Highly crystalline CsPbI2Br films for efficient perovskite solar cells via compositional engineering

All-inorganic CsPbI2Br shows high thermal stability for promising application in perovskite solar cells (PSCs). The performance of PSCs is significantly affected by their morphology and crystallinity induced by compositional ratio, solvent/anti-solvent engineering and post thermal annealing. In this study, the compositional ratio effect of two precursors, PbI2 and CsBr, on the power conversion efficiency (PCE) of a device with ITO/SnO2/CsPbI2Br/Spiro-MeOTAD/Au structure was investigated. With the assistance of anti-solvent chlorobenzene, perovskite with a PbI2 : CsBr ratio of 1.05 : 1 showed a high quality thin film with higher crystallinity and larger grain size. In addition, the molar ratio of precursors PbI2 and CsBr improved the PCE of the PSCs, and the PSCs fabricated using the perovskite with an optimal ratio of PbI2 and CsBr exhibited a PCE of 13.34%.


Introduction
Due to its appropriate optical band gap, high optical absorption coefficient, good ion migration ability and long carrier life, organic-inorganic hybrid perovskite is considered as an ideal material for use in the light absorption layer of solar cells. [1][2][3][4][5] In the past decade, new breakthroughs have been made in the research of perovskite solar cells (PSCs). [6][7][8] So far, the highest power conversion efficiency (PCE) of organic-inorganic hybrid perovskite solar cells has reached 25.2%, 9 which is higher than the efficiency of commercial silicon solar cells. However, the poor long-term stability seriously limits the large-scale commercial use of organic-inorganic hybrid perovskite solar cells. [10][11][12] Although some studies have effectively improved the stability of hybrid perovskite solar cells, 13,14 they still cannot meet the requirements for practical applications. But the allinorganic halide perovskite with good thermal stability can be obtained by completely replacing the organic components (MA + , FA + ) of mixed perovskite with Cs + . 15 Typical all-inorganic halide perovskite materials mainly include CsPbI 3 , 16 CsPbI 2 Br, 17 CsPbBr 3 (ref. 18) and CsSnI 3 . 19 Among them, CsPbI 3 and CsPbI 2 Br of cubic phase have appropriate optical band gaps, 1.73 eV and 1.92 eV, respectively. 16,20 Therefore, these two halide perovskites have been extensively studied and the PCE of devices has been rapidly improved. 20 Nevertheless, there is still a big gap between the current maximum efficiency and the theoretical schottky limit, 20 which suggests that the all-inorganic perovskite solar cells still have great room for improvement and research value in photoelectric conversion efficiency. On the other hand, the allinorganic halide perovskite usually suffers a severe phase transition problem. In ambient air environment, CsPbI 3 and CsPbI 2 Br of cubic phase are easily converted into non-perovskite phase, which will seriously affect their photoelectric performance and practical utilization of devices. 21 Studies have shown that, compared with CsPbI 3 , CsPbI 2 Br is easier to maintain the cubic phase structure, which has intrigued great attention. 22 Chen et al. conrmed that the aphase and b-phase CsPbI 2 Br have better thermal stability and phase stability than CsPbI 3 due to the mixing of halide elements through theoretical calculations. 23 Through interface engineering, 24 solvent engineering, 25 precursor engineering 26 and other strategies, the photoelectric performance of CsPbI 2 Br has been effectively improved. For example, Guo et al. 27 successfully synthesized Rb-doped CsPbI 2 Br lm with improved crystallinity and light absorption. Based on this, the prepared all-inorganic perovskite solar cells without hole transport layer achieved stabilized PCE of over 11%. Bai et al. 28 improved the crystallinity of the CsPbBrI 2 lm by preheating the precursor solution at 100 C, and reached a high efficiency of up to 14.81% with the device structure of FTO/TiO 2 /CsPbBrI 2 /CsPbBrI 2 QDs/PTAA/Au. Chen et al. 29 precisely controlled the growth of a-CsPbI 2 Br crystal by the synergistic effect of gradient thermal annealing and anti-solvent method, and obtained a high-quality lm with few defects and good stability, achieving a record PCE of 16.07%. Liu et al. 30 successfully designed a novel structure of solar cell by using ZnO@C 60 bilayer as the electron transport layer, achieving a high PCE of 13.3% with good thermal stability. Dong et al. 31 used EtOH anti-solvent and optimized CsBr deposition cycle to obtain a high-purity phase and highquality CsPbI 2 Br lm, successfully achieving a record PCE of 10.21% for the a-CsPbI 2 Br PSCs without hole transporting material.
Herein, we studied the effect of raw material stoichiometric ratio on the properties of CsPbI 2 Br. Four perovskite solar cells with ITO/SnO 2 /CsPbI 2 Br/Spiro-MeOTAD/Au were fabricated by changing the molar ratios of PbI 2 and CsBr (0.95 : 1, 1.00 : 1, 1.05 : 1, 1.10 : 1). We found that excessive PbI 2 would lead to obvious PbI 2 peak in the XRD pattern, and a small amount of PbI 2 would signicantly affect the morphology and crystallinity of as-prepared CsPbI 2 Br perovskite lms. In addition, different proportions of PbI 2 and CsBr will also lead to changes in optical band gap and light absorption properties of perovskite thin lms, as well as differences in CsPbI 2 Br phase stability. Our results show that solar cell devices have the best power conversion efficiency when the molar ratio of PbI 2 and CsBr is 1.05 : 1. The enhanced PCE originates from the higher crystallinity and defects-free of perovskite thin lm with appropriate grain size characterized by X-ray diffraction (XRD) and can electronic microscopy (SEM).

Precursor preparation
1 M CsBr and X M PbI 2 (X ¼ 0.95, 1.00, 1.05, 1.10) were dissolved in a mixed solvent (DMSO : DMF ¼ 9 : 1) according to certain stoichiometric ratios, and stirred at 70 C for 12 h. The CsPbI 2 Br precursor solution was then obtained by ltration through a 0.22 mm PTFE lter.

Solar cells fabrication
The ITO glasses were successively cleaned with acetone, ethanol and deionized water respectively by ultrasonic cleaning for 15 minutes, respectively. Before the spin coating, each ITO glass was blown dry with nitrogen gun and cleaned with plasma for 60 s. SnO 2 was selected as the electron transport layer. We diluted the SnO 2 colloid precursor to 5% with deionized water, then spin-coated at 3000 rpm for 30 s, and thermal annealed at 150 C for 30 min to obtain an electron transport layer. Then the 1 M CsPbI 2 Br precursor solution was deposited on the SnO 2 layer at the speed of 1000 rpm for 10 s and 4500 rpm for 35 s in a glove box, annealing at 260 C for 10 minutes. For deposition of hole-transporting material, Spiro-MeOTAD was dissolved in chlorobenzene with a concentration of 80 mg mL À1 , then 35 mL of lithium bis (triuoromethanesulfonyl)imide in acetonitrile (260 mg mL À1 ) and 30 mL of 4-tert-butylpyridine was added into the Spiro-MeOTAD solution. The mixture was coated onto the perovskite lm at 3500 RPM for 30 s to form a Spiro-MeOTAD hole-transporting layer. Finally, the counter electrode was deposited by thermal evaporation of 80 nm-thick gold under a pressure of 2 Â 10 À6 mbar. The active area was measured to be 0.045 cm 2 .

Characterization of perovskite thin lms
UV-vis absorption and photoluminescence (PL) spectra of thin lms were recorded on a HP 8453 spectrophotometer and FLS920 spectrouorimeter (Edinburgh Instruments), respectively. A 150 W, ozone-free xenon arc lamp was used in PL measurements. Scanning electron microscope (SEM) images were obtained by using a eld emission scanning electron microscope (JEOL-7401). Thicknesses of thin lms were measured by Dektak 150 surface prolometer. X-Ray Diffraction (XRD) patterns were measured by an X-ray diffractometer Bruker D8 Advance using Cu Ka radiation source with a scan rate of 10 min À1 .

Characterization of perovskite solar cells
The J-V characteristics of PSCs were measured by a Keithley model 2400 source measure unit (Newport, Oriel AM 1.5 G, 100 mW cm À2 ). The light intensity of 100 mW cm À2 was calibrated by a silicon reference cell. The external quantum efficiency (EQE) spectra of PSCs were performed on a DSR100UV-B spectrometer with a bromine tungsten light source, a SR830 lock-in amplier and a calibrated Si detector. Fig. 1 The XRD patterns of perovskite thin films with different molar ratio of PbI 2 : CsBr.

Results and discussion
To investigate the inuence of compositional ratio of PbI 2 -: CsBr on the structural evolution of all-inorganic perovskite, Xray diffraction (XRD) of these thin lms are conducted in Fig. 1   formula, indicating the small adjusted component ratio of the PbI 2 added into the precursor solution doesn't change crystal size signicantly which is benecial to improve the charge transport in the solar cells. The new peak located at 12.5 detected from both perovskite with PbI 2 : CsBr (1.05 : 1 and 1.10 : 1) is likely to due to the peak of excess unreacted PbI 2 . It was reported that moderate excess PbI 2 can functioned as role of passivating the surface defects for reducing the charge carrier recombination. Fig. 2a showed the absorption spectrum of perovskite thin lms, all of which exhibited the typical absorption prole for cubic phase CsPbI 2 Br thin lm with the typical absorption peak  This journal is © The Royal Society of Chemistry 2019 of 625 nm, which is consistent with other report. 31,33 However, the perovskite lm deposited with PbI 2 : CsBr ratio of 1.10 : 1 has the absorption peak at 635 nm. The slightly red shied absorption peak indicates the excess PbI 2 actually inuences the CsPbI 2 Br perovskite thin lms. In addition, the peak is observed at 405 nm for the d-perovskite with PbI 2 : CsBr of 1.10 : 1, implying that the black a-CsPbI 2 Br phase could have been partially transitioned into the yellow d-CsPbI 2 Br phase. Fig. 2b depicted the steady-state photoluminescence (PL) spectra of these four perovskite thin lms. The emission peak of the CsPbI 2 Br thin lm is at 653 nm is well consistent with absorption edges for corresponding perovskite thin lms (Fig. 2a). It is also observed that a red shied PL peak for perovskites with the molar ratio of 1.10 : 1.
The surface morphology and crystalline size of CsPbI 2 Br thin lms deposited with these four different molar ratios of PbI 2 -: CsBr is further analyzed by SEM images, as shown in Fig. 3. It is obviously that all samples display densely stacked grains on their surfaces which is consistently with the calculated average crystal size in nanoscale range. However, brighter dots on the surface of the perovskite grains are clearly observed when the molar ratio of PbI 2 : CsBr is 1.10 : 1 (Fig. 3d), proong the existence of the unreacted PbI 2 particles could affect the thin lm quality.
To investigate the effect of composition ratios of PbI 2 : CsBr on the photovoltaic performance of PSCs, devices with a planar heterojunction structure of ITO/SnO 2 /CsPbI 2 Br/Spiro-MeOTAD/ Au (Fig. 4a) with the CsPbI 2 Br shown in Fig. 4b were fabricated where ITO acts as the cathode electrode, solution-processed SnO 2 thin lm is used as the electron extraction layer, Spiro-MeOTAD acts as the hole extraction layer and Au acts as the anode electrode, respectively (Fig. 4c). The current densities versus voltages (J-V) characteristics of PSCs under one-sun illumination with the light intensity of 100 mW cm À2 , at the scan rate of 0.20 V s À1 and the reverse scan direction, are shown in Fig. 5a. The device performance parameters are summarized in Table 1. The PSCs based on the perovskite with PbI 2 : CsBr of 0.95 : 1 and 1.00 : 1 show PCE of 11.49% (V OC of 1.09 V, J SC of 14.8 mA cm À2 and FF of 71.2%) and 12.52% (V OC of 1.12 V, J SC of 15.01 mA cm À2 and FF of 74.5%). The PSCs based on the perovskite with PbI 2 : CsBr of 1.05 : 1 shows an enhanced performance upon slightly increasing with PbI 2 . Specically, the PCE is increased to 13.34% with a V OC , J SC , and FF of 1.12 V, 15.78 mA cm À2 , and 75.5%, respectively. It is worth noted that the simultaneously enhanced V OC , J SC and FF is achieved. These enhancements are attributed to the denser, higher crystallinity CsPbI 2 Br grains which could effectively suppress charge recombination (Fig. 3c). However, further addition of much more PbI 2 causes a negative performance change. For example, when the molar ratio of PbI 2 : CsBr is increased to 1.10 : 1, the PCE is decreased to 10.91% (V OC of 1.10 V, J SC of 14.01 mA cm À2 , FF of 70.8%). The excess PbI 2 has reduced the sunlight absorption efficiency (Fig. 2a) and could act as recombination or trap centers with unnecessary spots on the active CsPbI 2 Br thin lm (Fig. 3d), severally reducing the photocurrent and FF ( Table 1). The strong correlation between J SC and series resistance (R s ) could be validated that with more and more PbI 2 added, the measured R s has consistently reduced when the molar ratio changes from 0.95 to 1.05 (Table 1), indicating the improved crystallinity does increase the charge separation and carrier transport efficiency. Thus the molar ratio of PbI 2 : CsBr in the precursor solution is a crucial parameter for determining device performance.
EQE of all the perovskite solar cells is shown in Fig. 5b. The perovskite solar cells exhibit a photoresponse in the wavelength ranging from 300 to 650 nm, suggesting absorbed photos being converted into charge carriers, and further generates into photocurrent in the solar cells. In addition, the champion device shows a maximum value of about 93.4% at l ¼ 480 nm, the integrated J SC values from the EQE spectra are 14.17 mA cm À2 , 15.16 mA cm À2 , 15.51 and 13.22 mA cm À2 for the PSCs fabricated by perovskite with PbI 2 : CsBr of 0.95 : 1, 1.00 : 1, 1.05 : 1 and 1.10 : 1, respectively, which is about 5% error between the J SC from EQE spectra and J-V curve, implying the accuracy of device performance. Furthermore, as the stability is the priority concern for the PSCs, the shelf stability of the PSCs is investigated. The devices are stored in N 2 atmosphere without any encapsulation. As shown in Fig. 6, the PSCs fabricated by perovskite with PbI 2 -: CsBr of 1.05 : 1 showed a better shelf stability than that of PSCs fabricated by perovskite with PbI 2 : CsBr of 1.00 : 1. Aer being stored for 240 h, PSCs fabricated by perovskite with PbI 2 : CsBr of 1.05 : 1 retained 84% of the initial PCE, while the PCE of PSCs fabricated by perovskite with 1 : 1 ratio of the two precursors dropped to 75% of the initial value in the same storage condition and time. The enhanced shelf stability which we attribute to the surface passivation of unreacted PbI 2 . 34-36

Conclusions
In summary, we investigated the effect of compositional engineering on the lm properties and device performance of inorganic perovskite CsPbI 2 Br by changing the stoichiometric ratios of PbI 2 and CsBr (0.95 : 1, 1.00 : 1, 1.05 : 1 and 1.10 : 1).
The results show that the molar ratio of the raw materials is 1.05 : 1, the CsPbI 2 Br lm shows the best crystallinity and the corresponding solar cell (ITO/SnO 2 /CsPbI 2 Br/Spiro-MeOTAD/ Au) achieves the highest efficiency of 13.34%. XRD and SEM results show that once the molar ratio of raw materials is increased to 1.10 : 1, the resultant lm contains more PbI 2 , which may provide more recombination centers for free electrons and holes, thus signicantly reducing the efficiency of the device. Our study highlights the importance of compositional engineering and offers a meaningful reference for the preparation of stable and efficient solar cell devices.

Conflicts of interest
There are no conicts of interest to declare.