A Study on Mn-Fe Catalysts Supported on Coal Fly Ash for Low-Temperature Selective Catalytic Reduction of NO X in Flue Gas

: A series of Mn 0.15 Fe 0.05 / ﬂy-ash catalysts have been synthesized by the co-precipitation method using coal ﬂy ash (FA) as the catalyst carrier. The catalyst showed high catalytic activity for low-temperature selective catalytic reduction (LTSCR) of NO with NH 3 . The catalytic reaction experiments were carried out using a lab-scale ﬁxed-bed reactor. De-NO x experimental results showed the use of optimum weight ratio of Mn / FA and Fe / FA, resulted in high NH 3 -SCR (selective catalytic reduction) activity with a broad operating temperature range (130–300 ◦ C) under 50000 h − 1 . Various characterization methods were used to understand the role of the physicochemical structure of the synthesized catalysts on their De-NO x capability. The scanning electron microscopy, physical adsorption-desorption, and X-ray photoelectron spectroscopy showed the interaction among the MnO x , FeO x , and the substrate increased the surface area, the amount of high valence metal state (Mn 4 + , Mn 3 + , and Fe 3 + ), and the surface adsorbed oxygen. Hence, redox cycles (Fe 3 + + Mn 2 + ↔ Mn 3 + + Fe 2 + ; Fe 2 + + Mn 4 + ↔ Mn 3 + + Fe 3 + ) were co-promoted over the catalyst. The balance between the adsorption ability of the reactants and the redox ability can promote the excellent NO x conversion ability of the catalyst at low temperatures. Furthermore, NH 3 / NO temperature-programmed desorption, NH 3 / NO- thermo gravimetric-mass spectrometry (NH 3 / NO-TG-MS), and in-situ DRIFTs (Di ﬀ use Reﬂectance Infrared Fourier Transform Spectroscopy) results showed the Mn 0.15 Fe 0.05 / FA has relatively high adsorption capacity and activation capability of reactants (NO, O 2 , and NH 3 ) at low temperatures. These results also showed that the Langmuir–Hinshelwood (L–H) reaction mechanism is the main reaction mechanism through which NH 3 -SCR reactions took place. This work is important for synthesizing an e ﬃ cient and environmentally-friendly catalyst and demonstrates a promising waste-utilization strategy.


Introduction
Coal, as a fuel, has played a vital role in the development of the world's economy. Coal fly ash production resulting from coal combustion is ecologically problematic, and every year millions of tons of fly ash are produced [1]. Also, the emission of environmentally harmful gases, e.g., NO x , has been a global concern, where coal combustion is responsible for most of the emissions. In 2017, the NO x emissions of China alone reached 12.59 million tons, with 67.6% from stationary sources (China Statistical Yearbook 2019). High concentrations of NO x gases in the atmosphere is environmentally

The Morphology and Structure of Catalysts
The physical structure of the catalysts was studied by SEM technique and the N 2 physisorption analysis. The results are given in Figures S1 and S2 (summarized in Table 1). These results correlate to a great extent. From the SEM images ( Figure S1) of the various catalysts, a large number of debris or/and cracks can be observed on the surface. FA had a smooth surface with minimum irregularities resulting in low surface area (7.7 m 2 ·g −1 ) and total pore volume (0.1 cm 3 ·g −1 ). Fe 0.10 /FA had larger openings or holes on the surface. Its surface area and total pore volume reached 35.8 m 2 ·g −1 and 0.1 cm 3 ·g −1 , respectively. The presence of micro-particles on Mn 0.15 /FA surface resulted in a high surface area value of 78.0 m 2 ·g −1 . For Mn x Fe 0.10 /FA, when the loading amount of Mn was lower than 0.15, more pores were observed. However, when Mn content was further increased to 0.20, dense and bonding pores were formed. For Mn 0.15 Fe y /FA, when the loading amount of Fe/FA was lower than 0.1, less porous structures were formed. The increase of FeO x led to the formation of a highly porous structure. In comparison, the surface morphology of Mn 0.15 Fe 0.05 /FA varied significantly with the growth of numerous screw-like structures appearing on the surface, as shown in Figure S1g, resulting in a high surface area value of 91.6 m 2 ·g −1 and a total pore volume of 0.15 cm 3 ·g −1 . The interaction between MnO x , FeO x , and the carrier changes the micro-morphology and the physical structure of the samples [23,24]. The N 2 physisorption analysis was used to understand the change of the physical structure of the catalysts caused by metal oxides. The N 2 adsorption/desorption isotherms of different samples are given in Figure S2. Notably, the curves of all samples are following the type IV isotherms with type H3 hysteresis loops, which corresponds to the micro-and mesoporous structure [25,26]. The presence of micro-and mesoporous structure is evident in Figure S2, indicating that the majority of the pores have an opening pore diameter of 6 nm. The main factor contributing to the formation of these microand mesoporous structures is the diffusion of iron oxides and manganese oxide into the lattice of the carrier during the calcination process [23]. These mesoporous materials provide abundant active sites for the adsorption of reactants, which facilitates the mass transfer process at the gas-solid phase boundary [26]. Efficient mass transfer plays a vital role during the bulk reaction and lowering the energy required for the NH 3 -SCR reaction. The N 2 desorption capacity of all samples followed the order of Mn 0. 15  oxide, the internal structure of samples became more copious than that of the FA, and the larger specific surface area of Mn 0.15 Fe 0.05 /FA provides more active sites than others. Figure 1 shows the catalytic activities of Mn x Fe y /FA catalysts at different temperatures. Where x is the weight ratio of Mn to FA, and y is the weight ratio of Fe to FA. The activity of Fe 0.10 /FA by Mn doping with "x" ranging from 0.05 to 0.2 at a GHSV of 50,000 h −1 is shown in Figure 1a. The as-obtained FA showed a weak De-NO x activity at the designed experimental temperatures, with the highest De-NO x efficiency of ∼23.95% recorded at 300 • C. This is due to the natural structure and surface property of FA [15], which will be demonstrated in the following sections. After the addition of Fe (weight ratio of 0.10), slightly higher De-NO x activity was recorded. These results indicate that the introduction of Fe did not contribute much to the De-NO x activity of Fe 0.10 /FA. However, the introduction of Mn into Fe 0.10 /FA notably improved its SCR activity, and the graphed NO x conversion data of all the samples show a "volcanic" shape ( Figure 1a). As the ratio of Mn was increased to 0.15, De-NO x activity increased, and further addition of Mn led to a decrease in conversion efficiency. De-NO x efficiency of different samples increased as follows: FA < Fe 0.10 /FA < Mn 0.05 Fe 0.10 /FA < Mn 0.20 Fe 0.10 /FA < Mn 0.10 Fe 0.10 /FA < Mn 0.15 Fe 0.10 /FA. The temperature range in which the NO x conversion efficiency reaches 80% (T 80 ) was considered as a primary indicator, and when conversion efficiency has reached 50% (T 50 ), that temperature is considered as the activation temperature of NH 3 -SCR activity. Among these samples, the catalytic activity of Mn 0.15 Fe 0.10 /FA shows high conversion efficiency, with the lowest T 50 at 108 • C and the broadest T 80 at 145-300 • C. Indicating that Mn 0.15 Fe 0.10 /FA exhibits great potential as an LTSCR catalyst (100-250 • C), and the mass ratio of Mn/FA was fixed at 0.15.  Figure 1 shows the catalytic activities of MnxFey/FA catalysts at different temperatures. Where x is the weight ratio of Mn to FA, and y is the weight ratio of Fe to FA. The activity of Fe0.10/FA by Mn doping with "x" ranging from 0.05 to 0.2 at a GHSV of 50,000 h −1 is shown in Figure 1a. The asobtained FA showed a weak De-NOx activity at the designed experimental temperatures, with the highest De-NOx efficiency of ∼23.95% recorded at 300 °C. This is due to the natural structure and surface property of FA [15], which will be demonstrated in the following sections. After the addition of Fe (weight ratio of 0.10), slightly higher De-NOx activity was recorded. These results indicate that the introduction of Fe did not contribute much to the De-NOx activity of Fe0.10/FA. However, the introduction of Mn into Fe0.10/FA notably improved its SCR activity, and the graphed NOx conversion data of all the samples show a "volcanic" shape ( Figure 1a). As the ratio of Mn was increased to 0.15, De-NOx activity increased, and further addition of Mn led to a decrease in conversion efficiency. De-NOx efficiency of different samples increased as follows: FA < Fe0.10/FA < Mn0.05Fe0.10/FA < Mn0.20Fe0.10/FA < Mn0.10Fe0.10/FA < Mn0.15Fe0.10/FA. The temperature range in which the NOx conversion efficiency reaches 80% (T80) was considered as a primary indicator, and when conversion efficiency has reached 50% (T50), that temperature is considered as the activation temperature of NH3-SCR activity. Among these samples, the catalytic activity of Mn0.15Fe0.10/FA shows high conversion efficiency, with the lowest T50 at 108 °C and the broadest T80 at 145-300 °C. Indicating that Mn0.15Fe0.10/FA exhibits great potential as an LTSCR catalyst (100-250 °C), and the mass ratio of Mn/FA was fixed at 0.15.

Catalytic Reactivity during Low-Temperature SCR Reactions
Then, the ratio of FeOx in the Mn0.15Fey/FA catalyst was varied and optimized to improve its efficiency and obtain the most efficient LTSCR catalyst. The De-NOx activity of Mn0.15Fey/FA is given in Figure 1b. Compared with Mn0.15/FA, the NOx efficiency of all catalysts was further improved at the desired temperatures (100-300 °C). The conversion efficiency improved as the Fe content increased to 0.05, and then declined with a further increase in the ratio of Fe. The best activity was obtained from the Mn0.15Fe0.05/FA during the whole operating temperature range, which has the lowest T50 (100 °C) and the broadest T80 (130-300 °C). Moreover, over 90% of NOx conversion was obtained between 147 and 300 °C. The final weight ratio of FeOx (0.05) and MnOx (0.15) achieved a high catalytic activity. In comparison (Table 2), the Mn0.15Fe0.05/FA has been successfully synthesized by a straightforward method. The stability results of the catalyst are shown in Figure 1c. It can be seen that after 4 hours and three cycles, the Mn0.15Fe0.05/FA catalyst exhibited excellent stability with long stable catalytic activity and good regenerability.   Then, the ratio of FeO x in the Mn 0.15 Fe y /FA catalyst was varied and optimized to improve its efficiency and obtain the most efficient LTSCR catalyst. The De-NO x activity of Mn 0.15 Fe y /FA is given in Figure 1b. Compared with Mn 0.15 /FA, the NO x efficiency of all catalysts was further improved at the desired temperatures (100-300 • C). The conversion efficiency improved as the Fe content increased to 0.05, and then declined with a further increase in the ratio of Fe. The best activity was obtained from the Mn 0.15 Fe 0.05 /FA during the whole operating temperature range, which has the lowest T 50 (100 • C) and the broadest T 80 (130-300 • C). Moreover, over 90% of NO x conversion was obtained between 147 and 300 • C. The final weight ratio of FeO x (0.05) and MnO x (0.15) achieved a high catalytic activity. In comparison (Table 2), the Mn 0.15 Fe 0.05 /FA has been successfully synthesized by a straightforward method. The stability results of the catalyst are shown in Figure 1c. It can be seen that after 4 hours and three cycles, the Mn 0.15 Fe 0.05 /FA catalyst exhibited excellent stability with long stable catalytic activity and good regenerability.

Surface Chemical States
The XRD (X-ray Diffraction) results of the Mn x Fe y /FA catalysts are shown in Figure S3. For FA, only the diffraction peaks of SiO 2 and weaker Fe 2 O 3 were observed. As different ratios of Mn and Fe oxide were doped, no additional peaks besides those of FA were detected. However, ICP-OES (Inductively Coupled Plasma Optical Emission spectroscopy) results (Table 1) show that the mass content of Fe and Mn within all samples is approximately equal to the calculated values, which indicates that the added metal oxide is present in the catalyst. XRD results also showed that the intensity of peaks and the average crystal size assigned to SiO 2 decreased with Mn and Fe doping, and the peak of SiO 2 slightly shifted towards lower 2θ values. The average crystal size of SiO 2 was calculated based on the diffraction peak (1,0,0) according to the Scherrer formula (Table 1). These results indicated that the presence of defective lattice and a decrease in the crystalline size of SiO 2 , which were caused by the introduction of Mn and Fe oxide. This result also shows the strong interaction between FeO x /MnO x and the carrier. The defective lattice and the interaction between metal oxides and the carrier led to a more unbalance charge state and oxygen vacancy, providing abundant active sites during the SCR process and contributing to the higher NO conversion ability of the catalyst [5]. The valence state of the added metals is of significant importance in the SCR reaction. To further examine the surface chemical states and the types of ion species over catalysts, XPS (X-ray photoelectron spectroscopy) measurement of Fe 2p, Mn 2p, and O 1s in fresh catalysts were conducted ( Table 1). The results indicate that the surface concentration of Mn is slightly higher than that in bulk catalysts (from the ICP-OES results), while the concentration of Fe shows the opposite phenomenon. This indicates that Fe could promote the enrichment of Mn on the surface, providing more active sites during low-temperature De-NO x by NH 3 . The recorded spectra of different elements were fitted into multiple sub-bands according to the multivalent oxide species on the surface by Gaussian-Lorentz fitting of XPS Peak 4.1, and examples of a peak-fitted graph are given in Figure S4. The relative atomic ratios of different elemental states are listed in Table 3. The overlapped Mn 2p peaks were deconvoluted into three pairs of peaks. The fitted peaks are assigned to Mn 4+ (643.58 and 654.71 eV), Mn 3+ (642.26 and 653.6 eV), and Mn 2+ (641.21 and 652.5 eV). Figure S4a shows that three valence states of Mn species within the series of Mn x Fe y /FA vary significantly with different ratios of MnO x and FeO x . It was found that the ratio of Mn 4+ /Mn n+ first increased and then decreased, and the ratio of Mn 3+ /Mn n+ slightly increased as the ratios of MnO x and FeO x increased. Mn 0.15 Fe 0.05 /FA has the highest ratio of Mn 4+ /Mn n+ value (39.59%) and a high ratio of Mn 3+ /Mn n+ (48.38%) within all samples. Among the catalysts, most of the Mn exists at high valence states (Mn 3+ , Mn 4+ ), which is caused by the oxidation of lower valence state (Mn 2+ ) contemporary leading to an electronic transfer, such as Mn 4+ ↔ Mn 3+ ↔ Mn 2+ . The order of the LTSCR activity from high to low is as follows: [30]. Furthermore, Mn has a high self-redox tendency leading to the formation of a large amount of higher valence state MnO x [31]. Many studies have indicated that Mn 4+ is the main factor in promoting the oxidation of NO to NO 2 during LTSCR. The co-existence of NO 2 and NH 3 in one circle can promote the LTSCR process through "fast reaction" [32]: The Fe 2p spectra were deconvoluted into different valence states of Fe: Fe 3+ (~725.1 and~711.8 eV), Fe 2+ (~723.5 and 710.1), and the satellite peak located at 718.5 eV [33]. Table 3 shows that most Fe species exist in the valence state of Fe 3+ . With the addition of MnO x into Fe 0.10 /FA, the ratio of Fe 3+ /Fe increased first and then decreased when the value of Mn/FA was higher than 0.15. However, when FeO x was added to Mn 0.15 /FA, the ratio of Fe 3+ /Fe also increased. This phenomenon indicates that the ratio of Mn/Fe is a critical factor in determining the catalytic capability of a catalyst during LTSCR reactions, and the optimum ratio of Mn/Fe could promote the cycle of redox as follows: The transfer of electrons creates an unbalanced charge and underfilled chemical bonds, making it easier to produce oxygen vacancies and highly mobile oxygen. Highly mobile oxygen that is chemically adsorbed on the surface is one of the critical active species during the NH 3 -SCR reaction [24]. The O 1s spectra were fitted with three sub-peaks [34]. The peak at~529.9 eV represents the lattice oxygen (named as O β ). Defective metal-oxide (~531.4 eV) and chemisorbed water (~532.7 eV (OH − or H 2 O)) [35,36] are assigned to the chemically absorbed oxygen species on the surface (O α ). Table 3 shows that the chemically absorbed oxygen is the main form of oxygen species and is more reactive due to higher mobility, lower bonding energy to the surface, and a higher tendency to form NO 2 by oxidizing NO; which could promote "fast reaction" [37]. Furthermore, chemically adsorbed oxygen can activate oxygen gas [38], providing sufficient active oxygen species, ensuring fast recovery from low-state metal ion: [39], which facilitate the excellent LTSCR.
Among all the synthesized catalysts, the active components within Mn 0.15 Fe 0.05 /FA possess the lowest energy value (Table S1), caused by high electron cloud density. This phenomenon may have been caused by two factors: (i) The interaction between Mn and Fe could increase the redox cycle ability, which is significantly vital to activate the reactant; (ii) the co-existence of active metal species with a different state could provide unsaturated and unbalanced chemical bonds, which enhances NO x removal [40]. Mn 0.15 Fe 0.05 /FA also has a high ratio of O α /O t , acting as an active site for NH 3 adsorption to form. The high valence state metal (Mn 4+ , Mn 3+ and Fe 3+ ) provides active sites for the reactant adsorption and activation and promotes the redox cycle combined with active oxygen species explains the excellent NH 3 -SCR activity of Mn 0.15 Fe 0.05 /FA.

Reducibility for the Low-Temperature Activity
The existence of active species with different valence states within catalyst promotes the oxidation/reduction cycle proven by XPS results (Table 3). Thus, the redox property is an important index to evaluate the De-NO x ability of the catalyst. The H 2 -TPR experiments were performed to examine the reducibility of different catalysts (Figure 2a,b). The H 2 -TPR curve of FA had a broad peak at around 400-650 • C, attributed to the reduction of FeO x inherently present in FA [41]. When metal oxides were added, the intensity of this peak increased gradually and shifted to a higher temperature. This result is caused by electronic transfer and healthy interaction between metal oxides and carrier, as indicated by XRD results that the metal oxide was doped into the SiO 2 crystal structure. Fe 0.10 /FA produced two main peaks at around 419 and 572 • C that were associated with the reduction of Fe 2 O 3 to Fe 3 O 4 , and Fe 3 O 4 to FeO [41], respectively. Three peaks ((α) 332, (β) 429, and (δ) 488 • C) appeared in Mn 0.15 /FA curve (Figure 2b), responsible for the reduction of the manganese oxide species [42]. All bimetal oxides containing catalysts produced peaks in three zones and were denoted as α, β, and δ, respectively. Based on a previous study [43], these zones were assigned to MnO 2 → Mn 2 O 3 (α), The primary objective of this work is to understand the reducibility of NO x by the catalysts at lower temperatures. Therefore, the first two zones were considered (Figure 2a,b). Furthermore, semi-quantitative H 2 -TPR analysis was conducted by calculating the peak area, and the results are given in Figure 3. For Mn x Fe y /FA, the onset of reduction shifts to lower temperatures increasing the peak area of all peaks with the addition of MnO x or FeO x . This phenomenon is caused by the interaction of MnO x , FeO x, and the carrier within the catalysts. Figure 3 shows that with the increase in MnO x or FeO x concentration, the peak area responsible for low-temperature reducibility (α) gradually increased. Mn 0.15 Fe 0.10 /FA and Mn 0.15 Fe 0.05 /FA have the best De-NO x activity within the series of the Mn x Fe 0.10 /FA and the Mn 0.15 Fe y /FA, respectively. These two catalysts have medium reducibility. Therefore, the medium reducibility of Mn 0.15 Fe 0.05 /FA is an essential factor contributing to its outstanding De-NO x activity.
Generally, the adsorption of gas reactants is the first step for a gas-solid reaction. Furthermore, SCR reactions are well established to be involved in both redox and adsorption active sites. To examine the NH 3 and NO adsorption and activation ability of the catalyst, temperature program desorption, in-situ DRIFTs, and TG-MS techniques were used. The primary objective of this work is to understand the reducibility of NOx by the catalysts at lower temperatures. Therefore, the first two zones were considered (Figure 2a,b). Furthermore, semiquantitative H2-TPR analysis was conducted by calculating the peak area, and the results are given in Figure 3. For MnxFey/FA, the onset of reduction shifts to lower temperatures increasing the peak area of all peaks with the addition of MnOx or FeOx. This phenomenon is caused by the interaction of MnOx, FeOx, and the carrier within the catalysts. Figure 3 shows that with the increase in MnOx or FeOx concentration, the peak area responsible for low-temperature reducibility (α) gradually increased. Mn0.15Fe0.10/FA and Mn0.15Fe0.05/FA have the best De-NOx activity within the series of the MnxFe0.10/FA and the Mn0.15Fey/FA, respectively. These two catalysts have medium reducibility. Catalysts 2020, 10, x FOR PEER REVIEW 9 of 18 Therefore, the medium reducibility of Mn0.15Fe0.05/FA is an essential factor contributing to its outstanding De-NOx activity. Generally, the adsorption of gas reactants is the first step for a gas-solid reaction. Furthermore, SCR reactions are well established to be involved in both redox and adsorption active sites. To examine the NH3 and NO adsorption and activation ability of the catalyst, temperature program desorption, in-situ DRIFTs, and TG-MS techniques were used.

The Adsorption and Desorption Behavior of NH3/NO
The surface acid property of the catalyst is highly essential for NH3 adsorption on acidic sites. This is the first step of the NH3-SCR reaction. Thus, the strength and number of acidic sites were calculated by NH3-TPD technology. Figure 2c,d shows that all the NH3 desorption profiles showed one broad peak, including two desorption processes [44]: (1) NH3 desorption by weak acid sites (100-200 °C) and (2) NH3 desorption peak (200-400 °C) attributed to medium acid sites. The peak area allows for a semi-quantitative understanding of acid sites, and the results are shown in Figure 3. The carrier can provide a small number of acid sites because FeOx is inherently present in FA. For all catalysts, the total acidic sites of Fe0.10/FA is the lowest but is the highest for Mn0.15/FA. However, bimetal-containing catalysts have similar total acidic sites. Mn0.15Fe0.05/FA has a slightly lower number of weak acidic sites, but it has the highest de-NOx activity. It can be concluded from this result that mediocre surface acid property is just one of the factors influencing De-NOx activity because excess and heavy adsorption of NH3 could have an inhibitory effect for De-NOx efficiency [45]. Figure 2e,f shows the temperature-programmed desorption of NOx to test the adsorption capacity of NO. All samples present a broad desorption peak at the temperature range of 50-400 °C. This temperature range can be divided into two segments: (1) At temperatures lower than 250 °C which is the desorption of physisorbed NO and disintegration of unstable nitrite species occurs, and (2) temperatures higher than 250 °C represents NO2 desorption due to the decomposition of stable thermal nitrates, such as bridged nitrate or bidentate bitrate that have high thermal stability [39]. The area of desorption peak could represent the semi-quantitative amount of NOx species, and the results are shown in Figure 3. For all catalysts, the desorption amount of NO and NO2 increased with MnOx addition. In contrast, the desorption amount of NO increased and then decreased as the concentration

The Adsorption and Desorption Behavior of NH 3 /NO
The surface acid property of the catalyst is highly essential for NH 3 adsorption on acidic sites. This is the first step of the NH 3 -SCR reaction. Thus, the strength and number of acidic sites were calculated by NH 3 -TPD technology. Figure 2c,d shows that all the NH 3 desorption profiles showed one broad peak, including two desorption processes [44]: (1) NH 3 desorption by weak acid sites (100-200 • C) and (2) NH 3 desorption peak (200-400 • C) attributed to medium acid sites. The peak area allows for a semi-quantitative understanding of acid sites, and the results are shown in Figure 3. The carrier can provide a small number of acid sites because FeO x is inherently present in FA. For all catalysts, the total acidic sites of Fe 0.10 /FA is the lowest but is the highest for Mn 0.15 /FA. However, bimetal-containing catalysts have similar total acidic sites. Mn 0.15 Fe 0.05 /FA has a slightly lower number of weak acidic sites, but it has the highest de-NO x activity. It can be concluded from this result that mediocre surface acid property is just one of the factors influencing De-NO x activity because excess and heavy adsorption of NH 3 could have an inhibitory effect for De-NO x efficiency [45].
Figure 2e,f shows the temperature-programmed desorption of NO x to test the adsorption capacity of NO. All samples present a broad desorption peak at the temperature range of 50-400 • C. This temperature range can be divided into two segments: (1) At temperatures lower than 250 • C which is the desorption of physisorbed NO and disintegration of unstable nitrite species occurs, and (2) temperatures higher than 250 • C represents NO 2 desorption due to the decomposition of stable thermal nitrates, such as bridged nitrate or bidentate bitrate that have high thermal stability [39]. The area of desorption peak could represent the semi-quantitative amount of NO x species, and the results are shown in Figure 3. For all catalysts, the desorption amount of NO and NO 2 increased with MnO x addition. In contrast, the desorption amount of NO increased and then decreased as the concentration of FeO x increased; the desorption amount of NO 2 increased continuously. Therefore, it can be concluded that the balance between the adsorption of reactants (NH 3 and NO x ) and the redox ability of metal oxide plays a significant role in LTSCR.
To further investigate the decomposition characteristics of surface adsorbed species over Mn 0.15 Fe 0.05 /FA, the pre-adsorbed catalyst with NH 3 was studied with TG-MS ( Figure S5). TG thermograms ( Figure S5a) show a slight weight loss caused merely by the desorption of gas because the sample was calcinated at 500 • C for 3 h. Mn 0.15 Fe 0.05 /FA can readily adsorb NH 3 at 50 • C, and then release NH 3 species during the entire heating process, including NH 3 , NH 2 , NH 4 + , and N 2 O.
When the temperature increased, the NH 3 was released. Because some of the NH 3 was physically adsorbed and/or weakly chemically adsorbed on the Lewis acid sites, nevertheless, some ammonia species are chemically adsorbed on the active sites of Mn 0. 15 [48]. Figure S6a shows the result of a mass change of pre-adsorbed NO on Mn 0.15 Fe 0.05 /FA, and the desorbed species were recorded by TG-MS ( Figure S6b-d), such as H 2 O, NO, and NO 2 . The TG thermogram ( Figure S6a) shows two weight loss stages. The first stage at around 100 • C was assigned to dehydration and was recorded by MS ( Figure S6b). The NO was released during all the weight loss stages, which the weakly adsorbed NO was released on the catalyst surface ( Figure S6c). NO 2 was disrobed in the second stage ( Figure S6d), which was caused by gaseous NO reacting with the surface chemical oxygen to produce NO 2 over Fe n+ and Mn n+ sites at a lower temperature [25,49], and promotes the De-NO x activity by "fast reaction". This reaction pathway is one of the reasons Mn 0.15 Fe 0.05 /FA has an excellent low-temperature activity.

In-Situ DRIFT Analysis
SCR catalytic performance is highly dependent on the adsorption capability of the catalyst. Gases such as O 2 , NO, and NH 3 are primary participants during SCR reaction. The adsorption capability of the catalyst was discussed in detail in the previous section. However, it is essential to understand the reactions occurring on the surface of the catalyst after the adsorption of primary participants. To understand the mechanism of the reactions on the surface of Mn 0.15 Fe 0.05 /FA catalyst by adsorption and reaction of reactants at active sites, the intermediates, and adsorbed species, four different experiments were conducted using the in-situ DRIFT at 150 • C: (1) NH 3 adsorption; (2) NO + O 2 adsorption; (3) the reaction of NO + O 2 with pre-adsorbed NH 3 ; and (4) the reaction of NH 3 with pre-adsorbed NO + O 2 .
The NH 3 adsorption spectra are given in Figure 4a. Several peaks appeared immediately when the NH 3 was introduced into the reaction cell after 1 min, such as 1621, 1405, 1349, 1229, 966, and 929 cm −1 . This phenomenon shows that Mn 0.15 Fe 0.05 /FA catalyst has a strong surface acidity, which can make NH 3 readily adsorbed and then facilitate the De-NO x reaction. During the 30 min adsorption of NH 3 , many adsorption peaks gradually increased in intensity. Studies show that these peaks can be divided into four groups: (1) The strong bands centered at 1621, 1270, 1227, and 1086 cm −1 were assigned to symmetric-asymmetric bending vibrations of the N-H bonds within NH 3 coordinately associated with Lewis acid sites [50]; (2) the 1694 and 1405 cm −1 weak bands were attributed to the N-H bending vibration of NH 4 + species on Brønsted acid sites [51]; (3) the two bands at 1530 and 1349 cm −1 can represent −NH 2 species [52], which were formed by the intermediate oxidation of adsorbed ammonia species. From the results of Figure S5, the formation of the activated amide (−NH 2 ) groups from dehydrogenation of NH 3 is evident [46]; (4) the groups centered at 966 and 929 cm −1 were assigned to loosely adsorbed NH 3 [44], indicating that NH 3 can be readily adsorbed and activated on Mn 0.15 Fe 0.05 /FA.
[53], formed as a result of NO reacting with O2 and/or the reaction of NO and chemisorbed oxygen [54]. The bands at 1438, 1337, and 1274 cm −1 were assigned to bridged nitrate, monodentate nitrate, and bidentate nitrate [54]. The bands at 1379 and 1048 cm −1 are responsible for trans-and cis-N2O2 [47]. When the pre-adsorbed (NO + O2) catalyst was purged by nitrogen, most of the peaks appeared weaker than before, indicating that the adsorbed NOx species are easily formed and decomposed, facilitating the cycle of nitro compounds, making a good foundation on the reaction. The in-situ DRIFT spectra of the reaction between NO + O2 and pre-adsorbed NH3 at different time intervals are shown in Figure 5a. As mentioned above in (Figure 4a), after the adsorption of NH3, the catalyst exhibits Lewis (L) acid sites related bands (1621, 1269, 1227, and 1086 cm −1 ), Brønsted (B) acid sites (1694 and 1405 cm −1 ), loosely adsorbed NH3 (966 and 929 cm −1 ) and -NH2 (1530 and 1349 cm −1 ). When NO + O2 mixture was introduced, both the acid sites (L and B) and active intermediate bands decreased immediately, and many new nitrate species bands appeared and increased gradually during the reaction process. These changes indicated that almost all of the adsorbed NH3 species could react with NO + O2 to form new species, such as bridged nitrate (1616 cm −1 ), bidentate nitrate (1696, 1568, 1270, and 1102 cm −1 ) [55], monodentate nitrate (1355 cm −1 ), trans-and cis-N2O2 2-(1403 and 1031 cm −1 ), unstable nitrate species (1227 cm −1 ) [56]. Over time, these peaks became more pronounced. However, the intensity of loosely adsorbed NH3 (967 and 929 cm −1 ) slightly decreased, which indicates that it reacted with NOx species and formed a small number of nitrate species.
The other transient reaction experiment was also carried out. In this reaction, NO + O2 was coadsorbed at first, and then NH3 was added. The results are given in Figure 5b. As observed, before the addition of NH3, the bands of NOx species (NO2 (1627 cm −1 ), bridged nitrate (1438 cm −1 ), monodentate nitrate (1337 cm −1 ), bidentate nitrate (1274 cm −1 ), trans-and cis-N2O2 2− (1379 and 1048 cm −1 ) were identical to the previous experiment shown in Figure 5b. After the addition of NH3, the following phenomena were observed: (1) the intensity of the peak assigned to the adsorbed NO2 (NO oxidation by active oxygen-species) and nitrate species decreased immediately; (2) some of the NH3 species (NH3 coordinate to Lewis acid sites (1085 cm −1 ), trans-N2O2 2− (1379 cm −1 ) and loosely adsorbed NH3 (967 and 929 cm −1 )) appeared, and the intensity increased with time; (3) formation of new nitrate   [53], formed as a result of NO reacting with O 2 and/or the reaction of NO and chemisorbed oxygen [54]. The bands at 1438, 1337, and 1274 cm −1 were assigned to bridged nitrate, monodentate nitrate, and bidentate nitrate [54]. The bands at 1379 and 1048 cm −1 are responsible for trans-and cis-N 2 O 2 [47]. When the pre-adsorbed (NO + O 2 ) catalyst was purged by nitrogen, most of the peaks appeared weaker than before, indicating that the adsorbed NO x species are easily formed and decomposed, facilitating the cycle of nitro compounds, making a good foundation on the reaction.
The in-situ DRIFT spectra of the reaction between NO + O 2 and pre-adsorbed NH 3 at different time intervals are shown in Figure 5a. As mentioned above in (Figure 4a), after the adsorption of NH 3 , the catalyst exhibits Lewis (L) acid sites related bands (1621, 1269, 1227, and 1086 cm −1 ), Brønsted (B) acid sites (1694 and 1405 cm −1 ), loosely adsorbed NH 3 (966 and 929 cm −1 ) and -NH 2 (1530 and 1349 cm −1 ). When NO + O 2 mixture was introduced, both the acid sites (L and B) and active intermediate bands decreased immediately, and many new nitrate species bands appeared and increased gradually during the reaction process. These changes indicated that almost all of the adsorbed NH 3 species could react with NO + O 2 to form new species, such as bridged nitrate (1616 cm −1 ), bidentate nitrate (1696, 1568, 1270, and 1102 cm −1 ) [55], monodentate nitrate (1355 cm −1 ), trans-and cis-N 2 O 2 2− (1403 and 1031 cm −1 ), unstable nitrate species (1227 cm −1 ) [56]. Over time, these peaks became more pronounced. However, the intensity of loosely adsorbed NH 3 (967 and 929 cm −1 ) slightly decreased, which indicates that it reacted with NO x species and formed a small number of nitrate species. The other transient reaction experiment was also carried out. In this reaction, NO + O 2 was co-adsorbed at first, and then NH 3 was added. The results are given in Figure 5b. As observed, before the addition of NH 3 , the bands of NO x species (NO 2 (1627 cm −1 ), bridged nitrate (1438 cm −1 ), monodentate nitrate (1337 cm −1 ), bidentate nitrate (1274 cm −1 ), trans-and cis-N 2 O 2 2− (1379 and 1048 cm −1 ) were identical to the previous experiment shown in Figure 5b. After the addition of NH 3 , the following phenomena were observed: (1) the intensity of the peak assigned to the adsorbed NO 2 (NO oxidation by active oxygen-species) and nitrate species decreased immediately; (2) some of the NH 3 species (NH 3 coordinate to Lewis acid sites (1085 cm −1 ), trans-N 2 O 2 2− (1379 cm −1 ) and loosely adsorbed NH 3 (967 and 929 cm −1 )) appeared, and the intensity increased with time; (3) formation of new nitrate species such as bridged nitrate (1617, 1458 cm −1 ), bidentate nitrate (1568 and 1538 cm −1 ), monodentate nitrate (1337 cm −1 ), unstable nitrate species (1242 cm −1 ) and trans-N 2 O 2 2− (1379 cm −1 ).
Among these, the bands of monodentate nitrates (1337 cm −1 ) were always present and increased with time, suggesting that the monodentate nitrates did not react with ammonia species, and the other NO x species (bidentate nitrate and trans-N 2 O 2 2− ) reacted with NH 3 to form monodentate nitrates. However, the bands of weakly adsorbed NO 2 (1627 cm −1 ), bidentate nitrate (1274 cm −1 ) and cis-N 2 O 2 2− (1048 cm −1 ) quickly vanished, indicating that they reacted with the introduced NH 3 .

Mechanisms and Reaction Pathways
NH3/NO-TPD and pre-adsorbed NH3/NO-TG-MS results showed that Mn0.15Fe0.05/FA could quickly adsorb and activate the reactants at 150 °C, which is promoted by the addition of a proper amount of metal ion providing the optimum amount of acid and active sites. The primary function of active sites is to adsorb and activate the reactants (NH3, NOx, and O2) to produce the NH3 species and nitrates. Moreover, in-situ DRIFTs analysis showed that NH3 could be easily oxidized to NH2 as an active intermediate were adsorbed on Lewis and Brønsted acid sites. With the introduction of NO + O2, most of the absorbed ammonia species decreased, and new species were formed (Figure 5a). The highly reactive NH2 species reacted with NOx (NO, NO2) to form NH2NO or NH2NO2, which rapidly discomposed into N2 and H2O [57]. As given in Figure 5b, the adsorbed NO2 or nitrates species reacts with adsorbed ammonia species when the NH3 was confirmed to follow the Langmuir-Hishelwood (L-H) mechanism.
According to several previous studies [58] combined with our findings, the L-H reaction pathway mainly occurs during the reaction as follows: (M denotes Fe and/or elemental Mn)  (Figure 5a). The highly reactive NH 2 species reacted with NO x (NO, NO 2 ) to form NH 2 NO or NH 2 NO 2 , which rapidly discomposed into N 2 and H 2 O [57]. As given in Figure 5b, the adsorbed NO 2 or nitrates species reacts with adsorbed ammonia species when the NH 3 was confirmed to follow the Langmuir-Hishelwood (L-H) mechanism.

Preparation of Catalysts
The catalyst samples were obtained by a co-impregnation technique using FA as the catalyst carrier. The The catalysts were prepared in the following order: Firstly, a different concentration of manganese nitrate and iron nitrate was dissolved in de-ionized water. Secondly, 4 g of FA was added to the solution with the ultrasonication of the slurry for 1 h at 80 • C. Thirdly, aqueous ammonia solution (75%) (Sinopharm Chemical Reagent Co. Ltd, (Shanghai, China)) was added drop by drop until a basic mixture was obtained (pH~10). Then, the mixture was dried at 105 • C for 10 h prior to the calcination process. The catalyst sample was calcined in an electric oven at 500 • C at 10 • C/min for three hours. After that, the dried product with a particle size range of 250-400 µm was used for further analysis. The obtained samples were named Mn x Fe y /FA, in which the value of 'x' ranges from 0 to 0.2, and the value of 'y' ranges from 0 to 0.10, representing the weight ratios of manganese/FA and iron/FA, respectively. For comparison purposes, fly ash was treated via the same preparation method, named FA.

Low-Temperature SCR Experiments
The NH 3 -SCR activity was conducted in a fixed-bed quartz reactor (20 mm i.d.). The experimental setup ( Figure 6) consists of three parts: the flue gas simulation system, reaction system, and tail gas test system. The reaction was conducted under a mixture of gases (with (NO) and (NH 3 ) of 0.1 vol.%, and (O 2 ) of 5 vol.%), and N 2 (1200 mL·min −1 ) was used as a balance gas, equivalent to a GHSV of 50,000 h −1 . Activity tests were carried out each 50 • C in the range of 100 to 300 • C, which was heated using a tubular furnace. The concentration of NO x (NO and NO 2 ) was measured by a flue gas analyzer (MRU 5, Germany) and were recorded at each test temperature after stabilization for half an hour. NO x conversion is calculated by Equation (14).
where C NO x in and C NO x out are the NO x concentration at the reactor inlet and outlet, respectively.

Characterization of Catalyst Samples
The morphology of the materials was analyzed by scanning electron microscopy (SEM, Zeiss SIGMA HD, (Oberkochen, Germany)). The physical properties of the catalysts were checked by an N 2 -physisorption analyzer using Kubo X1000 (Beijing Builder Electronic Technology Co. LTD, (Beijing, China)). The X-ray diffraction patterns were determined (Ultima IV, Rigaku, (Tokyo, Japan)) using a Cu K α radiation to obtain the crystal phase structure. The composition of Fe and Mn within the catalysts were determined by the inductively coupled plasma atomic emission spectroscopy (ICP-AES) technique, using Horiba Ultima 2 (Horiba, LTD, (Kyoto, Japan)). A high-performance electron spectrometer (ESCALAB 250, Thermo Fisher Scientific, Inc., (Waltham, MA, USA)) was used to carry out the X-ray photoelectron spectroscopy (XPS) experiments with monochromatic Al Kα radiation (1486.8 eV, 150 w) and the C 1s (284.8 eV) as a reference to calculate the binding energies to investigate the elemental state of catalyst surfaces. The XPS peak 4.1 software was used for curve fitting (Shirley background, Gaussian-Lorentzian ratio fixed to 80/20).
The NH3-SCR activity was conducted in a fixed-bed quartz reactor (20 mm i.d.). The experimental setup ( Figure 6) consists of three parts: the flue gas simulation system, reaction system, and tail gas test system. The reaction was conducted under a mixture of gases (with (NO) and (NH3) of 0.1 vol.%, and (O2) of 5 vol.%), and N2 (1200 mL·min −1 ) was used as a balance gas, equivalent to a GHSV of 50,000 h −1 . Activity tests were carried out each 50 °C in the range of 100 to 300 °C, which was heated using a tubular furnace. The concentration of NOx (NO and NO2) was measured by a flue gas analyzer (MRU 5, Germany) and were recorded at each test temperature after stabilization for half an hour.
NOx conversion is calculated by Equation (14).
where and are the NOx concentration at the reactor inlet and outlet, respectively. The temperature-programmed desorption (NH 3 /NO-TPD) and the temperature program reduction (H 2 -TPR) experiments were conducted on the chemisorption analyzer with a TCD (PCA-1200, Beijing Builder Electronic Technology Co. LTD (Beijing, China)). Before each TPD experiment, all the samples were de-gassed in He (30 mL·min −1 ) at 300 • C for 1 h. The absorption segment was carried out in 10 vol.% NH 3 /He (or 10 vol.% NO/He) at 50 • C for 1 h, and then the samples were purged by He for half an hour to reach a stable baseline. Finally, the temperature of the catalyst was increased to 400 • C (10 • C·min −1 ). However, the H 2 -TPR experiment started with a stable baseline by TCD under H 2 flow (10 vol.% H 2 /Ar). Then, the TCD test was carried out during the heating of the catalyst to 800 • C (10 • C·min −1 ). Desorption and decomposition properties of the surface species of the catalyst were tested by a simultaneous thermal analyzer (STA 8000, PerkinElmer, Inc., (Waltham, MA, USA)) attached with a Hiden HPR20 mass spectrometer (MS) (Hiden Analytical, Inc., (Peterborough, NH, USA)). Before each run, the sample was first adsorbed with NO or NH 3 , similar to the TPD process. Then, the temperature was ramped at a similar heating rate (10 • C·min −1 ) under helium. The in-situ DRIFT spectroscopy was conducted on FTIR spectrometer (Nicolet IS50, Thermo Fisher Scientific, Inc., (Waltham, MA, USA)) equipped with a ZnSe reaction cell. The sample was first de-gassed at 300 • C for 1 h under nitrogen gas (100 mL·min −1 ) to remove any adsorbed species (e.g., air or vapor). During the experiment, the background spectrum was obtained at the corresponding temperature, and then the reactant gas ([NH 3 ] = 1000 ppm or [NO] = 1000ppm combine with [O 2 ] = 5 vol.%) was introduced.

Conclusions
The Mn 0.15 Fe 0.05 /FA catalysts supported on coal fly ash had an excellent LTSCR catalytic performance in the temperature window between 130 • C and 300 • C. The optimum SCR activity in this study was obtained in the catalyst containing 0.05 and 0.15 weight ratios of Mn/FA and Fe/FA, respectively.
The great De-NO x activity of Mn 0.15 Fe 0.05 /FA was attributed to the following favorable properties: (1) The doped FeO x can promote the production of more high valence states of Mn that enriched these Mn 4+ species on the surface; (2) the interaction between the doped species (MnO x , FeO x ) and FA provides an abundant amount of active sites, promoting adsorption and activation of reactants; (3) the proper amount of acid property, reactant adsorption, and redox ability together are the primary factors contributing to the excellent LTSCR efficiency for Mn0.15Fe0.05/FA; (4) the synthesis of Mn 0.15 Fe 0.05 /FA provides a potential high value-added application of coal fly ash, not only to decrease NO x emissions but also to promote the development of cost-effective low-temperature NH 3 -SCR catalysts.
The presence of SO 2 (even in a small amount) may strongly impact the performance of low-temperature SCR systems. Future research should be focused on improving the SO 2 poisoning resistance of the catalysts by introducing additives and changing the structure of catalysts.