Abstract
In automotive assembly, laser brazing (LB) is important for enabling high-precision, light-weight car-body design with numerous fine details in body-in-white components. To ensure the production of defect-free LB seams, there are automated non-destructive examination (NDE) techniques for real-time inspection. Interpreting real-time NDE data using artificial intelligence (AI) powered processing systems represents a major step towards zero-defect manufacturing. As input/training dataset(s) are necessary to feed machine learning algorithms for AI development, the present work targeted emulating, on a lab scale, a broad variety of LB joints that can occur in a manufacturing line. The main focus of the present study was to optimize the braze seam geometry and investigate the braze seam imperfections, including a wavy LB seam, spatter formation around the seam edge, and porosity as substantial LB defects, and correlating these with variations in the process parameters, including laser power (P), travel speed (TS) and wire feed speed (WFS). LB was conducted on double-flanged joints made from a zinc coated hot-dip galvanized interstitial free steel using a Si-Bronze (CuSi3Mn1) filler wire for automotive applications. The results of this investigation showed that the LB seam quality and spatter formation were impacted by the heat input, as well as the TS/WFS ratio. It was observed that the lowest allowable heat input of 52.6 J/mm provided a defect-free LB seam, while only a 11% increase in the heat input to above 58.3 J/mm resulted in the appearance of a wavy LB seam, as well as spatter formation. Further increases in the heat input to 61.5 J/mm – resulting from the highest allowable P and TS – diminished the spatter occurrence. Also, it was noticed that keeping the ratio of TS/WFS close to 0.9 eliminated the spatter; however, the occurrence of porosity was observed when the TS was higher than 50 mm/s, regardless of maintaining the TS/WFS ratio at 0.9. An important metallurgical phenomenon was the formation of intermetallic compound (IMC) phases; these should be mostly formed at the interface of the brazing seam/substrate, but were also dispersed as islands inside the LB seam for heat inputs above 58.3 J/mm. Also, an increase in the TS led the IMC islands to be formed closer to the LB seam centre rather than the edges. Considering the inherent brittleness of IMC phases, these differences may influence crack initiation behavior and joint strength performance; this is an ongoing research area to optimize the LB process window.