In production-scale injection moulding of polybutylene terephthalate (PBT), measured post-ejection warpage on a 150 mm × 100 mm × 2 mm centre-gated plaque processed through a 25 mm three-zone screw with 20:1 L/D and an 800 kN clamp force frequently ranges from 1.0 mm to 2.4 mm for unfilled grades when the mould temperature is held at 60 °C to 80 °C and melt temperature at 250 °C to 260 °C. The distortion is not attributable to a single shrinkage coefficient. Differential scanning calorimetry according to ISO 11357-1 places the PBT melting peak near 223 °C to 225 °C and the maximum crystallization rate during cooling between 160 °C and 190 °C; this means that cavity wall temperatures below the crystallization-rate maximum freeze a highly oriented skin while the core continues to crystallize after gate seal. The resulting through-thickness density gradient, combined with packing-pressure decay along the flow path, produces a bending moment that exceeds the stiffness of thin sections.

After ejection, the volumetric shrinkage of unfilled PBT is typically 1.4 % to 2.0 % parallel and perpendicular to flow when measured in accordance with ISO 294-4:2018. That isotropy does not prevent warpage because the midplane and outer layers do not shrink at the same time or against the same constraint. The solid skin formed at the cavity wall develops orientation in the flow direction; the core solidifies later under lower pressure and with less orientation. If the moving and fixed mould halves are controlled at different temperatures, the temperature differential can produce a curvature on the order of 0.2 mm to 0.5 mm per 10 °C difference in a 2 mm thick part. This is why process capability studies for unfilled PBT connector components often specify a mould-temperature tolerance of ±2 °C per cavity plate, not merely a nominal setpoint.

Differential scanning calorimetry can understate the processing sensitivity because the cooling rate in an injection mould is 50 K/min to 200 K/min at the wall, whereas standard DSC cooling rates are 10 K/min or 20 K/min. At the cavity wall, the PBT skin may leave the crystallization-rate maximum before the core has cooled, creating a crystallinity difference of 5 % to 15 % absolute between skin and core. This crystallinity difference corresponds to a local shrinkage mismatch and is sufficient to produce measurable flatness error in parts with thickness below 2 mm. The same effect is stronger in unfilled PBT than in glass-filled PBT because fibres dilute the crystallizable volume but do not eliminate the matrix crystallization exotherm.

Why Does Unfilled PBT Warp Despite a Moderate Mould Shrinkage Range?

Unfilled PBT warps even when the nominal mould shrinkage is isotropic because the solidification sequence, not the final shrinkage value alone, governs the development of internal stress. In a cavity with a nominal wall thickness of 1.6 mm to 2.0 mm, the gate seal time for unfilled PBT at a melt temperature of 255 °C and mould temperature of 80 °C is often 4 s to 6 s. If the holding pressure is terminated before gate seal, molten core material can flow backward through the gate, reducing cavity pressure near the sprue and causing density loss in the gate region. The ejection temperature then exceeds 90 °C to 120 °C at the thickest section, and post-ejection crystallization continues for 24 h to 48 h at ambient temperature. This after-shrinkage is commonly 0.05 % to 0.15 % and contributes to delayed flatness failure in printed circuit board connectors.

When the moving half is held at 60 °C and the fixed half at 90 °C, the faster-cooling surface solidifies earlier and develops higher orientation, while the slower-cooling surface remains rubbery and accumulates more crystallinity later. The part bows toward the hotter surface because the colder surface shrinks less after ejection. The sensitivity of unfilled PBT to this imbalance is high: a 10 °C differential across the cavity can shift warpage by more than 0.5 mm on a 100 mm span. Production lines therefore use dual-zone mould temperature control units with ±1 °C stability and independent cooling circuits per cavity plate.

For unfilled PBT, the operational boundary is a narrow mould-temperature window. Mould temperatures below 80 °C reduce cycle time but freeze excessive orientation and increase the risk of post-mould dimensional change. Mould temperatures above 120 °C improve dimensional stability but can cause ejection problems, plate-out, and cycle-time increases of 20 % to 30 %. The preferred production range for many unfilled PBT electrical connectors is 90 °C to 110 °C, with the actual setting validated by flatness measurements after 48 h because after-crystallization shifts the part shape even when initial ejection is acceptable.

Glass Fibre Orientation as the Dominant Warpage Tensor in Reinforced Grades

In short-glass-fibre PBT compounds, the E-glass fibres have a coefficient of linear thermal expansion of approximately 4.9 × 10⁻⁶ K⁻¹ to 5.0 × 10⁻⁶ K⁻¹, while the PBT matrix exhibits 80 × 10⁻⁶ K⁻¹ to 100 × 10⁻⁶ K⁻¹ below the glass transition. Because the fibres do not undergo crystallization shrinkage, they act as internal constraints. A 30 % glass-fibre PBT grade tested to ISO 294-4 typically shows parallel-to-flow shrinkage of 0.2 % to 0.5 % and perpendicular-to-flow shrinkage of 0.8 % to 1.1 %. The anisotropy ratio therefore reaches 2.5:1 to 4:1, whereas unfilled PBT remains near 1:1. This difference is the primary reason glass grades warp differently: the final part embeds a residual stress field generated by orientation-dependent shrinkage, not merely a reduced overall shrinkage.

The orientation distribution is not uniform through the thickness. In moulding simulation, the fibre orientation tensor in the shell can show a flow-direction component a11 above 0.8, while the core component may drop below 0.3. Near the cavity surface, fountain flow and high shear align fibres in the flow direction; in the core, fibres are more transverse or randomly oriented. For a 2 mm wall, the oriented shell may be 0.15 mm to 0.35 mm per side, leaving a core of 1.3 mm to 1.7 mm. The shell attempts to shrink 0.3 % in the flow direction while the core attempts to shrink 0.9 % to 1.0 %. This mismatch produces in-plane tensile stresses in the shell and compressive stresses in the core. A flat part can then develop saddle-shaped or corner-lifted distortion because the two principal curvatures have opposite signs.

The ranges in the table are processing responses, not intrinsic properties. Warpage depends on gate location, packing time, and fibre length distribution after compounding. Fibre length after injection moulding is typically 200 µm to 400 µm for a compound initially supplied with 3 mm chopped strands; severe screw shear can reduce this to below 150 µm, lowering the restraint effect. Published data for this exact plaque geometry is limited; ranges reflect compounder technical datasheets and ISO 294-4 shrinkage specimens.

Between 15 % and 30 % glass loading, a property cliff-edge appears. At 10 % to 15 % glass, the fibres reduce overall shrinkage but do not form a continuous restraint network in the flow direction; warpage is still controlled mainly by crystallinity and pack pressure. At 30 % and above, the flow-direction shrinkage can drop below 0.5 % while transverse shrinkage remains near 1 %, so the anisotropy becomes the dominant warp driver. Further increasing glass loading to 45 % to 50 % may reduce total shrinkage but can raise melt viscosity so much that fibre breakage and filling imbalance increase warpage variation. Published data for the exact transition point varies with gate geometry and fibre length, but the step change between 15 % and 30 % is regularly observed in technical datasheets and moulding trials.

When 30 % Glass-Fibre PBT Is Centre-Gated Instead of End-Gated

In a 30 % glass-fibre PBT automotive connector housing moulded on a 1200 kN injection moulding machine with a two-cavity hot-runner valve-gate system, the cavity gate type is often a stronger determinant of warp direction than the nominal mould temperature. If the same housing is end-gated, fibres orient predominantly along the flow direction from the gate, producing greater transverse shrinkage in the width direction; the side walls then bow inward. If the housing is centre-gated, radial flow generates a more complex orientation field: the oriented shell in the radial direction restricts radial shrinkage, while the transverse circumferential direction shrinks more, leading to corner lift or saddle-shaped distortion. Process tests at 265 °C melt temperature and 80 °C mould temperature show that changing from a 1 mm side gate to a 1 mm centre gate can reverse the sign of curvature, even though the material compound and total shrinkage remain unchanged.

The same mechanism affects glass-filled PBT connector housings with snap-fit latching features. Warpage in these components often appears as a difference in latch engagement depth between the two ends, measured at 0.3 mm to 0.8 mm during assembly. This is not a single shrinkage event but a curvature generated by anisotropic orientation. Because glass-filled PBT has a higher tensile modulus, 9000 MPa to 10000 MPa by ISO 527-2, the same geometrical distortion can generate higher assembly forces, making latch damage more likely than in unfilled PBT.

At a shear rate of 1000 s⁻¹, capillary rheometry according to ISO 11443 shows that a 30 % glass-fibre PBT at 265 °C has a melt viscosity approximately 1.3 to 1.8 times that of unfilled PBT at the same temperature. The increased viscosity thickens the frozen skin and shifts the fibre orientation distribution toward the core. Because the shear stress in the shell is higher, the local orientation tensor component a11 increases, while the core remains less oriented. This viscosity-driven orientation gradient is more sensitive to injection speed than unfilled PBT: raising injection speed from 50 mm/s to 150 mm/s can change the shell thickness by 0.05 mm to 0.15 mm and alter flatness on a 2 mm plaque by 0.3 mm to 0.7 mm.

Measurement of warpage on glass-filled PBT should follow ISO 294-4 for shrinkage and be supplemented by flatness measurement after 48 h at 23 °C ± 2 °C and 50 % ± 10 % relative humidity to capture after-shrinkage. ASTM D955-21 provides a similar mould-shrinkage method for thermoplastics; however, its plaque dimensions and gate configuration differ from ISO 294-4, so the two methods should not be interchanged without a correlation study. For glass-filled PBT, the standard shrinkage specimen is often too narrow to reflect production warpage because fibre orientation is strongly influenced by cavity thickness and gate location. A production flatness fixture with three-point support and a dial indicator or laser scanner with ±0.05 mm resolution is required to quantify saddle, twist, and corner lift. If warpage is reported only as average shrinkage, the anisotropic mechanism is masked.

Because PBT undergoes hydrolytic chain scission above 0.02 % moisture, drying conditions impose an operational boundary. PBT and its glass-filled compounds require pre-drying to a residual moisture level below 0.02 % prior to melt processing, typically using a desiccant dryer with a dew point of −40 °C at 120 °C for 3 h to 4 h. At ambient relative humidity above 60 %, regrind and virgin material should be dried immediately before use because hydrolytic degradation reduces molecular weight and changes crystallization rate, which in turn alters post-mould shrinkage and warpage. Melt temperatures above 270 °C or residence times longer than 8 min to 10 min can generate tetrahydrofuran and cause yellowing, gas splay, and warpage variation due to degraded matrix domains. Avoid amine-based additives and certain halogenated flame-retardant systems in glass-filled PBT because they can accelerate hydrolytic attack at processing temperatures.

The fibre-matrix interface also modifies the shrinkage tensor. In glass-fibre PBT compounded with an amino-silane coupling agent, transverse shrinkage is lower than in an uncoupled glass-fibre compound at the same fibre loading because the interface restrains matrix contraction more effectively. The reduction is small in the flow direction but can be 0.1 % to 0.3 % in the transverse direction depending on fibre loading and compound quality. Poorly coupled or severely degraded fibre compounds may show higher transverse shrinkage and lower tensile strength, with tensile strength measured by ISO 527-2 dropping from approximately 130 MPa to below 100 MPa for 30 % glass-fibre PBT. Tensile modulus measured under the same standard also falls by 15 % to 25 % when fibre-matrix adhesion is lost. These changes in mechanical properties should be read as indirect evidence of a weakened interface, because the same interface controls the local shrinkage anisotropy that generates warp.

On the production floor, the difference between unfilled and glass-filled PBT warpage becomes operationally visible in injection pressure and cushion control. Glass-filled grades require higher melt pressure, typically 80 MPa to 120 MPa hydraulic, and produce more screw wear in the check ring. The non-return valve must seal within 0.1 s to 0.2 s to prevent unstable packing; a leaking check ring in glass-filled PBT causes shot-to-shot cushion variation of 1 mm to 3 mm, which directly transfers to warpage variation. Maintenance intervals for the check ring assembly in 30 % glass-fibre PBT are often reduced by 50 % compared with unfilled PBT. These process effects are not material shrinkage properties but they influence the orientation state and packing density that determine final flatness.