In-situ low-temperature sulfur CVD on metal sulfides with SO₂ to realize self-sustained adsorption of mercury

Abstract

Capturing gaseous mercury (Hg⁰) from sulfur dioxide (SO₂)-containing flue gases remains a common yet persistently challenge. Here we introduce a low-temperature sulfur chemical vapor deposition (S-CVD) technique that effectively converts SO₂, with intermittently introduced H₂S, into deposited sulfur (S_d⁰) on metal sulfides (MS), facilitating self-sustained adsorption of Hg⁰. ZnS, as a representative MS model, undergoes a decrease in the coordination number of Zn–S from 3.9 to 3.5 after S_d⁰ deposition, accompanied by the generation of unsaturated-coordinated polysulfide species (S_n^2–, named S_d^*) with significantly enhanced Hg⁰ adsorption performance. Surprisingly, the adsorption product, HgS (ZnS@HgS), can serve as a fresh interface for the activation of S_d⁰ to S_d^* through the S-CVD method, thereby achieving a self-sustained Hg⁰ adsorption capacity exceeding 300 mg g⁻¹ without saturation limitations. Theoretical calculations substantiate the self-sustained adsorption mechanism that S₈ ring on both ZnS and ZnS@HgS can be activated to chemical bond S₄ chain, exhibiting a stronger Hg⁰ adsorption energy than pristine ones. Importantly, this S-CVD strategy is applicable to the in-situ activation of synthetic or natural MS containing chalcophile metal elements for Hg⁰ removal and also holds potential applications for various purposes requiring MS adsorbents.

Introduction

Mercury pollution has aroused global concern mainly ascribed to the volatility, insolubility, and long-range transport of gaseous elemental mercury (Hg⁰), which is the focus of mercury abatement^1,2,3,4. Effective control of Hg⁰ pollutant necessitates the use of functional materials that exhibit high activity, stability, and tolerance to flue gas conditions^5,6,7. Recent developments in materials for Hg⁰ capture include carbon-based^8,9, oxide-based¹⁰, and noble metal-based materials^11,12. However, the presence of SO₂ in any flue gas has been known to negatively impact Hg⁰ removal, leading to surface sulfation or the occupation of active sites¹³. More unfortunately, the non-ferrous metal smelting industry is considered as the largest single source of Hg⁰ emissions, where high concentrations of SO₂ and Hg⁰ co-exist¹⁴. Therefore, achieving large-capacity adsorption of Hg⁰ at high concentration of SO₂ remains a significant challenge.

While many sulfur-based materials have shown a degree of resistance to SO₂ in Hg⁰ adsorption^15,16, their capacities often experience significant suppression due to active site depletion or deactivation, particularly at high SO₂ concentrations¹⁷. This limitation frequently requires off-line regeneration under harsh conditions (e.g., heating or acidification treatments with irreversible destruction) or even necessitates the replacement of adsorbents^18,19. To address the challenge of active depletion or deactivation, a more cost-effective and convenient method involves the continuous replenishment of active sites in situ at the interface of sulfur-based materials. Recognizing that the performance of metal sulfides (MS) relies heavily on the quantity of active sulfur sites²⁰, a viable strategy is to directly convert SO₂ to sustainably replenish surface active sites, thereby turning the negative effect of SO₂ into a positive one. However, the high average S−O bond energy of SO₂ (548 kJ mol⁻¹) necessitates high-temperature conditions (>2000 °C) for its decomposition^21,22, whereas the preferential reaction of SO₂ with flue gas O₂ impedes the feasibility of this pathway²³. Notably, the assistance of H₂S can lower the S−O bond breaking energy barrier (139 kJ mol⁻¹) and further reorganize the S−S bond to generate elemental sulfur through the Claus reaction^24,25. Fortunately, H₂S or its raw materials (Na₂S or NaHS) are easily accessible and commonly used for heavy metals removal from various wastewaters in non-ferrous smelters^26,27.

However, two key challenges persist in achieving our objectives. Firstly, it involves effectively generating fresh sulfur on MS surface. More importantly, it pertains to activating the deposited sulfur (S_d⁰) for site replenishment instead of allowing it to aggregate into the inert S₈⁰ state (octatomic ring structure with poor Hg⁰ adsorption activity^28,29). To address these challenges, we have developed a sulfur chemical vapor deposition (S-CVD) method using the Claus reaction between excessive SO₂ in flue gases and intermittently added H₂S. This method facilitates the deposition of gas-phase sulfur species with high controllability and scalability. Furthermore, an active interface is crucial to bonding with S_d⁰ to create unsaturated coordination sites rather than coordination-saturated S₈. Notably, the incorporation of anchoring sites to bond with sulfur can maintain its unsaturated state³⁰. Chalcophile elements exhibit a natural tendency to lose outer electrons to form an 18-electron outermost structure (s²p⁶d¹⁰), which in turn combines with sulfur (3s²3p⁴) to form an ionic compound under ambient conditons^31,32. Thus, MS containing chalcophile metals emerges as promising candidates to bond with S_d⁰, preventing it from falling into a saturated-coordinated ring structure.

Hence, this work employs the proposed in-situ S-CVD method on chalcophile MS to counteract the negative effects of SO₂ and achieve self-sustained Hg⁰ adsorption, enabling in-situ reactivation without the need to replace spent adsorbents. Variety of experimental conditions and characterization methods, such as scanning electron microscopy (SEM), X-ray absorption fine structure (XAFS), and density functional theory (DFT) calculations, were devoted to evaluating the self-sustained adsorption performance, revealing the deposition process of S_d⁰, identifying the formation of unsaturated coordination environments, and calculating the energy changes to interpret the self-sustained adsorption mechanism. The results indicate that S_d⁰ can be efficiently activated to polysulfide (S_n²⁻, named S_d^*) species by chalcophile MS, including the formed HgS itself, ensuring self-sustained Hg⁰ adsorption. This in-situ S-CVD approach provides a promising solution to active site depletion and poisoning issues and offers an avenue for efficient and continuous heavy metal removal using MS materials.

Results

Establishment of in-situ S-CVD method for SO₂ deposition

The in-situ S-CVD method was established for flue gas SO₂ deposition. To initiate S-CVD, a small amount of H₂S (100 ppm) was injected into the SO₂-containing flue gas upstream of Al₂O₃@MS adsorbents (Supplementary Fig. 1). In actual non-ferrous smelting processes, flue gas particle-bond mercury (Hg^p) and oxidized mercury (Hg²⁺) can be respectively removed by an electrical precipitator and scrubber, resulting in a subsequent flue gas with high concentrations of SO₂ and Hg⁰ (Supplementary Fig. 2). Extraction of approximately 0.1‰ of total SO₂ for on-site conversion to H₂S^27,33 can satisfy the needs of S-CVD, and Hg⁰ will be removed by the proposed self-sustained adsorption method on adsorbents (Fig. 1a). In this method, MS functions as the fresh surface for the initial S-CVD and Hg⁰ adsorption, and then the spent MS (i.e., MS-HgS) acts as new surface for further S-CVD and Hg⁰ adsorption, ultimately achieving the sustained adsorption by HgS itself (Fig. 1b).

Fig. 1: Proposed in-situ S-CVD strategy for Hg⁰ self-sustained adsorption on metal sulfides.

a Hg⁰ removal through proposed in-situ S-CVD strategy in smelting flue gas. b Schematic illustration of the Hg⁰ self-sustained adsorption on metal sulfides. b1 Initial stage, S_d⁰ activated only by MS; b2 Transition stage, S_d⁰ activated by MS and/or HgS; b3 Sustained stage, S_d⁰ activated by HgS itself.

During the S-CVD process, the reaction ratio of H₂S and SO₂ was monitored as 2.1: 1 and chemical composition analysis exhibited that the sulfur content in different Al₂O₃@MS increased by 2.1%−3.0% after 180 min of S-CVD (Supplementary Fig. 3). These confirm the occurrence of the Claus reaction on the Al₂O₃@MS surface, which is a critical step in the S-CVD process. Furthermore, the optimal adding sequence of H₂S and SO₂ was investigated to understand the generation mechanism of S_d⁰. The results demonstrated that the activity of Al₂O₃@MS pretreated with H₂S followed by SO₂ was significantly lower than that pretreated with SO₂ followed by H₂S (Supplementary Fig. 4). Meanwhile, pretreatment only by SO₂ cannot enhance the activity of Al₂O₃@MS. This finding indicates that the formation of S_d⁰ on Al₂O₃@MS surface followed the Eley-Rideal mechanism, in which SO₂ is first adsorbed on adsorbent surface and then reacts with gaseous H₂S to produce S_d⁰ (Eq. 1, 2):

$${{{{{{\rm{SO}}}}}}}_{2}({{{{{\rm{g}}}}}})+{{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@MS}}}}}} \to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@MS}}}}}}{\mbox{-}}{{{{{{\rm{SO}}}}}}}_{2}({{{{{\rm{ads}}}}}})$$

(1)

$${{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@MS}}}}}}{\mbox{-}}{{{{{{\rm{SO}}}}}}}_{2}({{{{{\rm{ads}}}}}})+{2{{{{{\rm{H}}}}}}}_{2}{{{{{\rm{S}}}}}}({{{{{\rm{g}}}}}})\to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@MS}}}}}}{\mbox{-}}{3{{{{{\rm{S}}}_{{{\rm{d}}}}^0}}}}+{2{{{{{\rm{H}}}}}}}_{2}{{{{{\rm{O}}}}}}$$

(2)

Importantly, the negative Gibbs free energy (ΔG⁰ = −91 kJ mol⁻¹ at 25 °C) of the Claus reaction and the much lower concentration of H₂S (100 ppm) compared to SO₂ (≥5000 ppm) used in S-CVD guarantee the sufficient reaction of added H₂S.

Deposited sulfur activation on MS for Hg⁰ removal

The surface properties of metal sulfides play a vital role in S-CVD process. Both synthetic and natural MS served as the deposition surface (Fig. 2a). In light of Goldschmidt geochemical classification of the elements, metals with chalcophile nature have higher affinity towards sulfur, which could accelerate the stimulation of S_d⁰ (refs. ^31,32). Thus, a range of typical chalcophile metal elements, including Cu, Zn, In, Cd, Pb, and Sn, were chosen as candidates for the synthesis of MS, while some siderophile metals, including Mn, Fe, Co, and Ni, were offered as contrasts (Fig. 2b). As depicted in Fig. 2c and Supplementary Table 1, after 15 min of S-CVD, various Al₂O₃@MS-S_d showed substantial differences in enhancing their Hg⁰ adsorption performances, of which all the chalcophile Al₂O₃@MS-S_d exhibited significantly increase in their adsorption capacities. Notably, Al₂O₃@CuS-S_d demonstrated a remarkable increase the Hg⁰ adsorption capacity, rising from 178.9 to 640.4 mg mol⁻¹(within 180 min and normalized to MS molar mass). Similarly, Al₂O₃@ZnS-S_d exhibited a significant enhancement, with adsorption capacity increasing from 11.0 to 457.3 mg mol⁻¹. Al₂O₃@CdS-S_d, Al₂O₃@In₂S₃-S_d, Al₂O₃@PbS-S_d, and Al₂O₃@SnS-S_d also showed considerable improvements in their Hg⁰ adsorption capacities, reaching 454.4, 431.7, 401.6, and 275.9 mg mol⁻¹, respectively. However, the adsorption capacities of Al₂O₃@Fe₂S₃-S_d, Al₂O₃@CoS-S_d, Al₂O₃@MnS-S_d, and Al₂O₃@NiS-S_d were comparatively lower, reaching 207.4, 165.3, 139.7, and 125.4 mg mol⁻¹, respectively. In addition, the Al₂O₃-S_d, Al₂O₃@ZnSO₄-S_d, and Al₂O₃@Na₂S-S_d did not show improved performance compared to their raw materials (Supplementary Fig. 5), implying that the isolated presence of metal or sulfur sites cannot directly activate the deposited S_d⁰.

Fig. 2: Enhanced performance of different metal sulfides using in-situ SCVD activation.

a Proposed in-situ S-CVD strategy on synthesized metal sulfides or natural sulfide ores. b Geochemical classification of elements containing chalcophile and siderophile elements. c Hg⁰ adsorption capacities of different Al₂O₃@MS and Al₂O₃@MS-S_d (M = Cu, Hg, Zn, Cd, In, Pb, Sn, Ni, Co, Fe, and Mn). Reaction conditions: adsorbent mass = 0.3 g, total flow rate = 360 mL min⁻¹, SO₂ concentration = 5000 ppm (during S-CVD process), H₂S concentration = 100 ppm (during S-CVD process), S-CVD time = 15 min, Hg⁰ concentration = (1.5 ± 0.05) mg m⁻³, reaction temperature = 80 °C, and reaction time = 180 min.

Further, to quantitatively explore the role of metal sites in MS on the activation of S_d⁰, we further construct the relationship between the Hg⁰ adsorption capacity increment (Q_i = Q_Al2O3@MS-Sd − Q_Al2O3@MS) and metal-sulfur (M−S) bond energy (measured by bond length³⁴, Supplementary Table 2) of the investigated metal sulfides. As depicted in Supplementary Fig. 6, the Q_i of chalcophile MS showed a negative relationship with the increase of M−S bond length. However, for siderophile MS, higher M−S bond length instead led to relatively higher Q_i. It can be deduced that the activity of S_d⁰ on the sulfide interface highly depends on the geochemical characteristics of metal element and the M−S affinity. Notably, Hg also belongs to chalcophile elements (Fig. 2b), and it is supposed that HgS itself could potentially play a role in activating S_d⁰ (see later in Fig. 5e).

We then choose ZnS as a representative model to optimize the reaction conditions owing to its significant performance enhancement (~42 times) and simple metal and sulfur speciation. The sulfur content in Al₂O₃@ZnS-S_d exhibited a three-stage growth pattern with increasing S-CVD time (Fig. 3a). The growth rate of S_d⁰ decreased from 2.5 mg g⁻¹ min⁻¹ at 0−5 min to 0.2 mg g⁻¹ min⁻¹ at 5−30 min then to 0.06 mg g⁻¹ min⁻¹ at >30 min (Supplementary Table 3). The decreasing formation rate indicates that the fresh surface of the adsorbent was gradually covered by S_d⁰, and once the surface was completely covered, subsequent S_d⁰ would generate on the existing S_d⁰ layer, resulting in a final slow but steady growth rate. Figure 3b presents the relationship between S_d⁰ increment and Hg⁰ removal efficiency of Al₂O₃@ZnS-S_d. The results showed that the removal efficiency gradually increased to 71.7% when the ratio of S_d⁰/Zn increased to 2.7 (corresponding to 60 min of S-CVD); while, as the ratio further increased to 4.1 (240 min), 6.0 (480 min), and 13.5 (1440 min), the removal efficiency remained at a stable level of around 83%. This indicates that excess deposited S_d⁰ did not work, presumably attributed to the inevitable aggregation of excess S_d⁰ into inert S₈⁰ (ref. ²⁹). When the adsorption capacity is normalized to the mole of S_d⁰ (red curve in Fig. 3b), it reached a maximum of 467.5 mg mol⁻¹ at the S_d⁰/Zn ratio of 2.1 (15 min) and then gradually decreased to 89.3 mg mol⁻¹ as the ratio increased to 13.5. Therefore, 30 min of S-CVD was chosen as the optimal condition by integrating the Hg⁰ removal efficiency and S_d⁰ utilization. Besides S-CVD time, reaction temperatures and other flue gas components also played significant roles. Thermogravimetric analysis (TGA) results indicated the high thermal stability of Al₂O₃@ZnS-S_d below 200 °C (Supplementary Fig. 7). The increase in adsorption temperature from 60 to 120 °C improved the Hg⁰ removal efficiency of Al₂O₃@ZnS-S_d from 23.8% to 89.9% within 180 min (Supplementary Fig. 8). However, further elevation of temperature to 140 °C and 160 °C resulted in decreasing activity to 86.6% and 75.9%, respectively, presumably due to the re-decomposition of partially captured Hg⁰ (Supplementary Fig. 9). Moreover, Al₂O₃@ZnS-S_d demonstrated high tolerance to different gas components (Supplementary Fig. 10). The addition of 5000 ppm SO₂ in adsorption process resulted in enhanced removal efficiency by 3.2% at 120 °C, and further addition of 5% O₂ or 100 ppm NO had slight influence with a reduction of 2.6% and 6.8%, respectively. The introduction of 5000 ppm SO₂ + 4% H₂O showed negative effect on Hg⁰ adsorption performance of Al₂O₃@ZnS-S_d, presumably due to the competing adsorption between H₂O and Hg⁰ on the active sites³⁵ and the hydrophilicity of Al₂O₃ (ref. ³⁶); however, its removal efficiency maintained at a stable level of ~70% without reduction for 180 min adsorption.

Fig. 3: Influencing factors of in-situ SCVD strategy for Hg⁰ adsorption over Al₂O₃@ZnS-S_d.

a S_d⁰ increment in Al₂O₃ and Al₂O₃@ZnS with the increase of S-CVD time. The red, green and blue-gray highlights represent three stages of different S_d⁰ growth rates. b The relationship between S_d⁰ increment and Hg⁰ adsorption capacity of Al₂O₃@ZnS-S_d. c The effect of SO₂ concentration on Hg⁰ adsorption breakthrough curve of Al₂O₃@ZnS-S_d. d The effect of intermittent addition of SO₂ on the Hg⁰ adsorption curve over Al₂O₃@ZnS-S_d. Reaction conditions: adsorbent mass = 0.3 g, total flow rate = 360 mL min⁻¹, Hg⁰ concentration = (1.5 ± 0.05) mg m⁻³, SO₂ concentration = 5000 ppm (for a and b), H₂S concentration = 100 ppm (during the S-CVD process), temperature = 80 °C (for a), and S-CVD time = 30 min (for b and c).

Long-term experiments were conducted to clarify the SO₂ effect on Hg⁰ removal during the S-CVD and adsorption process under the optimal condition (H₂S = 100 ppm, temperature = 120 °C, S-CVD time = 30 min) (Fig. 3c). In stage I (S-CVD process), without the presence of SO₂, Al₂O₃@ZnS initially exhibited a temporary increase in Hg⁰ removal but rapidly decreased once the S-CVD was stopped. This improvement might be attributed to the sulfuration of residual metal salt precursors on the surface. However, when SO₂ was added during the S-CVD process, the Hg⁰ removal sharply increased to around 90%. Furthermore, SO₂ continued to exhibit an extraordinary effect on the performance of Al₂O₃@ZnS-S_d after the S-CVD process. As shown in stage II (without H₂S), the slope of curve without SO₂ addition in the adsorption process (black line) was lower than that with SO₂ addition (red one), indicating a positive effect of SO₂, which is contrary to our traditional perceptions. Moreover, as the SO₂ concentration increased from 0 ppm to 5000 ppm and to 33,000 ppm, the Hg⁰ removal after 1440 min reaction increased from 46.1% to 64.0% and to 75.0%, respectively. To verify this unexpected positive effect, SO₂ was added intermittently to the SO₂-free system in stage II (Fig. 3d). The results showed that once 5000 ppm of SO₂ was introduced, the Hg⁰ removal efficiency was instantly improved by about 35%; however, once the SO₂ was turned off, it dropped back down to a lower level immediately. Thus, SO₂ was identified as the positive component for Hg⁰ adsorption over the Al₂O₃@ZnS-S_d. Through Hg⁰ temperature-programmed desorption (Hg⁰-TPD) experiments, we analyzed the changes in de-mercury products (Supplementary Fig. 9). The desorption peaks of spent Al₂O₃@ZnS-S_d in the absence of SO₂ located at 210−265 °C, which are attributed to the decomposition temperature of β-HgS³⁷. While, after adsorption in the presence of SO₂, there appeared a new peak centered at higher temperature of 365 °C, which is assigned to the decomposition temperature of α-HgS³⁸, thereby improving its Hg⁰ adsorption stability. Additionally, the performance of Al₂O₃@MS without S-CVD assistance was significantly reduced at high concentrations of SO₂, excluding the promotional effect of SO₂ on Al₂O₃@MS itself (Supplementary Fig. 11).

Self-sustained adsorption performance of Al₂O₃@MS-S_d

Although Al₂O₃@ZnS-S_d exhibited high performance and SO₂ can further promote the Hg⁰ adsorption, it still has a limited number of active sites and requires replacement of the adsorbent once depleted. If we can periodically replenish the S_d⁰ on the spent Al₂O₃@ZnS-S_d surface, self-sustained adsorption of Hg⁰ can be achieved without the need for adsorbent replacement. Hence, simulated flue gas ((2.5 ± 0.05) mg m⁻³ Hg⁰, 5000 ppm SO₂, 4% H₂O, total flow rate = 300 mL min⁻¹, and reaction temperature = 120 °C) was applied to investigate the self-sustained adsorption performance of Al₂O₃@ZnS-S_d. 100 ppm of H₂S was intermittently injected into the flue gas for 30 min per 3, 6, and 24 h to replenish S_d⁰ (Fig. 4a–c). The Hg⁰ removal of Al₂O₃@ZnS-S_d can reach around 95% after each round of S-CVD except the first round (83.2%). After the supplemented S_d⁰ was consumed by Hg⁰, the removal efficiency gradually decreased. Meanwhile, as the number of S-CVD increased, the 3, 6, and 24 h breakthrough ratios gradually converged to ~20%, ~35%, and ~78%, respectively. Additionally, owing to the higher performance of Al₂O₃@CuS-S_d at low temperatures (Supplementary Fig. 12), it displayed the self-sustained adsorption performance at 60 °C, with an initial Hg⁰ removal efficiency of ~90% and a 24 h breakthrough ratio of ~50% in each round of reaction (Supplementary Figs. 13). Moreover, Al₂O₃@ZnS-S_d can restrore its original activity after Hg⁰ desorption and secondary S-CVD (Supplementary Fig. 14). Additionally, we evaluated the Hg⁰ re-emission of Al₂O₃@ZnS-S_d after ten rounds of reaction. At the reaction temperature, the Hg⁰ re-emission concentration can be reduced from 1.7 mg m⁻³ to 0.05 mg m⁻³ (emission standard for non-ferrous smelting flue gas in China) in 30 min with the assistance of S-CVD, and once the temperature dropped to room temperature, the Hg⁰ concentration rapidly decreased to 0 mg m⁻³ (Supplementary Fig. 15). Thus, the S-CVD strategy not only can fulfill the self-sustained adsorption of Hg⁰, but also inhibit the re-emission of adsorbed mercury.

Fig. 4: Self-sustained adsorption performance of Hg⁰.

Hg⁰ adsorption breakthrough curves of Al₂O₃@ZnS-S_d assisted with (a) 3 h, b 6 h, and c 24 h intermittent S-CVD. Hg⁰ adsorption breakthrough curves of (d) natural chalcopyrite ore and (e) sphalerite ore assisted with intermittent S-CVD. Insets in d and e: photographs of the corresponding natural ores. Reaction conditions: adsorbent mass = 0.4 g (for a–c) or 1 g (for d and e), temperature = 120 °C (for a–d) or 40 °C (for e), [Hg⁰] = (2.5 ± 0.05) mg m⁻³, [SO₂] = 5000 ppm (for a–c) or 6% (for d and e), [H₂O] = 4%, [H₂S] = 100 ppm (during S-CVD), and total flow rate = 300 mL min⁻¹. f Photographs of Al₂O₃@ZnS in different reaction stages. g EDS mapping images of the cross-section of spent Al₂O₃@ZnS-S_d. Inset: line scanning results of the selected position.

Besides synthesized Al₂O₃@MS, natural sulfide ores that contain chalcophile metal elements, such as chalcopyrite and sphalerite, also enable their potential for self-sustained Hg⁰ adsorption. As depicted in Fig. 4d, at conditions of near-actual flue gas SO₂ concentration (6%), natural sphalerite ore achieved ~95% Hg⁰ removal after each round of S-CVD, and the breakthrough ratio converged to ~80%. Chalcopyrite exhibited enhanced Hg⁰ adsorption performance with intermittent S-CVD at a lower temperature (40 °C), which reached a 100% initial Hg⁰ removal efficiency and had a 24 h breakthrough ratio of ~50% in each round of reaction (Fig. 4e). Thus, directly utilization of natural sulfide ores as adsorbents can effectively lower the cost for Hg⁰ pollution control as well as improve the adsorption capacity taking advantage of their self-sustained adsorption properties.

Figure 4f shows the macrophotographs of Al₂O₃@ZnS in different reaction stages. After the first round of S-CVD, the color of Al₂O₃@ZnS changed from white to pale yellow, verifying the formation of S_d⁰. Then, the color of spent Al₂O₃@ZnS-S_d gradually turned black after 140 h reaction (six rounds of S-CVD and Hg⁰ adsorption). The cross-sectional view of spent Al₂O₃@ZnS-S_d pellets (after ten rounds) in Fig. 4f depicted an obvious black shell. The energy dispersion X-ray spectroscopy (EDS) mapping images and the selected line scanning curves of the cross-section of Al₂O₃@ZnS revealed that the contents of S and Zn elements on Al₂O₃ pellet gradually decreased from outside in (Supplementary Fig. 16). After reaction, S and Hg elements were more scattered in the outer layer and exhibited synchronous linear increase at the boundary layer (~50 μm); while, there presented a decline in Zn element at this boundary layer (Fig. 4g). In addition, the S and Hg contents on the surface of the spent Al₂O₃@ZnS-S_d was much higher than the Zn content (Supplementary Fig. 17). This indicates a layer-by-layer outward deposition of S_d⁰ and adsorption of Hg⁰ on Al₂O₃@ZnS surface. Moreover, assume that the Zn: S ratio in raw Al₂O₃@ZnS is 1:1, the ratio of Hg to S_d⁰ in spent Al₂O₃@ZnS-S_d was calculated as 1.04 in light of the ESD result of the cross-section (Supplementary Fig. 18), giving an indication that S_d⁰ atoms were fully utilized for Hg⁰ adsorption.

Mechanism for self-sustained Hg⁰ adsorption on Al₂O₃@MS-S_d

Identifying the S_d⁰ activation and Hg⁰ adsorption behaviors on Al₂O₃@ZnS contributes to a profound insight into the self-sustained adsorption mechanism of Hg⁰. The Brunauer−Emmett−Teller (BET) surface area, total pore volume, and average pore size of Al₂O₃, Al₂O₃@ZnS, and Al₂O₃@ZnS-S_d were not significantly different (Supplementary Table 4), suggesting that the coating of ZnS and the deposition of S_d⁰ did not affect the pore structure of Al₂O₃. The X-ray diffraction (XRD) pattern of Al₂O₃@ZnS-S_d showed that, besides the diffraction peaks assigned to γ-Al₂O₃ (JCPDS no. 79-1558) and sphalerite ZnS (JCPDS no. 77-2100), no peaks related to elemental sulfur emerged (Fig. 5a), suggesting the amorphous structure of formed S_d⁰. The Raman spectrum of Al₂O₃@ZnS-S_d charactered the peaks at 155.6, 223.9, 443.8, and 475.3 cm^–1, which are related higher than Raman shift of S₈⁰ (Fig. 5b). This indicates that the surface ZnS can change its S–S vibration of S_d⁰ and prevent its aggregation³⁹. In the X-ray photoelectron spectroscopy (XPS) S 2p spectrum of Al₂O₃@ZnS-S_d, aside from the peaks ascribed to S²⁻ (161.7 eV and 162.9 eV)⁴⁰ and SO₄²⁻ (168.4 eV and 169.6 eV)⁴¹, new peaks related to S_n²⁻ (163.4 eV and 164.6 eV)⁴² occurred (Fig. 5c). Additionally, the proportion of S_n²⁻ in Al₂O₃@CuS-S_d increased from 21.6% to 29.7% (Supplementary Fig. 19, Supplementary Table 5), demonstrating the consistency of the sulfur chemical state on chalcophile metal sulfides. However, in comparison, Al₂O₃-S_d featured its characteristic peaks of S 2p_3/2 at binding energies of 164.0 eV and 167.8 eV (Supplementary Fig. 20), which were ascribed to S₈⁰ and adsorbed unreacted SO₂ species, respectively⁴³. Moreover, the binding energy of Zn²⁺ 2p_3/2 in Al₂O₃@ZnS-S_d shifted from 1022.0 eV to 1021.8 eV after S_d⁰ deposition (Fig. 5d), indicating the formation of unsaturated coordination environments⁴⁴. To better observe the microscopic changes and the dynamic evolution of deposited S_d⁰, pure ZnS was further synthesized in the same way without adding Al₂O₃ pellets to directly serve as support for S-CVD process. The transmission electron microscope (TEM) images of ZnS-S_d showed a decrease in the contrasts of zinc atoms (Supplementary Fig. 21), indicating the formation of Zn defects¹⁹. The XAFS S L-edge spectra confirmed the formation of S_n²⁻ species in ZnS-S_d (Supplementary Fig. 22a). The extended XAFS (EXAFS) Zn K-edge spectra further revealed a decrease from 3.9 to 3.5 in the coordination number of Zn to S atoms in ZnS-S_d compared to that in pristine ZnS (Supplementary Fig. 22b and c, and Supplementary Table 6), further certifying the formation of unsaturated coordination sites. The XPS depth profiling results depicted that with the Ar⁺ etching depth increased to 6 mm, the average valence state of S_n²⁻ decreased, suggesting a shortening of the S_n²⁻ chain length close to the ZnS surface (Supplementary Fig. 23). Furthermore, in-situ Raman spectra revealed conversion of S₈ to S_n²⁻ in ZnS-S_d at elevated temperatures (Supplementary Fig. 24), which is also demonstrated by Fourier transform infrared spectroscopy (FTIR) and ¹³C nuclear magnetic resonance (NMR) using propylene as an indicator (Supplementary Fig. 25). Therefore, these imply that S_d⁰ does not simply physically accumulate on the Al₂O₃@ZnS surface in form of S₈⁰ (Eq. 3), but can be activated by Zn atoms and generated chemically bonded S_n²⁻ (S_d^*) with unsaturated coordination environments (Eq. 4):

$${{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}+{{{{{\rm{S}}}_{{{\rm{d}}}}^0}}}\to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{\mbox{-}}{{{{{\rm{S}}}_{{{\rm{d}}}}^0}}}\to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{\mbox{-}}{{{{{\rm{S}}}_{{{\rm{8}}}}^0}}}$$

(3)

$${{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@ZnS}}}}}}+{{{{{\rm{S}}}_{{{\rm{d}}}}^0}}}\to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@ZnS}}}}}}{\mbox{-}}{{{{{{\rm{S}}}}}}}_{{{{{\rm{d}}}}}}^{\ast }$$

(4)

Fig. 5: Self-sustained adsorption mechanism.

a XRD patterns of Al₂O₃@ZnS, Al₂O₃@ZnS-S_d, and spent Al₂O₃@ZnS-S_d. b Raman spectra of Al₂O₃@ZnS, Al₂O₃@ZnS-S_d, spent Al₂O₃@ZnS-S_d, and pure S₈. c S 2p and d Zn 2p XPS spectra of Al₂O₃@ZnS, Al₂O₃@ZnS-S_d, and spent Al₂O₃@ZnS-S_d. e Hg⁰ adsorption breakthrough curve of Al₂O₃@HgS-S_d assisted with intermittent S-CVD. Reaction conditions: adsorbent mass = 0.4 g, [Hg⁰] = (2.5 ± 0.05) mg m⁻³, [SO₂] = 5000 ppm, [H₂O] = 4%, [H₂S] = 100 ppm (30 min per 24 h), and total flow rate = 300 mL min⁻¹. f S 2p XPS spectra of Al₂O₃@HgS and Al₂O₃@HgS-S_d. g Linear fitting of the adsorption rates of Al₂O₃@ZnS-S_d and Al₂O₃@HgS-S_d at 120 °C. h Schematic illustration of in-situ intermittent S-CVD on Al₂O₃@MS surface for Hg⁰ multilayer adsorption. i DFT calculations of S_d⁰ activation and Hg⁰ self-sustained adsorption behaviors on ZnS (111) surface.

The kinetic simulation revealed that the Hg⁰ adsorption over Al₂O₃@ZnS-S_d closely followed the pseudo-first-order kinetic model (Supplementary Fig. 26), emphasizing the important role of the external surface area for Hg⁰ adsorption progress. Notably, after Hg⁰ adsorption, there brought out the crystal of β-HgS (JCPDS no. 75-1538), which characterized its main diffraction peaks at 26.4°, 30.6°, 43.8°, and 51.9°, in the XRD pattern of spent Al₂O₃@ZnS-S_d (Fig. 5a). The Raman spectrum of spent Al₂O₃@ZnS-S_d also verified the formation of Hg−S bonds that located at 249.6 cm⁻¹ and 343.6 cm⁻¹ (Fig. 5b)⁴⁵. Besides, the proportion of S_n^2– species in XPS spectra of spent Al₂O₃@ZnS-S_d decreased from 36.7% to 3.0%, while the proportion of S^2– increased from 35.4% to 82.9% (Fig. 5c, Supplementary Table 5). Moreover, the Hg 4f spectrum of spent Al₂O₃@ZnS-S_d exhibited the characteristic peaks of Hg²⁺ 4f_7/2 and 4f_5/2 centering at 100.3 eV and 104.3 eV, respectively (Supplementary Fig. 27). This signifies that adsorbed Hg⁰ can combine with the active S_d^* on Al₂O₃@ZnS-S_d to form HgS:

$${{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@ZnS}}}}}}{\mbox{-}}{{{{{{\rm{S}}}}}}}_{{{{{\rm{d}}}}}}^{\ast }+{{{{{{\rm{Hg}}}}}}}^{0}\to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@ZnS}}}}}}{\mbox{-}}{{{{{\rm{HgS}}}}}}$$

(5)

As the 24 h breakthrough rate increased and stabilized with increasing S-CVD rounds, and HgS crystal was observed in the spent Al₂O₃@ZnS-S_d, we suspect that the adsorbent surface was gradually covered by produced HgS, which may also possess the ability to activate S_d⁰ owing to the chalcophile nature of Hg element⁴⁶. Upon this, Al₂O₃@HgS was synthesized and used to testify the performance of Al₂O₃@HgS-S_d for Hg⁰ adsorption. As shown in Fig. 5e, Al₂O₃@HgS-S_d exhibited an initial Hg⁰ removal efficiency of ~96% and an 24 h breakthrough ratio of ~80% in each round of experiment, which is close to that of Al₂O₃@ZnS-S_d at the tenth round. As shown in Fig. 5f, compared with Al₂O₃@HgS, the XPS S 2p spectrum of Al₂O₃@HgS-S_d brought out new sulfur species located at 163.7 eV and 164.9 eV, which were between the binding energies of S_n^2– in Al₂O₃@ZnS-S_d and S₈⁰ in Al₂O₃-S_d. This indicates that the average chemical valence of S_d^* on Al₂O₃@HgS-S_d was slightly higher than that on Al₂O₃@ZnS-S_d, explaining the decrease in Hg⁰ adsorption on Al₂O₃@ZnS with increasing S-CVD rounds. The Hg 4 f spectra showed that the locations of Hg²⁺ in Al₂O₃@HgS-S_d shifted to lower binding energies compared to those of Al₂O₃@HgS (Supplementary Fig. 28). Moreover, the relationship between the Q_i of Al₂O₃@HgS-S_d and the Hg–S bond energy likewise fell within the negative correlation trend of chalcophile MS (Fig. 2c, Supplementary Fig. 29, Supplementary Table 2).

Notably, as the segmented linear fitting results in Fig. 5g present, with the reaction proceeded, the adsorption rate of Al₂O₃@ZnS-S_d at 120 °C decreased from 1.6 mg g^–1 h^–1 to 1.1 mg g^–1 h^–1, which converged to that of Al₂O₃@HgS-S_d (1.1 mg g^–1 h^–1). The adsorption rate of Al₂O₃@CuS-S_d at 60 °C decreased from 1.5 mg g^–1 h^–1 to 1.4 mg g^–1 h^–1, which was also converged to that of Al₂O₃@HgS-S_d (1.3 mg g^–1 h^–1) (Supplementary Fig. 30). This validates the gradual transition of S_d⁰ activation and Hg⁰ adsorption from the raw MS surface to the self-sustained HgS surface, in accordance with the concept presented in Fig. 1b. This verifies the promoting effect of MS on S_d⁰ is gradually replaced by HgS. Importantly, disregarding the effect of increasing adsorbent size caused by the formation of HgS layer, theoretically, the Hg⁰ adsorption capacity of Al₂O₃@ZnS-S_d can be continuously increased due to the self-sustained adsorption performance of HgS itself (Fig. 5h). Thus, according to the adsorption rate, the adsorption capacity of Al₂O₃@ZnS-S_d can achieve (54.2 + 25.1d) mg g⁻¹ (d > 6 days), thereby breaking the saturation limitations and realizing multilayer adsorption.

DFT calculations were applied to elaborate the Hg⁰ self-sustained adsorption mechanism on ZnS. ZnS (111) model was established and optimized to investigate the Gibbs free energy (ΔG) of S_d⁰ activation and the adsorption energy (E_ads) of Hg⁰. Given that S_n^2– dominated in the Hg⁰ adsorption process, the ΔG from S₈ ring to S_n (n = 6, 4, 2, and 1) chains were first calculated. As presents in Supplementary Fig. 31, the ZnS-S₈ can spontaneously convert to ZnS-S₆ and then to ZnS-S₄ with a negative ΔG of −77.9 kJ mol⁻¹, while the conversion of ZnS-S₄ to ZnS-S₂ has a positive ΔG of 158.6 kJ mol⁻¹. This indicates the most stable structure of ZnS-S₄. Moreover, compared with ZnS, the Zn–S bond length in ZnS-S₄ surface increased from 2.31 Å to 2.38–2.48 Å (Supplementary Fig. 32), in line with the decrease in Zn–S coordination number in ZnS-S_d. The E_ads of Hg⁰ adsorption on ZnS, ZnS-S₈, and ZnS-S₄ were calculated as –46.2, –67.6, and –125.0 kJ mol⁻¹, respectively (Fig. 5i). The highest E_ads of ZnS-S₄ verifies the important role of S₄ chain on Hg⁰ adsorption. Further, considering the formation of HgS on Al₂O₃@ZnS-S_d after self-sustaied adsorption of Hg⁰, we constructed the ZnS@HgS structure for subsequent S_d⁰ activation and Hg⁰ adsorption. The negative ΔG (–68.4 kJ mol⁻¹) from ZnS@HgS-S₈ to ZnS@HgS-S₄ demostrated its spontaneous convertion process. The E_ads of Hg⁰ on ZnS@HgS-S₄ (–73.1 kJ mol⁻¹) was higher than those of ZnS@HgS (–26.0 kJ mol⁻¹) and ZnS@HgS-S₈ (–27.9 kJ mol⁻¹). These confirm the role of HgS in the activation of S_d⁰ and further adsorption of Hg⁰, supporting the proposed Hg⁰ self-sustained adsorption mechanism.

Discussion

Figure 6a and Supplementary Table 7 compare the Hg⁰ adsorption capacities of Al₂O₃@ZnS-S_d with various reported adsorbents. The adsorption capacity of carbon- and oxide-based adsorbents were generally lower than 2 mg g^–1 and 10 mg g^–1, respectively, and their performance was severely inhibited by SO₂. Among the reported sulfide-based adsorbents, nano-CuS exhibited the highest capacity of 122.4 mg g^–1; however, it decreased to 89.4 mg g^–1 in the presence of SO₂ and H₂O (ref. ¹⁶). Additionally, our previous work found that under scaled-up conditions, ~1 mm Al₂O₃@CuS only had the normalized saturated adsorption capacity of 21.0 mg g^–1 (ref. ⁴¹). Another recently reported in-situ acid etching method can boost the adsorption capacity of ZnS to 53.8 mg g^–1 (ref. ¹⁹), but it decreased to 1.4 mg g^–1 under the scale-up conditions (Supplementary Fig. 33). Impressively, the self-sustained Al₂O₃@ZnS-S_d not only reversed the poisoning effect of SO₂ but also reached a Hg⁰ adsorption capacity of 303.9 mg g^–1 (normalized to ZnS coating amount, and with 24-h breakthrough ratio of ~80%) after 10 rounds of reaction, which is over 250, 60, and 8 times higher than the average of reported carbon-, oxide-, and sulfur-based adsorbents, respectively. Moreover, as the self-sustained adsorption process continues, the Hg⁰ adsorption capacity of Al₂O₃@ZnS-S_d will be increasing, thereby breaking the capacity limitations.

Fig. 6: Performance comparison and mechanism schematic.

a Comparison of Hg⁰ adsorption capacities between Al₂O₃@ZnS-S_d and other reported adsorbents. The raw data can be found in Supplementary Table 7. b Schematic illustration of the proposed in-situ S-CVD technology for self-sustained and multilayer adsorption of Hg⁰. S_d⁰ is deposited and activated to S_d^* on the Al₂O₃@MS by the reaction between flue gas SO₂ and intermittently added H₂S; when S_d^* is consumed by Hg⁰, it can be replenished by repeated S-CVD, resulting in self-sustained and multilayer adsorption of Hg⁰.

In summary, the Al₂O₃@MS-S_d adsorbents assisted with intermittent S-CVD can reverse SO₂ poisoning effects and have great potential for efficient and cost-effective Hg⁰ removal from SO₂-containing flue gases. This study demonstrates the crucial role of HgS, whether synthesized or formed by Hg⁰ adsorption on Al₂O₃@MS-S_d, in activating of S_d⁰ into S_d^* (Eqs. 6 and 7). Therefore, this enables the self-sustained adsorption of Hg⁰ on Al₂O₃@MS-S_d surface-like chain reactions (Eqs. 8–10), realizing the multilayer adsorption (Fig. 6b).

$${{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@HgS}}}}}}+{{{{{\rm{S}}}_{{{\rm{d}}}}^0}}}\to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@HgS}}}}}}{\mbox{-}}{{{{{{\rm{S}}}}}}}_{{{{{{\rm{d}}}}}}}^{\ast }$$

(6)

$${{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@MS}}}}}}{\mbox{-}}{{{{{\rm{HgS}}}}}}+{{{{{{\rm{S}}}}}}}_{{{{{{\rm{d}}}}}}}^{0}\to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}({{{{{\rm{@MS}}}}}}{\mbox{-}}{{{{{\rm{HgS}}}}}}){\mbox{-}}{{{{{{\rm{S}}}}}}}_{{{{{{\rm{d}}}}}}}^{\ast }$$

(7)

$${{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}({{{{{\rm{@MS}}}}}}{\mbox{-}}{{{{{\rm{HgS}}}}}}){\mbox{-}}{{{{{{\rm{S}}}}}}}_{{{{{{\rm{d}}}}}}}^{\ast }+{{{{{{\rm{Hg}}}}}}}^{0}\to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@MS}}}}}}{\mbox{-}}{({{{{{\rm{HgS}}}}}})}_{2}$$

(8)

$${{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@MS}}}}}}{\mbox{-}}{({{{{{\rm{HgS}}}}}})}_{2}+{{{{{{\rm{S}}}}}}}_{{{{{{\rm{d}}}}}}}^{0}\to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@MS}}}}}}{\mbox{-}}{({{{{{\rm{HgS}}}}}})}_{2}{\mbox{-}}{{{{{{\rm{S}}}}}}}_{{{{{{\rm{d}}}}}}}^{\ast }$$

(9)

$${{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@MS}}}}}}{\mbox{-}}{({{{{{\rm{HgS}}}}}})}_{{{{{{\rm{n}}}}}}{-}1}{\mbox{-}}{{{{{{\rm{S}}}}}}}_{{{{{{\rm{d}}}}}}}^{\ast }+{{{{{{\rm{Hg}}}}}}}^{0}\to {{{{{{\rm{Al}}}}}}}_{2}{{{{{{\rm{O}}}}}}}_{3}{{{{{\rm{@MS}}}}}}{\mbox{-}}{({{{{{\rm{HgS}}}}}})}_{{{{{{\rm{n}}}}}}}$$

(10)

Methods

Materials

The raw materials used in this work were purchased in chemical purity (>99.5%) from Sinopharm Chemical Reagent Co., Ltd and used without purification. Natural sulfide ores (chalcopyrite and sphalerite) were provided by provided by a non-ferrous smelter in Henan, China.

Preparation of adsorbents

Different metal sulfides (MS, M = Cu, Zn, Cd, In, Hg, Pb, Sn, Mn, Fe, Co, and Ni) were synthesized via a one-step hydrothermal method as the interface for S-CVD and Hg⁰ adsorption. According to our previous study, ~1 mm commercial γ-Al₂O₃ pellets were chosen as the support for MS (named Al₂O₃@MS) to improve gas mass transfer⁴¹. Taking Al₂O₃@ZnS as an example, firstly, 3.50 mL of H₂O containing 2.50 mmol of ZnSO₄·7H₂O was added into 4.75 g of γ-Al₂O₃ pellets followed by 30 min ultrasonic treatment. Then, the mixture was dried at 60 °C for 9 h to obtain Al₂O₃@ZnSO₄. After that, the pellets were poured into 50.0 mL of H₂O containing 2.50 mmol of Na₂S·9H₂O and transferred to a 100 mL Teflon-lined autoclave. After reacting at 120 °C for 12 h, the resultant Al₂O₃@ZnS was collected by vacuum filtration, washed with deionized water and ethanol, and dried at 60 °C in an oven for 12 h. Other Al₂O₃@MS adsorbents were synthesized in the same way, except that different metal precursors, including CuSO₄·5H₂O (2.50 mmol), CdCl₂ (2.50 mmol), InCl₃·4H₂O (1.67 mmol), HgCl₂ (2.50 mmol), (CH₃COO)₂Pb (2.50 mmol), SnCl₂ (2.50 mmol), MnSO₄·H₂O (2.50 mmol), FeCl₃ (1.67 mmol), CoCl₂·6H₂O (2.50 mmol), and NiCl₂·6H₂O (2.50 mmol), were used instead of ZnSO₄·7H₂O.

In-situ low-temperature sulfur chemical vapor deposition

The in-situ S-CVD was conducted in a self-made fixed-bed reactor system as shown in Supplementary Fig. 1. To prevent the pre-reaction between H₂S and SO₂ in the pipeline (heat protection at 120 °C), a three-way quartz reaction tube was used to introduce different gas components, of which the main tube (10.0 mm inner diameter) was used to feed 5000 ppm SO₂, and the side tube (4.0 mm inner diameter) embedded in the main reaction tube was used to feed 100 ppm H₂S. The H₂S concentration was detected by a gas chromatograph (Agilent GC 8860) equipped with a flame photometric detector. To verify the feasibility of the S-CVD strategy, we used SiO₂ wafer as substrate to conduct the reaction. The SEM images of SiO₂-S_d exhibited a homogeneous surface with a few grooves and the ESD mapping images showed the uniform distribution of S element (Supplementary Fig. 34a–f). Moreover, the SEM image of the cross-section of SiO₂-S_d showed the formation of a deposition layer (Supplementary Fig. 34g, h).

Specifically, in this work, single S-CVD (15 min) was used to evaluate the enhanced effect of S-CVD on the activities of different Al₂O₃@MS adsorbents, while Al₂O₃@MS or natural sulfide ores treated with intermittent S-CVD (30 min each round) were applied to assess their self-sustained adsorption performance for Hg⁰.

Characterizations

The chemical composition of the as-prepared adsorbents was analyzed by X-ray fluorescence spectroscopy on an Epsilon 3X instrument (Netherlands). The thermal stability of the adsorbents was analyzed by a thermogravimetric analyzer (NETZSCH, STA 2500 Regulus) at a heating rate of 5 °C min⁻¹ from 20 °C to 500 °C in N₂ atmosphere. The BET-specific surface area and pore structure were detected by an automatic porosity analyzer apparatus (Quantachrome, Autosorb-iQ, USA). The XRD patterns were obtained on Shimadzu XRD-6100 (Japan) with Cu Kα radiation (scan speed: 8° min⁻¹, and 2θ range of 10−80°). The SEM images of adsorbents were performed on ZEISS Sigma 300 (Germany), and EDS/mapping images using Oxford Xplore 30 (UK). XPS spectra were determined on an X-ray photoelectron spectrometer (Thermo Scientific ESCALAB Xi+, USA) equipped with a mono Al Kα X-ray source. TEM images were carried out on FEI Tecnai F20. The XPS depth profiling is measured by Ar⁺ ion etching material surface layers at different depths. XAFS spectra of S L-edge and Zn K-edge were detected by synchrotron radiation light sources (see detail in Supplementary Methods 1). Raman spectra were measured on a Raman spectrometer (Horiba LabRAM HR Evolution, Japan) with 532 nm line of Ar⁺ laser for excitation. FTIR spectra were recorded on a Fourier-Transform infrared spectrometer (Nicolet 6700, USA) using the potassium bromide pellet technique. ¹³C NMR spectra were performed on a nuclear magnetic resonance spectrometer (Bruker Avance Neo 400WB, Germany).

Gaseous mercury adsorption assessment

Hg⁰ adsorption performances of as-prepared adsorbents were assessed in the same reactor system. The Hg⁰ vapor was obtained by an Hg⁰ permeation device (VICI Metronics) loaded in a U-shaped glass tube. The H₂O vapor was produced by a steam generator. The concentrations of both Hg⁰ and H₂O were controlled by adjusting the water bath temperature and carrier gas (N₂) flow rate. Other gas component including O₂, SO₂, and NO, were obtained directly from compressed gas in cylinders. The gas components of the simulated flue gas, including Hg⁰ (1.5−2.5 mg m⁻³), SO₂ (5000−60,000 ppm, when used), and H₂O (4%, when used), O₂ (5%, when used), and NO (100 ppm, when used) were mixed evenly before entering the fixed-bed reactor. The total flow rate was controlled at 300−360 mL min⁻¹, the adsorbent mass used in each experiment was 0.3−1.0 g. The detailed experiment conditions were summarized in Supplementary Table 8. Lumex RA 915+ was used to record the inlet and outlet Hg⁰ concentrations and before each experiment, the inlet concentration should remain stable (±0.05 μg m⁻³) for more than 10 min. The exhaust gas was absorbed by 0.1 mol L⁻¹ potassium permanganate solution and activated carbon before discharged.

The Hg⁰ removal efficiency (η, %) and Hg⁰ adsorption capacity (Q, mg·g⁻¹) were calculated according to following equations:

$${{{{{\rm{\eta }}}}}}=\frac{{{{\mbox{Hg}}}}_{{{\mbox{in}}}}^{0}{{{-}}}{{{\mbox{Hg}}}}_{{{\mbox{out}}}}^{0}}{{{{\mbox{Hg}}}}_{{{\mbox{in}}}}^{0}}\times 100\%$$

(11)

$${{\mbox{Q}}}={\int }_{{{{\mbox{t}}}}_{1}}^{{{{\mbox{t}}}}_{2}}\frac{{{{\mbox{Hg}}}}_{{{\mbox{in}}}}^{0}{{{{-}}}{{\mbox{Hg}}}}_{{{\mbox{out}}}}^{0}}{{{\mbox{m}}}}\times {{\mbox{fdt}}}$$

(12)

where Hg⁰_in and Hg⁰_out (mg m⁻³) are the inlet and outlet Hg⁰ concentrations, respectively, m (g) is the adsorbent mass, f (mL min⁻¹) represents the total flow rate, and t (min) donates the adsorption time.

Hg⁰-TPD experiments were conducted to identify the Hg⁰ species adsorbed on the adsorbent surface and its stability. A certain amount of spent adsorbent was heated from 50 to 450 °C with a heating rate of 5 °C min⁻¹ in pure N₂ to desorb the adsorbed mercury. The signal of desorbed Hg⁰ was also detected by Lumex RA 915+.

Kinetics and theory calculation

Hg⁰ adsorption kinetic models and theoretical calculations of S_d⁰ activation and Hg⁰ adsorption behavior are detailed in Supplementary Methods 2 and 3.