Supported Inverse MnO x /Pt Catalysts Facilitate Reverse Water Gas Shift Reaction

: Catalytic conversion of CO 2 to CO via the reverse water gas shift (RWGS) reaction has been identified as a promising approach for CO 2 utilization and mitigation of CO 2 emissions. Bare Pt shows low activity for the RWGS reaction due to its low oxophilicity, with few research works having concentrated on the inverse metal oxide/Pt catalyst for the RWGS reaction. In this work, MnO x was deposited on the Pt surface over a SiO 2 support to prepare the MnO x /Pt inverse catalyst via a co-impregnation method. Addition of 0.5 wt% Mn to 1 wt% Pt/SiO 2 improved the intrinsic reaction rate and turnover frequency at 400 ◦ C by two and twelve times, respectively. Characterizations indicate that MnO x partially encapsulates the surface of the Pt particles and the coverage increases with increasing Mn content, which resembles the concept of strong metal–support interaction (SMSI). Although the surface accessible Pt sites are reduced, new MnO x /Pt interfacial perimeter sites are created, which provide both hydrogenation and C-O activation functionalities synergistically due to the close proximity between Pt and MnO x at the interface, and therefore improve the activity. Moreover, the stability is also significantly improved due to the coverage of Pt by MnO x . This work demonstrates a simple method to tune the oxide/metal interfacial sites of inverse Pt-based catalyst for the RWGS reaction.


Introduction
The ever-growing CO 2 emissions into the atmosphere have caused negative impacts on the global ecological environment, such as global warming, sea level rise, and ocean acidification [1].CO 2 capture followed by utilization has been identified as a necessary approach for the reduction of CO 2 emissions and the mitigation of global warming.Catalytic conversion of the captured CO 2 with renewable H 2 to value added products, such as CO, methanol and ethanol [2][3][4], has been widely explored recently.Among these products, conversion of CO 2 to CO via the reverse water gas shift (RWGS) reaction is of particular interest, as CO can be used as a feedstock for further synthesis of various chemicals.Nevertheless, the activity of CO 2 conversion at low temperatures is constrained by the high thermodynamic stability of CO 2 [5].In addition, the selectivity toward CO may be influenced by the competition reaction of methanation, since the latter reaction is strongly exothermic (−164.0kJ•mol −1 ) while the RWGS reaction is mildly endothermic (41.3 kJ•mol −1 ) [6].Hence, catalysts with high low-temperature activity and selectivity for the RWGS reaction remain to be developed.
Catalysts of metal supported on reducible metal oxide have been widely applied for various reactions, including the RWGS reaction.For example, Pt/TiO 2 and Pt/CeO 2 catalysts were studied for the RWGS reaction and it was found that the oxygen vacancies on the support play a key role in CO 2 activation [7][8][9][10].During the reaction, the partially reduced (oxophilic) metal oxide activates the C-O bond, and the metal catalyzes hydrogenation, which synergistically enables the reduction of CO 2 to CO accompanied with H 2 O formation.In this case, the metal-oxide interfacial perimeter sites are the active sites since they provide both functionalities.Similar sites are required for hydrodeoxygenation of phenolic compounds [11][12][13][14][15], which also requires C-O breakage and hydrogenation.Hence, finely tuning the interactions between metal and reducible metal oxide, as well as their interface, could be essential to improve the activity for the RWGS reaction.However, little work has been performed to tune the interface between metal and metal oxide for the RWGS reaction.
The interaction between the metal-oxide, or the metal-oxide interface, can be tuned by the phenomenon of strong metal-support interaction (SMSI), i.e., metal encapsulation by reducible oxide overlayers, which typically occurs during high-temperature reduction pretreatment [16][17][18][19][20].Besides the reduction temperature, the SMSI can be influenced by the identity of the metal oxide, the facet of metal oxide, and the metal particle sizes [11,12,[21][22][23][24].On the other hand, the metal-oxide interface may also be tuned by deposition of metal oxide on the metal surface during catalyst preparation, i.e., the concept of inverse catalyst [25][26][27][28].In this case, the surface of bare metal or metal supported on an inert support is modified by reducible metal oxide promoter [29].In both cases, electronic and geometric modification of metal oxide/metal can occur and may determine catalytic performance.A number of previous reports showed that the inverse oxide/metal catalysts can effectively catalyze the water gas shift reaction [26,27,30,31], in which the H 2 O dissociates on the oxygen vacancies of the metal oxide and CO adsorbs on the metal.The inverse oxide/metal catalysts, such as, Cr 2 O 3 /Cu [32], ZnO/Cu [33] and Y 2 O 3 /Cu [34], have been explored for the RWGS reaction with high activity.
Pt-based catalysts have been extensively studied for the RWGS reaction owing to the hydrogenation ability [8,9,[35][36][37], but the low oxophilicity of Pt renders low activity for the RWGS reaction [38].In addition, manganese has been widely reported as a promoter in hydrogenation reactions [39][40][41][42].MnO x readily forms oxygen vacancies under reducing conditions, which may facilitate the adsorption and activation of CO 2 .To the best of our knowledge, there have been few studies on inverse Pt-based catalyst for the RWGS reaction.Herein, we prepared MnO x /Pt inverse catalyst supported on SiO 2 by an incipient wetness co-impregnation method.SiO 2 was selected as the support, due to the following: (1) it is relatively cheap and readily available; (2) it is inert, so that one can focus on the interaction between MnO x and Pt, and exclude the influence of support.Our results indicate that the optimal coverage of Pt by MnO x results in an optimal MnO x /Pt interface, leading to improved activity and stability for the RWGS reaction.

Catalytic Performance
The catalytic performance of 1PtxMn catalysts (representing 1 wt% Pt − x wt% Mn/SiO 2 ) with varying Mn content (0-5 wt%) for the RWGS reaction was evaluated at 400 • C with a gas hourly space velocity (GHSV) of 60 L•g −1 •h −1 and a CO 2 /H 2 molar ratio of 1/4.At 400 • C and H 2 /CO 2 molar ratio of 4, the thermodynamic equilibrium CO 2 conversion was 41% toward CO [5].A high GHSV of 60 L•g −1 •h −1 was selected to ensure the CO 2 conversion (<11%) is distant from the thermodynamic limitation, and therefore to compare properly the activities of different catalysts.As illustrated in Figure 1, the CO 2 conversion increased with increasing Mn first and then kept constant.Specifically, Pt/SiO 2 exhibited a low CO 2 conversion of 5.1%.The CO 2 conversion increased gradually to 6.9%, 8.7%, 9.5%, and 10.0% on 1Pt0.1Mn,1Pt0.15Mn,1Pt0.2Mn, and 1Pt0.5Mn,respectively, and then leveled off at ~10% when Mn loading was >0.5 wt%.The selectivity towards CO was 100% without CH 4 being detected in any sample.Notably, Mn/SiO 2 was inactive under such reaction conditions (consistent with the literature result [39]), suggesting the Mn species acted as a promoter.Considering the high activity and moderate Mn loadings, the 1Pt0.5Mnsample was selected for further study.The mass based intrinsic reaction rate, measured under differential conditions with CO 2 conversion < 10% by adjusting the GHSV, on 1Pt0.5Mn(3.01 µmol•g −1 •s −1 ) was 2 times higher than that on Pt/SiO 2 (1.54 µmol•g −1 •s −1 ).
Catalysts 2024, 14, x FOR PEER REVIEW 3 of 14 leveled off at ~10% when Mn loading was >0.5 wt%.The selectivity towards CO was 100% without CH4 being detected in any sample.Notably, Mn/SiO2 was inactive under such reaction conditions (consistent with the literature result [39]), suggesting the Mn species acted as a promoter.Considering the high activity and moderate Mn loadings, the 1Pt0.5Mnsample was selected for further study.The mass based intrinsic reaction rate, measured under differential conditions with CO2 conversion < 10% by adjusting the GHSV, on 1Pt0.5Mn(3.01 µmol•g −1 •s −1 ) was 2 times higher than that on Pt/SiO2 (1.54 µmol•g −1 •s −1 ).The effect of GHSV on the conversion of CO2 over Pt/SiO2 and 1Pt0.5Mn is plotted in Figure 2. In the case of 1Pt0.5Mn, the CO2 conversion increased from 10.0% to 27.8% when the GHSV decreased from 60 to 15 L•g −1 •h −1 , due to the increased residence time of reactants.Although a similar trend was observed for Pt/SiO2, the CO2 conversion on Pt/SiO2 was consistently inferior to that on 1Pt0.5Mn at the same GHSV, which verified the better catalytic performance of 1Pt0.5Mn.It is worth mentioning that CO was the only product detected over both Pt/SiO2 and 1Pt0.5Mneven at a low GHSV (or long residence time), suggesting further hydrogenation of the produced CO to CH4 did not occur.
The stability test was then performed at 400 °C (Figure 3).To achieve a similar initial CO2 conversion, the GHSVs of 15 and 60 L•g −1 •h −1 were selected for Pt/SiO2 and 1Pt0.5Mn,respectively.For Pt/SiO2, continuous deactivation was observed, with the CO2 conversion dropping from the initial 13.6% to 11.5% when the reaction time was increased to 8 h.In contrast, the 1Pt0.5Mncatalyst was rather stable with conversion at ~10% throughout the 8 h test.This result suggested the stability of Pt catalyst was improved with the addition of MnOx.To further confirm the stability of 1Pt0.5Mn, the reaction was also performed at The effect of GHSV on the conversion of CO 2 over Pt/SiO 2 and 1Pt0.5Mn is plotted in Figure 2. In the case of 1Pt0.5Mn, the CO 2 conversion increased from 10.0% to 27.8% when the GHSV decreased from 60 to 15 L•g −1 •h −1 , due to the increased residence time of reactants.Although a similar trend was observed for Pt/SiO 2 , the CO 2 conversion on Pt/SiO 2 was consistently inferior to that on 1Pt0.5Mn at the same GHSV, which verified the better catalytic performance of 1Pt0.5Mn.It is worth mentioning that CO was the only product detected over both Pt/SiO 2 and 1Pt0.5Mneven at a low GHSV (or long residence time), suggesting further hydrogenation of the produced CO to CH 4 did not occur.
Catalysts 2024, 14, x FOR PEER REVIEW 3 of 14 leveled off at ~10% when Mn loading was >0.5 wt%.The selectivity towards CO was 100% without CH4 being detected in any sample.Notably, Mn/SiO2 was inactive under such reaction conditions (consistent with the literature result [39]), suggesting the Mn species acted as a promoter.Considering the high activity and moderate Mn loadings, the 1Pt0.5Mnsample was selected for further study.The mass based intrinsic reaction rate, measured under differential conditions with CO2 conversion < 10% by adjusting the GHSV, on 1Pt0.5Mn(3.01 µmol•g −1 •s −1 ) was 2 times higher than that on Pt/SiO2 (1.54 µmol•g −1 •s −1 ).The effect of GHSV on the conversion of CO2 over Pt/SiO2 and 1Pt0.5Mn is plotted in Figure 2. In the case of 1Pt0.5Mn, the CO2 conversion increased from 10.0% to 27.8% when the GHSV decreased from 60 to 15 L•g −1 •h −1 , due to the increased residence time of reactants.Although a similar trend was observed for Pt/SiO2, the CO2 conversion on Pt/SiO2 was consistently inferior to that on 1Pt0.5Mn at the same GHSV, which verified the better catalytic performance of 1Pt0.5Mn.It is worth mentioning that CO was the only product detected over both Pt/SiO2 and 1Pt0.5Mneven at a low GHSV (or long residence time), suggesting further hydrogenation of the produced CO to CH4 did not occur.
The stability test was then performed at 400 °C (Figure 3).To achieve a similar initial CO2 conversion, the GHSVs of 15 and 60 L•g −1 •h −1 were selected for Pt/SiO2 and 1Pt0.5Mn,respectively.For Pt/SiO2, continuous deactivation was observed, with the CO2 conversion dropping from the initial 13.6% to 11.5% when the reaction time was increased to 8 h.In contrast, the 1Pt0.5Mncatalyst was rather stable with conversion at ~10% throughout the 8 h test.This result suggested the stability of Pt catalyst was improved with the addition of MnOx.To further confirm the stability of 1Pt0.5Mn, the reaction was also performed at The stability test was then performed at 400 • C (Figure 3).To achieve a similar initial CO 2 conversion, the GHSVs of 15 and 60 L•g −1 •h −1 were selected for Pt/SiO 2 and 1Pt0.5Mn,respectively.For Pt/SiO 2 , continuous deactivation was observed, with the CO 2 conversion dropping from the initial 13.6% to 11.5% when the reaction time was increased to 8 h.In contrast, the 1Pt0.5Mncatalyst was rather stable with conversion at ~10% throughout the 8 h test.This result suggested the stability of Pt catalyst was improved with the addition of MnO x .To further confirm the stability of 1Pt0.5Mn, the reaction was also performed at a low GHSV of 15 L•g −1 •h −1 .The CO 2 conversion was much higher, and was indeed slightly increased from 27.8% to 31.3% during the 8 h reaction, with CO as the only product.
a low GHSV of 15 L•g −1 •h −1 .The CO2 conversion was much higher, and was indeed slightly increased from 27.8% to 31.3% during the 8 h reaction, with CO as the only product.The Ea was derived from the Arrhenius plots of intrinsic reaction rates in the temperature range of 360 to 440 °C.Note that CO was the only product under these reaction conditions.As shown in Figure 4, the derived Ea of 1Pt0.5Mn(66.5 kJ•mol −1 ) is substantially lower than that of Pt/SiO2 (81.1 kJ•mol −1 ), indicating that the addition of Mn species to Pt may lower the barrier of the rate limiting step in the RWGS reaction.Such difference may imply different active sites in the two samples [43], i.e., bare Pt sites and interfacial perimeter sites of MnOx/Pt, respectively, as is discussed below.

Catalytic Characterization
To unveil the promotional role of Mn in the RWGS reaction on Pt/SiO2, the catalysts were characterized by various techniques.The XRD patterns of calcined 1PtxMn and Pt/SiO2 catalysts (Figure 5A) showed diffractions at 2θ of 39.8°, 46.2°, 67.6°, and 81.3°, which correspond to metallic Pt(111), ( 200), (220), and (311) facets (PDF-#04-0802), respectively [44].This result indicated that most of the Pt precursors had been decomposed to metallic Pt during calcination.Interestingly, no diffractions of MnOx were present for 1PtxMn and Mn/SiO2 samples, indicating that the MnOx species were highly dispersed.To probe the structure of the MnOx species, Raman spectra of 1PtxMn were collected.As shown in Figure 6A, all samples exhibited a broad band centered at 462 cm −1 , and this band was little affected by the loadings of Mn, reflecting that this band mainly originated The E a was derived from the Arrhenius plots of intrinsic reaction rates in the temperature range of 360 to 440 • C. Note that CO was the only product under these reaction conditions.As shown in Figure 4, the derived E a of 1Pt0.5Mn(66.5 kJ•mol −1 ) is substantially lower than that of Pt/SiO 2 (81.1 kJ•mol −1 ), indicating that the addition of Mn species to Pt may lower the barrier of the rate limiting step in the RWGS reaction.Such difference may imply different active sites in the two samples [43], i.e., bare Pt sites and interfacial perimeter sites of MnO x /Pt, respectively, as is discussed below.
Catalysts 2024, 14, x FOR PEER REVIEW 4 of 14 a low GHSV of 15 L•g −1 •h −1 .The CO2 conversion was much higher, and was indeed slightly increased from 27.8% to 31.3% during the 8 h reaction, with CO as the only product.The Ea was derived from the Arrhenius plots of intrinsic reaction rates in the temperature range of 360 to 440 °C.Note that CO was the only product under these reaction conditions.As shown in Figure 4, the derived Ea of 1Pt0.5Mn(66.5 kJ•mol −1 ) is substantially lower than that of Pt/SiO2 (81.1 kJ•mol −1 ), indicating that the addition of Mn species to Pt may lower the barrier of the rate limiting step in the RWGS reaction.Such difference may imply different active sites in the two samples [43], i.e., bare Pt sites and interfacial perimeter sites of MnOx/Pt, respectively, as is discussed below.

Catalytic Characterization
To unveil the promotional role of Mn in the RWGS reaction on Pt/SiO2, the catalysts were characterized by various techniques.The XRD patterns of calcined 1PtxMn and Pt/SiO2 catalysts (Figure 5A) showed diffractions at 2θ of 39.8°, 46.2°, 67.6°, and 81.3°, which correspond to metallic Pt(111), ( 200), (220), and (311) facets (PDF-#04-0802), respectively [44].This result indicated that most of the Pt precursors had been decomposed to metallic Pt during calcination.Interestingly, no diffractions of MnOx were present for 1PtxMn and Mn/SiO2 samples, indicating that the MnOx species were highly dispersed.To probe the structure of the MnOx species, Raman spectra of 1PtxMn were collected.As shown in Figure 6A, all samples exhibited a broad band centered at 462 cm −1 , and this band was little affected by the loadings of Mn, reflecting that this band mainly originated

Catalytic Characterization
To unveil the promotional role of Mn in the RWGS reaction on Pt/SiO 2 , the catalysts were characterized by various techniques.The XRD patterns of calcined 1PtxMn and Pt/SiO 2 catalysts (Figure 5A) showed diffractions at 2θ of 39.8 • , 46.2 • , 67.6 • , and 81.3 • , which correspond to metallic Pt(111), ( 200), (220), and (311) facets (PDF-#04-0802), respectively [44].This result indicated that most of the Pt precursors had been decomposed to metallic Pt during calcination.Interestingly, no diffractions of MnO x were present for 1PtxMn and Mn/SiO 2 samples, indicating that the MnO x species were highly dispersed.To probe the structure of the MnO x species, Raman spectra of 1PtxMn were collected.As shown in Figure 6A, all samples exhibited a broad band centered at 462 cm −1 , and this band was little affected by the loadings of Mn, reflecting that this band mainly originated from SiO 2 [45].For 1Pt0.1Mn, a weak band at 603 cm −1 was present, which is attributed to the vibration of Mn-O-Si or Mn-O-Pt, suggesting MnO x is below monolayer coverage at low loadings [46].As the Mn content increased from 0.1 to 1 wt%, this band shifted gradually to 636 cm −1 and grew in intensity.Since MnO 2 , Mn 2 O 3 , and Mn 3 O 4 show strong Mn-O vibration between 640 and 650 cm −1 [46][47][48], the band at 636 cm −1 likely originated from a mixture of MnO x oxides.This result also indicates that the polymerization degree of MnO x increases with Mn content and the small well-crystallized bulk MnO x particles are formed when the Mn content exceeds 0.5 wt%.Since no XRD diffractions of MnO x were detected, the particle size of MnO x should be smaller than the XRD detection limitation of 3 nm.
Catalysts 2024, 14, x FOR PEER REVIEW 5 of 14 from SiO2 [45].For 1Pt0.1Mn, a weak band at 603 cm −1 was present, which is attributed to the vibration of Mn-O-Si or Mn-O-Pt, suggesting MnOx is below monolayer coverage at low loadings [46].As the Mn content increased from 0.1 to 1 wt%, this band shifted gradually to 636 cm −1 and grew in intensity.Since MnO2, Mn2O3, and Mn3O4 show strong Mn-O vibration between 640 and 650 cm −1 [46][47][48], the band at 636 cm −1 likely originated from a mixture of MnOx oxides.This result also indicates that the polymerization degree of MnOx increases with Mn content and the small well-crystallized bulk MnOx particles are formed when the Mn content exceeds 0.5 wt%.Since no XRD diffractions of MnOx were detected, the particle size of MnOx should be smaller than the XRD detection limitation of 3 nm.3).H2-TPR was used to analyze the reducibility of 1PtxMn samples (Figure 7).For monometallic Mn/SiO2, two peaks centered at 284 and 378 °C are present, which are typically assigned to stepwise reduction of MnO2 and/or Mn2O3 [42,[49][50][51][52].That is, reduction of MnO2 or Mn2O3 to Mn3O4 first, and then reduction of Mn3O4 to MnO.The corresponding theoretical H2 consumptions, i.e., the H2/Mn molar ratios, are 0.67 or 0.17, and 0.33, respectively.The quantified H2/Mn ratio of the peak at 284 °C is 0.25, which is much lower than 0.67 while it is slightly higher than 0.17.This difference indicates that the Mn/SiO2 sample contains a mixture of Mn2O3 and MnO2, with the former as the dominant component.Our result is also consistent with previous work, which showed the formation of a mixture of MnO2/Mn2O3 on various supports when Mn(NO3)2 was used as the precursor [53].The H2/Mn ratio of the peak at 284 °C is 0.36, which agrees with the value of 0.33,  from SiO2 [45].For 1Pt0.1Mn, a weak band at 603 cm −1 was present, which is attributed to the vibration of Mn-O-Si or Mn-O-Pt, suggesting MnOx is below monolayer coverage at low loadings [46].As the Mn content increased from 0.1 to 1 wt%, this band shifted gradually to 636 cm −1 and grew in intensity.Since MnO2, Mn2O3, and Mn3O4 show strong Mn-O vibration between 640 and 650 cm −1 [46][47][48], the band at 636 cm −1 likely originated from a mixture of MnOx oxides.This result also indicates that the polymerization degree of MnOx increases with Mn content and the small well-crystallized bulk MnOx particles are formed when the Mn content exceeds 0.5 wt%.Since no XRD diffractions of MnOx were detected, the particle size of MnOx should be smaller than the XRD detection limitation of 3 nm.H2-TPR was used to analyze the reducibility of 1PtxMn samples (Figure 7).For monometallic Mn/SiO2, two peaks centered at 284 and 378 °C are present, which are typically assigned to stepwise reduction of MnO2 and/or Mn2O3 [42,[49][50][51][52].That is, reduction of MnO2 or Mn2O3 to Mn3O4 first, and then reduction of Mn3O4 to MnO.The corresponding theoretical H2 consumptions, i.e., the H2/Mn molar ratios, are 0.67 or 0.17, and 0.33, respectively.The quantified H2/Mn ratio of the peak at 284 °C is 0.25, which is much lower than 0.67 while it is slightly higher than 0.17.This difference indicates that the Mn/SiO2 sample contains a mixture of Mn2O3 and MnO2, with the former as the dominant component.Our result is also consistent with previous work, which showed the formation of a mixture of MnO2/Mn2O3 on various supports when Mn(NO3)2 was used as the precursor [53].The H2/Mn ratio of the peak at 284 °C is 0.36, which agrees with the value of 0.33, H 2 -TPR was used to analyze the reducibility of 1PtxMn samples (Figure 7).For monometallic Mn/SiO 2 , two peaks centered at 284 and 378 • C are present, which are typically assigned to stepwise reduction of MnO 2 and/or Mn 2 O 3 [42,[49][50][51][52].That is, reduction of MnO 2 or Mn 2 O 3 to Mn 3 O 4 first, and then reduction of Mn 3 O 4 to MnO.The corresponding theoretical H 2 consumptions, i.e., the H 2 /Mn molar ratios, are 0.67 or 0.17, and 0.33, respectively.The quantified H 2 /Mn ratio of the peak at 284 • C is 0.25, which is much lower than 0.67 while it is slightly higher than 0.17.This difference indicates that the Mn/SiO 2 sample contains a mixture of Mn 2 O 3 and MnO 2 , with the former as the dominant component.Our result is also consistent with previous work, which showed the formation of a mixture of MnO 2 /Mn 2 O 3 on various supports when Mn(NO 3 ) 2 was used as the precursor [53].The H 2 /Mn ratio of the peak at 284 • C is 0.36, which agrees with the value of 0.33, confirming the reduction of Mn 3 O 4 to dominantly MnO.Differently, no obvious reduction peak is observed for Pt/SiO 2 , which could be explained by the fact that the majority of the Pt species were reduced during the stabilization of the TCD signal at room temperature.Interestingly, neither H 2 consumption peaks of PtO x nor MnO x were observed for 1PtxMn catalysts.This result could be interpreted by H spillover from the Pt surface towards vicinal MnO x species and then reduction of MnO x [35].This result also suggests that the MnO x should be in close proximity to Pt to enable the reduction.
Catalysts 2024, 14, x FOR PEER REVIEW 6 of 14 confirming the reduction of Mn3O4 to dominantly MnO.Differently, no obvious reduction peak is observed for Pt/SiO2, which could be explained by the fact that the majority of the Pt species were reduced during the stabilization of the TCD signal at room temperature.Interestingly, neither H2 consumption peaks of PtOx nor MnOx were observed for 1PtxMn catalysts.This result could be interpreted by H spillover from the Pt surface towards vicinal MnOx species and then reduction of MnOx [35].This result also suggests that the MnOx should be in close proximity to Pt to enable the reduction.When the samples were reduced at 400 °C, the XRD patterns (Figure 5B) were similar to those of calcined samples, which allowed us to exclude the formation of PtMn alloy and large MnOx crystallite.The estimated Pt particle sizes using the Scherrer equation are 4.6, 5.2, 4.8, and 4.9 nm for 1Pt/SiO2, 1Pt0.05Mn,1Pt0.5Mn, and 1Pt1Mn, respectively, indicating that the addition of MnOx has little effect on the particle size of Pt.To explore the interactions between Pt and MnOx, the samples were tested by CO chemisorption.The Pt/SiO2 shows a CO uptake of 9.52 µmol•g −1 , which corresponds to a Pt dispersion of 18.6% and agrees well with the value estimated from the Pt particle size.However, no CO uptake was observed for 1Pt0.5Mn,likely due to the low sensitivity of the TCD detector.Thus, the CO adsorption was further followed by FTIR.As shown in Figure 8, no CO adsorption was observed on Mn/SiO2, reflecting CO did not adsorb on MnOx under such conditions.In contrast, a strong band of linear-bound CO at 2071 cm −1 accompanied with a weak band of bridge-bound CO at 1817 cm −1 [9, [54][55][56][57] was present on Pt/SiO2.Notably, the former band was strongly reduced by 84% and the latter band was reduced to being barely observable on 1Pt0.5Mn,indicating the surface of the Pt particle was strongly covered by MnOx since the Pt particle size was similar in the two samples.In combination with the results of CO chemisorption and FTIR, the Pt dispersion was estimated to be 3.0% on 1Pt0.5Mn.Additionally, the linear CO band was blue-shifted to 2075 cm −1 on 1Pt0.5Mnrelative to Pt/SiO2, which likely stemmed from the charge transfer from Pt to MnOx [6,58].One may expect the electron deficient Pt due to vicinal MnOx to weaken the back-donation of the Pt d-orbital toward the antibonding 2π* orbital of CO, resulting in weakened Pt-C bond while strengthening the C-O bond (band blue shift) [55,59,60].As such, the desorption of CO is more favorable on 1Pt0.5Mn.When the samples were reduced at 400 • C, the XRD patterns (Figure 5B) were similar to those of calcined samples, which allowed us to exclude the formation of PtMn alloy and large MnO x crystallite.The estimated Pt particle sizes using the Scherrer equation are 4.6, 5.2, 4.8, and 4.9 nm for 1Pt/SiO 2 , 1Pt0.05Mn, 1Pt0.5Mn, and 1Pt1Mn, respectively, indicating that the addition of MnO x has little effect on the particle size of Pt.To explore the interactions between Pt and MnO x , the samples were tested by CO chemisorption.The Pt/SiO 2 shows a CO uptake of 9.52 µmol•g −1 , which corresponds to a Pt dispersion of 18.6% and agrees well with the value estimated from the Pt particle size.However, no CO uptake was observed for 1Pt0.5Mn,likely due to the low sensitivity of the TCD detector.Thus, the CO adsorption was further followed by FTIR.As shown in Figure 8, no CO adsorption was observed on Mn/SiO 2 , reflecting CO did not adsorb on MnO x under such conditions.In contrast, a strong band of linear-bound CO at 2071 cm −1 accompanied with a weak band of bridge-bound CO at 1817 cm −1 [9, [54][55][56][57] was present on Pt/SiO 2 .Notably, the former band was strongly reduced by 84% and the latter band was reduced to being barely observable on 1Pt0.5Mn,indicating the surface of the Pt particle was strongly covered by MnO x since the Pt particle size was similar in the two samples.In combination with the results of CO chemisorption and FTIR, the Pt dispersion was estimated to be 3.0% on 1Pt0.5Mn.Additionally, the linear CO band was blue-shifted to 2075 cm −1 on 1Pt0.5Mnrelative to Pt/SiO 2 , which likely stemmed from the charge transfer from Pt to MnO x [6,58].One may expect the electron deficient Pt due to vicinal MnO x to weaken the back-donation of the Pt d-orbital toward the antibonding 2π* orbital of CO, resulting in weakened Pt-C bond while strengthening the C-O bond (band blue shift) [55,59,60].As such, the desorption of CO is more favorable on 1Pt0.5Mn.
The interactions between Pt and MnO x were further observed by TEM. Figure 9 shows the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and energy dispersive spectroscopy (EDS) mapping of the 1Pt0.5Mnsample.It can be seen that the Pt aggregated into particles, while the Mn was uniformly distributed (Figure 9A-C).As highlighted in white circles (Figure 9D), the Pt and Mn elements were concentrated in the same area, confirming the direct interaction between Pt with MnO x .Consequently, MnO x may partially cover the surface of Pt and thus lower the accessible surface Pt sites.The interactions between Pt and MnOx were further observed by TEM. Figure 9 shows the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and energy dispersive spectroscopy (EDS) mapping of the 1Pt0.5Mnsample.It can be seen that the Pt aggregated into particles, while the Mn was uniformly distributed (Figure 9A-C).As highlighted in white circles (Figure 9D), the Pt and Mn elements were concentrated in the same area, confirming the direct interaction between Pt with MnOx.Consequently, MnOx may partially cover the surface of Pt and thus lower the accessible surface Pt sites.To further study the electronic interactions between Pt and MnOx, XPS analysis was performed.As shown in Figure 10A, the reduced Pt/SiO2 exhibited two Pt valence states, Pt 0 and Pt δ+ , with the BE of Pt 4f7/2 at 71.2 and 72.8 eV, respectively [10,61,62].Relative to Pt/SiO2, the BE of Pt 0 for 1Pt0.5Mn is slightly higher, which reflects the decreased electron density.The intensity ratio of Pt δ+ /(Pt 0 + Pt δ+ ) is 8.7% for Pt/SiO2, whereas it increases to 15.8% for 1Pt0.5Mn.Taken together, these results reflect the charge transfer from Pt to MnOx in close proximity.As exhibited in Figure 10B, the XPS of the Mn 2p region for Mn/SiO2 can be fitted with three peaks, i.e., Mn 4+ , Mn 3+ , and Mn 2+ at BE of 643.5, 641.9, and The interactions between Pt and MnOx were further observed by TEM. Figure 9 shows the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and energy dispersive spectroscopy (EDS) mapping of the 1Pt0.5Mnsample.It can be seen that the Pt aggregated into particles, while the Mn was uniformly distributed (Figure 9A-C).As highlighted in white circles (Figure 9D), the Pt and Mn elements were concentrated in the same area, confirming the direct interaction between Pt with MnOx.Consequently, MnOx may partially cover the surface of Pt and thus lower the accessible surface Pt sites.To further study the electronic interactions between Pt and MnOx, XPS analysis was performed.As shown in Figure 10A, the reduced Pt/SiO2 exhibited two Pt valence states, Pt 0 and Pt δ+ , with the BE of Pt 4f7/2 at 71.2 and 72.8 eV, respectively [10,61,62].Relative to Pt/SiO2, the BE of Pt 0 for 1Pt0.5Mn is slightly higher, which reflects the decreased electron density.The intensity ratio of Pt δ+ /(Pt 0 + Pt δ+ ) is 8.7% for Pt/SiO2, whereas it increases to 15.8% for 1Pt0.5Mn.Taken together, these results reflect the charge transfer from Pt to MnOx in close proximity.As exhibited in Figure 10B, the XPS of the Mn 2p region for Mn/SiO2 can be fitted with three peaks, i.e., Mn 4+ , Mn 3+ , and Mn 2+ at BE of 643.5, 641.9, and To further study the electronic interactions between Pt and MnO x , XPS analysis was performed.As shown in Figure 10A, the reduced Pt/SiO 2 exhibited two Pt valence states, Pt 0 and Pt δ+ , with the BE of Pt 4f 7/2 at 71.2 and 72.8 eV, respectively [10,61,62].Relative to Pt/SiO 2 , the BE of Pt 0 for 1Pt0.5Mn is slightly higher, which reflects the decreased electron density.The intensity ratio of Pt δ+ /(Pt 0 + Pt δ+ ) is 8.7% for Pt/SiO 2 , whereas it increases to 15.8% for 1Pt0.5Mn.Taken together, these results reflect the charge transfer from Pt to MnO x in close proximity.As exhibited in Figure 10B, the XPS of the Mn 2p region for Mn/SiO 2 can be fitted with three peaks, i.e., Mn 4+ , Mn 3+ , and Mn 2+ at BE of 643.5, 641.9, and 640.1 eV, respectively [51,52,63].The presence of Mn 4+ and Mn 3+ likely resulted from surface oxidation during sample transfer, which reflects the oxophilic property of MnO x which thus may activate the C-O bond of CO 2 .Compared with Mn/SiO 2 , slight shifts in BE of Mn 3+ and Mn 2+ to lower values were observed for 1Pt0.5Mn,indicating charge transfer from Pt to MnO x .In addition, although Mn 3+ is the dominant state for both samples, the 1Pt0.5Mnhas more Mn species with lower valence state than Mn/SiO 2 (Table 1), indicating that the presence of Pt facilitates the reduction of MnO x .Notably, the estimated Mn/Pt atomic ratio on the catalyst surface is 3.44, which is much higher than the theoretical value of 1.78.This result indicates the accumulation of Mn species on the surface, and confirms the coverage of Pt by MnO x .Note that the O 1s spectrum was dominated by O of SiO 2 due to the low loadings of Mn, which makes the analysis of O species in MnO x to be not possible.
BE of Mn 3+ and Mn 2+ to lower values were observed for 1Pt0.5Mn,indicating charge transfer from Pt to MnOx.In addition, although Mn 3+ is the dominant state for both samples, the 1Pt0.5Mnhas more Mn species with lower valence state than Mn/SiO2 (Table 1), indicating that the presence of Pt facilitates the reduction of MnOx.Notably, the estimated Mn/Pt atomic ratio on the catalyst surface is 3.44, which is much higher than the theoretical value of 1.78.This result indicates the accumulation of Mn species on the surface, and confirms the coverage of Pt by MnOx.Note that the O 1s spectrum was dominated by O of SiO2 due to the low loadings of Mn, which makes the analysis of O species in MnOx to be not possible.

Discussion
Based on the catalytic performance and characterization results, an inverse MnOx/Pt structure model of 1PtxMn catalysts is proposed.As shown in Figure 11, the reduced MnOx partially covers the surface of Pt particles, which reduces the number of accessible Pt sites while creating new MnOx/Pt interfacial perimeter sites.In addition, such coverage also results in charge transfer from Pt to MnOx, resulting in weakened CO adsorption.This structure resembles the phenomenon of SMSI, i.e., encapsulation of metal particles by partially reduced metal oxide, which is typically observed during high temperature reduction of metal on a reducible metal oxide support.In this work, the structure was achieved on an inert support of SiO2 with the surface Pt being partially covered by MnOx.Compared with SMSI, the coverage degree can be simply tuned by the amount of MnOx doped to avoid the usage of expensive bulk MnOx support which may reduce the cost of catalyst.In combination of Raman results (Figure 6A) and catalytic performance (Figure 1), it seems that 0.2 wt% Mn (Mn/Pt molar ratio of 0.71) is a critical point of coverage of Pt by MnOx.At low Mn loadings of <0.2 wt%, i.e., below monolayer coverage, the MnOx species is highly dispersed, which is likely in an isolated monomeric state as shown by the

Discussion
Based on the catalytic performance and characterization results, an inverse MnO x /Pt structure model of 1PtxMn catalysts is proposed.As shown in Figure 11, the reduced MnO x partially covers the surface of Pt particles, which reduces the number of accessible Pt sites while creating new MnO x /Pt interfacial perimeter sites.In addition, such coverage also results in charge transfer from Pt to MnO x , resulting in weakened CO adsorption.This structure resembles the phenomenon of SMSI, i.e., encapsulation of metal particles by partially reduced metal oxide, which is typically observed during high temperature reduction of metal on a reducible metal oxide support.In this work, the structure was achieved on an inert support of SiO 2 with the surface Pt being partially covered by MnO x .Compared with SMSI, the coverage degree can be simply tuned by the amount of MnO x doped to avoid the usage of expensive bulk MnO x support which may reduce the cost of catalyst.In combination of Raman results (Figure 6A) and catalytic performance (Figure 1), it seems that 0.2 wt% Mn (Mn/Pt molar ratio of 0.71) is a critical point of coverage of Pt by MnO x .At low Mn loadings of <0.2 wt%, i.e., below monolayer coverage, the MnO x species is highly dispersed, which is likely in an isolated monomeric state as shown by the observation of the Si-O-Mn or Pt-O-Mn vibration band (Figure 6).In this range, the conversion of CO 2 increases almost linearly with increasing Mn due to increased interactions between Pt and MnO x .When the Mn content is >0.2 wt%, bulk MnO x may start to form and most of these species may stay distant from Pt, which has little promotional effect on the RWGS activity.When the Mn loading is ≥0.5 wt%, the CO 2 conversion does not change with further increase of Mn, likely due to formation of MnO x distant from Pt.
The RWGS reaction requires the breakage of one C-O bond of CO 2 and formation of H 2 O by hydrogenation.It is well known that Pt based catalysts are very active for hydrogenation while having low oxophilicity for activation of the C-O bond of CO 2 , resulting in hydrogenation assisted C-O breakage (carboxyl pathway) as the dominant pathway [38].The formation of MnO x /Pt interfacial perimeter sites in 1PtxMn catalysts may provide both the hydrogen functionality on Pt and C-O activation functionality on oxophilic MnO x in close proximity, which may synergistically facilitate the RWGS reaction (Figure 11).Although >84% surface Pt was covered by MnO x in the 1Pt0.5Mnsample, the intrinsic reaction rate is doubled compared with Pt/SiO 2 .Moreover, the estimated turnover frequency (TOF) in terms of accessible Pt sites on the 1Pt0.5Mn(1.97 s −1 ) is twelve times higher than that on Pt/SiO 2 (0.16 s −1 ), and the E a is substantially lowered on 1Pt0.5Mn,confirming the MnO x /Pt interfacial perimeter sites are the active sites.Collectively, these results point to the interfacial MnO x /Pt perimeter sites as being likely to be the active sites, which provide balanced hydrogenation and C-O activation functionalities due to close proximity of the two kinds of sites.Although the TOF in current work is lower than that on Pt-Re/SiO 2 (3.48 s −1 at 400 • C), CH 4 is formed on the latter catalyst [35].The TOF in the current work is also comparable with K promoted Pt/zeolite catalysts (0.1-2.25 s −1 varying with K/Pt atomic ratio), on which CH 4 is also formed [64].Thus, the SiO 2 supported inverse MnO x /Pt catalysts can achieve both high RWGS reaction activity and excellent CO selectivity.
Catalysts 2024, 14, x FOR PEER REVIEW 9 of 14 observation of the Si-O-Mn or Pt-O-Mn vibration band (Figure 6).In this range, the conversion of CO2 increases almost linearly with increasing Mn due to increased interactions between Pt and MnOx.When the Mn content is >0.2 wt%, bulk MnOx may start to form and most of these species may stay distant from Pt, which has little promotional effect on the RWGS activity.When the Mn loading is ≥0.5 wt%, the CO2 conversion does not change with further increase of Mn, likely due to formation of MnOx distant from Pt.The RWGS reaction requires the breakage of one C-O bond of CO2 and formation of H2O by hydrogenation.It is well known that Pt based catalysts are very active for hydrogenation while having low oxophilicity for activation of the C-O bond of CO2, resulting in hydrogenation assisted C-O breakage (carboxyl pathway) as the dominant pathway [38].The formation of MnOx/Pt interfacial perimeter sites in 1PtxMn catalysts may provide both the hydrogen functionality on Pt and C-O activation functionality on oxophilic MnOx in close proximity, which may synergistically facilitate the RWGS reaction (Figure 11).Although >84% surface Pt was covered by MnOx in the 1Pt0.5Mnsample, the intrinsic reaction rate is doubled compared with Pt/SiO2.Moreover, the estimated turnover frequency (TOF) in terms of accessible Pt sites on the 1Pt0.5Mn(1.97 s −1 ) is twelve times higher than that on Pt/SiO2 (0.16 s −1 ), and the Ea is substantially lowered on 1Pt0.5Mn,confirming the MnOx/Pt interfacial perimeter sites are the active sites.Collectively, these results point to the interfacial MnOx/Pt perimeter sites as being likely to be the active sites, which provide balanced hydrogenation and C-O activation functionalities due to close proximity of the two kinds of sites.Although the TOF in current work is lower than that on Pt-Re/SiO2 (3.48 s −1 at 400 °C), CH4 is formed on the latter catalyst [35].The TOF in the current work is also comparable with K promoted Pt/zeolite catalysts (0.1-2.25 s −1 varying with K/Pt atomic ratio), on which CH4 is also formed [64].Thus, the SiO2 supported inverse MnOx/Pt catalysts can achieve both high RWGS reaction activity and excellent CO selectivity.
More importantly, the optimal 1Pt0.5Mnappears to be very stable during the RWGS reaction (Figure 3).Despite the GHSV being four times higher for 1Pt0.5Mn, the Pt particle size estimated from XRD (Figure 5) remained almost constant (4.8 vs 4.9 nm) after 8 h reaction.In contrast, it grew from 4.6 to 5.7 nm for Pt/SiO2.This difference indicates that the presence of adjacent MnOx prevented the sintering of Pt.In addition, the absence of D and G bands (at 1360 and 1580 cm −1 , respectively) of carbon in the Raman spectra (Figure 6B) of used samples excluded the formation of obvious carbon deposition.In conclusion, we demonstrated a simple approach to generate and tune interfacial perimeter sites of More importantly, the optimal 1Pt0.5Mnappears to be very stable during the RWGS reaction (Figure 3).Despite the GHSV being four times higher for 1Pt0.5Mn, the Pt particle size estimated from XRD (Figure 5) remained almost constant (4.8 vs. 4.9 nm) after 8 h reaction.In contrast, it grew from 4.6 to 5.7 nm for Pt/SiO 2 .This difference indicates that the presence of adjacent MnO x prevented the sintering of Pt.In addition, the absence of D and G bands (at 1360 and 1580 cm −1 , respectively) of carbon in the Raman spectra (Figure 6B) of used samples excluded the formation of obvious carbon deposition.In conclusion, we demonstrated a simple approach to generate and tune interfacial perimeter sites of MnO x /Pt, which synergistically enhance the activity and stability of the RWGS reaction with CO as the only product.

Catalyst Preparation
The 1PtxMn/SiO 2 catalysts with different Mn loadings were prepared by an incipient wetness co-impregnation method.The amount of Pt loading was fixed at 1 wt%, while the amount of Mn was varied from 0 to 5 wt%.SiO 2 (Sigma-Aldrich, Burlington, MA, USA, 99.8%, specific surface area of 200 m 2 •g −1 ) was co-impregnated with the appropriate amount of aqueous solution of H 2 PtCl 6 •6H 2 O (Kermel, AR, Tianjin, China) and Mn(NO 3 ) 2 (50% in water, Damao, AR, Tianjin, China) at room temperature for 12 h.The mixture was then dried overnight at 100 • C followed by calcination in air at 450 • C for 4 h.These catalysts were denoted as 1PtxMn, where x is the mass ratio of Mn/Pt.In addition, reference samples of Pt/SiO 2 (1 wt% Pt) and Mn/SiO 2 (1 wt% Mn) were prepared by following the same procedure.

Catalyst Characterization
X-ray diffraction (XRD) patterns were recorded on a MiniFlex 600 diffractometer (Rigaku, Tokyo, Japan), which was operated at 40 kV and 20 mA, with a Cu Kα radiation source (λ = 1.54056Å).The samples were scanned in the 2θ range of 10-90 • at a rate of 2 • •min −1 .Raman spectra were recorded on a RM-1000 instrument (Renishaw, London, UK) equipped with a visible excitation of 532 nm.All samples were ground to powder and dried for 12 h prior to measurement.X-ray photoelectron spectroscopy (XPS) was conducted using a Thermo Scientific K-Alpha system (Waltham, MA, USA) with a chamber base pressure of 2 × 10 −7 mbar.This system was equipped with a monochromated Al Kα (1486.6 eV) X-ray source.All spectra were calibrated with the binding energy (BE) of the adventitious C 1s peak at 284.80 eV.
Hydrogen temperature programmed reduction (H 2 -TPR) was performed using a Chemisorb 2750 (Micromeritics, Norcross, GA, USA) equipped with a thermal conductive detector (TCD) to monitor the consumption of H 2 .A sample of 100 mg was placed in a Utube quartz reactor and pretreated by the flow of Ar (30 mL•min −1 ) at 300 • C for 1 h.After pretreatment, the temperature was cooled down to 30 • C and a 10% H 2 /Ar (30 mL•min −1 ) was introduced to the system.After stabilization of the TCD signal, the temperature was ramped up to 800 • C at a rate of 10 • C•min −1 .CO dynamic pulse chemisorption was conducted on the same equipment.The sample (200 mg) was fixed in a U-tube quartz reactor, pre-reduced at 400 • C in flowing 10% H 2 /Ar for 1 h, held at the same temperature in flowing He for 1 h, and then cooled to room temperature in flowing He.Pulses of 5 vol.%CO/He (500 µL) were injected to the catalyst until saturation.
Fourier transform infrared spectroscopy (FTIR) of CO adsorption was recorded on a PerkinElmer Frontier spectrometer (Waltham, MA, USA) with an in situ transmission reaction cell (Harrick) and a DTGS detector.All spectra were recorded averaging 64 scans and a resolution of 4 cm −1 .The sample wafer was reduced in situ at 400 • C for 1 h, flushed with Ar (90 mL•min −1 ) for another 1 h, and then cooled to room temperature, at which a background spectrum was collected.After that, a 5 vol.%CO/He stream was introduced into the cell for 40 min to saturate the surface of the catalyst.Subsequently, physically adsorbed CO was removed by flowing Ar (25 mL•min −1 ) for 120 min.

Catalytic Activity
The catalytic performance for reduction of CO 2 was evaluated at 400 • C and atmospheric pressure.The catalyst sample (40-60 mesh) was placed in the center of a fixed bed quartz tubular reactor with an outer diameter of 6 mm, and was then reduced at 400 • C for 1 h prior to reaction.Subsequently, the reactants (50 mL•min −1 , molar ratio of CO 2 /H 2 /Ar = 1/4/5) were introduced for a typical run.The outflow passed through an ice-water trap and silica gel column to remove H 2 O, and was then analyzed online on a gas chromatograph (Agilent 7890B, Santa Clara, CA, USA) equipped with a TCD detector and two packed columns (Porapak Q and 5A molecular sieves).To measure the intrinsic reaction rate and apparent activation energy (E a ), the reaction was conducted under differential conditions (conversion < 10%) to exclude both internal and external mass transfer by adjusting GHSV.The carbon balance was maintained at >95% for each run.The mass based intrinsic reaction rate was defined as moles of converted CO 2 per second per gram catalyst.The CO 2 conversion and selectivities of CO and CH 4 were calculated using the following equations [35]

Conclusions
In summary, a series of MnO x /Pt inverse catalysts with varying Mn content were prepared by a co-impregnation method to convert selectively CO 2 to CO via the RWGS reaction.Addition of 0.5 wt% Mn to 1 wt% Pt/SiO 2 improved the intrinsic reaction rate and turnover frequency by two and twelve times, respectively.Characterization results indicate that MnO x partially encapsulates the surface of the Pt particles and the coverage increases with increasing Mn content.Although the surface accessible Pt sites are reduced, new MnO x /Pt interfacial perimeter sites are created, which provide both hydrogenation and C-O activation functionalities synergistically due to close proximity between Pt and MnO x at the interface.Consequently, the apparent activation energy is lowered on the MnO x /Pt catalyst, resulting in improved activity.Moreover, the stability is also significantly improved due to the coverage of Pt by MnO x .

Figure 5 .
Figure 5. XRD patterns of 1PtxMn catalysts: (A) catalysts calcined at 450 • C for 4 h; (B) catalysts reduced at 400 • C for 1 h, as well as used catalysts after the RWGS reaction (see reaction condition in Figure3).

Figure 5 .
Figure5.XRD patterns of 1PtxMn catalysts: (A) catalysts calcined at 450 °C for 4 h; (B) catalysts reduced at 400 °C for 1 h, as well as used catalysts after the RWGS reaction (see reaction condition in Figure3).

Figure 11 .
Figure 11.Schematic illustration of MnOx/Pt interface of 1PtxMn catalysts and RWGS reaction at the interface.

Figure 11 .
Figure 11.Schematic illustration of MnO x /Pt interface of 1PtxMn catalysts and RWGS reaction at the interface.

Table 1 .
Figure 10.XPS spectra of reduced Pt/SiO2 and 1Pt0.5Mncatalysts: (A) Pt 4f, (B) Mn 2p.Valence state distribution of Pt and Mn, as well as atomic ratios derived from XPS.

Table 1 .
Valence state distribution of Pt and Mn, as well as atomic ratios derived from XPS.