Accelerat ing t he world's research.

All-solid-state dye-sensitized solar

cells with high efficiency
Byunghong Lee

Nature

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

CsSnI 3 : Semiconduct or or Met al? High Elect rical Conduct ivit y and St rong Near-Infrared Phot …
Christ os D Malliakas, Hao Li

Solid St at e ChemiSt ry and it S appliCat ion 2014 Ant hony R. West

Do Quang T hang

Energy barrier at t he N719-dye/CsSnI 3 int erface for phot ogenerat ed holes in dye-sensit ized solar cells
Kai Shum
 Article

pubs.acs.org/JACS

CsSnI3: Semiconductor or Metal? High Electrical Conductivity and

Strong Near-Infrared Photoluminescence from a Single Material.
High Hole Mobility and Phase-Transitions
In Chung,† Jung-Hwan Song,‡,⊥ Jino Im,‡ John Androulakis,† Christos D. Malliakas,† Hao Li,†
Arthur J. Freeman,‡ John T. Kenney,§ and Mercouri G. Kanatzidis*,†
†
Department of Chemistry, Northwestern University, Evanston, Illinois 60208, United States
‡
Department of Physics and Astronomy, Northwestern University, Evanston, Illinois 60208, United States
§
OmniPV Inc., 1030 Hamilton Court, Menlo Park, California 94025, United States
*
S Supporting Information

ABSTRACT: CsSnI3 is an unusual perovskite that undergoes

complex displacive and reconstructive phase transitions and
exhibits near-infrared emission at room temperature. Exper-
imental and theoretical studies of CsSnI3 have been limited by
the lack of detailed crystal structure characterization and
chemical instability. Here we describe the synthesis of pure
polymorphic crystals, the preparation of large crack-/bubble-
free ingots, the refined single-crystal structures, and temper-
ature-dependent charge transport and optical properties of
CsSnI3, coupled with ab initio first-principles density functional
theory (DFT) calculations. In situ temperature-dependent single-crystal and synchrotron powder X-ray diffraction studies reveal
the origin of polymorphous phase transitions of CsSnI3. The black orthorhombic form of CsSnI3 demonstrates one of the largest
volumetric thermal expansion coefficients for inorganic solids. Electrical conductivity, Hall effect, and thermopower
measurements on it show p-type metallic behavior with low carrier density, despite the optical band gap of 1.3 eV. Hall
effect measurements of the black orthorhombic perovskite phase of CsSnI3 indicate that it is a p-type direct band gap
semiconductor with carrier concentration at room temperature of ∼ 1017 cm−3 and a hole mobility of ∼585 cm2 V−1 s−1. The
hole mobility is one of the highest observed among p-type semiconductors with comparable band gaps. Its powders exhibit a
strong room-temperature near-IR emission spectrum at 950 nm. Remarkably, the values of the electrical conductivity and
photoluminescence intensity increase with heat treatment. The DFT calculations show that the screened-exchange local density
approximation-derived band gap agrees well with the experimentally measured band gap. Calculations of the formation energy of
defects strongly suggest that the electrical and light emission properties possibly result from Sn defects in the crystal structure,
which arise intrinsically. Thus, although stoichiometric CsSnI3 is a semiconductor, the material is prone to intrinsic defects
associated with Sn vacancies. This creates highly mobile holes which cause the materials to appear metallic.

■ INTRODUCTION combination of thermal and e-beam evaporation methods,

CsSnI3 is a unique phase-change material that exhibits four show a direct energy gap at 1.3 eV and an emission at 950 nm.4
polymorphs.1,2 Two polymorphs exist independently at room Organic analogues of CsSnI3 have been extensively studied.
temperature: one has a one-dimensional double-chain structure For example, a homologous series of
and is yellow in color (Y), and another has a three-dimensional (C4H9NH3)2(CH3NH3)n−1SnnI3n+1 compounds undergo a
perovskite structure and is black in color (B-γ). When heated semiconducting (n < 3)-to-metallic (n > 5) transition with
above 425K, the Y phase transforms to a black cubic perovskite increasing n.5 Its end member (n = ∞), cubic perovskite
phase (B-α), which, on subsequent cooling, converts to a black CH3NH3SnI3 compound is claimed to be a low-carrier-density
tetragonal phase (B-β) at 426 K (instead of Y) and a black p-type metal.5 The isostructural lead halide compounds
orthorhombic phase (B-γ) at 351 K. This behavior was studied CH3NH3PbX3 (X = Br, I) efficiently sensitize TiO2 for
by X-ray powder diffraction (XRD) and differential thermal visible-light conversion to realize high voltages close to 1.0
analysis (DTA).1 The CsSnI3 system is unique in combining V.6 Theoretical calculations on ASnX3 (A = Cs, CH3NH3,
two generally contra-indicated properties, strong photolumi- NH2CHNH2; X = Cl, I) suggested that their electronic
nescence (PL) and high electrical conductivity. Electrical properties strongly depend on the structure of the inorganic
conductivity of the black CsSnI3 phase was measured and
shown to be inversely proportional to temperature, indicating Received: February 16, 2012
metallic behavior.3 Thin films of CsSnI3, deposited by a Published: May 11, 2012

© 2012 American Chemical Society 8579 dx.doi.org/10.1021/ja301539s | J. Am. Chem. Soc. 2012, 134, 8579−8587
Journal of the American Chemical Society Article

Table 1. Crystallographic Refinement Details for four CsSnI3 Polymorphs

CsSnI3
code B-α B-β B-γ Y
color black black black yellow
temp/K 500 380 300 300
crystal system cubic tetragonal orthorhombic orthorhombic
space group Pm3̅m P4/mbm Pnma Pnma
unit cell a = 6.2057(7) a = 8.7182(8) a = 8.6885(5) a = 10.350(1)
dimensions/Å b = 8.7182(8) b = 12.3775(4) b = 4.7632(5)
c = 6.1908(6) c = 8.6384(6) c = 17.684(2)
Z 1 2 4 4
V/Z, Å3 238.99(5) 235.27(2) 232.25(2) 217.94(4)
d (calcd)/g cm−3 4.393 4.463 4.521 4.818
crystal dimensions/mm3 0.076 × 0.075 × 0.020 0.027 × 0.032 × 0.243
λ/Å 0.71073
μ/mm−1 16.018 16.271 16.482 17.565
F(000) 264 528 1056 1056
θmax, deg 27.5 27.46 26.99 29.15
total/unique reflections 2004/81 3641/320 6924/1059 7977/1311
no. parameters 6 13 49 32
refinement method full-matrix least-squares on F2
final R indices [I > 2σ(I)], R1a/wR2b 0.0289/0.0690 0.0656/0.01520 0.0431/0.0954 0.0288/0.0536
R indices (all data), R1/wR2 0.0345/0.0707 0.0822/0.1599 0.0553/0.0991 0.0421/0.0563
goodness-of-fit on F2 1.171 1.128 1.192 1.049
largest diff peak and hole/e·A−3 4.444/−1.483 2.091/−1.651 1.360/−1.709 0.731/−1.349
a
R1 = ∑||Fo| − |Fc||/∑|Fo|. bwR2 = {∑ [w(Fo2 − Fc2)2]/∑[w(Fo2)2]}1/2.

cage and on the formation of the perovskite octahedral semiconductor with a band gap of 1.3 eV. We demonstrate that
network.7 the CsSnI3 is a unique system that combines two generally
Despite its interesting physical and chemical properties, our contra-indicated properties, strong PL and high electrical
understanding of the origin of such properties in CsSnI3 is still conductivity. The data show that the hole mobilities in this
limited. Additionally, many previous reports understate its high material are among the highest known in medium gap
sensitivity to air, moisture, and organic solvents, resulting in an compound semiconductors (∼585 cm2 V−1 s−1). Remarkably,
incorrect description of the phase change behavior and its near-infrared PL emission at 950 nm is remarkably strong at
unreliable characterization results.1,3 Some authors also room temperature and orders of magnitude greater than that of
revealed the inclusion of impurities in their syntheses.2 Most a standard commercially available InP single-crystalline wafer
importantly, the black phases of CsSnI3 have never been which also emits in the same wavelength. It is important to note
structurally characterized because of the extreme difficulty of that solid-state materials emitting near-IR light at room
single-crystal X-ray diffraction (XRD) studies. As a result, temperature are highly desirable for noninvasive investigations
theoretical and experimental studies have been severely limited, of paintings and artifacts,8 optical amplifiers,9 fiber-optic
and the origin of its phase transition has never been systems for telecommunications,10 tunable solid-state laser
understood. systems, and biomedical applications.11 Currently available
Here we report the synthesis, single-crystal structure materials for such applications are compounds containing
refinements, charge transport and optical properties, and lanthanides12 and some III/V and II/VI semiconductors such
theoretical calculation studies of all polymorphic forms of as InP, GaAs, and CdTe. However, they show low emission
CsSnI3. We determined the single-crystal structures of the three efficiency. Therefore, materials that consist of environmentally
black polymorphs by in situ temperature-dependent single- friendly and inexpensive elements with high emission efficiency,
crystal XRD studies for the first time. Examination of the such as CsSnI3, are highly sought after.
coordination sphere of Cs+ in the three perovskite frameworks Temperature-dependent electrical conductivity of the black
suggests that the transitions are the result of the desire of the phase shows metallic behavior, and, surprisingly, the con-
Cs+ cation to achieve closest packing and the remarkable ductivity and PL intensity gradually increase upon heat
flexibility of the perovskite network thanks to the plasticity of treatment. The results of ab initio density functional theory
the Sn−I−Sn angle. Based on an in situ temperature-dependent (DFT) calculations suggest that such phenomena may result
synchrotron powder XRD study, we probed in detail the from facile formation of Sn vacancies in the structure, which
reversible phase transitions. The interconversions of the four form hole carriers only in the black polymorphs of CsSnI3.
polymorphs proceed under specific conditions along with the
influence of air, moisture, and organic solvents. Surprisingly, the EXPERIMENTAL SECTION
black polymorphs exhibit one of the largest volumetric thermal
■ Synthesis. The synthesis of the pure black orthorhombic phase of
coefficient for inorganic solid materials, close to that for liquid. CsSnI3 was achieved by reacting a stoichiometric mixture of CsI and
We also show with optical absorption spectroscopy, Hall SnI2 in an evacuated Pyrex tube at 550 °C for 1 h, followed by cooling
effect, and Seebeck coefficient measurements that the black for 6 h to room temperature. The synthesis of the pure yellow phase
orthorhombic perovskite phase is a p-type direct band gap was achieved by reacting a stoichiometric mixture of CsI and SnI2

8580 dx.doi.org/10.1021/ja301539s | J. Am. Chem. Soc. 2012, 134, 8579−8587

Journal of the American Chemical Society Article

Figure 1. Illustration of phase change details of four CsSnI3 polymorphs and their crystal structures. How phase transitions of CsSnI3 polymorphs
proceed is displayed with synthesis conditions and conversion temperatures obtained by in situ temperature-dependent synchrotron powder XRD
studies. Crystal structures of the respective polymorph are shown: (a) black cubic (B-α) at 500 K; (b) tetragonal (B-β) at 380 K; (c) orthorhombic
(B-γ) at 300 K; yellow (Y) phase at 300 K. Disordered atoms in (c) are omitted for clarity. Cs atoms, blue; Sn atoms, yellow; I atoms, violet. Green
dashed lines represent a unit cell.

(total ∼0.3 g) in ethylenediamine (0.2 mL) in an evacuated Pyrex tube ∼2 mm × 3 mm × 8 mm under N2 atmosphere. Electrical conductivity
at 140 °C for 3 d. Large crack- and bubble-free ingots (ca. 6 g) were and Seebeck coefficient were measured simultaneously under He
prepared using a modified vertical Bridgman technique, and the atmosphere from room temperature to 650 K on a ULVAC-RIKO
growth condition was guided by the DTA results. Pure CsSnI3 powder ZEM-3 instrument system. The mechanical electrodes made of Rh/Pt
in an evacuated fused silica tube was passed down a single-zone and Pt were directly contacted to the ingot with no use of a chemical
Bridgeman furnace at a rate of 2 cm h−1, maintained at 550 °C. The paste. Six heating and cooling cycles were repeated for a given sample.
phase purity of the ingot obtained was confirmed by powder XRD Hall Effect Measurements. The Hall coefficient was measured on
analysis (Supporting Information, Figure S1). a polycrystalline bar-shaped ingot with four cold-pressed In contacts
Powder X-ray Diffraction. Powder XRD analysis was performed (one on each side of the ingot) positioned in a standard Hall
using a calibrated CPS 120 INEL powder X-ray diffractometer (Cu Kα geometry. The sample was subjected to a flow of nitrogen gas to
graphite-monochromatized radiation) operating at 40 kV/20 mA and protect it from oxidation during measurements. A dc current source
equipped with a position-sensitive detector with flat sample geometry. (6220 Keithley Instruments) was utilized to supply a constant 100 mA
In Situ Temperature-Dependent Synchrotron Powder X-ray current through the sample, while an external magnetic field up to
Diffraction. Fine powders of a black CsSnI3 (B-γ) compound were ±1.4 T was applied in steps of 0.25 T. The voltage response, in both
sealed in an N2-filled Roentgen glass tube. In situ synchrotron XRD (λ the positive and negative directions of the magnetic field, was recorded
= 0.413975 Å) patterns were recorded from room temperature to 673 at each respective step by a nanovoltmeter (2182A Keithley
K at a rate of 5 K h−1 using at the 11-BM beamline at the Advanced Instruments), and half the difference of the two signals was recorded
Photon Source. as the Hall voltage.
Single-Crystal X-ray Crystallography. Intensity data for black Optical Absorption Spectroscopy. Optical diffuse reflectance
and yellow CsSnI3 were collected at 300(2)−500(2) K on a STOE measurements were performed at room temperature using a Shimadzu
IPDS 2T diffractometer with Mo Kα radiation operating at 50 kV and UV-3101 PC spectrometer operating in the 200−2500 nm region. The
40 mA with a 34 cm diameter imaging plate under N2 atmosphere. reflectance versus wavelength data generated were used to estimate the
Individual frames were collected with a 3 min exposure time and a band gap of the material by converting reflectance to absorption data
1.0ω rotation. The X-AREA, X-RED, and X-SHAPE software packages according to the Kubelka−Munk equation: α/S = (1 − R)2(2R)−1,
were used for data extraction and integration and to apply empirical where R is the reflectance and α and S are the absorption and
and analytical absorption corrections. The SHELXTL software scattering coefficients, respectively. The details of the energy gap
package13 was used to solve and refine the structure. The parameters measurements have been discussed elsewhere.14
for data collection and the details of the structural refinement are given Theoretical Calculations and Electronic Structures. The
in Table 1. Fractional atomic coordinates and displacement parameters electronic structures of CsSnI3 were calculated by the full potential
are given in Supporting Information, Tables S1−S8. The selected linearized augmented planewave (FLAPW) method15 based on DFT;
bond distances and angles are given in Tables S9 and S10. experimental lattice constants and atomic coordinates were employed.
Charge Transport and Thermopower Measurements. The The muffin-tin radii for Cs, Sn, and I atoms were chosen to be 2.90,
ingot was cut and polished into a rectangular shape with dimensions of 2.70, and 2.50 Bohr, respectively. The cutoff energies for the

8581 dx.doi.org/10.1021/ja301539s | J. Am. Chem. Soc. 2012, 134, 8579−8587

Journal of the American Chemical Society Article

planewave basis and star function were 13 and 144 Ry, respectively; with cubic Pm3m ̅ space group (Figure 1a). All Cs, Sn, and I
5×5×3 and 5×9×3 meshes of special k-points were used in the atoms reside on the special positions of the space group. The
irreducible Brillouin zone for the B-γ and Y phases, respectively. Sn2+ center sits in an ideal octahedral geometry with six I−
Within the local density approximation (LDA), the Hedin−Lundqvist anions, resulting in stereochemically inactive 6s2 lone pair
form16 of the exchange-correlation functional was used, and spin−orbit
coupling (SOC) was included by a second variation method.17 In electrons. The more common coordination geometries for Sn2+
order to improve the accuracy of the band gap, which is usually are trigonal pyramidal found in SnCl2 and SnS and square
underestimated within LDA, the nonlocal scheme of the screened- pyramidal found in SnO and α-SnWO 3 . The [SnI6/2] −
exchange LDA (sX-LDA) method was also employed.18 This method octahedra condense to form a three-dimensional cubic
leads to good agreement with experimentally measured band gaps and framework via corner-sharing with the Cs+ countercations
improves the conduction band topology as well.14,19 residing at 12-coordinate interstices within the network made
Defect Formation Energy Calculations. The defect formation by eight [SnI6/2]− octahedra. The Sn−I distance is normal at
energy is a function of the chemical potentials of atoms and electron. 3.1028(4) Å, and the bond angle of Iap−Sn−Ieq is 90°.
The set of chemical potentials is determined by the experimental
When cooled, the cubic form undergoes successive displacive
growth condition. Based on the total energy from the ab initio
calculation, we calculate the defect formation energy, ΔHD,q,20 defined phase transitions to lower symmetry with no bond breaking
by the eq 1: (Figure 1b,c). The Pm3̅m symmetry is lowered to tetragonal
P4/mbm at 380 K (B-β) and orthorhombic Pnma at 300 K (B-
ΔHD , q = [ED , q − E H] + q(ΔE F + Ev ) + ∑ nα(μα0 + Δμα ) (1) γ). The B-β phase results from the tilting of the neighboring
[SnI6/2]− octahedra in only the ab plane to give Sn−I(2)eq−Sn
In this formula, the first term on the right-hand side defines the = 167.989° (Glazer tilting system: a0a0c−) (Figure S2). The
binding energy ΔE of the defect D in the host system. The second Sn−I distances slightly differ: Sn−I(1)ap, 3.0954(3) Å; Sn−
term represents the energy of the charge q change from the Fermi level I(2)eq, 3.0994(5) Å. Neither tilting of the octahedra viewed
EF = ΔEF + Ev, where Ev is the energy of the valence band maximum of
host system. The third term is the energies of atoms of species α added
down the c axis nor the bent bond angle in the [SnI6/2]−
to (or removed from) reservoirs of elements with chemical potential octahedron is observed: Sn−I(1)ap−Sn, 180°; I(2)eq−Sn−
nα = μ0α + Δμα, where μ0α is the reference energy of the most stable I(2)eq, 90°; I(1)ap−Sn−I(2)eq, 90°. The new cell is enlarged to
phase of each element. Possible ranges of Δμα are determined by the a √2a×√2a×a supercell with Z = 2 (Z is the number of
thermodynamic stable condition of host material, and ΔEF is obtained formula units per unit cell). In the B-γ phase at room
as a function of temperature within the charge neutrality condition. To temperature, octahedral tilting occurs along both the apical and
reduce the effect caused by finite cell size, we considered a large equatorial directions (a+a−a−) to give bond angles of 157.6(4)°
supercell which includes 360 atoms. For each defect, the internal and 166.0(3)° for Sn−I(1a)ap−Sn and Sn−I(2a)eq−Sn,
coordinates of atoms are relaxed. Relaxation and total energy of
respectively (Figure S1). Some internal bond angles of the
pristine and each defected supercell are calculated by using a plane
wave basis and projector augmented wave method21 which is [SnI6/2]− octahedron are also slightly distorted to give the
implemented in the Vienna ab initio simulation package.22 The energy I(1a)eq−Sn−I(2a)eq, 88.6(2)° and 88.9(1)°, and cis-I(2a)ap−
cutoff for the plane wave basis set was set to 300 eV. For the k-point Sn−I(2a)eq, 89.95(3)°, while those of I(1a)ap−Sn−I(1a)ap and
sampling, a 1×1×1 mesh is used in the relaxation, and a 3×3×3 mesh the trans-I(2a)eq−Sn−I(2a)eq remain at 180°. The distances of
is used for total energy calculations. The generalized gradient Sn−I(1a)ap and Sn−I(2a)eq deviate more than those in the B-β
approximation (GGA) is used for the exchange-correlation functional phase: 3.124(5) and 3.118(1) Å, respectively. Consequently,
within the Perdew−Burke−Ernzerhof (PBE) formalism;23 the SOC the √2a×2a×√2a supercell is obtained with Z = 4. Note that
interaction is not considered here. investigation of the reciprocal lattice using the extended X-ray
exposure did not reveal any superstructure or twinning.
■ RESULTS AND DISCUSSION When exposed to air or organic solvents such as acetonitrile,
Synthesis, Crystal Structure, and Phase Transitions. the B-γ phase undergoes a significant reconstructive phase
The yellow phase of CsSnI3 was isolated and structurally transition to the Y phase at room temperature in 1 h (Figure
characterized in 1980,24 whereas the black phases were only 1d). The composition of the latter is better described as
studied by powder XRD two decades ago.1 We observed that Cs2Sn2I6. The edge-sharing [SnII2/2I3/3]− octahedra condense
obtaining suitable single crystals for XRD analysis of black to form infinite one-dimensional double chains of 1∞[Sn2I62−]
phases is extremely difficult. When melts of CsSnI3 cool to form running down the crystallographic b axis, separated by Cs+
crystals, a phase transition typically occurs anisotropically from cations. There is one crystallographically independent Sn and
higher to lower symmetry, resulting in inhomogeneous Cs atom and three I atoms. The [SnI6/2]− octahedra are
crystallographic domains in the crystals. Starting materials of distorted to give I−Sn−I angles ranging from 91.28(2)° to
CsI and SnI2 that contain even very minor impurities give 96.32(2)°. The Sn−I distances are 3.2243(6) and 3.1968(6) Å
mixtures of the Y and B-γ phases in solid-state reactions, with to bridging I(1) and (2) atoms, respectively, and 2.9518(8) Å
single crystals of only the Y phase. When exposed to air, the B-γ to the terminal I(3) atom. It is important to note that the B-γ
crystals promptly lose single crystallinity, rapidly transform to phase does not spontaneously transform to the Y phase under
the polycrystalline yellow phase, and ultimately decompose. an inert atmosphere. Some previous reports incorrectly
We determined the single-crystal structures of the three indicated that grinding the black phase gave the yellow
metastable black polymorphs for the first time (Table 1). To polymorph.1 Such a transition results from exposure to air,
obtain the high-temperature B-α phase, a melt of CsSnI3 was not pressure. The Y phase transforms to the B-α phase on
quenched at 623 K. The lustrous black block-like crystal was heating above 425 K under inert atmosphere.
immediately sliced to a thin plate to minimize stress by the Local Coordination Environment. Examination of the
phase transitions and loaded on a single-crystal diffractometer, coordination environment and anisotropic displacement
maintained at 500 K. The single crystal chosen showed a single parameters (ADPs) of Cs atoms obtained from single-crystal
crystallographic domain, indexed to a cubic unit cell. The B-α XRD studies provided important clues on what drives these
phase forms in the ideal three-dimensional perovskite structure phase transitions (Tables S5−S8). At 500 K, the Cs atoms
8582 dx.doi.org/10.1021/ja301539s | J. Am. Chem. Soc. 2012, 134, 8579−8587
Journal of the American Chemical Society Article

reside in the ideal, 12-coordinated cuboctahedral geometry with distances, including the shortest of 3.902(9) Å, become
a Cs−I distance of 4.3881(5) Å. Note that this distance is significantly shorter than that at 500 K of 4.3881(5) Å. The
significantly longer than that found in CsI (3.94 Å) and the sum longer distances to the opposite side of the shifted Cs(a) atoms
of the ionic radii of I− and 12-coordinated Cs+ (4.08 Å).25 The are as long as 5.19 Å, where Cs(b) occupies the void.
ADPs of Cs atoms at this elevated temperature are unusually Cs atoms in the yellow phase no longer rattle in the
high: U11 = U22 = U33 = 0.154 Å2; U23 = U13 = U12 = 0 Å2, but oversized cages but instead are eventually stabilized in a nine-
are isotropic by symmetry. These data indicated that the Cs coordinated tricapped trigonal prism geometry, surrounded by
atoms randomly rattle inside the oversized perovskite cages, as three adjacent 1∞[Sn2I62−] chains. The Cs−I distances narrowly
observed in clathrate compounds (Figure 2a). The ADPs of I range from 3.8760(7) to 4.0687(9) Å, much shorter than those
of the black phases, and the APDs of all atoms are nearly ideal.
The cell volume per formula unit decreases throughout the
phase transitions from cubic perovskite to yellow phases: B-α,
239.99(5) Å3; B-β, 235.27(2) Å3; B-γ, 232.25(2) Å3; Y,
217.94(4) Å3 (Table 1 and Figure S4). The mass density
calculated from the structural models presented above increases
from B-α, 4.393 g cm−3; B-β, 4.463 g cm−3; B-γ, 4.521 g cm−3;
to Y, 4.818 g cm−3. It is also interesting to note that the phase
change between the two room-temperature forms involves a
remarkable 6.6% increase in density. Thus, the yellow phase is
the best packing model for the CsSnI3 stoichiometry and likely
the thermodynamically stable form.
To confirm our assumption that stabilization of Cs atoms in
CsSnI3 plays a key role in phase transitions, we examined
relevant phases that include different size combinations of
component atoms. Presumably due to the smaller size of Rb,
the perovskite phase of RbSnI3 does not exist (the cage for the
Rb atoms is too large) and instead only forms the one-
dimensional structure isomeric to yellow CsSnI3.3,26 In contrast,
the bigger, asymmetric CH3NH3+ cation can better support the
cages and only forms a cubic perovskite with no phase
transition near room temperature.27 Indeed, [CH3NH3]SnI3
Figure 2. Coordination environments of Cs atoms in the 3∞[SnI3−] shows even larger cell volume per formula unit of 243.4(1) Å3
perovskite cage in the black phases at (a) 500 K (B-α), (b) 380 K (B- than black CsSnI3 at room temperature.27 With a smaller cage
β), and (c) 300K (B-γ). (d) Cs atoms are surrounded by three of 3∞ [SnBr3−], CsSnBr3 can only form the perovskite structure.2
1 2−
∞[Sn2I6 ] chains in the yellow phase (Y). Split models of I and Cs In Situ Temperature-Dependent Synchrotron X-ray
atoms are shown in (c). I atoms are labeled. Cs atoms, blue; Sn atoms, Diffraction. To track each phase transition, we performed in
yellow; I atoms, wine in (a), (b), and (d). In (c), Cs(a) atoms, blue; situ temperature-dependent synchrotron powder XRD studies
Cs(b), light blue; Sn atoms, yellow; I(1a,2a) atoms, wine; I(1b,2b)
atoms, orange. The Cs−I bond distances are given in Å. The thermal on the B-γ phase from room temperature to 673 K. Figure 3
ellipsoids are shown at 80% probability. shows the temperature dependence of XRD patterns on heating
at a rate of 5 K h−1 with a 2θ range of 5.3−6.9° for selective
temperatures. The XRD pattern at room temperature
atoms indicate large thermal vibrations perpendicular to the confirmed the purity of the B-γ phase. The doublet around
Sn−I−Sn bonds: U11 = 40(1), U22 = U33 = 0.215(2) Å2; U23 =
U13 = U12 = 0 Å2. Correspondingly, the Sn−I−Sn bonds are
flexible and can bend in order to distort the cages and get Cs
and I atoms closer.
When cooled to 380 K, the 3∞[SnI3−] framework compress
slightly along the ab axis to make the Sn−I(1)ap−Sn angles
bend, but still with high ADPs in the I atoms (Figure 2b). Cs
atoms vibrate ellipsoidally (U11 = U22 = 0.134(3) Å2, U33 =
0.120(4) Å2) along the elongated direction (a and b axes) of
the cages. The Cs atoms sit in a distorted cuboctahedral
geometry, and, because of this distortion, four of the Cs−I
bonds become significantly shorter than those at 500 K: four
each of 4.146(3), 4.3591(4), and 4.6030(3) Å (Table S10).
At room temperature, Cs and I atoms finally have to be
modeled with split positions and their occupancies refined
(Figure 2c). As a result, the ADPs of Cs and I atoms are close Figure 3. In situ temperature-dependent synchrotron XRD patterns of
to normal. The occupancy ratios of Cs(a):Cs(b), I(1a):I(1b), black CsSnI3 on heating. Changes in the Bragg reflection peaks
and I(2a):I(2b) were refined to 76:24, 75:25, and 77:23, indicate temperature-driven phase transitions to higher symmetry on
respectively. The Cs(a) atom shifts to one side of the distorted heating: orthorhombic (B-γ) to tetragonal (B-β) above 362.1 ± 5.4 K
cage to effectively interact with I atoms, as does the Cs(b) atom and the latter to orthorhombic (B-α) above 440.5 ± 3.5 K. The
in the opposite direction. As a result, nine of the Cs(a)−I representative XRD peaks that reflect phase transitions are indexed.

8583 dx.doi.org/10.1021/ja301539s | J. Am. Chem. Soc. 2012, 134, 8579−8587

Journal of the American Chemical Society Article

6.11−6.13° (indexed as the (201) and (102) reflections) starts

merging into a single peak at 6.09° (indexed as the (210)
reflection of the tetragonal cell) above 338.0 ± 3.4 K. Above
372.1 ± 4.6 K, the characteristic Bragg peaks of the
orthorhombic cell completely disappear: (210), (031), (211),
and (112), indicative of forming the tetragonal B-β phase. On
subsequent heating, the (200) and (210) peaks vanish,
indicating the formation of the cubic B-α phase above 440.5
± 3.5 K, which is stable up to 673.0 K. On cooling, the opposite
trend is observed. At 420.1 ± 5.8 K, the (200) and (210) peaks
evolve to restore the B-β form. Below 352.0 ± 5.3 K, the Bragg
peaks of the B-γ phase start to appear (Figure S3). The results
confirm the reversibility of the phase transition among the black
phases. Figure 4. (a) Temperature-dependent electrical conductivity. The first
The cell volume changes smoothly as a function of run (●) is enlarged in the inset for clarity. (b) Seebeck coefficients of
temperature with no noticeable discontinuity by the phase black CsSnI3 in (a). In both plots the second run (■) was performed 1
transition (Figure S4). For the black phase, the coefficients of week after the first run.
the linear thermal expansion are αa = 5.01 × 10−5 K−1, αb =
1.42 × 10−5 K−1, and αc = 7.71 × 10−4 K−1. The coefficient of performed a second run on the same ingot. The electrical
the volumetric thermal expansion at 298 K is measured to be β conductivity at room temperature shows a value similar to that
= 1.26 × 10−4 K−1. This value is unusually large for solid and obtained for the fourth cycle, indicating that the increased value
close to that for liquid. For example, those for PbTe, Hg, and of the electrical conductivity is sustained. Over the course of six
distilled water at 293 K are 0.2 × 10−4, 1.82 × 10−4, and 2.07 × heating cycles, conductivity rises drastically from 16 to 208 S
10−4 K−1, respectively. Since the observed value is one of the cm−1 at room temperature. Therefore, heat energy boosts the
largest for solid-state inorganic materials, we present theoretical carrier concentration.
calculations of bulk modulus below in the Electronic Structure Thermoelectric power measurements gave positive Seebeck
Calculation section to explain these observations. coefficients, suggesting p-type conduction, which increase with
Figure 1 illustrates the phase transition details of the four increasing temperature. Their values steadily decrease through
polymorphs. The transition between the black phases is all the heating cycles, consistent with the increase in the
reversible, driven by temperature under inert atmosphere, as electrical conductivity. The value at room temperature
shown by in situ synchrotron XRD studies. The transition from dramatically diminishes from 242 to 99 μV K−1 over these
the B-γ to the Y phase, typically activated by exposure to air, measurements. Hall effect measurements of the pristine
moisture, or organic solvents, readily proceeds at room material show a low carrier concentration at room temperature
temperature, whereas the reverse requires heating above 425 of p < 1017 cm−3, indicative of p-type conduction and consistent
K. Under inert conditions, the B-γ phase, however, is stable for with the positive Seebeck coefficient. The calculated resistivity
more than a year. The black phase congruently melts at 708 K and mobility are 0.9 Ω cm and 585 cm2 V−1 s−1, respectively,
and recrystallizes at 718 K (measured at the onset, Figure S5). and are comparable to those obtained from transport property
It should be noted that the pure black phase is obtained only by measurements: ρ = 1 Ω cm and μ = 400 cm2 V−1 s−1. This
solid-state reaction, whereas the pure yellow phase is prepared mobility is one of the highest observed among p-type
only by solution or solvothermal reactions. Several authors semiconductors with comparable band gaps. In comparison,
incorrectly reported a compound greenish in color. The GaAs (Eg = 1.45 eV) and InP (Eg = 1.3 eV) exhibit hole
greenish color is typically the Y phase contaminated by the mobility of 400 and 700 cm2 V−1 s−1 at room temperature,
B-γ and/or Cs2SnI628 that includes SnIV. Cs2SnI6 is also a respectively.29 Discontinuities arising from the phase transition
typical impurity in the solution synthesis of the Y phase and is are not observed in the electrical conductivity or thermopower
black in color. measurements, consistent with the result from synchrotron
Charge Transport Properties. For the charge transport XRD studies (Figures S3 and S4).
property measurements, we prepared large, crack-free poly- Optical Properties. The two distinct polymorphic forms at
crystalline ingots (ca. 8 g). Our crystal growth conditions were room temperature (B-γ and Y) display a striking contrast in
guided by the DTA results. Sample preparation and all optical band gaps because of the significant structural
measurements were carried out under inert atmosphere. differences. Optical absorption spectra reveal sharp absorption
Mechanical contacts made of Pt/Rh and Pt were directly edges at 1.3 eV for the B-γ and 2.55 eV for the Y phases (Figure
applied to the rectangular ingot with no application of a paste 5). A discussion on what causes the optical contrast is given
that typically contains organic solvents. The temperature below in the ab initio theoretical calculations section. When
dependence of the electrical conductivity and the Seebeck excited with 533 nm light irradiation at room temperature, the
coefficient over six independent heating cycles at a rate of 5 K B-γ phase exhibits strong near-IR PL at ca. 950 nm. The PL
h−1 between individual steps show highly unusual behavior intensity of as-prepared powders is several times larger than
(Figure 4). Although the electrical conductivity decreases with that of a single-crystalline InP wafer (Figure 5). Materials
increasing temperature over the entire temperature range, exhibiting strong near-IR emission operating at room temper-
indicating metallic behavior, with each heating cycle the overall ature are rare and highly desirable. For example, emission at
conductivity rises. 950 nm is preferred for biomedical analysis because IR light
Because of the rise in electrical conductivity after each better penetrates human tissues than UV and visible lights to
heating cycle, we attempted to determine if the increase was give higher resolution images, and wavelengths between 700
permanent. One week after measuring four cycles, we and 950 nm are least absorbed by the tissues.30 Such an
8584 dx.doi.org/10.1021/ja301539s | J. Am. Chem. Soc. 2012, 134, 8579−8587
Journal of the American Chemical Society Article

emission is also highly desirable for current silica-based

telecommunication systems having standard telecommunica-
tion windows at 980 nm.9
We examined the optical emission and absorption spectra as
well as the powder XRD patterns after the same heat treatment
cycles employed in the charge transport property measure-
ments. Surprisingly, the PL intensities gradually increased after
each cycle, similar to the trend in the electrical conductivity.
However, no noticeable changes in the optical absorption edges
and XRD patterns were observed, indicating no structural
Figure 5. Optical absorption and photoluminescence spectra of black change or deformation in the compound.
CsSnI3 (B-γ) at room temperature. PL (λmax ≈ 950 nm) and optical
Electronic Structure Calculations. To understand the
absorption spectra of the B-γ phase showing a band edge at 1.3 eV are
displayed. PL intensity steadily increases over heat treatment, which is nature of the band gaps of the black and yellow phases and
compared with that of a single-crystalline InP wafer. their striking difference in properties, we performed ab initio
DFT calculations using the FLAPW method.15 Experimental
lattice constants and atomic coordinates were employed.
Within the LDA, the Hedin−Lundqvist form16 of the
exchange-correlation functional was used, and SOC was

Figure 6. Calculated band structures of (a) the B-γ and (c) Y phase using the sX-LDA method, showing the direct and indirect band gap nature,
respectively. The projected density of states (PDOS) for individual atoms calculated with the sX-LDA method for (b) the B-γ and (d) Y phases is
shown. (e) Diagrams of calculated defect formation energies for VSn, VCs, SnI, and VI in the B-γ form. Charge state, q, is denoted by +, 0, and −. The
energy level of the VBM is set to zero. The regions below VBM and above CBM are shaded in gray color. The defect formation energies of SnI are
higher than 2.5 eV in the Sn-poor, I-rich and the Cs-poor, I-rich conditions.

8585 dx.doi.org/10.1021/ja301539s | J. Am. Chem. Soc. 2012, 134, 8579−8587

Journal of the American Chemical Society Article

included by a second variation method.17 In order to improve ∂P

B = −V
the accuracy of the band gap, which is usually underestimated ∂V (2)
within LDA, the nonlocal scheme of the sX-LDA method was
also employed.18 This method leads to good agreement with where P is pressure and V is volume. The bulk modulus is
the experimentally measured band gaps and improves the calculated by fitting calculated total energies in various volumes
conduction band topology as well.14,19 with the Birch−Murnaghan equation of state.31 The obtained
The band structure and projected density of states (PDOS) bulk modulus of B-γ phase is 19.84 GPa. This value is extremely
for orthorhombic black CsSnI3 (B-γ) are presented in Figure small considering those of other crystalline solid materials: 70
6a,b. The band structure calculated by the sX-LDA method GPa for BeTe,32 and 322 GPa for TiN33 and liquid Hg (25.3
with SOC determines a direct band gap between the valence GPa). Note that perovskite CsPbI3 that is isostructural to black
band maximum (VBM) and the conduction band minimum CsSnI3 shows 23 GPa.34 Since the bulk modulus is inversely
(CBM) at the Γ point. The VBM mainly consists of p-orbitals proportional to the coefficient of volumetric thermal expansion
of I, and the CBM is composed of p-orbitals of Sn. Thus, the via the Gruneisen constant, the obtained bulk modulus
band gap originates from the p−p mixing between Sn and I. supports the extremely large value of the coefficient of thermal
expansion of B-γ phase.
Our sX-LDA calculation with SOC yields conduction band
Defect Formation Energy. To better understand the p-
dispersion quite different from the previous report,7 which, we
type nature and unusual observations in transport and optical
believe, was calculated within the GGA only. The bandwidth of
properties of black CsSnI3, we calculated the defect formation
the conduction band increases and suggests a small electron
energy. We regarded point defects, a Sn vacancy (VSn), a Cs
effective mass. The partial hybridization between Sn (s- and p- vacancy (VCs), and Sn replacing I (SnI), as possible intrinsic p-
orbitals) and I (p-orbitals) in the valence band (between −5 type sources. We also considered the defect formation energy
and 0 eV) is consistently shown in both studies. of an I vacancy (VI) that might act as an n-type source since it
Figure 6c,d shows the band structure and PDOS of yellow can compensate for the hole carrier. Figure 6e shows the defect
CsSnI3 (Y), respectively. The indirect band gap of 2.46 eV is formation energies of VCs, VSn, SnI, and VI as a function of the
between the X point and the middle of the Γ and Y points. Fermi level under three different growth conditions: (Sn-poor,
However, the direct band gap (about 2.53 eV) is located at the I-rich), (Cs-poor, I-rich), and (Sn-rich, I-poor). Under the
nearly flat band along Γ to X, which is only 0.07 eV larger than three growth conditions, the defect formation energy of VSn
the indirect band gap. This should give rise to a sharp increase generating hole is lower than those of any other defects, which
in the intensity of the optical transition near the band edge, means that VSn is the most favorable intrinsic defect. The
while the optical transition of the indirect gap is expected to be possible compensation of hole by VI is unfavorable due to its
relatively small. Thus, the calculated direct band gap shows higher formation energy. These results strongly indicate that a
better agreement with the experimental value of 2.55 eV. The major p-type carrier source is VSn, and the black compound is a
overall composition of the PDOS is similar to that of the B-γ nearly intrinsic p-type semiconductor. The facile generation of
phase (Figure 6b,d). The band gap of the Y phase also Sn vacancies would be even better activated by heat energy,
originates from the hybridization between the p-orbitals of Sn resulting in continuously rising electrical conductivity and PL
and I. Unlike the case of the B-γ phase, the PDOS of the Y intensities of black CsSnI3. Considering observations of the
phase shows two kinds of I atoms in yellow CsSnI3. For I(1) close correlation of PL and charge transport behavior, coupled
and I(2), the hybridization of I p-orbitals with Sn s-orbitals at with DFT results, the observed near band gap PL plausibly
the top of the valence band and Sn p-orbitals at the bottom of originates from defect centers made by a Sn vacancy, rather
conduction band is more enhanced, whereas I(3) has more than a transition between the band edges.
lone-pair p-electrons that appear in the middle of the valence
band (between −2.5 and −1.0 eV). This difference can be ■ CONCLUSION
explained by the different local environments of I(1,2) and We elucidated what drives the facile phase transitions of
I(3); I(1) and I(2) are located in the square plane of the CsSnI3, whose understanding has been elusive. A surprising
octahedron and are bonded to two Sn atoms with a 90° bond finding is that two distinct polymorphs that are stabilized at
angle. On the other hand, the terminal I(3) is bonded to only room temperature readily convert to each other by heating or
one Sn atom. The difference in the number of nearest-neighbor exposure to air, despite involvement of significant structural
atoms accounts for the different electronic structures of I atoms reconstruction. One of them is a highly conducting p-type
explained above. semiconductor and black in color, in contrast to its yellow
The band gap contrast between the B-γ and Y phases is counterpart with an indirect band gap. They show a dramatic
mainly caused by major structural difference between the two optical band gap difference of 1.25 eV. Their phase transition is
phases. As shown in Figure 1, the 3D network of octahedra in derived from stabilization of rattling Cs atoms in the oversized
the perovskite structure is changed to a pseudo-1D structure perovskite cages. Therefore, it is controllable by introducing
built of edge-sharing octahedra chains. The lower dimension- other congeners or more covalent Ag, Cu, or Tl atoms that
ality in the Y phase gives rise to narrower bandwidths (see should better interact with the cages or organic cations for alkali
Figure 6), which leads to an increase in the optical band gap. It metals.
is important to note that the total energy of the latter is 0.15 eV In going from one perovskite phase form to the next at lower
per formula unit lower than that of the B-γ phase, indicating the temperature and ultimately to the yellow phase, the density
stability of the Y phase over the B-γ phase. gradually increases as better structural packing is achieved; the
In order to explain the extremely large value of the coefficient yellow phase gains 7−10% in density compared to the other
of thermal expansion of B-γ phase, we calculated the bulk phases. CsSnI3 is a rare case of a material that combines high
modulus (B). It represents the resistance of material to uniform electrical conductivity and strong PL. Electrical conductivity
compression, defined by eq 2: and thermopower measurements indicate black CsSnI3 to be a
8586 dx.doi.org/10.1021/ja301539s | J. Am. Chem. Soc. 2012, 134, 8579−8587
Journal of the American Chemical Society Article

p-type metal, despite the obvious optical band gap of 1.3 eV. (12) Bünzli, J. C. G.; Piguet, C. Chem. Soc. Rev. 2005, 34, 1048−1077.
Very unusually, it becomes more conducting by heating (13) Sheldrick, G. M. Acta Crystallogr. A 2008, 64, 112−122.
treatment, which is coupled with the enhancement of near-IR (14) (a) Chung, I.; Song, J. H.; Jang, J. I.; Freeman, A. J.; Ketterson, J.
PL at room temperature. These observations possibly result B.; Kanatzidis, M. G. J. Am. Chem. Soc. 2009, 131, 2647−2656.
(b) Trikalitis, P. N.; Rangan, K. K.; Bakas, T.; Kanatzidis, M. G. J. Am.
from the readily changing Sn vacancy centers in the structure Chem. Soc. 2002, 124, 12255−12260.
according to our ab initio first-principles DFT calculations. (15) Wimmer, E.; Krakauer, H.; Weinert, M.; Freeman, A. J. Phys.
They also suggest that PL originates from the Sn defect centers Rev. B 1981, 24, 864−875.
rather than electron transfer between conduction and valence (16) Hedin, L.; Lundqvist, B. I. J. Phys. C: Solid State Phys. 1971, 4,
bands. The ability of CsSnI3 to emit strong near-IR PL at room 2064−2083.
temperature is a rare property and also important since it is free (17) MacDonald, A. H.; Pickett, W. E.; Koelling, D. D. J. Phys. C
of rare-earth metals. This study opens up opportunities to 1980, 13, 2675−2683.
discover visible and IR light-emitting materials at room (18) Bylander, D. M.; Kleinman, L. Phys. Rev. B 1990, 41, 7868−
temperature, which is highly desirable in achieving break- 7871.
through applications in photovoltaics,35 various radiation (19) (a) Chung, I.; Biswas, K.; Song, J. H.; Androulakis, J.;
Chondroudis, K.; Paraskevopoulos, K. M.; Freeman, A. J.; Kanatzidis,
detectors, and light-emitting diodes because the perovskite
M. G. Angew. Chem., Int. Ed. 2011, 50, 8834−8838. (b) Chung, I.;
structure can incorporate a very broad range of elements. Song, J. H.; Kim, M. G.; Malliakas, C. D.; Karst, A. L.; Freeman, A. J.;
Weliky, D. P.; Kanatzidis, M. G. J. Am. Chem. Soc. 2009, 131, 16303−
■ ASSOCIATED CONTENT 16312. (c) Bera, T. K.; Song, J. H.; Freeman, A. J.; Jang, J. I.;
* Supporting Information
S Ketterson, J. B.; Kanatzidis, M. G. Angew. Chem., Int. Ed. 2008, 47,
Fractional atomic coordinates, displacement parameters and 7828−7832. (d) Kim, M.; Freeman, A. J.; Geller, C. B. Phys. Rev. B
selected bond distances and angles for CsSnI3 polymorphs; 2005, 72, 035205.
(20) Van de Walle, C. G.; Neugebauer, J. J. Appl. Phys. 2004, 95,
diagrams of volumetric thermal expansion and DTA. This
3851−3879.
material is available free of charge via the Internet at http:// (21) Blochl, P. E. Phys. Rev. B 1994, 50, 17953−17979.
pubs.acs.org. (22) Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, 558−561.
(23) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77,
■ AUTHOR INFORMATION 3865−3868.
Corresponding Author (24) Mauersberger, P.; Huber, F. Acta Crystallogr., Sect. B: Struct. Sci
[email protected] 1980, 36, 683−684.
(25) Shannon, R. D. Acta Crystallogr. A 1976, 32, 751−767.
Notes (26) Thiele, G.; Serr, B. R. Z. Kristallogr. 1995, 210, 64−64.
The authors declare no competing financial interest. (27) Takahashi, Y.; Obara, R.; Lin, Z.-Z.; Takahashi, Y.; Naito, T.;
⊥
Deceased Inabe, T.; Ishibashi, S.; Terakura, K. Dalton Trans. 2011, 40, 5563−
5568.
ACKNOWLEDGMENTS (28) Tudela, D.; J. Sanchez-Herencia, A.; Diaz, M.; Fernandez-Ruiz,
■
This paper is dedicated to the memory of Jung-Hwan Song.
R.; Menendez, N.; D. Tornero, J. J. Chem. Soc., Dalton Trans. 1999,
4019−4023.
Financial support was provided by the National Science (29) Bockris, J. O. M.; Reddy, A. K. N.; Gamboa-Aldeco, M. E.
Foundation (Grant DMR-1104965). Use of the Advanced Modern electrochemistry: Fundamentals of electrodics, Vol. 2, 2nd ed.;
Photon Source was supported by the U.S. Department of Springer: New York, 2000.
Energy, Office of Science, Office of Basic Energy Sciences, (30) Frangioni, J. V. Curr. Opin. Chem. Biol. 2003, 7, 626−634.
under Contract Nos. DE-AC02-06CH11357 and DE-FG02- (31) Birch, F. Phys. Rev. 1947, 71, 809−824.
88ER45372. (32) González-Díaz, M.; Rodríguez-Hernández, P.; Muñoz, A. Phys.
Rev. B 1997, 55, 14043−14046.
REFERENCES (33) Stampfl, C.; Mannstadt, W.; Asahi, R.; Freeman, A. J. Phys. Rev.
■(1) Yamada, K.; Funabiki, S.; Horimoto, H.; Matsui, T.; Okuda, T.;
B 2001, 63, 155106.
(34) Chang, Y. H.; Park, C. H.; Matsuishi, K. J. Korean Phys. Soc.
Ichiba, S. Chem. Lett. 1991, 20, 801−804. 2004, 44, 889−893.
(2) Scaife, D. E.; Weller, P. F.; Fisher, W. G. J. Solid State Chem. (35) Chung, I.; Lee, B. H.; He, J.; Chang, R; Kanatzidis, M. G. Nature
1974, 9, 308−314. 2012, DOI: 10.1038/nature11067.
(3) Yamada, K.; Matsui, T.; Tsuritani, T.; Okuda, T. Z. Naturforsch.,
A: Phys. Sci. 1990, 45, 307−312.
(4) Shum, K.; Chen, Z.; Qureshi, J.; Yu, C.; Wang, J. J.; Pfenninger,
W.; Vockic, N.; Midgley, J.; Kenney, J. T. Appl. Phys. Lett. 2010, 96,
221903.
(5) Mitzi, D. B.; Wang, S.; Feild, C. A.; Chess, C. A.; Guloy, A. M.
Science 1995, 267, 1473−1476. Mitzi, D. B.; Feild, C. A.; Harrison, W.
T. A.; Guloy, A. M. Nature 1994, 369, 467−469.
(6) Kojima, A.; Teshima, K.; Shirai, Y.; Miyasaka, T. J. Am. Chem. Soc.
2009, 131, 6050−6051.
(7) Borriello, I.; Cantele, G.; Ninno, D. Phys. Rev. B 2008, 77,
235214.
(8) Romani, A.; Clementi, C.; Miliani, C.; Favaro, G. Acc. Chem. Res.
2010, 43, 837−846.
(9) Kuriki, K.; Koike, Y.; Okamoto, Y. Chem. Rev. 2002, 102, 2347−
2356.
(10) Kido, J.; Okamoto, Y. Chem. Rev. 2002, 102, 2357−2368.
(11) Hemmila, I.; Webb, S. Drug Discovery Today 1997, 2, 373−381.

8587 dx.doi.org/10.1021/ja301539s | J. Am. Chem. Soc. 2012, 134, 8579−8587