highly crystalline c8‐btbt thin‐film transistors by …the first step is the growth of...

CommuniCation

1703864 (1 of 8) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advmat.de

Highly Crystalline C8-BTBT Thin-Film Transistors by Lateral Homo-Epitaxial Growth on Printed Templates

Robby Janneck,* Nicolas Pilet, Satya Prakash Bommanaboyena, Benjamin Watts, Paul Heremans, Jan Genoe, and Cedric Rolin*

R. Janneck, S. P. Bommanaboyena, Prof. P. Heremans, Prof. J. Genoe, Dr. C. RolinIMECKapeldreef 75, Leuven B-3001, BelgiumE-mail: [email protected]; [email protected]. Janneck, Prof. P. Heremans, Prof. J. GenoeKU LeuvenESATKasteelpark Arenberg 10, Leuven B-3001, BelgiumDr. N. Pilet, Dr. B. WattsPaul Scherrer Institut5232 Villigen PSI, Switzerland

The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adma.201703864.

DOI: 10.1002/adma.201703864

The most promising methods such as meniscus-guided coating techniques[10–15] or thermal gradient annealing,[16] all employ a unidirectional solidification approach that separates nucleation and subsequent growth. Even though these techniques yield thin films with large single crystalline regions, these layers exhibit many morphological defects such as cracks,[16,17] voids, or parallel ribbon growth[12–14] that affect the film over multi ple length scales. Such defects lower the electrical performance and are a major cause for spread in transport characteris-tics. Despite intensive efforts to optimize growth recipes and improve setups,[12–15] the formation of perfectly smooth and uni-form single crystalline films has remained elusive in these kinetically driven unidi-rectional solidification processes.

Here we report on a double-step method to grow a closed and dense highly crystalline thin film over sev-eral mm2. The first step is the growth of individual, noncon-nected single-crystal ribbons from solution by zone-casting, a meniscus-guided coating technique. The second step is the growth of typically ten monolayers of the same organic semi-conductor by thermal evaporation. The grown layer is found to be epitaxially templated on the crystal seeds of the ribbons, such that it can be considered homo-epitaxial. Importantly, the mole-cules landing in the gap between ribbons diffuse to the ribbon edge and condense there, forming a homo-epitaxial layer of sev-eral monolayers high that closes cracks, voids, and gaps over distances up to several microns between parallel single-crystal-line ribbons present in the initial film. This method is success-fully applied to the growth of the small molecule 2,7-dioctyl[1]benzothieno[3,2-b][1]benzothiophene (C8-BTBT). Resulting transistors have much superior characteristics to transistors formed by simple meniscus-guided crystalline thin-film semi-conductor layers.

We distinguish three main types of defects that can be formed during unidirectional crystallization. First, cracks in the semiconductor film are often caused by thermal contrac-tion[16,17] or a phase change[18,19] when the film cools down after processing above room temperature. Second, in meniscus-guided coating techniques, a stick–slip motion of the meniscus gives rise to defect patterns parallel to the receding contact line.[20,21] Third, transversal “fingering” instabilities of the solid-ification front result in branched defects that are orthogonal to

Highly crystalline thin films of organic semiconductors offer great potential for fundamental material studies as well as for realizing high-performance, low-cost flexible electronics. The fabrication of these films directly on inert substrates is typically done by meniscus-guided coating techniques. The resulting layers show morphological defects that hinder charge transport and induce large device-to-device variability. Here, a double-step method for organic semiconductor layers combining a solution-processed templating layer and a lateral homo-epitaxial growth by a thermal evaporation step is reported. The epitaxial regrowth repairs most of the morphological defects inherent to meniscus-guided coatings. The resulting film is highly crystalline and features a mobility increased by a factor of three and a relative spread in device characteristics improved by almost half an order of magnitude. This method is easily adaptable to other coating techniques and offers a route toward the fabrication of high-performance, large-area electronics based on highly crystalline thin films of organic semiconductors.

Thin-Film Transistors

The latest generation of organic semiconductors display excel-lent characteristics, with charge mobilities surpassing those of amorphous silicon thin film transistors (TFTs) that are com-monly used in today’s flat panel displays.[1–5] The integration of organic TFTs (OTFTs) into real applications requires high performance and low spread of the electrical characteristics.[6] As transport properties are greatly influenced by the micro-structure of the organic layer, single crystalline films offer the greatest potential for high-performance OTFTs.[7–9]

Therefore, researchers focus on fabrication techniques to form highly crystalline thin films on large area inert substrates.

Adv. Mater. 2017, 29, 1703864

© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim1703864 (2 of 8)

www.advmat.dewww.advancedsciencenews.com

the receding contact line: the parallel ribbons.[20–23] The origins of thermal cracks is well understood and engineering strate-gies exist to limit their occurrence.[19,24,25] On the other hand, the two other types of defects result from kinetic instabilities arising in the nonequilibrium unidirectional solidification system.[26–28] Understanding this system requires developing further multiscale, multiphase, shape-changing physical models of the fluid and solidification dynamics.[29,30] Despite interesting developments in instrumentation,[10,14,31] it is doubtful that meniscus-guided coating techniques can be engi-neered to deliver defect-free, uniform single-crystalline thin films of organic semiconductors (see Note 1 in the Supporting Information for further elaboration on defect formation).

To illustrate this, we grew thin films of C8-BTBT by zone-casting using previously optimized processing conditions.[12] At low magnification in polarized optical microscopy, the films appear highly ordered as they display single-crystalline domains covering areas of several mm2 (see Figure S1, Supporting Infor-mation). At closer inspection however, these films are far from ideal and present many defects (see Figure S2, Supporting Information): stick–slip thickness modulation, cracks and, above all, ubiquitous parallel ribbons (see Figure 1a). Ribbons typically have a width of 2–5 µm and are separated by slightly narrower gaps. The ribbon thickness is ≈24 nm, corresponding to ≈8 stacked monolayers (ML) of C8-BTBT. Parallel ribbons have the same crystal orientation when they stem from the same nucleation event. In C8-BTBT, ribbons tend to grow either along the [100] or the [110] direction that is set by the propaga-tion of fast-growing facets in a mass-supply limited system.[17] Depending on the orientation of the initial nucleus, the direc-tion best aligned with the coating direction is favored. A mixture of the two competing directions, resulting in a chevron-shaped morphology, is also often observed (see Figure S2d, Supporting Information).

In order to repair these omnipresent defects, we depos-ited supplementary C8-BTBT by thermal evaporation in a

high-vacuum chamber. Figure 1 shows the evolution of parallel ribbons morphology with an increasing amount of evaporated monolayers. The growth on the top surface of the ribbons is classical[32]: 1 ML thick islands first nucleate and then grow laterally by diffusion-limited aggregation (Figure 1b) until they coalesce and form a continuous film that covers almost all of the ribbon (Figure 1c). The next ML nucleates and starts to grow slightly before the completion of the previous ML (Figure 1c).[33] Beyond a certain evaporated thickness (>5 ML), we observe the nucleation of 3D elongated needles, a common feature in the growth of many organic semiconductors.[34] The growth in the gaps between ribbons proceeds along a different path. No nucleation is ever observed within the gaps: Molecules impinging on the substrate surface have sufficient diffusion length (at least >1 µm) to reach ribbon sidewalls before they can participate in a nucleation event. Therefore, gap closing proceeds by an expansion of the >5 ML tall ribbon sidewalls. Interestingly, these walls remain steep until they meet the opposing sidewall. Just before the ribbons are merged, we observe only a very thin gap separating the ribbons (Figure 1d). With further evaporation, neighboring ribbons start to merge in random spots, first leaving behind small holes (Figure 1e) and then fully bridging the initial gaps (Figure 1f,i). Finally, once the gaps are closed, overlaying monolayers formed on top of the ribbons propagate sideways (Figure 1g,h), displaying the terraced morphology visible in Figure 1j.

The growth by thermal evaporation of C8-BTBT on a bare substrate is a good reference to interpret growth patterns (see Figure S3, Supporting Information). In this case, at low nominal thickness (0.5 ML), individual islands with 2 ML thickness are separated by several µm, confirming the long dif-fusion length of adsorbed molecules.[35] Upon further growth, islands thicken up to 5 ML before they coalesce with neigh-boring islands, forming a thick film with deep holes. These holes tend to close with additional growth, and a terraced growth of overlaying single monolayers starts to cover the film.

Adv. Mater. 2017, 29, 1703864

Figure 1. Morphological enhancement by thermal evaporation on a zone-cast templating layer. AFM images of zone-cast C8-BTBT ribbons with increasing amount of evaporated C8-BTBT monolayers (ML). a) Zone-cast sample only. Zone-cast samples with increasing evaporated thickness: b) 1/2 ML, c) 1 ML, d) 3 ML, e) 5 ML, f) 7 ML, g) 10 ML, h) 20 ML. i) AFM image of the gap area in between two ribbons after evaporation of 7 ML. j) Thickness profile along the profile line of (i). Inset of (j) shows the chemical structure of C8-BTBT.



Finally, at a nominal thickness >5 ML, 3D needles become vis-ible. The ability of C8-BTBT to propagate the steep edge of a thick (>5 ML) island directly in contact with the substrate is an important feature that explains the closing of the gaps between parallel ribbons.

The X-ray diffraction of C8-BTBT films evaporated on bare and zone-cast substrates reveals for both cases, an out-of-plane d-spacing of 2.9 nm (see Figure S4, Supporting Information). This corresponds to the step height measured by atomic force microscopy (AFM) in Figure 1j. The peaks correspond to the crystal structure as reported previously,[36] indicating that the c-axis of the crystal is vertical to the substrate, while the high-mobility ab-plane is parallel to the substrate. Both C8-BTBT films display the same layer-by-layer microstructure. Hence, evaporating C8-BTBT on the zone-cast template does not modify the vertical ordering of the thin film.

To image the in-plane orientation of the films, we performed scanning transmission X-ray microscopy (STXM) measure-ments for the samples utilizing zone-casting only, evaporation only and their combination (see Figure 2). STXM can image molecular orientation at a lateral resolution of about 30 nm.[37,38] STXM images were recorded near the carbon K edge energy (285.3 eV), in resonance with the C1s→π* mole cular transition oriented perpendicular to the plane of the aromatic moieties of C8-BTBT. The signal can be obtained from the full film thick-ness by detecting the transmitted X-ray photo ns (Figure 2a–c) or from its surface only by detecting photo-emitted electrons (Figure 2d). In these monochromic images, each shade of gray corresponds to a same molecular orientation. Contrast differences due to a change in topography were extracted by normalizing the images with measurements taken at energies off-resonance (320 eV), giving thickness-dependent signals

(see Figure S5, Supporting Information). The three ribbons imaged in the zone-cast film (Figure 2a) are monochromatic: they are single crystals with a same orientation that stem from a unique nucleation event that was split into parallel ribbons during the subsequent unidirectional solidification. Next, the STXM image of the C8-BTBT layer evaporated on a bare sub-strate demonstrates its polycrystalline character indicated by the mosaic of grains with different shades of gray (Figure 2b). The grains have sizes in the range of a few micrometers and random orientations that coincide with the morphology evolu-tion observed by AFM in Figure S3 (Supporting Information). In comparison, the image of the C8-BTBT layer evaporated on a zone-cast template displays a quite uniform gray shade (Figure 2c). Here, the initial ribbon region is no longer distin-guishable from the initial gap region: Ribbons have coalesced during evaporation to form a highly textured layer with a unique azimuthal orientation, corresponding to the monochro-matic background of the image. We nevertheless distinguish three types of defects in the scan of Figure 2c: First, at the junc-tion line between widened ribbons some gaps are still visible that are indicative of a shallower region that, however, still pre-sents the same orientation as the rest of the film. Second, the short black lines are caused by the strong X-ray absorption of the tall 3D needles visible in the morphological AFM images. Third, a few randomly located darker and brighter spots are visible, indicative of regions with different azimuthal orienta-tions. We attribute these to grains that nucleated on top of the ribbons with a crystal orientation that is different from that of the underlying ribbon. Indeed, contrarily to the polycrystalline case in Figure 2b, these spots show stronger contrast in the surface sensitive image (Figure 2d), indicating that they are thin surface defects.

At this stage, we propose a growth scenario that explains the main features of the C8-BTBT templated growth (see Figure 3). The widening of the ribbons during C8-BTBT evapo-ration (Figure 1) is similar to the lateral growth of thick islands on bare substrates (see Figure S3, Supporting Information): both happen through the expansion of a steep sidewall that is >5 ML high. STXM images in Figure 2b,c show that in both cases, this lateral growth is homo-epitaxial as the azimuthal orientation is conserved during island and ribbon widening, leading to the formation of highly textured films oriented along the seed layer. While vertical homo-epitaxial growth on highly crystalline seed layers has been described earlier for several organic semiconductors,[39–41] to the best of our knowledge, the lateral homo-epitaxial propagation of a tall crystal front observed here has not been reported before. This concept is key to our method since it repairs the defects present in the zone-cast template. The material supply leading to C8-BTBT side-wall expansion consists of: i) molecules adsorbed in the gaps, on the bare substrate, that diffuse and attach to the sidewall; ii) molecules adsorbed on the top surface of the ribbon that fall off the step edges and bind to the wall. The latter implies a low Ehrlich–Schwöebel barrier for diffusion over step edge,[42] which is required to explain the complete closure of most of the gaps. This is supported by the observation in Figure 1b that nucleation is suppressed along the edges on the top surface of the ribbons: if diffusing molecules were completely reflected by the step-edge (i.e., a high Ehrlich–Schwöebel barrier),

Adv. Mater. 2017, 29, 1703864

Figure 2. In-plane crystallinity of the different methods. STXM images for bulk molecule orientation of: a) zone-cast film, b) evaporated film, and c) evaporated film on a zone-cast template; (d) shows the surface molecule orientation of (c).



supersaturation should be reached over the whole ribbon area, including its edges.

In contrast to the bare substrate case, the evaporation of C8-BTBT on the top surface of ribbons leads to the nucleation and growth of 1 ML high islands. As islands coalesce, a 1 ML thick film forms that, on most of the surface, displays a unique crys-tallographic orientation corresponding to the azimuthal orien-tation of the ribbon underneath. This bears some similarity to previously reported homo-epitaxial behavior for this material.[43] As seen in Figure 2c,d, defects may however nucleate and propagate. They are either 2D islands with their ab-plane par-allel to the substrate surface but no epitaxial relationship with the underlying crystal, or 3D needles with a microstructure that is completely disconnected from that of the underlying film. These defects remain localized in the upper regions of the film and do not impact the crystalline order of the film at the inter-face with the substrate where charge transport occurs in TFTs.

The structural coherence of the thin C8-BTBT films formed by the double-step homo-epitaxial method remains an open question: Are the films highly textured with disordered domain boundaries where ribbons and individual islands coalesce? Or are the films truly monocrystalline with a perfect continuity of crystalline order at the coalescence points? An assessment of these questions requires advanced microstructural characteriza-tion techniques such as grazing incidence X-ray diffraction and is left to further studies. Here, we concentrate on the electrical properties of these highly crystalline lateral homo-epitaxial films.

In order to study the electrical characteristics of our films, we fabricated bottom gate–top contacts OTFTs using a highly

doped silicon wafer as common gate. Figure 4 compares typical electrical characteristics of the lateral homo-epitaxial film with reference zone-cast and evaporated C8-BTBT films. Detailed transfer and output characteristics of the same devices are given in Figure S6 (Supporting Information). The saturation transfer characteristics are well-behaved with on–off ratios >106, small onset voltages Von, and small hysteresis (see Table S1 for a summary of performance). The effective mobility in saturation µeff,sat is extracted using the conventional gradual channel approximation model. We use this method with cau-tion as nonlinearity in device characteristics have been shown to cause serious mobility overestimation.[44,45] It can be seen in Figure 4b that µeff,sat saturates with gate voltage VG for the three different films, and the plateaus in the on region define the aggregate mobility that is representative of device effective mobility.[44] The aggregate mobility of the lateral homo-epitaxial film is almost twice that of the evaporated film and three times that of the zone-cast film. Absence of charge scattering and trapping at grain boundaries is the main cause for the doubling in µeff,sat when stepping from the polycrystalline evaporated film to the highly crystalline lateral homo-epitaxial film.[46] With

Adv. Mater. 2017, 29, 1703864

Figure 3. Schematic illustration of the lateral homo-epitaxial growth. Green shows the original 8 ML high ribbons obtained by zone-casting. The growth happens both laterally inside the gap (blue) and regularly on top of the ribbons (red). Possible movements of diffusing adsorbed molecules are shown (red dots and arrows). The dark red area depicts the growth of a grain with an orientation different from that of the underlying crystal.

Figure 4. Comparison of typical OTFT electrical characteristics for C8-BTBT films fabricated by zone-casting only, evaporation only, and the combination of both: lateral homo-epitaxy (LH-Epitaxy). a) Transfer curves in saturation and b) the corresponding mobility versus gate voltage curves.



an average grain diameter of about 2 µm in the evaporated film, charge percolation pathways must cross at least 50 grain boundaries in order to bridge source and drain contacts in a TFT with a 190 µm long channel, as used in this work. On the other hand, zone-cast films present single-crystalline rib-bons that fully bridge the TFT channels without grain bounda-ries. These ribbons, however, only cover ≈2/3 of the transistor channel. Besides, some ribbons are interrupted by cracks, so that they are entirely lost to charge transport. Finally, ribbons are not necessarily oriented along the channel, thereby length-ening the current path between contacts. Therefore, for TFTs based on zone-cast films, the geometrical W overestimates the effective W, and the geometrical L underestimates the effective L. As a result, the extracted macroscopic µeff,sat heavily under-estimates the mobility within a single-crystal ribbon. Addition-ally, the randomness of above defects is responsible for a large spread in µeff,sat. Microscopic techniques could be used to esti-mate the effective W and L of each device. While this approach helps the extraction of individual ribbon electrical characteris-tics, it is highly unpractical for statistical studies and further device integration into circuits. In conclusion, the absence of grain boundaries allied with a full channel coverage explains the substantial improvement of the aggregate mobility in lateral homo-epitaxial films.

The evolution of the aggregate mobility with the evaporated thickness both on bare substrates and on zone-cast templates is shown in Figure 5. For the films evaporated on bare substrates, we observe a sharp mobility increase above 4 ML thickness. This corresponds to the coalescence of individual islands in the polycrystalline film (see Figure S3, Supporting Information).

The performance peaks at 7 ML and then slowly decreases due to the contact resistance Rc that increases with film thick-ness.[47] We see a similar trend in our lateral homo-epitaxial films: Up to 3 ML evaporation, no significant increase of µeff,sat is observed despite evidence of ribbon widening (Figure 1e). But beyond 5 ML evaporation, when ribbons start to coalesce (Figure 1f), mobility sharply increases. We conclude that it is the ribbon coalescence rather than ribbon widening that brings the large improvement in electrical performance. Further thickening improves electrical performance and µeff,sat peaks at 10 ML when gaps between ribbons have mostly disappeared (Figure 1h). Beyond this peak, µeff,sat slightly decreases due to higher Rc in thicker films. We have assessed Rc using the trans-mission line method (see Figure S7, Supporting Information). Despite its higher thickness, the lateral homo-epitaxial film pre-sents an Rc of 11.3 kΩ cm (VG = −40 V) that is lower than in reference samples thanks to a more efficient current crowding effect. This important contact resistance, typical for this mate-rial system, justifies our reporting of long channel TFTs only (L = 190 µm) that are little affected by this Rc (see Note 2 in the Supporting Information for more details).

The histograms in Figure 6 show the spread in aggregate mobility of 90 OTFTs per sample for the three types of C8-BTBT films examined so far. With a relative spread (standard deviation with respect to mean value) of 6%, the evaporated film shows the most uniform electrical characteristics. This is justified by the small dimensions of individual grains compared to device size. The spread of the zone-cast film is the worst, at 44%. This spread results from two factors: i) the spread in effective channel dimensions compared to the geometrical dimensions W and L; ii) the spread in crystal azimuthal orientation in the TFT channel combined with the anisotropy of transport proper-ties of C8-BTBT single crystals.[48] By filling the gaps between the ribbons, lateral homo-epitaxy solves the first cause of varia-tion, which reduces spread in µeff,sat by a factor of more than 4, down to 10%. The remaining 10% spread in the lateral homo-epitaxial film is attributed to the spread in orientation of the

Adv. Mater. 2017, 29, 1703864

Figure 5. Evolution of OTFT aggregate mobility with the number of evaporated monolayers. C8-BTBT is evaporated on bare substrates (red circles) or on zone-cast templates (blue stars). The latter method is lat-eral homo-epitaxy (LH-Epitaxy). One evaporated monolayer represents the evaporation of 2.9 nm C8-BTBT. Mobility is extracted in the saturation regime from devices with W/L = 2035/190 µm. Error bars represent the standard deviation over 90 devices in each condition.

Figure 6. Histogram of saturation mobilities of 90 transistors for each thin C8-BTBT film fabrication method. Average values are 3.4 ± 1.5 cm2 V−1 s−1 (spread = 44%), 5.9 ± 0.3 cm2 V−1 s−1 (spread = 6%), and 10.2 ± 1.0 cm2 V−1 s−1 (spread = 10%) for the zone-cast, evaporated, and lateral homo-epitaxial (LH-Epitaxy) samples, respectively.



parallel crystallites composing the channel. Strategies to control nucleation and crystal orientation[14] or crystals with isotropic in-plane transport properties[49] are still required to further lower parameter spread. Most importantly, the few devices that barely function in the zone-cast film are efficiently repaired by the subsequent evaporation step: the worst device of the lateral homo-epitaxial film is better than the best performing ones from both zone-cast and evaporated films.

To demonstrate the reproducibility of our method, we fabri-cated five films by lateral homo-epitaxy on different dates. Aggregate mobilities were averaged over 90 devices on each sample (see Table S2, Supporting Information). The com-pounded average delivers a µeff,sat = 10.0 ± 1.1 cm2 V−1 s−1. This spread is similar to the spread obtained on a single sample, showing that the method is easily reproducible. The best per-forming device reached a µeff,sat = 13.0 cm2 V−1 s−1. Table S3 (Supporting Information) shows a comparison of these results with C8-BTBT OTFT mobilities reported in the literature. We limited the selection to those OTFT characteristics that dis-play minimal nonlinearity, because it is known that nonlin-earity flaws the mobility extraction.[44,45] With the best sample reaching an average aggregate mobility of 10.2 ± 1.0 cm2 V−1 s−1, lateral homo-epitaxy results in the best performance published so far for C8-BTBT (see Table S3, Supporting Information). Moreover, the spread within our devices is much smaller than is typically obtained in single-crystalline devices reported in the literature.

Finally, as OTFTs based on C8-BTBT films suffer from con-tact issues, we performed gated van der Pauw (gVDP) meas-urements[50] on lateral homo-epitaxial films to independently verify the high charge carrier mobilities. The gVDP method is a simple gated four-point probe measurement that accu-rately determines the mobility of the semiconducting layer independently from contact effects. The mobility µgVDP and threshold voltage Vth,gVDP are obtained from a linear fit of the evolution of the sheet conductance σs of the semiconductor film with the average potential across the gate dielectric (see Figure S8, Supporting Information). From this analysis, we obtain µgVDP = 7.8 cm2 V−1 s−1 and Vth,gVDP = −13.8 V, that are slightly below the average values measured in regular TFTs (µeff,sat = 10.0 ± 1.1 cm2 V−1 s−1, Vth,sat = −11.4 V ± 1.7 V). Inter-estingly, the gVDP analysis reveals an anisotropy in transport properties since the sheet conductance along the zone-cast direction is higher than perpendicular to the zone-cast direc-tion. As the gVDP method averages along all directions of the device, this anisotropy explains the slightly lower µgVDP value compared to the µeff,sat of TFTs with all channels oriented along the zone-cast direction. An in-depth analysis of these gVDP results for a complete assessment of charge transport anisotropy is left to further work as this requires extended theoretical developments. This analysis also shows that the long-channel TFT reported here hardly suffer from contact resistance and are a valid probe for electrical characterization.

In conclusion, we have developed a double-step method to fabricate highly crystalline thin films of small molecular organic semiconductors. The first step relies on a directional solidification method such as zone-casting to produce large area single-crystalline films. These films are however plagued by many morphological defects whose formation is inherent

to the dynamics of growth. The second step closes the gaps between ribbons by lateral homo-epitaxy templated on the single-crystal ribbons by means of thermal evaporation of the same molecule. We employ this approach to fabricate thin films of the small molecule C8-BTBT. Growth of seven to ten homo-epitaxial monolayers is shown to suffice to close the micron-sized gaps, delivering highly textured films with low defect density and good uniformity. Thin film transistors based on these films deliver excellent electrical characteristics: The average aggregate mobility measured over 90 devices is higher than 10 cm2 V−1 s−1 with a spread of 10%. This is a significant improvement over the performance of zone-cast-only films. Besides, this is the highest aggregate mobility for the com-monly used small molecule C8-BTBT. This method should be applicable to other organic semiconductors and possibly even perovskite materials. One of the boundary conditions of this approach is that the semiconductor needs to be processable both in liquid and vapor phases.

Experimental sectionMaterials and Substrates: C8-BTBT (supplied by Nippon Kayaku

Co.) was purified before usage in a tri-zone purification oven from Creaphys. Solutions were prepared in anhydrous heptane (99%, Sigma-Aldrich) with a concentration of 0.25 wt%. Highly doped silicon substrates with a 125 nm thick thermally grown SiO2 layer served as the substrates, providing a common gate and dielectric for the OTFT fabrication. The samples were cleaned in subsequent ultrasonic baths of detergent, deionized water, acetone and isopropanol. After a 15 min UV treatment, the samples were placed in a silane oven at 140 °C for 1 h to form a self-assembled monolayer of trichloro(phenethyl)silane. STXM measurements were performed on silicon nitride membranes from Silson Ltd. with 100 nm membrane thickness. Before C8-BTBT film deposition, membranes also underwent the UV treatment and silane treatment.

Film Preparation and Characterization: For solution processing, zone-casting experiments were conducted on a homebuilt slot-die coater as previously described.[12] All experiments were carried out at room temperature. An initial drop of 60 µL solution was injected into the slot-die blade by an automatic syringe pump in order to fill the dead space inside the blade and form the initial drop in contact with the substrate. Afterward, the substrate was translated by a stepper motor at a speed of 24.7 µm s−1. To keep the meniscus shape constant during the coating process, solution was added by the syringe pump at a resupply rate of 15–20 µL min−1. For vapor processing, C8-BTBT was thermally evaporated in a high-vacuum chamber (at ≈5 × 10−8 mbar) at a deposition rate of 0.15 Å s−1 while the substrate remained at room temperature. If not otherwise defined, evaporated thicknesses are 30 nm. AFM studies were performed using a Bruker dimension edge scanning probe in tapping mode. XRD measurements were done on a PANalytical X’Pert Pro Materials Research Diffractometer using Cu Kα radiation. STXM measurements were performed at the NanoXAS beamline at the Swiss Light Source, Paul Scherrer Institut, Villigen, Switzerland.[37] Transmission measurements were detected by a silicon avalanche photo diode. Surface sensitive electron detection was performed simultaneously using a channeltron. Details of the analysis followed that described by Watts et al.[51]

Transistor Fabrication and Characterization: 60 nm thick gold source and drain contacts were deposited by vacuum evaporation through a shadow mask at a deposition rate of 1 Å s−1 and a substrate temperature of −5 °C. The channel length and width were 190 and 2035 µm, respectively, and the channel length was aligned with the casting direction so that the zone-cast ribbons tend to form bridges between the source and drain contacts. To ensure good electrical

Adv. Mater. 2017, 29, 1703864



Adv. Mater. 2017, 29, 1703864

characteristics, the samples were annealed in an N2-filled glove box at a temperature of 50 °C for 40 h. Electrical characterizations were done using an Agilent Agt1500 in dry air. Field-effect mobilities were evaluated in the saturation regime using the conventional transconductance analysis given by (2 / )(1/ ) /eff ,sat D G

2L W C I Viµ ( )= ∂ ∂ , where Ci is the gate

capacitance per unit area. For 125 nm SiO2, Ci = 27.6 nF cm−2 was used, as verified by impedance spectroscopy. For gVDP device fabrication, all methods and materials were the same as for TFT fabrication. gVDP samples had a slightly thicker dielectric thickness of 156 nm SiO2. Devices with fourfold symmetry and lateral dimension of 2 mm were obtained by the evaporation of gold contacts through a shadow mask. The organic film was then patterned in a clover-leaf shape by scratching with a probe needle. Five-probe standard gVDP electrical measurements were then performed with an HP4156 parameter analyzer, as described previously by Rolin et al.[50]

Supporting InformationSupporting Information is available from the Wiley Online Library or from the author.

AcknowledgementsThis work received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007–2013)/ERC grant agreement no. 320680 (EPOS CRYSTALLI) and from the Research Foundation Flanders (FWO Vlaanderen) under the FWO-ARRS research collaboration program/grant number G0B5914N (ORSIC-HIMA). The authors acknowledge the Paul Scherrer Institut, Villigen, Switzerland for provision of synchrotron radiation beamtime at the NanoXAS beamline of the SLS and thank Blagoj Sarafimov for technical assistance. The authors thank Nippon Kayaku Co. for supplying the C8-BTBT used in this study.

Conflict of InterestThe authors declare no conflict of interest.

Keywordshighly crystalline, homo-epitaxy, meniscus-guided coating, organic semiconductors, organic TFT

Received: July 11, 2017Revised: August 25, 2017

Published online: October 10, 2017

[1] T. Hasegawa, J. Takeya, Sci. Technol. Adv. Mater. 2009, 10, 24314.[2] H. Sirringhaus, Adv. Mater. 2014, 26, 1319.[3] M. J. Kang, I. Doi, H. Mori, E. Miyazaki, K. Takimiya, M. Ikeda,

H. Kuwabara, Adv. Mater. 2011, 23, 1222.[4] T. Okamoto, C. Mitsui, M. Yamagishi, K. Nakahara, J. Soeda,

Y. Hirose, K. Miwa, H. Sato, A. Yamano, T. Matsushita, T. Uemura, J. Takeya, Adv. Mater. 2013, 25, 6392.

[5] K. Takimiya, I. Osaka, T. Mori, M. Nakano, Acc. Chem. Res. 2014, 47, 1493.

[6] S. De Vusser, J. Genoe, P. Heremans, IEEE Trans. Electron Devices 2006, 53, 601.

[7] J. Rivnay, L. H. Jimison, J. E. Northrup, M. F. Toney, R. Noriega, S. Lu, T. J. Marks, A. Facchetti, A. Salleo, Nat. Mater. 2009, 8, 952.

[8] S. Wo, R. L. Headrick, J. E. Anthony, J. Appl. Phys. 2012, 111, 73716.

[9] G. Horowitz, M. E. Hajlaoui, Synth. Met. 2001, 122, 185.[10] Y. Diao, L. Shaw, Z. Bao, S. C. B. Mannsfeld, Energy Environ. Sci.

2014, 7, 2145.[11] L. Shaw, Z. Bao, Isr. J. Chem. 2014, 54, 496.[12] R. Janneck, F. Vercesi, P. Heremans, J. Genoe, C. Rolin, Adv. Mater.

2016, 28, 8007.[13] J. Jang, S. Nam, K. Im, J. Hur, S. N. Cha, J. Kim, H. Bin Son, H. Suh,

M. A. Loth, J. E. Anthony, J. J. Park, C. E. Park, J. M. Kim, K. Kim, Adv. Funct. Mater. 2012, 22, 1005.

[14] Y. Diao, B. C.-K. Tee, G. Giri, J. Xu, D. H. Kim, H. A. Becerril, R. M. Stoltenberg, T. H. Lee, G. Xue, S. C. B. Mannsfeld, Z. Bao, Nat. Mater. 2013, 12, 665.

[15] J. Soeda, T. Uemura, T. Okamoto, C. Mitsui, M. Yamagishi, J. Takeya, Appl. Phys. Express 2013, 6, 76503.

[16] G. Schweicher, N. Paquay, C. Amato, R. Resel, M. Koini, S. Talvy, V. Lemaur, J. Cornil, Y. Geerts, G. Gbabode, Cryst. Growth Des. 2011, 11, 3663.

[17] P. Fesenko, C. Rolin, R. Janneck, S. P. Bommanaboyena, H. Gaethje, P. Heremans, J. Genoe, Org. Electron. 2016, 37, 100.

[18] I. Cour, P. V Chinta, C. M. Schlepütz, Y. Yang, R. Clarke, R. Pindak, R. L. Headrick, J. Appl. Phys. 2013, 114, 93501.

[19] Y. Li, J. Wan, D. Smilgies, N. Bouffard, R. Sun, R. L. Headrick, Sci. Rep. 2016, 6, 32620.

[20] S. Maheshwari, L. Zhang, Y. Zhu, H. C. Chang, Phys. Rev. Lett. 2008, 100, 44503.

[21] M. Byun, R. L. Laskowski, M. He, F. Qiu, M. Jeffries-EL, Z. Lin, Soft Matter 2009, 5, 1583.

[22] J. Huang, F. Kim, A. R. Tao, S. Connor, P. Yang, Nat. Mater. 2005, 4, 896.

[23] T. P. Corrales, M. Bai, P. Homm, P. Ferrari, A. Diama, C. Wagner, H. Taub, K. Knorr, M. Deutsch, M. J. Retamal, U. G. Volkmann, P. Huber, ACS Nano 2014, 8, 9954.

[24] K. Fukuda, T. Yokota, K. Kuribara, T. Sekitani, U. Zschieschang, H. Klauk, T. Someya, Appl. Phys. Lett. 2010, 96, 21.

[25] H. Iino, T. Usui, J. Hanna, Nat. Commun. 2015, 6, 6828.[26] M. C. Cross, P. C. Hohenberg, Rev. Mod. Phys. 1993, 65, 851.[27] R. V. Craster, O. K. Matar, Rev. Mod. Phys. 2009, 81, 1131.[28] J. S. Langer, Rev. Mod. Phys. 1980, 52, 1.[29] U. Thiele, Adv. Colloid Interface Sci. 2014, 206, 399.[30] M. Wilczek, W. B. H. Tewes, S. V. Gurevich, M. H. Köpf, L. F. Chi,

U. Thiele, Math. Modell. Nat. Phenom. 2015, 10, 44.[31] F. Molina-Lopez, H. Yan, X. Gu, Y. Kim, M. F. Toney, Z. Bao, Adv.

Funct. Mater. 2017, 27, 1605503.[32] R. Ruiz, B. Nickel, N. Koch, L. C. Feldman, R. F. Haglund,

A. Kahn, F. Family, G. Scoles, Phys. Rev. Lett. 2003, 91, 136102.

[33] F.-J. Meyer zu Heringdorf, M. C. Reuter, R. M. Tromp, Nature 2001, 412, 517.

[34] S. Verlaak, S. Steudel, P. Heremans, D. Janssen, M. Deleuze, Phys. Rev. B 2003, 68, 1.

[35] B. Stadlober, U. Haas, H. Maresch, A. Haase, Phys. Rev. B 2006, 74, 165302.

[36] H. Ebata, T. Izawa, E. Miyazaki, K. Takimiya, M. Ikeda, H. Kuwabara, T. Yui, J. Am. Chem. Soc. 2007, 129, 15732.

[37] N. Pilet, J. Raabe, S. E. Stevenson, S. Romer, L. Bernard, C. R. McNeill, R. H. Fink, H. J. Hug, C. Quitmann, Nanotechnology 2012, 23, 475708.

[38] T. Schuettfort, B. Watts, L. Thomsen, M. Lee, H. Sirringhaus, C. R. McNeill, ACS Nano 2012, 6, 1849.

[39] B. Verreet, P. Heremans, A. Stesmans, B. P. Rand, Adv. Mater. 2013, 25, 5504.



Adv. Mater. 2017, 29, 1703864

[40] A. Sassella, A. Borghesi, M. Campione, S. Tavazzi, C. Goletti, G. Bussetti, P. Chiaradia, Appl. Phys. Lett. 2006, 89, 261905.

[41] X. Zeng, L. Wang, L. Duan, Y. Qiu, Cryst. Growth Des. 2008, 8, 1617.[42] S. Bommel, N. Kleppmann, C. Weber, H. Spranger, P. Schäfer,

J. Novak, S. V Roth, F. Schreiber, S. H. L. Klapp, S. Kowarik, Nat. Commun. 2014, 5, 5388.

[43] D. He, Y. Zhang, Q. Wu, R. Xu, H. Nan, J. Liu, J. Yao, Z. Wang, S. Yuan, Y. Li, Y. Shi, J. Wang, Z. Ni, L. He, F. Miao, F. Song, H. Xu, K. Watanabe, T. Taniguchi, J. Xu, X. Wang, Nat. Commun. 2014, 5, 5162.

[44] E. G. Bittle, J. I. Basham, T. N. Jackson, O. D. Jurchescu, D. J. Gundlach, Nat. Commun. 2016, 7, 10908.

[45] T. Uemura, C. Rolin, T. H. Ke, P. Fesenko, J. Genoe, P. Heremans, J. Takeya, Adv. Mater. 2016, 28, 151.

[46] X. Li, A. Kadashchuk, I. I. Fishchuk, W. T. T. Smaal, G. Gelinck, D. J. Broer, J. Genoe, P. Heremans, H. Bässler, Phys. Rev. Lett. 2012, 108, 1.

[47] P. V. Pesavento, K. P. Puntambekar, C. D. Frisbie, J. C. McKeen, P. P. Ruden, J. Appl. Phys. 2006, 99, https://doi.org/10.1063/1.2197033.

[48] H. Kobayashi, N. Kobayashi, S. Hosoi, N. Koshitani, D. Murakami, R. Shirasawa, Y. Kudo, D. Hobara, Y. Tokita, M. Itabashi, J. Chem. Phys. 2013, 139, 14707.

[49] C. Reese, M. E. Roberts, S. R. Parkin, Z. Bao, Appl. Phys. Lett. 2009, 94, 162.

[50] C. Rolin, E. Kang, J. Lee, G. Borghs, P. Heremans, J. Genoe, Nat. Commun. 2017, 8, 14975.

[51] B. Watts, T. Schuettfort, C. R. McNeill, Adv. Funct. Mater. 2011, 21, 1122.

https://doi.org/10.1063/1.2197033

https://doi.org/10.1063/1.2197033

highly crystalline c8‐btbt thin‐film transistors by …the first step is the growth of...

Documents