Iron-derived solid redox system for effective and sustainable sulfur recovery from sulfide-bearing wastewater

Abstract

Elemental sulfur recovery from sulfide-bearing wastewaters presents a promising strategy for achieving both a sustainable elemental sulfur supply and pollutant removal. However, existing methods face low efficiency or heavy chemical usage, limiting their practicality. Here, we report an iron-derived solid redox sulfur recovery system, capable of efficiently converting sulfides to elemental sulfur and separating elemental sulfur from the iron catalysts (ferrihydrite) via simple gravity-based methods. Using iron-derived solid redox system, we achieve a record sulfide removal rate of 102.9 ± 3.1 kg m⁻³ d⁻¹ with an elemental sulfur selectivity of 93.5 ± 0.9%. The produced elemental sulfur is then efficiently separated from the solid mixture through a hydraulic sorting process, attaining 97.7 ± 0.2% purity. Mechanistic analysis identifies the iron and sulfur species governing the solid redox process and elucidates that the separation behavior of elemental sulfur is likely a result of surface-state-dependent elemental sulfur agglomeration and repulsion between elemental sulfur and iron flocs. Moreover, we verify the effectiveness of iron-derived solid redox system in recovering elemental sulfur from two real wastewaters and estimate the economic benefits of elemental sulfur recovery to exceed 180 United States dollars per ton. This work suggests a practical approach to enable sulfide-bearing wastewaters as a sustainable source for elemental sulfur supply efficiently and cost-effectively.

A solid Fe(II/III) redox system is developed to remove sulfide in sulfide-bearing wastewater and recover it as S⁰. The recovered S⁰ is purified by physical method based on the particle size difference induced by hydrophobic flocculation.

Introduction

Water contamination and resource scarcity represent formidable impediments to global sustainable development. Sulfide is a ubiquitous pollutant commonly found in diverse wastewater streams, including those from oil and gas extraction, petrochemical plants, viscose rayon manufacturing, tanneries, coal gasification processes, pharmaceutical production, textile mills, landfill leachate, pulp and paper industries, and mine drainage^1–7. Its propensity to generate malodorous nuisances, release toxic gases, exert biological toxicity, induce structural corrosion, and contribute to acid rain formation renders it a significant environmental concern in global wastewater treatment practices^1,8,9. Meanwhile, elemental sulfur (S⁰) is recognized as a widely used and essential raw material in various industries, including agricultural fertilizers, construction materials, rubber, and batteries^10–14. Currently, over 85.8% of S⁰ is sourced as a by-product from the petroleum and natural gas industries, raising concerns about a potential sulfur crisis, as increasing global decarbonization efforts have limited production in these traditional energy sectors¹⁵. Research indicates that by 2040, the unsustainable supply of sulfur could result in an 80% shortfall in sulfuric acid¹⁶—an indispensable raw material for both industry and agriculture¹⁰. Therefore, recovering sulfur from sulfur-bearing wastewater (SBW) has emerged as a crucial strategy for both pollutant removal and bolstering the sustainability of sulfur supply. However, it remains a significant challenge.

For S⁰ recovery from SBW, biological methods have been extensively studied, in which sulfur-oxidizing bacteria¹⁷ (SOB) oxidize sulfide to S⁰ with either oxygen or nitrate as the electron donor^18,19. Although biological sulfur recovery technology has advanced significantly in recent decades, several inherent challenges limit its application². First, there is a trade-off between reaction selectivity and rate; i.e., SOB typically have the capability to fully oxidize sulfide to sulfate^20,21, necessitating oxygen-limiting conditions to achieve high selectivity²², which likely compromises the oxidation rate as the dissolved oxygen concentration in this condition was suggested to be lower than 100 nM (~0.03 mg L⁻¹)²³, considerably lower than the reported oxygen affinity constant (0.1 mg L⁻¹) of S-oxidizing biomass²⁴. Second, the produced S⁰ remains challenging in settling and recovering²⁵, as it is colloidal and hydrophilic²⁶, with reported settling times of up to several days²⁷; Finally, SOB are sensitive to various toxic compounds (e.g., BTEX²⁸, endocrine disruptors²⁹, Cr (VI)³⁰, etc.). In addition to biological methods, S⁰ recovery from sulfide can also be achieved through chemical catalytic methods utilizing metal-derived redox systems (e.g., EDTA-Fe(III)/Fe(II) and VO₃⁻/V₄O₉²⁻)^31–33. Unlike their biological counterparts, chemical catalytic S⁰ recovery methods are highly selective and do not require limiting oxygen supply, allowing for high reaction rates simultaneously³⁴. Furthermore, the produced S⁰ is more hydrophobic compared to biogenic S⁰³⁵, facilitating the formation of larger S⁰ particle clusters that enhance sedimentary separation³⁶. However, the application of these chemical methods necessitates costly procedures³⁷ (e.g., acidification and stripping) to separate sulfide³⁸ (in the form of H₂S) from wastewater beforehand, as directly dosing these aqueous redox chemicals into wastewater can lead to significant secondary pollution³⁹. Therefore, there is an urgent need to explore new strategies for more efficient and cost-effective S⁰ recovery from SBW.

$$2\mathrm{FeO}\left(\mathrm{OH}\right)\cdot {\mathrm{nH}}_2\mathrm{O}+{3\mathrm{HS}}^{-}\to \left(3-\mathrm{x}\right){\mathrm{S}}^0+{\mathrm{FeS}}_{\mathrm{x}}+\mathrm{Fe}{\left(\mathrm{I}\mathrm{I}\right)}_{\mathrm{solid}}+{3\mathrm{OH}}^{-}+\left(2\mathrm{n}+1\right){\mathrm{H}}_2\mathrm{O}$$ 1

$${\mathrm{FeS}}_{\mathrm{x}}+\mathrm{Fe}{\left(\mathrm{I}\mathrm{I}\right)}_{\mathrm{solid}}+\left(2\mathrm{n}+1\right){\mathrm{H}}_2\mathrm{O}+3/{2\mathrm{O}}_2\to 2\mathrm{F}\mathrm{e}\mathrm{O}\left(\mathrm{OH}\right)\cdot {\mathrm{nH}}_2\mathrm{O}+{\mathrm{xS}}^0$$ 2

Herein, we present an iron-derived solid redox (IDSR) system for S⁰ recovery from SBW, which operates through a redox cycle involving ferric (hydr)oxides and ferrous sulfides, as shown in Eqs. (1) and (2). We have demonstrated that the proposed IDSR system maintains the advantages of a chemical catalytic method in terms of high reaction rate and selectivity when converting sulfide to S⁰. Moreover, this system can treat SBW directly without requiring prior sulfide stripping, as it avoids the use of aqueous chemicals. Importantly, we have observed a self-separation phenomenon between ferric (hydr)oxides and the produced S⁰, achieving S⁰ recovery with a purity of up to 97.7 ± 0.2% through a simple gravity-derived separation method. In addition to performance metrics, the mechanism of the IDSR process has been systematically studied, identifying the actual players involved in the solid redox system and elucidating the self-separation behavior between S⁰ and iron species. Finally, we have demonstrated the effectiveness of the IDSR system for S⁰ recovery from two different real SBWs and shown the significant economic benefits of this system. This study provides a strategy for S⁰ recovery from SBW, enabling sustainable S⁰ supply from this widespread waste source more efficiently and cost-effectively.

Result

Design of the IDSR system

The IDSR system is designed as a three-step sequencing batch procedure for wastewater desulfurization with S⁰ production, followed by a solid separation procedure for S⁰ recovery (Fig. 1). Regarding the wastewater desulfurization, the three-step sequencing batch procedures include: (i) Step-Ⅰ: SBWs injection and anaerobic mixing, where the sulfide reacts with the ferric (hydr)oxides (in this work, six-line ferrihydrite) following Eq. (1), with part of the sulfide converting to S⁰; (ii) Step-Ⅱ: aeration, where the iron-associated sulfide is further oxidized to S⁰ with the regeneration of ferric (hydr)oxides, following Eq. (2); (iii) Step-Ⅲ: sedimentation and water discharge, where liquid-solid separation occurs through sedimentation, enabling the discharge of desulfurized water while minimizing the loss of ferric (hydr)oxides and the produced S⁰. These three steps are repeated in multiple cycles to sustain the SBWs treatment and continuously accumulate S⁰ in the system. Once S⁰ is accumulated to a certain extent, the solid mixture is removed from the desulfurization reactor, and S⁰ separation is performed through either gravity settling separation or hydraulic sorting. In this procedure, the separated S⁰ is recovered as a sustainable product, while the ferric (hydr)oxides are reused in the desulfurization process.

Fig. 1. Schematic diagram of iron-derived solid redox S⁰ recovery system.

a Iron-derived solid redox S⁰ recovery process: conversion of sulfide to elemental sulfur and its accumulation in the reactor through injection, mixing, aeration, and settling from sulfide-bearing wastewater. b S⁰ purification process: after sufficient accumulation of S⁰, the S-Fe mixture is transferred to a gravity settling or hydraulic sorting reactor for the purification of S⁰.

Performance of the IDSR system

The performance of IDSR system was initially tested in a single cycle of the 3-step batch mode with a water discharge ratio of 75%. Sulfide-bearing wastewaters (SBWs) with different total sulfide (S(-Ⅱ)-S) concentrations (ranging from 0.1 g L⁻¹ to 5.0 g L⁻¹) were employed for testing. In Step-Ⅰ (SBWs injection and mixing anaerobically), around half of the S(-Ⅱ) was converted to S⁰ after reacting with the pre-synthesized Six-line ferrihydrite (6LFh). This process was found progressing very fast, which can be completed in just 2 min (Fig. 2a, b). Accompanied by the conversion of sulfur species, Fe(III) derived from 6LFh was confirmed to be reduced to Fe(II) stoichiometrically (Supplementary Fig. 1). Thereafter, the operation was turned to Step-Ⅱ (aeration). As shown in Fig. 2a, b, the remaining S(-Ⅱ) from Step-Ⅰ was depleted within 40 min, accompanied by the increase of S⁰. Meanwhile, the regeneration of Fe(III) occurred but required longer time (maximum to about one-hour, Supplementary Fig. 1). By collecting the results obtained from these two steps, IDSR was demonstrated to be ultra-highly efficient for converting S(-Ⅱ) to S⁰, as evidenced by 93.5 ± 0.9% of the selectivity and especially a reaction rates up to 102.9 ± 3.1 kg m⁻³ d⁻¹ (Fig. 2c) which is at least one order of magnitude higher than those reported in biological methods⁴⁰ (Supplementary Table 1). Finally, in Step-Ⅲ (sedimentation and water discharge), once the aeration was switched off, solids sedimentation occurred immediately with the floc-water interface dropping down visibly. For all sets, the floc-water interface could drop to the stable reactor’s height (a location for discharging treated wastewater) in 40 min (Supplementary Fig. 2). This sedimentation time is considerably shorter than that reported in biological methods²⁷ (several hours to days), further strengthening the superiority of IDSR system.

Fig. 2. Performance of elemental sulfur production by the iron-derived solid redox S⁰ recovery method.

a Sulfide concentration in batch tests for sulfide-bearing wastewater (SBW) treatment using the iron-derived solid redox (IDSR) system at different sulfide concentrations. b S⁰ concentration in batch tests for SBW treatment using the IDSR system at different sulfide concentrations (Step Ⅰ means that the mixing of iron oxyhydroxide with sulfide is completed, while Step Ⅱ means after aeration). c Sulfide removal efficiency, sulfide removal rate and S⁰ selectivity in batch tests for SBW treatment using the IDSR system at different sulfide concentrations. d Sulfide and S⁰ concentration in the system over ten cycles of SBW treatment at 0.4 g L⁻¹ S(-Ⅱ)-S concentration using the IDSR method. e Sulfide removal efficiency, sulfide removal rate and S⁰ selectivity over ten cycles of SBW treatment at 0.4 g L⁻¹ S(-Ⅱ)-S concentration using the IDSR method. f Total sulfur balance over ten cycles of SBW treatment at 0.4 g L⁻¹ S(-Ⅱ)-S concentration using the IDSR method. Each value in (a–f) is the mean of three replicates, and error bars indicate standard deviation (samples n  =  3). Source data are provided as a Source Data file.

To evaluate performance repeatability, the IDSR systems were investigated under a multi-cycle operational mode with the influent S(-Ⅱ)-S concentration at 0.4 g L⁻¹, which falls into a typical range of S(-Ⅱ)-S concentration in real SBWs (0.06–2.0 g L⁻¹, Supplementary Table 2). As illustrated in Fig. 2d, consistent profiles of S(-Ⅱ) depletion were observed across the ten tested cycles. Concurrently, S⁰ was continuously accumulated in the reactors, with its concentration showing an almost linear increase. These results clearly indicate the good repeatability of IDSR. Further, according to Fig. 2e, there are no statistically significant differences in sulfide removal efficiency (all cycles yielded identical results as the sulfide concentrations in the effluent were below the detection limit), removal rate (p > 0.05), and S⁰ formation selectivity (p > 0.05) over ten cycles. Other detailed reaction process indicators for different cycle counts also show a similar reaction trend. (Supplementary Figs. 3–5). Moreover, element balance analysis was performed (Fig. 2f). The results showed 93.7 ± 0.2% of S(-Ⅱ) was selectively oxidized to S⁰, with the vast majority retained in the reactor (accounting for 93.4 ± 0.2% of S(-Ⅱ) removal). In addition to S⁰, more oxidized products, thiosulfate and sulfate, were detected, but just accounting for 4.2 ± 0.2% and 2.1 ± 0.1% of the S(-Ⅱ) removal, respectively. Regarding the 6LFh, the loss was detected due to the discharge of treated water (3.5 ± 0.3 mg L⁻¹, Supplementary Figs. 5, 6). Nevertheless, the total loss of iron within ten cycles only accounts for 6.7 ± 0.2% of that initially dosed, indicating the high reusability of the 6LFh. To further understand the long-term feasibility of the IDSR system, additional tests were conducted by extending the operation cycles to 60 with a constant 6LFh concentration. In this case, no obvious performance deterioration was found in the IDSR reactor despite the mass accumulation of S⁰ being twenty times greater than the iron catalyst (Supplementary Figs. 7, 8). This result suggests the S⁰ accumulation does not affect the iron-derived catalytic reactions and highlights the good stability of the proposed IDSR system.

To investigate the hypothesized self-separation behavior between ferric (hydr)oxides and the produced S⁰, we conducted additional multi-cycle operation experiments focused primarily on solids sedimentation in the IDSR systems. After the fifth cycle, we observed a distinct layered sedimentation phenomenon, which became particularly clear by the end of the tenth cycle (Fig. 3a). Scanning electron microscopy with energy dispersive spectrometer (SEM-EDS) analysis revealed the bottom sedimentary layer was consisted of high-purity S⁰ agglomerates with the size reaching up to ~100 µm (Fig. 3d, e), while the upper sedimentary layer contained ferric (hydr)oxides and small amounts of S⁰ with the agglomerate sizes smaller than ~20 µm (Fig. 3b, c). Quantitative analysis of the S⁰ content indicated that 96.5 ± 1.0% of S⁰ was recovered into the bottom sedimentary layer, with purity increasing from 77.7 ± 1.9% (before separation) to 90.0 ± 1.3% (Fig. 3g). In addition to gravity settling separation, we evaluated another gravity separation method using hydraulic sorting. As shown in Fig. 3f, the separability of S⁰ and ferric (hydr)oxides was negligible in the first cycle but improved with successive operational cycles. By the end of the tenth cycle, hydraulic sorting resulted in a maximum S⁰ purity of 97.7 ± 0.2% at a flow velocity of 1.1 m min⁻¹, with a measured S⁰ recovery ratio of 93.9 ± 0.2%. Furthermore, we confirmed that the S⁰ recovered from hydraulic sorting could be further refined, achieving a purity upgrade to 99.8 ± 0.1% by washing with hydrochloric acid to remove iron species. Regarding the ferric (hydr)oxides, we calculated the iron loss by considering contributions both from water discharge and S⁰ recovery (Fig. 3h). As expected, scenarios with higher S⁰ purity minimized the residual ferric (hydr)oxides in the recovered S⁰, resulting in a lower corresponding iron loss ratio. In the scenario involving hydraulic sorting followed by hydrochloric acid washing, since the acid-washed iron was transformed into FeCl₃, a precursor for preparing ferric (hydr)oxides, this process was also considered a means to reduce iron loss. Consequently, this scenario exhibited the lowest 6LFh-Fe loss per gram of S⁰ recovered (0.02 ± 0.001 g), with nearly all contributions arising from water discharge.

Fig. 3. Self-separation phenomenon of S⁰ and Ferric (hydr)oxides and gravity settling separation and hydraulic sorting methods.

a Pictures of the stratification of S⁰ and Ferric (hydr)oxides after ten cycles of sulfide-bearing wastewater (SBW) treatment at 0.4 g L⁻¹ S(-Ⅱ)-S concentration. b SEM image of the top sedimentary layers. c Energy-dispersive X-ray spectroscopy (EDS) analysis of the top sedimentary layers. The inset shows the atomic mass proportions of Fe, S, and O. d SEM image of the bottom sedimentary layers. e EDS analysis of the bottom sedimentary layers. The inset shows the atomic mass proportions of Fe, S, and O). f Separation efficiency of hydraulic sorting at different flow rates and cycles for S⁰ and Ferric (hydr)oxides produced from SBW treatment at 0.4 g L⁻¹ S(-Ⅱ)-S concentration. g S⁰ purity and recovery ratio using gravity settling separation, hydraulic sorting and acid washing after hydraulic sorting. h, Iron loss per gram of recovered S⁰. Each value in (f–h) is the mean of three replicates, and error bars indicate standard deviation (samples n  =  3). Source data are provided as a Source Data file.

To further expand the understanding of IDSR system performance under broader conditions, additional tests were conducted with a low influent S(-Ⅱ)-S concentration of 0.05 g L⁻¹. It was discovered that the mass ratio of ferric (hydr)oxides concentration in the reactor to the influent S(-Ⅱ) concentration (referred to as the Fe/S ratio) became critical under this scenario. The Fe/S ratio of 1.2, previously set for high influent S(-Ⅱ) concentrations, was inadequate for low influent S(-Ⅱ) concentrations, necessitating increased ratios (≥4.0) to ensure effective sedimentation of both iron floc and the produced S⁰ (Supplementary Note 1 and Supplementary Figs. 9, 10). Furthermore, the low influent S(-Ⅱ) concentration resulted in less S⁰ accumulation per operational cycle, requiring more cycles (≥10) to observe the stratification between ferric (hydr)oxides and S⁰. Based on these observations, a twenty-cycle test of the IDSR system with an Fe/S ratio of 6.0 was conducted at an influent S(-Ⅱ)-S concentration of 0.05 g L⁻¹ to verify its feasibility in treating low S(-Ⅱ) SBW. The results demonstrated the remaining of IDSR’s high efficiencies in sulfide removal (100%), S⁰ production and retention (90.0 ± 3.2%), and S⁰ separation (with S⁰ purity reaching 94.2 ± 0.4% and 97.0 ± 0.2% after gravity settling separation and hydraulic sorting, respectively) (Supplementary Figs. 11, 12).

Mechanism of S⁰ formation aided by IDSR

To gain insight into the mechanism of S⁰ formation by IDSR, we investigated the species involved in the redox reactions of solid-phase Fe(III)/(II) using a series of spectral methods. In the reaction between sulfide and pre-synthesized 6-line ferrihydrite (6LFh) (Step-Ⅰ of the IDSR first cycle), Fe(II)-polysulfides (FeS_x) were produced, as evidenced by the sextet Mössbauer spectra (Fig. 4a, Supplementary Table 3). This finding is further supported by cryogenic XPS spectra, which revealed Fe(II)-S bonds and S-S bonds in S_x^2- within the Fe 2p spectra (Fig. 4b, Supplementary Table 4) and S 2p spectra (Fig. 4c, Supplementary Table 5), respectively. Additionally, the XPS Fe 2p spectra indicate the presence of a Fe(II)-O bond, accounting for 47.5% of the total Fe(II) (Fig. 4b, Supplementary Table 4). This observation aligns with the behavior of sulfur and iron species in this step, where the produced Fe(II) was about twice the amount of remaining sulfide in molar terms (9.6 mM vs 4.6 mM, calculated from Fig. 2a and Supplementary Figs. 1, 13). Subsequently, this solid Fe(II) mixture (FeS_x and Fe(II)-O-like phase) was oxidized by aeration (Step-Ⅱ of the IDSR first cycle). Based on Mössbauer spectra (Fig. 4d), the oxidized product was identified as a mixture of ferric (hydr)oxides, comprising 89.7% 6LFh and 10.3% goethite (Supplementary Table 3). Additional XRD characterization further corroborated these results (Fig. 4e). Beyond the first cycle of IDSR, characterizations of these iron species were also conducted in the second and tenth cycles, revealing consistent profiles in the Mössbauer and XRD spectra (Fig. 4a, d, e). These results suggest that IDSR operates as a reversible redox system composed of a Fe(II) mixture (FeS_x and Fe(II)-O-like phase) and a ferric (hydr)oxide mixture (6LFh and goethite) representing the reduced and oxidized states, respectively. The reversible characteristics of these iron species likely explain the excellent repeatability of the IDSR process across multiple cycles (Fig. 2e).

Fig. 4. Mechanisms of iron-derived redox conversion and S0 recovery.

a Mössbauer spectra of solids before aeration after mixed SBW during the first, second and tenth cycles. b XPS Fe 2p spectrum of solids before aeration after mixed sulfide-bearing wastewater (SBW). The inset shows the proportions of components. Curve fitting based on parameters from Supplementary Table 4. c XPS S 2p spectrum of solids after SBW mixing and subsequent oxidation processes. The inset shows the proportions of components. Curve fitting based on parameters from Supplementary Table 5. d Mössbauer spectra of solids after aeration during the first, second and tenth cycles. e XRD pattern of Fe solids after aeration during the second and tenth cycles. f TEM image of suspended solid during the mixing stage of the first cycle. The inset shows HR-TEM image. g TEM image of suspended solid during the aeration stage of the first cycle. The inset shows HR-TEM image. h TEM image of suspended solid during the mixing stage of the second cycle. The inset shows HR-TEM image. i TEM image of suspended solid during the aeration stage of the second cycle. The inset shows HR-TEM image. j XRD pattern of solids after mixing and aeration. k Mechanistic diagram of S⁰ production in iron-derived redox conversion. The unedited TEM images and Source data are provided in a Source Data file.

In addition to iron species, the progress of the solid sulfur species during S⁰ formation was also probed. In Step-Ⅰ of IDSR process, according to the XPS spectra (Fig. 4c, Supplementary Table 5), the produced S⁰ was found to be mostly in the form of polysulfides-sulfane (S⁰-S_x²⁻ and S⁰-S₂²⁻) rather than the crystalline cyclooctasulfur (S⁰-S₈). This result aligns with the XRD spectra, which exhibit an amorphous state of the solids (Fig. 4j). The TEM results show that no crystalline S⁰ was formed in this step and that no distinct crystal lattice was observed on the surface (Fig. 4f), indicating that ferric (hydr)oxides can only convert sulfides into zero-valent sulfur existing as polysulfide chains, rather than crystalline S⁰. After the IDSR process was operated as aeration (Step-Ⅱ of IDSR), the XPS spectra showed a decreasing trend in S²⁻, S⁰-S_x²⁻, S⁰-S₂²⁻, alongside a substantial increase in S⁰-S₈ (Fig. 4c, Supplementary Table 5). This indicates that the aeration process not only converts residual S(II) into S⁰ but also promotes the formation of crystalline S⁰ from polysulfide chains. The formation of the S⁰-S₈ crystal structures was further confirmed by XRD characterization (Fig. 4j), and the lattice of the (0 4 0) crystal plane was observed in TEM images (Fig. 4g). Crystallized sulfur is spherical morphology; however, it is worth noting that the surface of S⁰ is not fully crystallized, but rather possesses an ~6.5 nm-thick amorphous layer (Fig. 4g). In the step-Ⅰ of the second cycle, the newly generated zero-valent sulfur was deposited onto the surface of the incompletely crystallized spherical S⁰ produced in the first cycle, leading to the formation of spindle-shaped S⁰, which is considered the complete morphology of orthorhombic sulfur (Fig. 4h). The surface of the spindle-shaped sulfur still contained an amorphous layer. After re-aeration treatment, we observed irregularly shaped S⁰ with a completely crystallized surface; the thickness of its surface amorphous layer decreased to 1.3 nm, and it was completely unattached to ferric (hydr)oxides (Fig. 4i). Meanwhile, after aeration, spherical sulfur (completely identical to that after aeration in the first cycle) was also observed. This indicates that the second cycle, in addition to promoting the shaping and complete crystallization of elemental sulfur, also repeated the two-stage process of S⁰ generation process of the first cycle. Subsequent third cycles maintained a similar trend (Supplementary Fig. 14). Accordingly, the complete process of S⁰ production consists of four stages across two cycles (Fig. 4k): (i) reaction between ferric (hydr)oxides and sulfides to form amorphous polysulfide chains; (ii) aeration to produce spherical S⁰; (iii) stacking of zero-valent sulfur to form spindle-shaped sulfur; and (iv) re-aeration to yield fully crystallized, irregularly shaped S⁰. Throughout multiple cycles, stages (i) and (iii) occur simultaneously as a repetitive process in Step-Ⅰ, while (ii) and (iv) repeat synchronously within Step-Ⅱ.

Solids separation mechanism

One of the key findings of this study is that solid separation occurs spontaneously and efficiently in both water-solid separation and the separation of solid iron and sulfur species (Fig. 3 and Supplementary Fig. 15). To understand this unique property, we investigated the behavior of solid particle agglomeration at the end of aeration over multiple cycles of the IDSR process. Microscopic photographs clearly show several small S⁰ particles (black, mean size: 9.7 ± 8.2 µm) agglomerating with iron flocs (brown, mean size: 39.7 ± 11.3 µm) during the first cycle (Fig. 5a and Supplementary Fig. 16). Subsequently, some S⁰ particles agglomerated independently, forming larger S⁰ flocs (mean size: 58.8 ± 28.7 µm), as evidenced by the microphotograph from the fifth cycle (Fig. 5b). Finally, microphotographs taken from the tenth cycle suggest that these S⁰ flocs can grow up to 125.3 ± 64.7 µm and are located away from the iron flocs (mean size: 32.6 ± 15.7 µm) (Fig. 5c and Supplementary Fig. 16). SEM-EDS analysis further confirmed that the small S⁰ particles attached to the iron flocs in the first cycle, while the large S⁰ flocs formed in the tenth cycle existed independently with negligible mixing with iron (Supplementary Fig. 17). These findings indicate that the S⁰ particles/flocs exhibit a “small-attached-large-independent (SALI)” behavior in relation to the iron flocs. This SALI behavior of S⁰ particles/flocs with iron flocs provides a robust explanation for the effective solid separation observed in the IDSR process. For the smaller S⁰ particles that would not readily settle, their attachment to larger iron flocs facilitated their separation from water. The large S⁰ flocs, on the other hand, due to their size exceeding that of iron flocs, made gravity separation possible between the two solid phases, either through sedimentation or water flush.

Fig. 5. Mechanism of separation for S⁰ and Ferric (hydr)oxides.

a Optical microscopy image and particle size distribution (inset) of S⁰ in the first cycle. b Optical microscopy image and particle size distribution (inset) of S⁰ in the fifth cycle. c Optical microscopy image and particle size distribution (inset) of S⁰ in the tenth cycle. d XPS Fe 2p spectrum of solids in the first cycle. The inset shows the proportions of components. Curve fitting based on parameters from Supplementary Table 4. e TEM image of S⁰ embedded in ferric (hydr)oxides from the upper layer after stratification of ferric (hydr)oxides and S⁰ following ten treatment cycles. f HR-TEM and magnified HR-TEM (inset) of the upper layer after stratification of ferric (hydr)oxides and S⁰ following ten treatment cycles. g The violin plot shows a comparison of contact angle of S⁰ in first, fifth, and tenth cycle and stratified top and bottom layers in the tenth (n  =  6). The data were obtained from the measurements of two samples taken from each of three replicate experiments. Inside each violin, the white dot represents the median, the thick black bar represents the interquartile range (IQR, 25th–75th percentiles), and the whiskers extend to the adjacent values within 1.5 × IQR from the box edges. h TEM image of S⁰ from the bottom layer after stratification of ferric (hydr)oxides and S⁰ following ten treatment cycles. i HR-TEM and magnified HR-TEM (inset) of the bottom layer after stratification of ferric (hydr)oxides and S⁰ following ten treatment cycles. j XPS S 2p spectrum of solid in in first, fifth, and tenth cycle. The inset shows the proportions of components. Curve fitting based on parameters from Supplementary Table 6. k Schematic diagram of the occurrence states and mechanisms of S⁰ and ferric (hydr)oxides in the wastewater treatment system. CA, ZP, and 6LFh represent contact angle, zeta potential, and six-line ferrihydrite, respectively. The unedited TEM images and Source data are provided in a Source Data file.

To gain insight into the SALI behavior, surface state analysis was performed on different solid samples. For exploring the association of small S⁰ particles with iron flocs, we first excluded charge neutrality between iron and S⁰ species as a possible reason, since the zeta potentials of six-line ferrihydrite (6LFh) and the separated S⁰ particles (through acid washing of the S⁰-iron flocs mixture) were both measured as negative (−13.2 ± 2.1 mV for 6LFh and −25.8 ± 2.5 mV for S⁰, respectively; Supplementary Fig. 18). Using cryogenic XPS to examine the surface species of the S⁰-iron flocs mixture in the first cycle, we observed a mixed surface state for S⁰ particles, comprising S₂(-II), S_x(-II), and S₈, according to the S 2p spectra (Fig. 5j, Supplementary Table 6). Concurrently, the Fe 2p spectra exhibited the presence of surface Fe(III)-S species (Fig. 5d, Supplementary Table 4). These findings suggest that the unoxidized surface S(-II) state of S⁰ particles contributes to their combination with iron flocs. Subsequently, XPS characterizations were conducted to examine the samples collected across multiple cycles of the IDSR process to understand the formation of independently existing S⁰ flocs (Fig. 5j, Supplementary Table 6). The S 2p spectra displayed that the surface state of S⁰ in the S⁰-iron flocs mixture was gradually unified to be S₈ along with the increase in the operation cycle number. The loss of surface S(-II) state indicates a weak binding force between the S⁰ and iron flocs at this stage (Supplementary Fig. 19, Table 4). Additionally, the possible repulsive effect between these two solids (i.e., due to the negative zeta potentials as mentioned above) may further facilitate the S⁰ species away from the iron flocs, forming an independent solid phase. Nevertheless, the classic DLVO theory may not support the S⁰ species in forming flocs, because of the measured zeta-potential (−28.9 ± 2.0 mV) closing to the threshold value (±30.0 mV) (Supplementary Fig. 18)⁴¹. To address this concern, we further examined the S⁰ samples in terms of surface hydrophobicity, another factor known to lead to particle agglomeration. According to the measured contact angles of the acid-washed S⁰ particles/flocs taken from the first (68.8 ± 3.6°), fifth (79.7 ± 0.9°) and tenth cycle (102.0 ± 2.1°) (Fig. 5g), the surface hydrophobicity was suggested to be continuously enhanced. This can be elucidated by the fact that the change of surface state of S⁰ species across the multicycle operation ended up as a nonpolar S₈ form (Fig. 5j, Supplementary Fig. 20). Moreover, we observed the S⁰ samples with the lowest hydrophobicity exhibited as colloids after it dispersed into the water, while that with the highest hydrophobicity formed distinct flocs (Supplementary Fig. 21). This result clearly indicates the hydrophobic interaction governs the formation of S⁰ flocs.

Polysulfide chains exist as a surface amorphous layer in the generated S⁰. The thickness of this layer determines whether the surface remains attached to ferric (hydr)oxides via Fe(III)-S bonds or completely separates and self-aggregates as a hydrophobic surface in the nonpolar S₈ form. The newly generated spherical S⁰ during the aeration process was fully attached to the ferric (hydr)oxides, with its surface amorphous layer measured 6.5 nm in thickness (Fig. 4g). The sulfur adhering to ferric (hydr)oxides at the top of the tenth cycle layer similarly showed an amorphous layer thickness of 6.5 nm (Fig. 5e, f). In contrast, the S⁰ forming hydrophobic flocs in the bottom layer exhibited an amorphous surface layer thickness of only 1.3 nm (Fig. 5h, i). The reduction in the thickness of the amorphous layer implied a decrease in the polysulfide chains on the sulfur surface, which led to the cleavage of Fe(III)-S bonds and subsequent detachment from the ferric (hydr)oxides. Simultaneously, the increased content of nonpolar S₈ rendered the surface more hydrophobic. In fact, S⁰ with a thinner amorphous layer was detected as early as the second cycle (Fig. 4i); however, its concentration became the limiting factor restricting the formation of large-scale flocs (Supplementary Fig. 22). As the cycles progressed, the declining polysulfide ratio and rising sulfur concentration indicated the gradual accumulation of crystalline S⁰ (Figs. 2d and 5j). This resulted in the formation of large hydrophobic flocs (exceeding ferric (hydr)oxides in size), enabling the layer-by-layer deposition of sulfur and iron (Fig. 3a). Potential factors facilitating the reduction of surface polysulfide chains and the decrease in amorphous layer thickness include H⁺ and aeration. Further comparative analysis revealed that the reduction in the thickness of the amorphous layer was driven by the aeration process (Supplementary Fig. 23). Collecting these findings, we proposed a surface-state-dependent mechanism to elucidate the SALI agglomeration behavior as illustrated in Fig. 5k.

Application and economic evaluation of real wastewater

As a further step towards real-world applications, we tested the IDSR system using two types of industrial wastewater as examples: spent caustic and electronics manufacturing wastewater, which were the most common in industrial wastewater. These wastewaters typically contain complex organic matter, ions and suspended solids (Supplementary Tables 7, 8). All performance indicators remained stable during the treatment of both types of SBW (Fig. 6a–c), confirming the strong anti-interference capability of the IDSR system. Photographs of the produced S⁰ are shown in Fig. 6e. Furthermore, an economic analysis of the complete S⁰ recovery process was conducted (Fig. 6d, Supplementary Note 2 and Table 9). The results showed that recovering one ton of S⁰ from real SBW using the IDSR system yields a positive net profit of USD 185.2 and USD 201.0 for wastewater A and B, respectively (Fig. 6f). In addition, we compared the economic performance of S⁰ recovery by IDSR system and traditional biological oxidation²². The results indicated that the net profit generated by biological oxidation was approximately half that of the IDSR method, primarily because the former requires a substantial dosage of flocculants to facilitate S⁰ precipitation. The low purity of S⁰ caused by flocculants may affect its marketability. This comparison further strengthens the superiority of the proposed IDSR system over its biological counterpart.

Fig. 6. Application and economic evaluation of real sulfide-bearing wastewater.

a Sulfide removal rate and settling time of simulated wastewater (0.4 g L⁻¹), wastewater A (spent caustic) and wastewater B (Electronics manufacturing wastewater). b Sulfide removal efficiency and S⁰ selectivity of simulated wastewater, wastewater A and B. c S⁰ recovery ratio and S⁰ purity of simulated wastewater, wastewater A and B. d Simplified flow diagram of the S⁰ recycling process. The raw material inputs are shown in red, and the products are shown in blue. e Photographs of recycled S⁰ from simulated wastewater (S), spent caustic (A) and Electronics manufacturing wastewater (B). f Simplified economic summary. The analytical methods and parameters were based on Supplementary Note 2, and the detailed results of the relevant analyses are presented in Supplementary Table 9. Each value in (a–c) is the mean of three replicates, and error bars indicate standard deviation (samples n  =  3). Source data are provided as a Source Data file.

Discussion

In this study, we have developed a strategy for S⁰ recovery from wastewater, which demonstrated several advantages that address the limitations of existing technologies. Due to its hydrophilic nature and colloidal form, the separation of biogenic S⁰ typically relies on the addition of coagulants or a filtration process^25–27. In contrast, the IDSR system leverages the unique surface-state-dependent SALI agglomeration behavior of S⁰ in relation to iron flocs, facilitating straightforward separation through simple gravity-driven methods, which will be crucial for the usability of the recovered S⁰. The SALI behavior also ensures that the ferric (hydr)oxides remain uncovered by the elemental sulfur, thereby maintaining their sustained catalytic activity. In addition to S⁰ separation, the selective transformation of sulfide to S⁰ is also a limiting factor in biological S⁰ recovery technologies, as it necessitates oxygen-limiting conditions²², which compromises the reaction rate²⁴. By contrast, the IDSR system does not experience the trade-off between reaction selectivity and rate when converting sulfide to S⁰, and can be operated with an adequate oxygen supply, achieving both a high rate and selectivity. Moreover, in the biological S⁰ recovery process, the unexpected oxidation of accumulated S⁰ is a concern because the involved sulfide-oxidizing bacteria (SOB) can also use S⁰ as a substrate²¹. In contrast, although the tested wastewater in this study was enriched with nutrients suitable for SOB growth, this bio-derived S⁰ oxidation does not appear to occur in the IDSR system, as indicated by consistent S⁰ accumulation across 60 operation cycles (Supplementary Fig. 7). This result might be attributed to the generation of strong oxidative hydroxyl radicals during the oxidation of ferrous sulfides⁴², which is not conducive to bacterial growth.

The solid-solid separation, dominated by the S⁰ hydrophobic flocculation, is the cornerstone for sulfur recovery in the IDSR system. Given that similar large-diameter hydrophobic self-flocculates of S⁰ have not been previously reported, we identified that Stages (iii) and (iv) of the sulfur generation process in the IDSR system are the distinctive and indispensable stages that differentiate it from prior sulfur formation mechanisms. In Stages (i) and (ii), the prototypes of spherical S⁰ were formed. These were similar to the sulfur particles (80–100 nm) generated via conventional liquid chelated-iron H₂S removal systems⁴³, and the particle size reported in this study was larger (~200 nm). Although the S⁰ reported in previous liquid chelated-iron systems is also hydrophobic, owing to its formation via the direct hydrolysis of polysulfides, it can only form 1–10 μm particles through clustering⁴⁴ (which we speculate are likewise hydrophobic flocculates). This limited particle size occurs because the S⁰ generated in chelated-iron systems was discharged immediately after the reduction of Fe(III) to Fe(II), thereby lacking the opportunity for further growth within the desulfurization environment. In contrast, the spherical sulfur undergoes the superposition of zero-valent sulfur in Stage (iii) of the IDSR system, which promotes the growth of spherical particles into 1–3 μm spindle-shaped particles from 200 nm. This stage ensures a sufficient size for the individual grains. In Stage (iv), aeration induces the “closed-loop” transformation of the surface amorphous layer, resulting in fully crystallized hydrophobic sulfur. The synergistic effect of these two stages ensures that the S⁰ aggregates into particles exceeding 100 μm, making them easily separable. In fact, the enhanced separability between ferric (hydr)oxides and sulfur achieved through this growth mechanism provides a potential pathway for recovering sulfur from H₂S gases using iron solids, thereby eliminating the dependency on chelating agents in conventional chelated-iron methods^45,46.

The findings of the engineering system developed in this study may also shed light on a better understanding of the iron-related sulfur transformation in natural systems. The reaction of ferric (hydr) oxides and sulfide in natural aquatic systems has been extensively documented⁴⁷. In natural aquatic environments, the reaction products between ferric (hydr)oxides and sulfides are conventionally thought to be mackinawite, which subsequently transforms into pyrite⁴⁸. However, the regenerability of ferric (hydr)oxides within the IDSR system confirms that in the presence of oxygen, the final product remains ferric (hydr)oxides. Simultaneously, this process facilitates the catalytic mineralization of sulfides into S⁰. These two products have likely been overlooked in the current understanding of iron-sulfur transformations in aquatic systems. The regeneration of ferric (hydr)oxides could exert a considerable influence on the adsorption and migration of pollutants. Meanwhile, the formation of S⁰ suggests the potential existence of potential sources of sulfur in similar environments.

Methods

Chemicals

All chemicals were of analytical reagent grade unless otherwise stated. Sodium sulfide nonahydrate (Na₂S·9H₂O, purity ≥98%), magnesium sulfate heptahydrate (MgSO₄·7H₂O), ammonium chloride (NH₄Cl), iron(III) chloride nonahydrate (FeCl₃·6H₂O, purity ≥99%), ferrous chloride (FeCl₂), and carbon tetrachloride (CCl₄) were purchased from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Sodium bicarbonate (NaHCO₃), potassium dihydrogen phosphate (KH₂PO₄), sodium thiosulfate (Na₂S₂O₃), sodium hydroxide (NaOH), and sulfur (S₈, purity ≥99%) were obtained from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Disodium hydrogen phosphate dodecahydrate (Na₂HPO₄·12H₂O) was supplied by Macklin Biochemical Co., Ltd. (Shanghai, China). Anhydrous sodium sulfate (Na₂SO₄) was sourced from Xilong Chemical Co., Ltd. Hydrochloric acid (HCl, 37% solution) was provided by Harbin Chemical Reagent Factory. All reagents were used as received without further purification.

Synthetic sulfide-bearing wastewaters (SBW) were prepared using mineral solutions amended by a Na₂S and a mineral solution. The mineral solution contained 1.7 g L⁻¹ NaHCO₃, 0.06 g L⁻¹ MgSO₄ ∙ 7H₂O, 0.04 g L⁻¹ Na₂HPO₄ ∙ 12H₂O, 0.03 mg L⁻¹ KH₂PO₄, 0.1 g L⁻¹ NH₄Cl, 1.0 ml L⁻¹ trace element solution as described by Pfenning and Lippert⁴⁹. The pH of SBW was adjusted to 6.2 ± 1.0 using 6.0 N HCL prior to use, as a pre-testing of the IDSR system showed the optimal pHs of SBW falling into 5.0–7.0 (details can be found in Supplementary Note 3 and Supplementary Fig. 24). To minimize H₂S volatilization and oxidation in SBW preparation, two strategies were employed. First, aluminum foil gas bags were used to prepare and store the SBW, which can avoid the formation of headspace. Second, during the preparation of SBWs, deoxygenation by N₂ sparging was performed on the mineral solution before adding Na₂S stock solution to minimize the loss of sulfide.

Six-line ferrihydrite⁵⁰ was synthesized by dissolving 100.0 g of FeCl₃ ∙ 9H₂O into 2.0 L of deionized water (DI) and maintained at a temperature of 75 °C with rapid stirring for 10–12 min. During this process, the formation of a homogenous dark reddish suspension can be observed. Then, the suspension was rapidly cooled in an ice bath to room temperature and titrated to pH 6.2–6.5 with 1.0 M NaOH to get a ferrihydrite stock solution. Synthetic hematite and magnetite were prepared according to the standard methods of Cornell and Schwertmann⁵¹. Iron minerals were selected through pre-testing, and ferrihydrite was chosen for subsequent experiments due to its optimal properties (details can be found in Supplementary Note 4 and Supplementary Fig. 25).

Single-cycle batch tests of IDSR system for treating SBWs with varied sulfide concentrations

Single-cycle batch tests of IDSR system were conducted using 500 mL SCHOTT bottles equipped with a tube-connection module in an anaerobic glove box. The experiments were set up with SBWs and ferric (hydr)oxides solutions in a volume ratio of 3.0:1 (300 mL and 100 mL, respectively). The S(-II)-S concentrations in SBWs were set as 0.05, 0.1, 0.4, 1.0, 2.0 and 5.0 g L⁻¹, respectively. Accordingly, ferric (hydr)oxides solutions were used, resulting in the molar ratio of Fe to S(-II) falling into a range from 0.6:1 to 8.0:1. After the injecting SBWs and ferric (hydr)oxides solutions into the SCHOTT bottles, a magnetic stirrer was activated at 700 rpm for 5 min to ensure thorough mixing for the reaction between Fe(Ⅲ) and sulfide. Subsequently, the aeration process was carried out at an air flow rate of 5 L min⁻¹ by connecting the air inlet and outlet tubes to the outside of the glove box. The outlet gas was passed through a bottle containing 400 mL of 2.0 M NaOH solution to capture any released H₂S. All sampling and sample storage were done in a glove box. After aeration was completed, all solutions were transferred to a 500 ml measuring cylinder to measure the settling rate of the suspended material. For the above experiments, each test condition was set in triplicate.

IDSR reactors and operation in multi-cycles

The IDSR reactor was constructed from acrylic panels and has a total volume of 4 L, consisting of a 2 L cylindrical reaction zone and a 1 L conical settling zone. The water inlet was positioned at the middle of the settling zone, while the water outlet was located at the bottom of the reaction zone. The aeration inlet and gas outlet were situated at the bottom and top of the reactor, respectively. All gas and water ports were equipped with one-way valves. The gas outlet was connected to a gas-collecting bag and a bottle containing 2.0 M NaOH solution to capture any released hydrogen sulfide. The reactor was completely enclosed and tested for air-tightness before use to prevent sulfur imbalance due to hydrogen sulfide escape during the experiment. A mechanical agitator was mounted on top to ensure thorough mixing of the reactor contents.

The operation of IDSR reactors commenced with the introduction of 1 L freshly produced ferric (hydr)oxides solution, followed by a three-step sequencing batch procedure for multiple cycles. In Step-Ⅰ (SBWs injection and anaerobic mixing), 2 L SBW was pumped into the reactor and mixed with the solids in the reactor at an agitation of 500 rpm for 5 mins. Subsequently, Step-Ⅱ (aeration) was initiated, during which the reactor was aerated using an aeration pump at an air flow rate of 15 L min⁻¹ under stirring at 700 rpm. The final step (Step-Ⅲ) involved sedimentation and water discharge. Here, aeration was immediately ceased, and the agitation rate was gradually reduced to 0 rpm over 5 mins. After a well settling of solids, 2 L of treated desulfurized water was discharged.

The IDSR reactor was tested with two S(-II)-S concentrations in the SBWs (0.4 g L⁻¹ and 0.05 g L⁻¹). For the test of relative high S(-II) concentration, the molar ratio of Fe in the pre-filled ferric (hydr)oxides solution to the loaded S(-II) from SBW (referred to Fe/S ratio) was set at 1.2:1. Additionally, the aeration time in Step-Ⅱ and sedimentation time in Step-Ⅲ were set as 60 mins and 30 mins, respectively. For the test with low S(-II) concentrations, the Fe/S ratio, aeration time and sedimentation time were set as 6.0:1, 30 min and 60 min, respectively. These parameters were selected based on results obtained from single-cycle batch tests of IDSR system (detailed explanation can be found in Supplementary Note 1 and Supplementary Figs. 9 and 10). For each tested S(-II) concentration, the IDSR reactor was run three times for statistical analysis.

S⁰ separation and purification

The S⁰ separation through gravity settling was conducted in a cylindrical tube with a diameter of 5 cm and a height of 80 cm. The mixed S⁰ and ferric (hydr)oxides suspensions, taken from the IDSR reactor, were transferred into the tube and left to stand until a stable stratified interface of S⁰ and ferric (hydr)oxides was observed. Subsequently, the upper-layered ferric (hydr)oxides were removed via an adjustable discharge port set in the middle of the settling tube. Tap water was then added to the tube to resuspend the solids, followed by another run of gravity settling. This process was repeated three times, after which the S⁰ was drained from a discharge port set at the bottom of the settling tube.

The S⁰ separation via hydraulic sorting was carried out using a cylindrical tube with a diameter of 5 cm and a height of 160 cm, equipped with a water outlet and inlet at the top and bottom, respectively. The mixed S⁰ and ferric (hydr)oxides suspensions taken from the IDSR reactor were transferred into the tube. Subsequently, tap water was pumped into the tube through the water inlet with an incrementally increasing flow rate. As a result, the water flow velocities in the tube were set at 0, 0.08, 0.3, 0.4, 0.6, 0.8, 0.9 and 1.1 m min⁻¹, respectively. Solids washed out at each velocity were collected separately for further analysis. The switch of flow velocity depended on the absence of visible solids being flushed out. After hydraulic sorting, the remaining solids in the tube were washed with 1.2 M hydrochloric acid for further purification.

To conduct a statistical analysis of the separation experiments described above, the solids taken from the IDSR reactor were evenly divided into three portions and tested sequentially.

Applications in industrial wastewater

The spent caustic (wastewater A) used in this experiment was obtained from the absorption solution of hydrogen sulfide exhaust gas in a petroleum processing company in Handan, China. The electronics manufacturing wastewater (wastewater B) was sourced from the printed circuit board (PCB) etching wastewater, treated with sodium sulfide to remove heavy metals, in Shenzhen, China. Detailed water quality parameters for both types of wastewaters are provided in Supplementary Tables 7, 8. A total of 100 L of each type of wastewater was pumped into aluminum foil air bags to serve as influent. The iron redox process in both wastewaters followed the operational procedures described for the IDSR reaction system using simulated wastewater. Due to the sufficient alkalinity in the actual wastewater, which buffers the acidity introduced by FeCl₃ addition, the reagent was directly added as FeCl₃, dissolved in water under neutral conditions, forming ferric (hydr)oxides. Ten-cycle experiments were conducted using the IDSR reaction system. The producing solid was then transferred to a hydraulic sorting reactor for flushing at a flow rate of 1.1 m s⁻¹. The two types of real wastewaters underwent three replicate experiments consecutively.

Analytical methods

Aqueous samples were regularly collected for wet chemical analysis, with the sampling method consistently applied across all processes. Total sulfide was analyzed through an acid-purging process followed by an absorption process under alkaline conditions, and subsequently measured using the methylene blue method⁵² (details can be found in Supplementary Note 5). 10.0 ml samples for dissolution SO₄²⁻, S₂O₃²⁻, SO₃²⁻_, S²⁻ analyses were immediately spiked with 5.0 µl of 1.0 mol L⁻¹ NaOH solution. Dissolved S²⁻ was measured immediately by the methylene blue method after filtration (10 nm, Polycarbonate). Samples for SO₄²⁻, S₂O₃²⁻ and SO₃²⁻ were stored at 4 °C in airtight, air-free vials after filtration (10 nm, Polycarbonate) and analyzed within 4 h using an ion chromatograph (883 Basic IC plus, Metrohm, Switzerland) equipped with a Metrosep A Supp 5-250 column. The pH of the system was measured using a pH probe meter (WTW, pH).

Sample for zero-valent sulfur was extracted by pretreating 1.0 ml of unfiltered sample with 50.0 μL of ZnAc (c = 1.0 mol L⁻¹) to precipitate free polysulfide, following a modified procedure from Kamyshny et al.⁵³. After 10 min, 10.0 mL of methanol was added to the suspension. The sample was shaken for 4 h and then filtered (0.2 μm, Nylon). All zero-valent sulfur elements have been dissolved in methanol. The samples were analyzed by UPLC (Agilent G7114B, Germany) after separation on a C18 column (Zorbax, SB-C18, 80 A, 5 μm, 250 × 4.6 mm) with isocratic elution using a methanol and water mixture (97:3, v/v) at a flow rate of 2.0 mL min⁻¹, with detection at 255 nm.

Samples for total Fe(Ⅱ) were analyzed after extracting both the precipitate and the suspension with HCl (c = 0.5 mol L⁻¹) and filtering (10 nm, Polycarbonate) using the phenanthroline method⁵⁴. Dissolved Fe(Ⅱ) was determined immediately after filtration (10 nm, Polycarbonate) using the same method. Total Fe was measured after digesting the sample with 6.0 M HCl and subsequently reducing it with 10% hydroxylamine hydrochloride.

After completing aeration for all cycles, 40.0 mL of solution was transferred to a 50 mL measuring cylinder to measure the settling rate of the suspended material. After settling was complete, the overlying water was drained in the same proportion, and the sediment was discharged into the reactor to participate in subsequent reactions.

Details on the calculations for sulfide removal efficiency and rate, as well as the selectivity and purity of S⁰, are provided in Supplementary Note 6.

Characterization

Field-emission scanning electron microscopy (FESEM, Hitachi S4800) and optical microscopy (Mshot, MF43-M) were used to observe the microstructure of suspended solids in the reaction. The particle size of the suspended material was determined using a Laser Particle Sizer (Malvern, Mastersizer 3000). The largest projection length was defined as the suspended particle size in this study and employed to evaluate the floc size. The particle size of S⁰ and ferric (hydr)oxides was calculated based on colors differentiation from microscopy images, using Image-Pro Plus 6.0, respectively. For imaging observations, samples from three independent replicate experiments were each observed to verify the reproducibility. A Zeta potential analyzer (Malvern Model Nano-Z Zetasizer) was utilized to assess the chargeability of materials in solutions. Contact angle measurements were conducted with a contact angle determinator (Lunderskov). The suspended material was tested after being filtered, rinsed, air-dried and then compacted using a tablet press. An automatic titration of 8 μl droplets was deposited on the surface of the material, and images were acquired every 50 ms.

The wide-angle powder X-ray diffraction technique (XRD, Ultima IV) with Cu-Kα radiation was used to determine the crystal phases present in the samples. The cryogenic X-ray photoelectron spectrometer (XPS, Thermo ESCALAB 250XI) was employed to determine the elemental composition and valence states of atoms within the samples. Throughout the analysis, the pressure and sample temperature were maintained at 10⁻⁹ Torr and −135 °C, respectively. All binding energy measurements were calibrated using the C 1 s spectrum (284.8 eV). Mössbauer measurements were performed at 13 K using a conventional spectrometer (Germany, Wissel MS-500) in transmission geometry with constant acceleration mode. A ⁵⁷Co(Rh) source with activity of 25 mCi was used. The velocity calibration was conducted with a room temperature α-Fe absorber. The spectra were fitted using the software Recoil with Voigt-based Fitting (VBF) Analysis. Samples susceptible to oxidation were prepared in an anaerobic glove box to prevent oxidation, and transfers were made through at least three layers of plastic sealing. Given that a S(-II)-S concentration of 0.4 g L⁻¹ is more representative, all characterizations were conducted based on the solid produced during treatment at this concentration.

Reporting summary

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

Author contributions

Conceptualization: D.H. Ren, A.J. Wang, H.Y. Cheng. Investigation: D.H. Ren, W.K. He, X.C. Ba, Y.X. Cen, S.K. Yuan. Methodology: D.H. Ren, W.K. He, X.C. Ba, Y.X. Cen, S.K. Yuan, J.M. Xu, N. Zhang. Validation: D.H. Ren, W.K. He. Data curation: D.H. Ren, W.K. He, X.C. Ba, Y.X. Cen, S.K. Yuan. Formal analysis: D.H. Ren, W.K. He. Visualization: D.H. Ren, W.K. He, X.C. Ba, J.M. Xu. Resources: A.J. Wang, H.Y. Cheng. Supervision: A.J. Wang, H.Y. Cheng. Writing - original draft: D.H. Ren, H.Y. Cheng. Writing - review & editing: D.H. Ren, Z.Q. Zuo, G.J. Zhang, H.Y. Cheng. Project administration: G.J. Zhang, A.J. Wang, H.Y. Cheng. Funding acquisition: A.J. Wang, H.Y. Cheng.

Peer review

Peer review information

Nature Communications thanks the anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

This study was supported by National Natural Science Foundation of China (No.52321005 and No.52500028), National Key R&D Program of China (No. 2023YFC3207100), Basic Research Project of Shenzhen Natural Science Foundation (No. JCYJ20230807094315032), and GuangDong Basic and Applied Basic Research Foundation (No. 2023A1515110670), and State Key Program of Shenzhen Natural Science Foundation (No. QNXMC20250701092036009).

Data availability

The raw data generated in this study have been deposited in the Figshare repository (https://doi.org/10.6084/m9.figshare.32657256). Source data are provided in the Source Data file. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76480-w.