Promoting Effect of the Core-Shell Structure of MnO 2 @TiO 2 Nanorods on SO 2 Resistance in Hg 0 Removal Process

: Sorbent of αMnO 2 nanorods coating TiO 2 shell (denoted as αMnO 2 -NR@TiO 2 ) was prepared to investigate the elemental mercury (Hg 0 ) removal performance in the presence of SO 2 . Due the core-shell structure, αMnO 2 -NR@TiO 2 has a better SO 2 resistance when compared to αMnO 2 nanorods (denoted as αMnO 2 -NR). Kinetic studies have shown that both the sorption rates of αMnO 2 -NR and αMnO 2 -NR@TiO 2 , which can be described by pseudo second-order models and SO 2 treatment, did not change the kinetic models for both the two catalysts. In contrast, X-ray photoelectron spectroscopy (XPS) results showed that, after reaction in the presence of SO 2 , S concentration on αMnO 2 -NR@TiO 2 surface is lower than on αMnO 2 -NR surface, which demonstrated that TiO 2 shell could effectively inhibit the SO 2 diffusion onto MnO 2 surface. T hermogravimetry-differential thermosgravimetry (TG-DTG) results further pointed that SO 2 mainly react with TiO 2 forming Ti(SO 4 )O in αMnO 2 -NR@TiO 2 , which will protect Mn from being deactivated by SO 2 . These results were the reason for the better SO 2 resistance of αMnO 2 -NR@TiO 2 . After SO 2 introduction, the values of R 2 of the pseudo-second order model for αMnO 2 -NR and αMnO 2 -NR@TiO 2 are 0.997 and 0.992, which are still much higher than those of the pseudo-first order model. These results show that Hg 0 adsorption process in the presence of SO 2 atmosphere are also dominantly controlled by chemisorption.


Introduction
The emission of mercury from coal-fired power plants has drawn wide public concern in modern society. Mercury emissions are a long-term threat to human health and the environment because of extreme toxicity, persistence, and bioaccumulation. Therefore, controlling mercury emitted from coal-fired power plants has practical significance. Mercury in coal combustion flue gas is mainly present in three forms: Elemental mercury (Hg 0 ), oxidized mercury (Hg 2+ ), and particulate-bound mercury (Hg p ). Particulate-bound mercury (Hg p ) can be removed by electrostatic precipitators (ESP) and fabric filters (FF), while oxidized mercury (Hg 2+ ) can be captured by wet flue gas desulfurization system (WFGD). However, existing air pollution control devices can hardly remove Hg 0 due to its high volatility and low solubility.
Hg 0 capture with specific adsorbents is a usual way to control Hg 0 emissions from coal-fired power plants [1]. Activated cabon (AC) has been widely used for the adsorption of Hg 0 in coal-fired flue gas [2,3]. However, a huge amount of AC needs to be injected into flue gas because of its low Hg 0 capture capacity, which leads to a high operating cost of this technology. Sulfur or halogen modification can enhance adsorption ability of AC [4,5]. However, the injected AC is usually captured together with fly ash by particulate control device, and the Hg 0 adsorbed on AC will influence the fly ash utilization [6]. Therefore, alternative economic sorbents with high Hg 0 removal efficiency are necessary.
Oxides, such as CuOx [7,8], FeOx [9,10], CeOx [11,12] and MnOx [13][14][15], with high redox properties, exhibit great potential for Hg 0 adsorption. Among these oxides, MnOx is a commonly available and inexpensive material has received extensive attention due to the redox couples of Mn 2+ /Mn 3+ and Mn 3+ /Mn 4+ [16]. Electronic shift between the different valence states of Mn is active and leads to a high redox capacity. Stefano Cimino et al. [14] investigated the Hg 0 removal performance of Mn/TiO2 and found that Hg 0 capture efficiency was about 57% at 70 °C. After modification by some other transition metal oxides, Mn-based materials, such as Mn-FeOx [15], Mn-ZrOx [17], Mn-CeOx [18], and Mn-CuOx [19] can remove Hg 0 better. Furthermore, it has been reported that the shape and crystallographic phases of Mn based sorbents have serious effects on Hg 0 removal performance. Xu et al. [20] synthesized three different crystallographic phases of MnO2 and found that α-MnO2 had the highest capacity due to its larger surface area and oxidizability. Chalkidis et al. [21] pointed out that MnO2 nano-rods possessed good Hg 0 removal capacity owing to the higher surface adsorbed oxygen species.
However, Mn-based sorbents usually have a poor SO2 resistance as SO2 can easily react with Mn, thereby forming MnSO4 and leading to a largely suppressed Hg 0 removal activity. Even a little amount of SO2 will results in serious inhibited effects on Hg 0 removal process. Our previous work has indicated that Ce-Zr modified Mn sorbent will be totally deactivated in 1h after the introduction of 50 ppm SO2 due to SO2 poisoning Mn forming MnSO4 [22]. TiO2 is a traditional way to enhance the SO2 resistance of MnOx [23] as TiO2 can inhibit the deposition of sulfates on sorbents surface [24]. But the Hg 0 removal activity of MnOx/TiO2 is unsatisfactory because the active component of Mn is still exposed in SO2 atmosphere. Core-shell is a structure with active component core and supporting components shell. The shell can inhibit the interaction between SO2 and sorbent surface and efficiently protect active component core [25]. Therefore, synthesizing a core-shell structure with MnOx core and TiO2 shell may obtain a better SO2 resistance.
Inspired by this, αMnO2 nanorods and αMnO2 nanorods coating TiO2 shell were synthesized in the present work to investigate the Hg 0 removal efficiency in the presence of SO2. Thermo-gravimetric (TG) and X-ray photoelectron spectroscopy (XPS) were performed to determine the role of SO2 in the Hg 0 oxidation and adsorption processes and a probable mechanism of SO2 influence was deduced based on XPS and TG results. The kinetic model of the Hg 0 adsorption process was examined as well.

Structure Characterization
Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were performed to investigate the morphologic and structural properties of αMnO2-NR and αMnO2-NR@TiO2. Figure 1a,a' show SEM and TEM images of αMnO2-NR. It can be seen that αMnO2-NR has a uniform nanorod structure with an average diameter of about 100 nm. As shown in Figure 1b, for αMnO2-NR@TiO2, the uniform nanorod structure is well-retained after being coated with TiO2 and the packing state of this sample is similar to αMnO2-NR. The surface of αMnO2-NR@TiO2 are rougher when compared to αMnO2-NR, and the average diameter increases to 150 nm due to the TiO2 coating. The average length of the αMnO2-NR@TiO2 is about 2-3 μm (shown in Figure 1c). As shown in Figure  1b', an obvious dividing line can be detected between MnO2 core and TiO2 shell, and the shell with thickness of about 30 nm is well dispersed outside of the αMnO2-NR. N2 sorption-desorption isotherms of the samples are shown in Figure 2. Both αMnO2-NR and αMnO2-NR@TiO2 exhibit a type IV adsorption isotherm, according to the definition of IUPAC, which means that αMnO2-NR and αMnO2-NR@TiO2 have a mesoporous structure. The surface areas, pore volumes, and average pore diameters of the sorbents are illustrated in Table 1. BET surface areas of the two sorbents are similar, suggesting that TiO2 coating does not change the structure of αMnO2-NR a lot. This result consists with SEM results.  X-ray diffractometer (XRD) patterns of the two catalysts are shown in Figure 3. All the peaks in XRD pattern of αMnO2-NR and αMnO2-NR@TiO2 were indexed to cryptomelane type α-MnO2 (JCPDS 44-0141, tetragonal, I4/m, a = b = 0.978 nm, c = 0.286 nm). The intensity of diffraction peaks for the two samples is almost the same. It means that TiO2 shell does not influence the dispersion of αMnO2-NR, which is great agreement with BET and SEM results.

Hg 0 Adsorption Performance
Breakthrough experiments were performed to investigate the Hg 0 adsorption performance of the two sorbents. A blank test was also performed and the results is shown in Figure S1. It can be seen that the outlet Hg 0 concentration is stable when no sorbent was loaded in the fixed-bed reactor. As shown in Figure 4, the Hg 0 removal efficiency of αMnO2-NR is about 92% at the beginning of the test and it decreases to 41% after 130 min reaction. When it comes to αMnO2-NR@TiO2, the Hg 0 removal efficiency at the beginning of the test is about 81% which is lower than that of αMnO2-NR. But it is about 43% at the end of the test suggesting a more stable removal activity. These results indicate that TiO2 shell does not inhibit the Hg 0 diffusion from gas phase to the surface of αMnO2-NR.  Figure 5 shows the effects of SO2 on Hg 0 adsorption performance. For αMnO2-NR, Hg 0 removal efficiency sharply declines from 55% to 14% during the 35 min reaction, when SO2 is injected into flue gas. However, for αMnO2-NR@TiO2, the downward trend of Hg 0 removal efficiency is much slower and decreases from 76% to 43% in a 30 min test, and still has a Hg 0 removal efficiency of 25% after 80 min. These results confirm that TiO2 shell can inhibit the direct interaction between SO2 and MnO2 surface, which will efficiently protect MnO2 core from SO2 poisoning. αMnO2-NR@TiO2 was used to investigate reusability for Hg 0 removal. The results are shown in Figure 6. After 10 h Hg 0 adsorption test, αMnO2-NR@TiO2 reaches a Hg 0 adsorption equilibrium. And then, the sorbent was heated at 450 °C for 2 h to release the HgO on sorbent surface. It can be found that, after heated treatment, the Hg 0 adsorption efficiency and capacity of αMnO2-NR@TiO2 recovers to its original level. After two recycling, it still shows a good Hg 0 adsorption efficiency. Furthermore, SEM results of the fresh and used αMnO2-NR@TiO2 (shown in Figure S2) show that recycle have no effect on the microstructure. These results suggest an outstanding reusability of αMnO2-NR@TiO2. The Hg 0 adsorption capacity of αMnO2-NR@TiO2 is 0.11 mg/g, it is good enough compared to other sorbents (shown in Table S1). The surface areas of the sorbents in the present work are relatively low thereby lowering the available surface active sites. αMnO2-NR@TiO2 with higher surface area will be studied in our following works, and may give a better Hg 0 adsorption capacity. Hg 0 adsorption test of αMnO2-NR@TiO2 at different Hg 0 concentration was also investigated and the results are shown in Figure S3. With a doubled Hg 0 concentration, the breakthrough curve gets steep suggesting that αMnO2-NR@TiO2 will easily reach Hg 0 adsorption equilibrium at a higher Hg 0 concentration.

Structure-Activity Relationship
Fourier Transform Infrared Spectrometer (FTIR) was used to confirm the kind of surface active site for Hg 0 adsorption. As can be seen in Figure 7, the peaks at 429, 503, and 700 cm −1 correspond to Mn-O vibration [26], which becomes much weaker after reaction. It suggests that Mn-O group participates in Hg 0 adsorption process. According to previous work, the surface active oxygen species in Mn-O group should be the active sites for Hg 0 adsorption.

Models of Adsorption Kinetics
In order to better illustrate the Hg 0 adsorption mechanisms of αMnO2-NR and αMnO2-NR@TiO2, two popular models of pseudo-first order and pseudo-second order kinetic models, which have been widely used to investigate the adsorption process [27], were employed to fit the above experimental data. These two kinetic equations are displayed as follows [28]: where qe and qt are the adsorption capacity of Hg 0 on the sorbents at equilibrium, and at reaction time t (min), respectively. The parameters k1 (min −1 ) and k2 (g/(μg·min)) are the rate constants of the pseudo-first order, and pseudo-second order models, respectively. The fitting results are shown in Figure 8, and the obtained values of correlation coefficient (R 2 ) are summarized in Table 2. The values of R 2 of the pseudo-second order model for αMnO2-NR and αMnO2-NR@TiO2 are 0.991, and 0.995, respectively, which are higher than those of pseudo-first order kinetic model (0.944 and 0.938 for αMnO2-NR and αMnO2-NR@TiO2). It indicates that the pseudosecond order model can better fit the experimental data and Hg 0 removal process are dominantly controlled by chemisorption. After SO2 introduction, the values of R 2 of the pseudo-second order model for αMnO2-NR and αMnO2-NR@TiO2 are 0.997 and 0.992, which are still much higher than those of the pseudo-first order model. These results show that Hg 0 adsorption process in the presence of SO2 atmosphere are also dominantly controlled by chemisorption.

The Mechanism of SO2 Effects on Hg 0 Adsorption
XPS analysis was employed to explore the relative proportion of elements on the sample surface. The XPS spectra of Mn 2p, O 1s and S 2p for the fresh and used samples are shown in Figure 9. The surface atomic concentrations and surface atomic ratios are summarized in Table 3.   [29,30]. As shown in Table 3, the ratio of Mn 4+ /Mn is about 37.8% for the fresh αMnO2-NR and it decreases to 33.4% after the SO2 resistance test. Compared to αMnO2-NR, Mn 4+ content is almost constant for αMnO2-NR@TiO2 before, and after, SO2 resistance test. These results indicate that, for αMnO2-NR, Mn 4+ is easily reduced to Mn 2+ during SO2 resistance process via the reaction between SO2 and MnO2 [31]. For αMnO2-NR@TiO2, the interaction between SO2 and MnO2 is inhibited by the TiO2 shell structure, which can efficiently protect active component Mn 4+ in the core. Figure 9b shows O 1s XPS spectra. For the fresh catalysts, O 1s bands can be split into two peaks, corresponding to lattice oxygen (peak at 529.5 eV, denoted as Oα) and chemisorbed oxygen (peak at 530.8 eV, denoted as Oβ), respectively [32]. Whereas, a new peak appears around 532.3 eV after SO2 treatment, which corresponds to SO4 2− (denoted as Oγ) [33]. The intensity of the peak around 532.3 eV for αMnO2-NR@TiO2 is weaker than that for αMnO2-NR suggesting a lower amount of SO4 2− on the used αMnO2-NR@TiO2 surface. Furthermore, the peaks of Oα and Oβ in αMnO2-NR have an obvious slight shift to higher binding energy after SO2 treatment. It might be due to the formation of sulfate salts during the sulfating process [34].
To determine the above deduction, S 2p bands was further investigated and the results are shown in Figure 9c. For the fresh αMnO2-NR and αMnO2-NR@TiO2, two peaks around 162.2 eV and 163.2 eV attributed to S 2− and S2 2− can be detected [35,36], which may come from MnSO4 (the precursor of MnO2). But for the used αMnO2-NR and αMnO2-NR@TiO2, two new peaks at about 168.8 eV and 170.0 eV are observed, which may be assigned to SO4 2− , and HSO4 − , respectively [37,38]. The peak intensity of the used αMnO2-NR is much higher than that of αMnO2-NR@TiO2. As shown in Table 3, for αMnO2-NR, the surface atomic concentrations of S increases from 3.17% to 4.97% after SO2 teatment while it increases from 2.27% to 2.66% for αMnO2-NR@TiO2. These results confirm that TiO2 shell can inhibit the S accumulation on catalyst surface.
To obtain more information about the SO2 poisoning mechanism, Thermo-gravimetricdifferential thermos-gravimetry (TG-DTG) was performed to investigate the weight loss of αMnO2-NR and αMnO2-NR@TiO2 after SO2 treatment, and the results are presented in Figure 10. It can be seen that the used αMnO2-NR has an obvious weight loss step in the temperature range of 680−780 °C with a weight loss of about 2.4%, which can be attributed to manganese sulfate decomposition [39][40][41]. There is no weight loss step between 680−780 °C with respect to αMnO2-NR@TiO2, but there is a new weak step around 780-850 °C can be detected, and it may be due to the decomposition of Ti(SO4)O [42]. This result demonstrates that SO2 tends to react with titanium oxides instead of manganese oxides over αMnO2-NR@TiO2. Based on these results, TiO2 shell can lead to the preferential adsorption of SO2 on Ti surrounding forming Ti(SO4)O to protect Mn active component from being deactivated.  Figure 10. Thermo-gravimetric (TG) and differential thermos-gravimetry (DTG) of spectras of αMnO2-NR and αMnO2-NR@TiO2 after SO2 treatment.

Catalysts Preparation
The αMnO2 nanorods were synthesized through a hydrothermal method [43]. KMnO4 (2.5 g, AR) and MnSO4·H2O (1.05 g, AR) were dissolved in 80 mL distilled water. The mixed solution was transferred into a Teflon-line stainless steel autoclave, sealed, and kept in an oven at 160 °C for 12 h. After cooling to room temperature, the precipitates were filtered off, washed several times using deionized water and dried at 110 °C overnight. Finally, the product was calcined at 400 °C in a muffle furnace for 4 h and the obtained sample is denoted as αMnO2-NR.
MnO2@TiO2 core-shell nanorods were synthesized through a versatile kinetics-controlled coating method [44]. αMnO2-NR (0.075 g) and aqueous ammonia (0.28 mL, 28 wt.%) were dispersed in 100 mL absolute ethanol under ultrasound for 30 min. Afterwards, titanium tetrabutoxide (TBOT) (0.75 mL) was added drop-wise into the mixture and then kept at 45 °C for 24 h. The mixed solution was filtered, washed and dried at 60 °C for 12 h. Finally, the solid was calcined under flow air at 500 °C for 2 h to obtain the sample (denoted as αMnO2-NR@TiO2).
ALL reagents are from Aladdin company, Shanghai, China.

Hg 0 Adsorption Experiments
The Hg 0 removal test has been described in detail in our previous work [45]. The experimental reactor contains a gas distribution system, a Hg 0 vapor generating device, a fixed-bed quartz reactor (ID = 8 mm), an online mercury analyzer and a tail gas treating unit. The mercury permeation tube was placed in a U-shape glass tube, which was immersed in a water bath at a constant-temperature (38 °C) to ensure a constant Hg 0 permeation rate. The total gas flow was 600 mL/min, and the sorbent volume was generally 0.2 mL, resulting in a GHSV of 1.8 × 10 5 h −1 . The concentrations of Hg 0 and SO2 were monitored by a VM-3000 online mercury analyzer (Mercury Instruments, München, German), and flue gas analyzer (KM950, Kane International Ltd., London, United Kingdom), respectively.
During each test, the Hg 0 gas first bypassed the fixed-bed reactor, and then introduced into the reactor for 2 h to obtain a stable Hg 0 concentration. Hg 0 breakthrough ratio was quantified by the following formula, where C and C0 represent the inlet and outlet Hg 0 concentrations (μg/Nm 3 ) in the fixed-bed reactor.

Characterization
The morphology and microstructure of the samples were observed using SEM (Nova NanoSEM 450, FEI) and TEM (Tecnai G2 F30 S-Twin, FEI). The surface areas and pore parameters of the samples were determined by Nitrogen adsorption/desorption method at liquid nitrogen temperature at −196 °C on an automated gas sorption analyzer (Autosorb-iQ-C, Quantachrome Instruments, Boynton Beach, FL, USA). The pore size and pore volume were derived from the desorption branches using the Barrette-Joynere-Halenda (BJH) model. The crystal structures of the samples were characterized by an XRD (XRD-7000S, SHIMADZU Corporation, Kyoto, Japan) operating at 40 kV and 100 mA using a Cu Kα radiation. The scanning range (2θ) was from 10° to 90° with a scan speed of 5°/min. The element (Mn, O, and Hg) valence state was analyzed by XPS (ESCALAB250 Thermo Fisher Scientific, Wilmington, DE, USA) with a monochromatic Al Kα source. The C 1s binding energy value of 284.8 eV was used to calibrate the observed spectra. TG was performed on TGA/DSC1 analyser (METTLER TOLEDO, Schwerzenbach, Switzerland), under a nitrogen flow of 20 mL/min, using a heating rate of 10 °C/min from room temperature to 900 °C (NETZSCH Corporation, Selb, Germany). DTG analysis was obtained based on residual weight of the sample with respect to time. FTIR spectra were obtained on a Nicolet Magana-IR 750 spectrometer to measure the surface groups of the samples (Thermo Nicolet Corporation, Madison, WI, USA).

Conclusions
αMnO2-NR@TiO2 was prepared by versatile kinetics-controlled coating method to compare with αMnO2-NR in the Hg 0 removal process. SEM, BET, and XRD results showed that TiO2 shell did not change the structure of αMnO2-NR. Therefore, the two sorbents had similar Hg 0 removal performance in N2 atmosphere. When SO2 was introduced, αMnO2-NR@TiO2 had a much better performance than αMnO2-NR. XPS and TG-DTG results showed that αMnO2-NR@TiO2 had lower surface S concentration after treatment of SO2, and no manganese sulfate could be detected in αMnO2-NR@TiO2. It suggests that the TiO2 shell can effectively protect MnO2 from being deactivated by SO2. Adsorption kinetic results showed that Hg 0 adsorption process over both the two sorbents obeys pseudo-second order model with, or without, SO2.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Figure S1: Outlet Hg 0 concentration without sorbent, Figure S2: The image of αMnO2-NR@TiO2 after adsorption, Figure S3: Breakthrough curve of αMnO2-NR@TiO2 with different Hg 0 feed concentration, Table S1: Comparison of the adsorption capacities of the sorbents.