Application of 3D PEEK in skull defects: what goes wrong on the machine

• 1
  Symptom → likely cause → how to fix it
• 2
  Why 3D PEEK in skull defects behaves differently from metal
• 3
  What makes a cranial PEEK implant machinable
• 4
  Cutting parameters that hold the tolerance
• 5
  Blank selection and the paperwork behind it
• 6
  Cases where machined PEEK is the wrong answer
• 7
  Step by step: from blank to seated implant

PEEK is a semi-crystalline polymer. It is stiff for a plastic, but its thermal conductivity is roughly 0.25 W/m·K, which is far below aluminium. Heat from the cutting edge has nowhere to go except into the chip and the workpiece. On a cranial implant with a 0.8 mm wall, that heat softens the surface layer and the tool starts to rub rather than shear.

The second difference is anisotropy. A printed or extruded blank has a grain direction. When you machine a curved implant that spans the temporal region, the load direction changes across the part. A tool path that works on a flat plate can pull the thin rim out of shape on a dome.

The third difference is inspection. A metal implant can be checked with a touch probe and a micrometer. PEEK at 0.8 mm deflects under light contact, so a probe force of 0.5 N can read low. In practice, non-contact scanning plus a light-touch check on the seating rim gives a truer picture.

None of this makes PEEK a bad material for cranial repair. It makes the machining window narrow. Feed, speed, depth of cut, and workholding all have to be chosen together, and the blank has to be stable before the first cut.

• 1
  Low thermal conductivityHeat stays in the cut. Use coolant and conservative depths.
• 2
  Directional strengthThe blank grain and the load path must be aligned.
• 3
  Soft under probingLight touch or optical scanning for final checks.

A machinable design has a wall thick enough to survive the cutting force and a seating rim wide enough to carry the load. In our experience, walls below 0.8 mm become a finishing problem. They can be cut, but the process window shrinks to a few hundred rpm and one wrong stepover will chip the edge.

Curvature matters as much as thickness. A dome with a continuous surface and gradual transitions lets a ball end mill follow the shape without sudden direction changes. A design with a sharp step next to a thin rim forces the tool to change engagement mid-cut. That is where chatter starts.

Screw holes and fixation slots should sit on thicker bosses, not on the thinnest section of the shell. Moving a hole 1 mm inward from a thin edge often removes the break-through risk without changing the surgeon's plan. This is a design change, not a machining trick, so it belongs in the DFM review.

We run a DFM analysis on every uploaded file and return it within 12 hours. The output is specific: minimum wall, tool reach, and any feature that will need a second setup. That review is free and it happens before the quote is finalized.

• 1
  Wall thicknessKeep the shell at 0.8 mm or above where possible.
• 2
  Surface continuityGradual curvature beats sharp steps next to thin rims.
• 3
  Hole placementPut fixation holes on bosses, not on the thinnest shell.

For PEEK cranial shells we use sharp carbide tools with 2 or 3 flutes, polished flutes, and a helix angle around 30°. A 6 mm flat end mill for roughing and a 3 mm ball end mill for finishing covers most dome geometry. Runout should stay under 0.01 mm. Above that, one flute does most of the cutting and the finish turns cloudy.

Roughing removes material in 0.3 mm radial and 0.5 mm axial steps. Spindle speed stays near 8,000 rpm with a feed per tooth of 0.05–0.1 mm. Deeper cuts are possible on thick sections, but on a 1 mm shell they generate enough heat to move the surface. Flood coolant is the default, not an option.

Finishing uses a smaller stepover, 0.1–0.2 mm, to keep the scallop height low. The target surface is Ra 0.8–1.6 μm. If the finish comes out above Ra 3.2 μm, the usual cause is a dull tool or a feed that is too low for the rpm. Raising the feed per tooth and replacing the cutter fixes it more often than slowing the spindle.

Tolerance is held at ±0.005 mm on critical seating surfaces. That number is realistic on a rigid setup with a stable blank. It is not realistic on a thin shell held only by the vise. Vacuum fixturing or a custom soft jaw that supports the full underside is what makes the number repeatable.

• 1
  Roughing0.3 mm radial, 0.5 mm axial, flood coolant.
• 2
  Finishing0.1–0.2 mm stepover, ball end mill, sharp edges.
• 3
  Tolerance±0.005 mm on seating surfaces with full support.

Implant-grade PEEK is supplied as rod, plate, or near-net printed blanks. Machining from plate gives predictable properties and a clear material certificate. Printed blanks allow a shape closer to the final dome, which cuts cycle time, but the layer orientation has to be considered in the tool path. Both routes are valid. The choice depends on the geometry and the documentation the hospital needs.

For medical work we machine under ISO 13485:2016. Our shop also holds ISO 9001:2015, IATF 16949:2016, and ISO 27001:2022. The ISO 27001 certificate covers how patient-linked files and CAD data are stored and transferred. Uploads are handled as confidential, and an NDA is available on request.

Traceability runs from the material lot to the finished part. Raw material is checked on receipt, dimensions are monitored during the run, and every part gets a final inspection before shipment. Reports are available on request. That is the same flow we use for other medical device components, and it is what makes a cranial implant order auditable.

We do not claim to hold a specific implant clearance or a regulatory approval for any finished device. We machine parts to the customer's design and provide the dimensional and material records that support their own submission.

• 1
  Plate vs printed blankPlate is predictable; printed saves cycle time on domes.
• 2
  CertificationsISO 13485, ISO 9001, IATF 16949, ISO 27001.
• 3
  TraceabilityMaterial lot to finished part, reports on request.

PEEK is not the right material when the defect is large and the shell would need to span an unsupported opening with almost no curvature. A very flat, very thin panel has low bending stiffness no matter how well it is machined. In that geometry, a different material or a different fixation scheme should be discussed with the clinical team.

It is also a poor fit when the schedule leaves no room for a stable blank. If the material arrives warped and there is no time to stress-relieve it, the part will move after machining. We would rather flag that early than ship a shell that does not seat.

Finally, PEEK is not the answer when the design has features smaller than the tool can reach. A slot 0.5 mm wide and 5 mm deep cannot be cut cleanly. In that case the feature should be redesigned, or the part should be built additively and finished only on the critical surfaces.

The honest position is this: PEEK solves a lot of cranial repair problems, but it is not universal. The DFM review exists to say so before metal or polymer chips are made.

• 1
  Large flat defectsLow stiffness regardless of machining quality.
• 2
  No time to relieve stressThe blank will move after the last cut.
• 3
  Sub-millimeter deep featuresRedesign or switch to additive, then finish.