Predicting in silico electron ionization mass spectra using quantum chemistry.

Area of Science:

• Computational Chemistry
• Spectroscopy
• Chemical Informatics

Background:

• Mass spectrometry is crucial for compound identification, but curated electron ionization (EI) mass spectral databases are limited.
• Existing databases like NIST and MoNA contain fewer than 300,000 EI spectra, despite over 102 million compounds in PubChem.

Purpose of the Study:

• To evaluate quantum chemistry methods combined with molecular dynamics (QCEIMS) for generating in silico EI mass spectra.
• To assess the accuracy of predicted mass spectra against experimental data for diverse chemical classes.

Main Methods:

• Generated in silico EI mass spectra for 451 small molecules using QCEIMS.
• Employed molecular dynamics (MD) simulations and statistical methods.
• Compared predicted spectra to experimental data from the NIST 17 mass spectral library, optimizing parameters like initial temperature and impact excess energy (IEE).

Main Results:

• QCEIMS successfully predicted 70 eV EI mass spectra from first principles for tested compounds.
• Prediction accuracy varied by chemical class, with organic oxygen compounds showing lower matching accuracy.
• Computation time increased exponentially with molecular size; conformational flexibility did not impact prediction accuracy.

Conclusions:

• QCEIMS is a viable method for generating EI mass spectra computationally.
• Further advancements in calculating potential energy surfaces are needed for large-scale generation of spectra for novel molecules.