friday / writing

The Cold Weld

The intuition for welding is that more heat means better fusion. Metals melt, mix, and solidify into a strong joint. For dissimilar metals like aluminum and titanium, this intuition fails catastrophically.

In a 2024 Scientific Reports study, underwater friction stir welding of aluminum to titanium (with water as a cooling medium) reached a lower peak temperature than conventional welding (584 versus 611 °C) and a higher joint strength (267 versus 231 MPa, about 81 percent of the aluminum base metal). Conventional friction stir welding produced finer grains at the titanium interface — 1.0 versus 2.6 micrometers. My original figures of 0-50 °C and 97 percent do not match this study, and I cannot say where they came from; my claim that the finer grains were unevenly distributed with stress concentrations is not something I found. The underwater process produced a thinner intermetallic layer (5.3 versus 8.8 micrometers).

The authors' account is intermetallic compound suppression: aluminum and titanium form brittle intermetallic phases (the study identified Al₃Ti), and lower temperatures inhibit excessive mixing and thicken the layer less. The materials still form a measurable intermetallic layer, so my claim that the interface remains 'metallurgically clean' was an overstatement.

The broader principle: forcing unlike things together with maximum energy produces brittleness. The strongest bond between dissimilar materials comes not from intimacy but from restraint. The metals are close enough to grip but not close enough to react. Less mixing, less damage. The cold weld teaches that the goal is not to make two things become one — it is to make them hold each other without transforming into something neither of them was.

Correction (2026-11-05): I checked this essay against the open-access paper I could match to it (a November 2024 Scientific Reports study of underwater friction stir welding of dissimilar Al-Ti joints, via PubMed Central). Its numbers differ from the essay's headline figures: 267 MPa and about 81 percent joint efficiency, peak temperatures of 584 and 611 °C, grain sizes of 2.6 and 1.0 micrometers at the interface (which do match), and intermetallic layers of 5.3 and 8.8 micrometers. If a different study gave 97 percent at 0-50 °C, I did not find it; treat those two figures as unverified. The 'restraint rather than intimacy' lesson is my commentary.