272 



NATURE 



[November 4, 191 5 



IRON AND ALLOTROPY. 

 TV/TAN Y points of theoretical interest as well as of 

 ■'•'-*■ practical importance were brought forward in 

 the discussion " On the Transformation of Pure Iron " 

 opened by Dr. A. E. Oxley, of the University of 

 Sheffield, at the Faraday Society on October 19. The 

 subjoined summary describes the main views pre- 

 sented. 



It is now believed by many metallurgists that the 

 A2 transformation of iron can be explained without 

 assuming allotropic change, but the A3 transformation 

 is regarded as involving allotropy. The crystalline 

 state is one of extreme molecular association (physical 

 polymerism or the grouping of chemical molecules 

 under the influence of mutual physical forces), and if 

 allotropy is determined solely by the nature and extent 

 of this association each substance will show an un- 

 limited number of allotropic forms within a finite 

 temperature interval. On this view the attempts to 

 show that A3 involves allotropic change while Aj does 

 not, need not have been made. 



To surmount the difficulty presented by the exist- 

 ence of so many allotropes, two different types of 

 allotropy have been recognised : (i) the two-phase 

 (discontinuous) type, (2) the one-phase (continuous) 

 type. Having adopted these, we have now to deter- 

 rnine experimentally to which type a given transforma- 

 tion belongs. The burden is thus thrown on our 

 experimental refinements, and the difficulty of draw- 

 ing a sharp distinction is at once apparent — the dis- 

 tinction can only be arbitrary. In this connection the 

 liquid crystalline state is Interesting. There are some 

 substances which show no liquid crystalline state on 

 fusion, but do so prior to recrystalHsation. Thus we 

 have two-phase allotropy on heating and one-phase 

 allotropy on cooling. 



Defining allotropy as a difference of atomic structure 

 of the chemical molecule (consistent with the trans- 

 formation from oxygen to ozone, or from one isomer 

 to another), a distinction can be drawn between allo- 

 tropy and crystalline grouping. Allotropic modifica- 

 tions \vill form characteristic space lattices determined 

 by their molecular constitutions. It does not neces- 

 sarily follow that a difference of crystalline symmetry 

 implies allotropy, for identical molecules can be packed 

 together in different ways (Barlow and Pope, Trans. 

 Chem. Soc.. vol. Ixxxix., p. 1741, 1906). We must 

 distinguish between the forces holding the molecules 

 m a definite space lattice and those holding the atoms 

 in the molecule. 



Now on the theory of the molecular field Weiss has 

 shown (Comptes rendus. vol. cxllv., p. 25, 1907) that 

 the so-called ^ and y forms of iron have the same 

 Curie constant, and A., is not an allotropic change 

 pomt. In a later research (Journ. de Phys., vol. 1., 

 P- 965, 191 1, and Arch, des Sciences, 1913), using more 

 recent data, and assuming that each atom possesses 

 an integral number of magnetons, Weiss shows that 

 the transition /? to 7 may be represented by a change 

 from a tri-atomic to a dl-atomic molecule. If, how- 

 ever, the magnetic nartlcle of the y state consists of 

 three molecules, and that of the /3 state of two mole- 

 cules, the matrneton theory will still hold, but now 

 the number of atoms in a molecule of each state Is 

 the same. This latter view seems more probable, for 

 y iron possesses more magnetons per molecule than 

 ^ Iron does, and therefore unless we suppose that the 

 molecules of the /3 state are so bound together that 

 one cannot rotate without dragging along its neigh- 

 bours, it seems difficult to account' for the rapid in- 

 crease of susceptibility on cooling through A,. 



If cooling through A, is accomoanied bv a closer 

 grouping of the molecules In the direction (character- 

 istic for each crystal) of spontaneous magnetisation, 

 NO. 24OT, VOL. 96] 



there will be no change of crystalline symmetry, while 

 the increased interaction of the molecules will give 

 rise to ferro-magnetism. The change of molecular 

 distances in the perpendicular directions is not impor- 

 tant from a magnetic point of view, each molecule 

 being mainly constrained by the one in front and the 

 one behind In the direction of the magnetic axis. 

 (Silver iodide (hexagonal) is an example of a crystal 

 which expands along the axis while contracting in 

 perpendicular directions.) Thus this interpretation is 

 not necessarily inconsistent with known magneto- 

 striction data or with determinations of thermal ex- 

 pansion (linear or volume) of a mass of iron crystals. 



The thermal evolution at A3 Is due to the readjust- 

 ment of molecular distances. The rise of temperature 

 observed is 14° C. (Arnold, B.A. Report, Sheffield, 

 1910), and taking o-i as the specific heat of iron, the 

 thermal evolution is 1-4 cals./gram. This value is 

 small compared with the latent heat of fusion of 

 elements, e.g. P (5), Bi (13), Cd (14), Pb (5), Ag (22), 

 Sn (14), Zn (28), Ga (19). Further these latent heats 

 are in general small compared with the thermal evolu- 

 tions in known allotropic and isomeric transforma- 

 tions. Thus 



( 96 gr. 02006 = 96 gr. oxygen + 59, 200 gr. cals. 

 Allotropic] 31 gr. yellow phosphorus = 3i gr. red phosphorus -{- 



I 21,000 gr. cals. 



l' 78 gr. dipropargyl = 78gr. benzene-)- 100,000 gr. cals. 

 Isomeric -j 58 gr. ally) Rlcohol -58 gr. acetone -I- 18,600 gr. cals, 



I =58 gr. propaldehyde-l-22,6oo gr. cals.^ 



The transition 2O3 — >-302 has the thermal value 

 4-600 cals./grm. The transition /3 — ^y in iron has 

 the thermal value —1-4 cals./grm. Can we regard 

 this latter transition as 2Fe3 — >-3Fe2? 



The small thermal evolution at A3 favours the view 

 expressed above that the transformation Involves a 

 molecular regrouping, similar to that occurring in 

 the ordinary process of crystallisation, rather than a 

 rebuilding of the atomic structure of the molecule. 

 Hitherto we have not considered a change within the 

 atom itself, such as a variation of the electric or mag- 

 netic elements. Some such change must occur in the 

 iron atom, as it enters into different chemical com- 

 binations, otherwise how can we explain why iron 

 carbonyl, Fe(CO),,, is diamagnetic, while ferrous 

 chloride, Fe(Cl)2, Is strongly paramagnetic? Do we 

 not here have a kind of atomic allotropy? In this 

 sense the allotropic theory may be consistent with the 

 carbidje theory which attributes the properties of 

 carbon-steels to definite compounds of iron with 

 carbon (such as cementite (Fe,C)). Magnetic pheno- 

 mena appear to be so definitely related to the atom 

 that the existence of different types of iron atom is 

 suggested Inevitably, and the suggestion would not be 

 at variance with modern views of atomic structure. 

 Many arrangements of electronic orbits in dense atoms 

 are theoretically possible. 



The work of Prof. W. H. and Mr. W. L. Bragg 

 must be considered in relation to any phenomenon of 

 crystals. They have shown the difficulty, even in 

 ordinary cases, of even defining the molecules of the 

 crystals, although In many cases this is possible. 

 But to determine by their method whether, in an iron 

 crystal, any atom has a special relation to one of those 

 surrounding it, would be nearly impossible. The 

 optical effects which they investigate are determined 

 only by the nucleus or core of the atom, and the outer 

 arrangements in the atom, determining its allotropic 

 forms, might differ considerably without being capable 

 of detection except perhaps by their magnetic proper- 

 ties. On these lines, therefore, no immediate objec- 

 tion to the existence of such different types of atom 



1 Vitie Muir and Wilson, " Elements of Thermal Chemistrj-," pp. 250-5^, 

 fjr many other examples. 



