AMORPHOUS CARBON AND GRAPHITE By H. L. RILEY D.Sc. F.R.I.C. (PROFESSOR OF INORGANIC AND PHYSICAL CHEMISTRY KING’S COLLEGE NEU-CASTLE-UPON-TYNE) GEOMETRICAL factors p1ay an important r d e in the molecular architecture of solids molecules atoms and ions all possess definite size and shape ; homopolar valency bonds are directed in space and it is therefore difficult to conceive hon- a solid particularly one in which homopolar bonds play a major part can liavc a completely random crystal structure. The examina- t ion b\- S-rag diffraction inethocls of substances which show no obvious crJ-stnlliiic cliaracterist ics nud t)rc\-iously were considered amorphous e.g. vit reoils niid highly dispersed solids niiimnl and vegetable fibres soaps ctc. has shoi\-~i thiit t hc great ninjoritjr posscss structural characteristics which are t>-picnl of thc cr\-stnlline state.As n rcsnlt of the X-ray examina- tion of a nuni1)cr of specimens P. De1jJ-e and P. Scherrer coiiclucled that amorphous carbon is J I ~ C I T ~ J - gra1)liite in a state of sub-tlivision so fine that it coultl iievcr lm rcachctl 11)- incc~liniiici~l means. P. P. ven Weimarn and T. Hagiivara mailit iiine(1 froin cslwrimeiit s on the precipitation of barium sulphate that ewii 1~1leii X-ray tiiffr action photographs indicate that a substance is aniorl)lious tlii.3 niii\t not be taken as cz proof that it is so. Glassy or vitreous so1 ids inalio 1111 ail iiiiportant group of the so-called amorphous substances. -1 cornparison of X-ray powder photographs of silica wollastonitc sodium ciiboratc selenium potash and soda felspars boric oxicle complcx silicates glucose aiitl sucrose in the vitreous and the crystalline state led J.T. Randall H. Y . Rooksby and B. S. Cooper to put forward the “ cr\-stallitc ” theory of tlic vitreous state. They showed for example that the main featiires of the diffraction pattern of vitreous silica can be accouiitecl for by the assumption that it consists of exceedingly minute crystals of crj-stoljnlite wit 11 avcragc linear dinicnsions of the order of 15 A . and lattice coiistants sonic G.G% grcater than those of large crysto- balite crystals. pointed out that this suggestion leads t o discrcpaiicies bctwcen the observed and the calculatcd densities and is not in accord with tlic cliwacteristic mechanical and thermal properties of the glasses he suggested that the ultimate condition for the formation of a glass is that the substance can form czii extended three-dimciisional network of atoms lacking periodicity with an energy content comparable with that of the corresponding crystal network.put forward what is perhaps a compromise betwccn tlic above views namely that thc formation of a glass is due to the presence of large or irregular groups of atoms too cumber- some for direct addition to the lattice. RI. L. Huggins Kuan-Han Sun and A. Silverman6 have pointed out that a solid will readily assume the If7. H. Zacliariasen G. Haigg Physikal. Z. 1917 8 291. 2 KriBt. 1930 75 196. I . Anier. Chem. SOC. 1932 64 3841. J . Chem. Physics 1935 3 42. .J. Anier. Cerclni. SOC. 1943 26 393. 59 2 Kolloid-Z. 1926 38 120 ; Kolloidclicnr. Ueih. 1927 23 400.60 QUARTERLY REVIEWS amorphous state if the regular network structure (crystalline) has practically the same energy content as the irregular in certain glasses besides homo- polar valency bonds hydrogen bonds also probably have an important function in building up the irregular network. In an attempt to show how far our knowledge of the structure of the so-called " amorphous " carbons is compatible with the above views we shall first of all consider the crystal structure of graphite then the crystal chemistry of amorphous carbon and of its transition into graphite and fhally discuss its relationship with other amorphous substances. Throughout we shall use the term " amorphous carbon " in its usual sense viz. for those forms of black carbon other than macro-crystalline graphite which possess no obvious crystalline characteristics.The Graphite Lattice.-The graphite crystal structure which first received wide acceptance was that deduced by J. D. Bernal and by 0. Hassel and H. Mark 8 and later confirmed by C. Mauguin and H. Ott lo it consists of carbon atoms arranged in flat honey- comb-like layers which are stacked parallel to and equidistant from each other in such a way that half the atoms in one lnyerlie normally above half the atoms in the layer beneath while the other half are normally above the centres of the hexagons of the layer beneath altcrnate layers lie atom for atom normally above each other. (a) (61 The C-C spacing in the layers is 1-42 A. whilst adjacent layer-planes are 3.35 A. apart.ll D. S. Laidler and A. Taylor,f2 however pointed out that this structure does not account for the presence of certain faint lines in the X-ray powder photographs of both natural and artificial graphites these lines had previously been reported by G.I. Finch and H. Wilman l3 in electron-diffraction photographs of graphite. To explain the presence and intensity of these additional diffractions H. Lipson and A. R. Stokes l4 suggested that many graphites have a cornyosite structure made up of about 80% of the ordinary (Bernal) structure 6% of a disordered structure (turbo- stratic) and 14% of the structure originally suggested by Debye and Scherrer (Zoc. cit.) in this structure the flat honeycomb planes are stacked parallel to each other as in the Bernal structure but instead of alternate layers being normally above each other the layers follow an abcabc sequence the third layer being symmetrically related to the planes above and below (see Fig.1 a and b ) . These two ways of stacking hexagon layer-planes are FIG. 1 reminiscent of the manner in which close-packed planes of spheres can be Proc. Roy. Soc. 1924 A 106 749. Bull. Soc. franp. Min. 1926 49 32. 8 Z . Physik 1924 25 317. lo Ann. Physik 1928 85 81. l1 J. B. Nelson and D. P. Riley Proc. Physical SOC. 1945 67 477. l2 Nature 1940 146 130. l 4 Ibid. 1942 A 181 101. l a Proc. Roy. SOC. 1936 A 155 345. RILEY AMORPHOTTS CARBON AND GRAPEUTE 61 13.9 2.45 11.7 ' 2.28 10.8 j 1.99 11.5 2-02 13.2 1 2-02 stacked to give either the close-packed hexagonal or the face-centred cubic structure according to whether the stackings follow the ababab or the abcabc sequence respectively.The phenomenon may have some connection with the metallic character of the inter-planar valency bonding in graphite. The lattice structure of graphite however is not yet fully elucidated J. Gibson l 5 has recently obtained powder photographs of ash-free graphites both natural and artificial which show some 16 faint lines which cannot be indexed on the basis of either of the two structures described above. In addition to the crystallographic there is also physical and chemical evidence in support of the layer-lattice structure of graphite. Its pro- nounced electrical inagnet'ic and mechanical anisotropy is in keeping with this structure so also are the close parallelism between the chemical properties of graphite and those of the triarylrnethyls,ls and the lamellar reactions of graphiteel' Amorphous Carbon.-Certain types of amorphous carbon show crysto- chemical reactions similar to those of macrocrystalline graphite ; e.g.0. Ruff and 0. Bretschneider 18 state that norit an active carbon freed from oxygen and water vapour by ignition reacts with elementary fluorine at 280" to form a solid monofluoride containing 57.3% of fluorine [theory for (CF)n 61.27y0] which they suggest has fluorine atoms intercalated between hexagon layer-planes of carbon atoms as in the compound made from macrocrystalline graphite. U. Hofmann and A. Frenzel19 have shown that certain apparently amorphous carbons give good yields of graphite oxide when oxidised with potassium chlorate in the presence of concentrated sulphuric and nitric acids (see Table I). Sugar charcoal however when treated similarly was 93.6 I 92.3 83.3 87.6 20.2 TABLE I Composition and Yield of Graphite Oxide Carbon used.Ceylon graphite . . . . . . Acheson graphite. . . . . . Carbonmonoxidecarbon . . . Retort carbon . . . . . . Petroleum soot . . . . . . 1 Percentage composition. Percentage yield of 1 graphite oxide 1 i 1 calculated on I 1 C. i H,O. 1 C 0. carbon present. 53.4 53.3 ~ 52-3 oxidised practically completely to gaseous and soluble colloidal products no graphite oxide being formed. This reaction has been used as a means of lb Nature 1946 158 752. l6 H. L. Riley J . Inst. Fuel 1937 10 149. l7 Detailed summaries of the lamellar reactions and compounds of graphite have been given by W. Ritdorff Wien. Chem. Ztg. 1944 47 172 and H. L. R,iley FveZ 1946 24 1. 1* 2.anorg. Chem. 1930 217 1. l b Ber. 1930 B 63 1248. 62 QUARTERLY REVIEWS differentiating amorphous from graphitic carbon. M. Berthelot in 1870 defined graphite as “toute varikt6 de carbon susceptible de fournir par oxydation un oxyde graphitique.” W. A. Selvig and W. C. Ratliff 2o used the Brodie reaction (oxidation with fuming nitric acid and potassium chlorate) to determine the proportion of graphitic carbon in coals and cokes and from their results concluded that the former contain no graphitic carbon whilst the latter contain only a small amount (< 1%). Hofmann and Frenzel (Zoc. cit.) on the basis of their X-ray results (see below) however argue that these differences in graphite oxide yield are due t o differences in crystallite size and packing rather than to any fundamental difference in the crystal structures of graphitic and amorphous carbons the much smaller crystallites of the chars and cokes are more readily oxidised to soluble products than the larger crystallites in graphitic carbons.It is significant however and we shall return to this point again that the amorphous carbons which give relatively large yields of graphite oxide are those made by deposition from the gas phase. K. Fredenhagen and G. Cadcnbach 21 found that potassium rubidium and czsium react not only with macro-crystalline graphite but also with soot pure arc carbon wood charcoal and active carbon ; whereas sodium reacts with soot but not with graphite. From more detailed investigations on the interaction of potassium and a soot made by the incomplete com- bustion of a tar oil K.Fredenhagen and H. Suck 2 2 concluded that under the same conditions graphite and soot take up practically the same amount of potassium the establishment of equilibrium with the soot required much longer than with the graphite ; the pretrcatrnent of the soot played some part in determining the velocity of the absorption and desorption of the potassium. These results and those of the X-ray study of (C,K) and (CI6K),$ by A. Schlecde and 31. Wellmann 23 suggest that lamellar structures are present in soot. The intercalation of bromine between the layer-plancs of the graphite crystal lattice has been investigated by W. R t i d ~ r f f ~ ~ who found that graphites take up about 80% of their weight of bromine whatever their state of subdivision active carbon takes up about 170% of its weight of bromine probably because of its morc highly developed crystnllite surface.H. H. 1,owvry and S. P. Jloigan 25 prepared graphite in a highly dispersed state by first oxidising it to graphite oxide and then subjecting the product to thermal decomposition and repeating this treatment on the resulting graphite the final product possessed from one-third to one-quarter of the adsorptive power of the best active charcoal. E. Herl K. Andress L. Rein- hardt and W. Herbert 26 convliicled from X-ray and electrical conductivity measurements that there is 110 truly amorphous carbon in active carbons its properties differ from thosc of graphite because of the size and mesomor- phous character of its crystallites ; the elccbrical conductivity of samples 20 J. Physical C‘/ic111.1!125 29 1103. 21 Z. anorg. Chevt. 1926 158 549. 2 2 Ibid. 1929 178 353. 23 2. physikal. Chem. 1932 B 18 1. 2 4 2. a,Lorg. C’/ic)ti. 1941 245 383. 26 2. p k y s i l i ~ l . Chem. 1932 158 273. 25 J . Physical Chem. 1925 29 1105. RILEY AMORPHOUS CARBON AND GRAPHITE 63 prepared a t high temperatures way approach that of macrocrystalline graphite. The above properties of amorphous carbon are qualitatively in keeping with the DebyeScherrer hypothesis that they consist of exceedingly minute graphite crystallites. There are however amorphous carbons which possess properties which this hypothesis cannot explain. Cokes and chars always contain an appreciable proportion of chemically combined foreign matter ; e.g. carbonisation of carbohydrates up t o 1000" gives chars containing 1-2% of combined oxygen which is not eliminated by prolonged heating in a vacuum at the same or a lower temperature ; only prolonged heating a t much higher temperatures eliminates this oxygen.That it plays an important r6le in the structure of amorphous carbon is indicated by studies of nitrogenous and sulphurous carbons. Amorphous carbon does not take up nitrogen when strongly heated in the gas ; if' how- ever it is heated in gaseous ammonia hydrogen cyanide is formed and the carbonaceous residue may contain up to 3% of nitr~gen.~' E. Terres 28 studied the carbonisation of nitrogenous organic compounds such as glycine asparagine etc. and obtained nitrogenous cokes. W. Hook 29 and H. E. Blayden J. Gibson and H. L. Riley 30 have contrasted the analysis of chars prepared at various temperatures from carbohydrakes on the one hand and from glycine on the other it is significant that the former even those prepared a t lOOO" all contain appreciable quantities of oxygen ; the latter all contain nitrogen but oxygen is present only in the chars prepared a t the lower temperatures.The 1200" glycine char contains 3.64% of nitrogen but no oxygen. Sulphurous chars have been prepared by heating sucrose char with elementary s ~ l p h u r ~ l by carbonising organo-sulphur compounds 32 and by heating semi-coke filter-paper and wood in a current of sulphur dioxide.33 The sulphur in these products cannot be removed by exhaustive extraction with sulphur solvents and in some cases amounts to several per cent. I n addition to oxygen nitrogen or sulphur cokes and chars even when prepared at high temperatures always contain appreciable amounts of hydrogen These foreign atoms all probably play an important part in disordering the atomic arrangement in amorphous carbons.The X-ray powder photographs of amorphous carbons are strikingly different from those of macrocystalline graphites not only are the diffrac- tions much broader but they are also fewer in number most specimens giving only two or three highly diffuse haloes instead of the numerous sharp diffractions in the graphite X-ray photograph. A much deeper insight into the crystallographic character of amorphous carbon has been obtained from 27 W. G. Mixter Amer. J . Sci. 1893 45 363. 28 J . Qaibekucht. 1916 59 619. 2B Carnegie Schol. Mem. Iron and Steel Imt. 1936 25 81. so " The Ultra-Fine Structure of Coals and Cokes," B.C.U.R.A.London 1944 s1 J. P. Wibaut Rec. Trav. chim. 1919 38 159 ; Proc. K . Akad. Wetensch. Amster- 32 R. Ciusa Gazzetta 1922 52 130 ; ssF. Fischer and A. Pranschke Brennstoff-Chem. 1928 9 361. 176-231. dam 1921 24 92. 1925 55 385. 64 QUARTERLY REVIEWS the detailed study of these X-ray powder photographs. The diffracting units are of colloidal dimensions and therefore because of their low resolving power give diffuse diffractions there is a quantitative relation between this diffuseness and the size of the diffracting unit. The first explanation of the disappearance of all the hkl diffractions and most of the high-order diffrac- tions in bhe powder photographs of amorphous carbons was put forward by Berl Andress Reinhardt and Herbert 26 and by H. Arnfelt 34 it is briefly as follows.I n these finely crystalline or so-called amorphous carbons the hexagon layer-planes are stacked parallel t o each other in the individual crystallites with a spacing approximately equal to or somewhat greater than the corresponding spacing in the graphite crystal ; the individual hexagon layer-planes however are otherwise orientated in a completely random manner. J. Biscoe and B. E. Warren 36 have coined the word turbostratic to describe such a crystallographic structure. It can give rise to only 001 and hk0 (more correctly hk) diffractions the hk bands are cross- lattice diffractions formed by the individual hexagon layer-planes acting as two-dimensional diffraction gratings. Such diffractions are not symmetrical but show " tails " towards greater values of 0 .B. E. Warren 36 has discussed in detail this type of diffraction and derived a formula relating the broadening of the cross-lattice band with the linear dimension of the diffracting net-plane ; this is of the same form as the Schemer formula for three-dimensional crystallites but the numerical constant instead of being approximately unity is 1-84 1,843 Linear dimension of crystallite L = ___ p cos 8 where 3 is the wave-length of the X-radiation p the breadth of the difiaction a t half its peak intensity (half-peak width) and 0 the Bragg angle. Warren has also shown that with very small crystallites there is a small displacement of the diffraction maximum of the cross-lattice band towards a larger value of 8 0.163. d(sin0) - - L If a correction is not made for this displacement the calculated C-C spacing within the hexagon layer-planes will be too The assessment of the average.size and shape of the cryatallites present in amorphous carbons from the broadening of the diffraction bands in X-ray powder photographs has been attempted by several workers.Unfortunately differeiit niet hods of calculation have been employed pa&icularly in correct- ing the iticasured hdf-pcilk u-idth for the width of a diffraction from a similar specinien made up of large crystals the results reported by dieerent workers t herefow arc often not strictly comparable. Nevertheless the results reported for similar carbons are of the same order of magnitude. Because 3 4 A r k . Nat. Ast. Fys. 1932 23B No. 2. 37 Cf. for examplo H. E. Blayden H. L. Riley and A. Taylor J .Arner. Chem. S O C . 19.10 62 180 ; .J. H. de Boor Rec. Trav. chim. 1940 59 828 ; U. Hofmann XL'atrrrwis.y. 1944 260. J . Appl. Physics 1941 9 492. 313 Physical Rev. 1942 49 693. RILEY AMORPHOUS CARBON AND GRAPHITE 65 usually only two diffraction bands are of sufficient intensity for accurate measurement it has become customary to express the results of measure- ments of this kind in terms of a hypothetical average cylindrical crystallite the diameter La of which is termed the a dimension and the height L, the c dimension. Some typical results 38 are shown in Table 11. These and the results of other workers indicate that the original Debye-Schemer theory of the crystallographic nature of amorphous carbon requires modifica- tion in a t least two respects 'uiz. the small crystallites in amorphous carbon differ from macro-crystalline graphite in that (1) they are turbostratic and TABLE I1 Crystallite Dimensions (Hofmann and Wilm 38) Carbon.Ceylongraphite . . . . . Carbon monoxide carbon 700" . 9 9 , 550" . 9 9 1 , 420" . Y f Y , 400" . Acetylene soot (explosive decom- position) . . . . . . . Retort carbon . . . . . . Acetylene soot. . . . . . s t , calcined . . . , activated . . . Actiyemrbon AKTIV . . . 9 7 , , , calcined. 9 , , , activated Supranorit . . . . . . . , calcined . . . . , recalcined 30 hrs.. . ) activated . . . . Carboraffi calcined . . . . , aotivated . . . . Gas-mask carbon calcined . . 9 7 , act.iveted . . Sugar carbon . . . . . . 7 , activated . . . C. 99.1 94.9 95.0 91.8 - 99.7 96.9 95.0 99.1 98.5 88.5 96-5 94.4 89.3 97.3 97.7 96.2 93.0 97.4 96.5 98.1 98.0 L Analysis (jo.H. 0.0 0.1 0.1 0.1 - 0.3 0.3 0.8 0.3 0.0 1.7 0.35 0.0 0.6 0.15 0.0 0.6 0.0 0.6 0.0 0.7 0.0 - Ash. 0.3 4 3 2 - 0.0 0.6 0.3 0-3 0-5 1.1 1.1 2.0 0.3 0.4 0.9 1 . 9 3.3 0-5 1.0 0.0 0.0 - Layer- plane spacing C/? A. 3-35 3.4 3.4 3.4 3.45 3.43 3.45 3.55 3.55 3-6 3.5 3.45 3.65 3.5 3.45 3.55 3.7 3.G 3.7 3-55 3.6 3.6 3.6 Crystallite tliriiension A . L O . a 120 150 60 40 45 45 21 26 22 17 25 18 18 26 26 24 13 19 20 17 21 18 &. ca. 200 180 160 70 35 50 40 13 14 16 11 10 8 8 7 7 7 10 7 10 8 9 7 (2) the hexagon-layer spacing is not constant a t 3.35 A. but increases as the number of layer-planes in the crystallite decreases. Even when allowance is made for the uncertainty in the fundamental accuracy 39 of these results the extreme minuteness of these crystallites is striking in the smaller crystallites the individual hexagon layer-planes are not much larger than certain polynuclear aromatic molecules of known structure and the c dimensions indicate that the crystallites can be built up from a.s few a's three or four layer-planes.38 U. Hofmann and D. Wilm 2. EEektrochem. 1936 42 504. 89 a. H. Cameron and A. L. Patterson Amer. SOC. Testing Mat. Symposium on Radiography and X-ray Diffraction 1937. E 66 QUARTERLY REVIEWS It would appear from the above that the process of charring or carbonisa- tion must involve the progressive transformation of the more or less complex organic matter into the polynuclear aromatic ring systems and their subse- quent stacking and growth. Some such process is indicated by the experi- ments of R.C. Smith and H. C. Howard 40 who charred pure cotton cellulose at temperatures ranging from 190" to 400" and subjected the products t o exhaustive oxidation with alkaline permanganate the magnitude of the yields of aromatic acids from the various chars indicated progressive aromatisation as the charring temperature increased. X-Ray investiga- tions have shown however that this simple picture is far from complete. Fig. 2(a) summarises the results of crystallite-size determinations on chars 40 J . Amer. Chem. SOC. 1937 59 234. RILEY AMORPHOUS CARBON AND URAPHITE 87 prepared from cellulose at various temperature^.^^ Results of an identical character have been reported by U. Hofmann and F. Sinkel 42 who worked with sucrose chars. The striking feature of these results is the constmcy of the average height ( c dimension) of the crystallites (experimentally the half-peak width of the 002 diffraction band) over a wide range of carbonising temperature.Crystallites built up merely from polynuclear aromatic layer-planes held together by relatively weak van der Waals forces would not be expected to behave in such a manner but rather to vary their degree of packing with the speed and temperature of carbonisation. This variation would be reflected in changes of the c dimension. Other carbohydrates lignin and wood i.e. organic compounds containing a relatively high propor- tion of oxygen all give chars containing crystallites with c dimensions between 9 and 10 A. Carbonising conditions can be varied between wide limits with respect to both speed and temperature (up to lZOO") without any consequential change in the half-peak width of the 002 difiaction of the resulting char.Cellulose carbonised in an atmosphere of ammonia and also glycine behave in a similar manner but the constant c dimension is somewhat greater viz. between 12 and 13 A. This constancy of the c dimension is highly suggestive and appears to indicate some kind of cross- linking between the layer planes of the crystallites. This hypothesis appears more probable when we consider the X-ray crystallographic behaviour on carbonisation of certain organic compounds which have a relatively small oxygen content. Fig. 2(b) shows the c-dimen- sion curve obtained with cokes made over a range of temperatures from that part of a bituminous coal-tar pitch which is soluble in carbon tetrachloride Similar c-dimension curves 43 are given by pitches certain bitumens bituminous coals the pyridine-chloroform soluble extracts of bituminous coals and certain pure polynuclear aromatic compounds e.g.dibenzanthrone Caledon jade-green etc. Such c-dimension curves appear to be character- istic of the presence in the parent carbonaceous matter of large thermally stable lamellar aromatic molecules. Aromatic hydrocarbons of this type are volatile and apparently the presence of some oxygen and/or nitrogen in the parent molecule is necessary for a large carbon yield the proportion however must not be too large otherwise charring occurs at a low tempera- ture. These c-dimension curve8 have been interpreted as follows. The ascending part of the curve between room temperature and about 50O0 indicates a unidimensional crystallisation brought about by the increased packing of thermally stable aromatic molecules under the influence of thermal vibration.The maximum at about 500" reflects the conversion of a system consisting of stable independent organic molecules held together by weak dispersion forces (molecular lattice) into a " carbon " crystallite [rigid three- &%ensionally cross-linked ( ?) lattice]. Dibenzanthrone after being heated slowly up to 475" and cooled still gives its characteristic X-ray difiaction I1 H. E. Blayden H. L. Riley and A. Taylor J. 1939 67. Ia 2. a w g . Chem. 1940 245 85. u H. E. Blayden J. Gibson and H. L. Riley J . Inst. FueZ 1945 War Time Bulletin 117. 68 QUABTERLP BEVIIOWS pattern made up of several more or 106s sharp diffkactione; after heatixq to 500" the pattern characteristic of the dibenzanthrone crystal lattice haa completely disappeared and been replaced by two diffuse haloes character- istic of amorphous carbon.This change molecular lattice -+ carbon occurs over a relatively narrow range of temperature the height of which appears to be a function of the proportion of oxygen in the parent carbonaceous material. Substances like cellulose form rigid (cross-linked ?) carbon crystallites at temperatures little above 200" whereas certain coking cod bitumens retain their molecular identity up to temperatures as high as 550". The interpretation of the descending portion of the c-dimension curve between 550" and 900" [Fig. 2(b)] is not quite so clear. The phenomenon is however of general occurrence whenever a carbonaceous substance on being heated to some temperature below 550" gives a product which contains crystallites having an average c dimension greater than about 12 A.then a further increase of temperature brings about a decrease in the c dimension (experimentally a broadening of the 002 diffraction). The increasing dis- order over this temperature range may be due to the evolution of oxygen as oxides of carbon and a consequent reduction in the extent of tho cross- linking between the hexagon layer-planes. The subsequent increase in the c dimension above 900" is perhaps facilitated by the remnants of the ordered structure which the system possessed at 550". A further type of c-dimension curve is shown in Fig. 2(c) it is character- ised by an abrupt increase in the average c dimension in the temperature range 800-1000".Crystallite growth of this kind is shown by peats to a less extent by brown coals and lignites and also by " bituminised " cellulose and lignin Le. cellulose or lignin which has been heated in an autoclave to about 300" with an excess of aqueous alkali.43 In the case of peat the crystallite growth of a part of the char between 800" and 1OOO" is so great as to suggest some profound atomic rearrangement perhaps a kind of polymorphic change. Although the true nature of this crystallographic change is still not clear J. Gibson 11. Holohan and H. L. Riley44 have pointed out that besides the diamond and graphite lattices there is a third possible spatial arrangement of carbon atoms which may play an important part in the atomic architecture of amorphous carbon.This arrangement can be pictured as follows. Imagine a graphite layer-plane to be made up of discrete hexagons connected together by valency bonds as in Fig. 3(a) ; break the necessary bonds and rotate the hexagons A,B,C etc. inwards through 60" [Fig. 3(b)] ; it will be found that the valencies so freed can be attached without strain to those of similarly tilted hexagons in adjacent planes above and below This can be seen in Fig. 3(c) which shows the spatial relationship of 4 hexagons. The structure 80 formed is a three- dimensional network of the " skeleton lattice '' type full of large holes and long channels and reminiscent of the structure of certain zeolites. The disappearance of the flat layer-planes suggests that the resonance energy of this structure would be less than that of the graphite lattice and its existence therefore less probable.The new structure however contains an orderly 44 J. 1946 458. 69 network of conjugated double linkages running throughout the whole three- dimensional lattice and its resonance energy would therefore be of an appre- ciable magnitude. The existence of such a structure is made more probable by the discovery by I. L. Karle and L. 0. Brockway,45 by means of electron- diffraction measurements) of the non-co-planarity of o-tetraphenylene [see Fig. 3(c)]. The suggested structure is a three-dimensional repetition of o-tetraphenylene residues. RILEY AMORPHOUS CARBON AND GRAPHITE \ FIG. 3 A consideration of carbonising processes which would possibly tend to give some structure other than the flat honeycomb network of carbon atoms led Gibson Holohan and Riley to prepare carbon by (a) carbonising hexa- iodobenzene and (b) the interaction of hexachlorobenzene and sodium amalgam ; (a) gave a carbon which showed practically no coherent scattering of X-rays and ( b ) gave a carbon the X-ray powder photograph of which although highly diffuse showed new features which can be explained by the existence of small volumes of the three-dimensional network described above.The constancy of both the a and the c dimensions [Fig. 2 (d)] when a carbon prepared by method ( b ) (temperature of preparation did not exceed 300") was heated to higher temperatures provides additional evidence of a rigid cross-linked lattice structure These X-ray investigations leave little doubt that the idea that black a,morphous carbons are merely graphite in an extremely fine state of sub- d5 J .Amr. C h . SOC. 1944 66 1974, 70 QUARTERLY REVIEWS division is inadequate and needs further qualification. It is evident that the structure of a particular sample may be greatly influenced both by the nature of the parent carbonaceous material and by the method of carbonisa- tion. It is difficult to conceive for example how a system consisting of relatively free and independent lamellze could fail to develop some degree of order under the influence of thermal vibration and yet carbons can be made which give no coherent scattering of X-rays. Before outlining a modified view of the crystallographic nature of amor- phous carbons we must consider the process of graphitisation i.e. the transition amorphous carbon + graphite.Graphitisation.-The high thermal stability of pure amorphous carbon is shown by the results of Biscoe and Warren 35 (Table 111) who studied the TABLE I11 C'rystallite Dimensions of Carbon Black (Biscoe and Warren) Time and temp. of heating. - 2 hrs. at 760" 2 hrs. at 1040" 2 hrs. at 1500" 2 hrs. at 2000" 2 hrs. at 2800" 2 hrs. at 2000" 2 hrs. at 1040" Layer-plane spacing A. 1 Crystallite dimension A. Frorn002. 1 From001. 3.55 3.53 (3.65) 3.48 3.46 3.45 3,47 3.55 3 4 1 3-43 3.47 3-45 3-44 3.44 3.43 3.46 La. 20.0 22.6 28.0 44.2 55.8 65.2 60.5 29.8 Lc * 12.7 14.4 14.9 24.9 32-3 40.0 35.5 14.9 effect of high temperature on the crystallographic character of a commercial carbon black. Similar results were obtained by P. Corriez 46 who studied the effect of high temperature on the crystallite dimensions of sucrose chars.I n the case of pure carbons it is only a t extremely high temperatures that pronounced thermal recrystallisation occurs. U. Hofmann A. Ragoss and F. Sinkel 47 have shown by means of X-ray diffraction electron-microscope and adsorption studies that the small polycrystalline spherical particles of carbon black on heating for 24 hrs. a t 3000" undergo thermal recrystallisa- tion to an extent limited by the size of the original spherical particle i.e. the indiyidual polycrystalline particles probably become single crystals. This thermal stability of the amorphous carbon crystallites is due of course to the high resonance energy of the hexagon network planes of carbon atoms. It is only when these networks are broken down by some chemical mechanism that pronounced crystal growth can occur at temperatures below 2000".This was realised by E. G. Acheson long before the crystal structure of graphite was known in his original patent on the manufacture of graphite 4% Cornpt. rend. 1935 201 1189. 47 Kolloid-Z. 1941 96 231 ; s00 also D. Wilm and U. Hofmann ibid. 1935 70 21 ; A. Ragoss U. Hofmann and R. Holst ibid. 1943 105 118 ; M. von Ardenne and U. Hofmann 2. physikal. Chenz. 1941 B 50 1. RITAEY ttJIORPHOT7S CARBOX ,IN D GR~IP~IITE 71 he states 4 8 " I hnvc also discoverctl that in order to produce pure graphite from carbonaccous materials thcre is nri indirect conversion and that the act of formation of the graphite is morc in thc nature of an act of dissociation of the carbon from its combination with other materials than a conversion of the ordinary carbon into graphite and that as a preliminary step the carbon has to be combined chcniically with some other material.Thus I have found that if the carbonaccous material or carbon used in the process contains a considerable proportion of mincral matter or if it is mixed with a certain proportion of oxide or oxides such as silica clay alumina mangan- ese lime or oxide of iron and subjected to treatment as hereinafter set forth (high temperature) the yield of graphite is enormously increased and the product is most satisfactory." It is significant too that processes which involve the formation of carbon by deposition from the gas phase often give at relatively low temperatures products which are highly crystalline soe for example the crystallite dimensions of the carbons formed on a catalyst by the reaction 2CO = C + CO, quoted in Table 11.The growth mechanism of the carbon crystallites formed in such reactions apparently involves the initial formation of single carbon atoms which subsequently attach themselves to pre-formed graphite nuclei. Other reactions are known which give rise to highly crystalline graphite at temperatures far below that of thermal recrystallisation they all involve the liberation of individual carbon atoms e.g. the graphitisation of austenite,49 and the decomposition of carbides 5O and calcium cyanamide. 51 Carbons deposited from hydrocarbon gases in the iieighbourhood of lOOO" are not so highly crystalline as those deposited from carbon monoxide a t the same temperature the presence of hydrogen appears to inhibit crystal growth presumably by saturating the valencies of the carbon atoms at the surface of the crystallites.It is reasonable t o suppose however that carbons formed by any process which involves the participa- tion of single carbon atoms will have the most stable structure i e . the flat layer-plane graphite-like structure rather than any cross-linked structure of rather hlgher potential energy. JI. Holohan and R. Iley 52 have shown that in spite of' their highly dispersed state carbon blacks and carbon deposited from the gas phase on vitreous silica a t lOOO" have ignition temperatures which approach and may even exceed those of natural and artificial graphites. Tho X-ray-scattering power of carbons of this type is 48U.S.P. 568,323 29.9.1890 quoted by F.-4. J. Fitzqcrald J . SOC. Chem. Id. 1901 20 443 ; see also H. Ditz Chem. Z t y . 1904 28 167 and W. C. Arsem Met. Chem. Eng. 1911 9 536. 48 F. Wust and C. Geiger StuhZ zl. Eisen 1905 25 1134 1196 ; I<. Iokibe Sci. Rep. Tdhoku Inip. Unip. 1920 9 273 ; L . Sorthcott J . Iron Steel Inst. 1923 108 491. 5 0 H. Ishikawa Elect. Rev. Japan 1931 19 419 493 690 726 824 884 ; C. Hahn and A. Strutz Metallurgie 1906 3 72.3 ; Chem. Ztg. 1907 31 Rep. 33 ; J. West- becker Metullurgie 1904 1 137 ; 2. Elektrochenz. 1904 10 837 ; A. Frank Versamm. Ges. deutschen Naturforscher u. Aerzte Sept. 1905 ; Chem. Ztg. 1905 29 1044 ; V. M. Weaver U.S.P. 1,576,883 16.3.26. 61 E. Collett and M. Eckardt E.P. 5713 7.3.1911 ; A. Reme16 and B. Rassow Z . angew. Chem. 1920 33 139; N. Kameyama J.Chem. Ind. Japan 1921 24 1131. 5a Unpublished work of the Northern Coke Research Laboratory. 72 QUARTERLY REVIEWS much greater than that of chars of comparable crystallite size sulphuroue chars (see above) have a still lower scattering power. Although these Wer- ences in scattering power have as yet only been studied semi-quantitatively they are of such a magnitude that there is little doubt of their si@cance. Conclusion.-The above considerations lead to the conclusion that amorphous carbons are built up from two types of structure vix. (I) the turbostratic lamellar graphite-like structure and (11) the disordered three- dimensionally cross-linked structure (Fig. 3). Amorphous carbons which contain a preponderance of structure I are usually much purer than those made up largely of I1 it is possible that the foreign atoms hydrogen together with oxygen nitrogen or sulphur play some part in stabilising the type I1 disordered structure.Such a view of the crystallographic character of amorphous carbon appears to offer a reasonable basis for the explanation of the great diversity of its properties e.g. the highly specific nature and range of its adsorptive properties the great variation in its chemical reactivity the dif€erences in its mechanical properties and scattering power for X-rays etc. Carbons deposited from the gas phase which may contain little or no combined oxygen will be made up largely of type-I structures whilst cham and cokes in general will consist of intimate associations of both types. A large proportion of oxygen in the parent carbonaceous matter appears to favour the formation of structure 11 whereas carbonaceous matter containing little oxygen and a large proportion of benzenoid carbon e.g.pitches bitumens etc. will give carbons with a higher proportion of structure I. Both struc- tures possess high thermal stability structure I1 appears to persist in Borne specimens at temperatures well above 1000". No mention has been made of the importance of the secondary structure i.e. the cohesion and degree of agglomeration of the crystallites in deter-. mining the properties of amorphous carbon. Carbon blacks used in the rubber industry have been investigated in detail ; 53 little is definitely known however of the secondary structure of chars and cokes their accessible ~urface can be measured but the open lattice present in both structures I and I1 makes the interpretation of such measurements ditlicult.Such is the diversity of the crystal structure of solids that no single comprehensive theory of the amorphous state is possible. Metals and salts can be so disordered by mechanical strain or dispersion that they give diffuse X-ray diffractions. Crystals which are made up of long homopolar chains exist in an amorphous condition when the chains become tangled together e.g. vitreous sulphur and selenium rubber borate-glass etc. Systems in which the molecules are large lamellae can assume the amorphous (or meso- morphous) state because of the ditliculty these cumbersome units have in packing into an ordered lattice e.g. pitches bitumens polynuclear aromatic compounds. Finally there is the disordered three-dimensional network present in certain glasses and vitreous solids.Amorphous carbons should probably be placed in a class intermediate between the disordered lamella and the disordered cross-linked lattice. 68 Columbian Carbon Company " The Partiole Size and Shape of Colloidal Csrboa w Revealed by the Electron Microscope," New York 1940. AMORPHOUS CARBON AND GRAPHITE By H. L. RILEY D.Sc. F.R.I.C. (PROFESSOR OF INORGANIC AND PHYSICAL CHEMISTRY KING’S COLLEGE NEU-CASTLE-UPON-TYNE) GEOMETRICAL factors p1ay an important r d e in the molecular architecture of solids molecules atoms and ions all possess definite size and shape ; homopolar valency bonds are directed in space and it is therefore difficult to conceive hon- a solid particularly one in which homopolar bonds play a major part can liavc a completely random crystal structure.The examina- t ion b\- S-rag diffraction inethocls of substances which show no obvious crJ-stnlliiic cliaracterist ics nud t)rc\-iously were considered amorphous e.g. vit reoils niid highly dispersed solids niiimnl and vegetable fibres soaps ctc. has shoi\-~i thiit t hc great ninjoritjr posscss structural characteristics which are t>-picnl of thc cr\-stnlline state. As n rcsnlt of the X-ray examina- tion of a nuni1)cr of specimens P. De1jJ-e and P. Scherrer coiiclucled that amorphous carbon is J I ~ C I T ~ J - gra1)liite in a state of sub-tlivision so fine that it coultl iievcr lm rcachctl 11)- incc~liniiici~l means. P. P. ven Weimarn and T. Hagiivara mailit iiine(1 froin cslwrimeiit s on the precipitation of barium sulphate that ewii 1~1leii X-ray tiiffr action photographs indicate that a substance is aniorl)lious tlii.3 niii\t not be taken as cz proof that it is so.Glassy or vitreous so1 ids inalio 1111 ail iiiiportant group of the so-called amorphous substances. -1 cornparison of X-ray powder photographs of silica wollastonitc sodium ciiboratc selenium potash and soda felspars boric oxicle complcx silicates glucose aiitl sucrose in the vitreous and the crystalline state led J. T. Randall H. Y . Rooksby and B. S. Cooper to put forward the “ cr\-stallitc ” theory of tlic vitreous state. They showed for example that the main featiires of the diffraction pattern of vitreous silica can be accouiitecl for by the assumption that it consists of exceedingly minute crystals of crj-stoljnlite wit 11 avcragc linear dinicnsions of the order of 15 A .and lattice coiistants sonic G.G% grcater than those of large crysto- balite crystals. pointed out that this suggestion leads t o discrcpaiicies bctwcen the observed and the calculatcd densities and is not in accord with tlic cliwacteristic mechanical and thermal properties of the glasses he suggested that the ultimate condition for the formation of a glass is that the substance can form czii extended three-dimciisional network of atoms lacking periodicity with an energy content comparable with that of the corresponding crystal network. put forward what is perhaps a compromise betwccn tlic above views namely that thc formation of a glass is due to the presence of large or irregular groups of atoms too cumber- some for direct addition to the lattice.RI. L. Huggins Kuan-Han Sun and A. Silverman6 have pointed out that a solid will readily assume the If7. H. Zacliariasen G. Haigg Physikal. Z. 1917 8 291. 2 KriBt. 1930 75 196. I . Anier. Chem. SOC. 1932 64 3841. J . Chem. Physics 1935 3 42. .J. Anier. Cerclni. SOC. 1943 26 393. 59 2 Kolloid-Z. 1926 38 120 ; Kolloidclicnr. Ueih. 1927 23 400. 60 QUARTERLY REVIEWS amorphous state if the regular network structure (crystalline) has practically the same energy content as the irregular in certain glasses besides homo- polar valency bonds hydrogen bonds also probably have an important function in building up the irregular network. In an attempt to show how far our knowledge of the structure of the so-called " amorphous " carbons is compatible with the above views we shall first of all consider the crystal structure of graphite then the crystal chemistry of amorphous carbon and of its transition into graphite and fhally discuss its relationship with other amorphous substances.Throughout we shall use the term " amorphous carbon " in its usual sense viz. for those forms of black carbon other than macro-crystalline graphite which possess no obvious crystalline characteristics. The Graphite Lattice.-The graphite crystal structure which first received wide acceptance was that deduced by J. D. Bernal and by 0. Hassel and H. Mark 8 and later confirmed by C. Mauguin and H. Ott lo it consists of carbon atoms arranged in flat honey- comb-like layers which are stacked parallel to and equidistant from each other in such a way that half the atoms in one lnyerlie normally above half the atoms in the layer beneath while the other half are normally above the centres of the hexagons of the layer beneath altcrnate layers lie atom for atom normally above each other.(a) (61 The C-C spacing in the layers is 1-42 A. whilst adjacent layer-planes are 3.35 A. apart.ll D. S. Laidler and A. Taylor,f2 however pointed out that this structure does not account for the presence of certain faint lines in the X-ray powder photographs of both natural and artificial graphites these lines had previously been reported by G. I. Finch and H. Wilman l3 in electron-diffraction photographs of graphite. To explain the presence and intensity of these additional diffractions H. Lipson and A. R. Stokes l4 suggested that many graphites have a cornyosite structure made up of about 80% of the ordinary (Bernal) structure 6% of a disordered structure (turbo- stratic) and 14% of the structure originally suggested by Debye and Scherrer (Zoc.cit.) in this structure the flat honeycomb planes are stacked parallel to each other as in the Bernal structure but instead of alternate layers being normally above each other the layers follow an abcabc sequence the third layer being symmetrically related to the planes above and below (see Fig. 1 a and b ) . These two ways of stacking hexagon layer-planes are FIG. 1 reminiscent of the manner in which close-packed planes of spheres can be Proc. Roy. Soc. 1924 A 106 749. Bull. Soc. franp. Min. 1926 49 32. 8 Z . Physik 1924 25 317. lo Ann. Physik 1928 85 81. l1 J. B. Nelson and D. P. Riley Proc.Physical SOC. 1945 67 477. l2 Nature 1940 146 130. l 4 Ibid. 1942 A 181 101. l a Proc. Roy. SOC. 1936 A 155 345. RILEY AMORPHOTTS CARBON AND GRAPEUTE 61 13.9 2.45 11.7 ' 2.28 10.8 j 1.99 11.5 2-02 13.2 1 2-02 stacked to give either the close-packed hexagonal or the face-centred cubic structure according to whether the stackings follow the ababab or the abcabc sequence respectively. The phenomenon may have some connection with the metallic character of the inter-planar valency bonding in graphite. The lattice structure of graphite however is not yet fully elucidated J. Gibson l 5 has recently obtained powder photographs of ash-free graphites both natural and artificial which show some 16 faint lines which cannot be indexed on the basis of either of the two structures described above.In addition to the crystallographic there is also physical and chemical evidence in support of the layer-lattice structure of graphite. Its pro- nounced electrical inagnet'ic and mechanical anisotropy is in keeping with this structure so also are the close parallelism between the chemical properties of graphite and those of the triarylrnethyls,ls and the lamellar reactions of graphiteel' Amorphous Carbon.-Certain types of amorphous carbon show crysto- chemical reactions similar to those of macrocrystalline graphite ; e.g. 0. Ruff and 0. Bretschneider 18 state that norit an active carbon freed from oxygen and water vapour by ignition reacts with elementary fluorine at 280" to form a solid monofluoride containing 57.3% of fluorine [theory for (CF)n 61.27y0] which they suggest has fluorine atoms intercalated between hexagon layer-planes of carbon atoms as in the compound made from macrocrystalline graphite.U. Hofmann and A. Frenzel19 have shown that certain apparently amorphous carbons give good yields of graphite oxide when oxidised with potassium chlorate in the presence of concentrated sulphuric and nitric acids (see Table I). Sugar charcoal however when treated similarly was 93.6 I 92.3 83.3 87.6 20.2 TABLE I Composition and Yield of Graphite Oxide Carbon used. Ceylon graphite . . . . . . Acheson graphite. . . . . . Carbonmonoxidecarbon . . . Retort carbon . . . . . . Petroleum soot . . . . . . 1 Percentage composition. Percentage yield of 1 graphite oxide 1 i 1 calculated on I 1 C. i H,O. 1 C 0. carbon present. 53.4 53.3 ~ 52-3 oxidised practically completely to gaseous and soluble colloidal products no graphite oxide being formed.This reaction has been used as a means of lb Nature 1946 158 752. l6 H. L. Riley J . Inst. Fuel 1937 10 149. l7 Detailed summaries of the lamellar reactions and compounds of graphite have been given by W. Ritdorff Wien. Chem. Ztg. 1944 47 172 and H. L. R,iley FveZ 1946 24 1. 1* 2. anorg. Chem. 1930 217 1. l b Ber. 1930 B 63 1248. 62 QUARTERLY REVIEWS differentiating amorphous from graphitic carbon. M. Berthelot in 1870 defined graphite as “toute varikt6 de carbon susceptible de fournir par oxydation un oxyde graphitique.” W. A. Selvig and W. C. Ratliff 2o used the Brodie reaction (oxidation with fuming nitric acid and potassium chlorate) to determine the proportion of graphitic carbon in coals and cokes and from their results concluded that the former contain no graphitic carbon whilst the latter contain only a small amount (< 1%).Hofmann and Frenzel (Zoc. cit.) on the basis of their X-ray results (see below) however argue that these differences in graphite oxide yield are due t o differences in crystallite size and packing rather than to any fundamental difference in the crystal structures of graphitic and amorphous carbons the much smaller crystallites of the chars and cokes are more readily oxidised to soluble products than the larger crystallites in graphitic carbons. It is significant however and we shall return to this point again that the amorphous carbons which give relatively large yields of graphite oxide are those made by deposition from the gas phase.K. Fredenhagen and G. Cadcnbach 21 found that potassium rubidium and czsium react not only with macro-crystalline graphite but also with soot pure arc carbon wood charcoal and active carbon ; whereas sodium reacts with soot but not with graphite. From more detailed investigations on the interaction of potassium and a soot made by the incomplete com- bustion of a tar oil K. Fredenhagen and H. Suck 2 2 concluded that under the same conditions graphite and soot take up practically the same amount of potassium the establishment of equilibrium with the soot required much longer than with the graphite ; the pretrcatrnent of the soot played some part in determining the velocity of the absorption and desorption of the potassium. These results and those of the X-ray study of (C,K) and (CI6K),$ by A.Schlecde and 31. Wellmann 23 suggest that lamellar structures are present in soot. The intercalation of bromine between the layer-plancs of the graphite crystal lattice has been investigated by W. R t i d ~ r f f ~ ~ who found that graphites take up about 80% of their weight of bromine whatever their state of subdivision active carbon takes up about 170% of its weight of bromine probably because of its morc highly developed crystnllite surface. H. H. 1,owvry and S. P. Jloigan 25 prepared graphite in a highly dispersed state by first oxidising it to graphite oxide and then subjecting the product to thermal decomposition and repeating this treatment on the resulting graphite the final product possessed from one-third to one-quarter of the adsorptive power of the best active charcoal.E. Herl K. Andress L. Rein- hardt and W. Herbert 26 convliicled from X-ray and electrical conductivity measurements that there is 110 truly amorphous carbon in active carbons its properties differ from thosc of graphite because of the size and mesomor- phous character of its crystallites ; the elccbrical conductivity of samples 20 J. Physical C‘/ic111. 1!125 29 1103. 21 Z. anorg. Chevt. 1926 158 549. 2 2 Ibid. 1929 178 353. 23 2. physikal. Chem. 1932 B 18 1. 2 4 2. a,Lorg. C’/ic)ti. 1941 245 383. 26 2. p k y s i l i ~ l . Chem. 1932 158 273. 25 J . Physical Chem. 1925 29 1105. RILEY AMORPHOUS CARBON AND GRAPHITE 63 prepared a t high temperatures way approach that of macrocrystalline graphite. The above properties of amorphous carbon are qualitatively in keeping with the DebyeScherrer hypothesis that they consist of exceedingly minute graphite crystallites.There are however amorphous carbons which possess properties which this hypothesis cannot explain. Cokes and chars always contain an appreciable proportion of chemically combined foreign matter ; e.g. carbonisation of carbohydrates up t o 1000" gives chars containing 1-2% of combined oxygen which is not eliminated by prolonged heating in a vacuum at the same or a lower temperature ; only prolonged heating a t much higher temperatures eliminates this oxygen. That it plays an important r6le in the structure of amorphous carbon is indicated by studies of nitrogenous and sulphurous carbons. Amorphous carbon does not take up nitrogen when strongly heated in the gas ; if' how- ever it is heated in gaseous ammonia hydrogen cyanide is formed and the carbonaceous residue may contain up to 3% of nitr~gen.~' E.Terres 28 studied the carbonisation of nitrogenous organic compounds such as glycine asparagine etc. and obtained nitrogenous cokes. W. Hook 29 and H. E. Blayden J. Gibson and H. L. Riley 30 have contrasted the analysis of chars prepared at various temperatures from carbohydrakes on the one hand and from glycine on the other it is significant that the former even those prepared a t lOOO" all contain appreciable quantities of oxygen ; the latter all contain nitrogen but oxygen is present only in the chars prepared a t the lower temperatures. The 1200" glycine char contains 3.64% of nitrogen but no oxygen. Sulphurous chars have been prepared by heating sucrose char with elementary s ~ l p h u r ~ l by carbonising organo-sulphur compounds 32 and by heating semi-coke filter-paper and wood in a current of sulphur dioxide.33 The sulphur in these products cannot be removed by exhaustive extraction with sulphur solvents and in some cases amounts to several per cent.I n addition to oxygen nitrogen or sulphur cokes and chars even when prepared at high temperatures always contain appreciable amounts of hydrogen These foreign atoms all probably play an important part in disordering the atomic arrangement in amorphous carbons. The X-ray powder photographs of amorphous carbons are strikingly different from those of macrocystalline graphites not only are the diffrac- tions much broader but they are also fewer in number most specimens giving only two or three highly diffuse haloes instead of the numerous sharp diffractions in the graphite X-ray photograph.A much deeper insight into the crystallographic character of amorphous carbon has been obtained from 27 W. G. Mixter Amer. J . Sci. 1893 45 363. 28 J . Qaibekucht. 1916 59 619. 2B Carnegie Schol. Mem. Iron and Steel Imt. 1936 25 81. so " The Ultra-Fine Structure of Coals and Cokes," B.C.U.R.A. London 1944 s1 J. P. Wibaut Rec. Trav. chim. 1919 38 159 ; Proc. K . Akad. Wetensch. Amster- 32 R. Ciusa Gazzetta 1922 52 130 ; ssF. Fischer and A. Pranschke Brennstoff-Chem. 1928 9 361. 176-231. dam 1921 24 92. 1925 55 385. 64 QUARTERLY REVIEWS the detailed study of these X-ray powder photographs. The diffracting units are of colloidal dimensions and therefore because of their low resolving power give diffuse diffractions there is a quantitative relation between this diffuseness and the size of the diffracting unit.The first explanation of the disappearance of all the hkl diffractions and most of the high-order diffrac- tions in bhe powder photographs of amorphous carbons was put forward by Berl Andress Reinhardt and Herbert 26 and by H. Arnfelt 34 it is briefly as follows. I n these finely crystalline or so-called amorphous carbons the hexagon layer-planes are stacked parallel t o each other in the individual crystallites with a spacing approximately equal to or somewhat greater than the corresponding spacing in the graphite crystal ; the individual hexagon layer-planes however are otherwise orientated in a completely random manner. J. Biscoe and B. E. Warren 36 have coined the word turbostratic to describe such a crystallographic structure.It can give rise to only 001 and hk0 (more correctly hk) diffractions the hk bands are cross- lattice diffractions formed by the individual hexagon layer-planes acting as two-dimensional diffraction gratings. Such diffractions are not symmetrical but show " tails " towards greater values of 0 . B. E. Warren 36 has discussed in detail this type of diffraction and derived a formula relating the broadening of the cross-lattice band with the linear dimension of the diffracting net-plane ; this is of the same form as the Schemer formula for three-dimensional crystallites but the numerical constant instead of being approximately unity is 1-84 1,843 Linear dimension of crystallite L = ___ p cos 8 where 3 is the wave-length of the X-radiation p the breadth of the difiaction a t half its peak intensity (half-peak width) and 0 the Bragg angle.Warren has also shown that with very small crystallites there is a small displacement of the diffraction maximum of the cross-lattice band towards a larger value of 8 0.163. d(sin0) - - L If a correction is not made for this displacement the calculated C-C spacing within the hexagon layer-planes will be too The assessment of the average.size and shape of the cryatallites present in amorphous carbons from the broadening of the diffraction bands in X-ray powder photographs has been attempted by several workers. Unfortunately differeiit niet hods of calculation have been employed pa&icularly in correct- ing the iticasured hdf-pcilk u-idth for the width of a diffraction from a similar specinien made up of large crystals the results reported by dieerent workers t herefow arc often not strictly comparable.Nevertheless the results reported for similar carbons are of the same order of magnitude. Because 3 4 A r k . Nat. Ast. Fys. 1932 23B No. 2. 37 Cf. for examplo H. E. Blayden H. L. Riley and A. Taylor J . Arner. Chem. S O C . 19.10 62 180 ; .J. H. de Boor Rec. Trav. chim. 1940 59 828 ; U. Hofmann XL'atrrrwis.y. 1944 260. J . Appl. Physics 1941 9 492. 313 Physical Rev. 1942 49 693. RILEY AMORPHOUS CARBON AND GRAPHITE 65 usually only two diffraction bands are of sufficient intensity for accurate measurement it has become customary to express the results of measure- ments of this kind in terms of a hypothetical average cylindrical crystallite the diameter La of which is termed the a dimension and the height L, the c dimension.Some typical results 38 are shown in Table 11. These and the results of other workers indicate that the original Debye-Schemer theory of the crystallographic nature of amorphous carbon requires modifica- tion in a t least two respects 'uiz. the small crystallites in amorphous carbon differ from macro-crystalline graphite in that (1) they are turbostratic and TABLE I1 Crystallite Dimensions (Hofmann and Wilm 38) Carbon. Ceylongraphite . . . . . Carbon monoxide carbon 700" . 9 9 , 550" . 9 9 1 , 420" . Y f Y , 400" . Acetylene soot (explosive decom- position) . . . . . . . Retort carbon . . . . . . Acetylene soot. . . . . . s t , calcined . . . , activated . . . Actiyemrbon AKTIV .. . 9 7 , , , calcined. 9 , , , activated Supranorit . . . . . . . , calcined . . . . , recalcined 30 hrs.. . ) activated . . . . Carboraffi calcined . . . . , aotivated . . . . Gas-mask carbon calcined . . 9 7 , act.iveted . . Sugar carbon . . . . . . 7 , activated . . . C. 99.1 94.9 95.0 91.8 - 99.7 96.9 95.0 99.1 98.5 88.5 96-5 94.4 89.3 97.3 97.7 96.2 93.0 97.4 96.5 98.1 98.0 L Analysis (jo. H. 0.0 0.1 0.1 0.1 - 0.3 0.3 0.8 0.3 0.0 1.7 0.35 0.0 0.6 0.15 0.0 0.6 0.0 0.6 0.0 0.7 0.0 - Ash. 0.3 4 3 2 - 0.0 0.6 0.3 0-3 0-5 1.1 1.1 2.0 0.3 0.4 0.9 1 . 9 3.3 0-5 1.0 0.0 0.0 - Layer- plane spacing C/? A. 3-35 3.4 3.4 3.4 3.45 3.43 3.45 3.55 3.55 3-6 3.5 3.45 3.65 3.5 3.45 3.55 3.7 3.G 3.7 3-55 3.6 3.6 3.6 Crystallite tliriiension A . L O . a 120 150 60 40 45 45 21 26 22 17 25 18 18 26 26 24 13 19 20 17 21 18 &.ca. 200 180 160 70 35 50 40 13 14 16 11 10 8 8 7 7 7 10 7 10 8 9 7 (2) the hexagon-layer spacing is not constant a t 3.35 A. but increases as the number of layer-planes in the crystallite decreases. Even when allowance is made for the uncertainty in the fundamental accuracy 39 of these results the extreme minuteness of these crystallites is striking in the smaller crystallites the individual hexagon layer-planes are not much larger than certain polynuclear aromatic molecules of known structure and the c dimensions indicate that the crystallites can be built up from a.s few a's three or four layer-planes. 38 U. Hofmann and D. Wilm 2. EEektrochem. 1936 42 504. 89 a. H. Cameron and A. L. Patterson Amer. SOC. Testing Mat. Symposium on Radiography and X-ray Diffraction 1937.E 66 QUARTERLY REVIEWS It would appear from the above that the process of charring or carbonisa- tion must involve the progressive transformation of the more or less complex organic matter into the polynuclear aromatic ring systems and their subse- quent stacking and growth. Some such process is indicated by the experi- ments of R. C. Smith and H. C. Howard 40 who charred pure cotton cellulose at temperatures ranging from 190" to 400" and subjected the products t o exhaustive oxidation with alkaline permanganate the magnitude of the yields of aromatic acids from the various chars indicated progressive aromatisation as the charring temperature increased. X-Ray investiga- tions have shown however that this simple picture is far from complete.Fig. 2(a) summarises the results of crystallite-size determinations on chars 40 J . Amer. Chem. SOC. 1937 59 234. RILEY AMORPHOUS CARBON AND URAPHITE 87 prepared from cellulose at various temperature^.^^ Results of an identical character have been reported by U. Hofmann and F. Sinkel 42 who worked with sucrose chars. The striking feature of these results is the constmcy of the average height ( c dimension) of the crystallites (experimentally the half-peak width of the 002 diffraction band) over a wide range of carbonising temperature. Crystallites built up merely from polynuclear aromatic layer-planes held together by relatively weak van der Waals forces would not be expected to behave in such a manner but rather to vary their degree of packing with the speed and temperature of carbonisation.This variation would be reflected in changes of the c dimension. Other carbohydrates lignin and wood i.e. organic compounds containing a relatively high propor- tion of oxygen all give chars containing crystallites with c dimensions between 9 and 10 A. Carbonising conditions can be varied between wide limits with respect to both speed and temperature (up to lZOO") without any consequential change in the half-peak width of the 002 difiaction of the resulting char. Cellulose carbonised in an atmosphere of ammonia and also glycine behave in a similar manner but the constant c dimension is somewhat greater viz. between 12 and 13 A. This constancy of the c dimension is highly suggestive and appears to indicate some kind of cross- linking between the layer planes of the crystallites.This hypothesis appears more probable when we consider the X-ray crystallographic behaviour on carbonisation of certain organic compounds which have a relatively small oxygen content. Fig. 2(b) shows the c-dimen- sion curve obtained with cokes made over a range of temperatures from that part of a bituminous coal-tar pitch which is soluble in carbon tetrachloride Similar c-dimension curves 43 are given by pitches certain bitumens bituminous coals the pyridine-chloroform soluble extracts of bituminous coals and certain pure polynuclear aromatic compounds e.g. dibenzanthrone Caledon jade-green etc. Such c-dimension curves appear to be character- istic of the presence in the parent carbonaceous matter of large thermally stable lamellar aromatic molecules. Aromatic hydrocarbons of this type are volatile and apparently the presence of some oxygen and/or nitrogen in the parent molecule is necessary for a large carbon yield the proportion however must not be too large otherwise charring occurs at a low tempera- ture.These c-dimension curve8 have been interpreted as follows. The ascending part of the curve between room temperature and about 50O0 indicates a unidimensional crystallisation brought about by the increased packing of thermally stable aromatic molecules under the influence of thermal vibration. The maximum at about 500" reflects the conversion of a system consisting of stable independent organic molecules held together by weak dispersion forces (molecular lattice) into a " carbon " crystallite [rigid three- &%ensionally cross-linked ( ?) lattice].Dibenzanthrone after being heated slowly up to 475" and cooled still gives its characteristic X-ray difiaction I1 H. E. Blayden H. L. Riley and A. Taylor J. 1939 67. Ia 2. a w g . Chem. 1940 245 85. u H. E. Blayden J. Gibson and H. L. Riley J . Inst. FueZ 1945 War Time Bulletin 117. 68 QUABTERLP BEVIIOWS pattern made up of several more or 106s sharp diffkactione; after heatixq to 500" the pattern characteristic of the dibenzanthrone crystal lattice haa completely disappeared and been replaced by two diffuse haloes character- istic of amorphous carbon. This change molecular lattice -+ carbon occurs over a relatively narrow range of temperature the height of which appears to be a function of the proportion of oxygen in the parent carbonaceous material. Substances like cellulose form rigid (cross-linked ?) carbon crystallites at temperatures little above 200" whereas certain coking cod bitumens retain their molecular identity up to temperatures as high as 550".The interpretation of the descending portion of the c-dimension curve between 550" and 900" [Fig. 2(b)] is not quite so clear. The phenomenon is however of general occurrence whenever a carbonaceous substance on being heated to some temperature below 550" gives a product which contains crystallites having an average c dimension greater than about 12 A. then a further increase of temperature brings about a decrease in the c dimension (experimentally a broadening of the 002 diffraction). The increasing dis- order over this temperature range may be due to the evolution of oxygen as oxides of carbon and a consequent reduction in the extent of tho cross- linking between the hexagon layer-planes.The subsequent increase in the c dimension above 900" is perhaps facilitated by the remnants of the ordered structure which the system possessed at 550". A further type of c-dimension curve is shown in Fig. 2(c) it is character- ised by an abrupt increase in the average c dimension in the temperature range 800-1000". Crystallite growth of this kind is shown by peats to a less extent by brown coals and lignites and also by " bituminised " cellulose and lignin Le. cellulose or lignin which has been heated in an autoclave to about 300" with an excess of aqueous alkali.43 In the case of peat the crystallite growth of a part of the char between 800" and 1OOO" is so great as to suggest some profound atomic rearrangement perhaps a kind of polymorphic change.Although the true nature of this crystallographic change is still not clear J. Gibson 11. Holohan and H. L. Riley44 have pointed out that besides the diamond and graphite lattices there is a third possible spatial arrangement of carbon atoms which may play an important part in the atomic architecture of amorphous carbon. This arrangement can be pictured as follows. Imagine a graphite layer-plane to be made up of discrete hexagons connected together by valency bonds as in Fig. 3(a) ; break the necessary bonds and rotate the hexagons A,B,C etc. inwards through 60" [Fig. 3(b)] ; it will be found that the valencies so freed can be attached without strain to those of similarly tilted hexagons in adjacent planes above and below This can be seen in Fig.3(c) which shows the spatial relationship of 4 hexagons. The structure 80 formed is a three- dimensional network of the " skeleton lattice '' type full of large holes and long channels and reminiscent of the structure of certain zeolites. The disappearance of the flat layer-planes suggests that the resonance energy of this structure would be less than that of the graphite lattice and its existence therefore less probable. The new structure however contains an orderly 44 J. 1946 458. 69 network of conjugated double linkages running throughout the whole three- dimensional lattice and its resonance energy would therefore be of an appre- ciable magnitude. The existence of such a structure is made more probable by the discovery by I. L. Karle and L.0. Brockway,45 by means of electron- diffraction measurements) of the non-co-planarity of o-tetraphenylene [see Fig. 3(c)]. The suggested structure is a three-dimensional repetition of o-tetraphenylene residues. RILEY AMORPHOUS CARBON AND GRAPHITE \ FIG. 3 A consideration of carbonising processes which would possibly tend to give some structure other than the flat honeycomb network of carbon atoms led Gibson Holohan and Riley to prepare carbon by (a) carbonising hexa- iodobenzene and (b) the interaction of hexachlorobenzene and sodium amalgam ; (a) gave a carbon which showed practically no coherent scattering of X-rays and ( b ) gave a carbon the X-ray powder photograph of which although highly diffuse showed new features which can be explained by the existence of small volumes of the three-dimensional network described above.The constancy of both the a and the c dimensions [Fig. 2 (d)] when a carbon prepared by method ( b ) (temperature of preparation did not exceed 300") was heated to higher temperatures provides additional evidence of a rigid cross-linked lattice structure These X-ray investigations leave little doubt that the idea that black a,morphous carbons are merely graphite in an extremely fine state of sub- d5 J . Amr. C h . SOC. 1944 66 1974, 70 QUARTERLY REVIEWS division is inadequate and needs further qualification. It is evident that the structure of a particular sample may be greatly influenced both by the nature of the parent carbonaceous material and by the method of carbonisa- tion. It is difficult to conceive for example how a system consisting of relatively free and independent lamellze could fail to develop some degree of order under the influence of thermal vibration and yet carbons can be made which give no coherent scattering of X-rays.Before outlining a modified view of the crystallographic nature of amor- phous carbons we must consider the process of graphitisation i.e. the transition amorphous carbon + graphite. Graphitisation.-The high thermal stability of pure amorphous carbon is shown by the results of Biscoe and Warren 35 (Table 111) who studied the TABLE I11 C'rystallite Dimensions of Carbon Black (Biscoe and Warren) Time and temp. of heating. - 2 hrs. at 760" 2 hrs. at 1040" 2 hrs. at 1500" 2 hrs. at 2000" 2 hrs. at 2800" 2 hrs. at 2000" 2 hrs. at 1040" Layer-plane spacing A. 1 Crystallite dimension A.Frorn002. 1 From001. 3.55 3.53 (3.65) 3.48 3.46 3.45 3,47 3.55 3 4 1 3-43 3.47 3-45 3-44 3.44 3.43 3.46 La. 20.0 22.6 28.0 44.2 55.8 65.2 60.5 29.8 Lc * 12.7 14.4 14.9 24.9 32-3 40.0 35.5 14.9 effect of high temperature on the crystallographic character of a commercial carbon black. Similar results were obtained by P. Corriez 46 who studied the effect of high temperature on the crystallite dimensions of sucrose chars. I n the case of pure carbons it is only a t extremely high temperatures that pronounced thermal recrystallisation occurs. U. Hofmann A. Ragoss and F. Sinkel 47 have shown by means of X-ray diffraction electron-microscope and adsorption studies that the small polycrystalline spherical particles of carbon black on heating for 24 hrs. a t 3000" undergo thermal recrystallisa- tion to an extent limited by the size of the original spherical particle i.e.the indiyidual polycrystalline particles probably become single crystals. This thermal stability of the amorphous carbon crystallites is due of course to the high resonance energy of the hexagon network planes of carbon atoms. It is only when these networks are broken down by some chemical mechanism that pronounced crystal growth can occur at temperatures below 2000". This was realised by E. G. Acheson long before the crystal structure of graphite was known in his original patent on the manufacture of graphite 4% Cornpt. rend. 1935 201 1189. 47 Kolloid-Z. 1941 96 231 ; s00 also D. Wilm and U. Hofmann ibid. 1935 70 21 ; A. Ragoss U. Hofmann and R. Holst ibid. 1943 105 118 ; M. von Ardenne and U.Hofmann 2. physikal. Chenz. 1941 B 50 1. RITAEY ttJIORPHOT7S CARBOX ,IN D GR~IP~IITE 71 he states 4 8 " I hnvc also discoverctl that in order to produce pure graphite from carbonaccous materials thcre is nri indirect conversion and that the act of formation of the graphite is morc in thc nature of an act of dissociation of the carbon from its combination with other materials than a conversion of the ordinary carbon into graphite and that as a preliminary step the carbon has to be combined chcniically with some other material. Thus I have found that if the carbonaccous material or carbon used in the process contains a considerable proportion of mincral matter or if it is mixed with a certain proportion of oxide or oxides such as silica clay alumina mangan- ese lime or oxide of iron and subjected to treatment as hereinafter set forth (high temperature) the yield of graphite is enormously increased and the product is most satisfactory." It is significant too that processes which involve the formation of carbon by deposition from the gas phase often give at relatively low temperatures products which are highly crystalline soe for example the crystallite dimensions of the carbons formed on a catalyst by the reaction 2CO = C + CO, quoted in Table 11.The growth mechanism of the carbon crystallites formed in such reactions apparently involves the initial formation of single carbon atoms which subsequently attach themselves to pre-formed graphite nuclei. Other reactions are known which give rise to highly crystalline graphite at temperatures far below that of thermal recrystallisation they all involve the liberation of individual carbon atoms e.g.the graphitisation of austenite,49 and the decomposition of carbides 5O and calcium cyanamide. 51 Carbons deposited from hydrocarbon gases in the iieighbourhood of lOOO" are not so highly crystalline as those deposited from carbon monoxide a t the same temperature the presence of hydrogen appears to inhibit crystal growth presumably by saturating the valencies of the carbon atoms at the surface of the crystallites. It is reasonable t o suppose however that carbons formed by any process which involves the participa- tion of single carbon atoms will have the most stable structure i e . the flat layer-plane graphite-like structure rather than any cross-linked structure of rather hlgher potential energy.JI. Holohan and R. Iley 52 have shown that in spite of' their highly dispersed state carbon blacks and carbon deposited from the gas phase on vitreous silica a t lOOO" have ignition temperatures which approach and may even exceed those of natural and artificial graphites. Tho X-ray-scattering power of carbons of this type is 48U.S.P. 568,323 29.9.1890 quoted by F. -4. J. Fitzqcrald J . SOC. Chem. Id. 1901 20 443 ; see also H. Ditz Chem. Z t y . 1904 28 167 and W. C. Arsem Met. Chem. Eng. 1911 9 536. 48 F. Wust and C. Geiger StuhZ zl. Eisen 1905 25 1134 1196 ; I<. Iokibe Sci. Rep. Tdhoku Inip. Unip. 1920 9 273 ; L . Sorthcott J . Iron Steel Inst. 1923 108 491. 5 0 H. Ishikawa Elect. Rev. Japan 1931 19 419 493 690 726 824 884 ; C. Hahn and A. Strutz Metallurgie 1906 3 72.3 ; Chem.Ztg. 1907 31 Rep. 33 ; J. West- becker Metullurgie 1904 1 137 ; 2. Elektrochenz. 1904 10 837 ; A. Frank Versamm. Ges. deutschen Naturforscher u. Aerzte Sept. 1905 ; Chem. Ztg. 1905 29 1044 ; V. M. Weaver U.S.P. 1,576,883 16.3.26. 61 E. Collett and M. Eckardt E.P. 5713 7.3.1911 ; A. Reme16 and B. Rassow Z . angew. Chem. 1920 33 139; N. Kameyama J. Chem. Ind. Japan 1921 24 1131. 5a Unpublished work of the Northern Coke Research Laboratory. 72 QUARTERLY REVIEWS much greater than that of chars of comparable crystallite size sulphuroue chars (see above) have a still lower scattering power. Although these Wer- ences in scattering power have as yet only been studied semi-quantitatively they are of such a magnitude that there is little doubt of their si@cance.Conclusion.-The above considerations lead to the conclusion that amorphous carbons are built up from two types of structure vix. (I) the turbostratic lamellar graphite-like structure and (11) the disordered three- dimensionally cross-linked structure (Fig. 3). Amorphous carbons which contain a preponderance of structure I are usually much purer than those made up largely of I1 it is possible that the foreign atoms hydrogen together with oxygen nitrogen or sulphur play some part in stabilising the type I1 disordered structure. Such a view of the crystallographic character of amorphous carbon appears to offer a reasonable basis for the explanation of the great diversity of its properties e.g. the highly specific nature and range of its adsorptive properties the great variation in its chemical reactivity the dif€erences in its mechanical properties and scattering power for X-rays etc.Carbons deposited from the gas phase which may contain little or no combined oxygen will be made up largely of type-I structures whilst cham and cokes in general will consist of intimate associations of both types. A large proportion of oxygen in the parent carbonaceous matter appears to favour the formation of structure 11 whereas carbonaceous matter containing little oxygen and a large proportion of benzenoid carbon e.g. pitches bitumens etc. will give carbons with a higher proportion of structure I. Both struc- tures possess high thermal stability structure I1 appears to persist in Borne specimens at temperatures well above 1000". No mention has been made of the importance of the secondary structure i.e.the cohesion and degree of agglomeration of the crystallites in deter-. mining the properties of amorphous carbon. Carbon blacks used in the rubber industry have been investigated in detail ; 53 little is definitely known however of the secondary structure of chars and cokes their accessible ~urface can be measured but the open lattice present in both structures I and I1 makes the interpretation of such measurements ditlicult. Such is the diversity of the crystal structure of solids that no single comprehensive theory of the amorphous state is possible. Metals and salts can be so disordered by mechanical strain or dispersion that they give diffuse X-ray diffractions. Crystals which are made up of long homopolar chains exist in an amorphous condition when the chains become tangled together e.g.vitreous sulphur and selenium rubber borate-glass etc. Systems in which the molecules are large lamellae can assume the amorphous (or meso- morphous) state because of the ditliculty these cumbersome units have in packing into an ordered lattice e.g. pitches bitumens polynuclear aromatic compounds. Finally there is the disordered three-dimensional network present in certain glasses and vitreous solids. Amorphous carbons should probably be placed in a class intermediate between the disordered lamella and the disordered cross-linked lattice. 68 Columbian Carbon Company " The Partiole Size and Shape of Colloidal Csrboa w Revealed by the Electron Microscope," New York 1940.