Adaptive LIBS analysis for estimating concentrations of alloy elements in heterogeneous metal scrap recycling streams

Metal scraps pose an economic and ecologically viable source for secondary resource supply to our industries, which call for more independence from global crises and strategic uncertainties. Well advanced technologies exist for steel and aluminum based on mechanical sorting using basic physical properties in order to split the major Fe- and Al rich fractions. However, many high-tech products require a precise composition specified by narrow acceptable ranges of alloy elements to achieve distint performances of a given alloy type. Here, traditional recycling stream processing bears limitations due to the generation of sorting fractions that contain mixes of variable alloy types, both, in steel as well as aluminum sorting products. Metallurgical processing of such mixed alloys, especially mixed aluminum alloys, leads to lower quality metals with less defined performance specifications and hence, the material is then lost for high-tech industries as a secondary resource. A more detailed, quantitative identification of specific alloy elements provides a solution, which allows for the differentiation between and consequent separation of alloy types. Here, laser-induced breakdown spectroscopy (LIBS) has shown enormous potential for trace (alloy) element detection. The remaining challenge or limitation lies in the strong matrix dependence of LIBS. This means, that a well pre-defined and homogeneous material stream is required for the accurate application of LIBS for element quantification and associated alloy identification.

We propose a hierarchical system to adapt LIBS analysis in a flexible way to the requirements of heterogeneous scrap recycling streams. We developed a clustering method to first identify the metal type, steel or aluminum, in mixed recycling products. The identified metal type provides the information on matrix conditions. Using then the respective calibration model for this matrix condition allows estimating precise alloy element concentrations in order to identify the alloy type. In repetition experiments, we could document high accuracies and precisions for specific diagnostic alloy elements, while few others show medium accuracies and precisions. The complementary information of elemental concentrations provides solid ground for an improved alloy detection and strategically points towards further options for dynamic thresholds in scrap processing procedures.