Vertically oriented 1D/3D heterojunction for efficient and stable inverted perovskite solar cells Authors: Hongyu Chen, Xueliang Zhu, Kangwei Mo, Ning Ma, Shuting Pang, Bianyuan Ren, Xinyu Jiang, Man Yang, Yibo Tu, Guodong Li, Mingzhou Yu, Wensheng Yan, Gongxun Bai Published: 2026 ABSTRACT 1D/3D heterostructure perovskite solar cells have attracted considerable attention owing to their exceptional stability. However, conventional 1D perovskites typically align parallel to the substrate or in disordered orientations, hindering longitudinal carrier transport and limiting device efficiency. Here, we introduce a bilayer interface engineering strategy to induce a vertically oriented 1D capping layer on the surface of the 3D perovskite surface. This vertically ordered 1D/3D heterojunction enables extended carrier lifetime with rapid interlayer charge transport. More importantly, the vertically oriented 1D capping layer not only tunes energy level alignment but also significantly reduces interface trap density, thereby minimizing photovoltage deficit to 350 mV. With these merits, the champion device achieves a maximum power conversion efficiency of 25.9% (26.0% certified). Under high-temperature 85 °C and continuous illumination conditions, the 1D/3D devices maintain 83% and 86% of their initial efficiency after 500 and 1200 hours, respectively, demonstrating their excellent practical durability. Introduction Metal halide perovskite solar cells (PSCs) have attracted significant interest due to their excellent photoelectric properties, low cost, and tunable bandgaps1–3. Within just over a decade, the power conversion efficiency (PCE) has rapidly increased from an initial 3.8%4 to over 27%5, comparable to the best crystalline silicon cells. In particular, PSCs exhibit advantages in applications such as flexible devices, tandem solar cells, and building-integrated photovoltaic, positioning them as strong candidates for next-generation photovoltaic technologies6–8. Nevertheless, under damp heat and light conditions, three-dimensional (3D) perovskites tend to generate abundant defects at grain boundaries and surfaces, accompanied by severe ion migration, leading to rapid device performance degradation9–11. Hence, in terms of operational stability, the commercialization of PSCs still faces substantial challenges. Coating two-dimensional (2D) capping layer on the surface of 3D perovskite films is a widely recognized strategy for eliminating surface defects and suppressing ion migration12–15. This approach has been shown to improve the humidity stability of the device. However, the reconstructed 2D/3D heterojunction usually contains broad-band quasi-2D phases with various n values, potentially inducing interfacial energetic disorder16–18. Moreover, recent research has revealed that the 2D/3D perovskites still exhibit certain limitations under harsh aging conditions19. Elevated temperatures or prolonged illumination can accelerate the diffusion of 2D organic cations into the 3D perovskites, leading to a rapid collapse of the interfacial heterojunction20–22. In stark contrast, one-dimensional (1D) perovskites exhibit higher lattice strength and lower interlayer packing density, and are protected by organic cations, which help relieve residual interfacial stress23–26. These features endow 1D/3D heterojunctions with better stability, especially in terms of thermal stability27–29. Unfortunately, the 1D perovskite crystals tend to grow parallel to the substrate or in random orientations, leading to difficulties in charge transfer and extraction, strongly limiting the PCE (24–25%)26,30,31. Consequently, the realization of vertically aligned 1D perovskite growth on a 3D perovskite substrate not only significantly enhances carrier dynamics but also improves device efficiency while maintaining high operational stability, representing a significant breakthrough for 1D/3D PSCs. Controlling crystal growth orientation has been a challenge in low-dimensional (LD) perovskites. In this work, we report a bilayer interface engineering strategy that successfully induces vertical alignment of 1D perovskite crystals, thereby significantly enhancing the photoelectric performance of the device. Firstly, we employ the diethylamine hydrochloride (DEACl) to introduce 1D cation (DEA+) on the surface of 3D perovskite through dissolution-recrystallization reconstruction process. Subsequently, secondary post-treating using phenethylammonium iodide (PEAI) is performed to induce the formation of vertically oriented 1D DEAPbI3 along (001) plane. The crystal orientation of the 1D capping layer is perpendicular to the 3D perovskite substrates, facilitating carrier transport. Besides, the 1D perovskite not only passivates surface defect states but also modulates energy levels, thereby reducing non-radiative recombination and open-circuit voltage losses. As a consequence, the champion PCE of the 1D/3D device reaches 25.9%, which is significantly higher than that of the 3D device (23.6%). More importantly, the light and thermal stability of the 1D/3D device are also remarkably improved. Results Morphological and structural characterization As demonstrated in previous studies, direct post-treatment of the 3D perovskite surface with DEAI/DEACl solutions may lead to the formation of either a 1D phase (DEAPbI3) or Ruddlesden-Popper phase structure with mixed n-values based on the general formula DEA2PbnI3n+132. Regarding the crystal structure, DEAPbI3 is a 1D perovskite composed of face-sharing [PbI6]4- octahedral chains that are spatially separated by DEA+ cations33. The -NH3+ group of DEA+ can form strong hydrogen bonds with the inorganic chains, thereby ensuring structural stability and expanding ion channels34. However, these LD perovskites tend to grow parallel to the substrate or exhibit random orientations, severely limiting the photoelectric performance of devices31,35. To address this issue, we propose a bilayer interface engineering strategy to construct a 1D/3D perovskite heterojunction, enabling a vertically aligned orientation of the 1D capping layer at the top interface, as illustrated in Fig. 1a. Specifically, spin-coating DEACl onto 3D perovskite films to induce surface dissolution, thereby incorporating DEA+ cations into the surface and grain boundaries. This process leads to the formation of metastable DEA-Pb-Cl/I intermediate phases with high chemical reactivity. Subsequently, PEAI is drop-cast onto the reconstructed perovskite film to promote ion-exchange, stabilizing the [PbI6]4- octahedral chains at the surface. Simultaneously, a phase-pure and vertically oriented 1D capping layer is further formed on the 3D perovskite surface through a recrystallization process. A more detailed discussion of the bilayer interface engineering is provided in the Supplementary Note 1.Fig. 1Formation of 1D/3D perovskite heterojunction and morphological evolution.a Schematic of forming 1D/3D perovskite heterojunction via bilayer interface engineering. The top-view SEM images and corresponding surface pictures of (b) control, (c) DEACl and (d) target, respectively (scale bar: 200 nm). The cross-sectional SEM images of (e) control, (f) DEACl and (g) target, respectively (scale bar: 300 nm). Formation of 1D/3D perovskite heterojunction and morphological evolution. a Schematic of forming 1D/3D perovskite heterojunction via bilayer interface engineering. The top-view SEM images and corresponding surface pictures of (b) control, (c) DEACl and (d) target, respectively (scale bar: 200 nm). The cross-sectional SEM images of (e) control, (f) DEACl and (g) target, respectively (scale bar: 300 nm). To distinguish the samples, the untreated 3D perovskite (FA0.9MA0.06Cs0.04PbI3) film is named as control, the sample treated only with DEACl is named as DEACl, the sample treated only with PEAI is named as PEAI, and the sample treated simultaneously with DEACl and PEAI is named as target. Initial characterization using scanning electron microscopy (SEM) reveal significant differences in the sample surface microstructure. Compared with the control, DEACl post-treatment induces a dissolution process at the 3D perovskite surface, primarily attributed to the smaller ionic radius of Cl- and its defect-passivation capability (Fig. 1b, c)36,37. By stark contrast, a significant recrystallization phenomenon is observed on the surface of the target sample after combined DEACl and PEAI treatment (Fig. 1d). Compared with the control, the target film shows a markedly reduced lateral domain size, with grain boundaries covered by dense nanostructures, similar to 1D perovskite reported in the literature38. The large-area top-view SEM images further confirm the uniform coverage of the 1D perovskite (Supplementary Fig. 1). A reconstructed surface is clearly observed over micron-scale areas. Although the surface morphology differs from that of the control, the 1D capping layer uniformly covers the underlying 3D perovskite film. This phenomenon is not observed in the PEAI sample (Supplementary Fig. 2). Cross-sectional SEM analysis reveals continuous grain structures extending from top to bottom in both the control, PEAI, and DEACl samples (Fig. 1e, f and Supplementary Fig. 3). Strikingly, the target sample displays a distinct heterojunction architecture, with pronounced differences in grain size and morphology between its upper and lower regions (Fig. 1g). This bilayer interface engineering alters only the surface morphology, while preserving the intrinsic crystal integrity of the underlying 3D perovskite. Notably, the capping layer has a thickness of approximately 30-50 nm, obviously thicker than a conventional 2D capping layer (Supplementary Fig. 4)39,40. In addition, we observe a significant change in the film color of the target sample relative to the control sample (insets in Fig. 1b–d). This variation is likely attributed to possibly attributed to optical diffraction between materials with different refractive indices, as well as differences in grain size and surface texture. Crystal orientation and reaction mechanism To confirm the crystal structure after surface dissolution- recrystallization, X-ray diffraction (XRD) is employed to characterize the phase of all perovskite films (Fig. 2a). Obviously, we observe a diffraction peak at 2θ = 8.1° for the target sample, which is a characteristic signal of 1D DEAPbI3 phase along (001) plane33,34. Nevertheless, control experiments conducted under identical conditions using other ammonium iodides show no detectable 1D characteristic peak, confirming that this effect is not universal (Supplementary Fig. 5). Additionally, the absence of this phase in the DEACl sample may be due to the relatively low concentration of DEACl post-treatment. At low concentration, DEACl primarily induces surface dissolution, thereby forming a weakly crystalline and metastable DEA-Pb-Cl/I intermediate that is nearly undetectable by XRD. As the concentration of DEACl post-treatment increased from 1 to 2 mg mL-1, multiple LD phase XRD peaks appear between 5° and 10° (Supplementary Fig. 6). These non-highly oriented LD phases may hinder interlayer charge transfer and reduce the photoelectric performance of device41. Besides, we also observe that no LD phases are formed in the PEAI sample (Supplementary Fig. 7). Figure 2b shows the grazing incidence X-ray diffraction (GI-XRD) patterns of the target sample films at different incident angles ranging from 1° to 3°. As the X-ray penetration depth increases, the peak intensity of the 3D perovskite gradually increases, while the peak intensity of DEAPbI3 slightly decreases. This suggests that the formed DEAPbI3 is primarily concentrated on the surface of the 3D perovskite, consistent with the cross-sectional SEM observations. To decipher the structural evolution, we prepare the pure 1D DEAPbI3 perovskite films. The XRD results show that DEAPbI3 exhibits two diffraction peaks at 7° and 8.1°, corresponding to the (110) and (001) crystal planes, respectively (Supplementary Fig. 8). Based on the reported growth behavior of LD perovskites, the inorganic framework tends to align parallel to the substrate28,42. Accordingly, the (110) plane observed in the pure DEAPbI3 can be reasonably assigned as parallel to the substrate, indicating that the 1D inorganic chains primarily grow along the substrate plane. Interestingly, we find that the introduction of PEAI into DEAPbI3 precursor significantly suppresses the growth of the (110) plane, while promoting preferential orientation along the (001) plane (Fig. 2c). This orientation change indicates that the inherent tendency of 1D DEAPbI3 to grow parallel to the substrate is suppressed, with a preferential alignment along the vertical direction.Fig. 2Structural and orientation characterization of the 1D/3D perovskite heterojunction.a XRD patterns of the control, DEACl, and target samples. b The GI-XRD patterns of the target sample at different incidence angles. c XRD patterns of the DEAPb3 and DEAPbI3 + PEAI samples. d Pole-figures of the target film for (001) and (110) plans. The color bar transitioning from blue to red indicates the increasing intensity. e GIWAXS patterns of the control and target films. f The corresponding HRTEM images (scale bar: 5 nm). Structural and orientation characterization of the 1D/3D perovskite heterojunction. a XRD patterns of the control, DEACl, and target samples. b The GI-XRD patterns of the target sample at different incidence angles. c XRD patterns of the DEAPb3 and DEAPbI3 + PEAI samples. d Pole-figures of the target film for (001) and (110) plans. The color bar transitioning from blue to red indicates the increasing intensity. e GIWAXS patterns of the control and target films. f The corresponding HRTEM images (scale bar: 5 nm). The vertically oriented 1D capping layer is first supported by cross-sectional SEM imaging, which reveals a distinct columnar morphology of the surface 1D perovskite extending perpendicular to the substrate (Supplementary Fig. 9). This morphology provides direct macroscopic structural evidence for the preferential vertical growth orientation of the 1D perovskite. At the microscopic scale, the crystal orientation is further confirmed by XRD pole-figure (PF) measurement (Fig. 2d). For the target sample, the PF (001) exhibits a highly intense peak at the center (ψ ≈ 0°), indicating that the inorganic chains of 1D perovskite preferentially orient vertical to the substrate, consistent with recent reports on vertically oriented 2D perovskite35. In contrast, the PF (110) shows a broad and nearly isotropic intensity distribution without distinct poles, indicating the absence of a preferentially aligned (110) plane. The clear contrast between the PF (001) and PF (110) excludes the formation of horizontally or randomly oriented 1D phase, establishing a direct correlation between crystal orientation and vertical alignment. Grazing incident wide-angle X-ray scattering (GIWAXS) further corroborates this preferential vertical orientation (Fig. 2e). Compared with the control film, the target film shows a pronounced enhancement of the (001) scattering feature along the out-of-plane (qz) direction, whereas the in-plane (qr) signal is significantly suppressed. This highly anisotropic scattering pattern verifies the 1D phase growth perpendicular to the substrate. High-resolution transmission electron microscopy (HRTEM) imaging provided definitive evidence for the presence of the 1D phase (Fig. 2f). In the target film, surface lattice fringes with a spacing of ~10.5 Å, corresponding to the 1D (001) planes, coexist with the underlying 3D perovskite (200) lattice spacing of ~3.1 Å, confirming the formation of a 1D/3D heterointerface. Collectively, these results consistently demonstrate that the 1D perovskite grows vertically with respect to the substrate. Such vertically oriented LD perovskites can form well-aligned carrier transport channels, contributing to enhance device efficiency. To gain insight into the surface reaction mechanism, we calculate the electrostatic potential (ESP) distribution of DEACl and PEAI molecules (Fig. 3a). Both organic cations contain positively charged basic amino groups (-NH3+), with the positive charge primarily localized on the nitrogen atom. Notably, DEA+ exhibits a more localized EPS compared to PEA+, which enables it to form stronger coordination bonds with the [PbI6]4- octahedra. This strong interaction can selectively interrupt the [PbI6]4- framework through spatial site-barrier effects, thus promoting the formation of 1D perovskite phases. Consistently, the X-ray photoelectron spectroscopy (XPS) shows that the I 3 d peaks of the target sample shift to lower binding energy, indicating the presence of interactions between DEA and perovskite, which can help to suppress I- migration (Supplementary Fig. 10)43. The significant enhancement of N 1 s peak at 402 eV further confirms a direct interaction between DEA and perovskite, thereby passivating surface defects (Supplementary Fig. 11)44. A similar shift is also observed in the Pb 4 f, implying the increase in electron density around Pb2+ (Fig. 3b). Notably, the characteristic peaks of metallic Pb0 are detected in the control sample, which is attributed to the existence of iodide and cation vacancies on the 3D perovskite surface33. Nevertheless, the target sample exhibits complete elimination of Pb0 signals. These results indicate that the 1D capping layer not only suppresses halide ion migration and component volatilization but also delays the formation of Pb0, thereby significantly reducing the surface defects.Fig. 3Chemical interactions and interfacial reaction pathways enabling the formation of vertically oriented 1D perovskite.a ESP distributions of the DEA and PEA. b XPS spectra of Pb 4 f for control and target films. c, FTIR spectra of the DEACl, PEAI, and DEACl+PEAI films. d NMR spectra of the DEACl, PEAI, and PEAI+DEACl mixture. e NMR spectra of the FAI, PEA0 and FAI + PEA0 mixture. Chemical interactions and interfacial reaction pathways enabling the formation of vertically oriented 1D perovskite. a ESP distributions of the DEA and PEA. b XPS spectra of Pb 4 f for control and target films. c, FTIR spectra of the DEACl, PEAI, and DEACl+PEAI films. d NMR spectra of the DEACl, PEAI, and PEAI+DEACl mixture. e NMR spectra of the FAI, PEA0 and FAI + PEA0 mixture. To further elucidate the mechanism of vertical orientation, PEAI, DEACl, and their mixture are dissolved in isopropanol and analyzed by Fourier transform infrared spectroscopy (FTIR) (Fig. 3c). A characteristic peak at 2468 cm−1 is observed in the DEACl sample, which is attributed to the stretching vibration of NH⋯Cl45. Conversely, this peak completely disappeared upon the addition of PEAI, confirming the ion-exchange reaction between I- and Cl-. Correspondingly, the XRD peaks of the DEAPbI3 film shift by approximately 0.1° toward lower degree, indicative of lattice expansion caused by the removal of Cl-, again verifying the ion-exchange reaction (Supplementary Fig. 12). Afterward, nuclear magnetic resonance (NMR) analysis reveals that the resonance associated with the protonated amine in DEACl (δ = 8.9 ppm) shifts to a higher field and broadens upon introduction of PEAI (Fig. 3d). Concurrently, signals corresponding to protons on the carbon atoms adjacent to the amino group in PEAI also exhibit noticeable shifts. These spectral changes demonstrate that the ion-exchange process markedly alters the chemical environment of the N–H bonds between DEACl and PEAI. Importantly, PEA+ cations are prone to deprotonation during high-temperature annealing, generating PEA0, which can undergo an addition-elimination reaction with FA+ to form PEAMA+ (Fig. 3e and Supplementary Fig. 13)46. This cation possesses a more stable molecular configuration and a higher low acid dissociation constant pKa (12.8), which is conducive to effective passivation of grain boundaries and surface defects in the perovskite. Moreover, PEA+ is reported to preferentially adsorb on the lateral facets of the crystals47, where it undergoes reaction to form PEAMA+. Owing to the large steric hindrance and rigid molecular structure of PEAMA+, lateral crystal growth is suppressed, thereby promoting preferential vertical growth of 1D perovskite. Charge carrier dynamics and defect analysis To investigate the carrier recombination dynamics and defect states, we measure steady-state photoluminescence (PL). The target sample exhibits a markedly stronger PL intensity than the control, indicating suppressed non-radiative recombination at the interface (Fig. 4a). Compared with the control sample, the PL peak of the target sample exhibits a slight blue shift, suggesting effective passivation of deep-level defects. In-situ PL measurements further corroborate these observations, where the target film consistently displays stronger PL intensity during measurement, reflecting improved radiative recombination efficiency (Fig. 4b, c). Time-resolved photoluminescence (TRPL) measurements are carried out to further analyze the carrier dynamics. The TRPL decays can be well fitted with a double exponential function: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f\left(t\right)={A}_{1}\exp (-t/{\tau }_{1})+{A}_{2}\exp (-t/{\tau }_{2})+C$$\end{document}ft=A1exp(−t/τ1)+A2exp(−t/τ2)+C, where the fast component (τ1) is associated with defect-related non-radiative recombination and the slow component (τ2) corresponds to radiative recombination of charge carriers48. Compared with the control film, the target sample exhibits pronounced increases τ2, leading to a significantly prolonged average lifetime τave from 0.58 to 10.31 μs (Fig. 4d and Supplementary Table 1). The corresponding TRPL lifetime imaging further supports this trend, showing uniformly extended carrier lifetimes across the film (Supplementary Fig. 14). Although films treated with either DEACl or PEAI also exhibit extended lifetimes, it is less effective in suppressing non-radiative recombination than that of the target sample (Supplementary Fig. 15). These results indicate that the vertically oriented 1D capping layer effectively passivates surface defects and promotes interfacial electron extraction at the interface between 3D perovskite and C6049.Fig. 4Impact of 1D capping layer on carrier dynamics and defect characteristics.a PL spectra of perovskite film samples. The in-situ PL spectra of the (b) control and (c) target perovskite films. The color bar transitioning from blue to red indicates the increasing intensity. d The corresponding TRPL decays. e, TRPL decays of control (perovskite/C60) and target (perovskite/DEACl/PEAI/C60) films with an electron-transport layer. TA spectra of the (f) horizontal and (g) vertical samples upon excitation at 420 nm. h Arrhenius plot constructed from the characteristic transition frequency derived from the derivative of admittance spectra. i Trap state density (NT) of PSCs measured at 323 K. Impact of 1D capping layer on carrier dynamics and defect characteristics. a PL spectra of perovskite film samples. The in-situ PL spectra of the (b) control and (c) target perovskite films. The color bar transitioning from blue to red indicates the increasing intensity. d The corresponding TRPL decays. e, TRPL decays of control (perovskite/C60) and target (perovskite/DEACl/PEAI/C60) films with an electron-transport layer. TA spectra of the (f) horizontal and (g) vertical samples upon excitation at 420 nm. h Arrhenius plot constructed from the characteristic transition frequency derived from the derivative of admittance spectra. i Trap state density (NT) of PSCs measured at 323 K. To directly assess charge extraction of the vertically oriented 1D capping layer, we carry out PL and TRPL measurements on perovskite films with an electron-transport layer (C60). Compared with the control, target film exhibits a quicker PL quenching, indicating more efficient charge extraction at the perovskite/C60 interface (Supplementary Fig. 16)50,51. Consistently, TRPL measurements reveal a faster decay for the target film, reflecting accelerated interfacial charge transfer (Fig. 4e). These observations confirm that the vertically oriented 1D capping layer is beneficial for charge extraction. Then, we further perform transient absorption (TA) measurements on films with vertically oriented and randomly oriented 1D layers. Both samples exhibit a blue-shifted ground-state bleaching feature at 781 nm under 420 nm excitation, accompanied by broadband excited-state absorption in the 600–720 nm region (Supplementary Fig. 17). Notably, the vertically oriented 1D film exhibits a substantially faster interfacial charge transfer (τ3 ≈ 14.2 ps) compared to the randomly oriented (τ3 ≈ 78.2 ps), directly evidencing more efficient carrier extraction along the vertical orientation (Fig. 4f, g and Supplementary Table 2). Additionally, the shorter diffusion-related lifetime (τ2) together with the reduced recombination lifetime (τ4) indicates more efficient carrier transport and suppressed interfacial recombination at the perovskite/C60 interface. The corresponding PL, TRPL, transient photocurrent (TPC), current density-voltage (J-V) curves, and external quantum efficiency (EQE) results consistently support our conclusion (Supplementary Note 2 and Supplementary Figs. 18–22). Given that highly oriented crystal structures are known to suppress defect formation52. Subsequently, we conduct temperature admittance spectroscopy (TAS) measurements under dark conditions with temperatures ranging from 300 to 350 K to analyze the defects profile. The activation energies (Ea) of defect states in the control and target samples are calculated to be 0.839 eV and 0.551 eV, respectively (Fig. 4h). Correspondingly, the energy-level distribution indicates that the 1D capping layer effectively reduces the trap depth from 0.46 eV to 0.43 eV, as well as the defect density from 17.7 × 1022 to 11.5 × 1022 cm−3 eV−1 (Fig. 4i). Furthermore, we estimate trap-state density (Nt) from space charge limited current (SCLC) measurements using a hole-only device (Supplementary Fig. 23). At lower bias voltages, the dark current-voltage (I-V) curves exhibit a linear ohmic region. With increasing bias, a sharp rise in current is observed at the trap-filled limit voltage (VTFL), indicating the transition onto the trap-filled limit53. The Nt is calculated using \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${N}_{{{{\rm{t}}}}}=2\varepsilon {\varepsilon }_{0}{V}_{{{{\rm{TFL}}}}}/(e{L}^{2})$$\end{document}Nt=2εε0VTFL/(eL2), where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon$$\end{document}ε, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varepsilon }_{0}$$\end{document}ε0, e, and L represent the vacuum permittivity, relative permittivity, elemental charge and film thickness, respectively. The control exhibits a VTFL of 0.61 V, corresponding to Nt of 7.49 × 1017 cm−3, whereas the target displays a reduced VTFL of 0.52 V and a lower Nt of 6.39 × 1017 cm−3. These results are consistent with the TRPL and TAS analyses, further confirming effective suppression of trap-assisted recombination. Photovoltaic performance of device We further explored the effect of 1D capping layer on device photovoltaic characteristics by employing the inverted p-i-n device configuration of FTO/NiOX/Me-4PACz/Perovskite/DEACl/PEAI/C60/BCP/Cu (Fig. 5a). The J-V curves of the control and target devices are measured under AM 1.5 G illumination in Fig. 5b. Notably, the target device exhibits a high PCE of 25.9% (26.0% certified; Supplementary Fig. 24) with open-circuit voltage (VOC) of 1.18 V, short-circuit current density (JSC) of 26.0 mA cm−2 and fill factor (FF) of 84.5%, under reverse scan. These performance parameters are significantly better than those of the control, DEACl, and PEAI devices (Supplementary Fig. 25, and Supplementary Table 3). Compared to the control, the target device shows a lower photovoltage deficit (VOC, loss), approximately 350 mV at a band gap of 1.53 eV. Statistical analysis based on over 30 independently fabricated devices confirms the excellent reproducibility of VOC, loss (Supplementary Fig. 26). Moreover, the EQE integration JSC of target device is 25.2 mA cm−2, with an error of less than 5% compared to the J-V scan result, confirming the authenticity of PCE (Supplementary Fig. 27). The target device displays a much higher stabilized power output (SPO) of 25.3% than that of control (23.2%) for 60 s, matching well with the PCEs from J-V curves (Fig. 5c). Similarly, the average PCE of target devices is significantly higher than that of the control devices, stemming from the coinstantaneous improvement VOC and FF (Fig. 5d and Supplementary Fig. 28). To our knowledge, the PCE of target device has reached state-of-the-art level among reported 1D/3D PSCs to date (Fig. 5e and Supplementary Table 4).Fig. 5Photovoltaic performance and stability of PSCs.a The schematic structure of target device. The (b) J-V characteristic curves and (c) SPO of the control and target champion cells. d Statistical distribution of the PCE values of 25 cells per type. e Evolution of the efficiency of 1D/3D PSCs in recent years. f Energy-level diagram of perovskite films derived from UPS spectra. g MPP tracking for unencapsulated devices under continuous light illumination at 50 ± 5 °C in N2 atmosphere (initial PCE: 25.2% for target and 23.1% for control). Photovoltaic performance and stability of PSCs. a The schematic structure of target device. The (b) J-V characteristic curves and (c) SPO of the control and target champion cells. d Statistical distribution of the PCE values of 25 cells per type. e Evolution of the efficiency of 1D/3D PSCs in recent years. f Energy-level diagram of perovskite films derived from UPS spectra. g MPP tracking for unencapsulated devices under continuous light illumination at 50 ± 5 °C in N2 atmosphere (initial PCE: 25.2% for target and 23.1% for control). To elucidate the mechanisms of performance improvement, we carry out a series of device characterizations and analyses. Ultraviolet photoelectron spectroscopy (UPS) and surface energy diagram reveal that the Fermi level (EF) of target film is shifted markedly closer to the conduction band minimum (CBM) compared with the control, indicating an enhanced n-type doping effect induced by the 1D capping layer (Fig. 5f and Supplementary Fig. 29). The resulting band bending facilitates interfacial electron transport while impeding hole transfer toward the surface, thereby establishing a favorable energy alignment at the PSC interface54. This phenomenon is further supported by Kelvin probe force microscopy (KPFM) measurements (Supplementary Fig. 30). Relative to the control, the target film exhibits a lower average potential, consistent with the UPS results. Notably, the target film shows localized regions of lower surface potential (deep purple regions), attributable to the formation of 1D perovskite at the surface. Owing to its distinct crystal structure and organic spacer cations incorporation, the 1D perovskite typically exhibits a lower surface potential compared with the underlying 3D perovskite55. Collectively, the electrochemical impedance spectroscopy (EIS), capacitance-voltage (C-V), and light-intensity dependent analyses further reveal that 1D capping layer effectively suppresses non-radiative recombination and improves interfacial carrier transport, thereby boosting device performance (Supplementary Note 3 and Supplementary Figs. 31–33). According to the ISOS-L-1 protocol established by the International Summit on Organic Photovoltaic Stability (ISOS)56, we further conduct maximum power point (MPP) tracking under continuous AM 1.5 G illumination in N2 atmosphere to evaluate the long-term operational stability of devices (Fig. 5g). The target device retains over 86% of its initial PCE (25.2%) after 1200 hours of operation, whereas the control device suffers a rapid decline, falling below 50% of its initial PCE (23.1%) after 600 hours. Besides, we also perform light-soaking stability tests on unencapsulated devices. It can be found that the target device exhibits better stability performance compared with control cells (Supplementary Fig. 34). The degradation trend is generally consistent with MPP tracking measurements, further reflecting the reproducibility of results. We then evaluate the thermal stability of the corresponding unencapsulated PSCs at 85 °C in a N2 glovebox (Supplementary Fig. 35). The control, DEACl, and PEAI devices undergo rapid degradation, with their PCEs dropping below 40% of the initial values after 500, 100, and 300 h, respectively, indicating device failure. Especially, the inferior thermal stability of the DEACl devices originates from the formation of a Cl-rich metastable intermediate during the first-step treatment (Supplementary Fig. 36 and Supplementary Note 4). By contrast, the target device retains 83% of its initial PCE (25.6%) after 500 hours. The significantly enhanced thermal stability of the target device can be attributed to the generation of a 1D capping layer, which consumes the precipitate PbI2 during the aging process28. This performance is notably competitive compared to most reported PEAI-treated devices under unencapsulated conditions (Supplementary Table 6 and Supplementary Note 5), highlighting the effectiveness of bilayer interface engineering in suppressing thermal degradation. To further elucidate the degradation pathways, we first evaluate the thermal stability of perovskite films at 85°C in N2 atmosphere. XRD analysis reveals that the control, DEACl, and PEAI films generate a PbI2 phase at 2θ = 12.6° after 72 hours, indicative of thermally induced decomposition that is detrimental to device stability (Supplementary Fig. 37)57. Unlike traditional PEAI-modified films, the 1D capping layer can effectively passivate the vacancies/defects induced by MA+ escape and suppress the formation of PbI2 phase, thereby enhancing the thermal stability of films (Supplementary Fig. 38). Notably, the slight enhancement of the 1D diffraction features reflects thermally induced interfacial reconstruction and further completion of the 1D phase, which is beneficial for device stability. By comparing the TRPL spectra before and after thermal aging, we further clarify the degradation mechanisms (Supplementary Fig. 39). The control film exhibits an obvious acceleration of TRPL decay after thermal aging, indicating the formation of additional non-radiative recombination centers, primarily associated with thermally activated ion migration58,59. In contrast, the TRPL decays of target film before and after aging remain largely superimposed, indicating that the vertically oriented 1D capping layer effectively suppresses the formation of defect state and halide ion migration. To verify this point, we perform cross-sectional SEM energy dispersive X-ray spectroscopy (EDS) measurements (Supplementary Fig. 40). After thermal aging, the control device shows a significant redistribution of iodide, with diffusion toward the top electrode, evidencing severe halide ion migration. By contrast, the target device with the 1D capping layer shows a nearly unchanged I distribution profile before and after thermal aging, demonstrating effective suppression of halide ion migration. Moreover, the spatial distribution of Ag remains well preserved in the target device compared to the control, further demonstrating that a 1D capping layer can serve as a robust ion diffusion barrier, thereby enhancing the long-term stability of the device. Discussion In summary, we propose an ingenious bilayer interfacial engineering strategy that overcomes the limitations of conventional parallel-oriented 1D perovskite growth on substrates, enabling the construction of a vertically ordered 1D capping layer within 1D/3D perovskite heterojunctions. Such vertical orientation facilitates interlayer carrier transport, prolongs carrier lifetimes, and improves charge extraction efficiency. Besides, the 1D capping layer effectively suppresses interfacial non-radiative recombination by reducing defect density at the 3D perovskite/C60 interface, thereby minimizing VOC losses (350 mV). As a result, the optimized 1D/3D device achieves a champion PCE of 25.9%, representing the state-of-the-art performance for this architecture. More importantly, the device demonstrates enhanced stability under high-temperature (85 °C) and MPP tracking. This work provides valuable insights for the orientation control of LD crystals and the design of high-performance 1D/3D PSCs. Methods Materials N,N-dimethylformamide (DMF, 99.8%), dimethyl sulfoxide (DMSO, 99.7%), Chlorobenzene (CB, 99.8%), isopropanol (IPA, 99.5%), and anhydrous ethanol were purchased from Sigma Aldrich. Formamidinium iodide (FAI, 99.99%), Nickel oxide particle (NiOX), and 2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline (BCP, 98%) were purchased from Advanced Election Technology Co., Ltd. Cesium iodide (CsI, 99.999%), phenethylammonium iodide (PEAI, ≥99.5%), and methylammonium iodide (MAI, 99.99%) were purchased from Xi’an Yuri Solar Co. Diethylamine hydrochloride (DEACl) and Diethylamine Hydroiodide (DEAI) were purchased from Aladdin. [4-(3,6-Dimethyl-9H-carbazol-9-yl)butyl]phosphonic acid (Me-4PACz, 99%) and lead iodide (PbI2, 98%) were purchased from TCI. Buckminsterfullerene (C60, 99.5%) was purchased from Nano-C. PTAA was purchased from Borun New Material Technology Ltd (Borun New Material Technology). The fluorine-doped tin oxide (FTO) coated glass was purchased from Suzhou ShangYang Solar Technology. The glue dispenser/spin coater (Brand-REESEEN, PvS-mini7) used in the lab is from Jiangyin J. Wanjia Technology Co., Ltd. Device fabrication Perovskite solar cells (PSCs) were fabricated using a p-i-n structure: FTO/NiOX/SAMs/ FA0.90MA0.06Cs0.04PbI3/C60/BCP/Cu. The FTO substrates were cleaned sequentially with deionized water, acetone, and ethanol in an ultrasonic bath for 20 minutes. Then, the substrates were treated with plasma for 15 min to remove any chemical residuals and enhance the surface wettability of the FTO substrates. NiOX nanoparticles were dissolved in deionized water at a concentration of 10 mg mL−1 to form NiOX ink. The NiOX ink was filtered through a 0.45 μm PTFE filter prior to use, spin-coated onto FTO substrates at 3000 rpm for 30 s, and annealed at 150 °C for 20 min in air. After annealing, the FTO/NiOX substrates were immediately transferred to an N2-filled glovebox for subsequent interlayer deposition. Me-4PACz (1.0 mg mL−1 in anhydrous ethanol) was spin-coated at 3000 rpm for 30 s and annealed at 100 °C for 10 min. The FA0.90MA0.06Cs0.04PbI3 precursor solution was prepared by dissolving 224.4 mg FAI, 13.8 mg MAI, 9 mg MACl, 1 mg 3F-PEAI, 15 mg CsI, and 668.5 mg PbI2 in 1 mL mixed solvent of in DMF and DMSO at a volume ratio 4:1, forming a 1.45 M stoichiometric solution. The perovskite solution was spun on the substrates at 1000 rpm for 10 s and 4000 rpm for 40 s. Anti-solvent CB was dropped at 45 s (i.e., 5 s before the end of spin coating). The precursor perovskite films were then transferred to a hotplate and annealed at 100 °C for 60 min in the glovebox. Then, the DEACl (1.0 mg mL−1 in IPA) was spin-coated on the top of the perovskite, and annealed at 65 °C for 1 min. We further apply a post surface treatment with PEAI (1.0 mg mL−1 in IPA), spin coated at 4000 rpm for 30 s, and annealed at 100 °C for 10 min. Finally, the C60 (20 nm, 0.15 Å s−1), BCP (8 nm, 0.2 Å s−1), and Cu (100 nm) were deposited via thermal evaporation (vacuum degree <4 × 10−4 Pa). Device characterizations The scanning electron microscopy (SEM) images were measured using a scanning electron microscopy (GeminiSEM 360). X-ray diffraction (XRD) patterns were acquired using an X-ray diffractometer (Bruker, D8 Advance, Germany) with a Cu Kα radiation source operated at 40 kV and 44 mA. The time-resolved photoluminescence (TRPL) spectra were measured using the time-correlated single-photon counting technique (TimeHarp 260 PICO Single, PicoQuant GmbH), and the excitation was provided by a picosecond diode laser (405 nm, PiL040X, NKT photonics). The Photoluminescence (PL) signal was collected using the Thermo Scientific Luminescence Fluorescence Spectrometer (LF-1303003). A Fourier transform infrared (FTIR) spectrometer was measured by Lambda9 from PerkinElmer. The Kelvin probe force microscopy (KPFM) images were all recorded on tapping mode (Bruker, Multimode 8, Germany). Ultraviolet-visible absorption (UV-vis) spectra of perovskite films were characterized by a PerkinElmer Lambda 950UV/VIS/NIR spectrometer. The ultraviolet photoelectron spectroscopy (UPS) and X-ray photoelectron spectra (XPS) were measured by Thermo Fisher Scientific, ESCALAB 250xi, USA. The external quantum efficiency (EQE) spectra were measured by QE-R Quantum Efficiency Measurement System (Enli Technology Co., Ltd.). The current density versus voltage curves (J-V curve), dark J-V curves, steady output current curves, dark J-V curves of electron-only devices for the space charge limited current (SCLC) analysis, and the light intensity dependence curves of open-circuit voltage (VOC) and short-circuit current density (JSC) were all measured with an Enlitech 3 A solar simulator with simulated sunlight (100 mW cm−2, AM 1.5 G). The devices have an active area of 0.1 cm2, with an aperture area of 0.06 cm2 defined by a metal mask for measurements. The data from all the above-mentioned tests were collected by the SRC-0053 source meter. To evaluate the operating stability, ISOS-L-1 testing was carried out, in which unencapsulated devices (in the fixture) were subjected to continuous 1-sun equivalent illumination under the LED array and held at the maximum power point (MPP) voltage at room-temperature with a controller (MPP Tracking-4BR, Qingyuan Zhengying Electronic Technology Co., Ltd, China). The Mott-Schottky (M-S) curves, transient photocurrent (TPC), and the electrochemical impedance spectroscopy (EIS) Nyquist plots were recorded on a Chenhua CH760E electrochemical workstation. The femtosecond transient absorption (TA) measurements are performed using a fiber laser (1030 nm, 100 kHz repetition rate, 9.5 μJ pulse−1, YF-FL-10-100-IR, Yacto-Technology, China) as the laser source and a femto-TA100 spectrometer (Time-Tech Spectra, China). The 400 nm pump beams are generated by the optical parametric amplifier (YactoOPA, China). Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance II 400 by locking the signal on the deuterated solvent and using tetramethylsilane as an internal standard. All chemical shifts are quoted using the δ scale, and all coupling constants (J) are expressed in Hertz (Hz). Hall coefficients were obtained using a Hall effect tester (ECOPIA HMS-5300). Pole-figure (PF) analysis was carried out using X’Pert Texture 1.3 software from PANalytical. Transmission electron microscopy (SEM) images were measured by JEM 2100 F from JEOL at an acceleration voltage of 200 kV FE (field emission). Stability tests All stability measurements were performed on unencapsulated devices to evaluate their thermal and light stability. For thermal aging tests, the devices were placed on a hotplate inside a nitrogen-filled glovebox and maintained at 85 °C, during which devices were periodically taken out for J-V characterization to monitor the evolution of device performance. For light-soaking stability tests, unencapsulated devices were continuously exposed to one-sun equivalent LED illumination in air, with periodic J-V measurements performed to track device degradation behavior. Furthermore, MPP tracking stability tests were carried out using a dedicated setup under continuous illumination and nitrogen protection following the ISOS-L-1 protocol, with the operating temperature maintained at 50 ± 5 °C, ensuring a reliable assessment of long-term operational stability. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review file Supplementary Information Reporting Summary Transparent Peer Review file Source data Source Data Source Data