RTV Encapsulation Molds: Thermal vs Dielectric Specs

Thermal conductivity and dielectric strength pull a potting mold in opposite directions. You rarely maximize both in one compound. Fillers that move heat (alumina, boron nitride, zinc oxide, some graphites) raise viscosity, lower elongation, and usually cut dielectric performance compared with an unfilled platinum silicone. Design the electrical spec and the heat spec as a pair, then design the RTV jacket around the paste you actually have to pour.

This is potting and encapsulating of electronics: connectors, sensors, LED boards, power modules. The rubber that *shapes* the potting is typically platinum RTV. The encapsulant may be a pourable platinum gel or a filled paste. LSR overmold at volume is a steel tool on a parts OEM. HCR is uncommon inside a housing. Tin-cure is the wrong default around copper and around any part that must stay electrically stable in storage.

The two numbers, without fake datasheets

Dielectric. You care about dielectric strength (often discussed in kV/mm on a standard specimen), dielectric constant, and volume resistivity. Unfilled or lightly filled platinum silicones are used as insulators for a reason: they are stable, hydrophobic, and electrically quiet over a wide temperature range. A thin flash of filled paste in the wrong place is still an insulator *until* it is a void, a carbon-filled grade, or a moisture path.

Thermal conductivity. Unfilled silicone is a poor heat spreader. To move heat you add ceramic or other fillers. Conductivity rises; the mix becomes a paste; it may slump less or more depending on thixotrope; it can be abrasive to a mold face. Graphite and some carbon systems can raise conductivity and ruin dielectric behavior. Do not assume “thermally conductive silicone” is still an insulator. Read the filler family.

Ask for both properties on the same datasheet line, at the same cure state, with the test methods named. A thermal number from a pressed slab and a dielectric number from an unfilled cousin grade are not a pair. Do not invent an SGS report number to close the gap.

What the mold has to do that the compound cannot

Encapsulation failures that look like material problems are often mold and process:

• Voids at the board. Air trapped under a BGA, a connector hood, or a tall capacitor. The RTV jacket needs a fill path that pushes from one side and vents the last corner, plus time or vacuum. Vacuum potting is a process choice. A pretty cavity will not degas a blind pocket.
• Wicking into connectors. Low-viscosity unfilled RTV encapsulant will run into a pin field you wanted dry. Use a dam, a gasket, a thixotropic grade, or a two-shot mold that keeps the connector out of the pour.
• Strain on solder joints. A hard, high-filler potting that shrinks or expands against a large PCB can crack joints on thermal cycle. Soft unfilled gels protect mechanically and insulate; they do not move heat. Sometimes the right answer is a thin thermally conductive pad plus a soft gel, not one brick of filled rubber.
• Cure inhibition. Platinum against sulfurated rubber feet, some flux residues, amines, and certain tapes. The mold and the board prep are part of the encapsulant spec. An RTV jacket that was just sprayed with a sulfur-bearing release is a new failure mode.

RTV pour jackets versus LSR overmold tools

Low-volume RTV. The “mold” may be the housing itself, or a platinum RTV jacket that defines the potting line. If you build a platinum RTV mold to shape an encapsulant, that mold rubber must not inhibit the encapsulant (platinum against platinum is the usual safe pair; test anyway). Draft the housing so the cured brick can come out if it is not meant to stay.

High-volume LSR overmold. This is a steel tool, often a two-shot or insert-mold around the PCB. Cold-runner LSR, hot cavity — different quote. The dielectric/thermal grade must be an injectable LSR, not a pourable 1:1 paste. Filled LSRs need larger gates and more injection pressure. Do not quote a pourable filled paste into a micro-gate LSR tool.

Thermal paths the mold can help

Metal heat-spreaders, exposed pads, and thin silicone sections over a hot component beat dumping more filler into a grade that then will not fill. The RTV jacket can:

• Hold a heat sink in location with pins or nests so the silicone thickness over the die is controlled.
• Keep silicone off a thermal pad that must seat metal-to-metal.
• Define a thin bond line of conductive silicone where you do want rubber in the path, and a thick gel elsewhere.

A uniform 8 mm brick of “thermally conductive” rubber around a board is often a worse cooler than a 1 mm defined path plus air or gel.

Electrical creepage the mold can hurt

Flash and overflow into a creepage gap will shorten the insulating distance even if the rubber is a good dielectric. RTV splits that flash into a pin-to-pin slot need keys and a designed overflow, not more clamp on a floppy jacket. Carbon-filled or dark grey “thermal” grades in that gap are worse. If the drawing has a creepage, the mold must not put rubber there, or the rubber in that gap must be the insulating grade, not the conductive one.

Common mix-up: injecting a 3 W paste through a micro-gate

Viscosity and abrasive filler will tell you in the first shot. That is a LIM RFQ, and usually the wrong one.

What to ask when you're buying

• Is the RTV the jacket that shapes potting, or is this LIM overmold?
• Thermal and dielectric numbers on the same datasheet line, named methods, same cure state?
• Must the silicone remain an insulator?
• Vacuum potting yes/no; dam or thixotrope for pin fields?
• Platinum mold vs platinum encapsulant — inhibition test?

Worked example: sensor board, 50 units vs 50,000

50 sensor boards, potting line defined by a platinum RTV jacket, unfilled gel plus a 1 mm conductive path over the die, vacuum chamber. Sister PO: 50,000 LSR overmolds, cold-runner — different quote. Same “encapsulation mold” email. A uniform 8 mm “thermal” brick was the worse cooler.

When this article is the wrong quote

If you need LIM overmold, send the parts OEM. A tiny dam in the housing can skip a poured jacket. Geometry can override: a deep connector hood still needs a vent path, not more filler.

Related reading: skin-safe vs industrial encapsulant, flash at the RTV split, what RTV tooling means, custom RTV molds.

Put both requirements on the RFQ as methods

Write the thermal conductivity target with the method (for example a guarded-hot-plate or transient method the supplier actually uses) and the specimen thickness. Write the dielectric strength or resistivity with its method and thickness. Add voltage, altitude or humidity if they matter, the hottest component temperature, and whether the silicone must remain an insulator. Name platinum RTV jacket versus LSR overmold. Ban tin-cure unless there is a specific, tested reason.

If the supplier offers a single “electronic grade” without those two numbers, you do not yet have a compound. You have a color.

Send the board envelope, creepage keep-outs, and whether you need a jacket or a LIM cell. We will quote the RTV dam and tell you when the paste belongs on a different RFQ.

We would not recommend one compound to maximize heat and insulation

We would not recommend tin-cure around copper. We would not raise Shore A first to “fill better.” We would not paste a pourable 3 W class into a micro-gate. We would not spray sulfur-bearing release on a platinum jacket.

RUUIPON Mold Structure Decision: the jacket defines the potting line. Flat dam? open-back. Deep hood that cannot peel? two-part / plug. Soft gel plus a wet pour? mother mold if the jacket is the tool. If the silicone *leaves* a hot LIM tool, stop.

Typical shop values

Reference values only. Confirm material-specific limits against the supplier’s technical data and agree project acceptance criteria before production.

| Item | Typical shop value | | --- | --- | | Jacket chemistry | Platinum RTV 1:1 or 10:1; vacuum degas; not tin | | Jacket Shore | Often 10A–20A with a rigid support | | Platinum shrink | About 0.1–0.3% after full cure | | Contrast: LSR *part* shrink | About 2–3% — do not put this on the jacket master | | Uniform thick "thermal" brick | Often worse than a ~1 mm defined path plus gel |

FAQ

Can one silicone maximize thermal conductivity and dielectric strength? Rarely. Ceramic fillers that move heat raise viscosity, cut elongation, and usually hurt dielectric performance versus unfilled platinum. Graphite or carbon systems can raise conductivity and ruin insulation. Demand both numbers on the same datasheet line, same cure state, named methods.

Why does a pretty cavity still leave voids under a BGA? Air trapped under a hood or tall component is a fill-path and vacuum problem. The mold must push from one side and vent the last corner. A housing will not degas a blind pocket by looking clean. Low-viscosity unfilled RTV also wicks into pin fields you wanted dry; dam, gasket, or thixotrope.

Can I inject a pourable filled paste through a micro-gate LSR tool? Do not quote that. High-volume overmold needs an injectable LSR, larger gates, and more pressure — a different RFQ. Weld lines in a thermally conductive grade can be a mechanical weak line and a heat bottleneck. Tin-cure is the wrong default around copper and electrically stable storage.

How can the mold help cooling without dumping more filler into the grade? Hold a heat sink so silicone thickness over the die is controlled, keep rubber off a pad that must seat metal-to-metal, and define a thin conductive bond line plus a soft gel elsewhere. A uniform thick brick of "thermally conductive" rubber is often a worse cooler than a millimetre path.

Why is flash into a creepage gap an electrical failure even if the rubber is insulating? Overflow shortens the insulating distance. Carbon-filled thermal grades in that gap are worse. If the drawing has a creepage, the mold must not put rubber there, or the rubber in that gap must be the insulating grade. A sulfur-bearing release on the jacket is also a new platinum-inhibition mode.