Enhancement of spectral model transferability in LIBS systems through LIBS-LIPAS fusion technique

Jiayuan Zhou; Lianbo Guo; Mengsheng Zhang; Weihua Huang; Guangda Wang; Aojun Gong; Yuanchao Liu; Harse Sattar

doi:10.1016/j.aca.2024.342674

Enhancement of spectral model transferability in LIBS systems through LIBS-LIPAS fusion technique

Anal Chim Acta. 2024 Jun 22:1309:342674. doi: 10.1016/j.aca.2024.342674. Epub 2024 May 4.

Authors

Jiayuan Zhou¹, Lianbo Guo², Mengsheng Zhang¹, Weihua Huang¹, Guangda Wang¹, Aojun Gong¹, Yuanchao Liu³, Harse Sattar⁴

Affiliations

¹ Wuhan National Laboratory for Optoelectronics (WNLO), Huazhong University of Science and Technology, Wuhan, Hubei, 430074, China.
² Wuhan National Laboratory for Optoelectronics (WNLO), Huazhong University of Science and Technology, Wuhan, Hubei, 430074, China. Electronic address: lbguo@hust.edu.cn.
³ Department of Physics, City University of Hong Kong, Kowloon, 999077, Hong Kong SAR, China.
⁴ School of Integrated Circuits, Huazhong University of Science and Technology (HUST), Wuhan, Hubei, 430074, China. Electronic address: harissattar@hust.edu.cn.

PMID: 38772657
DOI: 10.1016/j.aca.2024.342674

Abstract

Background: Laser-induced breakdown spectroscopy (LIBS) is extensively utilized a range of scientific and industrial detection applications owing to its capability for rapid, in-situ detection. However, conventional LIBS models are often tailored to specific LIBS systems, hindering their transferability between LIBS subsystems. Transfer algorithms can adapt spectral models to subsystems, but require access to the datasets of each subsystem beforehand, followed by making individual adjustments for the dataset of each subsystem. It is clear that a method to enhance the inherent transferability of spectral original models is urgently needed.

Results: We proposed an innovative fusion methodology, named laser-induced breakdown spectroscopy fusion laser-induced plasma acoustic spectroscopy (LIBS-LIPAS), to enhance the transferability of support vector machine (SVM) original models across LIBS systems with varying laser beams. The methodology was demonstrated using nickel-based high-temperature alloy samples. Here, the area-full width at half maximum (AFCEI) Composite Evaluation Index was proposed for extracting critical features from LIBS. Further enhancing the transferability of the model, the laser-induced plasma acoustic signal was transformed from the time domain to the frequency domain. Subsequently, the feature-level fusion method was employed to improve the classification accuracy of the transferred LIBS system to 97.8 %. A decision-level fusion approach (amalgamating LIBS, LIPAS, and feature-level fusion models) achieved an exemplary accuracy of 99 %. Finally, the adaptability of the method was demonstrated using titanium alloy samples.

Significance and novelty: In this work, based on plasma radiation models, we simultaneously captured LIBS and LIPAS, and proposed the fusion of these two distinct yet origin-consistent signals, significantly enhancing the transferability of the LIBS original model. The methodology proposed holds significant potential to advance LIBS technology and broaden its applicability in analytical chemistry research and industrial applications.

Keywords: Laser-induced breakdown spectroscopy; Laser-induced plasma acoustic spectroscopy; Plasma acoustic-spectrum fusion; Spectral non-transferability; Spectral radiation model.