Automated methods for background estimation in laser spectroscopy

A new automated method for spectral background estimation in laser spectroscopy ensures the accuracy of quantitative analysis with minimal human intervention.

When using laser-induced breakdown spectroscopy in spectral analysis, scientists may encounter various obstacles. The most common challenge faced by scientists when conducting elemental analysis is to optimize the interaction between the laser and the sample, pay attention to changes in laser energy, and the convergence of environmental noise, which helps to create different backgrounds in the collected spectra. All these obstacles will have a significant impact on the analysis.

In a recent study published in the Journal of Spectroscopy Part B: Atomic Spectroscopy, a research group from Jiangnan University introduced a new LIBS method aimed at automatically estimating and removing different spectral backgrounds. Under the leadership of Chen Hao from the School of Mechanical Engineering at Jiangnan University, researchers proposed a method that utilizes window functions, differential concepts, and piecewise cubic Hermite interpolation polynomials.

In this experiment, Chen and his team conducted a series of simulation experiments to evaluate background correction methods. They found that their proposed method performs better than existing techniques such as asymmetric least squares and modelless background correction. By utilizing window functions, Pchip, and differential concepts, the new method improves the ability to eliminate white noise and baseline distortion, achieving a better signal-to-noise ratio than previous methods.

The research team also found that their method improved the processing of background baseline jumps.
The researchers applied their method to seven different aluminum alloys and observed a correlation between spectral intensity and magnesium concentration.

It is worth noting that in the experiment of measuring magnesium concentration in aluminum alloys, the correlation coefficient between predicted concentration and actual concentration significantly improved after correction.

The coefficients for ALS and model free methods are 0.9913 and 0.9926, respectively, while the coefficients for this new method have decreased from the initial 0.9943 to 0.9154.

These findings not only validate the effectiveness of this automated method, but also pave the way for future research to improve the accuracy of LIBS spectral analysis.

Source: Laser Network