Polybrominated diphenyl ethers (PBDEs) and novel brominated flame retardants (NBFRs) are widely used as important flame retardants in electronic products such as copper laminates. In this study, a simple Bayesian model combined with the XGBoost algorithm was used to analyze the contamination characteristics of PBDEs (polybrominated diphenyl ethers) and NBFRs (novel brominated flame retardants) in the sorting residues of copper laminates. GC/MSD was applied to detect and analyze the residues from nine sampling sites. The results showed that BDE99 was detected at all monitoring sites with a detection frequency of 100%, and its concentration ranged from 0.158 to 0.498 ng/L, with the highest concentration of 1.657 ng/L at site H1. Among the eight PBDEs monomers detected, the detection rate of BDE47, BDE99, BDE100, and BDE209 was 100%, and the Σ7PBDEs the contents ranged from 9.166~88.326ng·g-1, and the median value was 29.092ng·g-1. Among the novel brominated flame retardants, the detection rate of BPA was 84.648%, and the detection rates of BPB and BPAF were both 61.548%. Correlation analysis showed that there was a significant positive correlation between BPA and BPAF (r=0.54, p=0.048<0.05). The time trend analysis showed that the ratio of Σ26PBDEs/Σ5NBFRs showed a decreasing trend from 4.233 in 2011 to 2.073 in 2021, which indicated that the new brominated flame retardants were gradually replacing the traditional PBDEs.The machine-learning based analysis method effectively identified the main controlling factors of the contamination characteristics, and provided scientific basis for the management of the contaminated sites and the control of risks.
