Highly Stable Ni–Red Mud Catalysts for CO₂-Free Hydrogen and Valuable Carbon from Natural Gas

Abstract

The utilization of red mud as a catalyst support has been investigated to produce high-value carbon and CO_x-free hydrogen from natural gas. Nickel impregnation between 10 wt% to 20 wt% over red mud generates more active species in the form of nickel oxide; however, nickel–red mud interaction also generates less active spinel species (NiFe₂O₄). The red mud itself deactivates quickly during the production of hydrogen from the decomposition of methane; however, nickel-based red mud-supported catalysts have shown significant improvement in the activity results. For instance, the catalyst with 20 wt% nickel supported by red mud demonstrates a stable methane conversion as high as 75%. The reduction kinetics analysis demonstrated the lowest reduction in activation energy of 83 kJ/mol for 20Ni-PRM which played a major role in the excellent activity and stability of this catalyst. The post-reaction catalyst characterization results indicate the formation of multi-walled carbon nanotubes, as evidenced by high resolution transition electron microscope and thermogravimetric analyses.

1. Introduction

Aluminum production from the Bayer process produces a waste material with strong alkalinity known as red mud (RM) [1,2,3], and no effective utilization has been found for this waste material; hence, RM is discarded either in landfill or the sea. However, RM manifests environmental repercussions, largely associated with the higher contents of toxic elements as well as its high alkalinity, with a pH ranging from 10 to 12.5, and high content of toxic trace elements leading to the contamination/pollution of soil, groundwater, and marine life [1].

RM primarily comprises compounds based on iron, alumina, silica, titania, alkali, and alkaline earth metals. The possibility of separating RM components may open an opportunity to recycle RM in an economical way. RM can also be processed to achieve a material with a targeted application. For instance, the synthesis of hierarchical porous microspheres of alumina from RM serves as an efficient adsorbent for dye removal [4]. The high iron contents of RM can also find its utilization as a volatile organic compound (VOC) oxidation catalyst [5,6,7]. Under a reductive environment, iron contents of RM can play a role in breaking carbon–carbon and/or carbon–hydrogen bonds, i.e., cracking/pyrolysis of hydrocarbons [8,9,10].

The production of pure hydrogen with zero carbon footprint via catalytic methane decomposition (CMD) serves as an excellent technology and has gained significant attention in recent years [11,12,13,14,15]. Among the transition metal-based catalysts investigated for catalytic methane decomposition, iron-based catalysts [16,17,18,19,20,21,22,23,24,25,26,27,28] have been found to demonstrate excellent activity during CMD, regardless of the type of reactor, i.e., fixed-bed [16,17,21,22,23,24,25] and/or fluidized-bed reactors [18,19,26,27,28]. Therefore, scientists have made efforts to test RM, which is rich in iron, for catalytic methane decomposition. The utilization of RM as a catalyst during the catalytic decomposition of ethylene produced multi-walled carbon nanotubes (MWCNT) at 650 °C in a fluidized bed reactor [29]. It was found that RM generated MWCNT as high as 3.75 gC/g_cat. In another study, Balakrishnan et al. [30] reduced the oxides of iron present in RM, stepwise, at 800 °C under methane and tested for the catalytic decomposition of methane. They reported a carbon deposition of 47.7% during 6 h time-on-stream at a similar reaction temperature of 650 °C. The evaluation of a catalyst synthesized from modified RM demonstrated a methane conversion of 26% at 800 °C [31]. Geng et al. [32] investigated the role of residual sodium oxide (Na₂O) in RM during catalytic methane decomposition and found that the amount of sodium oxide not only suppresses the activation of methane over the catalyst surface but also reduces maximal activity. They concluded that sodium oxide dispersed over iron oxide caused an inhibition effect leading to reduced activity of RM-based catalysts.

The earlier discussion infers that no work has been reported for metal-impregnated red mud catalysts and/or utilization of red mud as a support for catalytic methane decomposition. Hence, in this work, the role of RM as support for nickel-based catalysts with various amounts of nickel (10 to 20 wt%), synthesized by the deposition–precipitation method, was analyzed for pure hydrogen production from catalytic methane decomposition. The as-synthesized catalysts were also characterized to identify their crystalline nature, reduction behavior, metal-support interaction, and porosity.

2. Results and Discussion

2.1. Characterizations of Catalysts Prior to Reaction

Figure 1 shows the X-ray diffraction data of PRM-supported nickel-based catalysts with varying amounts of nickel. XRD profiles exhibit various peaks that represent components of both PRM, nickel, and their combination. For instance, XRD peaks of PRM support show diffractions at 2-θ values of 27.5 and 47.6° that are attributes of TiO₂ while peaks assigned to hematite are found at 2-θ values of 32.5, 35.5, 41, 49.5, 54.5, 63, and 64.5° [25,26,33]. Similarly, the diffraction peaks appearing at 2-θ values of 14, 19, 22, 28, 32.5, and 52° are assigned to alumina and calcium/sodium aluminosilicates. In the case of PRM-supported nickel-based catalysts, added peaks represent both nickel oxide and the combination of nickel oxide with hematite in the form of NiFe₂O₄ [34]. The nickel oxide crystallite size was measured using the Scherrer equation (size = 0.9λ/βCosθ) and sizes were found to be 9, 11, and 13 nm for 10Ni-PRM, 15Ni-PRM, and 20Ni-PRM, respectively, indicating slight agglomeration with an increase in nickel amounts.

To explore the structure, nature, morphology, and size of the fresh catalyst’s particles, HRTEM analysis along with selected area electron diffraction (SAED) was conducted for fresh 15Ni-PRM and 20Ni-PRM, as presented in Figure 2. The dispersed nickel oxide particles (for 15Ni-PRM) over the support surface can be seen in Figure 2a where particle diameter varies between 2 and 20 nm. Moreover, the high-resolution image (Figure 2c) indicates crystal fringes corresponding to a d-spacing value of 0.24 nm representing a nickel oxide crystal plane of (111). XRD data in Figure 1 is further established with the SAED polycrystalline ring shown in the inset of Figure 2c.

In the case of 20Ni-PRM, the size of nickel oxide particles is found to be increased (5–30 nm), which could be assigned to metal particle aggregation (Figure 2b). The role of larger nickel particles during the CDM reaction is crucial and discussed in Section 3.2.

Hydrogen-based temperature-programmed reduction serves as a useful technique to evaluate the extent of interaction between metal particles and support as well as to measure the degree of reducibility since both factors play a significant role during CDM reaction. Figure 3 demonstrates the reducibility patterns of PRM-supported Ni-based catalysts, where two reduction peaks are found for pure PRM.

The XRD results of Figure 1 show that hematite was the main component of PRM that requires a two-step reduction in hematite to be reduced to iron metal form, as indicated by two reduction peaks. The first broader peak appearing between 385 and 740 °C (with peak maxima at 610 °C), could be ascribed to the first step of reduction where isolated Fe₂O₃ reduce to Fe₃O₄. The second relatively sharper peak within the temperature range of 740 and 900 °C (with peak maxima at 810 °C), represent a second reduction step where Fe₃O₄ either reduced to FeO or Fe [35,36]. The Ni-impregnated catalysts indicated a clear difference in their reduction profiles in comparison with PRM. For instance, the lower loading 10Ni-PRM catalyst showed the first reduction peak, with a slight shift towards left, between 360 and 730 °C (with peak maxima at 580 °C). Similarly, the second reduction peak also shifts to a lower temperature, appearing between 730 and 900 °C (with peak maxima at 790 °C). This observation suggests that lower Ni loading promoted metal oxide reduction. However, the amount of hydrogen consumed for 10Ni-PRM, i.e., 2.77 mmol/g, is slightly less than the hydrogen uptake of PRM, i.e., 2.82 mmol/g, revealing a slight loss of reducible species after Ni impregnation. In the case of higher loading catalysts, i.e., 15Ni-PRM and 20Ni-PRM, both of these catalysts have approximately similar reduction peaks, where first reduction peak appears between 370 and 760 °C (with peak maxima at ~610 °C) and second reduction peak is observed within temperature range of 760 to 900 °C (with peak maxima at 845 and 840 °C for 15Ni-PRM and 20Ni-PRM, respectively). These findings suggest that metal-support interaction is promoted in the case of higher loading catalysts, resulting in NiFe₂O₄ spinel formation. Furthermore, the amount of hydrogen consumed for 15Ni-PRM, i.e., 3.33 mmol/g, as well as 20Ni-PRM, i.e., 3.40 mmol/g, remained higher than the hydrogen uptake amount of PRM, i.e., 2.82 mmol/g, leading to an increased number of reducible species in higher loading catalysts. In summary, metal oxides are easily reduced at a lower loading of nickel catalyst, i.e., 10Ni-PRM, while reduction in metal oxide becomes crucial once nickel loading is increased. Moreover, the varying nickel amount could also affect the reduction kinetics (as discussed below), and all these factors would play a role in influencing the catalytic performance of these catalysts.

The hydrogen consumed during the experiment (H₂-TPR) is determined by evaluating the area under the curve for each peak corresponding to each of the analyzed catalysts. The integration of these distinct areas under the curve corresponding to reduction peaks results in the amount of hydrogen consumed. Therefore, the hydrogen consumption was evaluated using three alternate tools including the built-in software of Micromeritics AutoChem-II V 4.03 (Micromeritics Instrument Carporation, Norcross, GA, USA), MATLAB (version 23.3) programming software for numerical integration, and Origin Pro (9.1) graphical software, and for all these cases the analysis trend remained the same.

The hydrogen consumption trends are compared, and it is obvious that the calculated areas under the curve correspond to ascending order of 20Ni-PRM > 15Ni-PRM > PRM > 10Ni-PRM, in agreement with the hydrogen uptake of 3.40, 3.33, 2.82, and 2.77 mmol/g for 20Ni-PRM, 15Ni-PRM, PRM, and 10Ni-PRM, respectively, and these consistent results acknowledge the calculation reliability. Upon comparison of the results, it is found that the 20Ni-PRM catalyst with a hydrogen uptake of 3.40 mmol/g displays the consumption of highest amount of hydrogen among the analyzed catalysts, while the support (PRM) also shows a relatively significant area under the curve. Since the degrees of metal oxide reduction and hydrogen consumption are directly correlated, the quantitative evaluation of the amount of hydrogen consumed during H₂-TPR becomes significant.

In the case of solid catalysts (in oxide form), the overall reaction rate expression as a function of the composition of species in the gas phase as well as extent and/or degree of solid catalyst reduction [$f\left(ϵ\right)$] represents the kinetics model based on TPR data [37]:

$${\displaystyle \frac{d\left[ϵ\left(t\right)\right]}{dt}}=k\left(T\right)f\left(ϵ\right)f\left(p_{H_2},p_{H_{2}O}\right)$$

(1)

where ($ϵ$) corresponds to the progress of the reaction and the associated expression of reaction progress is dependent upon existing measured variable(s). Generally, the degree and/or extent of reduction reaction is defined several ways such as by evaluating the variation in sample mass, detecting the change in gas being evolved or being consumed, and/or measuring the heat consumed or evolved.

In this work, H₂-TPR data are utilized to evaluate the degree of reduction and/or activation. The following Equation (2) represents the calculation of transient conversion [$ϵ\left(t\right)$] related to the solid-state:

$$ϵ(t)={\displaystyle \frac{\Delta n_t}{\Delta n_{total}}}$$

(2)

where $\Delta n_t$ refers to the molar amount of hydrogen uptake at any time $t$ (in minutes) and $\Delta n_{total}$ represents the total molar amount of hydrogen that is required for a complete reduction in the metal oxides present in the catalyst.

The function associated with partial pressures, i.e., [$f\left(p_{H_2},p_{H_{2}O}\right)]$, in Equation (1) can be ignored based on constant flow rates and feed composition along the length of the differential reactor that leads to a reduced form of Equation (1) in the form of Equation (3);

$${\displaystyle \frac{d\left[ϵ\left(t\right)\right]}{dt}}=k\left(T\right)f\left(ϵ\right)$$

(3)

where the rate constant $k\left(T\right)$ can be defined by the famous Arrhenius equation as shown in Equation (4)

$$k=k_0\mathit{exp}\left[{\displaystyle \frac{-E_a}{R}}\left({\displaystyle \frac{1}{T}}-{\displaystyle \frac{1}{T_c}}\right)\right]$$

(4)

where $k_0$ denotes the pre-exponential factor and $E_a$ represents the activation energy. The centering temperature, i.e., $T_c$, can facilitate the minimizing of parameter cross-correlation. The evaluation of both $E_a$ and $k_0$ necessitate the split of $k\left(T\right)$ and $f\left(ϵ\right)$ in Equation (3). In fact, a kinetic model aimed at measuring the expression of rate of reaction does not directly affect $f\left(ϵ\right)$. In other words, $f\left(ϵ\right)$ is hardly a function of such a kinetic model. The earlier reported kinetic models comprise empirical and mechanistic models. The reported works have also evaluated the complications involved in overall reduction reaction mechanism. The steps involved in the reduction of a metal oxide, in sequence, are also described in earlier works [37,38] and is as follows. (i) The induction period allows the initial dissociation of molecular hydrogen, where metal oxide plays the major role in the dissociation of hydrogen. The earlier reduced metal oxide particles facilitate grain growth that in turn enhances the hydrogen dissociation process. In other words, hydrogen dissociation initiated by the metal oxide throughout the induction period is further promoted by previously reduced metal oxides while the reduction front interacts with the grain. (ii) The diffusion of atomic hydrogen through the solid surface ends up at a reduction/activation center. (iii) Metal oxide bond alienation occurs. (iv) Metal atoms nucleate to form clusters of metal species. And finally, (v) the resulting metallic clusters are subjected to further growth to turn into metal crystallites. These steps significantly affect the reduction rate.

Regarding heterogeneous catalysis, the accessibility of grains of metal oxides significantly influences the rate of reduction reactions. Similarly, factors such as the shape and size of metal crystallites, along with the catalyst’s pore size distribution, play a key part in determining the availability of oxide sites. The innate chemical properties of the support material within a heterogeneous type of the catalyst have a profound influence on aligning the active component, which in turn affects the reduction behavior of the catalyst [39]. The kinetics of solid–gas reaction were investigated, and alternating kinetics models of various type were also proposed [40]. The profile of degree of conversion of reduction reaction represented by [ϵ(t)] as a function of time (t) provides a base for the selection of the appropriate kinetic model of a solid–gas reaction. The profile of [ϵ(t)], in the form of S shape, observed during the catalyst’s TPR reaction indicates a multistep gas–solid reduction reaction, involving stages such as nuclei formation, successive progress and/or growth, particle accumulation/agglomeration, displacement, and formation of point defect. In the current work, we carried out parametric estimation via various kinetic models related to a solid-state kinetics such as the power law (PL) model, two- and three-dimensional Avrami–Erofeev (AEM) model, and random nucleation (RN) model. These models are mathematically expressed in

Table 1. Mathematical expressions of kinetic models examined for [$f\left(ϵ\right)$].

No.	Mechanism	Model Formulation
1	Random nucleation	$f\left(ϵ\right)=1-ϵ$
2	Power law model	$f\left(ϵ\right)={\left(1-ϵ\right)}^n$
3	Avrami–Erofeev model	$f\left(ϵ\right)=n\left(1-ϵ\right){\left[-\mathit{ln}\left(1-ϵ\right)\right]}^{\left(n-1\right)/n}$
4	Two-dimensional nuclei growth *	$f\left(ϵ\right)=2\left(1-ϵ\right){\left[-\mathit{ln}\left(1-ϵ\right)\right]}^{1/2}$
5	Three-dimensional nuclei growth **	$f\left(ϵ\right)=3\left(1-ϵ\right){\left[-\mathit{ln}\left(1-ϵ\right)\right]}^{2/3}$

* 2D AEM kinetic model ** 3D AEM kinetic model.

.

The experimental data for reduction conversions within 0 and 1 (0 < $ϵ$ < 1) and/or numerical fitting of Equations (3) and (4) leads to the expression of Equation (5):

$${\displaystyle \frac{d\left[ϵ\left(t\right)\right]}{dt}}=k_0\mathit{exp}\left[{\displaystyle \frac{-E_a}{R}}\left({\displaystyle \frac{1}{T}}-{\displaystyle \frac{1}{T_c}}\right)\right] f\left(ϵ\right)$$

(5)

To estimate parameters of Equation (5) through a MATLAB program, least square curve fitting was applied to the [$f\left(ϵ\right)$] expressions of the nuclei growth as well as the RN model from Table 1. The data generated by multiple repetitions of the experiments were highly reproducible, as evidenced by 1.1% standard deviation. A higher degree of freedom was ensured by conducting parametric analysis at a 95% confidence limit and using one thousand data points from each experiment to estimate the parameters. This indicates that a substantial amount of experimental data were utilized during the model parameter iteration process, resulting in precise and accurate model predictions.

Discrimination between various models, while implementing these models, was made by utilizing coefficient of correlation (R²), coefficient of cross-correlation (γ*) confidence intervals (with lower discretion), and minimum sum of squares. The R² values and parametric limits afforded better curve fitting of the AEM kinetic model when the value of $n$ was unity. The RN kinetic model, amongst the various models shown in Table 1, with a similar unity value of $n$, inherently assuming insignificant crystal size variation during repeated redox cycles, was particularly selected to manifest a hypothesis of approximately invariable crystal size. This indicates that during recurring redox cycles, the crystal growth is restricted [38]. The promising results of RN kinetic model as compared to the two-dimensional AEM kinetic model were associated with larger spans (increased from 2 to 5 to 10–20%) and the minimum R² and sum of square residual values for $n=2$.

The comparison of the RN kinetic model-based predicted values with experimental conversions is shown in Figure 4. The higher values of R² and residual normal distribution were found when experimental conversions were compared with model-based predicted conversions.

Table 2. Estimation of parameters for RN kinetic model demonstrating non-isothermal reduction.

Catalyst Sample	${\mathit{E}}_{\mathit{a}}$ (kJ/mol)	${\mathit{k}}_0$ × 103	γ*	R2
PRM	109 ± 0.12	480 ± 0.7	−0.85	0.999
10Ni-PRM	87 ± 0.19	910 ± 0.2	−0.63	0.998
15Ni-PRM	84 ± 0.14	730 ± 0.2	−0.74	0.999
20Ni-PRM	83 ± 0.13	670 ± 0.1	−0.77	0.999

Note: Repetition showed a standard deviation of ±2.2%; γ*: Cross-correlation coefficient.

summarizes the parametric estimations including the values of various parameters (with confidence interval >95%), the values of coefficient of determination (R²), and values of coefficients of cross-correlation (γ*). Table 2 also presents the values of activation energies of support as well as nickel-based catalysts and the data reveals that the support material PRM exhibited a higher value of activation energy, i.e., ~109 kJ/mol. The values of activation energies of nickel-based catalysts remained between 83 and 87 kJ/mol. The value of activation energy also facilitates classification of rates of reduction into three groups, as demonstrated by earlier reported works on reduction kinetics of metal-oxide [38,41]. For instance, the first group characterized by 15 to 25 kJ/mol values of activation energies highlights the role of external resistance or resistance caused by strong pore diffusion in controlling the overall reduction reaction. Pore diffusion also plays its part in the controlled or structured reduction regime with activation energies of 40 and 50 kJ/mol defining the second group. In addition to pore diffusion, chemical reactions also control the reduction regime that involves activation energy values from 65 to 135 kJ/mol and is included in the third group.

The activation energy values in Table 2 clearly illustrate that the support (PRM) and nickel-based catalysts, i.e., 10Ni-, 15Ni-, and 20Ni-PRM, are represented by the third group where reduction is controlled by the chemical reaction in agreement with the assumption made, i.e., mass transfer limitation is negligible.

The reduction activation energy data in Table 2 also indicate that nickel-based catalysts displayed reduction activation energy values (83 to 87 kJ/mol) much lower than the industrial waste-based support material, i.e., PRM (109 kJ/mol). The relatively lower reduction activation energy could be assigned to the uniform composition of PRM in all the catalysts emphasizing the nickel promoter impact after impregnation on PRM. These outcomes are in line with the increased uptake of hydrogen generally found in nickel-based supported catalysts during TPR analyses. An addition of cobalt to Ni, as demonstrated in a previous work, generated a noted improvement in the reduction in Ni species, facilitating improved Ni dispersion even in low surface area catalyst samples [42]. In a different study, incorporation of a second metal assisted in weaker metal-support interaction, resulting in the enhanced reducibility of mixed oxides emerging from two metals, and, in some cases, a second metal played was a barrier, averting the development of less reducible species of metal oxides [43].

The data in Table 2 reveal that the nickel-based catalyst with 20 wt% loading, i.e., 20Ni-PRM, showed the lowest activation energy, i.e., 83 kJ/mol, that was ascribed to fewer metal–metal local interactions and localized interactions between metal and nearby metal oxide species [42,44]. Extensive research has provided multiple explanations for the reduction behavior of Ni-based catalysts. These results align with previous research that investigated, via various characterizations, the role of the oxide nature of nickel oxide in alumina-supported Ni-based catalysts in affecting the reduction behavior of catalysts [45]. For instance, the presence of nickel oxide and nickel aluminate spinel (NiAl₂O₄) phases was identified by XRD. The reduction behavior study by H₂-TPR showed that these oxides/spinel species reduce in various ways, e.g., amorphous NiO freely available on the catalyst surface can be easily reduced at a reduction peak maxima of 330 °C, while a higher reduction temperature (537 °C) is needed to reduce spinel species (NiAl₂O₄). In a different work aimed at synthesizing Ni-based oxygen carriers supported on alumina for their application in chemical looping combustion (CLC) discovered that stable spinel species (NiAl₂O₄) formed over the catalyst surface resulting in a loss of oxide species leading to reduced conversion of reduction reaction [46].

2.2. Activity Performance

The activity tests are performed at 700 °C (1 atm pressure) using 90% methane diluted with argon at a gas hourly space velocity of 5000 mL/h/g_cat. Prior to the reaction, the catalysts were activated using pure hydrogen (20 mL/min) by raising the temperature from room temperature to 700 °C and keeping the catalyst at this temperature for 15 min. The conversion versus time-on-stream data in Figure 5 shows that PRM support and lower loading nickel-based catalysts (10Ni-PRM) deactivated immediately within one and a half hours after demonstrating an initial activity. However, higher nickel-loading catalysts showed varying trends with respect to activity, e.g., 15Ni-PRM deactivated during first 90 min before eventually remaining stable at ~8% conversion for 5 h time-on-stream. On the contrary, 20Ni-PRM exhibited stable performance, with ~75% methane conversion throughout the reaction duration of 5 h.

The catalytic activity performance results can be explained based on temperature-programmed reduction patterns, the size of active metal particles, and the dispersion of active metal nanoparticles. The metal–support interaction, degree of reduction, and/or reducibility play a critical role in directing the performance of the catalysts during the methane decomposition reaction [13,14]. The catalytic activity results (Figure 5) can be explored and elaborated using trends in reduction kinetics (Figure 4) as well as the amount of hydrogen consumed. The activation energy needed for reduction and/or activation of Ni-based PRM-supported catalysts is reduced when the amount of nickel is increased from 10 wt% to 20 wt%, as indicated by the data in Table 2. This reduced activation energy results in affecting the methane conversion or hydrogen production as shown in Figure 5. In fact, the number of active sites available for reaction are controlled by the reduction activation energy as analyzed in TPR kinetics. The impact of these available active sites associated with the reduction of activation energies is reflected in the catalytic activity results, where an increase in Ni content yielded a decrease in activation energy that in turn increased methane conversion and hydrogen production. For instance, 20Ni-PRM having an activation energy value of 83 kJ/mol did not only show significant improvement in methane conversion and hydrogen production rate but also remained stable during the course of reaction in comparison with 10Ni-PRM and 15Ni-PRM, since the latter catalysts deactivated over time. The role of nickel in improving the reducibility and/or easier reduction in metal oxides is fairly demonstrated by correspondingly enhanced catalytic performance, as reported in earlier works [13,14,47]. For example, Al-Fatesh et al. [48] evaluated the performance of m-modified (m = Ni, Co, and Mn within 3 to 10 wt%) 15 wt% iron-based magnesia (MgO)-supported catalysts for methane decomposition and discovered that Ni-modified (with only 3 wt% Ni) iron-based catalyst demonstrated significant improvement in the reducibility of Ni-free iron-based catalysts that also played a role in affecting the catalytic performance of the Ni-modified catalysts [48]. The shift in initial reduction peak, manifesting free (non-interacting) NiO species over the surface of the catalyst, from ~410 to 390 °C when Fe supported on MgO catalyst was modified with 3 wt% Ni indicated the role of Ni modification in the enhanced activity of catalysts even at a higher reduction temperature of 950 °C. In this work, the reduction profiles and amount of hydrogen consumed have demonstrated that Ni addition has facilitated the reduction in PRM, and additional hydrogen consumption could be associated with the reduction in surface NiO species having medium-to-strong interaction with the support. For instance, total hydrogen consumption of PRM is 2.82 mmol/g, including 2.18 mmol/g for first peak and 0.62 mmol/g for second peak. With Ni addition, two possibilities arise: (i) Ni facilitates hematite reduction as depicted by the first peak, (ii) the formation of spinel species NiFe₂O₄ occurs which is difficult to reduce and indicated by the second peak. Hence, hydrogen consumption associated with the first peak follows the trend 20Ni-PRM (3.10 mmol/g) > 15Ni-PRM (3.01 mmol/g) > 10Ni-PRM (2.24 mmol/g) > PRM (2.18 mmol/g). Moreover, the second peak assigned to either the reduction of iron oxide to metallic iron or a reduction in spinel species follows the trend PRM (0.64 mmol/g) > 10Ni-PRM (0.53 mmol/g) > 15Ni-PRM (0.32 mmol/g) > 20Ni-PRM (0.30 mmol/g). This trend suggests that Ni mainly contributed to facilitating the reduction in hematite that played a role during reaction in agreement with previous reports [13,14,47,48].

Generally, metal particle sintering and carbon encapsulation are major factors behind catalyst deactivation during methane decomposition. This suggests that both the sintering and the encapsulating carbon covering the active sites over PRM and PRM-supported nickel-based catalysts might have played a role in the quick deactivation of catalysts (10Ni-PRM and 15Ni-PRM) with otherwise high initial activity. However, the extent of sintering during synthesis might play a different role than metal particle agglomeration during reaction. For instance, we noticed from Figure 2 that the 20Ni-PRM catalyst demonstrated relatively larger particle size associated with sintering in comparison with the 15Ni-PRM catalyst. Moreover, crystallite size evaluated from XRD patterns for fresh catalysts showed 9, 11, and 13 nm for 10Ni-,15Ni-, and 20Ni-PRM, respectively, indicating agglomeration. However, 20Ni-PRM not only outperformed the rest of the catalysts, but it also remained stable for 5 h. TEM of the spent catalysts (Section 2.3) also demonstrated the metal particle size remained 6–35 and 10–50 nm for 15Ni-PRM and 20Ni-PRM, indicating that though particle size increased, sintering during reaction is the least contributing factor towards the performance of this catalyst. In the case of PRM and 10Ni-PRM, metal particle sintering along with carbon formation during reaction could also contribute to their deactivation. This can be established by characterizing the spent catalysts as discussed in Section 3.3.

A comparison of this work with earlier reported red mud catalysts investigated for methane decomposition reaction [9,30,31,32] is presented in

Table 3. Comparison of this work with earlier reported works.

Catalyst	Operating Conditions		Activity		Ref.
	Reaction Temperature (°C)	GHSV (L/h·gcat)	Max. Methane Conversion (%)	Max. Hydrogen Rate (µmol/g/s)
RM4	800	9	19.8	18	[30]
RM6			-	7.7
RM7			-	3.5
Geothite	800	9	-	5.7	[9]
RM7			-	5.7
DW1	750	12	16.8	11.5	[32]
DW3			17.6	12.0
DW6			18.7	12.7
Dry RM	800	4.8	21.2	5.8	[31]
MRM			25.0	6.8
20Ni-PRM	700	5	75 *	19.2	This work

* Catalysts demonstrated no deactivation over time.

and it is obvious from the activity results that nickel catalysts supported on red mud have outperformed earlier reported catalysts. Previous studies have focused on evaluating the performance of red mud without impregnating any metal and discussed various characteristics of red mud. For instance, Sushil et al. [9] studied the phase transformation of red mud during its exposure to methane. An in situ XRD study revealed that red mud samples witnessed phase transformation during temperature ramp under methane and these variations in the phases, e.g., of major component (hematite) to magnetite and formation of iron and iron carbide as well as graphite play a vital role in affecting hydrogen production rate (Table 3). They also observed the formation of carbon oxides as well as enhanced specific surface areas of spent catalysts associated with the formation of carbon nanospheres and multi-walled carbon nanotubes. However, the catalysts in this work did not show any significant formation of carbon oxides. In a similar work, the role of red mud samples’ composition and variation in alkali metal content during hydrogen production from methane discovered that the sample with high titania content had lower activity as compared to the red mud sample with similar iron content [30]. They also reported the formation of magnetic carbon, with the potential of wastewater treatment, as a co-product. The modification of red mud to synthesize a nano mesoporous sample was investigated for its application in methane decomposition reaction [31]. The modification process induced physical and chemical changes in the dry red mud sample. The resulting modified red mud sample showed a higher specific surface area (190.6 m²/g) and pore volume (0.39 cm³/g) that facilitated the higher methane conversion (~26%) of the modified red mud sample in comparison with the dry red mud sample (11.9%). Geng et al. [32] investigated the role of residual sodium oxide (Na₂O) in influencing the catalytic behavior of red mud samples during methane decomposition. The catalytic activity performance results showed that the amount and dispersion of residual sodium had an inhibitory impact on methane activation and maximal conversions. All the catalysts with the least amount of residual sodium went through a conversion maximum before they deactivated. In summary, all red mud-based samples were examined for various factors including the modification of red mud and the impact of phase transformation, composition, and sodium content; however, all red mud-based samples exhibited deactivation over time. On the contrary, nickel-based catalysts in this work have not only outperformed earlier reported works but also remained stable during the course of reaction.

2.3. Characterizations of Catalysts Post Reaction

A high resolution transition electron microscope (HRTEM) is used to identify the morphology of the carbon formed during the methane decomposition reaction. It can be seen from the HRTEM images (Figure 6) of 20Ni-PRM that nanotubes of carbon are formed. The nanotubes vary in length and diameter. Some of these carbon nanotubes are in the form of a chain-like structure while others are hollow tubes. Furthermore, metal particles are found to be on both the tip and base of the nanotubes. This suggests that carbon nanotubes are formed by following both the tip-growth as well as bas-growth mechanisms. It can also be seen that no impurities are found, reflecting the formation of pure carbon nanotubes over 20Ni-PRM.

TGA results from Figure 7 indicate no significant weight loss above 200 °C (a temperature up to which any weight loss is generally assigned to moisture removal) in the case of PRM and 10Ni-PRM spent samples. This shows that the quick deactivation of these samples is assigned to metal particles sintering rather than carbon deposition. In the case of 15Ni-PRM and 20Ni-PRM spent samples, a significant amount of weight loss as a function of carbon deposition is observed, suggesting the formation of carbon nanomaterials in the form of nanotubes, as evident from SEM and TEM.

3. Materials and Methods

3.1. Catalyst Preparation

The nickel (Ni)-based catalysts were synthesized by impregnating Ni on the pretreated red mud support (named as PRM), that largely comprises Fe₂O₃ (50%), Al₂O₃ (20%), SiO₂ (20%) and remaining CaO/Na₂O. Prior to Ni impregnation, the PRM was subjected to calcination at 800 °C which yielded the support with a specific surface area of 20.0 m²/g and a pore volume of 0.0540 cm³/g. During catalyst synthesis, the desired quantity of PRM was dispersed in 10 mL of deionized (DI) water under sonication for 20 min. A mixture of potassium hydroxide (KOH) and potassium carbonate (K₂CO₃) was separately made. Subsequently, the calculated amount of Ni precursor (Ni(NO₃)₂·6H₂O) was added to 6 g of already prepared PRM dispersed in DI water. Next, the mixture of KOH and K₂CO₃ was added. The resulting solution was subjected to heat treatment at 80 °C for 1 h under magnetic stirring. Then, the slurry was dried in an oven for another hour. Finally, the dried catalyst was calcined at 550 °C for 5 h. The rest of the catalysts were synthesized in a similar way with different amounts of Ni.

3.2. Catalyst Characterization

X-ray diffraction (XRD) patterns of the catalysts were carried out using a Rigaku (Sunnyvale, CA, USA) diffractometer having Cu Kα beam as the radiation source, and the machine was operated at 40 kV and 15 mA. Moreover, a scanning range with 2-θ values of 10–90° and a step size of 0.02 were used. The phases were identified using X’pert (version 5.3) software.

To evaluate the reducibility profiles and activation/reduction kinetics, temperature programmed reduction using hydrogen as a probe gas was conducted in Auto Chem II 2920 V 4.03 (Micromeritics Instrument Corporation, Norcross, GA, USA). Initially, nearly 70 mg of the fresh catalyst was subjected to pretreatment at 150 °C to eliminate any moisture/impurities from the catalyst’s surface and afterward cooled down to ambient temperature before the temperature was increased from room temperature to 900 °C at 10 °C/min under a continuous flow (30 mL/min) of 10% H₂/Ar gas.

A high resolution transmission electron microscope (HRTEM) was used to analyze the morphological nature of the carbon formed over the surface of spent catalysts via a JEOL JEM 2100F (Tokyo, Japan). The solid samples were dispersed in ethanol using a sonication bath and later the dispersed samples were added onto a carbon grid by pouring drops of sample. The carbon grids were air-dried overnight before placing them in HRTEM for analysis.

Thermo-gravimetric analysis (TGA), TA Instruments (New Castle, DE, USA) SDT Q600 analyzer, was utilized to identify and quantify the type and amount of coke deposited over the surface of spent catalysts. The catalyst samples (8–10 mg) were subjected to temperature ramp (20 °C/min) from ambient temperature to 800 °C under an oxidative atmosphere by flowing air at 50 mL/min. Finally, the data measuring weight loss with respect to temperature was recorded.

3.3. Catalyst Evaluation

The CMD reaction study was performed in a PID micro-activity fixed-bed reactor (I.D. = 9.1 mm) using 0.30 g of fresh catalyst under atmospheric pressure. A K-type thermocouple was utilized to monitor the reactor temperature at the center of the catalyst bed. The catalyst was reduced prior to the reaction by flowing pure hydrogen at 20 mL/min for 15 min at the reduction temperature of 700 °C with subsequent purging using argon for 15 min. Afterwards, feed gas containing 90 vol% of methane balanced with argon was flown through the reactor at 25 mL/min at a reaction temperature of 700 °C. The reactor outlet stream was analyzed using an Agilent (Santa Clara, CA, USA) gas chromatograph (GC-7890B) equipped with a flame ionization detector as well as a thermal conductivity detector. The expression for CH₄ conversion, hydrogen production rate, and carbon yield is given by the following equations:

$${\mathrm{C}\mathrm{H}}_4 \mathrm{c}\mathrm{o}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}={\displaystyle \frac{{\mathrm{C}\mathrm{H}}_{4,\mathrm{i}\mathrm{n}}-{\mathrm{C}\mathrm{H}}_{4,\mathrm{o}\mathrm{u}\mathrm{t}}}{{\mathrm{C}\mathrm{H}}_{4,\mathrm{i}\mathrm{n}}}}\times 100\%$$

(6)

$${\mathrm{H}}_2 \mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}={\displaystyle \frac{{\mathrm{F}}_{\mathrm{C}\mathrm{H}4}\times {\mathrm{X}}_{\mathrm{C}\mathrm{H}4}}{m}\to \mathrm{mol}/\mathrm{g}/\mathrm{s}}$$

(7)

$$\mathrm{C}\mathrm{a}\mathrm{r}\mathrm{b}\mathrm{o}\mathrm{n} \mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}={\displaystyle \frac{\mathrm{w}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t} \mathrm{o}\mathrm{f} \mathrm{c}\mathrm{a}\mathrm{r}\mathrm{b}\mathrm{o}\mathrm{n} \mathrm{f}\mathrm{o}\mathrm{r}\mathrm{m}\mathrm{e}\mathrm{d}}{\mathrm{w}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t} \mathrm{o}\mathrm{f} \mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{m}\mathrm{e}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{s}}}\times 100\%$$

(8)

where F_CH4 is molar flow rate of methane, X_CH4 is methane conversion, and m is catalyst mass.

4. Conclusions

This work investigates the role of novel industrial waste-based catalysts for their performance during pure hydrogen production from natural gas. The characterization results of fresh catalysts indicated that the presence of active metal nanoparticles in the form of hematite and NiO, as well as NiFe₂O₄, played a role during reaction. Moreover, the amount of Ni also contributed to establishing the stable catalyst when Ni was loaded up to 20 wt% (20Ni-PRM). The lower loaded catalysts, i.e., 10Ni-PRM and 15Ni-PRM, suffered deactivation associated with sintering and carbon deposition. The formation of high-purity carbon nanotubes was confirmed by HRTEM. This study highlights the utilization of industrial waste in producing pure hydrogen along with high-value carbon nanomaterials.