Surface Engineering-Induced 2D/3D Interfacial Reconstruction for Stable Low-Light Perovskite Solar Cells

Abstract

Interfacial defects and ion migration critically limit the efficiency and operational stability of perovskite solar cells (PSCs), particularly under low-light conditions where interfacial recombination dominates. Herein, a postsurface treatment using the secondary amine dibutylammonium bromide (DBABr) is introduced to induce controlled 2D/3D interfacial reconstruction through the formation of a thin quasi-two-dimensional perovskite layer on a three-dimensional absorber. This dimensionally reconstructed interface effectively passivates surface defects, suppresses ion migration, and improves energy-level alignment without hindering charge extraction while remaining compatible with both organic and inorganic hole-transport layers (HTL). Under standard one-sun illumination (AM 1.5G), DBABr-treated devices exhibit a significant enhancement in power conversion efficiency (PCE), increasing from 13.82% to 15.52% relative to the control. Furthermore, the reconstructed interface enables efficient indoor energy harvesting, delivering a PCE of 30.57% for 0.09 cm² active-area devices and improving the PCE from 25.64% to 27.77% for 1 cm² devices under 1000 lx LED illumination, demonstrating scalability across device areas. Benefiting from the combined effects of the hydrophobic 2D surface layer and the carbon electrode, the devices retain approximately 90% of their initial efficiency after 1000 h of operation under ambient conditions. These results establish interfacial dimensional reconstruction via secondary-amine surface engineering as an effective strategy to simultaneously enhance efficiency, low-light performance, and operational durability of carbon-based PSCs, contributing to sustainable photovoltaic technologies aligned with Sustainable Development Goal 7 (affordable and clean energy).

1. Introduction

Organic inorganic perovskite solar cells (PSCs) have undergone continuous and rapid development in recent years, achieving a power conversion efficiency (PCE) of 26.95% for a single junction device, while carbon-based perovskite solar cells (C-PSCs) have recently reached a PCE of 24.84%. This remarkable performance arises from their tunable bandgap, high absorption coefficient, mechanical flexibility, long carrier diffusion lengths, high charge-carrier mobility, low exciton binding energy and low-cost solution processing. , These outstanding properties have attracted significant attention, establishing PSCs as a front-running photovoltaic technology for next-generation applications. Despite this progress, solution-processed polycrystalline perovskite films are often associated with intrinsic bulk and surface defects, which cause energy losses and limit performance to levels below the Shockley–Queisser theoretical limit. , Moreover, the redistribution and migration of charged point defects and ions contribute to the operational instability of PSCs, while oxygen and moisture can penetrate the perovskite film through the defect sites, compromising the device stability under ambient conditions. Beyond outdoor photovoltaics, PSCs have emerged as promising power sources for low-light energy harvesting, where artificial light typically operates at illumination level of 200–1000 lx, enabling self-powered sensors and Internet of Things (IoT) devices.

Defects are prone to facilitating trap-assisted recombination. Structural imperfections and trap states generated during the rapid solution processing of perovskites are major sources of nonradiative carrier recombination. The predominant defects are found at the surface and along grain boundaries, with defect densities approximately 2 orders of magnitude higher than those in the bulk perovskite film. Recombination occurs both at interfaces and within the perovskite bulk; however, the dominant carrier loss mechanism arises from interfacial recombination at the perovskite/transport layer interfaces. Therefore, passivating these defects is crucial for achieving high-performance PSCs.

Surface passivation, particularly strategies based on forming 2D/3D perovskite heterojunctions, has proven effective in reducing defects and suppressing nonradiative recombination. , Incorporating a small amount of hydrophobic, larger cations (arylammonium or alkyl) into the A-site of a 3D ABX₃ perovskite lattice can induce dimensional reconstruction, transforming the surface into a 2D or quasi-2D layered structure via dimensional engineering, , resulting in remarkable structural and environmental stability while maintaining high efficiency. Moreover, the chemical versatility of 2D perovskites allows for fine-tuning of optoelectronic properties through spacer-cation design, enabling improved photoresponse and enhanced moisture resistance by selecting appropriate ammonium spacers. − When operating under low-light illumination, the markedly reduced photon flux shifts the performance limitation from bulk absorption to interface recombination processes. Accordingly, precise control of the perovskite surface and interface becomes essential for efficient indoor light energy harvesting.

Extensive studies have investigated several spacer cations, primarily primary amines, including octylammonium (OA), butylammonium (BA), and phenethylammonium (PEA). Liu et al. demonstrated that the formation of 2D/3D heterostructures is primarily driven by interactions between [PbI₆]^4– octahedra and ammonium groups, whereas long-term stability depends on interactions between the entire spacer cations and the [PbI₆]^4– framework. Caiazzo et al. introduced 2H-benzotriazol-2-ylethylammonium bromide to form a nonuniform 2D perovskite layer, which effectively passivated defects, enhanced charge extraction, and reduced nonradiative recombination. n-Butylammonium has been employed to achieve an impressive open-circuit voltage (V _OC) of 1.31 V for a bulk perovskite absorber with a bandgap of 1.72 eV. Pei et al. utilized guanidinium chloride and oleylammonium iodide to assemble a binary 2D structure comprising Ruddlesden–Popper and alternating-cation interlayer phases, which proved beneficial for both tandem devices and single-junction PSCs. Azmi et al. reported exceptional thermal stability using oleylammonium iodide.

Exploration of secondary amines for this purpose is required as their use offers significant potential for fine-tuning material properties through molecular engineering, as they enable dual functionalization, in contrast to the single functionalization provided by primary ammonium cations. In addition, their fewer active hydrogens on the amino group can reduce hydrogen bonding between the ammonium in the perovskite active layer and moisture in the ambient environment, thereby enhancing environmental stability. Cui et al. employed benzothiophene-based passivators bearing primary and secondary ammonium cations to investigate structure-passivation relationships. They found that the additional hydrophobic methyl group on the secondary ammonium cation modifies the electron density of the ammonium unit, which further influences passivating hydrogen bonds and electrostatic interactions, leading to more effective defect passivation and improved device performance. Widely used primary ammonium spacers such as BA⁺ and PEA⁺ exhibit intrinsic chemical instability, as they can undergo deprotonation under thermal and illumination stress, leading to degradation reactions with FA-based perovskites and reduced device stability. Therefore, it is imperative to investigate the formation of 2D/3D heterostructures induced by secondary amines and to evaluate their impact on both outdoor and indoor photovoltaic performance. Under the orders-of-magnitude lower irradiance associated with indoor conditions, trap-assisted recombination and interfacial energetics dominate device operation, making interfacial reconstruction essential for efficient indoor energy harvesting.

In this work, we report a surface engineering approach based on posttreatment with the secondary amine dibutylammonium bromide (DBABr, C₈H₂₀BrN) to induce the formation of a thin 2D Ruddlesden–Popper perovskite layer at the perovskite/HTL interface in C-PSCs. This surface engineering effectively passivates interfacial defects, suppresses ion migration, and improves energy-level alignment while preserving efficient carrier transport. Optimization of the DBABr treatment increases grain size, reduces grain-boundary density, and minimizes carrier trapping and nonradiative recombination. Consequently, DBABr-treated devices achieve a PCE of 15.52%, representing a 12% improvement over the control device, and deliver an indoor power output of 111.9 μW cm^–2 with a PCE of 30.57%. To examine scalability beyond small-area devices, this surface engineering strategy was applied to a large-area configuration comprising four individual 1 cm² subcells fabricated on the same substrate. This multicell design allows a direct evaluation of device uniformity and reproducibility over a practically relevant area while preserving the original device architecture. The DBABr-induced interfacial modification maintains its performance benefits across all subcells, indicating that this surface engineering approach is compatible with scalable fabrication. Combined with improved stability under various operational conditions, these results underscore the potential of DBABr-based surface engineering for stable and efficient C-PSCs intended for indoor and outdoor applications.

2. Experimental Section

2.1. Methods

First, a 2D precursor solution was prepared by dissolving DBABr and PbI₂ in a 2:1 molar ratio to form (DBA)₂PbI₂Br₂. The solution was stirred at 50 °C, then spin-coated onto a glass substrate and annealed at 130 °C for 10 min. Subsequently, equal volumes of the (DBA)₂PbI₂Br₂ solution and the Cs_0.17FA_0.83Pb­(I_0.83Br_0.17)₃ perovskite precursors were mixed, spin-coated onto a glass substrate, and annealed at 130 °C for 15 min for analysis. After that, various concentrations of DBABr dissolved in IPA were spin-coated onto the perovskite film surfaces to modify the perovskite/HTL interface in C-PSCs. Furthermore, different DBABr concentrations were examined to enhance the overall quality of the perovskite films.

2.2. Materials

Commercial TiO₂ nanoparticles (anatase phase, 12 nm) were procured from Nanografi Nano Technology, processed, and used as the electron transport layer (ETL). Similarly, spiro-OMeTAD obtained from Sigma-Aldrich was modified and employed as the hole transport layer (HTL). Other reagents-including dimethyl sulfoxide (DMSO, 99.9%, anhydrous), lead bromide (PbBr₂, 99.999%), cesium iodide (CsI, 99.9%), chlorobenzene (99.8%, anhydrous), ethanol, N,N-dimethylformamide (DMF, 99.8%, anhydrous), isopropanol (IPA, 99.5%, anhydrous), 4-tert-butylpyridine (4TBP, 96%), acetonitrile (anhydrous, 99.8%), and anisole (99.7%) were purchased from Sigma-Aldrich. Lead iodide (PbI₂, 99.9985%) and lithium bis­(trifluoromethanesulfonyl)­imide (LiTFSI) were obtained from Ossila, and formamidinium iodide (FAI, 99.99%) and dibutylammonium bromide (DBABr) were supplied by Greatcell Solar Materials, while diethyl sulfide (DES, 96.0%) and copper­(I) thiocyanate were purchased from Ossila.

2.3. Preparation of ETL Precursor Solutions

TiO₂ nanoparticle precursor solutions were prepared by diluting them in ethanol at a 1:18 volume ratio and stirring continuously for 24 h before use.

2.4. Preparation of Perovskite Precursor Solutions

The perovskite precursor solution was prepared inside a nitrogen-filled glovebox, incorporating a slight excess of PbI₂, as previously reported to enhance device performance. The solution contained mixed cations (FAI, CsI, and PbBr₂) with 4% excess PbI₂, dissolved in a 4:1 solvent mixture of DMF and DMSO to yield a Cs_0.17FA_0.83Pb­(I_0.83Br_0.17)₃ perovskite precursor. The mixture was stirred for 48 h and subsequently filtered through a 0.22 μm PTFE filter before use.

2.5. Preparation of DBABr Solutions

Solutions of DBABr at different concentrations (0.75, 2, and 6 mg mL^–1) were prepared by dissolving the compound in IPA and stirring at room temperature for 2 h before use.

2.6. Preparation of HTL Precursor Solutions

Spiro-OMeTAD (65.07 mg) was dissolved in 900 μL of chlorobenzene, followed by the addition of 15.75 μL of Li-TFSI (prepared by dissolving 520 mg of Li-TFSI in 1 mL of acetonitrile) and 25.92 μL of 4-tert-butylpyridine. The resulting solution was stirred for 2 h before use. For the alternative HTL, 35 mg of CuSCN was dissolved in 1 mL of DES and stirred at room temperature for 2 h to form a transparent precursor solution.

2.7. Device Fabrication

FTO glass substrates (sheet resistance < 15 Ω/sq) were laser-patterned and sequentially cleaned by ultrasonication in detergent, deionized water, acetone, ethanol, and isopropanol for 15 min each. After air-drying, the substrates were heated at 135 °C for 15 min to remove organic residues, followed by a 15 min ultraviolet ozone (UVO) treatment to enhance hydrophilicity. TiO₂ nanoparticles were spin-coated at 3000 rpm for 40 s and annealed in air at 165 °C for 30 min. An additional 5 min UVO treatment was applied to further improve surface wetting. Cs_0.17FA_0.83Pb­(I_0.83Br_0.17)₃ perovskite films were deposited using a two-step spin-coating procedure (1000 rpm for 13 s, followed by 4000 rpm for 30 s), with 180 μL of anisole antisolvent dropped during the final 5 s, and subsequently annealed at 130 °C for 15 min. , For DBABr -passivated devices, a DBABr solution in IPA was spin-coated onto the annealed perovskite at 4000 rpm for 30 s and heated at 100 °C for 5 min. Previous studies have confirmed that IPA alone does not alter perovskite morphology, and the annealing temperature for Cs_0.17FA_0.83Pb­(I_0.83Br_0.17)₃ was optimized in prior work. Subsequent fabrication followed the same procedures as control devices. The HTL was spin-coated at 4000 rpm for 30 s after cooling to room temperature. All layers were processed at low temperatures in a dry air-filled glovebox (24–28% humidity). Finally, to prevent potential stability challenges associated with metal-electrode integration, self-adhesive carbon electrodes prepared via the ethanol interlacing method were applied to HTL using an automatic heat press at 0.6 MPa and sequentially pressed at 25 °C for 300 s and 50 °C for 180 s. For 1 cm² devices, an ITO substrate was placed on top of the carbon electrode prior to heat pressing to ensure reliable large-area electrical contact.

2.8. Characterizations and Measurements

The morphology of perovskite thin films was examined using a field emission scanning electron microscope (FESEM, JSM-7001F), and film thicknesses were determined with ImageJ software. Surface topography was characterized by atomic force microscopy, and KPFM work function mapping was performed using the KPFM mode of the atomic force microscopy (AFM, Park NX-10) in contact mode using a FORTA probe (61 kHz resonance, 1.6 N/m spring constant), with scans performed at 0.25 μm/s and a 0.5 nN set point. Hydrophobicity was evaluated using contact angle measurements (Kino SL250). Each photograph was taken 10 s after depositing a water droplet onto the perovskite surface. Crystallinity and structure were analyzed by X-ray diffraction (XRD, Rigaku) with Cu Kα radiation (λ = 1.5406 Å), while molecular composition was assessed with Fourier-transform infrared spectroscopy (FTIR, Vertex 80 V, ATR mode). Chemical composition and core-level binding energies were determined using monochromatic Al Kα X-ray photoelectron spectroscopy (1.4 keV). Optical absorption was measured with a UV–vis spectrophotometer (Lambda 35, PerkinElmer). Photoluminescence (PL) spectra were recorded over 532–900 nm with 1 nm increments using emission monochromators (AvaSpec-2048TEC-USB2-2). The excitation source and detector were positioned at 90° relative to each other, with the sample at 60° to minimize reflection. Time-resolved PL (TRPL) was measured using a 405 nm pulsed laser diode. Both PL and TRPL were performed on Glass/perovskite and FTO/TiO₂ NPs/perovskite structures with and without DBABr treatment to evaluate carrier lifetimes and defects. Trap densities in hole-only devices (FTO/PEDOT:PSS/perovskite/Spiro-OMeTAD/Carbon) were determined via the space-charge-limited current (SCLC) method under dark conditions (0–2.5 V). Photovoltaic J–V characteristics were recorded under AM 1.5 illumination (100 mW/cm²) using a Keithley 2400 Source Meter calibrated with a silicon reference cell. Measurements were conducted in reverse (1.2 → −0.2 V) and forward (−0.2 → 1.2 V) scans with 0.05 V steps and 15 s delay, without preconditioning. Laser patterning defined the carbon electrode and active area (0.09 cm² mask) and 1 cm², with overall device dimensions of 2.5 × 2.5 cm². Aging tests at the maximum power point (MPP) were conducted for 500 s, and external quantum efficiency (EQE) was measured using a QER system (Enlitech) to observe the light response. Mott–Schottky analysis was performed via capacitance–voltage measurements (0–1.2 V bias, dark) using an Autolab PGSTAT 302N to determine built-in potentials. Electrochemical impedance spectroscopy (EIS) was carried out under 0.6–0.85 V bias (dark) over 100 kHz to 0.1 Hz, with equivalent circuit modeling to extract recombination parameters. Indoor performance was evaluated using an 8 W, 6500 K Philips LED at illumination levels of 1000 lx (0.3651 mW cm^–2), 500 lx (0.1815 mW cm^–2), and 200 lx (0.07226 mW cm^–2). The emission spectrum and light intensity were characterized using an Ocean Optics USB4000 spectrometer Ultraviolet photoelectron spectroscopy (UPS) measurements were performed at beamline 3.2 Ua/b of the Synchrotron Light Research Institute, Thailand, using a photon energy of 60 eV to probe energy-level distribution. For charge carrier dynamics analysis, transient photovoltage (TPV) and transient photocurrent (TPC) measurements were performed under a Paios LED spectrum. The light pulse intensity was set to 80%, corresponding to 576 W m^–2 (720 W m^–2 × 80)/100. A pulse length of 10 μs was used for TPV measurements, while a pulse length of 5 μs with an offset voltage of 1.20 V was applied for TPC measurements. All signals were recorded using a digital oscilloscope.

2.9. Statistical Analysis

TRPL, TPV, and TPC data were normalized, and their decay curves were fitted with a biexponential function to provide insight into recombination dynamics and charge-carrier extraction. Stability test data were also normalized, while other measurements were processed using the corresponding formulas.

3. Results and Discussion

3.1. Formation of a 2D/3D Perovskite Heterostructure

To confirm the formation of a 2D/3D perovskite heterostructure, the 2D and 3D precursor solutions were mixed in equal volumes and spin-coated on glass, as described in the Methods section. The XRD pattern of the resulting film (Figure S1) shows a distinct low-angle peak at 6.3°, together with a weaker reflection at 8.2° (marked by stars), both characteristics of 2D perovskites layer. These features indicate the formation of a (DBA)₂PbI₂Br₂-related phase on the 3D perovskite surface. The increased intensities of the (100) and (200) reflections further suggest improved crystallinity, whereas the enhanced PbI₂ peaks may be attributed to excess PbI₂ introduced into the mixture through the incorporation of (DBA)₂PbI₂Br₂. The pristine 3D film displays well-defined polycrystalline grains (Figure S2a), while the 2D/3D film shows a newly formed surface layer with a more porous texture and reduced grain-boundary contrast as observed from the top SEM images (Figure S2b). Cross-sectional SEM images reveal a thicker film for the 2D/3D structure compared with the 3D control (Figure S2c, d), confirming the presence of an additional surface layer. Together, these results verify the successful formation of the 2D/3D heterostructure, providing a foundation for subsequent surface engineering and device optimization.

3.2. Surface Morphology

After confirming the formation of the 2D/3D heterostructure, various concentrations of DBABr were dissolved in IPA, which has been reported to facilitate surface reconstruction and applied as a posttreatment on the perovskite films. The surface morphology of the films was examined using scanning electron microscopy (SEM). Top-view SEM images (Figure a–d and Figure S3a-d), together with the corresponding grain-size histograms, show that crystalline PbI₂ features present on the control film surface gradually diminish with increasing DBABr concentration, indicating a reduction in residual PbI₂. At the same time, the surface becomes more uniform, with enlarged grains and fewer grain boundaries at moderate DBABr concentrations. These changes are attributed to interactions between the ammonium species and under-coordinated surface and grain-boundary sites of the corner-sharing [PbX₆]^4– framework (X = I, Br), which promote surface reconstruction and secondary grain growth through DBABr adsorption. , The most favorable morphology is observed at DBABr concentrations of 0.75 and 2 mg mL^–1, where the grain-size distribution becomes more homogeneous. In contrast, excessive DBABr treatment (6 mg mL^–1) results in a reduced average grain size of approximately 567 nm, suggesting that higher concentrations disrupt optimal crystallization.

1.

(a,b) Top-view SEM images with corresponding grain-size distributions, (c,d) cross-sectional SEM images, (e,f) water contact angle measurements of the control and DBABr-treated perovskite films.

Cross-sectional SEM images further show that DBABr treatment at 0.75 mg mL^–1 slightly increases the perovskite film thickness from 725 nm (control) to 795 nm (Figure S4), resulting in a denser and more uniform microstructure (Figure d). The thickness increases marginally to 801 nm at 2 mg mL^–1 but decreases sharply to 582 nm at 6 mg mL^–1 (Figure S4c, d), indicating that excessive DBABr can adversely affect film formation and continuity. While film thickness influences optical absorption and carrier transport, optimal device performance depends primarily on achieving balanced absorption and efficient charge extraction rather than maximum thickness alone. , In addition to morphological changes, DBABr treatment significantly enhances the surface hydrophobicity of the perovskite films. Water contact angle measurements (Figure e,f and Figure S5a-d) show higher contact angles for all DBABr-treated films compared with the control. This enhancement arises from the hydrophobic butyl chains of DBABr and the formation of a thin 2D/3D capping layer, which suppresses moisture ingress and protects the underlying 3D perovskite, thereby mitigating ambient degradation.

3.3. Structural Characterization and Chemical States Analysis

The crystallinity of the perovskite films was examined using X-ray diffraction (XRD), as shown in Figure a and Figure S6. All films exhibit distinct diffraction peaks at 14.0°, 20.0°, 24.5°, 28.4°, and 31.9°, corresponding to the (100), (110), (111), (200), (210) planes of Cs_0.17FA_0.83Pb­(I_0.83Br_0.17)₃, respectively. , After DBABr treatment, the diffraction intensity of the residual cubic PbI₂ peak at 12.7° decreases noticeably, indicating reduced surface PbI₂ content, consistent with the SEM observations. At DBABr concentrations of 0.75 and 2 mg mL^–1, the (100), (110), and (200) reflections become slightly more intense (Figure a and Figure S6), suggesting improved crystal ordering and effective defect passivation. This trend is further supported by the reduction in the full width at half-maximum (fwhm) of the (100) peak, which reaches a minimum at 0.75 mg mL^–1 (Figure S7), indicating optimized crystallinity. At low diffraction angles, the 2D/3D-related peak initially observed at 8.2° (Figure S1) shifts to 7.8° after DBABr treatment (Figure a), implying reduced residual stress and enhanced structural ordering. In contrast, a pronounced diffraction peak at 6.3° appears at a higher DBABr concentration (6 mg mL^–1; Figure S6), confirming the formation of a thicker 2D phase. Such excessive 2D coverage is known to hinder charge extraction, which is consistent with the reduced device performance observed at this concentration (Figure S8). ,, The corresponding photovoltaic parameters are summarized in Table S1, and their statistical distributions are shown in Figure S9. Based on these results, a DBABr concentration of 0.75 mg mL^–1 yields optimal device performance and is therefore referred to as DBABr in the following discussion.

2.

Structural and surface chemical characterization of control and DBABr-treated perovskite films, (a) XRD patterns, (b) XPS C 1s spectra, (c) XPS N 1s spectra, (d) XPS Br 3d spectra, (e) XPS Pb 4f spectra, (f) XPS I 3d spectra, (g) FTIR spectra, and (h,i) surface KPFM potential images.

The surface chemistry of the control and DBABr-treated perovskite films was investigated using X-ray photoelectron spectroscopy to identify their surface composition and bonding characteristics. In the control film, the C 1s spectrum shows two peaks at 285.39 and 288.74 eV, assigned to C–C/CH₃ and CN bonds, respectively. After DBABr treatment, the spectrum changes: the C–C/CH₃ peak increases in intensity and shifts to 285.71 eV, while the CN peak narrows and shifts to 288.49 eV (Figure b), indicating the presence of dibutylammonium cation anchored on the surface of the 2D/3D perovskite heterostructure. In the N 1s region, the main peak at ∼400.9 eV, associated with N­(C-NH₂ ⁺) from the formamidinium cation, shifts to 400.77 eV upon DBABr treatment. In addition, a new peak at 402.46 eV appears, corresponding to N­(NH₃ ⁺) from the dibutylammonium cation (Figure c). The enhanced C–C/CH₃ signal and the emergence of the N­(NH₃ ⁺) component confirm the incorporation of long-chain DBA⁺ cations at the film surface. , The Br 3d_3/2 and Br 3d_5/2 peaks, originally located at 69.89 and 65.85 eV, respectively, increase in intensity after DBABr treatment due to the additional Br introduced by DBABr (Figure d). Both peaks also shift to lower binding energies, which can be attributed to N–H···Br hydrogen bonding between DBA⁺ cation and Br^– ions within the perovskite lattice. The Pb 4f spectra (Figure e) show that the Pb 4f_5/2 and Pb 4f_7/2 peaks shift from 143.65 and 138.75 eV to 143.49 and 138.65 eV, respectively, indicating increased electron density around surface Pb²⁺ sites. , This shift suggests coordination of DBABr with under-coordinated Pb²⁺ defects, leading to effective surface passivation. A similar downshift is observed in the I 3d_3/2 and I 3d_5/2 peaks (Figure f), consistent with reduced Pb 4f-related binding energies. These results indicate that DBABr interacts with under-coordinated Pb²⁺ and forms intermolecular N–H···I^– hydrogen bonds with surface iodide ions, effectively compensating iodide vacancies and promoting larger grains formation. This thereby minimizes lattice disorder at the interface and suppresses defect formation. The O 1s spectrum of the control film shows a peak at 533.36 eV, assigned to adsorbed oxygen species, which is strongly suppressed after DBABr treatment (Figure S10). This result agrees with the enhanced hydrophobicity observed in water contact angle measurements, indicating improved resistance to moisture and oxygen exposure. Further evidence of DBABr incorporation is provided by Fourier-transform infrared (FTIR) spectroscopy, which shows an increased CN vibration at 1705 cm^–1, attributed to the FA⁺ cation, along with a slight enhancement of N–H stretching modes at 3417 and 3261 cm^–1 (Figure g). Elemental distribution was examined by energy-dispersive spectroscopy (EDS). Uniform spatial distributions of C, N, and Cs are observed across both films (Figure S11), with no evidence of elemental agglomeration. Quantitative analysis (Figure S12) reveals an increase in carbon content and a slight decrease in Cs content after DBABr treatment. All these results indicate that DBABr interacts chemically with the perovskite surface through hydrogen bonding and coordinative interactions, thereby suppressing nonradiative recombination and enhancing device stability.

Kelvin probe force microscopy (KPFM) was used to probe the surface potential of the control and DBABr-treated films (Figure h, i). The control film exhibits pronounced surface potential fluctuations across grains (Figure S13a), reflecting local energy-level misalignments associated with defects or lattice disorder. In contrast, the DBABr-treated film shows a more uniform surface potential distribution (Figure S13b), indicating improved electronic homogeneity. The average work function decreases from 4.72 eV for the control film to 4.66 eV after DBABr treatment, corresponding to an increased electrochemical potential and more favorable band bending for charge transfer. , Atomic force microscopy (AFM) measurements show no significant change in surface roughness for either film (Figure S14), confirming that the observed electronic improvements arise primarily from interfacial chemical modification rather than morphological roughness changes.

3.4. Photovoltaic Performance

Figure a presents the PCE distribution of control and DBABr-treated devices. The champion control device delivers a PCE of 13.82%, with a J _SC of 19.72 mA cm^–2, a V _OC of 1.06 V, and an FF of 65.88%. After DBABr treatment at optimized concentration, the device exhibits a 12% relative improvement in efficiency, reaching a PCE of 15.52%, together with an increased J _SC of 20.93 mA cm^–2, a V _OC of 1.07 V, and an FF of 69.69%. In contrast, devices treated with a higher DBABr concentration (6 mg mL^–1) show a clear degradation in performance, which is attributed to the accumulation of excess organic species at the surface and grain boundaries (Table S1). A comparison of the photovoltaic parameters with previously reported 2D/3D heterostructure-based PSCs is provided in Table S2. The operational stability under continuous illumination was evaluated by maximum power point (MPP) tracking. As shown in Figure b, the DBABr-treated device maintains an MPP efficiency of 14.09% for 600 s, indicating stable output under working conditions. The external quantum efficiency (EQE) spectra (Figure c) further reveal enhanced photoresponse for the DBABr-treated device, particularly in the 460–730 nm range, reflecting more efficient charge collection resulting from improved crystallinity, trap-state passivation, and favorable energy-level alignment. , The EQE-integrated current density (J _int) values are 18.77 and 19.44 mA cm^–2 for the control and DBABr-treated devices, respectively. The small mismatch between J _int and the J _SC values obtained from J–V measurements (within 8%) is commonly observed in carbon-based PSCs. ,, To assess scalability, devices with an active area of 1 cm² were fabricated. As shown in Figure d and Figure S15, DBABr treatment improves the champion PCE from 9.34% to 10.05%, demonstrating that the performance enhancement is preserved at larger device areas. Under indoor low-light conditions, measured following our previous protocols, , the DBABr-treated device achieves a champion PCE of 30.57% and delivers a power output of 111.9 μW cm^–2, compared with 108.1 μW cm^–2 for the control device under 1000 lx LED illumination (Figure S16 and Table S3). These output levels are sufficient to power low-consumption indoor electronics operating in standby mode. At 500 and 200 lx LED illumination, the DBABr-treated device achieves PCE values of 30.53% and 34.23%, respectively (Figure S16 and Table S3). The enhanced indoor performance is further supported by the higher J _int values obtained by integrating the EQE spectra over the LED photon flux (Figure S17a–c), confirming improved light harvesting and suppressed charge recombination in the DBABr-treated devices. The discrepancy between J _int and the J _SC derived from LED J–V measurements (Figure S16) is attributed to wavelength-dependent variations in light intensity. For 1 cm² devices under 1000 lx indoor illumination, DBABr treatment increases the power output from 92.6 to 99.4 μW cm^–2 and improves the champion PCE from 25.64% to 27.77% (Figure a and Table S4), while significantly reducing the hysteresis index (HI, Table S5), calculated as

$$\mathrm{HI}=\frac{{\mathrm{PCE}}_{\mathrm{Reverse}}-{\mathrm{PCE}}_{\mathrm{Forward}}}{{\mathrm{PCE}}_{\mathrm{Reverse}}}$$ 1

3.

(a) PCE statistical histogram, (b) stabilized power output, (c) EQE spectra and integrated current density, and (d) J–V characteristics measured with an active area of 1 cm² for control and DBABr-treated C-PSCs under 1 sun (AM 1.5G, irradiance of 100 mW cm^–2) illumination.

4.

(a) Box plot of photovoltaic PCE measured under 1000 lx illumination for devices with an active area of 1 cm², (b) PCE statistics of devices with a CuSCN HTL measured under 1 sun (AM 1.5G, 100 mW cm^–2) illumination, (c) SCLC measurements of hole-only devices, and (d) dark J–V curves of control and DBABr-treated C-PSCs.

Generally, the carrier density in PSCs under indoor illumination is significantly lower than under outdoor conditions, resulting in distinct charge transport and recombination mechanisms. Under high illumination, bimolecular recombination between electrons and holes dominates in PSCs. As the light intensity decreases, this recombination pathway is suppressed, which can enhance the FF. However, at very low light intensities, the reduced carrier density increases the relative contribution of trap-assisted recombination, leading to a decline in FF. The impact of series resistance under standard one-sun illumination (AM 1.5G) and low-light (1000 lx LED) conditions was evaluated following Steim et al. and Cui et al. by introducing resistors of 10, 22, 100, and 1 kΩ·cm². Higher resistance values cause a sharp decrease in J _SC and FF under AM 1.5G conditions (Figure S18a), whereas the effect of series resistance on photovoltaic performance is nearly negligible under low-light conditions (Figure S18b). A comparison of low-light indoor applications with previously reported literature is provided in Table S6. The compatibility of the DBABr-induced 2D/3D heterostructure with different hole-transport layers was further evaluated by replacing Spiro-OMeTAD with inorganic CuSCN in the n-i-p architecture FTO/TiO₂ NPs/Cs_0.17FA_0.83Pb­(I_0.83Br_0.17)₃/CuSCN/Carbon. As shown in Figure b, the DBABr-treated devices exhibit an increased maximum PCE of 13.05%, compared with 12.38% for the control devices. This improvement mainly arises from enhanced J _SC and FF (Figure S19), highlighting the robustness of the DBABr surface engineering strategy across different HTLs and operating conditions.

3.5. Defect Density and Charge Carrier Dynamics

The trap density (N _d ) was quantified under dark conditions using the space-charge-limited current (SCLC) method on hole-only devices, as described in the Methods section. N _trap was calculated using eq :

$${\mathit{N}}_d=\frac{2\epsilon {\epsilon }_0{V}_{\mathrm{TFL}}}{e{L}^2}$$ 2

where ε is the relative dielectric constant of the perovskite (∼35), , ε₀ is the vacuum permittivity, V _TFL is the trap-filled limit voltage, e is the elementary charge, and L is the film thickness. , As shown in Figure c, the DBABr-treated device exhibits a substantially reduced trap density compared with the control, indicating suppressed trap-assisted nonradiative recombination. Consistently, the DBABr-treated device shows a lower dark current density (Figure d), reflecting reduced leakage current and diminished recombination losses. Steady-state photoluminescence (PL) and time-resolved PL (TRPL) measurements were conducted to probe charge-carrier dynamics and film quality. For perovskite films deposited on glass substrates without charge-transport layers, the DBABr-treated film shows a stronger PL intensity than the control (Figure a), indicating suppressed nonradiative recombination and reduced defect density due to effective surface passivation. A slight blue shift is also observed, consistent with DBABr incorporation and in agreement with the XPS analysis. The TRPL decay curves were fitted using a biexponential model (eq ), and the fitting parameters are summarized in Table S7.

$$y=y_0+A_1\exp (-\frac{x}{{\tau }_1})+A_2\exp (-\frac{x}{{\tau }_2})$$ 3

5.

(a,c) Steady-state photoluminescence spectra, (b,d) time-resolved photoluminescence decay curves, (e) normalized transient photocurrent (TPC), and (f) normalized transient photovoltage (TPV) responses of control and DBABr-treated PSCs.

The average carrier lifetime (τ_avg) increases from 29.83 to 48.20 ns after DBABr treatment (Figure b), confirming synergistic defect passivation and reduced recombination losses, as calculated using eq : ,

$${\tau }_{\mathrm{avg}}=\frac{\sum A_i{t}_i^2}{\sum A_i{t}_i}$$ 4

When the perovskite films are deposited on FTO/TiO₂, pronounced PL quenching is observed for the DBABr-treated film (Figure c), suggesting more efficient electron extraction at the perovskite/TiO₂ interface. This trend is further supported by TRPL measurements (Figure d), where the average recombination lifetime decreases from 45.39 ns (control) to 38.56 ns (DBABr-treated), indicating accelerated charge transfer to the electron transport layer. ,

Transient photocurrent (TPC) and transient photovoltage (TPV) measurements were employed to further analyze carrier transport and recombination kinetics. The charge transport time extracted from TPC measurements is shorter for the DBABr-treated device (0.81 μs) than for the control (1.06 μs) (Figure e), reflecting faster carrier extraction and transport, which contribute to the enhanced Jsc and FF. Meanwhile, the longer recombination lifetime derived from TPV measurements (Figure f) indicates suppressed carrier recombination, consistent with the reduced defect density and increased V _OC. Mott–Schottky analysis was performed to evaluate the built-in potential (V _bi) (Figure a). The relationship between capacitance and voltage is expressed as (eq ):

$$\frac{1}{C^2}=\frac{2}{A^{2}q\epsilon {\epsilon }^0\mathrm{Nd}}\times (V_{\mathrm{bi}}-V)$$ 5

where A, C, q, ε, ε ⁰, and V denote the electrode area, depletion-layer capacitance, elementary charge, relative dielectric constant, vacuum permittivity, and applied voltage, respectively. The V _bi was determined by extrapolating the linear region of the 1/C ²–V plot to the voltage axis. The DBABr-treated device exhibits a higher V _bi (1.11 V) than the control (1.07 V), indicating an enhanced internal electric field and a more favorable band alignment. This strengthened driving force promotes efficient charge separation, enlarges the depletion region, facilitates exciton dissociation, and suppresses both hole and electron backflow, thereby contributing to a higher average V _OC. , Electrochemical impedance spectroscopy (EIS) measurements further elucidate the recombination and charge-transport behavior. The Nyquist plots (Figure b) exhibit a single high-frequency semicircle for each device, which was fitted using a Voigt two-component equivalent circuit model. After DBABr treatment, the series resistance (R _s) decreases from 83.04 Ω to 66 Ω, while the recombination resistance (R _rec) increases from 6609 Ω to 8152 Ω, indicating suppressed charge recombination and improved carrier transport. The R _rec values measured at different applied voltages are shown in Figure c. These electrical improvements are consistent with the observed enhancements in the FF and overall device efficiency.

6.

(a) Mott–Schottky plots, (b) Nyquist plots with the fitted equivalent circuit (inset), (c) recombination resistance (R _rec), (d) UV–vis absorption spectra, (e) Tauc plots, (f,g) schematic illustrations of UPS-derived energy band structure changes, (h) UV light stability of unencapsulated C-PSCs, and (i) long-term operational stability of control and DBABr-treated C-PSCs.

Ultraviolet–Visible absorption spectra reveal similar absorption profiles for both films, with a slight enhancement at longer wavelengths for the DBABr-treated film (Figure d). Both films exhibit an identical optical bandgap of 1.64 eV (Figure e), indicating that DBABr treatment does not alter the bulk bandgap.

Ultraviolet photoelectron spectroscopy (UPS) was employed to probe the electronic band structure and elucidate the enhanced surface potential and accelerated charge transfer in DBABr-induced 2D/3D perovskite devices. The cutoff energy (E _cutoff) and onset energy (E _onset) were determined by linear extrapolation of the corresponding binding-energy regions for both films (Figure S20). The work function (WF) was calculated using the relation WF = 60.00 eV – E _cutoff and was found to decrease slightly from 4.19 to 4.17 eV after DBABr treatment (Figure S19), consistent with KPFM results indicating a reduced ionic defect density. The valence band maximum (E _V) was determined using E _V = E _F – E _onset, where E _F is the Fermi level, and the conduction band minimum (E _C) was subsequently calculated as E _C = E _V + E _g. The E _V shifts from −5.67 to −5.61 eV, enhancing the driving force for hole extraction due to the increased built-in potential, , while the E _C shifts from −4.03 to −3.97 eV, thereby suppressing electron recombination. The resulting energy-level alignment is schematically illustrated in Figure f, g. Collectively, these results demonstrate that DBABr treatment simultaneously suppresses nonradiative recombination and promotes efficient charge transfer.

3.6. Operational and Device Stability

Currently, the commercialization of PSCs remains hindered by stability challenges. Figure h presents the UV irradiation stability test conducted on unencapsulated devices under 365 nm illumination at an intensity of 523 mW cm^–2, following the ISOS-L-1 protocol. The DBABr-treated device exhibits significantly enhanced photostability compared with the control device. It also demonstrates improved maximum power point (MPP) tracking stability under continuous illumination during MPP aging tests conducted at 85 °C in ambient air (ISOS-L-2) as shown in Figure S21. Long-term ambient stability was further evaluated using unencapsulated devices stored in air at a relative humidity of 30–35% and a temperature of 25 °C, following the ISOS-D-1 protocol. The normalized PCE evolution is shown in Figure i. The control device shows a rapid decline in PCE between 672 and 1008 h, retaining only 73% of its initial PCE, whereas the DBABr-treated device retains approximately 90% over the same period. This improvement reflects enhanced environmental stability arising from the incorporation of the bulky DBA⁺ spacer, which increases surface hydrophobicity, improves film quality, suppresses ion migration and moisture ingress in the DBABr-induced 2D/3D C-PSCs.

A comparison of our results with recent advances in 2D/3D heterostructures in C-PSCs is provided in Table S8.

4. Conclusion

In summary, we establish a postsurface treatment with dibutylammonium bromide as a robust interfacial engineering strategy to induce controlled 2D/3D interface reconstruction in PSCs. The formation of a thin quasi-2D perovskite layer on the 3D absorber enhances crystallinity, passivates interfacial defects, suppresses ion migration, and optimizes energy-level alignment, collectively leading to prolonged carrier lifetimes, reduced nonradiative recombination, and accelerated charge transport. This interfacial reconstruction results in a pronounced improvement in device performance under standard one-sun illumination and enables highly efficient operation under low-light indoor conditions, while maintaining effectiveness in centimeter-scale devices. Moreover, enhanced operational stability under ambient conditions highlights the protective role of the surface-engineered 2D/3D interface in conjunction with the carbon electrode. While excessive surface treatment can lead to overly thick 2D layers that hinder charge extraction, precise control of the surface engineering process enables an optimal balance between passivation and transport. Overall, this study provides clear mechanistic insight into surface-driven interface reconstruction and identifies surface engineering as a scalable pathway for advancing PSCs toward stable outdoor and indoor energy-harvesting applications.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.6c09772.

• Device assembly procedure; details about the characterization techniques; additional figures regarding XRD patterns, top-view SEM images, grain size statistics, cross-sectional SEM images, water contact angle, fwhm of the (100) diffraction peak, J–V characteristics with an active area of 0.09 cm², photovoltaic parameters with spiro-OMeTAD HTL, XPS spectra of O 1s, EDS, elemental content histograms, KPFM, AFM, photovoltaic parameters with an active area of 1 cm², J–V characteristics under low light, emission spectrum of a Philips 8W LED lamp at 1000, 500, and 200 lx intensity and integrated current density spectra, effect of series resistance on the J–V characteristics, photovoltaic performance parameters with CuSCN HTL, UPS spectra, ISOS-L-2 conditions at 85 °C, photovoltaic parameters under one-sun illumination, photovoltaic performance metrics of previously reported 2D/3D heterostructures, photovoltaic performance under LED illumination, reverse and forward J–V scans under 1000 lx LED, photovoltaic performance metrics of previously reported indoor C-PSCs, biexponential fitting parameters for TRPL, recent literature on 2D/3D interface engineering in C-PSCs (PDF)

The authors declare no competing financial interest.