Shiftless Fermi Level in Iron, Nickel, and Cobalt

Walmsley has suggested a method whereby one can deduce the relative density of states at the Fermi level for spin-up and spin-down electrons. This method compares the shift in Fermi level with applied magnetic field for a magnetic metal with that of a nonmagnetic metal by measuring the voltage induced across the plates of a capacitor, one plate of which is of each metal. His measurements indicate that all Fermi surface electrons have moments parallel to the bulk magnetization. We have repeated and extended these measurements and get different results. We have measured the voltage induced in Fe-Al, Ni-Al, and Co-Al capacitors (evaporated magnetic metal on one side of a high-permittivity ceramic wafer and Al on the other) as a function of applied field. We have also measured a series of Fe-Cu capacitors of different thicknesses of iron, and a capacitor constructed of rolled iron foil and aluminized Mylar. Voltages appear which relate to the change of magnetization rather than to applied field. A magnetostriction-mechanical strain-piezoelectric voltage is consistent with these results. The only reliable data is thus that taken at high fields approaching or exceeding saturation. Local demagnetization or anisotropy in a thin film deposited on rough ceramic increases the field necessary for saturation, which may account for Walmsley's result. Thicker iron films show a decrease in these voltage changes and the foil capacitor shows essentially nothing. We find the shift in Fermi level in iron, cobalt, and nickel to be less than ½ μV/kOe at high fields, and the ratio of spin-up to spin-down density of states at the Fermi level to be less than 1.18.