Vacuum Assist in Silicone Compression Molding: Eliminating Trapped Air in Deep Cores

A deep core in a compression tool is an air pocket with a deadline. Once the HCR charge hits the land and the cavity seals, that pocket has nowhere to go except into the rubber.

Vacuum assist is not a fancier vent. It is a timed evacuation of the cavity *before* the compound knits across the last air path. Compression molding of millable silicone (HCR) puts a preform on a hot plate, closes, and squeezes. LSR injection can vacuum the closed tool and then shoot a liquid. Compression does not have that luxury unless you design the sequence: load, (optional bump), pull vacuum, close to pressure, cure. If you skip the sequence and hope perimeter vents will empty a 40 mm blind well, you get a knit line, a blister on the core, or a sponge that only shows up after post-cure.

This is HCR compression (and some transfer). It is not room-temperature RTV poured in a vacuum chamber, and it is not LIM with a cold deck. Those three air problems look similar on a cut-up part and are tooled differently.

Why deep cores lose the venting race

Air is displaced by rubber, not absorbed by it. On a shallow gasket the charge sits in the open and the closing mold pushes air out across a wide land. On a deep core, boot, bellows convolution, or hollow with a long mandrel, the geometry closes a trap:

• The rubber front meets around the core before the bottom of the well has emptied.
• A blind pocket at the top of a core (last to fill in a closing press) compresses remaining air into a bubble that prints as a void or a shiny unfilled patch.
• Transfer from a pot can jet past a core and seal the far end first, which is the same trap with a different fill direction.

Perimeter vents at a few hundredths of a millimetre might still work on a thin flash land. They do not empty a closed well. Overflow grooves help only if they connect to the actual trap. Most drawings put overflow where it is easy to machine, not where the air is.

If you ignore the trap location, the defect is a good-looking OD and a porous ID that fails a leak test.

What vacuum assist actually is on a compression press

A vacuum compression cell is a press plus a sealed envelope plus a pump plus a sequence. The envelope is usually:

• A vacuum box or shroud around the mold, or
• O-ring / seal grooves in the mold plates so the *cavity volume* can be pumped down once the tool is nearly closed but not yet on full tonnage.

Typical sequence that actually works:

1. Load a preform that has itself been freshened (warm, not scorching) and shaped so it does not bridge the well and seal air in from the first millimetre of close.
2. Close to a bump / breath gap, or to a position where seals engage but the land is not crushed.
3. Pull vacuum. You need the cavity, not just the room around the press, below a pressure that leaves a harmless residual bubble. Shops argue about the number; the test is whether the cut-up core is solid, not whether the gauge looked impressive.
4. Close to molding pressure and hold through vulcanization.
5. Break vacuum in a controlled way so you do not suck flash into seal grooves or slam plates.

A pump on the press with a leaking shroud is theatre. Watch the vacuum *in the cavity* (or at a port that sees the cavity) versus time. If the curve will not hold when the valve closes, you are evacuating the factory.

LSR injection vacuum is a cousin: evacuate the closed, sealed tool, then inject. Compression vacuum has to happen while there is still a path, which is why the bump-and-pull step exists. Do not copy an LIM vacuum recipe onto an HCR press and call it done.

Charge placement is still the first vacuum

Vacuum will not save a biscuit dropped so it caps the core. Preform shape is DFM:

• A ring charge around a core, not a pancake on top of it, leaves a path down the well as the mold closes.
• A transfer pad feeding from the bottom of a deep feature pushes air toward a vent or vacuum port at the top, which is where you wanted it.
• Too little charge makes the press close on air and then on a short. Too much charge makes an early seal and a flash explosion that looks like you needed more vacuum.

Tin-cure or platinum RTV prototypes of the same geometry can be poured under a bell jar. That is a useful visual of trap locations. It is not a process qualification for HCR compression. The millable compound does not flow like mixed RTV, and the press does not have minutes of open time.

Tool details that make vacuum worth the pump

• Seal the tool. Dedicated vacuum grooves outside the flash land, with channels that cannot be packed shut by the first shot of rubber. If rubber can reach the vacuum port, it will, and then you have a cured plug in a hose.
• Put the port on the trap. A vacuum channel to the bottom of a deep core (or to the last-fill pocket) beats a single port on the parting line.
• Do not substitute vacuum for a vent you still need. Residual gas and peroxide decomposition products (on peroxide HCR) still want a micro-path. Platinum HCR has less of that chemistry, but air remains air.
• Core finish and draft. A deep core with a mirror polish and zero draft vacuum-locks the *part* on the way out. That is a demold problem, not a fill problem. Vacuum assist on close does not excuse a barb on open.
• Multi-cavity balance. Eight deep cores need eight paths. One manifold feeding a hero cavity will leave the far one porous.

If you ignore sealing the vacuum circuit from the compound, the defect is a pump full of silicone crumbs and a slow loss of vacuum over the first week of production.

RFQ notes for vacuum compression

Name the process: HCR compression with vacuum, not “vacuum molding.” Send:

• Core depth, draft, and whether the well is blind.
• Preform concept (ring, slug, transfer).
• Where you will tolerate overflow versus a flashless land.
• Whether the compound is peroxide or platinum HCR (off-gassing and cure chemistry differ).
• Leak-test or cut-up criteria for the core region, not only a visual on the OD.
• Cycle sequence you expect: bump, vacuum time, close, cure.

A shop that only vacuums LSR LIM tools may still be the wrong bidder for a deep-core compression boot. Ask to see a vacuum port in a compression cavity, not a cold-deck brochure.

FAQ

Can I skip vacuum if I add more overflow grooves around a deep core?

Only if the overflow actually connects to the trapped well before the rubber seals. Grooves on the parting line do nothing for air sitting at the bottom of a blind core. Vacuum plus a port on the trap is the usual fix. Extra overflow without a path just makes more flash.

Should vacuum start before the mold touches or after a bump close?

Pull after seals can hold vacuum and before the land is fully closed. An open tool vacuums the room; a fully closed tool has already trapped the well. Time that bump window on the press; do not copy an LSR closed-tool vacuum recipe.

Does platinum HCR need vacuum less than peroxide HCR?

Platinum systems give off less cure gas, so they are kinder on porosity from chemistry. They do not remove the air that was in the cavity. Deep cores still trap that air. Treat vacuum as an air-evacuation tool, not as a cure-chemistry band-aid.

Will vacuum assist stop knit lines in a multi-piece preform?

It helps the voids that are air. It will not weld two cold biscuits that never saw pressure and heat together. Knit lines are a charge-layout and temperature problem. Vacuum after a bad load still leaves a visible weld in the rubber.

How do I know the vacuum is reaching the core, not just the shroud?

Instrument a port that sees the cavity, or run a short-shot / bump trial and cut the core. A gauge on the pump with a leaking box will look fine. Solid rubber at the bottom of the well is the acceptance test, especially on a leak-critical ID.