• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

A Process to Recover Carbon Fibers From Polymer-Matrix Composites in End-of-Life Vehicles

Posted on: Tuesday, 17 August 2004, 06:00 CDT

Because of their high strength-to-weight ratios, carbon-fiber- reinforced polymer-matrix composite (PMC) materials are being evaluated for use in the automotive industry. The major barriers to their widespread use are their relatively high cost and the uncertainty about whether they can be recycled. A process to recover carbon fibers from obsolete PMC materials has been developed at Argonne National Laboratory. The process was tested using PMC samples made with different thermoset or thermoplastic substrates. For most mixtures of PMCs, the process can be energy self- sufficient using the polymer substrate as an energy source. An evaluation of the recovered samples found that the fibers appear to have retained good properties and characteristics and are suitable for short fiber applications. This paper describes the process and the characteristics and properties of the recovered fibers.

INTRODUCTION

Because of their high strength-to-weight ratios, carbon-fiber- reinforced polymer-matrix-compos i te (PMC) materials are being evaluated for use in the automotive industry. The major barriers to the widespread use of PMCs are their relatively high cost and the uncertainty about whether they can be recycled. Argonne National Laboratory has developed and tested a process to separate carbon fibers from PMC scrap materials. Three process options-chemical degradation, thermal shock, and thermal treatment-were experimentally evaluated using PMC materials made with thermoset substrates. Because thermal treatment was found to be the most technically and economically promising method, its development was continued.

Figure 1. A conceptual design of the chemical degradation method.

THE CHEMICAL DEGRADATION METHOD

In the chemical degradation method, the PMC scrap was treated with chemicals including solvents, acids, and bases at various temperatures to degrade and break down the polymeric substrate in order to liberate the fibers. For example, treatment of some PMC scrap containing a urethane-basecl substrate, using atriethyleneglycol/water solution, at temperatures >240C resulted in the complete degradation of the polymer substrate and liberation of the carbon fibers. However, such treatment was not successful in liberating the carbon fibers from epoxy-based substrates. Therefore, it was apparent that different chemicals and solutions will be required to process different types of polymer substrates. In addition, the liberated fibers require washing to remove residual chemicals and solvents from the surface of the recovered fibers. This will generate liquid waste, which makes the process more expensive and less attractive. A design of the process was also developed (Figure 1) as well as a preliminary cost estimate. The process was found to be more costly than the thermal method. Therefore, its development was discontinued.

Figure 2. Carbon fibers recovered from a PMC sample containing a urethane-based substrate at 677C in air for 5 min.

Figure 3. Fibers recovered from a 10 cm by 30 cm panel.

Table I. Surface Chemistry of the Recovered and Virgin Fibers (wt.%)

THE THERMAL SHOCK METHOD

This method involved exposing the PMC scrap to rapid cycling between temperature extremes in an attempt to utilize the differences in the thermal expansion coefficients of the fibers and the polymer substrate to break the adhesion between the two. In some of the experiments, PMC scrap pieces were immersed in liquid nitrogen for a few minutes and then rapidly transferred to boiling water in a containment reactor. This process did not result in any visible separation of any of the PMC pieces that were tested. Pieces that were initially treated at elevated temperatures, resulting in significant but incomplete decomposition of the polymer, were also immersed in liquid nitrogen after they were allowed to cool to room temperature. This also failed to produce clean fibers. Further testing of this method was discontinued because it did not appear to be technically feasible.

Figure 4. The weight loss during thermal treatment of a PMC panel made with an epoxy resin under air or nitrogen environments.

Figure 5. The weight loss during thermal treatment of a PMC panel made with a polyurethane-based epoxy resin in argon gas.

Figure 6. The weight loss during thermal treatment of a PMC panel made with a polyurethane-based resin in air.

Figure 7. Scanning electron microscopy pictures of (a) recovered fibers and (b) virgin fibers.

Figure 8. The relative concentrations of functional groups based on areas of the fitting curves/peaks of high-resolution carbon spectra (XPS binding energy), sputtered (beneath the top surface).

THE THERMAL TREATMENT METHOD

This method involved heating the scrap to elevated temperatures under different environments. Experiments were condueted at temperatures in the range of 260C to 954C, and under an inert environment as well as in air under different flow rates. Polymer- matrixcomposite scrap made with different polymer substrates was also tested. Some test samples were as large as 10 cm by 30 cm and some as small as 1.2 cm by 1.2 cm. Tests were conducted in a batch oven and in a continuous thermal reactor.

Batch Experiments

Initially, bench-scale experiments were conducted in an oven at different temperatures in order to investigate the technical feasibility of the process, identify the effective temperature ranges, and investigate the impact of the environment under which the scrap is processed on the quality of the recovered fibers.

The bench-scale experiments demonstrated that the process is technically feasibleforprocessing PMC scrap made with different polymer substrates in the same batch. Separation of the carbon fibers from their polymer matrices could be achieved in one step and at residence times on the order of minutes, with only a small loss of the carbon fibers. The actual residence time depends on the treatment temperature and on the substrate material. Scrap made with epoxy substrates required the longest treatment time at a given temperature. When a mixture of PMC scrap containing a variety of thermoplastic and thermoset substrates is treated in the same batch, as would be the case in an actual recycling operation, the treatment temperature and residence time are selected such that the recovered fibers contain the minimum amount of non-fiber residue, fiber loss is kept to a minimum, and energy requirements and volatile organic compound (VOC) emissions are minimized.

Table II. Comparison of the Density and Diameter of Virgin and Recovered Fibers

It was also observed that over 90% of the polymeric substrate will be removed in a matter of a few minutes, generally less than 10 min. for most polymeric substrates. However, removal of the remaining portion may involve prolonged treatment and may require different treatment temperatures and/or oxygen environments.

A sample of carbon fibers recovered from the bench-scale experiments were examined under a microscope and, as shown in Figure 2, the recovered fibers appeared to have minimal residue on their surfaces. This sample contained a urethane-based substrate. Other samples were submitted to Oak Ridge National Laboratory for preliminary analysis and evaluation.1 The results indicated that the recovered carbon fibers had properties that compare favorably with those of virgin carbon fibers produced from polyacrylonitrile (PAN). This is significant although the samples processed may not have been originally produced from PAN. For example, the intrinsic density of the recovered fibers was 1.8473 g/cm^sup 3^ and their electrical resistivity was 0.001847 ohm-cm, compared to an intrinsic density of 1.75-1.9 gm/cm^sup 3^ and an electrical resistivity of 0.00020 - 0.002 ohm-cm for virgin fibers produced from PAN. A comparison of the mechanical properties of the recovered fibers (without surface treatment) with those of surface-treated virgin fibers from PAN revealed that the modulus for the recycled fibers (2157 .7 MPa) was about the same as that for the virgin PAN fibers (214.4 MPa). However, the ultimate tensile strength and the elongation at brake values are about 1/3 the values for the virgin fibers. A preliminary economic analysis of the process was also conducted, based on the results of the bench-scale experiments. That analysis suggested that a potential payback of less than two years is likely. Therefore, the development of this process was continued.

The bench-scale experiment demonstrated also that under high air- flow rates, oxidation of the carbon fibers takes place while at low air flow rates the residue on the fibers after the initial treatment will be significant. Therefore, this parameter has to be controlled properly.

Continuous Thermal Reactor Experiments

The authors also conducted a limited parametric study in a continuous thermal reactor to further define the process and had more samples characterized in order to determine the quality of the recovered fibers and to assess the impact of the treatment process on the carbon fiber properties and characteristics. Several types of PMC scrap were used in these experiments including PMC panels with well-known properties, wellcharacterized fibers, and wellcharacterized polymer substrates. For example, the fiber properties included: density (1.7791-1.7928 g/cc), mass per unit length (0.44\18-0.4446 g/m), tensile strength (5284.3-5580.8 MPa), tensile modulus (274.6-281.3 GPa), and elongation (1.76-1.79%).

Figure 9. The recovered polymers retained their morphological properties. The 002 reflection from virgin and recovered fiber samples.

The panels' properties included panel identification, number of plies (eight plies), thickness (0.1076-0.1140 cm), tensile strength (4656.9-5147.8 MPa), fiber volume (57.16-60.19%), resin weight (0.504-0.542 g/cc), fiber weight (1.020-1.074 g/cc), and resin weight (31.91-34.59%). Properties of the recovered fibers were then compared with the values for the virgin fibers. A photograph of the recovered fibers from a 30 cm by 10 cm panel is shown in Figure 3.

IMPACT OF TEMPERATURE

AND TREATMENT ENVIRONMENT

Figure 4 shows the results for a PMC panel made with an epoxy resin (33 wt.% resin) when treated in air or in nitrogen. When treated in nitrogen (except for minimal air leakage into the reactor) at 6770C, about 85% of the resin was removed in about 10 min. and not much more was removed upon continuing the treatment for an additional 20 min. This is probably due to thermal decomposition of the pyrolytic products resulting in the formation of carbon (soot) that deposited on the fibers. This was more obvious on the inner layers of the fibers where the evolving gaseous pyrolytic products were trapped for a longer period of time. Increasing the treatment temperature to 816 C removed more of the polymeric substrate. The increased air leakage into the reactor at the higher temperature due to natural convection might have assisted in this process.

Table III. Mechanical Properties of Virgin and Recovered Fibers

When a similar panel was treated in air at 677C, over 99% of the polymer substrate was removed in less than 10 min. Further heating resulted in the oxidation of some of the fibers. Therefore, it is necessary to continue the treatment at lower temperatures and/or reduced air flow conditions.

Figures 5 and 6 show the results for a PMC panel made with a polyurethanebased resin (44 wt.% resin) when treated in argon gas and in air, respectively. When treated in argon gas (except for minimal air leakage into the reactor) at 677C, about 99% of the resin was removed in about 3 min. Further heating resulted in the further removal of the residual substrate. However, it appears that after about 6 min., slow oxidation of the fibers has started. This substrate material was mucheasiertoremovethan the epoxy substrate. When the treatment of a similar sample was conducted in air, essentially all of the substrate was removed in less than 3 min. at temperatures greater than or equal to 677C. However, at the higher temperatures, oxidation started early and continued at a slow rate.

These data show that for mixtures of different PMCs, the initial treatment can be carried out at an elevated temperature and in the presence of air to remove the bulk of the polymeric substrate. Further treatment is needed to remove the residual substrate.

CHARACTERIZATION AND EVALUATION OF THE RECOVERED FIBERS

Recovered Fiber Analysis Results

Oak Ridge National Laboratory1 and Hexcel Corporation2 evaluated the recovered fibers to determine their suitability for reuse. Figure 7 shows the results of a scanning electron microscope (SEM) analysis of (a) recovered fibers at 5,000 and (b) virgin fibers at 3,00Ox.1 The rough edge of the recovered fibers is due to cutting the fibers with scissors. The fibers were from a PMC panel made with epoxy resin. Figure 7 shows small amounts of residual material on the surface of the fibers. No other damage to the surface is apparent.

Oak Ridge National Laboratory1 also used x-ray photon spectroscopy (XPS) to determine the concentration (atomic percentage) of principal elements and functional groups found on un- sputtered fiber samples. Samples NBASR-I and NBASR-2 are recovered fibers, and the two IM7 -7 samples are virgin treated and untreated fibers. Table 1 and Figure 8 summarize the results.

The carbon concentration on the surface of the recovered fibers is essentially the same as that for the treated virgin surface. Interestingly, the oxygen concentration for the recovered fibers was higher than that for the treated virgin surface. This is important because complete re-treatment of the recovered fibers may not be necessary. This will save about $0.10-$0.15 per pound of fibers. The nitrogen is lower for the recovered fibers than for the treated virgin fibers. However, the nitrogen concentration is not critical. The trace amount of silicon observed on one of the recovered fibers may be a contamination from the thermal reactor, which was also used for the recovery of glass fibers.

Figure 10. A conceptual process flow diagram.

Further, the same functional groups were found on both the virgin and recycled surface (Figure 7). These groups are: aliphatic carbon and hydrocarbon (C-C and C-H), hydroxyl, ether, aromatic carbon, and single and double nitrogen bonds (C-OH, C-O-C, C=C, C-N, C=N), carbonyl (C=O), and carboxyl and ester (COOH, COOR).

Other characteristics that were evaluated include morphology, density, diameter, and mechanical properties of the recovered fibers. Oak Ridge National Laboratory1 evaluated the morphology of the recovered fibers and compared that with the morphology of virgin fibers (Figure 9) using wide angle x-ray diffraction. The morphology of the recovered fibers was nearly identical to that of virgin fibers of the same type. When the diameter and density of the recovered fibers were determined1 and compared with those of the virgin libers of the same type (Table 2), both the density and the diameter of the fibers were unaffected by the thermal treatment. The mechanical properties of recovered fibers were evaluated by Hexcel Corporation2 and were compared with the mechanical properties of virgin fibers from the same lot (Table III). The recovered fibers retained acceptable mechanical properties.

Table IV. Process Economics (Annual Basis) for Carbon Fiber Recovery from Obsolete PMC Materials

Conceptual Design of the Process

A conceptual design of the process is shown in Figure 10. The PMC scrap is fed into the thermal reactor along with some fuel when needed. For PMC scrap containing about 30% by weight or more polymeric substrate, the process will be energy self sufficient, except for small amounts of energy for the after burner to achieve temperatures high enough to break down VOCs that will be generated during the treatment process. The thermal reactor could be a modified commercially available heat-treating unit. The rest of the equipment is commercially available. The design shown in Figure 10 and the equipment information were used to develop an economic analysis of the process.

The results of the economic analysis are summarized in Table IV. This analysis did not assume any credit for the heating value of the substrate material. A potential payback of less than 2 years is possible.

CONCLUSIONS

The experimental work, limited product characterization, and economic analysis conducted so far on the thermal treatment method for recovering carbon fibers from obsolete PMC scrap indicate that the process is technically feasible and potentially economically attractive. Proper design of the treatment reactor to control the flow of air into the reactor during treatment is necessary in order to avoid oxidation of the carbon fibers. The recovered fibers retained good properties and characteristics and are suitable for short fiber applications.

ACKNOWLEDGEMENTS

This work is sponsored by the U.S. Department of Energy, Office of FreedomCAR and Vehicle Technologies.

References

1. Analysis conducted by Felix Paulauskas, Oak Ridge National Laboratory.

2. Analysis conducted by Mohammad Abdallah, Hexcel Corporation.

Bassam J. Jody, Joseph A. Poinykala, Ji:, and Edward J. Daniels are with Argonne National Laboratory in Argonne, Illinois. Jessica Greniinger is a graduate student who conducted some experiments at Argonne National Laboratory

For more information, contact Bassam J. Jody, Argonne National Laboratory, 9700 S. Cass Ave, Building 362, Argonne, IL69 049 39; (630) 252-4206; fax (630) 252-1342; e-mail bjody@anl.gov.

Copyright Minerals, Metals & Materials Society Aug 2004

More News in this Category