Organic-inorganic halide perovskite solar cells have attracted great attention in recent years. The general formula for perovskite is ABX3 (A cation, B cation, X anion). In 2012, the first all-solid-state solar cell [1] was introduced with a power conversion efficiency (PCE) of 9% [1] which is now increasing up to 22% [2]. These perovskite solar cells are mainly based on methylammonium lead iodide (MAPbI3) [3,4,5,6,7,8] and formamidinium lead iodide (FAPbI3) [9, 10]. Different halogens are used as anions (I, Br, Cl) [11] and inorganic cesium (Cs) is also used as cation with methylammonium(MA) and formamidinium(FA) in perovskite solar cells (PVCs) [12]. All of these materials have toxic lead, which is harmful to human health. This limits the commercial use of perovskite solar cells. Scientists have been looking for a non-toxic elements to replace lead in perovskites [16,17,18,19,20,21,22,23,24,25,26]. Some tried to mix the bivalent cations Sn2+ and Pb2+ as CH3NH3SnxPb(1−x)I3 [13, 14] and others mixed the monovalent A cations along with mixed bivalent cations, i.e. FA0.8MA0.2SnxPb1−xI3 [15], but these perovskites are still toxic. In 2014 [16] Snaith and co-workers first developed complete lead-free perovskite solar cells based on methylammonium tin triiodide (CH3NH3SnI3) and achieved a PCE around 6%. In the same year, Kanatzidis and co-workers [17] investigated CH3NH3SnI3-xBrx and received almost the same PCE. MASnI3 is very unstable in air according to previous reports [16]. Later, researchers attempted to use formamidinium tin triiodide (FASnI3) with the additive of SnF2 to delay the oxidation of Sn2+ to Sn4+ [18, 19]. They found that FASnI3 was more stable than MASnI3. Recently, Seok and co-workers [20] obtained a smooth and dense FASnI3 perovskite layer using SnF2-pyrazine complex as an additive. With SnF2 additives and diethyl ether dripping in a solvent engineering process to synthesize FASnI3 perovskite thin films [21], Dewei Zhao and co-workers [21] achieved an improved PCE up to 6.22% for lead-free Sn-based perovskite solar cells .

Similar as FASnI3, CsSnI3 also shows perovskite phase at room temperature. CsSnI3 has four phases at different temperatures [22], but only the black orthorhombic phase B-γ-CsSnI3 is the perovskite phase. Kumar et al. [23] fabricated solar cells using CsSnI3 as the perovskite layer between a TiO2 electron transfer layer and a Spiro-OMeTAD hole transfer layer and achieved a PCE of 2%. Zhou et al. [24] modulated B-γ-CsSnI3 grain sizes by employing different annealing temperatures and chose the optimal architecture for perovskite solar cells. They achieved a PCE of 3.31%. Marshall et al. [25] proved that the existence of SnCl2 resulted in higher film stability and they achieved a PCE of 3.56% from PSCs without any hole-selective interfacial layer. CsSnI3 can be used as the perovskite absorption layer as well as the active region in lead-free perovskite infrared LEDs [26]. Generally, Spiro-OMeTAD is used as a hole transport material (HTM), which typically contains acetonitrile and lithium (Li) and/or cobalt (Co) salts that may change the morphology of Sn-based perovskite films and form the undesirable Cs2SnI6 polymorph [21, 22].

While SnF2-pyrazine complex [20] and SnF2 [21] have been used as additives into FASnI3 solution and resulted in good performance on stability and efficiency, SnCl2 also can be used as an alternative additive. The mechanism is similar to other tin halides (SnF2, SnCl2, SnBr2, SnI2) additives which are chosen in CsSnI3-based perovskite photovoltaics [25]. In this report, we chose SnCl2 as an additive of FASnI3 solution and SnF2 as an additive of CsSnI3 solution to investigate the stability of the perovskite films, respectively. The measurement of the evolution of absorption spectra at different time courses and other experimental results (SEM, photos etc.) showed that the stability of the films was improved by both additives. With different anti-solvent dripping during spin-coating, we obtained some new findings about the surface morphology and obtained complete coverage of perovskite films.


The synthesis method of FASnI3 follows reference [21]: 372 mg of SnI2 (Sigma-Aldrich) and 172 mg of formamidinium iodide (FAI) were dissolved in 800 μl anhydrous dimethylformamide (DMF, Sigma-Aldrich) and 200 μl anhydrous dimethyl sulfoxide (DMSO, Sigma-Aldrich). For this precursor solution, 10 mol% SnCl2 was added and then stirred. Via spray pyrolysis at 500 °C, a compact layer of TiO2 substrate was deposited on FTO glass. The films were annealed at 500 °C for 15 min and then cooled down to room temperature. The mesoporous TiO2 scaffold was spin-coated at 4500 rpm for 20 s and then heated at 500 °C for 1 h. FASnI3 films were synthesized by spin-coating the precursor solution with SnCl2 additives at 4000 rpm for 60 s in a glove box. During the process of spin-coating, the anti-solvent (diethyl ether, toluene, chlorobenzene) was dripped and then the perovskite films were annealed at 70 °C for 20 min.

The synthesis method of CsSnI3 is described in a previous paper [23]: 0.6 M of CsSnI3 that contained equimolar quantities of CsI and SnI2 without or with 10 mol% SnF2 additives, respectively, was added to DMSO and stirred overnight at 70 °C. Sixty microliters of the precursor solution was spin coated onto the TiO2 substrate at 4000 rpm. The substrates were then annealed at 70 °C for 10 min, and mirror-like black perovskite films were formed.

Results and Discussion

Under the effect of different anti-solvent dripping, the FASnI3 films with 10 mol% SnCl2 additives exhibit different film morphologies. Figure 1 shows the SEM images of FASnI3 perovskite films on TiO2 with different anti-solvents dripped. Figure 1a shows discontinuous nucleation, partial coverage and the presence of pinholes on the surface of the FASnI3 film without (w/o) any anti-solvent dripped. Dripping with diethyl ether (Fig. 1b), the formed FASnI3 film likes a net which has a lot of holes on it and spreads on the TiO2 substrate. Dripping with toluene or chlorobenzene (Fig. 1c and d, respectively), the surface morphology of the FASnI3 film has been further improved, and the film is highly uniform and dense with full coverage on the substrate. Using toluene as the anti-solvent, the average size of crystal particles is bigger than that of films fabricated by using chlorobenzene as the anti-solvent. These results are consistent with those reported in other articles [20, 21]. Without anti-solvent dripped, the film does not change color during spin-coating, and after continuous annealing at 70 °C, the film turns to black immediately and leads to the formation of a rough surface. When the film is dripped with diethyl ether, toluene or chlorobenzene, it changes into a reddish color immediately. After thermal annealing at 70 °C for 20 min, the film turns whitish with diethyl ether dripping and black (the left insert in Fig. 1d) with toluene or chlorobenzene dripping. No matter what kind of anti-solvent is dripped, films from the back view of the FTO glasses are brownish red (the right insert in Fig. 1d).

Fig. 1
figure 1

SEM images of perovskite films prepared (a) without anti-solvent dripping, (b) with diethyl ether dripping, (c) with toluene dripping, and (d) with chlorobenzene dripping

To investigate whether the anti-solvents dripping would lead to any crystal phase transition or not, we measured the XRD patterns. As shown in Fig. 2, all of the FASnI3 films that were formed on TiO2 crystallize in the orthorhombic structure and random orientation, which is consistent with other reports [20, 21].

Fig. 2
figure 2

XRD patterns of FASnI3 perovskite films under different anti-solvents dripped

Figure 3a shows the scanning electron microscopy (SEM) images of TiO2 substrates. The cross-sectional SEM image (Fig. 3b) of a structure of FTO/compact TiO2/mesoporous TiO2/FASnI3 clearly displays the stacked layers. From the figure, the thickness of FASnI3 is about 250 nm.

Fig. 3
figure 3

a SEM top view of TiO2 surface. b The cross-sectional SEM image of a structure of FTO/compact TiO2/mesoporous TiO2/FASnI3

The optical absorption spectra of the FASnI3 perovskite thin film with 10 mol% SnCl2 additives under the effect of different anti-solvent dripping are shown in Fig. 4a. The absorption onset occurs at 900 nm, and this result is consist with that reported by other groups [21]. As shown in Fig. 4, there are various absorption peaks of the films prepared by dripping with different anti-solvents. The absorption strength can indirectly reflect the quality of perovskite films. It is known that FASnI3 can automatically degrade to FA2SnI6 in air, [18, 21] and the absorption coefficient of the latter in the visible spectrum is smaller than that of the former. Degradation of the film in air with respect to time can be measured from absorption spectra. As shown in Fig. 4b, UV-vis absorption as a function of time was measured. The changed intensity of absorption reflected the process of degradation. Note that, relative humidity of the environment was about 46% and the room temperature was 15 °C. From optical pictures, we can see that FASnI3 degrades quickly during the first few hours. After 17 h, the absorption peak corresponding to FA2SnI6 becomes obvious. This result confirms that FASnI3 can degrade to FA2SnI6 in air.

Fig. 4
figure 4

Absorption spectra of FASnI3 + 10% SnCl2 films (a) with different anti-solvents dripping on TiO2 and (b) with different time courses in air

Figure 5a shows a cross-sectional SEM image of the FASnI3 PSCs with a structure of FTO/cp-TiO2/mp-TiO2/FASnI3/Spiro-OMeTAD/Au. Figure 5b shows the J-V curves measured with different anti-solvent effects. Although the PCEs of these PSCs are quite low, some characteristics still offer new insights of the fabrication of lead-free perovskites. From Fig. 1a, we know that the nucleation of the film without anti-solvent dripping is discontinuous, which leads to nano-radiative recombination of electrons and holes and causes a large leakage current between TiO2 and FASnI3. The result of the films used diethyl ether as the anti-solvent was approximately the same as the untreated sample. The leakage current is large, but the coverage of the film is improved. With toluene and chlorobenzene as anti-solvents, the coverage of the film has been further improved, and larger crystal particles can create less grain boundaries to enhance the charge separation and collection of electrons and holes, thus leading to the highest PCE.

Fig. 5
figure 5

a Cross-sectional SEM image of a completed device. b J-V curves of the FASnI3 perovskite solar cells using different anti-solvents

There are several challenges that hinder performance improvement of CsSnI3 perovskite solar cells: (1) Sn2+ oxidizes to Sn4+ easily, which seriously affects the photoelectric properties of CsSnI3 perovskite solar cells. (2) It is difficult to synthesize uniform and fully covered lead-free Sn-based thin films. Even with different additives, there are many pinholes existing on the crystallite surface which may short electron transfer layer and hole transport layer, leading to an enormous leakage current. (3) Lead-free Sn-based PSCs are often prepared in regular cell structures [18, 23]. In the following content, we have made some preliminary studies on the CsSnI3 films.

Without any additives, the color of CsSnI3 precursor solution was more yellowish than the solution with 10 mol% SnF2 additive, as presented in Fig. 6. This indicates that oxidation occurred more easily for pure CsSnI3.

Fig. 6
figure 6

a Pure CsSnI3 without any additives. b 10 mol% SnF2 additives in CsSnI3 precursor solution

We also plotted the evolution of absorption spectra at different time courses to investigate the degradation of CsSnI3 thin films with and without additives in air. As shown in Fig. 7, the black vertical line indicates the direction of change with increasing time in ambient air. Without additives, the CsSnI3 thin films degraded quickly when it was exposed to air at relative humidity of 57% and temperature of 13 °C. The degradation rate of the film was very fast in the beginning, but it slowed down a lot after 1 h. The degeneration process of CsSnI3 with 10 mol% SnF2 additive showed some difference. During the first few minutes, the film was quite stable and no oxidation occurred. Meanwhile the absorption peaks were at the same position. Few minutes later the oxidation rate accelerated and slowed down after an hour. Therefore, it can be deduced that the stability of the film is improved by the addition of SnF2.

Fig. 7
figure 7

Absorption spectra of (a) pure CsSnI3 films and (b) CsSnI3 + 10% SnF2 films at different times in ambient air


In summary, we have studied the different morphology characteristics of FASnI3 films prepared by using different anti-solvents and 10 mol% SnCl2. The aforementioned experimental results show that toluene and chlorobenzene are the best anti-solvents for improving the quality of the films and allow the film to completely cover the substrate. Using toluene as anti-solvent, we can gain the highest PCE of PSCs. The stability of FASnI3 can be kept for several hours, while CsSnI3 can only be stable for few minutes. So if we want to develop alternative Pb-free Sn-based perovskite films, the most critical issue is to stabilize the material, namely, suppressing the oxidation of Sn2+ within the crystal. This will enhance the long-term stable operation of perovskite films. This technique may offer a promising approach to fabricate the high efficiency of Pb-free Sn-based perovskite solar cell over the next few years.