This journal is© the Owner Societies 2017 Phys. Chem. Chem. Phys., 2017, 19, 20619--20626 | 20619

Cite this:Phys.Chem.Chem.Phys.,

2017, 19, 20619

Tuning the gap of lead-based halide perovskitesby introducing superalkali species at the cationicsites of ABX3-type structure

C. Paduani *† and Andrew M. Rappe

We study with scalar relativistic density functional theory (DFT) calculations the effect of changes in the

ionicity of the bonding mechanism and charge donation on the structure and electronic properties of

new lead halide perovskites which show promising performance in optoelectronic applications as long-

wave infrared detectors and thermoelectrics. Our results provide evidence that the band gap of these

compounds can be tuned upon the introduction of appropriate superalkali moieties at the cationic

A-sites in the CsPbI3-type structure. The computed band gap is 0.36 eV (direct) and 0.41 eV (indirect) for

[Li3O]PbI3 and [Li3S]PbI3, respectively. By changing the chemical environment in the Pb-halide perovskite

structure, we see drastic changes in the shape of both valence and conduction bands, as compared to

CsPbI3. Introducing superalkali cations produces extra electronic states close to the Fermi level which

arise from the formation of delocalized energy states, where a strong hybridization is identified between

Pb and Li s-states near the top of the valence band. This can promote the hole mobility and increase

the exciton diffusion length at longer wavelengths. Berry phase calculations show rather significant

spontaneous polarization of 34 and 15 mC cm�2 along the x-axis in both [Li3O]PbI3 and [Li3S]PbI3,

indicative of ferroelectric behavior.

Introduction

Despite the fact that perovskites with ABX3 structure have longbeen studied, only recently have these compounds been con-sidered as the most promising candidates for applications inphotovoltaics and other optoelectronic devices.1–6 The recentinterest in photovoltaic materials has been piqued with theadvent of the organic–inorganic hybrid perovskites (OIHPs),which are low-cost emitters with very high color purity. TheOIHPs attract even greater focus because of their easy solution-based synthesis, high conversion efficiency, and outstandingperformance in applications using their luminescence andlight-emission properties, photodetection with tunable opticalband gaps, as well as ferroelectricity and piezolectricity.7–10

While the ideal perovskite is cubic11 (space group Pm%3m), themajority of perovskites are in fact distorted. Recent studiesfrom density functional theory (DFT)-based calculations indi-cate that the electronic properties, including the band gap, thegap deformation potential, and the exciton binding energy aswell as the chemical stability of OIHPs can be traced back to

their corresponding molecular motifs.12 New OIHP admixtureshave therefore been proposed by considering the replacementof halogens with superhalogen moieties having compatibleionic radii.13

The hybrid organic lead halide with the methylammonium(MA) molecular cation CH3NH3

+ in MAPbX3 has attracted aparticularly great deal of attention from researchers because,allied to its relatively low cost, it possesses remarkable solar-cellefficiency in photovoltaic applications, besides being an efficientionic conductor.1,2 It exhibits pronounced piezoelectricity due toa weak bonding interaction and rotation of the MA molecules,which have a permanent dipole due to the charge imbalancebetween the methyl radical and the ammoniumyl radical.The piezoelectric effect of MAPbI3 has been attributed to thedipole reorientation.14–16 However, the role of the MA moleculeincludes not only its own dipole and off-center displacement,17

but also the transverse shift between the Pb and I atoms.The latter enhances the separation between the positive (i.e.,Pb atoms) and negative charge centers (i.e., I atoms) inside thePbI6 octahedra, providing another contribution to the piezo-electric response of MAPbI3. The ionic disorder actually hasbeen considered to be more significant than electronic disorderin this material.17–19 However, the major drawback to its wide-spread use is its inherent instability and degradation upon expo-sure to oxygen or water. It has been shown that point defects

Department of Chemistry, University of Pennsylvania, 231 S. 34th Street,

Philadelphia, Pennsylvania 19104-6323, USA. E-mail: clederson.paduani@ufsc.br;

Fax: +55 (48) 3721 9946; Tel: +55 (48) 3721 9234

† Permanent address: DF-UFSC, Florianopolis, CEP 88040-900, SC, Brazil.

Received 31st March 2017,Accepted 19th July 2017

DOI: 10.1039/c7cp02091k

rsc.li/pccp

PCCP

PAPER

Publ

ishe

d on

19

July

201

7. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

29/0

8/20

17 0

0:06

:16.

View Article OnlineView Journal | View Issue

20620 | Phys. Chem. Chem. Phys., 2017, 19, 20619--20626 This journal is© the Owner Societies 2017

(involving Pb, I or MA) lead to shallow traps near the valenceand conduction band edges20 which may provide unintentionalp- or n-type doping, and therefore organic components otherthan MA will contribute to trap states. The size and charge ofthe molecule that occupies the perovskite A sites will obviouslyinfluence the energy barrier for ionic transport, in addition tothe fact that the size of the organic molecule is also essential forstabilizing the perovskite structure. Recent studies using amulti-scale approach for a comprehensive investigation overseveral materials have been undertaken as an effort to identifythe key parameters and mechanisms that control the basicproperties and the stability of HPs.12,13

Results of cluster calculations21–28 mimicking the alkalihalides in several materials have shown that it is possible toget quick estimations for the properties of new compounds onthe basis of simple molecular models. It has been verified thatthe structures of organic SAs indeed remain practically con-stant in all materials and hardly change from their geometriesin the gas phase. Here, we demonstrate that by changing thebuilding blocks from atomic ions to super-ions, we can createHPs with unusual properties, with the basic properties of thesematerials largely determined by the ionic radii and the bondingionicity of the super-ions occupying the cationic sites. Severalstrategies have been proposed to counter the degradation bymoisture: the lowering of the binding energy between the watermolecule and the SA halide (in order that the water moleculedoes not get trapped between the ions easily), the enhancementof the bonding between the M and X units within the MX3

structure (so that the water molecules cannot easily remove anyof the X atoms from the octahedra), and the replacement of theorganic cations of the HPs with inorganic cations having com-parably large size and proper electronegativity.

Another recent example of perovskite tunability is the rapidlygrowing family of layered perovskites, which has attracted con-siderable attention due to unique crystal structures and opto-electronic properties. Recently a solution-based cation infiltrationprocess has been developed to deposit layered perovskitestructures onto methylammonium lead iodide films to producea perovskite/perovskite heterojunction solar cell.29 By partiallysubstituting small methylammonium cations for bulkierones, the resulting compound resembles a multi-quantum wellstructure with alternating layers of corner-sharing lead halideoctahedra and sheets of long-chained hydrophobic cations.Tailoring the bulkier cation leads to a modulation of the densityof states of the material,30,31 thereby giving access to a widevariety of new optoelectronic devices. An intriguing questionwhich still persists is: what are the driving factors that controlthe band gaps of HPs? In this contribution, we investigate thestability and electronic properties of new perovskites formed byintroducing polyatomic yet aprotic SA species at the cationicA-sites of CsPbI3. We chose as candidates the SA moieties Li3Oand Li3S, which we expect to serve as promising options for theproduction of HPs for optoelectronic applications as infrareddetectors or thermoelectrics. By avoiding the H-bonding of theother OIHPs, it is also expected that these compounds couldhave good resistance to moisture.

Method

The calculations are based on DFT as implemented in theQuantum ESPRESSO code.32 The electron exchange and corre-lation are described within the semilocal generalized gradientapproximation (GGA), where the adopted pseudopotentials arethe RRKJUS (ultrasoft) type with non-linear core corrections,generated by employing the PBEsol functional. This latter isintended to improve on PBE for equilibrium properties such asbond lengths and lattice parameters. Approximate van der Waalscorrections for the treatment of dispersion interactions resultingfrom dynamical correlations between fluctuating charge distri-butions were accounted for with the zero damping DFT-D2 semi-empirical method of Grimme,33,34 where the van der Waalsinteractions are described via a simple pair-wise force field.35

A supercell model is adopted for the studied compounds whereinitially a cubic (Pm%3m) primitive-cell was used as the startingconfiguration for the SA-PbI3 perovskites. Several different initialgeometries have been considered in each case, representing thevarious possibilities of distortions, tiltings and rotations of thePbI6 octahedra as well as for the orientation of the SA cluster atthe cationic A-sites. For modeling the new perovskites, the SAcluster was then initially placed at the corners of the unit cell inseveral inequivalent orientations. The calculations for the geo-metry optimization are thus performed starting from a simplecubic supercell but however considering a variable cell duringthe relaxation procedure along the x, y, z directions. We want tostress that, in order to assure that the obtained equilibriumgeometries are indeed the ground-state structures, the relaxationprocess was performed from several different starting configura-tions. The plane wave cutoff was taken at 78 Ry. All geometries werefully relaxed, and thresholds for convergence of the total energy andforces were set to 1 � 10�6 eV per cell and 1 � 10�2 eV �1,respectively. Crystal structures are visualized using the XCrysdenprogram suite.36

Results and discussion

Examples of SA are Li2F, Li3O, Na3O, Li3S, and Li4N. In Fig. 1,we show the equilibrium geometries of the SA clusters Li3O andLi3S. These have been studied in a previous work,37 and have alsobeen experimentally obtained.38,39 In this study, these clustersare inserted at the cationic A-sites, replacing the Cs atoms in theCsPbI3 material. The final crystal unit cell geometry was obtainedby fully relaxing the unit cell. Initially, we performed extensivecalculations in supercells built up by stacking two unit cells alongall three orthogonal directions. We have tried several differenttiltings and rotations of the PbI6 octahedra, as well as six differentorientations of the SA clusters in the structure, in order to findthe lowest-energy equilibrium geometry. A volume relaxationof the initially cubic parent phase resulted in equilibriumgeometries which exhibit small tetragonal distortions forboth Li3OPbI3 and [Li3S]PbI3 (see Table 1). For cubic CsPbI3,a = 6.2894 Å.40 It is interesting to note that the structures of bothLi3O and Li3S at the cationic A-sites are essentially preservedin the equilibrium geometries of the perovskites [Li3O]PbI3

Paper PCCP

Publ

ishe

d on

19

July

201

7. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

29/0

8/20

17 0

0:06

:16.

View Article Online

This journal is© the Owner Societies 2017 Phys. Chem. Chem. Phys., 2017, 19, 20619--20626 | 20621

and [Li3S]PbI3. Nevertheless, the Li–O bond length increasesfrom 1.67 Å to 1.73 Å in [Li3O]PbI3, whereas the Li–S bondlength increases from 2.18 Å to 2.21 Å in [Li3S]PbI3. In the latter.we see the PbI6 cage slightly bent in the �y direction.

The Goldschmidt tolerance factor (a) is a dimensionless numberwhich is an indicator for the stability and distortion of crystalstructures, originally only used to describe the perovskite structure;

a � rA þ rXð Þ=ffiffiffi2p

rA þ rXð Þ� �

, where rA is the ionic radius of theA-cation, rB is the radius of the B-cation, and rX is the radius of theanion. The perovskite structure is favored by the following tolerancefactors: a = 0.90–1.0 for cubic perovskites, a = 0.80–0.89 for distortedperovskites, ao 0.80 for other types and a4 1.00 for hexagonal ortetragonal structures. In the present study, a = 0.97 for [Li3O]PbI3

and 1.07 for [Li3S]PbI3, suggesting that they are viable alternativesfor polyatomic cations in stable perovskites.

The calculated band structure of these perovskites at high-symmetry points of the Brillouin zone are presented in Fig. 2(the Fermi energy is shifted to zero). Observe that in [Li3O]PbI3

the valence band maximum and conduction band minima occurexactly at the R point, whereas in [Li3S]PbI3 we see an indirectband gap close to R. Within 1 eV below the Fermi energy EF

one sees flattened bands, a feature even more pronounced for[Li3O]PbI3. The remarkable result is that the band gap is 0.36 eVand 0.41 eV, for [Li3O]PbI3 and [Li3S]PbI3, respectively (see Table 1).The experimental results for the band gap for CsPbI3 and MAPbI3

are 3.1441 and 1.5 eV,42 respectively.

The calculated densities of states (DOS) are depicted inFig. 3. The uppermost part of the valence band spreads outover �4 to 0 eV. From these diagrams we see clearly that, in theenergy range �2 to 0 eV, the major contributions to the totalDOS of [Li3O]PbI3 arise from O 2p states, whereas in the range�4 to 2 eV these are derived from I 5p states. Correlating theseDOS results with Fig. 2, we identify the dominant contributionsto the bands in these energy ranges as arising from O andI p states. Above EF, the Pb 6p states are dominant. Similarly, for[Li3S]PbI3 we can see overwhelming contribution arising fromS 3p states in the energy range �1 to 0 eV, while the I p statesare responsible for the main contribution to the total DOS inthe range �4 to 2 eV. Furthermore, we observe in both casesthat the predominant contribution to the s DOS right belowEF comes from the Li atoms, which are seen to be stronglyhybridized with the Pb s states. Just above EF, the s states aremostly I s-states, which are also seen to be strongly hybridizedwith the Pb p states. As a common feature in these compounds,we observe that, close to the Fermi energy, the valence bands aregoverned by I and O/S p states, whereas above EF the Pb p statesdominate the lowest conduction bands.

Ferroelectric materials are characterized by spontaneouselectric polarization that is lost as the material is heated aboveTC. The permanent dipole associated with the molecular cationsin OIHPs may lead to spontaneous polarization and thus ferro-electricity and piezoelectricity. The application of ferroelectricmaterials in solar cells has long been considered, from the firstobservations of the anomalous photovoltaic effect and the bulkphotovoltaic effect43 until the recent successful application ofOIHPs in solar cells emphasizing that polar semiconductorsindeed can be used in conventional photovoltaic architec-tures.44,45 Theoretical studies have estimated the value of thepolarization density for the ferroelectric phase of MAPbI3 atroom temperature to be of the order of a few mC cm�2. Polariza-tion is induced by the rotation of the methylammonium ionsinside the PbI6 cages. A recent study shows that the piezoelectriccoefficient depends sensitively on the molecular ordering.16 Bycomparing organometallic halide perovskites with differentatomic substitutions, it has been observed that the displacementof the B-site cation contributes to the piezoelectric response aswell as the competition between A–X hydrogen bonds and B–X

Fig. 1 Top: Optimized geometries for (a) Li3O and (b) Li3S; bottom: cubicABX3 perovskite structures with superalkali clusters occupying the A sites,Pb atoms occupying the B sites, and I atoms occupying the X sites.

Table 1 Calculated parameters for the lead halide perovskites

Latticeconstants (Å)

Cellvolume (Å3)

Electric polarization(mC cm�2)

Gap(eV)

a b c O Px Py Pz Dg

[Li3O]PbI3 6.93 6.88 6.99 333.27 34.0 9.2 7.4 0.36[Li3S]PbI3 6.62 6.61 6.51 284.86 15.0 �5.5 7.8 0.41

Fig. 2 Band structure along symmetry directions for (a) [Li3O]PbI3, and(b) [Li3S]PbI3.

PCCP Paper

Publ

ishe

d on

19

July

201

7. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

29/0

8/20

17 0

0:06

:16.

View Article Online

20622 | Phys. Chem. Chem. Phys., 2017, 19, 20619--20626 This journal is© the Owner Societies 2017

metal–halide bonds. According to the modern theory ofpolarization,46 the spontaneous polarization of a ferroelectriccrystal or the macroscopic polarization of a crystalline dielectriccan be calculated as a Berry’s phase of the electronic Blochwavefunctions. We performed Berry’s phase calculations to findthe electric polarization. The components of the polarizationvectors are listed in Table 1. For [Li3O]PbI3, one can see a rathersignificant polarization in the xy-plane, with the larger compo-nent along the x direction (34 mC cm�2). For [Li3S]PbI3, we seealso a similar in-plane polarization, with largest component of15 mC cm�2, also along the x direction.

We also performed a Lowdin population analysis in orderto determine partial atomic charges in these compounds.

For [Li3O]PbI3, the valence electronic charges assigned to the Li,O, Pb and I atoms are 0.27 e, 7.00 e, 3.41 e, and 7.32 e (average),respectively, which imply net charges of +0.73 e, �1.00 e, +0.59 e,and �0.32 e, respectively. The Li–O bonding thus remainsprimarily ionic in these perovskites. For [Li3S]PbI3, the valenceelectronic charges assigned to the Li, S, Pb and I atoms are0.35 e, 6.93 e, 3.43 e, and 7.29 e, which suggest net chargesof +0.65 e, �0.93 e, +0.57 e, and �0.29 e, respectively. In[Li3O]PbI3, the net charge on the SA cluster is thus +1.19 e,whereas in [Li3S]PbI3 it is +1.02 e. From our previous study,37 thevertical ionization potentials (VIPs) for Li3O and Li3S are 3.86 eVand 4.52 eV, respectively. The lower VIP of [Li3O]PbI3 directlyleads to its larger donation of charge as an A-site cation. The net

Fig. 3 Projected densities of states for [Li3O]PbI3 (left) and [Li3S]PbI3 (right).

Paper PCCP

Publ

ishe

d on

19

July

201

7. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

29/0

8/20

17 0

0:06

:16.

View Article Online

This journal is© the Owner Societies 2017 Phys. Chem. Chem. Phys., 2017, 19, 20619--20626 | 20623

charge in the PbI3 cage in this perovskite is �0.37 e, whereas in[Li3S]PbI3 this is �0.30 e.

Despite the greater charge transfer in [Li3O]PbI3, its unit cellvolume is larger by E14%. This is due to the larger dimensions ofthe planar trigonal Li3O cation as compared with the pyramidalLi3S (Fig. 1). This translates into larger PbI6 cages in [Li3O]PbI3,where the Pb–I bond lengths are larger (3.47 Å) than in[Li3S]PbI3 (3.38 Å). We therefore observe that, in these materials,the polarization is due to a combined effect of moderately-enhanced effective ionic charges in addition to the alignmentof the polar SA moieties. The primarily ionic character of thebonding mechanism and the asymmetric orientations of thepolyatomic ions in the structure lead to tetragonal distortion,and thereby lack of inversion symmetry (noncentrosymmetry) ofthese compounds.

As shown in Fig. 2, the introduction of the SA clusters Li3Oand Li3S replacing the Cs atoms in CsPbI3 leads to the for-mation of new bands in the energy range �2 to 0 eV in both[Li3O]PbI3 and [Li3S]PbI3 compounds, as compared to CsPbI3.47

Within 1 eV below the valence band edge, these are ascribed tothe O and S p states, and as a common feature we see stronghybridization with the Li 2s states. In addition, we see tendencyfor the formation of flattened bands below �2 eV. In the con-duction band, we see accumulation of bands around the Rpoint in both [Li3O]PbI3 and [Li3S]PbI3. From a close inspectionof the shape of the band plots in Fig. 2 we can see that theintroduction of SA species replacing Cs atoms promotes thecarrier mobility and increases the exciton diffusion length atlonger wavelengths, relatively to CsPbI3. Moreover, there is aremarkable decrease of the computed band gap in these newmaterials (0.36 eV and 0.41 eV), as compared to CsPbI3 (2.27 eV).47

The accumulation of bands close to EF is even more pronouncedin [Li3S]PbI3 (see Fig. 2b), which has a stronger hybridizationbetween Pb and Li s states. Furthermore, a distinctive feature forthese new compounds is the combination of the two peaks in the[Li3O]PbI3 DOS below EF into a prominent broader peak at about�0.5 eV for [Li3S]PbI3, as one replaces S for O. The change in thedonation of bonding charge which arises from the SA substitutionfor the Cs atoms affects drastically the band structure of thevirtual bound states (above EF) in both [Li3O]PbI3 and [Li3S]PbI3,and therefore we see a shift upwards of the conduction bandfor I s- and Pb p-states, which is particularly seen even morepronounced at the R-point in [Li3S]PbI3.

We performed test cluster calculations with the Gaussian-09package48 at the B3LYP/6-311+G(2df) level of theory in order tostudy the structure and stability of clusters formed by the SAspecies Li3O or Li3S in the presence of a water molecule. InFig. 4 we show HOMO (highest occupied molecular orbital)plots and natural atomic orbital (NAO) occupancies computedby the population analysis. The Li3O cluster keeps its structuralintegrity after the addition of the water molecule, despite somechanges in the NAO charges on the Li and O atoms. Note thatboth Li3O and H2O are neutral. On the other hand, the Li3S clusteris significantly distorted and is disrupted upon the addition of theH2O molecule. One H atom strongly bonds with the S atom,whereas the OH radical attracts two Li atoms, giving an Li2OH

moiety with a net charge of +0.38 e. The characteristics of thebonding mechanism are depicted in the HOMO plots, where wecan see much weaker bonding at the HOMO level for Li3O�H2O.These features are indicative of less moisture sensitivity for thiscluster. The dissociation energies of the Li3O and Li3S clustershave been calculated in a previous work:37 2.1 eV and 1.5 eV,respectively. These give an estimate of their thermodynamicstability relatively to decomposition into plausible small frag-ments. The significant calculated values indicate good stabilityfor these clusters.

Furthermore, in order to get more insight on the stability ofthese compounds, we performed also cluster calculations at theB3LYP/SDD/6-311+G(2df) level to study the enthalpy changeDH (internal energy/per atom) for fragmentation along selectedchannels. The clusters adopted to represent the solid phasesare built up with the same geometry (when available) of thesolid phases where all dimensions are accurately preserved.For LiOPbI3 and LiSPbI3 we performed geometry optimizationprocedure. We run also frequency calculations in order to checkfor imaginary frequencies, which indeed do not occur. InTable 2 we can see that the largest DH (�0.72 eV) for [Li3O]PbI3

is seen along the decomposition channel Li3OPbI3 - LiOPbI3 +2Li. On the other hand, for [Li3S]PbI3 we see that the largest DH(�0.60 eV) takes place along the reaction Li3SPbI3 - LiSPbI3 + 2Li.These results show that both [Li3O]PbI3 and [Li3S]PbI3 systemsare somewhat unstable. However, the results of the frequencycalculations with all vibrational modes having positive frequencysuggest that these hypothetical structures are moderately unstableand at least locally stable (metastable). In addition, we performedalso DFT calculations of the vibrational frequencies (normalmodes/phonons) with Quantum ESPRESSO. The calculatedRaman spectra for our compounds are showed in Fig. 5. Theinfrared (IR) spectra are very weak, as one should expect for thesenoncentrosymmetric systems which have missing inversionsymmetry. For a molecule with a center of symmetry, vibrationswhich are infra-red active are Raman forbidden, and vice versa.

Fig. 4 HOMO plots and NAO charges for (a) Li3O� � �H2O, and (b) Li3S� � �H2O.

PCCP Paper

Publ

ishe

d on

19

July

201

7. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

29/0

8/20

17 0

0:06

:16.

View Article Online

20624 | Phys. Chem. Chem. Phys., 2017, 19, 20619--20626 This journal is© the Owner Societies 2017

There can also be vibrations forbidden in both, but noneallowed in both effects. In the present study, the obtainedresults show normal modes for weaker bands at the low-frequency side of the spectrum for both materials. They getrapidly stronger as the displacement gets larger for the stretch-ing modes, thus indicating strong absorption. In the Ramanplots of Fig. 5, for both compounds, we see that the low-frequency modes, which are indicative of inherent instability,have very low amplitudes. The stronger resonances are asso-ciated with higher frequencies, which, allied to the absenceof negative frequencies, are thus indicating thermodynamicstability. Besides, these results also indicate that a larger stabilityis expected for [Li3O]PbI3, which present smaller number of low-frequency modes.

It is noteworthy to mention that, according to the Laporterule, in a centrosymmetric environment, electronic transitionsthat conserve parity between like atomic orbitals such as s–s,p–p, d–d, or f–f transitions are forbidden. Actually, mostforbidden transitions are only relatively unlikely, and althoughelectric dipole transitions can be nominally forbidden, there is

a small probability of their spontaneous occurrence. Dipole-forbidden transitions between energy levels may also occurbased on other mechanisms such as quadrupole transitions.In the present study, we show that the SA substitution intro-duce drastic changes in the shapes of both valence and con-duction bands, comparatively to CsPbI3, and gives rise to theonset of extra energy states close to EF. The formation ofdelocalized energy states is favored through a strong hybridiza-tion between Pb and Li s-states, in addition to the s–p hybri-dization, near the top of the valence band. Hence, despite theoverwhelming contribution of p-orbitals close to the Fermienergy, the reduced symmetry in this noncentrosymmetricenvironment allows the valence orbitals to mix and hybrids toform, the lobes of which point along the bonding directions. Theso formed bonding molecular orbitals have symmetry character-istics in common with respect to the axis of the bond. Theresulting breaking of symmetry is thus a determining factor forallowing optical transitions which are parity forbidden. Never-theless, one expects that the p–p transition is disallowed by thesymmetry rule when both p orbitals are on the same atom.Finally, the results above show that the introduction of SAspecies at the cationic A-sites in the CsPbI3-type structure oflead-based halide perovskites is favorable for reducing the bandgap and thereby extending optical absorption and photovoltaicresponse to long wavelength radiation. These features areexpected to contribute to the lowering of the effective mass ofholes, improving the hole mobility and increasing exciton diffu-sion length. We hope this study will stimulate further experi-mental investigations to exploit new possibilities for synthesizingnovel structures.

Conclusions

We investigate the structure and electronic properties of newlead-based halide perovskites formed by introducing the SAclusters Li3O and Li3S in place of the Cs atoms in the cubicstructure of CsPbI3. The results show that the equilibriumcrystal structures of both [Li3O]PbI3 and [Li3S]PbI3 are slightlytetragonally distorted. A remarkable result is that the band gapsin both systems are much smaller than that of cubic CsPbI3. Thesmaller gaps arise from the strong hybridization between Li andPb s-states at the top of the valence band. The band gap isslightly larger in [Li3S]PbI3 because there occurs a shift of bothconduction and valence states towards higher energies, relativelyto [Li3O]PbI3. Cluster calculations indicate that the [Li3O]PbI3

compound is expected to be more resistant to water exposure.The present results demonstrate unequivocally that the bandgap in lead halide perovskites indeed can be reduced signifi-cantly by inserting appropriate A-site cations in the ABX3 struc-ture. These new materials have noncentrosymmetric structures(lacking inversion symmetry) which classifies them as potentiallypromising candidates in applications as ferroelectrics, commu-nications, remote sensing in an extended spectrum region intothe near and short-infrared spectrum, which is important in thenight vision as long-wave infrared detectors.

Table 2 Enthalpy (internal energy/atom) change on decomposition alongselected channels calculated at the B3LYP/SDD/6-311+G(2df) level

Channel DH/atom (eV)

Li3OPbI3 - Li2O + PbI2 + LiI �0.32Li3OPbI3 - LiOPbI3 + 2Li �0.72Li3OPbI3 - 3LiI + PbO �0.33

Li3SPbI3 - Li2S + PbI2 + LiI �0.26Li3SPbI3 - LiSPbI3 + 2Li �0.60Li3SPbI3 - 3LiI + PbS �0.16

Fig. 5 Calculated Raman spectra.

Paper PCCP

Publ

ishe

d on

19

July

201

7. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

29/0

8/20

17 0

0:06

:16.

View Article Online

This journal is© the Owner Societies 2017 Phys. Chem. Chem. Phys., 2017, 19, 20619--20626 | 20625

Acknowledgements

This research was supported by grants from the ConselhoNacional de Desenvolvimento Cientıfico e Tecnologico (CNPq),Brazil. Computational support from the Centro Nacional deProcessamento de Alto Desempenho da Universidade Federaldo Ceara (CENAPAD-UFC) and Centro Nacional de Super-computaçao da Universidade Federal do Rio Grande do Sul(CESUP), Brazil are gratefully acknowledged. A. M. R. acknowl-edges support from the U.S. Department of Energy, Office ofBasic Energy Sciences, under grant DE-FG02-07ER46431.

References

1 C. Eames, J. M. Frost, P. R. F. Barnes, B. C. O’Regan, A. Walshand M. S. Islam, Ionic transport in hybrid lead iodide perov-skite solar cells, Nat. Commun., 2015, 6, 7497–7503.

2 J. M. Azpiroz, E. Mosconi, J. Bisquert and F. De Angelis,Defect migration in methylammonium lead iodide and itsrole in perovskite solar cell operation, Energy Environ. Sci.,2015, 8, 2118–2127.

3 P. Delugas, A. Filippetti and A. Mattoni, Methylammoniumfragmentation in amines as source of localized trap levelsand the healing role of Cl in Hybrid Lead-Iodide Perovskites,Phys. Rev. B: Condens. Matter Mater. Phys., 2015, 92, 045301.

4 W. Xu, L. Liu, L. Yang, P. Shen, B. Sun and J. A. McLeod,Dissociation of Methylammonium Cations in HybridOrganic-Inorganic Perovskite Solar Cells, Nano Lett., 2016,16, 4720–4725.

5 X. Y. Zhu and V. Podzorov, Charge Carriers in HybridOrganicInorganic Lead Halide Perovskites Might BeProtected as Large Polarons, J. Phys. Chem. Lett., 2015,6(23), 4758–4761.

6 J. Li and P. Rinke, Atomic structure of metal-halide perovs-kites from first principles: The chicken-and-egg paradox ofthe organic-inorganic interaction, Phys. Rev. B: Condens.Matter Mater. Phys., 2016, 94, 045201.

7 K. T. Butler, J. M. Frost and A. Walsh, Ferroelectric materialsfor solar energy conversion: photoferroics revisited, EnergyEnviron. Sci., 2015, 8, 838–848.

8 H. Cho, S.-H. Jeong, M.-H. Park, Y.-H. Kim, C. Wolf, C.-L.Lee, J. H. Heo, A. Sadhanala, N.-S. Myoung, S. Yoo, S. H. Im,R. H. Friend and T.-W. Lee, Overcoming the electrolumi-nescence efficiency limitations of perovskite light-emittingdiodes, Science, 2015, 350, 1222–1225.

9 X.-Y. Zhu and V. Podzorov, Charge Carriers in HybridOrganic-Inorganic Lead Halide Perovskites Might BeProtected as Large Polarons, J. Phys. Chem. Lett., 2015, 6,4758–4761.

10 S. Sherkar and L. J. A. Koster, Can ferroelectric polarizationexplain the high performance of hybrid halide perovskitesolar cells?, Phys. Chem. Chem. Phys., 2016, 18, 331–338.

11 C. Paduani, Electronic structure of the perovskite-type nitrideRuFe3N, J. Magn. Magn. Mater., 2004, 278(1–2), 231–236.

12 H. Fang and P. Jena, Super-ion inspired colorful hybridperovskite solar cells, J. Mater. Chem. A, 2016, 4, 4728–4737.

13 H. Fang and P. Jena, Molecular Origin of Properties ofOrganic-Inorganic Hybrid Perovskites: The Big Picture fromSmall Clusters, J. Phys. Chem. Lett., 2016, 7, 1596–1603.

14 G. Giorgi, J.-I. Fujisawa, H. Segawa and K. Yamashita,Cation Role in Structural and Electronic Properties of3D OrganicInorganic Halide Perovskites: A DFT Analysis,J. Phys. Chem. C, 2014, 118(23), 12176–12183.

15 Y.-B. Lu, X. Kong, X. Chen, D. G. Cooke and H. Guo, Piezo-electric scattering limited mobility of hybrid organic-inorganicperovskites CH3NH3PbI3, Sci. Rep., 2017, 7, 41860, DOI:10.1038/srep41860.

16 S. Liu, F. Zheng, I. Grinberg and A. M. Rappe, Photoferro-electric and Photopiezoelectric Properties of OrganometalHalide Perovskites, J. Phys. Chem. Lett., 2016, 7(8), 1460–1465.

17 O. Yaffe, Y. Guo, L. Z. Tan, D. A. Egger, T. Hull, C. C.Stoumpos, F. Zheng, T. F. Heinz, L. Kronik, M. G. Kanatzidis,J. S. Owen, A. M. Rappe, M. A. Pimenta and L. E. Brus, Localpolar fluctuations in lead halide perovskite crystals, Phys. Rev.Lett., 2017, 118, 136001.

18 A. Walsh, D. O. Scanlon, S. Chen, X. G. Gong and S. H. Wei,Self-Regulation Mechanism for Charged Point Defects inHybrid Halide Perovskites, Angew. Chem., 2015, 127(6),1811–1814.

19 O. Hentz, Z. Zhao and S. Gradecak, Impacts of Ion Segrega-tion on Local Optical Properties in Mixed Halide PerovskiteFilms, Nano Lett., 2016, 16, 1485–1490.

20 W.-J. Yin, T. Shi and Y. Yan, Unusual defect physics inCH3NH3PbI3 perovskite solar cell absorber, Appl. Phys. Lett.,2014, 104, 063903.

21 C. Paduani, A prospective DFT study for assembling highlyreactive clusters based on lithium boranes and alanates,Mol. Phys., 2014, 112(2), 316–323.

22 C. Paduani, DFT study of gadolinium aluminohydrides andaluminofluorides, Chem. Phys., 2013, 417, 1–7.

23 C. Paduani and P. Jena, Role of Ti-based catalysts in thedehydrogenation mechanism of magnesium borohydride:A cluster approach, Int. J. Hydrogen Energy, 2013, 38(5),2357–2362.

24 C. Paduani and P. Jena, Structure, stability and super-halogen properties of sodium and magnesium borohydrides,Chem. Phys. Lett., 2013, 556, 173–177.

25 C. Paduani and P. Jena, Super and hyperhalogen behavior inMgX n and GdX n(X = F, BF4) clusters, J. Nanopart. Res.,2012, 14(9), 1035.

26 C. Paduani and A. M. Rappe, Design of New Complexes ofInorganic Salts Based on Lithium and Magnesium Hydroxidesand Carbonates for Usage as Propellants and Flame Retardants,J. Phys. Chem. A, 2016, 120(39), 7764–7770.

27 C. Paduani, High-spin multiplicities in ferromagneticground states of supramolecular halide complexes basedon the gadolinium chloride, Phys. E, 2016, 77, 199–205.

28 C. Paduani, Highly reactive complexes with promising strongmagnetic response based on gadolinium borates and oxyanions,Chem. Phys., 2015, 463, 52–57.

29 Y. Hu, J. Schlipf, M. Wussler, M. L. Petrus, W. Jaegermann,T. Bein, P. Muller-Buschbaum and P. Docampo, Hybrid

PCCP Paper

Publ

ishe

d on

19

July

201

7. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

29/0

8/20

17 0

0:06

:16.

View Article Online

20626 | Phys. Chem. Chem. Phys., 2017, 19, 20619--20626 This journal is© the Owner Societies 2017

Perovskite/Perovskite Heterojunction Solar Cells, ACS Nano,2016, 10(6), 5999–6007.

30 D. B. Straus, S. Hurtado Parra, N. Iotov, J. Gebhardt, A. M.Rappe, J. E. Subotnik, J. M. Kikkawa and C. R. Kagan, J. Am.Chem. Soc., 2016, 138, 13798–13801.

31 J. Gebhardt, Y. Kim and A. M. Rappe, Influence of theDimensionality and Organic Cation on Crystal and ElectronicStructure of Organometallic Halide Perovskites, J. Phys.Chem. C, 2017, 121, 6569–6574.

32 P. Giannozzi, S. Baroni, N. Bonini, M. Calandra, R. Car andC. Cavazzoni, et al., Quantum ESPRESSO: a modular andopen-source software project for quantum simulations ofmaterials, J. Phys.: Condens. Matter, 2009, 21, 395502URL.

33 T. Thonhauser, S. Zuluaga, C. A. Arter, K. Berland,E. Schrder and P. Hyldgaard, Spin Signature of NonlocalCorrelation Binding in Metal–Organic Frameworks, Phys. Rev.Lett., 2015, 115, 136402; T. Thonhauser, V. R. Cooper, S. Li,A. Puzder, P. Hyldgaard and D. C. Langreth, van der Waalsdensity functional: Self-consistent potential and the nature ofthe van der Waals bond, Phys. Rev. B: Condens. Matter Mater.Phys., 2007, 76, 125112.

34 D. C. Langreth, B. I. Lundqvist, S. D. Chakarova-Kack, V. R.Cooper, M. Dion, P. Hyldgaard, A. Kelkkanen, J. Kleis, L. Kong,S. Li, P. G. Moses, E. Murray, A. Puzder, H. Rydberg, E. Schrdeand T. Thonhauser, A density functional for sparse matter,J. Phys.: Condens. Matter, 2009, 21(8), 084203.

35 S. Grimme, Semiempirical GGA-type density functionalconstructed with a long-range dispersion correction, J. Comput.Chem., 2006, 27, 1787–1799.

36 A. Kokalj, XCrySDen-a new program for displaying crystal-line structures and electron densities, J. Mol. GraphicsModell., 1999, 176, 17–26.

37 C. Paduani and A. M. Rappe, Assemblage of SuperalkaliComplexes with Ever Low-Ionization Potentials, J. Phys.Chem. A, 2016, 120(32), 6493–6499.

38 O. M. Neskovic, M. V. Veljkovic, S. R. Velickovic, L. T.Petkovska and A. A. Peric-Grujic, Ionization Energies ofHypervalent Li2F, Li2Cl and Na2Cl Molecules Obtained by

Surface Ionization Electron Impact Neutralization MassSpectrometry, Rapid Commun. Mass Spectrom., 2003, 17,212–214.

39 K. Yokoyama, N. Haketa, H. Tanaka, K. Furukawa andH. Kudo, Ionization Energies of Hyperlithiated Li2F Mole-cule and LinFn�1 (n = 3, 4) Clusters, Chem. Phys. Lett., 2000,330, 339–346.

40 D. M. Trots and S. V. Myagkota, High-Temperature StructuralEvolution of Caesium and Rubidium Triiodoplumbates,J. Phys. Chem. Solids, 2008, 69, 2520–2526.

41 O. N. Yunakova, V. K. Miloslavskii and E. N. Kovalenko,Exciton Absorption Spectrum of Thin(KI)1�x(PbI2)x films,Funct. Mater., 2013, 20, 59–63.

42 T. Baikie, Y. Fang, J. M. Kadro, M. Schreyer, F. Wei, S. G.Mhaisalkar, M. Graetzel and T. J. White, Synthesis andCrystal Chemistry of the Hybrid Perovskite (CH3NH3)PbI3

for Solid-State Sensitized Solar Cell Applications, J. Mater.Chem. A, 2013, 1, 5628–5641.

43 L. Z. Tan, F. Zheng, S. M. Young, F. Wang, S. Liu andA. M. Rappe, Shift current bulk photovoltaic effect in polarmaterials-hybrid and oxide perovskites and beyond, npj Comput.Mater., 2016, 2(1–12), 16026.

44 Y. Zhou, F. Huang, Y. B. Cheng and A. Gray-Weale, Photo-voltaic performance and the energy landscape of CH3NH3PbI3,Phys. Chem. Chem. Phys., 2015, 17, 22604–22615.

45 Ferroelectric Materials for Solar Energy Conversion: Photo-ferroics Revisited,K. T. Butler, J. M. Frost and A. Walsh,Energy Environ. Sci., 2015, 8, 838–845.

46 R. D. King-Smith and D. Vanderbilt, Theory of polarizationof crystalline solids, Phys. Rev. B: Condens. Matter Mater.Phys., 1993, 47, 1651–1654.

47 R. A. Jishi, O. B. Ta and A. A. Sharif, Modeling of LeadHalide Perovskites for Photovoltaic Applications, J. Phys.Chem. C, 2014, 118(49), 28344–28349.

48 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A.Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucciand G. A. Peterssonet al., Gaussian 09, Revision E.01, Gaussian,Inc., Wallingford, CT, 2009.

Paper PCCP

Publ

ishe

d on

19

July

201

7. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

29/0

8/20

17 0

0:06

:16.

View Article Online