Polycarbonate earned its reputation in riot shields, machine guards and safety glazing. The chains are rigid bisphenol units joined by carbonate links, which gives an amorphous polymer that is stiff, optically clear and glass-transitions at 144 °C — Covestro publishes 144 °C for Makrolon by ISO 11357 and Stratasys 142.5 °C for its FDM grade by ASTM D7426. The method matters more than it looks: the same polymer measures 161 °C by DMA, so a figure quoted without its method is a figure that can be nineteen degrees out. In sheet form it absorbs an impact that would shatter acrylic, and it does so at temperatures where most plastics have already gone soft.
A printed polycarbonate part is a different object, and the gap between the two is where most disappointment lives.
Why a printed PC part is not a bulletproof PC part
Polycarbonate is unusually notch sensitive. Its toughness comes from the material's ability to yield and draw locally under load, and a sharp stress concentration robs it of the chance — a notched impact figure for PC can be a small fraction of its unnotched one. That is a well-known moulding constraint, and it is why PC parts are designed with generous internal fillets.
A printed part is full of notches by construction. Every layer boundary is a plane of reduced bonding, every internal corner is a sharp re-entrant, and every void between beads is a crack waiting to be a crack. So a printed PC bracket does not behave like a moulded one, and testing it by hitting it is how people discover that.
Two consequences worth designing around:
- Fillet every internal corner you possibly can. On PC it changes the outcome, not just the appearance.
- Orient for load. PC's interlayer strength is well below its in-plane strength, and it is the direction of loading rather than the wall count that decides whether the part survives.
Water in the melt does permanent chemical damage
This is the failure that catches people who have printed nylon successfully and assume they have the moisture problem beaten. PC hydrolyses: the carbonate link is attacked by water at melt temperature and the chain is cut. Shorter chains mean a part that is cloudy where it should be clear, brittle where it should be tough, and weak in a way no reprint of the same spool will fix until the spool is dried.
Dry at 90 °C for 6 to 12 hours, which is above what a food dehydrator comfortably reaches and is a genuine equipment requirement rather than a nicety. A dry box while printing is not optional either, because a 300 °C melt gives absorbed water plenty of opportunity.
It warps harder than ABS, and for an additional reason
Shrinkage of 0.5% to 0.8% is in ABS's territory, but PC combines it with far higher stiffness and a much higher glass transition. Stiffness means the contracting part cannot relieve stress by deforming gently; it stores the stress and then releases it all at once. The result is the most aggressive corner lift of any common filament, capable of pulling a print off a hot bed with an audible crack.
Nozzle 270 to 310 °C, so an all-metal hotend is mandatory and any machine capped at 260 °C is out. Bed 100 to 120 °C, enclosure required, and a chamber that is genuinely warm rather than merely closed. PC-specific build sheets exist because plain PEI does not hold it.
What that costs to run
Heated bed and chamber together are a real electrical load, so this is one of the few pages where the electricity term is worth computing rather than waving at. A 200 g part running 10 hours at 350 W, on $60-per-kilogram filament, with machine wear at $0.50 an hour and a 15% failure rate, totals $20.72:
- Filament, $12.00
- Machine time, $5.00
- Failure allowance, $3.11
- Electricity, $0.61 — 3.5 kWh at the US average rate
Note the ordering. Even at 350 W over ten hours, the power bill is the smallest of the four terms and the money set aside for warped prints is five times larger, because polycarbonate fails at a rate that would be alarming in PLA. A quote that omits that line loses money roughly every seventh job. 350 W is a chamber-heated machine working hard and it is still the smallest line here, so if you want polycarbonate to cost less, the field to move in the print cost calculator is the failure rate — and the way to move it is an enclosure, not a cheaper tariff.
The solvent trap
Polycarbonate is vulnerable to environmental stress cracking: a part under internal stress — and a printed part is always under internal stress — can craze or crack on contact with solvents it would tolerate unstressed. Isopropyl alcohol, the default cleaning fluid on every printer bench, is a documented offender, and so are many oils, greases and adhesives.
The practical rule is to clean PC prints with warm soapy water and nothing else, and to test any adhesive on a scrap before trusting it with a part.
What polycarbonate is genuinely bad at
- Sunlight. It yellows and slowly embrittles under ultraviolet. Commercial PC glazing carries a co-extruded UV layer; filament does not.
- Any open-frame machine, unconditionally.
- Hotends limited to 260 °C, PTFE-lined hotends of any kind, and machines whose bed tops out below 100 °C.
- Solvent contact, per above.
- Cheap production. At $40 to $80 per kilogram plus a high failure rate, the material cost of a successful PC part is well above the sticker price.
- Beginners. It is difficulty five out of five for good reasons, and a first attempt usually teaches you about warping rather than about polycarbonate.
Where it is genuinely worth it
Parts that must hold shape near boiling water, light housings and lenses where its clarity survives heat that would fog other plastics, guards and covers that take repeated knocks, drone and RC structures, and fixtures that live near a heated bed or a soldering station. Heat deflection between 110 and 140 °C is the property you are buying, and almost nothing else on a desktop machine reaches it.
If the requirement is toughness rather than temperature, the blend is usually the better answer — PC-ABS gives up some heat resistance and most of the warping in exchange.