Nuclear orientation studies of spin-lattice relaxation and hyperfine fields in ferromagnetic dilute alloys

Nuclear magnetic resonance experiments on impurity atoms in a ferromagnetic host have shown that the measured spin-lattice relaxation time of those nuclei located in domains is strongly dependent on the degree of magnetic saturation of the host material (1, 2, 3). The relaxation time increases as the applied magnetic field is increased and reaches a constant value for a magnetically saturated specimen. Wall nuclei show a much shorter relaxation time than those in the bulk. This fact, together with the increased number of walls present in a magnetically non-saturated specimen could explain the observed field-dependent decrease of the relaxation time if an increasing fraction of wall nuclei is observed. Nuclei located in walls experience a much larger enhancement than those in domains. Therefore special techniques have to be applied to exclusively observe nuclei located in the bulk (1, 4). For this reason some uncertainty exists in the interpretation of the nuclear magnetic resonance measurements. The theory of the spin-lattice relaxation in ferromagnetic metals (5) gives an estimate for the relaxation rate observed in magnetically saturated specimens. No field dependence the relaxation time is predicted. Partly due to the uncertainty in the NMR results, this theoretical problem has received little attention so far. We therefore have employed low temperature nuclear orientation which predominantly measures bulk nuclei to investigate this problem. In most of these experiments the combined technique of nuclear orientation and nuclear magnetic resonance (NMR/ON) (6) has been applied td prepare the initial state from which the relaxation takes place. Some experiments have also been performed by an entirely non-resonant technique (7). Our experimental results on ⁶⁰Co-Fe, ⁵⁴Mn-Fe and ⁵⁴Mn-Ni clearly confirm the field dependence of the relaxation time observed in nuclear magnetic resonance experiments (8). Thus the need for a detailed theoretical study is evident. Performing an NMR/ON experiment the resonance is detected by a change in the observed y-ray intensity. Resonance lines for ⁶⁰Co-Fe, ⁵⁴Mn-Fe and ⁵⁴Mn-Ni have been recorded. We have for the first time observed that their full widths at half maximum show a strong field dependence. An explanation in terms of a local distribution in the demagnetizing field is offered. We have also measured the intensity of the resonance line as a function of the applied field. An estimate shows that this is inadequately explained in terms of the expected field dependence of the enhancement factor. The distribution of hyperfine fields has never before been studied by NMR/ON. We have employed this technique successfully to investigate an alloy of one atomic percent ⁵⁹Co-Fe which has been doped with a small amount of ⁶⁰Co. A strong, well resolved satellite line of the impurity nuclei is observed. These data are interpreted in terms of the effect of near neighbor impurity nuclei on the hyperfine field (9, 10). We have computed a theoretical curve based on parameters given in the literature (10). This provides a moderately good fit for most portions of our spectra. This pilot study demonstrates that NMR/ON is indeed a valuable tool for the investigation of hyperfine field distributions. The advantages over nuclear magnetic resonance studies are that essentially only bulk as compared to wall nuclei are studied and that the sensitivity is independent of the alloy concentration. Based partially on our own data we present a short discussion of the question whether a spin temperature is maintained by the impurity nuclei during relaxation. Finally we offer a comparison between relaxation data measured by NMR/ON and other nuclear orientation techniques (11). For ⁶⁰Co-Fe the relaxation times measured by NMR/ON are found to be almost 50% longer than those measured by techniques in which the initial condition is known. This discrepancy is generally attributed to the incomplete knowledge of the initial conditions when the NMR/ON technique is employed. We have computed theoretical relaxation curves for a number of initial conditions and find that the resulting spread in relaxation time for those curves that allow a good fit to the measured curve is larger than the difference obtained from the experiments. Thus our model indeed could explain the observed discrepancy.Science, Faculty ofPhysics and Astronomy, Department ofGraduat