Micro-Molding Custom Silicone Parts: Maintaining Tolerances Below 0.05mm

0.05 mm is not a silicone superpower. It is a stack of tool, grade, heat, and metrology that only some micro parts can hold.

Micro-molding here means small LSR (liquid silicone rubber) parts: features on the order of a millimetre or less, shot weights in the milligram-to-small-gram band, often in multi-cavity cold-runner tools. A print that says “±0.05 mm everywhere” on a 12 mm seal is a different conversation from ±0.05 mm on a 0.8 mm valve slit. ISO 3302-1 already exists for rubber dimensions. Class M1 is tight. Sub-0.05 mm on a micro feature can sit inside or outside that class depending on the nominal. Treating 0.05 mm as a shop default is how an RFQ gets quoted as “yes” and measured as “no.”

HCR compression and RTV pours can make small parts. They are not the process that holds this band in volume. This article is LSR injection in steel, with aluminum tryout as a warning.

The 0.05 mm budget is spent before you open the box

Four movers eat the same hundredths:

Shrinkage of the grade. LSR linear shrink is typically in a roughly 1.5-3% band, and it moves with cavity temperature, hold, post-cure, and filler. A 2% shrink on a 2.5 mm OD is 0.05 mm by itself. If the shrink number on the datasheet was measured on a different plaque, a different heat, and a different post-cure, your micro OD is a guess. Compensate the steel from *your* process window, then freeze the window.

Hot steel versus cold rubber. The cavity is at about 160-200 °C. The part is measured at room temperature. Steel growth and rubber shrink both live in the 0.05 mm band on small inserts. A dimension cut to print at 20 °C on a CMM of the *tool* is not the dimension of the part.

Flash and vestige. LSR leaks through gaps on the order of 0.005 mm. A “±0.05 mm” OD that includes a 0.03 mm flash flake is already spent. Gate vestige on a micro part can be the whole tolerance. Valve-gated cold runners exist to make that vestige small and repeatable, not zero by slogan.

Metrology on a soft part. A contact CMM probe will indent 10-30 Shore A and report a small dimension. Optical / vision / CT are kinder, and they still argue about edge detection on a flash line and a radius the CAD called sharp. If you ignore how the part is measured, you and the customer are not talking about the same 0.05 mm.

If you ignore post-cure, the defect is a first article that passed green and drifted after the oven.

Tooling that can even try

Micro LSR tooling is not a scaled-down plastic toy mold:

• Hardened, well-guided inserts. Plate shift of 0.02 mm is a cosmetic step on a bumper and a fail on a micro slit.
• Shut-offs and vents in the micron-to-tens-of-microns depth band so you fill without flooding the land. Vacuum on the cavity is common because a bubble is a 0.05 mm void.
• Valve-gated cold deck with drops sized for milligram shots. A waste sprue can outweigh the part and unbalance fill.
• Heater control that represents the *insert*, not the block. A 10 °C swing changes shrink and cure on a thin micro wall.
• Cavity count you can balance. Sixteen micro cavities with one hero and one starved cavity will never all sit in ±0.05 mm.

Aluminum micro inserts are for learning flow and gate vestige. They will not hold this tolerance band for a production life. Steel, then process lock, then capability.

DIN ISO 3302-1 / ISO 3302-1 should be on the drawing as the language of rubber tolerances. If you need tighter than the class for a named feature, name that feature. A blanket ±0.05 mm on every radius and flash land is not a micro-molding spec. It is a fight.

Process window is the tolerance

Fill, cure, and cool-down (the part cools after eject; the tool stays hot) all move dimensions:

• Short cure: green part, extra shrink later, tear on demold that looks like a dimensional nick.
• Over-fill: flash, then an OD that depends on how the deflasher felt that shift.
• Cavity temperature high: faster cycle, different shrink, possible scorch at the gate on a micro drop.
• Grade change or color paste: viscosity and shrink shift. Micro tools feel that immediately.

Capability (Cpk) on a micro OD is a process study, not a machining quote. Cut the steel steel-safe on the features that can be recut, then walk in with measured parts from the *production* press, grade, and post-cure. First shots off a cold tool are not the study.

Walls and slits below 0.2 mm are a fill problem first (see thin-wall practice). Holding ±0.05 mm *on* a 0.2 mm wall is a harder problem than holding it on a 1.2 mm boss. State both the nominal and the tolerance.

What “below 0.05 mm” is honest for

Honest:

• A single critical OD, ID, or slit width on a small LSR part, after process lock, measured optically, on one grade.
• Matched pairs of cavities that were walked in together.
• Features with enough metal around them that the insert does not breathe.

Not honest:

• Every dimension on a 40 mm part.
• HCR compression with hand deflash.
• RTV parts from a printed master.
• A drawing that mixes ISO 3302-1 M3 on the title block with ±0.03 mm on twenty radii.

Medical micro parts add ISO 13485 process control and, if patient-contact, ISO 10993 on the *grade*. Neither standard creates 0.05 mm. They demand that you can show how you held what you claimed.

RFQ notes when 0.05 mm is real

Send the critical few dimensions that must sit under 0.05 mm, the measurement method, the grade (or the shrink you will freeze), post-cure yes/no, and whether this is a tryout aluminum tool or a production steel micro tool. If you need ISO 3302-1 plus extras, mark extras on the print. If you ignore measurement method, the first-article argument will be about the probe, not the mold.

FAQ

Is ±0.05 mm a standard LSR tolerance for small parts?

No. ISO 3302-1 classes are the rubber language, and 0.05 mm may be tighter than the class for that nominal. Some micro features can hold it after the steel is compensated and the process is frozen. A blanket ±0.05 mm on every dimension is not a standard; it is a request to be priced and then walked back.

Why did the tool CMM pass but the parts miss 0.05 mm?

The tool was measured cold. The part shrinks from a hot cavity and then maybe a post-cure oven. Probe indent on soft rubber also lies. Compensate from measured parts in the production window, and agree an optical method for micro edges.

Can aluminum micro molds hold this tolerance in production?

They can teach you fill, vestige, and a first shrink number. They will not hold a sub-0.05 mm band for a production life. Thermal cycling, shut-off bruise, and handling chew aluminum. Cut steel for the capability study you will sell.

Does a valve-gated cold runner improve micro tolerances?

It improves vestige and shot-to-shot fill, which is often where a micro tolerance is actually spent. It does not cancel shrink. A cured sprue that outweighs the part will also fight balance. For milligram shots, the deck is usually part of the tolerance stack, not a luxury.

How should we measure a 0.05 mm spec on 20 Shore A silicone?

Prefer non-contact optical or CT with a written edge rule, on parts that have reached dimensional equilibrium after any post-cure. Contact probes indent. If a customer CMM must be used, agree force, tip, and fixturing, or you will argue about 0.03 mm of rubber deflection forever.