Field Effect Transistors with Organic Semiconductor Layers

Dec 14, 2006 - 101, I-40129 Bologna, Istituto per lo Studio delle Macromolecole, Consiglio ... di Chimica Fisica e Inorganica and INSTM-UdR Bologna,...
1 downloads 0 Views 721KB Size
2030

Langmuir 2007, 23, 2030-2036

Field Effect Transistors with Organic Semiconductor Layers Assembled from Aqueous Colloidal Nanocomposites Chiara Dionigi,*,† Pablo Stoliar,*,† William Porzio,‡ Silvia Destri,‡ Massimiliano Cavallini,† Ivano Bilotti,§ Aldo Brillante,§ and Fabio Biscarini† Istituto per lo Studio dei Materiali Nanostrutturati, Consiglio Nazionale delle Ricerche, Via P. Gobetti 101, I-40129 Bologna, Istituto per lo Studio delle Macromolecole, Consiglio Nazionale delle Ricerche, Via Bassini 15, I-20133 Milano, and Dipartimento di Chimica Fisica e Inorganica and INSTM-UdR Bologna, UniVersity of Bologna, Viale Risorgimento 4, I-40136, Bologna, Italy ReceiVed August 10, 2006. In Final Form: October 16, 2006 We demonstrate field effect transistors based on organic semiconductor molecules dispersed in a self-organized polystyrene (PS) latex bead matrix. An aqueous colloidal composite made of PS and tetrahexylsexithiophene (H4T6) is deposited with a micropipet into the channel of a bottom-contact field effect transistor. The beads self-organize into a network whose characteristic distances are governed by their packing. The semiconductor molecules crystallize in the interstitial voids, leading to the growth of large interconnected domains. Depending on the bead size and the ratio between H4T6 and PS, the fraction of the different phases in the polymorph can be controlled. In the transistors where the H4T6 metastable “red phase” is the largest, the device response and the charge mobility are comparable to those of sexithienyl thin films grown by high-vacuum sublimation.

Introduction Organic semiconductors are materials of great interest for the development of novel electronics devices, such as field effect transistors (FETs), diodes, and photovoltaic junctions. The charge transport layer in such devices consists of a thin film of conjugated polymers or molecules. Polymer semiconductor thin films are deposited from solution by spin coating or patterned by ink-jet printing.1 Conjugated molecules are mostly processed by highvacuum sublimation.2-4 Molecular materials exhibit larger charge mobility values, because of the enhanced molecular order in the transport channel; hence, they are more attractive for FETs.5,6 Charge transport is correlated to the morphology of the active layer at length scales larger than the molecular ones. This was shown clearly in the case of high-vacuum sublimed oligomers.3 Controlling the molecular order at length scales comparable to the channel length of the device is extremely difficult for any established thin film technology.7-9 An attractive possibility consists of exploiting self-organization at all length scales. This can be pursued with materials, for instance semiconductor discotic liquid crystals, where cooperativity is * To whom correspondence should be addressed. E-mail: c.dionigi@ bo.ismn.cnr.it (C.D.); [email protected] (P.S.). † Istituto per lo Studio dei Materiali Nanostrutturati, Consiglio Nazionale delle Ricerche. ‡ Istituto per lo Studio delle Macromolecole, Consiglio Nazionale delle Ricerche. § University of Bologna. (1) Horowitz, G. J. Mater. Res. 2004, 19, 1946. (2) Dimitrakopoulos, C. D.; Malenfant, P. R. L. AdV. Mater. 2002, 14, 99. (3) Dinelli, F.; Murgia, M.; Levy, P.; Cavallini, M.; Biscarini, F.; de Leeuw, D. M. Phys. ReV. Lett. 2004, 92, 116802. (4) Stallinga, P.; Gomes, H. L.; Biscarini, F.; Murgia, M.; de Leeuw, D. M. J. Appl. Phys. 2004, 96, 5277. (5) Videlot, C.; Ackermann, J.; Blanchard, P.; Raimundo, J.-M.; Fre`re, P.; Allain, M.; de Bettignies, R.; Levillain, E.; Roncali, J. AdV. Mater. 2003, 15, 306. (6) Alik, M.; Klauk, H.; Zschieschang, U.; Schmid, G.; Ponomarenko, S.; Kirchmeyer, S.; Weber, W. AdV. Mater. 2003, 15, 917. (7) De Boer, R. W. I.; Gershenson, M. E.; Morpurgo, A. F.; Podzorov, V. Phys. Status Solidi A 2004, 201, 1302. (8) Leufgen, M.; Bass, U.; Muck, T.; Borzenko, T.; Schmidt, G.; Geurts, J.; Wagner, V.; Molenkamp, L. W. Synth. Met. 2004 146, 341. (9) Saya, D.; Coleman, A. W.; Lazar, A. N.; Bergaud, C.; Nicu, L. Appl. Phys. Lett. 2005, 87, 103901.

Figure 1. Scheme of the H4T6 molecular structure.

built into the design of the material.10 Recently, it was demonstrated by means of a combined fluidics/printing deposition of bilayer stripes that the charge mobility of self-organized nanostructures is two orders of magnitude larger with respect to that of the corresponding spin-cast material.11 This is the result of the better control in time and in space of supersaturation and precipitation in confined environments, which leads to the selforganization of the molecular semiconductor into very large molecularly ordered and oriented domains. In this paper, we present a “chimie douce” approach to form organic semiconductor layers with enhanced charge transport properties in organic FETs. Our process relies on the selforganization of the semiconductor molecules, viz., tetrahexyl(10) Roussel, O.; Kestemont, G.; Tant, J.; De Halleux, V.; Aspe, R. G.; Levin, J.; Remacle, A.; Geerts, Y. Mol. Cryst. Liq. Cryst. 2003, 396, 35. (11) Cavallini, M.; Stoliar, P.; Moulin, J. F.; Surin, M.; Leclere, P.; Lazzaroni, R.; Breiby, D. W.; Andreasen, J. W.; Nielsen, M. M.; Sonar, P.; Grimsdale, A. C.; Mullen, K.; Biscarini, F. Nano Lett. 2005, 5, 2422.

10.1021/la062371k CCC: $37.00 © 2007 American Chemical Society Published on Web 12/14/2006

Field Effect Transistors with Semiconductor Layers

Figure 2. Adsorption effect for 150 nm (squares) and 270 nm (circles) bead diameters. Two different scales are used to visualize the effects of both diameters: the right and top axes are for the 270 nm beads, the left and bottom axes are for the 150 nm beads. The linear fit (green line) for 270 nm PS is reported in the graph. The experiments were performed in the same laboratory conditions, keeping constant the volume of the H4T6/PS colloidal dispersion (see the text). In the text, the samples are named with the capital letters reported in the graph.

Langmuir, Vol. 23, No. 4, 2007 2031

Figure 4. Optical image (top) of dendritic domains under polarized light in a dried colloidal composite made of H4T6 and 150 nm PS and SEM images of a domain (bottom left) and a zoom of it (bottom right).

Figure 3. Schematic model of the formation process of a colloidal composite from a suspension of beads and a saturated solution of H4T6.

sexithiophene (H4T6),12-16 in a confined environment defined by the assembly of mesoscopic polystyrene beads, hereafter called PS, decorated with H4T6. The decorated beads are cast in the channel of field effect transistors at ambient conditions. H4T6 molecules are slowly released on the dielectric surface of the transistor channel, and crystalline domains grow in the interstitial cavities. The mesoscopic and nanoscale channels in the template dictate the length and time scales for the deposition and nucleation. These devices exhibit a charge mobility of 10-3 cm2/(V s), viz., 1 order of magnitude lower than that of high-vacuum grown sexithienyl thin films,3,4 a threshold voltage close to 0 V, a subthreshold slope of 0.2 V/decade, and on/off ratios in excess of 102. The enhanced device response is correlated to the dominance of two polymorphs (red (R) and meso (M) phases) by means of X-ray diffraction (XRD) and confocal Raman spectroscopy. (12) Botta, C.; Destri, S.; Porzio, W.; Bongiovanni, G.; Loi, M. A.; Mura, A.; Tubino, R. Synth. Met. 2001, 122, 395. (13) Destri, S.; Ferro, D. R.; Khotina, I. A.; Porzio, W.; Farina, A. Macromol. Chem. Phys. 1998, 199, 1973. (14) Borghesi, A.; Sassella, A.; Tubino, R.; Destri, S.; Porzio, W. AdV. Mater. 1998, 10, 931. (15) Porzio, W.; Giovanella, U.; Botta, C.; Pasini, M.; Destri, S.; Provasi, C.; Rossi, V. Org. Electron. 2004, 5, 59. (16) Botta, C.; Destri, S.; Porzio, W.; Sassella, A.; Borghesi, A.; Tubino, R. Opt. Mater. 1999, 12, 301.

Figure 5. XRD patterns of samples C and D (shifted for clarity). The peaks indicated by M, Y, and R map typical orientations of the three crystalline phases of H4T6, also observed in films grown by high-vacuum evaporation onto the silica substrate, under different conditions.

Experimental Section Decoration of PS with H4T6. Syntheses of both monodisperse PS and H4T6 are reported elsewhere.13,17 H4T6 is a semiconductive oligomer stable in the solid phase under ambient conditions (see Figure 1); it is insoluble in water, (17) Dionigi, C.; Nozar, P.; Di Domenico, D.; Calestani, G. J. Colloid Interface Sci. 2004, 275, 445.

2032 Langmuir, Vol. 23, No. 4, 2007

Dionigi et al.

Figure 6. H4T6 phase diagram (top row) and crystal size (bottom row) by XRD analysis for the composite with bead diameters of 150 nm (left column) and 270 nm (right column). The calculated size of the tetrahedral cavities is indicated by the solid line. Geometrical calculations demonstrate that the size of the perpendicular bisector of the largest inscribed equilateral triangle into the sphere interstices in close-packing configuration (either hexagonal or face-centered cubic) is equal to 0.233D, the sphere diameter. partially soluble in ethanol (∼0.5 mg/mL), and totally soluble in organic solvents such as acetone. In our work, the bead decoration is driven by highly hydrophobic interactions between H4T6 and polymeric nanobeads in a hydrophilic medium. When an acetone/ethanol solution of H4T6 is mixed with water, H4T6 starts to nucleate. We found that a precise water/organic ratio is able to induce the nucleation of H4T6 into clusters that make the liquid phase opaque. PS decoration was performed by mixing 10 mg of H4T6 (previously dissolved in 4 mL of acetone/ethanol (1:3) by volume) with 20 mL of an aqueous PS dispersion containing 20 mg of PS (D ) 150 ( 4 nm, D ) 270 ( 7 nm). The final solvent mixture resulted in a 5:1 (v/v) water:organic solvent ratio. H4T6 clusters were collected by polymeric nanobeads, and a homogeneous colloidal suspension of decorated beads was formed. Photon correlation spectroscopy (PCS) analysis and Z-potential measurements confirmed the adsorption of H4T6 on the nanobeads, as a decrease in both the external nanobead charge and hydrodynamic diameter was observed. Decorated beads were separated from the liquid phase by centrifugation at 700-3000g acceleration. The dry supernatant was dissolved in acetone and analyzed by spectrophotometry at λ ) 411 nm, to estimate the amount of adsorbed H4T6. The wet deposit spontaneously dispersed as a highly viscous colloidal suspension of the H4T6-decorated beads. A total of 58%

of the starting H4T6 was contained in the surnatant liquid phase, while 42% remained in the viscous phase, whose volume is 10% of the total volume. These proportions are highly reproducible. Preparation of Transistors. The FET test patterns were given by Philips Research Laboratories (courtesy of Dr. D. de Leeuw). These are prepared on top of a heavily n-type-doped silicon substrate, acting as a gate, with a 200 nm thick thermally grown SiO2 insulating layer which is primed with hexamethyldisilazane (HMDS). The insulating layer has a capacitance per area unit of 17 nF/cm2. Two interdigitated 150 nm thick evaporated gold pads on a Ti adhesion layer form the drain and source contacts and define a transistor channel 10 µm long and 10 mm wide. Before the deposition, the test pattern was cleaned by using boiling acetone and then acetone vapor and immediately dried by using filtered dry nitrogen. A drop of approximately 1 µL was deposited on top of each test pattern with a tapered glass micropipet. After drying, a semispherical deposit a few hundred micrometers thick was formed, completely covering the test pattern. In the case of incomplete coverage, the channel width value has to be properly corrected. XRD Measurements. Room-temperature XRD measurements were performed in Bragg-Brentano geometry using a Siemens D-500 apparatus equipped with Soller slits and narrow windows (0.3°/ 0.15°). Cu KR radiation at 40 kV × 40 mA power was used. Raman Measurements. Raman spectra were recorded with the Jobin Yvon spectrometers T64000 (excitation from a Kr laser, 647.1 nm) and LABRAM (excitation from a He/Ne laser, 632.8 nm). The

Field Effect Transistors with Semiconductor Layers

Langmuir, Vol. 23, No. 4, 2007 2033

Figure 8. Transfer (top) and output (bottom) characteristics of the transistor shown above, with a channel length of 10 µm, channel width 1450 µm, and dielectric capacitance 18 nF cm-2. The transfer characteristic was obtained with Vds ) -21 V.

Figure 7. (Top) Colloidal composite deposition on FET. The drying process promotes the growth of H4T6 crystals inside the tetrahedral cavities. Cavity projections are represented by cyan triangles. The triangle height (h) is calculated from the bead diameter. (Bottom) H4T6/PS composite as an FET channel: optical image under polarized light of sample C (labeled according to Figure 2) deposited on a transistor prototype. In the inset the micropipet with the bead dispersion approaching the test pattern for the deposition on the FET is shown. laser output power focused on the sample was reduced with a neutral filter, whose optical density was selected in each experiment to avoid phase transformation (yellow (Y) to R) produced by heating. A confocal optical microscope (Olympus BX40) was interfaced to the spectrometers, and 50× or 100× objectives were used. The spatial resolution, given by the laser spot area, ranged from 0.88 to 1.05 µm, depending on the numerical aperture of the microscope objective and on the wavelength of the laser line. The point to point variation of the Raman spectra was obtained by scanning the surface, typically a few tenths of micrometers, with steps as close as 2 µm, with an xy motorized stage. Confocal Raman mapping was then automatically achieved after selection of the area to be sampled and the number of points to be measured, their density being chosen according to the size of the specimen and the microscope objective used. Theoretical Basis: Measurement of the Transistor Response and Extraction of the Parameters. The mobility is extracted from the transfer characteristic Id ) f(Vgs) in the saturation regime (|Vds| > |Vgs - Vt|): µsat ) 2(L/WCi)(dId/dVgs)2, where Id is the drain current, W and L are the channel width and length, µsat is the saturated charge carrier mobility, Ci is the dielectric layer capacitance per area unit, Vt is the threshold voltage, and Vgs and Vds are the gate-source and the drain-source potentials. The subthreshold slope, on/off ratio, and threshold voltage shift were extracted from the transfer characteristic for Vds ) -21 V.

The transfer characteristics were measured at ambient conditions in darkness. The values of the drain current were measured by using a Keithley 6430 subfemtoammeter, which also was used to polarize the channel. During the acquisition, a Keithley 3930A multifunction synthesizer connected to a homemade voltage amplifier, varying with a rate of 100 V/s, supplied the gate voltage.

Results and Discussion The building blocks are formed upon the effect of highly hydrophobic interactions18 between H4T6 and PS in an aqueous medium formed by an acetone/ethanol solution of H4T6 mixed with water. This process is different from most of the chemical methods for core-shell bead synthesis which exploit ionic or polar interactions.19-24 In these conditions, H4T6 nuclei form in the solution and then adsorb on PS,25 forming a stable colloidal suspension. The relative amount of H4T6 adsorbed on PS depends critically on the bead size and the initial H4T6/PS ratio, as shown in Figure 2 for two different bead diameters. The 150 nm beads exhibit a constant adsorption yield in the range from 20% to 44% by weight of initial H4T6/PS. This evidence is consistent with the process depicted in Figure 3 where first PS collects H4T6 nuclei after mixing of the aqueous PS suspension with the H4T6 (18) Kohut-Svelko, N.; Reynaud, S.; Dedryevre, R.; Martinez, H.; Gonbeau, D.; Francois, J. Langmuir 2005, 21, 1575. (19) Caruso, F.; Lichtenfeld, H.; Giersig, M.; Mo¨hwald, H. J. Am. Chem. Soc. 1998, 120, 8523. (20) Radtchenko, I. L.; Sulkhorukov, G. B.; Gaponik, N.; Kornowski, A.; Rogach, A. L.; Mo¨hwald, H. AdV. Mater. 2001, 13, 1684. (21) Donath, E.; Sulkhorukov, G. B.; Caruso, F.; Devis, S. A.; Mo¨hwald, H. Angew. Chem., Int. Ed. 1998, 37, 2202. (22) Caruso, F. AdV. Mater. 2001, 13, 11. (23) Jang, J.; Oh, J. H. AdV. Funct. Mater. 2005, 15, 494. (24) Khan, M. A.; Armes, S. P. AdV. Mater. 2000, 12, 671. (25) Rigaud, J. L.; Levy, D.; Mosser, G.; Lambert, O. Eur. Biophys. J. 1998, 27, 305.

2034 Langmuir, Vol. 23, No. 4, 2007

Figure 9. Evolution of the on/off ratio vs rescaled mobility and conductivity for the samples indicated in Figure 2. In sample A the amount of H4T6 is not sufficient to form an electrical percolation path. Samples B-D show an enhancement of the field effect mobility of the composite. The apparent field effect mobility of the device largely increases, without resulting in a correlated change in the bulk conductivity. Moreover, as sample D shows, any further increment of the initial H4T6/PS is disadvantageous for the on/off ratio because it produces a much larger increment of the bulk conductance with respect to the augment in the mobility. Point E corresponding to a flocculation regime (Figure 2) does not belong to the same scenario as points C and D that represent the fully infilled bead array.

ethanol/acetone solution and then forms a close-packed structure upon centrifugation. The H4T6 nuclei carried by the beads fill the cavities, as 26% of the volume in a close-packed structure is empty. This explains the plateau at about 20% in weight (the densities of the two materials are comparable). Above a critical initial H4T6/PS ratio, the available PS surface is insufficient to collect the excess nuclei, and these coalesce independently of the beads. This parasitic flocculation of H4T6 is a competing parallel process that manifests itself in the linear rise of the adsorption curve. For 270 nm beads, the same conditions as for 150 nm beads were used, but the linear dependence throughout the range suggests that H4T6 infilling of the close-packed beads is not achieved. In this case, the H4T6 flocculation hinders the infilling process because of the decreased surface area. After centrifugation, a wet H4T6/PS deposit is collected. This deposit spontaneously disperses in the residual moisture in the form of a “slurry”, hereafter termed a colloidal composite, which contains H4T6 adsorbed on and between the beads. The colloidal composites of 270 and 150 nm beads were cast on silicon oxide substrates with a pipet. Depending on the amount of adsorbed H4T6, the dry deposited film of the colloidal composite of 150 nm beads forms large dendritic domains. These domains exhibit birefringence under a polarizing microscope. Domains of the H4T6/PS composite larger than 50 µm are visible in the optical image in Figure 4 (top). The morphology of the 150 nm deposit is shown in the scanning electron micrographs (Figure 4 (bottom)). The scanning electron microscopy (SEM) images show dendritic domains made of

Dionigi et al.

densely packed beads embedded in an H4T6 matrix. The birefringent domains are not observed when films are drop cast from H4T6 solution without beads or from colloidal composites made of larger beads (270, 340, and 415 nm diameter). XRD patterns on PS/H4T6 composites deposited on glass slides (Figure 5) demonstrate the presence of crystals of H4T6. The three phases reported earlier, viz., Y, R, and M phases,12-16 are present. The known phases, already present in the dry colloidal composite, appear in different proportions depending on both the size and the amount of PS in the composite, as shown in Figure 6. This is a consequence of the different thermodynamic stabilities of the known phases. The Y phase is the most thermodynamically stable at room temperature, while the R and M phases are kinetically favored.12-16 The crystallization rate is controlled by the evaporation of the liquid phase, which determines the transition from thermodynamic to kinetic control. In the colloidal composite made of 150 nm beads, the volume of liquid phase infilling the cavities is small and is quickly evaporated. The deposition is under kinetic control; thus, R and M phases are preferentially formed. In the colloidal composite with 270 nm beads and a high H4T6/PS ratio, the amount of liquid in the interstitial space or surrounding the flocculated deposit is larger and the solvent evaporation process takes place in a longer time. The system is under thermodynamic control, and the Y phase is preferentially formed. The coherence length of the crystals, viz., the mean domain size, extracted from analysis of the XRD data, is controlled by the crystallization inside the interstitial sites as shown in Figure 6. There is a clear correlation between the size of the tetrahedral sites in the close-packed structures of the beads26 and the coherence length of the H4T6 crystals infilled in PS structures, independently of the phase. Our experimental evidence demonstrates that the composite composition, semiconductor crystal phases, crystal size, and distribution arise from a spontaneous process. The H4T6/PS composite has been then used to form the transport layer of an FET device. Colloidal composites made of H4T6 and PS of diameter sizes of both 150 and 270 nm were deposited on top of a bottom contact FET transistor test pattern as shown in Figure 7. In the case of H4T6/PS composites made with 150 nm beads, we explored systematically the effect of the amount of H4T6 infilled. A clear FET behavior was observed in the condition of “full infilling” of the cavities of the PS close-packed array (samples C and D). Similar electrical characteristics were observed throughout the range where the adsorbed H4T6/PS concentration is constant at about 20% by weight. The electrical data demonstrate clearly an excellent reproducibility of the method. Figure 8 presents the transfer characteristics of a typical transistor made in conditions similar to those of sample C. The charge field effect mobility extracted from the transfer characteristics is 4 × 10-3 cm2/(V s) in the saturation regime. The subthreshold slope is 0.2 V/decade. The device exhibits a low hysteresis that manifests itself as a shift in the threshold voltage of 260 mV depending on the direction of the gate sweep. The on/off ratio exceeds 100, due to a relatively large off current.27 The behavior of all the 150 nm series is summarized in Figure 9 where the field effect mobility is plotted versus the bulk conductance (viz., off-current/bias voltage) of the composite (26) Hulteen, J. C.; Treichel, D. A.; Smith, M. T.; Duval, M. L.; Jensen, T. R.; Van Duyne, R. P. J. Phys. Chem. B 1999, 103, 3854. (27) Brown, A. R.; Jarrett, C. P.; de Leeuw, D. M.; Matters, M. Synth. Met. 1997, 88, 37.

Field Effect Transistors with Semiconductor Layers

Langmuir, Vol. 23, No. 4, 2007 2035

Figure 10. (Top left) Raman spectra of the red (R, lower trace) and yellow (Y, upper trace) phases of H4T6. (Top right and bottom) Raman mapping referred to the selected area (optical pictures in the insets, with the scale bars indicating 10 µm) of samples B, C, and E on FET. The phase purity is monitored, and an immediate identification of the crystal phase with any amount of physical impurities is found (transistors C and E are shown). When a significant phase mixing is present (transistor B), a patchwork of colors appears. Raman mapping also proves that the material is present all over the transistor. Raman mapping can instead probe the identification of the phase purity, not easy to obtain from optical images. The color scale is conventionally selected to emphasize the phase mixing in the crystal and is expressed as the intensity ratio R/Y of selected Raman bands as described in the text.

material for different H4T6/PS ratios. The transport properties are rescaled by the device geometrical parameters. The transport layer in the FET is formed by the very first few monolayers in contact with the dielectric.3,4,28 When the amount of H4T6 is below the percolation threshold, no current flows. Consequently, the increment of the initial H4T6/PS (from sample A to sample C) enhances the on/off ratio by increasing the µ/σ ratio.27 In Figures 2 and 6 we label as “working transistors” the devices whose on/off ratio is in excess of 10. In the case of composites prepared with 270 nm beads, no FET behavior over the detection limit was observed upon an equally systematic analysis. In this case an incomplete coverage of the device channels is possibly responsible for the lack of an FET response; viz., the H4T6 crystal domains inside the tetrahedral cavities between the beads are not connected. From a morphological point of view (Figure 7), we found that the working transistors typically appear largely covered by dendritic domains, whose branches stretch across the channel. From the optical images, a 60-70% coverage of the channel was estimated. This dendritic morphology does not always determine the transistors’ operations. FET behavior was found also in transistor channels where no connected dendritic domains were “optically” detected. We have applied confocal Raman microscopy to FET transistors to establish a relationship between the crystal structure inside the FET channel and device behavior. Confocal Raman spectromicroscopy allows us to identify the known crystal phases of H4T6 and map the spatial distribution of the phase inhomo(28) Ruiz, R.; Papadimitratos, A.; Mayer, A. C.; Malliaras, G. G. AdV. Mater. 2005, 17, 1795.

geneity of each sample29 in the transistor channel. The phase identification by Raman spectra relies on the different spectral profiles in the region of the totally symmetric ring vibration.12 It is quite common to observe that Y and R phases can also coexist in different crystal grains or thin films of deposited material. Phase mapping of different devices has been obtained by monitoring the spectral regions where each structure shows its most intense Raman band, 1450 and 1462 cm-1 for the R and Y phases, respectively. For a given sample we represent in Figure 10 the relative amount of both phases by reporting the ratio R/Y in a conventional scale of colors (red, R phase; green/yellow, Y phase). A deconvolution of the vibrational bands with background suppression results in a reliable representation of the data. When no phase mixing is present, a basically homogeneous color image is observed. In Figure 10 the optical images of selected transistors, with the corresponding Raman image, are shown. These data suggest that electrical conduction is not directly related to a percolation layer formed by the dendritic domains. Instead, the percolation path is formed by H4T6 material undetected by optical means, and only in the case of adsorbed H4T6/PS below 16% a real percolation path does not occur. Even if the transport in the reported FET configuration occurs in two dimensions (2D),3 it is peculiar that we can compare our 2D percolation threshold, 16%, with the three-dimensional percolation system reported in ref 30. The working transistor provides direct evidence that H4T6 molecules assemble into an organized layer in the proximity of (29) Brillante, A.; Bilotti, I.; Della Valle, R. G.; Venuti, E.; Masino, M.; Girlando, A. AdV. Mater. 2005, 17, 2549. (30) Mezzenga, R.; Ruokolainen, J.; Fredrickson, G. H.; Kramer, E. J; Moses, D.; Heeger, A. J.; Ikkala, O. Science 2003, 299, 1872.

2036 Langmuir, Vol. 23, No. 4, 2007

the dielectric surface. This strongly supports a scenario where the H4T6 aggregates carried by the beads crystallize inside template channels owing to the slower solvent evaporation. The variability of domain formation with respect to the bead diameter can be ascribed to the two scales of the process: (a) a submicrometer scale which is imposed by the templating beads; (b) a nanometer scale characteristic of diffusion/aggregation/ crystallization of H4T6 molecules and clusters. The latter process should not be critically influenced by the mesoscopic selfassembly of the templating beads as long as their characteristic size does not restrict flow or diffusion of H4T6. However, the size of the empty space among the beads (tetrahedral cavities) critically influences the time of the solvent evaporation discriminating both phase content and crystal size. Raman spectroscopy shows that the morphological and compositional characteristic of the dry PS/H4T6 composite in the FET channel is in agreement with a dominant concentration of R and M phases as detected by XRD of composites on a glass substrate. The strength of π-stacking decreases in the order R > M > Y as confirmed by the trend of the intermolecular distances.12-16 The orientation of the active molecules in the FET channel5,6,31 is the factor affecting the charge mobility. Even if we do not have direct evidence of the molecular orientation in proximity of the substrate in the FET channel, the orientations of both R and M phases in the bulk PS/H4T6 are the same as those observed in thin films of H4T6 obtained by high-vacuum evaporation onto dielectric substrates using different deposition rates and substrate temperatures.16,32 In all the bulk phases the assumed orientations are the most suitable for large charge mobility (Figure 5) with the molecules (31) DeLongchamp, D. M.; Jung, Y.; Fischer, D. A.; Lin, E. K.; Chang, P.; Subramanian, V.; Murphy, A. R.; Fre´chet, J. M. J. J. Phys. Chem. B 2006, 110, 10645. (32) Porzio, W.; Giovanella, U.; Pasini, M.; Botta, C.; Destri, S.; Provasi, C. Thin Solid Films 2004, 466, 231. (33) Dinelli, F.; Moulin, J. F.; Loi, M. A.; Da Como, E.; Massi, M.; Murgia, M.; Muccini, M.; Biscarini, F.; Wie, J.; Kingshott, P. J. Phys. Chem. B 2006, 110, 258.

Dionigi et al.

standing up onto the substrate and facing each other to maximize π-stacking.12-16,32 As a comparison, thin film transistors by spin-casting of H4T6 solutions exhibit mobilities of 10-5 cm2/(V s). It has to be said that, in the case of these solutions, the wetting of the substrate is a problem, since H4T6 wets preferentially the gold pads and it is very difficult to cast a continuous film between the drain and source. As a result spin casting on the test pattern yields a percolation path not sufficiently connected.

Conclusions We have demonstrated the possibility to induce H4T6 molecular ordering on an FET by controlling the crystallization of an organic semiconductor within a template made of selfassembled dielectric beads. The crystals of all the phases formed within the tetrahedral interbead cavities are oriented with respect to the S/D electrodes along directions allowing charge transport. It is important to underline that the conjugated homologue sexithienyl grown as a highly ordered26,33 thin film on a device structure3 by ultra-high-vacuum sublimation in an OMBD apparatus exhibits on the same test patterns a charge mobility on the order of 0.01-0.05 cm2/(V s). Our dispersion-cast transistors with 150 nm beads exhibit a charge mobility that is 2-10 times smaller. The method we have proposed for depositing the active layer from bead carriers is completely general for soluble semiconductors. The complete use of water-based solutions offers the interesting opportunity of fabricating organic electronic devices by a “chimie douce” approach. Acknowledgment. This work was supported by the EU Integrated Project NAIMO (Grant NMP4-CT-2004-500355). We thank Dr. Maria Antonietta Loi for help with the confocal laser scanning microscopy. P.S. acknowledges support of the ICTP Programme for Training and Research in Italian Laboratories, Trieste, Italy. LA062371K