Time-dependent transformation routes of perovskites CsPbBr3and CsPbCl3under high pressure

Article abstract of DOI:10.1039/d1ta01875b

All-inorganic halide perovskites are prospective materials for diverse applications in photovoltaic and optoelectronic devices. Their high performance is associated with good operational stability, which is the key problem of hybrid organic-inorganic perovskites. However, for these materials only fragmentary information is available on the mechanical robustness and response to external stress, fundamentally important for strain engineering in multilayers, pressure-assisted technologies, and flexible panels applications. Here we show that all-inorganic perovskites CsPbX3 (where X = Cl, Br) undergo various types of pressure-induced transformations, including reversible phase transitions, irreversible chemical reactions reducing the dimensionality of PbX6 frameworks, and amorphization. The transformation routes depend on the mode of the applied stress and are related to the kinetics of transitions to the most stable phases. The slow-kinetics transformations in a moderate pressure range of technological importance, between 0.5 and 1.5 GPa, can require days or even weeks, depending on the sample quality and external stimuli. The pressure-induced narrowing and widening of energy gaps has been explained by the mechanism combining Pb-X bond lengths and PbX6 octahedra tilts with the electronic structure of the crystals.

Full text of DOI:10.1039/d1ta01875b

Journal of
Materials Chemistry A
View Article Online
View Journal | View Issue
PAPER
Time-dependent transformation routes of
perovskites CsPbBr3 and CsPbCl3 under high
pressure†
Cite this: J. Mater. Chem. A, 2021, 9,
10769
a
b
and Kenny Stahlc
˚
*
Marek Szafranski,
Andrzej Katrusiak
´
All-inorganic halide perovskites are prospective materials for diverse applications in photovoltaic and
optoelectronic devices. Their high performance is associated with good operational stability, which is the
key problem of hybrid organic–inorganic perovskites. However, for these materials only fragmentary
information is available on the mechanical robustness and response to external stress, fundamentally
important for strain engineering in multilayers, pressure-assisted technologies, and flexible panels
applications. Here we show that all-inorganic perovskites CsPbX3 (where X ¼ Cl, Br) undergo various
types of pressure-induced transformations, including reversible phase transitions, irreversible chemical
reactions reducing the dimensionality of PbX6 frameworks, and amorphization. The transformation
routes depend on the mode of the applied stress and are related to the kinetics of transitions to the
most stable phases. The slow-kinetics transformations in a moderate pressure range of technological
importance, between 0.5 and 1.5 GPa, can require days or even weeks, depending on the sample quality
and external stimuli. The pressure-induced narrowing and widening of energy gaps has been explained
by the mechanism combining Pb–X bond lengths and PbX6 octahedra tilts with the electronic structure
of the crystals.
Received 3rd March 2021
Accepted 11th April 2021
DOI: 10.1039/d1ta01875b
rsc.li/materials-a
[CH3NH3]+).8,9 This nding renewed a wide interest in all-
inorganic halide perovskites, which currently are considered
Introduction
not only for photovoltaic applications, but also for light-
emitting diodes,10,11 multicolor displays,12 lasers,13 and high-
energy detectors.14
The discovery of superior optoelectronic properties of organic–
inorganic lead halide perovskites1–3 has opened new prospects
for low-cost, high-performance photonic devices. These mate-
rials have great potential for numerous devices such as light-
emitting diodes, photodetectors, and solar cells. Particularly
impressive is a quick rise of power conversion efficiency of
single-junction solar cells based on perovskite absorbers that
currently exceeds 25%.4 Although the optical and electronic
parameters of hybrid perovskites are excellent, their practical
application is still hampered by long-term stability problems
related to the resistance to atmospheric conditions such as
moisture, temperature and UV illumination.5–7 In the search for
better materials the organic cation was replaced with cesium,
Cs+, in order to obtain the inorganic CsPbBr3 perovskite struc-
ture of improved stability and photovoltaic properties compa-
Recently, a growing interest is observed in the strain engi-
neering aimed at enhancing the performance of the perovskite-
based optoelectronic devices.15–17 This approach requires the
understanding of relationships between the strain-induced
crystal structure modications and the associated properties.
One of the best tools to effectively study such relationships in
controlled conditions is hydrostatic pressure, capable of
precisely manipulating the crystal structure and tuning the
electronic parameters of the crystals. Moreover, the pressure-
assisted techniques for fabrication perovskite wafers or thick
lms on substrates18,19 can generate signicant changes in their
structure and properties. Therefore, the knowledge about the
structural response to the external stress is of primary impor-
tance. The high-pressure studies of inorganic metal halide
perovskites were reported by several groups,20–28 but it is char-
acteristic that all diffraction experiments were performed for
powdered samples only, which are not ideally suited for accu-
rate structural determinations for highly pseudo-symmetric
materials, nor for spectroscopic studies and visual micro-
scopic observations. Presently, we have undertaken this study
based on the single-crystal X-ray diffraction experiments, sup-
ported by microscopic observations, and single-crystal
rable to those of analogous hybrid MAPbBr3 (MA
¼
a
´
Faculty of Physics, Adam Mickiewicz University, Uniwersytetu Poznanskiego 2, 61-614
´
Poznan, Poland. E-mail: masza@amu.edu.pl
b
´
Faculty of Chemistry, Adam Mickiewicz University, Uniwersytetu Poznanskiego 8, 61-
´
614 Poznan, Poland
cChemistry Department, Technical University of Denmark, Building 207, DK-2800
Lyngby, Denmark
† Electronic supplementary information (ESI) available. CCDC 2060421–2060427,
2060567–2060576. For ESI and crystallographic data in CIF or other electronic
format see DOI: 10.1039/d1ta01875b
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 10769–10779 | 10769

View Article Online
Journal of Materials Chemistry A
Paper
measurements of the absorption edge of CsPbBr3 and CsPbCl3.
Owing to this approach, we have revealed new phases for both
these materials in hydrostatic conditions. We have also reex-
amined the stability regions of the phases observed, the pres-
sure effects on their structures, as well as the related changes in
the optical properties.
Results and discussion
X-ray diffraction experiments, dielectric measurements and
microscopic observations of CsPbBr3 under pressure
At atmospheric pressure above room temperature CsPbBr3
undergoes two successive phase transitions:29,30 the prototypic
perovskite phase I of cubic space group (s. g.) Pm3m at 403 K
transforms to phase II of tetragonal s. g. P4/mbn, and at 361 K it
transforms further to phase III of orthorhombic s. g. Pmna. Both
these critical temperatures increase when the crystal is
compressed,20 and therefore none of these transitions is ex-
pected to occur in high-pressure experiments in this study,
performed at room temperature. The hydrostatic compression
of a single-crystal phase III of CsPbBr3 resulted in monotonic
changes in the lattice parameters up to about 1.3 GPa, as shown
in Fig. 1a. At still higher pressure a phase transition damaged
the single-crystals used for the X-ray diffraction measurements.
It is worth noting that already above 0.5 GPa the lattice strain
generated micro cracks and other defects, which could be seen
through the optical microscope. This strain resulted in a clear
broadening or splitting of Bragg reections and hence the
structural parameters derived from the SCXRD data collected in
this pressure region were determined with a lower accuracy, as
indicated by increased standard deviations in Fig. 1b. However,
the most intriguing was the slow transformation of the crystals
le under the pressure between 0.5–1.3 GPa, illustrated by
Fig. 2 Microscopic observations of a single crystal of 3D-CsPbBr3
phase III compressed in isopropanol at 0.10, 0.44 and 0.60 GPa. A ruby
chip for pressure calibration lies at the left edge of the DAC chamber.
a series of crystal photos in Fig. 2. The transformation was
manifested by a formation of strongly defected areas which
expanded to all the crystal sample over time. The new form in
the sample is clearly distinguished by its white color from the
3D-CsPbBr3 orange form. It should be noted that time required
for this process varied between different crystals samples, but it
could be accelerated by heating the high-pressure cell to 330–
350 K. The similar slow-kinetics transformations were observed
in our earlier high-pressure experiments on CH3NH3PbBr3 and
CH3NH3PbCl3,31 but those transitions were fully reversible. The
transformed CsPbBr3 samples remained colorless aer the
pressure release, and no reverse transformation to the orange
phase III was observed even for months.
Fig. 1 Compression of the unit-cell volume and lattice parameters of 3D-CsPbBr3 phase III (a), and pressure dependence of structural
parameters (b and c) measured by SCXRD. The average Pb–Br distance was calculated for the three independent bonds (Fig. S1†). The inset
shows the Pb–Br–Pb angles connected to the tilts of the PbBr6 octahedra.
10770 | J. Mater. Chem. A, 2021, 9, 10769–10779
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry A
Owing to the irreversible character of the transformation of conditions assured by isopropanol it transforms to white 1D-
CsPbBr3 to the white form, the sample could be recovered from CsPbBr3 and to 2D-CsPb2Br5 (see Fig. 3b). The latter compound
the DAC and its high-resolution PXRD data was recorded. It is formed according to the chemical reaction: 2CsPbBr3
/
turned out that the PXRD data consists of two phases that could CsPb2Br5 + CsBr. The byproduct CsBr was not detected, as it
be indexed separately and for each of them the crystal structure dissolved in isopropanol (possibly with a small amount of
has been determined. The Rietveld renement (Fig. 3a and S2, water) and this solution leaked out aer opening the DAC
Table S1†) has shown that the sample contains both the white chamber. Consequently, it was absent in the analyzed part of
form of 1D-CsPbBr3,32 of orthorhombic s. g. Pmnb, Z ¼ 4, a ¼ the sample.
ꢀ
ꢀ3
4.5948(2), b ¼ 9.6985(6), c ¼ 16.7763(10) A, V ¼ 747.60(13) A ,
In order to better characterize the slow transformation of
and 2D-CsPb2Br5,33 of tetragonal s. g. I4/mcm, Z ¼ 4, a ¼ CsPbBr3, we have used another technique with another type of
ꢀ
ꢀ3
8.4930(6), c ¼ 15.2072(7) A, V ¼ 1096,92(21) A . Thus, the pressure-transmitting medium. The polar isopropanol was
moderate pressure of 0.5–1.5 GPa destabilizes the perovskite replaced with gaseous helium and the electric permittivity of
structure of orange 3D-CsPbBr3 phase III and in hydrostatic the polycrystalline sample was used as a probe of structural
Fig. 3 The orange phase III of CsPbBr3 transformed into white 1D-CsPbBr3 and 2D-CsPb2Br5 at 0.60 GPa in isopropanol (a and b), and the 3D–
1D transition observed in a pellet (shown in the inset) under pressure of gaseous He (c and d). The real part of electric permittivity 30 under
different pressures measured on heating (red lines) and cooling (blue), at the probing electric field frequency of 100 kHz. The white phases were
mechanically extracted and identified by PXRD using Rietveld refinement.
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 10769–10779 | 10771

View Article Online
Journal of Materials Chemistry A
Paper
changes. The results of dielectric measurements (Fig. 3c) show compressed in the DAC (see also Fig. S6†). These observations
that increasing pressure suppresses the dielectric response of were fully conrmed by XRPD experiments where pressure was
CsPbBr3, which matches the general tendency observed for changed with time intervals shorter than 1 h (Fig. 4b). The
other materials. However, most remarkable is the comparison PXRD has shown that in the pulverized sample the transition
of the heating and cooling runs. They are identical at 0.27 GPa, onset occurs between 1.3 and 1.54 GPa. Moreover, the
somewhat different at 0.47 GPa, but at 0.70 GPa the electric substantially increased background in the diffraction pattern
permittivity spectacularly drops in the cooling run, reecting measured at 1.54 GPa marks an increase of the crystal structure
the dramatic changes in the crystal structure. Aer releasing disorder, while the positions and intensities of reections
pressure and removing a part of one electrode, the white islands suggest a coexistence of phases III and IV. Pure phase IV is
were revealed in the originally fully orange pellet (see the inset represented in Fig. 4b by the diffraction pattern collected at
in Fig. 3d). The subsequent PXRD analysis showed that this is 2.12 GPa. This pattern shows a partial loss of the long-range
the pure white 1D-CsPbBr3, with no traces of 2D-CsPb2Br5 order of the crystal, but a few broad reections indicate that
(Fig. 3d and S3, Table S1†). This experiment conrmed the the average residual crystal structure approximates that of
instability of the 3D perovskite structure of CsPbBr3 in the phase III. With these features, phase IV can be described as
moderate pressure and temperature range. Noteworthy, at a partly amorphous version of phase III. The transition to phase
ambient pressure an inverse situation takes place: the white IV is fully reversible, as testied by the diffraction pattern
form 1D-CsPbBr3 can be crystallized together with the orange collected aer the pressure release to 0.05 GPa (Fig. 4b).
form 3D-CsPbBr3 from the solution of specic stoichiometry,32
but it transforms irreversibly to the orange isomer at 424 K.34
Absorption edge of compressed CsPbBr3
Thus, we can conclude that at atmospheric pressure the free
energy of 3D-CsPbBr3 is only somewhat lower compared with
The absorption edge of CsPbBr3 was measured on a single-
crystal plate approximately 2.5 mm thick, during the compres-
1D-CsPbBr3, but these minima are separated by an energy
sion and decompression cycles. The representative spectra are
barrier. Under pressure the energy barrier clearly lowers and the
shown in Fig. 5a and b. The determined pressure dependence
Eg(p), plotted in Fig. 5c, shows that the bandgap initially
narrows linearly, but above circa 1.0 GPa the pressure effect
saturates and about 1.5 GPa the Eg value becomes nearly
relation between the minima reverses in favor of 1D form.
The time-dependent PXRD experiments performed under
0.77 GPa, using the DAC lled with aprotic Galden as the
hydrostatic liquid, evidenced that 3D-CsPbBr3 transforms into
constant. It contrasts to the previously reported24 gradual blue
1D-CsPbBr3, similarly as in gaseous helium (Fig. S4, Table S1†).
shi of Eg starting at 1 GPa and continued to 2 GPa (see the
A similar transformation was observed recently in the pellets
green points in Fig. 5c; these data were obtained from the
experiment where the silicon oil, hydrostatic to about 0.5 GPa
pressed in oil.35
Our microscopic observations showed that when the single
only, was applied as a pressure medium). The turning point at
crystals of orange CsPbBr3 phase III are compressed quickly,
1 GPa was interpreted as resulting from the isostructural phase
without longer intervals, the white phase is not formed and
transition,24 but we have not observed such a transition at
around 1.5 GPa the crystals reversibly transform to a new light-
1.0 GPa. The isostructural phase transition must be discontin-
yellow phase IV. The transition is associated with numerous
uous and associated with a jump-wise volume change, whereas
cracks and other defects, as illustrated in Fig. 4a for the crystal
the evidently continuous changes in the lattice parameters were
Fig. 4 Single crystal of orange 3D-CsPbBr3 quickly compressed in isopropanol in a DAC (a), and PXRD patterns for the powder quickly
compressed and after releasing the pressure (b). Isopropanol was used as the hydrostatic liquid.
10772 | J. Mater. Chem. A, 2021, 9, 10769–10779
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry A
observed in the considered pressure range.24 Also, no charac-
teristic features of isostructural phase transitions can be found
in our structural data around 1.0 GPa (Fig. 1) nor in all range up
to 1.5 GPa.
The nonlinear Eg(p) can be rationalized by the crystal struc-
ture response to hydrostatic pressure and by the related
changes in the structural parameters. The calculations of the
electronic structure of CsPbBr3 have shown that electronic
states responsible for the absorption edge mainly involve
orbitals Pb 6s, 6p and Br 4p, whereas the contribution of Cs
cations is secondary. Therefore, similarly as in hybrid lead
halides,36 the shortening of Pb–Br distance is associated with
a stronger overlapping of the crucial atomic orbitals (Pb 6s and
Br 4p), which results in the band gap narrowing, whereas the
Pb–Br–Pb bending reduces this overlapping and leads to the
band gap widening. Both these two types of structural trans-
formations absorbing the crystal strain under pressure and
their competing effects on the bandgap are present in CsPbBr3.
To about 0.8 GPa the contraction of the Pb–Br bonds predom-
inates and hence the red shi of the absorption edge. Similar
effect was observed in the low-pressure range for MAPbBr3 and
for other hybrid organic–inorganic perovskite photovoltaics.31,36
Under higher pressure the compression of Pb–Br bonds is
reduced (see Fig. 1b) and consequently the Eg(p) dependence
becomes attened. Thus, for 3D-CsPbBr3 phase III, this is not its
phase transition, but the nonlinear change in the pressure
dependence of the Pb–Br distance, which is responsible for the
anomalous energy gap behavior between 1 and 1.5 GPa.
The microscopic distortion of the compressed CsPbBr3
phase III, described above as the Pb–Br bonds shortening and
the Pb–Br–Pb angles bending, can be rationalized in terms of
the structural features of this crystal, illustrated in Fig. 6 and
S7.† The tilts of PbBr6 octahedra (measured by the Pb–Br–Pb
angles) are expected to require less energy to change for
accommodating the compressed volume, compared to the
covalent Pb–Br bonds shortening, which surprisingly is quite
signicant in the initial stages of the compression (Fig. 1).
However, there is less space on the convex side of the Pb–Br–Pb
angles, compared to their concave side, as clearly seen from the
Br–Cs distances indicated in Fig. 6. Owing to this surrounding
of the Br anions, the crystal volume compression is accommo-
dated by the simultaneous squeezing of Pb–Br bonds and the
increased tilts of PbBr6 octahedra. But the PbBr6 reorientations
toward the sharpening of Pb–Br–Pb angles, expected from the
mechanical point of view and also favoured by the electrostatic
attraction to the closer Cs+ cations, are opposed by the repul-
sion due to the Br–Cs distances shortening to below the sum of
+
ionic radii of Brꢀ and Cs (equal to 3.61 A according to Gold-
ꢀ
ꢀ
schmidt and 3.64 A according to Pauling). Hence the consid-
erable shortening of bonds Pb–Br, connected with the initial
Eg(p) reduction in phase III.
Fig. 5 Selected absorption spectra of a single-crystal plate of 3D-CsPbBr3
(a and b) and the energy gap (c) determined from the absorption edge
measured in the compression (blue) and decompression (red) runs,
compared with the literature data24 (green). The insets show the crystal just
below the transition at 1.44 GPa and in the phase transition region at
1.58 GPa, where the coexistence of phases III (orange) and IV (slightly
yellowish) is clearly visible due to the strong piezochromic effect. The
spectra (a and b) are shifted vertically to avoid the overlapping.
The quickly compressed orange 3D-CsPbBr3 phase III at
around 1.5 GPa transforms to slightly yellowish phase IV. This
transition is associated with a strong piezochromic effect
illustrated in Fig. 5c. In the transition region phases III and IV
coexist within 0.2–0.3 GPa both during the compression and
decompression cycles. In the absorption spectra the coexistence
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 10769–10779 | 10773

View Article Online
Journal of Materials Chemistry A
Paper
As seen in Fig. 4b, the Bragg reections signicantly broaden
and the strong diffused scattering appears in the PXRD patterns
in the phase transition region, and became very pronounced for
the pure phase IV at 2.12 GPa. This indicates that the crystal
structure of phase IV is strongly disordered, which can be
interpreted as the beginning of amorphization. The process can
proceed through the local symmetry breaking and distortions in
the framework built of the corner-sharing PbBr6 octahedra. As
a result, the Pb–Br distances and Pb–Br–Pb angles are, starting
from the transition point, stochastically distributed with
different values in the distorted structure. Therefore, the elec-
tronic structure of CsPbBr3 cannot be faithfully modelled for
the average structural parameters and such calculations clearly
failed.23,24 The calculations predicted a continuous evolution of
Eg at the transition to phase IV, while the experiment performed
in hydrostatic conditions shows evidently the jump-wise blue
shi of the absorption edge (Fig. 5), corroborating the sudden
changes of structural parameters and the abrupt modication
of electronic bands of the crystal. The compression of phase IV
leads to the continuous blue shi of the edge and to the linear
increase of Eg. It is remarkable that the pressure reduction
results in some hysteresis in the values of the bandgap in phase
IV, as illustrated in Fig. 5c. This pressure memory effect
disappears aer the reverse transition to phase III, which leads
to the perfect agreement between the Eg values determined
during the compression and decompression cycles. These
results conrm that the transition between phases III and IV is
fully reversible.
Fig. 6 Crystal structure of CsPbBr3 phase III at ambient conditions,
viewed along [y], with the voids (shown in gold) calculated for the
probing radius 0.8 A and step 0.1 A The shortest Br–Cs distances are
ꢀ
ꢀ
indicated.
of phases III and IV is reected as two absorption edges, as it is
shown in Fig. 5a at 1.58 GPa, and in Fig. 5b at 1.44 and 1.17 GPa.
The results presented in Fig. 5 indicate a rst-order nature of
the transition with a characteristic sudden change of the
absorption edge and an apparent pressure hysteresis between
the compression and decompression cycles. The large blue shi
of the absorption edge and the bandgap widening suggest that
the structure profoundly changes. At the same time a part of the
diffraction peaks characteristic of phase III is also preserved in
phase IV in the diffraction pattern collected at 2.12 GPa
(Fig. 4b), which might indicate an isostructural phase transi-
tion. Such a transition was the main consideration for
explaining the anomalous changes in the optical properties of
3D-CsPbBr3,23,24 but this approach seems to be oversimplied.
Pressure dependence of structural parameters in CsPbCl3
At ambient pressure CsPbCl3 undergoes three successive phase
transitions in a narrow temperature range of about 10 K. The
prototypic cubic phase I of s. g. Pm3m transforms at 320 K to the
tetragonal phase II, s. g. P4/mbm, and further at 315 and 311 K
Fig. 7 Compression of structural parameters of CsPbCl3: unit-cell volume and lattice parameters of CsPbCl3 (a), the average Pb–Cl length of
three independent bonds (b), and Pb–Cl–Pb angles (c). Where not indicated, the standard deviations are smaller than the plotted points.
10774 | J. Mater. Chem. A, 2021, 9, 10769–10779
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry A
to the orthorhombic phases III (s. g. Cmcm) and IV (s. g. Pnma),
respectively. These transitions occur above the room tempera-
ture and their Tc values raise with pressure.20 All our high-
pressure experiments were performed at room temperature,
i.e. below these phase transitions. The SCXRD studies (Fig. 7)
were limited to about 0.65 GPa because of profound deteriora-
tion of the quality of samples under higher pressures. We
observed single defects forming under pressure of c.a. 0.5 GPa,
but sometimes even at lower pressure, depending on the crystal
sample. The microscopic observation of the pressure- and time-
dependent erosion in a single crystal is illustrated in Fig. 8a. In
the pressure range below 0.4 GPa the crystal was stable without
any noticable formation of defects, at least within several days.
The rst single defects captured under optical microscope
appeared at 0.48 GPa, but their size and distribution were
hardly time-dependent. Only the pressure increase to 0.70 GPa
resulted in a quick development of the pulverised portions.
These observations indicate that the crystal undergoes a slow
transformation associated with a large lattice strain. The rate of
this transformation varied from sample to sample between
hours to even weeks, suggesting a key role of defects formed at
the crystallization stage.
To conrm the pressure-induced slow transformation of the
CsPbCl3 crystal structure, we performed two series of X-ray
powder diffraction experiments applying two different modes
of compression. In the rst scenario, aer the data collection in
phase IV at 0.11 GPa, pressure was increased to 0.65 GPa, i.e. to
the range of the crystal slow transformation. Under this pres-
sure the changes in the diffraction pattern were monitored as
a function of time during two weeks. The representative PXRD
patterns in Fig. 8b show that just aer the pressure increase to
0.65 GPa the sample remained in phase IV, but in the course of
time substantial changes arose. The increased number of
reections, and particularly the low-2q ones, indicate that the
crystal transforms to a lower-symmetry phase V. This slowly
formed phase V remains stable at still higher pressures, as
evidenced by the PXRD measurements at 2.20 and 3.88 GPa
(Fig. 8b). The pressure release to 0.07 GPa resulted in a reverse
transition to phase IV, but a full restoration of the initial phase
required some time as testied by a coexistence of phases IV
and V just aer the pressure release. Nevertheless, the slow-
kinetics transition in CsPbCl3 is fully reversible as opposed to
CsPbBr3.
In the second series of the high-pressure PXRD measure-
ments, aer the data collection for phase IV at 0.46 GPa, the
pressure was increased immediately to 2.20 GPa. The results
plotted in Fig. 8c reveal a transition of phase IV to another high-
pressure phase VI, distinctly different than phase V. Both
phases V and VI can exist in a similar pressure range, but their
formation depends on the compression route. The sample
remained in phase VI for a long time and at still higher pres-
sure, which is consistent with the earlier PXRD study,27 most
likely performed in a quick compression mode, as most of
synchrotron measurements. Aer we released pressure to
0.08 GPa phase IV was restored (Fig. 8c). Compared to phase IV,
the PXRD patterns of phase VI contain less reections, all of
them strongly broadened, as well as clearly increased diffused
Fig. 8 Time-dependent transformation of a single crystal of CsPbCl3
compressed in the DAC (a), as well as PXRD patterns of CsPbCl3 slowly
transforming between phases IV and V at 0.65 GPa (b), and of CsPbCl3
quickly compressed to phase VI (c), as described in the text.
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 10769–10779 | 10775

View Article Online
Journal of Materials Chemistry A
Paper
scattering. These features indicate a partial reduction of the
crystal structure long-range order, and they are similar to those
in 3D-CsPbBr3 phase IV.
Pressure dependence of the absorption edge in CsPbCl3
The absorption spectra of CsPbCl3 were measured for a thin
single-crystal plate. The pressure was increased quickly to
prevent the transition to phase V, which would hamper trans-
missive measurements. During the measurements in the
compression and decompression cycles the crystal optical
quality was controlled by using the microscope. The exemplary
spectra are plotted in Fig. 9a. The compression of the crystal
resulted in a gradual red shi of the absorption edge up to
about 1.3 GPa, where the onset of the transition to phase VI
takes place. This transition is associated with a large jump-wise
blue shi of the edge. The spectrum recorded at 1.29 GPa shows
a clear coexistence of phases IV and VI, manifested by two
absorption edges corresponding to these phases. This coexis-
tence was observed in the phase transition region within 0.2–
0.3 GPa. The subsequent increase of pressure up to 4.13 GPa
caused a continuous blue shi of the edge. According to the
absorption edge behavior, the pressure dependence of the
energy gap (Fig. 9b) shows a jump-wise change at the transition
point, indicating a strong rst-order character of the transition.
This is contrary to the previously reported27 continuous changes
of the bandgap (green points in Fig. 9b) and of the position of
the photoluminescence band. These signicant discrepancies
in the energy gap of 3D-CsPbCl3 phase VI between our and the
literature data can be due to anisotropic strain generated by
silicone oil (hydrostatic to about 0.5 GPa only) used in the
previous experiments.27 The linear decrease of Eg as a function
of pressure up to the transition point well correlates with the
linear pressure dependence of the average Pb–Cl distance
(Fig. 7b). It testies that in the compressed 3D-CsPbCl3 phase IV
the Pb–Cl bonds shortening is a dominant structural change
absorbing the pressure-generated strain and determining the
electronic bands responsible for the absorption edge.
The energy gap values determined during the decompres-
sion of CsPbCl3 crystal are substantially larger than those
during the compression (see Fig. 9b). This pressure-memory
effect is similar to that in CsPbBr3. Although in CsPbCl3 this
effect disappears aer the reverse transition to phase IV, the
features of absorption spectra recorded at 0.83 and 0.48 GPa
clearly indicate that only a part of the crystal sample trans-
formed back to phase IV, while the other part remained in partly
amorphous phase VI. Some traces of the high-pressure phase
were also seen just aer the pressure release, and only aer c.a.
0.5 h the complete restoration of the initial spectrum, charac-
teristic of phase IV, was recorded.
Fig. 9 Absorption edge of CsPbCl3 under pressure: (a) selected
absorption spectra of a single-crystal plate of 3D-CsPbCl3 recorded
during its compression (bottom) and decompression (top), shifted
vertically to avoid their overlapping. (b) The pressure dependence of
the energy gap compared with the literature data.27
The understanding of changes observed in the diffraction
patterns and in the absorption spectra of CsPbCl3, induced by
the transition at 1.3 GPa, requires a deeper analysis. It seems
that the postulated isostructural phase transition is insufficient
to explain these changes, especially in the light of the disput-
able evidence for step-wise volume change,25 necessary for such
type of transition. The crucial are the amorphization symptoms,
detected in the powder-diffraction experiments (Fig. 8c), which
indicate a local structure collapse, resulting in a distribution of
the Pb–Cl lengths and Pb–Cl–Pb angles. Such distortions in the
framework of PbCl6 octahedra leads to the broadened diffrac-
tion peaks situated at similar positions as in phase IV, and to
10776 | J. Mater. Chem. A, 2021, 9, 10769–10779
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry A
the increase in the background scattering. But most impor- of the crystals nucleation was controlled to start below 315 K for
tantly, the abrupt collapse of the crystal structure profoundly CsPbCl3 and below 360 K for CsPbBr3, in order to avoid the
modies the electronic bands of the material and its optical crystals twinning resulting from the ferroelastic phase transitions
properties.
occurring in these compounds at higher temperatures.30,37
High-pressure single-crystal and powder X-ray diffraction
Conclusions
The single-crystal X-ray diffraction (SCXRD) and powder X-ray
diffraction (PXRD) high-pressure experiments have been per-
formed with a Merrill–Bassett diamond anvil cell (DAC)38
modied by mounting the diamond anvils (with 0.8 mm culets)
directly on steel supporting discs with conical windows. The
gaskets were made of 0.1–0.3 mm thick tungsten foil with
a spark-eroded hole 0.3–0.4 mm in diameter. The carefully
selected single crystals of CsPbBr3 or CsPbCl3 were glued to the
culet of one anvil. For powder-diffraction experiments the
crystals were pulverized just before loading to the pressure cell.
A ruby chip glued inside the pressure chamber served for the
pressure calibration by the uorescence method.39 Isopropanol
and Galden were alternatively used as the hydrostatic uids. All
high-pressure X-ray diffraction experiments were performed
using an Oxford Diffraction Gemini A Ultra diffractometer
operating with graphite-monochromated MoKa radiation. For
the high-pressure cell centering the gasket-shadowing method40
was applied. The CrysAlisPro soware41 was used for the data
collection and processing. The structures were solved and
rened using SHELX soware.42 Crystallographic information
les (CIFs) for CsPbCl3 (2060421–2060427) and CsPbBr3
(2060567–2060576) have been deposited with the Inorganic
Crystal Structure Database.
Our systematic study shows that under pressure perovskites
CsPbCl3 and CsPbBr3 are prone to time-dependent structural
transformations, which proceed in various time scales. No such
pronounced kinetic effects were observed for these crystals as
a function of temperature, although they undergo a series of
phase transitions at ambient pressure. Both CsPbCl3 and
CsPbBr3 undergo pressure-induced transitions that differ in
character, depending on the time scale and method of the
applied external stimuli. The slow-kinetics transformations
shed new light on the stress-related response and stability of
these all-inorganic perovskite structures. Our study shows the
importance of genuinely hydrostatic conditions for the clarity
and correctness of the discussion on the structural and optical
properties of these materials. The hydrostatic results can dras-
tically differ from the high-pressure experiments performed in
non-hydrostatic conditions, systematically affected by the
pressure gradients and regions under different anisotropic
stress in the sample, which confuse the conclusions assuming
the homogenous stress and uniform sample. The non-
hydrostatic effects can in turn lead to heuristic rules or to
advanced modelling and simulations based on the averaged
crystal structures strained in a different way than the samples
for which the physical properties were recorded. There are many
such reports in prominent journals, which confuse the trans-
parent description of the mechanisms of strain-energy gap
relations in the important eld of perovskites for photovoltaic
and optoelectronic applications. Our present report only signals
these problems for two all-inorganic perovskites. The obtained
results show that the long-time kinetics transformations
revealed in CsPbCl3 and CsPbBr3 have much in common with
the previously described transformations in analogous hybrid
perovskites [CH3NH3]PbX3.31 Surely, more careful research on
the structure and properties of perovskite photovoltaic mate-
rials is needed.
The PXRD analyses of the samples recovered from the high-
pressure cells were carried out on a high-resolution Bruker D8
diffractometer equipped with a Johansson monochromator
ꢀ
selecting the characteristic l(CuKa1) ¼ 1.54059 A line. The
sample was gently smeared over about 4 mm2 of the central
surface of a single-crystal silicon plate, oriented to eliminate the
background. In this way high-quality PXRD patterns were
recorded even for very small amount of the sample.
Dielectric measurements under pressure
High-pressure dielectric studies were performed on the pellets
of compressed polycrystalline samples. Gold electrodes were
deposited on the at surfaces of the pellets by the sputtering
method using a Scancoat SIX (BOC Edwards). Additionally, the
electrodes were covered with a silver paste to attach the wires.
The sample was mounted in a beryllium–copper high-pressure
cell lled with gaseous helium. The pressure was generated by
a gas compressor U11 (Unipress) and calibrated by means of
a manganin gauge with an accuracy of ꢁ3 MPa. The tempera-
ture was controlled inside the cell by a copper–constantan
thermocouple. Measurements of electric permittivity were
carried out with a Hewlett-Packard 4192A impedance analyzer
Methods
Synthesis and crystals growth
Crystals CsPbBr3 and CsPbCl3 were grown by a low-temperature
method from the aqueous solutions, as described previ-
ously.14,29 Cesium bromide and chloride were prepared through
reaction between Cs2CO3 (Aldrich, 99%) and HBr (Acros
Organics, 48% water solution) or HCl (POCH, 35% water solu-
tion, specially pure), respectively, while PbBr2 and PbCl2 were
synthesized in the reaction between Pb(CH3CO2)2$3H2O (Aldrich,
99.999%) and the appropriate acid. The precursor solutions were
prepared by dissolving stoichiometric amounts of CsBr and
PbBr2 or CsCl and PbCl2 in hot strongly acidied water. The
at a rate of temperature changes 0.5 K minꢀ1
.
Optical absorption spectra
solutions were saturated and le to crystallize during slow cool- The optical absorption spectra were measured using a Jasco
ing and water evaporation at room temperature. The temperature MSV-5100 microspectrometer in the 300–600 nm range. The
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 10769–10779 | 10777

View Article Online
Journal of Materials Chemistry A
Paper
measurements were performed on single-crystal plates 2–3 mm 16 C. Zhu, X. Niu, Y. Fu, N. Li, C. Hu, Y. Chen, X. He, G. Na,
thick, placed in the DAC equipped with type IIa diamond anvils.
The pressure chamber was lled with isopropanol. The absor-
bance was measured using the incident beam diameter of 30
P. Liu, H. Zai, Y. Ge, Y. Lu, X. Ke, Y. Bai, S. Yang, P. Chen,
Y. Li, M. Sui, L. Zhang, H. Zhou and Q. Chen, Nat.
Commun., 2019, 10, 815.
mm, the continuous scan speed of 200 nm minꢀ1, and the 17 Y. Chen, Y. Lei, Y. Li, Y. Yu, J. Cai, M.-H. Chiu, R. Rao, Y. Gu,
spectral bandwidth 5 nm. Direct bandgap Tauc plots43 were
applied for the energy gap (Eg) determination.
C. Wang, W. Choi, H. Hu, C. Wang, Y. Li, J. Song, J. Zhang,
B. Qi, M. Lin, Z. Zhang, A. E. Islam, B. Maruyama,
S. Dayeh, L.-J. Li, K. Yang, Y.-H. Lo and S. Xu, Nature,
2020, 577, 209–215.
Conflicts of interest
18 S. Shrestha, R. Fischer, G. J. Matt, P. Feldner, T. Michel,
A. Osvet, I. Levchuk, B. Merle, S. Golkar, H. Chen,
There are no conicts to declare.
¨
¨
S. F. Tedde, O. Schmidt, R. Hock, M. Ruhrig, M. Goken,
W. Heiss, G. Anton and C. J. Brabec, Nat. Photonics, 2017,
11, 436–441.
Acknowledgements
M. S. acknowledges the nancial support from the Polish 19 S. Tie, W. Zhao, D. Xin, M. Zhang, J. Long, Q. Chen, X. Zheng,
National Science Centre, Grant Opus 16 No. 2018/31/B/ST3/
02188.
J. Zhu and W. -H. Zhang, Adv. Mater., 2020, 32, 2001981.
20 K. Gesi, K. Ozawa and S. Hirotsu, J. Phys. Soc. Jpn., 1975, 38,
463–466.
21 U. Schwarz, H. Hillebrecht, M. Kaupp, K. Syassen and
H.-G. von Schnering, J. Solid State Chem., 1955, 118, 20–27.
References
1 A. Kojima, K. Teshima, Y. Shirai and T. Miyasaka, J. Am. 22 U. Schwarz, F. Wagner, K. Syassen and H. Hillebrecht, Phys.
Chem. Soc., 2009, 131, 6050–6051. Rev. B: Condens. Matter Mater. Phys., 1996, 53, 2545–12548.
2 H.-S. Kim, C.-R. Lee, J.-H. Im, K.-B. Lee, T. Moehl, 23 G. Xiao, Y. Cao, G. Qi, L. Wang, C. Liu, Z. Ma, X. Yang, Y. Sui,
A. Marchioro, S.-J. Moon, R. Humphry-Baker, J.-H. Yum,
W. Zheng and B. Zou, J. Am. Chem. Soc., 2017, 139, 10087–
¨
J. E. Moser, M. Gratzel and N.-G. Park, Sci. Rep., 2012, 2, 1–7.
10094.
3 H. J. Snaith, J. Phys. Chem. Lett., 2013, 4, 3623–3630.
4 Best Research-Cell Efficiencies, https://www.nrel.gov/pv/
24 L. Zhang, Q. Zeng and K. Wang, J. Phys. Chem. Lett., 2017, 8,
3752–3758.
assets/pdfs/best-research-cell-efficiencies.
www.nrel.gov/pv/assets/pdfs/best-research-cell-
efficiencies.20200925.pdf, accessed: November 2020.
https:// 25 Y. Nagaoka, K. Hills-Kimball, R. Tan, R. Li, Z. Wang and
O. Chen, Adv. Mater., 2017, 29, 160666.
26 G. Yuan, S. Qin, X. Wu, H. Ding and A. Lu, Phase Transform.,
5 P. Holzhey, P. Yadav, S.-H. Turren-Cruz, A. Ummadisingu,
2017, 91, 1–10.
M. Gratzel, A. Hagfeldt and M. Saliba, Mater. Today, 2019, 27 L. Zhang, L. Wang, K. Wang and B. Zou, J. Phys. Chem. C,
29, 10–19. 2018, 122, 15220–15225.
6 A. Aziz, N. Aristidou, X. Bu, R. J. E. Westbrook, S. A. Haque 28 D. Liu, Q. Li, H. Jing and K. Wu, RSC Adv., 2019, 9, 3279–
and M. S. Islam, Chem. Mater., 2020, 32, 400–409.
3284.
7 K. Domanski, E. A. Alharbi, A. Hagfeldt, M. Gratzel and 29 C. K. Møller, Mat.-Fys. Medd. - K. Dan. Vidensk. Selsk., 1959,
W. Tress, Nat. Energy, 2018, 3, 61–67. 32(No. 2), 1–27.
8 M. Kulbak, S. Gupta, N. Kedem, I. Levine, T. Bendikov, 30 S. Hirotsu, J. Harada, M. Iizumi and K. Gesi, J. Phys. Soc. Jpn.,
G. Hodes and D. Cahen, J. Phys. Chem. Lett., 2016, 7, 167–172. 1974, 37, 1393–1398.
9 X. Li, Y. Tan, H. Lai, S. Li, Y. Chen, S. Li, P. Xu and J. Yang, 31 M. Szafranski and A. Katrusiak, J. Phys. Chem. Lett., 2017, 8,
ACS Appl. Mater. Interfaces, 2019, 11, 29746–29752. 2496–2506.
10 K. Lin, J. Xing, L. N. Quan, F. P. Garcıa de Arquer, X. Gong, 32 A. Marstrander and C. K. Møller, Mat.-Fys. Medd. - K. Dan.
J. Lu, L. Xie, W. Zhao, D. Zhang, C. Yan, W. Li, X. Liu,
Vidensk. Selsk., 1966, 35(No. 5), 1–12.
Y. Lu, J. Kirman, E. H. Sargent, Q. Xiong and Z. Wei, 33 O. Nazarenko, M. R. Kotyrba, M. Worle, E. Cuervo-Reyes,
¨
¨
´
´
¨
Nature, 2018, 562, 245–248.
11 J. Sun, J. Yang, J. I. Lee, J. H. Cho and M. S. Kang, J. Phys.
Chem. Lett., 2018, 9, 1573–1583.
S. Yakunin and M. V. Kovalenko, Inorg. Chem., 2017, 56,
11552–11564.
¨
34 M. Aebli, B. M. Benin, K. M. McCall, V. Morad, D. Thony,
¨
12 H. Zhang, J. Yao and H. Fu, Chem. Mater., 2020, 32, 9602–
9608.
H. Grutzmacher and M. V. Kovalenko, Helv. Chim. Acta,
2020, 103, e2000080.
13 X. He, P. Liu, H. Zhang, Q. Liao, J. Yao and H. Fu, Adv. Mater., 35 A. Noculak, S. C. Boehme, M. Aebli, Y. Shynkarenko,
2017, 29, 1604510.
14 C. C. Stoumpos, C. D. Malliakas, J. A. Peters, Z. Liu,
K. M. McCall and M. V. Kovalenko, Helv. Chim. Acta, 2021,
104, e2000222.
M. Sebastian, J. Im, T. C. Chasapis, A. C. Wibowo, 36 L. Kong, G. Liu, J. Gong, Q. Hu, R. D. Schaller, P. Dera,
D. Y. Chung, A. J. Freeman, B. W. Wessels and
M. G. Kanatzidis, Cryst. Growth Des., 2013, 13, 2722–2727.
15 X. Li, Y. Luo, M. V. Holt, Z. Cai and D. P. Fennin, Chem.
Mater., 2019, 31, 2778–2785.
D. Zhang, Z. Liu, W. Yang, K. Zhu, Y. Tang, C. Wang,
S.-H. Wei, T. Xu and H.-k. Mao, Proc. Natl. Acad. Sci. U. S.
A., 2016, 113, 8910–8915.
10778 | J. Mater. Chem. A, 2021, 9, 10769–10779
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry A
37 H. Ohta, J. Harada and S. Hirotsu, Solid State Commun., 41 CrysAlisPro: Data Collection and Processing Soware for X-Ray
1973, 13, 1969–1972.
38 L. Merrill and W. A. Basset, Rev. Sci. Instrum., 1974, 45, 290–
294.
39 G. J. Piermarini, S. Block, J. D. Barnett and R. A. Forman, J.
Appl. Phys., 1975, 46, 2774–2780.
40 A. Budzianowski and A. Katrusiak, in High-Pressure
Diffractometers, Agilent Technologies, Santa Clara, CA, 2010;
Oxford Diffraction CrysAlisPro version 171.34.40: Data
Collection and Processing Soware for X-Ray Diffractometers,
Agilent Technologies, Santa Clara, CA, 2010.
42 G. M. Sheldrick, A Short History of SHELX, Acta Crystallogr.,
Sect. A: Found. Crystallogr., 2008, 64, 112–122.
Crystallography, ed. A. Katrusiak and P. F. McMillan, 43 J. Tauc, Mater. Res. Bull., 1968, 3, 37–46.
Kluwer Academic Publishers, Dordrecht, The Netherlands,
2004, pp. 101–112.
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 10769–10779 | 10779