Real-time layer detection and closed-loop control using femtosecond laser-induced breakdown spectroscopy (LIBS) and machine learning

This work introduces a real-time, closed-loop system for detecting chemical layer transitions during femtosecond (fs) laser micromachining of multi-layer metallic substrates using Laser-Induced Breakdown Spectroscopy (LIBS). The goal is to autonomously stop the laser upon reaching a new material interface—such as aluminum (Al), copper (Cu), or stainless steel (SS)—to prevent damage to underlying layers. An integrated setup combines a high-repetition-rate fs laser with a spectrometer that continuously monitors plasma emission. Spectral data are streamed to a Python-based pipeline, where a Partial Least Squares Discriminant Analysis (PLS-DA) model instantly classifies the ablated material. Upon detecting a layer change, the system sends a stop command to the laser controller. The approach achieves reliable layer identification and process interruption within 500 ms. Results validate the integration of fs-LIBS with real-time feedback for intelligent, precision-controlled micromachining in multi-material systems.