Repair Tears in Custom RTV Molds Without Distortion

Stop stretching the tear to see how far it runs. Photograph it, mark both ends with a silver pen, and decide repair versus recast before you open a glue bottle. Most distorted “repairs” fail because someone pulled the split open, filled a gap, and then wondered why the cavity grew.

This procedure is for RTV mold rubber (platinum or tin block, glove, and brush-on skins). It is not how you weld a torn LSR production part, and it is not how you salvage a cracked HCR compression tool. Those are recasts or steel repairs on a parts OEM.

Decide in five minutes

Repair is reasonable when all of these are true:

• The split is in a wall, flange, or low-detail region, not across a sealing land, a sharp logo, or a dimensional datum.
• You can close the faces so they kiss without a gap. If you need to force them, the mold has already taken a set.
• The rubber is the same family you can still bond: platinum-to-platinum with a platinum adhesive or a thin platinum skin; tin-to-tin with a tin-compatible adhesive. Do not glue platinum to tin and expect a structural joint.
• The mold is not oil-soaked, chalky from UV, or crumbling from age. Rotten rubber will not hold a bond.

Recast when the tear crosses the pour face of a precision cavity, when a glove mold has split on a high-stretch undercut for the third time, or when closing the tear changes the outline by a visible amount. A cheap repair that throws every subsequent casting off-size is not cheaper.

What you need on the bench

• The original grade if you still have it, or a platinum (or tin) adhesive sold for that cure family, plus a thin wash of the mold rubber for a skin repair.
• Isopropanol, lint-free wipes, and a sharp blade. No shop silicone aerosol.
• Mixing sticks, a card to knife on the adhesive, and a fixture: the mother mold, a 3D-printed jacket, or plaster that holds the cavity in its as-tooled shape.
• Clamps or elastic bandage. Not vises on the rubber.
• Release on any rigid surface the adhesive must not lock to.

If the mold was made in Shore 10A, do not repair it with a 40A adhesive fillet. The hard bead becomes a hinge-stress raiser and the next tear starts beside it.

Procedure

1. Map the split. With the mold in its shell, unstretched, mark the two ends. If the tear is longer on the inside than the outside, believe the inside. Look for a second start at a nearby undercut. One ignored satellite split will reopen the job.

2. Clean without swelling the joint. Wipe IPA. Let it evaporate. Do not soak the tear in solvent; some grades swell and then the fit you just judged is a lie. Degrease any release or resin film. Bonding to oil is how repairs peel in a week.

3. Prepare the faces, do not excavate them. For a clean split, you want original face against original face. Do not V-groove a cosmetic cavity wall; you will fill that groove with extra rubber and print a ridge. For a ragged edge in a non-cosmetic flange, a shallow scrape to fresh rubber is enough. Cut away only crumbs that would hold the joint open.

4. Dry-fit in the geometry fixture. Seat the mold in the mother mold or jacket. Close the tear with finger pressure only. The cavity must look like the master. If a letter fills or a wall goes wavy, stop. Recast. A clamp that “makes it close” is distortion you are about to freeze.

5. Mix a small batch. Adhesive or a brushable mix of the original rubber, catalyzed as the datasheet says. Degassing a tiny adhesive shot is often skippable; bubbles in a 0.2 mm bond line are not your main risk. Skin-coat repairs on a glove need a mix thin enough to wet, not a paste that tents.

6. Wet both faces thinly. Knife or brush a film. More adhesive does not mean more strength in silicone. Excess becomes a spacer, and spacers change the cavity. On a glove skin, work from the outside when you can so you do not bleed into detail.

7. Close in the fixture and hold. Seat the mold, close the shell, and restrain it in the as-tooled shape until cure. Do not stretch the joint to “test it” at the green stage. Cure at the temperature the adhesive wants. A platinum RTV that needed 24 hours at room temperature will not be a tool after 40 minutes under a lamp.

8. Post-cure only if the grade requires it, and only in the fixture or on a flat support. Heat-softened rubber plus a hanging mold equals a new banana shape.

9. Trim flush, do not sand the cavity. Flash of adhesive on the pour face is sliced with a blade. Abrasive on a silicone cavity dulls the copy and leaves a white, stretched surface that sticks.

10. Qualify with one casting, not with a finger poke. Cast the actual resin, wax, or plaster. Check the tear line for a witness, check a critical dimension against the last good part, and inspect the opposite wall for a bulge you introduced by clamping. If the part is off, you did not repair the mold. You made a new, worse cavity.

Special cases

Glove molds. Invert only after the bond has full cure. Repair on a mandrel or on the master if you still have it, so diameter does not shrink. A ring of adhesive around a wrist opening is fine. A stripe across a face is usually a recast.

Block molds with a mother mold. Always bond inside the closed shell. That is the entire anti-distortion trick.

Tin-cure library molds. A tin repair on an already-shrunk mold locks in today’s size. If the tool has been shrinking for a year, recast in platinum rather than stitching a moving target.

Food-contact molds. A bench repair with an untested adhesive takes the article out of 21 CFR 177.2600 / LFGB qualification. Recast in the qualified compound, or retest the repaired article. Do not patch a chocolate mold with aquarium silicone.

Common mix-up: stretching the tear “to see it”

Pulling the split opens a gap you will then fill, and the cavity grows. The analog of a steel weld on an LSR land is not more glue. It is a recast against the master.

What to ask when you're buying a repair vs recast

• Is the tear on a pour face, logo, or datum — or a flange?
• Platinum or tin, and do we still have that grade?
• Does the jacket still close without force?
• Food-contact: recast in qualified compound, or retest?
• Who owns the master if we recast?

Worked example: 12-cavity bear, letter tear

12-cavity, 18 mm gummy, platinum 10A–20A, jacket. Tear across a chest letter. Adult call: recast. A 40A fillet on that 10A face is a hinge and a proud ridge in every dump. A flange tear off the dump rim, faces kissing in the jacket: thin platinum film, hold until full cure, one witness dump vs 4.0 g.

When this article is the wrong quote

If the torn object is an LSR *part* or a cracked HCR plate, this glue procedure is the wrong SOP. A tiny shallow tray with a printed master reprints faster than a pretty repair. Geometry can override: third tear on a high-stretch undercut is a split/plug change, not more adhesive.

Related reading: salvage vs replace the kit, modification vs new master, food-contact rules, custom RTV molds.

What to send for a repair/recast call

Photos of the seated skin (unstretched), both ends marked, jacket closed. Name platinum vs tin. Say whether the master still exists. We will say bond in the jacket or recast — not “try aquarium silicone.”

We would not recommend a 40A fillet on a 10A mold

We would not recommend stretching the tear to inspect it. We would not patch a chocolate face with untested adhesive. We would not raise Shore A first to “stop tearing” when the peel path is wrong. We would not hang a glove to post-cure a bond.

RUUIPON Mold Structure Decision: if the tear is on a peel that was never going to work as one-piece, recast as two-part / plug / core. Flat-back flanges can bond. Large wet casts still need the jacket as the fixture.

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 | | --- | --- | | Skin chemistry | Platinum 1:1 or 10:1 (match family); vacuum degas on recast | | Cavity Shore | 10A–20A; repair at the same Shore, not 40A | | Platinum shrink on recast | About 0.1–0.3% after full cure | | Tin library already shrinking | Recast platinum; do not lock today's size | | Bond line | Thin film; excess is a spacer | | Qualify | One actual-media casting, not a finger poke |

FAQ

Why shouldn't I stretch a tear open to see how far it runs? Pulling the split opens a gap you will then fill, and the cavity grows. Photograph it, mark both ends unstretched, and decide repair versus recast before glue. This procedure is for RTV mold rubber, not a torn LSR part or a cracked HCR steel tool.

Can I glue a platinum mold with tin adhesive or mix families? Do not expect a structural joint. Match chemistry: platinum-to-platinum or tin-to-tin. Also match hardness; a 40A fillet on a 10A mold is a hinge-stress raiser. Rotten, oil-soaked, or UV-chalky rubber will not hold a bond.

Why bond inside the mother mold or jacket? That is the anti-distortion trick. Dry-fit so faces kiss without force and the cavity still looks like the master. A clamp that "makes it close" freezes a new, worse cavity. Excess adhesive is a spacer; thin film on original faces, then hold until full cure.

May I patch a food-contact chocolate mold with aquarium silicone? No. An untested adhesive takes the article out of 21 CFR 177.2600 / LFGB qualification. Recast in the qualified compound or retest the repaired article. Recast also when the tear crosses a sealing land, a logo, or a datum, or when closing it changes the outline.

When is recast cheaper than a pretty repair? When the qualifying cast does not match the last good part, when a glove has split on a high-stretch undercut for the third time, or when a tin library mold has already been shrinking for a year. Geometry lives in the rubber that copied the master, not in stacked glue.