Engineering Answers / Design & DFM
The material will happily go thinner than the tool can fill and thinner than the part can survive being stripped. Two separate limits, and design guides usually quote only the first.
A practical minimum is around 0.4 mm for LSR and 0.8 mm for HCR, with 0.5 mm and 1.0 mm as the numbers we would rather design to. Below that, the limit stops being the material and becomes the tool: LSR will not reliably fill a long thin section because air cannot escape ahead of the flow, and a thin wall tears at demolding where the strain concentrates at a sharp base.
Ask how thin silicone can go and most answers quote a single number. There are actually two limits, and they fail in different ways.
The fill limit is set by the tool and the process. LSR is injected at low viscosity, but it is still displacing air, and in a thin section the air has nowhere to go if the venting is poor. The part does not come out thin — it comes out short, with a rounded, unfilled edge at the far end.
The tear limit is set by geometry and demolding. The part fills fine, then tears at the base of the thin wall when it is stripped, or tears in service because the section has no support.
You can hit either one first depending on the design. A short thin wall on a well-vented tool hits neither. A long thin wall in a poorly vented tool hits the fill limit. A short thin wall with a sharp base hits the tear limit.
| Process | Absolute minimum | Working minimum | Typical failure below it |
|---|---|---|---|
| LSR injection | 0.3 mm, small area | 0.5 mm | Short shot at the far edge; air traps |
| HCR compression or transfer | 0.6 mm | 1.0 mm | Incomplete fill; trapped air blisters |
| Compression molding, high hardness | 1.0 mm | 1.5 mm | Tear at the base; dimensional drift |
The flow ratio is the number that predicts a short shot before the tool is cut. In LSR, a flow length around 100 times the wall thickness is a reasonable ceiling for a simple, well-vented cavity. At 0.5 mm that means 50 mm of flow from the gate; a 120 mm long thin strip needs either more gates or a thicker section.
When a part is demolded, the thin wall does not tear along its length. It tears where it meets something thicker, because that is where the strain concentrates.
Two geometry moves fix almost all of it:
There is also a support question that drawings rarely answer: what holds the thin wall in service? A 0.5 mm silicone membrane that is unsupported and loaded in one direction will deflect until it touches something, and then it wears. If the design intent is a diaphragm, fine — but a diaphragm needs a defined travel and a stop, not just a thin section.
Designing to 0.5 mm changes more than the drawing.
Whether the section is supported, and whether the base radius is at least the wall thickness. Those two decide whether the part survives production far more often than the wall number does. A 0.6 mm wall with R0.6 at the base and a proper vent outruns a 0.8 mm wall with a sharp corner every time.
Some parts are specified thin for the wrong reason — to save material, or because a competitor's part felt thin. Silicone is not a material where thinness saves meaningful cost; the compound is a small share of the unit price, and the tooling, venting and inspection work that a thin wall demands usually costs more than the material it saves.
If the requirement is flexibility, a lower Shore A at a normal wall thickness is almost always the better route. If the requirement is weight, the same applies. Thin walls are for geometry that genuinely needs to be thin: a membrane, a living hinge, a sealing lip, a light pipe.
A section view with the thin dimension marked, the flow length from the gate, the hardness, and a sentence on what the section has to do. That is enough to say whether it fills, whether it strips, and what the tool needs at the base. If the answer is "it will not fill", it is much cheaper to hear that now than after the cavity is cut.
It can, in a small area, with a tool built for it. We have run sections around 0.3 mm where the area was small, the flow length short and the part was supported underneath. What changes at that thickness is not the material's ability to fill, it is the process window: venting has to be near-perfect, injection pressure climbs, and the demolding step has to be gentle enough not to stretch the section. It is a special case, not a design rule.
Yes, by roughly a factor of two. LSR is injected at low viscosity and can fill a 0.4–0.5 mm section, while HCR is a high-viscosity gum that has to be pressed or transferred into the cavity and realistically needs 0.8–1.0 mm. The difference is flow, not strength — both materials tear at similar strains once molded.
Almost always the base, not the wall. When a part is stripped from the core, the strain concentrates at the transition between the thin section and whatever it attaches to. A sharp internal corner there acts as a stress riser and the tear starts at the corner and runs along the wall. A radius of at least the wall thickness at that transition removes most of the problem, and a slight taper in the thin section removes the rest.