Cerium nitrate Ce(NO₃)₃・6H₂O (Sigma-Aldrich, 99.99%), zirconium(IV) oxynitrate hydrate ZrO(NO₃)₃・xH₂O (Sigma-Aldrich, 99.99%), ethanol C₂H₅OH (Sigma-Aldrich, ≥99.8%), tetraethyl orthosilicate Si(OC₂H₅)₄ (TEOS) (ACROS, 98%), acetic acid CH₃COOH (Sigma-Aldrich, ≥99.7%) and palladium acetate (Pd(OAc)₂) (Fluka, 35.5% Pd) are used as chemicals in this work.

The (Ce_0.33Zr_0.66O₂)SiO₂ support was prepared by sol-gel method with variable Ce and Zr loadings (Zr % = 0, 2, 4 and 8 wt.%) and a fixed molar ratio Zr/Ce = 2. Ce(NO₃)₃・6H₂O, ZrO(NO₃)₃・xH₂O, CH₃COOH, Si(OC₂H₅)₄(TEOS), ethanol and deionized water were mixed at 40˚C. The molar ratios of H₂O/TEOS = 15 and CH₃COOH/TEOS = 1. The obtained sol was maintained under constant stirring until a spongy and transparent gel was formed. The solvent was then removed by evaporation under supercritical conditions of ethanol (T_c = 240.9˚C, P_c = 6.14 MPa). Finally, the obtained solid was calcined at 550˚C for 4 h under oxygen flow (30 mL/min).

The Pd/(Ce_0.33Zr_0.66O₂)SiO₂ catalysts (CeZry, where y is the Zr loading) were prepared by the impregnation method. The appropriate amounts of palladium acetate and (Ce_0.33Zr_0.66O₂)SiO₂ solid were ground in an agate mortar for 10 min. The chosen loading of Pd was 0.5 wt.%. Then, acetone was added (1 mL/g) to obtain a paste which was dried at 60˚C and calcined at 550˚C for 2 h under oxygen flow (30 mL/min).

The BET specific surface area and the average pore diameter were determined from N₂ adsorption-desorption measurements using an automatic Micrometrics ASAP 2020 device (error percentage: 5%). Hydrogen chemisorption measurements were performed at 100˚C in a Micromeritics ASAP 2020C equipment after an in-situ reduction treatment under hydrogen at 300˚C for 2 h. Temperature programmed reduction (TPR) was performed with H₂ using a quartz U-tube reactor, coupled to a thermal conductivity detector (TCD). The catalyst (0.05 g) was dried at 250˚C during 0.5 h under argon flow (AGA, 99.99%) and reduced with 10 v/v % H₂/Ar flow (30 ml/min) from 25˚C to 400˚C (10˚C/min). TEM studies were performed on a TECNAI G2 instrument operating at 200 kV (error percentage: 2%). The XPS analyses were conducted on calcined samples using a Kratos Analytical AXIS UltraDLD spectrometer (the error percentage was below 1%). The catalytic activity for methane combustion was determined over the calcined sample (0.1 g) in a dynamic micro-reactor.

A flow comprising 1 vol.% methane, 4 vol.% oxygen and balanced with helium was mixed and regulated at a total flow of 100 mL/min. The reactor effluent was then analyzed by a thermal conductivity detector at different reaction temperatures.

The methane conversion and the turnover frequencies (TOF) were calculated by the following equations:

Conversion ( % ) = P CO 2 P CH 4 + P CO 2 × 100

P CH 4 and P CO 2 are respectively, the partial pressures of methane and carbon dioxide.

TOF = A × M Pd l × D

A = P CO 2 × D T × 273 22.4 × T r × m

A: catalyst activity (%), M_Pd: atomic mass of palladium (106.42 g/mol), l: Pd loading (wt%), D_T: total gas flow (L/h), T_r: room temperature, m: catalyst weight (g) and D the dispersion (%).

The N₂-physisorption results of the Pd/(Ce_0.33Zr_0.66O₂)SiO₂ catalysts summarized in Table 1 show that the textural properties of the solids were significantly modified when the Ce and Zr loadings were increased. Indeed, the pore volume varied between 0.83 and 1.68 cm³/g, the average pore diameter between 45.7 and 83.6 Å and the BET specific surface area between 615 and 744 m²/g, (see Table 1). As can be noted, the surface area of Pd/(Ce_0.33Zr_0.66O₂)SiO₂ was larger than that of Pd/SiO₂. These results showed that there is an interaction between CeO₂-ZrO₂ and SiO₂, which resulted in an increase in surface area of support, similar phenomenon has been observed in the literature [17] . Which suggest that this interaction can enhance also the thermal stability of palladium supported catalysts. In addition, it is important to note, that all the adsorption-desorption isotherms were of type IV in the BDDT classification (Figure 1) and that hysteresis loops apparent on the all isotherms was identified as being of the H1 type (IUPAC), which is characteristic of cylindrical pores. These results suggest that the preparation method adopted in this work is able to produce mesoporous ceria-zirconia-silica supports with improved surface area when the Ce and Zr contents increase. In the present work, 8% of Zr and 6% of Ce allows yielding a Pd/(Ce_0.33Zr_0.66O₂)SiO₂ catalyst with a relatively high surface area equal to 744 m²/g.

^aFrom N₂ chemisorption at 77 K using the BET equation. ^bTotal pore volume estimated at reduced pressure P/P₀ = 0.99, accuracy ± 0.01 cm³/g. ^cBased on H₂ chemisorption measurements. ^dEstimated according to the TEM images.

Furthermore, we can note that the average pore diameter (Dp) and the total pore volume (Vp) increase with the Ce and Zr loadings up to the CeZr4 catalyst, and then decrease on the CeZr8 sample. The highest Dp and Vp values obtained on the CeZr4 catalyst may affect the catalytic activity through the improvement of the matter and heat transfer limitations.

The hydrogen chemisorption analyses were conducted on the Pd/(Ce_0.33Zr_0.66O₂)SiO₂ in order to determine the dispersion and the particle sizes of palladium. The obtained results gathered in Table 1, show that the palladium dispersion was slightly improved when the Zr and Ce loadings increased (see Table 1). Indeed, the palladium dispersion increased from 14% on the CeZr0 sample to 18 % on the CeZr8 catalyst. These results suggest that the Pd dispersion could be improved by the increasing the ceria-zirconia amounts [18] , According to the previous work in our research group [13] , this results explained by the sintering inhibition of the PdO particles. This improvement of thermal stability could be favored by the enhancement of the PdO-support interaction. In our work, the Zr and Ce loadings seem to be insufficient to improve significantly the Pd dispersion.

The TEM images of the Pd/(Ce_0.33Zr_0.66O₂)SiO₂ samples are shown in Figure 2. Homogeneous distribution of the Pd particle size and a good dispersion of the

palladium particles are shown on CeZr0 micrograph. According to the particle size distribution, the average Pd particle size is about 6.9 ± 2 nm (Figure 2(a)), which in a good agreement with data obtained from hydrogen chemisorption results. For the CeZr2 CeZr4 and CeZr8 micrographs, it is found therein that the morphologies of these catalysts are similar with each other. A careful inspection reveals the existence of crystals with different sizes, dispersed over an amorphous matrix with different contrasts. The average particle sizes of these crystals are respectively about 12, 32 and 68 nm. It is very difficult to know if the observed contrast is associated to CeO₂, ZrO₂ or PdO particles, the accurate measurement of the particle size of Pd cannot be achieved [19] . Ye Yuan et al. [20] indicated that most of the Pd particles on a CeO₂ support could not be observed even when HR-TEM was utilized. But according to the chemisorption results the Pd particle size did not exceed 7 nm, which does not make such particles visible on the TEM images. For this reason, the CeO₂ and ZrO₂ identification can be based on their particle growth. Contrarily, to the Pd particles which are more dispersed. These results are in agreement with those of Y. T. Kim et al. [21] . The existence of palladium particles is confirmed by EDX analysis (Figure 2(e)), which confirms the existence of Pd, Ce and Zr. This result suggests that metallic oxides are well dispersed [13] .

XPS results indicated that palladium is only present as PdO(Pd²⁺) species for all catalysts. The Pd3d_5/2 binding energies for the all Pd/(Ce_0.33Zr_0.66O₂) SiO₂ catalysts fall in the range of 336.4 - 337.0 eV (Table 2). Generally, the Pd3d_5/2 binding energy for Pd⁰ and Pd²⁺ are about 334.9 ± 0.1 eV and 336.4 ± 0.1 eV, respectively [19] . Here, the Pd3d_5/2 binding energies on CeZr8 CeZr4, CeZr2 and CeZr0 samples are respectively 336.8, 337.0, 336.5 and 336.4 eV (Figure 3). According to the literature, values between 336.8 and 337.0 eV are slightly too high for indicating the presence of pure PdO(Pd²⁺, 336.4 ± 0.1eV). This increase can be related to the interaction between PdO and the CeZr-O₂/SiO₂ support [16] . Furthermore, the relatively high Pd3d_5/2 BE on the CeZr4 catalysts (337.0 eV) indicates that optimum amounts of Zr(4 wt.%) and Ce(3 wt.%) are needed to enhance the interaction between Pd species and the support. According to the XPS results, no Pd⁰ is detected on catalyst surface. Ciuparu et al. [22] , explained

this result by the phase with lower surface tension tends to encapsulate the phase with higher surface tension. Knowing that oxides have lower surface tension than metals, so the absence of metallic Pd contribution may be explained encapsulated by PdO oxides.

Furthermore, it is important to note that the surface loadings of palladium remains quasi-constant (about 0.02%) with the increase of Ce and Zr amount.

The TPR profiles of the Pd/(Ce_0.33Zr_0.66O₂)SiO₂ catalyst are shown in Figure 4. The negative signals observed below 50˚C, designated as peak α and peak β correspond respectively to the reduction of the larger PdO species dispersed on the catalyst surface and to the reduction of the smaller PdO species having higher interaction with the support [19] [23] . Therefore, they are more stable than the larger PdO species. However, the positive signal, designated as peak γ is attributed to the decomposition of the PdHx species formed during the PdO reduction [23] . According to the obtained results (Table 3), the introduction of Zr and Ce on the Pd/SiO₂ catalyst affects the PdOx reducibility. This reducibility increased in the following order: CeZr0 < CeZr8 < CeZr2 < CeZr4. Consequently, the introduction of ceria and zirconia improved the PdO redox property [24] . However, an optimum amounts equal to 4 wt.% of Zr and 3 wt.% of Ce are needed to maximize the PdO reducibility which is a key factor determining the oxidation performance of Pd-based catalysts [25] .

The methane conversion curves of the Pd/(Ce_0.33Zr_0.66O₂)SiO₂ catalysts prepared with different Zr and Ce loadings are presented in Figure 5. The T₂₀, T₅₀ and T₁₀₀ values (respectively temperatures for 20%, 50% and 100% of methane conversion) (Table 4), decreased when the Zr loading increased from 0% to 4% and

^aTurnover frequency (TOF) at 325˚C.

then increased with the further Zr loading increase to 8%. Thus the presence of cerium and zirconium oxides would obviously improve the thermal stability of PdO phase of the Pd/SiO₂ catalysts and enhance the catalytic activity. This is in agreement with our previous research. In fact, I. B. Said et al. [13] show that the addition of Ce and Zr to the Pd/MCM-41 enhance the catalytic activity for methane combustion, and an appropriate amount of these oxides can improve significantly its catalytic activity. In our case, this optimum corresponds to 4% of Zr and 3% of Ce. Considering the similar textural properties of CeZr4 and CeZr2 samples, the differences in the catalytic performance is mainly dependent on the reduction and structural properties. The TPR results show that improved reducibility is obtained with increasing the doping amount of Zr in the following order: CeZr0 < CeZr2 < CeZr4. Furthermore these results suggest that the activity of the catalytic sites is enhanced by the improvement of the PdO reducibility. In addition, according to the XPS results, the increase in Zr and Ce amounts stabilizes palladium particles in PdO state and allows a strong interaction of PdO with the support, which enhances the re-oxidation properties of PdO.

Moreover, in the case of CeZr8 catalyst, the increase of Ce and Zr amount decrease the catalytic activity compared to that of CeZr4 and CeZr2 samples. As can be noted, similar dispersion and textural propriety are obtained for all catalysts. Thus, the reason how could mainly explain the activity drop is the Pd reducibility.

The turnover number frequency (TOF) has been calculated to investigate the effect of Ce and Zr introduction in the activity per site. As summarized in Table 4. it is clearly observed that the CeZr4 sample exhibit a relatively high TOF (1588 h⁻¹), in the combustion of methane at 325˚C, followed by CeZr2 (1243 h⁻¹), CeZr8 (875 h⁻¹) and CeZr0 (157 h⁻¹). From the obtained results, the TOF evolution can explain the activity decreases. The TOF is inversely proportional to the reduction temperature order. The higher the reduction temperature, the lower is the turnover number. As a conclusion, optimum values of Ce and Zr should be reached to maximize activity in methane combustion.