The Best Filament for 3D Printing

Two clear suggestions with complicated commentary

There are numerous videos reviewing 3D printer filaments to help you determine which is the best. I’ve spent way too much time watching these and running over data and can quickly summarize all the information relevant to most people doing 3D printing: Use PLA. There, you’re finished. If you want or need more information or are interested in running tests yourself read on.

There are two big components of the ‘strength’ of a material: stiffness and toughness. Stiffness refers to how hard it is to bend (or stretch/break) while toughness refers to how well it recovers from being bent (or stretched). These can be further broken down into subcategories, like whether the material successfully snaps back after getting bent or is permanently deformed. An important thing to understand is that the measures used aren’t deep ethereal properties of material, they’re benchmark numbers based on what happens if you run particular tests. This isn’t a problem with the tests, it’s an acknowledgement of how complex real materials are.

For the vast majority of 3D printing projects what you care about is stiffness rather than toughness. If your model is breaking then most of the time the solution is to engineer it to not bend that much in the first place. The disappointing thing is that PLA has very good stiffness, usually better even than the exotic filaments people like experimenting with. In principle proper annealing can get you better stiffness, but when doing that you wind up reinventing injection molding badly and it turns out the best material for that is also PLA. The supposedly better benchmarks of PLA+ are universally tradeoffs where they get better toughness in exchange for worse stiffness. PLA is brittle, so it shatters on failure, and mixing it with any random gunk tends to make that tradeoff, but it isn’t what you actually want.

(If you happen to have an application where your material bending or impact resistance is important you should consider TPU. The tradeoffs of different versions of that are complex and I don’t have any experience with it so can’t offer much detailed advice.)

Given all the above, plus PLA’s generally nontoxic and easy to print nature, it’s the go-to filament for the vast majority of 3D printing applications. But let’s say you need something ‘better’, or are trying to justify the ridiculous amounts of time you’ve spent researching this subject, what is there to use? The starting place is PLA’s weaknesses: It gets destroyed by sunlight, can’t handle exposure to many corrosive chemicals, and melts at such a low temperature that it can be destroyed in a hot car or a sauna. There are a lot of fancy filaments which do better on these benchmarks, but for the vast majority of things PLA isn’t quite good enough at PETG would fit the bill. The problem with PETG is that it isn’t very stiff. But in principle adding carbon fiber fixes this problem. So, does it?

There are two components of stiffness for 3d printing: Layer adhesion and bending modulus. Usually layer adhesion issues can be fixed by printing in the correct orientation, or sometimes printing in multiple pieces at appropriate orientations. One could argue that the answer ‘you can engineer around that’ is a cop-out but in this cases the effect is so extreme that it can’t be ignored. More on this below, but my test is of bending modulus.

Now that I’ve finished an overly long justification of why I’m doing bending modulus tests we can get to the tests themselves. You can get the models I used for the tests over here. The basic idea is to make a long thin bar in the material to be tested, hang a weight from the middle, and see how much it bends. Here are the results:

CarbonX PETG-CF is a great stronger material, especially if you want/need something light weight spindly. It’s considerably more expensive than PLA but cheaper and easier to print than fancier materials and compared to PLA and especially PETG the effective cost is much less because you need less of it. The Flashforge PETG-CF (which is my stand-in ‘generic’ PETG-CF as it’s what turns up in an Amazon search) is a great solution if you want something with about the same price and characteristics as PLA but better able to handle high temperatures and sunlight. It’s so close to PLA that I’m suspicious that it’s actually just a mislabeled roll of PLA but I haven’t tested that. I don’t know why the Bambu PETG-CF performed so badly. It’s possibly it got damaged by moisture between when I got it and tested it but I tried drying it thoroughly and that didn’t help.

Clearly not all carbon fiber filaments are the same and more thorough testing should be done with a setup less janky than mine. If anybody wants to use my models as a starting point for that please go ahead.

The big caveat here is that you can engineer around a bad bending modulus. The force needed to bend a beam goes up with the cube of its width, so unless something has very confined dimensions you can make it much stronger by making it chunkier. You can do it without using all that much more material by making I-beam like structures. Note that when 3D printing you can make enclosed areas no problem so the equivalent of an I-beam should have a square, triangular, or circular cross section with a hollow middle. The angle of printing is also of course very important.

The conclusion is that if you want something more robust than PLA you can use generic PETG engineered to be very chunky, or PETG-CF with appropriate tradeoffs between price and strength for your application.

A safety warning: Be careful to ventilate your space thoroughly when printing carbon fiber filaments, and don’t shred or machine them after printing. Carbon fiber has the same ill effects on lungs as asbestos so you don’t want to be breathing it in. In my tests the amount of volatile organic compounds produced are small, but it’s a good idea to be careful.