Resolution of on flow LC/NMR data by multivariate methods - a comparison

Data from LC/NMR have some distinct characteristics, differing from those of LC/MS and LC/DAD. This paper compares several approaches for resolution on both a three-component and a seven component case study. Non-iterative methods based on principal component analysis (EFA, WFA, HELP) and iterative approaches using constraints (ALS, Gentle) are compared with a new proposed method, constrained key variable regression (CKVR), which involves refining key variables and then using constraints of non-negativity and equality to obtain optimized spectral resolution. In LC/NMR the reconstruction of detailed spectral profiles is important, requiring specific refinement of existing tools. Copyright (C) 2002 John Wiley Sons, Ltd.