HPLC – 51st International Symposium on High Performance Liquid Phase Separations

M.R. Euerby^abc, R. Wong^c, V. Paget^d, E. McKoy^c, J. Hogbin^d, S.N. Berry^d, J.K. Field^bc, AM. Smith^e,
V.E. Zwicker^f

[a] Strathclyde Institute of Pharmacy and Biomedical Sciences, University of Strathclyde, Glasgow,
UK, G4 0RE
[b] The Open University, Faculty of Science, Walton Hall, Milton Keynes, United Kingdom MK7 6AA
[c] Shimadzu UK, Milton Keynes, Buckinghamshire, United Kingdom MK12 5RE
[d] Advanced Chemistry Development, UK Ltd. (ACD/Labs), Venture House, Arlington Square, Downshire Way, Bracknell, Berkshire, RG12 1WA
[e] Advanced Chemistry Development, Inc, 8 King St E, Toronto, ON, M5C 1B5, Canada
[f] Advanced Chemistry Development, Germany GmbH, Hahnstraße 70 60525 Frankfurt am Main

Computer-assisted modeling of retention time is a powerful tool to accelerate method development for analytical and preparative separations. Building retention models using RPLC is dependent upon peak tracking and component identification. Dwell volume is an important variable in predicting analyte retention. Here, it has shown that computer-assisted modeling can also be of great diagnostic value. We present the ability to switch the order of elution of analytes based on organic modifier and stationary phase combinations, and the ability to optimize the final separation considering the % initial, final, and gradient time.

In wave 1, various stationary phases are combined with different organic modifiers in gradient RPLC at a fixed gradient slope and column temperature. In wave 2, a two-dimensional gradient time versus temperature resolution model is created with the LC Simulator module in ACD/Method Selection Suite software. This constructs a resolution response and predicts optimum separation conditions with 9 experiments from 3 temperatures and 3 gradient times.

In Wave 1, using both PDA and MS detection, all components in the mixture could be peak tracked in Method Selection Suite. The ability to switch the order of elution of analytes based on organic modifier and stationary phase combinations was also observed.

In Wave 2, peak movement as a function of temperature and gradient time is observed. The resolution map shows that the accuracy of the retention model is %Δt_R < 0.3%, with improved separation observed at temperatures below 35°C. Switches in peak elution are indicated by the resolution map. LC simulator was used to investigate the behavior of the analytes in the test mixture (composed of 8 analytes of varying hydrophobicity). Excellent retention time predictions were achieved when the initial %B was kept the same as that used to generate the retention model input data. When there were larger differences in the initial %B optimized and %B used to create the model, anomalous retention predictions were observed. This was caused by analytes with low to medium retentivity that migrated isocratically down the column, here small errors in dwell volume can cause larger inaccuracies in retention time prediction. More accurate retention time predictions were obtained using Method Selection Suite software to try different dwell volumes that fit the retention model, compare the residuals obtained, and select the dwell volume that gave the lowest residual. It was also seen that analytes that were more polar exhibited this inaccuracy to a greater degree compared to analytes that were more non-polar. This observation was tested further with an additional test mixture of phenone standards which showed the same behavior.