Experimental design optimization and confidence interval estimation in multiexponential NMR relaxation measurements

Numerical simulations were used to estimate the precision of parameters responsible for NMR spin-lattice relaxation in a coupled two-spin system where two relaxation mechanisms, dipole-dipole and chemical shift anisotropy, are present. Confidence intervals in nonlinear least-squares fitting, necessary for multiexponential relaxation, can be obtained with restrictive assumptions. The simulations allow one to investigate the effect of experimental variables on parameter precision without performing many lengthy experiments. Precision depends upon may factors including the nucleus observed, how the recovery curves are sampled, the sample temperature stability, and the number of simultaneously estimated parameters.