Discovering lead-free perovskite solar materialswith a split-anion approach†
Yi-Yang Sun,*a Jian Shi,b Jie Lian,c Weiwei Gao,d Michael L. Agiorgousis,a
Peihong Zhangd,e and Shengbai Zhang*a
Organic–inorganic hybrid perovskite solar materials, being low-cost and high-performance, are promis-
ing for large-scale deployment of the photovoltaic technology. A key challenge that remains to be
addressed is the toxicity of these materials since the high-efficiency solar cells are made of lead-contain-
ing materials, in particular, CH3NH3PbI3. Here, based on first-principles calculation, we search for lead-
free perovskite materials based on the split-anion approach, where we replace Pb with non-toxic
elements while introducing dual anions (i.e., splitting the anion sites) that preserve the charge neutrality.
We show that CH3NH3BiSeI2 and CH3NH3BiSI2 exhibit improved band gaps and optical absorption over
CH3NH3PbI3. The split-anion approach could also be applied to pure inorganic perovskites, significantly
enlarging the pool of candidate materials in the design of low-cost, high-performance and environ-
mentally-friendly perovskite solar materials.
Introduction
The discovery of organic–inorganic halide perovskite materialsas light absorbers to make high-efficiency solar cells1–8 hasstimulated a surge of research in the past several years.9,10 Theearth-abundance of the constituent elements and the low-temperature synthesis of these materials11 make them highlypromising for low-cost photovoltaic applications. One of thekey issues faced by the current halide perovskite materials isthe use of lead (Pb) to achieve high efficiency. Both elementalPb and its halides are well known to be highly toxic, prohibit-ing large-scale deployment of the Pb-based materials. Acurrent focus of research is to replace Pb with tin (Sn).12–15 Sofar, the efficiency achieved by Sn-based materials has not beencomparable with that by Pb-based materials. Unfortunately, forhalide (or I–II–VII3) perovskite materials, it appears that noother alternative elements across the periodic table can replacePb while maintaining suitable band gaps for solar absorberapplication. Thus, it is of great interest to explore new perovs-
kite materials beyond the composition of I–II–VII3, e.g., chalco-genide perovskites, which have recently been proposed as solarmaterials.16
Cation-splitting has been a successful approach in design-ing solar materials. The high-performance I–III–VI2 chalcopyr-ite materials, e.g., CuInSe2 or Cu(In,Ga)Se2,
17 can beconsidered as derived from II–VI zinc-blende structures bysplitting two 2+ cations into one 1+ and one 3+ cation. Theapproach can be further carried out by splitting two 3+ cationsinto one 2+ and one 4+ cation, leading to the kesteritematerials, of which Cu2ZnSnSe4 is a successful example.18,19
The advantage of this approach is that the crystal structureand local chemical environment of the parent materials arelargely preserved, resulting in similar electronic and opticalproperties in the derived materials. The cation-splittingapproach has also been applied to oxide perovskite materialsshowing promising results for photovoltaic applications.20 Incontrast, anion-splitting has been rarely studied for developingnew solar-cell materials. Recently, nitride-oxide perovskitematerials have been synthesized for photocatalytic water split-ting.21 This work shows great promise of split-anion perovs-kites for photovoltaic applications. The success of the halideperovskite materials, as represented by CH3NH3PbI3, hasinspired us to explore the anion-splitting approach with theaim of developing Pb-free perovskite solar materials.
Because of the high efficiency already achieved by the Pb-based materials, it is preferred to replace Pb with an elementthat is close to Pb in the periodic table with the hope that thedesirable properties of the Pb-based materials can be pre-served. Bismuth (Bi) is thus a candidate of great interest.
†Electronic supplementary information (ESI) available: Detailed descriptions onthe structure optimization and quasi-particle GW calculation. See DOI: 10.1039/c5nr04310g
aDepartment of Physics, Applied Physics & Astronomy, Rensselaer Polytechnic
Institute, Troy, New York 12180, USA. E-mail: [email protected], [email protected] of Materials Science and Engineering, Rensselaer Polytechnic Institute,
Troy, New York 12180, USAcDepartment of Mechanical, Aerospace, and Nuclear Engineering, Rensselaer
Polytechnic Institute, Troy, New York 12180, USAdDepartment of Physics, University at Buffalo SUNY, Buffalo, New York 14260, USAeBeijing Computational Science Research Center, Beijing 100084, China
Moreover, in terms of toxicity, Bi is much safer than Pb formassive use. In this paper, based on first-principles calcu-lations, we study the Bi-based organic–inorganic perovskitematerials for photovoltaic applications. Starting withCH3NH3PbI3, we apply the split-anion approach by replacingone I per formula with Se or S, while replacing Pb with Bi, tosatisfy the charge neutrality. In general, this approach leads tothe I–III–VI–VII2 compounds. By using hybrid functional calcu-lations including spin–orbit coupling, we show thatCH3NH3BiSeI2 and CH3NH3BiSeI2 could possess a direct bandgap of 1.3 to 1.4 eV, ideal for single-junction solar cells. Onecould also envision a large family of new compounds based onthe split-anion approach, which have been largely unexplored.We will discuss one such pure inorganic compound, CsSnS2Cl,for photovoltaic applications.
Method
Our first-principles calculation was based on the density func-tional theory (DFT) as implemented in the VASP program.22
The generalized gradient approximation of Perdew, Burke andErnzerhof (PBE)23 was used for the exchange–correlation func-tional. To better describe the interaction between the organiccation and the inorganic framework, we include the van derWaals interaction through the PBE-D3 method.24 A 2 × 2 × 2supercell containing eight formula units (f.u.) of I–II–VII3 orI–III–VI–VII2 compounds was used in our calculation. Thelattice constants and the internal coordinates were obtainedby performing variable-cell optimization. To ensure convergedstructural parameters, a cutoff energy of 544 eV for the plane-wave basis set, a 3 × 3 × 3 k-point grid, and a force criterion of0.025 eV Å−1 were used. Further details on optimizing theatomic structures are given in the ESI.† Qualitative analysis ofelectronic structures, such as band structures and density ofstates, was also conducted at the PBE level. To accuratelypredict the band gaps, we employed the Heyd–Scuseria–Ernzer-hof (HSE) hybrid functional,25,26 as discussed below. The effectof spin–orbit coupling (SOC) was also considered when calcu-lating the band gaps. In the HSE + SOC calculations, acutoff energy of 340 eV and four special k-points, (0, 0, 0), (12,
12, 0),
(0, 12,12), and (12, 0,
12), were found to yield converged band gaps.
Results and discussion
Fig. 1a and b illustrate the idea of the split-anion approach. Biis the neighbor of Pb in the periodic table and has no issuewith toxicity. However, simply replacing Pb in CH3NH3PbI3with Bi will introduce one extra electron per f.u. (similar to adonor) and make the material metallic. To compensate for theextra electron, we replace one I atom per f.u. with one Se atom,serving as an acceptor, so that the charge neutrality, hence thesemiconductivity, can be preserved. Thus, the anion sites aresplit into two groups occupied by the group-VI and group-VIIelements, respectively.
At room temperature, CH3NH3PbI3 stabilizes in a tetragonalstructure with the I4/mcm space group,27 where the I atomsoccupy the 4a (or apical) and 8h (or equatorial) Wyckoff sites.Here, we consider the Se atoms occupying the 4a sites, while Iatoms occupy the 8h sites, as shown in Fig. 1b. To evaluate thestability of the proposed split-anion perovskite material, weperformed an ab initio molecular dynamics (AIMD) simulationon CH3NH3BiSeI2. The simulation was in the NVT ensemblewith the temperature controlled at 300 K using the Nosé ther-mostat.28 Because of the presence of the light element H, weused a small time step of 0.25 fs. The total simulation time at300 K is 50 ps. We averaged the atomic positions of Bi, Se, andI atoms over the last 40 ps simulation and found that theyform a framework of the perovskite structure, while theorganic cations appear to freely rotate in the void region (i.e.,around the A-sites). No structural instability was observed inour simulation.
Next, we discuss the electronic and optical properties of thesplit-anion perovskite materials. We first evaluate their bandgaps. For CH3NH3PbI3, the band gap calculated with the PBEfunctional is 1.57 eV, which fortuitously agrees with the experi-mental value (∼1.55 eV).29 This agreement is known to be aresult of error cancellation between an underestimation due tothe standard DFT band gap error and an overestimation due tothe exclusion of the SOC effect.30 Here we employ the HSEfunctional,25,26 a screened-exchange hybrid functional, tocorrect the DFT band gap error. The standard HSE functionalwas found to yield a smaller band gap (∼1.13 eV) for
Fig. 1 Atomic structures of (a) CH3NH3PbI3 and (b) CH3NH3BiSeI2 and aschematic illustrating the split-anion approach to replace Pb inCH3NH3PbI3. (c) and (d) show the calculated band gaps of CH3NH3PbI3and CH3NH3BiSeI2, respectively, using improved methods from PBE, HSEto HSE + SOC. The alignment of the band edge positions was obtainedby assuming that the reference potentials from different methods arethe same.
CH3NH3PbI3 than the experimental value if considering theSOC effect. There are two parameters in the HSE functionalα and ω controlling, respectively, the mixing and screening ofthe Hartree–Fock exchange. Here we fix α at 0.25 following anargument based on the adiabatic connection formula,31 whileemploying a smaller ω (0.05 Å−1) than the standard value(0.2 Å−1). With the modified ω, our HSE + SOC calculationyielded a band gap of 1.54 eV for CH3NH3PbI3, in good agree-ment with the experiment. Below, we will use the same para-meters for the Bi-based organic–inorganic perovskites.
Using the structures optimized by the PBE-D3 method, wecalculated the band gap at the Brillouin zone center (Γ point)using improved methods from PBE, HSE to HSE + SOC (i.e.,HSE including spin–orbit coupling). Our results show that thesplit-anion Bi-based perovskite materials (including thematerials discussed below) also have direct band gaps at the Γpoint of the 2 × 2 × 2 supercell. Fig. 1c and d show the calcu-lated band gaps of CH3NH3PbI3 and CH3NH3BiSeI2, where thealignment of the band edge positions was made by assumingthat the reference potentials from different methods are thesame. This approximation was shown32 to be valid for typicalsemiconductors because given the same atomic structure,different DFT functionals usually yield similar charge den-sities, which determine the average potential of the unit cell.Based on these alignments, it is shown that the HSE calcu-lation opens up the band gap by down-shifting the valenceband maximum (VBM) and up-shifting the conduction bandminimum (CBM). The down-shift of the VBM is roughly twotimes more prominent than the up-shift of the CBM. In con-trast to the effect of HSE, the SOC reduces the band gap.However, the magnitude of the down-shifting of the CBM byincluding the SOC effect is much more prominent than theup-shifting of the VBM because the CBM is mainly composedof the Pb or Bi 6p states (discussed below), which are more sig-nificantly affected by the SOC than the VBM (mainly Se 4p andI 5p states). The asymmetric change in the band gap by theHSE and SOC results in overall down-shifts of the band edges.
Experimentally, alloying with different anions has beencommonly used to fine tune the material properties.33–35
We have considered the combinations of S, Se and Te forgroup-VI elements and Cl, Br and I for group-VII elements.The calculated band gaps for the nine compounds are shownin Fig. 2. It can be seen that similar to the Pb-based halide per-ovskites, the iodides, especially CH3NH3BiSI2 and CH3NH3Bi-SeI2, yield the optimal band gaps (1.3–1.4 eV) for solarabsorbers according to the Shockley–Queisser theory.36 Weevaluated the possible errors in the calculated band gaps. Onesource of error is the random orientation of the organiccations. To consider this, we took 10 snapshots from the AIMDsimulation (see the ESI†). The variation in the band gap isfound to be less than 0.1 eV. Another source of error is thepossible disorder of Se atoms on the anion sites. We con-sidered this by generating 10 supercells with randomly distrib-uted Se atoms. It was found that the disordered supercellshave smaller band gaps than the ordered ones. However,as long as the total energy is not higher than the ordered struc-
ture by 0.5 eV per supercell, the reduction in the band gap isless than 0.2 eV.
Fig. 3 compares the band structures, densities of states(DOS), and imaginary parts of the dielectric function (ε2) ofCH3NH3PbI3 and CH3NH3BiSI2. From the band structure inFig. 3a, it can be seen that the bottom of the conduction bandof CH3NH3BiSI2 shows significant dispersion suggesting goodelectron transport properties. The top of the valence band con-tains one dispersive band while the other bands are less dis-persive. The effective masses of electrons and holes ofCH3NH3BiSI2 are 0.32 and 0.40m0 (electron rest mass), respect-ively, as calculated using the PBE functional and averaged overthe four directions shown in Fig. 3. It has been shown thatadding SOC and quasi-particle GW correction could furtherreduce the effective masses.30 It is thus expected that theeffective masses of CH3NH3BiSI2 could be comparable withthose of CH3NH3PbI3 (0.19 and 0.25m0 for electrons and holes,respectively).30 From the DOS in Fig. 3b, it can be seen that thetop of the valence band of CH3NH3BiSI2 contains two mainpeaks. The lower peak is mainly contributed by I 5p states,while the higher peak is contributed by hybridized statesbetween S 3p and I 5p states, which are the less-dispersivestates. Similar to CH3NH3PbI3, the bottom of the conductionband of CH3NH3BiSI2 is contributed by the 6p states of Bi,which are responsible for the dispersive bottom conductionbands.
Fig. 3c shows a comparison of the calculated ε2 ofCH3NH3PbI3 and CH3NH3BiSI2. The results are obtained byusing the PBE functional with a scissor operator applied toCH3NH3BiSI2 by shifting the conduction bands up by 0.3 eV toobtain the band gap calculated by HSE + SOC. The ε2 providesa direct measure of optical absorption. From this comparison,CH3NH3BiSI2 is expected to exhibit stronger optical absorptionthan CH3NH3PbI3 below 3 eV. According to the AM 1.5 solarspectrum, about 94% of the solar radiation reaching the earthsurface is below 3 eV. Thus, CH3NH3BiSI2 would be a stronger
Fig. 2 Calculated band gaps of CH3NH3BiXY2 compounds (with X = S,Se, or Te and Y = Cl, Br, or I) using the HSE functional with spin–orbitcoupling. The dashed line marks the optimal band gap for a single-junc-tion solar cell according to the Shockley–Queisser theory.
solar absorber than CH3NH3PbI3. Also, given that the calcu-lated band gap of 1.38 eV is closer to the ideal value for single-junction solar cells, CH3NH3BiSI2 could be a promising candi-date PV material.
We have considered above the I–III–VI–VII2 compounds todemonstrate the split-anion approach. Other possible split-anion compounds could also have chemical compositions ofII–II–VI–VII2, I–IV–VI2–VII, and II–III–VI2–VII. The split-anionapproach thus significantly enlarges the pool of elements forthe design of new perovskite materials, not only for theorganic–inorganic hybrid perovskites, but also for pure-in-organic perovskites. Currently, lead-free perovskite solarmaterials are mainly based on the replacement of Pb by Sn. Insuch materials, Sn is in the 2+ oxidation state (SnII), part ofwhich could be oxidized to the 4+ state (SnIV) under ambientconditions. The SnIV defects in the SnII compounds could bedetrimental to the device performance.12,13 The split-anionapproach provides a possibility of designing all SnIV-based per-ovskite solar materials to avoid further oxidation.
Taking CsSnI3 as an example, which has a nearly idealband gap of 1.3 eV and has been used to make dye-sensitized
solar cells with an efficiency up to 8%,37 we now demonstratethe design of SnIV-based perovskite solar materials. We notethat, while Cs is used here for the A-site cation as an example,other 1+ cations could also be considered as candidates.Similar to Fig. 2, we considered nine combinations of theanions. It was found that CsSnS2Cl possesses a suitable bandgap as a solar absorber, while the band gaps of other materialsare smaller than 1 eV.
Fig. 4a and b show the structure of CsSnS2Cl in the dis-torted perovskite phase, which has the symmetry of the I4/mcmspace group, i.e., the symmetry of CH3NH3PbI3 at room tem-perature. Fig. 4c and d show the Brillouin zone and the bandstructure calculated using the standard hybrid HSE functional(i.e., α = 0.25 and ω = 0.2). Here we used the quasi-particle GWmethod38 to calibrate the hybrid functional and found that thestandard HSE functional already yielded a band gap in agree-ment with the GW calculation within 0.01 eV (see the ESI†).CsSnS2Cl is an indirect-gap material with the CBM located atthe Γ point and the VBM at about one-third distance from Γ toM. The HSE band gap is 0.98 eV. It is worthwhile to note thatan indirect gap may not necessarily be a disadvantage for solarabsorber materials since indirect-gap materials have a weaker
Fig. 4 Atomic structure of CsSnS2Cl from the top (a) and side (b) views.(c) Brillouin zone with labels of high-symmetry k-points. (d) Band struc-ture of CsSnS2Cl calculated using the HSE functional. (e) Imaginary partof dielectric function (ε2) of CsSnS2Cl, CsSnI3 and Si, all of which havecalculated band gaps at about 1.0–1.1 eV using the HSE functional.
Fig. 3 Comparison of (a) band structures, (b) densities of states (DOS)and (c) imaginary parts of dielectric constants of CH3NH3PbI3 andCH3NH3BiSI2. The band structures were obtained using a supercell con-taining eight formula units of CH3NH3PbI3 and CH3NH3BiSI2. Thek-points Γ, X, M, Z, and R have coordinates of (0, 0, 0), (0, 1
2, 0), (12,
12, 0),
(0, 0, 12), and (12,12,
12), respectively. The DOS of conduction bands in (b) for
both CH3NH3PbI3 (upper panel) and CH3NH3BiSI2 (lower panel) areincreased three times to clearly show the 6p states of Pb or Bi.
radiative recombination than direct-gap materials, whichcould improve the carrier lifetime in the regime where defect-mediated (or non-radiative) recombination is weak.
While having an indirect band gap, CsSnS2Cl showsdifferent behaviors in optical absorption from typical indirect-gap solar materials, such as Si. Fig. 4e compares the calculatedε2 of CsSnS2Cl, CsSnI3 and Si. It can be seen that Si, as is wellknown, exhibits a slow increase in ε2 until about 3 eV. In con-trast, CsSnS2Cl exhibits a sharp increase when the photonenergy is greater than the band gap. This is because the directband gap of CsSnS2Cl at the Γ point is 1.11 eV. In comparison,the direct gap of Si at the Γ point is 3.34 eV based on our HSEcalculation. According to Fig. 4e, CsSnS2Cl could have evenbetter optical absorption than the direct-gap CsSnI3 below3 eV. This is partly because of the relatively large density ofstates near the top of the valence bands due to the presence ofdual anions, similar to the case of CH3NH3BiSI2 shown inFig. 3b.
Conclusions
In summary, we have explored the split-anion approach tosearch for lead-free perovskite solar absorber materials. Start-ing from the high-performance organic–inorganic hybrid per-ovskite CH3NH3PbI3, we replace Pb with non-toxic Bi whilesplitting the three I per f.u. into two I and one Se or S. Theresulting materials, CH3NH3BiSI2 and CH3NH3BiSeI2, areshown to preserve or even improve the properties ofCH3NH3PbI3, as solar absorber materials. The split-anionapproach could also be applied to pure inorganic perovskitematerials. As an exploratory example, we studied CsSnS2Cl,which has an indirect band gap of about 1 eV according to ourhybrid functional calculation, but promising optical absorp-tion even higher than CsSnI3. The split-anion approach withpossible compositions of I–III–VI–VII2, II–II–VI–VII2, I–IV–VI2–VII, or II–III–VI2–VII significantly enlarges the pool of elementsfor the design of environmentally benign and high-perform-ance solar absorber materials.
Acknowledgements
This work was supported by the US National Science Foun-dation (NSF) under Grant No. CBET-1510948. SZ acknowledgessupport from the US Department of Energy (DOE) under GrantNo. DE-SC0002623. JL acknowledges the financial support ofthe US NSF under Grant No. DMR 1151028. PZ acknowledgessupport from US NSF under Grant No. DMR-0946404 andDMR-1506669, and the National Natural Science Foundationof China under Grant No. 11328401. The supercomputer timewas provided by the National Energy Research Scientific Com-puting Center (NERSC) under DOE Contract No. DE-AC02-05CH11231, the Center for Computational Innovations (CCI)at RPI, and the Center for Computational Research (CCR) atthe University at Buffalo, SUNY.
References
1 A. Kojima, K. Teshima, Y. Shirai and T. Miyasaka, J. Am.Chem. Soc., 2009, 131, 6050–6051.
2 J.-H. Im, C.-R. Lee, J.-W. Lee, S.-W. Park and N.-G. Park,Nanoscale, 2011, 3, 4088–4093.
3 H.-S. Kim, C.-R. Lee, J.-H. Im, K.-B. Lee, T. Moehl,A. Marchioro, S.-J. Moon, R. Humphry-Baker, J.-H. Yum,J. E. Moser, M. Gratzel and N.-G. Park, Sci. Rep., 2012, 2,591.
4 J. Burschka, N. Pellet, S.-J. Moon, R. Humphry-Baker,P. Gao, M. K. Nazeeruddin and M. Gratzel, Nature, 2013,499, 316–319.
5 M. Liu, M. B. Johnston and H. J. Snaith, Nature, 2013, 501,395.
6 H. Zhou, Q. Chen, G. Li, S. Luo, T.-b. Song, H.-S. Duan,Z. Hong, J. You, Y. Liu and Y. Yang, Science, 2014, 345, 542–546.
7 N. J. Jeon, H. G. Lee, Y. C. Kim, J. Seo, J. H. Noh, J. Lee andS. I. Seok, J. Am. Chem. Soc., 2014, 136, 7837–7840.
8 J. H. Heo, D. H. Song, H. J. Han, S. Y. Kim, J. H. Kim,D. Kim, H. W. Shin, T. K. Ahn, C. Wolf, T.-W. Lee andS. H. Im, Adv. Mater., 2015, 27, 3424–3430.
9 M. Gratzel, Nat. Mater., 2014, 13, 838–842.10 Z. Lin, N.-G. Park and G. Li, J. Mater. Chem. A, 2015, 3,
8924–8925.11 T.-B. Song, Q. Chen, H. Zhou, C. Jiang, H.-H. Wang,
Y. Yang, Y. Liu, J. You and Y. Yang, J. Mater. Chem. A, 2015,3, 9032–9050.
12 F. Hao, C. C. Stoumpos, D. H. Cao, R. P. H. Chang andM. G. Kanatzidis, Nat. Photonics, 2014, 8, 489–494.
13 N. K. Noel, S. D. Stranks, A. Abate, C. Wehrenfennig,S. Guarnera, A.-A. Haghighirad, A. Sadhanala, G. E. Eperon,S. K. Pathak, M. B. Johnston, A. Petrozza, L. M. Herzand H. J. Snaith, Energy Environ. Sci., 2014, 7, 3061–3068.
14 M. H. Kumar, S. Dharani, W. L. Leong, P. P. Boix,R. R. Prabhakar, T. Baikie, C. Shi, H. Ding, R. Ramesh,M. Asta, M. Graetzel, S. G. Mhaisalkar and N. Mathews,Adv. Mater., 2014, 26, 7122–7127.
15 D. Sabba, H. K. Mulmudi, R. R. Prabhakar,T. Krishnamoorthy, T. Baikie, P. P. Boix, S. Mhaisalkar andN. Mathews, J. Phys. Chem. C, 2015, 119, 1763–1767.
16 Y.-Y. Sun, M. L. Agiorgousis, P. Zhang and S. Zhang, NanoLett., 2014, 15, 581–585.
17 I. Repins, M. A. Contreras, B. Egaas, C. DeHart, J. Scharf,C. L. Perkins, B. To and R. Noufi, Prog. Photovoltaics: Res.Appl., 2008, 16, 235–239.
18 S. Chen, X. G. Gong, A. Walsh and S.-H. Wei, Appl. Phys.Lett., 2009, 94, 041903.
19 T. K. Todorov, K. B. Reuter and D. B. Mitzi, Adv. Mater.,2010, 22, E156–E159.
20 I. Grinberg, D. V. West, M. Torres, G. Gou, D. M. Stein,L. Wu, G. Chen, E. M. Gallo, A. R. Akbashev, P. K. Davies,J. E. Spanier and A. M. Rappe, Nature, 2013, 503,509–512.
21 S. H. Porter, Z. Huang, S. Dou, S. Brown-Xu, A. T. M.G. Sarwar, R. C. Myers and P. M. Woodward, Chem. Mater.,2015, 27, 2414–2420.
22 G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6,15–50.
23 J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett.,1996, 77, 3865–3868.
24 S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J. Chem.Phys., 2010, 132, 154104.
25 J. Heyd, G. E. Scuseria and M. Ernzerhof, J. Chem. Phys.,2003, 118, 8207–8215.
26 J. Heyd, G. E. Scuseria and M. Ernzerhof, J. Chem. Phys.,2006, 124, 219906.
27 A. Poglitsch and D. Weber, J. Chem. Phys., 1987, 87, 6373–6378.
28 S. Nosé, J. Chem. Phys., 1984, 81, 511–519.29 T. Baikie, Y. Fang, J. M. Kadro, M. Schreyer, F. Wei,
S. G. Mhaisalkar, M. Graetzel and T. J. White, J. Mater.Chem. A, 2013, 1, 5628–5641.
30 P. Umari, E. Mosconi and F. De Angelis, Sci. Rep., 2014, 4,4467.
31 J. P. Perdew, M. Ernzerhof and K. Burke, J. Chem. Phys.,1996, 105, 9982–9985.
32 D. West, Y. Y. Sun and S. B. Zhang, Appl. Phys. Lett., 2012,101, 082105.
33 S. D. Stranks, G. E. Eperon, G. Grancini, C. Menelaou,M. J. P. Alcocer, T. Leijtens, L. M. Herz, A. Petrozza andH. J. Snaith, Science, 2013, 342, 341–344.
34 N. J. Jeon, J. H. Noh, W. S. Yang, Y. C. Kim, S. Ryu, J. Seoand S. I. Seok, Nature, 2015, 517, 476–480.
35 J. H. Heo, D. H. Song and S. H. Im, Adv. Mater., 2014, 26,8179–8183.
36 W. Shockley and H. J. Queisser, J. Appl. Phys., 1961, 32,510–519.
37 I. Chung, B. Lee, J. He, R. P. H. Chang andM. G. Kanatzidis, Nature, 2012, 485, 486–489.
38 M. S. Hybertsen and S. G. Louie, Phys. Rev. B: Condens.Matter, 1986, 34, 5390–5413.
