Enhancing halide ion homogeneity in blade-coated wide-bandgap perovskite for perovskite-organic tandem solar cells

Abstract

The shared processing technologies of perovskite and organic semiconductors make them ideal partners for constructing perovskite-organic tandem solar cells. However, the different crystallization rates of bromide and iodide ions lead to inhomogeneous vertical halide distribution within wide-bandgap (WBG) perovskite films, causing notable open-circuit voltage (V_OC) loss. In this study, we developed an approach to blade-coat halide-compositionally homogeneous WBG perovskite films by introducing a hydrogen-bonding donor solvent (formamide). The formamide effectively modulates crystallization kinetics via balancing the differential hydrogen-bonding interactions of formamide with bromide and iodide ions. The resultant WBG perovskite solar cells achieved a power conversion efficiency of 18.9% with an exceptional V_OC of 1.41 volts. Last, a perovskite-organic tandem solar cell was fabricated, achieving a high efficiency of 26.3% (certified as 25.6%) and retaining 92% of its initial efficiency after being illuminated for 1000 hours. This work provides an approach for the large-area fabrication of halide-compositionally homogeneous WBG perovskite films, paving the way for the industrialization of tandem devices.

Formamide modulates crystallization kinetics to fabricate halide-homogeneous wide-bandgap perovskite films.

INTRODUCTION

Among various perovskite-based tandem solar cells (TSCs), perovskite-organic TSCs have emerged as promising candidates despite their lower power conversion efficiency (PCE) relative to other tandem configurations (1–4). This growing research momentum stems from the breakthroughs of narrow-bandgap organic photovoltaic materials, which have substantially extended the theoretical PCE limit of perovskite-organic TSCs, recently achieving a certified efficiency exceeding 26% (5–8). Moreover, the unique advantage of using orthogonal solvents for the perovskite and organic absorbers offers a promising solution to overcome the challenges associated with fully solution-processed fabrication, potentially enabling more efficient and scalable manufacturing processes (1, 9).

In state-of-the-art perovskite-organic TSCs, wide-bandgap (WBG) perovskites with I/Br mixed halide compositions, typically exhibiting bandgaps exceeding 1.85 eV, are predominantly used as the front subcell for current matching with the narrow-bandgap organic absorber (6, 10). However, substantial open-circuit voltage (V_OC) losses in WBG perovskite subcells remain a critical bottleneck that severely limits the further development of perovskite-organic TSCs (11), primarily due to the halide phase segregation observed in halide mixed WBG perovskites with a high Br ratio beyond 50% (6).

Most studies on phase segregation in WBG perovskites focus on phase stability characterization under continuous illumination (12, 13) while also developing various approaches to enhance phase stability via compositional engineering or surface passivation (14–17). Only a few studies focus on the initial elemental and phase homogeneity of freshly formed WBG perovskite films (18–20). In contrast to photo-induced halide phase segregation, which represents a microscopic local phenomenon driven by thermodynamic processes, the halide inhomogeneity observed in the as-prepared WBG perovskite films constitutes a macroscopic phenomenon. Such macroscopic phase separation phenomenon mainly stems from the distinct crystallization kinetics between Br⁻ and I⁻ ions during film formation, with the Br-rich phase rapidly forming at the film surface (18, 21). For instance, Janssen et al. revealed that rapid, antisolvent-driven nucleation of Br-rich domains imprints a lateral halide gradient across the film. This intrinsic heterogeneity seeds sub-bandgap electronic defects that steadily erode device efficiency and hasten light-induced degradation (18). He et al. observed a vertical halide distribution in spin-coated WBG perovskite films and introduced a 2-(9H-carbazol-9-yl)ethyl phosphonic acid bilayer that templates bottom-up crystallization, yielding a homogeneous halide phase (19). While these studies focus primarily on spin-coated small-area perovskite films, the initial halide homogeneity in large-area WBG perovskite films prepared via roll-to-roll compatible techniques remains unexplored and poorly understood.

In this study, we systematically investigated the vertical distribution of halide ions in large-area WBG perovskite films fabricated using a gas-quenching blade-coating method. Our analysis revealed notable vertical halide inhomogeneity within the initial WBG perovskite films, with Br-rich and I-rich phases predominantly accumulating at the film surface and bottom regions, respectively. To enhance the halide homogeneity of the blade-coated WBG perovskite films, we incorporated hydrogen-bonding donor solvent (HBDS) into the WBG perovskite precursor alongside the conventional N,N′-dimethylformamide (DMF)/dimethyl sulfoxide (DMSO) mixed solvent system to modulate the crystallization kinetics of I⁻ and Br⁻ ions. These HBDS have a high acceptor number (A_N) value, which effectively enhances their hydrogen-bonding interactions with halide ions, especially with the electron-rich Br⁻ ions (22). Among the three studied HBDS additives, formamide (FM) demonstrates an ideal interaction strength, which effectively slows the crystallization rate of the Br⁻-rich phase to facilitate synchronized crystallization of the Br⁻/I⁻phase, resulting in vertically homogeneous WBG perovskite films. On the basis of the FM-optimized WBG perovskite films, the corresponding WBG PSCs (1.86 eV) achieved a distinct PCE of 18.9% with an exceptional Voc of 1.41 V. By combining them with an organic material, perovskite-organic TSCs were constructed and achieved a high PCE of 26.3% (certified as 25.6%) with prolonged photostability.

RESULTS

Vertical inhomogeneity of halide ion distribution

Here, WBG perovskite films with a composition of FA_0.8Cs_0.2PbI_1.5Br_1.5 (bandgap of 1.86 eV) were prepared via gas-quenching blade-coating for the fabrication of perovskite-organic TSCs (Fig. 1, A and B). This blade-coating method is inherently compatible with roll-to-roll processing, thereby offering a scalable and high-throughput approach for large-scale industrial production (23, 24). Considering that the halide homogeneity of the WBG perovskite film plays a key impact on the photovoltaic performance of the final devices, we systematically evaluate it using time-of-flight secondary-ion mass spectroscopy (ToF-SIMS) and energy-dispersive spectrometry (EDS). As shown in Fig. 1 (C to F), both ToF-SIMS and EDS images reveal pronounced accumulation of Br⁻ ions at the surface of the WBG perovskite films, while I⁻ ions accumulate at the bottom. This nonuniform distribution of halide ions indicates a heterogeneous distribution of halide ions within the perovskite films. To further confirm this phenomenon, x-ray diffraction (XRD) and depth-dependent photoluminescence (PL) spectroscopies were further conducted. As shown in fig. S1, the XRD spectra exhibit two distinct split peaks of the (200) crystal plane at 28.7° and 29.1°, corresponding to the I-rich composition at the film bottom and the Br-rich composition at the surface of the WBG perovskite films, respectively. In the depth-dependent PL spectra (Fig. 1G and fig. S2), the PL peak position exhibits a blue shift from the bottom to the surface of the perovskite films, indicating a Br-rich composition at the surface. The PL mapping also reveals a pronounced spatial heterogeneity of halide ions, with Br-rich phase dominating the surface and I-rich phase concentrated at the bottom of the perovskite films (fig. S3). On the basis of the above investigation, we conclude that the freshly blade-coated perovskite film already exhibits vertical inhomogeneity of halide ions.

Fig. 1. Characterization of halide distribution in control WBG perovskite films.

(A) Schematic illustration of the preparation of WBG perovskite films using gas-quenching blade-coated techniques. (B) Photograph of blade-coated WBG perovskite films. (C and D) ToF-SIMS study of Br (C) and I (D) distribution of the control WBG perovskite films prepared by a DMSO/DMF-based precursor. (E and F) EDS study of Br (E) and I (F) distribution of the control WBG perovskite films. (G) Depth-dependent PL spectra of the control WBG perovskite films. (H) Schematic illustration of the crystallization process of the control WBG perovskite films based on DMSO/DMF-based precursor.

To understand its mechanism, the structure of the intermediate complex PbX₂•yDMSO in these WBG perovskite films is experimentally determined by XRD and thermogravimetric analysis (TGA). As shown in fig. S4A, the characteristic peaks at approximately 9.1° and 9.7° correspond to PbI₂•DMSO, while the stronger peaks around 10.2° are attributed to PbI₂•2DMSO (25). On the basis of the relative peak intensities, PbI₂•2DMSO appears to be the dominant species among the complexes. Furthermore, the TGA result (fig. S4B) shows a mass loss of approximately 26% between 120° and 220°C, which corresponds to a PbI₂-to-DMSO molar ratio of about 1:2.1 in the as-synthesized complex. Combining the above studies, we can get the conclusion that PbI₂•2DMSO is the predominant complex in the perovskite films.

Subsequently, we propose the intrinsic mechanism for the vertical inhomogeneous distribution of halide ions in blade-coated WBG perovskite films, as shown in Fig. 1H. When a DMSO/DMF-based precursor is used to prepare WBG perovskite films, nitrogen-gas quenching drives the preferential formation of Br-rich phases at the surface since Br⁻ ions crystallize more rapidly than I⁻ ions. On the basis of the dominant complex structure of PbX₂•2DMSO, our theoretical density functional theory (DFT) calculations reveal that the binding energy (−22.12 kcal mol⁻¹) of PbBr₂•2DMSO is significantly lower than that (−26.67 kcal mol⁻¹) of PbI₂•2DMSO (fig. S5). This low binding energy induces faster release of Br⁻ ions from the solvate complex, which then immediately bind with FA⁺ cations to form Br-rich phase nuclei that undergo rapid growth. In contrast, a large number of I⁻ ions exhibit slower crystallization kinetics due to their delayed release from the solvate complex. In addition, Br⁻ ions, with a smaller ionic radius, have a lower nucleation energy barrier, resulting in an accelerated nucleation process. We therefore attribute the sequential I/Br phase crystallization to the differences in PbI₂/PbBr₂–DMSO binding energies coupled with distinct nucleation energy barriers. Consequently, this sequential crystallization process induces a vertical inhomogeneity in halide ion distribution, thus leading to compositional gradients in the final WBG perovskite films (26).

Crystallization modulation of WBG perovskite films

After knowing the reason for the halide ion inhomogeneous distribution within the as-prepared WBG perovskite films, we try to address it by enhancing the electrostatic interaction with Br⁻ ions in the perovskite precursor, thereby balancing the crystallization rate of I⁻ and Br⁻ ions. Considering that the commonly used coordinating solvent DMSO has a relatively low A_N value of 19.3 kcal mol⁻¹, which results in a small hydrogen-bonding effect with Br⁻ ions. We introduce three HBDS with higher A_N value to restrict Br⁻ ions, including FM (39.8 kcal/mol), formic acid (FA, 83.6 kcal/mol), and trifluoroacetic acid (TA, 105.0 kcal/mol), respectively (Fig. 2A). The dipole moments of DMSO, FM, FA, and TA were calculated to be 4.02, 3.95, 3.91, and 2.23 D, respectively (Fig. 2B). Notably, DMSO demonstrates a distinct concentration of negative charge, endowing it with a robust coordination capability with Pb²⁺ ions. This is further confirmed by DFT calculations (fig. S6), which show that the binding energy for the PbBr₂•DMSO interaction (−17.47 kcal/mol) is significantly higher than that (−8.67 kcal/mol) for PbBr₂•FM. However, its positive charge density is relatively weak, resulting in a limited capability to form hydrogen bonds. In contrast, the molecules of FM, FA, and TA show progressively enhanced positive charges compared to DMSO, suggesting their stronger coordination capabilities with halide ions.

Fig. 2. Property of HBDS and their impact on WBG perovskite films.

(A) A_N value of different coordination solvents of DMSO, FM, FA, and TA. (B) Molecular structures and electrostatic potential (ESP) of DMSO, FM, FA, and TA; μ represents the dipole moment of the molecule. (C) DFT calculation of HBEs between different solvents and Br⁻/I⁻ ions. (D) Photographs of WBG perovskite precursor solution with 10 vol % solvent of DMSO, FM, FA, and TA, respectively. (E to G) XRD spectra [(E) and (F)] and 3D morphology (G) of WBG perovskite films prepared with DMSO, FM/DMSO, FA/DMSO, and TA/DMSO-based perovskite precursors after annealing.

Figure 2C presents the calculated hydrogen-bonding energies (HBEs) between HBDS and Br⁻/I⁻ ions. Consistent with the evolution of A_N value, the HBEs between I⁻ ions and FM, FA, or TA show a progressive increase from −2.19 to −7.58 kcal/mol. In contrast, the HBEs between Br⁻ ions and these molecules (FM, FA, and TA) increase more rapidly from −2.86 to −13.49 kcal/mol. In summary, the HBEs between HBDS molecules (FM, FA, and TA) and Br⁻ ions are stronger than those for I⁻ ions by −2.81, −5.54, and −5.91 kcal/mol, respectively. These differences are substantially larger than those observed for DMSO, which is merely 0.67 kcal/mol. Moreover, given that HBDS can theoretically form hydrogen bonds with formamidinium ions as well, we used FM as a representative solvent to clarify the predominant interaction via DFT calculations. The result show that the HBEs between FM and formamidinium ions are −4.38 kcal/mol, which is lower than that between FM and Br⁻ ions (−5.61 kcal/mol; fig. S7). Collectively, these results indicate that HBDS molecules exhibit a preferential binding affinity toward Br⁻ ions. The additional binding energy from hydrogen bonds hindered crystal growth by introducing an energy barrier that must be overcome to break the “hydrogen-bonding lock” between HBDS and Br⁻ ions (27). Figure 2D shows the WBG perovskite precursor solutions containing FM, FA, and TA solvents. The color of the perovskite precursor solution remained unchanged after the introduction of FM solvent. In contrast, the addition of FA and TA resulted in significant color changes to orange-yellow and orange-red, respectively, which arise from charge transfer between Br⁻ ions and the solvent, mediated by too strong hydrogen bonding (28).

Characterizations of perovskite films

To understand how the hydrogen-bonding strength affects the quality of WBG perovskite films, XRD analysis was first performed. As shown in Fig. 2E, the XRD patterns of all samples show specific peaks at 14.2°, 20.4°, and 28.9°, corresponding to the (100), (110), and (200) crystal planes of the WBG perovskite, respectively. Compared to the control DMSO-based film, the FM-based sample exhibits enhanced crystallinity, particularly along the (100) and (200) planes. However, the crystallinity of FA- and TA-based WBG perovskite films deteriorates significantly, accompanied by the gradual growth of non-perovskite δ-phase 2H and 4H hexagonal polytypes as indicated by the XRD peak at 11.1° (29). Figure 2F shows a magnified view of the XRD pattern around 29°, corresponding to the (200) plane. The control film exhibits two split peaks for the (200) plane at 28.7° and 29.1°, indicating the presence of phase segregation. In contrast, the FM-, FA-, and TA-based films display a single diffraction peak, suggesting a homogeneous I/Br perovskite phase. Then, the morphologies of these perovskite films were characterized by three-dimensional (3D) surface profilometry (Fig. 2G) and scanning electron microscopy (SEM, fig. S8). The control film displays a wrinkled surface morphology with the arithmetic mean roughness (R_a) of 35.0 nm. This morphology may arise from the formation of perovskite films exhibiting a bilayer structure, which implies a nonsynchronous crystallization process between the surface and bottom of the perovskite film (30). In contrast, the FM-based sample exhibits a much smoother surface with an R_a of 13.0 nm, indicating a reconstructed crystallization process. Conversely, introducing FA and TA into the precursor solution degrades film morphology, resulting in increased R_a of 26.7 and 135.8 nm, respectively. This deterioration stems from enhanced hydrogen-bonding strength, which forcibly restricts Br⁻ ions. Collectively, XRD analysis, 3D surface profilometry, and SEM characterization demonstrate that introducing FM solvent with proper HBE effectively enhances the crystallinity and homogeneity of halide ions in WBG perovskite films.

To acquire deeper insights into the alterations in element distribution, structure, and energy level of the WBG perovskite films along the vertical direction, several depth-dependent characterizations were conducted. As shown in Fig. 3 (A to D), both EDS and ToF-SIMS analyses reveal an enhanced homogeneity of halide ion distribution throughout the depth of FM-based WBG perovskite films compared to the control sample (Fig. 1, C to F). This observation is further supported by depth-dependent PL spectra. Figure 3E and fig. S9 show constant PL peak position at 675 nm from 0- to 450-nm depth at multiple locations across a large-area FM-based perovskite film (52 mm by 60 mm), rather than a shifted peak in the control sample (Fig. 1G). Along with the similar PL mapping at both surface and bottom regions (fig. S10), we further confirm the homogeneity of halide distribution within the films. Vertical structural differences between control and FM-based perovskite films were further investigated through 2D grazing-incidence wide-angle x-ray scattering (GIWAXS). By progressively increasing the x-ray incident angle from 0.1° (surface-sensitive, ~5-nm depth) to 1.0° (bottom-sensitive) (fig. S11A) (31), depth-resolved analysis revealed a distinct phase segregation in the control film, with the (001) perovskite peak of (001) plane shifted from q = 1.046 Å⁻¹ (d = 6.00 Å, Br-rich surface phase) at 0.1° incidence to q = 1.039 Å⁻¹ (d = 6.05 Å, I-rich phase) at 1.0° (Fig. 3F and figs. S11B and S12) (32). In stark contrast, the FM-based sample maintained a constant peak position of (001) plane at q = 1.042 Å⁻¹ (d = 6.02 Å) across all incident angles, demonstrating homogeneous halide distribution throughout the film depth.

Fig. 3. Characterization of halide ion distribution in FM-based WBG perovskite films.

(A and B) EDS study of Br (A) and I (B) distribution of WBG perovskite films prepared with FM-based perovskite precursors after annealing. (C and D) ToF-SIMS study of Br (C) and I (D) distribution of FM-based WBG perovskite films. (E) Depth-dependent PL spectra of FM-based WBG perovskite film. (F) GIWAXS analysis of WBG perovskite crystal prepared by FM-based perovskite precursors after annealing (The original data are presented in fig. S8).

Furthermore, we performed depth-profile ultraviolet photoelectron spectroscopy (UPS) to map the internal band alignment arising from the out-of-plane halide gradient (fig. S13). At the film surface, the valence band maximum (VBM) of control and FM-based films are −5.66 and −5.44 eV, respectively, revealing a wider bandgap when Br⁻ ions are enriched (fig. S14). Progressively deeper into the control film, the VBM shifts to −5.66, −5.51, and − 5.42 eV, whereas the FM-treated film keeps a nearly constant VBM of −5.50 eV (fig. S15). This fact indicates that introducing the FM solvent can enhance the energy level alignment to minimize energy losses in the corresponding WBG PSCs (fig. S16) (33).

Characterizations of the hydrogen-bonding effect

The hydrogen-bonding interaction between the FM molecule and halide ions was investigated by Fourier transform infrared (FTIR) spectrometry. As shown in Fig. 4A, a distinct peak at 1615 cm⁻¹, arising from the intrinsic N–H bending vibration [δ(N–H)] of the excess FM, was consistently observed. Upon the introduction of Br⁻ ions, a previously unidentified peak emerged at 1607 cm⁻¹, indicative of N–H···Br hydrogen bond formation. This peak arises from a decrease in the force constant induced by the hydrogen-bonding effect, which lowers the wave numbers of the corresponding groups (34). In contrast, the δ(N-H) vibrations of the FM solution with I⁻ ions remain almost unchanged in the FTIR spectra, implying a relatively weak hydrogen-bonding strength between FM and I⁻ ions. Hydrogen bonding was further characterized by proton nuclear magnetic resonance (¹H NMR) spectroscopy. As shown in fig. S17, the ¹H NMR spectrum of pure FM exhibits two peaks at ~8.0 parts per million (ppm) (H_a) and 7.2 to 7.5 ppm (H_b). Upon addition of Br⁻ ions, two distinct peaks appear at 10.1 and 11.1 ppm due to hydrogen-bonding interactions (Fig. 4B). We speculate that hydrogen bonding alters the chemical environment of H_b in FM, resulting in a downfield chemical shift (deshielding effect) from 7.2 to 7.5 ppm to ~10.1 to 11.1 ppm (35). Similarly, the introduction of I⁻ ions induces a chemical shift of H_b from ~4.3 to 9.9 to11.0 ppm. These ¹H NMR results are consistent with the conclusion from FTIR spectroscopy. On the basis of the above characterization, the mechanism by which HBDS solvents homogenize the out-of-plane distribution of halide ions is illustrated in Fig. 4C. The presence of FM facilitates the formation of strong hydrogen bonds with Br⁻ ions in the precursor. This effectively suppresses the preferential formation of Br-rich phases at the film surface during gas quenching, resulting in a halide-compositionally homogeneous WBG perovskite film after subsequent annealing. Time-resolved PL measurements further confirmed that these WBG perovskite films exhibited excellent photostability under illumination, with no observable halide phase segregation (fig. S18).

Fig. 4. Characterizations of the hydrogen-bonding effect.

(A) FTIR spectra of FM solvent, FM·Br⁻ and FM·I⁻ compounds. (B) ¹H NMR spectra of the FM solvent, FM·Br⁻ and FM·I⁻ compounds at low field, respectively. (C) Schematic illustration of the crystallization process of WBG perovskite films based on FM-based precursor. a.u., arbitrary units.

Photovoltaic performance of perovskite-organic TSCs

To investigate the effect of halide ion homogeneity within WBG perovskite films on device performance, we fabricated single-junction WBG PSCs with a structure of glass/indium tin oxide (ITO)/NiOx/Me-4PACz {[4-(3,6-Dimethyl-9H-carbazol-9-yl)butyl]phosphonic acid}/WBG perovskite/C₆₀/bathocuproine (BCP)/Cu. As shown in Fig. 5A and table S1, the typical control WBG PSC shows a PCE of 16.9% with a V_OC of 1.31 V, a short-circuit current (J_SC) of 16.1 mA/cm², and a fill-factor (FF) of 0.799. As the ratio of added FM solvent in the precursor solution increases from 0 to 9 vol %, the PCE of the WBG PSCs initially increases, peaking at 6 vol %, and then decreases upon further increase to 9 vol %. At the optimal FM concentration of 6 vol %, we achieved a champion PCE of 18.9% with a V_OC of 1.41 V, a J_SC of 16.3 mA/cm², and an FF of 0.822. The J_SC of the optimized device was further confirmed by the integrated current density from external quantum efficiency (EQE) with a deviation of ~2% (fig. S19). Subsequently, we further fabricated WBG perovskite mini-modules with an aperture area of 10.4 cm². As shown in fig. S20, the control WBG perovskite solar module (PSM) exhibits a PCE of 14.8%, with a V_OC of 5.15 V, a short-circuit current (I_SC) of 39.6 mA, and an FF of 0.755. In contrast, the FM optimized mini-module achieved a higher PCE of 15.9%, with a V_OC of 5.50 V, an I_SC of 39.1 mA, and an FF of 0.769. The PCE improvement further confirmed that the introduction of FM effectively regulates the halide homogeneity of large-area WBG perovskite films, thereby translating into improved device performance at the mini-module scale.

Fig. 5. Photovoltaic performance of the perovskite-organic TSCs.

(A) J-V curves of the WBG PSCs fabricated with varying ratios of added FM solvent in the perovskite precursors. (B and C) Dependence of V_OC on light intensity (B) and TPV decay (C) of control and FM-based WBG PSCs. (D and E) Schematic illustration (D) and cross-sectional SEM image (E) of a TSC. (F) J-V curves of WBG and organic single-junction solar cells and their tandem device measured under AM1.5G illumination. (G) Statistical PCE distribution of the TSCs from 64 samples. (H) EQE spectra of the perovskite-organic TSC. (I) Long-term photostability of the perovskite-organic TSC measured under 1-sun intensity illumination from a full-spectrum LED lamp in ambient conditions with a relative humidity of 30 to 40%.

After comparison, we can find that the enhanced PCE primarily arises from the improved V_OC. Considering that the V_OC is closely associated with carrier recombination in PSCs, we investigated the ideality factors and carrier lifetimes of these devices. Figure 5B presents the V_OC evolution as a function of light intensity, which serves to uncover the ideality factor of the devices. Typically, the ideality factor of a diode falls within the range of 1 to 2, with values closer to 1 indicating a higher proportion of radiative recombination (36). The results reveal that the ideality factor of the control and FM-based devices is simulated to be 1.89 and 1.34, respectively, suggesting that nonradiative recombination was significantly suppressed after introducing FM solvent into the perovskite precursor. In addition, as shown in the transient photovoltage decay (TPV, Fig. 5C), the FM-based devices show longer carrier lifetimes of 5.8 μs than that (2.9 μs) of the control device, further confirming a reduced charge recombination and efficient charge extraction. In the comparison with other reported WBG PSCs with similar bandgaps around 1.8 to 1.9 eV (table S2), our device also shows an advanced V_OC (1.41 V) and v × f (0.622). Here, v and f represent the voltage ratio (v = Voc/V_SQ) and the FF ratio (f = FF/FF_SQ), respectively. Crucially, the FM-optimized WBG PSCs demonstrated outstanding photostability. The PCE retained 104% of its initial value after being illuminated under AM1.5G for 4800 s at the maximum power point (MPP) (fig. S21). For the long-term photostability, the PCE of WBG PSCs climbed to 120% of its initial value in the first 200 hours, a gain attributed to light soaking–induced trap filling and lattice-strain relaxation that collectively drive self-passivation in the pristine devices (37). The efficiency then declined by ~4% (fig. S22) in the following 1000 hours of continuous illumination, demonstrating excellent photostability.

Building on the excellent photovoltaic performance of our WBG PSCs, we fabricated perovskite-organic TSCs by combining efficient organic solar cells with an active material of PM6:BTP-eC9 (bandgap of 1.35 eV). As shown in Fig. 5D, the structure of the TSCs is ITO/NiOx/Me-4PACz/perovskite/C₆₀/BCP/Au/MoO₃/2PACz/PM6:BTP-eC9/C₆₀/BCP/Ag. Figure 5E shows the cross-sectional SEM image of the perovskite-organic TSC, with WBG perovskite and organic layers’ thicknesses of 350 and 120 nm, respectively, to achieve the best current match. Given the relatively low PCE of the blade-coated organic solar cell (OSC) (16.6%; fig. S23), the spin-coated device was therefore selected for the fabrication of the TSC. The spin-coated OSC exhibits a PCE of 18.7% with a V_OC of 0.84 V, a J_SC of 28.1 mA/cm², and an FF of 0.795 (fig. S24). The champion perovskite-organic TSC exhibits an impressive PCE of 26.3% under reverse scan (a V_OC of 2.194 V, a J_SC of 14.6 mA/cm², and an FF of 0.82) and PCE of 26.1% under forward scan (a V_OC of 2.194 V, a J_SC of 14.7 mA/cm², and an FF of 0.81) (Fig. 5F, fig. S25, and table S3). To validate the reliability of our measurements, the perovskite-organic TSC was independently certified by the National Center of Inspection on Solar Photovoltaic Products Quality, which exhibits a PCE of 25.6% under reverse scan (fig. S26), representing one of the highest certified efficiencies reported to date (table S4). Furthermore, our devices exhibit excellent reproducibility, achieving an average PCE of 26.0 ± 0.17% with a narrow distribution (Fig. 5G). The EQE spectra confirm a well-balanced photocurrent of subcells in the TSCs, with integrated J_SC values of 14.6 and 14.5 mA/cm² for the perovskite and organic subcells, respectively (Fig. 5H). Last, we evaluated the operational stability of encapsulated perovskite-organic TSCs under different conditions. As shown in Fig. 5I, the tandem device showed no degradation after 600 s of MPP tracking under AM1.5G illumination. Furthermore, the TSC retained approximately 92% of its initial PCE after 1000 hours of continuous illumination under a full-spectrum white light-emitting diode (LED) lamp in ambient air (figs. S27 and S28).

DISCUSSION

In summary, we developed an approach to prepare halide-homogenized WBG perovskite films by modulating hydrogen-bonding interactions within the perovskite precursor. We found that solvent FM with appropriate hydrogen-bonding capability effectively balances the crystallization kinetics of I- and Br-based intermediate phases. Specifically, FM delays the crystallization of the Br-based phase, enabling a synchronous crystallization process that yields halide-homogenized WBG perovskite films. Leveraging this advance, we fabricated WBG PSCs that achieved a high PCE of 18.9% with an exceptional V_OC of 1.41 V. Last, a perovskite-organic TSC was fabricated, realizing a high efficiency of 26.3% (certified as 25.6%) with excellent photostability. This work provides a previously unexplored strategy for fabricating high-quality, halide-homogenized WBG perovskite films, promoting the development of highly efficient perovskite-based TSCs.

MATERIALS AND METHODS

Materials

All solvents and chemical materials were used as received without any further purification. Formamidinium iodide (FAI) (99.999%) was purchased from GreatCell Solar Materials. Caesium iodide (CsI) (99.9%) was purchased from Thermo Fisher Scientific. Lead bromide (PbBr₂)(99.999%, trace metals basis) was purchased from Energy Chemical. Lead iodide (PbI₂)(99.99%, trace metals basis), isopropanol (99.9%), DMF (99.8% anhydrous), and DMSO (99.8% anhydrous) were purchased from Sigma-Aldrich. NiO nanocrystals (VZHT02) were purchased from Shanghai Weizhu New Material Technology. PEDOT:PSS aqueous solution (Clevios P VP AI 4083), fullerene (C₆₀)(99.7%), and BCP (99.6%) were purchased from Xi’an Polymer Light Technology. Me-4PACz (>98%) and 2PACz (>98%) were purchased from TCI. FM (≥99.5% anhydrous), FA (Standard for GC, >99%), and TA (95%) were purchased from Beijing InnoChem Science & Technology Co. Ltd. MoO₃, PM6, BTP-eC9, and 1,4-diiodobenzene were purchased from Jiaxing Hepu Optoelectronic Technology Co. Ltd.

Preparation of perovskite and organic precursor solution

For WBG perovskite precursor solution (FA_0.8Cs_0.2PbI_1.5Br_1.5), the composition was formulated with molar ratios of FAI:CsI:PbI₂:PbBr₂ at 0.8:0.2:0.25:0.75. WBG perovskite precursor solution (1.6 M) was prepared using a mixed solvent comprising DMF (60% v/v) and DMSO (40% v/v). HBDS were then added to the WBG perovskite precursor solution at optimized volume fractions. For the organic precursor solution, PM6 and BTP-eC9 were dissolved in chloroform at a ratio of 1:1.2 (w/w), with PM6 at a concentration of 7.5 mg/ml. In addition, 1,4-diiodobenzene (10 mg/ml) was added as a solid additive.

Fabrication of WBG PSCs and TSCs

Patterned ITO substrates were sequentially cleaned using ultrasonication in deionized water, acetone, and isopropanol for 15 min each. After that, they were dried with a nitrogen flow and treated with ultraviolet (UV) ozone for 15 min. A NiO_X aqueous solution (10 mg/ml) was then spin-coated onto the substrates at 5000 rpm for 30 s, followed by annealing at 120°C for 30 min in air. After cooling to room temperature, a 1 mM solution of Me-4PACZ in isopropanol was spin-coated onto the NiO_X film at 4000 rpm for 30 s, and the sample was annealed at 100°C for 5 min. Subsequently, the blade-coater settings, including the gap between the blade and substrates, the operating speed, and the nitrogen (N₂) knife air pressure, were adjusted to 60 μm, 2 mm/s, and 60 psi, respectively. With these parameters, the WBG perovskite precursor was blade-coated onto the substrates and annealed at 105°C for 10 min. Subsequently, 20 nm of C₆₀, 5 nm of BCP, and 120 nm of Cu were deposited on top via thermal evaporation. For the fabrication of the subcells in the series connection of the WBG perovskite mini-modules, the ITO electrode, NiOx/Me-4PACz/WBG perovskite/C₆₀/BCP, and 80-nm Ag electrode were patterned using a laser scriber (GT-LS150-G) to create P1, P2, and P3 lines. The widths of the active area, P1, P2, and P3 are 6.7 mm, 30 μm, 120 μm, and 100 μm, respectively. The aperture areas of small devices and mini-modules are 0.061 cm² and 10.40 cm², respectively.

For the fabrication of TSCs, 1-nm-thick Au clusters and 10 nm MoO₃ film were deposited onto the substrates of ITO/NiO_X/Me-4PACZ/perovskite/C₆₀/BCP using thermal evaporation. Subsequently, 2-PACZ (0.3 mg/ml) was spin-coated on MoO₃ at a speed of 3000 rpm, with a thermal annealing of 100°С/5 min. The organic precursor solution was spin-coated on 2-PACZ at a speed of 2500 rpm, with a thermal annealing of 100°С/7 min. A PNDIT-F3N layer was spin-coated on the organic layer at a speed of 3000 rpm without annealing. The PNDIT-F3N was dissolved in methanol at a concentration of 1 mg/ml. Last, a Ag (120 nm) top electrode was thermally evaporated to complete the TSC. The device area is 0.08 cm².

Device characterizations

The J-V curves of all devices were tested under the illumination of AM 1.5G (100 mW cm⁻²) by a Keithley 2400 source meter together with a 3A solar simulator (CME-Sol 8040-3A, Microenerg Beijing Technology Co. Ltd). The areas of the solar cells were determined by a shade mask with an aperture area of 0.055 mm². The light intensity was calibrated using a 2 cm–by–2 cm reference monosilicon cell (Oriel PN 91150 V, Newport, USA), which was itself calibrated by the National Renewable Energy Laboratory. These measurements were performed in a nitrogen-filled glovebox with a scanning rate of 0.02 mV s⁻¹ and a delay time of 10 ms. The EQE measurements were carried out on a TRACQ-BASIC system using a lock-in amplifier coupled with a monochromator and a 500-W Xenon lamp. The system was verified with a calibrated silicon detector (PRL-12, Newport, USA) that has a known photo-response. For the measurement of the WBG perovskite and organic subcells, bias illumination from high-brightness LEDs with emission peaks at 550 and 780 nm, respectively, was used to saturate the other junction. Long-term photostability measurements were conducted under homebuilt LabVIEW-based LED illumination, simulating 1-sun intensity, in ambient conditions with a humidity range of 30 to 40%. The LED light source used in this study covers the wavelength range from 350 to 1050 nm (fig.S28) (38). The devices were encapsulated with cover glass and UV epoxy, which was cured under a UV LED lamp with a peak emission at 365 nm for 1 min.

Other characterizations

XRD patterns were generated using a D8 Advance x-ray diffractometer with monochromatized Cu-Kα radiation as the x-ray source, scanning at a rate of 6° per min. ¹H-NMR spectra were recorded on d6-DMSO with a Bruker 500-MHz instrument. FTIR spectra were obtained using a Spectrum 100 spectrophotometer in reflective mode. SEM images were collected via a RISE-MAGNA microscope with an electron beam accelerated at 5 kV. The SEM-EDX measurement in BSED (Z contrast) mode was performed to acquire morphology and elemental distribution at 15 kV. Depth-dependent PL spectra were captured using a confocal micro-Raman spectrometer (Renishaw inVia Qontor) with excitation at a wavelength of 532 nm. ToF-SIMS spectra were detected by an ION ToF-SIMS 5-100 system, with Bi³⁺ as the primary ion beam at 30 keV, an incident angle of 45°, a scanning area of 150 μm by 150 μm, a pixel resolution of 128 by 128, and a beam current of 0.48 pA. Depth-dependent UPS measurements were conducted using an AXIS UltraDLD system. GIWAXS measurements were performed at the beamline BL14B1 of the Shanghai Synchrotron Radiation Facility. Incident angles of 0.1°, 0.5°, and 1.0° were used. Data were collected every 0.5 s with an exposure time of 50 s. GIWAXS data were analyzed using Fit-2D and MATLAB software and presented in q-space coordinates. The optical transmission was characterized using an optical microscope (Leica Microsystems CMS Co. Ltd., LEICA DM2700M).

DFT calculation

The DFT calculation for all molecules was performed using Gaussian 16 [Revision C.01. Wallingford, CT (2016)]. All the molecules were optimized with the PBE0 functional combined with the basis set of def2-SVP, and the single point energy was calculated with the basis set of def2-TZVP. The dispersion corrections were carried out through the DFT-D3BJ method. The solvation effect was concerned by SMD method, and the solvent was defined as DMF. The binding energy is defined as the difference between the electron energy of the complex and the sum of the electron energies of the individual molecules in the free state, ignoring the thermodynamic correction.

Supplementary Materials

This PDF file includes:

Note S1

Figs. S1 to S28

Tables S1 to S4

References