The phospholipid membrane is a self-assembled, dynamic molecular system that may exist alone in association with only water, or in complex systems comprised of multiple lipid types and proteins. In this dissertation the intra- and inter-molecular forces responsible for the atomistic, molecular and collective equilibrium structure and dynamics are studied by nuclear magnetic resonance spectroscopy (NMR). The multinuclear NMR measurements and various experimental techniques are able to provide data that enable the characterization of the hierarchical spatio-temporal organization of the phospholipid membrane. The experimental and theoretical studies conducted target membrane interactions ranging from model systems composed of only water and lipids, to multiple component domain forming membranes that are in association with peripheral and trans-membrane proteins. These measurements consisit of frequency spectrum lineshapes and nuclear-spin relaxation rates obtained using 2 H NMR, 13 C NMR, 31 P NMR and 1 H NMR. The changes of these experimental observables are interpreted within a statistical thermodynamic framework that allows the membrane structure, activation energies, and correlation times of motion to be determined. The cases presented demonstrate how fundamental principles of NMR spectroscopy may be applied to a host of membranes, leading to the biophysical characterization of membrane structure and dynamics.

Abstract This thesis reports simulations of direct observables in electron and nuclear spin relaxation experiments in an example paramagnetic system, as well as polarization transfer occurring in a spin-exchange optical pumping (SEOP) experiment. Studies of paramagnetic relaxation are important, e.g., in the development of agents used for enhanced contrast in magnetic resonance imaging. SEOP is used to produce hyperpolarized noble gases, which are then used to, e.g., enhance sensitivity in structural studies of matter with nuclear magnetic resonance. Presently the theory, available software and hardware for such computational modeling have reached a state in which quantitative reproduction of the experimentally observed magnetization decay is possible from first principles. The present multiscale computations are carried out from first principles combining molecular dynamics simulations of atomistic motion and quantum-chemical electronic structure calculations of the spin interaction parameters that enter the effective spin Hamiltonian. A time series of the spin Hamiltonian is then explicitly used to propagate spin dynamics in the system, and dynamical time constants of the magnetization are obtained through ensemble averaging. The complete decay of electron spin magnetization could be followed directly within the duration of the simulation, whereas the nuclear spin relaxation rates were extracted using Kubo’s theory regarding generalized cumulant expansion and stochastic processes. The extracted electron and nuclear spin relaxation rates for the chosen prototypic system, the aqueous solution of Ni²⁺, are in quantitative and semi-quantitative agreement, respectively, with the available experimental results. The simulations of polarization transfer corroborate the empirical observations on the importance of van der Waals complexes and binary collisions in the spin-exchange process. Long van der Waals complexes represent the overwhelmingly most significant kind of individual events, but the short binary collisions can also give a relatively important contribution due to their vast abundance. This thesis represents a first study in which first principles-calculated trajectories of individual events could be followed. The simulations reported in this thesis were run without any empirical parametrization and thus represent a significant step in first-principles computational modeling of magnetization dynamics.