Laser-induced breakdown spectroscopy analysis of steel structures : an improved single-shot quantitative analysis technique to investigate porous and non-adhesive layers related to the early stage rust behaviour in mild steel

A previously used technique on particulate materials was adopted on composite solid steel structures to determine surface concentrations. The technique was improved with modelled differential order fits that adjusted to the characteristic shot-to-shot spectra variations. This was to further establish the practicability of appropriate, quantifiable parameters, to characterise porous and non-adhesive layers related to the complex early stage rust behaviour in mild steel, using single-shot Laser-Induced Breakdown Spectroscopy (LIBS) technique. In comparison to the mean baseline spectra fit, the quadratic and linear trend fits provided suitable baseline value variations, with the analyte signal, that corresponded to analyte concentrations. The most improved results were obtained from the quadratic fits. This was as a result of adequate and characteristic higher variations obtained from the quadratic reference baseline fit. The significant intrinsic and shot-to-shot variations were eliminated from the successive plots to provide accurate results. The quadratic baseline fit showed improved values when compared to the mean and linear fits with smaller calibration curve assessment measures, particularly, lowest LOD values which provided better validation and accuracy in predicting low surface concentrations related to loss of material and leaching. The reduction was indeed recognised with the quadratic baseline fit's competency in adjusting the intrinsic and shot-to-shot variations from the successive plots. The related calibration curve also showed the lowest statistical error and most consistent measures for both low and high analyte concentrations. This was useful in properly analysing surface concentrations related to inherent scale layers and/or leaching dynamics related to rust. The lesser analyte concentration measures associated with rust mild steel sample profiles provided evidence of leaching and surface quantifiable losses. These were assumed with previous semi-quantitative findings. The quadratic fit related calibration curve atypically demonstrated lesser uncertainties in concentration predication for a reduced sensitivity in comparison with the mean and linear fits. The ideal factor for a reduced sensitivity in the Plᵢ vs Rbᵢ plots was about 2.1 to produce low and consistent statistical errors and measures for all baseline fits. Comparing signal intensity parameters for both surfaces showed a noticeable distinction, although, temperature variations were minimal with good precision observed for the temperature measurements. This was as a result of the bulk matrix line intensities used for the Boltzmann plots in measuring the plasma temperature as they revealed lesser variations compared to the analyte signals. The effect of pulse energy on the variations and behaviours relating to bulk matrix and analyte signal intensities and optimum pulse energy requirements for reduced effect of bulk matrix and analyte signal behaviour interaction were also previously investigated.