FERROMAGNETISM] MAGNETISM and for Fe io or I 1. The fact that a ferro-cobalt, of composition has a higher saturation intensity than Fe or Co is strikingly indicated on the curves. (This was discovered by Preuss in 1912, and ferro-cobalt has been used for the pole pieces of elec tromagnets.) Iron loses its ferromagnetic properties at about 769° C. (The temperature varies considerably according to impurities present.) Below that temperature it is known as a, above as (3 iron; both these modifications have the same crystal structure (body centred cubic), and it is probable that there is no change other than that due to the loss of the spontaneous magnetization suggest e d by Weiss's theory. 13 iron is para magnetic, the inverse of the sus ceptibility varying approximately linearly with the temperature, up to 92o°, when there is a sud den decrease in the susceptibility, due to the formation of y iron (face centred). At 1,395° there is again a sudden change due to the formation of 8 iron (body centred). The magnetic charac teristics above the Curie point are shown in fig. 39, the numeri cal value of the constants in the formula x = Cbeing given in the table.
Magnetic Constants of Iron, Cobalt and Nickel above the Curie Point (The magnetic moment per atom is calculated from the Curie constant in the usual way, and given as p Weiss magnetons.) The magnetic moments of the atoms deduced from measure ments above the Curie point do not agree with those deduced from the low temperature sat uration measurements (fig. 38), the p value for nickel, for ex ample coming out as about 8 (for NO 1, between 412° and goo° C) in one case, and about 3 in the other. For some sub stances there may be a change in the magnetic carriers with change of temperature (as un doubtedly occurs when iron changes from the 0 to the y state), but it has been shown by E. C. Stoner that it is not necessary to assume that such a change invariably occurs in pass ing from a low temperature up THE CURIE TEMPERATURES to the Curie point. In carrying out calculations in the usual way
different p values per atom may be deduced, although there is no structural change involving a variation in the relative numbers and the nature of the magnetic carriers present. In an ordinary para magnetic salt (such as the magnetic carrier is the metallic ion (e.g., Ni"), and the number of electrons it contains is known. In a metal, however, it is not known what are the actual ions present ; but there is no reason for supposing they are all similar.
Transfer and sharing of electrons may take place among groups of atoms. The number of atoms in these groups will not neces sarily be of any significance in connection with the crystal struc ture of the metal, but each group may be regarded as containing the minimum number of atoms required to give rise to the same mean magnetic effect as is observed for the crystal as a whole. If such groups of atoms are assumed, with the aid of the quantum modification of the Langevin theory, which is successful in deal ing with normal paramagnetism, it is possible to account for the different p values at high and low temperatures without postulat ing any change of structure, for which there is no independent evidence. It is also possible to explain the fact that a p value of 3 is obtained for nickel, at low temperatures, which at first sight appears to conflict with the fact that the unlit magneton value suggested by the Bohr theory corresponds to a p value of 5. (See Paramagnetism.) The existence of (fig. 38) sug gests that in nickel itself there may be groups of five atoms, some of which lose electrons, forming ions. If among five atoms, there is one Ni" ion (with a moment corresponding to two Bohr magne tons) and one Ni+ ion (with one Bohr magneton), the mean mo ment deduced from low temperature saturation when the ele mentary magnets are aligned will be (2X5)+(1 X5) — 5 3, remembering that one Bohr magneton is approximately equal to 5 Weiss magnetons.