Ice 3D Printing Uses Vacuum, Not a Freezer

Ice 3D printing is a physics-lab story this week, not a new desktop freezer: University of Amsterdam researchers published a PNAS paper showing a 16-micrometer water jet that freezes in a 2–3 mbar vacuum with no cold plate and no liquid nitrogen. TechXplore’s September 3, 2026 write-up walks through the same setup. For a classroom, that is a visible phase-change demo. For a shop, it is not a FibreSeeker material, and it is not a part you leave on the bench overnight.

What's Happening

Menno Demmenie, Stefan Kooij, and Daniel Bonn replaced the thermoplastic extruder on a commercial ROOK MK1 with a Luer-lock HPLC adapter and a silicon-nitride chip that makes a laminar water jet about 16 micrometers across. The gantry sits in a transparent acrylic chamber pumped to roughly 2–3 mbar. Water evaporates so fast that the remaining liquid supercools, then freezes after it lands. High-speed imaging in the paper shows drops still liquid on impact, then solid in about 0.5 seconds. Trace NaCl went in the water only to cut electrostatic charging. The bed is a sanded plastic plate. It is not refrigerated.

Two print modes are in the paper. Layer-by-layer built an 8-centimeter Christmas tree with a 6-centimeter base in 26 minutes; measured walls were about 600 micrometers wide and 200 micrometers tall for one setting. Slow the head relative to the jet and the ice climbs: pillars lean from about 3 to 36 mm/s of travel, and a 6-centimeter face profile printed support-free in 8 seconds, down to about 14 degrees at the neck. Vent the chamber to 1 bar and the tree melted in about 13 minutes. The authors say this vacuum route runs about 100 times faster than earlier drop-on-demand ice printing on a cold stage, and that it avoids the thermal bottleneck of conducting heat through a thickening ice stack. Treat those comparisons as the authors’ reading of their own method versus prior lab work, not an independent bake-off.

They also flag limits that a STEM class should hear first. Jet speed is much higher than gantry travel, so droplets pile up and walls come out thicker than the orifice. Accuracy is limited by backlash and vibration on a stock printer. Sublimation still happens because the chamber is at room temperature. Mars and Moon pressures sit in a range where evaporative freezing can occur; that is a physics note in the paper, not a habitat contract.

Why Ice 3D Printing Matters for Classrooms and Labs

Most classroom printers teach layer lines, bed adhesion, and “don’t touch the nozzle.” This apparatus teaches something else: latent heat, vapor pressure, and why a part can exist only while the pump is on. You can watch freeze and melt through the acrylic. That is useful even if no student ever builds the chamber. It also draws a clean line between a sacrificial process and a keep-the-part process. Ice is the mold. Plastic and fiber are the thing you still have after lunch.

A shop that already prints jigs knows the keep-the-part side: a clamp, a nest, a cover that has to sit at room temperature and take load. Ice templating is the opposite job. You want the geometry to disappear without solvent residue. Do not mix the two on the same mental BOM. A continuous-tow fixture that holds a pump or a cold plate is still a polymer part. It is not an ice tree, and it does not become a microfluidic channel when you warm the room.

How Ice 3D Printing Differs From a Desktop Tow

Job	Typical process	What you actually get
Classroom freeze/melt demo	Water jet in 2–3 mbar vacuum	Ice that lasts while the pump runs
Sacrificial microchannel	Print ice, cast, melt or sublime	A proposed lab template, not a SKU
Prior 3D-ICE work	Drop-on-demand onto a −35 °C stage	Microscale ice; needs a cold platform
Bench fixture / clamp	Desktop thermoplastic + continuous tow	A designed load path that stays solid
PLA classroom model	Standard FFF	A keep-the-part print, not a phase-change lesson

What the Research Says

Demmenie, Kooij, and Bonn’s PNAS paper is the source for the vacuum numbers above: 16-micrometer jet, 2–3 mbar, 0.5-second freeze after impact, 26-minute tree, 8-second face, and the claim that evaporative cooling at the liquid surface bypasses conduction through the ice already printed. They also report that water on ice keeps a finite contact angle, which they say is what stops the puddle from running down the wall before it freezes (Demmenie et al., 2026).

Garg, Yerneni, Campbell, LeDuc, and Özdoğanlar’s 3D-ICE process is the cold-stage contrast the Amsterdam team is answering. They jetted water onto a −35 °C platform, printed overhangs and branched features down to about 50 micrometers without supports, and used the ice as a sacrificial core inside resin. That work needed the refrigerated platform the new paper is trying to drop. It is still the peer-reviewed template for “print ice, cast, remove water.” Do not treat either paper as a catalogue for classroom hardware (Garg et al., 2022).

Frequently Asked Questions

What is ice 3D printing with evaporative cooling?

It is a lab method that shoots a thin water jet into a 2–3 mbar vacuum so evaporation pulls heat out of the remaining liquid until it freezes. University of Amsterdam physicists report no cold plate and no liquid nitrogen. Vent the chamber and it melts. That is a physics demo, not a desktop filament job.

Can a desktop carbon fiber printer print ice like the Amsterdam setup?

No. The PNAS apparatus is a modified FFF gantry inside an acrylic vacuum chamber, with an HPLC pump feeding a 16-micrometer nozzle. A desktop printer that lays thermoplastic and a continuous tow makes room-temperature fixtures. It does not evaporate water in vacuum, and it does not leave a sacrificial ice channel you melt out of a mold.

What can you use a 3D-printed ice template for?

The authors point to sacrificial templates: cast a material around the ice, then melt or sublimate it to leave a clean void for microfluidics or tissue scaffolds. That is a proposed use, not a product you can order. Treat the Christmas tree and face as lab geometries. Do not treat them as a classroom kit or a Mars habitat.

Fibricate's Place in This Story

A vacuum ice jet and a desktop continuous-tow printer answer different lab questions. Companies like Fibricate, whose FibreSeeker 3 continuous carbon fiber 3D printer lays polymer with a tow from feedstock such as the continuous carbon fiber spool, are not printing ice trees and are not melting channels out of a mold. Use the desktop machine when a fixture, clamp, or cover has to stay solid and carry load at room temperature. Leave evaporative freezing, HPLC jets, and sacrificial ice to the physics group that published the chamber. Do not treat a carbon-fiber jig as a microfluidic template, and do not treat a PNAS Christmas tree as a FibreSeeker spec.

What to Watch Next

Watch whether anyone copies the chamber with a stiffer gantry and a slower, smaller jet so walls can approach the orifice size. The useful public signals are independent freeze-time measurements, how long a thin strut lasts before sublimation wrecks the geometry, and whether a cast-and-melt microfluidic chip shows up with channel dimensions, not just a holiday sculpture. Over 12–24 months, expect more lab demos that make phase change visible, and more shops that still print the clamp that holds the real pump. The split that will still matter in 2027 is parts you keep versus ice you are supposed to lose.

References & Further Reading

1. Demmenie, M., Kooij, S., & Bonn, D. (2026). Three-dimensional printing of ice structures via evaporative cooling in vacuum. Proceedings of the National Academy of Sciences.
2. Garg, A., Yerneni, S. S., Campbell, P. G., LeDuc, P. R., & Özdoğanlar, O. B. (2022). Freeform 3D Ice Printing (3D-ICE) at the Micro Scale. Advanced Science.
3. Supercooled water enables 3D printing of ice. TechXplore / University of Amsterdam. Retrieved September 4, 2026.