ON THE STRUCTURE OF CARBON. by ROSALIND E. FRANKLIN An X-ray investigation of some « amorphous » carbons and graphites has revealed certain new fea- tures which it is the purpose of this note to deseribe. As a preliminary to the wider problems of carbon structure and the dependence of structure on the origin and treatment of the material, a detailed quan- titative study of a single carbon was made, in order to find out just how much information the diffuse X-ray diagram could be made to yield. The mate- rial was prepared by pyrolysis of polyvinylidene chloride at 1.000°, and is more than 99 °% carbon. The following results were obtained. 65 % of the carbon is in the form of highly perfect sraphite-like layers. The mean diameter of these layers is only 16 A. Of the graphite-like layers, about 45 %, show no mutual orientation and 55 % are grouped in parallel pairs with spacing 3,7 A, the number of parallel-layer groups containing more than 2Jayers per group being very small. The remaining 35 % of the carbon is in a form so disordered as to give only a gaz-like contribution to the tolai X-ray scattering. Application of the Fourier transform to the very extensive low angle scattering reveals a mean inter- particulate distance of 25 A, x The details of this structure are, of course, pecu- lar to the carbon investigated, but the results sugges- ted two properties which it might be of interest to investigate in other carbons. ‘The first, a very for- tunate result, was the sharp separation observed between the ordered and disordered parts of the Structure, It might have been expected that in a sarbon Showing such a low degree of crystallinity there would be present all degrees of partial disorder, but this is not so. Apart from the small, perfect, B'aphite-like layers, only highly disordered material ‘S Present. It thus seems clear that the proportion of ordered and disordered material is an important “ature of the structure of such carbons. The other Point of interest is-the spacing, 3,7 A, observed bet- Layer agmeter 4 ween pairs of small parallel graphite-like layers, the spacing in true graphite being 3,35 A. Investigation of a number of other « amorphous » carbons showed that the sharp separation between the ordered and disordered parts is of general occurence. All the X-ray diagrams obtained can be satisfactorily interpreted by supposing the existence only of small, perfect, graphite-like layers together with some highly disordered material. For carbons of widely different origin there is a general relationship between the diameter of the graphite-like layers and the proportion of amor- phous material. This is shown infigurel, For car- he { : i t { ] 7 4 CeCly” 70008 2 (ly Hy Clo), 1000" 3 Sugar fo0o* 4 (C,H;Cl), 1000? “5 (Cp Hy Cla} = BOGS* 22 20 & x 8 42 I { i | I 40 Zo 3e 4O So % Disareered Pig. 1. bons which contain a measurable proportion of disor- dered material the layer diameter is less than 25 A. (fhe method of X-ray analysis used is directly applicable only to carbons which are nearly pure, This explains the absence, in figure 1, of carbons which are more than 50 °% disordered or in which the. layer diameter is less than 12 A.,). nok 574 a For the same carbons, the inter-layer spacing and the mean number of layers per parallel group (as indi- cated by the form of the (002) band) have been measured. It is found that for few (less than 5) layers per group the spacing decreases sharply as the num- ber of layers increases. mean value of 2-3 layers it is 3,6 A, and for carbons having 4 to 5 layers per group it is 3,44 to 3,445 A. With further increase in the number of layers per group the change is small. It will be noted that this value, 3,44 A, is still markedly different from 3,35, the spacing in graphite. The carbons mentioned so far are all substances which show only 2-dimensional order. That is, the gra~ phite-like layers are grouped parallel to one another and equidistant, but are not otherwise mutually orientated. Vhe X-ray diagrams of such materials show only (0 0 J crystalline reflections and 2-dimen- sional (kh k) bands of the type described by Warren (Phys. Rev, 59, 693, 1941). During further graphi- tisation:- induced by thermal treatment — there is a eradual and ‘continuous deformation of each (A k) band tending towards the transformation of the band into a,series of (h kD) crystalline reflections, which are af first very. diffuse but become more sharply defined as the process proceeds, All carbons which show undeformed (hk) pands have an inter-layer spacing nol less than 3,44 A, When deformation of. the (h k) bands sets in, showing the existence of some degree of 3-dimensional order, the inter-layer spacing again decreases. All carbons showing deforined {h k) bands or diffuse (A k Q reflections have appa- rent inter-layer spacings intermediate between 3,44 A Doar of | CRYSTALLINE ~ PUR TARE DDE nc a an ne SPALING . {Random Tease a von on Roravar } SPacwe 3,44 a and 3,35 A.. In reality it appears that fhe ¢hange in spacing from 3,44 to 3,85 A is discontinuous. . When neighbouring layers are mutually orientated as in a graphite crystal the spacing is 3,35. A. - When this For 2 layers it is 3,7 A, for a ROSALIND E, FRANKLIN orientation is destroyed by a random translation or rotation of a layer in its own plane, then the spacing is increased to 3,44 A (See fig. 2). The apparent intermediate spacings observed are in reality average values. This is shown by the figure 3. From the 344 T “r T + T — “T + aaa, é : 343b ‘4 3d2h , 4 o é , 3425 7 4 , é = 8407 , 4 3 - a A ® 330b a 4 5 ? = sash “ve 4 5 : ’ 2 ’ on a R87b uc 4 *, cae 3365 ev? 4 Ler ASB 4 i i 2. i. i Z 4 i. 2 ar) 3 4 5 6 7 8 9 Fraction of layers displaced Fig. 3. form of the deformed (Ak) bands (or diffuse (hk D reflections) it is possible to calculate the proportion ‘of layers at which a translation or rotation, equiva- lent to a break in the crystalline order, or «mistake» . occurs (fig. 2). This calculation -has been carried out, and the apparent inter-layer spacing (given by the (002) line) measured, for a number of carbons of different origin heated to temperatures between 2.300 and 3.000°C, In Figure 3 it is seen that the apparent inter-layer spacing is a function of the pro- portion of « mistakes » or displaced Jayers in the structure. For a mistake at each layer --- that is, in the absence of any crystalline order -— the spacing 3,44-A is obtained. The spacing obtained by Nelson and Riley (Proc. Phys. Soc., 57, 477, 1945) for the best-cerystallised natural graphite available to them is 3.354 A, and this value has been confirmed during the course of the préscnt work. * we Up ‘to the present no attempt has been’ made to correlate the structural factors deseribed- above with v ON THE STRUCTURE OF CARBON 575 the chemical behaviour of the carbons, but there are certain obvious lines which it is hoped to follow in the nearfuture. In the case of the «amorphous » carbons it would be of interest to investigate whether the ordered or disordered part is attacked preferentially in combustion or, other reactions. In the case of carbons showing some 3-dimensional order it may be asked whether or not planes showing the 3,44 A spacing are substantially more available to attack than those which have the true graphite spacing of DISCUSSION SUR LES DEUX COMMUNICATIONS PRECEDENTES _ M. Riley. — In an amorphous structure, it is net permissible to Jeave valency bonds unsatisfied, in other words, it is not merely a chaotic jumble of carbon atoms. A stracture is invoived, but one which, is not sufficiently ordered to give coherent scattering of X-rays. Miss Franklin’s paper suggests that we must identify the suggested three-dimensionally, cross-linked, aromatic structure with the disordered phase in’ the samples studied. M. Bangham. The studies of fuel carbons by X-ray methods made by Riley, Franklin and others have indicated that ordered crystallites of graphite form only part of Lhe whole structure. It is a little tempting to identify the « amorphous » carbon as the portion which contributes most to the « reacti- vity » Instead of doing so, however, I would like to offer the suggestion that the most reactive carbon atoms are those In transition between the amorphous (or other metastable) state and the stable graphitic phase. Whilst it is known that above a certain temperature the ordered portion. increases at the expense of the disordered, it is well-nigh impossible to obtain data as to the rale al which this process takes place; this is “because one cannot, in practice, readily change the touipera- ture of a sample of carbon from one steady value to anotherin an instant of time. Some data of H, L. Riley make it pro- bable that the change associated with a marked rise of tempe- rature occurs fairly rapidly —- in a matter of minutes, perhaps, but not of days. In considering how the more ordered structure is evolved from the less ordered, it appears necessary to suppose that Migration, cither of carbon atoms or of small groups thereof, can take place. These migrating atoms (er groups) will surely contribute very much to the reactivity towards gases. [would emphasise the iniportance of this transition state, bearing in mind the distinction made by Grone between the feactions undergone by a fuel particle which is being heated up, and those associated with its-burning out; with technical fuels the carbon generally has not been previously heated to temperatures such as are attained in fuel beds. It would perhaps be interesting (though difficult) to study, from the theoretical point of view, the magnitudes of the energy barriers separating (say) amorphous from graphitic carbon, making different assumptions as to the size of the migrating unit. Tt might be found, for example, that the migration of individual carbon atoms is the most probable mechanism of the change; or, alternatively, that several atoms move as a group. The progress of the combustion of a sample of commercial carbon must provide conditions which favour the production -of these migrating fragments. We are thus led, as in the study of homogencous gas-phase reactions, to think in terms of the elementary events which may control (a) the formation and (bh) the destruction of these migrating reaction centres, Viewed from this stand point, the «reaction vessel » becomes a two-dimensional mobile surface phase in which the simple gas-phase species (QO, CO, etc...) are somewhat concentrated hy the action of ordinary Van der Waals adsorption: One must.remember that in such a phase — approximating to a. condensed phase -- dissipation of energy by three-body colli-. sions would be frequent. M. Goldfinger. —- Ll résulte de ces discussions, que les théo- riciens aussi bien que les expérimentateurs admettent que le graphite réagit d'une maniére un peu semblable A celle des radicauk libres. Plus précisément, les indices de valence libre de M. DAUbDEL indiqueraient quwil s’agit @une réactivilé inter- médiaire entre celle des radicaux libres et des molécules satu-. rées, Li semble que de tels radicaux « serni-libres » existeraient aussi én phase gazeuse ¢t nous espérons pouvoir publier pro- chainement des résultats a ce sujet. Serait-il possible d’étudier ecla par la conversion H,-para ? ou bien a-t-on déja fait des mesures de conversion sur du graphite traité de différentes miantéres, ou de réactivité différente ? M. Duval. — Réponse & M. Goldfinger. —- Tl semble bien Vaprés les expériences de Bontiomrrer, Farkas et RomMMEL quil existe une-conversion ortho-para due seulement.aux pro- prictés paramagnéliques des atomes de carbone non saturés, ii faut seulement craindre qu’éa basse température la réaction ne soit perturbée par des traces doxygéne paramagnélique (adsorbé moléculairement). Aux températures supérieures A la température ordinaire, d’autre. parl, la réaction peut étre accompagnée de la conversion par dissociation de l’hydrogéne (par adsorption activée), dans laquelle les impuretés peuvent | jouer un réle, M. Brusset. —- Les communications préeddentes viennent dinsister sur Vinsportance de Vétat cristallin et la complexité de la structure dans le cas du carbone amorphe. di faut tenir compte dans la comparaison de réactions d’oxy- dation du carbone non seulement de létat général cristallin miais aussi d’effets locaux @orientation privilégiée. LA encore Vexamen aux rayons X peut élre d’um certain secours. Avec du carbone vraiment a état de graphite on voit que les facteurs qui permettent de caraclériser Pétat et Varchitec- ture du solide carbone soumis 4 Voxydation sont moins nom- breux et plus facilement déterminables,