Challenges: Producing a food-contact–safe silicone mold with high dimensional fidelity for repeatable edible casts; excluding non-food-safe materials, including release agents; generating a watertight master model and achieving void-free infiltration of deep crevices; and realizing a detailed, castable brain geometry with extreme negative draft and high-aspect-ratio cavities.
Why: The project aimed to translate MRI data into a castable, food-safe, high-fidelity model for education and demonstration, while evaluating low-cost moldmaking, silicone cure behavior, and release-agent-free demolding strategies under practical constraints.
Learned: Established a food-safe, release-agent-free process, verified substrate compatibility to prevent cure inhibition, and optimized pouring and vacuum degassing to reduce voids. Validated silicone injection and lost-master casting for complex geometries by removing severe negative-draft features, and employing draft-angled, keyed shells; developed proficiency in organic-mesh cleanup, manifold repair, and retopology in Meshmixer, MeshLab, and Fusion 360 for watertight STL/OBJ export.
The initial brain model, segmented from MRI data, retained interior channels and cavities that were larger internally than at their surface openings. Such negative-draft topologies prevent complete mold fill and demolding, making the part unsuitable for casting. To ensure castability, the geometry was edited to include only features observable from the exterior. The resulting mesh contained regions of reduced wall thickness that occasionally failed during additive manufacturing; therefore, a targeted post-print sealing step was implemented to close porosity and prevent silicone leakage into the master during mold making.
A low-density mesh of the MRI-segmented brain was used as a reference to CAD two complementary mold shells incorporating draft angles and keys. This approach reduced silicone consumption and produced precise, self-aligning placement of the elastomeric mold during casting.
To screen for cure inhibition, silicone patch tests were performed on representative surfaces, including FDM prints, water-based acrylic paint, SLA resin, casting wax, sulfur-free modeling clay, and hot-melt adhesive (“hot glue”). Food-safe mold release agents were also tested, such as Vaseline and cooking oil.
Smooth-On Sorta Clear 18 was selected as the casting elastomer. Per technical datasheet specifications, it had a Shore A hardness of 18, was rated food-contact safe, and showed an elongation at break near 500%.
Air entrapment was reduced by subjecting the mixed silicone to vacuum degassing for ≥2 minutes, with periodic venting/agitation to promote bubble coalescence and release.
Before Degassing
After Degassing
The master consisted of a two-part PETG print. Surface voids and parting-line gaps were packed with sulfur-free modeling clay, after which the entire assembly received a water-based acrylic coating. Upon casting, the silicone showed partially uncured regions, consistent with either inhibitor-mediated cure suppression at the interface or insufficient mixing/stoichiometric imbalance of the silicone components.
The master consisted of a one-piece PLA print. Gaps were filled with molten, sulfur-free modeling clay. The silicone jacket was then cast as a thin shell in staged pours: five small batches to cover the surface and limit entrapped air, followed by three larger batches to reach the final wall thickness.
Degassed silicone was introduced in small aliquots to fill the internal cavities of the brain master. The master mold was reoriented between pours to direct flow and improve wetting of recessed features.
After a sufficient silicone build-up had been achieved on the master, it was seated in the rigid plastic support shell without release agent; large-volume pours were then used to completely fill one half of the mold.
Two large-volume pours completed the second half of the mold. A syringe body affixed to the empty shell enabled plunger-driven injection of mixed, degassed silicone. The shells were bolted together to achieve a leak-tight seal and maintain dimensional accuracy.
For food safety, no mold release was applied; demolding was accomplished by raising the shell material above its glass transition temperature (Tg), softening the thermoplastic and permitting mechanical separation from the elastomer without damaging the silicone surface.
Given the extensive negative-draft overhangs and the decision to forgo mold release, removal was executed at temperatures exceeding the PLA Tg, softening the shell and preventing elastomer damage. The silicone was intentionally parted with a wavy (keyed) cut to increase interfacial area, provide self-registration, and improve sealing on reassembly.
The gel’s mechanical response was governed by Bloom strength, formulation pH, and solute concentration. Room-temperature stability was achieved using either 11 wt% medium-Bloom gelatin or 4 wt% 300-Bloom gelatin. To avoid loss of gel strength, no fruit juices or acidifiers were incorporated. Opacity was introduced with evaporated milk; coloration with food-grade dye; and palatability with sucrose and aromatic extracts.
After an overnight gelation period, the gelatin brain cast was released from the flexible silicone mold.
While 4 wt% 300-Bloom gelatin is sufficient for room temperature stability, there are more variables to consider. Throughout multiple recipes, the concentration of gelatin, sucrose, and corn syrup was varied to reach the desired physical properties, alongside texture and taste.
The upside to using >4 wt% 300-Bloom gelatin is that any spills while casting are set in minutes, even at room temperature, and can easily be peeled off, leaving no sugary residue.
My clay sculpture, "Steve" from "Clay Sculpture to 3D in 2 Images" (Other projects), was broken in half. The lower half was exceptionally helpful during silicone casting and the brain molding.
"GELATIN HANDBOOK", Gelatin Manufacturers Institute of America, 2012, PDF Link
To be continued... If I remember to take pictures