Suppressing Multi‐Dimensional Defects in Cs_0.05FA_0.95PbI₃ Single Crystals Enables Efficient and Stable Back‐Contacted Perovskite Photovoltaics

ABSTRACT

Back‐contacted architectures offer cost and stability advantages for perovskite solar cells (PSCs), yet their efficiencies have plateaued at ∼12% due to defect‐induced recombination and limited carrier diffusion in thin single crystals. Herein, a multi‐dimensional defect suppression strategy is reported to overcome this bottleneck by incorporating N‐methylformamidinium (MFA⁺) into Cs_0.05FA_0.95PbI₃ (FA = CH(NH₂)₂ ⁺) crystals. MFA⁺ strengthens interaction between A‐site cations with iodide ions, thereby suppressing iodide vacancies (point defects), relieving tensile microstrain, and eliminating dislocations and surface wrinkles (line and plane defects). This approach yields high‐quality crystals with extended electron diffusion lengths (∼400 µm). As a result, an impressive efficiency of 17.35% is obtained, representing a substantial advance over reported back‐contacted PSCs. Moreover, the devices exhibit excellent operational stability with no performance degradation after 1350 h of continuous light illumination. This work highlights the importance of suppressing multi‐dimensional defects for enhancing carrier transport, which is instructive for developing efficient back‐contacted PSCs.

Multi‐scale defects including iodide vacancies, dislocations, and surface wrinkles are suppressed by MFA⁺ incorporation, enabling back‐contacted perovskite solar cells with an efficiency of 17.35% and stable operation over 1350 h.

1. Introduction

Over the past decade, metal halide PSCs have attracted tremendous interest owing to their remarkable power conversion efficiency (PCE) and low manufacturing cost [1, 2, 3]. The certified PCE of single‐junction PSCs has now exceeded 27%, rivaling that of commercial single‐crystal silicon solar cells, which underscores their potential for commercial application [4, 5, 6]. High‐efficiency PSCs predominantly adopt a vertical device architecture, where the perovskite layer is sandwiched between two electrodes [7, 8]. In this configuration, the built‐in electric field is aligned with the direction of halide ion migration, which facilitates field‐driven ionic transport [9, 10, 11]. The halide ion migration, especially at the perovskite/electrode interface, can lead to chemical reactions between metal electrode and halide ions, which will accelerate degradation of PSCs [12, 13, 14]. Mitigation of halide ion migration remains one of the critical challenges for achieving long‐term operational stability of PSCs [15, 16].

By contrast, back‐contacted PSCs, which feature both electrodes on the same side of the perovskite layer, offer advantages for suppressing interfacial ion migration [17, 18]. In this architecture, the built‐in electric field is nearly orthogonal to the ion migration pathway at the interface, substantially reducing the driving force of ion migration [19]. Additionally, the absence of transparent electrodes such as indium tin oxide (ITO) not only lowers material cost but also increases light‐harvesting efficiency. To date, both polycrystalline and single‐crystalline perovskites have been investigated for back‐contacted PSCs [20, 21, 22]. On one hand, the better carrier transport properties of single crystals than polycrystalline films make them more suitable for effective carrier collection when involving large electrode spacing of tens of micrometers. On the other hand, the absence of grain boundaries and high ion migration activation energy (E _a) of single crystals are beneficial for suppression of bulk ion migration, further strengthening potential of back‐contacted architecture for stable PSCs when combining the weak interfacial ion migration [19].

The first demonstration of back‐contacted single‐crystal PSCs by Dong et al. reported a PCE of 1.8% under 0.25 sun illumination [23]. Subsequent advances in crystal thickness, composition, and growth control have incrementally improved the PCE to ∼12% [24], however, which still lags far behind that of vertical‐structure counterparts. This performance gap primarily arises from the mismatch between the relatively short carrier diffusion length (L _D) of thin perovskite single crystals and the long carrier transport distance required in back‐contacted device architecture [25]. Reducing the vertical thickness of perovskite single crystals to the micrometer scale helps confine carrier transport laterally, thereby shortening the effective carrier path length and improving device efficiency distinctly [26]. However, such thin single crystals often suffer from a range of multi‐dimensional crystalline defects—including point, line, and plane defects—that collectively impair carrier transport properties and device performance [7, 27, 28]. Among these, halide vacancies represent the most pervasive point defects. They are commonly observed in both solution‐ and melt‐grown crystals, and remain one of the critical challenges limiting the performance of single‐crystal perovskite optoelectronic devices [29, 30]. In terms of planar defects, surface wrinkles are frequently observed in thin crystals and can severely hinder lateral carrier transport across the device, thereby deteriorating overall photovoltaic performance [31, 32].

In this work, we demonstrate that the incorporation of N‐methylformamidinium (MFA⁺) into Cs_0.05FA_0.95PbI₃ thin single crystals enables effective suppression of multi‐dimensional defects, thereby significantly enhancing the performance of back‐contacted PSCs. The introduction of MFA⁺ strengthens the interaction between A‐site cations and iodide ions, which mitigates the formation of iodide vacancies. Simultaneously, the stronger A‐I interaction compresses the [PbI₆]⁴⁻ octahedra, relieving tensile microstrain that would otherwise give rise to dislocations and surface wrinkles, representative of line and plane defects, respectively. This multi‐dimensional defect suppression leads to a remarkable improvement in crystal quality and extends the electron L _D to 400 µm. As a result, back‐contacted PSCs with an impressive PCE of 17.35% and significantly reduced V _OC deficit are achieved, representing a substantial advance over previously reported back‐contacted PSCs. Furthermore, the suppressed defects raise the E _a from 0.41 to 0.55 eV, which enables superior device operational stability with negligible efficiency degradation after 1350 h of continuous maximum power point (MPP) tracking at 60°C under an inert atmosphere. These findings highlight an effective strategy for defect management in thin perovskite single crystals, which is important for enhancing carrier transport and performance of back‐contacted PSCs.

2. Results and Discussion

2.1. Incorporation of MFA⁺ to Eliminate Iodide Vacancies of Cs_0.05FA_0.95PbI₃ Thin Single Crystals

In this work, the Cs_0.05FA_0.95PbI₃ composition was selected for single‐crystal growth, as it has been extensively studied in both polycrystalline and single‐crystal PSCs with reported PCEs exceeding 25% and excellent operational stability [33]. As shown in Figure S1, Cs_0.05FA_0.95PbI₃ single crystals were grown using the inverse temperature crystallization (ITC) method [34]. To ensure effective carrier collection in back‐contacted PSCs, thin crystals with micrometer‐scale thickness are required, which were obtained using a space‐confined growth strategy (Figure S1).

In general, halide vacancies universally exist on the surface of perovskite single crystals grown by both solution‐ and melt‐based methods, which poses a significant challenge to the performance of diverse single‐crystal perovskite optoelectronic devices, including solar cells, x‐ray detectors, and light‐emitting diodes. As illustrated in Figures 1a and S2, scanning electron microscopy coupled with energy‐dispersive x‐ray spectroscopy (SEM‐EDS) reveals an I/Pb atomic ratio of 2.75 for the as‐grown Cs_0.05FA_0.95PbI₃ crystals. This value deviates from the ideal stoichiometric ratio of 3 and is consistent with earlier reports on lead‐iodide perovskite crystals [35, 36], indicating a substantial density of iodide vacancies on the crystal surface (Figure 1b). These vacancies provide channels for nonradiative recombination and ion migration, both of which limit device performance and long‐term stability.

FIGURE 1.

Enhanced interaction and reduced iodide vacancies in Cs_0.05MFA_XFA_0.95‐XPbI₃. (a) I/Pb ratios at the surface of Cs_0.05FA_0.95PbI₃ and Cs_0.05MFA_XFA_0.95‐XPbI₃ (x = 0.01, 0.015, 0.025) thin single crystals calculated from the SEM‐EDS measurements. Error bars represent the estimated relative quantification uncertainty of SEM‐EDS. (b) A schematic illustration of surface crystal structure of Cs_0.05FA_0.95PbI₃ and Cs_0.05MFA_XFA_0.95‐XPbI₃ (x = 0.01, 0.015, 0.025) to show that the incorporation of MFA⁺ can inhibit the formation of iodine vacancies. (c) Schematic diagram of molecular structure of (left) FA⁺ and (right) MFA⁺. (d) XPS spectra of I 3d in Cs_0.05FA_0.95PbI₃ and Cs_0.05MFA_XFA_0.95‐XPbI₃ (x = 0.01, 0.015, 0.025) thin single crystals. (e) Superoxide yield plot for the Cs_0.05FA_0.95PbI₃ and Cs_0.05MFA_0.015FA_0.935PbI₃ thin single crystals.

The formation of surface iodide vacancies may stem from weak interaction between A‐site cations and iodide anions. To strengthen this interaction, we synthesized N‐methylformamidinium iodide (MFAI) and substituted FAI partially (Figure 1c). The ¹H‐NMR spectrum of MFAI displays a major peak at 2.82 ppm and a minor one at 2.97 ppm, with an integral ratio of approximately 10:1 (Figure S3) [37, 38]. The actual composition of MFA⁺ in the single crystals is measured by 1H‐NMR spectroscopy on the polished single crystals (Figures S4–S6), which also indicates the incorporation of MFA⁺ into the crystal lattice. Incorporation of MFA⁺ has negligible impact on the crystal morphology, optical absorption range, and crystallographic orientation (Figures S7–S9). However, a minor δ‐phase impurity emerges when the MFA⁺ content exceeded 2.5%, thus limiting the upper bound of substitution. Notably, both Cs_0.05MFA_0.015FA_0.935PbI₃ and Cs_0.05FA_0.95PbI₃ crystals exhibited excellent phase stability after 4000 h of storage (Figure S10), confirming the size compatibility of MFA⁺ incorporation.

Compared to FA⁺, the –CH₃ group on MFA⁺ introduces a hyperconjugation effect that increases electron density on the ${\pi }_3^4$ molecular backbone, potentially strengthening N─H···I hydrogen bonds and thereby enhancing A‐site–halide interaction. This enhanced interaction is evidenced by a shift of I 3d XPS peaks toward higher binding energy (Figure 1d), attributed to reduced electron cloud density from stronger hydrogen bonding. Density functional theory (DFT) calculations further support this interpretation, showing that the binding energy between MFA⁺ and I⁻ is approximately twice that of FA⁺ and I⁻ (Figure S11), indicating substantially stronger interaction. Consequently, a reduction in surface iodide vacancy density can be anticipated. Indeed, SEM‐EDS analysis shows that the I/Pb atomic ratio on the crystal surface increases from 2.75 to 2.91, 2.98, and 2.96 upon incorporation of 1%, 1.5%, and 2.5% MFA⁺, respectively (Figures 1a and S12), approaching the ideal stoichiometry gradually. To further validate this trend, surface‐sensitive x‐ray photoelectron spectroscopy (XPS) measurements were performed. The XPS‐derived I/Pb atomic ratio increases from 2.73 for Cs_0.05FA_0.95PbI₃ to 2.91, 2.97, and 2.96 after incorporation of 1%, 1.5%, and 2.5% MFA⁺, respectively (Figure S13), which is consistent with the SEM‐EDS results. This reduction in surface point defects is further validated by a diminished generation rate of superoxide species, as measured using a hydroethylamine fluorescent probe (Figures 1e and S14), corroborating the suppression of surface iodide vacancies [39, 40].

2.2. Elimination of Surface Wrinkles of Cs_0.05FA_0.95PbI₃ Thin Single Crystals

As shown in Figure 2a, pronounced wrinkles are observed on the surface of as‐grown Cs_0.05FA_0.95PbI₃ thin single crystals. SEM images reveal extensive cracking along these wrinkles (Figure S15), which severely impedes lateral carrier transport between electrodes. The photoluminescence (PL) mapping indicates uniform peak positions across the crystal surface, suggesting negligible elemental or phase segregation (Figure S16). Nevertheless, the PL intensity at the wrinkle regions is significantly suppressed (Figure 2b), indicating a high density of nonradiative trap states.

FIGURE 2.

Elimination of surface dislocation and wrinkles in Cs_0.05FA_0.95PbI₃ thin single crystals. (a) Optical microscope image, (b) PL intensity distribution map, and (c) AFM topography image of the wrinkle regions of Cs_0.05FA_0.95PbI₃ thin single crystals. (d) Magnified XRD patterns of the (110) plane of Cs_0.05FA_0.95PbI₃ and Cs_0.05MFA_XFA_0.95‐XPbI₃ (x = 0.01, 0.015, 0.025) thin single crystals. (e) Magnified XRD patterns of the (100) plane of ground powders of Cs_0.05FA_0.95PbI₃ and Cs_0.05MFA_0.015FA_0.935PbI₃ thin single crystals. (f) Optical microscope image, (g) PL intensity distribution map, and (h) AFM topography image of Cs_0.05MFA_0.015FA_0.935PbI₃ thin single crystals.

To elucidate the origin of these wrinkles, atomic force microscopy (AFM) was employed to probe the local microstructure around the wrinkle regions. As shown in Figures 2c and S17, a high density of dislocation lines is observed near the wrinkles, gradually diminishing with increasing distance from the wrinkled regions. According to crystallographic dislocation theory, the accumulation of lattice strain can drive the formation of such dislocation lines (Figure S18) [41]. The subsequent migration and interaction of these dislocations with other line defects or obstacles can result in irreversible plastic deformation and microcracking. In FAPbI₃‐based perovskites, tensile microstrain is widely reported due to the relatively large ionic radius of FA⁺, which expands the [PbI₆]⁴⁻ octahedral framework [42]. In the Cs_0.05FA_0.95PbI₃ system, the low substitution ratio of small‐sized Cs⁺ is insufficient to fully relieve this strain, which may result in the persistent formation of wrinkle‐associated line defects. Thus, we speculate the emergence of surface wrinkles is attributed to the accumulation of unresolved tensile microstrain [43, 44, 45].

Upon incorporation of 1% and 1.5% MFA⁺, the x‐ray diffraction (XRD) peaks of the crystals shift toward higher angles (Figure 2d), suggesting lattice contraction and reduced tensile microstrain. This peak shift is consistently observed in the ground powder form of the same crystals (Figures 2e and S19). On one hand, MFA⁺ interact stronger with iodide ions than FA⁺, which tends to compress the crystal lattice (Figure S20), shifting XRD peak to higher angle. On the other hand, MFA⁺ has larger ionic size than FA⁺, which can cause expansion of crystal lattice, leading to XRD peak shift to lower angle. When low concentration of MFA⁺ is introduced (1% and 1.5%), the XRD peak shift to higher angle, which should be due to the dominating effect of the stronger interaction. To further distinguish these two effects, we prepared Cs‐content‐dependent control crystals without MFA⁺ incorporation. As shown in Figure S21, decreasing the Cs⁺ content causes a monotonic lower‐angle XRD shift, which indicates that lattice expands upon increasing the average A‐site cation size. Therefore, the lattice contraction at low concentration of MFA⁺ should be attributed to the MFA⁺‐specific strengthened A‐site–iodide interaction, rather than to the ionic‐size effect alone. The excessive MFA⁺ cation induces lattice expansion and local structural distortion, which destabilizes the 3D perovskite framework and promotes the formation of δ‐phase impurities (Figures S9 and S19) [42, 46]. Similar phenomenon is also observed in Cs_xFA_1‐xPbI₃ crystals where lattice expansion occurs and δ‐phase forms at high FA⁺ content (Figure S21).

Considering I/Pb ratio still deviates from stoichiometry of 3 at 1% MFA⁺ and the onset of phase instability at 2.5%, we focus on 1.5% MFA⁺ for further investigation. As demonstrated in Figures 2f,h, and S15, the incorporation of 1.5% MFA⁺ effectively eliminates the macroscopic wrinkles, microscopic cracks, and dislocation lines. In addition, PL mapping (Figure 2g) reveals a homogeneous intensity distribution across the entire crystal surface, confirming the mitigation of nonradiative recombination centers.

Collectively, these results indicate that the formation of surface wrinkles in Cs_0.05FA_0.95PbI₃ crystals may originate from accumulated tensile microstrain, which can be effectively relieved through moderate incorporation of MFA⁺, thereby improving both the structural and optoelectronic uniformity of the single crystals.

2.3. Improvement of Crystallinity and Carrier Transport Properties

The incorporation of MFA⁺ suppresses surface iodide vacancies, dislocation lines and wrinkles effectively, which will contribute to improved crystallinity and enhanced charge transport. As shown in Figure S22, the full width at half maximum (FWHM) of the XRD rocking curve for Cs_0.05MFA_0.015FA_0.935PbI₃ single crystals is as narrow as 0.023°, smaller than that of Cs_0.05FA_0.95PbI₃ single crystals (0.031°). Notably, this FWHM value is comparable to that of high‐quality bulk single crystals (Table S1), confirming the superior crystallinity achieved through MFA⁺ incorporation. In parallel, the trap density of the crystals was evaluated using the space‐charge limited current (SCLC) method. As shown in Figure S23, the hole trap density decreases from 4.57 × 10¹² cm⁻³ to 3.2 × 10¹¹ cm⁻³ after MFA⁺ incorporation. The value of the trap density and its change after improvement are comparable to those of many reported perovskite thin single crystals [24, 47, 48, 49]. This enhanced crystallinity and reduced trap density will facilitate suppression of nonradiative recombination, which is evidenced by stronger PL emission and extended time‐resolved PL (TRPL) lifetimes (from 123 to 658 ns) (Figure S24).

In back‐contacted PSCs, photogenerated carriers must laterally traverse distances on the order of tens of micrometers before reaching the electrodes. Since the built‐in electric field in such devices is not fully aligned with the direction of lateral carrier transport (Figure S25), a large L _D is essential for efficient charge collection. Based on prior studies, L _D can be extracted by fitting the relationship between short‐circuit current density (J _SC) and crystal thickness using the following equation [50, 51]:

$$J_{\mathrm{SC}}=q\underset{\lambda }{\overset{\infty }{\int }}\left[1-R\left(\lambda \right)\right]\mathrm{\Phi }\left(\lambda \right)\mathrm{d}\lambda$$ (1)

$$\frac{J_{\mathrm{SC}}}{J_{\mathrm{SC},0}}=\frac{1}{\cos \frac{d}{L_{\mathrm{D}}}}$$ (2)

where q is elementary charge, R is the part of incident photons (the percentage lost when photons reach perovskite), λ is wavelength of incident light, Φ _sun is the spectrum of sunlight photons, J _SC,0 is the short‐circuit current density of solar cells with the smallest crystal thickness, J _SC is the short‐circuit current density of solar cells with different crystal thickness, d is the thickness of single crystals, and L _D is the carrier diffusion length. In previous and this works, the J _SC,0 is defined as the J _SC of solar cells based on 20‐µm‐thick single crystals.

The device structure used for extracting L _D follows a typical vertical configuration: ITO/MeO‐2PACz/perovskite single crystals/fullerene (C₆₀)/bathocuproine (BCP)/copper (Cu) (Figure 3a), consistent with previous reports. Current density–voltage (J–V) curves of devices with varying crystal thickness (Figure S26) based on the control Cs_0.05FA_0.95PbI₃ and target Cs_0.05MFA_0.015FA_0.935PbI₃ thin single crystals under AM 1.5G illumination are presented in Figure 3b,c. As expected, J _SC decreases with increasing thickness, a trend also confirmed by external quantum efficiency (EQE) measurements (Figures S27 and S28). Notably, the J _SC values of Cs_0.05MFA_0.015FA_0.935PbI₃ devices are consistently higher than those of Cs_0.05FA_0.95PbI₃ at large thicknesses, indicating longer diffusion lengths (Figure S29). By fitting the relationship between J _SC/J _SC,0 and crystal thickness for the control and target devices (Figure 3d,e), an electron diffusion length of ∼400 µm is obtained for Cs_0.05MFA_0.015FA_0.935PbI₃ crystals, which is approximately two times longer than that of Cs_0.05FA_0.95PbI₃ (∼200 µm) [33]. This enhancement is consistent with the suppression of iodide vacancies and elimination of surface wrinkles afforded by MFA⁺ incorporation, confirming its positive role in improving carrier transport properties.

FIGURE 3.

Carrier transport properties of Cs_0.05FA_0.95PbI₃ and Cs_0.05MFA_0.015FA_0.935PbI₃ thin single crystals. (a) Device structure of the vertical single‐crystal solar cells. J–V curves of (b) Cs_0.05FA_0.95PbI₃ and (c) Cs_0.05MFA_0.015FA_0.935PbI₃ devices with different crystal thicknesses under AM 1.5 light illumination. Calculated average electron L _D by fitting J _SC/J _SC,0 and crystal thicknesses of (d) Cs_0.05FA_0.95PbI₃ and (e) Cs_0.05MFA_0.015FA_0.935PbI₃ devices.

2.4. Impact of MFA⁺ Incorporation on Device Performance of Back‐Contacted PSCs

To evaluate the effect of MFA⁺ incorporation on device performance, back‐contacted perovskite solar cells (PSCs) were fabricated with the structure Au/MoO₃/single crystal/C₆₀/BCP/Cu and electrode spacing of 50 µm (Figures 4a and S30). J–V curves under AM 1.5 G illumination are presented in Figure 4b,c for devices based on Cs_0.05FA_0.95PbI₃ and Cs_0.05MFA_0.015FA_0.935PbI₃ thin single crystals. A metal photomask was used during measurement to confine the effective working area (Figure S30). The control Cs_0.05FA_0.95PbI₃ devices exhibit a PCE of 11.03%, with a J _SC of 25.92 mA cm^{− 2}, an open‐circuit voltage (V _OC) of 0.706 V, and a fill factor (FF) of 60.3%. This efficiency is comparable to previously reported back‐contacted PSCs (Table 1). In comparison, the Cs_0.05MFA_0.015FA_0.935PbI₃‐based devices achieve a significantly improved PCE of 17.35%, accompanied by a J _SC of 26.62 mA cm^{− 2}, a V _OC of 0.913 V, and an FF of 71.4%, establishing a new benchmark efficiency for back‐contacted PSCs. Statistical data from five devices (Figure S31) confirm the reproducibility of this enhancement. The increase of J _SC and FF can be attributed to the enhanced carrier transport properties (Figure 3d,e), which is further verified by fitting the relationship between J _SC and light intensity. As shown in Figure S32, the control devices exhibit a low α value of 0.87, reflecting obvious recombination loss, while the α value of Cs_0.05MFA_0.015FA_0.935PbI₃ devices is 0.98, close to the ideal value of 1, indicating minimal bimolecular or trap‐assisted recombination. It is worth noting that the PCE of the Cs_0.05MFA_0.015FA_0.935PbI₃ devices reaches 18.29% at 0.75 sun (Figure 4d), suggesting suppressed bimolecular recombination under lower illumination, which is advantageous for real‐world photovoltaic applications with fluctuating sunlight. The significantly improved PCE is mainly due to the dramatic increase of V _OC, indicating distinctly suppressed nonradiative recombination, which is consistent with the suppression of multi‐dimensional defects. To further validate this inference, light‐intensity dependent V _OC measurements were conducted. As shown in Figures 4e,f and S33, the ideality factor of the single‐crystal devices decreases from 2.85 to 1.23 after MFA⁺ incorporation, reflecting the reduced nonradiative recombination and improved diode behavior in the back‐contacted devices. The high ideality factor of the control devices is comparable to that of reported MAPbI₃ (MA = methylammonium) back‐contacted single‐crystal PSCs [24].

FIGURE 4.

Photovoltaic performance of back‐contacted single‐crystal solar cells. (a) Device structure of the back‐contacted single‐crystal solar cells. J–V curves of the champion (b) Cs_0.05FA_0.95PbI₃ and (c) Cs_0.05MFA_0.015FA_0.935PbI₃ devices. (d) J–V curves of back‐contacted Cs_0.05MFA_0.015FA_0.935PbI₃ single‐crystal solar cells under different light intensity. Light intensity‐dependent V _OC of the solar cells based on (e) Cs_0.05FA_0.95PbI₃ and (f) Cs_0.05MFA_0.015FA_0.935PbI₃ thin single crystals.

TABLE 1.

Summary of device parameters of representative back‐contacted PSCs.

Materials	Electrode spacing (µm)	PCE (%)	Illumination intensity	References
MAPbI3 single crystal	50	1.88	0.25sun	[23]
MAPbI3 polycrystal	2	3.52	1sun	[20]
MAPbI3 single crystal	33	4.83	1sun	[21]
MAPbI3 single crystal	50	5.9	0.25sun	[10]
MAPbI3 single crystal	1.5	7.4	1sun	[22]
MAPbI3 single crystal	50	11.5	1sun	[24]
FA0.75MA0.25PbI3 single crystal	50	12.6	1sun	[19]
Cs0.05MFA0.015FA0.935PbI3 single crystal	50	17.35	1sun	This work

2.5. Impact of MFA⁺ Incorporation on ion Migration and Device Stability

In perovskite single crystals, halide ion vacancies provide pathways that facilitate ion migration and device degradation. The incorporation of MFA⁺ contributes to reduction of iodide ion vacancies, which is expected to suppress ion migration and retard device degradation. To validate this analysis, temperature‐dependent conductivity measurements in lateral devices using Au/single crystals/Au configuration were conducted to monitor the change of E _a which was calculated using the Nernst–Einstein equation:

$$\sigma \left(T\right)=\left({\sigma }_0/T\right)\exp \left(-E_{\mathrm{a}}/\mathrm{KT}\right)$$ (3)

where K is the Boltzmann constant, σ ₀ is a constant, and T is the temperature. As shown in Figure 5a,b. The calculated E _a of Cs_0.05FA_0.95PbI₃ thin single crystals is increased from 0.41 to 0.55 eV after the incorporation of MFA⁺, indicating a higher ion migration barrier. To further confirm the suppression of ion migration, both back‐contacted devices were subjected to a high electrical field (400 V mm⁻¹) to accelerate ion migration for 24 h, followed by surface SEM‐EDS mapping analysis. As shown in Figure S34, the halide distribution is uniform in both devices before applying the bias. However, obvious iodine loss and iodine accumulation are observed on the Cu and Au electrode of the Cs_0.05FA_0.95PbI₃ devices, while the Cs_0.05MFA_0.015FA_0.935PbI₃ devices exhibit negligible iodine change after aging (Figure 5c,d), verifying suppressed iodide ion migration laterally. The EDS analysis of the peeled Cu/BCP/C₆₀ layers reveals iodine contents of 1.17%, 11.61%, and 0.98%, 1.09% for fresh, aged Cs_0.05FA_0.95PbI₃ and fresh, aged Cs_0.05MFA_0.015FA_0.935PbI₃ devices, respectively (Figures S35 and S36), indicating suppressed interface iodide ion migration from perovskite to Cu electrode. The suppressed ion migration should lead to improvement in operational stability of the back‐contacted PSCs, which was investigated by monitoring steady‐state power output under voltage at MPP (V _mpp) under one‐sun illumination in inert environment at about 60°C without device encapsulation. As shown in Figure 5e, the control Cs_0.05FA_0.95PbI₃ devices can retain only 90% of its initial efficiency after 700 h, while the Cs_0.05MFA_0.015FA_0.935PbI₃‐based devices show negligible efficiency degradation after 1350 h of continuous illumination. The slight increase in PCE during the initial MPP tracking stage can be attributed to the slow release of trapped solvent from the single‐crystal thin films [52, 53].

FIGURE 5.

Suppressed ion migration and enhanced stability of the back‐contacted PSCs. Temperature‐dependent electrical conductivity of the (a) Cs_0.05FA_0.95PbI₃ and (b) Cs_0.05MFA_0.015FA_0.935PbI₃ single‐crystal devices. Top‐view SEM‐EDS mapping images of (c) Cs_0.05FA_0.95PbI₃ and (d) Cs_0.05MFA_0.015FA_0.935PbI₃ single‐crystal devices after aging for 24 h at 400 V mm⁻¹ under light illumination. (e) Operational stability of the two back‐contacted single‐crystal solar cells.

3. Conclusions

In summary, we report incorporation of MFA⁺ into Cs_0.05FA_0.95PbI₃ single crystals to suppress multi‐dimensional defects, including iodide vacancies, microscopic line dislocations and macroscopic surface wrinkles. The as‐grown Cs_0.05MFA_0.015FA_0.935PbI₃ single crystals exhibit improved crystallinity, reduced nonradiative recombination, and extended electron L _D of ∼400 µm. As a result, the back‐contacted PSCs based on Cs_0.05MFA_0.015FA_0.935PbI₃ single crystals exhibit low V _OC loss, resulting in a significantly enhanced PCE of 17.35%, setting a new efficiency benchmark for back‐contacted PSCs. Moreover, the iodide ion migration is inhibited distinctly, leading to superior device operational stability with negligible efficiency degradation after 1350 h of continuous illumination. Since the thin perovskite single crystals are also promising for single‐photon x‐ray detectors, self‐powered photodetectors, field‐effect transistors and light emitting diodes [47, 54, 55, 56], our work is instructive for developing high‐performance and stable single‐crystal perovskite optoelectronic devices.

Author Contributions

Delong Han: conceptualization, data curation, software, investigation, validation, formal analysis, supervision, writing – original draft, and methodology. Hailong Liu: formal analysis and validation. Dalin Li: formal analysis and data curation. Mingxuan Lv: software and data curation. Nianqiao Liu: formal analysis and investigation. Xiao Cheng: conceptualization, writing – review and editing, resources, and project administration. Zhaolai Chen: writing – review and editing, writing – original draft, funding acquisition, visualization, project administration, resources, and conceptualization. William W. Yu: funding acquisition, writing – review and editing, resources, and project administration.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie73474‐sup‐0001‐SuppMat.docx.

Data Availability Statement

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