PEEK in 3D Printing: A Durable and Demanding Material

• 1
  Key takeaways
• 2
  What PEEK actually is, in numbers
• 3
  Why PEEK in 3D printing is hard: melt, flow, then freeze
• 4
  Direction matters more than the datasheet
• 5
  Where printed PEEK pays off, and where it does not
• 6
  How to choose between printing and machining PEEK
• 7
  Running PEEK on a high-temperature printer
• 8
  Printed PEEK vs machined PEEK: pick by requirement

Polyetheretherketone is a semi-crystalline thermoplastic built from aromatic rings linked by ether and ketone groups. That backbone is stiff and thermally stable, which is why the numbers look unusual for a plastic. Glass transition sits at 143 °C and the crystalline melting point at 343 °C. A part can run continuously near 250 °C and still hold most of its room-temperature stiffness.

The aromatic rings also resist attack. PEEK shrugs off most fuels, hydraulic fluids, weak acids and steam, which is why it shows up in pump internals, valve seats and aircraft brackets. Unfilled PEEK has a tensile strength around 90–100 MPa and a modulus near 3.6 GPa. Add 30% carbon fiber and the modulus roughly triples while wear resistance improves.

For an engineer, the useful way to think about PEEK is as a metal substitute that is 4–5 times lighter than steel. It does not conduct heat or current. It does not gall. Those three properties, not raw strength, are usually what drives the material choice.

The catch is that all of these properties assume a fully dense, fully crystallized part. FDM printing gives you neither by default. It gives you a stack of bonded beads with whatever crystallinity the cooling history allowed.

PEEK is printed almost exclusively by fused deposition modeling. The filament is pushed through a nozzle held between 380 °C and 430 °C. At that temperature the melt viscosity is still high, roughly an order of magnitude above PLA or ABS at their own print temperatures. High viscosity means the polymer does not flow into corners the way a low-viscosity melt does, so toolpaths need wider extrusion widths and slower speeds.

The second requirement is chamber temperature. If the deposited bead cools below 143 °C before the next layer lands, it cannot co-crystallize with the layer above. The result is a weak, glossy interface that splits under load. High-temperature printers therefore hold the build chamber at 200 °C or higher, often with a heated bed above 120 °C. That is the single feature that separates a PEEK-capable machine from a general-purpose one.

The third factor is shrinkage during crystallization. Semi-crystalline polymers contract measurably as the disordered melt orders into crystals. PEEK does this over a temperature range that overlaps with the printing window. Large flat parts and thick sections therefore curl and delaminate unless the chamber is uniform and the geometry is managed.

None of this is a defect in the material. It is simply the price of a polymer that stays useful at 250 °C.

• 1
  Nozzle380–430 °C, hardened or high-temp hotend, 0.4–0.6 mm typical
• 2
  Chamber200 °C or above for layer bonding and crystallinity
• 3
  Bed120–160 °C, often on a PEI or PEEK build plate
• 4
  Post-processControlled anneal, then slow cool to avoid stress cracking

A printed PEEK part is not isotropic. In the XY plane, where the extruded bead runs continuously, tensile strength can reach 80–90% of the injection-molded value. Across the layers, in Z, it often falls to 40–60%. The gap widens if the chamber ran cool or the layer time was too long.

That ratio drives design decisions more than any single material number. A bracket loaded in shear across its layers will delaminate long before the same bracket loaded in-plane. If the load path must cross layers, thicken the section, raise the extrusion temperature slightly, and keep the layer time short so each bead lands on a still-warm surface.

Porosity is the second hidden variable. PEEK outgasses and picks up moisture, and any water in the filament flashes to steam in the nozzle. The result is microvoids along the bead, which act as crack starters in fatigue. Drying the filament before a run is not optional. Most PEEK grades should sit in a 120–150 °C oven for several hours before printing.

A printed part can be strong. It is rarely strong in every direction.

Printed PEEK earns its cost in low-volume, high-value parts where the geometry is complex and the environment is hostile. A ducting elbow with internal channels, a custom impeller, a surgical guide, a test fixture that runs inside an autoclave. These are parts where machining would need many setups or would not reach the internal features at all.

It also wins during development. A printed PEEK manifold can be on a test rig within days, letting a team validate flow and fit before committing to tooling. The printed part may not be the production part, but it answers the same engineering questions.

Printed PEEK does not win on tight tolerances. FDM holds roughly ±0.2 mm on a good day, and warping pushes that wider on tall or flat geometry. It does not win on void-free walls either, and it does not win when the part sees cyclic load across layers.

In those cases the answer is usually machined PEEK stock. Compression-molded or extruded rod and plate are fully dense and fully crystallized. Milling or turning that stock holds ±0.005 mm, produces no layer interface, and can be inspected like any metal part.

Start with the load direction. If the highest stress crosses layer planes, machining is the safer default. If the load stays in the plane of the beads, printing stays in play.

Then check the tolerance callouts. Anything tighter than ±0.1 mm on a mating feature points to machining, because printed PEEK will need post-machining anyway and that defeats the purpose of printing it.

Then look at the internal geometry. Printing is the only practical route when a part needs a curved internal passage, a lattice, or a cavity a cutter cannot reach. Machining wins when the part is a turned bushing, a flat seal face, or a manifold that can be drilled from three sides.

Finally, count the parts. Printing makes sense for a handful of units or a design that is still moving. Above a few hundred identical pieces, the setup cost of machining spreads out and the per-part economics flip.