Using Chemical Ionisation Mass Spectrometry to Measure Oceanic Emissions of Trace Gases

In this thesis the capability of chemical ionisation mass spectrometry (CIMS) to measure emissions of trace gases.

Rate constants of several oxygenated volatile organic compounds (OVOCs) and benzene are reported for H3O+, NO+ and O2+ reagent ions in nitrogen carrier gas, using a permeation tube calibration source. Higher fragmentation was observed due to changes to the lens voltages of the instrument to tune for higher sensitivity, altering the reagent ion energies. The impact of increasing flow tube temperature and voltage on the sensitivity and ionisation of benzene, butanone and butanal was shown under dry and humid conditions. It is demonstrated that a higher flow tube temperature reduces the water adduct formation, simplifying compound concentration determination.

The iodide dependence of the production of methyl iodide is debateable under natural conditions as studies that report this dependence (Y. Chen et al., 2020; Moore and Zafiriou, 1994) use iodide levels toward the top of the natural range. Iodide dependent reactions were measured in only two samples, PYSML1 and PFSML3. Further work is required to ascertain the cause of this reaction, though it is known that it is restricted the SML and only currently occurs in estuarine Plymouth samples.

PTR-ToF-MS is used to measure emissions from seawater samples exposed to both light and ozone. Significant emissions (37.7 – 40.3 TgC yr-1) of VOCs were measured from photolysis of the seawater surface. These were higher than those reported in literature but could potentially be explained through the sample composition being unrepresentative of average global composition. Future work is required to confirm this magnitude of VOCs to the marine atmosphere and to provide more complete quantification and identification of the species produced. This would allow for better modelling of SOA formation, OH reactivity and concentration of VOCs in the marine atmosphere that are all often underestimated in models.