An event-driven correction method for enhancing signal stability in laser-induced breakdown spectroscopy using a dynamic vision sensor

Laser-induced breakdown spectroscopy (LIBS) has attracted considerable research interest and found wide application across diverse fields. However, LIBS signals often suffer from poor stability due to several factors, most notably the matrix effect, which remains a major obstacle to achieving accurate quantitative analysis. Existing methods are frequently ineffective, complex, or costly. To address these challenges, this study proposes a novel and cost-effective method for high-precision spectral correction. This study utilized event data from a dynamic vision sensor (DVS) to extract plasma features, specifically the number of events and plasma area, which characterize the plasma temperature and total particle number density, respectively. Based on these features, the DVS-T1 correction model was developed and applied to carbon steel and brass samples. The calibration curves obtained after correction for the Fe I 355.851 nm, Mn I 403.076 nm, Cu I 327.396 nm, and Zn I 328.233 nm lines achieved R² values of 0.994, 0.999, 0.995, and 0.999, respectively, surpassing those of the original data and spectral normalization. The mean relative standard deviation of the corrected signals decreased by 82.7 %, 81.3 %, 79.4 %, and 32.9 %, respectively, compared to the original data and by 77.8 %, 68.1 %, 78.1 %, and 25.8 %, respectively, compared to data with spectral normalization. Leave-one-out cross-validation demonstrated a significant reduction in the absolute relative error, mean absolute error, and root mean square error. The application of DVS-extracted plasma parameters in the DVS-T1 model significantly reduces signal fluctuations and enhances analytical accuracy. This low-cost, efficient approach provides new insights for LIBS development and application.