Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Nature Energy (2026)
7
Metrics details
Antimony chalcogenide (Sb2(S,Se)3) is a promising candidate for next-generation photovoltaic materials due to its optoelectronic properties, high absorption coefficient and material availability. Hydrothermal deposition has advanced the technology, but there is a limited understanding of the underlying reaction mechanisms, often resulting in non-ideal valence band maximum gradient across the absorber thickness and high concentration of deep-level defects. Here we introduce sodium sulfide as an additive in the precursor solution to control reaction kinetics. This strategy enables a more uniform depth-dependent elemental distribution, flattens the unfavourable valence band maximum gradient across the depth and suppresses the formation of deep-level defects. We demonstrate an improvement in Sb2(S,Se)3 material quality, achieving a power conversion efficiency of 11.02%, with a certified value of 10.7 ± 0.37%. This work advances the efficiency for Sb2(S,Se)3 solar cells and provides insights to optimize the hydrothermal synthesis for this technology.
Antimony selenosulfide (Sb2(S,Se)3) has emerged as a particularly compelling absorber material. It features a simple binary composition and a quasi-one-dimensional (Q1D) crystal structure that imparts excellent optical absorption characteristics with absorption coefficients exceeding 105 cm−1 across the visible and infrared spectrum. This allows efficient power conversion efficiency (PCE) with ultrathin (hundreds of nanometres) absorber layers1. Its tunable bandgap (1.1–1.7 eV) and feasible semi-transparent architecture also opens the possibility of integration into tandem solar cells, offering a pathway to surpass the Shockley–Queisser limit of single-junction solar cells2. In particular, wide-bandgap compounds (>1.5 eV) can be a well-suited top cell candidate for Si-based tandem solar cells. On the other hand, its low-temperature processing compatibility further enhances its potential for up-scalable manufacturing. As such, Sb2(S,Se)3 is increasingly recognized as a promising candidate among next-generation chalcogenide photovoltaic materials, especially in the pursuit of tandem and ultrathin solar technologies.
Among various synthesis methods, hydrothermal deposition has gained increasing attention due to its low processing temperature, solution-based scalability and ability to produce highly crystalline, preferably oriented Sb2(S,Se)3 absorber layers. Notably, both the benchmark PCEs—10% for 1.5 eV-bandgap Sb2(S,Se)3 in 20202 and 10.7% for 1.3 eV-bandgap Sb2(S,Se)3 in 20253—are achieved using the hydrothermal method, highlighting its strong potential for efficient and low-cost device fabrication.
Despite this promise, further advancement of the efficiency remains hindered by fundamental challenges. Recent studies focus on improving bulk material quality through intentional doping4, band structure engineering5,6 and defect passivation7,8. However, the hydrothermal process occurs in a sealed autoclave under elevated temperature and pressure, making in situ observation difficult and thereby leaving the reaction pathway poorly understood. Whereas previous studies hypothesized that Sb and Se could bond through a reaction between antimony potassium tartrate, HSe− (a product formed when selenourea dissolves in an alkaline solution) and H+ without the involvement of sodium thiosulfate4, there still lacks systematic validation of the completed chemical reaction sequences either experimentally or theoretically. In particular, the decomposition mechanism of selenourea, and the competitive incorporation of S2− and Se2− anions into the growing Sb2(S,Se)3 lattice, remain elusive. Without a clear understanding of the underlying mechanism, state-of-the-art Sb2(S,Se)3 solar cells are still suffering from inhomogeneous depth-dependent composition induced unfavourable valence band maximum (VBM) gradient and high concentration of deep-level defects5.
To address these issues, we systematically investigate the hydrothermal reaction mechanism. Sodium sulfide (Na2S) is then accordingly introduced as an additive to regulate the hydrothermal reaction kinetics. The reversible hydrolysis of sulfide ions helps buffer the pH, preventing the time-dependent acidification and slowing the release of selenium. This controlled reaction kinetics leads to a homogeneous depth-dependent Se and S distributions, preventing the detrimental VBM gradient and benefiting hole transport. Furthermore, the formation of deep-level defects in this Sb-rich compound, SbS anti-sites and VS vacancies, are suppressed, reducing the trap centre density in Sb2(S,Se)3 bulk. By implementing this strategy, we achieve a champion PCE of 11.02%, with a certified efficiency of 10.70 ± 0.37%. These findings provide critical insights into the reaction kinetics of hydrothermal synthesis of Sb2(S,Se)3 and offer a viable pathway towards further performance improvements.
The hydrothermal synthesis of Sb2(S,Se)3 has traditionally been using antimony potassium tartrate (APT), sodium thiosulfate (STS) and selenourea (SU) as precursors, which we refer to as the control condition (denoted by ‘CT’ in the following discussion for convenience)2. It is evident that Sb2S3 is formed through the reaction between APT and STS without the involvement of SU (Path i in Fig. 1)9. However, when STS is excluded from the hydrothermal system, we find that SU alone is unable to react with APT to produce Sb2Se3, regardless of the solution pH (adjusted using a diluted NaOH). This finding suggests that SU is highly likely transformed into an intermediate species, assisted by STS, before reacting with APT to form Sb2Se3.
a, Raman spectrum of the residue obtained by vacuum-drying the solution after the hydrothermal reaction between STS and SU. The identified peaks at 343, 447 and 458 cm−1 represent typical Se–S stretching vibration of SeSO32−, S–S stretching vibration of S2O32 and v2 mode (symmetric bending vibration) of SO42−, respectively. Inset: the sample appearance. b, XRD pattern of the sediment after the two-step hydrothermal process. The suspension is centrifuged to collect the sediment, followed by post-annealing. The phase structure of stibnite Sb2S3 and Sb2Se3 is indexed by the PDF#42-1393 and PDF#15-0861, respectively. c,d, Schematics of the main hydrothermal synthesis paths under CT condition (c) and SS conditions (d). Blue, orange and mauve-pink arrows represent the three main synthesis paths over time, that is, Path i—Sb2S3 formation, Path ii—Sb2Se3 formation and the by-product path, respectively. Arrow width demonstrates the reaction rate. Intermediates, products and the by-products are boxed in green, light blue and red, respectively. Insets: the corresponding time profiles of reaction rate and pH, which reveal a steadier reaction rate under SS condition than CT condition.
To validate the hypothesis and identify the intermediate, we conducted a two-step hydrothermal experiment by separating the antimony precursor from chalcogenide precursors (details in Methods). In the first step, only STS and SU are added to the precursor solution. The cooled post-reaction solution remains transparent and clear. The Ramen spectrum of the residue after vacuum-drying the solution in Fig. 1a demonstrates typical Se–S stretching vibration of SeSO32− and S–S stretching vibration of S2O32− at 343 and 447 cm−1, respectively10,11,12,13. The overlapping peak at 458 cm−1 is attributed to the v2 mode (symmetric bending vibration) of SO42− 14. In the second step, APT is added to the remaining solution, and an fluorine-doped tin oxide (FTO)/CdS substrate is placed in the autoclave for another hydrothermal run. Brownish-red sediment forms at the bottom, and a thin film grows on the substrate, consistent with our typical one-step hydrothermal synthesis. X-ray diffraction (XRD) of the post-annealed sediment (Fig. 1b) confirms the presence of Sb2S3 and Sb2Se3. The same composition in the thin film is revealed by XRD (Supplementary Fig. 1) and Raman spectrum (Supplementary Fig. 2). Thermodynamic calculations indicates that the reaction between SU and STS happens spontaneously, leading to the formation of SeSO32− (Supplementary Note 1). On the basis of these findings, we conclude that SU reacts with S2O32− under the hydrothermal condition to form SeSO32− as an intermediate, which subsequently reacts with SbO+ to produce Sb2Se3 (Path ii in Fig. 1). This pathway is in line with the prior reports showing that Sb2(S,Se)3 can be synthesized when SU is replaced by Na2SeSO33,15.
We further investigate the reaction kinetics by assessing the effects of temperature, pH and precursor concentrations. According to the Arrhenius equation, the primary variables influencing the reaction rates in our system are temperature and reactant concentrations, including that of H+ (that is, pH). The pH of the precursor solution quickly drop to approximately 4.6 due to precursors’ hydrolysis16. As demonstrated in Fig. 1c,d, under these conditions, two competitive reaction pathways operate:
Path i (Reaction 1): direct formation of Sb2S3 from SbO+ and S2O32−.
Path ii (Reactions 2–3): SU reacts with S2O32− to form SeSO32−, which then reacts with SbO+ to yield Sb2Se3.
Path i is driven forward during the temperature ramp. However, low pH under the CT condition makes Reaction 1 proceed slowly but enables mild protonation at the C=Se bond in SU, which facilitates Reaction 2. As a result, S2O32− preferentially produces SeSO32− intermediate at the early stage. Because the Se–S bond in SeSO32− is weaker than the S–S bond in S2O32−, its cleavage lowers the Gibbs free energy barrier, making Sb–Se bond formation more favourable than Sb–S. As a result, SeSO32− is rapidly consumed via Reaction 3. Its exhaustion in turn drives Reaction 2 forward, leading to the rapid depletion of SU within a short period as illustrated in Fig. 1c and the inset plot. The subsequent pH drop and the shortage of SU at the late deposition stage may lead to a compositional gradient of anions across the Sb2(S,Se)3 thin film thickness, consistent with the EDS line scan presented later8. Notably, despite that the presence of a selenium gradient is of interest, sulfur remains the dominant anion in Sb2(S,Se)3, owing to its dramatically higher precursor concentration (80 mM STS versus 6 mM for selenourea).
To suppress the rapid and uncontrolled release of selenium, we introduce sodium sulfide (denoted by ‘SS’ in the following discussion) to increase and buffer the solution pH and to regulate above reaction kinetics (Fig. 1d). During precursor preparation (unsealed), dissolution and hydrolysis of sodium sulfide establish a dynamic equilibrium by generating HS−/H2S and OH−, and the pH stabilizes at around 6.2 (measured by a pH meter). The increased OH− promotes Path i (Reaction 1) while slowing Path ii for suppressing nucleophilic substitution at the C=Se bond in SU at the early deposition stage. As a result, this ensures a steadier Se supply throughout the hydrothermal process.
Beyond the desired products, a by-product, antimony hydroxide (Sb(OH)x), is inevitably formed during the hydrothermal process following Reaction 4 below4.
It would be converted to SbxOy upon post-annealing, but in the presence of HS−/S2−, in situ sulfuration occurs forming Sb2S3 through:
This is supported by the reduced oxygen content in the small-scale atomic percentage profiles across the absorber in Supplementary Fig. 5. Thus, Na2S not only buffers the pH and stabilizes Se release but also promotes in situ sulfuration, suppressing oxide by-products.
The inhomogeneous composition along the film thickness could be mitigated by the addition of Na2S to the precursor solution, as expected by the proposed chemical reaction kinetics. We thereby investigated the cross-sectional morphology and elemental distribution in Sb2(S,Se)3 films using transmission electron microscopy (TEM) and energy-dispersive spectroscopy (EDS) mapping and line scans.
Apparent voids are observed in the cross-section image of the CT sample in Fig. 2a. These voids may hinder carrier transport. Notably, the voids predominantly appear in the bottom half of the Sb2(S,Se)3 layer, which correspond to the fast early deposition stage when selenium is rapidly released from SU. The lower Gibbs free energy enables a preferable and faster formation of Sb2Se3, compared with Sb2S3. Such structural voids are effectively suppressed in the upper half of the Sb2(S,Se)3 film, where the deposition rate decreases as the selenourea becomes depleted. Contrastingly, upon the addition of Na2S, the formation of voids is almost completely prevented, as shown in Fig. 2b. This is attributed to a consistently slow deposition rate, resulting from a reduced concentration of effective reactants in the precursor solution and a slow but steady release of selenium. The resulting improvement on cross-sectional morphology is expected to enhance both fill factor (FF) and short circuit current density (Jsc) of the final device.
a,b, Cross-sectional TEM images and EDS mapping of the CT sample (a) and the SS sample (b). c,d, Atomic percentage profile of the CT sample (c) and the SS sample (d), along the yellow arrow shown in panels a and b, respectively. The cross-sectional samples are made using focused ionbeam (FIB). Elements including O, S, Se, Cd, Sn and Sb are monitored.
In addition to the morphology, the depth profile of anionic distribution is greatly modified. Anti-correlated elemental gradients are observed for sulfur and selenium in the CT sample in Fig. 2c, which aligns with the competitive incorporation behaviour of S and Se under fast Se release condition as aforementioned. By contrast, this compositional gradient is completely eliminated in the SS sample as depicted in Fig. 2d, owing to a steady concentration of SeSO32− maintained in the quasi-neutral solution throughout the entire deposition process. This ensures a consistent ratio of Sb2Se3 to Sb2S3 throughout the entire Sb2(S,Se)3 film, leading to improved compositional uniformity along the film depth.
The change in element distribution normally leads to a modulation of the energy band. The conduction band minimum (CBM) of Sb2(S,Se)3 is dominated by the density of states of Sb 5p orbital while the VBM is mainly determined by the density of states of S 3p and Se 4p orbitals17. Therefore, changes in the anionic composition may result in shifts in the valence band edges of Sb2(S,Se)3. This is supported by the cathodoluminescence (CL) measurements on the cross-sections of the CT and SS samples. The horizontally consistent intensity in the CL mapping of the CT sample in Fig. 3b indicates compositional uniformity at the same depth. However, a vertically gradual variation is evident, consistent with the elemental gradient, suggesting a bandgap gradient formed during the hydrothermal process. In comparison, this depth-profile gradient is not observed in the SS sample (Fig. 3e). Therefore, we extracted CL spectra along the vertical direction to compare the band structure and non-radiative recombination between the CT and SS samples.
a–c, SEM image (a), CL mapping (b) and CL spectra (c) along the distance from origin (as indicated by the orange arrow) of the cross section of the CT sample. d–f, Scanning electron microscopy (SEM) image (d), CL mapping (e) and CL spectra (f) along the distance from the origin (as indicated by the orange arrow) in the cross section of the SS sample. Origin is the starting point of the orange arrow in a and d. The ultrathin samples were made by FIB. In b and e, the brightness corresponds to the CL intensity within the selected band-pass wavelength window. Spectra line curves at varying depths are overlaid in a single diagram to visually illustrate the peak behaviours in Supplementary Figs. 7 and 8. The high energy peak in c and f originates from the CdS. g, Peak energy extracted from CL spectra in c and f. h, CL spectra at early deposition stage for both the CT and SS samples. i,j, Band-edge profiles of the CT (i) and SS (j) devices. The early-stage Sb2(S,Se)3 film was fabricated by reducing the hydrothermal reaction time to 20 min, with a final thickness of around 20 nm. The band positions of Sb2(S,Se)3 were measured by ultraviolet photoelectron spectroscopy (UPS) and UV–vis spectroscopy (details in Supplementary Fig. 10). Band-edge positions of other layers are from ref. 29. Blue circles with a plus sign represent free holes in the bulk, which need to transport along the arrows and finally collected by Spiro-OMeTAD. ΔE denotes the energy barrier for hole transport across Sb2(S,Se)3 in the CT condition due to the VBM gradient, which disappears under the SS condition. a.u., arbitrary units.
In the CT sample, a broad peak ranging from 850 to 1,000 nm is observed in the film deposited at early stage (near the CdS side), whereas a narrower peak shifting towards ~800 nm (indicating a higher bandgap and less Se incorporated) occurs in the subsequently deposited film, as shown in Fig. 3c. This trend confirms the proposed mechanisms that rapid Se release at the early hydrothermal stage leads to a depth-dependent Se composition under CT condition. In contrast, the CL peaks in the SS sample remain sharp and consistently centred around 850 nm in Fig. 3f, in line with a homogeneous Se/S ratio throughout the deposition process due to the controlled, gradual release of Se enabled by the Na2S additive.
The peak energy extracted from the CL spectra in Fig. 3g demonstrates the vertical bandgap gradient in the CT sample and its suppression in the SS sample. Moreover, spectral line curves at the early deposition stage reveals substantially higher CL intensity in the SS sample (Fig. 3h), where higher intensity indicates lower non-radiative recombination and thereby better carrier transport efficiency. This intensity gap occurs in all depths, as suggested in Supplementary Figs. 7–9, indicating suppressed non-radiative recombination in the SS sample. More recombination characteristics are discussed in ‘Deep-level defect suppression’.
The bandgap shift in the CT sample follows a quasi-linear relationship with depth, summarized from the CL spectra. The band-edge profiles of both CT and SS samples were thereby quantitively identified by measuring early-stage and full Sb2(S,Se)3 thin films (Supplementary Fig. 10). In the CT sample, where the Se/S ratio varies with depth, the VBM shifts downward from –5.47 eV to –5.57 eV, whereas the CBM remains relatively constant at around –4.02 eV, as illustrated in Fig. 3i. In this superstrate configuration, most short-wavelength photons are absorbed near the CdS/Sb2(S,Se)3 interface region, generating free electron-hole pairs on site. Because holes must travel through the entire absorption layer before being collected by the hole transport layer (HTL), the downward VBM gradient acts as a barrier, hindering hole transport and collection, which negatively impacts device performance6. In contrast, the SS sample exhibits a flat VBM and CBM profile at –5.54 eV and –4.02 eV, respectively (Fig. 3j), attributing to the stable Se/S ratio throughout the film. This flattened VBM gradient lowers the energy barrier for carrier transport, thereby improving the optoelectronic performance of a Sb2(S,Se)3 solar cell.
Optical Deep-level Transient Spectroscopy (O-DLTS) was employed to characterize point defects as CL results indicated severer non-radiative recombination in the CT sample than that in the SS sample. Figure 4a,b presents two negative peaks detected in both samples under different pulse voltages, corresponding to minority-carrier (hole) traps (H1 and H2) in the n-type Sb2(S,Se)3 layer, in line with findings in previous reports regarding the deep-level defects in hydrothermally deposited Sb2(S,Se)3 bulk2. The Arrhenius plots in Fig. 4c illustrate thermal emission rates as a function of reciprocal temperature, from which key parameters—including activation energy (ET), capture cross section (σ) and trap density (NT)—were extracted and are detailed in Supplementary Table 2. Notably, similar trends were observed across different pulse voltages, indicating consistency in defect characteristics. Therefore, the results obtained by using a 0.2 V pulse voltage were selected for further discussion.
a,b, O-DLTS signals from the CT (a) and SS (b) samples with pulse voltages of 0.2 V, 0.4 V and 0.6 V. Two negative peaks (H1 and H2) represent hole traps (donor-type defects) in Sb2(S,Se)3. c, Arrhenius plots derived from O-DLTS signals. vth, thermal velocity. The data points in c were obtained by calculating the internal transients included in the O-DLTS signals using the discrete Laplace transform. The solid lines represent linear fits to the data points. d,e, Schematic diagrams of energy bands and defect levels for the CT and SS samples under a pulse voltage of 0.2 V. Ec, Ev and EF stand for CBM, VBM and Fermi level, respectively. Energy levels are labelled with their corresponding values in electron volts (eV). The cone-shaped shaded regions illustrate the capture cross section area (opening width) and density (extent) of defect. f, Values of NT and (σ×NT) at H1 and H2 defects under a pulse voltage of 0.2 V. A break is applied on the right y axis to clearly display data that differ by one to two orders of magnitude.
Analysis of H1 and H2 energy levels revealed that these defects are positioned approximately 0.666–0.668 eV and 0.763–0.764 eV above the VBM in both samples as demonstrated in Fig. 4d,e. This indicates that the defect type remains unchanged upon the incorporation of Na2S. According to the first-principles calculations, H1 and H2 are assigned to sulfur vacancy VS and SbS anti-site defects given their relatively low formation energies and appropriate transition levels18,19,20. Both VS and SbS function as deep-level defects, possessing activation energies far exceeding 0.025 eV (ref. 21). Consequently, they are difficult to ionize but serve as effective recombination centres, facilitating trap-assisted Shockley–Read–Hall (SRH) recombination. The SRH model [τ ∝ (σ × NT)−1] indicates that carrier lifetime is inversely correlated with the product of capture cross section and trap density22,23. Thus, the SS sample is anticipated to possess a longer carrier lifetime owing to a notably lower σ × NT value, approximately two orders of magnitude lower than that of the CT sample as Fig. 4f shows (H1, 3.76 × 10−3 cm−1 versus 1.368 × 10−1 cm−1; H2, 3.268 × 10−2 cm−1 versus 1.872 cm−1). This substantial reduction in defect density suggests that the addition of Na2S in precursor solution effectively suppresses the formation of deep-level defects during the hydrothermal deposition of Sb2(S,Se)3.
The formation of deep-level sulfur vacancy VS and anti-site SbS defects is a critical issue in Sb2(S,Se)3 thin films, substantially reducing carrier lifetime and device efficiency. Rapid reaction kinetics, often induced by high precursor concentrations or elevated temperatures under the CT condition, tend to drive the system away from thermodynamic equilibrium, promoting the formation of unsaturated covalent bonds and such deep-level point defects due to insufficient incorporation of sulfur atoms into the lattice. In contrast, the slower reaction rate under the SS condition allows the precursor sources more time to incorporate into energetically favourable lattice sites. This facilitates the growth of films with improved stoichiometry and structural order, thereby effectively suppressing the formation of VS and SbS defects. Moreover, the effective suppression function also arises from the existence of S2− in the solution by restoring local stoichiometry and eliminating mid-gap states24. The formation of Sb2(S,Se)3 requires high activation energy to break the S–S/S–Se covalent bond in S2O32−/SeSO32− and ionize the non-central sulfur/selenium. In contrast, S2− ions can directly bond with Sb3+ with much lower activation energy, minimizing unsaturated covalent bonds in the crystal lattice during the deposition of Sb2(S,Se)3. This defect suppression mechanism reduces non-radiative recombination in the SS sample, leading to an increased carrier concentration and improved optoelectronic performance.
The carrier concentration for the CT and SS samples was calculated from the capacitance–voltage (C–V) measurements, as a function of the depletion region width under different voltage biases. At 0 V bias, the depletion region widths were determined to be 317 nm for the CT sample and 262 nm for the SS sample, in good agreement with the thicknesses of the Sb2(S,Se)3 absorber layers measured via cross-sectional TEM images in Supplementary Fig. 12. This confirms that both Sb2(S,Se)3 absorber layers are fully depleted, consistent with the previous report1. To ensure that the measurement reflects the intrinsic properties of the Sb2(S,Se)3 bulk, a positive bias of 0.4 V was applied to shift the depletion region boundary into the absorber layer. Under this condition, the net carrier concentration in the Sb2(S,Se)3 bulk increases from 4.83 × 1016 cm−3 in the CT sample to 1.72 × 1017 cm−3 in the SS sample (Fig. 5a), owing to the suppression of deep-level defects discussed earlier. The increased carrier concentration lowers the series resistance of the Sb2(S,Se)3 layer, thereby facilitating hole transport within the absorber, together with the elimination of the energy barrier via flattening unfavourable VBM gradient. These improvements contribute to a dramatic improvement of FF from 66.09% to 69.02% as shown in Fig. 5b (data summarized in Table 1).
a, Mott–Schottky plots demonstrating effective carrier concentration in Sb2(S,Se)3 bulk derived from C–V measurement, which is represented by NCV on the y axis. Wd on the x axis represents the depletion region width under the applied bias. b,c, J–V curves (b) and EQE curves (c) of CT and SS devices. d–g, Statistic distributions of Voc, Jsc, FF and PCE for the CT (n = 9) and SS (n = 22) devices. The minima, maxima, centre, bounds of box and whiskers and percentile of the box plots in d–g share the same definition, as demonstrated in the legend of g.
Source data
With illumination coming from the CdS side, the external quantum efficiency (EQE) in the short-wavelength region is primarily governed by the collection efficiency of minority carriers near the CdS/Sb2(S,Se)3 interface25. Holes are minority carriers in our n-type Sb2(S,Se)3 absorber layer and need to travel through the whole absorber before being extracted by the HTL1. Thus, the combined effect of the prevented unfavourable VBM gradient across the depth and the reduced series resistance facilitates more efficient hole transport across the bulk, as evidenced by the obvious improved EQE response in the short-wavelength region shown in Fig. 5c. Moreover, the suppression of deep-level defects in the bulk lowers the density of recombination centres, allowing a greater fraction of photogenerated carriers to be collected rather than to recombine. This effect is particularly pronounced for carriers generated at deeper regions by long-wavelength photons, leading to an improved EQE response in the long-wavelength region. As a result, the integrated Jsc increases from 22.35 mA cm−2 for the CT sample to 24.54 mA cm−2 for the SS sample, closely aligning with Jsc values measured under AM 1.5 G illumination—22.71 mA cm−2 for the CT sample and 24.96 mA cm−2 for the SS sample.
Therefore, even with a slight Voc loss due to the thinner absorber, the introduction of Na2S in the hydrothermal system improves both the band alignment and material quality. Benefiting from these advancements, our Sb2(S,Se)3 solar cell achieved a PCE of 11.02%, with a certified value of 10.7 ± 0.37%. The distributions of Voc, Jsc, FF and PCE for a batch of CT and SS cells are presented in Fig. 5d–g, revealing consistent trends in optoelectronic parameters and reproducibility by applying the Na2S additive strategy.
Our work explores the chemical mechanisms governing the hydrothermal deposition of Sb2(S,Se)3, revealing the synthesis pathways and the competitive interplay between sulfur and selenium and their impact on the film growth. Guided by this advancement in the understanding, Na2S was utilized as an additive in the precursor solution to buffer the pH and regulate reaction kinetics by its reversible hydrolysis. The consequently controlled release rate of Se from selenourea stabilizes the depth-dependent anionic distribution and flattens the unfavourable VBM gradient in the final Sb2(S,Se)3 film, thereby effectively improving the carrier transport efficiency. Moreover, the formation of deep-level VS and SbS defects is effectively suppressed, dramatically reducing the trap density and increasing the net carrier concentration in the Sb2(S,Se)3 bulk. Although Voc drops slightly due to the thinner absorber, these remarkable improvements in bulk material quality enhance Jsc and FF, ultimately yielding a PCE of 11.02%. Moreover, our insights on the fundamental chemical mechanisms governing the hydrothermal process provide a theoretical foundation for further optimization of Sb2(S,Se)3 solar cells fabricated via this method.
FTO-coated glasses (purchased from Liaoning Libra Tech.) were pre-cleaned by deionized (DI) water, isopropanol, acetone and ethanol (Materials without specifically pointed out were all purchased from Sigma-Aldrich Chemical Sciences) for 20 min sequentially. Further treatment by UV ozone equipment (Ossila, L2002A3)) was employed on the FTO layer before use. The cadmium sulfide layer was deposited on FTO substrate by the chemical bath deposition method2. Then CdCl2 (20 mg in 1 ml methanol) was spin coated at 3,000 rpm for 30 s and samples were heated at 400°C for 10 min in air to improve the grain size of the CdS layer and incorporate Cl and O into the film26. This results in a better surface condition for epitaxial growth of Sb2(S,Se)3 in an autoclave reaction. After that, the Sb2(S,Se)3 layer was deposited by the hydrothermal method. In the control condition, 20 mM antimony potassium tartrate, 80 mM sodium thiosulfate tetrahydrate and 30 mg selenourea (99.97%, Alfa Aesar) were added sequentially into a Teflon autoclave (45 ml nominal volume) with 40 ml DI water. Under the SS condition, 1.155 × 10−4 mol Na2S (delivered as a 0.03 g ml−1 pre-prepared solution) was added to the solution in the order after APT and STS but before SU. The hydrolysis of Na2S consumed approximately 2.5 × 10−5 mol of H+. This reaction also produces a maximum of 1.25 × 10−5 mol H2S, which remains fully dissolved in the solution, given its solubility in water (0.152 g 40 ml−1), preventing its escape from the reaction system. The solution turns orange and non-transparent immediately. Then 30 mg selenourea was added. The substrate inclined to the inside surface at 75°. Then the autoclave was sealed into a stainless-steel acid digestion vessel (Parr instrument company, 4749 PTFE) and heated in an oven (Across International, FO-19123) up to 135 °C for 140 min. After natural cooling to room temperature, the samples were taken out and swilled with DI water and ethanol before being dried for 1 min in a vacuum dryer (Across International, AT-26) where the temperature was kept at 110 °C and the air pressure was 10 kPa. Afterwards, samples were annealed at 350 °C for 10 min for crystallization. Subsequently, the spiro-OMeTAD was spin coated at 3,000 rpm for 30 s in a glovebox. The precursor solution was prepared by mixing 36.6 mg of spiro-OMeTAD powder (Lumitech), 14.5 μl of 4-tert-butylpyridine (tBP) and 9.5 μl of a 520 mg ml−1 lithium trifluoromethanesulfonyl (Li-TFSI) together in acetonitrile in 1 ml of chlorobenzene. Annealing in air on 100 °C for 10 min was conducted then. Finally, gold anode was deposited by thermal evaporation under a pressure of 5 × 10−4 Pa and anti-reflection film (MOSMITE, MITSUBISHI chemical group) was applied to the glass side to enhance the light entry into the absorber layer.
In the first hydrothermal step, 80 mM sodium thiosulfate tetrahydrate and 30 mg selenourea (99.97%, Alfa Aesar) were added into a Teflon autoclave (45 ml nominal volume) with 40 ml DI water. Then the autoclave was sealed into a stainless-steel acid digestion vessel (Parr instrument company, 4749 PTFE) and heated in an oven (Across International, FO-19123) up to 135 °C for 140 min. After natural cooling to room temperature, 3–5 drops were pipetted onto a glass slice and dried in a vacuum oven (Across International, AT-26) at 60 °C and 10 kPa for 30 min. The resulting white residue was used for Raman measurements. In the second step, 20 mM antimony potassium tartrate was added to the remaining solution and an FTO/CdS substrate (identical to that used in our typical one-step hydrothermal synthesis) was placed in the autoclave. The autoclave was sealed and heated in a way absolutely the same as that in the first step. After natural cooling to room temperature, the suspension was centrifuged to collect the sediment. The sediment and the thin film on the FTO/CdS substrate were then annealed at 350 °C in a N2-filled atmosphere for 10 min, followed by XRD and Raman measurements to determine the composition.
Raman spectra were measured by inVia Reflex Raman (532, 785, 830, 1064 nm) with 532 nm excitation laser. XRD measurements were conducted by Panalytical Xpert Materials Research diffractometer system. The cross-section TEM samples with thickness around 100 nm were prepared by a focused ion beam (FIB) equipped with micro-manipulator for in situ lift out (Thermo Fisher Helios G4 PFIB). The transmission electron microscopy (TEM) specimen was prepared using a plasma FIB (Thermo Fisher, Helios G4 PFIB) equipped with a micro-manipulator for in situ lift-out processes. The TEM images and element mapping were measured by a JEOL F200 system with a dispersive X-ray detector. The bandgap was measured by UV–vis (SOLID 3700). The cathodoluminescence (CL) specimen was also prepared by the plasma FIB. Band-edge positions were determined by using UV–vis light absorption (Perkin Elmer Lambda 1050 Spectrophotometer) and UPS (ESCALAB250Xi, Thermo Scientific). The specimen underwent ion beam polishing (Fischione, Nanomill 1040) at 500 V before measurement. Subsequently, spectral-resolved CL mapping was performed using the system (Delmic, SPARC), which was integrated with the FEI Nova Nano SEM 450 field-emission SEM.
J–V characteristics were determined using a Keithley 2400 apparatus under an AM 1.5 illumination with an intensity of 100 mW cm−2 provided by a standard xenon-lamp-based solar simulator (ABET Sun 3000 IV Tester). The voltage step size is 0.0047–0.0054 V. Before the test, the illumination intensity of a solar simulator was calibrated by a monocrystalline silicon reference cell (Oriel P/N 91150 V, with KG-5 visible colour filter), previously standardized by the National Renewable Energy Laboratory. The EQE was measured by a single-source illumination system (halogen lamp) coupled with a monochromator (PV Measurements QEX7 Spectral Response). The C–V characterization was performed using Keysight E4990A Impedance Analyser at a frequency of 10k Hz in the dark, ranging from −1V to 1 V. O-DLTS was conducted using Phystech FT-1230 HERA with a 10-mW and 635-nm wavelength laser source, a capacitance meter (Boonton 7200 (HF 1 MHz)), a temperature controller (JANIS 21308) from Janic Research Company and a Cryostat (Lakeshore 335,336). O-DLTS (optical deep-level transient spectroscopy) monitors the capacitance change (ΔC) of the device under short bias/voltage pulses, while sweeping temperature. Each peak corresponds to a thermally activated defect level releasing carriers at a characteristic rate. The shift with pulse voltage indicates how traps respond to different filling conditions (closer vs deeper traps in bandgap)27. To record the O-DLTS spectra of the solar cell devices, the frequency was varied in a range of 1 Hz to 75 kHz to measure the capacitance. The samples were placed in a liquid helium cryostat with a temperature scan ranging from 120 to 420 K at 2 K intervals. The reverse bias, optical pulse width and period width were −0.4 V, 10 ms and 100 ms, respectively. A pulse voltage was applied to modulate the depletion region width, enabling the extraction of defect information at various depths by analysing transient capacitance decay at different temperatures. To systematically compare defect states at different depths and determine an optimal pulse voltage, measurements were conducted at 0.2 V, 0.4 V and 0.6 V, which are lower than the open-circuit voltage. This approach also minimized the occurrence of erroneous peaks arising from capacitor bridge recovery delays. For each chosen pulse voltage, ΔC(T) is measured. Two curves per voltage usually correspond to different Laplace transform components (emission branches of traps) extracted from the raw transient. Here we overlay both the ‘raw Laplace-transformed spectrum’ and the ‘baseline-corrected version’, which produces apparent duplication28. Each trap emission rate ({e}_{n}) in O-DLTS follows an Arrhenius form:
Where, σ = capture cross section, vth = thermal velocity, NC = effective density of states in conduction band, Ea = activation energy of the trap, T is temperature and kB is the Boltzman constant.
O-DLTS peaks are assigned a characteristic emission rate at the peak temperature. Plotting (mathrm{ln}(frac{{e}_{n}}{{T}^{2}})mathrm{vs}frac{1}{T}) yields a straight line with a slope of (-frac{{E}_{{rm{a}}}}{{k}_{{rm{B}}}}). In Fig. 4c, those linear fits (H1, H2 for different voltages) are the Arrhenius plots. The points are extracted from the O-DLTS peaks (temperature positions), and the straight lines give the activation energies listed in Fig. 4d, e.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
The data that support the findings of this study are available within the Article and its Supplementary Information. Source data are provided with this paper.
Qian, C. et al. 9.6%-Efficient all-inorganic Sb2(S,Se)3 solar cells with a MnS hole-transporting layer. J. Mater. Chem. A 10, 2835–2841 (2022).
Article Google Scholar
Tang, R. et al. Hydrothermal deposition of antimony selenosulfide thin films enables solar cells with 10% efficiency. Nat. Energy 5, 587–595 (2020).
Article Google Scholar
Dong, J. et al. Carrier management through electrode and electron-selective layer engineering for 10.70% efficiency antimony selenosulfide solar cells. Nat. Energy 10, 857–868 (2025).
Article Google Scholar
Zhao, Y. et al. Regulating deposition kinetics via a novel additive-assisted chemical bath deposition technology enables fabrication of 10.57%-efficiency Sb2Se3 solar cells. Energy Environ. Sci. 15, 5118–5128 (2022).
Article Google Scholar
Huang, L. et al. Temperature-gradient solution deposition amends unfavorable band structure of Sb2(S,Se)3 film for highly efficient solar cells. Angew. Chem. 136, e202406512 (2024).
Article Google Scholar
Chen, J. et al. Se-elemental concentration gradient regulation for efficient Sb2(S,Se)3 solar cells with high open-circuit voltages. Angew. Chem. 63, e202409609 (2024).
Article Google Scholar
Li, J. et al. Defects passivation via potassium iodide post-treatment for antimony selenosulfide solar cells with improved performance. Adv. Funct. Mater. 33, 2211657 (2023).
Article Google Scholar
Che, B. et al. Thermally driven point defect transformation in antimony selenosulfide photovoltaic materials. Adv. Mater. 35, 2208564 (2023).
Article Google Scholar
Wang, S. et al. A novel multi-sulfur source collaborative chemical bath deposition technology enables 8%-efficiency Sb2S3 planar solar cells. Adv. Mater. 34, 2206242 (2022).
Article Google Scholar
Tugarova, A. V., Vladimirova, A. A., Dyatlova, Y. A. & Kamnev, A. A. Raman spectroscopic and TEM monitoring of selenite and selenate reduction by the bacterium Azospirillum thiophilum with the formation of selenium (0) nanoparticles: effects of sulfate. Spectrochim. Acta, Part A 329, 125463 (2025).
Article Google Scholar
Li, W. et al. Role of S∕Se ratio in chemical bonding of As–S–Se glasses investigated by Raman, X-ray photoelectron, and extended X-ray absorption fine structure spectroscopies. J. Appl. Phys. 98, 053503 (2005).
Ward, A. T. Molecular structure of dilute vitreous selenium-sulfur and selenium-tellurium alloys. J. Phys. Chem. 74, 4110–4115 (1970).
Article Google Scholar
Haigh, J., Hendra, P., Rowlands, A., Degen, I. & Newman, G. Raman spectroscopy of thiosulphates. Spectrochim. Acta, Part A 49, 723–725 (1993).
Article Google Scholar
Buzgar, N., Buzatu, A. & Sanislav, I. V. The Raman study on certain sulfates. Ann. Stiintifice Univ. Alexandru Ioan Cuza 55, 5–23 (2009).
Google Scholar
Dong, J. et al. Low-cost antimony selenosulfide with tunable bandgap for highly efficient solar cells. Small 19, 2206175 (2023).
Article Google Scholar
Huang, Y. et al. Chemical insight into the hydrothermal deposition of Sb2(S,Se)3 towards delicate microstructure engineering. J. Mater. Chem. A 10, 9892–9901 (2022).
Article Google Scholar
Dai, G. et al. Selenium-rich Sb2(S,Se)3 thin films deposited via sequential chemical bath deposition for high-efficiency solar cells. Adv. Funct. Mater. 35, 2415215 (2025).
Savory, C. N. & Scanlon, D. O. The complex defect chemistry of antimony selenide. J. Mater. Chem. A 7, 10739–10744 (2019).
Article Google Scholar
Huang, M. et al. More Se vacancies in Sb2Se3 under Se-rich conditions: an abnormal behavior induced by defect-correlation in compensated compound semiconductors. Small 17, 2102429 (2021).
Article Google Scholar
Huang, M., Xu, P., Han, D., Tang, J. & Chen, S. Complicated and unconventional defect properties of the quasi-one-dimensional photovoltaic semiconductor Sb2Se3. ACS Appl. Mater. Interfaces 11, 15564–15572 (2019).
Article Google Scholar
Lian, W. et al. Revealing composition and structure dependent deep-level defect in antimony trisulfide photovoltaics. Nat. Commun. 12, 3260 (2021).
Article Google Scholar
Hall, R. N. Electron-hole recombination in germanium. Phys. Rev. 87, 387 (1952).
Article Google Scholar
Shockley, W. & Read, W. Jr Statistics of the recombinations of holes and electrons. Phys. Rev. 87, 835 (1952).
Article Google Scholar
Zhang, S. et al. Eco-friendly Zn–Ag–In–Ga–S quantum dots: amorphous indium sulfide passivated silver/sulfur vacancies achieving efficient red light-emitting diodes. ACS Appl. Mater. Interfaces 15, 50254–50264 (2023).
Article Google Scholar
Li, J. et al. Defect-resolved effective majority carrier mobility in highly anisotropic antimony chalcogenide thin-film solar cells. Solar RRL https://doi.org/10.1002/solr.202000693 (2021).
Wang, L. et al. Ambient CdCl2 treatment on CdS buffer layer for improved performance of Sb2Se3 thin film photovoltaics. Appl. Phys. Lett. https://doi.org/10.1063/1.4932544 (2015).
Lang, D. Deep-level transient spectroscopy: a new method to characterize traps in semiconductors. J. Appl. Phys. 45, 3023–3032 (1974).
Article Google Scholar
Schroder, D. K. Semiconductor Material and Device Characterization (John Wiley & Sons, 2015).
Qian, C. et al. Bifacial and semitransparent Sb2(S,Se)3 solar cells for single-junction and tandem photovoltaic applications. Adv. Mater. 35, 2303936 (2023).
Download references
This research received funding from the Australian Renewable Energy Agency (ARENA) as part of ARENA’s Transformative Research Accelerating Commercialization (TRAC) Program. X.H. acknowledges financial support of the Australian Research Council Future Fellowship (FT190100756). T.C. acknowledges financial support from the Fundamental Research Funds for the Central Universities (WK2490000002) for the O-DLTS measurements. K.S. acknowledges the Australian Research Council Discovery Early Career Researcher Award (DE230100021). We acknowledge the facilities and the scientific and technical assistance of the Electron Microscope Unit (EMU) and the surface analysis laboratory (SAL), SSEAU, MWAC at the University of New South Wales (UNSW). The views expressed herein are of the authors but not of the Australian Government, ARENA or ARC.
Open access funding provided through UNSW Library.
Australian Centre for Advanced Photovoltaics, School of Photovoltaic and Renewable Energy Engineering, University of New South Wales, Sydney, New South Wales, Australia
Chen Qian, Kaiwen Sun, Jialiang Huang, Jialin Cong, Mingrui He, Zhen Li, Ziyue Feng, Xu Liu, Martin Green & Xiaojing Hao
Department of Materials Science and Engineering, School of Chemistry and Materials Science, University of Science and Technology of China, Hefei, People’s Republic of China
Junjie Yang, Rongfeng Tang & Tao Chen
Deep Space Exploration Laboratory, Hefei, China
Tao Chen
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
X.H. supervised the project at the University of New South Wales. C.Q. conceived the original idea, designed the experiments, fabricated the devices, conducted all characterizations except those specifically assigned to others, analysed data and wrote this paper. J.C. prepared cross-section samples for EDS and CL mapping. J.H. conducted the EDS, CL and C–V measurements, processed data and contributed to the analysis. J.Y. conducted O-DLTS measurements and processed the data. M.H. conducted Raman of the final thin film after two-step hydrothermal synthesis. Z.F. and Z.L. conducted XRD measurements on the synthesized powder of the two-step hydrothermal process and provided relevant analysis. X.L. provided access to a glovebox for depositing spiro-OMeTAD. R.T. offered suggestions on establishing the baseline for the reference sample. M.G provided suggestions on the experiment design. X.H., T.C., J.H. and K.S. revised this paper with all authors commenting on the paper.
Correspondence to Chen Qian, Jialiang Huang, Tao Chen or Xiaojing Hao.
The authors declare no competing interests.
Nature Energy thanks Susanne Siebentritt and the other, anonymous, reviewers for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Notes 1–4, Figs. 1–14 and Tables 1 and 2.
Source data for J–V curves in Fig. 5b and statistical distribution of data in Fig. 5d–g.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Reprints and permissions
Qian, C., Sun, K., Huang, J. et al. Regulation of hydrothermal reaction kinetics with sodium sulfide for certified 10.7% efficiency Sb2(S,Se)3 solar cells. Nat Energy (2026). https://doi.org/10.1038/s41560-025-01952-0
Download citation
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s41560-025-01952-0
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Advertisement
Nature Energy (Nat Energy)
ISSN 2058-7546 (online)
© 2026 Springer Nature Limited
Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.
