Reorientation‐Driven Degradation in Oriented Perovskite Films: Shifting Facet Engineering to Thermodynamic Stability

ABSTRACT

Hybrid perovskite solar cells (PSCs) suffer from underexplored links between crystallographic orientation and thermal stability, especially in narrow‐bandgap devices. We fabricate highly oriented mixed Sn‐Pb perovskite films via an additive‐free two‐step method. Accelerated aging studies at 120°C reveal that high orientation paradoxically compromises stability, and PSCs built from highly oriented perovskite films retain only 73% of their initial power conversion efficiency (PCE), compared to 89% PCE in less‐oriented devices. Operando grazing‐incidence wide‐angle X‐ray scattering of the PSCs shows that thermal stress induces significant reorientation and lattice distortion in the oriented crystallites, accumulating pronounced microstrain that accelerates the PSC degradation. Structural analyses confirm progressive crystallographic transitions, including grain reconfiguration, shifts toward isotropy, and systematic diffraction migrations. Critically, we demonstrate that metastability is an intrinsic consequence of high crystallographic order, which is why the very high alignment strategies that enhance performance induce thermodynamic vulnerability. This necessitates redesigning crystal engineering priorities where suppressing instability requires engineering thermodynamic equilibrium states over maximizing alignment for stable perovskite photovoltaics.

The role of crystallographic orientation in the thermal stability of narrow‐bandgap perovskite solar cells remains largely underexplored. In Sn–Pb mixed perovskite films, strong orientation can inherently compromise stability under prolonged thermal stress, triggering reorientation‐induced degradation through microstrain accumulation. These findings underscore the importance of prioritizing thermodynamic‐equilibrium‐based engineering over maximizing crystal alignment alone to achieve stable photovoltaic performance.

1. Introduction

Hybrid organic–inorganic halide perovskite solar cells (PSCs) are considered as the most promising next‐generation light‐harvesting technology due to their remarkable advantages over silicon photovoltaics, such as longer charge carrier lifetimes (>1 µs), higher power conversion efficiencies (PCEs), lower fabrication costs, and tunable direct bandgaps (1.2–2.3 eV) [1, 2, 3]. Over the past decade, with advancements in interface engineering, solvent engineering, surface passivation, and additive engineering, single‐junction PSCs have achieved a significant breakthrough in their PCEs, increasing from 3.8% to 27.3% [4, 5, 6, 7]. Sn‐Pb alloyed perovskite photovoltaics have garnered significant research interest in recent years, owing to their tunable bandgap (1.2–1.4 eV) that approaches the Shockley–Queisser (SQ) optimal range for single‐junction solar cells [8, 9]. When strategically integrated into tandem architectures through bandgap complementarity, these narrow‐bandgap absorbers enable theoretical PCEs surpassing 40%, transcending the fundamental efficiency limits of single‐junction counterparts [10, 11]. Recent certification milestones demonstrate the rapid progress: Sn‐Pb perovskite subcells now achieve 24.13% PCE [12], while all‐perovskite tandem configurations reach 30.1% [7], positioning this material class as a transformative pathway to push perovskite photovoltaic efficiency boundaries toward commercial viability. Moreover, this exciting improvement in PCE has been closely associated with an enhanced understanding of the microstructure of perovskite materials [13, 14, 15, 16]. According to previous studies, the facet orientation is directly linked to the surface chemical, physical, and atomic potential landscapes because different crystal planes have distinct spacings, atomic coordinations, and arrangements [17, 18, 19]. Therefore, understanding the role of the perovskite crystal orientation in determining the photovoltaic performance is of critical importance.

Crystallographic orientation critically governs surface electronic states through facet‐dependent variations in atomic coordination and defect formation energies. According to the literature, the (001) orientation displayed the higher charge carrier mobility and the lowest trap density, while the (111) orientation showed the lowest charge carrier mobility and the highest trap density [15]. Similarly, the (110) and (100) orientations exhibit different charge carrier mobilities under an electric field due to A‐site cation migration, with the (110) facet orientation showing a lower mobility [20]. Additionally, the defects in the (111) crystal orientation reduce the open‐circuit voltage (V_OC ) and increase the hysteresis in JV measurements of PSCs, implying that growing perovskite with a specific crystal orientation can reduce the defect density and enhance the PSC performance [21]. Furthermore, the crystal orientation of the perovskite plays an important role in determining the interfacial properties between the perovskite layer and the charge transport layers [13]. An optimal crystal orientation can optimize the energy level alignment at the interface, thereby curtailing charge carrier recombination and boosting the fill factor (FF) [22]. The facet‐dependent optoelectronic properties, especially the PCE, of perovskite solar cells have been widely studied. When targeting real‐world applications, besides the photovoltaic performance, the device stability is also of essential importance, however, the role of the perovskite crystal orientation in determining the device stability is far less understood.

Growth of a preferentially oriented perovskite film can be achieved through the introduction of additives, antisolvents, and 2D‐seed‐induced growth [23, 24]. For example, Zheng et al. significantly reduced the efficiency gap in perovskite solar cells by incorporating trace amounts of surface‐anchoring alkylamine ligands as grains and interface modifiers. This approach prepared the highly oriented face‐on perovskite films, improved the film properties, and achieved a certified stabilized efficiency of 22.3% [25]. However, it is difficult to exclude the interference of the introduced additives in the resulting device photovoltaic performance. Therefore, in this work, no additives are used to exclude potential additive‐induced modulation of the crystal orientation. Instead, we fabricate preferentially oriented MA₀.₀₄FA₀.₉₆Sn₀.₃Pb₀.₇I₃ (MA = methylammonium; FA = formamidinium) narrow bandgap (1.33 eV) perovskite films through a controlled two‐step interdiffusion annealing without any additives. The two‐step approach eliminates the need for using an antisolvent and provides a simpler, more robust, and highly reproducible processing route. In contrast, one‐step deposition methods typically require large amounts of antisolvent to produce high‐quality perovskite films. The crystal facet orientation distribution in polycrystalline perovskite films is investigated with grazing‐incidence wide‐angle X‐ray scattering (GIWAXS), with a particular focus on orientation angle and the probed orientation material quantity (MQ). Operando GIWAXS measurements further reveal that highly oriented PSCs are more fragile and experience degradation at 120°C in comparison to less oriented PSCs, indicating that a lower degree of orientation is preferable for improving the stability of the solar cells. In addition, continuous current density‐voltage (J–V) measurements are performed at high temperature conditions, which consistently corroborates these findings. Thereby our work elucidates the high‐temperature degradation mechanisms in PSCs with different crystallographic orientations, providing critical new perspectives on crystallographic orientation engineering for the thermal stability of these devices.

2. Results and Discussion

2.1. Characterization of Mixed Sn‐Pb Perovskite Films

Figure 1a schematically illustrates the two‐step fabrication protocol for mixed Sn‐Pb perovskite films with a controlled crystalline orientation. For fabricating less oriented perovskite films, the Sn_0.3Pb_0.7I₂ thin film is prepared by spin coating without additional annealing. Afterward, MAI/FAI is further spin‐coated on top of Sn_0.3Pb_0.7I₂ thin film, followed by annealing at 150°C for 15 min (details in Experimental Section). In contrast, for the highly oriented films, the spin‐coated Sn‐Pb iodide film is further annealed at 70°C for 1 min, while the rest of the fabrication process remains the same.

FIGURE 1.

(a) Schematic diagram of the fabrication processes for perovskite films with less (top row) and high (bottom row) orientation. Reshaped 2D GIWAXS data of (b) Sn_0.3Pb_0.7I₂ film without annealing, (c) Sn_0.3Pb_0.7I₂ film with annealing, (d) perovskite prepared from Sn_0.3Pb_0.7I₂ film without annealing, and (e) perovskite prepared from Sn_0.3Pb_0.7I₂ film with annealing. Azimuthal tube cut data retrieved from cuts along (001) crystal plane of (f) perovskite prepared from Sn_0.3Pb_0.7I₂ film without annealing, and (g) perovskite prepared from Sn_0.3Pb_0.7I₂ film with annealing, along with the corresponding fits.

To elucidate the crystallographic evolution of as‐prepared Sn‐Pb iodide and perovskite films, GIWAXS measurements are performed with an incident angle of 0.6° (Figure 1b–g). The reshaped 2D GIWAXS data of non‐annealed and annealed Sn_0.3Pb_0.7I₂ films (Figure 1b,c) reveal that an additional annealing process promotes the preferential crystallographic orientation of the perovskite. Specifically, the (001) diffraction peak at q ≈ 0.9 Å⁻¹ demonstrates a pronounced out‐of‐plane alignment after annealing. Furthermore, the (101) diffraction feature at q ≈ 1.8 Å⁻¹ exhibits characteristic orientations at azimuthal angles of χ = ± 54° after thermal annealing [26]. Additionally, a diffraction ring at q ≈ 0.7 Å⁻¹, corresponding to the Sn_0.3Pb_0.7I₂·DMSO complex, maintains an out‐of‐plane orientation despite having a reduced intensity compared to the non‐annealed sample, consistent with a partial complex decomposition at the 70°C annealing temperature [27, 28]. Moreover, these results are also supported by XRD data of the Sn_0.3Pb_0.7I₂ layer with and without annealing, as shown in Figure S1a. Figure 1d,e display the reshaped 2D GIWAXS data of mixed Sn‐Pb perovskite films fabricated with non‐annealed and annealed Sn‐Pb iodide thin films, respectively. To quantitatively analyze the crystallographic orientation, azimuthal tube cuts are applied to the (001) diffraction plane of the less and highly oriented perovskite films achieved via absence or presence of the initial annealing step. The corresponding azimuthal profiles with Gaussian fits are presented in Figure 1f,g. The MA_0.04FA_0.96Sn_0.3Pb_0.7I₃ perovskite film, prepared from non‐annealed Sn‐Pb iodide, exhibits a complete isotropy in the (001) plane orientation (orange area, Figure 1f, hereafter referred to as less oriented perovskite). In contrast, the perovskite film fabricated with annealed Sn‐Pb iodide demonstrates a preferential orientation, with two intensity maxima at χ = ± 54° (pink area, Figure 1g, hereafter referred to as highly oriented perovskite). This angular distribution indicates a 54° tilt of the (001) crystal plane with respect to the substrate, characteristic of a corner‐on orientation configuration of the perovskite grains [29, 30]. The observed orientation correlation between Sn‐Pb iodide thin films and final perovskite films suggests template‐directed crystallization. In this process, during the cation exchange (FA⁺/MA⁺ replacing DMSO), the perovskite lattice inherits its orientation preferences from the oriented Sn‐Pb iodide matrix being initially prepared [31, 32]. This templating effect rationalizes both the enhanced (111) peak intensity in highly oriented perovskite (Figure S1) and the preferential orientation in final films.

The surface morphology and microstructural characteristics of less and highly oriented Sn‐Pb perovskite films are investigated through top‐view scanning electron microscopy (SEM) and atomic force microscopy (AFM), as presented in Figure S2. The SEM analysis reveals that less oriented films exhibit a higher density of pinholes and voids relative to the highly oriented films, suggesting that thermal annealing effectively suppresses the surface defect formation. The highly oriented perovskite thin film demonstrates a reduced root‐mean‐square (RMS) roughness (13.7 nm) compared to the less oriented samples (16.9 nm). Statistical analysis of 50 randomly selected grains from the AFM images, processed through Gaussian fits, indicates comparable average grain sizes of ∼0.9 µm in both types of perovskite films. Cross‐sectional SEM images (Figure S3) confirm a film thickness of ∼600 nm and a continuous, homogeneous film morphology without interfacial stacking in both types of films, implying minimal defect‐induced charge carrier recombination pathways.

UV–vis absorption spectroscopy is performed to analyze the optical properties of the perovskite layers with low and high orientation. As illustrated in Figure S4, both highly oriented and less oriented perovskite films exhibit comparable spectral absorption profiles across the measured wavelength range. Notably, the highly oriented architecture demonstrates a slight increase in absorption in the visible range compared to its less oriented counterpart. This is because the highly oriented perovskite films exhibit fewer defects (Figure S2), resulting in a reduced non‐radiative recombination. Additionally, the highly oriented films have a smoother surface morphology and slightly thinner thickness (Figures S2 and S3), which decreases light scattering within the film and increases the effective optical path length, contributing to the observed enhancement in absorption. As shown in Figure S4b, the optical bandgap of perovskite layers remains stable at approximately 1.33 eV regardless of the perovskite orientations. Notably, this bandgap value (1.33 eV) closely matches the ideal bandgap with the potential to achieve the maximum theoretical PCE for single‐junction PSCs, according to the Shockley‐Queisser limit [33].

2.2. Photovoltaic Characteristics of PSCs

Inspired by the enhanced film properties observed in highly oriented perovskite layers, we further explore their impact on the device performance. Planar single‐junction Sn‐Pb PSCs are fabricated using an inverted p‐i‐n architecture consisting of indium tin oxide (ITO) / poly(3,4‐ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) / perovskite / [6,6]‐phenyl C₆₁ butyric acid methyl ester (PC₆₁BM) / bathocuproine (BCP) / Ag, as depicted in Figure 2a. The target group, comprising PSCs incorporating highly oriented perovskite films, exhibits an enhanced photovoltaic performance. As presented in Figure 2b,c, the highly oriented perovskite films led to an improvement in the photovoltaic parameters. The V_OC increases slightly from 0.850 to 0.856 V, the short‐circuit current density (J_SC ) increases slightly from 30.47 to 30.75 mA cm⁻², and also the FF increases slightly from 79.04 to 79.85% when measured under a reverse voltage scan, as shown in the insert table in Figure 2b,c. Accordingly, the champion PCE of highly oriented Sn‐Pb solar cells is 21.02%, while the PCE of PSCs with less orientation is 20.47%. Since the differences are not large, we further present the statistical distributions of PCE, J_SC , V_{OC ,} and FF for both less and highly oriented perovskite solar cells in Figure 2d–g, respectively. The average PCE increases from 18.61 ± 0.60% to 19.00 ± 0.92%, which is ascribed to the collective enhancement of V_OC (from 0.838 ± 0.003 V to 0.850 ± 0.030 V), J_SC (from 29.47 ± 0.30 mA cm⁻² to 30.20 ± 0.24 mA cm⁻²), and FF (from 75.72 ± 0.93% to 77.31 ± 0.83%). The enhancement of PCE can be ascribed to the reduced nonradiative recombination, which can be corroborated by electrochemical impedance spectroscopy (EIS) and dark J–V curves. Figure S5 shows the Nyquist plots of the EIS data collected for both types of perovskite solar cells. The Nyquist plots show a main semicircle at high frequency (10²–10⁶ Hz), which can be attributed primarily to the recombination resistance (R_rec ) and capacitance [34, 35]. The estimated R_rec values of the PSCs with low and high orientation are 732 and 878 Ω cm², respectively. Because the recombination rate is inversely proportional to R_rec , the devices with high orientation have lower recombination rates. On the basis of these results, high orientation likely reduces the trap‐state density of the perovskite and the recombination rate of the perovskite films and devices. This finding is also supported by the dark current measurements, as shown in Figure S6a. Devices built from highly oriented perovskite films have lower dark currents at negative and positive bias voltages, suggesting an improved charge transport and decreased recombination loss in the devices [36]. As shown in Figure S6b, both of these two PSCs show a similar EQE trend. However, the less oriented sample shows lower EQE values from 400 to 900 nm compared with the highly oriented one (80%–90%), illustrating that the photogenerated charge carrier collection efficiency of low‐orientation thin films is relatively limited, which agrees with our other photovoltaic results.

FIGURE 2.

(a) Device layout of the PSCs. J–V curves with forward and reverse scans of the champion device (b) fabricated from less oriented perovskite films, and (c) from highly oriented perovskite films. Box plot of the distribution of (d) PCE, (e) J_SC , (f) V_OC , and (g) FF values of the PSCs fabricated from less oriented (blue) and highly (pink) oriented perovskite films (20 devices for each type).

2.3. Stability of PSCs

We examine the temporal device performance of mixed Sn‐Pb PSCs prepared from the two different orientations of the perovskite films under operation conditions at 120°C. The high temperature mimics an accelerated aging. The non‐encapsulated PSCs fabricated from less oriented perovskite films show a better device stability compared with those fabricated from highly oriented perovskite films, as shown in Figure 3a,b (normalized from five devices). All photovoltaic parameters of highly oriented perovskite‐based cells decrease dramatically compared with the devices with less oriented perovskite devices, especially the PCE, showing a reduction of 27% after 250 min of operation already. The loss in PCE is driven mainly by the decrease in the FF (12%). Alternatively, for the less oriented perovskite‐based devices, the V_OC remains at 97% of its initial value, while the J_SC and FF values show a decrease of 6% and 5%, respectively, resulting in a PCE decrease of 11%. This is because there is no obvious phase separation and phase transition during the thermal degradation, which is keeping the V_OC stable. Thus, the less oriented perovskite active layer displays a better operation stability at high temperatures. Additionally, we also investigate the operational stability of non‐encapsulated devices under ambient conditions. As shown in Figure S7, the PCE of the less oriented device retained 83% of its initial value after approximately 96 h of aging, whereas the highly oriented device decreases already to 75% of its initial PCE. In addition, we examine the surface morphology and XRD data of the perovskite films after being stored in ambient air at room temperature for 96 h (Figure S8). Notably, highly oriented perovskite film exhibited a greater presence of degradation products, indicating its inferior stability.

FIGURE 3.

Temporal evolution of PCE, V_OC , J_SC , and FF measured by the reverse scan from −0.9 to 0.2 V for (a) less oriented perovskite film, (b) highly oriented perovskite‐ based PSCs in air and at 120°C. The shaded area refers to the error bars derived from the standard deviation of normalized PCE, V_OC , J_SC , and FF values of five devices. Evolution of (c) normalized FWHM of less (blue) and highly (pink) oriented perovskite. (d) Normalized microstrain evolution during high‐temperature aging of less (blue) and highly (pink) oriented perovskite film in PSCs during operation extracted from Williamson–Hall analysis. Evolution of enlarged pseudo‐XRD data (q region from 0.7 to 1.8 Å⁻¹) of (e) less oriented perovskite film and (f) highly oriented perovskite film with the region of the PbI₂ (001) peak highlighted with a box.

To elucidate the structural origin of the observed phenomena, operando GIWAXS measurements are performed on PSCs containing mixed Sn‐Pb perovskite films with both types of orientation under thermal stress (120°C for 120 min). Representative 2D GIWAXS data are shown in Figure S9 and reveal coexisting diffraction signatures from both perovskite crystallites and residual PbI₂ phases. For quantitative phase evolution analysis, azimuthally integrated radial profiles (pseudo‐XRD patterns) are extracted and calibrated against the ITO substrate reference peaks, as shown in Figure S10 [37]. To perform a quantitative analysis of the GIWAXS data, we apply a Gaussian function to fit the (001) Bragg peak located at q ≈ 1.0 Å^{− 1} (Figure 3c) and evaluate the changes of its full‐width‐at‐half‐maximum (FWHM). In case of the highly oriented perovskite film, the FWHM steadily increases over time, reaching 1.4 times its initial value after 120 min, which is higher than in the case of the less oriented perovskite film. In contrast, the FWHM of the less oriented perovskite film rapidly increases to 1.1 times its initial value within the first 20 min at 120°C, followed by a slower growth, ultimately reaching 1.25 times of the initial value after 120 min. The increase in the FWHM of the (001) Bragg peak indicates that the crystallographic properties of the perovskite film degrade under the applied continuous high‐temperature operation conditions of the PSCs. We further explore the evolution of the microstrain through the Williamson‐Hall analysis, as shown in Figures 3d and S11 [38]. Notably, the highly oriented perovskite film develops a server microstrain at high temperature operation of the PSCs. A gradual accumulation phase during the initial 40 min is followed by an accelerated strain development over the subsequent 80 min. This kinetic profile strongly correlates with the temporal degradation patterns of the photovoltaic parameters measured in the operando experiment. The strain modulates interfacial energetics between crystallite domains in the textured perovskite structures, inducing FF and Jsc deterioration through charge carrier recombination losses [39, 40]. The comparative strain analysis reveals distinct strain responses between both types of perovskite films in the solar cells. The highly oriented perovskite film develops compressive strains, reaching 3.5 times of the initial values after 120 min stress exposure, whereas the less oriented control films maintain a nearly constant strain level. This strain‐mediated degradation mechanism highlights the critical role of the crystallographic alignment in the device stability. For further analysis, the magnified pseudo‐XRD spectra are presented in Figure 3e,f to explicitly illustrate the structural evolution in both types of perovskite films. No peak position shift or splitting of the (001) Bragg peaks is observed across all samples, suggesting the absence of detectable phase segregation in the perovskites [41]. By comparing the evolution of the pseudo‐XRD patterns over time for the perovskite films with different orientations, it is evident that in the highly oriented sample, a distinct PbI₂ (001) peak emerges at q ≈ 0.9 Å⁻¹ after 80 min of high‐temperature heating. Moreover, this peak intensifies over operation time, whereas no such peak is observed in the less oriented perovskite based PSCs. This finding indicates that the highly oriented perovskite films are less stable under high‐temperature operation conditions of the related PSCs.

2.4. Degradation Mechanisms of PSCs

To elucidate the crystallographic orientation evolution in the highly oriented perovskite films in PSCs during operation, azimuthal tube cuts are performed for both (001) and (111) crystallographic planes. The resultant orientation distributions are resolved through Gaussian deconvolution of azimuthal intensity profiles. Representative azimuthal integration patterns with corresponding fits for the (001) plane are presented in Figure 4a. Critical to this quantitative analysis is the implementation of the Lorentz factor correction (sinχ normalization) for the azimuthal intensity calibration [42, 43, 44, 45, 46]. The analysis of the (001) plane before thermal degradation (Figure 4a, 1) reveals that an orientation of approximately ± 54° relative to the substrate dominates the crystallographic configuration, corresponding to a corner‐on orientation within the pseudo‐cubic lattice framework (Figure S12). This preferential alignment inherently orients the (111) facet parallel to the substrate. Extended thermal exposure triggers a progressive structural reorganization. After 40 min of ageing (Figure 4a, 2), a subset of grains undergoes reorientation, as indicated by the emergent out‐of‐plane alignment of the (001) plane characteristic of the face‐on configuration. Quantitative evaluation of material quality (MQ) fraction changes (MQ/MQ _tot ) documents the evolution of the orientation populations, as shown in Figure 4b. The corner‐on fraction decreases from 81% to 75%, while the face‐on component rises from 0% to 3%, along with a slight increase in the number of isotropic domains. Thus, a divergent MQ evolution emerges under the applied thermal stress. It means that the corner‐on crystals exhibit a progressive MQ degradation proportional to the ageing duration, whereas the face‐on grains demonstrate a commensurate MQ enhancement. A detailed Gaussian analysis of χ ≈ ± 54° azimuthal profiles identifies two interdependent degradation signatures. A continuous broadening of the corner‐on peak coincides with a progressive angular shift toward in‐plane orientations, ultimately reaching χ ≈ ±65° after 120 mins of thermal treatment, as shown in Figure 4c. The observed crystal reorientation from ±54° to ± 65° suggests a gradual lattice distortion together with a phase reconfiguration (Figure 4d). Parallel investigations of the (111) plane kinetics (Figures S14–S17) confirm that this thermally driven reorientation process represents a universal degradation pathway rather than an isolated (001) plane behavior. As such, we infer that the metastable crystals in the highly oriented perovskite thin films experience a reconfiguration under thermal stress conditions, as driven by the microstrain suggesting orientation‐dependent degradation pathways.

FIGURE 4.

(a) Azimuthal integration patterns for the (001) crystal plane of the highly oriented perovskite film in PSCs during operation under 120°C at different time scales. (b) Time evolution of material quantity fraction for the (001) crystal plane of different orientations. (c) Azimuthal tube cuts as a function of polar angle for the (001) crystal plane of the highly oriented perovskite film at different time scales. (d) Schematic diagram of the lattice reorientation and distortion.

3. Conclusion

In summary, this work presents a simple method for controlling crystal orientation in mixed Pb‐Sn perovskites by utilizing pre‐structured Sn‐Pb iodide layers to guide crystallographic alignment during cation‐exchange processing. Photovoltaic devices fabricated from highly oriented perovskite films achieve an improved initial device performance, with champion PCE values reaching 21.02%, compared to less oriented counterparts. However, the devices built from highly oriented perovskite films exhibit an accelerated degradation under thermal (120°C) stress, retaining only 73% of their initial PCE, compared to 89% in the case of the less oriented perovskite film‐based devices under identical ageing conditions. The highly oriented perovskite films demonstrate a thermally activated lattice distortion under prolonged operation at high temperatures, with microstrain increase reaching 3.5 times those observed in the less oriented counterparts. Azimuthal integration analysis reveals progressive crystallographic reorientation during degradation as the reason behind this device degradation, characterized by concurrent structural transitions: approximately 6% of corner‐on grains undergo reconfiguration to face‐on alignment, while isotropic grain populations increase from 75% to 81% over 120 min of thermal stress. This reorientation process is accompanied by a systematic angular migration of diffraction peaks from ± 54° to ± 65°, indicative of cumulative lattice strain accumulation. In this study, we propose a new perspective regarding the impact of crystallographic engineering on the thermal stability of PSCs. The results demonstrate that high crystallographic orientation does not necessarily confer superior thermal stability of the PSCs. Paradoxically, highly oriented perovskite films exist in a metastable state, undergoing reorientation and orientation transition processes at elevated temperatures. The resulting more pronounced lattice strain consequently leads to degradation in the stability of the PSCs. These findings reveal that metastability is an intrinsic consequence of high crystallographic order, providing a critical redesign principle: suppressing orientation‐induced instability requires engineering thermodynamic equilibrium states rather than maximizing alignment of perovskite crystallites. This redefines crystal engineering priorities for stable perovskite photovoltaics. In addition, we have also noticed that crystallographic orientation is correlated with the moisture resistance of perovskite solar cells, and we plan to investigate this relationship further in our future work [18, 47].

4. Experimental Section

4.1. Materials

Lead iodide (PbI₂, 99.999%), tin iodide (SnI₂, 99.99%), formamidinium iodide (FAI, 98%), methylammonium iodide (MAI, 98%), methylammonium chloride (MACl), [6,6]‐phenyl C₆₁ butyric acid methyl ester (PC₆₁BM, 99.5%), bathocuproine (BCP, 99%), N,N‐anhydrous dimethylformamide (DMF, 99.8%), dimethyl sulfoxide (DMSO, 99.9%), isopropanol (IPA, 99.9%), chlorobenzene (CB, 99.8%) were purchased from SigmaAldrich. Poly(3,4‐ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS, BaytronPVPAI4083) was purchased from Ossila. All materials were used directly without further purification.

4.2. Device Preparation

The ITO glass substrates were washed with the sequence of diluted Hellmanex III (2:98), DI water, acetone, 2‐propanol, and ethanol using an ultrasonic cleaner for 15 min each step. Then, the substrates were dried with a nitrogen gas flow. Before further processing, the ITO substrates were treated with O₂ plasma cleaning for 15 min. The PEDOT:PSS layer was prepared by spin‐coating atop ITO substrates at 4000 rpm for 30 s and then annealing at 120°C for 15 min. The perovskite layer was prepared by a two‐step spin‐coating method in N₂ glove box. First, 1.05 m PbI₂ and 0.45 m SnI₂ in DMF:DMSO (4:1) solution was deposited on the substrate at 2000 rpm for 30 s, (then annealed at 70°C for 1 min for the highly oriented perovskite films); second, the FAI:MACl:MAI (90:9:6 mg in 1 mL IPA) solution was spin‐coated at 2200 rpm for 30 s, followed with annealing at 150°C for 15 min in ambient air (30%‐40% humidity). The PC₆₁BM solution (20 mg mL⁻¹ in CB) was dynamically spin‐coated on ITO/PEDOT:PSS/perovskite layers at 2500 rpm for 30 s. Afterward, the BCP (0.5 mg mL⁻¹ in IPA) solution was dynamically deposited at 5000 rpm for 30 s. Finally, a 100 nm silver electrode was evaporated with a mask. The effective area of each pixel was about 0.2 cm².

4.3. GIWAXS Measurement

Grazing‐incidence wide‐angle X‐ray scattering (GIWAXS) measurements were carried out at the P03/MiNaXS beamline of the PETRA III storage ring at DESY. A LAMBDA 9 m detector collected the GIWAXS data. The photon energy of the X‐ray beam was 11.79 keV, corresponding to a wavelength of 1.044 Å. The sample‐to‐sample distance (SDD) was set as 209.5 mm. An incidence angle (α_i) of 0.6 degrees was chosen for the operando GIWAXS measurements to enable a scattering signal from the ITO substrate, which was used for calibration. Image processing was done with the Python tool of INSIGHT, including transformation to q‐space, geometric calculations, intensity corrections for path attenuation, detector absorption, photon polarization, and solid angle [48]. Due to thermal expansion under light illumination, the SDD must be corrected based on the ITO substrate signal at q = 2.139 Å⁻¹ [49]. Radial cake cuts for pseudo‐XRD and azimuthal tube cuts for orientation analysis were carried out with INSIGHT. Radial cake cuts for pseudo‐XRD were done with χ = (−90, 90) and q = (0, 2.5) Å⁻¹ as cut limits. Azimuthal tube cuts for texture analysis were done with χ = (−90, 90).

4.4. Characterization

Cross‐sectional and top‐view scanning electron microscopy (SEM) imaging was performed using Zeiss Gemini NVision 40 and ZEISS EVO systems, with cross‐sectional samples analyzed at an accelerating voltage of 5 keV and a 6 mm working distance, while surface morphology characterization utilized a 30 keV beam energy and a 9 mm working distance. Atomic force microscopy (AFM) measurements were conducted using a Nanosurf FlexAFM system under ambient conditions. Optical absorption spectra were acquired with a PerkinElmer Lambda 35 UV–vis spectrophotometer at a 480 nm/min scan rate. Current density‐voltage (J–V) characteristics were measured under AM 1.5G illumination (100 mW/cm²) using a Newport Class ABA solar simulator with a Keithley 2611B source meter, calibrated by a certified Si reference cell (Fraunhofer ISE019‐2015). All devices were evaluated through a 0.079 cm² aperture mask using a 50 mV/s scan rate (10 mV voltage steps, 100 ms dwell time).

Author Contributions

X.C and X.J. conceived the idea, designed the research, performed the experiments, analyzed the data, and wrote the manuscript. K.S., G.P., Z.L., L.L., T.B., and A.B.‐A. performed the experiments. K.S., G.P., and Z.L. analyzed the data and discussed the results. S.K.V. and P.M.‐B. provided resources. P.M.‐B. provided funding, project administration, and supervised the research. All authors discussed the results, provided critical feedback, contributed and approved the final version of the manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smll72339‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available from the following public repository: https://doi.org/10.14459/2026mp1840060.