You are on page 1of 13

Organic Electronics 37 (2016) 61e73

Contents lists available at ScienceDirect

Organic Electronics
journal homepage: www.elsevier.com/locate/orgel

A comprehensive theoretical study of halide perovskites ABX3


Jingyu Qian, Bin Xu**, Wenjing Tian*
State Key Laboratory of Supramolecular Structure and Materials, Jilin University, Changchun, 130012, PR China

a r t i c l e i n f o a b s t r a c t

Article history: Solar cells based on halide perovskites have recently been attractive due to their excellent power con-
Received 6 April 2016 version efficiency (PCE), lower cost and simple manufacture. Here, a series of halide perovskites (ABX3:
Received in revised form A ¼ CH3NH3, CH(NH2)2, Cs, Rb; B ¼ Pb, Sn, Ge; X ¼ I, Br, Cl, F) were investigated by Density Functional
18 May 2016
Theory (DFT) calculations, together with Shockley-Queisser Maximum Solar Cell Efficiency (S-Q) and
Accepted 31 May 2016
Spectroscopic Limited Maximum Efficiency (SLME) mathematical models. The results indicate that: the
electronic structure of germanium perovskites bears a close similarity to that of lead perovskites with a
small energy difference between the nonbonding orbital and antibonding orbitals, but with a large
Keywords:
Perovskites
energy difference comparing with that of tin perovskites (0.6e1.7 eV for CsGeI3 at Z point of the Brillouin
Theoretical study zone, 0.7e1.4 eV for CH3NH3PbI3 and 1.4e2.2 eV for CH3NH3SnI3 at R point of the Brillouin zone), which
Electrical is attributable to the atomic level, where the 4s orbital energy of Ge (11.5 eV) is close to the 6s orbital
Optical energy of Pb (11.6 eV), but the 5s orbital energy of Sn (10.1 eV) is significantly high. Therefore,
Lead free germanium perovskites possess as high absorption coefficient around solar spectrum as lead perovskites,
while tin perovskites only have low absorption coefficient, which makes the short-circuit current of
CsGeI3 and CH3NH3PbI3 (0.017 Acm2 and 0.016 Acm2, simulated by SLME with a 200 nm absorber
under AM1.5G) are higher than that of CH3NH3SnI3 (0.015 Acm2) even if the bandgap of CsGeI3 and
CH3NH3PbI3 (1.51 eV and 1.55 eV) are larger than that of CH3NH3SnI3 (1.21 eV). Meanwhile, the effective
mass of electrons and holes are approximate for germanium perovskites and lead perovskites (0.14:0.19
for CsGeI3 and 0.12:0.12 for CH3NH3PbI3), indicating a balanced electrons and holes transport, whereas
the electrons transport is much slower than the holes transport for tin perovskites due to the effective
mass of electron is much larger than that of hole (0.17:0.04 for CH3NH3SnI3). As a result, the PCE of CsGeI3
(27.9%) and CH3NH3PbI3 (26.7%) is higher than that of CH3NH3SnI3 (19.9%).
© 2016 Elsevier B.V. All rights reserved.

1. Introduction drastically at a high temperature [7] about 425 K or takes a slow


process at room temperature [8], where CH3NH3PbI3 decomposes
Since the first photovoltaic device based on organometal halide into a solid (PbI2) and two gases (CH3NH2 and HI). On the other
perovskites (CH3NH3PbI3) with a power conversion efficiency (PCE) hand, environmental problems drag on the further development of
of 3.8% was reported by Kojima et al. [1], CH3NH3PbI3 based pe- halide perovskites solar cells based on CH3NH3PbI3 for the practical
rovskites has attracted more and more attention due to their rapid employment in daily life due to the heavy element, lead.
increase in PCE, for example 9.7% in 2011 [2], 15.4% in 2013 [3], and In order to overcome these two difficulties, much effort has been
20.1% in 2015 [4], and 22.1% in 2016 [5], which approaches the PCE made over the last years. For example, in 2014, Feng et al. reported a
of polycrystal silicon solar cells [6]. Although great achievements first attempt by using CH3NH3SnI3 instead of CH3NH3PbI3 with an
for perovskites solar cells have been made in the last six years, there efficiency of 5.7% [9]. Noel et al. fabricated another CH3NH3SnI3
are still some issues should be addressed when considering their perovskite lead-free solar cell with a PCE of 6.4% [10]. Kumar et al.
future commercial applications. On one hand, CH3NH3PbI3 is rela- presented a photovoltaic device by using CsSnI3 as the active layer
tively not stable because a decomposition reaction occurs and achieved an efficiency of 2%, but the entire device needed to be
fabricated and tested inside the glovebox due to the tendency for Sn
to get oxidized [11]. Sabba et al. reported a composite perovskite
* Corresponding author. CsSnI2$9Br0.1 and achieved a PCE of 1.76% [12]. Besides, Huang et al.
** Corresponding author. achieved an efficiency of 3.2% based on a CsGeI3 perovkite solar cell
E-mail address: wjtian@jlu.edu.cn (W. Tian).

http://dx.doi.org/10.1016/j.orgel.2016.05.046
1566-1199/© 2016 Elsevier B.V. All rights reserved.
62 J. Qian et al. / Organic Electronics 37 (2016) 61e73

[13]. than 0.001 eV/Å. For the density of states (DOS) for each perovskite,
For the theoretical investigations, most scientists focused on the width of the gaussian smearing was 0.1 eV, with 6  6  6 and
CH3NH3PbI3 and paid more attention to exploring of its outstanding 12  12  12 k-point meshes centered at the G point. For the band
performance, such as the optical properties [14] and electrical structure of halide perovskites ABX3, 100 k-points between high
properties [15]. There are some theoretical investigations about symmetric points were used. Spin-orbit coupling (SOC) was also
other halide perovskites. For example, Stoumpos et al. prepared a taken into account for both DOS and band structure, and all of the
CsGeI3 crystal and investigated its band structure with DFT/LDA parameters were in parallel to what has been mentioned above
method. In the meanwhile they found out the large second har- except the choice of k-points, for which a 6  6  6 k-point mesh
monic generation response is attributed to the high density of and 20 k-points between high symmetric points were chosen,
states in valence band due to sp-hybridization of the Ge and I or- respectively. To account for the underestimation of the band gap in
bitals [16]. Saparov et al. synthesized a lead-free layered perovskite standard DFT calculations, the accurate electronic structures were
Cs3Sb2I9 and calculated its electronic structures with DFT/GGA then calculated by using the HeydScuseria Ernzerhof (HSE06)
method. It showed that Cs3Sb2I9 possesses a nearly direct band gap hybrid functional [22], in which the cutoff energy was improved to
(less than 0.02 eV difference between the direct and indirect band 800 eV and the exact HartreeFock (HF) exchange contribution for
gaps) and a similar high-level of absorption compared to the hybrid functional was tuned to 50% and 75% to match experi-
CH3NH3PbI3 [17]. However, there is almost no report on the theo- mental values, as discussed later. Limited by computing power, only
retical study of the comprehensive and systematic analysis of the high symmetric points were calculated by HSE06.
halide perovskites as the active materials for solar cells. In CASTEP, Electronion interactions were described by the ul-
In this paper, we investigated a series of halide perovskites trasoft pseudopotential [23]. Except for GGA/PBE, the local-density
ABX3, where A can be CH3NH3, CH(NH2)2, Cs, or Rb; B and X belong approximation (LDA) was used for the electron exchange-
to the same main group of lead and iodine in the periodic table of correction functional, too. The cutoff energy, k-point mesh, SCF
the elements respectively (B ¼ Pb, Sn, Ge; X ¼ I, Br, Cl, F), by using tolerance and the residual force for per atom were the default,
DFT calculations, accompanying with Shockley-Queisser Maximum which was automatically set by package when the quality of
Solar Cell Efficiency (S-Q) and Spectroscopic Limited Maximum calculation was set as ‘ultra-fine’.
Efficiency (SLME) mathematical models. A systematical investiga- The charge distributions of Csþ, CH3NH3þ and CH(NH2)2þ were
tion of the optical and electrical properties indicated that the calculated at the DFT level using B3LYP hybrid functional [24] and
electronic structure of germanium perovskites bears a close simi- SDD basis set, as implemented in the Gaussian 09 package [25],
larity to that of lead perovskites with a small energy difference which was also used to describe the molecular orbital of CH3NH3þ,
between the nonbonding orbital and antibonding orbital, but with CH(NH2)2þ, PbI64 and Cs4PbI6.
a large energy difference comparing with that of tin perovskites
(0.6e1.7 eV for CsGeI3 at Z point of the Brillouin zone, 0.7e1.4 eV for 2.2. Semiempirical calculations
CH3NH3PbI3 and 1.4e2.2 eV for CH3NH3SnI3 at R point of the Bril-
louin zone). The germanium perovskites possess as high absorption The experimental optical band gap of halide perovskites ABX3
coefficient around solar spectrum as lead perovskites, while tin was taken into calculation for the absorption spectrum, with the
perovskites only have a low absorption coefficient, which makes use of scissor operator [26]. We calculated the dielectric frequency
the short-circuit current of CsGeI3 and CH3NH3PbI3 (0.017 Acm2 spectra firstly with the cutoff energy of 500 eV and a 12  12  12 k-
and 0.016 Acm2, simulated by SLME with a 200 nm absorber under point mesh, and the imaginary part was determined by a summa-
AM1.5G) higher than that of CH3NH3SnI3 (0.015 Acm2) although tion over empty states in the following equation [27]:
the bandgap of CsGeI3 and CH3NH3PbI3 (1.51 eV and 1.55 eV) are
 2
larger than that of CH3NH3SnI3 (1.21 eV). The effective mass of 4pe
ε2 ðuÞ ¼ j〈vjpjc〉j2 rcv ðZuÞ½f ðEv Þ  f ðEc Þ
electrons and holes are approximate for germanium perovskites me u
and lead perovskites (0.14:0.19 for CsGeI3 and 0.12:0.12 for
CH3NH3PbI3), indicating a balanced electrons and holes transport, Where v represents valence bands and c represents conduction
whereas the electrons transport is much slower than the holes bands. The real part could be obtained by the Kramers-Kronig
transport because tin perovskites due to the effective mass of relation:
electron is much larger than that of the holes (0.17:0.04 for
Z∞
CH3NH3SnI3). As a result, the PCE of CsGeI3 (27.9%) and CH3NH3PbI3 2 u0 ε2 ðu0 Þ
(26.7%) is higher than that of CH3NH3SnI3 (19.9%). ε1 ðuÞ ¼ P du
p u0 2  u2
0
2. Computational methods
In which P denotes the principal value. The frequency depen-
dent absorption coefficient was then given by:
2.1. First-principles calculations
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
uε ðuÞ þ ε2 ðuÞ þ ε2 ðuÞ
The crystal structure optimization and electronic structure cal- t 1 1 2
culations of halide perovskites ABX3 were performed by using the aðuÞ ¼ u
2
Vienna Ab-initio Simulation Package (VASP) [18] and the Cam-
bridge Sequential Total Energy Package (CASTEP) [19].
In VASP, Electronion interactions were described by the
projector-augmented wave (PAW) potentials [20] and the gener- 2.3. Shockley-Queisser Maximum Solar Cell Efficiency (S-Q) [28].
alized gradient approximation (GGA) parametrized by Per-
dewBurkeErnzerhof (PBE) [21] was used for the electron The standard AM1.5G and AM0 solar spectrum were used to
exchange-correction function. A cutoff energy of 500 eV was calculate the efficiency. The IeV curve under illumination was
used. The convergence threshold for self-consistent-field (SCF) described as I ¼ Ish þ I0 ½1  expðqV=kTc Þ, where Ish is the short
iteration was set at 106 eV, and the atomic positions were fully circuit current, I0 is the reverse saturation current, V is the voltage
optimized until all components of the residual forces were smaller between electrodes, Tc is the temperature of the solar cell, k is
J. Qian et al. / Organic Electronics 37 (2016) 61e73 63

Boltzmann constant, and q is electronic charge. Ish was calculated CH3NH3SnBr3, CH(NH2)2PbI3 and CH(NH2)2SnI3) as the investigated
R∞
by Ish ¼ q Eg Isun ðEÞdE, where Eg is the bandgap of material, and targets.
Isun(E) the sunlight photon flux at energy E. I0 was calculated by In order to get the initial structures, we collected the experi-
R∞
I0 ¼ q Eg Ibb ðE; Tc ÞdE, where Ibb(E,Tc) is the black-body spectrum at mental lattice constant of the existent halide perovskites at room
temperature Tc. temperature or near room temperature as listed in Table S1. When
A part of perovskite (ABX3) is an inorganic atom, most of the pe-
2.4. Spectroscopic Limited Maximum Efficiency (SLME) [29]. rovskites like CsSnI3, CsSnBr3, CsSnCl3, CsPbBr3, CsPbCl3, RbPbF3,
RbGeI3 and CsPbF3 (Fig. 1b) possess the symmetry of m3m and the
Developed from S-Q, the SLME method takes the photon ab- smallest repeat unit is a 1  1  1 cell, while CsGeI3, CsGeBr3 and
sorptivity aabs ðEÞ into account. In this case, CsGeCl3 (Fig. 1c) have the space group of R3m and the smallest
R∞ R∞ repeat unit is a 1  1  1 cell. When A part is organic ions
Ish ¼ q 0 aabs ðEÞIsun ðEÞdE and I0 ¼ q 0 aabs ðEÞIbb ðE; Tc ÞdE.
(CH(NH2)2 and CH3NH3), the crystal structures of perovskite (ABX3)
obtained from x-ray diffraction (XRD) only confirm the frame
3. Results and discussions
formed by B and X, so CH3NH3PbCl3, CH3NH3PbBr3, CH3NH3SnBr3,
CH3NH3SnI3 and CH3NH3PbI3 at high temperature (Fig. 1f) have a
3.1. Crystal structures
symmetry of m3m and the smallest repeat unit is a 1  1  1 cell
[41]. However, the organic ions (CH(NH2)2 and CH3NH3) arranged
The perovskite ABX3 is composed of three elements. There are
disorderedly in the lattice of perovskite, which has to be relaxed to
two requirements for the formation of ABX3 perovskite. On one
the fixed orientations during calculations. For perovskites with
hand, the valence of anions must be balanced with that of cations,
CH3NH3 (CH3NH3PbCl3, CH3NH3PbBr3, CH3NH3SnBr3, CH3NH3SnI3
i.e., 3N(X)þN(B)þN(A) ¼ 0, where N(x) represents the valence of
and CH3NH3PbI3 at high temperature), CeN bond should be aligned
ion x. On the other hand, the ionic radia of A, B and X need to meet
approximately with [111] direction (Fig. 1e) in order to reach the
the condition of the tolerance factor t [30]: 0.75 < t < 1, where
lowest energy [42]. For CH3NH3PbI3 at room temperature, the
t ¼ pffiffiRffi A þRB , and Rx is the ionic radius. So we selected B (þ2) and X frame formed by lead and iodinep possesses
2ðRB þRX Þ
ffiffiffi pffiffiffi a space group of I4/mcm
(1) from the same main group of lead (þ2) and iodine (1) in the and the smallest repeat unit is a 2  2  2 supercell, so the CeN
periodic table of the element (Ge (2þ) and Sn (2þ) for B; F (1), Cl bond should be aligned approximately along [001] direction
(1), Br (1) and I (1) for X), respectively, and A was selected from (Fig. 1h) to reach the lowest energy [43]. For CH(NH2)2PbI3, the
the first main group (Na (þ1), K (þ1), Rb (þ1) and Cs (þ1)). Then we frame formed by lead and iodine possesses
pffiffiffi paffiffiffispace
pffiffiffi group of P3m3
calculated the tolerance factor t of a series of perovskites composed and the smallest repeat unit is a 2  2  3 supercell. For
of elements we selected (Fig. S1a). From Fig. S1a, theoretically, we CH(NH2)2SnI3, the frame formed by tin and iodine possesses pffiffiffi apspace
ffiffiffi
would expect the existence of all perovskite ABX3 based on the group of Imm2 and the smallest repeat unit is a 2 2  2 2  2
elements we chose. However, only CsSnI3, CsGeI3, CsPbBr3, CsSnBr3, supercell (Fig. 1g). Different perovskites possess different repeat
CsGeBr3, CsPbCl3, CsSnCl3, CsGeCl3, CsPbF3, CsSnF3, CsGeF3, RbGeI3 units and crystal structures. Therefore, for the purposes of simpli-
and RbPbF3 exist in practice [12,31e38], and [39], because the fication and comparison, in the further calculations, an initial
tolerance factor t might be more tight (0.806 < t < 1) for the system structure as shown in Fig. 1e was applied for CH3NH3PbCl3,
of halide perovskites. CH3NH3PbBr3, CH3NH3SnBr3, CH3NH3SnI3 and CH3NH3PbI3, and the
According to the requirements for the formation of ABX3 smallest repeat unit is a 1  1  1 cell; an initial structure in Fig. 1d
perovskite, some organic ions, such as CH3NH3 (þ1) and CH(NH2)2 for CH(NH2)2PbI3 and CH(NH2)2SnI3, and the smallest repeat unit is
(þ1), carrying one positive charge and possessing a similar size as a 1  1  1 cell; an initial structure in Fig. 1b for CsSnI3, CsPbBr3,
Cs (þ1) can also serve as A part of perovskites so as to form CsSnBr3, CsPbCl3, CsSnCl3, CsPbF3, CsSnF3, CsGeF3, RbGeI3 and
CH3NH3PbI3, CH3NH3SnI3, CH3NH3PbBr3, CH3NH3PbCl3, RbPbF3 and the smallest repeat unit is a 1  1  1 cell; and an initial
CH3NH3SnBr3, CH(NH2)2PbI3 and CH(NH2)2SnI3. structure in Fig. 1c for CsGeI3, CsGeBr3, CsGeCl3 and the smallest
In fact, the implication of the tolerance factor equation is that repeat unit is a 1  1  1 cell.
the ionic radius of A should not be too large or too small (when A is In order to calculate the electronic structures, the optical and
too large or too small, B and X will self-assemble and unable to form electrical properties of halide perovskites, we optimized the atomic
a 3D network) so that the frame made up of B and X should be positions and lattice parameters by using GGA/PBE method with
stable (Fig. 1m). For example, Csþ with an ionic size of 1.67 Å, is not Vasp program, GGA/PBE method and LDA method with Castep
large enough to prop the frame built by Pb and I, so Pb and I in program, respectively. The optimized lattice parameters together
CsPbI3 form a 1D structure by self-assembly as shown in Fig. 1a. We with the experimental parameters are shown in Fig. 2. From Fig. 2,
estimated the equivalent ionic size of CH3NH3þ and CH(NH2)2þ at we found that the lattice parameters of perovskites increase obvi-
2e2.2 Å and 2.2e2.4 Å respectively (Fig. S1b). The size of ously when X part changes from F, Cl, Br to I, but increase slightly
CH(NH2)2þ is too big and reaches the critical point due to observing when B part changes from Ge, Sn to Pb. All the theoretical lattice
both 3D structure (Fig. 1j) and D structure (Fig. 1k) at room tem- parameters of perovskites calculated by GGA/PBE (Vasp) are almost
perature. The size of CH3NH3þ falls in between Csþ and larger than the experimental lattice constant [44], except for the
CH(NH2)2þ, which makes CH3NH3PbI3 has a twisty 3D structure perovskites (ABX3, B]Ge) because Ge perovskite possesses a twisty
(Fig. 1i) at room temperature (at 162.2e327.4 K, CH3NH3þ is unable 3D structure, and the bond length of Ge-X is not corresponding to
to support the frame built by Pb and I fully) and a neat 3D structure the lattice parameter.
(Fig. 1f) at high temperature (at >327.4 K, CH3NH3þ is able to By comparing the lattice parameters calculated by GGA/PBE
support the frame built by Pb and I). Compared with 3D perov- (Vasp), GGA/PBE (Castep) and LDA (Castep) and the experimental
skites, 1D perovskites have a larger bandgap and higher probability parameters, the relative errors between theory and experiment
of electron-hole recombination, which are not favorable to serve as were obtained as listed in Table S1. The relative errors between the
the active materials for solar cells [40]. So in this paper we only lattice parameters optimized by GGA/PBE (Castep) and the exper-
chose 3D perovskites (CsSnI3, CsGeI3, CsPbBr3, CsSnBr3, CsGeBr3, imental parameters are 0.9%e5.0% for CsGeI3, CsPbBr3, CsSnBr3,
CsPbCl3, CsSnCl3, CsGeCl3, CsPbF3, CsSnF3, CsGeF3, RbGeI3, RbPbF3, CsPbCl3, CsSnCl3, CsGeCl3, CsPbF3, CsSnF3, CsGeF3, RbGeI3, RbPbF3,
CH3NH3PbI3, CH3NH3SnI3, CH3NH3PbBr3, CH3NH3PbCl3, CH3NH3PbCl3, CH3NH3PbBr3, CH3NH3SnBr3, CH3NH3SnI3,
64 J. Qian et al. / Organic Electronics 37 (2016) 61e73

Fig. 1. Ball-and-stick representations (a) CsPbI3 (space group: Pbnm), (b) CsSnI3 (space group: Pm3m), (c) CsGeI3 (space group: R3m), (d) CH(NH2)2PbI3 (space group: P1), (e)
CH3NH3PbI3 (space group: P1), (f) CH3NH3PbI3 (space group: Pm3m), (g) CH(NH2)2SnI3 (space group: Imm2), (h) CH3NH3PbI3 (space group: P1), (i) CH3NH3PbI3 (space group: I4/
mcm), (j) CH(NH2)2PbI3 (space group: P3m1) and (k) CH(NH2)2PbI3 (space group: P3). The sectional views of (l) CH(NH2)2PbI3 and (n) CH3NH3PbI3. The relationship of A, B and X in
perovskites (ABX3) (m).

calculations. So the electronic structures, the optical and electrical


properties are calculated by GGA/PBE (Vasp) in this study.

3.2. Electronic structures

3.2.1. Density of states (DOS) and band structures


Based on the relaxed structures, we calculated the electronic
structures of the above halide perovskites ABX3, which are
commonly described by the density of states (DOS) and band
structures.
The typical DOS of CsPbF3, CsPbCl3, CsPbBr3, CH3NH3PbBr3,
CH3NH3PbI3, CH3NH3SnI3, CH(NH2) 2SnI3, CsGeI3, including the
similarities and differences, as shown in Fig. 3.
From the view point of similarities, the formation and the order
of bands are analogous because of the similar framework structures
formed by B and X and chemical constitutions of these halide pe-
rovskites. The valence band maximum (VBM) is derived from the p
orbitals of X and s orbitals of B, and their overlapping indicates the
Fig. 2. Relationship between the experimental lattice constant and theoretical lattice
significant hybridization with each other. The p orbitals of B and a
constant. The theoretical lattice constants of perovskites were calculated by using Vasp bit of p orbitals of X constitute the conduction band minimum
with GGA/PBE (green points), Castep with GGA/PBE (red points) and Castep with LDA (CBM). Below VBM, a mass of states derived from the p orbitals of B
(blue points). (For interpretation of the references to colour in this figure legend, the and d orbitals of X, which are situated in the same energy level.
reader is referred to the web version of this article.)
Within these p orbitals of B, most of them are nonbonding orbitals
(VBM-1) and the rest form the conventional coordinate bonds
CH3NH3PbI3, CH(NH2)2PbI3 and CH(NH2)2SnI3, and 11.3% for CsSnI3, (X/B) in terms of the classical valence bond theory, but ionic-
-2.0% for CsGeCl3 and -1.2% for CsGeBr3. However, the relative er- covalent bonds and coordinate bonds are not distinguishable
rors between the optimized results obtained by LDA (Castep) and here. For example, there are six Br atoms around one Pb atom with
experiment values are 0.1% ~ 3.4% for CsSnI3, CsGeI3, CsPbBr3, the same distance between Pb and Br (2.937 Å), which indicates
CsSnBr3, CsPbCl3, CsSnCl3, CsPbF3, CsSnF3, CsGeF3, RbGeI3, RbPbF3, homogeneous bonds rather than two ionic-covalent bonds
CH3NH3PbCl3, CH3NH3PbBr3, CH3NH3SnBr3, CH3NH3SnI3, together with four coordinate bonds.
CH3NH3PbI3, CH(NH2)2PbI3 and CH(NH2)2SnI3, and -6.4% for From the perspective of differences, the PDOS of part A (Cs,
CsGeCl3 and -5.6% for CsGeBr3. The relative errors of the theoretical CH3NH3, and CH(NH2)2) and the energy differences among bands
lattice parameters calculated by GGA/PBE (Vasp) are 1.7%e4.3% (for example: △ECBM, VBM, △EVBM, VBM-1) are various. Firstly, for
for CsSnI3, CsGeI3, CsPbBr3, CsSnBr3, CsGeBr3, CsPbCl3, CsSnCl3, part A of perovskite (ABX3), different compositions have different
CsGeCl3, CsPbF3, CsSnF3, CsGeF3, RbGeI3, RbPbF3, CH3NH3PbCl3, PDOS. The energy of the p orbitals of Cs is far below its VBM
CH3NH3PbBr3, CH3NH3SnBr3, CH3NH3SnI3, CH3NH3PbI3, (Fig. 3g). When it comes to the PDOS of CH3NH3 (Fig. 3e), we can see
CH(NH2)2PbI3 and CH(NH2)2SnI3, which are tolerated for further that sharp peaks belong to the s orbitals of hydrogen and p orbitals
of carbon and nitrogen, which is the localized covalent bond (CeH
J. Qian et al. / Organic Electronics 37 (2016) 61e73 65

Fig. 3. Density of states (DOS) and partial densities of states (PDOS) (a) CsPbF3, (b) CsPbCl3 (c) CsPbBr3 (d) CH3NH3PbI3 (e) CH3NH3SnI3 (f) CH(NH2) 2SnI3 (g) CsGeI3 (h) RbGeI3
calculated using PBE functions. The energy of valence band maximum is set to zero.

and NeH). CH(NH2)2 is also different from CH3NH3, whose LUMO, the relationship of energy difference between bands is △E(Pb)VBM,
HOMO and HOMO-1 are derived from the p orbitals of carbon and VBM-1 z△E(Ge)VBM, VBM-1 < △E(Sn)VBM, VBM-1. Therefore, the
nitrogen (Fig. 3f). However, from PDOS and DOS, part A does not electronic structure of Ge perovskites is found to be abnormal.
seem to play a direct role in determining the basic electronic In order to get the detailed information about the electronic
structures of frameworks formed by B and X and the important role structures, we calculated the band structure of halide perovskites
of A is donating an electron to the B-X frame and supporting the ABX3. Different high symmetrical points of the Brillouin zone (BZ)
perovskite structure. So, we ignore the effect of different part A and paths between the high symmetrical points are chosen ac-
when we discussed DOS. Secondly, the energy difference between cording to the symmetry provided by XRD data. For the crystal with
ECBM and EVBM (△ECBM, VBM) and the energy difference between the space groups of Pm3m, we described the band structure along
EVBM and EVBM-1 (△EVBM, VBM-1) are various. After comparing the the path X (0.5 0 0) / R (0.5 0.5 0.5) / M (0.5 0.5 0) / G (0
bandgap of CsPbF3 (2.55 eV), CsPbCl3 (2.10 eV), CsPbBr3 (1.66 eV) 0 0) / R (0.5 0.5 0.5), such as CsPbBr3 (Fig. 4c). For the crystal with
and CH3NH3PbI3 (1.53 eV) (Fig. 3) with the same Pb as the part B of a small organic molecule (possess Pm3m space group approxi-
perovskite (ABX3), the band gap becomes smaller when X part mately), we described the band structure along the path X (0.5
changes from F, Cl, Br to I, and the energy difference between VBM 0 0) / R (0.5 0.5 0.5) / M (0.5 0.5 0) / G (0 0 0) / R (0.5 0.5 0.5),
and VBM-1 becomes smaller (~2.1 eV for CsPbF3, ~1.0 eV for such as CH3NH3PbBr3. In addition, we chose the path G (0 0 0) / Z
CsPbCl3, ~0.8 eV for CsPbBr3, ~0.5 eV for CH3NH3PbI3). However, (0.5 0.5 0.5) / F (0.5 0.5 0) / G (0 0 0) to describe the band
when we choose the same I as part X (CsGeI3, CH3NH3PbI3 and structure of the crystal with R3m space groups, like CeGeI3 (Fig. 4f).
CH3NH3SnI3), we find the band gap for CsGeI3, CH3NH3PbI3 and From the view of similarities, we find all of them are the direct-
CH3NH3SnI3 is 1.07 eV, 1.36 eV and 0.46 eV, respectively, and the band-gap materials, which are different from Si, a typical indirect
energy difference between VBM and VBM-1 (~0.5 eV for CsGeI3, bandgap material whose the electronic transition from VBM to CBM
~0.5 eV for CH3NH3PbI3, ~0.8 eV for CH3NH3SnI3) shows the rela- needs the help of phonon [45]. As for DOS discussed above, we find
tionship: △E(Pb)VBM, VBM-1 z△E(Ge)VBM, VBM-1 < △E(Sn)VBM, VBM- the similarity of these halide perovskites in band structure and
1. In the periodic table of elements, the atomic number decreases steep VBM and CBM, which means good characteristics of the
from I, Br, Cl to F in VIIA and the energy difference of CsPbF3, electrons delocalization. In fact, a physical quantity related to the
CsPbCl3, CsPbBr3 and CH3NH3PbI3 between bands becomes smaller; electrons delocalization is the effective mass (m*) of electrons
while when the atomic number decreases from Pb, Sn to Ge in IVA, (holes), which will be discussed later. Unlike VBM and CBM, the
66 J. Qian et al. / Organic Electronics 37 (2016) 61e73

Fig. 4. Band structure of (a) CsPbF3 (b) CsPbCl3 (c) CsPbBr3 (d) CH3NH3PbI3 (e) CH3NH3SnI3 (f) CsGeI3 (g) CH3NH3PbBr3 calculated using PBE functionals with and without SOC (grey
dot line) and the energy of valence band maximum is set to zero, where red line represents CBM, purple line represents VBM and blue line represents other band. Charge dis-
tribution (h) of CH(NH2)þ þ þ
2 (left) CH3NH3 (middle) Cs (right), where blue represents positive charge and red represents neutral. Schematic diagram of the dipole polarization of
organic molecules (i). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

bands below VBM seem to be flat and reflect the nonbonding


properties of halide perovskites. b j¼E j
H (1)
0 0
In terms of differences, the energy difference among bands is
various. From CsPbF3 (Fig. 4a), CsPbCl3 (Fig. 4b), CsPbBr3 (Fig. 4c) to And Schrodinger’s equation with SOC is expressed as follows
CH3NH3PbI3 (Fig. 4d), we can see that the energy differences be- [47]:
tween VBM and VBM-1 (including four orbitals) at R point are  
_ _
different (3.7e4.2 eV for CsPbF3, 1.7e2.3 eV for CsPbCl3, 1.3e2.0 eV H0 þ H SO j ¼ ðE0  ms Bl Þj (2)
for CsPbBr3, 0.7e1.4 eV for CH3NH3PbI3). However, compared the
band structure of CsGeI3 and CH3NH3PbI3 to that of CH3NH3SnI3,
Where
CH3NH3SnI3 is rather different due to the larger energy difference R
E0  ms  Bl ¼ E0 þ 12  m80 Ze Z
2 2

between VBM and VBM-1 (0.6e1.7 eV for CsGeI3 at Z point of the pm2
 j* r13 jdt  ½jðj þ 1Þ  lðl þ 1Þ 
e

Brillouin zone, 0.7e1.4 eV for CH3NH3PbI3 and 1.4e2.2 eV for sðs þ 1Þ . The energy level (E0) defined by n and l splits into sub-
R
CH3NH3SnI3 at R point of the Brillouin zone). The relationship of the (E0  12  m80 Ze Z
2 2
levels pm2e
 j* r13 jdt  ðl þ 1Þ ¼ E0  DSOC and
energy difference between bands (F > Cl > Br > I, Pb z Ge < Sn)
m0 Ze2 Z2 R * 1
calculated by band structures is in accordance with that calculated E0 þ 2  8pm2  j r3 jdt  l) with different total moments, j ¼ lþ1/
1
e
according to the peaks of DOS. 2 and j ¼ l-1/2.

3.2.2. Spin-orbit coupling (SOC) Z4


SOC effects, which make both conduction and valence bands For a hydrogenlike atom; DSOC ∝ (3)
n3
near band edges more dispersive [46], need to be taken into
consideration when discussing the electronic structure of materials Where Z is the atomic number and n is the main quantum number.
containing element with a large atomic number. In these halide perovskites, CBM is three-fold degeneration if
Schrodinger’s equation without considering SOC is expressed as SOC is neglected, as shown in Fig. 4c. When the SOC effect is
follows: considered, CBM splits into a two-fold low-lying p1/2 manifold and
J. Qian et al. / Organic Electronics 37 (2016) 61e73 67

a four-fold high-lying p3/2 manifold if the spin degeneracy is exchange-correlation energy: △XC) caused by the failure of
counted. This splitting can induce a down-shift of CBM (△soc). describing excited state [44], which leads to the famous problem of
According to the analyses of DOS, the CBM of perovskite (ABX3) is underestimating bandgap of semiconductor, as shown in Fig. S3,
derived from the p orbitals of part B mainly, and we find the Z4/n3 and people set the error into the part of exchange-correlation en-
values of Ge (Z ¼ 32, n ¼ 4), Sn (Z ¼ 50, n ¼ 5) and Pb (Z ¼ 82, n ¼ 6) ergy (Exc) of GGA-DFT and LDA-DFT. Comparing △XC with
are 16,384, 50,000 and 2,09,315. We divide the Z4/n3 values of Ge, △XC þ△SOC, we can see the band gap of perovskites with Pb as part
Sn and Pb by 209315, and the ratio value of Z4/n3 s becomes 0.078: B calculated by PBE is in good agreement with experiments just
0.239: 1, which is in parallel the ratio of △soc for Ge perovskites because the losing sight of SOC effect offsets the underestimated
(0.04 eVe0.13 eV): △soc for Sn perovskites (0.29 eVe0.39 eV): exchange-correlation energy, △XC. Compared the bandgap value of
△soc for Pb perovskites (0.96 eVe1.62 eV) as listed in Table 1. It perovskites calculated by PBE and SOC, we can see that with the
might be a coincidence, but eq. (3) reflects the nature of SOC, i.e., same part B, the band gap of halide perovskites increases from I, Br,
the relativistic effect. When an atom has more protons, the Cl to F (0.37 eV for CH3NH3PbI3, 0.64 eV for CsPbBr3, 1.07 eV for
Coulomb force between atomic nucleus and electron, which is the CsPbCl3 and 1.59 eV for CsPbF3). However, compared with the
centripetal force when electrons revolve around the atomic nu- experimental bandgap (1.52 eV for CH3NH3PbI3, 2.25 eV for CsPbBr3
cleus, becomes stronger, resulting in the fast moving of electrons. and 3.00 eV for CsPbCl3), the sort of halogen has a deep influence on
When the speed of electrons is close to the speed of light, the the bandgap caused by the exchange-correlation energy. From
relativistic effect is more obvious, which is so called SOC. Table 1, we can see △XC of the halide perovskites grows from I, Br,
Comparing Ge, Sn and Pb perovskites, the SOC effect for Ge pe- Cl to F (for example, 1.15 eV for CH3NH3PbI3, 1.61 eV for CsPbBr3,
rovskites is weak, and that for Pb perovskites is strong, while Sn 1.93 eV for CsPbCl3 and 2.51 eV for CsPbF3).
perovskites is just in the middle due to the difference of part B. The core of DFT is the Kohn-Sham method [44]:
Perovskites containing organic ions (Fig. 4f and g) have different E ¼ EðNÞ þ ET ðrÞ þ EV ðrÞ þ EJ ðrÞ þ EXC ðrÞ, where r is electric den-
band structure with SOC, compared with perovskites containing sity, Exc(r) is the exchange-correlation energy which can be
inorganic ions (Fig. 4c). If the crystal has a center of inversion, each decomposed into exchange energy [Ex(r)] and correlation energy
band will preserve its spin degeneracy, like CsPbBr3. However, [Ec(r)]. To correct the system error of GGA-DFT and LDA-DFT, we
when we replace Csþ with CH3NH3þ, there is no center of inversion adopted HF/DFT method (HSE06), by adding a part of exact ex-
in CH3NH3PbBr3 and the p3/2 and p1/2 bands will additionally be change energy of HF into Exc:
split by spin-orbit interaction, removing the spin degeneracy HSE ¼ aESR ðm; rÞ þ ð1  aÞE PBE;SR ðm; rÞ þ E PBE;LR ðm; rÞ þ E PBE ðrÞ,
Exc x x x c
(Fig. S2). where a is the pecentage of HF exact exchange and m is the range-
The SOC has also been found to be important in other com- separation. The default pecentage of Ex in HSE06 is 25%, and we
pounds containing heavy 6p cations [48], which makes both con- adjusted it to 50% and 75% for different materials because perov-
duction and valence bands near band edges more dispersive and skites with smaller halogen radius need higher pecentage of EX.
the electrical conductivity better. By using HSE06, we can estimate the band gap of halogen
perovskite (ABX3). As a reference, the band gap of CsGeI3 (1.55 eV)
is calculated by HSE06, which is rather close to its experimental
3.2.3. Exchange-correlation energy
value (1.51 eV). The band gap of RbGeI3, whose absorption spec-
As shown in Table 1, we calculated the band gap of halide pe-
trum has never been reported and experimental band gap is still
rovskites by using PBE with and without SOC. The DFT approxi-
unknown, was speculated to be 0.85 eV.
mated by GGA or LDA has a system error (underestimated

Table 1
Band gaps of halide perovskites calculated by using PBE functionals with and without SOC and HSE06 functionals without SOC.

PBE PBE þ SOC △soc △xc HSE06 (25%HF) HSE06 (50%HF) HSE06 (75%HF) EXP.

CH3NH3SnI3 0.59 0.30 0.29 0.91 0.79 1.05 1.33 1.21a


CsSnI3 0.45 0.06 0.39 1.21 0.74 1.11 1.27b
RbGeI3 0.44 0.31 0.13 0.85
CH(NH2)2SnI3 0.65 0.33 0.32 1.08 1.00 1.40 1.41a
CH(NH2)2PbI3 1.72 0.10 1.62 1.38 2.15 1.48a
CsGeI3 1.07 0.99 0.08 0.52 1.55 1.51c
CH3NH3PbI3 1.49 0.37 1.12 1.15 1.83 1.52a
CsSnBr3 0.62 0.28 0.34 1.47 1.05 1.55 1.75b
CH3NH3SnBr3 0.77 0.47 0.30 1.68 1.11 1.52 1.98 2.15d
CH3NH3PbBr3 1.86 0.77 1.10 1.46 2.38 2.23e
CsPbBr3 1.75 0.64 1.11 1.61 2.35 2.25f
CsGeBr3 0.63 0.57 0.05 1.80 1.15 1.78 2.43 2.37g
CsSnCl3 0.98 0.64 0.33 2.26 1.50 2.09 2.73 2.90h
CsPbCl3 2.19 1.07 1.11 1.93 2.89 3.00i
CH3NH3PbCl3 2.30 1.19 1.11 1.92 2.93 3.63 3.11j
CsGeCl3 1.02 0.99 0.04 2.32 1.69 2.40 3.16 3.31g
RbPbF3 2.84 1.79 1.05 3.86
CsPbF3 2.55 1.59 0.96 3.54
a
Reported by Stoumpos et al. [33].
b
Reported by Sabba et al. [12].
c
Reported by Tang et al. [23].
d
Reported by Chiarella et al. [36].
e
Reported by Kulkarni et al. [49].
f
Reported by Stoumpos et al. [31].
g
Reported by Lin et al. [50].
h
Reported by Borriello et al. [51].
i
Reported by Yunakova et al. [32].
j
Reported by Baikie et al. [52].
68 J. Qian et al. / Organic Electronics 37 (2016) 61e73

3.2.4. Molecular orbital


In order to learn more about the electronic structures and
discuss the optical transition of perovskites, we took metal ionic
compounds like Cs4PbBr6 and PbBr64 as model compounds and
calculated their molecular orbitals as shown in Fig. 5. For PbBr64,
from the low energy to high energy, there are bonding orbitals
formed by the s orbitals of Pb and p orbitals of Br, nonbonding
orbitals formed by the p orbitals of Br (orbitals for coordinate bonds
are included.), antibonding orbitals formed by the s orbitals of Pb
and p orbitals of Br (Orbital1*), antibonding orbitals formed by the p
orbitals of Pb and p orbitals of Br (Orbital2*), and antibonding or-
bitals for coordinate bonds formed by the d orbitals of Pb and p
orbitals of Br. When we added four Cs atoms around PbBr64, the
assignment of the molecular orbitals changed due to the addition of
d orbitals of Cs insert between Orbital1* and Orbital2*.
According to the orbital energy of R point of BZ and PDOS, the Fig. 6. The formation of CsPbBr3 molecular orbitals. The energy of CsPbBr3 molecular
molecular orbital of CsPbBr3 was given in Fig. 6. The unit cell of orbitals based on the average orbital energy at special k point of BZ. Red line represents
CsPbBr3 has the same structure as that of CsSnI3 (Fig. 1b), where a occupied molecular orbital and blue line represents unoccupied molecular orbital.
Detail methods are described in Fig. S13. (For interpretation of the references to colour
Cs atom is located in the center of the unit cell, Br atoms on the edge
in this figure legend, the reader is referred to the web version of this article.)
of the frame and Pb atoms at the corner. As a result, the main bonds
are the ionic bonds formed between Cs and Br and the covalent
bonds formed between Pb and Br, while no bond is formed between 3.3. Absorption spectrum
Pb and Cs because of no direct contact with each other. The dis-
tribution of the orbital energy and orbital parity is rather important According to the transition matrix element and selection rules
for the absorption spectrum, and the situation is similar for [44], when hvn jpjcm i ¼ 0, transitions from the occupied state n (vn)
different perovskites. As an example, we only discuss CsPbBr3 here. to the unoccupied state m (cm) are forbidden; and when
According to the schematic diagram of the molecular orbitals hvn jpjcm is0, transitions from vn to cm are allowed. In other words,
formation of perovskites, different molecular orbital energies of allowed transitions require the parity of wave functions (occupied
perovskites boil down to different atomic orbital energies. We state and unoccupied state) to be opposite in a system with the
calculate the atomic orbital energy of F, Cl, Br, I, Ge, Sn and Pb as inversion symmetry. For example, if the wave function of the
shown in Fig. S4. The outmost p orbital energy of F, Cl, Br and I occupied state is odd function, the transitions to the unoccupied
increases from 11.1 eV, -8.4 eV, -7.7 eV to 6.8 eV. The outmost p state with an even wave function are allowed, but the transitions
orbital energy of Ge, Sn and Pb increases from 3.7 eV, -3.3 eV to are forbidden to the unoccupied state with odd wave function.
3.1 eV, and the outmost s orbital energy of Ge, Sn and Pb changes Taking CsPbBr3 as an example, if we set Pb as the center of sym-
from 11.6 eV, -10.2 eV to 11.5 eV. The change of the atomic metry, the wave function of the Antibonding Orbital1* (Ob1*)
orbital energies explains the changes of bandgaps (CsPbF3 formed by the s orbitals of Pb and p orbitals of Br is an even function
(2.55 eV), CsPbCl3 (2.10 eV), CsPbBr3 (1.66 eV) and CH3NH3PbI3 and that of the Antibonding Orbital2* (Ob2*) formed by the p or-
(1.53 eV)) as well as the changes of energy differences between bitals of Pb and p orbitals of Br is an odd function, so transitions
VBM and VBM-1 (~0.5 eV for CsGeI3, ~0.5 eV for CH3NH3PbI3, from Ob1* to Ob2* are allowed. There are so many Nonbonding
~0.8 eV for CH3NH3SnI3). orbitals (Non-Ob) located between Ob1* and Bonding orbital2 (Ob2),
where we can find both symmetrical and antisymmetric orbital
distributions, which means that the transitions from Non-Ob to
Ob2* are possible. The possibility of transitions from Bonding
orbital1 (Ob1) to Ob2* is existent, because the symmetry of Ob1 is
the opposite of Ob2*. While the transition from Ob2 to Ob2* is
impossible, since the wave functions of both Ob2 and Ob2* are odd
function.
In the perovskites we chose, the transitions from Ob1* to Ob2*
and from Non-Ob to Ob2* have been taken into consideration. As
shown in Fig. 7, the energy of transition from Ob1* to Ob2* de-
termines the value of the optical band gap and the energy of
transition from Non-Ob to Ob2* determines the intensity of molar
absorption coefficient within the scope of solar spectrum. A small
energy difference between the transition from Ob1* to Ob2* and the
transition from Non-Ob to Ob2* means a better absorption
spectrum.
The simulate absorption spectra of several representative pe-
rovskites have been shown in Fig. 8. By comparing Pb perovskites,
from CsPbCl3, CsPbBr3 to CH3NH3PbI3, the edge of the absorption
peak gives red-shift due to the various halogens. Comparing iodic
perovskites, we can see CsGeI3 and CH3NH3PbI3 have relatively
Fig. 5. Molecular orbitals of (a) PbBr64 and (b) Cs4PbBr6. Red line represents occupied
higher absorption coefficient than CH3NH3SnI3 in the range of solar
molecular orbital and blue line represents unoccupied molecular orbital. The energy of spectrum. When comparing the DOS of CH3NH3SnI3 (Fig. 3a) with
HOMO is set to zero. (For interpretation of the references to colour in this figure those of CH3NH3PbI3 (Fig. 3d) and CsGeI3 (Fig. 3c), we find the
legend, the reader is referred to the web version of this article.)
J. Qian et al. / Organic Electronics 37 (2016) 61e73 69

Fig. 7. Orbital energies of ABX3 with a small energy difference between Antiboding orbital1* and Nonbonding orbital (a) resulting in a good absorption spectrum fitting with solar
spectrum (b). Orbital energies of ABX3 with a large energy difference between Antiboding orbital1* and Nonbonding orbital (c) resulting in a poor absorption spectrum fitting with
solar spectrum (d). Blue dot line represents solar spectrum, green line represents transition 1 with gaussian smearing, pink line represents transition 2 with gaussian smearing, red
line represents transition 3 with gaussian smearing and grey line represents total absorption spectrum. (For interpretation of the references to colour in this figure legend, the
reader is referred to the web version of this article.)

3.4. Effective masses of electrons and holes

As the active material of solar cell, not only a good ability of


charge carrier transport is necessary but also a balanced electron
and hole transportation is rather important in order to avoid charge
accumulation [53]. In fact, benefitting from the 3D structure, the
carrier mobility of perovskites is relatively high. For example, the
experimental carrier mobility of both electrons and holes in poly-
crystalline CH3NH3PbI3 can reach to 10 cm2v1 s1 approximately
[54]. While CsSnI3 is p-type semiconductor with a hole carrier
mobility of 585 cm2v1 s1 [55]. In theory, the carrier mobility of
electrons (holes) is inversely proportional to the effective mass (m*)
of electrons (holes), which can be calculated by fitting the VCM
" #1
v2 εðkÞ
(CBM) to m* ¼ h2 v2 k
, where εðkÞ is the energy dispersion

relation function described by band structures, as shown in


Fig. 8. Simulate absorption spectrum of CsPbCl3 (green line), CsPbBr3 (red line), Table S2. We found out that the effective masses for electrons and
CH3NH3PbI3 (brown line), CH3NH3SnI3 (grey line) and CsGeI3 (blue line). AM1.5G is holes of perovskites (ABX3) with the same part A and part B become
drawn with a purple line. (For interpretation of the references to colour in this figure smaller from CsSnCl3 (0.28:0.09), CsSnBr3 (0.22:0.06) to CsSnI3
legend, the reader is referred to the web version of this article.) (0.17:0.04) and from CsPbF3 (0.45:0.26), CsPbCl3 (0.28:0.18) to
CsPbBr3 (0.22:0.14). The effective masses for holes of Sn perovskites
are smaller than those of Ge and Pb perovskites, resulting in the
energy difference between VBM and VBM-1 is ~0.5 eV for CsGeI3, higher hole mobility of Sn perovskites. Compared the effective
~0.5 eV for CH3NH3PbI3, and ~0.8 eV for CH3NH3SnI3, which means masses for electrons and holes of perovskites (ABX3) with the same
the n-Ob of iodine in CsGeI3 and CH3NH3PbI3 is closer to CBM than part B and part X, the difference of the effective masses for elec-
that in CH3NH3SnI3, so the intensity of the molar absorption co- trons and holes reduces when a polar CH3NH3þ replaces Csþ, like
efficients of CsGeI3 and CH3NH3PbI3 is greater than that of CsSnBr3 (0.22:0.06) and CH3NH3SnBr3 (0.23:0.07). While the
CH3NH3SnI3 within the scope of solar spectrum, fitting with the effective masses for electrons and holes are approximate for the
mechanism described in Fig. 7. This is one of the reasons why following halogen perovskites: CsGeI3 (0.14:0.19), CsPbBr3
CH3NH3SnI3 and CH3NH3PbI3 have the similar bandgap but (0.21:0.14), CH3NH3PbCl3 (0.19:0.19), CH3NH3PbBr3 (0.17:0.16),
different conversion efficiency. CH3NH3PbI3 (0.12:0.12) and CH(NH2) 2PbI3 (0.19:0.19).
70 J. Qian et al. / Organic Electronics 37 (2016) 61e73

3.5. PCE simulated by S-Q and SLME mathematic models near-earth orbit is stronger than that around the surface of the
earth (maximum 1.60 W m2 nm1 at 500 nm). According to S-Q
Shockley-Queisser Maximum Solar Cell Efficiency (S-Q) and mathematic model, the short-circuit current is described as
R∞
Spectroscopic Limited Maximum Efficiency (SLME) mathematical Ish ¼ q Eg Isun ðEÞdE, which means that a photon produces a pair of
models have been introduced to make a good understanding of the electron and hole and the short-circuit current is in proportion to
optical and electrical properties of halide perovskites. AM1.5G (the the light intensity. In addition, the open-circuit voltage depends on
solar spectrum around the surface of the earth where the solar the bandgap (from 0.5 eV to 4 eV) mainly and has little difference
energy might be used for daily life power supply), AM0 (the solar between AM1.5G and AM0 (from 0.3 eV to 3.5 eV). For the fill factor,
spectrum around the near-earth orbit where the solar energy might the same situation is obtained due to a perfect semiconductor
be used for artificial satellite power supply) and bandgaps serve as behavior assumption. When the band gap changes from 0.5 eV to
input parameters in S-Q, and AM1.5G, AM0 and absorption spectra 4 eV, the fill factor varies from 72% to 95%.
in SLME. Similar PCE of perovskites by using S-Q and SLME are In S-Q, the simulated PCE only depends on the band gap so that
show in Fig. 9. From Fig. 9, we can see that perovskites with the materials with the similar bandgap (for example, 1.51 eV,
different halogen gather together at the bottom, in the middle and 1.55 eV and 1.21 eV for CsGeI3, CH3NH3PbI3 and CH3NH3SnI3) have a
at the top of the curve of S-Q. From Cl, Br to I, the similar PCE travels near simulated PCE value (31.2%, 31.0% and 32.0% for CsGeI3,
from the foot of the ’hill’ (~5.0% (AM0), ~3.0% (AM1.5G)) to the CH3NH3PbI3 and CH3NH3SnI3 simulated by AM1.5G; 28.5%, 28.4%
mountainside (~17.5% (AM0), ~17.5% (AM1.5G)) to the ’mountain’ and 29.0% for CsGeI3, CH3NH3PbI3 and CH3NH3SnI3 simulated by
top (~27.5% (AM0), ~32.0% (AM1.5G)). The relative height of PCE AM0). In SLME, when the absorption spectra of materials are taken
simulated by AM1.5G and AM0 are polytropic and dependent on into consideration, materials with similar bandgap have obviously
the bandgap of halide perovskites due to the difference of AM1.5G different PCE. Furthermore, the simulated PCE depends on the
and AM0 (at ultraviolet region, the light intensity of AM1.5G is thickness of the active layer. According to Beer’s Law, lg(T/
lower than that of AM0 due to the absorption of O3). For details, the T0) ¼ Kabc, where T/T0 presents transmittance, Ka is the molar
photons at the ultraviolet region possess high energy but produce absorption coefficient, b is the absorption depth and c is the con-
the same output work (hvEg, where vEg is the band gap of material) centration. The percentage of photons which are absorbed by the
as photons in lower energy region. So, at 1e2 eV, PCE simulated by absorber depends on Ka and b, while c is a solid constant. A thicker
AM1.5G (from 30.1% to 32.5%e22.5%) is higher than PCE simulated absorber means a better absorbing ability. However, for a solar cell,
by AM0 (from 27.5% to 29.0%e22.5%), while, at 2e4 eV, PCE simu- the thickness of an absorber is constrained by the carrier diffusion
lated by AM1.5G (from 22.5% to 0%) is lower than PCE simulated by length. Once the thickness of the absorber goes beyond the carrier
AM0 (from 22.5% to 1.2%). Different from the PCE simulated by diffusion length, most of the carriers generated by photons may
AM1.5G and AM0, the short-circuit current density, open-circuit recombine and enormously reduce the conversion efficiency.
voltage and fill factor changes straightforward with the bandgap. Therefore there is an appropriate value of the thickness between
No matter the value of bandgap is large or small (from 0.5 eV to the absorption depth and the carrier diffusion length of the selected
4 eV), the short-circuit current density simulated by AM0 (from materials, which we set to 200 nm, a typical value in the experi-
0.060 Acm2 to 0 Acm2) is larger than that simulated by AM1.5G ment. By considering the absorption coefficient and thickness, the
(from 0.045 Acm2 to 0 Acm2) as shown in Fig. S5, because the PCE was calculated by using SLME (27.9%, 26.7% and 19.9% for
light intensity (maximum 2.20 W m2 nm1 at 490 nm) around the CsGeI3, CH3NH3PbI3 and CH3NH3SnI3 simulated by AM1.5G; 26.2%,
25.3% and 19.1% for CsGeI3, CH3NH3PbI3 and CH3NH3SnI3 simulated
by AM0). As shown in Fig. 9, the PCE simulated by SLME is lower
than that simulated by S-Q because there is no consideration of the
absorption in S-Q. For the short-circuit current density, open-circuit
voltage and fill factor of these materials (Table S3), CsGeI3
(0.017 Acm2 for JSC, 1.23 V for VOp and 89.9% for FF simulated by
AM1.5G; 0.022 Acm2 for JSC, 1.24 V for VOp and 90.0% for FF
simulated by AM0) and CH3NH3PbI3 (0.016 Acm2 for JSC, 1.27 V for
VOp and 90.1% for FF simulated by AM1.5G; 0.021 Acm2 for JSC,
1.28 V for VOp and 90.2% for FF simulated by AM0) possess higher
values than CH3NH3SnI3 (0.016 Acm2 for JSC, 1.02 V for VOp and
88.3% for FF simulated by AM1.5G; 0.020 Acm2 for JSC, 1.03 V for
VOp and 88.4% for FF simulated by AM0).
Putting all the theory calculations and simulations results and
experiments data together enables us to draw a clear overall picture
for the properties of the halide peroveskites. Based on the PCE of
halide peroveskites simulated by S-Q model, the results indicate
the PCE of perovskites (ABX3) increase from Cl to Br to I, because the
bandgap becomes narrower when X changes from Cl to Br to I. We
summarize the structure, optical properties, electrical properties,
as well as the PCE from the theoretical simulations and experi-
ments of iodic peroveskites in Table 2. From the results of Table 2,
Fig. 9. Solar spectra (top) and simulated PCE for halide peroveskites by using S-Q and we know that the structure of all iodic perovskites, including
SLME models (bottom). PCE for halide peroveskites simulated by using S-Q model with FAPbI3, MAPbI3, MASnI3, CsSnI3, FASnI3 and CsGeI3, possess 3D
AM0 are marked as blue points. PCE for halide peroveskites simulated by using S-Q framework, which benefits the charge carrier transport of these
model with AM1.5G are marked as red points. PCE for halide peroveskites simulated by perovskites.
using SLME model with AM0 are marked as blue triangles. PCE for halide peroveskites
simulated by using SLME model with AM1.5G are marked as red triangles. (For
In addition, except for CsSnI3, other iodic perovskites show the
interpretation of the references to colour in this figure legend, the reader is referred to polarity in their band structures which results in the internal
the web version of this article.) junction to aid electron-hole separation [59]. This is one of the
J. Qian et al. / Organic Electronics 37 (2016) 61e73 71

Table 2
A summarization the bandgap, transition behavior, absorption spectrum, effective mass, polarity of crystal and SOC of iodic perovskites.

FAPbI3c MAPbI3c MASnI3 CsSnI3 FASnI3 CsGeI3

Structure Framework 3D 3D 3D 3D 3D 3D
Polarity Yes Yes Yes No Yes Yes
Optical properties Bandgap Well Well Well Well Well Well
Transition Direct Direct Direct Direct Direct Direct
Absorption Well Well Poor Middle Middle Well
Electrical properties Effective mass of carriers Equal Equal Unequal Unequal Unequal Equal
Other properties SOC Strong Strong Middle Middle Middle Weak
PCEa Theoryb 26.4% 26.7% 19.9% 25.6% 26.9% 27.9%
Experiment 16.4% [56] 22.1% [5] 5.2% [9] 0.1e2% [11] 0.1% [58]
20.2% [57] 6.4% [10] 3.2% [13]
a
Simulated under AM1.5G.
b
Simulated by SLME model.
c
FA ¼ CH(NH2)2 and MA ¼ CH3NH3.

reasons that the stimulation of PCE shows the similar result for 4. Conclusions
CsSnI3 perovskites (25.6%) and MASnI3 perovskites (19.9%), but only
0.1% (2% doping by F) of the PCE for CsSnI3 perovskites was achieved We investigate the comprehensive properties of halide perov-
from the experiment, which is much lower than that of MASnI3 skites ABX3 (A ¼ CH3NH3, CH(NH2)2, Cs, Rb; B ¼ Pb, Sn, Ge; X ¼ I, Br,
perovskites (6.4%). Cl, F) by DFT calculations. The electronic structure, SOC, absorption
The bandgap of all the iodic perovskites is within 1e1.6 eV, spectrum, effective masses of charge of halide perovskites ABX3 are
which is the most suitable value for the absorber of solar cells based calculated to provide in-depth understanding of the relationship
on both S-Q and SLME models. All the iodic perovskites are the between the structure of halide perovskites and their photovoltaic
direct-band-gap materials, whose electronic transition from VBM properties. From the optical and electrical property point of view,
to CBM does not need the help of phonon. Ge perovskites possess a similar absorption spectrum and carrier
The absorption coefficient of the iodic perovskites containing Pb transport properties with Pb perovskites, however, Sn perovskites
and Ge near the solar spectrum is similar. But the absorption co- present a low absorption coefficient near the solar spectrum and an
efficient of Sn perovskites is lower than that of Pb and Ge perov- unbalanced electron and hole transport properties, which preju-
skites, and CH3NH3SnI3 shows very poor absorption coefficient, dices the photovoltaic properties.
especially. The theoretical values of PCE of halide perovskites are simulated
In addition, CH3NH3PbI3 and CsGeI3 perform a balanced by introducing S-Q and SLME model and compared with the
electron-hole-transport (0.14:0.19 for CsGeI3 and 0.12:0.12 for experimental values of halide perovskites. The enhanced exchange-
CH3NH3PbI3) due to the approximate effective mass of electrons correlation energy in halogen with small atomic number, from Cl to
and holes, while the effective mass for the holes of Sn perovskites is Br to I, decreases the bandgap and thus significantly increases the
rather smaller than the electronic effective mass (0.17:0.04 for PCE of perovskites. The introduction of I interstitials as the halide
CH3NH3SnI3), which may cause the unbalanced electron-hole- into perovskites will lead to the best performance of solar cells in
transport. both theoretical and experimental data. Notably, among the pe-
The differences of the optical and electrical properties of Pb, Sn rovskites with I interstitials, the theoretical PCE of Ge perovskites is
and Ge perovskites could be originated from the different electronic close to that of Pb perovskites, and a little higher than Sn perov-
structure of perovskites. Among the halide perovskites, the elec- skites. The investigation and understanding in the present work
tronic structure of Ge perovskites is almost the same as that of Pb will provide the theoretical foundation for exploring and devel-
perovskites with a small energy difference between the oping the lead free halide perovskites as high performance solar
nonbonding orbital and antibonding orbital, but is different from cells materials.
that of Sn perovskites because of the different outer-shell orbital
energy levels of Ge, Pb and Sn. Acknowledgements
The effect of SOC for Pb, Sn and Ge perovskites is strong, middle
and weak, respectively owing to the difference of the nuclear This work was supported by 973 Program (2014CB643506), the
charge number of Pb, Sn and Ge. Although Ge perovskites exhibit Natural Science Foundation of China (No. 21221063), Program for
the weak effect of SOC, which may reduce its conductivity, the in- Chang Jiang Scholars and Innovative Research Team in University
fluence of SOC on the photovoltaic performance is unclear. (No. IRT101713018) and Program for Changbaishan Scholars of Jilin
As a result, the experimental PCE of Pb perovskites is close to the Province. We thank the China Scientific Computing Grid (ScGrid)
theoretical value, i.e., the PCE of CH3NH3PbI3 is 19.7% in experiment for computing services.
and 26.7% in theory, and that of CH(NH2)2PbI3 is 20.2% in experi-
ment and 26.4% in theory, respectively. While Sn perovskites pre-
Appendix A. Supplementary data
sent a lower experimental PCE compared with the theoretical
values, due to the poor absorption, unbalanced electron-hole-
Supplementary data related to this article can be found at http://
transport and more defects (oxidation easily). It is worthy to note
dx.doi.org/10.1016/j.orgel.2016.05.046.
that CsGeI3 possesses the similar optical and electrical properties as
Pb perovskites. However, up to date, the best experimental PCE of
CsGeI3 is 3.2%, which still exhibits a huge gap from the theoretical References
value of 27.9% and needs to be optimized the design and fabrication
[1] A. Kojima, K. Teshima, Y. Shirai, T. Miyasaka, Organometal halide perovskites
of the related devices. as visible-light sensitizers for photovoltaic cells, J. Am. Chem. Soc. 131 (2009)
6050e6055.
[2] N. Cai, S.J. Moon, L. Cevey-Ha, T. Moehl, R. Humphry-Baker, P. Wang,
72 J. Qian et al. / Organic Electronics 37 (2016) 61e73

S.M. Zakeeruddin, M. Gratzel, An organic d-pi-a dye for record efficiency solid- CH3NH3PbI3 and CH3NH3SnI3 perovskites for solar cell applications, Sci. Rep.
state sensitized heterojunction solar cells, Nano Lett. 11 (2011) 1452e1456. 4 (2014), 4467e4467.
[3] M.Z. Liu, M.B. Johnston, H.J. Snaith, Efficient planar heterojunction perovskite [28] W. Shockley, H.J. Queisser, Detailed balance limit of efficiency of p-n junction
solar cells by vapour deposition, Nature 501 (2013) 395e398. solar cells, J. Appl. Phys. 32 (1961) 510e519.
[4] N.J. Jeon, J.H. Noh, W.S. Yang, Y.C. Kim, S. Ryu, J. Seo, Compositional engi- [29] L.P. Yu, A. Zunger, Identification of potential photovoltaic absorbers based on
neering of perovskite materials for high-performance solar cells, Nature 517 first-principles spectroscopic screening of materials, Phys. Rev. Lett. 108
(2015) 476e480. (2012) 1e5.
[5] Best Research-Cell Efficiencies, www.nrel.gov/ncpv/images/efficiency_chart. [30] S.C. Tidrow, Mapping comparison of Goldschmidt’s tolerance factor with
jpg (accessed: April 2016). perovskite structural conditions, Ferroelectrics 470 (2014) 13e27.
[6] H.S. Jia, L. Luo, B.L. Li, Z.H. Xu, X.K. Ren, Y.S. Jiang, L. Cheng, C.Y. Zhang, Per- [31] C.C. Stoumpos, C.D. Malliakas, J.A. Peters, Z. Liu, M. Sebastian, J. Im,
formance of polycrystal silicon color solar cells, Acta Phys. Sin. 62 (2013), T.C. Chasapis, A.C. Wibowo, D.Y. Chung, A.J. Freeman, B.W. Wessels,
168802e168802. M.G. Kanatzidis, Crystal growth of the perovskite semiconductor CsPbBr3: a
[7] S.D. Stranks, G.E. Eperon, G. Grancini, C. Menelaou, M.J.P. Alcocer, T. Leijtens, new material for high-energy radiation detection, Cryst. Growth Des. 13
L.M. Herz, A. Petrozza, H.J. Snaith, Electron-hole diffusion lengths exceeding 1 (2013) 2722e2727.
micrometer in an organometal trihalide perovskite absorber, Science 342 [32] O.N. Yunakova, V.K. Miloslavsky, E.N. Kovalenko, V.V. Kovalenko, Effect of
(2013) 341e344. structural phase transitions on the exciton absorption spectrum of thin
[8] I. Deretzis, A. Alberti, G. Pellegrino, E. Smecca, F. Giannazzo, N. Sakai, CsPbCl3 films, Low. Temp. Phys. 40 (2014) 690e693.
T. Miyasaka, A. La Magna, Atomistic origins of CH3NH3PbI3 degradation to [33] C.C. Stoumpos, C.D. Malliakas, M.G. Kanatzidis, Semiconducting tin and lead
PbI2 in vacuum, Appl. Phys. Lett. 106 (2015) 6050e6055. iodide perovskites with organic cations: phase transitions, high mobilities,
[9] F. Hao, C.C. Stoumpos, D.H. Cao, R.P.H. Chang, M.G. Kanatzidis, Lead-free solid- and near-infrared photoluminescent properties, Inorg. Chem. 52 (2013)
state organic-inorganic halide perovskite solar cells, Nat. Phot. 8 (2014) 9019e9038.
489e494. [34] Y. Takahashi, R. Obara, Z.Z. Lin, Y. Takahashi, T. Naito, T. Inabe, S. Ishibashi,
[10] N.K. Noel, S.D. Stranks, A. Abate, C. Wehrenfennig, S. Guarnera, A. Haghighirad, K. Terakura, Charge-transport in tin-iodide perovskite CH3NH3SnI3: origin of
A. Sadhanala, G.E. Eperon, S.K. Pathak, M.B. Johnston, Lead-free organ- high conductivity, Dalton Trans. 40 (2011) 5563e5571.
iceinorganic tin halide perovskites for photovoltaic applications, Energy En- [35] C. Bernal, K. Yang, First-principles hybrid functional study of the organ-
viron. Sci. 9 (2014) 3061e3068. iceinorganic perovskites CH3NH3SnBr3and CH3NH3SnI3, J. Phys. Chem. C
[11] M.H. Kumar, S. Dharani, W.L. Leong, P.P. Boix, R.R. Prabhakar, T. Baikie, C. Shi, 118 (2014) 24383e24388.
H. Ding, R. Ramesh, M. Asta, M. Graetzel, S.G. Mhaisalkar, N. Mathews, Lead- [36] F. Chiarella, A. Zappettini, F. Licci, I. Borriello, G. Cantele, D. Ninno,
free halide perovskite solar cells with high photocurrents realized through A. Cassinese, R. Vaglio, Combined experimental and theoretical investigation
vacancy modulation, Adv. Mater. 26 (2014) 7122. of optical, structural, and electronic properties ofCH3NH3SnX3thin
[12] D. Sabba, H.K. Mulmudi, R.R. Prabhakar, T. Krishnamoorthy, T. Baikie, P.P. Boix, films(X¼Cl,Br), Phys. Rev. B 77 (2008) 045129e045135.
S. Mhaisalkar, N. Mathews, Impact of anionic BreSubstitution on open circuit [37] R. Mosca, P. Ferro, T. Besagni, D. Calestani, F. Chiarella, F. Licci, Effect of hu-
voltage in lead free perovskite (CsSnI3-xBrx) solar cells, J. Phys. Chem. C 119 midity on the a.c. impedance of CH3NH3SnCl3 hybrid films, Appl. Phys. A 104
(2015) 1763e1767. (2011) 1181e1187.
[13] P.P. Boix, S. Agarwala, T.M. Koh, N. Mathews, S.G. Mhaisalkar, Perovskite solar [38] G. Murtaza, I. Ahmad, M. Maqbool, H.A.R. Aliabad, A. Afaq, Structural and
cells: beyond methylammonium lead iodide, J. Phys. Chem. Lett. 6 (2015) optoelectronic properties of cubic CsPbF3for novel applications, Chin. Phys.
898e907. Lett. 28 (2011) 117803e117807.
[14] X. Zhu, H.B. Su, R.A. Marcus, M.E. Michel-Beyerle, Computed and experimental [39] Y. Yamane, K. Yamada, K. Inoue, Mechanochemical synthesis and orderedis-
absorption spectra of the perovskite CH3NH3PbI3, J. Phys. Chem. Lett. 5 order phase transition in fluoride ion conductor RbPbF3, J. Phys. Chem. C
(2014) 3061e3065. Nanomater. Interfaces Solid State Ionics 179 (2008) 605e610.
[15] M.H. Du, Efficient carrier transport in halide perovskites: theoretical per- [40] W.J. Yin, T. Shi, Y. Yan, Unique properties of halide perovskites as possible
spectives, J. Mater. Chem. A 2 (2014) 9091e9098. origins of the superior solar cell performance, Adv. Mater. 26 (2014)
[16] C.C. Stoumpos, L. Frazer, D.J. Clark, Y.S. Kim, S.H. Rhim, A.J. Freeman, 4653e4661.
J.B. Ketterson, J.I. Jang, M.G. Kanatzidis, Hybrid germanium iodide perovskite [41] T. Baikie, N.S. Barrow, Y.A. Fang, P.J. Keenan, P.R. Slater, R.O. Piltz, M. Gutmann,
semiconductors: active lone pairs, structural distortions, direct and indirect S.G. Mhaisalkar, T.J. White, A combined single crystal neutron/X-ray diffrac-
energy gaps, and strong nonlinear optical properties, J. Am. Chem. Soc. 137 tion and solid-state nuclear magnetic resonance study of the hybrid perov-
(2015) 6804e6819. skites CH3NH3PbX3 (X ¼ I, Br and Cl), J. Mater. Chem. A 3 (2015) 9298e9307.
[17] B. Saparov, F. Hong, J.P. Sun, H.S. Duan, W.W. Meng, S. Cameron, I.G. Hill, [42] C. Bernal, K.S. Yang, First-principles hybrid functional study of the organic-
Y.F. Yan, D.B. Mitzi, Thin-film preparation and characterization of Cs3Sb2I9: a inorganic perovskites CH3NH3SnBr3 and CH3NH3SnI3, J. Phys. Chem. C 118
lead-free layered perovskite semiconductor,, Chem. Mater. 27 (2015) (2014) 24383e24388.
5622e5632. [43] M.L. Agiorgousis, Y.Y. Sun, H. Zeng, S. Zhang, Strong covalency-induced
[18] G. Kresse, J. Furthmüller, Efficient iterative schemes for ab initio total-energy recombination centers in perovskite solar cell material CH3NH3PbI3, J. Am.
calculations using a plane-wave basis set, Phys. Rev. B 54 (1996) Chem. Soc. 136 (2014) 14570e14575.
11169e11186. [44] R.G.P.a.W. Yang, Density-functional Theory of Atoms and Molecules, OUP,
[19] M.C. Payne, M.P. Teter, D.C. Allen, T.A. Arias, J.D. Joannopoulos, Iterative USA, 1995.
minimization techniques for ab initio total-energy calculations: molecular [45] W.J. Yin, J.H. Yang, J. Kang, Y. Yan, S.H. Wei, Halide perovskite materials for
dynamics and conjugate gradients, Rev. Mod. Phys. 64 (4) (1992) 1045e1097. solar cells: a theoretical review, J. Mater. Chem. A 3 (2014) 8926e8942.
[20] B. PE, Projector augmented-wave method, Phys. Rev. B 50 (1994) [46] J. Even, L. Pedesseau, J.M. Jancu, C. Katan, Importance of spin-orbit coupling in
17953e17979. hybrid organic/inorganic perovskites for photovoltaic applications, J. Phys.
[21] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made Chem. Lett. 4 (2013) 2999e3005.
simple, Phys. Rev. Lett. 77 (1996) 3865e3868. [47] H.C. Siegmann, Magnetism, Springer, 2010.
[22] J. Heyd, G.E. Scuseria, M. Ernzerhof, Hybrid functionals based on a screened [48] M.H. Du, D.J. Singh, Enhanced Born charge and proximity to ferroelectricity in
Coulomb potential, J. Chem. Phys. 118 (2003) 8207e8215. thallium halides, Phys. Rev. B 81 (2010).
[23] L.C. Tang, C.S. Chang, J.Y. Huang, Electronic structure and optical properties of [49] S.A. Kulkarni, T. Baikie, P.P. Boix, N. Yantara, N. Mathews, S. Mhaisalkar, Band-
rhombohedral CsGeI3 crystal, J. Phys. Condens. Matter 12 (2000) 9129e9143. gap tuning of lead halide perovskites using a sequential deposition process,
[24] C.T. Lee, W.T. Yang, R.G. Parr, Development of the Colle-Salvetti correlation- J. Mater. Chem. A 2 (2014) 9221e9225.
energy formula into a functional of the electron density, Phys. Rev. Lett. 37 [50] Z.G. Lin, L.C. Tang, C.P. Chou, Characterization and properties of infrared NLO
(1988) 785e789. crystals: AGeX3 (A¼Rb, Cs; X¼Cl, Br), J. Cryst. Growth 310 (2008) 3224e3229.
[25] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, [51] I. Borriello, G. Cantele, D. Ninno, Ab initio investigation of hybrid organic-
J.R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, inorganic perovskites based on tin halides, Phys. Rev. B 77 (2008),
H. Nakatsuji, M. Caricato, X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, 235214e223523.
G. Zheng, J.L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, [52] T. Baikie, N.S. Barrow, Y. Fang, P.J. Keenan, P.R. Slater, R.O. Piltz, M. Gutmann,
J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, S.G. Mhaisalkar, T.J. White, A combined single crystal neutron/x-ray diffrac-
J.A. Montgomery Jr., J.E. Peralta, F. Ogliaro, M.J. Bearpark, J. Heyd, E.N. Brothers, tion and solid-state nuclear magnetic resonance study of the hybrid perov-
K.N. Kudin, V.N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, skites CH3NH3PbX3 (X ¼ I, Br and Cl), J. Mater. Chem. A 3 (2015) 9298e9307.
A.P. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M. Cossi, N. Rega, N.J. Millam, [53] M.A. Green, Third generation photovoltaics. Advanced solar energy conver-
M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, sion, Phys. Today 57 (2004) 71e72.
R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, [54] W. Christian, G.E. Eperon, M.B. Johnston, H.J. Snaith, L.M. Herz, High charge
R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth, P. Salvador, carrier mobilities and lifetimes in organolead trihalide perovskites, Adv.
J.J. Dannenberg, S. Dapprich, A.D. Daniels, O. € Farkas, J.B. Foresman, J.V. Ortiz, Mater. 26 (2014) 1584e1589.
J. Cioslowski, D.J. Fox, Gaussian 09, Gaussian, Inc., Wallingford, CT, USA, 2009. [55] I. Chung, J.H. Song, J. Im, J. Androulakis, C.D. Malliakas, H. Li, A.J. Freeman,
[26] W.J. Yin, T.T. Shi, Y.F. Yan, Unique properties of halide perovskites as possible J.T. Kenney, M.G. Kanatzidis, CsSnI3: semiconductor or metal? High electrical
origins of the superior solar cell performance, Adv. Mater. 26 (2014) conductivity and strong near-infrared photoluminescence from a single ma-
4653e4658. terial. High hole mobility and phase-transitions, J. Am. Chem. Soc. 134 (2012)
[27] P. Umari, E. Mosconi, F. De Angelis, Relativistic GW calculations on 8579e8587.
J. Qian et al. / Organic Electronics 37 (2016) 61e73 73

[56] S. Wozny, M.J. Yang, A.M. Nardes, C.C. Mercado, S. Ferrere, M.O. Reese, [58] K. Thirumal, H. Ding, C. Yan, L.L. Wei, T. Baikie, L. Zhang, M. Sherburne, S. Li,
W.L. Zhou, K. Zhu, Controlled humidity study on the formation of higher ef- M. Asta, N. Mathews, Lead-free germanium iodide perovskite materials for
ficiency formamidinium lead triiodide-based solar cells, Chem. Mater. 27 photovoltaic application, J. Mater. Chem. A 3 (2015) 23829e23832.
(2015) 4814e4820. [59] J.M. Frost, K.T. Butler, F. Brivio, C.H. Hendon, M. van Schilfgaarde, A. Walsh,
[57] W.S. Yang, J.H. Noh, N.J. Jeon, Y.C. Kim, S. Ryu, J. Seo, I.S. Sang, High-perfor- Atomistic origins of high-performance in hybrid halide perovskite solar cells,
mance photovoltaic perovskite layers fabricated through intramolecular ex- Nano Lett. 14 (2014) 2584e2590.
change, Science 348 (2015) 1234e1237.

You might also like