摘要:

DISCUSSIONS OF THE FARADAY SOCIETY No. 26, 1958 IONS OF THE TRANSITION ELEMENTS THE FARADAY SOCIETY Agents f o r the Society’s Publications : The Aberdeen University Press Ltd. 6 Upper Kirkgate, Aberdeen Scotland@ The Faraday Society and Contributors, 1959 PUBLISHED . . . 1959 PRINTED IN OREAT BRITAIN AT THE UNIVERSITY PRESS ABERDEENA GENERAL DISCUSSION ON IONS OF THE TRANSITION ELEMENTS 9th-11th September, 1958 A GENERAL DISCUSSION on Ions of the Transition Elements was held in the Depart- ment of Chemistry, University College, National University of Ireland, Dublin, at the invitation of Trinity College (University of Dublin) and University College (National University of Ireland) from the 9th to the 11th September, 1958. Dr. J. W. Linnett, F.R.S., took the chair at the first scientific session and welcomed about 150 members and visitors. Prof.M. H. L. Pryce (Bristol University), Prof. Linus Pauling (California Inst. of Technology) and Prof. J. H. van Vleck (Harvard University) shared the duties of Chairman at the remaining three sessions.A GENERAL DISCUSSION ON IONS OF THE TRANSITION ELEMENTS 9th-11th September, 1958 A GENERAL DISCUSSION on Ions of the Transition Elements was held in the Depart- ment of Chemistry, University College, National University of Ireland, Dublin, at the invitation of Trinity College (University of Dublin) and University College (National University of Ireland) from the 9th to the 11th September, 1958. Dr. J. W. Linnett, F.R.S., took the chair at the first scientific session and welcomed about 150 members and visitors.Prof. M. H. L. Pryce (Bristol University), Prof. Linus Pauling (California Inst. of Technology) and Prof. J. H. van Vleck (Harvard University) shared the duties of Chairman at the remaining three sessions.CONTENTS PAGE GENERAL INTRODUCTION. By J. W. Linnett . . 7 I. OPTICAL AND MAGNETIC PROPERTIES- Introductory Paper-Physical Properties of Transition Ions. By M. H. L. Pryce . . 21 Optical Absorption in the Divalent Oxides of Cobalt and Nickel. By W. R. Doyle and G. A. Lonergan . . 27 The Absorption Spectrum of Vanadium Corundum. By M. H. L. Pryce and W. A. Runciman . . 34 The Line Spectra of Cr3+ Ion in Crystals. By S. Sugano and Y . Tanabe 43 The Absorption Spectra of Solid Hydrated Nickel Sulphate. By H. Hartmann and H.Muller . . 49 Experiments on Charge Transfer and Exchange Interactions. By J. Owen . 53 The Bonding and Valence Properties of Iron Group Impurities in NaF. By W. Hayes . . 58 Electron Transfers Among Transition Elements in Magnesium Oxide. By J. E. Wertz, P. Auzins, J. H. E. Griffiths and J. W. Orton . 66 Magnetic Resonance of Different Ferric Complexes. By J. F. Gibson, D. J. E. Ingram and D. Schonland . . 72 The Electronic Structures of Some First Transition Series Metal Por- phyrins and Phthalocyanines. By J. S. Griffith . . 81 GENERAL DIscussIoN.-Dr. J. S. Griffith, Dr. C. K. J~zrrgensen, Dr. R. Englman, Dr. R. J. P. Williams, Mr. R. P. Bell, Dr. J. Owen, Dr. W. Hayes, Dr. L. E. Orgel, Dr. D. J. E. Ingram, Dr. D. Schonland . . 87 11. ENERGETICS OF COMPLEXES- The Magnetic Behaviour of Regular and Inverted Crystalline Energy Levels.By J. H. van Vleck . . 96 The Magnetic Properties of Some de3, ds4, and de5 Configurations. By B. N. Figgis, J. Lewis, R. S. Nyholm and R. D. Peacock . . 103 The Interelectronic Repulsion and Partly Covalent Bonding in 5 Transition-Group Complexes. By Chr. K. J~zrrgensen. . . 1106 CONTENTS PAGE Orbital Modification in Metal Complexes. By D. P. Craig and E. A. Magnusson . . 116 The Influence of n-Bonding in some Transition Metal Complexes. By R. J. P. Williams . . 123 The Transmission of Electronic Effects through a Heavy Metal Atom. By J. Chatt, L. A. Duncanson, B. L. Shaws and L. M. Venanzi . 131 Ferroelectricity and the Structure of Transition-Metal Oxides. By L. E. Orgel . . 138 Crystal-Field Stabilization and Site Deformation in Crystals and Complexes Containing Transition Ions. By N. S. Hush . . 145 The Thermodynamics of the Formation of Complex Ions of Ethylene- diaminetetra-Acetic Acid. By L. A. K. Staveley and T. Randall . 157 Relations between the Polarographic Half-Wave Potentials and Optical GENERAL DIscussIoN.-Dr. C. K. Jerrgensen, Dr. J. Owen, Dr. J. S. Griffith, Dr. B. Figgis, Dr. R. J. P. Williams, Dr. L. E. Orgel, Dr. F. J. C. Rossotti, Dr. R. Englman, Dr. D. W. A. Sharp, Mr. E. A. Magnusson, Dr. M. G. Brown, Dr. J. Chatt, Dr. J. E. Prue, Mr. L. A. K. Staveley, Dr. A. G. Sharpe . . 172 Properties of Some Inorganic Complexes. By A. A. VlCek . . 164 Author Index . . 192

ISSN:0366-9033    

DOI:10.1039/DF9582600001

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

GENERAL INTRODUCTION BY J. W. Lnwmr Inorganic Chemistry Laboratory, Oxford Received 3rd July, 1958 The general object of this meeting is to consider the properties and electronic structures of the ions of the transition elements especially when they are included in complexes; the particular aim being to bring physicists and chemists together to discuss these problems. This introduction will be divided into two parts. The first will deal with the theoretical treatments of the electronic structures and the second with the various experimental methods that are being used for studying these compounds, often in order to discover more about the electronic structures. SECTION I The three procedures for treating the electronic structures are (i) the Pauling valence bond method; (ii) the molecular orbital method; and (iii) the ligand (or crystal) field method. PAULING’S METHOD Pauling dealt only with the electronic structure of the ground state,l and was primarily interested in the shapes and magnetic moments of complexes.He designated the orbitals of a complex in terms of the orbitals of the central atom. For the first long period the important valency shell orbitals were the five 3 4 one 4s and three 4p orbitals and he supposed that a number of these, equal to the number of ligands, were involved in bond formation. He devised a most suc- cessful means of presenting his ideas ; two examples are shown in fig. 1. Pauling’s FIG. 1 .-Examples of Paulhg-type representations for (a) Fe(CN);- and (6) NiCl, . ZP(C,H,),. main achievement was his systematization of the shapes of complexes in terms of the groups of central atom orbitals involved in accepting electrons from the ligands.For instance, the group d2sp3 led to an octahedral, and sp3 to a tetrahedral arrange- ment (cf. fig. 1). The directional qualities of these groups of orbitals may be presented in terms of either hybrids 2 or configurations of maximum probability.3 The results of the two viewpoints are necessarily the same. Pauling’s approach encountered difficulties when the number of orbitals needed to accommodate all the electrons was too small, either because of the total number of electrons present, or because of the number with parallel spin 78 GENERAL INTRODUCTION as indicated by magnetic measurements. This difficulty was met by supposing either that outer orbitals (e.g.some of the five 4d, or the 5s, in the first long period) had to be considered, or that the structure of the complexes was " ionic ". Ac- cording to the latter the electronic structures of the central ion and ligand groups were, in certain cases, sufficiently separate and independent that the orbitals of the ligands had to be considered as well as the central atom orbitals. This meant that the bonding electrons which, in covalent complexes, occupied orbitals in- volving those of the central ion, occupied, in ionic complexes, orbitals separate from the central ion leaving all nine orbitals of the central ion free to accommodate the other electrons. By Hund's rule these tended to be spread with parallel spin as widely as possible among the degenerate orbitals so that the high magnetic moments of ions such as FeFg- were explained.MOLECULAR ORBITAL METHOD This employs the same orbitals of the central ion as does the Pauling method but, in addition, the N orbitals of the N ligands which are directed towards the central atom are included. For example, if there are six ligands, there are in all fifteen orbitals available for the construction of the molecular orbitals by linear combination, this being the approximation that is used. Supposing that the ligands are octahedrally disposed along the Cartesian axes, there will be three degenerate non-bonding d-orbitals (usually labelled dwy, dyz and dzx having their four lobes directed between the artesian axes),* six bonding orbitals derived from the d2sp3 grouping (cf.Paulhg) and six corresponding anti-bonding orbtials with additional nodes in the wave function between the central atom and the ligands. By analogy with Pauling's method, the molecular orbitals may be repre- sented diagrammatically as in fig. 2. The way in which the molecular orbitals are occupied is shown for some cobaltous complexes in fig. 3. This may be com- pared with a diagram for these same compounds based on the Pauling system pub- ished by Nyholm.4 With Co(H20)26' the Pauling type representation is 3d2.3d2.3d. 3d. 3d. 49.4~2.4~9.4132-4d2.4d2 the last six pairs being the bonding ones. They have been underlined to indicate this. Each of the 4d atomic orbitals has one spherical node f so that, as regards the number and general form of the nodes, they resemble the molecular orbital which may be described as " 3d anti-bonding " for this has a node between the central atom and all the ligands.The Pauling and molecular orbital representa- tions are therefore quite different in character. A Pauling type representation which would be more like the molecular orbital one is 3d2.3d2.3d. 3 d 2 . 3 d 2 . 4 ~ 2 . 4 ~ 2 . 4 ~ 2 . 4 ~ 2 . 4 d . 4d. However, it does seem that, for ions requiring this number of orbitals to accom- modate the electrons, the molecular orbital method is much sounder and more straightforward. The Pauling method has only been applied to complex ions qualitatively. The molecular orbital method is, however, capable of dealing with the system quanti- tatively, and the energy level diagram for an octahedral complex will now be * The five d orbitals will be labelled d,,, dx,, d,,, dz2 and dx2 - ,2 The first three have four lobes which are directed between the Cartesian axes given in the subscript.The last is of the same form with positive lobes in the + x and - x directions and two negative lobes in the + y and - y directions. The fourth has positive lobes in the + z and - z directions and a smaller negative " frill " in the xy plane ; its variation is in fact (3z2 - r2). t Since the number of nodes is one less than the principle quantum number, a 4d orbital has one spherical and two other nodes. For 4d,, the two other nodes are the xz and yz planes.J . W. LINNETT 9 (a) Six octahedral ligands. f l a W T1 1 1 non- bonding - T T I , T L T1 1 1 T1 1'1 + dx, dyz dxy dzz dx=-y= Px Py br (b) Four planar ligands (d)T 1 T i T T T v tl T L T1 T b10 GENERAL INTRODUCTION considered.If n-bonding is ignored, orbitals of only four symmetry classes need be taken into account. These are (a) totally symmetric A1, orbitals (e.g. 4s of the central atom) ; (b) triply degenerate 7'1, orbitals (e.g. 4p) ; (c) doubly degener- ate Eg orbitals (e.g. d,z and d,Z-,,z); (d) triply degenerate T2, orbitals (e.g. d,,, dyz and dzx). The six ligand u-orbitals can be combined to give an Al,, a pair of Eg and a trio of TlU orbitals. These can combine respectively with the 4s, dtz and dxz - p, and 4p orbitals of the central atom. The orbitals in (d) remain un- changed and are non-bonding in this approximation in the complex.Consider, for example, the two component Al, orbitals (central atom and ligand). These will combine according to the simplest molecular orbital procedure to give two new orbitals of energies E given by where El and E2 are the energies of the two component orbitals and /3 is the so- called exchange integral, which is dependent, in part, on the extent they overlap. The lowering in the energy of the orbital of lower energy (which, at this approxima- tion, is equal to the rise in the energy of the orbital of higher energy) is greater the smaller (El - E2) and the larger 18. These quantities are estimated in various more or less empirical ways. The values of El and E2 are usually obtained from a consideration of ionization potentials while those of the various 18s are fixed in a more or less arbitrary but necessarily uncertain manner to reproduce a measured quantity.Because of doubts of the above type, it is not easy to be sure what precise order the various molecular orbitals will take in a given complex. Clearly the Al,, Eg and 7'1, bonding orbitals lie below the T2, non-bonding orbitals, while the anti-bonding orbitals (Alg, Eg and 7'1,) lie above them. Owen5 gives the order of the last three as E,, A1, and 7'1, for an ion in which the ligand is " an ion like Cl- ". On the other hand a calculation for Cr(CN)z- by Gilde and Ban,6 who have used the procedure adopted by W-olfsberg and Helmholtz7 for Mn04, CrOi- and ClO,, gives the order as Al,, TI, and E, (see fig. 4). The reason for the large splitting of the Eg orbitals in this complex is that an electron in the ligand orbitals and one in the 3d orbitals of the central atom have about the same energy and the corresponding exchange integral is not small.It is possible, of course, that the order of levels may differ in different complexes. Since these are the orbitals into which electrons can be excited from the T2, orbital on light absorption they are of interest. The energy of the anti-bonding E, orbital will be higher the greater the overlap of the 3d and the ligand orbitals, and this overlap is likely to be greater the smaller the effective nuclear charge on the ligand atom adjacent to the central atom. This provides an explanation of the basic form of the spectrochemical series (see later). The ionization potentials associated with the particular orbitals will also be important.As Orgel says, molecular orbital theory, in its complete form, includes all interactions between electrons and nuclei. It is therefore the utility of the theory in a reasonably simple form " which is at stake ".* The diagrams in fig. 3 present a similar approach to the basic stereochemistry of the complexes as that due to Pauling, but some h e r points also appear. Because only one of the degenerate pair of anti-bonding d-orbitals is occupied in Co(N0,):- and it must be in either the d , ~ or d,z - ,,z orbital, it will be expected that the full octahedral symmetry will be lost and that the two polar bonds will be longer or shorter than the equatorial ones according as the d,z or dxz - y~ is the one occupied. In the former case, the Pauling treatment equivalent to this would be that, while the equatorial bonds were formed by electron pairs, a three electron and an elec- tron pair bond resonated between the two polar ligands causing a weakness of these bonds.9 They differ in detail in that the approximate wave functions constructed according to the two hypotheses would differ, allowing to a different degree for electron correlation.* 10 In Co(H20)2,+ complete octahedral symmetry wouldJ .W. LINNETT 11 also be lost since a group of three degenerate orbitals arc not symmetrically 00 cupied (i.e. filled or half-filled). However, the effect will be small because the orbitals are non-bonding. On the other hand, in Co(diAs)i' the half-filled orbital is nan-degenerate for a square arrangement so that only bond strain will destroy the full symmetry of the ion.In CoC1:- the set of three degenerate orbitals are half-filled so that the ion is a regular tetrahedron. large overlap small overlap €9 a . . . . b .. .- TI, .. : ;'. 4 p *.=' :. .. .. *. b ' a' . : : ;. 4s ; ; . . * . 3d T29 ... -.. T2g .{ Jigand 'A 8.'. TI" Alg \.* *. .. .. . :. .'. . . .. .. .. *. TIU .. * . * . * . * . 8 . E9 * . A 19 €9 . . (4 atomic and ligand orbitals. FIG. 4.-Diagrammatic representation of possible molecular orbital energy levels of AX6 in relation to the energies of the atomic and ligand orbitals for two cases : (a) Owen 5 (X is C1-) ; (b) Gilde and Ban 6 (X is CN-). (b) Pauling classed complexes as " ionic " or " covalent " according as the electron pairs on the ligands occupied solely ligand orbitals or orbitals involving contribu- tions from the central atom (e.g.FeFz-, ionic; Fe(CN)i- covalent). Molecular orbital theory would treat these two within the same framework rather than as two distinct types. The bonding and anti-bonding molecular orbitals involving the d-orbitals of the central ion are represented by a linear combination of the atom and ligand orbitals. The difference between the energy of the anti-bonding and non-bonding d-orbitals will depend on the amount of mixing with the ligand orbitals in the anti-bonding orbitals (cf. fig. 4). If the mixing is small (Pauling's " ionic ") then the energy difference will be small and the electrons will be spread, in order to reduce inter-electronic repulsion, among the non-bonding and anti- bonding orbitals (FeFz-).If there is considerable mixing Cpauling's " covalent ") * By electron correlation is meant the tendency of electrons to keep apart from one another by virtue of the fact that they are all negatively charged and because of spin effects.12 GENERAL INTRODUCTION the energy of the anti-bonding orbitals will be much greater than that of the non- bonding orbitals and all the electrons will occupy the non-bonding orbitals (Fe(CN)g-). These two treatments are not therefore very different in concept, though they are different in detail. LIGAND FIELD METHOD This has grown out of the crystal field theory11 and envisages the complex as a central ion surrounded by other ions or molecules, the orbitals of which to a first approximation are essentially separate.To that extent, it treats all complexes like Pauling’s “ ionic ” complexes. For an octahedral complex, there are six ligand orbitals fully occupied and a number of orbitals of the central ion wholly or partly occupied. In fact, for transition elements, the theory con- centrates attention on the d-orbitals and considers the effect of the field due to the ligands on their energies, the degeneracy being destroyed by the field (an effect not included by Pauling). Ligand field theory therefore considers only those d-type orbitals which, in the molecular orbital diagrams in fig. 2 are labelled non-bonding and anti-bonding, the ligand orbitals replacing the bonding ones. The sequence of energies is qualitatively the same on the two theories. They differ, however, as to the causes of this splitting.A molecular orbital calculation introduces new interactions between the electrons occupying atomic orbitals and the additional nuclei and electron cores probably allowing also in an empirical way for the other electrons, the energies of a number of new one-electron molecular orbitals being obtained. The energy of the system is then the sum of these for the occupied orbitals. The molecular orbitals may be very different in form from the individual atomic orbitals. On the other hand, the ligand field theory supposes that the ligand orbitals are not affected on complex formation. It considers the effect on the energies of the central atom orbitals by the ligands, the effect resulting mainly from the field due to the electron pairs in the ligand orbitals adjacent to the central atom.The prime cause of the splitting is therefore treated differently in the two cases and it may be that these differences are important when attempting to account for, say, the spectrochemical series and the effects of different ligands on stability constants. The success of ligand field theory lies, to some extent, in the numerical results obtained. These are surprisingly successful considering the ruthlessness involved in the approximations. It has also benefited from a shift of interest towards electronic absorption spectra and their employment in deducing electronic struc- tures. Numerically ligand field theory depends partly on parameters which are assessed empirically and partly on quantities deduced from data obtained from the atomic spectra of the central atoms.Using again an octahedral complex as the example, the field due to the ligands will split the five d-orbitals into a low-lying triplet ( d ~ ~ ~ made up of dxy, dYz and dxz) and a higher energy doublet (dE, made up of dZ2 and dxz - ,,2) (see fig. 5). It has not been possible to calculate satisfactorily the splitting that will result in a given complex, though models in which the orbitals of the central atom are perturbed by six charges or dipoles have been proposed and tested.12 As a consequence the separation of the triplet and the doublet is usually estimated from the spectrum. The data for the hydrates of the di- and tri-valent ions are the most complete.For the former, in the firFt transition series, the splitting varies between about 8,000 and 14,000 cm-1 (23 and 40 kcal/mole) changing in a somewhat irregular manner with atomic number. For the hydrates of the trivalent ions the splitting is considerably greater though not in uniform ratio to the iso-electronic divalent ions. This is peculiar and Jorgensen says : 13 “ If the ligands are not much more polarised, this result can only be explained by treatments taking intermixing of molecular orbitals into account.” The splitting of the levels is greatest with CN- as the ligand. For Cr3+ Jorgensen gives the splittings as 26,300, 21,600, 17,400 and 13,300 for CN-, NH3, H20 andJ . W. LINNETT 13 C1-. This sequence can be understood on the basis of ligand field theory (and also on molecular orbital theory).Allowance has also been made for changes in inter-electron repulsion energy for different occupations of the orbitals. This is usually carried out by a procedure devised by Condon and Shortley 14 and used also by Racah 15 (cf. also Tanabe and Sugano 16). The parameters involved in these calculations can be evaluated from atomic spectra but it appears (Orgel 17 and Owen 18) that they are smaller (half or two-thirds) in the complexes than in the isolated atoms. This decrease in the inter-electronic repulsion indicates that the wave functions are spread over a larger volume in the complexes than in the isolated atoms. This could be due to a decrease in the effective electronegativity of the central atom leading to an ex- pansion of the atomic orbital, or to a mixing of the orbitals of the central atom with those of the ligands.T A 0 S FIG. 5.-Ligand field splitting of the five d-orbitals of the free atom A, by a regular octa- hedral field 0, by a square field in the xy plane or a distorted octahedron with the bonds in the z-direction extended S, and by a tetrahedral field T. Both ligand field and molecular orbital theory predict a splitting of the 3d orbitals on complex formation so that they can clearly both explain many phe- nomena depending on this. Molecular orbital theory provides potentially the more complete description but ligand field theory, by limiting the consideration to a few orbitals, can within this limitation, be carried through more completely and with fewer subsequent assumptions.It seems that in many cases the results of the calculations are most useful (cf. earlier quotation of Orgel). The two ap- proaches should, however, be regarded as complementary, their similarities as well as their differences being borne in mind. n-BONDING So far only u-bonding has been considered. In fact, n-bonding is also possible. For octahedral complexes perhaps the most important n-bonding is that involving the dxy, drz and d,, orbitals labelled in fig. 2 as non-bonding. This n-bonding will be possible when there are appropriate orbitals on the ligands with which the d-orbitals can combine. It is probably important in some cyanides and carbonyls. In these ligands the n-bonding orbitals are filled. Therefore, in ferrocyanide, there are 15 electron pairs to be disposed between 15 molecular orbitals derived from the 3 atomic orbitals on the iron atom and the 12 bonding 7r-orbitals of the ligands.As far as complex formation is concerned, of these molecular orbitals three are bonding, three are anti-bonding, and nine are non-bonding and all are14 GENERAL INTRODUCTION occupied, it does not seem that these can provide any resultant stabilization. However, their effect must be taken into account when considering electronic transitions, because one is concerned with transitions from the highest occupied orbital and the energy of this has been affected by the n-bonding. In ions such as Cr(CN)i- where the above orbitals are not filled there may be some resultant n-bonding. However, the precise form of the low-lying n-orbitals of this type does not appear to have been examined closely.Paramagnetic resonance measure- ments show that the ones occupied by the unpaired electrons do not involve the nitrogen atoms because there is no hypedine splitting due to the nitrogen nuclei.19 On the other hand, Raman spectra and bond lengths show that the strength of the CN bonds is not greatly affected. It is not entirely clear how both these two observations are to be accommodated. A class of ligand in which 7r-bonding is almost certainly particularly important is that exemplified by PEt3 which has a vacant d-orbital which can combine with and accept electrons from a dxr orbital of the central ion. SECTION I1 The second part of the introduction will deal in general terms with some of the more common experimental methods that are being used for studying the structure of transitional metal ions and their complexes.STEREOCHEMISTRY The stereochemistry of complexes has always interested the chemist and the classical method of identifying isomers was first used in studying it. More recently X-ray crystallography has been employed and many structures have been obtained including specially interesting ones such as those of Mo(CN);-,20 and TaF;-.21 The structures of ionic crystals are also of interest. For example, CrF2 has a distorted rutile structure,22 two opposed CrF distances being longer than the other four. This can be explained on crystal field theory because there is one electron in the pair of orbitals which would be degenerate for a regular octahedron and so a Jahn-Teller distortion occurs.Also MoS2 has a layer structure, each molybdenum atom being at the centre of a trigonal prism which has the sulphur atoms at the corners.23 MoS2 is diamagnetic so it appears that bonding may be achieved by a d4sp group of orbitals.24 Nevertheless one could wish that many more crystal structures involving com- plex ions and oxide, halide, etc., lattices had been determined. For example, it is known that the six oxygen atoms are arranged irregularly round the metal atoms in Cr03, Moo3 and WO3.25 But only for Moo3 has the precise structure been determined.26 In that case the three oxygen atoms in the + x, - x and + 3 directions lie on the Cartesian axes, the molybdenum being at the origin.Those in the + y and - y directions are displaced slightly in the - z direction, while that roughly in the - z direction is displaced considerably in the 4- x direction and is much further from the molybdenum atom than the others. This is fascin- ating and it would be interesting to know the irregularities in CrO3 and WO3. Numerous examples of our ignorance of particular structures could be given and the situation is being remedied only slowly. Moreover, more detailed and accurate knowledge of precise shapes would be valuable. For instance, informa- tion about small irregularities, or a series of accurate metal-ligand bond lengths would be most acceptable. INFRA-RED AND RAMAN SPECTRA Another means of measuring the relative strengths of metal ligand bonds might be by using infra-red and Raman spectra. For various reasons this has not been carried out as extensively as might have been expected.For example,J. W. LINNETT 15 the radiation usuaUy used for exciting Raman spectra, the 4358A mercury line, is often absorbed by transition metal complexes. Probably, with the development of alternative sources, this particular difficulty may be avoided. Chatt, Duncanson and Venanzi 27 have employed infra-red spectra to assess electron drifts in platinous complexes. They measured the NH stretching fre- quencies in compounds of the general formula trans-(L, am, PtC12) where L represents a variety of uncharged ligands (P(C3H7)3, C2H4, SbEt3, SeEt2, SEt2, etc.), and “ am ” various amines (e.g. 2 : 6-dimethylaniline, piperidine, etc.).Because the NH frequencies “increase as the nitrogen atom becomes more negatively charged” drifts in the system L-Pt-N-H can be assessed. They consider their observations in connection with the relative ability of the various L groups to direct an incoming substituent into the trans position.2* They con- clude that groups capable of accepting electrons by n-bonding from the dxy orbital of the platinum exert a strong trans-directing effect because there is a withdrawal of electrons from the dx,, orbital which therefore affects particularly the position tram to L, so that nucleophilic attack is favoured in that position. THERMODYNAMIC PROPERTIES To account for the actual and relative stabilities of different complexes must be one of the main objects for the chemist.The stability is best represented by the equilibrium (stability) constant governing the formation of the complex and this may be given as the free energy change AG. In most cases these quantities refer not to some idealized reaction but to the partial or complete conversion of the transition metal ion from the hydrated form in aqueous solution to a complex involving another ligand also in aqueous solution. Ideally one would wish to know, in addition, the contributions to AG of AH, the heat content change, and TAS where AS is the entropy change. Unfortunately only in a few cases are AG, hH and AS all known so, in considering stability constants theoretically, it has usually been necessary to assume that, for the same reaction of a closely related series of ions, AS remained constant and that changes in AG reflected changes in AH.That this can be a satisfactory assumption is indicated by some data due to Care and Staveley29 who showed that AS on the production of the ethylene diamine-tetracetic acid complexes of Ni2+, Cu2f and Zn2f at 20°C were 56.7, 56.4 and 56-3 cal/mole deg. respectively. On the other hand AS for the formation of the bis(ethy1ene-diamine) complexes of the same three ions is - 14.0, - 16.5 and - 9.8.30 Chatt and Wilkins 31 have also drawn attention to the differences in the entropy changes involved in the formation of cis and trans isomers suggesting that the difference may be due to the fact that the former has a dipole while the latter has not. It must therefore be hoped that, in the future, more values of hH itself will become available.An important observation regarding stability constants was made by Irving and Williams 32 and others.33 It was noted that, for many ligands with divalent metal ions in the first transition series, the stability constants increased in the series of divalent ions of Mn, Fey Co, Ni, Cu where a maximum was reached, the constants being smaller for Zn. Irving and Williams interpreted the data in terms of the s u m of the first and second ionization potentials of the metals. However, Orgel34 has pointed out that these ionization potentials do not refer to the same electronic changes in all the above ions because the copper atom, in its ground state, has only one 4s electron whereas all the others have two. Orgel has interpreted the effect in terms of the splitting of the d-orbitals of the central atom by the field due to the ligands, though the principle of the interpretation would not be altered if the d-orbital splitting were regarded in terms of molecular orbital theory.According to Orgel electrons in the lower-lying d-orbitals (the triplet, d ~ ~ ~ , for an octahedron) favour stability, while electrons in the higher ones (the doublet, dEg, for an octahedron) favour instability. The basic covalent bond strength is supposed to be increasing from Mn to Zn. From Fe to Ni the16 GENERAL INTRODUCTION ligand field stabilization is increasing as electrons are added to the lower orbital. In Cu, the extra electron is added to the upper orbital but the anti-bonding effect of this is counteracted by the greater splitting factor for Cu and also by the de- parture from full octahedral symmetry.In Zn, ligand field stabilization is zero because the 3d-orbital is filled both in Zn2+ and in the complex. Orgel 35 has introduced this type of interpretation when he considered the heats of hydra- tion of the divalent ions from Ti2+ to Zn2+ in a paper which, by applying crystal field theory to definitely chemical data, for the first time caused this theory to have a great impact on chemists. This series shows the same behaviour from Mn2+ to Zn2+ as had been noted by Irving and Williams but the behaviour is aIso the same in the fist half of the transition series preceding Mn2+. Effects of the above type may be illustrated in greater detail by Orgel's con- sideration of the first, second and third stability constants for the ethylene-diamine complexes of Mn2+ to Zn2+.36 His analysis is summarized in table 1.The experimental values of log Kt, log K2 and log K3 and their sum are listed; in TABLE l.-LoGmms OF THE EXPERIMENTAL AND INTERPOLATED VALUES OF THE FIRST, SECOND, THIRD AND TOTAL STABILITY CONSTANTS FOR THE ETHYLENE-DIAMINE COMPLEXES, TOGETHER Wi.TH THE DIFFERENCES BETWEEN EXPEFUhENTAL AND INTERPOLATED VALUES.36 M n Fe co Ni cu Zn log K1 expt. log Kl int. A log Ki log K2 expt. log KZ int. log K3 expt. log K3 int. A log K3 log K expt. log K int. A log K A log K2 2.73 2.73 0 2.06 2.06 0 0.88 0.88 0 5.67 5.67 0 428 3-33 0.95 3-25 2-58 0.67 1.99 1-05 0-94 9-52 696 2 5 6 589 3-92 1.97 4.83 3.10 1 -73 3-10 1 -22 1.88 13-82 8.24 5.58 7.52 4-52 3.00 6.28 3.62 2.66 4 2 6 1.38 2.88 18.06 9-52 8-54 1055 5.1 1 5.44 9.05 4-14 4.9 1 - 1.0 1-55 - 2.55 18.60 10.80 7.80 571 5.71 0 4.66 4.66 0 1 -72 1 -72 0 12-09 12.09 0 addition, the values these would have if there were a linear change with atomic number from Mn2+ to Zn2+.The differences between the experimental and theoretical values are also given. For Fe, Co and Ni the increase in the values of the difference is approximately in the ratio of 1 : 2 : 3 which would be expected as 1, 2 and 3 extra electrons are added to the lower triplet of the d-orbitals. With Cu2+ there is a large increase in A log K1 and A log K2 even though, in this case, the fourth electron is being added to the upper doublet. This greater effect, it is interesting to note, is in line with the larger ligand field splitting observed for Cu2' from optical spectra.Moreover, as long as only one or two ethylene- diamine molecules are co-ordinated, the electron in the upper d-orbital will con- tinue to be in an orbital directed towards water molecules and so there is little change in its situation on forming the mono- and di-ethylene-diamine complexes. When Cueng+ is formed the situation is quite different as the geometry of the ethylene-diamine molecule imposes an approximately regular form on the complex and, because there is an odd number of electrons in the anti-bonding d-orbitals ; this is an unsatisfactory situation as shown by the abnormal value of log K3 for copper. It should also be possible to account for the relative stability of octahedral and tetrahedral complexes since for the latter the crystal field splitting is the reverse of the former being into a lower doublet and a higher triplet. Orgel36 has discussed this and pointed out, for example, that it is possible to account for the existence of COX:- complexes because with the group of seven electrons full use is made of the lower doublet whereas in the octahedral complex onlyJ .W. LINNETT 17 partial use is being made of the lower triplet. However the situation here is that more data are needed about tetrahedral complexes (spectra, stability, etc.). The variation in lattice energies for a series of transition metal halides can be understood by taking account of the crystal field splitting of the d-orbitals of the metal ion.VISIBLE AND ULTRA-VIOLET SPECTRA 0 bservations of electronic transitions provide data regarding the energy separation of the higher-occupied and the more low-lying unoccupied orbitals. Such studies will clearly be of the greatest value in giving information about the electronic structures of complex ions. The results obtained can, in principle, be handled either by molecular orbital theory or by ligand field theory. Pauling’s theory is not applicable. Almost all the detailed study and interpretation of the spectra of complexes of the transition metals has taken place in the present decade, the pioneer papers being those of Finkelstein and van Vleck37 on Cr3+, of Abragam and Pryce 38 on Co2+ and of Ilse and Hartmann 39 on Ti(H20)2’, the last involving just one optical d-electron.However, before this, Tsuchida40 had remarked that the spectra are affected only by the first co-ordination sphere and that the ligands in this first shell have the same effect in causing shifts of the position of the absorption bands whatever the central ion. Tsuchida was able to construct a spectrochemical series of ligands such that the groups affected the positions of the absorption band in the order that they were listed. A shortened series is : I-, Br-, C1-, F-, H20, CzOi-, C5H5N, NH3, en, NOz-, CN-. This can be understood either in terms of ligand field theory or of molecular orbital theory. For both it seems likely that the ligand orbital (and the electrons in it) which is directed toward the central ion will be the most important.The more this is concentrated, or is drawn, towards the central ion the greater will be its effect on both theories. The basis of the series is halogen, 0, N, C which is in the order of decreasing effective nuclear charge and electronegativity so that the electrons will be capable of being drawn towards the central atoms as one goes from left to right. As regards the ligands containing nitrogen it is understandable therefore that the negatively charged ligand should have the greatest effect, that the nitrogen which is part of an aromatic ring should have the least, and that the nitrogen atom attached to an electron- releasing aliphatic group should have a slightly greater effect than the nitrogen in ammonia. For mixed complexes a “rule of average environment” obtains, the effect of the mixture of ligands being the mean of the effects of the separate ones.Of course, in certain cases, the effect of mixed ligands is such as to produce a field at the central atom which is so much different from an octahedral field that the multiplets (the doublet and triplet) are themselves split, and as a consequence the absorption bands divide into several components. This effect can be studied quantitatively from the absorption spectra. Jorgensen 41 has discussed the spectral and other effects of changing from octahedral to tetragonal symmetry and proposes that the “ tetragonality ” for different groupings can be measured by comparing the frequencies of the principal bands of the Cu2+ complex which will be tetragonal with that of the Ni2+ complex which will only have any tendency to go tetragonal by virtue of strain effects.Optical spectra provide values of the splitting of the d-orbitals by the ligands. Our knowledge of this has increased greatly during this decade. However, though our knowledge for hexa-aquo complexes is fairly complete there are still many gaps for other octahedral complexes. For tetrahedral complexes the situation is much less satisfactory. Other features derived from electronic spectra have been mentioned in the first part of this discussion.18 GENERAL INTRODUCTION MAGNETIC SUSCEPTIBILITY Measurements of the magnetic susceptibilities of salts containing complex ions have been of the greatest interest since their importance was stressed by Pauling. Since that time this has been much developed by Nyholm and others.With transition metals, the subject divides itself into two parts : (a) the first long period for which much data are available, and (b) the second and third long periods about which very much less is known. With (a) the best representation of the magnetic moment is that based on the view that the spin moment alone contributes. This is an excellent approximation when the d-shell is less than half full, but less good when it is more than half full. The orbital contribution is completely or largely quenched. That is, the ligand field orients the orbitals that are occupied so firmly that there is no effect of the magnetic field on their orienta- tion and hence no orbital contribution. However, because the interaction be- tween the spin and orbital motion is small the spin contribution is free.Con- sequently it is possible to use the measured susceptibility to make an immediate assessment of the numbers of unpaired electrons. With Fez+, Ni2+ and CU~+, and particularly &2+, there does appear to be an orbital contribution. It is interesting that the large orbital contribution for, say, hexa-aquo cobaltous salts is associated with a considerable anisotropy (30 %). With compounds involving tetrahedrally co-ordinated cobalt (e.g. CoCl:-) the orbital contribution and aniso- tropy are much less.42 In this case, the magnetic moment can, it seems, be used as a guide to the stereochemistry, a moment just above the spin only value in- dicating a tetrahedral complex while one considerably above indicates an octahedral one.For the elements of the second and third long periods the situation is more complicated because the spin-orbit coupling is greater because of the greater nuclear charge. This causes the moment to be much less than the spin only value. Kotani43 concludes that in these cases the effective magnetic moment will be temperature dependent and this has been demonstrated for ( N H ~ ) ~ O S B T ~ . ~ ~ More work is undoubtedly needed on the magnetic moments of complexes of the heavier elements. PARAMAGNETIC RESONANCE SPECTRA If a material containing molecules or ions in which there are unpaired electrons is placed in a suitable magnetic field, the substance absorbs microwave radiation, undergoing a transition between the Zeeman sub-levels separated by the field.From this, information can be obtained about the distribution of the odd electron in a paramagnetic complex using either (a) the frequency at which the micro-wave absorption occurs, or (b) the h e structure that is sometimes shown by the ab- sorption. The separation of the levels, with the magnetic field H, is given by gHp, where p is the Bohr magneton, and g, the splitting fact being given by g = 2.0023 - (SA/A) where X is the spin-orbit coupling constant and A is the energy separation of the dTlu and dEg orbitals.45 The last can be obtained from the spectrum so that h can be calculated from g which is obtained from the frequency absorbed and the magnitude of the magnetic field. For Cu(HzO)2,+, for example, the value of h is less than for the free ion, apparently because the electron is in a molecular orbital which is made up in part only of the orbital of the central ion, but to which there is some contribution from the ligand orbitals.The ratio of h for the com- plexes to that for the free ion gives the contribution to the molecular orbital of the appropriate orbital of the central ion. For Cu(H,O)Z' the value obtained is 84 %46 which is for an electron in the d anti-bonding orbital. For the cor- responding bonding orbital the situation must be reversed, it being primarily a ligand orbital.J . W. LINNETT 19 With IrC1;- the absorption line shows a fine structure which results from the small magnetic fields produced by the moments of the chlorine nuclei. The separation of the components is smaller than would be expected for an electron in the appropriate orbital on the chlorine. The magnitude of the splitting is a measure of the contribution of the chlorine orbital to the molecular orbital occupied by the odd electron.The result indicates that the main contribution (74 %) to the molecular orbital is that of the orbital of the iridium atom.47 If the central nucleus possesses a spin magnetic moment it should be possible to make similar deductions from the fine structure caused by this. An important conclusion from this work is that it is not valid to treat the d~~~ and d~~ orbitals as pure central atom orbitals as is done by simple ligand field theory. Moreover, the method does enable definite numerical values to be given to the extent of mixing of the component atomic orbitals.This is of the greatest possible value. Unfortunately this can only be applied to paramagnetic ions. OTHER EXPERIMENTAL METHODS Other methods for obtaining information about electronic structure are nuclear magnetic resonance, particularly high-resolution work and the determination of chemical shifts, optical rotatory power, rates of substitution reactions and nature of products, but space does not permit these being dealt with here. Finally, it must be stressed that there is still need for an increase in our knowledge of the general chemistry of several of the transition elements in the second and third long periods. In conclusion I wish to record my debt to the excellent reviews of Nyholm,48 Orgel,49 Jorgensen,l3 Owen 50 and Griffiths.11 I also wish to thank my colleague Venanzi for his helpful advice.1 Pauling, Nature of the Chemical Bond (Cornell University Press, 1939), p. 93. 2 Pauling, Nature of the Chemical Bond (Cornell University Press, 1939), p. 95. 3 Linnett and Mellish, Trans. Faraday Soc., 1954, 50, 657. 4 Nyholm, Report of the 10th Solvay Council (Brussels, 1956), p. 230. 5 Owen, Faraday SOC. Discussions, 1955, 19, 127. 6 Gilde and Ban, Acta Phys. Chem. Universitatis Szeged, 1957, 3, 42. 7 Wolfsberg and Helmholtz, J. Chem. Physics, 1952,20, 837. 8 Orgel, Report of the 10th Solvay Council (Brussels, 1956), p. 31 1. 9 Pauling, ref. (l), p. 239. 10 Linnett and Dickens, Quart. Rev., 1957, 11, 291. Pople, Quart. Rev., 1957, 11, 273. 11 van Vleck, J. Chem. Physics, 1935, 3, 807.Orgel and Griffiths, Quart. Rev., 1957, 12 Bjerrum, Ballhausen and Jorgensen, Acta Chem. Scand., 1954, 8, 1275. Hartmann 13 Jorgensen, Report of the 10th Solvay Council (Brussels, 1956), p. 375. 14 Condon and Shortley, Theory of Atomic Spectra (Cambridge, 1953). 15 Racah, Physic. Rev., 1949, 76, 1352. 16 Tanabe and Sugano, J. Physic. SOC. Japan, 1954,9, 753, 766. 17 Orgel, J. Chem. Physics, 1955, 23, 1004, 1819, 1824. 18 Owen, Proc. Roy. SOC. A, 1955, 227, 183. 19 Bowers, Proc. Physic. SOC. A, 1952, 65, 860. 20 Hoard and Nordsieck, J. Amer. Chem. SOC., 1939,61, 2853. 21 Hoard, J. Amer. Chem. SOC., 1939, 61, 1252. 22 Jack and Martland, Proc. Chem. SOC., 1957,232. 23 Wells, Structural Inorganic Chemistry (O.U.P., 1952), p. 397. 24 Linnett and Venanzi, unpublished work. 25 Braekken, 2. Krist., 1931, 78, 484. 26 Wooster, 2. Krist., 1931, 80, 504. 27 Chatt, Duncanson, and Venanzi, J. Chem. SOC., 1955,4456,4461 ; 1956, 2712. 28 Chenyaer, Ann. Inst. Platine, U.S.S.R., 1926, 4, 243 ; 1927, 5, 118. 11,381. and Fischer-Wasels, Z. physik. Chem., 1955, 4, 297.20 GENERAL INTRODUCTION 29 Care and Staveley, J. Chem. SOC., 1956, 4571. 30 Davies, Suiger and Staveley, J. Chem. SOC., 1954, 2304. 31 Chatt and Wilkins, J. Chem. SOC., 1952,4300 ; 1955, 525. 32 Irving and Williams, Nature, 1948, 162, 746; J. Chem. SOC., 1953, 3192. 33 Calvin and Melchior, J. Amer. Chem. Soc., 1948, 70, 3270. Schwarzenbach, 34 Orgel, Report of the 10th Solvay Council (Brussels, 1956), p. 292. 35 Orgel, J. Chem. SOC., 1952,4756. 36 Orgel, ref. (34), p. 302. 37 Finkelstein and van Vleck, J. Chem. Physics, 1940, 8, 790. 38 Abragam and Pryce, Proc. Roy. SOC. A, 1951,206, 173. 39 Ilse and Hartmann, 2. physik. Chem., 1951, 197,239. 40 Tsuchida, Bull. Chem. SOC. Japan, 1938, 13, 388,436,471. 41 Jorgensen, Acta Chem. Scand., 1955, 9, 1362. 42Krishnan and Mookherju, Proc. Roy. SOC. A, 1938, 237, 135. Nyholm, ref. (4), 43 Kotani, J. Physic. SOC. Japan, 1949,4, 293. 44 Lundberg and Johannsen, J. Amer. Chem. Soc., 1954,76,5349. 45 Owen, Proc. Roy. SOC. A, 1955, 227, 183. 46 Bleaney, Bowers and Pryce, Proc. Roy. SOC. A , 1955,228, 166. 47 Griffiths and Owen, Proc. Roy. SOC. A , 1954,226,96. 48 Nyholm, ref. (4), and Quart. Rev., 1953, 7, 377 ; 1957, 11, 339. 49 Orgel, ref. (8) and ref. (11). 50 Owen, ref. (5). Bowers and Owen, Reports Prog. Physics, 1955, 18, 304. Bagguley Ackermann and Prue, Nature, 1949, 63, 723. p. 258. and Owen, Reports Prog. Physics, 1957, 20, 304.

ISSN:0366-9033    

DOI:10.1039/DF9582600007

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

I. OPTICAL AND MAGNETIC PROPERTIES INTRODUCTORY PAPER PHYSICAL PROPERTIES OF TRANSITION IONS BY M. H. L. PRYCE Received 2nd October, 1958 In recent years the ions of transition elements have been attracting the attention of many workers in different fields, publishing, often unknown to one lanother in a wide variety of journals. Chemists, inorganic and physical, naturaly have, had an interest in transition ions fur a long time, not necessarily restricted to what may be termed more obviously chemical properties, but also including physical properties which may be used diagnostically to gain insight into chemical struc- tures, such as magnetism and spectra. Physicists might have been expected thirty years ago to have taken an interest in transition-ion spectra, at the time when spectroscopy appeared to be one of the most powerful tools of physics. And to some extent this happened,l-lo but the time was not then ripe for a proper under- standing.More recently, through magnetism, and particularly magnetic ieson- ance phenomena, physicists have been induced to study these very interesting systems, and this has led them to fields on the borders of chemistry and physics similar to those investigated by the chemists. This meeting serves a valuable dual purpose; first, in linking up the efforts of various groups of researchers by bringing their members together in one place, so that each may learn from the others what has been achieved in related fields; and secondly, in giving us all an opportunity to take stock of the situation in a subject which has reached a particularly interesting stage in its development, and of formulating the outstanding problems for future study.It should become clear by the end of this meeting that each aspect of the properties of transition ions is capable of throwing light on all the others, and that many apparently different properties are intimately interrelated. It is particularly fitting that the Faraday Society should be the sponsor of this synthesis of chemistry and physics. My task in this introductory talk is to try to convey the background of the physicists’ view of the nature of the problems, in as intelligible a way as possible. A good place to start, it seems to me, is to ask in what essential way the ions of transition elements differ from other ions, particularly other metallic ions.For it is not without reason that a whole meeting has been devoted to transition ions, to the exclusion of the more law-abiding cations. When we have identified the essential feature we shall then wish to study how it is responsible for the properties that we find interesting, such as the stability of their complexes, their colour (which is a direct visual manifestation of their interesting optical spectra), their magnetism, etc. I would say that the distinguishing feature lies in the arrangement of the elec- trons in the ion. Ions of the ordinary, non-transition, elements have their elec- trons in completed shells of energy levels. In the transition ions, however, the topmost shell of energy levels, or more precisely, orbitals, is incompletely filled. Before going any further, let me remind you that the language of orbitals which I have used-and which I propose to continue using-expresses an ap- proximation to the proper description of the electronic structure.It has the great merit of giving a simple mental picture: electrons are packed into the available energy states or orbitals, one orbital per electron. (I assume that the 2122 INTRODUCTORY PAPER specification of the orbital includes spin-otherwise each orbital can hold two electrons.) The real state of affairs is of course more complicated than is de- scribed by saying n electrons, n orbitals. But it often seems to be a rather good approximation-and one which readily forms the starting point for further re- finements when these are necessary.For my present purpose it is quite adequate. The orbitals of electrons in an ion fall into groups, or shells. In each shell the energies of the orbitals are rather close together; but Werent shells differ widely in energy. In a closed-shell ion the orbitals in the lowest shells are all filled, and those in the higher shells are all empty. Such an electron configuration has certain notable features. The charge distribution is spherically symmetrical ; there is no net angular momentum of the electrons, because to each occupied orbital which might contribute to electron rotation there corresponds one with rotation in the opposite sense, and there is full compensation; similarly there is no net magnetic moment, because the circulating currents all compensate, and the spins are all paired.Furthermore, it takes considerable energy to excite an ion from this configuration to its next highest energy level, since this requires the transfer of an electron from an orbital in the topmost occupied shell to one in a higher shell. Also, it generally requires considerable energy to detach an electron from such an ion, since even the topmost orbital is relatively tightly bound ; while at the same time the affinity for a further electron is relatively low, as only orbitals in the higher shells are available. Such ions therefore exhibit no paramagnetism; and are unable to absorb quanta of visible light, which have too little energy to excite them. Their compounds are therefore colourless and diamagnetic, while their stereochemistry tends to be dominated by their spherically symmetric charge distribution.When the topmost shell of orbitals is incomplete, however, the situation is radically different in all the respects just mentioned. Before going on to analyze in greater detail how this comes about it is well to remark on one important feature of the transition ions, namely, their degeneracy in the ground state. That is to say, there is more than one quantum state with the same lowest energy. In fact there is a whole infinity of states, whose wave-functions are made up by linear combination from a limited number of independent wave-functions. Consequently there is a great variety of charge configurations, each corresponding to the same lowest energy. It may be instructive to give an example.The Fez+ ion has six 3d-electrons in addition to those occupying closed shells. Now the 3d-shell contains ten independent orbitals, so that there are as many independent states of the ion made up in this way as there are ways in which one can arrange six objects in ten boxes, namely, 210. It is true that not all these states have pre- cisely the same energy, because although there are always six electrons in 3d- orbitals, their correlation in space is different, so that their mutual energy, arising from their electrical repulsion, is different. Nevertheless, a good deal of degeneracy remains, and in any case, the energy differences of the 210 levels are relatively small. In the example quoted, there is a lowest group of 25 levels within an energy span of 2,000 cm-1, the ground state being ninefold degenerate ; and the majority of the remaining levels lie within 40,OOO cm-1 (5 ev) of the ground state.That the ion now possesses magnetic moment is a natural consequence of the fact that with an incomplete shell the magnetic effects of circulating electric currents, or of electron spins, no longer necessarily cancel. The degeneracy we have just described is a manifestation of the different “ directions ” in which the magnetic moment can point, which are governed by the socalled ‘‘space quantization ”. What has just been said applies to the free ion, uninfluenced by external sur- roundings, and needs modifying for ions in combination-as in complexes or in inorganic crystals. The fact that in the free ion there are many different charge distributions for the same energy means that in an external force field Werent states will behave differently and have different energies.In other words theM. H. L. PRYCB 23 external surroundings remove the degeneracy-or partly. One may visualize this as follows. Some states have electron distributions which point strongly towards negatively charged regions in the surroundings (e.g. the ligands of a complex)- they will have a relatively high energy because of the mutual repulsion of negative charges. Others will have electron distributions avoiding the negative charges, and will consequently have a lower energy. This splitting of degenerate levels by the surroundings is the most important factor in determining the optical absorption of transition metal compounds, and hence their colour.The fact that the lowest energy level corresponds to a charge distribution adapted to fit snugly into the interstices between surrounding negative charges implies that a transition ion compares favourably with an ordinary ion of the same radius, which cannot so adapt itself.11-13 Its compounds and complexes are therefore stabilized in comparison. In most compounds the stabilization energy increases sharply with decreasing cation-anion (or cation-ligand) distance, and this results in the configuration of overall minimum energy being achieved for a shorter cation-anion distance than would happen for a simple cation.14-16 In other words, the effective ionic radius of a transition ion is decreased by its powers of adaptation to its surroundings.Similarly, since the stabilization energy also depends on the shape of the framework of surrounding ligands or ions, the requirement of minimum energy influences this shape. In complex ions where the Jahn-Teller effect is strong, this can be very important.12.17-21 A well-known example is Cu2+6H2O, in which the octahedron of water molecules is so distorted that four are much closer to the central ion and the other two are further from it, than would be expected for an aquo-complex in this part of the periodic table (Ni2+6H20 by contrast is quite regular.)13. 199 20 This state of affairs, which I have described in qualitative terms, can be given a quantitative expression. Energy differences, magnetic moments and spectro- scopic splitting factors such as are related to paramagnetic resonance observations, can all be related to models of the field set up by the ligands.ll.22-42 Relations between observable quantities are deduced, which are independent of the details of the model, and can be tested.If the ion is only loosely associated with its ligands, the orbitals for the electrons fall into two distinct classes : those localized on the ion ; and those localized on the ligands. And the orbitals on the ion are very little different from the orbitals of the free ion. Calculations based on the approximation that the ionic orbitals are unchanged by bonding are often referred to as “crystal field theory” or ‘‘ ligand field theory ”. They have the merit of being often quite simple and straightforward, and so of giving an insight into the outlines of the situation.For so-called “ ionic ” compounds, where the orbitals are reasonably well localized, ligand field theory gives a very good account of magnetic, optical and energetic properties. Even in ‘‘ covalent ” complexes, where the orbitals spread strongly from the central core to the ligand atoms, it makes a useful first approximation. One may see qualitatively how bonding affects the orbitals. Those orbitals originally localized on the ligands, which are directed towards the central ion, acquire a certain amplitude in the region of the ion, particularly where the 3d wave-function is important. If the ligand atoms are oxygen or nitrogen, the 2p-orbitals are the ones most affected in this way. Their energy is lowered, so that they become bonding orbitals.They are always completely filled in transition- ion complexes. The orbitals which in the free ion are pure 3d, and are directed toward the ligands, similarly overspill on to them. Their energy tends to be raised, as they correspond to anti-bonding hybrid orbitals. Such considerations of course apply to closed-shell ions as well as to transition ions, and to some extent modify the statement that the former’s stereochemistry is dominated by their spherical shape. It remains true, nevertheless, that closed-shell ions are less adaptable to their surroundings.24 INTRODUCTORY PAPER Since the magnetic properties and the optical absorption are connected with the orbitals of the unfilled shell, such aspects of them as are determined by the behaviour of the electrons close to the central nucleus will be reduced if the orbitals overspill on to the ligands.This provides a quantitative measure of the overspill, and in a certain sense may be taken as one measure of the covalent, as opposed to ionic, character of the complex. Effects which can be used in this way are the hyperfine coupling, and the effective spin-orbit coupling as revealed by g-values, in paramagnetic resonance ; and the wavelengths of the sharp lines in the optical absorption spectra. The former two arise from the interaction of electrons with either the nucleus of the ion, or with the electric field in its immediate vicinity, and are therefore reduced from the free-ion values proportionately to the electron probability- density.27.38s 439 44 The latter comes from the mutual repulsion between pairs of electrons in the unfilled shell, which again comes mainly from the probability density close to the nucleus.27.45 Jn relation to the sharp lines in the spectra it is worth making a few comments. They correspond to transitions between states which have, on the orbital ap- proximation, the same charge density, but in which the electrons are differently correlated. Since the charge density is the same in the two states, the energy difference is unaffected by variations in the ligand field. In particular the equilib- rium configuration of the ligand-framework in the excited state is the same as in the ground state. Consequently no vibrations are excited by the electronic transi- tions, and the energy required is just the electronic energy-difference.37 These sharp lines are always very weak, because they correspond to a rearrangement of the spin orientation, without change of orbital state, and such " intersystem combinations " as they are called in atomic spectroscopy, are weak.The main absorption of light, by contrast, is spread over a broad band of frequencies. The main reason for this is that, since the charge distribution in the upper state reached by the electronic transition when light is absorbed is different from that in the ground state, the ligands, which are near their equilibrium position in the ground state, are no longer in equilibrium in the upper state, and vibrations are set up. Since there are very many vibration modes in a crystal or in a solution, there is a wide spread in the energy going into vibration.The quantum energy of the light is the sum of the electronic energy-difference and the energy of vibration-so one sees that it is spread over a wide spectral region. Thermal agitation, which causes fluctuation in the crystal field, also contributes to the width, though not so importantly. The experimental approach has been along three main lines : thermodynamic, spectroscopic and magnetic. By thermodynamic I mean the broad approach towards measuring energies or heats of formation and reaction, stabilities of complexes.12* 46 Many complexing agents and most of the ions of the 3d group have been investigated.34-36.47148 Spectra have been studied in the solid state for simple hydrated com- pounds,l-9*49-53 and more extensively in solution for a wide variety of com- plexes.8-10s 54-58 The solid-state spectra show some fine structure if taken at low temperatures (14°K or below), which is smeared out at higher temperatures.Of recent years, measurements of oscillator strengths and band widths have proved interesting, in view of developments in the theory which enable calculations to be made.59-64 In solution spectroscopy the establishment of spectrochemical series of complexing agents has proved fruitful, throwing light on the mechanism Magnetic studies have dealt with many aspects. Susceptibility measurements have provided chemists with a tool for probing stereochemistry and the electronic structure of the complexes.22r 24.25.67.68 More recently, paramagnetic reson- ance has complemented this approach, and in favourable cases given information of a very detailed nature concerning the electronic structure.26127.38.39. 69-71 of bonding.65.66.34,35. 36.48M. H. L . PRYCE 25 The following list of references is far from complete. The reader is referred to the bibliographies at the end of the specialized papers of this Discussion, for a more comprehensiv: sirvey. 1 du Bois and Elias, Ann. Physik, 1908, (4), 27, 233. 2 Sauer, Ann. Physik, 1928 (4), 87, 197. 3 Snow and Rawlins, Proc. Camb. Phil. SOC., 1932, 28, 522. 4Deutschbein, Ann. Physik, 1932 (5), 14, 712, 729; 1934 (5), 20, 828; 2. Physik, 5 Joos and Schnetzler, 2. physik. Chem By 1933, 20, 1. 6 Spedding and Nutting, J. Chem. Physics, 1934, 2, 421 ; 1935, 3, 369.7 Gielessen, Ann. Physik, 1935, 22, 537. 8 Houston, Proc. Roy. SOC. Edin., 1911, 31, 538. 9 Dreisch and Trommer, 2. physik. Chem. B, 1937, 37,40. 10 Dreisch and Kallschener, 2. physik. Chem. B, 1939, 45, 19. 11 Penney, Trans. Faraday SOC., 1940, 36, 627. 12 Orgel, J. Chem. SOC., 1952, 4756. 13 Orgel, Report of 10th Solvay Council (Brussels, 1956), p. 289. 14 van Santen and van Wieringen, Rec. trav. chim., 1952, 71, 420. 15 Hush and Pryce, J. Chem. Physics, 1957, 26, 143. 16 Hush and Pryce, J. Chem. Physics, 1958, 28, 244. 17 Jahn and Teller, Proc. Roy. SOC. A , 1937, 161,220. 18 van Vleck, J. Chem. Physics, 1939, 7, 72. 19 Opik and Pryce, Proc. Roy. SOC. A, 1957,238,425. 20 Dunitz and Orgel, J. Physics Chem. Solids, 1957, 3, 20, 318.21 McClure, J. Physics Chem. Solids, 1957, 3, 3 11. 22 Schlapp and Penney, Physic. Rev., 1932, 42, 666. 23 Finkelstein and van Vleck, J. Chem. Physics, 1940, 8, 790. 24 Polder, Physica, 1942, 9, 709. 25 Kotani, J. Physic. SOC. Japan, 1949, 4, 293. 26 Abragam and Pryce, Proc. Roy. SOC. A , 1951, 205, 135. 27 Abragam and Pryce, Proc. Roy. SOC. A , 1951, 206, 164, 173. 28 Ilse and Hartmann, 2. physik. Chem., 1951, 197, 239. 29 Hartmann and Use, Z. physik. Chem., 195 1, 197, 1 16. 30 Hartmann and Use, 2. Naturforschung, 195 1 , 6a, 75 1. 31 Hartmann and Schlafer, 2. Naturforschung, 1951, 6a, 760. 32 Tanabe and Sugano, J . Physic. SOC. Japan, 1954, 9, 753, 766. 33 Tanabe and Sugano, J. Physic. SOC. Japan, 1955, 11, 864. 34 Jsrgensen, Acta Chem. Scand., 1954, 8, 1502. 35 Jsrgensen, Acta Chem.Scand., 1955, 9, 1166, 1362. 36 Jargemen, Acta Chem. Scand., 1956, 10, 500, 518, 887. 37 Orgel, J. Chem. Physics, 1955, 23, 1004. 38 Owen, Proc. Roy. SOC. A, 1955, 227, 183. 39 Bleaney, Bowers and Pryce, Proc. Roy. SOC. A , 1955,228, 166. 40 Bostrup and Jargensen, Acta Chem. Scand., 1956, 10, 1501. 41 Bjerrum and Jsrgensen, Rec. trav. chim., 1956, 75, 658. 42 Ballhausen, Rec. trav. chim., 1956, 75, 665. 43 Stevens, Proc. Roy. SOC. A, 1953, 219, 542. 44 Owen and Stevens, Nature, 1953, 171, 836. 45 Jsrgensen, this Discussion. 46 Care and Staveley, J. Chem. SOC., 1956, 4571. 47 Bjerrum, Adarnson and Bostrup, Acta Chem. Scand., 1956, 10, 329. 48 Schaffer and Jargensen, Ricerca Sci. Suppl., 1958, 28, 3. 49 Holmes and McClure, J. Chem. Physics, 1957, 26, 1686. 50 Pappalardo, Nuovo Cimento, 1957, 6, 392. 51 Pappalardo, Phil. Mag., 1957, 2, 1397. 52 Pappalardo, Phil. Mag., 1959, in press, 53 Low, 2. physik. Chem., 1957, 13, 107. 54 Tsuchida and Kobayashi, Bull. Chem. SOC. Japan, 1938, 13,476. 55 Kiss, Abraham and Hegedus, 2. anorg. Chem., 1940,244,99. 56 Rabinowitch and Stockmayer, J. Amer. Chem. SOC., 1942, 64, 335. 1932, 77, 489.26 INTRODUCTORY PAPER 57 Furmann and Garner, J. Amer. Chem. Soc., 1950, 72, 1785. 58 Hartmann and Schlafer, 2. Naturforschung, 1951, 6a, 757, 760. 59 van Vleck, J. Physic. Chem., 1937, 41, 67. 60 Broer, Gorter and Hoogschagen, Physica, 1945, 11, 231. 61 Liehr and Ballhausen, Physic. Rev., 1957, 106, 1161. 62 Koide and Pryce, Phil. Mag., 1958, 3, 607. 63 Koide, Phil. Mag., 1959, in press. 64 Englman and Pryce (to be published). 65 Fajans, Naturwiss., 1923, 11, 165. 66 Tsuchida, J. Chem. SOC. Japan, 1938,13, 388,426, 471. 67 Selwood, Magnetochemistry (Interscience, New York, 1956). 68 Nyholm, Report of 10th Sohay Council (Brussels, 1956), 230. 69 Griffiths and Owen, Proc. Roy. SOC. A , 1952, 213,459. 70 Griffiths and Owen, Proc. Roy. SOC. A, 1954, 226,96. 71 Griffiths, Owen and Ward, Proc. Roy. SOC. A , 1953, 219,526.

ISSN:0366-9033    

DOI:10.1039/DF9582600021

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

OPTICAL ABSORPTION IN THE DIVALENT OXIDES OF COBALT AND NICKEL BY W. P. DOYLE AND G. A. LONERGAN Dept. of Chemistry, University College, Upper Merrion Street, Dublin Received 9th July, 1958 The absorption spectra of cobaltous and nickelous oxides have been determined between 200 and 1000 mp by transmission through thin films and between 360 and 1000 mp by the technique of pressed potassium bromide discs. The first maxima in the thin film spectra, at 5.9 eV for cobaltous oxide and 4.4 eV for nickelous oxide, are interpreted as exciton transitions and the second maximum in nickelous oxide, at 5.2 eV, is attributed to the transition to the conduction band. The pressed disc spectra are discussed in terms of crystal field theory. It was not found possible to prepare thin films of ferrous oxide.During the past few years there has been considerable development in the interpretation of the optical properties of transition metal complexes. Ab- sorption in the visible and near ultra-violet has been discussed in terms of crystal field theory 1 and that in the further ultra-violet has been attributed to electron transfer processes.2 Interest has largely centred on octahedral complexes in solution and there is little data on solid compounds of the transition metals. In solids with the sodium chloride lattice the metal ion is octahedrally surrounded by negative ions so that it is in an environment analogous to that in an octahedral complex. Thus absorption data on such solids may be of interest and the object of the present investigation was to determine the absorption spectra between 200 and 1OOOmp of some transition metal compounds having the sodium chloride lattice.EXPERIMENTAL THIN FEM SPECTRA Thin films of the oxides studied were prepared by oxidation of evaporated metallic films. The evaporation chamber consisted of a glass bell-jar vacuum-sealed to a steel base-plate by an L-gasket. The base-plate was connected to the pumping-system and carried the leads for the heating element. The latter was a small boat of 0.002 in. thick tungsten strip and was cleaned before use by heating in hydrogen. Evaporation was carried out at a pressure of less than 10-4 mm of mercury and the evaporated layers were deposited on amorphous quartz discs supported vertically over the heating element at a distance sufficient to ensure uniformity of film thickness.For the evaporation of nickel, cobalt or iron, the metal is electroplated on to the heating element.3 Cobalt films were oxidized by heating in a stream of steam at 300"C, preheating and cooling being carried out in nitrogen. The films so obtained were grey in colour, uniform and free from turbidity. Three typical films were identified by electron diffraction as the divalent oxide. The normal COO diffraction pattern was obtained from one film while the patterns from the other two films, although certainly of COO, were somewhat dis- torted. Two of the films were oriented with the (100) plane parallel with the surface while in the third preferred orientation could just be detected. In the pattern from one film a few faint additional lines were present ; these could be due to a higher oxide and might represent approximately 5 % of the oxide.Impurity concentrations up to about 10 % are not detectable in the transmission spectra of thin films 4 and the spectrum of the film containing the higher oxide was identical with that of the films containing no higher oxide. The differing degrees of distortion and preferred orientation observed by electron diffrac- tion caused no differences in the absorption spectra. Films prepared by heating the metal in oxygen at 400°C and then heating in carbon dioxide at 950°C to dissociate any higher oxide 5 always contained signiscant quantities of higher oxides. 2728 OPTICAL ABSORPTION I N coo A N D NiO Nickel films were oxidized by heating in a stream of oxygen at 300°C. The films so obtained were grey-green in colour, uniform and free from turbidity. Two typical films were examined by electron diffraction.The diffraction pattern from one film, although badly distorted, was recognizable as that of the divalent oxide. With the other film a weak, but sharp, pattern was obtained which was unlike that of either NiO or nickel. Distortion prevented accurate calculation of lattice spacings and the film could not be identified. However, the absorption spectra of the two films were the same. The reason for this apparent discrepancy between the electron diffraction results and the absorption results is not known. The unidentified film was about one-third as thick as the NiO film but it is believed that this fact has not influenced the identification except in so far as it introduced greater accumulation of charge due to the insulating nature of the quartz mount.Ferrous oxide can be prepared by oxidation of iron at temperatures above 575°C at low oxygen pressures ; 6 at room temperature it is metastable with respect to higher oxides but the rate of transformation is slow.7 Iron films were heated in an evacuated tube to 9OO"C, oxygen at a pressure of 0.01 mm of mercury was admitted for 15 min, the tube was then re-evacuated and rapidly cooled. Films prepared in this way were always orange-brown in colour indicating that they consisted largely of higher oxides and not the divalent oxide which is black. Presumably in thin films the rate of transformation is greater than in the bulk material.The absorption of a film was determined in a calibrated Beckmann model DU spectro- photometer and the absorption coefficient calculated from the thickness determined by weighing as described previously.8 For each oxide, the absorption of at least three films covering a variation in thickness of a factor of at least two was measured and the ab- sorption curves were found to be reproducible. PRESSED DISC SPECTRA The transmission spectra of pressed potassium bromide discs containing 0.1, 0.2 and 0.5 % NiO and 0-15 and 0.3 % COO were measured, relative to that of a blank pressed disc, from 360 to 1OOOmp. NiO was prepared from the nitrate9 and the product was light green in colour and therefore reasonably stoichiometric.10 COO was prepared 5 by ignition of Co304.RESULTS The absorption spectra, from 200 to lo00 mp of typical films of COO and NiO are shown in fig. 1 and 2 respectively. In COO, the absorption increases rapidly from about 500 mp h c., .w 0.4 'd d a 3 .3 c 0 . 2 2 0 0 2 5 0 300 5 0 0 1000 wavelength in mp FIG. 1.-Absorption spectrum of COO film.29 (2-5 ev) and rises to a maximum at 210 mp (5.9 ev). The main features of the spectrum of NiO are the threshold at 600 mp (2.1 eV), the broad maximum at 280 mp (4.4 eV) W. P. DOYLE AND G . A . LONERGAN wavelength in mp FIG. 2.-Absorption spectrum of NiO film. 0 . 4 0 .d 8 a - 8 ._ Y a 0 0.35 0 . 3 3 0 0 400 6 0 0 1000 wavelength in mp FIG. 3.-Absorption spectrum of pressed KBr disc containing 0-1 % NiO. and the peak at 240 mp (5-2 eV). For both oxides the absorption coefficient in the region of maximum absorption is of the order of 106 cm-1.The absorption spectrum of the pressed potassium bromide disc containing 0.1 % NiO is shown in fig. 3. The steps in the range 550-36Omp are considered to be absorption30 OPTICAL ABSORPTION IN COO AND NiO maxima which appear flattened due to the underlying continuous rise in absorption which is probably du0 to the tail of the fundamental band. The only feature observed in the pressed disc spectrum of COO was a broad maximum extending from about 480 to 360 mp. DISCUSSION FIRST MAXIMUM IN THIN FILM SPECTRA For both COO and NiO the absorption coefficient in the region of maximum absorption is of the order of 106cm-1, which can occur only if the oscillator strength is very close to unity.Thus the absorption cannot be due to defects or impurities, nor can it be attributed to transitions in the cation which are forbidden in the free ion but become allowed with a small transition probability in the lattice as a result either of thermal oscillations or the symmetry-type of the lattice.11 Bands of such intensity must correspond either to fully permitted transitions in the cation or to electron transfer processes.12 For the Co2+ ion the first allowed transition (3d7 + 3d64p) is at 12.2 eV and for the Ni2+ ion the first allowed transi- tion (3d8 -+ 3d74p) is at 13.7 eV.13 In the crystal, the energy of transition may be expected to be somewhat reduced,ll by about 2eV, because the electrons in the excited state are able to employ the crystalline field to advantage.However, interpretation of the first maxima in COO and NiO as allowed cation transitions would require that the energy of transition be reduced by as much as 6 to 9 eV which appears unlikely. It is probable therefore that the maxima should be attributed to an electron transfer process. It has often been suggested that in binary ionic inorganic solids the first ab- sorption maximum corresponds to the transfer of an electron from the negative to the positive ion. For the alkali halides the energy of this process has been calculated 14 using the following cycle-a negative ion is removed from the crystal, converted to an atom and the atom replaced; then a positive ion is removed, converted to an atom and the atom replaced; the values so calculated agree well with the experimental data.For COO and NiO, in which both cation and anion are doubly charged, the cycle gives the energy of transfer of an electron from a negative to a positive ion as where CCM is the Madelung constant for the crystal, e, the electronic charge, Y, the interatomic distance, E2, the electron affinity of the 0- ion, 12, the second ioniza- tion potential of the metal. w, the energy of polarization resulting from the absorption act, is given by the expression 15 where a is the sum of the polarkabilities of the ions. For NiO, u was calculated from the refractive index using the Lorentz-Lorenz formula. For COO, whose refractive index has not been reported, the value of o was taken to be the same as for NiO. For the alkali halides, an additional term was introduced to allow for the fact that energy is required to insert an alkali atom in the lattice compared with the smaller positive ion which was extracted.For both nickel and cobalt, the singly charged ion resulting from the absorption act differs from the doubly charged ion in having an extra electron in the 3d shell ; thus there is little difference in size between the singly and doubly charged ions so that the interaction energy of the singly charged ion with the lattice is small and may be neglected. The quantities necessary for the cycle calculations and the resulting values of E are given in table 1. Electron transfer from negative to positive ion corresponds in the zone theory of solids to transitions from the highest full band to the first exciton level so that it may be necessary to consider band width.E calculated as described presumably E = (4uM - l)e2/r + E2 - I2 - W, o = 2.03 e2u/r4,W. P. DOYLE AND G. A . LONERGAN 31 corresponds with the separation between the peak:of the full band and the exciton level. Thus the energy of transitions from near the top of the full band will be less than E by an amount depending on the band width. In the alkali halides the full band is narrow and thus satisfactory agreement with experiment is obtained without considering band width; however, the bands for oxides are much wider and must have a considerable effect. There is no experimental evidence on the width of the highest full band for COO or NiO but soft X-ray emission spectra TABLE 1 COO ( 4 a ~ - l)e2/r, eV 408 12, ev 17.1 rr 8, 2.12 E2, ev - 9.1 nred tc W, eV I 3.7 NiO 208 415 - 9.1 18.2 2.18 2.36 X 10-24 3.7 E, eV 10.9 10.5 have shown 16 that in FeO the width of the oxide ion band is 19 eV, a large part of the width being due to ends where the electron density is very low.It appears that the band ends have not sufficient density to affect the optical behaviour of the crystal so that the effective optical width is the width without the low density ends.17 For FeO the effective width is 8 eV and the top of the band is then 4 eV above the peak. It is assumed that for COO and NiO the band width is the same as for FeO because the width is determined by the extent of overlapping of the oxygen ions and since there is little difference between the radii of the ferrous, cobaltous and nickelous ions, the extent of overlapping, and hence the band width, will be much the same.Thus the expected energy of the exciton transition is 4 eV less than E, that is 6.9 eV for COO and 6.5 eV for NiO. Considering the uncertainties of calculations of this type, these values agree sufficiently well with the experimental values of 5.9 eV and 4.4 eV respectively to confirm that the first absorption maxima are associated with electron transfer from negative to positive ion. For NiO, this interpretation is supported by an analysis of the conductivity, Seebeck effect and optical transmission between 440 and 2500mp from which it has been suggested 18 that the absorption band extending from about 460 mp to shorter wavelengths is probably associated with excitation of electrons from the top full band.SECOND MAXIMUM IN THIN FILM SPECTRA For the alkali halides, the second maximum in the absorption spectrum has been interpreted 19 as the series limit, that is as a transition from the full band to the conduction band and it seems probable that the second peak in NiO may be interpreted in the same way. It has been pointed out 20 that for these excited states of higher quantum number it is preferable to regard the electron in the field of the positive hole as a hydrogen-like system in a medium of dielectric constant KO, so that the separation between the energies of the higher states should be similar to those for the hydrogen atom decreased by a factor ~ 0 2 . Thus if the separation between the exciton level and the series limit is known for one crystal, it can be estimated 17 for others of known KO. Here, KO, the dielectric constant at infra-red frequencies, is estimated from observed indices of refraction.For NaCI, the energy separation between the first and second absorption maxima is 1-9 eV,4 the refractive index is 1.54 and that of NiO is 2-18. Thus, for NiO the32 OPTICAL ABSORPTION I N COO AND NiO expected energy separation is 0.5 eV which is sufficiently close to the experimental value, 0.8 eV, to suggest that in NiO also the second absorption maximum cor- responds to transitions from the full band to the conduction band. NiO does not show photoconductivity between 250 and 800 mp.21 Thus interpretation of the maximum at 280mp as an exciton transition requires that the exciton is not thermally dissociated at room temperature.Further, if the peak at 240 mp is associated with transitions to the conduction band, NiO should be photoconducting below 240 mp. However, no experimental evidence on this point is available. PRESSED DISC SPECTRA The maxima observed in the spectrum of NiO by the technique of pressed potassium bromide discs are given in table 2 (column 3) together with those observed in the transmission spectrum 18 of a 14 p thick film of NiO (column 1) and in the diffuse reflection spectra 22 of NiO + MgO mixed crystals (column 2). In column 4 of table 2 are given the maxima observed in the hydrated Ni2+ ion in aqueous solution which have been assigned 1 to the transitions predicted (columns 5 and 6) using a slightly modified crystal field theory.TABLE 2 observed maxima (cm-1) predicted transitions NiO film 8,070 15,400 From table containing the NiO + Mg0 NiO + K33r Nii$ frequency (cm-1) assignment 8,500 8,500 3A2g-3T2g 13,900 14,500 14,000 14,000 3A2g-3 TIg 21,500 22,200 22,500 22,500 3A2g-1 T2g 25,000 25,000 25,600 3A~g-1A lg 26,600 26,000 27,000 3A2g-3 Tlg 14,900 15,000 3A2g-1Eg 2 it can be seen that the transitions observed in the solid systems Ni2+ ion correspond well with one another and with both the experimental and predicted transitions for the aquo-complex. Thus the spectral data on solids may be interpreted in the same way as those for octahedral nickelous complexes in solution in terms of the transitions within the Ni2+ ion predicted from the crystal field theory.1 This correspondence may be expected as in all the solid systems studied the Ni2+ ion will be octahedrally surrounded by negative ions since NiO, MgO and KBr all possess the rock-salt lattice.The closeness in the frequencies observed for the solid systems and those of the aquo-complex indicate that the appropriate value of Dq for the Ni2+ ion in the solids studied is practically the same as that for the aquo-complex,l 850cm-1. A few mis- cellaneous points arise in connection with the data of table 2. First, the maximum observed at 25,000 cm-1 in both the NiO + MgO mixed crystals and the pressed disc corresponds quite well to the transition to the 1A1, state; this transition has not been observed in the aquo-complex. Secondly, the band at 8070 cm-1 in the NiO film has been interpreted 18 as due to excess oxygen.The correspondence of this band with the 8,500 cm-1 band in the aquo-complex would suggest that it also may be associated with the transition to the 3T2= state of the Ni2+ ion. Thirdly, the aquo-complex band at 14,000 cm-1 which has been assigned 1 to the transition to the 3Tlg state consists of two bands 23 with maxima at 13,800 and 15,100 cm-1. These two maxima correspond well to the transitions to the 3T1, state at 14,000 cm-1 and that to the 1E' state at 15,OOO cm-1. The broad band from about 21,000 cm-1 to about 26,000 cm-1 observed in the pressed disc spectrum of COO appears to correspond to the broad complexW. P. DOYLE A N D G . A . LONERGAN 33 band of the hydrated Co2+ ion which stretches from about 15,000 cm-1 to about 28,000 cm-1 and whose position i s in agreement 1 with the transitions expected from the crystal field theory.The electron diffraction studies were made by Miss M. A. Barrett in the Research Department, Imperial Chemical Industries, Ltd., Metals Division, by courtesy of Dr. N. P. Inglis. One of us (G. A. L.) is indebted to the Department of Education, Republic of Ireland, for a maintenance grant. We wish to thank Prof. T. S. Wheeler for his continued support and encouragement. 1 Orgel, J. Chem. Physics, 1955, 23, 1004. 2 Orgel, Quart. Rev., 1954, 8, 422. 3 Olsen, Smith and Crittenden, J. Appl. Physics, 1945,16,425. 4 Schneider and O’Bryan, Physic. Rev., 1937, 51,293. 5 Chufarov, Zhuravleva and Tatievskaya, Doklady Akad. Nauk S.S.S.R., 1950, 73, 6 Iimori, Nature, 1937, 140, 278. 7 Forestier, Ann. Chim., 1928, 9, 316. 8 Doyle, J. Physics Chem. Solids, 1958, 4, 144. 9 Cairns and Ott, J. Amer. Chem. SOC., 1933, 55, 527. 10 Le Blanc and Sachse, Z. Elektrochem., 1926, 32, 58, 204. 11 Seitz, Rev. Mod. Physics, 1951, 23, 328. 12 Rabinowitch, Rev. Mod. Physics, 1942, 14, 112. 13 Moore, Atomic Energy Levels, vol. I1 (Natl. Bur. Stand Circ. 467), 1952, 85, 103. 14 von Hippel, Z. Physik, 1936, 101, 680. 15 de Boer, Electron Emission and Adsorption Phenomena (University Press, Cam- 16 O’Bryan and Skinner, Proc. Roy. SOC. A, 1940, 176, 229. 17 Wright, Proc. Physic. SOC., 1948, 60, 13. 18 Morin, Physic. Rev., 1954, 93, 1199. 19 Mott, Trans. Faraday Soc., 1938, 34, 500. 20 Mott and Gurney, Electronic Processes in Ionic Crystals (University Press, Oxford, 21 de Boer and Verwey, Proc. Physic. SOC., 1937, 49, suppl. 59. 22 Kroger, Vink and van der Boomgaard, Physica, 1952, 18, 77. 23 Roberts and Field, J. Amer. Chem. SOC., 1950, 72, 4232. 1209. bridge, 1939, p. 241. 1940), p. 99.

ISSN:0366-9033    

DOI:10.1039/DF9582600027

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

THE ABSORPTION SPECTRUM OF VANADIUM CORUNDUM BY M. H. L. PRYCE* AND W. A. RUNCIMAN-~S Received 14th July, 1958 A crystal of A1203 (V) has been studied in absorption over the wavelength range 0.4-1.2 p. Measurements have been made at 4.2, 20 and 77"K, and polarization spectra obtained in the visible region. Two absorption bands have been resolved into finer struc- ture than has been previously reported, and one of the components has been further split by a magnetic field. The relevant crystal field theory for two d-electrons in a trigonal field has been developed, using an octahedral field to provide a preliminary classification of states. The main features of the spectra are discussed in terms of this theory and possible complications due to Jahn-Teller distortions are considered.1. INTRODUCTION Many solids owe their colour to small quantities of impurities. For instance, ruby consists of colourless corundum, Al2O3, with the addition of some trivalent chromium ions which substitute in the lattice for about 1 % of the aluminium ions. In the same way, vanadium corundum contains trivalent vanadium ions, which are assumed to substitute for aluminium ions. The colouring is distinctive, being bluegrey in daylight and plum-coloured in artificial light. Unlike ruby, it is not found in nature, but it can be grown by the Verneuil process. Cut gem- stones of this type are known as synthetic sapphires of the " alexandrite type ". They are recognized by their colouring and by a distinctive absorption line in the blue near 4750 A. Corundum is a uniaxial crystal of trigonal symmetry, with two molecules per unit cell.The aluminium ions are situated on the three-fold axis, and are co- ordinated to six oxygen ions which are at the corners of a deformed octahedron. Tanabe and Sugano 192 have calculated energy level diagrams for configurations d" in an octahedral field. These results form a useful starting-point for con- sidering the spectra due to impurities in the corundum lattice, and Tanabe and Sugano3 have studied ruby in some detail. However, vanadium corundum is a simpler case for study, since the V3+ ion has only two 3d electrons as compared with Cr3+ which has a d3 configuration. Low 4 has reported some absorption data for AlzO3(V+). The present paper includes high-resolution and polarization results together with a theoretical analysis.Before presenting experimental results it is desirable to summarize the results of the calculations for a regular octahedron. A single d-electron can either occupy a triply degenerate level t2, energy - 0.4 A, or a doubly degenerate level e, energy 0.6 A, where the difference in energy, A, is a measure of the octahedral field strength. The configuration d2 in this notation consists cf configurations t22, t2e and ez. The states of configuration t22 are 3T1, 1T2, 1E and 1A1 in increasing order of energy (fig. 1). The energy levels associated with t2e are also shown in this figure; but the levels associated with e2, namely, 3A2, 1E and 1A1 are at higher energies and are not included. The states are labelled according to the representations of the cubic group, as described for instance in reviews by Moffitt and Ballhausen and Runcirnan.4 * Department of Physics, University of Bristol.t Atomic Energy Research Establishment, Harwell, Didcot, Berks. Snow at the Department of Physics, University of Canterbury, Christchurch, New Zealand. 34M . H L . PRYCE AND W. A. RUNClMAN 35 A trigonal crystal field removes some of the degeneracy and splits the lowest cubic state 3T1 into an upper 3E state and the ground state 3A2, now labelling states, in italics, according to point group C3w. Furthermore, when the spin-orbit inter- action is included, the ground state 3A2 is split by a second-order effect into a doublet and singlet (fig. 2), the final degeneracy only being removed under the action of a magnetic field. This is the same situation as Abragam and Pryce 1 discussed for vanadium ammonium alum.3c 2c IC C FIG. l.-Energy levels of d2 in a trigonal field. Energy Units = lo3 cm-l. m=kl I I m=O FIG. 2.-The splitting of the ground state of vanadium corundum. The absorption spectrum in the visible and infra-red corresponds to transitions from the ground state 3 4 to excited states of the configuration d2. These transi- tions would be parity forbidden for the free ion, but are allowed in the crystal as the sites of the vanadium ions do not possess centres of symmetry. The site symmetry for the vanadium ions is C3, and the pseudo-symmetry C3w is used for labelling states for reasons stated in 3 3. However, for a discussion of the polar- ization rules it will be advisable to ignore the suffixes and to label states as simply A or E, since the intensity of the transitions depends on configuration interaction36 VANADIUM CORUNDUM resulting from odd-parity terms in the crystal field.These odd-parity terms are not needed for the energy level calculations, where the pseudo-symmetry C30 can be used. The spin prefix is dropped when allowing spin-orbit interaction to have mixed different spin states. Since the spin-orbit interaction is small, we may consider the spin states as nearly pure, and the most intense transitions are those involving no change in spin, i.e. triplet-triplet transitions. This is found experi- mentally and it is fortunate that in this case these transitions do not mask the weaker and sharper triplet-singlet transitions in the visible and near infra-red spectral regions.Polarization spectra are denoted T when the electric vector is parallel to the crystal trigonal axis, and u when the electric vector is perpendicular to the crystal axis. The selection rules for the electric-dipole absorption allow transitions A -+ A and E -+ E in T polarization, and allow transitions A -+ E, E -+ A and E --f E in D polarization. In those cases in which the components of absorption from the two lowest states are not resolved it is dificult to draw any definite con- clusions from the relative intensities in the two polarization spectra. 2. EXPERIMENTAL AND RESULTS The crystal was immersed in the cooling liquid which was contained in a glass Dewar with a transparent tip.For liquid helium (4-2°K) a double Dewar was used, the outer containing liquid nitrogen. The source used was a 6 V, 108 W ribbon filament lamp which was focused on the crystal. The spectra were obtained either on the glass optics of a Hilger large quartz and glass spectrograph or on 21 ft. grating spectrographs. (a) INFRA-RED SPECTRA These spectra were taken on the glass spectrograph using liquid nitrogen (77°K) as coolant. The Kodak Z plates were sensitized in 4 % ammonia solution, and dried after a rinse in methyl alcohol. Two absorption lines were found at 8770 and 966Ocm-1 and can be attributed to transitions to two of the three levels arising from the cubic states IT2 and 1E. (6) VISIBLE SPECTRA Spectra were obtained on the glass spectrograph with the crystal at 77°K and 4-2°K (fig.3, plate 1). The light direction was perpendicular to the crystal axis, and polariza- tion was obtained using a simple Polaroid filter. There is a broad absorption band covering the orange-yellow region and at 77°K a vibrational structure is resolved. There are bands at about 15890, 16060, 16250, 16420 and 1664Ocm-1. Owing to the large background absorption it is difficult to find the maxima accurately. The narrow band at 1589Ocm-1 is no doubt due to an electronic transition from the ground state to 3T2. Within the limits of accuracy the other terms may be ascribed to a simple vibrational series with Av w 180 cm-1. This is rather smaller than the frequency difference, Av rn 215 cm-1, found in ruby by Grechusnikov and Feofilov.8 This broad absorption band with its associated vibrational structure is not strongly polarized, but is slightly more intense in the o spectrum (fig.3, plate 1). At 4~2°K or 20°K the narrow band splits into two lines about 10 cm-1 apart. It was found difficult to measure the wavelengths of these lines accu ately as they are broad nd show signs of further partially resolved structure. These lines are polarized, the 15880 cm-1 line being mainly T polarization and the 15890 cm-1 line, which is stronger, being u polar- ized, This suggests that the upper state is an E state. It has long been known that there is an absorption line in vanadium corundum near 21000 cm-1. It is now found that this line is clearly resolved into two components at 77°K. The wavelengths of these lines were accurately determined on the 21-ft.plane grating spectrograph at Harwell and found to be 4758-2 and 4756.4 A, or as wavenumbers 21017 and 21025 cm-1 respectively. The line at 21017 cm-1 is u polarized and only half as strong as the 21025 cm-1 line which is T polarized. The polarization selection rules clearly indicate that these are transitions from the two components of the 3A2 ground state to the 'A1 state. Both lines showed satellites spaced at 1 or 2 cm-1 from the central component. These did not appear to be grating " ghosts " and are not fully understood(0) FIG. 3.-Plate 1 : absorption spectra of vanadium corundum (a) at 77"K, (b) at 4.2"K, (c) spectrum at 4.2-K, (ri) spectrum at 4-2 K. 2 l02,5c tn- ' 77°K 4 *2"Y21017 2IO25cm" I t FIG.4.-Plate 1 : (a) absorption lines at 21020 cni-1, without magnetic field ; (h) with a field of 9400 gauss parallel to the crystal axis. FIG. 3a, h and fig. 4a, b also include iron arc comparison spectra.M. H. L. PRYCE A N D W. A . RUNCIMAN 37 -they may well be due to vanadium ions which are close to another vanadium ion, for which the crystal field is slightly different. The only other feature in the visible region is a poorly resolved vibrational structure at the long-wavelength edge of a strong broad absorption band. At 77°K the first two maxima were estimated to be at 22370 and 22570 cm-1, and it is probably significant that the frequency difference Av M 200 cm-1 is not far different from that found for the other vibrational structure in the orange-yellow. This broad blue absorption band is ascribed to transitions to the 3T1 state and is stronger in T polarization.(c) ZEEMAN EXPERIMENT The line at 4758.2 8, is narrow for an absorption line in a solid, having a half-width of 0.4 8, (2 cm-1) at 77°K. In view of its interpretation it is expected to show a Zeeman effect, since it corresponds to a transition from a doubly degenerate lower level (see fig. 2) to a single upper level, and the degeneracy of the lower level is expected to be removed by a magnetic field. In a magnetic field H the energies of the three components of the ground triplet 3A2 are given by the eigenvalues of the spin-Hamiltonian, ( S = 1, D = 8 cm-1, 811 and gl roughly equal to the " spin-only " value 2, = Bohr magneton). Formulae for 811, gl and D are given by Abragam and Pryce.7 If the mag- netic field is parallel to the axis, the doubly degenerate level is split by 2gllPH.Hence the splitting is almost four times the " normal " Zeeman splitting. In a perpendicular field there is only a quadratic Zeeman effect. And in intermediate directions 0 the linear part of the splitting is 2gliPH cos 8. In view of the large splitting expected, it was thought worthwhile to try to detect the effect even though the maximum field available was 9400 gauss at just over an inch gap. With HIIC a splitting was detected using the 21-ft. concave grating at Imperial College (fig. 4, plate 1) ; but with H I C no change in the line was found. The splitting in the former case was that expected for a g value slightly less than 2.As expected, the line at 4756.4 8, remained unsplit with the field in either direction. 3. CRYSTAL FIELD CALCULATIONS Only potential energy terms of even parity up to fourth order mix states within the configuration d2. Therefore when calculating the energy levels of the con- figuration d2 in a trigonal field which has symmetry C3, only the harmonics r2Y20, r2Y40 and r2Yqf3 enter. By rotation about the threefold axis we can choose our axes such that the effective potential has the symmetry C3v. Hence if we neglect spin-orbit interaction, states may be labelled by the total spin multiplicity, 2S + 1, which will either be 1 or 3 ; and by the representations of the point group C30, which are Al, A2 or E (doubly degenerate). The theory of the crystal field splittings has been earlier formulated (Abragam and Pryce7,9), but will be redeveloped here in a more convenient form for this problem.In particular the cubic field wave functions will be used as a basis instead of Russell-Saunders LS wave functions. The non-cubic part of the potential energy of the crystal field can be written for a suitable orientation of axes as Y = A2[yz + zx + XU] + A4[yz(y2 + z2 - 6x2) + zx(z2 + x2 - 6y2) Both the terms in brackets are the sum of three functions which transform like the cubic representation T2. Normally one chooses as wavefunctions for a d-electron in a cubic field the functions, + XY(X~ + ~2 - 6~2)].38 VANADIUM CORUNDUM I When considering a distortion acting + along the (111) direction, however, it is advantageous to use the following G '-17 4- m'crl linear combinations: C I 1 I > I, 4.I 7 Y 4 d3 I> fo = --(cot, + 02ty + tJ, t- = -(tx + ty + tz), to = -(&x + oty + fz), d3 r c w 1 I Hrc) d3 + G + dly where + T I Irc) w = exp ( 2 ~ i / 3 ) . q' ,+ The apparently perverse signs in t+ I> and e+ are chosen for reasons of uni- formity with the phase conventions for 2 spherical harmonics,lo and have a con- -- k siderable advantage when angular momenta have to be discussed. Next I we define the strength of the trigonal 0 field by two parameters v and v' defined by the following relations between the 17 matrix elements of the one-electron + states in the crystal field: I 4 8 2 k .- I? I c 0 ' I < and (t+ 1 V I e+> = (t- I V I e-> =v'. (2) All other matrix elements are zero.The subscripts in t+, to, t- refer to the values + 1, 0, - 1 for the crystal quantum number p, where p M (mod 3) if M is the component of total angular momentum along the trigonal axis. As a preliminary to forming two- electron states, it is useful to write down the detailed representation products of the cubic group representations E and T2 as in table 1. The interpretation of the table is best described by an example. The product E+ x T2+ is given as - (Tz- + iT1-)/42. This means that a (normalized) two-electron wavefunction which is the product of two normalized + G *- I? + + I c E?M. H. L. PRYCE A N D W. A . RUNCIMAN 39 one-electron functions, respectively of type E+ and T2+, is the sum of two parts, belonging to the irreducible representations T 2 and Ti.The numerical factors represent the normalization and relative phases between components of a given representation. The phases are naturally to some extent arbitrary, only relative phases within a given representation having significance. It is now possible to construct the cubic field states for d 2 out of linear com- binations of determinantal two-electron states. For example one of the states of t22 is 1 1Al = -{- I t+L) + J to&)) - ] t-j+)>, 2/3 where a bar above the symbol denotes a spin component m, = -3, whereas un- barred symbols refer to m, = ++. The antisymmetric combination of the elec- trons is automatically formed by regarding the product as the diagonal term of a determinant formed by permutation of the states. A more complicated example occurring in the configuration t;?e is the 3T2+ state which has the following three-sub-states, 1 m, = 1 : a {I toe+) - I t-e-)), The crystal field matrix elements are calculated by operating on the cubic field states, expressed as a linear combination of determinantal product states, with the operator of the trigonal potential field.This is accomplished using the definitions (1). This operator does not act on the spin states, which remain un- changed. For example, the calculation for the 1Al state (3) is as follows : The subscripts a and b here refer to the configurations t22 and t2e respectively. In the following table of the trigonal field matrix elements calculated in this way the subscript a always refers to the configuration t22 whereas b either refers to t2e or e2.There is only one state of symmetry 3A1 arising from 3T2 and its matrix element is ( 3 ~ 2 0 I V I 3 ~ 2 0 ) = - +v. Similarly the only state of symmetry 1A2 arises from IT1 and has the matrix element <IT10 I vl %o> = - 33. The other matrices are naturally more complicated. They are given in table 2. It will be noticed that some matrix elements are imaginary. These can be made real by a trivial change of phase of the basic states 3T2& and 1 T l f , but we have preferred to leave them as they stand, for reference in any future work where the spin-orbit coupling matrices might be important, and where changing the phases would lead to further confusion.40 VANADIUM CORUNDUM TABLE 2.-TRIGONAL FIELD MATRIX ELEMENTS lT20n 0 'TlO(-- j V ) 1 'Tzfa / -;v ' T 2 f a 42v' 0 - jV 1 d2v' j d2V' 0 0 ' E f b 0 0 - v' 'f iv' 0 The trigonal field matrix elements were then added to the corresponding matrix elements for the cubic field and the coulomb forces,l as given in table 3.The following trial values of the parameters were used, where B and C are the linear combinations of Slater integrals as used by Racah.11 B = 540 cm-1, A = 17500 cm-1, C = 2600 cm-1 v = 1200 cm-1, v' = 0. This value of v is consistent with a spin-orbit splitting constant A of about 70 cm-1 when inserted in the formula of ref. (7) for the ground state splitting D, known experimentally to be 8 cm-1. For the parameters assumed this formula reduces to D w (1*25)02/v.M . H . L . PRYCE AND W. A . RUNCIMAN TABLE CO COULOMB AND OCTAHEDRAL CRYSTAL FIELD MATRIX ELEMENTS 41 1T2a 'Tzb i?: ( - 2 d 3 B 8 B + 2 C + 2 A ) , 9B + 2C - 22/3B 'T1 (12B + 2C + The secular equations were solved on an electronic computer.The spin-orbit interaction was not included at this stage, since it can be treated adequately as a perturbation, being small. Its effect is to split the 3E states into three components roughly 100cm-1 apart (this is an order of magnitude figure), and the 3A state by a much smaller amount, since the effect here operates only in second order. The energy levels relative to the ground state are in cm-1: 3A1 : 16,740, 3A2 : 0, 23,680, 34,640, 3E : 1,200, 17,290, 24330, 1A2: 28,420, 1A1 : 10,150, 19,000, 26,480, 57,170, 1E : 8,730, 9,970, 26,900, 29,190, 44,260. It is not suggested that the parameters used are the best possible, and in par- ticular v' may be considerable. However, it was not thought necessary to refine the parameters in view of uncertainties about the method, to be discussed in the next section.The tentative assignments of the observed levels are also shown in fig. 1, the three highest levels being taken from LOW.^ 4. DISCUSSION The moderately sharp lines in the infra-red and in the visible near 21,020 cm-1 axe precisely what one might expect from rather simple considerations. These lines are due to transitions between different states of the configuration t22, and hence are unlikely to be affected very much by the crystal field, since the charge distribution is practically the same in both. Likewise one would expect broad bands for the triplet-triplet transitions, where there is a marked change in the charge distribution. The amount of detail at the head of the band at 15,790 cm-1 is rather sur- prising and so is the existence of a well-defined vibrational series.Under the action of a trigonal field with spin-orbit interaction the 3T2 state should split into six levels. No lines are found due to five of these levels; yet if the absorption lines due to these levels were sharp they should show on top of the broad ab- sorption band associated with transitions to the lowest level coupled with the excitation of some crystal vibrations. One possible solution of this dilemma is that Jab-Teller distortions are very important for the position of the energy levels. The lowest level being almost a pure " spin " triplet is not affected and shows a second-order splitting due to the trigonal field plus the spin-orbit interaction.However, the excited states may be influenced by strong Jahn-Teller distortion. Such distortions will strongly couple the electronic transitions to certain asymmetric crystal vibrations which may be42 VANADIUM CORUNDUM partially localized near the impurity ions. This could account for the existence of a vibrational series and also for the lack of sharp lines due to the other levels as all the excited levels of the 3T2 state, for example, would be mixed by the Jahn- Tellzr distortions. The theory of the static Jahn-Teller effect has been developed by Opik and Pryce,l2 and for a simple case the dynamical theory has been used to calculate the en5rgy levels resulting from the electronic vibronic interaction (Longuet-Higgins, Opik, Pryce and Sack 13).Independently Moffitt and Liehr 14 have tackled the same aspect of the dynamical theory with similar results. Un- fortunately, detailed calculations of the type required for vanadium corundum would be of extreme complexity. In fig. 1 a comparison is made between observed and calculated energy levels. In this figure a comparison is made with the narrow bands at the long wavelength edges of the broad absorption bands for 3T2 and 3T1. On the other hand, a theoretical analysis indicates that the position of the electronic energy level in the crystal field of the ground-state configuration of the lattice corresponds to the mean frequency of the absorption band.This mean frequency corresponds fairly closely in practice to the peak of the absorption, and it is therefore more nearly correct to compare calculated energies with peak absorption positions, as done by Low 4 and Tanabe and Sugano.l.2 In view of the uncertainties here and on account of the Jahn-Teller effects it has not been felt worthwhile to complete detailed calculations of the spin-orbit interaction for all the states of d2 in a crystal field. On the experimental side there is the possibility of obtaining further Zeeman splittings when much higher fields of the order of 30 kilogauss are available. It would also be useful if paramagnetic resonance could be used to examine the magnetic splitting of the doublet situated 8 cm-1 above the ground state, as much greater accuracy would be obtained. There may be difficulties associated with the fact that this would be a magnetic quadrupole transition. Further progress may also arise from investigations on related substances, such as ruby which share some of the features here discussed. We wish to thank Dr. E. T. Richards for help while using the 21-ft. plane grating spectrograph at Harwell and Dr. J. Anderson at Imperial College, London, for similar help with the Zeeman experiment. Also we thank Mr. P. D. Preston for assistance with the computation of the energy levels, and all those who extended the use of the facilities at their disposal. 1 Tanabe and Sugano, J. Physic. Soc. Japan, 1954a, 9, 753. 2 Tanabe and Sugano, J. Physic. SOC. Japan, 1954b, 9,766. 3 Tanabe and Sugano, J. Physic. Soc. Japan, 1957, 12, 556. 4 Low, 2. physic. Chem., 1957,13, 107. 5 Moffitt and Ballhausen, Ann. Rev. Physic. Chem., 1956, 7 , 107. 6 Runciman, Rep. Prog. Physics, 1958, 21, 30. 7 Abragam and Pryce, Proc. Roy. SOC. A, 1951a, 205, 135. * Grechushnikov and Feofilov, J. Expt. Theor. Phys. U.S.S.R., 1955, 29, 384 (trans. 9 Abragam and Pryce, Proc. Roy. Soc. A , 1951b, 206, 173. 10 Condon and Shortley, The Theory ofAtomic Spectra (Cambridge Univ., 1935). 11 Racah, Physic. Rev., 1942, 62, 438. 12 Opik and Pryce, P y c . Roy. SOC. A , 1957,238,425. 13 Longuet-Higgins, Opik, Pryce and Sack, Proc. Roy. SOC. A, 1958, 244, 1. 14 Moffitt and Liehr, Physic. Rev., 1957, 106, 1195. in Soviet Physics JETP, 1956, 2, 330).

ISSN:0366-9033    

DOI:10.1039/DF9582600034

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

THE LINE SPECTRA OF Cr3+ ION IN CRYSTALS BY S. SUGANO* AND Y. TANABET Received 9th June, 1958 The line spectra of ions of the transition elements in crystals offer many interesting problems, among which the electronic origin of the line spectra seems to be the first problem to be attacked. This problem is treated here for Cr3+ impurities in A1203 (ruby), and assignments are made for the characteristic doublets R1, R2 (- 14,400 cm-1) and BI, B2 (- 21,000 cm-1) of ruby. The difficulties for Cr alums are discussed in com- parison with ruby. 1 . INTRODUCTION It is well known that several ions of the iron-group elements show sharp ab- sorption lines in the visible region even in crystals. The lines are sometimes so sharp that the Zeeman effect can be observed. The problem of interpreting the line spectra is interesting for several reasons.The study of the band spectra has led to the understanding of the electronic struc- ture of these ions in crystals, but the line spectra will provide more information than the band spectra. For example, the analysis of the observed vibrational structures associated with the electronic transitions will provide knowledge of the coupling between the electronic motion and the lattice. The observation of the Zeeman effect would give information not only about the properties of the ion in the crystal itself but also about the nature of magnetic coupling between the ions, as with paramagnetic resonance. Among the ions that show line spectra, Cr3+ ion is most outstanding and the absorption lines of Cr3+ ion in A1203 (ruby) and in aquo-complexes (alums) have been the subject of much experimental work.The purpose of this paper is to report some of the results of the experimental and theoretical studies on the line spectra of ruby 1 and to discuss certain diffi- culties encountered in the interpretation of the lines in the spectra of Cr alums. Detailed analyses of the line spectra of the ions of the transition elements have not been made as far as we know, and the problem of interpreting the absorption lines of Cr3+ ion in ruby seems well worth pursuing. 2. EXPERIMENTAL RESULTS A single crystal of ruby, whose colour is due to Cr3+ impurities in an A1203 crystal, shows similar spectrum to that of Cr alum. The spectrum in the visible region consists of two bands (the band with its peak at - 18,000 cm-1, called the U band hereafter and that with its peak at - 25,000 cm-1 called the Y band) and several groups of lines (located at - 14,400 cm-1 and - 21,000 cm-1).Among the lines, R1 (14,418 cm-I), R2 (14,447 cm-I), B1 (20,993 cm-1) and B2 (21,068 cm-1) are conspicuous.2 The broad bands show remarkable optical anisotropy known as the dichroism of ruby.3 Measurement of this optical anisotropy has been carried out by Kuwabara ef al. According to their preliminary result, the ratio of the dipole strengths w is w(Ul1) : w(U1) : w( Y11) : w( Y l ) = 1 : 3 : 6 : 3. (2.1) Ul and Y l denote the absorptions of U and Y when the electric vector E of an incident light is perpendicular to the optical axis C3, and UII and Yll those when E is parallel to * Dept. of Physics, Faculty of Science, Tokyo University, Tokyo, Japan.t Dept. of Applied Physics, Faculty of Engineering, Tokyo University, Tokyo, Japan. 4344 LINE SPECTRA OF Cr3* C3. The peaks of UII and Yll are on the shorter wavelength side of the corresponding peaks of Ul and Yl. The separation amounts to - 500 cm-1 for both U and Y. R and B also show anisotropy.L 3 The ratio of the dipole strengths for R is w(R# : w(R+) : w(R!j) : w(R1) = 2 : 4: 10 : 4 : 7, w(B$ : w(B+) : w(Bg : w(B+) = 10 : 2 : 8. (2.2) (2.3) and that for B is o b c d 82 /& o b c d E l C 3 a b c d a c b (a) case I: HoIIC3 ElC3(cr) (a') c a s e I : H,IIC3, E l C J ( c ) J, 4.2"K -d %---+- I f a e f e f 3, e f 4.2OK (b) c a r e n : HoII C3 , E tIC>(?T) (b') cases.H,IIC3, EIIC~(?T) 2 2 2 2 9 9 6 2 2 b i 2 (c') transition - diagram (c) tra n si ti0 n -d i a q ra rn FIG. 1.-The Zeeman patterns of R and B. Patterns of B are drawn in a scale different from that used for R. Observed splittings are given in /3Ho unit. Patterns expected from the theoretical transition diagrams are also shown (vertical lines). The numbers in the transition diagram (c) (or (c')) give the probability of each transition in an arbitrary unit. The unit in (c') is not related to that in (c). An elaborate experiment of the Zeeman effect on the characteristic doublet R1 and R2 has been carried out by Lehman.4 However, there remain some doubts about his results. Measurements were made, therefore, again for R1 and R2, and newly made for B1 and B2 at 20°K and 4-2"K by one of us (S.S.) and Tsujikawa. The results are shown in fig. 1. The patterns in the case of HO I C3 are omitted to save space. The patterns in this case indicate that the gLs of the corresponding excited states are almost zero. They thus represent the Zeeman splitting of the ground quartet. The relative intensities of the Zeeman components in each pattern also confirm the assignments given below.S . SUGANO AND Y. TANABE 45 3. THEORETICAL ANALYSIS Cr3+ ion in A1203 is surrounded by a distorted octahedron of six oxygen ions.5 The situation is similar to that in aquo-complexes of alums, where Cr3+ is sur- rounded by six water molecules. There are, however, two differences: all the Cr3+ ions in ruby are subject to the trigonal field directed along the optic axis C3, which coincides with the direction of the three-fold axis of the surrounding octahedron, The Cr(H20)$+ octahedrons in the alums are also subject to the trigonal fields, but the directions of these fields are different for each octahedron in the unit cell. Furthermore, the trigonal field of ruby does not have a centre of symmetry around the Cr3+ ion, in contrast to that of alum.3.1. ASSIGNMENTS We are interested in the electronic origin of the absorption lines R and B. The assignments 123 4A2 --+ split components of t23 2E or 2T1 are expected for R and t23 4A2 + split components of 2T2 for B, as with Cr alum.6 They are not split by the first-order perturbation of the trigonal field or the spin-orbit interaction, as long as we consider the strong (cubic) field limit or the pure t23 configuration. In fact, they are not perturbed to the first order by any field of lower symmetry than cubic.7 This fact is characteristic of the half-filled shells and we suppose that this will be the cause of the sharpness of the lines of Cr3+ (@), Mn2+ (t23e2) and Ni2+ (t26e2).The second-order perturbation theory applied with the assumption of the strong field limit leads to the following results : the positions of the split levels measured from the centre of gravity and the corresponding eigenfunctions are : Let us first consider the splittings of these doublets 2E, (2T1) and 2T2. W(Z(2E)) = - h/2, (3.1) K is the matrix element of the trigonal field : K = (t2Xf I %ig 1 t2Xf) (3.3) u, and x, denote the wave functions that should be multiplied by the factor * If 3 denote Ms.and [ is the spin-orbit coupling parameter. exp (2md/3) under the trigonal rotation.46 LINE SPECTRA OF Cr3+ The anisotropy of the absorption lines can be correlated to the anisotropy of the bands, if one adopts the usual assumption that the intensities of the lines are " borrowed " from those of the bands. It is then found that the following assign- ments lead to the explanation of the anisotropy of the lines : R~ : 4~~ -+ E ~ E ) , R2 : 4A2 -+ 22(2E), ~ 1 : 4 ~ 2 + 2 2 ( 2 ~ ~ ) , B2 : 4A2 --+ EU(2TZ). The consideration of anisotropy only does not, (3.4) however, exclude the possi- bility that R1 and R2 might correspond to the transitions to the split components cf 2T1. Definite evidence in favour of the above assignment is provided by the observed values of the spectroscopic splitting factors.The analysis of the ob- served pattern (fig. 1) shows that the g-values of the relevant excited states are : 8 where k is the orbital reduction factor.9 while the theory predicts the following values : The theory also predicts the relative intensities of the Zeeman components. Predicted patterns are also given in fig. 1 for comparison. The origin of the characteristic doublets of ruby has thus been clarified. There remain, however, several unsolved problems. The lines that would correspond to the components of 27'1 and Eb(ZT2) are not yet identified. The origins of the many weak lines are also left unanswered. 3.2. INFORMATIONS OBTAINED FROM THE ANALYSES (a) The strength of the trigonal field K estimated from the splitting of the (b) The spin-orbit coupling parameter 5 as determined from the separation (c) The orbital reduction factor k as determined from g11(21(2T2)) is 0-6.(d) The observed 811 shifts of the states G(2E) andT(2E), f 0.4 can be ex- plained as the orbital contribution (the matrix of Lz has an off-diagonal element between 2E and neighbouring 2Tl). From the sign and the magnitude of this bands is - 350 cm-1. of R1 and R2 is 140 cm-1. (Cfree = 270 cm-1.)S . SUGANO A N D Y . TANABE 47 shift, 2T1 is expecteed to lie at the shorter wavelength side of 2E, separated ap- proximately by 700 cm-1. (e) The analysis of the observed Zeeman pattern shows that the magnitude 6 of the natural splitting of the ground quartet is 0.36 f 0.03 cm-1, Kramers doublet of spin i- 3/2 being lower than that of spin f 1/2.(Paramagnetic reson- ance gives I 6 I = 0.38 cm-1.10) The latter fact does not contradict with the sign of K,* since it is found that 6 depends rather on the anisotropy of the spin-orbit interaction than on the trigonal field. In fact, it is possible to explain the sign and the magnitude of 6, taking the anisotropy of the g value of the ground state into account (811 = 1.9840 f 0-0006, gJ. = 1.9867 f 0.0006 10). 4. DIFFICULTIES IN THE PROBLEM OF Cr ALUMS One is tempted to ascribe the lines of Cr alums to the same origin as ruby, since the Cr(H2O)63+ octahedron in the alum crystal is also subject to the trigonal field. In fact, it seems possible to explain the Zeeman pattern of the doublet (- 15,000 cm-1) of ammonium sulphate alum in the high-temperature form reported by Tsujikawa, Jacquinot and Couture 12 with the assumption that the doublet corresponds to the split component of 2E.The situation is, however, not so simple. Phase transition occurs in ammonium sulphate alum 13 and the low-temperature form has six strong lines at the position of the doublet of the high-temperature form. The appearance of six lines is difficult to understand, since the paramagnetic resonance spectrum indicates the presence of only two kinds of ions in the unit cell exposed to the trigonal fields of different strengths at low temperatures.14 The low-temperature form of potassium sulphate alum has three strong lines near 15,000 cm-1.15 Similar difficulty arises in the interpretation of the number.Furthermore, the splitting factors of these lines as reported by Spedding and Nutting 16 are almost isotropic and close to 2 for the observed three angles 8 = 35", 55" and go", 6 being the angle between HO and the direction of the trigonal field. This is another difficulty, since any field other than the trigonal field is not expected from the results of the paramagnetic resonance experiment. Still difficult is the interpretation of the single line of methylamine sulphate alum observed at 14,501 cm-1. It should only be mentioned that the appearance of a single line might be in some way connected with the fact that the paramagnetic resonance spectrum has rhombic symmetry in this case. (The presence of different kinds of ions is not reported.) These problems cannot be understood if one is concerned only with the elec- tronic problem as with ruby.In order to understand the observed number of lines, it would be necessary to solve the vibronic problem (the Jahn-Teller effect in the excited state) and get an exact knowledge of the intensity and width in the optical spectrum. The isotropy of the g-values observed in potassium alum might also be related to such a dynamical effect, The investigation of the line spectra would thus enable us to get deeper insight into the mechanism of electron- nuclear coupling in crystals. The authors would like to express their sincere thanks to Prof. M. Kotani for his encouragement during this work, and to Dr. I. Tsujikawa (Tohoku Uni- versity) for his discussions on Cr alums and for his collaboration in the experi- mental studies on ruby. *van Vleck gives the following expression for 6 assuming the isotropic spin-orbit coupling :I148 LINE SPECTRA OF Cr3+ 1 More detailed discussions will be published in the J. Physic. SOC. Japan. 2 Deutschbein, Ann. Physik, 1932, (3, 14, 712, 729 ; 1934, (5), 20, 828. 3 Thosar, Physic, Rev., 1941, 60, 616; J , Chem. Physics, 1942, 10, 246. 4Lehman, Ann. Physik, 1934, (3, 19, 99. 5 Wyckoff, Crystar Structures (Interscience Publishers, New York, 1948). 6 Tanabe and Sugano, J. Physic. Soc. Japan, 1954, 9, 753, 766. 7 Tanabe and Kamimura, J. Physic. SOC. Japan, 1958, 13, 394. 8 Tsujikawa and Sugano, J. Physic. SOC. Japan, 1958, 13, 220. 9 Stevens, Proc. Roy. Soc. A , 1953, 219, 542. 10 Manenkov and Prokhorov, J.E.T.P., 1955, 28,762. 11 Van Vleck, J. Chem. Physics, 1939, 7, 61. 12 Tsujikawa, Jacquinot and Couture, Conference de physique des basses temperatures 13 Kraus and Nutting, J. Chem. Physics, 1941, 9, 133. 14 Bowers and Owen, Reports Prog. Physics, 1955, 18, 304. 15 Spedding and Nutting, J. Chem. Physics, 1934, 2, 421. 16 Spedding and Nutting, J. Chem. Physics, 1935, 3, 369. (Paris, Septembre, 1955), p. 399.

ISSN:0366-9033    

DOI:10.1039/DF9582600043

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

THE ABSORPTION SPECTRA OF SOLID HYDRATED NICKEL SULPHATE BY HERMANN HARTMANN AND HEINZ MULLER Institut fur phys. Chemie Universitat, Frankfurt am Main Robert Mayer Strasse 11 Received 9th June, 1958 In order to test Hartmann and Furlani's quantum-mechanical calculations of the energy level diagram of the complex ion Ni(H20)2,+ for octahedral and tetragonal symmetry, the spectra of the crystalline hexa- and heptahydrates of nickel sulphate have been determined at room temperature and at - 205°C. The number and sequence of main and intercombination bands satisfy the requirements of the theory. Distortion of the octahedral symmetry of the complex ion is shown by the splitting of the main bands. The extent of distortion can be deduced from the relative displacement of the absorption curve, The spectrum suggests the occurrence of multiplet splitting of the triplet terms.Hartmann and Use's theory of the light absorption of complex ions1 has given good agreement with experimental results in a large number of cases 2 and was therefore applied to the Ni(HzO)z+ ion. The shape of the absorption curve for this ion is more complicated than those of the ions investigated before. In order to give a theoretical account of the fine structure of this spectrum, distortion of the octahedral symmetry of the ligand arrangement and intercombinations were taken into consideration. The possibility of multiplet splitting of levels was also discussed. The results of these quantum-mechanical calculations 3 have now been tested by spectroscopic measurements.4 The crystalline hexa- and heptahydrates of nickel sulphate were used.These are obtained by crystallization from hot solutions and from solutions at room temperature, respectively. In both hydrates the structural units are octahedral Ni(H20)e and tetrahedral SOa- ions but the heptahydrate contains in addition an isolated seventh water molecule which is responsible for more extensive dis- tortion of the octahedra and tetrahedra than is met in the hexahydrate. Large crystals (several centimetres long) of both salts were grown. Parallel-sided plates were obtained by careful grinding of the crystals, carbon tetrachloride being used as cooling liquid. The spectroscopic investigations covered the region 2400-9100 A. A Carl- Zeiss 4 2 4 quartz spectrograph was used for the ultra-violet region, and at longer wavelengths a Steinheil Universal-Spektrograph GH with three glass prisms and a focal length of 1.6 m was used.Infra-red-sensitized plates were employed above 6000A. In order to bring out the fine structure, the spectra were taken not only at room temperature but also at the lowest possible temperatures. By applying Simon's desorption technique it was possible to attain temperatures down to - 205" to - 207.5" and to maintain them for the duration of the exposures. All four spectra have the same general form. In particular, the main absorp- tion peaks occur at analogous positions in all four curves. As an example, the spectrum of the hexahydrate at room temperature is shown in fig. 1. There are two main absorption regions of the Ni(H20)3+ ion in the wavelength range in- vestigated.They are indicated by the numerals II and III. Another absorption region occurs at still longer wavelengths for which the experimental technique employed was unsuitable. As this infra-red band is also important to the dis- cussion of the theoretical results we have reserved for it the symbol I. In addition 4950 SPECTRA OF NICKEL SULPHATE several weaker absorption bands, designated by small letters, can be recognized in fig. 1. Lastly, the main bands exhibit fine structure, an enlargement of which is reproduced in fig. 2. The main components of the strong absorption which F~G. 1.-Absorption curve of [Ni(H20)6]S04 at 20°C. FIG. 2.-Fine structure of the main bands 2 5 0 0 0 26OOOcm" 350 300mm of the hexahydrate at 20°C.recur in all the spectra are indicated by capital letters. The main difference between the individual spectra is a largely regular displacement of the entire ab- sorption curve along the abscissa. The average shift of the main maxima is schematically shown in fig. 3. The wavelength difference of the two low-tem- perature spectra is remarkably slight (39 cm-I), that of the hexahydrate occurring at shorter wavelengths. At room temperature both spectra are displaced towards longer wavelengths, the spectrum of the heptahydrate being shifted more than twice as far (367 cm-1) as that of the hexahydrate (142 cm-1). The broad, strong maximum of the heptahydrate which occurs at room temperature in place of the partial bands C and D is also noteworthy.This band is also a characteristic feature of the spectrum of the ion in aqueous solution. The shape and position of this spectnun agree almost exactly with that of the heptahydrate at room temperature. The absorption curves described are of special interest when compared with Hartmann and Furlani's quantum-mechanical calculations of the splitting of energy levels in an electrostatic field of six ligands. The calculated energy levels are shown in fig. 4.H . HARTMANN AND H. MULLER 51 In an electrostatic field of symmetry Oh the 3Fground state of the Ni2+ ion is split into three components. For the absorption spectra the next higher 3P level must also be taken into account. There are thus three transitions from the new ground state whose energies are of the order of light quanta.Since transitions from the triplet ground state to other triplet levels are possible only in combination with suitable vibration transitions, the intensities of the light absorptions must be lower than that caused by pure electronic transitions. The spectrum of the Ni(H20)4+ ion does indeed contain three main bands whose intensities are of the order of magnitude found in analogous spectral transitions of structurally NiSO+* 6H20 405' 367cm-I FIG. 4.-Energy level diagram of the Ni(H20);+ ion. NiSO+: 7H10 -205' In addition to the transitions between the levels produced by splitting of the triplet terms one may also expect transitions from the triplet ground state to the levels produced by splitting of the singlet terms. Although such " intercombinations " between terms of different multiplicity are forbidden, they occur in larger atoms with a low transition probability. The intensities of intercombination bands should therefore be low.It seemed plausible to attribute the bulges at the shorter wavelength side of the main band II and on both sides of the main band III to intercombinations of this kind. The cal- culations showed in addition that most intercombination bands must lie so close to the broad main bands as to be superimposed on them. In order to confirm them experimentally it seemed admissible to take the spectra under high dis- persion at low temperatures. Unfortunately the structure of the band regions is so complicated that it cannot be resolved into Gaussian error curves. However, they are sdciently intense to be recognizable even in the immediate vicinity of the most intense maxima under these rather favourable experimental conditions.Fig. 5 is a schematic diagram of the positions of the observed maxima and of the calculated transitions. The weaker bands are distinguished from the stronger ones by being drawn as lines of half length. The labelling of the bands is the same as in the earlier diagrams. The number and sequence of the observed weaker bands agrees well with the intercombinations required by theory. In the spectrum of the hexahydrate all main bands are split. The octahedral symmetry of the ligand field is therefore disturbed. The occurrence of two components indicate a resultant D4h symmetry, in agreement with crystal struc- ture analysis.It is peculiar that the spectrum of the heptahydrate has the same form as that of the hexahydrate. However, according to the crystallographic measurements, the Ni(H20);+ octahedron is more strongly distorted. The spectra52 SPECTRA OF NICKEL SULPHATE at room temperature do not permit any definite conclusions about the extent of band splitting, and hence about the degree of distortion. According to theory the displacement of the bands to shorter wavelengths is, in addition to the above splitting, a measure of the distortion. The remarkably large displacement of the absorption curve for the heptahydrate, as compared with the hexahydrate, clearly indicate therefore that the distortion is greater in the heptahydrate. All the details of the absorption curves discussed so far can be well reconciled with the requirements of the theoretical treatment of the Ni(&,O)~+ ion.Only the occurrence of the maximum in the main band I1 is not indicated by the theory and cannot be accounted for on the basis of the energy level diagram derived from the ligand-field splitting. To explain it one could invoke multiplet-splitting due to the interaction of spin and orbital angular momenta. NiSO;6H10 +lao 11 u I , I Ill I I I NiS0;6H10 -205' 1 1, I I l l I I Ill I NiSO~7H10t181 I II I1 AB CDEa b FGC d C f CD Ea b FGc d 1 1 Cf b FGc d AB CDEa UiS0,+.7HLO -205O )I 11 CDEa b FGc v 10000 2 0 0 0 0 3 0 0 0 0 4 0 0 0 0 crni FIG. 5.-Comparison of calculated transitions with observed absorption maxima. The difference in the symmetry of the complex ion in the two hydrates, which has been deduced from X-ray measurements at room temperature, is not brought out in the shape of the spectra. Even the low-temperature spectra give no in- dication that a symmetry lower than D4h occurs in the heptahydrate.On the contrary, their remarkable agreement suggests that at this temperature, there can be hardly any difference in the type of distortion of the complex ion in the two hydrates. Conclusions about the extent of distortion may, however, be drawn from the relative displacement of the spectra. This is slight at - 205" (39 cm-1). At this temperature the deformation in the two hydrates is therefore little different and must be much smaller than in the hexahydrate at 18", i.e. considerably less than 1 %. The displacement of the absorption curves on warming the hydrates to room temperature is so great that it cannot be attributed to the usual thermal dilatation but must be due to distortion. It is entirely plausible that the influence of temperature should be greater for the more loosely constructed heptahydrate (density = 2-07) than for the hexahydrate (density = 2.24). The magnitudes of the displacements (hexahydrate : 142 cm-1, heptahydrate : 367 cm-1) agree well with the changes in ligand distances found by X-ray crystallography. These amount to 1 % for the hexahydrate and nearly 10 % for the heptahydrate. The spectroscopic investigations have confirmed all the requirements of the theory. As theoretically calculated energy-level differences can be unreliable owing to the approximations made, we would point out that in the present case, the discrepancies are, on the average, only about 5 % and always less than 10 %. 1 Ilse and Hartmann, 2. physik. Chem., 1951, 197, 239. 2 Hartmann and Schlafer, Angew. Chem., 1958,70, 155. 3 Furlani, 2. physik. Chem., 1957, 10, 291. 4 Heinz Miiller, DipZomarbeit (Frankfurt am Main, 1957).

ISSN:0366-9033    

DOI:10.1039/DF9582600049

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

EXPERIMENTS ON CHARGE TRANSFER AND EXCHANGE INTERACTIONS BY J. OWEN Clarendon Laboratory, Oxford Received 21st July, 1958 A brief review is given of recent experiments concerning exchange interactions between paramagnetic transition group ions in crystals by the mechanism of electron transfer to intervening diamagnetic ions (superexchange). The approximate amount of electron transfer can be measured in favourable cases by both electron and nuclear magnetic resonance techniques. The exchange interaction can also be measured by electron magnetic resonance. Some examples are discussed, particularly chloroiridate salts. The purpose of this report is to summarize some recent developments in the application of magnetic resonance techniques to investigate exchange interactions between paramagnetic transition-group ions in crystals.One experiment relevant to this problem is the use of electron magnetic resonance to measure unpaired electron transfer between the paramagnetic ions and the adjacent ligands. This type of measurement was described 1 at the 1955 Faraday Discussion on Micro- wave Spectroscopy. An important advance since then has been the demonstra- tion that nuclear magnetic resonance from the ligand nuclei can also give a measure of the electron transfer. This was reported by Shulman and Jaccarino in 1956.2 A third type of experiment is the application of electron magnetic resonance to measure the superexchange interaction between neighbouring pairs of paramagnetic ions, which occurs because of the transfer of unpaired electrons to the intervening ligands.Measurements of this kind have been recently reported by Griffiths, Owen, Park and Partridge, 1957.3 This type of exchange mechanism was origin- ally suggested by Kramers in 1934,4 and is now thought to be the main origin of the exchange coupling in many antiferromagnetic transition group salts. To illustrate the experiments to be discussed, we consider first a simple system in which the paramagnetic ions M are surrounded by the ligands X, and each ligand is shared between two paramagnetic ions. The system can be represented diagrammatically by M-X-M. In the simplest case the energy of this system in an applied magnetic field H can be described by the Hamiltonian The first term represents the energy of the unpaired electron spins, S1 and S2, belonging to the M ions in the applied field, where is the Bohr magneton and g is the spectroscopic splitting factor.The second term represents the correspond- ing energy of the nuclear spin I belonging to the ligand X, where p~ is the nuclear magneton and gN the nuclear g factor. The nuclear spin of the M nucleus is neglected. The third term in A represents the hyperfine coupling between the electron spins S1 and Sz and the nuclear spin I ; this is only appreciable if there is appreciable unpaired electron transfer between M and X. The last term in J represents the exchange coupling between S1 and S;! which also depends on the electron transfer. The object of the magnetic resonance experiments described below is to measure A and J. 2' = gPH. (Sl + S2) - gj@NH.I + A1 . (S1 + SZ) + JS1. S2. (1) HYPERFINE STRUCTURE FROM LIGAND NUCLEI The technique is to grow magnetically dilute crystals in which most of the paramagnetic M ions are replaced by diamagnetic ions, so that one of the electrons 5354 CHARGE TRANSFER spins S1, S2 in eqn. (1) can be assumed to be zero, and the term in J is eliminated. The allowed electron resonance transitions corresponding to (1) then occur in a field H given by (2) where hv is the quantum of microwave energy, I, is the nuclear magnetic quantum number taking values I, I - 1, . . ., - I, and it is assumed, as is usually the case in practice, that gPH > A . Each electron resonance absorption line is thus split into 21 + 1 hyperfine components by each ligand nucleus contributing to the structure, and the magnetic field spacing between adjacent components is A/g/3.The M-X charge transfer is approximately related to A by A = p l o y where p is the probability of finding an unpaired electron in particular orbit on X, and A0 is the hyperfine structure constant due to an electron occupying this orbit in a free X atom. A shows symmetry properties corresponding to the type of ligand orbits which are occupied. Well-known examples of such hypefine structures are in (NlI4)2Ir, PtC16 where the magnetic complex is [IrC16]2-, and in Mn, Zn, F2 where the complex is [MnF6]4-. In the first case, there is 5 a very anisotropic structure which cor- responds to approximately 5 % transfer of unpaired electron spin from Ir to pn orbits on each C1. In the second case6 the principal contribution to the structure is isotropic, corresponding to the occupation of s orbits on F atoms which arises from o-bonding, and the total spin transfer to each F atom is estim- ated to be about 2 %.Further examples and a more detailed analysis are given by Hayes in a paper at this Discussion. Recent improvements in the resolution of the spectrum by using double resonance techniques have shown7 that it is also possible to measure the hyperfine structure due to second nearest neighbours from a magnetic centre. hv = gPH + AIz, SHIFT IN NUCLEAR RESONANCE FROM LIGANDS The second type of resonance experiment for detecting charge-transfer effects involves the shift in the nuclear magnetic resonance line from the ligand nuclei which arises from the hyperfine coupling discussed above. This shift has been recently found by Shulman and Jaccarino 2 in the F19 nuclear resonance in MnF2, and has since been observed in several other systems. The analysis given by Bleaneys and these authors2 is briefly outlined below.* The allowed nuclear resonance transitions corresponding to eqn.(1) are analogous to eqn. (2) and occur in a field H given by The sum 2 is taken over the number of neighbouring paramagnetic ions M whose electronic spins couple with the X nucleus ; this would be just two for our simple M-X-M system. The average value of the projection of S along the direc- tion of H must be taken because the electron spins change their orientations rapidly compared with nuclei. This average value is related to the total magnetic moment M of the N paramagnetic ions in the sample by M = XH = - Ng& where X is the paramagnetic susceptibility. Eqn.(3) can thus be written in terms of an effec- tive field He acting on the X nuclei in the form hv = gNpN(H He), (4) where He = [XH/(NgPg&)] Ai. The fractional shift in the resonance line is thus O! = He/H. * This shift and its analysis are very similar to the well-known Knight-shift in nuclear resonance from metals. It is also similar to nuclear resonance shifts arising from dipolar interactions.22J. OWEN 55 The experimental results are found2 to be in agreement with this analysis. The shifts are proportional to X and lead to values of the hyperfine constant A reasonably close to the values measured directly by Tinkham6 in Mn, ZnF2 as discussed above.They therefore lead to the same estimate of approximately 2 % unpaired electron on each F atom. The actual magnitude of the shift in MnF2 at 77°K for example, is about 7.5 %, and there is a smaller superimposed angular variation corresponding to dipolar interactions and to the fact that p - orbitals, as well as s-orbitals, on the F atoms are partially occupied by unpaired electrons. Similar nuclear resonance shifts in paramagnetic FeF2 and CoF2 have been measured by Baker and Hayes.9 All of these fluoride salts have strong M-F-M exchange interactions and go antiferromagnetic at low temperatures. Under these conditions Shulman and Jaccarino and co-workers 10 find the shifts to be very large since they are now proportional to the sublattice spontaneous magnet- ization.Further recent applications of the nuclear resonance shift technique have been to various transition-group ion-complexes in solution,ll and to nuclei which are second nearest neighbours from paramagnetic ions in crystals.12 MEASUREMENT OF EXCHANGE INTERACTION We now consider the exchange term JS1. S2 in eqn. (1). This represents the coupling between the paramagnetic ions M-M which results from spin transfer on to the intermediate ligand X. This superexchange mechanism has been dis- cussed in particular detail for pairs of the form Mn-0-Mn in the antiferro- magnetic MnO. Anderson 13 gave the first such detailed analysis, and a recent paper which includes a brief review of the present theoretical situation is by Yamashita and Kondo.14 This type of exchange is well illustrated by the electron magnetic resonance investigations by Grfiths, Owen, Park and Partridge 3 of (w4)2bc16 and K2IrC16.The magnetic complex is [bC16l2-, and each Ir4+ ion has one unpaired electron with spin S = 3. The structure of a nearest neighbour pair of Ir ions can be represented diagrammatically by Ir \Ir, and a detailed theoretical analysis of the Ir-Ir exchange interaction for such a pair has been made by Judd.15 For the present purpose we outline a simple order of magnitude estimate of the expected size of the exchange given by Griffiths et al. The I r 4 1 spin transfer already mentioned above can be pictured as a migration for about p = 5 % of the time of an unpaired electron from a Cl- ion to the Ir4+ ion, thus cancelling the unpaired spin on the Ir and leaving a neutral C1 atom.There is thus a probability of order 2p2 = 2 x (0.05)2 of finding an adjacent pair of Cl atoms in the ,Cl-Cl \Cl-Cl/ Ir<cl-cl\Ir structure, which looks something like a stretched out C12 molecule, c1-Cd Using a Morse-curve model, Griffiths et al. estimate that the separation & be- tween the ground-state singlet (spins antiparallel) and the first excited triplet (spins parallel) for such a " molecule " is of order 2000 cm-1. The order of magnitude of the singlet-triplet separation for the Ir-Ir pair, that is to say the size of the exchange interaction J, is then expected to be J - 2p2& - 10 cm-1. This value is perhaps in better agreement with the experimental results (table 1) than might have been expected considering the crudeness of the model.The singlet-triplet separation is found experimentally in the following way. Magnetically dilute mixed crystals of (NH4)2Ir, PtCl6 are grown with Ir : Pt M 1 : 20. There are then an appreciable number of nearest neighbour pairs of Ir ions which have no other Ir neighbours. The energy of the pair in a magnetic56 CHARGE TRANSFER TABLE 1 m 4 ) 2 I f l 1 6 K2IrCLj ref. spin transfer to each Cl, % 5 5 (5) bour Ir-Ir pair, J/k in OK 7.5 f 1 11.5 f 1 (3) (Weiss constant)/3 = 8/3 in OK 6.7 f 1 10-7 f 1.3 (21) Nkel temperature, Tn in OK 2.1 3.1 (21) isotropic exchange for nearest neigh- field H is described by the Hamiltonian of eqn. (l), with S1 = S2 = 3. This can be rewritten in the form Z' = .S + (J/2)[S(S + 1) - 3/21, ( 5 ) where the small nuclear terms have been neglected, and S is the total spin which can take values 1 or 0. As anticipated above, the energy levels are then a triplet (S = 1) at J/4, J/4 & gpH and a singlet (S = 0) at - 3J/4. In practice, there is also found to be a small zero field splitting between the levels of the triplet cor- responding to the presence of anisotropic exchange in addition to the isotropic exchange J, but this is neglected here. The electron resonance spectrum cor- responds to transitions between the levels of the triplet. The intensity I of the absorption lines varies with temperature Taccording to IcX(l/T)[3 + exp (J/kT)]-l corresponding to the depopulation of the triplet levels into the singlet as Tis lowered.Measurement of I as a function of T gives the values of J in table 1. The positive sign indicates that the triplet lies above the singlet, i.e. that the exchange is anti- ferromagnetic. Previous applications of electron resonance techniques to investigate exchange interactions between transition group ions includes the work of Bagguley and Griffiths,l6 and Pryce 17 on copper sulphate, and that of Bleaney and Bowers 18 on copper acetate. More general accounts are given in ref. (19) and (20). The results on the chloroiridate salts are of particular interest because the directly measured exchange for a nearest neighbour pair of Ir ions in the magnetic- ally dilute salt, can be compared with the bulk susceptibility of the concentrated salts where each Ir ion has twelve nearest neighbours in the face-centred cubic lattice.The high temperature susceptibility is then expected 13923 to follow a Curie-Weiss law X = C/(T + 8) with 8 = 3J/k. The experimental values of the Weiss constant 8 measured by Cooke et aZ.21 are in reasonable agreement with this relation as can be seen from the results in table 1. The observed antiferro- magnetic transition or Ndel temperatures T', also given in the table, are not yet fully understood. They may depend on the anisotropic part of the nearest neighbour exchange or on the exchange between next-nearest neighbour Ir ions. In conclusion, it may be said that the application of magnetic resonance techniques to the types of problem outlined in this report seem likely to lead to a more detailed understanding both of the mechanism of exchange interactions and of the bulk magnetic properties of transition group salts.The author would like to thank the General Electric Research Laboratory, Schenectady, for support. 1 Owen, Faraday SOC. Discussions, 1955, 19, 127. Tinkham, Faraday SOC. Discus- 2 Shulman and Jaccarino, Physic. Rev., 1956,103,1126 ; Physic. Rev., 1957,108,1219, 3 Griffiths, Owen, Park and Partridge, Physic. Rev., 1957, 108, 1345, and submitted 4 Kramers, Physica, 1934, 1, 182. 5 Stevens, Proc. Roy. SOC. A , 1953,219, 542. Grfiths and Owen, Proc. Roy. SOC. A , sions, 1955, 19, 174. for publication. 1954, 226, 96.J. OWEN 57 6 Tinkham, Proc. Roy. SOC. A , 1956,236, 535, 549. 7 Feher, Physic. Rev., 1957, 105, 1122. 8 Bleaney, Physic. Rev., 1956, 104, 1190. 9 Baker and Hayes, Physic. Rev., 1957, 106, 603. 10 Jaccarino and Shulman, Physic. Rev., 1957, 107, 1196. Jaccarino, Shulman, Davis and Stout, Bull. Amer. Physic. SOC., 1958, 3, 41. 11 Shulman, Bull. Amer. Physic. SOC., 1958, 3, 262. Bloembergen, J. Chem. Physics, 1957, 27, 595. 12 Mays, Physic. Rev., 1957, 108, 1090. 13 Anderson, Physic. Rev., 1950, 79, 350, 705. 14 Yamashita and Kondo, Physic. Rev., 1958, 109, 730. 15 Judd, submitted for publication. 16 Bagguley and Griffiths, Nature, 1948, 162, 538. 17 Pryce, Nature, 1948, 162, 539. 18 Bleaney and Bowers, Proc. Roy. SOC. A, 1952, 214? 451. 19 Baker and Bleaney, Conference on the Physics of Low Temperatures (Paris, Institut 20 Bagguley and Owen, Report Prog. Physics, 1957, 20, 304. 21 Cooke, Lazenby, McKim, Owen and Wolf, submitted for publication. 22 Bloembergen, Physica, 1950, 16,95. 23 van Vleck, J. Chem. Physics, 1941, 9, 85. International du Froid, 1955). Poulis and Hardeman, Physica, 1952,18,201 ; 1953, 19, 391.

ISSN:0366-9033    

DOI:10.1039/DF9582600053

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

THE BONDING AND VALENCE PROPERTIES OF IRON GROUP IMPURITIES IN NaF BY W. HAYES The Clarendon Laboratory, University of Oxford Received 30th June, 1958 The paramagnetic resonance spectra of iron group ions present as trace impurities in NaF have been investigated and the ions Cr+, Mn2+, Fe+, Co2+, Co+ and Ni+ have been observed. The monovalent ions are produced by irradiating a crystal containing the corresponding divalent impurity with prays, X-rays or electrons and are stable to about 140°C. Each impurity ion forms a weak covalent complex with the surrounding octa- hedron of fluorine ions resulting in a fluorine hyperfine structure in the resonance spectrum. The bonds are of the o-type and for Cr+, Mn2+ and Fe+ the amount of electron transfer is estimated. The properties of a crystalline lattice are considerably affected by the presence of impurity ions and these effects have been studied in a number of ways.The luminescence of many phosphors is associated with the presence of impurities and in some cases these are paramagnetic, for example ZnS and CdS activated with copper or manganese ions. Other phosphors of interest are the alkali halides activated with thallium and calcium fluoride with rare earths. The properties of impurities in these crystals have been investigated through their optical ab- sorption spectra, luminescence spectra, paramagnetic resonance spectra, dielectric loss and electrical conductivity. Interesting changes in the properties of crystals containing impurities may occur upon irradiation with ionizing radiations and the investigation of effects of this type has been largely confined to alkali halides.In these crystals, well-defined colour centres are found which are associated with the presence of impurity ions.1 Many impurities are effective electron traps and irradiation under suitable conditions may produce a metastable valence state. In general, an impurity ion will distort the host lattice to some extent; the degree of distortion will depend on its valence, bonding characteristics and size and will determine the ease with which the impurity is incorporated into the lattice. It is of interest to know whether the impurity ion is present substitution- ally or interstitially. In cases where the charge on the impurity cations and the host cations are not the same, charge compensation must occur since the crystal as a whole must remain neutral and the mechanism by which this is achieved is also of interest.When the impurity ion is paramagnetic and the electron resonance spectrum is observable it is sometimes possible to obtain precise answers to these questions. The nature of the resonance spectrum gives information about the spectroscopic state of the ion and the symmetry and strength of the crystalline electric field in which it is placed. With NaF a fluorine hyperfine structure is observed in the resonance spectra of iron group impurity ions due to the formation of a weak covalent complex with the six nearest neighbour fluorine ions ; this structure arises from the interaction of the magnetic electrons with fluorine nuclei through bond formation.Analysis of the fluorine h.f.s. enables one to estimate the per- centage covalent character of the bonds and gives aztailed information about the environment of the impurity ion. In the preseni paper we shall be concerned mainly with structures of this kind and with the cbange of valence of the impurity ions which occurs upon irradiation of the crystals. 58W. HAYES 59 EXPERIMENTAL The crystals of NaF used in these experiments were grown by the Stockbarger tech- nique2 with known quantities of impurity added. An “Extra Pure” grade of NaF produced by B.D.H. Ltd. was used. The appropriate amount of salt containing the element required as impurity was mixed with NaF powder before loading into the graphite crucible. The type of salt used and the molar concentration of impurity added are given in table 1.TABLE 1 salt added to molar concentration of molar concentration of the melt impurity added impurity observed CrCl3-nH20 -01 % Cr -003 % MnF2 a 0 1 % Mn -002 % FeS04-7H20 -01 % Fe -0001 % CoC12.6H20 1.0 % c o *0004 % NiC12-6H20 -02 % Ni 402 % The measurements were made at 9200 Mc/sec with a paramagnetic resonance spectro- meter of the type described by Llewellyn.3 A high sensitivity was achieved by modulating the external magnetic field at 465 Kc/sec. Estimates of the concentration of impurity present in the crystal were made from the strength of the resonance signal and these are given in table 1 ; the error in the estimates amounts to about a factor of five. The irradiation of the NaF crystals was performed with (a) 6OCo y-rays at 40°C, or (b) 150 kV X-rays at room temperature, or (c) about 5 pA of 1 MeV electrons for 2 min at room temperature.The observed effects produced by each of these irradiations were substantially the same. THE INTERACTION OF THE MAGNETIC IONS WITH THE LATTICE A brief summary of the work on iron group impurities in NaF has already been given.% 5 NaF has the cubic NaCl type structure and each sodium ion is surrounded by a regular octahedron of fluorine ions. When singly charged impurity cations enter the lattice substitutionally the octahedral cubic symmetry will in general be maintained since vacancies are not introduced. Consider, however, a solid solution of MnF2 in NaF in which Mn2+ ions occupy sites normally occupied by Naf ions.In order to conserve charge the crystal will in general contain one cation vacancy for each Mn2+ ion incor- porated into the lattice. Vacancies are expected to occur near doubly charged impurity ions because of the electrostatic attraction between them. The degree of association of the vacancy-impurity complex appears to depend on the method of crystal growth and the concentration of impurity added. A vacancy in positions P or Q (fig. 1) will introduce an axial component about a cube edge into the crystal field seen by the impurity ion at A ; the lower symmetry will be apparent in the resonance spectrum and three distinguish- able magnetic complexes will be observed since there are three distinguishable cube edges. A vacancy in position R or S will introduce rhombic symmetry with principal axes along two face diagonals and a cube edge and six distinguishable magnetic centres will be found.Charge compensation for doubly charged impurity cations may also be achieved by introducing doubly charged impurity anions, for example, 02-, S2-. An example of such a system is described by Watkins 6 who suggests that charge compensation for Mn2+ ions in NaCl results from the presence of a doubly charged impurity anion at T (fig. 1). With NaF a check on this type of system is possible since the number of nearest neighbour fluorine ions may be determined from an analysis of the fluorine h.f.s. The first detailed description of extra-nuclear h.f.s. in the spectrum of a paramagnetic ion was given for Ir4+(5&) in the complex FrC16]2-.7, * , 9 Subsequently a fluorine h.f.s. was observed in the spectra of iron group (34 impurity ions in ZnF2 by Tinkham 10 and in CaF2 by Baker, Bleaney and Hayes.11 The fluorine h.f.s.arises from a partial transfer of the unpaired d electrons to orbits on fluorine ions and its order of magnitude may be estimated from the product of the h.f.s. of the free fluorine atom and the probability of finding the d-electrons on fluorines. The following discussion of the bonding follows van Vleck,l2 Owen 9 and especially Tinkham.10 Consider a complex AX6 where A is an iron group ion (3d9 surrounded by a regular octahedron of fluorine ions X. If we consider the bonding to be perfectly ionic then the60 orbitals of the magnetic electrons on A are of the form dx2-y2, d322-r2, d,,, d,,, dZx.The first two orbitals (dy type) have a higher energy than the remaining three (de type) which form a ground state orbital triplet. For Mnz+(d5) each of these orbitals is singly occupied and the ground state (6s) has zero angular momentum. If we consider the mixing which occurs between the central dy orbitals on A and the 2s and 2p orbitals on X then the modified dy orbitals become IRON GROUP IMPURITIES IN NaF 03~2-r2 Nu{&2-r2 + [au/l2f](pfc - P: + P: - P: - 2~: + 2~ :))a oX2-,,2 = NU{dx2-y2 + 8.X- P: + P: + P; - ~:)}u. I + - 8 - + + + - + - + - + FIG. 1.-Methods by which charge compensation may be achieved for a doubly charged cation A in NaF. P, Q, R and S are positive ion vacancies ; T is a doubly charged anion. The numbers 1, 2, 3 and 4, 5, 6 refer to nearest neighbour fluorine positions on positive and negative x, y , z axes respectively.The p orbitals are made up of 2s and 2p orbital- on X and = a: + &. The mixing between de orbitals on A and pn orbitals on X will lead to augmented de orbitals of the form nxy = Nn{&y + P; - P;" + P; - P;)l*. The admixtures in this case are expected to be smaller than for a-bonds and do not con- tribute measurably to the bonding in the cases to be described. The interaction between the magnetic electrons and the fluorine nuclei may be expressed in the form where I = Q is the spin of the fluorine nucleus and the summation is over the six nearest neighbour fluorine ions. Fluorine h.f.s. components occur at energies displaced from the zero interaction position by N N where Bz, is the angle between the external magnetic field and the bond axis to the Nth fluorine ion.The term arises from the s electron contact interaction through the fluorine 2s orbital. When the orbital angular momentum is completely quenched and when, as for Cr+, Mn2+ and Fe+, both o-bonding orbitals are singly occupied this becomes In this expression p is the Bohr magneton, f l is the magnetic moment of the Nth fluorine nucleus, S is the true electronic spin and s(0) is the amplitude at the fluorine nucleus ofW. HAYES 61 the 2s fluorine orbital. Where the observed multiplicity of the ground state is 2 an effective spin S’ = 4 may be used to describe the resonance spectrum.13 In such cases a correction factor 2 < IS[ > = gs/2 must be introduced such that gs A: = A“.This correction factor is unity for Cr+, Mn2+ and Ni+ and 1.6 for Fe+ where S = 9. The term A: includes (a) the coupling A; with the Nth fluorine nucleus through the Pa orbital, and (b) the direct magnetic dipole interaction gn13,gp/r3 between the Nth fluorine nucleus and the electronic wave function located on the central ion ; g and g, are the electronic and nuclear g-factors and /3, is the nuclear magneton. Both interactions have the same angular dependence and in general are comparable in magnitude (g,flngfl/r3 = 2.0 x 10-4 cm-1 for g = 2). If the orbital angular momentum is completely quenched then the expression for Ap”. becomes (7) where Y is the radius of the 2p wavefunction. In the calculations which follow we assume n = 2 orbitals and take I s ( 0 ) 12 = 75 x 1024 cm-3 and <r-3 > = 38 x 1024 cm-3.14915 In cases where the quenching approximation is poor and the g-factor departs considerably from 2 then the interaction between the electronic orbital magnetic moment and the fluorine nuclear spin should be considered ; ignoring the effect will introduce errors of around 20 % into the calculation of the anisotropic bonding for Fe+ and Ni+ and it will be neglected.When the external magnetic field is in the (111) orientation, (3 cos2 BZ,, - 1) = 0, and the interaction with six equivalent fluorines will give a seven-line h.f.s. with intensities in the ratio 1 : 6 : 15 : 20 : 15 : 6 : 1 ; 4 the spacing of the lines gives a direct measurement of A,. The parameter A , may now be determined by examination of the spectra in other orientations.Since both a-bonding orbitals are occupied in the case of Cr+, Mn2+ and Fe+ the probability that the magnetic electrons will be found in a fluorine 2s orbital is +A’?: and in a 2pa orbital is *I$$,; values for these expressions are given in table 2. TABLE 2 probability of probability of occupation of a orbital orbital Cr + 12.8 f 0.2 0 f 1.5 -42 % - Mn2+ 14.4 f 0.3 0.8 rt 0.7 -47 % 0.7 % Fe + 28.3 f 1.0 6.0 f 1.0 -35 % 2.0 % Ni + type I fluorines 41 f 2 14 f 2 - - type I1 fluorines 10 f 1 O f 2 occupation of a AS $0 fluorine 2s fluorine 2p0 ion - - The values of A, and A,, are expressed in units of 10-4 cm-1. DESCRIPTION OF THE RESONANCE SPECTRA CHROMIUM Clear crystals of NaF to which Cr3f ions were added to the melt do not show a resonance down to 20°K; the absence of a resonance is not unexpected if chromium is present in the lattice as Cr2+.After irradiation, however, a single isotropic line with a resolved fluorine h.f.s. appears at g = 2.000. The crystals are now a pink colour which is associated with an optical absorption band at 500 mp and probably arises from M centres.1 The resonance spectrum is assigned to chromium since a hyperfine structure due to 53Cr ( I = 3/2, natural abundance 9.5 %) is observed. An analysis of the fluorine h.f.s. shows that the bonding is of the o-type and that the centre is present substitutionally and is surrounded by a complete octahedron of fluorine ions. A fine structure is observed in the spectrum characteristic of the effect of a cubic crystal field on a 6S ground term.All this suggests that the chromium is present as Cr+ which is isoelectronic with62 Mn2+ (3d5,6S). It appears that Cr2f ions are stable electron traps and chromium can exist in the NaF lattice at room temperature in the monovalent form. IRON GROUP IMPURITIES IN NaF MANGANESE Before irradiation a complex spectrum is observed due to Mn2+; there are two types of Mn2+ centre present, each presumably arising from a different degree of association of the vacancy-impurity complex. The stronger spectrum has axial symmetry about a cube edge and the fluorine h.f.s. indicates that the Mn2+ ion is on a cation site and is surrounded by a complete octahedron of fluorine ions. The weaker spectrum which contains about 1 % of the total number of centres has a principal axis of symmetry along a cube edge.The splitting of the fine structure is greater in this orientation than that observed in the stronger spectrum suggesting that the vacancy is on a closer lattice site in this case. An analysis of the fluorine h.f.s. in the weaker spectrum was not possible because of insufficient intensity. After irradiation the intensity of the Mn2+ spectrum is reduced by 90 % but no new resonances are observed down to 4°K. It may be that the irradiation induced reduction of intensity is due to the conversion of Mn2f ions to Mnf ions which are isoelectronic with Fe2+ (3d6, 5 0 ) . Mn+ transi- tions may not be observed in the presence of crystal field components of low symmetry. IRON When crystals of NaF to which Fe2+ ions had been added are irradiated and the resonance spectrum investigated at 20"K, a single line is observed with a resolved fluorine structure.The g-value (g = 4,344) may be compared with the g-value of Co2+- in MgO (g = 4.278).16 However, for Co2+ a large eight-line h,f.s. due to the nucleus of the cobalt ion is observed; in the present case no h.f.s. due to the nucleus of the magnetic centre is found. Since both NaF and MgO have the NaCl structure the results indicate that the irradiation induced centre is Fe+ which is isoelectronic with Co2+ (3d7). No resonance was observed before or after irradiation down to 4°K which could be attributed to Fez+. COBALT Before irradiation a complex spectrum due to Co2+ is observed at 20°K.There are six distinguishable magnetic ions with similar spectra. Each ion has rhombic symmetry with two face diagonals and a cube edge as principal axes. The fluorine h.f.s. was resolved but a detailed analysis was not possible. Irradiation removes completely the spectrum described above and two iso- tropic lines now appear. One line is observed at 20"K, has a g-value of 4.5 and is about 400 gauss wide. It appears that the centres involved are Co2+ ions in slightly distorted cubic surroundings ; the departure from cubic symmetry prob- ably arises from vacancies at distant lattice sites. The second line appears at 90°K with a g-value of 2-31 and has a flat top about 200 gauss long. The g-value is close to the g-value of Ni2+ in MgO (g = 2-22> 17 suggesting that the magnetic ion is Co+ which is isoelectronic with Ni2+ (d8).The absence of a resolved struc- ture on the line is due to the presence of cobalt h.f.s., fluorine h.f.s. and line broadening due to slight distortions of the crystal field. NICKEL After irradiation a spectrum is observed at 20°K ; there are three similar centres each with axial symmetry about a cube edge. The details of the spectrum are explained by assuming the centres are Ni+ ions which are isoelectronic with Cu2+ ions (d9). A cubic field leaves a " non-magnetic " orbital doublet (dxz-yz, d3zz-rz) lowest; the degeneracy is raised by a tetragonal distortion of the octahedron which in this case arises from a Jab-Teller effect 18 (cf. Bleaney, Bowers, Trenam 19). The g-values (811 = 2-766, gL = 2-114) indicate that the octahedron is extendedW.HAYES 63 along a cube edge. If the distortion of the octahedron is pure tetragonal then the single unpaired spin is expected to be in the dx2,,2 ground state and bonding only to the four fluorines (type I) in the plane perpendicular to the symmetry axis should occur. However, an appreciable bonding to the two fluorines (type II) on the symmetry axis is observed indicating an admixture of the d3z~-r2 level into the ~'z-,,z state. This may result from a rhombic distortion which is not observable however in the g-values. The charge transfer has not been calculated since the admixture coefficient is not known. No resonance is observed before or after irradiation which can be assigned to Ni2f. The spectrum of the Nif ion whose axis is directed along (001) is shown in fig.2 for the (110) orientation of the external magnetic field. The interaction with the 1.0 I I I I f = 9150 MC~EC- gauss 3000 3100 3 200 FIG. 2.-Chart recording of differential of absorption lines in the spectrum of Ni+ showing fluorine h.f.s. four type I fluorines splits the resonance line into 5 components with intensities in the ratio 1 : 4 : 6 : 4 : 1 ; a further splitting of each of these lines into 3 com- ponents with intensities in the ratio 1 : 2 : 1 is produced by the weaker interaction with the two type I1 fluorine ions. CONCLUSION Earlier investigations of the effect of divalent impurities (for example, Ca2+ and Sr2+) in alkali halides showed the presence of 21 optical absorption bands following irradiation.These bands were attributed by Seitz to electrons trapped on the divalent ions. It was further assumed that the positive ion vacancy which was initially associated with the divalent ion moves away, possibly to join the electron deficient centre left behind by the captured electron. In the present work the resonance spectra of two divalent ions, Co2+ and Mn2+ have been ob- served. In both cases the symmetry of the surroundings is less than cubic due presumably to association with positive ion vacancies. No spectra are observed which can be attributed to Cr2f, Fe2f or Ni2+ ions; vacancies associated with these ions may reduce the symmetry in such a way as to prevent observation of64 IRON GROUP IMPURITIES I N NaF a resonance. The observation of Cr+, Fe+ and Co+ ions with cubic symmetry indicates that vacancies which may have been originally associated with the di- valent ions have migrated.Some of the electrons released during the irradiation are trapped as F-centres and M-centres. No resonances have been observed which can be attributed to the electron deficient centres and the nature of these centres is unknown. The trapping efficiency of the divalent iron group ions may be partly due to the fact that the radii of the ions are less than that of Na+ and the strain energy of the lattice due to misfit of the impurity ion will be less in the monovalent form. The metastable monovalent ions are annealed out by heating to about 140°C. The analysis of the fluorine h.f.s. shows that the largest interaction comes from magnetic electrons in fluorine 2s orbitals.The smaller anisotropic contribution from 2p orbitals comes mainly from a-bonding ; the contribution from r-bonding is inside the error of measurement. These results may be compared with the work of Tinkham on iron group impurities in ZnF2,lo of Baker, Bleaney and Hayes on iron group impurities in CaF2,11 and of Shulman and Jaccarino on fluorine nuclear resonance shifts in MnF2.20 The measurements on Mn2+ are the most complete and the contact interactions are shown in table 3 ; included TABLE 3 Mn : ZnF MnF2 Mn : NaF Mn : CaF2 As 16.5 f 1.5 15.7 f 0.3 14.4 & 0.3 9.5 f 0-3 probability of occupation of a fluorine 2s orbital, % 5 3 -5 1 -47 -30 cation-anion distance, A 2.03 2.1 1 2.3 1 2.36 The values of A , are expressed in units of 10-4 cm-1.also are the amount of charge transfer to each fluorine 2s orbital and the separa- tions of the bonding ions. The bonding in Mn : ZnF2 and MnF2 should be similar since the systems are isomorphous; it is found that the contact interactions and o-bonding agree closely and for MnF2 it is estimated that the =-bonding is about 30 % of the o-bonding. The contact interaction in Mn : NaF is slightly smaller than that found in Mn : ZnF2 and MnF2 and in Mn : CaF2 is reduced by about 40 %. Inspection of table 3 shows a slight reduction of the contact interaction with increased bond length in the octahedrally co-ordinated compounds ZnF2, MnF2 and NaF and a sudden change in CaF2. However, for CaF2 each manganese ion is at the centre of a cube of eight fluorine ions so that the reduction in the total charge transfer is not so pronounced.The o-bonding orbitals are now d . rather than dy because of the different symmetry involved. The inability to resolve =-bonding in NaF and CaF2 indicates that for most of the impurity ions involved the r-bonding is less than 30 % of the a-bonding. In NaF the probability of finding the magnetic electrons of the impurity ions on a nearest neighbour fluorine ion is approximately 2 %. This value is con- sistent with the ionic nature of the crystal and should be a good indication of the amount of charge transfer which occurs in the pure crystal. I would like to thank Prof. B. Bleaney, Dr. B. R. Judd, Dr. M. C. M. O’Brien, Dr. J. M. Baker and Dr. J. Owen for valuable discussions. I am particularly indebted to Dr. D. A. Jones of the University of Aberdeen for the crystals and Mr. J. Orton and the Atomic Energy Research Establishment, Harwell, for the irradiation facilities. The work was done during the tenure of an 1851 Overseas Scholarship and the report was written while the author was a member of the staff of the Argonne National Laboratory.W. HAYES 1 Seitz, Rev. Mod. Physics, 1954, 26, 7. 2 Stockbarger, J. Opt. SOC. Amer., 1949,39,731. 3 Llewellyn, J. Sci. Instr., 1957, 34, 236. 4 Bleaney and Hayes, Proc. Physic. SOC. B, 1957,70,626. 5 Hayes and Jones, Proc. Physic. SOC., 1958, 71, 503. 6 Watkins, Bull. Amer. Physic. Soc., 1958, 3, 135. 7 Griffiths and Owen, Proc. Roy. SOC. A, 1954,226,96. 8 Owen, Proc. Roy. SOC. A , 1955,227, 183. 9 Owen, Faraday SOC. Discussions, 1955, 19, 127. 10 Tinkham, Proc. Roy. SOC. A , 1956, 236, 535, 549. 11 Baker, Bleaney and Hayes, Proc. Roy. SOC. A, 1958, 247, 141. 12 van Vleck, J. Chem. Physics, 1935, 3, 807. 13 Bowers and Owen, Reports Prog. Physics, 1957, 18, 314. 14 Hartree, Proc. Roy. SOC. A , 1935, 151, 96. 15 Lowdin, Physic. Rev., 1953, 90, 120. 16 Low, Physic. Rev., 1958, 109,256. 17 Low, Physic. Rev., 1958, 109, 247. 18 Jahn and Teller, Proc. Roy. SOC., A, 1937, 161, 220. 19 Bleaney, Bowers and Trenam, Proc. Roy. SOC. A, 1955,228, 157. 20 Shulman and Jaccarino, Physic. Rev., 1957, 108, 1219. 65

ISSN:0366-9033    

DOI:10.1039/DF9582600058

出版商:RSC    

年代:1958

数据来源: RSC

摘要:

ELECTRON TRANSFERS AMONG TRANSITION ELEMENTS IN MAGNESIUM OXIDE* B Y JOHN E. WERTZ,t$ PETERIS AUZINS,? J. H. E. GRIFFITHSS AND J. W. ORTON$ Clarendon Laboratory, Parks Road, Oxford Received 1st July, 1958 All single crystals of MgO which have been available to us contain iron-group im- purities in quantities easily detectable by e.s.r. absorption. For a given impurity, the fractions in the di- and tri-valent states depend in part upon the past heat treatment and upon the concentrations of the other impurities. Energetic radiation may cause electron transfer to occur among the impurities. Oxygen of the crystal may partake in these transfers. As a divalent analogue of the alkali halides, magnesium oxide is a nearly ideal medium for the study of crystalline defects by electron spin resonance (e.s.r.) techniques.Not only may one observe the e.s.r. spectra of such intrinsic defects as I;-centres 1 and trapped holes 2 but also of paramagnetic impurity ions.3 The radius of the Mg2+ ion is closely approximated by the di- or tri-valent forms of most iron-group ions, and hence they commonly substitute for it in the MgO structure. Iron, manganese and chromium have been detected in the purest single-crystal specimens available to us, while nickel, cobalt or vanadium may sometimes also be present. Because of the high dilution and the exact cubic symmetry (in most cases), one sees as narrow e.s.r. lines as in any other solid yet investigated. This narrowness permits the observation of well over two hundred such lines in some of our specimens, An impression of the appearance of the e.s.r.spectra of the ions discussed herein is obtained from fig. 1. This represents positions of the major lines when the magnetic field is applied along a principal crystal axis. Deviations from cubic symmetry about the impurity ions will alter their e.s.r. spectra and betray the presence of lattice defects such as vacancies.4 Since the localized levels of impurity ions occur at various positions in the energy gap of the crystal, the effects of electron (or hole) transfers are readily followed after a variety of treatments. EXPERIMENTAL Magnesium oxide crystals were obtained from the Norton Company (Canada), Infra- Red Development Company (England) and the General Electric Company (U.S.A.) . These were heated in vacuum or in oxygen by means of a tantalum furnace.The speci- mens to be vacuum-heated were enclosed in a box made from MgO in order to prevent their surfaces from becoming cloudy. Some samples were exposed to metal vapours at partial pressures of a few mm to 2 atm at temperatures of 1200°C or more. It was found 5 that uniform colorations could be produced if the crystals were first vacuum-heated. * This work was supported in part by the United States Air Force Office of Scientific Research under contract AF18(600)-479 with the University of Minnesota and in part by an Atomic Energy Research Establishment (Harwell) contract with the Clarendon Laboratory, Oxford. -f School of Chemistry, University of Minnesota. $ Clarendon Laboratory, Oxford. 5 Fulbright research scholar and John Simon Guggenheim fellow at the Clarendon Laboratory, Oxford, for 1957-58.66J . E . WERTZ, P. AUZINS, J . H. E . GRIFFITHS AND J . w. ORTON 67 Samples heated in magnesium vapour showed a Tyndall scattering, presumably due to aggregations of metal along dislocations. X-irradiation was performed with anode voltages ranging from 50 to 100 kV, with samples shielded from light. A germidical lamp emitting most of its energy at 4.9 eV was used for u.-v. irradiation. In some cases filters were used to eliminate light of lower energy. E.s.r. spectra were taken with simple 3-cm transmission cavity spectrometers except for a few measurements at 1.25 cm. Most spectra of Fe3+, Cr34 V2+ and Ni2+ were taken at room temperature, while Fez+ was obseried only at 4°K and Co2+, Fel+ and Nil+ at 20°K.V1+ II I l l I I1 I I I I I lo00 2000 3 0 0 0 4000 so00 Gauss FIG. 1.-Principal e.s.r. lines of impurity ions in MgO. RESULTS The present paper is largely devoted to interpretation of the qualitative displacement of the equilibria M2+ = M3+ + e when MgO is heated in vacuum, in oxygen or in metal vapours, or subjected to 4 9 eV or X-irradiation. Data are summarized in tables 1, 2 and 3. Positive or negative signs are used to indicate a displacement of the valence state in favour of the trivalent or divalent forms respectively. Valence changes were inferred from increase or decrease in the e.s.r. intensity of the observed lines (Fe3+, Cr3+, V2+, Ni2+ and Co2+). In principle one should be able to observe quantitatively changes in all the valence states of iron, but at present the e.s.r.spectrum of Fez+ is not well understood.3 The terms " cubic " and " axial " applied to chromium refer to the electric field symmetry about the Cr3+ ion under observation.4 Positive-ion vacancies must be at least several lattice spacings away from the Cr3+ ion for its electric field to be cubic, while an associated vacancy in [loo] or [110] directions leads to axial or rhombic symmetry. Manganese has not been included in these tables, since heating to 1200°C in oxygen or in vacuum, TABLE EF EFFECT OF HEAT TREATMENTS atmosphere duration Fe V Cr (cubic) + + - + vacuum hours - oxygen hours + + + rapid cooling + magnesium hours - - - Y, days - 9 , - - TABLE 2.-EFFEm OF IRRADIATION source Fe 4.9 eV + 4.9 eV (decay) f 50 kV X-ray + 50 kV (decay) - V Cr (cubic) - - + + + + - - Cr (1 00 axial) - - + + - Cr ( 1 0 0 axial) + + + -68 ELECTRON TRANSFERS IN MgO TABLE 3.-EFFECTS OF VACUUM HEATING ON FURTHER TREATMENT relative e.s.r.line intensity treatment sample Fe V Cr (cubic) Cr (100 axial) none (1) 32.6 heated in air 950°C (1) 110 (2) 2.0 X-rayed 3.5 h (1) 81.5 (2) 49.0 decay for 3 days (1) 81.0 (2) 43.0 4.9 eV for 5 h (1) 49.2 vacuum-heated (2) 0 (2) 12.0 0 0 0 0 0 1 0 1 0 1 180 140 172 130 32 0 60 0 138 40 3.0 11.9 6.8 11.9 4.3 8.0 5.2 8-3 3.8 9.6 or irradiation with 4.9 eV light or X-rays has little effect upon the intensity of the + 112 = - 1/2 e.s.r. lines. These are used as a measure of the Mn2+ concentration. However, the 5/2 = & 3/2 and & 3/2 = & 1/2 transitions are exceedingly sensitive to the presence of other impurities and to any treatments which alter the electric field symmetry.While manganese doubtless participates to some extent in electron transfer with other ions, the Mn2+ ion appears to be exceedingly stable in MgO. Few data are as yet available on the divalent e trivalent transformations of cobalt or nickel. DISCUSSION HEATING IN VACUUM A detailed discussion of the processes occurring upon heating in vacuum, oxygen or in metal vapours will be published elsewhere. An outline is presented here since an important group of valence changes involves the addition or loss of oxygen from the MgO, as well as addition of magnesium or other metal. Other valence changes which occur on heating (principally after irradiation) do not involve gain or loss of the intrinsic lattice components Mg2+ or 02-.At temperatures of 1200°C or higher molecular oxygen is lost from the surface of MgO when the partial pressure of the oxygen is less than the decomposition pressure. Appreciable loss of magnesium does not occur until the temperature is raised above 14OO0C.6 The oxygen escapes in molecular form since the 0 2 - ion is stable only in crystals.7 The electrons left behind by the oxide ions are trapped by trivalent iron group ions, principally Fe3+, though V3+ may also do so if the treatment is prolonged. Vanadium in the 23- state is converted by simple heating to the 3 f state, and hence the " + " in the first row of table 1 under V is largely independent of the atmosphere in which the heating takes place.This behaviour would seem to locate the level of the V2+ state in the energy gap of the crystal above that for Fe2+, with Cr2+ still higher. The conversion of most of the iron in the crystal to Fe2+ profoundly affects the optical absorption spectrum,8 the photoconductive behaviour9 as well as the absorption and e.s.r. centres in- duced upon X-irradiation.29 10 The Fez+ ion is a major reservoir of electrons under u.-v. or X-irradiation, as well as for oxygen which is to be incorporated into the structure in the form of 0 2 - ions by heating in an oxygen atmosphere. Chromium tends to remain in the 3+ state under heating in vacuum. As a mechanism of electron transfer we assume tentatively that a Fe3+ ion located near a positive ion vacancy may acquire an electron from a neighbouring oxygen ion, leaving a hole.The hole as well as the positive-ion vacancy will tend to move toward the surface, with the hole combining with an electron from the surface oxygen which is lost. HEATING IN OXYGEN Oxygen is taken up 11 by MgO at temperatures above 1000°C in amounts exceeding that required to form a chemisorbed layer. However, we reject theJ . E . WERTZ, P . AUZINS, J . H. E . GRIFFITHS AND J . w. ORTON 69 notion of “excess” oxygen11 contained within the crystal structure as an un- justified and misleading concept. Instead, we note from the e.s.r. data the tendency of divalent ions to lose an electron and become trivalent. This is con- sistent with the assumption that atomic oxygen on the surface may become in- corporated as ions by acquisition of electrons from interior divalent impurities.The outward migration of magnesium ions (and therefore the inward migration of positive ion vacancies) allows the process of oxygen take-up to continue as long as electrons are available from impurities for 0 2 - ion formation. Simul- taneously, the inward migration of magnesium-ion vacancies prevents formation of a space charge and provides local charge compension for the extra positive charge of trivalent ions. Since one such vacancy compensates for two trivalent ions, one must have at least one ion without an associated vacancy for every one associated, the ions being on the average too remote for simultaneous compensation. It is evident that this mechanism requires a decrease in density of the MgO crystal as the oxygen is taken up.It is likewise apparent that the possibility of conversion of an impurity from its divalent form to the trivalent form biases the formation of positive-ion vacancies, while discriminating against negative-ion vacancies which would aggravate the charge-compensation problem. It is known from another study that the number of negative-ion vacancies is normally very small, though these can be produced by drastic treatments.1 This is in disagreement with Clarke’s interpretation of absorption spectra of MgO.12 We presume that the transfer of electrons from divalent ions to oxygen on the surface (or vice versa on vacuum heating) proceeds by way of hole movement in the opposite direction. At the elevated temperatures at which these experiments are carried out, one does not expect appreciable localization of the holes, though trapped holes can be demonstrated at room temperature after 4-9 eV or X- irradiation.2 HEATING IN MAGNESIUM VAPOUR The effect of this treatment is to tend to convert the iron-group ions to their divalent states.That the effect is not specific to magnesium can be shown by heating in lithium, vanadium, manganese or aluminium, the net effect being qualitatively the same. The metal atoms are converted to ions by transfer of electrons, and a stable state will be reached with different ratios of divalent to trivalent ions for the several impurities. EFFECTS OF X-IRRADIATION The optical absorption induced by this treatment has been discussed recently.10 One gets perhaps the clearest indication of the processes occurring from a study of the changes in e.s.r.spectra, which reveal extensive electron transfer occurring. The results may be summarized by the following set of equations : (1) (2) (3) (4) ( 5 ) Fe2+ = Fe3f + e 0 2 - (at positive ion vacancy) = 01- + e e + Fe2+ = Felf e + Cr3+ = Cr2+ e + V3+ = V2f. We shall assume that before irradiation the concentrations of each ion in its several valence states are the equilibrium values. This is a reasonable assumption if the crystal has first been heated in vacuum at 1200°C or more and cooled slowly. After this treatment the predominant ions will be those on the left-hand side. The effect of absorption of energy will be to upset the equilibrium by causing the70 ELECTRON TRANSFERS IN MgO reactions to occur as written.It is possible that Mn2+ is converted to Mn3+, but if it is, the decay to the divalent form is so rapid that one sees no change when the irradiation has ceased. Other than the oxygen ions of the crystal, the Fez+ is the important source of electrons, some of which indeed it captures itself to form Fel+. It is thus understandable that in crystals containing a large concentration of chromium there is only a small fractional conversion from the Cr3+ to the Cr2+ state if the concentration of Fe2+ is not also large. Very commonly the Cr3+ em-. spectrum becomes undetectably weak after X-irradiation. In these processes of electron transfer by way of the conduction band, one is probably filling electron levels which lie close to it, while emptying low-lying levels.One infers again that the Cr2+ and V2+ levels lie considerably higher than the Fe2+ level. As noted in the section on vacuum treatment, the V2+ level appears to lie intermediate between Fe2+ and Cr2+. The Fel+ level probably lies within 2 eV of the conduction band. While this level is normally empty, it may be oc- cupied temporarily as a result of 4.9 eV or X-irradiation and serve as a source of photoconduction electrons at lower energies than the usual 5 eV band.9 ULTRA-VIOLET (4.9 ev) IRRADIATION Depending upon the content of impurity ions and the previous treatment of a crystal, MgO may show an optical absorption coefficient at 4.9eV ranging in order of magnitude from 1 cm-1 to 100 cm-1. There appear to be a number of overlapping bands in this region, none of which has yet been unequivocally assigned.One notes the formation of trapped holes2 as well as of Fel+, Fe3+, Nil+ and V2+ ions. It appears reasonable to assume that there is weak absorption by Fez+ in this region, leading to the formation of the observed Fe3+ and Fel+. Absorption bands are induced both in the visible and the ultra-violet regions, the latter being in part assigned 8910 to Fe3+. It is known that irradiation with sodium light will reverse some of the electron transfers induced by 4.9 eV or X-irradiation. Under some treatments (see table 3) it is noted that 4.9 eV irradiation leads to a reduction in the Fe3+ concentration. Whether reaction (1) proceeds in the forward or the reverse direction is intimately related to the concentration of positive-ion vacancies, as noted later in this paper.It is important to note that the results in tables 1 and 2 refer to specimens which have previously been heated in vacuum (except, of course, for the effects of vacuum treatment itself). Unless the crystals are so treated, the results of various treatments may be qualitatively different. This is illustrated by the data of table 3, which refer to two crystals cleaved from the same parent stock, one of which is vacuum heated and the other not. The effects are most pronounced with iron and chromium. The origin of these differences is probably associated with the different con- centration of positive-ion vacancies in the two cases. Vacuum heating will be expected to reduce greatly the number of such vacancies, and in accord with this view, one notes the markedly smaller tendency toward formation of Fe3+ after such treatment.Although the e.s.r. spectrum refers to the Fe3+ in a purely cubic electric field (which implies that there are no vacancies within a radius of at least several lattice spacings), it is likely that stable trivalent-ion formation will require some positive-ion vacancies within tens of Zngstroms for charge compensation. While the vacuum-heated sample showed a very slight increase in Fe3+ concentra- tion after the 950°C heating, the previously unheated sample showed a very large increase. One notes that the untreated specimen shows a decrease in Fe3+ e.s.r. intensity on X-irradiation, while the opposite is true of the vacuum-heated one.For Cr3+ in a purely cubic electric field, the vacuum-heated specimen shows a very marked tendency for conversion to the 2+ state on similar treatment. Having been formed, the Cr2+ state also appears to be much more stable in the vacuum- heated sample.I. E. WERTZ, P . AUZINS, J . H . E . GRIFFITHS AND J. w. ORTON 71 The formation of Fel+ as a moderately stable ion is probably of considerable importance in the consideration of the optical, conductive or photoconductive properties of MgO crystals of the purity thus far available (our purest specimen had 7p.p.m. of Fe). This ion, as well as Nil+, was detected in MgO after its presence in NaF was reported by Bleaney and Hayes.13 The existence of the Fel+ ion was postulated by Peria to explain his photoconductivity studies, although he had no direct evidence of its stable formation in MgO.9 When produced by 4.9 eV or X-irradiation, it slowly decays, so that its e.s.r.spectrum is usually not detectable one week after X-irradiation. However, a crystal which was heated in mamesium vapour three years ago still shows an easily-detected Fel+ e.s.r. absorption. The stabilization of the lower valence state in the presence of a metal is analogous to the stabilization for an indefinite time of the Cr2+ and the V2+ valence states in this same crystal. It will be noted that the formation of Fel+ represents one form of charge com- pensation (for trivalent ions) which may occur at a time when there are appreci- able numbers of electrons in the conduction band.One will then expect that one Fe3+ ion will be converted to Fe2+ or one trapped hole will be annihilated for every Fel+ ion which disappears. If the holes are eliminated from consider- ation, one expects one Fe3+ ion to be converted to the Fez+ state for every Cr3+ ion formed from Cr2+. Only a few aspects of electron transfer have been discussed here, due to space limitations. The correlation of changes in e.s.r. absorption intensity with other electronic properties, especially with optical absorption, will be discussed elsewhere. Thanks are due to Dr. G. K. Finlay of the Norton Company (Ontario) and to Dr. R. Hansler of the General Electric Company (Ohio) for supplying single crystals. Messrs. B. V. Haxby, R. G. Lye and J. P. Sturtz have kindly treated specimens for us and freely shared their data. Dr. G. Series (Oxford) supplied some of the optical filters used. We have benefited from discussions with A. J. Dekker, W. Peria, B. V. Haxby, R. W. Soshea, L. Orgel, W. N. Lipscomb and John Owen. 1 Wertz, Auzins, Weeks and Silsbee, Physic. Rev., 1957, 107, 1535. 2 Wertz, Auzins, Griffiths and Orton, to be published. 3 Low, Ann. N. Y. Acad. Sci., 1958, 72, 69. Numerous earlier references are cited in 4 Wertz and Auzins, Physic. Rev., 1957, 106,484. 5 Haxby, private communication. 6 Lye, private communication. 7 Yamashita and Kojima, J. Physic. SOC. Japan, 1952,7, 261. 8 Haxby, to be published. 9 Peria, submitted for publication. 10 Soshea, Dekker and Sturtz, J. Physics. Chem. Solids, in press. 11 Weber, 2. Physik, 1951, 130, 392. 12 Clarke, Phil. Mag., 1957, 2, 607. 13 Bleaney and Hayes, Proc. Physic. Suc. B, 1957,70,626. this paper.

ISSN:0366-9033    

DOI:10.1039/DF9582600066

出版商:RSC    

年代:1958

数据来源: RSC