P. K. GALLAGHER AND J . B . MACCHESNEY 47 GENERAL DISCUSSION Prof. N. N. Greenwood (Newcastle upon Tyne) said I was interested in Gallagher's observation of a single line for Fe2+/Fe3+ above the Nkel temperature but the appear- ance of two separate hyperfine fields below it due to the slowing down of electron exchange as a result of antiferromagnetic coupling. Whitfield and I have recently observed a single hyperfine field in cubanite (Cu+Fe2+Fe3+S3) between room tempera- ture and 77°K due to the ferromagnetic exchange interaction between iron ions as in magnetite Fe3+[Fe2+Fe3+]04. Is the occurrence of one or two 6-line spectra in mixed valency iron compounds a general criterion for ferromagnetism on the one hand and antiferromagnetism on the other? Prof. J. F. Duncan (Victoria University of Wellington N.Z.) said Would Gallagher and MacChesney agree that the designation of formal oxidation and valence states as Fe2+ Fe3+ and Fe4+ should not be undertaken without some caution? Whilst it was clear that different isomer shifts were obtained for the species designated in this way the safest conclusion to be drawn was that the electron densities at the 57Fe nucleus were different in these cases.It did not necessarily follow that this was due to successive removal of a single d-electron from the valence shell of the iron. Indeed the absence of quadrupole interactions for Fe4+ suggested that this could not be fairly designated as d4. An alternative possibility is that all five 3d orbitals were occupied for 4/5 of the time by five electrons; or yet again that due to electron delocalization of all five 3d electrons from the iron atom to other parts of the system the eEective screening of the electrons by each of the 3d electrons was only 4/5 of that normally obtained in a 3d5 structure.Many other cases of uncertain or intermehate oxidation states are known of which an exampIe is the nitroprusside ion. This can be written as [Felll(CN)S NO]'- or [Fe"(CN),N0+I3-. Chemically it behaves as iron (11) but the Mossbauer isomer shift 6 suggests iron (111) although it can be fitted on to a linear plot of quadrupole separation AE against 6 for compounds of the type [Fe"(CN),XI3- where X=H20 NH3 CN- NO2 NO+. Spectroscopic (ir.) evidence seems to favour iron (111). Dr. I?. K. Gallagher (Bell Telephone Lab. N.J.) said In reply to Duncan I agree that there are always certain reservations or conditions associated with the assignment and interpretation of oxidation numbers.For the iron compounds which I have 48 GENERAL DI S CUSS10 N discussed however I do not think that it is incorrect to refer to them as compounds of iron(1V). This seems consistent within the general limitations universally applied to such terminology. Admittedly it is a simplified picture but it is just as correct as to say niobium has a + 5 oxidation state in KNb03 or titanium is +4 in the anal- ogous SrTiO compound. One can show that departure from oxygen stoichiometry or La substitution gives rise to the formation of a peak (or peaks) in the Mossbauer spectrum at a position which is consistent with the formation of Fe(II1) at the expense of what I call Fe(1V).The absorption which I attribute to iron(1V) is well outside that of other valence states of iron in oxide systems and is reasonable for that of iron(1V) based on calculations by Danon. There is no indication of any peroxide or super- oxide ion formation based on the crystal structure. The formation of the perovskite structure also suggests a small radius for the iron ion consistent with iron(1V). Prof. J. Danon (Rio de Janeiro) said The values of the isomers shifts listed in table 2 of Gallagher et aZ.’s paper are those expected for the +4 oxidation state of iron according to the modified Walker Wertheim and Jaccarino calibration of the isomer shift of Fe57. Dr. G. M. Bancroft (Ufiiversity of Cambridge) said Could Gallagher comment on the apparent lack of quadrupole splitting in these compounds? Considering the electronic structure of Fe4+ a large quadrupole splitting might be expected.Dr. P. K. Gallagher (Bell Telephone Co. N.J.) said The question of the absence of quadrupole splitting is more interesting. I do agree that the picture of an electron occupying each d orbital four fifths of the time or possibly each dc orbital two thirds of the time is a possible hypothesis; however I do not agree that this does not repre- sent a 3d4 co&pration. This is analogous to the treatment of 3d6[Fe(II)] which does not exhibit quadrupole splitting in a cubic en~ironment.~’~ To date our studies of iron(1V) materials have been restricted to those in which the iron(1v) has cubic symmetry. The exception to this is the Srl-x La,FeO (x<0.5) in our paper. There may well be quadrupole splitting in this material.This is presently under investigation in connection with the determination of the activation energy associated with the electronic exchange in this system. Prof. N. N. Greenwood (Newcastle upon Tym) said Gallagher’s use of the symbol Fe(IV) to indicate an oxidation state of +4 is justified and is more than just a book- keeping convenience. The concept of oxidation state is precisely defined in chemistry and does not necessarily imply an electrostatic charge of +4 on the iron ions. The optical spectra and magnetic properties of Fe(1V) would be quite distinct from those of Fe(II1) for example and the species would contain a different number of unpaired d electrons. Oxidation state should therefore not be confused with electron density on an atom.A particularly clear-cut example of this distinction is the ferricyanide [Fe(CN)J3- and ferrocyanide [Fe(CN),I4- ions. The Mossbauer chemical shift of these two species is almost identical implying identical s electron density at the nucleus and hence identical shielding by the 3d electrons etc. However the oxidation state undoubtedly alters from Fe(III)d5 to Fe(II)d6 as shown by the diamagnetism of the l J. Danon in Applications of the hfossbauer Efect to Chemistry and Solid State Physics (Int. Atomic Energy Agency Vienna Austria) 1966 p. S9. G. K. Wertheim Mossbauer Efect PrincQdes and Application (Academic Press Inc. New York 1964) chap. 6. G. Wertheim €3. Guggenheirn J. Williams and D. Buchanan Physic. Reu. 1967 158,446. F. F. Ham Physic. Rev. 1967 160 328. GENERAL DISCUSSION 49 reduced species.Calculations suggest that although one electron is added to the iron d orbitals approximately 0.9 of an electron is delocalized at ligand-based orbitals via d -pn back donation. The overall electron density around the iron atom is thus simi- lar in the two species despite their differing oxidation states. In this sense Mossbauer chemical shift data provide a useful experimental verification of the old Pauling electroneutrality rule for low spin coordination complexes. It should perhaps be added that alteration of the oxidation state of low spin iron complexes does not always leave the chemical shift unchanged and sometimes the added electron density remains predominantly on metal-based orbitals. For example we have recently shown that reduction of the tris(maleonitriledithio1ate) iron anion from [Fe1vS6C6(CN)6]2- to [Fe111S,C6(CN),]3- increases the chemical shift by 0.15 mmsec-'.Mr. M. G. Clark (University of Cambridge) said In connection with the paper by Gallagher and MacChesney it seems a little surprising that even at low temperatures only small or negligible quadrupole splittings have been observed in Fe(1V) Mossbauer spectra. The simplest picture of the d electrons of high-spin Fe(1V) is a spherically symmetrical 3d5 core containing a positive hole which would be expected to lead to quadrupole splitting. It would be interesting to see how far this simple picture was supported by measurements of the electronic spectrum of Fe(1V). Dr. P. K. Gallagher (Bell Telephone Lab. N.J.) said The iron(1V) compounds which we have measured to date have always provided a cubic environment for the iron(1V).This would explain the absence of the splitting even though the iron is presumably 3d4. A simjlar example is RbFeF,. For the rhombohedral phase Sr0.5La0.5Fe03 (fig. 7 of our paper) there may well be splitting in the iron(1V) spectrum but it is not easily resolved. Dr. A. G. Freeman (Victoria University of Wellington N.Z.) (communicated) The Mossbauer spectra of several graphite-FeCl and graphite-FeCl3-A1Cl3 compounds at room temperature and at 78°K give the following typical results. isomer shift (rel. to natural iron) width at half maximum (mm/sec) (mm/sec) FeC13 (anhydrous) 78" 0.55 f0.04 0.30 f0.05 graphite-FeCl 78" 0.61 &0.04 0.30 d~0-05 graphite-xFeCl3-yA1Cl3 78" 0-62 f0.04 0-30 f0.05 0.1 tx/y<0*25.These observations discount the importance of such bonding species as C,+Cl-FeCl,. 3FeC1 and C:[FeCl4]-FeC1 which have previously been suggested to account for the existence of these compounds. The presence of FeCl would also be indicated by a quadrupole split doublet with rather higher isomer shift than is observed. The shift of the absorption peak to higher velocities is considered significant. It indicated that there has been a general but small increase in the electron density of the d shells of all the iron atoms involved. Presumably these electrons have been donated from the n-electron conductance band of the graphite although donation of graphite n-electrons into the d-shell of thechlorine atoms is probably also ofimportance in bonding the intercalated molecule to the graphite layers.R. G. Shulman and S . Sugano J. Chem. Physics 1965,42 39. T. Birchall N. N. Greenwood and J. A. McCleverty Nature 1967 215 625. 50 GENERAL DISCUSSION Dr. R. Krishnan (Laboratoire de Magn&isme France) said With regard to the paper by Gallagher et al. I wish to make the following remarks concerning the diff- erent valence states of Fe in oxide compounds. In certain cases there are other methods more sensitive than Mossbauer techniques such as ferrimagnetic resonance techniques and determination of thermoelectric properties to detect these valence states of Fe. For example the presence of Fe2+ even in very minute quantities affects pro- foundly the magnetic anisotropy and resonance line width in spinel and garnet ferrites. And lately in Y3Fe5 . OI2 doped with Ca2+ the presence af Fe4+ is indicated by its strong influence on the line width variation with temperature. In the same way elec- trical conductivity and Seebeck coefficient measurements enable a fairly accurate guess of the existing valence states of Fe to be made. The Fe4+(d4) ought to cause Jahn- Teller distortions of the octahedral sites as Mn3+(d4) which is well known for Spinel ferrites and manganites.