Elemental mercury removal from coal gas by CeMnTi sorbents and their regeneration performance*

Hui CAO, Jin-song ZHOU, Qi-xin ZHOU, Xin-yu XU, Cong XIE

Elemental mercury removal from coal gas by CeMnTi sorbents and their regeneration performance*

Hui CAO, Jin-song ZHOU†‡, Qi-xin ZHOU, Xin-yu XU, Cong XIE

†E-mail: zhoujs@zju.edu.cn

Ce and Mn modified TiO2sorbents (CeMnTi) were prepared by a co-precipitation method, and their ability to remove elemental mercury from coal gas in a fixed bed reactor was studied. Based on results of Brunauer-Emmett-Teller (BET), X-ray diffraction (XRD), scanning electron microscope (SEM), and X-ray photoelectron spectroscopy (XPS) studies, the modification mechanisms of the CeMnTi sorbents are discussed. Mn doping improved the specific surface area and dispersion of cerium oxides on the sorbent surface, while Ce doping increased the proportion of Mn4+in manganese oxides by a synergetic effect between manganese oxides and cerium oxides. The effects of the active component, temperature, and coal gas components on the mercury removal performance of the sorbents were investigated. The results showed that the CeMnTi sorbents exhibited high mercury removal efficiency. Ce0.2Mn0.1Ti adsorbed 91.55% elemental mercury from coal gas at 160 °C. H2S and O2significantly improved the ability of sorbents to remove mercury. Part of the H2S formed stable sulfates or sulfites through a series of oxidation reaction chains on the sorbent surface. HCl also improved the mercury removal performance, but reduced the promotion effect of H2S for mercury removal when coexisting with H2S. CO and H2had a minor inhibitory effect on mercury adsorption. The recycling performance of the sorbents was investigated by thermal regeneration. The thermal decomposition of the used sorbents indicated that mercury compounds were present mainly in the form of HgO and HgS, and higher temperature was beneficial for regeneration. The formation of sulfates and sulfites in the presence of H2S led to a decrease in mercury removal efficiency.

CeMnTi; Synergistic effect; Gas components; Mercury removal; Thermal regeneration

1 Introduction

Mercury is a highly toxic and volatile heavy metal with stable chemical properties, and accumulates in organisms and food chains for long durations. These properties have led to it being listed as a high priority for pollutant control. The main source of mercury pollution in the air is the combustion of fossil fuels, especially coal, natural gas and oil. According to International Energy Agency (IEA) projections (IEA, 2019), fossil fuels will continue to dominate global energy consumption in the next decade. Therefore, the harmful effects of mercury on humans and the environment will receive increasing attention by countries around the world.

As a promising clean coal utilization technology, coal gasification is characterized by high efficiency and cleanliness. It forms the basis of coal-based chemical synthesis, liquid fuel synthesis, oxygen production, poly generation systems, and other important process industries (Wang et al., 2009).Coal gasification transforms macromolecular coal into micromolecular combustible gases such as CO, H2, and CH4, which have a higher calorific value and cleaner combustion process than coal combustion. However, during the coal gasification process, the mercury contained in the coal is inevitably released into the coal gas. According to the literature(Lu et al., 2004), the mercury content in coal gas is much higher than that in flue gas. Moreover, the mercury in coal gas exists mainly in the form of elemental mercury, which is more difficult to remove than other forms of mercury. Currently, oxidation of Hg0to Hg2+is the most common and effective method for chemical removal. Nevertheless, it is more difficult to remove elemental mercury from coal gas than from flue gas owing to its reducing behavior with CO, H2, H2S, and other reducing gases. A number of studies have focused mainly on the emission and control of mercury in coal-fired flue gas. Industrial applications involving injection of activated carbon for flue gas purification are common, but almost all of the sorbents have a very low physical adsorption performance for Hg0. Therefore, research is focusing increasingly on the development of modified methods such as chemical adsorption to improve the performance of adsorbents for Hg0in flue gas (Liu and Adewuyi, 2016). For example, SO2, NO, NO2, the chlorine-based radicals in flue gas, and ozone strongly promote Hg0removal. Therefore, the use of these gases or the injection of corresponding substances can achieve efficient mercury removal or coordinated removal (Yuan et al., 2020). However, less attention has been paid to mercury removal from coal gas. Thus, with the rapid development of coal gasification technology, it is particularly important to strengthen basic research on mercury pollution control in the process of coal gasification, and to explore the mechanisms of mercury removal by sorbents in coal gas.

Sorbents with manganese oxides as their active component have been proved to exhibit an optimal mercury removal performance. Thus, the excellent catalytic performance of manganese oxides has been widely used for mercury removal from flue gas (Ji et al., 2008; Li et al., 2010; Yu et al., 2015). CeO2has been also widely studied and applied in the field of catalysis due to its strong oxygen storage capacity and superior catalytic performance (Reddy et al., 2003; Zhou et al., 2013). It has been observed to form Ce4+/Ce3+redox couples in the redox process, which have a good ability to store oxygen in oxygen deficient atmospheres and to release oxygen in oxygen excess atmospheres (Matsumoto, 2004). The combination of CeO2with other oxides such as manganese oxides may affect the surface structure of sorbents and the conversion of surface oxygen, thereby contributing to the adsorption and oxidation of mercury (Zhuang et al., 2012; Zhang HW et al., 2016; Li et al., 2018; Zhang SB et al., 2018).However, there has been no systematic study of the mechanism of mercury removal from coal gas by manganese cerium composite metal oxide sorbents.

Therefore, using TiO2with a defined pore structure and high specific surface area as carrier, sorbents have been modified by doping with Ce and Mn to enhance their adsorption capacity for mercury. In this study, a series of composite CeMnTi sorbents with different manganese and cerium contents were prepared by a co-precipitation method. Brunauer-Emmett-Teller (BET), X-ray diffraction (XRD), scanning electron microscope (SEM), X-ray photoelectron spectroscopy (XPS), and temperature-programmed reduction (TPR) were applied to characterize the structures and physicochemical properties of the sorbents, and the modification mechanisms were investigated. The efficiency of mercury removal from coal gas and the thermal regeneration performance of the sorbents were studied. The mechanisms of Hg0adsorption at different temperatures and in different coal gas atmospheres were also analyzed.

2 Experimental

2.1 Preparation of sorbents

The composite metal oxide sorbents were labeled CeMnTi, where Ce represents CeO2and Ce2O3, Mn represents MnO2and Mn2O3, Ti represents TiO2,represents the CeO2/TiO2mass ratio (=0, 0.2, which refers to the optimal ratio of CeTi sorbents for mercury removal from coal gas according to our previous research (Zhou et al., 2013)), andrepresents the MnO2/TiO2mass ratio (=0, 0.05, 0.1, 0.15, and 0.2). CeMnTi sorbents were prepared using a co-precipitation method. First, to attain different mass ratios, certain amounts of Ce(NO3)3·6H2O (analytical reagent (AR)), Mn(NO3)2·6H2O (AR), and nano-TiO2(Degussa P25, AR, Germany) were weighed and dissolved in deionized water by stirring. Subsequently, K2CO3solution (0.1 g/mL) was added to adjust the pH to 8–9 under stirring, followed by precipitation and ageing. Next, the solution was suction filtered at room temperature and washed to neutrality. Finally, the precipitate was collected and dried at 110 °C for 8 h. It was calcined at 500 °C for 2 h, followed by grinding and sieving to 60–80 meshes.

2.2 Mercury removal

The prepared sorbent was placed in a fixed-bed reactor (Fig. 1) for the adsorption experiments. As shown, the experimental system comprises mainly a simulated gas generation system, a mercury vapor generator, a fixed bed reactor, and an online mercury measurement system. About 50 mg of the sorbent was placed in a quartz tube reactor (10 mm in diameter) situated in a vertical tube furnace. The flow rate of the mercury vapor carrier N2and the temperature of the mercury vapor generator were adjusted such that the initial mercury concentration was stabilized at 55–60 μg/m3. The total gas flow was maintained and controlled at 1.2 L/min by the mass flow controller (MFC), and the reaction time was 120 min. Adsorption temperatures of 30, 80, 120, 160, and 200 °C were used. The simulated atmosphere included N2, H2S, CO, H2, and HCl. The mercury removal performance of the sorbents can be defined by the mercury removal efficiency:

Fig. 1 Fixed-bed experimental device

2.3 Thermal desorption and regeneration

The thermal desorption and regeneration of the sorbents were also studied in the fixed-bed reactor. The specific process was as follows: the used sorbent was placed in the quartz tube reactor. The carrier gas was adjusted to be pure N2, with the flow rate controlled to 1 L/min. The system was purged at room temperature for 30 min to remove the unstable mercury adsorbed on the sorbent surface, as well as the residual mercury vapor in the reactor. The carrier gas was adjusted again, and temperature programmed desorption regeneration was initiated.

2.4 Sorbent characterization

The characteristics of the sorbents and the modification mechanisms of Ce and Mn were studied by a series of characterization methods using fresh and used samples.

The specific surface area and pore parameters, which are key factors affecting the mercury removal performance of sorbents, were tested using an ASAP 2460 surface area and porosity analyzer (Micromeritics, the USA). The sorbent samples were degassed at 250 °C under vacuum for 4 h and subsequently subjected to N2adsorption at the liquid nitrogen temperature (196.15 °C). The specific surface area was calculated using the BET method, and the average pore diameter and total pore volume were calculated by the Barrett-Joyner-Halenda (BJH) method. XRD patterns were collected on a BRUKER D8 Advance X-ray diffractometer (Brook, Germany) to explore the atomic and molecular morphology of the sorbents. The target element and wavelength were Cu K-(=0.154 nm), the tube voltage was 40 kV, and the tube current was 40 mA. The scanning range was 10°–80°, and the step size was 0.01°. SEM analysis was carried out using a Zeiss Sigma 500 microscope (Carl Zeiss, Germany) to gain insights about the surface morphology and structure of the samples. XPS was carried out using a Thermo Fisher Scientific K-spectrometer (Thermo Fisher, the USA) to test the surface element valence and composition. The radiation source used in the XPS study was Al K-(=1486.8 eV, whereis the Plank constant andis the frequency) with a power of 150 W. The transmission energy was fixed using the beam spot energy analyzer. The charge effect correction of the sample surface was calibrated by the C 1s standard binding energy value of 284.8 eV, and curve fitting was performed using the peaking software XPS Peak 4.1. A TPR test was performed using a ChemBET Pulsar analyzer (Quantachrome, the USA). The gas was 10% CO/He (CO-TPR) and 10% H2/He (H2-TPR). The gas flow rate was 30 cm3/min. The heating rate was 10 °C/min, and the heating range was 50–900 °C.

3 Results and discussion

3.1 Characterization

3.1.1 BET analysis

The BET surface area, average pore diameter, and total pore volume of the pure TiO2and CeMnTi sorbents are shown in Table 1. Most of the modified sorbents had larger specific surface areas and pore parameters than pure TiO2, which meant that more active sites were available on the sorbents. The average pore diameter of nano-TiO2was 21.17 nm. The Ce and Mn improved the pore structure so that the pore sizes of the modified sorbents increased. The average pore diameters of all sorbents were in the range of mesopore, which facilitated the adsorption for macromolecular compounds such as HgO and HgS. In addition, the mesoporous structures could promote the mass transfer process in the catalytic oxidation reaction(Yu et al., 2015).Although micropores have a significant influence on the adsorption of mercury (the particle size of Hg is 2.6 μm), the physical adsorption of mercury molecules by micropores may not be as good as chemical binding by mesopores at high temperature.

Note that the pore parameters of Ce0.2Ti and Ce0.2Mn0.05Ti were improved while their specific surface areas decreased due to sintering and agglomeration of partial metal oxides. Compared with the CeTi sorbent, the BET results of CeMnTi sorbents showed a slight improvement, which can be attributed to an increase in the dispersion of cerium oxides on the sorbent surface due to Mn doping. Also, the pore diameter decreased as the mass ratio of Mn exceeded 0.1, which might be related to the partial blockage of micropores due to excessive metal oxide loading (Shen et al., 2012). The BET surface area, average pore diameter, and total pore volume of Ce0.2Mn0.1Ti reached 58.45 m2/g, 31.08 nm, and 0.4792 cm3/g, respectively. Overall, its textural properties were observed to be the best among the developed sorbents.

Table 1 BET surface area and pore parameters of sorbents

3.1.2 XRD analysis

The XRD diffractograms are presented in Fig. 2. The XRD patterns of the sorbents were similar due to the obvious TiO2characteristic diffraction peaks, present mainly in the anatase form in all sorbents. Weak CeO2characteristic peaks (228.5°, 33.1°, 47.5°, 56.6°, and 59.19°) in CeTi sorbent could be observed. The characteristic peaks of MnO2(2=28.6°, 37.3°, 42.7°, and 56.8°) were not significant in MnTi or CeMnTi sorbents, which may be due to the lower crystallinity of the manganese oxides on the sorbent surface (Wu et al., 2019). In addition, the intensity of the characteristic peak of CeO2decreased after loading Mn among CeTi and CeMnTi sorbents. This can be explained by monolayer dispersion theory(Mullins et al., 1998), which indicates that the active components have a high degree of dispersion on the surface of TiO2. The metal oxides on the sorbent surface interact during the doping process to produce more amorphous metal oxides, which have higher activity than the crystalline metal oxides (Qi and Yang, 2004; Gao et al., 2010).

3.1.3 SEM analysis

SEM micrographs of the sorbents are shown in Fig. 3. The particles of the Ce0.2Ti and Ce0.2Mn0.05Ti sorbents with a higher content of cerium oxides were uneven, and a fraction of oxides on the surface exhibited agglomeration and sintering. Also, Fig. 3a shows crystalline particles that were identified as metal oxides on the sorbent surface. In contrast, the Mn0.2Ti sorbent loaded with manganese oxides had a uniform surface and effective dispersion. The surface conditions of the sorbents carrying both cerium and manganese oxides were intermediate. On increasing the manganese content, the surface structure of the sorbents showed a tendency for densification, followed by loosening, indicating that the addition of manganese can increase the dispersion of CeO2on the TiO2support. This was consistent with the findings from BET and XRD.

Fig. 2 XRD patterns of the sorbents

Fig. 3 SEM micrographs of the samples

(a) Ce0.2Ti; (b) Ce0.2Mn0.05Ti; (c) Ce0.2Mn0.1Ti; (d) Ce0.2Mn0.15Ti; (e) Ce0.2Mn0.2Ti; (f) Mn0.2Ti

3.1.4 XPS analysis

The XPS spectra of Ce 3d, Mn 2p, and O 1s are shown in Fig. 4. The atomic ratios of Ce, Mn, and O on the sorbent surface were calculated (Table 2). Fig. 4a shows the XPS spectrum of Ce 3d. The spectrum can be classified into two types of peaks (Reddy and Khan, 2005): Ce 3d3/2and Ce 3d5/2, denoted byand, respectively. Among these, the0/0,2/2,3/3pairs of peaks represent Ce4+, and1/1indicates Ce3+. Two forms of Ce4+and Ce3+existed in the valence state of Ce on the sorbent surface. During the chemisorption process, the Ce4+/Ce3+electron pairs could participate in mercury removal by producing oxygen vacancies. Hence, the Ce4+/(Ce4++Ce3+) ratio was calculated to quantify the change in the Ce4+/Ce3+electron pairs. The Ce4+/(Ce4++Ce3+) ratios for fresh Ce0.2Ti and Ce0.2Mn0.1Ti sorbents were 0.8634 and 0.8598, respectively (Table 2), which indicated that CeO2was the main cerium oxide, and the effect of the introduction of manganese on the Ce4+/Ce3+electron pairs was not obvious. After mercury adsorption, the Ce4+/(Ce4++Ce3+) ratio decreased to 0.6355, demonstrating that part of the CeO2was transformed to Ce2O3during the mercury oxidation process.

Fig. 4b shows the Mn 2p XPS spectrum of the CeMnTi sorbent. The spectrum was fitted to two peaks (Zeng et al., 2017): Mn4+with a binding energy of 642.93 eV, and Mn3+with a binding energy of 641.69 eV. No peak of Mn2+could be observed; thus, MnO2and Mn2O3were regarded as the main manganese oxides on the sorbent surface. As higher valence manganese ions have a stronger ability to oxidize elemental mercury (Jampaiah et al., 2016), the Mn4+/(Mn4++Mn3+) ratio is important for mercury removal efficiency. The Mn4+/(Mn4++Mn3+) ratio was 0.3944 for fresh Mn0.2Ti, and 0.4564 for the fresh Ce0.2Mn0.1Ti sorbent, indicating that the addition of Ce increased the proportion of Mn4+on the sorbent surface, thereby improving the oxidation performance of manganese. The Mn4+/(Mn4++Mn3+) ratio for Ce0.2Mn0.1Ti changed from 0.4564 to 0.3291 during the mercury adsorption process, which may be attributed to the reduction of MnO2to Mn2O3for oxidizing mercury. Ding et al. (1998) proposed a synergistic mechanism by the reaction chain between CeO2and MnO2, which can be written as

where “[O]” represents the active oxygen.

Fig. 4 Ce 3d XPS spectrum (a), Mn 2p XPS spectrum (b), and O 1s XPS spectrum (c) of the CeMnTi sorbents

Fig. 4c shows the O 1s XPS spectrum of Ce0.2Mn0.1Ti sorbent. The O 1s peaks can be split into two peaks:OAwith a binding energy range of 531.3–531.9 eV, represents the chemisorbed oxygen, while OBwith a binding energy range of 529.3–530.0 eV, represents the lattice oxygen (Zhang et al., 2014). The OA/(OA+OB) ratio was 0.1810 before adsorption, and 0.4203 after adsorption. Compared with the fresh sorbent, the lattice oxygen of the Ce0.2Mn0.1Ti sorbent was consumed significantly in the reaction process.

Table 2 Atomic ratios of Ce, Mn, and O to the total concentration of sorbents

3.2 Effect of temperature

Adsorption temperature plays an important role in the removal of mercury from coal gas. The mercury removal efficiency of the sorbents increased initially, then decreased with increasing temperature in the experimental range (30–200 °C) (Fig. 5). This indicates that the sorbents exhibited two adsorption pathways, physical adsorption and chemisorption, to remove mercury from the coal gas. When the temperature was low, physical adsorption was dominant and the adsorption efficiency was positively related to the specific surface area. At higher temperatures, the physical adsorption weakened, and chemical adsorption became dominant due to the higher probability of molecular activity and collision. However, when the temperature exceeded 160 °C, the decomposition of HgO and HgS and the sublimation of active sulfur would be increased (You et al., 2014), leading to a decrease in mercury removal efficiency. Nevertheless, more than 83% Hg0removal efficiency could be obtained over Ce0.2Mn0.1Ti at 200 °C. The adsorption efficiency of the CeMnTi sorbent in the temperature range of 80–160 °C remained stable. In particular, Ce0.2Mn0.1Ti achieved the optimal physical and chemical adsorption performance, reaching anvalue of 91.55% at 160 °C.

Fig. 5 Effect of temperature and active components on the mercury removal efficiency of the sorbents

3.3 Effect of active components

The adsorption performance of TiO2as support was not effective, while the mercury removal efficiency of Ce0.2Ti significantly increased (Fig. 5). Compared with the single cerium loading sorbent Ce0.2Ti, the mercury removal efficiency of the composite metal oxide sorbents CeMnTi was also significantly improved. The Ce0.2Ti sorbent was noted to be less efficient than other sorbents owing to sintering agglomeration on the sorbent surface to form CeO2crystals, observed from the SEM and XRD patterns, which reduced the activity of the sorbent. On doping with Mn, the adsorption efficiency of the sorbents increased. However, when the Mn content was too high, the efficiency decreased to a certain extent, which may be related to the specific surface area of the sorbents and the dispersion of the active components. The mercury removal efficiency was consistent with the trend in the BET and SEM results, so it was inferred that the difference may be caused by the physical adsorption capacity of sorbents. Also, the Ce0.2Mn0.1Ti and Ce0.2Mn0.15Ti sorbents exhibited a higher mercury removal efficiency than Ce0.2Ti and Mn0.2Ti, which was attributed to the interaction between Ce and Mn (Eqs. (2) and (3)).

For the CeMnTi sorbents, the values ofwere stable,and Ce0.2Mn0.1Ti and Ce0.2Mn0.15Ti reached more than 85% in the temperature range of 80–160 °C. Based on the observed results, the Ce0.2Mn0.1Ti sorbent and an adsorption temperature of 160 °C were selected for further experiments.

3.4 Effect of coal gas components

3.4.1 Effect of H2S

The mercury removal efficiency of the Ce0.2Mn0.1Ti sorbent under pure N2conditions was 46.29% (Fig. 6). On introducing 80 ppm (1 ppm=1×10−6) H2S, the mercury removal performance was obviously improved to 65.47%. On continuously increasing the concentration of H2S to 400 ppm, the mercury removal efficiency rose to 91.55%. Obviously, the presence of H2S contributed to the removal of Hg0. The oxidation process of Hg0on the sorbent surface follows the Mars-Maessen mechanism (Granite et al., 2000), i.e. the gas phase Hg0is first adsorbed on the sorbent surface, followed by oxidization with the active oxygen in the sorbent to form HgO:

where “(ad)” represents the adsorbed state.

As a reducing gas, H2S was not able to oxidize Hg0by itself. In the presence of the Ce0.2Mn0.1Ti sorbent, the oxygen in the sorbent reacted with H2S to promote further oxidation of Hg0. As mentioned earlier, the presence of the Ce4+/Ce3+electron pair provides a large number of oxygen vacancies and reactive oxygen for the H2S promotion process, which is:

A small amount of light yellow matter was found at the end of the quartz tube reactor on introducing 400 ppm H2S. The melting point of elemental sulfur is about 126 °C and the decomposition temperature of HgS was relatively high (Lu et al., 2012). Therefore, this might be attributed to the condensation of partially sublimated active sulfur (Eq. (6)), which suggests that 400 ppm H2S was excessive for mercury removal.To verify the morphology of elemental sulfur species after mercury and H2S adsorption, XPS was applied to characterize the used samples. The peaks at 162.44 eV and 163.68 eV (Fig. 7) were attributed to S2−and S, respectively (Yang et al., 2018), which proved the generation of active sulfur and HgS (Eqs. (6) and (7)). In addition, the peaks at 167.14 eV and 168.71 eV were assigned to sulfite and sulfate, respectively (Andreu et al., 2015), indicating the formation of metal sulfites and sulfates. The metal sulfites and sulfates, such as manganese sulfates and cerium sulfates, are stable during regeneration (Cheah et al., 2009), resulting in a loss of active material. Therefore, future studies could further explore the chemical reaction chains and morphology of elemental sulfur over the CeMnTi sorbents to avoid the formation of sulfites and sulfates.

Fig. 6 Effect of coal gas components on the mercury removal efficiency of Ce0.2Mn0.1Ti at 160 °C

Fig. 7 S 2p XPS spectrum of the used Ce0.2Mn0.1Ti sorbent

3.4.2 Effect of CO and H2

CO and H2are the main components in coal gas. Thus, it is also necessary to study the effect of CO and H2on mercury removal by CeMnTi sorbents. Specifically, the effect of CO and H2on the mercury removal of the sorbents was investigated under pure N2and N2+H2S conditions. On introducing 20% CO or 30% H2into pure N2, the values ofdecreased from 46.29% to 16.08% and 34.51%,respectively. As strong reducing agents, CO and H2can reduce active components on the sorbent surface. However, as shown in Fig. 8, there was no significant difference between CO and H2when the temperature was in the range of 0–200 °C. Therefore, the reason for the stronger inhibition by CO might be its stronger competitiveness for the active adsorption sites.

On introducing H2S,CO and H2had no obvious influence on Hg0capture, which demonstrated that H2S played a key role in mercury removal from the coal gas. In the presence of H2S and CO, the value ofwas lower than that in N2with H2S. This was because CO reacted with H2S and active sulfur to form COS (Eq. (8)), which seized the active adsorption sites, thereby, inhibiting the adsorption of mercury (Wu et al., 2007). For H2, it was easily adsorbed on the sorbent surface and protonated to H+(Zhang et al., 2012), which consumed the active oxygen and suppressed the generation of active sulfur. Furthermore, Kobayashi and Flytzani-Stephanopoulos (2002) proposed that H2could partially consume active sulfur by reacting with H2S and CeO2(Eq. (9)), which was also not conductive to mercury removal.

Nevertheless, the coexistence of CO and H2promoted Hg0adsorption compared with CO. This may have been due to the increase in H2concentration, which hindered reaction Eq. (8), thereby reducing the consumption of active sulfur.

3.4.3 Effect of HCl

HCl is considered to be one of the most critical components affecting mercury oxidation in flue gas (Zheng et al., 2012). Therefore, the effect of HCl on the mercury removal of the CeMnTi sorbent in coal gas was also studied in this work. The mercury removal efficiency of the CeMnTi sorbent increased from 46.29% to 68.17% as 10 ppm HCl was introduced into pure N2(Fig. 6) indicating that HCl could promote the removal of mercury. HCl could react with the active oxygen to form active chlorine, followed by further chemical reaction with the adsorbed mercury (Li et al., 2011). However, in the presence of H2S, the introduction of HCl caused a decrease in mercury removal efficiency, which might have been due to competition between HCl and H2S. HCl and H2S may have competed with elemental mercury for the active adsorption sites, resulting in a reduced amount of mercury adsorbed on the surface (Hou et al., 2014). Alternatively, HCl may have consumed a part of the active oxygen to weaken the promotion of mercury removal by H2S.

3.4.4 Effect of O2

He et al. (2014) found that the gas phase O2in flue gas could supply the lattice oxygen consumed in the reaction process. Therefore, the effect of O2on mercury removal performance was tested. The introduction of oxygen facilitated the removal of mercury (Fig. 6). When 5% O2was introduced, the mercury removal efficiency of the sorbent increased to 53.25%. When the concentration of O2was raised to 10%, the mercury removal efficiency increased further to 69.80%. Thus, under aerobic conditions, the presence of oxygen was conducive to the adsorption and oxidation of mercury.

3.5 Thermal desorption and regeneration

The regeneration performance of the sorbent has an important influence on reducing the operational cost and environmental risks (Dong et al., 2009). Thus, the regeneration performance of the CeMnTi sorbent was investigated using the thermal desorption regeneration method.

To further explore the mercury desorption performance of the used Ce0.2Mn0.1Ti sorbents (the sorbents adsorbed mercury at 160 °C, N2+400 ppm H2S for 2 h), a series of temperature-programmed desorption experiments were carried out. The mercury desorption curves of 500 °C+N2, 500 °C+ 5% O2+N2, and 400 °C+5% O2+N2using a heating rate of 10 °C/min were collected (Fig. 9). The desorption pathway of mercury was observed to be as follows: mercury started to release from the sorbents at 180–200 °C, which suggests that it was chemisorbed on the used Ce0.2Mn0.1Ti surface. On increasing the heating temperature, the main desorption peak of mercury occurred at around 300 °C. In particular, the secondary peak was observed at about 500 °C. Note that during the desorption process, only a small amount of Hg2+was detected at the beginning of the mercury desorption phase. In other words, mercury was largely desorbed in the form of elemental mercury, indicating that the mercury compounds bound to the sorbent surface were decomposed into elemental mercury, followed by removal with other gases by the carrier gas.

Fig. 9 Mercury desorption curves for the Ce0.2Mn0.1Ti sorbent under different temperatures and atmospheric conditions. Reaction conditions: (a) 500 °C+N2; (b) 500 °C+5% O2+N2; (c) 400 °C+5% O2+N2

After studying the mercury desorption, the mercury adsorption performance of the sorbents was re-tested (Fig. 10). The regeneration temperature had a more significant effect on the mercury removal efficiency of the CeMnTi sorbent than the introduction of 5% O2. The mercury removal efficiency of the fresh Ce0.2Mn0.1Ti sorbent was 91.55%, while it was reduced to 72.83% after regeneration at 500 °C under N2with 5% O2. This shows that the higher regeneration temperature could desorb most of mercury compounds and restore the occupied active sites. After the introduction of 5% O2, the mercury removal efficiency improved slightly (from 70.25% to 72.83%). This may have been due to the O2molecules oxidizing the low valence oxides on the sorbent surface, thereby supplementing the active oxygen required for the adsorption reaction. Zeng et al. (1999) conducted research with modified ceria. Their ceria regeneration scheme consisted of:

A synergistic effect of Mn and Ce existing in the modification (Eq. (15)) also worked during the regeneration process.

Regeneration under different oxygen concentrations was studied (Fig. 11). The oxygen concentration had a limited effect on regeneration efficiency when it exceeded 5%. O2can oxidize low-priced metal oxides to restore the partly adsorbed active sites (Zhu et al., 2018). However, Table 3 shows that the sorbent provided a large amount of active oxygen and consumed less active oxygen in mercury removal. Further, the regeneration (under 5% O2) after the mercury removal reaction in pure N2(without H2S) was studied. The decrease in mercury removal efficiency was reduced compared with mercury removal in H2S (Fig. 12). Thus, we conclude that the smaller O2effect, as suggested earlier, may be due to the formation of some metal sulfates and sulfites, resulting in the loss of active components and hindering the decomposition of mercury compounds.

Fig. 10 Mercury removal efficiency after regeneration under different temperatures and atmospheric conditions. Mercury adsorption conditions: N2 with 400 ppm H2S, about 55 μg/m3 Hg0, 160 °C

Fig. 11 Regeneration under different oxygen concentrations

Fig. 12 Regeneration under different adsorption atmospheres

4 Conclusions

CeMnTi sorbents were prepared by a co-precipitation method, and their performance and mechanisms involved in removing mercury from coal gas were studied. Several meaningful observations were made in this study:

1. The Ce0.2Mn0.1Ti sorbent displayed higher mercury removal efficiency and thermal stability than the CeTi and MnTi sorbents. Its mercury removal efficiency reached 91.55% at 160 °C and remained above 83% at 200 °C.

2. Mn doping improves the BET surface area and dispersion of cerium oxides on the sorbent surface, while Ce doping could increase the proportion of Mn4+by the interaction with manganese oxides.

3. H2S significantly promoted the adsorption of mercury by generating active sulfur on the sorbent surface. However, H2S could react with active components to form sulfates and sulfites, making it difficult for sorbent regeneration. The presence of CO and H2inhibited mercury removal. HCl improved mercury removal performance, but had the opposite effect when combined with H2S due to competition for mercury.

4. The mercury compounds on the sorbent existed mainly in the form of HgO and HgS, and most could be desorbed at 500 °C. The main reason for a decrease in mercury removal efficiency may be the formation of sulfates and sulfites in the presence of H2S.

本研究提示，鼻塞、鼻部胀痛、烦躁不安、溢泪、鼻痒、打喷嚏、头面部疼痛及压迫感、睡眠困难、夜间睡眠质量等9项症状不填塞组较填塞组明显减轻。鼻内镜术后不填塞高膨胀海绵组的患者舒适度较好，同时不需要抽取填塞物，无明显疼痛及鼻腔再次出血。鼻内镜术后不填塞高膨胀海绵不但提高患者术后短时间内生活质量，而且无需取填塞物，24h可出院，可缩短患者的住院时间，减轻患者的经济负担。

Contributors

Hui CAO wrote the manuscript. Jin-song ZHOU and Qi-xin ZHOU provided the suggestions. Qi-xin ZHOU, Xin-yu XU, and Cong XIE helped to organize the manuscript.

Conflict of interest

Hui CAO, Jin-song ZHOU, Qi-xin ZHOU, Xin-yu XU, and Cong XIE declare that they have no conflict of interest.

Andreu N, Flahaut D, Dedryvère R, et al., 2015. XPS investigation of surface reactivity of electrode materials: effect of the transition metal., 7(12):6629-6636. https://doi.org/10.1021/am5089764

Cheah S, Carpenter DL, Magrini-Bair KA, 2009. Review of mid- to high-temperature sulfur sorbents for desulfurization of biomass- and coal-derived syngas., 23(11):5291-5307. https://doi.org/10.1021/ef900714q

Ding ZY, Li LX, Wade D, et al., 1998. Supercritical water oxidation of NH3over a MnO2/CeO2catalyst., 37(5):1707-1716. https://doi.org/10.1021/ie9709345

Dong J, Xu ZH, Kuznicki SM, 2009. Mercury removal from flue gases by novel regenerable magnetic nanocomposite sorbents., 43(9): 3266-3271.https://doi.org/10.1021/es803306n

Gao X, Jiang Y, Zhong Y, et al., 2010. The activity and characterization of CeO2-TiO2catalysts prepared by the sol-gel method for selective catalytic reduction of NO with NH3., 174(1-3): 734-739.https://doi.org/10.1016/j.jhazmat.2009.09.112

Granite EJ, Pennline HW, Hargis RA, 2000. Novel sorbents for mercury removal from flue gas., 39(4):1020-1029. https://doi.org/10.1021/ie990758v

He C, Shen BX, Chen JH, et al., 2014. Adsorption and oxidation of elemental mercury over Ce-MnO/Ti-PILCs.,48(14):7891-7898. https://doi.org/10.1021/es5007719

Hou WH, Zhou JS, Qi P, et al., 2014. Effect of H2S/HCl on the removal of elemental mercury in syngas over CeO2-TiO2., 241:131-137. https://doi.org/10.1016/j.cej.2013.12.047

IEA (International Energy Agency), 2019. World Energy Outlook 2019.https://www.iea.org/reports/world-energy-outlook-2019

Jampaiah D, Ippolito SJ, Sabri YM, et al., 2016. Ceria-zirconia modified MnOcatalysts for gaseous elemental mercury oxidation and adsorption., 6(6):1792-1803. https://doi.org/10.1039/c5cy01534k

Ji L, Sreekanth PM, Smirniotis PG, et al., 2008. Manganese oxide/titania materials for removal of NOand elemental mercury from flue gas., 22(4):2299-2306. https://doi.org/10.1021/ef700533q

Kobayashi M, Flytzani-Stephanopoulos M, 2002. Reduction and sulfidation kinetics of cerium oxide and Cu-modified cerium oxide., 41(13):3115-3123. https://doi.org/10.1021/ie010815w

Li HL, Wu CY, Li Y, et al., 2011. CeO2-TiO2catalysts for catalytic oxidation of elemental mercury in low-rank coal combustion flue gas., 45(17):7394-7400. https://doi.org/10.1021/es2007808

Li JF, Yan NQ, Qu Z, et al, 2010. Catalytic oxidation of elemental mercury over the modified catalyst Mn/α-Al2O3at lower temperatures., 44(1):426-431. https://doi.org/10.1021/es9021206

Li XQ, Zhou JS, Zhou QX, et al., 2018. Removal of elemental mercury using titania sorbents loaded with cobalt ceria oxides from syngas., 42(15): 12503-12510.https://doi.org/10.1039/C8NJ02069H

Liu YX, Adewuyi YG, 2016. A review on removal of elemental mercury from flue gas using advanced oxidation process: chemistry and process.,112:199-250. https://doi.org/10.1016/j.cherd.2016.06.024

Lu DY, Granatstein DL, Rose DJ, 2004. Study of mercury speciation from simulated coal gasification., 43(17):5400-5404. https://doi.org/10.1021/ie034121u

Lu H, Greenwood P, Chen TS, et al., 2012. The separate production of H2S from the thermal reaction of hydrocarbons with magnesium sulfate and sulfur: implications for thermal sulfate reduction., 27(1): 96-105. https://doi.org/10.1016/j.apgeochem.2011.09.007

Matsumoto S, 2004. Recent advances in automobile exhaust catalysts., 90(3-4):183-190. https://doi.org/10.1016/j.cattod.2004.04.048

Mullins DR, Overbury SH, Huntley DR, 1998. Electron spectroscopy of single crystal and polycrystalline cerium oxide surfaces., 409(2):307-319. https://doi.org/10.1016/S0039-6028(98)00257-X

Qi GS, Yang RT, 2004. Characterization and FTIR studies of MnO-CeO2catalyst for low-temperature selective catalytic reduction of NO with NH3., 108(40):15738-15747. https://doi.org/10.1021/jp048431h

Rallo M, Fuente-Cuesta A, Lopez-Anton MA, et al., 2014. Speciation of Hg retained in gasification biomass chars by temperature-programmed decomposition., 126:1-4. https://doi.org/10.1016/j.fuproc.2014.04.010

Reddy BM, Khan A, 2005. Nanosized CeO2-SiO2, CeO2-TiO2, and CeO2-ZrO2mixed oxides: influence of supporting oxide on thermal stability and oxygen storage properties of ceria., 9(3):155-171. https://doi.org/10.1007/s10563-005-7552-1

Reddy BM, Khan A, Yamada Y, et al., 2003. Structural characterization of CeO2-TiO2and V2O5/CeO2-TiO2catalysts by Raman and XPS techniques., 107(22):5162-5167. https://doi.org/10.1021/jp0344601

Rumayor M, Fernandez-Miranda N, Lopez-Anton MA, et al., 2015. Application of mercury temperature programmed desorption (HgTPD) to ascertain mercury/char interactions., 132:9-14.https://doi.org/10.1016/j.fuproc.2014.12.032

Shen BX, Ma HQ, Yao Y, 2012. Mn-CeO/Ti-PILCs for selective catalytic reduction of NO with NH3at low temperature., 24(3): 499-506. https://doi.org/10.1016/S1001-0742(11)60756-0

Wang FC, Yu GS, Gong X, et al, 2009.Research and development of large-scale coal gasification technology, 28(2):173- 180. https://doi.org/10.3321/j.issn:1000-6613.2009.02.001

Wu SJ, Oya N, Ozaki M, et al., 2007. Development of iron oxide sorbents for Hg0removal from coal derived fuel gas: sulfidation characteristics of iron oxide sorbents and activity for COS formation during Hg0removal., 86(17-18):2857-2863. https://doi.org/10.1016/j.fuel.2007.03.004

Wu X, Duan YF, Yao T, et al., 2019. Mercury removal performance and SO2resistance of Ce-Mn/TiO2sorbent., 39(6):2336-2343. https://doi.org/10.19674/j.cnki.issn1000-6923.2019.0278

Yang ZQ, Li HL, Liao C, et al., 2018. Magnetic rattle-type Fe3O4@CuS nanoparticles as recyclable sorbents for mercury capture from coal combustion flue gas., 1(9):4726-4736. https://doi.org/10.1021/acsanm.8b00948

You SL, Zhou JS, Hou WH, et al., 2014. Factors influencing the removal of elemental mercury by Mn-AC sorbent in syngas., 42(11):1324-1331. https://doi.org/10.3969/j.issn.0253-2409.2014.11.008

Yu XQ, Bao JJ, Jiang XX, et al., 2015. Performance and mechanism of catalytic oxidation for mercury by Mn-doped TiO2catalysts in flue gas., 35(13):3331-3337. https://doi.org/10.13334/j.0258-8013.pcsee.2015.13.016

Yuan B, Mao XZ, Wang Z, et al, 2020. Radical-induced oxidation removal of multi-air-pollutant: a critical review., 383:121162. https://doi.org/10.1016/j.jhazmat.2019.121162

Zeng XB, Xu Y, Zhang B, et al., 2017. Elemental mercury adsorption and regeneration performance of sorbents FeMnOenhanced via non-thermal plasma., 309:503-512. https://doi.org/10.1016/j.cej.2016.10.047

Zeng Y, Zhang S, Groves FR, et al., 1999. High temperature gas desulfurization with elemental sulfur production., 54(15-16):3007-3017. https://doi.org/10.1016/S0009-2509(98)00427-8

Zhang AC, Zheng WW, Song J, et al., 2014. Cobalt manganese oxides modified titania catalysts for oxidation of elemental mercury at low flue gas temperature., 236:29-38. https://doi.org/10.1016/j.cej.2013.09.060

Zhang H, Zhao JT, Fang YT, et al., 2012. Catalytic oxidation and stabilized adsorption of elemental mercury from coal-derived fuel gas., 26(3):1629-1637. https://doi.org/10.1021/ef201453d

Zhang HW, Chen JY, Zhao K, et al., 2016. Removal of vapor-phase elemental mercury from simulated syngas using semi-coke modified by Mn/Ce doping., 44(4):394-400. https://doi.org/10.1016/S1872-5813(16)30020-2

Zhang SB, Zhao YC, Díaz-Somoano M, et al., 2018. Synergistic mercury removal over the CeMnO3perovskite structure oxide as a selective catalytic reduction catalyst from coal combustion flue gas., 32(11): 11785-11795. https://doi.org/10.1021/acs.energyfuels.8b02518

Zheng JM, Zhou JS, Luo ZY, et al., 2012. Impact of individual acid flue gas components on mercury capture by heat-treated activated carbon.,13(9):700- 708.https://doi.org/10.1631/jzus.A1200112

Zhou JS, Hou WH, Qi P, et al., 2013. CeO2-TiO2sorbents for the removal of elemental mercury from syngas., 47(17):10056-10062. https://doi.org/10.1021/es401681y

Zhu YC, Han XJ, Huang ZG, et al., 2018. Superior activity of CeO2modified V2O5/AC catalyst for mercury removal at low temperature., 337: 741-749. https://doi.org/10.1016/j.cej.2017.10.115

Zhuang K, Qiu J, Xu BL, et al., 2012. Promotional effect of cerium oxide on the catalytic properties of Ce-Mn-Ti-O catalysts for selective catalytic reduction of NO., 28(3):681-685. https://doi.org/10.3866/PKU.WHXB201111141

Journal of Zhejiang University-SCIENCE A (Applied Physics & Engineering)

ISSN 1673-565X (Print); ISSN 1862-1775 (Online)

www.jzus.zju.edu.cn; link.springer.com

E-mail: jzus_a@zju.edu.cn

ORCID: Hui CAO, https://orcid.org/0000-0002-4253-150X; Jin-song ZHOU, https://orcid.org/0000-0003-2704-0419

Feb. 29, 2020;

Revision accepted Aug. 21, 2020;

Crosschecked Feb. 24, 2021

https://doi.org/10.1631/jzus.A2000079

TK16; X701

*Project supported by the National Natural Science Foundation of China (No. 51576173)