Recent progress in Pb-, Sn-, and Pb–Sn-based inorganic perovskite solar cells: toward enhanced stability and efficiency

The Goldschmidt tolerance factor (TF, t) and octahedral factor (OF, m) are common indicators for stable three-dimensional perovskites. The t and m are defined as and r_B/r_X, respectively, where r_A/B/X means the ionic radius of A (Cs⁺), B (Pb²⁺ or Sn²⁺) and X (I⁻), schematics of which can be found in Fig. 3a.⁴⁰ For ideal perovskites a tolerance factor in the range 0.9–1.0 yields a stable cubic structure, whereas significantly lower or higher values favor distorted or non-perovskite structures. There have been various efforts to use t to predict the stability of 3-dimensionally (3-D) connected perovskite structures such as α, β, and γ phases.^41–44 Travis et al.⁴⁴ suggested revised stable regimes of the tolerance factor (0.875 < t < 1.06) and octahedral factor (0.41 < m) with a revised ionic radius (r_Sn = 0.97 Å and r_Pb = 1.03 Å) for iodide compounds. The predicted stable area for Sn-based halide perovskites is presented in Fig. 3b.⁴⁰ The tolerance factor of CsPbI₃ is lower than that of CsSnI₃ due to the larger Pb²⁺ ionic radius. This low tolerance factor is a primary intrinsic reason that the corner-sharing (Pb/Sn)I₆ network cannot be maintained without distortion – the Cs⁺ cation is simply too small relative to the (Pb/Sn)I₃ framework, so the structure collapses to the δ-phase (with isolated (Pb/Sn)I₆ octahedra) at room temperature. Recent research by Bartel et al. introduced a significantly improved tolerance factor, achieving about 92% accuracy in predicting perovskite stability across diverse oxide and halide compositions. This revised factor incorporates ionic radii, oxidation states (n_A), and their ratios, enabling a more accurate and generalized stability prediction. Their work utilized a data-driven SISSO method to develop an improved tolerance factor defined using:

with a threshold value (τ < 4.18) for stable perovskites (Fig. 3c).⁴³

	Fig. 3 Goldschmidt's tolerance factor, geometrical effects and phonon effects on stability. (a) The structural description of an ideal ABX3-type perovskite structure40 (reproduced with permission from Solar RRL, 2020, 4, 2000513. Copyright 2020 Wiley-VCH GmbH). (b) The stable area prediction for Sn-based halide perovskites with regard to the tolerance factor (TF, 0.875 ≤ TF ≤ 1.06) and octahedral factor (m, 0.41 ≤ μ ≤ 0.895)40 (reproduced with permission from Solar RRL, 2020, 4, 2000513. Copyright 2020 Wiley-VCH GmbH). (c) Classification using the machine learned new tolerance factor43 (reproduced with permission from Science Advances, 2019, 5, eaav0693. Copyright 2019 AAAS). (d) Phonon-driven polymorphic phase transitions among black phases (α, β, and γ) of CsPbI3, highlighting small energy differences that govern structural stability45 (reproduced with permission from ACS nano, 2018, 12, 3477–3486. Copyright 2018 American Chemical Society). (e) Comparison of the impact of different exchange correlations to calculate phase transition temperatures46 (reproduced with permission from Chemistry of Materials, 2022, 34, 8561–8576. Copyright 2022 American Chemical Society).

Despite these advancements, approaches based on tolerance factors continue to demonstrate fundamental limitations. Inconsistencies arise from different research groups employing varied ionic radius values and stability criteria, creating confusion in predicting perovskite structural stability. Indeed, several studies have erroneously predicted CsPbI₃ stability based on tolerance factor calculations alone.⁴⁷ Most notably, the approach relies purely on geometric and ionic considerations, disregarding important electronic and chemical effects such as orbital hybridization, covalency, defect chemistry, and anharmonic lattice dynamics. These chemical and electronic factors critically influence stability in real systems, especially for compounds with significant covalent bonding character or polarizable ions, thus limiting the universal applicability of tolerance factors alone.

Within the harmonic approximation, lattice vibrations are assumed to be small oscillations around equilibrium positions with linear restoring forces. In this idealized scenario, phonon dispersion calculations using density functional theory (DFT) reveal soft-mode instabilities characterized by imaginary frequencies, especially prominent at the Brillouin zone boundaries.^45,48 Such instabilities indicate that the cubic high-symmetry phases (α-phase for CsPbI₃ and CsSnI₃) are inherently unstable at low temperatures, predicting spontaneous distortions to lower-symmetry orthorhombic or tetragonal structures.

However, the purely harmonic description does not fully capture temperature-dependent stability. This discrepancy is resolved by considering anharmonic phonon effects—nonlinearities in lattice vibrations arising from significant atomic displacements at finite temperatures. Anharmonicity manifests notably in phonon–phonon interactions, double-well potentials, and temperature-dependent frequency shifts, critically influencing the thermodynamic stability of phases as shown in Fig. 3d.^45,49,50 This dynamic stabilization explains the persistence of metastable black phases under conditions far from equilibrium, such as rapid quenching or lattice strain. Understanding phonon-driven stability mechanisms offers practical strategies for enhancing phase stability. For instance, compositional tuning through doping or alloying can modify phonon spectra, reducing anharmonic instabilities and suppressing undesirable phase transitions. Additionally, controlled lattice strain and microstructural engineering can selectively stabilize desired perovskite phases by altering vibrational modes and entropy contributions.⁵¹ Accurate modeling of phonon properties is essential; recent research combining random-phase approximation (RPA) calculations with machine learning potentials (MLPs) has demonstrated significant sensitivity in predicted phase transition temperatures. For example, employing different exchange–correlation (XC) functionals or inadequate structural optimizations can cause deviations in predicted transition temperatures exceeding 100 K. Specifically, precise calculations using RPA combined with MLP corrections yielded transition temperatures within 50 to 100 K of experimental values, highlighting the necessity of carefully chosen computational methods for accurate stability predictions (Fig. 3e).⁴⁶

	Fig. 4 Role of point defects. (a) Layer by layer phase transition processes between the γ-phase and MS1-phase (intermediate states having higher energy barriers)52 (reproduced with permission from Advanced Functional Materials, 2024, 34, 2308246. Copyright 2024 Wiley-VCH GmbH). (b) Transition barrier lowering by the formation of an iodine vacancy52 (reproduced with permission from Advanced Functional Materials, 2024, 34, 2308246. Copyright 2024 Wiley-VCH GmbH). (c) The conversion of subdomains from cubic to orthorhombic phases in defect-rich or fewer NCs when cooling down54 (reproduced with permission from Chemistry of Materials, ACS Materials Letters, 2019, 1, 185–191. Copyright 2019 American Chemical Society). (d) Schematic illustration of reducing the energy change from the d to the a phase by controlling Cs+ and Pb2+ vacancies55 (reproduced with permission from Chemistry of Materials, The Journal of Physical Chemistry C, 2019, 123, 9735–9744. Copyright 2019 American Chemical Society).

Experimental observations strongly support the above picture. Iodine vacancies tend to form readily under iodine-poor or destabilizing conditions (owing to their relatively low formation energy), and their presence correlates with accelerated phase degradation. For example, moisture is known to catalyze the black-to-yellow transition in CsPbI₃ by leaching iodide from the surface (via HI or I₂ formation), thereby amplifying the surface V_I concentration.⁵³ In contrast, vacancies on the cation sublattices (A-site Cs or B-site Pb vacancies) may have quite different effects. Cs vacancies, in particular, tend to stabilize the perovskite framework rather than destabilize it (Fig. 4d). Thermodynamic calculations by Kye et al. showed that the introduction of cation vacancies weakens the coupling between Cs⁺ ions and the PbI₆ octahedral network, thereby lowering the energy difference between the cubic/orthorhombic perovskite phase and the δ-phase.⁵⁵ However, Kye's theoretical study extrapolates macroscopic phase stability from single-defect formation energies, neglecting defect–defect interactions; moreover, the very high vacancy concentrations required for full stabilization have yet to be experimentally verified.

Lin et al. quantified the kinetics of this moisture-induced phase change: higher relative humidity dramatically increases the nucleation rate of the yellow phase, accelerating the black → yellow conversion (Fig. 5a and b).⁵⁸ Crucially, water itself is not consumed into new compounds (no permanent hydrates are formed); instead, it acts as a catalyst. Dastidar et al. measured a black → yellow phase-change enthalpy of ∼14.2 kJ mol⁻¹ and found that under dry inert conditions the trapped black phase remains stable up to ∼100 °C.⁵⁶ The presence of water vapor did not alter the phase equilibrium enthalpy, but sped up the transformation, confirming that moisture primarily catalyzes the phase transition rather than shifting the thermodynamics.⁵⁶ Interestingly, δ-CsPbI₃ can be easily converted into a perovskite phase by annealing, which is the distinguished feature of inorganic perovskites with hybrid perovskites. This reversible characteristic of CsPbI₃ enables its application in thermochromic solar cells (Fig. 5c and d).⁵⁹

	Fig. 5 Effect of moisture on inorganic perovskites. (a) PL micrographs of CsPbI3 at 90% RH over time under 375 nm laser excitation. Dotted lines: a guide to the eye highlighting the emergence of low-T phase crystals. Scale bars, 20 μm (ref. 58) (reproduced with permission from Matter, 2021, 4, 2392–2402. Copyright 2021 Cell Press). (b) Dependence of the nucleation rate (JN) on RH. Inset: the nucleation rate shown in the log plot58 (reproduced with permission from Matter, 2021, 4, 2392–2402. Copyright 2021 Cell Press). (c) Photograph of the low-T phase and high-T phase (orange-red-colored) thin films59 (reproduced with permission from Nature Materials, 2018, 17, 261–267. Copyright 2018 Springer Nature). (d) The stable and reversible switching of the absorption (550 nm) of the three CsPbIBr2 thin films over 100 phase transition cycles. (e) Schematic of VI and H2O driven fast δ-CsPbI3 growth59 (reproduced with permission from Nature Materials, 2018, 17, 261–267. Copyright 2018 Springer Nature). (f) Variation of H (enthalpy) during the γ-to-δ phase transition in the presence of H2O, H+, and OH− within CsPbI3. The shaded curve represents the pristine case60 (reproduced with permission from Chemistry of Materials, 2023, 35, 2321–2329. Copyright 2023 American Chemical Society).

Surface iodide vacancies serve as nucleation sites for δ-CsPbI₃ growth, with moisture amplifying these vacancies by strongly solvating halide ions at the interface (Fig. 5e).⁵³ Wylie et al. showed that surface treatment strategies are effective in improving moisture resistance, with CsI and CdI₂ treatments reducing the phase transition rate by approximately 5-fold by filling surface iodide vacancies.⁵³

Theoretical investigation revealed that H₂O and OH⁻ largely affect the kinetics by lowering the activation energy, whereas H⁺ has a negligible effect on the kinetics by calculating the phase transition barrier with DFT (Fig. 5f).⁶⁰ For CsSnI₃, moisture effects have not been thoroughly investigated as much as in the case of CsPbI₃, yet. Yang et al.'s DFT calculations presented that water adsorbed on the surface of γ-phase CsSnI₃ could induce charge transfers between the H atom in the water and adjacent I atoms and between the O atom and Cs⁺, leading to distortion of SnI₆ octahedra and structural deformation, which can be the plausible origin of phase instability.⁶¹

The recognition of water as a catalyst (rather than a reactant) was key to devising mitigation strategies: theory indicated that simply keeping films dry or introducing water-resistant interfaces could preserve the metastable black phase. Indeed, encapsulation and surface passivation (to prevent H₂O adsorption) emerged as effective measures to maintain black-phase stability in humid environments, aligning with the computational predictions.

By comparison, tin-based perovskites are extremely oxygen-sensitive. Oxygen exposure induces a chemical oxidation: Sn²⁺ (in the black perovskite) oxidizes to Sn⁴⁺, which destabilizes the perovskite lattice and generates a yellow or amorphous phase. This manifests as rapid degradation of the optoelectronic properties of CsSnI₃ in air. The pronounced oxygen sensitivity of CsSnI₃ stems from the synergy between redox chemistry and defect formation in its Sn-based lattice. Frost diagrams indicates that Pb²⁺ is in a deep asymmetric thermodynamic sink while Sn²⁺ is in a shallow sink (Fig. 6a), which can lead to oxidation pathways for Sn²⁺.⁶³ Therefore, adsorbed O₂ or O₂/H₂O rapidly extracts electrons and converts Sn²⁺ to Sn⁴⁺, which can result in V_Sn and lower the defect formation energy barrier. Tin vacancies raise the native p-doping level and accommodate the oxidized Sn⁴⁺ ions.

	Fig. 6 Effect of oxygen and light irradiation on inorganic perovskites. (a) Frost diagrams for Pb and Sn under standard conditions63 (reproduced with permission from ACS Materials Letters, 2021, 3, 299–307. Copyright 2021 American Chemical Society). (b) Cyclic degradation mechanism of a tin iodide perovskite under ambient air exposure64 (reproduced with permission from Nature Communications, 2021, 12, 2853. Copyright 2021 Springer Nature). (c) Growth of Cs2SnI6 from CsSnI3 (ref. 65) (reproduced with permission from Energy & Environmental Science, 2019, 12, 1495–1511. Copyright 2019 Royal Society of Chemistry). (d) Energy levels of Pb- and Sn–iodide perovskites. Adapted with minor modifications66 (reproduced with permission from Nature Communications, 2019, 10, 2560. Copyright 2019 Springer Nature). (e) Confocal false-color images with the red and blue colors corresponding to emission channels >600 and <600 nm (ref. 67) (reproduced with permission from Matter, Matter, 2022, 5, 1455–1465. Copyright 2022 Cell Press). (f) The total area of the I-rich regions plotted as a function of time at carrier generate rates of G1 (blue line) and G2 (red line), respectively, in comparison with the experimental data (black dots)68 (reproduced with permission from Matter, Matter, 2023, 6, 2052–2065. Copyright 2023 Cell Press).

Oxygen can trigger cyclic degradation of tin-based perovskites. Haque et al. revealed that oxidized Sn⁴⁺ complexes with I⁻, releasing SnI₄; SnI₄ hydrolyses to HI + SnO₂, and HI + O₂ ⇒ I₂ + H₂O. I₂ is itself a strong oxidant toward remaining Sn²⁺, creating a cyclic degradation loop (Fig. 6b).⁶⁴ Depletion of Sn destabilizes the already metastable black perovskite and drives it toward yellow δ-CsSnI₃ and then vacancy-ordered Cs₂SnI₆ or direct conversion into yellow Cs₂SnI₆ (Fig. 6c).⁶⁵ The high-lying Sn 5s and I 5p valence bands also facilitate hole transfer to O₂, so easy Sn oxidation, self p-doping, iodine autocatalysis, lattice metastability and favorable band energetics together make CsSnI₃ far more air-labile than its Pb counterpart (energy levels can be found in Fig. 6d).⁶⁶

Another phenomenon observed under continuous illumination is halide redistribution in mixed-halide perovskites. In CsPb(I_1−xBr_x)₃, strong light causes iodine-rich and bromine-rich regions to segregate, leading to local bandgap changes.^69,70 Peng et al. conclusively showed that light-induced Br/I segregation in single-crystalline CsPbBr_2.1I_0.9 proceeds by spinodal decomposition, quantifying both the carrier-dependent kinetics and the nanometre-scale coarsening length using in situ cryo-STEM cathodoluminescence and phase-field modelling (Fig. 6f).⁶⁸

Although this halide segregation is not directly connected with phase transformation, the ensuing I-rich micro-domains promote iodide-vacancy accumulation and local strain, which under severe operating conditions can nucleate the yellow δ-phase, so halide segregation—though structurally discrete—may indirectly accelerate phase degradation.

As mentioned above, phase transitions and instability in inorganic perovskites are induced by a variety of intrinsic and extrinsic factors, which in turn contribute to the degradation of both long-term operational stability and device efficiency. Accordingly, mitigating these instability-inducing factors, such as lattice distortions, defect formation, and environmental stressors, is essential for enhancing the long-term stability and photovoltaic performance of inorganic perovskite solar cells.

	Fig. 8 The review of stability enhancement via composition engineering with (a) GeI2 in CsPbI3 (reproduced with permission from Adv. Energy Mater., 12(10), 2103690),72 (b) Zn in CsSnI3 (reproduced with permission from J. Mater. Chem. A, 10, 23204–23211. Copyright 2022 Royal Society of Chemistry),74 and (c) GeI2 in CsPbI3 (reproduced with permission from Adv. Energy Mater., 12(10), 2103690. Copyright 2022 Wiley-VCH GmbH).73

Wang et al. proposed a facile and effective galvanic displacement reaction approach for stable CsSnI₃ perovskite solar cells.⁷⁴ Metallic Zn powder was introduced into the perovskite precursor film, which behaves as a favorable redox couple with Zn/Zn²⁺ and Sn²⁺/Sn⁴⁺ ions. The introduction of zinc powder thus successfully reduces the undesired Sn⁴⁺ species within the precursor solution and in turn results in reduced trap states in the final perovskite film. Moreover, the galvanic displacement reaction simultaneously incorporates small fractions of Zn²⁺ ions, saturating at 5% in the final perovskite lattice, affording enhanced inherent stability, as shown in Fig. 8b. The zinc incorporated CsSnI₃ perovskite film achieved 8.27% PCE with a carbon electrode and retained 86.3% PCE after 216 hours of storage in air without encapsulation.

Padture et al. demonstrated a stable lead-free perovskite light absorber through Sn and Ge alloys. A solid solution perovskite of CsSn_xGe_1−xI₃ was prepared to stabilize the Goldschmidt tolerance factor and octahedral factor.⁷³ The optimal inherent structural and chemical stability was achieved in CsSn_0.5Ge_0.5I₃, which retained the perovskite crystal structure even after exposure to humid air for 72 hours as shown in Fig. 8c. The CsSn_0.5Ge_0.5I₃ film was prepared through vacuum thermal evaporation of the alloy powder, and the successful Ge incorporation in the resulting film was confirmed with the optical bandgap reaching 1.50 eV. Notably, the Ge incorporated film spontaneously forms native GeO₂ oxide at the surface when exposed to air, providing a passivation layer against detrimental oxygen and moisture. Accordingly, the CsSn_0.5Ge_0.5I₃ alloy perovskite solar cell exhibited promising storage stability in air without encapsulation, maintaining 91% of its initial PCE after 100 hours, and demonstrated excellent operational stability, maintaining 92% of its initial PCE after 500 hours of continuous operation in an inert environment.

Min et al. also adopted halide compositional engineering by introducing CsCl within the precursor solution for stability enhancement of CsPbI₃ perovskite solar cells.⁷⁶ According to the time-of-flight secondary ion mass spectrometry spectra and X-ray diffraction pattern, the chloride ions remain at the bottom buried layer during the film formation stage, establishing a spontaneous Cs₂PbI₂Cl₂/CsPbI₃ interface. The as-grown CsPbI₃ film exhibits suppressed trap density and superior film crystallinity, contributing to the film stability. Moreover, the superior thermodynamic stability of the Cs₂PbI₂Cl₂ interlayer inhibits the δ-CsPbI₃ phase transition predominant at the buried interface during exposure to humid environments as shown in Fig. 7c. Thus, the CsPbI₃ perovskite solar cell with CsCl incorporation exhibits 20.6% PCE, maintaining approximately 80% of its initial performance after 1000 hours of light exposure in a N₂ environment.

	Fig. 9 The review of stability enhancement via additive engineering with (a) carbazide molecule in CsSnI3 (reproduced with permission from Adv. Mater., 35(26), 2300503. Copyright 2023 Wiley-VCH GmbH),81 (b) 4-fluorobenzothiohydrazide in CsPbI3 (Adv. Funct. Mater., 34(10), 2312638. Copyright 2023 Wiley-VCH GmbH),80 and (c) acetylhydrazine in CsPb0.6Sn0.4I3 (reproduced with permission from J. Energy Chem., 72, 487–494. Copyright 2022 Elsevier).82

Similarly, Liu et al. incorporated a 4-fluorobenzothiohydrazide (FBTH) molecule in CsPbI₃ perovskite solar cells to stabilize the redox reaction during aging and to passivate defects in the final perovskite film.⁸⁰ Owing to the Lewis acidic F, S, and N atoms, the FBTH additive stabilizes the precursor solution, maintaining uniform colloidal size distribution during storage and regulating the crystallization process, enhancing film morphology with larger grain size. Moreover, the FBTH molecule acts as an effective redox agent to suppress I₂ and Pb⁰ formation during aging of the CsPbI₃ perovskite film. In addition, the enhanced hydrophobicity of FBTH further contributes to the phase stability of the resulting film as shown in Fig. 9b. As a result, the CsPbI₃ perovskite solar cell prepared with TBFH demonstrated 92.6% and 85.63% PCE retention during 900 hours of storage and during 200 hours of operation at the MPP, respectively.

Yang et al. reported an acetylhydrazine additive in the CsPb_0.6Sn_0.4I₃ perovskite precursor solution to demonstrate an antioxidative solution processing method.⁸² The Sn²⁺ ions in the precursor solution are susceptible to spontaneous oxidation to the Sn⁴⁺ oxidation state, resulting in abundant deep trap states in the control film. The acetylhydrazine addition successfully reduces the Sn⁴⁺ oxidation states to desirable Sn²⁺ states and further prevents oxidation under thermal stress, as shown in Fig. 9c. The acetylhydrazine incorporation successfully suppresses trap states in the resulting film and further protects the resulting CsPb_0.6Sn_0.4I₃ film from oxidation in ambient air owing to the favorable adsorption energy on the perovskite surface compared to oxygen molecules. Accordingly, the CsPb_0.6Sn_0.4I₃ perovskite solar cell with acetylhydrazine demonstrates excellent storage stability in an N₂ environment for 3500 hours and in an ambient environment for 85 hours and further exhibits remarkable operational stability, maintaining 90% of its initial PCE for 1000 hours of operation.

	Fig. 11 The review of stability enhancement via surface post-treatment. (a) Post-treatment with Yb(TFSi)3 on an inorganic perovskite surface (reproduced with permission from Energy Environ. Sci., 2024, 17, 7271–7280. Copyright 2024 The Royal Society of Chemistry),104 (b) passivating surface defects by applying post-treatment with Boc-S-4-methoxy-benzyl-L-cysteine (BMBC) (reproduced with permission from Adv. Mater., 2023, 35, 2301140. Copyright 2023 Wiley-VCH GmbH),107 (c) suppressing Sn2+ oxidation and passivating defects in inorganic Pb–Sn-based perovskite through post-treatment with 4-fluorophenylcarbothioamide (F-TBA) (reproduced with permission from ACS Appl. Mater. Interfaces, 2023, 15(30), 36594–36601. Copyright 2023 American Chemical Society),112 (d) surface defect passivation through CsF post-treatment induced surface reconstruction (reproduced with permission from Nature Energy, 2023, 8, 372–380. Copyright 2023 Springer Nature),116 (e) suppression of Sn2+ oxidation and defect passivation in Pb–Sn-based inorganic perovskites through surface reconstruction induced by 1-(4-fluorophenyl) piperazine (1-4FP) additive incorporation and post-treatment (reproduced with permission from Chemical Engineering Journal, 2024, 479, 147554. Copyright 2023 Elsevier B.V.),91 and (f) formation of a moisture protection layer via surface treatment with 3-aminopropyltriethoxysilane (APTES) (reproduced with permission from Chemical Engineering Journal, 2024, 497, 154706. Copyright 2024 Elsevier B.V.).118

As shown in Fig. 11b, Zhang et al. proposed a method for passivating surface defects by applying post-treatment with Boc-S-4-methoxy-benzyl-L-cysteine (BMBC), a molecule containing multiple functional groups.¹⁰⁷ The multiple Lewis bases of –NH, –S, and –CO in BMBC formed strong interaction with undercoordinated Pb²⁺ through Lewis base–acid reactions and suppressed the formation of halide vacancies. Additionally, the tert-butyl group, which exhibits hydrophobic properties, was uniformly distributed on the surface, effectively preventing moisture penetration. The BMBC-treated unencapsulated device exhibited high thermal stability, maintaining 89.3% of its initial efficiency for 120 h at 85 °C in a nitrogen-filled glovebox. Additionally, it demonstrated excellent storage stability, retaining 91.8% of its initial efficiency for 720 h at a relative humidity of 20–35% at ambient temperature.

As shown in Fig. 11c, Zhang et al. suggested a strategy to suppress Sn²⁺ oxidation and passivate defects in inorganic Pb–Sn perovskite through post-treatment with 4-fluorophenylcarbothioamide (F-TBA).¹¹² The CS and NH₂ groups of F-TBA formed strong coordinated interactions with Sn²⁺, suppressing its oxidation and consequently reducing defects formed by Sn⁴⁺. In addition, the hydrophobic nature of fluorine enhanced moisture resistance. The F-TBA-modified cell retained 90% of its original efficiency after continuous operation at the MPP at 55 °C for 900 h and maintained over 72% of its initial performance for 100 h at 60% relative humidity at ambient temperature.

Tan et al. proposed a method for passivating surface defects by applying post-treatment with benzyl trimethylammonium bromide (BTABr).¹¹⁷ After post-treating the CsPbI₃ perovskite surface with BTABr and applying low-temperature annealing, Br⁻ ions diffuse into the CsPbI₃ film, passivating undercoordinated Pb²⁺ and inhibiting Pb cluster formation in both the bulk and on the surface of CsPbI₃. Moreover, through I/Br ion exchange, a gradient BTA⁺–CsPbI_3−xBr_x heterostructure was formed on the surface, improving energy level alignment. The BTABr-treated device exhibited good storage stability, maintaining over 90% of its initial efficiency for 1000 h under ambient conditions with 10–15% RH and 25 °C. Furthermore, it retained nearly 100% of its performance for 400 h under continuous white LED (6500 K) illumination and at a bias voltage of 0.85 V in a nitrogen-filled glove box.

As shown in Fig. 11e, Zhang et al. suggested a method to suppress Sn²⁺ oxidation and passivate defects by simultaneously applying 1-(4-fluorophenyl) piperazine (1-4FP) additive incorporation and post-treatment to Pb–Sn based inorganic perovskites.⁹¹ The surface post-treatment with 1-4FP reconstructed the film, transforming the pinhole-rich and poorly covered Pb–Sn inorganic perovskite film into a more uniform and smoother film. Additionally, 1-4FP passivated the Sn vacancy defect by inhibiting the oxidation of Sn²⁺ through strong interaction with Sn²⁺. The 1-4FP-treated perovskite film exhibited improved humidity and thermal stability, while the device demonstrated enhanced long-term storage stability.

As shown in Fig. 11f, Li et al. demonstrated an in situ surface reconstruction using siloxane surfactants to form a protective layer that inhibits moisture infiltration.¹¹⁸ When 3-aminopropyltriethoxysilane (APTES) was surface-treated on the perovskite surface, siloxanes reacted with moisture in the air to form a hydrophobic Si–O–Si cross-linked network layer, which impeded the infiltration of moisture and oxygen. The water contact angle of the APTES-treated CsPbI₃ film significantly increased from 52.25° to 76.5°, demonstrating enhanced hydrophobicity. Furthermore, the –NH₂ groups in APTES were converted to –NH₃⁺, interacting with I⁻ and anchoring to non-coordinated Pb²⁺, which demonstrated defect passivation and the suppression of ion migration. The APTES-treated inorganic perovskite film maintained the black phase for 216 h when exposed to air at 20–25 °C and 30–40% relative humidity. Additionally, the unencapsulated device retained over 80% of its initial efficiency for 500 h at around 30% relative humidity, while the encapsulated device maintained 95% of its initial efficiency for 500 h at 85% relative humidity, demonstrating improved stability.

Hea et al. synthesized OMXene plates by oxidizing Ti₃C₂T_x MXene and applied them as a surface post-treatment on CsPbI₃ to form a physical protective layer that prevents moisture ingress.¹²⁰ By adjusting the oxidation time of MXene, OMXene with a TiO₂ electron-transporting shell on its surface was synthesized. As a result, the electron selectivity was enhanced and the formation of a strong electric field at the OMXene interface improved charge extraction. The OMXene-treated CsPbI₃ mini-module encapsulation demonstrated robust stability maintaining 90% of its initial efficiency at 85 °C and 85% relative humidity environments during 1 sun light soaking for 1000 h.

As shown in Fig. 12a, Qiu et al. adapted dipolar chemical bridge (DCB) materials at the ETL/inorganic perovskite interface.¹²² The DCB material, 3-amino-1-propane-sulfonic acid (SA), generated a strong electric field at the interface, optimizing the interfacial energy-level alignment and facilitating charge extraction. SA passivated defects through strong interactions with uncoordinated Ti⁴⁺ on the TiO₂ surface, I⁻ ions and uncoordinated Pb²⁺ in the buried perovskite interface. Furthermore, by enhancing the contact with the perovskite, it effectively reduced non-radiative recombination. Furthermore, the interface treatment with SA improved the crystallinity and relieved tensile strain of the inorganic perovskite, thereby enhancing the durability of the inorganic perovskite film. The SA-treated unencapsulated device maintained 87.1% of its initial efficiency after 720 h of storage at 35 ± 5% RH. Additionally, it retained 91.1% of its initial efficiency after 300 h of operation under continuous 100 mW cm⁻² illumination.

	Fig. 12 The review of stability enhancement via interface optimization. (a) ETL/inorganic perovskite interface optimization via TiO2 surface treatment with 3-amino-1-propane-sulfonic acid (SA) (reproduced with permission from Angew. Chem. Int. Ed. 2024, 63, e202401751. Copyright 2024 Wiley-VCH GmbH),112 (b) suppression of Sn2+ oxidation and in situ formation of a SnO2 ETL via the surface Sn(IV) hydrolysis (SSH) technique in Pb–Sn-based inorganic perovskite solar cells (reproduced with permission from ACS Energy Lett., 2023, 8(2), 1035–1041. Copyright 2023 American Chemical Society),127 (c) modification of the SAM surface with 4-(aminomethyl)-N,N-diphenylaniline iodide (TPAI) to suppress SAM aggregation and improve molecular ordering (reproduced with permission from Angew. Chem., 2025, e202502221. Copyright 2025 Wiley-VCH GmbH),129 (d) suppression of Li+ ion migration and passivation of perovskite interface defects by incorporating the biocompatible molecule tryptamine (TA) into the HTL (reproduced with permission from Adv. Mater., 2024, 36, 2306982. Copyright 2023 Wiley-VCH GmbH),131 (e) surface defect passivation, lattice strain buffering, and environmental barrier effect achieved by forming a 3D/0D heterostructure through deposition of 0D Cs4PbBr6 nanocrystals on CsPbI3 (reproduced with permission from Adv. Mater., 2024, 36, 2408387. Copyright 2024 Wiley-VCH GmbH),134 and (f) optimization of the CsPbI3/carbon interface in HTM-free devices via CF3-PAI interface treatment (reproduced with permission from Small, 2024, 20, 2402061. Copyright 2024 Wiley-VCH GmbH).136

As shown in Fig. 12b, Hu et al. applied the surface Sn(IV) hydrolysis (SSH) technique in Pb–Sn based inorganic perovskite solar cells to suppress Sn²⁺ oxidation and form an in situ SnO₂ ETL, thereby enhancing charge transport and extraction.¹²⁷ As a first step, SnF₂ was coated onto the perovskite surface to reduce Sn⁴⁺ to Sn²⁺, while also inducing Sn²⁺ to fill defects on the surface. Subsequently, a second surface treatment was performed using an IPA solution containing a trace amount of H₂O, forming an ultrathin SnO₂ film. The in situ SnO₂ film formed through secondary treatment suppressed the oxidation of Sn²⁺ in Pb–Sn based inorganic perovskites and improved energy-level alignment, resulting in enhanced device stability and charge extraction capability. The unencapsulated device employing the SSH technique demonstrated long-term stability, maintaining over 90% of its initial efficiency for 958 h under 1 sun continuous illumination in a nitrogen-filled glovebox.

Recently, to address these issues, strategies have been reported that either replace conventional HTL materials with self-assembly materials (SAMs)¹⁴⁰ or enhance stability by suppressing chemical reactions through SAM surface treatment.¹²⁸ However, despite these advantages, the uneven distribution of the SAM layer due to agglomeration causes defects to form at the HTL/perovskite interface, ultimately degrading stability.^129,130

As shown in Fig. 12c, Liu et al. demonstrated that the modification of 4-(aminomethyl)-N,N-diphenylaniline iodide (TPAI) to the SAM surface enhances stability.¹²⁹ TPAI bonded to the SAM through π–π interactions, suppressing SAM aggregation and improving SAM order. In addition, TPAI passivated defects at the CsPbI₃ surface by interacting with Pb²⁺ through its –NH₃ group and optimized the energy-level alignment at the HTL/perovskite interface, enhancing charge extraction and suppressing charge recombination. The TPAI-treated perovskite solar cell maintained 96.71% of its initial efficiency after 1400 h of maximum power point (MPP) tracking and retained 95.88% of its initial efficiency even after 2000 h of storage in air, demonstrating improved stability.

Another approach is to modify the HTM itself. In the case of n–i–p structured devices, the commonly used HTL, spiro-OMeTAD, contains additives such as LiTFSi to enhance its conductivity. The LiTFSi doped spiro-OMeTAD is highly sensitive to humidity and temperature variations, leading to Li⁺ ion migration. Additionally, due to the high moisture affinity of Li⁺ ions, it can absorb humidity and react with the perovskite layer, causing its degradation.^131,132

As shown in Fig. 12d, Liu et al. proposed a method to suppress Li⁺ ion migration and passivate perovskite interface defects by incorporating the biocompatible molecule tryptamine(TA) into the HTL.¹³¹ The –NH₂ group of TA formed strong bonds with undercoordinated Pb²⁺ on the perovskite surface, reducing the interface defect density. Furthermore, TA formed TA:Li⁺ complexes, reducing moisture absorption and suppressing Li⁺ ion migration, thereby enhancing the stability of the perovskite layer. The device incorporating the TA-doped HTM maintained 90% of its initial efficiency after 800 h of storage at 25–35% RH. Additionally, in the thermal stability test, it retained 83% of its initial efficiency after 250 h at 65 °C.

Zhao et al. formed a 2D Cs₂PbI₂Cl₂ capping layer on the CsPbI₃ surface by applying CsCI post-treatment.¹³³ Through CsCl surface treatment, Cl⁻ ions passivated I vacancy defects, while the 2D Cs₂PbI₂Cl₂ capping layer increased the activation energy to 0.536 eV, thereby suppressing iodide ion migration from the perovskite to the HTL under thermal and light stress, ultimately enhancing stability. The encapsulated 3D CsPbI₃/2D Cs₂PbI₂Cl₂ device demonstrated a remarkable T₈₀ lifetime of 2100 h using the ISOS-L-3 protocol measured at the MPP under simultaneous light exposure (120 mW cm⁻²) and in a heating environment at 110 °C.

As shown in Fig. 12e, Heo et al. synthesized 0D Cs₄PbBr₆ nanocrystals on the CsPbI₃ surface and subsequently applied them via the spray coating process, forming a 3D/0D heterostructure.¹³⁴ 0D Cs₄PbBr₆ consists of isolated PbBr₆⁴⁻ octahedra surrounded by Cs⁺ ions, effectively passivating Pb²⁺ defects on the perovskite surface. Also, the Cs₄PbBr₆ lattice spacing, similar to that of 3D perovskites, acts as a lattice strain holder, improving both phase and thermal stability. Furthermore, it has been shown to function as a barrier against environmental stress. The encapsulated device with 0D Cs₄PbBr₆ treatment retained 92.48% of its initial efficiency under simultaneous 1 sun illumination and damp heat (85 °C/85% relative humidity) for 1000 h.

In contrast, the chemical inertness of carbon electrodes fundamentally prevents electrode corrosion while the HTL-free structure enhances moisture resistance, thereby improving stability. However, despite these advantages, the abundant surface defects of inorganic perovskites and the low hole selectivity of carbon electrodes lead to decreased durability and performance degradation.^{136,137,143,144}

As shown in Fig. 12f, Zhang et al. demonstrated a method utilizing a dipole electric field at the CsPbI₃/carbon interface to enhance device performance and stability.¹³⁶ The –NH₃ group of CF₃-PAI bonded to the perovskite surface, while the –CF₃ group interacted with the carbon electrode. This alignment generated a well-ordered dipole electric field within the perovskite/carbon electrode interface, enhancing hole selectivity and interfacial charge separation. In addition, CF₃-PAI passivated defects by interacting with uncoordinated Pb²⁺, which suppressed non-radiative recombination and contributed to the improvement of phase stability. The unencapsulated CF₃-PAI-treated HTM-free device maintained 90% of its initial efficiency over 1000 h under storage conditions with relative humidity below 20%. Moreover, under continuous illumination at 100 mW cm⁻² for 200 h, it retained 87.7% of its performance, demonstrating enhanced operational stability.

Zhang et al.¹⁴⁴ and Lin et al.¹³⁷ treated the inorganic perovskite surface with choline iodide (ChI) and thiocholine iodide (TChI), respectively, forming a low-dimensional perovskite layer, which passivated the inorganic perovskite surface and enhanced charge extraction capability by improving energy-level alignment. In particular, Zhang et al.¹⁴⁴ enhanced stability by applying ChI in two steps, first at the wet film intermediate phase and then as a second treatment on the dried film, effectively passivating both bulk and surface defects in the inorganic perovskite.

Another strategy for overcoming the low stability induced by hygroscopic additives to the HTL is to use self-standing inorganic HTLs such as CuSCN and phthalocyanine (CuPc) that do not require hygroscopic additives.¹⁴⁵ Moreover, inorganic HTLs are generally more robust both mechanically and chemically, compared to soft organic HTLs such as spiro-OMeTAD, thus providing a significant advantage in long-term thermal and environmental stability.¹⁴⁶ Similarly, inorganic ETLs such as SnO₂ can also provide robust stability when processed on top of the perovskite layer for inverted perovskite solar cells.¹⁴⁷ However, the number of reports on using inorganic CTLs for all-inorganic PSCs with high stability is extremely limited, largely due to the difficulty in processing inorganic CTLs on top of the inorganic perovskite surface, and many reports on “all-inorganic PSCs” unfortunately still rely on organic top-CTLs such as spiro-OMeTAD and PCBM. It is expected that there remains a significant amount of research to be explored in the development of “true” all-inorganic PSCs with inorganic CTLs for enhanced thermal and moisture stability.

Post-treatment of the perovskite film surface is one of the most widely used methods of defect passivation, especially since the perovskite surface is most vulnerable to defect formation during thin film fabrication.^{106–108,117,171,175,176} Gu et al. identified iodine vacancies as one of the most common types of surface defects for all-inorganic perovskite thin films and demonstrated the use of histamine (HA) to passivate the under-coordinated Pb²⁺ defects on the perovskite surface.¹⁷¹ HA possesses an amine group and an imidazole ring, which could coordinate with the Pb²⁺ defects while simultaneously forming a H-bond with neighboring halides, effectively resulting in increased V_OC and FF from 1.195 V and 80.9% to 1.233 V and 81.9% and a PCE increase from 19.5% to 20.8%. Similarly, many surface passivation studies thereafter have focused on using molecules that can effectively interact with under-coordinated Pb²⁺ defects, while adding extra functionalities to the passivator for added effects. Zhang et al. used a trifluoroacetamide (TFA) passivator that has two amine groups for stronger chelation with the Pb²⁺ defects.¹⁰⁸ Moreover, the trifluoro group not only applies a strong dipole to the TFA passivator, but also gives hydrophobic passivation capabilities, greatly increasing the PCE to 21.35%. In 2023, Wang et al. additionally identified iodine interstitial defects as another source of solar cell performance inhibition and used 2,6-diaminopyridine (2,6-DAPy) to strongly passivate the surface defects as shown in Fig. 16a.¹⁰⁶ The authors showed that the strong nucleophilicity of the pyridine ring is more effective in passivating the positively charged defects, effectively resulting in a large increase in V_OC and FF from 1.189 V and 80.03% to 1.252 V and 83.71% and a significant PCE increase from 19.6% to 21.8%. Amine groups are not the only functional groups that can passivate perovskite surface defects, as shown by Zhang et al. in 2023.¹⁰⁷ Here, the authors used a cysteine derivative, namely BMBC, that possesses multiple functional groups such as –NH, –S and –CO that can each interact with halide vacancy defects. Through the multiple passivation effect, the authors were able to confirm the effectiveness of this strategy through the enhanced V_OC of 1.249 V and high FF of 83.57%, yielding a high PCE of 21.75%. On the other hand, Tan et al. demonstrated the passivation of the perovskite surface with extra halide anions, supplied from an organic halide salt, BTABr.¹¹⁷ The authors showed that the bromide ions could not only passivate the surface defects but also diffuse into the perovskite bulk film and passivate deeper defects, while most of the BTA⁺ ions remain at the surface. The effectiveness of this strategy can be seen by the high efficiency of 21.31%.

	Fig. 16 The review of efficiency enhancement via reducing trap-induced recombination losses. (a) Perovskite surface defect passivation with 2,6-diaminopyridine (2,6-DAPy) (reproduced with permission from Angewandte Chemie, 2023, 62, e202305815. Copyright 2023 Wiley-VCH GmbH).106 (b) Buried interface defect passivation with 3-sulphonatopropyl acrylate potassium (SPA) (reproduced with permission from Advanced Materials, 2023, 35, 2207172. Copyright 2022 Wiley-VCH GmbH).126 (c) Bulk defect passivation via additive engineering with 2,6-pyridinedicarboxamide (PC) (reproduced with permission from Advanced Energy Materials, 2024, 14, 2400151. Copyright 2024 Wiley-VCH GmbH).94 (d) Inhibiting defect formation by releasing lattice stress with a 5-maleimidovaleric acid (5-MVA) additive (reproduced with permission from the Journal of Energy Chemistry, 2025, 100, 87–93. Copyright 2024 Science Press and Dalian Institute of Chemical Physics, Chinese Academy of Sciences).96 (e) Regulating crystal growth kinetics and transformation of the phase transition pathway with a cesium formate (CsFa) additive (reproduced with permission from ACS Energy Letters, 2024, 9, 5870–5878. Copyright 2024 American Chemical Society).161 (f) Preventing Sn2+ oxidation with a dicyandiamide (DCD) antioxidant additive (reproduced with permission from the Chemical Engineering Journal, 2023, 452, 139697. Copyright 2022 Elsevier B.V.).82

The bottom interface (or the buried interface) of perovskite thin films is also a well-known hotspot for detrimental defects, and their passivation must consider interactions of the passivating agent with both the perovskite defects and the CTL defects. Typically, a proper passivating agent would consist of functional groups such as amine groups or ammonium halides, for the passivation of the perovskite layer, and acids such as carboxylates and sulfonates for the passivation of the bottom layer, which is usually a metal oxide.^122,126,170 In 2022, Xu et al. treated a TiO₂ ETL with 3-sulphonatopropyl acrylate potassium (SPA) prior to coating the CsPbI₃ inorganic perovskite layer.¹²⁶ As shown in Fig. 16b, both the sulfonate and carboxylate groups can interact with either the oxygen vacancies of the TiO₂ layer or the halide vacancies of the perovskite layer, while the potassium cation would bind with the halides of the perovskite layer in order to inhibit formation of additional halide vacancies. Moreover, the increased hydrophilicity of the TiO₂ layer due to SPA treatment increased the perovskite crystallinity, leading to overall fewer defects, and hence an increased V_OC of 1.221 V, increased FF of 82.8%, and increased PCE of 20.72%. Meanwhile, Xu et al. treated the TiO₂/CsPbI₃ interface with a 4-amino-2,3,5,6-tetrafluorobenzoate cesium (ATFC) passivator, where the TiO₂ defects were passivated by the amine group of ATFC and the perovskite defects were passivated by the carboxylate group of ATFC.¹²³ In addition, the adjacent fluorine atoms on the benzene ring could provide additional interaction with both the TiO₂ defects and the perovskite defects, ensuring a stronger degree of passivation and a high PCE of 21.11% for the final device. Similarly, Qiu et al. used a dipolar passivator, named SA, for simultaneous passivation of TiO₂ and perovskite layers.¹²² Compared to the commonly used carboxylate groups, the sulfonate group could form a tridentate chelation with the TiO₂ defects for stronger passivation, while the hydrogen bonds between the amine group of SA and halides of perovskite passivated the CsPbI₃ layer. Overall, a high PCE of 21.86% was obtained. Similar strategies can also be applied to inverted inorganic perovskite solar cell devices as well. Zhang et al. applied 4-aminobenzenesulfonic acid (4-ABSA) to passivate the bottom NiO_x HTL surface and the buried perovskite layer interface simultaneously, achieving a PCE of 17.4% for a CsPb_0.6Sn_0.4I_3−xBr_x solar cell device.¹⁷⁰

While the majority of defects are usually congregated at interfaces, the impact of perovskite bulk defects such as point defects and grain boundaries must also be addressed. However, despite their similarity in nature to interfacial defects, since the physical location of such defects makes it difficult to passivate them through pre- or post-treatment methods, researchers have turned to applying additives to the perovskite precursor solution itself so that harmful defects are passivated within the bulk film during and after the thin-film annealing process.^{93–95,97,103,162,166,177} For example, a common approach is to use a formate additive such as a dimethylamine formate (DMAFa) or CsFa additive, since formate anions can passivate the Pb²⁺ defects that occur due to halide vacancies within the perovskite bulk film.^162,166 Similarly, in 2021, Du et al. introduced a low concentration of 1-ethyl-3-methylimidazolium hydrogen sulfate (EMIMHSAO₄) ionic liquid additive to the perovskite precursor solution.⁹⁷ The imidazolium (EMIM⁺) cations and HSO₄⁻ anions were shown to passivate electrophilic defects and under-coordinated Pb²⁺ defects, respectively, within the perovskite bulk film, resulting in a high performing device with 20.01% PCE. Meanwhile, Wang et al. used a single molecule DED or 2,2-dithienylketone (DTK) additive with the intent to passivate bulk Pb²⁺ defects with the collaborative chelating effect of the carbonyl and thienyl groups.⁹³ Here, they found that the bifacial ‘dione’ structure of DED allowed for more efficient passivation of the perovskite bulk defects, where the optimized devices yielded a PCE of 21.15%. A notable aspect of additive engineering is that the additives more often than not fulfill multiple niches, as shown in the studies of Wang et al. using a 2,6-pyridinedicarboxamide (PC) additive.⁹⁴ The extra stability to the perovskite lattice offered by multiple balancing interactions between the amide groups or the N-pyridine group of the PC additive and the Pb²⁺ defects, otherwise noted as a “hinge modulation” by the authors, ensured that the perovskite film with PC can withstand longer annealing times in air, thus resulting in a more crystalline perovskite film as depicted in Fig. 16c. As a result, not only did the optimized device display a high V_OC of 1.342 V, but the resulting PCE is also among the recently highest reported for inorganic perovskite solar cells at 22.07%.

The formation process of solution processed perovskite films consists of nucleation and crystal growth, where it is generally accepted that fast homogeneous nucleation and slow crystal growth are beneficial for achieving high crystallinity.^178,179 As such, Zhang et al. used a sacrificial dye Rhodamine B isothiocyanate (RBITC) to facilitate nucleation while deterring crystallization in order to increase the crystallinity of the perovskite film.⁹⁵ The authors showed that photo-thermal decomposition of RBITC yields benzoic acid molecules that act as anti-solvent to promote initial nucleation, while the SCN⁻ anions interact with the Pb sites to deter crystal growth, eventually resulting in a highly crystalline inorganic perovskite solar cell with a high PCE of 20.95%. In another study by Zai et al., DMAFa was added to the perovskite precursor, where the formate anion would interact with the exposed Pb²⁺ sites during the annealing process.¹⁶⁶ This interaction provided sufficient time for grain growth as the HCOO⁻ is gradually replaced by I⁻, and the final solar cell device with reduced defects exhibited increased V_OC, an increased FF and increased PCE at 20.40% compared to the pristine device with 18.87% PCE. Meanwhile, another risk of inorganic perovskites is the formation of undesired photo-inactive intermediate phases such as Cs₄PbI₆ that can degrade the overall photovoltaic performance, and thus it is crucial to regulate the crystallization process to prevent this. A common strategy is to incorporate the dimethylammonium (DMA⁺) additive into the precursor solution, which has been reported to aid in the formation of a stable photo-active black phase of CsPbX₃, while inhibiting the formation of Cs₄PbX₆.^163,167 A study by Cui et al. well illustrates this strategy where a dimethylamine acetate (DMAAc) molten salt additive supplies the precursor solution with extra Ac⁻ anions that can stabilize the DMAPbI₃ lattice structure, which reacts with the CsI precursors to produce a stable CsPbI₃ structure with volatile DMAI byproducts.¹⁶³ In particular, the authors emphasize the importance of the DMAAc additive that inhibits Cs₄PbI₆ formation, whereas without DMAAc, the reaction of DMAPbI₃ with CsI could produce Cs₄PbI₆. The optimized CsPbI₃ solar cell device shows a high PCE of 21.14%. Meanwhile, Sun et al. employed dimethylammonium iodide (DMAI) and lead acetate (PbAc₂) to substitute the original PbI₂ precursor.¹⁶⁷ The authors showed that the addition of DMAI increases the solubility of CsI within the precursor solution, after which the PbAc₂ introduces Ac⁻ ligands to the nuclei and slows down crystal growth as the anion is slowly exchanged with the iodide ion from DMAI. The formation of Cs₄PbI₆ is also shown to be inhibited in the optimized solar cell device, which showed a PCE of 20.17%.

Another inherent source of defects and trap-induced recombination loss in all-inorganic perovskite solar cells is the presence of lattice stress, which could potentially cause the formation of harmful defects within the perovskite layer. On one hand, the ion size mismatch of inorganic perovskites may lead to intrinsic tensile strain throughout the bulk perovskite film.^87,96,180 In 2024, Zhang et al. introduced graphdiyne (GDY), an sp-hybridized carbon framework, to be incorporated within the CsPbI₃ bulk film.⁸⁷ The authors found that GDY not only delays the crystal growth process for the formation of a highly crystalline perovskite film, but it also helps release lattice stress within the bulk film during the annealing process, as noted in depth-dependent crystallographic analysis. The optimized device showed impressive V_OC and FF values at 1.191 V and 83.96%, respectively, and a high PCE of 20.49%. Very recently, Sun et al. also noted the significance of releasing internal perovskite lattice stress and added a flexible molecule, 5-maleimidovaleric acid (5-MVA), as an additive within the inorganic perovskite precursor solution.⁹⁶ It was shown that, since both functional groups at either end of the 5-MVA molecule are effective electron donors, they could strongly interact with Pb²⁺ defect sites within the CsPbI_2.85Br_0.15 perovskite lattice and act as physical buffers to release tensile strain of the perovskite film, as seen in Fig. 16d. Its immediate effectiveness was shown by the depth-dependent analysis and the greatly improved photovoltaic characteristics resulting in a high PCE of 20.82%. On the other hand, the lattice mismatch and discrepancies in thermal expansion coefficients between the perovskite lattice and substrates such as metal oxides can cause greater lattice distortion at the substrate/perovskite interface, thus also harming the interfacial contact.^122,130 In the studies by Qiu et al., a flexible SA molecule based on a propyl chain that was treated on the TiO₂ surface helped release the lattice stress at the TiO₂/CsPbI₃ interface.¹²² The inverted p–i–n structured devices adopt a similar strategy through the widely accepted SAM treatment on the NiO_x substrate. Xu et al. realized that a poor SAM monolayer coverage on the NiO_x substrate could cause unnecessary lattice distortions at the perovskite buried interface and treated the SAM layer with niobium pentachloride (NCL) to prevent SAM agglomeration and improve its morphology.¹³⁰ As a result, the authors observed a decrease in interfacial lattice stress that also led to an increase in optimized solar cell PCE to 21.24%, which is among the highest PCEs reported to date for inverted inorganic perovskite solar cells.

The problem of rapid crystallization could be resolved by adapting similar strategies to regulating crystallization kinetics of Pb-based inorganic perovskite thin films discussed in the previous section.^{75,81,82,152,161,169,173} For example, Chen et al. first demonstrated the poor crystallinity of a pristine CsPb_0.55Sn_0.45I₂Br perovskite film and then showcased a CsPb_0.55Sn_0.45I₂Br–CsCl perovskite film with much improved crystallinity and large, micro-sized grains.¹⁷³ It was shown that the CsCl additive provided small halogen ions that could simply infiltrate the bulk film at the grain boundaries during fabrication and strongly bond with the Sn²⁺, effectively reconfiguring the grain boundaries and inhibiting the formation of halide vacancies. With the addition of a hydrophobic PbSO₄ passivation barrier at the perovskite surface, the optimized device displayed an impressive PCE of 10.39% with a highly improved V_OC and FF of 0.70 V and 71.82%, respectively. As such, due to the lower solubility and faster ion release of the Sn-based precursors compared to the Pb-based precursors, it is important to alleviate this discrepancy in order to form a homogeneously distributed perovskite film. Shang et al. did a good job in demonstrating this idea as they partially substituted PbI₂ with PbAc₂ in the perovskite precursor solution.¹⁶⁹ As shown in this work, the release rate of Sn²⁺ within the precursor solution is matched by the release of Pb²⁺ by PbAc₂, while an additional MAI additive supplies extra iodide anions for the fabrication of a homogeneous phase-pure Pb–Sn perovskite solar cell with a high PCE of 13.98%. Recently, Yu et al. fabricated CsSnI₃ perovskite solar cells with a CsFa additive in order to control the crystallization kinetics, yielding a high certified PCE of 13.68%, which is one of the highest reported efficiencies for CsSnI₃ devices to date.¹⁶¹ Here, the authors proposed that the introduction of a pseudohalide anion that could strongly bond with the Sn²⁺ cations could dramatically slow down crystallization by transforming the phase transition path as shown in Fig. 16e.

Meanwhile, the oxidation of Sn²⁺ to Sn⁴⁺ provided new challenges to the Sn-based inorganic perovskite solar cell community. To address this challenge, many studies have actively sought antioxidants or reducing agents in order to inhibit or reduce the Sn²⁺ oxidation.^{81,92,99,100,102,156,173,174} For example, some cases report the effectiveness of Zn-based additives to prevent Sn²⁺ oxidation based on the more suitable reduction potential of Zn²⁺ compared to Sn²⁺.¹⁷³ For example, zinc oxalate additives within the CsPb_0.7Sn_0.3PbIBr₂ perovskite precursor solution would not only reduce any present Sn⁴⁺ to Sn²⁺, but also the oxalate anions could strongly interact with the metal ions to regulate crystal growth. The resulting Pb–Sn-based solar cell device showed a champion PCE of 14.1%. However, such strategies would also have to consider the possible incorporation of Zn²⁺ within the inorganic perovskite lattice, which could alter the intrinsic properties of the perovskite thin film. A study by Wen et al. in 2023 used a dicyandiamide (DCD) additive to prevent oxidation of Sn²⁺.⁹⁹ The authors noted that the presence of –CN and –CNH functional groups would provide a strong interaction with the Sn²⁺ ions within the perovskite film, actively preventing oxidization and reducing deep level defects, for a high PCE of 14.17% in the optimized solar cell device as shown in Fig. 16f. Similarly, many researchers have incorporated antioxidative additives such as carbazide (CBZ),⁸¹ dimethyl ketoxime (DMKO),¹⁵⁶ phthalimide (PTM),¹⁰⁰ 1-ethyl-3-methylimidazolium acetate (EMIMAc),⁹² thiosemicarbazide (TSC),¹⁷⁴ and 1-(4-carboxyphenyl)-2-thiourea (CPT)¹⁰² within CsSnI₃-based precursor solutions and have shown promising results.

	Fig. 17 The review of efficiency enhancement via modifying interface charge carrier dynamics. (a) Improving energy level alignment at the perovskite/ETL interface via a p- to n-type transition at the perovskite surface with propylamine hydrochloride (PACl) treatment (reproduced with permission from Energy & Environmental Science, 2023, 16, 2572–2578. Copyright The Royal Society of Chemistry 2023).183 (b) Improving hole transfer at the perovskite/HTL interface via introducing a dipole layer with highly polar 2,2,2-trifluoroethylammonium iodide (F3EAI) treatment at the perovskite surface (reproduced with permission from Advanced Materials, 2022, 34, 2202735. Copyright 2022 Wiley-VCH GmbH).114 (c–e) Improved charge transfer induced by internal p–n heterojunctions at the inorganic perovskite layer. (c) 3D-CsPbI3/0D-Cs4PbI6 heterojunctions via surface reconstruction with benzyldodecyl-dimethylammonium bromide (BDABr) treatment at the perovskite surface (reproduced with permission from Science Bulletin, 2023, 68, 706–712. Copyright 2023 Science China Press).142 (d) Composition graded CsPbI3−xBrx heterojunction via spray deposition of CsPbI3 on top of a CsPbI2Br layer (reproduced with permission from Joule, 2021, 5, 481–494. Copyright 2020 Elsevier Inc.).168 (e) Phase heterojunction formed by a two-step process of spin coating β-CsPbI3 followed by vacuum evaporation of γ-CsPbI3 (reproduced with permission from Nature Energy, 2022, 7, 1170–1179. Copyright R. Ji, Y. Vaynzof et al. 2022).184

Other studies of interface modification have focused on treating the interface with molecules that exhibit strong dipoles such as when a dipole monolayer is formed, which could either facilitate charge extraction or hinder charge back-recombination.^{112,114,122,135,136,138} Zhang et al. treated the CsPbI₃ perovskite surface with highly polar 2,2,2-trifluoroethylammonium iodide (F₃EAI) molecules, which they expected to provide a favorable path for hole extraction at the perovskite/HTL interface and yield a high PCE of 20.50%, as described in Fig. 17b.¹¹⁴ It was shown that polar passivators would orient the positive dipole towards the perovskite surface, effectively attracting and passivating negatively charged defects that would otherwise trap holes and hinder extraction. Zhang et al. also treated the CsPbI₃/carbon–electrode interface with 4-trifluoromethyl-phenylammonium iodide (CF₃-PAI), which would align itself so that the positive dipole of the CF₃ group interacts with the perovskite surface.¹³⁶ Here, the authors showed that, while a direct contact between the perovskite and the carbon electrode layers would lead to back recombination since electrons could also cross the interface, the presence of the dipole layer would then deter electron transfer and ensure that only holes could selectively cross the interface. Meanwhile, some studies have demonstrated that dipole layer treatment at the buried interface could also be a viable strategy. In 2023, Wang et al. used polar molecules of 4-amino-1-naphthalene sulfonate (4A1N) to functionalize the TiO₂ substrate and subsequently interact with the CsPbI₃ perovskite layer via the amino group.¹³⁸ The resulting ion–dipole interaction yielded a large shunt resistance across the interface, even under low light conditions, thus preventing back-recombination across the interface and enabling a high PCE of 20.03% and a high indoor photovoltaic PCE of 41.21% with an output power of 137.66 μW cm⁻².

In 2023, Chen et al. treated a CsPbI₃ perovskite film with a benzyldodecyl-dimethylammonium bromide (BDABr) salt to demonstrate a surface reconstruction process where in situ 3-dimension-to-0-dimension phase transformation was observed.¹⁴² While the 0D Cs₄PbI₆ phase is generally detrimental to the perovskite film due to its photo-inactivity, in this case, the 0D phase would actually accumulate on the grain boundaries to passivate pre-existing defects and also slightly shift the surface energy level such that the abundant 3D/0D heterojunctions at the surface instead aided in efficient hole transfer, as shown in Fig. 17c. Due to the reduced defects and increased charge transfer, the optimized device showed a high PCE of 20.63%. A similar concept was later demonstrated by Heo et al. where 0D Cs₄PbBr₆ perovskite NCs were synthesized and spray-deposited on the surface of CsPbI₃ films.¹³⁴ The authors showed that the 3D/0D heterojunctions could effectively reduce lattice strain, while also providing an electron-selective junction for efficient electron extraction, and the resulting p–i–n solar cell devices yielded a high PCE of 18.78%.

Besides dimension-engineered heterojunctions, the formation of heterojunctions via compositional control was also studied. Heo et al. suggested a creative approach towards creating composition-graded heterojunctions, where a thin film of CsPbI₂Br was first spray-deposited on a TiO₂ substrate, followed by additional spray-deposition of CsPbI₃ in order to create a CsPbI_3−xBr_x film with an internal graded heterojunction where the halide ratios (and the effective energy levels) changed with depth, as shown in Fig. 17d.¹⁶⁸ The fully spray-coated process not only provided an ideal band structure alignment for efficient charge extraction through an enlarged depletion region, where the unit cell device exhibited the best PCE of 16.81%, but also allowed for the realization of a stable process for efficient fully scalable inorganic perovskite photovoltaic devices. In 2023, Chu et al. reported another method of achieving composition-graded heterojunctions for CsPbI_xBr_3−x solar cells with a high PCE of 21.02% and high V_OC of 1.27 V.¹¹⁶ Here, the authors showed that CsF salt treatment on CsPbI_xBr_3−x films, followed by annealing, could trigger a solid-state reaction between the perovskite surface and CsF salt, proven by the wider bandgap measured at the surface. Through optimized parameters, the authors reported a high PCE of 21.02%.

A unique merit of CsPbI₃ inorganic perovskites is that the perovskite material can exhibit multiple photo-active “black” phases, which can be aligned with each other for efficient charge carrier extraction. In 2022, Ji et al. reported PHJ solar cells by creating a γ-CsPbI₃/β-CsPbI₃ PHJ.¹⁸⁴ The PHJ was fabricated through a two-step process, where the β-CsPbI₃ layer was first solution-processed through a spin deposition procedure with a DMAI additive and then the subsequent γ-CsPbI₃ layer was vacuum evaporated for deposition on the β-CsPbI₃ layer, as seen in Fig. 17e. Through careful optimization of the thickness of each perovskite layer, the authors achieved a high PCE of 19.1%. Mali et al. also demonstrated γ-CsPbI₃/β-CsPbI₃ PHJ solar cells in 2023 and increased the device PCE to an impressive 21.59%.¹⁵⁹ In order to further stabilize both layers and optimize energy level alignment, the authors incorporated a bis(pentafluorophenyl)zinc (Zn(C₆F₅)₂) p-type additive within the bottom β-CsPbI₃ layer, while also co-evaporating a small amount of guanidinium iodide (GAI) alongside the other precursors for the fabrication of a GAI-incorporated γ-CsPbI₃ layer.