Adsorption Energy Difference‐Driven Synthesis of Perovskite Single Crystals With Tailored Exposed Facets

ABSTRACT

The precise synthesis of perovskite single crystals with tailored exposed facets is crucial for enhancing the performance of optoelectronic materials. However, the controlled preparation of perovskite single‐crystal with different exposed facets is challenging, because perovskite tends to grow thermodynamically stable crystal facets, hindering systematic investigations of facet‐dependent optoelectronic properties. Here, we first realized the synthesis of facet‐specific perovskite single crystals via a coordination competition strategy mediated by the crystal‐habit modifier, achieving morphological manipulation from cubic (100) to dodecahedral (110) facet‐dominant structures. The formation mechanism of non‐thermodynamic (110) facets originates from strong coordination between the crystal‐habit modifier and (110) facets, which selectively inhibits (110) facet growth and finally produces a (110) facet‐dominated single crystal. Optoelectronic characterization reveals distinct facet‐dependent properties, with (110) facets exhibiting higher defect density compared to (100) facets. Performance evaluations demonstrate that lower defect densities in (100) facets enhance photodetector performance, while the elevated defect densities in (110) facets improve photocatalytic performance. This study establishes a new paradigm for tailoring perovskite crystal facets and provides fundamental insights for designing high‐performance optoelectronic materials.

Precise synthesis of perovskite single crystals with tailored exposed facets is first realized based on the adsorption energy difference of crystal‐habit modifiers by atomic‐scale design, yielding cubic and dodecahedral morphology for (100) and (110) crystal facets. (100) facet‐based photodetectors exhibit higher responsivity due to reduced defect‐mediated recombination, while (110) facet‐based photocatalysts show improved photocatalytic performance owing to defect‐enhanced surface reactivity.

1. Introduction

The orientation of the crystal facets directly determines the arrangement of atoms, the distribution of electronic states, and the chemical activity of material surfaces, thereby governing their optical, electrical, catalytic, and mechanical behaviors [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]. Tailoring material properties by exposing specific crystal facets has emerged as a universal paradigm for atomic‐scale design of functional materials [11, 12, 13]. For instance, catalysts with different exposed facets exhibit significantly different catalytic and sensing performance due to the distinct active sites [14, 15, 16, 17, 18]. Nevertheless, precise control of the orientation of crystal facets undergoes complex competing effects between thermodynamics and kinetics, leading to the formation of non‐ideal crystal facets [2, 19, 20, 21]. Controlled growth of specific crystal facets with superior properties remains challenging.

Metal halide perovskites have garnered significant attention due to their low‐cost solution processing and exceptional optoelectronic properties, such as high absorption coefficients, tunable band gaps, and high carrier mobility [22, 23, 24, 25, 26, 27, 28]. Additive‐mediated interactions and interfacial chemistry have further enhanced the optoelectronic performance of perovskite‐based materials [29, 30, 31]. Particularly, single‐crystal perovskite possesses more obvious advantages, such as less defect density and longer carrier diffusion length, which greatly accelerate the development of optoelectronic devices, such as photodetectors, lasers, and solar cells [32, 33, 34, 35]. Deep defects typically impair device performance, whereas carefully designed and controlled shallow defects can enhance device performance [36, 37, 38]. To fabricate perovskite crystals with different exposed crystal facets, various crystal facet control strategies have been developed, including vapor‐phase synthesis, top‐down cutting, thermodynamic and kinetic control of crystallization, and so forth [39, 40, 41, 42, 43, 44]. Although the above methods can be used to control perovskite crystal facets, they still face some challenges [39, 40, 41, 42, 43, 44]. Vapor‐phase synthesis and top‐down cutting processing methods typically require strict experimental conditions, complex fabrication processes, high costs, and poor universality. Thermodynamic and kinetic methods for exposing specific crystal facets suffer from either limited selectivity or stringent experimental conditions, and poor universality. Meanwhile, the exposed crystal facets of perovskite single crystal devices are usually thermodynamically supported crystal facets, and it is impossible to realize the controllable growth of non‐thermodynamically supported crystal facets, which makes it difficult to systematically study the optical and electrical properties of perovskite with different crystal facets [3, 17, 45]. Importantly, Sebastian Polarz and his colleagues successfully achieved precise control of crystal facet orientation in perovskite microcrystals using an aerosol‐assisted synthesis technique, guiding research on the facet‐dependent structure‐property [46, 47, 48]. If the large‐scale, high‐throughput, and low‐cost perovskite single crystals with different crystal facets can be manipulated, it will further promote the research of facet‐dependent properties and the development of perovskite optoelectronic devices.

In this work, we first realized the precise synthesis of perovskite single crystals with tailored exposed crystal facets via competitive coordination mediated by the crystal‐habit modifier. Compared to previously reported methods, we not only established the relationship between crystal facet and properties but also demonstrated the universality of our method based on the coordination‐competition strategy, ultimately expanding the applications of perovskite‐based materials. The perovskite single crystals with (100) and (110) crystal facets present a cubic and dodecahedral morphology, respectively. Combined in situ crystallization monitoring and theoretical calculations elucidate the facet‐selective growth mechanism, in which the crystal‐habit modifier shows stronger adsorption affinity to (110) facets compared to (100) facets, thereby selectively inhibiting (110) facet growth. Comprehensive characterization, including steady‐state/transient absorption (TA) spectroscopy, photoluminescence, ultraviolet photoelectron spectroscopy (UPS), and space charge limited current (SCLC) measurements, confirms facet‐dependent optoelectronic properties. Quantitative analysis reveals a higher defect density on (110) facets ((7.34 ± 0.97) × 10¹¹ cm⁻³) versus (100) facets ((3.27 ± 0.76) × 10¹¹ cm⁻³). Optoelectronic performance evaluations demonstrate that (100) facet‐based photodetectors achieve higher responsivity due to reduced defect‐mediated recombination, while (110) facet‐based photocatalysts exhibit enhanced photocatalytic performance owing to defect‐enhanced surface reactivity. This innovative work provides a strategy and opportunity for the precise synthesis of perovskite single crystals with tailored crystal facets, pioneering a new insight for the fabrication of various high‐performance optoelectronic devices.

2. Results and Discussion

2.1. The Importance of Crystal Facet Manipulation

Semiconductor single crystals with different exposed facets exhibit distinct optoelectronic properties due to facet‐dependent atomic arrangements. To systematically investigate these facet‐dependent optoelectronic properties, precise synthesis of crystals with a controlled exposed facet is essential. Perovskites, as a class of ionic semiconductors, demonstrate particularly pronounced facet‐dependent variations in their anionic and cationic configurations (Figure 1a,b). Current research predominantly focuses on thermodynamically stable (100) facets (e.g., in FAPbI₃ and MAPbBr₃) due to their lowest surface energy, while the (110) crystal facets with the higher surface energy are difficult to fabricate, limiting comprehensive facet‐property studies [3, 49]. Addressing this limitation, we developed a coordination competition strategy based on classical nucleation theory, employing a crystal‐habit modifier, 1‐butyl‐3‐methylimidazolium bis(trifluoromethylsulfonyl)imide (BMIMTFSI), to control facet development thermodynamically. Our results demonstrate that TFSI anions exhibit strong coordination to (110) facets compared to (100) facets, effectively suppressing growth of the (110) facets. This differential coordination enables morphological evolution from (100)‐dominant cubes to (110)‐dominant dodecahedra (Figure 1c). Figure 1d,e, and Figures S1 and S2 show the perovskite single crystal morphology fabricated without the modifier and with the modifier, demonstrating the successful crystal facet modulation. The successful morphology manipulation of the perovskite single crystal is further verified by scanning electron microscopy (SEM) (Figures S3 and S4). In contrast to cubic single crystals without a modifier, perovskite bulk single crystals with dodecahedral morphology can also be fabricated by introducing a modifier (Figure S5). Furthermore, the crystal morphology of MAPbBr₃ perovskite single crystals with different exposed facets is derived from Bravais‐Friedel‐Donnay‐Harker (BFDH) calculations, which present consistent morphology with the experimental results, demonstrating the feasibility of the modifier (Figure 1f,g). X‐ray diffraction (XRD) analysis reveals a systematic transition from pure (100) to (110) crystallographic orientation with increasing modifier concentration (Figure 1h,i), demonstrating the feasibility of the coordination competition strategy. By precisely controlling the exposed facets from pure (100) to (110) crystallographic orientation, the optoelectronic properties of perovskite materials were tailored, realizing a conversion from high‐performance optoelectronic devices to high efficiency photocatalysts (Figure 1j).

FIGURE 1.

Crystal facets manipulation of perovskite single crystals. (a) The schematic diagram of perovskite crystals with different exposed crystal facets. (b) Schematic diagram of the (100) crystal facet and the (110) crystal facet, presenting different anionic and cationic compositions. (c) Schematic diagram of crystal facet manipulation, realizing the morphology manipulation of perovskite single crystals simultaneously. Morphology and photoluminescence photographs of MAPbBr₃ perovskite single‐crystal fabricated (d) without crystal‐habit modifier, and (e) with crystal‐habit modifier, showing a noticeable difference in morphology. The simulated crystal morphology of MAPbBr₃ perovskite single crystals with (f) (100) exposed facet and (g) (110) exposed facet based on the Bravais‐Friedel‐Donnay‐Harker (BFDH) method. (h) The x‐ray diffraction (XRD) peaks of perovskite single crystals with increased crystal‐habit modifier. (i) The percentage of (100) and (110) diffraction peaks extracted from the (h) figure without the crystal‐habit modifier and with the crystal‐habit modifier. (j) The significant importance of crystal facets manipulation, expanding its applications from high‐performance optoelectronic devices to highly active photocatalysts.

To elucidate the respective roles of anions and cations within the modifier for crystal facet modulation, we synthesized perovskite single crystals by incorporating cation‐dominant 1‐butyl‐3‐methylimidazolium bromide ([BMIM]Br) and anion‐dominant trifluoromethanesulfonimide (HTFSI) additives. Perovskite crystals with [BMIM]Br additives exhibit a cubic morphology, and no (110) diffraction peaks appear with increased amounts of additives, suggesting that cations are not responsible for the modulation of crystal facets (Figures S6 and S7). In contrast, perovskite crystals with HTFSI additives showed a dodecahedral morphology, and the (110) diffraction peaks became dominant with the increased additives, suggesting that anions are responsible for the modulation of crystal facets (Figures S8 and S9). In BMIMTFSI, the cation [BMIM]⁺ serves solely to balance the charge. X‐ray photoelectron spectroscopy (XPS) and the energy‐dispersive x‐ray spectroscopy (EDS) mappings of F and S elements further confirm the successful introduction of the crystal‐habit modifier (Figures S3, S4, and S10). For perovskite single crystals without a modifier, no binding energy peaks and EDS mappings of the F and S elements appear, whereas notable binding energy peaks and EDS mappings of the F and S elements appear after the introduction of a modifier, indicating the successful introduction of the modifier during crystal growth.

2.2. Mechanism of Crystal Facet Manipulation

To explore the impact of the TFSI anion on the non‐thermodynamically supported (110) facet, density functional theory (DFT) calculations were conducted. The computed surface energy (E _s) of (100), (110), and (111) facets of the MAPbBr₃ crystal are E _s(100) = 0.00343 eV·Å⁻², E _s(110) = 0.00869 eV·Å⁻², and E _s(111) = 0.01261 eV·Å⁻², respectively (Table S1 and Note S1). Among these, the (100) facet has the lowest surface energy. According to the periodic bond chain

(PBC) theory [50], MAPbBr₃ adopts a cubic crystal structure wherein Pb²⁺ and Br⁻ form an octahedral framework, with MA⁺ (CH₃NH₃ ⁺) cations occupying the interstitial sites for charge balance.‌ During the crystal growth process, Br⁻ forms single, double, and triple bonds with Pb²⁺ on the F‐facet {100}, S‐facet {110}, and K‐facet {111}, respectively (Figure 2a). This difference in bonding configurations leads to an increase in dangling bond density across these facets, correspondingly raising their surface energies. Facets with high surface energy tend to be eliminated during growth, whereas those with low surface energy are thermodynamically stable and become predominantly exposed. As a consequence, the natural crystal growth of MAPbBr₃ typically yields a cubic morphology.

FIGURE 2.

Mechanism of crystal facet manipulation with the crystal‐habit modifier. (a) Schematic diagram of periodic bond chain theory for MAPbBr₃ crystal. (b) The electrostatic potential (left) and stick structure of the TFSI anion with measured distances between oxygen atoms (right). O (red), S (yellow), N (blue), C (grey), F (cyan). (c) The average distance between bromine atoms in (100), (110), and (111) crystal facets. (d) Schematic representation of adsorption geometries of TFHI⁻ on (100), (110), and (111) crystal facets. (e) Calculated charge density distribution and adsorption energy of TFHI⁻ on (100), (110), and (111) crystal facets. Blue iso‐surface represents electron depletion, and the yellow stands for electron accumulation. (f) In situ observation, (g) schematic illustration, and (h) the XRD diffraction peaks of the crystal growth process‌‌ of MAPbBr₃ with the crystal‐habit modifier. (i) Infrared spectra of pure TFSI anion and the MAPbBr₃ perovskite with coordination addition. (j) The ¹H nuclear magnetic resonance spectroscopy (¹H NMR) of pure HTFSI and HTFSI mixed with MAPbBr₃ perovskite. The inset shows the structural formula of HTFSI and the MA⁺ cation. (k) X‐ray photoelectron spectroscopy (XPS) for Pb 4f core‐level spectra of the perovskite single crystal without the modifier and with the modifier.

The introduction of TFSI anion alters the crystal growth process, suppressing the traditional cubic morphology that exposes the (100) facet. The crystals evolve a dodecahedral morphology with exposure of the (110) facet. This crystal morphology evolution stems from the anisotropic modulation of surface energy by the TFSI anion, because the lone pair electrons of the oxygen atom in the TFSI anion have a strong coordination interaction with Pb²⁺. The spatial configuration and electrostatic potential of the TFSI anion were analyzed using DFT. The electrostatic potential indicates that the electron‐donating ability is concentrated at oxygen atoms. The distances of distinct oxygen atoms are 3.58 and 4.38 Å (Figure 2b). The average distances between neighboring bromine ions in the (100), (110), and (111) crystal facets of MAPbBr₃ are 5.92, 4.20, and 4.20 Å, respectively (Figure 2c, Figure S11, and Tables S2 and S3). The distance between oxygen atoms aligns with that between bromine atoms, allowing TFSI anion bidentate anchoring on the (100), (110), and (111) crystal facets (Figure 2d). This strong bidentate anchoring coordination interaction exhibits anisotropic modulation capability for the surface energy of distinct crystal facets. As illustrated in Figure 2e, the charge density distribution of the TFSI anion on the (100), (110), and (111) crystal facets shows a significant difference. The (110) facet displays a more notable change, suggesting enhanced charge transfer between the TFSI anion and the (110) facets. When TFSI anions were added, the surface energy after adsorption (E _s‐a) of each facet can be expressed as: E _s‐a = E _s + n _a·E _ads, in which E _s is the surface energy of each crystal facet, n _a is the adsorption density that positively correlated with the additive concentration, and E _ads refers to the adsorption energy of TFSI anions on each crystal facet [51, 52]. The E _ads values of TFSI anions on the (100), (110), and (111) facets were calculated as −5.05, −5.78, and −5.44 eV, respectively. The (110) facet exhibits the most negative adsorption energy, thus the lowest E _s‐a can be achieved at an optimal additive concentration.

The crystal growth process with a coordination additive was further revealed by in situ microscopy and XRD measurement. In situ microscopy observations show that the (111) facet disappeared first, followed by (100), ultimately exposing the (110) facet (Figure 2f,g). Meanwhile, the XRD pattern demonstrates that the addition of TFSI anion facilitates the exposure of (110) facet, which is consistent with the in situ microscopy observation (Figure 2h). The strong coordination between the TFSI anion and Pb²⁺ was also experimentally validated by Fourier‐transform infrared spectroscopy (FTIR), liquid‐state ¹H nuclear magnetic resonance (¹H NMR), and XPS. The characteristic vibrational peak of the S = O bond shifts from 1352 cm⁻¹ to a higher wavenumber of 1356 cm⁻¹ owing to the electron transfer from TFSI anion to Pb²⁺, indicating the formation of coordination bonds (S = O→Pb) (Figure 2i). For NMR measurements, HTFSI was employed as an additive. ¹H NMR spectra of pure HTFSI and HTFSI mixed with MAPbBr₃ perovskites show an obvious shift and a new peak appearance (Figure 2j), which can be attributed to three factors: (1) Pb²⁺ coordinates to the oxygen atoms of HTFSI, preventing HTFSI forming hydrogen bonds; (2) MA⁺ undergoes rapid proton exchange with HTFSI, which averages the proton ¹H NMR signal of HTFSI and leads to an increase in the chemical shift compare to PbBr₂:HTFSI mixture (Figure S12 ); (3) the protons of the free MA⁺ exhibit another distinct ¹H NMR signal [53, 54]. XPS spectra for Pb 4f core‐level spectra of the MAPbBr₃ single crystal with crystal‐habit modifier shifted towards lower binding energy due to the strong interaction between the S = O group of the additive and Pb²⁺. Peak deconvolution confirms the coexistence of two binding energy states, indicating that coordinated (binding energy shifted) and uncoordinated (binding energy unshifted) Pb²⁺ species are present together in the (110) oriented MAPbBr₃ crystals (Figures 2k, S13, and S14) [55]. The combined theoretical calculations and experimental results demonstrate that the addition of TFSI anions can anisotropically modulate surface energy through bidentate anchoring, promoting the exposure of non‐thermodynamically supported (110) facets that would otherwise be buried‌. The realization of facet modulation is the result of thermodynamics/dynamics regulation. Thermodynamically, the strong adsorption of the modifier on the (110) facet enhances the stability of this facet. Dynamically, it inhibits the binding of bromide and lead on the crystal surface through steric hindrance, thereby reducing the growth rate of the facet. Based on the above mechanism, the crystal‐habit modifier BMIMTFSI enabled facet modulation in FAPbBr_3, owing to the matched Br‐Br distance on its (110) facet (Figures S15 and S16). Moreover, bis(fluorosulfonyl)amide (HFSI) was identified as another effective modifier that induces (110)‐oriented dodecahedral MAPbBr₃ crystals (Figure S17), underscoring the universality of this competitive coordination strategy.

2.3. Comparison of Optoelectronic Properties for Different Exposed Crystal Facets

Upon the precise synthesis of crystal facets, we conducted systematic investigations into the facet‐dependent optoelectronic properties of perovskite materials. Different perovskite crystal facets are composed of different ratios of ions, which in turn give rise to distinctive optoelectronic properties [17, 56]. Figure 3a shows the Pb atomic densities of different perovskite crystal facets, where a higher Pb atomic density appears on the (100) crystal facet compared to the (110) and (111) crystal facets, in agreement with previous literature reports [57].

FIGURE 3.

Comparison of optoelectronic properties of perovskite single crystals with different crystal facets. (a) The Pb atomic densities of perovskite with (100) crystal facet, (110) crystal facet, and (111) crystal facet. The inset schematic shows the corresponding crystal structure of the different crystal facets. (b) The absorption spectra of the perovskite single crystals with the (100) crystal facet and the (110) crystal facet. (c) Time‐resolved photoluminescence spectroscopy of perovskite single crystals with the (100) crystal facet and the (110) crystal facet, revealing a longer lifetime for the (100) crystal facet. The transient absorption spectrum of the perovskite single crystals with (d) the (100) crystal facet, and (e) the (110) crystal facet. (f) Kinetics traces of the (100) crystal facet and the (110) crystal facet with a pump fluence of 5 µJ cm⁻². Ultraviolet photoelectron spectroscopy (UPS) results of (g) the (110) crystal facet, and (h) the (100) crystal facet. (i) The morphology and the corresponding energy level alignment of perovskite with the (100) crystal facet and the (110) crystal facet.

To further investigate the optical properties of different perovskite crystal facets, we measured steady‐state absorption (photoluminescence) and TA (photoluminescence). Figure 3b presents the absorption spectrum of the perovskite single crystal with the (100) and (110) crystal facets. Compared to the (100) crystal facet with stronger absorption, the absorption of the (110) crystal facet is slightly weaker, which can be attributed to the different atomic arrangement significantly affecting the density of surface states and the distribution of defects, which in turn affects the intensity of the absorption [42, 58]. Furthermore, we investigated the band gap of perovskite single crystals dominated by different crystal facets. Figure S18 shows the Tauc plot of the perovskite single crystal with the (110) and (111) crystal facets. In contrast to the band gap of 2.30 eV for the (100) crystal facet, the band gap of the (110) crystal facet shows a slight difference, increasing to 2.31 eV, which can be attributed to the crystal spacing difference [42, 59]. Meanwhile, compared to the (100) crystal facet, the weak photoluminescence, low photoluminescence quantum yield, and fast lifetime decay of the (110) crystal facet further indicate more defects within the crystal (Figures 3c, S19, and S20). The 4 nm blue shift of the photoluminescence peak demonstrates an increasing band gap for the (110) crystal facets, in agreement with the Tauc plot results. Figure 3d,e shows the TA spectra of perovskite single crystals with the (100) crystal facet and the (110) crystal facet. The TA measurements for the two facet‐oriented crystals were conducted under identical excitation conditions, ensuring that the observed differences arise from the samples rather than measurement conditions. The TA signals in perovskites generally contain contributions from multiple photoinduced processes, including hot‐carrier cooling, carrier trapping, and band‐edge recombination [60, 61]. For the (100) and (110) facet‐oriented crystals, the bleaching peaks are both located at 539 nm with no significant difference, which is consistent with the steady‐state absorption results. Figure S21 shows the time‐resolved TA spectra of perovskite single crystals. Besides the bleaching peaks centered at 539 nm, additional transient spectral features are observed in the adjacent non‐bleach regions. These additional signals at 440–510 nm and 550–580 nm may originate from photoinduced absorption (PIA) [60]. To quantify the facet‐dependent carrier dynamics, the kinetic traces at 539 nm were further fitted using a bi‐exponential function, and the extracted time constants together with their amplitudes were used to calculate an amplitude‐weighted average lifetime (Figures 3f and S22, Table S4). Compared to perovskite single crystals with the (100) crystal facet, the (110) crystal facet exhibits a shorter τ ₁ and a higher amplitude ratio a ₁, indicating stronger trap‐assisted recombination. Besides, as the excitation fluence increases, the slow component in the (100) facet become increasingly prominent, consistent with the enhanced involvement of bimolecular recombination after trap filling, further supporting the conclusion that the (100) facet possesses fewer defects than the (110) facet, which is consistent with the lower trap density derived from the SCLC measurements (Figures 3f and S23, Table S5) [62].

To further investigate the energy band structure differences of perovskite single crystals with different crystal facets, UPS was used to probe the change of band structure (Figures 3g,h and S24). Figure 3g,h shows the cut‐off binding energy (E _cut) and the surface Fermi edge (E _edge) for perovskite with the (100) crystal facet and the (110) crystal facet. The valence band (E _VB) of the perovskite single crystal can be defined as: E _VB = (E _cut—E _edge) − 21.22 eV, where 21.22 eV is the energy of the helium light source. The calculated values of the E _VB for the (100) and (110) crystal facets are −6.26 and −6.2 eV, respectively. Further combining the band gap measured by the Tauc plot result, the calculated values of the conduction bands (E _CB) for the (100) crystal facet and the (110) crystal facet are −3.96 and −3.89 eV, respectively. Furthermore, the crystal‐habit modifier on the surface of the perovskite crystal does not affect its band structure. (Figures S25–S28 and Note S2). The differences in band structure are primarily attributed to the influence of the crystal facets. Figure 3i shows the corresponding energy band diagrams of perovskite with the (100) and (110) crystal facets. Overall, the crystal facet modulation, which in turn affects the energy band structure demonstrated by Tauc plot, PL, and UPS characterization, plays a crucial role in the performance modulation of perovskite optoelectronic devices [17, 56].

2.4. Comparison of Optoelectronic Performance

Lastly, we systematically evaluate the performance of perovskite optoelectronics based on (100) and (110) crystal facets. During the optical experiments above, we have demonstrated that perovskite single crystals with (110) crystal facets possess higher defect densities. Defects in metal halide perovskites typically consist of a large number of shallow‐level and deep‐level defects [63, 64]. For more accurate characterization of the defect density and carrier mobility of different crystal facets, we performed SCLC tests. As shown in Figure 4a,b, the trap‐filled limit voltage (V _TFL) of 0.29 and 0.65 V was obtained from the I‐V curve for (100) and (110) crystal facets, respectively. The trap density n _traps of perovskite single crystal with different crystal facets was defined as: n _traps = 2 εε ₀ V _TFL/(qL ²), where ε, ε ₀, q, and L are the dielectric constant of the material (25.5 for MAPbBr₃), the vacuum permittivity, electronic charge, and the channel length (50 µm in our experiment), respectively [65]. Meanwhile, the carrier mobility can be extracted from the Child regime of the I‐V curve by the Mott‐Gurney law, J _D = 9εε ₀ µV _b ²/(8L ³), where J _D, μ, and V _b are the current density, the carrier mobility of the perovskite single crystal, and Child voltage, respectively [66]. To systematically evaluate the n _traps and μ of perovskite single crystal with different crystal facets, the statistical n _traps and μ are measured for (100) and (110) crystal facets based on five different devices (Figure 4c). Compared to the n _traps of (3.27 ± 0.76) × 10¹¹ cm⁻³ and μ of 72.3 ± 2.70 cm² v⁻¹ s⁻¹ for the (100) crystal facet, higher n _traps of (7.34 ± 0.97) × 10¹¹ cm⁻³ and lower μ of 48.5 ± 1.46 cm² v⁻¹ s⁻¹ were demonstrated for the (110) crystal facet, further revealing that the properties of perovskite optoelectronic devices vary significantly among different exposed crystal facets. The crystal‐habit modifier has a negligible effect on the carrier mobility and trap density of the (110)‐oriented perovskite single crystal (Figure S29). The crystal‐habit modifier has a negligible effect on the carrier mobility and trap density of the (110)‐oriented perovskite single crystal (Figure S29). Actually, the measured defect density and carrier mobility are averages of both holes and electrons, since “electron‐only” and “hole‐only” devices have not been fabricated [67, 68, 69]. Differences in optoelectronic properties are primarily attributed to variations in crystal facets, while the influence of crystal‐habit modifiers can be neglected.

FIGURE 4.

Performance comparison of the optoelectronic device of perovskite single crystals with different crystal facets. Space charge limited current (SCLC) tests of the perovskite single crystal devices with (a) the (100) crystal facet, and (b) the (110) crystal facet. The inset shows the crystal orientation of the corresponding devices. (c) The statistical carrier mobility and trap density of the perovskite single crystal devices with the (100) crystal facet and the (110) crystal facet. (d) The representative logarithmic I–V curves of perovskite single‐crystal with the (100) crystal facet and the (110) crystal facet under dark and a fixed excitation wavelength of 455 nm with illumination power of 155.4 mW cm⁻². (e) Response time of perovskite single‐crystal devices with the (100) crystal facet and the (110) crystal facet under a fixed illumination power of 155.4 mW cm⁻². (f) The operational stability of perovskite single‐crystal devices with the (100) crystal facet and the (110) crystal facet. The error bars represent the standard deviation.

Then, the two‐terminal photoconductive photodetector was fabricated. The photodetector consists of a photoconductive structure with two Au electrodes on the same side. Figure S30 shows a detailed schematic of the device configuration. Figure 4d and Figures S31a and S32a show the logarithmic I‐V curves of the perovskite single‐crystal detector with different crystal facets under light and dark conditions. Compared to the perovskite devices with the (110) crystal facet, the devices with the (100) crystal facet exhibit lower dark current (I _dark) and higher photocurrent (I _photo), which can be mainly attributed to their lower defect density. Figures S31b and S32b show the corresponding photocurrents and responsivities under different light intensities. The responsivity (R) is defined as R = (I _photo − I _dark)/P, where P is the illumination power. A higher responsivity of 1.1 × 10³ A W⁻¹ can be obtained for the perovskite devices with the (100) crystal facet, which is higher than the perovskite devices with (110) crystal facet. To more rigorous comparison of responsivities for the perovskite devices with (100) crystal facet and (110) crystal facet, statistics of responsivities from 5 different devices under the lowest detectable power were measured (Figure S33). Compared to the (110) perovskite devices with a responsivity of (0.73 ± 0.13) × 10³ A W⁻¹, the (100) perovskite devices exhibit a significantly higher responsivity of (1.10 ± 1.1) × 10³ A W⁻¹, which can be attributed to the fewer defects. Response speed is another important performance parameter for photodetectors, and its rise (fall) time is defined as the time from 10% (90%) to 90% (10%) interval of the maximum photocurrent (Figure 4e). The perovskite devices with the (100) crystal facet exhibit faster response with a rise time of 113 µs and a fall time of 141 µs. Furthermore, we monitored the environmental stability of the devices. After 32 days in an environment with 45% air humidity and a temperature of 25°C, the perovskite devices with (100) crystal facets show better environmental stability, which can be attributed to the fact that they have a lower defect density to block the damage of water and oxygen in the air (Figure 4f). Overall, perovskite devices with (100) crystal facets exhibit better performance due to their fewer deep‐level defects, which improve charge collection efficiency and reduce dark noise.

2.5. Comparison of Photocatalytic Performance

Given the variation in defect densities across different crystal facets of perovskites, we further evaluated their photocatalytic performance. Figure 5a shows the CO yield rate based on MAPbBr₃ perovskite photocatalysts with predominantly exposed (100) and (110) facets during photocatalytic CO₂ reduction. Compared to the photocatalyst with (100) facet, the photocatalyst with (110) facet exhibits a fivefold increase in the CO yield rate. The adsorption energy (E _ads) of CO₂ on the (100) and (110) facets was calculated as −2.06 and −3.10 eV, respectively (Figure 5b). The synergistic advantage of higher defect density and more negative adsorption energy on the (110) facet boosts the degradation efficiency of CO₂. Time‐dependent in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) of the photocatalytic reaction was performed to elucidate the CO₂ reduction mechanism of MAPbBr₃ perovskite photocatalysts. As shown in Figure 5c, with the increased time, the characteristic peaks of *COOH at 1397 cm⁻¹ increased progressively, peaks of •CO₂ ⁻ at 1252 cm⁻¹ decreased gradually, and the peak of HCO₃ ⁻ at 1420 cm⁻¹ remained stable. These spectroscopic observations confirm the continuous activation and protonation of CO₂, identifying the decomposition of *COOH to CO as the key step in the photocatalytic reduction process. Based on these insights, a reaction pathway for CO₂ reduction over MAPbBr₃ is proposed, wherein water vapor is oxidized on the catalyst surface to provide protons, while adsorbed CO₂ undergoes sequential proton‐coupled electron transfer to form CO (Figure 5d). ‌In addition, we investigated the C–H activation reaction performance of organic molecules based on perovskite catalysts with different crystal facets. ‌Figure 5e presents the reaction scheme of the C–H activation reaction between 4‐phenylmorpholine (4‐PhMor) and benzylidenemalononitrile (BMN). The NMR spectra confirm the successful reaction between 4‐PhMor and BMN molecules (Figure S34 ). The photocatalyst with (110) facet exhibits a superior yield in molecular synthesis than the photocatalyst with (100) facet (Figure 5f). DFT calculations further reveal higher adsorption energy of 4‐PhMor on the (110) facet compared to the (100) facet, facilitating more efficient electron transfer and thus higher catalytic activity on the (110) facet (Figures 5g and S35). The probable reaction mechanism involves 4‐PhMor and oxygen being activated by photo‐generated holes and electrons on the MAPbBr₃, respectively (Figure 5h). The activated 4‐PhMor species attack the vinyl group adjacent to the cyano moiety in the BMN molecule, triggering a coupling reaction. This step is followed by the cyclization of the coupled intermediate, which is facilitated by the presence of activated oxygen, ultimately affording the final product. The stronger adsorption energies of CO₂ and 4‐PhMor on the (110) facet, along with the higher surface energies, reveal that the inherent high surface energy of the (110) facet intrinsically leads to stronger adsorbate binding. This inherent property, coupled with a higher defect density, fundamentally underpins the superior catalytic activity observed for the (110) facet. The habit‐crystal modifier has a negligible influence on the adsorption of CO₂ and 4‐PhMor on the MAPbBr₃ surface. First, TFSI⁻ anions are partially adsorbed on the (110) facet of MAPbBr₃ crystals, leaving extensive regions of this facet exposed and available for CO₂ and 4‐PhMor adsorption (Figure S36). Under light irradiation, the photothermal effect also leads to desorption of the modifier during CO₂ photocatalytic reduction (Figure S37). Moreover, the modifier can be completely removed using dichloromethane (Figure S26). In the photocatalytic reduction of CO₂, water molecules required for CO₂ activation cause partial damage to MAPbBr₃ crystals. In contrast, in C–H activation reactions, no detectable crystal degradation occurs (Figures S38–S40). Therefore, halide perovskites are more suitable for the photochemical synthesis of organic materials in the future, without causing damage to their crystal structure. In summary, precise control over perovskite crystal facets not only advances the development of high‐performance optoelectronic devices but also offers a promising strategy for designing efficient photocatalysts.

FIGURE 5.

Performance comparison of perovskite photocatalysts with different crystal facets. (a) CO yield from the (100) crystal facet and the (110) crystal facet of MAPbBr₃ under different reaction times in a typical CO₂ photocatalytic system. (b) Adsorption energy of CO₂ on different crystal facets of MAPbBr₃ perovskite crystal. (c) Time‐dependent in situ DRIFTS spectrum of CO₂ based on the MAPbBr₃ perovskite photocatalysts. (d) Schematic illustration of the catalytic reaction process for CO₂ photocatalysis. (e) C–H activation reaction equation of 4‐PhMor with BMN under blue LED irradiation. (f) The yield of molecular synthesis with different crystal facets of MAPbBr₃ perovskite photocatalysts. (g) Adsorption energy of 4‐PhMor on the (110) facet of MAPbBr₃ perovskite crystal. (h) C–H activation reaction pathway catalyzed by the MAPbBr₃ perovskite photocatalysts.

3. Conclusion

In conclusion, we pioneer a competitive coordination strategy for the facet‐controlled synthesis of perovskite single crystals, realizing a dodecahedral morphology with (110) crystal facets. The stronger coordination of the crystal‐habit modifier with the (110) crystal facet is further revealed systematically by in situ observation of the crystallization process and theoretical calculations, which hamper the growth of the (110) crystal facet, thus generating perovskite single crystals with the (110) facet. The measurements of optoelectronic properties further revealed the presence of more defects on the (110) crystal facet, which enhances the catalytic performance. Conversely, the lower defect density of the (100) crystal facet leads to significantly improved performance of the photodetector. This innovative work not only provides fundamental insights into the facet‐dependent optoelectronic properties of perovskites but also establishes a new paradigm for tailoring perovskite crystal facets. This work has the potential to revolutionize the field of perovskite‐based optoelectronics and beyond.

Author Contributions

Yingjie Zhao: conceptualization, investigation, funding acquisition, methodology, validation, writing – review and editing, writing – original draft, formal analysis, data curation. Jinjin Zhao: software, data curation, writing – review and editing, writing – original draft, formal analysis, investigation, methodology. Xinzhang Lin: software, data curation, formal analysis, validation. Jingyuan Zhang: conceptualization, methodology, writing – review and editing, software, formal analysis, data curation. Mengru Zhang: software, data curation, validation. Meng Yuan: data curation, formal analysis, software. Kaixin Dong: software, data curation, validation. Xiangxiang Yang: data curation, validation. Ke He: software, data curation. Yuyan Zhao: software, data curation. Jianpeng Ma: software, data curation. Jiawang Zou: data curation, software. Junjie Liu: software, data curation. Dan Li: software, data curation. Xiao Wei: software, data curation. Lei Jiang: conceptualization, methodology, validation. Yuchen Wu: conceptualization, methodology, writing – review and editing, validation, funding acquisition. Yanlin Song: methodology, conceptualization, validation, writing – review and editing, investigation, funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie73676‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

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