Zirconium-91 chemical shifts and line widths as indicators of coordination geometry distortions in zirconocene complexes

Article abstract of DOI:10.1021/om950757c

⁹¹Zr NMR chemical shifts and line widths (Δυ_1/2) are reported for a number of ring-bridged and ring-substituted zirconocene dichloride, dibromide, and dimethyl complexes. Ab initio computations at the SCF level employing basis sets of moderate size suggest that the magnitude of the electric field gradient (EFG) at the Zr atom dominates Δυ_1/2 when the substituents X at Zr are varied (X = Br, Cl, Me). Substituents at the cyclopentadiene (Cp) rings affect the computed EFGs much less; in these cases, the line widths Δυ_1/2 are governed by the molecular correlation times τ_c, which were obtained for several zirconocene dichlorides from T₁(¹³C) measurements. Experimental trends in δ(⁹¹Zr) of zirconocenes are well reproduced computationally with the IGLO (individual gauge for localized orbitals) or GIAO (gauge including atomic orbitals) SCF methods employing large basis sets. Model calculations suggest that δ(⁹¹Zr), as well as the EFG, are quite sensitive to the inclination and twist angles of the Cp rings and, to a lesser extent, to the CpZrCp′ angle. A substantial deshielding, δ(⁹¹Zr) ca. 700 ppm, is predicted for (C₅H₅)₂ZrMe⁺, presumably the active olefin-polymerizing catalyst.

Full text of DOI:10.1021/om950757c

778
Organometallics 1996, 15, 778-785
Zir con iu m -91 Ch em ica l Sh ifts a n d Lin e Wid th s a s
In d ica tor s of Coor d in a tion Geom etr y Distor tion s in
Zir con ocen e Com p lexes†
Michael Bu¨hl,* Gudrun Hopp, and Wolfgang von Philipsborn*
Organisch-chemisches Institut, Universita¨t Zu¨rich, CH-8057 Zu¨rich, Switzerland
Stefan Beck, Marc-Heinrich Prosenc, Ursula Rief, and
Hans-Herbert Brintzinger*
Fakulta¨t fu¨r Chemie, Universita¨t Konstanz, D-78464 Konstanz, Germany
Received September 25, 1995X
91Zr NMR chemical shifts and line widths (∆ν1/2) are reported for a number of ring-bridged
and ring-substituted zirconocene dichloride, dibromide, and dimethyl complexes. Ab initio
computations at the SCF level employing basis sets of moderate size suggest that the
magnitude of the electric field gradient (EFG) at the Zr atom dominates ∆ν1/2 when the
substituents X at Zr are varied (X ) Br, Cl, Me). Substituents at the cyclopentadiene (Cp)
rings affect the computed EFGs much less; in these cases, the line widths ∆ν1/2 are governed
by the molecular correlation times τc, which were obtained for several zirconocene dichlorides
from T1(13C) measurements. Experimental trends in δ(91Zr) of zirconocenes are well
reproduced computationally with the IGLO (individual gauge for localized orbitals) or GIAO
(gauge including atomic orbitals) SCF methods employing large basis sets. Model calculations
suggest that δ(91Zr), as well as the EFG, are quite sensitive to the inclination and twist
angles of the Cp rings and, to a lesser extent, to the CpZrCp′ angle. A substantial deshielding,
δ(91Zr) ca. 700 ppm, is predicted for (C5H5)2ZrMe+, presumably the active olefin-polymerizing
catalyst.
In tr od u ction
netic shielding terms, to the optical excitation ener-
gies.12,15 Siedle and co-workers have utilized 91Zr NMR
spectroscopy to identify cationic zirconocene species.10
In a recent study, Bo¨hme, Thiele, and Rufinska have
reported 91Zr NMR data for several ring-substituted and
ring-bridged zirconocene dichloride compounds and have
correlated the 91Zr shifts with C5-ring centroid-Zr-
centroid angles in these complexes.11
Relations between the structures of ring-bridged and
ring-substituted zirconocene complexes and their prop-
erties as catalysts for olefin polymerization2-5 and other
C-C bond formation reactions6-9 are of current interest.
91Zr NMR spectroscopy has been used by several
research groups10-14 as a probe for the environment of
the Zr centers in complexes of this type. Following
initial studies by McGlinchey and co-workers,13,14 Benn
and Rufinska reported 91Zr NMR spectra for a number
of (C5H5)2Zr(diene) complexes and proposed that the
91Zr chemical shifts are related, through the paramag-
The line widths ∆ν1/2 of the 91Zr resonances are
conspicuously variable in different zirconocene com-
plexes, ranging from very sharp peaks with ∆ν1/2 ) 18
Hz [(C5H5)2ZrBr2] to resonances several kHz wide in
other complex species10-15 or even signals undetectably
broad, i.e. ∆ν1/2 > ca. 10 kHz. These large and variable
line widths represent a serious obstacle for quantitative
91Zr NMR studies and sometimes even for the detection
of individual zirconocene complex species.10 On the
other hand, Hafner and co-workers have reported that
49Ti NMR line widths for (C5H5)Ti(OR)2(allyl) complexes
with different carbohydrate ligands are a very useful
indicator for their structural flexibility and, as a con-
sequence, for their effectiveness in inducing stereose-
lective aldol reactions.16 We have thought it worthwhile
to investigate how 91Zr NMR parameters of typical
zirconocene complexes depend on their coordination
geometries and, therefore, determined the 91Zr chemical
shifts and line widths of a series of zirconocene com-
plexes. Ab initio calculations have been performed of
the electric field gradients (EFGs) of (C5H5)2ZrX2 species
(X ) Br, Cl, Me). Further computations have been
carried out for substituted and for artificially distorted
† Transition Metal NMR Spectroscopy. 29. Part 28: see ref 1.
‡ E-mail: buehl@oci.unizh.ch.
§ E-mail: beck@chclu.chemie.uni-konstanz.de.
X Abstract published in Advance ACS Abstracts, J anuary 1, 1996.
(1) Asaro, F.; Costa, G.; Dreos, R.; Pellizer, G.; von Philipsborn, W.
J . Organomet. Chem., in press.
(2) Mo¨hring, P. C.; Coville, N. J . J . Organomet. Chem. 1994, 479, 1.
(3) Sinclair, K. B.; Wilson, R. B. Chem. Ind. 1994, 857.
(4) Kaminsky, W. Catal. Today 1994, 20, 257.
(5) Brintzinger, H. H.; Fischer, D.; Mu¨lhaupt, R.; Rieger, B.;
Waymouth, R. M. Angew. Chem., Int. Ed. Engl. 1995, 34, 1143.
(6) Vostrikova, O. S.; Ibragimov, A. G.; Sultanov, R. M.; Dzhelimev,
U. M. Organomet. Chem. USSR 1992, 5, 377.
(7) Morken, J . P.; Didiuk, M. T.; Hoveyda, A. H. J . Am. Chem. Soc.
1993, 115, 6997.
(8) Negishi, E. I.; Takahashi, T. Acc. Chem. Res. 1994, 27, 124.
(9) Knight, K. S.; Waymouth, R. M. Organometallics 1994, 13, 2575.
(10) Siedle, A. R.; Newmark, R. A.; Gleason, W. B.; Lamanna, W.
M. Organometallics 1990, 9, 1290.
(11) Bo¨hme, U.; Thiele, K. H.; Rufinska, A. Z. Anorg. Allg. Chem.
1994, 620, 1455.
(12) Benn, R.; Rufinska, A. Angew. Chem. 1984, 98, 851.
(13) Sayer, B. G.; Hao, N.; De´ne`s, G.; Bickley, D. G.; McGlinchey,
M. J . Inorg. Chim. Acta 1981, 48, 53.
(14) Sayer, B. G.; Thompson, J . I. A.; Hao, N.; Birchall, T.; Eaton,
D. R.; McGlinchey, M. J . Inorg. Chem. 1981, 20, 3748.
(15) Benn, R.; Rufinska, A. J . Organomet. Chem. 1984, 273, C51.
(16) Hafner, A.; Duthaler, R. O.; Marti, R.; Rihs, G.; Rothe-Streit,
P.; Schwarzenbach, F. J . Am. Chem. Soc. 1992, 114, 2321.
0276-7333/96/2315-0778$12.00/0 © 1996 American Chemical Society

91Zr Chemical Shifts and Line Widths
Organometallics, Vol. 15, No. 2, 1996 779
(C5H10)2C2(C5H4)2Zr Cl2 (8a ). Yield: 41%. 1H NMR (600.1
MHz, CDCl3): δ 6.01 (m, 4H, R-Cp), 6.61 (m, 4H, â-Cp), 1.05-
2.13 (m, 20H, CH2). 13C{1H} NMR (150.9 MHz, CDCl3): δ
109.81 (4C, R-Cp), 123.55 (4C, â-Cp), 141.89 (2C, ipso-Cp),
23.18 (4C, CH2), 25.72 (2C, CH2), 35.85 (4C, CH2), 50.86 (2C,
bridge-C). Anal. Calcd for C22H28Cl2Zr: C, 58.13; H, 6.21.
Found: C, 57.51 ; H, 5.64.
Further complexes, Me2Si(C5H3-2-Me-4-tBu-)2ZrCl2 (9a ),44
Me2Si(1-indenyl)2ZrCl2 (10a),45 C2H4(1-indenyl)2ZrCl2 (11a),46,47
and Me2Si(2-Me-benz[e]indenyl)2ZrCl2 (12a ),48 were likewise
prepared as previously described.
zirconocenes in order to assess the geometric and elec-
tronic effects of substituents on the 91Zr NMR line
widths.
In addition, 91Zr chemical shifts have been computed
2-
for some zirconocenes and for ZrX6
(X ) F, Cl)
employing the IGLO17-20 (individual gauge for localized
orbitals) and GIAO21-23 (gauge including atomic orbit-
als) methods. While computations of NMR chemical
shifts are now routinely possible for the lighter, i.e. first-
and second-row nuclei (both at SCF and at electron-
correlated levels),19,24-28 much less is known regarding
the reliabilty of such calculations for compounds of
heavier elements. It has recently been shown that
electron correlation effects can be quite important in
chemical shift computations for heavier nuclei, e.g., for
Ga and Se compounds.29-31 On the other hand, experi-
mental trends for certain classes of compounds can often
be reproduced quite satisfactorily at SCF level, even
when fourth-row elements such as Mo, Cd, or Sn are
involved.32,33 As the SCF-based methods prove to
perform rather well for 91Zr NMR chemical shift calcu-
lations, interesting predictions of δ(91Zr) values can be
made for some cationic zirconocene complexes involved
in the catalytic olefin polymerization process.
The corresponding dibromide compounds 1b,49 2b,50 3b,51
4b, 5b,39 6b, 7b52 , and 8b were obtained by reaction of 1a -8a
with BBr3, as described by Druce and Lappert,49 in almost
quantitative yields.
(C5H4tBu )Zr Br 2 (4b). Yield: 85%. 1H NMR (600.1 MHz,
CDCl3): δ 6.33 (m, 4H, R-Cp), 6.53 (m, 4H, â-Cp), 1.33 (s, 18H,
tBu). 13C{1H} NMR (150.9 MHz, CDCl3): δ 112.62 (4C, R-Cp),
116.40 (4C, â-Cp), 143.65 (2C, ipso-Cp), 31.50 (6C, tBu), 33.65
t
(2C, Bu). Anal. Calcd for C18H26Br2Zr: C, 43.82; H, 5.31.
Found: C, 43.29 ; H, 5.20.
Me4C2(C5H4)2Zr Br 2 (6b). Yield: 91%. 1H NMR (600.1
MHz, CDCl3): δ 6.13 (m, 4H, R-Cp), 6.84 (m, 4H, â-Cp), 1.42
(s, 12H, Me). 13C {1H} NMR (150.9 MHz, CDCl3): δ 108.88
(4C, R-Cp), 123.24 (4C, â-Cp), 142.97 (2C, ipso-Cp), 28.34 (4C,
Me), 45.27 (2C, bridge-C). Anal. Calcd for C16H20Br2Zr: C,
41.47; H, 4.35. Found: C, 41.58 ; H, 4.45.
(C5H10)2C2(C5H4)2Zr Br 2 (8b). Yield: 90%. 1H NMR (600.1
MHz, CDCl3): δ 6.10 (m, 4H, R-Cp), 6.89 (m, 4H, â-Cp), 1.14-
2.17 (m, 20H, CH2). 13C{1H} NMR (150.9 MHz, CDCl3): δ
110.26 (4C, R-Cp), 124.01 (4C, â-Cp), 141.85 (2C, ipso-Cp),
23.54 (4C, CH2), 26.08 (2C, CH2), 35.19 (4C, CH2), 51.42 (2C,
bridge-C). Anal. Calcd for C22H28Br2Zr: C, 48.62; H, 5.19.
Found: C, 47.32; H, 4.59.
The dimethyl complexes 1c,53 2c,54 3c,55 4c,55 5c,56 6c,57 7c,56
and 8c were prepared by a method described by Samuel and
Rausch:53 To the zirconocene dichlorides, suspended in diethyl
ether at -50° C, a slight excess (2-3%) of 1.6 M methyllithium
in diethyl ether solution was added. After being heated slowly
Exp er im en ta l a n d Com p u ta tion a l Section
1. Zir con ocen e Syn th eses. The following zirconocene
chloride complexes were prepared by previously described
methods: (C5H5)2ZrCl2 (1a ),34 (C5H5)2Zr(Cl)Me (1d ),35 (C5Me5)2-
t
ZrCl2 (2a ),36 (C5H4Me)2ZrCl2 (3a ),37 (C5H4 Bu)2ZrCl2 (4a ),38
(C5H4SiMe3)2ZrCl2 (5a ),39 Me4C2(C5H4)2ZrCl2 (6a ),40 Me2Si-
(C5H4)2ZrCl2 (7a ).41,42 The dicyclohexyl-bridged complex
(C5H10)2C2(C5H4)2ZrCl2 (8a ) was prepared in analogy to the
corresponding titanium complex:43 Reaction of the dimagne-
sium salt (C5H10)2C2(C5H4MgCl)2‚4THF with ZrCl4 in toluene
gave 8a in a yield of ca. 40%.
(17) Kutzelnigg, W. Isr. J . Chem. 1980, 19, 193.
(18) Schindler, M.; Kutzelnigg, W. J . Chem. Phys. 1982, 76, 1919.
(19) Kutzelnigg, W.; Fleischer, U.; Schindler, M., In NMR Basic
Principles and Progress; Springer-Verlag: Berlin, 1990; Vol. 23, p 165.
(20) Meier, U.; Wu¨llen, C. v.; Schindler, M. J . Comput. Chem. 1992,
13, 551.
(38) Howie, R. A.; McQuillan, G. P.; Thompson, D. W.; Lock, G. A.
J . Organomet. Chem. 1986, 303, 213.
(39) Antin˜olo, A.; Lappert, M. F.; Singh, A.; Winterborn, D. J . W.;
Engelhardt, L. M.; Raston, C. L.; White, A. H.; Carty, A. J .; Taylor, N.
J . J . Chem. Soc., Dalton Trans. 1987, 1463.
(40) Schwemmlein, H.; Brintzinger, H. H. J . Organomet. Chem.
1983, 254, 69.
(21) Ditchfield, R. Mol. Phys. 1974, 27, 789.
(22) Wolinski, K.; Hinton, J . F.; Pulay, P. J . Am. Chem. Soc. 1990,
112, 8251.
(23) Ha¨ser, M.; Ahlrichs, R.; Baron, H. P.; Weis, P.; Horn, H. Theor.
Chim. Acta 1992, 83, 455.
(24) Gauss, J . J . Chem. Phys. 1993, 99, 3629.
(25) Gauss, J .; Stanton, J . F. J . Chem. Phys. 1995, 102, 251.
(26) Chesnut, D. B.; Rusiloski, B. E.; Moore, K. D.; Egolf, D. A. J .
Comput. Chem. 1993, 14, 1364.
(27) Bu¨hl, M.; Schleyer, P. v. R. J . Am. Chem. Soc. 1992, 114, 477.
(28) For a recent monograph on methods and applications see e.g.:
Tossel, J . A., Ed. Nuclear Magnetic Shieldings and Molecular Struc-
ture; Kluwer Academics: Amsterdam, 1993.
(41) Klouras, N.; Ko¨pf, H. Monatsh. Chem. 1981, 112, 887.
(42) Bajgur, C. S.; Tikkanen, W. R.; Petersen, J . L. Inorg. Chem.
1985, 24, 2539.
(43) Burger, P.; Diebold, J .; Gutmann, S.; Hund, H. U.; Brintzinger,
H. H. Organometallics 1992, 11, 1319.
(44) Wiesenfeldt, H.; Reinmuth, A.; Barsties, E.; Evertz, K.; Brintz-
inger, H. H. J . Organomet. Chem. 1989, 369, 359.
(45) Herrmann, W. A.; Rohrmann, J .; Herdtweck, E.; Spaleck, W.;
Winter, A. Angew. Chem., Int. Ed. Engl. 1989, 28, 1511.
(46) Wild, F. R. W. P.; Zsolnai, L.; Huttner, G.; Brintzinger, H. H.
J . Organomet. Chem. 1982, 232, 233.
(47) Wild, F. R. W. P.; Wasiucionek, M.; Huttner, G.; Brintzinger,
H. H. J . Organomet. Chem. 1985, 288, 63.
(48) Stehling, U.; Diebold, J .; Kirsten, R.; Ro¨ll, W.; Brintzinger, H.
H.; J u¨ngling, S.; Mu¨lhaupt, R.; Langhauser, F. Organometallics 1994,
13, 964.
(29) Loos, D.; Schno¨ckel, H.; Gauss, J .; Schneider, U. Angew. Chem.,
Int. Ed. Engl. 1992, 31, 1362.
(30) Bu¨hl, M.; Thiel, W.; Fleischer, U.; Kutzelnigg, W. J . Phys. Chem.
1995, 99, 4000.
(31) Bu¨hl, M.; Gauss, J .; Stanton, J . F. Chem. Phys. Lett. 1995, 241,
248.
(49) Druce, P. M.; Kingston, B. M.; Lappert, M. F.; Spalding, T. R.;
Srivastava, R. C. J . Chem. Soc. A 1969, 2106.
(50) Wochner, F.; Brintzinger, H. H. J . Organomet. Chem. 1986, 309,
65.
(51) Wielstra, Y.; Gambarotta, S.; Meetsma, A.; Spek, A. L. Orga-
nometallics 1989, 8, 2948.
(52) Ko¨pf, H.; Klouras, N. Z. Naturforsch. 1983, 38b, 321.
(53) Samuel, E.; Rausch, M. D. J . Am. Chem. Soc. 1973, 95, 6263.
(54) Manriquez, J . M.; McAlister, D. R.; Rosenberg, E.; Shiller, A.
M.; Williamson, K. L.; Chan, S. I.; Bercaw, J . E. J . Am. Chem. Soc.
1978, 100, 3078.
(32) Nakatsuji, H. In Nuclear Magnetic Shieldings and Molecular
Structure; Tossell, J . A., Ed.; Kluwer Academics: Amsterdam, 1993;
p 263.
(33) For recent applications of 129Xe chemical shift calculations
see: Freitag, A.; Wu¨llen, C. v.; Staemmler, V. Chem. Phys. 1995, 192,
267.
(34) Heyn, B., Hipler, B., Kreisel, G., Schreer, H., Walther, D., Eds.
Anorganische Synthesechemie:
Ein
integriertes Praktikum;
Springer-Verlag: Berlin, 1986.
(35) Wailes, P. C.; Weigold, H.; Bell, A. P. J . Organomet. Chem.
1971, 33, 181.
(36) Bercaw, J . E.; Marvich, R. H.; Bell, L. G.; Brintzinger, H. H. J .
Am. Chem. Soc. 1972, 94, 1219.
(55) Couturier, S.; Tainturier, G.; Gautheron, B. J . Organomet.
Chem. 1980, 195, 291.
(56) Siedle, A. R.; Newmark, R. A.; Lamanna, W. M.; Schroepfer, J .
N. Polyhedron 1990, 9, 301.
(37) J ordan, R. F.; LaPointe, R. E.; Bradley, P. K.; Baenziger, N.
Organometallics 1989, 8, 2892.
(57) Niemann, U.; Diebold, J .; Troll, C.; Rief, U.; Brintzinger, H. H.
J . Organomet. Chem. 1993, 456, 195.

780 Organometallics, Vol. 15, No. 2, 1996
Bu¨hl et al.
to room temperature, the solutions were stirred for an ad-
ditional 1 h. Solvents were removed in vacuo, and the residue
was extracted with pentane. Crystallization at -78 °C af-
forded the complexes as colorless crystals in yields of 60-90%.
(C5H10)2C2(C5H4)2Zr Me2 (8c). Yield: 57%. 1H NMR (600.1
MHz, CDCl3): δ 5.52 (m, 4H, R-Cp), 6.46 (m, 4H, â-Cp), 0.95-
1.65 (m, 20H, CH2), 0.03 (s, 6H, CH3). 13C{1H} NMR (150.9
MHz, CDCl3): δ 107.74 (4C, R-Cp), 115.34 (4C, â-Cp), 131.93
(2C, ipso-Cp), 23.60 (4C, CH2), 26.29 (2C, CH2), 34.66 (4C,
CH2), 49.53 (2C, bridge-C), 27.96 (2C, CH3).
2. NMR Mea su r em en ts. The 91Zr NMR spectra were
recorded on a Bruker AMX-600 spectrometer at 55.76 MHz
with a standard broad-band probe at 300 ( 1 K or 348 ( 1 K,
using a pulse width of 18 µs (ca. 90°) and a repetition delay of
150 ms. The number of scans were ca. 150 for narrow and
50 000 for broad signals, and the aquisition time was e0.05 s.
The 91Zr chemical-shift values, δ(91Zr), are reported relative
to a saturated solution of (C5H5)2ZrBr2 as an external refer-
ence. The FID was treated with a line-broadening function.
The line widths at half-height (∆ν1/2) of the 91Zr signals are
reported in Hz. The T1(13C) values were obtained from 0.1 M
degassed samples by the standard inversion-recovery method
with a triple resonance (1H, 13C, 15N) 5 mm probe on the same
instrument. The spectra were recorded with a recycle time of
ca. 10T1.
Sa m p le P r ep a r a tion . All manipulations were performed
under argon by the use of Schlenk techniques or in a nitrogen-
filled vacuum glovebox. NMR tubes and other glassware in
contact with zirconocene solutions were previously dried by
heating under a vacuum. Solvents were dried and destilled
over sodium/benzophenone or calcium hydride under argon
atmosphere. NMR solvents (d-chloroform, d6-benzene, d8-
toluene, d8-tetrahydrofuran) were dried in the same manner,
desoxygenated, and stored over molecular sieves (4 Å) in a
nitrogen atmosphere. Solutions of the zirconocene complexes
were prepared as concentrated as possible, transferred into 5
mm OD NMR tubes, degassed, and sealed in vacuo. The
samples for the high-temperature NMR measurements were
sealed under atmospheric pressure of argon.
3. Str u ctu r a l Da ta . Crystal structures have been re-
ported for zirconocene dichloro complexes 1a ,58 3a ,59 4a ,38 5a ,39
7a ,42 9a ,44 and 1c.60 The corresponding data for the permethyl
complex 2a have been approximated by those of the closely
related complex (C5Me4Et)2ZrCl2.61 Structural data for the
dibromo complex 1b are based on a crystal structure deter-
mination.
Cr ysta l Da ta for Cp 2Zr Br 2 (1b). C10H10Br2Zr: M ) 381.2,
a ) 7.984(5) Å, b ) 12.135(8) Å, c ) 12.527(9) Å, R ) 71.75(5),
â ) 79.34(5), γ ) 89.36(5)°, V ) 1131.4(12) Å3, triclinic, space
group P1h, Z ) 4, Dc ) 2.238 g/cm3, radiation Mo KR (λ )
0.710 73 Å), µ ) 7.910 mm-1, F(000) ) 720, T ) 243 K, crystal
size (mm) 0.3 × 0.3 × 0.3, diffractometer Siemens R3m/V.
Structure solution and refinement were with Siemens SHELX-
TL PLUS (VMS) software. Solution: Patterson method.
Absorption correction: semiempirical; min/max transmission
0.1199/0.1659. The cell contains two independent molecules.
The Cp ring of one molecule is disordered. All non-hydrogen
atoms were refined anisotropically, except for the disordered
carbon atoms. All hydrogen atom positions were calculated
and refined using the riding model technique with fixed
isotropic U. Final R ) 0.0605, wR ) 0.0732, and S ) 2.21.
Further structural details are in the Supporting Information.
Cr ysta l Da ta for Me4C2(C5H4)2Zr Cl2 (6a ). Complex 6a
cocrystallizes with one solvent molecule from CHCl3 as a
Me4C2(C5H4)2ZrCl2‚CHCl3 adduct. C17H21Cl5Zr: M ) 493.8,
a ) 10.441(2) Å, b ) 8.224(2) Å, c ) 23.143(4) Å, â ) 96.68-
(8)°, V ) 1973.6(8) Å3, monoclinic, space group P21/c, Z ) 4,
Dc ) 1.662 g/cm3, radiation Mo KR (λ ) 0.710 73 Å), µ ) 1.228
mm-1, F(000) ) 992, T ) 238 K, crystal size (mm) 0.3 × 0.3 ×
0.5, diffractometer Siemens R3m/V. Structure solution and
refinement were with Siemens SHELXTL PLUS (VMS) soft-
ware. Solution was by direct methods. All non-hydrogen
atoms were refined anisotropically, and all hydrogen atom
positions were calculated and refined using the riding model
technique with fixed isotopic U. Final R ) 0.0369, wR )
0.0589, and S ) 2.32. Further structural details are in the
Supporting Information.
4. Ab In itio Ca lcu la tion s of Electr ic F ield Gr a d ien ts
a n d 91Zr NMR Sh ifts. The principal values qii of the EFG
tensors have been computed at the SCF level employing the
Gaussian 92 program62 and the following basis sets. Basis A:
Zr (17s13p9d)63,64 contracted to [12s9p5d], including two sets
of diffuse p functions65 (Rp ) 0.11323 and 0.04108) and one
diffuse d-set (Rd ) 0.0382); Br relativistic ME-fit effective core
potential (ecp),66-68 valence [2s2p1d] basis;69 Cl, Si 3-21G(*);70
C, H 3-21G.70 Basis B: Zr same as basis A, decontracted to
[12s10p6d] and augmented with additional sets of diffuse p
and d functions (exponents chosen as one-third of those of the
most diffuse functions of basis A); Br same as basis A, all
others 6-31G*.70 The EFG values are given in atomic units (1
au ) 9.717 365 × 1021 V m-2) and are ordered so that |qzz| >
|qyy| > |qxx|. In addition, EFG tensors have been computed em-
ploying density functional theory (DFT) using the Becke 198871
and the Perdew 198672,73 gradient-corrected exchange-
correlation functionals together with basis B (denoted as
BP86/B).
Geometries have been either taken from X-ray structural
analyses (with C-H bond lengths set to 1.09 Å) or have been
fully optimized70 in the given symmetry at the SCF level
employing basis B′, i.e. same as basis B but using a relativistic
ME-fit ecp and a valence [5s5p2d] basis on Zr.74
NMR chemical shifts have been computed using SCF/B′
geometries, IGLO-SCF,17-19 or GIAO-SCF21,22 methods in their
direct versions,20,23 as implemented in the TURBOMOLE
program75 and the following basis sets. II: Zr same as A,
IGLO-basis II19 for the rest, i.e. Br (15s11p7d2f)/[10s9p5d2f];
Cl (11s7p2d)/[7s6p2d]; C (9s5p1d)/[5s4p1d]; H (5s1p)/[3s1p].
III′: Zr same as A, decontracted to (13s11p8d); Br same as II,
decontracted to [11s10p6d2f]; Cl (12s8p3d)/[8s7p3d]; C,H same
as basis II. Zr chemical shifts are reported relative to (C5H5)2-
ZrBr2, the experimental standard (C2 symmetry, SCF/B′
geometry), with absolute, computed shieldings σ76 of 2917.5
(IGLO-SCF/II),77 2359.5 (IGLO-SCF/III′), 2518.1 (GIAO-SCF/
II), and 2477.7 ppm (GIAO-SCF/III′).
(62) Frisch, M. J .; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
J ohnson, B. G.; Wong, M. W.; Foresman, J . B.; Robb, M. A.; Head-
Gordon, M.; Replogle, E. S.; Gomperts, R.; Andres, L.; Raghavachari,
K.; Binkley, J . S.; Gonzales, C.; Martin, R. L.; Fox, D. J .; DeFrees, D.
J .; Baker, J .; Stewart, J . J . P.; Pople, J . A., Gaussian92/DFT: Pitts-
burgh, PA, 1993.
(63) Horn, H. unpublished data.
(64) Scha¨fer, A.; Horn, H.; Ahlrichs, R. J . Chem. Phys. 1992, 97,
2571.
(65) Walch, S. P.; Bauschlicher, C. W.; Nelin, C. J . J . Chem. Phys.
1983, 79, 3600.
(66) Dolg, M.; Wedig, U.; Stoll, H.; Preuss, H. J . Chem. Phys. 1987,
86, 866.
(67) Dolg, M. Ph.D. Thesis, University of Stuttgart, 1989.
(68) Schwerdtfeger, P.; Dolg, M.; Schwarz, W. H. E.; Bowmaker, G.
A.; Boyd, P. D. W. J . Chem. Phys. 1989, 91, 1762.
(69) Kaupp, M.; Schleyer, P. v. R.; Stoll, H.; Preuss, H. J . Am. Chem.
Soc. 1991, 113, 6012.
(70) Hehre, W. J .; Radom, L.; Schleyer, P. v. R.; Pople, J . A. Ab initio
Molecular Orbital Theory; Wiley: New York, 1986.
(71) Becke, A. D. Phys. Rev. A 1988, 38, 3098.
(72) Perdew, J . P. Phys. Rev. B 1986, 33, 8822.
(73) Perdew, J . P. Phys. Rev. B 1986, 34, 7406.
(74) Andrae, D.; Ha¨ussermann, U.; Dolg, M.; Stoll, H.; Preuss, H.
Theor. Chim. Acta 1990, 77, 123.
(58) Soloveichik, G. L.; Arkhireeva, T. M.; Bel’skii, V. K.; Bulychev,
B. M. Metalloorg. Khim. 1988, 1, 226.
(59) Petersen, J . L.; Egan, J . W. Inorg. Chem. 1983, 22, 3571.
(60) Hunter, W. E.; Hrncir, D. C.; Bynum, R. V.; Penttila, R. A.;
Atwood, J . L. Organometallics 1983, 2, 750.
(75) Ahlrichs, R.; Ba¨r, M.; Ha¨ser, M.; Horn, H.; Ko¨lmel, C. Chem.
Phys. Lett. 1989, 162, 165.
(61) Kurz, S.; Hey-Hawkins, E. Z. Krist. 1993, 205, 61.

91Zr Chemical Shifts and Line Widths
Organometallics, Vol. 15, No. 2, 1996 781
Ta ble 1. 91Zr Ch em ica l Sh ifts a n d Lin e Wid th s of
Zir con ocen e Com p lexesa
Ch a r t 1
zirconocene complex
solvent T/K δ/ppm89 ∆ν1/2/Hz
Cp2ZrCl2 (1a )4
Cp2ZrBr2 (1b)4
Cp2ZrBr2 (1b)4
Cp2Zr(Cl)Me (1d )6
Cp2ZrMe2 (1c)6
Cp2ZrMe2 (1c)
CDCl3
CDCl3
C6D6
CDCl3
C6D6
300
300
300
300
300
348
-112
0
270
16
18
1480
2530
1070
0
126
386
387
C6D6
(C5Me5)2ZrCl2 (2a )5
(C5Me5)2ZrBr2 (2b)
(C5Me5)2ZrMe2 (2c)
(C5Me5)2ZrMe2 (2c)
CDCl3
CDCl3
C6D6
300
300
300
348
85
240
443
451
57
450
3910
2240
C6D6
(C5H4Me)2ZrCl2 (3a )7
(C5H4Me)2ZrBr2 (3b)
(C5H4Me)2ZrMe2 (3c)
(C5H4Me)2ZrMe2 (3c)
CDCl3
CDCl3
C6D6
300
300
300
348
-69
56
401
406
540
140
4030
1630
C6D6
(C5H4tBu)2ZrCl2 (4a )7
(C5H4tBu)2ZrCl2 (4a )
(C5H4tBu)2ZrBr2 (4b)
(C5H4tBu)2ZrMe2 (4c)
(C5H4tBu)2ZrMe2 (4c)
CDCl3
d8-Tol
CDCl3
C6D6
300
348
300
300
348
-61
-64
61
438
422
2810
1130
1700
8250
3230
C6D6
(C5H4SiMe3)2ZrCl2 (5a )56
(C5H4SiMe3)2ZrBr2 (5b)
(C5H4SiMe3)2ZrMe2 (5c)56
C6D6
C6D6
d8-Tol
300
300
348
-89
28
410
1250
260
3240
Me4C2(C5H4)2ZrCl2 (6a )
Me4C2(C5H4)2ZrBr2 (6b)
Me4C2(C5H4)2ZrMe2 (6c)
Me4C2(C5H4)2ZrMe2 (6c)
CDCl3
CDCl3
C6D6
300
300
300
348
-48
65
432
429
700
83
5310
2160
C6D6
Me2Si(C5H4)2ZrCl2 (7a )
Me2Si(C5H4)2ZrBr2 (7b)
Me2Si(C5H4)2ZrMe2 (7c)56
Me2Si(C5H4)2ZrMe2 (7c)
CDCl3
C6D6
d8-Tol
d8-Tol
300
300
300
348
-95
13
427
419
1820
360
4520
2500
(C5H10)C2(C5H4)2ZrCl2 (8a ) CDCl3
(C5H10)C2(C5H4)2ZrBr2 (8b) CDCl3
300
300
-32
80
1000
490
ZrCl4‚2THF (13)
Zr(NMe2)4 (14)
d8-THF 300
C6D6 300
628
893
2920
1130
a
No signals were detectable for the following complexes:
(C5H10)C2(C5H4)2ZrMe2 (8c), rac-Me2Si(C5H2-2-Me-4-tBu)2ZrCl2
(9a ), rac-Me2Si-(1-ind)2ZrCl2 (10a ), rac-Me2Si(ind)2ZrBr2 (10b),
rac-C2H4(1-ind)2ZrCl2 (11a), rac-Me2Si(2-Me-benz[e]-indenyl)2ZrCl2
(12a ), and rac-Me2Si(2-Me-benz[e]indenyl)2ZrBr2 (12b).
dibromide complexes 1b-8b [with the exception of (C5-
Me5)2ZrBr2 (2b), 240 ppm], 386 to 451 ppm for zir-
conocene dimethyl complexes, and finally 628 ppm for
ZrCl4‚2THF (13) and 893 ppm for Zr(NMe2)4 (14). The
(C5H5)2ZrX2 species (X) halogen) exhibit an inverse
halogen effect probably due to the dominance of the
paramagnetic term in the chemical shift tensor.12,15 The
strong deshielding of the permethylated complexes 3a
and 3b was found by Benn and explained by a lowering
of the mean excitation energies.15
Resu lts a n d Discu ssion
1. 91Zr NMR Sp ectr a . The experimental chemical
shift data for the investigated Zr complexes (Chart 1)
are compiled in Table 1.
91Zr NMR signals have been observed for complexes
1a -8a , 1b-8b, and 1c-7c. For complexes 8c, 9a -12a ,
and 10b-12b, 91Zr NMR signals could not be detected
even at 353 K; for these complexes we estimate line
widths ∆ν1/2 > 10 KHz. The 91Zr chemical shift range
is quite large (-112 to 893 ppm) and can be divided into
four regions: -112 to -32 ppm for the zirconocene
dichloride complexes 1a -8a , [with the exception of (C5-
Me5)2ZrCl2 (2a ), 85 ppm], 0 to 80 ppm for zirconocene
There is only a very small concentration effect observ-
able on the 91Zr chemical shift for a series of (C5Me5)2-
ZrCl2 (2a ) samples with concentrations ranging from
0.03 to 0.25 M in CDCl3, the variation ∆δ(91Zr) being
only (0.1 ppm. Likewise, the temperature has a small
effect on the chemical shift of 2a : 85.5 ppm (295 K) and
86.3 ppm (330 K). More pronounced temperature effects
are observed for 2c-7c. As is common for other quad-
rupolar nuclei, the line widths are considerably more
sensitive to temperature changes: for 2a , ∆ν1/2 de-
creases from 59 Hz at 295 K to 50 Hz at 330 K. Even
larger effects can be observed for the broad lines of the
zirconocene dimethyl complexes. The line width
(76) Τhe σ value denotes shielding relative to the bare nucleus and
is converted to δ values according to δ ) σ(standard) - σ. Note the
different sign convention for σ, larger values denote increasing
shielding.
(77) By default, core and valence orbitals are localized separately,
with the Zr 4s and 4p orbitals included in the former. If these orbitals
are localized together with the valence shell, somewhat different
absolute shieldings are obtained; the relative δ values, however, are
affected only by a few ppm.
t
∆ν1/2 of (C5H4 Bu)2ZrMe2 (4c) decreases from 8250 to
3230 Hz upon increasing the temperature from 300 to
348 K.

782 Organometallics, Vol. 15, No. 2, 1996
Bu¨hl et al.
Ta ble 2. Ab In itio Com p u ted Electr ic F ield
3.05, and 3.35 at SCF/A, SCF/B, and BP86/B levels,
respectively)! Likewise, the computed qzz ratios between
1a and Cp2ZrBr2 (1b) are rather similar for the various
methods (ca. 1.35, 1.42, and 1.16 at SCF/A, SCF/B, and
BP86/B, respectively). The computed |qzz| values for
(C5H5)2ZrX2 increase with the experimental line widths,
i.e. in the sequence Br < Cl < Me. It thus appears that,
unlike the absolute, quantitative numbers, the relative
trends in qzz from one molecule to the other can be
described with ab initio levels, at least qualitatively.
Subsequently, EFGs of several other zirconocenes
have been computed at a moderate theoretical level,
SCF/B, employing the solid-state structures. The re-
sults are summarized and related to the experimental
line widths in Table 3. The trends can be best compared
by looking at the ∆ν1/2(rel) and qzz2(rel) values, i.e. the
ratios with respect to the corresponding data for (C5H5)2-
ZrCl2 (last two columns in Table 3). In the series
(C5H5)2ZrX2 (X ) Br, Cl, Me), the ∆ν1/2(rel) and qzz2(rel)
values are very similar, covering almost 2 orders of
magnitude. It thus appears that for these molecules,
the variation of ∆ν1/2 is mainly governed by the changes
a
Gr a d ien ts qii for Zir con ocen es
compd
level
qzz
qyy
qxx
ηb
Cp2ZrBr2 (1b)
SCF/A
SCF/B
-0.2470 0.2336 0.0134 0.891
-0.2043 0.1776 0.0267 0.739
BP86/B -0.2056 0.1575 0.0481 0.532
Cp2ZrCl2 (1a )
SCF/A
SCF/B
-0.3345 0.2451 0.0895 0.465
-0.2903 0.2498 0.0404 0.721
BP86/B -0.2375 0.1839 0.0536 0.549
Cp2ZrMe2 (1c) SCF/A
-0.9950 0.6399 0.3551 0.286
-0.8859 0.5880 0.2979 0.327
SCF/B
BP86/B -0.7965 0.5966 0.1999 0.498
a
Eigenvalues of the diagonalized efg tensor in atomic units,
ordering |qzz| > |qyy| > |qxx|; X-ray-derived geometries employed.58,60
b
Asymmetry parameter η; for definition see text.
As quadrupolar relaxation governs the line width of
the 91Zr resonance, the greatly varying line widths of
the investigated zirconocene complexes are related to
the respective molecular structures via changes of the
electric field gradient components, qii, and of the iso-
tropic molecular correlation times τc.
η2
3
in qzz. For the substituted species (C5H4R)2ZrCl2 (R )
2
2
(
)
H, Me, tBu) (3a -c, 4a -c), however, the changes in qzz
-
∆ν1/2 qzz 1 +
τ
(1)
c
(rel) are much smaller than the variations in ∆ν1/2(rel).
These data, therefore, suggest that the electronic influ-
ence of the alkyl groups R is rather small and that the
In this product, qzz is the largest component of the
electric field gradient (EFG) and η is the asymmetry
parameter defined as (qxx - qyy)/qzz. Apart from tem-
perature and viscosity, τc depends primarily on the
molecular size and should increase with the latter. If
all other parameters were equal, (C5H5)2ZrBr2 (1b )
should, therefore, yield a broader 91Zr resonance than
(C5H5)2ZrCl2 (1a ), and ∆ν1/2 for (C5H5)2ZrMe2 (1c)
should be of the same order of magnitude. However,
the ∆ν1/2 values recorded for the (C5H5)2ZrX2 species are
18, 270, and 2530 Hz for X ) Br, Cl, and Me, respec-
tively, showing a different sequence and an unexpect-
edly large variation. It has thus been suspected that
variations in qzz and/or η are dominant in these cases.
2. EF G Ca lcu la tion s. EFGs and asymmetry pa-
rameters η for (C5H5)2ZrX2 (X ) Br, Cl, Me), computed
at various ab initio levels and employing X-ray-derived
geometries are collected in Table 2. The absolute qii
values (i ) x, y, z) depend strongly on the basis set and
on the theoretical level that is used. In general,
enlargement of the basis set decreases the computed qii
values considerably (compare SCF/A and SCF/B values
in Table 2). Inclusion of electron correlation also affects
qii, mostly resulting in further, smaller reductions
(compare SCF/B and BP86/B data). Thus, quantitative
calculations of EFGs appear very difficult for these
molecules. This also holds, more pronounced, for the
asymmetry parameter η which can even oscillate strongly
with the theoretical level (see Table 2). Since small
numbers and differences between these govern the
magnitude of η in zirconocenes, even qualitative esti-
mates of η based on ab initio data seem virtually
impossible at this time.
t
broadening of the signals for R ) Me and Bu is mainly
caused by larger correlation times τc due to the increas-
ing bulkiness of the ligands.
3. Cor r ela tion Tim es τc. Additional support for
this interpretation comes from isotropic τc values ob-
tained from T1 relaxation times and NOE values of the
13C nuclei of the cyclopentadienyl ligands. The 13C
relaxation time T1dd(13C) due to C-H dipolar interaction
is approximated by78
2
µ0 Nγ(1H)2γ(13C)2p2τ
1
)
c ) NKτc
(2)
6
T1dd(13C)
16π2rCH
K ) 2.147 × 1010 s-2
(cf. ref. 79)
where τc is the isotropic correlation time. Dipolar
interactions from nonbonded protons can be neglected
because of the much larger distances with respect to the
C-H bond lengths (cf. the 1/r6 dependence in eq 2), and
only N, the number of directly attached protons to the
carbon, is of interest. The C-H distances are assumed
as constant (1.09 Å). According to eq 2, τc can be
calculated from the dipole-dipole relaxation time T1dd
-
(13C), the interatomic distance rCH, the number of
directly attached protons N, and the magnetogyric ratio
γ(1H) and γ(13C). T1dd(13C) can directly be obtained from
the NOE factor and the relaxation time T1(13C) (eq 3).
2T1(13C)
T1dd(13C)
NOE ) 1 +
(3)
According to eq 1, however, η affects ∆ν1/2 via the
factor (1 + (η2/3)). Since η assumes values between 0
and 1, this factor can only vary between 1 and 1.33, and
the other parameters in eq 1, qzz in particular, may be
more important. In fact, the absolute numbers of qzz
computed for (C5H5)2ZrMe2 (1c) are ca. three times as
large as those obtained for (C5H5)2ZrCl2 (1a ) (Table 2),
irrespective of the theoretical level (the ratios are ca. 2.97,
T1(13C) values of the zirconocene compounds 1a , 2a , 3a ,
4a , 5a , and 7a have been determined with the standard
inversion-recovery NMR experiment. The NOE values
(78) Abragam, A., Ed. The Principles of Nuclear Magnetism; Oxford
University Press: Oxford, U.K., 1961; Chapter 9.
(79) µo ) 4π × 10-7 m kg s-2 A-2; γ(1H) ) 2.6752 × 108 kg-1 s A
rad; γ(13C) ) 0.6726 × 108 kg-1 s A rad; rCH ) 1.09 Å.

91Zr Chemical Shifts and Line Widths
Organometallics, Vol. 15, No. 2, 1996 783
Ta ble 3. Rela tion betw een Exp er im en ta l 91Zr NMR Lin e Wid th s ∆ν1/2, Com p u ted EF Gs,a a n d Tilt An gles τ
for Zir con ocen es
compd
∆ν1/2/Hz
qzz
∆ν1/2(rel)b
qzz2(rel)b
τ/deg
Cp2ZrBr2 (1b)
Cp2ZrCl2 (1a )
Cp2ZrMe2 (1c)
16
270
2530
-0.2043
-0.2903
-0.8859
0.1
1.0
9.4
0.5
1.0
9.3
(C5Me5)2ZrCl2 (2a )
Cp2ZrCl2 (1a )
57
270
540
2810
1250
1820
n.d.d
-0.1772
-0.2903
-0.2453
-0.4074
-0.1496
-0.4830
-0.7062
0.2
1.0
2.0
10.4
4.6
6.7
0.3
1.0
0.7
2.0
0.3
2.8
5.9
0.8
0.7
1.7
2.9
0.8
2.8
4.5
(C5H4Me)2ZrCl2 (3a )
(C5H4tBu)2ZrCl2(4a )
(C5H4SiMe3)2ZrCl2 (5a )
rac-Me2Si(C5H4)2ZrCl2 (7a )
rac-Me2Si(C5H2-2-Me- 4-tBu)2ZrCl2 (9a )c
a
Largest components, qzz, computed at the SCF/B level employing X-ray-derived geometries. b Ratio with respect to the value for Cp2ZrCl2.
c In the calculation, the solid-state structure has been employed with t-Bu and Me substituents replaced by H. Not detected.
d
Ta ble 4. Rela xa tion Tim es T1(13C), NOE, a n d
Cor r ela tion Tim es τc of Zir con ocen e Dich lor id e
Com p lexes
compd
T1/s
NOE T1DD/s 10-12τc/s τc(rel)
(C5H5)2ZrCl2 (1a )
(C5H4Me)2ZrCl2 (3a )
8.60 (CH) 2.30 13.2
3.5
5.3
6.3
11.0
17.1
19.4
19.7
11.2
8.7
1.0
1.8
4.9
5.60 (CH) 2.28
5.71 (CH) 2.55
2.41 (CH) 2.34
2.39 (CH) 2.79
8.8
7.4
4.2
5.4
2.4
2.4
4.2
5.4
(C5H4tBu)2ZrCl2(4a )
(C5H4SiMe3)2ZrCl2 (5a ) 2.00 (CH) 2.67
1.95 (CH) 2.65
Me2Si(C5H4)2ZrCl2 (7a ) 2.83 (CH) 2.36
3.41 (CH) 2.27
5.6
3.2
F igu r e 1. Schematic representation of zirconocenes Cp2-
ZrX2 indicating geometrical parameters.
suggests that geometry effects may be important for the
EFGs in zirconocenes. To study the influence of certain
geometrical parameters in more detail, EFGs have been
computed at the SCF/B level for specifically altered
(C5H5)2ZrCl2 geometries. Starting from the solid-state
structure, the orientations of the Cp ligands have been
varied by separately changing the CpZrCp′ angle, R, the
inclination angle, â, or the twist angle, θ (cf. the
schematic representation in Figure 1), leaving all other
parameters unchanged.
of the 13C atoms have been obtained from comparisons
of the intensities of the 13C resonances of inverse gated
and continously composite pulse 1H-decoupled 13C NMR
spectra (Table 4).
The relative correlation times τc(rel) for (C5H5)2ZrCl2
t
(1a ), (C5H4Me)2ZrCl2 (3a ), (C5H4 Bu)2ZrCl2 (4a ), (C5H4-
SiMe3)2ZrCl2 (5a ), and Me2Si(C5H4)2ZrCl2 (7a ) confirm
our suggestion that τc is the dominant factor for the
relative line widths in these cases. Even though larger
in size than the parent (C5H5)2ZrCl2 (1a ), (C5Me5)2ZrCl2
(2a ) shows a narrower line in the 91Zr NMR spectrum.
According to the data in Table 3, the smaller ∆ν1/2 of
2a is largely due to the decrease of qzz.
Closing R by 15° increases |qzz| (0.4197), while in-
creasing R by the same amount results in a smaller EFG
(0.1289). Larger variations in |qzz| are found upon
changes in â by +7.5° (0.7915) and by -7.5° (0.6281) or
upon twisting the Cp rings in opposite directions by θ
) 7.5° (0.7915). This indicates that the EFG is more
dependent on the magnitude of the deviation of the
normal to the Cp planes from the direction of the
zirconium-zentroid vector, i.e. the tilt angle τ, than on
the direction of this distortion. Experimental tilt angles
τ are available for a series of zirconocenes 1a -5a and
7a ,and 9a from X-ray crystallographic analy-
ses38,39,42,44,58,59 (for 2a we use the tilt angle determined
from the crystal structure of (C5Me4Et)2ZrCl261), and
they are listed in Table 3. The experimental 91Zr NMR
line widths ∆ν1/2, the computed qzz2 (rel) values, and the
tilt angle τ exhibit parallel trends. It appears that the
exact position of the Zr atom with respect to each Cp
ring plane is more important for the EFG than is the
CpZrCp′ angle (see also the chemical shift section).
To separate possible inductive effects from these
geometrical effects, model calculations have been per-
formed for the zirconocenes 2a -7a with the substitu-
ents at the Cp rings replaced by hydrogens, leaving all
other parameters unchanged. This procedure affords
the parent 1a but in the geometry of the various,
substituted derivatives. In most cases, only minor
changes are found for the computed qzz values: e.g.,
4. Coor d in a tion Geom etr ies. ansa-compounds
bearing a dimethylsilylene bridge between the two Cp
rings show broad 91Zr NMR resonances. ansa-Me2Si-
(C5H4)2ZrCl2 (7a ) gives a somewhat increased, computed
qzz2(rel) value (see Table 3), possibly contributing to this
line broadening. The signal of a highly substituted
derivative, ansa-Me2Si(C5H2-2-Me-4-tBu)2ZrCl2 (9a ), is
undetectably broad at room temperature. Compound
9a is rather large for ab initio computations; a calcula-
tion has been performed by employing the X-ray struc-
t
ture of 9a with Me and Bu groups replaced by H, i.e.
for a “distorted” 7a geometry. Interestingly, this “dis-
tortion” induced by the bulky groups increases the
computed relative qzz2 significantly (ca. by a factor of 2;
cf. the last two entries in Table 3). Taking into account
the additional line broadening due to the bulky substit-
uents (cf. the factor of ca. 10 in going from 1a to the
tBu derivative 4a ), ∆ν1/2 estimated for 9a would be much
larger than 10.000 Hz, i.e. too broad to be observable.
It is unfortunate that the excessive line widths of the
catalytically most interesting complexes 9a -12a pre-
clude any correlations between 91Zr NMR data and
catalytic properties.
The notable variation of the qzz values computed for
the same ansa-compounds in two different geometries
t
going from R ) Bu to R ) H in (C5H4R)2ZrCl2 (4a )

784 Organometallics, Vol. 15, No. 2, 1996
Bu¨hl et al.
Ta ble 5. Ab In itio Com p u ted 91Zr Ch em ica l Sh ifts
for Zir con iu m Com p lexesa
changes qzz from -0.4074 (Table 3) to -0.4718;80 i.e. the
t
“electronic effect” of the Bu group is small. This holds
for most other Cp-substituted derivatives, except for
(C5Me5)ZrCl2 (2a ): replacing all methyl groups by
hydrogens affords a change in qzz from -0.1772 (Table
3) to -0.3227, close to the value of 1a (-0.2903). In
this case, it appears to be the electronic effect exerted
by the highly basic C5Me5 ligand that is responsible for
compd (sym)
2-
2-
Cp2ZrCl2 (C5Me5)2ZrCl2 Cp2ZrMe2 ZrF6
ZrCl6
method
(C2)
(Cs)
(C2v
)
(Oh)
(Oh)
IGLO-SCF/II
IGLO-SCF/III′
-87
-91
37
434
435
-144
-162
588
566
GIAO-SCF/II
GIAO-SCF/III′
-85
-92
28
85
406
412
-12
-66
-191b
612
601
the reduction of |qzz| and, hence, of ∆ν1/2
.
exp
-112
386
601c
Because of the relevance to homogeneous catalysis,
structures and reactivities of zirconocenes have recently
been studied in great detail, both at the SCF and DFT
levels.81-83 A detailed discussion of the ab initio geom-
etries would be beyond the scope of this paper. Geom-
etry optimizations at the SCF/B′ level of (C5H5)2ZrX2
(X ) Br, Cl, Me) afford Zr-X and average Zr-C
distances that are somewhat overestimated, e.g. by ca
2 and 8 pm, respectively, for X ) Cl with respect to the
gas-phase structure.84 The relative conformations of the
two Cp rings are correctly reproduced at that level, i.e.
“staggered” for X ) Br and Cl (C2 more stable than C2v
by 0.2 kcal/mol each) and “eclipsed” for (C5H5)2ZrMe2
(1c) (C2v more stable than Cs by 0.4 kcal/mol).
a
δ values in ppm relative to Cp2ZrBr2 (C2 symmetry); SCF/B′
geometries employed. b [NH4]2[ZrF6]/D2O. c H2[ZrCl6]/concentrated
HCl.
Ta ble 6. P r in cip a l Va lu es of th e 91Zr Ch em ica l
Sh ift Ten sor s in Zir con ocen esa ,b
compd
δ11
δ22
δ33
δav
Cp2ZrCl2 (1a )
Cp2ZrBr2 (1b)
Cp2ZrMe2 (1c)
-244
-208
-180
-182
-138
-116
+152
+346
+1600
-91
0
435
aIn ppm relative to the isotropic value for Cp2ZrBr2, IGLO-SCF/
b
III′ level. Orientation of the principal axes: σ33 perpendicular
to the ZrX2 plane; σ22 along the 2-fold axis bisecting the XZrX′
angle; σ11 in the ZrX2 plane perpendicular to σ11 and σ22
.
When the ab initio geometries are employed for the
EFG calculations, the results depend somewhat on the
relative Cp conformations. When the lowest-energy
structures are used, the computed EFGs differ slightly
from those in Table 2, obtained for the solid-state
geometries, but the trends remain the same. In par-
ticular, the qzz2(rel) values are in good accord for both
sets of structures; cf. 0.3, 1, and 9.7 for X ) Br, Cl, and
Me, respectively (SCF/B′ geometries) vs 0.5, 1, and 9.3,
respectively (X-ray-derived geometries, Table 3).
sponding IGLO and GIAO values are apparent. Prob-
ably, even larger basis sets are required in this case to
approach the SCF limit. Note that the inverse halogen
effect, studied recently for Ti compounds in great
detail,85 is correctly reproduced computationally for
(C5H5)2ZrCl2 (1a ) and (C5H5)2ZrBr2 (1b).
The principal values of the 91Zr NMR chemical shift
tensors in the zirconocenes (C5H5)2ZrX2 (X ) Cl, Br, Me)
are given in Table 6. The components δ11 and δ22 are
always shielding and show only relatively small varia-
tions with X. Changes in δ33, the component perpen-
dicular to the ZrX2 plane, are much larger, in particular
when going to X ) Me. The principal axis systems of
the EFG tensors are oriented similarly as those of the
chemical shift tensors; the direction of largest deshield-
ing, δ33, coincides with that of the largest absolute
EFG, qzz.
Except for ZrF62-, the computed and experimental
isotropic δ(91Zr) chemical shifts agree reasonably well;
the maximum deviation is 57 ppm for (C5Me5)2ZrCl2
(GIAO-SCF/II level, Table 5), and mean absolute devia-
tions are on the order of 30 ppm, i.e. a few percent of
the total 91Zr chemical shift range (more than 1300
ppm). This performance is satisfactory, especially when
bearing in mind that electron correlation and relativistic
effects are neglected in the calculations. In addition,
theoretical values for isolated molecules are compared
to experimental data obtained in solution. Both the
experimental 91Zr chemical shift range and the relative
sequence of the zirconocene derivatives are correctly
reproduced at the SCF level. Thus, calculations at this
level should allow reliable predictions of qualitative
trends for 91Zr chemical shifts.
5. 91Zr Ch em ica l Sh ifts. In order to test how well
91Zr chemical shifts of zirconocenes can be described
theoretically, the δ(Zr) values of Cp2ZrX2 (X ) Br, Cl,
Me) (1a -c) have been computed with the IGLO-SCF
and the GIAO-SCF methods. The hexahalogeno com-
2-
plexes ZrX6 (X ) F, Cl) have also been considered
because they cover a somewhat larger chemical shift
range. It is known for first-row compounds that opti-
mized geometries often perform better in chemical shift
calculations than do experimental structures.26,27 This
may not always be the case; for a consistent description,
however, the IGLO and GIAO computations employed
ab initio geometries optimized at a comparable theoreti-
cal level, SCF/B′.
As can be seen from the results in Table 5, the trend
of δ(91Zr) is well reproduced for the zirconocenes at
IGLO-SCF/II and GIAO-SCF/II levels, over a range of
ca. 500 ppm. Further increase of the basis set affects
the computed δ values only slightly, up to ca. 10 ppm
(compare the corresponding II and III′ entries in Table
5). The only case with larger deviations between
theoretical and experimental data is ZrF62-, for which
also unusually large differences between the corre-
One such trend, postulated by Bo¨hme, Thiele, and
Rufinska,11 is the relationship between δ(91Zr) and the
CpZrCp′ angle, R; the Zr atom should be deshielded with
increasing R, cf. the 125 ppm downfield shift in going
from (C5H5)2ZrCl2 (1a ) to (C5H2-1,2,4-iPr3)2ZrCl2, which
is paralleled by an opening of R by ca. 5°. When R is
(80) The difference of the latter value compared to that computed
for 1a , -0.2903 (Table 3), is a geometrical effect.
(81) Yoshida, T.; Koga, N.; Morokuma, K. Organometallics 1995,
14, 746.
(82) (a) Fan, L.; Harrison, D.; Woo, T. K.; Ziegler, T. Organometallics
1995, 14, 2018. (b) Lohreuz, J . C. W.; Woo, T. K.; Fan, L.; Ziegler, T.
J . Organomet. Chem. 1995, 497, 91.
(83) Meier, R. J .; Doremaele, G. H. J . v.; Iarlori, S.; Buda, F. J . Am.
Chem. Soc. 1994, 116, 7274.
(84) Ronova, I. A.; Alekseev, N. V. Russ. J . Struct. Chem. (Engl.
Transl.) 1977, 18, 180.
(85) Berger, S.; Bock, W.; Frenking, G.; J onas, V.; Mu¨ller, F. J . Am.
Chem. Soc. 1995, 117, 3820.

91Zr Chemical Shifts and Line Widths
Organometallics, Vol. 15, No. 2, 1996 785
inactive.87,88 The Zr atom is shielded even though two
2c-2e bonds are formally replaced by two 3c-2e bonds;
i.e., electron density is actually removed from the Zr-C
bonds. The strong shielding of δ(91Zr) in 16 is paralleled
by a decrease in the computed EFG: the qzz2(rel) value
is only ca. half of that obtained for 1c. Thus, the extent
of the line broadening for 16 due to a larger τc should
be significantly reduced because of the smaller EFG,
and the δ(91Zr) NMR signal could well be observable.
Con clu sion
F igu r e 2. SCF/B′-optimized structure of the [Cp2ZrMe2-
AlMe2]+ complex (16, C2v symmetry imposed).
91Zr Chemical shifts and line widths of ring-bridged
and ring-substituted zirconocenes depend considerably
on electronic and steric effects that can be adequately
reproduced by ab initio computations. Signal detection
is limited to complexes with line widths < 10 kHz, thus
excluding the compounds of catalytic interest. Line
widths ∆ν1/2 are not solely determined by the correlation
times, τc (and, thus, by the molecular size), but also by
the magnitude of the electric field gradient, EFG, at the
Zr atom. Ab initio computations suggest that the EFG
is decisive for ∆ν1/2 when the substituents at Zr are
varied, e.g. in the series (C5H5)2ZrX2 (X ) Br, Cl, Me)
(1a -c). Substituents at the Cp rings usually exert a
much smaller effect on the computed EFGs at the Zr
atom. As a consequence, the line widths should be
dominated by τc in these cases, in accord with experi-
mental observations.
increased by as much as 15° in the SCF/B′ geometry of
1a , indeed a downfield shift is computed for δ(91Zr),
albeit only by 14 ppm with respect to the undistorted
structure. Closing R by 15° results in an increased,
computed shielding by -23 ppm. As found for the EFGs
(see above), tilting the Cp rings affects δ(91Zr) much
more: changing â by -7.5° and +7.5° affords computed
∆δ(91Zr) shifts of +14 and +78 ppm, respectively.
Likewise, twisting both Cp rings by θ ) 7.5° results in
a computed downfield shift of ∆δ(91Zr) ca. +49 ppm. It
therefore appears that 91Zr chemical shifts depend on
all geometrical parameters of the (C5H5)2Zr moiety and
not just on the CpZrCp′ angle, R. In particular, δ(91Zr)
seems to be more sensitive to the Cp inclination and
twist angles, â and θ, respectively. In contrast to the
suggestions of Bo¨hme et al., the theoretical 91Zr chemi-
cal shifts are almost unaffected by the ClZrCl′ angle:
changing this angle by +5 and -5° results in computed
∆δ(91Zr) shifts of only +4 and -1 ppm, respectively.
Another interesting aspect concerns the consequences
of electron deficiency on the magnetic properties of
cationic zirconocene complexes.10 Species of the type
(C5H5)2ZrR+ (R ) alkyl) are believed to be the key
intermediates in the catalytic olefin polymerization
process.2-5 It would be very interesting to probe the
active catalyst by means of 91Zr NMR spectroscopy, but
(C5H5)2ZrR+ species are far too reactive to be observable
on the NMR time scale. An electronic structure not
unlike that of (C5H5)2ZrMe+ (15) may be preserved in
complexes such as (C5H5)2ZrMe2‚B(C6F5)3;52 cf. the
canonical resonance formulation as [(C5H5)2ZrMe]+-
[MeB(C6F5)3]-. However, 91Zr NMR signals could not
be recorded for this or for similar complexes, presumably
due to very large correlation times, τc.
Experimental trends in 91Zr NMR chemical shifts of
zirconocenes are well reproduced computationally em-
ploying IGLO-SCF or GIAO-SCF methods and large
basis sets. Model calculations suggest that δ(91Zr), as
well as the EFGs, are quite sensitive to the structural
parameters, in particular to the inclination and twist
angles of the Cp rings, and, to a lesser extent, to the
CpZrCp′ angle. A substantial deshielding, δ(91Zr) ca.
700 ppm, is predicted for (C5H5)2ZrMe+ (15), the active
olefin-polymerizing catalyst. In contrast, the Zr atom
is strongly shielded in going to [(C5H5)2ZrMe2AlMe2]+
(16), formally a complex of 15 and AlMe3, δ(calc) ) -12
ppm.
Ack n ow led gm en t. This work has been supported
by the Swiss National Science Foundation, the Fonds
der Chemischen Industrie, BMFT, and the Volkswagen-
Stiftung. G.H. wishes to thank Dr. W. Kozminski, Dr.
D. Nanz, and Dr. A. Linden for helpful support. M.B.
gratefully acknowledges generous support from W.
Thiel. Calculations have been performed on IBM/RS
6000 workstations of the Organisch-chemisches Institut
and of the Rechenzentrum der Universita¨t Zu¨rich. S.B.
and H.H.B. thank Mr. T. Haselwander for providing
crystals of Me4C2(C5H4)2ZrCl2‚CHCl3.
According to the ab initio δ(91Zr) value predicted for
(C5H5)2ZrMe+ (15), 698 ppm (GIAO-SCF/II level em-
ploying the SCF/B′ geometry, Cs symmetry imposed),
the Zr atom is strongly deshielded in such cationic,
unsaturated complexes. Even larger Zr chemical shifts
are known, e.g. δ ) 893 ppm for Zr(NMe2)4 (14). In
contrast, the metal is strongly shielded when multi-
center bonding is involved: for [(C5H5)2ZrMe2AlMe2]+
(16), formally a complex of 15 and AlMe3 (Figure 2), a
91Zr chemical shift of -12 ppm is computed (GIAO-SCF/
II, C2v-symmetric SCF/B′ geometry).
Su p p or tin g In for m a tion Ava ila ble: ORTEP diagrams
and tables of X-ray parameters, atom coordinates, thermal
parameters, and bond distances and angles (21 pages). Order-
ing information is given on any current masthead page.
OM950757C
[(C5H5)2ZrMe2AlMe2]+[B(C6F5)4]- (16) is reported to
be stable at room temperature and is inactive as a po-
lymerization catalyst.86 The increase in Zr shielding
predicted for this cationic complex with respect to neu-
tral (C5H5)2ZrMe2 (1c) is noteworthy, ∆δ ca. -400 ppm.
In this context it is interesting to note that highly shield-
ed transition metal complexes are often catalytically
(86) Bochmann, M.; Lancaster, S. J . Angew. Chem., Int. Ed. Engl.
1994, 33, 1634.
(87) Bender, B. R.; Koller, M.; Nanz, D.; von Philipsborn, W. J . Am.
Chem. Soc. 1993, 115, 5889.
(88) Bo¨nnemann, H.; Brijoux, W.; Brinkmann, R.; Meurers, W.;
Mynott, R.; von Philipsborn, W.; Egolf, T. J . Organomet. Chem. 1984,
272, 231.
(89) The maximum uncertainty of the chemical shifts is (10 ppm
for broad lines.