Bubbles and Voids in Silicone: Air, Moisture and the Vacuum That Was Too Late

Two kinds of bubble, two different problems

A void in a silicone part is either air that was never removed or water that turned to steam. They look similar in a finished part and they are completely different problems, and the reason it matters is that one is fixed by the vacuum and the other is not.

Air voids are usually round, they are distributed through the part, and they are larger near the last point to fill. Moisture voids are also round but they tend to cluster near the surface and near thick sections, because steam forms where the temperature is highest and escapes towards the nearest free surface.

The distinguishing test is simple. Cut the part and look at the void wall. An air void has a smooth, glossy interior that looks like the outside of the part. A moisture void has a slightly rougher, sometimes whitish interior, because the steam has left a different surface behind.

The second test is to change one thing. Dry the compound and run again. If the voids go away it was moisture. If they do not, it was air, and the vacuum is where to look.

Table 1 — Telling the two apart on a cut section

Air: the vacuum was not too weak, it was too late

In compression or transfer moulding, the vacuum is applied to the tool before the material enters. The idea is to evacuate the cavity and the pot so that the material fills a space with no air in it. The variable that matters is when the vacuum starts relative to the injection, not how deep it goes.

Measured on a 100 by 100 mm block, starting the vacuum 20 seconds before injection gives 25 bubbles per 100 square centimetres. At 15 seconds it is 27, at 10 seconds 34, at 5 seconds 61, at 2 seconds 120, and with no vacuum at all it is 186.

The curve is steep between 0 and 5 seconds and almost flat after 10. That shape is the clue: the cavity is evacuated in a few seconds, and everything after that is the material itself degassing. So a vacuum that starts 2 seconds before injection has not evacuated the cavity, and a vacuum that starts 20 seconds before has evacuated it and had time to pull air out of the compound as well.

The other half of the air problem is where the air is. A cavity with a blind corner or a deep thin rib traps air that the vacuum cannot reach, and no amount of hold time will remove it. Those are the voids that appear in the same place on every part, and they are a venting problem rather than a vacuum problem.

Moisture: the one no vacuum fixes

Silicone rubber absorbs water. Not much, and not fast, but a bag of compound stored in a humid room will pick up 0.2 to 0.4 percent over a few weeks, and a two-part liquid silicone will pick it up faster because it is already a paste.

At moulding temperature that water becomes steam, and steam expands by a factor of over a thousand. A 0.1 percent moisture content in a 10 gram shot is 10 milligrams of water, which is about 12 cubic centimetres of steam at 170 C. In a cavity that holds 8 cubic centimetres, that is a void.

The numbers are brutal. At 0.02 percent moisture a block shows 9 voids per 100 square centimetres. At 0.05 percent it is 14, at 0.10 it is 31, at 0.20 it is 78, and at 0.35 it is 164. A vacuum cannot remove water that is dissolved in the compound, because the water is not in the gas phase until the material is hot, and by then the cavity is closed.

The fix is drying, and it is a controlled process rather than an hour in an oven. Solid silicone is dried at 80 to 100 C for 2 to 4 hours in a tray no more than 20 mm deep. Liquid silicone is harder, because it is a filled paste and the water diffuses slowly; a vacuum mixer or a dry-air blanket over the drum is the usual answer.

Storage matters as much as drying. A dried compound left in an open bag in a humid plant reabsorbs most of what it lost within a shift. Sealed foil bags, desiccant and a first-in-first-out rule are part of the process, not housekeeping.

Table 2 — What moisture costs, and what removes it

Where the air actually comes from

Even with a good vacuum, air gets in from four places. The first is the charge itself: a rolled preform or a cut slug carries air between its layers, and that air has to escape before the material consolidates. A preform that is rolled loosely carries more than one that is tight.

The second is the pot and the transfer channel. A transfer pot that is not evacuated holds air above the material, and that air is pushed into the cavity ahead of the slug. A pot with a vacuum port at the top and a seal at the plunger is a pot that does not do this.

The third is the tool. A parting line that does not seal, an ejector pin with clearance, a core that does not bottom: all of them are paths for air to be drawn in during the vacuum and pushed in during the injection. A tool that holds vacuum for 30 seconds is a tool that is sealed; one that loses it in 5 seconds is not.

The fourth is the material itself. A highly filled compound has air adsorbed on the filler surface, and it is released slowly. That is why a highly filled compound needs a longer vacuum hold than an unfilled one, and why a compound change can bring back voids that were solved.

Detection, and the cut test that settles it

The standard production check is a cut. Take a part, cut it through the thickest section, and count voids on the cut face against a comparator. It is destructive, it is slow, and it is the only method that sees inside the part.

X-ray and ultrasonic inspection will find voids non-destructively and they are used on medical and optical parts. They will not tell you whether the void is air or steam, and they will not tell you where in the process it came from.

A weight check catches gross porosity. A part with 5 percent voids weighs 5 percent less than a solid one, and a 0.5 percent weight tolerance will catch it. What it will not catch is a few large voids in a thick section, which is exactly the failure mode that matters for a seal.

The most useful production number is not the void count but the void count at a fixed location. A part with 20 voids scattered at random is a process that needs tuning. A part with 3 voids in the same corner every time is a tool that needs a vent.

The order to check them in

The order matters because the cheap checks are also the ones that are usually skipped. Start with the compound: measure the moisture on the lot in use. If it is above 0.05 percent, dry it and run again. That single check resolves the majority of void complaints and it costs ten minutes.

Then check the vacuum timing. Log when the vacuum valve opens relative to the injection signal. A 2 second lead looks fine on a stopwatch and gives 120 bubbles where 20 seconds gives 25.

Then check the tool seal. Pull a vacuum on an empty tool and watch the gauge for 30 seconds. A tool that loses vacuum is a tool that will make voids, whatever else is changed.

Then check the venting. Any void that appears in the same place every time is a venting problem, and no amount of vacuum, drying or hold time will move it.