Abstract

Poster Presentations

Day 1, June 10（Mon.）　　Room P1 (Multipurpose Hall)・Room P2 (Conference Room 101+102)

Pyrolysis-GC-MS (Py-GC/MS) is widely used for qualitative analysis of polymeric materials. When polymeric materials are measured by Py-GC/MS, many pyrolyzates derived from the polymers are observed. Because some of pyrolyzates are not registered in commercially available EI mass-spectral libraries, it is difficult to identify those compounds by library search. We attempted to solve the problem of pyrolyzates not registered in mass-spectral libraries by constructing a virtual mass-spectral library by combining a method of calculating the pyrolysis reaction of polymers with a method of predicting EI mass spectra using machine learning. Computational pyrolysis reaction of polymers is carried out as follows: first, an oligomer in which six monomers are connected is created. Next, the oligomer is computationally fragmented. Since many simple cleavages were reported in literature, we focused on simple cleavages. The EI+ mass spectra were predicted from the structural formulas of in silico pyrolyzates using a machine-learning model. The structural formulas and predicted EI mass spectra of the pyrolyzates were registered in a virtual mass-spectral library.

---