A Molten-Salt Pyrolysis Synthesis Strategy toward Sulfur-Functionalized Carbon for Elemental Mercury Removal from Coal-Combustion Flue Gas

Abstract

The emission of mercury from coal combustion has caused consequential hazards to the ecosystem. The key challenge to abating the mercury emission is to explore highly efficient adsorbents. Herein, sulfur-functionalized carbon (S-C) was synthesized by using a molten-salt pyrolysis strategy and employed for the removal of elemental mercury from coal-combustion flue gas. An ideal pore structure, which was favorable for the internal diffusion of the Hg⁰ molecule in carbon, was obtained by using a SiO₂ hard template and adjusting the HF etching time. The as-prepared S-C with an HF etching time of 10 h possessed a saturation Hg⁰ adsorption capacity of 89.90 mg·g⁻¹, far exceeding that of the commercial sulfur-loaded activated carbons (S/C). The S-C can be applied at a wide temperature range of 25–125 °C, far exceeding that of commercial S/C. The influence of flue gas components, such as SO₂, NO, and H₂O, on the Hg⁰ adsorption performance of S-C was insignificant, indicating a good applicability in real-world applications. The mechanism of the Hg⁰ removal by S-C was proposed, i.e., the reduced components, including sulfur thiophene, sulfoxide, and C-S, displayed a high affinity toward Hg⁰, which could guarantee the stable immobilization of Hg⁰ as HgS in the adsorbent. Thus, the molten-salt pyrolysis strategy has a broad prospect in the application of one-pot carbonization and functionalization sulfur-containing organic precursors as efficient adsorbents for Hg⁰.

1. Introduction

The excessive emission of mercury from industrial activities has caused consequential hazards to the ecosystem and human health [1,2,3]. Coal combustion is one of the largest industrial sources of mercury emission. The mercury emitted from typical coal-combustion flue gas generally existed in three forms, i.e., elemental mercury (Hg⁰), oxidized mercury (Hg²⁺), and particulate bound mercury (Hg^p) [4,5,6,7,8]. The Hg^p can be captured by using particulate matter control devices, while the Hg²⁺ can be removed by using wet flue-gas scrubbers due to Hg²⁺’s water solubility [9,10,11,12]. However, Hg⁰ is difficult to remove owing to its water insolubility and high volatility [13,14,15]. As a consequence, Hg⁰ is the primary species of mercury discharged into the atmosphere from coal-combustion flue gases. The highly efficient removal of Hg⁰ is a key challenge to reducing mercury pollution from coal-fired power plants.

Activated carbons (ACs) are the most widely researched and skilled mercury adsorbents for coal-fired power plants [16,17,18,19,20,21]. However, ACs were generally limited by adsorption kinetics and equilibrium capacities, hence causing large consumptions of ACs during Hg⁰ removal [22,23]. Moreover, the weak binding affinity of ACs toward mercury induced leaching risks of mercury when the ACs were dumped in landfills together with fly ashes [24,25]. Thus, active components that can accommodate Hg⁰ were generally introduced to the ACs to improve the Hg⁰ adsorption capacity. In nature, the mercury was tied to sulfides owing to its sulphophile affinity [26,27]. Inspired by this natural law, sulfur was widely employed to modify the Hg⁰ adsorption capacity of ACs [28,29]. This strategy was relatively simple, cheap, and effective, but was still found to suffer from drawbacks [30,31]. The most significant one is that the sulfur was not actually anchored to the carbon surface during the impregnation process, hence affecting the activity, abundance, and accessibility of the sulfur groups during Hg⁰ removals [32]. The impregnation method for introducing sulfur would plug the pores of ACs. As a result, the efficient diffusion of Hg⁰ on the sulfur-modified ACs would be limited [33]. In addition, the anchoring of sulfur on ACs was not firm, hence causing leaching risks of sulfur as well as mercury adsorption products when ACs were disposed of in landfills [32,34]. Therefore, a new strategy is urgently undergoing exploration to overcome the disadvantages associated with sulfur-impregnated ACs [35,36].

Traditionally, the sulfur-functionalized carbon materials were prepared by using two steps, i.e., the carbonization of natural products (e.g., cellulose, chitin, starch, alginic acid, and chitosan) as well as some synthetic polymers (e.g., poly-acrylonitrile, polyaniline, and phenolic resins) first and then introducing sulfur onto carbons by using an impregnation method [37,38,39,40]. A one-pot carbonization and functionalization step, which could guarantee the uniform distribution and firm fixation of sulfur functionality on the carbon matrix, was required. Very recently, a simple approach via the carbonization of small organic molecules with the assistance of transition metals was reported to prepare a series of functional carbon materials [37]. This approach was realized via the pyrolysis of a mixture of small organic molecules and transition-metal salts in a conventional tubular furnace, hence avoiding the barriers to large-scale production, such as complicated equipment and harsh conditions. The salts would act as a heat-transfer medium and provide an oxygen-free environment for pyrolysis [41]. Moreover, the salts catalyze the formation of a thermally stable intermediate polymerization structure, avoiding the direct sublimation of small organic molecules during heating [37,38]. More attractive, template-like SiO₂ could be mixed into the precursors and hence adopted to adjust the porosity of carbons. This strategy is commonly available and easy to use to control the surface sulfur functionalities, porosities, and morphologies of carbons.

In this work, porous sulfur-functionalized carbons were prepared and employed for Hg⁰ removal from coal-fired flue gas. The removal performance of sulfur-functionalized carbon (S-C) on Hg⁰ was studied and compared with that of commercial sulfur-loaded ACs (S/C). The Hg⁰ removal performances of S-C and S/C under various adsorption temperatures and flue gas conditions were studied. The excellent adsorption mechanism of Hg⁰ by S-C was further investigated. This work not only provides a promising trap for highly efficient Hg⁰ sequestration from coal-fired power plants but also illustrates a versatile platform for preparing functional carbon materials by using a one-pot carbonization and functionalization organic precursor.

2. Experimental Section

2.1. Sample Preparation

S-C was prepared by using a molten-salt pyrolysis strategy [38]. Two grams of 2,2-bithiophene and 2.0 g of SiO₂ were added into 150 mL of tetrahydrofuran and stirred at room temperature for 6 h. The mixture was then dried and then ground to a powder. After that, the powder was placed into a rail boat and covered with 5.0 g of Co(NO₃)₂·6H₂O, which was heated at 800 °C for 2 h. Then, the SiO₂ was removed by using 10% HF solution. To obtain different porosity of S-C, the sample was etched by using HF for different times (0, 24, 48 and 56 h). After drying at 105 °C for 10 h, the S-C was finally obtained.

2.2. Sample Characterization

The morphology of the sample was studied by using a scanning electronic microscope (SEM, FEI F50, New York, NY, USA). Transmission electron microscope (TEM, EOL JEM 2100F, microscope Tokyo, Japan) and high-resolution TEM (HRTEM) were used to study the morphology and structure of sample. The valence states of samples were characterized by using X-ray photoelectron spectroscopy (XPS, Thermo ESCALAB 250Xi, New York, NY, USA).

2.3. Hg⁰ Adsorption Activity Tests

The Hg⁰ adsorption activity test was measured on a fixed-bed reactor [42,43]. The adsorbent was placed in a quartz reactor, reaction temperature of which was controlled by using a tube furnace. The flue gas was composed of N₂, O₂, SO₂, NO, and H₂O, with the total flow rate of 1 L·min⁻¹. Hg⁰ was provided by a mercury permeation tube placed in a constant temperature water bath, delivering Hg⁰ by using N₂ to ensure a stable concentration of 65 µg·m⁻³. The concentration of Hg⁰ was monitored by using an online mercury analyzer (RA-915M, Lumex, Tianjin, China). The Hg⁰ adsorption capability was calculated by using the following equations:

$$Q=\frac{1}{m}{{\displaystyle \int }}_{t_1}^{t_2}(C_{in}-C_{out})\times f\times dt$$

(1)

where C_in (μg·m⁻³) and C_out (μg·m⁻³) represent the inlet and outlet concentrations of Hg⁰, Q (μg·g⁻¹) is the Hg⁰ adsorption capacity, m (g) is the sample amount, f (m³·h⁻¹) is the gas flow rate, and t (h) is the reaction time.

3. Description of Sorption Kinetic Models

3.1. Pseudo-First-Order Model

This model is described as follows [44]:

$$\frac{d{q}_t}{dt}=k_1(q_e-q_t)$$

(2)

Based on the initial conditions, i.e., t = 0 q_t = 0, Equation (2) is revised as:

$$q_t=q_e(1-e^{-k_{1}t})$$

(3)

where q_t and q_e represent the adsorbed mercury amount at any time t and equilibrium time (µg·g⁻¹). k₁ represents the rate constant (min⁻¹).

3.2. Pseudo-Second-Order Model

This model is described as follows [45]:

$$\frac{d{q}_t}{dt}=k_2{(q_e-q_t)}^2$$

(4)

On the basis of initial conditions, i.e., t = 0 and q_t = 0, Equation (4) is modified as:

$$q_t=\frac{t}{\frac{1}{k_{{}^2}{q}_e^2}+\frac{1}{q_e^{}}t}$$

(5)

where k₂ represents the rate constant (µg/(cm³·min)).

3.3. Elovich Model

This model is described as follows [46]:

$$\frac{d{q}_t}{dt}=\alpha \exp (-\beta q_t)$$

(6)

where α represents the initial rate and β is related to surface coverage and activation energy. If t is much larger than t₀, this equation is modified as follows:

$$q_t=\frac{1}{\beta }\ln \left(\alpha \beta \right)+(\frac{1}{\beta })\ln t$$

(7)

3.4. Intra-Particle Diffusion Model

This model is described by the following formula [47]:

$$q_t=k_{id}{t}^{0.5}+C$$

(8)

where k_id represents the diffusion rate constant within the particle and C is a constant, which is related to the boundary layer.

4. Results and Discussion

4.1. Preparation and Characterization of Samples

The S-C was prepared by using the molten Co(NO₃)₂·6H₂O-assisted carbonization of sulfur-containing small organic molecule precursors. SiO₂ nanoparticles were adopted as hard templates, followed by an HF etching step to remove the SiO₂ template. The porosity of S-C was adjusted by using the HF etching time (shown in Figure 1). As listed in

Table 1. Pore structure parameters of various S-Cs with different HF etching times.

HF Etching Time (h)	Total Surface Area (m2·g−1)	Surface Area of Micropores (m2·g−1)	Pore Volume (cm3·g−1)	Pore Diameter (nm)
0	180.92	143.33	0.29	6.32
5	267.66	263.68	0.50	7.43
10	318.50	335.88	0.58	7.31

, the HF etching step will generate mesopores on the S-C. The S-C with HF etching for 10 h possessed the largest BET surface area of 318.5 m²·g⁻¹, with an average pore size of 7.31 nm and total pore volume of 0.58 cm³·g⁻¹. Pore size affects the internal mass transfer process of Hg⁰ on adsorbents, thus affecting the removal rate of Hg⁰. As shown in Figure 2a, the N₂ absorption–desorption isotherms for S-C belonged to a typical II isotherm, indicating that the porosity of S-C was very limited. The S-C after the HF etching step presented a typical IV isotherm with a negligible absorption at lower pressures but significant absorption at higher pressures (p/p⁰ = 0.2-1.0). This indicates that S-C is composed of mesopores or macropores rather than micropores. The pore distribution curves shown in Figure 2b demonstrate that the pore size of the S-C was in the range of 2–10 nm. The mesoporous structure in the range of 2–50 nm is beneficial for Hg⁰’s diffusion to the inner surface of an adsorbent [48]. Thus, the pore structure of S-C is favorable for Hg⁰ adsorption.

Figure 3 shows the morphologies of S-Cs with different HF etching times. These images display the high-solution local microscopies of individual particles, in which the sizes of individual particles are in the range of 100–200 mesh. As shown, the morphology of S-C was significant dependent on the HF etching time. The S-C without HF etching displayed a dense surface, and there were negligible pores observed on the S-C surface. After etching with HF, abundant pores were generated on the S-C surface, and more pores were generated with the extension of the HF etching time from 5 to 10 h. Thus, the SiO₂ was embedded in the carbon structure as a template, which could be removed by using the HF solution to induce the formation of a porous structure. However, upon reaching a higher HF etching time, the resultant carbon framework collapsed, which might have affected the internal diffusion of mercury. Therefore, a SiO₂ template was used to achieve the adjustable preparation of porous carbons.

The surface-functional groups of S-C were studied by using FTIR. As shown in Figure 4, there were four main peaks on the FTIR spectra, which could be assigned to the C-OH (3392 and 615 cm⁻¹) and aromatic C = C (1626 and 1100 cm⁻¹) [49]. Figure 5 shows the XPS spectra for S-C. As shown in Figure 5a, the spectra for high-solution C 1s spectra could be divided into three peaks at 284.8, 286.0, and 288.8 eV, assigned to the characteristic peaks of C = C, C-O, and C = O [30]. Figure 5b shows the O 1s spectra for C-S, in which the peaks at 531.5, 532.3, and 533.3 eV are regarded as C-OH, C-O, and O-H [31]. Figure 5c shows the S spectra for S-C. The sulfur on the S-C existed in four forms. The two peaks at 164.2 and 165.3 eV corresponded to S 2p3/2 and S 2p1/2 for thiophenic sulfur (i.e., -C-S_x-C-, x = 1-2), the small peak at 169.1 eV could be assigned to sulfoxide [50], the peak at 166.2 eV could be assigned to C-S, while the peaks at 167.2 and 170.3 eV could be assigned to sulfate [51,52].

4.2. Hg⁰ Adsorption Capacity Tests

Figure 6 shows the Hg⁰ adsorption performances of S-Cs with various HF etching times. As shown, the normalized-outlet Hg⁰ concentration of S-C without etching climbed dramatically to above 0.90. Thus, the S-C without etching has a poor Hg⁰ adsorption performance, and the removal rate of Hg⁰ is only below 10%. However, after the HF etching, the S-C displayed far superior Hg⁰ removal performances. The HF etching times played significant roles in the Hg⁰ adsorption of S-C. After etching for 5 h, the concentration of Hg⁰ at the normalized outlet remained below 0.2. With the extension of the etching time to 10 h, the Hg⁰ adsorption performance of S-C can be further improved. However, a too-long etching time resulted in a decrease in Hg⁰ adsorption performance. This is attributed to the fact that HF etching can remove the SiO₂ template to generate abundant pores on S-C, which allows for accessible mercury molecule transportation. As a result, the sulfur in S-C can be accessible sufficiently for binding mercury. However, the excessive etching will result in the collapse of the framework of carbons, hence the weakening of the Hg⁰ adsorption on S-C.

The saturated adsorption capacity of an adsorbent is an important index to use to evaluate the adsorption performances of materials, so it is necessary to study the change in S-C adsorption capacity with time. The curve of Hg⁰ adsorption capacity changing over time is shown in Figure 7. When the reaction duration was 1200 min, the adsorption capacity of Hg⁰ exceeded 50% and reached 45.39 mg·g⁻¹. In addition, it can be seen from the figure that the adsorption rate increases first and then decreases with time, and the slope is zero when the adsorption is saturated. Adsorption kinetic models can be adopted to investigate the process of Hg⁰ adsorption and the dominant controlling factors. The pseudo-first-order-based kinetic model is the most common adsorption kinetic model. Ln(q_e-q_t) is used to plot t, and the adsorption mechanism conforms to the pseudo-first-order model if a straight line can be obtained. The pseudo-second-order model is based on the assumption that the adsorption rate is controlled by the chemisorption mechanism. Elovich describes a series of reaction mechanism processes, which are suitable for processes with large activation energy changes in reactions and for complex heterogeneous diffusion processes. The intra-particle diffusion model is commonly used to analyze the control steps in the reaction. Generally, the material adsorption process is divided into two processes: adsorbent surface adsorption and slow pore diffusion. If the fitting result fails to reach the origin, it indicates that the internal diffusion of the material is not the only step to controlling the adsorption process. As shown in Figure 8, the pseudo-first-order model was closest to the adsorption process of Hg⁰ on S-C, with an extremely high correlation coefficient (R² = 0.9982). The saturation adsorption capacity of S-C was simulated as 89.90 mg·g⁻¹. When the molar ratio of Hg:S is 1:1, it is equivalent to 55% of the sulfur accessibility in Hg⁰ adsorption. For comparison, the Hg⁰ saturation adsorption capacity of a sulfur-loaded commercial activated carbon (S/C) was also investigated. It should be noted that the S/C possessed a higher sulfur content and surface area compared with S-C. However, the saturation adsorption capacity of S/C was less than 1 mg·g⁻¹, which was much lower compared with its theoretical adsorption capacity of 352.1 mg·g⁻¹. Thus, most of the sulfur in S/C was not adopted sufficiently for binding mercury, although it possessed a higher surface area. It fully shows that the key to Hg⁰ adsorption improvements lies in the high dispersion of active sulfur species [53]. Thus, the molten-salt pyrolysis synthesis strategy toward sulfur-functionalized carbon would be more superior compared with other traditional methods, such as impregnation when significantly dispersing the active components (i.e., sulfur).

4.3. Impact of Operation Conditions on Hg⁰ Adsorption Capacity

An excellent adsorbent should have a good stability at various conditions. Figure 9 shows the influence of temperature on the Hg⁰ adsorption capacities of S-C and S/C. The normalized-outlet Hg⁰ concentration when passing through the S-C was maintained below 0.05 in a wide reaction temperature range of 25–125 °C. This suggests that an above 95% Hg⁰ adsorption efficiency was obtained; even the sorbent dosage was as low as 5 mg. This wide temperature range proves that the S-C can be applied flexibly at different scenes for Hg⁰ removal. In contrast, the S/C exhibited significantly different Hg⁰ adsorption capacities under various temperatures, in which relatively limited Hg⁰ adsorption capacities were obtained at low temperatures. Even at the optimum reaction temperature and the same amount of adsorbent, the concentration of Hg⁰ at the normalized outlet is still higher than 0.15, much higher than S-C. Depending on the preferred temperature range, setting the S/C upstream of the ESP is the ideal application in which the Hg⁰ adsorption capacity of S-C might be affected by various flue gas components, such as high concentrations of fly ash, SO₂, and NO₂.

According to previous studies, SO₂ and H₂O generally compete with Hg⁰ for adsorption sites, hence exhibiting inhibitive effects on the Hg⁰ removal over carbonaceous sorbents [54,55]. However, the S-C exhibited an excellent resistance to these detrimental flue gas components. Figure 10a–c show that, in the presence of SO₂, H₂O, as well as NO, the Hg⁰ removal performance was very similar to that under a pure N₂ atmosphere. Even adding 1200 ppm of SO₂, 12% H₂O, or 400 ppm of NO, the normalized-outlet Hg⁰ concentration was kept at less than 0.05. As a direct comparison, the adsorption performance of S/C for Hg⁰ was investigated. As shown in Figure 10d, the SO₂, H₂O, and NO significantly weakened the adsorption capacity of S/C. These results could fully demonstrate the excellent resistance of S-C to the detrimental impacts of flue gas impurities compared with commercial activated carbon, which would facilitate the real-world applications.

4.4. Reaction Mechanism

To study the mechanism of a Hg⁰ removal on S-C, the valence states of elements on spent S-C after adsorbing mercury were investigated by using XPS. Figure 11a shows that there is no significant change in the spectral binding energy of C 1s for the spent adsorbent, indicating that C did not participate in the Hg⁰ adsorption [56]. Figure 11b shows the O 1s spectra for the spent adsorbent. The spectra of O 1s can be divided into four separate corresponding peaks: C = O, C-O-C, COOH, and H₂O at 531.3, 532.5, 533.8, and 536.1 eV, respectively [57]. After the Hg⁰ adsorption, part of the C-O groups changed to C = O, and H₂O was generated on the adsorbent surface, which may have been caused by a charge imbalance after binding Hg⁰. Figure 11c shows the S 2p for the spent adsorbent, including only two peaks: C-S at 163.94 eV and thiophene at 165.22 eV [58]. The transformations of sulfur species on the spent adsorbent indicate that sulfur played a crucial role in the Hg⁰ removal. The absolute content of thiophene and C-S decreased after adsorbing Hg⁰, especially the peaks for sulfoxide, which disappeared compared with the fresh adsorbent. This variation indicates that thiophene (i.e., -C-S_x-C-, x = 1-2), sulfoxide, and C-S were beneficial to improving the Hg⁰ removal capacity. This is in line with a previous study [59], i.e., findings that the sulfur in sulfide existing in a low state as well as the reducing sulfur (such as thiosulfone) had a high affinity with the Hg⁰ atom. Electrons around the sulfur atom were easy to combine with Hg⁰, which was conducive to the removal of mercury. Figure 11d shows that the two peaks of Hg 4f corresponded to the peaks of 103 and 107 eV of HgS [60], further demonstrating that the sulfur in the adsorbent was active for binding Hg⁰.

5. Conclusions

Sulfur-functionalized carbon (S-C) derived from a molten-salt pyrolysis strategy was employed for a Hg⁰ removal. Abundant pores could be generated by using a SiO₂ hard template and a subsequent HF etching step. The HF etching time significantly affected the Hg⁰ removal performances of S-Cs. With an HF etching time of 10 h, the S-C presented the best Hg⁰ removal performance, the saturation Hg⁰-adoption capacity of which reached 89.90 mg·g⁻¹. This value far exceeded that of the commercial sulfur-loaded activated carbon (S/C), which was specialized for Hg⁰ removals. The S-C displayed a good applicability at 25–125 °C, while the S/C could be adopted around only 125 °C since the Hg⁰ adsorption performances decreased when deviating from this temperature. The SO₂, NO, and H₂O had negligible adverse effects on Hg⁰ adsorption over S-C. The good Hg⁰ removal performance of S-C was ascribed to the existence of reduced sulfurs, such as thiophene, sulfoxide, and C-S, which have high affinities toward Hg⁰, and the fact that gaseous Hg⁰ was converted into HgS after the adsorption. These results indicate that molten-salt pyrolysis can simultaneously achieve the carbonization and functionalization of sulfur-containing organic precursors and is an ideal method for preparing carbonaceous adsorbents for Hg⁰ removal.