Microwave spectra and the molecular structure of tetracarbonylethyleneiron

Article abstract of DOI:10.1021/ja983937x

Microwave spectra of seven isotopomers of tetracarbonylethyleneiron were recorded using a Pulse-Beam Fourier Transform Microwave Spectrometer. Rotational transitions for a c dipole moment with J′ ← J from 2 ← 1 to 6 ← 5 were measured in the 4-12 GHz range. Rotational constants were determined by fitting the measured microwave spectra to a Watson A reduced Hamiltonian with centrifugal distortion parameters. The measured rotational constants of the main isotopomer are A = 1031.1081(4) MHz, B = 859.8055(4) MHz, and C = 808.5675(3) MHz. Data were also obtained for three ¹³C-substituted species and two ¹⁸O-substituted species in natural abundance. Additional spectra were measured for an isotopically enriched sample of perdeuterated tetracarbonylethyleneiron. The moments of inertia of the seven isotopomers were used in a Kraitchman analysis and in two different least-squares fitting analyses to determine the molecular structure of the compound. The ethylene ligand exhibits significant structural changes upon complexation to iron, primarily an increase in C-C bond length with movement of the hydrogen atoms away from the metal center. The C-C and C-H bond lengths were found to be r_o = 1.419(7) and = 1.072(4) A, respectively. The C-C-H angle and the Fe-C-C-H dihedral angle were found to be 〈(C-C-H)_o = 120.6(5)° and 〈(Fe-C-C-H)_o = 103.6(9)°, respectively. The plane of the hydrogen atoms is displaced 0.217(2) A above the ethylene carbon atoms, along the c axis. Extensive DFT calculations were carried out prior to the experimental research. The calculated structure proved extremely valuable in obtaining accurate predictions for the spectra, and provided structural parameters in excellent overall agreement with measured parameters.

Full text of DOI:10.1021/ja983937x

J. Am. Chem. Soc. 1999, 121, 4023-4030
4023
Microwave Spectra and the Molecular Structure of
Tetracarbonylethyleneiron
Brian J. Drouin and Stephen G. Kukolich*
Contribution from the Chemistry Department, UniVersity of Arizona, Tucson, Arizona 85721
ReceiVed NoVember 12, 1998
Abstract: Microwave spectra of seven isotopomers of tetracarbonylethyleneiron were recorded using a Pulse-
Beam Fourier Transform Microwave Spectrometer. Rotational transitions for a “c” dipole moment with J′ r
J from 2 r 1 to 6 r 5 were measured in the 4-12 GHz range. Rotational constants were determined by
fitting the measured microwave spectra to a Watson “A” reduced Hamiltonian with centrifugal distortion
parameters. The measured rotational constants of the main isotopomer are A ) 1031.1081(4) MHz, B )
859.8055(4) MHz, and C ) 808.5675(3) MHz. Data were also obtained for three 13C-substituted species and
two 18O-substituted species in natural abundance. Additional spectra were measured for an isotopically enriched
sample of perdeuterated tetracarbonylethyleneiron. The moments of inertia of the seven isotopomers were
used in a Kraitchman analysis and in two different least-squares fitting analyses to determine the molecular
structure of the compound. The ethylene ligand exhibits significant structural changes upon complexation to
iron, primarily an increase in C-C bond length with movement of the hydrogen atoms away from the metal
center. The C-C and C-H bond lengths were found to be ro ) 1.419(7) and ) 1.072(4) Å, respectively. The
C-C-H angle and the Fe-C-C-H dihedral angle were found to be (C-C-H)o ) 120.6(5)o and (Fe-
C-C-H)o ) 103.6(9)o, respectively. The plane of the hydrogen atoms is displaced 0.217(2) Å above the
ethylene carbon atoms, along the c axis. Extensive DFT calculations were carried out prior to the experimental
research. The calculated structure proved extremely valuable in obtaining accurate predictions for the spectra,
and provided structural parameters in excellent overall agreement with measured parameters.
Introduction
simple complex. An early electron diffraction study11 (GED)
verified the basic trigonal bypyramidal structure, and this was
supported by the assignment of the vibrational spectrum.12 The
ethylene is bound on one of the equatorial sites of a near trigonal
bypyramidal structure, the C-C bond lies in the equatorial
plane, and the molecule has C2V symmetry with the C2 rotation
axis bisecting the olefin bond. Unfortunately, the GED struc-
ture11 does not provide hydrogen coordinates, and there were
difficulities in locating hydrogen atoms for an X-ray study8,13
of a similar compound. This solid-state X-ray structure is for
the closely related compound8,13 Fe(P(C6H5)3)(CO)3C2H4. This
X-ray structure includes a reasonable Fe-C-C-H dihedral
angle but the C-H distances are not well determined, indicating
that the direction of the bond was determined more accurately
than the length. No X-ray study of the title compound was found
in the literature, presumably because of the high sensitivity to
light, heat, and air. Accurate hydrogen coordinates for this
molecule are important because structural changes of the
ethylene upon complexation help to indicate how much the metal
center can activate the olefin. In a previous study, accurate
coordinates for the hydrogen atoms provided evidence for a
dramatic change in hybridization from sp2 to near sp3 for the
terminal H atoms of butadiene, when complexed to iron.14 The
X-ray structure of Fe(P(C6H5)3)(CO)3C2H4 indicates that the
C-H bonds are bent 8° out of the (former) ethylene plane. This
suggests that a similar hybridization change occurs for the
Olefin activation on metal catalysts is often used today in
organic1-5 syntheses. The commonly accepted mechanisms6 for
olefin isomerization involve metallocyclopropane intermediate
complexes. These short-lived intermediates are quickly con-
verted to either allylic or alkyl bound intermediates or released
from the metal to give a free olefin. Experiments are often
carried out on closely related model compounds since reaction
intermediates are notoriously difficult to study. In many cases,
the most closely related model compounds are still quite
reactive. This is certainly the case for the present example,
tetracarbonylethyleneiron.7,8 The ethylene ligand is very labile
in this compound as indicated by rapid decomposition under
light to form Fe3(CO)12. This compound is an example of a
simple metal bound olefin complex since it contains only one
olefinic bond formed from π and π* orbitals interacting with d
orbitals on the metal center. The characteristics and reactivity
of typical metal-olefin bonds9,10 are well represented in this
(1) Miller, M. E.; Grant, E. R. J. Am. Chem. Soc. 1985, 107, 3386.
(2) Wuu, Y.; Bensten, J. G.; Brinkley, C. G.; Wrighton, M. S. Inorg.
Chem. 1987, 26, 530.
(3) Kane, V. V.; Light, J. R.; Whiting, M. C. Polyhedron 1985, 4, 533.
(4) Roberts, B. W.; Wong, J. J. Chem. Soc., Chem. Commun. 1977, 20.
(5) Baar, M. R.; Roberts, B. W. J. Chem. Soc., Chem. Commun. 1979,
1129.
(6) Crabtree, R. H. The Organometallic Chemistry of the Transition
Metals; Wiley-Interscience Inc.: New York, 1994; Chapter 9.
(7) Murdoch, H. D.; Weiss, E. HelV. Chem. Acta 1963, 66, 1588.
(8) Lindner, E.; Schauss, E.; Hiller, W.; Fawzi, R. Angew. Chem., Int.
Ed. Engl. 1984, 23, 711.
(9) Barnhart, T. M.; Fenske, R. F.; McMahon, R. J. Inorg. Chem. 1992,
31, 2681.
(11) Davis, M. I.; Speed, S. J. Organomet. Chem. 1970, 21, 401.
(12) Andrews, D. C.; Davidson, G. J. Organomet. Chem. 1972, 35, 161.
(13) Lindner, E.; Schauss, E.; Hiller, W.; Fawzi, R. Chem. Ber. 1985,
118, 39 15.
(14) Kukolich, S. G.; Roehrig, M. A.; Wallace, D. W.; Henderson, G.
L. J. Am. Chem. Soc. 1993, 115, 2021.
(10) Beach, D. B.; Jolly, W. L. Inorg. Chem. 1983, 22, 2137.
10.1021/ja983937x CCC: $18.00 © 1999 American Chemical Society
Published on Web 04/06/1999

4024 J. Am. Chem. Soc., Vol. 121, No. 16, 1999
Drouin and Kukolich
ethylene ligand. Numerous theoretical studies15-17 have been
performed to assess the degree to which the tetracarbonyl iron
fragment can activate olefins. When detailed structural data are
available, the accuracy of the theoretical analyses can be better
evaluated. Further exploitation of the theory may help to
elucidate the remarkable properties of this molecule, including
a rich photochemistry,18 and catalytic activity in the isomer-
ization of alkenes.2
calculations in an effort to further test the application of DFT
in predicting gas-phase structures of organometallic complexes.
Experimental Section
The sample was prepared using the method of Murdoch and Weiss7
with only small modification. One gram of Fe2(CO)9 was placed in a
40 mL stainless steel bomb reactor with 10 mL of pentane. The bomb
was charged with 30 atm of ethylene by condensing 1.2 L of ethylene
at 1 atm into the bomb at -196 °C. The reaction was run for at least
2 days and then the excess ethylene was vented and the solvent was
removed under vacuum at -80 °C. The rest of the sample manipulation
was performed in a darkened laboratory, due to the high light sensitivity
of this compound.17 Exposure to fluorescent light caused rapid
decomposition as indicated by the production of a dark green compound
(Fe3(CO)12).
It is well-documented that one byproduct of this reaction, Fe(CO)5,
is extremely difficult to remove10,13 from the product C2H4Fe(CO)4.
Due to the nature of the microwave experiment, a highly pure sample
was not necessary for measurements of strong microwave transitions.
Therefore, a typical sample preparation involved only a partial
separation of the products. After solvent removal under vacuum at -80
°C, the bomb reactor was slowly warmed and the contents were
removed under vacuum into two fractions. The first fraction, containing
mostly pentacarbonyliron, was collected up to a temperature of -35
°C. The second fraction, containing mostly tetracarbonylethyleneiron,
was then collected from -35 to 0 °C. At this point little or no material
was observed to be leaving the bomb. Infrared analyses of the different
fractions revealed significant amounts of each product in both fractions,
and the overlap of peaks in the C-O and Fe-C stretching regions
was significant.
Microwave measurements were made in the 4-12 GHz range using
a Flygare-Balle type pulsed-beam Fourier transform microwave spec-
trometer.28 The sample from the second fraction was maintained at -35
°C, as the tetracarbonylethyleneiron was distilled into a small glass
sample cell by applying vacuum through the attached pulsed valve and
placing a small cup of liquid nitrogen (-196 °C) around the sample
cell. The pulse valve was then closed and the sample chamber filled
with 1 atm of Ne carrier gas. The cell was warmed to 0 °C. At lower
frequencies a pressure of 1.5 atm was seen to increase signal intensity.
Transitions measured are somewhat broadened due to unresolved
Doppler components and perhaps even unresolved spin-rotation cou-
pling. Greater broadening was observed for the perdeuterated sample,
presumably due to unresolved quadrupole splitting. The standard
deviations listed for the measured transitions listed in Tables 1-3 are
1σ. Systematic errors in the frequencies are expected to be much less
(1 part in 109) than the reported random errors because the spectrometer
is periodically calibrated to the frequency standard broadcast from
WWVH in Boulder, CO.
The title compound is structurally similar to the series of
dihydrides19-21 previously studied by this group, particularly
H2Fe(CO)4.20 The ethylene replaces the two hydride ligands
bound to the tetracarbonylferrate base. Both of these molecules
are thermal, light, and air sensitive. Chemically, the molecule
is much more similar to the η4- and η5-bound Fe systems also
studied by microwave spectroscopy.14,22-25 The tricarbonylcy-
clobutadieneiron,11,21 tricarbonylbutadieneiron,11,14 and tricar-
bonylcyclooctatetraeneiron23 species are all η4 type “piano-stool”
complexes with increasing stability in the order listed. The
sandwich compounds chloro- and bromoferrocene24,25 are both
very stable in comparison with the diene-iron-carbonyls, but
are less stable than the parent compound, ferrocene. Complete
gas-phase structures have been published for H2Fe(CO)419 and
14
C4H6Fe(CO)3 along with some structural data for C4H4Fe-
(CO)3 and the haloferrocenes.24,25 Previous electron diffrac-
21
tion11 work has shown that structural comparisons between these
species provide valuable insight into the characteristics of the
compounds and their reactivity. In the present work microwave
spectra for a large number of isotopomers are given and a very
detailed structure of the η2-bound ethylene complex is presented.
The experimental results, along with the DFT calculations
performed, provide a more complete view of structure and
bonding for the olefin-iron systems.
Density functional calculations are becoming increasingly
accurate and useful for describing structures of organometallic
systems. Previously the BPW9126 methods of Gaussian27 were
shown to reproduce the gas-phase structure of H2Fe(CO)419 with
great accuracy. It is useful to test the accuracy of quantum
chemical theories with relatively small systems to determine if
their application to larger structures is valid. Many of the iron
systems studied in this laboratory have been modeled using DFT
(15) Axe, F. U.; Marynick, D. S. J. Am.Chem. Soc. 1984, 106, 6230.
(16) Li, J.; Schreckenbach, G.; Ziegler, T. Inorg. Chem. 1995, 34, 3245.
(17) Albright, T. A.; Hoffman, R.; Thibeault, J. C.; Thorn, D. L. J. Am.
Chem. Soc. 1979, 101, 3801.
(18) Mitchener, J. C.; Wrighton, M. S. J. Am. Chem. Soc. 1983, 105,
1065.
Interference was not expected from the major contaminant, iron
pentacarbonyl, since it is not expected to exhibit a microwave spectrum
because it has no permanent dipole moment. The microwave spectrum
of C2H4Fe(CO)4 was most readily measured from the second fraction
when the sample cell was held at 0 oC. Evidence for Fe(CO)5
contamination in the early part of the experiment was revealed by a
slowly growing signal intensity as the Fe(CO)5 distilled off. Often, after
several hours of scanning the signal intensity would grow to 3-10
times the initial intensity, this would last for about an hour, and then
the signal would drop to below half of the initial intensity. This behavior
is attributed to the gradual fractionation of the more volatile Fe(CO)5
out of the sample chamber, leaving mostly C2H4Fe(CO)4. Since this
molecule is highly volatile even at 0 °C, the purified sample rapidly
evaporates leaving only a small portion mixed with decomposition
products. The primary decomposition products, ethylene and Fe3(CO)12,
are also volatile and can thus diminish signal intensity also. The samples
were stored at -196 °C, and appeared to be thermally stable up to
-20 °C for extended periods of time, but low volatility of C2H4Fe-
(CO)4 at this temperature required the scanning to be done above -10
(19) Kukolich, S. G.; Sickafoose, S. M.; Breckenridge, S. M. J. Am.
Chem. Soc. 1996, 118, 205.
(20) Drouin, B. J.; Kukolich, S. G. J. Am. Chem. Soc. 1998, 120, 6774.
(21) Lavaty, T. G.; Wikrent, P.; Drouin, B. J.; Kukolich S. G. J. Chem.
Phys. In press.
(22) Roehrig, M. A.; Wikrent, P.; Huber, S. R.; Wigley D. E.; Kukolich,
S. G. J. Mol. Spectrosc. 1992, 154, 355.
(23) Kukolich, S. G.; Breckenridge-Estes, S. M.; Sickafoose, S. M. Inorg.
Chem. 1997, 36, 4916.
(24) Drouin, B. J.; Cassak, P.; Kukolich, S. G. Inorg. Chem. 1996, 36,
2868.
(25) Drouin, B. J.; Lavaty, T. G.; Cassak, P. A.; Kukolich, S. G. J. Chem.
Phys. 1997, 107, 6541.
(26) (a) Becke, A. D. Phys. ReV. 1988, A 38, 3098. (b) ACS Symp. Ser.
1989, 394, 165. (c) Int. J. Quantum Chem. Symp. 1989, No. 23, 599. (d)
Perdew, J. P.; Wang, Y. Phys. ReV. 1992, B 45, 13, 244.
(27) Gaussian 94, Revision B.3; Frisch, M. J.; Trucks, G. W.; Schlegel,
H. B.; Gill, P. M. W.; Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.;
Keith, T.; Petersson, G. A.; Montgomery, A. J.; Raghavachari, K.;
Al-Laham, M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.; Peng,
C. Y.; Ayala, P. Y.; Chen, W.; Wong, M. W.; Andres, J. L.; Replogle, E.
S.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Binkley, J. S.; Defrees, D. J.;
Baker, J.; Stewart, J. P.; Head-Gordon, M.; Gonzalez, C.; and Pople, J. A.
Gaussian, Inc.; Pittsburgh, PA, 1995.
(28) Bumgarner, R. E.; Kukolich, S. G. J. Chem. Phys. 1987, 86, 1083.

Molecular Structure of Tetracarbonylethyleneiron
J. Am. Chem. Soc., Vol. 121, No. 16, 1999 4025
Table 1. Measured Transition Frequencies for C2H4Fe(CO)4 and
C2D4Fe(CO)4
with CDCl3 solvent. The tube was sealed and stored at -78 °C prior
a
to running the experiment. The proton decoupling was turned off so
that the 13C-1H spin-spin coupling constant could be determined. The
quantum nos.
C2H4Fe(CO)4
measured
3911.7638(18)
5358.7201(37) -0.0030 5125.5720(12)
5566.1820(09) 0.0009 5341.3113(24)
5672.7252(13) -0.0012 5411.5415(66)
C2D4Fe(CO)4
measured
13
center of the
C triplet was found at 34.9 ppm, in good agreement
et
JKpKo J′Kp′Ko′
dev
dev
with the proton decoupled value29 of 35.3 ppm. The 1:2:1 triplet pattern
had a spin-spin splitting of 1JCH ) 161(1) Hz. The 1H NMR spectrum
showed a singlet at 2.48 ppm with 13C sidebands at 155(5) Hz
separation. Peaks due to the reaction solvent pentane were also observed
at the same signal-to-noise ratio as the product compound.
1
2
10 220
02 312
0.0007
0.0009
0.0098
0.0161
211 321
2
3
2
2
3
3
12 322
22 414
20 330
21 331
12 404
21 413
5804.2278(26)
5986.7355(15)
5995.3931(15)
0.0049
Calculations. The DFT calculations were performed on an SGI
Origin 2000 computer using the algorithms of Gaussian94 revision
E.2.27 The density functional methods of Becke26a-c and Perdew and
Wang26d (BPW91) were applied for a geometry optimization of the
structure of tetracarbonylethyleneiron. All coordinates were varied in
both internal coordinate space and Cartesian coordinate space, each
leading to an identical minimum that has all real positive vibrational
frequencies, subject to C2V symmetry constraint. The ground-state
structure is in excellent agreement with previously published15 DFT
results using the ADF30-32 package; unfortunately, this study did not
report the complete hydrogen atom coordinates.
Data Analysis: Microwave Spectra. The measured rotational
spectrum for C2H4Fe(CO)4 was successfully analyzed as an asymmetric
top with only a “c”-type dipole moment. The measured lines are listed
in Table 1 along with the quantum number assignments. The deviations
of the “best fit” calculated transition frequencies from measured values
are also included to demonstrate the ability of the model to reproduce
the spectrum within experimental error. The lack of measured transitions
for “a”- and “b”-type dipole selection rules provides strong evidence
for the C2V symmetry in this complex. After the “c”-type transitions
were fit, the hypothetical “a”- and “b”-type transitions could be
calculated with greater accuracy. Many of these predicted transitions
are very close to regions containing “c” type transitions which had
been extensively scanned in search of 13C and 18O transitions. The
existence of only three sets of 13C lines and two sets of 18O lines further
verifies that the three pairs of carbon atoms and two pairs of oxygen
atoms lie in symmetry-related positions within the molecule. We
therefore believe that only “c”-dipole transitions are possible for the
main isotopomer.
The 13C and 18O transitions were measured in natural abundance
(2.2% and 0.4%, respectively) using the normal isotope sample, and
these are given in Tables 2 and 3. The perdeuterated sample was greater
than 80% D4, so a variety of transitions were readily measured, and
the transition frequencies are shown in Table 1 alongside the corre-
sponding main isotope transitions.
The structure of this complex is shown in Figure 1. The a inertial
axis is nearly in line with the axial carbonyl groups with the remaining
ligands located in the equatorial plane. The c inertial axis is coincident
with the C2 rotation axis of the molecule and lies in the equatorial
plane. The carbon atoms of the ethylene ligand lie in the bc (equatorial)
plane. From here on the carbon atoms in the ethylene ligand will be
referred to as Cet, and the carbon and oxygen atoms in the CO ligands
will be referred to as Ceq, Oeq, Cax, and Oax, denoting the equatorial
and axial carbonyls, respectively. Note that in Figure 1 the principal
axes (abc) correspond directly to the symmetry axes (xyz) for the C2V
point group.
0.0017 5697.4804(19) -0.0038
0.0014 5701.1674(13) -0.0055
6200.1532(38) -0.0001
6261.0940(42) -0.0022
441 533
440 532
7001.0737(76)
0.0049
7032.7745(32) -0.0029
7143.1090(23) -0.0009 6814.7490(59)
3
03 413
312 422
0.0077
7239.1109(27)
0.0013 6955.1697(40) -0.0087
4
4
32 524
23 515
7305.2943(67) -0.0047
7319.6564(29)
7415.8640(13)
7562.5340(56) -0.0034
7641.5320(19)
7652.6716(11) -0.0053
7680.3698(15) -0.0004 7331.0333(53)
7700.6236(49)
0.0001
313 423
31 523
0.0001 7078.4238(23) -0.0065
4
3
4
3
21 431
13 505
22 432
0.0004 7313.7021(38) -0.0038
0.0056
5
4
32 606
22 514
0.0015
7926.9064(43) -0.0003
3
3
30 440
31 441
8052.9087(24)
8054.0011(21)
8295.6270(49)
0.0019 7654.2221(29) -0.0051
0.0020 7654.5402(08) -0.0055
0.0031
550 642
303 431
533 625
8523.5437(15) -0.0016
8907.1688(14) 0.0014
4
4
4
4
4
4
4
13 523
04 514
14 524
22 532
23 533
31 541
32 542
8942.9717(24) -0.0012 8583.0771(30) -0.0082
8965.8631(35)
9181.4488(19)
0.0001
0.0025
9282.0181(89) -0.0015 8920.5697(28) -0.0012
9379.6367(13) -0.0001 8967.5165(90) -0.0023
9720.3487(51)
0.0009 9276.8988(55)
0.0019
0.0006
9727.5750(31) -0.0019 9279.0550(26)
5
23 615
9368.7648(46) -0.0010
10115.6700(29) -0.0006 9609.3683(17) 0.0072
4
4
5
5
5
5
5
5
5
5
5
5
40 550
41 551
14 624
05 615
23 633
15 625
24 634
32 642
33 643
41 651
42 652
50 660
10115.7876(28)
10685.0477(33)
0.0047
0.0026 10230.8835(35) -0.0038
10816.2091(19) -0.0013 10263.6681(70) -0.0001
10919.7230(15) -0.0007 10518.7883(35)
0.0087
0.0015
0.0016
10967.1489(14)
11097.2505(42)
0.0010
0.0007 10613.0840(69)
11378.5117(54) -0.0054
11405.0018(22) 0.0024 10904.5500(73)
11786.3176(34) -0.0013 11233.0694(69) -0.0060
11787.2907(27) -0.0023 11233.2777(51) 0.0038
12177.9478(21) 0.0014 11564.3312(35) -0.0010
a All measured frequencies and deviations (dev ) measured -
calculated) from calculated values are given in MHz, listed uncertainties
in the measured lines are 1σ.
°C. No signal was observed below -20 °C, the strongest signals were
observed at 0 °C, but no measurements were attempted at higher
temperatures due to thermal instability of the compound.
Measured rotational transitions were used in a nonlinear least-squares
fitting routine to determine the adjustable parameters of an asymmetric
top Hamiltonian with centrifugal distortion constants. The molecule
exhibits a typical nearly rigid-rotor spectrum with all distortion constants
well below 1 kHz. For the main isotopomer, many ∆Ko ) 2 lines were
measured along with one ∆Ko ) 4 line and one ∆Ko ) -2 line. These
transitions helped to independently define the five quartic distortion
The perdeuterated sample was prepared with perdeuterated ethylene-
d4, obtained from Cambridge Isotopes. One liter of C2D4 was combined
with 200 mL of C2H4 prior to condensation into the reaction vessel.
The intensity of the main isotope signal was a small fraction of the
C2D4Fe(CO)4 signal intensity. All sample manipulation for the deu-
terated sample was identical with that for the normal isotopic species.
For measurements of the 18O lines, it was necessary to further purify
the sample prior to transfer into the spectrometer sample cell. The
second fraction described above was slowly fractionated in a -30 °C
cold trap, and the remaining material was then used in the experiment.
This purified sample gave up to 200:1 signal-to-noise ratios for the
main isotope transitions. However, the increased volatility also caused
rapid loss of sample during the experiment.
(29) Lindner, E.; Jansen, R. M.; Schauss, E. Organometallic Syntheses;
Elsevier Publishing Co., Inc.: New York, 1988; Vol. 4, p 141.
(30) Baerends, E. J.; Ellis, D. E.; Ros, P. Chem. Phys. 1973, 2, 41.
(31) Ravenek, W. Algorithms and Applications on Vector and Parallel
Computers; te Rilel, H. J. J., Dekkar, Th. J., van de Vorst, H. A., Eds.;
Elsevier: Amsterdam, 1987.
(32) (a) Boerrigter, P. M.; te Velde, G.; Baerends, E. J. Int J. Quantum
Chem. 1988, 33, 87. (b) te Velde, G.; Baerends, E. J. J. Comput. Phys.
1992, 99, 84.
NMR spectra were recorded on a Unity 300 MHz NMR spectrometer
at 0 °C. The sample was vacuum distilled into the NMR tube along

4026 J. Am. Chem. Soc., Vol. 121, No. 16, 1999
Drouin and Kukolich
Table 2. Measured Transition Frequencies for the 13C Substituted Isotopomers Measured in Natural Abundancea
13
13
13
quantum nos.
C
C
C
et
eq
ax
JKpKo J′Kp′Ko′
measured
dev
measured
dev
measured
dev
2
02 312
5315.2613(34)
5526.5370(15)
5624.7603(56)
5934.2677(25)
5941.5321(34)
7080.3191(64)
7188.3418(71)
7353.4343(41)
7584.0571(24)
7617.0483(06)
7980.0365(23)
7980.8854(15)
8876.8034(32)
8880.5862(18)
9102.5319(36)
9220.2730(15)
9304.8082(32)
9640.9731(20)
9646.6097(37)
10022.8825(82)
10022.9498(30)
-0.0020
-0.0024
-0.0022
0.0014
-0.0008
0.0008
5347.3062(59)
5542.0028(41)
5649.6556(59)
5952.0154(07)
5961.3871(24)
7131.7376(54)
7214.1885(13)
7391.0288(56)
7601.9490(53)
7643.5900(83)
8005.4959(59)
8006.7527(61)
8919.1061(35)
8954.8870(36)
9155.2956(14)
9237.9591(83)
9341.0192(31)
9668.5725(29)
9676.8247(32)
10055.3688(84)
10055.5018(65)
0.0036
0.0029
-0.0015
-0.0008
-0.0013
0.0014
5334.3700(14)
5551.5284(15)
5658.2690(09)
5982.1754(22)
5990.5016(53)
7108.2804(53)
7213.5489(61)
7391.6478(24)
7628.3061(25)
7665.8476(44)
8048.1561(22)
8049.1690(15)
8905.6748(30)
8920.5855(65)
9147.4008(19)
9259.8779(13)
9355.0506(73)
9706.7121(52)
9713.4431(65)
10110.8257(37)
10110.9077(46)
-0.0004
-0.0040
-0.0024
-0.0015
0.0017
-0.0007
0.0007
-0.0077
0.0004
0.0024
0.0014
0.0006
0.0014
0.0011
0.0041
0.0020
-0.0017
-0.0016
0.0020
0.0089
-0.0099
211 321
2
2
12 322
20 330
221 331
3
3
03 413
12 422
0.0035
-0.0017
0.0018
313 423
-0.0014
-0.0014
-0.0004
-0.0020
0.0009
0.0000
0.0004
-0.0007
0.0004
0.0028
-0.0025
0.0006
0.0079
-0.0053
3
3
3
3
21 431
22 432
30 440
31 441
-0.0037
-0.0005
-0.0023
0.0006
-0.0049
-0.0005
0.0007
0.0051
0.0015
-0.0013
-0.0040
0.0044
413 523
04 514
4
414 524
4
4
22 532
23 533
431 541
432 542
440 550
441 551
0.0008
a All measured frequencies and deviations (dev ) measured - calculated) from calculated values are given in MHz, listed uncertainties are 1σ.
Table 3. Measured Transition Frequencies and Deviations from
Calculated Values for 18O Substituted Isotopomers Measured in
Natural Abundancea
available, thus all of the distortion constants were held fixed at the
values obtained for the main isotopomer, since these values are not
expected to change much upon isotopic substitution. For the 13C
isotopomers the parameters DJ and DJK could each be treated as variable
parameters to obtain values in agreement with the main isotopomer
values. To reduce correlation errors in the rotational constants for these
species all of the distortion constants were fixed at main isotopomer
values for the final fits. For the perdeuterated species only DJK was
fixed at the value for the main isotopomer. The value for DJK was not
well determined in the fit, presumably due to the lack of ∆Ko ) 2
transitions and the larger uncertainties in the line centers caused by
unresolved deuterium quadrupole coupling. The Hamiltonian parameters
obtained from the least-squares fitting analyses and the standard
deviations of the fits are given in Table 4.
18
18
quantum nos.
O
eq
O
ax
JKpKo J′Kp′Ko′
measured
dev
measured
dev
211 321
220 330
221 331
303 413
312 422
321 431
322 432
330 440
331 441
413 523
404 514
5442.7813(62)
5803.4305(32) -0.0045 5963.2893(02) -0.0039
5817.1347(66) 0.0003 5970.3676(62) 0.0023
7094.2726(28) -0.0019 7108.7282(37) -0.0010
7117.0123(32)
7432.5862(24)
7489.8100(52) -0.0066 7605.8571(50) -0.0001
7803.0955(38)
0.0071 5491.3845(67)
0.0043
0.0017
0.0100 7573.3379(64) -0.0022
0.0013 8028.5520(26)
0.0000
0.0015
7805.5376(29) -0.0021 8029.3019(21)
Molecular Structure. Data were collected on 7 isotopomers and
therefore 21 rotational constants were used for the structural analyses.
Two structures can be determined from the spectral data: Kraitchman
(rs) and least squares (ro). The Kraitchman structure, which gives only
8832.8068(34) -0.0128
8920.1084(74)
0.0092
a All measured frequencies and deviations (dev ) measured -
calculated) from calculated values are given in MHz, listed uncertainties
are 1σ.
coordinates of isotopically substituted atoms, is incomplete without 54
-
Fe substitution data. However, the iron c coordinate for the rs structure
can be obtained through the first moment equation for the c axis. The
least-squares structure also allows complete determination of all atomic
coordinates. The ro, or least-squares, structure was determined by two
independent methods denoted methods “a” and method “b”. In method
a, the internal coordinates of the molecule, defined by bond lengths,
interbond angles, and Cartesian coordinates, are varied as parameters
in a least-squares fit to the rotational constants. Method b employs
the program written by Schwendeman,33 in which the Cartesian
coordinates are varied in a fit to the experimental moments of inertia.
Both ro structures are in excellent agreement with each other, with the
Kraitchman structure, and with the theoretical structure obtained using
DFT.
Kraitchman Analysis. The substitution coordinates directly deter-
mined from Kraitchman’s equations are given in Table 5. For the
quadruple substitution of the four equivalent hydrogen atoms it was
necessary to develop substitution coordinate equations from first
principles. In general, the differences between the inertial moments
for the main and (multiply) substituted species can be expressed34 as
Figure 1. The molecular structure and principal axis system for
tetracarbonylethyleneiron. The axial carbonyl groups are at an angle
of 2.6° relative to the “a” principal axis.
(33) Schwendeman, R. H. Critical EValuation of Chemical and Physical
Structural Information; Lide, D. R.; Paul, M. A., Eds.; National Academy
of Sciences: Washington, DC, 1974; p 74. Thanks to Kurt Hillig for
continued updates of the code.
(34) Gordy, W.; Cook, R. L. MicrowaVe Molecular Spectra; Interscience
Publishers: John Wiley & Sons: New York, 1970; Chapter 13.
constants. The parameter δJ could only be determined using all the
available data, and the small value of 6(5) Hz indicates that it makes
a negligibly small contribution to “regular” ∆Ko ) 0 transitions. The
value of δJ was fixed at zero during all of the isotopically substituted
spectral fits. For the 18O isotopomers, only a limited data set was

Molecular Structure of Tetracarbonylethyleneiron
J. Am. Chem. Soc., Vol. 121, No. 16, 1999 4027
Table 4. Rotational Constants Derived from the Measured Transition Frequenciesa
13
13
13
18
18
parameter
C2H4Fe(CO)4
C
C
C
O
ax
O
eq
D4
et
ax
eq
A, MHz
B, MHz
C, MHz
DJ, KHz
DJK, KHz
DK, KHz
δJ, Hz
1031.1079(4)
859.8056(4)
808.5672(4)
0.094(7)
0.10(3)
-0.17(2)
6(5)
1021.2044(4)
853.7213(4)
807.8437(11)
0.094(F)*
0.10(F)*
-0.17(F)*
6(F)*
1031.0759(5)
855.0194(5)
804.3557(17)
0.094(F)*
0.10(F)*
-0.17(F)*
6(F)*
1024.6218(4)
858.6021(4)
805.6373(12)
0.094(F)*
0.10(F)*
-0.17(F)*
6(F)*
1030.9689(5)
835.1063(6)
786.7757(19)
0.094(F)*
0.10(F)*
-0.17(F)*
6(F)*
996.9317(15)
854.5308(14)
791.8870(60)
0.094(F)*
0.10(F)*
-0.17(F)*
6(F)*
977.4884(15)
826.3590(16)
796.3847(22)
0.12(2)
0.10(F)*
-0.17(4)
6(F)*
δK, KHz
σfit, KHz
0.37(5)
2.6
0.37(F)*
2.8
0.37(F)*
4.1
0.37(F)*
2.8
0.37(F)*
2.8
0.37(F)*
7.7
0.4(2)
6.7
a The Hamiltonian models an asymmetric top with five independent distortion constants. All errors in parameters are 2σ. The value of δJ was
held fixed at zero for all substituted isotopomer fits. Asterisks with (F) indicate fixed at values obtained in fitting the normal isotopomer.
Table 5. Results of the Cartesian Coordinate Least-Squares Structure Fit to the Experimental Moments of Inertia (Standard deviation of fit
0.004 amu‚Å2, or 28 kHz) and the Kraitchman Analysis Showing Absolute Values of Substituted Atom Coordinates in the Parent (FeH2(CO)4)
Principal Axis System (see Figure 1). (Uncertainties near 0.003 Å)a
atom
ao (Å)
bo (Å)
co (Å)
|as| (Å)
|bs| (Å)
|cs| (Å)
rcom (Å)
Fe
0.000a
(1.814(2)
(2.954(1)
0.000b
0.000b
0.000b
0.073(16)
0.148(32)
0.205(14)
-0.942(4)
-1.576(2)
2.067(2)
0.081
0.132
0.199
0.937
1.574
2.064
2.283
Cax
Oax
Ceq
Oeq
Cet
H
1.8107
2.9535
0.052ic
0.043ic
0.059ic
0.896
0.043ic
0.048ic
1.493
2.448
0.706
1.254
1.816
2.960
1.762
2.902
2.182
2.755
0.000b
(1.496(3)
(2.448(1)
(0.709(3)
(1.254(1)
0.000b
0.000b
(0.897(2)
2.2835(6)
a Listed errors are 2σ. b These values were fixed at zero during the fitting. c These coordinates are nonzero (and/or imaginary) due to vibrational
averaging.
The carbon and oxygen atom positions obtained using the Kraitch-
man equations for single substitution in an asymmetric top, the program
of Schwendeman,33 and one developed in this laboratory produced
identical results. It should be noted that the Kraitchman coordinates
for atoms in the ac and bc planes that should be zero are in fact nonzero
or imaginary. This effect is similar to inertial defects in planar molecules
in which zero-point vibrational movement of the atoms out of the plane
results in substitution coordinates that represent the root-mean-square
average vibrational displacements. In this molecule, the Kraitchman
analysis provides root-mean-square positions for the b coordinates of
atoms in the ac plane and the a coordinates of atoms in the bc plane.
The iron atom is the only nonsubstituted atom in the molecule; therefore
all of the substituted atom cs coordinates can be used to solve the c
axis first moment equation for the iron c coordinate. For the internal
coordinates listed in Table 6, it was necessary to assume all of the
small nonzero “out of plane” values to be zero. The errors reported for
internal substitution coordinates are derived from Costain’s35 model
by using the Schwendeman program.
shown in eq 1:
N
N
(
∆mbi)2
(
∆mci)2
∑
∑
i
i
∆Iaa ) ∆m(bs2 + cs2) -
-
(1)
M + ∆m
M + ∆m
where ∆m is the difference between the total mass of the parent (M)
and the substituted isotopomer. The subscript s refers to the coordinates
of the substituted atom(s). The second and third terms account for the
shift of the center of mass from the parent molecule. The analogous
terms for ∆Ibb and ∆Icc can be obtained by cyclic permutation of the
abc coordinates in eq 1. For single substitution, the second and third
terms are typically combined with their corresponding substitution
coordinate to give a single term involving the substitution reduced mass,
µ ) ∆m‚M/(∆m + M). In some cases, where multiple substitution
retains the initial symmetry of the parent molecule it is possible that
these terms will drop out of the difference moment equations entirely.
The C2V symmetry of tetracarbonylethyleneiron gives the molecule both
ac and bc mirror planes. The ac mirror plane forces the second term in
∆Iaa to zero, thus the first term is all that contributes to the bH coordinate.
Similarly, symmetry about the bc plane cancels the corresponding term
in ∆Ibb, and together these mirror planes cancel both terms in ∆Icc.
The third term in ∆Iaa and ∆Ibb remains, and when combined with ∆mcs2
can be shown to reduce to µ. Therefore, the difference moments for
quadruple deuterium substitution in this molecule can be expressed as:
Least-Squares Analyses. Two independent least-squares fitting
methods were employed in an effort to minimize errors resulting from
correlation between variable parameters and from inherent weighting
of experimental data.
In method a the internal coordinates of the molecule were defined
using the Fe-Cax, Fe-Ceq, Cax-Oax, Ceq-Oeq, Fe-Cet, and Cet-Cet
bond lengths and the Cax-Fe-Cax, Ceq-Fe-Ceq, Fe-Cax-Oax,
Fe-Ceq-Oeq interbond angles; the hydrogen Cartesian coordinates
were varied as three independent parameters. Within the C2V symmetry
of the molecule this scheme allowed variation of 13 independent
molecular parameters in a fit to the 21 measured rotational constants.
The standard deviation of the fit was 11 kHz, indicating that the final
structure reproduces the experimental rotational constants to within 1
order of magnitude of their experimental error (∼1 kHz). Final values
of the variable parameters and other interesting structural parameters
are given in Table 6.
In method b, the Cartesian coordinates of symmetrically related
atoms were tied together to give 13 independent parameters; the location
of the iron atom along the c inertial axis was a 14th dependent parameter
in the fit. Each set of carbon atoms required two variables (b and c for
Cet and Ceq, a and c for Cax), similarly the two sets of oxygen atoms
each required two variables, finally the H atoms required 3 variables
∆Iaa ) ∆mbH2 + µcH2, ∆Ibb ) ∆maH2 + µcH2, ∆Icc
)
∆m(aH2 + bH2); ∆m ) 4(mD - mH) (2)
Since the quadruple substitution does not rotate the principal axes there
are no off-diagonal terms in the inertial tensor for the substituted
molecule. This allows for rapid solution of the three difference moments
in eq 1 to give the hydrogen substitution coordinates:
∆Ibb + ∆Icc - ∆I
2∆m
∆Icc + ∆Iaa - ∆Ibb
2∆m
|a| )
aa, |b| )
,
x
x
∆Iaa + ∆Ibb - ∆Icc
2µ
|c| )
(3)
x
(35) Costain, C. C. J. Chem. Phys. 1958, 29, 864.

4028 J. Am. Chem. Soc., Vol. 121, No. 16, 1999
Drouin and Kukolich
Table 6. Structural Parameters Derived from the Experimental
The DFT study presented here is a remarkable example of
theory supporting experiment. During the initial phase of
scanning for microwave spectra, two predictions were ob-
tained: one given by the published GED structure and one from
the present DFT(BPW91) calculation. The two structures had
very similar (A + B)/2 values, but very different (B - C)/2
values. This encouraged us to look for the ∆J ) +1 Ko ) 0, 1
transitions that were predicted to be very close to each other.
The distance between the two Ko transitions could give a better
value for (B - C)/2 and the overall location and spacing between
consecutive ∆J transitions (R branches) could give a better value
for (A + B)/2. The J′ r J ) 3 r 2, Ko ) 0, 1 transitions were
predicted 30 MHz apart for the two structures, and were found
between the two predictions, closer to the DFT predicion. The
measured splitting between the two Ko components (9 MHz) is
also much closer to the DFT prediction (5 MHz) than the GED
prediction (1 MHz). After accurate determination of the main
isotopomer parameters, an approximate structure fit was done,
allowing only the Fe-Cet and Cet-Cet distances to vary, and
holding all other parameters fixed at values from the DFT
structure. This resulted in a surprisingly accurate structure that
allowed prediction of the 13C lines to within 2 MHz for most
transitions of all the 13C-substituted isotopomers. The new data
set containing 13C coordinates was quite insensitive to the
location of the hydrogen atoms, and thus a reliable prediction
for the quadruply substituted D4 complex was a concern. A study
of the dependence of the least-squares fit standard deviations
on the C-C-H and Fe-C-C-H angles showed a flat surface
with a shallow minimum very close to the DFT prediction. The
structure corresponding to this minimum was used to predict
the D4 substituted spectrum. This proved to be an accurate
analysis since the D4-substituted transitions were found within
3 MHz of the predicted frequencies. This indicates that the DFT
structure obtained using the functionals described by Becke,
Perdew, and Wang26 is a very accurate representation of the
gas-phase structure of this compound. In comparison to the ADF
DFT31,32 (using the B3LYP functional) work,16 the reported
parameters are in good agreement for the Cax and Cet bond
lengths and angles. The calculated Fe-Ceq and Fe-Cax bond
lengths are nearly equal (agree within experimental error). This
result is in better agreement with the experiment than the present
DFT study. The only reported parameter for hydrogen atoms is
the dihedral angle describing the bending of the C-H bond out
of the former ethylene plane. The B3LYP method gives an Fe-
C-C-H dihedral angle of 111.5° which is significantly larger
than the experimental value. This would indicate that the
BPW91 methods may better represent small structural changes
upon olefin complexation. It is also interesting to note that the
least well-determined parameter in the DFT study is the Fe-
Cet distance, which also has the largest uncertainty in the
experimental work. This is an indication of the diffuse metal-
olefin bonding. The electronic structure calculations give the
most difficulty with the metal-olefin bonding, and the rotation
of the molecule evidently causes more centrifugal distortion
along this coordinate. Despite these difficulties, the Fe-Cet bond
length was determined to 1% accuracy.
Rotational Constantsa
(a) Bond Lengths Determined in Structural Fits (a) and (b) and
with Kraitchman’s Equations and c(H-C), the Separation between
the Plane of the Ethylene H Atoms and the C-C Bond
bond
ra (Å)
rb (Å)
rs (Å)
r(C-H)
1.071(4)
1.072(3)
1.073(2)
r(Fe-Cax)
r(Fe-Ceq)
r(Fe-Cet)
r(Cet-Cet)
r(Cax-Oax)
r(Ceq-Oeq)
r(Fe-|)
1.815(2)
1.805(9)
2.118(14)
1.420(6)
1.142(2)
1.145(2)
1.996(14)
0.217(2)
1.815(2)
1.807(9)
2.116(14)
1.418(8)
1.142(3)
1.145(4)
1.994(17)
0.217(2)
1.811(1)
1.808(10)
2.105(17)
1.412(3)
1.145(2)
1.147(2)
1.983(17)
0.219(3)
c(H-C)
(b) Interbond and Dihedral Angles Determined in
Structural Fits (a) and (b) and with Kraitchman’s Equations
angle
Θa (°)
Θb (°)
Θs (°)
C-C-H
120.5(6)
113.7(6)
103.6(9)
174.8(25)
111.8(9)
39.2(5)
120.6(3)
113.6(5)
103.6(9)
175.3(19)
111.6(8)
39.2(4)
120.7(1)
113.3(2)
103.7(10)
176.8(13)
111.4(9)
39.2(3)
H-C-H
Fe-C-C-H
Cax -Fe-Cax
Ceq-Fe-Ceq
Cet-Fe-Cet
Fe-Cax-Oax
Fe-Ceq-Oeq
180.0(28)
179.5(7)
179.5(30)
179.4(6)
178.2(12)
179.4(5)
a Values determined in the internal/Cartesian coordinate least-squares
fit to the rotational constants are denoted by a subscript “a”. Values
determined by a global Cartesian fit to the first and second moments
of inertia are denoted by a subscript “b”. Values obtained from
Kraitchman’s equations are denoted by the subscript “s”. Errors for
the least-squares fits “a” and “b” are 2σ, errors in Kraitchman
coordinates are Costain35 errors.
(a, b, and c coordinates). The 14 Cartesian coordinate parameters were
varied to fit the 21 measured moments of inertia and forced to satisfy
the first moment equation for c coordinates. The fit has a standard
deviation of 0.0036 amu‚Å2, which gives a relative uncertainty on the
same order of magnitude (1 × 10-5) as the standard deviation obtained
in method a. The coordinates from method b and their respective
uncertainties are listed in Table 5 alongside the Kraitchman values.
The experimental microwave bond lengths and angles determined
from the two methods are given in Table 6. All three methods are in
excellent agreement with each other, indicating that differences in
correlation and weighting do not cause significant problems in the
structural fits. The close agreement of the Kraitchman parameters with
the least-squares structures indicates that the equilibrium values of the
structural parameters are not very different from the reported ro and rs
values. Unlike the similar compound H2Fe(CO)4, the assumption of
rigid bonding upon isotopic substitution seems to be accurate even in
the hydrogenic coordinates of the ethylene ligand.
Discussion
This paper presents the first detailed study of the structural
changes of ethylene upon complexation to iron. The displace-
ment of the plane of the ethylene hydrogen atoms away from
the ethylene C-C bond provides details of changes in bonding
and hybridization of the ethylene moiety. This displacement was
assumed to be small, but not accurately determined in previous
studies. The interbond angles for the Fe(CO)4 fragment of this
complex are considerably different than the corresponding angles
for H2Fe(CO)4. Bond lengths were found in excellent agreement
with GED data whereas bond angles were determined with better
accuracy and precision in the present work. Theoretical treatment
using DFT provided a very reliable structure for the search and
subsequent analysis of the microwave spectra.
There are some discrepancies in the literature11,13,15-17
concerning the iron-carbon bond lengths for the carbonyl
ligands in tetracarbonylethyleneiron. It is well-known that the
“piano-stool” compounds with η4 bound olefins have remarkably
different Fe-C bond lengths and Fe-C-O interbond bond
angles in the iron tricarbonyl base. In H2Fe(CO)4 either
intermolecular exchange or intramolecular pseudorotation causes

Molecular Structure of Tetracarbonylethyleneiron
J. Am. Chem. Soc., Vol. 121, No. 16, 1999 4029
Table 7. Structural Parameters Obtained in This and Various Other Studies of Tetracarbonylethyleneiron and of the Closely Related Species
Triphenylphosphinetricarbonylethyleneiron
structural
parameter
C2H4Fe(CO)4
GED
C2H4XFe(CO)4
X-rayb
C2H4Fe(CO)4
MWc
C2H4Fe(CO)4
DFT
r(Fe-Cet) (Å)
2.117(30)
1.796(35)
1.836(35)
1.46(6)
1.146(10)
1.146(10)
1.08a
2.099(7)b
1.759(6)
1.761(6)b
1.398(8)
1.164(4)
1.171(12)b
1.00(24)b
38.9(2)
176.4(2)
112.2(3)
178.8(6)
175.2(9)b
118(8)
2.117(14)
1.815(2)
1.806(9)
1.419(7)
1.142(3)
1.145(3)
1.072(4)
39.2(5)
175.1(22)
111.7(9)
179.8(29)
179.5(7)
120.6(5)
103.6(9)
2.1188
1.8006
1.7865
1.4186
1.1546
1.1576
1.0905
39.12
176.64
111.66
179.64
180.27
120.0
r(Fe-Cax) (Å)
r(Fe-Ceq) (Å)
r(Cet-Cet) (Å)
r(Cax-Oax) (Å)
r(Ceq-Oeq) (Å)
r(C-H) (Å)
(Cet-Fe-Cet) (°)
(X-Fe-Cax) (°)
(Ceq-Fe-Ceq) (°)
(Fe-Cax-Oax) (°)
(Fe-Ceq-Oeq) (°)
(C-C-H) (°)
40.3
180a
105.2(30)
180a
180a
116a
(Fe-C-C-H) (°)
90a
98
103.5
a These values were assumed during the fitting process. b X ) P(C6H5)3. The compound is not symmetric, thus averaged values of symmetry
related terms are listed; the number in parentheses represents the range if larger than 4σ. c This column lists average values of methods “a” and “b”
described in the text.
the CO groups to appear equivalent on the NMR36 time scale.
A 13C NMR study29 of tetracarbonylethyleneiron also shows a
single peak at 212 ppm for both sets of carbonyl groups,
indicating either an accidental chemical shift equivalence or that
an exchange process is occurring on the longer time scale of
the NMR experiment. The present experimental data, which
include two sets of 13C carbonyl isotopomers, indicates that this
complex is fairly rigid (no tunneling splittings were directly
observed) and has two distinctly different locations for the two
carbonyl groups, as was observed for H2Fe(CO)4.19 Interestingly,
the iron carbon bond lengths were found to be equal to within
experimental error. In the GED work an attempt was made to
separate these bond lengths, but the final values reported in that
study still have overlapping error bars. Furthermore, the
theoretical studies (including this one) have all indicated small
differences between these bond lengths. It should be noted that
the theoretical predictions for the differences in bond lengths
are in the opposite direction to the GED assignment, but do
agree with the trend indicated by the present data. The present
Figure 2. Structural parameters and internal coordinates for tetracar-
bonylethyleneiron showing the iron-ethylene fragment of the complex.
In the top view, ethylene parameters are shown. In the bottom view,
experimental data suggest errors for these bond lengths to be
0.009 and 0.002 Å for r(Fe-Cax) and r(Fe-Ceq), respectively,
and suggest that the two bond lengths are equal to within 0.01
Å. The DFT value for the difference in bond lengths is (0.014
Å) right at the error limit of the experiment, and thus there is
not yet a definitive answer as to whether these bond lengths
are equivalent to less than one hundredth of an angstrom. The
Fe-C-O angles were very nearly linear for both groups. Both
carbonyl groups were found to have collinear CO and Fe-C
bonds to within 1°. The Fe(CO)4 fragment has a structure much
closer to the Fe(CO)5 molecule than to H2Fe(CO)4.19 The
trigonal bipyramid structure of C2H4Fe(CO)4 deviated from the
the iron-carbon bond length (2.117 Å) and the dihedral angle (103.6°),
which would be 90° if the ethylene moiety remained planar, are shown.
the C-C bond length. There are only very minor changes in
the C-H bond length and C-C-H bond angle (see Figure 2).
The C-C bond is lengthened 0.07(1) to 1.419(7) Å, compared
to 1.339(1) Å for free gaseous ethylene.37 The previously
reported GED value11 (see Table 7) was much closer to a single
C-C bond length, whereas this study produces a more accurate
value, intermediate between typical single and double C-C bond
lengths. The C-C-H angles of C2H4 and C2H4Fe(CO)4 are
nearly identical at 121.1(3)o and 120.6(5)o. The C-H bond
lengths of 1.085(2) and 1.072(4) Å for C2H4 and C2H4Fe(CO)4,
respectively, demonstrate only a small contraction of 0.012(4)
Å. To the authors’ knowledge, only one other experimental
study13 of an ethylene-iron complex has determined the
dihedral angle that describes the out-of-plane bend of the C-H
bond. This X-ray diffraction study of triphenylphosphinetricar-
bonylethyleneiron determined this angle to be near 8°. The
uncertainties in hydrogen coordinates in X-ray diffraction studies
are often quite high, as can be seen by the erratic C-H bond
lengths reported13 (see Table 7). An unperturbed ethylene
structure would have the C-H bond “in the plane” perpendicular
to the plane containing the Fe atom and the Cet-Cet bond. This
D3h structure of Fe(CO)5 primarily in the contraction of the Ceq-
Fe-Ceq interbond angle from 120° to 111.7(9)°. The Cax-Fe-
Cax bond angle deviates from linearity by 5(2)°. The C2V
distortions are more pronounced in H2Fe(CO)4 where the Ceq-
Fe-Ceq and Cax-Fe-Cax angles are 99(4)° and 154(4)°,
respectively.19 The Fe-Cax,eq bond lengths in all three iron
carbonyl compounds are nearly identical at 1.81 Å.
The results of the experimental and theoretical analyses
indicate that the effects of complexation on ethylene are
primarily the movement of the plane of the ethylene hydrogen
atoms away from the carbon-carbon bond and elongation of
(36) Vancea, L.; Graham, W. A. G. J. Organomet. Chem. 1977, 134,
219.
(37) Duncan, J. L.; Wright, I. J.; Van Lerberghe, D. J. Mol. Spectrosc.
1972, 42, 463.

4030 J. Am. Chem. Soc., Vol. 121, No. 16, 1999
Drouin and Kukolich
structure would have an Fe-C-C-H dihedral angle of 90°.
The value determined in the present study, 103.6(9)o, is nearly
halfway between what would be expected (120°) for full sp3
hybridization of the Cet atoms and unperturbed ethylene. An
intermediate hybridization model is consistent with theory
describing cyclopropane,38 in the respect that the bonds char-
acterized by small angles between ring atoms retain some π
character. There is a crystal structure of the isoelectronic
compound, tetracarbonylethyleneosmium,39 in which the eth-
ylene appears to be more tightly bound to the metal center as
indicated by an even larger C-C bond length of 1.488 Å (which
would be much closer to sp3-type C-C bond lengths) and a
larger “out-of-plane” bending of the hydrogens to Os-C-
C-H ) 112°.
scheme places the bonding orbitals outside the triangle formed
by Fe and the two Cet’s; however, the sp2 hybridization scheme
places these partially within this triangle, thus allowing better
overlap with the metal orbitals. An intermediate hybridization
scheme allows favorable overlap between ligand and metal
orbitals and gives the correct displacement of the H atoms. The
H atom displacement may be an important part of the mecha-
nism for catalytic isomerization of olefins. With the H atoms
displaced out of the former ethylene plane, adjacent, bound
olefin molecules can approach the ethylene more closely. With
symmetry restrictions42 on olefin 2+2 addition removed, the
olefin exchange process can proceed.
Summary
The complete, three-dimensional structure of tetracarbonyl-
ethyleneiron was obtained by fitting and Kraitchman analysis
of high-resolution microwave spectra for seven isotopomers.
This is the first study of an organometallic complex for which
a significant number of transitions have been obtained for
natural-abundance 18O isotopomers. The structure of the ethylene
is modified through interaction with the iron tetracarbonyl
moiety. The C-C bond length for the complex is elongated by
0.07(1) Å relative to free ethylene and the plane of the ethylene
H atoms is displaced 0.217(2) Å away from the C-C bond.
The C-H bond length is reduced by 0.012(4) Å in the complex.
These effects are in excellent agreement with the results of the
DFT calculations. There are important correlations between
structures and reactivity. The changes in the ethylene geometry
observed here would correlate with reduced steric hinderence
and increased reactivity with another complexed olefin. The
present results should increase confidence in using DFT
calculations for larger organometallic complexes.
The values for the tetracarbonylethyleneiron proton chemical
13
shift (2.48 ppm) and
C chemical shift (35 ppm) both fall
et
between values for ethane (1H 0.8 ppm, 13C 5.3 ppm) and
ethylene (1H 5.3 ppm, 13C 122 ppm). This is a further indication
that the hybridization of ethylene, when complexed to iron, falls
between sp3 and sp2 hybridization.
Although arguments have been made40 for correlations of the
13C-1H spin-spin coupling constants (1JCH) with percent s
character in the hybridized carbon orbitals at the carbon atom,
1
the measured value of JCH ) 161(1) Hz for the present iron
complex indicates that the electronic structure (i.e., the percent
s character) near the Cet atoms is very similar to that of free
ethylene. Ethylene in CDCl3 solvent41 has a coupling constant
1
1
of JCH ) 156.3 Hz. The values for JCH for sp3 hybridization
are smaller (125 Hz). The coupling constant results do not give
a good indication of the hybridization at carbon, which could
be caused by interactions with the iron orbitals.
A qualitative inspection of valence orbitals indicates that the
interaction of ethylene π and π* orbitals with d metal orbitals
stabilizes the sp2-type configuration. An sp3 hybridization
Acknowledgment. We are extremely grateful to the National
Science Foundation (Grant CHE-9634130) for support of this
research. We thank Caroline Kriss for obtaining NMR spectra.
Jennifer Dannemiller, Paul A. Cassak, and Burzin Engineer
provided help with some of the spectra.
(38) Coulson, C. A.; Goodwin, T. H. J. Chem. Soc. 1971, 93, 1904.
(39) Bender, B. R.; Norton, J. R.; Miller, M. M.; Anderson, O. P.; Rappe,
A. K. Organometallics 1992, 11, 3427.
(40) Muller, N.; Pritchard, D. E. J. Chem. Phys. 1959, 31, 1471.
(41) Lynden-Bell, R. M.; Sheppard, N. Proc. R. Soc. London A 1962,
269, 385.
JA983937X
(42) Hoffman, R.; Woodward, R. B. J. Am. Chem. Soc. 1965, 87, 2046.