**1. Introduction**

Hydrogen is an energy vector and is mainly used in the synthesis of several chemicals such as methanol, ammonia, and liquid hydrocarbons via the Fischer–Tropsch process. Concerning carbon nanofilaments (CNF), several studies have shown that they have noteworthy properties, including high surface area, high mechanical resistance, and high electrical and thermal conductivities [1]. This is why research e fforts have focused on optimizing and controlling the formation of these types of carbon structures, instead of inhibiting their growth [2]. Although, usually, carbon formation on catalysts causes the deactivation of the catalyst [3], CNF are shown to grow in such a way that the catalytically active sites maintain their activity [4]. CNF properties make them a good substitute to high-cost materials used currently in various applications such as reinforcement of composites [5], manufacturing

of double-layer condensers [6], fabrication of anodes in lithium batteries [7,8], adsorption [9], support for catalysts [10], or catalysts themselves [11].

H2 is generated mainly from hydrocarbons via thermocatalytic processes such as steam reforming (SR), autothermal reforming (ATR), partial oxidation (POX), dry reforming (DR), and catalytic decomposition or cracking (CC) [12]. However, methane SR is the only industrial production technology used so far [13]. SR is an endothermic reaction and requires a high energy input. Temperatures as high as 950 ◦C and relatively high steam/C ratios are required to reach high H2 yields and avoid carbon formation and, consequently, premature catalyst deactivation. In the last decade, many researches have focused on DR that uses CO2 instead of H2O to produce not only H2 but also CNF. Hence, DR reaction has not only economic interests but also an environmental interest, which is the contribution on the sequestration of CO2—a greenhouse gas [14–17].

The feedstock composition, the choice of catalysts including the support and active metals, as well as the operating conditions, especially the temperature, are the main elements that have been largely studied to optimize H2 and CNF production [18]. Methane [19–21], n-octane [22], ethanol [15], and biogas [23] are the main reactants used. The use of pyrolytically-produced gases has rarely been cited in the current literature. Arena et al. [24] have developed an innovative process for mass production of multiwall carbon nanotubes (MWCNT) by pyrolysis of virgin or recycled polyolefins. Regarding CNF production, the literature is rather scarce. Svinterekos et al. [25] used lignin (a natural polymer found in plants) combined with recycled polyethylene terephthalate (PET) to make precursor fibers that are used for the electrospinning of CNF. The work presented here is part of a larger research endeavor aimed at the conversion of waste plastic streams into added plus-value products such as CNF. Since the gases produced by plastic pyrolysis are composed mainly of unsaturated hydrocarbons, the first step of this study is focused on using C2H4 as a surrogate molecule. The dry reforming reaction of ethylene is not ye<sup>t</sup> well reported in the literature; the products of this type of reaction are considerably dependent on the nature of the catalyst used. In the presence of a transition metal catalyst, carbon and synthesis gas are the products obtained from ethylene DR [26]. However, the Mn and Cr oxides convert ethylene to butadiene and propylene. For example, with a MnO/SiO2 catalyst at 850 ◦C, the products obtained are C4H6 with a selectivity of 25%, C3H6 with a selectivity of 18%, and traces of CH4, C3H8, and C4H8 [26].

The theoretical reaction of ethylene DR is given by Equation (1) below [26]:

$$\text{CH}\_2\text{H}\_4 + 2\text{CO}\_2 \rightarrow 4\text{CO} + 2\text{H}\_2\text{ (}\Delta\text{H}^\circ\text{}\_{298} = 292.5\text{ MJ/kmol}\text{)}\tag{1}$$

Other known reactions that take place during ethylene DR are Equations (2)–(5):

$$\text{Ethylene decomposition: }\mathrm{C}\_{2}\mathrm{H}\_{4} \rightarrow 2\mathrm{C} + 2\mathrm{H}\_{2}\text{ (}\Delta\mathrm{H}^{\circ}\mathrm{}\_{298} = -52.5 \,\mathrm{M}\text{)/kmol} \tag{2}$$

Reverse water gas shift reaction (RWGS): CO2 + H2 - CO + H2O ( Δ H◦298 = 41.0 MJ/kmol) (3)

$$\text{Boundaryard reaction: } 2\text{CO} \rightleftharpoons \text{CO}\_2 + \text{C} \text{ (}\Delta\text{H}^{\circ}\text{}\_{298} = -172.0 \text{ MJ/kmol} \text{)} \tag{4}$$

$$\text{Carbon magnification: C} + \text{H}\_2\text{O} \rightarrow \text{CO} + \text{H}\_2 \text{ (}\Delta\text{H}^\circ\text{}\_{298} = 131.3 \text{ MJ/kmol}\text{)}\tag{5}$$

In general, at the temperatures used for these reactions and in the presence of the chosen catalysts, hydrocarbon molecules (HC) are converted into free radicals in the gas phase or at the catalyst's surface (intermediates). The reforming agent, CO2 in the case of DR, is also dissociated into oxygen intermediates (O \*) and CO. Oxygen-containing intermediates oxidize HC intermediates to produce CO and subsequently produce CO2 and carbon via the Boudouard reaction (Equation (4)). The atomic carbon formed during the Boudouard reaction first di ffuses and dissolves into the metal particles until saturation is reached and then the graphitic carbon starts precipitating to form CNF [27]. For the CC reaction, HC intermediates self-decompose to produce H2 and carbon [28] and the atomic carbon formed follows the same process of di ffusion, saturation, and finally precipitation [29].

Transition metal-based catalysts, particularly iron and nickel, are recognized for their ability to decompose carbonaceous gases into filamentous carbon and hydrogen. This capacity for carbon formation is due to the high di ffusion rate of carbon in these metals at high temperatures.

The coe fficients of di ffusion of carbon into transition metals at 550 ◦C are 1.2 × <sup>10</sup>−7, 0.8 × <sup>10</sup>−7, and 0.2 × 10−<sup>7</sup> cm<sup>2</sup>/s, for Ni, Fe, and Co, respectively [29]. Consequently, the carbon yield would increase as follows: Ni-based catalyst > Fe-based catalyst > Co-based catalyst. This order, however, was not confirmed by Romero et al. [29], who studied the influence of these active metals (Co, Ni, Fe) and the influence of the zeolite type support on the synthesis of highly graphitized carbon nanofibers produced from the catalytic decomposition of ethylene. They found that the order is rather Ni > Co > Fe. They a ffirmed that this di fference is due to the zeolite support that has a di fferent synergistic e ffect, which explains the important role played by the support, on the activity of the catalyst.

Recently, our research group (GRTP-C & P) collaborated with Rio Tinto Iron and Titanium (RTIT) for the valorization of a mining residue (upgraded slag oxide (UGSO)) of the upgraded slag (UGS) process to produce titanium slag from ilmenite. Since UGSO is largely composed of iron oxides, in addition to Mg and Al oxides, it has been used to produce an e ffective Ni-functionalized spinel catalyst tested in methane DR, methane mixed reforming [30], and pyrolytic oils SR [31]. In this work, we investigate the e fficiency of this new catalyst in ethylene DR and CC reactions to produce H2 and CNF.

#### **2. Results and Discussion**

#### *2.1. Fresh Catalyst Characterization*

Table 1 illustrates the BET surface area, average pore volume, and average pore diameter for UGSO and Ni-UGSO with di fferent Ni wt.%. We can observe that Ni-UGSO has a smaller BET surface area, smaller pore volume, and smaller pore diameter than UGSO; this is due to the formation of other phases (spinels) as shown on the XRD pattern (Figure 1) that cause a rearrangemen<sup>t</sup> of UGSO structure (the signification of each symbol in XRD patterns are presented in Table 2). Regarding the e ffect of Ni wt.%, no statistically significant change was found. Generally, Ni addition leads to the reduction of specific surface area, pore volume, and pore size of the catalyst.


**Table 1.** Textural properties of Ni-UGSO with di fferent Ni contents (5, 10, and 13 wt.%).

a Pore volume was obtained from P/P0 = 0.97. b Pore diameter was obtained from Barret–Joyner–Halenda (BJH) desorption method. c Ni crystallite size was calculated from Scherrer Equation.


**Table 2.** XRD phase legend.

**Figure 1.** XRD analysis of Ni-UGSO with different Ni contents (0, 5, 10, and 13 wt.%).

We can also observe that the NiFe2O4 crystal sizes are nanometric and there is no difference in the crystal size in function of wt.% of Ni on the catalyst.

XRD patterns of fresh catalysts with different wt.% of Ni are shown in Figure 1, which shows that the patterns of the three catalysts are identical. The same family phases have been detected whatever the Ni percentage. In summary, the catalysts are mainly composed of two phases: spinels, in the most probable order of formation (figure of merit (FOM) smallest); MgFeAlO4, MgFe2O4, Fe3O4, NiFeAlO4, AlFe2O4, NiFe2O4; and monoxides (NiO, MgO), which coexist in their solid solution. When comparing to the pattern of fresh UGSO, the new crystalline phases are a clear indication that the Ni has been well integrated into the structure of the UGSO.

According to the TEM images (Figure 2a,c), the catalyst particles are faceted and have a size distribution ranging between 70 nm and 355 nm.

SAED patterns (Figure 2b,d) indicate that the catalyst is composed of the spinels NiFe2O4 and Fe3O4, and oxides NiO and (MgFe)O (Table 3). These results corroborate the XRD analysis results.


**Table 3.** Indexation of d-spacing measured by SAED.

**Figure 2.** TEM analysis of Ni-UGSO 13% (**<sup>a</sup>**,**<sup>c</sup>**) and its corresponding selected area electron diffraction (SAED) (**b**,**d**).

#### *2.2. Catalyst Activation and Characterization before DR and CC Reactions*

Before the DR and CC reactions, Ni-UGSO was activated by H2. Concerning structural properties, we notice that the activation has increased the BET surface area, pore volume, and pore diameter (Table 4). The effect of the activation is the reduction of metal oxides into metal particles as we can see in XRD pattern (Figure 3), especially into Ni and Fe metal and their alloys, which leads to a pore enlargement and a BET increase due to the nanometric size of the metallic species proved by TEM (Figure 4).


a Pore volume was obtained from P/P0 = 0.97. b Pore diameter was obtained from Barret–Joyner–Halenda (BJH) desorption method. c Ni crystallite size was calculated from Scherrer Equation.

**Figure 3.** XRD analysis of Ni-UGSO 13% before and after activation.

**Figure 4.** TEM analysis of activated Ni-UGSO 13%, (**a**) support particle size and (**b**) crystallite size.

When comparing the XRD patterns of the catalyst structure before and after activation by H2, we can observe the appearance of peaks attributed to the metallic phases Ni, Fe, and their alloys. Yu et al. [32] have shown that the reduction of catalysts containing Ni and Fe leads to the formation of their alloys such as tarnite and kamacite, and the proportion of Ni:Fe on the alloy after reduction depends on their initial mass ratio. We can also observe the presence of FeO, which means that the magnetite has been reduced partially into wüstite and iron.

TPR analysis was used to determine the reduction temperatures of the different metal species present in the catalyst. The TPR profiles for the three different catalysts with different wt.% (Figure 5) have the same shape with three distinctive peaks. The difference is in the amount of H2 consumed, which, as expected, increases with the wt.% of Ni in the catalyst.

**Figure 5.** TPR analysis for Ni-UGSO with different Ni contents (5, 10, and 13 wt.%).

In fact, the reduction temperature of metals depends on their interaction with the support and their location, as well as the structure to which it belongs (oxide or spinel). The reduction of Ni-UGSO by H2 essentially leads to the formation of metallic Fe and Ni particles in addition to their solid solution. Al and Mg are resistant to reduction and remain in their oxidized state.

The analysis of the Ni-UGSO 13% TPR pattern depicts the main reduction peak (the one in the middle) and two others. The first peak can be attributed to the reduction of free NiO (not in interaction with all other phases) and the reduction of Fe3<sup>+</sup> to Fe2+. The second peak can be assigned to the reduction of both Fe3<sup>+</sup> species to Fe2<sup>+</sup> and Fe, and Ni2<sup>+</sup> to Ni (NiO moderately interacting with other phases). The third peak can be attributed to the reduction of NiO strongly interacting with MgO or having a strong interaction with spinel MgFeAlO4 [30].

As shown in Figure 4, the reduction of the catalyst by H2 led to the formation of metal crystallites with different sizes on the surface of crystals that have not been reduced (Al and Mg oxides). Similar results were found by Romero et al. [29], who studied the reduction of zeolite-supported Ni- and Fe-based catalysts. They observed Fe and Ni crystallites with different distributions formed on the surface of the zeolite. In fact, the activation of catalysts by H2 led to the reduction of oxides into small metallic particles, which are the active phase for the growth of CNF.

As depicted in Figure 4b, the crystallite sizes are in the range of 10–30 nm. Yu et al. [32] have found that the reduction of the Ni:Fe (6:1) catalyst has an average crystallite size of 6 nm with a Gaussian-like distribution. The difference observed when comparing our results with those in the literature might be due to a different Ni:Fe ratio and/or to the reduction conditions (nature of the substrate and rate of heat and mass transfer). Some sintering seems to have taken place due to reduction because, if we compare Figures 4a and 2, we notice that the support particle size has increased (100–400 nm).

The EDX pattern (Figure 6) shows that these crystallites are composed mainly of Ni and Fe and no O has been detected. This proves that these crystallites are metallic and they are Ni, Fe, and/or Ni–Fe alloys, thus corroborating the already presented XRD results.

**Figure 6.** EDX analysis of activated Ni-UGSO 13%.

#### *2.3. Ni-UGSO Catalyst Performance*
