Low-temperature ALD-grown SnO_x interlayer for scalable and stable p–i–n perovskite solar cells and modules

Abstract

Inverted p–i–n perovskite solar cells (IPSCs) offer promise for next-generation photovoltaics. However, IPSCs utilizing solution-processed PC₆₁BM as the electron transport layer (ETL) remain less interface-optimized than conventional n–i–p configurations, restricting their efficiency, stability, and scalability. In this work, we introduce an ultrathin atomic-layer-deposited SnO_x (ALD-SnO_x) film, fabricated at a low temperature (80 °C), as a versatile interfacial modifier to address these shortcomings. This scalable, vapor-phase approach directly addresses the core instability in p–i–n architectures, effectively remedies morphological defects such as pinholes and phase segregation in PC₆₁BM, significantly enhancing interfacial contact and suppressing charge recombination. Consequently, the champion IPSC incorporating a 10 nm ALD-SnO_x interlayer yields a power conversion efficiency (PCE) of ∼19.2%, representing a remarkable 58% improvement over control devices (PCE ∼11.3%). The ALD-SnO_x interlayer effectively enhances moisture resistance, giving the IPSCs excellent environmental stability. Additionally, the redesigned IPSCs show scalability by effectively generating a large-area (∼12.1 cm²) mini-module with a high PCE (∼14.1%). These findings demonstrate the immense potential of this interfacial engineering approach for the commercial production of scalable, stable, and effective IPSCs.

---

Introduction

Perovskite solar cells (PSCs) have emerged as a transformative photovoltaic technology, achieving power conversion efficiencies (PCEs) exceeding 26.9%, surpassing conventional silicon-based solar cells.¹ However, despite their remarkable efficiency, poor scalability and long-term stability remain the key obstacles to their commercialization. To mitigate these challenges, researchers have explored several approaches, such as architectural optimization, additive engineering, interfacial passivation, and advanced encapsulation techniques.^2,3 Inverted p–i–n type perovskite solar cells (IPSCs) have unique advantages that make them attractive for industrial-scale production. Nevertheless, this architecture remains comparatively less optimized at critical interfaces than the conventional n–i–p configuration, limiting its efficiency and stability.⁴ Essentially, p–i–n device performance is governed by the interface between the perovskite absorber and the electron transport layer (ETL), which influences charge extraction dynamics and energy-level alignment.

Commonly employed ETL materials for IPSCs included fullerene derivatives, SnO_2, and TiO₂.^5,6 Among these options, [6,6]-phenyl-C₆₁-butyric acid methyl ester (PC₆₁BM) is preferred due to its optimal energy-level alignment with perovskite absorbers and dual functional capability, serving as (i) an efficient electron transport pathway and (ii) a protective barrier against ion migration between the perovskite layer and metal electrode.^7,8 However, PC₆₁BM is undesirable for large-scale manufacturing, where molecule aggregation impairs electron extraction efficiency and accelerates device degradation.⁹ Additionally, severe non-radiative recombination at the PC₆₁BM/electrode interface substantially compromises charge collection. To address these issues, researchers have developed various strategies, including dopant engineering and interface modification techniques, to enhance the morphology and charge transport properties of the PC₆₁BM layer to mitigate the inferior device performance and perovskite decomposition in IPSCs.^10,11

One of the most effective approaches involves incorporating an inorganic protective interlayer, such as tin oxide (SnO_x), atop PC₆₁BM to enhance charge transport, device performance, and environmental stability.^12,13 SnO_x is particularly promising as a buffer layer due to its low-temperature processability, high optical transparency, excellent carrier mobility, and robust stability.¹⁴ Among various deposition techniques, atomic layer deposition (ALD) has gained attention as a vapor-phase deposition technique capable of producing conformal, uniform SnO_x films with tunable stoichiometry and highly precise thickness in the atomic regime, rendering it ideal for industrial-scale production.¹⁵ The quality of atomic-layer-deposited SnO_x (ALD-SnO_x) films can be precisely controlled by adjusting key deposition parameters, including deposition duration, cycle number, and substrate temperature.¹⁶ Nevertheless, the incorporation of ALD-SnO_x into IPSCs presents notable challenges, primarily due to the thermal sensitivity of organic ETL, such as PC₆₁BM, and the perovskite film under typical ALD processing conditions. For instance, exposure to elevated temperatures within the ALD chamber can rapidly degrade these active layers, leading to poor device performance.¹⁶ Although strong oxidants, such as O₂ plasma or O_3, can facilitate the formation of high-quality SnO_x films, they often induce irreversible damage to underlying layers. In contrast, deionized water (H₂O) is a milder and more compatible oxidant for use in IPSC fabrication, offering a balance between SnO_x film quality and preservation of underlying layers.¹⁷

Most studies concerning ALD-SnO_x have concentrated on high-temperature deposition (>100 °C), often overlooking the thermal instability of perovskite materials and organic ETLs.¹⁸ It has been reported that ALD-SnO_x buffer layers deposited at 120 °C accelerated the decomposition of perovskite into PbI₂.¹⁹ To overcome this limitation, Andrea et al. investigated the deposition of SnO₂ buffer layers at 100 °C on organic ETLs (PC₆₁BM) pre-stacked over the perovskite film. Their findings revealed that the organic ETLs effectively shielded the underlying perovskite from thermal decomposition. However, during the ALD growth process, PC₆₁BM underwent involuntary chemical modification, which negatively impacted the device performance of IPSCs.²⁰ In another study, Chen et al. reduced the ALD deposition temperature to 85 °C by introducing an ALD-SnO_x layer atop the PC₆₁BM/BCP ETL bilayers. This modification significantly improved device performance, preserving more than 90% of the initial PCE after 600 hours under ambient conditions. However, the effectiveness of this architecture relies on a sufficiently thin BCP layer to allow efficient electron tunneling, an aspect that is difficult to realize using conventional solution-processing techniques.¹¹ These constraints underscore the critical need for a low-temperature ALD process capable of producing high-quality SnO_x film while preserving the integrity of the underlying perovskite and organic ETL layers to solve a fundamental roadblock to commercialization.

Furthermore, while the evaporated C₆₀/BCP bilayer remains the performance benchmark for p–i–n PSCs, achieving champion efficiencies >25%,⁶ its reliance on a slow deposition rate and aggregation during the evaporation process presents significant challenges for scalable, low-cost manufacturing. In contrast, solution-processable PC₆₁BM offers a more industrially compatible pathway but is plagued by morphological instability and inadequate moisture resistance, leading to faster degradation.⁹

To bridge this critical gap, we integrated a scalable, low thermal vapor-phase process to solve the fundamental incompatibility between the high-performance of ALD and sensitive organic/perovskite materials. We developed a low-temperature (∼80 °C) ALD process to deposit an ultrathin ALD-SnO_x buffer layer atop perovskite/PC₆₁BM stacks. By optimizing the ALD parameters at this reduced temperature, we successfully produced a high-quality, fully dense, and compact SnO_x buffer layer that effectively mitigates thermal degradation and enhances charge extraction. This approach not only improves interfacial contact but also remarkably enhances the long-term operational stability of the IPSC. As a result, the champion IPSC device achieved a PCE of 19.18% and retained ∼90–95% of its initial PCE after 3000 h of ambient stability testing. Furthermore, we successfully scaled up the IPSC fabrication to a mini-module using a low-cost mechanical scribing technique, achieving a PCE of ∼14.1% over ∼12.1 cm². This study highlights the potential of low-temperature, aqueous-phase ALD under vacuum conditions as a practical strategy for scalable production of high-performance, environmentally stable IPSC devices and modules.

Results and discussion

Atomic layer deposition (ALD) offers atomic-level control over film thickness, exceptional conformality, precise stoichiometric tuning, low-temperature processability, and scalability for large-scale manufacturing. These features are particularly suitable for depositing high-quality buffers or interfacial layers such as SnO_x in IPSC architectures. In this work, we employed a thermal ALD system (NexusBe Mini™ series) to deposit an ultrathin SnO_x buffer layer using tetrakis(dimethylamido)-tin (TDMASn) as the organometallic precursor and water vapor as the co-reactant. The ALD process is conducted by exposing the substrate surface to a sequential, self-limiting surface reaction carried out in a highly controlled atmosphere, enabling the deposition of a sub-monolayer of SnO_x with each ALD cycle. Fig. 1a schematically illustrates a single thermal ALD cycle for SnO_x film growth and the reaction mechanism between the precursors (TDMASn; H=O) involved.

Fig. 1 (a) Schematic representation of the growth process of the ALD SnO_x buffer and growth mechanism. (b) Growth rate and thickness of ALD-SnO_x films at different temperatures. (c) The thickness of ALD-SnO_x as a function of growth cycle number at 80 °C. (d) Refractive index versus wavelength. (e) Optical transmission spectra and Tauc plots (inset) of ALD-SnO_x films deposited at 80 °C with thicknesses of 10 nm and 20 nm.

Herein, a single cycle consists of four main stages: (1) precursor pulse, (2) first purge, (3) co-reactant pulse, and (4) final purge. In stage (1), the TDMASn precursor is pulsed into the reactor chamber for 1 s, where it chemisorbs onto the substrate surface (surface adsorption). Once the surface adsorption is achieved, the chamber is purged with inert nitrogen gas (N₂) for 15 s to remove unreacted precursor molecules (first purge). Subsequently, H₂O vapor is pulsed for 2 s into the reactor as the co-reactant, reacting with the available surface sites to form SnO_x film on the substrate (surface reaction). After the reaction is complete, the second N₂ purge lasting 15 s is executed to evacuate residual by-products and excess reactants from the chamber (final purge). Therefore, each complete ALD cycle was configured to last 33 s; this timing was optimized to ensure complete surface reactions while maintaining overall process efficiency. In particular, the extended 15 seconds N₂ purge durations were deliberately developed to preserve the self-limiting nature of the ALD process and suppress undesired chemical vapor deposition effects. Moreover, the use of water as a volatile and mild oxidant enables ALD at lower temperatures, significantly reducing residual contamination from organic ligands and resulting in a high-purity, contamination-free SnO_x film.¹⁹ By repeating the ALD cycle the appropriate number of times, the thickness of the resulting ALD-grown SnO_x (ALD-SnO_x) film can be precisely controlled. This confirms the efficacy of ALD in producing uniform, pinhole-free ALD-SnO_x buffer layers through a layer-by-layer growth mechanism in a low-temperature regime.

The optical properties of the ALD-SnO_x films were studied using spectroscopic ellipsometry and absorption spectroscopy. Fig. 1b shows the variations in growth rate and film thickness for ALD-SnO_x films deposited on Si wafers over 200 cycles at various deposition temperatures, as measured by ellipsometry. The deposition temperature significantly influenced the ALD-SnO_x film, with higher temperatures resulting in a linear reduction in both growth rate and film density. This may be due to incomplete surface reactions or condensation of the precursor species at elevated temperatures.²¹ This observation is well aligned with earlier studies that utilized the same TDMASn precursor and H₂O co-reactant, confirming slower film growth at high temperatures.²² Details of the ALD-SnO_x film thickness at each deposition temperature, used to determine the growth per cycle (GPC) of the ALD process, are summarized in Table S1.

To improve the performance and reliability of IPSCs, a low-temperature recipe for ALD deposition of SnO_x was required. Reducing the substrate temperature and deposition time of SnO_x film is essential to avoid perovskite decomposition and undesirable interfacial reactions within the ALD chamber, both of which can hinder the IPSC device performance.¹⁶ Given these considerations, the SnO_x buffer layer was deposited via ALD at a reduced temperature of 80 °C, with a varying number of ALD cycles. As displayed in Fig. 1c, the growth rate of SnO_x film exhibited a strong dependence on the number of ALD cycles at low temperature (80 °C). A gradual decline in growth rate was observed as the ALD cycle number increased until the deposition process reached saturation. The observed behavior may be attributed to the initially high density of reactive sites on the substrate, such as hydroxyl groups or defects, that interact readily with TDMASn precursors. During the early cycles, rapid ligand exchange and incomplete ligand removal due to insufficient thermal energy could result in a higher initial GPC, consistent with the findings reported by Richey et al.²³ Despite the early-stage variation, the SnO_x film thickness demonstrated a linear relationship with the number of ALD cycles, manifesting the layer-by-layer growth mechanism characteristic of ALD.²⁴ From the slope of the linear fit in Fig. 1c, the growth rate of the SnO_x film at 80 °C was 1.74 Å per cycle within the range of 60–300 cycles, which is relatively higher than the other reported GPC values at the same growth temperature using the same precursors.²⁵

To evaluate the optical properties of ALD-SnO_x films deposited at 80 °C, we first estimated the number of ALD cycles required to obtain film thicknesses of 10 nm and 20 nm on silicon wafers and glass substrates, based on the measured GPC. Then, we measured their refractive indices (RI) using spectroscopic ellipsometry, as shown in Fig. 1d. The results reveal that the RI of the ALD-SnO_x is strongly influenced by the number of cycles and, hence, by film thickness. At a wavelength of 633 nm, the RI values for the 10 nm and 20 nm films were approximately 1.73 and 1.83, respectively. These values are higher than those reported by Elam et al. for SnO_x films deposited via ALD at less than 100 °C, likely due to variations in film density and stoichiometry.²⁶

Compared to the 20 nm ALD-SnO_x films, the 10 nm film demonstrated significantly higher optical transmittance (∼91%) at a wavelength of 550 nm, while the 20 nm film showed a marginally lower value of ∼88%, as displayed in Fig. 1e. The high optical transparency in the UV-visible region highlights the compatibility of SnO_x as an efficient buffer layer in IPSCs, where optical clarity is essential for maintaining decent device performance and operational stability. The optical band gap (E_g) values, derived from Tauc plots (inset of Fig. 1e), were ∼3.87 eV and ∼3.71 eV for the 10 nm and 20 nm films, respectively. The slight decrease in E_g is attributable to the alterations in film stoichiometry, which governs the electronic band structure and optical characteristics of the SnO_x layers.²⁷

In addition to characterizing the standalone ALD-SnO_x films, further analysis was conducted to evaluate the optoelectronic properties of a 10 nm-thick ALD-SnO_x buffer layer deposited at 80 °C atop the PC₆₁BM ETL. This investigation aimed to verify the successful formation of the ALD-SnO_x layer and to assess any modifications in the optical properties of the underlying PC₆₁BM following the ALD process. For this purpose, individual PC₆₁BM, ALD-SnO_x, and bilayer PC₆₁BM/ALD-SnO_x films were sequentially prepared and characterized. As shown in Fig. S1a, the individual PC₆₁BM layer exhibited a transmittance at 550 nm of ∼78.2%. When SnO_x was deposited atop the PC₆₁BM layer, the resulting PC₆₁BM/ALD-SnO_x bilayer maintained high optical transparency, with only a marginal reduction in transmittance (T_550nm ∼77.1%). The minimal reduction in transmittance (ΔT < 1%) for the bilayer structure suggests negligible optical losses at the PC₆₁BM/ALD-SnO_x interface, which is critical for preserving photon harvesting efficiency and maintaining the overall device efficiency of IPSCs.¹¹ Moreover, the E_g values of the PC₆₁BM and PC₆₁BM/ALD-SnO_x films, as determined from Tauc plot analysis (Fig. S1b and S1c), were 1.89 eV and 2.03 eV, respectively. The observed increase in E_g for the bilayer structure is beneficial to an improved band alignment, which can facilitate more efficient charge extraction at the perovskite-ETL interface. Collectively, these findings justify the use of the ALD-SnO_x layer as a highly transparent and electronically compatible buffer interface in IPSC architectures.

To study the surface elemental composition of PC₆₁BM, ALD-SnO_x, and the bilayer PC₆₁BM/ALD-SnO_x films, X-ray photoelectron spectroscopy (XPS) was performed. The survey scan spectrum (Fig. S2) confirmed the presence of carbon (C) with minimal oxygen (O), nitrogen (N), and no detectable tin (Sn) in the pristine PC₆₁BM film. Contrarily, the SnO_x-containing samples were constituted of four primary elements, Sn, O, C, and N. For further analysis, the core-level spectra of the detected elements were deconvoluted, as illustrated in Fig. S3a–d. The high-resolution Sn 3d spectra (Fig. 2a) for both ALD-SnO_x and PC₆₁BM/ALD-SnO_x samples were deconvoluted into spin–orbit doublets corresponding to Sn 3d_3/2 and Sn 3d_5/2, respectively. The Sn 3d peaks for ALD-SnO_x (486.73 eV, 495.15 eV) and the bilayer (486.76 eV, 495.17 eV) match well with literature values, showing negligible chemical shifts (<0.03 eV), confirming minimal interfacial chemical modification during deposition on the PC₆₁BM layer.²⁸ The narrow Sn 3d_5/2 full width at half maximum (FWHM) in both samples (1.33, 1.37 eV) and the energy separation between the Sn 3d_5/2 and Sn 3d_3/2 peaks (8.4 eV) indicate a low defect density, minimal concentration of oxygen vacancies, and Sn⁴⁺ dominance.^28,29 As anticipated, the absence of Sn 3d peaks in the pristine PC₆₁BM film reaffirms its purely organic nature, as the fullerene derivative does not contain any Sn species.

Fig. 2 XPS core-level spectra of (a) Sn 3d, (b) O 1s, and (c) C 1s elements of PC₆₁BM, ALD-SnO_x, and PC₆₁BM/ALD-SnO_x. Secondary electron cut-off region and onset region from UPS spectra of (d) PC₆₁BM film and (e) ALD-SnO_x film. (f) Schematic energy band alignment of PC₆₁BM with an ALD-SnO_x buffer layer.

To gain deeper insight into the oxygen-related chemical environments, the O 1s core-level spectra of all three samples were analyzed (Fig. 2b). In the PC₆₁BM film, two minor O 1s peaks were identified at 533.4 eV and 532.1 eV, corresponding to ionized oxygen atoms in the single-bonded ether (C–O–C) and the double-bonded carbonyl (C=O) groups, respectively, which are characteristics of the fullerene side chains.³⁰ In both ALD-SnO_x and PC₆₁BM/ALD-SnO_x samples, the broad O 1s spectrum was deconvoluted into two distinct components: a dominant peak at 530.6 eV assigned to lattice oxygen in the Sn⁴⁺–O bonds, and a secondary peak at 532.3 eV corresponding to surface-adsorbed hydroxyl (–OH) groups and oxygen-related impurities. These components are typically observed in SnO_x films deposited at low temperatures.³¹ The corresponding atomic percentages for these components are summarized in Table S2. Interestingly, the O/Sn ratio increased from 2.10 in the standalone ALD-SnO_x sample to 2.18 in the PC₆₁BM/ALD-SnO_x bilayer. This slight increase is likely due to chemical interactions between TDMASn and ester functionalities on the PC₆₁BM surface, which may serve as a nucleation site during the ALD process, facilitating oxygen incorporation.²⁰

Fig. 2c presents a noticeable difference in the intensity of C 1s core-level spectra among the three samples. However, there is no measurable shift in binding energy for the C 1s spectrum of the PC₆₁BM layer before and after deposition of the low-temperature ALD-SnO_x, confirming that SnO_x deposition at 80 °C improves the electronic quality of the interface without causing chemical degradation or phase segregation of the PC₆₁BM layer. For more details, in the pristine PC₆₁BM sample, a prominent peak at 284.7 eV corresponding to C–C bonding was observed, indicating the hydrocarbon backbone of the fullerene structure. Two additional weaker peaks were observed at 286.0 eV and 288.6 eV, corresponding to C–O and C=O bonds, respectively. These features are characteristics of the ether and carbonyl functional groups present in the fullerene side chains and may also be associated with adventitious carbon contamination or partial surface oxidation due to air exposure.³⁰ Meanwhile, the C 1s spectra of the ALD-SnO_x-containing samples exhibit two primary peaks corresponding to C–C bonds (∼284.7 eV) and C–O bonds (∼286.0 eV), likely originating from residual carbon impurities or surface contamination (Fig. S3c). Furthermore, a weak N 1s signal is observed in both the ALD-SnO_x and PC₆₁BM/ALD-SnO_x bilayer samples (Fig. S3d). This nitrogen impurity is associated with incomplete removal of dimethylamino ligands (–N(CH₃)₂) derived from the TDMASn precursor. This finding is consistent with previous reports, which suggest that nitrogen impurities tend to increase at lower deposition temperatures due to reduced thermal energy, which limits ligand desorption.^32,33

In addition to XPS analysis, ultraviolet photoelectron spectroscopy (UPS) measurements were performed to investigate the interfacial energy band alignment between the PC₆₁BM ETL and ALD-SnO_x buffer layer. The UPS spectra for PC₆₁BM and ALD-SnO_x are presented in Fig. 2d and e, respectively. By analyzing the secondary electron cut-off and valence band regions, in conjunction with the E_g values derived from Tauc plots, the key energy band parameters for both samples were extracted using eqn (S1)–(S5), and are summarized in Table S3. For the PC₆₁BM film, the highest occupied molecular orbitals (HOMO) and lowest unoccupied molecular orbitals (LUMO) were −6.00 and −4.11 eV, respectively. Correspondingly, the valence band maximum (E_VBM) and conduction band minimum (E_CBM) were calculated to be −8.1 and −4.23 eV, respectively. These values confirm favorable energy level alignment between the ALD-SnO_x buffer layer and PC₆₁BM, allowing efficient electron extraction and transport from the perovskite absorber to the ETLs and onward to the Ag electrode. Moreover, the deeper E_VBM of the ALD-SnO_x (∼8.1 eV) functions as an effective hole-blocking layer in IPSCs. The overall energy level alignment and charge transfer dynamics across the multilayer structure are schematically illustrated in Fig. 2f, highlighting the essential role of the ALD-SnO_x buffer layer in optimizing interfacial energetics and enhancing photocurrent generation by mitigating charge recombination at the perovskite-ETL interface.

The X-ray diffraction (XRD) patterns in Fig. 3a exhibit no discernible sharp diffraction peaks of PC₆₁BM, ALD-SnO_x, and PC₆₁BM/ALD-SnO_x thin films, indicating their amorphous nature without the formation of new or separate phases after the deposition of ALD-SnO_x. Such amorphous characteristics are typical of organic semiconductors like PC₆₁BM and low-temperature ALD-grown metal oxides, consistent with previous studies.^5,34 Simultaneously, additional XRD measurements were conducted on PVK/PC₆₁BM/ALD-SnO_x (∼10 nm) multilayer stacks to assess the thermal stability of the triple-cation perovskite (PVK) layer upon ALD-SnO_x deposition at 80 °C. As depicted in Fig. 3b, the perovskite films capped with either PC₆₁BM or a PC₆₁BM/ALD-SnO_x bilayer displayed diffraction patterns nearly identical to those of the pristine perovskite film. All samples showed prominent crystallographic peaks corresponding to the cubic perovskite phase, with no detectable PbI₂ formation.³⁵ This finding confirms that ALD-SnO_x deposition at 80 °C does not degrade the underlying perovskite structure, which is crucial for preserving the structural integrity and device performance and stability of IPSCs.³⁶

Fig. 3 XRD spectra of (a) PC₆₁BM, ALD-SnO_x, and bilayer PC₆₁BM/ALD-SnO_x films and (b) PC₆₁BM and bilayer PC₆₁BM/ALD-SnO_x topping of perovskite films. (c) Optical absorption spectra and Tauc plot (inset) of perovskite and modified perovskite films. High-resolution FESEM images of (d) PVK, (e) PVK/PC₆₁BM, and (f) PVK/PC₆₁BM/ALD-SnO_x films. High-resolution 2D-AFM micrographs of (g) PVK, (h) PVK/PC₆₁BM, and (i) PVK/PC₆₁BM/ALD-SnO_x films.

To further investigate the thermal tolerance of the perovskite underlayer during ALD-SnO_x deposition, the crystallography of the PVK/PC₆₁BM/ALD-SnO_x stacks with the SnO_x buffer layer deposited at temperatures ranging from 60 °C to 130 °C was analyzed, and the related XRD patterns are shown in Fig. S4. Notably, no PbI₂-related diffraction peaks were detected at deposition temperatures up to 100 °C, demonstrating that the perovskite remains structurally intact within this low-temperature regime. However, at 100 °C, a faint PbI₂ peak was observed, and its intensity increased gradually with higher deposition temperatures. For samples exposed to an ALD process at ≥100 °C, the intensity of the signature perovskite peaks was quenched significantly. This behavior implies thermally induced decomposition of the perovskite underlayer, predominantly due to volatilization of organic cations such as methylammonium (MA⁺) and formamidinium cations (FA⁺) under prolonged thermal stress and low-vacuum conditions during the ALD process, which compromised the integrity of the perovskite lattice.³⁷ Collectively, XPS and XRD analyses confirm that the ALD-SnO_x buffer layer deposited at 80 °C offers the optimal balance between a high-quality SnO_x thin film and thermal degradation of the underlying perovskite layer. Fig. 3c displays the UV-vis absorption of three sample configurations: glass/PVK, glass/PVK/PC₆₁BM, and glass/PVK/PC₆₁BM/ALD-SnO_x. All perovskite films show similar absorption profiles with no significant shift in the absorption edge, suggesting that the ALD-SnO_x buffer layer at 80 °C does not impact the optoelectronic properties of the underlying perovskite layer. Additionally, the Tauc plot (inset) derived from the absorption spectra reveals an E_g of approximately 1.62 eV for the perovskite film.

As is widely recognized, the performance of high-efficiency IPSC devices is strongly dependent on the morphology and quality of the perovskite absorber layer. A high-quality perovskite film improves optical absorption and minimizes charge recombination losses. To examine the uniformity and surface features of perovskite films capped with either PC₆₁BM or PC₆₁BM/ALD-SnO_x layers, a field-emission scanning electron microscope (FE-SEM, NFEC-2021-01-267107) was employed. The top-view FE-SEM image in Fig. 3d depicts a pristine perovskite film with a compact, dense, and pinhole-free morphology with well-defined grain boundaries. Grain size analysis, as presented in the form of a histogram plot (Fig. S5), indicates an average grain size of (300 ± 38) nm for the perovskite film. The FE-SEM image in Fig. 3e illustrates that the spin-coated PC₆₁BM layer uniformly covers the entire perovskite surface. However, PC₆₁BM molecules tend to aggregate, leading to the formation of localized clusters and fine cracks (highlighted by the red circles), which compromise film uniformity. These morphological imperfections may hinder electron extraction efficiency and create direct contact between the perovskite and the top Ag electrode, forming shunt pathways. Consequently, shunt resistance R_sh decreases while leakage current increases, ultimately undermining the performance and stability of the IPSC devices.^11,38 To mitigate these issues, an ultrathin ALD-SnO_x buffer layer was deposited onto the PVK/PC₆₁BM stack at low temperature, serving as both a passivation and buffer layer to refine the morphology of PC₆₁BM. There is a substantial improvement in the PC₆₁BM morphology following ALD-SnO_x deposition (Fig. 3f), whereby the pinholes and phase segregation originally present in the PC₆₁BM layer were effectively eliminated, resulting in a conformal, pinhole-free, and smoother surface. This morphological refinement underscores the significance of the ALD-SnO_x buffer layer in reducing morphological defects of PC₆₁BM and promoting carrier transportation at the perovskite-ETL interface.

Two-dimensional atomic force microscopic (AFM) micrographs, acquired over a 2 × 2 µm² scanning area (Fig. 3g–i), provide further insight into the surface morphology evolution of the perovskite films upon sequential deposition of PC₆₁BM and ALD-SnO_x. The root-mean-square (RMS) surface roughness values consistently decreased from 10.6 nm for the pristine PVK film to 2.93 nm for the PVK/PC₆₁BM bilayer and further to 1.95 nm for the PVK/PC₆₁BM/ALD-SnO_x tri-layer. This progressive reduction in surface roughness demonstrates considerable morphological enhancement with each additional layer deposited. The ALD-SnO_x deposit noticeably smooths the surface and mitigates topographical inhomogeneities, improving the ohmic contact between the ETL and Ag electrode. This enhanced surface uniformity helps to reduce shunt pathways and enables more efficient charge carrier extraction and transfer.²⁵ Compared to solution-processed buffer layers, the ALD-SnO_x interlayer offers markedly superior conformity and ultra-low roughness, as evidenced in Fig. 3f and i. Such topography refinements not only optimize interfacial energetics for efficient charge transport but also enable scalable fabrication of defect-tolerant, large-area IPSC devices.

Next, we evaluated the influence of the ALD-SnO_x buffer layer on the photovoltaic performance of IPSCs with an architecture of ITO/NiO/MeO-2PACz/PVK/PC₆₁BM/Ag (Fig. 4a). The ALD-SnO_x films were deposited between the PC₆₁BM layer and Ag electrode at varying thicknesses and deposition temperatures. To identify the optimal thickness of the ALD-SnO_x layer for IPSC fabrication, a series of devices with ALD-SnO_x thicknesses of 0, 5, 10, 20, and 30 nm deposited at 80 °C were fabricated and evaluated. The results in Fig. 4b and Table S4 display the current density–voltage (J–V) characteristic curves of the devices under AM 1.5 G illumination and the extracted photovoltaic parameters, respectively. The results indicate that the control devices employing only PC₆₁BM ETL exhibited a modest PCE of ∼11.28%. Meanwhile, the IPSCs incorporating the ALD-SnO_x buffer layer demonstrated significant improvements across all key performance metrics. Among the tested thicknesses, the 10 nm ALD-SnO_x layer delivered the best performance, yielding a champion PCE of ∼19.18%. This optimization led to substantial enhancements in device parameters: V_oc improved from 0.855 V to 1.084 V, J_sc rose from 20.73 mA cm⁻² to 22.16 mA cm⁻², and the fill factor (FF) increased from ∼63% to ∼80% (Table S4). These simultaneous improvements imply more efficient charge extraction and suppression of non-radiative recombination. Consequently, the champion device showed ∼60% in PCE, highlighting the critical role of ALD-SnO_x layer integration in boosting the performance of IPSCs. Notably, the device with a suboptimal thickness of the ALD-SnO_x layer (5 nm) showed only modest improvements compared to the 10 nm counterpart (PCE ∼15.44%). Conversely, when the ALD-SnO_x thickness increased to 20 nm and 30 nm, the performance of the IPSCs deteriorated to 17.37% and 12.82%, respectively. As the ALD-SnO_x layer became excessively thick, it hindered the perovskite-ETL interfacial charge mobility.⁵ Therefore, the thickness of the ALD-SnO_x buffer layer was fixed at 10 nm for the fabrication of IPSCs based on various ALD-SnO_x deposition temperatures. To assess performance reproducibility, 20 IPSCs with varying thicknesses of ALD-SnO_x buffer layer were fabricated. The PCE histogram plots in Fig. 4c and S6 show that the IPSC devices incorporated with a 10 nm ALD-SnO_x layer delivered consistently greater performance than other counterparts.

Fig. 4 (a) A schematic of the PSC device. (b) J–V characteristic curves of IPSCs based on ALD-SnO_x buffer layer deposition at 80 °C with varying thickness (0, 5, 10, 20, 30 nm). (c) PCE distributions of PSCs based on a variation of ALD-SnO_x thickness deposited at 80 °C. (d) J–V curves under both reverse and forward scan directions for pristine PC₆₁BM, C₆₀/BCP, and the PC₆₁BM/ALD-SnO_x champion device using the ALD system at 80 °C. (e) TRPL spectra and PL spectra (inset) of the glass/perovskite, glass/PVK/PC₆₁BM, and glass/PVK/PC₆₁BM/ALD-SnO_x films, respectively. (f) PCE distributions of PSCs incorporating an ALD-SnO_x buffer layer at various deposition temperatures.

To provide a better comparison, we have fabricated and tested standard C₆₀/BCP ETL-based IPSCs and compared their performance with pristine PC₆₁BM and champion devices incorporating a low-temperature ALD-SnO_x interlayer on top of the PC₆₁BM layer. The J–V curves of the PC₆₁BM, C₆₀/BCP and PC₆₁BM/ALD-SnO_x bilayer devices under reverse and forward scan directions (Fig. 4d) and the summarized photovoltaic parameters (Table S5) reveal that the PC₆₁BM-based device with a 10 nm of ALD-SnO_x interlayer achieved the highest performance in all key photovoltaic parameters with significantly suppressed hysteresis compared to both pristine PC₆₁BM and the C₆₀/BCP ETL stack. Specifically, in the reverse-scan direction, our champion PC₆₁BM/ALD-SnO_x bilayer device achieved a comparable PCE of ∼19.2% with a low hysteresis index (HI) ∼0.06. In contrast, the standard C₆₀/BCP devices recorded only a PCE of ∼16.2% with HI ∼0.1 and the pristine PC61BM devices reached only ∼11.28% with HI ∼0.14 (eqn (S6)).The pronounced performance enhancement and hysteresis suppression are attributed to the synergistic effects of the low-temperature ALD-SnO_x interlayer, which provides conformal coverage, effective interfacial defect passivation, and improved the charge extraction. Importantly, these results demonstrate that a solution-processable PC₆₁BM-based ETL, enhanced with an ALD-SnO_x buffer layer, can address the interfacial challenges of the PC₆₁BM layer and surpass the performance of the C₆₀/BCP stack, emphasizing its potential for scalable manufacturing.

The charge transfer and recombination dynamics of charge carriers at the perovskite-ETL interface were studied using time-resolved photoluminescence (TRPL) and steady-state photoluminescence (PL) measurements. Although all perovskite films displayed nearly identical UV-vis absorption spectra (inset of Fig. 4e), greater than 85% quenching was observed when using PC₆₁BM as an ETL. This quenching effect improved further with a PC₆₁BM/ALD-SnO_x bilayer, indicating reduced defect states and more efficient charge transfer across the perovskite/ETL interface.³⁸ The TRPL decay profiles of all perovskite films (Fig. 4e) were fitted using a quadruple exponential decay model (eqn (S7)) to extract their photoluminescence decay lifetimes (τ_PL). As shown in Table S6, the bare perovskite film exhibited an average τ_PL of 185 ns. Introducing PC₆₁BM and PC₆₁BM/ALD-SnO_x bilayers reduced the τ_PL of the perovskite film to 77 and 16 ns, respectively. This substantial τ_PL reduction, along with the sharp PL quenching, indicates enhanced charge extraction and suppressed recombination at the perovskite/ETL interface. The notably shorter τ_PL in the PVK/PC₆₁BM/ALD-SnO_x structure reflects improved interface quality and minimal trap-assisted recombination. This improvement is further supported by the more favorable energy level alignment in the bilayer configuration. The 10 nm ALD-SnO_x layer showed ideal band alignment with the underlying layers, facilitating efficient electron extraction from the perovskite absorber to the electrode. Its deep valence band also effectively blocks hole leakage to the cathode. Overall, these results highlight how the ALD-SnO_x buffer layer facilitates carrier extraction, suppresses interfacial recombination, and minimizes current leakage to enhance IPSC device performance.

Although ALD-SnO_x deposition can boost the performance of IPSCs, the inherent thermal sensitivity of perovskite material necessitates careful control of the deposition temperature to avoid degradation of the underlying layers during ALD. To investigate the thermal limitations of the IPSC structure, a 10 nm-thick ALD-SnO_x buffer layer was deposited at various temperatures from 60 to 130 °C. The corresponding J–V curves are shown in Fig. S7. As illustrated in Fig. 4f, the ALD-SnO_x layer deposited at 80 °C yielded the best device performance, achieving an FF of ∼79% and a maximum PCE of ∼18.7%. This temperature appears to maintain an optimal balance, enabling the formation of a high-quality, defect-minimized ALD-SnO_x layer without compromising the structural and chemical integrity of the perovskite underlayers.³⁶ However, IPSC device performance began to deteriorate when the deposition temperature exceeded 100 °C. This degradation can be attributed to two primary factors: (i) degradation in ALD-SnO_x film quality and (ii) thermal instability of the perovskite absorber in a high-temperature regime. As the deposition temperature rises, the ALD growth rate decreases, possibly due to precursor condensation or insufficient surface reactivity that results in poor film composition, adhesion, and uniformity.²¹ Similarly, elevated deposition temperatures promote the degradation of volatile organic cations (MA⁺ and FA⁺) in the perovskite film, accelerating phase segregation and PbI₂ formation, as evidenced by XRD analysis. The PbI₂ impurity phase can induce additional charge transfer barriers that impede charge extraction at the perovskite/ETL interface. These findings align well with earlier studies demonstrating a massive decrease in the performance of the IPSC device when the ALD temperature exceeds 120 °C.⁷ Both degradation mechanisms can hinder the interfacial charge transfer process, increase recombination losses, and ultimately threaten IPSC device performance.³⁹

While the perovskite absorber layer showed no signs of degradation, the overall device performance was significantly reduced when using ALD-SnO_x deposited at 60 °C. This underperformance is directly attributed to the inferior electronic quality of the SnO_x layer formed at this lower temperature. Incomplete precursor reactions during the 60 °C ALD process lead to a porous film containing residual impurities. Such a poor-quality ETL results in insufficient charge extraction at the perovskite/ETL interface and enhances non-radiative recombination, collectively degrading device performance. This observation is consistent with the findings of Kuang et al., who reported similarly poor device performance for low-temperature (50 °C) ALD-SnO_x films, attributing it to high series resistance and inadequate electronic properties.³² These results further reinforce our conclusion that a dense, high-purity SnO_x film is essential for achieving efficient device operation. The photovoltaic parameters of IPSCs fabricated with ALD-SnO_x layers deposited at different deposition temperatures are summarized in Table S7.

Fig. 5a shows the external quantum efficiency (EQE) spectra and the corresponding integrated J_sc of IPSC devices fabricated with and without the ALD-SnO_x buffer layer. Both EQE spectra show a broad spectral response in the range of 300–850 nm, consistent with the signature light absorption characteristics of the perovskite layer. The IPSCs incorporated with ALD-SnO_x exhibited higher EQE values across the UV-visible region compared to the control device. The integrated J_sc values for the control and ALD-SnO_x-based devices were 20.56 and 22.06 mA cm⁻², respectively, correlating well with measured J_sc values obtained from the solar simulator. The EQE enhancement primarily arises from improved electron transport dynamics enabled by the ALD-SnO_x layer.

Fig. 5 (a) EQE and integrated J_sc plots of the PSCs w/o an ALD-SnO_x buffer layer and (b) SCLC model for defect density analysis of the only electron for devices w/o an ALD-SnO_x buffer layer (inset) image showing device structure of electron-only device. (c) Figure-of-Merit (FoM) plots of PSC devices based on a PC₆₁BM/ALD-SnO_x bilayer deposited at 100 °C and 80 °C.^{20,25,36,40,41} (d) Steady-state J_sc and PCE spectra at maximum power point (MPP) under continuous testing of the PC₆₁BM and ALD-SnO_x devices. Stability study of unencapsulated PSC devices stored in (e) ambient air and (f) an N₂-filled glovebox. (g) Normalized PCE of encapsulated PC₆₁BM and PC₆₁BM/ALD-SnO_x-based devices under a harsh environment (85 °C and RH of 85%). (h) Normalized PCE of unencapsulated PC₆₁BM and PC₆₁BM/ALD-SnO_x based devices under 100 mW cm⁻² white LED illumination (25–30 °C and RH of 40–50%).

To evaluate the defect passivation capability of ALD-SnO_x, space-charge-limited current (SCLC) measurements were performed using electron-only devices with a configuration of ITO/NP-SnO₂/perovskite/PC₆₁BM (w/o ALD-SnO_x)/Ag. The trap-filled limit voltage (V_TFL) and corresponding trap densities (N_trap) of the electron-only devices were determined using the Mott–Gurney law (eqn (S8)). The dark I–V curves in Fig. 5b show that the V_TFL values decreased from 0.33 V in the PC₆₁BM-only device to 0.26 V in the PC₆₁BM/ALD-SnO_x bilayer device, corresponding to N_trap values of 2.47 × 10¹⁵ and 3.14 ×10¹⁵ cm⁻³, respectively. The reduction in trap density indicates reduced trap-assisted Shockley-Read-Hall (SRH) mono-molecular recombination due to the defect-passivating effect of the ALD-SnO_x buffer layer on the PC₆₁BM ETL.⁵ This phenomenon is consistent with the PL quenching and TRPL lifetime results, which reflect facilitated interfacial and reduced SRH recombination that contributes to the enhanced V_oc. For comparison, Fig. 5c and Table S8 summarize the photovoltaic performances of recently reported IPSC devices employing a PC₆₁BM/ALD-SnO_x bilayer architecture.^{20,25,36,40,41} Among these, the IPSC incorporating an ALD-SnO_x buffer layer fabricated herein achieved the highest PCE, highlighting the efficacy of the mild-temperature ALD technique in optimizing the perovskite-ETL interface for better device performance.

The operational stability of the IPSCs was assessed by continuously recording the stabilized current density (J_stable) and efficiency (PCE_stable) at the maximum power point (MPP) under one-sun illumination. As shown in Fig. 5d, both control (PC₆₁BM-only) and ALD-SnO_x-modified devices maintained stable photocurrent and power output over 1000 s, confirming their excellent short-term operational stability. The ALD-SnO_x device exhibited higher J_stable and PCE_stable than the control device, which is ascribed to the compact, conformal, and pinhole-free nature of the ALD-SnO_x film. This high-quality buffer layer substantially reduces morphological defects within the PC₆₁BM layer and improves interfacial contact between the ETL and adjacent layers.

Furthermore, to evaluate the long-term stability of our IPSCs, we conducted a complete evaluation of both device types under varying conditions, including ambient air, an inert atmosphere, harsh environments, and exposure to illumination. For ambient stability testing, the IPSCs were aged under controlled conditions (18–23 °C; relative humidity, 20–40% RH) for 3000 h without encapsulation. The PCE of the PC₆₁BM devices decreased to less than 70% of their initial values, primarily due to moisture-induced perovskite degradation. Meanwhile, the PC₆₁BM/ALD-SnO_x devices retained greater than 90% of their initial PCE (Fig. 5e). This substantial improvement is attributed to the robust inorganic defect-free ALD-SnO_x layer achieved by our low-temperature ALD process, which acts as a barrier that inhibits oxygen and moisture permeation into the perovskite layer. Similarly, when stored under inert conditions within an N₂-filled glovebox (O₂ < 0.1 ppm; H₂O < 0.1 ppm), both types of devices showed improved stability. However, the PC₆₁BM/ALD-SnO_x devices showed the highest, retaining greater than 95% of their initial PCE, while the PC₆₁BM devices only showed ∼80% PCE retention (Fig. 5f). The mitigated PCE degradation in both cases under inert conditions indicates not only ALD-SnO_x as a highly efficient barrier layer but also that moisture and oxygen are the key factors driving IPSC performance decline. For the harsh conditions' stability testing, we subjected encapsulated devices to accelerated aging tests at 85 °C and 85% relative humidity (85/85) and monitored the normalized PCE over time (Fig. 5g). The PC₆₁BM/ALD-SnO_x-based device exhibits outstanding thermal and humidity stability, with a T₉₀ lifetime (time to 90% of initial PCE) exceeding 330 h, significantly surpassing the control device, which had a T₉₀ of just about 70 h. This substantial improvement is directly attributable to the ALD-SnO_x film, which acts as an effective encapsulation layer, reducing the permeation of atmospheric H₂O and O₂ into the perovskite layer. Additionally, photostability was tested under continuous illumination (100 mW cm⁻²) by a white LED in air for unencapsulated devices (Fig. 5h). The device with the PC₆₁BM/ALD-SnO_x bilayer exhibited remarkable photostability, retaining 93% of its initial PCE after 100 h of light exposure. This improved photostability can be attributed to the enhanced interface created by the ALD-SnO_x layer, which effectively passivates the top surface of the PC₆₁BM. Furthermore, this layer prevents metal diffusion from the electrode and establishes a more advantageous band alignment. Such alignment serves to mitigate charge accumulation and decrease non-radiative recombination at the critical interface between the perovskite and ETL. Overall, demonstrate that integration of the ALD-SnO_x buffer layer improves device performance and stability. These stability metrics meet essential benchmarks for commercial viability.

For practical applications in real life, large-area perovskite solar modules (PSMs) are preferred over single cells because of their ability to deliver significantly higher power output. Herein, PSMs were fabricated on 5 × 5 cm² pre-patterned ITO-coated glass substrates using pristine PC₆₁BM ETL and PC₆₁BM/ALD-SnO_x bilayer ETLs. Each module comprised six subcells connected in series through the conventional P1–P2–P3 scribing scheme. In this module architecture, P1 involves isolation of the bottom ITO electrode; P2 defines the active area of the patterned perovskite layer and establishes the electrical interconnections among the adjacent subcells; and P3 electrically isolates the adjacent cells following deposition of the top Ag electrode.⁴² The stacked functional layers (NiO/MeO-2PACz/perovskite/PC₆₁BM/ALD-SnO_x) were mechanically scribed using a metal needle at a moving speed of 1 mm s⁻¹ to create P2 and P3 grooves, as described in our earlier study.⁴³ The schematic layout of the P1–P2–P3 interconnection architecture in the PSMs is depicted in Fig. 6a.

Fig. 6 (a) Schematic representation of the six serially connected sub-cells in PSM structured via P1, P2, and P3 scripts. (b) An optical microscopic image of the P1, P2, and P3 grooves and dead areas on the fabricated PSM. (c) Real images of the PSM in various stages as seen, under lighting, and the final device after Ag deposition and P3 script. (d) J–V curves under both forward and reverse scan directions and (e) whisker plots of the PCE distributions of w/o ALD-SnO_x-based PSMs.

Moreover, the inter-electrode distances and line widths of the P1, P2, and P3 scribed grooves in the PSMs were measured via optical microscopy, as shown in Fig. 6b. Based on the optical images, the total active area of the PSM was calculated to be ∼12.13 cm², while the dead area was 1.87 cm², resulting in a high geometrical fill factor (photoactive area/total area) of∼86.6%. Fig. 6c illustrates photographic images of the mechanically scribed PSM (on a 5 × 5 cm ITO substrate) at different fabrication stages: (1) the well-defined grooves separating the active layer stacks following P2 scribing and before Ag deposition; (2) the P2-scribed PSM under illumination, illustrating the uniformity of the grooves and the perovskite layer; and (3) the fully assembled PSM following Ag electrode deposition and P3 scribing. These images confirm the uniform film formation and highly precise patterning in our PSMs fabrication process.

The J–V characteristics of the champion and control PSMs are presented in Fig. 6d and S8. The best-performing PSM incorporating the PC₆₁BM/ALD-SnO_x bilayer achieved a V_oc of 6.42 V, J_sc of 3.862 mA cm⁻², FF of 57.55%, and PCE_module of 14.12%, corresponding to a total power output of ∼171.3 mW. The PSM also exhibited a low hysteresis index of ∼0.07, indicating minimal charge accumulation at charge transport interfaces. In contrast, the control PSM based solely on PC₆₁BM ETL displayed markedly inferior performance, achieving a PCE_module of only 8.8%, approximately 40% lower than that of the ALD-SnO_x-modified counterpart. This notable performance deficit is attributed to the pronounced aggregation behavior of the PC₆₁BM ETL at larger fabrication scales, which was effectively suppressed through the incorporation of the conformal ALD-SnO_x buffer layer.⁹ Comprehensive photovoltaic parameters for both PSMs are summarized in Table S9. Additionally, the PSMs based on PC₆₁BM/ALD-SnO_x bilayers consistently outperformed their PC₆₁BM-only counterparts across multiple batches of fabrication (Fig. 6e), highlighting the improved device reproducibility. This improvement is ascribed to the improved homogeneity and reduced morphological defects in the PC₆₁BM layer enabled by deposition of ALD-SnO_x buffer. Overall, the excellent photovoltaic performance and stability, and batch-to-batch reproducibility of the PSMs herein demonstrate the scalability of the PC₆₁BM/ALD-SnO_x bilayer ETL design and the feasibility of integrating low-deposition (80 °C) ALD processing into perovskite photovoltaic fabrication. This approach holds a promising approach for the advancement of large-scale, high-efficiency IPSC device module manufacturing.

Experimental section

Atomic layer deposition (ALD) of SnO_x thin films

Atomically thin films of SnO_x f were deposited using a thermal ALD system (NexusBe Mini™ series) with tetrakis (dimethylamido)-Sn(IV) (TDMASn) as the metal–organic precursor and deionized water (H₂O) as the co-reactant. The TDMASn precursor was maintained at 40 °C in a bubbler-type canister, while the H₂O canister (self-type) remained at ambient temperature. The temperature of the substrate stage inside the deposition chamber varied between 80 and 150 °C during deposition. High-purity Argon (Ar, 99.99%) was used as the carrier and purge gas, with a fixed flow rate of 30 sccm. The chamber's working pressure stabilized at 500 mtorr. Each ALD cycle consisted of the following sequence: (i) TDMASn pulse for 1.0 s, (ii) Ar purge for 15.0 s, (iii) H₂O vapor pulse for 2.0 s, and (iv) Ar purge for 15 s. The gas flows were regulated by mass flow controllers (MFCs), while the temperatures of the substrate stage, precursor canisters, gas injector, and process lines were controlled by a computer-centralized system. Initially, SnO_x films were deposited on single-sided polished c-Si (100) wafers to determine the growth-per-cycle (GPC). Subsequently, SnO_x films with thicknesses precisely controlled through GPC calibration were deposited onto indium tin oxide (ITO)-coated glass substrates (25 × 25 mm²) to serve as a charge-selective electron transport interlayer in PSC device architectures. The basic characterizations were performed in the Centre for University-Wide Research Facilities (CURF) at Jeonbuk National University.

Fabrication of PSC devices

In this study, we fabricated inverted p–i–n structured PSC devices and modules with a device architecture of glass/ITO/NiO/MeO-2PACz/perovskite/PC₆₁BM/(with/without ALD-SnO_x)/Ag. Detailed fabrication procedures for both PSC devices and modules are provided in the experimental section of the SI.

Conclusion

In this work, we successfully demonstrate that integration of an ALD-processed SnO_x buffer layer significantly improves both the efficiency and stability of p–i–n PSCs. By optimizing the ALD process at a mild deposition temperature of 80 °C, the PCE of the IPSC device was boosted from 11.28% to 19.18%. This performance enhancement is attributed to the ability of the conformal ALD-SnO_x layer to address critical issues such as pinhole formation in the PC₆₁BM ETL and phase separation within the perovskite absorber induced by thermal stress. These improvements enable more efficient charge extraction and suppress non-radiative recombination losses. In addition to performance gains, the ALD-SnO_x layer also functions as a robust encapsulant that shields the perovskite underlayer from environmental degradation, allowing the IPSC devices to retain over 90% of their initial efficiency after more than 3000 h of storage in ambient air, underscoring the long-term device stability afforded by this interlayer. Importantly, we also fabricated a perovskite solar mini-module with an active area of ∼12.13 cm², achieving a PCE of 14.12%, validating the scalability and reproducibility of this device architecture. Overall, this study outlines a promising pathway for bridging the gap between high-performance lab-scale IPSCs and industrial-scale modules suitable for real-world deployment under concentrated sunlight conditions. The low-temperature ALD-SnO_x buffer layer introduced herein represents a critical step toward the commercial realization of perovskite photovoltaics.

Author contributions

Asmaa Mohamed: conceptualization, methodology, visualization, data curation, formal analysis, and writing the original draft. Hock Beng Lee: methodology, investigation, and writing – review and editing. Vinayak Vitthal Satale: formal analysis, review, and editing. Keum-Jin Ko: formal analysis. Barkha Tyagi and Do-Hyung Kim: methodology. Jae-Wook Kang: visualization, supervision, writing – review and editing, project administration, and funding acquisition.

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

The supporting information (SI) contains supplementary figures (Fig. S1–S8), tables (Table S1–S9), and equations (eqn S1–S8). Supplementary materials cover UV-vis spectroscopy, XPS, XRD, tables, equations, and PSC device performance. See DOI: https://doi.org/10.1039/d5se01332a.

Acknowledgements

This work was supported by National University Development Project at Jeonbuk National University in 2024. The basic characterizations were performed in the Centre for University-Wide Research Facilities (CURF) at Jeonbuk National University.

References

1. M. A. Green, E. D. Dunlop, M. Yoshita, N. Kopidakis, K. Bothe, G. Siefer, X. J. Hao and J. Y. Jiang, Prog. Photovoltaics, 2025, 33, 795–810.
2. C. Yang, W. Hu, J. Liu, C. Han, Q. Gao, A. Mei, Y. Zhou, F. Guo and H. Han, Light:Sci. Appl., 2024, 13, 227.
3. T. Danielyan, A. Asatryan, A. Sahakyan and H. Khachatryan, J. Power Sources, 2025, 630, 235995.
4. B. Li and W. Zhang, Commun. Mater., 2022, 3, 65.
5. A. Lan, H. Lu, B. Huang, F. Chen, Z. Chen, J. Wang, L. Li and H. Do, ACS Appl. Mater. Interfaces, 2024, 16, 64825–64833.
6. S. W. Liu, V. P. Biju, Y. B. Qi, W. Chen and Z. H. Liu, NPG Asia Mater., 2023, 15, 27.
7. E. M. Younes, A. Gurung, B. Bahrami, E. M. El-Maghraby and Q. Qiao, Org. Electron., 2022, 100, 106391.
8. X. Zhu, X. Zhao, L. Li, Y. Peng, W. Wei, X. Zhang, M. Su, Y. Wang, Z. Chen and W. Sun, Energy Power Eng., 2020, 12, 257–272.
9. C. Gong, H. Li, Z. Xu, Y. Li, H. Wang, Q. Zhuang, A. Wang, Z. Li, Z. Guo, C. Zhang, B. Wang, X. Li and Z. Zang, Nat. Commun., 2024, 15, 9154.
10. J. Yang, X. Luo, Y. Zhou, Y. Li, Q. Qiu and T. Xie, Int. J. Mol. Sci., 2022, 23, 11792.
11. B. B. Chen, P. Y. Wang, N. Y. Ren, R. J. Li, Y. Zhao and X. D. Zhang, J. Semicond., 2022, 43, 052201.
12. Y. Zhang, Y. Yang, M. T. Mbumba, M. W. Akram, E. K. Rop, L. Bai and M. Guli, Sol. RRL, 2022, 6, 2200823.
13. S. Bishnoi, R. Datt, S. Arya, S. Gupta, R. Gupta, W. C. Tsoi, S. N. Sharma, S. P. Patole and V. Gupta, Adv. Mater. Interfaces, 2022, 9, 2270109.
14. G. K. Dalapati, H. Sharma, A. Guchhait, N. Chakrabarty, P. Bamola, Q. Liu, G. Saianand, A. M. Sai Krishna, S. Mukhopadhyay, A. Dey, T. K. S. Wong, S. Zhuk, S. Ghosh, S. Chakrabortty, C. Mahata, S. Biring, A. Kumar, C. S. Ribeiro, S. Ramakrishna, A. K. Chakraborty, S. Krishnamurthy, P. Sonar and M. Sharma, J. Mater. Chem. A, 2021, 9, 16621–16684.
15. C. Altinkaya, E. Aydin, E. Ugur, F. H. Isikgor, A. S. Subbiah, M. De Bastiani, J. Liu, A. Babayigit, T. G. Allen, F. Laquai, A. Yildiz and S. De Wolf, Adv. Mater., 2021, 33, 2005504.
16. H. H. Park and D. J. Fermin, Nanomaterials, 2023, 13, 3112.
17. W. Zhao and Y. Duan, Adv. Photonics Res., 2021, 2, 2100011.
18. V. Zardetto, B. L. Williams, A. Perrotta, F. Di Giacomo, M. A. Verheijen, R. Andriessen, W. M. M. Kessels and M. Creatore, Sustainable Energy Fuels, 2017, 1, 30–55.
19. J. Li, Z. Xing, D. Li, Y. Wang, X. Hu, T. Hu and Y. Chen, Small, 2024, 20, 2305732.
20. A. E. A. Bracesco, J. van Himste, W. M. M. Kessels, V. Zardetto and M. Creatore, ACS Appl. Mater. Interfaces, 2024, 16, 59468–59476.
21. B. H. Yu, J. W. Zhang, Y. Z. Yang, D. S. Yu, Y. H. Mai and X. D. Chen, Energy Mater., 2024, 4, 400045.
22. D. W. Choi, W. J. Maeng and J. S. Park, Appl. Surf. Sci., 2014, 313, 585–590.
23. N. E. Richey, C. de Paula and S. F. Bent, J. Chem. Phys., 2020, 152, 040902.
24. A. Hultqvist, T. J. Jacobsson, S. Svanstrom, M. Edoff, U. B. Cappel, H. Rensmo, E. M. J. Johansson, G. Boschloo and T. Torndahl, ACS Appl. Energy Mater., 2021, 4, 510–522.
25. W. Wang, Z. Z. Yang, J. X. Ding, J. Q. Kong and X. Li, Sol. Energy Mater. Sol. Cells, 2022, 238, 111609.
26. J. W. Elam, D. A. Baker, A. J. Hryn, A. B. F. Martinson, M. J. Pellin and J. T. Hupp, J. Vac. Sci. Technol., A, 2008, 26, 244–252.
27. Y. Zakaria, B. Aissa, T. Fix, S. Ahzi, A. Samara, S. Mansour and A. Slaoui, Sci. Rep., 2022, 12, 15294.
28. L. Mai, D. Zanders, E. Subasi, E. Ciftyurek, C. Hoppe, D. Rogalla, W. Gilbert, T. L. Arcos, K. Schierbaum, G. Grundmeier, C. Bock and A. Devi, ACS Appl. Mater. Interfaces, 2019, 11, 3169–3180.
29. Y. Y. Cheng, H. Xie, F. L. Yu, J. L. Zhang, Y. F. Wang, X. M. Luo, B. Y. Shi and B. Y. Liu, Ionics, 2021, 27, 4143–4151.
30. K. Y. Chiu, S. H. Chang, W. C. Huang, H. M. Cheng, H. Shaw, S. C. Yeh, C. T. Chen, Y. O. Su, S. H. Chen and C. G. Wu, Nanotechnology, 2018, 29, 305701.
31. N. Ren, C. Zhu, R. Li, S. Mazumdar, C. Sun, B. Chen, Q. Xu, P. Wang, B. Shi, Q. Huang, S. Xu, T. Li, Y. Zhao and X. Zhang, Appl. Phys. Lett., 2022, 121, 033502.
32. Y. Kuang, V. Zardetto, R. van Gils, S. Karwal, D. Koushik, M. A. Verheijen, L. E. Black, C. Weijtens, S. Veenstra, R. Andriessen, W. M. M. Kessels and M. Creatore, ACS Appl. Mater. Interfaces, 2018, 10, 30367–30378.
33. A. J. Mackus, C. MacIsaac, W. H. Kim and S. F. Bent, J. Chem. Phys., 2017, 146, 052802.
34. Y. Zhong, K. Suzuki, D. Inoue, D. Hashizume, S. Izawa, K. Hashimoto, T. Koganezawa and K. Tajima, J. Mater. Chem. A, 2016, 4, 3335–3341.
35. V. V. Satale, H. B. Lee, B. Tyagi, M. M. Ovhal, S. Chowdhury, A. Mohamed, D. Kim and J. W. Kang, Chem. Eng. J., 2024, 493, 152541.
36. N. Mallik, J. Hajhemati, M. Frégnaux, D. Coutancier, A. Toby, S.-T. Zhang, C. Hartmann, E. Hüsam, A. Saleh, T. Vincent, O. Fournier, R. G. Wilks, D. Aureau, R. Félix, N. Schneider, M. Bär and P. Schulz, Nano Energy, 2024, 126, 109582.
37. H. H. Park, Nanomaterials, 2021, 11, 1–21.
38. T. C. Chang, C. Y. Liao, C. T. Lee and H. Y. Lee, Materials, 2023, 16, 5061.
39. M. T. Mbumba, D. M. Malouangou, J. M. Tsiba, L. Y. Bai, Y. F. Yang and M. Guli, Sol. Energy, 2021, 230, 954–978.
40. B. Yu, F. Tang, Y. Yang, J. Huang, S. Wu, F. Lu, W. Duan, A. Lambertz, K. Ding and Y. Mai, Adv. Mater., 2023, 35, e2202447.
41. K. O. Brinkmann, J. Zhao, N. Pourdavoud, T. Becker, T. Hu, S. Olthof, K. Meerholz, L. Hoffmann, T. Gahlmann, R. Heiderhoff, M. F. Oszajca, N. A. Luechinger, D. Rogalla, Y. Chen, B. Cheng and T. Riedl, Nat. Commun., 2017, 8, 13938.
42. B. Tyagi, N. Kumar, H. B. Lee, M. M. Ovhal, V. V. Satale, A. Mohamed, D. H. Kim and J. W. Kang, Small Methods, 2024, 8, e2300237.
43. H. B. Lee, A. Mohamed, N. Kumar, N. H. Zain Karimy, V. V. Satale, B. Tyagi, D. H. Kim and J. W. Kang, Small Methods, 2025, 9, e2400850.

---