Towards environmental friendly multi-step processing of efficient mixed-cation mixed halide perovskite solar cells from chemically bath deposited lead sulphide

Organic–inorganic hybrid perovskite is the most promising active layer for new generation of solar cells. Despite of highly efficient perovskite active layer conventionally fabricated by spin coating methods, the need for using toxic solvents like dimethylformamide (DMF) required for dissolving low soluble metal precursors as well as the difficulties for upscaling the process have restricted their practical development. To deal with these shortcomings, in this work, lead sulphide as the lead metal precursor was produced by aqueous chemical bath deposition. Subsequently, PbS films were chemically converted to PbI2 and finally to mixed-cation mixed halide perovskite films. The microstructural, optical and solar cell performance of mixed cation mixed halide perovskite films were examined. Results show that controlling the morphology of PbI2 platelets achieved from PbS precursor films enabled efficient conversion to final perovskite films. Using this processing technique, smooth and pin hole-free perovskite films having columnar grains of about 800 nm and a bandgap of 1.55 eV were produced. The solar cell performance consisting of such perovskite layers gave rise to a notable power conversion efficiency of 11.35% under standard solar conditions. The proposed processing technique is very promising towards an environmentally friendly method for the production of large-scale high efficient perovskite solar cells.

, PbSe 25 and PbS [26][27][28][29] , have been developed via different deposition approaches including chemical, physical and electrochemical methods. Table S1 summarizes different metal precursors, their processing towards the fabrication of PSC and highest PCE of the cell reported.
Lead sulphide (PbS) is a non-halide compound with a lower bond dissociation energy (3.3 eV) compared to oxide precursors 30 and therefore more chemically reactive which facilitates its subsequent reduction to final perovskite films. PbS can be deposited by various techniques such as vacuum evaporation 31 , spray pyrolysis 32 , successive ionic layer adsorption and reaction 33 , electrodeposition 34 , molecular beam epitaxy 35 , and chemical bath deposition (CBD) 36,37 .
Further to the methods already developed for depositing PbS, some of them have been employed in combination with solution processing to produce final lead halide perovskite film and PSCs. Sutherland et al. 26 have reported atomic layer deposition of lead sulphide (PbS) layer for the first time. The deposited PbS was a 75 nm thick film introduced as a precursor material to be subsequently converted into the final perovskite layer 26 . Successive exposure to iodine vapor and methylammonium bromide was used to directly convert two-dimensional (2D) PbS nanocrystals into nanocrystals of hybrid perovskite maintaining 2D morphology 38 . Radio-frequency sputtering assisted solution process has also been employed for depositing PbS films which were converted into the perovskite layer by exposing in an iodine atmosphere at room temperature, followed by immersing in a methylammonium iodide solution to be transformed to the perovskite layer 27 . CBD technique combined with chemical vapor deposition (CVD) has been proved to be applicable for synthesizing lead halide perovskite films whose power conversion efficiency has been reported 4.68% in a PSC 28 . We have recently demonstrated that lead halide perovskite films and cells were successfully fabricated based on electrodeposited PbS 29 with a champion efficiency of 7.72%. The proposed work on the electrodeposition of PbS and subsequent solution based processing towards the fabrication of lead halide perovskite films provided complex interfacial electrode reactions and nucleation and growth with cuboidal grain morphology with uneven interlayer junctions which delays the charge carrier transportation and increases the recombination rates before hitting the hole transport layer.
CBD has been established as a distinct method of depositing PbS with controllable morphology along with its scalability, simplicity, low-cost and low temperature [39][40][41] . As explained above there is only one report of using CBD for depositing precursor of lead sulphide, which is combined with CVD method for final conversion to lead halide perovskite. MAPbI 3 has been used as the final product in most of reports. As an alternative precursor material, FAPbI 3 is shown a potential candidate with its broader absorption spectrum and narrower bandgap material than MAPbI 3 , however, its morphology and phase structure have significant effects on its performance in perovskite solar cells. In most cases, the PCE of FAPbI 3 based inverted planar devices are lower than those of MAPbI 3 because of the inferior morphology of FAPbI 3 film. It is well known that the performance of devices based on MAPbI 3 would degrade greatly in high temperature or high humidity level. FAPbI 3 is more stable than MAPbI 3 , but it has an undesirable phase transition between δ-phase (yellow phase; more stable) and perovskite α-phase (black phase) at ambient condition 42 .
It has been shown that adding small amount of MA + induces crystallization of black phase of FA-based perovskite and inhibits the formation of its δ phase 43,44 . A more notable structural stability of the α-phase FAPbI 3 was achieved by developing a mixed-cation mixed halide perovskite 45 .
Therefore, in this work, we employed a toxic solvent-free route based on CBD to deposit PbS as the lead metal precursor film is introduced for fabricating perovskite films with the mixed cations (FA, MA) and mixed halides (I, Br, Cl). The process started with the chemical bath deposition of a PbS seeding layer. Then, PbS film was exposed to iodine vapor to be chemically converted to PbI 2 . Depending on the temperature of iodination, the morphology of the PbI 2 layer changed. The effect of different morphologies on final step i.e. chemical conversion to perovskite film as well as photovoltaic performance of achieved layers in the perovskite solar cell were studied. This study represents the versatility of the proposed route to produce a fully covered and high-quality perovskite surface film. After current optimization, the power conversion efficiency (PCE) of 11.35% was achieved under standard conditions (air mass (AM) 1.5, 100 mW cm −2 ) with potential enhancement after several trials in future.

Results and discussion
Chemical bath deposition of PbS films. Figure 1a shows the XRD pattern of chemically deposited PbS layer. As is evident from the XRD pattern, PbS thin film exhibits a face-centered-cubic (fcc) microstructure. SEM micrographs taken from top ( Fig. 1b) of the chemically deposited PbS shows a compact structure which consists of smooth, uniform spherical and densely packed grains. The film covers the substrate completely. The thickness of PbS layers deposited for various dipping times as well as the thickness of corresponding PbI 2 and perovskite films were measured and summarized in Table S2. www.nature.com/scientificreports/ Figure 1c represents the AFM image of the PbS thin film deposited on the mp-TiO 2 /bl-TiO 2 /FTO substrate. It indicates that the PbS thin film is composed of particles of granular nature which are smoothly and uniformly distributed on the surface. The root-mean-square roughness (Rq) of film is about 31.3 nm. Furthermore, the particle size distribution histogram calculated on the SEM image is shown in Fig. 1d. Accordingly, the average granular grain size of PbS is approximately 119 nm. The formation of uniform compact grains in the microstructure of PbS films is achieved which is a consequence of multiple nucleation followed by one-step growth 47 .
Evaluating the morphology and composition of the obtained film showed that the PbS thin film deposited through dipping the substrates inside the chemical bath for 60 min have low surface roughness and full coverage of electron transport layer (ETL) layer. Thus, it is a good candidate for using as a precursor layer for preparing final lead halide perovskite film.
Chemical conversion of PbS to PbI 2 films. The iodination of chemically deposited PbS films were carried out by putting PbS films in iodine vapor environment at two different temperatures of 120 and 155 °C. From XRD patterns (Fig. 2a,b), it is obvious that whole PbS is converted to PbI 2 and instead of initial PbS peaks, some new peaks can be seen which are in accordance with crystallite planes of PbI 2 (00-007-0235). The possible reaction can be 27 : The position of the main Bragg diffraction peaks are the same for two PbI 2 films achieved at the two temperatures. However, there are some differences in the intensity of the peaks related to PbI 2 crystallite planes, depending on the temperature of the reaction. The XRD pattern of 120 °C-PbI 2 represented almost all the characteristic peaks of PbI 2 , including (001), (100), (101), (102), (003), (110), (111), (103) and (201) reflections. Whereas the pattern of 155 °C-PbI 2 revealed four major peaks of (001), (100), (110) and (111). Analysis of the preferred orientation is performed through calculation of texture coefficients (TC). Table 1 shows the TC(hkl)s of the observed planes. For the 120 °C-PbI 2 , (001) plane has the highest TC(hkl) value. While for 155 °C-PbI 2 the (100) plane is dominant. The change in crystallographic direction against the temperature of substrate is in agreement with results observed in vapor-deposited PbI 2 crystals 48 . Provided that general conditions for producing diffraction peaks persist, i.e. Bragg's law and parallel normal vector of planes to the diffraction vector 49 , we assume that in our experiments XRD only have detected diffractions from the crystallographic planes oriented parallel to the surface of the sample 50 . Thus, we expect that this variation of crystallographic texture observed here may be originated from the difference in microstructure features.  www.nature.com/scientificreports/ Figure 3a,b show the SEM micrographs of PbI 2 films prepared at 120 and 155 °C, respectively. PbI 2 layers fabricated by our present method exhibit clear grain boundaries and incompact structure, different from the PbI 2 layers fabricated by conventional spin coating method which have fuzzy domain boundaries and few grain boundaries.
Furthermore, as expected from XRD results, there is an obvious difference amongst the structure of PbI 2 layers obtained at two different temperatures. As can be seen from Fig. 3a, 120 °C-PbI 2 has platelet crystals with hexagonal shape, which are lying parallel to substrate. This structure is consistent with the calculated TC coefficients. According to TC results shown in Table 1, the preferred orientation for 120 °C-PbI 2 is (001) plane that is perpendicular to the c-axis of the hexagonal crystal system. Thus, the PbI 2 is composed of a layered structure with basal planes parallel to the substrate. In contrast, in the case of 155 °C-PbI 2 , the highest TC is for the (100) plane which is parallel to the c-axis of the hexagonal crystal. Accordingly, as can be seen from Fig. 3b, many platelet-like grains are positioned perpendicular to the substrate. This peculiar orientation causes the formation of many intergranular voids which is the cause of the observed porous PbI 2 layer. The AFM images of aforementioned PbI 2 films are also taken and presented at Fig. 3c,d, respectively. The AFM images certifies the same morphology for achieved films.
Production of final mixed-cation mixed halide perovskite films. It has been well demonstrated that changing the morphology of PbI 2 film will provide a chance to modify the morphology and quality of perovskite layer, which may influence the PSC device performance 51 . In this regard, we have investigated the effect of two different PbI 2 films on the solar cell properties.
Perovskite active layers were formed by spin coating of a solution of FAI:MACl:MABr (85:10:10 mg in 1 ml IPA) on top of as-converted PbI 2 layers on mp-TiO 2 /bl-TiO 2 /FTO. As shown in Fig. 4a  www.nature.com/scientificreports/ peaks of δ-phase at 11.6°, indicating that selected composition and method is successful to achieve desirable α-phase perovskite. For perovskite film prepared from 120 °C-PbI 2 an additional peak at 12.8° corresponding to (001) lattice plane of PbI 2 suggests the incomplete conversion to perovskite. In contrast, in the case of perovskite from 155 °C-PbI 2 , the peaks of PbI 2 completely disappear, indicating the complete conversion into photoactive black phase. In addition, for the former perovskite all the peaks are enhanced in intensity than those of the latter perovskite. It means that modifying the precursor PbI 2 morphology may also increase the crystallinity of final perovskite film. The SEM images taken from the top surface of the final perovskite films, made from PbI 2 layer converted at two temperatures of 120 and 155 °C, are shown in Fig. 5a,b, respectively. The perovskite film produced from the 120 °C-PbI 2 layer exhibits the existence of disconnected grains with edges which may be unconverted PbI 2 since heavy-atom regions appear with brighter contrast in the SEM images 53 . EDX analysis of the white regions also confirms the speculated composition, (Fig. S1). In contrast, the perovskite film produced from the 155 °C-PbI 2 layer shows smooth and homogenous perovskite grains with size up to 800 nm which is comparable to that observed for perovskite films prepared by a conventional spin-coating method 54 . Furthermore, some spiral-shape lines can be seen on grains. According to the theory of crystal growth, it can be the result of a spiral growth mechanism, which leads to a step-like morphology 55 .
According to both XRD and SEM results, it can be concluded that the reaction rate of 155 °C-PbI 2 is faster than that of 120 °C-PbI 2 . In the case of 155 °C-PbI 2 , the vertically arranged nanoplatlets create a large degree of   Fig. 6a-c, respectively. Comparing Fig. 6a,b, it can be seen that the thickness value of PbI 2 is almost twice of PbS. According to Fig. 6c the thickness of FAPbI 3 -based perovskite after converting PbI 2 increased by a factor of 2 again. We interpret the thickening of the deposited films by the volume expansion happening after chemical conversion at each step 20,26 . The volume expansion is attributed to variation of lattice parameters of PbS, PbI 2 and perovskite. Figure 7 shows the UV absorption spectra of two samples prepared from two different PbI 2 layers. The higher absorption of perovskite film based on 155 °C-PbI 2 compared to perovskite film based on 120 °C-PbI 2 may be related to the higher amount of perovskite in the former one. Both spectra show onsets of absorption at 810 nm. The optical band gap almost remains the same (1.55 eV) for both films. A range of E g values between 1.51 and 1.55 eV have been reported for (FA/MA)Pb(I/Br) mixed-cation mixed halide perovskites [56][57][58][59][60][61][62] which is consistent with our present result. Therefore, tuning the composition of perovskite by partially substituting cation and halide resulted in an increase in the bandgap compared to pure FAPbI 3 (1.45-1.51 eV in thin films) 54 . Widening of the band gap can be correlated with the perturbing the organic cation size that can cause the contracting of the whole lattice and the change of B-X bond length, which has been shown to influence E g in ABX 3 structure 63 .
UV absorption spectra also confirm the formation of α-phase of FAPbI 3 without the presence of any trace of δ-phase. Because the yellow phase of FAPbI 3 has characteristic absorption peak below 500 nm 64 , whereas both the obtained spectra have shown a remarkable absorption in the visible range. Obtaining pure α-phase of FAPbI 3  www.nature.com/scientificreports/ is of great importance for photovoltaic applications. The yellow δ-FAPbI 3 is a non-photoactive phase and due to having a chain-like structure prevents the transport of electrons and in turn reduces photovoltaic performance 65 .

PSC fabrication and device performance.
In order to investigate the photovoltaic applicability of the synthesized perovskite films, devices with configuration FTO/bl-TiO 2 /mp-TiO 2 /perovskite/spiro-OMeTAD/Au were made by using perovskite films obtained from three-step-method. Figure 8a shows the cross sectional view of the complete perovskite cell and the right part of image was colored and labeled to make a guide to position and thickness of each layer. It can be seen that perovskite film prepared by multi-step method has columnar crystal grains which are elongated from bottom to top. The related current density-voltage (J-V) curves of the best perovskite fabricated from perovskite based on 120 °C-PbI 2 and 155 °C-PbI 2 measured under AM 1.5G solar illumination at 100 mW cm −2 and are plotted in Fig. 8b. Also Table 2 summarizes the photovoltaic parameters obtained from J-V curves, average values (based on six devices) and corresponding standard deviation. Tables S3 and S4 in the "Supplementary document S1" shows photovoltaic parameters and error analysis derived from J-V curves of devices. Furthermore, the corresponding box chart plots of J sc , V oc , FF and PCE for two different processing conditions can be found in Fig. S2a-d, respectively. As can be seen from Table 2, the best-performing cell fabricated via the proposed method, demonstrates acceptable performance with a J sc of 21.43 mA cm −2 , V oc of 0.918 V, FF of 57.67 and PCE of 11.35%. It can be seen that the J sc of PSC based on 155 °C-PbI 2 increased slightly compared to that of 120 °C-PbI 2 based perovskite. It can be related to PbI 2 -free and high crystalline perovskite synthesized from 155 °C-PbI 2 which leads to increment of visible light absorption. Furthermore, in the device based on 120 °C-PbI 2 , the leftover PbI 2 can hinder the electron transfer of perovskite film due to high resistivity of PbI 2 which in turn increases the possibility of charge carrier recombination and reduces V oc and FF.
Totally, the performance of the device prepared by the proposed three-step method is comparable with the performance of other devices made from uncommon metal precursors rather than PbI 2 and is the highest reported efficiency for CBD-based method of synthesizing perovskite. Nevertheless, the method is still less efficient compared to the conventional spin coating method and there are many rooms to modify different parameters of this method in order to enhance the efficiency. One of the main reasons for low efficiency might be the appearance of so-called S-shaped J-V curve, as is evident in Fig. 8b. This phenomenon is one of the many problems occurring when developing new materials and new methods for fabricating perovskite solar cell devices. In most cases, this observation is attributed to the presence of extraction barriers and large interface resistance 66 , which can lead to the accumulation of photo-generated charges and a charge reservoir. Whereas the occurrence of these interface barriers might have different origins, which needs more detailed and systematic study.

Conclusion
Briefly, a scalable processing technique with environmental-friendly approach by neglecting DMF from conventional fabrication processes of halide perovskite solar cells was successfully developed by using chemically deposited PbS from an aqueous solution as the precursor and created perovskite films with good quality and full coverage. The iodination temperature in the second step was changed in a controlled way and the influence on final film quality was evaluated. Finally, grown perovskite films with no unconverted PbI 2 was synthesized by iodination of the PbS layer at 155 °C and exposing to the FAI:MACl:MABr/IPA solution. This was subsequently applied in solar cell devices with the following structure: FTO/bl-TiO 2 /mp-TiO 2 /perovskite/spiro-OMeTAD/Au. The assembled solar cells represented the PCE of 11.35% with corresponding photocurrent of 21.43 mA cm −2 and open circuit voltage of 0.918 V. CBD is a widely used industrial method, therefore, the offered technique opens new avenues for large-scale production of perovskite solar cells. Moreover, avoiding the use of toxic solvent especially DMF facilitates the green fabrication of PSCs. It can be expected that the performance of devices made by this method will be enhanced via optimizing the various parameters of the deposition.

Materials and method
Preparation of electron transport layers.

Synthesizing of the perovskite film.
A three-step deposition method was employed to produce the perovskite film. Initially a PbS layer was coated on substrates via chemical bath deposition method. The deposition bath contained 85 ml of an aqueous solution, consisting of 2.5 ml Pb(CH 3 COO) 2 ·3H 2 O (Sigma-Aldrich) 1 M, 6 ml thiourea SC(NH 2 ) 2 (Sigma-Aldrich) 1 M, 10 ml NaOH (Aldrich) 1 M and 2 ml triethanolamine (Sigma) 1 M. The deposition was performed at room temperature. Afterward, the PbS layers and iodine chips (Sigma-Aldrich) were put inside one petri dish at different temperatures to chemically convert PbS to PbI 2 . The temperatures were chosen according to iodine phase diagram to obtain the iodine vapor. In fact, it is a known that iodine can be in vapor state at the atmospheric pressure in a range of 113.5-184.4 °C. In this range, the iodine liquefies first and then transforms to violet-colored gas. Therefore, two different temperatures in this range were chosen to simultaneously sublimate the iodine and react with the PbS. Next, PbI 2 films were exposed to a FAI:MACl:MABr solution (85:10:10 mg ml −1 ) and then spun at 5000 rpm for 30 s. The selection of solution was based on a comparison between the qualities of perovskite films obtained by different solutions as shown in Fig. S3. After two step annealing at 150 °C and 100 °C for 30 min per each, the conversion of PbI 2 to perovskite film was completed.
Characterization. The thickness of the deposited and chemically converted solid state thin films were measured by a stylus profiler (KLA Tencor). The morphologies of the films were examined using a scanning electron microscopy (SEM, S-4800, Hitach High-Technologies) and atomic force microscopy (AFM, MultiMode V Veeco). Energy dispersive X-ray analysis (EDX) is done in conjunction with SEM. Crystalline structure of the films were determined by using a X-ray diffractometer (XRD, D8 ADVANCE, Bruker AXS) with Cu Kα radiation (λ = 1.5405 Å). In order to estimate the degree of crystalline orientation in different directions, the texture coefficient (TC) was calculated by Eq. (6) 48 : where I(hkl) is the intensity of the measured peak, I 0 (hkl) is the intensity of reference data peak for random oriented powder, and N is the number of measured peaks.
The absorption spectrum of final perovskite layers was extracted using a UV-Vis spectrophotometer (Jasco V-780). In order to record current-voltage (J-V) graphs, A Keithley 2400 source meter was used and Illumination of air mass (AM) 1.5 G and a power of 100 mW cm −2 was applied by a solar simulator (Newport, Oriel Class A, 91195A). For setting the correct irradiance of the solar simulator, a calibrated Si reference cell certified by National Renewable Energy Laboratory (NREL) was used. For each condition, the photovoltaic parameters with corresponding error analysis extracted from J-V curves of all devices.