The long–short double-pulse laser-induced breakdown spectroscopy (LS-DP-LIBS) method was applied to qualitative and quantitative analyses of underwater steel samples to improve the detection ability of underwater measurement. The stable plasma intensity and discrete emission lines were detected using LS-DP-LIBS when comparing the measured results of single-pulse LIBS (SP-LIBS) and LS-DP-LIBS. The long pulse produces a cavitation bubble without plasma, and the short pulse induces the plasma of steel samples within the bubble. The detection features of LS-DP-LIBS for underwater steel samples were discussed in different intra-pulse delay time, long-pulse width, and delay time conditions when analyzing the measured spectra, the signal intensity of Fe(I) at 400.524 nm and 402.187 nm, Mn(I) at 404.136 nm, and intensity ratio of Mn(I) 404.136 nm/Fe(I) 402.187 nm. The results indicated that the plasma stability and spectral signal intensity were improved significantly with a long-pulse width of 80 µs in the intra-pulse delay time of 70 µs, which were appropriate for bubble formation and plasma generation. According to the discussion of the delay time effect, the state of generated plasma was almost stable from 650 ns to 850 ns. Manganese (Mn) contents in steel samples were analyzed quantitatively when measuring five steel samples with different Mn contents using LS-DP-LIBS in optimal experimental conditions. A strong linear dependence was observed with R2=0.9842, which demonstrated the feasibility and appropriateness of quantitative analysis for underwater measurement using LS-DP-LIBS.

No abstract

The influence of resolution on spectral analysis is of great significance to improve the measurement accuracy of laser-induced breakdown spectroscopy (LIBS). In this study, the low alloy steel samples were measured at different resolutions using dual-channel spectrometer simultaneously to discuss the plasma characteristics. The diffraction efficiency of the grating was different at different resolutions, which led to different spectral intensities measured at different resolutions. It affected the LIBS spectral analysis. For plasma spectral analysis, the experimental results showed that the Boltzmann plot method was suitable for calculating the plasma temperature using the low-resolution spectra with 0.076 nm/pixel. Based on the high-resolution spectra with 0.01 nm/pixel, Boltzmann double lines method was applied to calculate the plasma temperature, and the analysis line was IFeI395.668/IFeI400.524. Due to the influence of instrument broadening, Stark broadening could only be used to characterize the electron density using the high-resolution spectra. For quantitative analysis in LIBS spectral analysis, support vector machine regression (SVR) with different inputs was used to quantitatively analyse the Mn content in the low alloy steel. The prediction accuracy of the low-resolution spectra was poor compared to the high-resolution spectra. When the input was the target spectral intensities and the plasma state, the fitting accuracy and prediction accuracy were improved. It showed that SVR combined with the plasma state was an effective method to improve the accuracy of LIBS quantitative analysis of Mn content in low alloy steel.