Recycled polyolefin composites: why fiber length decides

In recycled polyolefin composites, glass fiber only reinforces when it stays long. Most of the loss happens inside the extruder, where shear snaps fibers before they entangle into a load-bearing network. A three-rotor extruder that replaces part of that shear with extension held number-average fiber length at 0.35 mm and limited length loss to about 30 percent, against roughly 50 percent on a twin-screw line. At matched fiber content, tensile strength came out up to 74.3 percent higher.

Why mixed polyolefin waste is the hard case

Sorting plastic waste gets discussed as a technology problem. With polyolefins it behaves more like an accounting problem. Polypropylene and polyethylene sit within a few hundredths of a gram per cubic centimeter of each other in density and carry nearly identical functional groups, so density separation and infrared identification both fail to split them cleanly. Each additional point of purity costs another sorting stage, and each stage adds energy and equipment. The pellet that comes out the far end often costs more than the virgin resin it replaces.

Which changes the question. If mechanical recycling cannot pay for the sorting, recovery and reinforcement stacked on top of it, the useful move is to skip the separation and reinforce the mixture as it arrives. Mixed polyolefin scrap becomes the feedstock instead of the obstacle.

The material is also degraded before it reaches the hopper. Waste polypropylene and waste polyethylene have been through several heat histories, photo-oxidation and mechanical abuse, so their chains are shorter than they started, and shorter chains mean less toughness, more brittleness and weaker melt strength. A compatibilizer repairs the interface between the two phases, which is why the elongation figures in this work move so sharply, but it cannot put molecular weight back. Strength and stiffness in recycled polyolefin composites have to come from an added reinforcement.

What a three-rotor extruder changes in recycled polyolefin composites

In glass fiber reinforced composites the bottleneck sits in the flow field, and in whether the glass survives it.

In a conventional twin-screw machine the melting zone loads the material in shear. Glass is brittle in shear, so past a certain stress it fractures instead of flowing, and every fracture shortens it. Below a critical length the fiber stops carrying load at all and turns into an expensive filler that adds density and nothing else. Nobody sees this happen. The fiber entering the die is already shorter than the fiber that went into the hopper.

A three-rotor extruder goes after that one step. Its screw elements are reconfigured so the melting section generates an extensional-shear coupled field rather than a shear-dominated one. Peak shear stress in the melt drops, and the fiber is loaded mainly in controlled extension and alignment. Extension orients a fiber and shear breaks it, so holding the fiber in the first regime is what preserves aspect ratio.

The consequences show up in every measurement in the study, and they line up well enough to put in one table.

Table 1. What changes when the melting zone stops being shear-dominated

The void row looks like an error and is not, for reasons that come up later.

How the composites were made

The base formulation stayed constant across every sample: waste polypropylene, waste polyethylene and a maleic anhydride grafted polyolefin elastomer (POE-g-MAH) at a mass ratio of 56 to 24 to 20. That elastomer is the compatibilizer, and 20 percent is a large dose by the standards of virgin polyolefin compounding, which says something about how badly these two waste streams need bridging.

Everything except the glass was dried at 70 degrees Celsius for 8 hours in a vacuum oven, then mixed for 5 minutes at 600 rpm. The blend went into the hopper, the glass went in through a side feeder with loading set by feeder speed, and the screw ran at 100 rpm. The barrel profile climbed from 120 to 200 degrees Celsius and backed off to 190 at the die. Extrudate was pelletized and compression molded at 200 degrees Celsius and 20 MPa for 5 minutes, then cold pressed at the same pressure for 5 minutes.

Table 2. Formulation and compounding conditions

The comparison is deliberately narrow. The same formulation ran through a conventional twin-screw line at the same screw speed and barrel profile, so the machine is the only variable separating the two series. That narrowness is also the study’s main weakness, and it gets its own section below.

What was measured and how

Table 3. Test and characterization methods

The fiber length method needs a caveat. Burning away the matrix and counting what is left on a glass dish gives a number-weighted picture of what survived the machine, not of what happened to any individual fiber. Counting at least 500 fibers per sample is what makes the number usable.

Fiber length after compounding

Fiber length is the measurement the whole argument rests on. At the lowest glass loading the three-rotor samples averaged 0.35 mm against 0.32 mm for the twin-screw samples. The gap is modest in absolute terms and meaningful in relative terms, since reinforcement scales with fiber aspect ratio and not with length alone. Both series lost length as glass loading rose: about 30 percent from its low-loading value on the three-rotor line, about 50 percent on the twin-screw line.

Table 4. Fiber length retention after compounding

The gap widens with loading for a mechanical reason. Every extra kilogram of glass is another chance for fibers to collide, agglomerate and get dragged through a high-shear region. In a shear-dominated machine, more fiber means more fracture events per unit volume. In an extension-dominated machine it means more entanglement, which is the outcome you want.

That entanglement is the point. Long fibers crossing each other build a three-dimensional skeleton inside the matrix, and that skeleton is the primary load path: stress transfer runs from fiber to fiber through the interface, spreading the load over a volume far larger than the crack tip can reach alone. Short fibers cannot build it. They sit in the matrix like gravel in concrete and stop contributing as soon as the crack finds a way around them.

Tensile strength at matched fiber content

Waste polypropylene alone broke at 22.77 MPa. Waste polyethylene alone broke at 11.24 MPa. Blended at the study’s ratio, the pair broke at 17.86 MPa with an elongation at break of only 10.03 percent, the signature of two immiscible phases pulling apart at the interface. Adding the POE-g-MAH compatibilizer pushed elongation to 81.76 percent with visible plastic deformation before failure. The compatibilizer repaired the interface and left the strength where it was.

Glass fiber then did the work the compatibilizer could not.

Table 5. Tensile strength with glass fiber, TRE against TSE at the same loading

The gain is not monotonic. It peaks at the middle loading and falls back at the highest one, and the low-loading composite at 22 MPa sits marginally under the 22.77 MPa of unmodified waste polypropylene. At that loading the glass is not yet earning its place: the composite weighs more, costs more and is no stronger than the unreinforced waste. Anyone planning a trial should treat level 1 as a floor to clear rather than a target to hit.

The middle loading is where the fiber skeleton becomes continuous enough to carry load across the whole section while the matrix is still able to wet and fill it. Past that point the fibers crowd each other, the matrix struggles to penetrate the bundles, and the marginal fiber adds less than the marginal void costs.

Elongation at break moved the other way throughout, falling steadily as glass content rose, which is what a rigid reinforcement does. The fall was steeper on the twin-screw series, and that is the same story told a second way: shorter fibers with poorer dispersion fail earlier.

Flexural strength rises and impact strength splits

Flexural strength climbed with glass content in both series, for the obvious reason that a higher-modulus phase was being added to a low-modulus matrix and because the fiber network density inside the specimen rose with it. The three-rotor samples sat above the twin-screw samples at every loading, peaking at 62.4 percent higher. The ordering and the mechanism were the same; only the loading mode differed.

Impact strength is where the two processes part company, and it is the most useful single result in the paper.

Table 6. Flexural and impact behavior against glass content

A twin-screw line adding more glass makes the composite more brittle. A three-rotor line adding more glass makes it tougher as well as stronger, which is not the trade most compounders expect.

The explanation is a length threshold. For this system the threshold for effective entanglement sits at roughly 200 to 250 micrometers. Above it, fibers interlock into a network that spreads impact energy across a large volume instead of concentrating it at a crack tip. Below it there is no network to absorb anything, so each added fiber is one more stress concentration and the crack moves faster. On the twin-screw line, higher glass content pushed average length further below the threshold and impact strength fell. On the three-rotor line, length stayed above it and impact strength rose.

Rheology shows the network from the outside

Polymer melt rheology is the cheapest independent check on whether a fiber network exists in a recycled polyolefin composite, because a network changes how the melt responds to being pushed.

Complex viscosity fell with angular frequency in every sample, which is ordinary shear thinning. At equal glass content, the three-rotor samples sat above the twin-screw samples. That difference is not about how much glass is present, since the loading is matched. It is about how much the glass restricts the polymer chains around it, and a tighter, longer, better-connected network restricts them more.

Storage modulus told the same story. It rose with frequency throughout, and the three-rotor samples were above their twin-screw counterparts, with the gap widest in the high-frequency region. Higher glass content raised storage modulus across the board, which is what a stiffer skeleton does: it stores more of the deformation energy instead of dissipating it.

Where the gap was largest carries the most information. The viscosity difference between TRE and TSE grew with glass content and was most pronounced at low frequency, and low-frequency response samples the longest relaxation times in the melt, which is where a connected fiber structure leaves its signature. The two processes differ most at the highest loading because that is where their retained lengths differ most. The mechanical data and the rheology measure the same structure from two directions, which is why neither one needed a separate explanation.

Thermal stability does not move

Thermogravimetry produced the least dramatic result in the study, which is a reason to report it.

Every sample followed the same decomposition path. Degradation began around 400 degrees Celsius. By 490 degrees Celsius the polyolefin matrix was gone. A residual fraction, attributed to inorganic contamination carried in with the waste stream, degraded around 670 degrees Celsius. The decomposition temperatures were essentially identical across the three-rotor and twin-screw series and across glass loadings.

That gives two conclusions. The extrusion route does not damage the polymer’s thermal stability, so the mechanical gains are not being bought with a hidden thermal penalty. The more practical one: thermogravimetry cannot tell a well-dispersed composite from a badly dispersed one, because both pass. A quality plan built on TGA alone would accept a bad batch. Fiber length after compounding and impact strength are the measurements that separate them.

The counterintuitive result: bigger voids and higher strength

Scanning electron microscopy of the fracture surfaces showed something that contradicts the usual reading of a composite micrograph.

The three-rotor samples contained larger defects in cross section, visible as unfilled regions. The explanation the authors give is that the fibers entangle so tightly that the matrix cannot fully penetrate the bundles before the melt solidifies. By the standard heuristic, where void content goes up, strength should go down.

It does not, and the reason matters for anyone reading a fracture surface. A void here is a local packing artifact inside a fiber bundle, while the network it belongs to spans a much larger fraction of the specimen volume than the equivalent network in a twin-screw sample. Strength in this system is governed by how long the fibers are and how well they bond to the matrix, and the three-rotor samples are better on both. Void fraction at this scale is a second-order term that the first-order terms overwhelm.

The fracture surfaces back this up. Fibers in the three-rotor samples were pulled flush with the surface rather than pulled out, with flat brittle breaks on the fibers themselves and drawing and plastic deformation in the matrix around them. Failure was controlled by fiber fracture, not by interfacial debonding. That is what strong interfacial adhesion looks like, and it means the compatibilizer did its job at the glass surface as well as at the PP and PE interface.

The same data set holds a second surprise. The highest glass loading did not give the highest strength: the middle loading reached 35 MPa and the highest fell back to 34 MPa. Reinforcement past the point where the matrix can still wet it stops paying, and the plateau is not an artifact. It is the point where voids and matrix starvation begin to compete with the fiber network.

Practical rules for compounders

• Do not buy sorting purity the application will not use. If the part is a non-structural profile or a pallet, a mixed polyolefin stream plus glass fiber may beat a sorted single-resin stream on cost per unit of strength.
• Size the glass loading to the length your machine can retain, not to the datasheet. A machine that loses half its fiber length needs a lower loading to stay above the entanglement threshold, not a higher one to compensate.
• Add the compatibilizer before the fiber sees the melt. POE-g-MAH lifted elongation from 10.03 to 81.76 percent on the unreinforced blend, and the interface it repairs is the same one the glass has to bond to.
• Measure fiber length after compounding on every batch of recycled polyolefin composites you care about. Ash the sample, count at least 500 fibers, and compare against the 200 to 250 micrometer threshold rather than against the supplier’s nominal length.
• Qualify impact strength separately from tensile strength. The two moved in opposite directions across the two processes in this study, so a tensile-only release test would have passed the worse material.
• Expect a plateau and find it deliberately. Run at least three loadings, because the best result here was in the middle and the highest loading was worse.
• Do not use thermogravimetry as a dispersion check. Every sample in this study decomposed identically regardless of how well the fiber was distributed.

What this study does not settle

The three glass loadings are reported as levels rather than mass fractions, so the absolute dosages behind each data point are not available from the source, and the peak at level 2 cannot be pinned to a specific weight percentage. Anyone reproducing this work will need to establish the loading window from scratch.

The twin-screw tensile strengths are given only as percentage differences from the three-rotor values, not as absolute figures. That is enough to rank the two processes and not enough to compare either one against a commercial glass-filled polypropylene grade.

The specimens were compression molded. Compression molding applies a mild, mostly unidirectional flow history compared with injection molding, where fibers align with the flow front, pack out under high pressure and break in the gate region. Fiber length retention measured on a plaque is an optimistic estimate for an injection-molded part, and the TRE advantage may narrow, widen or invert in a real mold depending on gate design.

One mixed waste stream at one fixed ratio of 56 to 24 to 20 was tested. Nothing in the design shows how sensitive the result is to a different PP to PE balance, a dirtier feed, or a stream with a different contamination profile, and real waste streams vary far more than a laboratory blend does.

Screw speed was held at 100 rpm and one barrel profile was used for both machines. Throughput and specific energy consumption were not measured, so the study says nothing about whether the three-rotor machine produces a cheaper kilogram. That is the number a recycler would actually need.

Fiber length was measured statistically after ashing, which gives a distribution and a mean but not a history. There is no direct observation of where in the barrel the fibers broke, so the mechanism is inferred from inlet and outlet states rather than traced.

Finally, no aging data were collected. Waste polyolefins are already partially oxidized, and a composite that performs well on the day it is molded may lose that performance faster than a virgin-resin equivalent under heat, moisture or UV.

Bottom line

Recycled polyolefin composites made from unsorted waste PP and PE can reach 35 MPa tensile strength with glass fiber, provided the fiber survives compounding long enough to build a network. Whether it survives depends on the extruder’s flow field rather than on the fiber loading. A machine that loads fibers in extension instead of shear retained about 0.35 mm at low loading and lost 30 percent of its length across the loading range where a shear-dominated machine lost 50 percent, and that difference produced gains of 74.3 percent in tensile strength, 62.4 percent in flexural strength and 41.2 percent in impact strength at matched fiber content.

The practical read is narrower than the headline. The optimum sits in the middle of the loading range, voids are not automatically the enemy, and thermal analysis will not tell you whether the batch is good. Fiber length and impact strength will.

FAQ

What makes recycled polyolefin composites different from virgin ones?

The matrix has already been degraded before it is compounded. Waste polypropylene and waste polyethylene have been through several heat histories plus photo-oxidation, so their chains are shorter and their melt strength is lower. Elongation at break for the unmodified blend is only 10.03 percent, against 81.76 percent once a compatibilizer is added.

How much fiber length does a three-rotor extruder save?

At low glass loading it retained 0.35 mm of number-average fiber length against 0.32 mm for a twin-screw line. The gap widens as loading rises. Across the loading range the three-rotor series lost about 30 percent of its length while the twin-screw series lost about 50 percent, which is the difference between clearing the entanglement threshold and falling below it.

Why does glass fiber break inside a twin-screw extruder?

The melting zone of a conventional machine loads the melt in shear, and glass is brittle in shear. Past a certain stress the fiber fractures instead of flowing, and each fracture shortens it. Below a critical length the fiber stops carrying load and behaves like a dense filler, so the damage is invisible at the die but permanent.

What fiber length does glass fiber need to reinforce a composite?

Published work on this system puts the threshold for effective entanglement at roughly 200 to 250 micrometers. Above that length, fibers interlock into a network that transfers stress and absorbs impact energy across a large volume. Below it, there is no network, so each fiber acts as a stress concentration instead of a load path.

How much stronger are these composites than twin-screw versions?

Tensile strength was 19.6 percent higher at the lowest glass level, 74.3 percent higher at the middle level and 30.5 percent higher at the highest level, all at matched fiber content. In absolute terms the three-rotor samples reached 22, 35 and 34 MPa. Flexural strength peaked 62.4 percent higher and impact strength 41.2 percent higher.

Why does impact strength fall on a twin-screw line and rise on a three-rotor line?

Because the two processes push fiber length in opposite directions relative to the entanglement threshold. Adding glass to a twin-screw compound shortens the fibers further below the threshold, so each addition increases brittleness. Adding glass to a three-rotor compound keeps length above it, so the energy-absorbing network gets denser and impact strength rises.

What does POE-g-MAH do in a waste polypropylene and polyethylene blend?

It acts as a compatibilizer at the interface between the two immiscible polyolefin phases. The effect is large and easy to measure: elongation at break went from 10.03 percent without it to 81.76 percent with it, and the material showed visible plastic deformation before failure. It repairs the interface but does not restore molecular weight, so strength still requires reinforcement.

Why do the strongest samples contain more voids?

The fibers entangle so tightly that the matrix cannot fully penetrate the bundles before the melt solidifies, which leaves unfilled regions in cross section. Those voids are local packing artifacts inside a bundle rather than distributed defects. Strength is governed by fiber length and interfacial bonding, and both are better in those samples, so the void content is a second-order effect.

Is more glass fiber always better in recycled polyolefin composites?

No, and the study shows the plateau. Tensile strength reached 35 MPa at the middle loading and fell to 34 MPa at the highest, because the matrix can no longer wet and fill the fiber bundles. Adding reinforcement past that point costs weight and money without adding strength, so at least three loadings should be tested to locate the optimum.

Does melt processing change the thermal stability of recycled polyolefin composites?

No. Every sample decomposed along the same path, starting near 400 degrees Celsius, with the matrix gone by 490 degrees Celsius and residual inorganic contamination degrading near 670 degrees Celsius. The processing route did not shift those temperatures. That also means thermogravimetry cannot distinguish a well-dispersed batch from a poorly dispersed one.

Can a three-rotor extruder handle other waste plastics?

The study only tested two commodity polyolefins, so that claim is an extrapolation rather than a result. The authors argue the mechanism should transfer to other waste polymer blends and to natural fiber or carbon fiber reinforcement, because all of those systems also lose aspect ratio to shear. Untested systems should be treated as hypotheses until fiber length is measured on them.

What are the limits of this evidence?

The specimens were compression molded rather than injection molded, so the fiber length retention figures are optimistic for real parts. Only one waste stream at one fixed ratio was tested, at one screw speed and one barrel profile. No energy, cost, throughput or aging data were collected, so the economics and the durability remain open questions.