Investigation of SO2 tolerance of Ce-modified activated semi-coke based catalysts for the NO + CO reaction

Activated semi-coke was loaded with Fe–Comixed oxides and doped with an optimized amount of cerium oxides. This prepared catalyst exhibited excellent NO removal (deNO) activity, and also showed outstanding SO2 resistance at 250 C. To understand the SO2 tolerance mechanism, the catalysts were characterized by scanning electron microscopy, energy-dispersive X-ray spectroscopy, X-ray diffraction, in situ Fouriertransform infrared spectroscopy, H2-temperature-programmed reduction, and SO2-temperatureprogrammed desorption, as well as CO–deNO activity testing under different conditions. The results indicate that the Ce (molar ratio 1⁄4 0.1) doped onto the Fe–Co binary oxide catalysts would promote the generation of Ce2(SO4)3. This generation could prevent the active metal oxides from being poisoned by SO2. Furthermore, this kind of sulfate would weaken the interaction between SO2 and NO, so that the adsorbed NO will have a better opportunity to react with CO.


Introduction
Selective catalytic reduction of NO x with NH 3 (NH 3 -SCR) has proven to be the most effective process for removing NO x from stationary sources and diesel engines. [1][2][3] Moreover, it is presently the most applied method for NO x removal in power plants. 1,4 However, the NH 3 -SCR process has some disadvantages, such as NH 3 leaks, the toxicity of vanadium catalysts, and use of air preheater blocks. 5,6 Recently, researchers have proposed that the leaked NH 3 will interact with SO 2 to form (NH 4 ) 2 SO 4 . This sulfate evolves in the natural atmosphere; furthermore, it is a major contributor to the formation of Chinese haze. 7 Therefore, it is extremely urgent to provide a substitute for the NH 3 -SCR process.
Among the processes for NO x removal from vehicle exhaust, selective catalytic reduction with hydrocarbon (HC-SCR) 6 and three-way catalyst technology (TWC) 8 have been widely investigated. These technologies make almost no contribution to the NH 3 leaks. If these could be applied to the NO x elimination process in power plants, the Chinese haze could be alleviated. In recent years, researchers have begun to focus on investigations such as these. [9][10][11][12] However, the application of the new processes has faced some difficulties: one is that the presence of O 2 will consume the majority of the reductant. [10][11][12] Therefore, in consideration of the negative effects of O 2 , the NO x adsorptionreduction process (model reactor in Fig. S1 †) was proposed by Professor Bi. [10][11][12] This technology could also achieve high deNO x efficiency. Developed from the TWC process, CO-deNO x is very suitable for NO x adsorption-reduction. This is because CO is inexpensive, can easily be produced, and cannot generate solid carbon deposits 10,13 upon reaction with NO. It is widely reported that transition metal oxides could catalyze the NO + CO reaction efficiently. In fact, Cu-, [14][15][16][17][18][19] Co-, [20][21][22] Fe-, 13 Ni-, 23 and Mn-based 14,24 oxides, as either loaded or non-loaded catalysts, exhibit excellent CO-deNO efficiency at temperatures between 150 and 300 C. Furthermore, CeO 2 , SnO 2 , or ZrO 2 mixed with transition metals and carbon supported metal catalysts also promote the efficient reaction of CO with NO. 17,[25][26][27] As for the mechanism of the NO + CO reaction, it is established that at relatively low temperatures, NO coordinates with the metal cation and generates nitrites. 5,13,28 Subsequently, the formed nitrites can react with CO to yield N 2 O and CO 2 . 5,28 However, as the temperature increases, the coordinated NO is transformed to nitrates that react with CO, generating N 2 and CO 2 . 5,13,27,28 In our previous study, CO-deNO (catalyzed by semi-coke based catalysts) were applied to a NO x adsorption-reduction process, and the deNO efficiency was relatively high. 5,9,27 However, in this process, SO 2 can be adsorbed onto the adsorbents more easily than NO. 5 Even if the deNO reactors are placed downstream of the desulfurization equipment, there is an appreciable quantity of SO 2 in the ue gas, which can severely decrease the deNO activity. As shown in our previous studies, 5,28 the addition of SO 2 reduces NO conversion by approximately 60% at 250 C, and the coordinated SO 2 induced the reversible catalyst deactivation. Therefore, the improvement of SO 2 resistance for semi-coke based catalysts should be investigated.
As for the deactivation of catalysts caused by SO 2 , the most accepted theses are as follows: (1) the competitive adsorption between NO and SO 2 can result in no active sites being available for NO on the catalyst surface; 29 (2) SO 2 can easily occupy the acid site that is the vital point for NH 3 -SCR; 1,30 (3) the interaction of SO 2 and H 2 O can generate some surface sulfates; for instance, Mn, 30 Cu, 31 Co 32 oxides are very actively drawn to SO 2 ; therefore, the SO 2 present will easily transform these oxides to sulfates; and (4) the formation of (NH 4 ) 2 SO 4 or NH 4 HSO 4 (NH 3 -SCR) will accumulate on the surface, which would reduce the available surface area and then deactivate the catalysts. 33,34 Therefore, to understand the SO 2 poisoning mechanism, researchers have focused on the investigation on the SO 2 tolerance of deNO x catalysts. Until now, breakthroughs in the investigation of SO 2 resistance have occurred in two domains, viz. the doping of SO 2 resistant metals onto active metals and the optimization of the micro-structure. From wide study, it is known that the doping of Ti-, 32,35 Zr-, 36 Sn-, 37 Ce-, 2,33 oxides onto the SCR catalysts as active components or supporters can reduce the thermal stability of the surface sulfates. The excellent decomposing performance of surface sulfates is the vital factor for SO 2 resistance. 2 Recently, it was established that SO 2 could directionally accumulate on the surface of CeO 2 affording some bulk-like sulfates, 1,2 which would be benecial for NO reacting with the active metals. As for the micro-structure optimization, researchers invented mipor supporters like SSZ-13. [38][39][40] The pore size of these supports is smaller than the molecular vibration radius of SO 2 , but larger than that of NO x , which can prevent the generation of surface sulfates.
In the present study, based on the previous investigation on the SO 2 tolerance of activated semi-coke based catalysts, 5 we coimpregnated Ce and Fe-Co binary oxides onto the coke based supports using a hydrothermal method. The CO-deNO results demonstrate that Ce doping can increase the SO 2 -resistant performance. A variety of methods were used to obtain insight into the mechanism, including scanning electron microscopy (SEM), X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), N 2 physisorption, temperature-programmed reduction of hydrogen (H 2 -TPR), and temperature-programmed desorption of SO 2 (SO 2 -TPD). The evolution of surface components was also detected by in situ diffuse reectance infrared Fourier transform spectroscopy (DRIFTS).

Catalysts preparation
Commercial semi-coke (Shaanxi Shenmu Coal Mine Co., Ltd., China) was rst ground and sieved into granules with diameters of 1.02-1.27 mm (labeled SC). Second, the SC particles were activated using nitric acid (30 wt%) at 80 C for 2 h. Next, aer the particles had been washed with deionized water, they were dried at 120 C for 6 h, followed by calcination in Ar at 700 C for 4 h (labeled ASC).
The supports (ASC) were loaded with transition metals using a hydrothermal method: ferric nitrate, cobalt nitrate, and cerium nitrate (analytical-reagent grade, Sinopharm Chemical Reagent Co., Ltd.) were rst dissolved in deionized water for use as precursors. Table 1 summarizes the loading-amount parameters of the prepared catalysts. Then, ASC (5 g) immersed in 30 mL of the precursor was transferred to a stainless-steel autoclave. The autoclave was maintained at 160 C for 24 h. Next, the activated coke particles were washed using deionized water and then dried at 120 C for 6 h, followed by calcination in Ar at 700 C for 4 h.

Characterization
The textural properties were evaluated by physical adsorption of N 2 at 77 K using an automatic surface analyzer (Quantachrome Autosorb 1C), and the specic surface areas and pore volumes were calculated using density functional theory (DFT) from the N 2 ad-/de-sorption isotherms. XRD detection was performed on a Rigaku D/max 2400 diffractometer using Cu-Ka radiation (l ¼ 1.5056Å) at a scanning rate of 8 min À1 with a step size of 0.02 over a 2q range of 10-80 . The surface atomic states of the catalysts were analyzed using XPS (Axis Ultra DLD ) with Al-Ka radiation (hn ¼ 1486.7 eV, 225 W, 15 mA, 15 kV). The binding energies were calibrated using the C 1s peak at 284.5 eV as a reference, and experimental data were tted with a Gaussian-Lorentzian mixed function as implemented in the Origin soware. H 2 -TPR (SO 2 -TPD) was performed using a Chemisorb instrument (Chembet Pulsar TPR/TPD 2139). These tests were conducted using a quartz U-type reactor, which was connected to a thermal conductivity detector. The module reductant gas was composed of 10 vol% H 2 balanced by Ar at a ow rate of 40 mL min À1 . Before the reduction, the sample (100 mg) was pretreated in a He stream at 300 C for 1 h, and then TPR was heated from room temperature to 900 C at a rate of 10 C min À1 . As for the TPD, before the test, the sample (100 mg) was pretreated in a He stream at 300 C for 1 h to eliminate surface impurities. The adsorption of SO 2 was performed at room temperature. In this process, the samples were rst treated with SO 2 (5 vol%, 40 mL min À1 ) for 2 h to reach saturation. Second, samples were puried using He (40 mL min À1 ) for 1 h, and then TPD was started at room temperature and heated to 1100 C at a rate of 5 C min À1 .
In situ DRIFTS spectra were recorded from 650 cm À1 to 4000 cm À1 at a spectral resolution of 4 cm À1 (number of scans, 100) on a Nicolet 6700 FTIR spectrophotometer equipped with a high-sensitivity MCT detector cooled by liquid N 2 . The DRIFTS cell (Pike) was tted with a ZnSe window and heating cartridge, which permits the heating of the sample to 500 C. Before the performance, all the samples were ground into ne powder (<2 mm) and diluted with KBr. The dilution factor was approximately 150. Then the powder (approximately 25 mg) was placed on a sample holder and carefully attened for IR reection. The sample was pretreated with a high-purity Ar stream at 400 C for 1 h to eliminate the physically absorbed water and other impurities. At each target temperature, the sample background was collected during cooling. For the steady state response, at each desired temperature, the sample was exposed to a controlled stream of 200 ppm SO 2 and/or 1000 ppm NO balanced by Ar at a ow rate of 100 mL min À1 for 0.5 h for saturation. For the transient response, the spectra were continuously collected in synchrony with the reaction time under each desire condition. The spectra were recorded at various target temperatures by subtracting the corresponding background reference.

Catalytic activity testing
The activity of the catalysts was investigated in a xed-bed reactor system, which consisted of a stainless steel tubular reactor (internal diameter: 12.7 mm), a gas supply and ow rate control unit (mass ow meter, Beijing Sevenstar Huachuang Electronics Co., Ltd.), a gas heating unit (Shandong Lulong furnace factory), and a gas analysis unit (GASMET DX4000, Finland). First, 2 g (approximately 5 cm 3 ) of the sample was loaded into the reactor and pretreated with N 2 at 300 C for 1 h to activate the samples and eliminate the surface impurities, followed by cooling to room temperature. The total ow rate of the mixed gas was 500 mL min À1 (GHSV ¼ 6000 h À1 ). The modeled ue gas was nitrogen, 1% NO balanced by N 2 , 2% CO balanced by N 2 , and 5000 ppm SO 2 balanced by N 2 (Deyang Gas, Ltd.). Tests under each reaction condition were completed aer more than 1 h, until a steady state had been reached, and the data were collected aer the outlet concentration had reached a steady state. The NO conversion and N 2 O selectivity were calculated from concentrations of the inlet and outlet ue gases using eqn (1).

Catalytic activity
DeNO efficiency is an important factor for evaluating the catalyst activity. Therefore, we determined the deNO efficiency of the prepared catalysts, and the results are displayed in Fig. S2 27 However, as the temperature increases, the N 2 O will be decomposed to N 2 . Although, the addition of cerium oxides can reduce the deNO efficiency, the SO 2 resistance of the catalysts doped by Ce increases. Fig. 1 shows the effect of SO 2 on the deNO activity. It can be found that for Fe 0.8 Co 0.2 Ce 0.05 /ASC, when 200 ppm of SO 2 is introduced into the dry ue gas, the NO conversion rst decreases and then increases: $80% / $70% / $95% (in Fig. 1(a)). The rst decreasing is speculated to be ascribed to the competitive adsorption between SO 2 and NO, according to the previous literature. Aer 1000 s of reaction, the NO conversion begins to decrease until reaching steady state at $55%; whereas, the SO 2 resistance for Fe 0.8 Co 0.2 Ce 0.1 /ASC and Fe 0.8 Co 0.2 Ce 0.2 /ASC are similar. Aer introduction of SO 2 , the NO conversion slightly decreases; then remains steady. For Fe 0.8 Co 0.2 Ce 0.1 /ASC, the steady conversion rate is approximately 80%, while that for Fe 0.8 Co 0.2 Ce 0.2 /ASC is $75%. When the Ce doped samples are compared with Fe 0.8 Co 0.2 /ASC ( Fig. 1(b)), it is easy to see that the doping of cerium oxides can improve the SO 2 tolerance.
As for the reason for the SO 2 tolerance improvement, it is speculated that a small amount of cerium oxide (Fe 0.8 Co 0.2 Ce 0.05 / ASC) may induce gaseous SO 2 to adsorb to its surface. The adsorbed SO x can generate some acid sites benecial for the reaction of NO + CO. Nevertheless, when the amount of adsorbed SO x increases to a threshold degree, surface sulfates may be produced that will result in the decrease of NO conversion. With the doping amount of cerium oxides increasing, the adsorbed SO x may transform to bulk-like sulfates 1,2 on CeO 2 that will have little inuence on the active components (Fe and Co). The evident mechanism will be discussed in the Characterization section.  Fig. 2(a), it can be observed that uniform spherical clusters are produced on the surface of Fe 0.8 Co 0.2 /ASC. With the doping of Ce, the sizes of the clusters begin to differentiate (Fig. 2(b)). When the molar ratio of cerium is $0.1, a very obvious accumulation phenomenon is observed. That is, the spherical clusters serry on the surface of ASC (Fig. 2(c)). Moreover, as the ratio is equal to 0.2, a block-like structure appears on the surface (Fig. 2(d)). It is speculated that with the increasing of the loaded metal oxides, the ACS surface structure will be occupied by the metal crystal. The reduction of the available surface area leads to an accumulation phenomenon. The results in surface morphology that is very similar to that in the N 2 physisorption. Table 2 summarizes the textural parameters of the prepared catalysts. It can be observed that the average pore size and surface area decrease obviously with increase in the loading amount of the cerium oxides. However, as the molar ratio of cerium is 0.1, the value of the average pore volume increases to 0.61 cm 3 g À1 . Maybe this is responsible for the improvement of the SO 2 tolerance.

XRD analysis of the prepared catalysts
In order to obtain the correlation of the SO 2 tolerance with the surface metal crystal, XRD was performed. Fig. 3 shows the XRD  Notably, it can be also observed that there are no characteristic peaks of cerium oxides when the molar ratio of Ce is 0.05. When the ratio increases to 0.1, a peak appears at approximately 28 , which should be assigned to the characteristic peak for the (111) crystal face of CeO 2 (JCPDF ¼ 82-0661). Moreover, the intensity of this peak increases to some extent with the loading amount of CeO 2 . As the ratio of Ce is 0.2, the sample shows the characteristic peaks for the (200), (220), and (311) crystal faces of CeO 2 . Comparing the degree of the intensity for the (110) crystal face of CoFe 15.7 , it can be observed that the introduction of Ce can reduce the intensity of this peak. That indicates the crystallization properties of CoFe 15.7 are diminished. This is speculated to be responsible for the slight decrease in the deNO activity.

H 2 -TPR and SO 2 -TPD analysis
H 2 -TPR is performed to investigate the reduction behavior of the catalyst surface components. Fig. 4 shows the TPR proles of the prepared catalysts. It can be observed that for all samples, there is a main peak at approximately 378 C. As we previously determined, this peak should be attributed to the interaction of Fe and Co oxides. 5,41,42 This result corresponds to the XRD analysis results (i.e., the major active component is CoFe 15.7 ). Additionally, along with the main peak, a shoulder peak is observed for each of the four samples. We can also see that the addition of Ce could change the position of this peak to a lower temperature. For Fe 0.8 Co 0.2 /ASC, the broad shoulder peak appears at approximate 520 C, while for Fe 0.8 Co 0.2 Ce 0.1 /ASC, the central temperature of this peak is about 450 C. It is speculated that the doping of Ce into the Fe-Co binary oxides can improve the activity of the low valent transition metal. However, for Fe 0.8 Co 0.2 Ce 0.2 /ASC, this shoulder peak is mixed with a higher temperature peak. This overlap peak can be decomposed into two peaks: one at approximate 450 C; the other at 625 C. For the other three samples, the highest temperature peak appears at $780 C. It is speculated that this peak is assignable to the complex components of solid carbon and metal oxides, while for Fe 0.8 Co 0.2 Ce 0.2 /ASC, the peak at 625 C should be ascribed to the excellent redox performance of CeO 2 . In conclusion, it can be seen that Ce doping can improve the reduction behavior of the low valent metal cation, which may be benecial for SO 2 tolerance. SO 2 -TPD is a signicant method for investigating the SO 2 resistance of CO-deNO catalysts. Fig. 5 shows the TPD proles of the prepared catalysts. It is observable that for Fe 0.8 Co 0.2 /ASC, a main peak appears at approximately 850 C. Along with the main peak, a broad weaker peak is also detected at $130 C. As reported, 43 the ferric sulfates has a decomposition temperature of 600-800 C, while the temperature for the cobalt sulfates is approximately 1200 C. According to our previous research, 5 the interaction between Fe and Co changes the position of the main peak to $850 C. The low temperature peak should be ascribed to desorption of coordinated SO 2 . Furthermore, for the other three samples, similar peaks are observed. However, the central temperatures of these peaks show some differentiation. It is easily seen that the position of the low temperature peak shis to the le with the introduction of Ce at approximately 100 C. This phenomenon indicates that CeO 2 on the surface could improve desorption of coordinated SO 2 . Besides, the intensity of the low temperature peak for Fe 0.8 Co 0.2 Ce 0.05 /ASC is stronger than that for other two catalysts, this demonstrated there is more amount of SO 2 adsorbing, i.e. the competitive adsorption of NO + SO 2 inevitably exists in the initial seconds. However, the high temperature peak shows a reverse trend (i.e., the central temperature shis to the right). For Fe 0.8 Co 0.2 Ce 0.05 /ASC, the temperature of the peak's maximum is $880 C, while this  temperature shis to $910 C for Fe 0.8 Co 0.2 Ce 0.1 /ASC. Nevertheless, when Ce molar ratio increases to 0.2, this temperature shis to 850 C. As reported in the literatures, 33,44,45 Ce(SO 4 ) 2 shows a decomposition peak at approximate 750 C, while Ce 2 (SO 4 ) 3 has decomposition peaks at 900 C. Therefore, we can speculate that when Ce is added into the catalyst, it will introduce some interactions among the metals. These interactions will produce different kinds of sulfates with SO 2 added into the ue gas. Compared the decomposition temperature of the sulfates, it is reasonable to demonstrate that the le shi is  attributed to the generation of Ce 2 (SO 4 ) 3 . The abnormal phenomenon for Fe 0.8 Co 0.2 Ce 0.2 /ASC is attributed to the interaction between Ce(SO 4 ) 2 and Ce 2 (SO 4 ) 3 , because on the surface of Fe 0.8 Co 0.2 Ce 0.2 /ASC, there should be more kinds of cerium sulfates. Owing to the excellent thermal stability, and rejection for SO 2 , Ce 2 (SO 4 ) 2 has been proved to improve SO 2 resistance of the catalysts. 2,33 Moreover, in the atmosphere of simulated ue gas, the generation of Ce 2 (SO 4 ) 3 occurs faster. Therefore, we speculate that on the surface of Fe 0.8 Co 0.2 Ce 0.1 /ASC, there is a larger amount of Ce 2 (SO 4 ) 3 . The rapidly generated Ce 2 (SO 4 ) 3 could inhibit the SO 2 poisoning of the active components.

XPS analysis
To demonstrate the opinion we obtained in Section 3.4, we carried out XPS on the surface components. Fig. 6-9 show the XPS results of Fe, Co, Ce, and S; and in addition, the percentages of the surface components are summarized in Table S3. † In these gures, the label "Fresh" indicates catalysts without pretreatment, the label "SO 2 -pretreated" represents samples purged with 200 ppm of SO 2 for 1 h at 250 C, and the label "SO 2 + NO" indicates samples pretreated by 200 ppm SO 2 + 1000 ppm NO for 1 h at 250 C. As shown in Fig. 6, we decomposed the overlapped peaks into three, viz. Fe 3+ at approximate 713.5 eV, Fe 2+ at $711 eV, and the satellite peak of Fe 2+ at $718.5 eV. [46][47][48] It can be observed that for the Fe 0.8 Co 0.2 /ASC, the percentage of the signicant component-Fe 3+ is about 46.49%. Moreover, this percentage rst increases and then shows a trend of decrease with increasing amount of Ce loaded. This is because even a small amount of Ce can induce an increase in the oxygen releasing performance, while a larger amount of Ce will generate more cerium oxide crystals. The producing of crystals can consume the oxidation performance. The SO 2 pretreatment of the catalysts results in a decrease of the Fe 3+ percentage for Fe 0.8 Co 0.2 /ASC, Fe 0.8 Co 0.2 Ce 0.05 /ASC, and Fe 0.8 Co 0.2 Ce 0.1 /ASC, but an increment for Fe 0.8 Co 0.2 Ce 0.2 /ASC. It is speculated that the sulfates on the surface can attract some electrons from FeO x . However, with increasing load of cerium, larger amounts of electrons will come from CeO 2 , due to its excellent redox performance.
The co-addition of SO 2 and NO to the reaction gas can induce an obvious decrease in the Fe 3+ percentage for all samples, and may demonstrate that the presence of NO can promote the generation of SO 4 2À . Similarly, the Co 2p 3/2 overlapped curves were also decomposed into three peaks: Co 3+ (783.5 eV), Co 2+ ($781 eV), and Co 2+ satellite peak ($788 eV). 49,50 However, the percentages of Co 3+ under different treatments remains stable for all samples (i.e., for the single sample, SO 2 or NO + SO 2 atmosphere has almost no inuence on the percentage of Co 3+ ). Otherwise, the addition of Ce can increase this percentage to some extent. Therefore, it is reasonable to establish that the poisoning mechanism of SO 2 to Fe-Co catalysts is the decomposition of Fe-Co crystal. The decomposition is ascribed to the interaction of Fe 3+ and SO 4 2À . Fig. 8 shows the Ce 3d curves of the prepared catalysts, and the binding energies of Ce 3d 5/2 and Ce 3d 3/2 are summarized in

DRIFTS analysis
To investigate the interaction of surface sulfates with NO or the surface metal ion, in situ DRIFTS was performed. First, we investigated the interaction of SO 2 and the catalyst surface. Fig. 10 shows the DRISTS spectra of the catalysts as a function of exposure time in a ow of 200 ppm SO 2 at 250 C. As shown in Fig. 10 ) coordinated with Fe 3+ or Co x+ . 33,36 Moreover, the bands at 1084 cm À1 , 1140 cm À1 and 1389 cm À1 grow in intensity with exposure time until they become stable; whereas, the  bands at 1535 cm À1 , 1638 cm À1 and 1709 cm À1 vary inconsistently. The intensity of the band at 1535 cm À1 rst decreases and then increases with time, while the positions of the other bands shi until the broad, strong peak appears at approximate 1680 cm À1 . It is speculated that during this variation, the surface sulfates evolve with the reaction time, in order to keep a steady state. Contrasting with the results in Fig. 10(a), Fe 0.8 -Co 0.2 Ce 0.1 /ASC shows fewer bands ( Fig. 10(b)): only four obvious bands at 1112 cm À1 , 1168 cm À1 , 1389 cm À1 and 1587 cm À1 . These bands will grow in intensity with time, but will not change positions. Therefore, we speculate that the sulfates on the surface of Fe 0.8 Co 0.2 Ce 0.1 /ASC are more puried (i.e., the major sulfates are cerous sulfates). The inuence of NO on the surface sulfates was also investigated, and Fig. 11 displays the spectra of the representative samples. It can be observed in Fig. 11(a), that there are some bands assigned to n of weakly adsorbed NO (1739 cm À1 ), n as (NO 3 À ) in bridge bidentate coordination (1632 cm À1 ), n as of nitro species (1383 cm À1 ), 8,15,18,51 and n of SO 4 2À (1532 cm À1 , 1114 cm À1 , 1068 cm À1 and 1002 cm À1 ). The intensity of these bands increases with time, indicating that the reaction gradually reaches a steady state. The bands of n as (nitro species) should theoretically be located at approximately 1375 cm À1 . This shiing is speculated to be related to the interaction of surface sulfates and NO 2 À . However, for Fe 0.8 Co 0.2 Ce 0.1 /ASC, the band of n as (nitro species) has almost no shiing. In addition to the band at 1378 cm À1 , bands also appear at 1698 cm À1 and 1123 cm À1 . The intensity and half peak width simultaneously increase with the adsorbing time, indicating some bulk-like sulfates may be produced.
In conclusion, it can be deduced that with no Ce doping, there will be more kinds of sulfates like Fe 2 (SO 4 ) 3 generated on the surface of Fe 0.8 Co 0.2 /ASC. However, the addition of Ce can attract these sulfates and transform them to cerous sulfates. Nitro acts as an important intermediate in the NO + CO reaction, 27 but the presence of SO 2 will affect the locating site of nitro species on the micro surface for Fe-Co binary catalysts. That will reduce the deNO activity. Meanwhile, the Ce addition can decrease the NO adsorption performance of the ASC-based catalysts in the presence of SO 2 , to some extent (this is also reasonable for the decrease of the NO conversion). Notably, cerium oxides show excellent performance for improving the generation of bulk-like cerous sulfates, which could remedy the decrease of CO-deNO activity, and further strengthen the SO 2 resistance.

SO 2 tolerance mechanism
Aer the analysis above, a possible scheme for the SO 2 resistance mechanism in the NO + CO reaction over ASC-based catalysts is proposed. As shown in Fig. 12, when the catalyst surface has no Ce, the adsorbed SO 2 will transform the active components to sulfate, and further deactivate the catalysts. Moreover, the SO 2 presence could also change the form of the intermediates of nitrates. However, Ce doping can improve the SO 2 tolerance in two aspects: one is protecting the active metal oxides; the other is stabilizing the form of the surface nitrates.

Conclusions
Aer a series of tests on the samples, the improvement in the SO 2 tolerance for Fe 0.8 Co 0.2 Ce 0.1 /ASC in the NO + CO reaction was revealed. The conclusions are as follows: (1) The presence of SO 2 in the feed gas can severely decrease the CO-deNO activity catalyzed by Fe-Co binary oxides over  activated semi-coke. The doping of Ce onto the Fe-Co binary catalysts will improve the SO 2 tolerance, and the optimizing doping molar ratio is approximately 0.1.
(2) The Ce on the surface of catalysts can improve the generation of Ce 2 (SO 4 ) 3 . These sulfates will directly accumulate on the Ce x+ site, affording some bulk-like sulfates, which can then protect the active metal from poisoning by SO 2 .
(3) Although the addition of Ce will decrease the NOadsorption performance of the Fe-Co binary catalysts, the presence of Ce plays an important role in protecting the formation of surface nitrates. Furthermore, this protection can alleviate the interaction of surface sulfates and nitrates.

Conflicts of interest
There are no conicts to declare.