Maria Lanzinger successfully defended her PhD-Thesis – Congratulations!

Rapid Element Quantification in Metallic Particles via Laser-Induced-Breakdown-Spectrometry

Summary

In the context of high-quality production, quality assurance and process control, the characterization of metal alloys and their respective particles is a commonly applied tool. One example for such an application is technical cleanliness analysis (TCA) in the automotive industry, where more and more sensitive and highly technologized products place increasingly high demands on the product quality. Already small impurities and (particle) contaminations in a size range between roughly 50 and 1000 μm, can lead to serious failures and damage, for example by blocking of mechanical components or short-circuiting of electrical systems. Therefore, limits for particle contamination are tightened in today’s and future’s automotive production to guarantee high-quality and safe goods. Even though TCA is known and executed in the automotive industry for over 20 years, with the shift to electromobility and to more sensitive components, the demand for more accurate and precise analytical tools in TCA increases accordingly.
Currently, the commonly used TCA techniques for material characterization are optical microscopy and automated scanning electron microscopy coupled with energy dispersive X-ray spectroscopy (SEM-EDX). However, these methods only provide information about rough material classes, like metallic-reflective and not metallic-reflective (optical microscopy), and alloy classes, like iron or aluminium based (SEM-EDX). With only this rough material classification the identification of a contamination source is not possible. Only with an in-depth characterisation of the particle, a fast and efficient origin recognition and solution process is enabled. Considering the fast-paced industrial production, also the time necessary for analysis and evaluation is a key factor in the context of TCA.
To increase the analytical performance and reduce analysis time, therefore, increasing efficiency, other methods than SEM-EDX and optical microscopy are under review. In this dissertation, laser-induced breakdown spectrometry (LIBS) studied in this context. LIBS offers a fast measurement, little to no sample preparation as well as high sensitivity for qualitative and quantitative analysis. Additionally, only small sample volumes are necessary for LIBS, enabling TC particle characterization.
In this thesis, the fitness of LIBS for the use in industrially applied particle analysis, like TCA, is studied und improved models for quantitative analysis are built. The quantification models are specifically developed for and applied to the characterization of aluminium alloy particles using the multivariate data analysis algorithm “Partial Least Squares” (PLS) -regression. Employing and training with 49 certified reference materials (CRM) quantification models for ten analytes in aluminium alloy matrices are established. By additionally pre-selecting evaluation wavelengths and implementing sub-calibrations, large concentration ranges are calibrated with low limits of quantification (LOQ). For validation cross-validation as well as samples not included into developing the calibration are utilized. This way, the stability of the models is presented for varying sample matrices.
The developed models are then applied to the analysis of in-house generated reference particles for the use in TCA. As these reference particles are shaved from bulk samples before being pestled and sieved, they are variable in size and composition, easily and economically produceable as well as similar to TC-particles found in the production environment. Using the previously developed calibration models it is shown that (i) the composition of the reference particles is not dependent on their size and (ii) the developed models are fit to be used for the analysis of TCA relevant particles.
Finally, the results of the developed LIBS PLS-regression quantification models are compared to widely established X-ray-based metallic particle characterization methods (SEM-EDX, μ-X-ray-fluorescence-spectrometry: μXRF). Four CRMs in particle form are analysed subsequently with all three methods to directly compare the quantitative results for different aluminium alloy matrices and compositions. Additionally considering the respective analysis volume for each method, the results showcased, a high comparability of LIBS to the established methods, independent of sample composition.
The results of this dissertation showcase the possibilities and advantages of applying LIBS in industrial particle analysis, like TCA. The developed multivariate models are fit to characterize metallic particles in detail, yielding comparable results to widely established analytical techniques with significantly reduced analysis time and sample preparation effort. By developing and optimizing further quantification models, for example for iron or copper matrices, LIBS can be further established as a rapid and reliable analysis method for particle characterization in industrial applications; both as an alternative as well as an addition to established metallic material characterization techniques.