Measurement of olefin bond rotation barriers in a series of vinylaniline complexes of the cyclopentadienyliron(II) dicarbonyl cation and their correlation to metal-olefin bond asymmetry

Article abstract of DOI:10.1021/om040092r

H NMR techniques (selective inversion, total line shape analysis, and T₁measurement) have been used to measure the olefin bond rotation rates in a series of compounds of the general formula CpFe(CO) ₂[η²-CH₁C(H)NH(p-C₈H ₄X)], where X = OMe (4), H (6), CN (9), N0₂ (10), Cl (11), COOMe (12). Complexes 4, 6, 11, and 12 enabled the use of all three methods to yield data sets that spanned a wide temperature range and allowed for reliable determination of the thermodynamic activation parameters for olefin bond rotation. When these three methods were combined with a bootstrap statistical analysis of the Eyring plots, it was possible to resolve the small differences in the ΔH? values between three of the four complexes. A clear trend is established between increasing electron donation from the aniline ligand and the facilitation of olefin bond rotation. These results are placed in the context of previous work, providing additional experimental evidence that electronic control from the para position of the aniline influences the overlap of the nitrogen lone pair with the adjacent olefinic carbon. This overlap controls the asymmetry in the metal-olefin bond by displacing the Fp⁺ moiety away from the nitrogen-bearing carbon, which is reflected in a drop in the ΔH? value for olefin bond rotation.

Full text of DOI:10.1021/om040092r

5440
Organometallics 2004, 23, 5440-5449
Measurement of Olefin Bond Rotation Barriers in a
Series of Vinylaniline Complexes of the
Cyclopentadienyliron(II) Dicarbonyl Cation and Their
Correlation to Metal-Olefin Bond Asymmetry
Stephen A. Matchett,* Guirong Zhang,† and David Frattarelli‡
Department of Chemistry, Grand Valley State University, Allendale, Michigan 49401
Received July 1, 2004
1H NMR techniques (selective inversion, total line shape analysis, and T1F measurement)
have been used to measure the olefin bond rotation rates in a series of compounds of the
general formula CpFe(CO)2[η2-CH2C(H)NH(p-C6H4X)], where X ) OMe (4), H (6), CN (9),
NO2 (10), Cl (11), COOMe (12). Complexes 4, 6, 11, and 12 enabled the use of all three
methods to yield data sets that spanned a wide temperature range and allowed for reliable
determination of the thermodynamic activation parameters for olefin bond rotation. When
these three methods were combined with a bootstrap statistical analysis of the Eyring plots,
it was possible to resolve the small differences in the ∆Hq values between three of the four
complexes. A clear trend is established between increasing electron donation from the aniline
ligand and the facilitation of olefin bond rotation. These results are placed in the context of
previous work, providing additional experimental evidence that electronic control from the
para position of the aniline influences the overlap of the nitrogen lone pair with the adjacent
olefinic carbon. This overlap controls the asymmetry in the metal-olefin bond by displacing
the Fp+ moiety away from the nitrogen-bearing carbon, which is reflected in a drop in the
∆Hq value for olefin bond rotation.
Understanding the factors that control olefin coordi-
nation in metal-olefin complexes is important to ex-
ploiting the diverse chemistry of these complexes.
Bonding in these complexes was originally described by
Chatt, Dewar, and Duncanson1 as a combination of σ
donation from the olefin to the metal and a π back-bond
from the metal to the π* orbital on the olefin. This model
is generally appropriate for an interaction in which the
metal is symmetrically placed along the olefin face.
Metal-olefin bonding, however, ranges from a fully
symmetric placement of the metal to strongly asym-
metric bonding wherein the metal approaches a purely
σ interaction with only one carbon of the double bond.
Examples from both ends of this continuum were
reported by Rosenblum and co-workers for a series of
cationic cyclopentadienyliron(II) dicarbonyl (hereafter
referred to as Fp) olefin complexes in the early 1980s.2
Crystal structures reported for complexes 1-3 show the
effect of increasing π-donation from the â-carbon sub-
stituent on the symmetry of the metal-olefin bond.2
Rosenblum quantified this asymmetry using the pa-
rameter δ, where δ ) (distance from Fe to Câ) -
(distance from Fe to CR).
As the π-donor ability of the â substituent increased
(Y) H to OMe to NMe2), the Fp fragment was displaced
toward the R-carbon (the value of δ increases). When
the donor group possesses a lone pair, the complexes
represent points along the resonance continuum be-
tween structures I and II.
Supported by the calculations of Hoffman and Eisen-
stein,3 Rosenblum proposed that the position along this
continuum was determined by the ability of a lone pair
on the donor to overlap with the developing LUMO that
forms on the â-carbon as the Fp fragment is displaced
along the olefin face.
* To whom correspondence should be addressed. E-mail: matchets@
gvsu.edu.
Caulton and co-workers4 recently updated and con-
firmed the structural characterization of Rosenblum’s
† Visiting Scientist from the Department of Chemistry, East China
Normal University, Shanghai, People’s Republic of China 200062.
‡ Undergraduate researcher at Grand Valley State University.
(1) (a) Dewar, M. J. S. Bull. Soc. Chim. Fr. 1951, 18, C71; Annu.
Rep. Prog. Chem. 1951, 48, 112. (b) Chatt, J.; Duncanson, L. A. J.
Chem. Soc. 1953, 2239.
(3) (a) Eisenstein, O.; Hoffmann, R. J. Am. Chem. Soc. 1980, 102,
6148. (b) Eisenstein, O.; Hoffmann, R. J. Am. Chem. Soc. 1981, 103,
4308.
(4) Watson, L. A.; Franzman, B.; Bollinger, J. C.; Caulton, K. G.
New J. Chem. 2003, 27, 1769.
(2) Chang, T. C. T.; Foxman, B. M.; Rosenblum, M.; Stockman, C.
J. Am. Chem. Soc. 1981, 103, 7361.
10.1021/om040092r CCC: $27.50 © 2004 American Chemical Society
Publication on Web 10/05/2004

Olefin Bond Rotation Barriers
Organometallics, Vol. 23, No. 23, 2004 5441
heteroatom-substituted Fp+ olefin complexes5 and ana-
lyzed the metal to olefin bonding asymmetry in the
series of Fp+ olefin complexes [CpFe(CO)2-n(PH3)n(H2Cd
CHX)] (where X ) F, OEt, NMe2 and n ) 0-2) using
density functional theory (DFT) calculations. These
calculations supported the conclusion that increased
π-donor ability from the heteroatom favors resonance
structure II, which dominated in the case of X ) NMe2.
They also found that as the σ acidity of the metal
fragment decreased with increasing PH3 substitution,
both the vinyl fluoride and the vinyl ether moved toward
more symmetric metal to olefin bonding. In the case of
the vinylamine, however, even the increased electron
density from two PH3 substituents on iron could not
significantly alter the extreme asymmetry in its metal-
olefin bonding6 caused by strong π donation from the
NMe2 group.
form IV, leading to a displacement of the Fp+ fragment
toward the R-carbon. Conversely, electron withdrawal
from the para position led to an increased contribution
from resonance form III.
We contend that substitution at the para position of
these vinylanilines is directly influencing the overlap
of the nitrogen lone pair with the potential LUMO that
develops on the â-carbon as the metal moves across the
olefin face.
To strengthen this argument, we sought an additional
physical measurement that would assay the relative
importance of resonance structure IV in response to
electronic changes at the para position. We reasoned
that, as the contribution from IV increased, the barrier
to rotation about the carbon-carbon bond of the olefin
should decrease:
Recent work in our laboratories has focused on
controlling the extent of this asymmetry in Fp-olefin
bonding, specifically with nitrogen-substituted olefins.
We recently reported7 synthesizing a series of para-
substituted η2-vinylaniline complexes of the Fp cation
(complexes 4-10). Our work demonstrated a correlation
Restricted bond rotation has been described in the
literature for many years, with the classic case being
amide bond rotation.9 Because restricted bond rotations
often produce exchanges between chemical environ-
ments on the NMR time scale, rotation barriers in these
systems have been measured using various NMR line-
broadening techniques.10 The preferred method has
often been total line shape analysis. This method is most
accurate in the vicinity of the coalescence temperature,
where the exchange process dominates the line width.9b
This limits the temperature range of measurable ex-
change rates, which can lead to inaccuracies in the
calculation of activation parameters from regression of
the Arrhenius and Eyring plots.11
Spin saturation transfer (selective inversion) experi-
ments, originally developed by Forsen and Hoffman,12
are capable of measuring exchange rates at tempera-
tures well below coalescence. When the 1H NMR signal
of one of the exchange sites is inverted, the system
returns to equilibrium magnetization by the normal
spin-lattice relaxation, T1, and by chemical exchange.
Curve-fitting routines allow one to separate the ex-
change rates from the T1 data, expanding the measur-
able rate range to include values approximately an order
of magnitude smaller than rates obtained through line
shape analysis.
between the 13C shifts of the olefin carbons and the σpara
constants for the para substituents on the coordinated
vinylaniline. This correlation, and the accompanying
X-ray crystal structures of 5 and 8, gave direct struc-
tural evidence8 for the ability to electronically control
the extent of asymmetry in the metal-olefin bonding
and hence its position along the continuum represented
by resonance structures III and IV. As the electron-
donating properties of the para substituent increased,
there was an increased contribution from resonance
In the region of fast exchange, Deverell and co-
workers13 demonstrated that measurement of T1F (spin-
lattice relaxation rate in the rotating frame) can be used
(5) They repeated the X-ray structure of [CpFe(CO)2(H2CdCHNMe2)]-
PF6 and solved the structure for [CpFe(CO)2(H2CdCHOEt)]PF6 in an
attempt to overcome the crystal disorder problems encountered with
Rosenblum’s original structure of [CpFe(CO)2(H2CdCHOMe)]PF6. Both
structures were improved and supported the original conclusions.
(6) Upon comparison of the DFT calculations to the actual X-ray
crystal structures for [CpFe(CO)2(H2CdCHNMe2)]PF6 and [CpFe(CO)2-
(H2CdCHOEt)]PF6, the DFT calculations tended to overestimate the
length of the Fe-CR bond in each case. The authors attribute the
discrepancy to the inability of DFT calculations to accurately model
weak interactions of this type. Although the bond lengths do not match
exactly, the calculations support the trend described.
(7) Matchett, S. A.; Schmiege-Boyle, B. R.; Cooper, J.; Frattarelli,
D.; Olson, K.; Roberts, J.; Thommen, J.; Tigelaar, D.; Winkler, F.
Organometallics 2003, 22, 5047.
(8) δ values from these crystal structures were 0.501 and 0.354 Å,
respectively.
(9) Stewart, W. E.; Siddall, T. H., III. Chem. Rev. 1970, 70(5), 517.
(10) (a) Johnson, C. S. Adv. Magn. Reson. 1965, 1, 33. (b) Sandstrom,
J. Dynamic NMR Spectroscopy; Academic Press: London, 1982. (c)
Dynamic Nuclear Magnetic Resonance Spectroscopy; Jackman, L. M.,
Cotton, F. A., Eds.; Academic: New York, 1975. (d) Kaplan, J. I.;
Fraenkel, G. NMR of Chemically Exchanging Systems; Academic
Press: New York, 1980.
(11) Bain, A. D.; Duns, G. J.; Ternieden, S.; Ma, J.; Werstiuk, N. H.
J. Phys. Chem. 1994, 98, 7458.
(12) (a) Foresen, S.; Hoffman, R. A. J. Chem. Phys. 1963, 39, 2892.
(b) Hoffman, R. E.; Forsen, S. Prog. Nucl. Magn. Reson. Spectrosc. 1966,
1, 15.
(13) Deverell, C.; Morgan, R. E.; Strange, J. H. Mol. Phys. 1970,
18, 553.

5442 Organometallics, Vol. 23, No. 23, 2004
Matchett et al.
to extract exchange rates in the range of 103-106 s-1
.
While total line shape analysis on coupled systems
has become more sophisticated,19 we simplified our
treatment by decoupling the â-proton. This simplified
the spectrum to yield two peaks (one for each R-proton)
at slower rotation rates, collapsing to a singlet in the
limit of fast exchange.20 The comparison of the coupled
and decoupled spectra in the fast-exchange region is
illustrated for complex 11 in Figure 2.
Other methods, including the CPMG14 sequence and
more recently offset saturation methods,15 have also
been used in the region of fast exchange.
While each of these techniques allows collection of
data in a limited range of rotation rates, the combina-
tion of multiple NMR techniques has increasingly been
demonstrated16 to extend the temperature range over
which meaningful data can be collected. Combining data
sets from selective inversion experiments in the limit
of slow exchange, total line shape analysis in the region
of coalescence, with offset-saturation techniques in the
limit of fast exchange into one Eyring plot, Bain et al.
have produced reliable values17 of the activation pa-
rameters for several different bond rotations.
In this paper we report a systematic determination
of the activation parameters for rotation about the
olefin bond in a series of para-substituted η2-vinyl-
aniline complexes of the Fp+ cation using three dis-
tinct NMR techniques. To effectively use all three
techniques, we also describe the synthesis of two more
derivatives in our series: complex 11 (p-Cl derivative)
and complex 12 (p-COOMe derivative). We have em-
ployed selective inversion,18 total line shape analysis,
and the T1F experiment to explore the correlation
between the activation barriers to olefin bond rota-
tion in these complexes and the electronic properties of
the para substituent on the aniline ring. Such a cor-
relation would lend strength to the argument that
the para substituent controls the metal-olefin bond
symmetry by controlling the availability of the nitrogen
lone pair.
Dynamic Motions in These Molecules. It is im-
portant to establish that the activation parameters
measured in this experiment are the direct result of
rotation about the CR-Câ bond of the olefin and not
some other dynamic process within the molecule. Rota-
tion of the Fp+ moiety as either classical olefin propeller
style rotation or rotation about the Fe-CR bond in the
limit of resonance structure IV must also be consid-
ered.21 We have no doubt that a rotation of the Fp+
moiety does indeed occur. At room temperature, while
most derivatives in the series show an equilibration of
the two CO ligands to a single CO peak in the 13C NMR,
derivatives 9, 10, and 12 show no CO peak. We feel that
this is because these latter derivatives are at or near
coalescence and the broadening of these peaks does not
allow for resolution of these notoriously weak peaks
from the baseline.7 The singlet observed for the other
derivatives reflects their lower barrier to CO equilibra-
tion, placing them in the limit of fast exchange at room
temperature. To confirm this, a series of 13C spectra
(acetone-d6) of complex 4 (p-OMe) was taken at tem-
peratures ranging from 5 to -85 °C. Figure 3 shows that
as the temperature decreased the singlet seen at room
temperature broadened and eventually separated into
two peaks in the limit of slow exchange.
Fp+ rotation (either classic propeller rotation or
rotation about the Fe-CR bond in IV) cannot equilibrate
either the CO peaks or the geminal protons in the
absence of some other molecular motion. It should be
noted that classic propeller-style rotation of a sym-
metrically bound olefin is not a reasonable model in this
series, where the ground-state preference for resonance
form IV has been previously established in solution by
13C NMR22 as well in the solid state by the X-ray crystal
Results and Conclusions
The case for restricted olefin bond rotation in these
complexes can be seen in their variable-temperature 1H
NMR spectra. The olefin protons in the series of deriva-
tives 4-10 all show an A2X style spectrum at or near
room temperature. At these temperatures the two
R-protons are equivalent in their coupling to the â-pro-
ton, yielding a low-frequency doublet and a high-
frequency doublet of triplets (the expected triplet is
further split by the adjacent N-H). Figure 1 shows (for
11, the p-Cl derivative) that, as the temperature is
lowered, the doublet for the R-protons coalesces and
eventually separates into an ABX pattern in the limit
of slow exchange.
2
structures of 2, 3, 5, and 8.7 To equilibrate the CO
peaks, the molecular motions must generate a plane of
symmetry that bisects the CO-Fe-CO angle. In order
for the geminal protons to collapse from an ABX system
to a clean doublet at room temperature, the cis, trans
(19) (a) Stephenson, D. S.; Binsch, G. J. Magn. Reson. 1978, 32,
145-152. (b) Binsch, G.; Kessler, H. Angew. Chem., Int. Ed. Engl. 1980,
19, 411. (c) See also ref 9b, Chapter 3.
(20) In the limit of slow exchange one can begin to freeze out the
geminal coupling between the two R-protons. The exchange rates in
this temperature range are, however, too slow to be accurately treated
by line shape analysis. The geminal coupling constant was included
during line shape simulation for accuracy, but the effects are not
evident in the temperature range where these simulations were
performed.
(21) The authors also considered the possibility of rotation about
the imine bond present in systems dominated by resonance structure
IV. This was readily dismissed, since such a rotation cannot in itself
equilibrate the cis, trans relationships between the R- and â-protons
of the coordinated olefin. Second, as the electron-releasing properties
of the para substituent are increased, the overlap of the nitrogen lone
pair with the â-carbon (to form the imine structure) should increase,
increasing the barrier to rotation. The opposite trend is observed,
supporting the contention that the data describe rotation of the CR-
Câ bond.
(14) (a) Allerhand, A.; Gutowsky, H. S. J. Chem. Phys. 1964, 41,
2115. (b) Reeves, L. W. In Dynamic Nuclear Magnetic Resonance
Spectroscopy; Jackman, L. M., Cotton, F. A., Eds.; Academic: New
York, 1975, pp 83-130.
(15) (a) Bain, A. D.; Ho, W. P. Y.; Martin, J. S. J. Magn. Reson. 1981,
43, 328. (b) Bain, A. D.; Duns, G. J. J. Magn. Reson. 1993, A109, 56.
(16) (a) Bain, A. D.; Duns, G. J.; Rathgeb, F.; Banderkloet, J. J. Phys.
Chem. 1995, 99, 17338. (b) Bain, A. D.; Hazendonk, P. J. Phys. Chem.
A 1997, 101, 7182. (c) See also ref 10.
(17) Errors in ∆Hq of less than 1 kJ/mol and in ∆Sq of less than 6
J/(mol K) have been reported.15 While ∆Sq values are often considered
suspect, ref 15b presents a case for placing increased faith in ∆Sq values
obtained by this method.
(18) (a) The authors used a combination of an initial nonselective
inversion to determine T1 values, followed by a selective inversion with
curve fitting to determine the exchange rates. This method has been
proposed to provide better quantitative kinetic data than the use of
the 2-D NMR technique of EXSY.18b (b) Bain, A. D.; Fletcher, D. A.
Mol. Phys. 1998, 95, 1091.
(22) Previous work on these complexes show the difference between
the 13C chemical shifts for CR and Câ can be as large as 155 ppm
(complex 4).7

Olefin Bond Rotation Barriers
Organometallics, Vol. 23, No. 23, 2004 5443
1
Figure 1. Variable-temperature H NMR spectra of the R-protons of complex 11 (p-Cl derivative), demonstrating the
collapse of the ABX splitting to an A2B pattern in response to olefin bond rotation.
relationships between the R- and â-protons must be
equilibrated during the exchange.23 No rotation of the
Fp+ moiety can be envisioned that would equilibrate
either of these positions in the absence of some other
dynamic process.
Although rotation of the CR-Câ bond will directly
equilibrate the geminal protons, this rotation will not
exchange the CO ligands. While the coalescence tem-
perature estimated for this CO exchange (Figure 3) is
similar to that reported in Table 1 for CR-Câ bond
rotation (-25.0 °C), rotation of this carbon-carbon bond
alone will not generate the symmetry necessary to
equilibrate the two CO peaks. The previously reported
X-ray crystal structures of complexes 5 and 87 both show
the Fp+ moiety to be displaced toward the R-carbon in
the solid state. In both cases, one CO is directed under
the olefin face toward the â-carbon, while the other is
directed toward the R-carbon. If only the CR-Câ bond
is allowed to rotate, the spatial relationship between
each CO and the rest of the molecule will not change.
We therefore believe that generation of the symmetry
(23) While the collapse to an A2B system could also occur if a plane
of symmetry developed between the two R-protons, only rotation of
the C-C bond could generate such a plane.

5444 Organometallics, Vol. 23, No. 23, 2004
Matchett et al.
Measurement of the Activation Parameters. Ide-
ally, rate measurements would be made on each of the
complexes (4-10) from our original 13C correlation
study.7 To make these measurements in all three
exchange regimes, one must balance solubility, coales-
cence temperature (vs solvent freezing/boiling point),
and overlap with other peaks in the spectrum. Complex
8 (p-COMe) gave only a limited line shape analysis data
set due to overlap from the acetyl methyl peak with the
R-carbon protons at some temperatures.24 Complex 9 (p-
CN) and complex 10 (p-NO2), as expected, had the
highest barriers to rotation (more contribution from
resonance form III). While these two derivatives gave
an excellent data set25 for the combination of total line
shape analysis with selective inversion, the boiling point
of the solvent (acetone-d6)26 prevented reaching the
temperature where rates determined by the T1F experi-
ment are most accurate. To extend the data, two more
derivatives of the original series, 11 (p-Cl) and 12 (p-
COOMe), were prepared by reaction of the ethyl vinyl
ether compound [CpFe(CO)2(CH2CHOEt)+PF6- 27 with
]
a slight molar excess of the appropriate para-substituted
aniline as the attacking nucleophile (as seen above for
4-10).
Figure 2. Comparison of the 1H-coupled and 1H-decoupled
spectra of the R-protons of complex 11 (p-Cl derivative) in
the region of fast exchange. The spectra on the left are 1H
coupled, while those on the right have been decoupled by
irradiation of the â-proton.
Measurement of the coalescence temperature and the
approximation of the rotation rates (kc ) (π|δν|)/x2)28
for derivatives 4, 6, and 9-12 clearly show the influence
of the para substituent. This approximation has been
demonstrated29 to give reasonable values for the ex-
change rates at the coalescence temperature. Table 1
lists these complexes (in order of increasing Hammett
σpara value) with their coalescence temperature (Tc),
exchange rates at Tc, and ∆Gq values at Tc. As the
electron-donating properties of the para substituent
decreased (σpara increased), the Tc value and, hence, the
barrier to rotation increased. This reflects a decreasing
overlap between the nitrogen lone pair and the â-carbon
and thus an increasing contribution from resonance
structure III. Using this same technique, Rosenblum
found2 the coalescence temperature for 3 (NMe2 bonded
directly to the olefin) to be -60 °C,30 well below that of
4 (the most electron-rich aniline complex in the current
q
Table 1. Rotation Rates and the ∆G Values for
Olefin Bond Rotation in Complexes 4, 6, and 9-12,
1
As Estimated from the H NMR Coalescence
Temperature
k at Tc
(s-1
∆Gq from Tc
(kJ/mol)
complex
Tc (°C)
)
4 (X ) OMe), σ ) -0.27
6 (X ) H), σ ) 0.00
11 (X ) Cl), σ ) 0.23
12 (X ) COOMe), σ ) 0.45
9 (X ) CN), σ ) 0.66
10 (X ) NO2), σ ) 0.78
-25.0
-14.0
-7.5
2.5
185.7
168.3
158.8
131.7
123.3
106.7
49.6
52.1
53.6
56.1
57.9
59.3
10.0
15.0
plane needed for CO exchange requires both rotation
of the Fe-CR bond (rotation 1 in the drawing below) and
rotation of the CR-Câ bond (rotation 2):
series). Rosenblum’s value was determined in CD
2Cl2
instead of acetone-d6, which makes the comparison more
qualitative.31 Despite this, it is evident that changing
the olefin substituent from a NMe2 group through the
series of increasingly electron-poor anilines raises the
(24) This overlap was extensive for all temperatures for complex 5
(p-Me), and no data were collected for this derivative.
(25) The Eyring plot for 9 had an R2 value of 0.998 for 11 data points.
The Eyring plot for 10 had an R2 value of 0.999 for 15 data points.
(26) Solubility limits the solvent choices to acetone-d6, methylene-
d2 chloride, or nitromethane-d3. Methylene-d2 chloride has similar
boiling point limitations, and nitromethane-d3 freezes above the
temperature where these complexes exhibit rates suitable for measure-
ment by selective inversion.
Since both rotations are facilitated by increased contri-
bution from resonance form IV, both are influenced by
the electronic properties of the para substituent on the
aniline.
While both rotations are necessary to explain the CO
exchange, the exchange of the geminal protons only
requires the rotation of the CR-Câ bond. Since Fp+
rotation is not required to explain the dynamic behavior
of the olefin protons, we contend that the data in Tables
1 and 2 directly reflect the influence of the aniline’s para
substituent on the barrier to CR-Câ bond rotation.
(27) Cutler, A.; Raghu, S.; Rosenblum, M. J. Organomet. Chem.
1974, 77, 381
(28) (a) Pople, J. A.; Schneider, W. G.; Bernstein, H. J. High-
Resolution Nuclear Magnetic Resonance; McGraw-Hill: New York,
1959; p 223. (b) δν is the separation of the peaks (in Hz) in the region
of slow exchange, or the peak width at half-height for the peak at
coalescence. The latter method was used here, since our simulations
showed a slight temperature dependence for δν.
(29) Kost, D.; Carlson, E. H.; Raban, M. J. Chem. Soc., Chem.
Commun. 1971, 656.
(30) From this he estimated ∆Gq at coalescence for rotation of the
carbon-carbon double bond to be 43.9 kJ/mol.
(31) A. D. Bain has reported a solvent effect on the internal rotation
barriers in furfural.15b

Olefin Bond Rotation Barriers
Organometallics, Vol. 23, No. 23, 2004 5445
Figure 3. Variable-temperature 13C NMR spectra for the metal carbonyl region of complex 4 (p-OMe derivative).
barrier to olefin bond rotation by decreasing the asym-
metry in the metal-olefin bonding (a fact previously
demonstrated by comparing X-ray crystal structures of
3, 5, and 8).7
Although the Tc values described a clear trend, we
wished to provide more fundamental activation param-
eters for this series of complexes. The four derivatives
4 (p-OMe), 6 (p-H), 11 (p-Cl), and 12 (p-COOMe) enabled
rate measurements to be made using three independent
NMR techniques: total line shape analysis, selective
inversion, and measurement of T1F. Figure 4 gives the
Eyring plot for these four derivatives, while Figure 5
gives the Arrhenius plot for the same compounds. Table
2 lists the activation parameters (determined from
Figures 4 and 5) for each of the derivatives measured,
along with the methods used. Incomplete entries reflect
an inability to collect T1F data for complexes 9 and 10.
A preliminary report of some of these data has been
recently published in the People’s Republic of China.32
Comparison of the Data by Method. Table 2
demonstrates that using only total line shape analysis
gave the lowest values for the parameters of the bond
rotation. Despite excellent linear agreement of the data
(32) Zhang, G. R.; Matchett, S. A.; Dai, L. Y.; Lu, J. X. J. East China
Normal Univ. 2003, 2, 46 (publication in Chinese).

5446 Organometallics, Vol. 23, No. 23, 2004
Matchett et al.
Table 2. Summary of the Activation Parameters for Olefin Bond Rotation in Complexes 4, 6, and 9-12 as a
1
Function of the Various H NMR Methods Useda
total line shape anal, (TLS)
∆Sq, ∆Gq
Ea,
selective inversion (SI) and TLS combined
∆Hq, ∆Sq, ∆Gq
Ea,
T1F, SI, and TLS
∆Sq,
Ea,
complex
(Hammett σ
param)
∆Hq,
,
,
∆Hq,
∆Gq
,
298
298
kJ/mol
298
kJ/mol
kJ/mol J/(mol K) kJ/mol
kJ/mol J/(mol K) kJ/mol
kJ/mol J/(mol K) kJ/mol
kJ/mol
4, X ) OMe
(-0.27)
6, X ) H
(σ ) 0.0)
11, X ) Cl
(0.23)
40.0 -38.8 42.0
51.5
39.6 -39.9 41.6
51.4
43.9 -21.7 45.9
50.4
((0.5) ((2.0) ((0.5)
37.5 -57.5 39.5
((0.9) ((3.5) ((0.9)
39.6 -53.0 41.8
((0.7) ((2.7) ((0.7)
-37.2 48.1
((5.4) ((1.4)
-21.5 54.3
((1.6) ((0.4)
-18.5 55.7
((10.1) ((2.8)
((1.1)b
54.6
((1.9)
55.4
((1.5)
57.0
((3.0)
58.4
((0.9)
58.9
((5.8)
((1.1) ((4.8) ((1.1)
45.8 -24.1 47.8
((0.9) ((3.8) ((0.9)
45.5 -30.3 47.5
((0.9) ((3.6) ((0.9)
54.9 -3.5 57.0
((0.7) ((2.7) ((0.7)
54.3 -12.8 56.4
((0.5) ((1.8) ((0.5)
61.6 10.9 63.8
((0.8) ((3.1) ((0.8)
((2.5)
52.9
((2.0)
54.5
((2.0)
55.9
((1.5)
58.1
((1.0)
58.4
((0.1)
((1.0) ((4.3) ((1.0)
46.6 -20.9 48.7
((0.6) ((2.3) ((0.6)
48.1 -19.4 50.3
((0.8) ((3.2) ((0.8)
55.1 -2.8 57.2
((0.5) ((2.1) ((0.5)
((2.3)
52.8
((1.3)
53.8
((1.8)
55.9
((1.1)
12, X ) COOMe 45.9
(0.45)
9, X ) CN
(0.66)
((1.4)
52.0
((0.4)
53.4
((2.8)
10, X ) NO2
(0.78)
a Errors are reported as (2 standard deviations. b Errors on the ∆Gq values are calculated from the formula σ(∆Gq) ≈ |σ(∆Hq) -
Tσ(∆Sq)|.17b
Use of the selective inversion technique extended the
data into the region of slow exchange (k ) 0.1-10 s-1
)
and generally acted to increase the values of ∆Hq.33 Data
sets with good linear agreement (R2 > 0.99) for a
combination of selective inversion with total line shape
analysis were obtained for derivatives 4, 6, and 9-12.
These data, however did not cleanly distinguish the ∆Hq
values between the various derivatives. Inclusion of the
T1F experiment (where possible) gave linear data sets
for derivatives 4, 6, 11, and 12 (R2 ) 0.994, 0.998, 0.996,
and 0.999, respectively) and changed the final values
slightly. Full data sets for these derivatives average 16
data points collected by 3 independent methods and
span a 102-104 fold range of rates.
Correlation of Bond Rotation Parameters with
the Electronic Influence of the Para Substituent.
1
Table 1 shows that the H NMR coalescence tempera-
ture for rotation about the olefin bond varies directly
with the Hammett σpara value for each of six derivatives.
While using only TLS or a combination of TLS and SI
to calculate the rotation barriers failed to distinguish
some of these six derivatives, extension of this general
trend to the ∆Hq value for bond rotation is reasonable.
Inclusion of the T1F data refines these measurements
and clearly establishes the trend for 4, 6, 11, and 12
(Table 2). As the electron-donating properties of the para
substituent increase, the ∆Hq value for olefin bond
rotation decreases. To establish the significance of the
differences between these values, it is important to take
a conservative look at the error in their measurement.
The error estimation in these quantities is often based
on the standard deviation from the linear fit of the
Eyring plot. These have the potential to be an optimistic
estimation of the error for such data sets. A normal
linear regression treats all points as having an equal
significance and uniform error, which may not be the
case. If the error for each value of k is known, a more
accurate estimation of the error in ∆Hq or ∆Sq could be
determined using a Monte Carlo resampling fit to the
data.34 The problem lies in the estimation of the error
in the individual determinations of k, particularly using
Figure 4. Eyring plot for complexes 4 (p-OMe), 6 (p-H),
11 (p-Cl), and 12 (p-COOMe). Data were collected by a
combination of selective inversion, total line shape analysis,
and spin lock (T1F) techniques.
Figure 5. Arrhenius plot for complexes 4 (p-OMe), 6 (p-
H), 11 (p-Cl), and 12 (p-COOMe). Data were collected by a
combination of selective inversion, total line shape analysis,
and spin lock (T1F) techniques.
(33) In the case of the p-OMe derivative, the numbers were almost
identical.
(34) By random sampling from within the error range of each data
point (k value) and continuous replotting of the data, a large set of
regressions could be produced from the same data set. This would
produce a range in the slopes and intercepts to the Eyring plot which
would give a more conservative estimate of the overall error in each
of the activation parameters.
in the region of intermediate exchange, each additional
NMR method extended the temperature range and
acted to refine the values determined for ∆Hq and ∆Sq.

Olefin Bond Rotation Barriers
Organometallics, Vol. 23, No. 23, 2004 5447
q
q
Table 3. Comparison of ∆H and ∆S Values
Obtained for Complexes 4, 6, 11, and 12 by Single
Linear Regression of the Eyring Plots versus
Multiple Regressions using the Bootstrap Methoda
linear regression
bootstrap method
∆Hq,
kJ/mol
∆Sq,
J/(mol K)
∆Hq,
kJ/mol
∆Sq,
J/(mol K)
complex
4 (X ) OMe) 43.9((1.0) -21.70((4.3) 43.8((1.4) -22.1((5.9)
6 (X ) H)
11 (X ) Cl)
12 (X )
46.6((0.6) -20.88((2.3) 47.2((0.8) -18.1((3.4)
48.1((0.8) -19.44((3.2) 48.1((1.0) -19.7((3.6)
55.1((0.5) -2.76((2.1)
54.2((0.8) -5.9((3.4)
COOMe)
a Errors are reported as (2 standard deviations.
the total line shape method. Typically, k values in this
range are determined by minimizing the ø2 value
between the experimental and calculated line shapes.
Thus, the σ value for each k value is not easily
determined. Estimations by visually determining error
in line shape methods have been reported with a value
of 10%.16 In the absence of σ values for each k, it is
possible to do a similar resampling analysis using the
bootstrap method.35 Here a random set of data points
from a given Eyring plot (equal to the number of data
points in that curve) are drawn from the original data
with the possibility that each point can be drawn more
than once. This gives a series of new data sets composed
of some fraction of the original points with some points
present as duplicates. Each set is then used to create
an Eyring plot and regressed to produce values of ∆Hq
and ∆Sq. These are then averaged to give values of ∆Hq
and ∆Sq with its corresponding σ value. The bootstrap
method has been confirmed36 as a meaningful method
for determining the errors in parameter estimations
where individual errors are not well defined.37 We ran
a series of 500 such resampled plots on each of our four
full data sets. The resulting activation parameters are
given in Table 3 along with the more conservative
estimation of their uncertainties. Figure 6 shows the
relationship between the Hammett σpara constants for
each para substituent and the ∆Hq value for these four
compounds determined by the bootstrap method. The
more conservative error bars in Figure 6 represent (2σ
based on applying a bootstrap analysis to the Eyring
plots. While errors inherent to the methodology prevent
one from cleanly distinguishing between the ∆Hq values
of 6 and 11, the overall trend remains clear.
Figure 6. Correlation between ∆Hq for olefin bond rotation
and the Hammett σpara parameter for complexes 4, 6, and
9-12.
duced. Combining three NMR methods has previously
been reported to allow determination of ∆Hq to within
1 kJ mol-1 and ∆Sq to within 10 J mol-1 K.38 To confirm
that the ∆Sq value for 12 is statistically different for
the values seen for the other complexes, we again return
to the bootstrap analysis of the error in these values.
Even using this conservative estimation of the error
(see Table 3 where reported errors are (2σ), the ∆Sq
value for complex 12 (X ) COOMe) differs by ap-
proximately 5σ from the next nearest ∆Sq value.39 Use
of our temperature extended data set in conjunction
with conservative error estimation makes a reasonable
case that the ∆Sq value for complex 12 is truly larger
than those seen in the other complexes and reflects a
difference in the transition state for rotation in that
compound. While one would expect the ∆Hq value to
change with the degree of π character in the double
bond, it seems at first surprising that the ∆Sq values
are not similar across the series. We contend that the
difference is in the ability of the p-COOMe group to
interact through resonance. In the crystal structure of
8 (p-COMe derivative), the molecule was remarkably
planar, more so than seen in the structure of 5 (p-Me
derivative).7 The latter showed approximately 15° greater
rotation in the dihedral angle between the imine
(resonance structure IV) and the plane of the aromatic
ring. The same dihedral angle in 8 was 178.3°. This
planarity would be a reasonable expectation for complex
12, on the basis of its similarity to 8. This extra
interaction would be disrupted upon rotation, allowing
greater rotational freedom and yielding a larger value
of ∆Sq. The small exchange data set for 5 does not allow
us to produce a ∆Sq value with the same level of
confidence, limiting our argument. Complexes 9 and 10
would also be capable of resonance but yielded only
selective inversion and total line shape data sets.
Although these smaller data sets are less reliable,
regressing the Eyring plot to find ∆Sq for 9 and 10 gave
comparatively larger values of ∆Sq (-12.8((1.8) and
An interesting feature of Table 2 is the ∆Sq values.
While complexes 4, 6, and 11 all regress to give ∆Sq
values of approximately -20 J/mol K, complex 12 is
clearly different at -2.76 J/mol K. This is seen in Figure
4, where the lines are regressed to the origin. Histori-
cally, ∆Sq values obtained by regressing the Eyring plot
have been viewed with skepticism, since small changes
in the slope can result in a large change in the intercept.
The use of combined NMR techniques has expanded the
temperature range of the data set and, in doing so,
increased confidence in the activation parameters pro-
(35) Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flannery, B.
P. Numerical Recipes in C: The Art of Scientific Computing, 2nd ed.;
Cambridge University Press: New York, 1992.
(36) Launer, R. L., Wilkinson, G. N., Eds. Robustness in Statistics;
Academic Press: New York, 1979.
(37) One caveat is offered. The Bootstrap method is best applied
when the data set is uniformly distributed across the range of points.
Since the three combined NMR methods each have a useable range,
there are some gaps in the distribution between these ranges.
(38) Bain, A. D.; Duns, G. J.; Ternieden, J. Ma, Werstiuk, N. H. J.
Phys. Chem. 1995, 99, 17338.
(39) Standard deviation on the difference between the values is
2
calculated as σdiff ) x(σ6 + σ122).

5448 Organometallics, Vol. 23, No. 23, 2004
Matchett et al.
retain solvent (particularly diethyl ether), making elemental
analysis more difficult. The new members 11 and 12 of the
previously characterized (including analyses)7 series of com-
pounds were prepared by the same method, and elemental
analyses were therefore not performed.
10.9((3.1) J/(mol K), respectively) than were observed
with 4 (-39.9((4.8), 6 (-24.1((3.8) J/(mol K)) and 11
(-30.3((3.6) J/(mol K)) when only selective inversion
and line shape data sets were used in the analysis.
Without the T1F data these numbers are less reliable,
but they do support a relationship between resonance
from the para position and a larger ∆Sq value.
Synthesis of Complexes 11 and 12. These compounds
represent two new members of the previously reported7 series
of Fp+ vinylaniline complexes and were prepared by identical
methods. These reactions yielded 11 and 12 in 58% and 70%
isolated yields,43 respectively.
In summary, this work demonstrates that there is a
direct correlation between the electronic properties of
the para group in these η2-vinylaniline complexes and
the NMR coalescence temperature for rotation around
the olefin bond. This supports the original hypothesis
that the para position is directly influencing the overlap
between the nitrogen lone pair and the â-carbon and,
hence, controls the position of the iron moiety along the
olefin face. Table 2 demonstrates that, despite linearity
in the region of coalescence, exclusive reliance on total
line shape analysis for calculating activation parameters
would lead to considerable underestimation in both ∆Hq
and ∆Sq. Combining three separate NMR techniques
over a wide range of temperatures, we have determined
the barriers to this olefin bond rotation for four com-
plexes in the series of Fp+ vinylaniline complexes. This
method has allowed us to measure small changes in the
∆Hq across this series. While the differences in ∆Hq
values were small enough to prevent unambiguous
distinction between two members of the series, the
correlation clearly seen in the coalescence temperatures
is reflected as a reasonable trend in the ∆Hq values.
Work in our laboratories has begun exploring the
competition between heteroatom π donors on both the
R- and â-carbons and how it influences both the position
of the iron moiety along the olefin face and the chem-
istry of these metal olefin complexes.
1
Characterization Data for the p-Cl Derivative 11. H
NMR (acetone-d6, 300 MHz): δ 2.67 (br d, 2H, J ) 8.0 Hz,
olefin CH2), 5.52 (s, 5H, Cp), 7.55 (s, 4H, aromatic), 8.8 (d of t,
1 H, JH-NH ) 12.6 Hz, J ) 8.8 Hz, olefin CHN), 10.6 (br d, 1H,
J ) 10.2 Hz, NH). 13C{1H} NMR (acetone-d6, 75.58 MHz): δ
11.0 (olefin CH2), 87.3 (CHCp), 120.8, 130.9, (CH aromatic),
132.5, 137.7 (C aromatic), 158.6 (olefin CHN), 213.6 (Fe-CO)
ppm. IR (KBr): 1995, 2049 cm-1
.
Characterization Data for the p-COOMe Derivative
12. 1H NMR (acetone-d6, 300 MHz): δ 2.78 (br d, 2H, J ) 8.0
Hz, olefin CH2), 3.89 (s, 3H, CH3), 5.62 (s, 5H, Cp), 7.63 (d,
2H, J ) 8.5 Hz, aromatic), 8.10, (d, 2H, J ) 8.5 Hz, aromatic),
8.9 (d of t, 1 H, JH-NH ) 14.3 Hz, J ) 9.4 Hz, olefin CHN),
10.5 (br d, 1H, J ) 14.0 Hz, NH). 13C{1H} NMR (acetone-d6,
75.58 MHz): δ 13.5 (olefin CH2), 52.4 (OCH3), 87.5 (CHCp),
118.6, 132.2, (CH aromatic), 128.6, 142.8 (C aromatic), 153.2
(olefin CHN), 166.3 (COOMe) ppm. Fe-CO peak(s) not located
due to dynamic exchange. IR (KBr): 1996, 2043 cm-1
.
1H NMR Exchange Experiments. Each derivative was
prepared as an approximately 0.10 M solution in acetone-d6
for analysis. Each of the three techniques was carried out on
the same sample. Sample tubes were stored in the refrigerator
between experiments, where they were stable (under N2) in
solution for weeks.
Selective Inversion. Prior to the selective inversion
experiments, the T1 values for both geminal protons were
determined using the standard nonselective inversion T1
experiment in the JEOL software package.44 This gives initial
T1 values for use in the nonlinear least-squares fitting of the
selective inversion data. The selective inversion experiments
were run using a relaxation-π/2-τ-π/2-variable delay-π/
2-acquisition pulse sequence with decoupling at the â-proton.
The pulse sequence for the decoupling modified version of the
standard selective inversion experiment was provided by the
software engineers at JEOL (see Acknowledgment). The τ
delay value was set at 1/(2∆νA-B), where ∆νA-B is the frequency
difference between the high-frequency and low-frequency
protons on the R-carbon of the molecule being measured. The
carrier frequency was set to invert the frequency of the
downfield proton. Values for the variable delays were typically
chosen to give a set of 10 values evenly spaced between 5 ms
and 10 s. The value of the exchange rate was extracted from
the best nonlinear least-squares fit of the data using CIFIT,45
a C version of the SIFIT program written by McClung and
Muhandiram.46 The combination of nonselective and selective
inversion experiments has been demonstrated to provide
reliable rate data in the limit of slow exchange.47
Experimental Section
General Procedures. All reactions were carried out under
N2 using standard Schlenk line techniques. Solvents were
distilled under a N2 atmosphere off of appropriate drying and/
or deoxygenating reagents (CH2Cl2, CaH2; Et2O, Na/benzophe-
none). Acetone-d6 was dried on 3 Å molecular sieves and
vacuum-transferred directly into the NMR tube. NMR spectra
1
were recorded on a JEOL Eclipse 300 at 300.52 MHz for H
and 75.57 MHz for 13C. All chemical shifts were referenced to
the residual protons in the deuterated solvents. The probe
temperature was calibrated using a plot of the chemical shift
differences between the OH and the CH3 resonances of
methanol as a function of temperature.40 Samples were allowed
to equilibrate for a minimum of 10 min at each temperature
setting, and the probe was then gradient-shimmed prior to
collection of each data point. Since all measurements were
performed while decoupling the proton on the â-carbon, an
1
initial H NMR spectrum was collected after shimming at a
Total Line Shape Analysis. The line shape analysis was
performed using MEXITER, an iterative version of Bain’s
Mexico program.48 The MEXITER program allows for inclusion
given temperature to accurately determine an irradiation offset
value for this proton. Values for the various σ parameters were
obtained from ref 41. Samples of the ethyl vinyl ether starting
material, CpFe(CO)2(CH2CHOEt)+PF6-, were prepared by
literature methods.42 All para-substituted anilines were sub-
limed prior to use. Samples of the known complexes 4-10 were
prepared by the literature methods.7 These Fp+ salts tend to
(43) These yields are after recrystallization.
(44) Delta NMR Software; JEOL USA Inc. Peabody, MA.
(45) (a) Bain, A. D.; Cramer, J. A. J. Magn. Reson. 1996, 118, 21.
(b) Copies of this program are available from A. D. Bain at the e-mail
address bain@mcmaster.ca.
(40) As described in: Braun, S.; Kalinowski, H.-O.; Berger, S. 150
and More Basic NMR Experiments; Wiley-VCH: New York, 1998; page
(46) Muhandiram, D. R.; McClung, R. E. D. J. Magn. Reson. 1987,
71, 187.
(47) Bain, A. D.; Cramer, J. A. J. Magn. Reson. 1993, 103A, 217.
(48) (a) Bain, A. D.; Dun, G. J. Can. J. Chem. 1996, 74, 819. (b) The
Mexico program is downloadable from http://www.chemistry.mcmas-
ter.ca/∼bain/exchange.html. (c) MEXITER was obtained upon request
from A. D. Bain, McMaster University.
136.
(41) March, J. Advanced Organic Chemistry; Wiley: New York,
1985; p 244.
(42) Cutler, A.; Raghu, S.; Rosenblum, M. J. Organomet. Chem.
1974, 77, 381.

Olefin Bond Rotation Barriers
Organometallics, Vol. 23, No. 23, 2004 5449
of the geminal coupling constant as well as the value of the
natural line width in the simulation. Experimental spectra
were imported into the program as ASCII text files. The
spectral simulation was run by refining the ν values for the
exchanging peaks49 until the ø2 value for the match between
the experimental and simulated spectra was minimized.50
While each derivative displayed some temperature dependence
in δν, such dependence is not uncommon51 and was reliably
modeled in MEXITER. Causes may include hydrogen bond-
ing,52 temperature-dependent changes in the dielectric con-
stant of the solvent,53 solute aggregation,54 and dynamic
behavior.55
T1G Experiment. Prior to the T1F experiments, the T1 value
for the nonirradiated proton was determined using the stan-
dard nonselective inversion T1 experiment in the JEOL
software package. The T1F pulse sequence, capable of decou-
pling the proton on the â-carbon, was provided by JEOL
software engineers.34 In this sequence, the spin-lattice relax-
ation time in the rotating frame is measured as a function of
the spin lock field strength.56 The decoupler irradiation offset
was set to the frequency of the â-proton. The spin lock field
attenuator was set at five to six levels, spaced between 15 and
50 dB. The spin lock pulse length was controlled by the pulse
program to give 25 evenly spaced values between 0.05 and 10
s. Data were analyzed using the standard JEOL T1/T2 data
analysis package. The chemical exchange rate was extracted
from the T1F value using the graphing method described by
Deverell.11 It should be noted that the T1F experiment may
cause heating of the sample if the spin-locking field is too
strong. This could cause error in the data if the measured rates
are in excess of 104 s-1 11
. All rates measured here were well
below this threshold except for one, the 25 °C data point for
4. This was measured to be 10 057 s-1, yet it remained well in
line with the rest of the T1F data, as well as the remaining
data measured independently by the other methods.
Acknowledgment. We wish to thank the Michigan
Space Grant Consortium and the Padnos School of
International Studies at Grand Valley State University
for their financial support of this work. Thanks are
given to Professor A. D. Bain from McMaster University
for use of the Mexiter program and his consultations,
which helped us greatly in our line shape analysis. His
suite of simulation and curve fitting programs were of
great value to this work. Thanks are also given to
Professor George McBane from Grand Valley State
University for his assistance with the Bootstrap error
analysis and for his consultations throughout the prepa-
ration of this paper. Special thanks are given to Ashok
Krishnaswami, Ph.D., Application Chemist, JEOL USA
Inc., Peabody, MA. We acknowledge the synthetic con-
tributions of undergraduate researcher Ryan Broekhui-
zen, who helped in preparing samples of 4 and 6.
(49) These values were always slightly less than the low-tempera-
ture exchange limit. The first run started with the low temperature
values and each subsequent run used slightly decreased values until
the ø2 value was minimized. Final refined values were used as the
starting values for simulating the next higher temperature.
(50) While the MEXITER program will fit the ν values, the exchange
rate, and the T1 values to minimize ø2, we found more consistent results
by manually entering ν values in a series of minimizations.
(51) Reference 10b, p 79.
(52) (a) Maricic, S.; Berg, U.; Frejd, T. Tetrahedron 2002, 58 3085
(the reference specifically refers to amide hydrogen bonding). (b)
Hirayama, F.; Horiuchi, Y.; Utsuki, T.; Uekama, K.; Yamasaki, M. Jpn.
Pharm. Res. 1993, 10, 208.
(53) (a) Golubev, N. S.; Shenderovich, I. G.; Tolstoy, P. M.; Shchep-
kin, D. N. J. Mol. Struct. 2004, 697, 9. (b) Takebayashi, Y.; Yoda, S.;
Sugeta, T.; Otake, K.; Nakahara, M. J. Phys. Chem. B 2003, 107, 9847.
(c) Garcia-Viloca, M.; Gonzalez-Lafont, A.; Lluch, J. M. Org. Lett. 2001,
3, 589.
(54) Macchioni, A.; Romani, A.; Zuccaccia, C.; Guglielmetti, G.;
Querci, C. Organometallics 2003, 22, 1526.
(55) It has been proposed that a temperature-dependent change in
the relative populations of endo and exo isomers interconverted by
propeller style rotation may exist in solution and that this may be
causing the observed temperature dependence of δν. This could mean
that the observed shifts in the 1H NMR are averages of these two,
which may add an additional component to the line broadening.
However, there is no evidence for populations of two isomers at -85
°C in the 13C NMR. Given that such temperature dependence is fairly
common and can alternatively be caused by many routine factors,51-54
we do not feel there is much evidence for this complicating factor. The
data set is linear as well as collinear with the data obtained by the
other methods, which are unaffected by such a possibility.
Supporting Information Available: A table of all mea-
sured rates by each of the methods. This material is available
free of charge via the Internet at http://pubs.acs.org.
OM040092R
(56) The parameter setup is described in: Braun, S.; Kalinowski,
H.-O.; Berger, S. 150 and More Basic NMR Experiments; Wiley-VCH:
New York, 1998; p 150.