Synthetic Metals, 3 (1981) 269 - 278 © Elsevier Sequoia S.A., Lausanne -- Printed in the Netherlands
269
E F F E C T OF REACTION CONDITIONS ON THE BASAL RE SIST IV IT Y OF G R A P H I T E - A C C E P T O R I N T E R C A L A T I O N COMPOUNDS*
M. J. MORAN**, J. W. MILLIKEN, C. ZELLER***, R. A. GRAYESKI and J. E. FISCHER Moore School of Electrical Engineering and Laboratory for Research on the Structure of Matter, University of Pennsylvania, Philadelphia, PA 19104 (U.S.A.) (Received May 19, 1980)
Summary In this paper we explore the effect o f variations in reaction conditions on the basal resistivity Pa of graphite intercalated with HNO3 and AsFs. In the former case, the reaction rate is varied over a wide range by diluting the reactant vapor with inert gas. In the latter case, several variations which might be e x p e c t e d to alter the fractional ionization are explored. None of these treatments produces a significant effect, from which we conclude that intrinsic values of pa are being obtained with these compounds. Furt her improvements in p~ will t he r ef or e rely on development of new materials.
Introduction One o f the more dramatic features of graphite intercalation compounds is the large e n h a n c e m e n t of electrical conductivity parallel to the layers (oa), relative to the pure graphite value. This e n h a n c e m e n t is now u n der s to o d as being due to the net effect of a large increase in free carrier c o n c e n t r a t i o n (or N ( E F), the density o f electronic states at the Fermi energy) [ 1 ] , and a more m o d e r a t e decrease in carrier mobility [2]. The highest o~ values at 300 K are currently achieved with acceptor-type intercalants; these do n o t give the largest N ( E F) but apparently the mobility is degraded less than with d o n o r intercalants [ 1 ]. For practical as well as fundamental reasons, it is i m port ant to ascertain wh eth er the intrinsic oa is achieved for a given intercalant. The large scatter in values r e por t ed by different groups at different times can, in principle, be partitioned between measurement inaccuracies and real variations among samples. In the first category, the advantages o f contact*Supported by the NSF Materials Research Laboratory Program, Grant No. DMR 79-23647. **Present address: Naval Research Laboratory, Washington, D.C. 20375, U.S.A. ***Present address: Pitney-Bowes Corp., Norwalk, CT 06852, U.S.A.
270 less a.c. methods have been well-documented [3], and several versions of inductively coupled apparatus have been described [3 - 5]. The latter category has received less systematic attention, and is the focus of this paper. Different ua values might be expected from different samples which are nominally the same chemically, for several reasons. First, the quality of the initial graphite plays a role, the most striking example being the difference between results obtained on intercalated fibers and HOPG. With commercial fibers, one presumes that the 300 K mean free path is defect limited, both before and after intercalation. On the other hand, the typical in-plane crystallite size of HOPG, ~- 1 pm, is large compared with the 300 K e l e c t r o n - p h o n o n mean free path, since oa (HOPG) is the same as the single crystal value. Intercalation decreases the phonon mean free path, so the crystallite size in HOPG should not be a limiting factor. Additional defect scattering might be introduced during intercalation. Several possibilities come to mind: (i) gross defects associated with the large c-axis expansion; (ii) edge dislocations created during the formation of islands [6] within which the graphite stacking must change from ABAB .... to AIA .... , where I denotes the intercalant layer; (iii) boundaries among rotationally-disordered islands with different intercalant stacking sequences [7], etc. The relative importance o f these and other phenomena would probably vary significantly from one intercalant to another. Finally, the chemical reactions which occur during formation of acceptor compounds suggest several possibilities for variable o~ values. The degree of oxidation of the graphite, as well as the possibility of side reactions, may both be expected to vary with reaction conditions. In particular, many acceptor dompounds of high stage can be synthesized by two different approaches, which may be termed "equilibrium" and "metastable". In the equilibrium approach, graphite is exposed to the equilibrium vapor pressure corresponding to the desired stage n, and the reaction goes to completion [5]. This requires some degree of temperature control. The metastable approach consists of using a pressure which would give stage N at equilibrium, but terminating the reaction prematurely at stage n > N. The advantage of the latter is that high stage compounds can be obtained without temperature control [8] (if the kinetics are slow enough), but there is no guarantee that the resulting compounds correspond to equilibrium compounds. The systematic investigation of even one of the above-mentioned phenomena would be a major undertaking. We report here the results of two series of experiments which provide qualitative information on the relative importance of some of these phenomena. The first series consists of a study of o~ (300 K) in HNO3 compounds with 2 ~ n ~ 9 as a function of intercalation rate. The principal results are that a~ (300 K) for stage 2 is inde. pendent of intercalation rate over a wide range, and that a~ is the same for
271 "metastable" and "equilibrium" preparations. The second series involves stages 1, 2, and 3 compounds CsNAsFs, for which values of aa (300 K) differing by a factor of two have been reported by different groups [5, 9]. Here we find t h a t the source of discrepancy is most likely the measurement technique rather than the method of preparation, and that variations in preparation chemistry which might be expected to affect the degree of oxidation have little or no effect on oa. Our overall conclusion, therefore, is that intrinsic oo values are, in fact, being obtained with these intercalants, and that further improvements in oa will come from new compounds and not from "improved" preparations of existing materials.
Experimental Compounds were prepared from 5 × 5 mm 2 plates of HOPG cleaved to thicknesses in the range 0.15 - 0.35 mm. Measurements of aa were made before and after intercalation using an r.f. induction method based on a slotted ferrite core. The early HNO3 measurements used the 100 kHz apparatus described in ref. 4, with the same calibration as was employed for the AsF5 measurements in ref. 9. The later HNO3 data, and all the AsF5 data reported here, were obtained with a 1 kHz version which eliminates skin depth problems. In the older apparatus, since the sample thickness exceeded the skin depth, 5, the actual sample thickness (including effects of exfoliation) was used to calculate o~ from the o u t p u t signal. The advantage of the 1 kHz apparatus is that t < 5, so the appropriate thickness is the ideal value, obtained either from the mass, area, and density of the initial graphite or from the interlayer spacing of the c o m p o u n d (determined from (00/) X-ray diffractograms). As an example of the performance of the 1 kHz apparatus, eight plates cleaved from the same piece of HOPG gave aa = 40.0 -+ 1.4 p~2 cm, a standard deviation of 3.5% (based on the ideal thickness, corrected for density deficit). Compounds of graphite nitrate were prepared by exposing HOPG to the saturated vapor of 100% HNO~ at 300 K in closed pyrex tubes. The reaction rate was reduced by adding a known Ar pressure to the tube before sealing. This procedure also permitted measurements to be performed on metastable high stage compounds. At equilibrium, stage 2 was obtained with both the c o m p o u n d and the acid at room temperature. The details of this procedure have been described elsewhere [8, 10]. Compounds of AsF5-graphite, of stages 3, 2, and 1, were prepared by a variety of methods. The basic procedure was that described by Falardeau et al. [11], i.e., exposing the HOPG, at 300 K, to 1 - 2 atm AsF 5 in a closed glass volume. They showed that material of uniform stage 3, 2 or 1 could reliably be obtained by stopping the reaction at the desired stage, as determined by X-ray and/or measuring the c-axis thickness. This is possible because the reaction is very much slower than with HNO~ in the absence of Ar. Uniform higher stage material cannot be obtained by this method. For some samples, the resistivity was measured continuously during intercala-
272 tion. In this w a y t h e same sample p r o v i d e d resistivity data for stages 3, 2, and 1 a n d f o r the " m i x e d " stages during the t r a n s i t i o n f r o m stage t o stage. The resistivity did n o t c h a n g e a p p r e c i a b l y in the time required t o ascertain the stage o f t h e sample (several minutes). I n d e e d , m a n y o f these " m e t a s t a b l e " p r e p a r a t i o n s required several days t o reach pure stage 1. S o m e samples were p r e p a r e d using m o d i f i c a t i o n s o f this basic procedure as discussed in individual cases in the Results section below. Results and discussion - - HNO3 R o o m t e m p e r a t u r e d a t a for 13 samples are s u m m a r i z e d in Table 1. With no a r g o n t o retard t h e reaction, we were unable t o observe pure stages higher t h a n n = 2 (sample # 3 0 ) . Most o f the o t h e r s were allowed to go t o c o m p l e t i o n , w h i c h required several days. S o m e samples were sealed o f f at high stage for o t h e r e x p e r i m e n t s ( # ' s 15, 34, 36, 38). C o n t i n u o u s m o n i t o r i n g o f (00l) X-ray p a t t e r n s (Mo K s ) p e r m i t t e d the m e a s u r e m e n t o f Pa for high stages. F o r t h e 100 k H z e x p e r i m e n t s we include o n l y t h o s e samples for w h i c h t h e t h e o r e t i c a l and observed thicknesses disagreed b y less t h a n 10%; the observed thickness is used in the analysis for Pa. This criterion led us t o e x c l u d e all samples m a d e u n d e r zero argon, d u e to the severe e x f o l i a t i o n (as m u c h as a f a c t o r o f 2 in thickness). TABLE 1 Basal plane resistivity Pa for graphite intercalated with HNO 3 to various stages, N, as indicated Values obtained at 300 K using an inductively coupled a.c. technique (ref. 4) operating at 100 or 1 kHz, as indicated. In most cases, the reaction was slowed down by adding Ar to the HNO 3 vapor in the sealed reactor; the various Ar pressures are indicated. Sample No
Ar pressure (arm)
Frequency (kHz)
Pa at stage N (p~'/cm)
3 12 15 18 Vogel ref. 12
2.O 2.0 2.0 2.0 0
100 100 100 100 100
2.59 2.36
30 31 32 33 34 35 36 37 38 Vogel ref. 12
0 3.1 3.1 3.1 5.6 1.8 5.6 5.6 5.6 0
1 1 1 1 1 1 1 1 1 1
4.01 3.57 3.63 3.72 3.53 4.44 3.1
3
4
5
6
3.82 3.93
4.23
3.48
3.80
2.80 2.11 2.8 -3.8
- 3.6
3.34
3.56
-
3.54
3.38 3.90
3.95
-
3.59
4.38 3.34
4.16 3.22
273 Using the 100 kHz method, the mean and standard deviation for stage 2 is 2.35 -+ 0.27 pgt cm. Vogel e t al. [12] report a large spread in 100 kHz values on an unspecified number of samples, 2.8 < Pa < 3.8 p~2 cm. We emphasize that the observed sample thickness was used in our analysis, which is appropriate if t < 5. The I kHz method permitted us to study the effect of intercalation rate, since reliable results were obtained with p(Ar) = 0 despite exfoliation, by using the ideal thickness (determined from the ideal thickness of the starting HOPG in conjunction with the X-ray results). With this lower frequency, the average of all six stage 2 samples is 3.82 + 0.35 ugt cm. (Eliminating the two poorest values (#'s 30 and 37) gives a much smaller spread, 3.61 -+ 0.08 p~2 cm.) Vogel e t al. report a range of 1 kHz values 3.1 < po < 3.6 pgt cm for stage 2 with no argon. Therefore, we find no significant differences in pa (1 kHz) between preparations with p(Ar) = 0 (our # 3 0 plus Vogel e t al. ) and p(Ar) = 1.8 or 3.1 atm. (The Pa value for #37 is anomalously large, but this is probably not characteristic of the greater p(Ar), since two other identical preparations (#'s 36 and 38) gave stage 3 and 4 values more in line with the 3.1 atm results.) This negative result is somewhat surprising, since the reduced intercalation rate (~1 h with p(Ar) = 0 vs. ~ 5 0 h with 5.6 atm) has a marked beneficial effect on the integrity of the sample. Rapid intercalation leads to edge fraying, highly distorted surfaces, non-uniform thickness expansion, and average thickness in excess of the theoretical value. All of these problems are largely alleviated with even small pressures of argon; the improvement seems to be directly related to reaction rate, since AsF5intercalated graphite requires times comparable with HNO3 with p(Ar) ~ 3 atm, and the former samples exhibit excellent gross morphology. Evidently the defects associated with these macroscopic distortions of the sample geometry have no effect on the room temperature scattering of charge carriers. One might have also postulated a beneficial result from requiring the c o m p o u n d to pass through successive uniform high stages e n r o u t e to stage 2. The structural reorganization of carbon layers could presumably be more complete if it began in coherent fashion over long distances (i.e., starting at high stage). In this context, the null result can be interpreted as being consistent with an essentially two
274 5
I
I
I
I
]
!
- ~ . . . ~ o . 4
__
l
0 0
~34 ~38
•
VOGEL
""'"-,.. "c-.........o ..........
d_ 3
--
GRt,PHITE - HNO3 300 K 8
I
7
]
6
I
I
5 4 STAGE
] 3
Fig. 1. Plots of P a v s . stage for 3 samples of HNO3-graphite. Data from ref. 12 pertinent to stage 2 are included for comparison. All data taken at 1 kHz. The differences between equilibrium samples (closed circle with error bar) and metastable samples (open triangles, circles and squares) are probably insignificant statistically.
tional to the fourth power of the basal plane dimension, d, for a square sample [4], so A p / p = 4 A d / d . Thus 2% errors in length measurement could explain the spread in stage 2 Pa values in Fig. 1. Errors of this magnitude are difficult to avoid, since the cut sides of the HOPG plates are seldom rigorously perpendicular to the cleavage faces. Errors in Pa associated with dimensional uncertainties also affect studies of pa vs. stage when a different sample is used for each stage (i.e., the equilibrium m e t h o d of Vogel e t al. [12] ). The metastable approach has an obvious advantage here, since dimensional errors produce only a shift of the entire p vs. stage curve without altering its shape. Thus, from Fig. I we observe that the minimum p, occurs at stages 3 or 4, the former obtaining with 3.1 atm argon (#'s 31 and 33), while the latter occurs with 5.6 atm (#'s 37 and 38). Clearly, this subtle effect of reaction rate on the location of p, (minimum) vs. stage can only be explored with the metastable approach. The trend observed here is consistent with Vogel's conclusion that p, (minimum) occurs "near stage 2" with p(Ar) = 0. The microscopic origin of this effect is not at all obvious, but the variations in Pa thus produced are hardly of practical significance. Finally, we call attention to the discrepancy between 100 kHz and 1 kHz values in Table 1. The synthesis methods were the same, apart from minor variations in p(Ar). The discrepancy in pa values is significant, about 5 times larger than either the 1 kHz or 100 kHz standard deviation. A similar discrepancy, albeit of smaller magnitude, was observed by Vogel e t al. [12]. Much effort was expended in the calibration of the 100 kHz instrument; in particular, exfoliation was simulated by a stack of copper foils separated by thin insulating sheets, and it was verified that the appropriate thickness to use in the analysis was the total value, not merely the fraction corresponding to conductor [4]. There is no reason to expect a large intrinsic frequency dependence of p~, so the discrepancy remains a mystery. We simply note that this same problem may be the origin of differing p~ data for AsF5 compounds, the 100 kHz values of Foley e t al. being roughly half those of
275 Interrante e t al. [5]. Our own results, described in the next section, agree much better with the latter than with the former. Results and discussion -- AsF~
The room temperature Pa values for AsF5-graphite are given in Table 2. These were all obtained using the 1 kHz technique described above. There are 53 values reported, mostly for pure stages. These are grouped in 3 categories depending on synthetic method. The first and largest category corresponds to metastable c o m p o u n d s prepared by the standard method described above [11]. The second group are derivatives in which the metastable compound is exposed to dynamic vacuum [13]. The chemical effect of this treatment is the removal of AsFs; in the case of stage 1, 60% of the initial AsF5 equivalents are removed. The final category contains samples for which the departure from standard technique was more radical, as noted in Table 2 and discussed below. Including all synthesis categories, the average values for pure stages 1, 2, and 3 are 3.23 -+ 0.37, 3.26 + 0.28 and 3.20 -+- 0.07 p~2 cm, respectively. On average, the stage dependence of pa is therefore nil. The 8 mixed stage samples give 3.63 + 0.44 p ~ cm, a slightly higher value. Trends show up more clearly b y following the progression of a single sample, because dimensional errors cancel. Consider samples 62-1 and 3-2. It is clear that p~ is significantly higher when the samples are mixed stage (3 + 2 or 2 + 1) than when they are pure stage (2 or 1). This effect has been noted earlier [9]. These two samples plus sample 10-2 all exhibit higher Pa at stage 1'than at stage 2, so the concentration dependence is probably real, although small. The effect of removing AsF~ from stage 1 and 2 c o m p o u n d s by pumping at room temperature has been reported in detail elsewhere [13]. The resistivity results of that study are included in Table 2 as the second category. The first 3 samples in this category were transformed from stage 2 c o m p o u n d s to stage 3', where the prime indicates that the sandwich thickness is ~ 0 . 5 A smaller than usual. These 3 samples (4-30-3,125-2 and 126-2) show that the 300 K p~ of stage 2 AsFs-graphite is not measurably increased when a large fraction (40%) of the AsF 5 is removed by pumping. (Resistivity values in parentheses correspond to pumped c o m p o u n d s exhibiting the reduced layer spacing noted above.) This suggests that the de-intercalated AsF~ was not active in electron transfer from the graphite. In the analogous experiment on stage 1, however, p~ is significantly increased b y pumping (samples 4-2-2, 4-2-4 and 149-2). We have shown [13] that a stage 1 c o m p o u n d , CsAsFs, transforms under pumping to a mixture of stages 1' and 2', with composition C20AsFs. (As in the transformation 2 -* 3', the layer spacings for stages l ' a n d 2' are ~ 0.5 A smaller than the corresponding stage 1 and 2 values.) The Pa increase encountered here is more likely analogous to the mixed stage behavior discussed above than to the removal of AsF 5 p e r se. Defects introduced during de-intercalation might also be considered as a source of the increased resistivity.
276 TABLE 2 AsF5-graphite Pa values, all obtained at 300 K and 1 kHz, corresponding to pure stages 1, 2 and 3 and to mixed stages Parentheses denote compounds exposed to dynamic vacuum, which leads to reduced interlayer spacing in addition to removing up to 60% of the intercalant. Sample No.
Pa at stage N ( p ~ cm) 1
RFG-45 RFG-51 RFG-47 62-1 3-2 6-2 10-2 14-2 RFG-602 RFG-702 RFG-603 9-21-A 9-21-B 9-21-C 9-21-D 9-21-E RFF-706 RFF-709 10-12-E 10-18-A 10-18-D 4-2-3 4-30-1 JM-7 JM-13 4-30-3 125-2 126-2 4-2-2 4-2-4 149-2 70-2 150-2 10-18-E JMMM-1 JM-14 JMMM-4 JM-8
1&2
3.1 3.3 3.1
3.4 3.8
3.2 3.2 2.7 2.8 3.1
2 3.2 3.0 3.7 3.2 3.0 3.4 3.1
2+3
3
Preparation of compound (see text) Standard method (ref. 11 )
3.6 3.2
3.1 3.2 3.5 3.5 3.0 3.2 2.8 3.8 3.1 3.2 3.1 3.5 4.2 3.6 3.1 3.5 3.4 3.5 3.5 3.1 2.8 3.5
(3.2) (3.3) (3.2)
Pumped to remove AsF 5 (ref. 13)
(4.4) (4.1) (3.5) 3.3 3.4 3.7 2.8 2.8
3.2
Liq. AsF 5 at --63 °C AsF 5 in AsF 3 Trace H20 Eq. pressure Eq. pressure Eq. pressure Eq. pressure
F i n a l l y , we t u r n to the effects of m o d i f y i n g the r e a c t i o n chemistry. S a m p l e 7 0 - 2 was p r e p a r e d i n a m u c h d i f f e r e n t m a n n e r f r o m u s u a l , b e i n g i m m e r s e d i n l i q u i d A s F 5 a t - - 6 3 °C f o r 3 d a y s . T h e r o o m t e m p e r a t u r e Pa o f t h i s s a m p l e is a m o n g t h e l o w e s t v a l u e s m e a s u r e d . H o w e v e r , m a n y s u c h
277 samples would have to be measured to determine if this is significant, since values as low (or lower, cf. RFG-602) are sometimes observed with the standard technique. In the standard technique, the AsF 5 is carefully purified of AsF3 by trap-to-trap distillation under vacuum. The results of sample 150-2 indicate that this is not important to the resistivity, however, since this sample was intercalated to lustrous blue stage 1 (confirmed by X-ray) with AsF5 dissolved in liquid AsF3. This shows that AsF3 is not an effective reducing agent for graphite oxidized by AsF5 [13], as has been previously suggested [14]. In the standard method, samples are typically prepared by scrupulously drying the glassware before intercalation by flaming out under dynamic vacuum. Sample 10-18-E, whose conductivity both at stage 3 and stage 2 are typical of the other results presented here, was intercalated in a glass reactor to which a drop of water had been added and which was only briefly pumped out to remove the liquid water. The remaining adsorbed water presumably reacted with some of the AsF5 to give HF and some arsenic oxide (a small but significant a m o u n t of white solid was seen on the glass). The resistivity results show that no significant change was effected by this treatment. The last group of four Pa values represent equilibrium stage 2 compounds, as distinct from all the other stage 2 values which were obtained on metastable compounds e n r o u t e to stage 1. Samples JMMM-1 and JM-14 were prepared by introducing just enough AsF5 vapor so that the final pressure at stage 2 would be near the equilibrium value 130 - 170 Torr. Samples JMMM-4 and JM-8 were prepared similarly, except that the equilibrium pressure was defined by the saturation vapor pressure [5] over a --78 °C reservoir of liquid AsF5 (136 Torr). In general, the values obtained in the 4 equilibrium syntheses are statistically in line with the much larger group of metastable values, as was found in the case of HNO3-graphite. Within the smaller group of equilibrium samples, there appears to be an advantage in using a --78 °C bath to control P (AsF5) rather than simply reducing the total a m o u n t of AsF5 in the reactor. Two possible reasons could be that the bath keeps P (AsF5) rigorously constant throughout the reaction, and/or that the liquid AsF 5 traps any AsF 3 that might be present. Neither of these factors appeared to be important in the large number of samples discussed earlier. It is not clear whether the low Pa values for JMMM-4 and JM-8 are statistically significant; similarly low values occur in 3 samples within the metastable group. One would have to do a large number of preparations with a --78 °C bath to clarify this question. In summary, the 300 K Pa of stages 1, 2, and 3 AsFs-gvaphite, measured on m a n y samples with the 1 kHz ferrite core technique, is ~ 3 . 2 ~ 2 cm. Changes in reaction parameters (presence or absence of traces of H20 or large excess of AsF3, liquid or gas phase reaction, equilibrium vs. metastable conditions, and even the removal of 40% of the intercalant) have no significant effect on pa. This implies that the average values reported here are
278
representative of the intrinsic performance of the material, and that improvements in Pa by small changes in reaction conditions will not prove fruitful. One possible exception to this conclusion could be the addition of fluorine b y exposing pure stage compounds to F2. Bartlett's work [14] suggests that such a procedure might increase the fraction of electrically active species b y converting AsF3 or AsF5 to AsF6-, although parallel direct fluorination of the graphitic part of the structure would be detrimental to Pa by destroying the resonant macromolecular character of the carbon sheets. These possibilitities will be explored in future work. The results reported here for AsF~-graphite agree very well with the data of Interrante et al. [5], who found a mean pa of 3.2 ~Q cm for stage 2 using t w o different a.c. techniques (both of which differ from ours). Both the present results and those of ref. 5 contradict the original report b y Foley e t al. [9] who gave a range of stage 2 values 1.9 ~ p~ ~ 1.5 p~2 cm. Since the synthetic methods in all three studies overlap, and since we have shown that rather radical departures from the standard techniques produce little or no effect on p,, we conclude that the data of Foley et al. contain a systematic error, despite their extremely careful calibration procedures. The same systematic error is probably the origin of the discrepancy between 100 kHz and 1 kHz values for HNO3 c o m p o u n d s (Table 1). References I 2 3 4 5 6 7 8 9 10 11 12 13 14
R e v i e w e d in J. E. F i s c h e r , Physica, 99B ( 1 9 8 0 ) 383. C. Zeller, L. A. P e n d r y s a n d F . L. V o g e l , J. Mater. Sci., 14 { 1 9 7 9 ) 2 2 4 1 . E. M c R a e a n d A. H 6 r o l d , Mater. Sci. Eng., 31 ( 1 9 7 7 ) 249. C. Zeller, A. D e n e n s t e i n a n d G. M. T. F o l e y , Rev. Sci. Instr., 50 ( 1 9 7 9 ) 6 0 2 . L. V. I n t e r r a n t e , R. S. M a r k i e w i c z a n d D. W. M c K e e , Synth. Met., 1 ( 1 9 7 9 / 8 0 ) 287. N. D a u m a s a n d A. H 6 r o l d , Bull. Soc. Chim. Fr., 5 ( 1 9 7 1 ) 1 5 9 8 . P. L a g r a n g e , D. G u 6 r a r d a n d A. H 6 r o l d , Ann. Chim. (Paris), 3 ( 1 9 7 8 ) 143. C . C . S h i e h , R. L. S c h m i d t a n d J. E. F i s c h e r , Phys. Rev. B , 2 0 ( 1 9 7 9 ) 3 3 5 1 . G. M. T. F o l e y , C. Zeller, E. R. F a l a r d e a u a n d F. L. V o g e l , Solid State C o m m u n . , 24 (1977) 371. S. L o u g h i n , R. A. G r a y e s k i a n d J. E. F i s c h e r , J. Chem. Phys., 69 ( 1 9 7 8 ) 3 7 4 0 . E . R . F a l a r d e a u , L. R. H a n l o n a n d T. E. T h o m p s o n , Inorg. Chem., 17 ( 1 9 7 8 ) 301. F. L. V o g e l , H. F u z e l l i e r , C. Zeller a n d E. M c R a e , Carbon, 17 ( 1 9 7 9 ) 2 5 5 . M . J . M o r a n , J. E. F i s c h e r a n d W. R. S a l a n e c k , J. Chem. Phys. 73 ( 1 9 8 0 ) 6 2 9 . E. M. M c C a r r o n a n d N. B a r t l e t t , t o b e p u b l i s h e d ; N. B a r t l e t t , B. M c Q u i l l a n a n d A. S. R o b e r t s o n , Mater. Res. Bull., 13 ( 1 9 7 8 ) 1 2 5 9 .
N O T E A D D E D IN PROOF Several groups now (see, for example, the previous 2 papers in this conference, and [5] ) have re-investigated the conductivity of AsF s-graphite compounds, after the early work by Foley et al. 19l • For "pure-stage" compounds, there is remarkable agreement among these groups that there is very little, if any, difference among the conductivities of stages 3, 2 and 1. As noted earlier [11], these three are the only stages which have been obtained " p u r e " : well-characterized samples of stage n ~ 3 have not yet been reported. The "high stage" vs. conductivity data reported in Fig. 7 of [5] must be considered with caution, since At/t o data alone axe not a reliable indication of stage. If these stage (3, 2 and 1) vs. conductivity data (on which several groups, using different measurement and sample preparation techniques, agree) have any bearing on the question of chargetransfer per intercalant, " f " , they lend at least prima facie support to the idea that " f " decreases as the stage changes from 3 to 2 to 1. If every three AsF s molecules, regardless of stage, created two holes in the graphite band, according to 3AsFs + 2e- = 2AsF6- + AsF3, then stage 2(Cl6AsFs) ought to have 50% more carriers than stage 3 (C24AsFs ), since it has 50% more interealant. The fact that the conductivity of stage 2 is equal to that of stage 3, and is not 50% higher, suggests the possibility that the charge transfer per intercalant is smaller for stage 2 than for stage 3.