Introduction

Metal halide perovskite solar cells (PSCs) have shown their superior optoelectronic properties to achieve a record efficiency of 26.1% for single junction devices1. The tunability of this material opens up its applications for light-emitting diodes, detectors, lasers, and also tandem solar cells for even higher power conversion efficiency (PCE)2,3,4,5. Notably, the monolithic two-terminal (2 T) perovskite/silicon tandem devices have successfully surpassed the theoretical limit of single-junction crystalline silicon (c-Si) by a certified efficiency of 33.9%, with an annual absolute efficiency growth rate of 1.47%1.

Engineering of perovskite interfaces, especially through the use of 2D/3D perovskite heterojunctions, is essential to enhance the performance and stability of PSCs and tandem devices6,7,8,9,10,11. For example, in single junction devices, it is effective to passivate perovskite absorbers (Eg < 1.6 eV) by incorporating Ruddlesden−Popper (RP) phases to form 2D/3D heterojunction, which leads to outstanding PCEs over 24% and promising stability12,13. The upper 2D perovskite layers are found to inhibit non-radiative recombination, but charge transfer across the device is limited due to the insulating nature of RP phases (with low n value)14,15. To increase the charge mobility, it is proposed to reduce the [PbI6]4– plane spacing in 2D perovskites, in which the alternating-cation-interlayer (ACI) phase is a promising candidate16,17,18. (GA)(A)nPbnX3n+1 (GA = guanidinium, A = methylammonium (MA), formamidinium (FA) and Cs, X = I, Br and Cl) represents ACI phase structure composed of smaller alternating cations, which is generally more electrically conductive as compared with RP phases. Nevertheless, it requires a comprehensive understanding of the ACI phase to utilize its application in 2D/3D heterojunctions for absorber passivation, particularly for wide-bandgap (WBG) mixed-halide perovskites (Eg > 1.6 eV) suffering severe non-radiative recombination19,20. Therefore, it is necessary to broaden the design horizon of 2D passivation for high-performance WBG perovskite and tandem solar cells.

In this work, we propose a strategy by assembling 2D perovskite phases to reduce non-radiative recombination with improved charge extraction in a ~ 1.68 eV bandgap perovskite Cs0.235FA0.765Pb(I0.867Br0.133)3 with 5% MAPbCl3. Specifically, we demonstrate a binary two-dimensional (B-2D) structure, consisting of ACI and RP phases, which was realized by reacting a binary mixture of organic salts with 3D perovskite. The perovskite absorbers are effectively passivated by both phases at the surface. The resultant 2D/3D heterojunction showed an n-type surface and appropriate energy level alignment, ensuring effective electron extraction and reduced carrier recombination. As a result, 1-cm2 single-junction PSCs showed a PCE of 20.95%, with a fill factor (FF) of 81.22% and open-circuit voltage (VOC) of 1.234 V. The tandem device based on this strategy achieved a steady-state efficiency of 30.65% at an independently accredited test center of Shanghai Institute of Micro-system and Information Technology (SIMIT). In addition, the tandem devices retained 96% of their initial performance after 527 h of operation at maximum power point (MPP) under 1-Sun illumination in air (unencapsulated device with an initial PCE of 30.34%, ~25 °C, ~25% relative humidity (RH)), and 98% after 1000 h of the damp-heat test at 85 °C and 85% RH.

Results

Construction of the B-2D passivator

The B-2D passivator is composed of both the RP and ACI 2D phases at the surface of 3D WBG perovskite absorbers (Fig. 1a). ACI 2D phase can also be formed in perovskite bulk due to the permeation of GA+. Thus, the B-2D strategy enables simultaneous passivation of perovskite bulk and surface. As part of the B-2D structure, RP 2D layers had been widely utilized to reduce defect densities and improve the reliability of single-junction PSCs12. Here, we introduced oleylammonium iodide (OAmI) to construct RP perovskite as one component for the binary structure. The powder X-ray diffraction (XRD, Supplementary Fig. 1) pattern of the OAmI-treated sample exhibits a characteristic peak at ~4.46°. And the steady-state photoluminescence (PL, Supplementary Fig. 2) spectra of these samples show a featured emission peak at 563 nm, indicative of the formation of an RP perovskite phase (n = 2)12.

Fig. 1: Structure of 2D perovskites.
figure 1

a Schematic illustration of B-2D strategy, showcasing dual passivation for perovskite bulk and surface. Evolution of XRD patterns (b) and steady-state PL spectra (c) for the perovskite films treated with varying concentrations of GACl. The ACI 2D phase was marked as #. d GIWAXS map of the B-2D perovskite film. e ToF-SIMS depth profiles of B-2D sample. The tracked ions are positively charged monovalent fragments. f AR-XPS spectrum for N 1 s of B-2D film. g Ration of ratio of integral characteristic peak areas of ACI (12.96°) and 3D (14.13°) extracted from the GIXRD.

Full size image

ACI 2D layers are scarcely reported to construct gradient 2D/3D heterojunctions in WBG perovskites. In ACI phase, two cations (GA and A site cations) alternatingly occupy the interlayer space for closer contacts of [PbnX3n+1] slab, which is more conductive than the RP phase with the same thickness16,21. We prepared the sample by treating 3D WBG perovskite films with GACl (see methods for details). Figure 1b shows the evolution of XRD diffraction patterns for GACl-treated WBG perovskites as the concentration of GACl increases. The main characteristic diffraction peaks of the untreated film locate at 14.13° and 12.64° are corresponded to the (100) crystal plane of 3D perovskite and PbI2, respectively. For the sample treated by GACl with a concentration of 2 mg/mL, a new diffraction peak appears at about 12.96° instead of the PbI2 peak at 12.64°. And the peaks at 11.55°, 9.32°, and 6.73° appeared successively with the increasing concentration. These low-angle diffraction peaks suggest that the introduction of GACl results in the formation of 2D structures.

To identify the origin of these new XRD peaks, we fabricated (GA)(FA)nPbnI3n+1 ACI 2D perovskite films with different n values (Supplementary Fig. 3). The characteristic diffraction peaks at 12.81°, 11.45°, and 6.46° corresponding to the ACI phases are observed, which are slightly left-shifted compared to the new peaks (12.96°, 11.55°, and 6.73°) in GACl-treated WBG perovskite films. This offset is due to the reduction of the lattice constant because of the partial cation/anion substitution by smaller Cs+, Br–, or Cl–. Based on these results, we reveal the formation of ACI 2D phases via GACl treatment. Moreover, we probed n-value distributions for ACI phases by PL measurements, in which their characteristic emission peaks at 625, 656, and 692 nm are corresponding to the different ACI phases with n = 2, 3, and 418,22,23, respectively (Fig. 1c and Supplementary Fig. 4). For simplicity, we designate the films/devices samples treated by OAmI and GACl as RP and ACI, respectively.

To construct the B-2D passivator, a simple one-step treatment with the binary mixture of OAmI and GACl was employed. The grazing-incidence wide-angle X-ray scattering (GIWAXS) measurements were performed to investigate the resultant crystal structure. We find the characteristic diffraction spots of the RP and ACI 2D phases appear in the samples treated with OAmI and GACl, respectively, which matches the 1D XRD results well (Fig. 1d and Supplementary Fig. 5). And these characteristic spots coexist in the B-2D sample, verifying the formation of a binary structure composed of the RP and ACI phases. The structure of B-2D passivator is in line with PL results (Supplementary Fig. 6). The strong and concentrated diffraction spots in the z direction are indicative of a highly lateral orientation of the B-2D perovskite layers.

To investigate the uniformity of the 2D phase on large-area films, we grew the 2D phase on a 4 × 4 cm2 perovskite film (Supplementary Fig. 7). The consistent B-2D signature peaks in 4 regions demonstrates the uniformity and scalability of our B-2D technology. In addition, the surface morphology and coverage of the B-2D passivator were investigated. The scanning electron microscopy (SEM) image shows lamellar generation and contrast variation in the B-2D sample, which combines morphological features of both ACI and RP films (Supplementary Fig. 8). As compared with the reference sample (Ref), the roughness variation of samples with different 2D surface layers is within 5 nm (Supplementary Fig. 9). Furthermore, there were no significant changes observed in the optical absorption, as well as the optical bandgap (Supplementary Fig. 10).

To further understand the structure of B-2D perovskite, we employed time-of-flight secondary-ion mass spectroscopy (ToF-SIMS) to probe the depth profiles of the atomic species within the ITO/perovskite/B-2D/C60 configuration (Fig. 1e and Supplementary Figs. 11, 12)24,25. OAm+, the spacer cation of RP phase, is found to exist on the perovskite surface. It is evident that GA+ and Cl– are distributed not only at the interface of the perovskite/C60 but also extensively within the bulk of the perovskite (Supplementary Figs. 13‒15 and Supplementary Note 1). Interestingly, the distribution curve of FA+ imitates the trajectory of GA+ at the upper surface of perovskite in B-2D sample, in which a bump is clearly shown. This phenomenon is not observed in the Ref, which is likely caused by the formation of ACI phase that composes FA+ and GA+ alternatively in their spacer layers.

Besides, we carried out X-ray photoelectron spectroscopy (XPS) to probe the chemical bonding information of the perovskite absorber at the surface (Supplementary Fig. 16). The binding energy at ~400.3 eV is assigned to the NH2+ group in FA+ (N1) for the Ref film. And an additional peak locates at 402.0 eV for the NH3+ group from OAm+ (N2) in RP films, which again confirms OAm+ exists at the upper surface. When the perovskite film is treated with pure GACl, the Cl 2p peaks exhibit enhanced intensity, suggesting more Cl‒ retains at the surface of the perovskite film. Noted that the B-2D film presents a similar N 1s signal to that obtained from the RP film. It further exhibited an increased intensity of Cl 2p peak similar to that in GACl film. These results indicate the successful application of the B-2D passivator in our strategy.

Moreover, the angle-resolved XPS (AR-XPS) measurement26 was performed to reveal the depth profile of RP phases by monitoring N 1s signals (Fig. 1f and Supplementary Fig. 17). With the takeoff angle increasing from 0° to 60° (the detection depth from 3 nm to 1 nm), we find the ratio of integral areas for N2 to N1 is increased. It suggests that the RP phase, with nanometers thickness, in the B-2D system is mainly concentrated at the upper surface of perovskite. Subsequently, we performed typical grazing incidence x-ray diffraction (GIXRD) measurement to assess the distribution depth of the ACI phase by further increasing the x-ray incidence angle from 0.5° to 3° (the detection depth from about 70 nm to 500 nm)27,28. As shown in Supplementary Fig. 18, the presence of the ACI characteristic peak when the incident angle is 3° implies that the ACI phase exhibits strong crystallinity or possesses a significant vertical distribution within the perovskite film. We further extracted ratio of integral characteristic peak areas of ACI (12.96°) and 3D (14.13°) from the GIXRD data to analyze the variation in the content of the ACI phase with increasing detection depth. We observed that as the depth increases, the ration of integral areas rapidly decreases until it levels off, which implied the decrease in ACI content from perovskite surface to bulk. The intersection point could represent as the approximate depth of ACI phases. As a result, the distribution depth range of the ACI phase was determined to be approximately 150 − 300 nm. Combining ToF-SIMS, AR-XPS and GIXRD results, we identify the phase structure of B-2D perovskites, in which the ACI and RP phases mainly coexist at the 3D perovskite surface. Additionally, there is a portion of ACI phase present within the bulk of the perovskite absorber induced by GACl infiltration. Furthermore, the cross-sectional high-resolution scanning transmission electron microscopy (HR-STEM) images in Supplementary Fig. 19 clearly revealed the gradient 2D/3D heterojunctions across the 3D/2D/C60 configuration.

Fig. 2: Formation kinetics, passivation effects, and regulation of carrier transport behavior of the B-2D perovskite.
figure 2

In situ PL contour plots of perovskite films undergoing surface treatment with IPA (a) or 2D reactants in IPA for the fabrication of RP (b), ACI (c), and B-2D (d) layers. e PL spectra of each film. The dashed box in PL spectra encloses the characteristic peaks of the ACI structure. The grounded lines indicate the absence of the ACI phases from the back side of the perovskite film. f Energy level scheme of perovskite films with and without 2D layers stacked extracted from IPES and UPS data. g KPFM map of the B-2D perovskite film. The scale bar is 1 μm. h The statistical distribution of surface potential. i–l TPC mapping of WBG devices with or without post-treatment. The scale bar is 0.2 mm.

Full size image

The formation of B-2D passivator and its effects

To figure out the formation kinetics of B-2D perovskite at the surface of WBG perovskite, we in situ monitored the PL of as-prepared perovskites during their post-treatment, with a 375 nm excitation wavelength. Here, the isopropyl alcohol (IPA) was used to dissolve organic cations, which was reported to assist surface reconstruction29. It is observed that the PL intensity of perovskite increased instantly (0–2 s) with a broadening emission peak upon IPA dropping (Fig. 2a). In general, the instantaneous increase in intensity is likely attributed to the lens effect of the droplet that disappears during solvent evaporation30. And the PL peak broadening and subsequent quenching of PL intensity (2–28 s) are indicative of increased non-radiative carrier recombination, which suggests the generation of FA vacancy29,31. As OAmI/IPA is used to treat the perovskite film, we also observe the lens effect at first (Fig. 2b). Subsequently, a slightly widening emission peak is observed without sharp reduction in PL intensity (4–28 s), which indicates defects are effectively passivated by OAmI promptly29.

When GACl/IPA is applied, the resultant ACI film also exhibits emission intensity decrease (3–5 s) after the lens effect. And its PL intensity increases with a synchronous broadening of emission peak (5–13 s), which implies a solution-processed Ostwald ripening (Fig. 2c)32. The above phenomena indicate that both OAmI and GACl could mitigate the surface non-radiative recombination of the 3D perovskite absorber. To be noted, further extending time (22 s), a slight decrease in PL intensity is observed in ACI film. It is presumably due to the penetration of GACl into the bulk of perovskite film, which causes insufficient repair of surface damage induced by IPA. Interestingly, when OAmI and GACl are employed together, they result in B-2D perovskites mixed with products from each single treatment. As shown in Fig. 2d, the evolution of PL spectra for perovskite by B-2D strategy includes both evolutionary features of individual OAmI and GACl treatment. Importantly, the film by B-2D passivation shows the best luminescence intensity right after the lens effect (4–28 s), as compared with any single treatment.

We further conducted steady-state PL measurements on the front and back sides of samples with the configuration of glass/perovskite/B-2D to investigate the defect reduction effect (Fig. 2e). The RP films show an enhanced PL from the perovskite side but a negligible increase from the back side, which indicates the defects at the upper surface is passivated by OAmI treatment. In contrast, the PL intensity of ACI samples excited from both sides increases considerably as compared with Ref films, and the PL emission peak is red-shifted by several nanometers. Notably, we didn’t observe apparent characteristic emission peaks of any 2D phases from the back sides, proving that the formed ACI 2D phases within bulk does not penetrate through the entire perovskite film. The penetration of GACl enables effective defect passivation inside the perovskite films. As expected, the B-2D films exhibit the highest luminescence intensity upon illumination from either side. Moreover, the time-resolved photoluminescence (TRPL) measurements reveal the longest carrier lifetime in the B-2D sample (Supplementary Fig. 20). These results illustrate an effective suppression of the non-radiative recombination by B-2D perovskite to passivate both the surface and bulk of the absorber, further confirmed by the highest photoluminescence quantum yield (PLQY, Supplementary Fig. 21).

To prove carrier dynamics within the device based on the B-2D strategy, we combined ultraviolet photoelectron spectroscopy (UPS) and inverse photoelectron spectroscopy (IPES)33 to examine the energy-level alignment at the perovskite/C60 interface (Fig. 2f and Supplementary Fig. 22). Compared with the 3D perovskite, the CBM of 3D/RP (3.38 eV) is upshifted, and the work function (WF) of 4.17 eV is lower. The 3D/RP form a Type-I alignment, which is not favorable for extracting electrons but reduces recombination loss34. In contrast, 3D/ACI showed a type II alignment, in which the CBM is 3.74 eV, which is ideal for extracting electrons but presents a large energy barrier for holes34. Interestingly, the 3D/B-2D maintains the type-II contact inherited from the 3D/ACI configuration, while the incorporation of RP leads to a further deepening of the valence band maximum (VBM). The band alinement of 3D/B-2D is contributing more to facilitating the carrier transport and inhibit non-radiative recombination at electron extraction interface. Moreover, the WF systematically shifts toward smaller values by coating the perovskite with both RP and ACI 2D layers. Notably, the adding up of these two 2D approaches further lift up WF (3.92 eV) of the B-2D film. Meanwhile, a wider gap of 1.61 eV between the VBM and Fermi level (EF) are observed, which indicates a significantly enhanced n-type character. Above results suggested that our B-2D strategy not only suppresses non-radiative recombination but also facilitates charge transfer. We noted that the electrical bandgap of the 3D/2D perovskite surface, determined by combining UPS and IPES, does not match the optical bandgap of the single 2D phases because the post-treated perovskite surface is a mixture of 3D and 2D phases or multiple 2D phases rather than a single 2D component. We further performed the Kelvin probe force microscopy (KPFM) to access the surface potential of 3D perovskites after different 2D layers stacking (Fig. 2g, h and Supplementary Fig. 23). A distinct increase in the surface potential with the tendency from Ref, RP, and ACI to B-2D implies a decrease in WF, which is consistent with the UPS results. The largest surface potential of the B-2D sample confirms a more n-type surface again. It is concluded that the B-2D strategy yields a favored energy level alignment for efficient charge transfer at the perovskite/C60 interface12,35,36.

Furthermore, we employed transient photocurrent (TPC) and transient photovoltage (TPV) measurements to investigate the carrier dynamics along the entire pathway in the completed cells (Fig. 2i–l and Supplementary Fig. 24). As shown in the mapping results, the B-2D devices exhibit longest average carrier recombination lifetime of 6.3 μs (Ref’s 1.2, RP’s 3.0, and ACI’s 4.4 μs) and faster carrier transport of 17.1 ns (Ref’s 65.0, RP’s 39.4, and ACI’s 30.3 ns). And the observed uniformity of carrier dynamics behavior in B-2D devices compared to Ref is beneficial for effective photocarriers extraction in a layers-tacked PSCs to avoid detrimental recombination. These results indicate that the B-2D strategy could effectively reduce charge recombination and enhance charge extraction occurred in devices, as further confirmed by light intensity-dependent TPV and TPC measurements (Supplementary Fig. 25).

Performance of single junction devices

Next, we applied the above-mentioned absorbers to fabricate inverted WBG PSCs with a structure of glass/ITO/PTAA/perovskite/C60/BCP/Ag. We systematically optimized the recipes for the B-2D perovskite (Supplementary Figs. 26–29). Finally, the optimal proportioning for the B-2D strategy is determined to be 2 mg of GACl and 0.4 mg of OAmI in 1 mL of IPA (namely 2/0.4). As shown in Supplementary Fig. 30, a well-balanced distribution of two distinct characteristic morphologies (lamellar ACI phase and a burr-like RP phase) can be clearly observed in 2/0.4 film. The best devices performance in 2/0.4 composition has also been verified by the highest PL intensity and the longest carrier lifetime, suggesting the most effective suppression of nonradiative recombination processes (Supplementary Fig. 31).

Figure 3a shows the current density-voltage (J-V) curves of the optimal Ref, RP, ACI, and B-2D devices by reverse scan, and the detailed parameters are summarized in Supplementary Table 1. The formation of either ACI, RP or B-2D perovskites leads to a slight decrease in short-circuit current density (JSC) due to the insulating properties of their larger spacer cations, which was consistent with the corresponding external quantum efficiency (EQE) measurements (Supplementary Fig. 32). Whereas the VOC and FF were substantially enhanced by conducting 2D/3D heterojunctions. Among the three 2D/3D heterojunctions, B-2D resulted in the best performance values of VOC (1.23 V) and FF (84.12%) and thus PCE (21.58%). The B-2D PSCs’ reverse- and forward-sweep J-V curves (Supplementary Fig. 33) had negligible hysteresis. The statistical histograms of VOC and FF for each batch of 52 individual devices reveal the improvement of PCE stemmed from these two parameters, which are highly reproducible (Fig. 3b and Supplementary Fig. 34). The higher VOC and FF observed for the B-2D devices should be partially ascribed to suppressed non-radiative recombination within the perovskite bulk and perovskite/C60 interfaces. In addition, the B-2D interlayer induces better energy level alignment as well as a more n-type character that also plays a crucial role.

Fig. 3: Performance of single-junction WBG perovskite solar cells.
figure 3

a J-V curves under reverse scan for devices with and without 2D layers passivation. The aperture area is 0.102 cm2. b Histogram of VOC and FF fitted with Gaussian distributions from a batch of 52 devices for Ref and B-2D PSCs. c J-V curves for the champion B-2D device (1 cm2). d Statistics of PCE of B-2D PSCs. The B-2D strategy was implemented using spin-coating or bath immersion methods. e Thermal stability at 85 °C in N2 and (f) long-term operational stability under 0.8-Sun illumination of WBG perovskite solar cells with and without post-treatment.

Full size image

Furthermore, we fabricated 1 cm2 devices based on our B-2D approach to demonstrate its feasibility for large-area processing. B-2D-fabricated large-area PSCs exhibits an average PCE of 19.68% (obtained from 37 individual devices, Supplementary Fig. 35). Champion device achieved a PCE of 20.95% (VOC = 1.234 V, JSC = 20.90 mA cm–2, FF = 81.22%) by reverse scan and a PCE of 20.56% (VOC = 1.225 V, JSC = 20.88 mA cm–2, FF = 80.41%) by forward scan, and demonstrated sustained high efficiency of 20.79% with no decay in the steady-state output for 300 s in a nitrogen atmosphere (Supplementary Fig. 36). The J-V parameters are shown in Fig. 3c. Furthermore, regarding the narrow processing window of post-treatment by spin-coating approach, we developed a bath soaking method to extended processing window, making it feasible for mass production (Supplementary Note 2). As shown in Fig. 3d, devices obtained by bath soaking process exhibits comparable performance to spin-coating method, and the processing window is extended to about 10 s.

The stability of WBG perovskite films and devices with regard to light, temperature, and electrical bias were also examined (Supplementary Figs. 37, 38). We compared the thermal stability by XRD evolution of WBG perovskite films with and without 2D passivation at 85 °C for ten days. Clearly, the B-2D sample revealed a much lower ratio difference of the PbI2/perovskite (100) than the reference after 10 days of thermal aging, suggesting that the improved thermal stability for B-2D strategy. Consequently, as depicted in Fig. 3e, the corresponding B-2D device exhibited superior thermal stability by retaining 95% of its initial PCE after annealing at 85 °C for 1044 h, outperforming RP (92%), ACI (89%), and Ref (53%). Then, the stability of perovskite films with regard to light (1-Sun illumination for 10 days) and electrical bias (1 V) were evaluated by PL spectra evaluation. The ion-migration-induced phase segregation in mixed-halide perovskite, leading to PL spectra shift or split. Significant PL shifting under continuous illumination and splitting under electrical bias was observed in reference perovskite. The 2D/3D perovskites showed slight PL spectra change. Especially the B-2D perovskite exhibited negligible change throughout the measurement, affirming that the B-2D strategy significantly suppress the ion-migration-induced phase segregation. This is further supported by the improved long-term photostability results, as depicted in Supplementary Fig. 39 and Fig. 3f. Especially the B-2D devices demonstrate impressive durability, maintaining initial efficiencies of 80% and 92% after 300 h illumination under 1-Sun and 780 h continuous operation under 0.8-Sun, respectively. Above results demonstrate a prominent way to achieve enhanced thermal, light and electrical bias stability of WBG perovskite films by B-2D strategy. The enhancement is attributed to the passivation of bulk and surface defects of perovskite films, improving the ion migration activation energy (Supplementary Fig. 40) and thereby suppressing phase separation and degradation of perovskites during aging37,38.

Performance of perovskite/silicon tandem devices

We integrated the B-2D perovskite films in 1-cm2 monolithic 2 T tandem solar cells on double-side textured silicon heterojunction (SHJ) bottom cells (Fig. 4a). The surface and cross-sectional SEM images of the bottom silicon solar cell with a submicron random pyramid structure were shown in Supplementary Fig. 41. As shown in Supplementary Fig. 42, PTAA, as hole transport layer, could realize full-coverage and conformal on the textured silicon. We utilized a concentrated (1.7 M) precursor solution to fully cover the submicron-sized pyramids, resulting in a micrometer-thick perovskite film (Supplementary Fig. 43). The increase in the concentration of the perovskite precursor did not affect the performance output of the single-junction devices (Supplementary Fig. 44). Furthermore, considering the difference in the direction of incident light between tandem and single-junction devices, it is necessary to examine the impact of 2D passivation on optical absorption. We conducted optical absorption measurements on both the front and back surfaces of perovskite thin films prepared from precursors with different concentrations (1 M and 1.4 M; Supplementary Fig. 45). After 2D modification, there was just a slight reduction in front-side optical absorption before 500 nm. This indicates a minor effect of 2D passivation when testing the tandem devices with front-side illumination.

Fig. 4: PV characteristics and stability test of 1 cm2 2 T perovskite/silicon tandems.
figure 4

a Schematic of the monolithic tandem structure based on a double-side textured silicon heterojunction cell. b–e Statistics of VOC, JSC, FF, and PCE of the Ref and B-2D tandem solar cells. J-V curves (f) and SPO (g) of the champion B-2D tandem solar cell, including the performance parameters and a photograph of the tandem cell. h EQE spectra of the perovskite top cell and silicon bottom cell of the B-2D tandem device. R is reflectance. i Evolution of PV performance of the encapsulated B-2D tandem device when subjected to a damp-heat test at 85 ˚С and 85% relative humidity. j Long-term operation track of unencapsulated B-2D tandem device under full spectrum continual AM 1.5 G illumination in air (~25% RH) at a temperature of 25 ˚С.

Full size image

The tandem device performance statistics are summarized in Fig. 4b–e, wherein substantial increase is observed in VOC and FF. The champion B-2D tandem device displays the hysteresis-free J-V curves, with a PCE of 31.42%, VOC of 1.899 V, JSC of 20.54 mA cm–2, and FF of 80.59% (Fig. 4f). The VOC value is consistent with single-junction cell results (perovskite top cells in Supplementary Fig. 35; c-Si bottom cell in Supplementary Fig. 46). We noted that the c-Si bottom cell employed in our work is worth optimizing to further enhance the performance of the tandem device. The stabilized power output (SPO) of 31.14% was also recorded for the tandem device after continuous illumination at MPP under simulated light (1-Sun, AM 1.5 G) for 5 min (Fig. 4g). Moreover, the current density integrated from EQE spectra agrees well with the JSC obtained in J-V measurement (Fig. 4h). One such unencapsulated solar cell was measured at SIMIT, showing a steady-state PCE of 30.65%, with a reverse PCE of 30.85%, JSC of 20.25 mA cm–2, VOC of 1.877 V, and FF of 81.19% (Supplementary Fig. 47).

We performed dark J-V characterizations on the Ref and B-2D tandem devices to investigate their reverse-bias robustness (Supplementary Fig. 48). Both the Ref and B-2D perovskite/silicon tandem showed negligible changes until −10 V, which aligns with the results found in the previous literature39. To further compare the stability of the Ref and B-2D tandem under prolonged reverse biasing, we biased each device at a constant reverse voltage of −10 V for 20 min. The transient current of the Ref tandem device showed an abnormal spike after only 200 s, followed by a sudden decrease after being sustained for 100 s. However, it then gradually increased again, suggesting that the device underwent shunting. By contrast, our B-2D perovskite/silicon tandem showed negligible degradation after biasing at −10 V for 20 min (Supplementary Table 2). The results indicate the 2D passivation would help to improve the reverse bias stability of perovskite/silicon tandem.

Finally, we subjected our encapsulated tandem device to damp-heat testing (85 °C with 85% RH, IEC 61215:2021 standard; Fig. 4i and Supplementary Table 3). The device retained 98% of its initial efficiency after more than 1000 h, indicating that our 2D/3D heterojunction perovskite itself and the interfacial layers are sufficiently tolerant to thermal stress. In addition, we investigated the operational reliability of unencapsulated tandem devices aged under continuous 1-Sun illumination (AM 1.5 G, 100 mW cm‒2) at ~25 °C in ambient air (~25% RH). Our tandem cell maintains 96% of its initial efficiency of 30.34% after 527 h tracking at MPP (Fig. 4j and Supplementary Table 5). The performance of the B-2D tandem devices is among the best perovskite/silicon tandem solar cells ever-reported (Supplementary Table 6). Above results show the perovskite absorber with B-2D passivation is durable in these harsh stability tests.

Discussion

In summary, we have explored a promising B-2D strategy, by taking the advantages of both RP and ACI 2D phases, to reconcile the contradiction between defect passivation and efficient charge transport at the electron extraction interface. The ensemble effect of inhibition of non-radiative recombination, more n-type doping at the surface and well-matched interface energy level alignment boost the VOC and FF of single-junction WBG PSCs and relevant perovskite/silicon tandem devices. As a result, WBG single junction device with a stabilized PCE of 20.79% (1 cm2) have been obtained. Moreover, the monolithic 2 T perovskite/silicon tandem device enables a stabilized PCE of 30.65% (measured by a third party). The B-2D strategy improves the reverse bias stability of perovskite/silicon tandem. And the perovskite/silicon tandem devices with B-2D treatment also exhibited remarkable long-term stability, retaining 96% of their original PCE after 527 h of operation at MPP and 98% after 1000 h of damp-heat testing (85 °C with 85% RH). We anticipate that these findings will open new avenues for developing high-performance perovskite-based PV technologies.

Methods

Materials

PbBr2 (99.99%), PbCl2 (99.99%), CsI (99.9%), guanidinium chloride (GACl), guanidinium Iodide (GAI), oleylammonium iodide (OAmI), phenethylammonium iodide (PEAI), methyl-ammonium chloride (MACl), C60, bathocuproine (BCP) and Poly(bis(4-phenyl) (2,4,6trimethylphenyl)-amine) (PTAA) were all purchased from Xi’an Polymer L.T. PbI2 (99.999%) was purchased from TCI. Formamidium iodide (FAI) was purchased from Greatcell Solar Materials. The dimethylformamide (DMF, 99.99%, anhydrous), dimethylsulfoxide (DMSO, 99.9%, anhydrous), isopropanol (IPA, 99.9%, anhydrous), methanol (99.9%, anhydrous) and chlorobenzene (CB, 99.9%, anhydrous) were all purchased from Sigma Aldrich. All materials were directly used without post-treatment.

WBG Perovskite precursor solution

The 1.4 M Cs0.235FA0.765Pb(I0.867Br0.133)3 with additional 5% MAPbCl3 (relative to 1.4 M Pb, formulated by mixing MACl and PbCl2) perovskite precursor solution was prepared by dissolving a mixture of FAI (184.2 mg), CsI (85.5 mg), PbI2 (516.3 mg), PbBr2 (102.5 mg), MACl (4.7 mg), and PbCl2 (19.5 mg) in a mixed solvent of DMF and DMSO with a volume ratio of 3:1. 0.5% PEAI relative to 1.4 M Pb was further added in precursor. Before use, the precursor solution was stirred at 50 °C until clarified and then filtered via a 0.22 μm PTFE membrane.

WBG perovskite solar cell fabrication

The configuration of the single-junction device is ITO/HTL (hole transport layer)/Perovskite/C60/BCP/Ag. The ITO substrate was successively cleaned with ultrapure water, acetone, and ethanol, each for 20 mins in an ultrasonic bath. All the spin-coating processes were conducted in a nitrogen-filled atmosphere. After 15 minutes of UV-Ozone treatments on the ITO surface, the HTL (PTAA, 4 mg/mL in CB) was spin-coated on the substrate at 4000 rpm for 30 s, then annealed at 100 °C for 10 min. A pure DMF was then spin-coated at 5000 rpm for 20 s followed by spin-coating the perovskite layer. The WBG perovskite precursor was spin-coated at 1000 rpm for 5 s with an acceleration of 200 rpm/s, then followed by 5000 rpm for 20 s with an acceleration of 1000 rpm/s. Chlorobenzene of 100 μL was dropped in the center of the substrates 8 s before the end of the spin-coating process. The substrates were then annealed for 20 minutes on a hotplate at 110 °C. For 2D treatments, the GACl, OAmI, or GACl/OAmI were all dissolved in IPA with different concentrations and spin-coated onto the cooled perovskite surface at a spin rate of 5000 rpm for 30 s, the rate ramp to get 5000 rpm is 5000 rpm/s, followed by thermal annealing at 100 °C for 5 min. Note that, the immediate rotation of the substrate when the post-treatment solutions contact the perovskite surface is necessary to ensure an appropriate degree of reaction. Afterward, 30 nm C60 and 6 nm BCP were thermally evaporated at a respective rate of 0.2 and 0.1 Å s-1 onto the perovskite films sequentially. The devices were finished by thermal evaporation of 100 nm Ag under vacuum.

Si bottom cell fabrication

Silicon heterojunction (SHJ) bottom solar cells were made from 300 μm thick float-zone (FZ) n-type crystalline silicon (c-Si) wafers. Both sides of the wafer were treated with alkaline solution for textured surfaces. Afterward, the wafers were cleaned in RCA solutions and dipped in hydrofluoric acid to remove the silicon oxide layer. ~5 nm intrinsic a-Si:H, ~5 nm n-type, and 15 nm p-type a-Si:H were deposited by plasma-enhanced chemical vapor deposition (PECVD). ITO layers were sputtered from an In2O3 target onto both sides of cells, with 20 nm acting as the recombination layer and 70 nm on the rear contact through a shadow mask to define 1 cm×1 cm square cells. A 250 nm thick Ag was thermal evaporated through the same shadow mask on the rear surface. Finally, the wafers were laser-cut (Huzhou Phoenixolar Co. Ltd) into 2.2 cm × 2.2 cm square substrates for tandem fabrication.

Perovskite/silicon two-terminal monolithic tandem fabrication

The processes for HTL and 2D passivation layers are the same as that on single-junction perovskite devices, with a few adaptations:

  1. 1.

    Perovskite precursor with an enhanced concentration from 1.4 to 1.7 M was conducted to completely cover the texture silicon substrate. The precursor was spin-coated at 300 rpm for 5 s with an acceleration of 300 rpm/s, then followed by 3500 rpm for 20 s with an acceleration of 500 rpm/s. Chlorobenzene of 200 μL was dropped in the center of the substrate 8 s before the end of the spin-coating process.

  2. 2.

    The C60 ETL layer thickness is reduced to 15 nm to decrease the parasitic absorption in the top cell.

  3. 3.

    The 6.6 nm SnO2 was then deposited by atomic layer deposition (ALD) with N2 as the gas carrier. During ALD deposition, the substrate temperature was maintained at 85 °C and tetrakis(dimethylamino)tin(IV) (TDMASn) precursor source at 55 °C and the H2O source at room temperature. The dosing and purging time for TDMASn is 0.3 and 12 s, and for H2O is 0.2 and 12 s. 60 cycles were performed.

  4. 4.

    After the SnO2 layer, a 70 nm ITO layer was then sputtered with 60 W power (3-inch target) on top of the SnO2, and 1μm of Ag fingers and busbars were thermally evaporated on top of the ITO. Finally, a 110 nm MgF2 layer through thermal evaporation was used as the antireflective coating.

Device encapsulation and stability tests

The black butyl rubber sealant is used at the edges in a sandwiched construction of Glass/Tandems/Glass. This configuration was vacuum-laminated in an industrial laminator (WYT500-2 Wei Yi Te (Hebei) energy-saving Equipment Technology Co., Ltd.) at 120 °C for 20 min. Tinned plated copper strips, secured with Cu tape, were used to connect the upper and lower electrodes of the tandem devices; and were extended to the outside of the cover glass. The damp heat test was conducted by placing the tandem devices inside an environmental chamber with a temperature of 85 °C and relative humidity of 85%, then retrieving them for periodic J-V measurements. The operation stability of tandem devices was performed at MPP under a full spectrum 1Ssun irradiation (WaveLabs Sinus 220) in the air with a controlled temperature of 25 °C and relative humidity of about 25%. The sample structure used for electrical bias aging tests is ITO/perovskite/BCP/Ag. A bias voltage of 1 V is applied to the device using an electrochemical workstation, while keeping it in the dark.

Perovskite film characterizations. The crystal structure of perovskites with and without 2D passivation was analyzed using X-ray diffraction (XRD) measured by PANalytical X’Pert Pro X-ray powder diffractometer with Cu Kα radiation (λ = 1.54050 Å). Grazing incidence x-ray diffraction (GIXRD) data were measured using a Bruker D8 Advanced, equipped with Cu Kα radiation (λ = 1.54060 Å). The optical properties of the perovskite-based films/devices were obtained by a UV–visible diffuse reflectance spectrophotometer (UV–vis DRS, Japan Hitachi UH4150). The steady-state photoluminescence (PL) and time-resolved PL (TRPL) spectra were characterized by FLS1,000 (Edinburgh Instruments Ltd.), equipped with a Xe lamp and photomultiplier tube (PMT) detector. The feature of perovskite surface structure after 2D passivation was investigated by 2D synchrotron radiation grazing incidence wide-angle X-ray scattering (GIWAXS) at the BL14B beamline, Shanghai Synchrotron Radiation Facility. The processes of kinetic formation and passivation effects for 2D perovskites were monitored by in situ PL spectra with an excitation at 375 nm recorded on a Du-100 dynamic spectrum (Puguangweishi Co.Ltd). The PLQY was conducted on the samples by using a commercialized system (LQE-100-EL, Enlitech) with laser excitation from a 405 nm LED. Scanning electron microscope (SEM) images were obtained by Hitachi S4800 field-emission scanning electron microscopy. Ultraviolet photoelectron spectroscopy (UPS) measurement was carried out on an XPS AXIS Ultra DLD (Kratos Analytical). Atomic force microscope (AFM) and kelvin probe force microscopy (KPFM) measurements were conducted to investigate the surface morphology and potential of perovskite films with and without 2D passivation using an MFP-3D bio (Oxford Instruments Asylum Research) with an NSG01/Pt probe. AFM-IR (Nano-FTIR) experiments were carried out at the Bruker Dimension Icon IR using the PRUM-TNIR-D-10 probe. High-resolution scanning transmission electron microscopy (HR-STEM) images were obtained on a FEI Talos F200X G2 electron microscope operated at an accelerating voltage of 200 kV and equipped with a field emission source. Cross-sectional electron-transparent lamella for the HR-STEM study was prepared using a focused ion beam (FIB) on the FEI Helios G4 UC dual beam system. 1H NMR spectra were measured using a Bruker AVANCE III 300 MHz NMR Spectrometer in the designated deuterated solvent.

Time-of-flight secondary-ion mass spectroscopy

ToF-SIMS measurements were conducted with a PHI nano ToF II (ULVAC-PHI Inc., Japan). Bi3 + + beam (30 kV) was used as the primary beam to detect the samples and the sputter etching was performed using an Ar+ gun (2 kV, 100 nA) to obtain the desired depth profile. The area of analysis was 50 × 50 μm2 while the sputtering area was 400 × 400 μm2.

X-ray photoelectron spectroscopy (XPS) and angle-resolved XPS

XPS measurement was carried out on a STARPES system (Specs) with a monochromatic SPECS XR-MF Microwave X-Ray Source (Al Ka = 1486.7 eV). The measurement geometric diagram of angle-resolved XPS is shown in Supplementary Fig. 17. The λcosθ represents the probing depth, where λ is the mean diffusion length of electrons in perovskite film, increasing with the decrease of takeoff angle, indicating that the larger take-off angle can be applied to reflect the elemental composition closer to the film surface.

Transient photovoltage/current (TPV/TPC) measurements

The TPV/TPC Mapping setup primarily consisted of three parts: a continuous/transient light module, mapping motors, and a control unit. The light module integrated both an intensity-tunable white light (~0.6-sun equivalent) and a pulse laser (~0.06-sun equivalent), aligned by a coupling lens. The Pulse Width Modulation (PWM) mapping motors were fixed in three dimensions, allowing height adjustment to match the laser’s focus spot and horizontal movement for sample mapping. The control unit, equipped with a 2 GHz Processor, facilitated movement control, state-control of light, and signal acquisition, achieving a high resolution of up to 0.5 ns in timescale.

Solar cell characterizations

The J-V measurements of single-junction devices were performed using Keithley 2400 via the solar simulator (SSF5-3A, Enlitech) along with AM 1.5 G spectra whose intensity was calibrated by a certified standard silicon solar cell (SRC-2020, Enlitech, 2019/02/11) at 100 mW/cm2. The J-V characteristics were obtained by scanning along the reverse and forward scan direction in the range of –0.1 to 1.26 V with a scanning speed of 200 mV/s. Two metal masks with 0.102 and 1 cm2 aperture area for single-junction opaque cells with different sizes were used. J-V measurements on tandem devices were performed with four-probes in the air under an LED-based solar simulator (WaveLabs Sinus 220). The photograph of the probes’ connection (Method 2) is shown in Supplementary Figs. 50, 51 and Supplementary Table 7. Additional details could be found in Supplementary Note 3. The J-V characteristics and parameters of a silicon bottom cell are shown in Supplementary Fig. 46 and Supplementary Table 4, respectively. A scanning speed of 200 mV/s and a range of –0.2 to 1.9 V were set and no preconditioning was used in this work. A mask with an aperture area of 1.0 cm2 for tandem devices was used. Light intensity was calibrated with a National Institute of Metrology (China) calibrated Si reference cell. The stabilized power output (SPO) of tandem devices was measured by recording the continuous current output at the maximum power point (MPP) of J-V scans. The external quantum efficiency (EQE) curves were recorded by an Enli Technology (Taiwan) EQE measurement system. The light intensity for EQE measurement at different wavelengths has been calibrated with standard single-crystal Si (300‒1100 nm) and Ge (800‒1200 nm) reference solar cells. EQE spectra of perovskite top cell or Si bottom cell were measured independently by saturating the bottom/top cell with continuous biased light from a white light equipped with a long-pass (850 nm)/short-pass (550 nm) filter, respectively.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.