Why Aluminium and Copper Terminations Differ

Why this matters

Aluminium gets blamed for things that are not its fault and excused for the one thing that is. Techs who learned the trade on copper carry a set of habits that work because copper's own properties cover for them, then apply the same habits to aluminium and get a joint that is fine on the day and degraded in two seasons. The useful way to hold this is not "aluminium is worse", it is that a copper termination gets two things for free that an aluminium termination has to be given deliberately by the terminal, the compound and the installation instructions. Most of this article is about what is not the difference, because that is where the wasted effort goes.

Before opening any of these joints

Every inspection and repair described here happens on a de-energized, locked, tagged and proven-dead circuit under 29 CFR 1910.333(b)(2) for general-industry work or 29 CFR 1926.417 for construction, proven live-dead-live per NFPA 70E-2021, 120.5, which binds through your employer's electrical safety program in the edition the employer adopted. Aluminium terminations are common on service and feeder conductors, which is the highest available fault energy in most buildings and the worst place to take a shortcut.

What is not the difference: conductivity

Aluminium's conductivity is about 61 percent of copper's on the standard annealed-copper scale, and that fact is fully handled before the termination exists. The adopted code's ampacity tables already carry separate aluminium columns, which for the same ampacity generally lands an aluminium conductor about two trade sizes larger than the copper equivalent, and the sizing you take from those tables in the edition your jurisdiction adopted is the answer. A properly sized aluminium conductor is not working harder than a copper one, is not running hotter along its length, and has no penalty at the terminal on account of its bulk conductivity.

If your termination is failing, conductivity is not why. Do not upsize your way out of a joint problem.

What is not the difference: galvanic attack, in a dry enclosure

Aluminium and copper are far apart on the galvanic series, and in a wet environment that gap is a real corrosion driver. Inside a dry indoor enclosure it is rarely the mechanism that opens the joint first, because galvanic corrosion needs a continuous electrolyte and there is none. The dissimilar-metal joint has its own article and the corrosion mechanism itself is owned elsewhere in this library; take the mechanism from there rather than re-deriving it here. What matters for this article is that treating an indoor aluminium termination failure as a corrosion problem sends you looking for water that was never there.

What actually differs: two properties, both mechanical

Aluminium creeps under sustained pressure at ordinary temperatures. Hold aluminium under a steady compressive stress and it slowly flows out from under the load. Copper does this too, far less, and far more slowly. Every clamped aluminium joint is losing a little of the material that was under the clamp, forever.

Aluminium's oxide is hard, instant and insulating. Expose bare aluminium to air and an oxide film forms in seconds. It is a good insulator and it is harder than the parent metal, so it does not simply crush out of the way when you clamp it. Copper oxidizes too, but the film is softer and slower, and the clamping action itself breaks through it well enough that copper terminations mostly self-solve this without anybody thinking about it.

Everything in an aluminium termination system exists to answer one of those two.

The ratchet, with numbers

This is the part worth understanding, because it explains why an aluminium joint gets worse specifically at the moments you would expect it to be fine.

Near room temperature, aluminium expands roughly 13 parts per million per degree Fahrenheit, copper about 9.2, and steel about 6.5. Those are the common alloy values around room temperature and they shift by a few percent with alloy and with temperature, so treat them as the relationship rather than as a specification.

Take a set-screw lug with an effective clamp length of 0.5 inch through a steel screw and lug body, holding an aluminium conductor, and put the joint through a 100 F rise from a cold start to a hot summer afternoon under load.

• Aluminium in the grip grows 13e-6 times 0.5 inch times 100 F, which is 0.00065 inch.
• Steel in the grip grows 6.5e-6 times 0.5 times 100, which is 0.000325 inch.
• The aluminium therefore tries to grow 0.000325 inch more than the loop of steel holding it.

Now compare that against the stretch already in the screw. A steel screw at a working preload stress of around 50,000 psi, with steel's elastic modulus near 29 million psi, is strained about 0.00167, which over a 0.5 inch grip is 0.00083 inch of stretch. Adding 0.000325 inch of interference on top of that is roughly a 39 percent increase in the screw's strain, and therefore in the clamping force.

That figure treats the screw as the only elastic member and the aluminium and lug body as rigid, which is an upper bound rather than a prediction: a real stack shares the strain among all three, so the actual rise is smaller. The direction and the order of magnitude are what matter. Heating the joint does not loosen it. Heating the joint makes it substantially tighter, and that is the problem.

At the hot end of the cycle the aluminium under the screw is carrying a compressive stress well above what you set, and above its creep threshold, so it flows sideways. At the cold end of the cycle everything contracts back, but there is now less aluminium under the screw than there was, so the clamping force lands lower than it started. Repeat that with every cooling season, every night, every start and stop. The joint ratchets its own preload down.

Run the same arithmetic for copper: 9.2 minus 6.5 is 2.7 parts per million per degree F, so 2.7e-6 times 0.5 times 100 is 0.000135 inch of differential, which is about 41 percent of aluminium's 0.000325. Copper generates less than half the extra interference and resists flowing under it far better. The two effects compound, and that is the whole gap between the two metals at a terminal.

What that means for what you install

• The terminal must be identified for aluminium. Under NFPA 70 (National Electrical Code) Article 110, in the edition your jurisdiction adopted, connectors have to be identified for the conductor material, and a lug marked for copper only is not made legal by any amount of care. On 15 and 20 amp receptacles and switches the corresponding marking is CO/ALR, and a device with no aluminium marking at all is a copper-only device.
• The building wire alloy matters and it is not the alloy from the 1960s. The branch-circuit trouble that produced that era's reputation involved an earlier alloy and devices that were never designed for it, and it was addressed by both a different conductor alloy family and by devices identified for aluminium. Modern aluminium building conductors are the AA-8000 series required by the adopted code's conductor article. An old aluminium branch circuit found in a house is a real condition to evaluate and it is not evidence about the aluminium feeder you are installing today.
• Torque comes from the connector's own instructions and is set once with a calibrated tool. There is no general aluminium torque table.
• Spring load belongs where the design calls for it. Bolted aluminium bus and pad connections often specify conical spring washers precisely because they keep a useful clamping force as the aluminium creeps, converting a stiff stack into a compliant one. Where the connector's instructions call for them, they are not optional hardware.

Five things that look like the fix and are not

More torque. Higher clamping stress means faster creep, not less. Over-torque also extrudes the conductor at the screw edge and can shear strands. The number on the connector is the number.

A scheduled re-torque. This is the most common well-intentioned error in the trade. Many listed connectors are designed and tested to be torqued once; re-torquing a joint that has already cold-flowed presses the screw into metal that has moved, sets a new preload on a smaller cross-section, and buys another cycle at the cost of a further reduction. Where a manufacturer or a maintenance specification does call for a re-check, that instruction governs and it will say so. Where it does not, a re-torque is a modification you have made to a listed assembly.

Compound as a substitute. An oxide-inhibiting compound answers the oxide half and nothing else. It cannot stop creep. It also is not universally required: some listed aluminium connectors are supplied with it pre-filled, some require it, some do not, and the instructions decide. Applying it where the instructions do not call for it is not neutral, because it can affect the clamping surfaces the connector was tested with. Treat it as a chemical while you are handling it: chemical-resistant gloves and eye protection, and read the safety data sheet before opening it.

Wire-brushing alone. Brushing bare aluminium breaks the hard oxide, and a new film starts forming immediately, so brushing without immediately applying the specified compound and closing the joint accomplishes very little. Wear eye protection while brushing, and keep a brush that has been used on other metals away from aluminium, because embedded particles of a different metal are exactly the contamination you spent the effort removing.

Treating the conductor as the suspect. When an aluminium termination has failed, the conductor is usually sound half an inch back. The failure is in the interface and the terminal, which is the weakest element in that load path, and the repair is a new connector on freshly prepared conductor, not a length of copper spliced in to avoid the issue.

The transition point

Where aluminium has to meet copper, the joint gets its own dedicated hardware: a connector listed for both materials, or a bimetallic transition made at the factory with the aluminium and copper permanently joined and each side terminated to its own metal. The reason is not just corrosion, it is that the copper side wants a copper preload regime and the aluminium side wants an aluminium one, and no single screw can be right for both. Put the transition inside a listed device rather than inside your lug.

Verifying you got it right

Set the marked torque with a calibrated tool, then witness-mark the screw against the lug body so the next visit reads movement instead of guessing at it. Record the drop across the joint with the load current beside it the first time the equipment runs under real load, and again a season later at a comparable current. That is energized work, and on a service or feeder it is the worst arc-flash exposure in the building, so it proceeds only under 29 CFR 1910.333(a) where de-energizing is infeasible or introduces a greater hazard, with the boundaries established and PPE selected for the incident energy at that enclosure. Where the equipment can be scheduled down instead, take a resistance reading de-energized and skip the exposure entirely. A copper joint that has not moved reads about the same both times. An aluminium joint that is ratcheting reads higher the second time at the same current, and that is the only early warning this failure mode gives you before it becomes visible as discoloration.

References

• NFPA 70 (National Electrical Code), Article 110 for connectors identified for the conductor material and installation to the listed torque, and the conductor article for aluminium building-wire alloy requirements, both in the edition your authority having jurisdiction adopted.
• 29 CFR 1910.333(b)(2), general industry, and 29 CFR 1926.417, construction, for de-energizing and verifying before contact work; NFPA 70E-2021, 120.5 for live-dead-live, binding through the employer's electrical safety program in the edition adopted.
• Connector manufacturer instructions, which own the torque value, the compound requirement, the spring-washer requirement and any re-check interval.
• See related: What a Dissimilar Metal Joint Needs From You; How to Make a Terminated Connection That Lasts; Why a Loose Electrical Connection Heats and Then Fails.