Analysis of Fatty Acids by High Performance Liquid Chromatography and Electrospray Ionization-Mass Spectrometry
Not peer reviewed
MetadataShow full item record
Fatty acids (FA) have been traditionally analyzed by gas chromatography (GC) as fatty acids methyl esters (FAME) and more recently using mass spectrometry (MS) detection. Since high performance liquid chromatography (HPLC) presents some advantages like the possibility to analyze them as underivatized compounds, the purpose of this work has been to investigate to which extent HPLC-MS can be a replacement or a complement technique to GC-MS. A direct infusion (DI)-MS and an HPLC-MS method to analyze FAs were developed. Fragment diagnostic ions used for structure elucidation, are usually obtained when FAMEs are analyzed by GC-MS with electron ionization. When FAs were analyzed by HPLC-MS with electrospray ionization, this technique gave almost no fragmentation and no adducts even with collision induced. HPLC-MS therefore provides information about the molecular mass, which is often missing in GC-MS. A limitation found with HPLC-MS is that it was not possible distinguish between some isomers, which for quantification purposes limit the use of the technique to cases where no separation of isomers is needed. It was also noticed that fatty acids of different chain length have different ionization efficiencies and these depends in some extent on the mobile phase used. Chromatographic selectivity, efficiency and retention were also investigated applying HPLC-MS. These parameters can be explained by Purnell and van Deemter equations in isocratic and isothermal chromatography. Since the retention factor (k) and number of theoretical plates (N) are not valid concepts in programmed chromatography, equivalent chain length (ECL) and peaks per carbon (PPC) were the parameters used to explain selectivity and efficiency, respectively, by HPLC with gradient elution. The variability of ECL with different chromatographic conditions (methanol, acetonitrile, acetone or tetrahydrofuran in the mobile phase, temperature and gradient time) was studied, applying factorial design and response surface methodology to build models to predict ECL. Root mean squared errors for predictions (RMSE) were below 0.04 for all the solvents analyzed, which resulted in less than 10% of a peak width. It was also found that ECL varies with the selection of the solvent and to some degree with the temperature, and that gradient time (steepness of the gradient) has almost no effect. Partial least square regression (PLSR) was also applied to build models to predict ECL based on the chemical structure of the molecule and based on GC retention data. Again, good prediction models were found with errors that were a fraction of a peak width. The PPC concept was used as a measure of efficiency and is defined as the inverse of peak width in retention index units. The highest efficiency was obtained when methanol was used as solvent. Efficiency can be improved by decreasing column temperature or increasing gradient time, which results in higher time of analysis. The maximum value for PPC obtained by HPLC-MS was around 7.
PublisherThe University of Bergen
Copyright the Author. All rights reserved