MaV 2 1, 1896J 



NA TURE 



57 



tube, and in both cases, the initial concentration of 

 the alloy, denoted by a c, from which diffusion pro- 

 ceeded, was the same, so that the area, a c c d, 

 represents the total amount of gold or platinum em- 

 ployed in the experiment, the whole quantity of either 

 metal being initially below the linef/c. The final state 

 of complete diffusion would be represented by the area 

 II /> g'f, which is the same as a c c </, since the quantity 

 of gold or of platinum remains unaltered. In the same 

 manner the area ii y x f, would represent the distribu- 

 tions of the gold at the end of the experiment, and conse- 

 quently in experiments which have lasted for equal times, 

 the nearer the cur\e approximates to the line /', ^f,', the 

 more rapid is the diffusion of the metal it represents. 

 It will be evident from the distribution of the spheres of 

 gold and platinum that diffusion can be accurately 

 measured iri molten metals. 



P.\RT \\.—DiJfiision of Solid Metals. 



The second part of the investigation was devoted to 

 the consideration of the diffusion of solid metals, ^tuch 

 of the evidence is historical, for there has long' been a 

 prevalent belief that diffusion can take place in solids, and 

 the practice in conducting important industrial operations 

 supports this view. In this connection two truly vener- 

 able "cementation" processes may be cited. The object 

 in the first of these is the removal of silver from a solid 

 gold-silver alloy, while the second is employed in steel 

 making by the carburization of solid iron. In both of 

 these processes, however, a gas may intervene, though the 

 carburization of iron by the diamond, which, in l88g, I 

 effected /'// vacuo, suggests that if a gas does intervene 

 in the latter case, its quantity must be very minute. In 

 connection with the mobility of various elements in iron 

 the work of Colson, of Osmond, and of Moissan must be 

 carefulh' kept in view. 



The electro-deposition of metals also affords evidence 

 of the interpenetration of metals. I observed in 18S7 that 

 an electro-deposit of iron on a clean copper plate will 

 adhere so firmly to it that when the metals are severed 

 by force, a copper film is actually stripped from the copper 

 plate and remains on the iron, thus aftbrding clear 

 evidence of the interpenetration of metals at the ordinary 

 temperature, and this interpenetration of copper and iron 

 will take place through an interxening film of nickel. 



My friend Dr. George Gore has gi\ en me the follow- 

 ing interesting reference to the penetration of gold and 

 platinum at a temperature below redness, which is re- 

 corded in " Weldon's Register'' for July 1863 by Edward 

 -Sonstadt, who states that he gilded a platinum crucible 

 " inside and out . . . but no sooner was the platinum 

 warmed than it began to change colour, and before the 

 crucible attained visible redness not a vestige of the 

 gilding remained." 



This is interesting in connection with the earlier 

 observation of Faraday and .Stodart, who in 1820 showed 

 that platinum will alloy with steel at a temperature at 

 which e\en the steel is not melted, and they expressed 

 their interest in the formation of alloys by cementation, 

 that is by the union of solid metals. 



The i-emarkable view expressed by (Jraham, in 1863, 

 that the " three conditions of matter (liquid, solid, and 

 gaseous) probably always exist in e\ery liquid or solid 

 substance, but that one predominates over the other," 

 affords ground for the anticipation that metals will 

 diffuse into each other at temperatures far below their 

 melting points. The important work by Spring, in 1SS6, 

 on the lead-tin alloys, showed that they retain a certain 

 amount of molecular acti\ity after they become solid, 

 and special importance will alwrns be connected 

 with the proof afforded by him (iSS^-, that alloys may 

 be formed either by the strong compression of the finely 

 divided constituent metals at the ordinary tempera- 

 ture, or (1894) by the union of solid masses of metal 



NO. 1386, VOL. 54] 



compressed together at temperatures which varied from 

 1 80' in the case of lead and tin, to 400° in the case of 

 copper and zinc ; tin melting at 227' and zinc at 415'. 



Early evidence as to the volatilisation of solid metals 

 may be traced to the expression of Robert Boyle's belief, 

 that even such solid bodies as glass and gold might 

 respectively " have their little atmospheres, and might in 

 time lose their weight," and Merget's experiment on the 

 exaporation of frozen mercury is specially interesting in 

 relation to Gay-Lussac's well-known discovery that the 

 \apours emitted by ice and water both at o' C., are of 

 exactly equal tension. Demarijay's experiment on the 

 volatilisation of metals in vacuo at comparatively low 

 temperatures is, moreover, connected with the evidence 

 afforded by Spring (1894), that the interpenetration of 

 two metals at a temperature below the melting point of 

 the more fusible of the two is preceded by volatilisation. 



It is well to remember, however, that interesting as the 

 results of the earlier experiments are, as affording evidence 

 of molecular interpenetration, they do not, for the purpose 

 of measuring diffusivity, come within the prevailing con- 

 ditions in the ordinary diffusion of liquids, in which the 

 diffusing substance is usually in the presence of a large 

 excess of the solvent, a condition which was fully main- 

 tained in the experiments on the diffusion of liquid metals 

 described in the first part of the Bakerian Lecture.. 

 \'an 't Hoff has made it highly probable that the osmotic 

 pressure of substances existing in a solid solution is 

 analogous to that in liquid solutions, and obeys the same 

 laws ; and it is probable that the behaviour of a solid 

 mixture, like that of a liquid mixture, would be greatly 

 simplified if the solid solution were very dilute. 



The experiments on the diffusion of solid metals are of 

 the same nature as in the case of fluid metals, except that 

 the gold, which was the metal chosen for examination, was 

 placed at the bottom of a solid cylinder of lead instead of 

 a fluid one. 



In the first series of experiments, cylinders of lead, 

 70 mm. long, with either gold, or a rich alloy of gold and 

 lead at their base, were maintained at a temperature of 

 251° (which is 75' below the melting point of lead; for 

 thirty-one days. At the end of this period the solid lead 

 was cut into sections, and the amount of gold which had 

 diffused into each of them was determined in the usual 

 way. Other experiments were made, in which the lead 

 was maintained at 200', and at various lower tempera- 

 tures down to that of the laboratory. The following are 

 the results in sq. cm. per day: — 



The experiments at the ordinary temperature are stil 

 in progress, but there is evidence that slow diffusion of 

 gold in lead occurs at the ordinary temperature. If clean 

 surfaces of lead and gold are held together /// vacuo at a 

 temperature of only 40' for four days, they will unite 

 firmly, and can only be separated by the application of a 

 load equal to one-third of the breaking strain of lead 

 itself The nature of welding, however, remains to be in- 

 vestigated, as there is probably interlocking of molecules 

 and atoms, which precedes true diffusion. It may be 

 considered remarkable that gold placed at the bottom of 

 a cylinder of lead, 70 mm. long (which is to all appear- 

 ance solid), will diffuse to the top in notable quantities at 

 the end of three days. At 100 the diffusivity of gold in 

 solid lead can readily lie measured, though its diffusivity 

 is only 1/100,000 of that in fluid lead at a temperature of 

 500', and experiments which are stfll in progress show 

 that the diffusivity of solid gold in solid silver, or copper, 

 at Soo' is of the same order as that of gold in solid lead 

 at 100°. 



