Polyetheretherketone is a chain of aromatic rings joined by ether and ketone links. Those rings are rigid and the links are thermally robust, which is why PEEK survives continuous service at temperatures that destroy every other polymer on this site, resists almost every chemical, and is one of the few plastics implanted permanently in the human body.
It is also the only material here where two people printing the same spool can end up with genuinely different materials, and the variable is not the filament.
Crystallinity is set on the machine, not at the factory
PEEK is semi-crystalline. As it cools through the range where chains can still move, they either fold into ordered crystalline regions or they do not, and which happens depends almost entirely on how slowly the part cools.
- Cooled fast — in a cool or unheated chamber — PEEK stays largely amorphous. The part comes out translucent amber-brown, and it is comparatively weak, more prone to solvent attack, and softer at temperature.
- Cooled slowly in a genuinely hot chamber, the chains crystallise. The part is opaque tan or beige, stiffer, stronger, far more chemically resistant, and holds its properties much closer to the temperatures the datasheet promises.
The colour tells you which one you got, which is a rare case of a printed part reporting its own internal structure. If your PEEK prints are brown and brittle, the chamber is the problem and no amount of nozzle tuning addresses it.
Post-print annealing can recover some crystallinity when the chamber could not, at the cost of further shrinkage. Anneal a test coupon and measure it before you anneal a part that has to fit.
The temperatures involved
Nozzle 380 to 440 °C, bed 130 to 160 °C, and a heated chamber that is not optional in any sense. That combination rules out essentially every machine discussed on this site: it needs an all-metal hotend rated for the temperature, a heater and thermistor that can reach it, and a chamber with active heating rather than a closed box.
Shrinkage of 1.0% to 2.0% is among the largest here, and — because it is crystallisation-driven — it is not a fixed figure you can compensate for once. It moves with the chamber temperature that also sets the crystallinity, so dimensional compensation and material properties are coupled in a way they are not for anything amorphous.
The stored heat deflection range of 140 to 160 °C is worth reading carefully. Heat deflection is a specific test at a specific load; PEEK's continuous service capability is considerably higher, and the two numbers answer different questions. If a part must hold a load at temperature, the deflection figure is the honest one to design against.
What it costs to press "print"
Sixty grams. That is a component you could hold between two fingers, and printing it over 8 hours on a machine whose hotend, bed and chamber together draw 1000 W comes to $66.75.
Where the money goes: $36.00 into filament at $600 per kilogram, $16.00 into machine time at $2.00 an hour, $13.35 into a 20% failure allowance, and $1.40 into electricity for 8 kWh at the US average rate.
Two things are worth noticing. First, 8 kWh is more than many households use in half a day, and it is still almost negligible against the material. Second, the money set aside for failures is nearly as large as the machine time — because PEEK prints fail, and at this price a failed print is expensive in a way a failed PLA print never is. At $600 a kilogram the usual conclusion of this site inverts: here it is worth spending hours and kilowatt-hours to protect the plastic rather than the other way round. Enter a longer, slower, safer print into the print cost calculator and the bill still shrinks, provided the failure percentage shrinks with it.
What PEEK is genuinely bad at
- Being printed at all on a normal machine. A hotend that stops at 300 °C is not close, and a passively enclosed printer is not a heated chamber.
- Cost per gram. At $400 to $900 per kilogram it is the most expensive material on this site, and prototyping in it is rarely justified when a geometry check in PLA costs pennies.
- Dimensional repeatability, because shrinkage and crystallinity both track the thermal history of the individual print.
- Bed adhesion without purpose-made surfaces. It is chemically inert, which is exactly what makes it hard to stick down.
- Ultraviolet, where it discolours, although its mechanical properties hold up better than most.
- Being justified by enthusiasm. Almost every hobby application people propose for PEEK is served by nylon, polycarbonate or PPA at a fraction of the cost.
Where nothing else will do
Chemical-plant components in contact with aggressive process fluids. Bearings, bushings and seals running hot and dry, where PEEK's low friction and wear resistance are exceptional. Semiconductor handling parts that must be repeatedly cleaned in solvents. Aerospace and motorsport components under sustained heat. Medical instruments and implants, because it can be steam-sterilised over and over and is biocompatible in implant grades.
The honest test before buying a spool is a single question: does the part have to work above roughly 150 °C, or in a chemical environment that would dissolve a polyamide? If not, PPA or polycarbonate will do the job for a small fraction of the money and on a machine you might actually own.