Plasma Oxidation of H 2 S over Non-stoichiometric La x MnO 3 Perovskite Catalysts in a Dielectric Barrier Discharge Reactor

In this work, plasma-catalytic removal of H2S over LaxMnO3 (x = 0.90, 0.95, 1, 1.05 and 1.10) has been studied in a coaxial dielectric barrier discharge (DBD) reactor. The non-stoichiometric effect of the LaxMnO3 catalysts on the removal of H2S and sulfur balance in the plasma-catalytic process has been investigated as a function of specific energy density (SED). The integration of the plasma with the LaxMnO3 catalysts significantly enhanced the reaction performance compared to the process using plasma alone. The highest H2S removal of 96.4% and sulfur balance of 90.5% were achieved over the La0.90MnO3 catalyst, while the major products included SO2 and SO3. The missing sulfur could be ascribed to the sulfur deposited on the catalyst surfaces. The non-stoichiometric LaxMnO3 catalyst exhibited larger specific surface areas and smaller crystallite sizes compared to the LaMnO3 catalyst. The non-stoichiometric effect changed their redox properties as the decreased La/Mn ratio favored the transformation of Mn3+ to Mn4+, which contributed to the generation of oxygen vacancies on the catalyst surfaces. The XPS and H2-TPR results confirmed that the Mn-rich catalysts showed the higher relative concentration of surface adsorbed oxygen (Oads) and lower reduction temperature compared to LaMnO3 catalyst. The reaction performance of the plasma-catalytic oxidation of H2S is closely related to the relative concentration of Oads formed on the catalyst surfaces and the reducibility of the catalysts.


Introduction
The emission of odors from various sources including wastewater treatment and municipal solid waste (MSW) treatment facilities have become a public concern due to their negative effect on air quality and human health, especially on sensitive or sick people [1].As a result, air quality control in the waste treatment facilities is important to ensure a comfortable environment for the workers and local residents.Great efforts have been devoted to the research and development of odor abatement technologies including wet scrubbing, active carbon adsorption, incineration and biofiltration, etc. [2][3][4].However, these technologies are not cost-effective when dealing with low concentrations of odors in high-volume waste gas streams.For example, biological treatment is not flexible for the variation of odor loadings and volume of waste gas streams [5].
Recently, non-thermal plasma (NTP) has been regarded as a promising alternative for deodorization due to its unique characteristics of fast reaction, compact system and adaptability towards complex working conditions [6,7].Typically, air plasma could generate electrons and various chemically reactive species, which could react with the odor molecules, leading to the purification of the odor-containing waste gas streams.Anderson et al. performed an on-site test of a pilot scale plasma reactor to purify a 138 m 3 •h −1 odor-containing waste gas stream with the main compositions of acetic acid, propanoic acid, trimethylamine and indole, etc.The average odor removal efficiency of 97% could be achieved at an input power of 140.8 W [8]. Kuwahara et al. developed a dielectric barrier discharge (DBD) reactor with a laminated film-electrode for odor control, while the complete removal of 100 ppm odor was achieved at a discharge power of 10 W and a flow rate of 5 L•min −1 [9].Lu et al. reported the decomposition of ammonia and hydrogen sulfide with self-designed gliding arc plasma reactor, while a removal efficiency of 100% was achieved at the applied voltage of 11 kV and the velocity of 4.72 m•s −1 [10].Some researcher also performed scale-up studies of plasma technologies in odor control [11].Dobslaw et al. performed a 35-day-long scale-up test of plasma odor removal in two industrial sites in Germany, while the odor concentration was significantly reduced by 95.9-98.3%[12].They also used the scale-up plasma units to enhance the performance of a biotrickling filter, the odor removal efficiency of using plasma alone reached up to 93.9% [13].However, the relative low selectivity towards the desired final products in the NTP process remains the main challenge for the use of NTP-based technologies in environmental clean-up.
In the last three decades, the combination of non-thermal plasma and heterogeneous catalysis, known as "plasma-catalysis" has been demonstrated as a promising emerging process for the removal of low concentration gas pollutants including odors with reduced formation of by-products and enhanced process performance [14][15][16][17].Although LaMnO 3 perovskite type catalysts have been used in thermal catalytic reactions due to their relatively low cost, comparable activity, and thermal stability [18][19][20], the use of these catalysts in plasma chemical reactions for environmental clean-up or the synthesis of fuels and chemicals has been very limited.Hueso et al. found that the combination of lanthanum based perovskite catalysts and a microwave discharge plasma enhanced the conversion of low concentration methane (3600 ppm) by around 30% compared to the plasma process in the absence of a catalyst at a same energy density of 10 W•m −2 [21,22].Vandenbroucke et al. reported that the removal of trichloroethylene was increased by 13.9% when packing a Pd/LaMnO 3 catalyst into a negative DC corona discharge reactor at a specific energy density (SED) of 460 J•L −1 compared to the plasma reaction without a catalyst.They also found that the coupling of the plasma-catalyst coupling significantly reduced the formation of major by-product CHCl 3 [23].It is well recognized that the substitution of A and B sites of the perovskite catalysts could result in a structural non-stoichiometry, which might generate excess oxygen species on the catalyst surfaces and consequently affect the reaction performance [24].To the best of our knowledge, the use of perovskite catalysts in plasma-catalytic odor control has not been reported yet, while the knowledge about the underlying mechanisms of the non-stoichiometric effect of the perovskite catalyst on the plasma-catalytic oxidation processes are still missing.
In this work, hydrogen sulfide (H 2 S) is chosen as a model pollutant since it accounts for over 90% of the total mass concentration in the odor emissions [25].The non-stoichiometric effect of La x MnO 3 catalysts on the plasma-catalytic removal of low concentration H 2 S was investigated in a DBD plasma reactor in terms of the removal of H 2 S and sulfur balance.The physicochemical properties of the catalysts were determined using various characterization techniques, including Brunauer-Emmett-Teller (BET) surface measurement, X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS) and temperature programmed reduction of H 2 (H 2 -TPR) to understand the structure-activity relationships between the La x MnO 3 catalysts and the plasma-catalytic process and the role of these catalysts in the plasma-catalytic process.

Physicochemical Properties of the Catalysts
Table 1 show the results of N 2 adsorption-desorption experiments.The non-stoichiometric effect of the La x MnO 3 catalysts enlarged the specific surface areas (S BET ) by 5.7% to 23.5% compared to that of the LaMnO 3 catalyst (12.3 m2 •g −1 ), while the La 0.9 MnO 3 catalyst had the highest S BET .Figure 1 shows the XRD patterns of the La x MnO 3 catalysts.The formation of the perovskite phase was observed in all the catalysts and no obvious segregated phases of La 2 O 3 and MnO x were found.The stoichiometric LaMnO 3 catalyst exhibited a typical cubic perovskite structure (JCPDS 75-0440) of LaMnO 3 .The cubic structure of the La x MnO 3 catalysts remained unchanged when the La doping amount was decreased, while the orthorhombic structure of LaMnO 3 (JCPDS 89-2470) was observed for the La-rich catalysts (x > 1).According to Table 1, the cell volume of the La x MnO 3 catalysts increased with the increase of La/Mn molar ratio.The expansion of unit cell could be attributed to the transformation of Mn 4+ into Mn 3+ and the incorporation of La 3+ into the perovskite lattice of La x MnO 3 considering the larger ion radius of La 3+ (1.06 Å) and Mn 3+ (0.65 Å) compared to that of Mn 4+ (0.54 Å).The characteristic diffraction peak of LaMnO 3 (1 2 1) located at around 32.6 • was slightly shifted to a lower 2θ value for the La-rich catalysts, while this peak in the XRD pattern of the Mn-rich catalysts (x < 1) was shifted to a higher 2θ value [26].The non-stoichiometric catalysts had a smaller crystallite size compared to the LaMnO 3 catalyst (16.1 nm), while the La 0.90 MnO 3 catalyst had the smallest crystallite size of 15.4 nm.Table 1 show the results of N2 adsorption-desorption experiments.The non-stoichiometric effect of the LaxMnO3 catalysts enlarged the specific surface areas (SBET) by 5.7% to 23.5% compared to that of the LaMnO3 catalyst (12.3 m 2 •g −1 ), while the La0.9MnO3 catalyst had the highest SBET. Figure 1 shows the XRD patterns of the LaxMnO3 catalysts.The formation of the perovskite phase was observed in all the catalysts and no obvious segregated phases of La2O3 and MnOx were found.The stoichiometric LaMnO3 catalyst exhibited a typical cubic perovskite structure (JCPDS 75-0440) of LaMnO3.The cubic structure of the LaxMnO3 catalysts remained unchanged when the La doping amount was decreased, while the orthorhombic structure of LaMnO3 (JCPDS 89-2470) was observed for the La-rich catalysts (x > 1).According to Table 1, the cell volume of the LaxMnO3 catalysts increased with the increase of La/Mn molar ratio.The expansion of unit cell could be attributed to the transformation of Mn 4+ into Mn 3+ and the incorporation of La 3+ into the perovskite lattice of LaxMnO3 considering the larger ion radius of La 3+ (1.06 Å) and Mn 3+ (0.65 Å) compared to that of Mn 4+ (0.54 Å).The characteristic diffraction peak of LaMnO3 (1 2 1) located at around 32.6° was slightly shifted to a lower 2θ value for the La-rich catalysts, while this peak in the XRD pattern of the Mn-rich catalysts (x < 1) was shifted to a higher 2θ value [26].The non-stoichiometric catalysts had a smaller crystallite size compared to the LaMnO3 catalyst (16.1 nm), while the La0.90MnO3 catalyst had the smallest crystallite size of 15.4 nm.

Redox Properties of the Catalysts
The chemical states of major elements (Mn 2p and O 1s) in the LaxMnO3 catalysts were examined using XPS, as shown in Figure 2. The binding energies of La 3d5/2 (838.1 eV and 855.0 eV) were observed for all the catalysts (not shown), while those of La 3d3/2 were located at around 834.8 eV and

Redox Properties of the Catalysts
The chemical states of major elements (Mn 2p and O 1s) in the La x MnO 3 catalysts were examined using XPS, as shown in Figure 2. The binding energies of La 3d 5/2 (838.1 eV and 855.0 eV) were observed for all the catalysts (not shown), while those of La 3d 3/2 were located at around 834.8 eV and 851.5 eV.The binding energies and the spin-orbit splitting of La 3d were close to those of pure La 2 O 3 , indicating that the lanthanum ions were in trivalent state of the La x MnO 3 catalysts [27].
concentration of Mn 4+ of these catalysts varied from 38.2% to 43.2%, and the La0.90MnO3 catalyst showed the highest value of 43.2%.The relative Mn 4+ concentration of the catalysts decreased with the increase of nominal La/Mn molar ratio.
The deconvoluted O 1s spectra of the LaxMnO3 perovskites suggested the coexistence of various oxygen species on the catalyst surface (Figure 2b).The peaks located at around 532.9 eV, 531.2 eV and 529.4 eV could be attributed to the formation of oxygen-containing groups: hydroxyl/carbonate species, surface adsorbed oxygen (Oads) and lattice oxygen (Olat), respectively [30].The nonstoichiometric effect did not significantly change the binding energy of these three major oxygen species.Table 1 presents the relative concentration of Oads of the catalysts, defined as Oads/(Oads + Olat).The La0.90MnO3 catalyst showed the highest relative Oads concentration of 60.7%, while the relative concentration of Oads of the catalysts decreased with the increase of nominal La/Mn molar ratio.Note that the variation of Oads/(Oads + Olat) with the La/Mn molar ratio followed the same trend as Mn 4+ /(Mn 3+ + Mn 4+ ).The reducibility of the LaxMnO3 catalysts was analyzed using H2-TPR experiment (Figure 3).For all the LaxMnO3 catalysts, two major reduction peaks were observed.The first broad peak located between 200 °C and 500 °C could be divided into three sub-peaks.The shoulder peak centered between 200 °C and 300 °C could be ascribed to the reduction of weakly adsorbed oxygen species, Figure 2a shows the XPS spectra of Mn 2p 3/2 of the La x MnO 3 catalysts.The co-existence of Mn 2+ , Mn 3+ and Mn 4+ species in the LaMnO 3 -based perovskite catalysts has been reported [28].However, in this work, Mn 2+ species were unlikely existed due to the missing satellite peaks at around 648.8 eV [29].The XPS signals of Mn 2p 3/2 can be deconvoluted into two major peaks.The XPS peaks at 641.6 eV were assigned to the generation of Mn 3+ cations, while the peaks at around 643.0 eV was associated to the formation of Mn 4+ cations.The relative concentration of Mn 4+ in the La x MnO 3 catalysts, defined as Mn 4+ /(Mn 3+ +Mn 4+ ), was calculated based on the deconvoluted peaks (Table 1).The relative concentration of Mn 4+ of these catalysts varied from 38.2% to 43.2%, and the La 0.90 MnO 3 catalyst showed the highest value of 43.2%.The relative Mn 4+ concentration of the catalysts decreased with the increase of nominal La/Mn molar ratio.
The deconvoluted O 1s spectra of the La x MnO 3 perovskites suggested the coexistence of various oxygen species on the catalyst surface (Figure 2b).The peaks located at around 532.9 eV, 531.2 eV and 529.4 eV could be attributed to the formation of oxygen-containing groups: hydroxyl/carbonate species, surface adsorbed oxygen (O ads ) and lattice oxygen (O lat ), respectively [30].The non-stoichiometric effect did not significantly change the binding energy of these three major oxygen species.Table 1 presents the relative concentration of O ads of the catalysts, defined as O ads /(O ads + O lat ).The La 0.90 MnO 3 catalyst showed the highest relative O ads concentration of 60.7%, while the relative concentration of O ads of the catalysts decreased with the increase of nominal La/Mn molar ratio.Note that the variation of O ads /(O ads + O lat ) with the La/Mn molar ratio followed the same trend as Mn 4+ /(Mn 3+ + Mn 4+ ).
The reducibility of the La x MnO 3 catalysts was analyzed using H 2 -TPR experiment (Figure 3).For all the La x MnO 3 catalysts, two major reduction peaks were observed.The first broad peak located between 200 • C and 500 • C could be divided into three sub-peaks.The shoulder peak centered between 200 • C and 300 • C could be ascribed to the reduction of weakly adsorbed oxygen species, while the second peak locate at around 350 • C was associated to the reduction of Mn 4+ to Mn 3+ with concomitant of surface adsorbed oxygen species.The third peak indicated the reduction of Mn 3+ with coordination-unsaturated compose, while the peaks at higher temperature (above 600 • C) can be attributed to further reduction of Mn 3+ to Mn 2+ [27,31].The lowest reduction temperature of the first broad peaks (335 • C and 392 • C) was observed over the La 0.90 MnO 3 catalyst.Increasing the nominal La/Mn molar ratio slightly shifted the reduction peaks to higher temperatures.For example, the reduction temperatures of the LaMnO 3 catalyst were 342 • C and 393 • C, while those for La 1.10 MnO 3 were 351 • C and 401 • C, respectively.This phenomenon indicated stronger interactions between La and Mn species, which inhibited the oxygen mobility on the La x MnO 3 catalysts, and consequently decreased the reducibility of the catalysts.
Catalysts 2018, 8, x FOR PEER REVIEW 5 of 12 while the second peak locate at around 350 °C was associated to the reduction of Mn 4+ to Mn 3+ with concomitant of surface adsorbed oxygen species.The third peak indicated the reduction of Mn 3+ with coordination-unsaturated compose, while the peaks at higher temperature (above 600 °C) can be attributed to further reduction of Mn 3+ to Mn 2+ [27,31].The lowest reduction temperature of the first broad peaks (335 °C and 392 °C) was observed over the La0.90MnO3 catalyst.Increasing the nominal La/Mn molar ratio slightly shifted the reduction peaks to higher temperatures.For example, the reduction temperatures of the LaMnO3 catalyst were 342 °C and 393 °C, while those for La1.10MnO3 were 351 °C and 401 °C, respectively.This phenomenon indicated stronger interactions between La and Mn species, which inhibited the oxygen mobility on the LaxMnO3 catalysts, and consequently decreased the reducibility of the catalysts.

Plasma-Catalytic Oxidation of H2S
The effect of the LaxMnO3 catalysts on the removal of H2S and the sulfur balance of the plasmacatalytic process was shown in Figure 4.The removal of H2S increased monotonically with the increasing SED regardless of the presence of the catalysts.In the plasma reaction without a catalyst, the H2S removal increased significantly from 10.7% to 41.5% when increasing the SED from 304.8 J•L −1 to 604.0 J•L −1 , while the sulfur balance of the plasma process decreased from 98.2% to 74.3%.The input energy in a typical air plasma is mainly used to generate highly energetic electrons and chemically reactive species, including O, OH•, and nitrogen excited states N2(A), which contributed to the decomposition of the H2S through direct electron-impact dissociation of H2S and radical attack [6].The dissociation of H2S molecules by electrons with sufficient energy forms H• and SH• radicals [32], while the collision of H2S with radicals (e.g.O and OH•) and excited nitrogen species also takes place in the plasma reaction [33]: The generated H• radicals could be oxidized to form OH• by the oxidative species and participate

Plasma-Catalytic Oxidation of H 2 S
The effect of the La x MnO 3 catalysts on the removal of H 2 S and the sulfur balance of the plasma-catalytic process was shown in Figure 4.The removal of H 2 S increased monotonically with the increasing SED regardless of the presence of the catalysts.In the plasma reaction without a catalyst, the H 2 S removal increased significantly from 10.7% to 41.5% when increasing the SED from 304.8 J•L −1 to 604.0 J•L −1 , while the sulfur balance of the plasma process decreased from 98.2% to 74.3%.The input energy in a typical air plasma is mainly used to generate highly energetic electrons and chemically reactive species, including O, OH•, and nitrogen excited states N 2 (A), which contributed to the decomposition of the H 2 S through direct electron-impact dissociation of H 2 S and radical attack [6].The dissociation of H 2 S molecules by electrons with sufficient energy forms H• and SH• radicals [32], while the collision of H 2 S with radicals (e.g., O and OH•) and excited nitrogen species also takes place in the plasma reaction [33]: e + H 2 S → e + H• + SH• (1)  Clearly, the LaxMnO3 catalysts played a crucial role in the enhancement of the performance of the plasma-catalytic system.The non-stoichiometric effect slightly increased the specific surface area of the LaxMnO3 catalysts by 3.2 to 20.6% compared to pure LaMnO3 catalyst (12.6 m 2 •g −1 ), while the crystalline size of the LaxMnO3 catalysts was decreased by 0.9-9.8%(16.1 nm for the LaMnO3 catalyst).The changes in the specific surface area and crystalline size of the catalysts indicated that the nonstoichiometric effect contributed to the strong interactions between the La and Mn species of the catalysts.Larger specific surface area could offer more adsorption sites and active sites in the plasmacatalytic reaction, which prolonged the residence time of H2S and intermediates in the plasma reaction [42].Previous study also confirmed that smaller crystallite sizes of the catalysts favored the exposures of active sites on the catalyst surface [43].To sum up, the changes in the physicochemical properties of the LaxMnO3 catalysts would increase the possibilities of effective collisions between the reactive species and the pollutant including H2S and intermediates, which benefited the reaction performance of the plasma-catalysis system.
The XPS spectra confirmed the coexistence of Mn 3+ and Mn 4+ cations in the LaxMnO3 catalysts.Charge imbalance may occur in the non-stoichiometric LaxMnO3 regarding the change of the nominal La/Mn molar ratio in the perovskite structure [18,44].Taking the La0.90MnO3 catalyst as an example, the decrease of La 3+ content could form La vacancies in the perovskite lattice.The form of La vacancies could be compensated by the conversion of Mn 3+ to Mn 4+ to maintain the neutrality, which was evidenced by the production of highest relative concentration of Mn 4+ of 43.2% in the La0.90MnO3 catalyst.Similar results were reported for the La-rich samples as more Mn 3+ can be generated to maintain the electron neutrality in the presence of excess La cations [45].The redox cycle between Mn 3+ and Mn 4+ was closely associated with the redox properties of the catalysts.The value of Mn 4+ /(Mn 3+ + Mn 4+ ) decreased from 43.2% to 38.2% when the La/Mn molar ratio increased from 0.90 to 1.10 in this work.The presence of Mn 4+ species was also correlated with the formation of oxygen vacancies on the surface of the LaxMnO3 catalysts [46].Oxygen vacancies acted as the adsorptiondesorption centers for the generation of Oads from gas-phase oxygen species including O2 molecules and O radicals in the plasma process.The oxygen species adsorbed on the oxygen vacancies (Oads) played an important role in the plasma-induced surface reactions for the oxidation of H2S and the intermediates due to the higher mobility of Oads compared to the lattice oxygen (Olat) [47].Thus, perovskite catalysts with abundant oxygen vacancies were more active for the oxidation of pollutants on the catalyst surface in the plasma region.The La0.90MnO3 catalyst possessed the highest Oads/(Oads + Olat) value of 60.7%, while increasing the nominal La/Mn molar ratio reduced the value to 59.9% for LaMnO3 and 55.2% for La1.10MnO3, respectively.Note the decrease of Oads/(Oads + Olat) at higher La/Mn molar ratio was in line with the decreased Mn 4+ /(Mn 3+ + Mn 4+ ).The generated H• radicals could be oxidized to form OH• by the oxidative species and participate the consequent plasma-induced reactions [6].The SH• radicals were unstable in the oxidative environment and could be reacted with O and OH• radicals to form the major reaction product of SO 2 : The inner energy of the rotational and vibrational excited species, which were insufficient to break chemical bonds of H 2 S molecules, could be transferred to the electronically excited species and accelerate the plasma-catalytic reactions [34].NTP generated by the dielectric barrier discharge reactor consisted of numerous micro-discharges in the plasma region.The number of micro-discharges in each discharge period was increased at a higher SED with fixed gas flow rate, which creates more reaction channels for the decomposition of H 2 S and intermediates, resulting in the enhanced removal of H 2 S [35][36][37].It is worth noting that no ozone was detected in this work, which might be consumed by local heating, catalytic effect, or plasma oxidation reactions [38].The decreased sulfur balance at a higher SED can be ascribed to the formation of sulfur via: (1)  The oxidation of SO 2 to SO 3 by highly oxidative species present in a plasma was reported by Jarrige et al. [39]: The presence of the La x MnO 3 catalysts in the plasma region significantly enhanced the removal of H 2 S and the sulfur balance compared to the plasma processing of H 2 S without a catalyst.Figure 4 shows that the removal of H 2 S increased from 34.3% to 91.6% in the plasma reaction over the stoichiometric LaMnO 3 when increasing the SED from 303.6 J•L −1 to 573.6 J•L −1 , while the sulfur balance of the plasma-catalytic process increased from 59.2% to 87.0% under the same operating conditions.The increase of sulfur balance could be attributed to the enhanced oxidation of solid state sulfur to SO 2 at a higher SED, whilst the formation of a small amount of SO 3 was also observed.However, the formation of SO 3 in the plasma process was relatively low in this work (less than 8%), which was comparable to the results reported by Chang [40] and Kim et al. [41].The non-stoichiometric effect of the La x MnO 3 catalysts imposes a distinct effect on the plasma-catalytic removal of H 2 S. Coupling the plasma with the La 0.90 MnO 3 catalyst showed the highest H 2 S removal of 96.4% and highest sulfur balance of 90.5% at a SED of 593.7 J•L −1 .The increase of nominal La/Mn molar ratio inhibited the performance of the plasma-catalytic system for H 2 S removal.The removal of H 2 S and the sulfur balance follows the order of La 0.90 MnO 3 > La 0.95 MnO 3 > LaMnO 3 > La 1.05 MnO 3 > La 1.10 MnO 3 > NTP in the tested SED range, which is the opposite to the order of La/Mn molar ratio.Part of the missing sulfur element was found on the surface of the spent catalysts by XPS, which was also confirmed by a previous work [14].
Clearly, the La x MnO 3 catalysts played a crucial role in the enhancement of the performance of the plasma-catalytic system.The non-stoichiometric effect slightly increased the specific surface area of the La x MnO 3 catalysts by 3.2 to 20.6% compared to pure LaMnO 3 catalyst (12.6 m 2 •g −1 ), while the crystalline size of the La x MnO 3 catalysts was decreased by 0.9-9.8%(16.1 nm for the LaMnO 3 catalyst).The changes in the specific surface area and crystalline size of the catalysts indicated that the non-stoichiometric effect contributed to the strong interactions between the La and Mn species of the catalysts.Larger specific surface area could offer more adsorption sites and active sites in the plasma-catalytic reaction, which prolonged the residence time of H 2 S and intermediates in the plasma reaction [42].Previous study also confirmed that smaller crystallite sizes of the catalysts favored the exposures of active sites on the catalyst surface [43].To sum up, the changes in the physicochemical properties of the La x MnO 3 catalysts would increase the possibilities of effective collisions between the reactive species and the pollutant including H 2 S and intermediates, which benefited the reaction performance of the plasma-catalysis system.
The XPS spectra confirmed the coexistence of Mn 3+ and Mn 4+ cations in the La x MnO 3 catalysts.Charge imbalance may occur in the non-stoichiometric La x MnO 3 regarding the change of the nominal La/Mn molar ratio in the perovskite structure [18,44].Taking the La 0.90 MnO 3 catalyst as an example, the decrease of La 3+ content could form La vacancies in the perovskite lattice.The form of La vacancies could be compensated by the conversion of Mn 3+ to Mn 4+ to maintain the neutrality, which was evidenced by the production of highest relative concentration of Mn 4+ of 43.2% in the La 0.90 MnO 3 catalyst.Similar results were reported for the La-rich samples as more Mn 3+ can be generated to maintain the electron neutrality in the presence of excess La cations [45].The redox cycle between Mn 3+ and Mn 4+ was closely associated with the redox properties of the catalysts.The value of Mn 4+ /(Mn 3+ + Mn 4+ ) decreased from 43.2% to 38.2% when the La/Mn molar ratio increased from 0.90 to 1.10 in this work.The presence of Mn 4+ species was also correlated with the formation of oxygen vacancies on the surface of the La x MnO 3 catalysts [46].Oxygen vacancies acted as the adsorption-desorption centers for the generation of O ads from gas-phase oxygen species including O 2 molecules and O radicals in the plasma process.The oxygen species adsorbed on the oxygen vacancies (O ads ) played an important role in the plasma-induced surface reactions for the oxidation of H 2 S and the intermediates due to the higher mobility of O ads compared to the lattice oxygen (O lat ) [47].Thus, perovskite catalysts with abundant oxygen vacancies were more active for the oxidation of pollutants on the catalyst surface in the plasma region.The La 0.90 MnO 3 catalyst possessed the highest O ads /(O ads + O lat ) value of 60.7%, while increasing the nominal La/Mn molar ratio reduced the value to 59.9% for LaMnO 3 and 55.2% for La 1.10 MnO 3 , respectively.Note the decrease of O ads /(O ads + O lat ) at higher La/Mn molar ratio was in line with the decreased Mn 4+ /(Mn 3+ + Mn 4+ ).
The reducibility of the La x MnO 3 catalysts was also significantly affected by the non-stoichiometric effect.Figure 3 shows that the reduction temperatures of the Mn-rich catalysts were much lower compared to the La-rich catalysts and LaMnO 3 , indicating that on the Mn-rich catalysts had a higher oxygen mobility.As mentioned before, the compensation effect of the insufficient or excess La species could promote or inhibit the formation of oxygen vacancies, which could contribute to the formation of surface adsorbed oxygen (O ads ) with higher oxygen mobility and consequently contributes to the enhanced oxidation of the adsorbed H 2 S and intermediates to the final product of H 2 O, SO 2 , sulfur and possibly SO 3 in the plasma-catalytic system.It is noteworthy that the order of the reduction temperature of these catalysts was in consistent with the sequence of O ads /(O ads + O lat ) and Mn 4+ /(Mn 3+ + Mn 4+ ).Combining the results of XPS and H 2 -TPR of these catalysts, it is clear that the redox properties of the catalysts play a determining role in the reaction performances of the plasma-catalytic removal of H 2 S, while the non-stoichiometric effect at higher nominal La/Mn ratio inhibited the removal of H 2 S in the plasma-catalytic system.The possible reaction mechanisms in the plasma-catalytic oxidation of H 2 S over the La x MnO 3 catalysts have been summarized in Figure 5.
Catalysts 2018, 8, x FOR PEER REVIEW 8 of 12 much lower compared to the La-rich catalysts and LaMnO3, indicating that on the Mn-rich catalysts had a higher oxygen mobility.As mentioned before, the compensation effect of the insufficient or excess La species could promote or inhibit the formation of oxygen vacancies, which could contribute to the formation of surface adsorbed oxygen (Oads) with higher oxygen mobility and consequently contributes to the enhanced oxidation of the adsorbed H2S and intermediates to the final product of H2O, SO2, sulfur and possibly SO3 in the plasma-catalytic system.It is noteworthy that the order of the reduction temperature of these catalysts was in consistent with the sequence of Oads/(Oads + Olat) and Mn 4+ /(Mn 3+ + Mn 4+ ).Combining the results of XPS and H2-TPR of these catalysts, it is clear that the redox properties of the catalysts play a determining role in the reaction performances of the plasma-catalytic removal of H2S, while the non-stoichiometric effect at higher nominal La/Mn ratio inhibited the removal of H2S in the plasma-catalytic system.The possible reaction mechanisms in the plasma-catalytic oxidation of H2S over the LaxMnO3 catalysts have been summarized in Figure 5.

Catalyst Preparation
The non-stoichiometric LaxMnO3 (x = 0.9, 0.95, 1, 1.05 and 1.1) perovskite-type catalysts were synthesized using a citric acid method, while lanthanum nitrate hexahydrate and 50 wt % manganese nitrate solution were used as the precursors [1].All chemicals were analytic reagent and purchased from Aladdin Co. Ltd. (Shanghai, China) The synthesis procedures of the LaxMnO3 catalysts were: (1) weighted amounts of all precursors and deionized water were mixed to obtain 0.1 M solutions; (2) a 50% excess amount of citric acid (comparing to the total amount of the metal cations) was added to the solution obtained in step 1 as the ligand; (3) the solution was thoroughly stirred for 3 h and evaporated in an 80 °C water bath to form viscous gel; (4) the gel was dried in an oven overnight at 110 °C and then calcined at 700 °C for 5 h; (5) the samples was then pressed and sieved to 35-60 meshes for testing.

Catalyst Characterizations
The specific surface area, average pore size and pore volume of the LaxMnO3 catalysts were measured using N2 adsorption-desorption experiments (Quantachrome Autosorb-1, Boynton Beach, FL, USA) at 77 K.The LaxMnO3 catalyst samples were degassed at 200 °C for 5 h before each test.The specific surface area (SBET) of the LaxMnO3 catalyst was obtained by the Brunauer-Emmett-Teller (BET) equation.
The X-ray diffraction (XRD) measurements were carried out on a Rikagu D/max-2000 X-ray diffractometer (Tokyo, Japan) using Cu-Kα as the radiation source.
The X-ray photoelectron spectroscopy (XPS) experiments were performed with a Thermo ESCALAB 250 spectrometer (Waltham, MA, USA) using an Al K X-ray source (h = 1486.6eV).The XPS results were calibrated using the C 1s spectra at a binding energy (B.E.) of 284.6 eV.

Catalyst Preparation
The non-stoichiometric La x MnO 3 (x = 0.9, 0.95, 1, 1.05 and 1.1) perovskite-type catalysts were synthesized using a citric acid method, while lanthanum nitrate hexahydrate and 50 wt % manganese nitrate solution were used as the precursors [1].All chemicals were analytic reagent and purchased from Aladdin Co. Ltd. (Shanghai, China) The synthesis procedures of the La x MnO 3 catalysts were: (1) weighted amounts of all precursors and deionized water were mixed to obtain 0.1 M solutions; (2) a 50% excess amount of citric acid (comparing to the total amount of the metal cations) was added to the solution obtained in step 1 as the ligand; (3) the solution was thoroughly stirred for 3 h and evaporated in an 80 • C water bath to form viscous gel; (4) the gel was dried in an oven overnight at 110 • C and then calcined at 700 • C for 5 h; (5) the samples was then pressed and sieved to 35-60 meshes for testing.

Catalyst Characterizations
The specific surface area, average pore size and pore volume of the La x MnO 3 catalysts were measured using N 2 adsorption-desorption experiments (Quantachrome Autosorb-1, Boynton Beach, FL, USA) at 77 K.The La x MnO 3 catalyst samples were degassed at 200 • C for 5 h before each test.The specific surface area (S BET ) of the La x MnO 3 catalyst was obtained by the Brunauer-Emmett-Teller (BET) equation.
The X-ray diffraction (XRD) measurements were carried out on a Rikagu D/max-2000 X-ray diffractometer (Tokyo, Japan) using Cu-Kα as the radiation source.
The X-ray photoelectron spectroscopy (XPS) experiments were performed with a Thermo ESCALAB 250 spectrometer (Waltham, MA, USA) using an Al Kα X-ray source (hν = 1486.6eV).The XPS results were calibrated using the C 1s spectra at a binding energy (B.E.) of 284.6 eV.
The hydrogen temperature-programmed reduction (H 2 -TPR) were performed using a Micrometrics Autochem II 2920 instrument (Ottawa, ON, Canada).For each test, 50 mg La x MnO 3 catalyst samples were loaded and pretreated at 250 • C in a flowing N 2 for 1 h to remove the weakly adsorbed impurities and then cooled down to 50 • C. The samples were then heated from room temperature to 800 • C in a 5 vol % H 2 /Ar flow (40 mL•min −1 ).The heating rate was constant at 10 • C•min −1 .The H 2 consumption was calculated based on the H 2 -TPR profiles.

Experimental Set-Up
The schematic experimental set-up is shown in Figure 6.The details of the DBD reactor were described elsewhere [26].The reactor was co-axial type with the discharge length of 60 mm.The inner diameter of the discharge tube was 8 mm, while the outer diameter was 10 mm.The diameter of the stainless high voltage rod was 2 mm.Thus, the total plasma discharge volume was 2.26 mL.The carrier gas (high-purity air) and H 2 S was generated from gas cylinders purchased from Jingong, Hangzhou, China.All gas streams were regulated by mass flow controllers and mixed prior to the plasma reactor.The total flow rate was 2 L•min −1 , while the initial H 2 S concentration was 100 ppm.In this work, 100 mg La x MnO 3 catalyst was used in each test.The catalysts was held in place by quartz wool, while only quartz wool was packed in the plasma region in the cases using plasma alone.The reactor was energized by an AC power supply (Suman CTP2000-K, Nanjing, China) and the discharge frequency was fixed at 10 kHz.The temperature at the outer wall of the reactor was measured using an infrared thermometer (Omega OS540, Norwalk, CT, USA).During the experiment process, it gradually increased to approximately 80 • C, while the outlet gas temperature was about 30 The hydrogen temperature-programmed reduction (H2-TPR) were performed using a Micrometrics Autochem II 2920 instrument (Ottawa, ON, Canada).For each test, 50 mg LaxMnO3 catalyst samples were loaded and pretreated at 250 °C in a flowing N2 for 1 h to remove the weakly adsorbed impurities and then cooled down to 50 °C.The samples were then heated from room temperature to 800 °C in a 5 vol % H2/Ar flow (40 mL•min −1 ).The heating rate was constant at 10 °C•min −1 .The H2 consumption was calculated based on the H2-TPR profiles.

Experimental Set-Up
The schematic experimental set-up is shown in Figure 6.The details of the DBD reactor were described elsewhere [26].The reactor was co-axial type with the discharge length of 60 mm.The inner diameter of the discharge tube was 8 mm, while the outer diameter was 10 mm.The diameter of the stainless high voltage rod was 2 mm.Thus, the total plasma discharge volume was 2.26 mL.The carrier gas (high-purity air) and H2S was generated from gas cylinders purchased from Jingong, Hangzhou, China.All gas streams were regulated by mass flow controllers and mixed prior to the plasma reactor.The total flow rate was 2 L•min −1 , while the initial H2S concentration was 100 ppm.In this work, 100 mg LaxMnO3 catalyst was used in each test.The catalysts was held in place by quartz wool, while only quartz wool was packed in the plasma region in the cases using plasma alone.The reactor was energized by an AC power supply (Suman CTP2000-K, Nanjing, China) and the discharge frequency was fixed at 10 kHz.The temperature at the outer wall of the reactor was measured using an infrared thermometer (Omega OS540, Norwalk, CT, USA).During the experiment process, it gradually increased to approximately 80 °C, while the outlet gas temperature was about 30 °C.The applied voltage across the plasma-catalytic reactor was measured using a Tektronix 4034B digital oscilloscope connected to a Tektronix 6015A voltage probe (1000:1).An external capacitor was connected in series with the grounded electrode across which the voltage was measured using a TPP500 voltage probe (Tektronix, Beaverton, OR, USA).The average discharge power was calculated by integrating the voltage waveforms by Q-U Lissajous method.The specific energy density (SED) of the plasma-catalysis process can be obtained as: where P is the discharge power of plasma-catalytic process and Q is the total gas flow rate.H2S was measured by using a PTM-400 H2S analyzer (Yiyuntian, Shenzhen, China) with the measuring range of 0-200 ppm and an accuracy of ±2%, while the concentration of formed SO2 was measured by using a Testo 350XL gas analyzer (Lenzkirch, Germany) with an accuracy of ±5%.The concentration of outlet SO3 was measured using a SO3 analyzer (RJ-SO3-M) (Ruijing Tech., Shenzhen, The applied voltage across the plasma-catalytic reactor was measured using a Tektronix 4034B digital oscilloscope connected to a Tektronix 6015A voltage probe (1000:1).An external capacitor was connected in series with the grounded electrode across which the voltage was measured using a TPP500 voltage probe (Tektronix, Beaverton, OR, USA).The average discharge power was calculated by integrating the voltage waveforms by Q-U Lissajous method.The specific energy density (SED) of the plasma-catalysis process can be obtained as: where P is the discharge power of plasma-catalytic process and Q is the total gas flow rate.H 2 S was measured by using a PTM-400 H 2 S analyzer (Yiyuntian, Shenzhen, China) with the measuring range of 0-200 ppm and an accuracy of ±2%, while the concentration of formed SO 2 was measured by using a Testo 350XL gas analyzer (Lenzkirch, Germany) with an accuracy of ±5%.The concentration of outlet SO 3 was measured using a SO 3 analyzer (RJ-SO3-M) (Ruijing Tech., Shenzhen, China) based on the US EPA method 8 (specific IPA absorption method for SO 3 measurement).The detection range of the SO 3 analyzer was 0-200 ppm, while the average measuring error was less than 10% [48].All data given in this work was the average value of three repeated measurements.The removal of H 2 S (η H 2 S ) and sulfur balance of the plasma-catalytic oxidation of H 2 S can be defined as: Sul f ur balance(%) = c SO 2 + c SO 3 c in − c out × 100% (16) where c in and c out are the H 2 S concentration at the inlet and outlet of the reactor, respectively, while c SO 2 and c SO 3 are the outlet SO 2 and SO 3 concentrations.

Conclusions
A series of non-stoichiometric La x MnO 3 catalysts were evaluated in the plasma-catalytic oxidative removal of H 2 S using a non-thermal plasma DBD reactor in terms of H 2 S removal and sulfur balance of the plasma process.The coupling of the DBD plasma with the La x MnO 3 catalysts enhanced both H 2 S removal and sulfur balance compared to the plasma process without using a catalyst.The highest H 2 S removal of 96.4% was achieved at a SED of 593.7 J•L −1 when placing the La 0.90 MnO 3 catalyst in the DBD plasma reactor.In addition, the combination of the plasma with the La 0.90 MnO 3 catalyst also showed the highest sulfur balance in the plasma-catalytic process.The non-stoichiometric effect of the La x MnO 3 catalysts significantly affected the removal of H 2 S since the reaction performance of the plasma-catalytic process decreased with the increasing nominal La/Mn molar ratio.Compared to pure LaMnO 3 catalyst, the non-stoichiometric La x MnO 3 catalyst showed enhanced specific surface area and smaller crystallite size.The decreased La/Mn molar ratio in the La x MnO 3 catalysts favored the transformation of Mn 3+ cations to Mn 4+ , which contributed to the formation of oxygen vacancies on the catalyst surfaces.The results of XPS and H 2 -TPR demonstrated that the Mn-rich catalysts possessed higher relative concentration of surface adsorbed oxygen (O ads ) and better reducibility (lower reduction temperature) compared to pure LaMnO 3 , which suggests that the relative concentration of O ads and reducibility of the catalysts played a key role in determining the reaction performance of the plasma-catalytic removal of H 2 S.

Figure 4 .
Figure 4. Effect of SED on plasma-catalytic removal of H2S over LaxMnO3 catalysts: (a) removal efficiency; and (b) sulfur balance of the plasma-catalytic process.

Figure 4 .
Figure 4. Effect of SED on plasma-catalytic removal of H 2 S over La x MnO 3 catalysts: (a) removal efficiency; and (b) sulfur balance of the plasma-catalytic process.

Figure 5 .
Figure 5. Reaction mechanisms of plasma-catalytic removal of H2S over the LaxMnO3 catalysts.

Figure 5 .
Figure 5. Reaction mechanisms of plasma-catalytic removal of H 2 S over the La x MnO 3 catalysts.

Figure 6 .
Figure 6.Schematic diagram of the experimental setup.

Figure 6 .
Figure 6.Schematic diagram of the experimental setup.

Table 1 .
Physicochemical properties of La x MnO 3 catalysts.