2.4.4. TEM-EDX Analysis

• DR Reaction Sample

Examination of the carbon formed in TEM demonstrated that the carbon formed consists entirely of cylindrical and straight CNF. However, all CNF produced are of the fishbone type (graphene sheets at a certain angle relative to the hollow core fiber main axis). Romero et al. [29] have indicated in their article that Fe-based catalysts are responsible for the growth of two types of CNF, tubular (sheets parallel to fiber axis), and platelets; while Ni-based catalysts are responsible for the growth of fishbone type CNF only. However, in an earlier work of our research group [15] where steel was used as a dry reforming catalyst to produce CNF, it has been found that different forms of CNF were formed during the reactions, including fishbone ones. Yu et al. [32] have done a study in conditions similar to ours, which consists of decomposing a C2H4/H2/CO mixture with a Ni–Fe bimetallic catalyst and have found that the CNF formed during their study were of the fishbone type. They have reported that the fishbone is probably formed when a Boudourd reaction took place. In light of these results, we can deduce that the catalyst is not the only factor that influences the type of carbon formed; other factors, such as the type of gas supplied and the temperature, also have an influence as discussed further in Section 2.5.

When the graphitic sheets stack with one another, the angle formed between the sheets and the fiber axis was not always the same for all of the produced CNF. Different angles (11◦, 17◦, 23◦) are observed. It also seems that the diameter of the hollow core depends on this angle: the bigger the angle, the smaller the diameter (Figure 23). Since the catalyst's structure changes over TOS, it is rather impossible to control the width of the CNF as it is also discussed further on (Section 2.5).

**Figure 23.** TEM analysis of carbon deposited on Ni-UGSO 13% after the DR reaction at 650 ◦C for 2 h TOS, (**a**) CNF with d = 30 nm, (**b**) CNF with d = 36.7 nm, (**c**) CNF with d = 26 nm, (**d**) CNF with d = 53 nm.

The distance between the graphitic sheets is 0.340 nm (Figure 24), which is higher than the corresponding distance of graphite (0.335 nm). This means that CNF have structural defects and their structure is only ordered locally, not globally [29]. These defects are shown in as waved lines. The fact that the CNF are less ordered and have defects has been verified by TGA profiles (Figure 17), where the latter demonstrates that lower temperatures are necessary to oxidize CNF. Some zones that are darker than others can also be observed; they are due to layers not being stacked identically, and whose local density is different. The degree of graphitization (*g*) is calculated using this Equation: *dhkl* = 3.354 + 0.086 (1 − *g*), where dhkl is the interplanar distance [41]. Thus, the *g* -value of the CNF produced in this study is *g* = 46.5%. Romero et al. [29] have found that CNF produced from ethylene decomposition over a Fe-Ni-based catalyst have an interplanar distance of 3.42 Å, therefore *g* = 23%, which means that they are less graphitized than those obtained in this work.

**Figure 24.** Interplanar distance of graphene sheets.

SAED analysis confirmed that the interplanar distance of CNF is 0.340 nm. The second dhkl can be attributed to planes (102), (220), and (031) of Fe3C according to JCPDF File # 35-0772, or to planes (111) and (110) of Ni and Fe, respectively. From the presence of Fe3C, it can be deduced that CNF have grown on Fe. The existence of the Ni atoms in the metallic particles at the bottom of the CNF, as it is proven by SAED (Figure 25) and confirmed by the EDX analysis (Figure 21), confirms that the Ni has mainly participated in one of the stages of the growth of the CNF, which is the decomposition of the HC, while the iron is the main contributor in the second and third stages (dissolution and precipitation). In fact, it is known that Ni and Fe differ in their ability to decompose HC and solubilize carbon. Ni rapidly dehydrogenates the adsorbed HC while Fe is slower, and iron solubilizes carbon better than Ni. Indeed, the solubility of the carbon in the Ni in the range of temperature at which we worked is very low. Lander et al. [42] have experimentally developed an Equation that gives the solubility of carbon in nickel between 700 ◦C and 1300 ◦C, which is as follows: *lnS* = 2.48–4.880 *T* , where S is the solubility in grams of carbon per 100 gr of nickel and the temperature is in ◦C. The S value at 700 ◦C is relatively low (2.5%). This will also be discussed further in Section 2.5.

**Figure 25.** (**a**) Metallic particle at the tip of CNF. (**b**) SAED of this particle.

#### • CC Reaction Sample

For the CC reaction, we observed that there are different types of CNF, such as the tubular shape (layers are parallel to fiber axis like MWCNT) with a hollow core (Figure 26b,d), where we can also observe that some layers have torn ends. It seems that the sheets tended to connect to fill the inside of the CNF structure. Another type of structure is the bamboo type (Figure 26c). We were also able to observe that CNF formed with irregular stacking of graphene planes (Figure 26a), and we can observe that the graphene planes started out parallel to one another and that the angle of inclination with the fiber axis subsequently changed. The appearance of these CNF is quite similar to those formed by the decomposition of C2H4/H2 over Fe:Ni catalyst studied by Park and Baker [43].

**Figure 26.** TEM analysis of carbon deposited on Ni-UGSO 13% after CC reaction at 750 ◦C for 2 h TOS, (**a**) CNF formed with irregular stacked graphene planes, (**b**) and (**d**) tubular CNF with hollow core, (**c**) bamboo CNF.

In Figure 27, we observe that the metal particle is not on the tip of the filament contrary to what was found in DR, but it is encapsulated inside the filament. This could be explained by the fragmentation of the main particle, as its fragments could have been entrained within the body structure of the filament during the growth phase. The same behavior was found by Park and Baker [43] who worked in conditions similar to those used for this work (decomposition of C2H4 over Ni–Fe catalyst). The metal particle that was encapsulated by the CNF during the CC reaction appears to have a smooth globular morphology, in contrast to the structure of the catalyst particles at the end of the CNF produced in the DR reaction, which have more angular forms.

**Figure 27.** TEM analysis showing a catalyst particle inserted in two different nanofilaments.

The EDX analysis presented in Figure 28 shows Fe and Ni peaks in the pattern as well as C peak, which proves that the metallic particle is composed of both metals and encapsulated by carbon.

**Figure 28.** EDX analysis of carbon deposited on Ni-UGSO 13% after CC reaction at 750 ◦C for 2 h TOS.

#### *2.5. Mechanistic Understanding for the Growth of CNF*

The results of this study, which aimed to test the catalytic performance of a new catalyst derived from a mining residue (Ni-UGSO), have shown that Ni-UGSO is also a good catalyst for CNF production from ethylene cracking and dry reforming.

The influence of the catalyst composition as well as of the precursor gas composition and reaction conditions are discussed below.

#### 2.5.1. Influence of the Catalyst on the Growth of CNF

The effects of Ni and Fe on the growth of CNF is different depending on whether it is a CC reaction (decomposition of C2H4) or a DR reaction (decomposition of C2H4 and disproportionation of CO). Park et al. [43] have mentioned in their work that Ni-based catalysts are good for the decomposition of ethylene but were not as potent for catalyzing the Boudouard reaction, whereas Fe-based catalysts exhibited the opposite behavior. This fact was verified by our work, where we found that Ni is responsible for ethylene decomposition while Fe is responsible for the growth of CNF. They studied a bimetallic Ni–Fe catalyst for the decomposition of C2H4 and CO in the temperature range 600–725 ◦C, and they proved that increasing the ambient temperature improves the decomposition of C2H4 while the Boudouard reaction is favored thermodynamically by temperatures of around 550 ◦C.

It has also been found that the crystallographic orientation of the metal atoms plays an important role in the ability of the catalyst to decompose the reactive gases [44]. Zhu et al. [45] showed that the Ni (111) plane adsorbs ethylene and acetylene dissociatively, which was proven in this work by SAED results where the Ni (111) facet was found on the metal particle on the top of CNF (Figure 25). Their calculations show that the rate of di ffusion of carbon on the Ni (110) plane is the fastest step. However, the carbon deposited on the facet (110) is poorly crystallized because the distance of this plane does not correspond to that required for forming a graphite network. To form a good crystalline carbon structure, the atoms resulted from the decomposition of HC must first di ffuse through Ni to dissolve and then precipitate onto the adequate Fe (110) facet (Table 9), which is required to form a graphite network. The energy di fference between poor and well-crystallized carbon is the driving force that leads to transfer from one side of a metal to another [44].


**Table 9.** Indexation of D-spacing measured by SAED.

2.5.2. Influence of Catalyst Particle Sizes on the CNF Diameter

Rodriguez [46] has studied the interaction between a metallic surface and carbon. Figure 29 is a schematic representation of the forces involved in the interaction of a metal catalyst particle with a graphite support in the presence of a gaseous environment. The contact angle θ is determined by the surface energy of the graphite support (YSG), the surface energy of the metal (YMG), and the metal-graphite interfacial energy (YMS), and is expressed in terms of Young's Equation:

$$\mathbf{Y\_{SG}} = \mathbf{Y\_{MS}} + \mathbf{Y\_{MG}} \cos \theta$$

**Figure 29.** Interaction between surface metal and graphite [46].

It presents changes in the shape of the metal particles as a function of the catalyst wetting degree on the graphite:


Two forms of catalyst particles associated with nanotubular carbon products, which are clearly di fferent from one another, are commonly observed and presented in the literature [47]: one is conical and the other one is spherical. The conical particles are usually found at the end of the nanofilaments, as it was proven in this work (Figure 25), and the almost spherical particles are observed at the end of the nanotubes [48]. For conical particles, the adhesion exceeds the cohesion inside the particle, which leads the metal to spread on the graphite surface and, after precipitation, the carbon takes the form of piled up stacked cones from the particle, determining the shape of the particle's bottom (Figure 30). In addition, when weak forces occur between the metal and the graphite and the contact angle is <sup>&</sup>gt;90◦, nanotubes are formed (Figure 30).

**Figure 30.** Conical and spherical metal particles on the top of CNF and CNT, respectively.

A sequence of "stop-action" images [48] shows that after a few seconds of initial growth, the particle is pushed upward by the carbon flux and lengthens. As growth continues, the surface in contact with the carbon begins to tilt upward until it forms a conical or tear-shaped form, the tip of the cone being oriented toward the growing carbon nanostructure and pointing in the direction of carbon di ffusion. This observation leads to the conclusion that the commonly accepted belief that the catalyst particle determines the size and shape of the product is false. It is more likely the opposite [49].

#### 2.5.3. Influence of Gas Composition on CNF Growth

The type of CNF formed is di fferent for DR reactions and CC reactions; nevertheless, it also depends on other factors, namely the metal type and temperature. Luo et al. [50] have used Ni–La2O3 in a flow of CH4/N2, CO/N2, and CO2/CH4/N2, and they observed the production of both encapsulating carbon and CNF. The former was mostly formed in a CH4/N2 atmosphere whereas the latter was formed in a CO/N2 or CO2/CH4/N2 atmosphere. These results are in accordance with our findings. When using only C2H4, CNF with irregular forms as well as encapsulated carbon was formed; while, during DR reactions (where CO is present), only fishbone-type CNF were formed.

In fact, the composition of the gas a ffects the composition of the surface of metal particles because of the preferential segregation behavior of one of these components, which a ffects the arrangemen<sup>t</sup> of the atoms in the crystallographic face. This critical characteristic determines the mode of adsorption and decomposition of the reactive gas [51]. It has been found that when CO is present in the reactive gas, particles tended to have a faceted form, which leads to the formation of fishbone CNF [51].

## 2.5.4. CNF Precursor

As it appeared on the images of the TEM analysis (Figure 25), the metal is located in the tip of carbon nanofilaments, which indicates that carbon has grown in a crystallographic face of the metal. Several authors have tried to find which phase is the one responsible for CNF growth. First, Baker et al. [52] report an activation energy that suggests that carbon diffuses through the reduced metal and, therefore, they indicate that the reduced metal is the growth crystal. Subsequently, Oberlin et al. [53] studied CNF growth on iron. They used TEM to identify growth crystals and reported that cementite and alpha iron were the only ones present in their work, which led them to conclude that not only is the active metal responsible for the growth of CNF but that it also contributed to the formation of metal carbides. In other research, in order to confirm which solid phase of iron is the most catalytic for carbon formation, Sacco et al. [36] worked on phase diagrams. They experimented by heating iron foils under a stream of hydrogen at 900 K, then fed hydrocarbon gas mixtures of different compositions into the reactor for each experiment. They had a mass gain that corresponds to carbon formation only in the area where Fe3C is thermodynamically favored. Mass gain does not occur in α-Fe region, which proves that carbides, at least initially, are needed for carbon formation. In another study, it was shown that Fe3C supported on graphite and exposed to acetylene did not catalyze carbon formation [27]. There are two assumptions to explain this: Fe3C does not catalytically break up acetylene, or it is necessary to have a Fe3C/Fe dual phase metal interface to provide the solubility difference needed for carbon diffusion and thus the growth of the nanofilament. The results found in this work, which confirm the presence of the Fe3C peaks in the XRD pattern and SAED, confirm the assumption that the Fe3C is the responsible growth crystal for CNF.
