Zn doping modulation of carrier transport properties at the back interface of antimony sulfide solar cells

INTRODUCTION

In recent years, antimony-based solar cells have achieved rapid development due to their environmentally friendly characteristics and excellent physical properties^[1,2]. Among them, antimony sulfide (Sb₂S₃) has gained significant attention as a promising light-absorbing material for thin-film solar cells, thanks to its near-ideal bandgap of 1.7 eV, high visible-light absorption coefficient (> 5 × 10⁴ cm^-1), low toxicity, and excellent environmental stability^[3,4]. Since Savadogo’s pioneering work on chemically deposited Sb₂S₃ films and photoelectrochemical cells in 1992^[5], substantial progress has been made^[6,7]. Solution-based techniques, particularly chemical bath deposition (CBD) and hydrothermal synthesis, have dominated due to their low equipment requirements and scalability. Notably, in 2025, Shen et al. achieved a record-breaking 8.26% power conversion efficiency (PCE) by developing ethylenediaminetetraacetic acid disodium salt as an additive to regulate the reaction kinetics for Sb₂S₃ deposition^[8]. However, these efficiencies remain significantly below the Shockley-Queisser (S-Q) limit (28.6%)^[9]. The S-Q analysis predicts maximum values of open-circuit voltage (V_oc) = 1.402 V, short circuit current density (J_sc) = 22.5 mA/cm², and fill factor (FF) = 91%. In contrast, state-of-the-art devices (e.g., 8.26% PCE) exhibit substantially lower parameters: V_oc = 0.707 V (50.4% of limit), J_sc = 17.91 mA/cm² (79.6%), and FF = 64.81% (71.2%)^[8]. The severe deficit in V_oc is particularly critical, attributed primarily to interfacial recombination, inefficient carrier collection, and intrinsic material defects. Among them, suppressing interface recombination to enhance the extraction efficiency of electrons and holes has become an effective way to improve device performance.

Annealing, as a crucial step after the deposition of Sb₂S₃ thin films, should be carried out at 300-450 °C, which drives the transformation of the amorphous phase to the crystalline phase through thermal rearrangement and simultaneously enhances the ion diffusion rate^[10]. However, this process faces multiple challenges: Firstly, the essential annealing at high temperatures, aggravating the re-evaporation of sulfur and Sb₂S₃ due to their high saturated vapor pressure, leads to surface defects, film non-uniformity, and even exposure of the underlying CdS layer, which finally disrupts the integrity of the p-n junction. Secondly, residual oxygen reacting with antimony forms harmful Sb₂O₃ secondary phases at high temperature, which not only disrupts interface flatness and continuity but also introduces carrier recombination centers, significantly reducing device efficiency^[11]. Thirdly, the re-crystallization of Sb₂S₃ delivers a low-level position of valence band energy which creates severe band mismatch with the back electrode. To address the above challenges, researchers have explored the introduction of stable metal oxide buffer layers at the back interface during annealing. For instance, a Ni/Ti co-doped MoO₂ hole transport layer (HTL) can effectively extract holes from the metal halide perovskite layer, achieving a PCE of 18.1% in perovskite solar cells^[12]. When WO_3-x is used as the HTL, the formation of Sb-W bonds at the Sb₂Se₃/WO_3-x interface creates an interface dipole, which not only reduces the hole extraction barrier but also effectively suppresses back interface recombination in Sb₂Se₃ devices^[13]. Additionally, first-principles calculations by Professor Chen's team demonstrated that Zn doping in Sb₂S₃ can lower the Fermi level and increase the conductivity^[14,15]. And Tang et al. further confirmed that solution-processed Zn doping can elevate the back surface energy level of Sb₂S₃, improve carrier transport, and reduce interface recombination^[16]. These strategies based on oxide buffer layers or metal element doping all provide effective ways to improve the photovoltaic performance of Sb₂S₃.

Building on interface engineering and doping strategies, this work employs an ultrathin ZnO layer as an oxygen diffusion barrier, referred to as a “breathing membrane”. This layer enables controlled oxygen passivation of the underlying Sb₂S₃ film via a lattice-vacancy-mediated mass transfer mechanism, effectively inhibiting the formation of Sb₂O₃ and reducing interface defects, etc. Critically, the selection of ZnO also considers potential Zn²⁺ substitutional doping at the Sb₂S₃/oxide interface, which modulates interfacial energy levels and enhances carrier concentration. This synergistic approach concurrently optimizes the phase composition, morphology, and electrical properties of the Sb₂S₃ back surface, thereby improving carrier transport efficiency. Ultimately, a PCE of 7.00% was achieved in a fully inorganic fluorine-doped tin oxide (FTO)/CdS/Sb₂S₃/PbS/Carbon/Ag architecture, offering a novel route toward more efficient and stable Sb₂S₃ solar cells.

EXPERIMENTAL

RESULTS AND DISCUSSION

The surface morphology of Sb₂S₃ precursor films, with and without the ZnO modification layer, was first analyzed by scanning electron microscopy (SEM). As depicted in Supplementary Figure 1, the deposition of an ultrathin ZnO layer did not significantly alter the initial morphology; both samples displayed characteristic large-grained structures formed from agglomerated nanocrystals, typical of hydrothermally grown Sb₂S₃ films. However, post-annealing morphological evolution revealed stark contrasts, as shown in Figure 1. The unmodified sample (W/O ZnO) formed ~50 nm Sb₂O₃/Sb₂S₃ particles and ~40 nm Sb₂O₃ nanobelts at grain boundaries. By contrast, ZnO-coated films exhibited notable morphological refinements: the 20 s ZnO sample showed smaller particles (~30 nm) and narrower nanobelts (~20 nm), with coarsening observed at longer deposition times. This coarsening is attributed to ZnO-mediated suppression of Sb₂S₃ volatilization, which increases the local Sb₂S₃ availability for oxidation and promotes Sb₂O₃ growth supported by subsequent compositional analysis.

Figure 1. Surface and cross-sectional morphologies of Sb₂S₃ thin films (A) without ZnO and with ZnO deposited with (B) 20 s, (C) 40 s, (D) 80 s, (E) 200 s and (F) 400 s. SEM: Scanning electron microscope; W/O: without ZnO.

Cross-sectional SEM imaging (the inset of Figure 1) corroborated these findings. The control film (W/O ZnO) displayed a highly uneven thickness (~350 nm) due to extensive Sb₂S₃ loss during annealing, along with the presence of large Sb₂O₃ surface particles. Incorporation of a 20 s ZnO layer mitigated volatilization, resulting in a more uniform film with increased thickness (~500 nm). Extended ZnO deposition (≥ 40 s) further improved thickness (up to ~600 nm) and significantly reduced Sb₂O₃ surface features. Elemental analysis via energy-dispersive X-ray spectroscopy (EDS) quantified atomic percentages of Sb, S, O, and Zn [Supplementary Figure 2A]. Key trends include: (1) relatively stable Sb content, (2) a non-monotonic variation in S content, (3) an initial decrease followed by an increase in O content, and (4) a clear upward trend in Zn concentration (from ~0.1% to ~1.2 at.%). The elevated oxygen and decreased sulfur levels at 400 s ZnO are attributed to thick ZnO layers contributing additional oxygen. The increasing Zn/Sb atomic ratio [Supplementary Figure 2B] further confirms progressive Zn incorporation into the Sb₂S₃ matrix with extended ZnO exposure.

To evaluate the photovoltaic impact of ZnO incorporation, Sb₂S₃ solar cells were fabricated using a hydrothermally deposited PbS HTL, paste-coated carbon and Ag contacts. The statistical results of the average values of J-V measurements [Figure 2A and Table 1] show that the enhancement effect of the ZnO modified devices is consistent. The V_oc remained above 700 mV for all ZnO-treated samples, peaking at 40 s due to reduced Sb₂S₃ degradation and minimized leakage pathways. The optimal 40 s ZnO condition coincides with minimized resistive Sb₂O₃ formation and improved bulk conductivity, facilitating efficient carrier extraction. The J_sc increased by 17% at 40 s ZnO (14.6 mA/cm²) compared to the control, then declined with thicker ZnO layers. Similarly, the FF reached an optimal average of 59% at 40 s, but dropped to ~50% at ≥ 80 s, consistent with increased series resistance and impeded hole transport from thicker ZnO layers. Consequently, the highest PCE of 7.00% was attained under 40 s ZnO conditions (V_oc = 729 mV, J_sc = 15.6 mA/cm², FF = 61.5%), representing a 20% improvement over the unmodified reference device. Although this device performance still lags behind that of Spiro-OMeTAD/metal electrodes and Sb₂S₃-based tandem devices^[17,18], this study provides a practical strategy for the low-cost fabrication of solar cells employing all-inorganic carbon electrodes.

Figure 2. (A) J-V curves and (B) XRD patterns of Sb₂S₃ doped with ZnO with different deposition times; (C) The magnified XRD pattern of the Sb₂S₃ films; (D) Sb 3d, (E) S 2p and (F) Zn 2p XPS spectra of Sb₂S₃ thin films with and without ZnO. High-resolution (G) Sb 3d, (H) S 2p, (I) Sb MN XPS spectra. J-V: Density-voltage; XRD: X-ray diffraction; XPS: X-ray photoelectron spectroscopy; PDF: powder diffraction file; MN: X-ray photoelectron spectrum peak of Sb; W/O: without ZnO.

X-ray diffraction (XRD) analysis [Figure 2B] verified the crystallographic composition of annealed films, identifying orthorhombic Sb₂S₃ (PDF#42-1393) as the dominant phase, with minor peaks corresponding to cubic Sb₂O₃ (PDF#71-0365). Notably, the intensity of the Sb₂O₃ (222) peak at 27.68° decreased systematically with longer ZnO deposition [Supplementary Figure 3A and B], indicating suppressed secondary oxide formation. A weak ZnO (100) peak appeared only at 80 s ZnO, while no ZnO diffraction was detectable at shorter deposition times (≤ 40 s), underscoring the ultrathin and amorphous nature of ZnO in those cases. These results confirm that the ZnO interlayer effectively retards Sb₂O₃ phase formation during air annealing.

To understand the origin of these enhancements, grazing-incidence XRD (GIXRD) at 2^o incidence [Supplementary Figure 4] revealed no residual ZnO phase in the 40 s ZnO film post-annealing, suggesting Zn substitution within the Sb₂S₃ or Sb₂O₃ lattices. Detailed peak analysis [Figure 2C] showed systematic shifts of major Sb₂S₃ and Sb₂O₃ peaks toward higher angles, consistent with lattice contraction per Bragg’s law (d = λ/2sinθ). When larger Zn²⁺ (134 pm) replaces Sb³⁺ (140 pm), the smaller ionic radius causes lattice contraction, which is the direct origin of the aforementioned peak shift. Further observation also reveals that the diffraction peak shift in Sb₂O₃ is more pronounced and the lattice distortion is more significant compared with the counterpart of Sb₂S₃. This can be attributed to the higher surface content of Sb₂O₃ and a higher probability of reaction with Zn. Based on the above analysis, it can be preliminarily determined that after 40 s of ZnO deposition and subsequent annealing, Zn elements enter both the Sb₂S₃ and Sb₂O₃ lattices in the form of substitutional doping, providing atomic-scale insights into the performance enhancement mechanism.

X-ray photoelectron spectroscopy (XPS) was employed to elucidate the Zn chemical state. Survey scans [Supplementary Figure 5] confirmed elemental presence of C, Sb, S, O, and Zn. High-resolution Sb 3d spectra [Figure 2D] revealed that ZnO deposition reduced Sb₂O₃ content from 79% (control) to 65% (40 s ZnO), consistent with SEM/XRD trends. The S 2p spectrum [Figure 2E] showed typical spin-orbit splitting indicative of S^2- in Sb₂S₃, with a shift to higher binding energies upon ZnO deposition, suggesting reduced sulfur vacancies. Zn 2p peaks [Figure 2F] were absent in the control but faintly detected at 1,022.11 eV and 1,045.50 eV in the 40 s ZnO sample. This chemically shift relative to crystalline ZnO indicates incorporation rather than phase separation. Concurrent shifts in Sb MNN Auger peaks further suggest Zn-S bonding and altered Sb electronic environments.

To probe the depth distribution of Zn doping, XPS depth profiling was performed on the Zn-doping sample (40 s ZnO) versus the control (W/O ZnO) using Ar⁺ sputtering at 0.03 nm/s. The Sb 3d_5/2 binding energy [Figure 2G] decreased with sputtering time, stabilizing beyond 200 s (~6 nm depth). Quantifying the Sb 3d_5/2 shift relative to the unsputtered surface (Supplementary Figure 6, red spheres) showed an exponential decay (y = -0.91 + 0.73e^-x/60), with stabilization beyond 200 s (~6 nm depth). This confirms a thin Sb₂O₃ surface layer serving as a hole-blocking but tunnel-permissive oxide. This structure is beneficial for suppressing interfacial recombination while enabling charge extraction. Concurrently, smaller S 2p shifts [Figure 2H and Supplementary Figure 6] and analogous Sb MNN Auger peak displacements (Figure 2I, Supplementary Figure 6, fit: y = -1.52 + 1.72e^-x/42) indicate Zn_Sb substitutional doping confined primarily to the top ~10 nm. Zn_Sb substitution was confined to the top ~10 nm, supported by S 2p and Sb MNN trends [Figures 2H and I, Supplementary Figure 6]. This confirms that the ZnO cap limits Sb₂O₃ formation to an ultrathin ~6 nm layer, functioning as a wide-bandgap hole-blocking layer. Then it suppresses interfacial recombination while enabling efficient hole tunneling due to its minimal thickness, consistent with enhanced J_sc. Collectively, these depth-resolved analyses demonstrate the dual role of the ZnO interlayer: (1) restricting Sb₂O₃ to nanoscale surface regions, and (2) enabling shallow Zn doping in Sb₂S₃.

To correlate these interfacial modifications with carrier dynamics, light-intensity-dependent J_sc and V_oc measurements were conducted [Figure 3]. According to the power-law relationship [Equation 1]^[19], the fitting can obtain the α value.

Figure 3. (A) J_sc, (B) V_oc as a function of light intensity (C) 1/C²-V curves of Sb₂S₃ devices; (D) Dark J-V curves and (E) Nyquist plots of Sb₂S₃ devices; (F) Raman spectra of control and Zn doping Sb₂S₃ thin films; (G and H) AFM images and (I and J) C-AFM images of control and Zn doping Sb₂S₃ thin films. J-V: Density-voltage; V_oc: open-circuit voltage; C-AFM: conductive atomic force microscopy; R_tr: charge transport resistance; R_rec: recombination resistance; J_sc: short-circuit current density; R_s: series resistance; C_tr and C_rec: capacitor.

Where I represents the percentage of irradiation light intensity, and α is the power exponent. As shown in the J_sc and irradiation light intensity relationship graph in Figure 3A, yielded α = 0.90 (control) versus 0.91 (Zn-doping), indicating marginally improved carrier collection efficiency. More significantly, the function relationship between V_oc and light intensity satisfies^[20]:

where T, k_B (1.38 × 10^-23 J/K), and q (1.60 × 10^-19 C) are the Kelvin temperature, Boltzmann constant, and elementary charge, respectively, and n is the ideality factor related to recombination. The diode ideality factor n extracted from decreased from 2.43 (control) to 1.87 (Zn-doping), demonstrating substantial suppression of Shockley-Read-Hall (SRH) recombination via Zn-induced defect passivation. These carrier transport enhancements directly explain the observed performance gains.

The C-V characteristics of photovoltaic devices under dark conditions provide critical insights into defect-related properties. We thus measured dark C-V curves at 2 kHz with a 30 mV alternating current (AC) amplitude [Supplementary Figure 7]. By plotting 1/C² versus V [Figure 3C], the built-in potential (V_bi) is determined by linearly extrapolating the voltage axis intercept according to^[21,22]:

Here, C is the capacitance, V is the applied DC bias, S (0.09 cm²) is the cell area, ε₀ (8.854 × 10^-14 F/cm) is the vacuum permittivity, and ε_p (6.67) is the relative permittivity of Sb₂S₃. The control device exhibited a lower V_bi (1.13 V) than the Zn-doping counterpart (1.22 V), aligning with the trend in V_oc. The hole concentration (N_A) of Sb₂S₃, derived from the slope of the linear fit, increased from 1.32 × 10¹⁶ cm^-3 (control) to 1.47 × 10¹⁶ cm^-3 (Zn-doping). This enhancement is attributed to shallow acceptor states (Zn_Sb) induced by P-type doping of Zn, which improves surface conductivity and carrier extraction efficiency, thereby boosting J_sc^[15].

Dark J-V characteristics further elucidate diode behavior [Figure 3D]. The reverse saturation current (J₀) decreased significantly from 5.06 × 10^-5 mA/cm² (control) to 9.78 × 10^-6 mA/cm² (Zn-doping), indicating suppressed carrier recombination at the back interface. To gain further insight into the underlying carrier recombination and transport dynamics, we performed EIS measurements under dark conditions and acquired the corresponding Nyquist plots (Figure 3E, inset showing the equivalent circuit). Among them, the charge transport resistance (R_tr) characterizes the carrier collection capability, while the recombination resistance (R_rec) is inversely proportional to the carrier recombination rate of the device. Fitted parameters [Supplementary Table 1] revealed that reduced charge transport resistance (R_tr = 3.2 kΩ target vs. 5.2 kΩ control) confirms enhanced carrier collection, and increased recombination resistance (R_rec = 17 kΩ target vs. 10 kΩ control), which inversely correlates with recombination rates^[21]. In summary, dark J-V and EIS analyses demonstrate that Zn doping concurrently suppresses carrier recombination and enhances transport/extraction efficiency, directly corroborating the observed performance improvements.

The foregoing analysis demonstrates that Zn doping at the Sb₂S₃ back surface enhances carrier concentration, suppresses recombination, and improves carrier transport efficiency, particularly boosting J_sc. To elucidate the underlying physical origins, we systematically investigated the chemical bonding, electronic properties, and energy-level alignments of Zn-doping Sb₂S₃ films. Raman spectroscopy [Figure 3F] revealed no ZnO characteristic peaks (dashed lines in Figure 3F) in either control or Zn-doping films. All observed peaks correspond exclusively to Sb₂S₃ vibrational modes^[23]. The Raman spectrum of Sb₂S₃ arises from four vibrational modes (2A₁ + 2E) of its tetrahedral SbS₃ units (C_3v symmetry): symmetric stretch ν_s (A₁), asymmetric stretch ν_a (E), symmetric bend δs (A₁), and asymmetric bend δ_a (E), with typical intensity ordering ν_s > ν_a and δ_s > δ_a^[24]. As shown in Figure 3F, both films exhibit dominant peaks at 281 cm^-1 [ν_a (Sb-S)] and 304 cm^-1 [ν_s (Sb-S)], alongside weaker features at 127 cm^-1 (A_g), 156 cm^-1 (B_1g), 189 cm^-1 [δ_a > (S-Sb-S)], and 239 cm^-1 [δs (S-Sb-S)]. Critically, the intensity ratio I₃₀₄/I₂₈₁ of symmetric-to-asymmetric stretching vibrations inversely correlates with sulfur vacancy (V_S) concentration. Spectral deconvolution [Supplementary Figure 8A and B] yielded I₃₀₄/I₂₈₁ = 1.05 (control) versus 1.30 (Zn-doping), confirming reduced V_S defects in the Zn-doping film. This reduction is attributed to suppressed Sb₂S₃ volatilization and enhanced crystallinity during ZnO-capped annealing, consistent with diminished non-radiative recombination observed in carrier dynamics.

To further probe the local electrical properties, conductive atomic force microscopy (C-AFM) was performed under dark conditions on both control and Zn-doping Sb₂S₃ films [Figure 3G-J]. Significantly elevated current signals were detected within the large-grained structures of the Zn-doping film compared to the reference. Concurrently, regions passivated by Sb₂O₃ nanobelts at grain boundaries exhibited further suppressed current leakage. These observations collectively demonstrate that enhanced bulk conductivity in Zn-doping Sb₂S₃ directly contributes to higher J_sc. While effective defect passivation at grain boundaries via oxygen incorporation, mitigating carrier recombination. These nanoscale electrical characteristics align consistently with the macroscopic device performance enhancements.

To elucidate carrier transport mechanisms, ultraviolet photoelectron spectroscopy (UPS) was performed on control and Zn-doping Sb₂S₃ films. The Fermi level (E_F) and VBM positions were determined by subtracting the secondary electron cutoff edge from the He I excitation source (21.22 eV) [Figures 4A and B]. The control film exhibited E_F = -3.48 eV and VBM = -6.04 eV, while Zn doping shifted these to E_F = -3.70 eV and VBM = -5.99 eV. Given the mixed-phase surface composition (> 80% Sb₂O₃), the conduction band minimum (CBM) was referenced to Sb₂O₃’s bandgap. The proposed energy band diagram (Figure 4C, partial data adapted from Ref.^[25]) reveals that the control device possesses high CBM and deep VBM at the back surface form an electron-blocking barrier, which suppresses recombination and transport holes through the oxide tunnel. For Zn-doping device, a 0.22 eV downward shift in E_F and 0.05 eV upward shift in VBM/CBM collectively enhance back-surface field strength. Coupled with the ultrathin oxide layer (< 10 nm), this configuration promotes hole separation, directly explaining the enhanced V_oc and J_sc and PCE.

Figure 4. UPS spectrum of (A) control and (B) Zn doping Sb₂S₃ thin films; (C) schematic band diagram and (D) the summarized PCE of previous works of Sb₂S₃ solar cell. UPS: Ultraviolet photoelectron spectroscopy; FTO: fluorine-doped tin oxide; CdS: cadmium sulfide; PCE: power conversion efficiency.

The successful incorporation of Zn and the concomitant suppression of Sb₂O₃ underscore the dual functionality of our ultrathin ZnO layer strategy. This feature distinguishes it from conventional single-function strategies. For instance, prior approaches to alleviate back-contact issues often relied on introducing distinct metal oxide buffer layers (e.g., WO_3-x, MoO₂) primarily for hole transport or energy level alignment^[13,17], or focused solely on elemental doping (e.g., solution-processed Zn doping) to tailor electrical properties^[16]. While effective, these strategies typically address only one aspect of the complex interfacial challenges in Sb₂S₃-based devices. Our approach uniquely integrates a diffusion barrier and a dopant source into a single sub-nanometer-scale layer, simultaneously tackling phase purity, defect passivation, and energy level modulation. This synergistic effect is evidenced by the concurrent improvement in N_A, R_rec, and V_oc.

Through the strategic implementation of a ZnO protective layer, we optimized both the Sb₂O₃ thickness and achieved surface Zn-doping in Sb₂S₃, culminating in a 7.00%-efficient, fully inorganic carbon-electrode-based Sb₂S₃ solar cell with an FTO/CdS/Sb₂S₃/PbS/Carbon/Ag architecture. Statistically compiled literature data [Figure 4D] position this efficiency at the state-of-the-art performance tier for planar Sb₂S₃ thin-film devices {Surpassing the 7% benchmark indicated by the gray dashed line, and the references are presented in Supplementary Materials [1-90]}. Notably, previously reported high-efficiency Sb₂S₃ devices predominantly employ unstable organic hole-transport layers (e.g., Spiro-OMeTAD or P3HT) and cost-prohibitive Au electrodes, fundamentally limiting their practical viability. In contrast, our approach leverages ambient air-annealing and low-cost carbon electrodes to construct an all-inorganic device, which exhibits the competitive efficiency approaching the current Sb₂S₃ record. Based on the enhanced operational stability from inorganic constituents and scalable manufacturability via solution-processable components^[26], this work establishes a transformative pathway toward economically viable, high-performance thin-film solar cells.

CONCLUSION

In this work, we aimed to address the critical challenges of sulfur volatilization and detrimental Sb₂O₃ formation at the back interface of Sb₂S₃ solar cells during high-temperature annealing. By introducing an ultrathin ZnO protective layer, we developed a dual-functional strategy that simultaneously achieves controlled oxygen passivation and effective Zn doping. This synergistic modification suppresses interfacial recombination, optimizes energy level alignment, and enhances carrier transport. Ultimately, in comparison with unmodified devices, the ZnO-modified solar cells exhibit a 17% enhancement in J_sc and a significant improvement in PCE, culminating in a state-of-the-art PCE of 7.00% for fully inorganic, carbon-based architectures. This scalable and air-processable strategy provides a promising route toward high-performance, stable, and low-cost Sb₂S₃ solar cells.