ARTICLES

A sensitivity-enhanced field-effectchiral sensorLUISA TORSI1,2*, GIANLUCA M. FARINOLA1, FRANCESCO MARINELLI1, M. CRISTINA TANESE1,OMAR HASSAN OMAR3, LUDOVICO VALLI4, FRANCESCO BABUDRI1,3, FRANCESCO PALMISANO1,2,P. GIORGIO ZAMBONIN1,2 AND FRANCESCO NASO1,3*1Dipartimento di Chimica, Universita degli Studi di Bari, 70126, Bari, Italy2Centro di Eccellenza TIRES, Universita degli Studi di Bari, 70126, Bari, Italy3CNR ICCOM Bari, Dipartimento di Chimica, Universita degli Studi di Bari, 70126, Bari, Italy4Dipartimento di Ingegneria dell’Innovazione, Universita degli Studi di Lecce, 73100, Lecce, Italy*e-mail: torsi@chimica.uniba.it; f.naso@chimica.uniba.it

Published online: 20 April 2008; doi:10.1038/nmat2167

Organic thin-film transistor sensors have been recently attracting the attention of the plastic electronics community for theirpotential exploitation in novel sensing platforms. Specificity and sensitivity are however still open issues: in this respect chiraldiscrimination—being a scientific and technological achievement in itself—is indeed one of the most challenging sensor bench-tests.So far, conducting-polymer solid-state chiral detection has been carried out at part-per-thousand concentration levels. Here, a novelchiral bilayer organic thin-film transistor gas sensor—comprising an outermost layer with built-in enantioselective properties—isdemonstrated to show field-effect amplified sensitivity that enables differential detection of optical isomers in the tens-of-parts-per-million concentration range. The ad-hoc-designed organic semiconductor endowed with chiral side groups, the bilayer structure andthe thin-film transistor transducer provide a significant step forward in the development of a high-performance and versatile sensingplatform compatible with flexible organic electronic technologies.

Organic thin-film transistors (OTFTs) have been proposedas high-performance sensors1–5 with interesting perspectivesas sensing elements in plastic electronic arrays6–8. Althoughconducting polymers (CPs) have been implemented in differentchemical sensors9, the use of polycrystalline CPs, acting both astransistor channel materials and as sensitive layers, is peculiaronly to OTFTs. In this configuration, highly repeatable responseswere measured by properly gate-biasing the device1–3, whereasbroad chemical selectivity was conferred either by side groups,covalently bound to the CP’s backbone10, or by means ofCP blends11. Specificity and sensitivity are however still openissues. CP-based solid-state chiral detection12–15 has been carriedout so far at part-per-thousand concentration levels. Chiraldiscrimination is a challenging scientific and technological researchfield with huge potential for applications in key areas such aschemical synthesis, catalysis, pharmaceutics and biomedicals. Thechallenges arise from the fact that the two enantiomers havemirror-image molecular structures and show physical–chemicalproperties that can change differently, exclusively on interactionwith another chiral entity, most commonly a molecular receptor.Biological and physiological effects of such enantio-specificinteractions can be dramatically different. Currently, opticalisomer separation and detection is mostly carried out usingoff-line analytical techniques such as chromatography. Chiralsolid-state gas sensors hold the potential to carry out on-lineenantiomer discrimination, with exciting perspectives in processmonitoring. In a seminal work, thickness share-mode resonatorand reflectometric interference spectroscopy solid-state transducerswere proven to achieve enantiomeric recognition by using

standard chromatographic column materials as active layers16,17.Their implementation is foreseen in silicon-based single-chipgas-sensing microtransducers18.

Here, a new bilayer OTFT sensing technology is presented ashighly versatile and capable of an elevated level of performance.A field-induced sensitivity enhancement is reported, eventuallyleading to chiral differential detection at an unprecedentedly lowconcentration of terpene flavour molecules of biological origin.A liquid- or gas-sensing platform based on OTFTs can also takeadvantage of the rapid developments occurring in the field of plasticelectronics, where OTFTs have been successfully implemented inflexible circuits and also for sensing purposes19. Downsizing tothe nanoscale is a key driver as well, and OTFT nanosensors forgaseous20,21 and liquid4 analytes have been demonstrated. Theircompatibility with microfluidics has also been proven22.

Two different conjugated oligomers are implemented ina bilayer sensing OTFT able to combine field-effect andchiral-recognition properties. The two layers consist of similaralkoxyphenylene–thiophene oligomers, ad hoc designed andsynthesized, differing just in the side group structure, and namedPTO and PTA or PTG in Fig. 1. The alkoxy-substituted phenylenering acts as a general and convenient structural unit, which enablesthe introduction of a wide variety of covalently bound substituents,including biological molecules, to conjugated oligomer andpolymer backbones, via general synthetic strategies23,24. The-phenylalanine amino acid and the β--glucosidic substituents,deputed to the chiral recognition in PTA and PTG respectively, werechosen as examples of easily available and cheap chiral materialsfrom biological sources. The bilayer structure was adopted as the

412 nature materials VOL 7 MAY 2008 www.nature.com/naturematerials

© 2008 Nature Publishing Group

ARTICLES

O

O

S SS S

O

O

O HHN

O

H NH

tBOC

tBOC

O

O

S SS S

AcOAcO

O

OAc

OAcO

O

OAcOAc

OAcO

AcO

S SS S

L-phenylalanine

PTA

PTA or PTG

G

SiO2

Si

S D

PTO

D-glucose

PTG

-citronellol (R)-(+)/(S)-(–)or carvone (R)-(–)/(S)-(+)β

Figure 1 Bilayer OTFT chiral sensor structure. The transistor has a bottom-gate device structure that consists of a highly n-doped silicon wafer (resistivity 0.02–1� cm−1)coated by a 300-nm-thick SiO2 thermal oxide (capacitance per unit area C i = 10 nF cm2). Each sample was fabricated on a cleaved wafer piece of about 1.5 cm2. The siliconsubstrate with a gold pad is the gate (G) contact and the silicon dioxide is the gate dielectric. To improve device performance the SiO2 surface was alkyl functionalized bytreatment in 1,1,1,3,3,3-hexamethyldisilazane saturated vapours for 24 h. The dielectric surface was covered by the stacking bilayer composed of ten Langmuir–Schafer (LS)transfers of PTO and five of PTA, resulting in a total thickness of about 280 nm. An analogous procedure was adopted for the PTO–PTG bilayer (the total thickness was about360 nm in this case). A single PTO layer, constituted of 15 LS transfers as well (thickness about 170 nm), was used as sensing layer in an achiral OTFT reference device.A series of gold source (S) and drain (D) contacts was defined, by thermal evaporation through a shadow mask, directly on the sensing (single- or bilayer) films. The areabetween each pair of closest pads, called the channel region, was 200 µm long (L) and 4mm wide (W ). The analyte, in a nitrogen stream at a controlled concentration, wasdelivered through a nozzle (positioned at a fixed distance of a few millimetres from the device channel surface) directly onto the active layer surface. The sensors werefabricated and measured in a standard laboratory environment and operated at room temperature.

sole PTA or PTG OTFTs did not show field-effect amplified current,most probably because of the steric hindrance of the amino-acid orthe glucose units. This is a general issue to be faced when developingTFT-based bio-sensing devices, especially if receptors are covalentlybound to the CP main chain. The bilayer structure, here proposedfor the first time in OTFT sensors, is a suitable way around it.

The PTO–PTA-bilayer OTFT drain–source-current/gate–source-voltage, Ids–Vg, transfer characteristics are reported inFig. 2, and typical device Ids–Vds characteristics are reported inthe Supplementary Information. The solid black curve in Fig. 2ashows the current flowing in the channel region between the sourceand the drain contacts at a fixed drain–source bias (Vds = −50 V)while sweeping Vg between 20 and −100 V. Much lower biasingpotentials (down to few volts) can be used to induce the samecharge by just lowering the gate dielectric thickness to a fewtens of nanometres1,2. As alkoxyphenylene–thiophene conjugatedsystems are p-type semiconductors25, negative values of Vds andVg(|Vg| > |Vt|) drive the device in the on state (accumulationmode), whereas Vg values below Vt (OTFT threshold voltage, thedevice turn-on gate bias) generate a regime of charge depletion1

(off state or depletion mode). An OTFT operated in the off statecan in principle be expected to behave as an equivalent resistor.The field-effect mobility (µ) and Vt are extracted from the relevantsquare-root(Ids) versus Vg plot. A typical example is provided inthe Supplementary Information file. The mobility, averaged overmore than 30 devices, is in the (1.6 ± 0.5) × 10−4 cm2 V−1 s−1

range, whereas the Ids dynamic range (amplification) extendsover three orders of magnitude at most. Vt is, generally, below−20 V. The figures of merit are lower than state-of-the-artOTFT ones most likely because of the twisted nature of thealkoxyphenylene–thiophene backbone25. Although bio-substituted

CP structures, with improved charge delocalization properties(and eventually OTFT performance level), could be designed, atrade-off between effective conjugation length and bio-recognitioncapabilities must invariably be faced.

The transfer characteristics measured in N2 and in differentcitronellol or carvone atmospheres were used to derive thesensor responses. This is a convenient procedure, alreadyadopted in previous studies4,26, that enables us to operate OTFTsensors from the depletion to the accumulation mode whileexposed to different analyte concentrations. From these curvesmultiparametric information can be extracted26. Besides, becausethe device is driven in the depletion regimes before eachmeasurement run, response repeatability is expected to improve.The dark-blue curve in Fig. 2 is the PTO–PTA OTFT transfercharacteristic measured while a nitrogen stream, containing30.6 p.p.m. of (S)-(−)-β-citronellol, continuously flows over thedevice channel during the 45-s-long Vg scan. The comparisonbetween the transfer characteristics measured in N2 and incitronellol (Fig. 2a) shows that a current decrease is observed, asalready reported for different systems2,3, in a citronellol atmospherein the whole on-state regime. A closer inspection of Fig. 2areveals that for gate biases in the depletion regime an Ids currentincrease occurs in the analyte atmosphere. Such an effect has beenseen frequently on substituted thiophene or phenylene–thiopheneorganic semiconductor exposed to organic vapours. A study carriedout on an α,ω-dihexylhexathiophene (DH-α6T) OTFT sensorshows that the current decreases can be ascribed to channelresistance changes whereas the increases are due to contact-relatedresistance changes, the latter showing a lower relative differentialchange affecting just the subthreshold regime. The extent of thecurrent differential change, 1I = [Ids(N2) − Ids(citronellol)], that

nature materials VOL 7 MAY 2008 www.nature.com/naturematerials 413

© 2008 Nature Publishing Group

ARTICLES

ΔI

N2

(S)-(–)- - citronellol

|I |(A

)

I ds (n

A)

Vg (V)

0 –20 –40 –60 –80

Vg (V)

Vg (V)

0

0

–20

–20

–40

–40

–60

–60

–80

–80

10–11

10–10

10–9|ΔI|

(A)

10–8

10–10

10–9

0

–2

–4

–6

–8

–10a

b

β

Figure 2 PTO–PTA-bilayer OTFT sensor response to (S)-(−)-β-citronellolvapours. The Ids–Vg transfer characteristics were measured by biasing the device ina common-source configuration by sweeping Vg (from positive to negativepotentials) at a fixed Vds, namely −50 V. The current, Ids, flowing in the channelregion was measured in a nitrogen flux (black line) and in a flux of 30.6 p.p.m. of(S)-(−)-β-citronellol in N2 (dark-blue line). Each of the two steps lasted for 45 s andin both cases the total flux was 200ml min−1. Before starting each run, the unbiaseddevice was conditioned by exposure to the analyte atmosphere for 45 s. A strictcontrol of the citronellol vapour concentration in N2 was achieved with a systemcomposed of three (in-series) bottles placed in a thermostatic bath maintained at291.6±0.1 K. The N2 maximum concentration reached, measured bygas-chromatographic (GC) analysis, was 51 p.p.m. for the β-citronellol enantiomers.Lower concentrations were obtained by dilution of the stream at maximumconcentration with pure N2 through a system of computer-controlled flow-meters.The lowest concentration is 2.5 p.p.m., corresponding to a maximum concentrationvapour fraction of 5%. Sensing measurements were all made at room temperature ina standard laboratory environment. The inset of a shows the log plot of Ids versus Vgat Vds = −50 V. b, The absolute value of the differential response, |1 I|, evaluatedby subtracting (at each Vg) the Ids values measured in the absence and in thepresence of the analyte respectively. Red points are relevant to positive 1 I values.

is, the sensor response, is a function of the gate bias, as can beappreciated from the data in Fig. 2b, where red points stand fornegative values.

The achiral single-layer PTO OTFT was exposed to boththe (S)-(−)-β-citronellol and (R)-(+)-β-citronellol vapoursseparately and at different concentrations. The relevant 1I datapoints (averaged over three replicates), error bars and interpolatingcurves are reported in Fig. 3. Neat linear regression curves (R betterthan 0.993) are derived for both the enantiomers. Single-point1I response repeatability (expressed as relative standard deviation,RSD) was lower than 2% in accordance with what has alreadybeen reported1,2 when a reverse-gate bias sweep is applied at thebeginning of each measurement run. The slope of the calibrationcurves does not significantly differ for the two enantiomers asexpected, given the achiral nature of the PTO layer. A detectionlimit of 3 p.p.m. (signal-to-noise ratio = 3, noise taken as thestandard error of the fit) was estimated for both enantiomers.

ΔI (n

A)

5 10 15 20 25 30

-citronellol (p.p.m.)

0 350

–0.5

–1.0

–1.5

–2.0

–2.5(R)-(+)- -citronellol

(S)-(–)- -citronellolβ

β

β

Figure 3 PTO achiral sensing transistor exposed to (S)-(−)-β-citronellol and(R)-(+)-β-citronellol. 1 I values at Vg = −100 V and Vds = −50 V at differentβ-citronellol concentrations in the 2.5–30.6 p.p.m. range. Dark-blue squares areon-state 1 I responses to the (S)-(−)-β-citronellol enantiomer whereas blue circlesare relevant to (R)-(+)-β-citronellol. Solid lines are the interpolating calibrationcurves. The squares and the circles are the average data values over threereplicates and the error bars are taken as one standard deviation. All the data havebeen taken from the same device.

In Fig. 4 the slope values of the calibration curves for the(R)-(+)-β-citronellol and (S)-(−)-β-citronellol are shown inthe whole inspected Vg range along with the associated errorbars. Once more, as PTO is achiral, the two data sets are notsignificantly different. Interestingly, a sensitivity enhancement ofseveral orders of magnitude is readily observable when the devicemoves from the off to the on state. Similar data were gathered forDH-α6T, as well as for differently substituted thiophene oligomersexposed to organic and inorganic species, showing that this is ageneral property of OTFT sensors. It is of note that the OTFTthreshold voltage is often also the turn-on gate bias for the OTFTsensitivity enhancement. Also, the slope figures have different signs,and it is straightforward to associate the negative slopes withthe channel resistance variation and the positive ones with thecontact barrier changes. This means that contact resistance changes(possibly related to some sort of Schottky barrier) can dominatein the subthreshold regime, whereas channel resistance changesare responsible for the much higher sensitivity in the on state.Also, the calibration curve linearity improves in the on state, asthe slope-value-associated error is much lower. This effect, stillunder investigation, could be due to instabilities associated withthe subthreshold regime. Although response enhancement has beenalready reported1–3,10,27,28 for different OTFT sensors, this is the firstdirect evidence of gate-field-induced sensitivity enhancement. Thisenables interesting possibilities for the use of microscopic OTFTsas ultrasensitive electronic transducers that could also operate assensing switches.

The field-induced on-state responses of the PTO–PTA-bilayerOTFT to both the citronellol enantiomers as well as to theracemic mixture are reported in Fig. 5a. Here the calibration curves,relevant to responses acquired while operating the transistor in theaccumulation regime, are again all linear, with R better than 0.991,and each 1I response is repeatable within 5% RSD. Responsereversibility, also in the case of the bilayer OTFT, is therefore almostcomplete. The sensitivity differences in the calibration curves forthe PTO–PTA OTFT against the two enantiomers are readilyapparent. The values are mR = −0.08±0.01 nA per p.p.m. for the(S)-(−)-β-citronellol and mS = −0.04 ± 0.01 nA per p.p.m. for(R)-(+)-β-citronellol. Interestingly, the sensitivity for the racemic

414 nature materials VOL 7 MAY 2008 www.nature.com/naturematerials

© 2008 Nature Publishing Group

ARTICLES

(R)-(+)- -citronellol

(S)-(–)- -citronellol

m (A

per

p.p

.m.)

0 –20 –40 –60 –80

Vg (V)

20 –10010–14

10–13

10–12

10–11

10–10

β

β

Figure 4 Sensitivity enhancement with OTFT sensors exposed to citronellolenantiomers. The slopes of the calibration curves of a PTO OTFT exposed to the(R)-(+)-β-citronellol and (S)-(−)-β-citronellol at different gate biases, along withthe error bars (taken as described in Fig. 3). Red points are relevant to positiveslope values.

mixture (50% of each enantiomer) falls right between the previoustwo, being mrac = −0.06 ± 0.01 nA per p.p.m.. PTO–PTA-bilayerOTFT sensors are therefore capable of enantiomeric differentialdetection of β-citronellol at the p.p.m. level. This example isparticularly relevant, as citronellol enantiomer discrimination isdifficult even using common chromatographic techniques. Inperfect analogy to what is measured in the case of the PTOOTFT, the chiral bilayer OTFT calibration curve slopes (mR, mS

and mrac) are also switched by at least one order of magnitudeas the device is driven from the depletion to the accumulationmode. In fact, m versus Vg plots, similar to that reported inFig. 4, were also recorded for the PTO–PTA OTFTs. It is clearthat much worse sensitivities and, eventually, chiral discriminationcapabilities should be expected with a chemiresistor comparedwith an OTFT sensor operated in the on state. Comparison withthe literature seems to confirm this result, as chiral polymerchemiresistors could carry out chiral differential detection in thepart-per-thousand range, that is, at enantiomer concentrationsthree orders of magnitude higher14,15.

To prove the wider applicability of the system proposed,an alkoxyphenylene–thiophene oligomer bearing a -glucosidicchiral recognition element, PTG in fact, was used to sense ketoneenantiomers, namely (S)-(+)- and (R)-(−)-carvone. The bilayerOTFT was fabricated by stacking a PTG Langmuir–Schafer (LS)film onto the usual PTO underlayer. The calibration curves arereported in Fig. 5b. Also in this case it is apparent that thedevice sensitivity changes as the two enantiomers are separatelysensed. The slope values are mS = −0.16 ± 0.01 nA per p.p.m.for the (S)-(+)-carvone and mR = −0.21±0.01 nA per p.p.m. for(R)-(−)-carvone. R values are once more better then 0.992 andeach 1I response is repeatable within 6% RSD. The detection limitis very low also in this case, 9 p.p.m.

The calibration curves of the PTO–PTA OTFT exposed tocarvone isomers and those of the PTO–PTG OTFT exposed tocitronellol ones have been measured. In both cases 1I responseslinearly increase with concentration, and are highly repeatableand enhanced by the gate field. However, no chiral differentialdetection could be carried out in these configurations, as themeasured sensitivities were identical for the two enantiomers ofcarvone on the PTO–PTA OTFT and for those of citronellol onPTO–PTG. This system-dependent selectivity is not surprising, as

ΔI (n

A)ΔI

(nA)

(S)-(+)–carvone

(R)-(–)–carvone

0

–0.5

–1.0

–1.5

–2.0

–2.5

–3.0(R)-(+)- -citronellol

(S)-(–)- -citronellol

Racemic mixture

0

–2

–4

–6

–8

–10

–12

–14

604020

105 15 20 25 30

-citronellol (p.p.m.)

0 35

Carvone (p.p.m.)

0 80

a

b

ββ

β

Figure 5 Calibration curves of the chiral bilayer OTFTs exposed to(S)-(−)-β-citronellol, (R)-(+)-β-citronellol and to the racemic mixture as wellas to (R)-(−)- and (S)-(+)-carvone. a, PTO–PTA OTFT 1 I values measured atVg = −100 V and Vds = −50 V while exposing the device to β-citronellolenantiomers and to their racemic mixture at concentrations in the 2.5–30.6 p.p.m.range. Dark-blue squares are on-state 1 I responses to the (S)-(−)-β-citronellolenantiomer and blue circles are relevant to (R)-(+)-β-citronellol. The response tothe racemic mixture is reported as green triangles. Solid lines are the interpolatingcalibration curves. b, PTO–PTG OTFT 1 I values at Vg = −100 V and Vds = −50 Vfor carvone enantiomers at concentrations in the 10.8–64.7 p.p.m. range. Dark-bluesquares are on-state 1 I responses to the (R)-(−)-carvone enantiomer and bluecircles are relevant to the (S)-(+) one. All the data points are the average valuesover three replicates and the error bars (barely visible in b) are taken as onestandard deviation. All the data in each panel have been taken from the same device.

weak intermolecular interactions, probably responsible for chiraldiscrimination, critically depend on structural differences of themolecules involved. On the other hand, in an array-type sensingsystem, the different responses of the PTA- and PTG-based OTFTsto the same analyte could be exploited to improve the systemchemical recognition capabilities.

A simple model able to explain the experimental evidencepresented so far can be depicted as follows. PTO thin-filmmorphology can be worked out on the basis of atomic forcemicroscopy data recently gathered on PTO-like polymer thinfilms, of comparable thicknesses and deposited by the same LSprocedure25,29. It is shown that such thin films are composed ofcontiguous grains and voids roughly 50–100 nm wide. Even moreimportant, the granular structure is seen from the very first deposit(LS transfer). This means that the whole in-depth PTO layer is

nature materials VOL 7 MAY 2008 www.nature.com/naturematerials 415

© 2008 Nature Publishing Group

ARTICLES

probably made of stacking contiguous grains and comparably largevoids. Similar granular morphologies were also seen for glucose-substituted poly(p-phenyleneethynylene) thin films30. Additionally,the formation of chiral supramolecular aggregates was reported31.On this basis, it can be deduced that the in-depth granular structureapplies to the whole bilayers as well. The role of the active-layer grain size in the sensing mechanism was also previouslyinvestigated, showing that the larger the grains the lower thesensor response32,33. Hence, a morphology comprising nanograinsis beneficial to an OTFT for sensing applications but can negativelyaffect electronic performance. Also in this respect a trade-off is to besought in developing an OTFT sensor technology, as better OTFTmobility is observed with larger grains.

When OTFTs are exposed to the vapours, the analyte moleculediffuses into the organic semiconductor solid phase. Analytemolecule permeation into the grains sounds unlikely: this isexpected for swellable membranes generally showing spongelikemorphologies. Indeed, swelling of CP semiconductors withgranular morphology on exposure to organic vapour was notobserved1,10,33. Molecules therefore largely percolate through thevoids around the grains till the interface with the dielectricis reached. Such diffusion is likely to be fast and the OTFTresponse and recovery times support this hypothesis. Eventually,the molecules are adsorbed at the grain surface. The degree ofphysisorption is a function of the molecules’ aliquot statisticallypresent in the organic semiconductor phase. At equilibrium, thepartition coefficient quantifies the partitioning of the analytemolecules between the solid and the vapour phase. The degree ofanalyte/organic semiconductor chemical affinity, reflecting on thepartition coefficient, can be modulated by choosing appropriateside groups, as is well known in gas chromatography, and as wasdemonstrated in OTFT sensors too10. This strategy also appliesto chiral analytes whose discrimination requires chiral receptors.For this, an external layer of semiconducting molecules covalentlyfunctionalized with chiral side groups was specifically appointedto modulate the partition coefficient of the two enantiomers.Eventually, this confers the required selectivity by modulating thealiquot of analytes that passes through the external layer reachingthe inner one. The bilayer structure does not in this case—andin principle should not in general—play any role in enhancingthe sensitivity of the device as claimed in a recent publication11.The ability of CP systems bearing -glucosidic or -phenylalanineunits to differentially detect enantiomers of chiral compounds inthe vapour phase has already been assessed by measuring thedifferential mass uptake properties in preliminary quartz-crystalmicrobalance studies30,34. A report fully addressing all the chemicaland mechanistic aspects is in preparation.

Once the molecules in the solid phase have reached thedielectric interface, they can affect the two-dimensional transportproperties. It can be further postulated that, as the room-temperature transport-rate-determining step in OTFTs is thethermionic emission through grains35, a change in the potentialbarriers at grain boundaries can have a large impact on the Ids

current drifting in the channel region. Preliminary data on thecomparison of the activation energies, evaluated from temperature-dependent transfer-characteristic curves of a pentacene OTFT,show that a barrier increase is in fact observed on exposure topentanol36. Apparently physisorption at grain boundaries has theeffect of enhancing energy barriers, and this is critical. Eventually,this effect lowers the on-state Ids current flow, giving rise to atwo-dimensional sensing process. This quite fascinating hypothesiscan be supported by proving the thickness independence of anOTFT’s response. In a recent publication29 it was demonstratedthat, for a PTO-like polymer, a thickness-independent field-effectconductance holds, in accordance with several other similar studies

95 nm290 nm470 nm

950 nm

(S)-(–)- -citronellol

5 10 15 20 25 300 35

1

2

3

4

5

6

7

ΔI (n

A)

0

8

-citronellol (p.p.m.)β

β

Figure 6 Thickness-independent PTO OTFT 1 I responses. 1 I values relevant toPTO OTFT with active layer thickness ranging from 95 to 950 nm (5 to 50 LS transfer)exposed to different (S)-(−)-β-citronellol concentrations along with the regressioncurves. The data points are the averages over three replicates and the error bars (notvisible as they are smaller than the symbols) are taken as one standard deviation.

on OTFTs37. This implies that the gate-field-induced transport isconfined at the organic-semiconductor/dielectric interface, that is,it is two dimensional. The 1I responses of PTO OTFTs withactive layer thicknesses ranging from 95 to 950 nm exposed todifferent concentrations of (S)-(−)-β-citronellol are reported inFig. 6. Again, each data point is averaged over three replicates andthe error bars are very small, falling in fact within the area ofthe symbols. The thickness-independent nature of the responsesis apparent as no appreciable sensitivity difference can be seenon thickness variation over one order of magnitude. The sensingprocess in OTFTs is therefore strictly connected to the two-dimensional transport occurring at the interface with the gatedielectric, whereas the enantioselective recognition takes place onlyin the chirally functionalized external layer. It is also importantto note the following: in a chemiresistor the three-dimensionaltransport properties are involved in the sensing process, whereas inan OTFT two-dimensional transport is affected. The field-inducedsensitivity enhancement reported here can be seen as an effectcorrelated to the larger number of charges drifting in the device’schannel in the on-state accumulation regime with respect to thecharges drifting in the bulk in the device off state. In the off-state sensing spurious effects, also connected to contact resistancevariation, can play a role.

In this study, chiral detection is accomplished by appropriatelysubstituted alkoxyphenylene–thiophene oligomers and can beextended in principle to receptors other than the -phenylalanineand -glucosidic moieties used here. Although some figuresof merit of the transistor sensors are susceptible to furtherimprovements, this study represents a significant step forward inthe development of a high-performance and versatile chemical andbio-sensing electronic platform.

METHODS

Two octyloxy chains characterize the PTO system whereas six-carbon-atomalkoxy chains linked through an ester bond to the carboxylic group of anN-BOC-protected -phenylalanine molecule are the PTA substituents. InPTG, per-acetylated -glucose molecules are directly bound to the centralphenylene ring through a β-glucosidic linkage. The PTO, PTA and PTG thinfilms were deposited by the LS technique, which enables, similarly to theLangmuir–Blodgett one, a stratum-by-stratum transfer of a molecular assemblyon a substrate. It generally enables an effective thickness and morphologicalcontrol. PTO is deposited in direct contact with the gate dielectric, creating

416 nature materials VOL 7 MAY 2008 www.nature.com/naturematerials

© 2008 Nature Publishing Group

ARTICLES

the interface where the two-dimensional field-effect transport takes place37.PTA or PTG LS layers are staked on the PTO, forming the PTO–PTA andPTO–PTG bilayers (Fig. 1). Both single layers and double layers comprise 15 LStransfers overall and the relevant thicknesses are a few hundred nanometres.A detailed description of the synthesis and characterization of PTO, PTA andPTG molecular semiconductors, their LS deposition procedure and details ofthe device electrical properties are reported in the Supplementary Information.

The sensing measurements were made by directly exposing the bilayerOTFTs to vapours of a chiral alcohol and a ketone, namely to (S)-(−)-and (R)-(+)-β-citronellol or (S)-(+) and (R)-(−)-carvone enantiomersseparately, as well as to the vapours of the β-citronellol racemic mixture. A PTOsingle-layer OTFT was used as achiral reference. It is interesting to note thatβ-citronellol enantiomer discrimination is difficult by chiral chromatography.Their separation was, in fact, either not achieved or accomplished at very low(≤1) resolution factor38,39. The analyte concentration that impinged on thesole PTO or on either one of the two bilayer structures was in the low p.p.m.range. The analyte, in a nitrogen stream at a controlled concentration, wasdelivered through a nozzle directly onto the active layer surface. The OTFTsensors were always operated at room temperature and in a standard laboratoryenvironment. In agreement with what has already been reported for OTFTsexposed to organic volatiles, response and recovery times were very short,falling in the tens–hundreds of seconds range at most1,2,11. More informationconcerning OTFT fabrication and sensing measurements is summarized in thefigure captions.

The role of the source and drain contact resistance in the sensingmechanism was also investigated. This is definitely an issue for carbon-nanotube TFT sensors exposed to NO2 (ref. 40), whereas no effect was seen onthe same devices exposed to alcohols41. It has been already commented thatthis phenomenon in OTFTs affects mainly the subthreshold regime. However,to prove its minor role to a further extent, PTO–PTA OTFT sensor source anddrain contacts were passivated by depositing a SiO2 coating and 1I valueswere measured at 20 p.p.m. of both β-citronellol enantiomers separately.The response gathered in the device on state are in every way comparable tothose obtained with uncovered electrodes. Moreover, the modelling of thedevice characteristics carried out in the framework of the model developed byHorowitz and co-workers35 showed that the channel conductance, gds, is linearwith Vg (see the Supplementary Information). These two pieces of evidencerender unlikely that the sensing mechanism involved in the systems proposedcan be dominated by contact-related effects while operated in the on state.

In this study, chiral differential detection was also evaluated on a series ofthree different samples of PTO–PTA OTFTs exposed to citronellol enantiomers.In this case the average device selectivity coefficient (σ = mR/mS) was σ = 1.72,with an RSD of 15%.

Received 20 June 2007; accepted 12 March 2008; published 20 April 2008.

References1. Torsi, L. & Dodabalapur, A. Organic thin-film transistors as plastic analytical sensors. Anal. Chem.

77, 380A–387A (2005).2. Crone, B. et al. Electronic sensing of vapors with organic transistors. Appl. Phys. Lett. 78,

2229–2231 (2001).3. Bernards, D. A. et al. Enzymatic sensing with organic electrochemical transistors. J. Mater. Chem. 18,

116–120 (2008).4. Star, A., Gabriel, J.-C. P., Bradley, K. & Gruner, G. Electronic detection of specific protein binding

using nanotube FET devices. Nano Lett. 3, 459–463 (2003).5. Torsi, L. Organic thin-film transistors as analytical and bioanalytical sensors. Topic issue. Anal.

Bioanal. Chem. 384, 309 (2006).6. Mcculloch, I. Thin films: Rolling out organic electronics. Nature Mater. 4, 583–584 (2005).7. Forrest, S. R. The path to ubiquitous and low-cost organic electronic appliances on plastic. Nature

428, 911–918 (2004).8. Jenekhe, S. A. The special issue on organic electronics. Chem. Mater. 16, 4381–4390 (2004).9. Janata, J. & Josowicz, M. Conducting polymers in electronic chemical sensors. Nature Mater. 2,

19 (2003).10. Torsi, L. et al. Side-chain role in chemically sensing conducting polymer field-effect transistors.

J. Phys. Chem. B 107, 7589–7594 (2003).11. Huang, J., Miragliotta, J., Becknell, A. & Katz, H. E. Hydroxy-terminated organic

semiconductor-based field-effect transistors for phosphonate vapor detection. J. Am. Chem. Soc. 129,9366–9376 (2007).

12. Lemaire, M., Delabouglise, D., Garreau, R., Guy, A. & Roncali, J. Enantioselective chiralpoly(thiophenes). J. Chem. Soc. Chem. Commun. 658–661 (1988).

13. Huang, J. et al. Enantioselective discrimination of - and -phenylalanine by chiral polyaniline thinfilms. Adv. Mater. 15, 1158–1161 (2003).

14. Severin, E. J., Sanner, R. D., Doleman, B. J. & Lewis, N. S. Differential detection of enantiomericgaseous analytes using carbon black–chiral polymer composite, chemically sensitive resistors. Anal.Chem. 70, 1440–1443 (1998).

15. de Lacy Costello, B. P. J., Ratcliffe, N. M. & Sivanand, P. S. The synthesis of novel 3-substituted pyrrolemonomers possessing chiral side groups: A study of their chemical polymerization and theassessment of their chiral discrimination properties. Synth. Met. 139, 43–55 (2003).

16. Bodenhofer, K. et al. Chiral discrimination using piezoelectric and optical gas sensors. Nature 387,577–580 (1997).

17. Bodenhofer, K. et al. Chiral discrimination in the gas phase using different transducers: Thicknessshare mode resonators and reflectometric interference spectroscopy. Anal. Chem. 69,3058–3068 (1997).

18. Hagleitner, C. et al. Smart single-chip gas sensor microsystem. Nature 414, 293–296 (2001).19. Crone, B. et al. Organic oscillator and adaptive amplifier circuits for chemical vapor sensing. J. Appl.

Phys. 91, 10140–10146 (2002).20. Wang, L., Fine, D. & Dodabalapur, A. Nanoscale chemical sensors based on organic thin-film

transistors. Appl. Phys. Lett. 85, 6386–6388 (2004).21. Wang, L., Fine, D., Torsi, L. & Dodabalapur, A. Nanoscale organic and polymeric field-effect

transistors as chemical sensors. Anal. Bioanal. Chem. 384, 310–321 (2006).22. Someya, T., Dodabalapur, A., Gelperin, A., Katz, H. E. & Bao, Z. Integration and response of organic

electronics with aqueous microfluidics. Langmuir 18, 5299–5302 (2002).23. Babudri, F., Farinola, G. M. & Naso, F. Synthesis of conjugated oligomers and polymers: The

organometallic way. J. Mater. Chem. 14, 11–34 (2004).24. Naso, F. et al. Thin film construction and characterization and gas-sensing performance of a tailored

phenylene–thienylene copolymer. J. Am. Chem. Soc. 125, 9055–9061 (2003).25. Tanese, M. C. et al. Poly(alkoxyphenylene–thienylene) Langmuir–Schafer thin-films for advanced

performance transistors. Chem. Mater. 18, 778–784 (2006).26. Torsi, L., Dodabalapur, A., Sabbatini, L. & Zambonin, P. G. Multi-parameter gas sensor based on

organic thin-film transistors. Sensors Actuators B 67, 312–316 (2000).27. Tanese, M. C., Fine, D., Dodabalapur, A. & Torsi, L. Interface and gate bias dependent response of

sensing organic thin-film transistors. Biosens. Bioelectron. 21, 782–788 (2005).28. Someya, T., Katz, H. E., Gelperin, A., Lovinger, A. J. & Dodabalapur, A. Vapour sensing with

α-ω-dihexylquaterthiophene field-effect transistors: The role of grain boundaries. Appl. Phys. Lett.81, 3079–3081 (2002).

29. Tanese, M. C. et al. Nanostructural depth-profile and field-effect properties ofpoly(alkoxyphenylene–thienylene) Langmuir–Schafer thin-films. Thin Solid Films 516,3263–3269 (2008).

30. Tanese, M. C. et al. Conjugated phenylene–ethynylene polymers bearing glucose substituents aspromising active layers in enantioselective chemiresistors. Sensors Actuators B 100, 17–21 (2004).

31. Babudri, F. et al. Synthesis and chiroptical characterization of an amino acid functionalizeddialkoxypoly(p-phenyleneethynylene). Macromolecules 39, 5206–5212 (2006).

32. Horowitz, G., Hajlaoui, M. E. & Hajlaoui, R. Temperature and gate dependence of hole mobility inpolycrystalline oligothiophene thin film transistors. J. Appl. Phys. 87, 4456 (2000).

33. Torsi, L. et al. Correlation between oligothiophene thin film transistor morphology and vapourresponses. J. Phys. Chem. B 106, 12563–12568 (2002).

34. Tanese, M. C. et al. A poly(phenyleneethynylene) polymer bearing amino acid substituents as activelayer in enantioselective solid-state sensors. Proc. SPIE 6192, 61921E (2006).

35. Horowitz, G. & Haifaoui, M. E. Mobility in polycrystalline oligothiophene field-effect transistorsdependent on gate bias. Adv. Mater. 12, 1046 (2000).

36. Dodabalapur, A. et al. Sensitive four-terminal silicon/organic field-effect chemical vapor sensors.Part of Proc. SPIE 6659, 185 (2007).

37. Dodabalapur, A., Torsi, L. & Katz, H. E. Organic transistors: Two-dimensional transport andimproved electrical characteristics. Science 268, 270–271 (1995).

38. Kreis, P. & Mosandl, A. Chiral compounds of essential oils. Part XII. Authenticity control of rose oils,using enantioselective multidimensional gas chromatography. Flavour Fragrance J. 7, 199–203 (1992).

39. <http://www.restek.com/Fantasia/pdfCache/59889.pdf>.40. Zhang, J., Boyd, A., Tselev, A., Paranjape, M. & Barbara, P. Mechanism of NO2 detection in carbon

nanotube field effect transistor chemical sensors. Appl. Phys. Lett. 88, 123112 (2006).41. Someya, T., Small, J., Kim, P., Nuckolls, C. & Yardley, J. T. Alcohol vapor sensors based on

single-walled carbon nanotube field effect transistors. Nano Lett. 3, 877 (2003).

Supplementary Information accompanies this paper on www.nature.com/naturematerials.

AcknowledgementsWe are indebted to E. A. Chandross and G. Horowitz for advice. L. Sabbatini, P. Iliade andM. D. Angione are acknowledged for discussions. We are grateful to L. Dimo for assistance in the LSdepositions. Italian MIUR ‘Bando DM 1105 del 9/10/2002 Progetto n.ro 100/2’, Progetto FIRB 2003‘SYNERGY RBNE0 3S7XZ 001’ and ‘PRIN-06 Project—2006037708—Plastic bio-FET sensors’ areacknowledged for partial financial support. While the present manuscript was under revision anotherpaper was published (Huang, J., Miragliotta, J., Becknell, A. & Katz, H. E., J. Am. Chem. Soc. 129,9366-9376 (2007)) proposing a two-layer OTFT sensor.

Author contributionsAll authors have agreed to all the content of the manuscript, including the data as presented.

Author informationReprints and permission information is available online at http://npg.nature.com/reprintsandpermissions.Correspondence and requests for materials should be addressed to L.T. or F.N.

nature materials VOL 7 MAY 2008 www.nature.com/naturematerials 417

© 2008 Nature Publishing Group