Three-Dimensional Charge Transport in Organic Semiconductor Single Crystals

Organic semiconductors are the key component in organic electronics and determine the device performance. There is great interest in the fundamental understanding of charge-transport mechanisms and the structure–property relationships in organic semiconductors.1 To explore the intrinsic charge-carrier transport properties of organic semiconductors, the most promising method is to fabricate a single-crystal field-effect transistor (FET) because the organic single crystal has perfect molecular order, is free of grain boundaries, and has a minimal concentration of charge traps.2 Up to now, there has been significant progress in the growth of organic single crystals, the fabrication of single-crystal FETs, and the investigation of the structure–property relationship of organic semiconductors.3, 4 However, to fully understand the charge transport mechanisms, it is necessary to investigate the anisotropy of charge transport in organic semiconductor single crystals, because the anisotropic molecular packing in the crystals can lead to a strong anisotropy in the field-effect mobility. To date, the anisotropic charge transport properties of a few organic single crystals have been measured. For example, Sundar et al., Reese and Bao, and Zeis et al. succeeded in measuring the mobility anisotropy in single-crystal rubrene (5,6,11,12-tetraphenyltetracene).5 Mannsfeld et al. measured the anisotropic field-effect mobility of dicyclohexyl-α-quaterthiophene (CH4T) single crystals.6 Li et al. fabricated transistors based on individual microcrystals with multichannels along different crystal axes and crystal planes by using a “two-dimensional organic ribbon mask” technique to probe the charge-transport anisotropy.7 The anisotropic field effect mobilities of pentacene and tetracene were also studied.8, 9 However, all of the above results cover only one- and two-dimensional charge transport anisotropy, which is not enough to provide sufficient insight into the correlation between molecular packing and charge transport in organic semiconductor materials because many crystals have a three-dimensional (3D) nature. Therefore, it is necessary to investigate the 3D charge transport anisotropy in organic semiconductors. In this Communication, we report on the 3D charge transport anisotropy in organic semiconductor single crystals. By controlling the growth conditions, single crystals can be grown either parallel or perpendicular to the substrate.10 These two types of morphology provide a good platform for the fabrication of differently oriented transistor devices and for the investigation of 3D charge transport in organic semiconductor single crystals (Figures 1a,d).

The candidate compound for the study of 3D anisotropic charge transport in this research is 4,4′-bis((E)-2-(naphthalen-2-yl)vinyl)-1,1′-biphenyl (BNVBP), which was synthesized by the single-step Wittig –Horner reaction (see Scheme S1, Supporting Information) and was purified by sublimation. BNVBP single crystals were obtained through the physical vapor transport (PVT) method and single-crystal X-ray analysis revealed that the BNVBP molecules are arranged in a herringbone pattern, which is very similar to that of pentacene (Figure S1c, Supporting Information).

The basis for the fabrication of organic single-crystal transistor devices is the growth of the single crystals, followed by fabrication of electrodes on the crystals. As is well known, thin small crystals are more suitable for the fabrication of high performance devices and better reflect the essential nature of the material.6, 7b One of the biggest challenges for the investigation of 3D mobility anisotropy is to control the growth of the crystal for suitable dimensions and orientation, which may be why the 3D mobility anisotropy of single crystals has not yet been reported in FET configuration. Two types of morphological crystals were grown under controlled conditions using a PVT technique (Figure S2, Supporting Information).11, 12 The process of primary nucleation is competition between the thermodynamic driving force (a volume effect and the enthalpic lowering of the free energy by favorable intermolecular interactions) and the energy penalty associated with surface effects (between the organic layer and substrate), which plays a crucial role in the morphology of the crystals.4b, 11, 13 During the process of slow heating from 200 to 250 °C, nucleation occurs preferably with molecules standing up on the substrate surface (Figure 1b). Of course there are also a small number of nuclei with molecules arranged edge-on to the substrate, but they are gradually merged as the plate grows at 250 °C for 1 h. So a sample with nearly 100% plate morphology is obtained in the crystal growth zone with Ar as carrier gas (Figure S3a, Supporting Information). The micrometer-sized single-crystal plates have a regular hexagonal shape (or elongated hexagonal shape) with sizes as large as 0.1 mm × 0.1 mm and a typical thickness of 80 nm –1 μm. When heating is rapid, or the material is put directly into the tube furnace at 270–280 °C for 1 h, a nearly 5% fraction of micrometer-sized rods appears at the front of the crystal growth zone (Figure S2a, Supporting Information). This quick growth method utilizes a larger temperature gradient, which disrupts optimal orientation choice as the crystal grows. With Ar and H₂ in a ratio of 1:1 as carrier gas, rods account for 20% of the total at the slower flow rate (Figure S3b, Supporting Information). The micrometer-sized rods are thicker than the plates at a value above 300 nm, with a maximum length of 30 μm (Figure 1d). Rapid cooling under vacuum conditions is more conducive to the growth of rod-like crystals. Otherwise rods will grow along the other two directions and evolve into a standing hexagonal crystal (Figure S2e, Supporting Information).

Figures 1c,f show selected-area electron diffraction (SAED) patterns; the diffraction pattern did not change as the electron beam was moved. In light of the crystal data, it can be concluded that the single-crystalline BNVBP plate grew along the [100] and [010] directions, and that the [100] direction supports the faster growth (the hexagonal crystal is elongated along the [100] direction). Also, the rod grew faster along the [100] direction, which indirectly proved that the rod could evolve into the standing hexagonal crystal under slow cooling. X-ray diffraction (XRD) experiments were also carried out on the two different crystal morphologies to identify the growth orientation (Figure 1g). The peak at 6.64° (2θ) is attributed to the 004 reflection for the BNVBP structure, and corresponds to a lattice plane distance of ca. 13.33 Å, which is almost a quarter of the crystallographic c-axis lattice parameter (Table S1, Supporting Information). This indicates that the BNVBP molecules stood up nearly perpendicular to the substrate surface at an angle of 70.4° (Figure S1d, Supporting Information). One additional peak at 23.6° (2θ) appears from the mixture of rods and plates, which is attributed to the (020) plane. It corresponds to a lattice plane distance of 3.76 Å, which is just half of the b-axis lattice parameter. It is therefore clear that the molecules of rod-like crystals are arranged edge-on to the substrate plane along the short axis (Figure 1e). Two different film morphologies are also obtained, which coincide with the plate and rod crystals (Figure S4, Supporting Information).

In order to study the performance of the OFETs, micrometer-sized plate and rod crystals were measured by the “two-dimensional organic-ribbon mask” technique using the same device architectures.7 A single wire of 2,8-di(4-trifluoromethylpehnyl)-5,11-di-n-octylindolo [3,2-b]carbazole was used as the mask because it is long and thin enough and has excellent flexiblility.14 Gold source and drain contacts about 70 nm thick were deposited. All the devices exhibited p-type transistor behavior. As shown in Figure 2, the highest mobility measured was 2.49 cm² V⁻¹ s⁻¹. The device channel is perpendicular to the a-axis. This is consistent with what was previously reported in the case of rubrene and pentacene, in that the direction of the maximum mobility coincides with the fastest crystal-growth direction.5–9 Figure 3a shows an elongated hexagonal plate oriented along the a- axis, with a thickness of ~93 nm. Mobility along the a-axis (μ_a) was measured to be 2.37 cm² V⁻¹ s⁻¹, see Table 1. The transistors exhibited much lower mobilities along the [110] and [100] directions (μ₁₁₀ and μ₁₀₀) of about 0.65 and 0.75 cm² V⁻¹ s⁻¹, respectively. These values are less than a third of μ_a (Figure 3b). When charge is transported along the b-axis ([010] direction), the carrier mobility (μ_b) is higher (1.17 cm² V⁻¹ s⁻¹) than μ₁₁₀, which is half μ_a (2.11 cm² V⁻¹ s⁻¹) (Figures 3c,d). Moreover, all transistors exhibited very high threshold voltages (V_T), exceeding –25 V. It is believed that the higher mobility results from the closer stacking of adjacent molecules.7, 15 From the crystal packing structure, the molecular packing along the [100] direction is obviously closer than that along the [010] and [110] directions, which coincides well with the fact that most crystals were elongated along the [100] direction (Figure 3e). The molecular packing along the [010] direction is closer than along the [110] direction (Figures S5a,b, Supporting Information). These findings relate well with our experimental results, where the mobility was the highest along the [100] direction and the lowest along the [110] direction. The molecules in the perpendicular direction (the c-axis) do not interact and are more likely to have electrically insulating properties, which is a disadvantage for charge injection.16 Hence, all the devices have high V_T.

The same method was used to fabricate and measure the micrometer-sized rod single-crystal devices (Figure 4). The device (mobility of 0.54 cm² V⁻¹ s⁻¹ and V_T of only –8 V; Table 1) was measured along the [100] direction within the (010) plane, with a rod thickness of 400 nm (Figures 4a,b). In addition, a rod device oriented along the [001] direction was also obtained (Figure 4c,d). The carrier mobility is 0.016 cm² V⁻¹ s⁻¹, but V_T is only –13 V. Comparison of dozens of experiments indicates that the mobility along the [100] direction is at least ten times that along the [001] direction. From the crystal packing structure, the distance between adjacent molecules along the [001] direction is much larger than that along other directions (Figure 3e and Figure S5, Supporting Information). Thus, the mobility of the rods along the [001] direction is very small. The low threshold voltage is determined by the strength of the C-H–π interactions in the charge-injection layer. In theory, owing to molecules forming a two-dimensional network by intermolecular C-H–π interactions within the ab plane, the charge traps are adequately passivated no matter whether in charge-injection layers or in transport layers, a situation which leads to higher carrier mobility along the [100] direction within the (010) plane. However, the carrier mobility is small, mainly due to the crystal thickness. The thickness of the injection layer has a crucial effect on the mobility and V_T.

In order to investigate the relationship between the performance and packing structure, micrometer-sized plate and rod devices of the same thickness were studied. When the thickness is greater than 400 nm, the mobility of the plate is reduced to 0.13 cm² V⁻¹ s⁻¹ and V_T is –50 V along the a-axis (Figure S6, Supporting Information). Hence, the performance of the rod devices along the [100] direction is significantly better than that of thick crystal plate devices of the same size. As is well known, organic transistors are usually operated in the accumulation mode and charge transport typically takes place in the first few monolayers (<26 nm for BNVBP) of the organic semiconductor adjacent to the gate dielectric.17, 18 Therefore, the 400 nm thickness appears as an additional resistance in the stack and not as a storage element, because no interlayer interaction is excited. The transistor performance of the plate can more easily be influenced owing to the molecular packing of the charge-injection layer. Of course, different packing modes lead to different positions of the Fermi level between the organic layer and Au electrodes. By electrochemical analysis, BNVBP has a lower HOMO level (–5.3 V) than Au (–5.1 V). Therefore, charge injection is greatly influenced by the interaction between the gold electrode and the organic layer. From the atomic force microscopy (AFM) image shown in Figure S5, more island nuclei can be detected on the gold interface than on the Si/SiO₂ substrate, which suggests that the rod devices have a low energy level alignment at the metal/organic semiconductor interface. To a certain degree, this situation leads to lower contact resistance and more efficient charge injection when the molecules are arranged edge-on to the substrate plane along the short axis.16 Therefore, higher carrier mobility and lower V_T were measured in micrometer-sized rod devices of the same thickness along the [100] direction (Figure S7, Supporting Information). All of the above devices exhibit good stability (Figure S8, Supporting Information).

In conclusion, crystals with two kinds of morphology and orientation (two-dimensional highly regular hexagonal micrometer-sized single-crystal plates and one-dimensional rods) were grown under controlled conditions using a PVT technique. Well-performing organic transistors were fabricated and the 3D anisotropy of the carrier mobility and V_T of the crystals were measured for the first time. The results demonstrate that the transistors exhibit mobility as high as 2.49 cm² V⁻¹ s⁻¹ along the direction of fast growth (a-axis) in the single-crystal plate devices; the micrometer-sized rod devices (channel perpendicular to [100]) displayed better performance than micrometer plates of the same thickness, because of the advantage of low contact resistance at the metal/organic interface, and stronger intermolecular interactions in the charge-injection and transport layers. The results indicate specific steps can be taken in molecular design and device preparation to achieve good performance.

Experimental Section

Synthesis of 4,4′-bis((E)-2-(naphthalen-2-yl)vinyl)-1,1′-biphenyl (BNVBP): At 0 °C, potassium tert-butoxide (t-BuOK; 0.85 g, 7.6 mmol) in tetrahydrofuran (THF; 20 mL) was added dropwise to a mixture of 4,4′-bis(diethylphosphonomethyl)biphenyl (0.35 g, 0.77 mmol) and 2-naphthaldehyde (0.312 g, 2 mmol) in THF (20 mL) under nitrogen. The reaction mixture was stirred overnight at room temperature. After filtration, the residue was purified by sublimation twice to give a green solid (0.297 g, yield 84.2%). Anal. Calcd for C₃₆H₂₆S₂: C 94.29, H 5.71. Found: C 94.27, H 5.73.

Characteristics of microstructures and devices: The AFM images were obtained using a Digital Veeco Instruments atomic force microscope operating in the tapping mode. The XRD investigations were carried out with a Bruker D8 advanced diffractometer using Cu Kα radiation (λ = 1.5406 Å). The data were collected using a Ni-filtered Cu-target tube at room temperature in the 2θ range from 5° to 30°. SEM images were obtained with a Hitachi S-4300SE (Japan). TEM and SAED measurements were carried out on a JEOL 2010 (Japan). The FET measurements were carried out with a Süss MicroTec probe station at room temperature in air, and recorded using a Keithley 4200 SCS semiconductor characterization system.

Crystallographic data: CCDC 803867 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

Supporting Information

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

Acknowledgements

We thank the National Natural Science Foundation of China (Grant No. 51021062 50990061 50802054 and 51002086), 973 program of PR China (Grant No. 2010CB630702) for financial support. We are grateful to Prof. Robert I. Boughton, Wenping Hu, and Dr. Lang Jiang for helpful discussions.