?8o 



NA TURE 



[August 17, 1905 



of metallic glasses and films is determined, not only by 

 the absolute size of the metal particles, but also by the 

 proportion of the total volume they occupy in the medium 

 in which they are diffused. The results of Mr. Garnett's 

 calculations arc in close agreement with a number of 

 the observations on the colour and microstructure of thin 

 metal films which I had already recorded, and they appear 

 to me to supply the explanation of much that had appeared 

 puzzling before. My own observations lead me to think 

 that the actual microscopic particles which are to be seen, 

 and the larger of which can also be measured, in films 

 and solutions or suspensions, do not in any way represent 

 the ultimate units of structure which are required by Mr. 

 Garnett's theory, but that these particles are aggregates 

 of smaller units built up in more or less open formation. 



That a relatively opaque substance like gold may be so 

 attenuated that when disseminated in open formation it 

 becomes transparent is contrary to all our associations 

 with the same operation when performed on transparent 

 substances like glass or crystalline salts. The familiar 

 experiment of crushing a transparent crystal into a per- 

 fectly opaque powder would not prepare us for the effect 

 of minute subdivision on the transparence of metals. At 

 first it might be supposed that this difference is due to 

 the very rough and incomplete subdivision of the crystal 

 by crushing ; but this is not the case, for the perfectly 

 transparent oxide of magnesium may be obtained in a 

 state of attenuation comparable with that of the gold, 

 by allowing the smoke from burning magnesium to deposit 

 on a glass plate. The film of oxide obtained in this way 

 is found to be built up of particles quite as minute as 

 those of which the gold films are composed, yet the opacity 

 of the oxide film is relatively much greater. The minute 

 particles of the dielectric, magnesium oxide, scatter and 

 dissipate the light waves by repeated reflection and re- 

 fraction, while the similar particles of the metallic con- 

 ductor, gold, act as electrical resonators which pass on 

 some of the light waves while reflecting others. Specimens 

 of films of gold and silver and of magnesium oxide are 

 exhibited on the table and on the lantern screen. When 

 the metallic particles are in this state of open formation 

 and relative transparence, it was found that the electrical 

 conductivity of the films had completely disappeared. 

 Films of this description were found to have a resistance 

 of more than 1,000,000 megohms as compared with only 

 six ohms in the metallic reflecting condition. 



Molecules in the Solid State. 



My examination of gold films and surfaces has revealed 

 the fact that during polishing the disturbed surface film 

 behaves exactly like a liquid under the influence of surface 

 tension. At temperatures far below the melting point 

 molecular movement takes place under mechanical dis- 

 turbance, and the molecules tend to heap up in minute 

 mounds or flattened droplets. These minute mounds are 

 often so shallow that they can only be detected when the 

 surface is illuminated by an intense, obliquely incident 

 beam of light. I have estimated that these minute mounds 

 or spicules can be seen in this way in films which are 

 not more than five to ten micro-millimetres in thickness. 

 .\ film of this attenuation may contain so few as ten to 

 twenty molecules in its thickness. 



When moderately thin films of gold are supported on 

 glass and heated at a temperature of 400°-5oo°, thcv be- 

 come translucent, and the forms assumed under the in- 

 fluence of surface tension can be readily seen by trans- 

 mitted light. It was in this way that the beautiful but 

 puzzling spicular appearance by obliquely reflected light 

 was first explained as due to the granulation of the sur- 

 face under the influence of surface tension. Photo- 

 micrographs of these films are exhibited. 



Turning now to the mechanical properties of metals, 

 we find that gold has proved itself of great value in the 

 investigation of some of these. It has long been recog- 

 nised as the most malleable and ductile of the metals, 

 whilst its chemical indifference tends to preserve it in a 

 state of metallic purity throughout any prolonged series 

 of operations. 



The artificers in gold must very early have learned that 

 its malleability and ductility are not qualities which in- 

 definitely survive the operations of hammering and wire- 

 drawing. A piece of soft gold beaten into a thin plate 



NO. 1868, VOL. 72] 



dees not remain equally soft throughout the process, but 

 spreads with increasing difliculty under the hammer. If 

 carelessly beaten it may even develop cracks round its 

 edges. We may assume that the artificers in gold very 

 soon discovered that by heating, the hardened metal might 

 be restored to its former condition of softness. 



In connection with the study of the micro-metallurgy 

 of iron and steel during recent years it has been recognised 

 that heat annealing is, as a rule, associated with the 

 growth and development of crystalline grains, and Prof. 

 Ewing and Mr. Rosenhain have shown that overstrain is 

 often if not invariably associated with the deformation of 

 these crvstalline grains by slips occurring along one or 

 more cleavage planes. This hypothesis, though well sup- 

 ported up to a point by microscopic observations on a 

 variety of metals, olTers no explanation of the natural 

 arrest of malleability or ductility which occurs when the 

 overstrain has reached a point at which the crystalline 

 grains are still, to all appearance, only slightly deformed. 

 At this stage there is no obvious reason why the slipping 

 of the crystalline lamells should not continue under the 

 stresses which have initiated it. But far from this being 

 the case, a relatively great increase of stress produces 

 little or no further yielding until the breaking point is 

 reached and rupture takes place. 



The study of the surface effects of polishing, already 

 referred to, had shown that the thin surface film retained 

 no trace of crystalline stiucture; while it also gave the 

 clearest indications that the metal had passed through a 

 liquid condition before settling into the forms prescribed 

 by surface tension. Froin this it was argued that the 

 conditions which prevail at the outer surface might equally 

 prevail at all inner surfaces where movement had occurred, 

 so that every slip of one crystalline lamella over another 

 would cause a thin film of the metal to pass through the 

 liquid phase to a new and non-crystalline condition. By 

 observations on the effects of beating pure gold foil, it 

 was found that the inetal reached its hardest and least 

 plastic condition only when all outward traces of crystal- 

 line structure had disappeared. It was also ascertained 

 that this complete destruction of the crystalline lamellae 

 and units could only be accomplished in the layers near 

 the surface, for the hardened substance produced by the 

 flowing under the hammer appears to encase and protect 

 the crystalline units after they become broken down to a 

 certain size. By carefully etching the surface in stagcv 

 by means of chlorine w'ater cr cold aqua regia, the 

 successive layers below the surface were disclosed. The 

 surface itself was vitreous ; beneath this was a layer of 

 minute granules, and lower still the distorted and broken- 

 up remains of crystalline lamellfe and grains were em- 

 bedded in a vitreous and granular matrix. The vitreous- 

 looking surface layer represents the final stage in the 

 passage from soft to hard, from crvstalline to amorphous. 

 By heating the beaten foil, its softness was restored ; and 

 on etching the annealed metal it was found that the 

 crystalline structure also was fully restored. Photo- 

 micrographs showing these appearances are exhibited. 

 These microscopic observations were fully confirmed by 

 finding well-marked thermo-clectrical and electro-chemical 

 distinctions between the two forms of metal, the hard and 

 soft or the amorphous and the crystalline. The determin- 

 ation of a definite transition temperature at which the 

 amorphous metal passes into the crystalline metal further 

 confirms the phase view of hardening by overstrain and 

 softening by annealing. 



It was subsequently proved that the property of passing 

 from crystalline to amorphous hy mechanical floiv, and 

 from amorphous to crystalline by heat at a definite transi- 

 tion temperature, is a general one which is possessed by 

 all crystalline solids which do not decompose at or below 

 their transition temperature. The significance of this fact 

 I venture to think entitles it to more than a passing 

 reference. It appears to' me to mean that the transition 

 from amorphous to crystalline is entitled to take its place 

 with the other great changes of state, solid to liquid, liquid 

 to gas, for like these it marks a change in the molecular 

 activity which occurs when a certain temperature is 

 reached. It is entitled to take this place because there is 

 every indication that the change is as general in its nature 

 as the other changes of state. Compare it, for instance, 

 with the allotropic changes with which chemists have been 



