a 
ON DIFFUSION IN SOLIDS. 363 
although cementation to a very slight extent at so low a temperature 
as 250° has been recorded.** 
The diffusion of silicon into iron has often-.been observed.*’ At 
1,200° it is possible to convert iron completely into a silicide by heating 
in contact with solid silicon, although in this instance the vapour of 
silicon has been considered to play a part in-the reaction.°® The diffi- 
eulty of observations of this kind is illustrated by the case of sulphur. 
An attempt to measure the diffusion of sulphides through steel led at 
first to very inconclusive results,’ owing to the escape of fusible 
sulphides, but further experiments showed that true diffusion took 
place, and that when cuprous sulphide was used a part of the copper 
was also transported into the metal.°* 
When iron was placed in contact with alloys of iron containing 
other elements and heated at about 1,000° in vacuo, diffusion was 
observed when the element studied was sulphur, phosphorus, or nickel, 
but not when it was manganese, silicon, chromium, aluminium, tung- 
sten, arsenic, or copper.*® 
The case-hardening of iron and steel is a process of superficial 
cementation, and what has been said as to the latter process applies 
also to the former. Many important details as to the velocity of car- 
burisation and diffusion are to be found in Giolitti’s memoirs. 
- The decarburisation of iron-carbon alloys-by heating in an oxidising 
medium is the reverse of cementation. Oxides of iron are commonly 
employed as the medium. The process was described in detail by 
Réaumur.'°°. It is known that the removal of the carbon is preceded 
or accompanied by the decomposition of the carbide into iron and 
temper carbon, and the reaction has been explained 1°! as one of oxida- 
tion of this free carbon by carbon dioxide, formed by reactions between 
the oxidising material and the carbide in the superficial layer of iron. 
The explanation is, however, imperfect, and it is certainly possible 
for carbon to be removed without previous precipitation as temper 
carbon.?°? It is therefore most probable that removal of carbon by true 
diffusion actually takes place, in accordance with the theory formerly 
held, and that decarburisation by gas, although taking the principal 
share in some forms of the technical process, is not absolutely essential. 
Segregation and Recrystallisation. 
Diffusion in solid metallic alloys is well illustrated by the segrega- 
tion of constituents. A new phase commonly makes its first appear- 
ance in a finely divided form, and although chemical equilibrium may 
have been attained, physical equilibrium is not reached until the smail 
particles have united to form larger aggregates. Steel offers a good 
94 A, Colson, Compt. rend., 1881, 98, 1074. 
% A. Colson, ibid., 1882, 94, 26. 
% H. Moissan, ibid., 1895, 121, 621; P. Lebeau, Bull. Soc. Chim., 1902 [iii], 27, 44 
7 EK. D. Campbell, Jour. Iron and Steel Inst., 1897, ii. 80. 
9% Tbid., 1898, ii. 256; S. A. Grayson, ibid., 1910, i. 287. 
9 J. O. Arnold and A. McWilliam, Joc. cit. 100 Op. cit. 
10 KF. Wiist, Metallurgie, 1908, 5, 7; H. Becker, tbid., 1910, 7, 41. 
102 W.H. Hatfield, Jour. Iron and Steel Inst., 1909, i. 
