Toward a circular economy: zero-waste manufacturing of carbon fiber-reinforced thermoplastic composites

The study addresses how to enable circularity for carbon fiber-reinforced thermoplastic composites, focusing on retaining mechanical performance through successive recycling and remanufacturing steps. The context is the growing use of composites for lightweighting in aerospace, wind, and increasingly cost-sensitive automotive sectors, where end-of-life regulations (e.g., EU ELV Directive mandating 95% reuse/recovery and 85% recyclability) challenge traditional thermoset composites that are not recyclable. Recycled carbon fiber offers large energy and cost savings versus virgin fiber, with typical recovery of >95% of fiber properties and significant landfill diversion. However, existing recycling and remanufacturing methods often reduce fiber length below critical values and lead to microstructural inhomogeneity, degrading properties. The research tests whether industrially viable size-exclusion (sieving) to control fiber length distribution, combined with common molding processes, can reduce variability and retain tensile performance, and investigates the roles of fiber length and dispersion on thermal and mechanical properties.

Prior work on injection-molded short-fiber thermoplastics shows property losses upon recycling, often linked to progressive fiber length reduction: for 30 wt% CF/PA66, tensile strength decreased ~20% and modulus ~10% after recycling (Colucci et al.). For 35 wt% GF/PA66 EOL radiator parts, tensile strength and modulus decreased ~30% and ~24%, with further reductions on re-recycling due to fiber shortening (Pietroluongo et al.). Incorporating grinds from continuous CF/PEEK into neat PEEK injection compounds dramatically increased strength and modulus at high fiber loadings (Schinner et al.), but short fiber length still limited strength relative to theoretical maxima requiring fibers above the critical length Lc = σf rf / τm. Compression molding of scrap to preserve fiber length has been explored: recycled E-glass/PP woven scrap showed strong strength reductions vs. virgin laminates (Moothoo et al.), while work on shredded CF/PPS highlighted that sieving and mixing parameters control fiber length and clustering (Vincent et al.). The literature identifies two key determinants of recycled composite performance: (1) retaining fiber length above critical length to realize fiber strength and (2) achieving homogeneous dispersion to minimize agglomeration-induced stress concentrations.

Materials: Recycled carbon fiber (rCF) nonwoven preforms (areal density ~184 g m−2; fiber length 38.8 ± 22.2 mm; strength 4426 ± 386 MPa; modulus 206 ± 14 GPa) from a commercial pyrolysis process (Gen 2 Carbon); precision-chopped virgin CF (12 mm; ~4150 MPa; 230–255 GPa; <2 wt% sizing); PPS films (Ryton QC160P, thickness 0.127 mm, density 1.34 g cm−3) for organosheets; PPS pellets (Ryton QA200P) for injection compounding; a commercial CF/PPS compound (Electrafil PPS CF50 3DP, 50 wt% CF) as benchmark. Organosheet manufacturing: Recycled CF preform sandwiched between two PPS films; compression molded at 300 °C, 1 MPa, 5 min; cooled under pressure to form 250 × 250 mm organosheets. Sheets were cut to 100 × 100 mm and stack-molded at 300 °C, 1 or 4 MPa, 5 min, cooled to ≥60 °C under pressure to obtain ~2 mm plates. Scrap (~36%) retained for recycling. Waste preparation and sieving: Scrap was size-reduced by either hand-cutting into 12.7 × 12.7 mm platelets or hammer milling (PelletMasters CF198). Hammer-milled material was sieved through 6.35 mm, #4 (4.75 mm), #8 (2.36 mm), #10 (2.0 mm), #20 (0.85 mm), and #40 (0.425 mm) meshes (two 5 min cycles; flipping top sieve midway). Most recyclate was 1–5 mm. Contaminants were removed via water bath with surfactant (Tomamine AO-14-2), skimming floatables; rinsed on #325 (45 µm) sieve until bubble-free; vacuum dried 4 h at 100 °C. Zero-waste compression molding compounds: Plates (100 × 100 mm, target 2 mm thick) were molded at 5 MPa from: 12.7 mm platelets; #8; #20; and a 1:1:1 mass mixture (12.7 mm, #8, #20). A neat PPS plate was molded as control. Injection molding compounds: Recyclate fractions #4, #10, #40, and <#40 were each dry-blended 1:1 by weight with neat PPS pellets and extruded as filament on a single-screw Filabot EX2 (3 mm nozzle, 300 °C); air-cooled by fan; straight filaments (<300 mm) were pelletized (Filabot Pelletizer). Pellets were dried (150 °C, 3 h) and injection molded into ASTM D638 Type V dogbones using HAAKE MiniJet Pro: 315 °C for composites (305 °C for neat PPS), mold 149 °C, injection 693 bar for 2 s, hold 672 bar for 2 s. A commercial 50 wt% CF/PPS compound was also molded. Wet-laid (WL) nonwovens: Material retained on 6.35 mm sieve was processed via wet-laid papermaking. Due to dispersibility challenges, recyclate was mixed 1:4 by fiber mass with 12 mm virgin CF to target fiber volume fraction similar to compression-molded compounds. Slurry: 20 L water at 40 °C, mixed 20 min (Thwing-Albert tank; Ghossein mixer), with 4 g each of Nalco 8493 (alkyl amine surfactant) and Nalclear 7768 (anionic flocculant). Mats (305 × 305 mm; 302.3 ± 0.7 g m−2) were dried (104 °C, 30 min), stacked with 0.127 mm PPS films, and compression molded at 300 °C at 1 or 4 MPa; a virgin-CF-only nonwoven (areal density 184 g m−2) was molded at 1 MPa for comparison. Thermal analysis: DSC (TA DSC 2500) from 25→325→25 °C at 10 °C min−1 to determine crystallinity Xc via modified rule of mixtures using ΔHf = 88.37 J g−1; Tm and Tc from peak positions. TGA (TA TGA 550) in N2 from RT→600 °C at 10 °C min−1 to obtain fiber mass fraction, T5%, and residual mass at 600 °C. Mechanical testing: Compression-molded plates tested per ASTM D3039: 12.7 mm width × 100 mm length; fiberglass tabs; crosshead 2.0 mm min−1; strain via extensometer over 25.4 mm gauge on MTS 858 (25 kN). Injection-molded dogbones tested per ASTM D638 at 1 mm min−1 on MTS 312 (25 kN); strain via DIC (dual 9 MP cameras, Vic-3D, 2–10 Hz). Two-sample t-tests used for pairwise comparisons of strength, modulus, and strain to failure. Microstructure characterization: Polished cross-sections imaged on Keyence VHX-7000; fiber and void volume fractions quantified by image analysis. Dispersion index (DI) computed as ratio of mean matrix-to-nearest-fiber distance to that of an ideal hexagonal array at same Vf (minimum 1.0 = perfect dispersion). Fractography by SEM (Zeiss Auriga, 5 kV; Au sputter) to assess failure modes and fiber pull-out. Mean fiber length distributions measured from pyrolized specimens (≥2000 fibers/sample).

• Thermal behavior (compression molding): Adding CF increased PPS crystallinity (e.g., Xc from 39.7% in neat PPS to 47–55% with recyclate), slightly increased Tc (to ~224–229 °C), and modestly increased T5% (to ~510–512 °C). Higher molding pressure increased residual mass (indicative of higher fiber content) and T5% differences.
• Thermal behavior (injection molding): Xc generally lower than compression molding for recyclate blends, except the commercial 50 wt% CF/PPS, which showed highest Xc (~55.1%). Tc for injection molded was higher (~240–250 °C) due to cooling-rate effects. Residual mass (proxy for fiber content) was highest for the commercial compound; recyclate blends had significantly lower fiber content than compression-molded plates.
• Tensile properties (compression molded): Zero-waste recyclate plates exhibited significantly lower tensile strength and modulus than organosheet plates (which were anisotropic). Strength and modulus decreased with decreasing recyclate size (fiber length). Representative strengths (MPa): 12.7 mm platelet 82 ± 11; #8 68 ± 8; #20 65 ± 3; mixed 57 ± 17. Organosheets: 1 MPa molding 139 ± 13 (MD) and 268 ± 32 (CD); 4 MPa 109 ± 10 (MD) and 238 ± 7 (CD). Wet-laid: 89 ± 15 (recyclate 1 MPa), 61 ± 8 (recyclate 4 MPa), 57 ± 12 (virgin WL). Zero-waste samples showed notable plastic deformation and localized failure compared with linear-elastic behavior of neat PPS and organosheets.
• Tensile properties (injection molded): Recyclate-based injection-molded specimens achieved higher tensile strengths than compression-molded recyclate and comparable modulus. Strengths (MPa): #4 161 ± 18; #10 161 ± 7; #40 132 ± 9; <#40 127 ± 5; commercial 116 ± 5. Higher strengths stemmed from fiber alignment and fewer defects; commercial compound had higher modulus due to higher Vf but lower strain-to-failure from shorter fibers and agglomerations.
• Variability and sieving: Sieving reduced mechanical property variability and enabled targeted property control (longer fractions for higher performance; shorter for lower). Mixed recyclate modulus was predicted well by rule-of-mixtures (predicted 16.6 GPa vs measured 17.2 GPa, 3.6% error), but strength of mixed and wet-laid composites did not follow simple mixing rules due to dispersion-induced inhomogeneity.
• Microstructure and dispersion: Sieved recyclate compression-molded plates had very low void fractions (<1%), whereas organosheet and wet-laid plates showed higher voids due to reduced pressure and longer fibers hindering flow. Dispersion Index (DI) correlated with strength reductions: #8, #20, and mixed recyclate strength decreased with increasing DI; wet-laid composites had the highest DI (poor dispersion) and low strengths despite long fibers and comparable Vf. Injection-molded composites showed better dispersion than recyclate compression molding compounds; decreasing fiber length tended to improve dispersion but increased fraction below critical length.
• Critical fiber length: With interfacial shear strength 14.16 ± 2.96 MPa and fiber diameter 7.3 ± 0.23 µm, critical fiber length was ~1141 µm. Injection-molded mean fiber lengths (≈128–270 µm) and fine recyclate were below Lc, leading to prevalent fiber pull-out observed by SEM and explaining reduced strengths for finer fractions.
• Economics and sustainability: Mechanical size reduction and sieving are low-energy, low-cost steps (hammer milling 0.03–1.33 kWh/kg; $0.0025–0.1173/kg electricity), with labor dominating costs ($1.71/kg). Mechanical recycling is at demonstration TRL and can be made more viable by length retention and dispersion control. Pyrolysis recovers longer fibers (often >10 mm) but at higher energy (2.9–9.9 kWh/kg) and with CO2 emissions.

The findings show that achieving circularity in CF/PPS composites depends critically on preserving fiber length and ensuring homogeneous dispersion during remanufacture. Sieving provides an industrially practical means to control fiber length distributions and reduce variability, enabling property targeting across product lifecycles. In compression molding with randomly oriented fibers, stiffness is primarily governed by fiber length and volume fraction, while strength is penalized by dispersion-induced inhomogeneity (higher DI correlating with lower strength) and localized matrix-rich regions. In injection molding, fiber alignment boosts tensile strength even when mean fiber lengths are below critical length, though strength still decreases with further length reduction and poor dispersion; higher fiber content (commercial compound) lifts modulus but not strength due to short fibers and agglomerations that trigger premature failure. The interplay between length, orientation, and dispersion explains observed deviations from simple mixing rules for strength, while modulus remains more predictable by rule-of-mixtures when Vf and orientation are known. Economically, low-energy mechanical processing combined with sieving can elevate the value of recyclate; however, without strategies to retain length across cycles, eventual attrition below critical length limits structural performance. These insights directly address the research question by identifying processing levers (size-exclusion, alignment, dispersion control) required to retain performance in remanufactured composites and advance a circular economy.

This work demonstrates a zero-waste pathway to recycle and remanufacture CF/PPS composites using industrially accessible processes (sieving, compression molding, injection molding), achieving: (1) full use of waste material, (2) property retention through control of recyclate geometry and dispersion, (3) reduced variability via size exclusion, and (4) effective reuse of previously recycled carbon fibers. Key contributions include establishing the quantitative roles of fiber length and dispersion (DI) on strength and stiffness, showing that sieving enables targeted properties and that fiber alignment in injection molding elevates strength despite sub-critical lengths. Future research should develop recycling/remanufacturing methods that preserve fiber length at each cycle, enhance fiber alignment during processing, and actively improve dispersion (e.g., low-shear mixing, improved wet-laid dispersion chemistry, compatibilized sizings). Integrating length-preserving recovery (e.g., optimized mechanical processing or hybrid with low-CO2 thermal methods) with robust dispersion control is essential to sustain structural performance and realize a circular economy for fiber-reinforced thermoplastics.

• Material specificity: Results are for CF/PPS systems with particular PPS grades; the commercial compound’s PPS grade is unknown, complicating direct thermal comparisons.
• Fiber content differences: Injection-molded recyclate had substantially lower fiber content than organosheets due to 1:1 blending with neat PPS, affecting comparability.
• Edge effects and sieving bias: Scrap heterogeneity (dry-fiber and resin-rich edge regions) and sieving altered fiber/matrix proportions (e.g., #20 fraction enriched in neat PPS), impacting measured Vf and properties.
• Scale and processing windows: Lab-scale molding and specific pressures/temperatures were used; industrial-scale processing and broader parameter spaces may yield different dispersion and consolidation outcomes.
• Fiber length attrition: While sieving reduces variability, it does not prevent progressive fiber shortening across cycles; performance will ultimately be limited once fibers fall below critical length.