Characterization of Fiber Dust Resulting from Recycling of Carbon Fiber-Reinforced Thermoplastics (CFRP) and Their Cell Toxicity

Carbon fiber-reinforced thermoplastics (CFRP) have excellent specific strength and rigidity, which has made them a popular material for lightweight con-struction. The growing demand for fiber-reinforced plastics (FRP) leads to the problem of the sustainable handling of FRP at the end of their life cycle. The aim of the research project was to gain knowledge about the shredding of FRP concerning the optimal machine and process design of the shredding process and the possible formation of harmful, inhalable dust fractions and WHO fibers. Toxicity should be investigated at the cellular level. The investigated shredding parameters influence the amount and length of fiber dust produced, both when shredding with a cutting mill and when shredding with a single-shaft shredder. In all investigations, an increasing rotational speed leads to an increase in the fiber dust mass or the fiber concentration. The proportion of short, respirable fibers increases, but raising the are only a minor health hazard. Formally, the detected carbon fiber (CF) fragments meet the fiber definition of the legislator. However, carbon fibers currently have no specific limit value.


Introduction
Carbon fiber-reinforced thermoplastics (CFRP) are increasingly used. With the growing need for these materials, the problem of sustainable handling of residual materials arises. Material recycling systems may enable sustainable handling and use of this resource. Recycling is a legal requirement in the EU [1]: a manufacturer is obliged to prepare an exposure assessment for a product over the entire life cycle.
A sustainable recycling process is mechanical recycling [2] [3], where the whole composite material can be reused as regrinding and further processed into a recyclate. Therefore, the composite is shredded followed by a compounding or extrusion process and granulated into a short fiber-reinforced regrind. Several studies demonstrated the potential to reuse the recycled thermoplastic fiber-reinforced material [4] [5]. Recycling requires shredding to a suitable size by single or multi-shaft shredder to a particle size of <50 mm or by fast-rotating cutting or hammer mills to a particle size < 10 mm for further processing [6], e.g. in the injection molding or extrusion process. During the shredding of thermoplastic fiber-reinforced plastics inhalable dust including fibers in the micrometer range as well as splitters with WHO fiber characteristics can occur. The possible formation of WHO fibers requires a high level of occupational safety. Inhaled asbestos fibers for example are classified as human carcinogens [7] [8]. The toxicity of such fibers is largely determined by their inflammatory properties. Other bio-persistent, inflammatory fibers are also suspicious due to similarities in structure and biological effects [9] [10]. The same is feared for new fibrous materials. Fibers of different geometry and bio-persistence must therefore be assessed carefully regarding possible toxic effects. However, there are no toxicological studies on the dust and fibers that occur when shredding fiber-reinforced thermoplastics.
Toxicity in animal experiments is rather low, following inhalation [11] [12] [13] [14], or intratracheal injection [15]. However, the bio-persistency seems to be pronounced [11] and inhalation studies can underestimate toxicity since longer fibers cannot be inhaled by the rat. The migration of inflammatory cells into the lungs is a reliable toxicological endpoint for the toxicity of particles [16]. This can be measured by a so-called broncho-alveolar lavage (BAL, flushing of the bronchi and lungs), which is equally stressful for animals and humans. Moreover, such an experimental procedure is far too complex for extensive Journal of Materials Science and Chemical Engineering comparative investigations. Accordingly, in vitro test that depicts the inflammatory effects of particles and fibers is needed. In this research project, we, therefore, applied the particle-induced cell migration test (PICMA) [17] [18]. This approach assumes that predominantly inflammatory fiber effects lead to malignant cell changes and cancer. In addition, cytotoxic effects were examined.
This research project aims to achieve knowledge on fiber dust formation during the shredding of fiber-reinforced thermoplastics. One focus is on occupational health and safety, and a second focus concentrates on optimizing the shredding process.

Material
For these investigations, polycarbonate with carbon fiber-reinforced (PC-CF) organic sheets Tepexdynalite210fr-C200(x) 45% from Bond Laminates GmbH (Lanxess, Germany) have been used. The sheets have a thickness of 1 mm and are cut into squared pieces, whose size depends on the respective shredder used.

Shredding Process in the Cutting Mill and Analysis of the Fiber
Length in the Ground Material For the investigations in the cutting mill SM300 (Retsch GmbH, Germany) equipped with a 10 mm screen is used while the rotational speed, and the feedstock size are varied. The organic sheets are cut into pieces of 30 × 30 × 1 mm or 15 × 15 × 1 mm size to analyze the effect of different feedstock sizes. Samples of 50 g each are prepared with the respective size and are then cut with the cutting mill for 1 minute each. The rotational speed is varied from 800 rpm, 1500 rpm to 2500 rpm. The ground material is sieved for 20 minutes to receive the particle size distribution and to separate the fiber dust from the rest of the material. The matrix of the fraction 0.125 mm is ashed to retain the pure fibers. The length of the fibers is analyzed using the scanner Epson Perfection V750 Pro (EPSON Deutschland GmbH, Germany) and the software FiVer (SKZ-KFE gGmbH, Germany). The results are shown in the particle size distribution ( Figure 1) and the fiber length distribution ( Figure 2).    were removed by centrifugation with 400 g for 5 minutes and particles were removed by centrifugation with 15,000 g for 10 minutes at room temperature. The supernatants were immediately used for the migration tests.

Chemotaxis Assay
Cell migration was investigated according to Boyden [20], with the modifica-

Quantitative SEM Analysis of WHO Fiber Fragments
In order to determine the number of airborne fibers that meet the WHO convention in terms of particularly hazardous fibers, air samples have been collected by using a FAP fiber sampling device (GSA Messgerätebau GmbH, Germany) at a distance of approximately 35 cm from the feeding shaft of the shredder. A gold-coated polycarbonate filter with a diameter of 37 mm and a pore size of 0.8 µm has been used for these investigations. The air flow is 2.0 L/minute and the collection time is 15 minutes in each case; only for the tests with high rotational speed and small screen size has the sampling time been reduced to 10 minutes. The fibers on the gold-coated filters have then been counted by using a scanning electron microscope (SEM) S-2300 (Hitachi, Ltd., Tokyo, Japan) including an energy dispersive X-ray analysis (EDX) and the resulting fiber concentration has been calculated according to the guideline of the DGUV [21]. For this purpose, a Kevex Si (Li) detector system (eumeX Instrumentebau GmbH, Heidenrod, Germany) at 1000× magnification was used.

Effect of Shredding Parameter on Fiber Dust Formation
The following figures show the effect of different rotational speeds and feedstock sizes on the particle size distribution of the ground material ( Figure 1) as well as on the fiber length distribution in the fiber dust after the ashing process ( Figure  2). It can be seen, that most of the particles (at least 65%) have a particle size bigger than 1 mm. The relative frequency decreases with decreasing particle size but slightly increases again for the particles < 0.125 mm. For every parameter combination, the relative frequency is higher compared to the fraction > 0.125 mm. The fiber length distribution shows a small amount of fibers with a length < 30 µm, whereas the fibers with a length of 30 -60 µm provide the highest frequency. With increasing fiber length, the fiber length distribution shows a decreasing trend.
No significant effect of the feedstock size on the particle size distribution of the ground material can be observed (Figure 1). In the fiber length distribution ( Figure 2), a tendency of less short fibers (< 60 µm) can be seen when shredding the organic sheets with a smaller feedstock size. This is in accordance to the percentage of fibers in the short particle fraction in Figure 3. Here, shredding the pieces with a feedstock size of 30 × 30 mm leads to a higher percentage of fiber mass, except for the high rotational speed of 2500 rpm. Material with a smaller feedstock size needs fewer cutting steps until it is small enough to leave the cutting chamber through the screen. This could explain the smaller amount of short  a smaller feedstock size comes along with a higher number of pieces for the 50 g sample. Therefore, the particle-particle interaction in the cutting chamber is enhanced, which could be a reason why the favorable effect of the smaller feedstock size can just slightly be seen in the fiber length in Figure 2. In a similar investigation, Shuaib and Mativenga [22] also found no significant effect of the feedstock size on the fiber length and in [23] it was found that the effect of the feedstock size on the particle mass and the fiber length is material dependent.
Rising the rotational speed during shredding leads to a higher number of small particles (<0.125 mm) independent of the feedstock size ( Figure 1). This can be seen in the other fractions as well. Polycarbonate is a thermoplastic material and therefore provides viscoelastic material properties. With increasing load speed, the capability to creep as well as the stress relaxation is reduced so that the material properties become more brittle [6]. A more brittle fracture behavior leads to uncontrolled crack propagation [24] and thus a broader particle distribution with a large amount of small particles [25]. In brittle fracture, the crack propagates through the material, so that uncontrolled fracture behavior is dominant. This is the reason for the increased quantity of fine particles with higher rotational speed. In Figure 2, no clear effect of the increased rotational speed on the fiber length distribution can be seen. Nevertheless, the percentage of fibers in the fraction < 0.125 mm increases with rising rotational speed ( Figure 3), which is in accordance to the findings from the particle size distribution and is a result of the more brittle breaking behavior. This suggests that a higher rotational speed increases the number of fine particles in the ground material, but has no effect on the fiber length distribution in the dust. The absolute number of short fibers is enhanced, but the relative frequency is not significantly altered.
The investigations on shredding the PC-CF organic sheets with the single screw shredder enable the analysis of the airborne particle concentrations. The results are represented in Table 1 The sample, which is shredded at high speed and with small screen size, shows increased particle concentrations in the nanometer scale compared to the shredding with a large screen size. Reducing the screen size leads to a smaller maximum Journal of Materials Science and Chemical Engineering Figure 3. Percentage of fiber mass after ashing the particle fraction < 0.125 mm on the total ground material mass over the rotational speed while shredding. The two different feedstock sizes are presented.  Initial investigations were carried out with the cutting mill ( Figure 2) in order to determine the conditions for the large-scale test with the single-shaft shredder (Table 1). Increasing the rotational speed and feedstock size did not lead to any changes in the toxicity using the cutting mill ( Figure 4).
Since cytotoxicity remains unchanged for all examined parameters and rotational speed, feedstock size and screen size were the main influencing factors concerning the particle quantity, these were varied for the large-scale investigation using the single-shaft shredder. Even under these conditions, the Alamar-Blue test showed no increased cell toxicity, even at very high dust concentrations ( Figure 5). The fluorescence is plotted against the amount of CF particles (µg/cm 2 ) exposed to the macrophages. Cell toxicity can only be determined for the highest added quantity of 500 µg/cm 2 . As this is a very large amount, the observed toxicity is considered to be very low. The fluorescence is plotted against the amount of CF particles (µg/cm 2 ) exposed to the macrophages. No cell toxicity is evident up to and including the highest level tested.

Investigation of the Induced Cell Migration by Particles Obtained from the Large-Scale Test with the Single-Shaft Shredder
Rotational speed, feedstock size and screen size were the main influencing factors concerning the particle quantity and the AlamarBlue did not show any significant cytotoxic effects, regardless of how they were obtained. Particle induced cell migration was therefore investigated for particles that were obtained for different speed and feedstock size in the large-scale test with the single-shaft shredder ( Table 1). Particle induced cell migration was similar for all examined parameters. The PICMA, therefore, shows no increased effects, even at very high dust concentrations as well as the AlamarBlue test ( Figure 6).
Comparison with the historical control ( Figure 6 and Figure 7) shows that the induction of cell migration by carbon fibers is very weak and comparable to that of inert particles such as barium sulfate. Consistent with the results from the cell toxicity test, the migration test provides no evidence that the inflammatory or toxic effects of carbon fibers are significantly influenced by the type of shredding.

Quantitative Analysis of WHO Fibers Resulting from Shredding
PC-CF with a Single-Shaft Shredder High fiber concentrations have been counted for all shredding parameters, as well as a high proportion of WHO fibers (Table 2). Figure 8 presents an example of the collected fibers on the gold-coated core pore filter after shredding with different parameters.
EDX analysis of the collected fibers detected only the element carbon in addition to the sputtering material gold (Figure 9).     The results support the findings from the investigations in the micro-and nanoscale range that higher dust concentrations are produced during shredding at high rotational speed. The additional reduction in screen size significantly increases the amount of fibers. Furthermore, it is shown that the number of WHO fibers is not affected by the shredding parameters. This is in accordance with the results from the toxicological studies (Figures 4-6). Despite the different total fiber concentrations, no influence of the shredding parameters on toxicity can be detected here, since the potentially toxic WHO fiber fragments do not differ significantly in number.
Formally, the detected carbon fiber fragments meet the fiber definition of the legislator. However, carbon fibers currently have no specific limit value. Prevention is based on the provisions of TRGS 521 [26]. Thus, a protective measure concept takes effect that is based on the decision concentrations of 50,000 and 250,000 F/m 3 (no limit values). In the future, the adaptation to this protection concept should be based on the actual health risk from carbon fiber dust.

Conclusions
Irrespective of the shredding parameters, fiber fragments of different dimensions are produced, but they are of low toxicity and have very little impact on cell migration. This indicates minor inflammable effects of the dust formed after inha-Journal of Materials Science and Chemical Engineering lation. This is in accordance with the results of the animal experiments that report rather low toxicity. As a result, the optimization of the shredding processes should mainly focus on the amount of particles and their dimensions. The biological data indicate as well that the WHO fibers resulting from the shredding are only a minor health hazard. Most of these fibers are probably too thick to have any appreciable toxicity. Slight differences in the induction of cell migration are visible, but to an extent that it is not important to take them into account for the risk assessment. However, a pronounced bio-persistency of inhaled CFRP fragments must be assumed. Therefore, exposure to CFRP dust should be minimized in terms of preventive health protection.
In summary, it can be stated that during shredding fiber-reinforced thermoplastics, inhalable fiber fragments can arise, which probably do not pose any health risks that go well beyond that of granular dust. The proportion of these fibers can be reduced by a low speed or a large screen size. Especially when cutting carbon-fiber-reinforced thermoplastics and materials with a brittle matrix, special safety precautions should be taken, even if the dust produced shows only low cell-toxic effects in the tests.