Decomposition of electron ionization mass spectra for space application using a Monte-Carlo approach.

Abstract

Quadrupole mass spectrometers equipped with an electron ionization (EI) sources have been widely used in space exploration to investigate the composition of planetary surfaces and atmospheres. However, the complexity of the samples and the minimal calibration for the fragmentation of molecules in the ionization chambers have prevented the deconvolution of the majority of the mass spectra obtained at different targets, thus limiting the determination of the exact composition of the samples analyzed. We propose a Monte-Carlo approach to solve this issue mathematically. We decomposed simulated mass spectra of mixtures acquired with unit resolving power mass spectrometers and EI sources into the sum of the single components fragmentation patterns weighted by their relative concentration using interior-point least-square fitting. To fit compounds with poorly known fragmentation patterns, we used a Monte-Carlo method to vary the intensity of individual fragment ions. We then decomposed the spectrum thousands of times to obtain a statistical distribution. By performing the deconvolution on a mixture of seven different molecules with interfering fragmentation patterns (H2 O, O2 , CH4 , Ar, N2 , C2 H4 , and C2 H6 ) we show that this approach retrieves the mixing ratio of the individual components more accurately than regular mass spectra decomposition methods that rely on fragmentation patterns from general databases. It also provides the probability density function for each species's mixing ratio. By removing the solution degeneracy in the decomposition of mass spectra, the method described herein could significantly increase the scientific retrieval from archived space flight mass spectrometry data, where calibration of the ionization source is no longer an option.