Recently, all inorganic lead halide perovskites (CsPbX3, X=Cl, Br, I) have received tremendous attention in optoelectronic devices because of their advantages including a large light absorption coefficient, low fabrication cost, tunable bandgap, and high carrier mobility [1–3]. The most interesting feature of this group of material is the ability to obtain high optical gain in the whole visible range, making them highly desirable in tunable lasers, broad band amplifiers, and response band adjustable photodetectors [4–6]. Generally, mixed halide perovskites (CsPbClxBr3−x or CsPbBrxI3−x) can be prepared by adding or mixing appropriate PbX2 salts as precursors; thus the bandgap is correspondingly modulated from near ultraviolet (UV) to the red spectrum through fast anion exchange [7]. Huang et al. have synthesized a series of CsPbClxBr3−x (x=0–2.5) samples with emission wavelength from 525 nm to 430 nm through a solution method [8]. A vapor-assisted post-synthesis chlorination procedure is another efficient strategy to convert the green emitting CsPbBr3 into blue emitting CsPbClxBr3−x through controlling the conversion temperature and time [9]. Typically, the low-dimensional structures like quantum dots, nanocrystals, and microplates are promising in realizing the low-threshold amplified spontaneous emission (ASE) because of the quantum confinement effect [10]. CsPbClxBr3−x quantum dots and nanofilm have proved their ASE behavior with pumping thresholds of 20–56μJ/cm2, and the ASE peak centers are tuned to the blue band (440–495 nm) [11–13]. Compared with the low-dimensional structures, a large sized crystal is more difficult in the realization of ASE but has more potential in practical application. Kim et al. report the one-photon pumped ASE in CsPbBr3 single crystals, while the limited penetration depth and reabsorption of the crystal induce a large threshold of 1.38mJ/cm2 [14]. Zhao et al. reduce the ASE threshold (0.65mJ/cm2) in millimeter sized CsPbBr3 crystal using two-photon excitation, but the full width at half maximum (FWHM) is as large as 7 nm [15]. To date, studies regrading blue band ASE in millimeter sized lead halide perovskites are still rare. On the other side, as another kind of optoelectronic device that can convert the light into electrical signal, an all inorganic lead halide perovskite (CsPbX3, X=Cl, Br, I) photodetector plays a key role in optical imaging, environmental monitoring, forest fire alarms, and space secure communication [16–18]. However, a strong and narrow response band is normally observed in single crystal perovskites [19,20], which can be attributed to the CCN mechanism [21]. Therefore, it is highly desired if a low-threshold ASE and broadband photoresponse can be simultaneously realized within a large sized CsPbX3 crystal.

In this work, millimeter sized CsPbClxBr3−x crystals are synthesized through a simple anti-solvent process at room temperature. High surface defect density is observed and determined in the as-prepared sample with low photoluminescence quantum yield (PLQY) of only 0.35%. Nevertheless, ASE behavior still can be observed with a relatively low threshold of 106μJ/cm2 at 50 K, which verifies the high defect tolerance of the all inorganic perovskite. In addition, compared with the emission center of CsPbBr3 crystal, the ASE emission peak blue shifts from ∼529 to ∼456nm because of the anion exchange of Br by Cl. Subsequently, a symmetric MSM type photodetector based on CsPbClxBr3−x is prepared with InGa as the Ohmic contact. At 10 V, the photodetector demonstrates good performances with a responsivity of 3.95 mA/W, a detectivity of 7×1010 Jones, and fast response speeds of 96 μs (rise time) and 34 μs (decay time).

Cesium bromide (CsBr, 99.9%), cesium chloride (CsCl, 99.9%), and lead bromide (PbBr2, 99.9%) were purchased from Alfa Aesar (China) Co., Ltd. Dimethyl sulfoxide (DMSO, 99.5%), methanol (99.5%), and ethanol (99.5%) were purchased from Sinopharm Chemical Reagent Co., Ltd. All reagents and solvents were used without any purification.

The CsPbClxBr3−x sample was synthesized by an infiltrating inverse solvent re-crystallization method. 0.93 g CsBr, 0.93 g CsCl, and 0.93 g PbBr2 were simultaneously loaded in 8 mL DMSO solution and then stirred for 1 h. After precipitating for 1 h, the supernatant was transferred into a vial, and then the vial was loaded inside a beaker filled with methanol. Subsequently, the beaker was sealed by paraffin film in order to maintain the methanol atmosphere. The growth temperature is at room temperature, and yellow CsPbClxBr3−x crystals were obtained after 5 days. Finally, the CsPbClxBr3−x crystals were washed by ethanol and dried in the oven at 60°C for 30 min. It is worth noting that the repeatability of the method at a small-batch scale (gram scale) is good. CsPbClxBr3−x crystals with the similar size and the same component ratio can be collected within different batches under the same growth parameters. According to a previous study on the production of kilogram-scale PbI2 crystals using the similar solution method [22], large-scale production of CsPbClxBr3−x crystals for practical applications is expected with the upgrade and expansion of the synthesized facilities.

The XRD pattern of the CsPbClxBr3−x sample was collected using a multifunctional X-ray diffractometer (XRD, Bruker, D8 Advance) with a Cu−Kα line (1.54 Å). The Raman spectrum of the CsPbClxBr3−x sample was excited by a 532 nm laser and collected through a back scattering configuration. The room temperature photoluminescence (PL) spectrum was measured to study the luminescence behavior of the CsPbClxBr3−x sample. The photoluminescence quantum yield (PLQY) was determined (Edinburgh Instruments, FLS-1000) to study the PL efficiency of CsPbClxBr3−x. X-ray photoelectron spectroscopy (XPS, Thermo Fisher Scientific, ESCALAB XI) and ultraviolet photoelectron spectroscopy (UPS) were used to investigate the chemical state and energy band structure.

A 355 nm laser (100 fs, 1 kHz) with different densities was used as the excitation source to excite the CsPbClxBr3−x sample. The measurements were conducted at both room temperature and 50 K. Temperature-dependent PL spectra under excitation density of 5.9μJ/cm2 were also collected. During the measurement, the emission was collected by a spectrometer with resolution of 0.09 nm (Princeton Instrument, SpectraPro HRS-300).

InGa electrodes were employed as the contact electrodes for the CsPbClxBr3−x MSM photodetector. The response characteristics of the device were investigated by using an electrochemical workstation (Ivium Vertex One). A Xe-lamp equipped with a monochromator was utilized as the excitation source. The photoresponse measurements were performed in an air environment with a relative humidity (RH) of ∼57%. The time-resolved photoresponse curve was measured by using a continuous-wave (CW) laser with 442 nm wavelength as the excited source and an oscilloscope as the data collector. The CW laser was chopped and modulated by a programmable module.

The XRD pattern of the as-prepared CsPbClxBr3−x crystal is shown in Fig. 1(a). Eight distinct peaks located at 15.59°, 15.7°, 21.17°, 31.38°, 31.6°, 35.37°, 35.56°, and 38.7° are detected. By comparing with the standard cards of CsPbBr3 (PDF #18-0364) and CsPbCl3 (PDF #18-0366), these peaks can be ascribed to the diffraction from the (001), (100), (101), (002), (200), (102), (201), and (112) facets of CsPbClxBr3−x. The crystal constants of CsPbBr3 (5.827Å×5.827Å×5.891Å) are slightly larger than that of CsPbCl3 (5.584Å×5.584Å×5.623Å). Thus, compared with the standard XRD peaks of pure CsPbBr3, these eight XRD peaks shift to the larger angle side, and compared with the XRD peaks of pure CsPbCl3, they locate at the left angle side. Compared with the extremely high-crystal-quality hybrid perovskite MAPbX3 single crystal [23], the growth parameters including synthesized temperature, nucleation rate, reagent ratio, and additives need to be optimized in order to further improve the quality of CsPbClxBr3−x crystal in the future. The optimized crystal quality can greatly improve the performance of the CsPbClxBr3−x based optoelectronic devices. Figure 1(b) presents the Raman spectrum of the as-prepared CsPbClxBr3−x sample, from which only two distinct Raman peaks centered at 93.6cm−1 and 167.1cm−1 are probed. According to previous Raman studies on CsPbCl3 [24,25], the first vibrational peak is relevant to the Pb(ClxBr1−x)6 octahedron and the second one is associated with the motion of Cs+ cations. Compared with the Raman peaks on CsPbBr3, blue shifts can be determined owing to the substitution of a Br atom by a Cl atom [26,27]. The room temperature PL spectrum of the as-prepared CsPbClxBr3−x crystal is exhibited in Fig. 1(c), from which a broad blue band emission is probed. Two distinct peaks denoted as peak A and peak B are detected; the former one is relevant to the bandgap emission and the latter one is induced by the donor defects. The optical images of the as-synthesized CsPbClxBr3−x crystal are presented in Fig. 7 (Appendix A), suggesting a great number of crystal boundaries located within the surface of the sample. This will introduce a high concentration of surface defects, which supports the result from the PL measurement. It is worth noting that one key advantage in this work is the simple and low-cost equipment; only two beakers with different capacities are required, which makes the equipment easily upgraded. Then the CsPbClxBr3−x crystals can be mass produced once the synthesized processes are mature. In addition, different from previous efforts on the high-quality perovskite crystal, defects and grain boundaries are intentionally introduced in order to study the defect tolerance of the optical pumping property, which will be discussed later. The bandgap of CsPbClxBr3−x at room temperature is estimated as ∼2.76eV (∼448nm) by calculating the wavelength center of the bandgap emission, which is ∼0.45eV larger than the bandgap of CsPbBr3 [28] and ∼0.31eV smaller than the bandgap of CsPbCl3 [29,30]. Quantum efficiency is one important parameter to measure the performance of emitters and the corresponding PLQY result is exhibited in Fig. 1(d). The PLQY of the sample is as low as 0.35%, which is quite similar to previous work reported by Gong et al. [31]; they only detect a low PLQY of 1% in a CsPbBr3 single crystal. Typically, different from CsPbX3 nanocrystals or quantum dots, bulk single crystal usually displays very low PLQY (typically ∼1% or even lower). One reason is owing to the presence of indirect tail states below the direct transition edge caused by Rashba splitting in bulk CsPbBr3 single crystal [32]. The other reason of the low PLQY is the high defect concentration in our CsPbClxBr3−x crystal. Theoretically, the improvement of crystal quality is one effective strategy for improving the PLQY of CsPbClxBr3−x crystal. A lower defect concentration can naturally decrease the nonradiative recommendation centers, which brings the positive influence on the PLQY. Another strategy for improving the PLQY is decreasing the dimension of CsPbClxBr3−x to quantum size (few nanometers). However, our aim is focusing on the large sized crystal; thus improving the crystal quality of CsPbClxBr3−x crystals will be performed. Four strategies are considered in the future: (i) the synthesized temperature is enhanced from room temperature to 60°C–80°C, which provides sufficient thermal energy for the chemical reaction; (ii) a post-annealing process in the inert or halogen/lead atmosphere is considered; (iii) a specific organic molecule is selected and acts as the passivator of defects; (iv) the Bridgman method operated at high temperature is considered for growing high-quality and large sized CsPbClxBr3−x crystal, and then the sample is cut into the targeted size and polished. The high crystal quality will improve the PLQY of the crystal, and then decrease the threshold of the ASE. For photodetectors, undoubtedly, high crystal quality can lower the dark current of the device, and enhance the quantum efficiency and responsivity of the device. Especially, high crystal quality can naturally avoid the effect of trap states on the carrier dynamic, which contributes to the rapid response speed of a photodetector.

Figures 2(a)–2(d) present XPS spectra of Cs-3d, Pb-4f, Cl-2p, and Br-3d core electrons, in which C-1s peak (284.8 eV) is used as calibration. Apparently, all the orbitals can be well fitted with two standard Gaussian peaks, indicating the high-purity phase of CsPbClxBr3−x crystal. The spin-orbital splittings (Cs-3d5/2 and Cs-3d3/2; Pd-4f7/2 and Pd-4f5/2; Cl-2p3/2 and Cl-2p1/2; Br-3d5/2 and Br-3d3/2) are calculated to be 14 eV, 4.87 eV, 1.6 eV, and 1.04 eV, agreeing well with the results from the standard XPS database. The energy band diagram is the key factor for semiconductor devices. The valence band scanning [Fig. 2(e)] and UPS measurement [Fig. 2(f)] are employed to determine the band structure of CsPbClxBr3−x. Deriving from the cutoff and onset of the UPS spectrum, the work function of CsPbClxBr3−x is calculated as ∼3.8eV and the Fermi energy level locates ∼2.22eV above the VBM. The UPS study aligns well with the Fermi energy level (EF−EVBM=∼2.3eV) confirmed by the valence band scanning spectrum. Therefore, the energy band diagram of CsPbClxBr3−x can be summarized in Fig. 8 (Appendix A). The energy difference between VBM and Pb-5d7/2 orbital is 17 eV, and the Fermi energy level locates close to CBM.

Figure 3(a) displays the PL spectra under various excitation densities at room temperature, in which two distinct emissions (peak A and peak B) can be clearly observed. As further summarized in Fig. 3(b), the PL intensity at room temperature can be expressed by a power law (I∝Pα), where I is the PL intensity, P is the excitation density, and α is the exponent factor. The α factors are calculated to be 0.34 (peak A) and 0.45 (peak B), which are smaller than one. Theoretically, when 1<α<2, excitonic recombination dominates; when α is less than one, an impurity or defect (donor and acceptor) is involved in the transitions [33–36]. Thus, the small exponent factors indicate a large proportion of radiative recombination induced by defects. Theoretically, three factors including growth temperature, the ratio of the reaction solutes, and the growth rate (decided by the evaporation rate of methanol) have direct influence on the crystal quality. Low temperature (RT) cannot supply sufficient thermal energy for the migration of atoms, which will easily introduce defects in the crystal. Additionally, the methanol atmosphere can also affect the nucleation rate of CsPbClxBr3−x crystals. The large diffusion rate of methanol atmosphere at a higher temperature will lead to the reduction of the growth time and also affect the crystal quality of the CsPbClxBr3−x crystals. Nevertheless, in this work, CsPbClxBr3−x crystals with a high defect concentration are intentionally synthesized, which is demonstrated by the low α value. As previously mentioned, a high concentration of defects is probed in the as-synthesized CsPbClxBr3−x crystal, agreeing well with the emission spectra at room temperature. Figure 3(c) presents the excitation-intensity-dependent photon energies of peak A and peak B; both peak energies monotonically decrease with the increase of the excitation density. Typically, thermally induced lattice expansion can naturally reduce the bandgap of semiconductors [37,38]; thus the accumulated heat under higher excitation density is responsible for the red shift of the emission peaks at room temperature. Figure 3(d) shows the temperature-dependent PL spectra of the CsPbClxBr3−x crystal at a low excitation (5.9μJ/cm2). With the increase of temperature, a blue shift of peak A and red shift of peak B are observed, which agrees well with previous reports [39,40]. Figure 3(e) summarizes the PL intensity at 5.9μJ/cm2 as a function of the temperature. The PL intensity of peak B becomes larger than that of peak A when the temperature is below 220 K. Figure 3(f) presents the temperature-dependent photon energies of peak A and peak B under the excitation density of 5.9μJ/cm2. A slight blue shift on peak A and distinct red shift on peak B can be probed. The blue shift of peak A (bandgap emission) is unusual by comparing with conventional semiconductors like CdSe [41], ZnO [42], and GaN [43], but it is quite normal in perovskites [44–46]. A blue shift of ∼10meV (peak A) is determined with temperature increasing from 50°C to 295°C. Generally, the temperature-dependent bandgap can be well fitted by a Varshni model [45,47]: Eg(T)=E(0)−(γ×T2)/(β+T),(1)where Eg(T) and E(0) are the photon energies at T and 0 K; γ and β are shift parameters. The Varshni fit yields a γ parameter of −3.74×10−5K−1. The negative value indicates the slight blue shift characteristic, and the β parameter at 300 K is corresponding to the Debye temperature. This behavior can be dominated by the bandgap renormalization effect in perovskite, which is completed by the thermal expansion and the interaction with phonons [48,49]. Normally, thermal expansion in perovskite materials will lead to the blue shift of the bandgap, while the interaction with phonons will induce the red shift of the bandgap [50,51]. Both volume dilation and excitation of the lattice vibration (phonon) are relevant to the temperature, which can produce the shifts in energy levels and then change the energy bandgap. The electron-phonon interaction can induce the shift of the valence band and conduction band, resulting in a quadratic variation of the energy gap at low temperature and a linear shift at high temperature [50]. The conduction band is estimated from the electron and the filled band should be decided by the hole with a negative sign in the effective mass. For perovskite materials, volume dilation has a higher effect on producing the temperature variation of the bandgap than the lattice vibrations. Therefore, a blue shift induced by thermal expansion dominates over the bandgap renormalization, thus leading to a blue shift of the bandgap as the temperature increases. Undoubtedly, the bandgap blue shift with the increase of temperature has been expected in a great number of perovskite materials.

The schematic diagram of the CsPbClxBr3−x photodetector is presented in Fig. 5(a), from which InGa metal is employed as the bottom electrodes. Figure 5(b) shows the I-V curves of the photodetector under dark and illumination by 400 nm light with different intensities. The nearly linear relationship between applied voltage and dark current indicates the Ohmic contact between InGa and CsPbClxBr3−x. According to previous UPS measurement, the work function of CsPbClxBr3−x is determined to be ∼3.8eV, which is close to that of the InGa electrode (∼4.1eV) [62]. Combining the surface defect states in the sample, the Ohmic contact property can be naturally expected between the metal and semiconductor. Under irradiation of the 400 nm light [Fig. 5(b)], the photocurrent prominently increases, and it is enhanced with the increase of the light intensity. The response speed is also an important parameter for the photodetector, which represents its signal tracking ability. As presented in Fig. 5(c), the response curve is asymmetrical with a rise time of 96 μs and a decay time of 34 μs, which reveals a fast response performance. The 3 dB bandwidth can be further calculated as 140 kHz. Additionally, the decay of the response curve can be well fitted with a second order exponential equation [63]: y=y0+A1e−τ/τ1+A2e−τ/τ2,(3)where y0 is the maximum voltage, A1 and A2 are fitting constants, and τ1 and τ2 are the first order decay time and second order decay time. As exhibited in Fig. 5(d), τ1 and τ2 are well fitted as 25.6 μs and 31.3 μs, which means two physical processes are responsible for the recovery time once the illumination is turned off. Nevertheless, both processes show fast response behavior. As discussed above, the defect energy level of VBr is shallow in CsPbBr3 while that of VCl is relatively deep in CsPbCl3; the deep level of VCl may bring a negative effect on the recovery time of the photodetector. However, in the CsPbClxBr3−x crystal, the deep-level property of VCl will be partly suppressed. Therefore, the fast response speed of the device can still be obtained in the low-quality CsPbClxBr3−x crystals. Typically, three factors including the RC time constant, the transit time, and the excess life time of the trap carriers are responsible for the decay trace [64]. Herein, the RC time constant is originated from the large capacitance from the PN junction or Schottky junction, which can be ruled out in our photoconductive device. The transit time is determined by the space of two electrodes, the applied voltage, and the carrier mobility [65]. According to the parameters of the device, the transit time is estimated as 27.9 μs, which may be the most possible reason for the τ1 value. Additionally, the trapped carriers induced by the defects in the surface or the grain boundaries will lead to the excess life time as the incident light is turned off; this mean life time can be indexed to the τ2 time constant. Therefore, the decay time of the photodetector is contributed by the transit time and the excess life time of the trap carriers induced by defects.

The energy band diagram of the symmetric structure InGa/CsPbClxBr3−x/InGa photodetector can be observed in Fig. 6(a). The Fermi energy level locates ∼0.54eV below the CBM, implying the n-type conductivity of the perovskite. The energy band offset between the work functions of InGa and CsPbClxBr3−x is as small as ∼0.3eV. Considering the energy level broadening and surface state or defect on the surface of CsPbClxBr3−x, the shallow barrier can be easily ignored. So, the Ohmic contact property is probed in Fig. 5(b). Figure 6(b) exhibits the comparison of response spectra between the CsPbBr3 photodetector and the CsPbClxBr3−x photodetector. Apparently, substitution of Br by Cl can widen the bandgap of CsPbBr3; the response spectrum range of CsPbClxBr3−x is correspondingly modulated toward the shorter wavelength region. Additionally, the CsPbBr3 single crystal normally reveals a narrow band (∼30nm) response characteristic, which is explained by the CCN mechanism [21]. Nevertheless, this phenomenon is not probed in the CsPbClxBr3−x photodetector, which harvests broad spectra from UV to the blue band. This may be ascribed to the change of band structure as Br is substituted by Cl. The responsivity under different applied voltages is provided in Fig. 6(c) and the responsivity illuminated by 450 nm light is extracted in Fig. 6(d). At 10 V, the responsivity at 450 nm is calculated as 3.95 mA/W and it linearly depends on the working voltage. Correspondingly, the detectivities of the device under different voltages are further studied and displayed in Figs. 6(e) and 6(f). Similarly, the voltage-dependent detectivity also follows the linear tendency, and the detectivity at 10 V (@450 nm) is 7×1010 Jones. The linear dependency indicates the carrier extracted efficiency is linearly depending on the external bias voltage. Figure 9 (Appendix A) exhibits the external quantum efficiency (EQE) of the device at different bias voltages. At 10 V, the EQE at 450 nm is as small as ∼1.1%, indicating the low collection efficiency of the photo-generated carriers. In addition, the dark current of the device is relatively high and the light-dark current ratio is low. As mentioned above, a high defect concentration and numerous grain boundaries are probed on the as-synthesized CsPbClxBr3−x crystals. The defects in the crystal normally act as donors, which leads to the high background carrier concentration, and the scattering and nonradiative recombination induced by the defects and grain boundaries significantly reduce the EQE of the photodetector. Therefore, the crystal quality of the CsPbClxBr3−x crystal needs to be further improved in order to enhance the photodetection performance. Synthesized parameters such as growth temperature, reagent ratio, and nucleation rate should be carefully considered. During the photoelectronic measurement, the device is exposed in air conditioning with RH of ∼57%. The photodetector CsPbClxBr3−x shows good stability at the air atmosphere (room temperature, RH<∼57%) and the photoresponse performance can be well maintained for over 6 months. However, once the RH of the environment is >60%, the photocurrent of the device slowly attenuates because of the hydrolysis of the perovskite. Owing to the intrinsic ionic feature, the interaction between perovskite and water is the main reason of the degradation of the perovskite devices [66], which is named as the well known instability in halide perovskites. For example, though a capping layer is introduced for isolating the water, the power conversion efficiency of the perovskite is reduced to 23.5% at 80% RH, which is lower than that at 20%–60% RH [67]. In addition, the oxidation of the surface is another unfavorable factor that can also degrade the photoelectronic performance of the perovskite devices, while the oxidation process typically needs an extremely long time at room temperature in air atmosphere. Therefore, water is the most unstable factor for perovskite devices. Industrial grade packaging with resin or a polymer layer can effectively protect the perovskite active layer and prolong the life time of the perovskite devices.

Herein, the highly defect tolerant CsPbClxBr3−x crystals are synthesized via a room temperature anti-solvent precipitation process. Spontaneous and amplified spontaneous emissions are both studied under different temperatures and pumping densities. The blue band ASE emissions centered at ∼456nm are probed with a relatively low threshold of 106μJ/cm2 at 50 K. Though the as-prepared CsPbClxBr3−x crystal reveals a high defect concentration, ASE is still realized between the competition of radiative recombination and defect-related nonradiative recombination, indicating the highly defect tolerant CsPbClxBr3−x as a promising material for optoelectronic devices. Furthermore, an MSM structure CsPbClxBr3−x photodetector is prepared and shows a fast response performance with response speeds of 96 μs (rise time) and 34 μs (decay time). Our findings enhance the understanding of the light amplification properties in all inorganic perovskite, as well as the fabrication of a next generation high-performance photodetector.

Acknowledgment. The XRD, PLQY, and XPS data were obtained using the equipment maintained by Dongguan University of Technology Analytical and Testing Center. The authors thank Dr. F. Yi for the helpful discussion on this manuscript.