J. MATER. CHEM., 1991, 1(5), 735-738 Chemical Reduction of FeCI,-Graphite Intercalation Compounds with Potassium-Naphthalene Complex in Tetrahydrofuran Ali Messaoudi,a Michio Inagaki" and Franqois Beguinb a Faculty of Engineering, Hokkaido University, Kita-ku, Sapporo, 060 Japan Centre de Recherche sur la Matiere Divisee, CNRS la, rue de la Ferollerie, 45071 Orleans Cedex 02, France The reduction reaction between FeCI, intercalated into graphite and potassium has been studied. The reduction of FeCI, to metallic Fe and KCI occurred in the graphite gallery and, as a consequence, the reduction products were supposed to be confined within the graphite matrix. By coupling energy-dispersive X-ray spectroscopy (EDX) and X-ray diffraction (XRD) analyses of the products obtained during the reaction, it was concluded that the reaction between tetrahydrofuran(THF)-solvated K and FeCI, was the rate-determining step, and not the diffusion of THF-solvated K ions into the graphite gallery. Experimental results indicated that a strong constraint for the reaction was the limited space in the graphite gallery.When heated at different temperatures, the flakes of the product showed remarkable exfoliaton with subsequent appearance of metallic Fe and KCI as reaction products on the surface. Keywords: Graphite; Intercalation; Reduction; lron(II1) chloride; Exfoliation It has been pointed out by various authors that the catalytic activity of transition metals depends not only on their disper- sion state but also on the nature of their support.So far, several studies have been carried out on the preparation of catalysts of finely dispersed transition-metal particles on graphite using graphite intercalation compounds (GICs) as starting materials. Principally, two reactions have been used to disperse fine metallic particles on graphite: reduction of metal chlorides by a binary K-GIC, KC8, in an organic solvent' -3 and reduction of transition-metal chloride-GICs by potassium vapo~r.~.~ In the former, the metallic particles are formed on the surface of graphite, but are easily oxidized during handling in The catalysts thus prepared, there- fore, have two disadvantages: they cannot be handled in air and their support has a small specific surface area.The latter reaction results in very stable products in which the fine metallic particles formed are mostly included in the graphite matrix; however, they do not show high activity as catalysts. In the present work, the reducing reaction of iron(II1) chloride in the graphite gallery with a potassium-naphthalene complex in THF at room temperature was followed kinetically to prepare finely dispersed metal particles. The metal particles formed were mostly included between the graphite layers and, as a conse- quence, were inert to air. However, these metal particles could be activated by exfoliation of the graphite support. Experimental The pure FeCl,-GICs used as starting materials were synthe- sized by heating mixtures of anhydrous iron(@ chloride and natural graphite powder with an average particle size of 400 pm in a sealed tube; the stage (1) compound was prepared at 300 "C with a graphite: FeCl, ratio of 4 and the stage (2) at 500 "C with a graphite: FeCl, ratio of 10.The GICs were then separated from the unreacted FeCl, by washing with water. A large particle size for the host graphite flakes was chosen to make it easier to follow the kinetics of the GIC reducing reaction. The ideal chemical formulae for stage (1) and (2) FeC1,-GICs have been shown to be C6FeCl, and CI2FeCl,, respectively. However, the starting GICs used in the present work have a lower content of FeCl,, as shown by the presence of free graphite in the XRD patterns (Fig. 1 after 7hI 12 after 13h I 1 I 1 I I 10 20 30 40 50 60 261' Fig.1 Evolution of X-ray powder pattern from stage (1) FeC1,-GIC with reaction time in the THF solution of K-naphthalene radicals and 2); the free graphite is formed by decomposition during washing. After washing, the GICs were very stable, with no change in the XRD patterns over long periods of time, as reported previously. The aromatic anion radical, potassium-naphthalene, was prepared under an argon atmosphere in anhydrous THF, the FeC13-GIC, a. stage 2 o 0. hl al0 cl c h after 1 h I I I 1 I I 10 20 30 40 50 60 201' Fig. 2 Evolution of X-ray powder pattern from stage (2) FeC1,-GIC with reaction time in the THF solution of K-naphthalene radicals concentration of K being roughly 0.5 rnoldm-,.THF was purified in advance by refluxing under an argon atmosphere with a sodium wire for 5 h and then distilled. The formation of the aromatic anion radical was instantaneous and so the solution turned green immediately after adding potassium to the THF solution of naphthalene. The FeC1,-GIC was then added to the thus prepared K-napthalene-THF solution. A large excess of K in THF solution was used to reduce intercalated FeCl, completely. The suspension of GIC par- ticles was stirred at room temperature under an argon atmos- phere. During the reaction, a small amount of the solution was taken out by pipetting; the solid products were filtered off and then washed several times with acetone and finally with a water-ethanol mixture in order to remove adsorbed potassium and naphthalene.The reaction products thus sampled were analysed by XRD and EDX. They were heated up to different temperatures in the range 300-1000 "C under a flow of pure nitrogen gas and examined by scanning electron microscopy (SEM). In order to avoid any species being adsorbed on the surface of the graphite particles, the flakes of the products were sometimes cleaved by using an adhesive tape just before EDX analysis. Results Reduction of FeC1, in the Graphite Gallery The XRD diagram in Fig. 1 and 2 correspond to the reaction products obtained from the stage (1) and (2) FeC1,-GICs, respectively, after different reaction times. In both cases we can observe progressive decreases in diffraction intensities of 001 lines of the starting GICs.The reaction seems to proceed faster from the stage (2) GIC than from the stage (1). Note that the decrease in diffraction intensities for the starting intercalation compounds is accompanied by the appearance and growth of new diffraction lines which can be J. MATER. CHEM., 1991, VOL. 1 easily attributed to the reaction products, Fe metal and KC1. The diffraction lines due to Fe metal were weak because of the extremely small size of the particles, as will be discussed later. In any case, no lines attributable to Fe-GICs, as stated by Volpin et aL9 were identified. After complete destruction of the starting intercalation structure [13 h for the stage (I) and 6 h for the stage (2)], the XRD patterns showed only trace amounts of graphite, suggesting extensive destruction of the graphite structure during the reducing reaction.These changes in the XRD patterns, also imply that the starting- stage structure persists until complete decomposition, and any observable stage transformation is not apparent. Both Fe metal and KCl thus formed were very stable; no oxidation of Fe and no decrease in the relative amount of KCl were observed during washing. This suggests that neither could be leached away by water; they are probably included in the graphite matrix despite its extensive destruction as mentioned above. It was impossible to remove KCl even by washing with concentrated HCl solution (6 mol dm-3).The particle sizes of KCl and Fe in the final reduction products, calculated from the line broadening of XRD, were ca. 16 and 9 nm, respectively. Fig. 3 and 4 show representative results for the distribution of K, Fe and C1 in the GIC flakes, measured by EDX. After a short reaction time (e.g. 10 min), the distribution profiles of Fe and C1 are almost the same, consistent with XRD results, but, surprisingly K is detected even at the centre of the flakes of the stage (1) GIC as well as those of the stage (2)[Figs 3(a) and 4(a), respectively]. For the final products obtained after a long reaction time of 20 h, the EDX analysis Fig. 3 The distribution of (b) K, (c) C1 and (d) Fe along the line indicated by the arrows on the flakes (a) obtained from stage (1) FeC1,-GIC: Left-hand side, after reaction for 10 min; right-hand side, after reaction for 20 h J.MATER. CHEM., 1991, VOL. 1 Fig. 4 The distribution of (b) K, (c) C1 and (d) Fe along the line indicated by the arrows on the flakes (a) obtained from stage (2) FeC1,-GIC: Left-hand side, after reaction for 10 min; right-hand side, after reaction for 20 h indicates that the flakes still contain Fe, C1 and K, the amount of K increasing [Fig. 3(b) and qb)]. The chemical compo- sitions of the final product obtained from the stage (1) GIC was measured as C19.3FeK2.6C12.5, which is consistent with the XRD pattern showing the presence of KCl and Fe. The distribution profiles of K in the flakes seem to be very similar to those of C1 and Fe, which suggests the existence of three elements as KCl and Fe.For intermediate products, the coexistence of three elements with similar distribution profiles was also observed in the same flakes. These EDX observations reveal very rapid diffusion of THF-solvated potassium towards the centre of the flakes. However, XRD shows that it did not immediately destroy the structure of the starting GIC. The diffraction lines of the starting GIC became broader and weaker with reaction time (Fig. 1 and 2), which indicates a decrease in the amount of GIC and also the increase of stacking disorder in GIC structure. This experimental result suggests that the reaction of FeCl, with THF-solvated potassium in the graphite gallery is not as fast as that of free FeCl, with K in THF.From the SE micrographs presented in Fig. 5(a)and (b),the reaction in THF is found to provoke a pronounced decrease in the size of the flakes. This seems to be consistent with the assumption that the intercalated FeCl, is reduced with the THF-solvated K in the graphite gallery, which results in the destruction of the layer stacking in the graphite structure. Morphology Change of the Reaction Product When the final products were heated up to high temperatures under nitrogen, appreciable exfoliation of the flakes occurred Fig. 5 SE micrographs of the flakes: (a) the starting stage (1) FeC1,-GIC; (b) after complete reduction for 20 h (the final product); (c) after heating the final product at 300 "C for 1 h; (6)after heating at 400 "C for 1 h; (e) after heating at 500 "C for 1 h; and (f)-(h) after heating at 1000 "C for 1 min [Fig.S(c)-(f)]; this was more remarkable the higher the tempera- ture. At temperatures higher than 300 "C, minute particles appear on the surface of the exfoliated flakes. When the products were heated at 1000 "C for 1 min, one could see clearly cubic particles of different size and minute particles with undefined shapes [Fig. 5(f)-(h)].The cubic particles were characterized by EDX to be KCl. The reduced product obtained from the stage (2) FeC1,-GIC showed the same exfoliation behaviour. In addition to the exfoliation of the graphite matrix and the appearance of the particles, the restoration of the graphite structure by heating was observed on XRD patterns.The potassium chloride in the product exfoliated at 500 "C could be removed easily by washing with a mixture of water and ethanol, but at the same time Fe metal was quickly oxidized to a-Fe203, which were clearly detected by XRD. Discussion For short reaction times (10 min), XRD patterns of the products show that the starting structure is retained, but EDX analysis indicates the presence of potassium even at the centre of the flakes. Potassium must, therefore, be in the same gallery as FeCl,. The reaction between intercalated FeC1, and potass- ium is reasonably supposed not to occur at this stage, because the structure is retained and it is known that KC1 is not intercalated into the graphite gallery." At this reaction stage, potassium is expected to be solvated by THF molecules in the gallery." Therefore, the present work reveals that the diffusion of THF-solvated potassium ions, which have the relatively large thickness of 0.55 nm, is very fast even when the graphite galleries are occupied by FeCl, in advance.It can be assumed that the diffusion of THF-solvated potassium ions is much easier for an unoccupied gallery than for an occupied. If so, a kind of bi-intercalation compound must be obtained from the stage (2) FeC1,-GICs. However, this was not the case under our experimental conditions because the stage (2)structure was retained even though large amounts of potassium were detected by EDX. So, the THF- solvated potassium ions seem to diffuse preferentially into the occupied gallery.After complete reduction of FeC1,-GIC, the reaction prod- ucts, Fe and KCl, were detected, but only a trace of graphite. This is most likely due to the disturbance of the parallel stacking of the graphite layers by the inclusion of the reaction products Fe and KC1. After exfoliation at high temperatures, the 002 line of graphite reappeared, owing to the partial recovery of the graphite structure by expulsion of the included Fe and KC1 particles. On the way to complete reduction of FeCl, intercalated into graphite, the starting stage structure was preserved, though the size decreased. The reaction between FeC1, and THF-solvated potassium in the gallery seems to be the rate- determining step in the whole reducing reaction because the diffusion of the solvated K ions into the graphite gallery is extremely fast as discussed above.This is unexpected because the reducing reaction between free FeC1, and K in solution is instantaneous. In the present case, therefore, the limited space within the graphite gallery must be a strong constraint for the reaction. The following experimental results also suggest the strong effect of limited space: (1) there was not much change in the distribution of Fe and C1, the latter bonding with Fe in the beginning of the reaction and with K at the end of the reaction; (2)the fragmentation of the graphite layers occurred during the reaction; (3) a certain number of THF molecules were probably trapped in the graphite gallery, which caused exfoliation of the graphite matrix on heating.A similar exfoliation phenomenon has been observed on ternary GICs with K and THF.12 J. MATER. CHEM., 1991, VOL. 1 It is supposed that the reduction product of metallic Fe particles is confined within the graphite gallery together with KC1 and possibly THF molecules. It was reasonable to expect the metal particles to be inactive because they were enclosed within the graphite matrix; indeed the reaction products could be handled in the air. These metallic particles could be activated by heating to exfoliate the graphite matrix. After exfoliation at 300 "C the size of Fe metal particles was so small that it was difficult to identify them under SEM, and they were so active that they were immediately oxidized by washing with water.The present products prepared from the reduction of metal chloride intercalated into graphite by THF- solvated potassium seem to give us the following advantages over other similar catalysts: (i) they can be handled in air and (ii) they can be converted to finely dispersed metal particles supported on graphite with high surface area by simple heating just before their use. High activity can be expected because of the small size of the metallic particles and the high surface area of the graphite support after exfoliation. This work was partly supported by a grant for the Inter- national Joint Research Project from the NEDO, Japan.References 1 P. Braga, A. Ripamonti, D. Savoia, C. Trombini and A. Umani- Ronchi, J. Chem. SOC., Chem. Commun., 1981,40. 2 M. Inagaki, Y. Shiwachi and Y. Maeda, J. Chim. Phys., 1984, 81, 847. 3 A Messaoudi, R. Erre and F. Beguin, Carbon, 1991, 29, 515. 4 G. Bewer, W. Wichmann and H. P. Boehm, Mater. Sci. Eng., 1977, 31, 73. 5 R. Erre, A. Messaoudi and F. Beguin, Synth. Met., 1988, 23,493. 6 P. Kaiser, A. Messaoudi, D. Bonnin, R. Erre and F. Beguin, J. Chim. Phys., 1989,86, 1787. 7 A. Furstner and H. Weidmenn, J. Chem. SOC., Dalton Trans., 1988,2023. 8 Z. D. Wang and M. Inagaki, Synth. Met., 1988, 26, 181. 9 M. E. Volpin, N. Novikov, N. D. Lapkina, V. I. Kasatochkin, Y. T. Struchov, M. E. Kazakov, R. A. Stukan, V. A. Povitskij, S. Karimov and A. V. Zvarikina, J. Am. Chem. SOC., 1975, 97, 3366. 10 M. Inagaki and Z. D. Wang, Synth. Met., 1987, 29, 1. 11 M. Nomine and L. Bonnetain, C. R. Acad. Sci., 1967, 264. 12 M. Inagaki, K. Muramatsu and Y. Maeda, Synth. Met., 1983,8, 335. Paper 1/00333J; Received 23rd January, 1991