Stainless Fastener Galling: Why Threads Cold-Weld and How to Stop It
28 min·Yaxiio Engineering Alliance

Stainless Fastener Galling: Why Threads Cold-Weld and How to Stop It

Stainless steel galling: why threads cold-weld mid-assembly, how it cuts preload at the same torque, and which lubricants and release methods work.

What Galling Does to a Joint

A stainless M10 nut that turns by hand for half its length and then stops is not cross-threaded. If backing it off leaves a bright, torn smear on the bolt thread, the two surfaces have cold-welded. On a site where the joint must come apart for inspection, that half turn decides whether the job takes minutes or hours.

Three answers first:

  • Separate the flanks with something that shears before the stainless does.
  • Reduce pressure and sliding distance: coarser pitch, slower drive, shorter engagement.
  • Change the material pair only after checking the galvanic couple.

What follows is the mechanism, the preload loss, and how to release a seized fastener.

The Cold-Weld Feedback Loop

Galling is adhesive wear between two metals sliding under pressure without a separating film. In a thread it is a loop, and each pass makes the next worse.

  1. Austenitic stainless carries a chromium-rich oxide film a few atomic layers thick, which re-forms quickly where oxygen is present.
  2. Under preload the flanks touch at asperities, not over the full nominal area.
  3. Local pressure ruptures the film at those points.
  4. Inside the engaged thread oxygen is scarce, so the film does not rebuild during the pass.
  5. Clean metal meets clean metal and adheres at the true contact spots. This is cold welding, a bond formed without melting.
  6. Rotation continues. The welded junctions are stronger than the surrounding metal, so shear occurs just below the interface, tearing metal from one flank onto the other.
  7. The torn flank is rougher and work-hardened, so the next turn has more asperities and more places to weld.

The nut does not become hard to turn because it is tight. The interface has changed from two sliding surfaces into a set of welded junctions. Once torque is spent shearing them instead of advancing the nut, more torque damages the thread faster than it advances the joint.

Why Stainless Steel Galling Is Different

Three properties separate austenitic stainless from a zinc-plated carbon steel bolt.

The film needs oxygen, and the thread has none. Stainless is self-healing in air, which is why a scratched rail does not rust overnight. Between two engaged flanks the ruptured film cannot heal during the pass. The property that protects the surface in service leaves it unprotected at assembly.

The crystal structure work-hardens instead of breaking away. Austenitic stainless is face-centred cubic and hardens rapidly under deformation. A sheared asperity does not crumble like a brittle oxide; it hardens, drags, and takes metal with it. The transfer layer left on the opposite flank is work-hardened too.

Heat stays where it is made. Austenitic stainless conducts heat poorly compared with carbon steel, so friction heat from a power tool concentrates at the contact patch. A warmer patch adheres more readily.

A zinc-plated carbon steel bolt can gall too, but it usually fails differently. The zinc is softer and acts as a sacrificial separating layer; it wears or corrodes away first, and the joint gives warning as red rust. Zinc-flake coatings fail at the barrier level: the flake layer and binder wear through, then red rust appears. Neither cold-welds as readily as bare stainless, because the coating is designed to shear.

Joint Separating layer First failure What the operator sees
Bare stainless Passive film only Film ruptures, metal transfers Torque rises sharply mid-thread
Zinc-plated steel Soft zinc Zinc wears, steel corrodes Red rust later
Zinc-flake steel Flake layer in a binder Barrier wears or chips Red rust at damage

Conditions That Trigger Galling

Galling needs pressure, sliding, and bare metal. Raising any one moves the joint closer to seizure.

  • Contact pressure. Preload is carried by the first few engaged threads, so a small diameter at high preload concentrates stress on a narrow band.
  • Sliding speed. A high-speed nutrunner moves the flanks fast and puts friction heat into a small area.
  • Surface roughness. FED-STD-H28/21B (1984) §4.4.2 limits thread texture to 2.5 µm Ra for cut threads and 1.6 µm Ra for rolled and ground threads. The standard’s note ties rough texture to cross-threading; the same asperities are where the film ruptures first.
  • Dry friction. Solvent-cleaned threads have no separating film.
  • Assembly method. MIL-STD-1312-15 (1984) §5.1 makes hand torque with a tensile machine the referee method, because the torque-tension relation depends on how the fastener is driven.
  • Thread length and pitch. A fine thread packs more turns into the same engagement, so the flanks slide further; long engagement adds turns. Both add chances for a weld.
  • Thread condition. Burrs, nicks, heat tint and re-used nuts start the loop earlier. A tight fit is not safer: FED-STD-H28/2B (2019) §50.3 notes that a close-fitting assembly can fail where a looser fit would work.

The standard metric fit is 6H internal and 6g external (FED-STD-H28/21B §5.1.1). That is a starting point, not a galling control.

What Galling Does to Preload

Torque is an indirect measure of preload:

Formula
T=K×F×d
F=T/(K×d)

T is torque, F preload, d nominal diameter, and K a dimensionless nut factor bundling thread friction, under-head friction, and thread geometry. K is a property of the joint as assembled, not a material constant.

For M10 at 40 N·m (d = 0.010 m):

  • K = 0.20: F = 40 / (0.20 × 0.010) = 20,000 N
  • K = 0.30: F = 40 / (0.30 × 0.010) = 13,333 N

That is 6,667 N less, or 33.3% of the preload at K = 0.20. The wrench reading did not change; the clamp did.

Galling raises K twice: it adds a friction term for shearing welded junctions, and it roughens the flanks and the bearing face. The operator’s usual response, adding torque, raises the shear load and makes the next turn worse.

The 0.20 and 0.30 values are planning values, not measured constants. For safety-critical preload, measure the torque-tension relation for the actual lot and lubricant; MIL-STD-1312-15 defines the test. Carbon steel torque tables do not transfer to stainless. A property class such as 8.8 encodes nominal tensile strength and yield ratio under GB/T 3098.1-2010 §5, while austenitic stainless is designated under another system and its friction depends on surface treatment.

Prevention: Effect, Cost, and Limits

Effect, cost, and limits for each measure:

Method Effect Cost Do not use where
Nickel anti-seize Separates flanks at high temperature Consumable; recalculate torque Oxygen service; controlled nickel contamination
Copper anti-seize High load capacity Consumable; residue Oxygen service; near aluminium or in seawater (galvanic); above filler service range
MoS2 paste Low friction under pressure Consumable; residue Oxygen service; wet marine joints; elastomers the carrier attacks
PTFE or wax dry film Clean, repeatable friction Coater process step Above its temperature rating; very high pressure
Coarse thread Fewer turns, more clearance Design change Thin walls or adjustment joints; stress area is smaller
Lower drive speed Less heat, repeatable K Slower cycle High-volume lines without low-speed nutrunners
Controlled finish Fewer asperities Receipt inspection Not a substitute for lubrication
Dissimilar pair, such as a 316 nut on a 304 bolt Different hardness at the interface Part number change; strength penalty Wet or marine service without protection
Passivation Restores the passive film Batch process Cannot replace descaling; does not change friction

Two rows need qualification. Changing the alloy fixes the thread directly and is the easiest way to create a corrosion cell. MIL-STD-889C (2016) §5.2 sets the pairing rule: a stainless screw fastening aluminium sheet puts the smaller area on the cathodic member, which is the acceptable direction; the reverse is not. §5.1.6 requires joint edges to be sealed where stainless contacts itself, to limit galvanic and crevice attack. That clause is about corrosion, not galling.

The second is coating choice. GB/T 5267.1-2023 §1 notes that electroplating on stainless fasteners is intended for lubrication and to avoid thread seizure, a different purpose from corrosion protection on a carbon steel bolt. Plated high-strength fasteners bring a separate risk: hydrogen embrittlement, covered in the hydrogen embrittlement guide. The coatings and platings library page covers the systems themselves.

What Passivation Can and Cannot Fix

Passivation is chemical, not mechanical. QQ-P-35C (1988) defines it for corrosion-resistant steel.

What it fixes. Forming, machining, tumbling and lapping smear or embed iron particles. §6.1.1 states that these must be removed or they appear as rust or stain spots, and that nitric acid with sodium dichromate dissolves them and lets a thin, transparent passive film form. Cleaning comes first: §3.2 requires a surface free of oil, grease, rust and scale, descaled to ASTM A380. Acceptance is by water immersion or high humidity without free-iron staining (§3.7.1), or salt spray and copper sulfate (§3.7.2, §3.7.2.1), with no etching, pitting or frosting (§3.6).

What it cannot fix.

  • Heat tint and weld scale. Removed in pretreatment (§3.2), not by the bath.
  • Carburized surfaces. §6.7: cannot be passivated, because carbon ties up chromium as carbides.
  • Nitrided stainless. §6.8: should not be passivated, because the treatment corrodes the nitrided case.
  • Torn thread geometry. Passivation cannot restore metal torn from a flank, remove the transfer layer, or reverse work hardening.

A passive film is not a lubricant. A clean, passivated, dry thread can gall on the next assembly as before.

QQ-P-35C Table I defines nitric-acid types by concentration, temperature and minimum time; Types I, III, IV and V are withdrawn. The type must be specified on the drawing or purchase order.

Releasing a Seized Fastener

Once the nut stops, the task changes from tightening to separating the parts without destroying the bolt.

  1. Stop adding torque. More torque loads the junctions and tears more metal. If it moved less than about one turn and stopped, treat it as seized.
  2. Work it back and forth. A few degrees each way with a hand tool. A junction that will not shear one way may shear the other.
  3. Apply penetrant. A low-viscosity oil needs a gap. A cold weld has none, so penetrant helps at the boundary, not at the weld. It is cheap to try, not a repair.
  4. Apply heat within limits. Differential expansion can break the joint, but heat reaches seals, coatings and adjacent parts first. GB/T 5267.1-2023 §7.3 notes that high temperature affects corrosion resistance and uses 120 °C and 150 °C thermal cycles among its test options; those are a reference for what a coated fastener should survive, not a target. Never use an oxy-acetylene flame on stainless: carbon pickup and sensitisation damage the part. Use induction or a heat gun with a thermocouple, and check the seal rating first.
  5. Use axial impact. A few blows on the head can break the junction. Tangential impact rounds the hex.
  6. Cut as a last resort. A nut splitter or abrasive disc removes the nut with least damage to the bolt. Protect the thread, replace the nut, inspect for transfer layers. Drilling is the final option.
1Has the nut moved?
Less than one turn, then stopped
✅
treat as seized; reverse a few degrees
Turns with rising effort
⚠️
stop, lubricate, reduce speed
↓
2Will heat reach a seal or coating?
Yes
✅
penetrant and reverse motion only, then cut
No
⚠️
controlled heat with a thermocouple
↓
3Is the nut accessible?
Yes
✅
split it and protect the bolt thread
No
⚠️
drill and extract; plan to replace the bolt

After removal, inspect both threads. A bright smeared patch is metal transfer; a torn profile means the bolt lost material. Chase the thread with the correct 6H or 6g tool only if damage is light; replace the pair if the profile is torn. Reassemble with the missing lubricant or coating and reduce drive speed.

Putting It Into Practice

Three decisions settle most cases. Choose the separating layer before the torque value, because K changes with lubricant and coating. Measure K for the actual lot if preload matters, because a 0.10 shift costs a third of the clamp. Keep passivation in its place: it restores corrosion resistance, it does not repair a torn thread.

If a batch has seized, review the assembly instead of raising the torque setting. Send the fastener grade, size and pitch, the mating material, the lubricant or coating, the drive method and speed, and the required preload. That is enough to check the galvanic pair and the galling conditions.

References

Key Standards

Standards and clauses referenced by this article.

  1. Passivation Treatments for Corrosion-Resistant Steel (§3.2, §3.6-3.7, §6.1.1, §6.7-6.8)
  2. Dissimilar Metals (§5.1.2, §5.1.6, §5.2)
  3. Metric Screw Threads (§4.4.2, §5.1.1)
  4. Unified Inch Screw Threads (§50.3)
  5. Fastener Test Methods, Method 15: Torque-Tension (§1.1, §5.1)
  6. Fasteners, electroplated coatings (§1, §7.3)
  7. Mechanical properties of bolts, screws and studs (§5)

References

Sources used for fact checking and background context.

  1. QQ-P-35C (1988) — Passivation Treatments for Corrosion-Resistant SteelU.S. Department of Defense
  2. MIL-STD-889C (2016) — Dissimilar MetalsU.S. Department of Defense
  3. FED-STD-H28/21B (1984) — Metric Screw ThreadsU.S. General Services Administration
  4. MIL-STD-1312-15 (1984) — Fastener Test Methods, Method 15: Torque-TensionU.S. Department of Defense
  5. GB/T 5267.1-2023 — Fasteners, electroplated coatingsSAMR

Deep Reading

More systematic selection, procurement, or inspection guides.

This page is one entry in the fastener engineering reference. The library covers standards, dimensions, materials, and failure analysis; every figure is traced back to its source.

Continue reading in the library →