Title:

Nuclear magnetic resonance in liquid metals and alloys

Nuclear magnetic resonance measurements in pure lead and in dilute binary alloys o£ lead are reported. (i) Measurements of the 207pb linewidth and Knight shiftt (K) in solid and liquid lead are presented. The linewidth shows a narrowing at 550 K and this is interpreted in terms of a motional narrowing phenomenon. The value obtained for the pseudodipolar contribution to the linewidth is ̃/3 x 104 tesla and this is shown to be consistent with estimates available in the literature. The values of (l/K)(dK/dT) for both the solid and liquid phases are +6.3(±0.4} x 10 5 and 7.6(±0.9) x 10 5 (kelvin).1 respectively. The result in the solid is consistent with a value deduced £rom spot measurements available in the literature. The absolute value of the 207pb Knight shift in pure lead, together with the values £or 12 other pure metals, is discussed in terms of a zeroorder pseudopotential formulation. Holland has recently discussed the evaluation of K from a single O.P.W. formulation with the approximation that the pseudofunction is constant over an ion core. This approximation is now removed and good agreement with experiment is obtained using interacting electron spin susceptibilities (X p ). It is shown that very little change in K is expected on melting if "X remains cons tan t. P It is thus concluded that when K remains constant through the melting point the spin susceptibility, and therefore the density of states, usually nearly the freeelectron value in the solid. The extension of the Holland theory introduces a dependence of the Knight shift on the Fermi wave vector, k F. The temperature dependence of K, produced by expansion effects , is evaluated but in every case is an order of magnitude smaller than is observed. It is shown that the major temperature dependence arises from the term to firstorder in the pseudopotential and this has been evaluated for 6 liquid metals from the expression of Faber. (ii) Measurements of the 207pb Knight shift in dilute liquid alloys of lead with Ag, Cd, In, Sb, Au, Hg, TI and Bi are reported. The results for the liquid at 625K are larger in magnitude and of opposite sign to those on lead solid solutions at room temperature. The way in which these differences occur is investigated for the two alloys Pb  18% Bi and Pb  20% TI. The experimental results for the liquid are discussed in turn, in terms of the zeroorder pseudopotential theory, the firstorder theory and the Friedel theory. It is deduced that a reasonable explanation can be obtained from the Friedel model if the semiempirical choice of phase shifts, deduced by Flynn, is used. (iii) Measurements of the 207pb and 209Bi Knight shifts and 209Bi linewidth in the Pb  Bi system are reported. Experimental studies of. the concentration dependence and the temperature dependence are presented. The 209Bi linewidth versus concentration shows a maximum at approximately 50~ and around this maximum, the linewidth versus temperature plot shows a sharp minimum. This is interpreted in terms of a quadrupolar contribution to the linewidth. This quadrupolar relaxation is also observed in pure bismuth and the enhancement on alloying is discussed in terms of a crude ddiffusing ion model. On the basis of this model a qualitatively correct description of the alloying effects is obtained. Finally, in a general discussion, it is concluded that the nearlyfreeelectron picture is adequate for a description of Knight shifts in liquid metals and alloys if allowance for electronelectron interactions is included into the spin susceptibility.
