First-principles hybrid functional study of the electronic structure and charge carrier mobility in perovskite CH3NH3SnI3
Wu Li-Juan1, Zhao Yu-Qing1, Chen Chang-Wen2, Wang Lin-Zhi1, Liu Biao1, Cai Meng-Qiu1, †,
School of Physics and Electronics Science, Hunan University, Changsha 410082, China
Yali Middle School, Changsha 410007, China

 

† Corresponding author. E-mail: mqcai@hnu.edu.cn

Project supported by the National Natural Science Foundation of China (Grant No. 51172067), the Hunan Provincial Natural Science Fund for Distinguished Young Scholars, China (Grant No. 13JJ1013), the Specialized Research Fund for the Doctoral Program of Higher Education, China (Grant No. 20130161110036), and the New Century Excellent Talents in University, China (Grant No. NCET-12-0171.D).

Abstract
Abstract

We calculate the electronic properties and carrier mobility of perovskite CH3NH3SnI3 as a solar cell absorber by using the hybrid functional method. The calculated result shows that the electron and hole mobilities have anisotropies with a large magnitude of 1.4 × 104 cm2·V−1·s−1 along the y direction. In view of the huge difference between hole and electron mobilities, the perovskite CH3NH3SnI3 can be considered as a p-type semiconductor. We also discover a relationship between the effective mass anisotropy and electronic occupation anisotropy. The above results can provide reliable guidance for its experimental applications in electronics and optoelectronics.

1. Introduction

The recent emergence of halide perovskites as light harvesters and transport materials has revolutionized the scenario of emerging photovoltaic technologies. Organic–inorganic hybrid perovskite based on metal halides adopts the ABX3 perovskite structure (B cation = Sn2+ or Pb2+; X is typically Br1 −, or I1 −; and the A cation is selected to balance the total charge and it is a small organic cation) solar cells are revolutionizing the field because of the potential for high efficiency and long dispersion length.[15] Therefore, in order to realize the sustainable development of energy and commercialization of technology, it is necessary to reach analogous optical and photovoltaic performance developing environmentally friendly solar cells. As is well known, lead poisoning is a type of metal poisoning and does harm to both the human body and the natural environment.[6] Motivated by non-toxic atoms of Sn in the same group compared with Pb, one substituted Sn2+ for Pb2+. However, a recent study[7] has reported that the strong acidification induced by the SnI2 is found to be more harmful than the combined effect of milder acidification and the expected Pb intoxication induced by PbI2, indicating that Sn-based perovskite may not be the ideal Pb surrogate. Fortunately, our experimental group has substituted Ge for Pb and Sn, considering this problem, tuning the superior solar cell performance of effective mass and absorption in perovskite CH3NH3GeCl3 under hydrostatic pressure.[8] Recent implementation of CH3NH3SnI3 (X = Cl, Br, I) perovskite absorbers enabled power conversion efficiency of 12%–15%, which has been recognized as “The Next Big Thing in Photovoltaics”. This makes them a possible light absorber in energy conversion or detector devices.[2] The perovskite CH3NH3SnI3 has an optical band gap of 1.3 eV,[9] indicating a significant red shift compared with the CH3NH3PbI3 (about 1.55 eV).[10]

Recently, lead-free solid-state organic–inorganic hybrid halide perovskite solar cells based on methylammonium tin iodide perovskite semiconductor as the light harvester, such as CH3NH3SnI3, has been identified as a promising photovoltaic (PV) material.[2] Bandgap engineering was implemented by chemical substitution in the form of CH3NH3SnI3-xBrx solid solutions, which can be controllably tuned to cover much of the visible spectrum, thus enabling the realization of lead-free solar cells with an initial power conversion efficiency of 5.73% under simulated full sunlight.[9] Thus, many studies have focused on CH3NH3SnX3. For example, Lang et al.[11] studied the chemical trends of the electronic properties in halide perovskites and proposed that CH3NH3SnBr3 could be a promising light-harvesting material because of appropriate band gap and optical absorption. Feng et al.[12] reported that the band gaps of CH3NH3SnX3 can be engineered to match the visible light spectrum under the low hydrostatic pressure (including compression and expansion). Umari et al. predicted that CH3NH3SnI3 has better electron transport properties than CH3NH3PbI3.[13] The efficiency is expected to be further increased by optimizing and better fundamental understanding of the internal electron dynamics and corresponding interfacial engineering. Papaassiliou and Koutselas[10] reported that perovskite solar cells represent a step towards the realization of low-cost, environmentally friendly solid-state solar cells. Despite prosperous applications of the Sn-based hybrid perovskites in solar cells, a systematic fundamental understanding of their electronic properties and band-edge positions is important for further optimizing their material properties. Jia et al.[14] reported improving light trapping and conversion efficiency of a solar cell by modifying and randomly distributing ZnO nanorods. Carrier mobility is the central issue for optoelectronic semiconducting materials. However, there is rare information about quantitative analysis of the carrier mobility of CH3NH3SnX3 theoretically.

In this work, we investigate the electronic properties, including the effective mass values of hole and electron at the Γ point. According to the calculated effective mass values, we further predict the charge mobility for nontoxic methylammonium tin halide (CH3NH3SnI3) at the equilibrium geometry by using the Heyd–Scuseria–Ernzerhof (HSE06) hybrid functional method.[15] We also clarify the relationship between the carrier mobility anisotropy and electron occupation anisotropy in p orbit of Sn and I atoms to offer constructive guidance to tune the carrier mobility along a certain direction in future.

2. Computational details

In this work, first-principles computations on the basis of density functional theory (DFT) are performed in the orthorhombic phase of MASnI3 lattice by using the Vienna ab initio simulation package (VASP).[16] Projector-augmented plane wave (PAW) approach is used to describe the ion–electron interactions.[17,18] We used the generalized gradient approximation (GGA) involving Perdew–Burke–Ernzerhof (PBE) functional and the Grimme vdW correction, which can accurately describe the weak interactions.[19,20] A cutoff energy of 400 eV for the plane-wave basis set is adopted in all computations. The k-point mesh of the Brillouin zone is set to be 4 × 4 × 4 for geometry optimization. The convergence threshold for self-consistent-field iteration is set to be 10−6 eV, and the atomic positions are fully optimized until all components of the residual forces are smaller than 0.001 eV/Å. The density of states (DOS) of perovskite CH3NH3SnI3 is calculated by using the tetrahedron method with Blöchl corrections.[17] As is well known, the standard DFT usually underestimates the band gap of semiconductor due to the inherent lack of derivative discontinuity and the self-correlation error of electrons.[21] Yuan et al.[22] reported that the band gap for cubic phase of CH3NH3SnI3 is just 0.244 eV, which underestimates about 1 eV compared with the experimental data (1.3 eV). To account for the underestimation of the band gap in standard DFT calculations, the accurate electronic structures are then calculated by using the Heyd–Scuseria–Ernzerhof (HSE06) hybrid functional.[15] The scalar-relativistic effects are included in the PAW Pseudopotentials. In addition, we realize that strong spin–orbit coupling (SOC) effects play a significant role in reducing the band gaps of compounds consisting of heavy elements. However, according to Bernal and Yang’s work,[23] we can ignore SOC effects in Sn-based perovskites. So, the band calculated by the HSE06 methods does not consider the spin–orbit coupling (SOC). As is well known, the more serious distortion may induce stronger ferroelectricity, which might play a vital role in the high carrier mobility. So, we use an orthorhombic cell to approximate the realistic structure. We start our calculations by focusing on the orthorhombic phase of CH3NH3SnI3,[24] as shown in Fig. 1. Based on the calculated band structures, we calculate the effective mass values along three different directions by fitting the band edge according to the following equation:

where E(k0) represents the energy at the band extreme, that is, the value at the k0 in the Brillouin zone. The tensor Sij is related to the effective mass tensor by

where the total effective mass m* can be computed from m* = 1/Sij. For orthorhombic Γ structures, the band extreme is located at the Γ point, thus, k0 = 0. Due to the symmetry of the crystal structure, the effective mass tensors of orthorhombic CH3NH3SnX3 (X = Cl, Br, and I) have the following forms:

Crystal structures of orthorhombic in CH3NH3SnI3 perovskite. The left panel is a side view, and the right panel shows the structure in polyhedron graphs (dark: tin; purple: iodine; brown: carbon; blue: nitrogen; pink: hydrogen).

In the matrix (Eq. (3)), the S11, S22, and S33 refer to the inverses of effective mass values of either the hole or electron in the [100], [010], and [001] directions, respectively. By fitting the obtained band dispersions in these three principal directions near the Γ point within Eq. (1), we compute the effective mass values of holes and electrons for orthorhombic phase CH3NH3SnI3. Based on the calculated effective mass values, the transport properties are evaluated theoretically using a simple approach based on the deformation potential theory as shown below[25]

where ħ is the reduced Planck constant, e is the element charge, Cii is the elastic matrix constant, m* is the effective mass of charge, kB is the Boltzmann constant, T is the temperature, and E1 is the deformation potential.[2628] T is set to be room temperature (T = 300 K). Parameters m*, E1, and Cii for CH3NH3SnI3 can be acquired by first-principles computations, which can be obtained in Appendix A.

3. Results and discussion
3.1. A brief introduction to the structural properties

Like MAPbX3, MASnX3 also exhibits a very rich phase diagram as a function of temperature, i.e., the crystal structure goes from cubic, tetragonal, orthorhombic, and monoclinic to triclinic class by cooling.[17,23,2729] Hao et al. reported a tetragonal CH3NH3SnI3 phase at room temperature.[2] It was found that CH3NH3SnI3 compound crystallizes into monoclinic phase at low temperature experimentally. However, the unit cell of monoclinic phase deviates from the orthorhombic crystal class very slightly.[28,29] The complicated phase transition behavior of CH3NH3SnI3 is attributed to the disordering of (CH3NH3)+ cations and the distortions of (SnX3) octahedra at finite temperature. The geometrical structure of the hybrid compound is illustrated in Fig. 1. In perovskite CH3NH3SnI3, each Sn atom is surrounded by six I atoms, with four I atoms in the equatorial direction and two I atoms in the apical direction. The dipolar organic methylammonium ions are embedded in the octahedron SnI6 cages. The previous work[30] revealed that arrangements along different directions play an important role in the stability and electronic structure. Thus the wake interaction between inorganic and organic molecule must be taken into account as performed in VASP with the nonlocal vdW density functional. We first estimate the equilibrum lattice constants of the hybrid perovskite compound by using DFT + D2 calculations in the frame of the PBE–GGA density functional, and the results are listed in Table 1, which are consistent with previous results obtained by Feng and Xiao.[12] Meanwhile, the calculated averaged bond length and the bond angles of Sn–I–Sn along the equatorial and apical direction are shown in Table 2, which are in agreement with previous results.[11] From Table 1, the calculated bond lengths of Sn–I for orthorhombic phase in CH3NH3SnI3 perovskite are about 3.09 Å and 3.12 Å–3.13 Å which are consistent with the experimental data.[12] The corresponding Sn–I–Sn angles are 89.77° and 94.86° in the equatorial and the apical direction respectively. These results reveal that CH3NH3SnI3 has more distorted structure than CH3NH3PbI3, which may be the origin of ferroelectricity.

Table 1.

Lattice constants of orthorhombic phase in perovskite CH3NH3SnI3 compared with Feng and Xiao’s work.[12]

.
3.2. Electronic properties of CH3NH3SnI3

For orthorhombic phase in perovskite CH3NH3SnI3, the energy band structure is shown in Fig. 2, and the calculated energy band gap is 1.27 eV by using HSE06 functional method as shown in Table 2. Our calculations reveal that the CH3NH3SnI3 is a direct gap semiconductor, and the calculated results are consistent with those reported in previous research.[31] Figures 2(a) and 2(b) show the band structures of the orthorhombic phase in CH3NH3SnI3 calculated by standard DFT and HSE06 functional methods respectively. For this phase, they all exhibit a direct band gap at the Γ point of the Brillouin zone. The difference could be seen easily from Figs. 2(a) and 2(b), the calculated band gaps are 0.74 eV by standard density functions theory and 1.27 eV by HSE06 method, respectively, which are listed in Table 2.

Fig. 2. Band structures for orthorhombic phase in CH3NH3SnI3 perovskites by standard DFT and HSE06 calculations, respectively. The Brillouin zone is also shown.
Table 2.

Calculated band gaps of DFT-D2 and HSE06, the bond lengths of Sn–I and the bond angles of Sn–I–Sn along the equatorial and apical direction.

.

As is well known, the standard DFT usually underestimates the band gap of semiconductors due to the inherent lack of derivative discontinuity and the self-correlation error of electrons.[32] Therefore, results (1.27 eV) obtained by the hybrid functional within the framework of density functional theory (DFT-D2) method are closer to the experimental data (1.3 eV).[9] We ascribe the difference between the calculated data and experimental data to the following reasons: (i) the experimental materials are flawed, while the calculated structure is considered as an ideal model; (ii) though HSE06 calculations improve the exchange action, the improved exchange effect is still from a semi-empirical theory. According to the calculated band structure, the carrier effective mass values are shown in Table 3. The effective mass values of orthorhombic phase in perovskite CH3NH3SnI3 at Γ along three principal directions, i.e., [1 0 0], [0 1 0], and [0 0 1] direction are evaluated respectively by using Eq. (1) as the effective mass values of hole and electron are associated with the slope of the band edge. The calculated results compared with those of Feng’s work are listed in Table 3. From the Table 3, it is clearly seen that carrier effective mass values along three principal directions are anisotropic, which are consistent with Feng and Xiao’s[12] results. Owing to our calculated band gap more closer to the experimental data, we believe that the effective mass values of electrons and holes are considered as more reliable results. Also, the results reveal that the orthorhombic phase in CH3NH3SnI3 shows that the hole effective mass is smaller than the electron’s.

Table 3.

Calculated effective mass values of hole and electron of CH3NH3SnI3 for orthorhombic phase compared with the results of CH3NH3SnI3 (orthorhombic phase) calculated by Feng and Xiao’s work.[12]

.

The density of states (DOS), and the projected density of states (PDOS) of CH3NH3SnI3 are shown in Fig. 3. The calculated results show that the valence band maximum (VBM) for orthorhombic phase in CH3NH3SnI3 is mainly due to the 5p orbitals of the I and partly due to the 5s orbital of Sn, while in the conduction band minimum (CBM), the main contribution is from the 5p orbitals of Sn and little contribution is from the 5p orbitals of I. Also, it is noted that the organic cations (CH3NH3)+ are far from the Fermi level. Therefore, the emergency of the photocurrent is mainly generated by the electron transition from 5p orbitals of I and 5s orbital of Sn to the 5p orbitals of Sn. A similar structure such as CH3NH3PbI3 also shows that the strong hybridization of s–p orbital in valence band maximum (VBM) plays an important role in the structure stability and the small effective mass.[24] More detailed analyses of the chemical bonding mechanisms of those hybrid organic–inorganic pervoksite materials can be found in previous publications.[3336]

Fig. 3. The calculated PDOS from top to bottom in CH3NH3SnI3 perovskite.

In order to further illustrate our results, the calculated projected DOS in valence band maximum and conduction band minimum are shown in Fig. 4 to explain the electron and hole carrier effective mass values along three principal directions. Here, we present the theoretical discovery of a new relationship between the carrier effective mass anisotropy and the electronic DOS anisotropy along three principal directions. As mentioned above, the valence band maximum (VBM) is mainly due to the I-p states and the conduction band minimum (CBM) is mainly due to the Sn-p states. Thus we just discuss the electron occupation probabilities of I-p states in valence band maximum and Sn-p states in conduction band minimum, respectively, as shown in Fig. 4. The above Table 2 reveals that the hole effective mass values of orthorhombic phase in CH3NH3SnI3 along the x, y, and z directions are 0.35, 0.11, and 0.27 respectively. From Fig. 4(a), increasing sequence of hole effective mass values along different directions is which is consistent with the decreasing sequence of electron occupation probabilities in the p orbital of I along the x, y, and z directions px < pz < py in valence band maximum. Similarly the calculated values of electron effective mass , and are 1.77, 0.15, and 1.75 respectively. Increasing sequence of electron effective mass values along different directions is which is also consistent with the decreasing sequence of electron occupation probabilities in the p orbital of Sn along the x, y, and z directions as follows: py > pz > px in the bottom conduction-band depicted in Fig. 4(b). There might be such a relationship that the higher the occupation probabilities in the p orbital along the x, y, and z directions, the smaller the effective mass is, hence the higher the carrier mobility.

Fig. 4. Detailed PDOS from top to bottom for orthorhombic phase in CH3NH3SnI3 perovskite. The left panel shows the VBM, and the right panel shows the CBM.
3.3. High mobilities of electrons and holes

In order to evaluate the transfer efficiency of the carrier in perovskite CH3NH3SnI3, we systematically investigate the carrier mobility of CH3NH3SnI3. From experimental measurement, the system of CH3NH3SnI3 was reported to have high electron mobility (μe ∼ 2320 cm2· V−1·s−1),[8] which is higher than CH3NH3PbI3 (μe ∼ 66 cm2· V−1·s−1).[8] The charge mobility is calculated according to Eq. (2). It should be noticed that reliable band structure is highly essential for the calculations of carrier mobilities. All the calculated results are listed in Table 4. Figure S1 (see Appendix A) shows the energy changes of VBM and CBM band edge of CH3NH3SnI3 with the applied strain along the x, y, and z directions, respectively. By simulating the linear relation between band edge energy (Eedge) and strain (δ) exerted in the x, y, and z directions, we can obtain the deformation potential (DP) constant E1. The DP constant is a characterization of the coupling strength of the electron or hole to the acoustic phonon. It is found that the DP constant E1 values of electrons and holes are comparable in CH3NH3SnI3 perovskite (see Table 4). The elastic matrix constant along the y direction is 637 kbar (1 bar = 105 Pa), which is about twice that along the z direction, and the effective mass values for the hole and electron are the minimal (along z direction) among three directions. Therefore, it is clearly seen that the carrier mobility along the y direction is highest because of high DP and small effective masses. From Eq. (2), we can see the main difference between the electron and hole mobilities in CH3NH3SnI3 derives from the small effective masses, rather than the DP constant E1. It can be clearly seen that CH3NH3SnI3 perovskite shows high carrier mobilities of 14470 cm2· V−1·S−1 and 5445 cm2· V−1·S−1 for holes and electrons, respectively, which are rather higher than the experimental data (2230 cm2· V−1·S−1).[8] The difference can be attributed to the following reasons: (i) there are inevitably impurities and defects in the experimental material; (ii) the phonons have little influence on the charge transport properties at low temperatures; (iii) the Sn2 + ion can be oxidized into Sn4 + state easily, which may underestimate the carrier mobilities in the experimental measurement. The carrier mobility (17 and 545, 11 and 1663 cm2· V−1·S−1) for holes and electrons in the x and z direction are two or three orders lower than that along the y direction, respectively, and so is the anisotropic carrier mobility. Our calculated carrier mobility may lay the solid foundation and the theoretical support for the application of the device.

Table 4.

Calculated values of deformation potentials E1c and E1v for electron and hole, respectively, elastic constant (cii) , and carrier mobilities μe and μh for electron and hole, respectively at the Γ point of the Brillouin zone. Three directions are considered for the orthorhombic phase.

.
4. Conclusions

In this work, we employ DFT + D2 and HSE06 to study the equilibrium lattice parameters, electronic structures and the intrinsic charge carrier mobility of orthorhombic phase in CH3NH3SnI3. We conclude that CH3NH3SnI3 is semiconductor with direct band gaps at the Γ point. The anisotropy of carrier effective mass can be attributed to the anisotropy of electron occupation probabilities along the three principal directions, so can the anisotropy of carrier mobility. The results reveal that the high carrier mobility is ascribed to the intrinsically small effective mass, in addition to a relatively weak electron–phonon coupling. Owing to the hole mobility higher than the electron mobility, the CH3NH3SnI3 perovskite can be considered as p-type semiconductor in electrical conduction. The suitable energy bad gaps and impressively large carrier mobilities reported here indicate that CH3NH3SnI3 is a very promising material in the field of application and designing for solar cell devices.

Reference
1Chung ILee BHe J QChang R P HKanatzidis M G 2012 Nature 485 486
2Hao FStoumpos C CCao D HChang R P HKanatzidis M G 2014 Nat Photon. 8 489
3Kojima ATeshima KShirai YMiyasaka T J 2009 Am. Chem. Soc. 131 6050
4Kim H SLee C RIm J HLee K BMoehl TMarchioro AMoon S JHumphry-Baker RYum J HMoser J EGratzel MPark N G 2012 Sci. Rep. 2 11203
5Lee M MTeuscher JMiyasaka TMurakami T NSnaith H J 2012 Science 338 643
6Gidlow D A 2004 Occup. Med-Oxford 54 76
7Aslihan BDinh D TAnitha EJean MMarc MHans-Gerd BBert C 2016 Sci. Rep. 6 18721
8Zhao Y QWu L JWang L ZLiu BHe P BCai M Q 2016 J. Power Source 313 96
9Stoumpos C CMalliakas C DKanatzidis M J 2013 Inorg. Chem. 52 9019
10papaassiliou G CKoutselas I B 1995 Synthetic Met. 71 1713
11Lang LYang J HLiu H RXiang H JGong X G 2014 Phys. Lett. 378 290
12Feng JXiao B 2014 J. Phys. Chem. 118 19655
13Umari PMosconi EDe Angelis F 2014 Sci. Rep. 4 563
14Jia Z NZhang X DLiu YWang Y FFan JLiu C CZhao Y 2014 Chin. Phys. 23 046106
15Heyd JScuseria G EErnzerhof M 2003 J. Chem. Phys. 118 8207
16Kresse GHafner J 1993 Phys. Rev. 47 558
17Blöchl P EJepsen OAndersen O K 1994 Phys. Rev. 50 17953
18Kresse GJoubert D 1999 Phys. Rev. 59 1758
19Perdew J PBurke LErnzerhof M 1996 Phys. Rev. Lett. 77 3865
20Grimme S 2006 J. Comput. Chem. 27 1787
21Takahashi YObara RLin Z ZTakahashi YNaito TInabe TIshibashi STerakura K 2011 Dalton 40 5563
22Yuan YXu RXu H THong FXu FWang L J 2015 Chin. Phys. 24 116302
23Bernal CYang K 2014 J. Phys. Chem. 118 24383
24Feng JXiao B 2014 J. Phys. Chem. Lett. 5 1719
25Chung ISong J HIm JAndroulakis JMalliakas C DLi HFreeman A JKenney J TKanatzidis M G 2012 J. Am. Chem. Soc. 134 8579
26Wei S HZunger A 1998 Phys. Rev. 60 5404
27Li Y HGong X GWei S H2006Appl. Phys. Lett.884520
28Franceschetti AWei S HZunger A 1994 Phys. Rev. 50 17797
29Wallentin JAnttu NAsoli DHuffman MAberg IMagnusson M HSiefer GFuss-Kailuweit PDimroth FWitzigmannBXu H QSamuelson LDeppert KBorgstrom M T 2013 Science 339 1057
30Geng WZhang LZhang Y NLau W MLiu L M 2014 J. Phys. Chem. 118 19565
31Kresse GFurthmüller J 1996 Mater. Sci. 6 15
32Bardeen JShockley W 1950 Phys. Rev. 80 72
33Kawamura YMashiyama HHasebe K 2002 J. Phys. Soc. Jpn. 71 1694
34Yin W JShi T TYan Y F 2014 Adv. Mater. 26 4653
35Wang YGould TDobson J FZhang H MYang H GYao X DZhao H J 2014 Phys. Chem. Chem. Phys. 16 1424
36Huang L YLambrecht W R L 2013 Phys. Rev. 88 1702