Charge transfer in acceptor graphite intercalation compounds

Charge transfer in acceptor graphite intercalation compounds

Synthetic Metals, 34 (1989) 399-404 399 CIIARGE TRANSFER IN ACCEPTOR GRAPIIITE INTERCALATION COMPOUNDS S. FLANDROIS Centre de Recherche Paul Pascal...

302KB Sizes 1 Downloads 30 Views

Synthetic Metals, 34 (1989) 399-404

399

CIIARGE TRANSFER IN ACCEPTOR GRAPIIITE INTERCALATION COMPOUNDS

S. FLANDROIS Centre de Recherche Paul Pascal, Domaine Universitaire de Bordeaux-I, 33405 Talence C6dex (France) C. HAUW Laboratoire de Cristallographie, Universit~ de Bordeaux-I, 33405 Talence Cedex (France) R.B. MATItUR National Physical Laboratory, New Delhi 110 012 (India)

ABSTRACT We briefly review the various charge transfer mechanisms proposed for acceptor-graphite compounds. As there is no general consensus regarding even the amount of charge transfer in the case of bromine~GIC, we have studied the charge transfer during bromine desorption, using an empirical relationship based on C-C bond lengths. The data show a maximum of charge transfer for stage-4 samples, which could correspond to the presence of Br~- entities. INTRODUCTION Any intercalation reaction involves a charge transfer (CT) between host and guest species, which determines the electronic properties of the compound. Due to its amphoteric nature, graphite is able to accept electrons from donor intercalates or to yield electrons to acceptor intercalates. In the case of donor compounds (metals or alloys, and essentially alkali metals), even if some difficulties remain concerning the amount of CT or the electronic states implied in the process, the mechanism of CT is well established, as resulting from band effects (overlap of graphite and intercalate bands). For acceptor compounds, however, things are more complicated, due in part to the diversity of possible intercalates (halogens, halogenides, oxides, acids,...). The CT mechanism may depend on the nature of the species (molecular, ionic) or on their ability to give several oxidation degrees. Generally, the proposed mechanisms do not take into account the electronic levels of the intercalate. To our knowledge, only one attempt to consider band effects has been made recently for bromine compounds [1]. On the other hand many problems connected with the amount of CT have still to be solved. For instance, is the CT per intercalated molecule dependent on the stage? What is the influence of parameters such as electronic affinity or redox potentials? Is there a limit of the CT per carbon atom? This last question was especially examined by Milliken and Fischer I2l who concluded that in graphite-fluoroarsenate intercalation compounds 0.05 electron per carbon atom at the most can be extracted by oxidation. Is this limiting value general? Finally, in some cases, there is no general consensus regarding even the amount of CT betwen guest molecules and host graphite. Particularly typical is tile case of bromine: 0379-6779[89/$3.50

© Elsevier Sequoia/Printed in The Netherlands

400

the fractional negative charge per bromine molecule, referred to as f, estimated from various electronic properties, is being reported to be in the range 0.02 < f < 2 [3]. In this paper, we briefly review the various CT mechanisms proposed for acceptor compounds and the methods of C T estimates. Then we will give new data obtained from X-ray diffraction for graphite-bromine compounds, during bromine desorption. C H A R G E T R A N S F E R IN A C C E P T O R C O M P O U N D S Among acceptors, the halide group is particularly representative of the variety of mechanisms involved. As no iodide and few bromides have been intercalated [4], we will restrict our discussion to chlorides and fluorides. Stoichiometric measurements of metal chloride-GICs such as NiCI2 - G I C yield an excess of Cl, e.g. Cll.3NiCl2.13 [5]. As the metal ion keeps its oxidation degree, here N i 2+, the excess of negative charge (0.13 c - ) is compensated by graphite oxidation (0.13 positive charge for 11.3 carbon atoms). It has been shown [6] that the excess of Cl results from the intercalation of small islands (O ~ 150-200 ,~). Thus the mechanism of charge transfer is based on the nucleation of these islands and their diffusion between the graphite layers. Such a mechanism is certainly valid for other dichlorides such as CoCl2 or CdCl2. But for Mn C l 2 or CuCl2, metal vacancies must be taken into account in addition to islands [7]: the excess of Cl is too large for it to be due only to excess C l - ions surrounding the islands. It is noteworthy that these compounds give the largest CT to our knowledge, close to 0.07 positive charge per carbon atom. Although intercalation of metal dichlorides is understood in terms of the chlorine excess, the same is not true for FeCla - GIC. The chlorine excess is typically ten times smaller and the in-plane filling coefficients are very high, typically close to 0.9. From MSssbauer spectrometry, there is conclusive evidence that about 20% of the iron is in a 2+ ionization state [8]. This means a CT of 0.2 electron per iron atom, which seems independent of stage, filling coefficient and biintercalation partner [9]. Rancourt [10] has proposed a mechanism based on the creation of vacancy/interstitial pairs in the iron layers. Another possible mechanism in halide intercalation is based on disproportionation reactions. These reactions are common in metal fluorides, such as AsF~ which disproportionate into AsFa and A s F 6 [11], and occur at least in one chloride, SbCls, giving SbCl3 and SbCl 6 species [12]. The result is the coexistence of neutral and ionic entities in the intercalated layers. This coexistence of neutral and ionic species is typical of fluoride-graphite compounds. In addition to disproportionation mechanism, it can occur either without change in oxidation state (e.g., OsFs -~ O s F 6 [13], or TiF4 -* TiF4 + T i F 6 - [14]), or with total reduction of the metal (e.g., RuFs -* RuF4 + R u F 6 - [13]). The incorporation of neutral species seems also to occur in the intercalation of acids. For example, stage-1 H 2 S 0 4 - G I C s have the composition C + H S O ~ . x H 2 S 0 4 [15] and stage-1 I I N 0 3 - e l C s , C~o_24N0~.3 IINO3 [16]. The case of halogens is more controversial. Ahnost nothing is known concerning the CT in chlorine-graphite compounds. Fluorine is known to give covalent bonds with carbon by reaction at high temperatures. However, room temperature intercalation is possible in the presence of traces of fluorinated compounds [17]. A m a x i m u m of conductivity is observed for a F / C ratio of 0.075. The decrease of o at higher fluorine concentration must correspond to the formation of covalent F-C bonds. The same behavior is observed in the reaction of graphite with BrF5 : BrF5 intercalates first, then F is fixed covalently [18 I.

401

It is worth noticing that polyacetylene has been found to fix chlorine in the same way during FeCl3 intercalation, when the CT reaches 0.055 electronic charge per C H [191. As mentioned above, the amount of CT in bromine-graphite compounds varies largely according to the authors. There is even no general consensus regarding tile exact stoichiometry or size of the unit cell: stoichiometries C3.5,~Br, C4,~Br and CTnBr, where n is the stage index, have been proposed, based on weight uptake or X-ray diffraction. In the following, we will present new data obtained by X-ray diffraction during bromine desorption.

EXI'EItlMENTAL METltODS OF CIIARGE T R A N S F E R ESTIMATE Many more or less direct methods have been used for the determination of CT in G1Cs: coulometry, NMR, measurement of magnetic susceptibility, optical reflectivity, transport properties, quantum effects (De Haas-Van Alphen, Shubnikov-de Haas)... Most of these methods, which give the Fermi energy, need the use of a model [20 I. Another method is based on the accurate determination of in-plane C-C bond lengths. It ha~s been previously shown that the C-C bond length cilanges by intercalation and that the variation can be correlated with the amount of CT [5t. For electron-donor GICs the bond length increases, whereas for electron-acceptor GICs a contraction of ttlese bonds is observed. Several theoretical studies have been devoted to this problem [21] and an empirical relationship between CT and C-C bond length has been proposed for acceptor compounds, based on the chlorine excess observed in metal chloride-GICs [71. This relationship is: d c _ c ( ) t ) 1.4209 - 0.072 p where p is the CT per carbon atom and 1.4209 is the C-C bond length in pristine graphite. It is shown in Fig. I together with the ttleoretical results (dashed lines) from Pietronero and Str'hssler (PS) and Chan et al. (CKIIE). We used this relationship to estimate the CT in bromine graphite compounds.

RESULTS AND DISCUSSION High purity natural graphite |lakes (~ .5 mm din) from Madagascar have been choserl for intercalation with bromine in liquid phase. The sample was allowed to be in contact with bromine for various timings ranging from 6 to 144 hrs. After intercalation the excess bromine was removed by vacuum desorption at room temperature. Immediately after the operation tile tube wa.s sealed and weighed, in order to lind out the amount of bromine uptake. A part of tile sample was taken in another weighed tube to determine weight loss of the sample with time after desorption at room temperature. X-ray diffraction studies on tile desorption kinetics were carried out on an Inel X-ray diffractometer with curved position sensitive detector and multichannel analyzer. Tile sample (small flakes) was taken in a Lindemann tube and its position was not changed during tile desorption. X-ray diffraction studies on a single crystal graphite flake have been carried out on a Siemens three-circle diffractometer. The sample was kept in the Lindemann tube fixed on a goniometer head. Table t shows tile amount of bromine uptake by graphite flakes kept in liquid bromine for different timings.

402

TABLE 1

Sample

Treatment time (hrs}

Bromine uptake (% by weight)

Nominal stoichiometry

40 74 80

C16.6Br CgBr Cs.3Br

6 78 144

Samples 2 and 3 are close to the stoichiometry CsBr given by Rudorff [22] for stage-2 compound, whereas sample 1 correspond to stage-4. However, sample 2 and 3 lose about 30 % bromine immediately when exposed to air atmosphere. Then the desorption is very slow. Thus, it appears that the stage-4 has a certain stability compared to lower stages. A characteristic of the X-ray patterns is the occurrence of two peaks at about d = 4.2 A and d = 2.8 A. With subsequent desorption from the stage-4 compound, the first one at d ~ 4.2 A splits up into two well defined peaks at d "" 4.3 A and d -~ 4.1 A respectively. Both the peaks shift towards higher d values for longer desorption time, whereas the peak at d = 2.8 A shifts towards lower d values. Such a behavior has been observed by Bardhan et al. [23] as a function of temperature on a sample of dilute HOPG - Br2. It was attributed to a commensurate-incommensurate phase transition occurring at 340 K, before the melting of the intercalate layers at 374 K. If this interpretation is correct, it means that the bromine desorption produces a change in in-plane structure from commensurate for stage-4 to incommensurate for higher stages. In order to determine the exact value of CT from the C-C bond length we estimated accurately the shifts in the (110) reflections with respect to pristine graphite. (00£) reflections were also recorded for the stage determination. Table II depicts the results.

TABLE II Charge transfer estimates

Stage

d c - c ( ~4)

Charge transfer (e-/C atom)

II (lst sample) II (2nd sample) I1 + IV IV IV + VI V I (lst sample) V I (2nd sample)

1.4198(2) 1.4196(2) 1.4192(2) 1.4187(2) 1.4202(2) 1.4204(2) 1.4204(2)

0.015 0.018 O.023 0.031 0.010 0.007 0.007

* ass~/ming a stoichiometry CsnBr (see text).

4, 44, ± 4, 4, 4,

0.003 0.003 O.OO3 0.003 0.002 0.002 0.002

Charge transfer (e-IBr atom)* 0.15 0.18 0.35 0.62 0.25 0.21 0.21

4- 0.03 4- 0.03 ± 0.04 ± 0.06 5= 0.05 4, 0.05 5= 0.05

403

This table shows: i) the good reproductibility of the results obtained for two different single crystals of stage-2 and 6, ii) the CT is maximum for stage-4 compounds and would approximately correspond to Br~- species. The CT per Br atom has been calculated for a stoichiometry Cs,~Br, because from a comparison of the intensities observed and calculated for the (00~) peaks this stoichiometry gave the best agreement, whatever the stage. In addition, it is in the middle of the range previously proposed (from C3.snBr to C~nBr). A maximum of CT is at first view surprising. It is generally assumed that the CT is higher for highstage compounds. Thus, a decrease from stage-4 to stage-2 is expected. It must be noted that the larger stability found for stage-4 in our desorption study is in agreement with this finding. The decrease of CT from stage-4 to stage-6 should be correlated to the occurrence of an incommensurate structure. Such decreases in CT at a commensurate-incommensurate transition have been observed in other GIC with K or RuF5 for example. Finally, a CT maximum for stage-4 could explain the conductivity maximum generally observed for this stage in acceptor compounds.

Gr,phit= 1.420

? -o

""~'~"~""'6~CI3 Fig. 1. Charge transfer per carbon atom as a function of C-C bond length. Full line: empirical relationship from ref. [7], based on the data points obtained for several metal halide GICs. Dashed lines: theoretical calculations [211 from Pietronero and Str~ssler (PS) and Chan et at. (CKIIE).

1.415

,PS 0.05 I

030 1



C.T. / C atom

REFERENCES 1 M.S. Dresselhaus and G. Dresselhaus, Synth. Met., 12 (1985) 79. 2 J.W. Milliken and J.E. Fisher, J. Clmm. Phys., 78 (1983) 5800. 3 G. Dresselhaus and M.S. Dresselhaus, Mater. Sci. Eng., 31 (1977) 235. 4 E. Stumpp, Mater. Sci. Eng.. 31 (1977) 53. E. Stumpp, Intern. Coll. on Layered Compounds, Ed. by D. Gudrard and P. Lagrange, Pont-&-Mousson (1988) p. 41. 5 S. Flandrois, J.M. Masson, J.C. Rouillon, J. Gaultier and C. Hauw, Synth. Met., 3 (1981) 1. 6 S. Flandroks, A.W. Hewat, C. H a u w and R.H. Bragg, Synth. Met., 7 (1983) 305. 7 F. Baron, S. Flandrois, C. H a u w and J. Gaultier, Solid State Commun., 42 (1982) 759. 8 M. Prietsh, G. Wortmann, G. Kaindl and R. SchlSgl, Phys. Rev. B, 33 (1986) 7451. 9 D.G. Rancourt, B. Hun and S. Flandrois, Can. J. Phys., 66 (1988) 776.

404

10 D.G. Rancourt, in Chemical Physics of Intercalation, Ed. by A.P. Legrand and S. Flandrois, Plenum, N.Y. (1987) p. 79. ] 1 N. Bartlett, B. McQuillan and A.S. Robertson, Mat. Res. Bull., 13 (1978) 1259. 12 P. Boolchand, W.J. Bresser, D. McDaniel, K. Sisson, V. Yeh and P.C. Eklund, Solid State Commun., 40 (1981) 1049. 13 S. Flandrois, J. Grannec, C. Hauw, B. Hun, L. Lozano and A. Tressaud, J. Sol. St. Chem., 77 (1988) 264. 14 S. Flandrois, B. Hun, J. Grannec, A. Tressaud and C. Hauw, Synth. Met.~ 23 (1988) 435. 15 A. M(~trot and J.E. Fischer, Synth. Met., 3 (1981) 201. 16 S. Loughin, R.A. Grayeschi and J.E. Fischer, J. Chem. Phys., 69 (1978) 3740. 17 M. Touhara, Y. Goto, N. Watanabe, K. Imaeda, T. Enoki, tI. Inokuchi and Y. Mizutani, Synth. Met., 23 (1988) 461. 18 A.M. Danilenko, A.S. Nazarov and I.I. Yakoviev, Bull. Acad. Sci. USSR, 37 (1988) 827. 19 S. Flandrois, A. Boukhari, A. Pron and M. Zagorska, Solid State Commun., 67 (1988) 471. 20 J.P. lssi, in "A travers la Physique", published by les Editions de Physique, Paris (1983) p. 365. 21 L. Pietronero and S. Str~sler, Phys. Rev. Left., 47 (1981) 593; M. Kertesz, Mol. Cryst. Liq. Cryst., 125 (1983) 103 C.T. Chan, W.A. Kamitakahara, K.M. Ho and and P.C. Eklund, Phys. Rev. Lett., 58 (1987) 1528. 22 W. Riidorff, Z. Anor~. All~. Chem.~ 245 (1941) 383. 23 K.K. Bardhan, J.C Wu and D.D.L. Chung, Synth. Met., 2 (1980) 109.