Optimizing Citrate Combustion Synthesis of A-Site-Deﬁcient La,Mn-Based Perovskites: Application for Catalytic CH 4 Combustion in Stoichiometric Conditions

: LaMnO 3 -based perovskites are widely recognized as promising catalysts for several oxidation reactions, but the ﬁnal physicochemical and catalytic properties can be greatly inﬂuenced by the adopted synthesis procedure. In this work, a series of A-site-deﬁcient perovskites of composition La 0.8 MnO 3 and La 0.8 Mn 0.9 B 0.1 O 3 (B = Ni, Cu) were prepared through the citrate combustion route with variations in two synthesis parameters: a citric acid/metal cations molar ratio (CA/M) of either 1.1 or 1.5 and either acidic (given by HNO 3 + citric acid) or neutral (after NH 3 addition) pH of the precursor solution. The obtained samples were characterized by XRD, H 2 -TPR, O 2 -TPD, N 2 physisorption, SEM-EDX and XPS. Acidic pH coupled with a CA/M ratio of 1.1 clearly emerged superior among all the other combinations of the two parameters, resulting in smaller crystallite size, higher surface area and porosity, enhanced Mn 4+ reducibility and the ability to release oxygen species; these features were even further improved by B-site substitution with 10 mol% Ni and Cu cations. The synthesized catalysts were tested in CH 4 oxidation to CO 2 under stoichiometric O 2 , conﬁrming the great superiority of samples prepared in acidic pH with a CA/M ratio of 1.1. Ni and Cu doping had a beneﬁcial effect on catalytic activity, which, however, was more evident for less optimized perovskites (acidic pH and CA/M ratio of 1.5), without signiﬁcance differences among the two dopants.


Introduction
Perovskite oxides with the general formula ABO 3 have attracted much attention in the last decades as catalysts in many reactions because of their stable crystal structure, hydrothermal stability, redox activity, great composition tunability and defect engineering [1,2].Among them, LaMnO 3 -based perovskites have been widely studied as promising catalysts for several oxidation (to CO 2 ) reactions, of, e.g., CO, volatile organic compounds (VOCs) and hydrocarbons (HCs) [3][4][5].Factors like exposed surface area, cation reducibility, the content of highly valent cations (e.g., Mn 4+ compared to the main Mn 3+ ) and the presence of mobile surface and lattice oxygen species have been reported to be decisive in perovskites to improve their catalytic activity in oxidation reactions [6][7][8][9][10][11].However, such physicochemical properties can be greatly influenced by the adopted synthesis protocol, which must be properly optimized.The citrate combustion (or citrate sol-gel) route [12] has been extensively employed for the preparation of high-purity and crystalline oxide powders at a relatively low calcination temperatures (e.g., 600-900 • C), benefiting from the ability of citrate anions to complex and homogenously disperse metal cations, with the formation of a gel, which is then ignited at a suitable temperature through an in situ redox reaction between citric acid (fuel) and nitrates (oxidizer) used as precursors.The composition of the obtained gel, its combustion rate and heat generated upon burning depend on the fuel/oxidizer molar ratio, also affecting the final perovskite's textural properties.For instance, Taguchi et al. prepared LaMnO 3 with a citric acid/sum of metal cations (CA/M) molar ratio of between 0.016 and 1.07 (moles of citric acid from 0.0023 to 0.015 in the paper), showing that the crystallite size reached a minimum and the surface area a maximum at CA/M of about 0.5-0.6 (0.007 and 0.008 moles of citric acid), above which the combustion heat increased, negatively affecting the surface area [13].Ghiasi et al. prepared LaMnO 3 with a citric acid/sum of nitrates ratio from 1.0 to 5.0 (corresponding to CA/M of 2.5 to 12.5), highlighting that the stoichiometric amount of citric acid (1.0 vs. nitrates, 2.5 vs. metal cations) was the best for lowering crystallite size and particle agglomeration, whereas too high an amount of citric acid degraded the perovskite structure because of excessive heat of combustion [14].Sihaib et al. prepared LaMnO 3 with a CA/M ratio from 0.5 to 2.0, evidencing that intermediate CA/M values of 1.2 to 1.5 led to the lowest crystallite size, enhanced surface area, greater Mn 4+ reducibility, the highest Mn/La surface ratio and, accordingly, better catalytic activity in toluene and propene oxidation [15].When NH 3 was added to the LaMnO 3 precursor solution to neutralize the pH, as in the work of Li et al. with citric acid/nitrates ratio ranging from 0.5 to 3.0 (corresponding to CA/M of 1.25 to 7.5), the lowest particle size was obtained with a lower amount of citric acid (0.5 and 1.0 vs. nitrates, 1.25 and 2.5 vs. metal cations), although the phase purity was the maximum at a citric acid/nitrate ratio of 1.5 (CA/M = 3.75) [16].These works point out the necessity of finding the right fuel/oxidizer molar ratio, which can vary depending on the other synthesis parameters (like precursor solution pH and calcination temperature).Furthermore, charge imbalances at the A-site, achieved by aliovalent doping (e.g., Sr 2+ partially replacing La 3+ ) or by means of A-site cation deficiency, are known to increase the number of oxygen vacancies, which is useful for improving reactant adsorption and enhancing oxygen mobility [1,5,17].A-site deficiency can also be desirable to counteract the usually observed La surface enrichment (non-redox active) and to increase the B-site cation oxidation state and reducibility [9,18,19].Accordingly, LaMn 1.2 O 3+δ (or La 0.83 MnO 3 ) was found to be more active than stoichiometric LaMnO 3 towards CO and C 3 H 8 oxidation [19].A partial replacement of Mn at the B-site by either Ni or Cu cations was also shown to improve HCs' oxidation capability, thanks to improved surface area, reducibility and surface oxygen mobility [20,21] Based on the above considerations, in the present work, a series of A-site-deficient La 0.8 MnO 3 perovskites were synthesized by the citrate combustion route, with variations in terms of two synthesis parameters: a CA/M ratio of either 1.1 or 1.5 and either an acidic (given by HNO 3 and citric acid) or neutral (upon NH 3 addition) pH of the precursor solution.B-site doping with Ni and Cu cations was performed as well, preparing perovskites with the composition La 0.8 Mn 0.9 Ni 0.1 O 3 and La 0.8 Mn 0.9 Cu 0.1 O 3 .The precursor gel decomposition process was analyzed by TGA, and the calcined samples were deeply characterized by XRD, N 2 physisorption, SEM-EDX, XPS, H 2 -TPR and O 2 -TPD techniques.The influence of the synthesis parameters (pH, CA/M ratio) and of B-site doping on physicochemical properties and on catalytic activity towards CH 4 combustion with stoichiometric O 2 was evaluated.

Structural and Morphological Features
A list of the eight prepared perovskites and their synthesis conditions are reported in Table 1.The subscript 'H + ' is used to refer to samples synthesized in acidic precursor solution before the gel formation, as determined by HNO 3 and citric acid, whereas the subscript 'NH 3 indicates samples prepared through pH neutralization by NH 3 addition before the gel formation (see Section 3.1 Synthesis protocol).The molar ratio citric acid/sum of metal cations (either 1.1 or 1.5) is indicated by the notation 'CA1.1 and 'CA1.5 .Let us note that the variation in the amount of citric acid (CA1.1 or CA1.5) in the acidic precursor solution did not substantially change the final pH, which was always <1 and probably even negative.This was primarily governed by the strong acid HNO 3 .The ignition temperature was varied depending on whether or not NH 3 was introduced into the precursor solution, as further explained in Section 2.1.2.The chosen calcination temperature (750 • C) was a compromise between the necessity of approaching the conditions employed in the catalytic test (up to 800 • C) and the need to avoid excessive sintering induced by high temperatures.XRD patterns of the synthesized perovskites are reported in Figure 1.For all of them, the recorded diffractograms match with the hexagonal perovskite phase La 0.887 MnO 3 (space group R-3c:H, COD reference 1531294), with the characteristic reflection splitting of hexagonal LaMnO 3 -type structures (see Figure S1).Additionally, for all La 0.8 MnO 3 samples except H + -CA1.1, weak reflections at 2θ = 36.1 • and 2θ = 59.9 • are probably ascribed to the presence of small Mn 3 O 4 impurities, making these compositions less A-site-deficient than expected by stoichiometry.In the case of Ni-doped samples, no evident secondary phases emerged from the XRD patterns, which suggests the complete Ni 2+ (or possibly Ni 3+ ) incorporation inside the perovskite lattice at 10 mol% doping.Instead, Cu-doped samples seem to display a weak reflection at 2θ = 36.0• ascribed to the CuMn 2 O 4 spinel phase.Thus, Cu 2+ cations are not completely incorporated in the perovskite framework, probably because of a larger size difference with respect to Mn 3+ main cations than in the case of Ni doping (0.72 Å Mn 3+ , 0.87 Å Cu 2+ , 0.83 Å Ni 2+ , 0.74 Å Ni 3+ [22]).It is interesting to note that samples prepared in acidic pH, and especially those with CA1.1, exhibit broader reflections and a less defined splitting, suggesting the presence of smaller crystallites with respect to samples prepared with NH 3 neutralization.Indeed, average crystallite size values reported in Table 2, calculated by means of the Williamson-Hall method, support this observation: the grain size tends to decrease for acidic pH synthesis and especially when there is less citric acid (CA1.1 samples), and the size tends to be even smaller after Ni and Cu doping compared to undoped La 0.8 MnO 3 prepared with the same synthesis protocol, as already found in the literature [20,21].The lowest crystallite size, 23 nm, was obtained for La 0.8 Mn 0.9 Ni 0.1 O 3 -H + -CA1.1; the highest, 82 nm, for La 0.8 MnO 3 -NH 3 -CA1.5.For the sake of comparison, Ghiasi et al. obtained LaMnO 3 with an average grain size ranging from 30 to 70 nm depending on the amount of citric acid and the calcination temperature [14]; Gholizadeh prepared LaMnO 3 calcined at 900 • C with an average size of 53 nm [23]; and Frozandeh-Mehr et al. obtained a crystallite size of just 14 nm for LaMnO 3 with a low calcination temperature of 600 • C [24].Thus, crystallite sizes obtained in the present work are in reasonable agreement with those reported in literature for similar compositions and synthesis methods.In Figure 2, N 2 adsorption-desorption isotherms of the eight samples are reported, and the corresponding pore size distributions calculated from the desorption branches are shown in Figure S2.Specific surface area (SSA) and total pore volume, determined by BET and BJH methods, respectively, are listed in Table 2. Adsorption isotherms are similar to type III, with no evident "point B", indicating a weak N 2 -adsorbent interaction; however, a type H3 hysteresis also appears, suggesting the presence of macro/mesopores probably formed by loose aggregates of particles [25].Despite the similar shape of isotherms and hysteresis loops regardless of the synthesis protocol, SSA and pore volume can vary to a relatively large extent.Among the four La 0.8 MnO 3 samples, H + -CA1.1 clearly emerges superior for its much higher SSA (25 m 2 /g) and pore volume, both values being at least five times higher than those of the other three samples (4-5 m 2 /g), which is related to a larger amount of mesopores, as shown in Figure S2.In general, the synthesis in acidic pH seems to be preferable compared to synthesis with NH 3 neutralization, although a higher amount of citric acid (1.In some cases, La-deficient compositions were reported to have higher surface area values compared to stoichiometric ones.For instance, Chen et al. obtained a value of 23 m 2 /g for La 0.9 MnO 3 , compared to 20 m 2 /g for LaMnO 3 , both calcined at 700 • C; a marked increase in surface area was obtained for La 0.7 FeO 3 compared to LaFeO 3 as well [26].The SSA values obtained in this work are therefore comparable to the findings in the literature.At a parity of CA, B-site doping with Ni or Cu cations is even more beneficial in enhancing SSA and mesoporosity compared to benchmark La and Mn perovskites, pushing SSA values close to or even higher than 40 m 2 /g in the case of less CA (1.1), whereas the enhancement is less marked with CA of 1.5 (SSA of 7-11 m 2 /g).All these observations correlate well with the crystallite size determined from XRD patterns: a decrease in grain size, especially for H + -CA1.1 samples, is accompanied by a marked increase in surface area values and pore volume.Increases in SSA upon Ni or Cu doping were observed in the literature as well, e.g., 26 m 2 /g for LaMn 0.9 Ni 0.1 O 3 compared to 20 m 2 /g for LaMnO 3 , both calcined at 750 • C [20], and 15 m 2 /g for LaMnCu 0.1 O 3 compared to 8 m 2 /g for LaMnO 3 , calcined at 700 • C [21].The SSA values of 44 and 37 m 2 /g obtained in this work for La 0.8 Mn 0.9 Ni 0.1 O 3 -H + -CA1.1 and La 0.8 Mn 0.9 Cu 0.1 O 3 -H + -CA1.1, respectively, are therefore even higher than those in many works in the literature.
Catalysts 2023, 13, x FOR PEER REVIEW 5 of 19 In Figure 2, N2 adsorption-desorption isotherms of the eight samples are reported, and the corresponding pore size distributions calculated from the desorption branches are shown in Figure S2.Specific surface area (SSA) and total pore volume, determined by BET and BJH methods, respectively, are listed in Table 2. Adsorption isotherms are similar to type III, with no evident "point B", indicating a weak N2-adsorbent interaction; however, a type H3 hysteresis also appears, suggesting the presence of macro/mesopores probably formed by loose aggregates of particles [25].Despite the similar shape of isotherms and hysteresis loops regardless of the synthesis protocol, SSA and pore volume can vary to a relatively large extent.Among the four La0.8MnO3 samples, H + -CA1.1 clearly emerges superior for its much higher SSA (25 m 2 /g) and pore volume, both values being at least five times higher than those of the other three samples (4-5 m 2 /g), which is related to a larger amount of mesopores, as shown in Figure S2.In general, the synthesis in acidic pH seems to be preferable compared to synthesis with NH3 neutralization, although a higher amount of citric acid (1.5) reduces the benefits in terms of SSA and porosity.Similar values ranging from 5 to 26 m 2 /g, depending on CA/M ratio, were obtained by Sihaib et al. for LaMnO3 calcined at 750 °C [15], and Taguchi et al. obtained intermediate surface areas of 13-16 m 2 /g (again as a function of the amount of CA) by calcining LaMnO3 at 700 °C [13].In some cases, La-deficient compositions were reported to have higher surface area values compared to stoichiometric ones.For instance, Chen et al. obtained a value of 23 m 2 /g for La0.9MnO3, compared to 20 m 2 /g for LaMnO3, both calcined at 700 °C; a marked increase in surface area was obtained for La0.7FeO3 compared to LaFeO3 as well [26].The SSA values obtained in this work are therefore comparable to the findings in the literature.At a parity of CA, B-site doping with Ni or Cu cations is even more beneficial in enhancing SSA and mesoporosity compared to benchmark La and Mn perovskites, pushing SSA values close to or even higher than 40 m 2 /g in the case of less CA (1.1), whereas the enhancement is less marked with CA of 1.5 (SSA of 7-11 m 2 /g).All these observations correlate well with the crystallite size determined from XRD patterns: a decrease in grain size, especially for H + -CA1.1 samples, is accompanied by a marked increase in surface area values and pore volume.Increases in SSA upon Ni or Cu doping were observed in the literature as well, e.g., 26 m 2 /g for LaMn0.9Ni0.1O3compared to 20 m 2 /g for LaMnO3, both calcined at 750 °C [20], and 15 m 2 /g for LaMnCu0.1O3compared to 8 m 2 /g for LaMnO3, calcined at 700 °C [21].The SSA values of 44 and 37 m 2 /g obtained in this work for La0.8Mn0.9Ni0.1O3-H+ -CA1.1 and La0.8Mn0.9Cu0.1O3-H+ -CA1.1, respectively, are therefore even higher than those in many works in the literature.3a and S3a) appears to be composed of particles with a size distribution from tens to hundreds of nm, which leave pores of variable dimension upon aggregation.In the case of H + -CA1.1 samples (Figures 3b,c and S3c), individual grains are not clearly discernible, in agreement with their lower size calculated from XRD patterns, and many small pores in the meso-range are evident, especially for B-site-doped perovskites (Figures 3c and S3c), in agreement with N2 physisorption results.H + -CA1.5 samples (Figures 3d and S3b,d) have somewhat intermediate characteristics, with a more compact morphology compared to H + -CA1.1 samples, as evidenced by the presence of several-µm-wide plate-like aggregates (Figure S3b,d) and of discernible grains (Figures 3d and S3b), although a certain mesoporosity can still be appreciated, especially for Ni-and Cu-doped samples (Figures 3d and S3d).This is again in agreement with both XRD and N2 physisorption results (e.g., intermediate SSA values and pore volume).Representative SEM micrographs of the eight prepared perovskites are shown in Figures 3 and S3.The morphology of NH 3 samples (Figures 3a and S3a) appears to be composed of particles with a size distribution from tens to hundreds of nm, which leave pores of variable dimension upon aggregation.In the case of H + -CA1.1 samples (Figures 3b,c and S3c), individual grains are not clearly discernible, in agreement with their lower size calculated from XRD patterns, and many small pores in the meso-range are evident, especially for B-site-doped perovskites (Figures 3c and S3c), in agreement with N 2 physisorption results.H + -CA1.5 samples (Figures 3d and S3b,d) have somewhat intermediate characteristics, with a more compact morphology compared to H + -CA1.1 samples, as evidenced by the presence of several-µm-wide plate-like aggregates (Figure S3b,d) and of discernible grains (Figures 3d and S3b), although a certain mesoporosity can still be appreciated, especially for Ni-and Cu-doped samples (Figures 3d and S3d).This is again in agreement with both XRD and N 2 physisorption results (e.g., intermediate SSA values and pore volume).

Precursor Decomposition
Since the calcination temperature was kept equal for all catalysts, the differences in morphological properties at varying synthesis parameters, as evidenced by XRD, N2 physisorption and SEM, are clearly dependent on how the exothermic reaction occurs between nitrate oxidizer and citrate fuel during the perovskite synthesis process.By empir-

Precursor Decomposition
Since the calcination temperature was kept equal for all catalysts, the differences in morphological properties at varying synthesis parameters, as evidenced by XRD, N 2 physisorption and SEM, are clearly dependent on how the exothermic reaction occurs between nitrate oxidizer and citrate fuel during the perovskite synthesis process.By empirical observation, we generally noted that gels of samples prepared in acidic pH required temperatures lower than 250 • C (sometimes just above 150 • C) to start decomposing, with the evolution of gases and a change in color from yellowish/greenish to dark brown, as if carbonized.On the other hand, gels obtained after the addition of NH 3 seemed to require a temperature above 250 • C for this process to occur, at first with the evolution of gases, then with a clearly visible flame, which was not observed in the former case, producing a grey/black ash-like solid.To try to better understand the process of gel decomposition, thermogravimetric analysis (TGA) was carried out on four selected gel precursors after being dried at 120 • C; the resulting weight loss and corresponding first derivative profiles as a function of temperature are reported in Figure 4.As a general trend, a small weight loss of 5-10 % at up to 120-150 • C is ascribed to the removal of residual water of hydration.The main weight loss of up to 50-70 %, occurring between about 150 and 350 • C, is due to the exothermic nitrate-citrate redox reaction, and further weak weight losses can be ascribed to the decomposition of residual carbonates up to ≈500 • C, as well as to the formation of the final perovskite structure with some oxygen release by further increasing the temperature [15].The gel of the La 0.8 MnO 3 -H + -CA1.1 sample shows a very sharp and one-step weight loss at 195 • C because of the nitrate-citrate reaction.For La 0.8 MnO 3 -NH 3 -CA1.1,and especially for La 0.8 MnO 3 -H + -CA1.5, such decomposition occurs in more than one step, extending through a wider temperature range (the former up to 270 • C, the latter up to 350 • C).The delayed onset of decomposition when increasing the CA/M ratio (with acidic pH syntheses) was observed in previous works as well [13,15,16], explained by the lower amount of nitrate oxidizer in proportion to the citrate fuel; the same authors reported that either the longer reaction time or the higher amount of citrate fuel resulted in higher local temperatures, which produced bigger crystallites and decreased the final surface area.The values of crystallite size and SSA that we obtained for La 0.8 MnO 3 -H + -CA1.1 and La 0.8 MnO 3 -H + -CA1.5 are in line with these observations.In the case of the La 0.8 MnO 3 -NH 3 -CA1.1 sample, the TGA profile suggests that the decomposition of the gel's citrate matrix starts to occur before the expected temperature of 250 • C, although the sharpest weight loss occurs right after this temperature, which we postulate might be due to the exothermic decomposition of the NH 4 NO 3 formed in situ [27], further triggering the complete decomposition of the gel.The highest total weight loss observed for this sample is likely due to the initial presence of NH 3 .The higher crystallite size and poorer SSA values observed for NH 3 samples could be again due to the longer decomposition time as observed in TGA plots, or perhaps to the occurrence of the flame, which would be characterized by a very high local temperature, resulting in particle sintering and a decrease in SSA, with only a weak dependency on the CA/M ratio.Even for the gel of the La 0.8 Mn 0.9 Ni 0.1 O 3 -H + -CA1.1 sample, a multiple-step weight loss extending to almost 350 • C was noticed, which is thus above the temperature at which the gel carbonization was observed (as it was for La 0.8 MnO 3 -H + -CA1.5); in fact, such carbonization might occur with the first weight loss centered at ~190 • C. The reason for the lower final crystallite size and higher SSA achieved through Ni doping (as well as Cu doping), compared to La 0.8 MnO 3 -H + -CA1.1, seems not to be clarified by TGA alone; it could be more related to the perovskite unit cell distortion induced by dopant cations rather than to the gel decomposition process.
was noticed, which is thus above the temperature at which the gel carbonization was observed (as it was for La0.8MnO3-H + -CA1.5); in fact, such carbonization might occur with the first weight loss centered at ~190°C.The reason for the lower final crystallite size and higher SSA achieved through Ni doping (as well as Cu doping), compared to La0.8MnO3-H + -CA1.1, seems not to be clarified by TGA alone; it could be more related to the perovskite unit cell distortion induced by dopant cations rather than to the gel decomposition process.

Bulk and Surface Composition
The average perovskite composition until a depth of ~1 µm was explored by energydispersive X-ray analysis (EDX), and the average surface composition (depth of 5-10 nm) was explored by X-ray photoelectron spectroscopy (XPS).Element percentages are reported in Table S1 and Table 3, with and without considering the oxygen content, respectively, which is always excessive at the surface as well as in bulk, as usually occurs for perovskite oxides [28,29].For all samples, all the expected elements (O, La, Mn, Ni/Cu, if any) were detected by both EDX and XPS, without evidence of a significant amount of impurities.In general, EDX reveals an enrichment in La and a deployment of Mn with respect to the nominal amounts, almost irrespective of the synthesis protocol, although this phenomenon seems more prominent in samples prepared in acidic pH.On the contrary, XPS reveals an opposite Mn enrichment at the expense of La, which is an effect of the A-site understoichiometry and could in principle be desirable for catalysis, with the surface being enriched with the redox active species (Mn n+ ) [18].Regarding Ni and Cu dopants, with EDX, there seems to be a slightly deficient amount with H + -CA1.1 synthesis and a slightly excess amount in the case of H + -CA1.5.At the surface, Ni is always deficient and Cu always in excess, likely confirming the complete Ni n+ incorporation inside the perovskite framework and the partial Cu 2+ segregation as CuMn 2 O 4 spinel, as revealed by XRD.Generally speaking, there are no evident correlations between synthesis parameters (CA/M ratio, pH) and the resulting surface and bulk composition.

Bulk Reducibility and Oxygen Mobility
Samples' reducibility was tested by H 2 -TPR.The recorded profiles are reported in Figure 5, and reduction temperatures and quantification of H 2 consumption are listed in Table 4. Starting from the benchmark La 0.8 MnO 3 composition (Figure 4a), two main reduction processes are discernible: Mn 4+ → Mn 3+ reduction in the lower temperature range (~150-550 • C), in parallel with the removal of the excessive lattice oxygen with respect to metal cations, and Mn 3+ → Mn 2+ reduction in the higher temperature range (~600-900 • C), leading to perovskite decomposition into single oxides/hydroxides (La 2 O 3 or La(OH) 3 + MnO) [9,19].Although the described reduction events are the same for all samples, peaks' characteristics (shape, area and position) are quite variable depending on the adopted synthesis protocol.In particular, the peaks tend to shift towards lower temperatures when decreasing the amount of CA (1.1) in the precursor solution and with acidic pH (H + ) versus neutral pH (NH 3 ), indicating a more ready reducibility.Indeed, La 0.8 MnO 3 -H + -CA1.1 is by far the most reducible catalyst among the four, with the Mn 4+ → Mn 3+ reduction anticipated by at least 40 • C compared to the other three; in parallel, its low-temperature H 2 uptake is much enhanced (at the expense of the high-temperature uptake), indicating a greater lattice stabilization of Mn 4+ at the expense of Mn 3+ , whereas the three samples of H + -CA1.5, NH 3 -CA1.1 and NH 3 -CA1.5 show approximately the same H 2 uptake in both T ranges.These results can be well interpreted in terms of the lower crystallite size and much greater surface area and porosity of the H + -CA1.1 samples [15,19], leading to easier adsorption and activation of H 2 during TPR experiments in comparison to the other three samples, which displayed a very similar surface area and a weak Mn 3 O 4 phase segregation, likely decreasing the amount of Mn 4+ in the perovskite lattice.Furthermore, the low-temperature shoulder at 200-250 • C, evidently appearing only in H + -CA1.1 samples, could be attributed to the reduction in highly reducible Mn 4+ species, likely located at the abundant surface or pore sites of this sample; a similar result was obtained for nanocrystalline high-surface area CeO 2 , which was shown to display an additional TPR peak at low temperatures compared to bulk low-surface area CeO 2 [30].
decreasing the amount of CA (1.1) in the precursor solution and with acidic pH (H + ) versus neutral pH (NH3), indicating a more ready reducibility.Indeed, La0.8MnO3-H + -CA1.1 is by far the most reducible catalyst among the four, with the Mn 4+ → Mn 3+ reduction anticipated by at least 40 °C compared to the other three; in parallel, its low-temperature H2 uptake is much enhanced (at the expense of the high-temperature uptake), indicating a greater lattice stabilization of Mn 4+ at the expense of Mn 3+ , whereas the three samples of H + -CA1.5, NH3-CA1.1 and NH3-CA1.5 show approximately the same H2 uptake in both T ranges.These results can be well interpreted in terms of the lower crystallite size and much greater surface area and porosity of the H + -CA1.1 samples [15,19], leading to easier adsorption and activation of H2 during TPR experiments in comparison to the other three samples, which displayed a very similar surface area and a weak Mn3O4 phase segregation, likely decreasing the amount of Mn 4+ in the perovskite lattice.Furthermore, the lowtemperature shoulder at 200-250 °C, evidently appearing only in H + -CA1.1 samples, could be attributed to the reduction in highly reducible Mn 4+ species, likely located at the abundant surface or pore sites of this sample; a similar result was obtained for nanocrystalline high-surface area CeO2, which was shown to display an additional TPR peak at low temperatures compared to bulk low-surface area CeO2 [30].Considering, now, the Ni-and Cu-doped samples, the above written considerations about the influence of textural properties on reducibility are confirmed: at a parity of composition, H + -CA1.1 samples are more reducible and display a greater lattice Mn 4+ stabilization with respect to H + -CA1.5 samples, thanks to lower crystallite size, abundant surface area and porosity, as obtained by XRD and N 2 physisorption.The effect of Cu B-site doping (Figure 3c) is quite evident: Cu 2+ reduction to Cu 0 in two steps, with Cu + as an intermediary, appears on the TPR profile with two shoulders at very low temperatures (range 100-220 • C), confirmed by an increase in the total H 2 uptake at low temperatures as compared to un-doped perovskites.Furthermore, Mn 4+ reduction is greatly anticipated (at more than 100 • C) compared to benchmark La 0.8 MnO 3 and starts right after the complete reduction of Cu, thanks to the phenomenon of H 2 spillover from pre-formed Cu 0 particles to the perovskite support [21].In contrast, Ni doping (Figure 3b) has a less marked effect: there are no evident peaks ascribed to Ni n+ reduction; based on previous literature with similar compositions [31], Ni 2+ → Ni 0 reduction might be located at around 600-650 • C, possibly hidden by the rising part in the Mn 3+ reduction peak.In any case, Ni doping seems to render Mn 4+ species slightly more reducible by anticipating their peak temperature of 20-30 • C compared to benchmark La 0.8 MnO 3 , although it is not clear whether this is an effect of Ni n+ cations themselves, or of the improved surface area and porosity arising from Ni n+ insertion in the lattice, as could be inferred by the enhancement of the low-temperature shoulder previously ascribed to highly reducible Mn 4+ species located at surface/pore sites.Despite such improvement in Mn 4+ reducibility, the total H 2 uptake in the low T region decreases, and that in the high T region increases upon Ni substitution, suggesting the presence of a greater amount of Mn 3+ in the lattice at the expense of Mn 4+ , as found elsewhere in the literature [32].
Apart from cation reducibility, oxygen mobility was further investigated by O 2 -TPD on five selected samples.The O 2 desorption profiles are shown in Figure 6, and the calculation of evolved O 2 amounts at low and high temperatures is reported in Table 5.Let us note that the latter calculation cannot be considered very accurate because of the uncomplete desorption process at the maximum tested temperature of 900 • C, but it has been added for the sake of comparison among the different catalysts.Below 500 • C, the release of weakly bound surface oxygen species (α 1 -O 2 ), likely chemisorbed on oxygen vacancies, can be detected; at intermediate temperatures of approximately 550 to 700 • C, lattice oxygen still coming from the surface or near the surface (α 2 -O 2 ) is desorbed; finally, above 700 • C, the release of bulk lattice oxygen (β-O 2 ) can be appreciated [3], even occurring in more than one step [20].Notably, H + -CA1.1 samples display the highest amount of α 1 -O 2 desorption, especially the Ni-doped sample, in agreement with their easy reducibility at low temperatures and superior surface area values.La 0.8 MnO 3 -H + -CA1.1 has the most prominent α 2 -and β-O 2 release as well, whereas these decrease upon Ni doping; this suggests that the release of bulk lattice oxygen is accompanied by bulk Mn 4+ → Mn 3+ reduction, whose H 2 -TPR peak was less marked on La 0.8 Mn 0.9 Ni 0.1 O 3 compared to La 0.8 MnO 3 .La 0.8 MnO 3 -NH 3 -CA1.1 and La 0.8 MnO 3 -H + -CA1.5 samples do not show an evident α 1 -O 2 peak; it is likely very weak and hidden by experimental noise due to low surface area and delayed Mn 4+ reducibility compared to H + -CA1.1 samples, whereas La 0.8 Mn 0.9 Cu 0.1 O 3 -H + -CA1.5 samples display a higher amount of oxygen desorption in the whole range compared to undoped La 0.8 MnO 3 -H + -CA1.5, thanks to the somewhat improved surface area and Mn 4+ reducibility arising from B-site doping.

Catalytic Activity
The prepared A-site-deficient perovskites were tested as catalysts for the combustion of CH4 to CO2 under a stoichiometric amount of O2, as a model r curring on methane-fed vehicles' three-way catalytic converters (TWC) [33].No ment of the catalyst was performed, but further tests with a pre-reduction und H2 before exposure to the reaction mixture are currently ongoing, possibly mod final catalytic performances.CH4 light-off conversion curves are reported in Fig the values of Tx (temperature necessary to achieve a certain value x of CH4 c and the reaction rates at 600 °C are listed in Table 6.

Catalytic Activity
The prepared A-site-deficient perovskites were tested as catalysts for the complete combustion of CH 4 to CO 2 under a stoichiometric amount of O 2 , as a model reaction occurring on methane-fed vehicles' three-way catalytic converters (TWC) [33].No pretreatment of the catalyst was performed, but further tests with a pre-reduction under diluted H 2 before exposure to the reaction mixture are currently ongoing, possibly modifying the final catalytic performances.CH 4 light-off conversion curves are reported in Figure 7, and the values of T x (temperature necessary to achieve a certain value x of CH 4 conversion) and the reaction rates at 600 • C are listed in Table 6.Before examining the activity of the different samples, it is useful to introduce the mechanism of CH4 (and other HCs) combustion on perovskite oxides.The reaction is reported to follow the Mars-Van Krevelen (MvK) mechanism [34], in which oxygen species on the perovskite surface are the primary active species for the reactant activation and complete oxidation to CO2; the release of oxygen species induces the formation of oxygen  Before examining the activity of the different samples, it is useful to introduce the mechanism of CH 4 (and other HCs) combustion on perovskite oxides.The reaction is reported to follow the Mars-Van Krevelen (MvK) mechanism [34], in which oxygen species on the perovskite surface are the primary active species for the reactant activation and complete oxidation to CO 2 ; the release of oxygen species induces the formation of oxygen vacancies (V O ) on the catalyst surface and the parallel reduction in B-site cations (in this case, Mn 4+ → Mn 3+ ); finally, gas-phase oxygen is dissociatively adsorbed on the perovskite surface, thus replenishing V O and reoxidizing the reduced cations.However, such oxygen species can be different depending on the temperature range: at low temperatures, the mechanism is suprafacial, meaning that the reaction involves weakly bound oxygen species directly adsorbed on the catalyst surface, particularly on V O , which is easily exchangeable with gaseous O 2 ; at high temperatures, the diffusion of lattice oxygen becomes thermally activated, and thus lattice oxygen species present in the sub-surface layers can migrate to the reaction sites to oxidize the reactant and/or replenish the generated V O .Here, the mechanism is referred to as intrafacial [6][7][8]19,35].Therefore, in addition to the amount of exposed surface active sites, other important factors for catalytic activity are the content of highly valent Mn 4+ ions and their ease of reducibility, in connection with a facile release of weakly adsorbed oxygen species at low temperatures, and with a fast mobility of lattice oxygen at high temperatures.
All catalysts produce CO 2 with selectivity close to 100% as a product of CH 4 oxidation (a very low and hardly quantifiable amount of CO was detected in a few cases), since the stoichiometric amount of O 2 in the gaseous mixture and the ability of perovskites to release active oxygen in general prevent uncomplete combustion.Among undoped La 0.8 MnO 3 compositions, the H + -CA1.1 sample is by far the most active, having an anticipated T 50 value of more than 100 • C compared to the other three samples; it is also the only sample that is able to achieve full CH 4 conversion at 800 • C.This is evidently an effect of the superior surface area, porosity, Mn 4+ reducibility and surface and bulk oxygen mobility and is also possibly due to the absence of Mn 3 O 4 segregated phases, as revealed by physicochemical characterizations.The other three samples, NH 3 -CA1.1,NH 3 -CA1.5 and H + -CA1.5, have quite similar activities, since their physicochemical properties are not much different (e.g., SSA = 4-5 m 2 /g for all), although the NH 3 -CA1.1 sample seems to perform slightly better than the other two above 600 • C, possibly thanks to a slightly easier Mn 4+ reducibility (T max at TPR of 387 • C compared to 403-406 • C of H + -CA1.5 and NH 3 -CA1.5).The calculation of specific reaction rates (i.e., at parity of the catalyst weight) at 600 • C confirm the superiority of the H + -CA1.1 sample and the similar performances of the other three samples.However, the trend in the intrinsic rate values (normalized by surface area and thus linked to the single active sites' activity) is not in agreement with a specific rate trend, with H + -CA1.1 having the lowest intrinsic rate.This likely supports the thesis that specific surface area, implying a greater exposure of active sites, is one of the main factors in improving catalytic activity, since it also influences active Mn 4+ ions' reducibility and the ability to release active oxygen species.
Concerning Cu-and Ni-doped perovskites, analogous considerations about the influence of different synthesis protocols on catalytic activities can be formulated.Samples prepared with a lower CA/M ratio (1.1 vs. 1.5) perform much better, thanks to more favorable morphological features and Mn 4+ reducibility.H + -CA1.5 samples are indeed not able to reach complete CH 4 conversion in the tested temperature range and have much greater T 50 values (of at least 85 • C) compared to H + -CA1.1 samples.The comparison with undoped La 0.8 MnO 3 compositions reveals some activity gain upon Ni or Cu doping, which is more evident in the case of H + -CA1.5 samples: the T 50 values are 681 and 664 • C for La 0.8 Mn 0.9 Ni 0.1 O 3 and La 0.8 Mn 0.9 Cu 0.1 O 3 , respectively (vs.750 • C for undoped composition), with Cu doping performing slightly better than Ni doping in the higher temperature region.The better performances are always related to better surface area and porosity resulting from B-site doping, confirmed by the calculation of reaction rates at 600 • C: specific rates (per unit of mass) are greater for doped samples compared to undoped ones, and the reverse is true for intrinsic rates (per unit of exposed surface).O 2 -TPD further proved that La 0.8 Mn 0.9 Cu 0.1 O 3 -H + -CA1.5 can release a higher amount of active surface and bulk lattice oxygen species compared to La 0.8 MnO 3 -H + -CA1.5.The activity enhancement achieved through B-site doping is instead almost negligible for H + -CA1.1 samples, appearing only at low temperatures and to a slight extent: T 10 = 455 • C for both La 0.8 Mn 0.9 Ni 0.1 O 3 and La 0.8 Mn 0.9 Cu 0.1 O 3 , with respect to 489 • C for La 0.8 MnO 3 , whereas T 50 and T 80 are very similar for the three samples.In this case, the improved surface area only brings benefits to the low-temperature activity by enhancing the amount of surface oxygen release, as shown by O 2 -TPD, thus facilitating the suprafacial mechanism; conversely, the mobility of surface and lattice oxygen is not enhanced by B-site doping (at least by Ni doping), thus not accelerating the intrafacial mechanism.A temperature value of around 600 • C might be the threshold for the change in the CH 4 oxidation mechanism under the tested conditions (stoichiometric amount of O 2 ), roughly corresponding to the temperature needed to activate surface lattice oxygen (α 2 -O 2 ) at TPD.Overall, the most active catalyst is La 0.8 Mn 0.9 Ni 0.1 O 3 -H + -CA1.1, which is the catalyst with the highest specific surface area.

Synthesis Protocol
La 0.8 MnO 3 perovskites were prepared according to the citrate combustion route.The selected amount of La 2 O 3 (≥99.9%Sigma-Aldrich, Beijing, China) and Mn(acetate) 2 •4H 2 O (≥99% Sigma-Aldrich, Schnelldorf, Germany) were dissolved in deionized water containing the stoichiometric amount of HNO 3 (≥65% Sigma-Aldrich, Schnelldorf, Germany), as if the corresponding La(NO 3 ) 3 and Mn(NO 3 ) 2 salts were used as precursors.Nitrate salts were generated indirectly in the precursor solution by HNO 3 addition because of the high hygroscopic nature of these salts, making the precise amount of cations more controllable with the chosen precursors La 2 O 3 and Mn(CH 3 COO) 2 .Citric acid monohydrate (CA, ≥99.0%Sigma-Aldrich, Vienna, Austria) was added according to a molar ratio citric acid/sum of metal cations (CA/M) equal to 1.1 or 1.5.For the synthesis under acidic pH, the solution was directly evaporated overnight at ~120 • C, and a gel was formed, which was then decomposed in an oven at 200-230 • C for 1 h, obtaining a dark solid.For the synthesis under neutral pH, the necessary amount of NH 3 (32%, Sigma-Aldrich, Darmstadt, Germany) was added prior to solvent evaporation as above, and the resulting gel was ignited at 350 • C directly on the hotplate for few minutes, obtaining a dark ash-like solid.In both cases, the obtained solid was grinded and ultimately calcined at 750 • C (ramp 6 • C/min) for 6 h under static air.The same protocol was adopted for the preparation of La 0.8 Mn 0.9 Ni 0.1 O 3 and La 0.8 Mn 0.9 Cu 0.1 O 3 by dissolving the appropriate amount of Ni(NO 3 ) 2 •6H 2 O (≥97.0%Sigma-Aldrich, Schnelldorf, Germany) or CuO (≥99.0%Sigma-Aldrich, Tokio, Japan), together with La-and Mn-precursors, in the starting solution.For Ni, the chosen precursor was the nitrate salt and not NiO, since the latter is hardly dissolvable in HNO 3 .A synthesis with NiO was attempted, but it resulted in lower crystalline purity of the final perovskite.For doped samples, only the synthesis in acidic pH was carried out, with a CA/M molar ratio of either 1.1 or 1.5.In the Results and Discussion section, the subscript H + is used to indicate the synthesis under acidic pH, the subscript NH 3 to indicate the synthesis under neutral pH, and the molar ratio of citric acid/metal cations is indicated by CA1.1 and CA1.5 (see Table 1).

Characterizations
The decomposition of some selected dried gels was studied by thermogravimetric analysis (TGA) on a LINSESIS 1000 instrument (Linseis Messgeraete GmbH, Selb, Germany).About 10-15 mg of sample, placed in an alumina crucible, were heated under static air from RT to 750 • C at 10 • C/min.
H 2 temperature-programmed reduction (TPR) was carried out to investigate the materials' reducibility on a Micromeritics Autochem II 2920 instrument (Norcross, GA, USA).50 mg of sample were loaded on a U-shaped quartz reactor, above a layer of quartz wool, and heated from RT to 950 • C (ramp 10 • C/min) under a 50 scm 3 /min flow of 5% H 2 /Ar.The H 2 consumption was monitored with a TCD detector after H 2 O absorption by a cold trap.
O 2 temperature-programmed desorption (TPD) was performed on some selected perovskites with the same set-up as for H 2 -TPR experiments.About 200 mg of sample were pretreated at 500 • C for 30 min in pure He (50 scm 3 /min) and subsequently for 30 min in 5 vol% O 2 /He (20 scm 3 /min).After cooling down to 40 • C in the same environment, O 2 was allowed to be adsorbed for 30 min.The atmosphere was switched to 30 scm 3 /min of He and, after a purging at 40 • C for 15 min to remove physiosorbed oxygen, the sample was finally heated up to 900 • C at 10 • C/min.
Scanning electron microscopy (SEM) images were acquired with a Zeiss SUPRA 40 V P microscope (Zeiss, Oberkochen, Germany), setting the electron acceleration voltage at 5 or 10 kV.Energy-dispersive X-ray analysis (EDX) was coupled with SEM for elemental quantification at 20 kV electron acceleration voltage, probing a large area of sample (rectangle of hundreds of µm on each side); X-ray peaks considered for quantification were O Kα, Mn Kα, La Lα, Ni Kα and Cu Kα.

Catalytic Tests
The prepared catalysts were tested for complete oxidation of CH 4 to CO 2 under stoichiometric O 2 .A fixed-bed quartz reactor (ID 6 mm) was employed, loading 50 mg of sample between two layers of quartz wool, and a K-type thermocouple was placed in contact with the catalyst bed to monitor its temperature.The reaction mixture was composed of 2 vol% CH 4 , 4 vol% O 2 , 16 vol% N 2 and He balance, with a total flow rate of 100 scm 3 /min (weight hourly space velocity WHSV = 120 L/h/g).The catalytic tests were conducted in the temperature range of 300-800 • C without any pre-treatment of the catalyst, heating up at 2 • C/min.The outlet gas mixture was analyzed by an online GC (Agilent 7890A, Santa Clara, CA, USA) equipped with a TCD detector and two packed columns, MolSieve 13X (60/80 mesh, 1.8 m) and Porapak Q (80/100 mesh, 1.8 m), using He as a carrier gas.
Reaction rates of CH 4 conversion were calculated assuming the mass balance of a plug flow reactor (PFR) and a first-order kinetic with respect to the methane concentration, leading to the following Equations ( 1) and (2) for the specific rate (r speci f ic , normalized by catalyst mass) and intrinsic rate (r intrinsic , normalized by catalyst surface area), respectively [36].X CH4 is the methane conversion, m the catalyst mass (g), F CH4,0 the inlet molar flow rate of methane (mol/s) and SSA (m 2 /g) the catalyst specific surface area as determined from the BET method.

Figure 1 .
Figure 1.XRD patterns of (a) La0.8MnO3 samples, with zoom highlighting the presence of a weak Mn3O4 reflection at 2θ = 36.1°;(b) Ni-and Cu-doped samples, with zoom highlighting the presence of a weak CuMn2O4 reflection at 2θ = 36.0°on samples containing Cu.

Figure 1 .
Figure 1.XRD patterns of (a) La 0.8 MnO 3 samples, with zoom highlighting the presence of a weak Mn 3 O 4 reflection at 2θ = 36.1 • ; (b) Ni-and Cu-doped samples, with zoom highlighting the presence of a weak CuMn 2 O 4 reflection at 2θ = 36.0• on samples containing Cu.
5) reduces the benefits in terms of SSA and porosity.Similar values ranging from 5 to 26 m 2 /g, depending on CA/M ratio, were obtained by Sihaib et al. for LaMnO 3 calcined at 750 • C [15], and Taguchi et al. obtained intermediate surface areas of 13-16 m 2 /g (again as a function of the amount of CA) by calcining LaMnO 3 at 700 • C [13].

Figure 2 .
Figure 2. N2 adsorption isotherm of the prepared catalysts: (a) the four La0.8MnO3 samples, with zoom on the high-pressure range of the low SSA samples; (b) Ni-and Cu-doped samples with CA1.1, compared to La0.8MnO3-H + -CA1.1;(c) Ni-and Cu-doped samples with CA1.5, compared to La0.8MnO3-H + -CA1.5.Representative SEM micrographs of the eight prepared perovskites are shown in Figures 3 and S3.The morphology of NH3 samples (Figures3a and S3a) appears to be composed of particles with a size distribution from tens to hundreds of nm, which leave pores of variable dimension upon aggregation.In the case of H + -CA1.1 samples (Figures3b,cand S3c), individual grains are not clearly discernible, in agreement with their lower size calculated from XRD patterns, and many small pores in the meso-range are evident, especially for B-site-doped perovskites (Figures3c and S3c), in agreement with N2 physisorption results.H + -CA1.5 samples (Figures3d and S3b,d) have somewhat intermediate characteristics, with a more compact morphology compared to H + -CA1.1 samples, as evidenced by the presence of several-µm-wide plate-like aggregates (FigureS3b,d) and of discernible grains (Figures3d and S3b), although a certain mesoporosity can still be appreciated, especially for Ni-and Cu-doped samples (Figures3d and S3d).This is again in agreement with both XRD and N2 physisorption results (e.g., intermediate SSA values and pore volume).

Figure 5 .
Figure 5. H 2 -TPR profiles of: (a) the four La 0.8 MnO 3 samples; (b) Ni-doped perovskites, compared to undoped ones prepared with the same protocol; (c) Cu-doped perovskites, compared to undoped ones prepared with the same protocol.

Table 1 .
Prepared samples and synthesis conditions.CA/M* = citric acid/metal cations molar ratio.

Table 2 .
Specific surface area (SSA), total pore volume and average crystallite size (dXRD) of the eight prepared perovskites.

Table 2 .
Specific surface area (SSA), total pore volume and average crystallite size (d XRD ) of the eight prepared perovskites.

Table 3 .
Bulk (EDX) and surface (XPS) composition of the prepared perovskites, compared to the respective nominal amount according to sample stoichiometry.
* Metal percentage without considering the oxygen content.

Table 4 .
Peak temperatures and quantitative analysis of H 2 -TPR experiments.

Table 6 .
CH 4 oxidation performance of the prepared catalysts: temperature of 10%, 50% and 80% CH 4 conversion, specific and intrinsic reaction rates at 600 • C.