Machining PEEK Plastic: Speeds, Feeds, Heat, and Stress Relief

PEEK (polyether ether ketone) is the plastic that gets specified when nothing else will do: continuous service temperatures around 250 °C, chemical resistance that shrugs off most solvents, and a strength-to-weight ratio that puts it alongside some metals. It also machines differently from every other engineering plastic, and the difference is not subtle. The free-cutting habits that work on Delrin or nylon will smear the surface, weld chips to the flute, and leave a part that relaxes out of tolerance the moment it leaves the vice. This article lays out the cutting parameters, heat management, and stress-control steps used when a print calls for machined PEEK, written as working ranges rather than a single magic number.

Why PEEK behaves differently at the tool

PEEK is a semicrystalline thermoplastic with a glass transition around 143 °C and a melt point near 343 °C. Those two numbers decide almost everything about how it machines. Below the glass transition the material is stiff and cuts cleanly; once the cut zone drifts above roughly 140 °C the polymer softens locally, the chip stops shearing and starts smearing, and the freshly cut surface goes glossy and torn. That softening is the biggest difference from amorphous plastics, which tend to melt rather than shear. A machinist coming from acrylic or ABS will read the smell of burning as the danger signal, but with PEEK the damage happens well before smoke — by the time you see discoloration you have already run the cut zone past its useful window.

The semicrystalline structure also means PEEK has memory. It stores strain energy from the cut, and that stored energy is released slowly as the part sits on the bench. A flat milled face can pull into a gentle dish over a day or two, and a thin-walled sleeve can drift off round by a few hundredths of a millimeter with no load at all. You cannot see this happening on the machine, which is why PEEK parts are often measured twice — once fresh off the spindle and once after a settle period. The practice that saves the most scrap is simple: treat the part as something still moving after the last pass finishes, not as a rigid block you locked to size. Plan the tolerance for the relaxed state, not the just-machined state.

Moisture and fillers change the picture further. Natural PEEK is mildly hygroscopic and machines best after the stock has been dried, because surface moisture turns to steam at the edge and pits a finish you thought you had under control. Glass- and carbon-filled grades cut with more abrasive load and wear edges faster, but they also relax less after machining because the filler stiffens the matrix. None of these behaviors is a reason to avoid the material; they are just variables you set the process around instead of discovering halfway through a batch.

Speeds and feeds that actually hold

Typical working ranges for unfilled PEEK on a conventional three- or four-axis mill run a surface speed of about 150–350 m/min on carbide, with finish passes often near the upper end of that band. Feeds land around 0.02–0.08 mm per tooth depending on tool diameter and rigidity, and depths of cut stay shallow on finishing — 0.1–0.4 mm axial is common because the goal is to keep the cut cool and let the chip carry heat away. On a lathe, turning speeds of 100–250 m/min with a similar light feed read well for most diameters. These are starting points, not fixed values; the right number is the one that keeps the cut zone below the glass transition and produces a continuous, curly chip instead of a pasty ribbon.

Tool geometry matters as much as the numbers. A sharp, positive-rake carbide end mill with generous clearance and polished flutes sheds PEEK instead of dragging it. Dull edges generate heat exactly where you do not want it, and a rubbed edge on PEEK smears faster than it cuts. For filled grades the wear climbs and you trade some finish speed for a tougher edge, but the heat discipline stays the same. What kills a first article is rarely the spreadsheet of numbers — it is a tool that went dull, a feed that got timid, or a coolant line that stopped hitting the flute. The table below shows the ranges reached for first; every job gets a short prove-out cut before a batch runs.

Spindle speed follows directly from surface speed and tool diameter, and on small tools that puts you at high rpm with a light chip load — exactly the condition where the cut stays cool if the edge is sharp. The mistake to watch is lowering speed to "be safe" while the feed stays light; that combination rubs instead of shears and is the fastest route to a smeared wall. Keep the speed up and the feed meaningful, and let the shallow depth of cut do the heat control.

Heat is the enemy — keeping the cut zone cool

The quickest way to ruin a PEEK part is to let the tool do work that turns into heat instead of chips. Heat builds at the cutting edge from two sources: rubbing where the edge is not sharp, and plastic deformation of the chip as it curls. Both rise with dulling and with too light a feed, which is the counterintuitive part. A feed that is too small lets the edge burnish instead of shear, so the part heats and the surface smears — and operators often respond by slowing the spindle, which makes it worse. The fix is usually to feed harder within the tool's limit and keep the speed high enough that each tooth takes a real bite.

Cooling method is a choice, not a default. Flood coolant works and is the safe pick for deep pockets and long runs, because it carries heat out of the cut directly. But many shops run PEEK dry or with compressed air, and for good reason: some coolant formulations leave a residue that matters on parts that will be bonded or coated later, and a dry cut keeps the workholding simpler. Compressed air or mist aimed at the flute does most of the job for shallow work. What you must not do is let chips recut — a nest of hot PEEK swarf welding to the tool is how a clean job turns into a gouged one. Clear chips aggressively and keep air moving across the cut zone.

Tool path is part of the cooling strategy too. Long, uninterrupted engagements hold heat in one spot; stepping the engagement down with trochoidal or layered passes spreads the load and gives the chip a chance to evacuate. In a deep pocket this is the difference between a part that stays below the glass transition and one that goes glossy at the bottom of the wall. None of it requires exotic equipment — it is feed, speed, coolant aim, and path selection used together, with the glass transition as the line you refuse to cross at the edge.

Controlling residual stress and warpage

The warpage that bites PEEK users is rarely from the cut itself — it is from stress that was already in the blank. Extruded and molded stock carries orientation and cooling stress, and a near-net shape can sit for a long time before someone machines it. When you remove material unevenly, you unbalance those locked-in stresses and the part springs. The pattern is predictable: thin sections move more than thick ones, and material removed from one side only will bow that side open. Symmetric stock removal — taking the same depth on opposing faces in the same setup where you can — is the cheapest insurance against it.

The other source is heat soak. A part that warms during roughing and is finished while warm will relax as it cools, so the measured size on the machine is not the size the customer gets. A staged approach helps: rough, let the part return to ambient, then finish. For tight-tolerance work this settle step is not optional — it is the difference between a part that holds and a part that looked right at the spindle and drifted by morning. Fixturing also plays a role; clamping that pinches the part distorts it elastically, and that distortion springs back the instant the clamps come off, so light, even workholding paired with a good locating datum beats a hard vice grip every time.

"On a thin PEEK plate, flatness measured right off the machine is not flatness you can ship. Release the stock stress first, then machine, then let the part settle before the finish pass — the material is rarely the problem, the residual stress in the blank is."

Measurement discipline closes the loop. A PEEK part should be gauged after it has cooled and settled, not the minute the program ends, and for thin or wide parts a second check a day later is worth the bench time. This is not a quality-system requirement so much as physics: the part is still moving, and quoting a tolerance against the just-machined size simply hides the movement until it shows up as a return.

Annealing — before and after the cut

Annealing PEEK relaxes the internal stress in the stock before you cut, and it is the most reliable step for holding flatness and roundness on thin or wide parts. A typical stress-relief anneal for stock sits around 150 °C for a few hours, held just above the glass transition and well below the melt, then cooled slowly — the slow cool is the part people skip, and skipping it just locks a new stress pattern in. Blanks that will become large flat plates or thin rings get annealed before any machining, because the alternative is watching the finished part curl after the work is already done.

Post-machining anneal is a separate decision. It can relieve the strain introduced by cutting and stabilize the part, which helps if the component will see heat or load in service. But it also relaxes whatever tight tolerance you just held, so you anneal first and finish-machine second, or you accept that the anneal will move the size and plan the tolerance around it. For most prototyping and low-volume work the cleanest route is anneal the blank, machine to finished size with a settle between rough and finish, and skip the final anneal unless the application demands thermal stability. The grade matters here: unfilled PEEK relaxes more freely than carbon-filled, which is stiffer, moves less, and is harder on tools.

• Dry the stock first. Bake natural PEEK to remove surface moisture before the first pass, or expect pitted finishes from steam at the edge.
• Anneal wide and thin blanks. Plates and rings get a stress-relief anneal before machining, slow-cooled, not quenched.
• Remove material symmetrically. Take matching depth on opposing faces in one setup to keep locked-in stress balanced.
• Rough, settle, finish. Let the part return to ambient between roughing and finishing on any tight-tolerance feature.
• Keep the edge sharp. Replace tools before they rub; a dull edge smears PEEK faster than it cuts and drives heat up.
• Feed hard, cut shallow. Meaningful feed with light finish depth holds the cut below the glass transition better than a slow timid pass.
• Aim the coolant or air. Hit the flute, not the table. Clear chips so they cannot recut and weld to the tool.
• Measure after settling. Gauge PEEK after cool-down, and re-check thin parts the next day before you ship.

Tooling, chips, and getting a clean surface

A clean PEEK surface comes from sharp tools and a chip that leaves, not from a slow careful scrape. High-helix, polished-flute carbide with a positive rake throws the chip up and out of the cut; a slow, dull, or zero-rake tool drags material across the wall and leaves the smeared, hazy finish that reads as a defect even when the size is right. Speeds on the higher side of the range keep the shear zone crisp, and a light finish depth keeps the heat per pass low. Surface finish in the low single-digit micrometre Ra is routine on a well-set PEEK job; pushing below about 0.4 µm Ra usually means a secondary process rather than hoping the cutter will do it.

Chip control is the unglamorous skill that separates a clean PEEK run from a messy one. The material is tough and stringy, so it wants to make long ribbons that wrap the tool and redeposit heat. Trochoidal or peck-style toolpaths that keep engagement and evacuation steady beat a long straight slot, especially in deep pockets. For filled grades expect faster edge wear and plan tool changes before the finish pass, because a slightly worn edge on carbon-filled PEEK smears fast. None of this is exotic — it is the same discipline as any heat-sensitive plastic, applied with the specific knowledge that PEEK's glass transition is the line you must not cross at the cut.

The payoff for respecting that line is a part that holds its size after it cools and a surface that looks machined rather than melted. PEEK rewards process control and punishes shortcuts, which is exactly why it earns its place in the applications where metal would be heavier or chemical resistance matters more than raw stiffness. Set the speeds and feeds for the grade, keep the cut cool, anneal the stock that needs it, and measure the relaxed part — and machined PEEK stops being difficult and becomes just another material you can hold to print.

If your next project calls for a PEEK component, send the drawing and we will walk the stock, anneal step, and tolerance sequence with you before the first article runs — see how our materials capability handles heat-sensitive plastics.