Supporting Information...0.02–1 ohm∕cm) acting as the gate material. The gate contact was taken through an evaporated gold pad in contact with the silicon substrate. The gate dielectric

Supporting InformationAngione et al. 10.1073/pnas.1200549109SI TextSI Materials and Methods. FBI-OFETs Detailed Fabrication Procedure.The Functional BioInterlayer Organic Field-Effect Transistors,whose structure is reported in Fig. 1A, have been fabricated start-ing from a highly n-doped silicon substrate (resistivity0.02–1 ohm∕cm) acting as the gate material. The gate contactwas taken through an evaporated gold pad in contact with thesilicon substrate. The gate dielectric is thermally grown SiO2,either 100 or 300 nm thick, whose nominal capacitance per unitarea is 27 and 9 nF∕cm2, respectively. The FBI-SiO2 stackinglayers’ capacitance per unit area does not differ significantly fromthat of the bare SiO2, as the latter holds comparatively lower ca-pacitance than the sole FBIs. SiO2 was chosen in fact to minimizecapacitive effects in the evaluation of the FBI-OFET bioelectro-nic properties. Source-drain and source-gate biases (VDS and VG)ranged 20–100 V for the 300-nm thick SiO2 and 0–40 V for thethinner SiO2. Maximum voltage biases on the two dielectrics werescaled ad hoc to keep the maximum induced charge density con-stant. The SiO2 surface was cleaned through a rinsing procedureinvolving treatment with solvents of increasing polarity (1), thecontact angle of cleaned SiO2 surface being 72°� 2° for a2 μL deionized water droplet. The FBI layers have been subse-quently deposited, directly on the cleaned SiO2, by very slow-speed spin coating. Specifically, the following procedures wereimplemented:

Phospholipid (PL) Bilayers: An aqueous suspension of single uni-lamellar vesicles (SUVs) of PLs was prepared by dissolving 10 mgof soybean lecithin (EPIKURON 200, Cargill) in chloroform,subsequently dried under vacuum. When required, 10 μg of di-hexadecanoyl-phosphatidyl-ethanolamine labeled with a fluoro-phore (Texas-red DHPE from Invitrogen) were added to thechloroform solution. The phospholipids were suspended in1 mL of distilled water and sonicated on ice for 10 minutes.The resulting multilamellar vesicles suspension was repeatedlysubjected to extrusion through a polycarbonate filter having poresizes of 100 nm using the Avanti® mini-extruder (Avanti Polar),to obtain an evenly dimensionally distributed SUVs with an aver-age diameter of 80 nm as assessed by dynamic light scattering(ZetasizerNano Malvern Instruments). 50 μL of the SUVs sus-pension was deposited on the cleaned SiO2 surface, by spin coat-ing at 200 rpm for 20 min. It is widely reported in literature that,after deposition on a solid substrate, spontaneous fusion of thevesicles occurs simultaneously to adsorption of the double layeron the substrate (1).

Purple Membrane (PM): Purple membranes are discrete mem-brane patches in the archea Halobacterium salinarium consistingof approximately 75% bacterioRhodopsin (bR) and 25% lipid (2,3). These patches consist of an hexagonal 2D crystalline lattice ofuniformly oriented bR trimers that can be isolated by osmoticallylysing the cell followed by differential centrifugation. In the pre-sent work, lyophilized PMs have been purchased from MunichInnovative Biomaterials GmbH. The dry PMs were suspended(3 mg∕mL) in distilled water and a uniform suspension wasachieved by mild sonication on ice (5 cycles of 5 s with 2 min in-terval). The suspension was then diluted with water to a final PMconcentration of 10 μg∕mL and 60 μL of this PM suspensionwere spin coated at 150 rpm for 90 min directly on the cleanedSiO2 surface (or glass support in the case of optical measure-ments). Lower PM concentrations resulted in proportionally thin-ner film.

Streptavidin (SA): A 10 μg∕mL aqueous solution of SA (pur-chased from Sigma-Aldrich) was spin coated at 200 rpm for40 min directly on a bare SiO2 surface. The optimization ofcapture proteins immobilization was performed through fluores-cence imaging. The homogeneity of the immobilized biomoleculelayer on the SiO2 surface and the effect of chloroform spreadingwas evaluated also by fluorescence imaging techniques on a strep-tavidin—Alexa Fluor 488—fluorescent conjugate (provided byInvitrogen). A concentration of 10 μg∕mL assured a sufficientexcess of deposited capture proteins allowing to minimize errorsconnected to intersample variations of protein immobilizationefficiency. As control experiment also BSA, known for being aprotein not binding specifically to biotin, was used as FBI layer.To a similar aim, a saturated SA-biotin complex was used as non-specific FBI layer, while as positive control, an antibiotin mono-clonal antibody (Biotin, Mouse IgG1-cod. 033700, Invitrogen)was used. All the elicited biological systems were spin coatedfrom a 10 μg∕mL water solution using same conditions as for SA.

Organic semiconductor: The organic semiconductor (OS) washighly regioregular poly(3-hexylthiophene-2,5-diyl)—P3HT(RR > 98%, BASF Sepiolid P200). The polymer was purified ac-cording to an assessed standard procedure (4). The P3HT waspoured into a cellulose thimble and extracted in a Soxhlet appa-ratus, first with methanol and then with hexane. The purifiedP3HT was then dissolved in chloroform at a concentration of2.6 mg∕mL. The solution absorption spectrum was dominatedby a single band centered at 452 nm. The solid-state spectrumof purified P3HT features three vibronic bands located at 520,555 and 602 nm. Both these data are consistent with a (average)high molecular weight of the order of 10 kDa (5). The depositionof the OS was performed by spin coating at a spin rate of2,000 rpm for 30 sec. The P3HT film thickness was about 20 nmand its uniformity, inspected by optical microscopy, was generallyvery high. At the nanoscopic level, a granular morphology is seento comprise voids up to ten-hundred nm wide. The X-ray dataindicate that the RR P3HT film grows with the h100i axis prefer-entially oriented normal to the film surface, according to a lamel-lar model with a lamellar period d ¼ 1.63 nm. This is in goodagreement with published data for a high-quality RR P3HT film.The measured contact angle, being 97°� 3°, is typical of a hydro-phobic surface. It has been demonstrated, however, that dopingcan increase P3HTwettability as the contact angle can be loweredby more than 10° (6).

Source (S), drain (D), and gate (G) contacts were deposited bythermal evaporation (8 × 10−7 torr) of gold through a shadowmask. The geometry used to define the S and D contacts resultsin a sequence of several rectangular pads spaced by 200 μm; thisspacing is indicated as “L” in Fig. 1A, being the channel length,while the channel width,W ¼ 4 mm, is the width of the pads. Nopatterning of any layer, but the contacts, is implemented, thusmaking the fabrication of FBI-OFETs extremely easy and com-patible with ink-jet fabrication procedures. The devices have beenoperated in the common-source configuration and critical elec-tronic performance parameters (field-effect mobility, μFET, cur-rent amplification—on/off ratio, and threshold voltage VT) wereextracted from experimental current-voltage characteristics usingassessed procedures (7), by plotting the square-root of the source-drain current measured at fixed source-drain bias while sweepingthe gate bias.

Angione et al. www.pnas.org/cgi/doi/10.1073/pnas.1200549109 1 of 10

Electronic Response Measurements. The electronic responseshave been evaluated by measuring the OFET drain-sourcecurrent − gate-source voltage, (IDS −VG) transfer-characteris-tics, by sweeping the gate bias and measuring the current flowingin the OFET channel (IDS) while keeping the source-drain bias,VDS, fixed at −80 V. The transfer characteristics were first mea-sured in a N2 or deionized H2O environment (blank) and then inthe presence of the different analyte concentrations. This is a con-venient procedure, already adopted in previous studies (7), thatallows to operate OFET sensors from the depletion to the accu-mulation mode, eventually improving the response repeatabilityin reversible interactions (7). Both the bare P3HT OFET devicesas well as the FBI-OFETwere then exposed to the species to beprobed. The transfer characteristics normalized source-drain cur-rent changes (ΔI∕Io) are given by:

½IðIDS exposed to the analyteÞ− IoðIDS in the blankÞ�∕IoWhile the whole transfer-characteristic is measured, the Io and Ivalues are taken at VG ¼ −100 V. The ΔI∕Io is the electronicresponse at a given concentration, and the relevant dose curveis built by plotting the data points at all investigated concentra-tions as the average values over three replicates while the errorbars are taken as the standard deviation. To perform the sensingexperiments, different procedures were used for the detection ofvolatile and liquid substances.

Procedure 1: Determination of Halothane and Volatile Organic Va-pors. Io was evaluated measuring the transfer-characteristics ofthe bare P3HT OFETsensor in a N2 flux. Afterward, on the verysame transistor, a controlled concentration flow of the analytewas delivered and the IDS current was measured. See Figs. 2Cand D as examples. The controlled concentration of the volatilespecies was obtained by bubbling the inert carrier gas, N2, intotwo bubblers in series containing the analyte in its liquid form.This allows to obtain a nitrogen flow having a partial pressureof the vapors equal to the pressure of saturated vapor of the ana-lyte at the working temperature, the latter being controlled bydipping the bubblers into a cryothermostat. Flow control wasachieved through a system of two computer-controlled flow-meters (Brooks Smart DMFC model 5850C). The described ex-perimental apparatus allows to control independently andsimultaneously the flow of the carrier gas and that of the nitrogensaturated with the analyte. Different analyte concentrations wereobtained by subsequent dilutions changing the rate of the twoflows. The analyte concentrations were delivered in a random se-quence. A single dose curve was taken from the same device asthe P3HT interaction with the chosen volatile analytes was fullyreversible. All the measurements were carried out in the dark.

Procedure 2: Determination of Biotin in Water. To perform the bio-sensing measurements in water, a droplet (2 μL) of deionizedwater was deposited directly on the P3HTsurface of a FBI-OFETbetween two contiguous source and drain pads and dried under anitrogen flow. The transfer characteristics were then measured ina N2 flux and the Io current value was taken. The solution con-taining the biotin analyte, at a randomly chosen concentration,was then deposited on the same device and incubated for15 min. Subsequently, the unbound excess analyte was removedby washing with 2 μL droplet of deionized water three times. Thedevice was then dried under a nitrogen flow, and the transfercharacteristic was measured. At a given concentration, three dif-ferent I-values were measured and the average ΔI∕Io value wasplotted on the dose curve along with error bars corresponding tothe associated standard deviation. In this case, as the involvedinteractions are irreversible, each data point was taken on a dif-ferent device. Overall 15 different OFETs were tested for eachcurve, three for each concentration. These devices, although dif-

ferent, lied on the same chip, therefore had been fabricated with-in the same batch. Biotin was also tested on bare P3HT-OFET(blank experiment) and on a BSA-embedding OFET (control ex-periment) as well as on a SA-biotin complex and an antibiotinantibody. Also in these cases, 15 different devices were testedto gather data for each dose curve.

Morphological, Spectroscopic, and X-ray Characterization Apparata.Atomic Force Microscopy (AFM) measurements have been rea-lized with a XE-100E (Park System Corp., Korea) atomic forcemicroscope in Non-Contact Mode using silicon tips (curvatureradius <10 nm force constant 40 N∕m).

UV-visible absorption spectroscopy. The effect of light on theabsorption spectrum of PM was assessed by an Agilent 8453diode array spectrophotometer. The full spectrum from 190 nmto 1100 nm was recorded within 1 s with a spectral resolution of1 nm. The PM-P3HT film, deposited on glass, was placed at 45° toallow illumination with both the measuring beam and the excitingyellow light (supplied by a cold light source—filtered through ayellow filter—λ > 500 nm, density power on the sample beingca. 1.86 mW∕cm2).

Photoluminescence spectra were obtained by focusing the476.2 nm line of a Krþ laser to a spot of about 2 micrometer dia-meter with an 80× microscope objective. The Rayleigh scatteringgenerated by the laser light was suppressed using an interferencenotch filter. Laser-induced overheating of the samples was mini-mized by keeping the incident power density below 104 W∕cm2.The PL signal was dispersed using a 0.64 m monochromator anddetected with a Si charge-coupled device detector (CCD) cooledto 140 K.

Fluorescence (emission) spectra of liquid samples were mea-sured using a Cary Eclipse FL spectrophotometer (Varian).

The surface chemical characterization of the deposited singlelayer of PL and PL-P3HT was performed in a standard modeoperation by means of X-ray photoelectron spectroscopy(XPS) using a Thermo VG Theta Probe spectrometer equippedwith a micro-spot monochromatized Al Kα source (spot size ¼400 μm). Survey and high-resolution spectra were acquired infixed analyzer transmission mode with pass energies of 150and 100 eV, respectively.

X-ray specular reflectivity (XSR) was employed to assess thethicknesses of the P3HT, of the as deposited phospholipid layersas well as of the PL-P3HT stacking layers. Reflectivity curves(θ∕2θ scans) were collected for all samples by using a D8 Dis-cover diffractometer by Bruker, equipped with a Göbel mirror,using CuKα radiation, an Eulerian cradle and a scintillator de-tector.

Grazing Incidence Small Angle X-Ray Scattering (GISAXS)measurements were carried out with a S3-MICRO SWAXS cam-era system (HECUS X-ray Systems, Graz, Austria). Cu Kα radia-tion of wavelength, λ ¼ 1.542 Å, was provided by an ultra-brilliant point micro-focus X-ray source (GENIX-Fox 3D, Xe-nocs, Grenoble), operating at a maximum power of 50 W (50 kVand 1 mA). The sample-to-detector distance was 269 mm. Thevolume between the sample and the detector was kept under va-cuum during the measurements to minimize scattering from theair. The Kratky camera was calibrated in the small angle regionusing silver behenate (d ¼ 58.38 Å) (8). Scattering curves wereobtained in the Q-range between 0.01 and 0.54 Å−1, q beingthe scattering vector q ¼ 4πðsin θÞ∕λ, and 2θ the scattering angle.

Atomic Force Micrographs of P3HT in FBI-OFETs. Fig. S1A shows amicrograph taken on the P3HT deposited directly on the atom-ically flat SiO2. The data clearly show that the morphology iscomposed of granular domains, leaving voids of ca. 10–100 nm.This is in full agreement with what has already been published onRR-P3HT morphology (9). The Fig. S1B surface refers to a PMstack (1 μm thick) covered by the 20 nm P3HT film. The morphol-

Angione et al. www.pnas.org/cgi/doi/10.1073/pnas.1200549109 2 of 10

ogy is much more structured in this case. A granular surface mor-phology ascribable to the covering P3HT, is discernible. InFig. S1C, a SA deposit exhibits aggregated of SA molecules100 nm in size. Coating of SA film with P3HT is featured inFig. S1D exhibiting a very structured surface resembling the un-derneath biodeposit. Also in this case, the granular features of theP3HTare clearly seen on the very top. For all the FBI voids arepresent that are likely to allow analyte percolation.

FBI-OFET Stacking-Layers Structure. PL-P3HT OFET Stacking-LayersStructure. Photoluminescence Spectra. The assessment of the actualcomposition and stacking multilayer structure of the OFET inte-grating the PL layer was quite critical as the phosphatidylcholinePLs are soluble in chloroform and the dissolution of the depos-ited PL layer was likely upon deposition of the P3HT layer from achloroform solution. In fact, this was not the case and the OSdeposition resulted in a PL layer thickness drastic reductionbut not in its complete removal. An experimental evidence forthis is provided by the photoluminescence spectra reported inFig. S2. To improve the signal intensity, a fluorophore labeledPLs (Texas-red DHPE) was used.

The photoluminescence spectrum of just the phospholipidsfilms (blue curve) is particularly intense due to the presence ofthe fluorophore. The equally spaced fringes superimposed tothe main luminescent peak are attributed to interference effects,the distance between subsequent maxima being consistent withthe as deposited PL average thickness of approximately 5.3 μm.A photoluminescence signal can be recorded also for the bareP3HT film (black curve), even though less intense and red shifted.The photoluminescence line shape of the phospholipidic layercovered by the organic semiconductor (red curve), clearly showsthe contributions from the phospholipid layer.

In-Depth XPS Analysis. The spectroscopic and X-ray investiga-tions (vide infra) of the PL-P3HT can provide assessment onthe presence of both the phospholipid and the poly-3-hexylthio-phene layer but not necessary of a stacking structure with the bio-layer buried underneath the OS. Angle resolved XPS data (10) onthe PL-P3HTsample can definitively prove such a structure. Thephosphorous (P2p) and oxygen (O1s) XPS signals were taken asmarkers for the phospholipid layer while the sulfur (S2p) marksthe presence of the outer P3HT film. The abundance of each tar-get element associated to the PL inner layer was divided by thesulfur abundance. For the sake of comparison, ratios have beenfurther normalized by their maximum value. The data in Fig. S3show the elicited species normalized elemental ratios as a func-tion of the sampled depth. The lower the d∕dmax value, the moresurface sensitive the sampling is. The normalized ratios show that,as the sampling proceeds from the inner region (0.9 d∕dmax) tothe surface (0.2 d∕dmax), the P and O relative abundance (being infact correlated) drops evidencing how the phospholipid biolayeris, in fact, segregated underneath the organic semiconductor.

X-ray Specular Reflectivity (XSR). Experimental reflectivity curvesof P3HT and phospholipid as deposited layers, as well as thePL-P3HT stacking layers, are reported in Fig. S4 A.

The layer thickness d is derived from the thickness modulationfringes as d ¼ 2π∕ΔQ (11). ΔQ ¼ ð4π∕λÞ sin θ is the scatteringvector difference between two consecutive minima of the reflec-tivity curve, being λ the X-ray wavelength and θ the incidenceangle (half the scattering angle). In all curves of Fig. S4A, twomain intensity modulation frequencies are clearly visible: Alow frequency modulation ascribable to the thickness of thebioorganic layer and a high frequency one related to the under-lying silicon oxide layer. Although the data do not allow to readilydiscriminate between the P3HT and the phospholipid contribu-tions, their thickness values can be derived (from the comparisonwith the single layer blank measurements) to be 21.0 nm for the

bare P3HT layer, 5.6 nm for the bare PL laying underneath and27.6 nm for the PL-P3HT stacking-layers. Such data were alsoconfirmed by the simulations of the whole reflectivity curves (graycolor). Reflectivity measurements performed on a much thickerphospholipid layer show that it is actually formed by bilayersabout 4.5 nm thick stacked along the surface normal, as indicatedby the presence of sharp equally spaced peaks (Fig. S4B). Thischaracteristic feature shows that thicker PL films are composedby lipid molecules arranged in a lamellar structure. The samestacking in a lamellar structure is therefore inferred also forca. 6 nm thick phospholipid layers in the PL-P3HT sample.

PM-P3HTOFET Stacking-Layers Structure.The PM layer was retainedunderneath the OS after deposition as clear changes have beenseen upon illumination in the P3HT-PM optical spectrum(Fig. 1B). Such changes are ascribed to the photo-activation ofthe bR proton-pumping activity as no relevant changes couldbe seen on the bare P3HT film. XRS analysis of the PM-P3HT system gives a 1 μm thickness of the FBI layer in this case.The structure of the PM layer was studied by means of GISAXS.Here, the monochromatic X-ray beam is directed on a planar sur-face with a very small incident angle with respect to the surface.The scattered intensities can be recorded in the qz and qr direc-tions (out-of plane and in-plane scattering). A 1D-PSD-50 M de-tector (HECUS X-ray Systems) containing 1,024 channels ofwidth 54.0 μm and oriented perpendicular to the wafer plane,was used for the wafer alignment and for the detection of scat-tered signal only in the out-of-plane direction. Fig. S5 shows theout-of-plane GISAXS scattering coming from both the PM sam-ple and the PM-P3HT stacking layers deposited on Si∕SiO2. TheGISAXS curve obtained, in the same experimental conditions, forthe bare P3HT layer was identical to the curve obtained for thebare Si∕SiO2 wafer, hence was not reported. Both curves exhibitthe presence of an interlayer ordering typical of lamellar struc-tures, with the layers oriented parallel to the wafer plane direc-tion. First three orders are visible and have been labeled in thefigure. These peak repetitions are in agreement with an interlayerspacing of about 5.0� 0.5 nm which is close to the value reportedin the literature for a PM-dried phase (3). The PM-P3HT systemholds an even more ordered and defined structure as evidencedby the narrower peaks registered in the relevant curve, suggestingthat OS deposition generates a very smooth interface. Consider-ing the PM whole thickness (1 μm) and each lamella spacing(5 nm), about 200 lamellae are composing the whole PM layerin the PM-P3HT sample. Though each lamella is parallel tothe SiO2 surface, they can be directed upward or downward withrespect to the substrates, this meaning that the bR proteins in-cluded into the lamellae randomly expose the cytoplasmic(CP) to the extracellular (EC) side to the OS interface.

SA Functionality After Chloroform Spin Spreading. To evaluate theeffects of the treatment with chloroform also on the streptavidinlayer, particularly concerning its biotin-binding activity, a fluor-escence-quenching assay was performed on the SA-P3HTsystem.The biotin-binding capability of streptavidin can be evaluatedby measuring the intrinsic fluorescence of the protein beforeand after the biotin addition (12). A reduction in the proteinemission spectrum intensity is observed when biotin is boundto streptavidin due to the quenching of endogenous-tryptophanfluorescence. To perform the experiment, a 10 μg∕mL aqueoussolution of streptavidin was deposited by spin coating on theSi∕SiO2 substrate. The chloroform solvent was then spin spreadon the device at a rate of 2,000 rpm for 30 s, mimicking the or-ganic semiconductor deposition. The deposited protein was re-suspended in 100 μL of deionized water and the emissionspectra before and after the addition of 5 ng of biotin were re-corded in a microvolume fluorescence cuvette (Helmann). An in-cubation time of 15 min was allowed, before measuring the

Angione et al. www.pnas.org/cgi/doi/10.1073/pnas.1200549109 3 of 10

spectrum. The spectra were excited at a wavelength of 280 nmand the results are reported in Fig. S6. Here the comparisonof the resuspended SA treated with chloroform, before (bluecurve) and after biotin interaction (magenta curve) is reported.As it can be seen, the streptavidin-binding properties is retainedafter the treatment as a fluorescence-quenching is clearly seenwhose extension is comparable to that of a resuspended SA,not treated with chloroform and of a SA freshly preparedsolution.

FBI-OFET Sensors Electronic and Analytical Performances. FBI-OFETElectronic Performances. Typical FBI-OFET current-voltage char-acteristics (IDS −VDS) at different gate biases are reported inFig. S7. In particular, Fig. S7 A, B, and C are relevant to thePL-P3HT, PM-P3HT, and SA-P3HT stacking layers, respectively.PL-FBI OFET had a 100 nm thick SiO2 dielectric while the othertwo comprised a 300 nm thick SiO2. The panels placed below theI-V curves in Fig. S7 report the sqrtjIDSj vs. VG plots taken atVDS ¼ −80 V (VDS ¼ −30 V for the PL FBI-OFET). All theI-V characteristics exhibit linear and saturated regions as wellas current modulation and a quite low leakage current at lowsource-drain voltage is seen. This is remarkable as the fabricationprocedure does not involve any patterning of the organic activelayer or of the dielectric layer. Besides, to evidence the occur-rence of an extremely small hysteresis, the I-V characteristicsof the PM FBI-OFET were measured in the double-run mode.The device figure of merits has been extracted from thesqrtjIDSj vs. VG curve using the equation:

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiIDS

SATq

¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiW2L

Ciμ

r· ðVG −VTÞ

where Ci is 9 or 27 nF cm2, L ¼ 200 μm and W ¼ 4 mm. Thegraphical extrapolation of the data, averaged over 10 samples,resulted in the figures reported in Table S1, along with the datarelevant to a bare P3HT OFET, for comparison.

Permeability of the P3HT Film to Small Molecules. The analyte cap-ability to pass through the P3HT layer reaching the underneathbiological recognition element was demonstrated using immobi-lized horseradish peroxidase enzyme (HRP) interacting with achemiluminescent system constituted of luminol/H2O2/p-iodo-phenol. This choice was driven by the fact that luminol has a sizecomparable to biotin. To perform the determination, a100 μg∕mL HRP solution in water was deposited by spin coatingon the cleaned SiO2 surface, using the same procedure adoptedfor SA. After enzyme immobilization, the P3HTwas spin depos-ited from chloroform and finally the chemiluminescent systemwas added as a water solution droplet deposited on the OS sur-face. A clear chemiluminescent signal was indeed seen on HRP-P3HT system after few seconds following the chemiluminescentsolution addition. This proves the following: The enzyme biolayeris not removed by the spinning of the OS layer; the analyte, car-ried by water, is able to pass through the organic semiconductor,reaching the biolayer deposited underneath; and OS depositionand therefore the enzyme embedding, keeps this biospecies fullybioactive.

As already addressed, the P3HT is composed of grains and thepresence of voids between the grains can allow the percolation ofspecies down to the underneath biolayer. A P3HTannealing pro-cess is known to increase the grains’ size, lowering grain bound-aries area and voids size. As a further step, we wanted also toprove that a small biological molecule, such as insulin (molecularweight 5808 Da) can in fact pass through the P3HT. To this aim,P3HTwas spin deposited from chloroform on pressed KBr pelletand 4 μL of an insulin aqueous solution (1 μM) were dispensed onthe P3HT layer and incubated for 30 min at room temperature.Osmotic pressure drives the water flow through the OS layer into

the salt pellet. To quantify the amount of insulin adsorbed, theKBr pellet was washed with chloroform, to remove the organicsemiconductor film, and dissolved in 1 mL of water. The insulinpresence in this solution was monitored by measuring, uponexcitation at 280 nm, the intrinsic fluorescence of the peptide hor-mone occurring at ca. 340 nm. The data are reported in Fig. S8Aas a red curve. To evaluate the effect of P3HT film morphologyand thickness, an annealed (at 70 °C overnight) and twice spincoated P3HT films were tested too. The relevant spectra arethe green and blue curves, respectively. The emission spectrumof a 10 nM insulin solution was recorded (black curve) as a re-ference for quantification.

The obtained results clearly indicate that insulin molecules areable to flow through the P3HTeven when its permeability is sig-nificantly reduced as in the case of annealed or twice spin coatedfilms. However, the amount of insulin recovered after the diffu-sion through the P3HT film was satisfactory (higher than 80%)only in the case of the pristine organic semiconductor. In Fig. S8B,the red curve is again the spectrum of the dissolved KBr pellet foran insulin solution passed through a pristine P3HT film, excitedat 280 nm and recorded in a wider range (300–700 nm). This time,the data are reported as a log plot to evidence the presence ofsmall contributions. The peak at 340 nm is relevant to the insulinemission, while no appreciable signal is seen in the range 450–700 nm where the P3HT emission is expected (13). In fact, theweak peak at 560 nm is ascribable to the second order Bragg’sscattering (2 × 280 nm). In addition, excitation at 540 nm, knownto excite the P3HT fluorescence as well, results in negligible emis-sion above 550 nm, further proving that no significant amount ofP3HT is left after the chloroform washing. Further on, experi-ments (not shown) performed in absence of insulin give zerofluorescence independently from the excitation wavelengths.

Selectivity of SA FBI-OFET to Biotin. To evaluate if the electrical re-sponse of the SA-OFET biosensor after the biotin addition is onlydue to the specific interaction of the streptavidin with biotin mo-lecules, a saturated streptavidin-biotin tetrameric complex wasused as FBI. To prepare the saturated complex solution, the fol-lowing procedure was adopted. In Fig. S6, it was already shownthat a fluorescence quenching is associated to the binding of bio-tin to SA. The SA-biotin complex formation kinetic can be in factprobed until the equilibrium is reached. The data, for a stoichio-metric ¼ SA/biotin ratio, are shown in Fig. S9A, where thequenching is reported as a function of time. After ca. 25 min,the SA-biotin saturated tetrameric complex formation was com-pleted, and such a solution was used to deposit the completelyformed complexes on the OFET device following the conditionsused for the bare SA film. The electrical response of this sensor,after the addition of biotin solutions at different concentrations,was evaluated. The dose curve is reported as magenta trianglesFig. S9B. This panel reports exactly the same data of Fig. 2B, buton a different y-axis scale, to evidence the low responses datapoints. Such data clearly show that when an already saturatedSA protein is used as FBI, the device loses completely its sensi-tivity to biotin, acting very much like a not-binding system, such asthe BSA protein (red diamonds). The data in Fig. S9B, thanks tothe zooming of the y scale, clearly show also that the responses toBSA and SA-biotin complex overlay. Moreover, the data shownwith the black squares are taken on a bare P3HT device exposedto biotin. No response was seen in this case as well. These experi-ments strikingly prove that the SA FBI-OFET, besides being ex-tremely sensitive is also selective. Besides all the error barsassociated to each data point are better evidenced comparedto Fig. 2B, showing that the device is also highly repeatable.

As a further proof of selectivity, a positive control experimentwas performed in which a protein different from streptavidin butselective for biotin is used. In that, an antibiotin monoclonal anti-body (Biotin, Mouse IgG1-cod. 033700, Invitrogen) is used as

Angione et al. www.pnas.org/cgi/doi/10.1073/pnas.1200549109 4 of 10

FBI. This antibody is specific for biotin and has been used inplace of streptavidin. A 10 μg∕mL antibiotin antibody aqueoussolution was spin coated at 150 rpm for 50 min directly on thecleaned SiO2 surface. The P3HTorganic semiconductor was thendeposited directly on the antibody layer, using the method al-ready described in the FBI-OFETs Detailed Fabrication Proceduresection. The biosensing measurements were performed using bio-tin at different concentrations according to the procedure re-ported in the Electronic response measurements section. TheIDS −VG curves obtained adding pure water and two differentbiotin concentrations on antibody embedding device are reportedin Fig. S10. A response to biotin as current decrease can be ob-served and the ΔI∕Io mean values determined for the biotin sam-ples using the antibiotin are consistent with those obtained usingstreptavidin as receptor. These experiments further prove theFBI-OFET selectivity.

The SA-P3HT Limit of Detection. The biosensor detection limit(LOD) was evaluated by measuring 10 ppt biotin responses oneight different SA FBI-OFET devices. The same analysis was per-formed on other eight SA FBI-OFET devices by depositing justdeionized water droplets (blank sample with no biotin). For eachsample, the ΔI∕Io was calculated. The results obtained, the ΔI∕Iomean values, as well as the standard deviations (SDs) and relativestandard deviations (RSDs), calculated at 0 and 10 ppt biotinconcentration level, are reported in Table S2. The comparisonof the ΔI∕Io mean values determined for 0 and 10 ppt biotin sam-

ples showed that the biosensor response for 10 ppt biotin is ca. 9times higher than what obtained for the blank samples. In addi-tion, the within-assay RSD for 10 ppt biotin was below 15%,which is an outstanding result considering that no optimizationprocedure was carried out at this stage to improve device repro-ducibility.

For the sake of completeness, all the ΔI∕Io and SD connectedwith the data reported in Fig. 2B are reported in Table S3.

Selectivity of PL FBI-OFET to Anesthetics. In Fig. S11, the analyticalsensitivities, taken as the angular coefficient of the linear calibra-tion curves (ΔI∕Io vs. concentration), are reported for a PL andPM FBI-OFET. Analytes include halothane and diethyl-ether(both archetype anesthetics) as well as acetone. For a fair com-parison, the dose curves were plotted as a function of the volatileorganic saturated vapor fraction (14). Remarkably, PL-P3HTOFET sensitivity to both the anesthetics is higher than that toacetone. This datum is even more convincing considering thatwhile the bare P3HT OFETshows a comparably much lower sen-sitivity to halothane, a sensitivity, albeit low, is seen for the bareP3HT OFET to acetone. The latter being attributed to the effectof acetone adsorption on the bare polymer (14). A similar effectwas seen also for the interaction of the P3HTwith diethyl-ether.A very high sensitivity (opposite in sign to all the others in theplot, see main text) is seen also for the PM-P3HT OFET to ha-lothane.

1. Tero R, Watanabe H, Urisu T (2006) Supported phospholipid bilayer formation on hy-drophilicity-controlled silicon dioxide surfaces. Phys Chem Chem Phys 8:3885–3894.

2. Oesterhelt D, Stoeckenius W (1971) Rhodopsin-like protein from the purple mem-brane of Halobacterium halobium. Nature New Biol 233:149–152.

3. Shen Y, et al. (1993) Stabilization of the membrane protein bacteriorhodopsin to140 °C in two-dimensional films. Nature 366:48–50.

4. Urien M, et al. (2007) Field-effect transistors based on poly(3-hexylthiophene): Effectof impurities. Org Electron 8:727–734.

5. Verilhac JM, et al. (2006) Effect of macromolecular parameters and processing condi-tions on supramolecular organisation, morphology, and electrical transport propertiesin thin layers of regioregular poly(3-hexylthiophene). Synth Met 156:815–823.

6. Robinson L, Isaksson J, Robinson ND, Berggren B (2006) Electrochemical control of sur-face wettability of poly(3-alkylthiophenes). Surf Sci 600:L148–L152.

7. Torsi L, Dodabalapur A (2005) Organic thin-film transistors as plastic analytical sensors.Anal Chem 77:380A–387A.

8. Blanton T, et al. (1995) JCPDS—International Centre for Diffraction Data round robinstudy of silver behenate. A possible low-angle X-ray diffraction calibration standard.Powder Diffr 10:91–95.

9. Cho S, et al. (2006) Thermal annealing-induced enhancement of the field-effectmobility of regioregular poly(3-hexylthiophene) films. J Appl Phys 100:114503

10. Dilks A (1981) in Electron spectroscopy, theory, techniques and applications, edsBrundle CR, Baker AD (Academic, New York).

11. Pietsch U, Holy V, Baumbach T (2004) in High resolution X-ray scattering: From thinfilms to lateral nanostructures, 2nd Ed. (Springer, Berlin).

12. Gruber HJ, Kada G, Marek M, Kaiser K (1998) Accurate titration of avidin and strep-tavidin with biotin-fluorophore conjugates in complex, colored biofluids. BiochimBiophys Acta 1381:203–212.

13. Wang G, Swensen J, Moses D, Heeger AJ (2003) Increased mobility from regioregularpoly(3-hexylthiophene) field-effect transistors. J Appl Phys 93:6137–6141.

14. Torsi L, et al. (2002) Correlation between olgothiophene thin film transistor morphol-ogy and vapor responses. J Phys Chem B 106:12563–12568.

Angione et al. www.pnas.org/cgi/doi/10.1073/pnas.1200549109 5 of 10

Fig. S1. AFM images of (A) a pristine P3HT deposited on a SiO2 surface; (B) PM-P3HT layers; (C) SA deposited on a SiO2; (D) SA-P3HT layers.

Fig. S2. Photoluminescence spectra of the deposited Texas-Red labeled phospholipid layer (blue curve), the single P3HT layer (black curve) and the PL-P3HTstacking layers (red curve).

Angione et al. www.pnas.org/cgi/doi/10.1073/pnas.1200549109 6 of 10

Fig. S3. Angle-resolved XPS analysis of PL-P3HT layers. Normalized elemental ratios of P2p, O1s, and S2p taken on a PL-P3HT sample as a function of thesampled depth. d∕dmax 0.9 probes the inner region while d∕dmax 0.2 probes an outer region.

Fig. S4. X-ray specular reflectivity. Reflectivity curves of (A) P3HTand phospholipid single layers, and PL-P3HT stacking-layers; (B) as deposited, thicker PL film.

Fig. S5. Out-of-plane GISAXS intensity distribution for the PM and PM-P3HT.

Angione et al. www.pnas.org/cgi/doi/10.1073/pnas.1200549109 7 of 10

Fig. S6. Fluorescence spectra obtained for the streptavidin solution before (blue line) and after the biotin addition (magenta line). The measurements havebeen performed after the resuspension of the protein deposited on the device.

Fig. S7. Current-voltage characteristics, along with the relevant sqrtjIDSj vs. VG plots, for the three FBI-OFETs studied in this report. The I-V curve of the PMFBI-OFET has been measured in the double-run mode (barely evident).

Fig. S8. (A) Emission spectra (excited at 280 nm) of the dissolved KBr pellet solution after P3HT removal with chloroform; the black curve is the reference blankexperiment where a bare insulin solution is deposited directly on the KBr pellet; the other curves are relevant to the insulin solution that has passed throughthe pristine (red line), annealed (green line) and two times spin coated (blue line) P3HT layers. (B) Emission spectra of the KBr as for the red curve of panel (A)but excited at 280 nm (red curve) and 540 nm (green curve) and inspected over a larger spectral range.

Angione et al. www.pnas.org/cgi/doi/10.1073/pnas.1200549109 8 of 10

Fig. S9. (A) The formation of the SA-biotin tetrameric complex was monitored by a fluorescence quenching assay to evaluate the saturation. Biotin was addedto a SA solution and the complexation process reached the saturation after 25 min. (B) The dose curves of Fig. 2B (main text) are reported on an enlarged y scaleto evidence the contributions coming from the BSA FBI-OFET (red diamonds), the SA-biotin FBI-OFET complex FBI (magenta triangles) and bare P3HT OFET(black squares).

Fig. S10. I-V curves for an antibiotin embedding P3HT OFET exposed to water and to different concentrations of biotin.

Fig. S11. Relative analytical sensitivities (ΔI∕Io × ½saturated vapor fraction�−1) of PL-P3HT OFETsensor exposed to diethyl-ether, to halothane and acetone andPM-P3HT OFET exposed to halothane.

Angione et al. www.pnas.org/cgi/doi/10.1073/pnas.1200549109 9 of 10

Table S1. Figures of merit extracted from the FBI-OFETs and bare P3HT-OFET I-V curves. Values are expressed as the average over 10replicates and errors are the relevant standard deviations

P3HT-OFET PL FBI-OFET PM FBI-OFET SA FBI-OFET

ave (n ¼ 10) best ave (n ¼ 10) best ave (n ¼ 10) Best ave (n ¼ 10) Best

μFET (cm2 V−1 s−1) ð2.8� 0.79Þ × 10−3 4.0 10−3 ð8.6� 0.81Þ × 10−3 9.4 10−3 ð1.28� 0.28Þ × 10−3 2.7 10−3 ð1.17� 0.36Þ × 10−3 1.95 10−3

VT (V) 7.7 ± 3 5.5 −2.05 ± 0.67 −3.4 17 ± 7 7 27 ± 2 24Ion/off 1,168 ± 310 1890 2,444 ± 923 3500 181 ± 40 237 97 ± 19 146

Table S2. Table showing ΔI∕Io (average over eightreplicates) standard deviation (SD) and relative standarddeviation (RSD) obtained after the analysis of eight 0 and10 ppt biotin samples

ΔI∕Io (n ¼ 8) SD RSD%

Blank (H2O) 0.021 0.006 30Biotin 10 ppt 0.185 0.027 15

Table S3. ΔI∕Io (average over three replicates) and standard deviation (SD) data reported in Fig. 2B.

SA P3HT BSA SA-biotin complex

ΔI∕Io (n ¼ 3) SD ΔI∕Io (n ¼ 3) SD ΔI∕Io (n ¼ 3) SD ΔI∕Io (n ¼ 3) SD

Biotin 10 ppt 0.12 0.02 0.01 0.00 0.00 0.00 0.00 0.00Biotin 102 ppt 0.16 0.02 0.01 0.02 0.00 0.00 0.02 0.01Biotin 103 ppt 0.43 0.03 0.01 0.00 0.03 0.01 0.02 0.01Biotin 104 ppt 0.82 0.02 0.02 0.02 0.04 0.02 0.03 0.02Biotin 105 ppt 0.89 0.04 0.05 0.00 0.04 0.01 0.04 0.02

Angione et al. www.pnas.org/cgi/doi/10.1073/pnas.1200549109 10 of 10