Abstract
The increasing use of carbon fibre reinforced polymers requires suitable disposing and recycling options, the latter being especially attractive due to the high production cost of the material. Reclaiming the fibres from their polymer matrix however is not without challenges. Pyrolysis leads to a decay of the polymer matrix but may also leave solid carbon residues on the fibre. These residues prevent fibre sizing and thereby reuse in new materials. In state of the art, these residues are removed via thermal treatment in oxygen containing atmospheres. This however may damage the fibre’s tensile strength. Within the scope of this work, carbon dioxide and water vapour were used to remove the carbon residues. This aims to eliminate or at least minimize fibre damage. Improved quality of reclaimed fibres can make fibre reuse more desirable by enabling the production of high-quality recycling products. Still, even under ideal recycling conditions the fibres will shorten with every new life-cycle due to production-based blending. Fibre disposal pathways will therefore always also be necessary. The problems of thermal fibre disintegration are summarized in the second part of this article (Part 2: Energy recovery).
Introduction
Production of carbon fibres
Carbon fibres are produced based on organic polymers. Depending on the mechanical and electrical properties, different precursor materials are used. Typical precursors are polyacrylonitrile (PAN), pitch and cellulose. The majority of commercially available carbon fibres are produced on the basis of PAN, since with this precursor the highest carbon yields can be achieved (Frank et al., 2014). Despite PAN in theory consisting entirely of acrylonitrile monomers (C3H3N), in fact impurities and co-monomers present during production are introduced into the structure. The precursors are subjected to a series of thermal treatment processes, during which volatiles are eliminated. Figure 1 outlines the production process based on PAN fibre. After pyrolytic treatment at about 1600°C, the fibres consist of up to 98% of carbon. Further treatment at temperatures up to 3000°C is also possible, but only a small fraction of fibres are subjected to this. The carbon content is thereby increased and the surface structure changed.
Scheme of carbon fibre production (Frank et al., 2014), modified.
The production parameters, especially the temperature, are dependent on the desired fibre characteristics (Cherif, 2011). Up to temperatures of 2000°C nitrogen is used as atmosphere. At higher temperatures argon is used, to prevent the reaction of carbon fibres and nitrogen to hydrogen cyanide (DePalmenaer et al., 2014). Independent of production parameters, fibres never consist of 100% pure carbon. Heteroatoms cannot be flawlessly integrated into the carbon structure and cause deficiencies instead.
Waste management situation
Carbon fibre reinforced polymers (CFRPs) pose a challenge to classic waste handling pathways. Deposition without prior treatment is restricted and even prohibited in some countries due to the high organic carbon content in the matrix. Fibre reuse is largely attractive because of the high production cost (87 to 307 US$/kg for CFRP components; Sauer et al., 2017), while the high calorific value of about 32 MJ/kg suggests energy recovery as a valid disposal path.
Carbon fibre wastes are usually classified as fibre reinforced polymer (FRP) wastes. However, CFRPs are only a small portion of FRPs. The majority of FRP wastes are glass fibre reinforced polymers (GFRP), as they account for 99% by mass of all produced FRPs (Gutiérrez and Bono, 2013). However, with an increase in CFRP usage, the share of CFRP in the FRP waste is also increasing (Oliveux et al., 2015).
A survey of carbon fibre content in waste streams by evaluating waste statistics is impossible due to the lack of waste allocation of CFRPs. The European list of wastes contains multiple possible waste codes for the classification of CFRPs.
In each case recycling is the preferable pathway. Carbon fibres are inert against corrosion and material fatigue (Assmann and Witten, 2013; Preller, 2011; Tötzke, 2005) and can in theory go through multiple life cycles. Analogous to the recycling of paper fibres, it has to be assumed, that fibre lengths shorten with each life cycle. This is based on basic cutting processed during composite production and waste treatment. Therefore, while from an ecological and economic standpoint recycling is preferable, a pathway for energy recovery by thermal processes of too short or damaged fibres will always also be necessary. The current state of thermal treatment will be discussed in the second part of this work (Part 2: Energy recovery). With reference to this fibre behaviour, Figure 2 illustrates the life cycle of carbon fibres and CFRPs.
Carbon fibre life cycle [authors’ own illustration].
Recycling
Potential recycling pathways for CFRP wastes are based on mechanical, chemical and thermal treatment. These processes can be used as single approaches or in combination. Each pathway provides unique challenges and opportunities. This work focusses on thermal recycling processes. Information about mechanical and chemical recycling can be found elsewhere (overview articles of Oliveux et al. (2015) and Job et al. (2016) are recommended, and their data have been used in Limburg et al. (2017)).
Thermal recycling of CFRP is defined by the decomposition of the polymer matrix at elevated temperatures. This decomposition is a thermal process that occurs by heat supply in an inert atmosphere (pyrolysis), with oxygen as a reaction partner (autothermal gasification for substoichiometric oxygen concentrations and incineration for at least stoichiometric oxygen concentrations) or under oxidizing non-oxygen atmospheres (allothermal water vapour or carbon dioxide (CO2) gasification). During thermal recycling, the fibres are usually heated in an inert atmosphere; thereby, the polymer matrix and polymer sizing are decomposed into short-chained gaseous hydrocarbons. Those can be burned to supply the heat for the pyrolysis. The fibres are inert in non-oxidizing atmospheres at the typically applied temperatures. Different polymer matrices have different decomposition temperatures, so for mixed materials the highest decomposition temperature of a single material has to be applied to the whole mixed fraction. While energetically inefficient, higher temperatures pose no problem for the fibres themselves in an inert atmosphere. A common problem for matrix materials is that they do not fully decompose into gaseous products but leave some solid residues on the fibres. These residues commonly consist of pyrolytic amorphous carbon that cannot further be removed in the inert atmosphere of the recycling process. The resulting products are brittle and need further processing before they can be recycled into high quality materials. The common standard is to remove these residues by adding small amounts of oxygen during the thermal treatment or subjecting the fibres to a second step of thermal treatment in an oxidizing atmosphere (usually air). The oxygen not only removes the carbon residues, but also damages the fibres resulting in a loss of about 30% tensile strength (Kümmeth et al., 2012).
Comprehensive studies about pyrolysis of CFRPs have been conducted by the Technische Universität Hamburg. These conclude that a two-step thermal treatment is most effective. First, the matrix is decomposed in inert atmosphere at about 600°C within a few minutes (pyrolysis). The residence time is matrix-dependant but generally has a minor influence. Examinations via Raman spectroscopy show, that during pyrolysis some matrix polymers are partially converted into char residues. Subsequently, the char residues have to be removed via oxidation. During this second treatment phase temperature and dwell time are of critical importance. If dwell times are chosen too long, the fibres may be severely damaged. The second treatment phase can be adjusted to the respective fibres. For example, dwell times of 2 hours in 400°C with air as oxidizing atmosphere have shown to be promising. Fibre surfaces are generally oxidized before sizing to improve bonding properties (Meyer, 2007, 2011). Oxidation can therefore not only lead to damage but may also have a positive impact on the recycling product by enhancing the fibres’ bonding properties for an optimized incorporation into the matrix (Planitz-Penno et al., 2017).
In a technical feasibility study, the heat input via microwaves was evaluated. Generally, this approach seems feasible, and the reclaimed fibres were clean, with no carbon residues on the surface. The treated fibres however, did show changes in their surface topology. Reclaimed fibres showed a decrease in average roughness from 8.99 nm to 5.97nm (Lester et al., 2004).
Gasification instead of pyrolysis has been taking into consideration as well. The University of Nottingham, UK, examined the fluidized bed process, which was initially developed for the recycling of GFRPs. In this procedure, a fluidized sand bed is heated with hot air to temperatures between 450°C and 550°C. CFRPs at grain sizes between 6 and 10 mm are introduced into the furnace. The matrix is burned and the reclaimed fibres are reportedly clean but show signs of surface contamination with matrix or carbon residues. The mechanical properties are 20% inferior, compared to new fibres (Kennerley, 1998; Pickering, 2004). Another approach with promising results is the pyrolysis in a tin bath. Pyrolysis conditions are 500°C at a dwell time of at least 1 minute. Remarkably, the tensile strengths of fibres increased about 20% compared to newly produced fibres. However, the stability of composite products was 10 to 20% lower than the stability of composites consisting of newly produced fibres (Tötzke, 2005).
In the scope of the research cluster MAI (Munich, Augsburg and Ingolstadt) Carbon, the possibility for the development of a pyrolysis plant was examined. The plan was for refurbishing an existing municipal waste pyrolysis plant and running industrial scale experiments there. Preliminary experiments in the plant were performed and showed problems in material intake and outtake that would require bigger modifications to the plant. Along with pyrolysis, the whole value chain was examined and evaluated (Kreibe et al., 2015; Kümmeth, 2012)).On an industrial scale, three pyrolysis plants for CFRP recycling are known worldwide. Each of those plants has a capacity of about 1000 Mg/a CFRP wastes. In all these plants, a conveyor furnace with exterior heating is used. The furnace area is usually purged with inert gas and dwell times are – depending on the input – around 5 to 15 minutes, at temperatures between 500 and 700°C. Following is a (thermally or mechanically) post-treatment and packaging of the desired products.
Each recycling pathway brings unique advantages and challenges. The ideal pathway depends on the fibre, matrix and desired properties of the recycling product. Thermal recycling is especially applicable for unknown or mixed materials. Thermal decomposition works for each matrix and results in the same problem – carbon residues on the fibres’ surface.
Materials and methods
Materials
All experiments within this work were performed under set atmospheres. Nitrogen of 99.999% purity and CO2 of 99.998% purity was used. Water vapour was created by evaporating demineralized water in the retort.
The experiments in CO2 and nitrogen atmosphere were performed in a Shimadzu TGA-50H thermogravimetric analyser (TGA) with sample masses of about 20 mg. Experiments under CO2, nitrogen and water vapour were performed in a round retort with a diameter of 113 mm and a height of 315 mm. The retort sample sizes were around 100 g. Volumetric flow rates of gasses were adjusted via Fisher & Porter rotameters. For microscopic evaluation, a Keyence VHX-2000 and VHX-5000 digital microscope with VH-Z500R lenses and 500-to 5000-fold magnification were used.
For the initial experiments in the TGA, four CFRP materials of known fibre and matrix composition and one unsized virgin fibre were used. Two of the materials consisted of high tensile strength fibres, type T800H, once with a matrix of Araldite LY 556 and once with Biresin CR141, both with unidirectional weave. The other two consisted of a different high tensile strength fibre, type T700, both with a REM TM14 matrix but once with unidirectional weave, the other with bidirectional weave. All used matrices are bisphenol-A–epichlorohydrin resins, therefore solely consisting of carbon, oxygen, hydrogen and nitrogen with ratios depending on the mixture and hardener used. For the retort experiments, a shredder fraction of materials from an actual practical carbon fibre recycler was used, in order to examine the effects on a broad range of different waste materials.
All fibres are produced based on PAN and are commonly used in an industrial scale. All matrices are common and widely used thermoset epoxy matrices. In order to examine the effects of different weave structures, samples C1 and C2 were chosen with the same fibre and matrix materials, by the same manufacturers, both used as tubing but with different weave directions. The samples were cut to lengths between 5 and 6 mm for the thermogravimetric analysis. This was achieved by initially cutting the materials with a metal saw and then shortening them to the desired lengths with a wire cutter.
Samples for the retort experiments were prepared from shredder fractions obtained from an industrial CFRP recycling plant in order to represent actual occurring wastes. To create homogeneous samples (about 100 g for each experiment), these shredder fractions were sorted by hand into similar material groups and samples, as similar to another as possible. Plastic sheets, used to prevent prepreg materials from prematurely hardening, were prominently present in the shredder fractions, and were removed before sampling to minimize impurities.
Methods
TGA experiments
The experiments aim at reducing the pyrolytic carbon residues remaining on fibres after pyrolysis.
Initially, about 20 mg of all CFRP-fractions were heated in nitrogen atmosphere at a heating rate of 10 K/min to determine the decomposition temperatures of the matrices. All matrices decomposed at temperatures below 500°C, so for further experiments, temperatures between 500°C and 1000°C were chosen. Near 500°C the Boudouard equilibrium favours CO2 and solid carbon, and near 1000°C the equilibrium favours carbon monoxide. Since no carbon monoxide was introduced into the reaction chamber and carbon monoxide resulting from the reaction was instantly purged with the gas flow of the controlled atmosphere, there is no formation of new carbon via the Boudouard reaction. For each chosen temperature, the sample was heated at a rate of 50 K/min, the maximum possible heating rate the instrument can perform, to the desired temperature and held there for 5 minutes. The first series of experiments was performed with a nitrogen atmosphere, and the second series with a CO2 atmosphere. In a third series of experiments, the sample was initially heated in a nitrogen atmosphere until the desired temperature was reached. Then the atmosphere was switched to CO2 and the temperature was held there for 5 minutes. The atmosphere was then switched back to nitrogen while cooling the sample. This was done to ensure a fixed dwell time of the material in a CO2 atmosphere at each temperature.
All products were analysed under the microscope. Therefore, the whole sample was first examined at a magnification of 500 ×. Randomly, at least five (to ten) locations were further examined at magnifications of 2000 ×. Additionally, noticeable locations showing damages or residues were also further examined at magnifications of up to 3000 ×. Previous experiments, during which fibres have been incinerated, have shown that fibres often exhibit strongly varying degrees of decomposition within only a few micrometres of each other. Sampling only small areas can easily lead to false conclusions. This risk was minimized by examining the entire sample. The samples were thereby also not further subjected to mechanical stress while being prepared for scanning electron microscopy, where a monolayer of fibres is desired.
All experiments were performed at least three times. If significant discrepancies occurred, further experiments were performed and the discrepancies noted in the corresponding experimental description.
Retort experiments
In order to evaluate the technical feasibility and the influence of non-ideal conditions, the experiments were scaled up to a retort with larger inputs. Additionally, while the TGA was not equipped to handle a water vapour atmosphere, experiments in the retort could also be performed under water vapour as an additional means of gasification.
For all experiments in the retort, about 100 g of the manually sorted shredder fraction were given into the retort and heated in atmospheres of nitrogen, CO2 and water vapour. For nitrogen and CO2 volume flows of 1 L/min were used. For the creation of water vapour the sample was initially heated with a nitrogen flow of 1 L/min to 200°C, then a flow of liquid water was slowly and steadily increased while constantly controlling the temperature. The nitrogen flow was steadily reduced during this process. At 300°C, significantly before the first reactions occur, the atmosphere was completely switched from nitrogen to water vapour. Temperatures within the retort were measured at three positions – the bottom, middle and top of the sample – with the bottom temperature generally being the highest and therefore used to set the maximum temperature. The heating rate of the oven was set to 10 K/min but the actual temperature increase in the retort varied strongly for different samples. Especially for experiments in water vapour, the actual heating rate was decreased by the water evaporation. Experiments were performed for maximum temperatures of 500, 600 and 700°C and these temperatures held for 15 minutes before switching off the heat source. During cooldown the retort was purged with nitrogen at 1 L/min, removed from the insulated furnace and additionally cooled with a vent.
Samples of three different material mixes were used for a series of experiments under different atmospheres each. The resulting products were analysed akin to the TGA experiments.
Results and discussion
TGA in different atmospheres
To test the potential of using CO2 as an oxidizing agent that does not damage the fibres, unsized raw fibres were heated to 1000°C at a rate of 50 K/min in this atmosphere and the temperature was held for 5 minutes. There was no weight-loss observed for both, nitrogen and CO2 atmospheres. In contrast to sized dry fibres, unsized fibres tend to break under even small mechanical stress. The resulting very light fibre fragments lead to continuous electrical problems, especially with the TGAs’ thermocouple, showing significantly wrong temperatures. Cleaning the thermocouple removed these errors, but to spare the equipment potential lasting damage, further experiments with unsized fibres were not performed.
The TGA experiments in nitrogen and CO2 atmospheres resulted in the matrix decomposition temperatures listed in Table 1. The decomposition temperatures are derived from graphic evaluation of the TGA curves and represent the initial decomposition temperatures conventionally referred to as onset temperatures. It is noteworthy, that for Araldite and Biresin the decomposition temperatures were slightly higher than in nitrogen atmosphere.
Onset thermogravimetric analyser temperatures for matrix decomposition and their standard deviations (mean ± standard deviation (SD)) in nitrogen (N2) and carbon dioxide (CO2) atmospheres. Temperatures mediated from three measurements, except Biresin CR14 and REM TM14 with bidirectional fibres, here mediated from five measurements.
The products of samples with Araldite and REM matrix were highly stiff and brittle. Attempts at manual fibre separation lead to the products breaking, but there was no separation of the fibres. This matches what is stated in the literature (Kreibe et al., 2015; Meyer et al., 2009), according to which the fibres are encased in pyrolytic carbon.
The decomposition temperatures of the Biresin matrix varied to a stronger degree between measurements than the other matrices – this was confirmed by additional measurements. Treated at 500°C in nitrogen, the fibres from a Biresin matrix stuck together, but could manually be separated with little mechanical influence and without the fibres breaking. At temperatures above 600°C, the fibres were already separated. Microscopic evaluation also showed that higher treatment temperatures lead to fewer carbon residues on the fibres. This matrix material only forms small amounts of pyrolytic carbon in nitrogen atmosphere, which lead to residue free reclaimed fibres.
The influence of the matrix material is however significantly higher than the influence of temperature. At 700°C Biresin only left small traces of residue on the fibre (Figure 3 left), while REM left large amounts of residue (Figure 3 right).
Left: fibres with Biresin matrix; and right: fibres with REM matrix, both treated at 700°C in nitrogen atmosphere.
The experiments with CO2 atmosphere performed were analogous to the experiments in nitrogen. Table 1 shows the decomposition temperatures of matrix materials in CO2 also. High standard deviations in REM decomposition temperatures occurred in experiments with bidirectional fibres. Additional measurements with bidirectional fibres were performed, for a total of five measurements. The bidirectional fibres provided further challenges during the experiments. These are detailed later in this Section.
For Araldite and Biresin the mean decomposition temperatures in CO2 are higher than in nitrogen atmosphere, but the decomposition temperatures for REM however are slightly lower.
Biresin could already be removed without residues under nitrogen. There is no significant difference, both microscopically and macroscopically between nitrogen and CO2 atmosphere.
The products of Araldite and REM samples after treatment in CO2 atmosphere exhibit significant optic and haptic differences from the results in nitrogen atmosphere as shown in Figure 4 for the example of Araldite. While samples treated in nitrogen atmosphere were stiff and brittle, those treated in CO2 atmosphere were present as loose, easily bendable fibres.
Fibres from an Araldite matrix treated at 750°C, left: in nitrogen atmosphere; and right: in carbon dioxide atmosphere.
Fibres from an Araldite matrix treated in CO2 were loosely oriented as a twine, while fibres from a REM matrix were still in their original form, but were flexible and could manually be separated. As in the experiments under nitrogen atmosphere, the fibres exhibited fewer carbon residues at higher temperatures (Figure 5). At 500°C the Boudouard reaction strongly favours the reagents, so even with the high concentration of CO2 the reaction only occurs slowly. In this experimental setup, higher temperatures also lead to higher dwell times in CO2. Since a constant heating rate of 50 K/min was applied a temperature difference of 100°C increases the dwell time by two minutes. This means a significant increase, compared to the hold times of five minutes at maximum temperatures. Therefore, in another experimental series, the heating and cooling phases were performed under nitrogen atmosphere and the atmosphere was changed to CO2 for the hold times only.
Fibres from an Araldite matrix treated under carbon dioxide. Left: at 500°C, and right at 650°C.
Since the Boudouard equilibrium favours the decomposition of carbon at higher temperatures, a positive effect by application of higher temperatures was to be expected. The lowest temperatures for residue free fibres are listed in Table 2.
Optimum treatment temperatures of the sample materials – lowest temperatures at which fibre separation was possible.
The results from the experiments with initial pyrolysis and subsequent treatment under CO2 showed no significant difference from the previous experiments under only CO2. Higher temperatures lead reliably to a better removal of carbon residues and the ideal temperatures were in the same region as previously. The experiments however showed, that CO2 does not or not only prevent or minimize the formation of carbon residues, but also removes residues formed during a previous pyrolysis.
The optimum temperatures for thermal recovery of the fibres are listed in Table 2. Those temperatures on the fibres were flexible, separable and free from carbon residues in microscopic examination. For Araldite and REM, no ideal treatment temperatures in nitrogen could be determined, since no fibre separation was possible.
In order to test the fibre stability, experiments in CO2 at higher temperatures than necessary were also performed. This resulted in one single defect being visible on a T700 fibre from a Biresin matrix at 700°C (Figure 6). At this treatment temperature, no other fibres with similar damages could be found. It is possible, this damage occurred in another way than during thermal processing.
Surface damage of a T700 fibre at 700°C in carbon dioxide atmosphere.
At a temperature of 1000°C however, while most fibres still were completely undamaged, a few fibres exhibited extensive surface damage as shown in Figure 7. Only a small portion of fibres was affected, showing damage along the whole surface. This leads to the conclusion, that the affected fibres share a surface condition that makes those fibres more susceptible to damage. Microscopic imaging shows now difference in surface topology of those fibres towards undamaged fibres. A possible reason might be the chemical compounds, ingredients, materials and/or crystalline structure of these fibres. Since these fibres are generally affected throughout their whole surface area, it can be theorized that the presence of these surface defects leads to the inclusion of more deficiencies during fibre production or fibre decomposition. In accordance with Gibbs et al. (1979) it may be assumed, that alkali metals from co-monomers of the PAN spinning process are introduced into the fibre, which then catalyse the fibre degradation.
Surface damage on singular T700 fibres at 1000°C in carbon dioxide atmosphere.
A special case presented itself for the unidirectional and bidirectional T700 fibres in the REM matrix. Despite being the same fibres and matrices, by the same manufacturer, they exhibited vastly different results in decomposition and residue reduction. The matrix decay for unidirectional Biresin and Araldite fibres in the TGA showed graphs as largely expected.
For REM, a smaller weight loss started at temperatures of about 200°C reproducibly, until the matrix decomposed in an additional larger step. The bidirectional fibres exhibited reproducibly an additional step in their mass degradation. Akin to the unidirectional REM fibres, the first mass decay starts at about 200°C and leads to the main matrix decomposition. The decay speed however slows again in another step until the matrix degradation is fully completed – as shown in Figure 8.
Thermogravimetric analyser curves of matrix decay for unidirectional fibres in Biresin matrix, unidirectional fibres in REM matrix, and bidirectional fibres in REM matrix.
The matrix decay during pyrolytic conditions is only dependent on the temperature and heat transport and should not be influenced by material transport. The layer thickness of the unidirectional compound was greater than the layer thickness of the bidirectional compound. The longer required dwell times can be explained by the higher matrix content in the bidirectional compound. The change in the decomposition speed however indicates heat transport deficiencies most likely caused by the fibres’ weave direction.
The experiments with bidirectional REM fibres also exhibited an increase in the electronic defects within the TGA, especially the thermocouple. Those problems occurred incidentally with all materials and could generally be fixed by cleaning the TGA and the thermocouple. During experiments with unsized raw fibres and bidirectional woven fibres however those defects were the rule, not the exception. Further experimentation was aborted – after various failed attempts – with the bidirectionally woven fibres in order to not risk damage to the equipment.
There is no single standardized way to comprehensively evaluate a reclaimed fibre’s quality. Measuring the tensile strength is recommended in the literature, but as detailed in Emmerich and Kuppinger (2014) where via chemical recycling fibres with tensile strengths akin to new fibres were recovered or Tötzke (2005) where tensile strengths during treatment even increased, this can lead to false conclusions. In both studies the high quality fibres did not result in CFRP products of a quality compared to new fibre products. This likely originates in the fibre surface changing during treatment and thereby not bonding as strongly to the new matrix or sizing.
In the current state of industrial recycling fibres are either contaminated with carbon or show reductions in fibre diameter or damages on the surface. In contrast the method applied here leaves no residues on the fibres and neither causes a reduction of fibre diameter nor damages to the surface. As the results for Biresin CR141 show, a direct comparison with other studies can only be achieved in a limited scope, as even with otherwise insufficient means certain matrices can be completely dissolved. For a conclusive evaluation larger amounts of fibres will need to be recovered and reintroduced into a new polymer matrix.
Retort experiments
The results from the TGA experiments could be verified for the shredder fraction in retort experiments. While the chosen dwell times in CO2 atmosphere were not always sufficient for a full removal of char residues, the fibres could in all cases be manually separated with little effort. Treatment in nitrogen leads to a brittle and stiff product, which could not be separated. Compounds with multiple layers of CFRPs generally showed lower degrees of processing – the inner layers still largely contained matrix residues and were harder to separate than mono-layered materials after treatment in CO2 and water vapour. Material and heat transport into the inner layers are insufficient at the given dwell times. Figure 9 shows the products of thermal treatment of a bidirectionally woven composite, once in nitrogen, once in water vapour. In both cases the general weave structure was retained and single strains could be removed without problems. The application of mechanical stress caused strains treated in nitrogen atmosphere to break and those treated in water vapour to separate into single fibres.
Carbon fibre composited treated at 700°C. Left: in nitrogen atmosphere; and right: in water vapour.
The experiments under water vapour lead to results largely comparable to CO2 with two major differences. Firstly, the intrusion depths into the layered laminates were higher than for CO2. Comparable layered composites were better separated at the same temperatures and dwell times. The latter might also be subject to material inhomogeneity. Secondly, the temperature profile within the retort was less homogeneous. This means the temperatures in the upper areas of the retort were significantly lower than for CO2. This led to single compound pieces with matrix residues on parts that were in the retort’s upper zone and areas completely free of matrix and carbon residues from the lower regions.
A huge advantage in the thermal processing with CO2 or water vapour lies in the universal applicability. The thermal treatment of CFRPs in inert atmosphere leads to – depending on the matrix – completely separable fibres or the formation of pyrolytic carbon encasing the fibres. At temperatures up to 750°C CO2 had no impact on the fibre integrity of the fibres tested in this article, while pyrolytic carbon could be visibly reduced.
Matrices yielding high amounts of pyrolytic carbon and large layered compounds naturally require longer dwell times – even longer than the 15 minutes tested in the experiments. The principal results however indicate that a full removal of char residues is possible. Dwell times and temperatures are easily adjustable to the specific needs of certain matrix materials and the application of a basic high temperature and long dwell time enable treatment of unknown compounds and mixtures.
Conclusion
With an increase in carbon fibre compound usage, sustainable material and energy recovery strategies for CFRP production wastes and end-of-life wastes are mandatory. Solid residues left on the fibres during pyrolysis and the problematic thermal behaviour of fibres in oxygen are among the major challenges provided.
The formation of solid char residues during pyrolysis could be reproduced for most matrix materials discussed in this article. With the exception of Biresin, all used known and unknown matrix materials left carbon residues during pyrolysis. Material and temperature thereby majorly influenced the amount of carbon formed.
These residues could be removed in CO2 and water vapour atmospheres at temperatures upwards of 600°C without damaging the fibres. The carbon residue reduction was macroscopically obvious and could be proven microscopically. Higher temperatures lead to shorter required dwell times. There was no visible fibre damage up to temperatures of 750°C and even at 1000°C most fibres showed no signs of damage. However, at 1000°C in CO2 atmosphere, some fibres were massively damaged along the whole surface. The reason for this damage is so far unknown. It may be assumed, that impurities and co-monomers introduced during fibre production cause irregularities in the surface structure and are responsible for the susceptibility to damage.
A multitude of influencing parameters affects the reclamation of carbon fibres – matrix material, matrix content, temperatures, atmospheres, dwell times, but even the weave structure – which complicates the recycling process control to reclaim high quality fibres. Further research into the decomposition behaviour of certain commonly used polymers for optimum individual treatment is required.
The thermal treatment and energy recovery of CFRP residues, which cannot be recycled anymore, will be the subject of the second part of this article (Part 2: Energy recovery).
Footnotes
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.