The Nephelauxetic Effect — Calculation and Accuracy of the Interelectronic Repulsion Parameters II

BAND 27 b ZEITSCHRIFT FÜR NATURFORSCHUNG HEFT 1

The Nephelauxetic Effect — Calculation and Accuracy of the Interelectronic Repulsion Parameters II. Application to d3 and d7 Single Crystal Spectra at Cryogenic Temperatures *

E. KÖNIG

Institut für Physikalische Chemie II, Universität Erlangen-Nürnberg, 8520 Erlangen, Germany

(Z. Naturforsch. 27 b, 1—5 [1972] ; received September 28, 1971)

Expressions are reviewed which may be used to determine 10 Dq and B from the spin-allowed bands in the optical spectra of d3 and d7 electron systems within octahedral and tetrahedral sym metry. Application to low-temperature single crystal spectra demonstrates that (i) the semi-empirical ligand field theory reproduces transition energies with sufficient accuracy; (ii) differences in the values of 10 Dq and B observed with different fitting methods may be attributed to the in- accuracy of experimental data; (iii) there are generally valid values of B35 and /?33 for each complex ion.

The semi-empirical ligand field theory provides those complex ions are considered where all three means to completely determine the electronic d — d spin-allowed d — d bands are observed. Room tem- spectra of transition metal ions of octahedral sym- perature solution and single crystal spectra of al- metry in terms of three parameters: the octahedral most fifty complexes and impurity ions of the splitting parameter 10 Dq ( = A) and the inter- transition metals were subject of the analysis. The electronic repulsion parameters (= Racah para- results 4 may be summarized as follows: meters) B and C which are linear combinations of (i) The accuracy of B and 10 Dq depends on the the Condon-Short ley parameters F2 and 1. method adopted to their calculation. Conse- An analogous statement applies to tetrahedral sym- quently, certain methods may be selected which metry and one or two additional parameters (e. g. provide the "best" possible fit to the experi- Ds and Dt in tetragonal environment) are required mental data; if the symmetry is lower than cubic. In general, the (ii) Given a specific method of calculation, an un- numerical values of the parameters B and C, as systematic variation in the deviations between determined from d — d spectra, are lower than the calculated and observed transition energies is values in a free transition metal ion. This observa- often encountered. It was suggested that this tion is well known as the nephelauxetic effect 2' 3. is due to insufficient accuracy of the experi- In the first part of this study 4, the author has re- mental room temperature (solution) data. cently reviewed and tested methods which may be In the present contribution, the same methods as used to determine 10 Dq, B, and C from electronic used previously4 will be applied to single crystal spectra. To focus the attention on the value of B, spectra measured at cryogenic temperatures. It will these methods were applied to the spin-allowed d — d be demonstrated that results somewhat different bands in high-spin d2, d3, d7, and d8 complexes of from those of room temperature spectra are ob- octahedral and tetrahedral microsymmetry. In the tained. expressions of the corresponding transition energies, the parameter C does not occur5. In addition, I. Ligand Field Theory of d3 and d7 Ions 10 Dq may always be fixed by a suitable choice of in Cubic Fields the calculation method. A convenient check on the accuracy of the employed numerical procedure is The general treatment of ligand field theory is provided by calculating the extra band energy, if adequately covered in several textbooks 7-9 to which

Requests for reprints should be sent to Doz. Dr. E. KÖNIG, * For the first part of this study refer to E. KÖNIG, Struct. Institut für Phys. Chem. II der Universität Erlangen-Nürn- Bonding 9, 175 [1971]. berg, D-8520 Erlangen, Fahrstr. 17.

G ZT-I'H)^

Dieses Werk wurde im Jahr 2013 vom Verlag Zeitschrift für Naturforschung This work has been digitalized and published in 2013 by Verlag Zeitschrift in Zusammenarbeit mit der Max-Planck-Gesellschaft zur Förderung der für Naturforschung in cooperation with the Max Planck Society for the Wissenschaften e.V. digitalisiert und unter folgender Lizenz veröffentlicht: Advancement of Science under a Creative Commons Attribution Creative Commons Namensnennung 4.0 Lizenz. 4.0 International License. E. KÖNIG reference is made here. With respect to the single (b) fitting the first and third band, crystal data available at present, we will concentrate 10 Dq = 2 ?>! — j>3 + 15 B, (8) primarily on the theory of d3 and d7 ions in cubic fields. The relevant energy expressions have been B = jo[-{2Vl~ ± ^ ~ + *»* + "1 >'3>,/!] , derived previously4. For convenience, we will (c) fitting the second and third band, briefly introduce those quantities and list explicitly those expressions which will be needed in the sub- 10 Dq= I (2 v2 — v3) +55, sequent numerical calculations. 3 ß= 2 Thus in the octahedral d configuration three 510 2) ±3{81rs -16^(r2-ra)}''•]. 4 spin-allowed transitions from the A2g ground state (9) to the excited states 4T2g, a 4Tlg, and b 4Tig are (d) fitting the difference between the first and expected. Within the approximation considered second band, here, the energy of the lowest transition is always 10 Z)g = — J'i, determined as vx (4A2g —> 4T2g) = 10 Dq. The energies of the two higher transitions follow from B=(v2 + V3-3V1)/15. (10)

>'2,3= 2 (15B + 3O0<7)+-2 t(15ß-10^)2 II. Application to Single Crystal Spectra + 12 B • 10 Dq]1'1. (1) In order to asses the accuracy of the parameter The parameter B may then be obtained according values of 10 Dq and B, the equations listed in sec- to four different methods: tion I will be applied below to some recent low tem- (a) fitting the second band, perature single crystal spectra. Following the first 2 2 5 = (2v1 + v2 -3vlv2)/ (15 v2 — 21 vt), (2) part of this study 4 it will be assumed, for the sake (b) fitting the third band, of argument, that the three-parameter (10 Dq, B, C)

B = (2 Vi2 + f32 - 3 n r3) / (15 v3 - 27 vt), (3) theory is valid exactly. The question then arises about the significance of the calculated transition (c) fitting the sum of the second and third band, energies. In ligand field theory, all energy dif- B=(v2 + r3-3v1)/15, (4) ferences are calculated at a constant value of 10 Dq (d) fitting the difference between the second and (viz. "vertical" transitions in the Tanabe-Su- third band, g a no diagram). Since the relation 10 Dq ~ R~5 holds to a reasonable approximation 10, this is equi- B = ^ [3 f! ± (25 0>3 - v2)2 -16 V)I/2] • (5) valent to a fixed metal-ligand distance, R. In the spin-allowed d — d transitions considered here, the The expressions (1) to (5) apply to tetrahedral d7 states involved originate in different strong field ions as well. configurations g eg and, consequently, the potential In the octahedral d7 configuration, the ground minima of the excited state and the ground state do state is a4Tlg and the excited quartest states are, in not coincide. The calculated transition energy cor- the order of increasing energy, 4T2g, 4A2g, and responds, therefore, to the energy of a transition b 4Tlg . The energy of the three spin-allowed transi- from the zero-point vibrational level of the elec- tions is determined according to tronic ground state to an excited vibrational level of the excited state (cf. "vertical" transition ac- "i (a4Tig 4T2g) = I (10 Dq -15 B) + , cording to the Franck-Condon principle). v2 (a4Tig -> 4A2g) = vx +10 Dq , (6) As far as the comparison between theoretical and r3(a4Tlg-^b4Tig) = [(10Z)q + 15 5)2 experimental energies is concerned, two limiting — 12 5-10 Dq]1/!. conditions may be distinguished: There are again four different methods which may (i) In centrosymmetric (e.g. octahedral) com- be employed to obtain the parameters 10 Dq and B: plexes, all d — d transitions are rigorously forbidden (a) fitting the first and second band, on the basis of parity. The forbidden electronic

10 Dq — v2 — vl, transitions may gain intensity through coupling n B= (2v12-v1v2)(Uv2-27 vt), (7) to odd vibrations (vibronic mechanism ). At low THE NEPHELAUXETIC EFFECT 3 temperatures, each band thus consists of a progres- III. Results and Discussion sion in one or more even vibrations superimposed upon one quantum of the odd ("permitting") vibra- Results of the present analysis are compiled in tion. The no-phonon (0" —> 0') band is absent or Tables 1 to 3. For each compound, experimental of very weak intensity 12. Therefore, within reason- transition energies determined according to sec- able approximation, the calculated vertical transition tion II are listed in line 1. Subsequent lines contain energy should be associated with the maximum of the calculated transition energies, their deviation the vibronic band determined, in principle, at 0 °K. from the corresponding experimental value, dv — (ii) In non-centrosymmetric (e. g. tetrahedral) ''calc (in cm 1 and in percent), and the values complexes, the d — d transitions become partly al- of the parameters B3- and ßS5 . In Table 2, values lowed on account of mixing with odd-parity states of 10 Dq and of the deviation, (5(10Z)g), from the of the central ion (e.g. p states11). One observes, value of v2 — vt are listed in addition. Each line ap- at low temperatures, a progression in the totally plies to a different method marked with reference symmetric vibrational mode originating in the no- to section I. phonon (0"—>-0') band. Normally, the highest In the octahedral d3 configuration, the only well intensity would be expected in one of the higher evidenced low-temperature single crystal spectrum 4 4 vibrational sub-bands 0"->n/eveil. However, in where the A2g —> b Tig transition has been unequi- the example studied at present13'14, the most pro- vocally assigned is that of VC12 16'17. If the quasi- minent band is associated with the no-phonon molecular model is assumed type (i) behaviour transition. This situation is encountered if ground is expected. No fine structure has been observed in and excited state potential minima occur at the same the 4A2g —> 4T2g and 4A2g a4Tig bands, and thus internuclear distances12. It has been suggested13, the band maxima listed in Table 1 are associated therefore, that the geometry of the Co2® ion in the with the vertical transition energies. In the 4A2g-> relevant excited states is not greatly different from b4Tlg band where a vibrational progression in a that in the ground state. Here it may be more ap- mode of ~ 234 cm-1 is encountered, the first pro- propriate to approximate the calculated vertical minent vibrational component (viz. 22,244 cm-1) transition energy by the average of the energies of seems to well approximate the centroid of the band the no-phonon bands in the most intense spin-orbit (viz. Fig. 1 of 1. c.18). The parameter values calcu- components determined again, in principle, at 0 °K. lated by KIM et al.16 are incorrect, since applying If the temperature is increased, in both cases eq (1) the fit was based on the estimated energy of higher vibrational levels of the electronic ground the no-phonon transition. No suitable data on the state become successively populated and the cor- d3 configuration in tetrahedral environment are responding band in the electronic spectrum is pro- known. gressively shifted to lower energy 15. The magnitude Recently, low temperature single crystal spectra of the shift is dependent on the distribution of the of RbCoCl3 and of the Co2® ion in several chloride vibrational levels in the electronic ground state and lattices became available 19. The a4Tig ground state on the intensity of the resulting hot bands and will of the octahedral d7 configuration is split by spin- differ from compound to compound. Therefore, in orbit interaction into the -T6, r8a, .T8b and .T7 general, spectra measured at higher than cryogenic levels in the order of increasing energy20. Since temperatures cannot compare favorably with theory. and are Kramers doublets, the a4Tig term

Compound Method vi, cm-1 j>2, cm-1 cm-1 öv B35, cm-1 ^35

4A2g 4T2g 4A2g^a4Tlg 4A2g —> b 4Tig [cm-1] [%]

VC12 (22°K) exp 9300 14,220 22,244 a 10 Dq flitted 22,231 - 7 0.03 570.1 0.74 b 10 Dq 14,226 fitted + 6 0.04 571.3 0.75 c 10 Dq 14,224 22,240 ± 4 0.02 570.9 0.75 d 10 Dq 14,231 22,255 + 11 0.06 572.3 0.75

Table 1. Observed and Calculated Transition Energies of Octahedral Vanadium(II) (B^*® = 766 cm-1). 4 E. KÖNIG

Compound Me- vi, cm-1 i>2, em-1 j'3, cm-1 6 V £35 ßäö 10 Dq d (10 Dq) thod a4Tlg^4T2g a4Tlg->4A2g a4Tig->b4Tig [cm"1] [%] [cm-1] [%]

RbCoCls exp. 6450 13,500 17,210 b fitted 13,855 fitted + 355 2.56 781.0 0.80 7405 + 355 4.79 c 6278 fitted fitted - 172 2.74 791.8 0.82 7223 + 173 2.4.3 d 6134 13,184 fitted - 316 3.77 757.3 0.78 7050 0 0

KMgCl3: Co2© exp. 5740 12,000 16,475 b fitted 12,367 fitted + 367 2.97 774.8 0.80 6627 + 367 5.54 c 5563 fitted fitted - 177 3.18 785.8 0.81 6437 + 177 2.7.3 d 5415 11,675 fitted - 325 4.39 750.3 0.77 6260 0 0 CsMgCls: Co2© exp. 6450 13,300 17,190 b fitted 13,854 fitted + 554 3.99 779.6 0.80 7404 + 554 7.48 c 6181 fitted fitted - 269 4.35 796.5 0.82 7119 + 269 3.78 d 5957 12,807 fitted - 493 6.06 742.7 0.77 6850 0 0 MgClo : Co2© exp. 6450 13,570 17,025 b fitted 13,849 fitted + 279 2.01 768.2 0.79 7399 + 279 3.77 c 6314 fitted fitted - 136 2.15 776.8 0.80 7256 + 136 1.87 d 6202 13,322 fitted - 248 2.93 749.7 0.77 7120 0 0

CdCl2: Co2© exp. 6040 12,500 16,930 b fitted 13,001 fitted + 501 3.85 787.4 0.81 6961 + 501 7.20 c 5797 fitted fitted - 243 4.19 802.5 0.83 6702 + 242 3.61 d 5596 12,056 fitted - 444 5.81 754.0 0.78 6460 0 0

CsCdCl3: Co2© exp. 5263 11,236 16,400 b fitted 11,376 fitted + 140 1.23 799.1 0.82 6113 + 140 2.29 c 5196 fitted fitted - 67 1.29 803.2 0.83 6040 + 67 1.11 d 5140 11,113 fitted - 123 1.75 789.8 0.81 5973 0 0 LiCl: Co2© exp. 6386 13,570 17,300 b fitted 13,727 fitted + 157 1,14 791.2 0.82 7341 + 157 2.14 c 6310 fitted fitted - 76 1.20 796.0 0.82 7260 + 76 1.06 d 6247 13,431 fitted - 139 1.63 780.8 0.80 7184 0 0

Table 2. Observed and Calculated Transition Energies of Octahedral Cobalt (II) in Chloride Lattices at 5 °K Co2® (ßfree = 971cm-).

Compound Method v\, cm-1 i>2, cm-1 j% cm-1 ö v B35, cm-1 ^35 4A2-^4T2 4A2 a4Ti 4A2->b4Ti [cm"1] [%]

ZnO: Co2© exp. 4140 7186 15,384 (4.2°K) a 10 Dq fitted 17,108 + 1724 10.08 790.4 0.81 b 10 Dq 7106 fitted - 62 0.87 671.4 0.69 c 10Dq 7109 15,443 ± 59 0.61 675.5 0.70 d 10 Dq 7103 15,319 - 65 0.67 666.9 0.69

ZnAl204: Co2© exp. 4015 6914 16,210 (1.8°K) a 10 Dq fitted 15,467 - 743 4.80 689.1 0.71 b 10 Dq 6940 fitted + 26 0.38 740.3 0.76 c 10 Dq 6939 16,185 ± 25 0.26 738.6 0.76 d 10 Dq 6941 16,237 + 27 0.28 742.2 0.76

Table 3. Observed and Calculated Transition Energies of Tetrahedral Cobalt (II) in Oxidic Lattices (#free — 971 cm *). has been effectively stabilized against the action estimated. It should be observed that, due to its low of the Jahn-Teller effect. Thus, assuming type intensity ("two-electron jump"), a high experimen- (i) conditions, the experimental energies included tal uncertainty should be assumed for the in Table 2 were obtained directly from the reported a4Tlg (t|g e2 ) 4A2g (4 e4) band. spectra. If a vibrational structure was observed (viz. In the single crystal spectra of tetrahedrally co- 20 the a4Tlg->4T2g band in CsMgCl3 : Co2® and in ordinated Co ions, type (ii) conditions are pre- LiCl : Co2®), the centroid of the vibronic band was valent and the experimental energies are determined THE NEPHELAUXETIC EFFECT 5 accordingly. Thus, in ZnO: Co20, it is the energies taken into account, the results of octahedral Co2® of the highest intensity (no-phonon) vibrational ions are similar to those discussed above. bands which are listed in Table 3, since individual vibrational components at higher energies are con- IV. Summary and Conclusions siderably weaker in intensity14. In ZnAl204 : Co2®, the transition energies employed in Table 3 are We have reviewed expressions which were derived mean values of the most intense no-phonon bands in previously, on the basis of the semi-empirical ligand each of the electronic transitions13. It should be field theory, to determine 10 Dq and B from the kept in mind that, due to the complicated structure spin-allowed bands of d3 and d7 electron systems of of the bands, the totality of the weak vibrational octahedral and tetrahedral stereochemistry. These transitions may still contribute a significant fraction equations were applied to low-temperature single to the overall transition energy 21. crystal spectra of suitable compounds and the extra Inspection of the results which have been col- band energy was calculated. lected in Tables 1, 2, and 3 reveals considerable Provided the spectral data employed are repre- differences to the results from room temperature sentative for most systems of the studied electron (solution) spectra 4. If those results are disregarded configuration, the conclusions arrived at are as fol- which were obtained by the most unfavorable lows : methods [viz. (a) in Table 3], the agreement be- (i) the semi-empirical ligand field theory repro- tween calculated and observed transition energies is duces quite accurately the transition energies, surprisingly good. On the other hand, there is not at least in spin-allowed bands of d3 and d7 much that could be gained by choosing between systems of cubic symmetry; methods (b), (c), and (d) in octahedral d3 and (ii) the differences in the parameter values 10 Dq tetrahedral d7 spectra. Most important, the deviation and B resulting from the application of dif- between the values of B35 calculated by different ferent fitting methods are due essentially to in- methods is negligible and that of ß35 is practically accuracy of the experimental data; non-existent. In Table 2, differences between results (iii) there exist generally valid values of the inter- obtained by methods (b), (c), and (d) are some- electronic repulsion and nephelauxetic para- what larger but still considerably smaller than in the meters B35 and ß35 for each complex ion which, room temperature spectra 4. In addition, the smallest however, may be determined only if sufficient- differences are encountered in those spectra where ly accurate experimental data are available. This study has been sponsored in part by research the best resolution in the a4Tl!? 4A2g. band has grants of the Deutsche Forschungsgemeinschaft, the been achieved, cf. CsCdCl3 : Co2® and LiCl : Co2®. Stiftung Volkswagenwerk, and the Fonds der Chemi- Thus, if the complications inherent in the data are schen Industrie whose support is gratefully appreciated.

1 E. U. CONDON and G. H. SHORTLEY, The Theory of Ato- 11 C. J. BALLHAUSEN, Progr. inorg. Chem. 2, 251 [I960], mic Spectra, Cambridge University Press, Cambridge 12 G. HERZBERG, Molecular Spectra and Molecular Structure, 1959. Vol. 3, Van Nostrand, New York 1966. 2 C. K. JORGENSEN, Progr. inorg. Chem. 4, 73 [1962]. 13 J. FERGUSON, D. L. WOOD, and L. G. VAN UITERT, J. chem. 3 C. E. SCHÄFFER and C. K. JORGENSEN, J. inorg. nuclear Physics 51, 2904 [1969]. Chem. 8, 143 [1958]. 14 R. PAPPALARDO, D. L. WOOD, and R. C. LINARES, J. chem. 4 E. KÖNIG, Struct. Bonding 9, 175 [1971]. Physics 35, 2041 [1961]. 5 In this way certain problems associated with the treatment 15 J. LEE and A. B. P. LEVER, J. molecular Spectroscopy 26, of C are eliminated. Thus there is evidence that the para- 189 [1968]. meter C is not susceptible to the nephelauxetic effect to the 16 S. S. KIM, S. A. REED, and J. W. STOUT, Inorg. Chem. 9, same extent as B 6. 1584 [1970]. 8 H. WITZKE, Theor. chim. Acta 20, 171 [1971]. 17 Recently, SMITH 18 likewise studied single crystal spectra 7 C. J. BALLHAUSEN, Introduction to Ligand Field Theory, of VCl, and of the V2® ion in several chloride lattices at McGraw-Hill, New York 1962. 6 °K. However, since quantitative energies were not re- 8 J. S. GRIFFITH, The Theory of Transition Metal Ions, ported, these results could not be included into the present University Press, Cambridge 1961. study. SMITH 18 claims that ligand field theory fits his re- 9 H. L. SCHLÄFER U. G. GLIEMANN Einführung in die Ligan- sults rather well. This would support and extend our pre- denfeldtheorie, Akademische Verlagsgesellschaft, Frank- sent conclusions. furt a. M. 1967. 18 W. E. SMITH, J. diem. Soc. [London] A 1969, 2677. 10 E. KÖNIG and K. J. WATSON, Chem. physic. Letters 6, 457 [1970]. 6 H. BREUER UND H.-H. PERKAMPUS

19 F. C. GILMORE, US At. Energy Comm. ORNL-TM-2507 21 Apparently, the centers of gravity reported 14 or estimated 4 [1969], in these two tetrahedral cobalt (II) systems are not ap- 20 A. D. LIEHR, J. physic. Chem. 67, 314 [1963]. propriate to the present analysis.

Auswertung nmr-spektroskopischer Messungen der Mischassoziation yon Nucleosid-Derivaten bei vergleichbaren Konzentrationsverhältnissen

Evaluation of NMR-spectroscopical Results about Co-association of Nucleoside Derivatives in Comparable Concentrations

H. BREUER * und H.-H. PERKAMPUS

Institut für Physikalische Chemie der Universität Düsseldorf

(Z. Naturforsch. 27 b, 6—12 [1972] ; eingegangen am 31. August 1971)

In this paper we described a method how to calculate from NMR the association — caused by H-bonds — of two components. These components may be present in equal concentration and may be partially self-associated. The may act as H-donors and acceptors. The method is demonstrated using some nucleoside derivatives as examples.

In einer vorangehenden Arbeit haben wir über schiedener Umgebung, es müssen also auch zwei NMR-spektroskopische Untersuchungen zur Misch- verschiedene chemische Verschiebungen für das assoziation von vier Nucleosid-Derivaten berichtet1. „reine" Mischassoziat existieren, die jeweils die Die Auswertung erfolgte mit Hilfe einer modifizier- Lage der zwei beobachteten Signale beeinflussen. Sie ten Benesi-Hi ldebrand - Auftragung: gehen dabei mit dem einfachen Gewicht der Misch- i.i i + assoziations-Konzentration ein, wenn ein 1:1- Vb~ Vmb Vxb — Vmb (t>xb~i>mb) '^x Cma Mischassoziat vorliegt. Die folgende Abb. 1 veran- Hierin bedeuten: v^ gemessene chemische Verschie- schaulicht dies. Dabei bedeuten die gestrichelten bung der Komponente b, i>mb extrapolierte chemische Linien die hypothetischen Signale der einzelnen

Verschiebung des Monomeren b, fxb berechnete che- Spezies, und die ausgezogenen Linien entsprechen mische Verschiebung der Mischassoziate, Kx Misch- ~ca = cma+2cna + cab assoziations-Konstante in 1/Mol, cma Monomeren- konzentration der Komponente a.

cma läßt sich leicht berechnen, wenn die zugehö- ~cb=cmb+2cnb + cab rige Eigenassoziations-Konstante bekannt ist. Um dieses Verfahren anwenden zu können, müs- sen einige Voraussetzungen erfüllt sein. Für die Substanz im Überschuß muß gelten: gute Löslichkeit 'cmb und ideales Verhalten sowie genaue Kenntnis der ~cab 'Cab Parameter der Eigenassoziation. Die Substanz im ~2cn Unterschuß muß ein H-Brücken-Protonensignal lie- fern, das möglichst weit von Signalen ihrer Mi- J2cnb schungspartners entfernt liegt. Die Verbindungen v v v v v v v v sollen ferner eine geringe Eigen-, aber eine starke xa na a nb ma xb b mb Mischassoziation zeigen. chemische Verschiebung Abb. 1. Schematisches Spektrum der H-Brücken-Protonen- Herleitung der Auswertemehode signale für die Mischassoziation, v, vn , i>m gemessene Ver- In den beiden Molekülen eines Mischassoziats be- schiebung, Eigenassoziat-Verschiebung, Monomerenverschie- bung, c, cn , cm Einwaage-, Eigenassoziat- und Monomeren- finden sich die sauren Protonen in chemisch ver- Konzentration.

Sonderdruckanforderungen an Prof. Dr. H.-H. PERKAMPUS, * Jetzige Anschrift: Dr. H. BREUER, 34 Göttingen, Theodor- Institut für Physikal. Chemie der Univ. Düsseldorf, D-4000 Heuss-Str. 18. Düsseldorf, Fa. Henkel, Gebäude Z 10.