Machining
Annealing acrylic: when it is necessary, and what it does not fix
Stress locked into an acrylic part does not show up in inspection. It shows up in service, as a crack from a hole or a white haze along a bond.
Machining
Stress locked into an acrylic part does not show up in inspection. It shows up in service, as a crack from a hole or a white haze along a bond.
Every operation on acrylic adds stress: machining removes material unevenly, drilling concentrates it, bending locks it into the bend, solvent bonding releases it. The part looks fine, it measures correctly, and it fails three weeks later in a way that looks like a material defect.
Annealing is the controlled heating and slow cooling of the part so that stress can relax. It is a standard step in optical and medical acrylic work and it is often skipped in general fabrication, which is exactly where the failures come from.
| Symptom | Where it appears | Likely cause |
|---|---|---|
| Crack starting at a hole | Fixing points | Drilling stress plus a tight fixing |
| White haze along a bond line | Solvent joints | Solvent releasing sheet stress |
| Fine crack network (crazing) | Near holes, bends, machined edges | Stress plus a solvent or a cleaner |
| Panel bows after machining | Flat parts with pockets or slots | Unbalanced material removal |
| Cracks along a bend | Heat-bent parts | Bend stress not relieved |
| Shatters rather than cracks on impact | Thin parts and edges | High internal stress |
The principle is slow in, hold, slow out. Standard cast acrylic anneals at around 80 degrees Celsius for thick sections and slightly lower for thin ones, held long enough for the whole part to reach temperature, roughly one hour per 3 mm of thickness, then cooled at no more than about 10 degrees per hour.
| Thickness | Soak temperature | Hold time | Cooling |
|---|---|---|---|
| Up to 6 mm | 75–80 °C | 1–2 hours | 10 °C per hour or slower |
| 6–12 mm | 80 °C | 2–4 hours | 10 °C per hour or slower |
| 12–25 mm | 80 °C | 4–8 hours | 8 °C per hour or slower |
| Bonded assemblies | 70 °C, lower to protect the joint | 4 hours or more | Very slow, in the oven |
These are working figures, not a specification. The correct cycle depends on the sheet grade and the geometry, and a thick or complex part needs a slower ramp. Above the glass transition temperature acrylic will distort under its own weight, so the part must be supported flat during the cycle.
A flat panel cut from cast sheet, with a routed outline and no machined faces, has very little added stress. A simple display riser bonded on one face is unlikely to need annealing. A part that lives indoors, is never cleaned with a solvent and carries no load is not worth the extra process and the extra days.
The judgement is: what will the part be exposed to, and what happens if it cracks? If the answer is solvent, temperature cycling or a safety function, anneal it.
The correct order is machine, anneal, then finish. Polishing before annealing means the polished surface may dull or distort in the oven, and the part will need polishing again. Bonding after annealing means the bond introduces new stress, so a bonded and machined assembly is annealed after bonding and before final polishing.
ANNEAL AFTER MACHINING: 80 °C, 4 h, cool at 10 °C/h or slower on a machined part.ANNEAL AFTER BONDING BEFORE FINAL POLISH on an assembly.STRESS-RELIEVED SHEET REQUIRED if the starting material matters, for example for an optical part.Without one of those notes, a shop will do what is standard for its own workflow, which may be nothing. Annealing adds days to a schedule, so it should be agreed at quotation rather than discovered at inspection.
Not reliably. A domestic oven cycles around its set point by 10 to 20 degrees, which is enough to distort a part or overshoot the softening range. Annealing needs a controlled oven with a slow ramp and a slow cool.
No. It relaxes internal stress without changing the polymer properties. A properly annealed part has the same hardness and the same optical properties, with less tendency to craze.
Typically two to four days for a normal production batch, because the cooling ramp cannot be accelerated. Large or thick parts take longer. It is quoted as a separate step so it is visible.
Partly. The smell is residual volatiles at the cut edge; annealing drives some of them off, but it is not a deodorising process. Venting the parts before packing is more effective.
It can, but extruded sheet also relaxes its flatness in the process, so a flat panel may come out slightly wavy. Starting with cast sheet is more reliable than annealing extruded sheet.
Annealing is the step that separates a part that survives a year from one that crazes in a warehouse, and it is also the step that is most often left out of a purchase order because nobody wrote a number down. The numbers below are the schedule we run: air-circulation oven, whole part supported flat, slow ramp both ways.
| Thickness | Soak temperature | Hold at temperature | Ramp rate | Cool to 40 °C |
|---|---|---|---|---|
| up to 5 mm | 70 °C | 2 h | 15 °C/h in, 10 °C/h out | 4 h |
| 6–10 mm | 75 °C | 3 h | 12 °C/h in, 8 °C/h out | 5 h |
| 12–20 mm | 80 °C | 4 h | 10 °C/h in, 6 °C/h out | 8 h |
| over 20 mm | 80 °C | 1 h per 5 mm | 8 °C/h in, 5 °C/h out | 12 h |
Three things in that table are load-bearing. The soak temperature must stay below the point where the sheet starts to sag under its own weight — 80 °C is the practical ceiling for cast sheet, and going to 90 °C to save an hour produces parts with a wave in them. The hold is what actually relaxes the material; a part taken to temperature and brought straight down has been warmed, not annealed. And the ramp down matters more than the ramp up, because fast cooling puts a new stress gradient in exactly where you removed the old one.
What annealing does not fix is worth stating in the same breath, because it prevents most of the disappointments. It does not remove stress introduced after it — drilling, routing, sawing or vapour attack after the anneal will put new stress in, so the sequence is machine, then anneal, then bond but do not machine again. It does not stop crazing caused by solvent exposure: a part annealed perfectly and then wiped with the wrong cleaner will still craze. And it will not straighten a part that has already bowed; a re-flatten requires a press or a vacuum bag and a separate cycle.
For a purchase specification, the sentence that works is: "Parts to be annealed after all machining and before bonding, at 80 °C for 4 h for thicknesses of 12–20 mm, ramp ≤ 10 °C/h, with a dimensional check after cooling." That is checkable, it names the sequence, and it gives both sides something to point at if a part fails. Without it, "annealed" means whatever the supplier decided on the day.
If it will see solvent, temperature cycling or a safety function, we will quote annealing as part of the process rather than as an afterthought.