Rib and Boss Design Guidelines to Prevent Sink Marks in Silicone Tooling

Sink in silicone is real, just quieter than in thermoplastic.

LSR marketing likes to say sink is almost nonexistent. That is true next to a 3 mm ABS housing. It is not true on a thick HCR pad, a post-cured LSR lump, or a rib that sits on a 6 mm wall you then photograph under raking light. Flexible RTV molds have a different “sink”: the cavity bulges because the rubber tool moved, which prints as a fat boss on the casting. Same word, opposite physics.

If you ignore the process, you will add thermoplastic-style rib ratios to an RTV glove that did not need them, and you will skip them on an HCR bumper that did.

Where the dent actually comes from

Heat-cured LSR in metal. The cavity is filled; the rubber expands in the hot tool, then shrinks ~2.5-3% as it cools (often another ~0.5-0.7% in post-cure). Thick sections pack and shrink more in a way that can dish a show face opposite a rib or boss. LSR shear-thins and fills odd walls, so the fill is rarely the problem. The mass is. A rib that is as thick as a 5 mm wall is a heat sink and a shrink sink.

HCR compression/transfer. More thermoplastic-like. Charge knit, thick-to-thin, and millable compound shrink (~1.5-3% typical band) show as flats, sinks, and knit shadows at bosses. Uniform walls matter more here than in LSR.

RTV product (the part is poured silicone). A boss in the master becomes a pit in the mold and a boss on the part. “Sink” on the part is often air in the boss or a soft mold wall that ballooned. Vacuum and a mother mold fix those. Rib ratios copied from ABS CAD do not.

If you ignore mass, the defect is a dimple opposite every screw boss on a cosmetic LSR wearable, blamed on “the polish.”

Ribs: start at half the wall

A working LSR/HCR rule that matches what injection shops already quote:

• Rib thickness 0.5× the adjoining wall as the start.
• Up to 1.0× if the rib is structural and the face is not cosmetic, or if the durometer is high and you need stiffness.
• Height: keep ribs from becoming walls. A 10 mm-tall 0.5 mm rib in 50A will fold and may trap gas at the root.

Fillet the rib root. A sharp inside corner in steel is a tear in the part and a porosity site in LSR (inside fillets on the order of the wall, or at least ~0.25-0.4× wall, are the usual ask). On a flexible RTV *mold*, a sharp rib in the master is a sharp groove that rips when you demold the casting.

Draft the ribs. LSR will often strip a shallow 0° rib. A tall rib with texture will not. 0.5-1° on untextured LSR ribs is cheap. More with grain.

If you ignore the 0.5× start, the defect is a groove on the show face that looks like a sink and measures as a local shrink.

Bosses: core them like they had a screw

A solid boss is a lump. Treat it as a cycle-time and shrink problem in LSR, and as a knit problem in HCR.

• Core out from the back or from the open face so the remaining wall is near nominal.
• Keep the floor under a cosmetic face as even as you can. A 2 mm floor next to an 8 mm boss is a sink invitation.
• Gussets at 0.5× wall beat a fatter boss.
• Inside radius at the boss root so LSR does not trap gas in a square pocket.

Hot LSR in a thick boss skins on the steel and stays green in the core if you open on a thermoplastic cycle. The “sink” then is a collapsed or shredded boss, not a cosmetic dish. Slow the cure or remove the mass.

RTV: a tall boss on a master needs a vent at the last fill when you pour the mold, and again when you cast the part. That is air, not shrink.

Gas traps look like sink

LSR and HCR will leave a pale, dull, or dimpled patch at a rib intersection that never vented. Designers call it sink. The toolmaker calls it a dead end.

Put a vent at the last fill, or rotate the rib so it feeds from thick to thin toward a parting vent. Do not add more rubber to “pack it out.” Extra pack on LSR is how you flash, not how you fill a blind rib.

Cosmetic faces vs structural faces

If the camera sees the face opposite the rib, you are in thermoplastic discipline: thinner ribs, cored bosses, maybe a texture that hides a 0.05 mm dish. If the face is a gasket groove, a 1.0× rib is fine.

Transparent and translucent LSR shows everything. A sink you would ignore in 50A red will look like a lens defect in 30A clear.

Flexible RTV molds of a cosmetic master: polish the master, support the mold. Do not “fix sink” by thickening the CAD boss. You will copy a bigger bump.

What to change on the CAD before steel

1. Nominal wall called out.
2. Ribs at 0.5× unless a note explains why not.
3. Bosses cored to near-nominal wall.
4. Fillets at roots.
5. Draft on ribs.
6. Process: LSR / HCR / RTV. The same STEP should not wear all three rule books.

Prototype in the process you will ship if the show face is the risk. An RTV pour of a 1.5 mm wall will not predict LSR sink on a 6 mm pad.

FAQ

Does LSR really not sink?

It sinks less than a thick ABS wall, not never. Thick HCR pads, post-cured LSR lumps, and a rib as thick as a 5 mm wall can dish the show face. LSR shear-thins so fill is rarely the issue; mass and shrink are.

Why is a pale dimple at a rib intersection often not sink?

That patch is often a gas trap at a dead end. Vent the last fill or rotate the rib so it feeds thick-to-thin toward a parting vent. Extra pack on LSR makes flash; it does not fill a blind rib.

Should I copy thermoplastic rib ratios into an RTV glove?

No. "Sink" on a poured RTV part is often air in the boss or a soft mold wall that ballooned. Vacuum and a mother mold fix those. Thickening a CAD boss on a flexible tool copies a bigger bump.

Why core a boss that will never take a screw?

A solid boss is a shrink lump in LSR and a knit problem in HCR. Core it so remaining wall is near nominal, keep the floor even under a cosmetic face, and fillet the root so LSR does not trap gas. Hot thick bosses also stay green in the core if you open on a thermoplastic cycle.

Will an RTV pour of a 1.5 mm wall predict LSR sink on a 6 mm pad?

No. Prototype in the process you will ship if the show face is the risk. Transparent LSR also shows dishes you would ignore in 50A red.