Engineering Answers / Design & DFM
Flash is not one defect with one fix. It is four different failures that look the same on the inspection bench — and the pattern of where it appears is usually enough to tell you which one you have.
Silicone flashes because the material reaches a gap in the tool while it is still liquid. Liquid silicone rubber is injected at 170–200 °C with very low viscosity, and it will flow into a shut-off gap smaller than 0.005 mm. So flash is produced by whichever of four things opens that gap: the tool's shut-off faces, where the parting line was placed, how hard the cavity was packed, or the part geometry that forces material into a thin section.
The same defect on a polypropylene part is often a cosmetic annoyance. On silicone it is a functional risk, because the material is soft enough that any raised edge is felt immediately, and because the parts that flash matters most on — seals, masks, mouthpieces, gaskets — are handled and worn by people.
The numbers explain the difference. In rigid plastic injection, flash typically appears once the gap between mating steel faces exceeds roughly 0.01–0.03 mm. High-consistency rubber (HCR) is a high-viscosity gel and holds closer to 0.02 mm. LSR is the outlier: it is shear-thinning, so under injection pressure its viscosity drops sharply, and it will find its way into a gap under 0.005 mm — about one twentieth of the width of a human hair.
This is why "just tighten the mold" is not a real answer. The tool has to be built to a tolerance that is at the edge of what conventional machining holds, and it has to stay there after a few hundred thousand cycles.
Every mold has faces that meet to stop material: the parting line, insert boundaries, slides, and the lands around each cavity. If those faces do not close to within a few microns, LSR fills the difference. On our own tools, cavities are machined by CNC high-speed milling and finished by mirror EDM, with clamping tolerance held within ±0.005 mm and verified on a 3D CMM at ±0.002 mm resolution before the tool runs.
The variable here is time, not just the initial build. Tool steel — NAK80, S136 stainless, H13 hardened to HRC 48–54 — is rated for 100,000 to 500,000+ cycles, but the number that matters is how the shut-off faces wear across that life. A tool that was clean at first article and flashes at 80,000 shots has a maintenance problem, not a design problem.
Most flash complaints are really about location. A parting line placed across a sealing bead, an optical surface or the edge that touches skin turns a normal molding artefact into a rejected part. Moving it 2 mm onto a non-functional face during DFM review costs nothing; moving it after the steel is cut means re-machining or a new insert.
This is a drawing decision as much as a tooling decision. If the drawing does not say which faces are functional, the tool designer will place the parting line where it is easiest to machine — which is rarely where it belongs.
Flash can be created by the press even when the tool is perfect. Injection pressure that overpacks the cavity, insufficient vacuum at the parting line, and a cure temperature that lets the material stay fluid longer than necessary all push silicone into gaps it would otherwise not reach. On a multi-cavity tool this often shows up as flash on the cavities nearest the feed, where pressure is highest.
Uneven wall thickness is the geometry cause. When a thick section feeds a thin one, the thin section fills last and is packed at the highest pressure in the cycle, which is exactly where flash appears. The design rules are unglamorous but effective: transition between wall thicknesses with a gradual taper so the thicker wall is no more than about 1.5× the thinner one, keep ribs at 0.5–0.6× the nominal wall, and radius internal corners to at least R0.5 mm so the material does not tear as it is demolded.
This is the part that saves the most time. Before opening the tool, look at where the flash is and how it repeats.
| What you see | Most likely cause | First check |
|---|---|---|
| Uniform thin film around the whole parting line | Shut-off gap opened up across the tool | Measure the shut-off faces on a CMM; compare against the build report |
| Flash on one cavity only, others clean | Cavity alignment or uneven clamp force | Per-cavity measurement, not a mixed sample; check clamp balance |
| Heavy flash near the gate, tapering away | Overpacking — injection pressure or shot size | Pressure curve at transfer; verify shot weight stability |
| Flash starting after N thousand shots | Wear on inserts or slides | Tool maintenance log; inspect wear on the highest-pressure faces |
| Flash only on a thin wall fed by a thick one | Geometry — wall transition or rib thickness | Wall thickness map against the 1.5× taper rule |
| Flash with a witness mark on a sealing edge | Parting line placement | Redraw the parting line off the functional face |
It is the sequence our tooling engineers run when a customer sends a photo of a flashed part. In most cases rows one and six account for the problem, and row six is the one that should have been caught at the drawing stage — which is why the DFM review happens before steel is cut, not after.
A drawing is only half of it. On the production side, flash control is a maintenance discipline: a defined interval for inspecting and re-machining shut-off faces, first-article dimensional reports on a CMM, and process capability held at Cpk ≥ 1.33 on the critical-to-quality dimensions rather than a visual pass on the last part of the shift.
Ask for those three things before placing a volume order, and ask what the tool maintenance interval is in shots, not in months. A moulder who cannot answer the second question is not measuring wear — here is what a shot-based maintenance schedule looks like.
Cryogenic deflashing is a legitimate process and we run it. Parts are frozen with liquid nitrogen so the thin flash film becomes brittle, then tumbled against non-abrasive polycarbonate media that shears the film away without cutting into the part. On cosmetic edges and non-functional parting lines it is fast, consistent and cheaper than re-cutting a tool.
It is the wrong remedy in three situations: on a sealing bead, where the geometry that seals is precisely what the process is impacting; on very thin walls, which can chip rather than clean; and on any part where the flash is caused by a shut-off gap that is still open, because you are paying to remove a defect the tool keeps producing. If flash returns on every lot, the answer is the tool, not the tumbler.
Send a photo of the flash with the location marked, the drawing or 3D file, and the shot count on the tool. That is usually enough for us to tell you which of the four causes you are looking at before anyone touches steel. Our engineering team returns DFM feedback within 24 hours.
Not on every surface. Flash is normally specified by zone rather than by one global number: sealing beads, optical faces and any skin-contact edge are held to a functional limit, while non-functional edges are usually accepted against an ISO 3302-1 class such as E1 or E2. Without a zoned callout, the moulder and the inspector are working to different definitions.
It depends on where the flash sits. Cryogenic deflashing freezes the part with liquid nitrogen so the thin flash film becomes brittle, then tumbles it against non-abrasive polycarbonate media which shears the film off without cutting the part. That works on cosmetic and non-functional edges. On a sealing bead or a thin wall it is the wrong remedy, because the process removes material by impact rather than by controlled cutting, and the geometry that seals is exactly what you cannot afford to alter.
A single-cavity result does not describe the tool. Uneven clamp force across a multi-cavity layout, cavity-to-cavity alignment drift after repeated cycling, and unequal wear on individual inserts all produce flash on one position while its neighbours stay clean. The check is per-cavity measurement of the shut-off faces with a CMM, not a visual pass on a mixed sample.