How a Food-Grade RTV Silicone Mold Is Made, Stepwise

If you're buying a food-grade custom RTV mold, the manufacturing process isn't industrial pouring with a prettier carton. It's a chain: a master that holds dimensional and surface truth, a mold box that sets pour height, platinum addition-cure RTV A/B formulated for food contact, splits and registration keys where the geometry needs them, a rigid support jacket so the soft cavity holds shape, a post-cure scaled to thickness, a wash, and then testing of that finished article under FDA 21 CFR 177.2600, LFGB/BfR XV, or both.

Skip a link and the flask describes a different cavity than the one you pack. BPA-free doesn't belong in the work instructions, because silicone isn't polycarbonate. “FDA approved” doesn't either; 177.2600 is extraction of the finished rubber article, not a product registration. Heated steel, cold-runners, and LSR meter-mix belong on a molded-*part* traveler, not this one.

Common mix-up: cure in the box is the whole process

A lot of RFQs treat “pour RTV, demold, ship” as manufacturing. In the shop that usually means skipped post-cure, no jacket, a dirty SLA master that inhibits platinum, and a flask that fails six weeks later.

Surface and size start on the master. CNC, machined metal, a finished sample, or SLA/DLP that's been fully UV post-cured, washed, and barrier-coated are the usual options. Layer lines and undercured SLA inhibition show up as sticky faces. The box sets wall thickness and pour head; a warped plywood flask prints as a tapered block. The kit has to be a food-contact platinum RTV formulation, pigment frozen, no unlisted oils “to help flow.” Tin-cure is the wrong start. Composition constraints in 21 CFR 177.2600 still bind before anyone extracts.

What to ask when you're buying the manufacturing sequence

• What's the master (CNC, printed, sample), how is it finished so it doesn't inhibit platinum, and who keeps it after the job?
• One-piece tray or keyed split? Hollow work: does the core have a pull path?
• Platinum RTV A/B ratio, Shore of the skin, and is there a rigid jacket (plaster, resin, fiberglass, printed shell)?
• Post-cure time and temperature for the thickest wall, recorded on a traveler with the gum lot?
• Wash step after post-cure, given that 177.2600 expects a cleansed article?
• Is a recast of the RTV skin included, and which regulation is on first article (177.2600, LFGB/BfR XV, both)?

Pour sequence: splits, keys, cores, then jacket

One-piece glove molds exist for simple trays. Anything with a bail, a letter with negative draft, or a two-sided logo needs a split. Keys (pips, bushes, or CNC bosses) are how the halves return to the same shut after a hot dump. Inner cores for hollow chocolates or ice spheres need a pull path; if the core can't extract, you designed a cut mold, not a production mold.

Mix platinum A/B to the stated ratio, degas unless the compounder says otherwise, and pour against the registered master in the box. Orientation and venting keep bubbles off the food-contact face. Demold only after the network can take the pull without tearing logos.

Soft RTV without a jacket slumps. For undercut-heavy food or gummy tools, the usual stack is a thin, high-tear skin against the master and a stiff jacket behind it. The rubber does the stretch; the jacket does the cavity. The jacket is manufacturing hardware, not a packing extra.

Cure against the master makes the network. Post-cure in a ventilated oven drives off cyclics and residues that 177.2600 flasks, BfR XV volatiles, and sensory panels find. A 2 mm gummy skin and a 15 mm muffin wall don't share a bake. Skipping post-cure to save oven hours is how a compound that could pass fails. Adding a lab-only bake that production will never run on recasts is how a report lies.

Platinum fails loudly when tin, sulfur, or some amines sit in the same air or on the same gloves. A food-contact pour table that shares a bench with tin-cure craft molds will eventually ship a sticky cavity. Isolate tools. Inhibition means the network never finished — post-cure won't convert a tacky face into a 177.2600 specimen.

Worked example: 16-cavity 25 mm dome chocolate

A confectioner ordered a 16-cavity dome, 25 mm diameter, two-sided logo, fatty chocolate, 600–1,200 dumps a month.

The traveler is a CNC master with radiused logo lines, a keyed two-piece platinum RTV pour in a square box, registration so the logo halves meet, a fiberglass jacket, post-cure scaled to the thickest web, wash, then 177.2600 hexane (and water if also claimed) plus LFGB/BfR XV if Germany is on the PO. A recast of the RTV skin against the same master is the replacement path when the face tears. An LSR cold-runner quote for a molded silicone *dome part* won't dump cocoa butter.

21 CFR 177.2600 expects finished rubber articles to be thoroughly cleansed before food contact. Lot logic ties compound lot, pigment, pour, oven load, master revision, jacket, and SKU. Change pigment or post-cure, or sand the master “to clean it up,” and the next recast is a new article.

Write on the PO who owns the master, who owns the jacket and keys, and whether a recast is included. Paying a tooling charge doesn't, by itself, give unrestricted transfer rights. Customer-owned masters with shop-owned jackets, or the reverse, are common. Amortized RTV tooling exists but is less common than in steel. If it isn't in the agreement, don't assume it.

When this article is the wrong quote

If you need thousands of molded silicone *parts*, this pour sequence is the wrong asset — quote HCR compression or LSR injection with a cold-runner. If the job is tin-cure resin or soap, don't drag food-contact post-cure and 177.2600 into the traveler. A tiny, shallow one-piece tray can skip the mother mold; a multi-core hollow chocolate with a rigid jacket can cost more, in calendar, than a small compression tool. Geometry can override quantity.

Related reading: DFM for food-grade RTV, platinum-cure chemistry, custom food-grade molds, custom RTV molds.

What to send to start a food-grade RTV traveler

Copy this. “Make me a silicone mold” without these lines isn't comparable.

• Master: CNC, SLA/DLP (post-cure + barrier coat), or physical sample — who keeps it
• A/B ratio, Shore of the skin, vacuum degas yes/no, detail coat if logos are fine
• One-piece vs keyed split; core pull path if hollow; jacket material
• Food, dump/oven temperature, monthly quantity
• Post-cure for the thickest wall; wash; 177.2600 and/or LFGB/BfR XV on first article
• Recast of the RTV skin: included or extra; pigment freeze

Send master photos or the STL, finished dimensions, and the food you'll cast. We'll review the split, jacket, inhibition risk on a printed master, and whether post-cure belongs on the first-article gate.

We would not recommend shipping a cavity that only cured in the box

We would not recommend skipping post-cure because the face feels dry, or sharing a pour bench with tin-cure craft molds. We would not raise Shore A to hide a missing jacket. We would not sand the master "to clean it up" between the lab pour and production recasts.

RUUIPON Mold Structure Decision: flat back → open-back; deep undercuts that cannot peel → two-part / plug / core; large wet casting or soft Shore → mother mold.

flowchart TD
  M[Master finished, box square] --> X[Mix platinum A/B 1:1 or 10:1]
  X --> V[Vacuum degas until foam collapses]
  V --> D[Detail coat on fine logos if needed]
  D --> P[Pour, RT cure, demold]
  P --> J{Jacket?}
  J -->|soft Shore or wet dump| Y[Mother mold / support jacket]
  J -->|tiny shallow tray| S[May skip]
  Y --> C[Post-cure by max wall, wash]
  S --> C
  C --> T[177.2600 and/or BfR XV on this recast]

Typical shop values: food-grade RTV pour

| Item | Typical shop value | | --- | --- | | Mix | Platinum 1:1 or 10:1 by grade; vacuum degas | | Cavity Shore | 10A–20A skin | | Linear shrink | ~0.1–0.3% platinum | | Jacket | Required except tiny shallow trays; edge sag ~0.5–1.5 mm if omitted on a large tray | | Post-cure | Ventilated oven; 2 mm skin ≠ 15 mm muffin rim | | Recast | Same master, same keys; not a new steel tool |

Caption these as typical shop values.

FAQ

Can I pour food-contact platinum RTV on the same bench as tin-cure craft molds? Only with a contamination control plan that actually works: dedicated boxes and gloves, no shared oils, no tin residue on the master. Platinum is easy to inhibit. Many shops separate the pour tables.

Is post-cure optional if the cavity already feels dry out of the box? Feel isn't volatile organic matter and isn't a 177.2600 flask. Post-cure is a production step scaled to thickness. Skip it and you're manufacturing a different, dirtier article than the one in the report.

Do I need food-contact steel for the master? The poured RTV is the food-contact article. The master still must not rust, flake, or require illegal release onto the face. Name any sealer or barrier coat. Don't treat the master as a 177.2600 specimen unless it also touches food in use.

Where does BPA-free show up in the work instructions? It shouldn't. Silicone isn't polycarbonate. Work instructions should name platinum RTV A/B, master revision, box, jacket, post-cure, wash, and the regulation on the article.

If first article passes LFGB, can production drop the US flask on later recasts? LFGB/BfR XV and 21 CFR 177.2600 are parallel, so a German pass doesn't replace US extraction. Keep the market list on the traveler. Retest when the freeze breaks, including a master recut.