Better performing 3D prints with annealing, but... - Part 1: PLA

I guess everybody who’s owning a 3D printer for a while has heard about annealing prints to make them stronger and more heat resistant. The idea is pretty simple: You heat up your oven to a certain temperature, toss your parts in and after 30 to 60 minutes, you switch it off and let everything cool down slowly. There are plenty of videos about it here on YouTube and I also already made a couple and showed how that process can improve the heat resistance of PLA from 60 to even 180°C or increase stiffness. Though besides Tom’s video where he investigated the strength of annealed PLA in his Filaween series, there isn’t a lot of information about the change in strength around. This is why I finally thought it’s time to perform an investigation which is at least a bit scientific, in order to find out if rumors like if we can really fuse layers together and similar things really hold true. This will be a video series and, in this first video, we’ll take a look at the most common material, PLA. In the future I also want to find out if we can improve the mechanical properties of materials like PETG, ABS, Nylon or Polycarbonate.

Before I spoil you with my results, I’d like to know from you if you ever heat treated your 3D prints and what your experience was. Also let me know of any other investigations or paper I might have missed and put you experience and ideas down in the comments!

This video will be about PLA because this material has the special property that you can increase the crystallinity in the material by this process which is not possible in other materials like ABS or PETG. This means, that during the annealing process the polymer chains that are initially in a dominantly amorphous state, so unordered, rearrange in a more orderly fashion and the so called crystallinity increases. This change in structure also changes the properties of the material, predominantly heat resistance, stiffness and other mechanical properties. If you ever tried this process out on your own, you’ll know that the parts can also deform quite a bit.

In order to investigate the annealing of PLA, I printed a couple of test samples out of Formfuturas Premium PLA and then heat treated half of them to compare them to the reference later. I chose this material on purpose because it’s not a hugely modified PLA which is probably good for the crystallization process and due to its translucency, we’ll be able to see another optical change in the material. With higher crystallinity PLA becomes more opaque. I printed samples to test the strength and impact resistance of parts that were printed lying so where layer adhesion doesn’t have a huge impact but also standing specimens where we can see if the layers fuse together more. I also put a 3D Benchy and fan shroud on the print plate to nicely witness how a realistic part might change shape during that process.

For the annealing process I put all of the parts in my pre-heated convection oven at a set temperature of 100°C for 45 minutes then turned it off and let the parts cool down to room temperature over a couple of hours. Taking a look at the treated parts themselves already showed that something changed with the material. Most obviously some of the parts were really deformed and for example my layer adhesion samples were totally bent. Fortunately though, the standing test-hooks still looked usable. I measured a couple of parts and the dimensional change was up to 10%. The interesting thing is, that the parts contract in xy direction, so a 50mm part might come out only measuring 45mm but they expand in the z-direction for example from 50mm initially to 55mm after the process. Unfortunately, the amount of deformation is not always the same and can rarely be compensated by just scaling the parts. This is particularly obvious if we take a look at the fan shroud that just warped all over the place. There are a couple of ways how you can minimize that effect. One way, that I often use due to my removable print beds is, that I put the parts into the oven still attached to the base with all of the supports and that greatly reduces a lot of warping and keep at least the base straight. Another one that I tried out is submerging the parts in sand. I tired that in a couple of pre-tests where it helped a bit but not to the point that it totally eliminated warping. I’ll probably try it in the future with a bigger container so that the parts are more thoroughly held in position. Due to the thermal mass and the insolating properties of the sand this might be a possibility to even out the temperature change if you have an oven that’s not properly controllable. There are also specialized materials, like Formfuturas Volcano PLA around, that barely warps during the annealing process but still is able to improve the heat resistance significantly.

Untreated (left) and annealed (right) 3DBenchy

Another change that we see is that the parts become more opaque because they lose a bit of their amorphous structure. A good indication that the microstructure really changed.

But now let’s get to the real material tests you probably have all been waiting for. And if you enjoy these tests and learn something than also make sure that you’re subscribed and have clicked the bell to not miss any upcoming videos! Let’s at first start with the tests performed using my DIY Universal Test machine. The standard tension test specimens showed an ultimate tensile strength of 63.5MPa on average. All samples yielded a good bit until they failed. Next I tested the three specimens that I annealed while still attached to the printbed because the other ones were just too distorted. These samples failed at 68.2MPa on average. This is around 7.5% more than the untreated ones and with the low scatter statistically significant. They also failed quite similarly with a bit of plastic deformation before rupture. The plots look very similar and I can’t really say that there has been a huge increase in stiffness noticeable. Also previous tests that I made in the past did only show an increase in material stiffness of around 7%.

Next let’s take a look at the results of the test hooks that I use to test a more realistic load scenario with tension and bending. So at first the lying hooks. The untreated one failed at 75kg of load whereas the annealed one was able to bear a good 16.2% more and ripped at 87kg of force. This improvement is significant and very interesting but I have to say that due to the shrinkage of the hook the bending moment slightly decreased which spoils the result a little, but still a good improvement! The most interesting point of this investigation was the layer adhesion that we test with the hooks that were printed upright. The untreated ones failed on average at 42kg of load and all cracked right between the layers as one would suspect. Now let’s take a look at one that was annealed. They also all failed, on average, at 42kg of load. All right between the layers. So, this is quite a bummer but kind of confirms what I also already experienced in the past. At least in all of my investigations I never was able improve layer adhesion with this heat-treating process. The layers don’t melt together better. And trust me, I tried! In my pre-test I even tried higher annealing temperatures of 140°C and 180°C with parts that I submerged in sand. I had to use the sand for the higher temperatures because otherwise my test hooks just melted away directly. Also, for these, no real change. Some suggested stress-relieving the parts at first at temperatures right around the glass transition temperature. Did that for 5h at 55°C and then additionally annealed them at 100°C for 45 minutes. There was a change in the amount of warping I got, because that was reduced, but again, no significant improvement in layer adhesion.

Last but not least I tested the impact resistance using notched impact test specimens. The parts are struck by an impact hammer that has a known amount of kinetic energy before the impact. The amount it swings further after the strike determines how much energy was used during the impact thus telling us the impact resistance. Here, I would have suspected that the annealed specimens behave more brittle, but that wasn’t the case and all 6 specimens were able to absorb roughly the same amount of energy.

I don’t really know if I want to call annealing of PLA busted at this point. Well, this process definitely improves the thermal resistance of the materials significantly but in terms of mechanical properties the gain of around 10% is in my opinion only very small and I have definitely not seen anything in terms fusing the layers together, I’m sorry to give this verdict. My research hasn’t brought up any real way to fuse these layer boundaries together without re-meltimg them. This is especially all overshadowed by the warping problem that you usually have when annealing PLA. This can be tackled by material choice, stress-relief or maybe submerging the part in tightly packed sand but probably still will cause you some issues.

Still, I think what shouldn’t be overlooked and what is something that I will tackle in a future video is the stress-relief annealing. Due to the layer-wise build-up process and the shrinking that happens during cooldown our 3D printed parts still have quite some internal stresses and these can reduce strength and especially ductility of the material. This procedure is something that can be done with basically any material, will be very interesting and will have more impact on the mechanical properties than rather on the thermal resistance because it doesn’t involve a change in crystallinity.

Now, please let me know in the comments if the results of this annealing investigation is something that you have expected or if you made a difference experience in the past. If you think I should have done something differently I’d also be really happy to know and will investigate that in the future!