Galvanic Corrosion in Liquid Cooling Loops: Mixing Copper, Aluminum and Stainless Steel

01Three Ingredients

Galvanic corrosion needs three things at once: two metals with different electrochemical potentials, an electrical connection between them, and an electrolyte touching both. Remove any one and it stops. In a cooling loop the electrical path is the plumbing itself, or the chassis, and the electrolyte is the coolant, so the only ingredient the designer usually controls is the choice of metals.[1]

When the three are present, the more active metal (the anode) corrodes faster than it would alone and the more noble metal (the cathode) corrodes slower. MIL-STD-889D, the defense standard for dissimilar-metal design, names the usual victims and culprits directly: magnesium, steel, zinc and aluminum as anodes, in contact with copper, stainless steel and nickel as cathodes. An aluminum manifold feeding copper cold plates is that sentence, in hardware.[1]

Two more factors set the rate. The first is the potential gap: the farther apart two metals sit in the galvanic series, the harder the anode is driven. The second is area ratio. Corrosion current is set by the cathode and concentrated on the anode, so a small anode connected to a large cathode (an aluminum fitting on a copper manifold) is the worst case, and a large anode with a small cathode (a stainless screw in an aluminum block) is often tolerable. The standard's compatibility tables are built at a 1:1 area ratio, so a real design with a small anode does worse than the table says.[1]

02The Galvanic Series for Cooling-Loop Metals

MIL-STD-889D Table IV ranks materials by corrosion potential in artificial seawater (ASTM D1141). The standard is careful to say that the series identifies which member of a couple is the anode but is not by itself a compatibility rating. The table below is the subset that matters in cooling and power hardware, in the standard's order from most active to most noble.[1]

Three things stand out. Bare aluminum sits far up the active end and copper far down the noble end, with steel and tin between them. Chem-film treatment moves aluminum slightly more active, not less; it protects by sealing the surface, not by changing the couple. And graphite is more noble than any metal on the list, which is why a graphite-loaded gasket or seal against aluminum is a corrosion cell in its own right.[1]

03What the Standard Says About the Common Couples

Table I of MIL-STD-889D goes further than the series: it gives a measured corrosion-rate class for each bare couple, immersed in artificial seawater at a 1:1 area ratio. Class 0 is compatible (under 0.009 mil per year on the anode); classes 1 through 6 are incompatible, with class 5 meaning 10 to 99.99 mil per year and class 3 meaning 1 to 4.99. For the couples a cooling loop actually contains:[1]

These classes are for bare metal in seawater, the harshest common electrolyte, and the standard says plainly that they are not to be read as a risk level for a particular design. Read them as a ranking: the aluminum-to-copper problem is an order of magnitude worse than the copper-to-stainless one, and that ratio survives the change of electrolyte.[1]

04The Coolant Is the Electrolyte

A cooling loop does not run on seawater, and the difference is the whole reason mixed-metal loops exist. Deionized water is a poor conductor and slows every galvanic cell, but it is also aggressive toward copper and aluminum on its own and picks up ions as it goes. Water-glycol coolants for electronics are sold with corrosion-inhibitor packages that passivate the wetted metals, and the inhibitor chemistry is formulated for a stated list of metals. That list is the design document. A coolant that is qualified for copper, brass and stainless may or may not be qualified for aluminum in the same loop; the supplier's wetted-materials compatibility sheet says which.

Two consequences for the hardware. First, every wetted part needs to be on that sheet: not only the cold plates and manifolds but the fittings, quick disconnects, pump housing, tubing, seals and the solder or braze filler inside anything that was joined. Second, inhibitors deplete. A loop that was fine at commissioning becomes a bare-metal loop when the coolant is neglected, and then the MIL-STD-889D classes start to apply. Maintenance intervals are part of the corrosion design.

Single-phase dielectric fluids for immersion cooling are a separate case: with no water present there is no electrolyte and the galvanic mechanism does not run, which is one of their selling points. Compatibility there is about elastomers and plastics, not galvanic couples.

05Design Rules That Work

MIL-STD-889D's requirement for dissimilar metals that are not continuously immersed is short: insulate the joint electrically and keep the electrolyte out of it. For wetted parts in a loop, the electrolyte cannot be kept out, so the rules shift toward removing the couple or managing it.[1]

Crevice corrosion rides along with all of this. Stagnant coolant in a tight gap, under a gasket or in a dead-end passage, goes oxygen-poor and acidic and attacks even a compatible metal. The passage-layout advice in our manifold guide, no dead legs and passages that flush through, is corrosion advice as much as cleanliness advice.

06Choosing Loop Materials

On the alloys themselves: 6061 vs other aluminum grades is covered in the aluminum guide, and 303 vs 304 vs 316 in the stainless guide. Passivation of stainless wetted parts is cheap and is what puts them at the "passivated" positions in the series.

07Wetted-Materials Checklist