Solution-phase transition zones govern crystallographic perfection in metal halide perovskite single crystals

Abstract

Despite rapid progress in solution-grown metal halide perovskite single crystals for optoelectronic applications, further enhancement of their intrinsic crystallographic quality through solution engineering remains challenging. In particular, the fundamental links between precursor solution properties and crystal quality, as well as a universal mechanism governing crystallization from solution, remains poorly understood. Here, using in-situ microscopic spectroscopy, we identify key solution properties of viscosity, concentration, coordination number and growing temperature as critical determinants of crystal quality through their modulation of solution-phase transition zones, which are directly observed experimentally located 10 – 60 μm away from the crystal-solution interface. We reveal a direct correlation between crystal quality, quantified by the carrier mobility–lifetime product, and the width of the transition zone, providing a clear guideline for improving crystallographic quality by widening this zone. As a result, FAPbBr₃ single crystals with markedly improved crystallographic perfection are achieved via solution engineering, demonstrating a high charge-collection efficiency of 99% and an energy resolution of 1.32% for 662 keV γ-rays from a ¹³⁷Cs source.

In-situ spectroscopy identifies solution-phase transition zones as the missing link between precursor solution properties and perovskite crystallographic quality. Widening the transition zone via solution engineering enables FAPbBr₃ crystals with 99% charge collection and 1.32% γ-ray energy resolution.

Introduction

High-performance detection of ionizing radiation is essential for a wide range of technological applications, including medical diagnostics, nuclear security, space exploration, and high-energy physics^1–3. The demand for compact detectors capable of reliable room-temperature operation has therefore driven intense interest in new semiconductor materials that combine strong radiation absorption with efficient charge transport^4,5. In this context, metal halide perovskite (MHP) single crystals have emerged as a promising platform, combining strong γ-ray stopping power, efficient charge transport, and low-cost solution processability^6–8. Progress in this field has been exceptionally rapid, with the energy resolution (ER) of perovskite-based γ-ray detectors improving dramatically from 6.5% to 0.7% for 662 keV ¹³⁷Cs within less than a decade^9–16, approaching the benchmark set by commercial CdZnTe single crystals¹⁷. Among various MHPs, solution-grown FAPbBr₃ single crystals currently deliver state-of-the-art radiation-detection performance and, together with their scalable and cost-efficient all-solution fabrication, hold strong promise for next-generation γ-ray detectors^15,16.

Despite the rapid improvement in ER for γ-ray spectral detection, the best-performing FAPbBr₃ single-crystal detectors still depend on post-synthesis treatments, such as surface passivation or long-time bias preconditioning, to maximize device charge-collection efficiency (CCE)^15,16. This reliance points to an intrinsic limitation in the charge transport properties of pristine solution-grown crystals, which has emerged as a fundamental bottleneck for further performance advances. Improving the intrinsic materials quality of solution-grown FAPbBr₃ single crystals with reduced defect densities and enhanced carrier mobility-lifetime (μτ) products has therefore long been a central yet unresolved challenge^18–20. Critically, the microscopic mechanisms by which precursor solution conditions dictate crystallization behavior, defect formation, and ultimately device performance remain poorly understood. As a consequence, empirical tuning of growth parameters, such as temperature or precursor concentration, often yields inconsistent or even contradictory results^19–23. The absence of a fundamental mechanistic link between precursor-solution-controlled determinants and the crystallographic perfection of FAPbBr₃ single crystals significantly hampers the reproducible and scalable growth of high-quality crystals, underscoring the urgent need to identify the governing principles of solution-phase crystal growth.

In this work, we open the “black box” of solution-grown FAPbBr₃ single-crystal quality by identifying the key determinants that link precursor solution conditions to crystallization outcomes. Using in-situ microscopic fluorescence analysis, we directly observe a solution transition zone, extending 10–60 μm from the crystal-solution interface, which mediates particle dissolution and transport during crystal growth. The width of this transition zone is governed by core precursor solution parameters, including temperature, solvent viscosity, precursor concentration, and the coordination environment of solute clusters, and ultimately dictates the crystallographic quality of FAPbBr₃ single crystals. By establishing a semi-quantitative correlation between the transition-zone width and the μτ product, CCE and γ-ray ER of FAPbBr₃ single crystal detectors, this work provides a mechanistic framework and practical design principle for improving γ-ray detector performance through solution engineering.

Results

Solvent-dependent crystallization quality of FAPbBr₃ single crystals

For solution-grown FAPbBr₃ single crystals, γ-butyrolactone (GBL)/dimethylformamide (DMF) and propylene carbonate (PC)/DMF are the two most commonly used solvent systems because of their suitable inverse-temperature crystallization windows^15,24–27. Although both solvent systems enable supersaturation for inverse-temperature crystallization, differences in their molecular structure lead to distinct crystallization temperatures and solution viscosities, resulting in markedly different crystal qualities. Within the framework of the classical LaMer model, single-crystal growth from solution is generally considered to be diffusion-controlled, with the diffusion kinetics governed by parameters such as temperature, solvent viscosity, and solute concentration^28,29. However, elucidating the individual impact of each parameter on FAPbBr₃ single-crystal quality has remained challenging, as tuning one variable inevitably perturbs the others. In this context, we design a set of experiments aimed at decoupling the roles of these core parameters in determining the resulting crystal quality.

We first investigate crystal growth in the GBL/DMF = 1:1 v/v solvent system with solute concentrations ranging from 1.2 to 1.5 M, which are commonly used in the literature^16,18,19. As shown in Fig. 1a and Supplementary Table 1, varying the solute concentration simultaneously alters both the solution viscosity and the crystallization temperature. In particular, we observe a strong dependence of the crystallization temperature on the precursor concentration, reflecting the pronounced temperature dependence of the solvent–solute interactions, whereby elevated temperatures weaken the solvent–solute coordination interactions and consequently reduce the formation of solvent–solute complexes that govern the solubility of the precursor species³⁰. FAPbBr₃ single crystals grown under these conditions were then fabricated into γ-ray detectors using a standard gold (Au)/FAPbBr₃/bismuth (Bi) sandwich architecture (Supplementary Fig. 1), and their CCE and ER for ⁵⁷Co γ-ray detection were evaluated. Figure 1b presents the ⁵⁷Co γ-ray spectra of devices fabricated from crystals grown at different concentrations, and the spectral test parameters are shown in Supplementary Table 2. Notably, both CCE and ER improve as the solute concentration increases from 1.2 to 1.4 M, but deteriorate at 1.5 M. Accordingly, thermal admittance spectroscopy (TAS) measurement reveals a reduction in both trap bands II and III as the concentration increases from 1.2 to 1.4 M, followed by a pronounced increase primarily in trap band III for the 1.5 M crystal-based device (Fig. 1c). These results indicate that solute concentration alone does not determine the FAPbBr₃ single crystal quality, rather, the optimal performance at 1.4 M reflects the combined influence of solute concentration, solution viscosity, and crystallization temperature. Specifically, the extremely high precursor concentration of 1.5 M may hinder the diffusion of the relatively large FA⁺ cations within the dense precursor solution, leading to local FA⁺ depletion at the crystal growth front and ultimately the formation of FA vacancies, which are associated with trap band III in the as-grown crystals^15,31.

Fig. 1. Solvent-dependent crystallization quality of FAPbBr₃ single crystals.

Variation of solution viscosity and temperature versus concentration in the GBL/DMF solvent system (a) and solvent composition at the solution concentration of 1.4 M (d). Variation of the ⁵⁷Co spectra under −500 V operation of FAPbBr₃ crystal detectors with changes in concentration (b) and solvent composition (e), where the ⁵⁷Co source was irradiated from the bismuth side. All detectors have a similar thickness of 2 mm. tDOS of FAPbBr₃ crystals dictated by precursor concentration (c) and solvent conditions (f). Trap bands II and III are noted at the characteristic frequencies of around 10 and 0.1 kHz, respectively.

To further isolate the roles of individual parameters in determining crystal quality, we fixed the solute concentration at 1.4 M and systematically tuned the volume ratios of GBL/DMF and PC/DMF to achieve crystal growth under comparable viscosities or crystallization temperatures. As specifically summarized in Fig. 1d, conditions (1) PC/DMF = 2:3 (v/v) and (2) GBL/DMF = 1:1 (v/v) correspond to high- and low-viscosity solvent systems, respectively, at low temperature. Whereas conditions (3) PC/DMF = 1:2 (v/v) and (4) GBL/DMF = 2:3 (v/v) represent high- and low-viscosity systems, respectively, at elevated temperature. The corresponding device performance evaluated by CCE and ER for ⁵⁷Co γ-ray detection is shown in Fig. 1e. It is seen that at a fixed growth temperature, higher solution viscosity apparently yields superior device performance (conditions 1 vs. 2 and 3 vs. 4), indicating improved crystal quality potentially linked to high-viscosity conditions. Likewise, for comparable solution viscosities (conditions 1 vs. 3), lower growth temperature leads to enhanced crystal quality. Notably, comparison between conditions 2 and 3 further endorses this statement, where the reduction in growth temperature produces a more pronounced improvement in crystal quality than the adverse effect associated with reduced viscosity. These device-level trends are further corroborated by photoluminescence (PL) measurements on pristine crystals, which show stronger emission intensity and longer carrier lifetimes for crystals grown at lower temperature and higher viscosity (Supplementary Fig. 2).

Moreover, we evaluate the defect characteristics of these FAPbBr₃ single crystals using TAS. As shown in Fig. 1f, crystals grown at lower temperatures exhibit a markedly reduced density of trap band II. A similar trend is observed in the concentration-dependent growth series (Fig. 1c), where both growth temperature and solution viscosity vary concurrently. Previous studies have assigned trap band II to Br vacancy (V_Br⁺) defects in FAPbBr₃ single crystals¹⁵, and our current results provide additional insight into how V_Br⁺ defect formation evolves under different growth conditions.

Dependence of crystallographic perfection on coordination complexes

In addition to varying solution concentration and solvent composition, fine-tuning the solute itself also modifies its chemical environment and, at a more microscopic level, its coordination-complex structure in solution, thereby influencing the crystallization dynamics^32,33. In the solution growth of FAPbBr₃ single crystals, multiple Pb-Br coordinated species denoted as [PbBr_x]^(x-2)− coexist, with their relative populations governed by solvent coordination strength and the concentration ratio between Pb²⁺ to Br⁻ ions^34–37. To deliberately modulate these coordination complexes, we introduce excess FABr beyond the stoichiometric ratio (referred to as FABr-rich condition) and, guided by the Lewis acid-base theory, incorporate FACl as an additive, in which Cl^- ions stand as a stronger Lewis base than Br⁻ with the coordination ability to Pb²⁺ (referred to as FACl additive condition)^38,39.

These strategies enable controlled alteration of the [PbBr_x]^(x-2)− complexes, as schematically illustrated in Fig. 2a, and are evidenced by UV–vis absorption spectra (Fig. 2c), where the absorption peak at 313 nm assigned to [PbBr₃]⁻ species increases in intensity, accompanied by a concomitant decrease in the peak at 283 nm associated with low-coordination PbBr₂ species in both FABr-rich and FACl-additive solutions^40,41. This transition from low- to high-coordination complexes reflects a statistical increase in the average halide coordination number around Pb²⁺, leading to a reduced complex cluster size compared to lower coordinated complexes^42,43. Notably, no distinguishable peak shift potentially associated with the formation of [Pb(ClBr)_x]^(x-2)- complexes was observed in the 1% Cl⁻-incorporated sample, whereas such a shift became evident upon further increasing the Cl⁻ concentration to 2.5% (Supplementary Fig. 3). Here, we intentionally select Cl⁻ concentration at an extremely low level to modulate the Br coordination environment while minimizing potential side effects arising from Cl^- incorporation during the crystallization process.

Fig. 2. Dependence of crystallographic perfection on coordination complexes.

a Schematic of precursor solution chemistry governing coordination number for the stoichiometric, FABr rich and FACl additive solutions. b GIWAXS images of FAPbBr₃ single crystals grown under different precursor coordination environments. c Ultraviolet–Visible absorption spectra of FAPbBr₃ precursor solutions with different ratios. d ESI-MS spectral distribution as a function of precursor stoichiometry. Positive-ion mode spectra (25 °C) showing characteristic signals at m/z = 60, 96, and 149, assigned to the free FA⁺, [DMF+Na]⁺, and [PC+Na]⁺ adducts, respectively. e tDOS of FAPbBr₃ crystals depends on different precursor ratios. f Arrhenius plots of the inflection frequency in FAPbBr₃ crystals versus precursor stoichiometry. g Variation of the ⁵⁷Co spectra under −500 V operation of FAPbBr₃ crystal detectors with changes in the precursor stoichiometry, where the ⁵⁷Co source was irradiated from the bismuth side. All detectors have a similar thickness of 2 mm.

To further quantify the solute coordination environment, we performed electrospray ionization mass spectrometry (ESI–MS) on the diluted precursor solutions to probe the concentration of free FA⁺ ions (detailed in “Methods” and Supplementary Note 1), which reflects the population of uncoordinated FA⁺ relative to the charged [PbBr_x]^(x-2)− framework. As shown in Fig. 2d and Supplementary Fig. 4, the signal corresponding to free FA⁺ decreases markedly from the stoichiometric solution to the FABr-rich solution and is further suppressed upon FACl addition. These results further corroborate the formation of higher-coordination [PbBr_x]^(x-2)− complexes in FABr-rich and FACl-additive solutions, thereby promoting more effective incorporation of FA⁺ into the Pb-Br framework. It should be noted that the ESI‑MS samples were prepared by high dilution of the precursor solutions in ethanol, which does not fully preserve the native solvation equilibrium. However, all samples were treated identically, allowing comparative analysis to remain reliable. The use of a non‑coordinating diluent in future work could further validate the intrinsic solution species.

We next evaluate the crystallographic quality of FAPbBr₃ single crystals grown under these three conditions in terms of structural perfection, charge-transport properties, defect characteristics, and ultimately their optoelectronic device performance. X-ray diffraction (XRD, Supplementary Fig. 5) and rocking-curve measurements (Supplementary Fig. 6) show that all crystals exhibit highly oriented crystallization with well-defined diffraction features. Notably, the full width at half maximum (FWHM) of the rocking curve narrows under FABr-rich conditions relative to the stoichiometric case and is further reduced to ~ 60 arcsec for crystals grown with FACl additives, indicating enhanced crystalline perfection. Furthermore, it is worth noting that at such a low Cl⁻ introduction, no distinguishable XRD peak shift occurs (Supplementary Fig. 5). We have supplemented X-ray photoelectron spectroscopy (XPS) and energy-dispersive X-ray spectroscopy (EDX) measurements, neither of which detected discernible Cl signals (Supplementary Figs. 7 and 8). These results indicate that any potential Cl⁻ incorporation, if present, must be below the detection limit of approximately 0.1 at%, which is substantially lower than the concentrations reported in previous studies where intentional Cl doping significantly altered the electronic properties of MAPbBr₃ single crystals^11,44. Nevertheless, we acknowledge that even trace amounts of incorporated Cl⁻ may contribute to the observed behavior in addition to its primary role in modifying the precursor coordination environment during crystal growth. Such trace incorporation could influence the electronic and optoelectronic properties of the crystals through defect compensation, mixed-halide electronic effects, and/or suppression of ion migration.

To further probe structural coherence, grazing-incidence wide-angle X-ray scattering (GIWAXS) measurements were performed. As shown in Fig. 2b, the stoichiometric sample lacks the characteristic (110) diffraction signal at q = 1.4 and χ = 45°, whereas this feature appears distinctly in both FABr-rich and FACl-incorporated crystals⁴⁵. Moreover, diffraction spots in these halide-enhanced systems become progressively sharper, signifying reduced lattice disorder and improved structural uniformity.

Static and time-resolved PL (TRPL) measurements of the crystal bulk further reveal enhanced emission intensity and prolonged carrier lifetimes in FABr-rich and FACl-incorporated crystals (Supplementary Figs. 9 and 10), consistent with their improved crystallographic quality. Time-of-flight (TOF) measurements show that FACl incorporation significantly enhances the carrier mobility of FAPbBr₃ single crystals from 80 cm² V⁻¹ s⁻¹ (stoichiometric) to 120.9 cm² V⁻¹ s⁻¹, corresponding to a 1.5-fold increase (Supplementary Fig. 11). As a result, the μτ product of devices based on these crystals increases substantially from 5 × 10⁻⁴ cm² V⁻¹ to 8 × 10⁻³ cm² V⁻¹ upon FACl addition, consistent with the observed improvements in both carrier lifetime and mobility (Supplementary Figs. 12 and 13). We further examine their defect characteristics using TAS. As shown in Fig. 2e, introducing excess FABr into the precursor solution markedly reduces the density of trap band II, which is the V_Br⁺ defect. In addition, trap band III, which has been identified in earlier studies as originating from FA-vacancy (V_FA⁻), is also suppressed upon FABr or FACl incorporation¹⁵. This trend is consistent with our ESI-MS results, which reveal a reduction in free FA⁺ ions accompanied by an increase in coordinated FA⁺ species. Collectively, these results demonstrate that FACl incorporation yields FAPbBr₃ single crystals with superior crystallographic perfection, reduced defect densities, and markedly enhanced charge-transport properties.

Then we evaluate the optoelectronic performance of these crystals in γ-ray detection devices. Current density-voltage (J–V) measurements reveal a progressive suppression of dark current under reverse bias, following the order of stoichiometric, FABr-rich, and FACl-additive crystals (Supplementary Fig. 14). Interestingly, the J–V characteristics of different devices exhibited different electrical properties. To understand this phenomenon, we performed cross-sectional Kelvin probe force microscopy (KPFM) to determine their work functions (Supplementary Fig. 15). To minimize the surface-related influences on the KPFM measurements and obtain a more reliable estimate of the intrinsic bulk work functions of the crystals, all KPFM measurements were performed on freshly exposed crystal surfaces generated by cleaving the bulk crystals immediately prior to measurement. Test preparation was the same as the crystal bulk PL measurements (Supplementary Fig. 9). Conventionally grown Cl⁻-free FAPbBr₃ crystals show a work function of 4.824 eV, indicating a weak p-type doping nature, whereas crystals grown with added Cl⁻ exhibit a work function of 4.636 eV, close to intrinsic. The weakly p-type crystal forms a Schottky contact with the low-work-function Bi electrode, leading to hole depletion near the interface and typical rectifying behavior. However, for sufficiently thick, high-resistance crystals, most of the applied voltage could drop across the intrinsic bulk rather than the interfacial depletion region, especially at low doping levels. Consequently, as previously reported and observed here^11,46, devices with asymmetric electrodes but an almost intrinsic perovskite layer can exhibit relatively symmetric I–V characteristics, reflecting bulk-limited conductivity dominance. Meanwhile, given the majority-carrier-dominated nature of Schottky barrier devices⁴⁷, the Cl-optimized crystals, although exhibiting a weakened Schottky barrier, possess a drastically reduced doping density approaching intrinsic behavior, thereby maintaining low dark currents under both bias polarities. The observed behavior is likely the result of a synergistic combination of Cl^–-induced coordination engineering during crystal growth, which enhances crystallographic quality, and potential trace lattice-incorporated Cl⁻, which may further influence the defect landscape, carrier compensation, and electronic properties of the crystals. More importantly, we further acquired energy spectra under elevated bias voltages ranging from 200 to 500 V, which revealed no increase in dark noise; in particular, no enhancement of low-energy counts was observed under higher-bias conditions. Moreover, the full width at half maximum (FWHM) of the full-energy peak remained essentially unchanged with increasing bias voltage (Supplementary Fig. 16).

Furthermore, γ-ray spectra acquired from ⁵⁷Co sources show a pronounced enhancement in CCE from 52.9% to 91% and ER from 17.1% ± 1.2% down to 3.8% ± 0.8%, from devices based on stoichiometric crystals to those incorporating FACl additives (Fig. 2g), demonstrating the superior detector performance enabled by the improved crystal quality. Moreover, as shown in Fig. 2f, temperature-dependent J–V measurements indicate a substantial increase in the activation energy for ion migration in FABr- and FACl-incorporated crystals, evidencing significantly enhanced operational stability in coordination-engineered devices.

In situ characterization of solution-phase transition zone for crystal growth

The above results demonstrate that the quality of solution-grown FAPbBr₃ single crystals depends sensitively on precursor properties and growth conditions. However, whether and how the parameters of precursor solutions act through a unified, dominant mechanism, despite being accompanied by distinct and case-specific incremental effects, remains unresolved.

This motivated us to investigate the in situ crystallization dynamics under various solution-growth conditions. Here, we employed in situ fluorescence lifetime imaging microscopy (FLIM) combined with spectroscopic measurements to directly monitor the crystal growth process. Figure 3a schematically illustrates the in-situ FLIM setup, in which a 405 nm excitation laser is focused onto the solution-grown crystal confined between two glass substrates mounted on a customized temperature-controlled stage. PL spectra and their spatial distributions from the crystal and the surrounding solution are collected through a confocal microscopy system and recorded using a spectrometer equipped with wavelength-selective filters.

Fig. 3. In situ characterization of solution-phase transition zone.

a Schematic diagram of an in-situ microscopic fluorescence spectroscopy system. b Standard PL mapping image of a thin single crystal grown in the solution. c Point-scan PL full-spectrum results of the solution and crystal measure on the cross-marked points from (b). To allow a clearer visual comparison between the two spectra, the PL intensity of the solution spectrum was multiplied by a factor of 50. d PL mapping results of the precursor solution at room temperature (25 °C) and crystal growth temperature (55 °C). e The result mapping after overlaying the PL mapping of the 432 ± 20 nm solution and the 540 ± 20 nm crystal. The area outlined by the white dashed line in the figure is the transition zone. f Normalized line-scan PL intensity profiles at 430 ± 20 nm and 540 ± 20 nm along the white line from the crystal to the solution marked in (e). g–i Correlation between precursor composition and photoluminescence. g Normalized PL intensity profiles obtained from line-scan mapping at 430 nm (FWHM 40 nm) for solutions with varied precursor stoichiometry. h Concentration-dependent evolution of the integrated PL signal at 430 ± 20 nm from line-scan mapping. i PL intensity (430 ± 20 nm) profiles from line-scan mapping under different solvent conditions, demonstrating the modulation of luminescence by the local chemical environment. j Schematic diagram of solution growth transition zone.

Figure 3b presents a representative PL mapping image across the crystal-solution interface of a thin FAPbBr₃ single crystal, where the bright and dark regions correspond to the crystal and the surrounding solution, respectively. Full PL spectra were subsequently acquired at positions on the crystal and in the solution, as indicated in Fig. 3b, by scanning the detection window from the cut-on edge of a 410 nm long-pass filter to wavelengths beyond 600 nm. The resulting spectra are shown in Fig. 3c. Interestingly, in addition to the routine characteristic emission peak at 540 nm observed from the crystal, the solution exhibits a distinct PL feature peaked at around 432 nm with an asymmetric tail extending toward 480 nm.

To understand the origin of the 432 nm PL signal, we carried out wavelength-selective PL mapping centered at 432 ± 20 nm on the precursor solution alone at both room temperature (25  °C) and the crystal-growth temperature (55 °C), as shown in Fig. 3d. At both temperatures, the 432 nm emission is distributed throughout the field of view, with enhanced intensity localized at discrete particles ranging from hundreds of nanometers to a few micrometers in size. Time-resolved PL (TRPL) measurements collected from these fluorescent particles and from the surrounding background solution exhibit similar decay dynamics (Supplementary Fig. 17), indicating a common emissive origin. Notably, in pure Br-based perovskite solutions, dispersed [PbBr_x]^(x-2)− clusters or monomers typically possess band gaps too wide to be efficiently excited by a 405 nm laser^41,48. To further investigate the fluorescence characteristics of the precursor solution beyond the limitation of the excitation laser wavelength, additional solution-phase PL measurements were performed using a 266 nm excitation laser coupled with a 360 nm long-pass filter. As shown in Supplementary Fig. 18, the precursor solution exhibits a broad PL emission spanning from 380 to 500 nm. Further peak analysis using multi-peak Gaussian fitting resolved two distinct emission peaks at 405 nm and 432 nm, which are attributed to 1-monolayer and 2-monolayer perovskite nanoplate structures, respectively, in the Br-based perovskite precursor solution^49,50. We therefore adopt the 432 nm PL emission in the solution-phase PL mapping as an indicator of the aggregated species containing multilayer low-dimensional perovskite nanoplates in the solution.

Furthermore, by tracking the PL spectra of the solution along a linear trajectory toward the crystal-solution interface in the presence of a grown crystal, we observe that the intensity of the 432 nm emission gradually decreases as the distance to the interface is reduced from 30 to 6 μm (Supplementary Fig. 19). This behavior reflects the progressive depletion of emissive precursor species in the raw solution near the crystal interface, likely due to their transformation into more highly coordinated and more dispersed [PbBr_x]^(x-2)− clusters and eventually monomers at the crystal interface driven by the consumption of the solutes. A recent study by Shi et al. reported an interfacial region of 1.5–4 μm arising from the dissolution of [PbI_x]^(x-2)− clusters into monomers at the MAPbI₃ crystal interface⁴⁸, here our results provide a more comprehensive view of the solution-phase transition process via the direct capture of the fluorescence signals emitted from clusters in the solution phase, capturing the extended evolution of precursor species toward perovskite single-crystal crystallization.

To further elucidate the extended solution-phase transition process, we performed in situ PL mapping of the growing crystal and the surrounding solution simultaneously by employing a beam splitter to collect both the 540 ± 20 nm and 432 ± 20 nm emission channels. The two PL intensity maps were subsequently overlaid, as shown in Fig. 3e. Line profiles of the 540 nm and 432 nm emission intensities were then extracted along a direction perpendicular to the crystal-solution interface, spanning from the crystal into the solution, as indicated in Fig. 3e and summarized in Fig. 3f. A distinct solution-to-crystal transition zone is observed between the crystal edge and the bulk solution, within which the 432 nm emission intensity decreases progressively while the 540 nm emission intensity increases. It is noted that, for consistency of all the measurements to enable a reliable comparative analysis, the crystal edge is defined as the position where the solution PL signal first appears without reabsorption by the crystal. The apparent extension of the crystal PL emission beyond the geometric crystal boundary into the transition zone is attributed to: (1) the non-ideal and non-flat crystal surface approaching the solution with inherent geometric fluctuations; (2) geometric overlap between the solution phase and the crystal region; and (3) the finite optical resolution of both the excitation and collection optics.

Using this approach, we investigate the transition zone by analyzing the spatial profiles of the solution PL intensity under the various crystal-growth conditions described above (Supplementary Figs. 20–29). As shown in Fig. 3g–i, the width of the transition zone varies distinctly across different growth conditions. Specifically, the transition zone becomes significantly wider in precursor solutions containing a higher fraction of highly coordinated [PbBr_x]^(x-2)− clusters, as observed in FABr-rich and FACl-additive incorporated solutions (Fig. 3g). This trend supports a solution-phase transition process in which raw solution aggregates are progressively converted into more highly coordinated [PbBr_x]^(x-2)− species, with the formation of smaller and highly coordinated clusters extending the spatial range of this transition, as shown in Fig. 3j. To further exclude performance changes arising from halide-ion doping and to isolate the influence of the coordination environment on the transition-zone width, we selected acetate ions (Ac⁻) as representative coordinating additives. As well established in the literature^51,52, Ac⁻ ions interact strongly with Pb²⁺, promoting the formation of highly coordinated complexes in the precursor solution without being incorporated into the perovskite lattice as dopants. Upon the addition of Ac⁻, we observed an increased transition-zone width (Supplementary Fig. 30a) along with improved device performance (Supplementary Fig. 30b).

Meanwhile, the transition-zone width increases systematically under growth conditions characterized by lower temperature, higher viscosity (lower fluidity), and lower precursor concentration. It is noted that these factors broaden the transition-zone independently of the coordination environment, as evidenced by their nearly identical UV–vis spectra (Supplementary Fig. 31). Because all three conditions reduce mass-transport efficiency, a wider spatial region is required to accommodate solute transformation and transport toward the crystal interface.

In addition, importantly, we observe an experimental correlation between the measured transition-zone width and the crystallographic quality of the resulting FAPbBr₃ single crystals, suggesting that wider transition zones are consistently associated with higher-quality crystals. To further substantiate this correlation, we grew two FAPbBr₃ single crystals from the same precursor solution under different growth temperatures and tracked their transition-zone widths using the in situ FLIM setup. As shown in Supplementary Fig. 32, crystals grown at a lower quasi-equilibrium growth temperature (55 °C) and at a higher temperature (60 °C) yielded optically crack-free and cracked crystals, respectively, with the former exhibiting a markedly wider transition-zone profile than the latter (Supplementary Fig. 33). These results imply that the solution-phase transition zone plays a critical role in maintaining crystallographic order during growth, with its spatial extent governing the final crystal quality and structural uniformity.

Solution-phase transition-zone-guided crystallization optimization

To assess the practical guidance offered by the transition-zone width for crystal-quality optimization, we performed a statistical analysis correlating the transition-zone width with the μτ product of the resulting crystals, which reflects the intrinsic material quality and directly governs the CCE of device performance. As summarized in Fig. 4a, the analysis reveals that progressively wider transition zones consistently lead to systematically higher μτ values. Furthermore, we examine the relationship between device CCE and ER for detectors fabricated from crystals with different transition-zone widths, measured at a fixed electric field (E) of 250 V mm⁻¹. As shown in Fig. 4b, the CCE increases monotonically with increasing transition-zone width at constant E. Notably, this dependence can be well-fitted by the Hecht model equation using a constant E and the independent variable of μτ product⁵³, further corroborating the near-proportional relationship between the μτ product of the crystals and the transition-zone width. Based on extrapolation of the Hecht equation (right half of Fig. 4b), a further increase in transition-zone width is expected to enable near-unity CCE, or alternatively provide sufficient margin to operate at lower E while maintaining high CCE (indicated by the gray dashed line). Such flexibility is advantageous for enhancing long-term device stability and relaxing operational constraints. Consistent with the observed dependence of μτ and CCE on transition-zone width, statistical analysis further reveals a gradual improvement in ER from 27.9 ± 1.3% to 3.5 ± 0.7% for 122 keV ⁵⁷Co γ-ray detection by FAPbBr₃ single crystals grown with increasing transition-zone width (Fig. 4c).

Fig. 4. Solution-phase transition-zone-guided crystallization optimization.

a Dependence of μτ product on the transition zone width. The sample size is 2. The error bars indicate the standard deviation. b Dependence of CCE on the transition zone width (blue circles). The corresponding operating electric field is plotted by black circles. The left panel of the vertical dashed line displays the statistical results from the measurements, and the right panel denotes the extrapolated region. The horizontal gray dashed line represents CCE = 99%. The blue dashed line in the figure represents the Hecht equation fitting result. The gray-black dashed line on the right panel represents the electric field results extrapolated according to the Hecht equation. The sample size is 2. The error bars indicate the standard deviation. c Effect of transition zone width on ER of the single crystal detector. The dashed line illustrates the overall trend. The sample size of the transition zone width test is 2, and the error bars indicate the standard deviation. The error bars of the ER indicate the standard deviation. d μτ value of FAPbBr₃ single crystal devices grown with optimizing solution condition. e ¹³⁷Cs energy spectrum results of FAPbBr₃ single-crystal devices grown with optimizing solution condition, with the ¹³⁷Cs source irradiating from the bismuth electrode side. The thickness of the detector is 2 mm. f Schematic diagram of the relationship between solution conditions, transition zone width, and crystal quality. Four parameters, coordination number, fluidity (inverse of viscosity), concentration, and temperature define the vertexes of a pyramidal phase space representing its adverse influence on the transition-zone width. Growth conditions located near the bottom center of this pyramid render wider transition zones and superior crystal quality. Growth conditions located on the top of this pyramid result in narrower transition zones and inferior crystal quality.

Guided by the transition-zone-governed crystallization quality framework established for solution-grown FAPbBr₃ single crystals, we optimized the growth conditions by employing a precursor concentration of 1.4 M with the addition of 1% FACl in a PC/DMF = 2:3, v/v solvent system, of which the combination promotes higher coordination numbers of Pb-Br complexes, increased solution viscosity, and a balanced crystallization temperature and precursor concentration. Specifically, the incorporation of FACl primarily regulates the precursor coordination environment, the high-viscosity solvent mainly governs precursor transport kinetics, and the solute concentration predominantly influences crystallization thermodynamics and transport equilibrium. Their synergistic effects collectively broaden the transition zone, suppress defect formation, and ultimately enhance crystal quality and device performance. As a result, this solution-engineering strategy yields high-quality FAPbBr₃ single crystals with a μτ product exceeding those of suboptimal counterparts by more than an order of magnitude, reaching 1.2 × 10^–2 cm² V^–1 (Fig. 4d). γ-ray detectors fabricated from the optimized single crystals statistically exhibit a top-tier CCE of around 99% and state-of-the-art ER values of 3.5 ± 0.7% and 1.64 ± 0.32% (with a champion value of 1.32%) for 122 keV ⁵⁷Co and 662 keV ¹³⁷Cs sources, respectively (Fig. 4e and Supplementary Figs. 34–36). These values rank among the best reported for solution-grown FAPbBr₃ single-crystal detectors (Supplementary Table 3). Impressively, the optimized crystals, with significantly improved crystal quality and reduced ionic defect density, substantially suppress the preconditioning requirements associated with mobile-defect-induced charge dynamics (Supplementary Figs. 37–38). The corresponding devices exhibit only short operational bias-dependent transients of approximately 2 min, comparable to those observed in commercialized CZT detectors (Supplementary Fig. 38c), which are more likely associated with intrinsic polarization and space-charge effects. Moreover, the devices display excellent linear response over a broad energy range from 59.5 to 662 keV (Supplementary Fig. 39), as well as outstanding environmental stability, maintaining their initial CCE and ER over three months of ambient storage (Supplementary Fig. 40).

Discussion

Currently, our work proposes a semi-quantitative framework that links the crystal quality of solution-grown FAPbBr₃ single crystals to precursor solution conditions through a unified, dominant determinant of the solution-phase transition zone. Such a connection could be understood from combined thermodynamic and diffusion-kinetic perspectives, that a quasi-steady state balancing mass transport, solute transformation, and interfacial consumption is essential for perfect crystallization^54–56. For crystals grown under near-thermal-equilibrium growth conditions, where balanced mass transfer and crystallization ratio is established⁵⁷, a wider transition zone accommodates more gradual compositional evolution and smoother redistribution of coordination environments, thereby mitigating local supersaturation fluctuations and relieving structural perturbations that would otherwise promote defect formation^58,59. These effects are expected to become increasingly pronounced as solute species approach the crystal interface, where a sub-5 μm interfacial region more critically governs the facet stability from destruction, as recently reported⁴⁸. We believe this framework establishes a universal yet fundamental link between solution conditions and the resulting crystal quality, upon which additional strategies, such as additive incorporation with functional groups can exert incremental improvements in defect mitigation and overall crystallographic perfection^22,24,60,61.

Within this framework, key solution parameters, including but not limited to, temperature, viscosity, precursor concentration, and the coordination number of Pb-halide complexes, govern the final crystal quality through their modulation of the transition-zone width. However, the influence of these parameters cannot be straightforwardly ascribed to any single factor because of their strongly coupled and interdependent nature. In inverse-temperature crystallization systems of perovskite single crystals, lower precursor concentrations generally require higher growth temperatures and are associated with lower solution viscosities (higher fluidity), whereas lower growth temperatures are often accompanied by higher precursor concentrations, in the absence of solvent or additive engineering. This intrinsic coupling renders the rational pursuit of high-quality crystals nontrivial. Such mutually constraining interactions among the four key parameters are schematically illustrated in Fig. 4f, where each parameter defines a vertex of a pyramidal phase space representing its adverse influence on the transition-zone width. Growth conditions located near the bottom center of this pyramid, characterized by lower temperature, lower concentration, lower fluidity (higher viscosity), and higher coordination number generally correspond to wider transition zones and, consequently, superior crystal quality. In contrast, conditions deviating from this regime are increasingly dominated by the negative influence of one or more parameters, leading to reduced transition-zone widths and inferior crystallographic quality. We would also like to emphasize that, while our work establishes a semiquantitative relationship between the solution-phase transition zone and crystal quality, further improvements may be achieved by exploiting the unique chemical and electronic functionalities of different additives, provided that comparable transition-zone conditions are maintained during crystal growth.

To further demonstrate the generality of our optimization strategy, we extended the study to MAPbBr₃ single crystals. We first measured the PL of the MAPbBr₃ precursor solution (Supplementary Fig. 41). A clear PL peak at around 430 nm was observed, consistent with reported PL from low-dimensional MAPbBr₃ nanoplates^62,63. We then compared the absorption spectra of precursor solutions with different MABr: PbBr₂ molar ratios (Supplementary Fig. 42h). Excess MABr beyond stoichiometry increased the coordination number of Pb-based complexes, which is expected to stabilize highly coordinated intermediates and broaden the transition zone. Accordingly, transition-zone width measurements on MAPbBr₃ single crystals grown from solutions with higher coordination (excess MABr) showed a substantially wider transition region (Supplementary Fig. 42a–g). Devices fabricated from these crystals exhibited superior performance, mirroring our observations in the FAPbBr₃ system (Supplementary Fig. 42i). These consistent results across two perovskite compositions support the view that broadening the transition zone through enhanced precursor coordination is a general principle.

In summary, the crystallographic quality of solution-grown metal halide perovskite single crystals is critically limited by an incomplete understanding of how precursor solution conditions govern crystallization. Through direct in situ visualization, we identify a solution-phase transition zone that mediates solute transformation and mass transport during perovskite single-crystal growth. We show that its spatial extent integrates the coupled effects of viscosity, precursor concentration, coordination number, and temperature, and acts as a dominant determinant of crystallographic quality. Establishing a semi-quantitative link between transition-zone width and carrier transport metrics enables predictive optimization of crystal growth, culminating in FAPbBr₃ detectors with near-ideal charge collection and top-tier γ-ray spectroscopic performance.

Methods

Materials

Lead bromide (>98.0%), dimethylformamide (99.8%), propylene carbonate (99.7%), Chlorobenzene (99.8%), formamidine hydrochloride (97%) were purchased from Sigma-Aldrich. Formamidinium bromide (>99.99%) was purchased from GreatCell Solar. γ-butyrolactone (99%) and ethanol (≥99.7%) were purchased from Sinopharm Chemical Reagent Co., Ltd. All the raw materials and solvents were used as received without further purification.

Growth of FAPbBr₃ single crystals

For FAPbBr₃ crystals grown from precursor solutions of varying concentrations, FABr and PbBr₂ were dissolved in a 1:1 molar ratio in a mixed solvent of N,N-dimethylformamide and γ-butyrolactone (1:1 by volume), yielding clear precursor solutions with concentrations of 1.2, 1.3, 1.4, and 1.5 M. After stirring at room temperature for 24 h, the salts reacted completely to form a transparent FAPbBr₃ precursor solution. For seed crystal preparation, approximately 5 mL of the precursor solution was heated to about 65 °C and allowed to stand for 1 h to induce nucleation. Seed crystals smaller than 0.5 mm in diameter were carefully selected using clean tweezers. For bulk crystal growth, 10 mL of the precursor solution was filtered through a 0.2 μm PTFE membrane and transferred into a small culture vial. An intact seed crystal was placed at the center of the vial bottom. When the crystal seed neither dissolved nor exhibited further growth within 1 h, the corresponding temperature was defined as the crystallization temperature for that particular solute concentration. The vials were then transferred to ovens set at 52, 46, 43, and 40 °C for the 1.2, 1.3, 1.4, and 1.5 M solutions, respectively, each held for 2 h, followed by a gradual temperature increase of 1 °C per day. Within 3–5 days, centimeter-sized crystals (approximately 1.0 × 1.0 × 0.2 cm³) were obtained. Finally, the grown crystals were removed from the solution, immediately immersed in chlorobenzene, and dried under a stream of nitrogen gas.

For FAPbBr₃ crystals grown from precursor solutions in different solvent systems, FABr and PbBr₂ were dissolved in a 1:1 molar ratio in four mixed solvents (PC/DMF = 2:3 v/v, PC/DMF = 1:2 v/v, GBL/DMF = 1:1 v/v, GBL/DMF = 2:3 v/v) to prepare a 1.4 M precursor solution. After stirring at room temperature for 24 h, the salts completely reacted to form a clear FAPbBr₃ precursor solution. The remaining crystal growth process was the same as that for crystals grown from solutions of different concentrations.

For the growth of FAPbBr₃ crystals with excess FABr and added FACl, FABr and PbBr₂ were dissolved in a 1.3:1 molar ratio in a mixed solvent of PC/DMF = 1:2 (v/v) to prepare a 1.4 M FABr-rich precursor solution. Building on the FABr-rich precursor solution, an additional 1% FACl (relative to FABr) was added to prepare a 1.4 M FACl precursor solution. After stirring at room temperature for 24 h, the salts fully reacted to form a transparent FAPbBr₃ precursor solution. The remaining crystal growth operations were the same as those used for crystals grown from solutions of different concentrations.

For the optimal FAPbBr₃ crystal system, FABr, FACl, and PbBr₂ were dissolved at a molar ratio of 1.3:0.013:1 in a mixed solvent (PC/DMF = 2:3 v/v) to prepare a 1.4 M precursor solution. After stirring at room temperature for 24 h, the salts fully reacted to form a transparent FAPbBr₃ precursor solution. The remaining crystal growth operations were the same as those used for crystals grown in systems with different concentrations.

For thin crystal growth, drop 10 μl of the filtered solution onto the center of a preheated 25 × 25 mm² glass substrate, then quickly cover it with a 20 × 20 mm² cover glass. Usually, the initial nucleation for thin crystal growth requires keeping the solution for more than 30 min.

Detector fabrication

FAPbBr₃ single crystals with approximate dimensions of 10 × 10 × 2 mm³ were used for device fabrication. The crystals were first ground sequentially using 3 M aluminum oxide/silicon carbide sandpapers with grit sizes ranging from 400 to 8000, progressing from coarse to fine. Fine polishing was subsequently carried out with a commercial polishing compound (CHEMOMET, Buehler). Residual particles on the polished surface were removed by rinsing with chlorobenzene, followed by drying under a nitrogen stream. Surface defect passivation was performed by exposing the crystals to UV light for 10 min. For detector fabrication, an 80 nm Au electrode was thermally evaporated onto one side of the crystal, while an 80 nm Bi electrode was patterned on the opposite side using a mask. Electrical connections to the Bi electrode were established via gold wires. For devices employed in TAS measurements, thinner crystals (<1 mm thickness) were used and coated on both sides with planar 80 nm Au and Bi electrodes. For temperature-dependent current-voltage characterization, devices were fabricated by depositing interdigitated electrodes on a single side through a shadow mask, with an electrode spacing of 50 μm.

CCE calculation

We calculate the CCE of the FAPbBr₃ single-crystal devices based on the collected charges generated by γ-rays with a known deposited energy. The CCE is calculated by dividing the measured CN of the photopeak by CN₁₀₀.

γ-Ray spectral performance measurement and fitting procedure

For measuring the spectroscopic performance of the detector, we applied 0.8 μCi ²⁴¹Am 59.5 keV, 5 µCi ⁵⁷Co 122 keV, and 0.24 µCi ¹³⁷Cs 662 keV as the γ-ray source. As-prepared detectors were put in an aluminum box (Faraday cage) to shield noise, which was connected to a charge-sensitive amplifier (EV-550, Kromek). The high bias voltage was passed through EV-550 and applied to the detector. The shaping amplifier (972A, AMETEK) and multichannel analyzer (EASY-MCA 8k) were applied for signal processing and γ-ray spectrum generation. The shaping time used in the spectrum measurement was 2 µs. All Detectors have a similar thickness of 2 mm. To minimize fitting distortions arising from statistical fluctuations in the data, we performed repeated fittings using the Gaussian function and extracted the corresponding uncertainties (error bars) for each fitting result. The complete set of fitting parameters is summarized in Table S2 and Supplementary Fig. 43.

FLIM imaging and merged fluorescence-intensity mapping methods

FLIM imaging was performed on a FLIM system (MicroTime100, Picoquant). A pulsed laser operating at 405 nm with a repetition rate of 5 MHz served as the excitation source. Reflected laser light was blocked using a 405 nm dichroic mirror. The thin-crystal growth process was monitored in situ under ambient atmospheric conditions using a custom-designed stage equipped with a quartz viewing window and precise temperature control. Fluorescence was collected through a 50× objective (Olympus LMPlanFL N, NA = 0.50) and directed to a photon-counting detector (PMA Hybrid, Picoquant). To separate the solution-derived fluorescence (peak near 432 nm) from the crystal-derived fluorescence (peak near 540 nm), a monochromator (Picoquant) was used to isolate emission bands centered at 432 ± 20 nm and 540 ± 20 nm, respectively.

For merging fluorescence-intensity mapping, dual-channel in-situ FLIM data (432 nm and 540 nm) were exported as 200 × 200-pixels ASCII matrices (.dat) and merged in Python 3.9. A custom script reads the first 200 valid rows after skipping the three-line header, decodes each line (UTF-8 with Latin-1 fallback), and converts the entries to float. The two arrays are superimposed with adjustable transparency: the 432 nm signal is rendered with the “viridis” color map (alpha = 0.1–0.9), whereas the 540 nm signal is rendered with the “plasma” color map (alpha = 0.1–0.9). Five composite snapshots with alpha₂ = 0.1, 0.3, 0.5, 0.7, and 0.9 are automatically saved as 150 dpi PNG files; an interactive figure with live sliders (matplotlib 3.7) allows real-time optimization of the relative weighting. Contour and 3-D surface plots of either channel can be generated analogously for quantitative topography analysis. All the overlaid PL mapping images are 80 × 80 μm² in size.

In situ solution FLIM test

In situ solution FLIM imaging was performed on a FLIM system (MicroTime100, Picoquant). A pulsed laser operating at 405 nm with a repetition rate of 5 MHz served as the excitation source. The solution was heated from room temperature (25 °C) to 55 °C, increasing by 10 °C every 30 min. At each temperature, it was kept warm for 10 min, and PL mapping tests were conducted.

ESI-MS measurement

Take 5 mg of FABr, 100 µL of DMF, and 50 µL of PC, and dissolve each in 1 mL of ethanol to prepare standard solutions of the corresponding cations. Subsequently, 50 µL of the precursor solution is mixed with 1 mL of ethanol. After allowing sufficient time for precipitation, the mixture is filtered through a 0.2 µm PTFE membrane to obtain the test solution. Finally, the sample is analyzed using an Agilent 6545 Q-TOF mass spectrometer.

μτ product measurement

The μτ value was estimated by fitting the Hecht equation: $\eta =\frac{Q}{Q_0}=\frac{\mu \tau V}{d^2}\left(1-e^{-\frac{d^2}{\mu \tau V}}\right)$, where η is the CCE; Q and Q₀ are the maximum and theoretical saturated channel numbers of photopeak/shoulder, respectively; V is the bias; and d is the detector’s thickness.

TOF measurement

We put a 280 nm laser source with a detector in an aluminum box to shield from noise. The detector was biased, and the transient waveforms were recorded by the oscilloscope (Rigol DS1202Z-E). The mobility µ was calculated by recording pulse height using an oscilloscope. We used the average rising time under different biases to get the mobility according to this equation: $\mu =\frac{V_{dr}}{E}=\frac{L^2}{V\times t_r},$ where $V_{\text{dr}}$ is the drift velocity, $E$ is the electric field, $V$ is the applied bias, $L$ is the device thickness (2 mm), and $t_{\text{r}}$ is the carrier drift time. The drift time $t_{\text{r}}$ was extracted from the time interval during which the signal waveform rises from 10% to 90% of its full amplitude.

TAS measurement

An LCR (inductance (L), capacitance (C), and resistance (R)) meter (Agilent E4980A) was used to perform the TAS measurement for the trap density of states (tDOS) derivation. The d.c. bias (V) was fixed at 0 V, and the amplitude of the a.c. bias (δV) was 20 mV. The scanning range of the a.c. frequency (f) was 0.02–2000.00 kHz. The tDOS (N_t (E_ω)) is calculated using the equation $N_t=-\frac{V_{bi}}{qAW}\frac{dC}{d\omega }\frac{\omega }{k_{B}T}$. Here, q, k, T, ω, and C are the elementary charge, Boltzmann’s constant, temperature, angular frequency, and specific capacitance, respectively.

Other characterizations

The X-ray diffraction measurements were performed on a Malvern Panalytical X’ PERRT 3 MRD instrument with a rotating Cu Kα anode X-ray source at 40 kV and 30 mA. High-resolution X-ray rocking curves were also performed on a Malvern Panalytical X’ PERRT 3 MRD system. Ultraviolet-visible absorption measurements were recorded using a Shimadzu UV-2600i Plus instrument. The steady-state PL and TRPL spectra were measured with a PicoQuant FluoTime 100 spectrometer. A 405 nm pulsed laser with a repetition rate of 0.5–20.0 MHz was applied. The current density–voltage characteristics were recorded by a Keithley 2400 source meter in the dark. GIWAXS measurements were carried out with a Xeuss 2.0 SAXS/WAXS laboratory beamline using a Cu X-ray source (8.05 keV, 1.54 Å) and a Pilatus3R 300K detector.

Author contributions

Z.N. and D.Y. conceived the idea and supervised the project. Z.N. and Z.L. designed the experiments. Z.L. synthesized the crystals, fabricated the devices, and measured them. L.L. and T.Z. measured the TRPL data and provided helpful suggestions about crystal growth conditions. H.Z. carried out the PL Mapping measurements. F.Z. performed the refactoring of the solution PL mapping and the overlapping maps. L.W. and Y.F. provided helpful suggestions about γ-ray spectral performance. J.S. provided the schematic drawing. P.C., H.Z., and X.L. carried out the GIWAXS measurement. Z.L. and Z.N. wrote the paper, and all authors reviewed the paper.

Peer review

Peer review information

Nature Communications thanks Jinsong Huang and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the National Natural Science Foundation of China (Nos. 52303254 and 62474156), and the Baima Lake Laboratory Joint Fund of the Zhejiang Provincial Natural Science Foundation of China under Grant No. LBMHZ25F050001, and Central Guidance Funds for Local Science and Technology Development Projects (2025ZY01012), and the Fundamental Research Funds for the Central Universities (226-2026-00077).

Data availability

The main data supporting the findings of this study are available within the Article and the Supplementary Information. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76579-0.