C. E. JOHNSON 11 GENERAL DISCUSSION Prof. J. Danon (Riu de Janeiru) said Accurate measurements of isomer shifts show that this hyperfine interaction is sensitive to small changes arising from chemical bonding. Precisely with the iron (11) chlorides it has been found a systematic variation of the isomer shift with the hydration number.l These results indicate that the degree of covalency decreases with the increase in the number of coordinated water molecules around the iron atom. One of the difficulties for interpreting the hyperfine fields in terms of covalency has been pointed out by van Wieringen with the Mn2+ salts a small mixture of excited states involving s-shells such as (3s-4~) gives a very large contribution to the hyperfine field. The temperature variation of the electric field gradient as used in Johnson’s treatment is not sensitive to the presence of such states which involve s-electrons.J. W. Hurley Jr. R. C. Axtmann and Y. Hazony Bull. Amer. Physic. SOC. 1967 12 654. Dim. Faraday SOC. 1955 19 118. 12 GENERAL DISCUSSION Dr. C. E. Johnson (A.E.R.E. Harwell) said In reply to Danon the range of variation of the values of the hyperhe field with chemical bonding is in general larger than that of the isomer shift. It is in this sense that hyperfine field measurements are more sensitive. Regarding your second point the energy of the excited atomic states is surely too high to be affected directly by chemical bonding. One can therefore qualitatively associate a change in hyperfine field of nearly S-state ions with a change in d-electron spin density produced by covalency.Prof. J. Danon (Rio de Janeiro) said The following electronic mechanisms can affect the isomer shift of Fe5' in its compounds 1-3 (a) direct contribution from 4s bonding ; (b) indirect contribution from 3d bonding. From free-ion wave functions an increase in the number of 3d electrons increases the shielding of s electrons (mainly 3s) and thus decreases their density of the nucleus. Changes in 3dpopulation by bonding could have the same effect on the s-electron density at the nucleus. This can occur in two ways (i) covalency between d electrons and filed ligand orbitals which will increase the population of d electrons on the metal ; (ii) the bonding of d electrons with empty ligand orbitals will decrease the d-electron density at the metal ion because of back-donation.There is however some evidence suggesting that mechanisms (i) i.e. donation from the ligands to the 3d orbitals may not be an important one for determining the isomer shift in high-spin iron c~mplexes.~ TABLE 1. 3d 4s I.S. (cm/sec) Fe"C1 7.01 0.27 + 0.12 Fe"'C1 6.92 0.27 + 0.04 Table 1 lists the M.O. populations recently calculated for the ferrous and ferric tetrachlorocomplexes and the respective isomer shifts referred to a Cr source. The difference in isomer shift which is of similar magnitude as that observed with other + 2 and + 3 high-spin iron complexes cannot be explained on the basis of the M.O. populations. This can be interpreted by considering that the shielding of the 3s orbitals by the 3d orbitals may not be altered by the increase in the 3d orbital population due to bonding.A possible reason for this is the following as a con- sequence of the increase in the 3s population a spread of the 3d wave function occurs. It is known (nephelauxetic effect) that the average radius of the 3d shell is larger in bonded transition ions as compared to the free ion. From the point of view of the shielding of the 3s orbitals the increase in 3d population may be compensated by the spreading of this orbital. Evidence for this type of compensation mechanism has been recently discussed and it has been shown that reduced charge on the metal due to increased covalency compensates for most of the natural contraction of the 3d radial function with increasing nuclear charge. As a consequence the shielding of the 3s by the 3d orbitals remain unaffected by ligand-to-metal 3d bonding and its value is that corresponding to the free-ion configurations.When empty ligand orbitals are available for back-donation the spreading of the 3d charge will occur without any compensating increase in its value. In this situation the 3s shielding is markedly affected by the expansion of the 3d wave functions as has been shown with the complex ion cyanides and nitrosyl~.~* L. R. Walker G. K. Wertheim and S. Jaccarino Physic. Rev. Letters 1961 6 18. R. G. Shulmann and S. Sugano J. Chem. Physics 1965,42,39. J. Danon in Applications of Mossbaiier Efect in Chemistry and Solid State Physics (I.A.E.A. Vienna 1966) p. 89. H. Bash A. Viste and H. B. Gray J. Chern. Physics 1966 44 10. J. Danon and L. Iannarella J. Chem. Physics 1967 47 382. GENERAL DISCUSSION 13 Dr.R. V. Parish (University of Munchester Inst. of Sci. and Technology) said Danon suggests that the observed difference in isomer shift between FeC14 and FeCla- is better accounted for using the free ion 3d-populations (3d5 and 3d6) than those obtained by molecular orbital methods. The latter figures seem unreliable at present \ \ \ \ \ \ \\ \\ \ / / \ aln - as judged by the variety of figures available. I should like to suggest an additional factor which does not seem to have been taken into account and which may possibly explain why the effects of donation from the ligands to the 3d-orbitals may not be as large as one imagines. Consider the molecular orbital scheme for an octahedral complex. The bonding molecular orbitals represent the effects of ligand to metal donation.Since these orbitals we are told have some 10-20 % 3dcharacter to this extent we have an increase in the 3d-population over the free ion value. In a rough way one may say that the d-character of the bonding e orbitals is compensated by a corresponding degree of ligand character in the ez antibonding orbitals i.e. electrons in the e* orbitals are delocalized back to the ligands in a type of metal to ligand a-bacl- donation. In high-spin complexes there are two eg electrons both for ferrous and ferric complexes. Thus the effect of donating four electrons into the 3d,e orbitals is partially (roughly half) offset by back-donation of two electrons in e orbitals. Finally a treatment of this kind is much more difficult for tetrahedral complexes since in this case there is considerable mixing of 3d and 4p orbitals.Mi. M. Clear Prof. J. F. Duncan and Mi-.K.Trotter (Victoria University of Welling- fotz N.Z.) said A question was raised as to how covalency effects might be allowed for or investigated in relation to hyperhe fields. It was suggested that the effects of covalency could be more effectively investigated by application of external fields to compounds which are known to be covalent and these could then be compared with those which are not. The extent to which the nuclear hyperfine field is detectable depends on the relaxation time and the temperature. The observed field will also depend on the magnitude of the applied field. Even at room temperature the line width increases by 5-10 fold with applied fields of 25 kgauss depending on the spin 14 GENERAL DISCUSSION type.With high spin Fe2+ the increase is rather smaller than for low spin Fe" (e.g. [Fe(CN),]"-) even after making allowance for the quadrupole interactions of the former. The covalency present in low spin Fe" eliminates the spin and orbital contributions present in high spin Fe2+ so that the following features become apparent. (a) An estimate of the degree of covalency present in compounds of unknown electronic structure can be obtained by measuring the line broadening. This would be expected to be more sensitive at lower temperatures where hyperfine fields could be directly measured. (b) The hyperfine field is affected by the external field by at least two mechanisms as is evident from the relatively different effects of high and low external fields. Undoubtedly direct spin-field interactions are important but it seems likely that spin polarization plays a significant role as well.In view of the intense magnetic fields present in atoms one may well ask whether these play any role in banding (even though magnetic interactions are relatively weak compared to electrostatic inter- actions). An interesting case is iron (111) ammonium alum in which the line width is ob- served to decrease at about 2 kgauss external field but to increase as expected at higher fields. This does not occur however if the iron (111) is replaced by aluminium (111) ions until it is present in about 3 mole % only. Spin-spin or spin-lattice interactions are therefore apparently occurring which is equivalent to regarding the pure 5S state of the iron (111) atom as mixing with a higher energy state-i.e.covalency effects are present. The energy difference between the two states must however be small since the spectra are not detectably different at low fields when the absorber is cooled in liquid nitrogen. Dr. C. E. Johnson (A. E. R. E. Harwell) said In reply to the remarks of Duncan et al. I am not clear as to their meaning. The core polarization field H and the effective value of (r3) are both sensitive to covalency. But measurements 011 paramagnetic ions (e.g. Fe3+) in external magnetic fields at room temperature are not simple to interpret as the spectra are dominated by relaxation effects. The broadening cannot necessarily be described in terms of an effective magnetic field at the nucleus but may depend upon all the components of the magnetic hyperfine interaction tensor.To extract the values of these components (and hence H and (r3)) from the data is not straightforward and I would doubt that their conclusions about the determina- tion of the degree of covalency could be generally valid. The complexity of the behaviour of Fe3+ in the alums has been extensively studied by Housley and de Waardl and by Campbell and de BenedettL2 In low spin (i.e. S = 0) Fe" compounds one would expect no internal magnetic field at the nucleus and the broadening should be due solely to the external field at all temperatures. This is also true for high spin Fe2+ at room temperature and is shown in the spectra measured by Grant et aL3 So I do not understand your state- ment that high spin Fe2+ broadens less than low spin Fe" in a magnetic field.Regarding the role of these intense magnetic fields in chemical bonding it must be remembered that the core polarization field H is not a real magnetic field but is a convenient way of expressing the " contact " interaction between s-electrons and the nucleus. This field therefore only has a meaning at the nucleus and although its magnitude is affected by covalency it cannot be said to make a contribution to chemi- cal binding. Housley and H. de Waard Physics Letters 1966 21 90. Campbell and de Benedetti Physics Letters 1966 20 102. Grant et al. J. Chem. Physics 1966 45 1015. GENERAL DISCUSSION 15 Prof. J. I?. Duncan (Victoria University of Wellington N.Z.) said In reply to Johnson I agree with your remarks in general which put the situation very clearly but your reply makes it desirable that I should clarify a few aspects of our work.(a) The internal field observed with FeI1 has approximately the same slope at high fields as that predicted for a bare nucleus but up to about 4 kgauss the lines hardly seem to broaden at all. It is possible that this is an experimental artefact but we believe not. (b) In our work broadening is usually observed Fe" > Fe3+. Fez+ has a quadrupole interaction and therefore a correction is necessary to allow for this different feature. Fez+ then falls between Fe" and Fe3+. Prof. N. N. Greenwood (Newcastle upon Tyne) said I would ask Johnson whether his analysis for high spin d6 (four unpaired electrons) could be applied directly to high spin c14 (e.g. F P ) ; and if so what changes would be necessary. In Gallagher's paper the hyperfine field at FcIV in a perovskite is shown to be about 270 kOe which is similar to that in several of Jolinson's FeI1 compounds.I am not sure that I followed all the points made by Duncan but one point of principle bothers me. The single line is shown as splitting into six separate peaks at 0.46 0.36 0.28 0.20 0.1 1 and 0.00 nim sec-I. The separation between these peaks is considerably less than the Heisenberg natural line width for an ideal 57Fe source- absorber pair and they could not therefore be resolved even if the experimental line width approached the natural line width. Prof. J. F. Duncan (Victoria University of Wellington N.Z.) said The results on one compound viz. iron (111) ammonium sulphate were chosen to illustrate the kind which could be obtained. No claim was made that six lines were resolved indeed the slide showed that they were not.However there was certainly a systematic fluctua- tion in the observed intensity in the trough of the resonance which seemed to indicate the presence of separate resonances approximating to the values quoted by Green- wood. In the example shown the variation in intensity was of the order of statistical fluctuation but cases have been observed by us with identifiable peaks some five times larger than the standard deviation of the statistical fluctuation in magnetically lip-e-broadened Mossbauer spectra. One does not necessarily expect six lines to be observable in such spectra because in addition to the difficulty of separating the nuclear hyperfine peaks there is also the possibility of crystal field and spin-spin interactions which could affect the resonance energy.However in my comments on the spectra shown the significance of the resonance intensity was not discussed. The important feature was the line width which is cleaxly evident as broadening enor- inously on application of a magnetic field in a way which depends on the spin type of the species concerned. Dr. C. E. Johnson (A. E. R. E. Hawell) said In reply to Greenwood the two ions are similar ; high spin d6 may be considered as an electron outside a spherical d5 core while high spin d4 is a hole in the half-filled d-shell. FexV therefore has a spin orbit coupling parameter of opposite sign (i.e. positive). Dr. A. Simopsulos (Greek At. Energy Comm. Athens) said I would like to present some data on paramagnetic relaxation of Fe3+ ions.Samples of 57Fe impurities bound in LiOH and Ca(OH) were used as sources. Single line absorbers This work was carried out at the Soreq Nuclear Research Center Yavne Israel in collaboration with I. Pelah and P. Hillman. 16 GENERAL DISCUSSION of sodium ferrocyanide were used in order to detect the hyperhe splitting and the absolute effect in the sources. Mossbauer experiments were performed in the temperature range 20-300°K. Fig. 1 shows the temperature dependence of the LiOH ; Fe57 spectra. From the andysis these spectra consist of two parts one due to 57Fe in Co aggregates formed in the sample and the second due to paramagnetic hyperhe interaction of the Fe3+ ions. The shape of this second part is shown in fig. 2 where -0 - 2 - 4 - 6 x 0- x - .2 2- 8 2 % 4- E c) 5- - 2 - 4 -5 ii 222 O K +I0 +8 +6 +4 +2 0 -2 -4 -6 -8 -10 I I I I I I L - U velocity (mmlsec) FIG.1 .-The effect of temperature on the relaxation time of the LiOH Fe5’ polycrystalline source. The absorption is corrected for background. Absorption scales for each spectrum are shown next to the corresponding temperature. we have subtracted the part due to Co aggregates from the total spctrum. The temperature dependence of the spectra is typical of paramagnetic relaxation of the Fe3+ i0n.l Hyperfhe spectra due to the I & 5/2) spin state are clearly shown in the temperature range 20-200°K. The I f3/2) spin state displays hyperfine spectra around 20°K. The I f 1/2) spin state has too short a relaxation time to display any magnetic spectra in the temperature range examined but its effect is apparent from the broadening of the corresponding unsplit line at lower temperatures.The spin H. H. Wickman Mussbauer Methodology (Plennum Press 1966) vol. 2. GENERAL DISCUSSION 17 lattice relaxation time z was calculated from the broadening of the lines. The follow- ing temperature dependence fits these data with This temperature dependence is characteristic of Raman processes for impurities participating in a localized mode wo = kO/h The measured mode at 8 = 280°K is r1 = AT2 +B exp (- OiT) A = 400 sec-l deg.-2 ,B = lo7 sec-' and 8 = 280°K. -- 273 "K 0.9 m rn / s t c V E = 4% 235 "K IJ 1 I +I0 +8 +5 +4 +2 0 -2 -4 -6 -8 -10 velocity (minlsec) FIG. 2.-Mossbauer spectra of fig. 1 after subtraction of the Co aggregates part. The hyperfine structure for the I &5/2> state is shown by the broken line and along the bottom of the figure.J. Murphy Physic. Reu. 1966 145 241. 18 GENERAL DISCUSSION in agreement with optical modes of Li0H.I The crystal field splitting of Fe3+ was determined from depopulation effects of the I t5/2> spin state and was found to be D = 1.6 x eV. The Ca(OH) source displayed magnetic hyperfine spectra in the temperature range 20-300°K (fig. 3). Three different iron sites were assigned to these spectra. I I I 1 1 I I 4l 6 I 2 I 4 t G' G -235°K I t f ! I t 4 f 26OK GI C D - I I I I I I 1 +I0 +8 +6 +4 +2 0 -2 -4 -6 -8 -10 velocity (mmlsec) FIG. 3.-Mossbauer spectra of FeS7 in The absorption is corrected for background. The broken line shows the shape of the central part of the spectra after subtraction of the magnetic peaks corresponding to the I 5/2) state.Two of them are trivalent iron ions and the third is divalent iron. This complexity of the spectra did not permit a quantitative study of the temperature dependence of the spin-lattice relaxation time. However comparison with the spectra of the LiOH source shows that the relaxation time of the trivalent iron sites has a temperature dependence even weaker than that displayed by the iron site of the LiOH source. This is due to the fact that the energy of the optical modes of Ca(OH) is higher than that of Li0H.l The divalent iron displays only a quadrupole splitting. Due to the strong spin-orbit coupling the spin-lattice reIaxation of this ion is too short to give rise to magnetic hyperfine spectra. I. Pelah K. Krebs and Y. Imry J.Chem. Physics 1965,43 1864. GENERAL DISCUSSION 19 Mr. M. G. Clark (University of Cambridge) (communicated) In connection with the remark by Simopoulos the rapid spin-lattice relaxation of high-spin Fe2+ is in fact due to the orbital angular momentum of the corresponding free-ion term being non-zero. Thus there is strong orbit-lattice coupling which leads to rapid transi- tions. This point is discussed in more detail in a study of the role of electronic factors in the paramagnetic relaxation of high-spin Fe2+. It has also been shown that partial quenching of the orbit-lattice interaction of high-spin Fe2+ may sometimes occur e.g. in a square-planar envir0nment.l. Prof. R. H . Herber (Rutgers-The State Uniuersity) said I would like to make reference very briefly to two very recent papers from Drickamer’s group at Illinois and an ingenious idea first proposed to me by Dr.Y. Hazony at Princeton University which ties Drickamer’s work to the present discussion concerning 57C0 “ probe ” atoms in Mossbauer spectroscopy. What Drickamer and his colleagues have done is to subject a number of iron complexes (used as Mossbauer absorbers) to pressures up to -200 kbar and study the pressure dependence of the I.S. and Q.S. parameters. The results are quite dramatic especially the ferric complexes. In K,Fe(CN) 6 for example the isomer shift becomes more negative as the pressure is increased and at about 50 kbar suddenly becomes more positive by about 1*2mm/sec. Then as the pressure is further increased the isomer shift again decreases slowly almost to its original value at about 175 kbar.A similar “jump ” is observed in the pressure dependence of the quadrupole splitting. The authors identify the high-pressure form with ferrous iron-on the basis of its IS. and Q.S. parameters- which is formed reversibly under pressure. In a parallel study of Fe2(S04), FeP04 and ferric acetyl acetonate inter alia they find typically about 50 % of the iron in Fe2+ states at 150-200 kbar. These results may be related to what is observed in the Mossbauer spectra of 57C0 compounds (not doped !) where one sees in spectra produced with Co3+ sources resonance peaks which can be ascribed to Fe2+ Fe3+ and higher positive charge states. The latter are accounted for-at least qualitatively- by invoking an internal conver- sion-Auger effect cascade and a number of calculations dealing with the formation of such states have appeared in the literature.What has always been a mystery is how to account for charge states lower than those which obtain in the parent coin- pound and which are clearly seen in cobalticinium salts,4 cobalt (111) acetylacetonate and in the elegant work done recently by Sano in our laboratory on cobalt (111) cyanides. Hazony’s idea is the following in typical ionic compounds the ionic radius of Co3+ is slightly smaller than that of Fe3+. Therefore an iron atom in the 3+ charge state formed by the ex. decay of a 57C0 atom sitting in a Co3+ lattice site sits in a cavity which is slightly too small to contain it. In other words it experiences an effective pressure. We have done a few model calculations using compressibility data for typical ionic transition metal compounds where such information is available (e.g.for Fe20, the constant p = -(l/Vo)(aV/ap)T is 6 . 0 ~ lo-’ and for FeCO, /? = 1-0 x and the effective “ internal pressure ” which one calculates is 70-100 kbar. It is thus possible considering the data of Drickamer et al. that one can account M. G. Clark J. Chein. Physics 1968 in press. M. G. Clark G. M. Bancroft and A. J. Stone J. Chem. Physics 1967,47,4250. A. R. Champion R. W. Vaughan and H. G. Drickamer J. Chem. Physics 1967,47,2591. G. K. Wertheim and R. H. Herber J. Chem. Physics 1963,38 2106. G. K. Wertheim W. R. Kingston and R. H. Herber J. Chem. Physics 1962 37 687. H. Sano and R. H. Herber to be published. 20 GENERAL DISCUSSION for the apparent presence of Fe2+ in a Co3+ matrix by postulating such a mechanism Whether or not this explanation will prove to be the correct one cannot be predicted but the idea seemed an interesting one which merits further experimental testing and discussion.Finally possibly one bonus of the Hazony-Drickamer idea is that it would account neatly for the observation that anomalous charge states have never been observed in the 119Sn isomeric transition decay although we have looked for such effects-for which the a priori expectation is very high due to the large internal conversion co- efficient of the 65 keV transition-for many years. Experiments both at liquid nitrogen and at room temperature with a wide variety of ionic covalent Sn2+ and Sn4+ sources have always been negative.l Dr. R. V. Parish (University of Manchester Inst. of Science and Technology) said Would Herber inform me whether there are any data for low-spin cobaIt(II1) com- plexes decaying to high spin iron(II1) complexes? This would be a good system to investigate since the change in ionic radii will be large.Prof. N. N. Greenwood (Newcastle upon Tyne) said Three items puzzle me about Herber’s ingenious explanation of the pressure induced simulation of a ferrous chemi- cal shift by a ferric ion (i) the pressure would relax within the time available (- lo-’ sec) since an acoustic wave traverses the dimensions of one unit cell in a crystal within about 10-l2 sec; (ii) the influence of cationic polarization on the precise position of the contiguous anions has been neglected in calculating the pressure Fe3+ is likely to be less polarizing than Co3+ and this would again tend to reduce the calculated pressure ; (iii) in non-cubic environments high spin Fe2+ can readily be distinguished from Fe3+ by its much larger quadrupole interactions. Is it not possible for the iron ion to reach its preferred oxidation state in the lattice by the same process that is involved in replacing the large number of electrons which ‘‘ boil off” from the ion in the cascade process which follows the nuclear electron capture event which transmutes 57C0 into 57Fe? * l €3. Yoshida and R. H. Herber unpublished results. * see also comments by Gallagher and by Herber p. 101-2.