Manufacture of athletic footwear midsoles from ethylene-vinyl acetate copolymer foam requires simultaneous control of peroxide crosslinking and chemical blowing agent decomposition inside a hot injection mold. A representative production compound contains EVA with a vinyl acetate content of 18–28 wt%, dicumyl peroxide at 0.5–1.2 phr, azodicarbonamide at 2–5 phr, zinc oxide at 1–2 phr, zinc stearate at 0.5–1.0 phr, and a triallyl isocyanurate coagent at 0.2–0.5 phr. The apparent density target measured according to ISO 845:2009 is typically 0.20–0.25 g/cm³. On a production injection molding machine with a clamp force of 300–450 t, a screw diameter of 70 mm, an L/D ratio of 22:1, a shut-off nozzle, and barrel zones set between 85 °C and 95 °C, the compound is injected into a mold heated to 165–175 °C. Cavity pressure during filling, packing, and the subsequent pressure-quench foaming step determines gas solubility, bubble nucleation rate, local crosslink density, and final part density. The injection mold pressure therefore acts not as a single setpoint but as a time-dependent profile that must be matched to the decomposition kinetics of both dicumyl peroxide and azodicarbonamide.

In industrial practice, the pressure signal from flush-mounted piezoelectric cavity pressure transducers such as Kistler 6182C sensors is used to trigger the mold-opening or core-back sequence. The pressure at the gate may exceed 120 bar during filling while the last-fill region can remain below 30 bar until the mold is fully packed. This spatial gradient creates density and cell-size differences across the midsole unless the runner geometry, valve-gate timing, and injection speed profile are balanced. Because the melting point of EVA increases with pressure and the dissolved blowing gas plasticizes the melt, the relationship between pressure and cure is not linear. The pressure profile must therefore be interpreted at the cavity level, not at the hydraulic cylinder level.

Thermal and Pressure Coupling in Peroxide-Cured EVA Foam Systems

Dicumyl peroxide decomposes by first-order homolytic cleavage of the oxygen-oxygen bond. The half-life of dicumyl peroxide is approximately 10 h at 115 °C, 1 h at 135 °C, and 1 min at 171 °C. The activation energy for decomposition is approximately 157 kJ/mol. In EVA, the liberated cumyloxy radicals abstract hydrogen from the polymer backbone, producing polymer radicals that combine or react with the triallyl isocyanurate coagent to form a three-dimensional network. Crosslink density can be calculated from equilibrium swelling measurements in toluene using the Flory-Rehner equation: ν_e = -[ln(1-V_r) + V_r + χV_r²] / [V_s (V_r^(1/3) - V_r/2)], where V_r is the polymer volume fraction in the swollen gel, χ is the EVA-toluene interaction parameter, and V_s is the molar volume of toluene. In production audits, gel fraction determined by extraction in boiling xylene for 24 h is used as a simpler proxy for cure state. Higher cavity pressure alters the free volume available for radical termination and diffusion. This produces a secondary but measurable effect on gel fraction. Published pressure-specific kinetic data for injection-molded EVA foam are limited; however, process records from multi-cavity midsole tools indicate that a cavity pressure increase from 50 bar to 150 bar at the same mold temperature can shift the final gel fraction by 2–4 percentage points, with the higher pressure generally increasing cure efficiency through reduced macromolecular mobility and enhanced radical recombination. The effect is smaller than a 5 °C change in mold temperature, but it becomes relevant when the processing window is constrained to ±5 °C.

Dry-blended EVA compounds are mixed in an internal mixer with ram pressure 5–7 bar and fill factor 0.75–0.85, with a discharge temperature not exceeding 105 °C. After two-roll milling and granulation, the compound is dried at 60–70 °C for 4 h when ambient relative humidity exceeds 60 %. The dried granules are fed to the injection unit under a blanket of dried air. Barrel temperatures are maintained below 100 °C in the feed and compression zones to avoid scorch, while the nozzle is kept at 95–105 °C. Hydraulic back pressure is set between 2 bar and 6 bar to homogenize the gas-laden melt without inducing premature decomposition. Compression ratio of the screw is typically 2.2:1 to 2.8:1, with a low-shear mixing section to distribute zinc oxide and zinc stearate. Residence time of the melt above 100 °C must not exceed 15 min; exceeding this boundary leads to scorch, visible as hard gel particles in the molded midsole. Amine-based antioxidants are incompatible with this compound because amine residues accelerate peroxide decomposition and shift the scorch time to a lower temperature.

What Limits Cell Nucleation When Melt Pressure Exceeds 60 Bar?

Azodicarbonamide decomposes in the presence of zinc oxide and zinc stearate at temperatures of 150–170 °C, releasing a gas mixture dominated by nitrogen with smaller fractions of carbon monoxide, carbon dioxide, and ammonia. The theoretical gas yield is approximately 220 ml/g at standard temperature and pressure. In the injection barrel, the external pressure prevents the gas from forming bubbles because the saturation pressure of dissolved gas in EVA rises with pressure according to Henry’s law. When the cavity pressure exceeds the saturation pressure, the melt remains a single-phase polymer-gas solution. Bubble nucleation is governed by the classical nucleation rate J = A exp(-ΔG*/kT), where the critical free energy barrier ΔG* is proportional to 16πγ³/(3ΔP²). The term ΔP is the difference between the local gas pressure and the melt pressure. As long as the melt pressure remains high, ΔP is small and ΔG* is large, so nucleation is suppressed. Therefore, cell nucleation is not limited by the blowing agent decomposition itself but by the pressure-quench step that follows mold filling. If the melt pressure exceeds 60 bar throughout the cavity, nucleation is effectively postponed even at 170 °C. Once the mold opens or the core retracts, the melt pressure drops at rates that can exceed 50–200 bar/s, increasing ΔP and reducing ΔG* by orders of magnitude. The resulting cell number density in EVA midsoles can range from 10⁴ to 10⁶ cells per cm³ depending on the pressure drop rate, nucleation site density, and melt strength at the moment of expansion. Excessively high cavity pressure before mold opening also reduces the volume of free gas available for expansion because more gas remains dissolved; this shifts the final density upward and produces a thicker solid skin.

The injection mold pressure profile in a core-back or mold-breathing midsole tool consists of four distinct stages. In the first stage, the cavity is filled at high speed while the gas remains dissolved. In the second stage, a packing pressure is applied to compensate for thermal expansion of the melt and to hold the skin against the mold wall. In the third stage, the crosslinking reaction reaches the gel point and the mold is opened by 10–20 mm at a controlled speed. In the fourth stage, the part remains in the mold until sufficient crosslink density is obtained, then the mold is opened and the part is removed. The exact moment of the pressure-quench step is determined not by the cavity pressure alone but by the integral of cavity pressure and time, which accounts for the thermal history of the part. Off-line moving die rheometer data at 170 °C and 0.5° arc according to ISO 6502-2:2018 provide t_s2 scorch times and t_90 cure times that are used to set this trigger. In a six-cavity valve-gated midsole tool with cavity pressure sensors at the gate and the last-fill region, documented audits show that a cavity-to-cavity pressure variation of ±8 bar produced a density variation of ±0.02 g/cm³ and a cell diameter variation of ±20 µm. The last cavity in the runner chain normally operates at the lowest pressure and can exhibit premature nucleation at the flow front if the local pressure falls below the gas saturation pressure before the pressure-quench stage.

Gate design and gate freeze time interact with packing pressure delivery. In EVA foam injection, the mold is heated to 165–175 °C, so a cold runner gate may freeze slowly or remain open until after the pressure-quench step. If the gate remains open during mold opening, melt can be pushed back from the cavity into the runner, reducing cavity pressure and causing density asymmetry. Valve gates prevent this backflow but require precise needle timing. A valve gate in a midsole mold is typically closed after the packing phase but before the mold opening begins. Closure pressure at the gate can fall between 40 bar and 80 bar. If the gate closes too early, the part cannot be fully packed and the midsole develops low-density regions near the gate and high-density regions at the flow front. If the gate closes too late, the pressure-quench step is blunted by continued melt supply and the expansion ratio is reduced. The microcellular structure of the final midsole therefore depends as much on gate timing as on the absolute injection pressure.

When Packing Pressure Persists After Crosslinking Onset

When packing pressure persists after the peroxide-induced gel point has been reached, the crosslinking network forms while the gas is still dissolved and the melt is compressed. The result is a high-density foam with thick cell walls, a low expansion ratio, and a hard surface skin. If the mold is then opened after full cure, the residual gas pressure is insufficient to expand the network, and the part remains denser than the specification. Conversely, if the mold opens before the gel fraction reaches a critical range, the melt lacks sufficient extensional viscosity and cell walls rupture. The practical gel fraction window for mold opening in EVA midsoles is typically 40–60 % as measured by xylene extraction. At gel fractions below 40 %, cell coalescence and surface pinholes increase sharply. At gel fractions above 60 %, the expansion ratio decreases and shrinkage after demolding becomes more variable because the residual gas cannot be accommodated by the cured network. The exact value depends on vinyl acetate content, coagent type, filler loading, and local mold temperature. Process engineers therefore use a pressure-time integral trigger rather than a fixed timer: the mold-opening signal is generated when the area under the cavity pressure curve reaches a calculated threshold corresponding to the target gel fraction. This approach compensates for batch-to-batch variations in compound viscosity and mold temperature.

Process conflicts are most severe in thick midsole sections where the core reaches cure later than the skin. The skin adjacent to the mold wall crosslinks first because it is heated rapidly. The core remains molten and gas-laden longer. If the pressure-quench step occurs too early relative to the skin gelation, the skin tears and gas escapes through fissures. If the step occurs too late, the core cannot expand and the part contains a dense interior with an over-cured skin. Cavity pressure sensors used with a sampling rate of 100 Hz or greater capture the local pressure decay during mold opening. This data reveals that the core can remain at 5–15 bar positive pressure for several seconds after the skin has begun to solidify. The rate of mold opening has a direct effect on pressure drop rate; opening speeds of 5–20 mm/s are used for midsoles depending on thickness. At an opening speed above 20 mm/s, pressure drop rates can exceed 200 bar/s and produce fine cells but also increase the probability of internal tearing. At an opening speed below 5 mm/s, the pressure release is gradual and cell nucleation is sparse, producing coarse cells with average diameters above 150 µm.

Midsole Property Gradients Across Mold Thickness

The combination of a hot mold, a compressed melt, and a dissolved gas phase creates a density gradient across the midsole thickness. The outer skin is formed when the melt contacts the mold wall and crosslinks rapidly; this skin has a lower cell density and a higher polymer density than the core. The core expands as the pressure is released, producing larger, elongated cells. Density measured on sliced specimens according to ISO 845:2009 can vary by 0.03–0.06 g/cm³ from skin to core in a midsole of 20 mm thickness. Hardness measured with a Shore C durometer according to ASTM D2240-15 is typically 50–60 at the surface and 40–50 in the center. Compression set determined according to ISO 1856:2018 at 50 % deflection and 23 °C is usually lower near the surface where the polymer network is denser and higher in the expanded core. Rebound resilience determined according to ISO 8307:2018 can decrease by 8–12 % from skin to core because the larger core cells dissipate more energy through gas compression and cell-wall deformation. This gradient is not necessarily a defect; midsole designers often use a denser skin for abrasion resistance and a softer core for cushioning. However, uncontrolled pressure gradients along the flow path create asymmetrical cushioning and a product that fails the dimensional stability requirements of footwear quality control.

Dynamic mechanical analysis of pressure-molded EVA midsoles shows that the storage modulus at 25 °C falls between 3 MPa and 8 MPa depending on foam density and crosslink density. Tan δ at 25 °C typically ranges from 0.15 to 0.25. Lower tan δ is associated with lower heat buildup during cyclic loading and higher energy return in the midsole. The pressure history influences tan δ indirectly through cell size and skin density; a high packing pressure creates a dense skin that increases storage modulus and reduces the viscoelastic contribution from cell-wall bending. Fatigue testing of production midsoles under repeated compression at 3 Hz shows that samples molded with excessive cavity pressure above 120 bar can exhibit earlier surface cracking because the dense skin is highly crosslinked and less tolerant of strain. Samples molded below 50 bar often fail by internal cell collapse. These failure modes are evaluated by visual inspection and thickness loss measurement after 100,000 cycles.

Rheological behavior of the EVA compound before crosslinking is shear-thinning. Melt flow index measured at 190 °C under 2.16 kg according to ISO 1133-1:2022 typically ranges from 3 g/10 min to 10 g/10 min for midsole grades. The dissolved blowing gas and decomposition products reduce melt viscosity, while the onset of crosslinking increases viscosity. High injection pressure increases shear heating; in a 70 mm screw at high screw speed, the melt temperature can rise by 8–12 °C between the feed throat and the nozzle. This temperature rise is critical because dicumyl peroxide decomposition accelerates with temperature. If the melt entering the mold reaches 120 °C, the remaining safe residence time drops sharply and the compound may begin to scorch in the runner or gate. Injection speed profiles are therefore adjusted to limit shear heating while still filling the cavity before the mold heat initiates cure. A common profile uses a low initial speed of 10–20 % of maximum to fill the runner, a high middle speed of 70–85 % to fill the cavity, and a low final speed of 10–15 % to avoid flash and overpacking.

Gas counter pressure molding is used in some EVA midsole applications to suppress premature nucleation during filling. Nitrogen is introduced into the sealed cavity at 10–30 bar before injection; this pre-pressurization increases the local saturation pressure and prevents bubble formation at the flow front. After filling, the gas is vented rapidly and the pressure drop triggers uniform nucleation. The technique produces cell sizes below 80 µm and improves surface quality compared with conventional pressure-quench molding. The mold must be sealed with O-rings and the gas flow controlled by a dedicated pressure intensifier. The additional equipment cost is offset by a wider processing window for the injection pressure. When gas counter pressure is used, the injection pressure can be reduced by 10–20 % without sacrificing part density because the usual premature nucleation at the flow front is suppressed.

Mold design for EVA midsole injection requires venting channels that allow gas escape without allowing the melt to flash. Vent depths are typically 0.02–0.05 mm. Vents placed at the last-fill region remove air and decomposition gas before the pressure-quench step. Vacuum assist at -0.8 bar can reduce gas entrapment and improve skin quality. The mold parting line must withstand the expansion force generated during foaming. Clamp force required is calculated from the projected area and the maximum cavity pressure during expansion. For a four-cavity mold with 180–250 cm² projected area per cavity and a maximum cavity pressure of 80–120 bar, the required clamp force is between 300 t and 500 t depending on the number of cavities and platen deflection limits. Platen deflection above 0.15 mm can produce flash and change the local cavity pressure distribution, so the machine must be selected with adequate clamp stiffness.

Pressure also affects azodicarbonamide decomposition indirectly through its influence on local temperature. The decomposition exotherm can raise the local melt temperature by 5–15 °C depending on blowing agent concentration and heat transfer. In a sealed cavity under high pressure, the exotherm is constrained but not eliminated. The temperature rise accelerates dicumyl peroxide decomposition, creating a coupled reaction front. In thick midsoles, the center can reach a higher temperature than the mold setpoint because of the exotherm and poor heat conduction. Cavity pressure sensors record a pressure rise of 3–8 bar as the gas is released and the melt expands against the mold. This pressure rise can be used as an indirect indicator of blowing agent decomposition and helps determine the correct moment for pressure release. If the pressure rise is absent, the blowing agent may not have fully decomposed or the pressure is too high for gas evolution.

Microcellular characterization by scanning electron microscopy of fractured freeze-fractured specimens shows that cell diameter and cell density depend on the pressure drop rate and the pressure level before release. Cell wall thickness in midsole foams typically ranges from 5 µm to 20 µm. Thicker walls are associated with higher packing pressure. The cell size distribution narrows when the pressure drop rate is high and uniform across the cavity. If the pressure drop is non-uniform, bimodal cell size distributions appear. Image analysis is performed on a minimum of 500 cells per specimen to reduce sampling error. This method is not standardized in a single ISO method but is reported alongside physical property data to explain batch rejection.

Dimensional stability of the molded midsole is measured by leaving the part at 23 °C and 50 % relative humidity for 24 h before measuring length and width. Parts foamed with excessive residual gas pressure can grow after demolding, while parts with insufficient expansion can shrink. A process audit based on cavity pressure trace and post-mold dimensional data can establish a corridor of acceptable pressure-time integral values. If the integral falls below the lower boundary, the part is soft and oversized; if it exceeds the upper boundary, the part is dense and undersized. The acceptable corridor may be as narrow as ±10 % of the target integral for a given compound and mold temperature. This narrow corridor is one reason why injection molding of EVA foam midsoles is more challenging than compression molding.

Flow simulation of EVA foam injection molding can be performed with commercial software using measured PVT behavior and cure kinetics. The simulation requires pressure-dependent melt viscosity, blowing gas solubility, and the gel fraction as a function of time and temperature. Without these data, simulation results are indicative only. Plant validation of simulation results uses cavity pressure traces and density maps. The best correlation between simulation and production is obtained when the cavity pressure prediction error is below 10 bar across the flow path. Above this error, the predicted pressure-quench timing becomes unreliable and the mold-opening trigger must be confirmed experimentally.

Batch-to-batch variation in EVA melt viscosity and zinc oxide dispersion affects the pressure response. A viscosity increase of 10 % at the same machine injection pressure can reduce cavity pressure by 8–15 bar because of increased pressure drop in the runner and gate. The mold-opening trigger based on cavity pressure integral partially compensates for such variation. If the trigger relies only on time, a low-viscosity batch can crosslink too early for the planned pressure release, and a high-viscosity batch can open too late. That is why production-scale equipment behavior on midsole lines is monitored by in-mold sensors, not by hydraulic pressure alone. The sensor signal also reveals the repeatability of the pressure trace; a standard deviation of cavity pressure greater than ±5 bar across consecutive shots usually corresponds to an observable density variation and requires adjustment of check-ring sealing, barrel temperature profile, or back pressure.

Chemical blowing agent residues and peroxide decomposition products in the molded midsole must comply with applicable footwear material restrictions. Azodicarbonamide decomposition residuals and acetophenone are monitored in finished parts, and ventilation in the molding area must keep airborne concentrations below occupational exposure limits. Compliance declarations for the EVA compound commonly reference REACH Annex XVII and the RoHS Directive 2011/65/EU. Because these regulations do not define midsole physical performance, the mechanical tests are reported separately according to the ISO and ASTM methods cited above. The pressure history does not remove residual chemical species; it changes their distribution between the dense skin and the cellular core. High packing pressure can entrap volatile residues in the core and increase odor unless post-curing is sufficient. Post-curing at 60–70 °C for 2–4 h is used to reduce residual gas content and stabilize dimensions before assembly.

Troubleshooting of pressure-related defects requires direct cavity pressure data rather than machine hydraulic pressure. Hydraulic pressure does not compensate for runner pressure drop, gate restrictions, or viscosity changes caused by blowing agent decomposition. A midsole with low density near the gate and high density at the flow front often indicates a gate freeze before complete packing. A midsole with elongated cells and internal tears often indicates mold opening before the gel fraction reached 40 %. A midsole with excessive skin thickness and a hard feel often indicates mold opening after the gel fraction exceeded 60 % or packing pressure was too high. Surface pinholes at the last-fill region indicate that the local pressure fell below the gas saturation pressure during filling, allowing premature nucleation. Center blowholes indicate gas coalescence in the core before the network gained sufficient strength. Shrinkage after demolding can be measured according to ISO 2796:2004 and is minimized by post-curing the part in a ventilated oven at 60–70 °C for 2–4 h to allow residual gas to diffuse out while air diffuses in. The pressure profile must therefore be interpreted together with mold temperature, gate timing, compound rheology, and cure kinetics. Published pressure-specific kinetic data for this exact injection molding configuration remain limited, so production lines rely on instrumented molds and destructive testing to establish operating boundaries.