Fast Optimization of Supercritical Fluid Chromatography-Mass Spectrometry Interfacing Using Prediction Equations
Abstract
The effect of makeup solvent composition in ultrahigh-performance supercritical fluid chromatography-triple quadrupole mass spectrometry using electro-spray ionization was studied using a set of 91 compounds, 3 stationary phases, and 2 organic modifiers of the mobile phase. The 24 tested makeup solvents included pure alcohols and methanol in combination with commonly used additives such as water, formic and acetic acid, ammonia, and ammonia salts with varying molarity.
The behavioral trends for different makeup solvent additives were established in the first step. Subsequently, the correlations between physicochemical properties and the MS responses were calculated using the Pearson correlation test and matrix plots.
The regression analysis was performed using five descriptors: molecular weight, pK(a), log P, number of hydrogen donors/acceptors, and the MS responses obtained with methanol as the makeup solvent. The resulting regression equations had a high prediction rate calculated as R-2-predicted coefficient, especially when 10 mmol/L ammonium in methanol was used as an organic modifier of the mobile phase in positive mode.
The trueness of these equations was tested via the comparison between experimental and predicted responses expressed as R-2. Values of R-2 > 0.8 were found for 88% of the proposed equations.
Thus, the MS response could be measured using only one makeup solvent and the responses of other makeup solvents could be easily estimated. The suitability and applicability of determined regression equations was confirmed by the analysis of 13 blind probes, i.e., compounds not included in the original set of analytes.
Moreover, the predicted and experimental responses followed the same increasing/decreasing trend enabling one to predict makeup solvent compositions leading to the highest sensitivity.