Pressure-Induced Phase Transition versus Amorphization in Hybrid Methylammonium Lead Bromide Perovskite

The crystal structure of the CH3NH3PbBr3 perovskite has been investigated under high-pressure conditions by synchrotron-based powder X-ray diffraction. We found that after the previously reported phase transitions in CH3NH3PbBr3 (Pm3̅m→Im3̅→Pmn21), which occur below 2 GPa, there is a third transition to a crystalline phase at 4.6 GPa. This transition is reported here for the first time contradicting previous studies, which reported amorphization of CH3NH3PbBr3 between 2.3 and 4.6 GPa. Our X-ray diffraction measurements show that CH3NH3PbBr3 remains crystalline up to at least 7.6 GPa, the highest pressure covered by experiments. The new high-pressure phase is also described by the space group Pmn21; however, the transition involves abrupt changes in the unit-cell parameters and a 3% decrease of the unit-cell volume. Our conclusions are confirmed by optical-absorption experiments, by visual observations, and by the fact that pressure-induced changes up to 10 GPa are reversible. The optical studies also allow for the determination of the pressure dependence of the band-gap energy, which is discussed using the structural information obtained from X-ray diffraction.


INTRODUCTION
Over the last decades, hybrid organic−inorganic perovskites (HOIPs) have experienced a stunning development because they are outstanding light-absorbing materials for highly efficient photovoltaic cells. 1 Among HOIPs, methylammonium (MA, CH 3 NH 3 ) lead halide perovskites, such as MAPbBr 3 and MAPbI 3 , have received the most attention and perovskite solar cells have consequently achieved a photovoltaic efficiency of 25%. 2 The crystal structure of these compounds can be described as an inorganic sub-lattice of corner-sharing PbBr 6 (PbI 6 ) octahedral units hosting the MA molecules. 3 Because of the high compressibility of the inorganic host lattice and the organic molecule, the crystal structure of hybrid perovskites can be easily modified applying an external pressure. 1 In the case of MAPbBr 3 , two phase transitions have been reported below 4.6 GPa. 4−8 Pressure modifies not only the crystal structure but also the electronic properties, including the bandgap energy, thereby becoming a powerful tool to strengthen the current understanding of the properties of MAPbBr 3 as well as other HOIPs, and to provide information relevant for optimizing the photovoltaic performance. 1 Regarding the crystal structure, a recent single-crystal X-ray diffraction study revealed that MAPbBr 3 undergoes two phase transitions following the space-group sequence: Pm3̅ m→Im3̅ →Pmn2 1 . 4 The phase transitions occur at 0.8 and 1.8 GPa, respectively. Reference 4 unveiled the occurrence of the pressure-induced nonpolar/polar transition (Im3̅ →Pmn2 1 ) in MAPbBr 3 for the first time. Amorphization has also been reported to take place in MAPbBr 3 at higher pressures between 2.5 and 4 GPa, and it has been proposed that the difference in amorphization pressure relates to the degree of non-hydrostatic stress in the experiments. 5−7 In the case of MAPbI 3 , a gradual amorphization was believed to occur at 2.7 GPa. 9 However, it was more recently shown that amorphization does not occur in MAPbI 3 up to 10.6 GPa, with a phase transition to a crystalline phase occurring at lower pressures wherein a degree of disorder is introduced into the perovskite lattice by a pressure-induced freeze-out of the MA cation motion. 10 This concept of statistical disorder was confirmed by Raman, infrared, and Xray absorption spectroscopy. 11 Interestingly, the transition reported at 2.7 GPa in MAPbI 3 induces an abrupt increase of the band-gap energy. In this work, the pressure-induced amorphization of MAPbBr 3 has been re-examined by powder X-ray diffraction (XRD) up to 7.6 GPa. Instead of amorphization, we found a first-order isostructural phase transition at 4.6 GPa. The transition favors a smearing of the XRD patterns, but not the loss of crystallinity. These conclusions are supported by optical-absorption measurements and visual observations, which do not show any evidence of amorphization up to 10.6 GPa. The newly reported transition triggers an abrupt increase of 0.27 eV in the band-gap energy.

METHODS
PbBr 2 (98% purity, Fisher Chemical) and MABr (98% purity, Ossila) were dissolved in dimethylformamide (DMF, 99.8% purity, Sigma-Aldrich). The solution was stirred at ambient conditions until the precursors were dissolved. The product was then filtered with a 0.2 mm pore size filter, kept in a closed vial of 20 cm 3 , and heated up to 80°C in an oil bath. Then, it was kept at 80°C for 24 h. Reproducible size crystals are obtained with this method. A fine powder used for XRD measurements was obtained by grinding the resulting singlecrystal sample. Powder XRD experiments were carried out at the BL04-MSPD beamline of ALBA-CELLS synchrotron. 12 A membrane-type diamond anvil cell (DAC), with culets of 400 μm in diameter, was used to generate high pressure. A hole with a diameter of 200 μm drilled in the center of a preindented stainless-steel gasket was used to contain the finely ground powder obtained from the single crystals of the sample. Silicone oil was used as pressure-transmitting medium in order to minimize the possibility of chemical reaction with the sample, 13 and the ruby fluorescence method was used for pressure calibration. 14 In the pressure range covered by the experiments, silicone oil can be considered as a quasihydrostatic pressure medium. 13 The wavelength of the monochromatic X-ray beam was 0.4642 Å, and the spot size of the X-ray was 20 × 20 μm (full width at half maximum). A Rayonix SX165 CCD image plate was used to collect the diffraction patterns. The two-dimensional diffraction images were integrated to conventional XRD patterns using DIOPTAS. 15 The FullProf 16 suit was used to perform Rietveld refinements. 17 The protocol for refinement was as follows. The background of the XRD pattern was fitted with a Chebyshev polynomial function with six parameters, and the shape of Bragg peaks was modeled with a pseudo-Voigt function. The atomic positions and occupancy factors were fixed to the values determined previously from single-crystal XRD experiments. 4 Unit-cell parameters and the overall displacement factor were assumed as fitting parameters. Optical-absorption experiments were performed using the same high-pressure device. In this case, we used small single crystals, cut with a sharp knife from one as-grown crystal, of about 100 × 100 μm 2 in size and 10 μm in thickness. The sample-in and sample-out method was used to acquire the optical absorption spectra using the setup and methods described in our previous work. 18,19

RESULTS AND DISCUSSION
Powder XRD patterns of MAPbBr 3 at selected pressures are shown in Figure 1. The patterns up to 4.6 GPa have been examined in detail in our previous report. 4 In brief, at pressures lower than 0.9 GPa, the data can be well indexed by the ambient-pressure cubic crystal structure (phase I, space group: Pm3̅ m). Above this pressure, and up to 1.8 GPa, XRD patterns can be assigned to the cubic high-pressure structure (phase II, space group: Im3̅ ). Above 1.8 GPa, and up to 4.6 GPa, the XRD data can be satisfactorily explained by the recently proposed high-pressure orthorhombic structure (phase III, space group: Pmn2 1 ). To support this statement, we include Rietveld refinements at 0.1, 1.4, and 2.6 GPa in Figure 1. The goodness-of-fit-parameters obtained are R p = 1.04%, R wp = 1.42%, and χ 2 = 1.02 at 0.1 GPa; R p = 1.17%, R wp = 1.51%, and χ 2 = 1.04 at 1.4 GPa; and R p = 1.80%, R wp = 2.71%, and χ 2 = 1.13 at 2.6 GPa. At the pressure of 4.6 GPa, the observed peaks can also be assigned to the same orthorhombic structure. However, a notable decrease of the intensity of the diffraction peaks is observed. The smearing of the signal is even more significant at 5.1 GPa. At this pressure, there is also the merging of some of the Bragg peaks. Despite the loss of intensity in the XRD pattern, at 5.1 GPa and above, the peaks are sharp enough to support the existence of a crystalline phase with a long-range order (phase IV). In addition, we do not notice any relevant change in the background or the appearance of the typical amorphous halo 20,21 in the diffraction patterns up to 7.6 GPa.
The changes induced in the diffraction patterns at 5.1 GPa do not support amorphization and are more consistent with a gradual disordering of the crystal structure 22 introduced at different length scales in the perovskite lattice by the pressureinduced freeze-out of the MA cation motion as it occurs in MAPbI 3 . 10 Our conclusions are also in line with the study from Yesudhas et al. 23 where the onset of amorphization is set above 5.8 GPa according to Raman experiments. They are also in agreement with recent Raman and photoluminescence experiments that propose a third crystalline−crystalline phase transition and show that the fourth crystalline phase remains crystalline up to circa 7 GPa rather than becoming amorphous. 24 Furthermore, the pressure-induced structural changes observed in powder XRD patterns of MAPbBr 3 are totally reversible upon decompression (see Figure 1), which is unusual for pressure-induced amorphization. 25 Commonly, when releasing pressure after amorphization, the characteristic halo of a non-crystalline material is observed in addition to weak Bragg reflections of the stable polymorph. 25 This is because pressure-induced amorphization is a first-order transition with dramatic changes in the structure of the solids.
For the crystal structure of the new high-pressure phase (phase IV, purple in Figure 1), we have found that a crystal structure described by the same space group as the orthorhombic phase III (cyan in Figure 1, Pmn2 1 ) can account for the peaks in the XRD patterns measured from 5.1 to 7.6 GPa. This conclusion is supported by a Rietveld refinement shown in Figure 1. In this case, the goodness-of-fit parameters obtained are R p = 2.24%, R wp = 3.34%, and χ 2 = 1.02. The unitcell parameters at 5.1 GPa are a = 11.211(5) Å, b = 15.592(7) Å, and c = 7.860(3) Å. However, the atomic positions, which were chosen based on the structure of phase III, were not refined. Single-crystal XRD measurement would be probably needed to unambiguously determine the structural model of phase IV. A thermal annealing of the sample, to suppress/ reduce residual stress after three successive phase transition, would be beneficial to acquire good-quality single-crystal XRD data of phase IV. According to these parameters, there is an abrupt change in a, b, and c at the transition and a 3% reduction in the unit-cell volume. Therefore, the transition can be described as a first-order isostructural transition. Interestingly, b ≃ a 2 ≃ 2c. This makes the new high-pressure structure more symmetric, which is reflected by the merging of Bragg peaks, for instance, the most intense peaks at low angles. The change of unit-cell parameters at the transition will necessarily modify the tilting of PbBr 6 octahedra, reducing the size of cavities containing the MA molecules, which can spatially lock MA molecules in different directions. This local disorder will affect the intensity of XRD patterns. However, it does not destroy the long-range order as would be needed to trigger amorphization.
From powder XRD patterns, we have extracted the pressure dependence of unit-cell parameters. The results are presented in Figure 2. In the figure, vertical dashed lines indicate the transition pressure. However, we cannot rule out the phase coexistence between successive phases, at least in such a short pressure range, due to the pressure steps used in the experiment. In Figure 2a, it can be seen that there is a discontinuity in a, b, and c beyond 4.6 GPa. The discontinuity in the volume is more noticeable (see Figure 2b). In the new high-pressure phase, there is also a decrease of the compressibility. By fitting the four data points, we measured with a second-order Birch−Murnaghan equation of state (EOS); 26 the zero-pressure bulk modulus of phase IV is determined to be 33(6) GPa, which is 70% larger than the same parameter for the three phases measured below 4.6 GPa. 4 A similar reduction of compressibility has been previously reported for highly compressible halide compounds when firstorder isostructural transitions occur. 27 In addition to the XRD experiments, we have carried out a series of optical studies. In Figure 3, we show a collection of images taken under a microscope of a MAPbBr 3 crystal loaded in a diamond-anvil cell at different pressures. Images were captured under compression and decompression. Before the first phase transition (phase I), the color of the sample gradually changes moving from a light-orange color toward dark orange, indicating a red shift of the band-gap energy. After  the transition, and up to 4.1 GPa (i.e., in the first two highpressure phases, phases II and III), the sample gradually becomes light yellow indicating a blue shift of the band-gap energy. From 4.1 to 5.2 (i.e., in phase IV), there is an abrupt color change that corresponds to a strong blue shift of the band-gap energy. In the micrographs shown in Figure 3, it can be seen that there are no obvious changes in the shape or crystallinity of the sample, which is not expected if amorphization occurs between 2.5 and 4 GPa as previously reported. 5−7 Usually, amorphization is characterized by the developing of cracks and other microstructural features in the crystal, which can be visually observed under the microscope and are related to the creation of dislocation densities, which are precursors of the crystalline-to-amorphous transformation. 28,29 None of these features are observed in MaPbBr 3 in our experiments. Upon decompression, the changes in the color of the crystal are fully reversible.
Three independent high-pressure optical-absorption experiments have been carried out to determine the pressure dependence of the band-gap energy of MAPbBr 3 up to 10 GPa. The optical absorption spectra of one of the experiments (exp 2) is shown in Figure 4a. The absorption spectra have a sharp absorption associated with the fundamental band gap. Above 4.6 GPa (phase IV), at energies smaller than the optical gap, we did not observe the residual optical absorption typical of amorphous materials. 30 This supports the conclusions we made from powder XRD experiments. From the opticalabsorption experiments, we obtained the pressure dependence of the band-gap energy summarized in Figure 4b. As observed in Figures 3 and 4, the band-gap energy in phase I continuously red-shifts from ambient pressure until the occurrence of the first phase transition. After the transition, the band-gap energy blue-shifts under increasing pressure up to 4.6 GPa. At the phase transition pressure between phases III and IV, there is an abrupt increase of the band-gap energy of approximately 0.27 eV. Such a change supports the occurrence of a phase transition as determined from XRD experiments at the same pressure. From 4.6 to 10.6 GPa, the band-gap energy remains nearly constant. Consequently, the behavior of the band-gap energy supports the argument that there is no other phase transition up to 10.6 GPa. A similar increase of the bandgap energy has been reported for MAPbI 3 at 3 GPa. 10 However, the phase transition was originally misinterpreted as the pressure-induced amorphization. The analogous behavior of the two lead halide hybrid perovskites provides additional support to our interpretation of the pressure-induced crystal structural changes happening at 4.6 GPa in MAPbBr 3 .
In two of the optical-absorption experiments, we acquired measurements on sample decompression. In one of these experiments, the maximum pressure is 3.8 GPa (exp 3, green circles in Figure 4b), a pressure below that of third-phase transition pressure. In this experiment (green), the observed changes are fully reversible upon decompression. In the second experiment (exp 2, red circles), which includes data acquired upon decompression, the maximum pressure achieved was 10.6 GPa�a pressure that exceeds the phase transition pressure of 4.6 GPa. In this case, the band gap indicates that the sample remains in phase IV under pressure release down to 3 GPa. The subsequent data point has been measured at 0.7 GPa corresponding to phase II. In contrast with compression, during decompression, the pressure increments cannot be finetuned with the same precision. This is why there is a 2 GPa step between subsequent measurements on decompression below 3 GPa. The value of the band gap agrees with the value determined at the same pressure during compression. At ambient pressure, after the pressure is fully released, we measured in both cases (green and red) the same band-gap energy as before compression. This shows that the changes induced by pressure in the electronic band structure are fully reversible, which agrees with the reversibility observed in the structural changes.
Knowing that the valence band maximum of MAPbBr 3 at ambient pressure is mainly contributed to by the Br 4p orbitals and the conduction band minimum by the Pb 6p orbitals, 30 the non-linear pressure behavior up to 4.6 GPa has been explained in our previous study. 4 In particular, the observed behavior is a consequence of different distortions of the inorganic framework, with the decrease of the bond distance of Pb−Br favoring a band-gap decrease and the decrease of the Pb−Br− Pb angle favoring a band-gap increase. 4 The first phenomenon dominates the pressure dependence of the band gap in the lowpressure phase where the Pb−Br−Pb angle is constrained by symmetry to be 180°. The second phenomenon dominates the behavior from 0.9 to 4.6 GPa. In this pressure region, the Pb−

CONCLUSIONS
In conclusion, in this work, we have reported the results of powder XRD and optical-absorption experiments performed on MAPbBr 3 perovskite under high-pressure conditions. The two techniques, and visual observations, show that MAPbBr 3 does not become amorphous under compression up to at least 10.6 GPa. Instead of amorphization, a transition to a crystalline phase has been found at 4.6 GPa. The new phase is proposed to be orthorhombic with space group Pmn2 1 . The phase transition to this new phase induces an abrupt increase of the band-gap energy. Our results contradict previous studies 5−7 but are in agreement with studies on MAPbI 3 , which also ruled out amorphization 10 and found a transition to a crystalline phase at pressures where amorphization was previously reported, showing that the new high-pressure crystalline phase is stable up to 6 GPa. 10 Interestingly in both MAPbBr 3 and MAPbI 3 , the structural phase transitions, which occur instead of the expected amorphization, trigger abrupt blue shifts of the band-gap energy. The qualitative agreement between studies in MAPbBr 3 and MAPbI 3 suggests that the existence of a new high-pressure phase, which could be disordered, and has a wider band gap, could be a common phenomenon in hybrid lead halide perovskites. In fact, this will not be the first case where the absence of a previously alleged pressure-induced amorphization has been reported. 31 Therefore, the present work shows that the disappearance of the XRD signal above a given pressure is not sufficient to be taken as the sole criterion for amorphization. Methods such as pair correlation distribution functions are needed to confirm the existence of amorphous compounds. 32 One of the possible reasons for pressure-induced amorphization in previous studies on MAPbBr 3 and MAPbI 3 could be the fact that experimental data may be affected by a large degree of sample nonhydrostaticity. 33 However, this hypothesis is ruled out by the fact that our study and the study on MAPbI 3 10 were performed under similar quasi-hydrostatic conditions as experiments where amorphization was reported. The reason for the previously reported amorphization could be related to other issues like different sample preparation methods or sample manipulation to load them into the diamond-anvil cell, which include methods from cutting samples into smaller pieces to mechanical grinding, which might introduce structural defects in a soft material like hybrid perovskites. Another possibility is the loading of large amounts of sample in the pressure chamber favoring sample bridging between the diamonds, which could induce large deviatoric stresses in the samples. 34

Author Contributions
A.L. and R.T., experiments, data analysis, writing�original draft; C.P., XRD experiments; I.F.-G., R.A., and P.P.B., synthesis of samples; D.E., conceptualization, data analysis, writing�original draft. All authors contributed to the discussion, writing, and proofreading of the manuscript and have given approval to the final version of the manuscript.