Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
MIT researchers demonstrate rapid liquid metal 3D printing technique (tctmagazine.com)
100 points by Tomte on Feb 18, 2024 | hide | past | favorite | 47 comments



It is a shame we're seeing a divestment in 3D printing technologies. Relativity Space recently descoped their entire 3D printing department, but I believe they still may be using some commercially-acquired printers.

I hope we can see someone else take up the mantle on complex 3D printing manufacturing. There's an Indian space company that is trying to do what Relativity did. I hope it works out for them.


It's been amazing to watch the 3D printing space grow as far as it has. In just the last 10-20 years it's gone from something only dedicated people were able to build and get reasonable results to high precision devices any hobbyist can get their hands on for a few hundred bucks.

I have two resin printers and a PLA printer and I never expected at-home capability to get this far this fast.

But, that's still all effectively plastic we're talking about. I think the problem with metals is still well.. metals. The same types of metallurgy needed for 3D printing have been researched and hit almost a dead end with injection molding (I know there are some metal injection molding systems out there, but it's not hit anywhere near the strength of machined steel yet)


People are 3d printing metal right now with LPBF, DED, and Wire EDM. For LPBF the challenge is controlling keyholing and lack of fusion defects through the process parameters.


BBarn, how far along are the circularity tools, can I turn clean used plastic jugs back into feedstock worth using to make 3D stuff?


Something like a PET milk jug is harder since you’d need to shred it and remelt a good portion of the jug into filament since the handle area would be hard to deal with, which is hard to do in a DIY manner. Same for reprocessing failed prints and scrap support pieces of prints, plus various coloring agents and manufacturer additives make a blend of recycled bits inconsistent. It’s probably easier to DIY a high temp composter than to recycle PLA.

But there’s a ton of people out there with jig designs to spiral cut a normal cylindrical bottle and feed it into a hotend that creates filament from it. Here’s one example https://www.printables.com/model/768657-petalot-plastic-bott.... YouTube has dozens of videos of these in action. Generally speaking you won’t get as good of results as commercial filament since filament diameter needs to be carefully controlled and it affects flow rate which then affects resulting print quality. If you print something simple, large, and practical it’s fine, if you need something finely detailed it can be fiddly.


There are many designs for DIY filament makers that use 2L pop bottles.


Which kinds of plastic are reusable as filament?


HDPE and PET are the most common I’ve seen in the community, which probably accounts for the majority of beverage containers you’ve ever encountered.


I thought Relativity Space pulled back on their use of 3D printing for the major fuselage parts, saving it more for the engine components, etc., which SpaceX was already doing.


I believe you are correct. Relativity is now only printing engines, just like SpaceX, Rocket Lab, Astra, Firefly, Stoke, and others. They are no longer a company driving forward 3D printing technology and that's too bad.


- "It is a shame we're seeing a divestment in 3D printing technologies."

Is there other data to support that claim we can browse? I was under the impression the opposite was true but would love to know more if i'm missing something.


Check out Mantle


Neat to see this!

As the technology improves, I am skeptical of the necessity of sacrificing resolution. I can think of a half-dozen ways to improve on this as it moves from V0 to V1 to V2.

The video mentions post-machining, but aside from that, we have:

- More sophisticated media than uniform glass pallets. In particular, an outline could be preplaced, so it is more like casting. For example, one could have large glass pieces, smaller glass pieces, etc. arranged prior.

- Combined methods, where this prints the coarse shape, and slower techniques fill in details. This could even be the same technique but with successively smaller nozzles.

And simple process improvements (esp. refining nozzle design, temperature, and flow rate, so in some ways). I suspect moving from liquid to something more like extrusion, could help as well. Extrusion can be quite precise. Better process control, with some form of feedback loop, would help too. Imaging aluminum in realtime through the glass pallets in realtime should not be fundamentally hard.

By "not hard," I mean "known engineering process with known technologies" (e.g. solvable, but a serious multiyear engineering effort).

First press release + video I've seen from MIT which doesn't oversell / overhype results or grossly exaggerate potential impact. That's refreshing too.


Wasn't this on HN a few days ago?

There are already good metal printers. They're usually sintering systems - put down a powder and fire it in a furnace. There are systems for 3D printing sand molds for casting.[1] Both of those processes are much more precise than this one.

Metals do usually have a phase in which they're solid, but malleable, and can be worked with modest forces. "Modest" usually means hammers, large presses, or a rolling mill. A good metal extruder working with red-hot but not molten metal would need to be able to apply similar forces. That's what the 3D printers that work with wire and are similar to welders do.[2] There's directed energy deposition, which converts metal powder to molten metal for a fraction of a second at the deposition point.[3]

This liquid system has the same problem as the 3D printing systems for concrete that were being touted a few years ago. Some kind of molding or die is needed to guide the material at the point the metal becomes a solid, or the result is very rough.

[1] https://www.voxeljet.com/3d-printing-solution/sand-casting/

[2] https://www.youtube.com/watch?v=z3nNlVcoba8

[3] https://www.youtube.com/watch?v=oL7bMhPTtDI


Also the Mantle printers look promising



Cool idea. Impractical but cool.


This was posted before. Not really 3d printing. It's more of a quick casted mold maker, since it just uses a conventional CNC machine to "draw" the shape in sand and then it pours molten metal in the sand mold.

A clever enough idea, but the results are mixed. It's limited to fairly simple 2d shapes and the quality of the "prints" are quite poor and require a lot more post processing than true 3d metal printers.


Here's an older HN link from an MIT article.

https://news.ycombinator.com/item?id=39156463


Lots of progress, still clearly many improvements possible. Baby steps! This could be very useful in a lunar colony, quickly smelting parts and structures out of the lunar soil. It's around 8% aluminum (bauxite) which melts at a reasonable temperature. Strong enough for larger-scale building uses such as airlocks and beams.


I wonder if you could just 3D-print a mold and then pour the aluminum in as part of a 3d printer like this, rather than use a nozzle, and you'd get better quality/resolution.


Or a 3 step process: PLA mold -> silicone mold -> aluminum object.


Here’s a commercially available high resolution DLP 3D printer people use to make molds for casting gold and silver parts: https://www.junction3d.com/

There are downsides, too: printing is slow, and the build area is small, furniture parts not gonna fit there.


You would need a ceramic mold, but that might be easier to 3d print?

Right now it is in a bed of fine glass beads that the printer drags through as it goes, so this is kind of a similar approach. For one offs or prototypes this might be good, but when you print more than one of something than a mold likely makes more sense.


Aluminum sandcasting is what you're looking for.

Sand is cheap, and should be easy to 3d print, even for one-offs.


Sand molds have to be compacted.


I'm thinking about the other way around. Assuming that their aluminum 3D printer works, print the aluminum mold using their 3D printer. Then make plastic parts with injection molding.


Injection molding molds have precision requirements well beyond any 3d printing.

They're very expensive for that reason. It's hard even for precision machining.


You can print molds that you literally cannot machine.


For injection molding? Or other types of molding?

Injection molding processes I've seen have super-tight tolerances. Something like plaster or sandcasting do not.


But how do you get the product out!?


You break the mold


I've seen hobbyists do investment casting with FDM-printed patterns. I assume something similar in spirit is available to industry if they want to do it?


You can but then you lose some of the advantages of the 3d printing technique e.g. the ability to create complex internal geometries


A 3d-printed mold could actually do that. It'd just be a destructive process -- the mold would not survive.

Ceramic is cheap enough that this might be okay.

Removal would be a pain. Probably some chemical which melts ceramic but not metal. Perhaps something mechanical which relies on ceramic being brittle to break it up. Dunno.


Silicone is easy to remove.


This is already done using sand casting. However, it's low resolution.


Low resolution is a relative term. It certainly beats this.


What class of fabricated metal does it compare to? Cast metal? I didn't see it in the article.


Well, that clickbait headline certainly has James Cameronesque overtones, does it not?


How much energy does this use?


We know enough from the article to be able to ballpark estimate it. Heat capacity of aluminum is 0.9 kJ/kg °C, and the latent heat of fusion is 390kJ/kg. The temperature they heat the aluminum to is 700°C, from presumably room temperature around 20°C - so 0.9*680 + 390 kJ/kg = 1002 kJ/kg is going to be the basic heating requirement, or as near as dammit 1MJ/kg. Getting that heat into the aluminum and then maintaining that temperature during the print presumably introduces some inefficiencies (induction furnaces can heat aluminum with about 40-50% efficiency), plus they have to move the mass of aluminum around with their print motors... but the heat to melt the aluminum feels likely to be the main energy sink. Inductively heated, with good crucible insulation, 2MJ/kg sounds like a reasonable energy budget?

For a comparison, that's about ten times the amount of energy you use running your microwave to heat up a ready meal, for every kilogram of aluminum you want to form.


Ok. How much less efficient is this process compared to mass-fabrication techniques?


That energy cost is intrinsic to aluminum casting. If you’re recycling aluminum like this, you are going to have that cost anyway - even if you use a different technique like rolling or pressing or extrusion to form the final pieces (that said I bet induction heating aluminum scrap to melting point is a lot less efficient than heating an ingot)


Mass production is not only about energy efficiency. It is also about material efficiency and time efficiency. People want stuff NOW. Each process has its advantages. 3d printing can do stuff that mass fabrication techniques can't.


>It is also about material efficiency and time efficiency. People want stuff NOW.

Also, if you have a $1M machine then the machine-time really adds up.


Precisely. $1M and interest rates of 6% per year mean almost $7 per hour just to cover financing.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: