Glass Transition Depression in Large Castings Due to Slow Cooling

In casting operations where solidification proceeds across wall thicknesses exceeding 50 mm, the glass transition temperature measured at the core frequently departs from the manufacturer’s neat-resin specification by 8 °C to 18 °C. This phenomenon originates in the protracted thermal residence within the rubbery-to-glassy conversion window, which allows progressive densification and molecular rearrangement that would be kinetically arrested under the rapid quench profiles used to derive datasheet values. For an unfilled bisphenol-A diglycidyl ether cured with a cycloaliphatic anhydride, differential scanning calorimetry per ASTM E1356-23 on microtomed slices extracted from a 75 mm-thick casting revealed a core Tg of 118 °C versus a skin Tg of 134 °C, with the transition midpoint broadening from 8 °C to 22 °C full-width half-maximum. A production-line case encountered during manufacture of compression-moulded bushings for medium-voltage switchgear, where the tool steel mould was maintained at 140 °C and the casting exothermed to a peak of 198 °C, further illustrated the interplay: the thermal centre cooled from 198 °C to 110 °C over 4.7 hours, whereas the outer 3 mm traversed the same interval in 12 minutes. When these components were subjected to dynamic mechanical analysis from −50 °C to 250 °C at 1 Hz following ISO 6721-11:2019, the loss modulus peak at the core lagged by 16 °C and the rubbery plateau modulus was 1.8 MPa lower, confirming that slow cooling not only depresses the calorimetric Tg but also reduces the effective crosslink density available for load transfer at elevated temperature. Such gradients become structurally significant once the part thickness exceeds the critical dimension where Fourier-number-based cooling-rate disparity exceeds a factor of 5, and they cannot be remediated by post-cure alone because the vitrification front advances inward with a spatially dependent thermal lag that permanently locks in non-equilibrium free-volume distributions.

What governs the magnitude of Tg depression across the casting section?

The primary drivers are the instantaneous cooling rate through the glass transition region and the local degree of conversion upon vitrification. Cooling rate sensitivity follows a Vogel-Fulcher-Tammann dependence; a 20 mm-thick plate of a polyurethane casting system based on a poly(tetramethylene ether) glycol soft segment and a methylene diphenyl diisocyanate prepolymer, when cooled at the core rate of 0.3 °C/min versus the skin rate of 8 °C/min, exhibited a Tg shift from −38 °C to −51 °C as measured by modulated DSC with a ±0.5 °C amplitude and 60 s period per ASTM D7028-07(2024). In addition to the purely kinetic freezing of segmental mobility, the spatial conversion gradient contributes when the cure kinetics are sufficiently temperature-dependent that the core reaches vitrification before full conversion. For a cyanate ester resin cyclotrimerisation, where the activation energy of cure is 87 kJ/mol, a 60 mm casting with a mould-wall temperature of 177 °C showed a core conversion of 89 % by near-infrared spectroscopy of the cyanate band at 2270 cm⁻¹, while the skin exceeded 97 %. The resultant Tg difference of 27 °C—from 252 °C (skin) to 225 °C (core)—was partially recovered after a 6 h post-cure ramp to 260 °C, but a residual 9 °C depression persisted due to the topological constraints imposed on network densification by the earlier vitrification. On a twin-screw reactive processing line producing cast-in-place polyamide 11 liners for high-pressure hydrogen vessels, the cooling rate through the melt-to-solid transition is deliberately maintained above 15 °C/min by jacketed moulds circulating oil at −20 °C, because slower cooling rates of 2 °C/min result in a 5 °C depression in the α-relaxation temperature and a lowering of the yield stress by 12 % when tested at 60 °C per ISO 527-1:2019. It is the ratio of the characteristic relaxation time to the cooling time constant, often expressed via the Davies or Tool-Narayanaswamy-Moynihan parameters, that determines how far the observed Tg deviates from the limiting fictive temperature; for every order-of-magnitude reduction in cooling rate, a 3 °C to 5 °C depression is typical in network-forming polymer glasses devoid of significant side-chain β-relaxation interference. When tetrachloroethane replaces methylene chloride in immersion stripping of large cast epoxy tooling boards, the slow solvent evaporation and concurrent cooling at the core mimic the thermal conditions of thick-section moulding, and the resulting Tg depression must be accounted for in dimensional stability modelling. A metrology-grade epoxy board of 100 mm thickness, originally cured at 40 °C and post-cured at 80 °C with a reported Tg of 96 °C per ISO 11357-2:2020, was immersed in 1,1,2,2-tetrachloroethane at 25 °C for 24 h, then allowed to dry in still air. Thermal desorption analysis confirmed residual solvent of 0.9 wt% after 72 h, and DMA revealed a core Tg of 71 °C—a 25 °C depression attributable jointly to plasticisation and to the abnormally slow solvent evaporation front that retards the recovery of the equilibrium liquid structure. The processing window for final machining was consequently narrowed from ±12 °C around the nominal Tg to a floor of 55 °C, below which microcracking became probable during aggressive carbide-tool milling at 10,000 rpm spindle speed.

Measured Tg depression in a 70 mm-thick unfilled DGEBA/anhydride casting with varying post-cure and cooling regimes (DSC at 10 °C/min, ASTM E1356-23)

Condition	Skin Tg (°C)	Core Tg (°C)	ΔTg (°C)	Cooling rate core (°C/min)
Mould at 120 °C, air cool	128	111	17	0.4
Mould at 120 °C, cool in oven at 1 °C/min	130	120	10	0.8
Post-cure 4 h at 150 °C, air cool	136	121	15	0.3
Post-cure 8 h at 150 °C, cool at 0.5 °C/min	138	129	9	0.5

In production-scale reactive injection moulding of large glass-mat-reinforced dicyclopentadiene parts for agricultural machinery housings, the control logic of the mould temperature zones must compensate for the fact that the core of a 30 kg shot undergoing ring-opening metathesis polymerisation exhibits a residence time above the gel point exceeding 25 minutes, during which the temperature drifts downward at 0.6 °C/min once the exotherm subsides. That drift, recorded by embedded thermocouple arrays, generates a core Tg typically 8 °C below design, sufficient to reduce the heat deflection temperature under 1.82 MPa load (ISO 75-2:2013) from a target of 102 °C to 94 °C and thus failing the acceptance criterion for a driveline component exposed to continuous oil splash at 95 °C. Molders address this by programming a reverse thermal quench: the outer shell is cooled rapidly with 15 °C water while the core zone is held at 85 °C for an additional 12 minutes, which reduces the ΔTg across the wall to 4 °C but extends cycle time by 18 %—an economic penalty accepted only where the specified minimum Tg is contractually mandated by the OEM. A distinct risk emerges when amine-cured epoxies are cast in open moulds under uncontrolled ambient humidity exceeding 60 % RH. Moisture condensing on the slowly cooling surface undergoes reaction with isocyanate or amine sites, forming urea linkages and carbamate species that plasticise the network. For a 45 mm casting of a 3,3′-dimethyl-4,4′-diaminodicyclohexylmethane epoxy system exposed to 25 °C and 75 % RH throughout gelation and cooling, the core Tg sank to 104 °C compared with 127 °C for a dry nitrogen-blanketed control. The hygrothermal history also widens the β-transition, visible as a shoulder on the loss modulus curve between −70 °C and −40 °C, which correlates with a 30 % increase in room-temperature creep compliance under a 10 MPa static load applied for 24 h as per ISO 899-1:2017. Such casting environments therefore demand real-time dew-point monitoring and closed-loop humidity control to <35 % RH, as well as a pre-run bake-out of aggregate fillers at 110 °C to a moisture content below 300 ppm by Karl Fischer titration (ASTM D4672-22), to prevent the compounding of thermal and hygroscopic Tg depression mechanisms.

Thermal history fingerprints in industrial cast poly(methyl methacrylate) sheet

Continuously cast PMMA in thicknesses from 40 mm to 120 mm carries a permanent thermal history that is readable through the enthalpy overshoot at the glass transition. When a 80 mm sheet is cast between polished steel belts at a line speed of 0.15 m/min and subsequently annealed with a stepped cooling programme of 2 °C/min from 140 °C to 80 °C, the DSC first-heating scan shows an endothermic peak superimposed on the Tg step with an area of 2.1 J/g, while a water-quenched edge sample of the same chemistry shows a flat step with 0.3 J/g overshoot. This enthalpy recovery is not a cosmetic artefact; it directly affects the propensity for crazing under sustained tensile stress at 40 °C in air, with the quenched edge exhibiting a critical strain for craze initiation of 0.9 % compared with 1.4 % for the annealed core, measured by a constant-strain environmental stress cracking rig in accordance with the specimen geometry of ISO 22088-2:2006 and using ethanol as the active environment. From a manufacturing standpoint, the cast sheet destined for aircraft transparencies must be certified to MIL-PRF-25690C, which specifies a maximum acceptable biaxial residual stress evaluated through a photoelastic fringe count; the slower-core-cooling-induced density variation adds 0.5 to 1.0 fringe orders relative to an isothermally equilibrated coupon, and exceeding three fringes typically triggers rejection of the lift. Process engineers counteract this by inserting a 20-minute isothermal hold at the material’s fictive temperature of 108 °C before the final cool-down, effectively erasing the accumulated enthalpy and resetting the volumetric state without altering the optical clarity.

Compliance standards for glass transition measurement and acceptance in large cast polymer components

Standard	Method	Relevant clause / reporting parameter
ISO 11357-2:2020	Differential scanning calorimetry	Midpoint Tg, extrapolated onset, step height
ASTM E1356-23	DSC glass transition assignment	Section 10: Tg midpoint, report cooling rate
ASTM D7028-07(2024)	Dynamic mechanical analysis Tg	Loss modulus peak or tan δ peak
ISO 6721-11:2019	DMA of solid polymers	Storage modulus onset, loss modulus peak
ISO 75-2:2013	Heat deflection temperature	HDT at 1.82 MPa and 0.45 MPa
ISO 899-1:2017	Tensile creep	Creep modulus after 24 h
DIN EN ISO 22088-2:2006	ESC craze initiation	Critical strain in active environment

The situation becomes more severe in filled systems where the filler’s thermal diffusivity alters the macroscopic cooling pattern. A 90 mm-thick casting of a low-profile unsaturated polyester resin containing 35 wt% calcium carbonate (d₅₀ = 5 µm) and 25 wt% hollow glass microspheres exhibits a thermal conductivity of 0.38 W/(m·K), compared with 0.21 W/(m·K) for the unfilled resin. While the filler raises the nominal Tg by 3 °C due to restricted segmental mobility, the enhanced heat transfer reduces the core-to-skin cooling-rate ratio from 8.2 to 4.5, simultaneously lowering the spatial Tg gradient from 15 °C to 9 °C. This trade-off is exploited in the formulation of large bathroom vanity basins where a uniform surface hardness is demanded to prevent differential polishing marks; the specification calls for a maximum Barcol hardness variation of 3 points (ASTM D2583-13a) across the visible surface, and the filled formulation achieves 2.2 points compared with 5.1 points for the unfilled counterpart when both are cast in 60 mm-thick open moulds. Conversely, the incorporation of conductive carbon nanotubes at 0.5 wt% loading in a vinyl ester casting reduces through-thickness thermal resistance sufficiently that the core cooling rate increases from 0.3 °C/min to 0.7 °C/min, yet the localised cure exotherm acceleration near the nanotube-matrix interface creates nanometre-scale heterogeneities that manifest macroscopically as a splitting of the DSC glass transition into a low-shoulder at 101 °C and a primary step at 118 °C; published data for this specific nanofilled vinyl ester configuration remains limited to lab-scale specimens of 15 mm thickness, and extrapolation to production dimensions exceeding 50 mm is poorly validated. When casting massive poly(dimethylsiloxane) blocks for acoustic lens applications in medical ultrasound transducers, the extremely low glass transition of −123 °C means that ambient cooling rates of 1 °C/min versus 10 °C/min produce a detectable 2 °C shift in the dynamic Tg, which, though numerically small, alters the speed of sound through the lens by 0.3 %—enough to degrade the focal spot size by 8 µm when the acoustic path length is 90 mm. The manufacturing protocol therefore stipulates a controlled cool from the 150 °C cure plateau to −20 °C at exactly 5 °C/min, enforced by a multi-zone air oven with a uniformity of ±1 K across the 400 mm × 400 mm working volume, and a subsequent hold at −20 °C for 2 h to equilibrate the fictive temperature before demoulding. Any excursion below 4.5 °C/min triggers a rework cycle limited to two repetitions, after which the lot is scrapped because cumulative enthalpy relaxation irreversibly raises the storage modulus at 5 MHz by 4 %, violating the acoustic impedance specification of 1.45 ± 0.02 MRayl. This example underscores that in high-precision applications the operational boundary for cooling-rate tolerance may be narrower than the ±5 °C processing window often cited for thermoset composites; it is the interplay between glass transition depression and the functional property gradient that defines true process capability. In the absence of a header, the following observation is inserted directly: milled carbon-fibre-filled polyetheretherketone blanks for compressor plate valves, machined from a 65 mm-thick compression-moulded slab, demonstrate that the crystalline fraction and the rigid amorphous fraction both participate in the effective glass transition behaviour. Where the slab centre cooled at 2 °C/min from the 400 °C melt, the degree of crystallinity determined by differential scanning calorimetry (second heat, 20 °C/min) was 38 %, and the constrained amorphous region, quantified from the deficit between the observed and two-phase heat capacity increment at Tg, reached 24 %. At the water-quenched edge, crystallinity was 28 % and rigid amorphous fraction 31 %, resulting in a calorimetric Tg of 148 °C versus 153 °C at the core—a reversal of the typical core-depression trend but entirely consistent with the increased crystalline confinement limiting segmental relaxation. This interplay invalidates any attempt to use a single cooling-rate-dependent Tg model without incorporating the crystallisation kinetics, and is one reason why thick PEEK castings require individual part qualification by ultrasonic C-scan to correlate the acoustic velocity with the local dynamic mechanical Tg, per the internal acceptance test protocol derived from ISO 6721-11.