TiO2–SiO2 supported MnWOx catalysts by liquid-phase deposition for low-temperature NH3-SCR

NH3-SCR is an environmentally important reaction for the abatement of NOx from different resources. MnO2-based catalyst has attracted significant attention due to the excellent activity. In this paper, a series of MnWOx/TiO2–SiO2 catalysts were prepared by liquid-phase deposition method. The catalysts were characterized by N2 adsorption/desorption, XRD, TEM, XPS, FT-IR, H2-TPR, TG and water adsorption capacity. The existence of SiO2 improved the SO2 and H2O resistance of the MnWOx/TiO2–SiO2 catalyst without decreasing the NH3-SCR activity. Under the reaction conditions of 260°C and 60 000 ml gcata h−1 gas hourly space velocity (GHSV), the NO conversion was kept stable at about 95% for 140 min on stream. The excellent performance of MnWOx/TiO2–SiO2 catalyst is considered to be originated from the texture properties and active species dispersion improvement by SiO2 in the support and low-temperature preparation.


Introduction
Nitrogen oxides are one of the major pollutants that endanger the atmosphere air, which can cause severe environmental problems such as acid rain, smog and photochemical pollution. deNO x is a hot topic in the environmental fields in the past decades, and many researchers continue to be concerned till present. Comparing to the other routes, it is a desirable way to reduce NO x to harmless N 2 . Selective catalytic reduction of NO x via NH 3 (NH 3 -SCR) thereby has been widely studied and applied in the elimination of nitrogen oxides [1,2].
In general, NO x are mainly released from the combustion of fossil fuels at different stationary or mobile resources. The nature of the two types of exhaust gases is quite different. The catalysts are also evolved into two series, the transition metal oxide-based catalysts for the stationary resources, and the 2. Experimental procedure 2.1. Catalyst preparation MnWO x /TiO 2 -SiO 2 catalysts were prepared by two main procedures. The first step is to prepare TiO 2 -SiO 2 support by a sol-gel method. The method used TEOS (n-ethyl silicate) and TBOT (tetrabutyl titanate) as precursors. Taking 10% SiO 2 -90% TiO 2 as an example, a suitable amount of TEOS (1.0 ml) was mixed in anhydrous ethanol, then a small amount of distilled water (about 2.0 ml) and several drops of 1 mol l 21 HCl were added stirring for 30 min at room temperature to adjust pH ¼ 2 and got solution A. Meanwhile, 5.0 ml of acetic acid was mixed with 49.0 ml anhydrous ethanol, then a small amount of distilled water (about 2.0 ml) and several drops of 1 mol l 21 HCl were added to adjust pH 3 and got solution B. Solutions A and B were then uniformly mixed by stirring to get solution C. Then 13.5 ml anhydrous ethanol and 13.5 ml TBOT were uniformly mixed to get solution D. Solution D was added into solution C at a rate of 3 ml min 21 under vigorous stirring at room temperature. The resulting material was dried in an oven at 808C for 12 h to obtain the TiO 2 -SiO 2 dry gel. The gel was ground into powder to obtain the support.
Considering the dispersion of active components on the hydrophobic support, a modified liquid deposition method was used to load the MnWO x on the TiO 2 -SiO 2 . Manganese nitrate solution (50% Mn(NO 3 ) 2 ) and tungaline ((NH 4 ) 10 W 12 O 41 . xH 2 O) were used as precursors; 0.845 g (NH 4 ) 10 W 12 O41 . xH 2 O and 2.386 g Mn(NO 3 ) 2 solution were dissolved with the same molar oxalic acid in deionized water. TiO 2 -SiO 2 powder was added to the solution and the mixture was vigorous stirred royalsocietypublishing.org/journal/rsos R. Soc. open sci. 6: 180669 for 1 h. Ammonia water (0.5 mol l 21 ) was slowly added into the mixture to adjust the pH to 10 to achieve the precipitation of active species. The mixture was filtered, washed and the precipitate in the oven was dried at 1108C for 12 h. At last, the sample was moved into the muffle and calcined at 2008C for 2 h to obtain the target MnWO x /TiO 2 -SiO 2 catalyst. The samples were labelled as Mn x W y O x /TiO 2 -SiO 2 (n), where x and y are the molar ratio of Mn : W and n is the percentage of SiO 2 to (SiO 2 þ TiO 2 ).

Catalyst characterization
N 2 adsorption/desorption of the catalysts were measured by a Micromeritics 3Flex physical adsorption instrument. The samples were heated to 2008C under vacuum pressure and kept for 10 h before measurement. Specific surface area was calculated using the BET method.
Crystal structure of the catalysts were detected by an ARL SCINTAG X'TRA instrument (Shimadzu.) using Cu Ka radiation in 2u range of 10-808 with a step size of 0.028. H 2 temperature-programmed reduction (H 2 -TPR) was conducted on a chemical adsorption apparatus (Finetec Corp.). The samples were pretreated at 4008C in Ar for 40 min. The TPR analysis was carried out in a reducing gas mixture (30 ml min 21 ) consisting of 5% H 2 and balance Ar from 60 to 8008C at a rate of 108C min 21 . TCD detector temperature was 608C.
X-ray photoelectron spectroscopy (XPS) was conducted on a Kratos AXIS Ultra DLD clutches spectrophotometer. Excitation source was the monochromatic Al Ka radiation (hy ¼ 1486.6 eV). The power was 45 W. The working voltage was 15 kV. Scanning area was 300 Â 700 mm. Vacuum test was better than the 8.5 Â 10 29 mbar. The data were corrected using the C1 s 284.8 eV as the standard.
Skeleton FT-IR spectra of the catalysts were recorded using BRUKER VERTEX70 FT-IR apparatus, with the sample in KBr pellets.
Morphologies of the catalysts were disclosed by a Philips-FEI company Tecnai G2F30 S-Twin type high-resolution transmission electron microscopy. The samples were solved in ethanol solution to get the dispersed particles. The accelerating voltage was 300 kV.
The water adsorption capacity of the catalysts was determined by the saturated water adsorption experiment. The appropriate amount of dry sample was put into a weighing bottle and placed in a saturated water vapour environment. After the adsorption is balanced, the mass change was weighed with a electronic balance.
The coke of the catalyst after the reaction was analysed by an STA449 integrated thermal analyser. A total of 0.100 g of the sample was placed in an alumina crucible and heated from room temperature to 10008C at a rate of 208C min 21 in a 20% O 2 /80% N 2 atmosphere.

Catalytic activity tests
NH 3 -SCR activity was tested in a fixed-bed quartz tube reactor (i.d. ¼ 8 mm). The reaction conditions were as follows: 500 ppm of NO, 500 ppm of NH 3 , 5.0% O 2 and N 2 as balance, 10 vol% H 2 O (when used), 100 ppm SO 2 (when used). Total gas flow rate was about 500 ml min 21 , and calculated gas hourly space velocity (GHSV) of 60 000 ml g cata h 21 . Outlet gases were measured by an online MODEL T200H/M nitrogen analyser and Thermo trace 1300 gas chromatography. NO conversion (X) and N 2 selectivity (S) were calculated by the following equations: sample temperature could rise to very high (greater than 10008C) during the preparation process and affect the pore structure of the TiO 2 -SiO 2 support. Figure 1 shows that the activity of Mn 2 WO x /TiO 2 -SiO 2 (10) is somewhat worse than that of the Mn 2 WO x /TiO 2 with the same SHS procedure. Therefore, a modified liquid deposition method is used to prepare the Mn 2 WO x /TiO 2 -SiO 2 catalysts later. LPD is a method that through the reaction of the precursors salts solution forms a precipitate and deposit on the solid support pre-placed in the solution to realize the uniform loading of active components. It has been found that Mn(NO 3 ) 2 and (NH 4 ) 10 W 12 O 41 can not only react to produce a precipitation in the solution, but also form a MnWO 4 structure in the precipitation [29], which is considered to be favourable for the activity of Mn 2 WO x / TiO 2 catalysts. Here, the method is also adopted to prepare Mn 2 WO x /TiO 2 -SiO 2 . The sample exhibits rather high activity in NH 3 -SCR reaction. The NO conversion reaches 90% from 2008C to about 4008C, which is even broader than the results of the Mn 2 WO x /TiO 2 by SHS method.
The preparation procedure has a clear influence on the activity of Mn 2 WO x /TiO 2 -SiO 2 by LPD method. If the TiO 2 -SiO 2 gel was calcined at 5008C before the LPD, the activity of the as-prepared catalyst would decrease. That is to say, the NO conversion only keeps above 90% in a narrow range of 200 -2308C. It is because, after calcination, the pore structure of TiO 2 -SiO 2 gel is decreased to some extent and the hydrophobicity increases, which is not favourable for the dispersion and deposition of active components. Nonetheless, the LPD method can effectively load MnWO x active species on the TiO 2 -SiO 2 support and keep the activity of Mn 2 WO x /TiO 2 -SiO 2 catalyst in the right preparation procedure and conditions. Figure 2a shows the NH 3 -SCR activity and figure 2b shows the N 2 selectivity of the Mn x W y O x /TiO 2 -SiO 2 catalysts with different Mn : W ratios synthesized by LPD. It can be seen that the Mn : W ratio influences the temperature window of the MnWO x /TiO 2 -SiO 2 catalysts. Taking 90% NO conversion as the criterion, the working temperature window of the Mn x W y O x /TiO 2 -SiO 2 catalysts varies from 120 to 2008C, 120 to 3008C, 140 to 2808C, 200 to 3408C and 200 to 3708C when the Mn : W ratio decreases from 3 : 1, 2 : 1, 1 : 1, 1 : 2 and 1 : 3, respectively. Taking Mn : W ratio of 1 : 1 as a dividing line, when the manganese content is rich, the catalyst has better activity at low temperatures and worse activity at high temperatures; while when the tungsten is rich, it is just in the opposite way. In general, the Mn 2 WO x / TiO 2 -SiO 2 shows the widest working temperature window.
Comparing to activity, Mn : W ratio has a more pronounced effect on N 2 selectivity of the catalyst. In figure 2b, over the MnW 2 O x /TiO 2 -SiO 2 and MnW 3 O x /TiO 2 -SiO 2 , the N 2 selectivity is almost unchanged with the temperature rising; while over the other samples, the N 2 selectivity is decreased from about 100% to 87%, 82% and 75% when the Mn : W ratio increases from 1 : 1 to 3 : 1, respectively. The results indicate that Mn element is favourable for the low-temperature activity and W element is favourable for the N 2 selectivity. It is desirable to maintain a certain manganese tungsten ratio for the MnWO x /TiO 2 -SiO 2 catalyst to get a satisfying activity and N 2 selectivity.
The support properties often can be adjusted by the composite variation and influence the catalyst properties. TiO 2 -SiO 2 oxides with different SiO 2 contents from 5% SiO 2 /95% TiO 2 to 20% SiO 2 /80%  (20), are studied. Figure 3a shows the results of the activities of the samples in the NH 3 -SCR reaction. Taking 90% NO conversion as criterion, the increase of SiO 2 content in the support does not change the working temperature window of Mn 2 WO x /TiO 2 -SiO 2 too much. All the low-temperature boundaries of the working temperature windows are around 1708C. When the SiO 2 content increases from 5% to 20%, it only makes the high-temperature boundary move a slight degree to the high-temperature direction, and slightly widens the working temperature window. The working temperature windows of Mn 2 WO x / TiO 2 -SiO 2 (10) and Mn 2 WO x /TiO 2 -SiO 2 (20) are the same as 170-3708C.
The addition of SiO 2 also has a positive effect on the N 2 selectivity of the catalyst, as shown in figure 3b. It can be found that N 2 selectivity on the Mn 2 WO x /TiO 2 (without SiO 2 ) catalyst at 508C began to decrease rapidly with the increase of temperature. Nonetheless, the N 2 selectivity can be maintained at above 90% on all the Mn 2 WO x /TiO 2 -SiO 2 samples. Further, as the SiO 2 content increases, N 2 selectivity decreases slowly, and it decreases the slowest on the Mn 2 WO x /TiO 2 -SiO 2 (20) sample.

The resistance of MnWO x /TiO 2 -SiO 2 to H 2 O and SO 2
As is known, the flue gas usually contains SO 2 and water vapour, which often affects the NH 3 -SCR catalyst performance. It is the primary reason that the Mn 2 WO x /TiO 2 catalyst is modified by SiO 2 in this paper. Figure 4 shows the NH 3 -SCR results of Mn 2 WO x /TiO 2 -SiO 2 catalysts with different SiO 2 content in 100 ppm SO 2 at different reaction temperature. The SO 2 is introduced after 60 min of reaction time on stream.
It can be seen that SiO 2 has a significant effect on the sulfur resistance of the Mn 2 WO x /TiO 2 -SiO 2 catalyst. Taking the introduction time of SO 2 flow (60 min) as the initial point, for the Mn 2 WO x /TiO 2 catalyst, the NO conversion begins to decrease after 90 min (total 150 min time on stream) reaction after the introduction of SO 2 flow. It decreases with a slow rate till 240 min (total 300 min), and then  decreases with a significant rate to 55% at 420 min (total 480 min), where SO 2 is switched off. The NO conversion then rises back to about 65% in SO 2 -free flow after 120 min (total 600 min on stream). The performance of Mn 2 WO x /TiO 2 -SiO 2 (5) was similar to that of Mn 2 WO x /TiO 2 catalyst before 240 min, but the NO conversion decreased significantly slower than the Mn 2 WO x /TiO 2 catalyst. The NO conversion is still close to 80% at 420 min, and after SO 2 switches off for 120 min, it rises back to close to 90% again. For the catalyst Mn 2 WO x /TiO 2 -SiO 2 (10), the NO conversion decreases with a rather slow rate after 30 min of SO 2 flow introduction in a linear style till 300 min, where the NO conversion is kept at about 80%. In the remaining time, no significant change is observed even though the SO 2 is switched off. Finally, for the catalyst Mn 2 WO x /TiO 2 -SiO 2 (20), the NO conversion decreases clearly in a different way from other samples. A number of stable platforms appeared on the profile, mainly in the range of 160 -220 min, 260 -300 min and 330-420 min total time on stream. When the SO 2 is switched off, the NO conversion can rise back to about 70% after 120 min. The results show that the appropriate amount of SiO 2 is beneficial to improve the sulfur resistance of the Mn 2 WO x /TiO 2 -SiO 2 catalyst.
To further evaluate the effect, the performance of Mn 2 WO x /TiO 2 -SiO 2 (10) is then tested under 2308C and 2008C. However, the results in figure 4b are not very good. At 2308C, NO conversion has dropped below 50% after 420 min on stream; and at 2008C, NO conversion only is left no more than 20%. It means that the effect of SiO 2 is still limited and the catalyst needs to be more improved in the future.
Next, the activity of Mn 2 WO x /TiO 2 -SiO 2 (5) and Mn 2 WO x /TiO 2 -SiO 2 (10)       The presence of water has little effect on the texture properties of the catalysts. All the average pore sizes do not change after reaction, and the specific surface area of the catalysts decreases about 10 m 2 g 21 . SO 2 significantly affects the nature of the catalysts. All the specific surface area and pore volumes of the catalysts after the reaction are largely reduced. The largest decrement is from Mn 2 WO x /TiO 2 -SiO 2 (20). The specific surface area decreases 71.8 m 2 g 21 , and the pore volume decreases 0.046 cm 3 g 21 . The second is the Mn 2 WO x /TiO 2 , 67.3 m 2 g 21 and 0.048 cm 3 g 21 . Mn 2 WO x /TiO 2 -SiO 2 (10) and Mn 2 WO x /TiO 2 -SiO 2 (5) show somewhat better resistance to SO 2 . Their specific surface areas only decrease 42.4 and 48 m 2 g 21 . Especially, the pore volume of Mn 2 WO x / TiO 2 -SiO 2 (10) only decreases 0.022 cm 3 g 21 . The average pore size of Mn 2 WO x /TiO 2 -SiO 2 (5) sample is the only one enlarged after the reaction. Coexistence of water and SO 2 does not show a synergistic effect on the nature of the catalysts. The change in the values is close to the sum of the influence of the two compounds. Figure 6 shows the XRD pattern of the Mn 2 WO x /TiO 2 -SiO 2 catalysts. It can be seen that for the as-  results confirmed that the reaction to form manganese tungstate has occurred in the solution. The sample has well crystallinity with a uniform particles size distribution. Meanwhile, for the silicon-containing samples, the particles are also relatively small. But the lattice fringes are not very clear due to the high contents of the mixed crystal composition, which means that the SiO 2 affects the crystallinity of the MnWO x active species. Figure 8 shows the H 2 -TPR profiles of Mn 2 WO x /TiO 2 -SiO 2 catalysts with different SiO 2 contents. The reduction peaks may originate from the different active species and supports, which leads to the profiles somewhat complicated. In general, the profile can be divided into two reduction bands and one big peak. According to the previous discussions [30 -32], for the Mn 2 WO x /TiO 2 sample, in the lower temperature range of 250-4808C, the reduction bands or peaks are mainly originated from different MnO x species; in the higher temperature range of 500-8008C, the reduction band and peak are from different Ti or W oxides species. More detailed, the identified reduction peaks at about 3008C and 3408C are attributed to the reduction peaks of MnO 2 ! Mn 3 O 4 and Mn 3 O 4 ! MnO. The reduction band of 5208C-6008C is the contribution of Ti 4þ ! Ti 3þ , and the peak at 7808C is from W 6þ ! W 4þ . When the SiO 2 is introduced into TiO 2 , there are several significant changes observed. The first is that the reduction band in the low-temperature range moves to the high-temperature attitude; the second is the strength of the reduction band in the middle temperature increases and the peak at 7808C decreases. The movement of the band in the low-temperature range means that the MnO x species are inclined to be hard to reduce, which shall not be favourable for the NH 3 -SCR activity of the Mn 2 WO x /TiO 2 -SiO 2 catalysts. However, the NH 3 -SCR activity seems to be not influenced clearly. It might be amended by the dispersion increment of active species by the TiO 2 -SiO 2 support. The  royalsocietypublishing.org/journal/rsos R. Soc. open sci. 6: 180669 strength increment of the peak at the middle temperature illustrates that more TiO 2 can be reduced. The TiO 2 crystal structure might be broken and more amorphous or isolated species have been formed.

Catalysts characterization
In order to investigate the valence state of the Mn species and O species, XPS characterization was performed for the samples. Figure 9a shows the O1 s spectra of Mn 2 WO x /TiO 2 , Mn 2 WO x /TiO 2 -SiO 2 (5), Mn 2 WO x /TiO 2 -SiO 2 (10) and Mn 2 WO x /TiO 2 -SiO 2 (20), and table 3 lists the quantitative results via Gauss deconvolution method. After deconvolution, three sub-bands can be found on the spectra. As previous reports, the sub-bands around 529.7 eV can be attributed to the lattice oxygen (denoted as O b ); the sub-bands around 531.1 eV to surface absorbed oxygen (denoted as O a ) such as O 2 22 and O 2 , and the sub-band around 533.2 eV can be assigned to chemisorbed water (denoted as O a 0 ) [17,33]. With the SiO 2 introduction increases, the peak of O b decreases and the peak of O a and O a 0 increases gradually. As the O a is usually regarded as more reactive in redox reactions due to its higher mobility than lattice oxygen, the percentages of O a , O b and O a 0 are often viewed as an indicator for the redox ability of the catalyst. From   The weak water adsorption capacity of the Mn 2 WO x /TiO 2 -SiO 2 catalysts may come from the formation of silicon hydroxyl. The skeleton FT-IR of the Mn 2 WO x /TiO 2 -SiO 2 catalysts is then recorded and the results are shown in figure 10. It can be found that the hydroxyl vibration peaks at 1502 cm 21 are significantly changed after introduction of SiO 2 . For the Mn 2 WO x /TiO 2 -SiO 2 (20), it has changed from the shoulder peak to the independent peak, corresponding to its higher water adsorption. Meanwhile, the peak is weakened for the Mn 2 WO x /TiO 2 -SiO 2 (5) and the Mn 2 WO x / TiO 2 -SiO 2 (10) and their water adsorption content are somewhat lower too. Thus, the appropriate SiO 2 content can reduce the amount of hydroxyl groups on the catalyst surface, and the excessive SiO 2 increases the amount of hydroxyl groups on the catalyst surface.

Discussions
As mentioned earlier, this paper is intended to indirectly improve the SO 2 resistance of the MnWO x /TiO 2 catalyst by increasing its hydrophobicity. According to the activity tests, the SO 2 and water resistance of the MnWO x /TiO 2 -SiO 2 catalyst can be really improved under the premise that the NH 3 -SCR activity is maintained. The NO conversion is no longer decreased as soon as the catalyst contacts SO 2 , but it is maintained for some time on stream and then decreased slowly. Meanwhile, the water adsorption test shows that the water capacity of the catalyst decreases after SiO 2 introduction.
However, it is a little hasty to make this assertion, as SiO 2 introduction also changes some other properties of the catalyst. The specific surface area and pore volume of the MnWO x /TiO 2 -SiO 2 catalysts is clearly larger than the MnWO x /TiO 2 . The decreasing degree of the specific surface area and pore volume of the MnWO x /TiO 2 -SiO 2 catalysts after reaction is also lighter than the MnWO x / TiO 2 . Therefore, TG analysis has been performed after reaction and the results are displayed in

Conclusion
A series of TiO 2 -SiO 2 mixed oxides supports were prepared by sol-gel method, and the MnWO x active components were loaded on the supports by liquid deposition method to form a series of MnWO x / TiO 2 -SiO 2 catalysts. Owing to the reaction of manganese nitrate and ammonia tungstate to form MnWO 4 , the catalysts could be prepared at low temperature. The introduction of SiO 2 in the TiO 2 support increases the N 2 selectivity of the MnWO x /TiO 2 catalyst without decreasing the NH 3 -SCR activity. Meanwhile, with appropriate SiO 2 (5% and 10% percentage of TiO 2 ), the sulfur resistance of MnWO x /TiO 2 -SiO 2 is clearly improved, the decreasing rate and degree of NO conversion are both slowed down and retarded. By a series of BET, XRD and TEM characterization, it can be inferred that the texture properties of MnWO x /TiO 2 -SiO 2 are modified and the dispersion of MnWO x species is improved, which is beneficial for the NH 3 -SCR activity. Skeleton FT-IR, water adsorption capacity and XPS show that SiO 2 introduction decreases the water adsorption capacity especially at the high temperatures, which might be favourable for the water and sulfur resistance of the catalyst.
Data accessibility. Our data are available from Dryad Digital Repository: doi:10.5061/dryad.74pj317 [36]. Authors' contributions. W.Z.L. carried out the molecular laboratory work, participated in data analysis, carried out sequence alignments; Z.K.Z. participated in the design of the study and drafted the manuscript; H.F.L. conceived of the study, designed the study, coordinated the study and helped draft the manuscript. All authors gave final approval for publication.