Crystal Structure Features of CsPbBr3 Perovskite Prepared by Mechanochemical Synthesis.

We present a mechanochemical procedure, with solvent-free, green-chemistry credentials, to grow all-inorganic CsPbBr3 perovskite. The crystal structure of this perovskite and its correlations with the physicochemical properties have been studied. Synchrotron X-ray diffraction (SXRD) and neutron powder diffraction (NPD) allowed us to follow the crystallographic behavior from 4 to 773 K. Unreported features like the observed negative thermal expansion of the b unit-cell parameter stem from octahedral distortions in the 4-100 K temperature range. The mechanochemical synthesis was designed to reduce the impact energy during the milling process, leading to a defect-free, well-crystallized sample characterized by a minimum unit-cell volume and octahedral tilting angles in the low-temperature orthorhombic perovskite framework, defined in the Pbnm space group. The UV-vis diffuse reflectance spectrum shows a reduced band gap of 2.22(3) eV, and the photocurrent characterization in a photodetector reveals excellent properties with potential applications of this material in optoelectronic devices.

Introduction

All-inorganic cesium lead halide perovskites (CsPbX3, X = I, Br, and Cl) have attracted widespread attention because of their improved stability and balanced carrier mobility compared with their hybrid organic–inorganic counterparts. Nevertheless, the electrical and optical properties of these inorganic perovskites are strongly determined by their compositions, morphologies, and crystallographic phases.[1−4]

Among them, the best candidate for photovoltaic applications is CsPbI3, which shows a low band gap of 1.73 eV when the cubic phase is preserved. However, bulk CsPbI3 can only maintain the cubic perovskite structure (black phase) above ≈593 K,[5] and it transforms into an orthorhombic nonperovskite (yellow phase) material at room temperature, losing its photovoltaic property. The addition of bromide to the halide anion makes the black phase at room temperature more stable owing to the increased effective tolerance factor and a lower phase-transition temperature.[1,6] However, the larger electronegativity differences between the halogen and lead result in a more ionic bonding character, yielding shorter bond lengths and a larger band gap. This fact limits the short-circuit current (JSC) of perovskite solar cells (PSCs) but has great potential in tandem and semitransparent photovoltaic applications.

CsPbBr3 has attracted much interest because it possesses a stable crystalline structure (orthorhombic phase) at room temperature and, depending on its morphology, it can retain high carrier mobility, good optoelectronic properties, large photoluminescence quantum yield, and superior stability under humidity and thermal attacks. These properties make it suitable for applications in various optoelectronic devices such as light-emitting diodes, photovoltaic cells, photodetectors, and lasers.[1,7−11]

CsPbBr3 is usually prepared by reacting equimolar amounts of CsBr and PbBr2 through conventional wet procedures, while it was described to be prepared through dry methods in only four works. Stoumpos et al.[7] and Linaburg et al.[12] used solid-state reactions (milling and heating), while Posudievsky et al.[13] and Pal et al.[14] reported a mechanochemical procedure (without further heating).

In this work, CsPbBr3 was obtained by a mechanosynthesis procedure in a planetary ball mill at room temperature. This synthetic process involves simplicity, swiftness, and reproducibility in line with the green chemistry credentials (e.g., solventless solid-state synthesis). The combination of a moderate mechanical energy generated under mild ball-milling conditions and the inherent chemical modification of structures/surfaces makes this methodology extremely promising for greener perovskite syntheses, yielding well-crystallized powders with excellent photovoltaic and optoelectronic properties.

Experimental Section

CsPbBr3 was obtained as a microcrystalline powder from mechanosynthesis in a planetary ball mill, from stoichiometric amounts of CsBr (Strem) and PbBr2 (Alfa Aesar) processed in an N2 atmosphere. A total of 1.5 g of the reactants was milled using 30 zirconia balls of 5 mm diameter, with a final mass ratio of 8.6:1, for 4 h at 400 rpm in a Retsch PM100 mill. A laboratory X-ray diffraction (XRD) pattern was collected on a Bruker D5 diffractometer with Kα Cu (λ = 1.5418 Å) radiation. To study the crystallographic structure, a neutron powder diffraction (NPD) pattern at room temperature (298 K) was collected using the HRPT diffractometer of the SINQ spallation source (PSI, Paul Scherrer Institute, Villigen, Switzerland) with a wavelength of 1.494 Å. The crystal structure at lower temperatures was investigated from NPD patterns sequentially collected from 100 to 4 K in the D20 instrument (Institute Laue Langevin, Grenoble, France) with a wavelength of 1.540 Å. The sample, contained in a V cylinder, was introduced in a standard “orange” cryostat and measured at 100 K for 1 h, and then cooled down to 4 K while acquiring sequential patterns every 3 min. Finally, a good statistics pattern was collected at 4 K for 30 min. To investigate the high-temperature structural evolution, synchrotron X-ray powder diffraction (SXRD) patterns were collected at RT, 473, 673, and 773 K in the MSPD high-resolution diffractometer at the ALBA facility, Barcelona (Spain), selecting an incident beam with 38 keV energy (λ = 0.3252 Å). The high angular-resolution mode (MAD setup) was selected.[15] The polycrystalline powder was collected in quartz capillaries of 0.7 mm diameter, which were kept rotating during the acquisition time. In both cases, the refinement of the structure was performed by the Rietveld method using the Fullprof software.[16,17] A pseudo-Voigt function was chosen to generate the line shape of the diffraction peaks. The background was interpolated between regions devoid of reflections. The following parameters were refined in the final run: scale factor, background coefficients, zero-point error, pseudo-Voigt corrected for asymmetry parameters, positional coordinates, anisotropic displacement factors, and occupancy factors. For the neutron refinements, the coherent scattering lengths for Cs, Pb, and Br were 5.42, 9.405, and 6.795 fm, respectively. The scanning electron microscopy (SEM) images were obtained on a Hitachi instrument, model TM-1000. The optical diffuse reflectance spectrum was measured at room temperature using a UV–vis spectrophotometer Varian Cary 5000. The optoelectronic properties of the CsPbBr3 crystals were studied by fabricating a photodetector by drop-casting a CsPbBr3 suspension in dimethyl sulfoxide (ratio perovskite/solvent, 1:4, weight) onto a SiO2/Si substrate with pre-patterned gold electrodes separated by 10 μm (Ossila). The photoresponse was analyzed by illuminating the device with different light-emitting diode (LED) sources with wavelengths ranging from 420 to 1050 nm (1.18–2.95 eV). The light from the LED sources was focused to form a spot (400 μm in diameter) on the sample, and the intensity was adjusted to achieve a power density of 16 mW/cm2.

Results and Discussion

Initial Characterization

CsPbBr3 was obtained as a yellowish polycrystalline powder. The initial crystallographic identification of CsPbBr3 was carried out using laboratory XRPD. A Le Bail refinement, illustrated in Figure , shows that CsPbBr3 is pure and presents the characteristic distortion defined in the orthorhombic symmetry, space group Pbnm. The earliest crystal elucidations were made by XRD and NPD in the 1970s.[18] Recently, several structures have been reported using laboratory and/or synchrotron X-ray diffraction data,[7,12−14,19] but there are no recent measurements from NPD data.

Figure 1

Le Bail fit of a laboratory XRD pattern of CsPbBr3, prepared by ball milling.

Room-Temperature Combined Neutron and Synchrotron X-ray Diffraction Characterization

For a precise crystal structure resolution, both NPD and synchrotron SXRD were combined in a joint refinement; hence, both patterns were modeled in the mentioned Pbnm space group. The Cs+ and Pb2+ cations are located at 4c (x,y,1/4) and 4b (1/2,0,0) Wyckoff sites, while Br1 and Br2 atoms are placed in 4c (x,y,1/4) and 8d (x,y,z) sites. Figure illustrates the quality of the fit for both NPD and SXRD patterns, including anisotropic refinement of the displacement factors for all the atoms. The figure also includes a view of the crystal structure, highlighting the tilting of the PbBr6 octahedra. Table lists the main crystallographic data. As it is well known, the orthorhombic Pbnm crystal structure in perovskites consists of a three-dimensional (3D) framework of corner-sharing octahedra (PbBr6), tilted antiphase along the (100) and (010) directions of the pseudocubic cell and in-phase along the (001) direction, which correspond to a–a–b+Glazer’s notation as derived by Woodward for a simple perovskite.[20,21] The tilting angles, estimated as φ = (180° – θ), where θ = ⟨Pb–Br–Pb⟩, are 7.43 and 11.28° for the antiphase and in-phase tilts at RT, respectively. These compare well with the values of 6.8 and 11.35° found by Linaburg et al.[12] for CsPbBr3 at RT.

Figure 2

Observed (crosses), calculated (black line), and difference (blue line) profiles after the Rietveld refinement in an orthorhombic cubic unit cell for (a) NPD and (b) SXRD. Inset: view of the crystal structure enhancing the tilting of the PbBr6 octahedra and the anisotropic displacement factors.

Table 1

Crystallographic Data for CsPbBr3 Phase in the Orthorhombic System (Pbnm) from Combined NPD and SXRPD at RTa

	x	y	z	Ueq	occ
Cs	0.9927(7)	0.9710(7)	0.25	0.084(4)	1
Pb	0.5	0	0	0.026(1)	1
Br1	0.0464(8)	0.505(1)	0.25	0.086(6)	1
Br2	0.7929(5)	0.2070(5)	0.0251(4)	0.071(4)	1

a = 8.19154(2) Å, b = 8.24459(2) Å, c = 11.73993(2) Å, and V = 792.87(1) Å3. NPD: Rp = 3.69%, Rwp = 4.62%, Rexp = 4.29%, χ2 = 1.16, and RBragg = 8.54%. SXRPD: Rp = 8.41%, Rwp = 10.9%, Rexp = 9.24%, χ2 = 1.39, and RBragg = 6.23%.

Low-Temperature Neutron Diffraction Characterization

Additional NPD patterns were measured at 100 and 4 K; besides, several patterns were sequentially collected during the cooling process. These data reveal that the orthorhombic unit cell is maintained down to 4 K. Figure shows the a, b, and c unit-cell parameter variation as well as the Rietveld plot at 4 K. It is remarkable that a and c parameters decrease, whereas b increases upon cooling. This conspicuous effect of negative thermal expansion along the b-axis has not been reported before for CsPbBr3. Normally, negative thermal expansion in Pbnm perovskites is a consequence of magnetorestrictive effects, concomitant with magnetic ordering, for instance in rare-earth ferrites. In CsPbBr3, it deserves further analysis. Table lists the main crystallographic parameters from the pattern collected at 4 K. The Rietveld refinement and the crystallographic data at 100 K are displayed in Figure S1 and Table S1, respectively. The octahedral tiltings at 4 K are 10.97 and 14.11° for antiphase and in-phase tilts, respectively. The thermal evolution (including the RT and the range between 100 and 4 K) of the tilts shows a linear behavior for both phase and antiphase octahedral rotations.

Figure 3

Thermal evolution of (a) a, (b) b, and (c) c unit-cell parameters. (d) Rietveld NPD profiles at 4 K.

Table 2

Crystallographic Data for CsPbBr3 Phase in the Orthorhombic System (Pbnm) from NPD at 4 Ka

	x	y	z	Uiso	occ
Cs	0.9800(8)	0.9378(5)	0.25	0.021(1)	1
Pb	0.5	0	0	0.015(1)	1
Br1	0.0701(5)	0.5078(6)	0.25	0.027(1)	1
Br2	0.8020(4)	0.2007(4)	0.0369(3)	0.019(1)	1

a = 7.9734(7) Å, b = 8.3065(8) Å, c = 11.612(1) Å, and V = 769.1(1) Å3. Rp = 2.62%, Rwp = 3.40%, Rexp = 1.20%, χ2 = 8.03, and RBragg = 4.33%.

The unit-cell volume evolution is plotted in Figure a, which shows a constant reduction to reach a plateau below 20 K; this can be a compromise between the contractions of a and c and the expansion observed in b. A subsequent reduction is observed close to 4 K. Hence, other structural parameters such as the atomic displacement parameters (ADPs) and interatomic distances were examined as a function of temperature, showing that the ⟨Cs–Br⟩ distance exhibits a monotonic contraction, while the ⟨Pb–Br⟩ distances in the PbBr6 octahedra remain unaltered within the experimental errors (Figure b). This thermal evolution is reasonable considering the greater covalent component existing between Pb–Br with respect to Cs–Br interactions, implying that the unit-cell volume contraction stems from the rotation of quasi-rigid octahedra. Second, the ADPs exhibit a conventional reduction for Cs+ ions, whereas they remain almost constant for Pb2+; this observation can be a symptom of the much more covalent Pb–Br and rigid bond that locks the Pb displacement factors. This behavior close to 4 K can be driven by a structural limitation in the PbBr6 octahedral framework to follow the progressive reduction of the Cs–Br bonds, leading to octahedral distortions that account for the expansion of parameter b.

Figure 4

Thermal evolution of the unit-cell volume (a), interatomic distances, and atomic displacement factors (b). Inset: view of the crystal structure at 4 K.

High-Temperature Synchrotron X-ray Diffraction Characterization

For the high-temperature analysis, an SXRD experiment was performed at selected temperatures (473, 673, and 773 K). As mentioned above, the RT pattern confirms the already described orthorhombic symmetry; however, at 473 K and above, the structure can be defined in the cubic Pm3̅m space group. Figure shows a selected angular range that illustrates this phase transition. It is important to remark that between the orthorhombic and cubic phases (RT to 473 K) a transient tetragonal () phase has been previously reported,[7,18] which we could not identify.

Figure 5

Thermal evolution of selected regions of the SXRD patterns of CsPbBr3, in which an orthorhombic to cubic phase transition is evidenced.

At 473 K and above, the cubic symmetry is defined in the space group Pm3̅m. Cs atoms are placed at 1c (1/2,1/2,1/2) Wyckoff site, Pb at 1a (0,0,0), and Br at 3d (1/2,0,0). Figure plots the Rietveld refinements at different temperatures, showing an excellent agreement between the observed and the calculated profiles, including an inset with a view of the cubic crystal structure above 473 K. The main crystallographic data are listed in Table .

Figure 6

Rietveld refinement of synchrotron XRD patterns at (a) 473 K, (b) 673 K, and (c) 773 K. A view of the cubic crystal structure is given in the inset of (c).

Table 3

Crystallographic Data for CsPbBr3 Phase in the Cubic System (Pm3̅m) from SXRPD at High Temperature

	473 K	573 K	673 K
Unit Cell
a (Å)	5.87330(3)	5.91018(6)	5.92805(6)
V (Å3)	202.603(2)	206.444(3)	208.322(4)
Cs (0.5,0.5,0.5)
U11 = U22 = U33	0.013(2)	0.175(5)	0.198(5)
Pb (0,0,0)
U11 = U22 = U33	0.0383(8)	0.058(1)	0.071(2)
Br (0.5,0,0)
U11	0.037(4)	0.054(6)	0.067(6)
U22 = U33	0.234(5)	0.270(8)	0.292(8)
Reliability Factors
Rp (%)	10.3	11.5	11.6
Rwp (%)	13.3	16.0	16.0
Rexp (%)	9.82	10.5	10.6
χ2	1.85	2.35	2.26
RBragg (%)	3.59	8.18	8.97

The thermal evolution of the unit-cell parameters is illustrated in Figure as volume/Z versus temperature. On the other hand, the comparison of the unit-cell volume with those reported in the literature reveals significant differences, as illustrated in the inset of Figure . The obtained unit-cell volume is subtly lower than those of the previous reports. Additionally, a correlation with the synthesis method can be established. In general, the samples synthesized from the mechanochemical (MC) method (without heating) present a higher unit-cell volume, while those obtained from the solid-state (SS) reaction (with heating) exhibit a smaller unit-cell size. In contrast, our sample obtained by ball-milling exhibits the lowest unit-cell size, which is discussed below.

Figure 7

Thermal evolution of the normalized unit-cell volume of CsPbBr3 obtained by ball milling compared with other literature values from alternative synthesis techniques.

These subtle changes can be related to the defects in the crystals, since annealing at moderate temperatures may reduce the defects in the sample, producing quality crystals. Moreover, the sample obtained by Stoumpos et al.[7] in sealed ampoules at 600 °C presents a smaller cell than that obtained by Linaburg et al.[12] in air at 425 °C for 20 h. In this situation, the unit-cell size of the present sample is in the lower limit of the analyzed samples; this fact suggests that the present milling conditions yield a well-crystallized sample with a minimum number of defects. In the mechanosynthesis technique, the energy transferred to the mixture is a determining factor of the synthetic procedure. It depends on the different characteristics of the milling process, such as the balls/mixture mass ratio, grinding time, and rotation speed. These determine the crystallinity and defective nature of the perovskite structure and thus the lattice parameter and the unit-cell volume. If we compare our milling conditions, ball:mixture mass ratio, milling time, and rpm (8.6:1 mass ratio, 30 balls of 5 mm diameter, for 4 h at 400 rpm) with those described by Posudievsky et al.[13] (29:1 mass ratio, 30 balls of 10 mm diameter, for 4 h at 500 rpm), the latter are by far more energetic than those used in the present work, thus leading to a more defective material. Moreover, this can be quantified by estimating the ball-impact energy (ΔEb) and weight-normalized cumulative kinetic energy (Ecum). For the conditions of the present work, the estimated ΔEb and Ecum are 4.1 mJ/impact and 149 kJ/g, respectively. However, for Posudievsky et al.[13] ΔEb and Ecum are 20.4 mJ/impact and 826 kJ/g, respectively. The kinetic energy given to the sample per impact is approximately 5 times smaller in the present synthesis; hence, it is possible to infer that such moderate conditions for long times yield a better-crystallized, more defect-free sample. We have prepared, therefore, CsPbBr3 perovskite in mild milling conditions, beyond those typically used in literature.

Microstructure by Scanning Electron Microscopy (SEM)

Figure illustrates some typical views of the as-prepared CsPbBr3 polycrystals. From a mechanosynthesis process, involving the collision of high-energy ZrO2 balls with the specimens, one would expect a highly disaggregated product formed of small particles. However, surprisingly, we can observe a heterogeneous picture where quite large particles (10–20 μm) are mixed up with smaller fragments of undefined shapes (Figure a). However, in a larger magnification picture (Figure b), it is evident that each particle is indeed formed by an agglomeration of much smaller grains of typically 0.5–1 μm. We assume that these individual grains are monocrystalline, providing a sufficiently large diffraction domain that accounts for the good crystallinity of the sample versus neutron and X-ray synchrotron diffraction techniques. Altogether, this scenario illustrates that the growth of microcrystals is not perturbed by the dynamical motion of the reactants and ZrO2 balls after 4 h of reaction. This morphological evidence is in agreement with that previously deduced from the synthesis conditions in terms of ΔEb and Ecum energies.

Figure 8

SEM images of CsPbBr3 samples at 2500× (a) and 7000× (b) magnifications.

UV–vis–NIR Spectra

The absorption ability of CsPbBr3 powder was determined by diffuse reflectance UV–vis spectroscopy. Figure depicts the optical absorption coefficient related to the Kubelka–Munk function (F(R) = a = (1 – R)2/2 R, R is the reflectance) versus the wavelength in electronvolts. The band gap was calculated by extrapolating the linear region to the abscissa. The value obtained for CsPbBr3 (∼2.22(3) eV) is in agreement with data reported in the literature for its band gap at room temperature.[6,8,9] Moreover, there is a subtle reduction with respect to the value of 2.27 eV given by Linaburg et al.[12] for a sample prepared by solid-state reactions, which is convenient for use in solar cells.

Figure 9

Kubelka–Munk (KM) transformed diffuse reflectance spectrum of CsPbBr3. The inset shows an expanded zone of the absorption edge.

Optoelectronic Characterization

Figure a shows the time evolution of the current flowing through the device (with a bias voltage of 1 V applied between electrodes), while the illumination is switched on and off. This measurement allows one to determine the photocurrent (subtracting the current in the dark to the current upon illumination), as well as the response time of the photodetector device. By employing a wavelength of 420 nm, we obtain a photocurrent of 90 nA with response times of 170 and 90 ms for the rise and decay processes, respectively. In addition, a sizeable overshoot can be clearly seen in this figure, which has already been observed for these materials and has been attributed to a sudden generation of photogenerated charge carriers followed by a slow photocurrent decay toward a steady state when the equilibrium between the charge diffusion rate and its generation rate is achieved.[22−24]

Figure 10

(a) Time evolution of the current flowing through the photodetector (bias of 1 V) under alternating dark and light illumination with different wavelengths (power density of 16 mW/cm2). (b) Responsivity of the device as a function of the LED wavelength (bias of 1 V and power density of 16 mW/cm2). The inset shows the device investigated.

To determine the spectral response of our device, we extract the responsivity at different illumination wavelengths. The responsivity is a figure of merit that allows for a direct comparison between different photodetectors. This value is defined as R = Iph/P, where Iph is the photocurrent and P the effective power, which is given by P = Plight·Adev/Aspot, where Plight is the LED power, Adev is the area of the material covering the channel of the device and Aspot is the area of the spot.

Figure b exhibits the responsivity as a function of the illumination wavelength. The responsivity increases at shorter wavelengths, reaching a value of 3 A/W at 420 nm, which is higher than that reported, under similar illumination and biasing conditions, for other perovskites like CsPbCl3 (less than 0.5 A/W) or CsPbI3 (less than 0.4 A/W).[25,26] For CsPbBr3 photodetectors in the literature, the reported responsivity ranges from ∼0.005[24] to ∼10 A/W[27] and thus our device is in the upper bound of this range.

Conclusions

We have obtained a well-crystallized CsPbBr3 perovskite from a mechanochemical method under mild milling conditions. The unit-cell parameters at RT were smaller than those described for samples synthesized by ball milling, but they were similar to those of the specimens previously obtained from conventional solid-state reaction methods. The crystallographic features were analyzed from synchrotron X-ray and neutron powder diffraction in the 4–773 K temperature range. At RT, an orthorhombic superstructure of perovskite, defined in the Pbnm space group, is observed, which becomes cubic Pm3̅m above 473 K, in agreement with previous reports. At lower temperatures, the phase remains orthorhombic down to 4 K, and the collapse of the unit cell upon cooling is related to the progressive octahedral tilting; below 8 K, the structure distortion seems to reach a limit in the octahedral rotation. Finally, the optoelectronic properties of our CsPbBr3 specimen, implemented in a photodetector device, demonstrate a high and selective responsivity of 3 A/W at shorter wavelengths, improving the reported responsivity ranges.