A novel method of PM2.5 sample analysis via organic and elemental carbon (OC/EC) identification based on artificial intelligence and source apportionment

Organic and elemental carbons (OC/EC) are some of the main contributors (20-70%) to particulate matter (PM) mass. Current receptor modeling methods (e.g., Chemical Mass Balance (CMB), Positive Matrix Factorization (PMF)) for source apportionment (SA) are labor-intensive, expensive, and require advanced chemical processes (e.g., ICP-MS). However, the use of Artificial Intelligence (AI) in SA research has recently emerged, though at its early development stage, it still provides competitive results with conventional methods. This study will assess the use of image recognition methods via AI coupled with FTIR measurements to identify the OC/EC composition of atmospheric particles collected on filters.