Low Temperature deNOx Catalytic Activity with C 2 H 4 as a Reductant Using Mixed Metal Fe-Mn Oxides Supported on Activated Carbon

: The selective catalytic reduction of NO x (deNOx) at temperatures less than or at 200 ◦ C was investigated while using C 2 H 4 as the reductant and mixed oxides of Fe and Mn supported on activated carbon; their activity was compared to that of MnOx and FeOx separately supported on activated carbon. The bimetallic oxide compositions maintained high NO conversion of greater than 80–98% for periods that were three times greater than those of the supported monometallic oxides. To examine potential reasons for the signiﬁcant increases in activity maintenance, and subsequent deactivation, the catalysts were examined by using bulk and surface sensitive analytical techniques before and after catalyst testing. No signiﬁcant changes in Brunauer-Emmett-Teller (BET) surface areas or porosities were observed between freshly-prepared and tested catalysts whereas segregation of FeOx and MnOx species was readily observed in the mono-oxide catalysts after reaction testing that was not detected in the mixed oxide catalysts. Furthermore, x-ray di ﬀ raction and Raman spectroscopy data detected cubic Fe 3 Mn 3 O 8 in both the freshly-prepared and reaction-tested mixed oxide catalysts that were more crystalline after testing. The presence of this compound, which is known to stabilize multivalent Fe species and to enhance oxygen transfer reactions, may be the reason for the high and relatively stable NO conversion activity, and its increased crystallinity during longer-term testing may also decrease surface availability of the active sites responsible for NO conversion. These results point to a potential of further enhancing catalyst stability and activity for low temperature deNOx that is applicable to advanced SCR processing with lower costs and less deleterious side e ﬀ ects to processing equipment. / 1.0Fe3Mn NAC versus The bimetallic oxide catalysts had high initial activity and signiﬁcantly better conversion the mono-metal oxide catalysts. Some deactivation still for the bimetallic oxide catalysts after 2 h, in longer-term performance for synergistic e ﬀ ects between the Fe and Mn species. data from the as-prepared and reaction tested catalysts by surface and bulk analytical techniques, including BET surface area, porosity and pore volume analyses, SEM-EDS, XRD, Raman spectroscopy, and XPS. Overall surface areas, porosities and pore volumes of the supported mono-oxide or mixed metal oxide catalysts did not change as the concentrations of FeOx and MnOx were increased, and no di ﬀ erence existed between the as-prepared and tested catalysts. The SEM-EDS data indicated agglomeration of the FeOx and MnOx species in the mono-metal oxide catalysts during reaction testing while less, although some agglomeration, occurred in the mixed metal oxide catalysts after reaction testing. This di ﬀ erence agreed with the higher 80% NO conversion after 120 min of testing for the mixed metal oxide as compared to only 20% NO conversion at the same time for the mono-metal oxide catalysts. Simultaneously, the XRD and Raman spectroscopic data showed Fe 3 Mn 3 O 8 was formed during mixed metal oxide catalyst synthesis, and this compound was also identiﬁed in the reaction tested mixed metal oxide catalysts. This compound had higher crystallinity or ordering after reaction testing according to the Raman data, a factor which would decrease NO surface active sites and, thereby, decrease catalytic activity. Overall, the study supports the potential to synthesize active and stable HC-SCR catalysts for low temperature applications that are highly compatible with catalyst placement in reactor zones free of the deleterious gas phase species that occur in higher temperature reaction zones currently needing to be used for NH 3 -SCR catalysts.


Introduction
Nitrogen oxides (NO X ) are created during fossil fuel combustion and participate in the formation of acid rain and photochemical smog [1][2][3][4]. Consequently, deNOx technologies have been widely studied and commercially-practiced throughout the world [5]; these include NO X direct decomposition, selective catalytic reduction (SCR), and selective non-catalytic reduction (SNCR) [6]. Among the most successful and applied deNOx technologies is SCR of NO X using ammonia (NH 3 ) as the reductant (NH 3 -SCR).
The most widely used NH 3-SCR catalyst in stationary power plants and refineries is V 2 O 5 /WO 3 /TiO 2 [7][8][9][10][11]. Despite its high activity and stability during deNOx reactions, the need to maintain temperatures in a range of 300-420 • C requires additional external heating if the catalyst is located downstream of particulate collectors [12] and desulfurization units [13] where temperatures are normally below 200 • C. If located upstream of particulate collection and desulfurization units, the temperatures are compatible with those needed for V 2 O 5 /WO 3 /TiO 2 catalysts but significant particulate accumulation on the catalyst and sulfur poisoning will cause rapid deactivation [12][13][14]. Additionally, in this higher-temperature location, the presence of the NH 3 reductant causes the formation of ammonium salts (ammonium sulfate or ammonium bisulfate) [15] which crystallize and deposit [16] on downstream flow pipes of waste heat boilers; this deposition causes fouling and blockages that endanger the long-term operation of catalytic systems [17]. The stability and durability are critical in commercial catalyst application and the life time of current employed NH 3 -SCR catalyst is about 16,000 h. Ammonia is also a poisonous gas at concentrations higher than 500 mg/m 3 , and strict precautions are required for safety during transportation, storage and use [17,18]. Furthermore, spent vanadium-based catalysts are considered hazardous wastes, and require specialized post-treatment that is difficult and costly [19].
To overcome the issues and difficulties associated with V 2 O 5 /WO 3 /TiO 2 deNOx technology, it would be beneficial if operating temperatures of deNOx catalysts could be lowered to less than 250 • C. If successful, they could be installed downstream of a heat recovery steam generator (HRSG) unit to avoid deposition and the resulting blockage of heat exchangers. Interestingly, the temperature ranges of the flue gases from a majority of small-scale industrial boilers like coke ovens, steel furnaces, and glass kilns already fall within 200-300 • C and can be even lower than 200 • C after desulphurization units in these boilers [20]. Currently-used vanadium-based catalysts would be inefficient at these temperatures [8] Therefore, significant impetus exists to develop alternative SCR deNOx catalysts having high and stable performance at lower temperatures than now commercially practiced. For example, Zhu et al. [21] developed an excellent low temperature NH 3 -SCR catalyst based on Mn a Ce 0.3 TiO x (a = 0.1-0.3) oxides, and reported a 100% NO conversion and above 90% N 2 selectivity between 175 • C and 400 • C; Shih et al. [22] found that a Mn and Ti-containing catalyst provided a NO reduction rate of 83% at a relatively low temperature of 200 • C. Other methods such as the use of plasma could also enhance catalytic conversion of NO at low temperatures [23,24].
Hydrocarbons, such as C 2 H 4 , have been examined as a replacement reductive in place of NH 3 during deNO x reactions under lean conditions (labeled as HC-SCR). The overall reaction equation of HC-SCR can be written as: HC-SCR is promising for the removal of NOx under lean conditions since this method exploits unburned hydrocarbons already present in the exhaust gas stream [25]. However, there are some drawbacks along with hydrocarbons such as the increase in CO 2 emissions, the cost of reducing agent and hydrocarbons are also air pollutants. However, the CO 2 formed by C 2 H 4 is very minor compared with the coal combustion and can be neglected.
Current developed HC-SCR catalysts with excellent catalytic performance are mainly two types: Noble metal oxides and metal-exchanged zeolites. Xu et al. [26] found Ag/Al 2 O 3 exhibited almost 100% in the temperature region of 300-500 • C during H 2 -C 3 H 6 -SCR. However, as we have discussed before, the existence of SO 2 in the flue gas will cause the catalyst deactivated quickly and NOx conversion stabilized at 25%. The SCR catalyst used in power plants have strict requirement of its life to insure long-term operation. The stability and cost shortcoming of noble metal catalyst caused all the studies remained at laboratory conditions. Metal-exchanged zeolites also present good NOx conversion value in high temperature region (300-500 • C), but Valerie Houel et al. [27] showed that Cu/ZSM-5 catalyst was fairly insensitive to SO 2 compared with Ag/Al 2 O 3 . Therefore, the HC-SCR catalysts are far away from commercial application due to the stability and durability requirement in power plant and other small-scale industrial boilers. Non-vanadium catalysts have been examined in HC-SCR reaction testing and may provide an approach for circumventing the problems of current NH 3 -SCR technologies [17]. The basis of these catalysts include transition metal oxides containing manganese, iron or copper [28,29], supported noble metal oxides [30] and metal-exchanged zeolites like ZSM-5 [31]. Of these, the transition metal oxides have advantages of high activity at low-temperatures, environmentally acceptable handling and disposal, and cost effectiveness [20,28].
Manganese oxides (MnOx) have displayed excellent initial deNO X performance at lowtemperatures; for example, Mn/β-zeolite had 97.5% NO conversion at 240 • C [32]. Iron-based catalysts also have high deNOx activity, and very low toxicity compared to V 2 O 5 -WO 3 (MoO 3 )/TiO 2 catalysts [20,33]. Ren et al. [6] showed that pure Fe 2 O 3 exhibited 95% NO conversion although the operation temperature window was toward a higher range of 250 • C to 400 • C. Supports for these transition metal catalysts have been inert metal oxides, such as Al 2 O 3 , TiO 2 , and SiO2 [34][35][36] that promote high temperature resistance and mechanical strength [12]; as a downside, some of these have exhibited low surface areas which decreases activity [32]. Using carbon-based supports has also been attempted because of the potential for high specific surface areas (like in activated carbon-AC), the ease of modifying their surface properties and chemical stability [34,37], attractive porosity and relatively low costs [12,38].
The use of C 2 H 4 as the reductant for MnOx-based deNOx catalysts supported on AC showed NO conversions up to 100% at a temperature as low as 130 • C [39]. However, these mono-transition metal catalysts usually suffered from rapid deactivation; for example, the NO conversion over a 3.0 wt.% Mn/AC catalyst decreased by 70% after about 2 h of reaction testing [37]. Yang et. al. [37] studied the deactivation mechanisms of a CuOx/AC catalyst during C 2 H 4 -SCR reactions and found that the active metal oxide phase changed along with component aggregation that were attributed to catalyst deactivation. Tang et. al. also obtained more than 90% NO X conversion at 150-250 • C with MnOx/AC in NH 3 -SCR [37].
The development of commercial transitional metal-based HC-SCR catalysts is still in an early stage, especially for catalysts using AC supports. Hence, in this study, mono-transition metal oxides, i.e., MnOx or FeOx, and their mixtures were supported on a commercial AC and tested for NO conversion activity during C 2 H 4 -SCR reactions at temperatures below 200 • C; the freshly-prepared and reaction-tested catalysts were then analyzed using bulk and surface sensitive techniques to determine structural and chemical aspects relating to their activity and deactivation.

Catalyst Preparation
The activated carbon was purchased from Wanlong company, and had a particle size of 1000-2360 µm. It was treated with 10% HNO 3 for 4 h at room temperature, and then was washed with distilled water until the effluent pH became neutral; this AC was then dried at 140 • C for 14 h. It is denoted as nitric treated activated carbon (NAC).
The precursors used for impregnating MnOx and FeOx were Mn·(NO 3 ) 2 ·4H 2 O and Fe(NO 3 ) 3 ·9H 2 O. The NAC supports were impregnated with these solutions to achieve approximate Mn or Fe loadings of 3.0 wt.%, 5.0 wt.%, and 7.0 wt.% during which time the mixtures were sonicated in a bath for 2 h. After the mixtures were allowed to stand for 12 h, they were dried at 110 • C in a vacuum oven and then calcined at 400 • C for 2 h in a sealed muffle furnace under N 2 atmosphere. The synthesized materials were labeled as XMn/NAC, XFe/NAC where (X = 3, 5, 7).
The NAC was impregnated with an appropriate concentration of mixed precursor solution to achieve specified iron and manganese loadings within an ultrasonic bath for 2 h at room temperature. Then, the mixtures were allowed to stand for 12 h, after which they were dried in a vacuum oven at 110 • C and calcined at 400 • C for 2 h in a sealed muffle furnace under N 2 atmosphere. The synthesized catalysts were labeled as XFe3Mn/NAC, where (X = 0, 1.0, 1.5).

Catalyst Characterization
The catalysts were characterized to elucidate their physical and chemical properties before and after reaction testing. Surface areas and porosities were measured using a Micromeritics ASAP 2020 (ASAP 2020, Micromeritics Instrument Corporation, Norcross, U.S.A.) in which each sample was degassed overnight at 160 • C and then subjected to isothermal N 2 adsorption-desorption measurements at 77 K. Crystalline structures were examined by X-ray diffraction (XRD) using Cu Kα irradiation in a Rigaku SmartLab system with a 2θ range of 10-75 • . Microstructures were investigated using scanning electron microscopy (SEM). The SEM imaging was performed on a FEI Quanta 250 embedded with a Bruker Quantax energy dispersive spectrometer (EDS). Molecular speciation was examined using Raman spectroscopy (Horiba-Jobin Yvon LabRam HR) at a spectral resolution of 2 cm −1 while using 442 nm laser excitation to minimize sample fluorescence. The laser light was focused on the samples with a confocal microscope with a 50× objective (Olympus BX-30-LWD, NA = 0.5); the wavenumber calibration was checked using the silica vibrational mode at 520.7 cm −1 . X-ray photoelectron spectroscopy (XPS) was performed using an ESCALAB 250Xi instrument (Thermo Fisher, Waltham, U.S.A) with Al Kα excitation. The spectra were calibrated using the ubiquitous C1s peak at 284.8 eV.

Bench-Scale Testing for Catalytic Activity
The experimental setup for catalytic testing is shown in Figure 1. Ten grams of each sample was placed in a fixed-bed quartz tube reactor (i.d. = 20 mm) which was externally heated by a temperature-controlled furnace. Gas flow rates were controlled by mass-flow controllers (MFCs, MF SHY 400, SHY company, Suzhou, China), with the simulated gas mixture regulated to 500 ppm NO, 500 ppm C 2 H 4 , 3.0 vol.% O 2 and N 2 as the balance flow with a total flow rate of 1500 ml/min. Gas mixtures were injected into the reactor after catalyst temperatures reached a stable set point for each test. Inlet and outlet gas concentrations of NO and NO X were analyzed by a flue gas analyzer (MRU, NOVA Plus, Heilbronn, Germany). From these data, the catalytic activities were calculated according to the following equation: where the subscripts in and out indicate the inlet and outlet concentrations, respectively.
Energies 2019, 12, x FOR PEER REVIEW 4 of 14 measurements at 77 K. Crystalline structures were examined by X-ray diffraction (XRD) using Cu Kα irradiation in a Rigaku SmartLab system with a 2θ range of 10-75°. Microstructures were investigated using scanning electron microscopy (SEM). The SEM imaging was performed on a FEI Quanta 250 embedded with a Bruker Quantax energy dispersive spectrometer (EDS). Molecular speciation was examined using Raman spectroscopy (Horiba-Jobin Yvon LabRam HR) at a spectral resolution of 2 cm −1 while using 442 nm laser excitation to minimize sample fluorescence. The laser light was focused on the samples with a confocal microscope with a 50× objective (Olympus BX-30-LWD, NA = 0.5); the wavenumber calibration was checked using the silica vibrational mode at 520.7 cm −1 . X-ray photoelectron spectroscopy (XPS) was performed using an ESCALAB 250Xi instrument (Thermo Fisher, Waltham, U.S.A) with Al Kα excitation. The spectra were calibrated using the ubiquitous C1s peak at 284.8 eV.

Bench-Scale Testing for Catalytic Activity
The experimental setup for catalytic testing is shown in Figure 1. Ten grams of each sample was placed in a fixed-bed quartz tube reactor (i.d. = 20 mm) which was externally heated by a temperature-controlled furnace. Gas flow rates were controlled by mass-flow controllers (MFCs, MF SHY 400, SHY company, Suzhou, China), with the simulated gas mixture regulated to 500 ppm NO, 500 ppm C2H4, 3.0 vol.% O2 and N2 as the balance flow with a total flow rate of 1500 ml/min. Gas mixtures were injected into the reactor after catalyst temperatures reached a stable set point for each test. Inlet and outlet gas concentrations of NO and NOX were analyzed by a flue gas analyzer (MRU, NOVA Plus, Heilbronn, Germany). From these data, the catalytic activities were calculated according to the following equation: where the subscripts in and out indicate the inlet and outlet concentrations, respectively.

Catalytic Performance
The NO conversion data versus time at different temperatures are shown in Figure 2 while using the 3Mn/NAC catalyst. It was used to provide baseline data that could be compared with previously published data [39]. Nearly 100% NO conversion occurred during the first 20 min at all temperatures (150 °C, 180 °C and 200 °C), but it suffered rapid declines after this initial period and showed less than 20% conversion after 2 h of testing. The temperature dependence of the activity followed the order of 150 °C > 180 °C > 200 °C.

Catalytic Performance
The NO conversion data versus time at different temperatures are shown in Figure 2 while using the 3Mn/NAC catalyst. It was used to provide baseline data that could be compared with previously published data [39]. Nearly 100% NO conversion occurred during the first 20 min at all temperatures (150 • C, 180 • C and 200 • C), but it suffered rapid declines after this initial period and showed less than The NO conversion testing data for XFe/NAC (X = 3, 5 and 7%) at 150 °C are presented in Figure  3. Similar to the 3Mn/NAC data in Figure 2, the NO conversion declined dramatically after 15-20 min; it then leveled to values near 20% after about 80 min of testing, similar to the data for 3Mn/NAC. The Fe concentration had no significant effect on overall NO conversion. In summary, mono-metal oxide of either MnOx or FeOx on NAC had initial high NO conversion that was followed by rapid declines in activity after only 15-20 min from the beginning of reaction testing.  Figure 4 displays NO conversion data for the mono-metal oxides of 3Fe/NAC and 3Mn/NAC, and the bimetallic oxides 1.5Fe3Mn/NAC and 1.0Fe3Mn/NAC versus time at a 150 °C reaction temperature. The bimetallic oxide catalysts had high initial activity and significantly better conversion maintenance than did the mono-metal oxide catalysts. Some deactivation was still evident for the bimetallic oxide catalysts after 2 h, however, the improvement in longer-term performance was evidence for synergistic effects between the Fe and Mn species. The NO conversion testing data for XFe/NAC (X = 3, 5 and 7%) at 150 • C are presented in Figure 3. Similar to the 3Mn/NAC data in Figure 2, the NO conversion declined dramatically after 15-20 min; it then leveled to values near 20% after about 80 min of testing, similar to the data for 3Mn/NAC. The Fe concentration had no significant effect on overall NO conversion. In summary, mono-metal oxide of either MnOx or FeOx on NAC had initial high NO conversion that was followed by rapid declines in activity after only 15-20 min from the beginning of reaction testing. The NO conversion testing data for XFe/NAC (X = 3, 5 and 7%) at 150 °C are presented in Figure  3. Similar to the 3Mn/NAC data in Figure 2, the NO conversion declined dramatically after 15-20 min; it then leveled to values near 20% after about 80 min of testing, similar to the data for 3Mn/NAC. The Fe concentration had no significant effect on overall NO conversion. In summary, mono-metal oxide of either MnOx or FeOx on NAC had initial high NO conversion that was followed by rapid declines in activity after only 15-20 min from the beginning of reaction testing.  Figure 4 displays NO conversion data for the mono-metal oxides of 3Fe/NAC and 3Mn/NAC, and the bimetallic oxides 1.5Fe3Mn/NAC and 1.0Fe3Mn/NAC versus time at a 150 °C reaction temperature. The bimetallic oxide catalysts had high initial activity and significantly better conversion maintenance than did the mono-metal oxide catalysts. Some deactivation was still evident for the bimetallic oxide catalysts after 2 h, however, the improvement in longer-term performance was evidence for synergistic effects between the Fe and Mn species.  Figure 4 displays NO conversion data for the mono-metal oxides of 3Fe/NAC and 3Mn/NAC, and the bimetallic oxides 1.5Fe3Mn/NAC and 1.0Fe3Mn/NAC versus time at a 150 • C reaction temperature. The bimetallic oxide catalysts had high initial activity and significantly better conversion maintenance than did the mono-metal oxide catalysts. Some deactivation was still evident for the bimetallic oxide catalysts after 2 h, however, the improvement in longer-term performance was evidence for synergistic effects between the Fe and Mn species.  The difference between NOx and NO of all tested catalysts was very small and was near 0% during the first 20 min of testing, and remained at less than 1-5 ppm during the entire tests. The blank experiment showed that NAC support contributed a very small number of NO conversion, which was less than 2%, which was not discussed while nearly 90% NO conversion was achieved.
Compared with current commercial SCR catalysts, the present studied Fe-Mn-catalyst was in poor stability. Deactivation is the common problem for low temperature HC-SCR catalysts, thus ease of deactivation is the key of this study. The excited and interesting phenomena have been observed during such preliminary study, especially, the Fe-Mn-catalyst with C2H4 exhibited a promoted performance.

Active Components Dispersion
Surface morphologies of the catalysts were investigated by SEM and SEM-EDS to examine whether dispersion changes of MnOx and FeOx species were related to rapid conversion declines or maintenance. The SEM images of HNO3 treated AC, i.e., NAC without metal oxides, and of freshly prepared 1.5Fe3Mn/NAC in Figure 5 show readily accessible porous structures not noticeably changed as a consequence of incorporating the Fe and Mn oxides. In fact, as presented in Table 1, the surface areas and pore volumes of the NAC and metal oxide-containing catalysts were very similar to each other and near 650 m 2 /g and 0.4 cm 3 /g, respectively, before and after reaction testing (the aspurchased AC, before HNO3 treatment, had a surface area of 710 m 2 /g). Maintenance of the porous structure may have been beneficial for dispersing the active species or maintaining an effective pathway for access of reactants and products to and from the active sites; however, the surface areas and porosities of the mono-metal 3Mn/NAC also did not significantly change before and after reaction testing although its NO conversion activity rapidly declined after 15-20 min of testing. Hence, catalyst surface areas and porosities did not control NO conversion for these catalysts, in agreement with previous studies showing the surface area of supported MnOx catalysts was not a main cause of catalytic activity change during HC-SCR reactions [40].  The difference between NOx and NO of all tested catalysts was very small and was near 0% during the first 20 min of testing, and remained at less than 1-5 ppm during the entire tests. The blank experiment showed that NAC support contributed a very small number of NO conversion, which was less than 2%, which was not discussed while nearly 90% NO conversion was achieved.
Compared with current commercial SCR catalysts, the present studied Fe-Mn-catalyst was in poor stability. Deactivation is the common problem for low temperature HC-SCR catalysts, thus ease of deactivation is the key of this study. The excited and interesting phenomena have been observed during such preliminary study, especially, the Fe-Mn-catalyst with C 2 H 4 exhibited a promoted performance.

Active Components Dispersion
Surface morphologies of the catalysts were investigated by SEM and SEM-EDS to examine whether dispersion changes of MnOx and FeOx species were related to rapid conversion declines or maintenance. The SEM images of HNO 3 treated AC, i.e., NAC without metal oxides, and of freshly prepared 1.5Fe3Mn/NAC in Figure 5 show readily accessible porous structures not noticeably changed as a consequence of incorporating the Fe and Mn oxides. In fact, as presented in Table 1, the surface areas and pore volumes of the NAC and metal oxide-containing catalysts were very similar to each other and near 650 m 2 /g and 0.4 cm 3 /g, respectively, before and after reaction testing (the as-purchased AC, before HNO 3 treatment, had a surface area of 710 m 2 /g). Maintenance of the porous structure may have been beneficial for dispersing the active species or maintaining an effective pathway for access of reactants and products to and from the active sites; however, the surface areas and porosities of the mono-metal 3Mn/NAC also did not significantly change before and after reaction testing although its NO conversion activity rapidly declined after 15-20 min of testing. Hence, catalyst surface areas and porosities did not control NO conversion for these catalysts, in agreement with previous studies showing the surface area of supported MnOx catalysts was not a main cause of catalytic activity change during HC-SCR reactions [40].  The dispersions of MnOx and FeOx on NAC were further investigated by SEM-EDS mapping. These data for as-prepared 1.5Fe3Mn/NAC, Figure 6 show the Mn signal was evenly distributed over the entire surface that was imaged while the Fe signal had slightly less of an even distribution. These data suggest that before reaction testing the MnOx and FeOx are mostly distributed uniformly on the NAC surfaces.
Elemental concentrations were also acquired using EDS data, a spectrum for which is displayed in Figure 7 along with the elemental concentrations. These concentrations were very close to the preparation concentrations of 1.5Fe and 3Mn.  The dispersions of MnOx and FeOx on NAC were further investigated by SEM-EDS mapping. These data for as-prepared 1.5Fe3Mn/NAC, Figure 6 show the Mn signal was evenly distributed over the entire surface that was imaged while the Fe signal had slightly less of an even distribution. These data suggest that before reaction testing the MnOx and FeOx are mostly distributed uniformly on the NAC surfaces.
Elemental concentrations were also acquired using EDS data, a spectrum for which is displayed in Figure 7 along with the elemental concentrations. These concentrations were very close to the preparation concentrations of 1.5Fe and 3Mn.

Physical Properties Evolution
The SEM-EDS acquired Fe and Mn elemental distributions for 1.5Fe3Mn/NAC after reaction testing, as shown in Figure 8. Some agglomeration of Fe and Mn distributions can be found as compared to the distributions within the as-prepared catalyst in Figure 6. These Fe agglomerates are seen in the bright contrast areas on the catalyst surfaces toward the right and lower sides of the distribution maps. Agglomeration of Mn is also shown in Figure 8c but its extent is less than that observed for Fe. Hence, as compared to the EDS elemental distribution maps in Figure 6 of the asprepared 1.5Fe3Mn/NAC catalyst, the EDS data from tested catalysts indicated more Fe and Mn agglomeration. These aggregates could represent agglomeration sites of the originally active catalytic components, and thereby greatly affect active site dispersion to cause rapid deactivation. However, although this agglomeration was less than that observed for MnOx species in previously published data on a reaction tested 3Mn/NAC catalyst [39], its activity had decreased to near 20% NO conversion after 120 min of testing whereas the activity of 1.5Fe3Mn/NAC at the same time was near

Physical Properties Evolution
The SEM-EDS acquired Fe and Mn elemental distributions for 1.5Fe3Mn/NAC after reaction testing, as shown in Figure 8. Some agglomeration of Fe and Mn distributions can be found as compared to the distributions within the as-prepared catalyst in Figure 6. These Fe agglomerates are seen in the bright contrast areas on the catalyst surfaces toward the right and lower sides of the distribution maps. Agglomeration of Mn is also shown in Figure 8c but its extent is less than that observed for Fe. Hence, as compared to the EDS elemental distribution maps in Figure 6 of the as-prepared 1.5Fe3Mn/NAC catalyst, the EDS data from tested catalysts indicated more Fe and Mn agglomeration. These aggregates could represent agglomeration sites of the originally active catalytic components, and thereby greatly affect active site dispersion to cause rapid deactivation. However, although this agglomeration was less than that observed for MnOx species in previously published data on a reaction tested 3Mn/NAC catalyst [39], its activity had decreased to near 20% NO conversion after 120 min of testing whereas the activity of 1.5Fe3Mn/NAC at the same time was near 80% NO conversion. Thus, their presence did not significantly affect the catalytic performance but still cause a NO conversion drop during 2 h testing.

Interactions between Fe and Mn Oxides
XRD data from 1.5Fe3Mn/NAC are presented in Figure 9 in which the main peaks have been identified. The broad peaks at 2θ = 24° and 2θ = 43° (labeled as A) are in agreement with those of amorphous carbon [41]; their positions and widths were the same in the fresh and tested catalysts, pointing to the stability of the NAC amorphous structure. Three other peaks at 2θ = 33.5°, 2θ = 35.1°, and 2θ = 54.3° (labeled as B, C and D, respectively) were present in both the fresh and tested catalysts, and were not significantly altered in position or peak intensities as a result of reaction testing. The peaks at 2θ = 33.5° (peak labeled as B) is attributed to Fe2O3 [42] and peaks at 2θ=35.1° and 54.3° are

Interactions between Fe and Mn Oxides
XRD data from 1.5Fe3Mn/NAC are presented in Figure 9 in which the main peaks have been identified. The broad peaks at 2θ = 24 • and 2θ = 43 • (labeled as A) are in agreement with those of amorphous carbon [41]; their positions and widths were the same in the fresh and tested catalysts, pointing to the stability of the NAC amorphous structure. Three other peaks at 2θ = 33.5 • , 2θ = 35.1 • , and 2θ = 54.3 • (labeled as B, C and D, respectively) were present in both the fresh and tested catalysts, and were not significantly altered in position or peak intensities as a result of reaction testing. The peaks at 2θ = 33.5 • (peak labeled as B) is attributed to Fe 2 O 3 [42] and peaks at 2θ = 35.1 • and 54.3 • are attributed to Fe 3 Mn 3 O 8 [43]; although the intensities and widths of these Fe-related peaks were the same in the fresh and tested catalysts, it is rather difficult to determine whether Fe and Mn crystallinities have changed during testing because the XRD peaks are weak. Nevertheless, incorporating FeOx with the MnOx did cause the formation of Fe 3 Mn 3 O 8 solid solutions which may be an active species in the FeOx-MnOx/NAC catalysts for NO conversion [43,44].
Energies 2019, 12, x FOR PEER REVIEW attributed to Fe3Mn3O8 [43]; although the intensities and widths of these Fe-related peaks were the same in the fresh and tested catalysts, it is rather difficult to determine whether Fe and Mn crystallinities have changed during testing because the XRD peaks are weak. Nevertheless, incorporating FeOx with the MnOx did cause the formation of Fe3Mn3O8 solid solutions which may be an active species in the FeOx-MnOx/NAC catalysts for NO conversion [43,44]. Raman spectra from 1.5Fe3Mn/NAC were also acquired before and after reaction testing, as shown in Figure 10. The spectrum of the fresh catalyst contained bands at 1360 cm −1 (labeled D), 1610 cm −1 (G) and triplet structured bands near 3000 cm −1 (2D); these are attributed to the activated carbon support [40] and are present with similar intensities in the reaction tested catalyst. A peak at 650 cm −1 in the as-prepared catalyst is attributed to Fe-O-Mn bond vibrations [45]; this band also was observed in the reaction tested catalyst but is narrower and higher in intensity. The narrowing of this band after testing suggests a higher degree of order or crystallinity in the Fe3Mn3O8 species than before testing. Increased crystallinity as a result of testing points to a reordering or redistribution of the FeOx and MnOx species, a result also suggested by the SEM-EDS data from the as-prepared and tested catalysts. This redistribution would decrease the availability of active catalytic sites and could cause decreased NO conversion at the longer reaction times near 2 h.  Raman spectra from 1.5Fe3Mn/NAC were also acquired before and after reaction testing, as shown in Figure 10. The spectrum of the fresh catalyst contained bands at 1360 cm −1 (labeled D), 1610 cm −1 (G) and triplet structured bands near 3000 cm −1 (2D); these are attributed to the activated carbon support [40] and are present with similar intensities in the reaction tested catalyst. A peak at 650 cm −1 in the as-prepared catalyst is attributed to Fe-O-Mn bond vibrations [45]; this band also was observed in the reaction tested catalyst but is narrower and higher in intensity. The narrowing of this band after testing suggests a higher degree of order or crystallinity in the Fe 3 Mn 3 O 8 species than before testing. Increased crystallinity as a result of testing points to a reordering or redistribution of the FeOx and MnOx species, a result also suggested by the SEM-EDS data from the as-prepared and tested catalysts. This redistribution would decrease the availability of active catalytic sites and could cause decreased NO conversion at the longer reaction times near 2 h. attributed to Fe3Mn3O8 [43]; although the intensities and widths of these Fe-related peaks were the same in the fresh and tested catalysts, it is rather difficult to determine whether Fe and Mn crystallinities have changed during testing because the XRD peaks are weak. Nevertheless, incorporating FeOx with the MnOx did cause the formation of Fe3Mn3O8 solid solutions which may be an active species in the FeOx-MnOx/NAC catalysts for NO conversion [43,44]. Figure 9. XRD spectra of 1.5Fe3Mn/NAC catalyst before and after reaction testing.
Raman spectra from 1.5Fe3Mn/NAC were also acquired before and after reaction testing, as shown in Figure 10. The spectrum of the fresh catalyst contained bands at 1360 cm −1 (labeled D), 1610 cm −1 (G) and triplet structured bands near 3000 cm −1 (2D); these are attributed to the activated carbon support [40] and are present with similar intensities in the reaction tested catalyst. A peak at 650 cm −1 in the as-prepared catalyst is attributed to Fe-O-Mn bond vibrations [45]; this band also was observed in the reaction tested catalyst but is narrower and higher in intensity. The narrowing of this band after testing suggests a higher degree of order or crystallinity in the Fe3Mn3O8 species than before testing. Increased crystallinity as a result of testing points to a reordering or redistribution of the FeOx and MnOx species, a result also suggested by the SEM-EDS data from the as-prepared and tested catalysts. This redistribution would decrease the availability of active catalytic sites and could cause decreased NO conversion at the longer reaction times near 2 h. Figure 10. Raman spectra of 1.5Fe3Mn/NAC catalyst before and after reaction testing. Figure 10. Raman spectra of 1.5Fe3Mn/NAC catalyst before and after reaction testing. The XPS spectra of Mn2p, Fe2p, O1s and curve fittings to their peak envelopes of 1.5Fe3Mn/NAC before and after reaction testing are presented in Figure 11. The Mn spectra showed two main peaks at 641.6 and 653.3 eV, corresponding to Mn 2p 3/2 and Mn 2p 1/2 , respectively [46]. For Fe, the peak envelopes were at 711.1 and 724.6 eV, corresponding to Fe 2p 3/2 and Fe 2p 1/2 and have multiple shoulders within their envelopes that signify more than one Fe species; their satellite peaks were at 719.1 eV and 733.3 eV [47]. The O1s peaks were asymmetric and were deconvoluted into two symmetric peaks at 530.0 and 531.8 eV; these were assigned to lattice oxygen and absorbed oxygen, respectively [46].
Deconvolution of the Mn and Fe XPS data point to three Mn species, including Mn 4+ , Mn 3+ and Mn 2+ , and three Fe species, including one Fe 2+ and two Fe 3+ species, respectively [46]. The XPS spectra of Mn, Fe, and O from the catalyst were relatively stable for both the fresh and reaction tested samples, but some small changes were observed. For example, after testing the percentage of Mn 4+ dropped from 19.5% to 13.5% and the Mn 3+ increased by about 6.9%, while the percentage of Fe 2+ increased from 8.4% to 12.5%. However, the change of valence distribution for Mn and Fe after reaction were relatively small, and the Fe-Mn synergistic effect may help to stabilize the composition. The XPS spectra of Mn2p, Fe2p, O1s and curve fittings to their peak envelopes of 1.5Fe3Mn/NAC before and after reaction testing are presented in Figure 11. The Mn spectra showed two main peaks at 641.6 and 653.3 eV, corresponding to Mn 2p3/2 and Mn 2p1/2, respectively [46]. For Fe, the peak envelopes were at 711.1 and 724.6 eV, corresponding to Fe 2p3/2 and Fe 2p1/2 and have multiple shoulders within their envelopes that signify more than one Fe species; their satellite peaks were at 719.1 eV and 733.3 eV [47]. The O1s peaks were asymmetric and were deconvoluted into two symmetric peaks at 530.0 and 531.8 eV; these were assigned to lattice oxygen and absorbed oxygen, respectively [46].
Deconvolution of the Mn and Fe XPS data point to three Mn species, including Mn 4+ , Mn 3+ and Mn 2+ , and three Fe species, including one Fe 2+ and two Fe 3+ species, respectively [46]. The XPS spectra of Mn, Fe, and O from the catalyst were relatively stable for both the fresh and reaction tested samples, but some small changes were observed. For example, after testing the percentage of Mn 4+ dropped from 19.5% to 13.5% and the Mn 3+ increased by about 6.9%, while the percentage of Fe 2+ increased from 8.4% to 12.5%. However, the change of valence distribution for Mn and Fe after reaction were relatively small, and the Fe-Mn synergistic effect may help to stabilize the composition.

Conclusions
The use of mixed oxides of Fe and Mn impregnated onto activated carbon was shown to significantly improve catalytic activity and activity maintenance for decomposing NO with C2H4 as the reductant in comparison to the mono-oxides of FeOx and MnOx supported on activated carbon when temperatures were less than or at 200 °C. Reasons for these differences were sought by acquiring data from the as-prepared and reaction tested catalysts by surface and bulk analytical techniques, including BET surface area, porosity and pore volume analyses, SEM-EDS, XRD, Raman

Conclusions
The use of mixed oxides of Fe and Mn impregnated onto activated carbon was shown to significantly improve catalytic activity and activity maintenance for decomposing NO with C 2 H 4 as the reductant in comparison to the mono-oxides of FeOx and MnOx supported on activated carbon when temperatures were less than or at 200 • C. Reasons for these differences were sought by acquiring data from the as-prepared and reaction tested catalysts by surface and bulk analytical techniques, including BET surface area, porosity and pore volume analyses, SEM-EDS, XRD, Raman spectroscopy, and XPS. Overall surface areas, porosities and pore volumes of the supported mono-oxide or mixed metal oxide catalysts did not change as the concentrations of FeOx and MnOx were increased, and no difference existed between the as-prepared and tested catalysts. The SEM-EDS data indicated agglomeration of the FeOx and MnOx species in the mono-metal oxide catalysts during reaction testing while less, although some agglomeration, occurred in the mixed metal oxide catalysts after reaction testing. This difference agreed with the higher 80% NO conversion after 120 min of testing for the mixed metal oxide as compared to only 20% NO conversion at the same time for the mono-metal oxide catalysts. Simultaneously, the XRD and Raman spectroscopic data showed Fe 3 Mn 3 O 8 was formed during mixed metal oxide catalyst synthesis, and this compound was also identified in the reaction tested mixed metal oxide catalysts. This compound had higher crystallinity or ordering after reaction testing according to the Raman data, a factor which would decrease NO surface active sites and, thereby, decrease catalytic activity. Overall, the study supports the potential to synthesize active and stable HC-SCR catalysts for low temperature applications that are highly compatible with catalyst placement in reactor zones free of the deleterious gas phase species that occur in higher temperature reaction zones currently needing to be used for NH 3 -SCR catalysts.