December 8, 1923] 



NA TURE 



8 



Itures the thermal agitation of the atoms disturbs their 

 perfect ahgnment.even in a pure metal. Since it is 

 sutificient for one line or at most a few lines of atoms to 

 be perfectly straight at any given instance — since 

 such a single line would conduct infinitely well — super- 

 conductivity must set in at a temperature slightly 

 above and not only at actual absolute zero. In a 

 solid solution crystal, however, the atoms can never 

 attain perfect alignment, owing to the lattice-distor- 

 tion, and consequently the electrical conductivity of 

 a solid solution will always be relatively very low, 

 and even at absolute zero, real super-conductivity 

 cannot occur. Further, since the solid-solution lattice 

 is considerably distorted to begin with, the disturbing 

 effect of thermal agitation will be relatively much less 

 than in a pure metal ; in certain circumstances, indeed, 

 thermal expansion may partially relieve the distortion 

 — ^in those cases-, in fact, where solid solubility increases 

 with rising temperature. Consequently, in solid solu- 

 tion alloys the temperature coefficient of electrical 

 conductivity will be much lower than in pure metals, 

 while in some special cases it may even become negative. 

 The theory, as comparison of these inferences with 

 well-known facts at once indicates, offers at all events 

 a good qualitative explanation, and at a later stage 

 even quantitative prediction of electrical properties 

 should be possible. The difficulty here, and indeed 

 throughout the theory, in arriving at numerical results 

 lies in the fact that while the average distorting — i.e. 

 expanding or contracting effect of dissolved atoms on 

 a lattice — can be measured with considerable ease 

 and accuracy by the aid of X-ray spectrometry, the 

 maximum local distortion cannot as yet be deter- 

 mined directly. When this difficulty has been over- 

 come, considerable further progress should become 

 possible. 



We may now briefly consider inter-metallic com- 

 pounds. These are known to metallurgists from the 

 occurrence of certain kinds of singular points on 

 equilibrium diagrams and from characteristic features 

 of micro-structure and of physical properties, but 

 there are a number of alloys in which the existence 

 of definite compounds has hitherto been regarded as 

 doubtful. Again, the results of X-ray analysis, com- 

 bined with the indications of the above theory, prove 

 helpful. Very typical of inter-metallic compounds is 

 the body CuAlg found in copper-aluminium alloys. 

 It is a hard, brittle body, tending to crystallise in 

 well-formed long needles. Its atomic structure has 

 been determined by Dr. Owen and Mr. Preston at the 

 National Physical Lal^oratory. The lattice-structure 

 is shown in the accompanying diagram (F'ig. i). The 

 most striking feature is that certain pairs of aluminium 

 atoms approach one another within a range, centre 

 to centre, of only 2-42 Angstrom units. In an 

 aluminium crystal the lattice-constant is 4-85 A and 

 the closest approach is 2-86 A, and it would be quite 

 impossible, by the application of external pressure, 

 for example, to force the atoms so closely together 

 as they are placed in the compound. The inference, 

 which is justified by comparison with the known 

 lattice structures of other chemical compounds, is 

 that the very much closer approach of atoms in this 

 manner is a characteristic, if not the characteristic, 

 feature of chemical combination as distinct from the 



NO. 2823. VOL. I 12] 



" cohesion bonding " which occurs in the building up 

 of a crystal. It would seem, in the present case, that 

 the copper atom which is combined with the two 

 aluminium atoms has taken away or absorbed some- 

 thing from the aluminium atoms which now allows 

 them to come much closer together. This may well 

 be the absorption of certain exterior electrons by the 

 copper atom ; whatever the detailed mechanism may 

 be, it is probably the essence of chemical combination, 

 and furnishes us at once with a definite criterion for 

 distinguishing between solid solutions and compounds. 

 At first sight one might perhaps expect that inter- 

 mediate classes of structure should be found, in which 

 the inter-atomic distances might be only slightly less 

 than in the typical solid solutions. If our current 

 views of the structure of the atom in " shells " or 

 layers of electrons is correct, however, this should not 

 be the case ; we should find either substances in which 

 there is nothing more than " cohesion bonding " 

 without closer approach of the atoms, or bodies 

 in which the atoms are drawn closer by a definite 

 step. 



There is a further distinction which can be inferred 

 from the present theor}\ In a body of the solid 



2-42 



4-3- 



Fig. I. — Lattice structure of CuAlj 

 • Cu. O Al. 



solution type, atoms of one kind are readily replace- 

 able by atoms of the other ; in a compound, on the 

 other hand, it would be difficult to conceive of any 

 atom being replaced by an atom of the other con- 

 stituent. In the CuAlg structure, for example, it is 

 scarcely possible that any of the aluminium atoms 

 could be replaced by a copper atom. This very definite 

 inference is verified by reference to the equilibrium 

 diagrams of alloy systems in which typical well-defined 

 compounds are to be found — these bodies never 

 exhibit any appreciable amount of dissolving power 

 for their constituents. If we may extend this view 

 to those cases which, metallurgically, are still regarded 

 as doubtful, it will at once serve to classify them into 

 compounds and solid solutions respectively. A well- 

 known group of alloys of this kind is the copper-zinc 

 alloys (brasses), which exhibit a series of solid solutions 

 generally called the alpha, beta, and gamma phases. 

 These are micrographically distinct, and vary widely 

 in many of their properties, and it has been thought 

 that each was based upon a definite chemical compound 

 possessing a wide range of dissolving power for copper 

 and zinc. 



In one of the papers mentioned above (May lecture) 

 the writer suggested that these bodies need not be, 

 and probably were not, based on definite compounds, 

 but that they would probably be found to be based 

 upon what might be termed allotrope lattices of copper. 

 In the case of iron and nickel, for example, it is known 



