Filtern
Dokumenttyp
Sprache
- Englisch (4) (entfernen)
Schlagworte
- LIBS (4)
- Laser induced plasma (2)
- Plasma diagnostics (2)
- Plasma modeling (2)
- CF-LIBS (1)
- Calibration free LIBS (1)
- Cement (1)
- Cement-based materials (1)
- MC-LIBS (1)
- Particle size (1)
Organisationseinheit der BAM
- 1.9 Chemische und optische Sensorik (4) (entfernen)
Accuracy of calibration-free (CF) methods in laser-induced breakdown spectroscopy (LIBS) depends on experimental conditions and instrumental parameters that must match a CF LIBS model. Here, the numerical study is performed to investigate effects of various factors, such as the optical density, plasma uniformity, line overlap, noise, spectral resolution, electron density and path length on the results of CF-LIBS analyses. The effects are examined one-by-one using synthetic spectra of steel slag samples that fully comply with the mathematical model of the method. Also, the algorithm includes several new features in comparison with previously proposed CF algorithms. In particular, it removes limits on the optical thickness of spectral lines that are used for the construction of the Saha-Boltzmann plot; it retrieves the absorption path length (Plasma diameter) directly from spectral lines; it uses the more realistic Voigt line profile function instead of the Lorentzian function; and it employs the pre-calculated and tabulated thin-to-thick line ratios instead of approximating functions for selfabsorption correction.
The effect of particle grain sizes in different cement-based mixtures on the laser-induced plasma evolution is studied using two experimental methods: (i) temporal and spatial evolution of the laser-induced shock wave is investigated using shadowgraphy and two-dimensional plasma imaging, and (ii) temporal and spatial distribution of elements in the plasma is investigated using two-dimensional spectral imaging. This study is motivated by the interest in applying laser-induced breakdown spectroscopy (LIBS) for chemical analysis of concrete, and subsequently obtain information related to damage assessment of structures like bridges and parking decks. The distribution of grain sizes is of major interest in civil engineering as for making concrete different aggregate grain sizes defined by a sieving curve (64mm to 0.125 mm) are needed. Aggregates up to a size of 180 μm can be excluded from the data set, therefore only the amount of small aggregates with a grain size below 180 μm must be considered with LIBS. All components of the concrete with a grain size smaller than 0.125mm are related to the flour grain content. Tested samples consisted of dry and hardened cement paste (water-cement ratio w/z=0.5), which served as a reference. Aggregate mixtures were made by adding flour grains (size 40 μm) and silica fume (size 0.1 μm) in different ratios to cement: 10%, 30%, 50% and 60%, all combined to the remaining percentage of dry or hydrated cement. The visualization results show that a dependance in the evolution of the plasma as a function of sample grain size can be detected only in the initial stages of the plasma formation, that is, at the initial 3 μs of the plasma life. Spectral information reveals the elemental distribution of the silicon and calcium in plasma, in both neutral and ionized form. Here also, a significant effect is observed in the first 1 μs of the plasma lifetime.
Two calibration-free (CF) LIBS approaches are used for the quantitative analysis of cement samples: the CF-LIBS based on the Boltzmann plot method and the Monte Carlo (MC) LIBS based on the iterative spectrum fitting. In CF-LIBS, the inverse problem is solved, i.e. the elemental concentrations are determined by the reconstruction of plasma parameters from spectra. The MC-LIBS technique solves the direct problem by finding the highest correlation between the model-generated and experimental spectrum. The accuracy of both calibration-free LIBS methods suffers from factors such as inaccurately determined instrumental function, the deviation of experimental plasma from the mathematical model used, not taking into account the collection geometry, and from the uncertainty of spectroscopic data. The both calibration-free LIBS approaches are first applied to synthetic spectra which perfectly suit the mathematical model of the method, i.e. the model of the uniform, isothermal, and stationary plasma. This test yields the accuracy of both the approaches for the ideal case. In addition, the accuracy of both the methods is investigated for non-uniform and non-isothermal plasma, because real laser-induced plasma often has high gradients in temperature and particle number densities. Finally, both calibration-free LIBS approaches are applied to experimental spectra obtained from cement samples. The figures of merits of two approaches are compared when working with both synthetic and experimental spectra.
Two calibration-free LIBS techniques are used for the quantitative analysis of synthetic cement samples: the CF-LIBS based on the Boltzmann plot method and the Monte Carlo (MC) LIBS based on the iterative spectrum fitting. In CF-LIBS, the inverse problem is solved, i.e. the elemental concentrations are determined by the reconstruction of plasma parameters from spectra. The MC-LIBS technique solves the direct problem by finding the highest correlation between the model-generated and experimental spectrum. The accuracy of both calibration-free LIBS methods suffers from factors such as inaccurately determined instrumental function, the deviation of experimental plasma from the mathematical model used, not taking into account the collection geometry and from the uncertainty of spectroscopic data. Therefore, the both calibration-free LIBS approaches are applied to synthetic spectra which perfectly suit the mathematical model of the method. This test yields the accuracy of both the approaches for the ideal case. In addition, the accuracy of both methods is investigated for non-isothermal plasma, because real laser-induced plasma often has high gradients in temperature. Both methods assume an isothermal plasma.