3D Printed Metal Fail

This was months ago now, and I don’t normally show off my nutty ideas if they don’t work out, but I thought this was just crazy enough that it might be entertaining.

The basic idea is this:

  1. Have a pool of molten metal
  2. Dip the part into the metal just a little bit so that a new layer of metal solidifies on the bottom
  3. Machine away the unwanted metal
  4. Repeat to build up the object layer by layer (downward, like an SLA style printer)

The hope was that this could produce 3d printed metal objects directly. It seemed that each additional layer could naturally be slightly wider than the previous layer, meaning overhangs would be possible and it should be possible to get the same level of complexity that’s achievable with 3D printing. Supports might need to be very different because they need to support cutting forces, but something like benchy should be possible.

This picture shows the solder pot used to keep a pool of molten metal. I used bismuth-tin eutectic alloy that melts at about 135 C or 281 F. It is a bit more expensive than some other low-melting point alloys but it is nontoxic, which is a requirement if I am generating dust flying who knows where.

The little round plate on the Z axis has some metal screws in the bottom for the metal to adhere to, and the MDF provides some degree of thermal isolation so the plastic never gets too hot.

The router is mounted inverted and the endmill pokes up through a hole in a cardboard box, the hope being that I can capture and recycle the majority of the metal that’s carved away.


Alas, I never got this to work. In my mind I was thinking of how candles are dipped, and each layer sticks to the previous layer building up layer by layer. But in reality the metal really does not want to adhere to the metal that’s already been deposited. I started manually and eventually wrote some gcode to just dip repeatedly, allowing some time to cool between dips. And it would not build up. Once or twice I got a little bit (like the equivalent of a few drops of water) and when I tried milling, the whole thing broke away very easily as soon as it encountered the endmill.

Maybe this has to do with an oxidation layer and some flux or an inert atmosphere might help. Or maybe it’s just the crystal structure on the surface doesn’t grab. Or maybe (probably) something else that I’m not even aware of.

So I gave up. Disappointed of course, because 3d printing metal with ordinary tools would have been pretty awesome. But at least I got to cross it off my list and move on.

3 Likes

No worries, it’s true what ‘they’ say.

When you’re closing in on the finish line you won’t regret the things you did but the things you didn’t do.

Made me think of this:

The world is built on nutty ideas that don’t work…eventually some do work then become cutting edge. I wish I were more creative and ambitious to develop something completely new. Well done!

There was a kickstarter that welded (like the HAD article) and then milled away the extra. There were some serious challenges, like making something you can firmly mill, but also remove at some point.

Surely there are some more knowledgeable material scientists here. But I wonder if the key for attachment has to do with temperature. If you were sinking a lot of heat through the base, would it cool the metal enough to solidify while attached to the base. I suspect the real answer has more to do with the relative charge of the metals.

Interesting approach. It’s cool to see you trying things like this.

I’m seriously interested in trying to produce some “3d printed” metal parts, but I’m planning to try the much less adventurous lost PLA casting approach.

Don’t give up! This is a promising idea.

The troubles you encountered are almost certainly due to oxide build-up, as you guessed already. Here’s a table of oxide thickness measurements on various solder compositions. As you can see, Bismuth-Tin solder forms much thicker oxides than even regular Lead-Tin solder does. In industry, this type of solder is always used with either copious amounts of flux or inert nitrogen atmospheres. Also, temperature needs to be quite high (considerably above the melting point), so that you get good flow and the initial oxides dissolve inside the material (instead of forming an impenetrable shell).

Even assuming a nitrogen atmosphere, you’d still have issues keeping the molten metal running over the rest of the part via surface tension. What you’d need to do is to ‘passivate’ the rest of the material, except the top surface where you actually want metal to stick. Perhaps a way of doing this might be to have some kind of oxidizing solution into which you dip the part into. So the process would be: Dip the part into metal, dip into oxidizer, then mill away the top (exposing fresh metal). Repeat.

The other problem would be keeping the molten metal from just melting the rest of the material through heat transfer. You could make this work by having the molten metal at high temp, and dipping very quickly.

Interesting concept anyway.

3 Likes

It’s only semi-related, but BASF and The Virtual Foundry sell FDM filaments that consist of metal powder (in a plastic delivery vehicle) that one can print with a ghetto FFF printer, then anneal and finish in a kiln.

No direct experience with either set of products [yet], but it’s compelling stuff.

1 Like

Printing with metal powder and then debinding and sintering (not annealing) has been a thing for quite some time, a few companies offer full solutions for this, like Rapidia and Markforged, which are essentially just modified FFF printers.

The main problem is that debinding is really messy and hazardous and sintering requires a pretty decent (and expensive) furnace with good temperature control. You probably won’t get good results with a pottery kiln, lol. This is fairly specialized stuff, I have yet to see someone pull it off with just hobbyist-level equipment (I’d love to be proven wrong). The cheapest systems I’ve seen run at least a few tens of thousands of dollars in purchase price, and aren’t cheap to operate either.

Some companies offer a process where instead of embedding the powder in plastic, it’s delivered in a solvent slurry, so no debinding step is needed. This is slightly cheaper but still expensive.

1 Like

All interesting. Thanks. Another option that seems perhaps plausible is if there were a blend of say iron or zinc powder, mixed with bismuth+tin powder, would it be possible to use a low-power laser of the “5W” variety to melt the bismuth granules and fuse into the previous layer. This would be like a SLS process where a separate apparatus would drop the bed slightly and reapply a thin layer of powder for each layer.

One concern is whether the laser is powerful enough to heat the powder hot enough. It seems plausible since it can scorch wood. The other, perhaps bigger concern is what happens when it does melt, if the surface tension would move everything around and mess up the shape. Maybe if the particles are small enough and the percentage of bismuth powder is low enough so it just binds the powder without pooling into liquid puddles? All I know is I once saw a video of using a big magnifying glass and sunlight to fuse sand into an object layer by layer and the surface tension moved the sand grains significantly and the resulting product was pretty so-so.

Unfortunately this SLS concept is far enough down on my priority list of projects that it’s very unlikely that I will ever try it.