Hole-Transporting Low-Dimensional Perovskite for Enhancing Photovoltaic Performance

Halide perovskites with low-dimensionalities (2D or quasi-2D) have demonstrated outstanding stabilities compared to their 3D counterparts. Nevertheless, poor charge-transporting abilities of organic components in 2D perovskites lead to relatively low power conversion efficiency (PCE) and thus limit their applications in photovoltaics. Here, we report a novel hole-transporting low-dimensional (HT2D) perovskite, which can form a hole-transporting channel on the top surface of 3D perovskite due to self-assembly effects of metal halide frameworks. This HT2D perovskite can significantly reduce interface trap densities and enhance hole-extracting abilities of a heterojunction region between the 3D perovskite and hole-transporting layer. Furthermore, the posttreatment by HT2D can also reduce the crystal defects of perovskite and improve film morphology. As a result, perovskite solar cells (PSCs) can effectively suppress nonradiative recombination, leading to an increasement on photovoltage to >1.20 V and thus achieving >20% power conversion efficiency and >500 h continuous illumination stability. This work provides a pathway to overcome charge-transporting limitations in low-dimensional perovskites and delivers significant enhancements on performance of PSCs.


Introduction
Metal-halide perovskites have been tremendously developed over the past several years because they can offer the promise of easy fabrication, low-cost solution processability, flexible substrate compatibility, broad bandgap tunability, and integration possibility of tandem multijunction architecture [1][2][3][4][5]. Owing to the excellent intrinsic properties of perovskite materials, such as extremely high absorption coefficient and ultralong charge carrier diffusion distance, given by the unique three-dimensional (3D) ABX 3 framework of perovskite polycrystals [6,7], perovskite solar cells (PSCs) have achieved very impressive power conversion efficiencies (PCEs) already exceeding 25% [8]. Despite this remarkable achievement, however, the unacceptable vulnerability of 3D perovskites to humidity and ambient atmosphere rises to a barrier toward their market uptake. In overcoming the drawbacks, 2D and quasi-2D perovskites formed by inserting bulky organic spacer cations, which cannot fit into the octahedral network but can effectively passivate interfacial defects and vacancies outside ABX 3 frameworks, have been considered as promising materials to solve the stability issue of PSCs [9][10][11][12].
Herein, we reported a novel design strategy toward an efficient bifunctional organic salt (TA-PMA) consisting of two subunits: (i) a phenylmethylammonium (PMA) cation group (red part in Figure 1(a)) for building up RP layered phase perovskites and (ii) a triarylamine (TA) group (blue part in Figure 1(a)) for efficient hole extraction from RP perovskite to the hole-transporting layer (HTL). This tailored TA-PMA could induce a surface-2D/bulk-3D hierarchy perovskite structure, which not only can improve the stability of PSCs against damp environment due to the hydrophobic nature of formed RP perovskite but also is expected to enable a hole-transporting channel between RP perovskite slabs because of the hole-extracting ability of the TA functional group (Figure 1(b)). Moreover, the BA cation could naturally provide the desired passivation effect by forming an adduct with uncoordinated halide ions [27]. In addition, HT2D posttreatment can eliminate the miscellaneous phase of perovskite and thus improve the film morphology. Benefiting from these merits, the n-i-p planar PSC achieved a champion power conversion efficiency (PCE) of 20.71% as well as excellent operational stability exhibiting 92% of initial efficiency after 500 h continuous illumination without encapsulation.

Results
2.1. Characterization of TA-PMA. The TA-PMA was synthesized from a straightforward route with simple purifications and high yields ( Figure S1). The chemical structures of target TA-PMA was characterized by 1 H-NMR (Figure 1(c) Figure S2 and Figure S3. In order to identify the interaction of TA-PMA and lead iodide (PbI 2 ), we mixed them at a molar ratio of 2 : 1, then observed that 1 H a disappeared, and 1 H b shifted to 7.574 (d, 2H), 7.479 (d, 2H), and 7.404 (d, 2H) ppm, ( 1 H m ) shifted to 3.895 (s, 2H) ppm, but 1 H t was almost unchanged. These chemical shifts were attributed to varying distances between protonated ammonium and neighboring hydrogens, indicating that the reaction of TA-PMA and PbI 2 mainly occurred by the amine group and could potentially achieve a 2D perovskite (n = 1) phases at a molecular level. Moreover, to observe these 2D layered structures in solid state, X-ray diffraction (XRD) patterns of TA-PMA doped with PbI 2 at a ratio of 2 : 1 (mol) were measured using phenylmethylammonium (PMA) bromide and dimethoxy-triarylamine (TA) as references under the same condition ( Figure 1(d)). Only TA-PMA/PbI 2 and PMA/PbI 2 mixture films demonstrated significantly RP layered 2D peaks at 5.3°and 6.6°(marked in Figure 1(d)), respectively. The diffraction angle of PMA-treated 2D perovskite was similar to previous literature [28], whereas the smaller diffraction angle of TA-PMA-treated perovskite was following the tendency of its larger d-spacing distance owing to a bulkier structure than that of PMA [29]. In contrast, TA/PbI 2 films displayed no extra peaks due to the lack of a reactive cation group.  Figure S4. Moreover, for further understanding, the interfacial charge transportation mechanism was illustrated. It is observed that the partial charge densities of the highest occupied molecular orbital (HOMO) are mainly located at the domain of TA-PMA slabs (Figure 2(a)), whereas those of the lowest unoccupied molecular orbital (LUMO) are distributed in the perovskite region ( Figure 2(b)). These obvious distinctions between HOMO and LUMO charges indicates that the hole transportation direction at the interface is from perovskite to TA-PMA. Density of state is also calculated to illustrate the electronic change ( Figure S5). The relative energy difference of TA-PMA and MAPbI 3 is~1.7 eV, which is identical to the experimental energy gap between HOMO of TA-PMA (−5.6 eV) and LUMO of MAPbI 3 (−3.9 eV) [30].
Herein, we named this novel structure consisting of TA-PMA and perovskite as a hole-transporting 2D perovskite (HT2D). In other words, the spontaneously grown HT2D perovskite can make a contribution to the hole extraction from the perovskite photoactive layer to the holetransporting layer (HTL), thus restricting the interfacial charge recombination.

Structure Characterization of HT2D/3D Hierarchy
Perovskite. To investigate the hierarchy structure of HT2D/3D perovskite films, the synchrotron-based grazingincident wide-angle X-ray scattering (GIWAXS) was conducted on a series of samples, including neat HT2D perovskite, pristine 3D MAPbI 3 (3D), and hierarchy HT2D/3D perovskite based on TA-PMA posttreatment (see experimental details in Supporting information). GIWAXS measurements were performed as a function of the X-ray incident angle (α), providing depth-sensitive information in 2Dscattering patterns. X-ray beam could fully penetrate the hierarchy film samples with α = 0:20°, whereas it only analyzes the top surface if under critical angle α = 0:02°. In Figures 2(c) and 2(d), a neat HT2D sample prepared by stoichiometrical TA-PMA and PbI 2 exhibited scattering patterns at qz~0:19, 0.38, 0.57, and 0.76 Å −1 , corresponding to (100), (200), (300), and (400) crystal faces of a RP layered 2D (n = 1) perovskite, respectively. These diffraction data resulted in the lamellar stacking distance between perovskite octahedral units which was ca. 33 Å, which was in coincidence with simulated double RP layered distance as first-principle calculation results. Herein, the above HT2D (n = 1) can be formed by the following reaction): In Figures 2(e) and 2(f), the MAPbI 3 film demonstrated uniform 3D perovskite (110) and (220) diffraction rings in both surface and bulk phases. To form the HT2D/3D structure, TA-PMA in isopropanol (IPA) solution was spin-coated on the MA3D film and then, the posttreated film was annealed for 10 min. Figures 2(g) and 2(h) presented a direct evidence on this hierarchy HT2D/3D structure. In the bulk phase, this hierarchy HT2D/3D sample manifested obvious typical 3D perovskite (110) and (220) diffractions as well as quasi-2D diffractions at q z from 0.32 to 0.75 Å −1 , attributing to small-n quasi-2D phases (n = 2, 3, 4,…) [31,32]. Moreover, at the small grazing-incident angle, the 3D perovskite scattering intensities were significantly attenuated, whereas the layered diffraction displayed a significantly sharp peak at q z~0 :45 Å −1 with a narrow full width at half maxima, indicating a larger average grain size and higher ordered crystallite orientation caused by the quasi-2D perovskite phase on the top surface layer [28]. Thus, we inferred that small-n quasi-2D phases (n = 2, 3, 4,…) could be achieved by the following reaction: The resultant MAI could be dissolved in IPA and further disappeared during spin-coating and thermal annealing process [33]. In the GIWAXS patterns of triple-cation perovskites (CsFAMA3D), it could also be clearly detected that pristine CsFAMA3D (Figures 2(i) and 2(j)) changed to a hierarchy HT2D/CsFAMA3D (Figures 2(k) and 2(l)) structure after the TA-PMA posttreatment.
2.4. Hole-Transporting Properties of HT2D Perovskite. In order to further investigate the hole mobility of HT2D perovskite, space-charge-limited-current (SCLC) measurements ( Figure S6) based on hole-only devices were prepared as the following architecture: ITO/PEDOT:PSS/TA-PMA or HT2D/MoO 3 /Ag. The experimental hole mobility was significantly increased in more than one order of magnitude from 2:1 × 10 −4 cm 2 V -1 S -1 of TA-PMA to 2:6 × 10 −3 cm 2 V -1 S -1 of HT2D, which was comparable to doped Spiro-OMeTAD as HTL [34]. This enhanced hole mobility might be due to the following three reasons: (i) generally, the charge mobility of the 2D perovskite was relatively low owing to the insulating bulky organic layers which had a low hole mobility. In this work, the introduction of the triarylamine (TA) group could improve the hole mobility of the bulky organic molecules. Therefore, the TA-PMA-treated HT2D perovskite exhibited a high hole mobility. (ii) In addition, according to the DFT calculations, the TA units displayed a highly orientated arrangement along with the octahedral [PbI 6 ] 4− framework in HT2D perovskite, since the PMA cation groups in TA-PMA could fully embed into the A-site at the (010) lattice plane of ABX 3 3D perovskite, whereas the molecule arrangement in the neat TA-PMA film was randomly unordered owing to the lack of perovskite framework as the template. (iii) Furthermore, the azimuth angle integration curve of HT2D diffraction peaks at q~0:45 Å −1 displayed the appearance of this peak which was along the out-ofplane direction (q z ) and not the in-plane direction (q xy ), supporting the vertical packing of HT2D grown on substrate. These vertical packing HT2D can form a holetransporting channel and thus perform higher hole mobility.

Morphology and Photoelectric
Properties of TA-PMA Posttreatment Perovskite Films. We further investigated the morphology and crystalline phase changes of CsFAMA3D perovskite films induced by TA-PMA posttreatment. To simplify, in the following discussion, CsFAMA3D is abbreviated as 3D. As shown in Figure 3(a), the top-view SEM images of controlled 3D perovskites showed that many small grains of unreacted PbI 2 were distributed in the grain boundaries of perovskite polycrystals, due to the disordered growth in perovskite crystallinity during thermal annealing process [33][34][35][36][37]. Significantly, after TA-PMA posttreatment (Figure 3(b)), these small grains seemed to completely disappear and crystal grains were significantly enlarged, due to the process of the aforementioned reaction (equation (1) and equation (2)) and Ostwald ripening [38]. Moreover, root mean square (RMS) surface roughness estimated from AFM images ( Figure S7) demonstrated that the surface of the perovskite film became smoother and more compact for TA-PMA-modified 3D perovskites, ascribing to the formation of the HT2D perovskite layer [39]. Besides morphology modification, the surface-HT2D structure on the bulk-3D perovskite could also affect its work function and the interfacial band alignment. We performed photoelectron spectroscopy in air (PESA) on a series of samples (Figure 3(c)), exhibiting HOMO levels at −5.7 eV for 3D perovskite, −5.6 eV for HT2D, −5.5 eV for TA-PMA, and −5.2 eV for Spiro-OMeTAD as HTL, respectively. As shown in Figure 3(d), these gradually raised energy levels of 3D/HT2D/HTL architecture could effectively improve hole extraction and suppress interfacial nonradiative recombination between light-harvesting 3D perovskite and HTL in typical configuration. Furthermore, both steady-state photoluminescence (PL) (Figure 3(e)) and time-resolved photoluminescence (TRPL) (Figure 3(f)) spectra presented an effective enhancement on exciton/charge carrier extraction by using an intermediate HT2D layer. Compared to the 3D perovskite, the 3D/HT2D film showed a normalized PL intensity of 29.9% and calculated average PL decay time (τ avg ) of 45.16 ns, which should be ascribed to the holeextracting effect of the HT2D capping layer. Furthermore, these values of 3D/HT2D/HTL film could further reduce to 7.9% and 7.15 ns, which indicated that the gradual energy levels were more advantage for hole extraction compared to traditional 3D/HTL interface (13.8% and 9.61 ns). The fitted values of τ 1 , τ 2 , A 1 , and A 2 and the calculated τ avg of the samples are summarized in Table S1.
2.6. Photovoltaic Performance and Stability. The surface-2D/bulk-3D hierarchy perovskite films were further fabricated into PSCs with a conventional planar n-i-p configuration of FTO/SnO 2 /perovskite/Spiro-OMeTAD/Au. Figure 4(a) listed the current density-voltage (J -V) curves of HT2D/3D-and 3D-based PSCs. Noticeably, the mixed cation PSCs based on HT2D/3D delivered a high PCE up to 20.71% compared to the pristine 3D PSCs (PCE = 18:53%), achieving a photovoltage (V oc ) of 1.21 V, a current density (J sc ) of 22.81 mA cm -2 , and a fill factor (FF) of 75.08%. The photovoltaic parameters of HT2D/3D PSCs were significantly enhanced with the reducing defect density of the TA-PMA-treated perovskite layers. As shown in Figure S8, the trap filled limit voltage (V TFL ) of 3D and HT2D/3D perovskite films were 1.87 V and 0.70 V, respectively. According to the above logarithmic J − V analysis, the calculated defect concentration N defects of HT2D/3D perovskite films (7:96 × 10 15 cm −3 ) was an order of magnitude lower than the 3D counterpart (7:52 × 10 16 cm −3 ) [40]. The lower defect density of HT2D/3D perovskite films indicated in which fewer nonradiative recombination centers existed due to less miscellaneous phases. The PV parameters of the devices are listed in Table S2. Their typical J -V hysteresis curves are listed in Figure S9. The HT2D/3Dbased PSCs showed not only enhanced PV performance but also higher reproducibility than the pristine 3D-based PSCs ( Figure S10). HT2D/3D-type PSCs processed with different contents of TA-PMA were analyzed to demonstrate the effect of TA-PMA on device performance. As shown in Figure S11 and Table S3, TA-PMA-treated PSCs exhibited a synchronous enhancement of V oc , J sc , and FF by optimizing an appropriate thickness of the HT2D layer. Nevertheless, using a similar concentration of PMA and TA as counterparts, PMA-modified PSC showed a slight enhancement on device performance, because PMA possessed only a low-dimensional unit but no improvement on hole transportation. TA-treated PSC presented even poor performance since it could not form a 2D structure and thus insulated hole extraction from perovskite to HTL ( Figure S12 and Table S4). Incident photon-to-electron conversion efficiency (IPCE) measurements were carried out to validate the current densities of devices (Figure 4(b)). All IPCE spectra showed the onset of around 820 nm (~1% IPCE) and exhibited quantum efficiency values of over 80% from 400 to 780 nm. The integrated short-circuit current density (J sc ) calculated from the IPCE spectrum was 21.94 mA cm −2 for HT2D/3D PSC and 20.99 mA cm −2 for 3D PSC, which matched well (~5%) the J sc measured under AM 1.5G illumination. Furthermore, the HT2D/3D PSC also delivered a stabilized photocurrent of 21.67 mA cm −2 , corresponding to a stabilized efficiency of 19.68%, under a constant voltage bias near the maximum power point (0.91 V) (Figure 4(c)), which was very close to the J -V scan efficiency.
Electrochemical impedance spectroscopy (EIS) analyses of PSCs were carried out to further characterize the charge transfer in the full devices based on HT2D/CsFAMA3D perovskites (Figure 4(d)) [41][42][43]. The Nyquist plots were obtained in the dark with an applied bias voltage of 0.5 V, and the fitted parameters are summarized in Figure 4(d). As shown in the equivalent circuit, the internal series resistance (R s ) was related to the sheet resistance of the electrodes. The R s values of the HT2D/CSFAMA3D and pristine CsFA-MA3D PSCs were similar at 23.1 Ω and 31.8 Ω, respectively. On the other side, the charge-transfer resistance (R ct ) generally refers to the contact resistance at all the interfaces such as the electrode/transporting layer/perovskite interface, which is corresponding to the semicircle in the high-frequency region [44]. At the V oc of 0.5 V, the R ct value of HT2D/3D PSC at 195 Ω was less than half the R ct value of pristine CsFAMA3D PSC at 489 Ω, due to its better interfacial contact and charge transportation [45,46]. The Nyquist plots of HT2D/CsFAMA3D and pristine 3D PSCs from 0 V-0.5 V showed similar trends shown in Figure S13 and Table S5.
In addition to the enhanced PV performance, the HT2D/CsFAMA3D PSCs exhibit excellent stability. In the unencapsulated HT2D/CsFAMA3D PSCs, 91% of the initial PCE could remain and a small hysteresis between forward  [2]; (e, f) images showing PL (e) and TRPL (f) spectra of 3D, 3D/HT2D, 3D/HTL, and 3D/HT2D/HTL. and reverse JV sweep existed when stored in ambient condition with RH 25 ± 5% at 30°C for 1000 h (Figure 4(e) and Table S6). Contact angle analysis suggested that the high moisture stability of the TA-PMA-treated PSCs was ascribed to the more hydrophobic surface of the 2D structure, suggested by larger contact angles (64.2°) than 3D perovskite (48.6°) when using water droplets ( Figure S14) [47]. Furthermore, the HT2D/3D PSC showed better photostability and 91% of the initial PCE remained after continuous illumination with white LED light for~500 h in dry N 2 glove box, while the pristine 3D PSC was below 90% of initial PCE at 200 h (Figure 4(f)).

Discussion
In this work, we have designed and synthesized a novel hole transporting organic salt TA-PMA, which can induce a surface-2D/bulk-3D hierarchy perovskite structure in both MAPbI 3 and mixed-cation perovskite. The resulted HT2D perovskite on the 3D-bulk film surface could regulate the interfacial band alignment, suppress the interfacial charge recombination, improve the perovskite crystallization and film morphology, and enhance the hole mobility and charge extraction ability. As a result, all the photovoltaic parameters of the HT2D/3D PSC device including J sc , V oc , and FF were significantly enhanced, and the champion PCE of 20.71% was achieved along with the higher reproducibility than pristine 3D-based PSCs. Besides efficiency, another critical factor for photovoltaics was environmental stability and photostability under operating conditions. The stress tests demonstrated that HT2D/3D PSCs were much stabler over long-term operation against moisture and light soaking in contrast to the pristine 3D PSCs, owing to the hydrophobic nature properties of surface-2D layer. This work will offer a pathway to not only overcome the stability challenges for bulk 3D perovskite but also inspire the molecular design of novel hole-transporting 2D perovskite in the future.

Preparation of HT2D Perovskite
Films. The neat HT2D perovskite film was prepared as follows: 0.1 mmol of TA-PMA and 0.05 mmol of PbI 2 were dissolved in 1 mL of DMF, spin-coated on the substrate at 3000 rpm for 30 s, and then annealed at 100°C for 10 minutes.
where ε 0 is the permittivity of free space, ε r is the dielectric constant of the TA-PMA:PbI 2 , μ h is the hole mobility, V is the voltage drop across the device, and L is the TA-PMA:PbI 2 film thickness; in formula, V = V appl − V r − V bi , V appl is the applied voltage to the device, V r is the voltage drop due to constant resistance and series resistance across the electrodes, and V bi is the built-in voltage due to the difference in work function of the two electrodes. The current density versus voltage characteristics were recorded on a Keithley 2400 source meter.

Data Availability
The data that support the finding of this study are available from the corresponding author upon reasonable request.