Molecular interfaces drive vertical crystallization in Dion-Jacobson perovskite solar cells Authors: Rui Wang, Xiyue Dong, Yuting Ma, Liu Yang, Jiangnan Li, Yuping Gao, Yu Chen, Yu Zou, Wenjuan Feng, Ziyang Hu, Yongsheng Chen, Yongsheng Liu Published: 2026 ABSTRACT Two-dimensional perovskites are promising candidates for photovoltaics due to their intrinsic structural stability, but their efficiency is often limited by poor charge transport, in part due to unfavorable crystal orientation. Here, we report a molecular interface engineering strategy using dual-anchoring organic acids, croconic acid (CA) and squaric acid (SA), to direct vertical crystallization in Dion-Jacobson (DJ) perovskite films. These molecules form robust interlayers between the NiOx hole transport layer and TTDMA (thieno[3,2-b]thiophene-2,5-diyldimethanaminium)–based DJ perovskites (nominal n = 4), with SA exhibiting ordered vertical orientation via bidentate coordination. This templated interface promotes vertical orientation, reduces interfacial defects and lattice strain, and suppresses Ni3+-induced oxidation of I−. As a result, devices incorporating SA achieve a champion power conversion efficiency of 22.03% (certified 21.42%) along with outstanding operational stability. This study demonstrates a general molecular interface strategy to direct vertical crystallization and improve the performance of layered perovskite solar cells. INTRODUCTION Two-dimensional (2D) metal halide perovskites have attracted intense interest for photovoltaic applications due to their superior intrinsic structural and environmental stability compared to their 3D counterparts (1–5). Among 2D perovskite families, Dion-Jacobson (DJ) phases that comprise divalent organic spacers directly connecting adjacent corner-sharing [PbI6]4− layers offer improved mechanical rigidity by eliminating van der Waals gaps found in Ruddlesden-Popper (RP) phases. Despite recent advances, power conversion efficiencies (PCEs) of DJ perovskite solar cells (PSCs) still lag behind their 3D counterparts, (6–14) primarily due to strong excitonic effects and intrinsic anisotropic charge transport that hinder efficient carrier extraction (15). Incorporating spacers with conjugated structure and/or large dipole moments has proven to be an effective strategy for reducing excitonic effects (16–23). Beyond this, a critical factor limiting performance is the crystallographic orientation of the 2D perovskite layers, where vertical alignment is essential for efficient out-of-plane charge transport and minimizing recombination losses (24). To date, strategies to achieve vertical orientation have primarily focused on processing techniques, such as hot-casting, solvent engineering, and the incorporation of additives (24–30). A landmark study by Mohite and colleagues. demonstrated that hot-casting method could induce vertical orientation in RP perovskites, achieving a notable PCE of 12.52% with remarkable stability (24). This work catalyzed broad interest in morphological control for 2D perovskite films. Additives such as MACl and NH4SCN have since been used to modulate the crystallization kinetics and promote vertical alignment (26, 27). However, molecular-level strategies at the buried bottom interface to direct such vertical crystallization remain underexplored. Here, we present a molecular interface engineering strategy that directs vertical crystallization in DJ perovskite films and markedly boosts device performance. Specifically, two dual-anchoring molecules, croconic acid (CA) and squaric acid (SA), were explored as functional interlayer between NiOx hole transport layer (HTL) and DJ perovskite layer based on thieno[3,2-b]thiophene-2,5-diyldimethanaminium (TTDMA). These molecules exhibit strong interactions with both the NiOx surface and the overlying DJ perovskite. Notably, although CA forms a twisted, disordered layer due to multivalent coordination with NiOx, SA adopts an ordered vertical orientation via bidentate coordination, effectively serving as a molecular template that guides perovskite nucleation and vertical growth. This templating effect promotes the formation of highly crystalline of (TTDMA)(MA0.4FA0.6)n−1PbnI3n+1 (TTDMA-Pb, nominal n = 4) films with well-defined vertical orientation. In addition to structural regulation, the SA interface layer simultaneously passivated interface defects, relieves lattice stress, aligns interface energy levels, and suppresses Ni3+-induced I− oxidation. As a result, devices incorporating SA achieve a record PCE of 22.03% (certified 21.42%) for DJ PSCs, along with superior operational stability. The universality of this method that directs vertical crystallization was further validated in RP perovskites using thieno[3,2-b]thiophen-2-ylmethanaminium (TTMA) as the spacer. RESULTS Functional organic molecules with dual-anchoring properties The molecular structure of SA and CA are illustrated in Fig. 1A. Both contain multiple carbonyl (─C═O) and hydroxyl (─OH) groups capable of interacting with both NiOx and the perovskite layer in inverted PSCs. First-principles density function theory (DFT) calculations (fig. S1A) reveal that the ─C═O group exhibits a larger binding energy (Eb) when adsorbed onto the NiOx surface, indicating a stronger interaction with undercoordinated Ni sites. As a result, SA preferentially adopts a vertical orientation on the NiOx surface via bidentate coordination through two C═O⋯Ni bonds (Fig. 1A). In contrast, CA forms a twisted configuration due to its three C═O⋯Ni interactions, leading to a less ordered interface (Fig. 1A and fig. S1B). Compared to CA, the vertical orientation of SA molecules reduces steric hindrance and allows for higher molecular packing density (Fig. 1B and fig. S2). This enhanced packing leads to smoother surfaces and reduced surface potential differences, as verified by atomic force microscopy (AFM) and conductive AFM (c-AFM) analyses (figs. S3 and S4). Moreover, the NiOx/SA film exhibits a smaller water contact angle than NiOx and NiOx/CA (fig. S5), indicating improved hydrophilicity. This enhanced wettability facilitates better interaction with the perovskite precursor solution, promoting uniform film formation and improved crystallization as discussed below. Dual-anchoring properties of functional organic molecules. (A) Chemical structure and the optimal conformation for SA and CA adsorption at the interface between perovskite (PVSK) and NiOx. (B to D) XPS spectra of Ni 2p3/2 for NiOx, NiOx/CA, and NiOx/SA. a.u., arbitrary units. Benefiting from the robust coordination between the ─C═O groups and the NiOx surface, both CA and SA molecules can form robust self-assembled monolayers (SAMs) that are not easily removed by dimethyl sulfoxide (DMSO)–based perovskite precursor solution, as confirmed by ultraviolet-visible (UV-Vis) absorption spectra (fig. S6). The optimal dual-anchoring interactions of CA (and SA) between NiOx and perovskite are illustrated in Fig. 1A. Both SA and CA prefer to absorb on the NiOx surface via C═O⋯Ni interactions, exposing ─OH groups that subsequently interact with the overlying 2D perovskite layer. Notably, DFT calculations show that the O─Pb bond lengths between SA and the perovskite is at the range of 3.22 to 3.32 Å, which are narrower and more uniform than those CA (2.77 to 3.64 Å), indicating more homogeneous bonding with the perovskite (fig. S7). Furthermore, the molecular spacing of adjacent SA molecules (5.96 Å) is more compatible with the perovskite lattice (6.21 Å) compared to CA (6.67 Å) (31, 32). This better lattice match can enhance the anchoring and adsorption of SA onto the perovskite through chemical interactions while also guiding the alignment of perovskite crystals, thereby promoting a vertical orientation (31). Furthermore, the ─OH groups of SA exhibit a larger binding energy with perovskite (Eb = −0.152 eV) than CA (Eb = −0.150 eV), suggesting that SA is more effective in anchoring the perovskite crystals, passivating defects, and regulating the crystallization and orientation (33). The interaction between SA/CA and the NiOx layer were further supported by x-ray photoelectron spectroscopy (XPS) (Fig. 1, B to D). After SA modification, the Ni 2p3/2 peaks at 852.85 eV (Ni2+) and 854.48 eV (Ni3+) shifted to 853.35 eV (Ni2+) and 854.92 eV (Ni3+), respectively, consistent with coordination between Ni and the functional groups of SA. A similar trend was observed for CA. Fourier transform infrared (FTIR) spectra (figs. S8, A and B) show that the ─C═O stretching vibration peaks exhibit more pronounced shifts than ─C─O, indicating stronger interaction between NiOx and ─C═O than with ─OH. Upon addition of PbI2, the ─C─O stretching vibrations show a pronounced shift (fig. S8, C and D), indicating strong interactions with PbI2, consistent with DFT results (34). Crystal growth and orientation directed by interfacial molecules We performed in situ photoluminescence (PL) measurements of TTDMA-Pb films on various substrates to investigate the impact of functional molecules on perovskite crystallization (Fig. 2, A to F). At 40°C for 120 s (Fig. 2, A to C), the TTDMA-Pb film deposited on NiOx/CA exhibits an initial PL emission at ~750 nm after ~105 s, occurring earlier than those on indium tin oxide (ITO)/NiOx (~114 s) and ITO/NiOx/CA (~108 s) substrates. This observation further confirms accelerated nucleation as the vertically aligned, densely packed SA layer provides a high density of uniform nucleation sites (─OH coordination sites) that facilitate the initial assembly of perovskite precursors, whereas its enhanced wettability (fig. S5) lowers the heterogeneous nucleation barrier (35). During subsequent annealing at 100°C for 200 s (Fig. 2, D to F), the PL intensity underwent a characteristic evolution: an initial increase, followed by a decrease, and then a final increase. This trend signifies a crystallization-dissolution-recrystallization process (36). Notably, SA substantially narrows the kinetic window for recrystallization following surface crystal dissolution, promoting perovskite recrystallization and improving crystalline quality. Functional molecules guide the crystallization and morphology control of DJ perovskites. (A to C) In situ PL spectra of TTDMA-Pb films on different substrates at 40°C. (D to F) In situ PL spectra of TTDMA-Pb films on different substrates at 100°C. (G to I), Top morphologies of TTDMA-Pb films deposited on different substrates (insets are grain size distribution). (J to L) Bottom morphologies of TTDMA-Pb films deposited on NiOx, NiOx/CA, and NiOx/SA substrates. To evaluate the impact of substrate modification on the film quality of TTDMA-Pb, top-view scanning electron microscopy (SEM) measurements were performed. As shown in Fig. 2 (G to I), TTDMA-Pb films deposited on ITO/NiOx substrates modified with CA or SA exhibit obviously increased grain size compared to that on unmodified ITO/NiOx. Notably, the film on the ITO/NiOx/SA substrate displays a pronounced vertical growth with large, monolithic grains (fig. S9), which facilitate efficient charge transport between top and bottom electrodes. To gain deeper insights into the effects of SA and CA on the morphology of DJ perovskites, we obtained the SEM images of the buried bottom surfaces of TTDMA-Pb films. These surfaces were exposed by peeling off the perovskite layers using solidified adhesive. As depicted in Fig. 2 (J to L), the bottom surface of NiOx/TTDMA-Pb films exhibits obvious pinholes compared to the top surface. Notably, the bottom surfaces of the film on bare NiOx exhibits noticeable pinholes, indicative of poor interfacial coverage. In contrast, films deposited on ITO/NiOx substrates modified with CA or SA show notably improved bottom surface uniformity with minimal pinhole formation. In addition, a higher density of small NiOx particles was observed on the bottom surface of the film grown on CA or SA modified substrates compared to the unmodified control substrate. This indicates a strong interfacial interaction between the dual-anchoring molecules and both TTDMA-Pb and NiOx, consistent with the DFT calculation results. Note that we repeated the peeling and SEM characterization on multiple samples (fig. S10) and consistently observed these features, indicating that they predominantly originate from intrinsic perovskite-substrate properties rather than from SEM preparation artifacts. To investigate the depth-dependent crystallization behavior of TTDMA-Pb films, grazing incidence wide-angle x-ray scattering (GIWAXS) measurements were performed at various incident angles. As shown in Fig. 3 (A to C), the as-deposited wet films on different substrates exhibit a mixed-phase composition, comprising both the perovskite phase (q ≈ 1 Å−1) and an intermediate phase (q ≈ 0.5 Å−1), extending from the film surface into the bulk (37). Upon annealing, the intermediate phase gradually transforms into the perovskite phase. Notably, for films deposited on bare NiOx, crystallinity progressively decreases from the surface to the buried interface (fig. S11, A to C). In contrast, the incorporation of CA or SA at the TTDMA-Pb/NiOx interface substantially enhances perovskite crystallinity in the buried region, suggesting that interfacial functional molecules promote nucleation and improve film quality, consistent with the in situ PL results. To gain deeper insight into the orientation distribution of DJ perovskites, azimuth angle plots were derived by integrating the diffraction intensity along the (111) ring (q = 0.9 to 1.1 Å−1). As depicted in Fig. 3 (D to F), at a low incident angle of 0.1°, all TTDMA-Pb films exhibit preferred orientation at a polar angle (χ) of 0°, indicating an intrinsic tendency for vertical alignment at the surface, likely driven by the self-assembly of TTDMA spacer cations. With increasing incident angle (0.7° to 0.9°), the TTDMA-Pb film deposited on bare NiOx exhibits increased diffraction intensity at 40° < χ < 90° (Fig. 3D), suggesting the tilted crystal orientation at the buried interface. In addition, a broader orientation distribution emerges with progressive solvent evaporation (fig. S11D), implying increased disorder during film solidification. In contrast, films deposited on CA-modified NiOx exhibit a predominantly vertical orientation (χ = 0°) with a minor fraction of tilted grains (χ = 50°) at the buried interface (Fig. 3E and fig. S11E). Notably, the SA-modified substrate leads to highly oriented vertical crystallization throughout the film, as evidenced by a sharp, dominant diffraction signal at χ = 0°, which intensifies as solvent evaporation proceeds. (Fig. 3F and fig. S11F). Crystal growth and orientation of DJ perovskites. (A to C) Incident angle–dependent GIWAXS of TTDMA-Pb films on different substrates (left), along with their respective GIWAXS (T = 100°C) with an incident angle of 0.9° (right). (D to F) Radially integrated intensity plots along the (111) rings of TTDMA-Pb films on NiOx (D), NiOx/CA (E), and NiOx/SA (F). (G to I) Schematic illustrations of the crystal orientation of TTDMA-Pb on different substrates. These findings demonstrate that dual-anchoring functional molecules effectively drive the transition from randomly oriented to vertically aligned crystal growth in DJ perovskites. This underlying mechanism is further supported by combining the aforementioned DFT calculations and in situ PL results. As illustrated in Fig. 3G and figs. S12 and S13, in the absence of interfacial modulation, the isotropic environment at the buried bottom interface favors random crystal orientations (38). The introduction of CA molecules establishes a more ordered interfacial environment, enhancing alignment but still allowing for some degree of misorientation due to twisted CA on the NiOx surface (Fig. 3H). In contrast, SA molecules, owing to their compact structure and minimal steric hindrance, achieve higher surface coverage and vertical alignment on NiOx. Through bidentate coordination between their hydroxyl groups and Pb2+ ions, SA molecules stabilize local Pb2+ concentrations near the interface, promoting heterogeneous nucleation. Furthermore, the lattice matching between SA and perovskite facilitates epitaxial growth, leading to the formation of vertical orientated and high-quality TTDMA-Pb films (Fig. 3I). Electronic properties and lattice stress of perovskite films To investigate the impact of CA and SA on the electronic properties of NiOx, we conducted kelvin probe force microscopy (KPFM) and ultraviolet photoelectron spectroscopy (UPS) measurements. The KPFM results in Fig. 4 (A to C) show a sequential decrease in average surface potential from −46.7 mV for pristine NiOx to −139.1 and −209.5 mV for NiOx/CA and NiOx/SA, respectively. This trend is consistent with the downward shift in Fermi level (EF) for the corresponding films, as confirmed by UPS (fig. S14) (39–41). The EF values are −4.26 eV for NiOx, −4.46 eV for NiOx/CA, and −4.65 eV for NiOx/SA, with the corresponding valence band maximum (VBM) of −5.22, −5.39, and −5.49 eV, respectively. Consequently, the EF-VBM energy gap narrows from 0.96 eV (NiOx) to 0.93 eV (NiOx/CA) and further to 0.84 eV (NiOx/SA). The reduced EF-VBM energy gap could promote the excitation of electron in valence band to the conduction band, leading to increased hole concentration and conductivity of the NiOx film. Electronic properties of perovskite films. (A to C) KPFM images and statistical potential distributions of NiOx, NiOx/CA, and NiOx/SA films. (D to F) Energy diagram of NiOx, NiOx/CA, NiOx/SA, and TTDMA-Pb films. (G) GIXD spectra for TTDMA-Pb films on different substrates at different tilt angles. (H) d-spacing values obtained from the (111) plane as a function of incidence angle. Figure 4 (D to F) illustrate the energy band alignment at the NiOx/TTDMA-Pb interface with and without interfacial molecular modification. The EVBM offset (ΔE) between TTDMA-Pb and NiOx is reduced from initial 0.32 eV (pristine) to 0.15 eV with CA and further to 0.05 eV with SA. This progressive alignment enhancement facilitates more efficient hole extraction and transport, contributing to improvements in both open-circuit voltage (VOC) and short-circuit current density (JSC) as discussed below (42). Steady-state PL and time-resolved PL (TRPL) measurements further confirm this behavior (fig. S15 and table S1). TTDMA-Pb films on CA-modified and SA-modified NiOx exhibit stronger PL quenching and shortened carrier lifetimes, indicating more efficient hole extraction from TTDMA-Pb to the HTL. We note that the nominal n = 4 composition inherently forms a heterogeneous phase distribution (fig. S15), consistent with the literature report on solution-processed quasi-2D perovskites (43). Residual stress within perovskite films can compromise structural integrity and device longevity (44). To evaluate the effect of substrate properties on strain relaxation, we conducted grazing incidence x-ray diffraction (GIXD) measurements. As depicted in Fig. 4 (G and H), increasing the grazing incidence angle (ω) lead to a gradual shift of the (111) diffraction peak toward smaller angles in the NiOx/TTDMA-Pb film, indicating lattice expansion and stress accumulation (33). In contrast, this shift is evidently attenuated in CA-modified and SA-modified samples. The slopes of the d-spacing versus ω curve decreased from 0.053 (NiOx/TTDMA-Pb) to 0.033 (CA-modified) and 0.019 (SA-modified), confirming a reduction in residual stress. These results suggest that dual-anchoring molecules promote strain relaxation during crystallization, with SA being more effective due to its superior lattice matching with the perovskite phase. Photovoltaic performance To evaluate the photovoltaic performance of TTDMA-Pb-based DJ PSCs, we fabricated inverted devices using NiOx, NiOx/CA, and NiOx/SA as HTLs. The detailed photovoltaic parameters are summarized in table S2. Note that SA and CA do not function as effective SAM layers on bare ITO, and their beneficial effects arise only when they modify and passivate the NiOx surface, which facilitates efficient hole extraction and promotes improved perovskite nucleation and orientation. In addition, for the CA and SA deposition, we used spin-coating followed by an optimized annealing step (100°C for 3 min) to allow the molecules to form well-ordered interfacial layers on NiOx (fig. S16). As shown in Fig. 5A, the reference device using pristine NiOx exhibits a VOC of 1.09 V, a JSC of 23.11 mA cm−2, and a fill factor (FF) of 77.05%, achieving a PCE of 19.41%. After using CA as dual anchoring interfacial layer, the PCE was increased to 20.31%, with a VOC of 1.12 V, a JSC of 23.25 mA cm−2, and an FF of 77.99%. The SA-based device shows a record-breaking PCE of 22.03%, with a JSC of 23.78 mA cm−2, a VOC of 1.14 V, and a notable FF of 81.26%. This marks a substantial breakthrough for DJ PSCs (Fig. 5B). The NiOx/SA/TTDMA-Pb devices show excellent reproducibility and yield a certified PCE of 21.42% with negligible hysteresis (figs. S17 to S20). Note that this is the highest certified efficiency reported thus far for quasi-2D PSCs to the best of our knowledge (table S3) (7–9, 14, 16, 19, 26, 29, 30, 45–64). The successively enhancement in VOC from 1.09 V (control) to 1.12 V (CA modified) and 1.14 V (SA modified) correlates with an increase in built-in potential (Vbi) obtained from C-V measurement (Fig. 5C) (65). The superior efficiency of the SA-based device was further validated by the steady-state power output measurement (PCEmax = 21.79%) at the max power point under 0.98-V bias (fig. S19B). In addition, the JSC integrated from the external quantum efficiency (EQE) spectra are 23.16, 23.25, and 23.78 mA cm−2 for the corresponding devices, respectively, matching well with J-V results (fig. S19C). Device photovoltaic performance. (A) J-V curves of optimized TTDMA-Pb devices with or without functional layers. (B) Summary of efficiency (>15%) from previously published results on DJ PSCs (average n < 6) and this study. (C) Mott-Schottky plots of the corresponding devices. (D) MPP tracking of the unencapsulated devices under continuous light illumination at ~50°C in N2. (E) J-V curve of the device based on SA with an area of 1.1 cm2. Light intensity–dependent JSC measurements reveal reduced bimolecular recombination in the SA-based device, evidenced by a near-unity power-law exponent (α = 0.995), compared to CA-based (0.991) and control (0.989) devices (fig. S19A) (66). Similarly, the VOC versus light intensity plot (fig. S21B) shows a reduced slope for the SA-based device (1.47 kT/q) compared to CA-based (1.58 kT/q) control (1.66 kT/q) devices, indicating lower trap-assisted recombination (67). These findings are corroborated by reduced trap density (fig. S21C), prolonged photovoltage decay (fig. S22A), and enhanced hole mobility (fig. S22B). The interfacial instability arising from the redox reaction between I− and Ni3+ presents a notable challenge in achieving the high efficiency and stability of quasi-2D PSCs (68, 69). XPS analysis on both as-deposited and aged NiOx films were performed (fig. S23, A and B). Aged films were prepared by depositing the TTDMA-Pb perovskite on NiOx, aging under light-emitting diode (LED) illumination (100 mW cm−2, T = 40°C) for 5 days and then washing with N,N′-dimethylformamide (DMF) to expose the NiOx surface. The aged NiOx film shows a large drop in Ni3+/Ni2+ ratio from 3.17 to 1.71 for unmodified NiOx, indicating I− induced Ni3+ reduction. In contrast, this ratio remains higher for NiOx/CA (from 3.32 to 2.49) and NiOx/SA (from 3.66 to 3.13), confirming the superior protective effect of the dense SA layer on the overlying perovskite film. UV-Vis absorption of toluene extracts after light soaking further confirms suppressed I− oxidation in CA-modified and SA-modified films (fig. S23). Moreover, x-ray diffraction (XRD) tracking under ambient conditions reveals the emergence of PbI2 and low n-value phases in the control and CA-based film after aging, whereas SA-modified film remains undecomposed even after 360 hours (fig. S24). These results indicate that the dense SA interlayer effectively mitigates the I−/Ni3+ redox reaction, preserving a higher Ni3+/Ni2+ ratio and suppressing I− oxidation, thereby stabilizing the perovskite/NiOx interface and enhancing device durability. The operational stability is substantially improved with interface engineering. Under maximum power point (MPP) tracking (LED, 100 mW cm−2, N2), the unencapsulated bare NiOx-based device dropped to ~32% of its initial PCE after 470 hours. In contrast, devices modified with CA and SA maintained over 80 and 90% of their initial efficiencies after 580 and 990 hours, respectively (Fig. 5D). Note that the 85°C thermal aging results indicate that the SA-modified and CA-modified devices also have substantially improved thermal stability relative to the control (fig. S25). To further evaluate film uniformity, PSCs based on SA with active areas of 1.10 cm2 were also fabricated. As shown in Fig. 5E, the best-performing device exhibited a VOC of 1.17 V, a JSC of 23.46 mA cm−2, and an FF of 75.38%, delivering a high PCE of 20.69%. These results highlight the dual role of the SA interlayer in enhancing both efficiency and operational stability by improving crystallinity, suppressing charge recombination, and preventing interface degradation. Notably, the universality of this molecular interface engineering strategy in directing the vertical crystallization of DJ perovskite films was further demonstrated in the RP perovskite system by using TTMA as the spacer (figs. S26 and S27). This result highlights the broader applicability of the approach across different quasi-2D perovskite frameworks. DISCUSSION We have demonstrated molecular-scale regulation of DJ perovskite films by the incorporation of dual-anchoring organic functional molecules (CA and SA) at the buried interface between NiOx and TTDMA-Pb. Although CA forms a twisted configuration on NiOx, SA assembles into an ordered, vertical aligned interlayer that more efficiently guides vertical crystal orientation, leading to enhanced energy level alignment, mitigated lattice stress, and suppressed Ni3+-induced oxidation of I−. DJ perovskite devices incorporating SA achieves a champion efficiency of 22.03% (certified 21.42%) and improved operational stability. These results highlight the potential of interfacial molecular engineering to control crystal growth in layered perovskites and advance the performance of quasi-2D PSCs. MATERIALS AND METHODS Materials All the reagents and chemicals were used as received without further purification, including PbI2 (99.999%, TCI), poly (3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS, Baytron PVPAI 4083), phenyl-C61-butyricacidmethylester (PCBM; >99.5%, Advanced Election Technology Co. Ltd.), 2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline (BCP, 96%), DMSO (99.8%, Alfa Aesar), isopropanol (IPA; AR, Aladdin Industrial Corporation), chlorobenzene (CB; Sigma-Aldrich), CA and SA (≥99%, Bide Pharmatech Co. Ltd.), DMF (J&K), and nickel oxide (NiOx, 99.999%, Advanced Election Technology Co. Ltd.). Device fabrication The 2D RP perovskite (TTMA)2(MA0.4FA0.6)3Pb4I13 (nominal n = 4) precursor solutions were prepared using a stoichiometric molar ratio of 2:1.2:1.8:4 for TTMAI, MAI, FAI, and PbI2, respectively, dissolved in DMSO/DMF (4:6, v/v), with a Pb2+ concentration of 1.2 M. MACl was added to the precursor solution with a MACl/MAI weight ratio of 1:1. The 2D DJ perovskite (TTDMA)(MA0.4FA0.6)3Pb4I13 (nominal n = 4) precursor solutions were prepared using a stoichiometric molar ratio of 1:1.2:1.8:4 for TTDMAI, MAI, FAI, and PbI2, respectively, dissolved in DMSO/DMF (8:2, v/v), with a Pb2+ concentration of 1.4 M. The MACl was added to the precursor solution with a MACl/MAI weight ratio of 1:1. The optimized concentration of SA or CA solution is 2 mg/ml in IPA. The inverted PSCs were fabricated on ITO-coated glass substrates. The ITO substrates were sequentially cleaned in distilled water, acetone, and IPA, followed by UV-ozone treatment. The NiOx (25 mg/ml in H2O) was spin coated on ITO at 5000 rpm for 30 s and annealed in air at 150°C for 15 min. After cooling, the substrates were transferred into a N2-filled glove box. The SA or CA solution was spin coated on NiOx at 8000 rpm for 30 s and annealed at 100°C for 3 min. After cooling to room temperature, the substrates were rinsed with IPA by spin-coating at 8000 rpm for 30 s. Subsequently, the DJ perovskite precursor solution was spin coated at 7000 rpm for 50 s. During the spin-coating process, 120 μl of the antisolvent (IPA/CB = 1:1, v/v) was dropped onto the spinning substrate at 20 s before the end of the procedure. The resulting TTDMA-Pb films were annealed at 100°C for 35 min. Afterward, the samples were transferred to a glove box and a PC61BM solution (20 mg/ml in CB) was spin coated on the top of the samples at 1200 rpm for 40 s, followed by deposition of a BCP layer (0.6 mg/ml in IPA) at 1000 rpm for 40 s. Last, a 100-nm silver electrode was thermally deposited under a vacuum of 1 × 10−4 Pa. The effective area of the devices were 0.1 cm2 for small-area devices and 1.10 cm2 for large-area devices defined by shadow masks. Film characterization SEM images was obtained using the ZEISS MERLIN Compact at the accelerating voltage of 5 kV. XRD measurements were performed using a Rigaku Ultima IV x-ray powder diffractometer with Cu Kα radiation (λ = 1.54 Å) with a 2θ range of 3° to 50° at a scan rate of 20° min−1. Optical absorption spectra of the perovskite films were obtained using a Cary 5000 UV-visible-NIR spectrophotometer. Steady-state PL and TRPL measurements were performed by an FLS1000 spectrometer (Edinburgh Instruments Ltd.) with an excitation wavelength of 450 nm. XPS and UPS measurements were carried out using the Thermo ESCALAB 250XI. KPFM was carried out with an Asylum Research Cypher atomic force microscope equipped with a silicon probe coated with Ti/Ir (ASYELEC-01-R2), with a force constant of 0.5 to 4.4 nN/nm and a tip radius of 28 nm. The Nap mode was used, consisting of two passes: the first pass acquired surface topography in tapping mode, whereas the second pass lifted the first 40 nm above the surface and acquired the potential offset between the tip and the sample through a dc voltage feedback loop. All AFM measurements were carried out under ambient conditions with a relative humidity of ~20%. In situ PL In situ PL measurements were performed using a custom-built setup. The quasi-2D perovskite films were excited by a wavelength-tunable laser (λ = 450 nm), and the emitted PL was collected using a fiber-coupled spectrometer covering a spectral range of 400 to 1000 nm. Samples were placed in a temperature-controlled (40° and 100°C) chamber, regulated by a temperature control console (TU-300) with an accuracy of ±0.5°C. GIWAXS measurements Angle-dependent GIWAXS measurements were performed using a Xeuss 3.0 instrument (Xenocs, France) equipped with an x-ray source (λ = 1.54189 Å). These measurements were conducted on quasi-2D perovskite films deposited on various substrates during thermal annealing at 100°C (0 to 30 s). The GIWAXS pattern of the postannealed quasi-2D perovskite films were additionally obtained at the 1W1A Diffuse X-ray Scattering Station of Beijing Synchrotron Radiation Facility (BSRF-1W1A) using an x-ray wavelength of 1.54792 Å. The diffraction angles were calibrated using the NIST SRM660b (LaB6) standard. Device characterization The current density–voltage (J-V) curves and steady-state power output were recorded using a Keithley 2400 source-measurement unit under AM 1.5 G illumination (100 mW cm−2) provided by an SS-F5-3A (Enlitech) solar simulator. The light intensity was calibrated using a certified silicon reference diode, and the spectral irradiance of the simulator was verified using a calibrated spectrometer. EQE spectra were collected using a QER Solar Cell Spectral Response Measurement System (Enli Technology Co. Ltd.). J-V curves of representative devices were obtained under both forward (−0.1 V → +1.2 V) and reverse (+1.2 V → −0.1 V) scan directions. TPV and TPC measurements Transient Photovoltage (TPV) and Transient Photocurrent (TPC) measurements were carried out under a white-light bias generated by an array of LEDs (Molex 180081-4320) with an intensity of about 0.5 sun. A diode-pumped laser (Lapa-80) with a pulse duration of 10 ns and a repetition frequency of 20 Hz was used as the perturbation source. The perturbation light intensity was attenuated to maintain the transient VOC below 10 mV. Voltage and current transients were recorded using a digital oscilloscope (Tektronix MDO4104C). Voltages under open-circuit conditions and currents under short-circuit conditions were obtained across 1 MΩ and 50 Ω resistors, respectively. Trap density and mobility measurements The trap-state density (Nt) of DJ perovskite films were performed by the space charge–limited current (SCLC) method. Measurements were performed in a hole-only device configuration of ITO/NiOx/SA (or CA)/perovskite/spiro-OMeTAD/MoO3/Ag. Dark current density–voltage curves were recorded in the voltage range of 0 to 4 V and fitted using an SCLC model.