- Email: [email protected]
Intercalation compounds: covalent and ionic approach
Materials Science and Engineering, B3 (1989) 253-255
253
Intercalation Compounds: Covalent and Ionic Approach PAUL HAGENMULLER Laboratoire de Chimie du Solide du CNRS, Universit~ de Bordeaux I, 351 cours de la Libdration, F-33405 Talence C~dex (France)
(Received December 21, 1988)
Abstract
In contrast to amphoteric graphite, layer-type oxides and chalcogenides generally play the role of acceptors in intercalation reactions. Owing to the more ionic character of the M - O bonds, the structural evolution of the oxides may usually be explained on the basis of electrostatic considerations or in terms of cation oxido-reduction. In contrast, for the more covalent chalcogenides, occupancy of energy levels in the band structure by the transferred electrons constitutes a determining factor which strongly influences not only the structural changes but also the physical properties. Similar bonding considerations account for the strong tendency of the oxides to undergo 2D--" 3D transformations. Owing to its monolayer structure and to the existence of weakly bound rt-electrons, graphite has a number of specific properties: ability to play the role of either an acceptor or a donor; formation of a large variety of multistage intercalation compounds, following the cluster rule of Daumas and Herold [1], with strong repulsions between the carbon-layers (see e.g. ref. 2). Sandwich-type oxides and chalcogenides clearly differ from graphite intercalation compounds: as a rule the sheets have an acceptor character, i.e. they are Lewis acids; mostly they lead to stage 1 insertion in spite of the increasing negative charge of the sheets during the chemical or electrochemical intercalation process. In fact, the ionicity difference between the M - O bonds within the oxide sheets and the more covalent bonds in homologous 2D chalcogenides leads to behaviours which allow us to distinguish them from the structural point of view as well as by the resulting physical properties. This is particularly true when the intercalated species is an alkali cation, or even a 1B cation such as Cu ÷ or Ag + . 0921-5107/89/$3.50
It has been shown (e.g. by nuclear magnetic resonance Knight shift determinations) that in sheet compounds, as the result of an ionization process, these atoms occupy the van der Waals gap as cations, with only a very small local presence of external s-electrons [3]. As a consequence we have to distinguish between the influence in the intercalation compounds of the A + cations and that of the liberated s-electrons which occupy upper levels in the energy diagram of the sheets using the rigid band model. The A + cations essentially play a double role: they accommodate mechanical strain energy in pushing away the sheets from each other; they stabilize the materials via coulombic attraction with neighbouring sheets. The stronger influence of the first factor explains why, during the intercalation reaction, the parameter c usually increases, especially at the beginning of the process. As a rule, A c increases for homologous compounds with increasing size of the A + cation and decreasing value of the initial van der Waals gap [4]. The role of the inserted cations is more important for oxides since their anionic charge is higher than in the more covalent chalcogenides. This feature accounts for the fact that most AxMO 2 compounds only exist when x>0.5, M o O 3 being an exception owing to the strong covalency of the M o - O bonds [5]. Delmas et al. [6] have shown that the structural changes observed during electrochemical intercalation-deintercalation reactions can be explained in a rather simple way: no structural modification occurring within the sheets (the temperature is too low to allow breaking of the internal M - O bonds), but the sheets gliding with respect to. each other. This gliding process results from the modification of the electrostatic interactions between oxygen layers facing each other in the van der Waals gap. Two phase families related by simple sheet gliding and respecting each ABC circular © Elsevier Sequoia/Printed in The Netherlands
254 permutation of the oxygen packing have been shown to result either from an intercalation reaction or from a low temperature ion exchange process [7-9]. Such electrostatic phenomena clearly account for the fact that in an AxMO 2 o r A x M S 2 series, increasing x values favour the passage from a trigonal prismatic environment of A + to a more stable octahedral one, leading to strong weakening of the A + ion conductivity [10]. A significant example of this is provided by the investigations of Fouassier et al. on the NaxCoO2 series [11]. In a similar way Le Blanc et al. have shown that in KTiS 2 potassium has a trigonal prismatic environment owing to strong covalency of the TiS bonds, in contrast to the octahedral environment in KZrS2 where weakening of the bonds within the sheets increases the formal charge of sulphur [12]. Similar considerations explain the trigonal prismatic coordination of the large A + cations (rubidium, caesium) as opposed to the octahedral coordination of lithium in homologous series ]6, 13]. A tetrahedral environment occurs only for Li ÷ and Cu + cations, leading to strong covalent bonding. In delafossite-type phases the Cu* or Ag + ions are situated along the edges of trigonal oxygen prisms, sp hybridization leading to linear coordination (e.g. in CuA102). This particular structure is even observed for Pd + and Pt + in PdCoO 2 and PtCoO 2 respectively, probably as a result of d~s hybridization [14]. The influence of the outer electrons injected in the host lattice differs with the materials concerned. When they occupy discrete levels the upper intercalation limit depends not only on the available sites (as in the MOCI layer compounds) but also on the lowest oxidation state which is stable in the conditions of the reaction [15]. The intermediate compounds are characterized by a classical hopping mechanism. As an example, Nadiri and Delmas have shown that only two lithium or sodium atoms can be intercalated in the cage structure of Fe2(SO4) 3 or NaTi2(PO4) 3 where four empty sites are available [26]. The presence of alkali cations gives rise to unit cell shrinking whereas the reduced oxidation state of titanium or iron favours dilatation, but a careful examination of the precise variation of the cell parameters can bring evidence of the type of site occupied by the alkali cations in the host matrix [16]. In the Chevrel phases built up by Mo6 octahedra enclosed in chalcogen cubes only some of
the six available sites are occupied, since the Mo(, clusters provide 20 electrons for a maximum population of 24 in the discrete molecular levels. The intercalation is thus limited to two divalent or four monovalent cations [ 18]. Rouxel and coworkers have shown that for the chalcogenides, electronic occupation of the conduction band largely accounts for the reactivity of the MX 2 compounds as well as for the structure obtained and the corresponding physical properties, in particular the electrical properties [4, 19]. For instance, the existence of a large t2g-type conduction band explains the easy intercalation up to x = 1 of lithium in TiS2, the obtained LixTiS 2 compounds displaying a metallic behaviour. Easy intercalation of lithium in the p-variety of TaS 2 (with a trigonal prismatic coordination of tantalum) appears to result from the progressive filling of a low lying d~2 band, which compensates for a stronger repulsion between sulphur layers belonging to the same sheets. The material, as expected, is metallic, except when x-- 1 leads to full band occupancy. Another significant example is intercalation of ]lthium in MoS2, which has the structure of p-TaS2 but an already filled d~-, band: intercalation of lithium becomes difficult in comparison with TaS2 owing to occupancy by the extra electrons of an antibonding high energy band. MoS2, which is a semiconductor, becomes metallic. An obvious consequence of the bonding difference on the intercalation processes in oxides and chalcogenides is the 3D character of the oxides, which contrasts with the tendency of the intercalants to order in the van der Waals gap of the sulphides (as a consequence of strong cationcation repulsion [25]). This feature is clearly illustrated by the ease with which defect NaCltype lithium-intercalated layer oxides lead to spinels by taking advantage of the existence of a similar f.c.c, oxygen packing [20, 21]. Stabilization of Li + in tetrahedral sites is illustrated by lithiation of rutile-type fl-MnO2, which does not lead to an NiAs-type phase with the same h.c.p, oxygen packing but, after sheet gliding, to a spinel phase of formulation close to LiMn20 ~ [22]. A negative aspect of this tendency to 2D ~ 3D transformation is easy migration of the transition element into the van der Waals gap of the layer structure oxide, which indeed shrinks the parameter c and makes further A ÷ intercalation more difficult [23, 24]. A similar migration occurs in sulphides only for very high overvoltages.
255
References 1 N. Daumas and A. Herold, C.R. Acad. Sci. Paris, 268 (1971)373. 2 M.S. Dresselhaus, Phys. Today, 37 (1984) 60. 3 B.G. Silbernagei, Solid State Commun., 17(1975) 361. 4 J. Rouxel, Chem. Scr., 28 (1988) 33. 5 C. Fouassier, G. Matejka, J. M. Reau and P. Hagenmuller, J. Solid State Chem., 6 (1973) 532. 6 C. Delmas, C. Fouassier and P. Hagenmuller, Mater. Res. Bull., 11 (1976) 1483. 7 C. Delmas, J. J. Braconnier, C. Fouassier and P. Hagenmuller, Mater. Res. Bull., 3-4 (1981) 165. 8 J. Rouxel, M. Danot and J. Pichon, Bull. Soc. Chim. Fr. (1971)3390. 9 J. Rouxel, in F. A. Levy (ed.), Intercalated Layered Materials, Reidel, Dordrecht, 1979. 10 A. Maazaz, C. Delmas, C. Fouassier, J. M. Reau and P. Hagenmuller, Mater. Res. Bull., 14 (1979) 329. 11 C. Fouassier, C. Delmas and P. Hagenmuller, Mater. Res. Bull., 10 (1975) 443. 12 A. Le Blanc, M. Danot, L. Trichet and J. Rouxel, Mater. Res. Bull., 9(1974) 191. 13 J. Rouxel, J. SolidState Chem., 17(1976) 223. 14 J. P. Doumerc, A. Ammar, A. Wichainchai, M. Pouchard
and P. Hagenmuller, J. Phys. Chem. Solids, 48 (1987) 37. 15 G.A. Fatseas, E Palvadeau and J. E Venien, J. Solid State Chem., 51 (1984) 17. 16 A. Nadiri, Th~se de Doctorat ~s Sciences Physiques, Bordeaux I University, 1985. 17 G. Le Flem, private communication. 18 R. Sch611horn, Angew. Chem. English Edn., 19 (1980) 983. 19 W. Y. Liang, Physics and Chemistry of Electrons and Ions in Condensed Matter, NATO ASI, Reidel, Dordrecht, 1984, p. 130. 20 M. G. Thomas, W. I. David, J. B. Goodenough and E Groves, Mater. Res. Bull., 20 (1985) 1137. 21 L. A. De Picciotto and M. M. Thackeray, Mater. Res. Bull., 20 (1985) 187; Solid State lonics, 18-19 (1986) 773. 22 W. I. Bruce, M. M. Thackeray, E G. Bruce and J. B. Goodenough, Mater. Res. Bull., 19(1984) 99. 23 A. Maazaz, C. Deimas and E Hagenmuller, J. Incl. Phenom., 1 (1983) 45. 24 A. Maazaz and C. Delmas, C.R. Acad. Sci. Paris, 295 (1982) 759. 25 A.H. Thompson, Phys. Rev. Lett., 40 (1978) 1511. 26 A. Nadiri and C. Delmas, C.R. Acad. Sci. Paris, 304 (1987) 415.
253
Intercalation Compounds: Covalent and Ionic Approach PAUL HAGENMULLER Laboratoire de Chimie du Solide du CNRS, Universit~ de Bordeaux I, 351 cours de la Libdration, F-33405 Talence C~dex (France)
(Received December 21, 1988)
Abstract
In contrast to amphoteric graphite, layer-type oxides and chalcogenides generally play the role of acceptors in intercalation reactions. Owing to the more ionic character of the M - O bonds, the structural evolution of the oxides may usually be explained on the basis of electrostatic considerations or in terms of cation oxido-reduction. In contrast, for the more covalent chalcogenides, occupancy of energy levels in the band structure by the transferred electrons constitutes a determining factor which strongly influences not only the structural changes but also the physical properties. Similar bonding considerations account for the strong tendency of the oxides to undergo 2D--" 3D transformations. Owing to its monolayer structure and to the existence of weakly bound rt-electrons, graphite has a number of specific properties: ability to play the role of either an acceptor or a donor; formation of a large variety of multistage intercalation compounds, following the cluster rule of Daumas and Herold [1], with strong repulsions between the carbon-layers (see e.g. ref. 2). Sandwich-type oxides and chalcogenides clearly differ from graphite intercalation compounds: as a rule the sheets have an acceptor character, i.e. they are Lewis acids; mostly they lead to stage 1 insertion in spite of the increasing negative charge of the sheets during the chemical or electrochemical intercalation process. In fact, the ionicity difference between the M - O bonds within the oxide sheets and the more covalent bonds in homologous 2D chalcogenides leads to behaviours which allow us to distinguish them from the structural point of view as well as by the resulting physical properties. This is particularly true when the intercalated species is an alkali cation, or even a 1B cation such as Cu ÷ or Ag + . 0921-5107/89/$3.50
It has been shown (e.g. by nuclear magnetic resonance Knight shift determinations) that in sheet compounds, as the result of an ionization process, these atoms occupy the van der Waals gap as cations, with only a very small local presence of external s-electrons [3]. As a consequence we have to distinguish between the influence in the intercalation compounds of the A + cations and that of the liberated s-electrons which occupy upper levels in the energy diagram of the sheets using the rigid band model. The A + cations essentially play a double role: they accommodate mechanical strain energy in pushing away the sheets from each other; they stabilize the materials via coulombic attraction with neighbouring sheets. The stronger influence of the first factor explains why, during the intercalation reaction, the parameter c usually increases, especially at the beginning of the process. As a rule, A c increases for homologous compounds with increasing size of the A + cation and decreasing value of the initial van der Waals gap [4]. The role of the inserted cations is more important for oxides since their anionic charge is higher than in the more covalent chalcogenides. This feature accounts for the fact that most AxMO 2 compounds only exist when x>0.5, M o O 3 being an exception owing to the strong covalency of the M o - O bonds [5]. Delmas et al. [6] have shown that the structural changes observed during electrochemical intercalation-deintercalation reactions can be explained in a rather simple way: no structural modification occurring within the sheets (the temperature is too low to allow breaking of the internal M - O bonds), but the sheets gliding with respect to. each other. This gliding process results from the modification of the electrostatic interactions between oxygen layers facing each other in the van der Waals gap. Two phase families related by simple sheet gliding and respecting each ABC circular © Elsevier Sequoia/Printed in The Netherlands
254 permutation of the oxygen packing have been shown to result either from an intercalation reaction or from a low temperature ion exchange process [7-9]. Such electrostatic phenomena clearly account for the fact that in an AxMO 2 o r A x M S 2 series, increasing x values favour the passage from a trigonal prismatic environment of A + to a more stable octahedral one, leading to strong weakening of the A + ion conductivity [10]. A significant example of this is provided by the investigations of Fouassier et al. on the NaxCoO2 series [11]. In a similar way Le Blanc et al. have shown that in KTiS 2 potassium has a trigonal prismatic environment owing to strong covalency of the TiS bonds, in contrast to the octahedral environment in KZrS2 where weakening of the bonds within the sheets increases the formal charge of sulphur [12]. Similar considerations explain the trigonal prismatic coordination of the large A + cations (rubidium, caesium) as opposed to the octahedral coordination of lithium in homologous series ]6, 13]. A tetrahedral environment occurs only for Li ÷ and Cu + cations, leading to strong covalent bonding. In delafossite-type phases the Cu* or Ag + ions are situated along the edges of trigonal oxygen prisms, sp hybridization leading to linear coordination (e.g. in CuA102). This particular structure is even observed for Pd + and Pt + in PdCoO 2 and PtCoO 2 respectively, probably as a result of d~s hybridization [14]. The influence of the outer electrons injected in the host lattice differs with the materials concerned. When they occupy discrete levels the upper intercalation limit depends not only on the available sites (as in the MOCI layer compounds) but also on the lowest oxidation state which is stable in the conditions of the reaction [15]. The intermediate compounds are characterized by a classical hopping mechanism. As an example, Nadiri and Delmas have shown that only two lithium or sodium atoms can be intercalated in the cage structure of Fe2(SO4) 3 or NaTi2(PO4) 3 where four empty sites are available [26]. The presence of alkali cations gives rise to unit cell shrinking whereas the reduced oxidation state of titanium or iron favours dilatation, but a careful examination of the precise variation of the cell parameters can bring evidence of the type of site occupied by the alkali cations in the host matrix [16]. In the Chevrel phases built up by Mo6 octahedra enclosed in chalcogen cubes only some of
the six available sites are occupied, since the Mo(, clusters provide 20 electrons for a maximum population of 24 in the discrete molecular levels. The intercalation is thus limited to two divalent or four monovalent cations [ 18]. Rouxel and coworkers have shown that for the chalcogenides, electronic occupation of the conduction band largely accounts for the reactivity of the MX 2 compounds as well as for the structure obtained and the corresponding physical properties, in particular the electrical properties [4, 19]. For instance, the existence of a large t2g-type conduction band explains the easy intercalation up to x = 1 of lithium in TiS2, the obtained LixTiS 2 compounds displaying a metallic behaviour. Easy intercalation of lithium in the p-variety of TaS 2 (with a trigonal prismatic coordination of tantalum) appears to result from the progressive filling of a low lying d~2 band, which compensates for a stronger repulsion between sulphur layers belonging to the same sheets. The material, as expected, is metallic, except when x-- 1 leads to full band occupancy. Another significant example is intercalation of ]lthium in MoS2, which has the structure of p-TaS2 but an already filled d~-, band: intercalation of lithium becomes difficult in comparison with TaS2 owing to occupancy by the extra electrons of an antibonding high energy band. MoS2, which is a semiconductor, becomes metallic. An obvious consequence of the bonding difference on the intercalation processes in oxides and chalcogenides is the 3D character of the oxides, which contrasts with the tendency of the intercalants to order in the van der Waals gap of the sulphides (as a consequence of strong cationcation repulsion [25]). This feature is clearly illustrated by the ease with which defect NaCltype lithium-intercalated layer oxides lead to spinels by taking advantage of the existence of a similar f.c.c, oxygen packing [20, 21]. Stabilization of Li + in tetrahedral sites is illustrated by lithiation of rutile-type fl-MnO2, which does not lead to an NiAs-type phase with the same h.c.p, oxygen packing but, after sheet gliding, to a spinel phase of formulation close to LiMn20 ~ [22]. A negative aspect of this tendency to 2D ~ 3D transformation is easy migration of the transition element into the van der Waals gap of the layer structure oxide, which indeed shrinks the parameter c and makes further A ÷ intercalation more difficult [23, 24]. A similar migration occurs in sulphides only for very high overvoltages.
255
References 1 N. Daumas and A. Herold, C.R. Acad. Sci. Paris, 268 (1971)373. 2 M.S. Dresselhaus, Phys. Today, 37 (1984) 60. 3 B.G. Silbernagei, Solid State Commun., 17(1975) 361. 4 J. Rouxel, Chem. Scr., 28 (1988) 33. 5 C. Fouassier, G. Matejka, J. M. Reau and P. Hagenmuller, J. Solid State Chem., 6 (1973) 532. 6 C. Delmas, C. Fouassier and P. Hagenmuller, Mater. Res. Bull., 11 (1976) 1483. 7 C. Delmas, J. J. Braconnier, C. Fouassier and P. Hagenmuller, Mater. Res. Bull., 3-4 (1981) 165. 8 J. Rouxel, M. Danot and J. Pichon, Bull. Soc. Chim. Fr. (1971)3390. 9 J. Rouxel, in F. A. Levy (ed.), Intercalated Layered Materials, Reidel, Dordrecht, 1979. 10 A. Maazaz, C. Delmas, C. Fouassier, J. M. Reau and P. Hagenmuller, Mater. Res. Bull., 14 (1979) 329. 11 C. Fouassier, C. Delmas and P. Hagenmuller, Mater. Res. Bull., 10 (1975) 443. 12 A. Le Blanc, M. Danot, L. Trichet and J. Rouxel, Mater. Res. Bull., 9(1974) 191. 13 J. Rouxel, J. SolidState Chem., 17(1976) 223. 14 J. P. Doumerc, A. Ammar, A. Wichainchai, M. Pouchard
and P. Hagenmuller, J. Phys. Chem. Solids, 48 (1987) 37. 15 G.A. Fatseas, E Palvadeau and J. E Venien, J. Solid State Chem., 51 (1984) 17. 16 A. Nadiri, Th~se de Doctorat ~s Sciences Physiques, Bordeaux I University, 1985. 17 G. Le Flem, private communication. 18 R. Sch611horn, Angew. Chem. English Edn., 19 (1980) 983. 19 W. Y. Liang, Physics and Chemistry of Electrons and Ions in Condensed Matter, NATO ASI, Reidel, Dordrecht, 1984, p. 130. 20 M. G. Thomas, W. I. David, J. B. Goodenough and E Groves, Mater. Res. Bull., 20 (1985) 1137. 21 L. A. De Picciotto and M. M. Thackeray, Mater. Res. Bull., 20 (1985) 187; Solid State lonics, 18-19 (1986) 773. 22 W. I. Bruce, M. M. Thackeray, E G. Bruce and J. B. Goodenough, Mater. Res. Bull., 19(1984) 99. 23 A. Maazaz, C. Deimas and E Hagenmuller, J. Incl. Phenom., 1 (1983) 45. 24 A. Maazaz and C. Delmas, C.R. Acad. Sci. Paris, 295 (1982) 759. 25 A.H. Thompson, Phys. Rev. Lett., 40 (1978) 1511. 26 A. Nadiri and C. Delmas, C.R. Acad. Sci. Paris, 304 (1987) 415.