Spiro-OMeTAD single crystals: Remarkably enhanced charge-carrier transport via mesoscale ordering.

We report the crystal structure and hole-transport mechanism in spiro-OMeTAD [2,2',7,7'-tetrakis(N,N-di-p-methoxyphenyl-amine)9,9'-spirobifluorene], the dominant hole-transporting material in perovskite and solid-state dye-sensitized solar cells. Despite spiro-OMeTAD's paramount role in such devices, its crystal structure was unknown because of highly disordered solution-processed films; the hole-transport pathways remained ill-defined and the charge carrier mobilities were low, posing a major bottleneck for advancing cell efficiencies. We devised an antisolvent crystallization strategy to grow single crystals of spiro-OMeTAD, which allowed us to experimentally elucidate its molecular packing and transport properties. Electronic structure calculations enabled us to map spiro-OMeTAD's intermolecular charge-hopping pathways. Promisingly, single-crystal mobilities were found to exceed their thin-film counterparts by three orders of magnitude. Our findings underscore mesoscale ordering as a key strategy to achieving breakthroughs in hole-transport material engineering of solar cells.

INTRODUCTION

Spiro-OMeTAD [2,2′,7,7′-tetrakis(N,N-di-p-methoxyphenyl-amine)9,9′-spirobifluorene] has long been the prominent hole-transport material (HTM) in solid-state dye-sensitized solar cells (ssDSCs), ever since it provided a solid-state replacement to liquid electrolyte–based hole transporters (). More recently, spiro-OMeTAD HTM has played a key role in the rapid rise of perovskite solar cells (PSCs), including recent major breakthroughs in solar-to-electric power conversion efficiencies (, ), which now reach more than 20% (). In addition to energy levels that match with those of the absorber (for example, perovskite or sensitizer) and the electron-transport layer (ETL; for example, mesoporous TiO2 or ZnO), spiro-OMeTAD provides remarkable glass-forming properties related to its spiro-center (). When used in PSCs or ssDSCs with a mesoporous ETL, the spiro-center frustrates crystallization while securing sufficient infiltration of the material into the mesoporous layer upon spin-coating processed film formation. However, the glass-forming tendency of spiro-OMeTAD has precluded the understanding of its intermolecular π-π stacking and hole-transport pathways and mechanisms, which has prevented the material from reaching its ultimate performance limits. Thus, despite the many strategies to engineer this HTM, this challenge remains a major bottleneck in the advancement of PSC and ssDSC efficiencies. Particularly in the case of PSCs, whose rapid efficiency rise is primarily due to improvements in the crystallinity of the perovskite (–) and ETL layers (), spiro-OMeTAD appears to be the only device layer whose single-crystal structure is yet to be reported and exploited. Consequently, tremendous efforts have been made to design and synthesize new organic hole conductors as “enhanced” hole mobility alternatives ().

Thus, we were motivated to find a way to grow single crystals of spiro-OMeTAD to address the lack of fundamental understanding of their intermolecular packing and transport properties, particularly their intrinsic upper limit of hole mobility, and to develop a clear model for charge hopping and electronic structure that could be used to devise rational strategies for engineering this crucial device component ().

RESULTS AND DISCUSSION

Using an antisolvent vapor–assisted crystallization (AVC) strategy, we have succeeded in growing sizable high-quality spiro-OMeTAD single crystals (). Figure 1A displays the setup and working mechanism of the two-vial–based AVC. Successful application of this method for crystallization depends on the judicious selection of solvents according to the intrinsic properties of the materials under study. Considering that each spiro-OMeTAD molecule contains two π-conjugated fluorene fragments, intermolecular π-π stacking may occur spontaneously upon crystallization. To exclude all possible disruptive intermolecular (π-π) interactions between spiro-OMeTAD and the solvent molecules, we used dimethyl sulfoxide (DMSO) as the solvent instead of the commonly used chlorobenzene. In this case, the inner vial contains the solution of spiro-OMeTAD in DMSO (1 mg/ml) and the outer vial contains methanol as the antisolvent. Slow diffusion of the antisolvent vapor from the outer vial into the inner vial at room temperature (RT) gradually reduces the solubility of spiro-OMeTAD and eventually triggers crystallization once a supersaturation state is reached. The as-grown spiro-OMeTAD single crystals have overall crack-free morphology and well-defined borders (Fig. 1B). The high quality and macroscopic dimensions of the as-grown crystals enable their detailed structural and electrical characterization.

Fig. 1

Crystal growth, shape, and crystallography.

(A) Schematic diagram of the crystallization process. (B) Confocal optical microscopy image of a spiro-OMeTAD single crystal. (C) Unit cell of the single-crystal structure of spiro-OMeTAD (the fluorene fragments are highlighted in yellow).

We proceeded to characterize the structure of the crystals using single-crystal x-ray diffraction (XRD). Preliminary temperature-dependent single-crystal XRD measurements at both RT and 140 K yield the same lattice parameters for the as-grown crystal (), confirming that the crystal structures at these two temperatures are identical. Therefore, the structure refined directly from the data collected at 140 K represents the actual single-crystal structure of the as-grown crystals at RT. Direct single crystallography refinement on the as-collected data yields a low R factor of 0.0445 (), which is indicative of the high quality of the as-grown crystals.

Table 1 summarizes some key parameters of the crystal structure extracted out of the completely refined data set (). The crystal belongs to the triclinic space group, consistent with the crystal’s macroscopic long-sheet morphology with bevel edges (Fig. 1B). The single-crystal unit cell contains two spiro-OMeTAD molecules featuring intermolecular parallel π-π stacking, as highlighted in yellow in Fig. 1C. The vertical plane-to-plane distance between the two fluorene rings (yellow, Fig. 1C) is estimated to be ~3.8 Å, which is slightly longer than ~3.5 Å, a distance characteristic of closely packed π-π systems [for example, in pentacene ()], but is still liable to provide significant electronic coupling between stacked conjugated fragments (). Two types of steric hindrances which affect π-π stacking are manifested in the crystal structure of spiro-OMeTAD (Fig. 1C). The steric hindrance between the two molecules inside a single-crystal unit cell prevents the formation of close π-π stacking upon crystallization. On the other hand, the steric hindrance between the outer fragments of each unit cell prevents the formation of continuous π-π stacking. The latter type of hindrance has long been acknowledged as a result of the twisted central spiro-carbon, which spurred tremendous efforts to design and synthesize modified hole-transporting molecules that exclude the twisted central spiro-carbon or replace it with untwisted linking fragments such as a thiophene or bisthiophene core (). The discontinuity in the π-π stacking hinders the delocalization of charge carriers, and therefore charge transport occurs primarily via hopping from one spiro-OMeTAD molecule to the next one, an observation corroborated by theoretical modeling (vide infra).

Table 1

Key lattice parameters of the spiro-OMeTAD single crystal.

Space group	Cell length (Å)			Cell angle (°)
	a	b	c	α	β	γ
Triclinic P−1 (no. 2)	13.66	14.72	17.28	86.23	68.98	80.01

To determine the hole mobility of single-crystal spiro-OMeTAD, we prepared a field-effect transistor (FET) using the as-grown single crystals. The device was constructed on the basis of the well-laminated, long, sheet-like crystals on a bottom-gated configuration. Belt-like gold electrodes (20 nm thick), thermally evaporated on silicon substrate, were peeled off from the substrate and then pasted on top of the prelaminated crystals. The as-fabricated FET device of spiro-OMeTAD single crystals has a channel length (L) of 23 μm and a channel width (W) of 3.15 μm (fig. S1). For comparison, FET devices based on solution-processed, highly disordered spiro-OMeTAD thin films were also prepared with the same architecture (details are given in Materials and Methods). The spiro-OMeTAD thin films were prepared using a spin-coating method normally used in solar cell fabrication (, ). The top Au contact source and drain electrodes, with a channel length (L) of 20 μm and a width (W) of 160 μm, were thermally evaporated directly on the thin films (fig. S2). Both the single crystals and the thin films used for device fabrication were freshly prepared. Transfer characterizations were performed at the moment the devices were fabricated.

Figure 2 displays the transfer characteristics (Fig. 2, A and C) and output curves (Fig. 2, B and D) of the FET devices of single-crystal and thin-film spiro-OMeTAD. Both transfer and output characteristics show excellent scaling with the gate voltage, indicating the absence of any severe contact issues. Typical hole transport is exhibited in the transfer characteristics (Fig. 2, A and C). The threshold voltages (VT) are ~60 and ~90 V for the spiro-OMeTAD single crystal–based and thin film–based FET devices, respectively. One possible reason for the positive shift of VT could be the unintentional oxygen-induced hole-doping from the bottom dielectric surface, which underwent oxygen plasma treatment before deposition of the spiro-OMeTAD layer. Because the transfer characteristics are more sensitive to the top than to the bottom surface of the semiconducting layer of spiro-OMeTAD, the positive hole-doping effect on the VT shift is less apparent in the single crystal–based device because of its larger thickness. We have to remark that without oxygen plasma treatment, we could not deposit spiro-OMeTAD with full coverage and good electric contact. We can see that both transistors work in the saturation region when −VDS > VG − VT, as a result of depletion of charges in the channel. By fitting the square root of the source-drain current (IDS)1/2 plot versus gate voltage (VG) (black line in Fig. 2, A and C) in the saturation region, the slope (msat) was estimated as noted in Fig. 2 (A and C). The hole mobility was then calculated using the following equation: , where L/W is the ratio between channel length and width, which is equal to 7.302 for the single crystal–based device and 0.125 for the thin film–based device (inset values in Fig. 2, A and C, and figs. S1 and S2), and C is the gate capacitance per unit area. The hole mobility of μ ~ 1.69 × 10−6 cm2/Vs (inset in Fig. 2C) estimated for spiro-OMeTAD thin films is consistent with previous experimental results (, ). The hole mobility of the single-crystal spiro-OMeTAD is estimated to be μ ~ 1.30 × 10−3 cm2/Vs (Fig. 2A); remarkably, this value represents an increase of nearly three orders of magnitude over that of the spiro-OMeTAD thin film. This finding underlines that spiro-OMeTAD has still much untapped potential to offer in PSCs, and the future of device engineering of the HTM lies in enhancing its crystallinity. In other words, at least in the short term, it may not be necessary to design and synthesize radically new organic hole conductors as replacements to spiro-OMeTAD in PSCs.

Fig. 2

Comparison of transistor characteristics between spiro-OMeTAD single crystals and thin films.

(A) Transfer characteristics in the saturated regime for the single-crystal FET. (B) Output curve of the single-crystal FET. (C) Transfer characteristics in the saturated regime for the thin-film FET. (D) Output curve of the thin-film FET.

We also investigated the effects of doping with lithium salts and oxidation (two commonly used strategies in device engineering of the HTM) on the mobility of spiro-OMeTAD. Our findings confirm that lithium salts only have a negligible effect on the hole mobility of the bulk of spiro-OMeTAD (fig. S3), which is consistent with the view that lithium salts are mainly present at the interface between spiro-OMeTAD and other layers in practical solar cell devices (). Furthermore, we also found a negligible effect on the hole mobility of spiro-OMeTAD by oxidation, as shown in fig. S4. These results suggest that more attention should be directed toward studying the role of oxidation and lithium salts as potential interfacial additives.

To elucidate the charge-transport processes at the molecular level, we performed density functional theory (DFT) calculations for spiro-OMeTAD molecules extracted from the XRD-refined single-crystal structure. The calculations were performed with the tuned long-range corrected ωB97X-D functional and 6-31G** basis set, using the Gaussian 09 package (). As a result of the twisted spiro structure of the individual molecules, the two highest occupied molecular orbitals (HOMO and HOMO-1), which are responsible for hole transport, are nearly degenerate in energy and their wave functions are localized on one of the two fluorene rings. The values of the transfer integrals, the most critical parameter governing charge transport (), strongly depend on the relative orientations of adjacent fluorene moieties along the π-π stacking direction (a direction) of the spiro-OMeTAD crystal. The transfer integral between the two molecules within a unit cell is calculated to be different from that between adjacent unit cells. As a result, to map out the charge-transport channel in the spiro-OMeTAD crystal, a model consisting of at least three molecules (trimer) has to be taken into account, as shown in Fig. 3A. As in the single-molecule case, the HOMO and HOMO-1 levels of the molecular trimer are energetically nearly degenerate and their wave functions are distributed over two different pairs of the fluorene rings. This is related to the two large transfer integrals obtained for the corresponding fluorene pairs as a result of the coupling of the HOMO and HOMO-1 of the individual molecules (Fig. 3A). The magnitude of the transfer integrals is substantial, on the order of nearly 40 meV, even though it is lower by a factor of 2.0 to 2.5 with respect to the transfer integrals calculated for high-mobility rubrene (~100 meV) and pentacene (~85 meV) single crystals (). Such transfer integrals open up an efficient charge-transport channel within the spiro-OMeTAD single crystal, which explains the remarkably improved hole mobility as compared with that of the less ordered thin-film counterpart. The calculations also show that the hole geometry relaxation energy (polaron binding energy) is on the order of 145 meV, a value larger than that of the transfer integrals; this finding indicates that charge carriers tend to localize on one spiro-OMeTAD molecule (or a few molecules) to form small polarons (). For the sake of completeness, we have also evaluated the electronic band structure of the spiro-OMeTAD crystal at the DFT B3LYP-6-31G** level with the CRYSTAL14 package () (see Fig. 3B); the upper valence subband (which comes from the interactions among the HOMO and HOMO-1 levels of the two inequivalent molecules per unit cell) is calculated to be narrow. Therefore, all theoretical results point to the fact that hole transport in the spiro-OMeTAD crystal is expected to take place in the hopping regime, which is consistent with the measured hole mobility on the order of 10−3 cm2/Vs.

Fig. 3

Modeling of intermolecular charge transport.

(A) DFT-ωB97X-D/6-31G** energies and wave functions for the HOMO (left) and HOMO-1 (right) levels of a molecular trimer extracted from the spiro-OMeTAD single crystal along the a direction. Lower panels: Illustration of the transfer integrals estimated for adjacent molecules in the trimer; the blue squares highlight the fluorene rings for which wave-function overlap has a major contribution to intermolecular electronic couplings (transfer integrals). (B) DFT-B3LYP/6-31G** valence band structure and density of electronic states (DOS) of the spiro-OMeTAD single crystal; the crystallographic coordinates of high-symmetry points in the first Brillouin zone correspond to Γ = (0,0,0), X = (0.5,0,0), Y = (0,0.5,0), Z = (0,0,0.5), M = (0.5,0.5,0), and R = (0.5,0.5,0.5).

In summary, for the first time, single crystals of spiro-OMeTAD have been grown, and their crystal structure has been determined and shown to be contrary to the conventional wisdom that was built upon the material’s glass-forming properties (). Currently, most research efforts in the field are directed toward designing and synthesizing new HTMs, a much more time-consuming process of uncertain outcome. Our work not only elucidates a straightforward and assured strategy for creating a vastly improved hole transporter through improving the crystallinity of spiro-OMeTAD, one of the most widely used commercially available HTMs for photovoltaic and optoelectronic devices, but also highlights mesoscale molecular ordering as the key to promoting the material’s charge transport pathways. Although our method, which was developed to make structure-analysis quality single crystals, is not directly implementable in its current form in large-scale device fabrication, we foresee that the use of an antisolvent to trigger crystallization, as we did in this study, could potentially be adopted to enhance the crystallinity of the spiro-OMeTAD layer by simply immersing the as-deposited “wet” layer from solution into an antisolvent. Indeed, a similar strategy has been successfully used in preparing a hybrid perovskite layer with enhanced crystallinity and performance for PSCs, but it remains to be tested in spiro-OMeTAD (). In particular, PSCs with planar heterojunction TiO2/perovskite/HTM architecture, which are competing strongly with mesoporous-type solar cells in terms of efficiency but have the advantage of not requiring the infiltration of the HTM inside the perovskite layer, stand to benefit most from enhancing the hole mobility in the HTM layer by using a single-crystalline spiro-OMeTAD layer. More broadly, the issue of hysteresis in PSCs may be alleviated by the use of single-crystalline spiro-OMeTAD HTM because recent studies have found a strong correlation between reduced current density–voltage (J-V) hysteresis and improved electron and hole transport in PSCs ().

MATERIALS AND METHODS

Materials and solvents

All chemicals were used as received, unless otherwise stated. Spiro-OMeTAD (sublimated grade) was purchased from Borun New Material Technology Ltd. Bis(trifluoromethane)sulfonimide lithium salt (product number: 449504-50G), DMSO (product number: 276855-250 ml), methanol (product number: 34860-1L-R), and chlorobenzene (product number: 270644-1L) were purchased from Sigma-Aldrich.

Crystallization of spiro-OMeTAD

Spiro-OMeTAD (5 mg) was dissolved in DMSO (5 ml) under stirring in a closed vial (7 ml) at RT, yielding completely transparent solution with light yellowish color. The solution was further cleaned using a syringe filter through a Millipore membrane (100 μm). Then, the filtrated solution was equally divided into five vials (5 ml), and each vial was transferred inside a big vial (20 ml) containing methanol (8 ml). The outer vial was closed tightly, and the system was then transferred into a nitrogen-filled glove box. Slow evaporation of the antisolvent methanol in the solution of spiro-OMeTAD in DMSO (1 mg/ml) triggered the crystallization.

Single-crystal XRD

This characterization was performed at 140 K on a Bruker D8 Venture diffractometer equipped with a PHOTON 100 CMOS detector and Cu Kα monochromated radiation (λ = 1.54178 Å). The unit cell was determined using 9123 reflections. Preliminary temperature-dependent single-crystal XRD measurements at RT and 140 K yielded the same lattice parameters for the as-grown crystal [triclinic, space group P−1 (no. 2), a = 13.66 Å, b = 14.72 Å, c = 17.28 Å, α = 86.23°, β = 68.98°, γ = 80.01°], confirming no crystal structural changes in this temperature range. The crystal structure was refined using SHELXTL software. Detailed refinement parameters are given in the affiliated single crystal .cif file.

Silicon substrate cleaning

The silicon substrates used for FET devices were treated in a stepwise manner using the following procedures: (i) immersion in sulfuric acid (98%) at 100°C for 2 hours, (ii) thorough cleaning with Milli-Q water, and (iii) treatment with oxygen plasma for 30 min.

Single crystal–based FET device fabrication

The as-grown long sheet–like spiro-OMeTAD single crystals were washed thoroughly with methanol and then transferred directly onto the pretreated silicon substrate. The clustered crystals were separated and laminated on top of the substrate using tweezers. The prelaminated crystal samples were dried under vacuum at 60°C for 4 hours. Finally, predeposited belt-like gold electrodes on the silicon substrate were peeled off and transferred on top of the crystal samples using microprobes. Figure S1 displays a real photograph of the single-crystal FET device taken under an optical microscope. The channel width and channel length were estimated to be 3.15 and 23 μm, respectively.

Thin film–based FET device fabrication

Thin films were prepared using the spin-coating method. For the nondoped spiro-OMeTAD thin-film FET device, a drop of spiro-OMeTAD in chlorobenzene (0.2 mg/ml) was applied onto the as-prepared silicon substrate through a syringe filter equipped with a Millipore membrane (100 μm). After spinning (6000 rpm, 30 s) the excessive solution, a uniform thin film of spiro-OMeTAD on the silicon substrate was obtained and further dried under vacuum at 60°C for 4 hours. Gold electrodes (thickness, 20 nm; width of the gap between the two electrodes, 20 μm; length of each electrode, 160 μm) were thermally evaporated onto the thin film through a patterned mask. Figure S2 displays a real photograph of the thin-film FET device taken under an optical microscope. For the lithium salt–doped spiro-OMeTAD thin-film FET device, a solution of spiro-OMeTAD (0.2 mg/ml) and bis(trifluoromethane)sulfonimide lithium salt (saturated) in chlorobenzene was prepared. Then, the devices were prepared using exactly the same method as detailed for nondoped thin-film devices.