Scientists from Hokkaido University in Sapporo, Japan, are examining new more sustainable techniques detecting arsenic in minerals, publishing their findings in the journal Scientific Reports (1).
As the push for more sustainable energy grows, mineral-intensive technologies, such as renewable energy and electric vehicles, will be in high demand. However, the mining sector, especially copper mining, faces considerable difficulties due to growing demand, depletion of high-grade ore, and the rise of high-arsenic copper resources. These challenges not only make mineral processing more difficult, but it can also cause environmental and health concerns due to potential arsenic presence in wastewater and exhaust gas. There is a correlation between arsenic exposure and a range of health problems. Arsenic is a ranked as a toxic substance by the Agency for Toxic Substances and Disease Registry. However, there are 100 diverse species of arsenic in existence with diverse toxicological profiles (2). In fact, long-term exposure to inorganic arsenic through sources like water and food can lead to chronic arsenic poisoning. Symptoms of this can include vomiting, abdominal pain, skin lesions, cancer, or death (3).
With how dangerous arsenic exposure can be for human health, it is vital that ecologically viable methods are developed for the mining sector. In this study, the scientists investigated the feasibility of utilizing hyperspectral imaging combined with machine learning (ML) techniques for the identification of arsenic-containing minerals in copper ore samples. Specifically, they focused on practical application in sorting and processing operations. Hyperspectral data analysis is when a type of data that contains information about a surface is measured (5). Hyperspectral sensors, more commonly known as “imaging spectrometers,” collect spectral information across a continuous spectrum by dividing the spectrum into many narrow spectral bands (6).
For this study, the team experimented with various copper sulfide ores, and used neighborhood component analysis (NCA), which is a machine learning algorithm for metric learning. NCA learns a linear transformation in a supervised fashion to improve the classification accuracy of a stochastic nearest neighbor’s rule in the transformed space. This method was employed to select essential wavelength bands from hyperspectral data (4). The selected bands were subsequently used as inputs for machine learning algorithms to identify arsenic concentrations.
The results showed that by selecting a subset of informative bands using NCA, accurate mineral identification can be achieved with a significantly reduced the size of dataset, enabling efficient processing and analysis. When compared to other wavelength selection methods, NCA was deemed superior regarding the optimization of classification accuracy. Namely, the identification accuracy showed 91.9% or more when utilizing eight or more bands selected by NCA and was comparable to hyperspectral data analysis with 204 bands. These findings show potential for cost-effective implementation of multispectral cameras in mineral processing operations. The scientists also hope to research this situation further, with potential directions including refining machine learning algorithms, exploring broader applications across diverse ore types, and integrating hyperspectral imaging with emerging sensor technologies for enhanced mineral processing capabilities.
(1) Okada, N.; Nozaki, H.; Nakamura, S.; et al. Optimizing Multi-Spectral Ore Sorting Incorporating Wavelength Selection Utilizing Neighborhood Component Analysis for Effective Arsenic Mineral Detection. Scientific Reports 2024, 14, 11544. DOI: 10.1038/s41598-024-62166-0
(2) Hroncich, C. Using HPLC-ICP-MS for Arsenic Speciation in Freshwater Fish. MJH Life Sciences 2024. https://www.spectroscopyonline.com/view/using-hplc-icp-ms-for-arsenic-speciation-in-freshwater-fish (accessed 2024-6-11)
(3) Arsenic. World Health Organization 2022. https://www.who.int/news-room/fact-sheets/detail/arsenic (accessed 2024-6-11)
(4) Neighborhood Components Analysis. Scikit-learn Developers 2024. https://scikit-learn.org/stable/modules/generated/sklearn.neighbors.NeighborhoodComponentsAnalysis.html (accessed 2024-6-11)
(5) Hyperspectral Data. Elsevier B.V. 2024. https://www.sciencedirect.com/topics/computer-science/hyperspectral-data (accessed 2024-6-11)
(6) Basic Hyperspectral Analysis Tutorial. NV5 Geospatial Solutions, Inc. 2024. https://www.nv5geospatialsoftware.com/docs/HyperspectralAnalysisTutorial.html (accessed 2024-6-11)