Mössbauer effect studies of high-spin(5T2)⇌low-spin(1A1) transitions in transitions in frozen solutions of iron(II) complexes—I bis[2-(2-pyridylamino)-4-(2-pyridyl)thiazolato]iron(II)

Mössbauer effect studies of high-spin(5T2)⇌low-spin(1A1) transitions in transitions in frozen solutions of iron(II) complexes—I bis[2-(2-pyridylamino)-4-(2-pyridyl)thiazolato]iron(II)

J. inorg nucl (7hem.. 1977, Vol 3% pp. I I~l-lt3~. Pergamon Press. Printed in Great Britain MOSSBAUER EFFECT STUDIES OF HIGHSPIN(~T~).~LOW-SPIN('A,...

448KB Sizes 0 Downloads 2 Views

J. inorg nucl (7hem.. 1977, Vol 3% pp. I I~l-lt3~.

Pergamon Press.

Printed in Great Britain

MOSSBAUER EFFECT STUDIES OF HIGHSPIN(~T~).~LOW-SPIN('A,) TRANSITIONS IN FROZEN SOLUTIONS OF IRON(II) COMPLEXES--I BIS[2-(2-P YRIDYLAMINO)-4-(2-P Y R I D Y L ) T H I A Z O L A T O ] I R O N ( I I ) E. KONIG* and G. RITTER lnstitut fiir Physikalische und Theoretische Chemie and Physikalisches Institut, Abt. II, Universit~,t ErlangenNiirnberg, D-8520 Erlangen, W. Germany and H. A. GOODWIN School of Chemistry, University of New South Wales, Kensington, N.S.W.. Australia

(Received 2 September 1976) Abstract--The ~Fe M6ssbauer effect in frozen solutions of [~Fe(papt)2] is studied as a function of temperature. In a

DMF/glycerol glass between 77 K and the glass transition temperature T, ~ 160 K, the complex is in the 'A, ground state, whereas a high-spin(~T2)~low-spin('A0transition is found in the solid. Above T~, increasing conversion 'A, ~5T2 due to partial crystallization is observed. The Debye-Waller factor f shows a discontinuityof ~ 17% at T~. the Debye model being followed below T~ with O,~,~- 153.3K. In a CHCI~matrix and a CH3OH glass at 77 K, the complex is likewise in the 'A~ ground state. The high-spin(~T:)~low-spin('A,) spin equilibrium in solutions of [Fe(papt):] seems to be limited to temperatures between 160 and 233 K.

INTRODUCTION

In previous studies[I-4] we have shown that highspin(~T2)~low-spin(tA,)transitionsare responsible for the unusual magnetism encountered in certain iron(II) cornplexes of the ligand 2-(2-pyridylamino)-4-(2-pyridyl)thiazole (abbreviated paptH). The tridentate ligand may lose a proton from the bridging amino group and thus forms uncharged complexes of the type [Fe(papt)2]. This particular complex exists in two well-defined solvates, [Fe(papt)2]-4/3CHC13 and [Fe(papt)2]'C6H6 as well as in the unsolvated form. The 57Fe M/Sssbauer spectrum[I] of solid [Fe(papt)2] at 305K shows a single doublet (AEo=2.12+0.03mms ,' 8ts=+0.89_+0.04mms 1) typical for the high-spin ~~ ground state of iron(II). With decreasing temperature, a second spectrum (AEo = 1.54+0.04 mms -~, 6 ~s= +0.34-+ 0.05 mms-') characteristic for the low-spin JA~ ground state of iron(II) eventually appears. If the temperature is lowered further, the intensity of the ~A~ spectrum increases slowly, while that of the 5T2 spectrum is decreasing at the same rate. Below 100 K, no additional change in the spectrum is observed, the limiting area fraction at 4.2 K being, for a particular specimen of [Fe(papt)z], A(5T2):A('A~) =0.43:0.57. Evidently, the transition does not proceed to completion at low temperatures. Disregarding the effect of different Debye-Waller factors for the ~T2 and 'A~ ground states[l], the residual ST, fraction may be taken as n(STz)-0.43. For other samples, this value may be different, residual 5T2 fractions as low as n(ST2)=0.17 having been obtained for [Fe(papt)~.][1]. The ~7Fe M6ssbauer spectra of [Fe(papt)2].4/3CHCl3 and [Fe(papt)2]'C6H6 are similar, the n(*T2)-values at 4.2 K being 0.52 and 0.81, respectively, Numerous iron(II) complexes exhibiting highspin(~T2)~low-spin(IA,) transitions have been studied as solids within the last few years. However, the behaviour in these solid systems is often influenced by interactions

within the lattice, phase transitions of first order in combination with a high-spin(ST2)~-Iow-spin('A0 conversion being a particularly striking example[5-7]. It should be noted that the basic process is a thermodynamic equilibrium between the high-spin(ST2) and the lowspin(~A0 ground state of iron(II). It may be assumed that this equilibrium is indeed rapidly established if quasiisolated molecules in a dilute solution are considered. At present, detailed information is available for two iron(II) systems, whereby the first order rate constants for the processes k, 5T~ 'At. k , ' (1) have been determined. For the complex bis(hydrotris(parazolyl)borate) iron(II), Fe(HB(pz)3)z, measurements by a laser temperature-jump technique produced the values kt = 1 × 107 s t and k_~ = 2 × 107 s t in methanol-dichlormethane solution at 298 K[8]. For the complex [Fe(6-Mepy)2(py)tren](PF6)z where the hexadentate ligand is derived from tris{4-[(6-R)-2-pyridyl]-3aza-3-butenyl}amine, k~ - 6 × l05 s- ~ and k ~- 8 × l06 s were obtained in aceton-water solution at 294K[9]. Unfortunately, very few of the systems which show high-spin(STz)~low-spin(~A0 transitions in the solid state are sufficiently stable in solution to be studied by kinetic methods. Studies in solutions are nevertheless of prime importance in understanding spin equilibria. Spin conversions seem to be involved in the activity of certain metalloproteins[10, Ill as well as in intermolecular electron transfer processes[12]. In addition, the results of solution studies may contribute to the understanding of the more involved processes which are associated with spin transitions in solids. Recently, M6ssbauer effect studies on rapidly frozen solutions have provided important contributions to our knowledge of solution structure[13]. This is possible

1131

1132

E. K()NIG etal.

since, in terms of a widely discussed model[14, 15], a glass at a temperature T ~< Tg shows approximately the structure of the liquid at the glass transition temperature, Tr Thus, in order to obtain detailed information about spin equilibria in the liquid state, we have commenced M r s s b a u e r effect investigations on suitable glasses of those iron(II) complexes which are known to exhibit high-spin(ST2)~low-spin(1A0 transitions as solids. The present study deals, in particular, with investigations on frozen solutions of the complex [Fe(papt)2]. EXPERIMENTAL METHODS The complex bis[2-(2-pyridylamino)-4-(2-pyridyl)thiazolato]iron(II), [Fe(papt)2] was prepared by a modified and somewhat more efficient method than that previously reported [16]. Metallic iron (enriched by the isotope 57Fe to >90%) was treated with 1 M sulphuric acid until solution was complete and the stoichiometric amount of the ligand, paptH, dissolved in 1 M H2SO,, was added, Chloroform and then excess dilute ammonia solution were added to the mixture which was shaken vigorously. An intense red colour developed in the chloroform layer. This was separated, dried with anhydrous sodium sulphate and reduced in volume. The complex [Fe(papt)2].4/3CHCl3 crystallized on the careful addition of low boiling petroleum ether. It was recrystallized from a chloroform/petroleum ether mixture. The homogeneity and purity of the sample were verified by physical methods and results practically identical to those of unenriched samples[I,3] were obtained. The Mrssbauer spectrometer was of the constant acceleration type (Frieseke und Hoepfner FHT 800A) operating in the multiscaler mode. The source was 20-mCi ~Co in copper, giving a minimum observable line width of 0.23 mm s ~. The system was calibrated with a metallic iron absorber. All velocity scales and isomer shifts refer to the iron standard at 298 K. To convert to the sodium nitroprusside scale, add +0.257 mm s ~. Movement of the source toward the absorber corresponds to positive velocities. Variable temperature measurements between 77 and 309 K were performed using a small heating coil with the sample placed in a superinsulated cryostat. The temperatures were monitored by means of a calibrated copper/constantan thermocouple, a cryogenic temperature controller (Artronix model 5301-E) and liquid nitrogen as coolant. Measurements at 4.2 K were obtained with a separate spectrometer (Nuclear Data ND 2400, modulation of the pulse-height achieved by the driving wave form) and a suitable helium dewar. In order to obtain reliable values for the area fractions and hence for the effective thickness, all measurements were performed with the identical geometrical arrangement for source, absorber and detector. The resulting data were carefully corrected for nonresonant background of the -/-rays and fitted to Lorentzian line shape. The error bars shown in the spectra represent the statistical error only. Solubility and glass formation of [Fe(papt)~] were studied in various solvents and solvent mixtures. In most solvents where sufficient solubility was found, crystallization of the solute occurred at low temperatures. Thus in pure dimethylformamide (DMF), crystallization is observed at ~150K and in pure chloroform crystalline samples were obtained at 77 K. Glass formation suitable for the present purpose takes place if a DMF/glycerol mixture 9 : l v / v is employed and most measurements were performed in this solvent. Suitable absorbers were constructed using the following technique. Small disks of filter paper were soaked in the solution and glass formation was effected by fast dipping into liquid nitrogen and shaking. A complete absorber was then prepared from several disks partly separated by 5It aluminum foil. A typical arrangement thus comprised of 5/z AI--2 disks filter paper--5/, AI--2 disks filter paper--5/x AI--2 disks filter paper--5V, AI, the complete sandwich being placed in a small perspex container. During the procedure, the disks were kept and the assembling was performed, in liquid nitrogen. RESULTS The MiSssbauer spectrum of the solid complex [Fe(papt)2]'4/3CHC13 enriched to > 9 0 % in S7Fe was

measured between 85 and 309 K. The general behaviour of the spectrum conforms to that of unenriched [Fe(papt)2].4/3CHCl3, a high-spin(ST2)~low-spin(IAm) transition being involved over the temperature range 85-180 K, see Fig. 1. In particular, values of the quadrupole splitting AEo and the isomer shift a ~s for the two ground states ~T2 and ~A~ at the various temperatures are identical, within experimental uncertainty, to the values obtained previously[l]. The determination of area fractions was not attempted, the difficulties being high saturation corrections for the enriched absorber and the likelihood of granulation effects. The results obtained on a glass of the complex [Fe(papt)z].4/3CHCl3 in a DMF/glycerol mixture 9:1 v/v (6.5 mg solid in 0.5 ml solvent) are listed in Table 1. Values of the quadrupole splitting AEo and the isomer shift 8 ~s are not given in detail, since they correspond closely to the values found for the solid complex, viz. AEo :~,,;-.~.,',o~;,~. ~.,~. ~ ~,,~ _' e " ~; 180 K " : : i ! :: i : : : -: "! i: -~,~, ~: ~ a,a', ~ "~,/~, [ ~50 K ~ i J " ~i g_ ~ ~: .~ ~ ~ ,,"~,'~*~\ ~ ,-'~ #.. ;" ~ 12oK , " : i ' ii ~: ,." ~""~ 85 K

• ~ ~" '~' " ~ : ~; ~ : ! ; -; "

: ~: -3 -2 -1 0 ,1 ,2 ,3 Velocity (minis) Fig. 1. ~TFe Mfssbauer spectra of solid [Fe(papt)2]'4/3CHCl3 enriched to >90% in ~Fe at various temperatures between 180 and 85 K. Table 1. ~Fe M/~ssbauer effect data for [Fe(papt)2] in frozen solution of DMF/glycerol 9 : I v / v T Area fractions K A(~T2) A(~A~) 85 100 120 140 150 155 160 165 170 175 180 185 190

0.030 0.035 0.044 0.056 0.065 0.072 0.083 0.145 0.161 0.180 0.205 0.238 0.291

0.970 0.965 0.956 0.944 0.935 0.928 0.917 0.855 0.839 0.820 0.795 0.762 0.709

ts~2

t~

ts~-~ ts~2+t~A~

-ln.f,o,~

0.028 0.031 0.035 0.041 0.044 0.046 0.047 0.095 0.106 0.114 0.120 0.114 0.105

1.130 1.048 0.911 0.806 0.713 0.665 0.574 0.621 0.613 0.567 0.504 0.385 0.265

0.024 0.029 0.037 0.048 0.058 0.065 0.076 0.133 0.147 0.167 0.192 0.228 0.284

0.499 0.569 0.701 0.812 0.924 0.986 1.122 0.979 0.975 1.030 1.117 1.341 1.640

Mossbauer effect studies of high-spin('T2)~low-spin('A,) transitions in frozen solutions of iron(ll) complexes--I (~T2)=2.25-0.05mms -~. 8tS(ST2)=+0.91-+0.06mms ' and AEo(~A,)=l.53+-O.O4mms -', 8~S(~A0 = +0.34 -+0.05 mm s ', almost independent of temperature within the range studied. I n T a b l e l , t,r. andt,A, denotethe effective thickness for the ~T: and 'A, ground state, respectively. The values of these quantities were obtained from the normalized individual areas A(ST2)and A(IAO according to A(ST2) = l/2rrf~FL(Vr,) (2)

A(~At)=I/2~f~FL(ttA,) where F is the line width, fs the Debye-Waller factor of the source, and the saturation function L(t) is well approximated by

L(t)=t/(l+O.25t).

(3)

The factor 1/2*rfsF in eqn (2) has been determined by measurements on sodium nitroprusside absorbers of known effective thickness. Also, the effective thickness t is related to the product of the site fraction n and the Debye-Waller factor f of the corresponding molecular ground state according to Vr~ = 1/2N~do'of, r~n~r2 (4)

tL~=l/2N~do'of, A~n,a,. Here, N denotes the number of iron atoms per unit volume, /3 the isotopic abundance, d the absorber thickness and o'o the resonant cross section. In the glasses studied at present, the direct determination of n~r~and n'A, is not practicable. If, for the sake of argument, the possible difference [1] between f'r, and f'z, is neglected, an observed variation in VT-,and t'A, will reflect essentially the variation of n,r, and n'A,, respectively. The values, in Table 1, for 85 K were obtained after temperature stabilization immediately succeeding the glass formation, The subsequent data were collected by progressive increase in the temperature. Some interesting results

should be evident from curve a in Fig. 2, where the relative effective thickness for the ST,_ ground state, t~r~/t'T,+t'a,, has been plotted for increasing temperatures. Thus at 85K and somewhat above, the 5T~ fraction is negligible, viz. t,.r,Jtto~=O.024 at 85 K. It follows that, contrary to the result for the solid where a substantial residual contribution of the ST, state is found[l], the complex in the DMF/glycerol glass is, at low temperatures, essentially in the IA, ground state. There is a slight, though not significant, increase in t~T,/t,,~ with rising temperature, compare vr~/t,,,~= 0.058 at 150 K. At the glass transition temperature T ~ 1 6 0 K , transformation of the 'A~ into the 'T2 state sets in suddenly, compare vr~/t,o, = 0.133 and continues to increase if the temperature is raised further, compare th,jt,o, =0.284 at 190 K. At even higher temperatures, broadening of the spectra occurs due to increased diffusion and softening of the glass, the m.p. T,,, being around 200K. If now the solution is slowly cooled again, curve (b) in Fig. 2 results. Evidently, on slow cooling, crystallization sets in. the ratio of the ground state fractions adjusting, at each temperature, approximately to the fraction found in the solid. Although the situation in the absorber formed of filter paper soaked with the solution is certainly more complicated, the area fractions are not much different from those of solid [Fe(papt)2], absorber B, of our previous study[l[. Obviously, the small amount of chloroform from the solid [Fe(papt)2].4/3CHC13 cannot be of much influence after crystallization from the DMF/glycerol mixture. In Fig. 3,-In f,,,~,,is plotted as a function of temperature for the complex [Fe(papt)2] in DMF/glycerol mixture. Evidently, if a glass is formed as discussed above, -In f,o,,~ shows a discontinuity at the glass transition temperature, T, (see Fig. 3, curve (a)). The quantity f,o,,~, the total Debye-Waller factor of the system, has been obtained by a suitable standardization of the total effective thickness, V r : + t~A,, see eqn (4). It should be noted that, in the absorbers used for the present study, the quantity N • d is not easily available. A plot of -ln(VT2+ t,A,) vs ternperature produces, however, a straight line for 85 ~< T~< 140 K. Since, in the high-temperature limit of the Debye

o3

t5T2/ttot

0.1

0

I

I

1133

I

I

100 150 T,K 200 Fig. 2. Relative effective thickness VT2lt~T2+ t'A, for [Fe(papt)2] in a frozen solution of DMF/glycerol as function of temperature: curve (a) for rising temperatures, curve (b) for falling temperatures.

E. KONIG et al.

1134

1.6

-tn ftot 1.2

o 0.8 8 -- 153.3

. ~'~'" 0

#~'~" 0

~ ' 8 = 169K I 50

I 100

I "150

I 200

T,K

250

Fig. 3. Total Debye--Wallerfactor -In f,o~ for [Fe(papt)2]in a frozen solution of DMF/glycerolas function of temperature: curve (a) for rising temperatures, curve (b) for falling temperatures. model, it is -In [ = ~

(5)

the values of -In/tot~ have been obtained, in reasonable approximation, by shifting the straight line to zero intercept. The magnitude of the shift determines N x d (viz. 0.0704 mg cm-257Fe, i.e. 71% of the initially dissolved 57Fe in terms of the compound) and thus the values of - l n f~o~lfor any temperature easily follow. In eqn (5), Eo is the energy.of the y-ray, O the Debye temperature and, for simple atomic lattices, M is the mass of the absorbing atom. If, in the present case, M is identified with the mass of the ~TFeatom, O,A, = 153.3 K is obtained over the temperature range 85 to ~ 140 K. This value should be compared with the Debye temperature O,a~ = 165 K obtained for solid [Fe(papt)2] [1]. However, it should be observed that for a complicated molecule like [Fe(papt)2], a value for M higher than that employed above might be more appropriate, For falling temperatures, -In ~ota, shows no discontinuity and does not follow the Debye law in any range of temperature (see Fig. 3, curve (b)). It should be noted that, in solid [Fe(papt)2], the total Debye-Waller factor also could not be reproduced by the Debye model. This behaviour seems to be caused by the significantly

different[valuesfortheST2andIA, ground states in these systems. The observation of a discontinuity in the Debye--Waller factor at T~ conforms to the results obtained with frozen solutions of various other systems[13], and seems to be a general property of glasses. However, in contrast to studies on frozen aqueous solutions[13], no anomaly in the quadrupole splitting AEo or the line width F has been detected in the present system at T~. In a chloroform matrix at 89 K, the complex [Fe(papt)2].4/3CHC13 produced a clean M6ssbauer spectrum of the ~A~ ground state. The same result was obtained in a methanol glaSs~t 77 K, although irreversible formation of some 5T2 grotind State molecules occurred if

the system was warmed up to ~140K. Recall that the solid [Fe(papt):]-4/3CHC13 gives [1], at 77K, an area fraction for the high-spin(ST2)ground state, A(ST2) = 0.62. Magnetic hyperfine interactions were studied in a longitudinal magnetic field of 40kG at 4.2K. For [Fe(pap02] in DMF/glycerol, the sign of the electric field gradient was determined as V=(1A1)>0. This result is in full agreement with the finding on solid [Fe(papt)2][1]. DISCUSSION The results of Fig. 2, curve (a), clearly demonstrate that in the DMF/glycerol glass the complex [Fe(papt)2] is in the 1At ground state at all temperatures T ~
M6ssbauer effect studies of high-spin(~Tz)~low-spin('A,) transitions in frozen solutions of iron(II) complexes--I . . . . . . . .,,~,, ~ 85 K ! " ~ ~ : .. "~'~J 1~0 K

i\" f'v"~ .... •• ~:~ : " ~ ' ~ K • '~ " a : ~ ;:~,. ;,£, ~ ~ ~ -, .:,., 185K ,~ g : .f ~: ~, ,' :: '~ ~r ~ '~"~""~'*'~ ,/~¢'~"~' r~', -85K ~ ~! ; :: : : : . ,r~.-~;~,.9,~,~e,~r~. " 200 K "

85 K

: ~ :! ' ::

'. , :! :: . :

[Fe(papt)2] is high-spin in chloroform solution at 290 K, viz. /zen=5.15B.M.[18]. It follows that a highspin(ST2)~low-spin(~A,) transition is involved in solution, conversion to the ~A, ground state geing completed at or above T~. This result conforms with NMR studies of [Fe(papt)2] in chloroform solution[19] where a reversal of the change in shifts of ligand protons has been observed at about 233 K. This observation has been interpreted by an increasing conversion ~T2~'A~ at this temperature which is well above T~. It may then be concluded that, in solution, the high-spin(ST2)~low-spin('A3 transition in [Fe(papt)2] is taking place essentially between 160 and 233K. The apparent difference, apart from the temperature range, in the spin transition for samples in the solid state and in solution is that the conversion to the 'At state (with lowering of temperature) is virtually complete in solution. At least for this system then, the retention of a residual 5T2 fraction is a feature of the lattice of the solid. However, it has been demonstrated elsewhere[20] that there is, in addition, a fundamental difference in the mechanism of spin transitions in the solid state and in solution. Whether this difference is of general applicability is a subject for further study. Acknowledgements--The authors appreciate financial support by the Deutsche Forschungsgemeinschaft and the Bundesministerium fiir Forschung und Technologie. H.A. Goodwin is grateful to the Alexander-yon-Humboldt Foundation for an exchange fellowship.

-3 -2 -1 0 *1 +2 *3 Vetocity (mm/sl Fig. 4. 57Fe M6ssbauer spectra for [Fe(papt)2] in DMF/glycerol solution for a sequence of temperatures: 85 K, 140 K, 185 K, 85 K, 200 K (> Tin)and 85 K. The spectra were obtained in the sequence from top to bottom. ground state is obtained. When the temperature is increased over 140 to 185 K a superposition of the spectra of the JA~ and 5T2 ground states where t,T:/t,o, = 0.31t is obtained. Observe that T~ < 185 K < Tin. Cooling to 85 K then gives a spectrum with smaller though notable 5T2 contribution, viz. t~T2/t,o,=O.14. Cycling of the temperature between 185 and 85 K shows that these results are completely reproducible. If finally the temperature is raised to 200K, i.e. above T,,, with subsequent fast cooling to 85K, an almost pure spectrum of the 'A~ state is again observed. Clearly the crystalline phase once formed above T~ cannot disappear by lowering the temperature to 85 K, whereas cooling from above T,, produces a glass similar to that obtained initially. In the frozen solution, the spectrum shows that the 'A, ground state is alone present, as found before under the same conditions. A high-spin(ST0~low-spin('A,) transition is associated, in general, with a significant change of the optical spectrum[17]. The solid complex [Fe(papt)2] is known[3] to be red-brown at room temperature and, at 77 K, is much more intensely coloured and red-purple. A similar change of colour on cooling has been observed also for [Fe(papt)2].4/3CHC13 in the various solvents investigated in the present work. It is known from magnetic data that

tThe difference to the corresponding value of Table 1 is due to the fact that a lower concentration of glycerol was used.

1135

REFERENCES 1. E. K6nig, G. Ritter and H. A. Goodwin, Chem. Phys. 5, 211 (1974). 2. E. K6nig, G. Ritter and H. A. Goodwin, Chem. Phys. Lett. 44, 100 (1976). 3. R. N. Sylva and H. A, Goodwin, Aust. J. Chem. 21, 1081 (1968). 4. R.N. Sylva and H. A. Goodwin, Aust. J. Chem. 2 0 , 4 7 9 (1967). 5. E. K6nig and K. Madeja, lnorg. Chem. 6, 48 (1967). 6. E. K6nig, K. Madeja and K. J. Watson, J. Am. Chem. Soc. 90, 1146 (1968). 7. E. K6nig, G. Ritter and B. Kanellakopulos, Z Phys. C 7, 2681

(1974).

8. J. K. Beattie, N. Sutin, D. H. Turner and G. W. Flynn, J. Am. Chem. Soc. 95, 2052 (1973). 9. L.J. Wilson, D. Georges and M. A. Hoselton, lnorg. Chem. 14, 2968 (1975). 10. P. M. Champion, E. M/inck, P. G. Debrunner, P. F. Hollenbergand L. P. Hager, Biochemistry 12, 426 (1973). 11. T. lizuka, M. KotaniandT. Yonetani, J. BioLChem. 246,4731 (1971). 12. H. Taube, Electron Transfer Reactions of Complex Ions in Solution. Academic Press, New York (1970). 13. 1. Dezsi, M6ssbauer Effect Studies on Frozen Solutions, unpublished manuscript, presented at the Intern. Conf. Appl. M6ssbauer Effect, Israel (1972). 14. S. L. Ruby, Perspectives in M6ssbauer Spectroscopy (Edited by S.G. Cohen and M. Pasternak),p. 181. Plenum Press, New York (1973). 15. H. Rawson, Inorganic Glass-Forming Systems. Academic Press, New York (1967). 16. H. A. Goodwin, Aust. J. Chem. 17, 1366 (1964). 17. E. K6nig, Ber. Bunsenges. Phys. Chem. 76, 975 (1972). 18. H. A. Goodwin and D. W. Mather, Aust. J. Chem. 25, 715 (1972). 19. H. J. Keller, K. E. Schwarzhans, H. A. Goodwin and R. N. Sylva, Z. Natufforsch. 24B, 1058 (1969). 20. B. Kanellakopulos, E. K6nig, G. Ritter and W. lrler, J. Phys. (Paris) 37 suppl. C6-459 (1976).