhibited higher sensitivity.
We further investigated the selectivity of the heterostructured OFETs by comparing responses upon exposure
to different interference vapors with response to DMMP exposure. Much higher response was observed in
the case of DMMP, even when the concentration of DMMP vapor was much lower than other analytes.
Microstructures of OSC were characterized by scanning electron microscopy (SEM) and X-ray diffraction
(XRD), revealing that the organic mixture has similar crystal structure and surface morphology to those of
single component OSC films, indicating that the enhanced performance of the mixture is due to its chemical
properties, rather than microstructural effects.
Introduction
Analyte concentrations of 10-100 ppm in the vapor phase were
detected. A second virtual array study has recently appeared.7
Organic field-effect transistors (OFETs) are projected for
application in circuits of moderate complexity, such as display
drivers, radio frequency identification tags, or pressure mapping
elements.1-3 A compelling additional application of organic
electronics is chemical sensing, especially of vapors. This is
due to the ability to covalently attach receptors for compounds
of interest to the molecules that make up the organic semicon-
ductor (OSC), in locations where analyte-receptor binding will
strongly influence the current flowing across a transistor channel.
Responses are simply noted as changes in the output source
drain current for a given set of input drain voltages and gate
voltages as the vapor adsorbs onto the OFET.
Various OSCs including phthalocyanines and naphthalene-
tetracarboxylic dianhydride are sensitive to gases such as
oxygen, nitrogen dioxide, ammonia, carbon monoxide, and
hydrogen sulfide.4,5 A virtual array of 11 different OSCs in
OFETs showed distinguishable (0.8-0.3-fold reductions and
1.5-2-fold increases) responses for different classes of analytes,
in the absence of any particular receptor for the analytes.6
It was later found that devices that included grain boundaries
were much more sensitive to pentanol vapor than were the
single-crystal devices, indicating that grain boundaries were a
main adsorption site in this class of film.8 A following study
was published reporting distinctions in the nonspecific responses
of pentacene OFETs to pentanol depending on whether the
devices were grain boundary or single-grain dominated,9
consistent with the results of our preliminary study. Thus, the
sensitivity of OFETs is highly correlated with topographic
features in the films.
Some recent progress has been made using carbon nanotube
sensing elements for vapors,10,11 though rational vapor-receptor
interactions were not utilized. Inorganic and nanostructured
semiconductors are aimed at single-molecule detection but are
not as well suited for large-area sensing. In some cases, the
responses are irreversible and may be occurring at contacts rather
than the semiconductors themselves.12,13 Swellable polymers14-18
(6) Crone, B.; Dodabalapur, A.; Gelperin, A.; Torsi, L.; Katz, H. E.; Lovinger,
A. J.; Bao, Z. Appl. Phys. Lett. 2001, 78 (15), 2229-2231.
(7) Liao, F.; Chen, C.; Subramanian, V. Sens. Actuators, B 2005, 107, (2),
849-855.
(8) Torsi, L.; Lovinger, A. J.; Crone, B.; Someya, T.; Dodabalapur, A.; Katz,
H. E.; Gelperin, A. J. Phys. Chem. B 2002, 106, 12563-12568.
(9) Wang, L.; Fine, D.; Dodabalapur, A. Appl. Phys. Lett. 2004, 85, 6386-
6388.
† Department of Materials Science and Engineering.
‡ Applied Physics Laboratory.
(1) Kelley, T. W.; Baude, P. f.; Gerlach, C.; Ender, D. E.; Muyres, D.; Haase,
M. A.; Vogel, D.; Theiss, S. D. Chem. Mater. 2004, 16, 4413-4422.
(2) Parashkov, R.; Becker, E.; Riedl, T.; Johannes, H. H.; Kowalsky, W. Proc.
IEEE 2005, 93, 1321-1329.
(3) Someya, T.; Sekitani, T.; Iba, S.; Kato, Y.; Kawaguchi, H.; Sakurai, T.
Proc. Natl. Acad. Sci. U.S.A. 2004, 101 (27), 9966-9970.
(4) Guillaud, G.; Simon, J.; Germain, J. P. Coord. Chem. ReV. 1998, 180,
1433-1484.
(10) Snow, E. S.; Perkins, F. K.; Houser, E. J.; Badescu, S. C.; Reinecke, T. L.
Science 2005, 307, 1942-1945
(11) Staii, C.; Johnson, Alan T., Jr.; Chen, M.; Gelperin, A. Nano Lett. 2005, 5
(9), 1774-1778.
(5) Torsi, L.; Dodabalapur, A.; Cioffi, N.; Sabbatini, L.; Zambonin, P. G. Sens.
Actuators, B 2001, 77 (1-2), 7-11.
(12) Ramanathan, K.; Bangar, M. A.; Yun, M.; Chen, W.; Myung, N. V.;
Mulchandani, A. J. Am. Chem. Soc. 2005, 127, 496-497.
9
9366
J. AM. CHEM. SOC. 2007, 129, 9366-9376
10.1021/ja068964z CCC: $37.00 © 2007 American Chemical Society

OFETs for Phosphonate Vapor Detection
A R T I C L E S
Scheme 1
added dropwise to a solution of 2,2′-bithiophene (4.16 g, 25mmol) in
170 mL of THF at -78 °C with stirring in N2. After a white precipitate
formed, the mixture was warmed to room temperature and stirred for
1 h. Tri-n-butyltin chloride (14 mL, 52mmol) was added in the mixture,
and then the mixture was heated to gentle reflux for 1 h. The solution
was allowed to cool to room temperature, and then 400 mL of n-hexane
were added to the solution. The organic layer was washed with aqueous
Na2CO3 (5%, 125 mL) and water (125 mL twice). The organic layer
was dried over MgSO4 and evaporated in vacuo to give a dark brown
product (17.8 g, 96%). NMR (CDCl3) δ 0.91, 1,13, 1.36, 1.56 (Bu),
7.05, 7.29 (ABq, thiophene H).
The mixture of compound A (5.77 g, 21.2mmol), compound B 5,5′-
bis(tri-n-butylstannyl)-2,2′-bithiophene (6.9 g, 9.3mmol), and Pd(PPh3)4
(0.5 g, 0.7mmol) in 100 mL of DMF was stirred for 60 h under N2 at
80 °C. The dark brown precipitate was collected and washed with
MeOH and toluene, yielding a dark brown product. The crude solid
was hot-filtered and recrystallized with DMF to yield 0.94 g of yellow-
brown solid. 0.6 g of the crude product was further purified by vacuum
sublimation and gave 0.38 g pure yellow product. Mp 244-248. Mass
Spectrum: 550 (M+), 350 (base peak). Anal. Calcd for C32H34S2O4:
C, 69.78; H, 6.95; S, 11.6. Found: C, 68.99; H, 6.67; S, 12.2.
Analyte Compounds: 2-Octanone, butyl butyrate, acetic acid,
dibutylamine, and DMMP were purchased from Aldrich, and diiso-
propyl methylphosphonate (DIMP) was purchased from Alfa Aesar.
All of the compounds were used as received without further purification.
Several tests were performed to verify the purity and authenticity of
the DMMP. The pH value of the DMMP aqueous extract was found
to be neutral, indicating the absence of basic impurities. GC-MS was
also performed to check the purity of DMMP. With an S/N ratio of
100, the mass value of the major peak is 124, and no impurity was
detected. Because GC-MS cannot detect highly volatile impurities,
NMR measurement was performed additionally, and only a tiny peak
centered at 3.44 ppm in the NMR spectrum can be considered as an
impurity peak. The impurity peak was attributed to the CH3 group of
methanol. The peak height is less than 0.15% of DMMP OCH3 peaks,
suggesting that less than 0.15 mol % (<0.1 wt %) of methanol may be
present, the upper limit of possible impurity concentration. Further
checks for methanol interference were performed, as described in the
Results and Discussion.
have been used to form arrays with specific responses but are
limited by diffusion times within the polymers. Organic
semiconductors offer the opportunity to employ much more
facile deposition procedures and to covalently bind receptors
for analytes of interest at highly accessible regions within tens
of angstroms of the conductive path or “channel” in OFET
terminology. The binding chemistry is much better defined than
with inorganic semiconductors, and the binding position greatly
increases response speed relative to thicker polymer devices.
While OFET sensor work is ongoing at several institutions,
OFETs have not yet been designed for optimal binding ability
or film morphology for any particular analyte of interest, and
not specifically for nerve agent simulants that are of urgent
interest for security applications. The responses also have not
been correlated with theoretical predictions about sensitivity and
selectivity.
In this manuscript, we report initial studies of hole-transport-
ing phenylene-thiophene tetramers, with and without hydroxyl
functionalization, interacting with the weakly basic analyte
dimethyl methylphosphonate (DMMP) that simulates phospho-
nate nerve agents. The hydroxy group is the simplest receptor
that could be utilized for enhancing this interaction. We further
demonstrate the use of two-layer and mixed tetramer films for
the simultaneous optimization of sensitivity and response speed.
Experimental Section
Device Fabrication: Figure 1 is the schematic illustration of the
OFET “sensors”. Heavily doped silicon wafers with 300 nm of
thermally grown silicon dioxide served as substrates for all the devices.
The silicon dioxide layers on the corner of all devices were scratched
away to access the silicon as gate electrodes. Substrates were cleaned
by 20 min sonication in acetone followed by rinsing with 2-propanol.
Organic semiconductor thin films were deposited by vacuum evapora-
tion at a pressure < 10-5 mbar. The substrate temperature was kept
nominally at room temperature, rather than at elevated temperature, in
order to maximize grain boundaries.8,20 For the 6PTTP6 “device 1”,
50 nm of 6PTTP6 thin film were deposited. For a “blend” device, we
coevaporated the blend layers as follows: We used a thermal evaporator
with two source boats, with 6PTTP6 in one boat and HO6OPT in the
other one. The evaporation rates of each material were adjusted to be
1.2 Å/s before the shutter was opened. During evaporation, the total
evaporation rate of the two materials varied in the range 2.4 to 2.8
Å/s. With this method, two semiconductor materials can be evaporated
briefly at the same deposition rates. For the single layer blend (device
2), 6PTTP6 and HO6OPT were coevaporated and resulted in 50 nm of
film. For the two-layer blend (device 3), 35 nm of 6PTTP6 film were
evaporated, followed by the coevaporation of 15 nm of 6PTTP6 and
Materials: The semiconductor 5,5′-bis(4-n-hexyl-phenyl)-2,2′-
bithiophene (6PTTP6) was synthesized according to known pro-
cedures,19 and 5,5′-bis(4-hydroxyhexyloxyphenyl)-2,2′-bithiophene
(HO6OPT) was synthesized as shown in Scheme 1. To a solution of
15 g (82.8mmol) of 6-bromo-1-hexanol in 75 mL of DMF, 12.98 g
(75 mmol) of 4-bromo-phenol were added, followed by the addition
of 9 g (80.4 mmol) of potassium tert-butoxide. The reaction mixture
was stirred overnight at 70 °C in N2. After cooling to room temperature,
the mixture was poured into 200 mL of ether and washed with aqueous
NaOH (2%, 50 mL) and water (100 mL, 3 times). The organic layer
was dried over MgSO4 and evaporated in vacuo, yielding 15.5 g (56.7
mmol, 76%) of compound A as an oil. NMR (CDCl3) δ 1.4-1.5 (4H),
1.6 (2H), 1.8 (2H), 3.66 (t, 2H), 3.92 (t, 2H), 6.77 and 7.35 (ABq,
4H).
The reagent B 5,5′-bis(tri-n-butylstannyl)-2,2′-bithiophene was syn-
thesized as follows: 2.5 M n-BuLi in hexane (22 mL, 55mmol) was
(13) Chen, R. J.; Choi, H. C.; Bangsaruntip, S.; Yenilmez, E.; Tang, X.; Wang,
Q.; Chang, Y.-L.; Dai, H. J. J. Am. Chem. Soc. 2004, 126, 1563-1568.
(14) Lonergan, M. C.; Severin, E. J.; Doleman, B. J.; Beaber, S. A.; Grubb, R.
H.; Lewis, N. S. Chem. Mater. 1996, 8 (9), 2298-2312.
(15) Albert, K. J.; Lewis, N. S.; Schauer, C. L.; Sotzing, G. A.; Stitzel, S. E.;
Vaid, T. P.; Walt, D. R. Chem. ReV. 2000, 100 (7), 2595-2626.
(16) Gao, T.; Tillman, E. S.; Lewis, N. S. Abstracts of Papers of the American
Chemical Society 2004, 227, U504.
(19) Mushrush, M.; Facchetti, A.; Lefenfeld, M.; Katz, H. E.; Marks, T. J. J.
Am. Chem. Soc. 2003, 125 (31), 9414-9423.
(20) Someya, T.; Katz, H. E.; Gelperin, A.; Lovinger, A. J.; Dodabalapur, A.
Appl. Phys. Lett. 2002, 81 (16), 3079-3081.
(17) Gao, T.; Tillman, E. S.; Lewis, N. S. Chem. Mater. 2005, 17 (11), 2904-
2911.
(21) Guillaud, G.; Simon, J.; Germain, J. P. Coord. Chem. ReV. 1998, 178-
180, 1433-1484.
(18) Gao, T.; Woodka, M. D.; Brunschwig, B. S.; Lewis, N. S. Chem. Mater.
2006, 18 (22), 5193-5202.
(22) Horowitz, Gilles; Hajlaoui, M. E. AdV. Mater. 2000, 12 (14), 1046-1050.
9
J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007 9367

A R T I C L E S
Huang et al.
Figure 1. Schematic illustration of OFETs and semiconductor molecular structures: device 1, 6PTTP6; device 2, single-layer blend; device 3, two-layer
blend.
HO6OPT unless otherwise specified. Top-contact source and drain
electrodes were fabricated by thermal evaporation of gold through a
shadow mask with a spacing of L ) 270 µm and W ) 6.5 mm.
Measurement: As p-channel transistors, all the current-voltage (I-
V) curves of the devices were measured in the accumulation mode.
While the source was set to be common, the source-drain (Vd) was
swept between 0 and -80 V, the gate voltage was stepped from -20
V to -100 V with a -20 V step size unless described specifically,
and the saturation source-drain currents were recorded as the output
Figure 2. Scanning electron microscopic imagines of organic semiconduc-
signals of sensors. We did not start with zero gate voltage when we
tor films: (A) 6PTTP6; (B) blend of HO6OPT and 6PTTP6. Scale bars
denote 1 µm.
tested the sensor performance, because the source-drain current is too
small and the sensor performance is not very stable with Vg ) 0.
The sensors were tested in a 1 ft3 (28 L) chamber. For the
measurement of the sensor performance in DMMP vapor with
concentration varying over time, the chamber was initially filled with
air, and then a vial containing liquid DMMP was placed into the closed
chamber. The initial evaporation rate of DMMP was obtained by
determining the mass loss of the DMMP before and after the vial
containing DMMP liquid was kept in an open space for unit time. At
room temperature, the evaporation speed was measured to be 2.325
µmol/min, corresponding to a 1.8 ppm/min molar concentration change
in the 1 ft3 chamber. At room temperature, the saturation vapor
concentration of DMMP is 1620 ppm (mol/mol) at 21 °C.23 After an
hour of evaporation from the vial, the concentration of DMMP vapor
in the chamber should be no more than 60 × 1.8 ) 108 ppm (<7% of
the saturation concentration of DMMP vapor), considering the adsorp-
tion of DMMP on the chamber wall, and the decreasing evaporation
speed of DMMP due to the decreasing vapor pressure gradient between
the liquid surface and vapor phase in the chamber with time. Within 1
h, we can roughly consider the evaporation speed of DMMP to be
constant at 1.8 ppm/min and the DMMP vapor concentration to be
increasing linearly in the test chamber, since the change of the vapor
pressure gradient between the liquid surface and vapor phase in the
chamber is small.
For measurement at relatively constant analyte vapor concentration,
the test chamber was connected to a constant flow rate of gas, which
was switched between N2 and analyte vapors. Bubblers containing
analyte liquids were used as sources of analyte vapors. N2 gas flowed
above the liquid surface and delivered certain amounts of analyte vapor
to the test chamber. The mass of the bubblers was measured before
and after testing to determine the total mass of analyte delivered by a
specific volume of carrier gas and, thereby, calculate the average
concentration of analyte vapors in the test chamber. The sensors were
placed far away from the outlet of the analyte vapor to avoid direct
flush, so that the analyte vapors were delivered gently to the sample,
except as noted later. The setup was calibrated using a handheld
chemical detector (ProEngin, model: AP4C), which analyzes for
phosphorus atoms in gas-phase molecules using flame photometric
detection. The dynamic range of the AP4C is 24 ppb ≈ 13 ppm of
DMMP. We generated diluted DMMP vapor by mixing an N2 stream
with N2 blown over a DMMP source and measured the DMMP
concentration with the AP4C. The DMMP concentration of the vapor
generated by our setup was calculated by the rates of mass loss in
bubblers and gas flow into the chamber to be 7.3 ppm, while 7.02
ppm was the reading obtained using the AP4C. During our device tests,
150 and 20 ppm DMMP vapors were generated by the same dilution
method.
Where devices were exposed to air, the relative humidity was 20-
40% and did not change materially during the course of any experiment.
Results and Discussion
Microstructure Characterization: Various factors such as
film composition, OSC molecular structures, degrees of crystal-
linity, surface morphologies, and grain boundaries can influence
the sensing performance of organic thin films.21 We investigated
the microstructures of the organic semiconductor films in order
to understand if the variations in the performance of devices
are due to the film compositions or differences in film
microstructures.
Figure 2 shows the scanning electron microscopic images
(JEOL 6700F) of the pure 6PTTP6 film and the two-layer blend
of HO6OPT-6PTTP6 evaporated on SiO2/Si substrates at room
temperature. Both films show platelike structures, with the
topmost layer of plates partially standing perpendicular to the
surfaces. For the sensor application, surface structures like these
films are desired because of their larger surface area and more
adsorption sites compared to films with large bulk grains. The
densities of the upright plates of two films are different, but
(23) Kyle, C.; Kulkarni, R. D.; Kozlov, M.; Manohar, S. K. Nanotechnology
2006, 17, 4123-4128.
9
9368 J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007

OFETs for Phosphonate Vapor Detection
A R T I C L E S
the typical output currents. For most of the sweep, the leakage
current is a few tenths of a nA, of similar order as the charging
current: 10-8 F/cm2 oxide capacitance times 0.015 cm2 channel
area times 10 V/s gate sweep rate. The mobility of the single-
layer blend OFET is apparently much lower than that of 6PTTP6
and two-layer blend OFETs.
Performance of Nonfunctionalized Devices: 5,5′-Bis(4-n-
hexyl-phenyl)-2,2′-bithiophene (6PTTP6) Field Effect Tran-
sistor. The nonfunctionalized 6PTTP6 film is inherently sen-
sitive to DMMP. Figure 5a shows the response of the saturation
source-drain current of a 6PTTP6 OFET (device 1) as a
function of time at various gate voltages during DMMP
exposure. Device 1 was tested in a chamber filled with air. The
black solid line indicates the time when a vial with DMMP
liquid was placed into the closed chamber. The saturation
source-drain current Id decreased significantly during exposure
to DMMP. We also observed that the current drifts have an
obvious gate voltage dependence. In Figure 5b, saturation
source-drain currents with different gate voltages were normal-
ized by the value at time ) 0 min and plotted as a function of
time. The saturation source-drain current reduced to less than
5% of its original value with the gate voltage set to -20 V,
corresponding to a large relative response, while it only reduced
to around 30% with Vg set to -100 V.
The saturation source-drain current of this unfunctionalized
sensor reached the minimum value after 40 min of exposure to
DMMP vapor, during which time the DMMP concentration in
the chamber was considered to have increased linearly to 75
ppm. The kinetics of the response are controlled by the
evaporation rate of DMMP and slow diffusion rate of DMMP
within the OSC film.24,25 The adsorption of DMMP molecules
on the unfunctionalized semiconductor organic film surface is
probably a weak electrostatic force of the van der Waals type.26
Figure 3. X-ray diffraction patterns of the pure 6PTTP6 film, single-layer
blend film, and two-layer blend film.
both films are organized similarly by very thin layers stacking
together, which is a typical structure for oligomer semiconduc-
tors.22
X-ray diffraction measurements (Philips X Pert Pro) were
performed along the surface normal axis for all devices to
understand the degree of crystallization of the semiconductor
films. As shown in Figure 3, the pure 6PTTP6 film (device 1),
single layer blend film (device 2), and two-layer blend film
(device 3) exhibit essentially identical XRD patterns and
d-spacings, indicating the high similarity of the crystallinities
between pure 6PTTP6 and the blend films.
Current Voltage Characteristics of OFET Sensors: The
Id-Vd characteristics of all three OFET sensors are shown in
Figure 4. All of the three OEFT sensors exhibited obvious linear
and saturation regions and showed negligible gate leakage
current at Vd ) 0. The maximum leakage current for a two-
layer blend was 2 nA, at least 3 orders of magnitude lower than
Figure 4. Id-Vd characteristics of (A) 6PTTP6 OFET; (B) Single-layer blend OFET; (C) Two-layer blend OFET. Also, (D) leakage current for two-layer
blend with source and drain grounded and gate swept from 0 to -100 V.
9
J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007 9369

A R T I C L E S
Huang et al.
Figure 6. Drain current Id vs drain voltage Vd curves at various gate voltage
Vg of a HO6OPT field-effect transistor.
Figure 5. Responses of the saturation drain current of 6PTTP6 sensor as
a function of time at various gate voltages during DMMP exposure. The
black solid vertical line indicates the time when the device was exposed to
Figure 7. Response of the single-layer blend on exposure to DMMP vapor.
The black solid vertical line indicates the time when a vial with DMMP
liquid was placed into the closed chamber, and the dash vertical line indicates
the time when the vial was removed and the chamber was open to air.
DMMP: (A) saturation source-drain current; (B) relative response: Id-sat
/
Id-sat 0
.
Performance of Functionalized Devices: Organic Sensors
Based on Blends of HO6OPT and 6PTTP6. In order to
enhance the sensitivity and response speed of the sensor, we
synthesized a functionalized organic semiconductor material,
5,5′-bis(4-hydroxyhexyloxyphenyl)-2,2′-bithiophene (HO6OPT),
as receptor material. HO6OPT has a similar molecular structure
to 6PTTP6; however it has OH groups on the end of the
molecule. The OH group should form hydrogen bonds with
DMMP, leading to increased adsorption and electronic effects.
Figure 6 shows the typical Id-Vd curve of a field effect
transistor with HO6OPT as an active layer. The gate leakage
current was due to the use of a large substrate without gate-
electrode patterning. The field effect mobility of this device is
much lower and less precisely determined than that of 6PTTP6,
but it verified the semiconductor activity of the HO6OPT
compound. For the sensor application, it is more desirable to
have a higher current, so that the HO6OPT FET was not
optimized further and not employed as a sensor.
bottom layer is 35 nm of 6PTTP6 serving as charge transport
layer, and the top layer is 15 nm of the blend of two materials,
HO6OPT and 6PTTP6, which can provide a moderate distribu-
tion of OH groups as receptor sites. In order to better understand
this blend structure, we also fabricated the single layer blend
field effect transistor (device 2: Figure 1), which has a 50 nm
single layer of the blend of HO6OPT and 6PTTP6 as the active
layer.
As shown in Figure 4b, the single-layer blend OFET exhibited
good linear and saturation regions. Figure 7 represents the
response of the single layer blend (device 2) on exposure to
DMMP vapor in air under ambient conditions. The DMMP
vapor was again generated by placing the vial containing DMMP
liquid into the test chamber. The sensing measurements have
clearly indicated that the sensor with the blend film has a shorter
response time than that of the 6PTTP6 OFET. The saturation
source-drain current of this functionalized sensor decreased
dramatically right after the vial was placed into the chamber,
and it reached the minimum value after about 10 min of
exposure to DMMP vapor. The saturation current of this device
compared to device 1 is still 1 order of magnitude lower, which
is a disadvantage for the sensor application. We therefore
employed the two-layer blend structure of device 3.
Instead of the single component HO6OPT FET, we developed
a heterostructure (device 3; Figure 1), within which there are
two layers of organic semiconductors on the substrate: the
(24) Zimmermann, C.; Rebie`re, D.; De´jous, C.; Pistre´, J.; Chastaing, E.; Planade,
R. Sens. Actuators, B 2001, 76, 86-94.
(25) Zhou, R.; Josse, F.; Go¨pel, W.; O¨ ztu¨rk, Z. Z.; Bekarogˇlu, O¨ . Appl.
Organomet. Chem. 1996, 10 (8), 557-577.
The sensing response of the two-layer blend device 3 to
DMMP vapor in air under ambient conditions is shown in Figure
(26) Bouvet, M. Anal. Bioanal. Chem. 2006, 384, 366-373.
9
9370 J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007

OFETs for Phosphonate Vapor Detection
A R T I C L E S
Figure 9. Sensing responses of the two-layer blend and 6PTTP6 itself to
DMMP vapor generated by a vial containing DMMP liquid. The inset shows
the log plot of the same curves.
transport along the conducting channel, and the saturation
source-drain current of this two-layer blend should be com-
parable to that of device 1. Therefore, the response of the
heterostructure is probably due to the trapping or hindering of
charge carriers digressing from the channel or from a small
amount of the top layer that penetrates to the channel between
grains, where DMMP adsorbs.
Figure 9 compares the responses of the ratio of the normalized
saturation source-drain current Id/Id0 of device 1 and device 3
as a function of time during DMMP exposure, with Vg set to
be -40 V. The blank vertical line indicates the time when a
vial with DMMP liquid was placed into the closed chamber.
Comparing the two plots in Figure 9, we observed that device
3 has a much stronger and faster response than that from device
1 especially during the first period of the evaporation of DMMP
vapor from the vial, when the concentration of DMMP vapor
is presumably low.
Figure 8. Sensing response of the two-layer blend sensor to DMMP vapor.
The black solid vertical line indicates the time when a vial with DMMP
liquid was placed into the closed chamber, and the dash vertical line indicates
the time when the vial was removed and the chamber was open to air. (A)
The saturation source-drain current versus time; (B) the relative response
of the drain current versus time.
8 for various gate voltages. Again, the black solid vertical line
indicates the time when a vial with DMMP liquid was placed
into the closed chamber filled with air, while the dashed vertical
line shows the time we remove the vial and open the chamber
to air. We cycled these 3 times to show both the response and
subsequent recovery of the current. The absolute current is 1
order of magnitude higher than that of device 2, while the
variation of the saturation source-drain current versus time was
found to be similar to that for the single layer blend sensor
(device 2). The high current can be attributed to the bottom
layer of the high mobility semiconductor: 6PTTP6, which
provides charge transport. Studies27,28 have shown that, in the
accumulation regime of evaporated oligomer film, most mobile
charge carriers (more than 80% at high gate voltage)29 are
concentrated within the first two semiconductor monolayers at
the dielectric-semiconductor interface. XRD shows that the
monolayer spacing of 6PTTP6 is 3 nm, suggesting that the
effective thickness of conducting channel is around 6 nm for
this type of transistor. The film thickness of the bottom 6PTTP6
layer of this device is 35 nm, which is much thicker than the
effective thickness of the conducting channel. As a result, the
top blend layer should not practically influence the charge
Figure 10a shows the two-layer blend sensor responses to
150 ppm DMMP vapor. We vent the chamber with N2 and then
delivered 150 ppm DMMP with N2 for 6 min. The shadowed
areas in the plot indicate when the DMMP vapor is flowing
above the sensor. The sensor was heated at 45 °C for 2 min in
air and cooled down in N2 before test. The saturation drain
current decreased significantly when the DMMP vapor was
injected into the chamber. The observed response times are
comparable to many for reported sensors.10,30,31
In practical a sensing operation, heating the sensor at 45 °C
for 2 min is sufficient to fully recover the sensor. Figure 10
shows the reproducible responses and recovery of two-layer
blend sensors upon exposure to 150 ppm DMMP vapor. The
sensor was heated at 45 °C for 2 min in air and cooled down in
N2 between every measurement. Figure 10b was obtained from
a fresh two-layer blend FET, while Figure 10c shows the
response of a used sensor that has been tested many times
previously. The two sensors exhibited similar responses. The
same sensing tests were also performed with other two-layer
blend OFETs fabricated 6 months earlier, and similar results
were obtained.
(27) Horowitz, G.; Hajlaoui, R.; Bourguiga, R.; Hajlaoui, M. Synth. Met. 1999,
101, 401-404.
(28) Horowitz, G. J. Mater. Res. 2004, 19 (7), 1946-1962.
(29) Horowitz, G. Organic Transistors. In Organic Electronics; Klauk, H., Ed.;
Wiley-VCH: Weinheim, Germany, 2006; pp 3-32.
(30) Tomchenko, A. A.; Harmer, G. P.; Marquis, B. T. Sens. Actuators, B 2005,
108, 41-55.
(31) Patel, S. V.; Mlsna, T. E.; Fruhberger, B.; Klaassen, E.; Cemalovic, S.;
Baselt, D. R. Sens. Actuators, B 2003, 96, 541-553.
9
J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007 9371

A R T I C L E S
Huang et al.
Figure 11. Absolute current response of “thick” (top) and “thin” (bottom)
bilayer devices to 150 ppm of DMMP delivered via a hose 1 cm from the
device surface beginning at the left edge of the gray region of the plot.
Figure 10. Reversible responses of two-layer blend films upon exposure
to 150 ppm DMMP vapor. (A) Device preheated; (B) fresh device; (C)
used device.
We checked to see if the response times were determined by
convection through the large sample chamber or by diffusion
through the thickness of the two-layer film. Figure 11a shows
the response of 35 nm of 6PTTP6 with 15 nm of the blend on
top of 150 ppm DMMP, delivered through a hose located about
1 cm from the device surface. The time scale is similar to that
observed in Figure 10. On the other hand, the response speed
is much faster for a thinner bilayer. Figure 11b shows the
response of 6 nm of 6PTTP6 with 3 nm of the bilayer on top
under the same conditions. The time scale has decreased by an
order of magnitude. Thus, the response time is limited in the
thick devices by the time for vapor molecules to diffuse to the
positions where they have the maximum electronic effect. The
use of thin devices should result in adequate response times
for application. The maximum gate leakage current for the thin
device was 0.06 nA.
Figure 12. Mobility and threshold voltage response to DMMP vapor.
field effect mobility µsat and threshold voltage Vt can also be
used as output signals of OFET sensors.
In order to verify that the change of the drain current of the
two-layer blend sensor mainly resulted from DMMP, not from
the possible trace methanol impurity, we also did two control
experiments: A mixture of 1% of methanol and 99% of DMMP
and a DMMP liquid sample that had been bubbled with N2 for
30 min to remove possible volatile impurities were each used
as analyte sources. The responses of the two-layer blend sensor
to these two analytes were observed to be similar to that of
fresh DMMP, suggesting that the responses of the sensor we
observed are actually resulting from DMMP vapor.
From the same experiment shown in the middle curve of
Figure 10b, field effect mobility µsat and threshold voltage Vt
were extracted as shown in Figure 12. The shadowed area
indicates the time when DMMP vapor was delivered. When
the transistors are exposed to DMMP vapors, the change of
saturation drain current can be attributed to the variation of both
field effect mobility µsat and threshold voltage Vt. Hence the
9
9372 J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007

OFETs for Phosphonate Vapor Detection
A R T I C L E S
Figure 14. Response of the two-layer blend sensor to different analyte
vapors: (A) 200 ppm 2-octanone; (B) 400 ppm butyl butyrate; (C) 5500
ppm acetic acid; (D) 150 ppm DMMP.
interference analytes with similar polarities or vapor pressures
as those of DMMP and compared responses to that of DMMP
vapor. 2-Octanone, butyl butyrate, and acetic acid were chosen
to represent different groups of potential interference compounds
containing double bonded oxygen atoms. Figure 14 shows the
responses of the sensor to different analytes. The natural
logarithm of the ratio of normalized drain currents ln(Id/Id0) was
potted versus time. The shadowed areas indicate when the sensor
was exposed to 200 ppm 2-octanone, 400 ppm butyl butyrate,
5500 ppm acetic acid, and 150 ppm DMMP, respectively. We
used higher concentrations of interference analytes than that of
DMMP, to make the most conservative judgments about
selectivity. The response of the two-layer blend sensor to
DMMP is clearly much stronger than that to other interference
vapors.
The responses of the two-layer blend sensor to dibutylamine
(a secondary amine shaped like some mustard gases but
somewhat more basic) and to 110 ppm of diisopropyl meth-
ylphosphonate (DIMP) were also studied with the same setup.
As shown in Figure 15, the sensor exhibited a similar response
to DIMP as that to DMMP vapor, suggesting that the sensor is
sensitive to phosphonate compounds in general, not only to
DMMP.
When dibutylamine was delivered into the test chamber, the
OFET sensor was irreversibly quenched. In the practical
application of sensors for weak bases, strongly basic interferents
like amines will need to be filtered out in the sensor front end
to avoid permanent quenching of the sensor.
Finally, we tested the same kind of two-layer device for
humidity responses. Figure 16 shows the change in current as
the relative humidity of air in the chamber is raised from 33 to
55%. The change is linear and considerably lower in magnitude
than changes in response to DMMP. As a practical matter, any
environmental sensor would be corrected for humidity effects
by the use of membranes to filter water vapor at the front end,
Figure 13. Two-layer blend response to 20 ppm DMMP vapor. (A) Id-sat
/
d-sat,0 versus time. The shadowed areas indicate the time when the DMMP
vapor is flowing above the sensor; (B) the saturation drain current before
and after 10 min exposure to DMMP.
We have also obtained preliminary results of sensing DMMP
vapor on the scale of 20 ppm. This time, the bubbler containing
DMMP liquid was used as the source of DMMP vapor. N2 gas
flowed above the liquid surface at the rate of 4 L/min and
delivered certain amounts of analyte vapor to a flask. The
DMMP vapor was then further diluted by mixing with N2
flowing into the flask at a rate of 28 L/min. The combined gases
were directed into the test chamber. The mass of the bubbler
was measured before and after testing, showing that the average
concentration of DMMP vapors in the test chamber was 20 ppm.
The two-layer film of 35 nm 6PTTP6 with 15 nm of the blend
on top was preheated at 45 °C for 2 min followed by 2 min of
cooling in N2. The device was then tested in the 1 ft3 chamber
and showed an obvious response, though not as strong as that
shown for higher concentrations at analogous voltages in Figures
8 and 10. As shown in Figure 13a, the drain current kept
decreasing during the exposure time. In a second experiment,
the drain voltage was fixed at -80V, and the gate voltage was
swept from 20 to -80 V, with a -20 V step size. Figure 13b
compares the saturation drain current of the device before and
after 10 min of exposure to the 20 ppm DMMP vapor. The
result suggests that sub-ppm sensing is achievable, though rates
may be limited by diffusion through the films.31,32
(32) Ballantine, D. S.; White, R. M.; Martin, S. J.; Ricco, A. J.; Frye, G. C.;
Zellers, E. T.; Wohltjen, H. Acoustic WaVe Sensors: Theory, Design and
Physicochemical Application; Academic Press: Boston, MA, 1997; Chap-
ters 4 and 6.
Selectivity is a critical issue for chemical sensors. We
investigated responses of the two-layer blend to a set of potential
9
J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007 9373

A R T I C L E S
Huang et al.
Figure 17. (A) A schematic diagram of an OFET based chemical sensor.
(B) Dipole induced electric field at various distances in the direction parallel
to the direction of the dipole moment.
Figure 15. Two-layer blend FET response to 110 ppm DIMP.
is on the order of 108 V/m, and the longitudinal source-drain
field E(SD) used to drive current through the channel is on the
order of 106 V/m, with our device configuration. The DMMP
molecule has a highly polarized PdO bond. Previous studies
reported that the DMMP molecule has a dipole moment µ )
3-3.62 D, depending on the method of estimation and the
conformation of the molecule.10,34 In this paper, we use µ ) 3
D to get a conservative estimate of the dipole effect between
DMMP and sensors.
The adsorption and diffusion of DMMP molecules into the
semiconductor layer can induce strong electric fields in the
immediate vicinity of the molecules (not in the entire channel!),
which not only cause a portion of the mobile charges to be
trapped and lose their activity but also effectively slow down
the movement of charges that are still transported.35 This action
is probably mostly at the channel, resulting from vapor
molecules diffusing amid the grains to approach the dielectric
interface, or otherwise away from the channel, where charge
carriers have a small but finite probability of drifting, both cases
illustrated in Figure 17a.
Figure 16. Current as a function of relative humidity for the bilayer device,
with drain voltage held at -80 V. The experiment was performed over the
course of 0.5 h.
and/or by employing a separate sensor to track humidity
separately and correct the primary sensor for humidity effects.
Even more likely, a larger sensor array would be employed to
produce a specific response pattern to an analyte of interest.
Sensing Mechanism: The responses of the OFET based
sensors are consistent with electrostatic modeling that we have
done.33 Figure 17a shows a schematic diagram of an OFET
based chemical sensor and adsorbing DMMP molecules. E(g),
the transverse source-gate field used to induce mobile charges,
The electric field induced by a dipole with a dipole moment
of bp can be calculated by the equation
3K br (bp ‚ br)
K bp
BE )
-
r5
r3
Here K ) 9 × 109 N‚m2/C2 (Coulomb’s constant), and br is the
distance between the dipole and the position of concern.
(33) Katz, H. E.; Huang, J. Organic Semiconductor-based Chemical Sensors.
In Organic Electronics; Klauk, H., Ed.; Wiley-VCH: Weinheim, Germany,
2006; pp 411-421.
(34) Vishnyakov, A.; Neimark, A. V. J. Phys. Chem. A 2004, 108, 1435-1439.
(35) Tanese, M. C.; Fine, D.; Dodabalapur, A.; Torsi, L. Biosens. Bioelectron.
2005, 21, 782-788.
9
9374 J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007

OFETs for Phosphonate Vapor Detection
A R T I C L E S
Figure 17b plots the electric fields induced by a DMMP
molecule in the direction parallel to the dipole moment as a
function of the distance between DMMP and the position of
concern. We noticed that the magnitude of the dipole induced
field is comparable to E(g) and is much larger than E(SD), which
means the dipoles of DMMP molecules are certainly capable
of trapping mobile charges and significantly diminishing OFET
current by changing both the threshold voltage Vt and the field
effective mobility µsat.
Under the idealized assumptions including mobility indepen-
dent of gate voltage and gradual channel approximation ((E(g)
. E(SD) in the present case), the saturation drain current Isat
of the device is given by eq 136
Cµsat 1/2(Vg - Vt)
(1)
W
2L
1/2
Isat
)
[
]
When DMMP molecules are attached to semiconductors,
dipoles induce changes in µsat and Vt, noted as ∆µ and ∆Vt, so
that the saturation drain current Isat(DMMP) of the device with
DMMP becomes eq 2:
C(µsat + ∆µ) 1/2[Vg - (Vt + ∆Vt)] (2)
W
2L
1/2
Isat(DMMP)
)
[
]
Combining eqs 1 and 2, we can get the relationship between
Isat(DMMP) and Isat, shown as eq 3:
2
∆Vt
∆µ
µsat
I
sat(DMMP)/Isat ) 1 +
1 -
(3)
(
)
[
(
)
]
Vg - Vt
This equation is only valid when |∆Vt| < |Vg - Vt|, but it
expresses the gate voltage dependence of the sensor responses.
The value of Vg, Vt, and ∆Vt are all negative. For smaller values
of |Vg|, such as 40 V, the [1 - ∆Vt/(Vg - Vt)] term is smaller,
which results in a smaller value of the ratio of the saturation
source-drain currents Isat(DMMP)/Isat0, corresponding to a more
noticeable sensor response. On the contrary, if |Vg| is larger,
such as 100 V, the term [1 - ∆Vt/(Vg - Vt)] is larger than that
of the previous case, which results in a relatively larger value
of Isat(DMMP)/Isat0 and a smaller sensor response.
Figure 18. (A) Square root of the saturation drain current of a 6PTTP6
FET versus gate voltage before and during DMMP vapor exposure. (B)
The closed squares correspond to the ratio of saturation source-drain current
Isat(DMMP)/Isat0 obtained from experiment; the solid line is the least-squares
fitted curve according to eq 3.
Table 1. Field-Effect Mobility µsat and Threshold Voltage Vt of
6PTTP6 FET before and after Exposure to DMMP Vapor
Vt
µsat
(cm2/V
∆
(V)
Vt
∆µ
(cm2/V
‚s)
(V)
‚
s)
no DMMP
in DMMP
-5
0.0522
0.0245
Figure 18a plots the square root of the saturation drain current
of a 6PTTP6 FET versus gate voltages before and during
DMMP vapor exposure. This time, Vd was set to be -100 V
and Vg was swept from 20 to -100 V, with a -5 V step size.
We can see that at lower |Vg|, the current ratio is larger. We
can calculate the field effect mobility µ and threshold voltage
Vt of the device before and after exposure to DMMP vapor by
the slope of the curves and the intersection of the linear fitted
curves with the X axis, as shown in Table 1. In Figure 18b, the
closed squares correspond to the ratio of the saturation source-
drain current Isat(DMMP)/Isat0 obtained from experiment. We
performed a least-squares fit of those data, together with µ and
Vt obtained from Figure 18a, to eq 3, which gives us an estimate
of the value of ∆Vt and ∆µ. The estimated ∆Vt ) -15.3 V and
the estimated ∆µ ) -0.0283 cm2/VS are very close to the
values we obtained from experimental data, -16 V and -0.0277
cm2/VS, respectively. This estimation is qualitative since the
field effect mobility µ is actually also a function of Vg, and
therefore the dependence of µ on Vg also contributes to the gate
-16
-0.0277
-21
voltage dependence of the sensor responses. At high |Vg| and
|Vd|, the field effect mobility tends to saturate, and its gate
voltage dependence is not that significant. However, at low gate
voltage, µ increases linearly with Vg,36 which explains the
inconsistency between the experimental data and the fitted curve
when |Vg| is less than 25 V. When |Vg| is larger than 90 V, the
contact resistances cause much of the inconsistency between
experimental and estimated data. Equation 3 is thus useful for
estimating the voltage range over which a single sensing
mechanism applies, and also for predicting responses at voltages
not specifically included in a particular experimental dataset.
Conclusions
This study shows that simple organic field effect transistors
respond to DMMP vapor thorugh changes in mobility, threshold
voltage, and source-drain current. The devices exhibited a
greater response to DMMP than to other interferent analytes
we tested. We demonstrate for the first time that a device based
(36) Horowitz, G.; Hajlaoui, M. E.; Hajlaoui, R. J. Appl. Phys. 2000, 87 (9),
4456-4463.
9
J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007 9375

A R T I C L E S
Huang et al.
on a two-layer blend OFET has a higher sensitivity and shorter
response time compared to the case of an unfunctionalized pure
semiconductor serving as the active layer. All the responses have
an obvious gate voltage dependence. Higher absolute current
responses were observed at higher gate voltages, while the
optimal gate voltages for the greatest relative responses were
determined to be -20 V with our measurement. The two-layer
blend has a significant response to the tens of ppm level of
DMMP vapor. At optimal gate voltage, drain current decreased
to less than 1% of its original value in a flow of diluted DMMP
vapor, suggesting the capability of sensing ppb levels of
analytes. The response times of the sensors can be shortened
by reducing the film thickness. We plan to increase the Lewis
and/or proton acidities of the surface groups on the OFETs, to
increase the likelihood of effective adsorption of the analytes
of actual interest, such as nerve gases, which are likely to be
less basic than DMMP.
Acknowledgment. This work was performed with primary
support from the National Science Foundation SENSORS
program (Award No. 0528472). We also acknowledge support
from The Johns Hopkins University Applied Physics Laboratory
Partnership program. Helpful discussions with Professor Andreas
Andreou are greatly appreciated.
JA068964Z
9
9376 J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007