Mixed halide perovskites (MHPs) have attracted attention due to their tunability of optoelectronic properties and especially the bandgap, which is useful for tandem solar cells. Unfortunately, MHPs undergo phase separation under illumination. It can form low-bandgap iodide-rich phases as charge recombination centers, causing the reduction of open-circuit voltage (V_oc) and device photoinstability. To address this issue, many approaches have been used^[1].

In ABX₃ MHPs, A-site can be CH₃NH₃⁺ (MA⁺), CH(NH₂)₂⁺ (FA⁺) or Cs⁺, B-site can be Pb²⁺ or Sn²⁺, and X-site is dominated by halide anion (Cl^–, Br^– or I^–). The composition engineering on A/B/X sites can effectively alleviate the photoinduced phase separation. Photoinduced phase separation is common in MHPs with single cation at A-site, in which typical red shift in photoluminescence (PL) upon light illumination can be observed^[2]. A-site doping is an effective way to retard phase segregation. The introduction of Cs⁺ into A-site could inhibit photoinduced phase segregation^[3]. A time-dependent red-shift in PL was observed for MAPb(I_0.6Br_0.4)₃ film (Fig. 1(a)). But no significant red-shift in PL was observed in FA_0.83Cs_0.17Pb(I_0.6Br_0.4)₃ film after identical light illumination. Dang et al. observed the positive effect of Cs⁺ and Rb⁺ doping on phase segregation^[4]. A-site doping mechanism was explained by the polarization model^[5]. Electron–phonon coupling can induce lattice distortion, thus causing phase separation. Phenomenological model showed that lowering electron–phonon coupling can reduce the tendency of photoinduced phase separation^[5]. A strong correlation exists between Cs⁺ doping and photoinduced halide segregation. Excessive Cs⁺ doping could also cause photoinduced phase segregation (Fig. 1(b))^[6].

Adjusting B-site composition can improve phase stability under illumination. The photoinduced phase segregation was widely reported as the intrinsic instability of CsPbIBr₂. Li et al. inhibited photoinduced phase segregation and improved device stability by partially replacing Pb²⁺ with Sn^2+[7]. The PL peak of CsPbIBr₂ split into two peaks after illumination, corresponding to iodide- and bromide-rich phases (Fig. 1(c)). No obvious photoinduced phase segregation was seen in CsPb_0.75Sn_0.25IBr₂ film^[7]. The stabilization of I/Br phase in MAPb_0.75Sn_0.25(I_1–yBr_y)₃ was also reported by Yang et al.^[8], who further speculated that internal bonding environment was changed by partial Sn substitution to suppress halogen phase separation.

I^–/Br^–-mixed MHPs present increased chemical stability and suitable bandgap for tandem cells. Xu et al. announced efficient perovskite top cells (1.67 eV) by using triple-halide alloys (Cl^–/I^–/Br^–) to tailor the bandgap and stabilize perovskite under illumination^[9]. From double-halide to triple-halide alloy, an enhancement in optoelectronic properties was observed, e.g., remarkable improvements in photocarrier lifetime and mobility, significant suppression of photoinduced phase segregation under illumination up to 100 suns (Fig. 1(d)). The corresponding solar cells maintained >96% of their initial efficiency after 1000 h operation under white light illumination at 60 °C.

Photoinduced phase segregation likes to take place at grain boundaries. Hu et al. studied the role of grain boundary's area and grain orientation in phase separation of MHPs^[10]. The enhanced crystallinity and grain size of CH₃NH₃PbI_xBr_3–x films could stabilize these materials under one sun illumination, which enhanced the device efficiency and stability (Fig. 2(a))^[10]. The grain boundary defects, particularly halide vacancies in perovskite lattice, contribute expansive channels for photo-driven migration of halide ions. Interface passivation proves to be another effective approach to suppress photoinduced phase separation. Abdi-Jalebi et al. reported that the surface and grain boundary defects could be passivated by potassium^[11]. The external photoluminescence quantum efficiency (PLQE) as a function of time for (Cs,FA,MA)Pb(I_0.85Br_0.15)₃ films were measured at excitation densities equivalent to that of solar illumination (Fig. 2(b)), and the photoinduced migration was inhibited by increasing potassium content. Zhou et al. constituted a CsPbBr₃-cluster passivated perovskite structure with high-quality CsFAMA films, where CsPbBr₃-clusters were located at the surface/interface of CsFAMA grains (Fig. 2(c))^[12]. Belisle et al. indicated that surface modification was also an effective approach to suppress the photoinduced phase separation in MHPs^[13]. Coating a perovskite surface with electron-donating ligand trioctylphosphine oxide (TOPO) could not only reduce nonradiative recombination but also retard halide segregation in CH₃NH₃PbI₂Br films.

A kinetic model raised by Draguta et al. revealed that photoinduced phase separation in MHPs could be suppressed by reducing carrier diffusion lengths (l_e/h) because the rate of phase separation is related to l_e/h^[14]. By reducing l_e/hto a certain value, phase separation can become kinetically unfavorable. To verify this, CsPb(I_0.5Br_0.5)₃ thin film (with long l_e/h) and CsPb(I_0.5Br_0.5)₃ quantum dot (QD) film (with short l_e/h) were made. The PL peak for CsPb(I_0.5Br_0.5)₃ thin film presented time-dependent red-shift while that for nanocrystal-based film was almost unchanged (Fig. 2(d)), indicating negligible phase separation. When excluding the size effect, this result proved that the shortened l_e/h in nanocrystal-based films could effectively suppress phase separation.

The light intensity^[15], ambient temperature^[16] and pressure^[17] can affect photoinduced phase separation in MHPs. The distribution of halide ions correlates with light intensity. As increasing light intensity, the photoinduced phase separation becomes more obvious^[15]. The light provides energy for ion migration in perovskite lattice. Mao et al. thought that photoinduced phase segregation could be completely reversed under sufficiently high illumination (Fig. 2(e))^[18]. With a high photocarrier density, the halide ions distribution in MHP single crystals could be tuned from a segregated state to a homogeneous state. Besides, Elmelund et al. reported that ambient temperature affected phase segregation in MHPs^[19]. Nandi et al. indicated that photoinduced phase separation took place at a narrow temperature range and above a particular bromine content (Fig. 2(f))^[16]. Furthermore, Jaffe et al. reported that (MA)Pb(Br_xI_1–x)₃ (0.2 < x < 1) thin films presented reduced PL peak shift under high pressures ^[17].

To summarize, photoinduced phase segregation is detrimental to device performance and long-term stability. Several effective strategies have been applied, i.e. composition engineering, interface engineering, reducing carrier diffusion length, etc. Further efforts should focus on phase-separation mechanism.

L. Ke thanks the Natural Science Foundation of Hunan Province (2019JJ50776). L. Ding thanks the National Key Research and Development Program of China (2017YFA0206600) and the National Natural Science Foundation of China (51773045, 21772030, 51922032, 21961160720) for financial support.