Methane Dry Reforming over Ni/NiO Supported on Ce-, Zr-, and Al-Modified Y₂O₃ for Hydrogen Production

Abstract

In this work, Ce, Zr, and Al are used to promote Y₂O₃ as supports for Ni/NiO, with the expectation to obtain more efficient catalysts for DRM reaction. XRD and Raman results have testified that all the three cations have been doped into the lattice of Y₂O₃ to form a solid solution structure, thus obtaining supports with decreased crystallinity and improved surface areas. As a result, all the modified catalysts display evidently improved reaction performance. The Ni–support interaction of the modified catalysts is enhanced in comparison with the unmodified catalyst, thus having improved Ni dispersion. Moreover, the modified catalysts have improved alkalinity, which is beneficial to activate CO₂ and enhance the activity. In addition, it is found that all the modified catalysts possess a richer amount of surface active oxygen species (O₂^δ− and O₂⁻), which is critical to eliminate carbon depositions. It is believed that the interaction of these factors is responsible for the enhanced DRM performance of the modified catalysts. In situ DRIFTS results have confirmed that the addition of the secondary metals can improve the DRM activity of the catalyst by accelerating the conversion of formate intermediate species.

1. Introduction

The use of fossil energy produces a large amount of CO₂, which causes the global greenhouse effect and triggers a series of problems such as glacier melting, sea level rise, and extreme climate [1]. The CO₂ molecule has high stability and reaction inertia. To realize added-value CO₂ conversion, it is necessary to design a reasonable conversion route. Methane dry reforming (DRM) can convert two greenhouse gases, CH₄ and CO₂, into syngas. Moreover, the H₂/CO ratio in the produced syngas is around 1 [2], which can be directly used as raw materials for the Fischer Tropsch reaction [3]. Therefore, the implementation of the DRM reaction can not only alleviate the energy crisis, but also help to improve the living environment [4], whereas a highly active and effective catalytic system still needs to be developed to industrialize this reaction.

The DRM reaction process usually includes the following reactions:

CH₄ + CO₂ → 2H₂ + 2CO     ΔH^θ₂₉₈ = +247 kJ/mol

(1)

CO₂ + H₂ → H₂O + CO      ΔH^θ₂₉₈ = +41 kJ/mol

(2)

CH₄ → C + 2H₂          ΔH^θ₂₉₈ = +75 kJ/mol

(3)

2CO → C + CO₂          ΔH^θ₂₉₈ = −172 kJ/mol

(4)

As shown in Equation (1), DRM is a strong endothermic reaction with the formation of 2 mole of more gaseous products [5]. Therefore, it favors high temperature and low pressure in terms of the thermodynamic equilibrium. In the past two decades, Ni-based catalysts have been widely used in the DRM reaction due to its high initial activity, abundant resources, and relatively low price. However, rapid deactivation of Ni-based catalysts can take place during a DRM process due to severe coke deposition (Equations (3) and (4)) and Ni active site sintering at elevated temperatures. Therefore, it is very important to design and develop a high-efficiency DRM catalyst, which can tolerate high temperature and prevent coke deposition. Up until now, many efforts have been devoted to promote the DRM performance of Ni-based catalysts, for instance: (1) choosing a suitable support to enhance the Ni–support interaction [6]; (2) tuning the surface-acid–base properties and oxygen species by doping with a secondary metal [7]; (3) optimizing the synthesis method to obtain highly dispersed Ni active species [8].

Yttrium oxide, a rare earth oxide, has excellent chemical and thermal durability, which has been employed as catalysts, catalyst supports, or promoters for various reactions [9]. Compared with the acidic Al₂O₃ support and inert SiO₂ support, the alkaline Y₂O₃ support is more conducive to CO₂ adsorption. The activated CO₂ adsorbed on the catalyst surface will generate more adsorbed oxygen (O_ads), hence effectively inhibiting carbon deposition [10]. Takeshi et al. prepared nano-porous Ni/Y₂O₃ catalysts with a one-step dealloying technique, which showed significant anti-coking performance compared with traditional Ni-based catalysts [11]. Fatesh et al. loaded Gd and Ni on an Y₂O₃ support, which displayed significantly improved activity compared with the unmodified Ni/Y₂O₃ catalyst, and an 82.8% H₂ yield can be achieved at 700 °C. It was reported that the addition of a secondary active metal component to a DRM catalyst can prevent active site aggregation and coke formation [12]. Mahadi et al. prepared CeO₂, La₂O₃, Sm₂O₃, and Y₂O₃ doping Co/Al₂O₃ catalysts for DRM and found that the dispersion of Co was improved. The Y₂O₃-doped Co/Al₂O₃ catalyst exhibited the best activity and coking-resistance ability [13]. Wang et al. reported enhancing the Ni–support interaction in Ni/ZrO₂ catalysts through Y doping on ZrO₂, thus improving the activity and stability of DRM reaction [14].

Inspired by these previous works, Ce, Zr, and Al were selected as the secondary metals to dope a Ni/Y₂O₃ catalyst to improve its DRM performance. It is well known that CeO₂ has a good oxygen capacity and strong ability to disperse surface active components [15]; ZrO₂ is a metal oxide possessing acidic, basic, and redox properties, and it is easy to generate surface oxygen vacancies [16]; γ-Al₂O₃ is a porous material with good thermal stability and rich surface acidity [17]. By using multiple characterization techniques, it was found that the doping of these secondary metals can affect the Ni–support interaction, thereby improving the Ni dispersion on the surface [18]. The Ce-modified catalyst exhibits the best catalytic stability and the lowest carbon accumulation degree, which is mainly due to the increase in the quantity of active surface oxygen species and the inhibition of Ni sintering.

2. Results

2.1. XRD and Raman Analysis of the Supports

Figure 1A displays the XRD patterns of the supports. The diffraction peaks of the Y₂O₃ support with 2θ of 20.49°, 29.07°, 33.78°, 35.91°, 39.85°, 43.49°, 48.43°, and 57.38° correspond to (211), (222), (400), (411), (332), (431), (440), and (622) crystallite planes of cubic Y₂O₃ [19]. The ion radius of Y³⁺ in Y₂O₃ is 0.102 nm, and the corresponding radii of Ce⁴⁺, Zr⁴⁺, and Al³⁺ are 0.097 nm, 0.084 nm, and 0.054 nm. For Ce-, Zr-, and Al-doped Y₂O₃ supports, the diffraction peak is similar to that of pure Y₂O₃ without detectable CeO₂, ZrO₂, and Al₂O₃ signals, respectively. The maximum diffraction peaks at around 29° are extracted and enlarged beside the main figure. It is observed that Ce, Zr, and Al with a smaller ionic radius can cause lattice shrinkage after entering the Y₂O₃ lattice, resulting in a peak shift to higher angles [20,21]. As can be seen from cell parameters of the supports in Table S1, the side length decreases after doping with the secondary cations with smaller ion radius. The above results show that the modified Y₂O₃ solid solution supports are successfully synthesized. Meanwhile, the average crystallite sizes of the modified supports are calculated using the Scherrer formula, which are smaller than that of pure Y₂O₃. It can be seen that the crystallinity of the modified supports decreases after cation doping to form solid solutions, resulting in an increase in the surface area.

The structures of the supports are characterized by Raman spectroscopy in Figure 1B. Several main Raman peaks are observed at 377 cm⁻¹, 330 cm⁻¹, and 469 cm⁻¹, which are attributed to the typical T_g + A_g, E_g, and T_g stretching vibrations modes of the Y₂O₃ phase, respectively [22], and no peaks of other metal oxides are observed. This confirms that all the secondary cations are successfully incorporated into the Y₂O₃ matrix to form a solid solution structure, being well consistent with the XRD results.

2.2. Activity Evaluation

Figure 2 shows the DRM activity test results of the catalysts. All the data are collected after 30 min of stabilization at a corresponding reaction temperature, which can accurately reflect the initial activity at the temperature. As seen from Figure 2A–C, with the increase in the temperature, CH₄ conversion, CO₂ conversion, and H₂ yield increase evidently. From Figure 2D, one can see that the H₂/CO ratio gradually increases with a rise in temperature, which might be ascribable to the reverse water–gas shift reaction (CO₂ + H₂ → CO + H₂O) [23]. The reaction performance of the modified catalysts is generally better than that of the unmodified sample. 5Ni/Al_0.1Y_0.9O_x exhibits the best initial activity, which may be due to its largest surface area (Table S1) and highest Ni dispersion.

To explore the stability of the catalysts, they are tested at 750 °C for 50 h. As exhibited in Figure 3, the unmodified Ni/Y₂O₃ catalyst has low initial activity and poor stability, possibly due to the sintering of Ni active sites and coking. For all the modified catalysts, the 5Ni/Ce _0.1Y_0.9O_x exhibits the most stable CH₄ conversion and H₂/CO ratio, and there is no obvious decrease during the 50 h test, which may be ascribed to the excellent oxygen storage capacity of the CeO_x component. Although 5Ni/Al_0.1Y_0.9O_x and 5Ni/Zr_0.1Y_0.9O_x exhibit a similar initial activity to 5Ni/Ce_0.1Y_0.9O_x, more than a 10% drop in CH₄ conversion is detected, indicating that these two catalysts still experience deactivation, but slower than the unmodified Ni/Y₂O₃.

To further investigate the DRM performance of the catalysts, the apparent activation energy of each catalyst is calculated. The CH₄ conversion rates are always maintained below 10% by changing the catalyst amount, and the Arrhenius curves in Figure 3D are drawn with the reaction rates in the range of 450–500 °C. As calculated in

Table 1. DRM kinetics results of the catalysts.

Catalysts	Rw 450 °C (10−3 mmol·g−1·s−1) [a]	Rs 450 °C (10−4 mmol·m−2·s−1) [b]	TOF 450 °C (10−2·s−1)	Ea (KJ·mol−1)
5Ni/Y2O3	3.3	0.8	9.6	102.7
5Ni/Ce0.1Y0.9Ox	5.5	1.2	12.7	93.4
5Ni/Zr0.1Y0.9Ox	5.2	1.0	11.3	96.0
5Ni/Al0.1Y0.9Ox	6.4	1.3	13.6	92.8

^[a] Rw: CH₄ conversion rate based on catalyst mass. ^[b] Rs: CH₄ conversion rate based on catalyst surface area.

, the unmodified 5Ni/Y₂O₃ has the largest activation energy of 102.7 KJ·mol⁻¹, while the 5Ni/Al_0.1Y_0.9O_X has the smallest one of 92.8 KJ·mol⁻¹, which is consistent with the reaction performance and surface area results. Meanwhile, the turnover frequency (TOF), Rw, and Rs of CH₄ conversion at 450 °C are specifically calculated for each catalyst. Notably, the TOF is based on Ni loading and dispersion, Rw is normalized by the catalyst mass, and Rs is normalized by the catalyst surface area. Particularly, both TOF and Rs can reflect the intrinsic activity of the catalysts [24]. Obviously, the TOF and Rs rank as 5Ni/Al_0.1Y_0.9O_x > 5Ni/Ce_0.1Y_0.9O_x > 5Ni/Zr_0.1Y_0.9O_x > 5Ni/Y₂O₃, proving that the doping of all the cations can generate more active Ni sites, which enhances the intrinsic activity of the modified catalysts.

2.3. Carbon Deposition on the Spent Catalysts

TGA-DSC and Raman techniques are employed to analyze the coke formation on the spent catalysts after 50 h stability tests. In Figure 4A, the minor weight loss between 100 and 200 °C can be assigned to water desorption, and the major weight loss between 500 and 700 °C can be attributed to the loss of deposited carbon on the spent catalysts, which is accompanied by an evident exothermic peak around 490 °C [25]. As calculated in

Table 2. Quantified coking results measured by Raman and TGA-DSC.

Catalysts	Weight Loss of Carbon Deposition (%)	Coking Rate (mg·gcat−1·h−1) [a]	IG/ID
5Ni/Y2O3	11.3	2.3	1.10
5Ni/Ce0.1Y0.9Ox	3.5	0.7	0.67
5Ni/Zr0.1Y0.9Ox	4.8	1.0	0.97
5Ni/Al0.1Y0.9Ox	7.4	1.5	1.07

^[a] Temperature ranged from 300 °C to 800 °C from TGA.

, the weight loss of carbon deposition on the unmodified 5Ni/Y₂O₃ is 11.3%. In contrast, the weight loss of 5Ni/Ce_0.1Y_0.9O_x, 5Ni/Zr_0.1Y_0.9O_x, and 5Ni/Al_0.1Y_0.9O_x is 3.5%, 4.8%, and 7.4%, respectively, testifying that Ce, Zr, or Al doping can effectively suppress coking of the modified catalysts. The coking rates on the used catalysts are also calculated, which follows the order of 5Ni/Ce_0.1Y_0.9O_x < 5Ni/Zr_0.1Y_0.9O_x < 5Ni/Al_0.1Y_0.9O_x < 5Ni/Y₂O₃. 5Ni/Ce_0.1Y_0.9O_x displays the slowest coking speed, in good accordance with its most stable reaction performance in all the catalysts.

The Raman spectra are demonstrated in Figure 4B, in which two carbon peaks can be obviously observed. The peak at 1350 cm⁻¹ is named the D band corresponding to amorphous carbon deposits, which is easier to remove [26]. The peak at 1580 cm⁻¹ is named the G band with respect to graphitized carbon, which is more difficult to eliminate and is usually considered to be a main cause of catalyst deactivation in DRM [27]. The intensity ratio of the G band to the D band (I_G/I_D) can be used to determine the graphitization degree of coking, hence being calculated in Table 2. The I_G/I_D of the unmodified Ni/Y₂O₃ catalyst is largest, indicating that it has the highest graphitization degree among all the catalysts. The addition of all the secondary metal cations decreases the graphitization degree, as testified by the decreased I_G/I_D values. Notably, 5Ni/Ce_0.1Y_0.9O_x shows the smallest graphitization degree, testifying that most of its carbon deposit is in an amorphous state. Therefore, it is concluded that the Ce doping is the most effective to increase the coking-resistance ability of Ni/Y₂O₃. In brief, both TGA-DSC and Raman results are consistent with each other, which also agrees with the stability test data.

2.4. XRD and N₂-BET of the Freshly Calcined, Reduced, and Spent Catalysts

The XRD patterns of the freshly calcined, reduced, and spent catalysts are presented in Figure 5. Figure 5A shows the XRD results of the fresh catalysts. The typical diffraction peaks of NiO crystallites can be observed along with the initial diffraction peaks belonging to the Y₂O₃ phase, but with varied intensities for the modified and unmodified Ni/Y₂O₃ catalysts [19,28]. The NiO mean crystallite sizes of the catalysts are calculated by the Scherler formula with the NiO (111) peak, and listed in

Table 3. Surface areas and crystallite sizes of the catalysts before and after the reaction.

Catalysts	Fresh Catalysts		Reduced Catalysts		Used Catalysts
	Surface Area (m2·g−1) [b]	NiO Crystallite Size (nm) [a]	Surface Area (m2·g−1)	Ni Crystallite Size (nm)	Surface Area (m2·g−1)	Ni Crystallite Size (nm)
5Ni/Y2O3	39.7	11.5	37.9	12.6	33.5	22.5
5Ni/Ce0.1Y0.9Ox	43.4	10.3	41.1	11.0	39.5	14.5
5Ni/Zr0.1Y0.9Ox	43.8	10.0	42.0	10.9	39.8	15.7
5Ni/Al0.1Y0.9Ox	53.1	9.2	51.3	10.2	48.3	17.3

^[a] Calculated from XRD; ^[b] Determined by N₂-BET.

. 5Ni/Y₂O₃, the unmodified catalyst with the worst performance, has the largest NiO size of 11.5 nm. The NiO crystallite size decreases, and the surface area increases after metal cation doping.

In Figure 5B, all the reduced catalysts present characteristic diffraction peaks of the Ni phase, demonstrating that the original NiO has been fully reduced to metallic Ni. To facilitate comparison, the mean Ni crystallite sizes of the reduced catalysts are also calculated in Table 3. Obviously, 5Ni/Al_0.1Y_0.9O_x has the smallest size of 10.2 nm, corresponding to its highest initial activity among all the catalysts. In addition, 5Ni/Al_0.1Y_0.9O_x has the largest surface area in all the catalysts despite it being freshly calcined, reduced, or used.

Figure 5C demonstrates the XRD patterns of the catalysts after the 50 h stability test. It is observed that the intensities of the metallic Ni phase in the catalysts increase, indicating that the Ni grains aggregate during the long-term DRM tests. As shown in Table 3, the Ni crystallite size of 5Ni/Ce_0.1Y_0.9O_x catalyst increases slowly from 11.0 nm to 14.5 nm after the 50 h reaction, which is slower than the growth rate of Ni grains on the other three catalysts, indicating that the catalysts have the best thermal stability. Therefore, it is inferred that the Ni–support interaction of the 5Ni/Ce_0.1Y_0.9O_x catalyst could be the strongest among all the catalysts. Former studies have shown that the crystallite size of Ni plays a key role for the activity, stability, and anti-coking ability of a DRM catalyst. Usually, a smaller crystallite size relates to better activity, stability, and anti-coking ability [29].

2.5. TEM Analysis of the Ni Distribution over the Fresh Ni/Y₂O₃ Catalysts

TEM and HRTEM tests are carried out to study the morphology and particle size distribution of the catalysts. In Figure 6, all the modified and unmodified Ni/Y₂O₃ catalysts are composed of irregular particles without a special shape. In addition, it is noted that the NiO particle size from TEM is slightly larger than the crystallite size calculated by XRD, indicating that Ni has slight secondary aggregation [30]. The NiO particle sizes calculated by TEM are in the sequence of 5Ni/Y₂O₃ > 5Ni/Ce_0.1Y_0.9O_x > 5Ni/Zr_0.1Y_0.9O_x > 5Ni/Al_0.1Y_0.9O_x. The result is similar to the crystallite sizes measured by XRD, which confirms that the modification by the secondary metal cations improves the dispersion of the active Ni/NiO component. In addition, HRTEM images show that NiO crystallites are detected, which mainly expose the (200) crystal plane. In addition to this, only Y₂O₃ particles that mainly expose the (222) crystal plane are observed. No other metal oxide components are found. The HRTEM results correspond to the XRD patterns, which further proves the successful synthesis of the solid solution supports, with the Ce⁴⁺, Zr⁴⁺, and Al³⁺ cations being incorporated into the lattice of cubic Y₂O₃.

2.6. H₂-TPR and XPS Studies of the Freshly Calcined Catalysts

The H₂-TPR technique is employed to investigate the redox behaviors of the supports and the freshly calcined catalysts. As shown in Figure S1, all the supports, either unmodified or modified, do not display any evident reduction peak below 800 °C. This indicates that all the detected peaks on the Ni/Y₂O₃ catalysts in Figure 7 can be ascribed to the reducing of NiO species. The H/Ni atomic ratios in

Table 4. Quantified H₂-TPR results of the fresh catalysts.

Catalysts	H2 Consumption (mmol·g−1)
	α Peak	β Peak	Total	H/Ni
5Ni/Y2O3	0.37	0.49	0.86	2.0
5Ni/Ce0.1Y0.9Ox	0.16	0.71	0.87	2.0
5Ni/Zr0.1Y0.9Ox	0.30	0.59	0.89	2.1
5Ni/Al0.1Y0.9Ox	0.27	0.60	0.87	2.0

, which are normalized by NiO contents, are about 2 for all the samples, indicating that Ni species exist predominantly in the form of the Ni²⁺ state.

The low-temperature peak around 350 °C is attributed to the reduction of crystalline NiO grains, which has weak interaction with the support and is easy to reduce. The peak around 400 °C is relatively broad, which could relate to the NiO species having close contact with the support, and has stronger interaction, thus being more difficult to reduce [31]. During the reduction process, this part of NiO can form highly dispersed Ni active species on the surface of the catalysts because of the retarded nucleation process [32]. Based on the quantification results in Table 4, all the modified catalysts possess a larger amount of H₂ consumption of this peak than the unmodified 5Ni/Y₂O₃, thus obtaining smaller Ni crystallites, as detected by XRD. In more detail, the H₂-TPR results show that NiO and Ce_0.1Y_0.9O_x have the largest amount of H₂ uptake of this peak, implying that Ni/NiO can be strongly anchored on the Ce_0.1Y_0.9O_x support surface, thereby obtaining the smallest Ni/NiO crystallite size, which explains its superior stability.

XPS is thus employed to investigate the interaction between NiO and the supports. Figure S2A shows that the peaks at 855.0 eV and 860.8 eV are attributed to Ni 2p_3/2 and Ni 2p_1/2 of Ni²⁺ [33], respectively, in line with the H₂-TPR results. Figure S2B shows that the signals around 157.9 eV and 159.9 eV can be indexed to Y 3d_5/2 and Y 3d_3/2 of Y₂O₃, respectively [34]. With the modification by Ce, Zr, and Al, the Ni 2p peak is shifted to lower binding energies, suggesting the increase in Ni electron density. Considering the charge conservation law, other elements must show a decrease in electron density. As proved in Figure S2B, the binding energies of the two Y 3d peaks shift to a higher value. This finding is consistent with the H₂-TPR results, suggesting that the NiO–support interaction can be regulated and strengthened by the modification of Ce, Zr, and Al cations.

2.7. H₂ Adsorption–Desorption Analysis of the Freshly Reduced Catalysts

H₂ sorption is employed to investigate the Ni dispersion and active Ni surface area with the reduced catalysts. The dispersion of Ni has an important influence on the DRM reactivity, which will affect the activity, stability, and anti-coking ability of a catalyst [35,36]. The Ni contents of the four catalysts are determined by ICP-OES (

Table 5. Ni dispersion and metallic Ni surface areas of the reduced catalysts.

Catalysts	Ni Content (wt.%) [a]	H2 Desorption (µmol·g−1)	Ni Surface Area (m2·gcat−1)	Ni Surface Area (m2·gNi−1) [b]	Ni Dispersion (%)
5Ni/Y2O3	4.98	31.2	1.5	30.9	3.6
5Ni/Ce0.1Y0.9Ox	4.97	42.9	2.0	42.3	5.0
5Ni/Zr0.1Y0.9Ox	4.97	43.2	2.1	42.8	5.1
5Ni/Al0.1Y0.9Ox	4.98	46.8	2.3	46.4	5.5

^[a] Ni loading determined by ICP. ^[b] Based on the cross-sectional area of one surface Ni atom, 8.24 × 10⁻²⁰ m², 0.5% Pt/Al₂O₃(D = 34.5%) as the standard.

), which is the same as the original formulation calculation. The H₂ desorption amount, Ni active surface areas, and Ni dispersion of all reduced catalysts are listed in Table 5, which follow the order of 5Ni/Al_0.1Y_0.9O_x > 5Ni/Zr_0.1Y_0.9O_x > 5Ni/Ce_0.1Y_0.9O_x > 5Ni/Y₂O_3. Apparently, being consistent with the Ni crystallite sizes measured by XRD, the active metallic Ni surface areas of the modified catalysts are obviously improved by the doping of Ce, Zr, and Al cations to form solid solutions. This explains how all the modified catalysts show better reactivity than the unmodified 5Ni/Y₂O₃.

2.8. Surface Oxygen Properties of the Catalysts Investigated with XPS and O₂-TPD

The surface oxygen species are investigated by XPS for the catalysts, with the O 1s signals illustrated in Figure 8. By deconvolution, three kinds of oxygen species in different chemical states are obtained for all the catalysts. In detail, the bands around 528.8, 531.0, and 532.8 eV can be assigned to the surface lattice oxygen (O_latt), oxygen species related to CO₃²⁻ (O_carb), and adsorbed oxygen (O_ads), respectively [37]. It is worth noting that a large amount of surface carbonate has been formed on all the catalysts due to the alkalinity of the Y₂O₃ surface. The in situ DRIFTS results in Figure 9 also confirm this, which is discussed further in the related places.

The O_ads species is usually considered as an active oxygen species, which can activate the carbon deposition on the catalyst surface during the DRM reaction [38], and eliminate the carbon depositions. Therefore, the relative percentage of O_ads is calculated in

Table 6. Surface oxygen characteristics of the catalysts identified by XPS and O₂-TPD.

Catalysts	O 1s, B.E./FWHM (eV)			Oads/(Oads + Olatt + Ocarb) (%)	O2 Desorption Amount (µmol·g−1)
	Oads	Ocarb	Olatt		α Peak	β Peak	Total
5Ni/Y2O3	532.6/1.8	530.9/2.0	528.6/1.9	6.6	10.0	10.4	20.4
5Ni/Ce0.1Y0.9Ox	532.7/2.0	530.9/2.1	528.7/1.8	12.0	7.3	20.0	27.3
5Ni/Zr0.1Y0.9Ox	532.8/2.0	531.0/2.0	528.6/1.8	10.1	10.7	14.2	24.9
5Ni/Al0.1Y0.9Ox	532.8/1.9	531.1/2.0	528.8/1.8	7.7	10.3	11.8	22.1

. In comparison with the unmodified 5Ni/Y₂O₃, all the modified catalysts possess obviously increased O_ads proportion. This explains the TGA-DSC and Raman results, that all the modified catalysts have better anti-coking ability than the un-modified 5Ni/Y₂O₃. Among all the modified catalysts, 5Ni/Ce_0.1Y_0.9O_x owns the highest O_ads percentage, suggesting that it has a larger quantity of oxygen vacancies than the other two modified catalysts, thus resulting in the best stability and coking-resistance ability [39]. The oxygen adsorption process could be divided into the following three steps, O₂(ad) → O²⁻(ad) → O⁻(ad) → O²⁻(lattice). The presence of more surface vacancies generally favors the dissociation of gaseous O₂ to generate adsorbed oxygen species (O²⁻ and O⁻) [40].

The oxygen species and their states on the catalysts are further investigated by O₂-TPD, with the profiles shown in Figure S3. In O₂-TPD curves, the desorption peaks below 300 °C, at 300–550 °C, and above 500 °C could, respectively, be classified as weakly adsorbed oxygen, strongly adsorbed oxygen, and surface lattice oxygen [41]. According to the temperature, the desorption peaks are divided into α and β regions, which are quantitatively analyzed in Table 6. Similar to the XPS results, all the modified catalysts possess a larger amount of total active oxygen species than the unmodified 5Ni/Y₂O₃. Again, the catalyst modified by Ce owns the largest amount of active oxygen species, which is well consistent with its best reaction stability and anti-coking ability. Indeed, this proves further that the amount of active oxygen species has a tight relationship with the stability and anti-coking performance of a DRM catalyst.

2.9. Identifying the Surface Oxygen Species with In Situ DRIFTS

To distinguish the surface oxygen species, 5Ni/Ce_0.1Y_0.9O_x with the best stability and anti-coking performance is subjected to in situ DRIFTS tests after reduction. First, the sample is pretreated with 99.99% Ar at a flow rate of 30 mL·min⁻¹ at 500 °C for 1 h to eliminate surface impurities and physically adsorbed oxygen species. Subsequently, the temperature is lowered to 300 °C and 10% O₂/Ar is introduced. As shown in Figure 9A, the intensity of the peaks at 1270 cm⁻¹ and 1030 cm⁻¹ gradually increases with time extension, which are attributed to the vibrational peaks of the active oxygen species O₂^{δ− (0<δ<1)} and O₂⁻, respectively [42,43]. This indicates that gaseous O₂ can be adsorbed on the surface oxygen vacancies to form active oxygen species.

Afterward, the atmosphere is switched to 10% H₂/Ar. Figure 9B shows that the peaks of the active oxygen species gradually weaken, and disappear after 5 min. Then, the gas atmosphere is switched back to 10% O₂/Ar flow, and it is found that those active oxygen species recover.

After the oxygen adsorption is saturated, the gas atmosphere is changed to 10% CO₂/Ar flow. Figure 9C shows that the peaks of the active oxygen species gradually weaken and vanish, accompanying the generation of surface CO₃²⁻ species. This has demonstrated that the active oxygen species can react with CO₂ to produce carbonate. Therefore, it is postulated that O₂^{δ− (0<δ<1)} and O₂⁻ are the active oxygen sites on catalysts, which could be important for the DRM reaction.

2.10. CO₂-TPD of the Reduced Catalysts

The amount of surface alkaline centers also plays a crucial role in DRM reaction. According to the previous literatures, the surface alkaline sites can adsorb and activate CO₂ molecules to generate reactive intermediates and surface oxygen species [44]. The adsorbed oxygen species is active to remove the initially formed carbon-deposition in time, which is favorable for DRM reaction. Therefore, CO₂-TPD is used to characterize the surface alkalinity of the pre-reduced catalysts. Figure 10 shows that all the catalysts have three alkaline peaks with different intensities. It was reported that the peak below 250 °C is attributed to the desorption of CO₂ on the weak alkaline sites. Due to its easy desorption, this part of adsorbed CO₂ could have limited contribution to the reaction [45]. The peak between 250 and 500 °C is attributed to the desorption of CO₂ on the moderate alkaline sites, which could contribute significantly to the reaction [46]. In fact, the CO₂ in the DRM reaction could mainly adsorb on this part of moderate alkaline sites. However, the peak above 500 °C is attributed to the desorption of CO₂ on the strong alkaline sites, possibly due to the decomposition of carbonate. Due to its difficulty to be desorbed, this part of adsorbed CO₂ could also have little contribution to the reaction. The quantified results in

Table 7. CO₂ desorption amount of the reduced catalysts.

Catalysts	CO2 Desorption (µmol·g−1)
	Weak	Moderate	Strong	Total
5Ni/Y2O3	17.1	14.8	5.0	36.9
5Ni/Ce0.1Y0.9Ox	16.2	18.7	4.8	39.7
5Ni/Zr0.1Y0.9Ox	16.0	15.1	4.6	35.7
5Ni/Al0.1Y0.9Ox	15.1	15.0	4.8	34.9

have testified that all the modified catalysts possess more moderate alkaline sites than the unmodified 5Ni/Y₂O₃, which is consistent with their improved DRM performance. Notably, 5Ni/Ce_0.1Y_0.9O_x has the largest amount of CO₂ desorption, especially the largest amount of CO₂ on the moderate alkaline sites, proving that it has the best ability to adsorb and activate CO₂ molecules.

2.11. Probing the Reaction Intermediates with In Situ DRIFTS

To explore the possible intermediate species in the DRM reaction, in situ DRIFTS experiments are performed on the reduced 5Ni/Ce_0.1Y_0.9O_x catalyst in a CH₄/CO₂ feed, which simulates the kinetic reaction atmosphere. In Figure 11A, at lower temperatures, the typical adsorbed species on the catalyst surface are bicarbonate (1649 cm⁻¹, 1438 cm⁻¹) [47]. With the reaction temperature increase, the bicarbonate peak gradually weakens. When the temperature rises to 200 °C, the bicarbonate species almost disappears. At the same time, new peaks appear at 1376 cm⁻¹ and 1603 cm⁻¹, which are assigned to formate (HCOO*) species [48,49]. This implies that bicarbonate is gradually consumed during the reaction and could be an active intermediate species in the DRM reaction. The CO peak is clearly observed when the temperature is increased to 400 °C, and the peak of formate species reaches maximum intensity. Subsequently, as the temperature increases, the formate and CO peaks weaken at the same time, and the changes in the two with temperature are the same [50,51]. Hence, formate could be an active intermediate species that can be converted into CO. For the bidentate carbonate species at 1458 cm⁻¹, although its peak also gradually increases at the higher-temperature region (200–500 °C), the variation trend of this species with temperature is inconsistent with that of CO species. This testifies that the bidentate carbonate species does not participate in the reaction.

In order to illustrate more intuitively that the bidentate carbonate is not the active intermediates, another series of experiments are conducted by introducing only CO₂ gas onto the reduced 5Ni/Ce_0.1Y_0.9O_x catalyst. As shown in Figure 11B, under the same conditions with only CO₂ gas, the peak of bidentate carbonate species gradually increases in the higher-temperature range, but the bicarbonate species is converted into formate species gradually, until it disappears. This affirms that the bidentate carbonates are easily formed at higher temperatures and might not be active intermediate species.

Figure S4 compares the in situ DRIFTS spectra of 5Ni/Y₂O₃, 5Ni/Ce_0.1Y_0.9O_x, 5Ni/Zr_0.1Y_0.9O_x, and 5Ni/Al_0.1Y_0.9O_x catalysts recorded under the same condition of experiments in Figure 11A. By increasing the temperature, it is observed that the evolution of the intermediate species in the DRM reaction is basically the same, and the bicarbonate gradually converts into formate, indicating that the same DRM mechanism occurs on all the catalysts either unmodified or modified. However, the consumption rates of formate on the three modified catalysts are faster than that on the unmodified 5Ni/Y₂O₃, especially on the Ce- and Al-promoted catalysts. This indicates that the doping of the secondary metal cations can improve the DRM activity of the modified catalysts, through accelerating the conversion of formate species.

3. Materials and Methods

3.1. Catalyst Preparation

The Y₂O₃ supports unmodified or modified with Ce, Zr, and Al were prepared by a co-precipitation method. Taking the preparation of the Ce_0.1Y_0.9O_x support as an example here, the calculated amounts of Y(NO₃)₃·5H₂O (AR) and Ce(NO₃)₃·6H₂O (AR) were dissolved in 10 mL of DDI water at the molar ratio of Y/Ce = 9:1, and stirred at room temperature until the solution was clear. A certain amount of aqueous ammonia (25–28%) was slowly dripped into the solution to adjust its pH around 9.5. After aging for 3–5 h, the solution turned into a white turbid liquid, which was then centrifuged, vacuum-filtered, and washed with DDI water until the TDS of the filtrate was below 20. The filtered solid was dried overnight at 110 °C, then grounded and calcined at 750 °C in a muffle furnace in air for 3 h to obtain the final white product. The synthesis of Zr- or Al-modified Y₂O₃ supports was similar to that of Ce_0.1Y_0.9O_x except that the solution mixture of Zr(NO₃)₄·5H₂O (AR) or Al(NO₃)₃·9H₂O (AR) with Y(NO₃)₃·5H₂O was used. The prepared supports were denoted as Ce_0.1Y_0.9O_x, Zr_0.1Y_0.9O_x, and Al_0.1Y_0.9O_x, respectively.

The unmodified or modified 5 wt% Ni/Y₂O₃ catalysts were prepared by the impregnation method. First, a certain amount of a desired support was added into a suitable amount of Ni(NO₃)₂·6H₂O (>98.0%). After stirring at room temperature for 3 h, the catalyst was dried in a water bath at 80 °C and further dried overnight at 110 °C, which was finally calcined in an air atmosphere for 4 h at 650 °C. The unmodified catalyst was labeled as 5Ni/Y₂O₃, and the modified samples were labeled as 5Ni/Ce_0.1Y_0.9O_x, 5Ni/Zr_0.1Y_0.9O_x, and 5Ni/Al_0.1Y_0.9O_x, accordingly.

3.2. Activity Evaluation

The DRM reaction performance was evaluated in a quartz tube fix bed reactor having a 6 mm diameter at 1 atm. The average sizes of the used catalysts were 0.3–0.4 mm. The catalysts were first reduced in a 30 mL·min⁻¹ 10% H₂/Ar flow at 600 °C for 2 h, then switched to a gas with a CH₄ and CO₂ ratio of 1:1 (the total flow rate of the mixed gas was 30 mL·min⁻¹). The activity tests of the catalysts were carried out in the temperature range of 600–800 °C under the condition of a gas hour space velocity (GHSV) of 18,000 mL·g⁻¹·h⁻¹. The products were analyzed by an online GC7900 with a TDX-01 column and a TCD in a 99.999% Ar carrier gas flow to monitor the concentrations of H₂, CO, CO₂, and CH₄.

The CH₄ conversion is calculated by:

$$X_{{\mathrm{CH}}_4}=\frac{n_{{\mathrm{CH}}_4,in}-n_{{\mathrm{CH}}_4,out}}{n_{{\mathrm{CH}}_4,in}}\times 100\%$$

The CO₂ conversion is calculated by:

$$X_{{\mathrm{CO}}_2}=\frac{n_{{\mathrm{CO}}_2,in}-n_{{\mathrm{CO}}_2,out}}{n_{{\mathrm{CO}}_2,in}}\times 100\%$$

The H₂ selectivity is calculated by:

$$S_{{\mathrm{H}}_2}=\frac{n_{{\mathrm{H}}_2,out}}{2\times (n_{{\mathrm{CH}}_4,in}-n_{{\mathrm{CH}}_4,out})}\times 100\%$$

The H₂ yield is calculated by:

$$Y_{{\mathrm{H}}_2}=X_{{\mathrm{CH}}_4}\times S_{{\mathrm{H}}_2}=\frac{n_{{\mathrm{H}}_2,out}}{2\times n_{{\mathrm{CH}}_4,in}}\times 100\%$$

H₂/CO ratio is calculated by:

$$\frac{{\mathrm{H}}_2}{\mathrm{CO}}=\frac{n_{{\mathrm{H}}_2,out}}{n_{\mathrm{CO},out}}$$

It is noted that n_CH4, n_CO2, n_H2, and n_CO in the equations represent the quantity of a corresponding component calculated from multiplying its integrated area with its response factor on the basis of carbon balance.

By varying the catalyst dosage, the CH₄ conversion below 10% on all catalysts was collected over the temperature range of 450–500 °C to obtain differential reaction rates in the kinetic system.

Rw is the CH₄ conversion rate (mol·g⁻¹·s⁻¹) and the formula is as follows:

$$\mathrm{Rw}=\frac{\mathrm{RF}\times \mathrm{V}\%\times X_{{\mathrm{CH}}_4}}{{\mathrm{m}}_{\mathrm{cat}}\times 22.4}(\mathrm{mmol}·{\mathrm{g}}^{-1}·{\mathrm{s}}^{-1})$$

The RF is the gas flow rate (mL·min⁻¹), V% is the volume fraction of CH₄ in the reaction gas, X_CH4 is the CH₄ conversion rate, and m_cat is the catalyst mass.

The dispersion of Ni on the reduced catalyst was determined by H₂-sorption experiment. The turnover frequency (TOF) based on exposed Ni surface active sites for CH₄ conversion was calculated.

$$\mathrm{TOF}=\frac{{\mathrm{R}}_{\mathrm{w}}\times {\mathrm{M}}_{\mathrm{Ni}}}{{\mathrm{D}}_{\mathrm{Ni}}\times {\mathrm{X}}_{\mathrm{Ni}}}$$

M_Ni is the molar mass of Ni atoms, X_Ni is the content of Ni, and D_Ni is the degree of dispersion of Ni.

3.3. Catalyst Characterization

The detailed techniques, methods, and procedures for catalyst characterization are listed in the Supplementary Materials.

4. Conclusions

With the purpose to fabricate more efficient catalysts for DRM reaction, Ce-, Zr-, and Al-modified Y₂O₃ supports were prepared by the co-precipitation method to support Ni/NiO. XRD and Raman results have testified that all the three cations can be doped into the lattice of cubic Y₂O₃ to form solid solutions, thus resulting in modified supports with decreased crystallinity and improved surface areas. As a result, the modified catalysts display evidently improved DRM rection performance, including enhanced activity, stability, and coking resistance. To elucidate the reasons leading to the performance improvement, the catalysts were characterized with different techniques.

• H₂-TPR results have indicated that with the secondary metal doping, the Ni–support interaction is enhanced in comparison with the unmodified 5Ni/Y₂O₃ catalyst. As a result, the modified catalysts have decreased Ni crystallite sizes with improved Ni dispersion, as demonstrated by XRD, TEM, and H₂ adsorption results, which is important to enhance the activity, stability, and anti-coking ability in the DRM reaction. In addition, all the modified catalysts have improved alkalinity, which is beneficial to activate CO₂ and enhance the activity.
• O₂-TPD and XPS O 1s analyses have testified that all the modified catalysts possess a richer amount of surface active oxygen species (O₂^δ− and O₂⁻) than the unmodified 5Ni/Y₂O₃ catalyst, which obeys the order of 5Ni/Ce_0.1Y_0.9O_x > 5Ni/Zr_0.1Y_0.9O_x > 5Ni/Al_0.1Y_0.9O_x > 5Ni/Y₂O₃, and is well consistent with the coking-resistance sequence. This indicates that the surface active oxygen species is critical to eliminate carbon depositions.
• In situ DRIFTS results have confirmed that the addition of the secondary metals can improve the DRM activity of the Ni/Y₂O₃ catalyst by accelerating the conversion of formate intermediate species.
• Among all the catalysts, 5Ni/Al_0.1Y_0.9O_x owns the highest active Ni surface area, thus showing the best activity. In contrast, 5Ni/Ce_0.1Y_0.9O_x possesses the largest amount of surface alkaline sites and active oxygen species, hence displaying the highest stability and the best anti-coking ability.