Multivariate determination of 10B isotopic ratio by laser-induced breakdown spectroscopy using multiple BO molecular emissions
| Autor: | Lewis Johnson, Luke Helms, Luisa T. M. Profeta, Codjo A. Akpovo |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
010302 applied physics
Multivariate statistics Materials science Isotope Mean squared error 010401 analytical chemistry Analytical chemistry Isotopes of boron 01 natural sciences Atomic and Molecular Physics and Optics 0104 chemical sciences Analytical Chemistry Wavelength 0103 physical sciences Figure of merit Laser-induced breakdown spectroscopy Emission spectrum Instrumentation Spectroscopy |
| Zdroj: | Spectrochimica Acta Part B: Atomic Spectroscopy. 162:105710 |
| ISSN: | 0584-8547 |
| DOI: | 10.1016/j.sab.2019.105710 |
| Popis: | Measurements of boron isotopic ratios have been investigated using LIBS, and combined LIBS and Molecular laser-induced fluorescence (MLIF). Mixtures of various ratios of research grade H310BO3 and H311BO3 in pellet form were interrogated with a Q-switched Nd:YAG laser operating at 532 nm. The resulting plasma was examined for their BO molecular bands in the 254–262 nm and 266–276 nm wavelength ranges. A minimum of three band heads of the β system B2Σ → X2Σ transitions were identified. The rovibronic isotopic shifts of the corresponding bands were measured and were conclusively compared to the literature. MLIF was applied to LIBS emission lines covering 253–271 nm to enhance some band heads, selectively. Both LIBS and LIBS-MLIF emissions were subject to multivariate statistical analysis to predict 10B isotopic ratios. Two PLS regression calibration models were examined to better examine the effects, sensitivity, accuracy of the laser-produced plasma excitation modalities based models through the figure of merit. For the full calibration set (21 samples), the root mean square error of cross-validation (RMSECV), the pseudo univariate LOD (LODpu) and the LOD interval ([LODmin, LODmax]) for 10B isotopes were improved from 1.61%, 4.81%, [2.64%, 3.49%] 10B isotopic ratio, respectively for LIBS to 0.98%, 2.40%, [2.19%, 2.81%] 10B isotopic ratio, respectively for LIBS-MLIF. However, using a random subset of the samples (14) as a calibration set and the rest as a test set, the root mean square error of prediction of 10B isotopic ratio in the test set improved from 2.95% to 1.16% 10B isotopic ratio, respectively for LIBS and LIBS-MLIF; and the LOD interval improved from [2.45%, 2.69%] to [1.88%, 2.12%] 10B isotopic ratio when calculated with cross-validated residual variance. Moreover, the prediction error of a test set improved from 2.95% for LIBS to 1.16% for LIBS-MLIF. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect