Most sustainability claims in composites are marketing. Thermoplastic composite recycling is not, at least in its physical basis. The matrix is held together by entanglement and crystallinity rather than a permanent chemical network, so it melts and reshapes. A crosslinked thermoset does not. That difference is real, and it is exactly why the claim gets pushed past what the material can carry.
Suppliers call a polymer “infinitely recyclable”. A buyer hears that a scrapped part can come back as an equivalent part, forever. It cannot, at least not in any sense a designer would accept. What comes back has a different molecular weight distribution, a different fibre length distribution and a different interface from what went in. The chemistry allows reprocessing. It does not promise equivalence.
Below: which half of the claim holds, what actually degrades on each pass, and the questions that decide whether a regrind-containing part is safe in your application.
Where the “infinitely recyclable” claim comes from
The starting observation is correct. Heat a PP, PA6, PA12, PC, TPU, PPS or PAEK matrix above its melting or softening range and it flows. Cool it and it goes solid. No cure reaction is being reversed, so in principle the cycle repeats. That is why thermoplastics are called remouldable and thermosets are called set.
“Infinitely” is the word doing the damage. Reprocessing is a thermal history, and thermal history accumulates. Three variables set the practical ceiling: how many heat cycles the polymer has already seen, how much shear and oxygen it met during those cycles, and whether the fibre architecture survived the size-reduction step required to feed it back into a machine.
There is a second trap. A continuous-fibre laminate and a compression-moulded part do not recycle into the same product. Cut or mill a woven or unidirectional laminate to feed an extruder and the fibre is no longer continuous. What comes out is a short-fibre compound. That is a downgrade in architecture even when the polymer is intact, and it is the most common misreading of the word “recyclable” in a procurement meeting.
What degrades, how fast, and why the three losses differ
Three things degrade at different rates, and they have to be assessed separately. The mechanisms below are our assessment of how these polymer families behave under repeated thermal processing. They are not a statement of fact about any specific grade, and actual retention has to be measured per grade and per batch.
Matrix molecular weight comes first. Each pass through a melt adds chain scission and, where oxygen is present, thermo-oxidative branching. The signature is a fall in molecular weight and a rise in melt flow index. To a moulder that looks convenient, because the material flows more easily. The same shift reduces ductility, weakens weld lines and makes the melt more sensitive to moisture. Polyamides and polycarbonate are the least forgiving families here, because hydrolysis compounds the thermal load: residual moisture at processing temperature attacks the chain directly, so drying discipline decides how much of the loss is thermal and how much is hydrolytic. PPS and PAEK tolerate far more thermal history. That tolerance is part of what you pay for.
Fibre length and sizing degrade next, and faster than most people expect. Grinding and re-compounding break filaments and strip the sizing that bonds them to the matrix. Reinforcement efficiency in a short-fibre compound depends strongly on aspect ratio, so a moderate loss of length produces a disproportionate loss of stiffness. Once the sizing is gone too, the fibre stops wetting out and stops bonding, and the material behaves like a filled polymer instead of a reinforced one. A data sheet usually hides this loss, because fibre weight fraction can stay constant while fibre length collapses.
Contamination is the third, and the one that is hardest to see in a specification. A regrind stream carries whatever the original part carried: mould release agent, paint, foam core, metal inserts, machining dust, and often more than one polymer grade. Small fractions of an incompatible polymer form domains that start cracks. Which is why a recycled content figure by itself says very little. Regrind from a clean, single-grade, in-house scrap stream and regrind from mixed end-of-life parts can both be labelled 30% and still be different materials.
Crystallinity adds an effect that shows up on the shop floor rather than on the data sheet. Recycled polymer often nucleates differently, so cycle time, shrinkage and warpage move. A part that held tolerance in virgin material can drift out of it when the regrind ratio changes, even while tensile strength still looks acceptable.
Recyclability and recycled content are not the same claim
The two terms get used interchangeably in marketing and they answer different questions. Recyclability is a design property: can this part be economically disassembled, sorted and reprocessed into something useful? Recycled content is a mass balance: how much of this part came from recovered material? A part can be highly recyclable and contain no recycled content. It can also contain recycled content and be impossible to recover at end of life.
Commercial pressure is moving on both fronts at once. In October 2026 an EU-funded project reported that plant-fibre reinforced thermoplastic parts had reached series production on the order of ten thousand vehicles, with a reported 8% weight reduction per part and confirmed end-of-life recyclability (source: CINEA, published 13 October 2026). Recyclability is starting to be verified at production scale rather than asserted in a brochure.
| What the claim says | What it actually tells you |
|---|---|
| “Infinitely recyclable” | Reprocessing is possible. Equivalence is not promised, because molecular weight and fibre length both move with each pass. |
| “This part is recyclable” | A design property about disassembly and sorting. It says nothing about how much recovered material is inside. |
| “30% recycled content” | A mass balance only. Clean in-house scrap and mixed end-of-life regrind can carry the same label. |
| “Same grade, so same properties” | Fibre weight fraction can stay constant while fibre length collapses, and the data sheet will not show it. |
| “Recycled always means weaker” | Depends on the stream and the cycle count. Measure retention at the intended ratio instead of assuming either direction. |
For an engineer, the practical consequence is simple. “Recyclable” belongs in the design review. “Recycled content” belongs in incoming inspection. Neither belongs in a purchase order as a single adjective.
How to specify a regrind-containing part without guessing
Six questions separate a defensible specification from a hopeful one. Ask for written answers tied to the grade, not to the polymer family.
- Which polymer grade, and what is its melt flow index as supplied, not the virgin reference value?
- How many thermal cycles are documented for this stream, and was each one mechanical or solvent based?
- What is the fibre weight fraction and the fibre length distribution after compounding?
- What retention is measured against the virgin grade for tensile, flexural and notched impact, and on which specimen geometry?
- What contamination limits apply, and how is incoming regrind screened against them?
- What regrind ratio is allowed in the part, and does that limit change for thick sections or cosmetic surfaces?
Design rules follow the same logic. Regrind is easiest to justify where the part is stiffness-driven rather than strength- or fatigue-driven, where sections are thick enough that fibre length matters less, where the surface is not cosmetic, and where the duty cycle is not sealing-critical. Keep virgin material for fatigue-loaded structure, for pressure boundaries, and for anything that must weld to a controlled surface.
If the polymer is a polyamide, polycarbonate or TPU, put drying in the specification. Moisture limits before processing are not housekeeping; they are a property of the finished part. The same goes for regrind storage, since an open bin in a humid shop undoes a drying step that took hours.
A selection path that keeps the claim honest
Start from the function, not the percentage. A part that holds pressure, takes repeated load or must survive impact should be specified on properties measured at the intended regrind ratio. If nobody has measured them, that measurement is the first task, not the last. A part that is mainly a stiffener, a panel or a non-structural housing can often absorb regrind without a measurable penalty, provided the grade and fibre fraction are controlled.
Then match the material system to the process you actually run. Extrusion and compression moulding of short-fibre compounds take regrind far more readily than processes that depend on continuous fibre alignment. In a laminate, decide early whether recyclate goes into the core, into a non-structural layer, or nowhere, and write that decision on the drawing instead of leaving it to the shop floor.
Keep the vocabulary precise in your own documents as well. “Reprocessed once, 25% regrind, properties measured at that ratio” survives an audit and a customer question. “Recycled” invites both.
Where we can help is at the front of that chain. Our thermoplastic prepreg range covers CF/PP, CF/TPU, CF/PC, GF/PC, CF/PA6, GF/PA6, CF/PA12, CF/PES, CF/PPS and CF/PAEK, with carbon fabric from 3K to 24K tow, areal weights of 50 to 600 g/m² and widths up to 1270 mm, plus machinable laminates in 1.0, 1.5, 2.0 and 3.0 mm. If you are working through a regrind ratio or a recyclate specification and need a grade-level answer instead of a general one, send us the part function and the process. We will tell you what is confirmable per grade.