Halide perovskite single crystals arouse broad attention, since they can be grown from low-cost solution processes with great potential for future commercialization in optoelectronic devices, especially for radiation detectors. The unique physical and chemical properties of strong stopping power, absence of deep traps, large mu-tau [mobility (μ) and lifetime (τ), μτ] product, and easy crystallization from low-cost solution processes make perovskite single crystals suitable for next-generation ionization detection materials. The high performance and bright future of perovskite single crystals as X-ray imaging detectors, gamma ray detectors, and scintillators will be discussed in the review. While various methods appeared for growing perovskite single crystal, the crystalline processes are in line with traditional crystallization kinetic theory (Zhao et al., 2018; Yin et al., 2019), basically including nucleation and crystal growth. And the nucleation process, which is of paramount importance for the crystal quality, occurs at the beginning of the entire crystalline process. Crystalline speed and static equilibrium stability of saturation solution are the points for growing a high-quality and large-scale perovskite single crystal. As of now, various growth methods for perovskite single crystals with different composition, size, and exposed facets have been already reported. Herein, perovskite single crystals growth methods, including aqueous solution crystallization by cooling method (Yin et al., 2019), traditional high temperature Bridgman method (Stoumpos et al., 2013a), inverse temperature crystallization method (ITC) (Kadro et al., 2015; Saidaminov et al., 2015; Abdelhady et al., 2016; Zhumekenov et al., 2016), and anti-solvent vapor assisted method (ASV) (Shi et al., 2015; Rakita et al., 2016; Wei et al., 2016). The distinguished advantages of these methods will be discussed in detail.

Halide perovskite single crystal was first grown by gradually cooling the temperature of saturated solution from aqueous acid solution (Baikie et al., 2013; Stoumpos et al., 2013a; Pisoni et al., 2014; Dang et al., 2015). In the crystallization process, slowing down the cooling rate normally leads to larger single crystals with fewer seed crystals and higher quality single crystal with fewer defects (Figure 1A), which enables larger crystals dimensional sizes, smoother morphology, and a better optoelectronic performance (Yin et al., 2019). Similar to the controlled cooling processes, perovskite single crystals can be also grown from all-solid raw materials by the traditional high temperature Bridgman method (Stoumpos et al., 2013b; He et al., 2018). Despite the long amount of time and high energy consumption involved, raising temperature above solid's melting point is suitable for raw materials with low solubility, like all-inorganic perovskite. This method was always used to grow large-sized CsPbBr₃ single crystal serving as radiation detection (Stoumpos et al., 2013a; He et al., 2018; Figure 1B). Furthermore, perovskite materials are discovered to represent good solubility in highly polar organic solvent, such as N, N-Dimethylformamide (DMF), Dimethyl sulfoxide (DMSO), etc., and presents an inverse temperature crystallization process, which means lower solubility in higher temperatures within a certain temperature range. This crystallization process was so-called inverse temperature crystallization method (ITC); the crystallization velocity is an order magnitude faster than that of high temperature Bridgman method. ITC method attracted intensive investigations on perovskite crystallization.

ITC used for crystals with retrograde solubility allows the saturated solution to reach equilibrium and to be filtered at room temperature, as a result, maximizing the operation window, first. Then perovskite single crystals with inverse solubility can be grown at an elevated temperature, and growth velocity is always finely controlled by temperature (Figure 2A). A great deal of perovskite single crystals by ITC method have been reported as exhibiting excellent properties with fast growth velocity, tunable shape, and high optoelectronic performance, when compared with traditional crystal growth methods; and ITC method has been employed for kinds of high-quality perovskite crystals, such as methylammonium lead triiodide (MAPbI₃), methylammonium lead tribromide (MAPbBr₃), formamidinium lead triiodide (FAPbI₃), etc. Figure 2B shows different perovskite crystallization processes in which a crystal seed is added into the saturated precursor solution by ITC method, and the crystal grows at faster crystallization rate, resulting in larger crystal size. This is mainly caused by the critical Gibbs free energy G^* required for nucleation (Yang et al., 2015, 2017; Ummadisingu et al., 2017). Thus, when a seed occurs in the precursor solution (region II shown in Figure 2B), it can induce spontaneous crystal growth, shorten growth time, and result in a large-sized single crystal (Saidaminov et al., 2015; Han et al., 2016), otherwise, the crystal will disintegrate spontaneously in region I. Nevertheless, a continuously increased temperature during ITC method will inevitably cause the decomposition of perovskite and reduce the atomic utilization efficiency of raw materials.

Anti-solvent vapor-assisted crystallization (AVC) method can greatly improve the atomic utilization efficiency of raw materials. In this method, anti-solvent will gradually diffuse into perovskite solution and precipitate the crystals (Figure 2C). The crystal growth velocity is determined by the diffusion rate of anti-solvent, as well as the corresponding solubility difference in diverse mixture solvent. Another advantage of AVC method lies in that high-quality crystals can be grown at room temperature with low energy consumption (Rakita et al., 2016; Liu X. et al., 2018), which is more suitable for all-inorganic CsPbX₃ (X = Cl, Br, I) single crystals when compared to the traditional Bridgman and ITC processes. When mixing equal equivalent CsBr and PbBr₂ into the solvent, non-stoichiometry crystal nucleus easily appears first in the precursor solution, owing to the mismatched solubility of PbBr₂ and CsBr. Hence, to get single crystals with stoichiometric ratio, non-stoichiometry crystal nucleus is normally precluded first before entering into the crystal growth process. Then, the rest of the precursor solution with non-equivalent raw materials is filtered to grow crystals with CsPbBr₃ composition. Moreover, another approach to growing CsPbBr₃ single crystals is by directly changing the feeding ratio of the raw materials—the optimized ratio for CsBr:PbBr₂ being 1:1.5 (Zhang et al., 2017). However, the limitation of AVC methods lies in that it is not easily scaled up in industry, and the reproducibility is not as promising as other methods.

Based on these synthesis methods, different kinds of high-quality perovskite single crystals are grown from solution with the aim to understand the physical and chemical properties of perovskite materials for potential optoelectronic applications. Nevertheless, the applications of perovskite materials in optoelectronic devices are mainly focused on the thin film devices rather than single crystal devices, which are basically limited by their thickness and scaling up problem. A thick semiconductor layer often results in a high potential of charge carrier recombination for solar cells devices and a high possibility of reabsorption of the radiative emission from deep inside for light-emitting diodes, since the photo-electric conversions of both these devices are related to the low-energy UV-visible photons. However, perovskite single crystals are actually appealing for high-energy photon detection. In this sense, the thickness of single crystal is an advantage for high attenuation efficiency, and the charges generated inside the crystal can be extracted by applying a bias voltage on the opposite electrodes.

Semiconductor working in direct detection mode normally possesses high sensitivity and high spatial resolution of images. However, there are limited choices of semiconductors for direct detection mode, since it requires semiconductors of high atomic numbers for strong stopping power, large μτ product for high charge collection efficiency, low trap density for less noise and less non-radiative recombination. Halide perovskite single crystal has emerged as a new generation of ionization detection materials working in direct detection mode (Wei et al., 2016; Wei and Huang, 2019), which can directly convert the ionizing charges into readable current by driving the charges with a bias on opposite electrodes. Firstly, perovskite single crystals are easily grown from low-cost solution processes. The estimated price of a perovskite single crystal is about US$ 0.5–1.0 per cm³, when combined with the cost of raw materials and fabrication process (Yakunin et al., 2016). Secondly, perovskite single crystal advances in device performance as a candidate of radiation detector in terms of sensitivity, response speed, imaging spatial resolution, etc. The high atomic number of the elements Pb, Cs, Sn, I, and Br and the large density of perovskite result in comparable attenuation coefficient to CdZnTe, which is much higher than α-Se or Si semiconductors shown in Figure 3A (Huang et al., 2019). This is also the reason that Si cannot be employed for high-energy ionization detection material. Once again, the large μτ product and low trap density ensure a high charge collection efficiency under large bias, although there is polarization effect remaining unsolved for halide perovskite materials. Lastly, halide perovskite comprises a large family, including 3D materials with ABX₃ structure, 2D materials with R₂A_n−1B_nX_3n+1 structure, and double perovskite with A₂MM′X₆ structure. All of the A⁺, B²⁺, M⁺, M^′3+, and R⁺ ions are variable by following the Goldschmidt tolerance factor principle to stabilize the perovskite phase (Goldschmidt, 1926), which is closely related to the ion's radius. A large variety of chemical and physical properties can be obtained from perovskite by fine-tuning the ions and structures.

MAPbBr₃ single crystal has been demonstrated to be a promising X-ray imager, exhibiting excellent performance (Yakunin et al., 2015; Wei W. et al., 2017; Liu Y. et al., 2018). In addition, MAPbBr₃ single crystal was also successfully integrated on Si devices showing a sensitivity to X-rays over 1,000 times greater than that of α-Se detectors and an ability to image at a 15 to 120-fold weaker X-ray flux, as depicted in Figure 3C. The device can detect a very low dose rate of <0.1 μGy_airs⁻¹ with a high sensitivity of 2.1 × 10⁴ μC Gyair-1cm⁻² to 8 keV X-rays, providing a strategy to obtain active-matrix flat-panel imagers for further commercialization application as X-ray detector in medical and security check (Wei W. et al., 2017). Also, lead-free perovskite single crystals have developed rapidly in recent years, and Cs₂AgBiBr₆ single crystal detector has realized a minimum detectable dose rate as low as 59.7 nGy_airs⁻¹, comparable to the reported MAPbBr₃ (Pan et al., 2017). And, more stable 2D lead-free double perovskite single crystals of (BA)₂CsBiBr₇ and (NH₄)₃Bi₂I₉ have minimum detectable dose rate of 4.5 μGy_airs⁻¹ and 55 nGy_airs⁻¹, respectively, which are also comparable to that of MAPbBr₃ single crystal (Xu et al., 2019; Zhuang et al., 2019).

The decay of the majority of radioactive isotopes involves the irradiation of gamma (γ) photons with energies of ~200 keV to 10 MeV. Low-cost and highly sensitive detectors, which can operate at ambient temperature, are of critical importance for many applications, such as homeland defense, nuclear reaction inspection, research field, etc. (Milbrath et al., 2008; Sordo et al., 2009). Perovskite semiconductor is one type of promising radiation detection candidates with μτ product of 1 × 10⁻² cm² V⁻¹, stopping power, and bulk resistivity comparable or even better than state-of-the-art detector material Cd_1−xZn_xTe (x <20%, denoted as CZT) or thallium (I) bromide (TlBr) devices. Furthermore, halide perovskite crystals can be grown from low-cost solution processes based on earth-abundant raw materials, which is more suitable for industrial production when compared to the high-temperature Bridgeman methods. Perovskite single crystals with various composition (MAPbI₃, FAPbI₃, CsPbBr₃, I-treated/Cl-treated MAPbBr₃) were applied to serve as solid-state gamma-detecting materials with comparable energy resolution to standard scintillator detectors (Yakunin et al., 2016; Wei H. et al., 2017; He et al., 2018; Wei and Huang, 2019), owing to their superior device signal/noise ratio, which is benefited from a large μτ product of 1.0–1.8 × 10⁻² cm² V⁻¹ (Yakunin et al., 2016; Wei H. et al., 2017) and a low dark current and noise. All of these enable a promising device response to 59.5 keV ²⁴¹Am γ-source, 122 keV ⁵⁷Co source, and 662 keV ¹³⁷Cs source with a sufficient energy resolution of 3.9% (Figure 3B), which is comparable to the commercial NaI (Tl) scintillator. However, the ion migration phenomenon can be still observed in these detection devices, resulting in continuous dark current drift and degraded energy spectrum resolution (Luo et al., 2017).

Scintillator is another kind of ionizing detector working in indirect detection mode, which converts high-energy X-ray photons to low-energy UV-visible photons upon irradiation. A sensitive photodetector is integrated with the scintillator to detect the emitting UV-visible photons (Figure 4A). Therefore, it is important for a scintillator to have: (1) thick single crystal with strong stopping power to fully attenuate the high-energy photons, and transparent enough to emit light with less scattering; (2) large light emission yield to convert the irradiation energy to detectable energy as much as possible; (3) large stokes shift to avoid reabsorption of the emission from the thick scintillator crystals; and (4) fast radiation recombination speed to reduce the response time. In order to satisfy these requirements, scintillator can be composed of 2D materials or quantum dots materials to confine the ionizing charges around a local area to increase the radiative recombination velocity (Zhou et al., 2017, 2018). Foreign ions are often doped into the bulk single crystals to act as radiative emission center, which is a general strategy to enlarge the stokes shift and increase the light emission yield (Cavouras et al., 1996; Uchiyama et al., 2001).

MAPbI₃, MAPbBr₃, and 2D (EDBE)₂PbCl₄ [EDBE = 2,2′-(ethylenedioxy)bis(ethylammoniu-m)] (Figure 4B) single crystals have been studied for scintillator application, and photoluminescence (PL) peak of MAPbI₃ single crystal is centered at 780 nm. Meanwhile, PL peaks of MAPbBr₃ and 2D (EDBE)PbCl₄ locate at 550 and 520 nm, respectively (Figure 4B), which can match the common sensitive photodetector (Birowosuto et al., 2016). The bandgap emission of perovskite single crystal scintillator can be easily optimized through engineering of the composition. 2D perovskite structure with long chain organic spacer shows higher light yield (Li et al., 2019; Figure 4C). Recently, Rb₂CuBr₃ perovskite single crystal showed promising scintillator properties with a light yield of over 9 × 10⁴ Photons/MeV, higher than traditional bright scintillators of CsI:Tl (5.4 × 10⁴ Photons/MeV) and alginate-derived guluronate oligosaccharide (GOS) (6.0 × 10⁴ Photons-/MeV) (Figure 4D), and reaching a low X-ray detection limit of 121.5 nGy_airs⁻¹ (Yang et al., 2019; Figure 4E).

Similar to the semiconductor in direct detection mode, scintillator single crystals also need a low trap density to avoid the non-radiative recombination in the scintillator material, which often leads to a slow components in scintillation decay and afterglow. Therefore, low trap density single crystals are also necessary for high-performance ionization detection. However, the fluorescence from emission center in single crystal is isotropic, which normally ruins the detectable light emitting yield when there is an absence of an integrating sphere and also results in a low spatial resolution by mutual interference with neighboring pixels on a flat panel (Grim et al., 2012; Heo et al., 2018; Burdette et al., 2019).

Defects, which exist at both surface and bulk of perovskite single crystal, are detrimental for improving device performance in most cases despite absence of deep traps in halide perovskite materials (Shao et al., 2014; Wang et al., 2018; Chen et al., 2019). Passivating the defects in perovskite is a necessary requirement for high-performance devices. MAPbI₃, as a typical material in the halide perovskite family, is still rich of shallow defects such as Pb²⁺ interstitial and I⁻ vacancy, and so on. To reduce the intrinsic bulk defects in MAPbI₃ single crystal, large-sized A site cation was doped inside to tune tetragonal phase to more stable cubic phase (Figure 5A) by balancing opposite lattice distortion strains (Mitzi, 2001; Stoumpos et al., 2013b; Peng et al., 2016). Furthermore, long-chain organic cations as A site can result in the so-called quasi-2D perovskite structure, which is more ambient stable compared to the 3D structure halide perovskite (Peng et al., 2016). Moreover, 2D structure can also decrease electron–phonon coupling strength and increase material defects formation energy (Huang et al., 2019). Composition engineering, especially introducing large-sized cations of EA (ethylammonium), DMA (dimethylammonium), and GA (guaniddinium), can obviously increase the charge carrier lifetime and decrease the defect density (Figures 5B,C), i.e., increased μτ product (Peng et al., 2016; Huang et al., 2019), which is of great significance for radiation detectors. In addition to intrinsic bulk defects, crystal surface is also rich in shallow traps. UV-O₃ treatment was employed to passivate the defects on MAPbBr₃ single crystal top surface (Wei et al., 2016), revealing the Pb²⁺ dangling bonds passivated by forming oxygen–lead bonds after UV-O₃ treatment (Figures 5D,E).

Another obstacle that should be faced for perovskite X-ray detector is the ions migration phenomenon, which normally results in the polarization effect in ionization detection (Eames et al., 2015; Luo et al., 2017; Wei and Huang, 2019). Suppressing the ions migration is challenging in perovskite ionizing detector since a large bias is often applied, and the ionic intrinsic properties of perovskite determine the low activation energy of mobile ions. One strategy to suppress the ions migration is to block the ion mobile path and elevate the corresponding activation energy. Low dimensional perovskite single crystal has long-term stability under operation condition (Smith et al., 2014; Cao et al., 2015; Tsai et al., 2016; Yuan et al., 2016), and the generic structural formula of 2D perovskite is A₂B_n−1M_nX_3n+1 (n is integer), A⁺ is a primary aliphatic or aromatic alkylammonium cation, M²⁺ is a divalent metal, and X⁻ is a halide anion. [MX₆]⁴⁻ forms octahedra frame structure, suitably sized B⁺ cation insert into the octahedra lattice, and bulky A⁺ cations form an organic spacer existing between the semiconducting inorganic layers, which can effectively suppress the ions migration (depicted in Figure 5F).

As one of the high-performance semiconductors, perovskite is proposed to forge a bright future in the radiation detection field for its earth-abundant raw materials, low cost, feasible fabrication, and high-performance characteristics. However, the semiconductor's long-term instability limits its future application in many fields. To improve the long-term stability, researchers have reported various improvement strategies. 1, tuning the ratio of halide ions, such as chlorine incorporation into MAPbBr₃ and bromine incorporation into MAPbI₃, this can obviously improve the moisture stability of the perovskite (Noh et al., 2013; Wei H. et al., 2017; Boyd et al., 2019; Hieulle et al., 2019). 2, changing the halide into halide substitution, like totally or partly replacing halide with thiocyanate (SCN) ion (Jiang et al., 2015). 3, doping additives (i.e. quantum dots, polymer material and small organic molecules) in precursor (Wang et al., 2017; Hieulle et al., 2019; Yao et al., 2019). What‘s more, 2D perovskites [(A)₂(B)_n−1M_nX_3n+1 (n is integer)] have emerged as a more moisture stable material, and researchers designed more inter-molecular interactions in 2D perovskites by tunneling the A-site molecule, which have further improved the stability. On the other hand, researchers have reported lead-free perovskite as non-toxic substitutes, but the performance is still limited, so an urgent objective is to improve the performance of 2D perovskite and lead-free perovskite devices. Obtaining high-quality perovskite crystal materials is crucial for better device performance and stability, and the means of attaining a perovskite crystal with fewer defects and of a larger size remains obviously unsolved. More efforts are still needed to improve the synthesis techniques of the perovskite single crystals and to boost the performance of promising ionization detection applications.

HW and WP wrote the manuscript. All authors contributed to the commenting and review of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.