An Oxa[5]helicene-Based Racemic Semiconducting Glassy Film for Photothermally Stable Perovskite Solar Cells.

Attaining the durability of high-efficiency perovskite solar cells (PSCs) operated under concomitant light and thermal stresses is still a serious concern before large-scale application. It is crucial to maintain the phase stability of the organic hole-transporting layer for thermostable PSCs across a range of temperatures sampled during device operation. To address this issue, we propose a racemic semiconducting glassy film with remarkable morphological stability, exemplified here by a low-molecular symmetry oxa[5]helicene-centered organic semiconductor (O5H-OMeDPA). The helical configuration of O5H-OMeDPA confers the trait of multiple-dimension charge transfer to the solid, resulting in high hole mobility of 6.7 × 10-4 cm2 V-1 s-1 of a solution-processed glassy film. O5H-OMeDPA is combined with a triple-cation dual-halide lead perovskite to fabricate PSCs with power conversion efficiencies of 21.03%, outperforming the control cells with spiro-OMeTAD (20.44%). Moreover, the cells using O5H-OMeDPA exhibit good long-term stability during full-sunlight soaking at 60°C.

Introduction

Metal halide perovskite solar cells (PSCs) have stirred up intense research passions thanks to the steeply increasing power conversion efficiency (PCE) and cost-effective fabrication (Kojima et al., 2009, Kim et al., 2012, Burschka et al., 2013, Liu et al., 2013, Jeon et al., 2015, Hou et al., 2017, Luo et al., 2018). In a classical PSC, a hole-transporting layer (HTL) plays pivotal roles in both hole transporting and electron blocking. Moreover, a uniform and dense HTL can avert the penetration of moisture and oxygen from outside, which affects the long-term stability of the device (Wang et al., 2018a). Hitherto, an enormous amount of organic hole transporters has been developed (Calió et al., 2016, Wang et al., 2016), only a few of which, however, present superior photovoltaic efficiencies over the state-of-the-art spiro-OMeTAD, with PCEs exceeding 20% (Zhang et al., 2017a, Zhang et al., 2017b, Ge et al., 2018, Xu et al., 2017, Jeon et al., 2018, Saliba et al., 2016).

The spiro-OMeTAD-based PSCs undergo a severe PCE damping under a certain thermal stress (Domanski et al., 2018), which is essentially associated with the phase transition from the amorphous state to the crystalline state during device operation (Malinauskas et al., 2015, Zhao et al., 2017). The dynamic phase transition of the solution-processed spiro-OMeTAD film is imputed to the high molecular symmetry (D2d point group (Wang et al., 2018b)). Meantime, large-scale application of spiro-OMeTAD will encounter the issues of complicated preparation and purification cost. To date few stable PSCs using small-molecule hole transporters have been reported under the compounded stress of light and heat. In 2018, a benzodipyrrole-based hole conductor was used for PSCs with 17.2% PCE and good stability at 35°C (Shang et al., 2018). Later, Lee and his co-workers engineered a fluorene-terminated hole conductor for PSCs, which could maintain high stability at 60°C when stored in the dark (Jeon et al., 2018). Recently, Lin et al. designed three azahelicene derivatives as hole conductors for PSCs, among which the SY1-based device shows a PCE of 17.34% and an excellent ambient stability (Lin et al., 2018). Also, two [7]helicene derivatives with stable open-shell singlet biradical ground states were demonstrated as effective surface modifiers of the inorganic nickel oxide HTL in PSCs (Lee et al., 2019).

In this study, we explore the possibility of using a racemic glassy organic film composed of enantiomers, for the phase stability control of the HTL in PSCs. As a proof of concept, we design a low-molecule-symmetry helicene-based organic semiconductor, N5,N5,N9,N9-tetrakis(4-methoxyphenyl)dinaphtho[2,1-b:1′,2′-d]furan-5,9- diamine (abbreviated as O5H-OMeDPA, Figure 1A). O5H-OMeDPA consists of an oxa-[5]helicene core and two electron-donating bis(4-methoxyphenyl)amine segments (Figure 1A). As reported in the literature (Nakahara et al., 2013), oxa-[5]helicene itself distorts with a 25° torsion angle of the two outer benzene rings in single crystal. The slight torsion configuration reduces the molecular symmetry (C2 point group), which we think is beneficial for the formation of a glassy film. Also, the rigidity and twisting induced by the helical scaffold are considered to be favorable for the phase stability (Jhulki et al., 2016). Compared with the triangular pyramid configuration of spirobifluorene in spiro-OMeTAD, the relative planarity of helicene (Lin et al., 2018) may enhance the hole mobility. Our preliminary studies have shown that PSCs with O5H-OMeDPA as HTL present a high PCE of 21.03% and good stability under continuous full-sunlight soaking at 60°C.

Figure 1

Molecular Structure, Molecular Packing, and Energy Band

(A) Chemical structure of O5H-OMeDPA.

(B) Optical microscopic image of an O5H-OMeDPA single crystal.

(C and D) Crystal structures along the a axis view (C) and the b axis view (D), with color-coded molecules: blue, enantiomer P; red, enantiomer M.

(E) DFT-calculated vcalence band structure of the O5H-OMeDPA single crystal. General gradient approximation (GGA) was used with the Perdew–Burke–Ernzerhof (PBE) functional to describe the exchange-correlation potential. In the band graphic, high-symmetry points in the first Brillouin zone are labeled in the crystallographic coordinates as follows: (Γ) (0,0,0), (B) (0,0,0.5), (A0) (0.5,0.0,0.5), (Z) (0,0.5,0), (D) (0,0.5,0.5), (Y) (0.5,0,0), and (C) (0.5,0.5,0).

(F) HOMO of an a-laying dimer.

(G) HOMO of a bc-laying dimer.

Results

Synthesis, Molecular Packing, and Hole Transportation

The synthesis sequence leading to O5H-OMeDPA started with the acid-catalyzed cyclization of 1,1′-binaphthalene-2,2′-diol, which is followed by bromination and Buchwald-Hartwig cross-coupling (Scheme S1). The synthesis resulted in a total yield of 80%. The overall material cost (Table S1) for 1 g O5H-OMeDPA in our laboratory is estimated to be less than one-tenth of that for spiro-OMeTAD (Ge et al., 2018), which makes large-scale synthesis and application viable. The molecular geometry optimization by density functional theory (DFT) method reveals that the oxa[5]helicene segment distorts with a torsion angle of 27.92° and the highest occupied molecular orbital (HOMO) of O5H-OMeDPA is delocalized over the whole π-conjugated skeleton, which is decisive for intermolecular hole hopping (Figure S1).

High-performance organic HTLs are always amorphous to ensure a complete coverage of the underlying perovskite in PSCs. The thin films are usually grown rapidly via spin-coating to prevent the formation of crystallites. However, the alignment of adjacent molecules in these films likely resembles that in the single crystal to some extent (Rivnay et al., 2012, Baldo et al., 2001). To elucidate the molecular packing and charge transport characteristics of O5H-OMeDPA in the solid state, we grew its single crystals by slowly evaporating a saturated, racemic solution of dichloromethane/heptane. The as-grown O5H-OMeDPA single crystals present an overall crack-free morphology and distinct borders (Figure 1B). X-ray crystallographic structure analysis was carried out to define the molecular packing (Figure S2 and Table S2). As displayed in Figures 1C and 1D, a pair of enantiomers M and P are observed in the crystal. As Figure S3 presents, the oxa[5]helicene segment distorts with a torsion angle of 30.89°, which is slightly larger than that of DFT-optimized geometry (Figure S1). For the O5H-OMeDPA geometry from single crystal, HOMO is also wholly delocalized over the electronic skeleton (Figure S3B). The enantiomers form an antiparallel face-to-face dimer. The dimers stack on top of each other to form slipped columnar motifs along the a axis, via the C-H⋅⋅⋅π interaction of the terminal methoxyphenyl with the oxa[5]helicene core. Within the crystallographic bc-plane the dimers arrange in T-shape herringbone pattern, and their cohesion is achieved by London dispersion forces.

The electronic band structure of the O5H-OMeDPA crystal was calculated by DFT to gain insights on hole conductivity. From the results shown in Figures 1E and S4, one can notice that the band dispersion is quite small, as is typical of organic HTMs. Differences between adjacent energy levels show variations within 0.02 eV, depending on the directions of reciprocal space (i.e., ΓZ, ΓB, and ΓY). We also computed hole effective masses and electron transfer integrals (Table S3). Along the ΓY direction of the Brillouin zone (the crystallographic a axis), we recorded an effective hole mass of 8.35 m. The values are smaller along the ΓZ and ΓB directions of the Brillouin zone, being 5.17 and 6.67 m, respectively, suggesting the higher hole mobilities along the b and c axes of the lattice space (see Figures 1C and 1D).

Now we turn to the dimers A-B and B-C along the crystallographic a axis in Figure 1D. The calculated transfer integrals (Table S3) are 41.12 meV for dimer A-B (path 5 in Figure S5) and 1.32 meV for dimer B-C (path 6 in Figure S5). As Figure 1F shows, the HOMO of dimer A-B is fully extended on both A and B due to the strong electron coupling. However, the hole mobility along the a axis is not high, actually limited by the weak coupling region (B-C), which compensates the benefit of strong π-π interaction in dimer A-B. The transfer integral of path 2 and 4 (Figure S5) is 6.27 meV, contributing to a valid hole conduction in the bc plane, although the intermolecular electron coupling is not very strong (Figure 1G). The single crystal hole mobility of O5H-OMeDPA was theoretically calculated by using the Einstein relation μ = eD/kT based on the hopping model (Nguyen et al., 2015), and the calculated hole mobility reaches up to 4.7×10−2 cm2 V−1 s−1 (Figure S5 and Table S3).

The hole-transporting property of the O5H-OMeDPA films spun from chlorobenzene was investigated by measuring the space-charge limited currents of hole-only devices. The average hole mobility of a non-doped O5H-OMeDPA film is 3.3×10−5 cm2 V−1 s−1 (Figure 2A), over five times higher than that of 5.0×10−6 cm2 V−1 s−1 for the spiro-OMeTAD control, suggesting the merit of multiple-dimensional charge transfer pathways in a semiconducting glassy film. With lithium bis(trifluoromethylsulfonyl)imide (LiTFSI) and 4-tert-butylpyridine (TBP) as additives, the hole mobility of the spin-coated O5H-OMeDPA film was remarkably improved to 6.7×10−4 cm2 V−1 s−1 (Figure 2A), which is about three times higher than that of 2.1×10−4 cm2 V−1 s−1 for the spiro-OMeTAD counterpart.

Figure 2

Conductivity, Photovoltaic Characterization, and Photothermal Stability

(A) Space-charge limited currents hole mobilities of O5H-OMeDPA and spiro-OMeTAD spin coated on the substrate of poly(3,4-ethylenedioxythiophene) polystyrene sulfonate.

(B) J-V characteristic (reverse scan) of a champion PSC with O5H-OMeDPA as the hole-transporting layer measured under 100 mW cm−2, AM 1.5G illumination. The data for a control cell with spiro-OMeTAD are also included. The inset is the IPCE spectrum and integrated JSC from the IPCE curve for the O5H-OMeDPA based cell.

(C) Normalized PCEs of unencapsulated devices examined via MPP tracking under the continuous AM1.5 G equivalent light irradiation and nitrogen flow at 60°C. Error bars refer to the average deviations of four cells.

Device Efficiency and Stability

We adopted a triple cation perovskite, (FAPbI3)0.85(MAPbBr3)0.10(CsPbI3)0.05(PbI2)0.03, as the photoactive layer to fabricate PSCs on fluorine-doped tin oxide (FTO) substrates, based on the device structure: FTO/c-mp TiO2/perovskite/HTL/Au. In Figure 2B, we present the current density-voltage (J-V) curves for the champion cells, and the extracted photovoltaic parameters are tabulated in Table 1. The PCE of the O5H-OMeDPA-based cell reaches 21.03%, with a short-circuit current (J) of 24.48 mA cm−2, an open-circuit voltage (V) of 1.081 mV, and a fill factor (FF) of 0.788, outperforming the best spiro-OMeTAD control with a PCE of 20.44%. Furthermore, the maximum power point (MPP) tracking under AM 1.5G illumination (Figure S6) was conducted, affording a steady-state PCE of 20.03%. The hystrese of the corresponding cells were shown in Figure S7. The incident photon-to-electron conversion efficiency (IPCE) spectra of the cells are presented in the inset of Figure 2B. The O5H-OMeDPA cell achieves an integrated current density of 23.76 mA cm−2, which is in close agreement with the J-V scan. A good reproducibility can be perceived from the narrow cell parameter distributions of 10 devices (Figure S8).

Table 1

Photovoltaic Parameters of Champion PSCs Measured under an Irradiance of 100 mW cm−2, Simulated AM1.5G Sunlight

HTL	Light Intensity [mW cm−2]	JSCIPCEa [mA cm−2]	J_SC[mA cm−2]	V_OC[V]	FF	PCE[%]
O5H-OMeDPA	99.0	23.76	24.48	1.081	0.788	21.03
Spiro-OMeTAD	99.1	23.85	24.53	1.089	0.757	20.44

was derived with wavelength integration of the product of the measured IPCEs at the short-circuit and the standard AM1.5G emission spectrum (ASTM G173-03).

The operational stability tests were conducted by keeping PSCs at 60°C under 1 sun equivalent white-light-emitting diode (LED) illumination at MPP. After 100 h, the O5H-OMeDPA cell retains 86% of the initial PCE (Figure 2C). After the rapid initial burn-in decay, which is possibly caused by the migration of A-site cations of perovskite (Christians et al., 2018), the PCE remains practically stable. In contrast, the spiro-OMeTAD control has an over 40% loss in the initial PCE after 20 h (Figure 2C). The unstable behavior of spiro-OMeTAD is in agreement with the previous report (Domanski et al., 2018).

Film Morphology, Energy Level, and Hole Extraction

We performed atomic force microscopic (AFM) measurements to inspect the film morphology. The O5H-OMeDPA film deposited on the poly(3,4-ethylenedioxythiophene) polystyrene sulfonate substrate owns a smooth surface with a root mean square deviation of 0.39 nm (Figure S9). When being coated on the perovskite layer with LiTFSI and TBP as additives, the O5H-OMeDPA film still retains a continuous surface with full surface coverage (Figure 3A). By contrast, the spiro-OMeTAD control presents plenty of pinholes (the dark region in Figure 3B, marked with a red circle) and LiTFSI aggregates (the light region, marked with a green circle), in line with previous reports (Wang et al., 2018b).

Figure 3

Surface Morphology and Hole Extraction

(A and B) AFM images (1 μm × 1 μm) of O5H-OMeDPA (A) and spiro-OMeTAD (B) spin coated on the substrates of perovskite. Note that the hole-transporting films are doped with LiTFSI and TBP.

(C and D) Steady-state photoluminescence (PL) spectra (C) and time-resolved PL traces (D) of a pristine perovskite film deposited on a mesoporous alumina film and the counterpart covered with O5H-OMeDPA or spiro-OMeTAD.

To have a better understanding on the energy level alignment of O5H-OMeDPA with respect to the perovskite, UV photoelectron spectroscopy measurement was performed for the spin-coated thin films on an FTO glass. From the onset (Ei) and cutoff (Ecutoff) energy regions shown in Figure S10, we derived the HOMO energy level of solid O5H-OMeDPA, being −5.29 eV, which is 0.16 eV lower than −5.13 eV for spiro-OMeTAD. Compared with the bare perovskite film, photoluminescence (PL) was strongly quenched for the O5H-OMeDPA-coated sample, with a PL quenching (hole extraction) yield of 97.0% (Figure 3C), which is almost the same as the sample with spiro-OMeTAD (96.3%). The pristine perovskite film presents a long PL lifetime of 1.99 μs. The PL lifetime of an O5H-OMeDPA-coated perovskite film is reduced to 33.8 ns (Figure 3D), slightly shorter than that of the spiro-OMeTAD control (58.1 ns). Given the smaller driving force, the faster kinetics and higher yield of hole extraction suggest a stronger electronic coupling of O5H-OMeDPA with the perovskite than spiro-OMeTAD.

Discussion

To gain further insight into the interface interaction between the newly synthesized HTM and the perovskite surface, the binding energy (ΔEb) between O5H-OMeDPA and a modeled MAPbI3 perovskite slab (Figure 4A) was computed by DFT, see Supplemental Information for details. The obtained value (ΔEb) is −0.83 eV, which is significantly larger than that calculated for spiro-OMeTAD (−0.11 eV, Figure 4B). This study has unveiled the greater interfacial affinity of O5H-OMeDPA with perovskite, resulting in efficient hole extraction and higher interface stability. The isodensity plots in Figure 4C of valence band edge associated with the perovskite/O5H-OMeDPA confirm the large electronic coupling between the HTM and the perovskite, showing that the HOMO orbital of O5H-OMeDPA expands toward the MAPbI3 slab. Furthermore, the HOMO energy is perfectly aligned with the perovskite valence band as suggested by the projected density of states diagrams in Figure 4D.

Figure 4

Interface Interaction and Phase Stability

(A and B) Optimized geometries of O5H-OMeDPA (A) and spiro-OMeTAD (B) interacting with a MAPbI3 perovskite slab.

(C) Three-dimensional representation of valence band edge associated with the perovskite/O5H-OMeDPA interface.

(D) Projected density of states of O5H-OMeDPA molecule adsorbed on the (110) surface of MAPbI3.

(E and F) Comparison of experimental XRD patterns of the as-deposited films and the films aged at 60°C for 12 h with the calculated XRD patterns based on single-crystal parameters: O5H-OMeDPA (E) and spiro-OMeTAD (F).

The growth of large crystalline domains in the HTL with a thickness of a few tens of nanometers has been suggested as one of the degradation causes of PSCs at elevated temperatures (Malinauskas et al., 2015). Therefore we investigated the influence of continuous thermal stress on the microstructure of solution-deposited glassy films with the aid of X-ray diffraction analysis. As depicted in Figures 4E and 4F, the as-deposited thin films are amorphous for both O5H-OMeDPA and spiro-OMeTAD. Circular dichroism spectrum has proved the racemic feature of the solution-processed O5H-OMeDPA film. After 12-h heating at 60°C, the O5H-OMeDPA film maintains the amorphous characteristic, whereas the spiro-OMeTAD control evolves into a crystalline film verified by the occurrence of multiple X-ray diffraction (XRD) peaks. The crystalline phase of the annealed spiro-OMeTAD thin film matches well with the single-crystal structure reported by Shi et al. (Shi et al., 2016). During crystallization, molecules in the thin film will move, forming closely packed phases and cracks simultaneously. The heat-induced phase transition is likely to deteriorate the interfacial contacts, impacting the hole extraction and transportation behaviors. Also, the change in film morphology will weaken the blocking effect of the HTL on Au diffusion into the perovskite layer, which severely affects the device performance under working conditions (Domanski et al., 2016). In recent years, certain interface materials have been used to suppress metal diffusion (Hou et al., 2017, Chen et al., 2017, Christians et al., 2018, Wu et al., 2019), which is a promising approach to further improve the stability of O5H-OMeDPA-based devices. It is also reported that additives can accelerate the crystallization of spiro-OMeTAD (Zhao et al., 2017). Hence, we also compared the morphologies and XRD patterns of the doped O5H-OMeDPA films before and after thermal aging at 60°C. No remarkable change upon aging was observed from AFM images (Figure S13). Also, the 60°C aged film is still amorphous (Figure S14). Overall, the superior photo- and thermal stability of the O5H-OMeDPA-based PSCs demonstrated here highlights the grand potential of a racemic semiconducting glassy film of helicenes for organic optoelectronic devices.

Limitations of the Study

To validate the industrial pass-fail qualification tests IEC61215, the thermal stability of our PSCs was also evaluated at 85°C in the dark. The device degraded rapidly, which is likely associated with the crystallization of an amorphous O5H-OMeDPA film at 85°C (Figure S14), apart from the instability of the perovskite layer itself. This highlights the future need of developing new materials that can tolerate higher thermal stress.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.