Hydrogen Production through Bi-Reforming of Methane: Improving Ni Catalyst Performance via an Exsolution Approach

Abstract

Hydrogen production through the bi-reforming of methane over exsolution-derived Ni catalysts has been studied. Nickel-based catalysts were prepared through the activation of (CeM)_1−xNi_xO_y (M = Al, La, Mg) solid solutions in a reducing gaseous medium. Their performance and resistance to coking under the reaction conditions were controlled by regulating their textural, structural, morphological, and redox properties through adjustments to the composition of the oxide matrix (M/Ce = 0–4; x = 0.2–0.8; y = 1.0–2.0). The role of the M-dopant type in the genesis and properties of the catalysts was established. The efficiency of the catalysts in the bi-reforming of methane increased in the following series of M: M-free < La < Al < Mg, correlating with the structural behavior of the nickel active component and the anti-coking properties of the support matrix. The preferred M-type and M/Ce ratio determined the best performance of (CeM)_1−xNi_xO_y catalysts. At 800 °C the optimum Ce_0.6Mg_0.2Ni_0.2O_1.6 catalyst provided a stable H₂ yield of 90% at a high level of CO₂ and CH₄ conversions (>85%).

1. Introduction

Hydrogen is a versatile energy carrier and an important industrial raw material. About 75 million tons (Mt) of pure H₂ and 45 Mt of H₂ mixed with other gases are consumed annually [1]. The main consumer of pure hydrogen is the chemical sector (39 Mt), where H₂ is used in the production of ammonia, and oil refining (33 Mt), where it is needed for the hydrotreating and hydrocracking of oil [2]. Methanol synthesis and steel production are important sources of demand for hydrogen mixed with other gases. The large-scale use of hydrogen as an energy carrier requires the solution of various issues, including its efficient production, economical purification, safe storage, and transportation [3,4,5].

For H₂ production, conventional and renewable technologies are applied based on the use of fossil fuels (natural gas, heavy oils, naphtha, coal, etc.) and renewable resources (water, biomass, etc.), respectively [6,7,8]. At present, hydrogen is generally produced from natural gas (“gray hydrogen”) and coal (“brown hydrogen”), with small contributions from oil and electricity. “Gray hydrogen” is mainly manufactured through the steam reforming of methane (SRM, Reactions (1) and (2)), which is accomplished by CO₂ emission [9]:

CH₄ + H₂O ↔ CO + 3H₂                        Δ_rH⁰₂₉₈ = 206 kJ/mol

(1)

CH₄ + 2H₂O ↔ CO₂ + 4H₂                    Δ_rH⁰₂₉₈ = 165 kJ/mol

(2)

Today, H₂ production from fossil fuels emits about 900 Mt of CO₂ [10]. It is predicted that by 2050, the production of hydrogen from natural gas will be almost entirely based on low-carbon technologies: water electrolysis accounts for more than 60% of world production, and natural gas in combination with CCUS (carbon capture, utilization, and storage) accounts for almost 40% (“blue hydrogen”) [10]. However, the future of the latter technology is doubtful [6,9,11], since greenhouse gas footprints from “blue hydrogen” are only 18–20% less than those from “gray hydrogen” due to the use of fossil C-derived energy to implement the CCUS technology. Thus, the improvement of cost-effective processes of converting methane into a hydrogen-containing gas is still relevant.

The combined steam and carbon dioxide reforming of methane (bi-reforming of methane, Reaction (3)) is an attractive way to catalytically generate hydrogen-containing gas from natural gas with CO₂ utilization [12,13,14]:

2CH₄ + CO₂ + H₂O ↔ 3CO + 5H₂                    Δ_rH⁰₂₉₈ = 227 kJ/mol

(3)

This is a unique environmentally and climate-friendly process that allows one to simultaneously utilize three greenhouse gases (carbon dioxide, methane, and water vapor). The obtained syngas can be used directly in certain processes, such as methanol production or the direct reduction of iron, or they can be used to produce pure hydrogen. In this case, concentrated CO streams can be captured and used in the production of formic acid, acetic acid, or functional copolymers [15,16,17,18,19]. The bi-reforming of methane is a complex process [20,21]. It involves the reactions of steam (1) and (2) and carbon dioxide (4) and the conversion of methane, leading to the production of the target products (synthesis gas or hydrogen), a number of additional reactions that increase the yield of hydrogen (5) and (6), and side reactions of carbon formation (6)–(9).

CH₄ + CO₂ ↔ 2CO + 2H₂                    Δ_rH⁰₂₉₈ = 247 kJ/mol

(4)

CO + H₂O ↔ CO₂ + H₂                    Δ_rH⁰₂₉₈ = –41 kJ/mol

(5)

CH₄ ↔ C + 2H₂                                Δ_rH⁰₂₉₈ = 75 kJ/mol

(6)

2CO ↔ CO₂ + C                                 Δ_rH⁰₂₉₈ = –86 kJ/mol

(7)

CO + H₂ ↔ C + H₂O                          Δ_rH⁰₂₉₈ = –131 kJ/mol

(8)

CO₂ + 2H₂ ↔ C + 2H₂O                    Δ_rH⁰₂₉₈ = –90 kJ/mol

(9)

Extensive application in the reforming process was given to Ni-based catalysts due to their low cost, wide availability, and sufficiently high activity [22,23,24,25,26,27,28]. An obstacle that arises when using nickel catalysts is their deactivation under harsh reaction conditions and the impossibility of self-activation [29]. There are various strategies for improving the activity and stability of nickel catalysts: optimization of preparation mode [30,31] and support composition [32,33], application of different promoters (Mo, Pd, Pt, Rh, Re, or Mo) [34,35], variation of calcination and activation procedure [36,37,38,39]. For example, using La₂O₃ or MgO as basic support modifiers can prevent Ni agglomeration and promote CO₂ activation, which increases the activity and stability of the catalyst in the dry reforming of methane (DRM) [40,41].

Critical factors controlling catalyst performance and the rate of carbon deposits formation include the strength of metal–support interaction (MSI) and the nickel particle size. The Ni particle size significantly affects the carbon formation and removal reaction paths in DRM and, consequently, the amount and the structure/morphology of carbon deposits [42]. The amount of carbon deposition is insignificant for Ni crystallites sizes less than 10 nm and more than 100 nm and is greatest for Ni of about 20–30 nm in size [36]. On Ni particles of medium size (20–45 nm), the formation of carbon in the form of nanotubes prevails, and on small particles (<10 nm), amorphous carbon is typically found [36,42]. Carbon species (C-_S) formed during DRM can be removed through reaction with labile lattice oxygen (O-_L) of support and creating oxygen vacancies (V_o) that are replenished by the dissociative adsorption of CO₂ on the support (10) and (11) [42]:

C-_S + O-_L → CO + V_o

(10)

V_o + CO₂ → CO + O-_L

(11)

The rate of this carbon removal route is influenced by redox properties of the support and the dimension of the metal–support interface through which the migration of oxygen species from the support to the surface of Ni particles occurs [36]. Co-interaction of Ni⁰ and Ce³⁺(OH) through MSI facilitates oxygen transfer in Ni/CeO₂ [43]. Note that the metal–support interface contains specific highly reactive species [44,45,46,47,48]. In particular, in SRM over low-loaded Ni/CeO₂ catalysts, it has been shown that the activation of water at interfacial Ni sites facilitates the generation of OH species and opens additional OH-assisted CO-formation pathways via the COH intermediate formed from chemisorbed C atoms and OH groups, hindering carbon accumulation and catalyst deactivation [44]. The clusters of Ni at the ceria step edge with highly cationic characters of interfacial Ni atoms were found to have an outstanding activity in dissociating all the reactant molecules (CH₄, H₂O, and CO₂) [45]. Therefore, for high activity and superior decoking catalyst capability, it is advantageous to stabilize the strongly anchored small nickel nanoparticles on a support with appropriate oxygen storage/transport characteristics [36,44].

Among reforming catalyst carriers (Al₂O₃, MgO, ZrO₂, La₂O₃, SiO₂, CeO₂, and their mixtures), CeO₂ has great advantages due to its unique structural characteristics, the redox property of the Ce⁴⁺/Ce³⁺ pair, and the accompanying dynamic and total oxygen storage capacity (OSC) [25,49]. To tailor support crystal structure and morphology, thermal stability, surface basicity, oxygen defects, and lability, ceria is modified by aliovalent (Mg²⁺, Gd³⁺, La³⁺, and Al³⁺), isovalent (Zr⁴⁺, Si⁴⁺, and Ti⁴⁺) or aliovalent/isovalent (Pr^3+/4+, Tb^3+/4+, and Mn^2+/3+/4+) dopants [42,50,51,52,53,54,55]. The oxygen vacancies in CeO₂ arising during doping can serve as sites to anchor the active component as well as to activate the oxygen-containing bond in the reaction reagents [56]. A doping additive can also affect the form of the stabilization of the active component by controlling the strength of the MSI as well as create new ways of reactant transformation. Compared to undoped CeO₂, Ce_1−xM_xO_y (M = Gd, La, Mg, x = 0–0.5; 1.5 ≤ y ≤ 2.0) materials have a larger surface area, smaller crystallite size, and improved thermal stability, which contributes to Ni’s resistance to sintering and coking when used as support [53]. When the Ce/La ratio reached the optimum value (0.15), the Ce–La binary oxide had the highest CO₂ adsorption energy and demonstrated the best coke-removal performance [57]. By introducing in-ceria Zr as the dopant, the performance of Ni catalyst in DRM was enhanced due to a higher reducibility of the mixed-oxide support and hindering the particle sintering process of both Ni and ceria [58]. Thus, careful selection of the type and content of the dopant is necessary in order to achieve suitable physicochemical and functional properties of materials.

Ceria-based solid solutions not only act as carriers, but are increasingly used as Ni catalyst precursors [59,60,61,62,63]. In this case, the cations of the active component initially dissolve in the oxide lattice of the CeO₂-based support. When the solid solution is exposed to a reducing medium, the metal cations of the active component are reduced; the lattice oxygen is removed, and positively charged oxygen vacancies are formed in the lattice. Such an “exsolution” process is an effective approach to the formation of metal nanoparticles deposited on oxide surfaces [64,65,66]. In catalytic systems thus obtained, as a rule, a strong MSI is realized, which leads to high activity and stability of the catalyst [62,67,68]. Binary Ce-Ni-O systems are widely studied and applied for CO oxidation [69], water–gas-shift reaction [70], DRM [12], and autothermal reforming of ethanol (ATR of C₂H₅OH) [30]. It is known that the inclusion of Ni into the CeO₂ lattice introduces more defect sites and better OSC, ensuring the resistance of the catalyst to coking [31]. Ternary compounds Ce-M-Ni-O have been studied to a lesser extent. At the same time, the introduction of M strongly modifies the structure of nickel–cerium-based mixed oxides, changes the environment of active sites, and creates new Ni surface species [63]. Namely, results obtained on CeNi_xO_y (x = Ni/Ce, 0 < x ≤ 1) prepared by co-precipitation show coexisting nickel oxide and ceria modified by the insertion of Ni cations in its lattice, while on CeAl_zNi_xO_y they show a delay in the formation of NiO crystallites due to strong interaction between Al³⁺ and Ni²⁺ [63]. It was established [59] that La- or Mg-containing samples, compared to Ce_0.8Ni_0.2O_y, have a strongly defective structure and smaller crystallite sizes.

The method of preparation for mixed oxides is also an important point, since it can affect the material genesis and the mechanism of catalyst deactivation under the reaction conditions. It was shown that, resulting from the combustion synthesis (CS) in comparison to the co-precipitation, a greater fraction of Ni cations was incorporated into the Ce_xZr_1−xO₂ lattice-forming mixed metal oxides [31], and a smaller Ni particle size and stronger Ni-support interaction were realized [61]. At the same time, CS samples were more resistant to both Ni particle growth and coke formation, but suffered from Ni encapsulation by the support [31]. Great opportunities in the preparation of solid solutions are opened up by the polymerizable complex (PC) method [30,71,72]. Compared to samples obtained by impregnation, the Ni/CeO₂ catalysts obtained by the PC method were characterized by a smaller average CeO₂ crystallite size (5.5 vs. 11 nm), higher specific surface area (105 vs. 75 m²/g), defective structure, and reduced reducibility, and they were more resistant to the formation of carbonaceous deposits during the ATR of C₂H₅OH [30].

In this work, hydrogen production through the bi-reforming of methane was comprehensively studied over Ni catalysts derived from (CeM)_1−xNi_xO_y solid solutions of different compositions (M = Al, La, Mg; M/Ce = 0–4; x = 0.2–0.8; y = 1.0–2.0). The choice of doping cations was based on the ability of these elements to provide different characteristics to the oxide matrix and strength of MSI, which is expected to affect the physicochemical and functional properties of catalysts. The role of dopant type (M = Al, La, Mg) in the catalyst genesis and properties were elucidated using X-ray fluorescence analysis, thermal analysis, N₂ adsorption, ex situ and in situ X-ray diffraction, Raman spectroscopy, high-resolution transmission electron microscopy, scanning electron microscopy, EDX analysis, and temperature-programed hydrogen reduction and this provided novel insights about developing better catalysts. Careful control of the composition and activation conditions of Ni-Ce-containing solid solutions made it possible to create highly active and stable catalysts for hydrogen production through the bi-reforming of methane.

2. Results and Discussion

2.1. Characterization of Fresh Samples

The chemical composition and textural properties of fresh (CeM)_1−xNi_xO_y samples are given in

Table 1. Chemical composition and textural properties of (CeM)_1−xNi_xO_y samples.

Sample	Chemical Composition, wt.%			Textural Characteristics
	Ce	M	Ni	SBET, m2/g	Vp, cm3/g	Dp, nm
Unmodified CeO2
CeO2	82.6	–	–	83	0.20	9.6
Without M
Ce0.99Ni0.01O1.99 *	81.0	–	0.3	79	0.19	9.6
Ce0.95Ni0.05O1.95	80.1	–	2.2	107	0.20	7.3
Ce0.9Ni0.1O1.9	78.1	–	5.1	112	0.24	8.5
Ce0.8Ni0.2O1.8	72.1	–	11.0	106	0.23	8.8
Ce0.7Ni0.3O1.7	67.5	–	15.6	117	0.27	9.1
Ce0.3Ni0.7O1.3	40.8	–	39.5	98	0.26	10.8
Ce0.2Ni0.8O1.2	30.1	–	49.6	103	0.23	9.0
M = Al
Ce0.6Al0.2Ni0.2O1.7	66.2	3.2	10.1	115	0.19	6.9
Ce0.4Al0.4Ni0.2O1.6	54.6	10.6	9.8	80	0.22	11.3
Ce0.2Al0.7Ni0.1O1.5	32.5	25.1	10.0	25	0.11	16.7
Al0.9Ni0.1O1.5	–	46.3	9.9	20	0.11	22.0
M = La
Ce0.6La0.2Ni0.2O1.7	57.4	14.2	9.8	100	0.31	12.3
Ce0.4La0.4Ni0.2O1.6	36.5	36.2	9.4	48	0.18	15.6
Ce0.2La0.6Ni0.2O1.5	14.8	58.6	9.2	34	0.15	18.1
La0.8Ni0.2O1.4	–	74.2	10.0	24	0.10	17.1
M = Mg
Ce0.6Mg0.2Ni0.2O1.6	67.0	2.9	9.9	93	0.25	10.8
Ce0.4Mg0.4Ni0.2O1.4	57.0	9.9	9.2	36	0.15	17.0
Ce0.2Mg0.7Ni0.1O1.2	37.7	25.4	9.7	23	0.10	17.7
Mg0.9Ni0.1O	–	54.1	10.2	43	0.22	20.4

*—a formula used to designate the samples, where the coefficients indicate the mole fraction (usually rounded to tenths) of each element in the composition of the material.

. The chemical composition of the samples is in good agreement with the theoretically defined values.

Three series of (CeM)_1−xNi_xO_y samples were obtained using the polymerizable complex method, differing in the type (M = Al, La, Mg) and content (M/Ce = 0–4; x = 0.2; y = 1.0–2.0) of the dopant cation M, and one series based on an unmodified ceria, where the Ni/Ce molar ratio was varied over a wide range (0.01–4) and x was changed from 0.01 to 0.8.

When obtaining multicomponent solid solutions through the PC method (Figure 1), all metal cations are mixed at the molecular level in the water solution and then CA-complexes are homogeneously distributed in the gel and polymer resin. This preserves the initial stoichiometry of the molar ratio of cations in the final material, but does not exclude the coexistence of several amorphous or crystalline phases after the procedure of its thermal treatment [72].

The TA method was used to study the genesis of materials from a gel into a solid phase during thermal treatment. As typical example, Figure 2a illustrates the TG, DTG, and DTA curves for a Ce_0.8Ni_0.2O_1.8 sample. There are three main peaks on the DTG curve. The first peak at T < 200 °C is accompanied by an endothermic effect and can be associated with the desorption of water and volatile organic compounds. The second peak at T ≈ 300 °C and the third peak at 400 °C < T < 600 °C match well with the exothermic events observed in the DTA curve and can be assigned to multistage oxidation of the polymer matrix.

For (CeM)_1−xNi_xO_y samples, compared to Ce_1−xNi_xO_y, the profiles of the TG, DTG, and DTA curves have the similar behavior, but the temperature of organics burn-out shifts to the high-temperature region (Figure 2b). Thus, the introduction of M does not lead to a change in the sequence of decomposition of the organic matrix but rather slows it down. This effect increases in the series La < Al ≈ Mg and can be explained by differences in (i) the resistance of citrate–metal complexes to degradation and (ii) the catalytic activity of metals in the oxidation of organic compounds. The stability of citrate metal complexes increases in the series Mg (lgk₁ = 4.0) < Ni (lgk₁ = 5.4) < Ce (lgk₁ = 7.4) < Al (lgk₁ = 8.0) < La (lgk₁ = 8.4) [73], which has no correlation with the temperature of organic matrix degradation (Figure 2b). Apparently, the presence of a redox Ce⁴⁺/Ce³⁺ pair accelerates the oxidation reactions of organic residues, which reduces their burnout temperature with increasing content of Ce in the gel. Indeed, cerium-free samples have the highest organic matrix burnout temperature (Figure 2b). Note that, according to [74], hindered organic decomposition has a positive effect on the dispersion of the crystalline phase due to the capping around the particles and their protection from sintering. Accordingly, it can be expected that differences in the parameters of the thermal genesis of materials will affect their textural and/or structural properties.

The porous structure of the catalysts is formed under the condition of burnout of the organic matrix, accompanied by the release of gaseous combustion products. Data on low-temperature nitrogen adsorption indicate that (CeM)_1−xNi_xO_y are mesoporous materials. As evidenced by the type IV adsorption isotherm with an H3 hysteresis loop at P/P₀ > 0.6 (Figure 3), the samples have predominantly textural mesoporosity (pores between primary particles) [75,76]. The specific surface area of the Ce_1−xNi_xO_y samples practically does not change when the mole fraction of Ni varies within 0.05–0.8 and amounts to 110 ± 10 m²/g (Table 1), which represents a typical value for the Ce-Ni-O system [12,77]. It is noteworthy that, unlike samples obtained by impregnation, even at a high nickel content, the S_BET and V_p of the samples are higher than for unmodified ceria with S_BET = 83 m²/g (Figure 4a). In particular, at x = 0.3 (15 wt.% Ni) S_BET is equal to 117 and 65 for samples obtained by the polymerizable complex method and impregnation, respectively. The specific surface area and total pore volume of (CeM)_1−xNi_xO_y samples increased with decreasing M/Ce molar ratio (Table 1, Figure 4b). This trend intensifies in the series of M: Mg < La < Al. The average pore diameter depends on the composition of the material and varies from 7.3 to 22 nm. It increases with an increase in the M/Ce molar ratio. There are inhibition as well as promotion effects from dopants on the textural characteristics. In our case, the incorporation of Ni cations improves the textural properties of CeO₂, while for (CeM)_1−xNi_xO_y samples only for samples with M/Ce equal to 0.3, the textural characteristics are higher than for unmodified ceria. The resultant dependence of the textural properties of the samples upon their chemical composition may reflect the effects of dopant cations (Ni, Al, La, and Mg) of ceria on the sintering mode of the material and, consequently, phase dispersion and material porosity [77,78,79].

The morphological properties of the samples are close. They are aggregates of spongy particles (Figure 5). The formation of their porous macrostructure occurs under conditions of intense gas evolution due to the spontaneous burnout of the organic matrix.

Table 2. Phase composition and structural properties of (CeM)_1−xNi_xO_y samples.

Sample	XRD Data			Raman Spectroscopy Data
	Phase Composition	a, nm	CSR, nm	I570/I465
Unmodified CeO2
CeO2	CeO2	0.5412	11.0	0
Without M
Ce0.99Ni0.01O1.99	CeO2	0.5414	9.5	0
Ce0.95Ni0.05O1.95	CeO2	0.5416	8.5	0.4
Ce0.9Ni0.1O1.9	CeO2	0.5413	6.5	0.6
Ce0.8Ni0.2O1.8	CeO2	0.5412	5.5	1.0
Ce0.7Ni0.3O1.7	CeO2	0.5414	5.5	1.4
Ce0.3Ni0.7O1.3	CeO2 NiO	0.5412	3.2 5.0	7.4
Ce0.2Ni0.8O1.2	CeO2 NiO	0.5405	2.6 5.5	10.8
M = Al
Ce0.6Al0.2Ni0.2O1.7	CeO2	0.5409	4.5	1.1
Ce0.4Al0.4Ni0.2O1.6	CeO2	0.5412	2.8	4.0
Ce0.2Al0.7Ni0.1O1.5	X-ray amorphous phase	–	–	–
Al0.9Ni0.1O1.5	X-ray amorphous phase	–	–	–
M = La
Ce0.6La0.2Ni0.2O1.7	CeO2	0.5468	6.0	1.4
Ce0.4La0.4Ni0.2O1.6	CeO2	0.5553	4.0	–
Ce0.2La0.6Ni0.2O1.5	CeO2	0.5661	4.0	–
La0.8Ni0.2O1.4	La2O2CO3	–	–	–
M = Mg
Ce0.6Mg0.2Ni0.2O1.6	CeO2 MgO-NiO	0.5408 0.4182	5.5 4.0	1.1
Ce0.4Mg0.4Ni0.2O1.4	CeO2 MgO-NiO	0.5410 0.4204	3.5 3.6	0.8
Ce0.2Mg0.7Ni0.1O1.2	CeO2 MgO-NiO	0.5414 0.4212	5.5 8.5	1.4
Mg0.9Ni0.1O	MgO	0.4222	18.0	–

demonstrates the phase and structure data for (CeM)_1−xNi_xO_y samples. X-ray diffraction analysis indicates the presence of a CeO₂-based cubic phase in all Ce-containing samples (Figure 6). The only exception is the Ce_0.2Al_0.7Ni_0.1O_1.5 sample, for which the XRD pattern of an amorphous phase is observed. The Ni-containing phase is observed at a high nickel content (x = 0.7, 0.8) and also at M = Mg. In the first case, highly dispersed particles of NiO with an average size of ~5 nm are formed, and in the second case, a solid solution of NiO-MgO with a particle size of 4–8 nm is observed (Table 2, Figure 6). In the cerium-free samples—La_0.8Ni_0.2O_1.4 and Mg_0.9Ni_0.1O—the La₂O₂CO₃ and NiO-MgO solid solution phases are observed, respectively, while the Al_0.9Ni_0.1O_1.5 sample is X-ray amorphous.

The structural characteristics of a CeO₂-based phase depend on the content (x = 0.001–0.8, M/Ce = 0–4), ionic radius (r_Ni2+ = 0.69 Å, r_Al3+ = 0.68 Å, r_La2+ = 1.16 Å, r_Mg2+ = 0.72 Å), and degree of oxidation (2+ for Ni and Mg cations, 3+ for Al and La cations) of the dopant. It can be seen that in case of doping cations (Ni, Al, Mg) with a radius smaller than that of Ce⁴⁺ (r_Ce4+ = 0.97 Å), the unit cell parameter a of cubic CeO₂-based phase is 0.5412 ± 0.0004 nm and differs little from that for unmodified CeO₂. On the contrary, in the case of La, which has a larger radius, a noticeable increase in the unit cell parameter is observed (Table 2), which can indicate the incorporation of La³⁺ cations into the ceria lattice. For all samples, a decrease in the size of crystallites is observed in comparison with unmodified ceria. This effect is enhanced with increasing x, increasing M/Ce ratio and in the series La→Mg→Al (Figure 7), which correlates with the difficulty of burnout of the organic matrix (Figure 2b) and indicates the inhibitory effect of the cations introduced into the ceria structure on the growth of crystallites and an increase in the thermal stability to sintering of (CeM)_1−xNi_xO_y samples.

In the case of (CeLa)_1−xNi_xO_y samples, the absence of a nickel-containing phase, a decrease in the particle size, and an increase in the lattice parameter of the CeO₂-based phase clearly points to the formation of a solid solution. For the Ce_1−xNi_xO_y and (CeAl)_1−xNi_xO_y-series, no Ni-containing phases were observed either, and the CSR size for CeO₂ decreases with increasing x or M/Ce ratio. However, for these samples, the cell parameters remain virtually unchanged (Table 2). Based on the XRD data, it can be assumed that a solid solution is also formed in this case. But since the Ce³⁺/Ce⁴⁺ ratio increases with decreasing particle size [80], an increase in the concentration of Ce³⁺ with bigger ionic radius (r_Ce3+ = 1.14 Å) compensates for the expected decrease in the cell parameter of a CeO₂-based solid solution. According to [80], the cell parameter a increases from 0.541 to 0.554 Å as the CeO₂ particle size decreases from 70 to 3 nm. It was also shown [70] that differences in the formal charge of Ni²⁺ and Ce⁴⁺ promote the formation of oxygen vacancies and, as a consequence, lattice expansion. Less obvious from the XRD data is the formation of a solid solution for the (CeMg)_1−xNi_xO_y-series. In this case, although there is a decrease in the size of crystallites of the CeO₂-based phase, a NiO-MgO solid solution is present in the sample. Its unit cell parameter has values in the range of 0.4182–0.4212 nm, which lies between the values characteristic of individual oxides (0.4177 for NiO and 0.4213 for MgO) and increases with increasing Mg content in the sample (Table 2). This may indicate only a partial incorporation of Ni²⁺ and Mg²⁺ doping cations into the ceria lattice. The solubility limit of a dopant in the ceria lattice with the formation of a solid solution depends on the type of dopant and the method of its introduction. For Ce_1−xM_xO_y, a single-phase fluorite-like system is observed up to x = 0.1–0.5 for M = Ni [69,81,82], x = 0.6 for M = La [83,84,85] and x = 0.5–0.9 for M = Mg [86,87]. In addition, among the studied dopants (M = Al, La, and Mg), the Mg cations are characterized by the lowest stability of citrate complexes, which probably contributes to their less-successful entry into the ceria lattice.

Raman spectroscopy data provide additional information about the structural properties of the obtained samples (Table 2, Figure 8).

The spectrum of unmodified CeO₂ is characterized by one strong band at 465 cm⁻¹, which is due to the F2g vibration mode of the cubic fluorite structure [88,89]. In addition, it also exhibits two weak bands at about 250 and 600 cm⁻¹, which can be attributed to the non-degenerate transverse and longitudinal optical phonon modes of CeO₂, respectively [90]. The Raman spectral features of the (CeM)_1−xNi_xO_y samples were greatly affected by the content and type of dopant (Figure 8). For these samples, Raman spectra also include main bands at 452–464 cm⁻¹ and two weak bands at 230 cm⁻¹ and 550–650 cm⁻¹, which can be attributed to the displacement of oxygen atoms from their ideal fluorite lattice positions [91] and the presence of defects such as oxygen vacancies [92]. This shows [93] that the emergence of the band at about 560 cm⁻¹ (oxygen vacancies) results from the different oxidation state of the dopant compared to that of Ce⁴⁺, while the band at about 600 cm⁻¹ is caused by the different ionic radius of the dopant compared to that of Ce⁴⁺. The I₅₇₀/I₄₆₅ ratio, used to estimate the concentration of oxygen vacancies in a material [94], increases with increasing x, increasing M/Ce ratio, and in the series Mg→Al →La (Table 2). The higher the I₅₇₀/I₄₆₅, the more oxygen vacancies in the material are formed. In addition, a shift (465 → 452 cm⁻¹) of the position of the main band in Raman spectra for (CeM)_1−xNi_xO_y samples compared to the position of this band for pure CeO₂ was noted. This indicates a change in the binding energies due to the incorporation of Ni²⁺ and Mⁿ⁺ cations into the lattice of CeO₂ and confirms the formation of a solid solution. The considerable broadening of the bands observed for (CeM)_1−xNi_xO_y (Figure 8) could be attributed to the reduction in phonon lifetime due to the decrease in grain size [88], which correlates well with XRD data (Figure 6). Significant deformation of the spectrum shape for La-containing samples at La/Ce ≥ 1 points to a substantial rearrangement of the oxide structure compared to that of unmodified ceria. It was demonstrated [95] that with the increase of La concentration in Ce_1−xLa_xO_y, the Raman spectra are progressively broadened and shifted to lower energy (from 464 cm⁻¹ (La = 0%) to 458 cm⁻¹ (La = 20%)). These effects agree with the formation of homogeneous La-doped CeO₂ solid solutions.

The study of the nanostructure of the (CeM)_1−xNi_xO_y by TEM shows that the samples are represented by particles of two types: (1) thin polycrystalline plates, from 100 nm to several microns in length and 10–20 nm in width; (2) small, irregularly shaped polycrystalline agglomerates up to 100 nm in size. According to the EDX data, the composition of all particles is close to the declared one. The size of primary crystallites is very small (Figure 9a–d). For example, at x = 0.2 and M/Ce = 1 it is equal 1.5–2.5 nm for M = Al, 3–5 nm for M = La or Mg, and 5–8 nm for the sample without M. Analysis of the interplanar distances measured from the high-resolution images and microdiffraction patterns reveal the presence of a fluorite phase reflections corresponding to d₁₁₁CeO₂ (0.312 nm), d₂₀₀CeO₂ (0.271 nm) and d₂₂₀CeO₂ (1.913 nm). Also, in the case of the (CeMg)_1−xNi_xO_y sample, the reflection at 0.211 nm corresponding to d₂₀₀ of NiO-MgO particles is clearly visible in the electron diffraction pattern. In the Ce_1−xNi_xO_y and (CeLa)_1−xNi_xO_y samples, only solitary highly dispersed NiO particles can be found on the surface of the samples with a typical interplanar distance of 0.209 nm corresponding to (d₂₀₀NiO). In the case of M = Al, no reflections that would correspond to Ni-containing phases are visible. EDX mapping shows a uniform distribution of elements in the studied samples (Figure 9, e–h) implying that even in the (CeMg)_1−xNi_xO_y sample, there is a very intense mixing of crystallites of the Ce-containing phase and Ni(Mg)O nanocrystallites, which, under imaging conditions at selected magnification, does not allow one to distinguish inhomogeneities of the local composition.

Thus, solid solutions of various chemical compositions based on the cubic structure of ceria with a very-fine crystallite size were actually obtained. Both nickel and M cations were inserted into the structure of ceria through the synthesis of (CeM)_1−xNi_xO_y materials using the polymerizable complex method. This led to a significant decrease in the size of crystallites, an increase in the defectiveness of the material, and the formation of oxygen vacancies. Traces of highly dispersed Ni-containing species were also formed. This trend is intensified with an increase in the Ni content and with the use of Mg as a dopant. In the latter case, a MgO-NiO solid solution is formed. For the formation of catalytically active Ni⁰ nanoparticles, the appropriate activation of (CeM)_1−xNi_xO_y materials is needed.

2.2. Activation of Samples

As a rule, to activate nickel reforming catalysts, samples were reduced at a temperature of 600–800 °C for 1–3 h [37,96,97]. In order to study the features of the reduction of (CeM)_1−xNi_xO_y samples, methods of temperature-programed hydrogen reduction (Figure 10 and Figure 11) and in situ XRD (Figure 12 and Figure 13) were applied. On the TPR-H₂ curves of unmodified ceria, two regions of hydrogen consumption can be distinguished: in the temperature ranges of 100–600 °C (a peak maximum at T = 500 °C) and 600–1000 °C (a peak maximum at T = 880 °C), which are due to the reduction of cations Ce⁴⁺ localized on the surface and in the volume of particles, respectively [52,98]. The reduction profile of (CeM)_1−xNi_xO_y samples depends upon the amount of introduced nickel (Figure 10a) as well as the type and content of M (Figure 10b).

For Ce_1−xNi_xO_y-series, three regions of hydrogen consumption can be observed: low-temperature (T < 250 °C), medium (250 < T < 500 °C), and high-temperature (T > 550 °C). In the low-temperature region, weak consumption of H₂ is observed, the intensity and position of which is practically independent on the Ni content. The two peaks with maxima at ~150 °C and 230 ± 10 °C can be identified. In the medium-temperature region, the intensity of H₂ consumption increases with an increase in the nickel content. The peak is wide, without clearly defined maxima; its position slightly shifts towards higher temperatures with increasing x. In the high-temperature region, there is one peak, the maximum of which shifts from 880 to 780 °C as x increases from 0 to 0.8.

As shown by XRD, Raman spectroscopy, and TEM methods (Figure 6, Figure 7, Figure 8 and Figure 9, Table 2), Ni²⁺ cations in Ce_1−xNi_xO_y are stabilized in a CeO₂-based solid solution and fine NiO particles of ~5 nm in size. The share of the latter increases with the growth of x and becomes significant only at x ≥ 0.7. For comparison, Figure 10 also shows the TPR data for samples where Ni²⁺ ions are stabilized only in the form of well-crystallized NiO particles with an average size of ~50 nm. The first sample is a sample with x = 0.3 obtained by the impregnation method and calcined at 500 °C; the second is a sample with x = 0.2 obtained by the polymerizable complex method and calcined at 900 °C. The reduction profiles of these samples are similar and are described by three hydrogen consumption regions with maxima at 230, 350, and 830 °C. The fraction of [Ni-Ce-O] structures in such non-dispersed samples is insignificant. Note also that, in the impregnated samples, with an increase in the Ni content, the intensity of H₂ consumption increases only in the medium-temperature region [30].

Based on the analysis of literature data [12,98,99,100,101] and the results of studying samples by structural methods, the H₂ consumption in the low-temperature region indicates the reduction of [Ni-O-Ce] structures localized on the surface of the solid solution or at the interface between fine NiO particles and CeO₂ crystallites:

Ni²⁺ + O²⁻ + H₂ → Ni⁰ + H₂O + V

(12)

2Ce⁴⁺ + Ni⁰ → 2Ce³⁺ + Ni²⁺,

(13)

where V is an oxygen vacancy.

The H₂ consumption in the medium-temperature region indicates the reduction of Ni²⁺ cations in the bulk of CeO₂-based solid solution as well as NiO particles. The third region of hydrogen uptake is associated with the reduction of Ce⁴⁺ cations localized in the volume of particles. This process becomes easier as x increases. Samples of the Ce_1−xNi_xO_y-series are characterized by close values of the specific surface area (Table 1) and, as a consequence, a comparable content of easily reduced surface [Ni-O-Ce] structures in their composition, regardless of the different nickel contents. This type of structure is the main one at x ≤ 0.1. On the contrary, the fraction of the second type of Ni²⁺ increases with increasing x: for x < 0.7 it is mainly [Ni-O-Ce] structures in the bulk of CeO₂-based solid solution, while at x ≥ 0.7 the contribution of NiO particles grows. In any case, for M-free samples with a wide range of x = 0.01–0.8, the reduction of Ni²⁺ ends at a temperature not higher than 500 °C.

The introduction of M into the composition of the samples greatly changes their reducing properties (Figure 10b). For (CeM)_1−xNi_xO_y with M = Al or La, the profile of the TPR curve is similar to that of the sample without M, but it is shifted to the high-temperature region (ΔT = 100–200 °C). Considering the structural data (Table 2), this is mainly due to the reduction of Ni²⁺ cations in the [Ni–O–Ce] and [Ni–O–M] structures of the surface and the bulk CeO₂-based solid solution. For (CeMg)_1−xNi_xO_y, the profile is different. In this case, the intensity of H₂ consumption significantly decreases and it appears in the regions 150–350 °C and 500–700 °C. Taking into account the phase composition of Mg-containing samples (Table 2), it can be assumed that in the low-temperature region, surface and bulk [Ni-O-Ce] structures are reduced, while in the high-temperature region, NiO-MgO solid solution undergoes reduction [102]. Therefore, the feature of (CeM)_1−xNi_xO_y (M = Al, La, and Mg) series is that a significant part of the Ni²⁺ is reduced at T = 500–700 °C. This trend increases with increasing M/Ce molar ratio and for Mg-containing samples. An increase in the Ni²⁺ reduction temperature for (CeM)_1−xNi_xO_y indicates the increasing strength of the metal–support interaction [23]. For (CeM)_1−xNi_xO_y (M = Al, La) samples with a high M/Ce molar ratio, the idea that the formation of joint highly dispersed surface Al-Ni-O or La-Ni-O structures was reduced in the high-temperature region cannot be ruled out [60,103].

Total hydrogen consumption for (CeM)_1−xNi_xO_y samples during TPR increases with increasing x, which is associated with an increase in the Ni content, and in the series Mg→Al→La, which indicates an increase in the reducibility of samples (Figure 11). The lower hydrogen uptake of the Mg-containing system may be due to the presence of a hard-to-reduce NiO-MgO solid solution.

Figure 12 shows in situ XRD data for the reduction of (CeM)_1−xNi_xO_y samples in a 30% H₂/70% He mixture as a function of temperature. When the samples are reduced, the CeO₂-based phase is retained, and the formation of metallic Ni is observed. For M-free samples, traces of Ni⁰ phase are observed at 500 °C (Figure 12a) which corresponds to the temperature of full reduction of Ni²⁺ cations according TPR data (Figure 10a). This correlates with the data that Ni exsolution, atomic diffusion of Ni toward the surface, and the nucleation of Ni particles start and proceed at lower temperatures than anticipated from their appearance in the crystallized phase in XRD [104]. For (CeM)_0.8Ni_0.2O_y (M = Al, La, Mg) the temperature of the formation of the metallic nickel phase depends upon the type of M. The temperature at which reflections of metallic nickel in the diffraction pattern clearly appear, increases from 500 to 650 °C in the series Mg→Al→La (Figure 12). For Al and La-containing samples at this temperature (T > 600 °C), the reduction of Ni²⁺ cations is practically completed (Figure 10b). For the Mg series, on the contrary, at T = 500 °C, only a part of the Ni²⁺ cations is reduced—in the composition of the [Ni-Ce-O] structures. Nickel cations in the composition of NiO-MgO are reduced in a higher temperature region (Figure 10b). Additionally, for the Al-containing system, a peak at 2Ɵ = 26.4° was observed during reduction, which can be attributed to the oxygen-deficient CeO_y phase. Its low intensity indicates a small amount of this phase.

The evolution of the unit cell volume of a CeO₂-based solid solution during reduction exhibits a complex pattern (Figure 13). With thermal treatment, thermal expansion of the lattice is expected, while a sudden change in the slope may be associated with a chemical process [36,70,81,105]. Indeed, for the (CeM)_1−xNi_xO_y samples, there is a region where the expected growth of the unit cell volume due to thermal expansion does not occur. Note that this behavior is typical only for Ni-containing samples. For nickel-free oxides (dot lines in Figure 13), a continuous increase in the unit cell volume with increasing temperature is observed. The studied temperature range (up to 700 °C) is insufficient for the reduction of Ce⁴⁺ cations in the volume of crystallites. Therefore, the decrease in the lattice expansion rate may be associated with a change in the lattice composition due to Ni²⁺ exsolution from CeO₂ and the complicated activities of vacancies formed according to reaction (12) during the reduction process. The mobility of oxygen vacancies, their dimerization, clustering, or migration from the lattice to the surface to fill low-energy spaces, affects the structural parameters of samples [106].

With the exception of the Al-containing sample, the size of crystallites of CeO₂-based solid solution increases during the thermal treatment in H₂ (Figure 13). The degree of crystallite growth is higher for the M-free sample (6.2 → 21.3 nm) and increases in the series M La (4.4 → 7.8 nm) < Mg (3.6 → 11.2 nm). For an Al-containing sample, the size of solid solution crystallites remains at the level of 2–3 nm. Accordingly, the introduction of M into the composition of the samples effectively prevents sintering. This effect increases in the series Mg→La→Al.

The XRD data for (CeM)_1−xNi_xO_y samples after activation at 800 °C for 1 h in a 30% H₂/70% Ar flow is presented in

Table 3. Phase composition and structural properties of (CeM)_1−xNi_xO_y samples after activation.

Sample	Textural Characteristics			XRD Data
	SBET, m2/g	Vp, cm3/g	Dp, nm	Phase Composition	a, nm	CSR, nm
Unmodified CeO2
CeO2	7	0.07	39.4	CeO2	0.5414	50.0
Without M
Ce0.99Ni0.01O1.99	13	0.08	26.5	CeO2	0.5415	50.0
Ce0.95Ni0.05O1.95	21	0.06	11.0	CeO2	0.5419	25.0
Ce0.9Ni0.1O1.9	21	0.08	14.7	CeO2 Ni	0.5419	25.0 8.5
Ce0.8Ni0.2O1.8	20	0.07	15.0	CeO2 Ni	0.5419	25.0 15.0
Ce0.7Ni0.3O1.7	28	0.15	21.7	CeO2 Ni	0.5419	50.0 50.0
Ce0.3Ni0.7O1.3				CeO2 Ni	0.5426	50.0 50.0
Ce0.2Ni0.8O1.2	4	0.05	50.0	CeO2 Ni	0.5426	50.0 50.0
M = Al
Ce0.6Al0.2Ni0.2O1.7	23	0.14	24.8	CeO2 Ni	0.5424	6.5 6.0
Ce0.4Al0.4Ni0.2O1.6	11	0.15	55.0	CeAlO3 Ni		15.0
Ce0.2Al0.7Ni0.1O1.5	10	0.04	15.8	X-ray amorphous phase	–	–
Al0.9Ni0.1O1.5				X-ray amorphous phase	–	–
M = La
Ce0.6La0.2Ni0.2O1.7	10	0.11	44.1	CeO2 Ni	0.5490	25.0 15.0
Ce0.4La0.4Ni0.2O1.6	17	0.14	32.7	CeO2 Ni	0.5600	22.0 12.0
Ce0.2La0.6Ni0.2O1.5	20	0.25	50.6	La(OH)3	–	–
La0.8Ni0.2O1.4	29	0.23	30.9	La(OH)3	–	–
M = Mg
Ce0.6Mg0.2Ni0.2O1.6	12	0.16	53.1	CeO2 MgO (traces) Ni	0.5420	50.0 12.0
Ce0.4Mg0.4Ni0.2O1.4	12	0.10	25.8	CeO2 MgO Ni	0.5425 0.4222	22.0 12.0 12.0
Ce0.2Mg0.7Ni0.1O1.2	30	0.26	34.0	CeO2 MgO Ni	0.5417 0.4221	12.5 12.5 15.0
Mg0.9Ni0.1O	51	0.33	26.6	MgO	0.4221	19.5

and Figure 14.

It can be seen that the activated samples of the Ce_1−xNi_xO_y-series contain a CeO₂-based phase and the phase of metallic nickel. For the CeO₂-based phase, an increase in the unit cell parameter (Δa is equal to 0.0006 → 0.0019 at x = 0.1 → 0.8) is observed, which may indicate the release of nickel cations from the structure as well as an increase in the Ce³⁺/Ce⁴⁺ molar ratio. The average size of ceria crystallites amounts to 25–50 nm, which is significantly higher than in fresh samples. The average size of the formed Ni⁰ crystallites depends on the nickel content and increases from 8.5 to 50 nm with increasing x from 0.1 to 0.8.

Phase and structural transformations of (CeM)_1−xNi_xO_y samples as a result of reduction are more complex and depend on the type of M and the M/Ce molar ratio. In the case of M = Al, after activation, the CeO₂-based solid solution is retained only at Al/Ce = 0.3. For a sample with Al/Ce = 1, it transforms into the CeAlO₃ phase. The size of nickel crystallites in these samples is 6–15 nm. At higher Al/Ce molar ratios, the samples have an X-ray amorphous structure (Table 3). In the case of M = La, at La/Ce ≤ 3, the samples after activation contain a CeO₂-based phase and the phase of metallic nickel. A high cell parameter indicates the existence of the Ce-La-O-solid solution after Ni exsolution. At higher La/Ce, the CeO₂-based solid solution disintegrates as a result of the sample activation with the formation of lanthanum hydroxide (Table 3). Ce- and Ni-containing components are present in X-ray amorphous forms. For M=Mg, the changes resulting from activation are similar to those found for the M-free series. A distinguishing feature of this series is the presence of the magnesium oxide phase with an average crystallite size of 12 nm. Note that the lattice parameter of this phase indicates that the selected activation conditions ensure the reduction of the NiO-MgO solid solution present in the fresh sample. The absence of the Ni⁰-phase reflections in the XRD patterns of the samples with M/Ce >1 (M = Al or La) after activation indicates its high dispersion.

Additional information about sample transformation during activation is provided by HRTEM, EDX mapping (Figure 15), and SEM (Figure 16). On the surface of the (CeM)_1−xNi_xO_y samples, Ni species of two types are observed: nanocrystallites of 2–4 nm in size and nanoparticles of 8–15 nm in size. Larger particles have a core–shell structure; their surface is covered with a layer of NiO oxide after contact with air. Most likely the most-dispersed crystallites are also oxidized on the surface or completely upon contact with air but are easily reduced by the electron beam in the microscope. EDX mapping shows that in cases of M = Al or La, the homogeneous distribution of Ce and M is retained after Ni exsolution. This correlates with the XRD data, pointing to the existence of a Ce-M-O phase in the sample after activation. For M = Mg, the co-existing CeO₂ and MgO phases are observed, which indicates full destruction of the CeO₂-based solid solution along with Ni exsolution upon the sample activation.

After activation, the samples retained a spongy morphology (Figure 16) and mesoporosity. The average diameter of pores is within the wide range of 11–55 nm (Table 3). Nonetheless, the specific surface area of (CeM)_1−xNi_xO_y samples decreases as a result of high-temperature treatment during activation. To a lesser extent, S_BET declines in samples with an initially lower surface. For example, at M/Ce = 1, S_BET decreases by factors of 7.2 (80 → 11 m²/g), 2.8 (48 → 17 m²/g), and 2.4 (36 → 15 m²/g) for M = Al, La and M = Mg, respectively. Differences in the textural properties of the samples become less noticeable. Generally, all values of the specific surface area of activated (CeM)_0.8Ni_0.2O_y samples are within the range of 20 ± 10 m²/g (Table 3). Only for x ≥ 0.7 S_BET decreases to ~5 m²/g.

Thus, in the (CeM)_1−xNi_xO_y samples, as a result of activation, exsolution from fluorite structure and reduction of Ni²⁺ cations occur, and Ni⁰ particles are formed. The temperature range of Ni⁰ phase formation is higher for samples with a higher Ni content and increases with an increase in the molar ratio M/Ce and in the series M-free < Al = La < Mg. This indicates an increase in the Ni–support interaction upon the introduction of M into the sample. In the case of M = Al or La, the oxide matrix is the Ce–M–O joint phase, which is in contact with Ni⁰ nanoparticles. The situation is opposite in the case of M = Mg: there are coexisting CeO₂, MgO, and Ni⁰ phases. In situ XRD study in the temperature range up to 700 °C shows that the resistance to sintering of the CeO₂-based phase is higher in the case of doping M = Al or La, i.e., when a CeO₂-based solid solution continues to exist after activation. After activation at 800 °C, the dispersion of the Ni⁰ phase at the same content is comparable at M/Ce ≤ 1 for all studied M. There are two types of species: nanocrystallites (2–4 nm) and nanoparticles with an average size of 12–15 nm (x = 0.2). This is a fairly low nickel particle size at its content of 10 wt.%. For comparison, at the same Ni content and temperature of reduction, the Ni⁰ crystallite size varies from 19 to 33 nm for impregnated 10Ni/CeO₂ samples [71,105] and from 20 to 46 nm for 10Ni/CeO₂, prepared using the hydrothermal method [36]. As x increases from 0.1 to 0.8, the average particle size increases from 8 to 50 nm. At higher M/Ce molar ratio, the influence M becomes stronger. At M/Ce > 1, in contrast to Mg, the presence of Al or La in the composition of (CeM)_1−xNi_xO_y samples ensures a higher dispersion of the Ni⁰ phase as a result of activation. Thus, the activated (CeM)_1−xNi_xO_y samples have similar textural and morphological properties but different structural characteristics and resistance to sintering due to the different strength of Ni-support interaction, which can affect their functional performance.

2.3. Functional Properties

The functional properties (activity, stability, and anti-coking ability) of the (CeM)_1−xNi_xO_y samples were studied in the methane bi-reforming reaction when tested in screening and stability modes.

Figure 17 shows the temperature dependences of product concentrations during the bi-reforming of methane over (CeM)_1−xNi_xO_y samples. In the case of x = 0.05 (Figure 17a), the CH₄ and CO₂ conversion starts at 600 °C, but the quantity of reforming products—CO and H₂—is negligible (<10 vol.%). As the temperature rises, concentrations of methane and carbon dioxide decrease, while those of CO and hydrogen increase. This tendency is more pronounced for samples with x ≥ 0.2 (Figure 17b–e), for which higher values of H₂ and CO concentrations are achieved in the whole temperature range 600–850 °C. In the case of M = Al or Mg, the process parameters at 850 °C are close to thermodynamic equilibrium results (Figure 17f).

The parameters of the bi-reforming of methane over (CeM)_1−xNi_xO_y samples depend on the values of x (Figure 18) and type of M (Figure 19). It was found that with an increase in x from 0.05 to 0.8, both the conversion of reactants and the yield of products increased, reaching the following values at 750 °C: X(CH₄) 32 → 86%, X(CO₂) 46 → 73%, Y(H₂) 45 → 92%, and Y(CO) 43 → 85%. Catalyst productivity also grew with increasing x: from 8.1 to 12.0 L(H₂)/(g_cat·h) which may be connected with a higher content of Ni and, accordingly, a higher content of active Ni species per gram of catalyst. However, the efficiency of H₂ formation per mole of Ni decreases from 17.7 to 1.1 mole_H2/(mole_Ni·min) (Figure 18b). It can be seen that, in contrast to the conversion and yield indicators, the specific indicator (calculated per mole of the Ni active component) of the catalyst activity decreases with an increase in x. This may be due to an increase in the average particle size of the Ni active component (8 → 50 nm, Table 3), and, accordingly, a decrease in the fraction of available surface-active sites. The surface atoms/bulk atoms ratio rapidly decreases with increasing particle size: at a particle size of 3 nm, 50% of atoms or ions are on the surface, while at a size of 10 nm it is ~15% and at a size of 25 nm it is ~5% [107]. It can be concluded that at x = 0.2, a compromise is reached between sufficiently high catalyst productivity and a not-quite-low specific rate of hydrogen formation.

The performance of (CeM)_0.8Ni_0.2O_y in the bi-reforming of methane exceeds that of Ce_0.8Ni_0.2O_1.8 (Figure 19). An exception is the La-containing sample, in the presence of which, despite the high values of the conversion of the reagents, a reduced yield of hydrogen is observed. This may indicate the accumulation of carbonaceous deposits according to the reactions (8–9). Indeed, after screening tests (stepwise temperature-rise mode, 650→850 °C) the samples contain coke, the content of which increases in the series 0.8 wt.% (M = Al) ≈ 0.9 wt.% (M = Mg) < 2.7 wt.% (M-free) < 6.9 wt.% (M = La). At constant type of M, changing the M/Ce ratio has little effect on catalyst performance in screening tests. There is a slight tendency to a decrease in the hydrogen yield with an increase in the M/Ce molar ratio.

Note that the activity in screening tests of the studied (CeM)_0.8Ni_0.2O_y samples, prepared using the PC method, is comparable with that of impregnated samples of the same composition [108], while the degree of coking is significantly lower (Figure 20). For both cases, the Mg-containing samples are the most resistant to coking.

Table 4. Effect of Ni molar fraction (x), type of M, and molar ratio M/Ce on the parameters of the bi-reforming of methane at 800 °C over (CeM)_1−xNi_xO_y samples.

Sample	Time on Stream, h	XCH4	XCO2	YH2	YCO	H2/CO	Degree of Deactivation *, %	Carbon Accumulation Rate, mgC/gcat·h
Ce0.95Ni0.05O1.95	2	31	39	40	37	1.5	35	0
	24	19	31	26	25	1.4
Ce0.9Ni0.1O1.9	2	62	61	67	60	1.4	21	0
	24	48	48	53	49	1.5
Ce0.8Ni0.2O1.8	2	79	75	84	79	1.5	15	0
	24	67	66	72	67	1.5
Ce0.7Ni0.3O1.7	2	74	83	73	78	1.3	10	0.38
	24	68	67	75	70	1.4
Ce0.6Al0.2Ni0.2O1.6	2	87	84	99	88	1.4	0	0.16
	24	93	89	99	91	1.4
Ce0.6La0.2Ni0.2O1.7	2	92	94	92	90	1.4	5	0.46
	24	96	87	88	89	1.4
Ce0.4La0.4Ni0.2O1.6	2	91	90	91	87	1.4	0	1.33
	24	94	89	95	88	1.5
Ce0.2La0.6Ni0.2O1.5	2	92	89	83	85	1.4	0	1.71
	24	92	90	93	88	1.5
La0.8Ni0.2O1.4	2	93	86	99	89	1.5	10	3.79
	24	95	87	90	91	1.5
Ce0.6Mg0.2Ni0.2O1.6	2	90	86	90	87	1.4	0	0.08
	24	90	85	94	89	1.5
Ce0.2Mg0.7Ni0.1O1.2	2	70	74	80	77	1.3	0	0.21
	24	84	90	86	84	1.3
Mg0.9Ni0.1O	2	89	93	89	92	1.3	0	0.38
	24	88	89	91	87	1.4

* Relative decrease in hydrogen yield after 24 h of time on stream.

and Figure 21 demonstrate the effect of Ni molar fraction (x), type of M, molar ratio M/Ce, and preparation method on the parameters of the bi-reforming of methane at 800 °C during stability tests. A decrease in process performance over time is observed when M-free samples are used as catalysts or at M = La. The degree of sample deactivation, calculated as the relative decrease in H₂ yield over 24 h of time on stream, is at the maximum (35%) for a sample with a low Ni molar fraction (x = 0.05). For the La-series, it does not exceed 10% (Table 4). On the contrary, in the case of M = Al or Mg, the process values practically do not change with time on stream. As in the screening tests, in the stability tests, the M/Ce molar ratio has little effect on the activity of the samples. Among the studied types of M, the yield of hydrogen increases in the series M-free < La < Mg < Al (Figure 21). Better performance of methane reforming is observed for samples obtained by the PC method rather than by impregnation (Figure 21d).

Studied samples of optimal composition provide high and stable conversion of methane (X_CH4 ≈ 90%) and CO₂ (X_CO2 ≈ 85%), as well as a high hydrogen yield (Y_H2 > 90%) that is comparable to or better than the literature data for the bi-reforming of methane [109,110,111,112,113,114]. Namely, during the bi-reforming of methane at 800 °C over impregnated Ni/CeO₂ catalyst, X_CH4 is equal to 50% [109], at 800 °C over co-precipitated Ni/CeO₂ it is 80% [109], at 900 °C over Ni–Ce–Fe/Al₂O₃ it is 90% [110], at 800 °C over Ni/SBA-15 it is 70% [111], at 800 °C over Ni/MgAl₂O₄ it is 90% [112], at 800 °C over Ni/Ce_0.6Zr_0.4O₂ it is 60% [113], and at 700 °C over Ni/ZrO₂ it is 45–70%, depending on the type of ZrO₂ [114].

The study of spent catalysts by thermal analysis shows that an insignificant weight loss (0.2–0.4 wt.%) occurs in the low-temperature region (T < 200 °C) due to the desorption of water and volatile intermediate products (Figure 22). Further, at a temperature of 300–500 °C, the weight of the sample increases, which is associated with the oxidation of the Ni⁰ active component. The process is accompanied by an exothermic effect, the maximum of which shifts from 345 to 420 °C in the series Al < La < Mg < M-free. This indicates the decrease in Ni stability to re-oxidation in the presence of M, which correlates with improvement to the metal–support interaction. This effect is more pronounced for M = Al. There is weight loss in the region of 500–800 °C. The exception is M-free samples with x < 0.3, for which no change in weight is observed in the high-temperature region. The amounts of this weight loss vary from 0.2 to 9 wt.% and can be assigned to the oxidation of carbonaceous deposits. Higher weight values due to the burnout of carbon deposits are observed for samples with M = La. The carbon accumulation rate decreases with the decrease of the molar ratio of M/Ce and in the series La→Al→Mg. Among (CeMg)_0.8Ni_0.2O_y samples, the minimum rate of formation of carbonaceous deposits was observed for the Ce_0.6Mg_0.2Ni_0.2O_1.6 sample (Table 4). It is equal to 0.08 mgC/g_cat·h, which is significantly lower than the specific rate of H₂ formation: 6.8 × 10⁻⁵ mole_C/(mole_Ni·min) vs. 5.0 mole_H2/(mole_Ni·min).

The main pathways of carbon formation are cracking (6) and Boudouard reaction (7). According to thermodynamic data, the selectivity of coke formation during the bi-reforming of methane decreases with increasing temperature [21]. Therefore, carbon formation occurs more intensively in screening tests than in stability tests: 6.9 vs. 3.2 and 0.9 vs. 0.4 wt.% for M = La and Mg and at M/Ce = 1. In comparison to the literature data, the carbon formation level for spent samples with M = Al or Mg is low. For example, after the bi-reforming of methane, 12% Ni/Al₂O₃ catalyst includes 50 wt.% of carbonaceous deposits (C) [110], 12% Ni–5% Ce–5% Fe/Al₂O_3—2 wt.% of C [110], 5% Ni/MgAl₂O_4—3 wt.% of C [112], 10% Ni/SBA-15—6.4 and 3% B-10% Ni/SBA-15—1.5 wt.% of C [111], for 10% Ni/ZrO₂ the content of C varies from 0.2 to 47.9 wt.% for different types of ZrO₂ support [114]. Therefore, a significant advantage of the developed materials is their high resistance to coking even when a low steam/CO₂ molar ratio (H₂O/CO₂ = 0.5) is used in the initial reaction mixture. This stability of the catalyst to coking may be due to several factors. First, this is connected with the anti-coking properties of ceria support [49,115], and second, the large size of Ni crystallites, on which, as was shown in [36], the coking rate is lower than for nickel particles with a size of 20–30 nm. In addition to the influence of M on the rate of coke formation, its influence on the mode of change in the textural and structural properties of the samples (CeM)_1−xNi_xO_y as a result of the reaction is expected.

2.4. Characterization of Spent Samples

The harsh conditions of the catalytic reaction (high temperatures, presence of water vapor, etc.) inevitably affect the properties of the samples.

Table 5. Textural properties and phase composition of spent (CeM)_1−xNi_xO_y samples.

Sample	Textural Properties			Phase Composition *
	SBET, m2/g	Vp, cm3/g	Dp, nm
Without M
Ce0.95Ni0.05O1.95	3	0.02	29.9	CeO2 (50 nm, 0.5414 nm) ** Ni      (50 nm)
Ce0.9Ni0.1O1.9	4	0.03	34.8	CeO2 (50 nm, 0.5411 nm) Ni      (50 nm)
Ce0.8Ni0.2O1.8	3	0.02	23.6	CeO2 (50 nm, 0.5418 nm) Ni      (50 nm)
Ce0.7Ni0.3O1.7	5	0.03	23.9	CeO2 (50 nm, 0.5418 nm) Ni      (50 nm)
M = Al
Ce0.6Al0.2Ni0.2O1.7	11	0.14	51.9	CeO2 (12.0 nm, 0.5415 nm) Ni      (25.0 nm)
Ce0.4Al0.4Ni0.2O1.6	10	0.13	51.4	CeAlO3 Ni      (21.0 nm)
Ce0.2Al0.7Ni0.1O1.5	8	0.09	44.5	X-ray amorphous phase Ni      (8.0 nm)
Al0.9Ni0.1O1.5	21	0.12	22.5	Al-Ni-O (–, 0.8000 nm) Ni
M = La
Ce0.6La0.2Ni0.2O1.7	10	0.03	11.2	CeO2 (50 nm, 0.5476 nm) Ni      (traces)
Ce0.4La0.4Ni0.2O1.6	14	0.14	39.8	CeO2 (50 nm, 0.5598 nm) Ni      (traces)
Ce0.2La0.6Ni0.2O1.5	13	0.06	17.5	La(OH)3 (traces)
La0.8Ni0.2O1.4	8	0.02	11.2	La(OH)3 (traces) Ni      (50 nm)
M = Mg
Ce0.6Mg0.2Ni0.2O1.6	10	0.07	27.7	CeO2 (25 nm, 0.5415 nm) Ni      (25 nm)
Ce0.4Mg0.4Ni0.2 O1.4	11	0.05	18.0	CeO2 (25 nm, 0.5415 nm) Ni      (30 nm) MgO  (25 nm)
Ce0.2Mg0.7Ni0.1O1.2	19	0.12	26.0	CeO2 (25 nm, 0.5415 nm) Ni      (40 nm) MgO  (25 nm)
Mg0.9Ni0.1O	23	0.17	29.7	Ni      (18 nm) MgO

* In the samples after the reaction, the quartz phase is present in trace amounts. The free volume of the reactor is filled with quartz, and when the reactor is unloaded, it can get into the sample. ** In parentheses are the values of the coherent scattering region and the unit cell parameter.

presents the textural properties and phase composition of spent (CeM)_1−xNi_xO_y samples.

The specific surface area of Ce_1−xNi_xO_y samples decreases from 20 to 5 m²/g, while for the (CeM)_1−xNi_xO_y samples it remains at the level of the values for the samples after activation (15 ± 5 m²/g). This indicates a positive effect of M on the textural properties of the samples and a lower rate of thermal sintering of (CeM)_1−xNi_xO_y samples. Spent samples lose the porous spongy structure that was observed for fresh (Figure 5) and activated (Figure 16) samples and consist of large aggregates of irregularly shaped particles (Figure 23). The morphology of the spent M-free sample looked particularly dense. The average pore diameter increases, reaching 20–50 nm in size (Table 5).

XRD data show that spent Ce_1−xNi_xO_y samples contain CeO₂ and Ni⁰ phases (Figure 24, Table 5), which is similar to the phase composition observed after sample activation (Figure 14, Table 3). However, the phases become less dispersed (50 nm vs. 15–25 nm) as a result of sintering under the reaction conditions, which is typical for supported metal catalysts [31,116]. A similar picture is observed for the (CeM)_1−xNi_xO_y samples. A distinctive feature of the La-containing series is the preservation of a CeO₂-based solid solution after the reaction. Its average crystallite size increases from 20–25 to 50 nm. In contrast to M-free, Al- and La-containing samples, at M = Mg, the sample composition includes three phases (CeO₂, MgO, and Ni⁰) formed at the activation stage (Table 3 and Table 5). They grow larger as a result of the reaction. Nevertheless, the ceria phase has an average crystallite size of 25 nm, which is smaller than those for M-free and La-containing samples.

According HRTEM study, for all samples, in addition to large, well-crystallized big Ni particles, Ni nanoparticles of 1.5–5 nm in size were observed on the oxide surface. The particles were stabilized due to epitaxy on CeO₂ (a coincidence of the crystallographic directions [111] CeO₂ and [100] Ni is observed) (Figure 25). EDX-mapping of spent samples demonstrates that distribution behavior of M between phases of an individual oxide and CeO₂-based solid solution is controlled by the type of M. After 24 stability tests, La³⁺ cations remain in the solid solution, Al³⁺ cations only partially remain in the fluorite structure and form Al-enriched crystallites of needle shape, while Mg²⁺ cations prefer to exist outside of the CeO₂-based solid solution.

Thus, the high activation temperature and subsequent harsh reaction conditions smoothed out the initial textural and structural differences between samples with different values of x. After 24 h during the bi-reforming of methane, spent Ce_1−xNi_xO_y samples were characterized by the same very-low S_BET values (3–5 m²/g) and the phase composition, which is CeO₂ and Ni⁰ with a CSR of 50 nm. Nevertheless, with increasing x, first, a higher hydrogen productivity is achieved, and hence a higher hydrogen concentration in the mixture of reaction products, and second, lower deactivation rates are observed. However, the number of carbonaceous deposits is increasing: samples with a lower degree of deactivation (x = 0.3) have a higher rate of coke accumulation. It can be assumed that these differences in the activity and stability of the Ce_1−xNi_xO_y samples are more related to the differences in their anti-sintering properties, rather than in their resistance to coke formation. In situ XAS spectroscopy indicates [117] that under reducing conditions the size and morphology of Ni particles change, as they become flattened and strongly stabilized on the partially reduced CeO₂ surface which promotes their stability under reaction. Taking into account the absent of carbonaceous deposits for spent samples with x ≤ 0.2, the decrease in activity in the initial period of the reaction before reaching a steady state conversion seems to be mainly associated with phase sintering and a decrease in the number of available active sites. By improving the dispersion of Ni to isolated cations in as-synthesized samples, the Ni particle growth mechanism switches from crystal migration to atomic migration, which leads to higher particle growth rates [31]. Therefore, the sintering process is more pronounced for samples with low values of x, which have higher initial nickel dispersion. Among Ce_1−xNi_xO_y samples, the sample with x = 0.2 was chosen as the optimal one due to its ability to provide high and stable conversion of CH₄ and CO₂ to synthesis gas, high H₂ productivity, resistance to the formation of carbon deposits, and a moderate deactivation rate.

The introduction of M into the composition of the samples improves their activity during the bi-reforming of methane. In the case of hydrogen yield, this tendency increases in the series La < Mg < Al. In addition, (CeM)_1−xNi_xO_y samples provide more stable performance with time on stream. In comparison to Ce_0.8Ni_0.2O_1.8 samples, (CeM)_0.8Ni_0.2O_y is characterized by strengthening the Ni–support interaction and a lower rate of thermal sintering. Similar to spent samples without M, (CeM)_1−xNi_xO_y contain coarsely dispersed phases of Ni⁰ and CeO₂. In Al- and Mg-containing systems, the CSR of CeO₂ is lower than those in M-free and La-series samples (25 vs. 50 nm). In contrast to Ce_0.8Ni_0.2O_1.8, (CeM)_0.8Ni_0.2O_y samples are characterized by coke formation, the rate of which increases in the series Mg < Al < La. This is in good correlation with the structural properties of materials. To be specific, after activation at M = La, the solid solution is preserved, at M = Al, the Al³⁺ cations are only partly retained in the CeO₂ structure, and at M = Mg, the full destruction of solid solution occurs, and CeO₂ and MgO phases co-exist. Given that the M-free sample has better resistance to carbon deposits, this ability is reduced for CeO₂-based solid solutions due to a decrease in the oxygen mobility at a high doping concentration [118].

3. Materials and Methods

3.1. Sample Preparation

The procedure for the synthesis of (CeM)_1−xNi_xO_y (M = Al, La, Mg) solid solutions by the polymerizable complex method was adapted from our previous studies [59,60]. The molar ratio M/Ce varied within 0–4; the mole fraction of Ni in the sample (x) varied from 0.01 to 0.8. The molar fraction of oxygen (y) was calculated based on the principle of electrical neutrality and the assumption that the cations were in the form of Ce⁴⁺, Al³⁺, La³⁺, Mg²⁺, and Ni²⁺. The scheme for the synthesis of catalysts is shown in Figure 1. In these syntheses, cerium nitrate hexahydrate (Ce(NO₃)₃·6H₂O), aluminum nitrate nonahydrate (Al(NO₃)₃·9H₂O), lanthanum nitrate hexahydrate (La(NO₃)₃·6H₂O), magnesium nitrate hexahydrate (Mg(NO₃)₂·6H₂O), nickel nitrate hexahydrate (Ni(NO₃)₂·6H₂O), citric acid monohydrate (CA, C₆H₈O₇·H₂O), ethylene glycol (EG, C₂H₆O₂), and ethylenediamine (ED, C₂H₈N₂) were used as starting materials. The molar ratio of the components was constant and equal to CA:EG:ED:(Ce + M + Ni) = 4:12:4:1.

First, appropriate amounts of metal salts were dissolved in water and mixed with a solution of CA in EG, obtained at 70 °C, with stirring. Then, the resulting solution was stirred at 60 °C for 1 h, and ED was dropwise added to promote the polyesterification reaction. After evaporation at 70 °C for 48 h and the formation of a polymer resin, the sample was thermally treated in a muffle furnace at 500 °C for 4 h in air in order to eliminate the organic matrix. The heat treatment details are given in [60].

For comparison, samples with the same chemical composition were obtained by incipient wetness impregnation of the corresponding Ce_1−zM_zO_y (M = Al, La, Mg, z = 0–1, y = 1–2) supports by an aqueous solution of nickel nitrate with the required concentration. Supports Ce_1−zM_zO_y as reference samples were prepared following the procedure of the PC method used for (CeM)_1−xNi_xO_y. These materials were calcined also at 500 °C in air.

3.2. Sample Characterization

The chemical composition of the samples was determined by X-ray fluorescence analysis in an ARL ADVANT’X analyzer (ThermoTechno Scientific, Ecublens, Switzerland) with an Rh anode of the X-ray tube.

Thermogravimetry (TG) and differential thermogravimetry (DTG) with differential thermal analysis (DTA) of polymeric gel and spent catalysts were performed (NETZSCH STA 449C, Selb, Germany) at a heating rate of 10 °C/min up to 900 °C in flowing air.

The N₂ adsorption–desorption isotherms for the samples were measured on an ASAP 2400 automated volumetric instrument (Micromeritics, Norcross, GA, USA) at –196 °C. The specific surface area (S_BET) was calculated using the BET method. The total pore volume (V_p) and average pore size (D_p) were calculated using the BJH method applied to the desorption branch of the isotherm.

The scanning electron microscopy (SEM) analysis of the samples was conducted using a JSM-6390LA (JEOL, Tokyo, Japan) electron microscope.

The X-ray diffraction (XRD) analysis of the samples ex situ was carried out on an HZG-4C diffractometer (Freiberger Prazisionmechanik, Freiberg, Germany) with CoK_α radiation (λ = 1.79021 Å) at room temperature in the 2Ɵ ranging from 10 to 80° with a step of 0.1 degrees and an accumulation time of 6–15 s. Diffraction peaks were identified using JCPDS powder diffraction databases. Then, according to the Bragg–Wolfe equation, interplanar distances and the parameters of the unit cell (a) were calculated. The accuracy of the determination of the parameters of the unit cell was ±0.003 Ǻ. The coherent scattering region (CSR) was calculated by the Selyakov–Scherrer method from the broadening of the diffraction peak (111), fixed phases having a cubic structure of the fluorite type; CSR (NiO)—peak (200) phases of NiO; CSR (Ni)—peak (200) phases of Ni⁰.

X-ray phase analysis of the samples in situ was conducted on an AXS D8 diffractometer (Bruker, Karlsruhe, Germany) with CuK_a radiation (λ = 1.5406 Å) in the 2Ɵ ranging from 20 to 50° with a step of 0.05 degrees and an accumulation time in each point of 5 s. The heating and cooling rate was 12 °C/s, the exposure time of the diffraction pattern at a certain temperature was at least 30 min before each diffraction pattern was recorded. Parameters of the unit cell were determined using IK 2.1. software (BIC SB RAS, Novosibirsk, Russia) and JCPDS powder diffraction databases.

The Raman spectra of the samples were recorded using an excitation wavelength of 514.5 nm on a Renishaw Invia Raman spectrometer (Renishaw plc., Wotton-under-Edge, Gloucestershire, United Kingdom). The power of the laser radiation incident on the sample did not exceed 2 mW, and the accumulation time was 30 s.

Transmission electron microscopy (TEM) studies were carried out using JEM-2010 (JEOL, Tokyo, Japan), JEM-2200FS (JEOL, Tokyo, Japan) and Themis Z (Thermo Fisher Scientific, Waltham, MA, USA) electron microscopes operated at 200 kV. Images in Scanning-TEM (STEM) mode were acquired using high-angle annular dark field (HAADF) detectors. The local elemental analysis of the samples was studied by energy-dispersive X-ray (EDX) spectroscopy using a JEOL JED-2300 (JEM-2200FS) and Thermo Fisher Scientific Super-X (THEMIS Z) EDX detectors. The samples for the TEM studies were dispersed ultrasonically and deposited on copper grids covered with a holey carbon film.

The temperature-programed hydrogen reduction (H₂-TPR) was performed on a setup equipped with a flow reactor and a thermal conductivity detector [119]. To eliminate exothermic effects, a 100 mg sample with a size of 250–500 µm was mixed with 100 mg of quartz of the same size. The sample was then pre-treated at 450 °C for 1 h in a stream of air. The reduction was carried out at a heating rate of 10 °C/min from 25 to 900 °C in a flow of 10% H₂/90% Ar, 30 cm³/min.

3.3. Catalytic Activity Testing

The bi-reforming of CH₄ over prepared catalysts was studied in a fixed-bed flow quartz reactor with an inner diameter of 10 mm, at 1 atm, a gas flow rate of 200 ml_N/min, and a molar ratio between reagents of CH₄:CO₂:H₂O:He = 1:0.8:0.4:2.8. The molar ratio used was somewhat different from the stoichiometric CH₄:CO₂:H₂O molar ratio of 1:0.5:0.5. A higher CO₂/H₂O ratio (0.8 vs. 0.5) was applied in order to increase carbon dioxide utilization, while a higher O/C ratio (1.1 vs. 1) was used to reduce the rate of coking. Water was supplied by a piston pump at the required rate to the evaporator, where water vapor was mixed with other components (CH₄, CO₂, He) and quantitatively introduced into the fixed-bed reactor.

For catalytic activity testing, a 500 mg sample with a grain size of 250–500 µm was used. Initially, the activation of catalyst precursors was performed in situ by thermal treatment at 800 °C for 1 h in a 30% H₂/70% He flow.

The screening tests were performed in the stepwise temperature rise mode 650 → 850 °C. The heating rate was 10 degrees per minute; the holding time at each temperature was 40 min.

The stability tests were performed for 24 h at 800 °C.

The analysis of reaction mixtures was performed using the online automatic gas chromatography system Kristall 2000 m (Yoshkar-Ola, Russia) with a flame ionization detector and thermal conductivity detector.

The catalyst performance was characterized by CH₄ conversion ($X_{C{H}_4})$, CO₂ conversion ($X_{C{O}_2})$, yield of H₂ $(Y_{H_2})$, and yield of CO ($Y_{CO})$, which were calculated using the following formulas:

$${\mathrm{CH}}_4 \mathrm{conversion}: X_{C{H}_4}=100\times \left({\mathrm{F}}_{{\mathrm{CH}}_4}^{\mathrm{in}}-{\mathrm{F}}_{{\mathrm{CH}}_4}^{\mathrm{out}}\right)/{\mathrm{F}}_{{\mathrm{CH}}_4}^{\mathrm{in}}$$

$${\mathrm{CO}}_2 \mathrm{conversion}: X_{C{O}_2}=100\times \left({\mathrm{F}}_{{\mathrm{CO}}_2}^{\mathrm{in}}-{\mathrm{F}}_{{\mathrm{CO}}_2}^{\mathrm{out}}\right)/{\mathrm{F}}_{{\mathrm{CO}}_2}^{\mathrm{in}}$$

$${\mathrm{H}}_2 \mathrm{yield}: Y_{H_2}=100\times {\mathrm{F}}_{{\mathrm{H}}_2}^{\mathrm{out}}/\left(2{\mathrm{F}}_{{\mathrm{CH}}_4}^{\mathrm{in}}+{\mathrm{F}}_{{\mathrm{H}}_2\mathrm{O}}^{\mathrm{in}}\right)$$

$$\mathrm{CO} \mathrm{yield}: Y_{CO}=100\times {\mathrm{F}}_{\mathrm{CO}}^{\mathrm{out}}/\left({\mathrm{F}}_{{\mathrm{CH}}_4}^{\mathrm{in}}+{\mathrm{F}}_{\mathrm{CO}2}^{\mathrm{in}}\right)$$

where F_i is the molar flow rate of reagent (i) at the inlet (in) and outlet (out) of the reactor.

The thermodynamic equilibrium analysis was conducted using the software package IVTANTHERMO based on the minimization of the Gibbs free energy method [120,121].

4. Conclusions

Thus, (CeM)_1−xNi_xO_y (M = Al, La, Mg) materials served as precursors of Ni catalysts during the bi-reforming of methane. Their genesis features as well as textural, structural, and redox characteristics and functional properties were regulated by variation in the type and content of M. In comparison to the M-free sample, for (CeM)_1−xNi_xO_y with an optimal M/Ce ratio, the size of crystallites decreases, the defectiveness and the thermal stability of the material increases, and the temperature of formation of the active Ni⁰ phase rises (ΔT = 100–200 °C), while the temperature of its re-oxidation decreases (ΔT = 70 °C). This all results in the improvement of the strength of the metal–support interaction and provides a more stable and higher yield of hydrogen.

In the case of M = Al, the as-synthesized samples have a homogenous phase composition of a solid fluorite-like solution with mesoporous texture and spongy morphology. Among studied M, the Al-series is characterized by the highest burnout temperature of the organic matrix during synthesis by the PC method, a higher specific surface area, and the smallest solid solution crystallite size (CSR 2.8–4.5 nm). For higher specific surface area (80–100 m²/g) the Al/Ce molar ratio should be equal to ≤1. In (CeAl)_1−xNi_xO_y samples, the cations of Ni²⁺ are stabilized in the structure of a solid solution and their exsolution under a reducing atmosphere leads to the formation of the Ni⁰ phase at 600 °C. A feature of the Al-series is that, under activation and reaction conditions, a complete transformation of the oxide matrix occurs: CeO₂-based solid solution → Ni⁰ + cerium aluminate of perovskite structure (Al/Ce = 1). This is accompanied by a significant decrease in the specific surface area and pore volume. (CeAl)_1−xNi_xO_y samples provide the highest H₂ yield for the bi-reforming of methane and moderate stability against coking.

In the case of M = La and synthesis by the PC method, the formation of a solid fluorite-like solution (CSR 4.0–6.0 nm) with the highest concentration of oxygen vacancies is observed, and only an insignificant fraction of nickel is stabilized as NiO nanoparticles (3–5 nm) on the sample surface. Among studied M, for samples of the La-series, the burnout of the organic matrix occurs at the lowest temperature. They are characterized by a rapid decrease in the specific surface area and an increase of the defectiveness of the structure with increasing La/Ce molar ratio. A feature of the La-series is that the exsolution of Ni from solid solution is highly hindered, and the Ni⁰ phase appears only at a reduction temperature of 650 °C, while the Ce-La-O-solid solution remains. (CeLa)_1−xNi_xO_y samples provide the lowest H₂ yield for the bi-reforming of methane and poor stability against coking.

In the case of M = Mg, a two-phase system is already observed for the freshly synthesized sample, which may be due to the low stability of the Mg–citrate complex and partial segregation of the MgO phase at the earliest stage of the complex oxide formation. Therefore, in addition to a solid fluorite-like solution, the formation of a NiO-MgO solid solution (CSR 3.6–8.5 nm) is observed. Due to this, samples have the lowest reducibility. After activation, a three-phase system is formed, including Ni⁰, MgO, and a CeO₂-based fluorite-like phase. (CeMg)_1−xNi_xO_y samples provide high H₂ yield during the bi-reforming of methane and the highest stability against coking.

Taking into account three factors (activity, stability, and the coke accumulation rate), the efficiency of methane bi-reforming catalysts increases in the following order: M-free < La < Al < Mg. The composition Ce_0.6Mg_0.4Ni_0.2O_1.6 is considered the best. During the bi-reforming of methane over Ce_0.6Mg_0.4Ni_0.2O_1.6, a stable H₂ yield of 90% is achieved at a high level of CO₂ and CH₄ conversions (>85%).