N. Adnani et al., NUCLEAR MAGNETIC-RELAXATION IN DISORDERED SOLIDS - A MONTE-CARLO STUDY OF METAL HYDROGEN SYSTEMS, Journal of physics. Condensed matter, 6(16), 1994, pp. 2999-3012
The dipolar nuclear magnetic relaxation rate associated with the hoppi
ng diffusion of atoms in a disordered solid is calculated by Monte Car
lo methods. The model is intended to simulate the diffusion of hydroge
n atoms trapped at interstitial positions in the matrix of metal atoms
in amorphous alloys. The principal features of the model system are t
hat the atoms hop on a spatially disordered array of traps and the tra
pping energy varies from trap to trap so that the diffusion of the hyd
rogen is characterized by a distribution of jump rates. The effective
jump rate from a trap is assumed to have an Arrhenius dependence on te
mperature. Calculated at constant temperature, the characteristic peak
in the relaxation rate, which occurs in ordered solids when the produ
ct of the average jump rate and the Larmor frequency is approximately
unity, is found to be broadened and shifted in frequency, particularly
when the occupancy of the traps is high. The long-range diffusion con
stant is also calculated and used to evaluate the effect of atom-vacan
cy correlations. It is found that the shifts in the relaxation peak ca
nnot be accounted for solely by these correlation effects and it is su
ggested that multiple hopping of the more rapidly diffusing spins is a
contributory factor. The shifts have a profound effect on the tempera
ture dependence of the relaxation rate when the distribution of jump r
ates is also dependent on temperature. The adjustments to the peak in
the relaxation caused by the distribution are small when the temperatu
re dependence is taken into account, showing that experiments involvin
g only the temperature variation of the relaxation are unlikely to be
a sensitive method for detecting the presence of a jump-rate distribut
ion. This aspect of the results of the computer model is illustrated b
y comparison with experimental data.