Spontaneously formed high-performance charge-transport layers of organic single-crystal semiconductors on precisely synthesized insulating polymers

Organic field-effect transistors (OFETs) have been extensively studied as one of the most promising building blocks in printed electronic devices such as displays,¹ sensors,² and radio-frequency identification tags,^3,4 thanks to their unique functionalities such as flexible, light-weight, large-area coverage, and low-cost productions. To fabricate high performance devices, the key engineering would be to achieve high mobility, low subthreshold swing (S), and controlled threshold voltage (V_th), together with an excellent compatibility to solution- and low-temperature processes.

Because the charge transport in a field-effect transistor takes place at the interface between the semiconductor and the gate insulator, minimizing structural and energetic disorders at such an interface directly improves the device performance. One of the most classical approaches is to use a self-assembled monolayer (SAM) as an interfacial modification layer, which results in lower S and controllable V_th.^5,6 Alternatively, potentials of polymer blend methods have been highlighted for realizing an ideal transport interface.^7–19 During film growth in the admixture solution of an organic semiconductor and an insulating polymer, spontaneous phase separation is observed.¹⁷ Although the origin of the phase separation remains unclear, some advantages of polymer blend methods have been suggested; one is that the polymer layer acts as a buffer layer to relax the surface roughness of substrates,¹⁸ and another is that the polymer layer reduces traps at the dielectric surface, which has been estimated from hysteresis of transfer characteristics.¹⁹ Recently, it was reported that the weight-average molecular weight (M_w) of additive polymers has significant impacts on the film morphology and the device performance in OFETs based on the polycrystalline organic semiconductor.²⁰ 

Many of polymer blend methods focus on the production of polycrystalline films via conventional spin-coating which provides uniform thin films easily and quickly. However, even though the film morphology and the resulting transport property can be improved by polymer blend methods, the mobility of polycrystalline organic semiconductor materials is limited by unavoidable grain boundaries and defects.²¹ To overcome such a limitation, our group developed a concept to fabricate a single crystal semiconductor thin film, which is a combination of polymer blend methods and the continuous edge-casting method that has been also demonstrated in our laboratory.²² We have confirmed that an ideal ultra-thin film composed of a single-crystalline organic semiconductor was grown spontaneously on the top of an amorphous insulating polymer layer,²³ where the hole mobility was estimated to be 17 cm² V⁻¹s⁻¹, which is comparable to that obtained without polymer blending.

In this letter, we investigate the impact of M_w on the formation of a single-crystalline organic semiconductor film. Based on detailed transport characteristics, we conclude that the spontaneously formed phase-separated structure enables the ideal transport interface free from traps even without SAM modification and rather improves subthreshold swing.

A doped Si wafer with thermally oxidized SiO₂ (100 nm) was cleaned by sonication with acetone and 2-propanol for 8 min each. Subsequently, UV/O₃ treatment was performed before the SAMs of trimethoxy(2-phenylethyl)silane (β-PTS) were deposited on the SiO₂ surface by chemical vapor deposition at 120 °C for 3 h. After the deposition of β-PTS, the substrate was cleaned again by sonication with toluene and 2-propanol for 8 min each. Bare SiO₂ substrates without SAM modification were also prepared for comparisons. The polymer blend solution was prepared by mixing one of our benchmarked organic semiconductors, 3,11-didecyldinaphtho[2,3–d:2′,3′-d′]benzo[1,2–b:4,5–b′]dithiophene (C₁₀–DNBDT–NW, Figure 1(a))²⁴ and poly(methyl methacrylate) (PMMA) insulator polymer in 3-chlorothiophene (3-CT) at concentrations of 0.02 wt. % and 0.01 wt. %, respectively. In order to obtain single crystal thin films, the continuous edge-casting method,²² schematically shown in Figure 1(b), was employed under ambient air. The resulting films were dried at 80 °C in a vacuum oven. In the present work, PMMAs with three different M_w values were employed to investigate the influence on the film morphology and device performance. Because physical properties of polymers strongly depend on M_w, the molecular weight distribution and the polydispersity index (PDI) of PMMAs used in this work were precisely controlled by living anionic polymerization^25,26 to exclude the influences of PDI. The polymerization was carried out in a dried Schlenk flask filled with Ar gas. To a solution of LiCl in anhydrous tetrahydrofuran (THF), sec-BuLi was added at −78 °C in order to eliminate the remaining water. After quenching the excess sec-BuLi by immersing the flask in hot water, fresh sec-BuLi and 1,1-diphenylethylene were added as initiators. To the solution, distilled methyl methacrylate (MMA) was added at −78 °C, which was then stirred at the same temperature. After 2 h, the polymerization was terminated with MeOH. Colorless precipitate was collected by suction filtration, washed with MeOH, and dried in a vacuum at room temperature, yielding white powder (yields 51%, 62%, and 70% for M_w = 5000, 10 000, and 170 000, respectively). The products were characterized by ¹H nuclear magnetic resonance spectra and gel permeation chromatography. Table I summarizes the M_w and PDI of the obtained PMMAs. During the thin film growth, spontaneous phase separation between C₁₀–DNBDT–NW and PMMA layers can be expected as shown in Figure 1(c). In our previous report,²³ it has been confirmed that the obtained thin film forms a single crystal which is revealed by X-ray diffraction measurement. We have also confirmed that PMMA underlays the C₁₀–DNBDT–NW layer, where the thickness of PMMA is estimated to be 4 nm, which is revealed by atomic-force microscopy (AFM).

Figure 2 shows cross-polarized optical microscopy images of the thin films formed with PMMAs of two different M_w values under different conditions. By optimizing process conditions such as substrate temperature (T_substrate), rate of solution supply (V_solution), and shearing speed of substrate (V_substrate) during continuous edge-casting, uniform single crystalline films of C₁₀–DNBDT–NW can be obtained. Generally, these parameters should be optimized because they directly affect the evaporation rate of the solution and solubility of the solute, which are closely related to the speed of crystal growth of the semiconductor molecule. For example, if V_substrate is too fast under a certain T_substrate, the speed of crystal growth becomes slower than that of substrate movement, and resulting films are not continuous any more. Judging from Figure 2, the thin film growth with PMMA of M_w = 10 000 under the condition 1 is found to be the best among three attempted, whereas not the condition 1, but 2 is the best for M_w = 170 000. This result confirms that the optimal conditions largely depend on the M_w of PMMA in the continuous edge-casting method. This can be interpreted that M_w has an effect on the physical properties of the solution such as viscosity even though the concentrations of PMMA are fixed to 0.01 wt. %. However, even under the optimal conditions, aggregations of PMMA, which can be seen as blue and glossy lumps in Figure 2, are observed more frequently in the films with PMMA of M_w = 5000 and 170 000, compared to those in the film with M_w = 10 000. This is probably due to the compensation between an affinity of the polymer with the small molecule and a self-aggregation of the insulating polymer. When M_w is small, an entropic gain tends to suppress the phase separation and to produce an admixture system, which may prevent a single crystalline-domain growth and form defects.²⁷ When M_w is large, on the other hand, the PMMAs are likely to aggregate by themselves. Hence, in order to fabricate the uniform single crystalline thin film of the organic semiconductor, it is necessary to employ the polymer of optimal M_w.

To clarify the electrical property of the interface between the single crystalline organic semiconductor and the insulating polymer, OFETs with a bottom-gate top-contact configuration were fabricated (Figure 3(a)). 40-nm thick gold source and drain electrodes were deposited via thermal evaporation using a metal shadow mask. The devices were annealed at 80 °C in a vacuum for 10 h to lower the contact resistance.

A Keithley SCS 4200 semiconductor parameter analyzer was used to measure the electrical properties of the OFETs. All measurements were conducted under ambient air. Figures 3(d)–3(i) show transistor characteristics of the typical devices fabricated without and with PMMA of M_w = 10 000. The devices without the polymer layer were fabricated similarly via continuous edge-casting from the solution of C₁₀–DNBDT–NW in 3-CT. The device performances with PMMAs of M_w = 5000 and M_w = 170 000 are as high as that with M_w = 10 000 only when the truly single-crystalline domain without any aggregations is chosen. Because the reasonably large single-crystalline domain is rarely found in films of M_w = 5000 and M_w = 170 000, the overall device yield is found to be significantly low. The saturation mobility of the devices with the polymer layer (averaged among three different devices) is measured to be approximately 10 cm² V⁻¹s⁻¹, which is consistent with the previous report.²³ It is notable that the linear mobility of those is measured to be 7 cm² V⁻¹s⁻¹, which implies the negligibly small contact resistance and textbook-like characteristics, shown in Figures 3(h) and 3(i). Here, the value of 34.5 nF cm⁻² was used as the capacitance of the gate dielectric per unit area, in which the effects of PMMA and SAM were not included. This is because the thickness of PMMA and SAM is small enough to be ignored compared to that of SiO₂. These results suggest that the spontaneous phase separation provides the interface suitable for carrier transport. For practical applications, a subthreshold swing (S), which is the inverse of the logarithmic slope of the drain current (I_D) versus gate voltage (V_G) in the subthreshold regime, is an important characteristic in addition to the mobility.²⁸ In fact, the devices with smaller S can be operated by lower V_G. Given the assumption that the trap density in the bulk is zero, the trap density at the interface per unit area (D_it) can be estimated as follows:

where C_i is the capacitance of the gate dielectric per unit area, q the elementary charge, k_B the Boltzmann constant, and T the temperature. Note that S is related to the deep trap at the interface.²⁸ Table II summarizes the averaged S and D_it of three devices: for the device only with SAM, for the one with PMMA and SAM, and for the one only with PMMA (Transistor characteristics of typical devices are shown in Figure S1 in the supplementary material). Here again, the value of 34.5 nF cm⁻² was used as C_i in all types of devices. The values of S and D_it of the devices only with PMMA are the best among three. This can draw notable conclusion that the demonstrated polymer blend method minimizes interfacial traps through the one-step process, i.e., no SAM modification requires in the present continuous edge-casting to achieve the ideal transport interface. Generally, it has been believed that the SAM modification is the way to efficiently reduce traps at the interface between the organic semiconductor layer and dielectric layer.^29,30 This is presumably because of the difference in the surface roughness of and/or the surface energy of the substrate, the latter of which can influence the morphology of the polymer layer. It should also be pointed out that SAM treatment often requires multiple processes such as hydrophilic treatment of substrates and chemical vapor deposition. Hence, the present polymer blend method, which allows a simple “one-step” fabrication process of the ideal carrier transport layer, can be advantageous because it is free from such drawbacks of SAM treatment.

In conclusion, a carrier transport interface with low trap density can be formed by spontaneous phase separation from the organic semiconductor–polymer blend solution during the thin film growth by the continuous edge-casting method. Devices showed excellent performance with a saturation mobility up to 10 cm² V⁻¹s⁻¹ and a linear mobility around 7 cm² V⁻¹s⁻¹. In particular, a low subthreshold swing of 0.25 V dec⁻¹ was demonstrated even without any surface treatment. Further, the M_w of the polymer is important to achieve a single crystalline film with large-area coverage. It is worth noting that the polymer-blended continuous edge-casting method demonstrated here opens up a simple one-step fabrication process of the ideal carrier transport layer, which is also advantageous in terms of practical processability.

See supplementary material for typical transistor characteristics of the devices summarized in Table II.

S.W. thanks JST PRESTO “Hyper-nano-space design toward Innovative Functionality.” T.H. and T.O. also thank JST PRESTO Program “Molecular Technology and Creation of New Functions.” The authors are grateful to Dr. Kan Ueji for fruitful discussion. The authors would also like to thank Mr. Hiroki Takano for assistance with the experiments.