| Abstract [eng] |
Laser-induced breakdown spectroscopy (LIBS) is a rapid multi-element analysis technique requiring minimal sample preparation. Femtosecond LIBS is especially attractive for geological and paleoenvironmental investigations due to its localized, predominantly non thermal ablation and reduced matrix effects. However, detection of low-level trace elements remains challenging because ultrafast laser-induced plasmas are short-lived and weakly emissive. While ultrafast double-pulse LIBS has already demonstrated promising signal enhancement capabilities, GHz burst excitation for LIBS remains comparatively less explored. The aim of this work was to optimize femtosecond double-pulse LIBS experimental conditions for ppm-level trace-element detection in geological paleoenvironmentally significant samples and compare its performance with GHz burst excitation. The influence of double-pulse delay, pulse energy distribution, plasma gating conditions, and surrounding gas atmosphere on LIBS signal-to-noise ratio was investigated in order to determine conditions favorable for plasma reheating. Furthermore, the applicability of the developed setup for paleoenvironmental investigations was assessed by comparing LIBS spectral trends and elemental ratios with reference LA-ICP-MS measurements from different stratigraphic depths. The results demonstrated that independently controlled fs DP-LIBS achieved higher signal enhancement than GHz burst excitation under equal total energy conditions, with best performance obtained when more energy was supplied in the second pulse, confirming efficient plasma reheating. Fs DP-LIBS achieved enhancement factors of ~2.5, while commercial GHz burst excitation reached ~1.5. Argon atmosphere further increased SNR by ~1.6 compared to air. A broad optimum region of 300-450 ps double-pulse delay and 200-250 ns gating delay indicated robust operation. Correlation with LA-ICP-MS showed strong agreement, reaching R2 ~ 0.9. Cross-validation confirmed more realistic semi-quantitative performance with reduced, but still meaningful predictive capability. |
