Ic 401215 X

Article
pubs.acs.org/IC
Semiconducting Tin and Lead Iodide Perovskites with Organic

Cations: Phase Transitions, High Mobilities, and Near-Infrared
Photoluminescent Properties
Constantinos C. Stoumpos, Christos D. Malliakas, and Mercouri G. Kanatzidis*
Department of Chemistry, Northwestern University, Evanston, Illinois 60208, United States
*
S Supporting Information
ABSTRACT: A broad organic−inorganic series of hybrid

metal iodide perovskites with the general formulation AMI3,
where A is the methylammonium (CH3NH3+) or formamidi-
nium (HC(NH2)2+) cation and M is Sn (1 and 2) or Pb (3
and 4) are reported. The compounds have been prepared
through a variety of synthetic approaches, and the nature of the
resulting materials is discussed in terms of their thermal
stability and optical and electronic properties. We find that the
chemical and physical properties of these materials strongly
depend on the preparation method. Single crystal X-ray
diffraction analysis of 1−4 classifies the compounds in the perovskite structural family. Structural phase transitions were observed
and investigated by temperature-dependent single crystal X-ray diffraction in the 100−400 K range. The charge transport
properties of the materials are discussed in conjunction with diffuse reflectance studies in the mid-IR region that display
characteristic absorption features. Temperature-dependent studies show a strong dependence of the resistivity as a function of
the crystal structure. Optical absorption measurements indicate that 1−4 behave as direct-gap semiconductors with energy band
gaps distributed in the range of 1.25−1.75 eV. The compounds exhibit an intense near-IR photoluminescence (PL) emission in
the 700−1000 nm range (1.1−1.7 eV) at room temperature. We show that solid solutions between the Sn and Pb compounds
are readily accessible throughout the composition range. The optical properties such as energy band gap, emission intensity, and
wavelength can be readily controlled as we show for the isostructural series of solid solutions CH3NH3Sn1−xPbxI3 (5). The charge
transport type in these materials was characterized by Seebeck coefficient and Hall-effect measurements. The compounds behave
as p- or n-type semiconductors depending on the preparation method. The samples with the lowest carrier concentration are
prepared from solution and are n-type; p-type samples can be obtained through solid state reactions exposed in air in a
controllable manner. In the case of Sn compounds, there is a facile tendency toward oxidation which causes the materials to be
doped with Sn4+ and thus behave as p-type semiconductors displaying metal-like conductivity. The compounds appear to possess
very high estimated electron and hole mobilities that exceed 2000 cm2/(V s) and 300 cm2/(V s), respectively, as shown in the
case of CH3NH3SnI3 (1). We also compare the properties of the title hybrid materials with those of the “all-inorganic” CsSnI3
and CsPbI3 prepared using identical synthetic methods.
■ INTRODUCTION
Main group metal halometallates in general and iodometallates
role.6 The connectivity of the MI6 octahedra is also affected
varying from corner-sharing to edge-sharing to face-sharing.6
in particular, represent a large class of semiconducting materials The structural diversity of these compounds has been
which display intriguing optical and electronic properties. Of reviewed.7 Among these there are some remarkable series of
special interest are the group 14 iodometalates, since they can materials with the general formulas
adopt perovskitic structures that are amenable to compositional (RNH3)2(CH3NH3)n−1SnnI3n+1,8 (with n = 1−5) and IC-
variation.1 When the A-site in the perovskite formula, AMI3, is (NH2)2(CH3NH3)mSnmI3m+2 (with m = 2−4),9 which was
occupied by a small monovalent cation, such as Rb, Cs, reported to display a stepwise increase in conductivity with
methylammonium, or formanidinium, a three-dimensional increasing n or m, reaching its maximum value for the pure
(3D) framework is obtained.2 If a larger cation is used instead CH3NH3SnI3 perovskite. However, the isostructural com-
two-dimensional (2D) layered,3 or one-dimensional (1D) pounds of lead, (RNH3)2(CH3NH3)n−1PbnI3n+1,10 show a
chain4 materials are obtained with formulas AMI3, A2MI4, and much different behavior, displaying a very strong exciton-
A3MI5 which still retain their perovskite (or corner sharing) binding in the excited state. The energy bandgap of these
connectivity. 0D A2SnI6 and A4PbI6 perovskites have also been semiconductors can be easily tuned across a wide range of
obtained.5 The stoichiometry of the final material and the
metal-iodide linkages are extremely sensitive to the nature of Received: May 15, 2013
the cation with both size and functionality playing a crucial
© XXXX American Chemical Society A dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Inorganic Chemistry Article
energies by modifying either the dimensionality of the mm from the sample. For estimating the intensity of the signal and
inorganic framework or the functionality of the organic cation also to account for the excitation beam fluctuations, a disk of single-
(or both), and this is responsible for their unique properties. crystalline InP (930 nm emission peak) was measured before and after
Photoconductivity,2b ionic conductivity,11 electrical conductiv- the sample measurements. The excitation time used was 100 ms, and
the data were averaged for 5 independent scans. The data were
ity,8,9,12 photo-3c and electroluminescence,13 exciton forma- analyzed using the SpectraWiz or the OmniPV_PL software.23
tion,10,14 and phase transitions2 are but a few properties giving Vibrational Spectroscopy. Fourier-transformed infrared spectra
rise to applications ranging from thin film transistors (TFTs)3e (FT-IR) were recorded on a Nicolet 6700 IR spectrometer in the
to solar energy conversion.15 Moreover, the nonperovskite 400−4000 cm−1 spectral region using a diffuse-reflectance kit. The
structures also can exhibit interesting phenomena such measurement is conducted under a constant flow of nitrogen. The
ferroelectricity6i,16 and nonlinear optical properties.6d Solution spectrum is referenced toward a metal mirror used as the 100%
processability of such materials is another key property which reflectance standard.
makes them very attractive for device applications such as all- Charge Transport Properties. (a). Low Temperature Resistivity
Measurements. Four-probe low temperature measurements were
solid state dye-sensitized solar cells.15c,17 performed on a Quantum Design PPMS device in 5−330 K
Our interest in these materials arose from the observation temperature range using the four-probe technique on well-defined
that CsSnI3,18 in its perovskite modification,2g displays intense rhombic dodecahedral single-crystals grown from the solution method.
room temperature photoluminescence (PL) in the near- Current was applied on a rhombic face of the crystal, and the voltage
infrared spectral region when excited with a 532 nm was measured on two of the apexes of the polyhedron (c-axis). The
(Nd:YAG, frequency doubled) laser beam.19 In addition, the resistance was recorded as an average of both negative and positive
success of the CsSnI3 in enabling the efficient all solid state dye- current polarities.
sensitized solar cell motivated us to study the broader family of (b). High Temperature Resistivity and Seebeck Coefficient and
Hall-Effect Measurements. Four-probe room and high temperature
these perovskites as they too are expected to perform similarly. measurements were performed in the 300−550 K range. Measure-
We have been studying the parent compound CsSnI312,15 while ments were made for arbitrary current directions in the ab-plane using
simultaneously working on hybrid perovskite analogues which standard point contact geometry. A homemade resistivity apparatus
we describe here. In this article we report the synthesis and a equipped with a Keithley 2182A nanovoltometer, Keithley 617
broad set of physical properties of CH3NH3SnI3 (1),2d,i,j,8 electrometer, Keithley 6220 Precision direct current (DC) source,
HC(NH 2 ) 2 SnI 3 (2), 9b CH 3 NH 3 PbI 3 (3), 2e,3b,14,20 HC- and a high temperature vacuum chamber controlled by a K-20 MMR
(NH2)2PbI3 (4), and the solid solutions CH3NH3Sn1−xPbxI3 system was used. Data acquisition was controlled by custom written
(5) with x = 0.25 (5a), x =0.50 (5b), and x = 0.75 (5c). Our software. This setup is referred to as “conventional” throught the text.
studies focus on the structural, electrical, and optical properties Seebeck measurements were performed in the 300−550 K
temperature range on the same homemade apparatus using Cr/
of the materials and how they are affected by the inherent Cr:Ni thermocouples as electric leads that were attached to the sample
structural phase transitions and/or by the level of impurities. surface by means of colloidal graphite isopropanol suspension. The
The latter vary based on synthetic route. For comparison temperature gradient along the crystal was generated by a resistor on
purposes, we also prepared the cesium analogues CsSnI3 (6),2f,g the “hot” side of the crystal. The data were corrected for the
and CsPbI3 (7)2c,h to pinpoint the differences and similarities thermocouple contribution using a copper wire. Data acquisition was
with the related hybrid perovskites reported here. We show that controlled by custom written software.
the level of impurities is critical and determines the final Hall-effect measurements were performed at room temperature on a
physical/chemical properties of the materials. Precisely because 4-probe sample holder placed between the plates of an electromagnet.
of this tendency it is possible to tune/optimize the remarkable The magnetic field was varied in the 0.5−1.25 T range. A Keithley
6220 current source was used to excite the sample, and the Hall
emission properties of the materials as we successfully show in voltages were recorded on a Keithley 2182A nanovoltometer. Data
the case of the solid solution 5. The Sn analogues exhibit the acquisition was controlled by custom written software.
very rare characteristic of being simultaneously good electrical Single Crystal X-ray Diffraction Studies. Single-crystal X-ray
conductors and intense room temperature photoluminescent diffraction experiments were performed using either a STOE IPDS II
materials at near-infrared wavelengths. or IPDS 2T diffractometer using Mo Kα radiation (λ = 0.71073 Å)
■ EXPERIMENTAL SECTION
Synthesis. Experimental details and synthetic procedures for the
and operating at 50 kV and 40 mA. Integration and numerical
absorption corrections were performed using the X-AREA, X-RED,
and X-SHAPE programs. All structure were solved by direct methods
and refined by full-matrix least-squares on F2 using the SHELXTL
synthesis of 1−5 can be found in the Supporting Information.21
program package.24 The TwinRotMat, Rotax, and other PLATON
Optical Spectroscopy. Optical diffuse-reflectance measurements
functions25 of the WinGX platform26 were used to identify the
were performed at room temperature using a Shimadzu UV-3101 PC
twinning domains and the correct space group, respectively, for the
double-beam, double-monochromator spectrophotometer operating
low temperature crystal structures. The JANA200627 software was
from 200 to 2500 nm. BaSO4 was used as a nonabsorbing reflectance
employed for the refinement of the multiple twinning domains present
reference. The generated reflectance-versus-wavelength data were used
at low temperature structures. The ISOTROPY27,28 software was used
to estimate the band gap of the material by converting reflectance to
to calculate the group-subgroup relationships of the perovskite
absorbance data according to the Kubelka−Munk equation: α/S = (1
structures and to find the correct space group.
− R)2/2R, where R is the reflectance and α and S are the absorption
and scattering coefficients, respectively.22
PL Measurements. The samples were measured using an
OmniPV Photoluminescence system, equipped with a DPSS
■ RESULTS AND DISCUSSION
Synthetic Aspects. The synthesis of the materials can be
frequency-doubling Nd:YAG laser (500 mW power output, class 4) accomplished using a variety of approaches. This includes
emitting at 532 nm coupled with a bundle of 8 400 μm-core optical
fibers as an excitation source. The emission was measured in the 200− “crude” methods such as direct mortar and pestle grinding of
1700 nm range with two StellarNet spectrometers, covering the 200− starting materials at room temperature in air, high temperature
850 nm (CCD) and 900−1700 nm (InGaAs) areas, coupled with 400 fusion of the precursors in silica tubes, and solution growth of
μm-core optical fibers and integrated within the bundle (10 optical single crystals. All methods produce the title compounds,
fibers total). The backscattered light is collected at ∼180° angle at 3 though the purity and properties of the resulting materials vary
B dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
significantly with the method of synthesis. The previously 4a adopts the hexagonal prism shape. The crystal habit is clearly
studied CsMI3 analogues2c,12 were also prepared and used as reflected on the crystal structure of the solids (Figure 1).
reference materials for comparison with the AMI3 systems. Additional examples are given in Supporting Information,
The materials obtained from the grinding method (see Figure S1.
Supporting Information) usually are a mixture of products, the
main phase (visually homogeneous black solid) being the
desired perovskite, with significant amounts of unreacted
precursors as evidenced by the X-ray powder patterns of the
solids. This method is fast an efficient and is the method of
choice to observe the unusual optical properties of each
compound. The solids obtained through this method
consistently produce the most intense room temperature PL
signal, presumably due to the large degree of defects present or
due to interface effects (vide infra).
Another method that gives the perovskite phases quickly is
by heating the starting materials together at ∼350 °C in air or
under a gentle flow of nitrogen. A pure black melt is obtained in
all cases after 0.5−1 min and on solidification produces a shiny
black material. It is imperative to keep the reaction time short
to prevent the volatilization CH3NH3I and SnI4. The latter can
be seen on the cooler parts of the tube as an orange liquid.
HC(NH2)2I also decomposes at that temperature to the volatile Figure 1. Typical SEM images of 1−4 showing the crystal habits of the
sym-triazine and NH4I.29 The materials prepared in this fashion respective compounds, obtained from the solution method. (a)
are obtained in higher yield compared to the ones prepared at CH3NH3SnI3 (1), (b) HC(NH2)2SnI3 (2), (c) CH3NH3PbI3 (3), (d)
room temperature by grinding, but the presence of oxidation HC(NH2)2PbI3 (4a).
products (i.e., SnI4) and the loss of some of the organic cation
render these materials defective, not following the exact The materials obtained from the aqueous HI/H 3 PO2
stoichiometry. The presence of oxidized impurities is evident solution mixture are exceptionally pure as judged by X-ray
by thermogravimetric analysis (to be discussed in detail later) diffraction (XRD) (Figure 2 and Supporting Information,
by following the mass loss anomalies in the 150−200 °C, where
SnI4 is expected to melt and volatilize. For comparison, the Pb
compounds do not show any mass loss in this region (PbI4 is
not known). Instead, the oxidation mechanism on prolonged
heating for the Pb solids (in air) produces purple vapor of I2
with subsequent formation of PbO in the solid state.
Using solid state reactions we were able to isolate the
materials in pure form by grinding the precursors together,
followed by subsequent annealing at 200 °C inside vacuum-
sealed pyrex tubes for ∼2 h. Again, some condensation was
seen on the cooler parts of the tube (mainly formation of the
white organic iodide salt and/or orange SnI4 crystals in Sn- Figure 2. (a) X-ray diffraction patterns of samples of CH3NH3SnI3 as
containing compounds at the sand bath-air interface). There- functions of the preparative method. (b) X-ray diffraction patterns
fore, the materials prepared by this method are stoichiometri- showing the structural evolution of CH3NH3SnI3 standing in moist air
cally defective as well, but contain no oxidation byproducts over time. The extra peaks that appear most likely belong to the
compared to those from the open tube methods. (CH3NH3)2SnI6 compound, as judged indirectly by comparison with
The method of choice to obtain pure materials with exact the simulated pattern of cubic Cs2SnI6 (Fm3m ̅ ) which is expected to
have a similar structure. The effect is more pronounced following the
stoichiometries and lowest number of defects is the solution
extended exposure of the material to the atmosphere.
method. Through this route we were able to obtain high quality
single crystals that allowed X-ray crystal structure character-
ization. Our synthetic procedure is a modification on a reported Figures S2−S4). This is mainly attributed to the reducing
method18 and gives no observable impurities while it allows for nature of phosphinic acid which is sufficient to suppress any
easier manipulation of the materials. The crystals of the oxidation occurring by reducing I3− (formed by aerial oxidation
obtained materials have a well-defined habit which can even be of the solution) back to I−. This is evident from the
modified depending on the temperature profile of the disappearance of the dark red color of the triiodide ion from
precipitation. Namely, the products vary from polycrystalline solution (when oxidation occurs) which turns bright yellow
aggregates when precipitation starts at elevated temperatures when treated with phosphinic acid. The purging effects of
(ca. 130 °C) under vigorous stirring to discrete polyhedral phosphinic acid ensure that no stray oxidation products are
crystals when crystallization occurs by cooling slowly to room present during precipitation, while no traceable byproducts
temperature. The crystals have a strong tendency toward a were found as a result of the presence of H3PO2/H3PO3 in the
rhombic dodecahedral geometry in general, though each phase reaction. In most cases the sole product of the reaction was a
adopts a unique habit; thus, 1 and 3 adopt the elongated pure black crystalline phase. Interestingly, in the case of
rhombic dodecahedron (rhombo-hexagonal dodecahedron) HC(NH2)2PbI3 we were able to isolate two isomers; a black
shape, 2 adopts a more regular rhombic dodecahedron, while cubic perovskite phase (4a) and a yellow hexagonal perovskite
C dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Figure 3. Graphical scheme of the observed phase transitions in the Pb and Sn hybrid perovskites using compound 4 as a representative example.
Starting from the α-phase two possibilities arise: In the presence of a liquid interface (i.e. inside the mother liquor) it converts to the δ-phase upon
decreasing the temperature (below 360 K). When dry crystals of the α-phase are cooled, however, a different sequence of phase transitions occur
with conversion to the β- (below∼ 200 K) and γ- (below ∼130 K) phases. Precession images are drawn at [006] view.
phase (4b) having a different crystal structure. It appears that The stability of the materials in air varies. The Pb-containing
the black phase 4a is stable at high temperatures (>60 °C) phases are stable in air for months, though they are affected by
while at lower temperatures the black phase fully converts to humidity and lose their crystalline luster after a couple of weeks.
yellow phases inside the mother liquor. When the products are However, this is only a surface effect since the bulk properties
separated from the solvent and dried they interconvert only of the materials are retained. The Sn-containing materials are
slowly. air and moisture sensitive and partially decompose within 2 h
A complication that occurs when trying to isolate pure 4a is before total decomposition after 1 d. This behavior is detected
the partial decomposition of formamidine to ammonia and sym- in the powder diffraction pattern evolution over time (Figure
triazine in the strongly acidic reaction media using a prolonged 2b). The patterns show the appearance of extra Bragg
reaction time.29 Whereas the sym-triazine does not interfere in diffraction peaks that belong to the oxidized SnIV species
the reaction, the ammonium cation slightly complicates the (CH3NH3)2SnI6, as judged by comparing the diffraction pattern
reaction. NH4PbI3·2H2O,30 occurring as pale yellow needles, of the unindexed peaks with those of Cs2SnI6. A general trend
observed is that methylammonium-containing materials are
or,31 as yellow needles, can be detected in 4 as an impurity. The
considerably more stable than formamidinium-containing ones.
effect can be minimized by keeping the reaction time as short as
This reflects the respective stability of the cations themselves
possible and by avoiding temperatures >150 °C. and particularly the tendency of HC(NH2)2+ to dissociate to
It is important to point out that in the case of the Cs ammonia and sym-triazine.29
analogues CsSnI3 and CsPbI3 the solution method cannot be Structure Description. Single crystals of the materials
employed for the preparation of the black forms of these obtained from the solution method were analyzed by single
phases, and the yellow phases are isolated instead. The yellow crystal X-ray diffraction studies. The diffraction measurements
modifications of CsSnI3 (6a),21,12 CsPbI3 (7),2c are isostruc- were performed at room temperature, and in the range 100−
tural and crystallize in long, thin, needle-shaped crystals, in 400 K to determine the structural changes that occur during the
accordance with the 1D structure of these materials. The several phase transitions. The presence of phase transitions is
solution method is also capable of producing the black Cs2SnI6 consistent with the perovskite nature of the materials.1 The
phase5b (6b), an oxidized byproduct in the synthesis of CsSnI3 common trend for 1−4 is that there are two phase transitions
with a molecular salt structure. that take place leading to three distinct structural modifications.
D dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Figure 4. Perovskite building block of (a) α-CH3NH3SnI3, (b) α-HC(NH2)2SnI3, (c) β-CH3NH3PbI3, (d) α-HC(NH2)2PbI3. Color code: I =
purple, Sn = red, Pb = green, C = deep blue, N = pink.
These are classified as the high temperature α-phase, the consequence, these cations tend to be disordered in the cage. In
intermediate temperature β-phase, and the low temperature γ- a cubic unit cell this requires one atom to be placed exactly on
phase while the nonperovskite yellow phase is termed the δ- the special position and the other(s) to be disordered around
phase for consistency with the nomenclature used in the studies the 12 possible orientations inside the cube. Such a situation is
of CsSnI3.2f,g,12 The δ-phase denotes a 1D array of octahedra, not particularly appealing on chemical grounds, especially so if
and it occurs in δ-HC(NH2)2PbI3 (4b). A graphical scheme of one considers the hydrogen bonding capabilities of both
the phase transitions are summarized in Figure 3. CH3NH3+ and HC(NH2)2+. Moreover, when a polar organic
α-Phase Structures. The α-phase for 1 and 2 is found at cation is confined inside the cage its dipole moment cannot be
room temperature and consists of pseudocubic unit cell with canceled out, thus inducing noncentrosymmetric domains in
lattice parameter being a = 6.2307(10) Å and a = 6.3290(9) Å, the crystal.
respectively. The α-phase for 3 is also pseudocubic and forms With this in mind, in our consideration of the crystal
after a phase transition that occurs at temperature >50 °C42 structure determinations when an ambiguity occurs we prefer
with lattice parameter a = 6.3130(2) Å. Compound 4a is the noncentrosymmetric space group as the correct choice. Of
unique in adopting a trigonal phase with a = 8.9817(13) Å and course, when supercells arise from the phase transition an
c = 11.0060(20) Å at room temperature. We are as yet inversion center might emerge which will cancel the dipole
uncertain if this is the α-phase for 4a or if there is yet another
moments. At room temperature, the crystal structures of 1−4
higher symmetry phase at higher temperatures. For simplicity,
are noncentrosymmetric. The observed space groups (except
here we refer to 4a as the α-phase. The room temperature
for β-CH3NH3PbI3 which already has a supercell at this
structures of 1−4 are shown in Figure 4 while the detailed
crystallographic information is presented in the Supporting temperature) are classified into the “ferroelectric distortion”
Information. category, which means that they are subgroups of the ideal
Because phase transitions affect the PL properties of CsSnI3 Pm3̅m space group by abolishing certain symmetry elements
(which adopts the orthorhombic γ-phase at room temper- including the center of symmetry. In this case the symmetry
ature),19 we decided to investigate the phase transitions that elements are lost as a result of the displacement of the B cation
occur in the present hybrid perovskites at different temper- from the center of the octahedron, though the ferroelectric
atures. A valuable guide was the recent group-theoretical work displacement is only marginal (0.08 Å along c-axis for 1, 0.05 Å
on perovskite structures.32 The main difference between CsMI3 along b and c axes for 2, 0.05 Å along c-axis for 3 and ∼0.004 Å
inorganic perovskites and the organic hybrid perovskites is the for along c-axis for 4a from the ideal position). In P4mm, which
possibility of obtaining the ideal cubic Pm3̅m structure. is adopted by the higher temperature phase of CH3NH3SnI3
Whereas in the case of Cs+ a single atom occupies the 1b (1) and CH3NH3PbI3 (3), the displacement occurs along the
Wyckoff position in the center of the cube of the perovskite tetragonal c-axis while in Amm2, the space group adopted by
structure in a primitive unit cell, in the cases of CH3NH3+ and HC(NH2)2SnI3 (2) at high temperature, the M atom
HC(NH2)2+ this is not possible in this space group. As a displacement occurs along the a and b axes.
E dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
α-HC(NH2)2PbI3 (4a) is a special case because its space β-Phase Structures. Attempts to determine the crystal
group (centrosymmetric P3̅m1) seems to be the parent phase structures of the hybrid AMI3 materials revealed the second
for the family of perovskites termed “1:2 B cation ordering”.32b major well-known property of perovskites, namely, multiple
The peculiarity arises from the fact that there is only one type twinning on lowering of crystallographic symmetry.33 This
of B-cation -Pb- and the reason behind this ordering presently happens when moving from high to low symmetry distorting
is not understood. Following this distortion, there are three the unit cell. For the crystal to comply with such a distortion it
crystallographically inequivalent Pb cations inside the unit cell. “splits” into several domains, a fact that certainly complicates
In Figure 5 the symmetry elements of the space groups are the structure determination. All typical pseudomerohedral twin
highlighted with respect to the ideal parent structure. domains characteristic of the perovskite structure were
observed during structure refinement, along the [010], [101],
and [121] lattice directions.1 In all cases the effect was f ully
reversible and when the temperature was raised the crystal was
completely reassembled. Another interesting point here is that
when we move from the low to the higher symmetry phase
during a phase transition, this effect is not so severe, as it was
observed, for example, in the case of CH3NH3PbI3. Effectively,
this behavior suggests that the transition is smoother on
lowering the symmetry and more abrupt when increasing it, a
trend that is confirmed by the charge transport measurements
(vide infra).
The temperature range where the α-phase converts to the β-
phase lies between 300 and 150 K for the hybrid perovskites,
with the exception of CH3NH3PbI3 where the transition takes
place at 333 K. Lowering of the symmetry is achieved by means
of tilting of the octahedra. Pure octahedral tilting starting from
an ideal cubic Pm3m ̅ structure follows a specific sequence of
descending symmetry leading to a certain set of possible space
groups. Such a situation is described using the Glazer
notation.34 This notation describes the distortion about a, b,
c axes using the axis symbol accompanied by a “+” or “−”
symbol to denote if the tilting between two adjacent polyhedral
is in-phase or out-of-phase, respectively. Following the space
group sequence for CsSnI3, starting with the ideal Pm3̅m
structure, it follows a phase transition at ∼425 K to the
tetragonal structure, P4/mbm (double cell volume), before
obtaining an orthorhombic structure, Pnma (quadruple cell
volume) below 351 K. Such a sequence can be described by
considering a pure perovskite tilting model, ignoring ferro-
electric-type displacements, with an in-phase distortion along c-
axis initially and distortions along the a- and b-axes in the
orthorhombic phase. The situation for the hybrid perovskites
appears to be different because both ferroelectric and tilting
distortions should be taken into consideration. Crystallographic
data for the α- and β-phases of 1−4 are given in Tables 1 and 2.
Table 3 summarizes the crystallographic data for the related
compounds 4b, 5b, 6b, and 7. Atomic coordinates can be found
Figure 5. Unit cells of (a) the α-phases of CH3NH3SnI3 and in the Supporting Information, Tables S1−S24.
CH3NH3PbI3 (001 view), (b) HC(NH2)2SnI3 (001 view), and (c) Both CH3NH3SnI3 and CH3NH3PbI3 distort in a similar
HC(NH2)2PbI3 (011̅ view) and demonstrating slight and strong off- manner in their β-phase, showing out-of-phase tilting of the
centering distortions in the octahedra. Color labeling is according to polyhedra combined with ferroelectric type of off-centering
the Wyckoff positions of each space group. along the c-axis adopting the tetragonal I4cm space group
(Figure 5). The ferroelectric displacement component is once
again microscopic effecting a slight polarization of the
CH3NH3PbI3 adopts the β-phase at room temperature which octahedra along the c-axis. The tilting angle is 17.4° and
is distorted (vide infra), and for this reason we sought to solve 16.4° for CH3NH3SnI3 and CH3NH3PbI3 as measured at 150 K
the structure of the α-phase using the same specimen by raising and 293 K, respectively. At 180 K the framework of
the temperature above the reported transition at ∼330 K.20a,c HC(NH2)2SnI3 is found to be orthorhombic Imm2. The
We solved the crystal structure of α-CH3NH3PbI3 by measuring onset of the transition is at approximately 250 K. It displays two
a crystal at 400 K and found it to be isostructural to α- major distortions at 9.7° and 11.2° and a minor one at 2.2°.
CH3NH3SnI3. For the collection of good X-ray diffraction data HC(NH2)2PbI3 at 150 K adopts the P3 trigonal space group
sets we found it essential for the crystal to be maintained at the following a quadrupling of the room temperature cell volume.
collection temperature for at least 2 h (to complete the phase There is a broad array of distortions, due to significant off-
transition) before starting the data collection. centering of Pb, roughly averaged between the values of 9.0°
F dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Table 1. Crystallographic Data of the α-Phases for 1−4a

α-CH3NH3SnI3 (1) α-HC(NH2)2SnI3 (2) α-CH3NH3PbI3 (3) α-HC(NH2)2PbI3 (4a)
temperature 293(2) K 340(2) K 400(2) K 293(2) K
crystal system tetragonal orthorhombic tetragonal trigonal
space group P4mm Amm2 P4mm P3m1
unit cell a = 6.2302(10) Å, α = 90.00° a = 6.3286(10) Å, α = 90.00° a = 6.3115(2) Å, a = 8.9817(13) Å,
dimensions α = 90.00° α = 90.00°
b = 6.2302(10) Å, β = 90.00° b = 8.9554(11) Å, β = 90.00° b = 6.3115(2) Å, b = 8.9817(13) Å,
β = 90.00° β = 90.00°
c = 6.2316(11) Å, γ = 90.00° c = 8.9463(11) Å, γ = 90.00° c = 6.3161(2) Å, γ = 90.00° c = 11.006(2) Å,
γ = 120.00°
volume 241.88(7) Å3 507.03(12) Å3 251.60(2) Å3 768.9(2) Å3
Z 1 2 1 3
density 3.649 g/cm3 3.566 g/cm3 4.092 g/cm3 4.101 g/cm3
(calculated)
absorption 12.128 mm−1 11.579 mm−1 25.884 mm−1 25.417 mm−1
coefficient
F(000) 228 468 260 798
crystal size 0.053 × 0.032 × 0.021 0.176 × 0.163 × 0.136 0.035 × 0.022 × 0.017 0.041 × 0.026 × 0.021
(mm−3)
θ range 3.27 to 28.94° 3.22 to 29.11° 3.23 to 24.92° 1.85 to 29.19°
index ranges −8 ≤ h ≤8, −8 ≤ h ≤8, −7 ≤ h ≤7, −12 ≤ h ≤12,
−8 ≤ k ≤ 8, −10 ≤ k ≤ 12, −6 ≤ k ≤ 6, −12 ≤ k ≤ 11,
−8 ≤ l ≤ 8 −12 ≤ l ≤ 10 −7 ≤ l ≤ 7 −15 ≤ l ≤ 15
reflections 2164 2490 1628 7475
collected
independent 411 [Rint = 0.0458] 686 [Rint = 0.0517] 300 [Rint = 0.0700] 1645 [Rint = 0.0459]
reflections
completeness θ = 28.94°, 98.2% θ = 29.11°, 100% θ = 24.92°, 98.2% θ = 29.19°, 99.6%
to θ
data/restr./ 411/2/16 686/3/24 300/2/19 1645/7/44
param.
goodness-of-fit 1.262 0.861 1.176 0.908
final R indices Robs = 0.0327, Robs = 0.0323, Robs = 0.0417, Robs = 0.0398,
[>2σ(I)]
wRobs = 0.0576 wRobs = 0.0577 wRobs = 0.0981 wRobs = 0.0666
R indices [all Rall = 0.0451, Rall = 0.0615, Rall = 0.0417, Rall = 0.0892,
data]
wRall = 0.0613 wRall = 0.0632 wRall = 0.0981 wRall = 0.0807
2nd twin domain [−1 0 0 0 0 1 0 1 0]49(69) % [0 −1/2 −1/2 −1 −1/2 1/2 −1 1/2 −1/2] [−1 0 0 0 0 1 0 1 0] [−1 0 0 0 −1 0 0 0 1]
43(3) % 36(3) % 6.36(2) %
Flack parameter [−1 0 0 0 −1 0 0 0 −1] −4(45) % [−1 0 0 0 −1 0 0 0 −1] −50(32) % [−1 0 0 0 −1 0 0 0 −1] [−1 0 0 0 −1 0 0 0 −1]
12(3) % 20(15) %
2nd domain [1 0 0 0 0 1 0 1 0] 3(68) % [0 −1/2 −1/2 −1 1/2 1/2 −1 1/2 1/2] [1 0 0 0 0 1 0 1 0] 37(4) % not refined
Flack 13(3) %
parameter
extinction 0.0200(4) 0.016(2)
coefficient
largest diff. peak/ 0.895/−0.861 e·Å−3 0.394/−0.952 e·Å−3 1.871/−2.008 e·Å−3 1.104/−1.727 e·Å−3
hole
wavelength 0.71073 Å
refinement full-matrix least-squares on F2
method
a
R = ∑||Fo| − |Fc||/∑|Fo|, wR = {∑[w(|Fo|22 − |Fc|2)2]/∑[w(|Fo|44)]}1/2 and calc w = 1/[σ2(Fo2) + (0.0262P)2 + 0.0000P] where P = (Fo2 + 2Fc2)/3.
and 11.5°. For good measure, the 373 K distortion of β-CsSnI3 CH3NH3PbI3 this temperature was found to be at ∼150 K. The
is 18.4°. γ-phases are easily recognizable from diffraction studies because
From Figure 6 it is apparent that the methylammonium of an intense broadening and splitting of the Bragg peaks in
compounds have a tendency to distort the octahedra in the out- addition to the extra spots that appear. This can be clearly seen
of-phase mode, whereas formamidinium phases seem to distort in the precession images generated from diffraction data
in-phase, the latter being similar to β-CsSnI3. Contrary to the collected at 100 K (Supporting Information, Figure S5), as
purely inorganic phase, however, the distortions lead to lower compared to those created from room temperature data
symmetry of the lattice whence the tilting of the octahedra is (Supporting Information, Figure S6). We were unable to refine
less severe. the structure of the γ-phases mostly because about 50% of the
γ-Phase Structures. As the temperature continues to observed peaks remain unindexed during the unit cell
decrease further symmetry lowering occurs leading to the γ- determination. The unindexed peaks result from the formation
phases. The γ-phase is encountered close to 100 K, but for γ- of multiple twin domains which are created by the “freezing” of
G dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Table 2. Crystallographic Data of the β-phases for 1−4a

β-CH3NH3SnI3 (1) β-HC(NH2)2SnI3 (2) β-CH3NH3PbI3 (3) β-HC(NH2)2PbI3 (4a)
temperature 200(2) K 180(2) K 293(2) K 150(2) K
crystal system tetragonal orthorhombic tetragonal trigonal
space group I4cm Imm2 I4cm P3
unit cell dimensions a = 8.7577(15) Å, α = 90.00° a = 12.5121(9) Å, α = 90.00° a = 8.849(2) Å, α = 90.00° a = 17.7914(8) Å, α = 90.00°
b = 8.7577(15) Å, β = 90.00° b = 12.5171(8) Å, β = 90.00° b = 8.849(2) Å, β = 90.00° b = 17.7914(8) Å, β = 90.00°
c = 12.429(3) Å, γ = 90.00° c = 12.5099(9) Å, γ = 90.00° c = 12.642(2) Å, γ = 90.00° c = 10.9016(6) Å,
γ = 120.00°
volume 953.2(3) Å3 1959.2(2) Å3 990.0(4) Å3 2988.4(3) Å3
Z 4 8 4 12
density (calculated) 3.703 g/cm3 3.692 g/cm3 4.159 g/cm3 4.221 g/cm3
absorption coefficient 12.309 mm−1 11.986 mm −1
26.312 mm−1 26.159 mm−1
F(000) 912 1872 1040 3192
crystal size (mm−3) 0.158 × 0.111 × 0.089 0.018 × 0.014 × 0.009 0.005 × 0.003 × 0.002 0.038 × 0.030 × 0.023
θ range 4.65 to 24.94° 2.30 to 29.15° 3.26 to 29.03° 1.87 to 25.00°
index ranges −9 ≤ h ≤ 10, −17 ≤ h ≤ 17, −12 ≤ h ≤ 12, −21 ≤ h ≤ 20,
−10 ≤ k ≤ 9, −17 ≤ k ≤ 17, −12 ≤ k ≤ 10, −21 ≤ k ≤ 21,
−14 ≤ l ≤ 14 −17 ≤ l ≤ 17 −16 ≤ l ≤ 15 −12 ≤ l ≤ 12
reflections collected 3058 9462 4483 19512
independent reflections 521 [Rint = 0.1007] 2876 [Rint = 0.0464] 678 [Rint = 0.0664] 7026 [Rint = 0.0596]
completeness to θ θ = 24.94°, 99.2% θ = 29.15°, 99.7% θ = 29.03°, 97.1% θ = 25.00°, 100%
data/restr./param. 521/2/21 2876/1/47 678/2/18 7026/18/192
goodness-of-fit 1.142 0.963 1.268 0.839
final R indices [>2σ(I)] Robs = 0.0637, Robs = 0.0486, Robs = 0.0378, Robs = 0.0656,
R indices [all data] Rall = 0.0639, Rall = 0.0780, Rall = 0.0418, Rall = 0.1780,
2nd twin domain [−1/2 1/2 1/2 1/2 −1/2 1/2 1 1 0] 7.9(2) % [0 0 −1 0 −1 0 −1 0 0] [−1 0 0 0 1 0 0 0 −1] [−1 0 0 0 −1 0 0 0 1]
19.74(6) % 50(4) % 94.19(2) %
Flack parameter [−1 0 0 0 −1 0 0 0 −1] −34(23) % [−1 0 0 0 −1 0 0 0 −1] [−1 0 0 0 −1 0 0 0 −1] [−1 0 0 0 −1 0 0 0 −1]
9(5) % 0(10) % 51(5) %
2nd domain Flack [1/2 1/2 1/2 1/2 1/2 1/2 1 1 0] 16.7(2) % [0 0 −1 0 1 0 −1 0 0] not refined not refined
parameter 23.78(6) %
extinction coefficient 0.00247(6) 0.000808(18) 0.000498(8)
largest diff. peak/hole 2.002/−1.194 e·Å−3 2.606/−1.191 e·Å−3 0.927/−1.676 e·Å−3 3.391/−4.496 e·Å−3
refinement method full-matrix least-squares on F2
a
the cations (and subsequently the perovskite lattice) in different δ-Phase Structures. Interestingly, the δ-phase of HC-
orientations. From our attempts to solve the crystal structures, (NH2)2PbI3 (Figure 7) is markedly different from the δ-phase
we estimate that the γ-phases adopt a monoclinic crystal occurring in CsSnI3 and CsPbI3; in the latter case the structure
system. The methylammonium phases of Sn and Pb seem to comprises a double chain of MI64− octahedra.2c,36 The
acquire a C-centered cell whereas the formamidinium perov- intrachain linking of the polyhedra in the individual chains is
skites seem to adopt a primitive cell. It is however necessary to affected by edge-sharing of square bipyramidal units. The
highlight a distinction between the methylammonium and octahedral coordination occurs by interchain bonding by a
formamidinium behavior; whereas the γ-phase of the former relatively longer M−I bond with an iodide occupying a basal
contains many split peaks, for the latter the splitting is much position of the square pyramid, overall adopting the
less intense and instead superlattice ordering is observed NH4CdCl3-type structure. δ-HC(NH2)2PbI3 on the other
hand consists of face-sharing octahedra propagating along the
leading to multiplication of the parent unit cell. This may be
001 direction forming single chains, with the cations occupying
attributed to the point group symmetry of CH3NH3+ (C3v) as
the space between the chains, thus adopting a nearly hcp-lattice
opposed to that of HC(NH2)2+ (C2v), with the latter being (P63mc). The deviation from the ideal symmetry can be
more “flexible” to comply with the lattice distortions. The phase attributed to the HC(NH2)2+ cations which lack a center of
transitions of a similar compound, CH3NH3SnBr3, have been inversion despite the fact that the nitrogen atoms where treated
recently investigated by means of high resolution powder as disordered during the structure determination process. At
diffraction studies.35 The authors were able to solve the crystal this point we should stress that the black-to-yellow trans-
structure of the β-CH3NH3SnBr3 modification, but they too formation is a reversible effect. For example, when a single
were unable to refine the crystal structure of the γ-phase in a crystal of δ-HC(NH2)2PbI3 is heated at 400 K for ∼2 h a
satisfactory manner upon entering into the peak splitting polycrystalline black solid is formed instead, though not in a
regime. They concluded that the crystal structure may be single-crystal-to single-crystal manner. Interestingly, while the
triclinic based on density function theory (DFT) calculations. α- to δ-phase transformation can be repeated indefinitely in the
H dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Table 3. Crystallographic Data of the β-phases for 4b, 5b, 6b, and 7a
δ-HC(NH2)2PbI3 (4b) β-CH3NH3Sn0.46Pb0.54I3 (5b) α-Cs2SnI6 (6b) δ-CsPbI3 (7)
temperature 293(2) K 293(2) K 293(2 K) 293(2) K
crystal system hexagonal tetragonal cubic orthorhombic
space group P63mc I4cm Fm3m ̅ Pnma
unit cell dimensions a = 8.6603(14) Å, α = 90.00° a = 8.8552(6) Å, α = 90.00° a = 11.6276(9) Å, a = 10.4342(7) Å,
α = 90.00° α = 90.00°
b = 8.6603(14) Å, β = 90.00° b = 8.8552(6) Å, β = 90.00° b = 11.6276(9) Å, b = 4.7905(3) Å,
β = 90.00° β = 90.00°
c = 7.9022(6) Å, γ = 120.00° c = 12.5353(12) Å, γ = 90.00° c = 11.6276(9) Å, c = 17.7610(10) Å,
γ = 90.00° γ = 90.00°
volume 513.27(12) Å3 982.95(13) Å3 1572.1(2) Å3 887.78(10) Å3
Z 2 4 4 4
density (calculated) 4.096 g/cm3 3.870 g/cm3 4.842 g/cm3 5.393 g/cm3
absorption coefficient 25.384 mm−1 19.219 mm−1 17.925 mm−1 33.373 mm−1
F(000) 532 964 1912 1184
crystal size (mm−3) 0.062 × 0.046 × 0.031 0.191 × 0.160 × 0.109 0.071 × 0.053 × 0.044 0.265 × 0.022 × 0.013
θ range 2.72 to 24.99° 2.82 to 29.01° 3.03 to 28.96° 2.26 to 29.16°
index ranges −10 ≤ h ≤ 7, −10 ≤ h ≤ 12, −15 ≤ h ≤ 15, −14 ≤ h ≤ 14,
−9 ≤ k ≤ 10, −11 ≤ k ≤ 11, −14 ≤ k ≤ 15, −5 ≤ k ≤ 6,
−8 ≤ l ≤ 9 −17 ≤ l ≤ 17 −15 ≤ l ≤ 15 −22 ≤ l ≤ 24
reflections collected 1687 4728 3871 7985
independent reflections 354 [Rint = 0.0759] 788 [Rint = 0.0738] 143 [Rint = 0.0595] 1330 [Rint = 0.0733]
completeness to θ θ = 24.99°, 100% θ = 29.01°, 100% θ = 28.96°, 100% θ = 29.16°, 99%
data/restr./param. 354/3/17 788/2/22 143/0/7 1330/0/32
goodness-of-fit 1.055 1.162 1.483 1.069
final R indices [>2σ(I)] Robs = 0.0437, Robs = 0.0678, Robs = 0.0317, Robs = 0.0266,
R indices [all data] Rall = 0.0516, Rall = 0.0809, Rall = 0.0322, Rall = 0.0364,
2nd twin domain [−1 −1 0 0 1 0 0 0 −1] [−1/2 1/2 −1/2 1/2 −1/2 −1/2 −1 −1 0]
39(7) % 25(1) %
Flack parameter [−1 0 0 0 −1 0 0 0 −1] [−1 0 0 0 −1 0 0 0 −1] 28(5) %
0(3) %
2nd domain Flack not refined [1/2 1/2 −1/2 1/2 1/2 −1/2 −1 −1 0]
parameter 22(1) %
extinction coefficient 0.0081(15) 0.00084(10) 0.00247(12) 0.0127(3)
largest diff. peak/hole 1.975/−0.721 e·Å−3 2.980/−2.175 e·Å−3 1.005/−1.241 e·Å−3 0.887/−1.986 e·Å−3
refinement method full-matrix least-squares on F2
a
mother liquor by running heating and cooling cycles in the mass loss corresponding to loss of HI molecules. An example of
absence of a liquid medium, the transformation is irreversible; this is CH3NH3SnI3 (1), where the mass loss depends on the
that is, the polycrystalline black material of the α-phase preparative method employed (Figure 8a). It is clearly seen that
obtained from heating a crystal of the δ-phase does not convert the mass loss is associated with the purity of the material.
back to the yellow phase on cooling or it does so only very Therefore, it seems unlikely that the first decomposition step is
slowly. This is likely due to the partial loss of formamidinium associated with loss of the organic cation alone. For example,
resulting in a defective compound. when the material is prepared by grinding, the first step of
Thermal Analysis. The thermal analysis results for the decomposition displays a much greater mass loss as compared
hybrid perovskites presented here show similar trends for Sn to the material prepared by high temperature solid state or
and Pb. The experiments were conducted under a nitrogen flow solution methods. A plausible explanation for this behavior is
to avoid oxidation. Thermogravimetric analysis of the materials the presence of Sn4+ species in the compounds. In that case, the
shows that they decompose before melting. Decomposition mass loss could be associated with the liberation of volatile
starts at temperatures >300 °C (Figure 8) and proceeds in one SnI4. Contributing to the weight loss is the liberation of
step. The mass loss corresponds to loss of the organic cation CH3NH3I (in the form of CH3NH2 and HI) which is driven
and all the iodide content leaving a metallic residue. away due to the formation of Sn4+ species. It is worth pointing
The decomposition profiles seem to be dependent on the out that the decomposition temperature of CH3NH3I is
synthetic conditions used to prepare the compounds. When the roughly 280 °C, while the decomposition of partly oxidized
materials are prepared in air the decomposition follows a CH3NH3SnI3 (with weight loss) starts at ∼150 °C. Further
different pathway that includes two distinct weight loss steps. In support for the presence of Sn4+ species is provided by the
that case, the decomposition proceeds initially with a mass loss absence of a corresponding “two step” decomposition profile in
corresponding to “CH3NH3I” or “HC(NH2)2I”, possibly in the CH3NH3PbI3, although a two-step decomposition is indeed
form of the free amine and HI, followed by a second step in observed for the solution prepared sample. In the former case
I dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Figure 6. Unit cells of (a) the β-phases of CH3NH3SnI3 and CH3NH3PbI3 (001 view), (b) HC(NH2)2SnI3 (001 view) highlighting the tilting of the
octahedra. (c) The unit cell of β-CsSnI3 (ICSD-69995), given here for comparison purposes. (d) The unit cell of HC(NH2)2PbI3 (021̅ view). Color
labeling is according to the Wyckoff positions of each space group.
Figure 7. Unit cells of (a) δ-CsPbI3 and (c) δ-HC(NH2)2PbI3 viewed along the [010] and [100] directions, respectively. The corresponding building
blocks of the 1D chains are shown in (b) and (d). The nitrogen atoms of the formamidinium atoms are disordered down the 3-fold symmetry axis
with 2/3 partial occupancy in (c).
J dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX

Optical Properties. (a). Electronic Spectroscopy. All

materials display a strong, abrupt absorption in the visible
spectral region (Figure 9). The absorption has been attributed
Figure 8. (a) TGA analysis of CH3NH3SnI3 as a function of the

synthetic route employed. A two-step decomposition process can be
seen for the less “pure” phases. (b) TGA analysis of CH3NH3PbI3
prepared from the solution and open tube methods showing no
changes as a function of the preparative route. The measurements
were conducted under a constant flow of nitrogen to avoid oxidation.
The spike in the data that appears during the decomposition step is
from the abrupt gas evolution which perturbs the solid and
momentarily loses contact with the sample holder.
this could be explained by the inability of Pb2+ to oxidize to

Pb4+ in an all iodide environment.
As stated above, CH3NH3PbI3 does undergo a structural
phase transition going from tetragonal to pseudocubic
symmetry between 300 and 400 K as determined by single-
crystal diffraction studies. The reason for the absence of any Figure 9. Electronic absorption spectra of 1−4. Panels (a) and (b)
thermal signal for this transition may suggest that the transition highlight the band gap variability on the basis of the synthetic
is of second order as it has been predicted.37 The transition approach employed to prepare the sample. (c) The spectra of
itself does not involve chemical bond cleavage/formation; the specimens of 1−4 prepared with the solution method as compared
only real change is associated with lowering of symmetry. with CsSnI3 and CsPbI3 reference phases (dotted lines) prepared from
Subsequently, the change in the heat capacity of the material, C the open tube method. The value of the band gap is given next to each
= Q/ΔT, is expected to be small and explains the lack of signal compound.
in DTA and DSC. The transition has been reported to occur at
330 K20c though the sensitivity of our experimental setup did to the energy gap between the conduction and the valence
not allow us to reproduce this result. Plots of the thermal band, thus classifying the materials as medium-band gap
analysis data are given in Supporting Information, Figure S7. semiconductors. The Sn and Pb based materials display an
Nevertheless, the phase transition can be better detected by optical band gap between 1.2 and 1.6 eV, in good agreement
other techniques that are either (local) symmetry sensitive or with the black color of the solids.
detect properties that are affected by symmetry. Such examples The band gaps of the Sn-based perovskites, have mean values
involve but are not limited to dielectric constant measur- between 1.2 and 1.4 with CH3NH3SnI3 (1) having a
ements,20a,38 127I NQR,20b single-crystal resistivity,2i 1H NMR consistently smaller band gap with respect to HC(NH2)2SnI3
spin−lattice relaxation,20b,39 X-ray diffraction,40 and Raman (2) for every preparative method employed. We use the term
spectroscopy.41 An excellent discussion on the crystal chemistry “mean value” because as can be seen in Figure 9 there is a
of CH3NH3PbI3 (3) has appeared during the preparation of the variation of the band gap as a function of the preparative
present manuscript.42
method. There is no obvious trend on this variation besides the
In the case of compounds 3 and 4 a similar behavior can be
interesting fact that CH3NH3SnI3 samples prepared in an open
observed with the addition of a small mass loss associated with
the evaporation of formamidine or one of its decomposition tube (1.35 eV), prepared from solution, or in a sealed tube
products.25 The case of HC(NH2)2PbI3 (4a) is significantly reaction (1.21 eV) show very different properties. This is
more complicated as there is some contamination from phase despite the fact that their thermal behavior is very similar as is
(4b) which is probably responsible for the small exothermic their diffraction pattern. When compared with CsSnI3,12 1
mass loss observed at 270 °C (Supporting Information, Figure shows a slight narrowing on the gap whereas 2 shows the
S8). A decolorized (pale brown) residue remains, indicating the opposite trend which is consistent with the lowering in the
complete loss of iodine from the solid after the analysis. symmetry of the 3D [SnI3]− framework.
K dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Figure 10. PL spectra of (a) ASnI3 and (b) APbI3 as obtained from the open tube reaction. The data are normalized toward a single crystalline
sample of InP which is used as a benchmark. The Cs analogues, prepared in the same manner, are given for comparison purposes.
In contrast to the Sn compounds, the Pb-based phases have produces samples with high PL. Annealing of the ground
band gaps which are independent of the preparation method at material also produces solids with respectable luminescence,
1.45 and 1.52 eV for HC(NH2)2PbI3 (4a) and CH3NH3PbI3 slightly weaker than the ground ones, but the materials are
(3), respectively. Similar to the Sn-phases, a narrowing of the highly homogeneous. However, we observed that exhaustive
band gap is observed for the Pb compounds upon symmetry annealing, that is, > 24 h, reduces the PL emission (vide infra).
lowering, when compared to the room temperature absorption When synthesized under vacuum the materials obtained are
of the black perovskite form of CsPbI324 obtained by a solid pure and homogeneous, though the PL-properties are notice-
state reaction, before the material transitions to the yellow ably weakened compared to the samples prepared at ambient
nonperovskitic phase (Figure 9c). The absorption edge for both pressure. When exposed to the atmosphere, the materials
the Sn- and Pb-based phases is sharp, suggesting a direct band decompose within a few minutes, displaying an extreme
gap. This has been confirmed by theoretical calculations, where sensitivity. The solution-processed solids of both Sn and Pb
all electronic structure models reported predict a direct band perovskites are PL-inactive. This suggests that the presence of
gap for the Sn- and Pb-analogues.12,43 defects such as Sn4+ species in the structure which are more
Overall, the iodide perovskite materials absorb in the spectral likely to be present in the samples made by solid state methods
region between 1.2 and 1.7 eV. Semiconductors with direct is probably associated with the PL.
energy gaps between 1.0 and 1.5 eV are highly sought after As stated above, there is a significant variation in the optical
since they are optimal for solar cell applications yielding the absorption properties of these materials that depends strongly
maximum theoretical conversion efficiency.44 Thus, compounds on the preparation method. Therefore, it is not surprising that
1−4 and their Cs+ analogues having a nearly ideal band gap and the emission properties differ to some extent. The absence of
favorable charge transport properties (vide infra) are very emission from the well-formed solution-grown single crystals of
promising materials for use in high-efficiency solar cells. Sn and Pb perovskites is hard to explain. A rough attempt to
(b). PL Properties. The PL properties of specimens obtained account for this is to suppose that the materials are relatively
from all four synthetic approaches (solution, grinding, open PL-inactive in their higher symmetry phase (the α-phase),
tube melting, sealed tube annealing) were measured at room where the distortions are small. If PL originates from defects it
temperature using a green light source at 532 nm and a should increase as the symmetry is lowered via twinning,
specialized PL spectrometer. An intense PL emission was artificially creating defects. The temperature dependence of the
observed for all Sn and Pb compounds. The Pb analogues emit PL intensity/wavelength has been confirmed in CsSnI3.19b
between 700 and 800 nm and the Sn analogues between 850 When one of the other three methods (grinding, open tube
and 1000 nm, covering a broad region of near-IR spectrum. melting, sealed tube annealing-see Experimental Section) is
The emission wavelength is consistent with the experimentally employed, the resulting polycrystalline materials should have a
determined band gaps providing further evidence for the direct greater number of defects. We suppose that this might be the
nature of the band gap. source of PL enhancement. Another supporting indication of
Compounds that emit in the near-IR region at room the defect-induced PL model is that the only solution prepared
temperature are rare, and they are highly sought after for material that shows some PL-activity is HC(NH2)2PbI3 which
energy conversion applications such as Luminescent Solar undergoes a phase transformation from a perovskite structure
Concentrators (LSCs).45 Currently, the field of LSCs is to a nonperovskite structure incidentally forming a highly
monopolized by the venerable class of III−V semiconductors defective perovskite phase during transformation. Therefore,
with GaAs and InP being the major compounds of interest. We the defects likely play a crucial role in the expression of the PL,
have been using the latter as a benchmark material to evaluate either when these are created during the synthesis or by means
the relative efficiency of 1−4 and to estimate their potential as of temperature-controlled structural changes. This is supported
LSC materials. by the generally observed trend which is the following: the
The solution-prepared samples showed no PL with the cruder the synthesis the higher the PL intensity.
exception of HC(NH2)2PbI3, which typically is measured as In Figure 10, the PL properties of Sn and Pb hybrid materials
mixture of the two structural forms (Supporting Information, are presented along with the corresponding Cs phases. The
Figure S9). It is interesting that the materials prepared with any magnitudes of the emission are comparable with one another,
method, other than from solution, are PL-active. Figure 10 except for CsPbI3 in which the perovskite black phase, which is
shows typical emission spectra of the hybrid perovskites stable above ∼300 °C, very rapidly converts to the PL-inactive
prepared by grinding the precursors and heating the samples yellow phase. All samples shown in Figure 10 have been
at 350 °C for 20 s. Generally, the grinding method consistently prepared from the melt for the data to be comparable. There is
L dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
a small but real shift in the emission wavelength which matches behavior. This hysteresis comes as a consequence of the
well with the corresponding shifts in absorption energy edge. structural phase transitions. In fact, the structural phase
Charge Transport Properties. Given that the PL transitions are reflected as anomalies in the resistivity close to
properties arise from defects, possibly Sn4+ centers acting as the transition temperature. Thus, compound 1 changes to the
dopants in the structure, we performed charge transport β-phase in the 250 K region followed by one more transition to
measurements on 1−3. The presence of Sn4+ could give rise to the γ-phase around 130 K. A resistivity increase around 70 K
low resistivity by creating holes based on the Sn2+/Sn4+ couple. suggests more phase transitions on lowering the temperature.
Freshly prepared samples using the solution method were Similarly, for compounds 2 and 3 the β-phase is reached at 225
employed using isopropanol graphite suspension to make the K and 320 K, respectively, while the γ-phase is observed at 110
electrical contacts. This was found to be a necessary precaution K and 180 K, respectively. The transition temperatures are in
to avoid possible redox effects resulting from direct contact of good agreement with the transition observed in X-ray
the material with a metal surface. Single-crystals prepared from diffraction studies discussed above. In the case of 3 where the
the solution method (0.1−1 mm length) and cold-pressed α/β phase transition occurs above room temperature, resistivity
pellets from the annealing method (1.67 × 1.67 × 5.00 mm3) data were also collected in the 300−400 K region using the
were investigated. In both cases, the samples were found to conventional setup (see Experimental Section), revealing a
have a significant resistivity showing a semiconducting behavior. sharp transition with an onset at 310 K.
By contrast, when the open tube method was employed, the Room Temperature and High-Temperature Resistivity.
solids obtained had a much lower resistivity behaving as p-type Cold-pressed rectangular pellets (1.67 × 1.67 × 5.00 mm3) of
“metals”. The samples were measured using both a PPMS 4- 1−4 prepared from powders of the ground samples annealed at
probe setup for low temperature data (5−330 K) and a 200 °C for 2 h in a sealed tube and/or elongated rhombic
conventional 4-probe method for room and high temperature dodecahedral single-crystal (0.5 × 0.5 × 0.5 mm3) of 1−3
measurements (300−550 K). The measured resistivity displays prepared from the solution were also measured using the
a tendency to decrease after every measurement cycle until it conventional 4-probe method (Supporting Information, Figure
reaches an equilibrium value after several cycles which remains S10). A remarkable behavior can be easily spotted differ-
constant. This behavior can be attributed to contact annealing entiating CH3NH3PbI3 from its other perovskite counterparts.
or similar effects. Such a behavior has also been observed in the The difference lies in the strongly non-ohmic characteristics of
case of CsSnI 3.12 The resistivity values reported here 3 evident in the I−V curves of the measured materials
correspond to the equilibrium resistivity values. (Supporting Information, Figure S10). The CH3NH3SnI3,
Low-Temperature Resistivity. Resistivity data were collected HC(NH2)SnI3 and HC(NH2)PbI3 compounds display a nearly
in the 5−330 K range for single-crystals of 1−3 (Figure 11) perfect ohmic behavior with resistivity values of 0.49, 11.8 and
3.4 × 103 Ω cm at 300 K following a declining trend with
increasing temperature.
CH3NH3PbI3 is singled-out since it reproducibly displayed a
non-ohmic behavior with a parabolic I−V dependence. The
calculated resistivity value for a single-crystal of CH3NH3PbI3,
ignoring the nonlinear parts of the plot, is 51 MΩ·cm at room
temperature. This resistivity value is in general agreement with
the previously reported value of 6.6 MΩ·cm obtained for a
polycrystalline sample.46 The non-ohmic behavior of 3 is
graphically highlighted in Figure 12c as the I−V curve
completes a full hysteresis loop upon reversal of the current
polarization. Interestingly, the resistivity value varies slightly
with the polarity of the applied electric field with a higher slope
being obtained for negative polarity. The equilibrium value of
the voltage for zero applied current (remnant polarization) is
roughly −10 mV, but in general displays a strong dependence
on the crystal history. These characteristics strongly imply a
ferroelectric response (the point group 4mm allows for such a
Figure 11. Single-crystal temperature-dependent resistivity plots of (a)
CH3NH3SnI3, (b) HC(NH2)2SnI3, (c) CH3NH3PbI3 in the 5−330 K behavior). The resistivity graphs of compounds 1, 2, and 4 on
temperature range (PPMS) and (d) CH3NH3PbI3 in the 300−540 K the contrary display normal ohmic behavior.
range (MMR). The specimens were obtained from the solution In terms of absolute resistivity there is a discrepancy between
method. the low- and the high-temperature values for the single-
crystalline samples. We rationalize this based on the I−V
using the PPMS instrumentation where a similar behavior to characteristics of 1−4 (Supporting Information, Figure S10). At
the one described above was observed. The initial high room temperature, 1, 2, and 4a which adopt the α-phase
resistivity values obtained at 5 K gradually decrease as the display nearly perfect ohmic behavior which results in
temperature increases and at ∼330 K are 4 Ω cm for resistivities in the 10°−101 Ω cm range for the Sn compounds
CH3NH3SnI3, 34 Ω cm for HC(NH2)2SnI3, and 596 Ω cm and 103 Ω cm for the Pb compound. On the other hand, the β-
for CH3NH3PbI3). This resistivity vs temperature trend is phase of 3 displays a remarkable non-ohmic behavior which is
characteristic of undoped semiconductors. On cooling, responsible for a hysteresis loop obtained in its I−V plot. The
resistivity increases again down to 5 K. However, the absolute hysteretic behavior can be attributed to the reorientation of the
values were observed to slightly decrease in the case of 1 and permanent dipoles of CH3NH3+ on application of external
slightly increase in the case of 2 and 3 suggesting hysteretic electric field in addition to the resistance of the inorganic
M dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Figure 13. Temperature-dependence of the Seebeck coefficients on

single-crystals of 2 (b) and 3 (d). The respective resistivity values are
given for reference in panels (a) and (c). The single-crystals were
obtained from the solution method.
cold-pressed from annealed powders obtained from the

grinding method. All samples prepared this way had nearly
identical geometric dimensions which allowed for a direct
comparison of their electron transport properties. Interestingly,
with the exception of 4a which is p-type, all samples display a n-
type behavior. The Pb-containing samples invariably show an
extremely high Seebeck coefficient in the order of few (for 5b)
to several mV/K (for 3 and 4b) indicative of very low carrier
concentration. The polycrystalline pellets of the Sn-based
compounds (including a sample of similarly prepared CsSnI3)
Figure 12. Hysteresis loops for a sample of compound 3. (a) Voltage on the other hand show significantly lower Seebeck coefficients
and (b) temperature dependence of the coercive field and (c) in the order of few hundreds of μV/K indicating a higher carrier
polarization as a function of applied bias. The measured specimen was concentration (still low as an absolute value) with respect to
prepared via the solution method. the Pb analogues (Figure 14). The samples still behave as
{PbI3}− lattice. This is clearly a ferroelectric response. Thus, a

capacitance/inductance factor is introduced to the charge
transport which reflects the ability of the permanent dipoles to
align with the applied field while confined within the perovskite
“cage”. Field- and temperature-dependence of the coercive field
in the I−V hysteresis loops as well as polarization measure-
ments also suggest a motion of charge within the material
(Figure 12).
Thermopower. To assign the carrier type in the perovskite
compounds, we performed Seebeck coefficient measurements
on both solution derived single-crystals of 1−3 and cold-
pressed pellets of 1−4 fabricated from powders of the ground
samples annealed at 200 °C for 2 h (sealed tube). The
compound HC(NH2)PbI3 could only be measured in the latter
form, since the pressure applied to form the pellets (∼10 GPa)
was sufficient to stabilize the perovskite (black) form for several
hours. The thermopower of the single crystals for compounds 2
and 3 in the 300−400 K range, (Figure 13) are negative and
very large in magnitude indicating that the materials are n-type
semiconductors with very low levels of electron carriers. The
Seebeck coefficients were −2700 and −6500 μV/K at 380 K for
Figure 14. Temperature-dependence (a) resistivity and (b) Seebeck
2 and 3, respectively. These values are consistent with carrier coefficients on cold-pressed pellets of annealed polycrystalline samples
concentration of <1016 cm−3. For compound 3, the Seebeck of the compounds 1−4, 5b, and a CsSnI3 reference. The pellets were
value was sensitive to the phase transition, consistent with the annealed for 2 h at 200 °C under vacuum prior to the measurement.
overall change in the other transport properties. Negative Seebeck coefficients were converted to positive to fit in the
To gain further insight into the transport characteristics, a logarithmic scale. N-type compounds are named as such in the figure
complete collection of 1−4 and 5b (vide infra) was prepared as labels.
N dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX

Figure 15. Carrier-type transformations and PL-emission relationships in ASnI3 (A = Cs+, CH3NH3+ or HC(NH2)2+) effected by the different
synthetic methods employed.
semiconductors as evidenced by resistivity measurements. heating/cooling cycles during the measurements which
When the pressed pellets are made from samples that are not significantly reduces Sn4+ centers as explained above.
annealed, the Sn compounds are p-type semiconductors. A Combining resistivity and Hall effect data (measured for the
typical example of this p-type behavior is observed in CsSnI3.12 same specimens), the electron mobility was calculated to be
We have confirmed this p-type behavior on samples of 1 ∼2320 cm2/V·s for CH3NH3SnI3, ∼536 cm2/V·s for CsSnI3,
prepared from the open tube method (Supporting Information, ∼103 cm 2 /V·s for HC(NH 2) 2 SnI3 , ∼270 cm 2/V·s for
Figure S11) which show a similar resistivity (24 Ω cm) and a CH3NH3Sn0.5Pb0.5I3, and ∼66 cm2/V·s for CH3NH3PbI3
small and positive value of the Seebeck coefficient (+82 μV/K). using the expression μ = 1/neρ. These are remarkably high
These values, although they clearly indicate the change of the values and are comparable to those of classical semiconductors
carrier type on doping are still much higher than those including Si, GaAs, InP, and so forth. Such values for carrier
previously reported for CH3NH3SnI32i,8b suggesting a lower mobility are also consistent with the previously reported values
level of doping. for CsSnI3.12 The hole carrier mobility for the p-type sample of
Hall-Effect Measurements. Hall effect measurements also CH3NH3SnI3 prepared from the open tube method was found
performed on cold-pressed pellets of annealed polycrystalline to be ∼322 cm2/V·s (carrier concentration ∼1015), significantly
1−3, 5b, and a reference sample of CsSnI3 (prepared with the lower than the electron mobility, but still comparable to the
same method) to estimate the carrier concentration of the value obtained for p-type CsSnI3, ∼520 cm2/V·s.12
compounds. The specimens were measured under a variable Our present experimental charge transport results are quite
magnetic field of 0.5−1.25 T under a constant excitation of 10 different from those reported previously. For example, the
μA for the less resistive samples (1, 2, 5b) and 100 nA for resistivity for CH3NH3SnI3 is higher by at least one2i or two8b
compound 3 (Supporting Information, Figure S13). The Hall orders of magnitude from previously reported samples (for
coefficients were negative and confirm the negative sign of the both p- and n-type specimens), and this is attributed to the
carriers (n-type) for all the thus prepared compounds. The differences in the synthetic approach, as discussed above. More
carrier concentrations were determined based on the formula: n importantly, we find that the overall assessment of the Sn
= −BzIx/dVHe, where Bz is the magnetic field, Ix is the electric compounds as p-type “metals” needs to be revised since our
current, d is the sample thickness, e is the charge of electron, results clearly indicate that CH3NH3SnI3 and HC(NH2)2SnI3
and VH is the Hall voltage. They were in the order of 1014 cm−3 can behave as n-type semiconductors. The reason for the ability
for 1, 2, 5b and a similarly prepared sample of CsSnI3 while the of these compounds to be p-type conductors (e.g., as in the
respective carrier concentration for compound 3 was estimated case of CsSnI3) is the unique property of the materials to
at ∼109 cm−3. The data are in excellent agreement with the very accommodate Sn4+ centers and thus to transition to a doped
large Seebeck coefficients (Figure 14b). The very low carrier state with charge carriers being holes. In this work we were able
concentrations correspond to nearly intrinsic semiconducting to achieve samples with a very low carrier concentrations and
behavior. This finding is further corroborated by the fact that defects because of the presence of a powerful reducing agent
the specimens, except for p-type compound 4a, do not display (i.e., H3PO2) which inhibits/reverses the self-doping process
any PL signal after the measurements indicating depletion of (i.e., suppress formation of Sn4+) and allows the isolation of the
the defects, which also give rise to the charge carriers. This is pure stoichiometric compounds. P-type conducting behavior is
consistent with the proposal that the PL arises from Sn4+ observed when the preparation of the Sn compounds is carried
centers present in dopant concentrations in the PL-active out through solid-state methods. The absence of the reducing
samples. The disappearance of PL is probably caused by the agent renders the Sn2+/Sn4+ conversion irreversible and
thermal treatment during sample preparation and several therefore the obtained solids are inherently doped. The fact
O dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Figure 16. X-ray powder diffraction patterns of CH3NH3Sn1−xPbxI3 solid solution as obtained from (a) grinding, (b), open tube melt, (c) sealed tube
annealing, and (d) solution reactions. In the cases where there is ambiguity on the Sn:Pb ratio the values were obtained from EDS analysis.
that Sn oxidation is responsible for the “metal-like” behavior in material, which mask the properties of the inorganic [PbI3]−
the solid state is highlighted by a direct comparison with the framework.
Pb-analogues (which cannot undergo analogous Pb2+/Pb4+ CH3NH3Sn1−xPbxI3 Solid Solution Studies. In the context of
doping because of the instability of Pb4+ in an iodide the above results, we decided to study the properties of
coordination environment). In the latter system, the charge CH3NH3Sn1−xPbxI3 solid solutions to explore the structure and
transport properties are practically unaffected by the prepara- property trends as a function of x. As shown above, these two
tive route. phases follow the same phase transition path, albeit the
Interestingly, it is still possible to “un-dope” the semi- transitions occur at different temperatures. The materials were
conductors to a nearly carrier-depleted n-type state. This can be studied as a function of the preparation route as described
achieved with extensive thermal treatment at ∼200 °C, at which above. Powder X-ray diffraction patterns are presented in
temperature Sn4+ dopants are eliminated in the form of volatile Figure 16.
SnI4. Excess of the organic cations can also be partially removed The X-ray powder diffraction patterns show that the ground
at the annealing temperature in the form of the free amine and samples contain impurities. The other three preparative
HI gas. Therefore the doping levels within the Sn-based methods (open tube melting, sealed tube annealing, solution)
materials are controllable simply by choosing the appropriate provide sufficiently pure materials. The weak Bragg reflections
synthetic method to generate p- or n-type semiconductors in observed near the triplet of major peaks at 24.7°, 28.6°, and
the ASnI3 family of compounds. A schematic representation of 32.1° 2θ angles have been discussed earlier in the context of
Figure 2b as an oxidation impurity. For materials prepared from
the carrier type transformations is given in Figure 15.
the melt the three peaks appear already at the onset of data
In addition, the above results illustrate that the phase
collection. For samples obtained from solution synthesis and
transitions determine the transport characteristics of the
the annealed samples the minor reflections appear in the
semiconducting perovskite compounds. In their highest
powder diffraction patterns only after prolonged exposure in air
symmetry (α-phase), the perovskites display the maximum during data collection. The cleanest pattern was obtained from
conductivity, which is a consequence of the maximum orbital the annealed samples because of strong diffraction requiring a
overlap between the metal and the iodine (linear Sn−I−Sn considerably shorter collection time, so that substantial
bond). As the phase transitions progress at lower temperatures oxidation does not occur. Therefore, it is clear that these
(β- and γ-phases) the change in the M−I−M angles result in reflections are due to oxidation of Sn in the compounds since
reduced orbital overlap, thus hindering the charge transport they do not appear in the diffraction patterns of corresponding
through the inorganic framework. This trend is nicely pure Pb and Pb-rich phases.
illustrated by the symmetry lowering which is CH3NH3SnI3 The lattice parameters calculated from the diffraction
(C4v) > CsSnI3 (D2h) > HC(NH2)2SnI3 (C2v) and HC- patterns obtained from the annealed samples (Figure 15c),
(NH2)2PbI3 (C3v) > CH3NH3PbI3 (C4v) > CsPbI3 (C2h) for the are given in the Supporting Information, Table S25. The
Sn-series and Pb-series, respectively (Schönflies symmetry is evolution of the structure in this solid solution system is
given in parentheses). This trend is also reflected in the band consistent, irrespective of the method used. The Sn-rich
gap evolution throughout the series. The transport properties compositions adopt the pseudocubic tetragonal P4mm structure
of β-CH3NH3PbI3, in particular, where a capacitor-like behavior of the parent α-CH3NH3SnI3 compound up to x = 0.5. For x >
was observed, may be rationalized by the reorientation of the 0.5 the structure changes to that of the tetragonal β-
CH3NH3+ dipoles on application of current through the CH3NH3PbI3, I4cm. It must be noted here that the volume
P dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Figure 17. PL properties of CH3NH3Sn1−xPbxI3 solid solution as obtained from (a) mildly grinding the precursors (interface reaction) and (b) by
annealing the same specimen at 200 °C for 2 h in a sealed tube. The data are normalized toward single crystalline InP.
Figure 18. Optical absorption properties of CH3NH3Sn1−xPbxI3 solid solution as obtained from (a) grinding, (b), open tube melt, (c) sealed tube
annealing, and (d) solution reactions. In the cases where there is ambiguity on the Sn:Pb ratio the values were obtained from EDS analysis. The value
of the band gap is given next to each composition.
of the unit cell in the series changes only slightly, which is (Figure 10b). For x > 0.25 the shift is smaller as is the
consistent with previous observations that the cell dimensions intensity of the peak. Generally, the intensity seems to slightly
are dictated by the nature of the halide ions.47 decrease by annealing.
The PL properties of CH3NH3Sn1−xPbxI3 are strongly The red shift in the emission wavelength is reflected in the
affected by mixing the Sn and Pb sites (Figure 17). Again, electronic absorption UV−vis spectra which universally show a
materials prepared from solution were not luminescent. The shift of the band gap to lower energies (Figure 18). The largest
mixed-metal perovskites prepared from the melt were also band gap shift occurs in the materials obtained from solution,
inactive, in contrast to the pure phases prepared by the same on the Pb-rich side of the compositions reaching values of 1.1
method for x = 0 and x = 1. Intense PL was produced only eV (Figure 19). However, no emission could be detected in
when the materials were prepared by mild grinding so that these specimens. The annealed materials show an excellent
some unreacted solid remains (interface reaction). The PL agreement between absorption energy and emission wavelength
intensity decreases as the mixture is thoroughly ground to a and may indeed prove valuable in tuning the emission
visually homogeneous black powder and the reaction is almost properties of the iodide perovskites.
complete. The emission observed from this process corre-
sponds to the individual Sn and Pb centers showing two peaks
at the respective emission wavelengths (Figure 10a). However,
■ CONCLUDING REMARKS
A broad family of hybrid Sn and Pb iodide compounds with
when the material is placed in a sealed tube and annealed at 200 semiconducting properties bearing methylammonium or
°C, the formation of the solid solution is complete, and the Sn formamidinium cations can be universally stabilized in the 3D
and Pb active centers are fully mixed. As a result, the initial perovskite structure type, AMX3. The preparative route chosen
emission peaks disappeared, and a new red-shifted peak for the materials is of utmost importance as it governs their
appeared for x = 0.25 at 1030 nm. This is an overall red shift respective physical properties. The structural properties of these
of 70 nm with respect to the end member CH3NH3SnI3 phases strongly resemble the behavior of the perovskitic oxides
Q dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Figure 19. (a) Normalized lattice constants and (b) optical properties of CH3NH3Sn1−xPbxI3 as a function of composition from the samples
obtained through sealed tube annealing of ground samples showing the evolution of PL emission wavelength and the energy band gap. The
normalization factor is √2/2 and 1/2 for a- and c-axis, respectively.
by undergoing reversible, temperature-dependent phase tran- the lower is the ability to exhibit PL. Our findings open a path
sitions. The electrical properties can be controlled to a large for deeper understanding and further development of these
extent by selecting the preparation method or the thermal materials as components in various applications. The ability to
treatment. The entire family featured direct energy gaps be a good electrical conductor and strong emitter at room
spanning the range between 1.1 to 1.7 eV. This makes them temperature is a remarkable characteristic which places these
highly relevant as possible light absorbers in energy conversion materials in an elite class of solution processable semi-
or detector devices. The Pb and Sn containing members of the conductors. Although the initial studies of the materials date
family display strong PL at room temperature which is superior back to the 1970s, their complex behavior, exotic properties,
to classical semiconducting materials such as InP, but in and chemical and physical subtleties are only now beginning to
addition, the emission wavelength can be shifted through subtle
be understood. The interplay between p- and n-type semi-
changes in their compositions. Transport properties confirm
conducting behavior due to the presence or elimination of the
the semiconducting nature of the compounds showing
Sn4+ dopant is a characteristic example of this complexity and
additionally strong dependence upon structural changes. In
the case of β-CH3NH3PbI3, a capacitor-like behavior was versatility in the chemistry and physics of the hybrid metal
observed at room temperature which suggests that the other Sn iodide perovskites. The suitable energy band gaps and
and Pb analogues might show similar behavior upon impressively large carrier mobilities reported here and in
transitioning into the β-phase. Overall, we consider the hybrid previous publications on CsSnI312,15 and CH3NH3PbI315a,b,d,e
iodide perovskites as flexible multipurpose functional materials indicate that these materials are very promising for modern
emitting in the near-IR spectral region at room temperature. energy-related applications, such as solar energy conversion, as
They are easily accessible and scalable and display remarkable we have shown for example in the case of CsSnI3.15c Future
charge transport, optical absorption, and room temperature PL work should be focused on optimization of the PL intensity and
properties. The lower the number of defects in the structure, fabrication of thin films for use in devices.
R dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX
Inorganic Chemistry
■
Article
ASSOCIATED CONTENT H.-H.; Chen, Z.-R.; Cheng, L.-C.; Wang, Y.-J.; Feng, M.; Wang, M.
Dalton Trans. 2010, 39, 11000.
*
S Supporting Information
(7) (a) Papavassiliou, G. C. Prog. Solid State Chem. 1997, 25, 125.
Atomic coordinates and anisotropic thermal factors for (b) Mitzi, D. B. In Progress in Inorganic Chemistry; John Wiley & Sons,
compounds 1−7, as well as supplementary X-ray diffraction, Inc.: New York, 2007; p 1; (c) Billing, D. G.; Lemmerer, A.
thermal, and spectroscopic data. This material is available free CrystEngComm 2009, 11, 1549. (d) Mercier, N.; Louvain, N.; Bi, W.
of charge via the Internet at http://pubs.acs.org.
■
CrystEngComm 2009, 11, 720. (e) Wu, L.-M.; Wu, X.-T.; Chen, L.
Coord. Chem. Rev. 2009, 253, 2787.
AUTHOR INFORMATION (8) (a) Mitzi, D. B.; Feild, C. A.; Harrison, W. T. A.; Guloy, A. M.
Corresponding Author Nature 1994, 369, 467. (b) Mitzi, D. B.; Feild, C. A.; Schlesinger, Z.;
Laibowitz, R. B. J. Solid State Chem. 1995, 114, 159.
*E-mail: [email protected].
(9) (a) Mitzi, D. B.; Wang, S.; Feild, C. A.; Chess, C. A.; Guloy, A. M.
Notes Science 1995, 267, 1473. (b) Mitzi, D. B.; Liang, K. J. Solid State Chem.
The authors declare no competing financial interest. 1997, 134, 376.
■ ACKNOWLEDGMENTS
We thank OmniPV, Inc. for providing us with the
(10) Calabrese, J.; Jones, N. L.; Harlow, R. L.; Herron, N.; Thorn, D.
L.; Wang, Y. J. Am. Chem. Soc. 1991, 113, 2328.
(11) Mizusaki, J.; Arai, K.; Fueki, K. Solid State Ionics 1983, 11, 203.
(12) Chung, I.; Song, J.-H.; Im, J.; Androulakis, J.; Malliakas, C. D.;
instrumentation for the measurement of the PL properties. Li, H.; Freeman, A. J.; Kenney, J. T.; Kanatzidis, M. G. J. Am. Chem.
This work was supported in part from the ANSER Center, an Soc. 2012, 134 (20), 8579.
Energy Frontier Research Center funded by the U.S. (13) Chondroudis, K.; Mitzi, D. B. Chem. Mater. 1999, 11, 3028.
Department of Energy, Office of Science, Office of Basic (14) Tanaka, K.; Takahashi, T.; Ban, T.; Kondo, T.; Uchida, K.;
Energy Sciences, under Award No. DE-SC0001059. Miura, N. Solid State Commun. 2003, 127, 619.
■ REFERENCES
(1) Kay, H. F.; Bailey, P. C. Acta Crystallogr. 1957, 10, 219.
(15) (a) Kojima, A.; Teshima, K.; Shirai, Y.; Miyasaka, T. J. Am.
Chem. Soc. 2009, 131, 6050. (b) Im, J.-H.; Lee, C.-R.; Lee, J.-W.; Park,
S.-W.; Park, N.-G. Nanoscale 2011, 3, 4088. (c) Chung, I.; Lee, B.; He,
J.; Chang, R. P. H.; Kanatzidis, M. G. Nature 2012, 485, 486. (d) Etgar,
(2) (a) Moller, C. K. Nature 1957, 180, 981. (b) Moller, C. K. Nature
1958, 182, 1436. (c) Moller, C. K. Mat. Fys. Medd. Dan. Vid. Selsk. L.; Gao, P.; Xue, Z.; Peng, Q.; Chandiran, A. K.; Liu, B.; Nazeeruddin,
1959, 32. (d) Weber, D. Z. Naturforsch. 1978, 33b, 862. (e) Weber, D. M. K.; Grätzel, M. J. Am. Chem. Soc. 2012, 134, 17396. (e) Lee, M. M.;
Z. Naturforsch. 1978, 33b, 1443. (f) Yamada, K.; Tsuritani, T.; Okuda, Teuscher, J.; Miyasaka, T.; Murakami, T. N.; Snaith, H. J. Science 2012,
T.; Ichiba, S. Chem. Lett. 1989, 18, 1325. (g) Yamada, K.; Funabiki, S.; 338, 643.
Horimoto, H.; Matsui, T.; Okuda, T.; Ichiba, S. Chem. Lett. 1991, 20, (16) Hang, T.; Zhang, W.; Ye, H.-Y.; Xiong, R.-G. Chem. Soc. Rev.
801. (h) Trots, D. M.; Myagkota, S. V. J. Phys. Chem. Solids 2008, 69, 2011, 40, 3577.
2520. (i) Takahashi, Y.; Obara, R.; Lin, Z.-Z.; Takahashi, Y.; Naito, T.; (17) Mitzi, D. B. J. Mater. Chem. 2004, 14, 2355.
Inabe, T.; Ishibashi, S.; Terakura, K. Dalton Trans. 2011, 40, 5563. (18) Scaife, D. E.; Weller, P. F.; Fisher, W. G. J. Solid State Chem.
(j) Yamada, K.; Nakada, K.; Takeuchi, Y.; Nawa, K.; Yamane, Y. Bull. 1974, 9, 308.
Chem. Soc. Jpn. 2011, 84, 926. (19) (a) Shum, K.; Chen, Z.; Qureshi, J.; Yu, C.; Wang, J. J.;
(3) (a) Papavassiliou, G. C.; Patsis, A. P.; Lagouvardos, D. J.; Pfenninger, W.; Vockic, N.; Midgley, J.; Kenney, J. T. Appl. Phys. Lett.
Koutselas, I. B. Synth. Met. 1993, 57, 3889. (b) Papavassiliou, G. C.; 2010, 96, 221903. (b) Yu, C.; Chen, Z.; Wang, J. J.; Pfenninger, W.;
Koutselas, I. B. Synth. Met. 1995, 71, 1713. (c) Mitzi, D. B. Chem. Vockic, N.; Kenney, J. T.; Shum, K. J. Appl. Phys. 2011, 110, 063526.
Mater. 1996, 8, 791. (d) Guan, J.; Tang, Z.; M. Guloy, A. Chem. (c) Chen, Z.; Yu, C.; Shum, K.; Wang, J. J.; Pfenninger, W.; Vockic,
Commun. 1999, 1833. (e) Kagan, C. R.; Mitzi, D. B.; Dimitrakopoulos, N.; Midgley, J.; Kenney, J. T. J. Lumin. 2012, 132, 345.
C. D. Science 1999, 286, 945. (f) Mitzi, D. B.; Dimitrakopoulos, C. D.; (20) (a) Poglitsch, A.; Weber, D. J. Chem. Phys. 1987, 87, 6373.
Kosbar, L. L. Chem. Mater. 2001, 13, 3728. (g) Tang, Z.; Guan, J.; (b) Xu, Q. E., T.; Nakayama, H.; Nakamura, N.; Kishita, M. Z.
Guloy, A. M. J. Mater. Chem. 2001, 11, 479. (h) Knutson, J. L.; Martin, Naturforsch. 1991, 46a, 240. (c) Onoda-Yamamuro, N.; Yamamuro,
J. D.; Mitzi, D. B. Inorg. Chem. 2005, 44, 4699. (i) Sakai, K.-i.; O.; Matsuo, T.; Suga, H. J. Phys. Chem. Solids 1992, 53, 277.
Sonoyama, T.; Tsuzuki, T.; Ichikawa, M.; Taniguchi, Y. Chem. Lett. (21) Howie, R. A.; Moser, W.; Trevena, I. C. Acta Crystallogr., Sect. B
2005, 34, 212. (j) Sourisseau, S.; Louvain, N.; Bi, W.; Mercier, N.; 1972, 28, 2965.
Rondeau, D.; Boucher, F.; Buzaré, J.-Y.; Legein, C. Chem. Mater. 2007, (22) (a) Kortüm, G.; Braun, W.; Herzog, G. Angew. Chem. 1963, 75,
19, 600. (k) Takahashi, Y.; Obara, R.; Nakagawa, K.; Nakano, M.; 653. (b) McCarthy, T. J.; Tanzer, T. A.; Kanatzidis, M. G. J. Am. Chem.
Tokita, J.-y.; Inabe, T. Chem. Mater. 2007, 19, 6312. Soc. 1995, 117, 1294. (c) Larson, P.; Mahanti, S. D.; Kanatzidis, M. G.
(4) (a) Wang, S.; Mitzi, D. B.; Feild, C. A.; Guloy, A. J. Am. Chem. Phys. Rev. B 2000, 61, 8162. (d) Trikalitis, P. N.; Rangan, K. K.; Bakas,
Soc. 1995, 117, 5297. (b) Koutselas, I. B.; Mitzi, D. B.; Papavassiliou, T.; Kanatzidis, M. G. J. Am. Chem. Soc. 2002, 124, 12255.
G. C.; Papaioannou, G. J.; Krautscheid, H. Synth. Met. 1997, 86, 2171. (23) Pfenninger, W. OmniPV_PL; OmniPV Inc, 2008
(c) Mitzi, D. B.; Liang, K.; Wang, S. Inorg. Chem. 1998, 37, 321. (24) Sheldrick, G. Acta Crystallogr., Sect. A 2008, 64, 112.
(5) (a) Vincent, B. R.; Robertson, K. N.; Cameron, T. S.; Knop, O. (25) Spek, A. J. Appl. Crystallogr. 2003, 36, 7.
Can. J. Chem. 1987, 65, 1042. (b) Tudela, D.; J. Sanchez-Herencia, A.; (26) Farrugia, L. J. Appl. Crystallogr. 1999, 32, 837.
Diaz, M.; Fernandez-Ruiz, R.; Menendez, N.; Tornero, J. D. J. Chem. (27) Petricek, V.; Dusek, M.; Palatinus, L. Jana2006, The crystallo-
Soc., Dalton Trans. 1999, 4019. graphic computing system; Institute of Physics: Praha, Czech Republic,
(6) (a) Tang, Z.; Guloy, A. M. J. Am. Chem. Soc. 1998, 121, 452. 2006.
(b) Krautscheid, H.; Vielsack, F. J. Chem. Soc., Dalton Trans. 1999, (28) Stokes, H. T.; Hatch, D. M.; Campbell, B. J. ISOTROPY;
2731. (c) Lode, C.; Krautscheid, H. Z. Anorg. Allg. Chem. 2000, 626, Department of Physics and Astronomy, Brigham Young University:
326. (d) Guloy, A. M.; Tang, Z.; Miranda, P. B.; Srdanov, V. I. Adv. Provo, Utah, 2007; stokes.byu.edu/isotropy.html.
Mater. 2001, 13, 833. (e) Lode, C.; Krautscheid, H. Z. Anorg. Allg. (29) Schaefer, F. C.; Hechenbleikner, I.; Peters, G. A.; Wystrach, V.
Chem. 2001, 627, 841. (f) Lode, C.; Krautscheid, H. Z. Anorg. Allg. P. J. Am. Chem. Soc. 1959, 81, 1466.
Chem. 2001, 627, 1454. (g) Lode, C.; Krautscheid, H. Z. Anorg. Allg. (30) Bedlivy, D.; Mereiter, K. Acta Crystallogr., Sect. B 1980, 36, 782.
Chem. 2007, 633, 1691. (h) Wolff, M.; Harmening, T.; Pöttgen, R.; (31) Fan, L.-Q.; Wu, J.-H. Acta Crystallogr., Sect. E 2007, 63, i189.
Feldmann, C. Inorg. Chem. 2009, 48, 3153. (i) Zhao, H.-R.; Li, D.-P.; (32) (a) Howard, C. J.; Stokes, H. T. Acta Crystallogr., Sect. B 1998,
Ren, X.-M.; Song, Y.; Jin, W.-Q. J. Am. Chem. Soc. 2009, 132, 18. (j) Li, 54, 782. (b) Howard, C. J.; Stokes, H. T. Acta Crystallogr., Sect. B 2004,
S dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX

60, 674. (c) Howard, C. J.; Stokes, H. T. Acta Crystallogr., Sect. A 2005,
61, 93.
(33) (a) Nord, G. L. Phase Transitions 1994, 48, 107. (b) White, T. J.;
Segall, R. L.; Barry, J. C.; Hutchison, J. L. Acta Crystallogr., Sect. B
1985, 41, 93.
(34) Glazer, A. Acta Crystallogr., Sect. B 1972, 28, 3384.
(35) Swainson, I.; Chi, L.; Her, J.-H.; Cranswick, L.; Stephens, P.;
Winkler, B.; Wilson, D. J.; Milman, V. Acta Crystallogr., Sect. B 2010,
66, 422.
(36) Mauersberger, P.; Huber, F. Acta Crystallogr., Sect. B 1980, 36,
683.
(37) Landau, L. D., Lifshitz, E. M. Statistical Physics, Part 1, 3rd ed.;
Elsevier: Amsterdam, The Netherlands, 1980.
(38) Gesi, K. Ferroelectrics 1997, 203, 249.
(39) Wasylishen, R. E.; Knop, O.; Macdonald, J. B. Solid State
Commun. 1985, 56, 581.
(40) (a) Mashiyama, H.; Kurihara, Y.; Azetsu, T. J. Korean Phys. Soc.
1998, 32, S156. (b) Kawamura, Y.; Mashiyama, H.; Hasebe, K. J. Phys.
Soc. Jpn. 2002, 71, 1694. (c) Mashiyama, H.; Kawamura, Y.; Kasano,
H.; Asahi, T.; Noda, Y.; Kimura, H. Ferroelectrics 2007, 348, 182.
(41) Maaej, A.; Bahri, M.; Abid, Y.; Jaidane, N.; Lakhdar, Z. B.;
Lautié, A. Phase Transitions 1998, 64, 179.
(42) Baikie, T.; Fang, Y.; Kadro, J. M.; Schreyer, M.; Wei, F.;
Mhaisalkar, S. G.; Graetzel, M.; White, T. J. J. Mater. Chem. 2013, 1,
5628.
(43) (a) Borriello, I.; Cantele, G.; Ninno, D. Phys. Rev. B 2008, 77,
235214. (b) Chang, H., Y.; Park, H., C.; Matsuishi, K. J. Korean Chem.
Soc. 2004, 44, 5.
(44) Shockley, W.; Queisser, H. J. J. Appl. Phys. 1961, 32, 510.
(45) (a) Rau, U.; Glaeser, G. C. In Photovoltaic Energy Conversion,
Conference Record of the 2006 IEEE 4th World Conference, 2006; Vol. 1,
p 205; (b) Slooff, L. H.; Bende, E. E.; Burgers, A. R.; Budel, T.;
Pravettoni, M.; Kenny, R. P.; Dunlop, E. D.; Büchtemann, A. Phys.
Status. Solidi RRL 2008, 2, 257.
(46) Knop, O.; Wasylishen, R. E.; White, M. A.; Cameron, T. S.;
Oort, M. J. M. V. Can. J. Chem. 1990, 68, 412.
(47) Clark, S. J.; Donaldson, J. D.; Harvey, J. A. Chem. Mater. 1995, 5,
1813.
T dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX

Ic 401215 X

Uploaded by

Copyright:

Available Formats

Ic 401215 X

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ic 401215 X

Uploaded by

Copyright:

Available Formats

Article

Semiconducting Tin and Lead Iodide Perovskites with Organic

ABSTRACT: A broad organic−inorganic series of hybrid

Table 1. Crystallographic Data of the α-Phases for 1−4a

Table 2. Crystallographic Data of the β-phases for 1−4a

J dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX

Optical Properties. (a). Electronic Spectroscopy. All

Figure 8. (a) TGA analysis of CH3NH3SnI3 as a function of the

this could be explained by the inability of Pb2+ to oxidize to

Figure 13. Temperature-dependence of the Seebeck coeﬃcients on

cold-pressed from annealed powders obtained from the

{PbI3}− lattice. This is clearly a ferroelectric response. Thus, a

N dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX

S dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX

T dx.doi.org/10.1021/ic401215x | Inorg. Chem. XXXX, XXX, XXX−XXX

You might also like