A novel one-step synthesis for carbon-based nanomaterials from polyethylene terephthalate (PET) bottles waste

ABSTRACT

Nowadays our planet suffers from an accumulation of plastic products that have the potential to cause great harm to the environment in the form of air, water, and land pollution. Plastic water bottles have become a great problem in the environment because of the large numbers consumed throughout the world. Certain types of plastic bottles can be recycled but most of them are not. This paper describes an economical solvent-free process that converts polyethylene terephthalate (PET) bottles waste into carbon nanostructure materials via thermal dissociation in a closed system under autogenic pressure together with additives and/or catalyst, which can act as cluster nuclei for carbon nanostructure materials such as fullerenes and carbon nanotubes. This research succeeded in producing and controlling the microstructure of various forms of carbon nanoparticles from the PET waste by optimizing the preparation parameters in terms of time, additives, and amounts of catalyst.

Implications: Plastic water bottles are becoming a growing segment of the municipal solid waste stream in the world; some are recycled but many are left in landfill sites. Recycling PET bottles waste can positively impact the environment in several ways: for instance, reduced waste, resource conservation, energy conservation, reduced greenhouse gas emissions, and decreasing the amount of pollution in air and water sources. The main novelty of the present work is based on the acquisition of high-value carbon-based nanomaterials from PET waste by a simple solvent-free chemical technique. Thus, the prepared materials are considered to be promising, cheap, eco-friendly materials that may find use in different applications.

Introduction

The production, consumption, and waste generation rate of plastic solid waste have increased considerably since the first industrial-scale production of plastics took place in the 1940s. Particularly, polyethylene terephthalate (PET) consumption has recorded the fastest growth rate in the global plastic market due to the ongoing expansion of the PET bottle market. Thus, many researchers have placed a focus on plastic solid waste recycling in the past few decades, in order to develop innovative technologies for the conversion of these huge amounts of waste into new useful and value-added products (Zhuo and Levendis, 2014; Adibfar et al., 2014) instead of waste incineration and chemical recycling, which are the most common ways to deal with PET disposal problems (Mishra et al., 2003). The conversion of PET waste into valuable carbonaceous materials may be a notable way (Berkmansa et al., 2014; Esfandiari et al., 2012) due to PET wastes having high carbon content and low amounts of mineral matters and impurities. In addition, recycling PET waste is increasingly demanded for both environmental and technological reasons (Pol and Thiyagarajan, 2010, Parra et al., 2004).

Carbon is one of the most abundant elements found on Earth. It has unique properties due to its hybrid orbitals. The hybridized bonds formed by combining atomic orbitals s and p into new hybrid orbitals as sp, sp², and sp³ lead to many carbon allotropes, ranging from zero to three dimensions of carbon: for instance, zero-dimensional (0D) as buckyballs, one-dimensional (1D) as carbon nanotubes, two-dimensional (2D) as graphene sheets, and three-dimensional (3D) as diamond. All these make carbon a highly investigated material for a wide range of nanotechnology applications (Guisinger and Arnold, 2010). On the nanoscale, materials behave differently from their bulk counterparts, due to the effects of quantum confinement (Jin et al., 2013). From the perspective of materials science, these materials have fascinating chemical, physical, and electronic properties that have attracted scientists over the world to investigate advanced technological applications (Laszlo et al., 1999; Liu et al., 2015; Pol et al., 2009). As a result, the outcomes are materials known as nanoparticals (NPs). NPs include fullerenes, carbon tubes and nanoshells, and quantum dots (QDs). Carbon materials are known to have a wide range of structural and textural properties, and thus extensive applications (Hwang et al., 2013; Jin et al., 2013; Zeng et al., 2006). Among of the allotropes of carbon, fullerenes were one of the first nanoparticles discovered. Their most prominent representative is the C60 molecule, which has the form of a football; it consists of a graphene sheet, where some hexagons are replaced by pentagons, which cause a crumbling of the sheet and the final formation of a graphene sphere. This discovery was made in 1985 by a trio of researchers at Rice University, Richard Smalley, Harry Kroto, and Robert Curl, while its existence had been predicted before, in 1970, by the Japanese theoretician Eiji Osawa (Osawa, 1970).

Chemical vapor deposition, pyrolysis, hydrothermal treatment, templating techniques, heating polymer spheres in an inert atmosphere, reduction of glucose, and pressure carbonization are a few convenient methods employed to synthesize carbon nanostructure (Bazargan and McKay, 2012; Pol and Michael, 2011; Szabó et al., 2010). However, in the literature, most of the synthesis techniques to produce pure carbon nanostructure materials are limited by several factors. In a few cases, the proportion and yield of the produced materials are low, and one type of structure cannot be easily separated from the remaining carbon structures or carbon soot. In some of the cases, the catalysts are trapped and remained as additional impurities in the product, which needs further processing. Autogenic reactions, involving the decomposition of carbon-containing precursors, such as organic compounds and plastic waste, at elevated pressures and temperatures have recently been reported as a new approach to prepare spherical carbon particles (SCP) (Pol et al., 2009; Pol, 2010) and carbon nanotubes (CNT) (Pol and Thiyagarajan, 2010).

Polymers, such as polyethylene (PE), polypropylene (PP), polystyrene (PS), polyvinyl alcohol (PVA), polyvinyl chloride (PVC), phenol formaldehyde (PF), polycarbosilane (PCS), polytetrafluoroethylene (PTFE), and polyethylene terephthalate (PET) have been studied as low-cost feedstocks for CNT and graphene synthesis (Deng et al., 2016). Moreover, various catalysts have been examined, including transition metals in either chemical compound form (ferrocene, ferrous chloride, nickel oxides, or cobalt acetate) or in elemental form, such as iron or nickel, among others (Zhuo and Levendis, 2014).

The aim of this research is to discover the most environment-friendly solution to transform negative-valued wastes to valuable products. In this study, waste of polyethylene terephthalate (PET) bottles was chosen because it is one of the serious postconsumer plastic wastes that has caused lots of environmental problems. This paper describes a process for producing carbon nanostructure materials from PET waste using a solvent-free, one-pot process. This is a simple, reproducible, and affordable process that involves catalytic thermal decomposition of PET waste in a closed reactor under autogenic pressure. The main novelty of the present work is based on the acquisition of high-value carbon nanostructured materials from PET waste by an alternative simple technique comparable to chemical recycling to minimize the disposal problems.

Experimental methods

In this research, carbon nanostructure materials have been fabricated by fragmentation or “cutting” of PET bottle waste (top-down approach).

Materials and equipment

All reagents are commercially available and used without further purification. The instruments used are: balance with shield chamber, four digits (Sartorius, model CP225D, Germany); drying oven (Nabertherm, model TR60, Germany); and furnace (Carbolite, model MRF 16/22, UK). The autoclave jar, made of stainless steel (SS316), could be repeatedly used for such incineration processes. The reactor is designed to operate typically up to a maximum working pressure of about 130 bars and a maximum temperature of about 1000°C.

Samples preparation

Carbon nanostructure materials were prepared by the thermal decomposition of PET waste in an enclosed stainless-steel (SS316) autoclave having 50 mL capacity. The PET waste was crushed and sieved to obtain desired size fractions (1–3 mm) using a conventional sieve-shaker. Then 2 g of raw PET waste was introduced into the stainless-steel autoclave reactor with catalysts (Ferrocene, Sigma-Aldrich, 98%) in different weight ratios. The closed stainless-steel reactor filled with PET waste was placed inside the center of an electric furnace at room temperature; then the temperature of the furnace was increased to 800°C over 100 min and remained at this temperature for 20 hr, and the system was left to cool overnight. The reaction took place under the autogenic pressure of the decomposed PET. The dark product was collected and crushed. The chemical reaction that took place under created autogenic pressure during the thermolysis of PET in the presence of catalysts led to the growth of carbon nanostructure materials.

The yield of produced sample was calculated from eq 1 (Esfandiari et al., 2012):

(1)

The burn-off (BO) that was one of the most critical factors in physical method of carbon nano structure materials production was calculated as (Esfandiari et al., 2012)

(2)

where w₁ is weight of raw materials and w₂ is weight of product.

In the incineration of PET waste treatment technology, which involves the combustion of organic materials and/or substances, our present approach can be implemented. However, the design of the incinerator has to be modified to facilitate the fabrication of carbon nanostructure materials in the presence of ferrocene as a catalyst. Figure 1 describes the preparation setup, in addition to hydrocarbon feedstock, and it is indeed important to seal the reactor while pyrolysis of PET waste in the presence of catalyst is occurring. That allows hydrocarbons to be in close proximity to the catalyst, which facilitates the growth of carbon nanostructure materials. In the present system, not only the temperature but also the autogenic pressure plays an important role in the growth of carbon nanostructure materials. The carbon nanostructure materials settle nicely at the bottom of the autoclave for easy collection.

Figure 1. The scheme of the degradation of PET waste in a closed system to produce carbon nanostructure materials.

A series of experimental tests were initially performed to evaluate the influence of the catalyst and other parameters on the formation of carbon nanostructure materials. Table 1 contains the experimental conditions used during the tests.

Table 1. Operating conditions used during the experimental tests.

Test number	Sample code	PET weight	Ferrocene weight*	Run time	Run temperature	Other additives	Yield (%)
1	1Fe	2 g	0.01 g	1 hr	800ºC		36%
2	2Fe	2 g	0.5 g	22 hr	800ºC	20 mL H2O2(30%)
3	3Fe	2 g	0.5 g	20 hr	800ºC	15 mL H2O2 (30%)
4	4Fe	2 g	0.5 g	20 hr	800ºC	15 mL ultrapure water
9	9Fe	2 g	0.5 g	20 hr	800ºC	5 mL ultrapure water	17%
	9C	2 g		20 hr	800ºC	5 mL ultrapure water	27%
10	10Fe	2 g	0.5 g	20 hr	800ºC		13%
	10C	2 g		20 hr	800ºC		8%
11	11Fe	4 g	1.0 g	20 hr	800ºC		16%

*Fe content equal 30% of ferrocene [Fe(C₅H₅)₂] weight.

Carbon nanostructure materials characterization

The material was characterized by selected physical and chemical properties such as the following:

A transmission electron microscope (TEM) (TECNAI G20, Super twin, Double tilt, FEI, Netherland with EDX) was employed to obtain images for studying the nanostructure and elemental analysis (qualitative and semiquantitative analysis) of the carbon products. In order to prepare the samples for TEM analysis, approximately 1 mg of the dry black powder was dispersed in 15 mL ethanol using an ultrasonicator, and this yielded a dispersion of the carbon nanostructure material. Ten microliters of this dispersion was put on a coated copper grid. The dried grid was then examined under an electron microscope (HR-TEM; transmission electron microscope operating at 200 kV). Two different modes of imaging were employed: the bright field at electron accelerating voltage 200 kV using a lanthanum hexaboride (LaB₆) electron source gun, and diffraction pattern imaging. An Eagle CCD camera with 4k × 4k image resolution was used to acquire and collect transmitted electron images. TEM imaging and analysis (TIA) software was used for spectrum acquisition and analysis of EDX peaks.

Crystallographic information about the carbon micro structure materials was obtained from powder x-ray diffraction (XRD) data collected with a Schimadzu-7000 (USA) with CuKα radiation beam (λ = 0.154060 nm). The finely powdered samples were packed into a flat aluminum sample holder, where the x-ray is generated at 30 kV and 30 mA with a copper target. Scans are performed at 4° min⁻¹ for 2θ values between 10 and 80°.

Raman spectroscopy was performed on the dried samples at room temperature using a Senterra Raman spectrometer (Bruker, Germany) with a 514.5-nm excitation wavelength and power output of 10 mW to determine the extent of graphitic disorder within the carbon materials. Each measurement was taken for 60 sec and repeated four times in the range of wave numbers from 40 to 3500 cm⁻¹.

Results and discussion

Because of the significance of future commercial application of carbon nanostructured materials, this paper describes the as-obtained materials as being formed during thermal decomposition of PET waste. Here, transmission electron microscopy (TEM) was used primarily for the characterization of product morphologies where the observations were reported and discussed, along with x-ray diffraction (XRD) analysis, which was used for the determination of product structure (crystalline or amorphous), and Raman spectroscopy to determine the extent of graphitic disorder.

Effect of ferrocene on PET waste thermal decomposition

In this case, a study of different carbon nanostructure materials formed during thermal decomposition of PET waste in the presence of 25 wt% ferrocene as a catalyst (sample code 10Fe) was performed and compared with the resultant sample coded as 10C.

Morphologies of products

TEM investigations revealed that during thermal decomposition of PET, multilayered graphene sheets were formed, consisting of more than 10 layers (sample 10C, Figure 2a), while multishell fullerenes were formed in addition to graphene sheets when the ferrocene catalyst was added (sample 10 Fe, Figure 2b).

Figure 2. TEM images of the final product of autogenic thermal treatment of PET (a) without additives at 800ºC for 20 hr (sample code 10C), (b) with ferrocene (25 wt%) at 800ºC for 20 hr (sample code 10Fe), (c) with 5 mL ultrapure water at 800ºC for 20 hr (sample code 9C), (d) without additives at 800ºC for 1 hr (sample code 1Fe), and (e) EDX for part (b) and (f) EDX for part (d).

The experimental TEM images provide compelling evidence for the transformation of graphene to fullerene structure. However, a key to the initial step in the graphene to fullerene transformation is the loss of carbon atoms at the edge of graphene flake. Carbon atoms at the edge of a graphene flake are unstable, because only two bonds connect them to the rest of the structure. The loss of carbon atoms at the edge and the subsequent reconstruction do not cause any significant changes to the structure of a large graphene sheet (Chuvilin et al., 2010). However, small fragments of graphene, as our extended observations demonstrate, undergo drastic structural transformations, leading to the formation of fullerene cages (Smalley, 1992; Goroff, 1996; Lozovik and Popov, 1997). Fullerene is the most stable configuration for a finite number of sp² carbon atoms because the molecular cage has no open edges, and all the carbon atoms form three bonds. A theoretical study exploring the possibility of the transformation of a graphene sheet into a fullerene confirms that the formation of defects at the edge of graphene is the critical step in the process (Lozovik and Popov, 1997; Dresselhaus et al., 1996). The structural defects are based on a series of rearrangements that give rise to pentagonal rings. However, the energy barrier for such rearrangements appeared to be extremely high, making this pathway reasonable only at extremely high temperatures, significantly higher than the temperatures used in the experiment under investigation. However, the loss of the outermost carbon atoms in a graphene flake provides an applicable route for fullerene formation under realistic experimental conditions. The initial size of the graphene flake is important, because it determines the size of the fullerene cage that can be formed. If the flake is too large, in the region of several hundreds of carbon atoms, there will be a significant energetic consequence during the curving step associated with the van der Waals interactions between the underlying graphene sheet and the flake (Chuvilin et al., 2010). Its edges will continue to be etched until the flake reaches a size that enables the formation of fullerene. On the other hand, the transformation of very small flakes (less than 60 carbon atoms) into fullerenes will be suppressed by excessive strain on C–C bonds imposed by the high curvature of small fullerene cages and the violation of the isolated pentagon rule in fullerenes smaller than C₆₀ (Slanina et al., 2008). A number of multishell fullerenes have been observed, as shown in a higher magnification TEM image in Figure 3. Indeed, these experiments indicate that fullerene cages formed directly from graphene have a relatively wide range of diameters, averaging ~20 nm. It is interesting to note that the multishell fullerenes that we directly observed in this work represent an interesting and novel variety of carbon clusters. The multishell fullerenes are essentially different from much bigger and essentially graphitic carbon onions (Dresselhaus et al., 1996; Mordkovich, 2000). Figure 3 shows that the outer shells of multishell fullerenes do not resemble perfect rings that may refer to the higher fullerenes that have complicated shapes and that may look very different if viewed from different angles (Mordkovich, 2000). To further investigate the composition of the synthesized carbon nanostructure materials, energy-dispersive x-ray (EDX) was used to analyze the elemental composition for samples (10C and 10Fe). The synthesized nanostructured materials consisted of carbon only are shown in Figure 2e, indicating the high purity of the resulting products (the presence of copper is due to the sample holder and pthe resence of silicon is due to dust impurity).

Figure 3. TEM images of the final product of autogenic thermal treatment of PET with ferrocene (25 wt%) at 800ºC for 20 hr (sample code 10Fe).

Structure of products

The x-ray diffraction patterns of the synthesized carbon nanostructured materials are shown in Figure 4. The disordered structure is reflected by the broad x-ray diffraction peaks centered at approximately 26, 42.3, and 44.3 2θ that correspond to the (002), (100), and (101) reflections, respectively, for sample 10C. It is typical for highly amorphous graphite like carbons (Sergiienko et al., 2009; Zhang et al., 2016). The broad (002) peak, in particular, encompasses diffuse sets of interlayer distances that, on average, are larger than those in crystalline graphite (typically 0.344–0.355 nm) (Sergiienko et al., 2009). The presence of the (101) peak suggests a stacking ordering of graphene sheets (Shen and Lua, 2013). While for sample 10Fe as shown in Figure 4 several distinctive peaks are well indexed to hexagonal carbon (ICDD card no. 00-056-0159) such as the (101) plane, in addition a rhombohedral iron oxide (ICDD card No 01-071-5088) crystal and no iron carbide phase were observed in the XRD pattern.

Figure 4. XRD patterns of the final product at 800ºC for different samples at different conditions coded as in Table 1.

Raman

Figure 5 shows that the Raman spectrum of the synthesized samples is typical for a hard carbon, with a broad band at 1331–1338 cm⁻¹ representing a highly disordered (D) graphite arrangement and a band at 1573–1599 cm⁻¹, characteristic of a more ordered graphitic (G) structure. The D band has been attributed to the vibration of carbon atoms with dangling bonds for the in-plane terminated disordered graphite component. The G-band, corresponding to the E_2g mode, is closely related to the vibration of sp² bonded carbon atoms in a two-dimensional hexagonal lattice, as in graphene (Motiei et al., 2001; Sergiienko et al., 2009). When the bond lengths and angles of graphene are modified by strain, caused by the interaction with other graphene layers or due to external disorder, the hexagonal symmetry of graphene is broken (Malard et al., 2009; Ni et al., 2008) The G-band is therefore highly sensitive to strain effects in sp² nanocarbons and can be used to investigate any modification to the flat geometric structure of graphene, such as the strain induced by external forces, by one graphene layer on another in a few layers of graphene, or even by the curvature of the side wall as in the case of adding ferrocene, which leads to formation of fullerenes. The presence of the D peak in the Raman spectra of experimental powder samples (Figure 5) can be ascribed to the presence of finite-size crystals of graphite (nanographite crystals) (Dresselhaus et al., 2010). The intensity ratios of the D and G bands from Figure 5 are presented in Table 2, thus quantifying the relative levels of disorder (Sergiienko et al., 2009).

Table 2. Results of Raman spectra for the produced powder samples.

Sample	Raman peak position (cm^−1)		ID/IG
	D band	G band
1Fe	1333	1586	1
2Fe	1335	1593	1
3Fe	1338	1585	1.14
4Fe	1336	1599	1
9C	1337	1586	1.13
9Fe	1331	1596	0.94
10C	1338	1599	0.9
10Fe	1333	1588	1.15
11Fe	1335	1573	1.11

Figure 5. Raman patterns of the final product at 800°C for different samples at different conditions coded as in Table 1.

All kinds of sp² carbon materials exhibit a strong Raman feature that appears in the range 2500–2800 cm⁻¹. Together with the G band (1582 cm⁻¹), this spectrum is a Raman signature of graphitic sp² materials and is called the G′ band or 2D band to emphasize that it is a Raman allowed mode for sp² carbons (Dresselhaus et al., 2010). The 2D band provides a very sensitive probe for characterizing specific sp² carbon nanostructure. For example, the 2D band can be used for differentiating between single- and double-layer graphene (Malard et al., 2009). This band is also quite different from bulk graphite, for which the further increase in layers leads to a significant decrease of the relative intensity of the lower frequency peaks. For more than five layers the Raman spectrum becomes hardly recognizable from that of bulk graphite. Thus, Raman spectroscopy can clearly distinguish a single layer, from a bilayer, from a few (less than five) layers (Ferrari et al., 2006). Figure 6b shows a weak and wide beak in the range of the 2D band, which is slightly increased in intensity when PET has catalytic graphitization with ferrocene. This observation provides further proof that the 2D peak is a defect induced in graphene as fullerenes. In addition, the sample synthesized without any catalyst (sample 10C) exhibited lower I_D/I_G ratio and hence lower disorder for sample 10 Fe.

Figure 6. Raman spectra of graphene sheets for samples 10C and 10Fe: (a) D and G band peak, and (b) 2D band peak.

The role of ferrocene

At a temperature higher than 500°C, ferrocene decomposes completely spontaneously following the chemical reaction

as well as reactive hydrocarbons (Barreiro et al., 2006; Bhattacharjee et al., 2014). Thus, when ferrocene enters the reaction zone, iron clusters and reactive carbon in the gas phase are produced. These clusters act as catalyst nuclei for carbon nanostructure materials as fullerenes.

Effect of water on PET waste thermal decomposition

In order to investigate the effect of adding 5 mL ultrapure water on thermal decomposition of PET waste, a comparison between sample 9C and sample 10C was carried out using the following techniques.

Morphologies of products

The HRTEM image of sample 10C illustrated in Figure 2a shows a mixture of thick graphene (>10 layers) [marked with a black square] and graphite deposits made of both few and multilayer of graphene [marked with a white square]. However, for sample 9C, which is synthesized by adding 5 mL ultrapure water to the thermal decomposition reaction of PET waste, the formation of a few layers (4–10 layers) of graphene [appear as relatively transparent sheets rippled and entangled with each other marked with a black square] and both thick graphene sheets (>10 layers) [marked with a white square] as graphite deposits is observed, as shown in Figure 2c.

Structure of products

In both cases, the XRD profiles of the two samples have two broad peaks over the 2θ = 20–30º and 2θ = 40–50º ranges. The first broad peak is from the disordered carbon of the powder sample. The second peak arises from the small size of the carbon nanostructure, and some amorphous state could result in peak broadening (Sergiienko et al., 2009), as shown in Figure 4.

Raman

The Raman spectrum of graphene is dominated by three main features: the first-order D and G bands around 1335 and 1590 cm⁻¹, respectively, and the second-order 2D band at 2660 cm⁻¹, with each having different physical origins.

The characteristics of the 2D band vary with the number of layers in graphene flakes. The 2D band is hardly detectable in the spectra, as shown in Figure 5 for samples 9C and10C, and it is also in agreement with the formation of a few layered graphene sheets, which are observed in the HRTEM. However, the intensity ratio of the D and G bands from Figure 5 and Table 2 for samples 10C and 9C (I_D/I_G) are 0. 9 and 1.13, respectively, thus quantifying the relative levels of disordered. The barely broader D and G bands observed in the spectra of the experimental powder sample 9C in comparison with that for sample 10C indicate the smaller size of graphite clusters and obvious disorder of the carbon structure, as the powder sample contains quite a large amount of disordered carbon.

From these observations, it can be concluded that the graphene nanosheets formation occurs with less disorder (amorphous graphite) in the absence of H₂O in the same experimental conditions.

Effect of time on PET waste thermal decomposition

To clarify the effect of time, a comparison between sample 1Fe [iron to PET raw material about 0.0015 wt%, which was omitted] and sample 10C was carried out using the following techniques.

Morphologies and structure of products

A systematic investigation on the statistical distribution of the graphene sheet layers was conducted via careful examination of a large number of TEM micrographs obtained from HRTEM. The data reveal that the synthesized sample (1Fe) consists of 35% few-layer graphene sheets and 65% multilayer graphene sheets. It was observed, as shown in Figure 2d, that the graphene structure is identical to those shown in Figure 2a and the formed carbon nanostructures are sheet-like. To explore the composition of the synthesized carbon nanostructured materials, energy-dispersive x-ray (EDX) was used to analyze the elemental composition of the synthesized carbon nanostructured materials for sample 1Fe. The synthesized product consisted of carbon and traces of oxygen, as shown in Figure 2f. However, the XRD profile of sample 1Fe (Figure 4) looks like sample 9C and sample 10C, which represent a typical pattern of graphene.

Raman

Figures 7a and 7b represent a typical Raman spectrum of samples 1Fe and 10C. Several peaks are clearly visible at 1333 cm⁻¹ (D band), 1586 cm⁻¹ (G band), 2690 cm⁻¹ (2D band), and 2885 cm⁻¹ (D + G band), as shown. In particular, the 2D and G bands represent the key features of Raman spectroscopy for the identification of graphene sheets (Rao et al., 2009). The relative intensity ratio of I_2D/I_G can be used to distinguish the number of layers of graphene sheets. The integrity intensity ratio I_2D/I_G of <1 corresponds to many-layered graphene (Shan et al., 2012). With an increasing number of layers, the intensity ratio of I_2D/I_G decreases while the full width at half maximum (FWHM) of the 2D peaks increased (Shan et al., 2012; Shen and Lua, 2013). No single-layered graphene sheet was identified by the Raman test. The intensity of the D-band peak decreased with increasing number of layers (Shen and Lua, 2013). It was believed that the intensity of the D peak was closely related to the defects, that is, edges, ripples, and folders, in graphene sheets. From Figure 7a, it is noteworthy that the intensity of D band peak for samples 1Fe is higher than that for sample 10C. Furthermore, from Figure 7b, a weak and wide peak in the range of the 2D band peak for sample 1Fe has also higher intensity than that for sample 10C, at which further increase in layers leads to a significant decrease of the relative intensity of the lower frequency 2D peaks (Ferrari et al., 2006). This observation provides further proof that the number of formed graphene layers decreases with time decrease.

Figure 7. Raman spectra of graphene sheets for samples 1Fe and 10C: (a) D and G band peak, and (b) 2D band peak.

Effect of additives and catalyst on PET waste thermal decomposition

It is essential to first ask under what conditions different forms of carbon nanostructures are found when PET waste decomposes in an enclosed stainless-steel autoclave reactor at 800ºC under autogenic pressure. Two aspects appear to be important. First, the observation of the novel growth phenomena is tied to the presence of catalytic metal in the decomposition process. Second is the additives effect on the products. We describe the two aspects in this sequence in the following.

As shown in Figure 8a from the HRTEM image, most of the obtained nanostructure materials are shown as flat graphene films, but some look like transparent nanosheets coexisting with hollow quasi-spherical nanocarbon cages and some, like a string of beads, look like nanorods mixed with amorphous carbon. Each bead consists of a hollow core surrounded by near-spherical graphitic shells. Figure 8b for sample 4Fe reveals that the carbon nanostructure materials produced from using 15 mL ultrapure water contains fullerenes, graphene sheets, and strings of beads, which appear less dense, more intervening, and overlapping of the strings of beads with the same general character as in sample 3Fe.

Figure 8. TEM images of the final product of autogenic thermal treatment of PET raw material with ferrocene: (a) with 15 mL H₂O₂ at 800ºC for 20 hr (sample code 3Fe), (b) with 15 mL ultrapure water at 800ºC for 20 hr (sample code 4Fe), (c) with 5 mL ultrapure water at 800ºC for 20 hr (sample code 9Fe), (d) at inside surface of autoclave jar for sample 10Fe, and (e) EDX for part (d).

In order to understand the mechanism of strings of beads growth, the researchers consider H₂O₂ or ultrapure water with catalyst that stimulates growth at the interface, starting from amorphous material that converts into graphitic walls near the string. As the growing bead removes itself from the interface, this catalytic stimulus reduces, and the bead may follow its own growth dynamics. It may remove itself from the interface, and enclose to form a sphere, thus decreasing the surface energy associated with an open-ended structure. The newly generated bead joins the string behind it, and a new cycle is started at the interface (Seraphin et al., 1994). The catalyst position at the tip of the string could be incidental, or could result from the formation of numerous nucleation centers in one event at one initial time. From these observations, it can be concluded that the string-of-beads formation occurs when the oxidizing agent is used in large volume and is absent when water is employed under the same conditions.

As shown in a TEM micrograph (Figure 8c) for sample 9Fe, which was prepared by adding ultrapure water and ferrocene (see Table 1), transparent nanosheets, wrinkles, and ripples were observed on the synthesized graphene sheets. For sample 10Feprepared without adding additives (see Table 1) as shown in Figure 2b, multishell fullerenes coexist with graphene sheets. Furthermore, the TEM micrograph (Figure 8d) for the same run of sample (10Fe) and by taking a whit of soot, which is at the inside surface of the autoclave jar, shows a very obvious contrast between the graphene sheets and the iron catalyst particles. The black spots indicate the catalyst particles as approved by EDX analysis (the presence of copper is due to the sample holder and the presence of silicon is due to dust impurity), as shown in Figure 8e. The continuous and featureless regions are indicative of the central parts of the graphene sheets, while the slightly stretched black regions are indicative of the scrolls and folds of the graphene sheets.

However, the x-ray diffraction pattern, as in Figure 4, of the carbon nanostructures of samples 3Fe, 4Fe, 9Fe, and 10Fe shows several distinctive peaks to be well indexed to hexagonal carbon, in addition to a rhombohedral iron oxide crystal. In this study, the characteristic diffraction peaks of the carbide phase have not been detected and there is no significant difference between the four patterns. Moreover, the Raman pattern as in Figure 5 of the carbon nanostructures of samples 3Fe, 4Fe, 9Fe, and10 Fe represents a typical Raman spectrum for carbon nanostructure materials. Several peaks are clearly visible for the D band and G band, as illustrated in Table 2.

Effect of doubling reactants on catalytic thermal decomposition of PET waste

With reference to synthesis conditions for sample 10Fe, the researchers investigate the outcome (sample code 11Fe) of multiplying the reactants (as in Table 1). From TEM observations for sample 11Fe as shown in Figure 9a, aggregations of hollow quasi-spherical nanocarbon cages, with outer diameters of 12–70 nm and inner diameters of 7–50 nm, were observed. The crystallites of the carbon shell, defined as a stack of lattice borders, were randomly oriented and tangled in a complicated manner and not very large. Lattice borders continued through many crystallites, merging and branching several times, and they were denser than for sample 10C. However, the x-ray diffraction pattern as in Figure 4 of the carbon nanostructures of sample 11Fe shows several distinctive peaks to be well indexed to hexagonal carbon, in addition to a rhombohedral iron oxide crystal, which was like sample 10Fe, and there was no significant difference between the two patterns.

Figure 9. TEM images of the final product of autogenic thermal treatment of PET raw material with ferrocene: (a) without additives and with doubling reactants at 800ºC for 20 hr (sample code 11Fe), and (b) with 20 mL H₂O₂ at 800ºC for 22 hr (sample code 2Fe).

Furthermore, sample 10Fe and a sample prepared by doubling reactant (sample 11Fe) were compared with each other (Figure 5) using Raman spectrum patterns. As shown in Figure 5, the peak at about 1335 cm⁻¹ (which represents the D band) for sample 11Fe was stronger than that at about 1333 cm⁻¹ for sample 10Fe, which may be the result of a stronger lattice distortion of graphite layer in the sample 11Fe. For the G band at 1588 and 1573 cm⁻¹ for sample 10Fe and 11Fe, respectively, it was a doubly degenerate phonon mode (E_2g symmetry) related to ordered in-plane sp² carbon atoms. The measured I_D/I_G values of sample 10Fe and 11Fe were 1.15 and 1.11, respectively.

Effect of H₂O₂ and time on catalytic thermal decomposition of PET waste

In order to investigate the effect of heterogeneous oxidation on catalytic thermal decomposition products of PET waste, a brief study was done for carbon powder formed by catalytic thermal decomposition of PET waste with 20 mL of 30% H₂O₂ for 22 hr (sample 2Fe) and that produced with 15 mL of 30% H₂O₂ for 20 hr (sample 3Fe), using transmission electron microscopic observations, along with x-ray diffraction analysis and Raman spectroscopy.

As shown in Figure 9b and from HRTEM observations, most of the obtained nanostructured materials were shown as flat graphene films, but some flakes looked like transparent nanosheets and some like ribbon. Multiwall carbon nanotubes (MWCNT) were also observed in this sample as a minor structure. That can be explained, as, in the oxidation duration of formed graphite from PET decomposition, oxidation occurs at the edges of flake graphite first and then oxidants enter the inner layers (Xiu, 2013). Oxide radicals induced cutting along the edges and surface defects of graphite (Xiu, 2013; Li et al., 2009). Under thermal decomposition, formed graphite expanded rapidly because of the decomposing of the oxidants between the layers. It could be confirmed that H₂O₂ contributed to the expansion of the graphite flakes and helped to peel them into the thin grapheme-like nanosheets. Compared with other liquid-phase oxidizing agents, H₂O₂ is a relatively mild oxidant (Peng and Liu, 2006), and it is often used with metal ions such as nickel (Ni), iron (Fe), and copper (Cu), which can catalyze the degradation of H₂O₂ to form reactive hydroxyl radicals via the Haber–Weiss reaction (Torreilles and Guerin, 1990). Ni, Fe, or Cu catalysts are used for graphene synthesis using the chemical vapor deposition method (Reina et al., 2009). Thus, in this study the presence of ferrocene used during the synthesis may be catalyzing the degradation of the as-produced graphene during H₂O₂ treatment. Additionally, the solution of H₂O₂ decomposes thermally at 400ºC to form a gas phase of water and oxygen (Giguere and Liu, 1957).

However, the advantage of hydrogen peroxide over other oxidants in the oxidation of carbon nanostructure materials lies in H₂O₂ itself, which does not incorporate foreign metal elements into the carbon surface and can be used under neutral conditions. Thus, H₂O₂ oxidation could show special performances for purification and modification of carbon nanostructure materials (Peng and Liu, 2006).

On the other hand, the formation of MWCNT in sample 2Fe can be attributed to the increase of the yield of graphene, at which complete detachment of graphene sheets leads to the formation of carbon nanoscrolls with diameters of a few tens of nanometers (Viculis et al., 2003; Shioyama and Akita, 2003). The individual graphene sheet rolls around itself in order to minimize the surface energy (Shioyama and Akita, 2003). The newly formed curved layered structure resembles that of a multiwalled nanotube.

The carbon ordering can be determined using Raman spectroscopy. Figure 10 shows two strong peaks at 1593 and 1334 cm⁻¹ for sample 2Fe, whereas for sample 3Fe there are two peaks at 1585 and 1338 cm⁻¹. These could be attributed to the G band and disorder-induced D band, respectively, analogous to those of graphite. The D’ peak (1640 cm⁻¹) is believed to correspond to a strong maximum in the vibrational density of states (VDOS) of graphite (Peng and Liu, 2006). It is well known that in amorphous materials with no close order, the breakdown of the selection rules leads to Raman spectra that reflect the VDOS in their crystalline counterparts (Hoffman et al., 2005). The positions of the D and D’ bands may vary considerably, depending on the structure of the disordered carbon. In contrast, the relative intensities of D bands versus those of G peaks (including G and D’ bands) were significantly changed between two samples. The changes of the ratios of D to G are illustrated in Table 2, I_D/I_G, where they are increased by decreasing time and amount of H₂O₂ used. Increase in the I_D/I_G ratio for sample 3Fe can be attributed to an increase in the number of defects, which can be attributed to destruction of the multiple layers of graphite due to progressive increase in the defect sizes (Xing et al., 2014). Additionally, according to Figures 8a and 9b, the contents of ordered graphite structures seemed to decrease in sample 3Fe, which contained different graphite structures (few and multi graphene sheets; fullerenes; single-wall carbon nanotube filled with fullerenes). Furthermore, after careful analysis of transmission electron microscopy (TEM) micrographs, the physical natures of carbon nanostructure samples contributed little to the variance of the ratios of I_D/I_G. Thus, it could be confirmed that the oxidation occurred at ordered and disordered parts; additionally, disordered parts or low-organization carbon materials were preferentially removed under longer oxidation time.

Figure 10. Raman spectra for samples 2Fe and 3Fe.

These results imply the need for systematic studies to investigate the direct chemical interaction between H₂O₂ and graphene. The results also indicate that the rate of degradation of formed carbon nanostructure is dependent on H₂O₂ exposure concentration; an increase in concentration accelerates the formed nanocarbon structure’s degradation. Furthermore, reaction temperature may also influence the degradation rate. Additional studies are necessary to provide better understanding of chemical mechanism of H₂O₂-mediated degradation of carbon nanostructure (interplay of H₂O₂ concentration, reaction temperature, and presence of metal ions).

Conclusion

By optimizing the preparation conditions with the addition of appropriate amounts of ferrocene, water, and H₂O₂ it is possible to control types of the products, morphology, and size distribution of the particles produced and the process yield. These results can be summarized as in the following:

Multishell fullerenes coexisted with graphene sheets synthesized by adding only 0.5 g of ferrocene to the 2 g of PET and treated at 800ºC for 20 hr. Transparent wrinkle and ripple graphene nanosheets were produced by adding 5 mL water to the previous mixture. A mixture of fullerenes, graphene sheets, and strings of beads was produced by increasing the amount of water added, up to 15 mL. Furthermore, it was shown that the content of fullerenes depended on the conditions of the synthesis, and fullerenes yield up to 14% by doubling reactants in the catalytic thermal decomposition of PET waste with ferrocene. In conclusion, this study explains how PET waste can be transformed into high-value carbonaceous material via a simple controllable process. The proposed method has a significant number of advantages over reported methods for the synthesis of carbon nanostructure materials, including (a) simple reaction setup, (b) easy operational procedure, (c) an efficient and facile process, (d) one-pot, environmentally friendly and soft processing approach, and (e) this method could be easily scaled up and has the greatest future potential for industrial production of carbon nanostructure materials.

Additional information

Notes on contributors

Noha A. El Essawy

Noha A. El Essawy is Researcher at Advanced Technology & New Materials Research Institute, City of Scientific Research and Technological Applications (SRTA-City), Alexandria, Egypt.

Abdelaziz H. Konsowa

Abdelaziz H. Konsowa is Dean of Faculty of Engineering, Alexandria University, Alexandria, Egypt.

Mohamed Elnouby

Mohamed Elnouby is Assistant Professor at Advanced Technology & New Materials Research Institute, City of Scientific Research and Technological Applications (SRTA-City), Alexandria, Egypt.

Hassan A. Farag

Hassan A. Farag is Professor at Chemical Engineering Department, Faculty of Engineering, Alexandria University, Alexandria, Egypt.