Fusing Nickelocene and Cyclopentadiene by Two Silyl Bridges. Synthesis and *H, 13C, and 29Si NMR Investigation of a Paramagnetic Building Block for High-Nuclear Metallocenes Monika Fritz, Johann Hiermeier, Frank H. Köhler* Anorganisch-chemisches Institut, Technische Universität München, Lichtenbergstraße 4, D-85747 Garching Z. Naturforsch. 49b, 763-769 (1994); received March 16, 1994 Nickelocenes, Disilylcyclopentadiene. Lithium Cyclopentadienide, Paramagnetic NMR Spectra Two isomers of tetrahydro-4,4,8,8-tetramethyl-4,8-disila-s-indacene (LH2) were monode- protonated and treated with cyclopentadienyl anion and NiBr2(THF)i 5 to give a 72% yield of the mixed nickelocene CpNi(LH) where a cyclopentadiene is fused to a nickelocene. The analysis of the paramagnetic 'H, 13C, and 29Si NMR spectra demonstrated that the syn and anti isomer of CpNi(LH) formed in a ratio of 5/1. Both isomers could be deprotonated to yield the anion CpNi(L~). According to its 13C NMR spectrum the bridging ligand L is not planar.

Introduction tadienyl (Cp) ligands it was realized recently by Stacking of organometallic fragments or mol­ formal condensation of two Cps to conjugated six- ecules is a general strategy to obtain coordination membered rings [4], In previous studies we have polymers. The repeat unit within these polymers addressed stepwise stacking by using a building block concept where a metallocene is linked with or oligomers usually contains one or two jz ligands leading to different types of stacking. For instance, Cp_ through two silyl groups as represented by A with boron-containing ring systems linear stacks or [5]. The reaction of A with metal halides should fragments of linear stacks were obtained [1] which lead to trimetallic model compounds which allow are also known as oligodecker complexes. Linear a convenient study of the interactions between dif­ stacks were also formed by the reaction of organic ferent metallocenes. An obvious extension would be the analogous reaction of B to give coordi­ jt acceptors with metallocenes [2] and, finally, metallocenes could be assembled in a face-to-face nation polymers. For the diamagnetic case M = Fe arrangement by bridging with naphthalene [3], we have studied A and B in depth [5 a], and pre­ liminary results demonstrate that the building block concept also works for M = Ni [5 b]. Here we report the details of the paramagnetic nickelocene building block.

Results and Discussion A. Synthesis The isomeric silyl-bridged cyclopentadienes la and l b (Scheme 1) were converted to the monoanion 2 as described previously [6]. Further Stepwise stacking is an alternative approach reaction with an excess of Cp“ and with solvated which starts with bridged j z ligands. For cyclopen- nickel(II) bromide gave, after work-up, the mixed- ligand nickelocene 3 in 72% yield when di-^-bu- tylether was used as solvent. In THF the yield was * Reprint requests to Prof. Dr. F. H. Köhler. 53%. NMR spectroscopy showed that two diaster- eoisomers 3 a and 3 b were formed in the ratio 51 0932-0776/94/0600-0763 $06.00 © Verlag der Zeitschrift für Naturforschung, 1. This is in contrast to the iron analogue for which D-72072 Tübingen no anti isomer like 3b could be detected [5 a]. The

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. 764 M. Fritz et al. ■ Fusing Nickelocene and Cyclopentadiene mass spectrum and the elemental analysis were in full accord with the anticipated formula. 3a and 3b are highly soluble in hexane and sensitive to oxygen; water splits off the cyclopentadiene part of the molecule. The propensity of nickelocenes to react with alkyl lithium [7] led us to use lithium piperidide for the deprotonation of 3a and 3b. At -78 °C the reaction proceeded slowly as could be seen from the formation of the anion 4 which appeared as a green precipitate. However, when the mixture was heated to room temperature before the reaction was finished, the mixture turned black above Fig. 1. 29Si NMR spectrum of a 5/1 mixture of 3a and 3b dissolved in THF at 310 K. The feature near -9 0 0 -40 °C. Impurities were removed from 4 by ex­ ppm is expanded in the insert. S = standard (hexa- traction with hexane and the product was iden­ methyldisiloxane); G = signal from the glas tube; scale tified by 13C NMR spectroscopy. in ppm.

of ’H NMR signals yielding a ratio of 5/1. For the time being the bigger signals are assigned to 3 a because only the syn isomer was found for the cor­ responding ferrocene [5 a]. The paramagnetic sig­ nal shifts at 298 K are: -904 and -933 ppm for 3a and -889 and -911 ppm for 3 b. In order to establish the structure of 3 a and 3 b by *H and 13C NMR data series of temperature- dependent spectra had to be investigated because the assignment of the paramagnetically shifted sig­ nals proved to be non-trivial. Typical spectra are represented in Figs. 2 and 3 while the numerical / 3a data are given in the experimental part (Tables II and III). For each isomer fifteen 13C NMR signals were ex­ pected. The six nickelocene signals (C4a,5,6, 7,7a,

Scheme 1. a: /i-BuLi; b: Cp , NiBr2(THF)! 5; c: C5H 10NLi.

B. NMR Investigation o /3 a /b and 4 Analytically pure 3 gave a 29Si NMR spectrum (Fig. 1) which consisted of two sets of two signals near -900 ppm, a range which was known from (Me3SiCp)2Ni [8], The number and the pairwise different intensities of these signals indicated that the syn isomer 3 a and the anti isomer 3 b were Fig. 2. 13C NMR spectrum of a 5/1 mixture of 3a and 3b present. As the smaller signals appeared as shoul­ dissolved in THF at 210 K. S = solvent. The nuclei of 3a ders, integration was performed with several pairs are primed; scale in ppm. M. Fritz et al. • Fusing Nickelocene and Cyclopentadiene 765 and Cp) could not be localized owing to excessive /?-SiCH3. Molecular models suggest that the same broadening. Up to now corresponding signals were order applies for 3 b although the numerical values only found for nickelocenes which are more sym­ might change considerably. Thus the signals of a-/d- metric and/or more soluble [5 b, 9], Fig. 2 displays SiCH3 must have much larger shifts to high fre­ nine small signals for 3 b but only eight big ones for quency than those of y-//3-SiCH3. This is in good 3 a because two signals coincide accidentally near agreement with the experiment because only CH3 140 ppm. The coincidence could not be removed in groups are left for the assignment of the two big sig­ the temperature range 205 < T < 330 K, but the in­ nals near 800 ppm. Although further distinction of tegration leaves no doubt about the assignment. a- and (3-SiCH3 as well as y- and /3-SiCH3 is possible In the fully coupled spectrum multiplets due to by following 6 it must be regarded as tentative be­ one-bond CH couplings were observed in the ex­ cause the differences in 6 are now smaller. perimental shift range 250>d>-50. For 3 a the Another aid for the signal assignment is the peak with double intensity, and two more peaks extrapolation of the temperature-dependent signal gave doublets (C 1,2,3,3 a), two gave quartets (two shifts. When 1/T approaches zero the signal shifts SiCH3) and one remained a singlet (C8a). Similarly should be close to those of a diamagnetic analogue 3b gave three doublets (C 1,2,3), two quartets (two unless anomalies like antiferromagnetic exchange, SiCH3), and one singlet (C8a). The signals of the low-spin/high-spin equilibria, dynamic phenomena SiCH3 groups were further distinguished based on and the like are present. Thus when we apply the the dihedral angles 6 between the Si-CH3 bonds extrapolation to the above-mentioned three dou­ and the normals to the Cp planes. The normals are blets of 3a we obtain <3 = 145 ± 5, 134 ± 5, and approximately parallel to the spin-containing car­ 65 ± 10 which compares well with Cl/3, C2, and bon 2 pz orbitals of C 4 a/7 a which determine the hy- C3a of the iron analogue of 3a having ö = 140.7/ perconjugative spin transfer to the CH3 nuclei [10] 137.6, 132.4 and 57.7 [5 a], respectively. C 3 a can be and thus the signal shift <3para (<3para = d>o + Bcos20, identified independently by its one-bond CH coup­ d0 and B being constants). It was assumed that the ling of 130 ± 10 Hz which is smaller than that of the angles are similar to those determined for the iron vinyl-type carbon nuclei C 1-3 (160 ± 10 Hz) as ex­ analogue of 3a by X-ray crystallography [5 a] yield­ pected. No distinction is possible for Cl/3 although ing the following order of dihedral angles of the the coincidence of the signals is lifted owing to SiCH3 groups: a-SiCH3 < (3-SiCH3 << y-SiCH3< different reference shifts obtained from C 1 and

Table I. Paramagnetic NMR data3 of 3a, 3b, and 4.

Position 13C NMR ’H NMR <3('H) for lH NMR of nuc- T ---> o o C refe­ leusb 3a 3b 4 3a 3b 3a 3b rence0

1 0 -1 3 f -15.1 -2.4 -2.1 6.7 6.8 7.06 2 -18 63 16.8 -0.2g -0.4® 6.4® 6.2® 6.88 3 0 l f -15.1 1.1® 4.4® 6.1® 6.3® 6.88 3a -39 167 17.7 5.8 16.3 4.3 3.4 4.18 4 a/7 a e e 1525 5 - 7 e e 1525 -245h -244h j j 4.20 -242h -234h 4.39 -227h 4.20 8a -53 -35 17.7 a 685 545 509 18.2 16.2 -0.7 -0.4 -0.48 ß -17 89 31.9 -2.2' -2.7' o .r 0.2' 0.47 y 57 101 31.9 -o.s 1 -2.4' 0.2' 0.0' 0.45 ö 622 226 509 15.8 7.8 0.2 0.5 0.35 Cp e e 1525 -260 -260 j j 4.03

a In ppm relative to the data of the iron analogue of 3 a (cf Experimental), negative sign for shifts to low fre­ quency; see Scheme 1 for numbering; c extrapolated from the data given in Table III; all values ± 0.5 ppm; d data of the iron analogue of 3a in C6D

C3 of the iron analogue of 3 a. A diamagnetic ref­ and 3 ppm). They may be localized after changing erence compound for 3 b is unknown. However, the temperature as demonstrated in Fig. 3C. when we assume that the reference signal shifts do The SiCHj signals of 3 a are easily recognized not change very much on going from the syn to the by integration. From the l3C NMR results it is anti isomer C 1, C2, C3, and C3a can be identified clear that the signals of Ha and H6 must be more similarly. In particular the neighbouring signals shifted than those of H/3 and Hy. Further distinc­ near 300 ppm (Fig. 2) may be distinguished in this tion of these pairs of signals tentatively follows the way. Thus the d-l/T curve ( cf. Table II) of the less sequence of the ,3C NMR signals. The signal shifts shifted signal intersects the d axis near 60 ppm obtained after extrapolating the temperature-de- (C3a) whereas the curve of the more shifted sig­ pendent data for T —> oo may also be used for the nal has a smaller intercept (Cy). The overall as­ assignment. They are listed in Table I along with signment is summarized in Table I. the reference shifts of the iron analogue of 3 a. It In the ‘H NMR spectrum the strongly shifted is gratifying that the extrapolated shifts and the signals (Fig. 3A) of H5,6,7 are poorly resolved for reference data are well in accord. Nevertheless 3 a and 3 b, and those of the Cps coincide. The as­ ambiguities remain for H2 and H3 as well as Hß signment is based on the relative signal areas and and Hy. on the fact that for nickelocenes the signals of a Among the NMR data of 3 a and 3 b three points silylated Cp are less shifted than those of an un­ merit further comment. (1) Above 330 K some substituted Cp [5 c]. No distinction is possible for proton resonances start to coalesce. The reason is H5,6,7. Among the remaining signals shown in that in both molecules the cyclopentadiene moiety Fig. 3 B two are hidden by other signals (near 24 is engaged in 1,2-silatropic bond shifts. This phenomenon was not studied in detail because we have analysed it for the iron analogue of 3 a pre­ (■) viously [5a]. (2) For 3a the signal of H 3 a is much Cp’ broader than any other similarly shifted signal of the cyclopentadiene moiety (Fig. 3B). This is due to excessive dipolar relaxation which in turn must follow from a short distance between the para­ magnetic metal center and the nucleus under study [10]. From the X-ray structure of the iron analogue u of 3a [5a] we know that H3a selectively ap­ 21 19 17 proaches the metal. Hence the signal width of H3a is a proof for the syn isomer. (3) If hypercon­ jugation determines the spin transfer from C4a

2',3 and C7a to all other carbon atoms bound to Si4 and Si 8, respectively, then the sum of the d values for Ca, Cß, and C3a, and likewise for Cy, 05, and C8a should be similar. Table I shows that this is true for 3 a (629 and 626 ppm, respectively) whereas a strong deviation is found for 3 b (801 and 292 ppm, respectively). We conlcude that some additional distortion is present in 3 b, prob­ ably a different bending of the bonds Si4-C4a and Si8-C7a out of the neighbouring Cp plane. The product obtained after deprotonation of 3 a Fig. 3. 'H NMR spectra of a 5/1 mixture of 3a and 3b. and 3b (Scheme 1) was studied by I3C (Figure 4), (A) Range of the nickelocene ring proton signals; sol­ 29Si, and 7Li NMR spectroscopy. In the 13C NMR vent C6D6, temperature 305 K. (B) Range of all other spectra the one-bond CH coupling patterns and proton signals at 243 K. (C) Signals above 15 ppm at 283 K. Solvent for (B) and (C); acetone-d6. The nuclei the signal areas were in accord with the assign­ of 3 a are primed; all scales in ppm. ment of the signals between 20 and 150 ppm to a M. Fritz et al. • Fusing Nickelocene and Cyclopentadiene 767

1/3 relative to an external saturated solution of lith­ ium iodide. The evaluation of the paramagnetic

2000 1000 Instead there must be a folding at the vector Si 4 — Si 8 which leads to different dihedral angles 6 be­ Fig. 4. 13C NMR spectrum of 4 dissolved in THF at tween the Si-CH3 bonds and the nickelocene j t 298 K. S = solvent, X = impurity of Cp-. All scales in ppm. system as discussed for 3 a and 3 b. The resulting shift difference does not allow to distinguish be­ tween two folding versions of 4: Cp“ may be bent away from the nickelocene or both moieties may 1,2-disubstituted Cp“, which senses little spin, and approach. The latter version is suggested by X-ray to a methyl group. This meets the requirements of results obtained for the iron analogue of 4 [5 a]. 4. A second methyl signal was located near 500 Hence a- and (3-CH3 will be axial and have the ppm after selective phase adjustment. Finally an larger signal shift. unresolved and extremely broad feature was found near 1500 ppm which is typical for nickel­ Experimental ocene ring carbon atoms [9]. The details given in Table I show that the resonances of 4 appear in The syntheses and spectroscopic investigations were carried out under purified dinitrogen or ar­ shift ranges that are known from the precursors gon in dry and oxygen-free solvents. Equipment 3 a and 3 b. This is also true for the 29Si NMR sig­ purchased from Kronwald was used for medium nal which has a paramagnetic shift at 298 K of pressure liquid chromatography (MPLC) as de­ -824 ppm. scribed previously [5 a], The elemental analyses In the 7Li NMR spectrum a signal with a half were obtained from the microanalytical laboratory width of 60 Hz at 298 K was found at -4.1 ppm of this institute.

Table II. Temperature-dependent 13C NMR signal shifts3 of a 5/1 mixture of 3a and 3b dissolved in toluene-d8.

Temp. Ca Cß C2 C l/C 3 C3a C8a (K) CÖ c y 3a 3b 3a 3b 3a 3b 3a 3b 3a 3b 3a 3b

209.9 965 769 -26.4 128.3 103.6 218.2 135.8 122.1 -1.2 294 59.4 80.0 880 304 77.5 145.7 135.8 139.1 228.0 890 702 -23.3 117.9 106.5 212.2 136.9 123.6 4.2 275 69.4 87.7 808 283 72.7 134.0 136.9 139.8 242.5 840 663 -21.5 110.3 108.4 207.7 137.3 124.0 7.7 263 72.9 92.7 763 268 67.5 125.8 137.3 140.0 256.3 791 627 -20.0 104.0 110.0 204.0 138.2 125.5 11.0 253 78.2 97.6 722 255 65.6 118.9 138.2 141.0 298.0 679 539 -15.9 90.0 113.7 194.5 139.0 126.3 18.7 225 89.6 108.1 621 225 58.4 102.0 139.0 140.5 310.0 655 519 -15.1 85.7 114.6 191.8 139.2 127.9 20.3 220 91.8 110.0 597 220 56.6 98.4 139.2 140.6 314.6 638 509 -14.5 b 115.2 191.5 139.4 128.5 21.8 218 93.3 111.0 585 217 55.8 97.0 139.4 140.5

In ppm relative to TMS for carbon atoms other than C 4 a, 5, 6, 7, 7 a and Cp; b overlapping signals. 768 M. Fritz et al. • Fusing Nickelocene and Cyclopentadiene

The NMR spectra were recorded with a Bruker ml of THF while keeping the mixture at -78 °C. CXP 200 and a Jeol JNM GX 270 spectrometer by After slow heating and adding 60 ml of a 0.56 M using tubes with ground glass fittings and stoppers. solution of NaCp in THF the mixture became All signals were measured relative to solvent clear. It was diluted with 50 ml of h -Bu 20 and peaks, and the shifts were calculated relative to heated in an oil bath to 75 °C. When 7.0 g (21.4 TMS. Exceptions were 29Si and 7Li NMR for mmol) of NiBr2(THF)t 5 was added under stirring which (Me3Si)20 (d(29Si) = 6.9) and a saturated the colour changed immediately to dark green solution of Lil in water were used, respectively. and, when an excess of the nickel salt was present, Subsequent calculation relative to isostructural to blue-violet. The mixture was stirred over night, iron compounds [5 a] gave the paramagnetic signal the solvents were stripped, and the remainder was shifts. The temperature-dependent data are col­ extracted with 100 ml of pentane and washed sev­ lected in Tables II and III. In order to obtain eral times with 10 ml-portions of pentane. The standard shifts at 298 K data obtained near 298 K combined extracts were freed from the solvent, were transformed according to the Curie law or and most of Cp2Ni was removed from the solid by temperature-dependent measurements were inter­ sublimation (IPa, 40 °C bath temperature). The polated. The mass spectrum was recorded with a remainder was dissolved in a minimum of pentane Varian MAT 311 A instrument. and subjected to MPLC (column length/diameter 50/2.6 cm; silica Merck 60, 15-40 /im; eluent pen­ tane). The first small band contained Cp2Ni. It was Cyclopentadienyl(syn/anti-1,2,3,3 a,8a-rj5-3 a,4 ,7 a,8- followed by a green main band which, after re­ tetrahydro-4,4,8,8-tetramethyl-4,8-disila-s-indacene- moval of pentane, afforded 1.64 g (72% relative to 7a-yl)nickel (3 a and 3 b) la/b) of green microcrystals of 3a and 3b. In tolu­ 3.6 ml of a 1.72 M solution of n-BuLi in hexane ene solution the ratio of 3a/3b was 5/1 for was transferred to 1.5 g (6.2 mmol) of la/b in 50 213 K < T < 303 K (NMR). Sublimation (0.1-0.01

Table III. Temperature-dependent 'H NMR signal shifts3 of a 5/1 mixture of 3a and 3b dissolved in toluene-d8

Temp. H a Hß H2 HI H 3a (K) HÖ Hy H3 3a 3b 3a 3b 3a 3b 3a 3b 3a 3b

203 26.2 11.8 -2.5 -4.3 8.8 13.6 3.8 4.2 13.1 28.7 23.6 23.2 0.1 -2 .8 b b 213 25.0 11.2 -2 .4 -4.2 8.7 13.3 3.9 4.3 12.6 27.6 22.5 22.1 -0.1 -2.7 b b 223 23.8 10.7 -2 .4 -4.1 8.6 12.9 4.0 4.4 12.2 26.4 21.5 21.1 -0.1 b 6.8 b 233 22.8 10.3 -2 .3 -4 .0 8.5 12.7 4.1 4.5 11.8 25.5 20.6 20.2 -0.1 b 6.8 b 243 21.8 9.9 -2 .2 -3 .8 8.4 12.4 4.2 4.6 11.4 24.6 19.7 19.3 -0.1 b 6.8 b 253 20.9 9.6 -2.1 -3 .7 8.4 12.2 4.3 4.7 11.1 23.7 19.0 18.6 -0.1 b 6.7 b 263 20.1 9.2 -2 .0 -3.6 8.2 11.9 4.4 4.8 10.8 22.9 18.3 17.9 -0.1 b 6.7 b 273 19.3 8.9 -2 .0 -3 .4 8.2 11.7 4.5 4.9 10.6 22.2 17.6 17.3 -0.1 b 6.7 6.8 283 18.6 8.6 -1 .9 -3.3 8.1 11.6 4.6 4.9 10.4 21.6 17.0 16.6 -0.1 b 6.7 6.8 293 17.8 8.3 -1 .7 -3.2 8.1 11.3 4.6 5.0 10.1 20.8 16.3 15.9 -0.1 b 6.7 b 303 16.7 8.0 -1 .7 -3.1 8.0 11.2 4.7 5.0 9.9 20.3 15.8 15.5 -0.1 b 313 16.7 7.9 -1 .7 -2 .9 8.0 11.0 4.8 5.0 9.8 19.6 15.3 14.9 -0.1 b 323 16.2 7.8 -1 .6 -2 .8 7.9 10.9 4.8 5.1 9.6 19.3 14.9 14.3 -0.1 b

In ppm relative to TMS for protons other than H5,6,7 and Cp; b signal covered by other signals. M. Fritz et al. • Fusing Nickelocene and Cyclopentadiene 769

Pa/60 °C bath temperature) gave small cubes (m.p. hexane and added via canula to a cooled suspen­ 65 °C). Similarly 6.6 g (20.2 mmol) of sion of 3.7 mmol of lithium piperidide in 50 ml of NiBr2(THF)! 5, 6.2 mmol of 2, and 30.0 mmol of hexane freshly prepared from 0.25 ml (3.7 mmol) NaCp in 100 ml of THF gave 1.1 g (53% relative of piperidine and 5.3 ml of a 1.7 M solution of n- to 2) of 3a and 3b. BuLi in hexane. The mixture was stirred at -78 °C MS (EI, 70 eV): m/z (% ) 366 (100), M+; 351 (4), for 16 h and allowed to come to room temperature M+-CH 3; 300 (2), M+-C 5H6; 285 (3), M+-C H 3- within another 16 h. The green cloudy precipitate C5H6; 183 (3), M2+. Isotope pattern of M+, m/z which had formed was freed from the excess of (% calcd/found): 366 (100/100), 367 (31.5/30.3), neutral nickelocenes by extracting them with hex­ 368 (49.8/49.8), 369 (16.0/16.2), 370 (10.5/11.0), 371 ane until the extract was colorless. When the re­ (2.6/3.0), 372 (2.3/2.3). maining solid was dried in vacuo and dissolved in C19H24NiSi2 (367.3) THF the 13C NMR spectrum showed the signals Calcd C 62.14 H 6.59 Ni 15.98 Si 15.29%, of the anion 4. Found C 61.67 H 6.54 Ni 15.74 Si 16.6%.

Deprotonation of 3 a and 3 b This work was supported by the Fonds der The mixture of 3 a and 3 b obtained in the pre­ Chemischen Industrie, the Leonhard-Lorenz-Stif- vious section was dissolved at -78 °C in 300 ml of tung, and the Wacker Chemie GmbH.

[1] a) G. E. Herberich, U. Englert, D. Pubanz, J. Or­ b) P. Bergerat, J. Blümel, M. Fritz, J. Hiermeier, P. ganomet. Chem. 459, 1 (1993) and literature cited Hudeczek, O. Kahn, F. H. Köhler, Angew. Chem. therein; 104, 1285 (1992); Angew. Chem., Int. Ed. Engl. 31, b) R. N. Grimes, Chem. Rev. 92, 251 (1992); 1258 (1992); c) W. Siebert, Russ. Chem. Rev., Engl. Transl. 60, c) H. Atzkern, P. Bergerat, M. Fritz, J.Hiermeier, P. 784 (1991). Hudeczek, O. Kahn, B. Kanellakopulos, F. H. [2] a) J. S. Miller, A. J. Epstein, W. M. Reiff, Chem. Rev. Köhler, M. Ruhs, Chem. Ber. 127, 277 (1994). 88 , 201 (1988); [6] J. Hiermeier, F. H. Köhler, G. Müller, Organomet­ b) W. E. Broderick, J. A. Thompson, E. P. Day, allics 10, 1787 (1991). B. M. Hoffman, Science 249, 401 (1990); [7] T. I. Voyevodskaya, I. M. Pribytkova, Yu. A. Ustyn- c) G. T. Yee, J. M. Manriquez, D. A. Dixon, R. S. yuk, J. Organomet. Chem. 37, 187 (1972); S. Pasynk- McLean, D. M. Groski, R. B. Flippen, K. S. Na- iewicz, H. Lehmkuhl, J. Organomet. Chem. 289, 189 rayan, A. J. Epstein, J. S. Miller, Adv. Mater. 3, 309 (1985); H. Lehmkuhl, C. Krüger, Organometallics 7, (1991); 2038 (1988); H. Lehmkuhl, F. Danowski, G. Mehler, d) W. E. Broderick, B. M. Hoffman, J. Am. Chem. J. Poplawska, S. Pasynkiewicz, J. Organomet. Chem. Soc. 113, 6334 (1991); 363, 367 (1989); S. Pasynkiewicz, J. Organomet. e) D. M. Eichhorn, D. C. Skee, W. E. Broderick, Chem. 387, 1 (1990). B. M. Hoffman, Inorg. Chem. 32, 491 (1993). [8] F. H. Kohler, W. A. Geike, J.Magn. Reson. 53, 297 [3] a) B. M. Foxman, M. Rosenblum, V. Sokolov, N. (1983). Khrushchova, Organometallics 12, 4805 (1993); [9] a) F. H. Köhler, K.-H. Doll, W. Prößdorf, Angew. b) H. M. Nugent, M. Rosenblum, P. Klemarezyk, J. Chem. 92, 487 (1980); Angew. Chem., Int. Ed. Engl. Am. Chem. Soc. 115, 3848 (1993). 19, 479 (1980); [4] T. J. Katz, A. Sudhakar, M. D. Teasley, A. M. Gil­ b F. H. Köhler, W. Geike, J.Organomet. Chem. 328, bert, W. E. Geiger, M. P. Robben, M. Wuensch, 35 (1987). M. D. Ward, J. Am. Chem. Soc. 115, 3182 (1993); [10] J. Kowalewski, L. Nordenskiöld, N. Benetis, P. O. A. M. Gilbert, T. J. Katz, W. E. Geiger, M. P. Westlund, Progr. Nucl. Magan. Reson. Spectrosc. Robben, A. L. Rheingold, J. Am. Chem. Soc. 115, 17, 141 (1985) L. Banci, I. Bertini, C. Luchinat, Nu­ 3199 (1993) and references cited therein. clear and Electron Relaxation, VCH, Weinheim [5] a) M. Fritz, J. Hiermeier, N. Hertkorn, F. H. Köhler, (1991). G. Müller, G. Reber, O. Steigelmann, Chem. Ber. 124, 1531 (1991);