What dimensional tolerance can directly molded PTFE parts hold?

Directly molded and sintered PTFE parts should be treated as blank-grade or low-precision finished parts. Small, simple, thick-wall parts commonly hold about ±0.5% to ±1.5% of the nominal dimension, while larger or thin-wall parts typically need ±1% to ±3% or wider. A practical starting point is roughly ±1% for simple parts, widening to ±2% to ±3% for complex, thin-wall, or large parts. Critical dimensions are usually better guaranteed by post-machining rather than by molding alone.

How much does PTFE shrink after compression molding?

PTFE mold shrinkage is large compared with injection-molded plastics. Typical granular PTFE grades shrink from about 2.25% to 5.5%, with virgin grades often in the 2.25% to 4% range and modified grades around 4% to 5.5%. Shrinkage depends on resin particle size, bulk density, powder loading, preform pressure, sintering temperature, soak time, cooling rate, wall thickness, mold direction, filler system, and post-annealing. New molds are usually corrected from actual sintered and stabilized parts rather than from a single fixed shrinkage rate.

Why does temperature affect PTFE part dimensions so much?

PTFE has a high linear expansion coefficient, about 100 to 200 x 10^-6 per degree C. A 100 mm dimension can change 0.01 to 0.02 mm for every 1 degree C shift, and a 200 mm part across a 5 degree C swing can move 0.10 to 0.20 mm. PTFE also has a crystal transition near 19 degree C, where volume can swell 1% to 1.5%. For these reasons, tight tolerances on parts over 100 mm require inspection near 23 degree C with at least 24 hours of thermal stabilization.

What tolerance can machined PTFE parts achieve?

Machined PTFE parts are commonly held to about ±0.10 to ±0.15 mm for commercial work, ±0.05 to ±0.10 mm for finer work under 100 mm, and ±0.02 to ±0.05 mm for high-precision small local features with stress relief and agreed inspection. Tolerances of ±0.01 to ±0.02 mm are possible only on small features or short dimensions under special process and sample validation, and should not be used as a general tolerance for a whole part.

How does filled PTFE compare to pure PTFE for dimensional stability?

Filled PTFE is generally better than pure PTFE for dimensional stability. Fillers improve resistance to initial load deformation, stiffness, thermal conductivity, creep, and thermal dimensional stability, with creep and thermal stability often improving by roughly a factor of two. Glass-filled PTFE offers good rigidity and creep resistance but wears tools and can abrade mating faces. Carbon or graphite-filled grades improve wear and conduction, while bronze or copper-filled grades improve load and heat handling but need a chemical-compatibility check.

What inspection conditions are required for tight PTFE tolerances?

Tight PTFE tolerances only hold when the inspection is well defined. Measurement should be done at 23 plus or minus 2 degree C by default, or 23 plus or minus 1 degree C for precision parts, after the part has stabilized in that environment for at least 24 hours. Thin-wall or flexible parts need agreed fixtures or non-deforming supports, and measuring force should be limited because ordinary micrometers and hard fixtures can compress PTFE and read 0.13 to 0.25 mm small.

Why do PTFE parts sometimes fail measurement even when made correctly?

False rejects usually come from cold flow, creep, and excessive measuring force rather than from the molding itself. PTFE is soft and compresses under a micrometer or hard fixture, so inexperienced measurement can read undersize by 0.13 to 0.25 mm. Thin-wall rings may read oval when free yet function correctly when installed, and fixture distortion can spring back after unclamping. Defining the free versus assembled inspection state, limiting measuring force, and using soft-contact or optical gaging prevents most of these disputes.

PTFE Molded Part Tolerances: Realistic Dimensional Capability and How to Spec Them

If you’ve ever tried to hold a tight tolerance on a PTFE part, you already know the frustration. PTFE, the stuff most people call Teflon, is one of the most useful plastics we have for sealing, insulation, and low-friction surfaces. It’s also one of the trickiest to pin down to a number. Engineers who come from metal work or ordinary injection molding will often write ±0.05 mm on a drawing and expect it to hold. It won’t, and the fallout is the usual mess: scrap, arguments with the supplier, and parts that show up late.

So here’s what PTFE molded parts can actually do dimensionally, where the movement comes from, and how to write tolerances your shop can really hit. The honest summary is this: molding gives you shape, density, and a low cost. The accuracy comes later, from machining, from letting the part settle thermally, and from agreeing on how you measure it.

Why the dimensions are so hard to call

A finished PTFE part is several effects stacked on top of each other. The powder shrinks when you compact it. It shrinks again when you sinter it. It shifts as it cools, keeps drifting while internal stress bleeds out, and then changes size with the temperature of the room. Every one of those stages adds its own spread. That’s why a single tolerance number rarely tells the truth about a part.

I find it helps to think in layers:

Straight molded and sintered, no machining: fine for gaskets, thick blocks, non-critical insulators, rough locating pieces. Plan on percentage tolerances, around ±0.5% to ±1.5% on small simple parts, and ±1% to ±3% or wider as the part gets bigger or thinner.
Molded blank finished by ordinary turning or milling: this is where seals, bushings, valve seats, and insulating sleeves live. A commercial tolerance of ±0.10 to ±0.15 mm is realistic, opening up with size.
Molded blank, rough cut, stress relief, finish cut: precision sealing and instrument work. Small local features can sit at ±0.03 to ±0.08 mm, but only with temperature control, proper fixtures, and a defined inspection method.
Filled material or special fixturing at high precision: ±0.01 to ±0.03 mm on small features, judged one job at a time, never applied across the whole part.

There used to be a standard, ASTM D3297, that laid out tolerances for molded and machined PTFE. It’s been withdrawn, so don’t quote it as a contract requirement. The core idea is still sound, though: measure PTFE at a stable temperature, with equipment, fixtures, and a procedure that both sides have agreed on.

Shrinkage is the thing you respect first

PTFE shrinkage after molding is not subtle. Forget the 0.2% nudge you might use for an injection-molded part. Granular PTFE grades commonly run from about 2.25% to 5.5% mold shrinkage, and it depends on the resin. Virgin material tends to land between 2.25% and 4%; modified grades push toward 4%, 5%, and 5.5%. Those are typical figures, not promises you can write into a spec.

You can’t pick one shrinkage number, cut the mold, and forget it. Shrinkage moves with resin particle size and bulk density, how evenly you load the powder, preform pressure, dwell, sintering temperature, soak time, cooling rate, how much the wall thickness varies, which way the mold presses, the filler system, and how long the part anneals or sits in storage.

What actually works in the shop:

New mold on virgin PTFE: start around 2.5% to 4.5% shrinkage, and pull the number from supplier data for modified or special grades.
After the first parts, fix the mold from what you actually measured once the part is sintered and stable. Don’t assume the same tool stays put if you swap material.
Thick parts cool from the outside in, so expect core stress and a second shift later.
Thin rings are usually beaten by ovality, taper, and wavy edges long before linear shrinkage becomes the limit.
Filled PTFE changes how it shrinks, how it grinds, and how the surface looks, so you can’t borrow the pure-grade shrinkage rate.

What a directly molded part will hold

Treat a directly molded and sintered part as a blank, or at best a low-precision finished part. Old reference data put ring diameters around 1% (roughly 0.010 mm per mm), and gave even looser proportional numbers on height along the mold direction. The industry never treated an unmachined PTFE part as a precise copy of the tool.

A simple thick-wall part on a steady process with a corrected tool can beat those old numbers in places, but stay conservative when you quote and when you sign the drawing. These are reasonable places to start:

1 to 20 mm small thick-wall parts: ±0.10 to ±0.30 mm, or ±1%, whichever is the bigger number.
20 to 100 mm rings, blocks, sleeves: ±0.30 to ±1.00 mm, or roughly ±0.8% to ±1.5%.
100 to 300 mm parts: ±1 to ±3 mm, or about ±1% to ±2%.
Over 300 mm: ±2 to ±8 mm, or judged per job, usually set by warp, flatness, and ovality.
Thin rings, sheets, long strips: don’t lean on linear tolerances alone. Call out ovality, flatness, warp, and how you support the part while measuring.

The mistake I see most often is pinning a ring down on outside diameter, inside diameter, and wall thickness all at once with tight numbers. Pick the one dimension that actually does the job. If the seal works on the ID, control the ID and let the OD or wall relax. Otherwise every linear number can look fine on paper and you’ll still fight batch disputes, because cold flow, ovality, and the way you measure will eat the budget.

What you get after machining

PTFE machines cleanly. That doesn’t mean it’s dimensionally stable. The practical shop figure is about ±0.13 mm (roughly ±0.005 in) for normal turning, and you can go tighter, but it usually costs you a stress-relief step. For really tight work, heat the material a bit above where it will actually be used, but stay under 327°C, hold it by thickness, cool it slowly, leave 0.38 to 0.76 mm after roughing, relieve stress again, then finish.

A ladder that quotes well:

Commercial: ±0.10 to ±0.15 mm on pure PTFE, normal turning or milling, non-critical fits.
Finer: ±0.05 to ±0.10 mm for parts under 100 mm with sensible fixturing, sharp tools, and a stable inspection temperature.
High precision: ±0.02 to ±0.05 mm on small local features, thin finishing cuts, annealing when needed, agreed inspection.
Extreme: ±0.01 to ±0.02 mm, only on small features or short lengths, with sample validation, not as a general callout.

Measure turned parts in normal climate after they’ve reached thermal balance, and if you want a standard system, pick tolerances by an ISO IT grade. Under 50 mm, wall might reference IT10 and diameter IT11, with both loosening as the part grows.

Temperature is the variable that bites hardest

PTFE has a high linear expansion coefficient, usually quoted around 100 to 200 × 10⁻⁶ per °C. The number feels abstract until you run it. A 100 mm dimension moves about 0.01 to 0.02 mm for each 1°C change. A 200 mm part across a 5°C swing can shift 0.10 to 0.20 mm. So asking for ±0.03 or ±0.05 mm on anything over 100 mm, without a stable inspection environment near 23°C, isn’t really achievable.

There’s a second catch. PTFE has a crystal transition near 19°C. Older guidance called for measuring between 22 and 25°C and warned that 18 to 21°C is a difficult band because of that transition. It also said the part needs at least 24 hours at the inspection temperature or you’ll read it wrong. Move through the 19°C zone and the volume can swell by 1% to 1.5%, which matters the moment the dimension is critical.

A drawing note that saves arguments reads like this: “Unless otherwise specified, inspect at 23 ± 2°C after the part has stabilized in that environment for 24 hours. Measure thin-wall or flexible parts with agreed fixtures or supports that don’t deform them.” For precision work, tighten to 23 ± 1°C and agree the measuring force, the gage type, and the support method with your supplier.

Cold flow, creep, and measuring force cause false rejects

PTFE is low on strength and stiffness, and it will cold flow or creep under load. That alone can make some tolerances meaningless, but the real shock is in the measurement. Ordinary micrometers, calipers, and hard fixtures actually dent the material. An operator who doesn’t know this can read 0.13 to 0.25 mm small just from squeezing it.

Watch these:

Too much measuring force makes outside diameters read small and inside diameters read large. Limit the force, or use air gaging, optical measurement, or soft-contact gages.
Thin rings read oval when they’re free and work fine once they’re installed. Say plainly whether you’re checking the free state or the assembled state.
Fixture distortion springs back after you unclamp and the size drifts. Use soft jaws, enveloping supports, vacuum or dedicated fixtures, and light clamping.
Load over time drops seal preload and moves hole positions. Increase bearing area, avoid thin unsupported walls, and look at filled PTFE.
Different thermal histories widen the scatter inside one batch. Standardize the sintering, annealing, cooling, and stabilization time.

What filled PTFE does to the picture

Filled PTFE usually beats pure PTFE for holding a dimension. Fillers lift resistance to initial load deformation, stiffness, heat conduction, creep, and thermal dimensional stability, often roughly doubling creep and thermal stability in many mixes.

Pure PTFE: softest, most cold flow, most expansion. Give it the widest tolerances. Best where you need purity, strong corrosion resistance, insulation, or low friction.
Modified PTFE: better on deformation and permeation, usually steadier than pure, but you still verify its shrinkage on its own.
Glass-filled: good stiffness, creep, and thermal behavior, but it wears tools faster and can leave a rougher face that abrades the mating surface.
Carbon or graphite-filled: better wear, conduction, and creep resistance, but it may carry current and it’s sensitive to finish and tool condition.
Bronze or copper-filled: good load and heat handling and fairly steady, but heavier, and you have to check chemical compatibility.

Filled material specs often follow the ASTM D4745 system, built on virgin PTFE resin, split between non-pelletized compounds for general compression molding and free-flowing compounds for molding or ram extrusion.

How the design limits what you can hold

Thickness variation is the enemy. A part that jumps from thick to thin cools unevenly, builds stress at the junction, and warps or drifts. Keep walls as even as you can, make thick sections by machining later, put radii at the transitions, and leave stock on the faces that matter.

Thin rings need their own rules. The free size may not be the working size, so define exactly what you measure: free OD/ID, assembled OD/ID, gap, or compressed size. Don’t mix them.

Deep or slender holes are trouble in PTFE. Friction heat, stringy chips, and springback make the bore wander. Reaming is generally a bad idea; bore it or use dedicated tooling, don’t tighten the bore tolerance on a high length-to-diameter ratio, and use go/no-go gages for the functional size.

Flatness on plates and disks is often harder than length or width. Free, clamped, and in-service flatness can all differ. For a large thin sheet, mold a thick blank, rough both faces, stabilize, then finish both faces.

How much stock to leave

When a molded blank goes to machining, the allowance can’t be too thin or too fat. Too little exposes the molding shrinkage; too much invites residual stress and distortion. Single-side guidelines:

Under 50 mm: 0.5 to 1.0 mm
50 to 150 mm: 1.0 to 2.0 mm
150 to 300 mm: 2.0 to 3.0 mm
Over 300 mm: 3.0 to 6.0 mm, or judged from the measured blank
Thin or easily deformed parts: set the allowance from fixturing and how the distortion releases, not just a single-side number.

For precision work, the sequence that actually holds is: mold the blank, stabilize or anneal, rough, stabilize again, finish, rest at 23°C, then inspect.

What belongs on the drawing and in the purchase spec

Write these down so everyone measures the same thing:

Material: pure, modified, or filled, with the filler type and amount called out.
Process state: say “as-molded, sintered” or “machined from molded blank.”
Inspection temperature: 23 ± 2°C as default, 23 ± 1°C for precision.
Stabilization time: 24 hours at the inspection temperature before measuring.
Measurement method: define support and measuring force for thin-wall, soft, and ring parts.
Functional dimensions: control seal fit, groove width, lip, and bore first; let non-functional shape go loose.
Geometric tolerances: assign concentricity, roundness, and flatness by function, not by metal-part habit.
Sample validation: confirm real shrinkage, warp, and stabilized size on first articles before you freeze the tool or the process.

Just as important is what to leave off:

“Unspecified tolerance ±0.05 mm” is usually too tight across a whole PTFE part and just creates scrap.
“OD, ID, and wall all ±0.02 mm” over-constrains a ring past what measuring force and temperature allow.
“Inspect at room temperature” says nothing, because room temperature might be 18°C, 23°C, or 30°C, and PTFE will read differently at each.
“Machine to a metal-part drawing” ignores cold flow, expansion, and fixture distortion.
“Mold once to final, seal dimension ±0.03 mm” should usually become molded blank plus machining.

A five-layer rule for setting tolerances

First, function before everything. Find the dimensions that actually drive sealing, sliding, creepage distance, press fit, or locating, and relax the rest to what molding can do.

Second, split the processes. Let molding own shape and material use; let machining own the critical dimensions. Direct molding suits low-precision shapes or parts that don’t care about stiffness.

Third, size drives the tolerance. Under 100 mm you can talk in absolute numbers; over 100 mm you also have to weigh percent, temperature, and distortion. Small tolerances on big PTFE parts rarely pay for themselves.

Fourth, thin-wall parts are ruled by shape, not length. Roundness, flatness, warp, and how you support them set the limit, not diameter or length.

Fifth, precision demands inspection conditions. A ±0.02 to ±0.05 mm callout without 23°C stabilization, 24-hour equalization, low measuring force, and dedicated fixtures turns into a measurement fight, not a manufacturing one.

Where this lands

The straight reading on PTFE molded parts: molding locks in material density, basic shape, and cost, and the accuracy comes from machining, thermal stability, and inspection discipline.

As a rule of thumb:

Directly molded and sintered: about ±1% is a fair starting point. Complex, thin-wall, or large parts should move to ±2% to ±3% or be judged per project.
Commercial machined PTFE: ±0.10 to ±0.15 mm is safe.
Precision machined PTFE: small dimensions can reach ±0.03 to ±0.08 mm, but only with temperature control, light fixturing, stress relief, and agreed measurement.
±0.01 to ±0.02 mm is a case-by-case number for small features and short lengths under a special process, never a general tolerance.

Get these limits into the design up front and you’ll get parts that fit, seals that hold, and a supplier who can actually ship to the number on the drawing.