Efficient inverted FA-Cs perovskite solar modules fabricated by blade-coating on PET foil with robust encapsulation

Abstract

The commercial deployment of perovskite photovoltaics (PVs) hinges on bridging the gap between high-efficiency lab-scale devices and scalable, reliable modules. While flexible polymeric substrates offer a pathway to high-throughput roll-to-roll (R2R) manufacturing, achieving long-term operational stability on these platforms remains a critical challenge. Here, we report a robust strategy for the scalable manufacturing of efficient inverted formamidinium-cesium (FA-Cs) flexible perovskite solar modules (FPSMs), complemented by an industrially compatible encapsulation sequence. By employing a modified gas-assisted blade-coating process, combined with a gradual annealing protocol utilizing the thermal mass of a polydimethylsiloxane (PDMS)-covered copper plate (≤100 °C), we demonstrate a unified processing framework applicable to FA-Cs perovskite compositions of different bandgaps. In the device architecture, we employed a synergistic interface engineering approach utilizing the self-assembling molecule 4-((5H-diindolo[3,2-a:3′,2′-c]carbazole-5,10,15-triyl)tris(butane-4,1-diyl))tris(phosphonic acid) (TRIPOD-C4) as an effective hole-transporting layer, combined with propane-1,3-diammonium iodide (PDAI₂) for perovskite surface passivation. As a result, blade-coated FA-Cs PSCs with bandgaps of 1.61 eV and 1.76 eV achieved power conversion efficiencies (PCEs) of 20.4% and 17.4%, respectively. When scaled to modules, we recorded champion efficiencies of 17.0% and 15.1% for the medium- and wide-bandgap absorbers, respectively. Furthermore, we demonstrate a production-compatible vacuum lamination protocol utilizing ultra-high barrier foil. The encapsulated FPSMs exhibit promising durability, retaining 80% of their initial efficiency after 1026 hours of damp-heat aging (ISOS-D3, 85 °C/85% RH).

---

Broader context

The global transition to renewable energy demands photovoltaic technologies that go beyond rooftop and ground-mounted installations. Lightweight, flexible solar modules can be integrated onto building facades, vehicles, portable devices, and other curved or weight-sensitive surfaces – applications where conventional rigid silicon panels are impractical. Perovskite semiconductors, which can be deposited from solution at low temperatures, are uniquely suited for manufacturing such flexible devices using high-throughput roll-to-roll processes on inexpensive plastic foil. However, moving from efficient small-area laboratory cells to large-area modules that remain stable under real-world heat and humidity has proven difficult, and reports demonstrating the full fabrication-to-encapsulation chain on flexible substrates remain scarce. This work addresses that gap by presenting a complete, industrially compatible process for blade-coated perovskite solar modules on plastic foil, achieving module efficiencies above 15% for two different absorber bandgaps and demonstrating that encapsulated devices retain 80% of their initial performance after over 1000 hours of harsh damp-heat testing. By bridging laboratory-scale results with production-relevant fabrication and packaging methods, this study provides a practical blueprint for manufacturing durable, lightweight perovskite photovoltaics at scale.

Introduction

Metal halide perovskite solar cells (PSCs) have rapidly ascended as a frontrunner in the next generation of photovoltaics, demonstrating the fastest increase in certified power conversion efficiency (PCE) among all PV technologies.^1–3 PSCs are characterized by a unique combination of exceptional optoelectronic properties of the perovskite absorbing material and commercially attractive solution processability, with low-temperature post-processing.^4–7 This enables the utilization of lightweight polymeric foil, such as polyethylene terephthalate (PET) or polyethylene naphthalate (PEN), which in turn unlocks high-throughput roll-to-roll (R2R) manufacturing.⁸ The vast majority of reported data on perovskite PV devices is centered on those fabricated via spin-coating, a method not transferable to industrial production.^9,10 To enable large-scale manufacturing, it is essential to adopt scalable deposition techniques (such as slot-die coating, blade coating, spray coating, or inkjet printing) that minimize material waste and provide uniform coatings over larger areas.^10,11 These deposition techniques are well-suited for integration into a continuous R2R production process on flexible substrates.^12–14 Besides manufacturing innovations and implied cost efficiency, flexible PV devices offer a high power-to-weight ratio and large design versatility, which provide additional value propositions for numerous commercial use cases.^15–17 Mechanical conformability enables lightweight integration onto curved surfaces, textiles, and portable or airborne platforms.^18–20 While certified efficiencies for small-area cells now rival those of crystalline silicon, translating these performance metrics to large-area modules, particularly on flexible substrates, presents distinct engineering challenges.^10,21 The difficulty arises from the complex interplay of reduced control over nucleation and crystallization dynamics and a prolonged wet-film phase while processing large areas, especially when compared to spin-coating small-area samples.^22–24 This challenge is further exacerbated for perovskite compositions, such as the double-cation formamidinium-cesium (FA-Cs) systems, and those with high bromine content (typically used for wider bandgap perovskites), where the differences in solubilities and ionic radius of the precursor components result in rapid and inhomogeneous crystallization.^25–30 Furthermore, fabricating highly efficient devices on flexible substrates remains more demanding than on rigid counterparts.^8,13 This is related to the properties of the substrate itself, but also to the substrate handling. The most commonly used and most economical PET foil impose inherent thermal constraints on the device fabrication process, owing to their thermal deformations (thermal expansion, release of frozen-in strains, etc.).^8,31 Furthermore, the mismatch in thermal expansion coefficients between the polymeric substrate and deposited layers can generate substantial internal mechanical stresses within the device stack at elevated temperatures.^8,32 This mismatch must be carefully considered during both device fabrication and subsequent lamination. Most reported encapsulation protocols are optimized for rigid devices with glass–glass lamination stacks, and are not suitable for flexible architectures.^33–37 To maintain the key qualities of flexible devices, lamination protocols employing ultra-low-permeability barrier foil and mechanically compliant adhesives are essential.³⁸ Moreover, minimizing strain accumulation and reducing the temperature during the lamination process are critical. In this context, damp heat (DH; 85 °C/85% RH) testing serves as a rigorous, industry-standard accelerated-aging protocol.³⁹ It combines heat and moisture to activate the main extrinsic degradation pathways in perovskite PV devices. During DH stress, water vapor ingress and high temperature accelerate perovskite decomposition and lower the activation barriers for ion migration, thereby destabilizing interfaces and the bulk material.^40,41 Performance failure is often caused by the edge ingress: moisture enters through imperfect seals and spreads inward.^42–44 Flexible modules face additional challenges, as differences in thermal expansion can cause delamination and microcracks, creating new moisture entry points.^8,45 Bending and residual thermo-mechanical stresses further facilitate defect formation and interfacial separation, linking mechanical damage to faster moisture-driven chemical and ionic degradation.^35,45,46

In recent years, PSCs with an inverted (p–i–n) architecture have attracted significant research interest, driven by major advancements in device stack engineering and corresponding improvements in performance parameters, as well as their inherent compatibility with high-efficiency tandem configurations.^47,48 Nevertheless, there remains a notable knowledge gap regarding the scalable fabrication of flexible PV modules – a variant of perovskite technology with substantial commercial potential.^49–53 Mature photovoltaic technologies typically follow an inverse scaling law, with efficiency dropping by less than 1% with an increase in device area by an order of magnitude.⁵⁴ Although substantial progress has been made in narrowing the efficiency gap for flexible PSCs, achieving devices that are both sufficiently efficient and stable at scale remains a key challenge.

Herein, we report the scalable fabrication of efficient and stable inverted FA-Cs PSCs and PSMs. This work details the entire process, including solution blade-coating on large-area polymeric foil and subsequent device lamination with barrier foil, compatible with production-scale processes. Solution deposition was conducted directly on unsupported (free-standing) foil, with a maximum post-deposition thermal annealing temperature of 100 °C. Building on our previous work, we utilized the self-assembling molecule (SAM) 4-((5H-diindolo[3,2-a:3′,2′-c]carbazole-5,10,15-triyl)tris(butane-4,1-diyl))tris(phosphonic acid) (TRIPOD-C4) as a hole-transporting layer, which significantly improved the devices' fill factor.⁵⁵ We further incorporated propane-1,3-diammonium iodide (PDAI₂) to passivate the top surface of the perovskite layer. This approach enabled the fabrication of blade-coated FA-Cs PSCs with bandgaps of 1.61 eV and 1.76 eV, exhibiting PCEs of 20.4% and 17.4%, respectively. The optimized layer stack and processing sequence were successfully transferred to large-area modules, achieving champion efficiencies of 17.0% (E_g = 1.61 eV) and 15.1% (E_g = 1.76 eV). Furthermore, the laminated modules demonstrated robust reliability against moisture and elevated temperature, retaining up to 80% of their initial performance after 1026 hours of rigorous damp heat (ISOS-D3) testing (85 °C, 85% relative humidity). In this study, we combine blade-coating on free-standing PET foil, application of the same processing parameters to two perovskite compositions, vacuum lamination, and degradation analysis under damp-heat conditions within a single workflow.

Results and discussion

Solar cell blade-coating

We employed the inverted device architecture of the following structure: polyethylene terephthalate coated with indium tin oxide (PET-ITO)/nickel oxide (NiO_X)/self-assembling molecule (SAM)/perovskite/fullerene (C₆₀)/bathocuproine (BCP)/silver (Ag) (Fig. 1a). The NiO_X and SAM layers were sequentially blade-coated on 18.5 × 29.0 cm² substrates, which were subsequently sectioned into six equal pieces, each approximately 90 cm², for deposition of the perovskite layer using gas-assisted blade coating.

Fig. 1 (a) Scheme of the inverted WBG PSC layer stack, (b) scheme of the modified blade with a shielding element, (c) scheme of the gradual annealing step using a PDMS-covered copper plate, (d) picture of 90 cm²-sized blade-coated samples up to perovskite from which individual solar cells were extracted, (e) structure of TRIPOD-C4, and (f and g) statistics of the fill factor and efficiency for blade-coated WBG PSCs employing TRIPOD-C4 and Me-4PACz as a SAM.

Initially, we focused on blade-coating the wide-bandgap (WBG) perovskite. The perovskite composition, precursor solvent system, and additives were kept identical to those previously reported for WBG PSCs fabricated via spin-coating with gas-quenching, ensuring direct comparability between the two deposition methods and highlighting the importance of developing scalable fabrication techniques by mimicking large-area deposition conditions during small-area device fabrication.⁵⁵ The WBG perovskite of nominal composition Cs_0.2FA_0.8Pb(I_0.65Br_0.35)₃ displayed an optical bandgap of 1.76 eV (Fig. S1, SI). To ensure uniform hotplate annealing of the perovskite, the heating surface was covered with polydimethylsiloxane (PDMS), following the approach reported by Oh Yeong Gong et al.⁵⁶ This enabled sample planarization and prevented air gaps beneath the substrate, while allowing non-destructive transfer without inducing mechanical strain. To obtain compact and uniform films with low surface roughness, we systematically optimized the blade-coating speed and coating gap to control wet-film thickness. Two modifications to the conventional blade-coating process were introduced to improve control over the formation and conversion of the intermediate phase. First, a shielding element was mounted perpendicular to the deposition surface directly behind the blade, with a narrow gap maintained between its lower edge and the substrate (Fig. 1b). This element acts as a physical barrier between the liquid meniscus and the air-knife, decoupling the initial film deposition from the gas-quenching step and creating a more controlled evaporation environment around the deposition area. Second, a gradual annealing protocol was implemented in which the sample, planarized on a PDMS-covered copper plate pre-cooled to approximately 20 °C, was placed directly on a hot plate set to 100 °C (Fig. 1c). The thermal mass of the copper plate causes the sample temperature to increase gradually from room temperature to the target annealing temperature, effectively extending the low-temperature conversion window. This approach is particularly well-suited to flexible PET substrates: conventional gradient annealing typically requires transferring the sample between two hotplates held at different temperatures, which, for flexible foil, risks mechanical damage at each transfer step. Here, the sample remains continuously adhered to the heating surface throughout the annealing process thanks to the conformal adhesion of the PDMS layer, providing planarization and eliminating repeated attachment and detachment steps, making the protocol gentler and more practical for large-area flexible devices.

The impact of the two processing modifications on film quality and device performance was assessed. Introduction of the shielding element improved lateral film uniformity, as evidenced by spectroscopic ellipsometry thickness maps recorded over the central area of the substrate, which showed a reduced spread of thickness deviation from the mean value (Fig. S2, SI). Concurrently, XRD patterns of the annealed films showed reduced reflex intensities at 8.12° and 11.4° relative to the unshielded reference (Fig. S3, SI). We assign the feature at 8.12° to the solvent-coordinated intermediate phase (likely a solvated lead halide framework), consistent with reflections reported in the 8–9° 2θ range for both DMF and NMP-based perovskite precursor systems, and the feature at 11.4° to the δ-phase of the FA-Cs composition.^57–60 The persistence of the solvent-coordinated intermediate in the annealed film is consistent with the reported thermal stability of lead halide solvent adducts formed in DMF/NMP-based precursor systems, whose complete decomposition has been shown to require temperatures exceeding 100 °C.^59,60 The blade modification prevents non-uniform evaporation induced by turbulent gas flow from the air-knife. The concurrent decrease in both the 8.12° and 11.4° peaks suggests that the reduced volatilization rate affects the formation of the intermediate phase, supporting a transition to the desired α-phase rather than being trapped as a residual solvate or contributing to the formation of the δ-phase. However, the precise mechanism introduced by the shielding element requires further investigation. Additionally, a gradual annealing step was introduced to ensure controlled solvent removal and uniform film formation. Both the 8.12° and 11.4° reflexes were absent in the gradually annealed films (Fig. S4a, SI), indicating that the gradual temperature increase effectively extends the low-temperature crystallization window, providing a wider temporal range for the system to navigate the complex solvate-to-perovskite transition. While the suppression of the 8.12° reflex indicates a more complete conversion of the intermediate solvate, the concurrent disappearance of the 11.4° feature suggests that the gradual temperature increase reduces the δ-phase formation. We assume that this behaviour aligns with a kinetic regime where a slower, more controlled solvent removal favours direct crystallization into the α-phase. This is consistent with previous reports on intermediate-phase control in DMF/NMP-based precursor systems, where stabilising the solvent-coordinated intermediate phase is essential to suppress the formation of non-perovskite phases during crystallization.^59,61 SEM imaging revealed larger perovskite domains in the gradually annealed films (Fig. S4b, SI), consistent with a lower nucleation density when intermediate conversion occurs, as previously reported for multi-step thermal annealing approaches.^62,63 Photoluminescence (PL) mapping across different substrate positions showed improved spatial homogeneity of emission intensity, while single-spot PL spectra exhibited reduced spectral inhomogeneity and increased intensity (Fig. S5, SI). These improvements in film quality translated directly into enhanced device performance, with J–V characteristics showing increased open-circuit voltage (V_OC) and fill factor (FF) together with a reduced spread of the results, confirming that the gains in phase purity and spatial homogeneity are reflected at the device level (Fig. S6, SI). We obtained uniform NiO_X and WBG perovskite layers with average thicknesses of approximately 14 nm and 490 nm, respectively, as derived from ellipsometry mapping (Fig. S7, SI). Full processing details are provided in the Experimental section.

To evaluate the impact of the hole-selective interlayer, identical devices were fabricated employing two different SAMs: TRIPOD-C4 and (4-(3,6-dimethyl-9H-carbazol-9-yl)butyl)phosphonic acid (Me-4PACz), both prepared at equivalent molar concentrations. TRIPOD-C4 is a carbazole-based molecule with three phosphonic acid anchoring sites (Fig. 1e), which promotes favourable planarization of the conjugated aromatic system – a molecular design that is particularly advantageous for solution-based deposition over large areas.^55,64 We previously reported that spin-coated WBG PSCs utilizing TRIPOD-C4 exhibit improved efficiency and enhanced phase stability compared to those using Me-4PACz. These improvements were attributed to a more uniform coverage of NiO_X, improved hole extraction, and reduced non-radiative interfacial recombination.⁵⁵ Statistics of photovoltaic parameters derived from current density–voltage (J–V) characterization of WBG PSCs with distinct SAMs are summarized in Table 1. The devices utilizing TRIPOD-C4 demonstrated superior performance, primarily originating from a substantial increase in the fill factor (FF), which averaged 75.6% compared to 58.0% for Me-4PACz-based devices (Fig. 1f and g). The best-performing TRIPOD-C4 device delivered a PCE of 15.7%. This result extends the applicability of TRIPOD-C4 from spin-coated devices, as previously reported, to a blade-coated configuration, confirming that the multi-anchoring molecular design remains effective under scalable deposition conditions.

Table 1 Summary of photovoltaic parameters extracted from the J–V measurements of blade-coated WBG PSCs with different SAMs

SAM	PCE (%)	FF (%)	VOC (V)	JSC (mA cm^−2)
TRIPOD-C4	15.7 (14.8 ± 0.8)	77.7 (75.6 ± 1.5)	1.14 (1.10 ± 0.02)	18.7 (17.8 ± 0.8)
Me-4PACz	12.6 (11.2 ± 0.9)	61.6 (58.0 ± 2.5)	1.12 (1.08 ± 0.02)	18.4 (17.7 ± 0.6)

---

Additionally, we subjected WBG PSCs to a bending durability test, where devices retained over 90% of their initial performance after 10 000 bending cycles at a radius of 10 mm (Fig. S8, SI).

Next, to assess the universality of the developed approach, we reduced the bromide content in the perovskite composition from 35% to 10%, while maintaining the same precursor molarity, solvent system, and additives. As a result, a medium-bandgap (MBG) perovskite with an optical bandgap of 1.61 eV was obtained (Fig. S9, SI). The selection of these two compositions was motivated by their complementary relevance to the broader photovoltaic landscape: the WBG absorber (E_g ∼1.76 eV) is a key component in perovskite-based tandem architectures, where the top sub-cell harvests high-energy photons while matched with the narrow bandgap absorber, whereas the MBG composition (E_g = 1.61 eV) is well-suited for single-junction applications and represents a compositional regime with markedly more controlled crystallization dynamics. A central motivation of this work was to demonstrate that a single optimized processing framework, encompassing the same solvent system, additives, layer stack, and blade-coating parameters, can yield high-quality films and efficient devices across both compositions, despite the substantially different crystallization kinetics imposed by the bromide content.

The X-ray diffractograms of WBG and MBG perovskite films revealed intense reflexes at around 14.4° and 28.9° (2θ), corresponding to the pseudocubic (100) and (200) crystallographic planes, respectively (Fig. S10, SI). The WBG perovskite exhibited diffraction peaks shifted toward higher 2θ values relative to the MBG composition, consistent with lattice contraction induced by the increased bromide content. This shift arises from the smaller ionic radius of Br⁻ (1.96 Å) compared to I⁻ (2.20 Å), which results in contracted unit cell parameters and higher-angle Bragg reflections.⁶⁵ The diffractogram of the WBG perovskite composition additionally revealed a negligible peak at 12.8°, which corresponds to the (001) plane of residual hexagonal lead iodide (PbI₂).⁶⁶ Consistent with this observation, top-view SEM imaging revealed trace amounts of residual PbI₂, visible as bright intrusions protruding from the film surface (Fig. S4b, SI). In contrast, no PbI₂ reflection was detected in the diffractogram of the MBG perovskite film, reflecting the compositional dependence of the crystallization dynamics between the two perovskite compositions.^67,68 To further evaluate the optoelectronic quality and phase homogeneity of the two absorbers, steady-state photoluminescence (ss-PL) spectra were collected from the perovskite films prior to device completion (Fig. S11, SI). Both compositions exhibit single, well-defined emission peaks with no secondary features or shoulders indicative of halide phase segregation, confirming the phase purity of both blade-coated films. Nevertheless, the MBG perovskite film exhibits approximately one order of magnitude higher PL intensity than the WBG perovskite, indicating higher optoelectronic quality, consistent with the greater susceptibility of bromide-rich compositions to trap-state formation and compositional heterogeneity during crystallization.^67,68 Taken together, the XRD and PL data confirm that, although both absorbers are phase-pure, the WBG composition exhibits inherently lower phase homogeneity than the MBG composition.

Next, we fabricated MBG and WBG PSCs by applying the same optimized processing conditions and TRIPOD-C4 as a SAM. To further improve device performance, the perovskite top surface was passivated with propane-1,3-diammonium iodide (PDAI₂). PDAI₂ has been reported to form a surface dipole that increases the n-type character of the perovskite surface, thereby repelling holes near the electron-transporting layer (ETL) and suppressing interfacial recombination.^69–72 The PDAI₂ treatment increased the ss-PL intensity and elevated the V_OC of both WBG and MBG devices (Fig. S12 and S13, SI), suggesting suppressed non-radiative recombination at the perovskite/ETL interface. The schematic of the optimized PSC layer stack is shown in Fig. 2a. The blade-coated PSCs yielded average PCEs of 19.5% and 16.0% for the MBG and WBG perovskite, respectively (Fig. 2b). The champion MBG PSC delivered a PCE of 20.4% (FF = 76.6%, V_OC = 1.17 V, J_SC = 22.5 mA cm⁻²), while the best-performing WBG PSC yielded 17.4% (FF = 80.1%, V_OC = 1.22 V, J_SC = 17.7 mA cm⁻²), both with negligible hysteresis (Fig. 2c). The statistics of all photovoltaic parameters are listed in Table 2.

Fig. 2 (a) Scheme of the optimized inverted PSC stack, (b) statistics of efficiency for flexible, blade-coated solar cells employing WBG and MBG perovskite absorbers, (c) J–V curves (forward and reverse scan directions) of top-performing solar cells with the WBG and MBG absorbers, (d) scheme of the optimized inverted PSM stack, (e) statistics of efficiency for flexible, blade-coated solar modules employing WBG and MBG perovskite absorbers, and (f) J–V curves (forward and reverse scan directions) of the top-performing solar module with the WBG and MBG perovskite absorbers.

Table 2 Summary of photovoltaic parameters extracted from the J–V measurements of blade-coated solar cells and modules with WBG and MBG perovskite absorbers

Device type	PCE (%)	FF (%)	VOC (V)	JSC (mA cm^−2)
WBG PSCs	17.4 (16.0 ± 0.4)	80.1 (75.2 ± 2.0)	1.22 (1.19 ± 0.02)	17.7 (17.7 ± 0.3)
WBG PSMs	15.1 (13.7 ± 1.0)	73.8 (68.0 ± 5.3)	9.78 (9.65 ± 0.09)	17.4 (16.8 ± 0.6)
MBG PSCs	20.4 (19.5 ± 0.4)	76.6 (75.5 ± 2.0)	1.17 (1.14 ± 0.02)	22.5 (22.4 ± 0.2)
MBG PSMs	17.0 (15.9 ± 1.2)	72.5 (69.2 ± 4.9)	8.98 (8.87 ± 0.10)	21.5 (20.7 ± 0.5)

---

Fig. S14 in the SI presents an overview of the most recent reported PSCs fabricated using scalable techniques on both rigid and flexible substrates, categorized based on the perovskite bandgap. The data clearly demonstrate that devices processed on flexible substrates exhibit lower performance and represent a minority of the reported work. Similarly, the fabrication of flexible modules using scalable methods remains limited. This scarcity is particularly evident for wide-bandgap compositions (E_g ≥ 1.7 eV) on flexible substrates, where few reports employing scalable deposition methods are available. The present work contributes to this underexplored area by demonstrating efficient blade-coated devices at both 1.61 eV and 1.76 eV using a unified processing framework.

Solar module blade-coating

We leveraged our optimized architecture and blade-coating process to fabricate perovskite solar modules (PSMs). For the module stack, BCP was replaced with tin oxide deposited by atomic layer deposition (ALD-SnO_X) to enhance device stability. At elevated temperatures, BCP is prone to morphological degradation, leading to interface deterioration and performance loss.^73,74 In contrast, a dense, conformal ALD-grown inorganic layer provides an effective diffusion barrier, retarding thermally and humidity-driven degradation pathways.^74,75 Additionally, an ITO layer was introduced between ALD-SnO_X and the Ag electrode to prevent silver ion migration and silver electrode degradation caused by halide ions from the perovskite layer, which often leads to device failure.⁷⁶ The schematic of the module layer stack is shown in Fig. 2d.

We fabricated 8-cell modules with an active area of 15.2 cm². The WBG PSMs yielded an average PCE of 13.7%, whereas the MBG PSMs delivered 15.9% (Fig. 2e). The best-performing WBG module achieved a PCE of 15.1% (FF = 73.9%, V_OC = 9.77 V, J_SC = 16.9 mA cm⁻²) and the champion MBG module yielded 17.0% (FF = 73.1%, V_OC = 8.84 V, J_SC = 21.1 mA cm⁻²), both with a negligible hysteresis (Fig. 2f). The statistics of all photovoltaic parameters extracted from the modules' J–V measurements are listed in Table 2. Table S1, SI summarizes the state-of-the-art inverted PSCs and PSMs fabricated using scalable deposition methods on polymeric foil. The data reveal a cell-to-module PCE loss, which could be attributed to increased area and/or interconnection losses. Notably, the results achieved in this work represent some of the highest reported PCE values, with the lowest cell-to-module loss. The WBG module is of particular interest for use as a sub-cell in flexible all-perovskite tandem architectures. Notably, reports on scalable fabrication of inverted flexible modules employing WBG FA-Cs perovskite absorbers (E_g ≥ 1.7 eV) remain limited, highlighting the relevance of these results for the development of future flexible tandem devices.

Solar module lamination and stability testing

Next, the MBG perovskite modules were vacuum-laminated with foil having ultra-high barrier properties to preserve their inherent light weight and facilitate large-scale processing. Conventional silicon PV (Si-PV) encapsulation protocols are not directly transferable to perovskite PV devices due to their higher thermochemical sensitivity. Encouraging damp-heat stability has recently been demonstrated for flexible perovskite solar modules encapsulated with high-barrier foil.^77–79 However, the reported encapsulation architectures frequently rely on wide lateral margins rather than a dedicated edge sealant, thereby constraining the achievable aperture-to-total-area ratio and limiting compatibility with high-throughput manufacturing. The incorporation of a dedicated edge sealant alongside vacuum lamination, a process well-established in conventional silicon PV manufacturing, not only improves the aperture-to-total-area ratio but also mitigates the risk of residual gas entrapment. Nevertheless, reports on combining industrially scalable vacuum lamination, a dedicated edge sealant, and ultra-barrier foil for flexible perovskite modules remain exceptionally scarce in the literature. The encapsulation sequence reported here is therefore of practical significance, as it demonstrates the feasibility of adapting the established silicon PV lamination method to flexible perovskite modules while maintaining the mechanical compliance and low thermal budget required by polymeric substrates.

Here, the MBG PSMs were encapsulated with a commercial barrier foil, using a thermoplastic polyolefin (TPO) encapsulant adhesive. Both the barrier foil and the encapsulant film had thicknesses on the order of 200 µm and were applied symmetrically on both sides of the device. The edges were sealed with a polymer-based polyisobutylene edge-sealant. The use of a dedicated perimeter seal, rather than wide lateral encapsulant margins, supports a more favourable aperture-to-total-area ratio of the laminated module. Electrical interconnection to the external circuit was realized via a current-collector tape. The schematic of the employed encapsulation layer stack is shown in Fig. 3a.

Fig. 3 (a) Schematic of the encapsulation layer stack, (b) picture of the encapsulated PSM, (c) normalized efficiency evolution upon the DH stability test, (d) normalized series resistance evolution upon the DH stability test, and (e) normalized shunt resistance evolution upon the DH stability test. On the graphs, the solid symbol represents the data for the most stable module, and open symbols represent the average and associated standard deviations across all five tested modules.

Si-PV modules are typically vacuum-laminated at 135–150 °C under high vacuum, conditions that are not compatible with PSMs fabricated on PET foil.⁷⁵ Here, the vacuum lamination was performed at 120 °C with the pressure step omitted. To ensure reliable sealing between the edge-seal material and the barrier foil, the lamination was followed by a supplementary heated pressing step applied along the perimeter of the lamination stack. A picture of the encapsulated MBG PSM is shown in Fig. 3b. The PV parameters extracted from I–V curves of laminated modules are provided in Table S2, SI. The best-performing laminated MBG PSM delivered a PCE of 14.8% from the J–V scan (14.6% from maximum power point tracking (MPPT)). The WBG PSMs were not encapsulated in this study, as their target application is integration as a sub-cell in an all-perovskite tandem architecture. In such a configuration, the layer stack introduces additional thermo-mechanical constraints and distinct moisture ingress pathways that differ substantially from those of a single-junction package.^80,81 The encapsulated modules were subjected to an accelerated damp-heat stability test (ISOS-D3) at 85 °C and 85% relative humidity. The most resilient module maintained 80% of its initial performance over a 1026 hour test duration (Fig. 3c). Performance declined rapidly within the initial 66 hours, driven by a simultaneous increase in R_S and decrease in R_SH, as shown in Fig. 3d and e. Interestingly, after the initial 66 hour period, performance partially recovered, accompanied by a reduction in R_S. In contrast, R_SH showed a consistent, gradual decline throughout the entire test. The four remaining modules followed similar R_S and R_SH trends; however, a more pronounced increase in R_S and a steeper drop in R_SH led to their comparatively lower final stability. The PCE recovery after the initial dip at 66 h suggests that the early performance loss is unlikely to originate from irreversible chemical degradation of the perovskite material. Instead, it is consistent with thermo-mechanical stress arising from a mismatch in thermal expansion coefficients or the release of frozen strain in plastic foil within the encapsulation stack.

To complement the charge extraction and optoelectronic quality of the perovskite absorber, spatially resolved characterization, encompassing PL mapping under open and short-circuit conditions (PL_OC and PL_SC) was carried out (Fig. S15 and S16, SI). To investigate the spatial evolution of carrier extraction during the DH test, we calculated the extraction coefficient (EC) maps from PL_OC and PL_SC images, defined as EC = ((PL_OC − PL_SC)/PL_OC) × 100% (Fig. S17, SI). Horizontal line profiles were extracted at selected positions to evaluate the EC distribution across the cell stripes (Fig. 4). At 0 h, the EC profiles exhibited a flat, square shape within each stripe (∼80–85%), with sharp transitions at the stripe boundaries, and full spatial homogeneity, as confirmed by overlapping profiles at different y-positions. After 66 h, the EC dropped uniformly across the full stripe width to approximately 65–73%, consistent with the concurrent R_S increase derived from I–V measurements and indicating a homogeneous degradation of carrier extraction across the active area of the module. At 180 h, as R_S partially recovered, the EC level also increased; however, the stripe edges recovered more rapidly than the centres, resulting in a U-shaped profile. From 506 h onward, the EC profile inverted: the edges exhibited lower EC than the centers, yielding a dome-shaped distribution that was maintained through 1026 h, with central EC values of approximately 80–86%. This inversion occurred during the period when the monotonic R_SH decline became the dominant driver of PCE loss, suggesting a distinct degradation mechanism. In parallel, the absolute PL_OC signal decreased progressively over the test duration (Fig. S15). At the end of the aging test the EC remained relatively preserved but with reduced homogeneity. The PL_OC reduction reflects degradation of the optoelectronic quality of the perovskite absorber, consistent with the formation of PbI₂, as confirmed by XRD of the aged module (Fig. S18, SI).

Fig. 4 Spatially resolved extraction coefficient analysis of the most stable module at all measured time points during the DH test. (a) Illustration of the line profile extraction methodology: map of extraction coefficient of the module with the indicated positions of three horizontal line scans (z@y1, z@y2, and z@y3) and (b–h) extraction coefficient profiles along the x-direction recorded at three y-positions at (b) 0 h, (c) 66 h, (d) 180 h, (e) 343 h, (f) 506 h, (g) 718 h, and (h) 1026 h of the DH test.

Taken together, the EC analysis reveals three distinct stages: (i) a uniform EC reduction at 66 h, concurrent with the R_S increase, suggesting a homogeneous process across each stripe; (ii) a spatially non-uniform recovery (180–343 h), with faster extraction recovery at stripe edges, indicating position-dependent kinetics; and (iii) a progressive EC profile inversion from 506 h onward, correlating with dominant R_SH decline and absorber degradation. We attribute the initial uniform EC drop to thermo-mechanical stress originating from the encapsulation stack, causing partial loss of interfacial contact, an effect absent in unencapsulated modules aged at the same temperature in an inert atmosphere (Fig. S19, SI). The observed performance degradation mode highlights the dual role of the encapsulation stack, which must simultaneously provide an effective moisture barrier without inducing deleterious mechanical stress within the device.

Conclusions

In this work, we have demonstrated the fabrication and lamination of inverted perovskite PV devices using scalable techniques and a methodology suitable for polymeric substrates. Interface engineering, employing TRIPOD-C4 as a SAM and PDAI₂ to passivate the perovskite/ETL interface, enabled the fabrication of blade-coated FA-Cs PSCs with bandgaps of 1.61 eV and 1.76 eV to reach PCEs of 20.4% and 17.4%, respectively. The optimized layer stack and fabrication process were successfully transferred to large-area module manufacturing, achieving champion efficiencies of 17.0% (E_g = 1.61 eV) and 15.1% (E_g = 1.76 eV). Furthermore, the laminated modules exhibited robust environmental stability, retaining up to 80% of their initial performance after 1026 hours of damp heat testing (ISOS-D3, 85 °C, 85% RH). This stability result, achieved with a production-compatible vacuum lamination protocol on a flexible substrate, represents an encouraging step towards meeting the durability requirements of commercially viable flexible perovskite PVs. We have demonstrated that a single optimized blade-coating framework, incorporating a shielding element for controlled intermediate-phase formation and a gradual annealing protocol compatible with thermally sensitive PET substrates, can yield efficient devices and modules across FA-Cs perovskite compositions with bandgaps of 1.61 eV and 1.76 eV, despite the markedly different crystallization dynamics imposed by the bromide content. The synergistic interface engineering approach, employing TRIPOD-C4 and PDAI₂, proved effective for efficiency improvement under scalable deposition conditions. Taken together, this work provides an integrated demonstration of the scalable fabrication and environmental validation of flexible FA-Cs perovskite solar modules. The methodology presented in this work, from blade-coating on free-standing PET foil, through vacuum lamination, to ISOS-D3 stability testing, establishes a scalable processing route compatible with large-scale manufacturing. We anticipate that the insights into the degradation modes of flexible modules under damp heat conditions will facilitate the optimisation of encapsulation architectures, thereby advancing the long-term reliability of flexible perovskite photovoltaic devices toward commercial deployment.

Experimental section

Materials

Unless otherwise stated, all the materials were purchased and used as received. Indium tin oxide (ITO)-coated polyethylene terephthalate (PET) substrates (sheet resistance of 60 Ω/□) were bought from Eastman Chemical Company. Lead iodide (PbI₂, 99.99%), cesium bromide (CsBr, 99.9%), and [4-(3,6-dimethyl-9H-carbazol-9-yl)butyl]phosphonic acid (Me-4PACz, >99.0%) were purchased from Tokyo Chemical Industry Co. Ltd. Lead bromide (PbBr₂, 98%), cesium iodide (CsI, 99.9%), fullerene-C₆₀ (C₆₀, 99.9%), bathocuproine (BCP, 99.99%), and lead thiocyanate, (Pb(SCN)₂, 99.5%) were purchased from Sigma-Aldrich. Formamidinium iodide (FAI, 99%) was obtained from Ajay North America, and formamidinium bromide (FABr, 99.99%) from Dyenamo. 4-Fluorophenethylammonium iodide (4FPEAI) and propane-1,3-diammonium iodide (PDAI₂, >98.0%) were bought from GreatCell Solar. N-Metylopirolidon anhydrous (NMP, 99.5%), N,N-dimethylformamide anhydrous (DMF, 99.8%), dimethyl sulfoxide (DMSO, 99.9%), and 2-propanol (IPA, 99.9%) were purchased from Sigma-Aldrich. Ethanol (EtOH, 99.8%) was bought from ChemPur. Silver pellets (Ag, 99.99%) and the indium-tin oxide target (ITO, 99.99%) were purchased from Kurt. J. Lesker. TRIPOD-C4 was synthesized using a previously reported method.⁵⁵ NiO_X was synthesized internally, following the procedure described in the SI.

Perovskite precursor preparation

The WBG perovskite precursor solution (nominal perovskite composition: Cs_0.2FA_0.8Pb(I_0.65Br_0.35)₃) was prepared with a molarity of 1.15 M, following the recipe: 0.15 mmol of CsI, 0.08 mmol of CsBr, 0.6 mmol of FAI, 0.32 mmol of FABr, 0.4 mmol of PbBr_2, and 0.75 mmol of PbI₂, dissolved in 1 mL of DMF : NMP (9 : 1, v/v). The solution was continuously stirred overnight at around 20 °C. After overnight stirring, the additives were dissolved in the perovskite precursor solution. The molarities of the additives were 1.5 mol% with respect to perovskite (4FPEAI) and 2 mol% Pb(SCN)₂. The solution was left to dissolve for 1–2 hours without stirring. The precursor solution was filtered with a 0.22 µm polytetrafluoroethylene (PTFE) membrane before use. The MBG perovskite precursor solution (nominal perovskite composition: Cs_0.2FA_0.8Pb(I_0.90Br_0.10)₃) was prepared with a molarity of 1.15 M, following the same recipe and procedure as for the WBG perovskite precursor but with a reduced bromide fraction: 0.20 mmol of CsI, 0.02 mmol of CsBr, 0.8 mmol of FAI, 0.09 mmol of FABr, 0.12 mmol of PbBr_2, and 1.1 mmol of PbI₂.

Device fabrication

Perovskite solar cells and modules were fabricated in a p–i–n architecture. A blade-coater from Erichsen and an applicator from Zehntner were used for blade-coating. The substrates were planarized for deposition by applying IPA beneath the substrate to promote capillary-assisted seating on the plate. To fabricate the devices, a PET-ITO substrate measuring 18.5 × 29.0 cm was patterned with a near-infrared laser (1064 nm) to delineate the device design (cell or module). This large-area laser-patterned substrate was subsequently used to blade-coat NiO_X and SAM layers. Following the laser patterning, the substrate was cleaned using isopropanol, dried with nitrogen gas, and exposed to oxygen plasma for 2 minutes. The NiO_X powder was kept in a dry and inert atmosphere, and the Milli-Q water was kept refrigerated. The NiO_X dispersion was prepared by dissolving 20 mg of NiO_X powder in 1 mL of cold Milli-Q water, which was then sonicated for 5 minutes in a cold water bath. Finally, the dispersion was filtered through a 0.45 µm PTFE membrane prior to deposition. 350 µL of NiO_X dispersion was blade-coated in a clean-room environment with a blade gap set at 450 µm height and a 2.5 mm s⁻¹ speed, followed by annealing at 100 °C for 30 minutes in the oven. After NiO_X deposition, the substrate was transferred to a nitrogen-filled glovebox for SAM and perovskite layer deposition. The Me-4PACz solution was prepared inside a nitrogen-filled glovebox by dissolution of 1 mmol of Me-4PACz in 1 mL of ethanol. The TRIPOD-C4 solution was prepared inside a nitrogen-filled glovebox by dissolution of 1 mmol of powder in 1 mL of DMF. The TRIPOD-C4 solution was stirred overnight and filtered just before deposition with a polyvinylidene fluoride (PVDF) 0.45 µm filter. The SAM (TRIPOD-C4 or Me-4PACz) solution was blade-coated twice on the NiO_X layer in opposite directions. For each deposition, 800 µL of solution was applied using a blade with a 400 µm gap and a speed of 2.5 mm s⁻¹, followed by annealing at 100 °C for 10 minutes on a hotplate inside the glovebox. After SAM deposition, the 18.5 × 29.0 cm substrate was sectioned into six equal pieces, each about 90 cm², for deposition of the perovskite layer. The blade-coating procedure was the same for WBG and MBG perovskite absorbers. 120 µL of solution was applied using a blade gap set to 400 µm height and a speed of 3.5 mm s⁻¹, with nitrogen quenching. To promote solvent removal, gas quenching with nitrogen at 2.5 bar was applied perpendicular to the substrate using an air knife. The gas flow was terminated immediately after the pass. After a 20–30 s wait, the sample was placed on a PDMS-covered, room-temperature copper plate and transferred to a hotplate set at 100 °C for 30 min. After cooling down, the perovskite surface was passivated with PDAI₂. The PDAI₂ solution was prepared in a nitrogen-filled glovebox by dissolving 2 mg of powder in 1 mL of IPA. 100 µL of solution was applied using a blade gap set to 400 µm in height and a deposition speed of 10 mm s⁻¹, assisted by nitrogen at 1 bar. The sample was annealed at 100 °C for 10 minutes. Next, 30 nm of fullerene C₆₀ was deposited as an electron-transporting layer, using thermal evaporation at ∼10⁻⁶ mbar. In the case of solar cells, a 5 nm BCP buffer layer and 110 nm Ag electrode were deposited by thermal evaporation at ∼10⁻⁶ mbar. The metal electrode was deposited through a shadow mask, resulting in an active area of 0.1 cm². For the modules, deposition of C₆₀ was followed by a 10 nm-thick tin oxide layer via atomic layer deposition (ALD-SnO_X). The ALD-SnO_X process conditions follow a previously published procedure.⁸² Then, the laser ablation step was performed through the stack down to the TCO to create the series interconnection (P2). Finally, 40 nm ITO and 110 nm Ag were sputtered through a shadow mask (P3) to form the top electrode. The ITO sputtering details follow a previously published procedure.⁸² The module consisted of 8-cell modules connected in series, resulting in an active area of 15.2 cm² and a total area of 25 cm² (GFF 60.3%).

Module encapsulation

Vacuum lamination was performed using commercial barrier foil with WVTR <1 × 10⁻⁴ g m⁻² day⁻¹ (at 38 °C/90% RH) using a TPO-based lamination encapsulant. A barrier foil and an encapsulant film (each around 200 µm thick) were applied to both sides of the device stack. The lamination process was conducted under vacuum (<50 mbar) at 120 °C; the conventional pressure step applied after the vacuum step was omitted. The total residence time in the laminator was 480 s. The device perimeter was sealed with a polybutylene-based edge-seal tape (10 mm width, 0.6 mm thickness). To ensure proper sealing between the edge-seal material and the barrier foil, the lamination was followed by an additional heated pressing step applied to the package sides. External contacts were formed using a current-collector tape (4 mm width, 60 µm thickness).

Author contributions

A. W. M.: conceptualization, data curation, formal analysis, methodology, investigation, validation, visualization, writing – original draft, writing – review and editing; M. Ś.: data curation, formal analysis, methodology, validation, visualization, writing – review and editing; K. B.: data curation, formal analysis, methodology, validation, investigation, visualization, writing – review and editing; D. R. M.: methodology, data curation, validation, visualization, writing – review and editing; Ł. P.: methodology, writing – review and editing; K. D.: data curation, methodology, writing – review and editing; P. J.: methodology; K. R.: methodology; D. T.: data curation, formal analysis, methodology, writing – review and editing; A. D. C.: supervision, writing – review and editing; K. W.: supervision, project administration, funding acquisition, writing – review and editing.

Conflicts of interest

There are no conflicts to declare.

Data availability

The supporting data have been provided as part of the supplementary information (SI). Supplementary information: additional experimental methods, perovskite film characterisation, annealing- and passivation-study data, device statistics, a state-of-the-art performance comparison, and multi-modal analyses of the encapsulated modules during damp-heat testing. See DOI: https://doi.org/10.1039/d6el00050a.

Acknowledgements

This work was partly funded by the National Science Center as part of the “OPUS LAP” cooperation (PALACE project, grant no: 2021/43/I/ST5/03245), the National Center for Research and Development (grant no: LIDER13/0217/2022), and the European Union's Horizon Europe research and innovation program (Supertandem project, grant no. 101075605). A. D. C. and D. T. acknowledge the European Union's Horizon Europe research and innovation program under grant agreement (PEARL project, no. 101122283).

Notes and references

1. M. A. Green, A. Ho-Baillie and H. J. Snaith, Nat. Photonics, 2014, 8, 506–514.
2. L. Zhang, M. Zhang, H. Wang, Z. Li, Z. Zhang, Y. Song, X. Song, S. Wang, J. Chen, W. Li, H. Dong, F. Wang, Y. Liu, X. Li, J. Yuan, W. Ma, Y. Zhang, S. F. Liu, F. Gao, Z. Wu, Z. Yang, F. Hao, K. Zhao, Z. Jin, K. Yan, J. Chang, J. Cao, J. Wang, X. Zhang, Q. Dong, P. Gao, Y. Zhao, L. Xiao, Y. Ding and L. Ding, Adv. Mater., 2026, 38, e12221.
3. Best Research-Cell Efficiency Chart | Photovoltaic Research | NLR, https://www.nrel.gov/pv/cell-efficiency, accessed 24 January 2026.
4. L. Zhang, L. Mei, K. Wang, Y. Lv, S. Zhang, Y. Lian, X. Liu, Z. Ma, G. Xiao, Q. Liu, S. Zhai, S. Zhang, G. Liu, L. Yuan, B. Guo, Z. Chen, K. Wei, A. Liu, S. Yue, G. Niu, X. Pan, J. Sun, Y. Hua, W.-Q. Wu, D. Di, B. Zhao, J. Tian, Z. Wang, Y. Yang, L. Chu, M. Yuan, H. Zeng, H.-L. Yip, K. Yan, W. Xu, L. Zhu, W. Zhang, G. Xing, F. Gao and L. Ding, Nano-Micro Lett., 2023, 15, 177.
5. M. D. Bastiani, V. Larini, R. Montecucco and G. Grancini, Energy Environ. Sci., 2023, 16, 421–429.
6. K. J. Prince, H. M. Mirletz, E. A. Gaulding, L. M. Wheeler, R. A. Kerner, X. Zheng, L. T. Schelhas, P. Tracy, C. A. Wolden, J. J. Berry, S. Ovaitt, T. M. Barnes and J. M. Luther, Nat. Mater., 2025, 24, 22–33.
7. H. S. Jung and N.-G. Park, Small, 2015, 11, 10–25.
8. L. A. Castriotta, M. A. Uddin, H. Jiao and J. Huang, Adv. Mater., 2025, 37, 2408036.
9. B. Kang and F. Yan, Energy Environ. Sci., 2025, 18, 3917–3954.
10. A. Khorasani, F. Mohamadkhani, M. Marandi, H. Luo and M. Abdi-Jalebi, Adv. Energy Sustainability Res., 2024, 5, 2300275.
11. A. Agresti, F. Di Giacomo, S. Pescetelli and A. Di Carlo, Nano Energy, 2024, 122, 109317.
12. M. J. M. Marques, W. Lin, T. Taima, S. Umezu and Md. Shahiduzzaman, Mater. Today, 2024, 78, 112–141.
13. H.-C. Cha, S.-H. Huang, C.-F. Li, F.-Y. Tsai, W.-F. Su and Y.-C. Huang, Sustainable Energy Fuels, 2025, 9, 5962–6006.
14. H. Zhong, Y. Wang, L. Hou, X. Song and F. Zhang, ACS Appl. Mater. Interfaces, 2025, 17, 50168–50190.
15. V. Romano, A. Agresti, R. Verduci and G. D'Angelo, ACS Energy Lett., 2022, 7, 2490–2514.
16. A. Grabowska, R. Fuentes Pineda, P. Spinelli, G. Soto Pérez, F. A. Vinocour Pacheco and V. Babu, ACS Appl. Mater. Interfaces, 2024, 16, 48676–48684.
17. K. Wojciechowski, D. Forgács and T. Rivera, Sol. RRL, 2019, 3, 1900144.
18. B. Hailegnaw, S. Demchyshyn, C. Putz, L. E. Lehner, F. Mayr, D. Schiller, R. Pruckner, M. Cobet, D. Ziss, T. M. Krieger, A. Rastelli, N. S. Sariciftci, M. C. Scharber and M. Kaltenbrunner, Nat. Energy, 2024, 9, 677–690.
19. S. Aftab, G. Koyyada, Z. Ali, M. A. Assiri, J. H. Kim, N. Rubab and E. Akman, Mater. Today Energy, 2025, 51, 101872.
20. A. S. R. Bati, Y. L. Zhong, P. L. Burn, M. K. Nazeeruddin, P. E. Shaw and M. Batmunkh, Commun. Mater., 2023, 4, 2.
21. R. Yin, Y. Wu, Z. Huang, A. S. Vasenko, S. Xu and H. Zhou, Adv. Funct. Mater., 2025, 35, 2419184.
22. N.-G. Park and K. Zhu, Nat. Rev. Mater., 2020, 5, 333–350.
23. J. L. Starger, A. E. Louks, K. Schutt, E. A. Gaulding, R. W. Epps, R. C. Bramante, R. A. Kerner, K. Zhu, J. J. Berry, N. J. Alvarez, R. A. Cairncross and A. F. Palmstrom, Cell Rep. Phys. Sci., 2025, 6, 102655.
24. S. Ternes, F. Laufer and U. W. Paetzold, Adv. Sci., 2024, 11, 2308901.
25. Z. Jiang, F. Ren, Q. Zhou, B. Wang, Z. Sun, M. Li, Z. Zhao, Z. Liu and W. Chen, Adv. Energy Mater., 2025, 15, 2500598.
26. C. Dong, G. Chen, F. Yao, Z. Yu, K. Dong, S. Du, L. Huang, C. Wang, T. Wang, S. Wang, W. Ke and G. Fang, Adv. Opt. Mater., 2025, 13, 2401955.
27. J. Yang, D. K. LaFollette, B. J. Lawrie, A. V. Ievlev, Y. Liu, K. P. Kelley, S. V. Kalinin, J.-P. Correa-Baena and M. Ahmadi, Adv. Energy Mater., 2023, 13, 2202880.
28. X. Jia, K. Zhang, X. Gao, X. Liao, Y. Yang, W. Li, X. Lv, X. Zhao, J. Liu, Y. Ji, Z. Yan, Q. Du, F. Huang, Z. Ren, Y. Zhai, W. Huang, Y. Bai, C. Yao, Q. Lin, Y.-B. Cheng and J. Tong, Sci. Adv., 2025, 11, eadv4501.
29. Y. An, N. Zhang, Z. Zeng, Y. Cai, W. Jiang, F. Qi, L. Ke, F. R. Lin, S.-W. Tsang, T. Shi, A. K.-Y. Jen and H.-L. Yip, Adv. Mater., 2024, 36, 2306568.
30. R. Wang, X. Liu, S. Yan, N. Meng, X. Zhao, Y. Chen, H. Li, S. M. H. Qaid, S. Yang, M. Yuan and T. He, Nat. Commun., 2024, 15, 8899.
31. Q. Wali, I. E. Lee, M. Aamir, T. C. Chuah and R. Jose, Surf. Interfaces, 2025, 76, 107922.
32. H. Liang, W. Yang, J. Xia, H. Gu, X. Meng, G. Yang, Y. Fu, B. Wang, H. Cai, Y. Chen, S. Yang and C. Liang, Adv. Sci., 2023, 10, 2304733.
33. S. Ma, G. Yuan, Y. Zhang, N. Yang, Y. Li and Q. Chen, Energy Environ. Sci., 2022, 15, 13–55.
34. Q.-Q. Chu, Z. Sun, D. Wang, B. Cheng, H. Wang, C.-P. Wong and B. Fang, Matter, 2023, 6, 3838–3863.
35. P. Mariani, M. Á. Molina-García, J. Barichello, M. I. Zappia, E. Magliano, L. A. Castriotta, L. Gabatel, S. B. Thorat, A. E. Del Rio Castillo, F. Drago, E. Leonardi, S. Pescetelli, L. Vesce, F. Di Giacomo, F. Matteocci, A. Agresti, N. De Giorgi, S. Bellani, A. Di Carlo and F. Bonaccorso, Nat. Commun., 2024, 15, 4552.
36. Y. Wang, I. Ahmad, T. Leung, J. Lin, W. Chen, F. Liu, A. M. C. Ng, Y. Zhang and A. B. Djurišić, ACS Mater. Au, 2022, 2, 215–236.
37. T. Ahmad, S. Dasgupta, S. Almosni, A. Dudkowiak and K. Wojciechowski, Energy Environ. Mater., 2023, 6, e12434.
38. L. Mu, S. Wang, H. Liu, W. Li, L. Zhu, H. Wang and H. Chen, Adv. Funct. Mater., 2025, 35, 2415353.
39. M. V. Khenkin, E. A. Katz, A. Abate, G. Bardizza, J. J. Berry, C. Brabec, F. Brunetti, V. Bulović, Q. Burlingame, A. Di Carlo, R. Cheacharoen, Y.-B. Cheng, A. Colsmann, S. Cros, K. Domanski, M. Dusza, C. J. Fell, S. R. Forrest, Y. Galagan, D. Di Girolamo, M. Grätzel, A. Hagfeldt, E. von Hauff, H. Hoppe, J. Kettle, H. Köbler, M. S. Leite, S. F. Liu, Y.-L. Loo, J. M. Luther, C.-Q. Ma, M. Madsen, M. Manceau, M. Matheron, M. McGehee, R. Meitzner, M. K. Nazeeruddin, A. F. Nogueira, Ç. Odabaşı, A. Osherov, N.-G. Park, M. O. Reese, F. De Rossi, M. Saliba, U. S. Schubert, H. J. Snaith, S. D. Stranks, W. Tress, P. A. Troshin, V. Turkovic, S. Veenstra, I. Visoly-Fisher, A. Walsh, T. Watson, H. Xie, R. Yıldırım, S. M. Zakeeruddin, K. Zhu and M. Lira-Cantu, Nat. Energy, 2020, 5, 35–49.
40. D. Zhang, D. Li, Y. Hu, A. Mei and H. Han, Commun. Mater., 2022, 3, 58.
41. H. Zhu, S. Teale, M. N. Lintangpradipto, S. Mahesh, B. Chen, M. D. McGehee, E. H. Sargent and O. M. Bakr, Nat. Rev. Mater., 2023, 8, 569–586.
42. Q. Emery, M. Remec, G. Paramasivam, S. Janke, J. Dagar, C. Ulbrich, R. Schlatmann, B. Stannowski, E. Unger and M. Khenkin, ACS Appl. Mater. Interfaces, 2022, 14, 5159–5167.
43. Ł. Haryński, K. Trocki, S. Bykkam, G. del Pozo, B. Romero, B. Bochentyn and D. Glowienka, Sol. Energy, 2026, 303, 114131.
44. O. K. Segbefia, A. G. Imenes and T. O. Sætre, Sol. Energy, 2021, 224, 889–906.
45. M. Knausz, G. Oreski, M. Schmidt, P. Guttmann, K. Berger, Y. Voronko, G. Eder, T. Koch and G. Pinter, Polym. Test., 2015, 44, 160–167.
46. R. Cheacharoen, N. Rolston, D. Harwood, K. A. Bush, R. H. Dauskardt and M. D. McGehee, Energy Environ. Sci., 2018, 11, 144–150.
47. X. Zhang, S. Wu, H. Zhang, A. K. Y. Jen, Y. Zhan and J. Chu, Nat. Photonics, 2024, 18, 1243–1253.
48. Q. Jiang and K. Zhu, Nat. Rev. Mater., 2024, 9, 399–419.
49. W. Xu, B. Chen, Z. Zhang, Y. Liu, Y. Xian, X. Wang, Z. Shi, H. Gu, C. Fei, N. Li, M. A. Uddin, H. Zhang, L. Dou, Y. Yan and J. Huang, Nat. Photonics, 2024, 18, 379–387.
50. C. Gong, C. Wang, X. Meng, B. Fan, Z. Xing, S. Shi, T. Hu, Z. Huang, X. Hu and Y. Chen, Adv. Mater., 2024, 36, 2405572.
51. B. Fan, J. Xiong, Y. Zhang, C. Gong, F. Li, X. Meng, X. Hu, Z. Yuan, F. Wang and Y. Chen, Adv. Mater., 2022, 34, 2201840.
52. M. Park, S. C. Hong, Y.-W. Jang, J. Byeon, J. Jang, M. Han, U. Kim, K. Jeong, M. Choi and G. Lee, Int. J. Precis. Eng. Manuf. Green Technol., 2023, 10, 1223–1234.
53. L. A. Castriotta, R. Fuentes Pineda, V. Babu, P. Spinelli, B. Taheri, F. Matteocci, F. Brunetti, K. Wojciechowski and A. Di Carlo, ACS Appl. Mater. Interfaces, 2021, 13, 29576–29584.
54. Z. Li, T. R. Klein, D. H. Kim, M. Yang, J. J. Berry, M. F. A. M. van Hest and K. Zhu, Nat. Rev. Mater., 2018, 3, 18017.
55. A. Wąsiak-Maciejak, Ł. Przypis, W. Żuraw, K. Rycek, P. Janicka, M. Ścigaj, K. Dyk, H. Lai, A. Piejko, D. Pucicki, F. Fu, V. Kinzhybalo and K. Wojciechowski, J. Mater. Chem. A, 2025, 13, 7335–7346.
56. O. Y. Gong, G. S. Han, S. Lee, M. K. Seo, C. Sohn, G. W. Yoon, J. Jang, J. M. Lee, J. H. Choi, D.-K. Lee, S. B. Kang, M. Choi, N.-G. Park, D. H. Kim and H. S. Jung, ACS Energy Lett., 2022, 7, 2893–2903.
57. B. Zhou, P. Zhao, J. Guo, S. Hu, X. Guo, J. Liu and C. Li, Energy Environ. Sci., 2025, 18, 9865–9876.
58. M. Saliba, T. Matsui, J.-Y. Seo, K. Domanski, J.-P. Correa-Baena, M. K. Nazeeruddin, S. M. Zakeeruddin, W. Tress, A. Abate, A. Hagfeldt and M. Grätzel, Energy Environ. Sci., 2016, 9, 1989–1997.
59. T. Bu, J. Li, H. Li, C. Tian, J. Su, G. Tong, L. K. Ono, C. Wang, Z. Lin, N. Chai, X.-L. Zhang, J. Chang, J. Lu, J. Zhong, W. Huang, Y. Qi, Y.-B. Cheng and F. Huang, Science, 2021, 372, 1327–1332.
60. F. Cheng, X. Jing, R. Chen, J. Cao, J. Yan, Y. Wu, X. Huang, B. Wu and N. Zheng, Inorg. Chem. Front., 2019, 6, 2458–2463.
61. J.-W. Lee, Z. Dai, C. Lee, H. M. Lee, T.-H. Han, N. De Marco, O. Lin, C. S. Choi, B. Dunn, J. Koh, D. Di Carlo, J. H. Ko, H. D. Maynard and Y. Yang, J. Am. Chem. Soc., 2018, 140, 6317–6324.
62. V. Romano, L. Najafi, A. A. Sutanto, G. Schileo, V. Queloz, S. Bellani, M. Prato, S. Marras, M. K. Nazeeruddin, G. D'Angelo, F. Bonaccorso and G. Grancini, ChemPlusChem, 2021, 86, 1044–1048.
63. L. Huang, Z. Hu, J. Xu, K. Zhang, J. Zhang and Y. Zhu, Sol. Energy Mater. Sol. Cells, 2015, 141, 377–382.
64. M. A. Truong, T. Funasaki, L. Ueberricke, W. Nojo, R. Murdey, T. Yamada, S. Hu, A. Akatsuka, N. Sekiguchi, S. Hira, L. Xie, T. Nakamura, N. Shioya, D. Kan, Y. Tsuji, S. Iikubo, H. Yoshida, Y. Shimakawa, T. Hasegawa, Y. Kanemitsu, T. Suzuki and A. Wakamiya, J. Am. Chem. Soc., 2023, 145, 7528–7539.
65. B. Zhang, F. Guo, J. Xue, L. Yang, Y. Zhao, M. Ge, Q. Cai, B. Liu, Z. Xie, D. Chen, H. Lu, R. Zhang and Y. Zheng, Sci. Rep., 2017, 7, 17695.
66. B. P. Dhamaniya, P. Chhillar, A. Kumar, K. Chandratre, S. Mahato, K. P. Ganesan and S. K. Pathak, ACS Omega, 2020, 5, 31180–31191.
67. W. Zhu, M. Yang, T. Han, Y. Wang, X. Luo, W. Chai, H. Xi, L. Zhou, D. Chen, J. Zhang, C. Zhang and Y. Hao, Chem. Eng. J., 2024, 500, 156798.
68. R. Wang, X. Liu, S. Yan, N. Meng, X. Zhao, Y. Chen, H. Li, S. M. H. Qaid, S. Yang, M. Yuan and T. He, Nat. Commun., 2024, 15, 8899.
69. H. Chen, A. Maxwell, C. Li, S. Teale, B. Chen, T. Zhu, E. Ugur, G. Harrison, L. Grater, J. Wang, Z. Wang, L. Zeng, S. M. Park, L. Chen, P. Serles, R. A. Awni, B. Subedi, X. Zheng, C. Xiao, N. J. Podraza, T. Filleter, C. Liu, Y. Yang, J. M. Luther, S. De Wolf, M. G. Kanatzidis, Y. Yan and E. H. Sargent, Nature, 2023, 613, 676–681.
70. S. Li, Z. Liu, Z. Qiao, X. Wang, L. Cheng, Y. Zhai, Q. Xu, Z. Li, K. Meng and G. Chen, Adv. Funct. Mater., 2020, 30, 2005846.
71. C. Liu, Y. Yang, H. Chen, J. Xu, A. Liu, A. S. R. Bati, H. Zhu, L. Grater, S. S. Hadke, C. Huang, V. K. Sangwan, T. Cai, D. Shin, L. X. Chen, M. C. Hersam, C. A. Mirkin, B. Chen, M. G. Kanatzidis and E. H. Sargent, Science, 2023, 382, 810–815.
72. J. Wang, Y. Wu, J. Zhao, S. Lu, J. Lu, J. Sun, S. Wu, X. Zheng, X. Zheng, X. Tang, M. Ma, S. Yue, K. Liu, Z. Wang and S. Qu, Small Methods, 2024, 8, 2400043.
73. X. Zheng, T. Jiang, L. Bai, X. Chen, Z. Chen, X. Xu, D. Song, X. Xu, B. Li and Y. M. Yang, RSC Adv., 2020, 10, 18400–18406.
74. Z. Zheng, Z. Xue, K. Zhao, Y. Yang, X. Zhu, H. Li, S. Cheng, S. Li, N. Yan and Z. Wang, Sol. RRL, 2024, 8, 2301076.
75. R. Witteck, D. Nguyen Minh, G. Paul, S. P. Harvey, X. Zheng, Q. Jiang, M. Chen, T. Abzieher, A. F. Palmstrom, B. Habersberger, E. A. Gaulding, J. M. Luther and L. M. Wheeler, ACS Appl. Energy Mater., 2024, 7, 10750–10757.
76. P. Basumatary, J. Kumari and P. Agarwal, AIP Conf. Proc., 2021, 2369, 020022.
77. A. Mavlonov, Y. Hishikawa, Y. Kawano, T. Negami, A. Hayakawa and T. Minemoto, Sol. Energy Mater. Sol. Cells, 2025, 282, 113410.
78. N. Krobkrong, A. Mavlonov, Y. Hishikawa, Y. Kawano, T. Negami and T. Minemoto, Mater. Sci. Semicond. Process., 2026, 203, 110247.
79. A. Mavlonov, Y. Hishikawa, Y. Kawano, T. Negami, A. Hayakawa, S. Tsujimura, T. Okumura and T. Minemoto, Sol. Energy, 2025, 286, 113174.
80. S. Hu, J. Thiesbrummel, J. Pascual, M. Stolterfoht, A. Wakamiya and H. J. Snaith, Chem. Rev., 2024, 124, 4079–4123.
81. Y. Wang, R. Lin, X. Wang, C. Liu, Y. Ahmed, Z. Huang, Z. Zhang, H. Li, M. Zhang, Y. Gao, H. Luo, P. Wu, H. Gao, X. Zheng, M. Li, Z. Liu, W. Kong, L. Li, K. Liu, M. I. Saidaminov, L. Zhang and H. Tan, Nat. Commun., 2023, 14, 1819.
82. P. Spinelli, R. Fuentes Pineda, M. Scigaj, T. Ahmad and K. Wojciechowski, Appl. Phys. Lett., 2021, 118, 241110.

---