Why Your 3D Print Deforms In A Hot Car: Glass Transition Temperature Explained

A 3D printed phone holder left on a car dashboard can come back soft, bent, or completely out of shape by the afternoon. The part never went anywhere near its melting temperature. What it reached was its glass transition temperature, and for PLA that number is low enough to be a real problem in this climate.

This article explains what glass transition temperature is, why it matters more than melting point for almost everything we print, and how to read the heat numbers on a filament spec sheet without drawing the wrong conclusion from them. The last part matters more than it sounds, because the way these numbers are usually presented in filament comparison tables makes at least one common material look far worse than it actually is.

How hot a parked car actually gets

Most articles on this topic quote one big number and leave it there, which mixes up two different things. Air temperature inside the cabin and surface temperature of the parts sitting in direct sunlight are not the same, and the gap between them is large.

A study from Arizona State University parked cars on days in the region of 38°C and measured both. After one hour in the sun, cabin air averaged 47°C while the dashboard surface averaged 69°C. The steering wheel sat at 53°C and the seats at 51°C. The air in the car was hot. The dashboard was more than 20°C hotter than the air above it.

Closer to home, a Malaysian study on cabin thermal comfort reported cabin temperature climbing to around 60 percent above ambient within two hours in the sun. That is an awkward way to state a temperature, because a percentage of a Celsius reading depends on where you decide zero sits, so it does not convert into a usable figure. What it does confirm is that the effect is just as strong in our climate. The Arizona measurements are the ones with absolute numbers attached, so those are what I would design against.

Two things follow from this. Cabin air alone is already close to the limit for PLA. And a part clipped to a dashboard or stuck to a windscreen is not sitting in cabin air, it is absorbing sunlight directly, which is why dashboard surfaces run so much hotter. A black printed part in that position is the worst case, because dark surfaces absorb more of that radiation and heat up further.

What glass transition temperature means

Glass transition temperature, usually written as Tg, is the temperature range where a polymer changes from a hard, rigid state into a softer and more flexible one.

Below Tg, the long polymer chains that make up the plastic are effectively frozen in place. They can vibrate, but they cannot slide past each other. That is what makes the part feel stiff. As the material warms towards Tg, those chains gain enough mobility to start moving relative to one another, and the material loses a large part of its stiffness.

Nothing melts. PLA melts somewhere above 170°C, but it starts going soft in the 55 to 60°C range. That is a gap of more than a hundred degrees between “loses its shape” and “turns liquid”, and it is the gap that catches people out.

It is also worth saying that Tg is a range, not a switch. Materials are usually quoted as a single number because a single number fits in a table, but the softening happens gradually across several degrees on either side of it.

Whether the part actually deforms once it gets there depends on what else is happening to it. An unloaded decorative piece sitting on a shelf at 60°C may hold its shape indefinitely. A phone holder at 60°C is carrying a phone, and being pressed and pulled every time someone reaches for it, so it will sag long before an unloaded part would.

Why the same number means different things for different plastics

This is the part that most filament comparison tables skip, and it changes how the whole table should be read.

Polymers fall into two groups. Amorphous polymers have their chains arranged randomly, like cooked noodles in a bowl. Semi-crystalline polymers have regions where the chains fold into ordered, tightly packed structures, with disordered regions in between.

For an amorphous polymer, Tg is the practical ceiling. There is no ordered structure holding the part together once the chains start moving, so the material turns rubbery rather than falling apart, but the stiffness drop is large enough that a load-bearing part stops doing its job. PLA as printed, ABS, ASA and polycarbonate all behave this way.

For a semi-crystalline polymer, Tg only softens the disordered regions. The crystalline regions stay locked until you get close to the actual melting point, and they carry the load in the meantime. A semi-crystalline part above its Tg is less stiff than it was, but it is still a working part. Nylon, PEEK, PP and PET all fall into this group.

PLA is the interesting case, because PLA is capable of crystallising but barely does so at normal print settings. It cools too fast on the print bed. So a PLA print behaves like an amorphous material even though the raw polymer is not strictly one, which is exactly why Tg is such a hard limit for it. There is a way to change that, covered further down.

Glass transition temperatures of common filaments

Actual values depend on the specific formulation. Additives, fillers and blends move these numbers, so treat them as typical rather than as a specification for any particular spool.

NatureWorks quotes 55 to 60°C for its Ingeo PLA grades. Eastman lists values in the low 70s to around 80°C across its copolyester grades, which is where PETG sits. For polycarbonate, standard Makrolon from Covestro is around 145°C, but almost nobody is printing that. The PC filaments sold to makers are modified to print at reachable temperatures, and Polymaker lists 113°C for PolyMax PC, so the low end of that range is the number to plan around.

Nylon looks like the worst material in that table, and it is not

If you read the table quickly, PA6 at 60 to 70°C looks barely better than PLA, and worse than PETG. Plenty of filament comparison charts list it at “around 50°C” and leave it there, which puts it dead last. That conclusion is wrong, for two separate reasons.

The first is crystallinity. Nylon is semi-crystalline, so the section above applies. It keeps useful mechanical properties above its Tg, which is why nylon shows up in gears, bushings and under-bonnet automotive parts. It does lose stiffness on the way there, it just does not give up the way an amorphous part does. Judging nylon by its Tg alone gives you the wrong answer.

The second reason matters more in Malaysia. Nylon absorbs water from the air, and water acts as a plasticiser inside the polymer. NETZSCH measured PA6 with 1.2 percent moisture content at a Tg of 40.4°C, and reported more than a 70°C difference between a fully dry sample and one holding 4.9 percent water. That is an enormous swing, and it happens on the shelf, not in the printer.

So the practical rule for nylon here is that dry storage is not optional. A spool left open in Malaysian humidity is a different material from the one on the datasheet, both in how it prints and in how the finished part behaves. Moisture management matters more for us than it does for makers in drier countries.

Tg is not a maximum operating temperature

A material with a Tg of 80°C does not stay perfectly rigid at 79°C and collapse at 80°C. The stiffness comes off gradually, and how much of it a real part can spare depends on the load it carries, wall thickness, infill, print orientation, how long it is exposed, and the specific formulation.

This is why manufacturers publish other numbers alongside Tg. Heat Deflection Temperature, or HDT, is the most useful one for engineering parts, because it measures the temperature at which a standard test bar deflects by a set amount under a specified load. Vicat softening temperature does something similar using a needle pressed into the surface.

There is a catch with HDT that trips people up. It is measured at one of two standard loads, usually 0.45 MPa or 1.8 MPa, and the same material can differ by 10 to 20°C between the two. An HDT figure quoted without its test load is not comparable to anything. If you are checking a spec sheet, look for the load condition next to the number before you use it.

None of these numbers describe your specific part either. They describe injection moulded test bars. A printed part has layer lines, which are weak points, and its behaviour depends heavily on which direction the load runs relative to those layers. Treat all of them as indicators for narrowing down a shortlist, not as a pass or fail line.

What to use for a part that lives in a car

For anything decorative that stays indoors, PLA is fine. A part on a desk or a shelf is not going to approach 55°C, and PLA prints easily and looks good. The indoor exceptions worth knowing are a spot in direct sunlight through a window, a closed store room, and anything sitting in a roof space, all of which run much hotter than the room they belong to.

For a part going into a car, PLA is the wrong choice, and this is not a marginal call. Cabin air alone gets close to its Tg on an ordinary afternoon, and any part in direct sunlight goes well past it.

PETG gives real margin over PLA and handles cabin air comfortably, so it is a reasonable choice for something in a door pocket, under a seat, or in the boot. For a mount stuck to the windscreen or sitting on the dashboard in direct sun, I would not use it either. Dashboard surfaces measured at 69°C in a 38°C climate are already inside the PETG range.

ASA is what I would pick for a dashboard or windscreen part. Its Tg sits around 100 to 105°C, which is substantially more margin than PLA or PETG over the temperatures actually measured inside parked cars, and it resists UV far better than ABS. That second point matters as much as the first for a part sitting in permanent sunlight, because ABS in direct sun goes yellow and brittle over months regardless of how well it handles the heat.

ABS is the cheaper and more widely available alternative with similar heat performance, and it is a sound choice for parts that get hot but stay out of direct sunlight.

Polycarbonate is a step above both on heat, but printable PC blends give up some of the heat resistance that makes standard PC attractive, and they are demanding to print well. It is worth reaching for when a part needs both heat resistance and impact strength, not for heat alone.

PEEK exists for the extreme end and needs a printer most people do not have. For a car interior part, it solves a problem you do not have.

Getting more heat resistance out of PLA

Since PLA prints almost amorphous, forcing it to crystallise raises its heat resistance substantially. That is what annealing does. You hold the printed part at a temperature above its Tg but below its melting point, typically around 100°C, and the chains have enough mobility to fold into crystalline regions before you cool it back down.

The structural change is measurable. A 2026 study on eSUN PLA+ printed on an Ender-3 found as-printed parts sitting at 8.6 percent crystallinity, rising to 41.8 percent after annealing at 90°C for 80 minutes. That jump is what buys the heat resistance.

The same study is worth reading for a second finding. The strongest parts were not the most crystalline ones. Annealing at 70°C for 60 minutes gave the best tensile strength at 47.0 MPa against a 39.75 MPa un-annealed baseline, while the hotter cycles produced more crystallinity and weaker parts. Hotter and longer is not automatically better.

The cost is dimensional accuracy. In CNC Kitchen’s testing, Formfutura Premium PLA annealed at 100°C for 45 minutes in a preheated convection oven showed dimensional changes of up to 10 percent, with parts contracting in the horizontal plane and expanding vertically. A 50mm feature came out at 45mm in one direction and 55mm in another. The distortion varied too much between parts to correct by simply scaling the model, and leaving parts attached to the bed with supports reduced the warping without solving it. That is one filament under one set of conditions, so treat 10 percent as a warning about the size of the effect rather than a number to plan around.

So annealing is worth doing for a bracket where fit is loose and heat resistance matters, and it is not worth doing for anything that has to mate with another part accurately.

The alternative is buying a PLA that has already been formulated to crystallise more readily. Several manufacturers sell high temperature or nucleated PLA grades that reach useful heat resistance after a much gentler annealing cycle, or in some cases straight off the printer. If you like printing PLA and only occasionally need heat resistance, that is the less painful route.

What this means when you pick a filament

The useful shift is to stop asking which filament is strongest and start asking what environment the part will actually live in. Strength at room temperature tells you nothing about what happens on a dashboard at two in the afternoon.

Check the hottest condition the part will realistically see, including direct sunlight and not just air temperature. Check whether the part will be carrying a load while it is hot. Then pick a material with margin above that, using Tg to shortlist and HDT at a stated load to confirm, and remembering that a semi-crystalline material like nylon has more headroom above its Tg than the number alone suggests.

Get that right and the part lasts for years. Get it wrong and it becomes a bent piece of plastic after one afternoon in a car park.

Sources

• ASU study on parked car temperatures
• Assessment of Thermal Comfort in a Car Cabin Under Sun Radiation Exposure
• NETZSCH, Dry. Humid. Wet. Polyamide and Water
• Glass transition temperature versus structure of polyamide 6, a flash DSC study
• CNC Kitchen, Better performing 3D prints with annealing
• Post-Print Annealing of FDM-Printed Polylactic Acid, Polymers 2026
• Polyamide nylon, dry versus conditioned data explained