Abstract

Mapping with laser-induced breakdown spectroscopy (LIBS) can offer more than just the spatial distribution of elements: the rich spectral information also enables mineral recognition. In the present study, statistical approaches were used for the recognition of the spodumene from lithium pegmatite ores. A broad spectral range (280–820 nm) with multiple lines was first used to establish the methods based on vertex component analysis (VCA) and K-means and DBSCAN clusterings. However, with a view to potential on-site applications, the dimensions of the datasets must be reduced in order to accomplish fast analysis. Therefore, the capability of the methods in mineral identification was tested with a limited spectral range (560–815 nm) using Li-pegmatites with various mineralogical characters.

摘要

使用激光诱导击穿光谱 (LIBS) 进行测绘不仅可以提供元素的空间分布：丰富的光谱信息还可以识别矿物。在本研究中，统计方法用于识别锂伟晶岩矿石中的锂辉石。首先使用具有多条线的宽光谱范围 (280–820 nm) 来建立基于顶点分量分析 (VCA) 和 K-means 和 DBSCAN 聚类的方法。然而，考虑到潜在的现场应用，必须降低数据集的维度以完成快速分析。因此，使用具有各种矿物学特征的锂伟晶岩在有限的光谱范围（560-815 nm）内测试了这些方法在矿物鉴定中的能力。