Cr–O–In interlocking for window layer delamination resistance in operationally stable perovskite/silicon tandem solar cells

Abstract

Perovskite/silicon tandem solar cells have exhibited remarkable efficiencies but remain limited by poor operational stability, primarily due to the instability of fullerene/atomic-layer-deposited (ALD) SnO_X interfaces in the top window layer – an issue that is often negligible in single-junction devices but becomes critically important in tandems. This instability originates from poor adhesion, incomplete precursor reactions during the low-temperature ALD process, and oxidative exposure of the fullerene layer. Here, we introduce a Cr–O–In interlocking interface formed via chromium (Cr) diffusion into the transparent electrode and fullerene layer, which effectively mitigates window layer delamination and establishes robust interfacial contact. Notably, Cr does not diffuse across the fullerene layer due to its strong reaction with the C₆₀ layer, which effectively immobilizes the metal atoms shortly after deposition and impedes deep diffusion. In addition, this interlocking interface integrates optical transparency, favorable band alignment, and resistance to sputtering, while enabling n-type doping of the fullerene layer to enhance electron mobility and suppress across-interface nonradiative recombination. As a result, single-junction 1.67 eV perovskite solar cells exhibit a PCE of 23.27% (average PCE of 22.85%), and perovskite/silicon tandems exhibit 32.77%, with a certified stabilized efficiency of 32.51%, among the highest reported for monolithic perovskite/TOPCon tandems. Moreover, the Cr–O–In ionic interlocking interface enhances long-term operational stability, paving the way toward scalable, industrially applicable tandem photovoltaics.

---

Broader context

Perovskite/silicon tandem solar cells are at the forefront of next-generation photovoltaic technology, offering a direct path to surpass the efficiency limits of conventional single-junction cells and significantly reduce the cost of solar electricity. However, their commercial viability is critically hindered by poor interfacial adhesion, which leads to rapid degradation and operational failure. In this work, we address this fundamental challenge by introducing a thermally evaporated chromium oxide (CrO_X) buffer layer. First, thermal evaporation avoids sputtering damage during transparent conductive oxide deposition and ambient exposure, reducing interfacial non-radiative recombination. Second, residual Cr diffuses into C₆₀, forming a mixed interfacial region and n-doped C₆₀ that strengthens the buried Cr–C₆₀ adhesion and electron transport, respectively. Third, the Cr affinity of oxygen promotes strong Cr–O–In interlocking with IZO, significantly strengthening interfacial adhesion and device stability. Fourth, the oxides of Cr have a deep valence band and electron selectivity, enabling effective hole blocking and improved energy-level alignment. Consequently, single-junction perovskite and monolithic perovskite/TOPCon tandem cells exhibit 23.27% and 32.77% PCEs, respectively. Crucially, the devices exhibit exceptional operational stability for over 1400 h. This work addresses the persistent challenge of interfacial delamination and provides a simple, scalable, and low-cost pathway to accelerate the commercialization of highly efficient and durable perovskite/silicon tandem solar cells.

Introduction

Perovskite/silicon tandem solar cells (TSCs) have emerged as some of the most promising candidates to surpass the efficiency limits of single-junction devices and accelerate the deployment of next-generation photovoltaic technologies. In recent years, the certified power conversion efficiency (PCE) of monolithic TSCs has rapidly advanced, demonstrating remarkable potential for both academic research and industrial applications.¹ Despite these impressive achievements in efficiency, the operational stability of perovskite/silicon tandem devices remains a critical bottleneck. Although many single-junction perovskite solar cells (PSCs) have already demonstrated long-term stability under light-soaking or outdoor conditions—reaching tens of thousands of hours,^2–5 the reported stability of tandems typically lags far behind, often limited to only several hundred to one thousand hours.^6–9

One of the primary factors underlying this disparity in instability is the poor mechanical stability of transparent conductive oxide (TCO) window contacts. In most reported tandem architectures, atomic layer deposited SnO_X (ALD–SnO_X) has been widely adopted as a buffer layer due to its high transparency, electron selectivity, thermal stability and scalability, protecting the underlying fullerene and perovskite layers from sputtering damage during TCO deposition.^10–12 However, this commonly used inorganic/organic stacking strategy intrinsically suffers from weak interfacial adhesion.¹³ The large thermomechanical property mismatches between inorganic SnO_X and organic charge transport layers (primarily the C₆₀ fullerene layer), along with their weak van der Waals bonding, render the interface particularly vulnerable.¹⁴ During long-term operation, mechanical stresses accumulate from two main sources: (i) internal vapor pressures generated by the release of gaseous decomposition products from the perovskite absorber,¹⁵ and (ii) repeated thermal shocks during device processing and temperature cycling in outdoor operation. These stresses easily trigger delamination and crack formation at the fragile SnO_X/C₆₀ interface, particularly under illumination from this side in tandem devices. Such mechanical failures not only interrupt carrier extraction but also generate diffusion channels that accelerate the ingress of moisture and the outgassing of volatile organics, further undermining device stability and mechanical integrity.

In addition to adhesion challenges, the intrinsic processing limitations of ALD–SnO_X are also recognized as key factors restricting the long-term stability of tandem devices. Owing to the thermal sensitivity of perovskite absorbers, ALD–SnO_X deposition must occur at low temperatures, which often leads to incomplete precursor reactions and undesirable side reactions with the perovskite layer. For example, the precursor tetrakis(dimethylamino)tin (IV) (TDMASn) can undergo ligand exchange with formamidinium cations, leading to PbI₂ formation, higher series resistance, and reduced charge extraction.¹⁶ The buried C₆₀ layer is also highly vulnerable to oxygen exposure, either during chamber transfer or from oxygen-containing precursors (e.g., water vapor), which significantly decreases carrier mobility and stability.^11,17–20 Therefore, new buffer layers are urgently needed that combine high transparency, conductivity, and stability with strong adhesion and protection against oxygen-induced degradation of the C₆₀ transport layer.

In this study, we introduce a multifunctional CrO_X buffer layer deposited by thermal evaporation as a promising alternative to ALD–SnO_X. It is fully compatible with C₆₀ thermal evaporation without requiring chamber transfer, thereby eliminating ambient exposure and oxygen-induced defects. The CrO_X layer integrates high conductivity, excellent optical transparency, and favorable energy level alignment to enable ohmic contact and efficient electron extraction. Its dense morphology blocks oxygen permeation, while controlled chromium diffusion into the C₆₀ layer induces n-type doping, thereby enhancing conductivity. More importantly, this diffusion process promotes the formation of a metal–C₆₀ complex and a robust Cr–O–In interlocking structure with the zinc-doped indium oxide (IZO) top electrode, thereby enhancing adhesion, preventing delamination and ensuring stable interfacial contact. As a result, inverted single-junction PSCs deliver a champion PCE of 23.27%, the highest reported for ALD-free devices, using a perovskite absorber with a bandgap of ∼1.67 eV. When integrated into perovskite/silicon tandems, a certified 32.51% efficiency is achieved, among the highest reported for monolithic perovskite/TOPCon tandems. The tandem devices also exhibit outstanding operational stability after 1400 h of continuous illumination, which implies a reversible degradation behavior. This work thus presents not only a high-performance, ALD-free buffer layer but also a scalable, damage-free interfacial engineering strategy, paving the way for the commercialization of high-efficiency tandem photovoltaics.

Results and discussion

Fabrication and characterization of the CrO_X buffer layer

A schematic diagram of the thermal evaporation process for sequentially depositing C₆₀ and CrO_X films is shown in Fig. 1a, where the Cr metal was used as the evaporation source. Here, we selected Cr because its interfacial behavior with C₆₀ is fundamentally different from weakly interacting metals such as Ag, Cu, or Au, which easily diffuse deep into the fullerene matrix during evaporation and form metallic clusters.²¹ As a highly reactive transition metal, Cr undergoes strong localized chemical interactions with the C₆₀ π-system. Charge transfer and chemical interaction at the Cr/C₆₀ interface, consistent with the known coordinative capacity of fullerenes, result in a significantly reduced diffusivity of Cr into the organic material.^21–24 This confined intermixing prevents Cr from penetrating toward the C₆₀/perovskite buried interface and results in a highly adherent, robust interfacial layer at the C₆₀ surface. This unique adhesion-enhancing behavior is also the reason why Cr is widely used in microelectronics to improve metal adhesion.²⁵ In addition, the Cr surface can be readily oxidized into a dense CrO_X layer upon air exposure or during subsequent TCO deposition due to its strong oxygen affinity.²⁶ Such CrO_X layers exhibit excellent electron selectivity and strong adhesion, serving as effective electron-selective layers in perovskite solar cells.^27,28 Moreover, the formation energy of Cr oxides is much lower than that of chromium halides, enabling the CrO_X layer to effectively dampen halide diffusion from the perovskite into the electrode, thereby suppressing interfacial chemical reactions and improving device stability.²⁹

Fig. 1 Characterization of SnO_X and CrO_X on C₆₀ films. (a) Schematic diagram of the thermal evaporation process for depositing CrO_X films. (b) The XPS spectra of Cr 2p in the CrO_X film. (c) XRD pattern of the as-evaporated CrO_X film. (d) The transmittance of SnO_X and CrO_X films on the glass substrates. C-AFM current mapping images of C₆₀/SnO_X (e) and C₆₀/CrO_X (f) films. (g) The high-resolution Raman spectra of C₆₀ and C₆₀/CrO_X films. (h) The UPS spectra of C₆₀, C₆₀/SnO_X and C₆₀/CrO_X films. (i) The corresponding energy level diagram.

To investigate the composition of the thermally evaporated CrO_X film, the chemical states were first investigated by X-ray photoelectron spectroscopy (XPS). The peak at 573.6 eV corresponds to Cr 2p_3/2 of metallic Cr, whereas the peak at 582.9 eV relates to Cr 2p_1/2 (Fig. 1b). Besides, the peak of CrO_X was split into Cr 2p_3/2 at 576.3 eV and Cr 2p_1/2 at 586.2 eV, respectively.²⁷ The XPS peak of O 1s is deconvoluted into two peaks, where the peak at lower binding energy is related to O in CrO_X and the other peak at higher binding energy is attributed to contamination (Fig. S1). The crystal structure of the thermally evaporated CrO_X film demonstrates an amorphous characteristic without detectable diffraction peaks (Fig. 1c). We further study the effect of the CrO_X film on the optical transmittance (Fig. 1d). It is obvious that the 3 nm CrO_X film shows an excellent transmittance (300–1200 nm), which is comparable to that of the SnO_X film, indicating a negligible parasitic absorption, as evidenced by the photographs of C₆₀, C₆₀/SnO_X and C₆₀/CrO_X films. The glass color shows a minor change after depositing the CrO_X buffer layer. In conclusion, the resulting CrO_X film consists of a surface oxide layer and residual metallic Cr, which is prone to serve as an electron donor and endows the film with heavy n-type properties, thereby facilitating the electron transport.²⁷

Next, to study the electrical contact performance between CrO_X and C₆₀, we carried out the conductive atomic force microscopy (C-AFM) measurement to investigate the spatial distribution of the current (Fig. 1e and f). Compared to the SnO_X film, the CrO_X film exhibits a dramatically enhanced dark current from 3.83 to 472 pA. This increment is attributed to the superior conductivity of the CrO_X film because of the residual Cr metal, which is of benefit to carrier transport and extraction. Moreover, the roughness of the CrO_X film (4.30 nm) is diminished and its coverage is improved compared to that of SnO_X (7.27 nm) (Fig. S2). A smoother surface is conducive to successive deposition of transparent electrode IZO and forming a good contact with the adjacent IZO and ETL, reducing interfacial defects and interfacial non-radiative recombination.¹⁰ To further study the mechanism of improved conductivity, we performed Raman spectra analysis with high spectral resolution (Fig. 1g). As can be seen, the characteristic peak at 1459 cm⁻¹ of C₆₀ slightly changes after depositing CrO_X, probably indicating electron density transfer from metal to carbon cluster (n-type doping), consistent with previous work.^24,30 The CrO_X buffer layer not only suppresses the interfacial defect recombination, but also possesses a favorable energy level matching (Fig. 1h and i). Ultraviolet photoelectron spectroscopy (UPS) was carried out to study the electronic properties of SnO_X and CrO_X films (Fig. 1h and Fig. S3). The bandgap of C₆₀ is determined to be 2.24 eV (Fig. S4), and a bandgap of 3.9 eV was adopted for SnO_X, which is consistent with previous report.³¹ The positions of the conduction band edge (CBM), Fermi level and valence band edge (VBM) of SnO_X are 3.22 eV, 4.06 eV and 7.12 eV. The CBM of the CrO_X film is 3.89 eV, which is close to that of the C₆₀ layer, demonstrating the transportation of photogenerated carriers from the ETL of C₆₀ to the CrO_X buffer layer (Fig. 1i). Furthermore, the VBM of the CrO_X film is 7.39 eV lower than that of the SnO_X film, meaning that the CrO_X film can be utilized as an effective hole-blocking layer, which is in line with previous works.^27,28,32,33 These results illustrate that the CrO_X film possesses good electron selectivity and superior energy-alignment, which is energetically favorable for electron collection.

The enhanced interfacial adhesion

The C₆₀/SnO_X interface is identified as the mechanically weakest link owing to the presence of chemical bonding, prone to delamination under operational stress cycles (bias/thermal/light). To evaluate the adhesion strength of C₆₀, SnO_X and IZO in our system, we conducted tape-peeling tests (Fig. S5). The ALD–SnO_X film exhibited poorer adhesion to glass compared to the C₆₀ film. After sputtering and then peeling the IZO electrode, complete delamination of IZO occurred alongside partial removal of the underlying SnO_X, indicating that sputtering bombardment has a negative effect on the interfacial adhesion. To further investigate the delamination location, we did the tape-peeling tests on the perovskite films and tandem devices (Fig. S6 and S7). After tape-peeling, the color of the PVK/C₆₀/SnO_X/IZO film becomes different from those of pure perovskite and the PVK/C₆₀/SnO_X film, indicating that the interface between C₆₀ and SnO_X is the weakest (Fig. S6). As can be seen from Fig. S7, the top electrode fully delaminates upon tape-peeling on the tandem device. The colors outside the effective area which belongs to PVK/SnO_X are different from those of the IZO covered area and pure perovskite according to the tape-peeling test (Fig. S7a), implying that C₆₀ remains. Therefore, we believe that the delamination occurred at the interface between C₆₀ and SnO_X. To further understand the microscopic properties of the delamination, the cross-sectional SEM was performed on the peeled tandems fabricated on flat silicon substrates to avoid texture-related artifacts (Fig. S7b–d). A distinct gap can be observed at the IZO/SnO_X/C₆₀ interface (Fig. S7b), confirming a poor interfacial adhesion. At the peeling interface marked by a white circle in Fig. S7a, a thin layer remains atop the perovskite after tape removal, consistent with complete removal of IZO/SnO_X while the C₆₀ layer is largely retained (Fig. S7c and d). The observed thickness variation in Fig. S7c is attributed to the characteristic wrinkles formed on the perovskite surface during the crystallization process, which act as local stress concentrators and further promote delamination under external stimuli. Together, these results collectively imply that delamination preferentially occurs at the C₆₀/SnO_X junction. To evaluate the impact of CrO_X on interfacial adhesion, we performed 90° peel tests, which show that the target sample exhibits enhanced adhesion strength compared to control (Fig. 2a), indicating reduced sputtering ion bombardment and improved film adhesion. This enhanced mechanical integrity is further corroborated by the process of preparing the SEM cross-sectional sample: the control sample exhibits conspicuous delamination at the interface, whereas the target sample remains intact and resists separation, indicating robust contact and strong adhesion (Fig. 2b). We attribute this improvement to the Cr–O–In interlocking effect at the C₆₀/CrO_X/IZO interface, in which Cr diffusion into the C₆₀ surface forms a mixed interfacial region that strengthens the buried Cr–C₆₀ adhesion, while Cr simultaneously forms strong chemical bonds with the top IZO layer.

Fig. 2 The Cr–O–In interlocking and Cr diffusion mechanism. (a) The load–displacement curves of control and target perovskite films during the 90° peel tests. (b) The cross-sectional SEM images of control and target devices. (c) The related elemental maps of the tandem solar cell based on the CrO_X buffer layer. (d) The corresponding elemental variations with depth. (e) ToF-SIMS in-depth profiles (negative ions) of the ITO/C₆₀/CrO_X film. Depth profile XPS results of O, Sn or Cr in ITO/C₆₀/SnO_X (f) and ITO/C₆₀/CrO_X (g) films. (h) 3D ELF plot containing the chemical bond between Cr–O and In–O. (i) Schematic illustration of the mechanism of O₂-blocking and Cr diffusion.

To verify this mechanism, we first studied the elemental distribution using high-angle annular dark field (HAADF) STEM (scanning transmission electron microscopy) coupled with energy dispersive X-ray (EDX) mapping. IZO, CrO_X, and C₆₀ can be clearly identified, indicating the homogeneous distribution of these elements (Fig. 2c). The distribution of the representative elements across each layer demonstrates a clear spatial overlap of Zn, Cr and C signals, confirming the migration of Cr and the formation of an interlocking structure between IZO and the CrO_X film (Fig. 2d). Moreover, the high-resolution STEM image reveals a clear contrast between the IZO and CrO_X layers (Fig. S8). We estimated the region marked by a red arrow, which is ∼1.7 nm corresponding to the thickness of CrO_X. We also performed time-of-flight secondary-ion mass spectrometry (ToF-SIMS) to explore the relative element distribution throughout the entire film. As shown in Fig. S9, the intensity of the SnO₂⁻ signal exhibits a relatively fast decline at the interface of SnO_X and C₆₀, suggesting a distinct boundary with no significant interlayer mixing. In contrast, the CrO₂⁻ and Cr⁻ signals overlap with the C⁻ signal and it shows a slow decline, meaning that Cr diffuses into C₆₀, which is in line with the TEM results. Upon increasing the etching time, the intensity of the Cr⁻ signal is gradually higher than that of the CrO₂⁻ signal, indicating that the Cr metal is accumulated below the CrO_X layer. The CrO₂⁻ signal is an indicator for the CrO_X layer, and the Cr⁻ signal is related to the residual Cr metal.³⁰ The Cr⁻ signal intensity at the top of the C₆₀ layer is over 1 × 10² counts. This signal decays by over two orders of magnitude to approximately 1 × 10⁰ counts at the bottom of the C₆₀ layer. The decay of this magnitude (10² to 10⁰) strongly indicates that Cr diffusion is effectively blocked and does not undergo substantial diffusion across the entire fullerene layer. The signal detected in the middle of the C₆₀ layer (e.g., ∼1 × 10¹ counts after ∼150 s of sputtering) represents only a trace-level presence deeper in the C₆₀ film, since the decline involves a process. The dimerized Cr₂⁻ signal, which is a more specific indicator of the Cr–Cr fragment, mixes with C₆₀ and becomes undetectable shortly after sputtering for 150 s. These results indicate that Cr diffuses into a certain depth within the C₆₀ layer, and it will cease to diffuse due to its limited long-range mobility and chemical reaction with C₆₀, rather than aggregating on the surface of C₆₀.²¹ Such localized incorporation induces n-doping in C₆₀, thereby enhancing conductivity and electron mobility, in agreement with the C-AFM and Raman results.

During fabrication of tandems, the C₆₀ layer is unavoidably exposed to ambient air during transfer to the ALD chamber for SnO_X deposition, which results in the generation of interfacial defects. To probe the evolution of O and Sn or Cr elements in control and target films to evaluate the formation of a mixed interfacial region and O₂-blocking effects, the depth-dependent XPS analysis was further performed. The samples were exposed to ambient air during transfer for XPS measurement. However, as both the control and CrO_X modified samples were handled identically, the comparative XPS data reliably reveal the relative changes in the chemical state and element content. The C 1s peaks of the target film exhibit a significant shift toward lower binding energy with increasing etching depth, indicating the formation of a metal–C₆₀ complex that possesses a lower binding energy than those of other bonding states, consistent with previous findings (Fig. S10d).³⁴ The increasing metallic Cr fraction accompanied by a decreased CrO_X fraction further demonstrates the presence of the Cr metal at the bottom and the oxidation of the Cr surface (Fig. S10f). Moreover, the O content in both films was analyzed (Fig. 2f and g). Due to the presence of adventitious carbon contamination, the C concentration was not quantified, and the C 1s peak was not used for calibration. All reported atomic ratios are normalized accordingly. The O atomic ratio in the C₆₀/SnO_X film drops to 60.3% from 72.6%, whereas that of the C₆₀/CrO_X film decreases from 71.1% to 53.0%. The O content in the C₆₀/CrO_X film is apparently lower than that of the C₆₀/SnO_X film when etching to the C₆₀ layer and exhibits a fast decline, indicating that the exposure of the C₆₀ film to ambient air during transferring to the ALD chamber facilitates the permeation of oxygen. This permeation can lead to the formation of interfacial defects and non-radiative recombination.³⁵ Additionally, we compared two samples: (1) the standard sequentially deposited C₆₀/CrO_X, and (2) a sample where the C₆₀ layer was intentionally exposed to ambient air for 40 min prior to CrO_X deposition (C₆₀/Air/CrO_X). As this XPS measurement was performed with the sample loaded in an inert glovebox environment, the resulting elemental composition data are not directly comparable to those in Fig. 2f and g, where samples were exposed to air during transfer. At the initial etch stage (∼10 s), the C₆₀/Air/CrO_X sample shows a higher proportion of oxidized Cr species compared to metallic Cr (Fig. S11b and f) marked by red arrows, indicating that pre-adsorbed oxygen on C₆₀ influences the subsequent oxidation state of the Cr layer. Upon etching for about 100 s, the O signal exceeds the Cr signal and the In 3d peak appears, indicating that the C₆₀ layer has been completely sputtered through, revealing the underlying ITO substrate (Fig. S11d, h, i and j). Meanwhile, the O content in the C₆₀/CrO_X sample decreased more rapidly, and the lowest level was lower than that of C₆₀/Air/CrO_X. More importantly, the atomic ratio analysis (Fig. S11k and l) reveals that both the O/Cr and O/C ratios are significantly lower in the C₆₀/CrO_X sample than in the air-exposed C₆₀/Air/CrO_X sample. This direct comparison confirms that exposing C₆₀ to ambient air substantially increases the oxygen content within the interfacial region. The CrO_X layer effectively suppresses oxygen ingress. To verify this hypothesis, we compared the PL, PL mapping and SEM images of PVK/C₆₀ and PVK/C₆₀/CrO_X films before and after 6-hour aging in ambient air (Fig. S12–S14). The PVK/C₆₀ film exhibited a dramatic PL red-shift related to phase segregation and weakened carrier extraction, indicative of significant oxygen/water-induced degradation (Fig. S12). In stark contrast, the PVK/C₆₀/CrO_X film showed only minimal changes, which are consistent with the PL mapping results (Fig. S13). SEM imaging further revealed the formation of pinholes in the aged PVK/C₆₀ film, while the morphology of the PVK/C₆₀/CrO_X film remained largely unchanged (Fig. S14). Collectively, these results confirm that the CrO_X interlayer effectively blocks ambient-induced degradation and maintains stable interfacial conductivity. To elucidate the effect of chemical bonds between the CrO_X film and the top IZO layer, we performed the XPS analysis for IZO and CrO_X/IZO films. It can be seen that the XPS peaks of O 1s and In 3d shift to higher binding energy, indicative of chemical interaction between IZO and CrO_X (Fig. S15). The XPS spectra of C 1s and Zn 2p are shown in Fig. S16. To distinguish the chemical bonding type between CrO_X and IZO, the density functional theory (DFT) calculations modeling the interface were performed to obtain the interfacial energy and the behavior of the electron localization function (Fig. 2h and Fig. S17). The interfacial binding energy of Cr–O was calculated to be −0.121 eV Å⁻¹. This negative value signifies that the interface formation (Cr–O–In) is thermodynamically favorable and results in a stable heterojunction with strong interactions, making it resistant to delamination or failure. Moreover, a region of a high ELF value with a distorted spherical shape is observed around the O atoms, whereas almost no electron localization is evident around the Cr atoms. This is consistent with a predominantly ionic Cr–O bond, where the interaction is strong and clearly has no sharing of electrons, as directly visualized by the ELF analysis. Given the strong ionic bonds between Cr and oxides, we infer that the interlocking structure of Cr–O–In can be formed, which can enhance adhesion between TCO and C₆₀/perovskite.³⁴

The proposed mechanism illustrating Cr diffusion and O₂-blocking is depicted in Fig. 2i and Fig. S18. As reported, exposure of C₆₀ to oxygen leads to the generation of deep trap states and enhances non-radiative recombination.³⁵ This is particularly related to the device fabrication process. For instance, the subsequent deposition of the SnO_X layer via atomic layer deposition (ALD) typically employs water and TDMASn as co-reactants at low temperatures (80–120 °C). Transferring the C₆₀ film from the glovebox to the ALD chamber inevitably involves ambient exposure, potentially leading to interactions between C₆₀ and oxygen that significantly alter the electronic properties of the C₆₀ layer. To mitigate this issue, we employed a thermally evaporated CrO_X layer immediately after C₆₀ deposition. This process inherently shields C₆₀ from ambient air. More importantly, metallic Cr possesses a strong affinity for oxygen, resulting in the formation of a dense and smooth CrO_X layer on its surface.³⁶ This oxide layer acts as an effective barrier, thereby preventing the penetration of oxygen into the C₆₀ layer. The mechanism involves a dynamic interfacial process (Fig. S19). Owing to its high surface energy and reactivity, deposited Cr atoms initially interact with the C₆₀ layer, where the bulk diffusion is dominated. As deposition continues, the chemical reaction between Cr and C₆₀ leads to a self-limiting diffusion behavior that saturates the interface/bulk and suppresses deep diffusion, as evidenced by the sharp attenuation of the Cr⁻ and disappearance of Cr₂⁻ in ToF-SIMS. Consequently, even an ultra-thin Cr layer acts as an effective diffusion barrier. Subsequently, continued deposition results in the growth and oxidation of Cr at the surface. In essence, specific chemical interaction at the Cr/C₆₀ interface confines penetration to a shallow, trace level. The diffusion of Cr ions may induce n-type doping in the C₆₀ layer. Furthermore, the affinity of oxygen results in a robust interlocking interaction between CrO_X and IZO, enhancing the interfacial adhesion, thereby improving device stability.

The enhanced carrier transport and extraction

To further explicitly illustrate the effects of O₂-blocking and C₆₀ doping on electron transport and collection, fluorescence lifetime imaging microscopy (FLIM) was conducted (Fig. 3a–d and Fig. S20 and S21). A significant reduction in the τ_ave lifetime is observed upon incorporation of the C₆₀ layer compared to the pristine perovskite film, indicative of efficient electron extraction at the interface. When further introducing SnO_X or CrO_X as the buffer layer, the CrO_X-based perovskite film exhibits the shortest τ_ave lifetime (2.99 ns), in contrast to the SnO_X-based film (4.86 ns), which takes account both individual fitting components (τ₁ and τ₂) and their relative contribution (A₁ and A₂) shown in Table S1.^37,38 A significant reduction implies that CrO_X buffer could further improve charge collection, which can be attributed to improved conductivity and better energy alignment, as discussed in Fig. 1. The statistic distribution of lifetime τ₁ further reveals improved uniformity in the perovskite film with the CrO_X buffer layer, suggesting a reduction in interfacial defects (Fig. 3e). The FLIM results substantiate the enhanced electron selectivity and reduced carrier accumulation in the perovskite film with the CrO_X buffer layer. In Fig. 3f, a slightly high plateau at a longer delay time in the time-resolved photoluminescence (TRPL) spectra means a reduced defect density, which is in line with relatively enhanced PL intensity (Fig. S22). To further evaluate the role of the CrO_X buffer layer in enhancing the electron-transporting properties, the electron mobility was estimated using current–voltage (J–V) curves.³⁹ As shown in Fig. S23, the CrO_X-based device exhibits a higher slope than the C₆₀- and SnO_X-based devices. Furthermore, this enhanced slope for the CrO_X device shows no reduction after 24 hours of aging in ambient air attributed to the O₂-blocking effect and downward diffusion of Cr, while the control device degrades, evidencing that the CrO_X interlayer improves the overall conductivity and stability of the electron transport. A schematic of the proposed mechanism for electron extraction is presented in Fig. 3g. Upon incorporating the Cr buffer layer, the O₂-induced trap states can be suppressed, thereby reducing the non-radiative recombination at the interface and facilitating efficient electron extraction. To further figure out the enhanced electron collection, Kelvin probe force microscopy (KPFM) was conducted to explore the impact of CrO_X on the energy level alignment (Fig. 3h and i). The CrO_X-based perovskite film shows an increased contact potential difference (CPD), indicating an upward shift of the Fermi level and a more pronounced n-type surface. Thus, the incorporation of the CrO_X buffer layer can result in a cascade energy level, which is conducive to transferring the electrons. The PL spectrum of the perovskite film with IZO shows more pronounced PL quenching (Fig. S24a). And a fast delay time at early stages and a high plateau at longer time means more efficient electron extraction, in good agreement with the TRPL results of the perovskite film without IZO (Fig. S24b). These results indicate that the CrO_X buffer layer not only reduces the interfacial defects but also enables superior band matching.

Fig. 3 The carrier transport and extraction. (a)–(d) FLIM mapping images of PVK, PVK/C₆₀, PVK/C₆₀/SnO_X and PVK/C₆₀/CrO_X films, respectively, which indicate the lifetime of τ₁. (e) The statistic histograms of lifetime τ₁ extracted from the FLIM mapping. (f) The TRPL patterns of PVK, PVK/C₆₀, PVK/C₆₀/SnO_X and PVK/C₆₀/CrO_X films. (g) A schematic of the proposed mechanism for extracting electrons. The KPFM images of control (h) and target (i) perovskite films. (j) The XRD patterns of perovskite films.

Furthermore, the X-ray diffraction (XRD) analysis of four perovskite films was carried out to evaluate the perovskite quality after sputtering. After the deposition of IZO, the crystal orientation of PVK/C₆₀ without the protective layer undergoes noticeable changes, whereas the PVK/C₆₀/SnO_X and PVK/C₆₀/CrO_X films exhibit an inconspicuous variation (Fig. 3j). It has been reported that the magnetron sputtering can result in lattice strain in the perovskite film, as evidenced by the minor shifts in XRD peaks.⁴⁰ As shown in amplified XRD patterns (Fig. S25a), the diffraction peaks of the PVK/C₆₀/SnO_X/IZO film display an observable shift toward higher angles, while those of the PVK/C₆₀/CrO_X/IZO film show a slight change, indicating the negligible sputtering damage. Without the IZO deposition, the crystal orientation and peak positions of these films are identical, indicating that the shift in XRD peaks is indeed attributed to sputtering deposition (Fig. S25b and c). The top-view scanning electron microscopy (SEM) and AFM images of perovskite films after depositing IZO exhibit a similar morphology, further verifying the prevented damage from magneton sputtering (Fig. S26). These results corroborate that the thin CrO_X film can effectively inhibit the perovskite film from damage during the magneton sputtering, in line with previous study,^41,42 where 3 nm CuO_X can effectively inhibit the sputtering damage. We also investigated the surface morphology of perovskite films without depositing IZO (Fig. S27). It can be seen that the Cr-based perovskite film shows a smaller roughness, suggesting that the thermal evaporation of Cr enables a flatter surface, which is conducive to improving the interfacial contact.

Device performance

To examine the effectiveness of the interfacial optimization by the CrO_X buffer layer, we fabricated inverted single-junction PSCs based on an architecture of glass/ITO/4PADCB/perovskite/PDAI₂/C₆₀/buffer layer (SnO_X or CrO_X)/IZO (Fig. 4a). The cross-sectional SEM images are shown in Fig. S28. The composition of the perovskite absorber is Cs_0.05FA_0.8MA_0.15Pb(I_0.76Br_0.24)₃ with a bandgap of 1.67 eV (Fig. S29). Compared to the control device presenting a PCE of 22.79% with an open-circuit voltage (V_OC) of 1.234 V, the target device achieves overall enhancement in photovoltaic parameters, as shown in J–V curves (Fig. 4b). The champion target device exhibits a PCE of 23.27% with a V_OC of 1.251 V and a FF of 84.48%, and it can achieve a maximum V_OC of 1.263 V (Fig. S30). We attribute the marginal increment in the FF primarily to the disparity in extraction kinetics, where hole extraction (∼100 ns) is considerably slower than electron extraction (∼1 ns), thereby limiting further improvement.⁴³ Both devices exhibit no obvious hysteresis. This is among the highest PCE reported so far in inverted PSCs with a bandgap of approximately 1.68 eV. And the performance of our devices is state-of-the-art among devices without depositing SnO_X (Table S2). Compared to other buffer layers besides BCP and SnO_X, the CrO_X-based devices exhibit superior photovoltaic performance, indicating that the CrO_X film can be used as an efficient hole-blocking layer. It should be noted that the thickness of the CrO_X film is important for the device performance (Fig. S31). The 1 nm CrO_X film cannot inhibit the sputtering damage and the 5 nm CrO_X film may influence the electron transport and current density, thereby deteriorating the photovoltaic performance. The integrated short current density (J_SC) of control and target devices is calculated to be 21.53 and 21.63 mA cm⁻², respectively, which is extracted from the external quantum efficiency (EQE) spectra, showing agreement with the J_SC derived from J–V curves (Fig. 4c). The CrO_X buffer layer can serve as an optical spacer to regulate the light intensity and enhance the J_SC as previously discussed for TiO_X.^28,44 Moreover, the steady-state power output (SPO) of control and target devices at the maximum power points (MPP) for 200 s is calculated to be 22.51 and 23.24%, respectively, demonstrating the reliability of PCE enhancement (Fig. 4d). The statistic plots further validate the enhanced performance, indicating the reproducibility of the CrO_X buffer layer (Fig. 4e and Fig. S32). A statistical average PCE of 22.85% is obtained, which is notably higher than the 22.32% average of the control (Fig. S33).

Fig. 4 The photovoltaic performance of single-junction devices. (a) The device structure of a single-junction solar cell. (b) The J–V curves of the champion control and target devices under simulated AM1.5G solar illumination. The inserted table represents the corresponding parameters. (c) The EQE spectra and corresponding integrated J_SC of the champion control and target devices. (d) Steady-state PCE of control and target devices. (e) Statistical distribution of V_OC and PCE. (f) V_OCversus light intensity curves of control and target devices. The long-term operational stability (g) and dynamic stability (h) under MPP tracking upon exposure to cycled light illumination (12 h dark/12 h light).

A suite of measurements were carried out to uncover the mechanism behind the enhanced performance. The light intensity-dependent V_OC curves are shown in Fig. 4f, where V_OC exhibits a linear dependence on the seminatural logarithm of light intensity. The plot slopes of control and target devices are 1.40 k_BT/q and 1.28 k_BT/q, respectively. The reduced ideality factor in the target device suggests repressed trap-assisted non-radiative recombination at the interface, contributing to improved V_OC and FF. This is further supported by the increased external quantum efficiency of electroluminescence (EQE_EL) values (Fig. S34a). We performed the electronic impedance spectroscopy (EIS) measurements in the dark to acquire the recombination resistance (R_rec) (Fig. S34b). The conspicuously increased R_rec indicates a superior contact and reduced non-radiative recombination. Capacitance–voltage (C–V) measurement was then carried out to explore the separation of photoinduced carriers. Based on the Mott–Schottky formula detailed in the SI, the target device exhibits a higher V_bi than that of control (Fig. S35). The increased V_bi accelerates the generation of robust electric field in the depletion region and suggests an enhanced driving force for the transport and extraction of photogenerated carriers, thereby suppressing the electron–hole recombination, aligning perfectly with the lifetime mapping and energy band matching analyses, which would lead to a smaller V_OC deficit. Furthermore, to estimate the variation of defect state density, we performed the space charge limited current (SCLC) analysis. On the basis of the dark J–V curves of electron-only devices, the trap-filled limit-voltage (V_TFL) can be estimated to be 0.25 V and 0.18 V for control and target devices, respectively (Fig. S36). The corresponding defect-state density of control and target can be calculated to be 2.4 × 10¹⁵ and 1.73 × 10¹⁵ cm⁻³, respectively. These results demonstrate that the CrO_X buffer layer can reduce the interfacial defects and non-radiative recombination, in accordance with enhanced transport and extraction of electrons.

Notably, CrO_X at the surface of the Cr film has excellent stability to illumination, high temperature, and air, which is conducive to improving the operational stability of devices.³² To confirm this, the devices were operated in a N₂-flow box (25–35 °C) under white LED illumination with a power density of 100 mW cm⁻² (ISOS-L-1) (Fig. 4g). While both devices exhibit comparable stability during the initial 200 hours of continuous illumination, the target device demonstrated better operational stability beyond this point and it remains 90% of its initial PCE after 500 h aging. To highlight the influence of interfacial adhesion among different buffer layers during day–night operation, we further investigated the dynamic stability under cycled light illumination (12 h dark/12 h light), where the device was periodically held in the dark to facilitate the recovery (Fig. 4h). It can be seen that the PCE varies from a progressive enhancement trend to a sharp decline over cycling, suggesting reversible behavior in PCE. The control exhibits a comparable downward trend to the target during the initial 250 h, followed by a severe degradation, potentially attributed to delamination between the perovskite and top layers under cyclic light conditions. In contrast, the target device shows excellent dynamical stability, retaining 93% of its initial PCE after 650 hours of MPP tracking under cyclic light illumination, which can be attributed to enhanced interfacial adhesion. Moreover, the ambient air stability of perovskite films was also studied because the CrO_X layer possesses very promising ambient air stability (Fig. S37).²⁹ Obviously, the center of the perovskite film without the buffer layer suffered from obvious fade aging in ambient air with relative humidity of 60–80% for 165 h, indicative of the decomposition of perovskite. In sharp contrast, the perovskite films covered with the CrO_X buffer layer exhibited marginal degradation, indicating that the CrO_X buffer layer can effectively block the effect of water and oxygen on the perovskite film.

Encouraged by the promising performance of CrO_X-based single-junction perovskite solar cells, we integrated the perovskite top cell with a double-side textured silicon bottom cell to fabricate a monolithic tandem solar cell, as schematically presented in Fig. 5a. The tunnel oxide passivated contact (TOPCon) structure was utilized because of its superior thermal stability, relatively simple fabrication and the excellent surface passivation on the non-textured surface.⁴⁵ The cross-sectional and top view SEM images of the tandem solar cells are shown in Fig. S38, suggesting the conformal deposition of the perovskite film on the pyramid-shaped texture of the silicon substrate, endowing superior interfacial contact. The best performing tandem devices with an area of 1.0 cm² exhibit a PCE of 32.77%, a V_OC of 1.952 V a J_SC of 20.67 mA cm⁻² and an FF of 81.23% under reverse scan, featuring negligible hysteresis. The EQE spectra suggest a photogenerated current density of 20.61 mA cm⁻² for the perovskite top cell and 20.62 mA cm⁻² for the silicon bottom cell, which agrees well with the J_SC value obtained from J–V curves (Fig. 5c). Additionally, the steady-state PCE of 32.54% is recorded at the MPP of 1.66 V upon continuous 1-sun illumination for 400 s (Fig. 5d). Notably, one of the best performing tandem devices was measured by a third-party (Fig. S39). It displays a PCE of 32.51%, a V_OC of 1.9354 V and an FF of 80.55% under reverse scan. The impressive certified stabilized PCE of 32.51% is the highest value reported for monolithic perovskite/silicon tandems based on TOPCon (Fig. 5e and Table S3).

Fig. 5 The photovoltaic performance of monolithic perovskite/silicon tandem devices. (a) The device structure of monolithic perovskite/silicon tandem solar cells. (b) Typical J–V curves of the champion target device. The inserted table represents the corresponding parameters. (c) The EQE spectra and corresponding integrated J_SC. (d) Steady-state power output of target devices. (e) The V_OCversus PCE plot of the state-of-the-art performances of monolithic perovskite/silicon tandems based on the TOPCon reported in the literature. (f) The storage stability without encapsulation in N₂-filled glovebox at 25–35 °C. (g) The maximum power point tracking (MPPT) operational stability (1-sun illumination, at 25–35 °C, xenon lamp without ultraviolet filter, unencapsulated, N₂-flowed chamber, ISOS-L-1) of the target device. h. MPPT at 65 °C for three devices under identical conditions. Markers represent the normalized mean values, and the connecting traces indicate the SD range.

The improved carrier transport and energy-alignment are supposed to render the device more resilient to humidity and light exposure. We access the stability of tandem devices through recording their PCE variation. The target tandem device without encapsulation exhibits a decent storage stability (Fig. 5f). After being reserved in the N₂-filled glovebox for 1152 h, the PCE shows a slight degradation. For the MPP tracking operational stability, the tandem device maintains 82% of its initial PCE under the continuous 1-sun xenon lamp irradiation (without a UV filter) for 750 h. The subsequent performance decline beyond 750 h is attributed to the depletion of the N₂ atmosphere. Notably, the decreased stability can be gradually recovered to the initial state after re-introducing N₂, and the device stabilizes at its initial PCE following 1400 h of operation at MPPT. The reversible performance fluctuation observed during MPP tracking is primarily attributed to minor temperature variations within the test chamber. An interruption in the N₂ flow led to a rise in device temperature due to sustained illumination, which caused a temporary drop in efficiency. Upon resuming N₂ flow, the device cooled and its efficiency recovered to the baseline level. To confirm its superior stability, statistical analyses of the long-term stability at 65 °C for three devices are provided in Fig. 5h.

As can be seen, the tandem devices retain 97% ± 2.2% of their initial PCE after MPP tracking at 65 °C for 320 h. These results demonstrate the excellent operational stability of the CrO_X-based tandem solar cell.

Conclusions

In summary, we developed an efficient and robust ALD-free buffer layer using the thermally evaporated CrO_X film, which effectively blocks hole carriers, enhances interfacial adhesion, and prevents sputtering-induced damage. The CrO_X buffer layer exhibits high conductivity, optical transparency, and a deep VBM, thereby enabling ohmic contact formation, minimizing parasitic absorption, and facilitating energy-level alignment. It eliminates ambient exposure of the sensitive C₆₀, effectively suppresses oxygen permeation, and allows beneficial Cr diffusion into C₆₀, thereby leading to n-type doping, reducing trap density, and enhancing electron mobility. Furthermore, the formation of a Cr–O–In ionic interlocking interface with the IZO top electrode provides strong adhesion, suppresses delamination, and preserves the crystallinity and optoelectronic quality of the underlying layers. As a result, single-junction perovskite devices and monolithic perovskite/TOPCon tandems exhibited PCEs of 23.27% and 32.77%, respectively, alongside superior operational stability and environmental resilience. This multifunctional buffer layer thus offers a scalable and versatile strategy for advancing perovskite-based tandem photovoltaics.

Author contributions

Huan Li: conceptualization, investigation, data curation, and writing – original draft. Zhiqin Ying: conceptualization and supervision. Wenfeng Liu: supervision. Xin Li, Ziyu He, Haofan Ma and Yunyun Yu: data curation. Rui Li: visualization. Fanshu Kong, Meili Zhang, and Yan Zheng: investigation. Xi Yang: conceptualization, supervision, and writing – review and editing. Luyao Zheng, Jichun Ye and Yuheng Zeng: supervision.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.

Supplementary information (SI) is available. See DOI: https://doi.org/10.1039/d5ee07165h.

Acknowledgements

This work was supported by the National Key Research and Development Program of China (Grant No. 2024YFB3817304), National Natural Science Foundation of China (Grant No. U23A20354, 62574207, 52571233), Baima Lake Laboratory Joint Funds of the Zhejiang Provincial Natural Science Foundation of China (Grant No. LBMHD24E020002, LBMHQN26E020005), Key Research and Development Program of Zhejiang Province (Grant No. 2024C01092, 2025C01154), Zhejiang Provincial Natural Science Foundation of China (Grant No. LY24F040003, LZ26F040004, LQN26E020017, LQN26F040009, LQN26F040008), Key Research and Development Program of Ningbo (Grant No. 2023Z151), Ningbo Young Scientific and Technological Innovation Leading Talent Project (Grant No. 2024QL037, 2025QL033), and the China Postdoctoral Science Foundation (Grant No. 2025M770062).

References

1. G. Yang, C. Deng, C. Li, T. Zhu, D. Liu, Y. Bai, Q. Chen, J. Huang and G. Li, Nat. Photonics, 2025, 19(9), 913–924.
2. P. Zhu, C. Chen, J. Dai, Y. Zhang, R. Mao, S. Chen, J. Huang and J. Zhu, Adv. Mater., 2024, 36, e2307357.
3. Y. Xu, S. Wang, H. Liu and X. Li, Adv. Mater., 2024, 36, 2313080.
4. K. Zhao, Q. Liu, L. Yao, C. Değer, J. Shen, X. Zhang, P. Shi, Y. Tian, Y. Luo, J. Xu, J. Zhou, D. Jin, S. Wang, W. Fan, S. Zhang, S. Chu, X. Wang, L. Tian, R. Liu, L. Zhang, I. Yavuz, H.-f Wang, D. Yang, R. Wang and J. Xue, Nature, 2024, 632, 301–306.
5. D. Gao, B. Li, Q. Liu, C. Zhang, Z. Yu, S. Li, J. Gong, L. Qian, F. Vanin, K. Schutt, M. A. Davis, A. F. Palmstrom, S. P. Harvey, N. J. Long, J. M. Luther, X. C. Zeng and Z. Zhu, Science, 2024, 386, 187–192.
6. B. Yan, D. Zhang, R. Li, J. Wei, P. Hang, H. Xin, Z. Ni, M. Lei, D. Yang and X. Yu, Angew. Chem. Int. Ed., 2025, 64, e202509279.
7. Y. Luo, Y. Tian, K. Zhao, W. Mao, C. Liu, J. Shen, Z. Cheng, C. Değer, X. Miao, Z. Zhang, X. Sun, L. Yao, X. Zhang, P. Shi, D. Jin, J. Deng, M. Tian, I. Yavuz, N. Dong, R. Liu, R. Wang, D. Yang and J. Xue, Nat. Commun., 2025, 16, 4516.
8. H. Zhao, X. Zhang, K. Zhang, W. Zhang, R. Zhou, Y. Wei, J. Qu, Y. Chen, H. Li, X. Zong, S. Zhang, M. Liang, Y. Huang, H. Li, Y. Yang, W. Long, Y. Wang and S. Yang, Angew. Chem. Int. Ed., 2025, 64, e202504237.
9. Y. Chen, N. Yang, G. Zheng, F. Pei, W. Zhou, Y. Zhang, L. Li, Z. Huang, G. Liu, R. Yin, H. Zhou, C. Zhu, T. Song, C. Hu, D. Zheng, Y. Bai, Y. Duan, Y. Ye, Y. Wu and Q. Chen, Science, 2024, 385, 554–560.
10. M. Härtel, B. Li, S. Mariotti, P. Wagner, F. Ruske, S. Albrecht and B. Szyszka, Sol. Energy Mater. Sol. Cells, 2023, 252, 112180.
11. Z. Xiong, L. Wu, X. Zhou, S. Yang, Z. Liu, W. Liu, J. Zhao, W. Li, C. Yu and K. Yao, Small, 2023, 20, 2308024.
12. K. A. Bush, A. F. Palmstrom, Z. J. Yu, M. Boccard, R. Cheacharoen, J. P. Mailoa, D. P. McMeekin, R. L. Z. Hoye, C. D. Bailie, T. Leijtens, I. M. Peters, M. C. Minichetti, N. Rolston, R. Prasanna, S. Sofia, D. Harwood, W. Ma, F. Moghadam, H. J. Snaith, T. Buonassisi, Z. C. Holman, S. F. Bent and M. D. McGehee, Nat. Energy, 2017, 2, 17009.
13. M. De Bastiani, G. Armaroli, R. Jalmood, L. Ferlauto, X. Li, R. Tao, G. T. Harrison, M. K. Eswaran, R. Azmi, M. Babics, A. S. Subbiah, E. Aydin, T. G. Allen, C. Combe, T. Cramer, D. Baran, U. Schwingenschlögl, G. Lubineau, D. Cavalcoli and S. De Wolf, ACS Energy Lett., 2022, 7, 827–833.
14. M. Li, S. Johnson, L. Gil-Escrig, M. Sohmer, C. A. Figueroa Morales, H. Kim, S. Sidhik, A. Mohite, X. Gong, L. Etgar, H. J. Bolink, A. Palmstrom, M. D. McGehee and N. Rolston, Energy Adv., 2024, 3, 273–280.
15. A. A. Said, L. Mardegan, E. Ugur, Z. Song, P. Dally, B. Yildirim, D. S. Utomo, A. Razzaq, D. Turkay, C. M. Wolff, M. Morales-Masis, Y. Yan, H. J. Bolink, E. Aydin and S. De Wolf, ACS Appl. Energy Mater., 2025, 8, 13722–13731.
16. A. F. Palmstrom, J. A. Raiford, R. Prasanna, K. A. Bush, M. Sponseller, R. Cheacharoen, M. C. Minichetti, D. S. Bergsman, T. Leijtens, H. P. Wang, V. Bulović, M. D. McGehee and S. F. Bent, Adv. Energy Mater., 2018, 8, 1800591.
17. S. A. Johnson, K. P. White, J. Tong, S. You, A. Magomedov, B. W. Larson, D. Morales, R. Bramante, E. Dunphy, R. Tirawat, C. L. Perkins, J. Werner, G. Lahti, C. Velez, M. F. Toney, K. Zhu, M. D. McGehee, J. J. Berry and A. F. Palmstrom, Joule, 2023, 7, 2873–2893.
18. J. A. Raiford, C. C. Boyd, A. F. Palmstrom, E. J. Wolf, B. A. Fearon, J. J. Berry, M. D. McGehee and S. F. Bent, Adv. Energy Mater., 2019, 9, 1902353.
19. B. Yu, F. Tang, Y. Yang, J. Huang, S. Wu, F. Lu, W. Duan, A. Lambertz, K. Ding and Y. Mai, Adv. Mater., 2022, 35, 2202447.
20. H. Werner, T. Schedel-Niedrig, M. Wohlers, D. Herein, B. Herzog, R. Schlögl, M. Keil, A. M. Bradshaw and J. Kirschner, J. Chem. Soc., Faraday Trans., 1994, 90, 403–409.
21. F. Faupel, R. Willecke and A. Thran, Mater. Sci. Eng., R, 1998, 22, 1–55.
22. K. Lee, H. Song and J. T. Park, Acc. Chem. Res., 2003, 36, 78–86.
23. S. Nagao, T. Kurikawa, K. Miyajima, A. Nakajima and K. Kaya, J. Phys. Chem. A, 1998, 102, 4495–4500.
24. P. J. Fagan, J. C. Calabrese and B. Malone, Acc. Chem. Res., 2002, 25, 134–142.
25. N. J. Chou and C. H. Tang, J. Vac. Sci. Technol., A, 1984, 2, 751–755.
26. C. Cioarec, P. Melpignano, N. Gherardi, R. Clergereaux and C. Villeneuve, Langmuir, 2011, 27, 3611–3617.
27. M. Wang, Q. Tang, J. An, F. Xie, J. Chen, S. Zheng, K. Y. Wong, Q. Miao and J. Xu, ACS Appl. Mater. Interfaces, 2010, 2, 2699–2702.
28. S. Zheng, W. Li, T. Su, F. Xie, J. Chen, Z. Yang, Y. Zhang, S. Liu, M. P. Aldred, K. Y. Wong, J. Xu and Z. Chi, Sol. RRL, 2018, 2, 1700245.
29. M. Kaltenbrunner, G. Adam, E. D. Glowacki, M. Drack, R. Schwodiauer, L. Leonat, D. H. Apaydin, H. Groiss, M. C. Scharber, M. S. White, N. S. Sariciftci and S. Bauer, Nat. Mater., 2015, 14, 1032–1039.
30. W.-j Li, P. Wang, X.-X. Wang, J.-O. Wang, R. Wu, H.-J. Qian, K. Ibrahim, H.-Y. Li and H.-N. Li, Appl. Phys. Lett., 2014, 104, 191606.
31. G. Xie, H. Li, X. Wang, J. Fang, D. Lin, D. Wang, S. Li, S. He and L. Qiu, Adv. Funct. Mater., 2023, 33, 2308794.
32. J. Dong, J. Wu, J. Jia, X. He, Z. Lan, L. Fan, J. Lin and M. Huang, ChemSusChem, 2018, 11, 619–628.
33. N. Zhang, J. Sun, J. Sheng, W. Yang, X. Xue, L. Zhang, X. Sha, Z. Ying, C. Shou, X. Yang and J. Ye, Mater. Today Energy, 2023, 35, 101338.
34. F. Gaudette, S. Suresh, A. G. Evans, G. Dehm and M. Rühle, Acta Mater., 1997, 45, 3503–3513.
35. L. Zimmermann, D. Menzel, R. Gundermann, M. Simmonds, F. Scheler, T. Gries, E. Nandayapa, A. F. Castro Mendez, F. Mathies, A. Miaskiewicz, E. J. W. List-Kratochvil, P. Holzhey, A. Musiienko, F. Lang, L. Korte, E. Köhnen and S. Albrecht, Adv. Energy Mater., 2025, 15, 2501225.
36. N. Jiang and J. Silcox, Microsc. Microanal., 1999, 5, 166–167.
37. Y. Yuan, G. Yan, C. Dreessen and T. Kirchartz, Adv. Energy Mater., 2025, 15, 2403279.
38. M. Yuan, Y. Wang, Z. Liu, J. Zhang, S. Qin, M. Liu, Z. Chen, Y. Zhao, J. Li, X. Li, L. Meng and Y. Li, Nat. Commun., 2025, 16, 11330.
39. A. A. Grinberg and S. Luryi, J. Appl. Phys., 1987, 61, 1181–1189.
40. Q. Yang, W. Duan, A. Eberst, B. Klingebiel, Y. Wang, A. Kulkarni, A. Lambertz, K. Bittkau, Y. Zhang, S. Vitusevich, U. Rau, T. Kirchartz and K. Ding, J. Mater. Chem. A, 2024, 12, 14816–14827.
41. J. Wei, T. Huang, W. Peng, Y. He, W. Feng, Y. Hao, C. J. Brabec, Y. Mai and F. Guo, J. Mater. Chem. A, 2025, 13, 27101–27106.
42. R. A. Jagt, T. N. Huq, S. A. Hill, M. Thway, T. Liu, M. Napari, B. Roose, K. Gałkowski, W. Li, S. F. Lin, S. D. Stranks, J. L. MacManus-Driscoll and R. L. Z. Hoye, ACS Energy Lett., 2020, 5, 2456–2465.
43. A. Al-Ashouri, E. Kohnen, B. Li, A. Magomedov, H. Hempel, P. Caprioglio, J. A. Marquez, A. B. Morales Vilches, E. Kasparavicius, J. A. Smith, N. Phung, D. Menzel, M. Grischek, L. Kegelmann, D. Skroblin, C. Gollwitzer, T. Malinauskas, M. Jost, G. Matic, B. Rech, R. Schlatmann, M. Topic, L. Korte, A. Abate, B. Stannowski, D. Neher, M. Stolterfoht, T. Unold, V. Getautis and S. Albrecht, Science, 2020, 370, 1300–1309.
44. J. Y. Kim, S. H. Kim, H. H. Lee, K. Lee, W. Ma, X. Gong and A. J. Heeger, Adv. Mater., 2006, 18, 572–576.
45. X. Li, Z. Ying, J. Zheng, X. Wang, Y. Chen, M. Wu, C. Xiao, J. Sun, C. Shou, Z. Yang, Y. Zeng, X. Yang and J. Ye, Adv. Mater., 2023, 35, e2211962.

---