Polyethylene terephthalate (PET) is a thermoplastic condensation polyester of terephthalic acid and ethylene glycol, typically copolymerized with small amounts of isophthalic acid or diethylene glycol to suppress quiescent crystallization in bottle preforms. Bottle-grade resin is characterized by an intrinsic viscosity of 0.80–0.84 dL/g as measured by ASTM D4603-18 in a 60/40 phenol/1,1,2,2-tetrachloroethane solvent at 30°C; density is 1.33–1.40 g/cm³ per ASTM D1505 or ISO 1183-1, and the melt endotherm appears between 240–255°C under ASTM D3418. The recycled feedstock chain is summarized by the phrase PET feedstock for recycling: bottles, flakes and the collection gap.

Post-consumer PET bottle feedstock is not homogeneous. Bale composition, sorting technology, and regional collection architecture determine whether the bottle is converted to food-contact flake, non-food flake, fiber, or landfill. The collection gap is not only a mass-balance deficit; it is also a quality gate failure in which bottle volumes are displaced by non-bottle PET packaging, opaque containers, PVC sleeve labels, and residual liquids. Bottle-to-bottle flake lines therefore require a feedstock that can be processed into rPET pellets with adequate molecular weight, controlled color, and contaminant concentrations below melt-defect thresholds.

What Flake Quality Gates Prevent Hydrolytic and Contaminant Failure?

Flake entering extrusion must pass four operational gates: intrinsic viscosity, residual polymer contamination, moisture, and fines. Intrinsic viscosity is the primary molecular-weight indicator because post-bottle flake with IV below 0.70 dL/g generally requires extended solid-state polycondensation to reach bottle-grade values. The measurement is performed by dissolving flake in a 60/40 phenol/1,1,2,2-tetrachloroethane mixture and following ASTM D4603-18 or ISO 1628-5. A low IV gate is not a single reject point; rather, lower IV shifts the SSP residence time, temperature, and vacuum demand upward, reducing hourly throughput on continuous SSP units.

The PVC limit is especially severe because PVC degradation begins at PET melt-processing temperatures and generates hydrogen chloride, which accelerates ester bond cleavage. Near-infrared sorting arrays operating in the 1100–2100 nm region are used on whole bottles and flake to remove PVC and PLA; however, dark colored and carbon-black-loaded fractions can be misclassified unless the sorter is equipped with an additional color camera or X-ray channel. Melt filtration is a downstream safeguard but cannot correct gross PVC contamination. Candle filters or screen changers rated at 32–100 µm remove rigid particles, yet dissolved or partially degraded PVC can pass the filtration stage and deposit on extrusion dies.

Post-consumer bottle collection is not a uniform feedstock pool. Deposit return systems generate bales with lower adhesive contamination and fewer non-PET closures; single-stream material recovery facilities produce PET bales that contain thermoform trays, opaque PET, and multilayer barrier bottles at variable rates. The collection gap is therefore a two-component deficit: uncollected bottle mass and collected but nonroutable mass. Material recovery facility audits show that bale yield drops when optical sorter ejection rates rise in response to PVC sleeve labels and opaque packaging; published data for specific municipal configurations is limited, but the effect is consistently observed when feedstock complexity increases. Bale wire, ferrous closures, and glass shards introduced during collection also reduce granulator blade life and increase screen wear in wash lines.

Sink-Float Separation Limits and the PVC Density Overlap

Sink-float separation is effective for removing polyolefin closures and labels because PET sinks in water at 1.0 g/cm³ while polyethylene and polypropylene float at 0.90–0.96 g/cm³. The limitation is that PVC and PET occupy nearly the same density band: rigid PVC is typically 1.38–1.40 g/cm³, overlapping the upper PET range. Water-based separation cannot isolate PVC from PET flake with sufficient precision; therefore, PVC removal must occur before granulation by whole-bottle NIR ejection or after granulation by flake NIR or X-ray fluorescence. A secondary issue is filled polyolefin: talc-filled polypropylene caps can exceed 1.0 g/cm³ and sink, requiring downstream flake sorting. Multi-layer barrier bottles may contain polyamide or EVOH layers that alter flake density and can survive hot washing if delamination is incomplete.

Flake size reduction is normally performed with wet granulators using screen apertures between 10 mm and 14 mm. Undersized fines from brittle PET and surface dust are removed on vibratory screens; excessive fines increase moisture uptake and create dust-laden vacuum streams. After granulation, cold-wash or hot-wash sequences remove water-soluble contaminants and label adhesive residues. If the label adhesive is not removed, flake can carry paper fiber and adhesive particles into the extruder, leading to pressure build-up on the melt filter. Trial runs on twin-screw extruders show that screen-change frequency increases when hot caustic washing is bypassed, a signal that downstream filtration alone is not a substitute for upstream washing.

When Hot Caustic Washing Is Omitted, Flake-to-Pellet Extrusion Encounters Three Constraints

If hot caustic washing is omitted or operated outside its required window, three separate constraints emerge in melt conversion. First, residual label adhesives and paper fibers are not removed, increasing pressure drop across melt filters and shortening screen life. Second, organic residues and entrapped volatiles are not liberated before extrusion, so vacuum devolatilization must handle a higher volatile load. Third, flake moisture is less uniformly reduced before drying, because surface-contaminated flake holds water in adhesive films and fiber pockets; this creates local hydrolytic degradation in the extruder even when average flake moisture appears acceptable. Hot caustic washing systems typically operate with sodium hydroxide concentrations of 1.5–3.0 wt%, bath temperatures of 80–90°C, and residence times of 15–20 min; the operational boundary is narrow because excessive caustic concentration or temperature can hydrolyze the PET surface and raise fines generation.

Extrusion of bottle-derived PET flake is conducted on co-rotating twin-screw extruders with L/D ratios from 32:1 to 44:1. Vacuum devolatilization is applied at −0.8 bar to remove residual water, glycols, and volatile degradation products. The melt temperature is normally held between 275°C and 285°C; higher temperatures accelerate acetaldehyde formation and thermal degradation, while lower temperatures can leave unplasticized flake domains if feed compression is insufficient. Preform manufacturers specify acetaldehyde levels below sensory thresholds because acetaldehyde migrates into packaged water and affects flavor; resin specifications commonly require acetaldehyde below 5 ppm for water-grade rPET.

Solid-State Polycondensation and the Lower IV Boundary in Bottle-to-Bottle Feedstock

Solid-state polycondensation is used to rebuild intrinsic viscosity after extrusion. The reaction is conducted below the melting point, typically at 200–220°C under vacuum or nitrogen sweep. In this process, rPET pellets are crystallized and then held in a solid-state reactor where esterification and transesterification reactions increase chain length. The IV increase rate depends on starting IV, pellet size, temperature, and vacuum level. If flake IV is below 0.65 dL/g, the required SSP residence time becomes disproportionately long, reducing continuous throughput and increasing energy demand per kilogram of bottle-grade pellet. Conversely, flake with IV above 0.75 dL/g allows SSP to reach 0.82 dL/g in shorter cycles, which is why high-quality bottle bales retain a distinct feedstock value.

Moisture control remains the limiting variable. PET is hygroscopic; post-wash flake equilibrates with ambient humidity unless immediately dried. At relative humidity above 60%, flake can re-absorb surface moisture within minutes. A desiccant dryer operating at 160–170°C with a dew point of −40°C reduces moisture to below 50 ppm before extrusion. Drying time shorter than 4 h may be insufficient for cold flake bed temperatures, while over-drying above 6 h increases thermal history and can raise acetaldehyde formation in the finished pellet. These boundaries explain why bottle flake storage, drying, and extrusion are managed as a continuous moisture-control sequence rather than as separate unit operations.

The collection gap becomes most visible at the flake intake door. A bale containing 15% non-bottle PET, PVC sleeves, and residual liquids may still produce a measurable yield, but the yield is headed for fiber or strapping rather than bottle preforms. The difference between collected bottle mass and usable bottle flake mass is the operational expression of the collection gap, and it cannot be corrected solely by adding downstream melt filtration capacity.