Decoupling engineering of formamidinium–cesium perovskites for efficient photovoltaics

Abstract Although pure formamidinium iodide perovskite (FAPbI3) possesses an optimal gap for photovoltaics, their poor phase stability limits the long-term operational stability of the devices. A promising approach to enhance their phase stability is to incorporate cesium into FAPbI3. However, state-of-the-art formamidinium–cesium (FA–Cs) iodide perovskites demonstrate much worse efficiency compared with FAPbI3, limited by the different crystallization dynamics of formamidinium and cesium, which result in poor composition homogeneity and high trap densities. We develop a novel strategy of crystallization decoupling processes of formamidinium and cesium via a sequential cesium incorporation approach. As such, we obtain highly reproducible, highly efficient and stable solar cells based on FA1–xCsxPbI3 (x = 0.05–0.16) films with uniform composition distribution in the nanoscale and low defect densities. We also revealed a new stabilization mechanism for Cs doping to stabilize FAPbI3, i.e. the incorporation of Cs into FAPbI3 significantly reduces the electron–phonon coupling strength to suppress ionic migration, thereby improving the stability of FA–Cs-based devices.


INTRODUCTION
Metal-halide perovskites with superior photophysical properties and low-cost solution technology have emerged as promising candidates for different optoelectronic devices, including solar cells, lightemitting diodes, etc. [1][2][3][4][5][6]. For perovskite solar cells (PSCs), the certified power-conversion efficiency (PCE) has reached 25.7%, which is comparable to the current commercial crystalline silicon solar cells. ABX 3 perovskites with tailoring compositions, where A is an organic or inorganic cation, B is a metal cation and X is a halide anion, have been attempted for high efficiency and stable photovoltaic devices. Among these, formamidinium lead iodide (FAPbI 3 ) has exhibited great potential as the absorber layer, due to its optimal band gap of ∼1.5 eV and excellent thermal stability [7][8][9].
However, the photoactive FAPbI 3 black phase would easily transform into a non-photoactive yel-low δ-FAPbI 3 phase at room temperature, especially under humid conditions [10,11]. The poor phase stability challenges both the efficiency and long-term stability of the PSCs based on FAPbI 3 [12,13]. It is generally believed that the phase instability of FAPbI 3 perovskites originates from its unsuitable tolerant factor. To address this problem, alloying FA + with MA + /Cs + cations or partially substituting I − with Br − ions has been employed to tune the tolerant factor [14,15]. The resulting mixed-ion FA-based perovskites exhibit improved resistance to phase transition.
Among these different alloying approaches, formamidinium-cesium mixed-cation pure iodide (FA 1-x Cs x PbI 3 ) perovskites are particularly promising, because they avoid the concerns about volatile MA cations and phase segregation induced by mixed halide ions (Br-I) [16][17][18][19][20]. However, because of the complex crystallization kinetics of formamidinium and cesium, these pure iodide FA-Cs perovskites fabricated by one-step crystallization suffer from poor composition homogeneity and high defects/traps densities [21,22]. The PSCs based on these films are therefore facing relatively low efficiencies. Especially, strong non-radiative recombination in all reported FA-Cs-based PSCs limited the open-circuit voltage (V oc ) of the resulting devices [23][24][25].
Herein, we develop a novel sequential Cs incorporation (SCI) strategy to decouple the crystallization processes of formamidinium and cesium, and achieve highly efficient pure iodide FA 1-x Cs x PbI 3 (x = 0.05-0.16) perovskites (denoted as SCI-FA 1-x Cs x PbI 3 ). The ratio of FA and Cs in FA 1-x Cs x PbI 3 can be straightforwardly tuned by introducing different concentrations of cesium formate (HCOOCs) solution on the FA-based perovskite precursor film during the SCI process. A unique feature of our SCI-FA 1-x Cs x PbI 3 perovskites is the uniform distributions of Cs cations, in contrast to their poor uniformity in typical one-step (1S) crystallized FA-Cs perovskite films (denoted as 1S-FA 1-x Cs x PbI 3 ). As a result, the champion SCI-FA 0.91 Cs 0.09 PbI 3 PSCs yield a record PCE of 24.7% (certified 23.8%) with improved V oc and fill factor, which is the highest value for the pure iodide FA 1-x Cs x PbI 3 perovskites. Compared with FAPbI 3 , the SCI-FA 0.91 Cs 0.09 PbI 3 perovskite shows reduced electron-phonon coupling and lattice fluctuations, which suppress the formation of iodide-rich clusters and finally contribute to the excellent operational stability of the FA 0.91 Cs 0.09 PbI 3 -based PSCs. Figure 1a shows the schematic diagram of SCI-FA 1-x Cs x PbI 3 perovskite films prepared by decoupling the crystallization processes of formamidinium and cesium. A FAPbI 3 precursor film was first deposited by a typical anti-solvent method followed by annealing for 1 min. The Cs cation is sequentially introduced onto the FA perovskite film by spin-coating HCOOCs isopropanol (IPA) solution, followed by further annealing. For comparison, we employed different concentrations of HCOOCs solution (2.5, 5 and 10 mg mL -1 ) to fabricate SCI-FA 1-x Cs x PbI 3 perovskites. The final ratios of incorporated Cs in the above SCI-FA 1-x Cs x PbI 3 perovskite films, i.e. the value of x, are 0.05, 0.09 and 0.16, as confirmed by inductively coupled plasma-mass spectrometry (ICP-MS) analysis (Supplementary Table 1). The corresponding SCI-FA 1-x Cs x PbI 3 perovskite films are noted as x = 0.05, x = 0.09 and x = 0.16 in Fig. 1.

RESULT AND DISCUSSION
Optical and structural measurements of perovskite films indicate that Cs + from HCOOCs has successfully been incorporated into the lattice of FAPbI 3 perovskites. Figure 1b shows the ultraviolet-visible (UV-vis) spectra of SCI-FA 1-x Cs x PbI 3 perovskite films, in which the absorption edges of SCI-FA 1-x Cs x PbI 3 perovskites strongly depend on the amount of Cs + incorporation. When x increases from 0 to 0.16, the absorption edges of SCI-FA 1-x Cs x PbI 3 perovskites gradually blue-shift from 816 to 802 nm, and the corresponding photoluminescence (PL) peaks shift from 809 to 797 nm. The X-ray diffraction (XRD) measurements are carried out to investigate the crystal structure evolution of SCI-FA 1-x Cs x PbI 3 perovskites (Fig. 1c). All the SCI-FA 1-x Cs x PbI 3 perovskites exhibit stronger peak intensity than the pure FAPbI 3 at around both 14 • and 28 • , corresponding to (001) and (002) perovskite crystal planes. The inset image of Fig. 1c shows that the peak between 13.8 • and 14.1 • shifts to a higher degree, indicating that Cs ions are incorporated into the perovskite lattice. The lattice parameter decreases with increasing the amount of Cs ( Supplementary Fig. 1), further confirming the successful mixing of Cs + in the perovskite lattice. The tolerance factor of SCI-FA 1-x Cs x PbI 3 perovskites is also reduced compared with pure FA perovskite, potentially contributing to an stable perovskite structure ( Supplementary Fig. 2).
The Cs incorporation also significantly improves the film morphologies (Fig. 1d). All SCI-FA 1-x Cs x PbI 3 films show enlarged and pinhole-free grains compared with the FAPbI 3 film, which shows coarse grains and pinholes. As shown in the cross-sectional scanning electron microscopy (SEM) images, SCI-FA 1-x Cs x PbI 3 perovskite films (550-600 nm) with vertical growth of grains benefit efficient charge extraction.
Considering that x = 0.09 provides the optimal photovoltaic (PV) performance, we then chose this composition (denoted as SCI-FA 0.91 Cs 0.09 PbI 3 ) for detailed investigations on Cs incorporation and its role on film and device properties.
X-ray photoelectron spectroscopy spectra are conducted to explore the effect of SCI on the elements and their chemical states in perovskite films. All core-level peaks are assigned to Cs, Pb,    Fig. 5), the optimal efficiency is also obtained from the x = 0.09 sample, which hence will be used as the control sample for comparison with SCI-FA 0.91 Cs 0.09 PbI 3 . The absorption edge, characteristic XRD peaks and surface morphologies of the 1S-FA 0.91 Cs 0.09 PbI 3 perovskite films are consistent with SCI-FA 0.91 Cs 0.09 PbI 3 perovskite ( Supplementary Fig. 6).
The synchrotron-based grazing-incidence wideangle X-ray scattering (GIWAXS) and time-of-flight secondary ion mass spectrometry (ToF-SIMS) measurements are further employed to explore the crystal structure and internal composition in perovskite films. As shown in Fig. 2a Fig. 9), which can further passivate the defects on the bottom interface and improve the charge carriers' dynamics in the devices [27,28]. Uniform incorporation of Cs ions into FAPbI 3 has two positive effects: enhanced phase stability and decreased trap/defect densities. The enhanced phase stability is evidenced from the absence of color/structural changes under humid conditions. Under 60% relative humidity, the SCI-FA 0.91 Cs 0.09 PbI 3 perovskite maintains the black phase for 7 days without any changes (Supplementary Fig. 10), showing significant improvement compared with FAPbI 3 . The decreased trap/defect densities are demonstrated from photophysical measurements. The photoluminescence (PL) intensities of SCI-FA 0.91 Cs 0.09 PbI 3 perovskites are much stronger than those of 1S-FA 0.91 Cs 0.09 PbI 3 and pure FA perovskite films ( Supplementary  Fig. 11). In addition, the time-resolved PL (TRPL) spectra in Fig. 2d Fig. 13). The most striking difference is the V oc , which increases from 1.09 V in FAPbI 3 to 1. 18   lower efficiency of 23.1%. This comparison indicates that our crystallization decoupling engineering is beneficial for enhancing SCI-FA 1-x Cs x PbI 3 -based PSCs performance. The incident photon to electron conversion efficiency (IPCE) (Fig. 3b) is similar for both devices, with a high value of >90% in the wavelength range of 450-650 nm. The short-circuit current density (J sc ) of the SCI-FA 0.91 Cs 0.09 PbI 3 device is slightly decreased compared with the FAPbI 3 device, mainly due to slight increase in the band gap upon Cs incorporation. Figure 3c compares the PV parameters of FAPbI 3 -, SCI-FA 0.91 Cs 0.09 PbI 3 -and 1S-FA 0.91 Cs 0.09 PbI 3 -based PSCs for 18 devices, respectively, indicating that sequential Cs incorporation also improves the device reproducibility. In addition, the SCI-FA 0.91 Cs 0.09 PbI 3 -based PSCs exhibit a smaller hysteresis ( Supplementary Figs 13 and  14), resulting in a stabilized output power of 24.4% (Fig. 3d).
The significantly enhanced V oc of the SCI-FA 0.91 Cs 0.09 PbI 3 device is mainly due to suppressed non-radiative recombination, which can be quantified by measuring the external quantum efficiency of electroluminescence (EQE EL ) values [29]. As shown in Fig. 3f, at the injection current densities corresponding to J sc , the EQE EL value of the SCI-FA 0.91 Cs 0.09 PbI 3 device is 6.38%, while that of the FAPbI 3 device is 0.16%. We calculate the voltage losses due to non-radiative recombination ( V oc,non-rad ) based on the formula [30]: where k, T and q represent the Boltzmann constant, temperature and elementary electric charge, respectively. The difference in V oc,non-rad (0.09 V) matches well with the difference of device V oc (0.09 V). Suppressed non-radiative recombination in the SCI-FA 0.91 Cs 0.09 PbI 3 device is consistent with previous photophysical measurements on the films, which indicate that the sequential Cs incorporation can reduce the defects/traps. Further measurements on the devices also reach similar conclusions. The trap-filled limiting voltage in the space-charge limited current measurements decreases from 0.13 V in the FAPbI 3 device to 0.09 V in SCI-FA 0.91 Cs 0.09 PbI 3 device ( Supplementary Fig. 15), indicating suppressed traps/defects upon Cs sequential incorporation [31,32] Fig. 16), which show slower TPV decay (indicating longer recombination lifetime) and quicker TPC decay (indicating fewer trapping effects) in the SCI-FA 0.91 Cs 0.09 PbI 3 device [33,34]. In addition to improved PV performance, the SCI-FA 0.91 Cs 0.09 PbI 3 device also shows significantly enhanced stability. We first measure the shelf life by storing the unencapsulated devices in dark at 25 • C and 20% relative humidity. Figure 4a shows that the PCE of the FAPbI 3 device decreases by ∼30% after 3000 h of aging, whereas the SCI-FA 0.91 Cs 0.09 PbI 3 device shows a degradation of only 10% over 4500 h of aging. We then investigate the long-term operational stability of the PSCs by aging the unencapsulated devices under a nitrogen atmosphere, using maximum power point (MPP) tracking under simulated 1-sun conditions. As shown in Fig. 4b, the SCI-FA 0.91 Cs 0.09 PbI 3 based PSCs retains >90% of the initial PCE while the FAPbI 3 device maintains only 60% PCE after 1000 h of continuous illumination ( Supplementary Fig. 17). Especially, the sharp decline of the efficiency in FAPbI 3 PSCs in the initial stage should be attributed to the intrinsic instability of the FAPbI 3 perovskite layer and potential severe ion migration in the FAPbI 3 device.
A main reason for enhanced-stability PSCs is attributed to suppressed ionic migration. In Fig. 4c and d, we compare the I − ions distribution at the ∼300-nm depth of the perovskite layer for PSCs based on FAPbI 3 and SCI-FA 0.91 Cs 0.09 PbI 3 after 240-h MPP test. In the FAPbI 3 -based device, strong aggregation of I − clusters is observed in the perovskite absorber layers; in contrast, I − ions distribute uniformly in the SCI-FA 0.91 Cs 0.09 PbI 3based devices. This sharp contract indicates that the ionic migration in the SCI-FA 0.91 Cs 0.09 PbI 3 is much suppressed upon Cs sequential incorporation.
Suppressed ionic migration in SCI-FA 0.91 Cs 0.09 PbI 3 is consistent with suppressed electron-phonon coupling upon Cs incorporation. Figure 4e shows the full-width half-maximum (FWHM) of the PL peak of SCI-FA 0.91 Cs 0.09 PbI 3 and FAPbI 3 perovskites (Supplementary Fig. 16) ranging from 110 to 296 K. The wide broadening of the PL linewidth in FAPbI 3 perovskites arises from strong electron-phonon coupling [35,36]. The electron-phonon interaction is dominated by high energy longitudinal optical (LO) phonons in the high-temperature region, where the measured FWHM data could be fitted by the Boson model (Fig. 4c, Supplementary Fig. 18 and Supplementary Table 3). Compared with FAPbI 3 , both the electron-LO phonon coupling coefficient ( LO ) and LO phonon energy (hω) in the SCI-FA 0.91 Cs 0.09 PbI 3 are significantly reduced, indicating that the fluctuation of the PbI 6 octahedra cage in SCI-FA 0.91 Cs 0.09 PbI 3 is associated with much smaller energies upon the Cs sequential incorporation. This is consistent with the previous theoretical investigations, which indicate that mixed A-site cations could reduce the lattice fluctuations in halide perovskites [37]. As such, the suppressed lattice fluctuations and electron-phonon coupling in SCI-FA 0.91 Cs 0.09 PbI 3 rationalize suppressed ionic migration and hence enhanced stability in SCI-FA 0.91 Cs 0.09 PbI 3 PSCs, which agree with previous research results that the suppressed lattice fluctuations and reduced electron-phonon coupling could suppressed the formation of iodide-rich clusters to improve the stability of halide perovskites [38,39].

CONCLUSION
In summary, we successfully develop a novel SCI strategy to tackle the critical challenge of different crystallization dynamics of different cations in developing FA 1-x Cs x PbI 3 perovskite PSCs. The resulting pure iodide SCI-FA 1-x Cs x PbI 3 perovskites show more uniform composition distribution and reduced defects/traps density than FAPbI 3 and one-step crystallized 1S-FA 0.91 Cs 0.09 PbI 3 . Compared with FAPbI 3 , the SCI-FA 0.91 Cs 0.09 PbI 3 exhibits reduced electron-phonon coupling and lattice fluctuations, minimizing ion migration and hence enhancing the stability. As such, we have been able to achieve highly stable PSCs with a high efficiency of 24.7%, which is a record for SCI-FA 1-x Cs x PbI 3 PSCs. This work opens up new possibilities to develop high-quality mixed-cation perovskites, presenting a milestone towards the development of highly efficient and highly stable perovskites for various applications, including solar cells, light-emitting diodes and lasers.