Galvanic Corrosion and the Metals That Fight

Why this matters

A large share of what gets written up as a defective part is not a defect. It is a metal that was put next to the wrong metal with water in between, and it comes back as a warranty claim a season or two later. The shop eats the callback, the customer decides the brand is junk, and the replacement fails the same way because nobody changed the condition that killed the first one. Galvanic corrosion is worth learning properly because it is one of the few field failures that is fully predictable from things you can see and measure before you tighten anything.

The gate: three conditions, all of them required

A galvanic cell needs all three of these at the same joint. Evaluate them per joint, at the wettest condition that joint ever sees, not at the average condition:

1. Two metals at different electrochemical potential. Different alloys, not just different names. Cast iron and carbon steel are close enough to ignore. Copper and galvanized steel are not.
2. A continuous electrical path between them. Usually the metals touching, but a bolt, a bracket, a bonding jumper, or a shared chassis will do it just as well from three feet away.
3. One electrolyte wetting both at once. Tap water, condensate, rain, wet insulation, damp soil, a salted road film, mortar that has not cured. Dry contact does nothing. Two wet metals that share no continuous film of water do nothing either.

Break any one condition and the cell stops. That is the whole engineering problem, and it is why the fix is usually cheaper than the part.

The Boolean matters. All three, joined by AND. A single missing condition is a pass. That is different from most failure rules you carry, where any one of several conditions is enough to fail you.

Which metal loses

The metal that is more active (less noble) becomes the anode and gives up metal. The more noble metal becomes the cathode and is protected. Working from noble down to active, in ordinary water:

graphite and carbon, titanium, passivated 316 stainless, passivated 304 stainless, silicon bronze, copper, brass, tin, lead, active (crevice-starved) stainless, cast iron and carbon steel, cadmium, aluminum alloys, galvanizing and zinc, magnesium.

Two things in that list surprise people. Graphite is at the noble end, which is why a graphite gasket or a graphite-based anti-seize on a steel or aluminum flange in a wet location can drive corrosion on the metal it was supposed to help. And stainless appears twice, because stainless is only noble while its passive oxide film is intact. Starve that film of oxygen in a tight crevice full of stagnant water and the same alloy behaves like an active metal.

That ordering is not folklore. Published galvanic series are developed and used under ASTM G82, which is the standard guide for building one and for reading it as a prediction rather than a rate.

The further apart two metals sit in that list, the harder the cell drives. MIL-STD-889 puts numbers on this with an anodic index and gives allowable index differences by environment: tighter for harsh outdoor and marine exposure, looser for a controlled indoor space. Use the standard's own tables rather than a number from memory, because the allowance changes with the environment class and that is exactly the part people drop.

Area ratio is the multiplier, and it is the half everyone skips

The gate above tells you whether a cell exists. Area ratio tells you how fast it eats, and it swings the answer harder than the metal pairing does.

Corrosion current spreads across the anode. A large anode paired with a small cathode spreads a small current across a lot of metal, so local loss is slow and you see a haze. A small anode paired with a large cathode concentrates that current, and the small part disappears.

The field rule: the active metal should always be the larger, more replaceable piece. Fasteners, being small, should be the noble half of any couple. A stainless screw into an aluminum plate is a bad-looking pair that behaves acceptably. An aluminum screw into a stainless plate is the same pair, reversed, and it is a failure with a date on it.

Case one: passes the gate, survives anyway

A stainless machine screw with a stainless washer holds an aluminum equipment bracket on an exterior wall, under an overhang, in a region that gets wind-driven rain a few times a month.

Run the gate. Stainless and aluminum sit far apart in the list, so condition 1 is met. The screw threads bear directly on the aluminum, so condition 2 is met. Rain reaches both, so condition 3 is met. All three, so a cell exists. By the pairing alone this looks like a mistake.

Now the ratio. Call the wetted stainless contact area about 0.5 square inch for the screw head, washer face, and exposed shank (measure yours, these are illustrative figures for the arithmetic). Call the wetted aluminum about 40 square inches of bracket. The cathode-to-anode ratio is 0.5 divided by 40, or 0.0125, which is roughly 1 to 80 in the anode's favor.

The predicted result: a dull grey to white bloom on the aluminum immediately around the screw, spreading slowly, with no measurable section loss for years. That matches what you actually find on these brackets. The joint is fine. Leave it alone, and if you are replacing the screw, replace it with the same stainless rather than a plated steel screw that will rust and stain the wall.

Case two: same two metals, failed in a season

The same building, a different detail. A stainless or hot-dip-galvanized flashing panel is fixed with aluminum blind rivets, thirty of them across the panel.

Run the same gate. Same two metals, so condition 1 is met. Rivets clamp the panel, so condition 2 is met. It is a rain-washed exterior surface, so condition 3 is met. Identical verdict from the gate: a cell exists.

Now the ratio, and this is where the two cases split. Call the wetted aluminum at each rivet head and shank about 0.05 square inch, so thirty rivets give 1.5 square inches of anode. Call the wetted panel 40 square inches of cathode. The cathode-to-anode ratio is 40 divided by 1.5, or about 27 to 1, and now it runs against the aluminum.

Compare the two joints on the number that drives current density: 26.7 against 0.0125 is a factor of about 2,100. Anodic current density tracks that ratio to a first approximation, which is why one joint gives up a haze and the other gives up its fasteners. The rivet heads go chalky, then powdery, then the heads shear off in a wind event and the panel lifts. The panel is undamaged, so it gets diagnosed as a bad batch of rivets.

The gate answered yes both times, and it was right both times. The ratio is what separated the joint you walk past from the joint that pulls a panel off a wall, and nothing about the metal names told you which was which.

Breaking the circuit: which condition to remove

You have three conditions to attack, and they are not equally practical.

Removing condition 3, the electrolyte, is usually the most durable fix and the most often refused. Slope the surface so water sheds, add a drip edge, fix the drain that keeps the area damp, get the insulation dry and keep it dry. If a joint never gets wet, the metals do not matter. In the field, the electrolyte usually comes back, so treat this as a supporting fix rather than the whole answer.

Removing condition 2, the electrical path, is the fix people reach for and the one they most often get wrong. An isolating washer and sleeve, a nonconductive gasket, or a dielectric fitting all break the path through the joint itself. They do not break a path that runs somewhere else. In a building, two metal piping systems are commonly bonded together for electrical safety, so a dielectric union between them breaks nothing electrically while leaving both wetted. If you are chasing continuity in a system that includes a bonding conductor, do not disconnect a bonding jumper to test a theory; before working inside a panel or on any conductor you are treating as a source, de-energize and lock or tag the disconnect per 29 CFR 1910.333(b)(2) and prove the conductor dead with the live-dead-live sequence in NFPA 70E-2021, 120.5.

Removing condition 1, the pairing, is what you control at selection time and almost never after. Choose the second metal to be electrochemically close to the first, or accept the couple and fix the area ratio so the anode is large.

There is a fourth option the list does not contain, which is to add a deliberate sacrificial anode that is more active than either metal, so it is consumed instead. That is what a water-heater anode rod is, and it works because it is designed to be replaced on a schedule.

Before you open any corroded joint on a piping or pressure system, isolate it, relieve pressure to zero and confirm it at a gauge or an open vent rather than by feel, drain the line, and support anything the joint was holding up. Corroded fasteners fail unpredictably under load, and a joint that has been seized for years can release all at once. Where the isolating valve or disconnect could be operated by someone else while you are in the joint, lock or tag it and verify zero energy before loosening, per 29 CFR 1910.147.

Where this gate misleads you

Three cases where all three conditions are absent and the metal still fails, so you do not spend the visit looking for a second metal that is not there:

• Crevice corrosion. One alloy, a tight gap, stagnant water. The oxygen inside the crevice is consumed and never replaced, the passive film there breaks down, and the metal inside the crevice becomes anodic to the identical metal outside it. Stainless is the usual victim. There is no dissimilar metal to find.
• Concentration cells under a deposit. Same mechanism, driven by a scale patch, a wet gasket footprint, or debris sitting on the surface. The clue is that the damage stops exactly at the edge of the covered area.
• Erosion and flow attack. Copper tubing pitted on the inside of elbows from velocity or from grit, not from a couple. The pattern is directional and follows the flow, which a galvanic cell never does.

How to verify you got this right

Before you close the joint, answer these four out loud:

1. Name both metals. Not "steel and stainless," but which alloy family and whether either is plated or galvanized. A plating changes the surface that is actually in the cell, and once the plating wears through at the contact point, the metal underneath is what you have.
2. State which one you intend to lose. If you cannot name the anode, you have not chosen, you have guessed.
3. State the cathode-to-anode area ratio, roughly. If the anode is the smaller piece, you have not finished designing the joint.
4. Say when it gets wet, and from where. Rain, condensate, washdown, a leak nobody has fixed. If the answer is "it should not," treat that as "it will."

Then photograph the joint before you close it up. Six months on, a photo of clean bare metal at assembly is what tells you whether the bloom you are looking at is new or was there when you started.

References

• MIL-STD-889, Dissimilar Metals, for anodic index values and allowable index differences by environment class
• ASTM G82, Standard Guide for Development and Use of a Galvanic Series for Predicting Galvanic Corrosion Performance
• 29 CFR 1910.147 (general industry lockout/tagout for stored mechanical and pressure energy) and 29 CFR 1910.333(b)(2) (electrical de-energizing and lockout), with NFPA 70E-2021, 120.5 for the live-dead-live proving sequence
• Manufacturer documentation for isolating hardware, dielectric fittings, and approved fastener materials for a given base metal
• See related: Reading Rust and Corrosion Patterns; The Galvanic Series as a Field Tool