Detriment of air quality is one of the consequences of the rapid growth of different pollution sources. In recent years, research has been challenged in developing new monitoring methods that can identify pollutants efficiently. Laser Ablation Mass Spectrometry (LAMS) is a technique that has shown the potential to determine chemical speciation of airborne particles in real time [1]. However, in order to full exploit the online nature of the technique, it needs to be coupled with algorithms that can deal with the wealth of information produced by the system. In addition, results require cautious analysis to provide significant information on atmospheric conditions.
One of the algorithms that best suits the inherent features of the LAMS technique is the Adaptive Resonance Theory (ART) based neural networks. Artificial neural networks have been successfully implemented in chemical applications where the systems are non-linear in nature [2]. As Hopke demonstrated in his work [3], a variation of the ART family, the ART-2a neural network, performed well when used to classify mass spectra data in post-acquisition analysis.
The focus of t his work is the interface LAMS and ART-2a to perform online classification of mass spectra. The overall structure of the system is briefly discussed as it represents the necessary layout to accomplish this goal. The ultimate aim of this study is to contribute in developing a fully automated aerosol analysis station.