Recycled polyethylene terephthalate fibre is not a single material but a stream of post-consumer bottle bales in which flake colour carries the cumulative signature of polymer degradation, label pigments, cap resins, and multilayer barrier residues. When these flakes are melt-extruded into staple fibre or filament yarn, the measured CIE L*a*b* coordinates of the flake feed become the dominant predictor of yarn shade because molten PET is optically transparent and retains chromophores unless oxidative or hydrolytic processes bleach or yellow them further. A shift in flake b* from 2.0 to 6.5 under D65/10° illumination can produce a visible cream shift in undyed yarn even after 0.3 wt% optical brightener addition. Flake colour therefore functions as a pre-extrusion quality gate: it determines blend ratios, additive loadings, and the achievable shade bandwidth for downstream knitting and weaving.

Colour measurement of rPET flakes uses a sphere-based spectrophotometer configured with d/8° geometry, D65 illuminant, 10° observer, and specular component included or excluded depending on whether surface gloss from washed flake is to be separated. Flake samples are presented in a glass cell of fixed path length, and the reported values are typically the arithmetic mean of 30 individual flakes to reduce orientation and reflection variance. In bottle-to-fibre operations, clear flake is commonly controlled to L* ≥ 82, a* between -1.5 and +1.5, and b* ≤ 3.5. Light blue flake used as a shade adjuster may record b* between -4.0 and -1.0, while green flake can record a* between -9.0 and -3.0 and b* between 2.0 and 8.0. The yellowness index of clear flake measured by ASTM E313 typically rises from 2–4 for virgin pellet to 5–8 for washed clear rPET; this baseline shift limits brilliant whites unless optical brighteners or blue flake are used. These ranges are not absolute; they shift with feedstock geography and the proportion of hot-filled bottles, which often retain higher acetaldehyde and oligomer levels.

Table 1 summarises typical flake colour ranges and their observed undyed yarn shade behaviour.

Ranges are indicative of industrial bale audits, not single supplier specifications.

Which Feedstock Contaminants Shift Flake Colour Before the Spinning Line?

The main colour contributors in post-consumer rPET are not residual polyester degradation products alone. Bale audits frequently identify polypropylene caps, polyethylene liners, PVC shrink sleeves, acetaldehyde-containing closures, and paper labels. PVC is particularly consequential: at melt temperatures above 280 °C, PVC dehydrochlorination releases HCl, which autocatalyses PET hydrolysis and generates yellowing species. X-ray fluorescence or density bath separation limits PVC to below 50 ppm in bottle-grade rPET because chlorine above this threshold can shift b* by more than 2.0 units during a 12 min residence time. Colour sorting using visible-spectrum RGB cameras rejects coloured flakes when the camera’s normalised red-to-green ratio falls outside a set tolerance; near-infrared NIR sensors identify PET versus PVC and polyolefin fractions by the 1650–1750 nm C-H overtone region. However, visible sorting cannot detect low levels of yellowing caused by oxidative history. Metal residues from cap liners and machine wear, particularly iron above 10 ppm, catalyse oxidation and darken the melt.

Flake-grade decision thresholds used by sorting lines are not colorimetric pass/fail criteria alone; they are mass-flow control points. When clear flake is blended with 5 wt% green flake, the resulting melt can shift a* by approximately 0.4–0.8 units and b* by 0.3–0.6 units before melt filtration. This shift is measurable in the spun yarn, but it is not linear because PET’s optical absorbance depends on chromophore concentration and on particle scattering from gels and additives. Near-infrared sorters with 1–2 mm nozzle spacing eject coloured flakes by comparing the characteristic 1.9–2.1 µm reflectance peak of PET. However, flakes smaller than 5 mm are more likely to be lost in air jets and re-enter the accept stream. The resulting flake colour distribution in a super sack can have a coefficient of variation for b* of 15–30% if bale blending is not controlled.

If Intrinsic Viscosity and Carboxyl End Groups Drift, Colour Shift Is Already Locked In

Intrinsic viscosity loss during extrusion changes the rheological path and indirectly influences colour through oligomer formation and thermal oxidation. Flake with an input IV of 0.78 dL/g measured by ASTM D4603 in 60/40 phenol/tetrachloroethane at 30 °C can drop to 0.68 dL/g after drying at 165 °C for 5 h and spinning at 285 °C, depending on moisture and residence time. Carboxyl end group concentration measured by ASTM D740 often rises from 20 meq/kg to 35–45 meq/kg under hydrolytic conditions, and this increase correlates with a yellow shift because CEG formation accompanies chain scission and the release of acetaldehyde and vinyl ester species. A high CEG value also reduces dye uptake uniformity in disperse dyeing because terminal carboxyl groups alter the fibre’s surface charge. For undyed shade stability, many spinners specify input flake with CEG ≤ 30 meq/kg and IV ≥ 0.72 dL/g.

Process Conditions That Convert Residual Colour Bodies into Yarn Shade Deviation

Melt temperature, residence time, and screw design determine whether flake colour is preserved or amplified. A twin-screw extruder with 32:1–40:1 L/D and low-shear screw elements can process clear rPET at 275–290 °C with a residence time of 6–10 min. At the upper temperature limit, the b* value of extrudate can increase by 1.5–3.0 units compared with the flake feed. The effect is larger when the flake contains residual label adhesives, sugar-based deposits from beverage residues, or PVC fragments because these materials degrade into conjugated carbonyl species. Melt filtration with candle filters of 20–40 µm nominal rating removes some macroscopic particles but does not remove soluble yellow chromophores. Spin pack pressure upstream of the filter is typically maintained at 80–180 bar; a pressure rise above 180 bar indicates gel accumulation and can force premature pack changes, which themselves introduce shade variation. Pre-drying at 160–170 °C for 4–6 h to a target moisture of <30 ppm is required. If moisture exceeds 50 ppm, hydrolysis increases CEG and lowers melt viscosity, producing a grey-yellow yarn even from clear flake.

At the spin beam, quench air temperature and draw ratio do not alter the inherent chromophore concentration, but crystallite size and orientation modify the diffuse reflectance of undrawn versus drawn yarn. Drawn yarn can appear lighter or darker depending on fibre titre, with fine denier filaments scattering more light; a change from 1.5 dtex to 0.9 dtex filament titre can shift perceived L* without any change in flake colour. Crystallinity measured by differential scanning calorimetry per ISO 11357-3 increases from 5–10% in unstretched spun yarn to 30–40% after drawing, but this does not remove pre-existing chromophores.

Quantifying the Relationship Between Flake b* and Dye Uptake in Staple Yarn

Flake colour influences undyed yarn shade, but its effect on downstream dyeing is equally important for pale and pastel shades. A b* difference of 2.0 in feedstock can produce a difference of 0.4–0.8 units in dyed L* for a 0.5% disperse dye recipe because the base substrate is not optically neutral. Predisperse dye trials on knitted sleeves dyed at 130 °C for 45 min show that the same dye formula applied to clear-flake yarn and green-flake yarn can differ by ΔE*ab > 1.0, which exceeds typical retail shade tolerances for solid-colour garments. Shade correction in the dyehouse requires either weighted dye recipe adjustments or upstream flake blending to hold base a* and b* within ±0.5 units of the approved standard. When flake colour is not controlled, the dyehouse must increase rework and add shading dyes, which reduces dyehouse capacity. Disperse dye diffusion into rPET is also affected by the crystalline fraction developed during drawing and heat-setting at 140–160 °C. If flake colour is accompanied by a wider molecular weight distribution, the fibre may produce uneven dye penetration, visible as barré in fabric.

Dose-response data for optical brighteners in rPET show that a 0.1 wt% increase in a triazine-stilbene brightener raises CIE whiteness index by approximately 4–7 points but has diminishing returns above 0.3 wt% because of aggregation and exudation. Titanium dioxide added at 0.3–0.5 wt% masks yellow flake, but it reduces fibre tenacity by 0.15–0.25 cN/dtex and increases filter pressure. Therefore, shade correction through additives is bounded by mechanical property loss and spinability. The more robust control is upstream: keeping green flake below 3 wt% and yellow flake below 5 wt% of the blend before additive compensation.

Specific acceptance windows are operationally set; published data for a specific spinning configuration may vary.

Published data for the exact interaction between flake colour coordinates and yarn shade in high-speed staple spinning is limited because most commercial shading curves are proprietary. Industrial datasets nonetheless show that flake b*, L*, and the concentration of thermo-oxidative degradation products are the most transferable predictors of undyed yarn shade. Lines that do not track these parameters before silo blending typically compensate with higher colourant or brightener doses, which narrows the process window and increases melt filtration variability.