Keying Systems for Two-Part Silicone Mold Alignment

If you're buying a two-part custom RTV mold, a close that is 0.3 mm off is not a flash problem. It is a keying problem.

Mismatch shows up as a step across the parting line, crushed keys, torn RTV lips, and dumps with a ridge that no amount of polish on one master will fix. Alignment is a designed interface. A “flat butt and some clamps” is a wish on anything larger than a coin.

RTV two-part molds (platinum or tin blocks and jackets) need keys. Steel LSR/HCR leader pins are a different hardware set on a parts RFQ. Don't put rubber nubs on a 10A glove and call them pins.

Alignment is not a lip

A parting lip or a flash land is there to shut. If you ask it to locate as well, you grind it. Location and shut-off are two surfaces.

Location (keys): features that only fit in one XY (and often one rotation). They should touch before the cavity lands crush.

Shut-off: the land around the cavity that meets last, or meets on a designed crush for RTV.

Clamps / straps: the force. They don't translate a walking half into the correct place; they freeze the error.

If you ignore this: the operator “wiggles it until it looks even,” the keys never actually seat, and every dump has a 0.2 mm step that migrates around the split.

Common mix-up: four equal buttons in a square

Four hemispherical buttons in a square can still assemble 90° off if the cavity allows it. Use an odd count, a non-equilateral layout, or a unique extra key. Leftover sprue bumps crush and then locate worse than nothing.

ISO 3302-1 might class the *cast article*. It does not locate two halves, and it is not another word for precision.

Keys that work in RTV

Soft molds need keys the rubber can find with the hands, that survive opening, and that don't undercut so hard the mold tears.

Hemispherical buttons and cups. The classic. Pour or clay male hemispheres on one half; the other half copies them. They center. They don't clock rotation unless you use three and a non-equilateral layout.

Tapered cones / pyramids. More locating, more tear on peel if the taper is too shallow and too deep. Draft them like small cores.

Fences and stepped perimeters. A raised wall around the cavity that drops into a moat. Excellent against XY slide. Machine or clay it generously; a 1 mm fence on a 300 mm mold is a suggestion.

Puzzle tabs (one unique shape). Cheap rotation lock. Put it outside the pour so it does not print on the dump.

External jig. Pins in a wooden or printed jacket, the RTV halves sitting in pockets. The rubber can be almost keyless if the shell is keyed. This is often the right production RTV pattern: sloppy rubber, strict jacket.

Don't key with a few leftover sprue bumps. They crush and then you have less alignment than nothing.

Tin-cure versus platinum does not change key geometry. Hardness does: 10A buttons squash and stop locating. Put serious keys in the jacket, not in the glove. Mix 1:1 or 10:1, vacuum degas, Shore 10A–20A skin, rigid mother mold.

How many keys, and where

Minimum that usually works:

• Small RTV (hand-size): 2–3 buttons plus a unique rotation key, or a full perimeter fence
• Large RTV: fence + 4+ buttons + jacket hardware (bolts). Bolts are clamps, not locators, unless they sit in tight sleeves in the jacket

Place keys outside the flash land, far apart (leverage), not in a cluster on one edge. A cluster is one key that pretends to be four.

Height: keys should register while there is still air between cavity faces. If the cavity kisses before the keys have entered, you will get the nick that starts flash.

Steel leader pins, tapered interlocks, and insert keys are the parts-tool conversation. If your quote is full of bushings and a cold-runner, you are not buying this mold.

What a mismatch looks like (so you stop blaming silicone)

• A step along the split, constant around the dump: halves shifted in XY
• A tapered step: one corner high — dirt on a land, a crushed RTV key, or a warped jacket
• Flash on one side, crush on the other: rotation or a missing unique key, assembled 180° off
• Torn RTV at the same button every cycle: that button is doing shut-off duty or is too sharp; it is not “bad silicone”

Measure the closed empty mold with talc or a thin smear of contrasting clay before you blame shrink (~0.1–0.3% platinum). Open. Transfer should be even on keys first, then on lands. If the cavity paints before the keys, the keys are too short.

The dry-close test is the end of the argument. No compound. Close, open, inspect. Then pour. A keyed tool that only “fits” with rubber in it is already flashing; the rubber is the gasket for your alignment error.

What to ask when you're buying

• Two-part RTV, key type (buttons+fence / jacket pins), unique rotation?
• Keys engage before cavity lands?
• Jacket yes/no — are serious keys in the shell?
• Who keeps the master and the jacket if a key tears?
• Recast of the RTV skin included, same keys?

Worked example: two-part 40 mm hollow with jacket pins

A 40 mm hollow chocolate, two-part platinum RTV, about 800 dumps a month. Three non-equilateral buttons in the rubber plus pins in the printed jacket. Shore 10A–20A skin, mix 1:1, vacuum degas. Dry-close with clay: keys painted first. We would not have located on the flash land. A parallel LSR leader-pin quote does not dump chocolate.

When this article is the wrong quote

If you actually need steel two-part LSR/HCR tools, leader pins and bushings belong on a parts OEM RFQ. A tiny one-piece open-back tray can skip keys entirely. Geometry can override quantity: a two-sided logo at 80 dumps still needs a keyed split.

Related reading: custom RTV molds, self-demolding geometry, what tooling means, food-grade molds.

Keying lines to copy onto the RFQ

Copy this. If the quote is silent on keys, first article will write that section in scrap.

• Two-part platinum or tin RTV
• Key type: buttons+fence / jacket pins; unique rotation
• Keys engage before cavity; dry-close test
• Jacket yes/no; serious keys in the shell if Shore is 10A–20A
• Recast of the skin; who keeps jacket and master

Send split photos, undercut notes, and dump material. We will mark key layout, jacket hardware, and whether a one-piece glove is even honest.

We would not recommend locating on the flash land

We would not recommend four equal buttons in a square as the only rotation lock. We would not put serious keys in 10A rubber when a jacket exists. Raising Shore A first is the wrong fix for a walking half; add jacket pins. We would not buy LSR leader pins “so we can scale later” while you still need a cavity to dump into.

RUUIPON Mold Structure Decision: two-part? keys in the jacket, unique rotation. Flat back, no split? open-back, skip keys. Deep undercuts that will not peel? split or core — extra buttons do not remove a lock. Soft 10A–20A or wet dump? mother mold.

flowchart TD
  A[Two-part RTV?] -->|no| O[Open-back; skip keys]
  A -->|yes| B[Location vs shut-off are two surfaces]
  B --> C{Hardness 10A-20A?}
  C -->|yes| D[Serious keys in the jacket]
  C -->|30A+ block| E[Buttons plus unique key still]
  D --> F[Odd count or unique rotation]
  F --> G[Dry-close: keys paint first]
  A --> H{LSR/HCR steel?}
  H -->|yes| I[Stop: different part RFQ]

Typical shop values

Caption these as typical shop values, not a named lot.

| Item | Typical shop value | | --- | --- | | Cavity Shore | 10A–20A platinum RTV, keys in the jacket | | Mix | Platinum 1:1 or 10:1; vacuum degas | | Linear shrink | ~0.1–0.3% (platinum); do not blame shrink for a 0.3 mm step | | Small mold keys | 2–3 buttons plus a unique rotation key, or a fence | | Loaded-tray edge sag | About 0.5–1.5 mm if the jacket walks | | Dry-close | Talc or contrasting clay; keys transfer before lands |

FAQ

Why shouldn't the flash land also locate the two halves? A land is there to shut. If it also locates, you grind it. Keys should touch and register in XY (and rotation) before cavity faces crush. Clamps freeze whatever error is already there; they do not translate a walking half into place.

Can four hemispherical buttons in a square still assemble wrong? Yes. Four equal buttons can still clock 90 degrees if the cavity allows it. Use an odd count, a non-equilateral layout, or a unique extra key. Leftover sprue bumps crush and then locate worse than nothing.

Should 10A glove keys live in the rubber or the jacket? Hardness locates, not cure family. Soft buttons squash and stop working. Put serious keys in the rigid shell and let the glove sit in keyed pockets. Tin versus platinum does not change that geometry.

What does a dry-close test show that a flashed pour hides? Close empty, with contrasting clay on the RTV. Transfer should hit keys first, then lands. If the cavity paints before the keys, the keys are too short. A tool that only "fits" with rubber in it is already using the dump as a gasket for alignment error.

Why did my quote include leader pins and a cold-runner? That's a steel LSR or HCR *part* tool. Ask the supplier to requote a custom RTV mold if what you need is a keyed two-part cavity to cast food, wax, or resin.