J . Chem. SOC., Faraday Trans. 1, 1982, 78, 2369-238s Structure and Dynamics of Graphite Intercalation Compounds Part 1.-Neutron Diffraction and the Structure of C&, C,KH$ and C,KD+j BY TIMOTHY TREWERN, ROBERT K. THOMAS, GEOFFREY NAYLOR AND JOHN W. WHITE* Physical Chemistry Laboratory, Oxford University, South Parks Road, Oxford OX1 3QZ Received 7th August, 198 1 The neutron diffraction patterns from C,K, C,KDt and C,KHg have been measured at 300 K using samples prepared from a large variety of different graphites. The effect of sample texture on the degree of order that can be produced in the intercalation compounds is reported. A difference synthesis is made for the diffraction patterns of C,KDI and C,KHI and locates the hydrogen species in the median plane between the graphite sheets and potassium layers, confirming recent models.A model structure is fitted to the powder diffraction data and leads to a three-dimensional arrangement for a hydride ion species in the intercalated region of C,KHt materials. The intercalation compounds of graphite have properties, intermediate between metals and insulators, which may be varied systematically by choice of the intercalating and stoichiometry. There is currently much interest in their band structure4 and associated physical proper tie^,^?^ and their widely variable electron affinity gives them some properties of the transition metals, which may be responsible for their observed catalytic a ~ t i v i t y . ~ ~ ' 9 This series of studies concerns the ternary compounds of graphite, alkali metals and hydrogen and aims to resolve some outstanding questions on structure, kinetics of formation and excitations by studies with neutron diffraction and neutron-scattering spectroscopy.A study is presented of the texture of powder and single-crystal samples prepared by different methods. In this first paper neutron diffraction from C&, C,KHj and C,KDj is reported. Entry of potassium between layers of graphite increases their spacing from 3.35 to 5.4 A, a process which involves a translation of the carbon network so that planes on either side of an intercalate layer are superimposable. If we denote reactant layers by I, then the process of the first-stage formation can be denoted by ABABAB+AIAIAIA.. .. Structural studiesg have shown that in the first-stage compound every fifth intercalate layer is superimposable, so that it can now be represented as: A a ADA y A G A a ...giving a c-axis lattice parameter of 4 x 5.4 A = 21.6 A. The packing of the metal species within the layers for the stage 1 compound is a centred hexagonal arrangement. X-ray studies of the second- and subsequent-stage compounds have shown that the intercalate layers are disordered in a liquid-like fashion but pass through a transition into an ordered state on l1 2369 77 FAR 12370 GRAPHITE INTERCALATION COMPOUNDS TERNARY COMPOUNDS Two types of behaviour are seen in the reaction between the graphite intercalates C,M (M = K, Rb, Cs) and hydrogen or deuterium. In the first type (resembling physisorption) it has been reported by Watanabe et aZ.12 that the second-stage intercalates C,,M (M = K, Rb, Cs) take up hydrogen, deuterium and other gases such as nitrogen and methane at temperatures below 196 K.Although C,,K absorbs hydrogen or deuterium up to a limiting composition of C24K(Hz)2.1, C,K was found to be non-sorptive in this temperature range. The enthalpies of sorption (for half coverage) for this type of behaviour are ca. 30 kJ mol-l. A quite different behaviour at high temperature has been noticed by Herold et al.13914 who reported the action of hydrogen and deuterium on the first- and second-stage intercalates (C,M, C24M) of both natural and artificial graphite. The limiting composition for C8K is the dark blue C,KH+, which is obtained at a pressure < 1 atm. Replacement of potassium by caesium decreases the degree of hydrogenation at a given pressure.For C24K the product at a hydrogen pressure of 1 atm* had an approximate composition of C,,KH,,,,. For C,KH+ Colin and Herold14 have proposed an orthorhomic unit cell with parameters a = 4.92 A, b = 8.52 A, c = 47.52 A. According to them, intercalation of hydrogen into C,K increases the carbon interlayer spacing from ca. 5.4 to 5.94 A. Lagrange et aZ.15 have suggested that this process involves a major structural transformation whereby the first-stage intercalate passes to a second-stage material with a potassium bilayer. As evidence of this they found the ternary intercalate C,KH+ to be susceptible to attack by further alkali metal, forming a new stage 1 compound in the following manner: 2C,K&+M + C,KH+.C,M (M = K, Rb, CS).Transition of the stage 1 C,K to the stage 2 C8KHt completely empties the space between pairs of carbon planes. An objective of the work reported here was to verify this conclusion by measuring neutron diffraction along the crystalline c axis of C,KHt and C,KD#, if possible at the same time learning something of the lateral organisation of the hydrogen species by measurements on well dispersed, powdered samples. EXPERIMENTAL PREPARATION AND CHARACTERISATION OF SAMPLES The preparation and characterisation by neutron diffraction of all samples used for diffraction, kinetic and neutron inelastic scattering studies to be reported now and in subsequent papers are summarised. For the present work highly ordered single-crystal texture samples of C,K, C,KDg and C,KH, were required as well as samples of powdered C,K and its hydrogen intercalates.The two-bulb techniquegi l6, l7 was used for all preparations with a vacuum above the sample of always better than 10-5mmHgt at the start of experiments. Our modification of the apparatus for single-crystal samples is shown in fig. 1. Potassium, free from organic impurities, was obtained by vacuum distillation of carefully cleaned metal. Five different types of graphite were used (a) pyrolytic graphite (Union Carbide), (b) turbostatic graphitised carbon Graphon (86 m2g-l), (c) highly ordered pyrolytic graphite (Union Carbide, h.o.p.g.), ( d ) monochromator- grade pyrolytic graphite (Union Carbide), (e) exfoliated graphite (H,SO,) from A. Wedgewood, A.E.R.E.Harwell. * 1 atm = 101 325 Pa. t 1 mmHg = 13.5951 x 980.665 Pa.T. TREWERN, R. K. THOMAS, G. NAYLOR AND J. W. WHITE 237 1 A silica support to vacuum line PTFE tap\ graded :seal \( potassium *\ silica FIG. 1 .-Apparatus for preparing oriented specimens of intercalation compounds. The powder samples were prepared from powdered pyrolytic graphite of particle size ca. 0.2 mm. Following outgassing of the graphite at 623-673 K and 5 x lop6 Torr,* ca. 20% excess potassium was distilled in and the sealed system was maintained at 573 K for 18 h. The temperature was lowered to 473 K and the excess potassium was allowed to distill off by opening the tap. An analogous procedure was followed for otRer alkali intercalates. Incorporation of a PTFE tap allowed subsequent transfer of the product under vacuum.With potassium the product was always a copper-coloured, free-flowing powder. Thomy and Duvall* have reported that intercalation into graphitised carbon black is not easy, since the exposed crystal faces are almost exclusively basal planes and not the edges into which the guest species diffuses. With this in mind the heating time for samples using Graphon was extended to 36 h to ensure reaction. The products appeared a very dark brown colour and remained free-flowing powders. The preparation of single-crystal samples presented a difficulty with regard to their support. Since the crystal undergoes a net expansion of some 60% during intercalation of potassium it was necessary that the crystal support should also be able to expand.The solution adopted was to support the crystal between silica hooks as shown in fig. 1. The tension in the hooks was previously adjusted so as to grip the graphite firmly and yet allow expansion during reaction without fracture. The intercalation was carried out in a similar manner to that used for the polycrystalline materials. The temperature gradient within the furnace was such that there was no condensation of alkali metal on the surface of the graphite. A summary of the samples prepared and neutron experiments done is shown in table 1. * 1 Torr = 101 325/760 Pa. 77-22372 GRAPHITE INTERCALATION COMPOUNDS TABLE 1 .-SAMPLE SUMMARY sample no. graphite substrate neutron experimentsa 1 2 3 4 9 13 14 15 powdered pyrolytic graphite (p.p.g.) (Union Carbide) P*P*g.P.P*g* Graphon (surface area P*P*g- P.P*g- P*P.g- highly ordered pyrolytic graphite P.P*g- monochromator grade 86 m2 g-l) (h.0.p.g.) pyrolytic graphite h.0.p.g. 16 exfoliated graphite Curran 13. = 1.37 8, as C,K/H2/D2 beryllium filter as C,K/H, beryllium filter as C,K/D2 Curran 13. = 1.37 8, as C,K 4H5 13. = 4.8 8, as C,K 6H 13. = 5.8 A as C,K/H2 INIB as C,K/H2 HRPD 13. = 1.522 8, as C, C,K DIB as C,K/H,/D, Curran 13. = 2.63 8, Mk VI 13. = 1.09 (6) 8, w scan as C 0 scan as C,K HRPD 13. = 1.17 8, as C,K/H2 Mk VI 13. = 1.09 A w scan as C w 20 scan as C,K w and 020 scans as C,K o 20 scan as C,K/H, Mk VI A = 1.09 A w 20 scan as C,K/H2 Mk VI 13. = 1.09 8, w scan as C HRPD 13. = 1.29 8, w 20 scan as C,K w and 20 scans as C,K/D2 w 20 scan as C,K/H2 Curran 13.= 1.37 Mk VI 13. = 1.09 A a Curran, beryllium filter, 4H5, 6H, high-resolution powder diffractometer (HRPD), mark VI (Mk VI) are or were instruments at A.E.R.E. Harwell on the DIDO and PLUTO reactors. DIB and INIB are powder diffractometer and beryllium filter instruments at Institut Laue- Langevin, Grenoble. METHODS OF CHARACTERISING THE SAMPLES The quantities available from analysis of neutron diffraction patterns are the lattice parameters before and after intercalation, the mosaic spread of the crystals before and after intercalation and, in the case of powder samples, the dimensions of the two-dimensional crystals characterised by the parameters of Warren.lB To determine the mosiac spread, the mark VI single-crystal diffractometer of A.E.R.E.T. TREWERN, R.K. THOMAS, G. NAYLOR AND J. W. WHITE 2373 Harwell was used. A reflexion from the sample was chosen as near as possible in 28 to the take-off angle of the monochromator for the wavelength used, and rocking of the sample about the o axis was performed; the full width at half maximum, f.w.h.m., B, of the rocking curve is related to the mosaic spread B by where 2 is an instrumental parameter determined by measuring several reflexions at 28 values away from the take-off angle. The effect of 2 was found to be never more than 0.5O. The data are summarised in table 2 . (In all cases intercalation increases the f.w.h.m. of the rocking curve by ca. 5O.) Note, however, that in every case the crystal did not remain single but was extensively cleaved, so that relative misorientations account for a certain amount of the final width.The above data are relevant to current interest in graphite intercalates as monochromator crystals for long-wavelength neutrons. The reflectivities are quite good, in some cases the second orders of diffraction are very weak, and the possibility of sandwiching crystals with slightly different lattice parameters opens up the possibilities of band-pass monochromators. /P = B2-Z TABLE 2.-EFTECT OF POTASSIUM INTERCALATION ON THE ROCKING CURVES OF SOME GRAPHITE SPECIMENS rocking curve f.w.h.m. ~ ~~~ specimen diffractometer before intercalation after intercalation UCAR highly oriented HRPD A = 1 . 5 2 A pyrolytic graphite 10' sollers /? = 0.8 & 0.2 c = 6.716 A Curran A = 1.37 8, pyrolytic graphite Curran A = 2.63 A Mk VI A = 1.09 A (monochromator grade) pyrolytic graphite Mk VI A = 1.09 A exfoliated graphite exfoliated graphite Mk VI A = 1.09 A Mk VI A, = 1.09 A (1) (11) 0.88 f 0.02" 28 = 85.80 7.0 f 0.5O 28 = 50.6 (sample 8) 5.0 & 0.3 28 = 14.7 (sample 15) 0.7 f 0.1' 28 = 630 6.3 f 0.2O 28 = 47.60 (sample 13) 50 5.0' 1 1 .O f 0.5' 1 1.4 f 0.5" 28 = 18.60 28 = 18.60 2e= 18.60 15.5+ 1.0" 28 = 5.30 (sample 16) " After exposure to deuterium.PRELIMINARY RESULTS ON ORIENTED CRYSTALS The first oriented intercalate prepared, C,K (sample 8, table l), used a relatively large piece of ordinary pyrolytic graphite (dimensions $" x x g). Although the intercalation proceeded smoothly only three (001) diffraction peaks were seen on the high-resolution powder diffrac- tometer [A = 1.52 (1) A] at d = 5.39, 2.61 and 1.78 A.This gave a carbon<arbon spacing of 5.32 A. The loss of high orders was attributed to uneven intercalation in this large sample, which2314 GRAPHITE INTERCALATION COMPOUNDS was not further used. It is possible that the ordinary quality of the pyrolytic graphite contributed to the effect also. By starting with monochromator-grade pyrolytic graphite (sample 13) a better sample of C,K was obtained showing seven diffraction orders of (001) and a carbon layer spacing of 5.35 A using the Mk VI diffractometer [A = 1.09 (6) A]. This measurement showed up an important source of error in measurements on large mosiac crystals owing to the effects of counter aperture. The initial set of integrated intensities from sample 13 when corrected for the Lorentz factor were compared with those calculated from the accepted structure of C,K: I 1 2 3 4 5 6 7 It heory 100 67 35 36 23 28 20 *expt 100 56 22 17 8 7 4 We believe that the large discrepancy is due mainly to two effects: first no thermal factor has been included but a more important effect in this case is due to beam divergence.Implicit in the use of the Lorentz factor, 1 /sin 28, is the assumption that the detector is capable of accepting the entire diffracted beam, a questionable assumption for large mosiac spread crystals. Following Saxena and Schoenborn20 a suitable correction factor, L’, was applied which, with a physically meaningful thermal factor B = 0.9 (cf. B, for graphite = 1.321), gave good agreement between experiment and theory as shown: I 1 2 3 4 5 6 7 Itheory 100 55 22 17 8 7 4 Iexpt 100 56 22 17 8 7 4 Careful checks were made on all data to ensure that this phenomenon was adequately controlled. HYDROGENATED A N D DEUTERATED SAMPLES Admission of hydrogen (deuterium) at atmospheric pressure to the C,K powder samples at 373 K led to a colour change brown -+ blue in a matter of minutes.Samples were allowed to equilibrate overnight under 1 atm of hydrogen. Formation of the hydrogenated sample from large pieces of potassium intercalated pyrolytic graphite was much more difficult. Exposure of the C,K crystal (sample 13) at ca. 373 K to hydrogen led only to surface discolouration; diffusion of hydrogen into the bulk of the crystal occurred exceedingly slowly so that the preparation of the ternary intercalate in this case was not feasible.In order to overcame this problem the size of the next specimen (sample 15) was reduced. The dimensions of the highly ordered pyrolytic graphite substrate were al/ x a” x &” thick. As already described, potassium intercalation proceeded smoothly. Examination of the samples on the Curran diffractometer (A = 1.37) showed that intercalation had increased the mosaic spread from < lo to ca. 5O and the C-C separation from 3.35 to 5.35 A [derived from five orders of (001) peaks]. In spite of the ease of potassium intercalation, conversion to the hydrogenated intercalate was again inconveniently slow and was, in fact, never complete, as shown by the persistence of peaks due to C,K in the diffraction pattern.The final attempt to prepare an oriented specimen of C,KH, was made using exfoliated graphite (sample 16). Diffraction patterns were recorded on the HRPD [A = 1.29 (6) A], both after intercalation of potassium and following exposure at ca. 150 O C to deuterium for 18 h. In this sample C,K diffraction peaks were not visible, and further patterns were recorded on the Mk VI diffractometer [A = 1.09 (6) A], both of C,KDz and of the hydrogenated specimen. The statistics of the weaker peaks were improved by dividhg the angular range of 28 into three sections and successively increasing the monitor setting. The patterns from this crystal are shown in fig. 2.3 Q 2 2 2 C 0 1 0 T. TREWERN, R. K. THOMAS, G. NAYLOR A N D J. W. WHITE 2375 20 16 m 2 12 8 \ Y C a 4 Mon.= 1.25~1 L I I I I I I I I I I I I Mon. =2.5x106 I Mon.=5.0x106 I I I I I I I I I I 40 60 80 100 I I I I I Mon. = 6. 5x105 I I Mon. = 1 3 ~ 10' I Mon.=26x105 I I I ! I I I 0 20 40 60 80 100 2 e p FIG. 2.-Diffraction patterns of oriented C,KDI (A) and C,KHI (B) along the (001) direction measured with 1.09 A neutrons at 300 K. RESULTS APPEARANCE OF THE DIFFRACTION PATTERNS Fig. 2 (A) and (B) show CI), 28 scans along the c axis of single-crystal texture samples (sample 16) of C,KDi and C,KHi, respectively. Two points for consideration arise from the general appearance of the diffraction patterns: (a) There was a small proportion of a contaminant present in the specimen, as shown by a second series of weak diffraction maxima. The d-spacing of the first peak in this contaminant series corresponds to that of the second-stage material, C,,K.(b) The intensities of the2376 GRAPHITE INTERCALATION COMPOUNDS C,KHQ peaks fell rapidly with increasing scattering angle. It was again necessary to take into account the effect of beam divergence. The problem of contamination by C2,K was approached first. The observed integrated intensities are shown in table 3 to ether with the peak positions. The c-axis repeat spacing was determined as 11.88 x . For comparison the positions and TABLE 3.-TOTAL OBSERVED INTENSITIES FOR THE DIFFRACTION PATTERNS OF THE TERNARY INTERCALATES TOGETHER WITH DATA FOR THE SECOND-STAGE CONTAMINANT (Ao = 1.09 A) total integrated intensities order position C24K 1 2e CSKH, CSKD, 1 20 F (relative) 10 11 12 13 14 15 16 17 5.3 10.6 15.9 21.3 26.7 32.1 37.7 43.3 49.1 54.1 61.0 67.2 73.7 80.5 87.6 95.1 103.3 - 58.44 2.63 16.74 18.18 0.36 1.46 5.60 2.80 0.02 1.34 1.89 0.39 0.16 1.65 1.05 0.01 0.61 53.90 0.92 43.50 30.49 1.45 3.60 9.88 6.28 0.20 1.96 4.82 0.71 0.18 2.63 1.76 0.82 (0.00) 1 2 3 - - 5 6 7 8 9 10 11 12 - 7.2 14.4 21.7 - 36.5 44.1 52.0 60.1 68.6 77.6 87.1 97.5 - - - - 1 2a 56 92 - - - 97 42 34 100 4 85 64 8 - - - - a F2 values are those calculated; no account has been taken of the factors which will cause the intensities of these peaks to decrease rapidly with scattering angle, 28.calculated structure factors for C2,K are also shown. The only contaminant peak which was completely free from overlap was the fourth order which has a low structure factor and was experimentally unobservable.The method of separating the intensities eventually adopted involved the use of a Du Pont curve analyser. This allowed the observed profile to be constructed as the sum of two peaks of defined shape (based on the fifth order for C,KDj). This allowed the impurity peaks to be scaled relative to the C,KHj (Di) peaks and subtracted. The impurity was of the order of 8% by weight. Correction for the Lorentz factor (sin 28) and normalisation followed. The values of FF’ and FF’ are shown in table 4 and were used in several Fourier syntheses. All syntheses were simplified as shown by the following reduction 1 L l p(z) = - I= r;,, cos 2nlz n 1 - 1 = F,,,+2 C Fool cos 2nlzT. TREWERN, R. K. THOMAS, G . NAYLOR AND J. W. WHITE 2377 TABLE 4.-NORMALIZED STRUCTURE FACTORS CORRECTED FOR CONTAMINATION AND LORENTZ FACTOR 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 12.53 3.77 11.25 13.52 2.18 4.77 9.50 7.33 0.67 5.67 6.54 3.25 2.12 6.76 5.44 0.54 4.17 9.25 1.75 13.85 13.01 3.40 5.83 9.13 8.31 1.64 5.34 7.80 3.39 1.75 6.39 5.35 0.06 3.77 t ! 2.4SA A ! 3.3sA A FIG.3.-Model for the layer stacking in C,KHi. where the first n orders of (001) peaks are observed. In all cases the zero-order term was neglected. A Patterson difference synthesis was calculated first using coefficients I FF’ -FF’ 12. The intensity at z = 0 and 1 1.88 A suggested that the hydrogen/deuterium was located at a centre of symmetry. A more useful form of the Patterson difference in this case results from the use of the modified coefficients I FF’ l2 - I FF’ 12.Evaluation for the model (fig. 3) shows that the dominant peaks are those associated with interatomic vectors from hydrogen or deuterium [fig. 4(b)]. When the observed coefficients are used the pattern [fig. 4(a)] is seen to be dominated by the C-H (D) vector at ca. 4.3 A. This is fairly conclusive evidence for the central position of the hydrogen. Comparisons of Fourier maps from the observed intensities (a) with those predicted from the model (b) are shown in fig. 5 . On changing hydrogen for deuterium the scattering density markedly increases in the centre of the unit cell in both the calculated and the observed maps.2378 GRAPHITE INTERCALATION COMPOUNDS I I I I I ~ 0 4 8 1 dlA FIG. 4.4ne-dimensional Patterson difference functions from C,KH, and C,KD$ patterns (a) using coefficients 1 @’ - FF’ l2 and (b) using I e’ l2 FF’ 12.QUANTITATIVE TREATMENT In order to place the results on a more quantitative basis account was taken of the neutron beam divergence. Using parameters appropriate to the experimental conditions on the Mk VI instrument and following the treatment of Saxena and Schoenborn,28 the correction factor, L’, was calculated and applied to the observed intensities. The corrected structure factors FF and F , were then compared with those calculated from the model in a least-squares refinement. The temperature factors, B, were arbitrarily fixed at 1.0 for both carbon and potassium, 4.0 for hydrogen (by comparison with the values derived from the study of the motion of hydrogen in metals30) and 2.8 (4/ 4 2 ) for deuterium.Agreement between the calculated and observed intensities could be improved by slight adjustment of the contamination correction. A useful, but by no means infallible, measure of the fit between the observed and calculated structure factors is the residual index R. This is defined by The values of R calculated for the two sets of data are R, = 0.22 and R, = 0.15, and the agreement between the actual values is shown in table 5. The quality of the original data is probably insufficient to justify further reductions and interpretation. There is, however, some evidence that the carbon-carbon plane separation is lower than the usual 3.35 A. The lowest R factor (9%) was obtained with an interplane spacing of 3.31 A for C,KDi.This was approximately duplicated by the C,KHi data, although whether it is a genuine effect is not certain.T. TREWERN, R. K. THOMAS, G. NAYLOR A N D J. W. WHITE 2379 4 s I I I I 1 0 4 8 12 dlA s s 0 4 8 12 dlA FIG. 5.-One-dimensional Fourier syntheses for (A) C,KH3 and (B) C,KD3. TABLE 5 .-COMPARISON OF CALCULATED AND OBSERVED STRUCTURE FACTORS 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 56 11 106 100 31 63 98 81 24 68 97 26 41 78 59 3 57 55 11 99 103 29 54 92 88 18 63 96 43 23 89 77 1 57 70 25 91 112 18 51 108 70 15 76 89 33 46 72 63 7 53 75 25 81 108 19 45 96 78 7 68 82 42 29 94 79 8 642380 GRAPHITE INTERCALATION COMPOUNDS 6000- v1 c1 c g 4000- SUMMARY Preparation of an oriented specimen of C,KHj and C,KDj has allowed the layer-stacking sequence to be derived.The hydrogen (deuterium) species is situated mid-way between the potassium planes. The X-ray work of Herold on the carbon and potassium plane positions has been confirmed. Structure factor calculations agree reasonably well with observed values, especially in view of the rather severe corrections imposed by a contaminant and the mosiac spread of the sample. ( b ) . POWDER DIFFRACTION C8K Fig. 6(a) shows the neutron diffraction pattern from C,K powder (sample 1) at 295 K taken with 1.37 neutrons on the Curran diffractometer at A.E.R.E. Harwell. The sample was prepared from powdered pyrolytic graphite. For comparison the 6000 m U E s E 8 4000 0 2 000 0 20 40 60 80 2810 2elo FIG. 6.--(a) Powder diffraction pattern from C,K (a) prepared from graphite and (6) prepared from turbostatic graphite (Graphon).T. TREWERN, R.K. THOMAS, G. NAYLOR A N D J. W . WHITE 238 1 pattern from C,K originating in Graphon (sample 4) is shown in fig. 6(6). The chief differences lie in the intensities and sharpness of peaks at 28 = 30, 39 and 6 9 O , the peaks in the Graphon sample being invariably weaker and wider with strong evidence of compound structure. The most general feature in common for the two samples is the differentiability of peaks into two categories, the (001) series of narrow, more or less gaussian peaks, and the sawtoothed, asymmetric peaks with a sharp onset at low 28, characteristic of the two-dimensional layer latticelg of limited size. To analyse the powder patterns, intensities were corrected for multiplicity, Lorentz factor and a consistent thermal factor chosen. In the first instance the theoretical diffraction pattern of perfect, polycrystalline C,K was considered.An examination of the symmetry of the material led to the conclusion that the space group was not C222 (0;) as proposed by Wolten22 but was of higher symmetry: Fdd2(C3, in agreement with Nixon and Parry.23 Results are given in TABLE 6.-PREDICTED DIFFRACTION DATA FOR C,K, 2 = 1.37 A index Bragg angle, structure factor multiplicity Lorentz (FpL) h k l 20/O F P factor L (relative) - 0 0 4 0 0 8 0 4 0 , 2 2 0 0 4 4 , 2 2 4 0012 - 0 4 8 , 2 2 8 0412, 2 2 12 2 6 0 , 4 0 0 2 6 4 , 4 0 4 0416, 2216 2 6 8 , 4 0 8 0020 - 0 8 0 , 4 4 0 0 8 4 , 4 4 4 2612, 4012 0 8 8 , 4 4 8 - 0016 - 14.7 29.7 37.2 40.2 45.2 48.3 59.9 61.6 67.1 69.1 74.2 75.0 79.6 79.3 81.2 84.6 86.8 39.60 45.52 18-01 23.93 39.60 18.01 23.93 45.52 45.52 39.60 18.01 54.52 39.60 18.93 24.85 39.60 18.93 - 18.47 24.39 18.47 24.93 45.52 39.60 18.47 54.52 18.01 23.93 39.60 18.01 - - - 2 2 2+4 4+8 2 4+8 4+8 2 4+2 8+4 4+8 8+4 2 2+4 4+8 8+4 4+8 62.4 16.1 10.4 9.0 7.48 6.59 4.68 4.51 3.93 3.77 3.48 3.41 3.22 3.19 3.1 1 3.00 2.92 100 34 11 32 12 14 17 10 25 36 7 43 5 3 11 29 6 table 6.Recent studies by Lagrange et al.24 suggest that the space group is Fddd, but the observed diffraction peaks for the pyrolytic graphite samples were consistent with the Fdd2 space group with unit-cell parameters a = 4.96 A, b = 8.59 A, c = 21.4 A, a = a = y = 90°, and an attempt to fit Fddd was less successful than the fit to Fdd2.The c parameter derived from the pattern was 21.4 (2) A. Comparison of the experimental and theoretical intensities confirmed that the material was indeed turbostatic, no general (hkl ) reflexions being present and the observed reflexions falling into two separate sets. The experimental intensities are in reasonable agreement with the predicted values. From the powder pattern it is possible to obtain values for the average extent of the crystallites both within the basal planes and perpendicular to them by deconvoluting instrumental broadening. This is relatively easy if the final profile is analytically simple and especially if it can be assumed Gaussian. This assumption can be made for the2382 GRAPHITE INTERCALATION COMPOUNDS (001) reflexions but the form of the (hkO) reflexions is not simple and deconvolution has not been attempted. Analyses of peak widths show that for C8K and using the expression of Warren19 L, [from the (008) peak] is ca.160 A (or 110 A if instrumental effects are neglected). The riean value of La, derived from the two-dimensional peaks (100) and (1 10) and neglecting instrumental broadening, is 90 A. It is probably safe to say that minimum crystalline dimensions are ca. 100 A. L, = 1.84 A/B cos 8, L, = 0.89 L / B cos e. C8K/D$/H$ The C,KD$ pattern [fig. 7(a)] has a similar appearance to that of C8K in that, once again, both symmetric and asymmetric peaks are present. The C,KHg-pattern [fig. 01 4- E g 2000 -0 e, 0 c E 8 1000 0 28 I" 0 20 40 60 80 261" FIG. 7.-Powder diffraction patterns of (a) C,KD and (b) C,KHt taken with 1.37 8, neutrons at 300 k.T.TREWERN, R. K. THOMAS, G. NAYLOR A N D J . W. WHITE 2383 7(b)] is also similar, the most noticeable difference being the higher background due to the high incoherent scattering from the hydrogen. The sharp symmetric peaks are the (001) series of the expanded lattice, indexing on the d spacing of the first reflexion (1 1.9 A). The asymmetric peaks appear in roughly the same positions for all samples, implying that the carbon skeleton remains, to a first approximation, unchanged in the ab plane in both C,K and the hydrogen intercalates. The results of indexing are summarised in table 7. TABLE 7.-sUMMARY OF DIFFRACTION RESULTS FOR THE POWDER SPECIMENS 8.6 & 0.2 5.35 k0.15 2.84 & 0.06 2.69 f 0.02 2.12 & 0.03 1.78k0.01 1.35 0.01 1.23 f 0.01 1.07+0.01 - vw - vs - - W m S W - - vw VS W - - 3.98 & 0.03 2.95 f 0.02 - - C24K 0 0 8 0 4 0, 2 2 0 0 0 12 2.11 k0.02 1.71 kO.01 1.48 f 0.01 1.23f0.01 1.06 f 0.01 - - 0 0 16 2 6 0, 4 0 0 0 8 0 , 4 4 0 - vs I 1.55 f 0.45 5.80 k 0.20 s 3.93f0.08 s 2.92f0.03 - - vs vw m S - s 2.10f0.01 m 1.69f0.02 w 1.47 k 0.02 s 1.22f0.01 w 1.06f0.01 - - - - S W vw S W It is impossible to say whether crystallite dimensions are significantly changed upon intercalation of the hydrogen, although measurement of peak breadths yielded La values which were generally slightly lower than those for the potassium intercalate itself.INTERPRETATION OF THE POWDER DIFFRACTION PATTERNS Using the one-dimensional hydrogen structure and known stoichiometry, three- dimensional structures can be guessed for the ordered state of C,KD; or C,KHj. We suggest as one plausible ordered structure that a hydride ion sits in the approximately tetrahedral sites of the sandwich formed by juxtaposing two, 2 x 2 potassium/graphite, C8K structures.As drawn this has the stoichiometry of C,KHi, i.e. that of the compound most readily formed by the reaction of C,K and hydrogen.l3~l4 This structure is sketched and shown in plan in fig. 8, and should be considered as a basis from which a three-dimensional structure could be built. The higher hydrogen density in C,KH$ can be modelled by filling more of the unoccupied sites with H-, or on chemical arguments it may be reasonable to suppose filling of the sites with H; ions. We regard these as extreme models of the structure which cannot be distinguished on present evidence.If one takes as ionic radii 1.33 A (K+) and 1.54 A (H- and H;) the model of fig. 8 is plausible, since the predicted distance between carbon sheets on each side of the2384 GRAPHITE INTERCALATION COMPOUNDS FIG. 8.-Possible three-dimensional structure for C,KH$ based upon the oriented and powder diffraction patterns. TABLE 8 .-EFFECTIVE IONIC RADII FOR POTASSIUM IONS IN RELATED GRAPHITE INTERCALATION COMPOUNDS c axis compound stage spacing/A radius of K+/A ref. KCl - - 1.33 - C,KH, I1 8 x 5.94 1.32 (14) CZ4K(DZ)ZU I1 8.96 1.13 (12) C*K I 4 x 5.4 1.025 (9) G*K I1 8.67 0.99 (12) a Measured at 77 K. Molecular diameter of D, is 2.4 A (D; = H; assumed = 3.08 A). C-C plane/plane distance is always taken as 3.35 A.potassium bilayer is 8.95 A (found 8.61 A). By looking at related compounds, tat 8, there are grounds for taking a smaller K+ ionic radius, which would bring the: figures into close accord. These numbers suggest that the potassium ions are 'packed into' the graphite sheet or alternatively that the van der Waals distance between bilayer graphite sheets is < 3.35 A. This last possibility echoes the result from the least-squares refinement of the one-dimensional diffraction pattern of C,KDi (above). The above model has been tested against the powder diffraction patterns of C,KDi and C,KHi to attempt a characterisation of any proton disorder to be expected by analogy with C,K1'* l1 but without a conclusive result.T.TREWERN, R. K. THOMAS, G. NAYLOR AND J. W. WHITE 2385 GENERAL CONCLUSIONS The preparation of CEK, CEKHt and CEKDf samples with a wide range of textures is reported. The most crystalline samples give eighteen orders of the (001) series in neutron diffraction, allowing the hydrogen pention to be definitely placed within a bilayer of potassium ions in a second-stage intercalation structure. One possible model for a basic unit of this structure is proposed. The powder diffraction measurements support the model used but were inadequate to allow a three-dimensional structure to be proposed. Yu. N. Novikov and M. E. Vol’pin, Russ. Chem. Rev., 1971, 40, 733. A. Herold, M. Colin, N. Daumas, R. Diebold and D. Saehr, Chem. Soc. Spec. Pub., 1966, 22, 309. M. A. M. Boersma, Catal. Rev., 1975, 10, 243. W. Eberhardt, I. T. McGovern, E. W. Plummer and J. E. Fischer, Phys. Rev. Lett., 1980, 44, 200. F. L. Vogel, G. M. T. Foley, C. Zeller, E. R. Farlardeau and J. Gan, Muter. Sci. Eng., 1977,31,261. J. Parrod and G. Bienert, J. Polym. Sci., 1961, 53, 99. W. Rudorff and E. Schulze, Z. Anorg. Allgem. Chem., 1954, 277, 156. .5 A. R. Ubbelohde, L. C. F. Blackman and J. F. Mathews, Nature(London), 1959, 183, 454. ’ H. B. Kagan, Chem. Technol. 1976, 6, 510. lo G. S. Parry and D. E. Nixon, Nature (London), 1967, 216, 909. l 1 G. S. Parry, D. E. Nixon, K. M. Lester and B. C. Levene, J. Phys. C, 1969, 2, 2156. l2 K. Watanabe, T. Kondow, M. Soma, T. Onishi and K. Tamaru, Proc. R. Soc. London, Ser. A, 1973, l3 D. Saehr and A. Herold, Bull. Soc. Chim. Fr., 1965, 3130. l4 M. Colin and A. Herold, Bull. Soc. Chim. Fr., 1971, 1982. lS P. Lagrange, A. Metrot and A. Herold, C.R. Acad. Sci., Ser. C, 1974, 278, 701. l8 A. Herold, C.R. Acad. Sci., Ser. C, 1951, 232, 838. l7 A. R. Ubbelohde and F. A. Lewis, Graphite and its Crystal Compounds (Oxford University Press, 333, 51. 1960). A. Thomy and X. Duval, J. Chim. Phys. Physicochim. Biol., 1969, 66, 1966. *’ B. E. Warren, Phys. Rev., 1941, 59, 693. 2o A. M. Saxena and B. P. Schoenborn, Acta. Crystallogr., Part A, 1977, 33, 813. 21 R. Chen, P. Trucano and R. F. Stewart, Acta Crystallogr., Part A , 1977, 33, 823. 22 G. M. Wolten, Atomic Energy Commission Report NAA, SR4545 (1960). 23 D. E. Nixon and G. S. Parry, J. Phys. D, 1968, 291. 24 P. Lagrange, D. Guerard and A. Herold, Ann. Chim. E, 1978, 3, 143. (PAPER 1 / 1248)