Mixed Pt-Ni Halide Perovskites for Photovoltaic Application

1. Introduction

In recent years, perovskite photovoltaic technology has offered enormous viability and dimensionality in solar cell research. Perovskite, as a light-harvesting active layer, generated remarkable development in device efficiency of 25.6% in the single-junction solar cell, and over 33% in perovskite/silicon tandem solar cell [1,2]. Also, all-perovskite tandem solar cell is showing great potential in device performance and so far achieved a power conversion efficiency (PCE) of 26.4% with a wide bandgap (WBG) FA_0.8Cs_0.2Pb(I_0.62Br_0.38)₃ perovskite as the top subcell (1.8 eV) and thermally mixed Sn/Pb narrow bandgap (NBG) FA_0.7MA_0.3Pb_0.5Sn_0.5I₃ perovskite as the bottom subcell (1.2 eV) [3]. Transitioning photovoltaic technology from the laboratory to commercial products, high power conversion efficiency, low cost, long lifetime, and low toxicity are some of the critical factors to consider during material selection [4]. Pb-halide perovskites have been the most studied compositions in next-generation photovoltaics due to their excellent optoelectronic properties, such as the highest-power conversion efficiency (PCE) and ideal bandgap [5,6,7]. However, the practical relevance of these materials is hindered as they offer multifarious disadvantages, including toxicity, high water solubility and bioavailability, and thermodynamic instability of CH₃NH₃PbI₃ in air [4,6,8,9]. To address the toxicity issue, Pb-free perovskite compounds have been the mainstay of the perovskite research study. Among the several alternative cations to Pb from group-14 elements, the Sn-based perovskite absorbers are the widely studied alternative which work superior in achieving high efficiency due to its isoelectronic configuration of s²p² similar to Pb and smaller radius (1.35 Å and 1.49 Å in Sn²⁺ and Pb²⁺ respectively) [10,11]. Due to its smaller size than Pb, incorporating Sn in ASnPbX₃ system increases the tolerance factor and decreases the bandgap [12]. However, the high oxidation tendency from Sn²⁺ to Sn⁴⁺ upon exposure to air and the energy band levels mismatch between the perovskite and charge transport layers generate large recombination at the grain boundaries in Sn-based perovskites and hence lower PCE than that of Pb-based perovskites [10,13]. Recently, transition metal-based double perovskite, Cs₂PtI₆, has emerged as another potential alternative to Pb-free perovskite owing to their experimentally produced direct narrow bandgap of 1.37 eV [14] and 1.4 eV [15]. Our previous work reported an atmospherically processable Cs₂PtI₆ perovskite device with an excellent absorption coefficient of 4 × 10⁻⁵ cm⁻¹ and the highest efficiency of 13.88% [15]. A high V_oc comparable to Pb-based perovskite device and a high minority carrier lifetime of over 2.8 μs with ethylene diamine (EDA) achieved in our study revealing Cs₂PtI₆ as a competitive Pb-alternative for high-performance halide perovskite solar cells (HPSCs). Our theoretical investigation revealed key strategies, such as eliminating parasitic losses and optimizing band offset, to achieve a Cs₂PtI₆ photoabsorber with a PCE over 26% [16]. However, due to the high cost of Pt, the Cs₂PtI₆ perovskite is viable mostly as a model system [15]. Several other transition metals, such as copper (Cu), silver (Ag), nickel (Ni), palladium (Pd), etc., have been reported in a few studies. Nag et al. studied inorganic double perovskite with monovalent Ag and trivalent Bi (i.e., Cs₂AgBiX₆), showing analogous optoelectronic properties similar to CsPbX₃ with eco-friendly nature and long carrier lifetime [17]. However, it exhibits indirect bandgap with low optical properties leading to a low power conversion efficiency [18,19]. Soni et al. [20] numerically studied several transition metal-based halide double Cs₂ZSbX₆ perovskites for photovoltaic applications, with Z= Ag and Cu. The bandgap reduction by replacing Ag (2.542 eV) with Cu (1.299 eV) in the B-site and X= Cl was attributed to the high absorption of incident photons in the broad optical spectrum within the framework of their DFT analysis [20]. Having high dielectric constants of about 5.33 and 6.30 for Cs₂AgSbI₆ and Cs₂CuSbI₆ respectively and being optically active in the visible and ultraviolet regions, these materials can be productively utilized for optoelectronic devices [20]. CsNiCl₃ and CsNiBr₃ perovskites exhibiting low electronic bandgap and dispersive band edges are expected to offer attractive photovoltaic characteristics [21]. The use of precious metals impedes the commercialization progress of perovskite solar cells. Non-precious transition metals are promising candidates for the counter electrode of perovskite solar cells owing to their cost-performance ratio. Low-cost non-precious transition metals are investigated in several studies to replace expensive metals, such as gold (Au) or silver (Ag), as counter electrode materials in perovskite solar cells. Wang et al. [22] prepared perovskite solar cells with transition metals, such as Cu and Ni, and presented satisfactory performance with the power conversion efficiency of 13.04 and 12.18%, respectively, compared to that of 15.97% of the perovskite solar cell with Ag counter electrode. Ni, having a very close work function (∼5.04 eV) to that of Au (∼5.1 eV), showed a power conversion efficiency (PCE) of 10.4%, very comparable to devices with Au electrodes (11.6%) [23]. Cu-based perovskites are particularly advantageous because of their lower toxicity, magnetic properties [24], enhanced structural flexibility, and greater stability against light and humidity compared to Sn-based perovskites [25,26]. Several research groups have explored the use of Cu-based layered perovskites in the application of superconductors [25,27,28]. Cu, having a higher work function (~4.63 eV) than Ag (~4.23 eV), offers higher voltage output [22]. Perovskite materials also show promise in gas sensor applications due to their unique electrical and catalytic properties [29,30]. Among various target gases, the detection of hydrocarbons is significantly important in various applications. For example, monitoring dissolved gases (CH₄, C₂H₄, C₂H₂, CO, CO₂, and H₂) by gas-in-oil analysis in transformer provides important information about transformer status [31,32]. Moreover, the detection and control of ethylene gas in agriculture are extremely important as emissions of this gas indicate the maturity state of fruits [33]. Ni-based perovskites are potential candidates for gas sensing applications, as NiI₂ exhibits large impedance change in the ultra-low humidity environment [34]. Several research groups have developed lanthanum-transition metal perovskites, such as La-Co-based perovskite systems in the application of catalytic oxidation and combustion [35,36,37]. However, not enough information is available in the literature on Ni-based perovskites for photovoltaic applications. For the A-site cations, formamidinium (FA), methyl ammonium (MA), and cesium (Cs) are considered the most preferable elements to form perovskite structures due to the preferred tolerance factor in the range of 0.8 to 1. [FAPbI₃ (t ~ 0.99), CsPbI₃ (t ~ 0.8) MAPbI₃ (t ~ 0.9)] [38,39]. CH₃NH₃I has a high decomposition rate into CH₃I and NH₃ at low temperatures [40], and the unstable photoactive black cubic phase of FA-based perovskites transitions into the photoinactive yellow phase at room temperature [41]. Cation-enabled perovskite black phase stabilization by partially incorporating inorganic Cs⁺ cation has been proven effective to enhance the photo and moisture stability of perovskite [42]. The purpose of this study is to assess how partially replacing Pt in cesium platinum triiodide (CsPtI₃) perovskite with different concentrations of earth-abundant and low-cost nickel (Ni) influences its crystallographic and optoelectronic properties. Considering a partial replacement of Pt²⁺ sites by Ni²⁺ should not emerge severe lattice distortion due to a similar ionic radius of Ni²⁺ in comparison to the Pt²⁺ (Ni²⁺ vs Pt²⁺: 72 pm vs 80 pm) and we assume Ni²⁺ could likely incorporate within the perovskite crystal lattice, given high solid solubility of Pt and Ni. Cai et al. [43] performed first-principles calculations of halide perovskite-derived A₂BX₆ inorganic compounds to investigate the trends in bandgaps and energetic stability with chemical compositions providing guidelines for the design of halide A₂BX₆ compounds for potential photovoltaic applications. According to it, perovskite compounds with X=I and B=Ni exhibit large energy above hull (Ehull) (29 meV) and were not experimentally observed. All experimentally reported compounds have zero or small values of Ehull, such as Cs₂NiCl₆ and Cs₂PtI₆ have zero Ehull. Adjusting the molar ratio of PtI₂: NiCl₂ in the mixed-metal perovskite may afford new pathways to tailor a cost-effective system. To our best knowledge, this is the first study on the partial substitution of Pt by Ni performed experimentally for photovoltaic application.

2. Materials and Methods

In this study, we have chemically synthesized three different perovskite compositions referred to as PtI₂-based films, mixed PtI₂-NiCl₂-based films, and NiCl₂-based films. All the films were fabricated via precursor-based solution processing under atmospheric conditions. The synthesis process of the 3 types of films is expected to be led by the following solid-state reactions, respectively:

CsI+ PtI₂= CsPtI₃,

(1)

CsI + xPtI₂ + yNiCl₂ = Cs(Pt_x, Ni_y)(I, Cl)₃,

(2)

CsI+ NiCl₂= CsNi(I₁Cl₂),

(3)

where the x-to-y ratio of 50:50 has been used. The procedure and instruments for film fabrication and testing follow our previous works [15,44]. The precursor for the mixed PtI₂-NiCl₂-based films was prepared in a 0.25 M solution by the mixture of Cesium Iodide (0.06495 g), Platinum (II) Iodide (0.05611 g) and Nickel (II) Chloride (0.0162 g) with a molar ratio of 50:50 in 1 mL of pre-heated 50-50 Dimethylformamide (DMF)- Dimethyl sulfoxide (DMSO) solvent. The precursor mixture was heated at 75 °C for 1.5 h, followed by drop-casting on preheated Tec10 substrate. Doctor blade coating technique was used to spread the solution over the preheated substrates. Films were then annealed in a vacuum oven at -15 in Hg and 100 °C for 2 h. The same procedure has been followed to fabricate the PtI₂-based and NiCl₂-based films. The thermally annealed PtI₂-based films and mixed PtI₂-NiCl₂-based films turned out in dark reddish-black color, and NiCl₂-based films turned out orange color. For the stability testing, films were exposed to a dark thermal anneal test at 100 °C for 100 hours. All the solutes were purchased from Alfa-Aesar, Haverhill, MA, USA (Cesium Iodide, Alfa Aesar CAS: 7789-17-5; Platinum (II) Iodide, Alfa Aesar CAS: 7790-39-8; Nickel (II) Chloride, Alfa Aesar CAS: 7718-54-9; and solvents from Sigma-Aldrich, St. Louis, MO, USA (DMF, Sigma-Aldrich CAS: 68-12-2; DMSO, Sigma-Aldrich CAS: 67-68-5). The films were stored in a nitrogen-filled glove box before testing. X-ray diffraction spectroscopy (XRD) measurements were conducted on a Bruker diffractometer from Bruker Corporation Billerica, MA, USA under ambient conditions using Cu K radiation. EVA toolbox and Topas were used for XRD data analysis and phase identification. Shimadzu UV-2600 spectrometer was used to perform optical transmittance and reflectance measurements. The optical bandgap of the samples was determined by Tauc analysis. JEOL JSM-5610 was used to perform Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray Spectroscopy (EDS) analysis. Raman spectra were measured at room temperature using the customized Scope Foundry-based Raman spectrometer at Molecular Foundry, Berkeley National Laboratory. The microscope was in confocal geometry with 1800g/mm grating and equipped with a silicon CCD. A continuous wave 532nm laser was used for excitation with 100uW (0.366mW/um²) and the 532nm long wave pass filter was set in the output path.

Figure 1. Atmospheric synthesis of PtI₂, mixed PtI₂-NiCl₂, and NiCl₂-based films in 50:50 DMF: DMSO via solution processing.

Figure 1. Atmospheric synthesis of PtI₂, mixed PtI₂-NiCl₂, and NiCl₂-based films in 50:50 DMF: DMSO via solution processing.

3. Results and Discussion

3.1. Cost Analysis of Solutes for Perovskite Precursor and Encapsulation for HPSCs

This cost model determines the $/watt of several perovskite material compositions, considering the molarity (M), absorber layer thickness (t), active cell area, and power conversion efficiencies (PCEs) from the respective literature. In this cost analysis, we compared the $/watt value associated with the precursor solutes required to prepare these compositions and the additional cost of four different encapsulants. Cost per gm of solutes and properties of perovskite compositions are summarized in Table S1 and S2, respectively. The molarity used in the literature for respective material compositions remains constant throughout this analysis. We also evaluated the $/watt with the value of PCE (25.6 %) and t (~ 2 µm) optimized to the value of the state-of-art FAPbI₃ HPSCs [2]. The cost per unit volume is determined by the perovskite (product) volume, the proportional amount of reactants needed to achieve the product volume, and the cost of solutes per gram. The amount of perovskite, by moles, is determined using the known molecular mass (g/mol) and density (g/cm³) of the product seen in Equation (4).

$${\displaystyle \frac{MolesofPerovskite}{c{m}^3}}\left[mol /c{m}^3\right]={\displaystyle \frac{Density\left[g/cm\right]}{Molar mass\left[g/mol\right]}}$$

(4)

Figure 2 is the graphical representation of the comparison of the $/watt of the precursor solutes required to prepare several perovskite compositions considering two cost frameworks. The 1st cost framework (blue) represents the $/watt when both the PCE and absorber layer thickness (t) were derived from the corresponding literature respective to each perovskite composition. The 2nd cost framework (red) represents the $/watt considering each perovskite composition matches both the PCE of 25.6% and t of 2000 nm as reported for the state-of-art FAPbI₃ perovskite.

It is important to note that the overall cost associated with perovskite products depends on many other several things, such as the precursor synthesis methods, processing conditions, film stability, use of additives and encapsulations, etc. Our current analysis of $/watt only associates the cost of solutes used for the corresponding precursor. According to Figure 2 (blue), the $/watt of Cs₂PtI₆ is estimated to be ~144x more expensive than that of FaPbI₃ perovskite. The synergistic effect of a thicker absorber layer of 10,000 nm and the high cost of PtI₄ chemical compound (PtI₄= $226.55/gm, PbI₂= $1.176/gm) are responsible for the high $/watt of its precursor. The distinct impact of the PCE and t on the cost modeling is presented in Figure S1. According to it, if the t of the Pt-based composition remains constant at 10,000 nm and the PCE increases to 25.6%, the $/watt reduces by almost half times the initial $/watt value. Alternatively, if the PCE remains constant at 13.88% and t reduces to 2000 nm, the $/watt of Cs₂PtI₆ reduces by almost five times the initial $/watt value with the reported PCE and t. A thicker absorber layer requires more product and hence the t and cost of the solutes play a vital role in regulating the perovskite precursor cost. Our previous study numerically optimizes the thickness of several perovskite absorbers needed to reach the state-of-the-art PCE [16]. If both the parameters are optimized (PCE to 25.6% and t to 2000 nm), the $/watt of Cs₂PtI₆ is estimated to be only ~15x more expensive than the FaPbI₃ perovskite, as indicated in Figure 2 (red) and reduces by almost nine times the initial $/watt value with the reported PCE and t.

In the most efficient HPSCs with the best PV performance, PbI₂ is the main Pb-containing decomposition product, and thus is likely the main product to leak from broken solar modules due to its easy water solubility. the U.S. Environmental Protection Agency (EPA) has identified lead as one of 15 pollutants often found in publicly owned treatment works (POTW) and sewage that it considers a potential pollutant of concern [45]. The PbI₂ solubility in water is 0.76 g L^–1 at 20 °C [46], while the maximum accepted levels of Pb in drinking water are set as 5 orders of magnitude lower, at 0.000015 g L^–1 (15 ppb) by EPA [45]. These numbers manifest the importance to limit the possible leaching of dissolved Pb-containing products from HPSCs into the environment. A strategy to mitigate Pb leaching from HPSCs into the environment is through the use of encapsulants. Also, encapsulation for solar panels is a critical issue for the long-term operational stability of HPSCs. In the second phase of this cost analysis, we evaluated perovskite cost with four different cost-effective and commercially available polymer-based encapsulants viable for academic research, such as ethylene–vinyl acetate (EVA), Polyolefin (TPO), Polytetrafluoroethylene (PTFE) known by its trade name Teflon^®, and Polyethylene terephthalate (PET).

Currently the most common polymeric encapsulant material used in commercial silicon solar modules is EVA, due to its low-cost and easy processability [47,48]. It offers low water vapor transmission rate (WVTR) compared to some other encapsulants reported in the literature but high water diffusion rate causing a possible decline in module lifetime [49]. Also, its sensitivity to discoloration under UV radiation results in decreased light-transmittance, and therefore reduced solar cell power output [48]. Polyolefin is a commonly used encapsulant in academic research and has several advantages like good adhesion energy, creep failure resistance, low WVTR, low discoloration rate and better light transmittance compared to EVA [50,51,52].

Studies report the unique use of hydrophobic fluoropolymer, PTFE, for improving the perovskite crystallinity and passivating defects when used an optimum amount of it as an additive in the perovskite organic precursor [53]. Studies also reports that hydrophobic passivation of PTFE precursor solution prevents PbI₂ decomposition and improved moisture stability [54]. However, it is sensitive to electrophilic attack upon reaction with alkali metals when exposed to long hours of heat [55]. Several studies have used PET in a hybrid encapsulation framework in a combination with transparent metal oxide films (e.g., Al₂O₃) or inorganic/organic multilayers and reported its compatibility to be used on flexible substrates [56,57,58], however it comes comparatively costlier than the previously mentioned encapsulants. The $/m² of these encapsulants is summarized in Table S3. Recent improvements in perovskite stability through these encapsulants is summarized in Table S4. There are many other effective encapsulants being studied for stability improvement in perovskite. A comprehensive cost analysis with additional highly efficient encapsulants will be discussed in a future review paper.

Figure 3 is the graphical representation of the $ per watt of solutes with added encapsulant cost of several perovskite compositions considering each perovskite composition with optimized PCE of 25.6% and t of 2000 nm as reported for the state-of-art FAPbI₃ perovskite. If both the parameters are optimized, the $/watt of (solute+ encapsulant) Cs₂PtI₆ is estimated to be only ~1 %-1.4 % more expensive than the FaPbI₃ perovskite for four different encapsulants. Among them, The most expensive PET renders a cost of ~$3.6/watt for Cs₂PtI₆ ($3.3/watt for FAPbI₃) and the least expensive Teflon renders a cost of ~$0.81/watt for Cs₂PtI₆ ($0.516/watt for FAPbI₃), which are ~2.5x and ~4.7x cheaper than the $/watt value of solutes+ encapsulants with its reported PCE of 13.88% and t of 10,000 nm as outlined in Figure S2. The calculated precursor cost of FAPbI₃ with expensive encapsulants like PET ($3.29/watt) exceeds that of the unencapsulated Cs₂PtI₆ ($2.89/watt). Figures S3 and S4 represent the $/watt value (solute+ encapsulant) with the discrete effect of the optimized PCE and optimized absorber layer thickness reported for the Pb-based FAPbI₃ perovskite, respectively.

The long-term operational stability investigated using maximum power point (MPP) tracking under a simulated 1-sun illumination for the unencapsulated FAPbI₃-based PSC reports a loss of 15% of its initial efficiency, under continuous light soaking using a LED lamp for 450 hours at around 35 °C [2]. The unencapsulated Cs₂PtI₆-based Devices tested under AM1.5G at 65 °C for 500 h shows a loss of 23% of its initial efficiency. The shunts causing a decrease in Voc and FF after light soaking are expected to develop due to pinholes in the Cs₂PtI₆ films and can be improved with film quality [15]. Considering the high water solubility of the Pb-based compound, a strong and effective encapsulant system is needed which can potentially increase the overall cost. With comparable efficiency and t, Cs₂PtI₆ can be considered a suitable alternative to Pb-based perovskites despite the high cost associated with Pt-based solutes. However, considering the high cost associated with its current PCE and t, Cs₂PtI₆ is not viable for commercialization, and exploration for alternative compounds to replace Pt has immense research significance. In this paper, we have summarized the primary results corresponding to the partial replacement of Pt with Ni in the B-site.

3.2. Pt-Ni Mixing in Halide Perovskite

The bandgap analysis of our PtI₂-based films, mixed PtI₂-NiCl₂-based films, and NiCl₂-based films outlined in Figure 4a show a bandgap of 2.13 eV for PtI₂-based films. The bandgap increases to 2.35 eV when NiCl₂ entirely replaced PtI₂. Alteration of the halide anion changes the bond distance and/or angle of X–B–X, and incorporation of smaller X-anion, such as Cl replacing I, increases the bandgap [59]. Our bandgap analysis of NiCl₂-based films replacing PtI₂ supports this theory. The partial substitution of NiCl₂ with PtI₂ in mixed PtI₂-NiCl₂-based films is supposed to exhibit a larger bandgap than that of the NiCl₂-based films upon the incorporation of the larger Pt cation in the B-site. However, the reduced bandgap of 1.9 eV in mixed PtI₂-NiCl₂-based films is likely driven by the incorporation of the larger X-anion (I- in this case) rather than the larger B-cation. The XRD analysis of the 3 types of films is shown in Figure 4b. The XRD pattern shows mixed phases of many unidentified lower-intensity diffraction peaks with a partial match to Cs₂PtI₆ resulting in a higher relative noise level in the films prepared in CsI+PtI₂ precursor and indicates poor crystallinity. A peak shift to higher angles in XRD can be attributed to the incorporation of B-site cations with larger ionic radii, leading to an expansion of the perovskite lattice [60,61,62]. Similar result is observed in our study with the incorporation of bigger Pt in the mixed PtI₂-NiCl₂-based films. The formation of Cs(Pt,Ni)(I,Cl)₃ phase in the mixed PtI₂-NiCl₂-based films is confirmed by a peak shift of standard CsNiCl₃ (a= 7.118 Å, b= 7.118 Å, c= 5.9085 Å ) to higher 2*theta. The diffraction peaks of this new phase are located in between CsNiCl₃ and Cs₂PtI₆ phases, which implies the intercalation of Pt into perovskite lattice and formation of mixed Pt-Ni-based phase. (Table 1).

XRD spectra of a new phase of CsNi(I,Cl)₃ forms with the complete substitution of Pt with Ni, suggested by a peak shift of standard CsNiCl₃ to lower 2*theta. Peaks of CsNiCl₃ located at 20.8°, 25° and 32.74° shifts to 20.51°, 24.59° and 32.3°, respectively, and forms this mixed anion phase in CsI+NiCl₂ precursor. The peak located at 27.4° in the XRD pattern of the NiCl₂-based films was identified as CsI (27.7°) with a peak shift to lower angle (27.39°). The wide bandgap in NiCl₂-based films may also be ascribed to the presence of CsI. Cross-sectional SEM images of the 3 types of films are presented in Figure 4c–e. PtI₂-based films have a long needle-like structure, which is completely different than our previous observation in films prepared in CsI+PtI₄ precursor. This SEM morphology along with the XRD pattern and bandgap provides strong evidence that we are not making a pure phase of Cs₂PtI₆ with CsI+PtI₂ precursor [15]. Raman spectroscopy measurements were also conducted on the NiCl₂-based films and PtI₂ based film, as shown in Figure 4f and g, respectively. The sharp Raman peaks detected on the NiCl₂-based film around 107, 142, 195 and 267 cm^-1 are in line with reported values from literatures [65,66,67]. The peak at 267 cm^-1 matched well with the A1g modes of CsNiCl₃; The peak near 195 cm^-1 correlated to the E_2g mode while the peak 142 cm^-1 is attributed to the E_1g modes of CsNiCl₃ [65]. As the film system also contained I- (with the introduction of CsI), the unassigned peak at ~106.63 might come from Ni-I or CsNiI₃ structure [68]. The Raman peaks on the PtI₂-based film were also observed and fitted at 92, 129, 148, 164, and 265 cm^-1, respectively. These peaks are also in line with reported values from literatures [69,70,71]. The peak at ~129 and ~148 cm^-1 should be assigned to the symmetric Pt-I stretch in ν2 mode (Eg) and ν1 mode (A_1g), respectively [69,71]. And the Raman shift peaks at ~93 and 160 cm^-1 were also reported in Hexaiododiplatinate (II) salts A₂Pt₂I₆ [69]. This structure consists of Pt₂I₆ units and the corresponding cation (Cs in our case), while the anions form edge-shared squares [72]. The peak at ~93 cm^-1 might come from the asymmetric I-Pt-I bend in ν4 mode (T1u) [71]. In order to elucidate the phase formation in PtI₂-based films, we performed EDS analysis showing the statistical distribution of calculated average at % of precursor elements in Figure S5. It indicates a phase with Cs:Pt:I in the intended precursor ratio of 1:1:3. However, no standard XRD pattern is available for CsPtI₃ in the database, so our XRD analysis could not confirm the presence of this phase. Figure 4d shows two different microstructures, a dark rod-like structure, and a transparent plate-like structure, present in mixed PtI₂-NiCl₂-based films. Similarly, NiCl₂-based films have dark rod-like microstructure with white particles on the film surface presented in Figure 4e. The EDS analysis performed in various spectrums of these films shown in Figures S6 and S7 indicates the bright white plate-like features (solid columns) have Cs-I-rich morphology, and the dark rod-like features (patterned columns) have Cl-Ni-rich morphology, which suggests that the insolubility of NiCl₂ is responsible for the formation of the rod-like features. In our previous study on mixed Sn-Pb perovskite [44], a similar microstructure was observed as a result of the coagulation tendency of the Pb compound due to the insolubility of PbI₂. The presence of excess Cs and I particles observed in EDS analysis further validates the presence of CsI as identified in the XRD spectra of the NiCl₂-based films. The EDS analysis also suggests the presence of a higher at% of sulfur in the rod-like surfaces rather than in the white Cs-I-rich surfaces, both in the mixed PtI₂-NiCl₂-based and NiCl₂-based films. Even NiCl₂-based films have almost no sulfur present in their Cs-I-rich regions. However, the influence of sulfur in the formation of rod-like features is not clear yet. Moreover, we calculated the Goldshmidt and Bartel tolerance factor for the Cs(Ni,Pt)(I,Cl)₃ films, as shown in Figure 4h and I, respectively. It should be noted that optimal stability range of the 3D perovskite structure is indicated in the window 0.8<t<0.9 for Goldshmidt factor [73], while Bartel’s tolerance factor <4.18 predicts stable perovskite phase [74]. The tolerance factors of the Cs(Ni,Pt)(I,Cl)₃ films are outside the optical range, which might explain the phase separation we observed.

In order to demonstrate the effect of long thermal annealing, we performed a heat-stability test by annealing the films for 100 hours at 65°C and compared the optoelectronic features to that of the reference films that were annealed for 2 hours. Figure 5 indicates that the wide bandgap intermediate phase transforms to a stable Cs₂PtI₆ perovskite phase after the long hour of thermal annealing. Figure 5a exhibits a significantly reduced bandgap of 1.6 eV in the PtI₂-based films exposed to long thermal annealing, which is in reasonable agreement with the bandgap of the Cs₂PtI₆ film (1.4 eV) reported in our previous work [15]. Figure S8 exhibits noticeably red-shifted absorption spectra, compared to the reference films, with an absorption edge at close to 900 nm confirming the suppressed bandgap achieved in thermally exposed films. It is known that crystallinity is critical for perovskite stability because the main defect-induced degradation starts near the grain boundaries [2]. The poor crystallinity in the reference PtI₂-based films is consistent with poor optical measurements, which is improved with long hour thermal annealing. Upon thermal annealing, the PtI₂-based films become more crystalline with a better match to the diffraction pattern of the Cs₂PtI₆ perovskite (11.361 Å) with some unreacted Cs residue (Figure 5b). The Cs₂PtI₆ phase was determined by the (111), (200), (220), (222), (400), (440), and (622) peaks. The amorphous morphology becomes compact including some Cs-rich white particles on top as confirmed by the SEM analysis in Figure 5c. The EDS analysis in Figure 5d further validates the formation of Cs₂PtI₆ in the thermally annealed films showing a phase of Cs:Pt:I at an atomic ratio of 2:1:6 with some excess Cs, which is well aligned with our XRD analysis.

The effect of thermal annealing on the optoelectronic properties of mixed PtI₂-NiCl₂-based films is demonstrated in Figure 6. Figure 6a shows red-shifted bandgap spectra rendering a bandgap of 1.85 eV, very close to the bandgap (1.9 eV) of the reference films annealed for 2 hours. Figure S9 shows thermally exposed films have a slightly red-shifted absorption opening with an absorption edge at close to 700 nm similar to reference films, possibly implying its stability against thermal annealing. The peak broadening of the XRD pattern is inversely correlated with crystallite size. Perovskite films with large crystallite sizes can have reduced grain boundaries and restrained carrier recombination, which increases carrier mobility [75]. Films exposed to extended thermal annealing have improved crystallinity, and narrow refined XRD spectra with a shift to higher angles as shown in Figure 6b, which can be correlated to reduced lattice parameters and is similar to our previous observation [44]. Similar to the reference films, thermally exposed films exhibit diffraction pattern in the same orientation of Cs(Pt,Ni)(I,Cl)₃. The peak located at 37.56° in both of these films is attributed to the CsCl phase, which can be ascribed to the wider secondary bandgap. No significant change in the XRD pattern is observed, supporting our proposition of its thermal stability speculated through bandgap analysis. According to Figure 6c,d, the rod-like morphology transformed into a compact plate-like structure displaying white Cs-rich crystals on the surface. According to the EDS analysis in Figure 6e, a significant decline in the atomic distribution of sulfur content is observed after the thermal treatment. We attribute this to the evaporation of the DMSO solvent residue upon annealing which might be responsible for the refined and crystalline XRD spectra.

The effect of thermal annealing on the optoelectronic properties of NiCl₂-based films is demonstrated in Figure 7. Figure 7a exhibits a significantly narrower bandgap of 1.68 eV in the films annealed for 100 hours, compared to the films annealed for 2 hours (2.35 eV). Figure S10 exhibits noticeably red-shifted absorption spectra with an absorption edge shifted from only 500 nm to 750 nm confirming the suppressed bandgap achieved in the thermally exposed films. These films have improved crystallinity and better match to CsNiCl₃, confirmed by a shift of CsNi(I,Cl)₃ to higher angles as shown in Figure 7b. Similar to the mixed PtI₂-NiCl₂-based films, the peak located at 37.56º is attributed to the CsCl phase. According to Figure 7c,d, the rod-like features in the reference films transforms to a more plate-like compact morphology. EDS analysis in Figure 7e confirms the suppression of I- upon thermal treatment, suggesting an agreement with the suppressed CsI as detected in the XRD analysis. The reduction in the CsI intensity may be correlated to the reduced bandgap achieved after 100 hours of thermal annealing. Similar to PtI₂-based films and mixed PtI₂-NiCl₂-based films, a decline in the sulfur content was also observed in the thermally annealed NiCl₂-based films.

Several research groups studied the excellent performance of Ni-based compounds (e.g., NiI₂) in ultra-low humidity, which is attributed to their susceptibility to humidity and the resulting material transition characteristics. Zhang et al.[34] studied the moisture-induced discoloration of NiI₂ in humidity detection. Their study suggests that this moisture-induced material transformation of NiI₂ leads to a change in bulk resistance of materials due to a change in material composition, indicating that the crystal structure of NiI₂ is not directly affected by temperature. Rather, temperature promotes the rapid desorption of water molecules from materials. Their conclusion also suggests that the hydrophilicity of this compound is reversible, meaning the transition between NiI₂ and NiI₂•6H₂O due to the absorption of water in the presence of wet environment molecules and desorption of water by removing them is reversible. Studies have also confirmed that the dark state of the film (in dry environment) leads to higher absorbance compared to the transparent state (in wet environment) [34,76]. We also observed a transparent state of the NiCl₂-based films before thermal annealing and a reversible color change (black ↔ orange) after 2 hours of thermal annealing in a vacuum chamber, as depicted in Figure S11. The films exhibit a dark state for a remarkably long period after 100 hours of thermal annealing, which is attributed to the greater stability achieved through thermal annealing. The lighter state of the film exhibits close to 40-60% optical transmittance between 350 and 1400 nm, while the dark states exhibit less than 20% optical transmittance through the visible spectrum, with a gradually increasing transmittance through near IR spectrum (Figure S12). This observation in our study is well positioned to previous studies [34,76]. However, in the case of mixed PtI₂-NiCl₂-based composition, we observe neither the transparent state nor a rapid color transformation, which triggered us to explore the optical properties of the films upon NiCl₂ inclusion.

4. Conclusions

In this study, we first analyze the precursor materials cost for state-of-art Pb and Pb-free halide perovskites and encapsulation materials. Cs₂PtI₆ shows comparable optoelectronic properties to highly efficient FAPbI₃, however, due to the high cost of PtI₄ and reported absorber layer thickness, the precursor cost is estimated to be 15x higher. Commercially available encapsulants such as Teflon, PET, EVA, and polyolefin have been used for the cost analysis. We find a trade-off between encapsulant enhanced stability of FAPbI₃ and the cost of Pt in air-stable Cs₂PtI₆. The precursor cost of unencapsulated Cs₂PtI₆ ($2.89/watt) is estimated cheaper than encapsulated FAPbI₃ with expensive encapsulants like PET ($3.29/watt) and with comparable efficiency and absorber layer thickness, PET-encapsulated Cs₂PtI₆ ($3.59/watt) is expected to cost only 1.1x more expensive. Considering the high water solubility and toxicity of Pb-based perovskites, expensive encapsulants are needed for the commercialization of FAPbI₃ solar cells. To evaluate the replacement of Pt in Cs₂PtI₆, we also explore the substitution of Pt with Ni in the second part of this study. The structure, bandgap, and stability of Cs(Pt_xNi_1-x)(I,Cl)₃ thin-films are evaluated for x = 1, 0.5 and 0. We synthesize perovskite films using doctor-blade method with CsI, PtI₂, NiCl₂ in 50-50 DMF-DMSO. The PtI₂-based films result in a bandgap of 2.13 eV and the XRD pattern shows an unidentified mixed amorphous phase with possible matches to CsPtI₃ or Cs₂PtI₆. EDS analysis confirms PtI₂-based films have Cs:Pt:I atomic ratio of 1:1:3, however, due to lack of standard XRD patterns, CsPtI₃ phase is not confirmed. Upon thermal annealing for 100 hours, PtI₂-based films result in crystalline films with the XRD pattern showing a better match to Cs₂PtI₆, also confirmed by the average atomic ratios measured by EDS. With thermal annealing, the bandgap of the film reduces from 2.13 eV to 1.6 eV, with the latter being a closer match to previously reported Cs₂PtI₆ phase. Films deposited with CsI + NiCl₂ precursors result in a bandgap of 2.35 eV which is between the reported values of 0.8 eV for CsNiCl₃ and 3.86 eV for CsI. SEM shows a mixed morphology of 2 phases: a rod-like structure identified as CsNi(I,Cl)₃ and white particles of CsI; also confirmed by XRD. After annealing for 100 hours the bandgap of NiCl₂-based films reduces to 1.65 eV, and XRD shows primarily CsNiCl₃ phase. This observation is also confirmed by change in the appearance of the films from translucent in as-deposited state to dark brown after annealing. With a 50-50 mixture of PtI₂ and NiCl₂, resulting bandgap of 1.9 eV with XRD pattern showing a close match to CsNiCl₃ with a shift to higher 2*theta, confirming the substitution of Pt into CsNiCl₃ lattice. With thermal annealing, the films show improved crystallinity, and the bandgap is stable at 1.85-1.9 eV. Our study shows promise of creating earth-abundant halide perovskites as CsNiCl₃ or Cs(Pt,Ni)(I,Cl)₃ to address the stability and toxicity issues for FAPbI₃. Future work should include study of charge transport and other optoelectronic properties of Cs(Pt,Ni)(I,Cl)₃ which shows the best stability in our study.