The detection of trace amounts of residual trinitrotoluene in the environment using surface-enhanced Raman spectroscopy (SERS) can be hindered by the presence of structurally-analogous compounds, which can lead to erroneous spectral assignments. Here, the authors propose a SERS nose array approach integrated with machine learning to enhance the classification accuracy for detecting trinitrotoluene and structurally similar 2,4-DNPA, as well as in distinguishing between gases at different concentrations.
- Peitao Dong
- Haiyang Yang
- Xuezhong Wu