Preparing and Optimization of Cerium-Lanthanum-Cobalt Ternary Mixed Oxide as Catalyst for SO 2 Reduction to Sulfur

In this work, various ternary cerium oxide/lanthanum oxide/cobalt oxide (Ce/La/Co) nanocatalysts were synthesized by co-precipitation method based on response surface methodology (RSM). The optimum predicted surface area was found to be 67.6 m g at calcination temperature of 650°C, La content of 10.0 wt%, and Co content of 8.0 wt%. Average crystal size of optimum ternary Ce/La/Co catalyst was estimated 11.4 nm. The confirmation tests revealed that experimental data can be predicted well by the model. Furthermore, the prepared catalysts were evaluated by Brunauer-Emmett-Teller (BET), Barrett-Joyner-Halenda (BJH), X-ray diffraction (XRD), field-emission scanning electron microscopy (FESEM), energy-dispersive X-ray (EDX) and NH3 temperature programmed desorption (NH3-TPD) analyses. The characterization results confirmed that ternary Ce/La/Co catalysts were successfully synthesized. Also, the NH3-TPD result showed that total active sites of optimum ternary Ce/La(10)/Co(8) catalyst with La content of 10.0 wt% and Co content of 8.0 wt% was greater than that of single cerium oxide catalyst. The optimum synthesized catalyst was tested for SO2 reaction by methane to sulfur. SO2 conversion and selectivity of catalysts at various temperatures were determined. The better performance of Ce/La/Co optimum catalyst at different temperatures was obtained for SO2 reduction. Also the selectivity of the optimum catalyst for production of sulfur is better than other catalysts.


I. INTRODUCTION
Ceria-based metal oxide catalysts are widely used in various industrial processes, such as environmental pollution control reactions and reforming of hydrocarbons.Due to versatile properties of cerium oxide in the various chemical reaction applications, using binary and ternary combinations of this metal oxide has increased in recent years.The addition of second and third metals to cerium oxide improves the properties such as increasing the specific surface area or selectivity for the catalytic reactions.
Considering that by adding other metals and making triple compounds of cerium oxide, it is possible to improve the composite properties, the use of a triple compound, especially for catalyst applications, is growing steadily.In this work, combinations of Ce/La/Co as catalyst of SO2 reduction to sulfur by CH4 were synthesized.Due to the advantages of converting sulfur dioxide to sulfur, including the removal of pollutants without producing any inappropriate product and the production of the sulfur with the ability to be sold, this method has received considerable attention in recent years.
The main reaction for SO2 reduction by CH4 is as follows [21]: This reaction is complete methane oxidation by SO2, while partial methane oxidation by SO2 is an in-appropriate reaction and produces toxic H2S as [21]: Thus, good selectivity is the most important parameter for this process.
In the present work, Ce/La/Co ternary oxide nanoparticles were synthesized by co-precipitations method.BET specific surface areas of ternary oxide (as a measure of activity) were optimized by the response surface methodology (RSM).The operating parameters in RSM experimental design were found as percentages of lanthanum, cobalt, and the calcination temperature.Finally, the activity and selectivity of the optimized catalyst for sulfur dioxide reaction with methane were determined in a small packed bed reactor.

B. Catalyst synthesis
Co-precipitation method is used in the present work for synthesis of catalysts [7].After preparation of some mixed metal nitrate solutions, the catalyst particles were precipitated by addition of NaOH solution.Then, the nanoparticles were dried at 120℃ in an oven.Finally, the obtained catalysts were calcined at 616−784℃ for 3 h.It should be mentioned that ternary catalysts synthesized in this work were shown by Ce/La(A)/Co(B), where A and B are the La and Co contents in wt%, respectively.

C. Catalyst characterization
BET areas and adsorption isotherms of the nanoparticles were determined by nitrogen adsorption, at 77 K, using Autosorb-1MP from Quantachrome.BJH pore size distributions were extracted from the above mentioned N2 adsorption isotherms.
X-ray diffraction (XRD) was measured by Equinox 3000 from Inel (40 kV, and between 10−118°).Average crystal size of the nanoparticles was estimated using the Scherrer equation from XRD patterns.
Field emission scanning electron microscopy (FESEM) was used for catalyst particles shape observation, using Hitachi S-4160 scanning microscope.Moreover, compositions of metals in nanoparticles were measured by energy dispersive X-ray (EDX) analysis, using Tescan Vega/II/XMU.It should be mentioned that the appearance of Au peak in EDX graphs was due to gold coating of samples.
NH3 temperature programmed desorption (NH3-TPD) analysis was used to detect the total, weak, medium and strong acidic sites using Nanosord NS91 apparatus.Firstly, 50 mg of the prepared catalyst was located in a U-shaped quartz cell.Then, the sample was dried in a flow of nitrogen at 300℃ for 1 h and cooled to 100℃.After that, the sample was saturated by the adsorption of 5% NH3/He at 100℃ for 30 min.The loosely bounded ammonia on catalyst was removed by purging the sample in a He stream for 30 min at 100℃.Finally, thermal ammonia desorption was performed between 100℃ and 800℃ with 10℃/min in a flow of helium.
The catalyst average crystal size can be evaluated by the Scherrer equation as follows [27]: where d (nm) represents the crystal size of catalysts, β is line broadening determined at half height, λ is the wavelength, and θ denotes the diffraction angle.

D. RSM experimental design
Response surface methodology (RSM) is used to estimate the relative significance of effective factors and show the interactions between parameters.Also, to optimize the operational conditions to achieve the best surface area of ternary catalysts, RSM is applied.Furthermore, the Central Composite Designs (CCD) coupled with RSM was employed as an effective approach to evaluate the process conditions and estimate a quadratic model for the response variable [28].In this research, three independent factors including lanthanum content (8−16 wt%), cobalt content (8−16 wt%) and calcination temperature (650−750℃) were used for reaching to a maximum BET surface area.An optimal number of experiments (N) in CCD method can be calculated as follows [29]: where k is the number of variables; cp is the number of center points applied to estimate the standard deviation; 2 k is the factorial points and 2k is the axial points [29].It should be mentioned that the higher center points lead to a reduction in pure error.So, CCD with three independent factors involved 2 3 (= 8) full factorial tests, 6 axial points, and 6 replicates at the center points was used.The input variables including x1, x2, ⋯ , xn should be scaled to coded levels.In this study, all three factors in CCD investigated at five-coded levels including −α, −0.5, 0, +0.5, and +α.The coded values relate to the real ones as follows [28]: where represents the dimensionless coded value of the ith variable; Xi is the real value; X0 is center point real value; and ΔX shows the value of step change.The quadratic polynomial function was applied to fit the experimental data as follows [28]: where Y refers to the predicted response; β0, βi, βii, and βij indicate the regression coefficients (intercept, linear, second-order, and interaction effects, respectively); z is the number of factors; Xi represents the coded levels of the independent factors; XiXj shows the effect of interactions between parameters and Xi 2 represents the square effect.In order to consider the adequacy of applied regression models,  analysis of variance (ANOVA) was applied to BET surface area.The statistical significance of models was checked by F value and P value at 95% confidence.Design-Expert software was used to analyze the data.

E. Performance of catalysts
All performance tests were carried out in a small fixed-bed steel reactor.In each test, 300 mg of catalyst sample was used.The schematic flow diagram of the catalyst performance setup is illustrated in Figure 1.
At first, an inert gas flow is used for purging of the reactor.Then, the reactor is heated to the final appropriate temperature under combination of CH4, SO2, and argon streams with predefined compositions.SO2, CH4, and argon inlet flows were regulated by three mass flow controllers.A portion of outlet stream was conducted to a mass spectrometer (MS, from Leda Mass) for online analysis.
Mole fraction profiles of up to twelve different gases can be plotted versus time (with ppm sensitivity), after converting base peak heights to the partial pressures using cracking patterns and ionization sensitivities in MS [30].
For determination of SO2 conversion, the following relation was used: VSO2in and VSO2out are volumetric velocity of SO2 at the reactor inlet and outlet, respectively.

A. Optimization of BET surface area of ternary catalysts by RSM
The simultaneous effects of three independent variables including calcination temperature (X1), weight percentage of La (X2) and weight percentage of Co (X3) were studied using RSM according to CCD.The results of predicted and real BET surface area are given in Table 1.Final coded model is given as follows (Y is the BET surface area for ternary Ce/La/Co catalysts): Results depicted that the calcination temperature with highest coefficient (−9.699) showed the most important influence on the BET surface area.On the other hand, a negative sign for calcination temperature indicated that the surface area decreased with the increase of X1 parameter.ANOVA was applied to predict the surface area of ternary catalysts and the results are given in Table 2. Based on ANOVA analysis, P-values predicted for variables X1, X2, X3, X1X3, and X1 2 were less than 0.05, showing the importance of these variables in the surface area of catalysts.Also, P-values of X1X2, X2X3, X2 2 , and X3 2 were greater than 0.05, showing that they were insignificant terms in second-order model.The F-value and P-value of the applied model were found to be 86.23  good fitting of the model.The fit of polynomial Eq. ( 8) was evaluated by the correlation coefficient (R 2 ).In this study, R 2 value was found to be 0.987.This value indicated that only 1.3% of the total variables were not clarified by the obtained BET surface area.The actual and the predicted BET surface area are depicted in Figure 2. The predicted values were determined using the approximating functions and the actual values were experimental data.As shown, the points were located around the y = x line which displayed the correlation of the actual and predicted data and confirmed that the proposed model was applicable for the response prediction [31].Also, the lack of fit of model was greater than 0.05 which indicated that the quadratic model was adequate for predicting BET surface area.
The 3D surface plots indicate the interactive influences of two independent parameters (Figure 3). Figure 3(a) indicates the effect of calcination temperature and La weight percentage on BET surface area.As shown, the maximum surface area was obtained at 650℃ and La weight percentage of 10.0 wt%.The predicted response of surface area decreased with increasing the calcination temperature.This reduction can be due to the ternary metal oxide sintering at higher calcination temperatures [7].Also, the reduction of surface area at La > 10.0 wt% can be due to the larger crystal formation (confirmed by XRD results).
The effect of calcination temperature and Co weight percentage on the surface area is illustrated in Figure 3(b).As observed, the surface area decreased with the increase of calcination temperature and Co wt%.Therefore, the optimum Co wt% was found to be 8.0 wt%.
Figure 3(c) shows the effect of La and Co contents on BET surface area.This figure confirmed the results mentioned above.The highest predicted BET surface area was found to be 67.6 m 2 g −1 which obtained under optimum conditions (calcination temperature of 650℃, La content of 10.0 wt% and Co content of 8.0 wt%).As shown in Table 1, the optimum BET surface area was not included in the 20 runs designed by RSM.However, RSM method optimized the BET surface area according to these 20 runs.Run 21 in Table 1 was at these optimum conditions.
For verification of the applied model results, the confirmation experiments were accomplished at optimum conditions.The experimental value of surface area under optimum conditions was found to be 69.7 m 2 g −1 that is in close agreement of the predicted value within 3.2% of error.These estimated and acceptable errors confirmed the validity of the model.
As shown in Table 1, the real and predicted BET surface areas with Co contents lower than 8 wt% (5.27 wt%) were found to be 65.3 m 2 g −1 and 62.9 m 2 g −1 in Run 14, respectively.Also, the real and predicted BET surface areas with calcination temperature lower than 650℃ (615.9℃) were found to be 69.5 m 2 g −1 and 66.4 m 2 g −1 in Run 19, respectively.These runs showed that these obtained BET surface areas were lower than BET surface area under optimum conditions (Run 21 in Table 1).222), (400), and (311) for all catalysts corresponded to the cerium oxide phase.Similar crystal surface was obtained by others [27].A weak peak appeared at around 2θ = 31.0°was assigned to the one of the crystal surfaces of La oxide, other peaks at around 2θ = 28.5° and 57.0° were overlapped with the cerium oxide peaks [32].On the other hand, two very weak peaks appeared at around 2θ = 31.0°and 65−67.0°were attributed to the crystal surfaces of Co oxide phase [33].XRD results showed that the main structure of Ce/La/Co catalysts consisted of Ce oxide and also the presence of La and Co phases were confirmed.The average crystal size of single Ce2O3 and ternary Ce/La(12) /Co(12) and Ce/La(10)/Co(8) catalysts were estimated using Scherrer equation and they were found to be 26.4 nm, 20.5 nm, and 11.4 nm, respectively.
FESEM images of single Ce2O3 and optimum ternary Ce /La(10)/Co(8) nano-catalysts are depicted in Figure 5.As observed, some of the particles of single Ce2O3 catalyst are in spherical shape.[34,35].Therefore, the faster nucleation with a low-rate growth was performed in the crystallization process of ternary catalysts leading to the synthesis of smaller ternary nanoparticles.
Moreover, EDX analysis was performed to distinguish the chemical composition of Ce, La, O, and Co in the ternary Ce/La/Co catalysts, indicated in Figure 6.As observed, the corresponding EDX peaks proved the presence of Ce, La, O and Co elements in the structure of ternary Ce/La/Co catalyst.This showed that Ce/La/Co catalyst has been successfully synthesized for reduction reaction.Also, EDX analysis confirmed XRD results.
The weak, medium and strong acidic sites of synthesized catalysts were characterized by NH3-TPD analysis.Figure 9 indicates the NH3-TPD profiles of Ce2O3 and ternary Ce/La(10)/Co(8) catalysts.As shown, the first temperature region (a lower temperature) corresponded to the weakly bonded NH3, the second temperature region was attributed to the medium acid sites and the third temperature region (a higher temperature) was assigned to the strongly bonded NH3.Detailed temperature regions and the amounts of acidic sites are given in Table 3.As observed, weak, medium, and total acid sites of ternary Ce/La(10)/Co(8) catalyst was greater than that of pure cerium oxide.
In general, the acidity of catalysts corresponded to the unsaturated M cations (Ce, La, and Co) exposed on the surface [37].Entry of La and Co cations (La 3+ and Co 3+ ) in the structure of ternary catalyst affects the acidic sites.The La 3+ possesses empty 4f orbital and also Co cations have empty 3d orbital in their electron configuration structure to create some positions for the formation of weak and moderate Lewis acidic sites [37,38].Therefore, these cations (La and Co cations) increase the moderate acid sites of ternary Ce/La/Co catalyst compared to Ce2O3 catalyst.As shown in Table 3, the highest change in acidic sites was attributed to   Table 3: The amount of total, weak, moderate and strong acid sites of Cerium oxide and Ce/La/Co optimum catalysts.

C. Catalysts activity comparison tests
As mentioned, the main desirable reaction and the main side reaction between methane and sulfur dioxide are Reactions (1) and (2): The produced sulfur in Reaction ( 1) is in the form of a gas (sulfur boiling point is 444℃).Due to high operating temperatures of the reactor, sulfur is going out from the reactor in the form of a gas.But outside of the reactor, and as the temperature drops, it turns into liquid and then into solid crystals.
The kinetic study of these reactions has not been done so far.Of course, the catalyst role can be reducing the amount of activation energy, which reduces the energy barrier to start the reaction and causes a better breakdown of CH4 and SO2 molecules to start the reaction.
It seems that methane molecule adsorbed on surface of catalyst in the working temperatures and then it reacted with lattice oxygen atom (neighboring La or Co atoms) to form activated methyl radical.These La and Co atoms are Lewis acid sites.The produced CH3 * have free electron, and it adsorbs by Lewis acid site (La and Co).In fact, after adsorption of CH4, slow C−H bond dissociation occurred, resulting in Co−CH3 (CH3 * ) or La−CH3 (CH3 * ) and O−H (H * ) species.Co−CH3 (CH3 * ) or La−CH3 (CH3 * ) reacted with SO3 2− that produced from anionic sites on ceria (from SO2 adsorption) [39].
SO2 conversion plots for cerium oxide, optimum Ce /La(10)/Co(8) with the highest surface area and minimum mixed Ce/La(12)/Co (12) with the lowest surface area versus reactor temperature are given in Figure 10.As seen clearly in Figure 10, the reaction for all catalysts is strongly dependent on temperature, and as temperature rises, the SO2 conversion rate increases sharply.In all the temperatures, Ce-La-Co optimum catalyst shows better performance than pure cerium oxide.However, Ce-La-Co minimum catalyst performance is lower than cerium oxide.
This performance is quite expected due to their specific surface areas.For the best catalyst, the specific surface area was increased, and because of that, more active sites are available for the reaction.The reason for the sharp decrease in the SO2 reduction for minimum catalyst is that the active site is downgraded due to the reduction of the specific surface area.
The partial pressure curves for produced H2S and COS at different temperatures for all catalysts are compared in Fig-   As it can be seen from Figure 11(a), by increasing operating temperature, the produced H2S was increased for all of the catalysts.This is probably because, with increasing temperature, CS2 production increases through Reaction (9) [7]: Then the produced CS2 reacts with H2O to form H2S and COS according to Reaction (10): Moreover, produced H2 from Reaction (9) reacts with SO2 and sulfur [a product of Reaction ( 1)] through Reactions ( 11) and ( 12) respectively, causing some other H2S: It should be noticed that Reaction (12) accelerates from temperatures around 650℃ [40].
The amount of COS by-product was increased with increasing temperature, similar to hydrogen sulfide.This increase, especially at high temperatures (750℃ and 800℃), is severe.One of the reasons for increasing the amount of COS with increasing temperature, is production of CS2 from Reaction (9), and then COS was produced from this CS2 according to Reaction (12).
Another reaction that can produce COS, is reaction of CS2 with the CO2.This reaction is produces COS as follows: It is worth noting that the amount of H2S is higher than COS in various temperatures, and H2S are the main side product.This can be because the carbonyl sulfide is converted to hydrogen sulfide by the following reaction with water: Given that water is readily available due to Reaction (1), the probability of this reaction is high.Among all the catalysts, the optimum catalyst of Ce-La-Co shows the best performance in terms of the production of side products and selectivity, and the minimum surface area catalyst shows the lowest performance.This is due to the fact that the optimum catalyst has a higher active area, and so the reaction of sulfur dioxide with methane [Reaction (1)] is much higher (Figure 10).This causes methane and sulfur dioxide to be less available, and side reactions like Reactions (9) and (11) are less likely to occur.As a result, inappropriate side products are less produced.Finally, no traces of CS2 were observed as by-product in the experiments.
For example, for optimum Ce/La(10)/Co(8) catalyst at 800℃, the amount of H2S, COS, and SO2 at outlet are 125 ppm, 43 ppm, and 1800 ppm, respectively.This is while SO2 at inlet is 20000 ppm.This shows even at this temperature that the conversion is grow up to 91%, the sulfur selectivity is more than 99%.
It should be noted that the produced sulfur in the reactor conditions is in the form of gas, but while the temperature drops in the outlet tube, it becomes solid.Also, sulfur yellow crystals have been observed in the outlet tube.

D. Characterization of synthesized catalysts after reactor tests
XRD results after catalytic reactor tests revealed that the main peaks and structure of all catalysts was preserved (Figure 12).Nevertheless, the peak intensity of catalysts after reactor tests was slightly lower than that of fresh catalysts.Also, the average crystal size of single Ce2O3 and ternary Ce/La(12)/Co(12) and Ce/La(10)/Co(8) catalysts after reactor tests were found to be 27.3 nm, 23.9 nm, and 12.4  nm, respectively.Furthermore, the EDX results confirmed the presence of the main elements containing Ce, La, Co, and O in the structure of catalysts after reactor tests (Figure 13).Also, sulfur element was observed in the EDX peaks indicating that the sulfur was produced during catalytic reaction tests.It should be mentioned that the appearance of Au peak in EDX graphs was due to the gold coating on the surface of samples.

E. Comparing the catalysts
In our previous paper, ternary mixed Ce-La-Cu catalyst was used for SO2 reduction with methane [8].SO2 conversions for optimum Ce-La-Co and Ce-La-Cu catalysts are compared in Figure 14.
As it can be seen, at lower temperatures cobalt catalyst shows better performance.With increasing temperature, the conversion rate for copper catalyst is further increased and when the temperature reaches to 750℃, the conversion for this catalyst is equal with the cobalt catalyst.This is probably due to that the adsorption ability for SO2 and CH4 of active sites created in Ce-La-Co catalyst are better than the related sites of Ce-La-Cu catalyst at lower temperatures.

IV. CONCLUSIONS
In this work series of ternary metal oxides of Ce/La/Co were synthesized and optimized using RSM to achieve the maximum surface area for its application as catalyst.The optimal conditions for BET surface area of ternary catalyst were calcination temperature of 650℃, La content of 10.0 wt% and Co content of 8.0 wt%.This optimum surface area was 67.6 m 2 g −1 .The results revealed that the calcination temperature with the highest coefficient in the quadratic model showed the most important influence on BET surface area.Validity of the applied quadratic model was also con-firmed.
The characterization of Ce2O3 and ternary Ce/La/Co catalysts were carried out by BET, BJH, XRD, FESEM, EDX, and NH3-TPD analyses.XRD results showed the main peaks of Ce, La, and Co phases in the structure of ternary catalysts.EDX analysis confirmed XRD results.Also, NH3-TPD result showed that the total active sites of optimum ternary Ce/La(10%)/Co(8%) catalyst was greater than that of single cerium oxide catalyst.
The synthesized catalysts were tested for SO2 removal in the process of reduction with methane.The conversion of sulfur dioxide for Ce-La-Co optimum catalyst at different temperatures was better than of pure cerium oxide, which is expected due to the increase in its specific surface area and presence of other metals.Also, optimum Ce/La/Co catalyst produced less H2S and COS undesired materials than other catalysts.

Figure 1 :
Figure 1: Flow diagram of the reaction test system.

Figure 3 :
Figure 3: The 3D surface plots for interactive influences of two independent parameters.

Figure 2 :
Figure 2: The actual and the predicted BET surface area for different catalysts.

Figure 5 (
b) clearly showed that the particle size of ternary Ce/La/Co catalyst is significantly lower than that of single Ce2O3 catalyst [Figure 5(a)].This can be due to the change in crystallization step of Ce/La/Co catalyst compared to Ce2O3 catalyst.Indeed, the growth of crystallites in ternary catalysts was inhibited in the existence of La and Co oxides.This phenomenon can be due to the inhibitive effect of La and Co on the agglomeration and growth of individual phases of Ce oxide.Also, the chemical interaction between Ce, La, Co and O atoms leading to the formation of Ce−O−Co and La−O−Co bonds may be other reason.Since the atomic radius of Co atom (125 pm) is smaller than Ce and La atoms (182 pm and 187 pm, respectively), Ce/La/Co particles experience a contraction with the formation of Ce−O−Co and La−O−Co bonds and the crystal growth of Ce/La/Co particles is retarded because of the cobalt ions.The similar results for the synthesis of TiO2/SiO2 and ternary TiO2/ZrO2 /SiO2 catalysts were obtained in the literature

Figure 4 :
Figure 4: XRD patterns of the different catalysts.

Figure 10 :
Figure 10: SO2 conversion as a function of temperature for different catalysts.

Figure 11 :
Figure 11: Partial pressures of H2S (a) and COS (b) versus temperature for different catalysts.
ure 11(a, b).However, the amounts of these secondary materials from side reactions are very low.

Figure 12 :
Figure 12: XRD patterns of the different catalysts after reactor test.

Figure 13 :
Figure 13: EDX graphs of cerium oxide (a) and Ce-La-Co optimum catalyst (b) after reactor test.

Table 1 :
The predicted and real BET surface area values for ternary Ce/La/Co catalysts by using RSM according to CCD method.aCalcination temperature (℃).b La content in wt%.c Co content in wt%.

Table 2 :
and 2.7 × 10 −8 , respectively; indicating the The ANOVA analysis of the fitted quadratic equation for optimization of BET surface area of ternary Ce/La/Co catalysts.
a Degree of freedom.b Probability.