| Timetable |
| Download Conference Program |
| Download All Abstracts |
| Zoom Access |
| Corporate Program |
Poster Presentations
Day 3, June 24（Tue.）
Room P (Maesato East, Foyer, Ocean Wing)
- 3P-PM-20
Non-Targeted Analysis of Air Pollutants Using Thermal Desorption GC-HRTOFMS with Machine Learning Structural Analysis
(1JEOL Ltd., 2TOYO UNIVERSITY)
oMasahiro Hashimoto1, Chihiro Ueno2, Ryotaro Suzuki2, Katsuhito Yoshida2, Atsuyuki Sorimachi2, Masaaki Ubukata1
Electron ionization(EI) is the most commonly used in GC-MS, and EI mass spectral databases are often used to identify components. However, the molecular ion is often weak or not observed in EI data, making it difficult to identify components not registered in the database by EI alone. On the other hand, soft ionization(SI) is an effective ionization method for observing the molecular ion that is difficult to observe with EI. Therefore, SI and HRTOFMS analysis are effective for the analysis of unknown components. Additionally, we have recently developed a new workflow to predict the EI mass spectrum using machine learning. In this study, we utilized this new workflow to the non-targeted analysis of air pollutants. Samples were collected on a rooftop at Toyo University and analyzed by GC-HRTOFMS. 48 component peaks were detected, including additives such as phthalates. 16 components were not registered in the database and were analyzed using a new workflow, suggesting that they are terpenes and steroids of plant origin. Understanding these components in the atmosphere is expected to contribute to the suppression of air pollutants. Thus, GC-HRTOFMS and machine learning structural analysis were shown to be useful for the non-targeted analysis of air pollutants.