Synthesis of high-quality graphene sheets via decomposition of non-condensable gases from pyrolysis of polypropylene waste using unsupported Fe, Co, and Fe–Co catalysts

Producing high-quality graphene sheets from plastic waste is regarded as a significant economic and environmental challenge. In the present study, unsupported Fe, Co, and Fe–Co oxide catalysts were prepared by the combustion method and examined for the production of graphene via a dual-stage process using polypropylene (PP) waste as a source of carbon. The prepared catalysts and the as-produced graphene sheets were fully characterized by several techniques, including XRD, H2-TPR, FT-IR, FESEM, TEM, and Raman spectroscopy. XRD, TPR, and FT-IR analyses revealed the formation of high purity and crystallinity of Fe2O3 and Co3O4 nanoparticles as well as cobalt ferrite (CoFe2O4) species after calcining Fe, Co, and Fe–Co catalysts, respectively. The Fe–Co catalyst was completely changed into Fe–Co alloy after pre-reduction at 800 °C for 1 h. TEM and XRD results revealed the formation of multi-layered graphene sheets on the surface of all catalysts. Raman spectra of the as-deposited carbon showed the appearance of D, G, and 2D bands at 1350, 1580, and 2700 cm−1, respectively, confirming the formation of graphene sheets. Fe, Co, and Fe–Co catalysts produced quasi-identical graphene yields of 2.8, 3.04, and 2.17 gC/gcat, respectively. The graphene yield in terms of mass PP was found to be 9.3, 10.1, and 7.2 gC/100gPP with the same order of catalysts. Monometallic Fe and Co catalysts produced a mix of small and large-area graphene nanosheets, whereas the bimetallic Fe–Co catalyst yielded exclusively large-area graphene sheets with remarkable quality. The higher stability of Fe–Co alloy and its carbide phase during the growth reaction compared to the Fe and Co catalysts was the primary reason for the generation of extra-large graphene sheets with relatively low yield. In contrast, the segregation of some metallic Fe or Co particles through the growth time was responsible for the growth small-area graphene sheets.


Introduction
Nowadays, synthetic non-degradable polymers such as polyethylene (PE), polypropylene (PP), polyvinyl chloride (PVC), and polystyrene (PS) are becoming increasingly important in our daily lives, but they also cause numerous problems to human society [1][2][3][4]. PP is one of the most commonly used polymers in a wide range of industrial applications due to its high strength, heat resistance, and chemical resistance, thus generating massive amounts of waste [5,6]. The majority of this waste is either incinerated in power plants or dumped in landfills or oceans, leaving it with little or no value [5,7]. Landfilling consumes space, releases harmful gases such as CH 4 and CO 2 , and pollutes the groundwater [8]. Meanwhile, incineration emits CO 2 and toxic pollutants such as toxic dioxins, both of which are harmful to the environment [9,10]. Thereby, one of the most significant challenges facing scientists is the management of plastic waste and its conversion into value-added products, which is important both technologically and environmentally. The outstanding properties and vital applications of carbon nanomaterials (CNMs) such as carbon nanotubes (CNTs) and graphene strengthen plastic waste management technology, enabling a more productive value-added upcycling process [11,12].
At present, graphene and CNTs are commonly manufactured via the catalytic chemical vapor deposition (CCVD) technique using high-purity gases (e.g., CH 4 and C 2 H 2 ), which are expensive and explosive, limiting their advanced applications [12][13][14]. An alternative is to produce these valuable materials from plastic waste, which is plentiful, cheap, and easy to handle [14]. Currently, many researchers are focusing on converting various types of plastic waste into CNMs to reduce production costs while maintaining environmental sustainability. Based on the literature, a two-stage process is commonly used to produce CNMs from plastic waste [15][16][17][18]. In the two-stage process, the temperature in each stage can be easily and individually controlled, improving the interaction between pyrolytic gases and the catalyst, and facilitating the recovery of deposited carbon and catalyst [17]. Most importantly, pure hydrogen and syngas can be efficiently produced through this process as a valuable product along with the CNMs when using PP and PE wastes as carbon sources [19][20][21][22][23][24][25]. Besides, due to their high carbon content and the absence of inorganic components, PP and PE can be regarded as viable sources of high-purity solid CNMs [26].
Iron, cobalt, and nickel-based catalysts are well-reported to be the most effective catalysts for the CCVD synthesis of most CNMs, regardless of the type of carbon source [27][28][29][30][31]. Therefore, we will introduce in detail some relevant previous studies on the use of these transition metalbased catalysts for the production of CNMs from plastic waste via a two-stage process. Jia et al. [32] investigated the effect of metal-support interaction (MSI) on the activity of Ni catalysts supported by La, Mg, and Sr oxides during the conversion of plastic into CNTs. According to their findings, the Ni/La catalyst with a moderate MSI had the highest activity in terms of CNTs yield. The influence of decomposition temperature (600-800 °C) on the performance of Fe-Ni catalysts during the pyrolysis and decomposition of PP waste to CNTs was studied by Yao et al. [33]. According to their results, a temperature above 700 °C is beneficial for the production of CNTs with high quality and graphitization level. Previously, we investigated MgO-supported bimetallic Fe-Mo and Co-Mo catalysts with varying metal/Mo ratios for the production of CNMs via catalytic pyrolysis of PE waste [34,35]. According to the results, the change in the Fe/Mo and Co/Mo ratios had a significant impact on the yield and morphology of deposited carbon products. In another work, PP was thermally converted into CNTs via a two-stage process using La 2 O 3 -supported Ni and Ni-Cu catalysts [36]. The addition of Cu to the Ni/La 2 O 3 catalyst increased the catalytic decomposition activity, yielding a maximum carbon yield of 1458%.
As illustrated above, most studies focused on converting plastic waste into CNTs rather than graphene sheets (GS) via a two-stage process and transition metal-supported catalysts. However, few attempts have been reported to the selective synthesis of GS by pyrolysis-catalysis of plastic waste. For example, Ruan et al. used the CVD method to convert polystyrene waste into high-quality graphene using a Cu foil as a substrate [37]. Similarly, Sharma et al. [38] used the CVD process to create high-quality crystalline graphene from a mixture of PE (86%) and PS (14%) using Cu foil as a catalyst substrate. Gong et al. [39] investigated a simple method for producing a high yield of graphene flakes using PP waste as a carbon precursor in the presence of organomontmorillonite as a catalyst. In other studies, the same group prepared high surface area and high-quality GS by carbonizing real-world mixed waste plastics on organically modified montmorillonite and then activating them with KOH [40,41]. Another report used a solid-state chemical vapor deposition method to convert large amounts of plastic waste (PET, PVC, PE, PS, PP, and polymethyl methacrylate) to graphene foil with high electrical conductivity [42]. In the same trend, some studies for the production of graphene materials from plastic waste using a two-stage process have been reported [43][44][45][46].
As illustrated from the literature, earlier studies focused on the two-stage conversion of plastic waste into carbon nanotubes rather than graphene sheets. Furthermore, the use and comparison of unsupported Fe, Co, and Fe-Co catalysts as substrates for the CVD synthesis of graphene using PP waste as a carbon source have not been reported yet. Therefore, the present work was dedicated to synthesis high-quality graphene through the decomposition of non-condensable gases obtained from the pyrolysis of PP waste. In this regard, we prepared and evaluated unsupported Fe, Co, and Fe-Co catalysts for the first time in the current process. The activity of the catalysts towards the yield, quality, and morphology of graphene was widely discussed depending on the characterization of the prepared and reacted catalysts. Finally, we provided a simple growth mechanism for demonstrating the formation sequence of graphene sheets over the prepared catalysts and identifying the active centers responsible for its formation.

Catalyst preparation
The monometallic Fe and Co oxides as well as the bimetallic Fe-Co catalysts were prepared by the combustion method using citric acid as an ignition agent. In a typical preparation method, specific weights of metal nitrates and citric acid with a molar ratio of 1/0.75 were dissolved in 100 ml distilled water under continuous stirring to obtain a clear solution. For monometallic Fe and Co catalysts, about 13.46 g and 11.57 g of iron and cobalt nitrates were dissolved separately with the addition of 5.25 g and 6.26 g of citric acid, respectively. In this context, a mixture of 9.04, 6.51, and 7.05 g of iron nitrate, cobalt nitrate, and citric acid, respectively, were dissolved in distilled water to prepare the bimetallic Fe-Co catalyst. Each solution was transferred to a porcelain dish and thermally evaporated on a hot plate at ~ 90 °C with stirring to gain pasty-like material. The paste was then placed in a muffle furnace preheated to 600 °C for 3 h to produce stable metal oxide catalysts.

Graphene production
The fabrication of graphene sheets from PP waste was conducted by a dual-stage process using unsupported Fe, Co, and Fe-Co catalysts. The experimental setup consisted of vertical and horizontal quartz reactors used for the generation of non-condensable gases by PP pyrolysis and for the catalytic decomposition of pyrolysis products to form GNSs, respectively. The schematic diagram of the graphene production process has been thoroughly described in Fig. 1. The two reactors were heated by two separated furnaces. Before use, discarded PP drinking cups were washed and cut into small pieces. The discarded PP cups (15 g) were thermally pyrolyzed at 450 °C with a heating rate of 20 °C/min in a down-closed vertical quartz reactor (100 cm in length and 4 cm in outer diameter) and connected at the upper end by a long condenser (Fig. 1). The main function of the condensation system is to re-cracking the condensable vapors into non-condensable gases that are used as a graphene source. On the other hand, 0.5 g of the catalyst sample was uniformly distributed in the middle of the horizontal quartz reactor (100 cm in length and 2.5 cm in outer diameter) and reduced in situ at 800 °C under H 2 flow of 60 sccm for 1 h. Following that, the H 2 follow was stopped, and the temperature was raised to the growth temperature of 850 °C under N 2 flow of 50 sccm. At this moment, the pyrolysis of PP started, and the non-condensable gases left the vertical reactor to decompose on the catalyst surface within the horizontal reactor for 90 min to generate graphene sheets. Following the reaction, the system was cooled down to ambient temperature under a nitrogen flow of 100 sccm. The reproducibility test was carried out three times using the monometallic cobalt catalyst under the same operating conditions.
The weight difference between the pyrolysis reactor before and after the decomposition reaction determined the Fig. 1 Schematic diagram of the experimental system used for the production of graphene nanosheets from PP waste solid residue obtained from the pyrolysis process. Besides that, no liquid hydrocarbon products (condensable vapors) were detected as a result of the condensation system used in this study. Thus, the mass of the non-condensable gases was calculated by subtracting the mass of the residue from the initial mass of the PP waste. The non-condensable fraction was analyzed by gas chromatograph (Perkin Elmer Clarus 500, USA) using an Alumina BOND/Na 2 SO 4 capillary column (30 m long and 0.53 mm ID) and a flame ionization detector (FID), and the results are presented in Table 1 The graphene yield concerning the mass of catalyst, PP waste, and non-condensable gases were calculated using the following equations: where M NC , M cat , M pp , and M gases are the masses of net deposited graphene (without catalyst), catalyst before the reaction, PP waste, and non-condensable gas, respectively.

Characterization techniques
The X-ray diffraction (XRD) patterns of prepared catalysts and graphene were conducted by X'Pert PRO PANalytical apparatus using CuKa radiation. The measurements were determined using CuKa radiation (k = 0.1541 nm) within the 2θ range from 10° to 90° with a step size of 0.02°. The crystallite sizes of the materials were determined using the Debye-Scherrer equation (D = kλ/β cos θ), where D is the crystallite size, λ is the wavelength, β is the full width at half maximum (FWHM), and θ is the diffraction angle. The redox properties of the prepared catalysts were investigated by hydrogen temperature-programmed reduction (H 2 -TPR) using BELCAT-II equipment. Approximately 0.1 g of catalyst sample was placed in the quartz cell and pretreated for 1 h at 200 °C under He flow (50 sccm) before cooling to Graphene yield g C ∕100g gases = M NC ∕M gases × 100, 50 °C. TPR analysis was carried out by heating the sample from 100 °C to 1000 °C in 30 sccm of 5% H 2 /Ar mixture at a constant heating rate of 10 °C/min. Fourier transform infrared (FT-IR) analysis has been carried out in the transmission range of 400-400 cm −1 using the PerkinElmer spectrometer model 100 series. A field emission scanning electron microscope and energy-dispersive X-ray spectroscopy (EDX) (FESEM, ZEIZZ, Sigma 300 VP) was conducted to investigate the external morphology of the prepared catalysts. The graphene morphology was explored by transmission electron microscope (TEM, Model JEM-200CX, JEOL). A small amount of the sample was dispersed in ethanol, sonicated for 10 min, and then captured on a covered carbon grid. The quality and graphitization degree of graphene product was determined by Laser Raman spectra using SENTERRA Dispersive Raman Microscope (Bruker). The Raman spectra were recorded using diode Nd: YAG laser at a wavelength of 532 nm.

XRD analysis
The XRD analysis is used to investigate the phase composition and crystallinity as well as the purity of the prepared catalysts. Therefore, Fig. 2 presents the XRD patterns of  (Fig. 2). This indicates that the Fe oxide has fully interacted with the Co oxide to form the CoFe 2 O 4 spinel structure. Additionally, as compared to monometallic Fe and Co oxide catalysts, the relative intensity and sharpness of all diffraction peaks of the Fe-Co catalyst were dramatically reduced. As a result, we can confirm that the Fe-Co catalyst is primarily composed of the CoFe 2 O 4 phase, with a low amount of non-interacted Co 3 O 4 species present.
The average crystallite sizes of Fe, Co, and Fe-Co oxide catalysts were measured to be 37.6, 49.5, and 23.9 nm, respectively. The low crystallite size of the bimetallic Fe-Co catalyst in comparison to its individual components could be attributed to the facile formation of the CoFe 2 O 4 phase. This indicates that the combustion method facilitates the interaction between Fe and Co nitrates to generate their spinel ferrite structure.

H 2 -TPR analysis
H 2 -TPR study was performed to investigate the redox properties of both Fe and Co oxides, as well as their combined ferrite version, and the results are given in Fig. 3. As shown in Fig. 3, the H 2 -TPR profile of monometallic Fe oxide exhibits two reduction peaks throughout a temperature range of 300 °C to more than 900 °C. The first reduction peak, which occurs at a maximum temperature of 400 °C, is associated with the reduction of Fe 2 O 3 to magnetite (Fe 3 O 4 ). The second broad reduction peak, which extends from 468 °C to above 1000 °C, is principally due to the stepwise reduction of Fe 3 O 4 to wüstite (FeO) and finally to metallic iron. However, the extension of the second reduction peak to above 1000 °C indicates the incomplete reduction of bulk Fe oxide. On the other hand, the reduction of the Co oxide catalyst takes place in two stages, the first at 371 °C due to the reduction of Co 3 O 4 to CoO, and the second at 454 °C is rationally related to the reduction of CoO to metallic Co (Fig. 3) [49,50].
The TPR profile of Fe-Co catalyst displays three reduction peaks within the temperature range of 300 up to ~ 750 °C (Fig. 3). The weak reduction peak at 398 °C is mainly due to the first reduction step of non-interacted Co 3 O 4 species. This suggests the existence of minor non-interacted Co 3 O 4 species dispersed on the surface of the CoFe 2 O 4 spinel structure, which is consistent with the XRD analysis (Fig. 2). Meanwhile, the major reduction peak between 425 and 670 °C, with a maximum temperature of ~ 600 °C can be assigned to the gradual reduction of CoFe 2 O 4 species. This means that the peak encompasses the reduction of Co 2+ to metallic Co as well as the first reduction of Fe 3+ to Fe 2+ [51,52]. The last peak, around 695 °C, is primarily attributed to the reduction of Fe 2+ to metallic Fe species, and thus the Co-Fe alloy was readily formed (Fig. 3). According to the TPR results, the incorporation of Co oxide in the ferrite structure significantly facilitated the reduction of bulk Fe 2 O 3 , as reported in the literature [53]. This behavior can be attributable to the dissociation and activation of hydrogen molecules by the metallic Co, which promotes the reduction of Fe oxide to its metallic form via the spillover effect [52,54].
The H 2 uptake values were presented in the inset of Fig. 3. As depicted, the H 2 uptake values of Co and Fe-Co catalysts are nearly equal (~ 18 mol/g), indicating the complete reduction of these catalysts. In contrast, the H 2 uptake value of the monometallic Fe oxide catalyst was found to be low ( (Fig. 3 inset). This further confirms that the addition of Co 3 O 4 makes the reduction of Fe 2 O 3 easier.

FT-IR analysis
FT-IR analysis was conducted to explore the surface chemical composition of the prepared catalysts, and the resulting spectra are shown in Fig. 4. For all catalysts, the absorption bands in the range 700 and 400 cm −1 were attributed to the vibration of M-O bonds (metal-oxygen bonds) [55,56]. The weak bands at 3466 and 1645 cm −1 were assigned to the stretching and bending vibrations of adsorbed water molecules [57]. In addition, no bands corresponding to the organic bonds were detected in all catalysts, indicating the complete combustion of citrate precursor (Fig. 4). For Fe catalyst, the absorption bands at 541 and 470 cm −1 , were attributed to the stretching vibration of tetrahedral and octahedral Fe-O bond, respectively [55]. The FT-IR spectrum of the Co oxide catalyst reveals two strong resolved absorption bands at 665 and 573 cm −1 (Fig. 4), which correspond to the fingerprint of stretching vibrations of Co-O bonds in the Co 3 O 4 crystal [57,58]. This indicates that the preparation method and calcination temperature are appropriate for producing highly crystalline Co 3 O 4 nanoparticles, which agrees with the XRD results. On the other hand, the FT-IR spectrum of the Fe-Co catalyst shows also two bands at 665 and 573 cm −1 , which could be related to the stretching vibration of tetrahedral and octahedral M-O bonds (M = Fe or Co) in the CoFe 2 O 4 spinel structure (Fig. 4) [56,59].

SEM analysis
The exterior morphology of the freshly calcined Fe, Co, and Fe-Co catalysts was investigated using FESEM analysis, and the corresponding images are shown in Fig. 5. The flake-like structure with an extra-large size of up to several microns was observed for the Fe catalyst, indicating extreme agglomeration of Fe 2 O 3 particles, as shown in the low magnified image (Fig. 5a). Figure 5b confirms the presence of agglomerated particles. On the other hand, well-defined and nearly uniform particle sizes were observed for the monometallic Co catalyst (Fig. 5c,d). In addition, the particles appeared to be fine and near to the roundish shape with the presence of some agglomerated particles of varying sizes (Fig. 5d). For the bimetallic Fe-Co catalyst, the particles were aggregated to form a flake-like structure similar to the Fe catalyst with the appearance of several cavities, indicating the presence of porous material (Fig. 5e). The high magnification image showed that the CoFe 2 O 4 particles are strongly interweaving together to create a network with the presence of several narrow apertures (Fig. 5f). Undoubtedly, the absence of support materials is the primary cause of metal particle agglomeration, which is beneficial for the growth of graphene sheets.
The elemental composition of the Fe-Co catalyst and the establishment of CoFe 2 O 4 species were explored using EDX analysis, and the results are presented in Fig. 6. As shown in Fig. 6a, the significant peaks corresponding to Fe, Co, and O elements were observed in the EDX spectrum of the Fe-Co catalyst. The element contents by weight were determined to be 28.07, 31.53, and 40.4% for Fe, Co, and O, respectively, which agrees well with the atomic proportion in cobalt ferrite (Fig. 6a inset). Furthermore, the atomic weights (At%) of Fe and Co were nearly equal, indicating that the majority of Fe and Co oxides interacted extensively to form CoFe 2 O 4 species, as revealed by the XRD results (Fig. 2). Besides that, the elemental mapping revealed extensive overlap between all elements, indicating that CoFe 2 O 4 species were easily formed (Fig. 6b).

Characterization of the carbon product
The as-produced carbon materials obtained by thermal pyrolysis of PP waste over Fe, Co, and Fe-Co catalysts at 850 °C were characterized by XRD, TEM, and Raman spectroscopy to investigate their nature, microstructure, and morphology, as shown in the following discussions. Figure 7 displays the XRD results of as-deposited carbon (without purification) over Fe, Co, and Fe-Co catalysts. As shown in Fig. 7 Fig. 4 FT-IR spectra of the freshly calcined Fe, Co, and Fe-Co catalysts catalysts. This peak appears to be sharp, with the relative intensities of 100, 45.38, and 44.69 for Fe, Co, and Fe-Co catalysts, respectively (Table 2). Furthermore, the graphitic carbon d-spacing values in all catalysts were virtually identical (0.3377 nm, Table 2) and very close to the optimal space between graphene layers (0.3354 nm) [29]. This demonstrates that the prepared catalysts are capable of accumulating highly crystalline carbon materials with good quality and graphitization levels. The diffraction peak at 2θ = 44.52° was related to the metallic Fe phase (JCPDS No. 04-014-0258), indicating the Fe 2 O 3 catalyst undergone complete reduction during the CVD process. Also, the diffraction peaks corresponding to metallic Co nanoparticles were recorded at 2θ values of 43.68°, 50.97°, and 75.28° (JCPDS No. 04-020-5482) (Fig. 7). For bimetallic Fe-Co catalyst, the diffraction peaks at 2θ = 44.80°, 65.28°, and 82.66° were attributed to the Fe-Co alloy [60]. This behavior reveals that the CoFe 2 O 4 spinel structure was fully transformed into its alloy phase during the reduction process.

XRD analysis of deposited carbon material
It is worth noting that the crystallite sizes of graphitic carbon for Fe, Co, and Fe-Co catalysts were estimated to   14.45, 16.17, and 24.08 nm, respectively (Fig. 8). With the same catalysts order, the metallic phases had crystallite sizes of 37.9, 25.2, and 114.1 nm (Fig. 8). This suggests that the Fe-Co alloys were more stable during the growth reaction than the metallic Fe and Co phases, resulting in the formation of large-area graphene sheets, as we will see by the TEM results through the next part. The metallic Fe or Co nanoparticles or their carbide phases might be segregated into separated particles due to the quasi-liquid state nature at high temperatures. It is reported that the metal carbide phases are usually unstable at temperatures above 650 °C during the CVD synthesis of carbon nanomaterials [61,62].

TEM analysis of deposited carbon material
The XRD results demonstrated the successful growth of highly crystalline and high-quality carbon nanomaterials over Fe, Co, and Fe-Co catalysts at 850 °C using PP waste as a carbon source. Therefore, TEM analysis was carried out to investigate the type and morphology of as-grown carbon (without purification), and the resulting images are presented in Fig. 9. The general overview indicates the presence of graphene sheets with different sizes on all surfaces of the present catalysts. The deposited graphene sheets are semi-flat with the existence of some wrinkles and folds at the graphene edges. The active metallic or carbide particles appeared behind the graphene sheets, implying their high transparency, and the degree of transparency is affected by the thickness and percentage of stacked sheets. However, there were some differences in the deposited graphene morphology in terms of thickness, widths, and stacking extent depending on the type of catalyst used (Fig. 9). For the monometallic Fe catalyst, small graphene sheets or carbon nano-onions resembling quasi-spherical shapes were grown around the metallic Fe nanoparticles (Fig. 9a,b). Besides that, large-area graphene sheets with high thicknesses were also deposited on the Fe catalyst, as indicated in Fig. 9a,c. A Similar trend was observed for the deposited carbon on the metallic Co catalyst (Fig. 9d). Spherical graphene sheets accompanied by large-area graphene with relatively high thicknesses were grown on the surface of the Co catalyst (Fig. 9e,f). The presence of carbon nano-onion could be attributed to the fragmentation of some iron or cobalt carbide phases during the CVD process by the action of high reaction temperature. However, the size of fragmented carbide particles seems to be large enough to create carbon nano-onions. Meanwhile, the agglomeration of metallic Fe or Co particles due to the absence of support is the main reason for the growth of highly stacked graphene sheets with large diameters. This suggests that the particle sizes of Fe or Co catalysts are randomly changed during the reduction and reaction processes from small to extremely large particles due to the absence of structural supports. These results are well compatible with the XRD results shown in Fig. 8.
On the other hand, the Fe-Co catalyst exclusively produced extra-large graphene sheets with sizes reaching several hundred square nanometers. The as-produced graphene exhibited good quality, stability, and high transparency with the presence of a few folded and wrinkled parts (Fig. 9g-i). Moreover, the graphene sheets appeared to be partially detached from one another, generating semi-isolated sheets, as shown in Fig. 9h. Furthermore, the high transparency is a clear indication of the formation of few-layered graphene. Besides, the Fe-Co alloy particles have vanished completely, indicating that graphene sheets can easily detach from the catalyst surface. This behavior results in high-purity graphene products with relatively separated sheets, which is beneficial in advanced applications.

Raman spectroscopy analysis of deposited carbon material
Raman analysis was used to further evaluate the quality, crystallinity, and graphitization degree of the as-produced graphene sheets, and the Raman spectra and parameters are shown in Fig. 10. The characteristic D, G, and 2D bands of graphene product were detected in all spent catalysts at ~ 1350, 1580, and 2700 cm −1 , respectively. The D band is mainly related to the amorphous carbon or the structural imperfections of graphene sheets, whereas the G band is due to the sp 2 hybridized in-plane carbon-carbon stretching vibrations in graphitic sheets [63,64]. Meanwhile, the 2D is the second order of the D band and is associated with  [65,66].
Indeed, the presence of these bands, particularly the 2D band, is clear evidence of the formation of high-quality and crystallinity graphene sheets. The intensity ratios of I D /I G and I 2D /I G indicate the crystallinity degree and the number of graphene sheets, respectively. As shown in Fig. 10, the I D /I G ratios for all samples were less than one (0.20-0.41), indicating the formation of high crystallinity graphene sheets with rare defects and high purity. Moreover, the sharpness of the G band in all samples further reveals the presence of a well-crystallized sp 2 carbon network [60]. In this context, the I 2D /I G ratios for deposited graphene over Fe, Co, and Fe-Co catalysts were 0.60, 0.64, and 0.74, respectively (Fig. 10). This implies that the as-deposited graphene over the Fe and Co catalysts was multi-layered, whereas the Fe-Co catalyst was fewlayered. This also confirms that monometallic Fe and Co catalysts produced more stacked graphene sheets than the bimetallic Fe-Co catalyst. This behavior may be attributed to the remarkable stability of Fe-Co alloy and its carbide phase through the growth process at 850 °C, as revealed from XRD and TEM results. Generally, the bimetallic Fe-Co catalyst produced high-quality graphene sheets compared with the monometallic Fe and Co catalysts, as revealed by TEM and Raman results.

Growth mechanism
Based on the above discussions, we can predict the growth mechanism of graphene sheets over Fe, Co, and Fe-Co catalysts prepared via the combustion method, as illustrated in Fig. 11. After 1 h of thermal reduction under H 2 flow at 800 °C, both Fe 2 O 3 and Co 3 O 4 species can be reduced to their metallic phases (Fig. 11a). Following that, the metallic phases will be rapidly converted to metal carbides (Fe 3 C and Co 3 C) during the initial period of the decomposition reaction of non-condensable gases produced by the pyrolysis process. With the further decomposition of hydrocarbons, the deposited elemental carbon will then be rearranged and crystallized to generate graphene sheets of varying sizes. The Fe or Co carbides might be segregated into several particles with different sizes due to the impact of a high growth temperature of 850 °C (Fig. 11a). Even so, the sizes of the segregated metal carbide particles are still large enough to produce graphene sheets with different widths, as shown in the TEM images (Fig. 9a,d).
On the other hand, the bimetallic Fe-Co catalyst was transformed to the CoFe 2 O 4 spinel structure through the calcination step at 600 °C and then completely changed to Fe-Co alloy after the reduction step at 800 °C. The Fe-Co alloy can also be converted into carbide species during the initial stage of the reaction, followed by graphene growth (Fig. 11b). However, when compared to monometallic Fe and Co catalysts, the Fe-Co alloy appears to be more stable and resistant to segregation, resulting in the formation of extra-large graphene sheets (Fig. 9g-i).

Graphene yield
As shown in XRD, TEM, and Raman analyses, high-quality graphene sheets with different morphologies and crystallinity were successfully synthesized over Fe, Co, and Fe-Co catalysts using PP waste as a carbon source. Therefore, the graphene yield should be determined to evaluate the catalytic growth activity of the current catalysts.   Figure 12a displays the yield of pyrolysis products of PP waste at 450 °C for 90 min. As shown in Fig. 12a, the pyrolysis products were semisolid residue and non-condensable vapors with the absence of liquid hydrocarbons. For all catalysts, the yields of these products were nearly identical, measuring around 53-47% for solid residue and non-condensable gases, respectively (Fig. 12a). In this context, Fig. 12b displays the graphene yield obtained after 1.5 h of decomposing the non-condensable gases at 850 °C generated from pyrolysis of PP waste at 450 °C using Fe, Co, and Fe-Co catalysts. As shown, the monometallic Fe and Co catalysts yielded somewhat more graphene than the bimetallic Fe-Co catalyst. The amount of deposited graphene was 2.8, 3.04, and 2.17 g C /g cat using Fe, Co, and Fe-Co catalysts, respectively (Fig. 12b). Meanwhile, the graphene yield per mass of PP waste was calculated to be 9.3, 10.1, and 7.2 g C /100 g PP using Fe, Co, and Fe-Co catalysts, respectively (Fig. 12b). The graphene yield related to the mass of non-condensable gases which act the actual carbon source was calculated to be 19.7, 21.2 Fig. 12 (a) Pyrolysis product yield and (b) graphene yield obtained by pyrolysis of PP waste over Fe, Co, and Fe-Co catalysts. Reproducibility tests of Co catalyst (c) pyrolysis product yield and (d) graphene yield and 15.5 g C /100 g gases for Fe, Co, and Fe-Co catalysts, respectively (Fig. 12b).
As shown in Fig. 12b, the Co catalyst produced the highest graphene yield among the other catalysts. Therefore, the reproducibility testing of the current process was conducted individually using three batches of this catalyst under the same pyrolysis and growth conditions (Fig. 12c,d). The measurement results of all runs in terms of pyrolysis products were reasonably similar (Fig. 12c). The average yield of solid residue and non-condensable gases was calculated to be 53.32-46.68%, respectively. Correspondingly, the carbon yield associated with the masses of catalyst, PP, and noncondensable gases were almost similar (Fig. 12d). The difference between the maximum and minimum carbon yields was calculated to be 0.34 g C /g cat , 0.9 g C /100 g PP , and 1.48 g C /100 gases across all runs. The minor fluctuations in carbon yields could be attributed to a slight change in the weights of both fresh catalyst and PP waste. These findings demonstrated that the pyrolysis-catalysis process could be well reproduced using three batches of the catalyst.
The higher graphene yield over Fe and Co catalysts could be attributed to the splitting of metal carbide particles during the growth process, resulting in several particles with varying sizes that facilitate the deposition and growth of lowsized graphene sheets. On the other hand, the high stability of Fe-Co alloys and/or Fe-Co carbides against segregation during the growth period maintains their large size, leading to the formation of large-area graphene sheets. Indeed, the formation rate of low-size graphene sheets is rationally faster than that of large-area graphene. This means that the Fe-Co alloy takes longer to accumulate many graphene sheets than the monometallic Fe or Co catalysts. Finally, although the graphene yield from the bimetallic Fe-Co catalyst was lower than that from the monometallic Fe and Co catalysts, the quality, crystallinity, and graphitization degree were rather better.

Conclusion
The purpose of this study is to convert PP waste into highquality graphene via a dual-stage process using monometallic Fe and Co catalysts and their combined version (Fe-Co). The XRD and FT-IR results confirmed the formation of CoFe 2 O 4 after calcining the Fe-Co catalyst at 600 °C for 3 h. Besides, the H 2 -TPR analysis revealed the complete transformation of CoFe 2 O 4 into Fe-Co alloy through the thermal reduction process. It was found that high-quality few-layered graphene nanosheets with different sizes were deposited on the surface of all tested catalysts, as shown from TEM images and Raman spectra. The monometallic Fe and Co catalysts generated a mixture of small and large width graphene nanosheets, whereas the bimetallic Fe-Co catalyst only produced extra-large graphene sheets. In terms of graphene yield, the monometallic catalysts showed slightly higher catalytic growth activity than the bimetallic Fe-Co catalyst. On the other hand, in terms of graphene quality, the Fe-Co catalyst produced better graphene sheets with high graphitization than the other catalysts. The higher thermal stability of Fe-Co alloy and its carbide phase through the growth period was the main reason for the production of large-area graphene sheets with relatively low yield. In contrast, some of the Fe and Co carbide phases appeared to be separated into rather small particles, resulting in the formation of carbon nano-onion and large-sized graphene sheets.