TY - JOUR A1 - Gottlieb, Cassian A1 - Millar, Steven A1 - Grothe, S. A1 - Wilsch, Gerd T1 - 2D evaluation of spectral LIBS data derived from heterogeneous materials using cluster algorithm JF - Spectrochimica Acta Part B N2 - Laser-induced Breakdown Spectroscopy (LIBS) is capable of providing spatially resolved element maps in regard to the chemical composition of the sample. The evaluation of heterogeneous materials is often a challenging task, especially in the case of phase boundaries. In order to determine information about a certain phase of a material, the need for a method that offers an objective evaluation is necessary. This paper will introduce a cluster algorithm in the case of heterogeneous building materials (concrete) to separate the spectral information of non-relevant aggregates and cement matrix. In civil engineering, the information about the quantitative ingress of harmful species like Cl−, Na+ and SO2−4 is of great interest in the evaluation of the remaining lifetime of structures (Millar et al., 2015; Wilsch et al., 2005). These species trigger different damage processes such as the alkali-silica reaction (ASR) or the chloride-induced corrosion of the reinforcement. Therefore, a discrimination between the different phases, mainly cement matrix and aggregates, is highly important (Weritz et al., 2006). For the 2D evaluation, the expectation-maximizationalgorithm (EM algorithm; Ester and Sander, 2000) has been tested for the application presented in this work. The method has been introduced and different figures of merit have been presented according to recommendations given in Haddad et al. (2014). Advantages of this method will be highlighted. After phase separation, non-relevant information can be excluded and only the wanted phase displayed. Using a set of samples with known and unknown composition, the EM-clustering method has been validated regarding to Gustavo González and Ángeles Herrador (2007). KW - LIBS KW - Concrete KW - EM-clustering KW - Heterogeneity PY - 2017 DO - https://doi.org/10.1016/j.sab.2017.06.005 SN - 0584-8547 VL - 2017 IS - 134 SP - 58 EP - 68 PB - Elsevier B.V. AN - OPUS4-40644 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER -