The differences in crystal structure and phase of lead-free perovskite solar cell materials

Hybrid organic–inorganic lead halide perovskite materials have established themselves as a competitive solar cell technology, with high efficiencies and simple processing. However the main drawback of these materials is currently their stability, which is complicated by the potential to release the toxic element lead into the environment. Attempts to replace lead for less-toxic starting materials have generated significant research interest. Any candidate material must have enhanced stability and comparable performance to lead halide perovskites, and an understanding of the structure is key to developing these materials to be competitive. This review provides a brief overview and reference of experimentally determined crystal structures of lead-free perovskite materials. The most promising potential materials are identified as those that maintain a cubic structure, particularly double-perovskite type absorbers.


Perovskite solar cell introduction
The discovery of hybrid organic-inorganic lead-halide materials' photovoltaic activity has led to a significant new area of research: Perovskite Solar Cells (PSC) [1].This term is used for solar cell absorber materials that possess the perovskite crystal structure, originally based on CaTiO 3 [2].During their research journey, perovskite materials have found themselves to have many commercially beneficial properties, e.g.: abundant starting materials, simple processing methods, [3][4][5] and low energy costs of device production [6,7].Recently progress has been made in upscaling PSC device production, the solution processable nature of the starting materials lending itself to printing the absorber layer [8][9][10].
Perovskite solar cell efficiencies have also increased greatly since their inception in 2009, efficiencies of over 25% have now been recorded, making them competitive with commercially available technologies [1,11].One of the key developments that has increased cell efficiency is changing the ABX 3 components from the original methylammonium (MA) lead iodide [1].Through the use of additives such as formamidinium (FA), [12][13][14] caesium [15][16][17][18], among many others in various combinations [19][20][21][22][23][24], the structure and band gap can be altered to produce desirable efficiency and stability improvements [25][26][27].A more pronounced band gap change is observed when the halide anion is mixed; varying the ratio of iodide to bromide can lead to colour control from black, to red and yellow [28,29].
One constant feature of high-performance perovskite solar cells has been the cation that occupies the B-site of the perovskite, lead.Whilst minimal amounts of lead are used-and indeed it is not the largest environmental cost of manufacture, which is currently the device fabrication procedures-studies have shown that lead leached into the environment from perovskite solar cells can be taken up by organisms in nature [6,[30][31][32][33][34].This is exacerbated by the current longterm stability of perovskite absorber materials.Degradation mechanisms of the absorber material can be triggered by heat, [35,36] moisture [37], and light [38,39].Not only is this detrimental to device performance, but also increases the chances of lead being released into the environment through water soluble lead being washed away from the cells [40].
Tactics to improve the stability are currently manifold.The first of these is to alter the structure of the perovskite itself to prevent degradation through substitution.Previous studies have shown that cations of various size can be incorporated into the perovskite structure; Low molar ratio substitutions at the A, B and X sites can lead to significant improvements in performance and stability [23,41].However by altering the ions used, the crystal structure of the perovskite material can undergo significant changes, which can be predicted using the Goldschmidt tolerance factor.The tolerance factor predicts the viability of a combination of A, B and X components for the formation of a perovskite phase by calculating a structural ratio; a value between 0.9 and 1.0 would predict a cubic perovskite to form and can be calculated in different ways, for example including octahedral factor corrections to address changes in metal-halide bonding [42][43][44][45].As an example using the most basic approach, methylammonium lead iodide (Tolerance factor 0.9) possesses a tetragonal (I4/mcm) perovskite structure at room temperature, transitioning to cubic/pseudocubic at higher temperature and hexagonal at low temperatures, as shown in Fig. 1 [46].Formamidinium lead iodide (Tolerance factor 1.03) has a cubic (Pm 3 m) structure in its black phase [47].Weber et al. showed that a mixture of formamidinium and methylammonium as the A-site cation can form a continuous solid solution at various ratios of formamidinium:methylammonium, demonstrating phase change from tetragonal to cubic [48].
The ability to control the structure through ion substitutions is particularly important in the case of formamidinium lead iodide, which readily degrades into a non-photoactive δ-phase at room temperature [14,47].Stabilising the photoactive cubic phase through substitutions of bromide, caesium and methylammonium is what forms the state-of-the-art triple cation perovskite, Cs 0.05 (FA 0.83 MA 0.17 ) 0.95 Pb(I 0.83 Br 0.17 ) 3 , hence understanding the structure of perovskite materials is vital in developing high-performance devices [41].The importance of the higher symmetry cubic phase in FAPbI 3 is obvious, with the associated change in colour and therefore band gap, and photoactivity.The formation of cubic structures as opposed to lower symmetry tetragonal or hexagonal structures in perovskite materials leads to many differences [51].While it has been found that the tetragonal-cubic phase change through temperature does not contribute to a significant change in band gap [52], controlling the metal-halide bond distance chemically does; controlling the band gap to be closest to the ideal thermodynamic efficiency limit value of 1.34 eV will enable improved performances [28,53].The increased disorder of the methylammonium cation in the cubic phase could impact charge transport [54], has been attributed to performance increases, although it may affect overall stability [50,[55][56][57][58].
For enhancing perovskite stability, if the A-site cation substituted in is large enough, it can cause the separation of the structure into distinct 2D, 1D or 0D perovskites [59][60][61][62].2-Dimensional perovskites consist of layers of lead-halide octahedra separated by the large organic cation, which can act as a buffer layer preventing degradation-such techniques have been used to create devices that are stable over a year [63,64].
A second method for improving cell stability, and therefore preventing lead egress, is to alter the device architecture.Using a porous carbon electrode that can be backfilled with the absorber not only improves humidity resistance, it also means significantly less starting material is required (compared to e.g.spin-coating) [65][66][67].Carbon-electrode perovskite solar cells have been demonstrated to maintain stability over 1000 h [68].
Lead sequestration, using a chemical approach to absorb any lead lost from the perovskite material is a growing area of research that aims to prevent lead being lost to the environment [69].This can be achieved through the including phosphate salts within the perovskite material itself, that form a water insoluble salt with lead that prevents it washing into the environment [70].Alternatively interlayers or encapsulants that absorb lead can be used; which have shown no Fig. 1 a The tetragonal crystal structure of methylammonium lead halide, reproduced with permission from reference [49] and b PXRD demonstrating the phase changes with temperature, reproduced from reference [50] with the permission of the Royal Society of Chemistry impact on optoelectronic properties, and been able to significantly reduce any lead contamination [71,72].It has also been found that soil can effectively sequester lead-however some remains available to biological organisms within it.[73] Perhaps the simplest way to prevent lead being lost into the environment is to use alternative materials; the challenge within is to maintain the beneficial structure and properties of the lead-based perovskite materials.Figure 2a shows a pictorial representation of the alternative elements explored so far.In this review the structure of different lead-free perovskite materials will be summarised, with a focus on those synthesised and determined experimentally.Lead-free perovskites are a growing research topic within the perovskite solar cell community, and have seen a rise in publications since 2016 (Fig. 2b).At the start of each section a table is provided containing the abundance of the material and information on its toxicity, taken from materials safety data sheet information.Where data is not available, this does not necessarily mean they are non-toxic, it is that there is no data on these materials yet.

Group 14
As it is in the same group as lead, and only one row up in the periodic table, tin was the obvious first choice to replace lead in perovskite solar cell materials, and is currently the most extensively studied replacement.Tin possesses a similar ionic radius to lead, hence using tolerance factor as a guide can form a perovskite with many of the components already used in lead-based perovskite materials [44,75,76].Although tin is an abundant element, it is less so than lead, resulting in more expensive starting materials which would lower the effect of a key advantage of perovskite solar cells, their relatively low cost of production.There are also concerns about the environmental safety of tin-perovskite materials, showing similar toxicity to lead-based materials (Table 1), although the bioavailability of tin upon leakage into the environment has been demonstrated to be much lower than lead [31,77].Taking an additional step up group 14 in the periodic table brings you to another potential substitute for lead in germanium, an element even less abundant, with a significantly smaller ionic radius [44,75].Germanium does have fewer environmental concerns than both tin and lead [78].
Before their application in photovoltaics, organic tin-halide perovskites had been identified as potentially interesting materials for similar reasons to those for lead; that they are low cost, require simple processing methods and possess good transport properties [79,80].Early tin-halide perovskites tended to be lower dimensionality, with larger cations included to improve stability [79,81].The reduced charge transport capabilities inherent in lower dimensionality perovskite materials perhaps impeded the application of tin-halide perovskites for photovoltaics.
CsSnI 3 is one of the earliest recorded tin perovskites used in a solar cell-in this case a Schottky cell with no transport materials that exhibited 0.9% efficiency [82].At room temperature CsSnI 3 possesses an orthorhombic structure; a cubic structure is formed by X-site substitution to bromide (a monoclinic structure is made by CsSnCl 3 ) [83][84][85].High temperatures of over 150 °C are required to transition CsSnI 3 into a cubic phase [83].By switching from caesium to methylammonium, and incorporating the tin-perovskite absorber layer into a mesoporous titania scaffold, efficiencies of over 6% could be attained [86].The structure of methylammonium tin halide perovskites shows similar trends to some of their lead-analogues, i.e. there is a hexagonal phase and a cubic black perovskite phase; these transitions are displayed in Fig. 3 [87].
Both MASnI 3 and MAPbI 3 possess orthorhombic, tetragonal and cubic phases-MASnI 3 exists in the tetragonal (P4mm) structure at room temperature, with remarkably similar lattice parameters to MAPbI 3 [50,[86][87][88].The tetragonal-cubic phase transition occurs at a slightly lower temperature for MASnI 3 , at around 20 °C [89].Possessing phase-transition temperatures within the expected range a photovoltaic device may be expected to endure is not ideal, as it can affect properties such as resistivity [89].Interestingly, methylammonium tin iodide has been found to have some metallic properties, as opposed to being purely semiconducting [80,90].This has been attributed to hole-doping that occurs during material crystallisation and degradation to metallic tin [89,91].As is the case when Cs + is the A-site cation, using bromide instead of iodide produces a material with a cubic structure [92].Single crystals of formamidinium tin iodide have also been synthesised, which are orthorhombic at room temperature (space group Amm2), and as expected for a larger cation show a slightly expansion of the lattice compared to methylammonium [87,93].Currently, a significant barrier to more widespread research of tin-based materials is the rapid degradation caused by the tin cation greatly preferring the + 4 oxidation state [87,94,95].Issues of material stability have been addressed in lead-based materials by the formation of 2D materials or through surface passivation [96][97][98][99].Such phases can be formed with tin in the B-site using bulky organic cations such as phenylethenammonium (PEA); issues of charge transport inherent in materials where the inorganic backbone is disrupted were reduced slightly through artificial hole doping [100,101].Degradation to the + 4 oxidation state can also be slowed through other means.Using NABH 4 , a reducing agent, as an additive tin-perovskite materials based on FASnI 3 have been engineered to be stable for over 1000 h, with an initial efficiency of ~ 9% [102].
The next row up in the periodic table from tin is germanium -although germanium possesses a significantly smaller ionic radius than lead, at 73 pm compared to 119 pm.The result of this is that germanium iodide perovskites CsGeI 3 , MAGeI 3 and FAGeI 3 exhibit a trigonal/rhombohedral (space group R3m) structure, shown in Fig. 4 [103][104][105][106]. CsGeI 3 undergoes a phase transition to a cubic structure at 277 °C; if the iodide is substituted to chloride ions the phase transition temperature is brought below 200 °C [105].Reducing the size of the X-site anion also reduces the phase transition temperature in lead-based perovskites [107].A further similarity can be found when using large A-site cations, similar manipulation of the dimensionality of the structure can be obtained.2D-perovskites can be formed when an A-site cation such as n-butylammonium is used, overall producing an orthorhombic structure of germanium iodide sheets separated by the organic cation [100].Stoumpos et al.'s analysis of the structure of germanium based perovskites also discovered the presence of 1D perovskite structures when using organic cations such as guanidinium (Fig. 4c) and trimethylammonium [104].Germanium based materials are sparsely investigated, likely due to the increased instability of the + 2 state (even compared to tin).
An alternative approach to the problem of metal cation oxidation is to produce materials that rely on the B-site cation being in its more stable + 4 oxidation state to begin with.Materials such as Cs 2 SnI 6 (which is the degradation product of CsSnI 3 ) [87] form vacancy ordered double perovskites, illustrated in Fig. 5, a structure which essentially involves the removal of every other B-site cation [108,109].This is based on the antifluorite structure exhibited by e.g.Na 2 O-however double perovskite has become a common term for this particular structure in photovoltaic literature.Usefully tolerance factor calculations can still be applicable when looking at double-perovskites [110].Standard double-perovskites possess a cubic structure of space group Fm 3 m [111].While A 2 BX 6 materials have been less well investigated than ABX 3 (or indeed A 2 B'B''X 6 ), some structural analysis and variation of the crystal structure has been performed.Most tend to exhibit the same cubic structure as Cs 2 SnI 6 , this structure is preserved with halide substitution in the Cs-containing material [112,113].Hybrid organic-inorganic double perovskites have also been synthesised, MA 2 SnI 6 , MA 2 SnBr 6 and MA 2 SnCl 6 all form the standard cubic structure for this class of materials, with a slightly expanded lattice compared to caesium [112,114].The only experimentally determined exception to the cubic structure, Rb 2 SnI 6 forms a tetragonal structure (P4/mnc) at room temperature which transitions to a monoclinic lattice at low temperatures (< 120 K) [115,116].At the time of writing no equivalent materials using germanium have been synthesised, although computational calculations suggest that germanium iodide double perovskites would have suitable band gaps for single-junction solar cells [117].

Summary
The focus of using group 14 elements has been as a straight replacement for Pb, although currently Sn and Ge based materials have shown similar problems with stability and toxicity, with no efficiency improvements to compensate.Vacancy-ordered double perovskite materials may be an advantageous use of tin: these materials tend to exhibit high stability, and many have appropriate band gaps [108,118].While they have currently been employed as hole conductors rather than absorber layers, options for band-gap engineering could widen the possible application of these materials [119,120].

Group 15 and double perovskites
Taking one step to the right of the periodic table from lead comes to bismuth, which has been studied as a potential substitute to lead as it has a similar electronic configuration and has been assessed to be less environmentally damaging (Table 2) [121].For Group 15 materials, the + 3 oxidation state is the most favoured for elements such as bismuth and antimony, so the perovskite structure must accommodate this.This means a different ratio of A:B:X ions is prevalent.Importantly, the ionic radii of bismuth alongside other typical ions used in perovskites for photovoltaics fit tolerance factor calculations of what may produce a viable perovskite [122].
Perovskite materials based on Bismuth Iodide as the B and X ions generally possess a deep red colour; revealing a larger band gap in the above 1.9 eV.Single crystals made of methylammonium bismuth iodide and caesium bismuth iodide show the same structure.Hexagonal crystals are formed of space group P6 3 /mmc, in which two face-sharing octahedra of bismuth iodide, [Bi 2 I 9 ] [3−] are neutralised by three surrounding methylammonium or caesium cations, shown in Fig. 6 [123][124][125][126].The short bismuth to bismuth distance involved leads to some structural distortion in the face-sharing octahedra of methylammonium bismuth iodide [123].This type of structure is referred to as zero-dimensional, as there is no continuation in any axis direction.Changes to the crystal structure are observed when reducing the size of the A-site cation to ammonium/rubidium, or when reducing the size of the X-site anion to bromine.The structure of ammonium bismuth iodide was analysed by Sun et al. to be monoclinic; it also had a marginally lower band gap than methylammonium bismuth bromide at 2.04 eV [127].Rubidium bismuth iodide also produced a monoclinic crystal structure [128].
When replacing the iodide anion in methylammonium bismuth iodide with bromide, the single crystal structure changes; there are no longer face-sharing octahedra, individual [BiBr 6 ] 3− units are formed, with three methylammonium ions [127].Solar cells made using only bismuth as the B-site ion have shown some photovoltaic activity, but efficiencies still remain below 1% [122,[129][130][131].The stability of these materials is however improved compared to lead [132].
Antimony has also been investigated as the B-site anion, and has shown marginally higher efficiencies when used in solar cells compared to analogous bismuth materials [133][134][135].Antimony-based perovskite materials share a 0D structure with bismuth, furthermore MA 3 Sb 2 I 9 and (NH 3 ) 3 Sb 2 I 9 show same hexagonal and monoclinic structures respectively to their bismuth analogues [133,134].Rb 3 Sb 2 I 9 is also monoclinic, and was found to create more ordered layers of octahedra in the bulk material compared to an equivalent material using cesium [135].It should however be noted that antimony is also highly toxic, so does not immediately solve the problems presented by using lead-based materials [136,137].
Formation of a double perovskite structure can be achieved through the addition of a second, B' cation to the mixture [138].Previously we have looked at vacancy-ordered double perovskites when using B-site cations which strongly favour the + 4 oxidation state.The following double-perovskite materials all use a combination of two different cations, one in its + 3 oxidation state and another in a + 1 oxidation state, with a general formula A 2 B'B''X 6 [138][139][140].Iodobismuthate structures are potentially the most exciting of the lead replacement opportunities, as AgBi 2 I 7 , which forms a cubic structure, produces cells which have reached efficiencies of > 4% in thin-film solar cells [141,142].Silver bismuth materials have also found success when synthesised as quantum dots, which has the potential to create new ways to deploy solar cell technologies [143][144][145][146].While the synthesis, composition and shape of individual quantum dots can vary significantly, [139] this review is focussed on the structure of the bulk materials.As mentioned with the tin-based perovskites, double-perovskites based on an A 2 B'B''X 6 exhibit a cubic structure.While there is a large amount of possible substitution in A 2 B'B''X 6 , those with silver and bismuth in the B' and B'' position respectively share an Fm 3 m space group [147-149].Cs 2 AgBiBr 6 is the most extensively studied of these materials-see Fig. 5 for general structure (analogous Fm3m structure to A 2 B 2 X 6 ) [150]-and currently produces cells with ~ 2.5% efficiency [149,151].The metal halide backbone of Cs 2 AgBiBr 6 ideally consists of alternating octahedra with bismuth and silver at the centre; however defects and vacancies produced during synthesis can lead to disorder in A 2 B'B''X 6 type materials, producing changes in properties such as the band gap [149,[152][153][154]. Another common feature of silver-bismuth double perovskites to their lead and tin counterparts is the presence of a phase-transition from tetragonal to cubic.In the case of Cs 2 AgBiBr 6 , this transition happens well below room temperature, at ~ 120 K [155].The temperature of this phase transition can be altered by a few degrees Kelvin through partial substitution at the A-site with alkali metal cations such as rubidium and sodium [148].Such substitutions were also found to have minimal effect on the overall lattice parameters.
In the context of perovskite solar cells, which typically use low cost starting materials, silver is relatively expensive, and so alternatives for the B' site have been investigated.As alkali metals form ions with a suitable + 1 charge, they can also inhabit the B' position on a double perovskite.Additionally halide salts of these elements are generally abundant and cheap.Cs 2 NaBiCl 6 possesses the same cubic structure as Cs 2 AgBiBr 6 [147].One step down the group with potassium, (MA) 2 KBiCl 6 forms a rhombohedral structure R 3 m, with corner sharing KCl 6 and BiCl 6 octahedra in a 3D network [156].Copper, as the element above silver in the periodic table, is another viable candidate.However the starting materials CuCl, CuBr, and CuI all are hazardous to the environment.If the main goal of discovering new perovskite solar cell compounds is to alleviate the toxicity problems inherent with lead, alternative starting materials should also be critically analysed.Although copper has been predicted computationally to form a double perovskite with bismuth, at the time of writing no A 2 CuBiX 6 compounds have been synthesised [138,157,158].
Double perovskites can also be produced with indium as a substitute to bismuth, in Cs 2 AgInCl 6 the same cubic structure is formed, however the band gap at over 3.3 eV would not be suitable for photovoltaic applications [159].Cs 2 NaGaBr 6 and Cs 2 LiGaBr 6 have been computationally identified as potentially suitable materials with band gaps of 1.76 and 1.97 eV respectively [160].

Summary
Group 15 perovskite materials mainly feature bismuth.As a replacement to the readily available lead, there are issues of abundance and starting material cost that need to be considered when employing bismuth, or indeed antimony, based materials.There is a significant variety in potential materials; standard, single B-site perovskites tend to exist in lower dimensionalities which will likely cause issues with charge transport within the absorber layer.Double perovskites are a good alternative that possess an ordered, cubic structure and a continuous inorganic backbone.These materials typically have higher band gaps, so additional band-gap engineering should be explored, or they could be used in other devices such as LEDs or photodetectors [152,161].

Transition metals
Transition metals would be an ideal substitute for lead-based compounds, as top-row transition metals are largely abundant (Table 3).Titanium, iron, manganese, zinc, cobalt, nickel, zirconium, calcium and magnesium can possess suitable oxidation states, and a number of computational studies have identified a series of transition and group 2 metal perovskites as having significant potential for high-efficiency devices [162][163][164][165][166][167].Computational calculations have facilitated more targeted experimental synthesis of single and double perovskites, and can also provide information to aid transport layer selection (by calculating conduction band and valence band energies) [163].For example in the work by Jacobs et al., after generating potential perovskite materials their measurements firstly refine for thermodynamic stability, before refining based on band-gap, and finally solar cell efficiency.From computational calculations, materials using the relatively abundant magnesium and iron as the B-site ion have been found to have suitable structures (crystal structures of various magnesium perovskites using different A-site cations are shown in Fig. 7).KFeI 3 was identified as a material that could reach efficiencies of over 24% [162].However when simple, solution based synthesis was attempted, no material was formed; such materials may require more intense synthesis methods [168].For example Cs 3 Fe 2 Br 9 was synthesised in an autoclave for 3 days [169].This formed a hexagonal, zero-dimensional perovskite (P6 3 /mmc) similar to bismuth analogues, except with a more suitable band gap for photovoltaics of 1.65 eV.Due to the abundant nature of iron, such materials would be a useful avenue of further research.
In the 'original' perovskite, CaTiO 3 , titanium is in its + 4 oxidation state.This is also true of titanium halide perovskites for solar cells, which mostly follow the same cubic vacancy-ordered double perovskite structure, A 2 TiX 6 , shown in Fig. 5 Table 3 Abundance and toxicity of elements discussed in "Transition metals" Section, with lead included.

Element
A bundance (ppm in Earth's crust) 74    1 3 [179].Perovskite materials with titanium in the + 2 oxidation state have been synthesised, RbTiBr 3 and CsTiBr 3 both forming a hexagonal perovskite, although there have been few follow-up studies on optoelectronic properties or photovoltaic performance [179,180].Computational studies have suggested that both double perovskite Cs 2 TiX 6 and ATiX 3 materials may show good PV activity when paired with the correct transport layers [181,182].Fully inorganic Cs 2 TiBr 6 is the most well studied of the vacancy-ordered double perovskite titanium materials, with a band gap of 1.82 eV and an experimental efficiency of 3.3% [172,183].Although compared to an equivalent band-gap lead-based perovskite it has been shown that Cs 2 TiBr 6 is relatively stable to heat and light, [172] unfortunately it still degrades rapidly in air leading to the formation of the CsBr starting material peak on a timescale of 1-2 h [184,185].

Toxicity
While mostly studied as a dopant (including in bismuth perovskites), [186][187][188] perovskites using manganese as the sole B-site cation have also been synthesised [189].Methylammonium manganese bromide (MAMnBr 3 ) possesses a hexagonal perovskite structure, which on exposure to moisture forms a hydrated monoclinic structure in the same manner as hydrated MAMnCl 3 and hydrated CsMnCl 3 , shown in Fig. 8 [170,190].Using chloride anions can also produce a metal-poor MA 2 MnCl 4 material that exists in an orthorhombic structure [170,191].Switching to the much larger iodide cation produces an orthorhombic compound with a tetrahedral coordination of the Mn 2+ ion (Space group Pbca, Fig. 8b), [170] however manganese iodide as a starting material is still environmentally toxic.
In the same work, Daub et al. investigated various nickel-based perovskite materials [170].Caesium and potassium nickel halides had been previously analysed to exhibit hexagonal perovskite structures [192,193].When increasing the size of the A-site cation to methylammonium, an orthorhombic structure is formed [194,195].Exposure of these materials to moisture leads to the formation of 1-dimensional hydrated structures, which consist of chains of corner sharing octahedra that rely on hydrogen bonding to co-ordinate the separate chains.The water molecules involved cause an ~ 125° to form between octahedra [170].Such hydrated structures possess a monoclinic perovskite structure [190].
A detailed structural study of methylammonium cobalt halide materials found that (CH 3 NH 3 ) 2 CoX 4 also generally crystallised as a monoclinic structure [196].However in this case cobalt halide octahedra are not formed-instead the material has [CoX 4 ] 2− tetrahedra.
Similar structures have been discovered when using mercury or zinc as the metal [197,198].These tetrahedra do not share a corner or a face, and are stacked in the same orientation in the c axis.This structure holds when mixing chloride and bromide anions, and with mixed bromide/iodide materials.100% iodide containing compounds, and iodide mixed with chloride show two different structures, orthorhombic and an undefinable structure based on distorted hexagonal layers respectively.
Copper can also exist in appropriate oxidation states for perovskite formation and can show significant variance in crystal structure depending on the A or X-site cations used, or synthesis method of the crystal [199][200][201].Partial substitution of bromide into methylammonium copper chloride materials causes a lattice change from monoclinic to orthorhombic, accompanied by a significant change in the c lattice parameter compared to pure chloride, from 9.97 to 19.1 Ǻ for a material containing 50% bromide [201].It is the pure chloride methylammonium copper perovskite (exhibiting a monoclinic P2 1 /a structure) [202] that has so far shown the greatest photovoltaic efficiency at 2.41%; in this case (unlike with lead analogues) substitution of bromide or iodide serves to reduce the efficiency [177].Caesium copper halides tend to form in the orthorhombic Pnma structure, unless synthesised at room temperature at which point a tetragonal (I4/mmm) structure is formed [200].The only second row transition metal that has been analysed at the time of writing, is Zirconium-which also forms a double perovskite structure [203,204].While nanocrystals of Cs 2 ZrCl 6 and Cs 2 ZrBr 6 were found to be phase pure, bulk powders contained a degree of impurity phases [203].

Summary
While some transition metal replacements for lead would be beneficial due to relative abundance, cost and low-toxicity, they don't regularly form the structures that have been shown to be most beneficial for photovoltaic performance.Further to this transition metal perovskites tend to exhibit wide-bandgaps, which would not be suitable for high-efficiency single junction devices, but may find some use in multijunction solar cells.

Future outlook/conclusions
The most efficient lead-based perovskite materials generally possess a cubic structure.Table 4 shows a summary of some materials of interest throughout the report grouped by crystal structure.Of the materials discussed in this review, some tin materials and the 'double perovskite' materials all show the same cubic structure, and may be more relevant for future research than the strained/distorted trigonal or hexagonal structures shown by germanium and manganese respectively.Particularly those materials such as Cs 2 SnI 6 in which the B-site cation is already in its more stable, + 4 oxidation state may be of significant interest.Also band-gap engineering through the inclusion of multiple A-site or X-site cations on tin-based materials should be examined.On the topic of stability, lower-dimensional perovskites when using a suitable A-site cation can cause a longer lifetime in perovskite materials, however long-term studies of such lead-free materials are yet to be performed.Finally, there are a large number of potential perovskite materials that have been identified by computational studies but not yet synthesised.Attempting to produce these materials would generate broader knowledge of the material class as a whole, and help determine the viability of lead-free perovskite materials.
Acknowledgements S.R. Pering thanks Loughborough University for funding.
Author contributions S.P. wrote and edited manuscript.

Fig. 2 a
Fig. 2 a Periodic Table of the Elements.B-site elements discussed in this review are highlighted with respect to their relevant section and b representation of lead-free publications related to each group with starred value representing percentage of 'lead-free' perovskite solar cell publications compared to total perovskite solar cell literature, data taken from Web of Science-data for 2023 until June 2023

Table 1
Abundance and toxicity of elements discussed in "Group 14" Section, with lead included

Table 2
Abundance and toxicity of elements discussed in "Group 15 and double perovskites" Section, with lead included.Data for Exposure limits taken from SDS data (Merck) for the most relevant metal iodide salt and HSE.gov.ukwhere available.Elements with lower limits than lead highlighted in red Fig. 6 The structure of Methylammonium Bismuth Iodide, MA 3 Bi 2 I 9 .Reproduced from Ref. 99 with permission from the Royal Society of Chemistry Data for toxicity taken from SDS data (Merck) for the most relevant metal iodide salt (data for cobalt is taken from cobalt bromide) and HSE.gov.ukwhere available.Elements with lower limits than lead highlighted in red

Table 4
A summary of some materials of interest looked at throughout the report, grouped by crystal structure and highlighted in the colour of the corresponding section