Polyethylene Terephthalate (PET) is a semi-crystalline thermoplastic polyester whose repeating unit comprises ethylene glycol and terephthalic acid linked through ester bonds. Polyethylene Terephthalate fails in three interconnected ways relevant to beverage packaging: hydrolysis, stress cracking, and acetaldehyde taste. In injection stretch-blow moulded beverage containers, biaxially oriented PET achieves hoop and axial tensile yield strengths commonly measured in the 150 MPa to 200 MPa range and density-derived crystallinity of 25% to 35% per ISO 1183-1:2019. The failure modes that limit commercial shelf life are not ductile matrix yielding; they appear as molecular weight collapse from hydrolysis, environmental stress cracking in contact with conveyor lubricants and alkaline detergents, and sensory rejection caused by acetaldehyde migration. Each mode has a distinct thermodynamic and kinetic boundary, with allowable processing windows frequently narrower than ±5 °C in preform moulding and residual moisture limits below 50 ppm in melt-state conversion. Failure analysis therefore begins with the simultaneous measurement of intrinsic viscosity, carboxylic end group concentration, residual moisture, and preform acetaldehyde, because these variables interact during injection moulding and blow moulding.

What Causes Hydrolytic Scission During Melt Processing?

Hydrolysis of PET proceeds by random chain scission of the ester linkage, producing one carboxylic acid end group and one hydroxyl ester terminus per scission event. The reaction is autocatalytic; liberated carboxylic acid groups lower local pH and accelerate further ester cleavage, a feedback loop observed during intrinsic viscosity loss measured per ASTM D4603-18 and ISO 1628-5:1998. In melt-state conversion at barrel temperatures between 270 °C and 295 °C, residual moisture above 50 ppm is sufficient to generate measurable intrinsic viscosity loss within residence-time distributions of 2 min to 8 min. Solid-state polymerization of bottle-grade PET typically raises intrinsic viscosity from 0.60 dL/g to 0.82 dL/g, but this gain is erased if the resin is not dried before injection moulding.

Desiccant drying systems rated for a dew point of −40 °C feeding a heated hopper at 160 °C to 180 °C for 4 h to 6 h are the standard countermeasure; these units reduce pellet moisture to ≤30 ppm when inlet air flow is maintained at 3.5 m³/h per kg/h throughput. The critical control parameter is not hopper temperature alone but the dew point of air delivered to the resin; air at −40 °C dew point holds about 0.08 g water/kg dry air, while air at −20 °C dew point holds about 0.63 g water/kg dry air. Insufficient air flow or partially exhausted desiccant beds produce high pellet moisture even when the hopper setpoint is correct. On the injection moulding floor, a drop in intrinsic viscosity from 0.82 dL/g to below 0.70 dL/g and an increase in carboxylic end group concentration from 20 meq/kg to above 60 meq/kg are typical of hydrolytic damage; preforms then show low top-load strength and reduced base-burst pressure. On production lines without sealed hopper-inlet conveying, resin dried to 30 ppm can regain measurable surface moisture within 30 min when exposed to ambient air at 25 °C and 60% RH, and the first injection cycles after restart are the most frequent source of hydrolyzed preforms.

Vented extruders with L/D ratio 36:1 and vacuum vent pressure below 10 kPa absolute can strip some residual moisture from the melt, but this capability is not a substitute for pre-drying. Published melt-phase kinetic constants for commercial bottle-grade PET are limited, but the industrial operating limit is narrower than ±5 °C around 280 °C when hydrolysis and acetaldehyde generation are simultaneously controlled. Processors that rely solely on melt flow index per ISO 1133-1:2022 or on preform tensile yield per ASTM D638-14 Type IV specimens may miss hydrolytic scission. A preform can retain tensile yield above 50 MPa even when intrinsic viscosity has fallen to 0.70 dL/g; the molecular weight loss appears first as reduced base-burst pressure and environmental stress crack resistance, not as a simple tensile failure.

Environmental stress cracking in oriented PET containers is driven by the combination of residual hoop stress from carbonation pressure and surface-active agents that reduce cohesive entanglement density at craze fibrils. Injection stretch-blow moulded bottles tested at 4.0 volumes CO₂ and 22 °C under constant-tensile-load ESC conditions per ISO 22088-2 develop base cracks in the gate region within 24 h when conveyor lubricant concentration exceeds 0.1% and relative humidity remains above 85%. The crack path follows the boundary between injection-moulded amorphous islands and stress-crystallized oriented regions; propagation is fastest through low molecular weight zones where intrinsic viscosity has fallen below 0.72 dL/g. Production lines commonly use an aggressive alkaline bottle wash trial at pH 12 and 45 °C for 15 min to rank material lots; bottles that survive 1000 h of pressurised storage after wash are accepted for carbonated soft drink use. Published data for the exact lot-to-lot variation under this specific configuration is limited. The failure mode is often misdiagnosed as a moulding defect because the crack is sharp and splits along the preform gate crystallinity gradient, but scanning electron microscopy shows craze fibril remnants and secondary cracking characteristic of environmental stress cracking rather than impact damage.

Stress Crack Propagation at the Bottle Base: Fibril Rupture and Humidity Thresholds

The fracture mechanics of stress cracking in PET are governed by craze fibril yield and disentanglement ahead of a crack tip. Under hoop stress of 40 MPa to 60 MPa in sidewalls and up to 120 MPa at the base periphery, surface-active agents such as linear alkylbenzene sulfonates lower the craze fibril yield stress by plasticizing the amorphous phase. Reported critical stress intensity factor values for oriented PET in dry air at 23 °C are approximately 1.0 MPa·m^0.5 and drop below 0.5 MPa·m^0.5 in the presence of 0.05% sodium hydroxide. The reduction is not solely a surface-energy effect; the same agents increase equilibrium moisture sorption of the amorphous phase, and absorbed water acts as a crazing accelerant. Biaxial orientation improves stress crack resistance only when axial stretch ratio is maintained between 2.8 and 3.2 and radial blow ratio is kept between 3.0 and 3.4; out-of-window stretching creates a morphology with insufficient chain extension to resist fibril rupture. Preforms with high acetaldehyde but adequate intrinsic viscosity are not necessarily prone to stress cracking; however, hydrolyzed resin with carboxylic end group concentration above 50 meq/kg shows a measurable reduction in craze stability due to shorter chain length. The practical consequence is that a drop of 0.05 dL/g in intrinsic viscosity is detectable as a shorter time to base crack in pressurised storage at 4.0 volumes CO₂; accelerated trials at 40 °C and 85% relative humidity are used to compare lots.

Because stress cracking in carbonated bottles is accelerated by humid storage and caustic conveyor lubricants, preform design also contributes to failure. Gate geometry with sharp transitions creates residual stress concentrations that lower the apparent ESC threshold. Moulders therefore measure base crystallinity by density gradient column per ISO 1183-1:2019 and verify base wall thickness distribution before running full production. A preform with a gate vestige radius below 0.2 mm may fail base-crack testing even when the resin is within specification; this mechanical factor is separate from molecular weight but interacts with hydrolysis because low-intrinsic-viscosity resin has less capacity to redistribute stress around the gate notch. Published data for a specific gate-radius-to-IV interaction in commercial preforms remains limited.

Acetaldehyde generation is a parallel degradation route that does not require moisture. The aldehyde forms by intramolecular backbiting during thermal decomposition of the PET ester group and by oxidation of ethylene glycol residues. In injection moulding of preforms, acetaldehyde concentration in the melt scales approximately linearly with barrel temperature above 275 °C and rises sharply with residence time. Raising melt temperature from 280 °C to 290 °C increases acetaldehyde in preforms from approximately 3 µg/L to 8 µg/L as measured by headspace gas chromatography per ASTM F2013-10; extending residence time beyond 4 min at 290 °C raises acetaldehyde above 12 µg/L. Carbonated soft drink producers generally reject preforms above 6 µg/L residual acetaldehyde when filled with unflavoured sparkling water, while flavoured beverages may tolerate up to 10 µg/L. Water bottlers often set an internal preform acetaldehyde limit of 3 µg/L to maintain a wide safety margin for sensory thresholds. The same preform can therefore be acceptable for one beverage and rejected for another.

The sensory threshold for acetaldehyde in water is reported in the 20 µg/L to 50 µg/L range, but migration at 40 °C for 10 days into 3% acetic acid food simulant per EU No 10/2011 can exceed this range when preform acetaldehyde is above 8 µg/L. Acetaldehyde migration is not governed by an explicit numeric specific migration limit in EU No 10/2011; enforcement is sensory and typically predicated on organoleptic panel rejection thresholds rather than a numeric migration value. Residual acetaldehyde is not uniformly distributed in the bottle wall. The skin layers of a preform cool quickly and freeze in higher acetaldehyde than the core, while the core remains hot longer and can lose some aldehyde through venting or regenerated precursor chemistry. Headspace measurement on a ground preform therefore reports an average value that may under-represent the skin contribution to migration. For this reason, some filling lines correlate preform acetaldehyde with taste-test results using a trained panel and a triangle test protocol rather than relying solely on instrumentation. Published data for the exact skin-to-core acetaldehyde gradient in commercial preforms is limited.

When Acetaldehyde Formation Tracks Screw Residence Time and Melt Temperature

On a 48-cavity preform injection moulding machine with a 70 mm screw and L/D ratio of 24:1, the residence-time distribution broadens when shot weight falls below 30% of barrel capacity, increasing the fraction of melt exposed to temperatures above 285 °C for longer than 5 min. This condition is aggravated by hot-runner manifold dead spots and by screw-recovery delay during robot part removal. The result is a bimodal acetaldehyde distribution: first-shot preforms after startup contain 15 µg/L to 20 µg/L acetaldehyde, dropping to 5 µg/L within 8 cycles, then rising again if cycle time is extended beyond 20 s. To maintain acetaldehyde below 6 µg/L, melt temperature measured by an immersion thermocouple at the nozzle should be held between 270 °C and 280 °C, and cushion position should be maintained within 3 mm to 5 mm to minimize polymer hold-up. On a hot-runner system with 48 needle-valve nozzles, individual nozzle tip temperature variation of ±2 °C is sufficient to produce cavity-to-cavity acetaldehyde differences that exceed 2 µg/L; therefore manifold thermal uniformity is part of the organoleptic control strategy.

Acetaldehyde scavengers such as polyamides or acetaldehyde-scavenging additives may reduce preform acetaldehyde by 30% to 50%, but they increase haze and require revalidation of food-contact status under FDA 21 CFR 177.1630 and EU No 10/2011. The same barrel-temperature reduction that controls acetaldehyde also suppresses hydrolytic scission because both degradation modes accelerate with thermal input. However, lowering barrel temperature below 270 °C increases melt viscosity and can produce short shots, gate shear heating, and higher injection pressure; this is the central process conflict that makes acetaldehyde control nonlinear.

Material certification for bottle-grade PET requires simultaneous verification of molecular weight, moisture, and acetaldehyde precursors. A production lot with intrinsic viscosity of 0.80 dL/g per ASTM D4603-18 and carboxylic end group concentration below 30 meq/kg can still fail sensory acceptance if acetaldehyde exceeds specification. Incoming resin is sampled at hopper inlet and preforms are sampled at 4 h intervals for acetaldehyde. The following matrix gathers the test designations and common industrial limits used for failure analysis.

Published data for some specific configurations is limited because many production thresholds are proprietary internal specifications. Nevertheless, the failure boundaries in the table correspond to standardised analytical results and are reproducible when moisture, intrinsic viscosity, and acetaldehyde are measured at the hopper outlet and at the preform gate. Operators should not infer that a single pass/fail criterion is sufficient; simultaneous excursions in moisture and barrel temperature create coupled degradation modes that are not captured by individual test results.