2.2.2. Sizing

Sizing plays an important role in the recycling and reuse of carbon fibers. Presently, thermoset matrix materials dominate the market, so that a sizing specifically characterized for these materials is often used but not well-suited for thermoplastics. Production waste and residues, in particular, are usually not pre-treated but directly processed, which can lead to considerable difficulties during recycling and a lack of fiber–matrix adhesion in the final part [18].

Fibers from pyrolysis behave even more critically since the sizing is removed in the pyrolysis process [18]. Generally, the "bare" carbon fibers exhibit good fiber–matrix adhesion with PA6 but are difficult to process with the methods used in this study.

To improve the quality and handling of the fibers recovered by pyrolysis, preliminary sizing trials were carried out by the authors. Different sizing compositions were compared. The fibers were also processed without sizing. Samples were produced to analyze the resulting mechanical properties.

In the initial opening and processing trials of the carbon fiber variants, the need for sizing to minimize fiber damage was demonstrated. Significant reductions in fiber length by sizing carbon fibers could be achieved in the fiber preparation process and in fiber opening and sliver production.

The effects of four different sizing systems on the processing of recycling carbon fibers were compared: two industry-grade sizings (Stantex S 6256 PC from Pulcra Chemicals, Geretsried, Germany and Silastol CF1/CF2 from Schill + Seilacher GmbH, Böblingen, Germany), unsized, and original fibers, i.e., roving sections from production waste. Non-woven fabrics with defined carbon-fiber content were produced in a carding process, and the weight yield was recorded as an indicator of the processability of the fibers. Low carbon-fiber content indicates a high loss of fibers due to fiber breakage and entanglement, suggesting worse processing quality.

As shown in Table 2, sizing is necessary for a carding process; otherwise, a product, or "card web," cannot be produced (weight loss without sizing: 100%). If the fibers are sized manually, however, the losses are kept within limits. Manually sized fibers performed better compared to roving sections. Fibers from rovings need to be separated first, resulting in fiber loss, unlike the "fluffy" structure of fibers from pyrolysis [2–4,18].


**Table 2.** Carbon fiber weight yield after the carding process.

After processing, the card web was doubled and compacted to a non-woven-like structure. Two sample plates for each sizing were then produced: the non-wovens were heated to two different temperatures and pressed using a hydraulic parallel press. Microscopic images were then analyzed to determine the fiber–matrix adhesion. An example is given in Figure 3. The examination of the fiber–matrix adhesion was conducted with a scanning electron microscope (SEM). For evaluation, fracture surfaces were generated at temperatures of −196 ◦C. All tested sizings showed similar results. Since both commercial sizing types showed a comparable fiber–matrix adhesion and resulted in low fiber losses, one was randomly chosen for further tests in this study. Additionally, roving sections were used for sliver formation trails.

**Figure 3.** Microscopic images of the sizing Silastol CF1/CF2 after processing at 220 and 260 ◦C.

The processing of pyrolyzed carbon fibers without sizing proved to be not effective (see Section 1). However, the manual application of an appropriate sizing proved possible. Nonetheless, the processability could not be sufficiently improved, owing to a relatively uneven application. Therefore, roving sections were initially selected as the carbon fiber source for the sliver production.

## 2.2.3. Sliver Production

A sliver is a bundle of fibers that is usually spun into a yarn. In this study the sliver is used as a means to incorporate fibers into the injection molding process via the direct fiber feed module of an IMC. The production steps needed to produce the sliver will be presented in the following.

The first step toward realigning the recycled carbon fibers into a sliver and obtaining an infinite, homogeneous, semi-finished product is the preparation (mixing and opening) of the raw material. The second technological step involves machines that ideally dissolve and orient the fibers, which is called the carding process. Roller cards, like the one in this study, were originally used for processing wool. Wool fibers have low strength and have always been an expensive raw material. For these reasons, processing must be as gentle as possible without major fiber losses. Due to its processing properties, wool fibers are the most comparable textile fibers to carbon fibers.

Figure 4 shows a schematic cross-section of the textile-process chain to produce a sliver out of carbon fibers. The blue line illustrates the material flow through the machine. The general principle of the textile conversion of randomly oriented fibers into an aligned semi-finished product like the sliver is to transport, comb, and draw single fibers with metal pins, or "metallic card clothing." The size of the pins decreases as the process progresses to gradually separate and orient the fibers. The mechanical work required for orientation is applied step by step and more gently than if individual fibers are pulled from a large collection at the beginning of the process. Combing processes are generated through different surface speeds of the working elements.

**Figure 4.** Schematic illustration of the textile process steps used in this study for sliver production (**top**) and pictures of the material feed (**left**), the roller card (**middle**), and the sliver formation (**right**).

In addition to the orientation of the fibers, the carbon fibers were mixed with the thermoplastic matrix fibers (rPA6) to achieve a homogeneous blend within the sliver. The carding process separates and orients the fibers via mechanical forces, which leads to the breaking of the carbon fibers. At the same time, this separation process results in homogeneous mixing, with the matrix and carbon fibers lying directly next to each other, depending on the fiber diameters of the mixing components.

The process starts with a coarse blend of carbon and rPA6 fibers on an opener from the company Dilo Group, Eberbach, Germany (fiber pre-opening in Figure 4). The processability of the pre-opened fiber blend on the roller card resulted in almost trouble-free further processing with acceptable carbon fiber reduction when the fiber blend was opened twice (two passages, Figure 5). A fiber volume content of 55% for the sliver was specified as the mixing ratio. To compensate for possible losses during processing, the fiber flocks were mixed with a volume ratio of 65% carbon fibers to 35% rPA6.

(**a**) material feeding chute

(**b**) 1st opening step

(**c**) 2nd opening after feeding

**Figure 5.** Feeding (**a**) first (**b**) and second (**c**) opening and mixing process of carbon (black) and rPA6 (white) fibers.

The pre-opened and mixed fibers are further opened and mixed in the card's feeding chute Type FBK from Trützschler GmbH, Mönchengladbach, Germany. The feeding chute is followed by the carding machine of the company Maschinenfabrik Memmingen, Memmingen, Germany. The output of the card, the card web, is compacted through a cone and stored in a can by means of a can coiler. The sliver take-up and can coiler have been adapted to the card by the authors.

The resulting sliver is shown in Figure 6. On the right side, the oriented carbon fibers within the sliver and the homogenous blend of black carbon and white matrix fibers (grey) can be seen.

**Figure 6.** Sliver coil (**left**) and texture of the recycled staple fiber sliver (**right**).

#### *2.3. Production of Test Specimens*

To fabricate the test specimen, recycled carbon fiber sliver and polyamide matrix granules were processed via injection molding compounding on an IMC 200-1400 C2, KraussMaffei Group, Munich. The fiber content can be adjusted by the number and fineness of the fibers fed into the direct feed module of the twin-screw extruder of the IMC and by the screw speed *n* and the mass throughput .*m* of the compounder. The theoretical fiber content ϕ is calculated according to Equation (1), as described in the documentation of the IMC provided by KraussMaffei:

$$\text{l.p.} = \frac{\text{x-SDF} \cdot \text{n-k}}{(\text{x-SDF} \cdot \text{n-k}) + \dot{\text{m}}}; \text{IPF} = d\_f \cdot \text{l} \cdot \text{60, } \text{l.l.} = d\_{\text{sl}'} (1 + k) \cdot \pi,\tag{1}$$

where *x* corresponds to the number of rovings (in the case of this study, slivers) used in the feed module, *IPF* is the initial throughput factor, *k* is a correction factor in calculating the actual screw circumference *U* from the inner screw diameter *dsi*, and *df* is the fiber mass per kilometer in tex (g/1000 m).

To produce test specimens with a fiber content of 20% in the IMC, for example, the screw speed was set to 120 min−1, the mass throughput to 4 kg/h, and rovings with a fineness of 2.4 g/m were used. By increasing the screw speed to 160 min−<sup>1</sup> and reducing the mass throughput to 3 kg/h, a fiber content of 30% was achieved.

To further compare the results for recycled carbon fibers, test specimens with glass fiber rovings from Johns Manville, Denver, CO, USA, were also produced. The roving type E-glass PR 440 2400 871 (StarRov ®LFTPlus) was adapted for processing with polyamide matrices with a silane sizing.

#### *2.4. Characterization of Test Specimens*

The characterization of fiber-reinforced plastics was carried out with classical methods to determine mechanical properties. The tensile tests, according to ISO 527, were performed by a universal testing machine type 1455 from Zwick GmbH and Co. KG, Ulm, Germany on the injection-molded basic specimens of type 1A. The bending properties were determined on the same machine, according to ISO 178. Seven to ten test specimens per mixing ratio were tested and statistically evaluated.

Tests such as the determination of fiber content (ISO 1172 method A), fiber length (ISO 22314), and fiber–matrix adhesion were also carried out. Due to the limited resources, complexity, and time consumption involved in carbon fiber length determination in an injection molded part using a solvolysis method, only two test specimens were measured: specimen one with a fiber content of 27 wt-% ( . *m* = 3 kg/h, *n* = 300 min−1) and specimen two with a fiber content of 31 wt-% ( . *m* = 3 kg/h, *n* = 350 min−1). A specimen made of virgin fiber roving with a fiber content of 26 wt-% was also measured for comparison.

The single fiber length of the sliver was determined according to the standards DIN 53,803 and DIN 53808-1. For this purpose, 150 individual fibers were randomly and carefully removed from the fiber sample by hand and measured under a magnifying glass using a ruler.

#### *2.5. Summary of the Production Steps*

In the following section, the test evaluation is presented. Below is a short summary of the production process. The production cycle basically contains two steps:

