Originally published in Xattirc, 115, <)74 (1925) 



11. SELF-DIFFUSION IN SOLID METALS 



G. Hevesy and A. Obrutsheva 

 From tho Institute of Theoretical Physics, University of Copenhagen 



The "sagacity" with which atoms, or groups of atoms, oscillating about 

 fixed points in the crystal lattice, refuse to exchange position with 

 neighbouring atoms, is often regarded as one of the chief characteristics 

 of the crystalline state. On the other hand, numerous cases are recorded 

 in which crystalline bodies, for example, solid metals, penetrate into 

 each other, in which, therefore, a replacement of the atoms of one metal 

 by those of the other takes place. The classical experiments of Roberts- 

 Austen on the diffusion of gold in lead bars are widely known. At a 

 temperature as low as 100° he found the diffusion coefficient of gold 

 in lead to be 2 x 10"* cm^ day-i, being thus only about 100,000 times 

 smaller than that of sodium chloride in water. Several cases of inter- 

 penetration of solid metals have been recorded since, including the 

 interesting case of the diffusion of thorium in heated tungsten wires, 

 reported recently by Langmuir. But it must be noticed that from the 

 rate at which one metal like gold diffuses in another like lead, no con- 

 clusion can be drawn about the velocity with Avhich the atoms change 

 their position either in a bar of pure lead or of pure gold ; no conclusion 

 can be drawn on the rate of self-diffusion in these elements. 



The idea of self- diffusion was introduced by Maxwell, when calculat- 

 ing the rate of diffusion of gases. The calculation was very much simplified 

 by considering the case in which the molecules of the two diffusing 

 gases had the same properties, for example, the exchange of place of 

 molecules in a column of nitrogen. The use of the radioactive isotopes 

 of lead enabled one of the writers, in collaboration with J. Groh (Ann. 

 (I. Phys. 65, 216 [1921]) to realise a measurement of self-diffusion in 

 the case of liquid and solid lead, the diffusion in liquids and solids being 

 practically independent of the difference in the masses of the isotopes. 

 For the rate of the self-diffusion in molten lead, namely, of thorium B 

 in molten lead, close to the melting point, the value found was 2 cm- 

 day-^. In the solid metal, however, after heating a bar, the upper pari 

 of which was composed of radio-lead, for about a year at 280°, and 

 llien analysing the lower part with the electroscope, no diffusion could 



