Novel Nickel- and Magnesium-Modiﬁed Cenospheres as Catalysts for Dry Reforming of Methane at Moderate Temperatures

: Cenospheres from coal ﬂy ashes were used as support in the preparation of Ni–Mg catalysts for dry reforming of methane. These materials were characterized by means of XRD, H 2 -temperature-programmed reduction (H 2 -TPR), CO 2 -temperature-programmed desorption (CO 2 -TPD), and low-temperature nitrogen sorption techniques. The cenosphere-supported catalysts showed relatively high activity and good stability in the dry reforming of methane (DRM) at 700 ◦ C. The catalytic performance of modiﬁed cenospheres was found to depend on both Ni and Mg content. The highest activity at 750 ◦ C and 1 atm was observed for the catalyst containing 30 wt % Mg and 10, 20, and 30 wt % Ni, yielding to CO 2 and CH 4 conversions of around 95%.


Introduction
The policies of the European Union, with new laws and regulations which aim to protect the environment, have led to increased interest in finding new and efficient applications of coal combustion products. Currently, the total annual production of coal fly ash is around 780 million tons worldwide [1], which makes it necessary to carry out detailed studies on its alternative applications [2][3][4]. Cenospheres, present in fly ash, are hollow particles of about 10-1000 µm in diameter, with a density lower than 1 g/cm 3 [5]. These could become catalyst supports for certain chemical processes that require high thermal stability [6,7]. Cenospheres are composed of a mixture of aluminosilicates with some amount of K, Ca, Fe, Mg, and a small addition of Ti, Na, P, S, and trace elements [7]. Because of the high amount of Al 2 O 3 (23%-33%) and the fact that they were formed at temperatures over 1200 • C, cenospheres have high mechanical and thermal strength, and thus have potential applications in many branches of industry [8]. Cenospheres were examined and have exhibited positive results as catalysts, for example, for esterification of n-octanol with acetic acid [6], for deNOx processes [9], as well as in practical uses in the degradation of pollutants or for water cleanup [10]. Additionally, it was concluded that when used as catalysts, they can show high conversion and selectivity for some important organic reactions under solvent-free conditions [11].
Apart from the problem of solid waste, another important issue connected with energy production from fossil fuels is the high emission of carbon dioxide. The increase of CO 2 concentration in the atmosphere is considered to be the main reason for climate change [12]. Utilization and transformation of methane and carbon dioxide into valuable chemicals could be an interesting alternative to CO 2 sequestration [13]. Jeong et al. [14] tested catalysts based on Pd in catalytic oxidation of methane. Additionally, Zhang et al. [15] showed that bimetallic alloyed Pt 71 Co 29 catalysts exhibited high activity for methanol oxidation reactions. Torimoto et al. [16] tested Ni, Pt, Pd, and Rd catalysts in steam reforming of methane. The low-temperature heterogeneous catalytic reactions for hydrogen production are enabled with very high yield. Mota and Kim [17] described promising features of Ni-, Co-, and Fe-based catalytic systems on CO 2 reduction reactions (CO 2 RR), namely thermochemical, electrochemical, and photochemical. Thermochemically driven CO 2 RR has significant potential to be applied on a large scale.
The dry reforming of methane (DRM) is a chemical process that converts CH 4 and CO 2 to syngas with a H 2 /CO ratio of 1, which can be directly used in many important chemical reactions, among them Fischer-Tropsch synthesis [18][19][20][21].
The main reaction (CH 4 + CO 2 = 2CO + 2H 2 ) is efficient at temperatures higher than 730 • C, and is accompanied by several side reactions, among which the Boudouard reaction (2CO = C + O 2 ) and methane decomposition are especially responsible for carbon deposit formation during DRM [22]. This process has not as yet been applied on an industrial scale due to some complications connected with catalyst deactivation (carbon deposition formation) [18,22,23]. However, Linde and BASF installed the Pilot Plant in Wilsonville, Alabama, USA in 2015 [24]. The catalysts that have been most often examined for dry reforming of methane are nickel, ruthenium, and rhodium [18,25], supported on different carriers.
Although the noble metals are more resistant to carbon deposition, as well as more active than nickel, they are more expensive and less available. For this reason, in order to allow the DRM to become a fully mature industrial process, there is a need to develop a Ni-based catalyst which is more active and more stable, as well as low-cost and eco-friendly [26]. To the best of our knowledge, no studies concerning the use of cenospheres as catalysts for dry reforming of methane have been reported in the literature. These cenospheres as by-products of fly ashes are characterized by their high thermal stability [23,27,28], which is important for high-temperature processes such as reforming (steam, dry, oxy, etc.). Thus, the aim of this work was to use these cenospheres as supports for catalysts in DRM. Moreover, although the role of Mg in DRM is not yet fully clear, several reports (e.g., that of Dębek et al. [29]) have claimed that the high Lewis basicity of MgO has a beneficial effect, since CO 2 adsorption is enhanced on basic support. Thus, in this work, in order to improve the basicity of the studied catalysts, cenospheres were modified with MgO in addition to Ni.  Table 1, and are higher than for some other supported Ni-catalysts reported in the literature, such as Ni/Mg/Al hydrotalcite-derived materials [30] or Ni/Ce/Fe clay-based samples [31]. For the catalysts studied here, there is greater Ni content of 20%-30% as compared to 10%. However, the crystallite sizes for the catalysts studied in this work were smaller than those reported in the literature for 11% NiO/CaAl 2 O 4 catalysts (65 nm) [32].  Temperature-programmed reduction (H2-TPR) profiles for the catalysts after calcination are presented in Figure 2. A single symmetric reduction peak assigned to the reduction of nickel oxide is visible from 280 °C to 420 °C, with the maximum at ~350 °C, which is in good agreement with the literature [33,34]. For the catalysts with a loading of 30% Mg and the lowest nickel content, a shift to  Temperature-programmed reduction (H 2 -TPR) profiles for the catalysts after calcination are presented in Figure 2. A single symmetric reduction peak assigned to the reduction of nickel oxide is visible from 280 • C to 420 • C, with the maximum at~350 • C, which is in good agreement with the literature [33,34]. For the catalysts with a loading of 30% Mg and the lowest nickel content, a shift to higher reduction temperatures can be observed, which may indicate stronger interactions between the support and the active phase [30,35].

Catalyst Characterization
higher reduction temperatures can be observed, which may indicate stronger interactions between the support and the active phase [30,35]. H2 consumption (Table 1) increased with the loading of Ni and Mg, with the exception of Mg30Ni30 as compared to Mg20Ni30. It should be mentioned, however, that H2 consumption is rather low in comparison to some catalysts reported in the literature, for example, for 10 wt % Ni/CZ/SBA-15 (1634 µ mol/g) [19] or 15 wt % Ni-Ce-Fe clay (1470 µ mol/g) [31]. The obtained values in Table 1 are close to the ones that were reported for mixed oxides derived from double-layered hydroxides containing ~20% Ni and ~30% Mg [36].
The latter result indicates a lower reducibility of the microsphere-supported catalysts, possibly due to stronger interaction between NiO and MgO/microspheres. A CO2 temperature-programmed desorption (TPD) experiment was conducted to determine the basicity of the studied cenosphere catalysts. The results are reported in Table 1. The basicity of the calcined samples is relatively low, from ~18 to ~48 µ mol/g, in comparison to some catalysts reported in the literature, for example, nickel contacting layered double hydroxide-derived catalysts with a basicity of around 130 µ mol/g for Ni/La/Mg hydrotalcite [37] or 201 µ mol/g for Ni/MgO-ZrO2 [38], 350 µ mol/g for 12 wt % Ni/KIT-6, and 190 µ mol/g for (12 wt % Ni, 8 wt % Y)/KIT-6 [39].
This may be assigned to the lower specific surface areas of our studied catalysts. Figure 3 presents results for catalytic tests for the obtained samples. One can note that the conversions of CO2 and CH4 increase with the temperature, and that at 750 °C, the conversions of CH4 and CO2 are both higher than 93%, and the H2/CO is 0.98. For the sake of comparison, the supports were tested in DRM as a function of the temperature, and at 750 °C, and no significant activity was found.

DRM Catalytic Tests
In addition, activities in CO2, CH4, and H2/CO increased with the increase in Ni content, to an extent depending on the Ni/Mg ratio. This may be ascribed to the stronger metal support interaction, as well as to a higher stabilization of NiO species [40].  (Table 1) increased with the loading of Ni and Mg, with the exception of Mg30Ni30 as compared to Mg20Ni30. It should be mentioned, however, that H 2 consumption is rather low in comparison to some catalysts reported in the literature, for example, for 10 wt % Ni/CZ/SBA-15 (1634 µmol/g) [19] or 15 wt % Ni-Ce-Fe clay (1470 µmol/g) [31]. The obtained values in Table 1 are close to the ones that were reported for mixed oxides derived from double-layered hydroxides containing 20% Ni and~30% Mg [36].
The latter result indicates a lower reducibility of the microsphere-supported catalysts, possibly due to stronger interaction between NiO and MgO/microspheres. A CO 2 temperature-programmed desorption (TPD) experiment was conducted to determine the basicity of the studied cenosphere catalysts. The results are reported in Table 1. The basicity of the calcined samples is relatively low, from~18 to~48 µmol/g, in comparison to some catalysts reported in the literature, for example, nickel contacting layered double hydroxide-derived catalysts with a basicity of around 130 µmol/g for Ni/La/Mg hydrotalcite [37] or 201 µmol/g for Ni/MgO-ZrO 2 [38], 350 µmol/g for 12 wt % Ni/KIT-6, and 190 µmol/g for (12 wt % Ni, 8 wt % Y)/KIT-6 [39].
This may be assigned to the lower specific surface areas of our studied catalysts. Figure 3 presents results for catalytic tests for the obtained samples. One can note that the conversions of CO 2 and CH 4 increase with the temperature, and that at 750 • C, the conversions of CH 4 and CO 2 are both higher than 93%, and the H 2 /CO is 0.98. For the sake of comparison, the supports were tested in DRM as a function of the temperature, and at 750 • C, and no significant activity was found.

DRM Catalytic Tests
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 10 Thus, it can be concluded that both CO2 and CH4 conversions and H2/CO ratios depend on the Ni/Mg ratio, which is in agreement with the work of Dębek et al. [35].

Stability Test
The Mg30Ni20 catalyst which presented the highest activity in DRM as a function of temperature was selected for isothermal runs at 700 °C after an initial reduction under hydrogen at 800 °C ( Figure  3). Under isothermal steady-state conditions, only a slight deactivation was observed. CH4 conversion decreased from 91.4% to 86.6%, and CO2 conversion from 92.3% to 90.1%, while H2/CO ratio decreased by 6.1% after 7 h runs ( Figure 4).  In addition, activities in CO 2 , CH 4 , and H 2 /CO increased with the increase in Ni content, to an extent depending on the Ni/Mg ratio. This may be ascribed to the stronger metal support interaction, as well as to a higher stabilization of NiO species [40].
Thus, it can be concluded that both CO 2 and CH 4 conversions and H 2 /CO ratios depend on the Ni/Mg ratio, which is in agreement with the work of Dębek et al. [35].

Stability Test
The Mg30Ni20 catalyst which presented the highest activity in DRM as a function of temperature was selected for isothermal runs at 700 • C after an initial reduction under hydrogen at 800 • C (Figure 3). Under isothermal steady-state conditions, only a slight deactivation was observed. CH 4 conversion decreased from 91.4% to 86.6%, and CO 2 conversion from 92.3% to 90.1%, while H 2 /CO ratio decreased by 6.1% after 7 h runs (Figure 4).

Post-Test Characterization of the Catalysts
Our results for H2-TPR, as well as the particle sizes of Ni crystallites for the samples after DRM, are summarized in Table 1. H2 consumption indicates that a small amount of NiO was present after DRM. The presence of this reduction peak in the spent catalysts may point out that Ni particles were repeatedly oxidized and reduced during DRM process [35,41] (Table 1).
The XRD patterns for the catalysts after the DRM catalytic tests showed reflections of Ni 0 with the exception of Mg10Ni10 (Figure 1). Nickel crystallite sizes, calculated from XRD for the spent catalysts, are reported in Table 1.
In comparison to the reduced catalysts, the sizes of Ni particles are similar, proving the lack of sintering during DRM. For Mg20Ni20 and Mg20Ni30 the size of the Ni particle even decreased, which may have been caused by some redispersion of Ni crystallites. Similar effects were already observed for other types of mixed oxides, catalyst promoted or not [30,42]. Although the literature indicates carbon deposition in the form of graphite after DRM reactions on several types of catalysts-such as hydrotalcite-derived samples modified with Ce [30] or SBA-15-supported nickel catalysts promoted by Ce0.75Zr0.25O2 mixed oxide [19]-for the studied materials, graphite carbon deposits could neither be confirmed by XRD nor excluded because at the relevant 2θ (°) region, approximately 26°, reflections from AlXSiYOZ (mullite) and Al2O5Si (sillimanite) from microspheres are also present [5].
In order to confirm the presence of carbon, a thermogravimetric analysis was performed in air from 100 to 1000 °C (figure not shown). For Mg10Ni10, Mg20Ni10, Mg30Ni10, and Mg30Ni20, only traces of carbon deposits of 0.5%, 0.6%, 0.9%, and 1.1% were observed (Table 1).
Until now, cenospheres have not been studied in DRM. Table 2 compares the results with other catalysts. Catalysts with similar Ni content or similar reaction conditions were selected for comparison, results are fully comparable because the experiments were carried out under the same conditions, and the presented results were selected for the best catalysts. The cenospheres examined in this work obtained better catalytic results. Dębek et al. [43] obtained similar results for catalysts containing 20 wt % Ni. The only slight differences concern the H2/CO ratio.

Post-Test Characterization of the Catalysts
Our results for H 2 -TPR, as well as the particle sizes of Ni crystallites for the samples after DRM, are summarized in Table 1. H 2 consumption indicates that a small amount of NiO was present after DRM. The presence of this reduction peak in the spent catalysts may point out that Ni particles were repeatedly oxidized and reduced during DRM process [35,41] (Table 1).
The XRD patterns for the catalysts after the DRM catalytic tests showed reflections of Ni 0 with the exception of Mg10Ni10 (Figure 1). Nickel crystallite sizes, calculated from XRD for the spent catalysts, are reported in Table 1.
In comparison to the reduced catalysts, the sizes of Ni particles are similar, proving the lack of sintering during DRM. For Mg20Ni20 and Mg20Ni30 the size of the Ni particle even decreased, which may have been caused by some redispersion of Ni crystallites. Similar effects were already observed for other types of mixed oxides, catalyst promoted or not [30,42]. Although the literature indicates carbon deposition in the form of graphite after DRM reactions on several types of catalysts-such as hydrotalcite-derived samples modified with Ce [30] or SBA-15-supported nickel catalysts promoted by Ce 0.75 Zr 0.25 O 2 mixed oxide [19]-for the studied materials, graphite carbon deposits could neither be confirmed by XRD nor excluded because at the relevant 2θ ( • ) region, approximately 26 • , reflections from Al X Si Y O Z (mullite) and Al 2 O 5 Si (sillimanite) from microspheres are also present [5].
In order to confirm the presence of carbon, a thermogravimetric analysis was performed in air from 100 to 1000 • C (figure not shown). For Mg10Ni10, Mg20Ni10, Mg30Ni10, and Mg30Ni20, only traces of carbon deposits of 0.5%, 0.6%, 0.9%, and 1.1% were observed (Table 1).
Until now, cenospheres have not been studied in DRM. Table 2 compares the results with other catalysts. Catalysts with similar Ni content or similar reaction conditions were selected for comparison, results are fully comparable because the experiments were carried out under the same conditions, and the presented results were selected for the best catalysts. The cenospheres examined in this work obtained better catalytic results. Dębek et al. [43] obtained similar results for catalysts containing 20 wt % Ni. The only slight differences concern the H 2 /CO ratio.

Ni/Mg Cenosphere Catalyst Preparation
White commercial cenospheres (~32 wt % of Al 2 O 3 and~53 wt % SiO 2 ) from Cenospheres Trade & Engineering S.A. (Radom, Poland) were tested as supports for DRM catalysts. Microspheres were modified first by wet impregnation with Mg(NO 3 ) 2 , followed by the incipient wetness method using Ni(NO 3 ) 2 , as described by Kolebuk and Samojeden [45]. We then introduced 10, 20, or 30 wt % of Mg onto the samples, and then 10 wt %, 20 wt %, or 30 wt % of Ni. Each impregnation step was followed by drying at 100 • C and subsequent calcination at 500 • C for 6 h [45]. The catalysts were designated as MgxNiy, where x and y are amounts of introduced Mg and Ni, respectively; for example, Mg10Ni10 denotes the white cenospheres modified with 10 wt % of Ni and 10 wt % of Mg.

Catalyst Characterization
XRD analysis was conducted to examine the structure and phase composition of the samples, as well as to determine Ni crystallite sizes before and after DRM catalytic tests. The average crystallite size of NiO was calculated from Ni reflection at~44 • using the Scherrer equation. The XRD Empyrean diffractometer from Panalytical (Almelo, UK), equipped with a CuKα radiation source (λ = 0.154059 nm), was used for these measurements.
The S BET of the catalysts was determined by low-temperature N 2 sorption using a Belsorp Mini II apparatus (BEL, Osaka, Japan). Before each experiment, samples were degassed under vacuum for 3 h at 180 • C. To calculate the specific surface areas, the Brunauer-Emmett-Teller (S BET ) method was used.
Reducibility of the catalysts was evaluated by a temperature-programmed reduction (TPR) of H 2 (H 2 -TPR) with a BELCAT-M (from BEL, Osaka, Japan), equipped with a thermal conductivity detector (TCD). The sample (~60 mg) was loaded into a quartz reactor, and prior to the TPR measurements it was pretreated at 100 • C for 120 min in flowing He. Then, a 5% (v/v) H 2 /Ar gas mixture was introduced, and the catalyst was heated from room temperature to 900 • C at a rate of 7.5 • C/min, with a simultaneous measurement of hydrogen consumption.
The basicity of the catalysts was determined using CO 2 -TPD with the BELCAT-M. The materials (~60 mg) were pretreated under He flow at 500 • C for 120 min and then cooled down to 80 • C. Subsequently, a mixture of 10% (v/v) CO 2 /He was introduced for 60 min. After adsorption, the sample was flushed with He for 15 min to desorb the physically adsorbed CO 2 . Then the sample was heated up to 800 • C at the rate of 10 • C/min under He flow, and CO 2 desorption was measured by the TCD.
The thermogravimetric non-isothermal measurements were made using a DynTHERM Thermogravimetric Analyzer by Rubotherm in the air atmosphere (100-1000 • C).

Catalytic Tests
DRM activity tests were performed under atmospheric pressure in a U-shaped quartz microreactor under 100 mL/min with the following composition: CH 4 /CO 2 /Ar = 1/1/8 as function of temperature from 550 to 750 • C. For each temperature, the catalyst was kept for approximately 30 min, corresponding to steady-state measurements. Another DRM test was conducted at 700 • C for 7 h in order to favor