Electrification of the automotive industry has heightened demand for high-performance electric motors, where copper hairpin windings are favored for their superior electrical and thermal properties. However, autogenous welding of copper is challenging because of copper's high thermal conductivity and low absorptivity to infrared (IR) laser light, which often causes instability and defects in conventional IR laser welding. Blue diode lasers significantly improve copper absorption (up to ∼ 60%), enabling stable conduction-mode welds with minimal spatter. Despite this advantage, achieving repeatable, high-quality welds remains difficult due to the complementary near effects of process parameters. To address these challenges, we developed a hybrid optimization framework combining high-fidelity FLOW-3D thermal fluid simulations, systematic welding experiments, and artificial neural network (ANN) modeling. The trained ANN generates three-dimensional processing maps that predict welding bead width and penetration depth as functions of laser power, spiral scan speed, and elliptical scan speed. As a result, optimization parameters yield copper hairpin joints with high mechanical strength (210 - 260 N) and low electrical contact resistance (30 - 38 μΩ) based upon two criteria of welding width and penetration depth. In addition, the EDS analysis confirms that good welds exhibit minimal oxygen content and consistent copper composition. In contrast, over-welded regions display significant oxygen enrichment at multiple locations, although the highest mechanical strength and lowest electrical contact resistance are found. These findings indicate that excessive energy input promotes oxide entrapment and pore formation, underscoring the importance of optimized process parameters to maintain metallurgical cleanliness and joint integrity.
Quoc et al. (2026) studied this question.