Every FDM printer is, at bottom, a machine for depositing molten plastic in a place where gravity and a freshly printed layer below will hold it still until it cools. Take away that solid foundation and the whole scheme falls apart — which is why a writeup on Hackaday covering an embedded 3D printing project immediately stands out. The maker behind it, working from an original build documented on YouTube's Riley's Lab, replaced the print bed's job entirely: instead of stacking liquid onto solid, the extrusion happens inside a bath of gel that holds each bead in place until it hardens on its own. The proof piece is the hobby's default calibration object, the 3DBenchy tugboat — printed not in filament, but in two-part epoxy.
Extruding Into a Gel Instead of Onto a Bed
The trick that makes this possible is a material science curiosity called shear-thinning: a gel that behaves like a solid at rest but flows like a liquid the moment something moves through it. Squeeze a nozzle needle along a path inside the bath and the gel liquefies just ahead of and around the tip, parting easily; the instant the needle passes, the gel reverts to its resting viscosity and clamps back down around the newly deposited bead. Hackaday's summary of the technique describes a bath that will "contain the extruded liquid, yet flow as the extrusion needle slices through it" — support and freedom in the same substance, switching between the two states on demand. The bath itself is a home-mixed formulation: a combination of mineral oil, vegetable oil, and fumed silica, added specifically to make the mixture thick yet shear-thinning. Once a print is finished curing inside the bath, the part lifts out coated in oil, and a rinse in isopropyl alcohol strips the residue away.
Swapping the Hot End for a Syringe Pump
A gel bath solves the "where does the extrusion go" problem, but it does nothing for the second obstacle: ordinary FDM hot ends only know how to melt and push thermoplastic filament. To extrude a liquid like epoxy, the project's second major component is a 3D-printed syringe pump that stands in for the stock extruder entirely. Rather than driving filament through a heater block, it drives a plunger through a barrel — the same principle as a manual syringe, just motorized and mounted where the hot end used to live. That swap is what turns the machine from a plastic printer into something closer to a general-purpose liquid dispenser with X/Y/Z motion control, opening the door to any material fluid enough to be pumped through a needle.
Three Materials, One Escalating Test
According to Hackaday's account, the maker didn't jump straight to epoxy. The material tests moved in deliberate steps, each one raising the difficulty:
- Cream cheese served as the initial proof of concept — cheap, forgiving, and useful purely for confirming that the syringe pump and the shear-thinning bath could cooperate to hold a printed shape at all.
- Sylgard 184, a widely used two-part silicone, was the next step up in ambition but the first real setback: the printed layers showed poor adhesion to one another, and the part disintegrated when it was pulled out of the bath.
- Two-part epoxy was the material that actually worked. The printed Benchy held together as the bath was washed away in isopropyl alcohol, came across as decently shock-resistant afterward, and only showed some stringing as a cosmetic flaw.
That progression matters as much as the final result. Cream cheese proves geometry is possible; silicone proves geometry alone isn't enough — the material also has to bond to itself layer over layer, or the print is just a pile of separately shaped blobs held in formation by the bath and nothing else. Epoxy's success suggests its own chemistry, curing through a crosslinking reaction rather than simply cooling and setting, gives adjacent layers enough time and adhesion to fuse before the part is disturbed.
What It Means for Makers
Nobody is going to swap out their Benchy-shaped keychain fleet for an epoxy one — the process is slower, messier, and far less repeatable than standard FDM, and stringing in a rigid thermoset part isn't the kind of thing a hot end reprint or retraction tweak fixes. The value here is what the technique unlocks rather than what it replaces. Standard desktop printers are fundamentally limited to materials that melt cleanly and re-solidify by cooling: PLA, PETG, ABS, and their kin. A huge range of useful materials never make that cut, either because they cure via chemical reaction instead of cooling (epoxies, silicones, urethanes) or because they're liquid at room temperature with no filament form to begin with. An embedded support bath sidesteps the melt-and-cool requirement entirely. Any pumpable liquid that eventually solidifies — whether by curing, gelling, or a phase change — becomes a printable material, because the bath, not the material's own rigidity, is doing the job of holding the shape until it's ready to stand on its own. That's a meaningfully different design space: parts with the flexibility of silicone, the chemical resistance of epoxy, or properties no filament vendor has ever formulated, printed on hardware built largely from a syringe pump and a tub of oil-and-silica gel rather than a commercial resin printer or industrial dispensing rig.
For makers who already run a filament printer and want to experiment, the barrier to trying this is refreshingly low: the syringe pump is a printable part, and the bath ingredients — mineral oil, vegetable oil, fumed silica — are all available without a specialty supplier. The open questions are the ones any early-stage process faces: how far the shear-thinning bath scales to larger prints, whether other two-part chemistries (urethanes, RTV silicones with different cure profiles) fare better than Sylgard 184 did, and how much the stringing seen on the epoxy Benchy can be tuned out with pump timing or bath viscosity. None of that diminishes what's already been shown — that a home-built rig can print in a class of materials no ordinary FDM machine can touch.
