Strong Electron-Phonon Coupling and Self-Trapped Excitons in the Defect Halide Perovskites A3M2I9 (A = Cs, Rb; M = Bi, Sb)

Article abstract of DOI:10.1021/acs.chemmater.7b01184

The optical and electronic properties of Bridgman grown single crystals of the wide-bandgap semiconducting defect halide perovskites A₃M₂I₉ (A = Cs, Rb; M = Bi, Sb) have been investigated. Intense Raman scattering was observed at room temperature for each compound, indicating high polarizability and strong electron-phonon coupling. Both low-temperature and room-temperature photoluminescence (PL) were measured for each compound. Cs₃Sb₂I₉ and Rb₃Sb₂I₉ have broad PL emission bands between 1.75 and 2.05 eV with peaks at 1.96 and 1.92 eV, respectively. The Cs₃Bi₂I₉ PL spectra showed broad emission consisting of several overlapping bands in the 1.65-2.2 eV range. Evidence of strong electron-phonon coupling comparable to that of the alkali halides was observed in phonon broadening of the PL emission. Effective phonon energies obtained from temperature-dependent PL measurements were in agreement with the Raman peak energies. A model is proposed whereby electron-phonon interactions in Cs₃Sb₂I₉, Rb₃Sb₂I₉, and Cs₃Bi₂I₉ induce small polarons, resulting in trapping of excitons by the lattice. The recombination of these self-trapped excitons is responsible for the broad PL emission. Rb₃Bi₂I₉, Rb₃Sb₂I₉, and Cs₃Bi₂I₉ exhibit high resistivity and photoconductivity response under laser photoexcitation, indicating that these compounds possess potential as semiconductor hard radiation detector materials.

Full text of DOI:10.1021/acs.chemmater.7b01184

Article
pubs.acs.org/cm
Strong Electron−Phonon Coupling and Self-Trapped Excitons in the
Defect Halide Perovskites A M I (A = Cs, Rb; M = Bi, Sb)
3
2 9
Kyle M. McCall,†,‡ Constantinos C. Stoumpos, Svetlana S. Kostina, Mercouri G. Kanatzidis,
†
‡
†
,‡
and Bruce W. Wessels*
†
‡
Department of Chemistry and Department of Materials Science and Engineering, Northwestern University, Evanston, Illinois
0208, United States
6
*
S Supporting Information
ABSTRACT: The optical and electronic properties of Bridgman
grown single crystals of the wide-bandgap semiconducting defect
halide perovskites A M I (A = Cs, Rb; M = Bi, Sb) have been
3
2 9
investigated. Intense Raman scattering was observed at room
temperature for each compound, indicating high polarizability
and strong electron−phonon coupling. Both low-temperature
and room-temperature photoluminescence (PL) were measured
for each compound. Cs Sb I and Rb Sb I have broad PL
3
2 9
3
2 9
emission bands between 1.75 and 2.05 eV with peaks at 1.96 and
.92 eV, respectively. The Cs Bi I PL spectra showed broad
1
3
2 9
emission consisting of several overlapping bands in the 1.65−2.2
eV range. Evidence of strong electron−phonon coupling
comparable to that of the alkali halides was observed in phonon
broadening of the PL emission. Effective phonon energies obtained from temperature-dependent PL measurements were in
agreement with the Raman peak energies. A model is proposed whereby electron−phonon interactions in Cs Sb I , Rb Sb I ,
3
2 9
3
2 9
and Cs Bi I induce small polarons, resulting in trapping of excitons by the lattice. The recombination of these self-trapped
3
2 9
excitons is responsible for the broad PL emission. Rb Bi I , Rb Sb I , and Cs Bi I exhibit high resistivity and photoconductivity
3
2 9
3
2 9
3
2 9
response under laser photoexcitation, indicating that these compounds possess potential as semiconductor hard radiation
detector materials.
1
. INTRODUCTION
and X is a halide anion coordinated to the metal, forming a
framework of corner-sharing MX octahedra. Defect perovskites
6
Heavy metal−halide semiconductors have long been studied for
their optical and electronic properties.
halide perovskites intensified in 2009 with the discovery that
CH NH PbI was an efficient, low-cost light absorber for solar
cells, and since then devices based on these materials have
achieved efficiencies as high as 22%. Subsequently, a number
of organic−inorganic halide perovskite materials have shown
potential applications in hard radiation detection, LEDs,
1
−4
are derivative structures that possess the same corner-sharing
However, interest in
MX octahedra as the AMX aristotype but form different
6
3
structures due to cation deficiencies in order to satisfy charge
3
3
3
21
5
balance restrictions. Tetravalent cations form A M□X
2
6
6
structures (□ represents a vacancy) with 1/2 occupancy of
the M sites in the A M X perovskite formula, while trivalent
2
2 6
7
8,9
cations form A M □X structures with 2/3 occupancy of the M
3 2 9
25,26
9
,10
11
sites of the A M X perovskite formula.
In particular, the
3
3 9
lasing, and photocatalysis. These successes have motivated
studies of related hybrid materials such as the two-dimensional
A M X materials have recently attracted a great deal of interest
3
2 9
23,24,27−31
as solar cell materials.
This group of compounds
(
2D) organic−inorganic lead iodide analogues for increased
12
includes perovskite derivatives containing bismuth iodide,
which is a well-studied wide-bandgap semiconductor with
stability. However, the application of these materials has been
1
3
14
limited by the presence of toxic lead and stability issues
related to the organic cation. To bypass the stability issue, we
and other groups investigated the inorganic lead halide
perovskites CsPbX (X = I, Br, Cl), which have shown promise
for solar cells
32
potential applications in hard radiation detection and as the
3
3
active layer in photovoltaics. Bismuth is an especially
attractive candidate for replacing lead due to the isoelectronic
nature of the Bi3+ ion to Pb and its uniquely low toxicity
relative to the neighboring heavy metals Tl, Hg, and Pb.
Similarly, antimony is less toxic than lead while retaining the
3
2+
1
5−17
18−20
and radiation detectors.
However, the
resulting solar cells have not achieved the same performance as
the organic−inorganic perovskites and still contain lead. Thus,
investigators have turned to the so-called defect perovskites as
an avenue for eliminating toxic lead and improving device
21
Received: March 23, 2017
Revised: April 10, 2017
2
2−24
performance.
The halide perovskites share the formula
AMX where A is a monovalent cation, M is a divalent metal,
3
©
XXXX American Chemical Society
A
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
2
ns lone pair which is partially responsible for the exceptional
°C oven, yielding 34 g (95%) of Cs
3
Sb
2
I
9
. The obtained powder was
34
−3
sealed in a quartz ampule (13 mm o.d.) under vacuum (10 mbar)
and melted in a box furnace at 650 °C to form a dark red ingot.
Cs Bi I . Bi O (20 mmol, 9.32 g) was dissolved in 80 mL of 57%
performance of lead-based compounds. Thus, we chose to
investigate optical and electrical properties of the bismuth
iodide family of A Bi I compounds and the associated Sb-
analogues A Sb I , where A = Cs, Rb is an alkali metal.
Detailed single crystal structures have been reported for
3
2 9
2
3
3
2 9
aqueous HI to produce a bright red/orange solution in an exothermic
reaction. CsI (60 mmol, 15.59 g) was dissolved in 20 mL of water,
producing a colorless solution in an endothermic reaction. Addition of
the CsI solution to the Bi O solution led to spontaneous precipitation
3
2 9
3
5
36
24,37
Cs Bi I , Cs Sb I , and Rb Bi I ,
and very recently for
3
2 9
3
2 9
3
2 9
2
3
3
8
Rb Sb I . Optical measurements indicate these compounds
3
2 9
of a bright red solid in an essentially colorless supernatant liquid (pale
are semiconductors with wide band gaps in the 1.89−2.06 eV
yellow because of dissolved I ). The addition was done slowly under
constant stirring to avoid local concentration of CsI. After cooling to
room temperature, the solid was filtered through a fritted-dish funnel
2
10
11
range and accordingly they exhibit high resistivities (10 −10
2
5,39
Ω·cm).
The compound Cs Bi I (E = 2.06 eV) does not
3
2 9
g
under vacuum. The powder was washed with 100 mL of 5:1 HI:H O
form in a perovskite-derivative structure, as the BiI octahedra
share faces to form discrete [Bi I ] molecules, resulting in a
2
6
3
−
solution followed by 300 mL of methanol and dried overnight under
vacuum. The solid was further dried in an oven at ∼70 °C in air, with a
final yield of 38 g (97%). The obtained powder was sealed in a quartz
2
9
zero-dimensional (0D) molecular salt crystal structure (space
group P6 /mmc). Cs Sb I can form either a hexagonal
−3
3
3
2 9
ampule (13 mm o.d.) under vacuum (10 mbar) and melted in a box
3−
[
Sb I ] molecular dimer phase or a 2D layered structure
2 9
furnace at 650 °C to form a dark red ingot.
depending on the synthesis method. Crystals grown from a
concentrated HI solution form the dimer phase isostructural to
Cs Bi I , whereas crystals grown from the melt form a 2D
SbI . I (150 mmol, 38.07 g, Sigma-Aldrich, 3 N) and Sb shots (100
3
2
mmol, 12.18 g, Sigma Alrich, 5 N) were refluxed in 300 mL of toluene
overnight, yielding a clear orange solution. The solution was filtered
hot and left to cool slowly to room temperature, yielding red flakes.
The resulting solid was filtered under vacuum, washed with cold
toluene, and dried overnight. The obtained powder was sealed in a
3
2 9
3
6
̅
layered perovskite derivative structure (space group P3m1).
The single crystals used in the present study were grown by the
Bridgman method; therefore, only the layered structure of
−3
40,36
quartz ampule (13 mm o.d.) under vacuum (10 mbar) and sublimed
in a tube furnace with the hot zone set to 350 °C, yielding 45 g (90%)
of pure material. This process is required to remove residual Sb O
Cs Sb I is considered (E = 1.89 eV).
The compounds
3
2 9
g
Rb Bi I (E = 1.93 eV) and Rb Sb I (E = 2.03 eV) are
3
2 9
g
3
2 9
g
2
3
isostructural, with both forming distorted monoclinic layers
space group P2 /n) where the MI octahedra adopt an
and SbO I impurities.
x y
(
1
6
BiI . Bi(NO ) ·5H O (100 mmol, 48.51 g) was dissolved in 300 mL
3 3 3 2
irregular coordination approximating a rhombohedral elonga-
of warm ultrapure (Millipore) water acidified by 40 mL of
concentrated nitric acid, forming a colorless clear solution. In a
different beaker, KI (300 mmol, 49.80 g) was dissolved in 50 mL of
ultrapure (Millipore) water to form a colorless clear solution. Rapid
addition of the KI solution to the Bi(NO ) solution immediately
2
4
tion. Despite the importance of the related Pb-based
perovskites, the fundamental properties beyond band gaps
and resistivities are not well understood in these materials.
In this work, we describe the structural, optical, and
electronic properties of these heavy metal iodide perovskite
derivatives. Large single crystals of the A M I (A = Cs, Rb; M
3
3
resulted in a fine black precipitate and a colorless supernatant solution.
The BiI powder was filtered under vacuum, washed copiously with
3
3
2 9
ultrapure (Millipore) water, and dried overnight. Subsequently, the
BiI3 was washed with a dilute HI solution (1:20 concentrated
HI:water) to ensure removal of the nitrate ions and then refiltered and
rewashed as above. Yield was 55 g after purification (∼93%).
Rb Sb I . SbI (12.30 mmol, 6.18 g) and RbI powders (18.46 mmol,
=
Bi, Sb) materials were grown using the Bridgman method.
The crystal structure of Rb Sb I has been determined by single
3
2 9
crystal X-ray diffraction and is isostructural to Rb Bi I . The
3
2 9
structural relations of these compounds are used to interpret
the strong phonon modes observed via Raman spectroscopy. In
situ synchrotron X-ray diffraction over the melt temperature
range demonstrated that Cs Bi I is incongruently melting.
3
2 9
3
3.92 g, Sigma-Aldrich, 3 N) were crushed together using mortar and
pestle, sealed in a quartz ampule (13 mm o.d.) under vacuum and
reacted in a box furnace at 650 °C.
3
2 9
Rb Bi I . BiI (31.37 mmol, 18.5 g) and RbI powders (47.07 mmol,
0 g, Sigma-Aldrich, 3N) were crushed together using a mortar and
3
2 9
3
Comprehensive low-temperature photoluminescence (PL)
measurements are coupled with Raman spectroscopy to
demonstrate strong electron−phonon coupling. We propose a
model whereby small polarons are induced by the electron−
phonon interactions and trap charge carriers, resulting in self-
trapped excitons. Phonon-assisted recombination of these self-
trapped excitons is responsible for the broad PL emission bands
seen in Cs Bi I , Cs Sb I , and Rb Sb I . Photoconductivity
1
pestle, sealed in a quartz ampule (13 mm o.d.) under vacuum, and
reacted in a box furnace at 650 °C.
Crystal Growth. Crystal growth was performed using the
powdered raw materials prepared as described above. In a typical
procedure 6−20 g of material was sealed in quartz ampules (13 mm
−3
o.d.) under vacuum (10 mbar). Ingots were grown in a 2-zone
Bridgman furnace with the hot zone at 800 °C and the cold zone at
3
2 9
3
2 9
3
2 9
4
00 °C. The translation speed was 2 mm/h. The resulting ingots were
measurements under laser illumination indicate a strong
cleaved parallel to the layer plane (parallel to the growth direction for
photoresponse in Cs Bi I , Rb Bi I , and Rb Sb I , suggesting
3
2 9
3
2 9
3
2 9
Cs M I , approximately at a 30° angle for Rb M I ), coinciding with
3
2 9
3
2 9
that these materials are potential candidates for semiconductor
hard radiation detectors.
the van der Waals gap between the layers. The resulting thin wafers
(
0.2−2 mm) had mirror-like surfaces that required no further
processing.
2. EXPERIMENTAL METHODS
Powder X-ray Diffraction. Samples from the as-grown ingots
were crushed into fine powder for X-ray diffraction measurements.
XRD spectra were measured using a Rigaku Miniflex600 diffractometer
with a Cu Kα source (λ = 1.5406 Å) operating at 40 kV and 15 mA
with a Kβ foil filter.
Synthesis. Cs Sb I . Sb O (20 mmol, 5.83 g) was dissolved in 80
3
2 9
2
3
mL of 57% aqueous HI to produce a bright red/orange solution in an
exothermic reaction. Cs CO (30 mmol, 9.77 g) was added to 20 mL
2
3
of 57% aqueous HI to produce a colorless solution in an endothermic
reaction. Addition of the CsI solution to the Sb O solution leads to
Single-Crystal X-ray Diffraction. Single-crystal X-ray diffraction
experiments were performed at 300 K using a STOE IPDS II
diffractometer using Mo Kα radiation (λ = 0.71073 Å) operating at 50
kV and 40 mA. Integration and numerical absorption corrections were
performed using the X-AREA, X-RED, and X-SHAPE programs. Using
2
3
spontaneous precipitation of a bright orange solid in a pale yellow
solution. After cooling to room temperature, the solid was removed by
vacuum filtration through a fritted-dish funnel. The solid was washed
with 100 mL of 5:1 HI:H O solution followed by 300 mL of methanol
2
4
1
and dried overnight under vacuum. The solid was further dried in a 70
the Olex2 crystallography software package, the structure was solved
B
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
4
2
with the ShelXT structure solution program using Direct Methods
through a shadow mask using an e-beam evaporator (Edwards Auto
306). Cu wires (0.025 mm) were bonded to the sample contacts using
carbon paint (Ted Pella graphite), and these wires were attached to Cu
tape on a glass substrate. Alternatively, the Cu wires were attached
43
and refined with the ShelXL refinement package using Least Squares
minimization.
Scanning Electron Microscopy and Energy-Dispersive X-ray
Spectroscopy (SEM/EDX). SEM images and EDX analyses of the as-
grown cleaved crystal surfaces were acquired using a Hitachi S3400N-
II scanning electron microscope equipped with an Oxford Instruments
INCAx-act SDD EDS detector at an accelerating voltage of 25 kV. To
avoid excessive sample charging, measurements were taken under a
pressure of 40 Pa with the backscattering detector.
directly to the sample surface using carbon paint as a contact. The high
resistivity of these compounds (101
0−12
Ω·cm) permits the use of
carbon paint contacts, as the contact resistivity (104 Ω·cm) is well
within experimental error for these measurements.
−5
Resistivity and photoconductivity were measured in a setup using a
voltage source (Stanford Research Systems PS325) and an ammeter
(Keithley 2636A SourceMeter). For photoconductivity measurements,
In Situ Synchrotron X-ray Diffraction. High resolution
synchrotron powder X-ray diffraction data were collected at temper-
atures ranging from 25 to 600 °C using beamline 11-BM at the
Advanced Photon Source, Argonne National Laboratory, using an
average wavelength of 0.413614 Å. Discrete detectors covering an
angular range from −6 to 16° 2θ are scanned over a 28° 2θ range, with
data points collected every 0.001° 2θ and scan speed of 0.01°/s. For
2
a laser (OBIS 405 nm LX, 15,625 mW/cm ) was used as the excitation
2
source. The beam was operated at 7813 mW/cm during the
measurement.
3
. RESULTS AND DISCUSSION
18
3.1. Crystal Structure and Thermal Characterization.
further details on the sample stage see Stoumpos et al.
Single-Crystal XRD for Rb Sb I . Single-crystal X-ray diffraction
Differential Thermal Analysis Measurements. Differential
thermal analysis was performed on powder samples in an evacuated
fused silica tube using a Shimadzu DTA-50 thermal analyzer, with
alumina as a reference material.
3
2 9
was conducted on Rb Sb I at room temperature, and the
3 2 9
structure solution parameters can be found in Tables S1
through S4 (Supporting Information). The crystal structure of
Rb Sb I is monoclinic in the space group P2 /n (unit cell: a =
UV−vis Diffuse Reflectance Measurements. Optical diffuse
reflectance measurements were performed at room temperature using
a Shimadzu UV-3600 PC double-beam, double-monochromator
spectrophotometer operating from 200 to 2500 nm. BaSO4 was
used as a nonabsorbing reflectance reference. The generated
reflectance data was transformed to absorbance via the Kubelka−
3
2 9
1
1
4.591(3) Å, b = 8.1879(16) Å, c = 20.584(4) Å, β = 90.36(3)°,
3
V = 2459.1(8) Å , Z = 4) and is comprised of disordered layers
of corner-sharing SbI octahedra with Rb cations between them
6
(
Figure 1a). This is in good agreement with recent work by
38
4
4
2
Chang et al.
Munk equation α/S = (1 − R) /2R, where R is the reflectance and α
and S are the absorption and scattering coefficients, respectively. The
absorption edge was estimated by extrapolation of the linear regions.
Confocal Raman and Room-Temperature Photolumines-
cence (PL) Measurements. Raman spectroscopy was conducted
using a confocal Horiba LabRAM HR Evolution spectrometer with an
excitation wavelength of 785 nm. To reduce laser damage, the laser
2
intensity (100 mW, ∼1 270 000 mW/cm ) was lowered by a series of
neutral density filters to 12 700 mW/cm for Rb Bi I and Cs Bi I or
to 40 700 mW/cm for Cs Sb I and Rb Sb I .
2
3
2 9
3
2 9
2
3
2 9
3
2 9
Room-temperature PL was measured with the confocal Horiba
LabRAM HR Evolution spectrometer using an above-bandgap
excitation wavelength of 473 nm. To avoid damaging the samples
2
Figure 1. Rb
(b); the unit cell is outlined, and SbI
3
Sb
2
I
9
structure as viewed down b axis (a) and down c axis
octahedra are shaded for clarity.
with the laser, the laser power (25 mW, ∼318 000 mW/cm ) was
2
6
reduced to 3180 mW/cm using neutral density filters. Due to the
weak PL response at room temperature, further decreases of the laser
power led to a loss of signal. At this power the background increases
slightly as the measurement is taken from low to high energy,
indicating some sample damage is occurring.
The compound adopts a distorted variant of the layered
defect perovskite structure, where every third layer of the AMX3
structure is missing due to the 2/3 Bi occupancy of the M site.
Low-Temperature Photoluminescence Measurements. Ex-
citation intensity-dependent and temperature-dependent photolumi-
nescence (PL) were measured for each compound. The sample
mounting procedure and measurement setup were similar to those
The SbI octahedra are asymmetric, as each octahedron shares
6
three I atoms with three neighboring octahedra while the other
three terminate on the layer surface. The shared iodine atoms
have larger Sb−I distances (in the range of 3.15−3.18 Å) than
the terminal iodine atoms, which have smaller distances (2.84−
2.87 Å) nearer to the sum of the covalent radii of Sb and I. This
45
described by us previously.
A continuous wave diode laser (Excelsior One-405 nm, Spectra-
Physics) with a constant power of 50 mW and intensity of 3472 mW/
2
cm at 405 nm was used as the excitation source for PL measurements.
gives rise to the honeycomb motif formed by the SbI stacks
A set of calibrated neutral optical density filters was used to limit the
6
2
when viewed down the c axis (Figure 1b).
laser intensity between 35 and 700 mW/cm for temperature- and
power-dependent measurements. Temperature-dependent measure-
ments were taken at lower intensity to ensure reproducibility. For the
chosen laser powers, repeating the measurements yields the same
signal, indicating that the samples are undamaged. In the case of
Rb Sb I , the power-dependent measurements were performed using
Structure Overview in A M I (A = Rb, Cs; M = Sb, Bi).
3 2 9
Powder X-ray diffraction (Figure S1) and SEM-EDS (Figure
S2, Table S5) analysis both confirm that the compounds are
phase-pure, with no significant deviation from the A M I
3
2 9
3
2 9
stoichiometry observed. Similar to the aristotypical AMX3
perovskite structure, the A M I crystal structures are dictated
an OBIS CW diode laser (OBIS 405 nm LX 100 mW, 15 625 mW/
2
3
2 9
cm ) as the excitation source. The intensity was adjusted digitally from
by the behavior of the MI octahedra. In the case of Cs Bi I ,
2
6
3
2 9
7
8 to 3906 mW/cm .
3
−
the BiI octahedra share faces to form [Bi I ] anions, resulting
6
2 9
Electrical Properties Measurements. Devices for electrical
in a zero-dimensional (0D) molecular salt crystal structure
space group P6 /mmc). This structure possesses hexagonal
measurements were prepared from cleaved samples using several
contact materials (Au, Pd, carbon). All contacts were in parallel plate
configuration, either parallel or perpendicular to the cleaved surface of
the layers. Gold or palladium contacts (50 nm) were deposited
(
3
symmetry, as evidenced by the hexagonal channels filled by Cs
cations (Figure 2a), but sacrifices the corner-sharing octahedra
C
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
Figure 2. Crystal structure comparison of Cs Bi I (left), Cs Sb I (middle), and Rb Sb I (right) as viewed down the c-axis (a, b, c) and along the b-
3
2 9
3
2 9
3
2 9
axis (d, e, f).
characteristic of perovskite-related structures in favor of
examine the evolution of the structure on heating (Figure 3).
The temperature was raised from 25 to 600 °C, and X-ray
3−
disparate [Bi I ] anions (Figure 2d). Despite its apparent
2
9
structural simplicity, Cs Bi I exhibits unusual ferroelastic phase
3
2 9
transitions and optical properties which are hard to reconcile
40,46−48
with its molecular nature.
The crystal structures of Cs Sb I , Rb Sb I , and Rb Bi I are
3
2 9
3
2 9
3
2 9
2D layered derivatives of the perovskite structure possessing
the prototypical corner-sharing MI octahedra. Melt-grown
6
Cs Sb I forms a 2D defect perovskite structure (space group
3
2 9
P3
̅
m1) consisting of infinite, hollow layers of corner-sharing
3
−
3−
[
SbI ] octahedra. The [SbI6] octahedra are rhombohedrally
6
elongated along the crystallographic c-axis (perpendicular to the
layer plane), resulting in three short and three long Sb−I bonds
which reduce the structure to trigonal symmetry. The layers
form hexagonal channels (Figure 2b), bearing a strong
49
resemblance to the parent SbI (BiI -type) crystal structure.
3
3
Figure 3. Synchrotron PXRD of a polycrystalline Cs Bi I sample on
heating from 27 to 600 °C.
3
2 9
Contrary to the structure of SbI , the structural voids in
3
Cs Sb I are occupied by Cs cations. The cations can occupy
3
2 9
two unique positions in the crystal lattice: either in the center
of the hexagonal voids or at the edge of the layers bilateral to
the Sb atom plane. This configuration relates to the 3D
perovskite by cleavage through the (111) crystallographic plane
of the cubic perovskite aristotype. In contrast, Rb Bi I and
diffraction patterns were taken every 6−10 °C. Cs Bi I Bragg
3
2 9
peaks remained from 25 °C until decomposition at 502 °C,
whereupon the BiI (mp 408 °C) begins to dissociate and after
3
3
2 9
which the only remaining diffraction peaks are from the higher-
melting CsI (mp = 620 °C).
Rb Sb I adopt a reduced-symmetry monoclinic (space group
3
2 9
P2 /n) version of the Cs Sb I structure. The crystal structure
Rietveld refinement was conducted using the program
1
3
2 9
50
consists of distorted monoclinic layers in which the MI6
octahedra adopt an irregular coordination approximating a
rhombohedral elongation. Channels of Rb atoms still form in
JANA2006 to solve the structure at each temperature,
allowing the extraction of the thermal expansion parameters
(Figure S3, Supporting Information). The a-axis, c-axis, and
unit cell volume expansion parameters obtained for Cs Bi I are
the voids between MI octahedra, but they lie along irregular
6
3
2 9
−5
−1
−5 −1
directions with respect to the layers (Figure 2c). Similarly, the
α = 4.79 × 10 K , α = 4.83 × 10 K , and α = 1.51 ×
a c v
4
−
−1
regular I−Sb−I bond angles of Cs Sb I (Figure 2e) are
10 K , respectively. However, each parameter shows a
discontinuity near 410 °C, indicating a phase change. This is
likely the dissociation of Cs Bi I into CsI and BiI , as the
3
2 9
disrupted, resulting in offset iodine atoms when viewed along
the b-axis (Figure 2f). For a thorough analysis of the symmetry
relation of these structures to the perovskite aristotype we refer
3
2 9
3
melting and boiling points of BiI are 409 and 542 °C,
3
38
the reader to Chang et al.
In Situ Synchrotron PXRD Results for Cs Bi I . In situ
respectively, while those of the ionic CsI are much higher at
51
632 and 1280 °C, respectively. Differential thermal analysis
3
2 9
powder XRD was conducted on Cs Bi I in a synchrotron
(Figure S4, Supporting Information) provides further evidence
3
2 9
beamline (Argonne National Lab 11-BM, λ = 0.413614 Å) to
for BiI dissociation and CsI formation. There is a transition on
3
D
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
Figure 4. Raman spectra for the A M I compounds at 300 K using a 785 nm excitation laser.
3
2 9
heating which occurs near the boiling point of BiI at 542 °C,
Table 1. Raman Peaks and Phonon Energies of Cs Bi I
3
3
2 9
well before the final melt at 610 °C that nearly coincides with
the melting point of CsI (632 °C). Cs Bi I contains the same
Cs Bi I
3
2 9
3
2 9
freq
−1
energy
(meV)
BiI octahedra as BiI , so it is not surprising that BiI dissociates
6
3
3
vibration symmetry
description
(cm )
from the Cs Bi I structure at intermediate temperatures while
3
2 9
v1
A′1
E″
E′
terminal Bi−I symmetric
146.6
127.2
119.8
104.5
90.0
18.18
15.77
14.85
12.96
11.16
cesium and iodine atoms bond and cause an incongruent melt.
Similar results are expected for the other three A M I
stretch
3
2 9
v15
v10
terminal Bi−I asymmetric
compounds, as the low boiling points of SbI (bp 400 °C)
stretch
3
and BiI cause them to dissociate before the alkali salts RbI (mp
terminal Bi−I asymmetric
3
51
stretch
6
56 °C) and CsI reach their melting points. This increases
A1
bridge Bi−I symmetric
v2
1
the difficulty of growing single crystals and precludes the use of
horizontal zone refining and traveling heater methods due to
the tendency for phase separation. However, crystallization in
the vertical Bridgman configuration is possible because the
stretch
v7
E″
bridge Bi−I asymmetric
stretch
Bi−I bending mode
I ] stretching or other
57.5
48.3
7.13
5.99
crystallization point is separated by the melt from the BiI (or
3
not assigned: Cs-[Bi
bending mode
2 9
SbI ) vapor at the top of the tube, so the stoichiometry is
3
maintained at the solid−liquid interface during growth.
3
.2. Optical and Electronic Properties. Raman Spec-
per unit. Thus, the dominant vibrations in this crystal arise from
troscopy Results. Raman spectra for the A M I compounds at
room temperature using 785 nm laser excitation reveal
structural similarities between the three layered compounds
Cs Sb I , Rb Sb I , and Rb Bi I ), with significant differences
3 2 9 3 2 9 3 2 9
for Cs Bi I (Figure 4). Each compound was examined through
000 cm , and no peaks were observed above 200 cm . The
layered compounds have two major peaks near 150 cm and
several overlapping peaks below 85 cm . Cs Bi I shows the
same two major peaks above 120 cm , along with several
strong modes between 120 and 80 cm . This difference is
likely due to the molecular salt structure of Cs Bi I , as
additional Bi−I modes are allowed by the [Bi I ] anions
compared to the more uniform MI octahedra of the layered
compounds. The major peaks of Cs Bi I are summarized in
Table 1, and those of the layered A M I compounds are listed
3−
3
2 9
the strongly bound [Bi I ] unit and weaker modes from the
2
9
+
ionic interactions of that unit with the bridging Cs cations.
3−
In the [Bi I ] unit, the major vibrational modes are the Bi−
2
9
(
I stretching modes associated with the six terminal I atoms and
the three bridging I atoms which connect the two Bi atoms
together. All notation used to denote Raman modes is
3
2 9
−
1
−1
2
−1
53
consistent with that of Nakamoto. There are three Raman-
active stretching modes associated with each set of bonds (the
−
1
3
2 9
−
1
symmetric stretch A′ and two antisymmetric stretches E″, E′),
1
−
1
so there are six distinct stretching modes associated with the
3−
54
3
2 9
[Bi I ] units. The force constants for terminal Bi−I bonds
2
9
3
−
2
9
are stronger than the bridging Bi−I bonds because the bridging
I atoms share their bonds with two Bi atoms. Thus, the higher-
energy peaks are attributed to the stretching modes of the
terminal bonds. Of these modes, the symmetrical stretching
modes have higher degeneracy and thus relatively higher
intensity than the antisymmetric modes. Thus, the major peak
6
3
2 9
3
2 9
in Table 2. The Raman spectra of the Bi-containing compounds
are in good agreement with previous reports by Vakulenko et
5
2
−1
−1
al. The major peaks for Rb Bi I are near 145 and 130 cm ,
at 146.6 cm is assigned to the symmetric terminal Bi−I
3
2 9
and the spectrum of Cs Bi I shows a series of 9 peaks, with the
strongest peaks near 150, 105, and 60 cm .
A detailed analysis of the vibrational modes of the Cs Bi I
space group (P6 /mmc) indicates that they derive mainly from
the [Bi I ] anion. The crystal structure consists of
molecular [Bi I ] anions bound together by three Cs cations
stretch A′ (v ), and its antisymmetric partners are E″(v ) at
3
2 9
1
1
15
−1
−1
−1
127.2 cm and E′(v ) at 119.8 cm . The major peak at 104.5
10
−1
cm is the bridge Bi−I symmetric stretch A′ (v ), and its
3
2 9
1
2
−
1
antisymmetric partner E″(v ) is the 90.0 cm peak. These
3
7
3
−
52
frequencies are in good agreement with those previously
2
9
3−
+
55
53,56
reported, as well as those of related compounds.
2
9
E
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
Table 2. Raman Peaks and Phonon Energies of 2D Layered A M I Compounds
3
2 9
symmetry and description
Cs Sb I
Rb Sb I
Rb Bi I
3 2 9
3
2 9
3
2 9
vibration
*M = Sb, Bi
freq (cm−1)
energy (meV)
freq (cm−1
)
energy (meV)
freq (cm−1
)
energy (meV)
v1
v2
A1g: M−I sym. stretch
Eg: M−I asym. stretch
164.8
146.8
20.43
18.20
163.3
147.5
137.0
20.25
18.29
16.99
16.25
10.19
7.36
140.3
126.7
118.7
17.39
15.71
14.72
symmetry-breaking M−I Stretch
1
31.1
82.2
59.4
50.9
v5 F2g: M−I bending mode
83.9
67.6
10.40
8.38
6.82
73.4
64.0
56.7
9.10
7.93
7.03
6.20
not assigned: M−I bending or lattice
deformation mode
5
5.0
6.31
5
0.0
Interestingly, the E′(v ) peak appears absent for the bridging
interactions. This low intensity stands in stark contrast to
Cs Bi I , which possesses modes of high intensity in this region.
1
1
atoms, but the E″(v ) peak is wider than that of the terminal E′
7
3
2 9
and E″, possibly indicating that both peaks are present but
overlap strongly.
The low intensity of the bending modes in Cs Sb I is due to
3 2 9
the constraints imposed by the covalent connections between
the octahedra. The directional bonds composing the lattice
forbid many of the bending and rotational modes associated
The lower energy peaks are associated with various bending
3
−
modes and ionic interactions between the [Bi I ] units and
2
9
+
the surrounding Cs atoms and cannot be assigned without
additional polarization experiments and support from calcu-
lations. However, the peak centered at 57.5 cm is unusually
with independent MI octahedra, and accordingly the low-
6
energy modes are much weaker than the high-energy stretching
modes.
−1
strong relative to the higher energy modes. This is a result of
In the case of Rb Sb I and Rb Bi I , the dominant modes
3
2 9
3
2 9
3−
the molecular salt structure of Cs Bi I in which the [Bi I ]
are similar to those of Cs Sb I because the same 2D layers of
3
2 9
2 9
3 2 9
+
anions are not bound directionally by the bridging Cs atoms,
giving each unit the ability to twist and rotate. This freedom
permits the bending modes to occur as often as the stretching
modes and is a key difference between 0D compounds and
compounds with a covalently bonded lattice which constrain
such modes. This peak is likely due to the bending modes of
the Bi−I bonds, as the rotational symmetry and Bi−I coupling
are sufficiently strong that rotational bending of iodine atoms
around the major Bi−Bi axis is of the same order of magnitude
as the symmetrical stretching.
corner-connected MI6 octahedra are the motifs of the
monoclinic structure. Accordingly, the Raman spectra revealed
−1
two major high-energy peaks at 140.3 and 126.7 cm for
−1
Rb Bi I and at 163.3 and 147.5 cm for Rb Sb I . These peaks
3
2 9
3
2 9
again correspond to the symmetric A (v ) M−I stretching and
1
1
1
E (v ) antisymmetric Sb−I stretching modes, respectively. In
2
addition, the F (v ) bending mode can be observed at 73.4
2
g
5
−1
−1
cm in Rb Bi I and at 82.2 cm in Rb Sb I . However, it was
3
2 9
3
2 9
expected that additional modes would be allowed by the broken
symmetries of the monoclinic structure relative to the 2D
Cs Sb I structure. Accordingly, the spectra contain several
The Raman spectra of the layered compounds are quite
3
2 9
−1
different, with two major peaks above 100 cm and several
overlapping peaks of lower energy and much weaker intensity.
The spectrum of Cs Sb I was analyzed using previous studies
extra peaks and shoulders not seen in the spectra of Cs Sb I ,
3
2 9
−1
namely, at 118.7 cm− in Rb Bi I and at 137.0 and 131.1 cm
1
3
2 9
in Rb Sb I . There are also several strongly overlapping low-
3
2 9
3
2 9
of the isostructural CH NH Bi Br and Cs Bi Br as a
energy peaks in the spectra which are largely caused by the
various bending modes. These modes are again of low intensity
due to the constraints of the covalent connections between the
3
3
2
9
3
2
9
5
7−59
guide.
P3m1 is D , and the relations between the subgroups of site
symmetry and the positions of atoms are well-known.
The point group symmetry of the space group
3
3
̅
d
57,60
The
MI octahedra comprising the lattice, similar to the case of
6
structure contains six iodine anions as a part of the SbI6
octahedra and three corner-shared I atoms that act as bridges
between neighboring octahedra. There are 39 optically active
modes, nine of which are Raman-active: the symmetric
stretching (4A ) and asymmetric stretching of bonds (E ).
Cs Sb I .
3
2 9
These peak assignments are summarized with labeled Raman
spectra in Figure 5. Beyond revealing the structural details,
intense Raman scattering at room temperature is an indicator of
strong coupling between the excited electrons and Raman
active vibrations of the lattice. Raman spectroscopy detects
dipole−dipole polarizability, so strong scattering interactions
reflect the lattice’s proclivity to local fluctuations. This
propensity to form local charges then plays a role in the
charge transport, as the free charge carriers will cause the lattice
to deform. These displacements travel with the carriers,
forming polarons that increase the effective mass of charge
1g
g
Considering only the SbI octahedra, the intramolecular Raman
6
spectrum consists of five bands (2A1g + 3E ). Two of these
g
vibrations include Sb atoms, the A (v ) symmetric Sb−I
1
g
1
stretch, and the antisymmetric E (v ) Sb−I planar contraction
g
2
58
with stretching in the perpendicular plane. These can be
correlated with the two major peaks, as the highest-energy
modes arise from the Sb−I interactions. Of these, the higher
intensity mode in similar systems has been assigned to the
61
carriers. Note that this behavior is distinct from that recently
5
4,58
A (v ) mode due to its higher symmetry.
Therefore, the
reported for CsPbBr , in which local polar distortions cause a
1g
1
3
−1
1
64.8 cm band in Cs Sb I is tentatively attributed to the
strong central peak in the Raman spectra at room temper-
3
2 9
62
A (v ) symmetric Sb−I stretch and the 146.8 band to the
ature. No such peak is seen here, indicating that these
compounds are not prone to such fluctuations, possibly due to
their lower symmetry. To further probe these interactions and
get a complete picture of the optoelectronic properties of the
A M I compounds, the photoluminescence of each compound
1g
1
antisymmetric E (v ) Sb−I planar contraction. The peak at 83.9
g
2
−1
cm likely belongs to the F (v ) Sb−I bending mode, which is
2
g
5
53
−1
the other major Raman active mode. Below 80 cm lie
several peaks of relatively low intensity which strongly overlap,
and these are due to various bending modes and Sb−I−Sb
3
2 9
was studied. As some of these measurements were conducted at
F
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
exciton would dictate that the transition lies below the band
edge, implying that the band edge is actually above 2.6 eV.
Furthermore, the low electronic dimensionality of the 0D
65
Cs Bi I supports a higher band gap than the 2D compounds.
3
2 9
In addition to the n = 1 peak, Machulin et al. observe a second,
64
weaker peak near 2.3 eV in the reflectance spectrum at 4.2 K.
This peak may be a second exciton with an even higher binding
energy, or it may be an exciton related to the degenerate
conduction bands calculated by DFT with spin−orbit
2
4,29
correction.
The DFT results show several conduction
24,29
bands with ∼0.3 eV separation,
to the absorption.
all of which may contribute
The apparent discrepancy between band gap and crystal
color is due to the high exciton binding energy (279 meV at 5
64
K) measured by Machulin et al. This binding energy is similar
to exciton binding energies of the ionic alkali halide
3
Figure 5. Waterfall plot with labeled Raman peaks for A M2I9
compounds.
66
compounds. Such high binding energies lead to large Stokes
67
shifts between the absorption and emission spectra, which is
why no photoluminescence is detected near 2.6 eV (Figure 7).
low temperatures, it should be noted that the shift in Raman
mode energy from room temperature to low temperatures is
small enough to allow comparison between the two. In
particular, the difference is less than 10 cm (1.2 meV) for
Cs Bi Br and Rb Sb Br , which are isostructural to the
−1
57
3
2
9
3
2
9
layered compounds Cs Sb I , Rb Bi I , and Rb Sb I , and is
3
2 9
3
2 9
3
2 9
within 5 cm− (0.6 meV) for Cs Bi I .
1
55
3
2 9
UV−Vis Reflectance Measurements. The optical bandgaps
of A M I were determined with polycrystalline samples at
3
2 9
room temperature using UV−vis optical diffuse reflectance
spectroscopy (Figure 6). The measured band gaps are 1.89,
Figure 7. Confocal PL of A M I at 293 K using 473 nm laser
3
2 9
excitation.
This exciton binding energy is high enough that the excitonic
absorption peaks persist at room temperature where normally
they would be absent. These excitonic peaks are responsible for
the below-gap absorption, resulting in a significant difference
between the optical absorption edge and the electronic band
gap. These excitonic bands act as chromophores, resulting in
the red hue of the crystal that appears incongruent with the
high band gap. A previous report on the isostructural
compound (CH NH ) Bi I found an analogous absorption
Figure 6. Optical absorption spectra of A M I powder samples at 293
K (derived from diffuse reflectance measurements).
3
2 9
3
3 3 2 9
1
.93, 2.03, and 2.06 eV for Cs Sb I , Rb Bi I , Rb Sb I , and
feature at 2.51 eV due to an exciton with a binding energy of
3
2 9
3
2 9
3
2 9
68
Cs Bi I , respectively. These results reflect the structural
over 300 meV, along with a band edge near 2.8 eV. In this
case, the excitonic absorption also causes the crystals to appear
red.
3
2 9
differences of these compounds, such that deviations from
the ideal 2D layered structure of Cs Sb I to the monoclinic
3
2 9
Rb M I and further to the molecular salt structure of Cs Bi I
The absorption spectrum of Cs Sb I has a small peak near
3
2 9
3
2 9
3
2 9
result in an increased bandgap. The slope of the absorption
edges is similar for each compound, with no obvious trend as a
function of composition.
2.0 eV, which we attribute to an excitonic peak. Such features
69
are not uncommon in the 2D perovskites, as Cs Bi Br
9
3
2
(isostructural to Cs Sb I ) and the 2D hybrid Ruddlesden−
3
2 9
7
0
These values are in good agreement with previous reports
with the exception of Cs Bi I , which is reported to have a band
Popper perovskites both exhibit a prominent excitonic
absorption peak. In this case, the fundamental (electronic)
band edge is the higher energy absorption edge and the lower
3
2 9
25
63
gap value ranging from 1.80 eV or 1.87 eV to 2.857 eV at
64
70
room temperature. Looking to the color of the crystals of the
A M I compounds, Cs Bi I is the lightest red and Cs Sb I
2 9
energy edge is actually the tail of the excitonic absorption.
This leads to an underestimated band gap if the excitonic edge
is used, and it is especially difficult to determine the band gap in
the case of lower exciton binding energies. Analogous
absorption features have been observed in the 3D perovskites
(CH NH ) PbX (X = I, Br), but the fundamental and
3
2 9
3
2 9
3
the darkest red, consistent with the highest and lowest band
gaps, respectively. This is difficult to reconcile with the value of
2
.857 eV for Cs Bi I , which implies that the compound should
3 2 9
be white instead of bright red. However, there is also a
prominent peak in the absorption spectrum of Cs Bi I near 2.6
3
3 3
3
excitonic emissions overlap completely due to the relatively
3
2 9
71
eV corresponding to the proposed n = 1 excitonic transition
previously seen at 2.58 eV by Machulin et al. in reflection
low exciton binding energy. Similarly, an exciton with low
binding energy is likely responsible for the peak in the
absorption of Cs Sb I near 2.0 eV, in which case the true band
64
measurements on a single crystal. The classical definition of
3
2 9
G
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
Figure 8. (a) Broad Cs Sb I PL emission at 13 K, 5 mW (blue trace) modeled by a modified Gaussian peak (red). (b) Configuration coordinate
3
2 9
model where phonons are emitted as the center relaxes from B to C (absorption) or from D to A (luminescence). Here n is the number of phonons
involved in the transition, Eph is the phonon energy, and EZPL is the zero-phonon line energy. Nonradiative recombination can occur along the path B
to C to X to A via phonon emission and absorption.
gap is above the 1.89 eV edge. This agrees with the 1.96 eV
emission peak observed in room-temperature photolumines-
cence measurements (Figure 7), which should not lie above the
band gap. A similar feature may be present for Rb Bi I , but the
carriers to the lattice site (known as self-trapping) and broaden
73
the optical spectra substantially. Such broad bands are
7
4,75
commonly seen in the alkali halides
and related compounds
76
such as AgBr and the organic−inorganic halide perov-
3
2 9
77,78
excitonic nature of this peak cannot be confirmed because our
skites.
A recent report on organic−inorganic perovskites
Rb Bi I sample exhibits no photoluminescence at room
has also shown that the corrugated bilayer structure is more
3
2 9
temperature (Figure 7). No peak or shoulder was observed
for Rb Sb I . To better understand these features and check for
prone to structural deformation than the flat ⟨100⟩ perovskite
7
982
layer, leading to broad PL.
To determine the origin of these
3
2 9
radiative defect recombination, we performed photolumines-
emission bands, low-temperature PL was measured for each
cence measurements on these compounds.
compound.
Photoluminescence of A M I Compounds. We measured
Analogous to the PL measurements at 293 K, Rb Bi I
2 9
3
2 9
3
the photoluminescence (PL) properties of A M I single
showed no PL response in the range of 410 to 900 nm at 15
K. The lack of emission is presumably due to the presence of
nonradiative recombination pathways such as native defects and
grain boundaries, which can be favored by the indirect bandgap
3
2 9
crystals both at room temperature and at low temperatures
13−15 K). PL measurements at room temperature were
(
collected in the confocal configuration under 473 nm (2.62 eV)
excitation laser using 0.25 mW laser power (50 μm spot size). A
broad, weak emission in the 1.58−2.2 eV range was observed
for Cs Bi I , Cs Sb I , and Rb Sb I single crystals, while
24
of Rb Bi I . An indirect bandgap requires phonons to mediate
3
2 9
radiative transitions near the band edge and reduces the
probability of radiative recombination, allowing nonradiative
recombination to dominate.
3
2 9
3
2 9
3
2 9
Rb Bi I showed no response (Figure 7). Each compound
3
2 9
exhibited signs of laser damage due to the small spot size,
necessitating low laser powers and short exposure times for a
reliable measurement. Even under these conditions, the
background signal increases slightly as the measurement
continues, e.g., the Rb Bi I black trace in Figure 7, which
Low-temperature PL measurements for Cs Sb I , Rb Sb I ,
3
2 9
3
2 9
and Cs Bi I revealed one or more broad peaks, but no sharp
3
2 9
peaks characteristic of free exciton zero-phonon lines were
observed. This is likely due to the temperature (∼15 K) and
broadness of the peaks, as the reported intensity of the zero-
phonon line in Cs Bi Br (isostructural to Cs Sb I ) even at 4.2
3
2 9
has no PL response but shows a continuous rise in background
3
2
9
3
2 9
as the spectrum is taken from low to high energy.
K was quite weak in all samples and was not observed for some
80
The room-temperature PL of each compound (besides
Rb Bi I , which showed no emission) in Figure 7 consists of
samples. Timmermans et al. estimated an activation energy
less than 1.24 meV for quenching of the zero phonon emission.
This activation energy is too low in value to attribute to exciton
dissociation, as an exciton with lower binding energy than 1.2
meV would not be observed due to thermal energy. This
indicates that exciton trapping by defects or impurity centers is
responsible for the lack of free-exciton emission. Similarly, the
absence of free-exciton peaks in the A M I compounds is likely
3
2 9
very broad peaks with full widths at half-maximum (fwhm’s)
ranging from 315 to 403 meV. The PL spectrum of Cs Sb I
shows a broad, skewed emission band with a peak at 1.96 eV,
the same as the skewed band-edge emission observed
3
2 9
2
3
previously in Cs Sb I thin films at room temperature. The
3
2 9
PL spectra of Rb Sb I and Cs Bi I each show an asymmetric,
3
2 9
3
2 9
3
2 9
broad emission band with a maximum at 1.92 and 1.93 eV,
respectively.
due to carrier trapping which precludes direct recombination of
free carriers.
The electron−phonon coupling indicated by the intense
Raman scattering at room temperature affects the observed
luminescence since these interactions broaden the emission
bands via scattering. The emission and absorption of phonons
can alter the energy of transitions and broaden otherwise sharp
transitions into wide bands, and this is heavily dependent on
the strength of the electron−phonon coupling. The electron−
phonon interactions cause local distortions of the lattice, known
as polarons, to couple to carriers and increase their effective
mass. In the case of strong coupling, the polarons bind the
We begin this section by introducing a configurational
coordinate model of phonon-assisted self-trapped exciton
(STE) recombination, which depends strongly on the
electron−phonon coupling. The low-temperature PL results
indicate that this coupling is quite strong and demonstrate that
STE recombination is responsible for the broad PL bands
observed for the layered materials Cs Sb I and Rb Sb I . The
72
3
2 9
3
2 9
PL band of the molecular compound Cs Bi I is much more
3
2 9
complex with multiple overlapping peaks which render fitting
unreliable. However, the emission shares some similarities with
H
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
that of Cs Sb I and Rb Sb I , suggesting that recombination of
STEs with different binding sites may be the source of these
broad peaks.
Gaussian distribution, which is the convolution of an
exponential with a Gaussian:
3
2 9
3
2 9
85
2
⎡
⎤
⎥
⎛
⎜
⎞
z
2
I(E) = A exp
1
ω
⎟ −
E − Emax
1
−x /2
dx
2)
Low-Temperature Photoluminescence of Cs Sb I . The PL
3
2 9
⎢
3
2 9
μ
⎢ 2 ⎝ μ ⎠
μ
⎥
⎦
∫−∞
2π e
spectrum of Cs Sb I at 13 K shows a broad near band-edge
⎣
emission at 1.92 eV (Figure 8a), skewed toward lower energy
analogous to the room-temperature spectrum (Figure 7). While
asymmetry in the line shape at low temperatures can reflect
skewness in the impurity band responsible for the emission
(
where z is equal to
E − Emax
ω
μ
81
caused by fluctuations of the local potentials in the crystal,
z =
−
ω
(2a)
here the broadness and skewed shape of the band is attributed
to strong electron−phonon coupling which leads to charge
localization and emission of phonons that alter the energy of
the emitted photon. Such interactions can be modeled using a
configuration coordinate diagram (Figure 8b). Here the lower
band is the valence band and the upper band is the self-trapped
exciton (STE) band, and the conduction band (not pictured)
lies above the STE band. The absorption transition upon
excitation is from A to B and STE recombination emission is
from C to D. The absorption and emission edges begin at point
C, as this is lowest point in the excited state energy diagram.
This energy EZPL = E − E is known as the zero-phonon line
ω is the peak width of the Gaussian component, E and Emax are
the photon and peak energies, respectively, and A is the
amplitude of the emission. The modification/skewness factor μ
determines the direction and amplitude of the asymmetry in the
line shape. This expression has been previously used in analysis
86
of the skewed photoluminescence peak of Ge. Excellent
agreement between the experimental spectrum of Cs Sb I
3
2 9
(sample KM20B-PL) and calculated I(E) was observed (Figure
8a).
The dependence of PL emission intensity on photoexcitation
intensity was measured at 13 K to determine the recombination
kinetics (Figure 9a). An increase in the excitation intensity from
C
A
(
ZPL) of the emission band, as it is the energy absorbed or
emitted to go directly via a radiative transition from E to E .
C
A
Higher-energy absorption bands (with energies E = E
+
n
ZPL
Eph) are allowed with the subsequent emission of phonons as
the electron relaxes from points B to C. These peaks have
equidistant spacing given by the phonon energy E , the nth
ph
peak involving n phonons has energy E = E
+ nEph.
n
ZPL
Similarly, the luminescence spectrum consists of several
mirrored radiative transitions from C to D accompanied by n
phonons emitted as the center relaxes from point D to point A,
each with energy E = E − nE . This model can also explain
n
ZPL
ph
Figure 9. (a) Dependence of Cs Sb I PL emission on excitation
intensity at 13 K. (b) Plot of log(I) vs log(L) for sample KM20B-PL.
3
2 9
low-temperature thermal quenching of luminescence, as
efficient quenching can result from nonradiative recombination
where the bands cross (point X) provided phonons can make
83
the pathway B to C to X to A viable.
1
to 50 mW was accompanied by both an enhancement of
The resulting luminescence spectrum is the convolution of
emission intensity and a small 0.01 eV redshift in the peak
multiple Gaussians (one for each transition, offset by E ), the
energy. The dependence of the PL intensity I on the laser
ph
κ
intensities of which follow a Poisson distribution:
power L is given by the I ∼ L power law dependence,
87
described by the exponent κ. Excitonic (free or bound)
emission increases superlinearly with excitation intensity, where
the power law coefficient κ is in the 1 < κ < 2 range. For DAP
recombination or free-to-bound (FB) transitions the PL
intensity increases sublinearly, so the exponent is less than 1.
A plot of log(I) vs log(L) is linear with a slope κ of 0.89 ± 0.03,
indicating the recombination process is either DAP or free-to-
bound recombination (Figure 9b). However, the influence of
strong electron−phonon coupling can render this recombina-
tion model unreliable when the excitons are heavily localized by
small polarons in the lattice. Localization imposes severe
restrictions on recombination, limiting the rate at which it can
occur in the same way that free-to-bound and DAP transitions
can only occur at specific locations in the lattice. Such
interactions violate the rate equations used to derive the power
−
S
n
e S
n!
Int(n) =
(1)
where S is the Huang−Rhys parameter which describes the
72
magnitude of the electron−phonon coupling. The super-
position of these peaks is known as the phonon wing due to the
distinctive shape of the spectrum; see, for example, the green
8
4
emission of ZnO. In this case, the electron−phonon
interaction is strong enough that individual phonon replicas
are not distinguishable due to the influence of phonon
broadening of each Gaussian peak. S represents the mean
number of phonons emitted as the excited state relaxes along
the path B to C or D to A. The Poisson distribution approaches
a Gaussian distribution for sufficiently large values of S (≥100),
and below this threshold the distribution resembles a Gaussian
peak skewed to higher values of n. Here, the distribution of
luminescence energy is skewed to lower energies because high
n results in a lower energy E = E − nEph for the transition
87
law model, which does not account for localization effects in
the case of excitonic emission. Thus, excitonic emission cannot
be excluded based on the excitation intensity-dependence data
alone.
n
ZPL
accompanied by emission of n phonons.
This distribution matches the Cs Sb I PL, which is also
To determine the quenching behavior of the emission,
temperature-dependent Cs Sb I PL spectra were measured
3
2 9
3
2 9
skewed to low energy (Figure 8a). To account for this
skewness, the measured emission was analyzed using a modified
using an excitation intensity of 5 mW (Figure 10a). The peak
shifts irregularly with temperature, redshifting from 13 to 31 K
I
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
Figure 10. (a) Temperature-dependent PL spectra of Cs Sb I using 5 mW laser power. (b) Full width at half-maximum (fwhm) vs temperature
3
2 9
with solid line fit by eq 3. (c) Integrated PL intensity vs 1000/T where the solid line is the least-squares analysis of the data to eq 4.
and then blueshifting to 70 K until the emission quenches at
05 K. The temperature dependence of the full width at half-
maximum [fwhm] of the emission peak, denoted w(T), can be
I0
I(T) = 1
+ ϕ1T + ϕ2T exp(−E /k T)
1
3/2
3/2
i
B
(4)
88
modeled using the theory of Toyozawa. This theory models
the emission broadening resulting from electron−phonon
interactions using a configuration coordinate model where
the width of the peak w(T) depends on the average phonon
where I , ϕ , and ϕ are constants, k is the Boltzmann
constant, and E is the activation energy for thermal quenching.
The Arrhenius plot was analyzed according to eq 4 (Figure
0c), yielding an activation energy of 14 ± 1 meV. The values
0
1
2
B
i
1
8
9
energy:
−5
of I , ϕ , and ϕ were (1.59 ± 0.05) × 10 , 0.005 ± 0.001, and
0
1
2
0
.7 ± 0.2, respectively. The temperature-dependent data agrees
1
/2
⎡
⎢
⎣
⎛
⎞⎤
Eph
with the model well, but the lack of emission after 120 K
suggests that the calculated activation energy corresponds to a
nonradiative recombination process. If the activation energy
corresponded to ionization of a donor or acceptor level in DAP
recombination, a free-to-bound transition should be observed at
w(T) = 2.36 S E ⎢coth⎜
⎟⎥
ph
⎝2k T ⎠⎥
B
⎦
(
3)
Here S is the Huang−Rhys electron−phonon coupling
parameter and E is the effective phonon energy. This model
applies when there is sufficiently strong electron−phonon
coupling; it has been applied to the emission of KCl and
Cd1−xZn Te. A plot of fwhm (eV) vs temperature (Figure
0b) was analyzed according to eq 3, yielding S = 42.7 ± 3.2
and a phonon energy of Eph = 7.0 ± 0.8 meV. The electron−
phonon coupling is quite strong; S = 42.7 is comparable to that
observed in the alkali halides, in particular NaCl (S = 42), and
is significantly higher than the reported value for CsI (S =
ph
95
higher temperatures as the lower state ionizes, but instead the
90
emission quenches completely as the activation energy is
approached. This low activation energy indicates that the
nonradiative recombination pathway dominates at higher
temperatures, as 14 meV corresponds to the thermal energy
at 163 K, which is consistent with the quenching of the
emission at 120 K. Furthermore, the observed redshift in the
peak position with intensity is in contrast to the blueshift
8
9
x
1
72,96
9
1,92
characteristic of DAP recombination,
due to DAP recombination.
so the emission is not
1
2).
This value is unexpected given the covalency of the
Sb−I and Cs−I bonds, but it presumably arises from the
layered structure of the material, which tolerates more local
polarization than the 3D CsI structure. The layer interfaces are
The sublinear increase of the emission intensity with
increasing excitation intensity indicates that the emission of
Cs Sb I is due to either free-to-bound or DAP recombination,
so it seems that the emission must be a free-to-bound
transition. However, there is a peak in the UV−vis reflectance
spectrum (Figure 6) similar to the excitonic transition observed
not bonded as in the 3D case. Thus, the SbI octahedra have
3
2 9
6
more degrees of freedom since they are bonded to three other
octahedra in two directions, as opposed to six bonds in all three
directions of the 3D structure. This allows the 2D material to
distort substantially in the presence of charge carriers, resulting
in stronger electron−phonon coupling even with similar
bonding behavior. The measured effective phonon energy of
in the perovskite halides, including the isostructural Cs Bi Br ,
3
2
9
6
9
indicating that the emission should be excitonic. This
contradiction arises from the strong electron−phonon
interactions, which cause the excited carriers to become highly
localized so excitons are immediately trapped by lattice
deformations upon formation. The localization imposes severe
restrictions on recombination, violating the rate equations used
to derive the power law model. Thus, excitonic emission
cannot be excluded based on the excitation intensity-dependent
data alone and must be distinguished from free-to-bound
transitions using other observations. Free-to-bound transitions
are characterized by a donor or acceptor state responsible for
binding a carrier, which at higher temperatures would be
ionized and result in band-edge emission between free carriers.
This is not consistent with our observation of thermal
quenching of the emission which indicates that no such state
exists, so the transition should not be from free-to-bound
recombination. Furthermore, the excitonic transitions at room
temperature suggest a high exciton binding energy that cannot
be overcome via thermal activation, ruling out free exciton
7
.0 meV agrees quite well with the low-energy mode of 6.82
meV (55 cm ) observed in the Raman spectrum of Cs Sb I at
−1
3
2 9
room temperature. The higher energy modes are frozen out
below 50 K, so it is not surprising that the recombination
involves a lower-energy mode. The phonon broadening
diverges from the fitted line around 55 K (Figure 10b),
suggesting that a higher energy phonon is thermally activated
and dominates the recombination. However, there is
insufficient data to fit this higher energy phonon due to the
87
93
phase transition at 86 K, which shifts the fwhm and after
which the emission quickly quenches (105 K).
The thermal quenching of the emission intensity with
temperature was measured to analyze the recombination
process. The Arrhenius plot of integrated PL intensity vs
inverse temperature can be analyzed with the DAP equation,
which can describe the activation energy of the relevant level or
94
that of a competing nonradiative pathway:
J
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
Figure 11. (a) Broad PL emission of Rb Sb I fit by a modified Gaussian peak (dashed blue line) with residual plot (red trace, enlarged 100× for
3
2 9
clarity). (b) Dependence of PL emission on excitation intensity for Rb Sb I at 13 K. (c) Plot of log(I) vs log(L) for sample KM27-PL.
3
2 9
Figure 12. (a) Temperature-dependent PL spectra for Rb Sb I with 3 mW laser power. (b) fwhm vs temperature, with solid black line denoting the
3
2 9
least-squares fit to eq 3 through 105 K and blue line fit to eq 2 offset by 120 K. (c) Arrhenius plot of integrated PL intensity vs 1000/temperature fit
with an excitonic model (eq 5).
1
02
emission. Thus, the observed emission must involve bound
exciton recombination.
line broadening attributed to electron−phonon coupling.
Furthermore, self-trapping is strongly dependent on the
dimensionality of the system, with stronger self-trapping in
lower-dimensional materials. STEs have been shown to cause
broad emission covering the visible spectrum in the 2D hybrid
and α-(DMEN)-
PbBr4. As a corrugated 2D material, the structure of
Cs Sb I provides ample opportunities for trapping as the
Excitons can be bound by extrinsic defects, native defects, or
intrinsic lattice interactions. One such example of the former is
excitons bound to isoelectronic impurities, which are
responsible for emission in other semiconductors such as
77
103
halide perovskites (N-MEDA)[PbBr ]
4
7
2,76,97,98
82
GaP:N and AgBr.
In AgBr, excitons bound to iodine
6
impurity atoms produce a broad emission 10 times stronger
3
2 9
than that of the free exciton emission, even though there are
charges scatter between MI octahedra in the bilayers. Thus, the
6
7
∼
10 times more Br atoms than I atoms. However, this process
requires the presence of a substantial number of impurities, and
there is no indication that such defects exist in Cs Sb I . Our
broad band PL emission in Cs Sb I is assigned to
3
2 9
recombination involving self-trapped excitons and the emission
of multiple phonons.
3
2 9
explanation for the PL in Cs Sb I is recombination between
Low-Temperature Photoluminescence of Rb Sb I . PL
3
2 9
3
2 9
self-trapped excitons (STEs), where small polarons trap would-
spectra of Rb Sb I collected at 13 K for a single crystal reveal
3 2 9
67
be free excitons. The strong electron−phonon interactions
induce small polarons in the lattice that locally trap charge
carriers, resulting in self-trapped excitons. Self-trapping of
excitons often occurs in semiconductors with a large electron−
a broad, skewed peak at 1.96 eV, with what appears to be a
superimposed set of weak phonon replicas (Figure 11a). This
peak was successfully modeled using the exponentially modified
Gaussian distribution applied to the PL of Cs Sb I (Figure 11a,
3
2 9
72
phonon coupling.
blue dashed line). The residual plot (Figure 11a, red trace)
shows the difference between the experimental data and the
calculated modified Gaussian fitting, revealing several small
peaks with a separation of 62 ± 3 meV. These resemble
phonon replicas, and the energy spacing is a reasonable value
for a phonon. However, 62 meV lies well outside the observed
energy range (6−20 meV) for the optical phonons in the
Raman spectrum at room temperature so it is unlikely that
these peaks are phonon replicas. The intensities of these small
peaks are only 1% of the amplitude of the modified Gaussian,
so they will not affect the error significantly.
This STE recombination process matches the observed
emission behavior in every way, and self-trapped excitons have
been observed in many related materials. These PL emission
peaks are strongly dependent on electron−phonon interactions,
manifesting as narrow lines when the electron−phonon
coupling is weak (nearly free excitons, i.e., with a low S
parameter), and as broad bands when the coupling is strong
7
2
(
high S parameter). For Cs Sb I , both the phonon
3 2 9
broadening with a high Huang−Rhys parameter of S = 42.7
and intense Raman scattering at room temperature demon-
strate the strength of the electron−phonon coupling. The
configuration coordinate model can also explain the low-
temperature quenching of the PL, as efficient quenching can
come from nonradiative recombination where the bands cross
along the phonon-assisted pathway B to C to X to A (Figure
The dependence of PL emission intensity on laser power for
Rb Sb I (sample code KM27-PL) was measured at 13 K to
3
2 9
probe the nature of the broad band (Figure 11b). An increase
in the excitation intensity from 0.5 to 15 mW was accompanied
by an increase of the emission intensity but no shift in the peak
location. A plot of log(I) vs log(L) was linear with a slope κ of
1.08 ± 0.02 (Figure 11c), indicating an excitonic transition. The
excitation intensity-dependence of the emission is consistent
with both bound and free-exciton emission; however, the peak
8
3
8
b). This recombination process has been extensively studied
75,99,100
in the alkali halides,
observed in both SbI and BiI . The 3D hybrid organic−
inorganic hybrid halide perovskites also exhibit PL emission
and moreover STEs have been
101
3
3
K
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
is too broad for free exciton emission, which would be very
narrow at low temperatures. Thus, the low-T PL emission in
Rb Sb I is excitonic in nature and is due to recombination of
bound excitons and multiphonon emission. To further
characterize the emission behavior, temperature-dependent
PL spectra were taken at an excitation intensity of 3 mW
from 13 to 220 K (Figure 12a). The peak shifts irregularly with
temperature, redshifting from 13 K to 80 K, saturating until 140
K, and then redshifting again until 220 K after which it can no
longer be detected.
including radiative recombination of excitons bound to defects
such as isoelectronic impurities as described above, but they
require a substantial number of impurities to observe emission.
In particular, the high intensity of the bound exciton emission
supports the notion that the binding mechanism is intrinsic to
the structure, as the free exciton peak is entirely absent. Both
the intense Raman scattering at room temperature and the large
Huang−Rhys factor calculated from the fwhm broadening of
the PL emission by phonons at low temperature demonstrate
strong exciton−phonon interactions in Rb Sb I . Furthermore,
3
2 9
3
2 9
The temperature dependence of the fwhm can again be
modeled using the configuration coordinate model shown in
Figure 8b. Analyzing the data in Figure 12b with eq 3 yielded a
the spectral shape and photoluminescence peak behavior are
similar to those of Cs Sb I , supporting the assignment of
3
2 9
similar mechanisms to each emission. Thus, we assign the
broad band PL in Rb Sb I to recombination of self-trapped
coupling parameter S = 50.4 ± 1.7 and an effective phonon
1
3
2 9
energy of Eph1 = 7.1 ± 0.1 meV. This value corresponds to the
excitons.
Low-Temperature Photoluminescence of Cs Bi I . In
−1
low-energy phonon observed at 7.4 meV (59.4 cm ) in the
Raman spectrum of Rb Sb I at room temperature, as the
3
2 9
contrast to the two previous compounds, the low-temperature
(13 K) PL spectrum of Cs Bi shows several overlapping
3
2 9
higher-energy phonon modes are frozen out at low temper-
ature. This behavior, however, holds only to 105 K (black line),
after which another phonon dominates. Shifting the zero
temperature of eq 3 by 120 K to reflect the thermally activated
phonon, this phonon energy can be calculated (blue line, Figure
3
I
2 9
peaks in the 1.85−2.03 eV range, just below the room-
temperature absorption edge at 2.06 eV (Figure 13). The
1
2b), yielding a coupling parameter S = 21.2 ± 0.9 meV and
2
an effective phonon energy of Eph2 = 19.6 ± 0.1 meV for the
higher energy phonon. This value likely corresponds to the
dominant symmetrical stretching mode A (v ) at 20.25 meV
1
1
−1
(
163.3 cm ), accounting for the thermal shift of Raman peak
energy from high to low temperatures. The significant change
of the Huang−Rhys parameter S with temperature reflects the
change in the effective phonon energy. The parameter S
represents the average number of phonons of a single energy
emitted as the charge center relaxes from points D to A (Figure
Figure 13. PL spectra of Cs Bi I measured as a function of (a)
intensity, and (b) temperature.
3
2 9
8
b), so higher energy phonons reduce the number of phonons
required to mediate the same transition. In any case the
coupling is quite strong, S = 50.4 in particular matches that of
spectrum is in excellent agreement with the broad emission
observed from 1.63 to 2.23 eV at 293 K using the confocal
setup (Figure 7). This emission was measured as a function of
both excitation intensity and temperature (Figure 13a,b,
respectively). The emission shows strong thermal quenching,
with the signal vanishing above 105 K.
1
91,92
KBr (S = 51) and is higher than the value for RbI (S = 40).
This increase, as discussed previously, presumably results from
the layered structure, which tolerates more distortion than the
rigid 3D RbI lattice. At higher temperatures the coupling
91
weakens to S = 21.2, closer to the value of RbBr (S = 25).
2
The PL spectrum of Cs Bi I is comprised of at least four
3
2 9
To model the quenching of the PL intensity with increasing
temperature, we use the equation of Bimberg et al. for the
overlapping peaks. This strong overlap renders peak fitting
unreliable, as the peaks do not shift significantly with
temperature or excitation intensity and thus cannot be resolved.
Multiple peak analyses were attempted, but without the precise
number of peaks such analysis has limited utility and requires
the aid of calculations to assign the different peak energies to
specific defects or binding sites. A red emission peak in the
related compound Cs Bi Br was attributed to the presence of
104
thermal quenching of exciton emission given by
I0
I(T) = 1
+ C exp(−E /kT) + C exp(−E /k T)
(5)
1
1
2
2
B
where I , C , and C are constants and E and E are activation
0
1
2
1
2
3
2
9
80
energies. The Arrhenius plot of integrated PL intensity vs
inverse temperature was analyzed with eq 5 (Figure 12c),
yielding activation energies of 130 ± 50 meV and 30 ± 3 meV.
oxide impurities from the Bi O precursor, which may be one
2 3
of the peaks seen here. Nevertheless, fitting the emission with a
single-peak fit yielded a rough analysis of the overall emission
behavior and is included in the Supporting Information. Briefly,
this analysis indicates sublinear dependence of PL intensity on
excitation intensity, suggesting either a donor−acceptor pair
(DAP) recombination or free-to-bound transition. This is in
contrast with the observations of excitonic absorption bands via
UV−vis reflectance measurements (Figure 6), which indicate
that the emission is excitonic. The change in emission peak
width (Figure S5b) was attributed to phonon broadening with a
phonon of energy Eph = 5.9 ± 0.4 meV. This energy may
correspond to the strong bending mode phonon seen in the
−
3
The values of I , C , and C were (1.10 ± 0.01) × 10 , (1 ± 4)
0
1
2
6
×
10 , and 90 ± 40, respectively. The errors on C and E are
1 1
high because the high-temperature luminescence quenches
rapidly and the points cluster together. The observation of two
activation energies presumably indicates that there are two
nonradiative recombination mechanisms that dominate at
different temperatures, together quenching the emission around
2
20 K.
Similarly to the preceding case, the PL emission mechanism
in Rb Sb I is likely to involve recombination of STEs, where
3
2 9
−1
small polarons in the lattice bind free excitons. Other
possibilities exist which could account for this broad emission,
Raman spectrum at 7.1 meV (57.5 cm ), since the error in the
PL measurement is likely underestimated (see Supporting
L
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
Table 3. Photoluminescence Parameters for Cs Bi I , Cs Sb I , and Rb Sb I
2 9
3
2 9
3
2 9
3
absorption
edge
peak position
(293 K)
calc. fwhm (at
0 K)
calc. Ephonon
observed (Raman) Eph
Huang−Rhys
factor S
compound
Cs Bi I
peak position (13−15 K)
(meV)
(meV)
2.06 eV
1.89 eV
2.03 eV
1.93 eV
1.96 eV
1.92 eV
1.95 eV
1.92 eV
1.96 eV
125 meV
107 meV
119 meV
5.9
7.1
79.5
3
2 9
Cs Sb I
7.0
6.82
42.7
3
2 9
Rb Sb I
7.1, 19.6
7.4, 20.3
50.4, 21.2
3
2 9
Table 4. Conductivity Results for A M I Compounds
3
2 9
10
compound
Rb Sb I
sample ID
thickness (cm)
contact material
ρdark(+) (1010 Ω·cm)
σlight/σdark (h+)
ρdark(−) (10 Ω·cm)
σlight/σdark (e−)
KM27-1
KM20B-1
KM8R
0.033
0.037
0.094
0.067
Au
C
8.29
52.1
9.39
1.03
8.70
52.6
6.15
1.02
9.29
4.00
3
2 9
Cs Sb I
3
2 9
Cs Bi I
Pd
C
972
10.50
2.95
903
3
2 9
Rb Bi I
KM20-1
31.8
32.9
3
2 9
Information). The phonon broadening is an indication of
strong electron−phonon coupling, with a calculated Huang−
Rhys parameter S = 79.5 ± 3.8 which is presumably
overestimated due to the presence of multiple peaks. As a 0D
molecular structure, Cs Bi I is expected to have even more
candidates for radiation detection. Cs Sb I showed negligible
3 2 9
response to laser illumination. The other three materials
respond with a photocurrent 2−10× higher than the dark
current, summarized in Table 4 as the photoconductivity ratio
σlight/σdark. This response was consistent, with all samples
studied having response to laser greater than 2× the dark
current. While this photoresponse is relatively weak, it is
comparable to single crystals of the 3D perovskites
3
2 9
efficient self-trapping since charges can be easily bound to the
+
3−
charged Cs ions and [Bi I ] units. The Cs Bi I PL shares
2
9
3
3 9
many similarities to the broad emission of Cs Sb I and
3
2 9
106,107
108
Rb Sb I , the main difference being the observation of several
CsPbBr3
and MAPbBr3,
which show photocurrent
3
2 9
overlapping peaks. In this case there may be multiple binding
sites in the lattice for trapping the excitons, which would give
rise to several broad peaks slightly offset by the difference in
binding energy. These observations suggest that phonon-
assisted recombination of multiple self-trapped excitons is
responsible for the emission of Cs Bi I . The PL properties of
values roughly an order of magnitude higher than those of the
dark current. The lower values measured here could be due to
lower mobility or charge carrier generation and are the subject
of current investigation. Notably, one piece of the Cs Bi I
ingot exhibited a photocurrent over 100× the dark current,
indicating that these materials have higher potential that may be
accessible through increased purity or improved syntheses. The
consistent photoresponse indicates that self-trapping does not
prohibit charge transport and that these compounds merit
further study.
3
2 9
3
2 9
the three radiative compounds are summarized in Table 3
below.
Resistivity and Photoconductivity. Charge transport was
measured to determine how this charge trapping influenced the
photoconductivity in the A M I compounds and to evaluate
3
2 9
these materials as solar cell or detector materials. The resistivity
and photoconductivity under 405 nm laser excitation were
measured at room temperature using a two-probe method. The
4
. CONCLUSION
The optical and electronic properties of Bridgman grown large
single crystals of the defect halide perovskites A M I (A = Cs,
Rb; M = Bi, Sb) were examined. In situ powder X-ray
diffraction provided evidence for incongruent melting in
Cs Bi I with BiI dissociating at 460 °C. Intense Raman
emission is observed at room temperature for each compound,
indicating the high polarizability and strong electron−phonon
coupling of these materials. Photoluminescence spectroscopy
reveals broad red emission bands for Cs Sb I , Rb Sb I , and
1
0
12
resistivities are quite high, ranging from 10 to 10 Ω·cm for
each compound, with the highest and lowest resistivities
corresponding to Cs Bi I and Cs Sb I , respectively (Table 4).
3
2 9
3
2 9
3
2 9
While these resistivities are high for solar cell materials, they fall
3
2 9
3
within the desired range for hard radiation detector
1
05
materials.
A typical photoresponse to 405 nm laser excitation for
Cs Bi I is shown in Figure 14, which compares the dark
3
2 9
3
2 9
3
2 9
Cs Bi I below their respective absorption edges. A config-
conductivity to the photoconductivity. The compounds
Cs Bi I , Rb Bi I , and Rb Sb I are all photoconductive
3
2 9
uration coordinate model describes the multiphonon line shape
and temperature-dependent emission quenching behavior of
Cs Sb I and Rb Sb I . The broadness of the PL emission is
3
2 9
3
2 9
3
2 9
(
Figure S6, Supporting Information), making them potential
3
2 9
3
2 9
attributed to strong electron−phonon coupling comparable to
the alkali halides, with calculated phonon energies matching
those measured by Raman spectroscopy. These electron−
phonon interactions induce small polarons that bind charge
carriers and result in self-trapped excitons. The broad
luminescence of Cs Sb I and Rb Sb I is tentatively attributed
3
2 9
3
2 9
to phonon-assisted recombination of these self-trapped
excitons. The Cs Bi I PL spectrum shows several broad
3
2 9
overlapping peaks, with similar overall behavior that is
consistent with recombination of self-trapped excitons.
Rb Bi I , Rb Sb I , and Cs Bi I exhibited photoconductivity
3
2 9
3
2 9
3
2 9
Figure 14. I−V curves in the dark (black trace) and under 50 mW
under laser excitation, and all four materials demonstrated high
resistivity. The large band gaps, high resistivity, and high
laser (red trace) for Cs Bi I .
3
2 9
M
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
photoresponse of these compounds are key characteristics of
(5) Kojima, A.; Teshima, K.; Shirai, Y.; Miyasaka, T. Organometal
Halide Perovskites as Visible-Light Sensitizers for Photovoltaic Cells. J.
Am. Chem. Soc. 2009, 131, 6050−6051.
105
hard radiation detector materials,
and future work will
involve assessment of these materials for room-temperature
detection.
(6) NREL Solar Cell Efficiency Chart. http://www.nrel.gov/ncpv/
images/efficiency_chart.jpg (accessed Oct. 25, 2016).
7) Wei, H. T.; Fang, Y. J.; Mulligan, P.; Chuirazzi, W.; Fang, H. H.;
(
ASSOCIATED CONTENT
■
Wang, C. C.; Ecker, B. R.; Gao, Y. L.; Loi, M. A.; Cao, L.; Huang, J. S.
Sensitive X-ray detectors made of methylammonium lead tribromide
perovskite single crystals. Nat. Photonics 2016, 10, 333−339.
*
S
Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.chemma-
ter.7b01184.
(8) Tan, Z.-K.; Moghaddam, R. S.; Lai, M. L.; Docampo, P.; Higler,
R.; Deschler, F.; Price, M.; Sadhanala, A.; Pazos, L. M.; Credgington,
D.; Hanusch, F.; Bein, T.; Snaith, H. J.; Friend, R. H. Bright light-
emitting diodes based on organometal halide perovskite. Nat.
Nanotechnol. 2014, 9, 687−692.
Crystallographic refinement tables for Rb Sb I , PXRD,
3
2 9
and SEM-EDS analysis of the A M I compounds,
3
2 9
thermal coefficients of expansion for Cs Bi I derived
(9) Wong, A. B.; Lai, M.; Eaton, S. W.; Yu, Y.; Lin, E.; Dou, L.; Fu,
A.; Yang, P. Growth and Anion Exchange Conversion of
CH NH PbX Nanorod Arrays for Light-Emitting Diodes. Nano
3
2 9
from synchrotron PXRD data, DTA analysis of Cs Bi I ,
3
2 9
analysis of low-temperature PL spectra of Cs Bi I , and
3
3
3
3
2 9
Lett. 2015, 15, 5519−5524.
I−V curves in the dark and under laser for all four
compounds (PDF)
(10) Zhu, H.; Fu, Y.; Meng, F.; Wu, X.; Gong, Z.; Ding, Q.;
Gustafsson, M. V.; Trinh, M. T.; Jin, S.; Zhu, X. Y. Lead halide
perovskite nanowire lasers with low lasing thresholds and high quality
factors. Nat. Mater. 2015, 14, 636−642.
Crystallographic Information File for Rb Sb I structure
3
2 9
solution (CIF)
(11) Chen, Y.-S.; Manser, J. S.; Kamat, P. V. All Solution-Processed
Lead Halide Perovskite-BiVO Tandem Assembly for Photolytic Solar
4
AUTHOR INFORMATION
■
*
Fuels Production. J. Am. Chem. Soc. 2015, 137, 974−981.
Corresponding Author
(B.W.W.) E-mail: b-wessels@northwestern.edu.
(12) Cao, D. H.; Stoumpos, C. C.; Farha, O. K.; Hupp, J. T.;
Kanatzidis, M. G. 2D Homologous Perovskites as Light-Absorbing
Materials for Solar Cell Applications. J. Am. Chem. Soc. 2015, 137,
ORCID
7
(
843−7850.
Kyle M. McCall: 0000-0001-8628-3811
Constantinos C. Stoumpos: 0000-0001-8396-9578
Mercouri G. Kanatzidis: 0000-0003-2037-4168
Bruce W. Wessels: 0000-0002-8957-7097
13) Fabini, D. Quantifying the Potential for Lead Pollution from
Halide Perovskite Photovoltaics. J. Phys. Chem. Lett. 2015, 6, 3546−
3548.
(
14) Niu, G.; Guo, X.; Wang, L. Review of recent progress in
chemical stability of perovskite solar cells. J. Mater. Chem. A 2015, 3,
970−8980.
15) Kulbak, M.; Cahen, D.; Hodes, G. How Important Is the
Organic Part of Lead Halide Perovskite Photovoltaic Cells? Efficient
CsPbBr Cells. J. Phys. Chem. Lett. 2015, 6, 2452−2456.
16) Eperon, G. E.; Paterno, G. M.; Sutton, R. J.; Zampetti, A.;
Notes
8
(
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
3
■
(
This work is supported by the Office of Nonproliferation and
Verification Research and Development under the National
Nuclear Security Administration of the U.S. Department of
Energy under Contract DE-NA0002522. Use of the Advanced
Photon Source at Argonne National Laboratory was supported
by the U.S. Department of Energy, Office of Science, Office of
Basic Energy Sciences, under Contract No. DE-AC02-
Haghighirad, A. A.; Cacialli, F.; Snaith, H. J. Inorganic caesium lead
iodide perovskite solar cells. J. Mater. Chem. A 2015, 3, 19688−19695.
(17) Sutton, R. J.; Eperon, G. E.; Miranda, L.; Parrott, E. S.; Kamino,
̈
B. A.; Patel, J. B.; Horantner, M. T.; Johnston, M. B.; Haghighirad, A.
A.; Moore, D. T.; Snaith, H. J. Bandgap-Tunable Cesium Lead Halide
Perovskites with High Thermal Stability for Efficient Solar Cells. Adv.
Energy Mater. 2016, 6, 1502458.
0
6CH11357. This work made use of the EPIC and SPID
(18) Stoumpos, C. C.; Malliakas, C. D.; Peters, J. A.; Liu, Z.;
Sebastian, M.; Im, J.; Chasapis, T. C.; Wibowo, A. C.; Chung, D. Y.;
Freeman, A. J.; Wessels, B. W.; Kanatzidis, M. G. Crystal Growth of
facilities of the NUANCE Center at Northwestern University,
which has received support from the Soft and Hybrid
Nanotechnology Experimental (SHyNE) Resource (NSF
NNCI-1542205); the MRSEC program (NSF DMR-
the Perovskite Semiconductor CsPbBr : A New Material for High-
3
Energy Radiation Detection. Cryst. Growth Des. 2013, 13, 2722−2727.
(19) Sebastian, M.; Peters, J. A.; Stoumpos, C. C.; Im, J.; Kostina, S.
1
121262) at the Materials Research Center; the International
S.; Liu, Z.; Kanatzidis, M. G.; Freeman, A. J.; Wessels, B. W. Excitonic
emissions and above-band-gap luminescence in the single-crystal
Institute for Nanotechnology (IIN); the Keck Foundation; and
the State of Illinois, through the IIN.
perovskite semiconductors CsPbBr3 and CsPbCl . Phys. Rev. B:
3
Condens. Matter Mater. Phys. 2015, 92, 235210.
(20) Dirin, D. N.; Cherniukh, I.; Yakunin, S.; Shynkarenko, Y.;
REFERENCES
■
Kovalenko, M. V. Solution-Grown CsPbBr Perovskite Single Crystals
for Photon Detection. Chem. Mater. 2016, 28, 8470.
(
1) Stoumpos, C. C.; Kanatzidis, M. G. The Renaissance of Halide
3
Perovskites and Their Evolution as Emerging Semiconductors. Acc.
(
21) Aleksandrov, K. S.; Beznosikov, V. V. Hierarchies of perovskite-
like crystals (Review). Phys. Solid State 1997, 39, 695−715.
22) Lee, B.; Stoumpos, C. C.; Zhou, N.; Hao, F.; Malliakas, C.; Yeh,
Chem. Res. 2015, 48, 2791−2802.
(
2) Moller, C. K. The structure of perovskite-like caesium plumbo
(
trihalides. Mater. Fys. Medd. Dan. Vid. Selsk 1959, 32, 1−27.
(
3) Frohlich, D.; Heidrich, K.; Kunzel, H.; Trendel, G.; Treusch, J.
̈
̈
C.-Y.; Marks, T. J.; Kanatzidis, M. G.; Chang, R. P. H. Air-Stable
Molecular Semiconducting Iodosalts for Solar Cell Applications:
Cesium-trihalogen-plumbates a new class of ionic semiconductors. J.
Lumin. 1979, 18−19, 385−388.
Cs
15385.
2
SnI as a Hole Conductor. J. Am. Chem. Soc. 2014, 136, 15379−
6
(
4) Stoumpos, C. C.; Kanatzidis, M. G. The Renaissance of Halide
Perovskites and Their Evolution as Emerging Semiconductors. Acc.
Chem. Res. 2015, 48 (10), 2791−2802.
(23) Saparov, B.; Hong, F.; Sun, J.-P.; Duan, H.-S.; Meng, W.;
Cameron, S.; Hill, I. G.; Yan, Y.; Mitzi, D. B. Thin-Film Preparation
N
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
and Characterization of Cs Sb I : A Lead-Free Layered Perovskite
(41) Dolomanov, O. V.; Bourhis, L. J.; Gildea, R. J.; Howard, J. A. K.;
Puschmann, H. OLEX2: a complete structure solution, refinement and
analysis program. J. Appl. Crystallogr. 2009, 42, 339−341.
3
2 9
Semiconductor. Chem. Mater. 2015, 27, 5622−5632.
(
24) Lehner, A. J.; Fabini, D. H.; Evans, H. A.; Hebert, C.-A.; Smock,
́
S. R.; Hu, J.; Wang, H.; Zwanziger, J. W.; Chabinyc, M. L.; Seshadri, R.
Crystal and Electronic Structures of Complex Bismuth Iodides A Bi I
(42) Sheldrick, G. SHELXT - Integrated space-group and crystal-
structure determination. Acta Crystallogr., Sect. A: Found. Adv. 2015,
3
2 9
7
(
1, 3−8.
43) Sheldrick, G. M. Crystal structure refinement with SHELXL.
Acta Crystallogr., Sect. A: Found. Adv. 2015, 71, 3−8.
44) Kortum, G.; Braun, W.; Herzog, G. Prinzip und Meßmethodik
der diffusen Reflexionsspektroskopie. Angew. Chem. 1963, 75, 653−
61.
45) Kostina, S. S.; Peters, J. A.; Lin, W.; Chen, P.; Liu, Z.; Wang, P.
(
A = K, Rb, Cs) Related to Perovskite: Aiding the Rational Design of
Photovoltaics. Chem. Mater. 2015, 27, 7137−7148.
25) Peresh, E. Y.; Sidei, V. I.; Zubaka, O. V.; Stercho, I. P.
K (Rb ,Cs ,Tl )TeBr (I ) and Rb (Cs )Sb (Bi )Br (I ) perovskite
(
(
̈
2
2
2
2
6
6
3
3
2
2
9
9
compounds. Inorg. Mater. 2011, 47, 208−212.
26) Giustino, F.; Snaith, H. J. Toward Lead-Free Perovskite Solar
Cells. ACS Energy Lett. 2016, 1, 1233−1240.
27) Harikesh, P. C.; Mulmudi, H. K.; Ghosh, B.; Goh, T. W.; Teng,
6
(
(
L.; Kanatzidis, M. G.; Wessels, B. W. Photoluminescence fatigue and
inhomogeneous line broadening in semi-insulating Tl SeI single
(
6
4
Y. T.; Thirumal, K.; Lockrey, M.; Weber, K.; Koh, T. M.; Li, S.;
Mhaisalkar, S.; Mathews, N. Rb as an Alternative Cation for
Templating Inorganic Lead-Free Perovskites for Solution Processed
Photovoltaics. Chem. Mater. 2016, 28, 7496−7504.
crystals. Semicond. Sci. Technol. 2016, 31, 065009.
46) Aleksandrova, I. P.; Sukhovsky, A. A.; Aleksandrov, K. S. Novel
incommensurate phase in Cs Bi I . Solid State Commun. 1998, 105,
(
3
2 9
3
(
23−326.
47) Jorio, A.; Currat, R.; Myles, D.; McIntyre, G.; Aleksandrova, I.;
Kiat, J.; Saint-Gregoire, P. Ferroelastic phase transition in Cs Bi I : A
neutron diffraction study. Phys. Rev. B: Condens. Matter Mater. Phys.
000, 61, 3857−3862.
48) Motsnyi, F. V.; Smolanka, O. M.; Sugakov, V. I.; Peresh, E. Y.
(
́
28) Hoye, R. L. Z.; Brandt, R. E.; Osherov, A.; Stevanovic, V.;
Stranks, S. D.; Wilson, M. W. B.; Kim, H.; Akey, A. J.; Perkins, J. D.;
Kurchin, R. C.; Poindexter, J. R.; Wang, E. N.; Bawendi, M. G.;
́
3
2 9
́
Bulovic, V.; Buonassisi, T. Methylammonium Bismuth Iodide as a
2
(
Lead-Free, Stable Hybrid Organic−Inorganic Solar Absorber. Chem. -
Eur. J. 2016, 22, 2605−2610.
Giant thermodynamical optical effect near ferroelastic phase transition
(
29) Pazoki, M.; Johansson, M. B.; Zhu, H.; Broqvist, P.; Edvinsson,
point in Cs Bi I layered crystal. Solid State Commun. 2006, 137, 221−
3
2 9
T.; Boschloo, G.; Johansson, E. M. J. Bismuth Iodide Perovskite
Materials for Solar Cell Applications: Electronic Structure, Optical
Transitions, and Directional Charge Transport. J. Phys. Chem. C 2016,
2
24.
(
49) Trotter, J.; Zobel, T. The crystal structure of SbI and BiI . Z.
3 3
Kristallogr. - Cryst. Mater. 1966, 123, 67−72.
1
(
20, 29039−29046.
30) Hebig, J.-C.; Ku
Properties of (CH NH ) Sb I Thin Films for Photovoltaic
(50) Petrí ek, V.; Dusek, M.; Palatinus, L. Crystallographic
̌
c
̌
̌
̈
hn, I.; Flohre, J.; Kirchartz, T. Optoelectronic
Computing System JANA2006: General features. Z. Kristallogr. -
Cryst. Mater. 2014, 229, 345−352.
3
3
3
2 9
Applications. ACS Energy Lett. 2016, 1, 309−314.
31) Park, B.-W.; Philippe, B.; Zhang, X.; Rensmo, H.; Boschloo, G.;
Johansson, E. M. J. Bismuth Based Hybrid Perovskites A Bi I (A:
(51) Haynes, W. M. CRC handbook of chemistry and physics: A ready-
reference book of chemical and physical data, 96th ed.; CRC Press: 2015.
(52) Vakulenko, O. V.; Gubanov, V. O.; Kun, S. V.; Motsnyi, F. V.;
Peresh, E. Y.; Terekhov, V. A. Raman diagnostics of new types of
(
3
2 9
Methylammonium or Cesium) for Solar Cell Application. Adv. Mater.
015, 27, 6806−6813.
32) Han, H.; Hong, M.; Gokhale, S. S.; Sinnott, S. B.; Jordan, K.;
2
(
A B X layered crystals. Optical Diagnostics of Materials and Devices for
3 2 9
Opto-, Micro-, and Quantum Electronics; SPIE Proceedings 2259; SPIE:
1998; pp 351−354.
Baciak, J. E.; Nino, J. C. Defect Engineering of BiI Single Crystals:
3
(53) Nakamoto, K. Applications in Inorganic Chemistry. In Infrared
Enhanced Electrical and Radiation Performance for Room Temper-
and Raman Spectra of Inorganic and Coordination Compounds; John
Wiley & Sons, Inc.: 2008; pp 149−354.
ature Gamma-Ray Detection. J. Phys. Chem. C 2014, 118, 3244−3250.
(
́
33) Lehner, A. J.; Wang, H.; Fabini, D. H.; Liman, C. D.; Hebert, C.-
(54) Laane, J.; Jagodzinski, P. W. Low-frequency vibrational spectra
A.; Perry, E. E.; Wang, M.; Bazan, G. C.; Chabinyc, M. L.; Seshadri, R.
of bromo- and iodobismuthates and the observation of a trans effect.
Electronic structure and photovoltaic application of BiI . Appl. Phys.
3
Inorg. Chem. 1980, 19, 44−49.
Lett. 2015, 107, 131109.
(55) Motsnyi, F. V.; Vuychik, M. V.; Smolanka, O. M.; Peresh, E. Y.
(
34) Ganose, A. M.; Savory, C. N.; Scanlon, D. O. Beyond
Phase Transition in Cs Bi I Ferroelastic: Investigation by Raman
3
2 9
Methylammonium Lead Iodide: Prospects for the Emergent Field of
2
Scattering Technique. Funct. Mater. 2005, 12, 491−496.
ns Containing Solar Absorbers. Chem. Commun. 2017, 53, 20−44.
(56) Sheu, H.-L.; Laane, J. Trans Effect in Halobismuthates and
(
35) Lindqvist, O.; Johansson, G.; Sandberg, F.; Norin, T. The
Haloantimonates Revisited. Molecular Structures and Vibrations from
Crystal Structure of Caesium Bismuth Iodide, Cs Bi I . Acta Chem.
3
2 9
Theoretical Calculations. Inorg. Chem. 2013, 52, 4244−4249.
Scand. 1968, 22, 2943−2952.
(57) Yaremko, A. M.; Valakh, M. Y.; Trylis, O. V.; Baran, J.;
(
36) Yamada, K.; Sera, H.; Sawada, S.; Tada, H.; Okuda, T.; Tanaka,
Marchewka, M. K.; Ratajczak, H. Anharmonic effects and temperature
H. Reconstructive Phase Transformation and Kinetics of Cs Sb I by
3
2 9
dependence of the lattice spectra of ferroelectric crystals A B C . J.
Means of Rietveld Analysis of X-Ray Diffraction and 1 I NQR. J. Solid
27
3 2 9
Mol. Struct. 1999, 511−512, 57−68.
58) Bator, G.; Baran, J.; Jakubas, R.; Karbowiak, M. Raman studies
of structural phase transition in Cs Bi Br . Vib. Spectrosc. 1998, 16,
State Chem. 1997, 134, 319−325.
(
(
37) Sidey, V. I.; Voroshilov, Y. V.; Kun, S. V.; Peresh, E. Y. Crystal
3
2
9
growth and X-ray structure determination of Rb Bi I . J. Alloys Compd.
3
2 9
1
(
1−20.
59) Miniewicz, A.; Jakubas, R.; Ecolivet, C.; Girard, A. Raman
scattering in ferroelectric (CH NH ) Bi Br single crystals. J. Raman
2
(
000, 296, 53−58.
38) Chang, J.-H.; Doert, T.; Ruck, M. Structural Variety of Defect
Perovskite Variants M E X (M = Rb, Tl, E = Bi, Sb, X = Br, I). Z.
3
3
3
2
9
3
2
9
Spectrosc. 1994, 25, 371−375.
60) Valakh, M. Y.; Lisitsa, M. P.; Trilis, A. V.; Yaremko, A. M.;
Peresh, E. Y. Raman scattering in A B C ferroelectric crystals. Phys.
Anorg. Allg. Chem. 2016, 642, 736−748.
39) Peresh, E. Y.; Sidei, V. I.; Gaborets, N. I.; Zubaka, O. V.;
Stercho, I. P.; Barchii, I. E. Influence of the average atomic number of
the A TeC and A B C (A = K, Rb, Cs, Tl(I); B = Sb, Bi; C = Br, I)
(
(
3
2 9
Solid State 1997, 39, 1289−1291.
2
6
3
2
9
(61) Stoneham, A. M.; Gavartin, J.; Shluger, A. L.; Kimmel, A. V.;
Ramo, D. M.; Rønnow, H. M.; Aeppli, G.; Renner, C. Trapping, self-
trapping and the polaron family. J. Phys.: Condens. Matter 2007, 19,
255208.
compounds on their melting point and band gap. Inorg. Mater. 2014,
0, 101−106.
40) Ivanov, Y. N.; Sukhovskii, A. A.; Lisin, V. V.; Aleksandrova, I. P.
5
(
Phase Transitions of Cs Sb I , Cs Bi I , and Cs Bi Br Crystals. Inorg.
(62) Yaffe, O.; Guo, Y.; Tan, L. Z.; Egger, D. A.; Hull, T.; Stoumpos,
3
2 9
3
2 9
3
2
9
Mater. 2001, 37, 623−627.
C. C.; Zheng, F.; Heinz, T. F.; Kronik, L.; Kanatzidis, M. G.; Owen, J.
O
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
S.; Rappe, A. M.; Pimenta, M. A.; Brus, L. E. Local Polar Fluctuations
in Lead Halide Perovskite Crystals. Phys. Rev. Lett. 2017, 118, 136001.
(83) Ueta, M.; Kanzaki, H.; Kobayashi, K.; Toyozawa, Y.; Hanamura,
E. Theory of Excitons in Phonon Fields. In Excitonic Processes in Solids;
Springer Berlin Heidelberg: Berlin, Heidelberg, 1986; pp 203−284.
(84) Shi, S. L.; Li, G. Q.; Xu, S. J.; Zhao, Y.; Chen, G. H. Green
Luminescence Band in ZnO: Fine Structures, Electron−Phonon
Coupling, and Temperature Effect. J. Phys. Chem. B 2006, 110,
10475−10478.
(85) Grushka, E. Characterization of exponentially modified Gaussian
peaks in chromatography. Anal. Chem. 1972, 44, 1733−1738.
(86) Pezzoli, F.; Qing, L.; Giorgioni, A.; Isella, G.; Grilli, E.; Guzzi,
M.; Dery, H. Spin and energy relaxation in germanium studied by spin-
polarized direct-gap photoluminescence. Phys. Rev. B: Condens. Matter
Mater. Phys. 2013, 88, 045204.
(
63) Nila,
̆
A.; Baibarac, M.; Matea, A.; Mitran, R.; Baltog, I. Exciton−
phonon interactions in the Cs Bi I crystal structure revealed by
3
2 9
Raman spectroscopic studies. Phys. Status Solidi B 2017, 254, 1552805.
64) Machulin, V. F.; Motsnyi, F. V.; Smolanka, O. M.; Svechnikov,
(
G. S.; Peresh, E. Y. Effect of temperature variation on shift and
broadening of the exciton band in Cs Bi I layered crystals. Low Temp.
3
2 9
Phys. 2004, 30, 964−967.
(
65) Xiao, Z.; Meng, W.; Wang, J.; Mitzi, D. B.; Yan, Y. Searching for
promising new perovskite-based photovoltaic absorbers: the impor-
tance of electronic dimensionality. Mater. Horiz. 2017, 4, 206−216.
(
&
66) Kittel, C. Introduction to Solid State Physics, 8th ed.; John Wiley
Sons, Inc.: 2005.
(87) Schmidt, T.; Lischka, K.; Zulehner, W. Excitation-power
dependence of the near-band-edge photoluminescence of semi-
conductors. Phys. Rev. B: Condens. Matter Mater. Phys. 1992, 45,
(
67) Knox, R. S. Introduction to Exciton Physics. In Collective
Excitations in Solids; Di Bartolo, B., Ed.; Springer US: Boston, MA,
983; pp 183−245.
68) Kawai, T.; Ishii, A.; Kitamura, T.; Shimanuki, S.; Iwata, M.;
Ishibashi, Y. Optical Absorption in Band-Edge Region of
8
(
989−8994.
88) Toyozawa, Y. Further Contribution to the Theory of the Line-
Shape of the Exciton Absorption Band. Prog. Theor. Phys. 1962, 27,
9−104.
89) Stadler, W.; Hofmann, D. M.; Alt, H. C.; Muschik, T.; Meyer, B.
K.; Weigel, E.; Muller-Vogt, G.; Salk, M.; Rupp, E.; Benz, K. W.
Optical investigations of defects in Cd Zn Te. Phys. Rev. B: Condens.
1
(
8
(
(
1
CH NH ) Bi I Single Crystals. J. Phys. Soc. Jpn. 1996, 65, 1464−
3
3
3
2 9
468.
̈
(
69) Bass, K. K.; Estergreen, L.; Savory, C. N.; Buckeridge, J.;
Scanlon, D. O.; Djurovich, P. I.; Bradforth, S. E.; Thompson, M. E.;
Melot, B. C. Vibronic Structure in Room Temperature Photo-
luminescence of the Halide Perovskite Cs Bi Br . Inorg. Chem. 2017,
1‑x
x
Matter Mater. Phys. 1995, 51, 10619−10630.
(90) Morrison, S. R.; Rolfe, J. An anomalous absorption spectrum:
3
2
9
The missing zero-phonon line of the F -centre in KCl crystals. Solid
2
5
(
6, 42−45.
State Commun. 1976, 18, 617−619.
70) Stoumpos, C. C.; Cao, D. H.; Clark, D. J.; Young, J.; Rondinelli,
(91) Leung, C. H.; Song, K. S. On the luminescence quenching of F
J. M.; Jang, J. I.; Hupp, J. T.; Kanatzidis, M. G. Ruddlesden−Popper
Hybrid Lead Iodide Perovskite 2D Homologous Semiconductors.
Chem. Mater. 2016, 28, 2852−2867.
centres in alkali halides. Solid State Commun. 1980, 33, 907−910.
92) Dawson, R. K.; Pooley, D. F Band Absorption in Alkali Halides
as a Function of Temperature. Phys. Status Solidi B 1969, 35, 95−105.
93) Novikova, M. S.; Makarova, I. P.; Blomberg, M. K.; Bagautdinov,
(
(
71) Sestu, N.; Cadelano, M.; Sarritzu, V.; Chen, F.; Marongiu, D.;
(
Piras, R.; Mainas, M.; Quochi, F.; Saba, M.; Mura, A.; Bongiovanni, G.
Absorption F-Sum Rule for the Exciton Binding Energy in
Methylammonium Lead Halide Perovskites. J. Phys. Chem. Lett.
B. S.; Aleksandrova, I. P. Structure and phase transitions in trigonal
Cs Sb I crystals. Crystallogr. Rep. 2001, 46, 26−29.
3
2 9
(94) Krustok, J.; Collan, H.; Hjelt, K. Does the low-temperature
2
015, 6, 4566−4572.
72) Pelant, I.; Valenta, J. Luminescence Spectroscopy of Semiconductors;
Oxford University Press: New York, 2012.
73) Dean, P.; Fischer, B.; Herbert, D.; Lagois, J.; Yu, P.; Cho, K.
Excitons; Springer Science & Business Media: 2012; Vol. 14.
74) Mott, N. F.; Stoneham, A. M. The lifetime of electrons, holes
and excitons before self-trapping. J. Phys. C: Solid State Phys. 1977, 10,
391−3398.
75) Purdy, A. E.; Murray, R. B.; Song, K. S.; Stoneham, A. M.
Studies of self-trapped exciton luminescence in KCl. Phys. Rev. B 1977,
5, 2170−2176.
76) Czaja, W. Concentration-induced spectral variations of
luminescence in AgBr:I. J. Phys. C: Solid State Phys. 1983, 16, 3197−
213.
77) Song, K. S.; Williams, R. T.; et al. Materials, Extrinsic Self-
Arrhenius plot of the photoluminescence intensity in CdTe point
(
towards an erroneous activation energy? J. Appl. Phys. 1997, 81, 1442−
1
445.
(
(95) Leroux, M.; Grandjean, N.; Beaumont, B.; Nataf, G.; Semond,
F.; Massies, J.; Gibart, P. Temperature quenching of photo-
luminescence intensities in undoped and doped GaN. J. Appl. Phys.
(
1
(
999, 86, 3721−3728.
96) Pankove, J. I. Optical Processes in Semiconductors; Dover: New
York, 1971.
97) Wight, D. R. Green luminescence efficiency in gallium
phosphide. J. Phys. D: Appl. Phys. 1977, 10, 431−454.
98) Thomas, D. G.; Hopfield, J. J. Isoelectronic Traps Due to
Nitrogen in Gallium Phosphide. Phys. Rev. 1966, 150, 680−689.
99) Williams, R. T.; Song, K. S. The self-trapped exciton. J. Phys.
Chem. Solids 1990, 51, 679−716.
100) Fischbach, J. U.; Frohlich, D.; Kabler, M. N. Recombination
3
(
(
1
(
(
3
(
(
Trapping, and Low-Dimensional Systems. In Self-Trapped Excitons;
Springer Berlin Heidelberg: Berlin, Heidelberg, 1993; pp 351−373.
(
̈
luminescence lifetimes and the self-trapped excition in alkali halides. J.
Lumin. 1973, 6, 29−43.
(
(
78) Yangui, A.; Garrot, D.; Lauret, J. S.; Lusson, A.; Bouchez, G.;
Deleporte, E.; Pillet, S.; Bendeif, E. E.; Castro, M.; Triki, S.; Abid, Y.;
Boukheddaden, K. Optical Investigation of Broadband White-Light
Emission in Self-Assembled Organic−Inorganic Perovskite
101) Kurisu, H.; Komatsu, T.; Tanaka, T.; Akai, I.; Karasawa, T.;
Iida, T. Pressure effects on excitons, phonons and exciton-phonon
interactions in BiI and SbI single crystals. J. Lumin. 1994, 60, 816−
3
3
(
C H NH ) PbBr . J. Phys. Chem. C 2015, 119, 23638−23647.
6
11
3
2
4
818.
(
79) Cortecchia, D.; Neutzner, S.; Srimath Kandada, A. R.; Mosconi,
(102) Wright, A. D.; Verdi, C.; Milot, R. L.; Eperon, G. E.; Perez-
E.; Meggiolaro, D.; De Angelis, F.; Soci, C.; Petrozza, A. Broadband
Emission in Two-Dimensional Hybrid Perovskites: The Role of
Structural Deformation. J. Am. Chem. Soc. 2017, 139, 39−42.
Osorio, M. A.; Snaith, H. J.; Giustino, F.; Johnston, M. B.; Herz, L. M.
Electron-phonon coupling in hybrid lead halide perovskites. Nat.
Commun. 2016, 7, 11755.
(
80) Timmermans, C. W. M.; Blasse, G. On the Luminescence of
(103) Hu, T.; Smith, M. D.; Dohner, E. R.; Sher, M.-J.; Wu, X.;
Trinh, M. T.; Fisher, A.; Corbett, J.; Zhu, X. Y.; Karunadasa, H. I.;
Lindenberg, A. M. Mechanism for Broadband White-Light Emission
from Two-Dimensional (110) Hybrid Perovskites. J. Phys. Chem. Lett.
2016, 7, 2258−2263.
(104) Bimberg, D.; Sondergeld, M.; Grobe, E. Thermal Dissociation
of Excitons Bounds to Neutral Acceptors in High-Purity GaAs. Phys.
Rev. B 1971, 4, 3451−3455.
Cs Bi Br Single Crystals. Phys. Status Solidi B 1981, 106, 647−655.
3
2
9
(
81) Morgan, T. N. Broadening of Impurity Bands in Heavily Doped
Semiconductors. Phys. Rev. 1965, 139, A343−A348.
82) Mao, L.; Wu, Y.; Stoumpos, C. C.; Wasielewski, M. R.;
(
Kanatzidis, M. G. White-Light Emission and Structural Distortion in
New Corrugated Two-Dimensional Lead Bromide Perovskites. J. Am.
Chem. Soc., 2017, 139, 5210−5215.
P
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
(
105) Owens, A.; Peacock, A. Compound semiconductor radiation
detectors. Nucl. Instrum. Methods Phys. Res., Sect. A 2004, 531, 18−37.
106) Cha, J.-H.; Han, J. H.; Yin, W.; Park, C.; Park, Y.; Ahn, T. K.;
(
Cho, J. H.; Jung, D.-Y. Photoresponse of CsPbBr3 and Cs PbBr6
4
Perovskite Single Crystals. J. Phys. Chem. Lett. 2017, 8, 565−570.
(
107) Ding, J.; Du, S.; Zuo, Z.; Zhao, Y.; Cui, H.; Zhan, X. High
Detectivity and Rapid Response in Perovskite CsPbBr Single-Crystal
3
Photodetector. J. Phys. Chem. C 2017, 121, 4917−4923.
(
108) Fang, Y.; Dong, Q.; Shao, Y.; Yuan, Y.; Huang, J. Highly
narrowband perovskite single-crystal photodetectors enabled by
surface-charge recombination. Nat. Photonics 2015, 9, 679−686.
Q
DOI: 10.1021/acs.chemmater.7b01184
Chem. Mater. XXXX, XXX, XXX−XXX