In the field modulation technique for magnetic measurements, the voltage signalV(H) from a pickup coil surrounding a metal cylinder of permeability μ=1+K(H) and conductivity σ=σ0[1+S(H)] in the presence of an alternating fieldh0 cos ωt superimposed on a steady fieldH (both parallel to the axis of the cylinder) provides a measure ofK(H). This technique is particularly useful for study of the de Haas-van Alphen effect, in whichK(H) oscillates withH, and the paper is mainly concerned with the effect of eddy currents on the relation betweenV(H) andK(H) and the practical implications for the use of the technique. It is also shown that if ω is high enough,S(H) (e.g., the oscillations of the Shubnikov-de Haas effect) and also information on the orientation dependence of σ can be extracted from suitable measurements onV(H) and its orientation dependence. The calculations fall into two main parts, corresponding to weak and strong modulation, i.e., according ash0 is small or appreciable compared with the period of the field dependence ofK andS. For weak modulation, results are obtained for arbitrary values ofK andS over the whole range of ω. It is shown that ifK(H) is not small,V(H) no longer faithfully reproduces the form ofK(H) once ω is high enough to make the skin depth of the eddy currents comparable to the radius of the cylinder. In general,V(H) contains not only the fundamental frequency ω, but also its harmonics, i.e.,V(H)=∑n=1∞Vn(H)eniωt, but it is only for strong modulation that the higher harmonics become appreciable. The variation of theVn with modulation current and ω is calculated under the special assumptions thatK andS ≪ 1 and thatK andS are simple harmonic functions ofH; for these special assumptions the weak modulation results forV1 obtained earlier fall out as a limiting case of the results for arbitrary modulation. Some of the theoretical predictions were tested experimentally on sodium and potassium crystals. Good agreement was found at frequencies up to such that the skin depth became comparable to or smaller than the sample radius, when appreciable discrepancies appeared. Some of these could be attributed to helicon resonances, the possibility of which was ignored in the theory, and some to sample inhomogeneity, the effect of which are briefly discussed in an appendix.
Read full abstract