Choose acetal (POM) when the part must keep its size in damp air or water, carry a steady load as a clip or spring, or slide smoothly: it absorbs very little water. Choose nylon (PA66 or PA6) for higher heat, impact, abrasion and molded-in inserts, and design for it to swell and soften as it takes on moisture.
Nylon (polyamide, mostly PA66 and PA6) and acetal (polyoxymethylene, POM) are both semicrystalline engineering resins, both tough, both used for gears, bushings, clips and slides. On a dry datasheet they look close. The difference that settles most choices is what happens after molding: nylon's amide groups pull water out of the air, and acetal's chain holds far less.
DuPont's nylon design guide puts it plainly: PA66 settles at about 2.8% water in 50% relative humidity (the Zytel 101L datasheet, on 2 mm plaques, gives 2.6%) and about 8.5% at 100% RH, and that water swells the part by about 0.6% and 2.6% respectively. It also plasticizes the nylon, so stiffness and strength drop while toughness and elongation rise. The change reverses as the part dries, but slowly, because absorption and desorption both crawl through thick walls. A standard acetal copolymer, by comparison, takes up about 0.2% water at 50% RH and 0.65% at saturation, per Celanese's Hostaform C 9021 data. Even in water at 100 °C, Celanese reports it stays below roughly 1.6%.
So the first filter is simple. If a dimension, a fit or a stiffness value has to hold through humid summers, damp storage or water contact, acetal starts ahead. If the part has to run hot, take a hammer blow, shrug off abrasive grit or carry a molded-in metal insert, nylon starts ahead, and the design has to allow for growth and softening.
Comparing "nylon" with "acetal" in general hides how much each number depends on the grade and on moisture. The table uses two common unfilled grades. Nylon datasheets give two columns: dry-as-molded, and conditioned, meaning brought to equilibrium at 23 °C and 50% RH. Acetal datasheets give one, because moisture hardly moves its properties.
| Property (test) | PA66: Zytel 101L (dry / conditioned) | POM copolymer: Hostaform C 9021 |
|---|---|---|
| Density | 1.14 g/cm³ | 1.41 g/cm³ |
| Tensile modulus (ISO 527) | 3,100 / 1,400 MPa | 2,850 MPa |
| Yield stress (ISO 527) | 82 / 55 MPa | 64 MPa |
| Strain at yield | 4.5 / 25% | 9% |
| Creep modulus, 1 h and 1,000 h (ISO 899-1) | conditioned: 1,400 and 820 MPa | 2,500 and 1,300 MPa |
| Notched Charpy, 23 °C (ISO 179/1eA) | 5.5 / 15 kJ/m² | 6.5 kJ/m² |
| Notched Charpy, −30 °C | 4.5 / 3.0 kJ/m² | 6 kJ/m² |
| Water at 23 °C, 50% RH / saturated (ISO 62) | 2.6% / 8.5% | 0.2% / 0.65% |
| Heat deflection at 1.8 MPa / 0.45 MPa | 70 / 190 °C | 104 / 160 °C |
| Melting point | 262 °C | 166 °C |
| Mold shrinkage, flow / across (ISO 294-4) | 1.4 / 1.4% | 2.0 / 1.9% |
| Drying before molding | 80 °C, 2 to 4 h, to 0.20% max. | normally not needed; if wet, 100 to 120 °C, 3 to 4 h, to 0.15% max. |
| Melt / mold temperature | 280 to 300 / 50 to 90 °C | 190 to 210 / 80 to 120 °C |
Read the nylon column carefully. Dry-as-molded PA66 is stiffer than the acetal but has a yield strain of only 4.5% and a notched Charpy value close to the acetal's. After it picks up moisture, it is far tougher at room temperature and less than half as stiff. Which column describes your part depends on where it lives and how long it has been there. Cold impact is the exception: at −30 °C the conditioned nylon drops to 3.0 kJ/m², below the acetal's 6.
Creep is the slow extra deflection of a part that carries a steady load: a latch held closed, a spring arm preloaded against a cam, a boss clamped by a screw. The creep modulus is the stiffness to use for that case, and it falls with time. Both resins lose about 40 to 50% of it between 1 h and 1,000 h at room temperature, but the acetal starts higher and stays higher, and its value does not depend on the weather. DuPont lists "integrally moulded springs" and "stiffer and stronger in higher humidity" among the reasons to choose its Delrin acetal over its Zytel nylon, and positions Delrin 100 for highly loaded gears, plain bearings and snap-fits.
Snap-fits add a timing trap with nylon. A clip assembled straight off the press is in the dry state, with that 4.5% yield strain, so a snap that is fine on a conditioned sample can crack on the assembly line. Either size the snap for dry strain, or condition the parts before assembly. The snap-fit design guide covers the arm geometry, and the snap-fit calculator checks strain against a limit you enter.
The latch. Take a hypothetical cantilever latch arm 2.0 mm thick that is held deflected against a catch for the life of the product. Under a steady load, deflection scales with 1 divided by the creep modulus. At 1,000 h the conditioned PA66 arm has a creep modulus of 820 MPa and the acetal arm 1,300 MPa, so the nylon arm sags 1,300 ÷ 820 = 1.59 times as far. Arm stiffness scales with modulus times thickness cubed, so matching the acetal means a nylon arm 2.0 × (1.59)1/3 = 2.0 × 1.17 = about 2.33 mm thick. That extra third of a millimeter also changes the snap strain, so recheck it.
The gears. Now a pair of hypothetical 40 mm spur gears on fixed shafts in a metal frame, so the center distance does not move. In PA66, moving from dry-as-molded to equilibrium at 50% RH grows every dimension about 0.6%: 40 × 0.006 = 0.24 mm on each gear's diameter, so each pitch radius grows 0.12 mm and the mesh tightens by about 0.24 mm along the line of centers. If the gears end up wet or saturated, use 2.6%: 40 × 0.026 = 1.04 mm per gear, a large change for any gear mesh. For scale, Zytel 101L's thermal expansion in the flow direction is 1.0 × 10−4 per °C, so 0.6% of humidity growth equals heating the gear by 0.006 ÷ 0.0001 = 60 °C. An acetal gear absorbs about a thirteenth as much water at 50% RH (0.2% against 2.6%), so the same allowance is not needed.
The tool. The two resins also shrink differently in the mold. Using the datasheet values, a cavity for the 40 mm gear is about 40 × 1.014 = 40.56 mm for PA66 and 40 × 1.020 = 40.80 mm for the acetal, a 0.24 mm difference, the same size as the humidity growth above. To repeat the math on your own part, use the plastic shrinkage calculator; how much plastics shrink in the mold explains why the real number depends on wall thickness and gating.
Result for this example: acetal for the latch and the gear pair if the product sees humidity swings or water, unless the gears run hot or take shock loads. In that case nylon works if the center distance and backlash are set for the conditioned size, since DuPont notes nylon parts in ordinary air drift toward the moisture level of mid-range humidity.
Friction values depend on the test, so compare them only within one table. In DuPont's unlubricated thrust-washer test (23 °C, 2.1 MPa, 3 m/min), Delrin 100, 500 and 900 on steel gave a static coefficient of 0.20 and a dynamic coefficient of 0.35. Zytel nylon on steel ranged from 0.31 to 0.74 static and 0.17 to 0.43 dynamic. Delrin 500 running on Zytel 101L gave the lowest pair, 0.10 static and 0.20 dynamic. Acetal's static friction is lower than its dynamic friction, which DuPont points out limits stick-slip: useful in slides and small mechanisms that must start moving smoothly from rest.
Wear is a separate question from friction. DuPont rates nylon higher for abrasion resistance, impact and shock absorption, and quieter running, and rates acetal better for running against soft metals such as brass or aluminum at low load and speed. For any plastic bearing on a metal shaft, the guide says shaft hardness is the most important factor, and soft or stainless steel shafts wear plastic bearings quickly. One useful trick from the same source: running an acetal gear against a nylon gear gave lower wear and friction than either running against steel.

Neither resin likes strong acids, but they fail in different service. Celanese states its acetal copolymer withstands strong alkalis (50% caustic soda, even hot) but is attacked by oxidizing agents and strong acids below about pH 4. Its own chart rates 10% hydrochloric acid and roughly 10% chlorinated lime as not resistant. In drinking water, Celanese warns that chlorine and ozone reduce creep strength in stressed parts, and it sells special grades for acidic cleaners and chlorinated water. Its standard grades resist fuels, oils and most solvents, and Celanese describes them as not prone to environmental stress cracking.
Nylon resists oils, greases and fuels very well, and alkalis up to about 40% concentration, per DuPont, but strong mineral acids and oxidizing agents attack it rapidly, especially hot. Its weak point in water service is heat plus oxygen: DuPont notes a 15 °C rise can cut life by 40 to 50%, that oxygen-rich water from a heater is harsher than stagnant boiling water, and that lab chlorine levels up to 8 ppm may reduce service life by 20 to 30% with possible microcracking. For steam or hot water, DuPont points to hydrolysis-resistant PA66 grades, and to HSLR glass grades for hot glycol and water mixtures such as coolant.
The heat-deflection numbers seem to contradict each other until you read the load. At 1.8 MPa the acetal holds to 104 °C and the unfilled nylon only 70 °C, because PA66's glass transition sits near 60 °C dry. At 0.45 MPa the nylon goes to 190 °C and the acetal 160 °C, and nylon melts almost 100 °C higher. DuPont gives Delrin a useful range in air of −50 to 90 °C with intermittent use to 120 °C, and lists "highest end-use temperature" as a reason to pick nylon, with a heat-stabilized PA66 for long service at temperature. Rule of thumb: lightly loaded and hot, nylon; loaded, warm and dimension-critical, acetal, or a glass-filled grade of either.
Acetal comes in two families. Delrin is a homopolymer (POM-H); Celcon and Hostaform are copolymers (POM-C). Celanese explains that the comonomer units stop a breaking chain from unzipping back to formaldehyde, and credits the copolymer with better chemical, fuel and UV resistance and long-term thermal stability than the homopolymer. The homopolymer melts higher, 178 °C for Delrin against 166 °C for Hostaform C 9021, and DuPont reports that its high-viscosity Delrin 100 outperforms Delrin 500 by about 40% in gears, at the price of being harder to fill. For a loaded gear in dry indoor service, the homopolymer is worth asking for by grade. For hot water and detergent, Celanese backs its copolymer with immersion data and lists washing-machine and dishwasher parts among its uses, so start there.
Glass fiber raises stiffness, creep resistance and heat deflection in both resins, but it brings anisotropic shrinkage. DuPont measured glass-reinforced nylon at 0.2 to 0.4% shrinkage along the flow and 0.8 to 1.2% across it; its 25% glass Delrin 525GR lists 0.4% and 1.2%. That difference is where warp comes from. Glass also has costs specific to this comparison. DuPont advises against glass-filled grades for gears because the fibers wear both the plastic gear and its mate, and Celanese reports that glass-filled acetal loses tensile strength quickly in hot water, dropping back to about the level of the unfilled grade, because water attacks the glass-to-resin interface. The engineering and glass-filled resins page covers tooling wear and ordering by grade name.
| If the part needs | Lean toward | Why |
|---|---|---|
| Tight dimensions in humid air or water | Acetal | About a thirteenth of nylon's water uptake at 50% RH |
| A spring, clip or latch under steady load | Acetal | Higher creep modulus, unaffected by humidity |
| Smooth start-up, no stick-slip | Acetal | Static friction lower than dynamic |
| Running against brass or aluminum | Acetal | DuPont rates it better against soft metals |
| Strong alkalis or hot water with detergent | Acetal copolymer | Celanese data for caustic and hot-water service |
| Acids, bleach or chlorinated water | Neither by default | Test, or use a media-resistant grade |
| Hot, lightly loaded service | Nylon | Higher melting point and 0.45 MPa deflection temperature |
| Impact, shock and abrasive grit | Nylon (conditioned) | Tougher once it has absorbed moisture; best abrasion rating |
| Oils, greases, fuels | Either | Both resist them well |
| Molded-in metal inserts | Nylon | DuPont lists insert molding as a nylon strength; see insert molding |
The resin makers' own application lists are a useful sanity check. Acetal: complete gear trains in appliances and small precision reducers for clocks and timers, plain bearings, snap-fits and integral springs, fuel-system components, and washing-machine and dishwasher parts in the copolymer. Nylon: hot-water mixing valves, gears and baskets that see hot water, parts in contact with oils and fuels, components in alkaline batteries, and worm gears where abrasion would wear metal teeth.
If nylon's toughness is needed but its growth is not acceptable, two options narrow the gap: a lower-absorbing nylon such as PA612, which DuPont recommends for humid service because of its lower moisture uptake, or a mineral- or glass-reinforced nylon, which it describes as more dimensionally stable and creep resistant than unreinforced grades. If neither resin holds the load at temperature, step up to a glass-filled or high-temperature grade, and start from the resin selection guide or the plastic material selector. For housings rather than mechanisms, the trade-offs are different; ABS vs polycarbonate vs polypropylene covers those.
Delrin is a trade name for acetal homopolymer (POM-H). Celcon and Hostaform are trade names for acetal copolymer (POM-C). Both are polyoxymethylene and behave similarly. The homopolymer melts higher and DuPont recommends Delrin 100 for loaded gears, while Celanese credits the copolymer with better chemical and long-term thermal stability. Specify the grade, not just the word acetal.
Acetal for most small, precise gear trains in dry or humid indoor service, because it keeps its size and stiffness. Nylon when the gears run hot, take shock loads or face abrasive dirt, provided backlash is set for the conditioned size. Running an acetal gear against a nylon gear often wears less than either against steel. Avoid glass-filled grades for gears unless tested.
Nylon does. Zytel 101L, for example, calls for 2 to 4 h at 80 °C to reach 0.20% moisture or less, and DuPont notes only dry nylon molds successfully. Acetal is normally molded as received since it absorbs so little water, but it should be dried if bags were left open or it was stored badly.
Rarely well. Their datasheet shrinkages differ, about 1.4% for unfilled PA66 against about 2.0% for acetal copolymer, which on a 40 mm feature is roughly 0.24 mm. They also need different melt and mold temperatures, and nylon's later moisture growth moves dimensions again. A tool is normally cut for one resin; changing resin later usually means steel work.
Fresh nylon parts are dry. Dry PA66 such as Zytel 101L yields at about 4.5% strain and has a notched Charpy value of 5.5 kJ/m², against 25% and 15 kJ/m² once conditioned at 50% RH. A snap sized from conditioned data can break on the line. Size it for dry strain or condition the parts before assembly.
Send the part file and the quantity you expect to run, and we'll review it before you commit to tooling. NDA first if you want one.
Send a part for reviewBook a 30-minute call