The influence of layer height on the strength of FDM 3D prints

Most of you probably set the layer height to adjust the amount of detail of your 3D prints and how smooth the surface finish shall be at the end. Thinner layers give you less of a stairstep effect but will increase the print time. Actually, the printing time is inverse proportional to the layer height, so using layers half the thickness will more or less double the print time. Since most of my prints at least, don’t always have to be horribly pretty, I’ve been asking myself for years if it’s better for the strength of my parts to print with thinner or thicker layers. If I’m talking about strength at this point, I actually mean layer adhesion, so how good the individual layers of material bond together. But I also analyzed the strength of samples that were printed laying down to find out if we also have an effect there. If you do some research online there are contradictory results you can find. Some claim higher strength with thinner, some with thicker layers. Thinner layers could be stronger because the molten material is squished out more from the nozzle and due to the low distance between nozzle the previous layer warms the material and helps with bonding. Also, since you extrude less plastic in a given amount of time the material stays in the meltzone for longer and therefore properly and evenly warms up and melts. Also, the density of parts with thinner layers could be higher due to smaller gaps between lines of already printed material. There is also a very interesting article on flow math, so the ways the amount of material extruded is calculated, on the Slic3r wiki, to which I leave a link in the description.

Though especially the point that the material stays melted for a longer period of time and therefore potentially degenerates is an argument that thicker layers might give you parts with higher strength. With thicker layers the positioning tolerance of the Z axis also doesn’t play such a big role, because the tolerance is a smaller percentage of the total thickness. What I mean with that is that for example a 5um tolerance at 50um thickness is way worse and could result in -+10% over- or underextrusion. Whereas a 5um tolerance at 250um thickness gives you is only +-2%.

Then there is a statistical aspect. With thicker layers you consequently print less layers for the same part. This decreases the risk that in one layer something goes wrong, which in the end will be the location where a failure starts. This concept is also called statistical size effect. A chain that is longer has a higher probability to fail earlier than a shorter one. Because with more links, the probability is higher that one of the links is particularly weak. I’ve put a couple of articles and papers on this topic down below if you are interested in further reading.

In order to answer the question for myself, which layer height is the strongest, I printed a couple of dozens of my test hooks with varying layer heights and orientations and tested them on my DIY Universal Test Machine. If you like these kinds of investigations then really make sure that you are subscribed to the channel and have activated the bell, because still 4 out of 5 watching right now are not yet following the channel! Also consider supporting my work on Patreon, that helps me to spend more time on these topics.

So, the test setup was as follows: As the test geometry I used my test hook which I’ve also already been using for various other investigations in the past. This is a generic test part for me where the critical section is loaded in tension and bending, so a more realistic loadcase as just pure tension that you test on standard tension specimens. I didn’t use a standard tension test geometry because it’s harder to print in a standing position and I already noticed quite some scatter in the past, maybe because my test machine allows for some misalignment, which is not so bad for the test hook because it aligns itself. All parts were printed on my Original Prusa i3 Mk2.5 in Prusament PLA using a hardened steel nozzle and their default settings. I deliberately chose this material since Prusa Research claims a very precise filament diameter over the spool so flow should actually not vary, eliminating one more variable in the test. I chose the standard 0.15mm preset in Slic3r PE set 3 perimeters and then only varied the layer thickness and increased top and bottom layers to get a consistent wall thickness. I ended up testing 6 different layer heights, 0.05mm, 0.1mm, 0.15mm, 0.2mm, 0.3mm and 0.4mm. 50um is probably really on the small side and I’ve never been printing with layers this thin. 0.4mm is beyond what you should print with a nozzle that has the same diameter, but I wanted to see how bad the results really become. Before we continue, please click the Information Card in the upper right corner and vote for the layer height you think will be the strongest and leave a comment down below why you think this is the case!  For the standing hooks I always printed 3 at once for statistics and also that layer times are in a realistic range so that the material has time to cool down. I actually wanted to print the lying specimens all at once and if you didn’t know, you can actually have more than one layer height in a print job but this one failed due to some clogging problems at 0.05mm layer height. In the end I printed all of them separately.

As mentioned in the beginning, printing time is inversely proportional to the layer thickness but the first interesting result was that not all the hooks weighed the same. The higher the layer thickness the less the parts weight, which is probably explainable with the fact that thicker layers feature more gaps and are therefore less dense. We’ll keep these values in mind when we take a look at the strength later.

Sample weight VS layer height

Another result that was interesting for me is the print quality, which in my opinion didn’t improve the thinner the layers got. Yes, side surfaces usually looked better since the layers washed out but all in all, the sweet spot in my opinion was at 0.15mm where overhangs looked the best and also bridging was really nice. Another thing that I noticed was that the color of the parts started to get irregular especially at 0.05mm. This is probably because the material is in its molten state for the longest amount of time, messing with the pigments and maybe even the base polymer.

I put all of the samples, one by one into my DIY universal test machine and loaded them, very importantly, all at the same speed, until they failed and meanwhile recorded the load values. At first, let’s start with the samples that were printed laying down, so for which layer adhesion should actually only play a minor role. I initially thought that at this orientation we shouldn’t be able to see any major difference. I was wrong there because even though 0.05mm up to 0.2mm failed all in the range of 80kg, already the 0.3mm sample failed quite a bit earlier, not to speak of the 0.4 mm hook. If we take a look at the weight normalized values, for which I divided the failure load by the weight of the sample, the picture stays pretty similar, only the thick layers became a little better due to their reduced weight. The fracture surfaces are also very interesting because for the thicker layers the gaps between the filament lines are clearly visible and the thinner you go the more it looks like an injection molded part. So the first result is, that even though your part is not loaded in-between the layers, you shouldn’t go much above 0.2mm layer height because that will reduce the strength of your parts.

Next, we go to the way more interesting investigation, the layer adhesion test, where I tested the hooks that were printed in the standing orientation. First, all parts kind of broke at the same height which means that no severe printing errors happened which caused a premature failure. Still we can clearly see that all samples snapped perfectly in-between the layers, so layer bonding was a major weak point and I have to be honest, I have seen better results with other brands of PLA filaments in the past. Also if we really take a close look at some of the samples, we can see signs of under-extrusion, where the printing process wasn’t 100% perfect. This might also be the reason why we see some scatter in the results. Quite similar to the last samples, the strength of the 0.05mm to 0.2mm samples were the best with the medium layer height of 0.15mm being the strongest and then quickly plummeting with 0.3mm layers that were only able to bear half of the load and the 0.4mm layers basically holding nothing at all. Even the weight normalized values didn’t help the thicker layer samples and show the same image in the end. However, all of the values were at least 50% lower than the strength of the lying specimens, which shows that we do have quite some anisotropy in the material.

Layer adhesion samples: Failure load VS layer height

One of the papers I linked down below uses a, in my opinion, good comparison value for the layer thickness, because they take a look at the ratio between nozzle diameter to layer thickness. This enables us to generalize these results maybe a little better because not everyone uses a 0.4mm nozzle. Applying this on our test results means that starting at layer thicknesses more than half of nozzle diameter (ratio 2) we can expect that the strength of our parts will suffer. This means that for a strong print and 0.4mm nozzle, probably don’t go above 0.2mm layers. Interestingly, I would have thought that really thin layers would perform worse but that wasn’t the case. So even though we have seen, that the color of the material changed, the resin itself didn’t degenerate. But also, at least for the Prusament PLA, that means that the layers don’t stick better to each other when they are thinner. What do you think about these results?

Besides the layer height, there are many other parameters that will play a role in the strength of your 3D prints, especially for layer adhesion. Foremost there is the material itself and I have already tested many other materials where I also checked for layer adhesion, which can differ quite significantly. Then there is print temperature, which I have tested almost 2 years back, so check that video out if you are interested. Other significant parameters are probably printing speed, extrusion width and part cooling which I’ll probably test in future videos. If you have other ideas, let me know in the comments!