3D printing beautiful glass structures


[Read the post]


Nothing surprising. This is a variant on many of the manual glassworking technologies, except more automated and less varied. (But also pretty flexible.)

The glass object has to be kept above its annealing temperature, to avoid buildup of stresses and eventual cracking. At this temperature the glass is still pretty hard solid but the structure is still flexible enough to move around a tiny bit. Annealing kilns are common for glassworking; finished objects are put there so the stresses within the material can relax.

The feed has to be preheated, to be kept at the right viscosity for transport to the head, if supplied as liquid. (It could also be supplied as a glass rod, or a glass fiber. In both cases an extruder similar to the low-temperature (in comparison) plastic ones could be used. The rod could be actually made “online” - feed liquid glass through a nozzle, make a rod of a desired diameter, let it cool down to solidify enough to be pushed, and push it into the nozzle to extrude it.

Then there’s the nozzle. A refractory metal of suitable kind (common tech in extruded glass industry) is used, with appropriate heating. The extruded glass filament is then laid to the preheeated glass printout, and joins with the base material.

As the feed of molten glass to the moving head may be difficult, a design with stationary head and moving print bed (or head moving just along the z-axis) is the likely choice.

Not that difficult, in principle. I’d say the worst will be the fine-tuning of the parameters and finding a way to avoid bubbles.

A multihead extruder, working with several kinds of glass (different colors, or different refractory index or other optical properties, or maybe different chemical properties - think biocompatible surface for example, or leaching certain ions, or etchable…), could bring the ability to make objects impossible to be made in other ways.

Amorphous materials are pretty similar in behavior. So polymers are pretty similar to glass. If you play with hot glue sticks and flame (pretty useful for encapsulation of parts or gluing together things, can work way better than glue gun in certain scenarios), you have a good practice for lampworking glass.

Edit: And you can put all the actuators out of the furnace/kiln. All you need are sliders from refractory materials. All the tech is already mainstreamed by glassworking industry, decades old. The only thing new here is assembling it into a 3d printer (a look-and-see idea, rather obvious) and tuning it up (the real legwork is here, and major kudos for it).

A major improvement would be bringing the accuracy/precision to a fraction of wavelength of light, so optical components could be directly printed. As of now, however, the closest approximation is a near-net printout for subsequent grinding. It would save material and time, but a full-precision printing would allow combining multiple materials into photonic and even metamaterial structures.

Thought. Could this be achieved by a sol-gel method? An inkjet-like head printing different “colors” of gel precursors, layer by layer. Then cure the gel, dry it, it will isotropically shrink (so the printing can be done at lower precision as we’re working with a bigger structure). Optionally heat-treat but some gels can cure to amorphous oxide even without. You could use water (or other chemical) soluble material as support, so the lens (or laser rod or photonic crystal or microchannel plate…) is then liberated from the cured block by immersion in the “developer”. All sorts of physicochemical shenanigans are then possible; e.g. some parts could be made of oxides that can be easily reduced in e.g. hydrogen atmosphere, for precision positioned metal coating for e.g. said microchannel plates…


I’d agree on the ‘fundamental novelty’ point; but (given the number of fun ways to screw up even polymer extruder systems with a plastic chosen for playing nice); I suspect that ‘optical quality glass’ is a 3d printer design where a great many Devils lurk in more details than you expected, even if you went in with the assumption of unknown unknowns.

Now that they have this working, though, I’d love to see if they can get controlled combinations and internal boundaries of chemically distinct glasses or glass dopants. It isn’t uncommon for a nice lens to be a combination of multiple(sometimes a dozen or more) distinct elements made of different glasses to attempt to control chromatic aberration, get better uniformity across all relevant wavelengths, etc.

Being able to do that in a single chunk, without fabricating and assembling numerous elements would be pretty damn cool.


I think “optical quality” is a misnomer here. At least I understand “optical quality” as “put the model in, get an eyeglass or telescope or microoptics lens out”. Lighthouse-grade non-imaging Fresnel lens does not quite count.

Otherwise I agree, and that’s exactly what I meant by the kudos for tuning it up.

Exactly what I was suggesting, too! :smiley:
But this won’t be much likely to be achievable with molten filament deposition, at least not with macroscopic diameters. Rough blanks, maybe, but they won’t have the internal interfaces defined well enough so multimaterial printing of imaging lenses as we know them is out.

So far it looks like a good beginning for design and architectural and artsy fartsy and some technical applications.

Or for printing blocks for optical fibers; make a macroscopic block with roughly defined internal structures, then draw it thin.


Can we put the printed object over a pot of boiling HF in order to smooth out the surface?


Yes but you’ll likely get just a frosty appearance.

A mixture of sulfuric and hydrofluoric (NOT hydroflouric!) acid is used for chemical polishing. Not sure it could eat enough material for major smoothing, though.


This topic was automatically closed after 5 days. New replies are no longer allowed.