Custom RTV Ice Molds: Freeze, Flex, Jacket

If you're ordering a custom ice mold, you're buying a freeze-and-flex RTV cavity, not a small muffin pan. Manufacturing it well means a cube or sphere that leaves without a fork, a split that doesn't tear at -18 C, a mother mold that keeps the pocket from ballooning, and a food-contact report on the silicone face. Frozen water is still food contact: 21 CFR 177.2600 aqueous flasks and LFGB / BfR XV sensory still apply. Platinum RTV pours are the usual custom-mold process. Tin-cure isn't dual-market ice chemistry. Bakeware heat lines don't transfer. High-volume consumer trays as *parts* are typically LSR injection with a cold-runner — a different quotation.

Geometry, pour sequence, and why forks appear

Duty: fill, freeze, peel or twist, harvest, wash, repeat. Failures that aren't "cheap rubber": web tear, reverse-taper lock, lost key register, short post-cure odor, a fork because geometry won't release. Sequence: finished master (SLA fully UV post-cured or the platinum inhibits), rigid box, A/B platinum mix, vacuum degas, pour, keys on two- or three-piece tools, jacket, post-cure scaled to wall, wash.

Draft still helps in a flexible cavity. High polish can suction; ice often wants a controlled matte. Chocolate polish copied onto a sphere without a freeze trial is how you get a pry tool. Fill-line marks that are too sharp become tears. Lids in PP are separate plastic articles; the rubber flask isn't the lid.

Common mix-up: ice skips food-contact because it's cold

A lot of RFQs treat frozen water as a waiver. In the shop that usually means you've skipped 177.2600 water flasks and German sensory. Cold doesn't erase short post-cure odor. The first cubes are a sensory test whether you booked the panel or not.

What to ask when you're buying an ice tool

• Two-part keyed split or one-piece peel pocket, and is there a jacket?
• Draft and polish — have you freeze-trialed this geometry?
• Platinum RTV, post-cure scaled to wall, 177.2600 aqueous and/or BfR XV on this color?
• Does the lid have its own plastic food-contact file?
• Who owns the master, and is a recast included after a web tear?

Worked example: 50 mm two-part sphere

A beverage program wants a 50 mm clear-ice sphere, two-part platinum RTV, keyed splits, printed or fiberglass jacket, about 500 pulls a month, aqueous food-contact. Finished master, registered box, degassed platinum RTV, post-cure, wash, 177.2600 water flask and BfR XV sensory, fill-line that isn't a tear initiator. That's sampling-to-mid volume. It isn't a 50,000-piece LSR program.

When this article is the wrong quote

Consumer ice trays in the tens of thousands as molded *goods* belong on LSR injection. Alcohol pops and milk-fat popsicles are different food types than plain ice — if you only qualified water, don't claim cocktails. A muffin RTV cavity isn't an ice tool even in the same compound family. Geometry can override: a simple open cube tray can be a one-piece glove; a sphere almost always needs a split.

Related reading: custom RTV molds, food-grade molds, why platinum RTV, food-plant applications.

Ice-mold RFQ to copy

• Sphere/cube size, split vs glove, monthly pulls
• Master type; jacket; fill-line yes/no
• Platinum RTV, Shore, color, post-cure
• Food: water, flavored, alcoholic, milk-fat
• 177.2600 aqueous and/or LFGB/BfR XV; lid material separately
• Recast after web tear; no fork on the care card

Send the sphere diameter, monthly pulls, and whether a lid ships with it. We'll review the split, jacket, freeze trial, and which flask covers the rubber vs the lid.

We would not recommend a fork as a demold tool

We would not recommend copying chocolate gloss onto an ice sphere without a freeze trial, or pasting a bakeware oven number on the carton. We would not file the rubber flask as coverage for a PP lid. We would not run a one-piece glove on a locked-waist sphere.

Split vs glove for ice:

flowchart TD
  G[Geometry] -->|open cube peel| O[One-piece glove possible]
  G -->|sphere or reverse taper| S[Two-part keyed RTV plus jacket]
  S --> T[Freeze trial at -18 C]
  T -->|users still pry| D[Fix draft and flex; do not ship a fork]

Typical shop values: 50 mm two-part ice sphere

| Item | Typical shop value | | --- | --- | | Diameter | 50 mm two-part platinum RTV | | Monthly pulls | ~500 is a common custom/mid band | | Cavity Shore | 10A–30A with jacket | | Mix | Platinum 1:1 or 10:1; vacuum degas | | Shrink | ~0.1–0.3% platinum | | Heat claim | Not for oven; freezer duty |

Caption these as typical shop values, not a 50,000-piece LSR program.

FAQ

Why can't we run ice tools in the same RTV family as muffin cavities? You can share a platinum RTV compound family. You can't share geometry, polish, web, jacket, or a heat claim. Each SKU is its own article.

Does frozen water skip food-contact testing because it is cold? No. Aqueous contact still exists. 177.2600 water flasks are reflux methods, not a freezer simulation.

Platinum RTV or LSR for a 2-inch clear ice sphere? Usually platinum RTV in a two-part tool with a jacket for custom and mid volume. High-volume consumer spheres may move to LSR; that's a new RFQ.

Why do users puncture trays with forks? Geometry locked the cube. Fix draft, flex, and jacket. A fork cuts the food-contact face.

Can the silicone report cover a PP lid? No. The lid is a plastic food-contact article. The rubber flask is the RTV cavity.