One Percent That Decides Everything: Polyolefin Cross-Contamination and the Real Limits of Mechanical Recycling

Researchers at the University of Manchester tracked what happens to polymers during reprocessing across 41 HDPE:PP blends of controlled composition. Their central finding contradicts the industry’s working intuition: it is not large-scale commingling but low-level contamination that does the most damage — and in polypropylene-rich fractions, as little as 1% polyethylene is enough to change the chemical mechanism of degradation.

The study, published in Chem Circularity (Patel, Billington and Shaver, DOI: 10.1016/j.checir.2026.100112, CC BY licence), answers a question that waste management practice has been handling with rules of thumb for decades. The question: at what contamination level does mechanical recycling stop being able to preserve the technical value of secondary raw material? The answer is not a single number but a mechanism — and that mechanism has direct implications for sorting technology, quality control and the modulation of EPR fees.

Why HDPE and PP specifically?

Polyethylene and polypropylene sit so close together in density that classical density-based separation cannot reliably part them. The authors are unambiguous on this point: producing polyolefin recyclate free of cross-contamination requires stringent, expensive sorting — and even then, some contamination should be expected. The practical consequence is that most polyolefin recyclate in circulation is, in reality, a mixture of HDPE and PP.

The two polymers are thermodynamically incompatible: phase separation develops in the melt, and adhesion across the phase boundary is poor. The direct engineering consequence of a weak interface is increased susceptibility to structural failure, including environmental stress cracking (ESC). Until now, the literature has focused almost exclusively on PP contamination in HDPE. The reverse case — HDPE in PP — has remained essentially unexplored, even though PP recycling rates continue to rise and the recovery of trays, tubs and pots has become an explicit priority for separate collection systems.

This is precisely the gap the Manchester work fills.

The experimental design: simulated recycling inside a rheometer

The team prepared 41 blends of controlled composition, in places with single-percentage-point increments — which is what made it possible to identify thresholds rather than merely trends. Extrusion parameters (200 °C, 100 rpm) were deliberately carried over from the authors’ earlier study on polyethylene (Patel et al., Nature Communications, 2024) so that the two polymers could be compared directly.

A key methodological point is that thermal decomposition could be excluded as a confounder: the 5% mass-loss temperature of PP in air is 420 °C, meaning purely thermal degradation is negligible at 200 °C. What happens in the samples is therefore the result of thermo-oxidative and thermomechanical action — shear and oxygen — not temperature.

Degradation was tracked rheologically, through repeated frequency sweeps rendered as Van Gurp-Palmen (VGP) plots and Cole-Cole diagrams. From these the authors calculated the V_deg parameter, a quantitative value proportional to the degree of degradation a polymer has undergone. This parameter is the key to the paper: it not only diagnoses but, as becomes clear later, measures.

Neat PP: minute 195 and the fifth cycle

Before turning to the blends, the authors characterised the reprocessing of food-grade packaging PP on its own. The picture started out as expected: melt flow rate (MFR) rose from 20 to 28 across successive extrusion cycles, while complex viscosity and zero-shear viscosity fell. This is textbook chain scission — the dominant degradation pathway in PP, driven by abstraction of the tertiary methine hydrogens and subsequent β-scission.

Then came the interesting part. The researchers held PP under shear in air for 412 minutes, and at around minute 195 the process reversed direction: complex viscosity stopped falling and began to increase. Curvature in the terminal region of the VGP plot grew, and the previously semicircular, Maxwellian profile of the Cole-Cole diagram distorted into a distinct “tail” — the signature of a new relaxation mechanism in the melt.

In other words, the dominant degradation pathway shifted from chain scission to chain branching and, ultimately, crosslinking. The authors’ explanation is elegant: because temperature and shear force were held constant, the enthalpic contribution (the energy of cleaving covalent bonds) did not change either — so the shift must be entropic in origin. Chain scission produces an ever-larger population of short chains, making it progressively easier for the system to adopt ordered configurations, at which point recombination of the alkyl radicals present in the melt becomes thermodynamically more favourable than further cleavage.

The underlying radical chemistry is striking when quantified. In the presence of peroxyl radicals, primary, secondary and tertiary radicals form in PP at a ratio of roughly 1:14:18, and although tertiary radicals are chiefly responsible for β-scission, all types participate in recombination and disproportionation reactions. Allyl radicals recombine with primary PP radicals approximately one hundred times faster than they undergo disproportionation or chain scission. That factor of a hundred is why, above a critical concentration of allyl radicals or double bonds, crosslinking takes over as the governing mechanism.

One self-critical remark by the authors matters here: real-world recyclate does not undergo degradation of this severity. The practical limit lies elsewhere — meaningful use of recycled PP is unlikely beyond its fifth extrusion cycle.

Phase A: PP in HDPE — the 3% paradox

The authors divided the blends into three morphological regimes: in phase A, HDPE is the matrix with dispersed PP (0–25%); phase B is the co-continuous regime with no majority component; in phase C, PP is the matrix with dispersed HDPE (0–25%).

In phase A, adding 10–25% PP significantly accelerated chain branching relative to neat HDPE, and with it thermomechanical and thermo-oxidative degradation. Small additions (3–8% PP) were less damaging, although degradation was still accelerated compared with 100% HDPE. The lowest extent of branching after extensive processing was exhibited by the blend containing 3% PP.

This is where one of the paper’s most valuable theoretical contributions appears. The literature is contradictory: some reports find that small amounts (<5%) of virgin PP improve the thermal stability of polyolefin recyclate, others that a 10% PP addition stabilises PE blends, and others still that it delays the onset of degradation. Patel and colleagues argue that all of these observations can be valid — just not at the same stage of degradation. At the onset of decomposition, the contaminated blend does appear more stable, but this stabilising effect does not persist under extended thermo-oxidative degradation.

There is a very concrete circular-economy lesson in this. The more loops a packaging material completes, the less applicable are laboratory tests that characterise only the first stage of decomposition. Over the 195-minute simulation, blends containing 3–8% PP ended up with V_deg values below that of neat HDPE — the stabilising effect did not merely disappear, it reversed sign. Equally instructive: 100% HDPE and a 75:25 HDPE:PP blend showed similar extents of chain branching after prolonged degradation despite their radically different starting compositions. Beyond a certain point, the initial recipe loses its explanatory power.

Phase C: HDPE in PP — where 1% already counts

The paper’s most striking result comes from the other side. In a PP matrix, HDPE does not behave like a well-mannered minor phase: even 1% HDPE produces pronounced changes in the viscoelastic response of PP. From this the authors conclude that cross-contamination is more damaging to PP recycling than to the better-studied HDPE case.

The speed of the effect is equally notable. Whereas in neat PP the Cole-Cole “tail” indicating a new relaxation mechanism appeared only around minute 195, in a 75% PP blend containing 25% PE the same feature emerged as early as minute 44 — and the resulting long-chain-branched structure quickly became the dominant relaxation mechanism. PE dispersed in a PP matrix therefore strongly rewrites the melt’s viscoelastic behaviour, whether through recoupling of radical transfer products or via PP-g-PE grafted structures.

The detail of the V_deg(PP) profiles is telling from an engineering standpoint:

• 97% and 95% PP: the process begins with chain scission (minimal change in V_deg), then chain branching takes over, after which both its rate and extent decline and the system returns to chain scission. The explanation: the small amount of HDPE initially promotes branching reactions until the reactive sites are depleted, leaving a residual structure comparatively stabilised against further branching.
• >15 wt% HDPE: chain branching increases substantially, rapidly and persistently. There is no transient stabilisation here.
• Neat PP: the V_deg(PP) profile is flat. This does not mean degradation was absent — it means thermo-oxidative degradation strongly favoured chain scission.

Morphology as the real explanatory variable

The rheological observations were corroborated by SEM imaging, and the picture is consistent. Neat HDPE presents a highly ordered structure; adding PP forms immiscible, crystal-like domains that nevertheless do not significantly disrupt the crystalline ordering of the HDPE matrix. Critically, these domains are not uniformly distributed — which likely explains the uncomfortably wide property scatter so often observed in HDPE recyclate.

Neat PP, by contrast, forms a larger, more open and less ordered network. When HDPE is introduced into this matrix it disperses finely and forms large, immiscible “droplets”. That fine dispersion does several things at once: it increases interfacial surface area, thereby improving the accessibility and diffusion of oxidative radical species through the PP matrix; it provides sites that facilitate chain branching; and it creates micro-scale inclusions that act as stress concentrators.

Worth singling out: the authors report that interfacial deposition of as little as 3% PE is sufficient to disrupt spherulite formation, and that overall crystallinity fell measurably in the blends. Radicals propagate more readily through amorphous regions, particularly at elevated temperatures, which cascades into further chain scission.

One of the authors’ most important general conclusions is that the two phases largely degrade independently of one another. The blend does not decompose as a homogeneous material; two polymers decompose side by side — and what the macroscopic outcome will be is determined primarily by morphology, that is, by the distribution of the minor phase.

Mechanical consequences: which property you are protecting matters

The mechanical testing confirms that contamination does not simply “degrade” the material — it degrades it differently in the two matrices:

The lowest tensile strength — 31.7 MPa — was recorded, as expected, for the 50:50 PP:HDPE blend, which at the same time displayed a higher strain at break than neat HDPE. In addition, both crystallinity and oxidation induction time (OIT) fell substantially in the blends, accompanied by changes in activation energy. On the OIT decline the authors flag an important nuance: the antioxidant additive package typical of packaging-grade feedstocks is consumed during melt extrusion — the depletion of the stabiliser budget is part of the process, not an anomaly.

For packaging design this means the effect of contamination cannot be captured by a single “quality metric”. In an HDPE-dominant bottle or jerrycan application, the loss of strength and modulus is what matters; in a PP-dominant tray or living-hinge closure, it is the collapse of elongation and toughness that renders the product unusable — long before any strength limit is reached.

The measurement problem: DSC’s ±8% and rheology as a QC tool

The paper has an immediately industrially applicable layer as well. Industry typically determines the composition of polyolefin feedstocks by differential scanning calorimetry (DSC): the enthalpic contributions of the components are integrated and proportionally attributed to weight percentages. The method rests on two assumptions — that the phases do not interact, and that the total crystalline fraction is the sum of the crystalline contributions of the individual components.

The authors tested both assumptions, and the result is sobering: the method introduces errors of up to ±8%, particularly for materials containing more than 85% HDPE. As a decision-support tool this is a serious problem in a system where — as the paper itself demonstrates — compositional variations below 3% already influence performance materially. In other words, the uncertainty of the established measurement method is larger than the contamination level that matters.

The authors offer the V_deg rheological approach as an alternative. With the calculation modified to explicitly account for branching formed in the bulk polymer, the V_deg plot is capable of quantifying as little as 1% HDPE in PP — meaning the same tool that diagnoses damage can also serve as a quantitative analytical method. For variable-composition, PCR-type feedstocks, that dual function is valuable in its own right.

What this means in Hungarian and European practice

1. Sorting accuracy is not a fine-tuning question but a first-order one. If 1% cross-contamination is enough to switch mechanisms, then NIR calibration, the sequencing of optical separators and post-sorting steps are not nice-to-haves but determinants of the material’s technical quality. Because the density difference between polyolefins is negligible, only spectroscopic and machine-learning-assisted sorting can deliver results here.

2. PP fractions require separate treatment. In the Hungarian system, PP (tubs, trays, closures, film) has traditionally received second-tier attention behind PET and HDPE. This paper shows that the PP-dominant fraction is in fact more sensitive to contamination than the HDPE-dominant one — so if PP is intended for meaningful, non-downcycling recovery, it needs stricter, not looser, sorting specifications.

3. Deposit return systems change the composition of the residual fraction. Where a DRS removes clean, mono-material PET and HDPE bottles from the separately collected stream, the composition of the remaining polyolefin fraction shifts — typically toward PP and films. By the logic of this paper, that is not a neutral change: the material type that responds worst to cross-contamination becomes more dominant in what remains.

4. Modulated EPR fees have a technical basis. If HDPE contaminated with 3% PP and PP contaminated with 1% HDPE impair recyclability to different degrees, then fee modulation for multi-material, hard-to-separate polyolefin combinations (a PP hinge on an HDPE bottle, a PE liner in a PP closure) is not an aesthetic exercise but the tracking of a measurable material-quality effect.

5. Compatibilisation runs into an economic, not a technical, limit. The authors themselves note that compatibilisation can retain properties and mitigate phase separation, but at an additional cost that further worsens the economics of recycling. At the same time, the growing body of patents on HDPE/PP compatibilisation indicates that industry is actively working on the problem — the issue is simultaneously practically important and technically difficult.

6. The PPWR’s recycled-content obligations turn a quantitative target into a quality question. The Packaging Regulation prescribes rising recycled-content shares from 2030 and again from 2040. Those mandated shares, however, can only be met with recyclate that also satisfies functional requirements. If cross-contamination pushes strain at break or modulus below specification, then the available secondary raw material exists on paper but not in practice. The quality of material streams, not the tonnage collected, will be the binding constraint on compliance.

Limitations of the study

An honest reading requires several caveats. The work used blends prepared from virgin, packaging-grade feedstocks of controlled composition rather than real PCR material — so the effects of pigments, fillers, adhesive residues, paper and label contaminants, and heterogeneous processing histories are absent. The 412-minute simulated extrusion is a substantially harsher loading than real recyclate experiences, as the authors themselves acknowledge. At one point in their discussion of crystallinity they explicitly describe their own explanation as speculative and propose solid-state NMR or X-ray diffraction to resolve the question definitively. Finally, the study is confined to two polymers: in real polyolefin streams, the presence of LDPE, LLDPE and bimodal HDPE grades introduces further variables.

None of this weakens the central conclusion; it narrows the domain of validity. The description of the mechanism — that the dispersion of the minor phase and radical chemistry jointly determine the direction of degradation — holds even if the threshold numbers shift on real material.

Summary

The most valuable contribution of the Manchester work is that it connects polymer chemistry to waste management decisions. Four statements are worth carrying forward:

• Cross-contamination does not degrade material gradually; it switches the mechanism. A shift in the balance between chain scission and chain branching is a qualitative change, not a quantitative one.
• The largest changes occur at low contamination levels. In a PP matrix, 1% HDPE is already measurable; above 15% the result is rapid and persistent branching dominance.
• PP-dominant fractions are more vulnerable than HDPE-dominant fractions — the opposite of how the literature’s and the industry’s attention has been distributed to date.
• Morphology is the decisive variable, because the two phases degrade largely independently; sorting quality therefore translates directly into technical properties.

The practical message is simple, if uncomfortable: in the circular recovery of polyolefins, cost saved on sorting accuracy does not disappear — it is converted into deteriorating mechanical properties, a narrowing application envelope and material streams forced into downcycling. The good news is that the same rheological toolkit that reveals this deterioration also enables quality control at 1% precision. The diagnosis exists; the question is whether sorting facilities and reprocessors will write it into their specifications.

Source:

Patel, A.D., Billington, E.K., Shaver, M.P. (2026): Measuring the impact of cross-contamination on quality in polyolefin blends. Chem Circularity 1, 100112. DOI: 10.1016/j.checir.2026.100112. Open access article under a CC BY 4.0 licence. The research was funded by the Engineering and Physical Sciences Research Council (EPSRC) through the CAFE4DM Prosperity Partnership and the Manufacturing Hub for Sustainable Engineering Plastics; the work drew on facilities at the Henry Royce Institute and the Sustainable Materials Innovation Hub.