Effect of Different Catalyst Supports on the Quality, Yield and Morphology of Carbon Nanotubes Produced from Waste Polypropylene Plastics

The role of the effect of the support on the reactivity of heterogeneous catalysts cannot be over-emphasized. Therefore, the study documented in this article investigated the effect of different metal oxide supports (MgO, CaO and TiO2) and mixed oxide supports (CaTiO3) on the performance of a bimetallic NiMo catalyst prepared via the sol–gel method during the catalytic growth of carbon nanotubes (CNTs) from waste polypropylene (PP). Waste PP was pyrolyzed at 700 ◦C in a single-stage chemical vapor deposition reactor and off-gas was utilized in-situ as a cheap carbon feedstock for the growth of CNTs under similar conditions for all the prepared NiMo catalysts (supported and unsupported). The structures of the prepared catalysts and deposited carbon were extensively characterized using X-ray diffraction (XRD), temperature-programmed reduction (TPR), transmission electron microscopy (TEM), thermogravimetric analysis (TGA), etc. The catalytic performance of NiMo supported and unsupported catalysts was evaluated in terms of the yield, purity, and morphology of synthesized CNTs. The results revealed that the stabilizing role of supports is fundamental in preventing nanoparticle agglomeration and aggregation, thereby resulting in improved yield and quality of CNTs. Supported NiMo catalysts produced better aligned graphitic and high-quality CNTs. The NiMo/CaTiO3 catalyst produced the highest carbon of 40.0%, while unsupported NiMo produced low-quality CNTs with the lowest carbon yield of 18.4%. Therefore, the type of catalyst support and overall stability of catalytic materials play significant roles in the yield and quality of CNTs produced from waste PP.


Introduction
Polypropylene represents the second largest polymer market worldwide, accounting for more than 25% of the global polymer demand, on an estimated annual average growth rate of around 5.2% [1][2][3]. Recycling technologies for waste PP materials and products are usually low-grade since waste PP is limited to a small number of reprocessing and reuse cycles as these products lose some of their essential properties, such as strength and stability, after a number of recycling cycles; nevertheless, the carbon content remains intact [4]. From economic and environmental points of view, low-value waste PP has been widely studied as an inexpensive feedstock for high-value-added carbon nanotubes [5][6][7]. Recently, Cai et al. investigated and compared CNT production over an Fe/Al 2 O 3 catalyst during the catalytic pyrolysis of different types of plastic wastes comprising polypropylene, polyethylene (PE), i.e., low-density polyethylene (LDPE) and high-density polyethylene (HDPE), and omission could be attributed to the fact that different studies have employed different synthesis/reaction conditions. For instance, different authors have used different reaction times, concentrations or loadings, temperatures, gas (inert) environments, etc., during the conversion, thereby resulting in CNTs of different quality and yield. Additionally, different studies have employed different methods in the quantification of CNT yield and polymer converted to CNTs, etc. [22].
Motivated by the above statements, the effect of four different catalyst supports (metal oxides and mixed oxide materials: MgO, TiO 2 , CaO, and CaTiO 3 ) on the performance of a bimetallic NiMo catalyst during the synthesis of CNTs from waste PP was studied. The quality, yield, and morphology of PP waste-derived CNTs were explored to establish a relationship between the Ni-Mo catalyst, the different catalyst supports, and their corresponding catalytic activity. The catalytic performance of the NiMo catalyst, supported on different metal oxide and mixed oxide supports for CNT growth from waste PP, has not been explicitly investigated. This work aimed to systematically compare and present clear information on the effect of different catalyst supports, specifically on how it affects the yield, morphology, and crystallinity (quality) of CNTs produced from waste PP under similar growth conditions, using a single-stage CVD reactor. Therefore, the objective of the present work was to compare the activities of the NiMo parent catalyst supported on different supports for the synthesis of CNTs from waste PP plastics. The novelty of this study lies in its presentation of first-hand information on the performance of different supported NiMo catalysts towards the growth of waste PP-derived CNTs. The NiMo catalyst was chosen because optimum loading and composition have been carefully studied by other researchers [23,24]. All four supported NiMo catalysts under study were prepared using similar Ni, Mo, and support metal loadings.

Characterization of Catalysts
The XRD patterns of the Ni-Mo catalyst supported on CaO, TiO 2 , MgO, and CaTiO 3 and unsupported NiMo are comparatively shown in Figure 1. The respective phases of interest present in all the calcined catalysts are identified and labeled. There is a remarkable degree of similarity in the crystal position of some of the diffraction peaks of the catalysts under study. Characteristic diffraction peaks associated with cubic phase crystalline structures of NiO (PDF 00-432-9325) were detected at 2θ = 37.5 • , 43.0 • , 62.8 • , 75.4 • , and 79.2 • , which corresponded to the (111), (200), (2,2,0), (3,1,1), and (2,2,2) diffraction planes, respectively, for all the catalysts. For the NiMo/TiO 2 catalyst, well-resolved diffraction peaks at 2θ = 24.3 • , 33.7 • , 37.6 • , and 49.4 • were observed, which corresponded to the hexagonal crystalline structure of ilmenite (nickel titanate), NiTiO 3 (PDF 01-075-3757). Peaks associated with NiMoO4 species (PDF 01-086-0361) were observed for all catalysts under study. For the NiMo/CaTiO 3 catalyst, diffraction peaks of cubic perovskite-like Ca(TiO 3 ) structure phases overlapped with NiMoO 4 and NiO peaks, as observed in Figure 1E. Again, similar peaks were observed to coincide in the NiMo/MgO catalyst in Figure 1B. These overlaps could be attributed to the degree of dispersion of NiO and MoO 3 species on the surfaces of CaTiO 3 and MgO supports. The crystallite size of the NiO phase present in all prepared catalysts is shown in Table 1. NiMo/CaTiO 3 and NiMo/MgO catalysts displayed the highest metal crystallite-sized particles among all catalysts under study, probably due to the agglomeration of metal particles. It is also worth mentioning that a similar average crystallite size of the NiO phase (9.2-11.4 nm) was also reported for the NiMo-supported CaTiO 3 catalyst with analogously overlapped phases [25].  Nitrogen adsorption-desorption isotherms of NiMo supported and unsupported catalysts exhibited type IV isotherms with H3 hysteresis loops associated with mesoporous materials (2 nm < pore size < 50 nm); see Supplementary Materials Figure S1. The type-H3 hysteresis loop at high relative pressure P/Po, resulting from complete filling and emptying of mesopores by capillary condensation, indicated that all catalysts had slit-shaped mesoporous structures [26,27]. Apart from the NiMo/TiO 2 and NiMo/CaTiO 3 isotherms, other catalysts' isotherms show steep capillary condensation steps, indicating non-uniformity of their mesopores [28]. It could be seen that there was a sudden intersection of the desorption branch with the adsorption branch at a relative pressure of around 0.63 for the NiMo/MgO catalyst, which may be a result of pore blockage due to the different metal cations present in the bulk surfaces, resulting in pores being inaccessible [29,30]. The NiMo/CaTiO 3 catalyst showed hysteresis loop closure at P/Po = 0.88, indicating the presence of large mesopores. The surface area, pore size, and pore volume of all calcined catalysts under study are shown in Table 1. It could be observed that all catalysts displayed decreased BET specific surface area compared to the unsupported NiMo catalyst (72.5 m 2 g −1 ). The observed decrease in the BET surface area of the supported catalyst could be due to the blocking of mesopores arising from the introduction of more metal species from the support materials. A similar low-BET surface area of 23.68 m 2 g −1 was reported by [31] for a NiO/TiO 2 catalyst. Li et al. [28] also synthesized NiMo-supported TiO 2 catalysts promoted with potassium via the incipient wet impregnation technique and the authors reported a low BET surface area between 39.7 and 45.7 m 2 g −1 for their catalysts. In our previous study [25], NiMo-supported CaTiO 3 catalysts synthesized by the sol-gel method also showed a low BET surface area between 7.7 and 17.4 m 2 g −1 .
SEM images of all prepared supported and unsupported NiMo catalysts are shown in Figure 2. All catalysts showed either a dense platelet, flake-like, or foamy morphology. EDS analysis elucidated the elemental components of all catalysts under study, as shown in Table 2. Further insights into the morphology of the different NiMo-supported and unsupported catalysts were obtained using TEM analysis and the results are presented in Figure 3. Accordingly, unsupported NiMo particles showed agglomerated structures with an average particle size of 10-24 nm. Particles of NiMo/MgO and NiMo/TiO 2 are of spherical morphology with particle sizes of 10-26 nm and 8-22 nm, respectively. NiMo/CaO possessed well-dispersed, sponge-like particles with a particle size distribution of 6-16 nm, while the NiMo/CaTiO 3 catalyst showed fairly dispersed metal particles on the surfaces of the supports, with some patches of aggregated particles with sizes of 8-28 nm. The SAED pattern is shown in Figure 3F-J; all catalysts presented concentric circles of polycrystalline, nanostructured material for all NiMo catalysts under study. Diffraction rings are assigned to the characteristic hkl values or typical planes of Ni [32,33]. The obtained SAED data ( Figure 3F-J) correlate with the diffraction data obtained from XRD in Figure 1, which also reveal the existence of other phases in NiMo-supported and unsupported catalysts.   The H 2 -TPR profile of all NiMo catalysts studied is shown in Figure 4. All catalysts presented different reduction profiles, indicating differences in the interaction between the active phase and the different supports; nevertheless, all catalyst profiles exhibited a single, well-defined reduction signal below 750 • C. The reduction profile of the unsupported NiMo catalyst displayed a weak shoulder at 374 • C, indicating the reduction of Ni 2+ species to Ni 0 , and a major high-temperature reduction maximum peak at 659 • C. According to the literature [34][35][36], it is known that the reduction of Ni 2+ to its metallic stable state is a one-step reduction (i.e., Ni 2+ → Ni 0 ) at around 300-400 • C, while the reduction of MoO 3 species is more complex, Mo 6+ → Mo 4+ → Mo 0 , and as such requires higher reduction temperatures [37]. Therefore, the first shoulder peak could be attributed to the complete reduction of weakly bound nickel species in the NiMo and NiO phases from Ni 2+ to Ni 0 and could also be ascribed to the partial reduction of MoO 3 (Mo 6+ to Mo 4+ ). Li et al. [28] stated that the reduction of Mo 6+ to Mo 4+ occurs at temperatures below 450 • C. A welldefined high-temperature peak at 659 • C could be attributed to the complete reduction of the MoO 3 -that is, Mo 4+ to Mo 0 species. For NiMo-supported CaTiO 3 and MgO catalysts, the reduction temperatures for the major reduction peaks shifted to higher temperatures, as shown in Figure 4a,d; these results could be attributed to the increased strong interactions between the support and Ni-Mo oxide catalyst structure, resulting in overlapping phases. The strong interaction could be also due to the high structural stability constituted by both MgO and CaTiO 3 matrices; it is known that there is restricted reducibility of the NiMg complex [38], which is in agreement with the XRD result. Another notable and important observation was made in Figure 4b,c, for the NiMo/TiO 2 and NiMo/CaO catalysts, respectively. For the NiMo/TiO 2 catalyst, the well-resolved reduction peak at 577 • C is ascribed to the reduction of NiTiO 3 and a shoulder at 702 • C is also attributed to the complete reduction of metastable NiMoO 4 (MoO 3 ) phases, as observed in Figure 1 Figure 5 depicts TEM micrographs and outer diameter distribution histograms of all asproduced carbon nanomaterials. All deposited carbon on the surfaces of NiMo-supported and unsupported catalysts exhibited hollow filamentous nanostructures, suggesting multiwalled CNTs (MWCNTs). It could be observed that all as-grown MWCNTs displayed tip-growth mode, with the metal nanoparticle propagating at the tip of the tubes, except for MWCNTs deposited on the NiMo/CaTiO 3 catalyst-that is, (CNT NiMo/CaTiO3 )-which showed base growth mode, depicting a moderately strong interaction with the support, which agrees with the TPR result. Figure 5F-J depict the outer diameter distribution histograms of all nanotubes deposited over supported and unsupported NiMo catalysts. All MWCNTs obtained over supported NiMo catalysts exhibited nanotubes with almost near-uniform range diameters, except for CNT NiMo deposited on the surface of the unsupported NiMo catalyst, which displayed a wide average outer diameter of approximately 25.08 nm ± 7.538 nm, as shown in Table 3. Shen and Lua [40] and Ni et al. [41] observed that the size of Ni catalysts usually changes at temperatures above 500 • C, resulting in Ni catalysts being rapidly deactivated because of the thermal fluctuation during CNT production. Similarly, Zhang and co-workers [24] suggested that at high temperatures, originally small particle-sized Ni nanoparticles have a tendency to assemble and form large clusters of particles during CNT nucleation and growth. These large, clustered particles could result in the formation of CNTs with a large diameter range since, according to the literature, CNTs' diameter depends on the particle size of the active metal on the catalyst support [42,43]. Therefore, the non-existence of the stabilizing effect of the catalyst support in the unsupported NiMo catalyst could be the reason for the agglomeration of active metal particles during in-situ reduction and CNT synthesis, leading to the formation of CNTs with a larger diameter range [24,36]. As shown in Figure 5A,B after CNT NiMo/CaO and CNT NiMo/MgO were purified, the lengths of nanotubes were shortened: this is due to the acid treatment and purification process. Meanwhile, nanotubes deposited on the unsupported NiMo catalyst-that is, CNT NiMo -are bendy, with more encapsulated metal particles being observed. Therefore, the above results suggest that the morphology of MWCNTs strongly depends on the type of catalyst support.  A similar result was reported by Takenaka et al. [43], who studied the catalytic performance (activity, lifetime, and CNT yield) of Co-supported MgO, Al 2 O 3 , SiO 2 , and TiO 2 catalysts during methane decomposition at 500 • C. They suggested that the catalytic activity of Co-supported catalysts depended on the type of catalytic support. Wang and co-workers [44] also investigated the effect of different supports on the selectivity control towards single-wall carbon nanotubes' (SWCNTs) growth from carbonyl (CO), and they proposed that different supports either disrupt or enhance the synergistic effect between Co and Mo species, resulting in the formation of different Co phases, which were responsible for differences in their observed selectivity towards SWCNTs' growth. Furthermore, they indicated that the SiO 2 and MgO support improved the selectivity towards SWCNTs with a narrow diameter range, while Co-Mo supported on Al 2 O 3 and TiO 2 perturbs the interaction between Co and Mo species, resulting in the formation of predominantly MWCNTs; hence, they are not selective for SWCNTs' growth.

Effect of Different Supports on the Morphology of CNTs
XRD could be utilized to determine the morphology, degree of alignment, and structural features of CNTs. According to Cao et al. [45], a reduction in the peak intensity of (002), I (002) , describes higher nanotube alignment. In line with the XRD result in Figure 6, the intensities of (002) peak, I (002) , are shown in Table 3. It could be observed that CNT NiMo/CaTiO3 are more aligned, while CNT NiMo/MgO showed the least nanotube alignment compared to other as-synthesized CNTs deposited on the surfaces of all the catalysts under study. Similarly, I (002) could be used to relate the number of walls in CNTs, and the decreased intensity of the (002) peak is associated with thin-walled CNTs and vice versa [46,47]. Therefore, from the values obtained for I (002) , presented in Table 3, CNT NiMo/CaTiO3 had the smallest number of walls compared to CNT NiMo/MgO, which had the largest number of walls.

Effect of Different Supports on the Yield of CNTs
The carbon yield over each of the catalysts was obtained using Equation (1) and their corresponding results are presented in Table 3. It could be observed that deposited carbon (CNT NiMo ) on the surface of the unsupported NiMo catalyst demonstrated the lowest yield, while CNT NiMo/CaTiO3 deposited on the surface of the NiMo/CaTiO 3 catalyst displayed the highest yield. These results could be related to the reports by [43,44] in which the synergism between the Ni and Mo particles could be greatly impaired due to the nonstabilizing effect of the support. Hence, at high temperatures, large metal nanoparticles could be formed, which were reported to be inactive for CNT growth. Therefore, in this case, most of the aggregated active metals were not involved in the nucleation and growth of CNTs, resulting in a poor CNT yield.
The oxidation properties of nanotubes obtained on the surfaces of NiMo-supported and unsupported catalysts were examined by temperature-programmed oxidation (TPO) and are depicted in Figure 7. The more carbon content in the synthesized sample, the less metal content after oxidation and vice versa. The amount of metal catalyst remaining in each of the as-synthesized nanotubes deposited over unsupported and supported NiMo catalysts after oxidation was determined using TGA and results were obtained in the order CNT NiMo (58%), CNT NiMo/TiO2 (22%), CNT NiMo/CaTiO3 (15%), CNT NiMo/MgO (2%), and CNT NiMo/CaO (<0.5%). It could be observed that CNT NiMo displayed the highest amount of metal content present, suggesting that it had the lowest metal-assisted carbon growth, which could be the reason for its low CNT yield as shown in Table 3. This could also be attributed to the excessive agglomeration of active metal available for the nucleation and growth of CNTs, which could be also confirmed by the TEM result in Figure 5E. Conversely, CNT NiMo/CaO and CNT NiMo/MgO displayed the greatest weight loss during oxidation: this could suggest that acid purification is beneficial in removing all the metal catalyst present in as-produced carbon deposits [11] but also detrimental to the nanotube structures due to opening up of the tube tips, as observed in Figure 5A,B. The derivative TGA plot in Figure 7B showed that most of the MWCNTs begin to oxidize at temperatures above 450 • C. A similar finding was reported by Xu et al. [48]; they proposed that the oxidation of CNTs commenced at a higher temperature above 400 • C, while a lower temperature (200-400 • C) was related to the oxidation of amorphous carbon. The appearance of high-temperature shoulder peaks at~483 • C and~530 • C for CNT NiMo/MgO and CNT NiMo/CaTiO3 , respectively, could imply the decomposition of various species of carbon present (structured carbon species) or suggestive of the presence of two dissimilar lengths of CNTs in the samples [49,50]. This is in agreement with the XRD result presented in Figure 6 confirming the presence of other graphite phases.

Effect of Different Supports on the Purity and Graphitization Degree of CNTs
Raman analysis was performed to evaluate the purity (crystallinity), structure, and graphitization degree of MWCNTs deposited over supported and unsupported NiMo catalysts. Figure 7 displayed Raman spectra of as-synthesized MWCNTs. The D-band at around 1352 cm −1 is ascribed to the breathing vibrations of sp 3 hybridized carbon from defective or amorphous (non-crystalline) carbon in MWCNTs, while the G-band at around 1580 cm −1 is caused by the tangential vibrations of sp 2 hybridized carbon atoms of graphite layers. However, deviation in the G-band (dotted lines), as shown in Figure 8E, at around 1590.8 cm −1 could be attributed to a change in the structural nature of the nanotubes [51]; this is in good agreement with the XRD and TGA results. The shoulder on the right side of the G-band in Figure 8B could be attributed to the increase in the disordered graphite structure and/or variation in the pattern (transverse or longitudinal) mode in the nanotubes' direction [52]. Again, deconvolution was performed to reveal the position of hidden/overlapped peaks [53] (see the deconvoluted Raman spectra in Supplemenary Materials Figure S2). The 2D band is associated with two-phonon, secondorder scattering processes and is an indicator of the increasing defect density of CNTs [54]. The peak intensity ratio I G /I D is used to determine the graphitization degree of graphene planes [55], where a higher I G /I D ratio is an indicator of a higher degree of structural ordering and fewer defects inside the graphene sheet, while higher I D /I G suggests more defects in CNTs [24,56]. As shown in Figure 8 and Table 3, CNT NiMo/CaTiO3 possesses more graphitic CNTs with fewer defects and/or less amorphous carbon, while CNT NiMo had the most defective CNTs and more amorphous carbon within its graphite sheets-this result is consistent with the obtained TGA results.  The stability of carbon samples during oxidation is associated with the degree of crystallinity of CNTs and is linked to the existence of a purer tube shell and fewer defects at the ends of nanotubes. Generally, more thermally stable CNTs suggests higher structural ordering and a lower number of defects [12]. As could be observed from the derivative TGA plot in Figure 7B, the oxidation peaks of CNTs synthesized from supported NiMo catalysts shifted to higher temperatures, suggesting that the presence of the support in the NiMo parent catalyst resulted in increased thermal stability and graphitization degree of all as-produced MWCNTs. Hence, MWCNTs deposited over NiMo/CaTiO 3 (CNT NiMo/CaTiO3 ) showed the most thermally stable MWCNTs at 629 • C, while CNT NiMo , CNT NiMo/CaO , CNT NiMo/TiO2, and CNT NiMo/MgO displayed thermally stable MWCNTs at 603 • C, 601 • C, 606 • C, and 591 • C, respectively.
To further elucidate the phase structure and crystallinity of synthesized CNTs over supported and unsupported NiMo catalysts, XRD analysis was carried out as shown in Figure 6. Diffraction peaks of CNTs (PDF 00-058-1638) detected at 2θ = 26.28 • , which corresponds to the (002) diffraction plane, were obtained for all MWCNTs deposited on the surfaces of supported and unsupported NiMo catalysts. Other diffraction planes of CNTs were detected at 2θ = 43 • , 53.6 • , and 77.5 • , which corresponds to (100), (004), and (110) planes, respectively. From the XRD patterns, some other graphite peaks were obtained (PDF 00-041-1487, PDF 01-075-1621, PDF 00-056-0159, and PDF 01-073-598), most of which overlapped with CNT peaks. It is worth mentioning that for CNT NiMo/MgO , four (4) diffraction peaks typical of Mo 2 C (PDF 01-071-6028) were obtained, as shown in Figure 6D. Such peaks (Mo 2 C), as reported by many researchers [23,48,57], are instrumental in the growth of CNTs. The (100) peak, which reflects the existence of in-plane regularity and curvature within the graphene layers [58], was absent in CNT NiMo . This could be the reason for the shoulder in Figure 7B emanating from the different wave modes present within the graphene layers. The inter-planar distance, d (002) -that is, the distance between graphene layers-is used to relate the regularity between graphene layers and it is a measure of the graphitization degree of nanotubes [45]. The interlayer orbital interaction results in conductivity of the graphite and the presence of impurities located between graphene layers could result in structural roughness of CNTs [58]. All synthesized nanotubes have interlayer spacing between the graphene layers that is almost similar to the theoretical (0.34 nm) graphene layer spacing for MWCNTs [59], as shown in Table 3. As could be observed, CNT NiMo showed an interlayer distance d (002) value of 0.335 nm, which is typical of graphite, which indicates a related weak interaction between different tube (graphene) walls in MWCNTs [59,60]. CNT NiMo/CaTiO3 depicted a more ordered interlayer arrangement than CNT NiMo/CaO , CNT NiMo/MgO, and CNT NiMo/TiO2 . These results correlate with the TGA and Raman results.
The TEM images of the spent catalysts before purification are depicted in Figure 9, and their corresponding Ni crystallite sizes were obtained from XRD analysis. In addition, the average Ni particle size distributions obtained from TEM analysis using ImageJ software are given in Table 4. The XRD data showed that the Ni crystallite size in the unsupported NiMo catalyst was almost double (2 times) the crystallite size of NiO obtained for the calcined NiMo catalyst (see Table 1), suggesting that the Ni particles are thermally unstable during in-situ reduction and at CNT synthesis temperatures. These observations are supported by the report of Ni et al. [41], where Ni catalyst particle size was adversely affected at temperatures between 500 and 540 • C, resulting in low quality in the crystallinity of CNTs. For all supported NiMo catalysts, after in-situ reduction and subsequent CNT synthesis, there was no obvious change in Ni crystallite size or the average particle size distribution of the active phase. This indicates that Ni particles are thermally stable under reductive and synthesis temperatures due to the stabilizing effect of the support. Comparing peaks for residual catalysts in Figure 6, it could be observed that peaks in NiMo-supported catalysts clearly become significantly weaker, demonstrating that the degree of dispersion of the metal active sites is improved by reduction treatment, which resulted in decreased particle size in the catalyst and subsequent better activity.  Table 4. Ni crystallite size obtained from XRD and average Ni particle size distribution from TEM analysis of all spent catalysts after reduction and CNT synthesis.

Catalysts
Ni

Materials
Waste PP (e.g., food containers for household use) was obtained from the refuse deposition/collection center within the Doornfontein campus of the University of Johannesburg (Johannesburg, Gauteng, South Africa). Surface paint designs were cleaned with acetone in order to remove the painted design from the surfaces of obtained waste PP materials. Thereafter, waste PP was washed, dried, and cut into smaller sizes using a pair of scissors.

Catalyst Preparation
The sol-gel method was employed in the preparation of catalysts following the method reported by Bajad et al. [5], with some modifications. Metal loadings with molar ratios 5/0.1/1 were maintained for the Ni/Mo/M catalyst (where M is the corresponding metal content of metal oxide or mixed metal oxide support). The same ratio was maintained for all catalysts being investigated. Furthermore, for comparison, an unsup-ported bimetallic Ni-Mo catalyst was also prepared using a similar Ni/Mo metal loading with molar ratio 5/0.  4.0 g of anhydrous citric acid was added into the mixture and all content vigorously stirred using a magnetic stirrer at 60 • C for 6 h. The produced gel was dried overnight at 100 • C in an oven. Similar procedure and molar ratios were maintained for CaO-and TiO 2 -supported Ni-Mo catalysts. The CaTiO 3 -supported Ni-Mo catalyst was prepared analogously to the procedure described in our previous study [25]. Preparation protocol for all catalysts is shown in Table 5. All resultant oven-dried green products were ground to powder using an agate mortar and pestle, followed by calcination in air at 600 • C for 4 h to remove organic and inorganic matrix and obtain the desired constituent metal oxide system. Afterwards, calcined catalysts were sieved to particle sizes in the range of 50-180 microns and stored for further use. Obtained catalysts were denoted as NiMo/MgO, NiMo/CaO, NiMo/TiO 2 , and NiMo/CaTiO 3 .

Synthesis of CNTs
The production of CNTs from waste PP was carried out in a single-stage CVD reactor, as shown in Supplementary Materials Figure S1, according to the procedure already described in detail elsewhere [61]. Around 0.5 g of calcined catalyst was distributed evenly on a quartz boat placed at the center of the reactor. The furnace was ramped at 10 • C/min and was heated to a growth temperature of 700 • C. When the set temperature was reached, the catalyst was reduced in-situ for 1 h to obtain the catalyst in its metallic state under H 2 /Ar gas mixtures at gas flow of 110 mL/min. Afterwards, the boat containing around 1.0 g of waste PP as a carbon source was introduced some centimeters away (relatively low-temperature region) from the position of the boat containing the catalyst and the growth of carbon nanomaterial was allowed for 0.5 h. Thereafter, the setup was cooled down overnight under an argon atmosphere. Each experiment was repeated for accuracy of results and repeatability. The yield of CNT deposits on each of the catalysts was obtained based on the mass of feedstock (polymer used) using Equation (1), as described elsewhere [25]: CNTs Yield (%) = mass of carbon product − mass of catalyst used mass of polymer used × 100 % All grown CNTs were labeled according to the catalyst on which they were deposited, as shown in Table 3.

Characterization of Catalysts
Both surface and internal morphologies of catalysts were obtained using a VEGA 3 TESCAN scanning electron microscope (SEM) (TESCAN, Brno, Czech Republic) attached to an energy-dispersive X-ray spectroscope and JEM-2100 transmission electron microscope (TEM) (JEOL, Tokyo, Japan), respectively. Selected area electron diffraction (SAED) patterns of all catalysts were also acquired on TEM. Average particle size distributions of catalyst were obtained from the TEM images using ImageJ software.
Phase structures of all catalysts under study were obtained using the Rigaku Ultima 1V X-Ray Diffractometer (RIGAKU, Tokyo, Japan) operated at 40 kV and 30 mA using Cu Kα radiation (λ = 1.54 Å); samples were scanned at 2θ range of 10 • to 90 • . Peaks were identified and matched using PDXL software. N 2 physisorption analysis at −196 • C was carried out using Micromeritics TriStar II Plus (Micromeritics, Norcross, GA, USA) to check for the Brunauer-Emmett-Teller (BET) surface area and surface properties of all catalysts under study.
The reducibility of all calcined catalysts was determined using the H 2 -temperatureprogrammed reduction technique (H 2 -TPR). H 2 consumption by the metallic oxide part of all catalysts was tracked by a thermal conductivity detector (TCD) using an in-house built instrument. Precisely, 0.1 g of each catalyst sample was degassed under helium at 300 • C for 60 minutes and afterwards heated to 800 • C under 10 vol% H 2 /90 vol% Ar gas mixture for 2 h [62].

Characterization of CNTs
Microstructures of as-grown carbon on the surface of each catalyst were acquired using a JEM-2100 transmission electron microscope (TEM) (JEOL, Tokyo, Japan). Additionally, the outer diameter distribution of all obtained CNTs was obtained using ImageJ software.
Temperature-programmed oxidation (TPO) via TGA (thermogravimetric analysis) HITACHI STA-7200RV (HITACHI, Tokyo, Japan) was utilized to evaluate the thermal stability of all obtained carbon nanomaterials grown on the surfaces of each catalyst. During analysis, around 5.0 mg of each carbon sample was heated in air with a flow rate of 20 ml/min and ramped at 10 • C/min from 30 • C to 900 • C [15].
Raman spectroscopy of synthesized CNTs was carried out to evaluate their graphitic nature/quality; spectrograms were obtained at a Raman shift range of 1000 to 3000 cm −1 at 532 nm wavelength using a WITec focus innovations Raman Spectrometer (WITec, Ulm, Germany).
The crystallinity as well as morphology of all obtained carbon nanomaterials deposited on each catalyst were determined by XRD using the Rigaku Ultima 1V X-ray diffractometer (XRD) (RIGAKU, Tokyo, Japan) operated with Cu Kα (λ = 1.54 Å) radiation, in the 2θ range of 10 • to 90 • .

Conclusions
The effect of different catalyst supports in an NiMo catalyst during the production of CNTs has been demonstrated and reported in this article. All supported catalysts studied and reported displayed moderate metal-support interaction (i.e., the interaction was not too strong or too weak), attributable to the stabilizing effect of the supports, thereby enhancing the activities of the catalysts, as supported by the TPR and XRD results. Supports are instrumental to obtaining higher-yield CNTs. For example, the use of a supported NiMo/CaTiO 3 catalyst displayed the highest carbon yield of around 40.0% with goodquality and well-aligned CNTs, while unsupported NiMo displayed the lowest yield of 18.4%. The difference in the yield of deposited CNTs over supported and unsupported NiMo catalysts could be attributed to the large percentage of active metal nanoparticles that were not utilized in the nucleation and growth of CNTs in the unsupported NiMo catalyst, as showcased by the TPO results. The use of the NiMo/CaTiO 3 catalyst was more effective in the formation of higher-yield and better graphitized CNTs than using NiMo/TiO 2 , NiMo/CaO, and NiMo/MgO catalysts. The outstanding performance of the NiMo/CaTiO 3 catalyst utilized in growing CNTs during the decomposition of waste PP could be correlated to the prevalence of metastable phases present and the large, ordered pore sizes of the catalyst, which makes them less prone to catalyst deactivation, resulting in improved dispersion and activity compared to other catalysts. Therefore, the catalytic performance of the NiMo catalyst strongly depends on the type of support material employed during CNT growth from waste PP.