Polyoxymethylene (POM) is a semicrystalline engineering thermoplastic with a crystallinity of roughly 60% to 80% depending on homopolymer or copolymer grade, a tensile modulus near 2700 to 3100 MPa per ISO 527-2, and a linear coefficient of thermal expansion of approximately 100 to 120 µm/m·°C per ASTM E831. Injection-molded POM parts are rarely dimensionally stable immediately after demolding because melt-state shear, non-uniform mold cooling, and packing-pressure decay generate anisotropic crystallinity and residual stress. Post-molding annealing is a thermal exposure below the melt point that allows chain relaxation, secondary crystallization, and partial release of frozen-in free volume. The effect on dimensions is specific to the cause of the deviation: annealing can correct process-induced non-equilibrium shrinkage, but it cannot repair errors produced by tooling geometry, asymmetric filling, or predetermined machining stock loss.

Annealing POM Parts: When It Fixes Dimensions and When It Does Not

Annealing produces a measurable dimensional correction when the deviation is caused by process-induced anisotropy in the molded shell. In molds held below 70°C, the outer layer solidifies before the core can compensate for volume loss, creating a high-modulus shell with tensile stress and a delayed-shrinkage core. Air annealing in a recirculating oven at 150°C to 160°C for homopolymer, or 140°C to 150°C for copolymer, with a soak time of 30 min plus 5 min/mm of wall thickness, reduces residual stress and allows additional lamellar thickening. The resulting permanent shrinkage is typically 0.1% to 0.4% in the flow direction when measured according to ISO 294-4. However, the same annealing cycle will not correct a diameter that is oversized because the mold cavity was cut incorrectly; the part may shrink uniformly, but the nominal dimension remains outside tolerance.

Residual stress relief is confirmed by differential scanning calorimetry under ISO 11357-3. Annealed POM exhibits a narrower melting endotherm and a shift in the peak melting temperature of 1°C to 3°C relative to unannealed parts. The dimensional benefit is largest in parts molded with cold tooling and high injection velocity, because the non-equilibrium layer is thicker. In heated molds above 120°C, the as-molded crystallinity is closer to equilibrium, and post-annealing shrinkage is commonly below 0.05%. Thus annealing fixes dimensions when the deviation is dominated by thermal history and orientation, not when it is dominated by cavity geometry.

What Limits Dimensional Correction at the Fixture Level?

In a hot fixture, annealing cannot correct warpage that arises from asymmetric wall thickness, knit lines, or gate-induced orientation if the part is allowed to deform freely during heating. A flat, ribbed cover with a thickness ratio of 3:1 between rib and nominal wall will develop differential shrinkage during cooling; annealing relaxes stress but also releases the elastic strain that had held the part flat, increasing bow and twist. If a fixture is used to constrain the part, the contact pressure must be uniform below the creep modulus threshold; otherwise the annealed part takes the shape of the hot fixture rather than the intended geometry. Warpage measured after annealing is most repeatable when parts are supported on silica-free ceramic rods or pre-heated aluminum beds with a flatness deviation of less than 0.05 mm per 100 mm span. Aluminum fixtures introduce thermal expansion mismatch because aluminum expands at roughly 23 to 25 µm/m·°C, approximately one-fourth the rate of POM; rigid clamping can therefore indent or stretch the part during the thermal ramp. Published data for clamped POM fixtures with non-uniform contact pressure is limited; tooling shops therefore qualify annealing fixtures by dimensional capability runs rather than by predictive simulation alone.

From a thermal-history perspective, residual stress relaxation in POM is not a single-stage event. The first 20 min of annealing primarily release frozen-in free volume and oriented amorphous regions, while later stages involve secondary crystallization and lamellar thickening. A gear annealed for 30 min may show different pitch-diameter change than the same gear annealed for 120 min, even at the same setpoint, because the slower crystallization process continues after stress relaxation is largely complete. This is why dimensional control procedures specify both soak time and part thickness. A thin section of 1.5 mm reaches thermal equilibrium quickly, while a thick section of 8 mm may require ramp rates below 2°C/min to avoid overshoot and internal temperature gradients that create new residual stress.

Residual Stress, Secondary Crystallization, and Water Content Gradients

Thermal annealing does not always equal dimension correction because three independent processes proceed simultaneously: residual stress relaxation, secondary crystallization, and moisture desorption from surface layers. POM has low water uptake, typically 0.2% to 0.4% at saturation per ISO 62, so moisture-driven swelling is minor compared with polyamide. Nevertheless, a moisture gradient of 0.1% between surface and core can alter surface planarity by several tenths of a millimeter on a large flat part. Vacuum drying at 80°C for 2 h to 3 h is applied before annealing if parts were stored at relative humidity above 60%. Desorption during the first 20 min of annealing can create transient shrinkage that is mistaken for thermal correction. Stabilization against oxidative formaldehyde evolution is required above 150°C; thermal stabilizer packages are selected for long soak times to limit molecular-weight loss and surface pitting.

Across a batch of 50 parts, forced-air oven temperature variability as low as ±3°C can produce detectable dimensional scatter because the rate of secondary crystallization increases sharply near the alpha-relaxation of POM. Batch-to-batch variation in melt mass-flow rate, measured by ISO 1133-1 at 190°C/2.16 kg, changes orientation and post-anneal shrinkage. High-flow grades with MFR above 25 g/10 min orient less and may anneal with lower flow-direction shrinkage than low-flow grades with MFR below 7 g/10 min. Therefore a fixed annealing correction cannot be applied across material lots without dimensional capability data from at least 30 consecutive shots per cavity.

If Annealing Temperature Exceeds 160°C in Unstabilized Homopolymer

An upper temperature boundary exists. Annealing homopolymer POM above 160°C, or exceeding 180 min at 150°C, initiates oxidative degradation and surface pitting. The melt temperature of homopolymer POM is approximately 175°C per ISO 11357-3; the practical annealing limit is therefore maintained at least 15°C below the melt peak. At 170°C, parts soften sufficiently that self-weight deflection under a 0.45 MPa load can exceed the ISO 75 HDT limit, and thin sections can collapse. Degradation is autocatalytic because formaldehyde oxidation product forms formic acid, which lowers molecular weight. Acetal copolymer tolerates slightly higher thermal exposure because the -CH2-CH2- comonomer interrupts unzipping, but the safe air-annealing range remains 140°C to 150°C for most production grades. Parts with a wall thickness above 10 mm require staged heating at 2°C/min or less to prevent internal temperature overshoot.

Tooling compensation derived from annealed shrinkage coupons is valid only when the annealing cycle exactly matches the production cycle. The mold cavity is cut oversized by the sum of mold shrinkage and post-anneal shrinkage only if the part is oven-annealed after molding. If annealing is intended, the cavity should be cut to the high side of tolerance because post-shrinkage can move dimensions down by 0.05% to 0.15%. A gate diameter below 0.8 mm on a 3 mm wall will freeze before packing completes, creating a non-equilibrium skin that annealing cannot fully correct; the resulting post-anneal diametric tolerance is not recoverable.

Machined POM components respond to annealing differently from injection-molded parts because the residual stress state is compressive at the surface and tensile in the interior. A close-tolerance bushing machined from extruded POM and then annealed may exhibit bore closure or bellmouthing as machining stress is released. Published data for this specific configuration is limited; therefore boring after anneal is preferred when the final tolerance is below 0.02 mm. Annealing of a machined gear after hobbing can improve pitch-diameter stability in later thermal cycling, but it cannot correct a tooth-profile error produced by cutter runout. The anneal only shifts the residual stress state; it does not regenerate removed stock.

When dimensional deviation originates from surface degradation, acid exposure, or oxidative pitting, annealing does not restore the original surface profile. Strong acids such as hydrochloric acid and oxidizing agents can attack POM at the surface, causing weight loss and local dimensional erosion. The affected layer is not regenerated by thermal treatment; the part must be replaced or re-machined if stock remains. For this reason, annealing should be separated from cleaning operations that use acidic or amine-based agents, because residual surface contamination can accelerate degradation during the thermal soak. The dimensional inspection after annealing is performed at standardized laboratory conditions of 23°C and 50% relative humidity per ISO 291 to avoid transient thermal expansion masking the true annealed dimensions.