Crossover studies of methyl migration from oxygen to iron in the iron-manganese methoxycarbyne complex Cp(CO)Fe(μ-COCH 3)(μ-CO)Mn(CO)MeCp

Article abstract of DOI:10.1021/om0502640

Thermal decomposition of the iron-manganese methoxycarbyne Cp(CO)Fe(μ-CO)(μ-COCH₃)Mn(CO)MeCp (1a) occurs to give MeCpMn(CO)₃ and in low yield CpFe(CO)₂CH₃; in the presence of PPh₃, CpFe(CO)(PPh₃)CH₃ (2a) forms in high yield. The reaction is first-order in carbyne, but a side reaction that is also first-order in phosphine occurs to give [Cp(CO)Fe(μ-CO)₂Mn(CO)MeCp]^-[CH₃PPh ₃]⁺ (3a) as a byproduct. Crossover experiments between la and its bis-MeCp, CD₃ analogue 1b-d₃ result in scrambling of the methoxycarbyne methyl label between the products 2a and the MeCp analogue 2b. No alkyl exchange is seen in recovered starting materials in the reaction between 1a and bis-MeCp analogue 3b or (except after prolonged reaction) between the products 2a and 2b-d₃. Control crossover experiments between 1a and 2b-d₃ give complete alkyl scrambling. The data prove that alkyl exchange occurs among products after carbyne decomposition and presumably is induced by an intermediate that is formed by carbyne decomposition. Previous results show that the 16-electron intermediate CpFe(CO)CH₃ forms at essentially the same rate from 2a as from la, ruling it out as the exchange intermediate since 2a and 2b-d ₃ undergo only very slow exchange. A speculative proposal for the reactive intermediate is that it is the isomeric terminal methoxycarbyne CpFe≡COCH₃, and alkyl migration from oxygen to iron occurs after cluster cleavage. A detailed kinetic scheme for alkyl scrambling is proposed and tested by computer modeling: using an iterative procedure that couples numerical integration of the proposed rate equations with a simplex minimization algorithm designed to find the best rate constants, concentration data from several runs could be quantitatively fit to the proposed mechanism.

Full text of DOI:10.1021/om0502640

Organometallics 2005, 24, 4179-4189
4179
Crossover Studies of Methyl Migration from Oxygen to
Iron in the Iron-Manganese Methoxycarbyne Complex
Cp(CO)Fe(µ-COCH3)(µ-CO)Mn(CO)MeCp
William H. Hersh* and Raymond H. Fong†
Department of Chemistry and Biochemistry, Queens College and the Graduate Center of the
City University of New York, Flushing, New York 11367-1597, and Department of Chemistry
and Biochemistry, University of California, Los Angeles, California 90024
Received April 7, 2005
Thermal decomposition of the iron-manganese methoxycarbyne Cp(CO)Fe(µ-CO)(µ-
COCH3)Mn(CO)MeCp (1a) occurs to give MeCpMn(CO)3 and in low yield CpFe(CO)2CH3; in
the presence of PPh3, CpFe(CO)(PPh3)CH3 (2a) forms in high yield. The reaction is first-
order in carbyne, but a side reaction that is also first-order in phosphine occurs to give [Cp-
(CO)Fe(µ-CO)2Mn(CO)MeCp]-[CH3PPh3]+ (3a) as a byproduct. Crossover experiments
between 1a and its bis-MeCp, CD3 analogue 1b-d3 result in scrambling of the methoxycarbyne
methyl label between the products 2a and the MeCp analogue 2b. No alkyl exchange is
seen in recovered starting materials in the reaction between 1a and bis-MeCp analogue 3b
or (except after prolonged reaction) between the products 2a and 2b-d3. “Control crossover”
experiments between 1a and 2b-d3 give complete alkyl scrambling. The data prove that
alkyl exchange occurs among products after carbyne decomposition and presumably is
induced by an intermediate that is formed by carbyne decomposition. Previous results show
that the 16-electron intermediate CpFe(CO)CH3 forms at essentially the same rate from 2a
as from 1a, ruling it out as the exchange intermediate since 2a and 2b-d3 undergo only
very slow exchange. A speculative proposal for the reactive intermediate is that it is the
isomeric terminal methoxycarbyne CpFetCOCH3, and alkyl migration from oxygen to iron
occurs after cluster cleavage. A detailed kinetic scheme for alkyl scrambling is proposed
and tested by computer modeling: using an iterative procedure that couples numerical
integration of the proposed rate equations with a simplex minimization algorithm designed
to find the best rate constants, concentration data from several runs could be quantitatively
fit to the proposed mechanism.
Scheme 1. Methyl Migration in Methoxycarbyne
Complexes
Introduction
Many reactions in organometallic chemistry can be
explained and even predicted by a relatively small set
of now familiar mechanistic types, such as oxidative
addition, reductive elimination, migratory insertion, and
substitution.1 Mechanistic studies of newly discovered
reactions typically involve efforts to find combinations
of these known basic steps that can then explain the
new reaction. Because some of these elementary reac-
tion types require that the reaction be intramolecular,
a critical study is often the crossover experiment. That
is, does a particular fragment remain bound to the same
metal atom throughout the course of the reaction, or
does it migrate to another metal atom?2,3 Many years
ago we reported a new reaction, in which the methyl
group of a bridging methoxycarbyne ligand migrated
* Address correspondence to this author at Queens College. E-
mail: william_hersh@qc.edu.
from oxygen to one of the metal atoms (Scheme 1),4,5
and since that time we have reported a few more
examples of this reaction.6-8 In a formal sense, this
† Current address: Department of Chemistry, City College of San
Francisco, 50 Phelan Ave., San Francisco, CA 94112.
(1) Collman, J. P.; Hegedus, L. S.; Norton, J. R.; Finke, R. G.
Principles and Applications of Organotransition Metal Chemistry;
University Science Books: Mill Valley, CA, 1987.
(2) Bergman, R. G. Acc. Chem. Res. 1980, 13, 113-120.
(3) Schore, N. E.; Ilenda, C. S.; White, M. A.; Bryndza, H. E.;
Matturro, M. G.; Bergman, R. G. J. Am. Chem. Soc. 1984, 106, 7451-
7461.
(4) Fong, R. H.; Hersh, W. H. Organometallics 1985, 4, 1468-1470.
(5) Fong, R. H.; Lin, C. H.; Idmoumaz, H.; Hersh, W. H. Organo-
metallics 1993, 12, 503-516.
(6) Wang, B.; Hersh, W. H.; Rheingold, A. L. Organometallics 1993,
12, 1319-1330.
10.1021/om0502640 CCC: $30.25 © 2005 American Chemical Society
Publication on Web 07/13/2005

4180 Organometallics, Vol. 24, No. 17, 2005
Hersh and Fong
Figure 2. Plot of observed first-order rate constants (kobs
,
75 °C) for decomposition of 1a vs PPh3 concentration.
Figure 1. First-order plots of decomposition of 1a at 75
°C as a function of PPh3 concentration (L ) PPh3).
gration with concomitant metal-metal bond cleavage
yields MeCpMn(CO)3 and the 16-electron fragment
CpFe(CO)CH3. In the absence of PPh3, decomposition
of CpFe(CO)CH3 apparently leads to loss of CO, which
is then scavenged to give low yields of CpFe(CO)2CH3
and [CpFe(CO)2]2, while loss of the alkyl residue can
lead to the tetrameric cluster [CpFe(CO)]4.9 In the
presence of PPh3, the 16-electron species is trapped to
give CpFe(CO)(PPh3)CH3 (2a). Independent synthesis
of MeCpMn(CO)2PPh310 showed that it did not form in
these reactions of 1a,5 requiring that the carbyne
directly yields MeCpMn(CO)3 and CpFe(CO)CH3, rather
than MeCpMn(CO)2 and CpFe(CO)2CH3 which then
exchange a CO ligand; formation of MeCpMn(CO)2PPh3
from MeCpMn(CO)3 and PPh3 occurs photochemically
but not thermally (even at 140 °C).10-12 At high con-
centrations of PPh3, dealkylation is the major reaction,
giving [Cp(CO)Fe(µ-CO)2Mn(CO)MeCp]-[CH3PPh3]+ (3a).
On the basis of yields of MeCpMn(CO)3 and 2a (see
Supporting Information), the reaction pathway that is
zero-order in PPh3 yields iron-alkyl 2a, while the
reaction pathway that is first-order in PPh3 yields
[Cp(CO)Fe(µ-CO)2Mn(CO)MeCp]-[CH3PPh3]+ (3a). At
0.1 M PPh3 we calculate on the basis of eq 1 that ∼68%
of the decomposition of 1a occurs via the phosphine-
independent pathway to give 2a; for convenience ap-
proximately this concentration was used for all subse-
quent crossover reactions.
methyl migration can be thought of as a â-elimination
reaction, which if literally true would require the
reaction to be intramolecular. However, crossover ex-
periments showed the products to be due, formally, to
intermolecular migration. Methyl exchange was shown
to be occurring after product formation, leaving un-
settled the question of molecularity of the migration.
As indicated in Scheme 1, methyl exchange presumably
is mediated by a 16-electron “M-Me” fragment, which
in this case might most simply be imagined to be the
CpFe(CO)Me species. In this paper we describe the
mechanistic experiments designed to unravel these
migration processes.
Results
Kinetics of Thermal Decomposition. First-order
decomposition of 1a occurred at a convenient rate at 75
°C in C6D6 and was observed both in the absence and
in the presence of triphenylphosphine. Under pseudo-
first-order conditions (at least a 5-fold excess of PPh3),
linear plots of ln[1a] vs time were obtained (Figure 1);
all the reactions were monitored for three half-lives.
Observed rate constants are given in Table 1 (runs 1-5).
A plot of the pseudo-first-order rate constants (kobs) vs
[PPh3] was linear with a nonzero intercept (Figure 2),
which, within experimental error, gave the same rate
constant as that measured in the absence of PPh3 (run
1). That is, the phosphine-independent pathway has the
same rate constant as that of simple thermal decompo-
sition. The observed rate law is given in eq 1.
Carbyne Crossover Experiments. Crossover ex-
periments were carried out next as shown in Scheme
3. The doubly labeled analogue 1b-d3, in which the Cp
ligand on iron was replaced by the MeCp ligand and
the methoxy CH3 group by the CD3 group, was allowed
to decompose in the presence of 1a and ∼0.1 M PPh3
(Table 1, run 6). After 30 min reaction at 65 °C, ∼5%
decomposition of the starting carbynes had taken place,
and the alkyl label in the products 2a and 2b was found
to be essentially completely scrambled between the
-d[1a]
) (k1 + k2[PPh3])[1a]
(1)
dt
k1 ) (9.8 ( 0.4) × 10-5 s-1, k2 ) (4.64 ( 0.09) ×
10-4 M-1 s-1
1
CpFe and the MeCpFe centers, as judged by H NMR
The reaction products and the simplest mechanistic
interpretation of the above rate data are shown in
Scheme 2; characterizations of starting materials and
products have been described previously.5 Methyl mi-
examination of their methyl signals. At 200 MHz, the
(9) Landon, S. J.; Rheingold, A. L. Inorg. Chim. Acta 1981, 47, 187-
189.
(10) Nyholm, R. S.; Sandhu, S. S.; Stiddard, M. H. B. J. Chem. Soc.
1963, 5916-5919.
(7) Idmoumaz, H.; Lin, C. H.; Hersh, W. H. Organometallics 1995,
14, 4051-4063.
(11) Angelici, R. J.; Loewen, W. Inorg. Chem. 1967, 6, 682-686.
(12) Hershberger, J. W.; Klingler, R. J.; Kochi, J. K. J. Am. Chem.
Soc. 1983, 105, 61-73.
(8) Luo, W.; Fong, R. H.; Hersh, W. H. Organometallics 1997, 16,
4192-4199.

Crossover Studies of Methyl Migration
Organometallics, Vol. 24, No. 17, 2005 4181
Table 1. Summary of Experimental Data: Rate Constants for Carbyne Decomposition and Carbyne and
Product Crossover Resultsa
a
run
[1a] (M)
[PPh3] (M)
additive (M)
104k (s-1
)
crossoverb
1
2
3
4
5
6
0.028
0.010
0.025
0.021
0.032
0.022
0
0.937 ( 0.022
1.24 ( 0.04
1.57 ( 0.05
3.12 ( 0.12
4.64 ( 0.16
0.48 ( 0.01
0.0577
0.128
0.442
0.800
0.12
1b-d3 (0.022)
1b-d3 (0.017)
yes (∼100% exchange after 0.5 h, ∼5% conversion
of 1a and 1b-d3)
1b-d3: 0.316 ( 0.007
7
0.016
0.089
yes (1 h 75 °C, 65% exchange after 44 ( 5%
conversion of 1; no crossover in 1 (<2% exchange
of CH3 and CD3 in recovered 1)c
8
9
0.12
0.12
2a (0.017)
2b-d3 (0.014)
2b-d3 (0.028)
no (20 h; 51% exchange after 285 h, 80 °C)
0.026
0.026
0.69 ( 0.04
0.78 ( 0.04
yes (31% exchange after 0.5 h, 10% conversion
of 1a; ∼100% exchange after 8 h, 81%
conversion of 1a)
yes (47% exchange after 0.7 h, 23% conversion
of 1a; 64% exchange after 6 h, 85% conversion of 1a)
10
0.16
2b-d3 (0.063)
a All data collected in C6D6, at 75 °C for runs 1-5 and run 7, and at 65 °C for the rest. Except as noted, rate constants are for
decomposition of 1a. b Crossover of alkyl label between 2a and 2b. Complete crossover is 100% exchange (see Experimental Section).
c See Experimental Section for further details of this run.
Scheme 2. Thermal Methoxycarbyne
Decomposition
Scheme 3. Methoxycarbyne Crossover Experiment
molecular alkyl exchange in the mononuclear products,
it did not address the question of when or how the
exchange occurred. A variety of control experiments
were carried out to determine at what stage of the
reactions the methyl exchange was occurring.
Methoxycarbyne Crossover Control Experi-
ments. 1. Carbyne Crossover. Four control experi-
ments were conducted to investigate methyl exchange
prior to carbyne decomposition. A rational mechanism
would involve a chain reaction between trace amounts
observed doublets (due to coupling to the PPh3 phos-
phorus atoms) overlapped so that one line of each was
coincident, but comparison of the peak heights or
integrals of the first and third lines allowed the extent
of scrambling to be determined (see Experimental
Section). Monitoring of the reaction by 1H NMR showed
the two carbynes to decompose at comparable rates, and
the presence of 2a-d3 and 2b-d3 was inferred from
integration of the Cp and MeCp peaks. Simple first-
order plots gave excellent linear fits and the rate
constants shown in Table 1, although these values are
likely to be slightly erroneous due to the fact that only
a 3-fold excess of PPh3 was used in this run. The
reaction was followed to close to completion (∼13 h), and
complete methyl exchange occurred throughout.
+
of the CH3PPh3 salts of the dinuclear anions (3a-
+
CH3PPh3 and 3b-CH3PPh3+) formed in situ and the
carbynes as shown in Scheme 4. However, no crossover
was observed between 1a and 3b or between 1b-d3 and
3a, and as a check, 3a did not inhibit crossover in the
reaction of 1a and 1b-d3 (see Supporting Information
for details).
This anion exchange mechanism would give not only
product crossover but also crossover in the starting
materials. The standard control experiment, in which
the crossover reaction of 1a and 1b-d3 was run to partial
conversion and then the unreacted starting carbynes
examined for alkyl exchange, was carried out (eq 2,
While the crossover experiment between the doubly
labeled carbynes shown in Scheme 3 resulted in inter-

4182 Organometallics, Vol. 24, No. 17, 2005
Hersh and Fong
Scheme 4. Potential Anion-Mediated Crossover
Mechanism
2. Product Crossover. Alkyl exchange also can in
principle occur after carbyne decomposition occurs,
among the mononuclear products. The required control
experiment involves checking for crossover in the prod-
ucts 2a and 2b-d3. Under the same reaction conditions
used for the crossover reaction between 1a and 1b-d3,
that is, after 3 h at 65 °C in the presence of ∼0.1 M
PPh3, no signal due to the FeCH3 hydrogens of 2b was
visible (run 8, eq 3), and hence crossover is not due to
simple methyl exchange among the reaction products.
Even after 20 h at 65 °C no 2b was observed, but
prolonged reaction at 80 °C did lead to slow exchange
(run 8, eq 3), suggesting that 2a and 2b slowly yield an
intermediate that could lead to methyl crossover. How-
ever, no exchange could be induced by 3a or MeCpMn-
(CO)3, and in one run where 3a was introduced,
CpFe(CO)(PPh3)H was detected but still no exchange
occurred (see Supporting Information for details). In
conclusion, methyl exchange could not be induced to
occur in the absence of carbyne except after prolonged
heating. Examination of the mechanism of this ex-
change awaits further study, but it cannot be assumed
to be the same as that observed in the presence of the
carbynes.
Table 1, run 7), since this would allow the detection of
alkyl scrambling among starting materials regardless
of the molecular details of any exchange. After 44%
3. Carbyne/Product Crossover. The key crossover
experiment, dubbed the “control crossover”, was de-
signed to check the role of the carbyne rather than any
of the decomposition products in crossover. Here car-
byne 1a was mixed with “product” 2b-d3 (Scheme 5), and
in sharp contrast to the absence of reaction between
products 2a and 2b in eq 3, complete scrambling was
observed at 65 °C within 8 h (run 9). Since no 1b was
present, the crossover product 2b-h3 formed could not
arise due to intermolecular methyl migration from the
methoxycarbynes.
consumption of the carbynes, the extent of scrambling
of the methyl label in products 2a and 2b was 65%
complete, while no exchange was detected in the recov-
ered starting materials. The carbynes were isolated by
chromatography, and the mixture was analyzed by
electron ionization mass spectrometry. Large molecular
ion signals were observed for 1a (m/e ) 382) and 1b-d3
(m/e ) 399), but no peaks larger than 2% of these
molecular ions plus their M + 1 peaks were observed
at 396 corresponding to 1b or at 385 corresponding to
1a-d3.
Monitoring this reaction by 1H NMR showed that
while apparent equilibration of 2a and 2b occurred,
there appeared to be an induction period. That is, at 30
Scheme 5. Control Crossover Experiment
A final possibility could be that alkyl exchange among
the carbynes occurs concomitant with carbyne decom-
position, so that the unreacted carbynes would not
exhibit crossover. However, such a scheme would give
bimolecular decomposition kinetics, and only uni-
molecular kinetics are observed. While it can be argued
that NMR kinetics are not sufficiently precise to observe
the curvature in a log plot that would be indicative of
bimolecular kinetics, in fact similar rate constants have
been observed in some runs where fairly different
starting concentrations of the carbynes were used; for
instance in run 1 [1a] was 2.8 times [1a] in run 2.

Crossover Studies of Methyl Migration
Organometallics, Vol. 24, No. 17, 2005 4183
min the extent of exchange was 30% that expected at
equilibrium, and this gradually rose to about 90% after
3 h reaction, when half of the carbyne had been
consumed. In this experiment, as in virtually all of the
others, approximately equal amounts of the CH3 and
CD3 components were used, so at equilibrium, if in fact
an equilibration was involved, similar amounts of 2a
and 2b would be expected. If there is an equilibration
step, use of excess 2b-d3 might initially yield more 2b
than 2a. A control crossover experiment was therefore
run in which the initial ratio 2b-d3:1a was 2.5:1 rather
than 1:1 (run 10). The observed result was little changed
(Scheme 5): initially more 2a than 2b formed, and as
the reaction progressed, the amount of 2b formed
approached that of 2a, although in this run the final
extent of “equilibration” was less than 70%. The major
difference between runs 9 and 10 was found in the
relative amounts of 2a and 2a-d3: in the 1:1 2b-d3:1a
experiment (run 9), equal amounts of 2a and 2a-d3
formed, while in the 2.5:1 experiment (run 10), the ratio
of 2a-d3:2a was about 2:1. That is, the distribution of
methyl label in 2a did behave as though it was due to
equilibration, but the distribution of label in 2a and 2b
did not.
for dissociation of PPh3 from 2a to give 16-electron
intermediate Ia is slightly faster than carbyne decom-
position to give 2a. The rates for decomposition of 1a
(i.e., run 6 at 65 °C) in Table 1 include the bimolecular
pathway leading to anion 3a, while the path leading to
2a is less than ∼5 × 10-5 s-1 (vide infra). The key point
is that since no methyl exchange occurs between 2a and
2b-d3 in the absence of carbyne 1 under these condi-
tions, CpFe(CO)CH3 cannot be the intermediate formed
from 1a that induces methyl exchange.
There is an additional, albeit more subtle, kinetic
problem with any CpFe(CO)Me mechanism that allows
it to be rejected, along with any other kinetically related
mechanism in which a common intermediate leads to
both product formation and methyl exchange. The
degree of methyl exchange will depend on the relative
rates of trapping of intermediate CpFe(CO)CH3 by PPh3
to give product 2a and by 2b-d3 which regardless of the
details would be required for methyl exchange. If
trapping by PPh3 is faster, the result would be slower
formation of 2b than 2a throughout the reaction,
without attainment of equilibrium of the two products.
On the other hand, if trapping of CpFe(CO)CH3 by 2b-
d3 is faster, then the initially formed mononuclear alkyl
product from 1a would be the crossover product 2b, not
2a! The fact that three experimentssone starting with
equal amounts of 1a and 1b-d3, one with equal amounts
of 1a and 2b-d3, and one with 1a and 2.5 equiv of 2b-
d3sgive essentially 1:1 ratios of 2a and 2b makes it
unlikely that this result is due to some serendipitous
set of rate constants. The conflicting requirements of
product formation and equilibration require that a
common intermediate cannot lead to both pathways.
Further quantitative kinetic details may be found in the
Supporting Information.
3. The Reversible Cleavage Mechanism. A further
constraint on the mechanism was uncovered during
attempts to quantitatively model the CpFe(CO)Me
mechanism. Both the rate of carbyne decomposition and
yields of product were reproducibly correlated with the
amount of “product” initially present in solution. That
is, the two-carbyne experiments always gave relatively
lower rates and yields than the control crossover reac-
tions, and the 2.5:1 2b-d3:1a experiment gave the
highest rate and yield. For instance, derived values for
the first-order rates of decomposition of 1a when each
run was considered separately (see Supporting Informa-
tion for details) gave the methyl migration rates after
deducting the PPh3-induced bimolecular decomposition,
and these rates rose from 2.1 × 10-5 s-1 for the two-
carbyne experiment (run 6) to 5.0 × 10-5 s-1 for the
high-concentration control crossover (run 10). Alterna-
tively, examination of the raw datasthat is, first-order
carbyne decomposition rates that include the PPh3-
induced decompositionsshowed the same trend, pre-
sumably because [PPh3] differs little from run to run:
the rates rose from 4.8 × 10-5 s-1 for the two-carbyne
experiment (run 6) to 7.8 × 10-5 s-1 for the high-
Discussion
1. Constraints. The results described above are
complex and severely restrict the available mechanisms.
Facts that have been established are (1) methyl ex-
change occurs between carbynes only upon carbyne
decompositionsno exchange is detected in unreacted
carbynes, (2) methyl exchange is essentially complete
at early reaction time for the carbyne crossover, (3)
methyl exchange among the mononuclear products in
the first few hours of reaction occurs only in the
presence of carbynes, and (4) under all conditions,
roughly equivalent amounts of the CH3 methyl exchange
products form, even when the initial amounts of CH3
and CD3 starting materials differ substantially.
The most straightforward conclusion that follows from
the above facts is that a carbyne decomposition product
is involved in the methyl exchange reaction. We have
established that MeCpMn(CO)3 is extruded from the
carbyne without formation of any MeCpMn(CO)2PPh3
that would arise from MeCpMn(CO)2 (Scheme 2), leav-
ing behind the elements of CpFe(CO)CH3seither in this
form or some othersthat must also be extruded from
the carbyne. Since 2a can form by trapping of CpFe-
(CO)CH3 by PPh3, this specific intermediate is an
obvious candidate for involvement in the exchange
reaction, and one can readily imagine how this might
form by methyl migration from oxygen to iron.7
2. The CpFe(CO)Me Mechanism. The products 2a
and 2b-d3 do not exchange in the absence of 1a except
after prolonged reaction. Therefore, if CpFe(CO)Me and
MeCpFe(CO)Me are responsible for methyl exchange,
then loss of phosphine from 2a (and of course any of the
products 2b, 2a-d3, 2b-d3) to give these intermediates
must be much slower than their formation from carbyne
decomposition. While this assumption in fact seems
reasonable, it was tested by directly examining the
kinetics of phosphine substitution in 2a and was found
to be wrong.13 As shown in eq 4, rather than being
slower than carbyne decomposition, the rate at 65 °C

4184 Organometallics, Vol. 24, No. 17, 2005
Hersh and Fong
Scheme 6. Mechanism of Carbyne Cleavage and
Methyl Exchange
concentration control crossover (run 10). The final yield
of 2a and 2a-d3 rose from 39% in the two-carbyne
experiment to 61% in the 0.028 M 2b-d3 control cross-
over to 75% in the 0.063 M 2b-d3 control crossover
experiment. In all three cases, the rates and yields
appear to be correlated only with [2b-d3]; as this
concentration rises, more of the reaction is siphoned off
toward the reaction of interest and less to the PPh3 SN2
reaction. A plot of the rates of first-order decomposition
of 1a (k1 from Table S-4 for the individual runs in the
Supporting Information or the raw rate from Table 1)
vs [2b-d3]t)0 gave parallel straight lines (slope ) (4.5 (
0.5) × 10-4 M-1 s-1 and (4.7 ( 1.3) × 10-4 M-1 s-1
,
respectively, for runs 6, 9, 10), suggesting a bimolecular
component to the reaction. Since decomposition of the
carbynes is first-order, this result can only occur if the
carbyne is in reversible equilibrium with some inter-
mediate that is trapped by 2b-d3, and this trapping is
relatively slow compared to reversal back to starting
material. Nonetheless, this trapping will be faster at
higher initial concentrations of 2b-d3, and as more of
the reaction is diverted to this pathway, not only will
higher rates of reaction be observed, but the amount of
decomposition of the carbyne to the byproduct PPh3Me+
salt 3 (Scheme 2) will decrease, thereby leading to the
higher observed yields since less starting material will
be diverted by PPh3.
A mechanism based on reversible carbyne cleavage
is shown in Scheme 6. The initial step, production of
intermediate IA, is not kinetically distinguishable, but
is reasonable based on prior work in which rapid cis/
trans isomerization was observed; that is, following
carbyne unbridging (and carbonyl unbridging, which is
not shown here), rotation about the Fe-Mn bond can
occur. We have previously invoked this step for other
carbyne reactions in which cis/trans isomerization and
phosphine attack on reversibly formed dinuclear inter-
mediates occurs.6,8,14 Intermediate IA consists of MeCp-
Mn(CO)3 side-bound to the iron, and it is reasonable to
suggest that this species will rapidly decompose via
cleavage of the µ-CO-iron bonds to give the 16-electron
intermediate CpFeCtOCH3 (I1). Precedent for expulsion
of MeCpMn(CO)3 comes from the cleavage of isoelec-
tronic heterodinuclear compounds,15-18 which we have
previously compared to reactions of related hetero-
dinuclear carbyne complexes.6-8 Here, however, we
propose that the reaction is reversible; the subsequent
trapping by 2b-d3 is similar to trapping by MeCpMn-
(CO)3, and the fact that the best-fit rates for these
trapping steps, k6 and k7, will turn out to be quite
similar (see below) lends further support for this mech-
anism. It is reasonable to suppose that the trapping by
2b-d3 is similarly reversible, as shown by the dashed
arrow, although it is kinetically indistinguishable from
the irreversible trapping, and this step was not given
any further consideration. While many parts of this
mechanism are highly speculative, it is this portion of
the proposal alone that allows the decomposition rates
and the product yields to be fit reasonably well; all
variants on this mechanism that do not include this
reversible carbyne cleavage followed by competitive
trapping do not give reasonable fits. Following trapping
of I1 by the iron alkyls, as illustrated in Scheme 6 for
2b-d3 to give IB, we propose fast methyl migration from
oxygen to iron to give a diiron dimethyl species, IC1
,
which is in rapid equilibrium with IC2; rapid cleavage
to product and CpFe(CO)Me followed by trapping by
PPh3 then gives the final products. Methyl exchange in
this scheme does not involve equilibration of the methyl
groups among all the iron alkyl species in solution, but
rather only during the trapping step in a so-called “one-
time” methyl exchange.
In the absence of 2b-d3 initially, a route to some iron
alkyl must exist, since otherwise I1 could only revert
back to IA according to the mechanism thus far. A step
involving isomerization of carbyne I1 to CpFe(CO)CH3,
followed by trapping to give 2a, is reasonable, but there
is no obvious intuitive answer to whether this will be a
pathway that is competitive with the trapping pathway.
That is, significant production of 2a might come from
this isomerization pathway if k5 were large relative to
k6. Alternatively, slow formation of 2a along the k5
(13) Hersh, W. H.; Hunte, F.; Siegel, S. Inorg. Chem. 1993, 32,
2968-2971.
(14) Hersh, W. H. In Transition Metal Carbyne Complexes; Kreissl,
F. R., Ed.; Nato ASI Series C: Mathematical and Physical Sciences,
Vol. 392; Kluwer Academic Publishers: Dordrecht, 1993; pp 149-150.
(15) Leonhard, K.; Werner, H. Angew. Chem., Int. Ed. Engl. 1977,
16, 649-650.
(16) Werner, H.; Juthani, B. J. Organomet. Chem. 1981, 209, 211-
218.
(17) Werner, H. Pure Appl. Chem. 1982, 54, 177-188.
(18) Aldridge, M. L.; Green, M.; Howard, J. A. K.; Pain, G. N.; Porter,
S. J.; Stone, F. G. A.; Woodward, P. J. Chem. Soc., Dalton Trans. 1982,
1333-1340.

Crossover Studies of Methyl Migration
Organometallics, Vol. 24, No. 17, 2005 4185
Chart 1. Reaction Steps for Scheme 6
instance) MeCpMn(CO)3 would be expected to react with
I1 and I2 at slightly different rates, k7 and k7′; the
additional rate constants were found to be unnecessary.
A computer program was written to quantitatively
model the system, using known numerical routines to
(1) numerically integrate the coupled set of differential
equations for the reactions in Chart 1, using a fourth-
order Runge-Kutta method with adaptive step-size
control, (2) calculate the deviation of the values of the
calculated time points from the observed (see Experi-
2
mental Section for details of the error function øν ), and
(3) iteratively adjust the rate constants using a simplex
minimization algorithm to give the best-fit set of rate
constants simultaneously for all the runs.19 The pro-
gram, dubbed CRK2005 (“Complex Reaction Kinetics”,
updated from previous versions20,21), requires input of
the differential equations, the initial concentrations of
all species, and guesses for each rate constant. In
addition, since all the calculated concentrations depend
on the initial values input at zero time, CRK2005 allows
initial concentrations to be adjusted to give the best fit;
in practice, only the initial concentration of 1a was
varied (by up to ∼5%). Since this is a steady-state
system, only the ratio of the “fast” rate constants k5/k6
and k7/k6 can be determined, although in practice due
to the complexity of the equations, the steady-state
approximation was not used. Instead, rate equations for
each reaction species were used rather than algebra-
ically eliminating the intermediates, the concentrations
of which were constrained to be less than 0.01% of the
highest concentration in a kinetic run. While there is
no obvious precedent for a guess at the value of k6, it is
reasonably lower than that for the reaction of CpFe-
(CO)Me and PPh3. Use of Wrighton’s22 lower limit of 3
× 103 M-1 s-1 required unreasonable amounts of
computer time in order to successfully integrate equa-
tions with rate constants that differed by ∼8 orders of
magnitude, and the rate could be higher, given Ford’s23
value of 3 × 106 M-1 s-1 for trapping by CO. Neverthe-
less, values of k6 below even 1 M-1 s-1 gave identical
results with far less computer time, apart from the
concentrations of the intermediates, although in order
to keep these concentrations below the 0.01% level,
pathway would essentially give 2a as a catalyst for
reaction of I1. That is, as soon as any 2a forms, it would
catalyze the decomposition of I1 by trapping to give IB,
which then would yield more 2a. In the same way, 1b-
d3 will yield 2b-d3 and methyl exchange can then
proceed. Since this is an autocatalytic process, slow
isomerization of I1 to 2a will not be a bottleneck for
decomposition of 1a. In fact, as will be seen below, the
kinetic modeling clearly favors this autocatalytic path-
way, since rapid formation of 2a from I1 would give far
more 2a relative to 2b in the control crossover reactions.
The complexity of this system requires numerical
modeling. The reaction steps and rate constants needed
to model the mechanism in Scheme 6 are shown in
Chart 1. The kinetically distinguishable steps are (1)
reversible cleavage of the dinuclear carbynes to give
MeCpMn(CO)3 and the mononuclear 16-electron car-
bynes CpFetCOCH3 (I1) and MeCpFetCOCD3 (I2) by
k1 or k3 and k7, (2) trapping of the intermediate
mononuclear carbynes I1 and I2 by PPh3 to give products
2a and 2b-d3 by k5, and (3) trapping of the intermediate
mononuclear carbynes I1 and I2 by each of the iron alkyl
products in a one-time exchange event as an alternative
route to each of 2a, 2a-d3, 2b, and 2b-d3 by k6. Methyl
exchange occurs only following trapping of I1 by 2b-d3
(shown in Scheme 6) and of I2 by 2a (not shown in
Scheme 6) to give intermediate IB in the kinetically
significant k6 step, followed by several fast steps. Rather
than specify each of these fast steps, the mechanism was
modeled by substituting the fraction of exchange (f8) as
shown in Scheme 6 and Chart 1. That is, for instance,
in terms of the microscopic fast rate constants (see
Scheme 6 for k8 and k-8), if the rates of cleavage of IC1
and IC2 were equal, then the fraction of exchange would
be given exactly by f8 ) k8/(k8 + k-8), and the fraction
that does not give exchange by k-8/(k8 + k-8) ) 1 - f8.
If methyl exchange between IC1 and IC2 is rapid and
there is no thermodynamic preference for either inter-
mediate (i.e., K ) k8/k-8 ≈ 1) and if their rates of
cleavage are similar, then the fraction f8 should be close
to 0.5. In addition, use of this fraction emphasizes the
fact that there is only one independent variable for this
step. To minimize the number of different rate con-
stants, the same rate constants for each of the different
types of trapping steps were used even though (for
typical values of k6 were greater than 100 M-1 s-1
.
Use of the equations from Scheme 6/Chart 1 in
CRK2005 gave an excellent fit to the observed data for
runs 6, 9, and 10. The observed data and calculated lines
using rate constants k1-k7 and f8 are shown in Figure
2
3a-c, giving an overall deviation øν ) 0.952 for 191
nonzero data points (and an average error of 9.17%; see
Experimental Section for an explanation of these error
functions). A mechanism based on CpFe(CO)Me inter-
mediates, previously discarded as described above, could
also be numerically modeled and not surprisingly gave
2
much higher errors of øν ) 3.12 and an average error
of 16.7% (see Supporting Information for details). Of
(19) Press, W. H.; Flannery, B. P.; Teukolsky, S. A.; Vetterling, W.
T. Numerical Recipes: The Art of Scientific Computing; Cambridge
University Press: Cambridge, 1986.
(20) Bonnesen, P. V.; Baker, A. T.; Hersh, W. H. J. Am. Chem. Soc.
1986, 108, 8304-8305.
(21) Bonnesen, P. V.; Puckett, C. L.; Honeychuck, R. V.; Hersh, W.
H. J. Am. Chem. Soc. 1989, 111, 6070-6081.
(22) Kazlauskas, R. J.; Wrighton, M. S. Organometallics 1982, 1,
602-611.
(23) McFarlane, K. L.; Ford, P. C. Organometallics 1998, 17, 1166-
1168.

4186 Organometallics, Vol. 24, No. 17, 2005
Hersh and Fong
Figure 3. Best fit of mechanism in Scheme 6, using the equations from Chart 1, to observed concentration data for runs
6, 9, and 10. The calculated lines for [2a] + [2a-d3] and MeCpMn(CO)3 are coincident for runs 9 and 10, and the line for
[2b] + [2b-d3] at 0.063 M is not shown in run 10; it is an exact fit just as seen for the 0.028 M line for [2b] + [2b-d3] in
run 9.
interest is (1) the value of k1 is 4.66 ( 0.12 s-1, which
allows each of the three runs to fit the rate of decom-
position of 1a and is also less than the rate of dissocia-
tion of PPh3 from 2a (eq 3), (2) the data may be fit with
a wide range of values of k5/k6 with values of k5 that
are many orders of magnitude less than k6, so as
described above the k5 step is clearly indicative of an
autocatalytic reaction, (3) k6 and k7 are comparable in
value, consistent with trapping of reactive carbyne I1
with either 2a, 2b, or MeCpMn(CO)3, (4) the best fit
gives f8 ) 0.65 ( 0.03, not far from the expected value
of ∼0.5, and (5) allowing different rate constants to be
used for the different trapping steps (i.e., k7 for I1 and
MeCpMn(CO)3 and k7′ for I2 and MeCpMn(CO)3, or
variations in k6 for the different trapping combinations
of I1 and I2 with 2a and 2b, for instance) did not improve
the fit, thereby keeping the overall number of param-
eters required to a respectably small value.
4. Kinetics Leftovers. While the overall fit of the
data to the proposed mechanism is good, it is not perfect,
and so the question arises: is the mechanism still
incomplete or imperfect, or are the data imperfect? For
instance, the mechanism does not reproduce the ob-
served induction period in which more 2a forms than
2b at the outset of the reaction in the control crossover
reactions. Indeed, for the high-concentration 2b-d3 run,
the best fit leaves the concentration of 2b higher than
2a throughout the run, the opposite of what was
observed. Also as previously noted for run 10 (Scheme
5), the calculated ratio of 2a-d3:2a was lower than that
observed, for instance, as seen in Figure 3c, where the
calculated line for 2a-d3 + 2a is lower than the observed
data. Up to this point in the mechanism, there are no
“extra” steps or rate constants. That is, while common
wisdom is that any reaction can be fit given sufficient
number of mechanistic steps and rate constants, no
extraneous steps have been described thus far. Indeed,
we have attempted to minimize the proliferation of rate
constants, for instance as noted above by setting the
trapping rates for CpFe and MeCp analogues equivalent
to each other (i.e., k6 and k7). In this case, it is probably
best to conclude that while the data are sufficiently
consistent to point toward some mechanistic detail that
is yet uncovered, the data are also sufficiently imperfect
to not warrant further incorporation of additional
mechanistic steps.
5. Terminal Methoxycarbynes, Related µ-Car-
bynes, and Alkyl Migration. A brief review of related
carbynes, both terminal (observed and proposed) and
bridging, as well as their observed and proposed migra-
tion chemistry, is in order. We are aware of only one
report of a terminal methoxycarbyne, that is, a stable
18-electron analogue of CpFetCOCH3 (I1) and MeCpFet
COCD3 (I2), namely, Τp′(CO)2WtCOCH3 reported by
Templeton.24 A number of sulfur analogues, that is,
(24) Stone, K. C.; White, P. S.; Templeton, J. L. J. Organomet. Chem.
2003, 684, 13-19.

Crossover Studies of Methyl Migration
Organometallics, Vol. 24, No. 17, 2005 4187
terminal methylthiocarbynes, were described years ago
by Angelici,25,26 and terminal siloxycarbynes have been
described by Lippard.27,28 Templeton’s carbyne acts as
an electrophilic methylating agent, eliminating the
relatively stable anion Τp′(CO)3W-; such a pathway
could provide a mechanism for oxygen to methyl migra-
tion by intermolecular methyl transfer, but is unlikely
for the putative 16-electron iron carbynes, which would
have to eliminate the 16-electron anion CpFetCO- or
perhaps CpFe(PPh3)CO-, neither of which would be
reasonable and for good measure would be even less
likely in our case due to the use of benzene as the
solvent.
interconversion by 1H NMR spectroscopy.32 No reactions
of the carbyne were described, while the methyl complex
could be carbonylated to give an acyl complex. While
not A-frame compounds, Ruiz has reported the second
pair of isomeric µ-methoxycarbyne and in this case µ-Me
compounds, but no reactions of these dimolybdenum
PCy2-bridged compounds were reported.41 Cowie has
reported an extensive series of A-frame compounds,
some of which are cited above; perhaps the closest
analogues involve a diiridium system in which σ-methyl
and σ-methoxymethyl complexes were synthesized, and
in the latter case double C-H insertion gave an un-
reactive µ-methoxycarbyne dihydride.37 In addition,
Cowie has reported heterodinuclear compounds that
undergo methyl migration from one metal to the
other,35,39,40 so at least that step in our mechanism has
precedent.
In addition to the 1,2-oxygen to carbon alkyl migra-
tion reactions described above, a few related migration
examples have been described or proposed. One involves
reaction of metal anions with Fischer-type methoxy-
phenyl carbenes to give intermolecular methyl transfer
and an anionic acyl complex42 in a reaction that is
comparable to demethylation of Templeton’s methoxy-
carbyne with expulsion of Τp′(CO)3W-; as noted above
this pathway seems unlikely for our carbynes and was
shown not to occur between the dinuclear anions 3 and
the dinuclear carbynes 1. Several examples have been
proposed in which a silicon group migrates from iron to
an acyl oxygen to give a carbene,43,44 while in two
catalytic CO reductions by hydrosilanes, a step was
proposed that is the reverse of our methyl migration,
in which a silyl moiety migrates from metal to the
oxygen of a bound carbonyl to give a terminal siloxy-
carbyne.45,46 Decomposition of some silyl-substituted
Fischer-type carbenes has been proposed to proceed by
migration first of the silicon to the metal to give a
terminal alkoxycarbyne, followed by migration of the
alkyl group to the metal.47,48
We are not aware of any other reports of methyl
migration from oxygen to metal in µ-methoxycarbynes
other than those that we have described.6-8 Ford29 and
Seyferth30 have described examples of alkyl migration
in µ-alkoxycarbynes from oxygen to the neighboring
carbon to give µ-acyl compounds. Such an unconven-
tional 1,2-shift could serve as an alternative route to
methyl migration to the metal since subsequent “nor-
mal” carbonyl deinsertion could then occur via alkyl
migration from the acyl to the metal, or the “1,2-shift”
could be a product of our equally unconventional initial
alkyl migration from the carbyne oxygen directly to the
metal followed by “normal” carbonyl insertion via alkyl
migration back to the carbonyl. While we have consid-
ered acyl intermediates in our mechanistic schemes, we
see no advantage or necessity for such intermediates,
and there is no evidence for the formation of Cp(CO)-
31
(PPh3)FeC(O)CH3 in any of our reactions.
A potential example of methyl migration might rea-
sonably be sought from the variety of A-frame dimetal
compounds that have been reported with µ-methoxy-
carbyne, hydroxycarbyne, and methylene ligands, σ-alkyl
ligands, hydrides, and acyl ligands.32-40 In none of these
cases have interconversions of carbyne and alkyl ligands
been observed. Gladfelter reported the first set of strictly
isomeric µ-methoxycarbyne and σ-Me compounds, and
these diruthenium compounds did not exhibit any
One conclusion from the above is that dinuclear
systems with sufficient stability to exist both as the
isomeric carbyne and methyl carbonyl compounds32,41
may not exhibit interconversion of these species. How-
ever, the bridging acyls may be on the same reaction
pathway, although like our system, there is no detail
on the mechanism of the actual migration reaction
itself.29,30 The proposal of the intermediacy of a terminal
methoxycarbyne now has precedent with the observa-
tion of the first isolable example,24 and the intra-
molecular migration of the methyl from oxygen to iron
is supported by a small number of examples in both the
(25) Dombek, B. D.; Angelici, R. J. Inorg. Chem. 1976, 15, 2397-
2402.
(26) Greaves, W. W.; Angelici, R. J. Inorg. Chem. 1981, 20, 2983-
2988.
(27) Vrtis, R. N.; Liu, S.; Rao, C. P.; Bott, S. G.; Lippard, S. J.
Organometallics 1991, 10, 275-285.
(28) Vrtis, R. N.; Bott, S. G.; Lippard, S. J. Organometallics 1992,
11, 270-277.
(29) Friedman, A. E.; Ford, P. C. J. Am. Chem. Soc. 1989, 111, 551-
558.
(30) Seyferth, D.; Ruschke, D. P.; Davis, W. M. Organometallics
1994, 13, 4695-4703.
(31) Brookhart, M.; Tucker, J. R.; Husk, G. R. J. Am. Chem. Soc.
1983, 105, 258-264.
(32) Johnson, K. A.; Gladfelter, W. L. Organometallics 1990, 9,
2101-2105.
(33) Antonelli, D. M.; Cowie, M. Organometallics 1991, 10, 2550-
2559.
(41) Garc´ıa, M. E.; Melo´n, S.; Ramos, A.; Riera, V.; Ruiz, M. A.;
Belletti, D.; Graiff, C.; Tiripicchio, A. Organometallics 2003, 22, 1983-
1985.
(42) Toomey, L. M.; Atwood, J. D. Organometallics 1997, 16, 490-
493.
(34) Sterenberg, B. T.; Hilts, R. W.; Moro, G.; Mcdonald, R.; Cowie,
M. J. Am. Chem. Soc. 1995, 117, 245-258.
(35) Antwi-Nsiah, F. H.; Oke, O.; Cowie, M. Organometallics 1996,
15, 1042-1054.
(43) Brinkman, K. C.; Blakeney, A. J.; Krone-Schmidt, W.; Gladysz,
J. A. Organometallics 1984, 3, 1325-1332.
(44) Knorr, M.; Braunstein, P.; Decian, A.; Fischer, J. Organome-
tallics 1995, 14, 1302-1309.
(36) Alvarez, M. A.; Garc´ıa, M. E.; Riera, V.; Ruiz, M. A. Organo-
metallics 1999, 18, 634-641.
(37) Torkelson, J. R.; Oke, O.; Muritu, J.; McDonald, R.; Cowie, M.
Organometallics 2000, 19, 854-864.
(45) Chatani, N.; Fukumoto, Y.; Murai, S. J. Am. Chem. Soc. 1993,
115, 11614-11615.
(38) Alvarez, M. A.; Garc´ıa, M. E.; Riera, V.; Ruiz, M. A.; Robert, F.
Organometallics 2002, 21, 1177-1183.
(46) Chatani, N.; Shinohara, M.; Ikeda, S.; Murai, S. J. Am. Chem.
Soc. 1997, 119, 4303-4304.
(39) Trepanier, S. J.; McDonald, R.; Cowie, M. Organometallics
2003, 22, 2638-2651.
(47) Schubert, U.; Ho¨rnig, H. J. Organomet. Chem. 1987, 336, 307-
315.
(40) Rowsell, B. D.; McDonald, R.; Cowie, M. Organometallics 2004,
23, 3873-3883.
(48) Schubert, U. J. Organomet. Chem. 1988, 358, 215-228.

4188 Organometallics, Vol. 24, No. 17, 2005
Hersh and Fong
forward and reverse direction, for both silyl45,46 and
alkyl47,48 migration.
removing the sample from the water bath. The NMR tube was
cooled in water immediately after removing from the bath and
centrifuged prior to recording each NMR spectrum. All data
are collected in Table 1.
Conclusion
Kinetic runs of the decomposition of 1a were analyzed by
monitoring the disappearance of the combined MeCp singlets
of the cis and trans isomers at δ 1.84 and 1.78, relative to the
methine resonance of Ph3CH at δ 5.41. Crossover experiments
were analyzed by monitoring the appearance (or lack thereof)
of the CH3 doublet of 2b at δ 0.27 (JPH ) 6.7 Hz). At 200 MHz,
this doublet overlaps that due to 2a (δ 0.30, JPH ) 6.4 Hz),
giving the appearance of a triplet when the two are present
in equal amounts. The extent of methyl exchange given in
Table 1 is equal to the ratio of product to its equilibrium
concentration and can be calculated according to eq 5, where
We have discovered a series of heterodinuclear alkoxy-
carbyne complexes that undergo a novel migration
reaction in which the alkyl group migrates from oxygen
to iron.4-8,14 In the study reported here, we have carried
out our most extensive series of control experiments to
uncover the mechanism of this migration and the
concomitant methyl scrambling that occurs and coupled
that with the first detailed kinetic analysis of the
results. The computer methods used are applicable to
a broad range of organometallic mechanisms, and in this
case are necessary since the systems are too complex
to quantitatively determine if a given mechanism really
“fits”. In fact, the common wisdom that any mechanism
can be fit, given sufficient numbers of steps and rate
constants, seems to meet its match in this carbyne
decomposition, where the most surprising conclusion is
that no simple “natural” mechanism can yield equal
amounts of the methyl-scrambled products. The pro-
posal that a reactive 16-electron isomer of the venerable
CpFe(CO)CH3 intermediate forms, namely, the terminal
methoxycarbyne CpFetCOCH3, warrants scrutiny, al-
though it is not clear if it can be trapped in any way
that preserves the terminal methoxycarbyne moiety.
Finally, the question that we started with, namely, the
intra- or intermolecular nature of the methyl migration
from oxygen to iron, remains unanswered. We have
shown that the decomposing carbyne mediates methyl
exchange with the final reaction products, but we cannot
tell if the initial migration is to the iron of the terminal
methoxycarbyne or to another iron atom. However, since
we have also shown that the 16-electron intermediate
CpFe(CO)CH3 alone does not mediate methyl scram-
bling, isomerization of CpFetCOCH3 to CpFe(CO)CH3
can occur only after methyl scrambling has occurred,
and so the proposed terminal methoxycarbyne appar-
ently plays a key role in the methyl exchange reaction.
% exchange ) 2b/2be ) 2b/2bt)∞ ) [2b/(2a + 2b)]/
{[MeCp]/([MeCp] + [Cp])}t)0
) [ν3/(ν1 + ν3)]/{[MeCp]/([MeCp] + [Cp])}t)0
(5)
{[MeCp]/([MeCp] + [Cp])}t)0 is the initial percentage of
MeCpFe starting materials, and will therefore also be equal
to the MeCp products at equilibrium (t ) ∞).49 The relative
amounts of 2a and 2b were estimated by comparison of the
heights of the outer two peaks of the overlapping doublets, ν1
) ∼0.33 ppm (2a), ν2 ) ∼0.30 ppm (2a + 2b), and ν3 ) ∼0.27
ppm (2b). Integration, while somewhat qualitative due to the
lack of baseline resolution of the three peaks, clearly showed
that the height of ν2 was not an accurate indicator of its area.
The reaction of 1a and 1b-d3 in which the starting materials
were isolated for mass spectral analysis (run 7) was carried
out at the higher kinetic-run temperature (75 °C) than the
other crossover reactions (65 °C) in order to facilitate calcula-
tion of the halfway point. Methyl exchange was verified, but
then the unreacted starting materials were isolated by column
chromatography on silica using benzene as the eluting solvent.
A red band was collected and the solvent removed by vacuum.
Mass spectra (electron impact, 140 °C source temperature)
were obtained on the red solid at both 70 and 16 eV. At 70 eV,
1a (m/e ) 382) and 1b-d3 (m/e ) 399) were 8-9% of the base
peak due to Cp(MeCp)Fe (m/e ) 200), while crossover peaks
due to 1a-d3 (m/e ) 385) and 1b (m/e ) 396) were not observed,
each having an upper limit of 0.6% of the intensity of m/e )
382, 399; (MeCp)2Fe (m/e ) 214) was ∼77% of the base peak,
and Cp2Fe was not observed. At 16 eV, the carbyne M+ peaks
had the same absolute ion intensity as at 70 eV and were 50%
of the new base peak due to CpFe(µ-CO)(µ-COCH3)MnMeCp
(m/e ) 326); in addition MeCpFe(µ-CO)(µ-COCD3)MnMeCp
(m/e ) 343) was 70% of the base peak. No crossover peaks
due to 1b (<0.7% of m/e 396 + 397) or MeCpFe(µ-CO)(µ-
COCH3)MnMeCp (m/e ) 340) were observed, while a peak at
m/e ) 385, the same as crossover product 1a-d3, was 1.9% of
m/e ) 382 + 383 (the M+ and M+ + 1 peaks of 1a), and a peak
at m/e ) 329, the same as crossover peak CpFe(µ-CO)(µ-
COCD3)MnMeCp, was 2.2% of m/e ) 326 + 327 (the M+ and
M+ + 1 peaks of the base peak).
Experimental Section
General Procedures. Experimental methods were the
same as those described in detail previously.5 Compounds 1a,
+
1b-d3, 3a-Na+, 3a-CH3PPh3+‚CH2Cl2, and 3b-CH3PPh3
‚
CH2Cl2 were prepared as we have previously described,5 and
methods and data for 2a, 2b, and 2b-d3 have been reported
previously.5,6 Triphenylphosphine was recrystallized from
ethanol, Ph3CH was recrystallized from hexane, and CH3-
PPh3+Br- (Aldrich) and MeCpMn(CO)3 (Aldrich) were used as
received.
1H NMR Reactions. General. In the glovebox, reactants
were loaded into an NMR tube that had been sealed to a 14/
20 ground glass joint, and Ph3CH was added as an internal
NMR integration standard. For runs 1-6, the tube was fitted
with a vacuum stopcock, attached to a vacuum line, and
evacuated, and C6D6 was then added by vacuum transfer. The
tube was submitted to two freeze-pump-thaw cycles and then
sealed with a torch. For runs 7-10 the C6D6 was added in the
glovebox. The tubes were then attached to the vacuum line as
above. The samples were heated (the NMR tubes were
inverted) in a thermostated constant-temperature water bath.
The volume (in mL) used to calculate the concentrations was
determined according to the formula V ) π(0.213)2h, where h
is the height in cm of the solution measured immediately after
Computer Modeling. The Fortran program CRK2005 is
designed to allow a user to fit a set kinetic data to a set of
differential equations; while the size of the problem can be
increased without difficulty, the version in the Supporting
Information is dimensioned for five kinetic runs, each having
at most 30 time points, and a maximum of 30 variables (rate
constants and variable initial reactant concentrations), 30
species with initial concentrations, and up to 11 species with
observed concentrations specified at each time point. The
program reads in data from user-generated files that contain
information on the number of (1) differential equations, (2)
(49) Moore, J. W.; Pearson, R. G. Kinetics and Mechanism, 3rd ed.;
Wiley-Interscience: New York, 1981.

Crossover Studies of Methyl Migration
Organometallics, Vol. 24, No. 17, 2005 4189
observed species, (3) “steady-state” intermediates and their
maximum concentration as a percent of reactant concentration,
(4) rate constants, (5) reactant concentrations whose initial
values will be allowed to be varied, and (6) kinetic runs,
followed by for each run the number of time points, final time,
and standard deviation for each concentration (see below).
Initial guesses of rate constants are input, as are the identities
of the reactant concentrations that will be varied and by how
much; a separate data file of all initial reactant concentrations
(including for instance all intermediates which are initially
zero, and the intial value of [PPh3]) followed by values of
observed data points at each time is also input. The differential
equations written in Fortran are included as a separate
subroutine, and so the program must be compiled for each
mechanism to be tested. The initial rate constants and reactant
concentrations are then used to numerically integrate the rate
equations out to the specified final time, using a published
routine for fourth-order Runge-Kutta integration using adap-
tive step-size control in order to allow integration of rate
constants that vary by more than 4 orders of magnitude
(described as Runge-Kutta fifth-order quality-controlled
steps).19,50 A subroutine then uses these integrated results to
determine the calculated concentrations of each observed
species at each time point. These calculated values are then
compared to the observed values to give the error for the initial
set of variables. The error function that was minimized is the
reduced chi square function øν2 (eq 6),51 where n is the number
and calculated concentrations, respectively, and σi is the
estimated error in units of concentration for each run. This
error estimate was computed on the basis of estimates in errors
of weighing reactants, measurement of solvent volume, and
NMR integration, and was assumed to be constant for each
run. In practice, a “perfect” fit that involves only statistical
2
error equal to this estimate of σi will give øν ≈ 1. A value of
øν2 much less than 1 indicates that σi has been overestimated,
while a value of øν2 much greater than 1 indicates that σi has
been underestimated or more likely that the mechanistic model
is poor. To allow this statistical value to be compared to a more
intuitive error estimate, a percent error was also calculated
(eq 7) by dividing the standard deviation of the calculated
([ci(obs) - ci(calcd)]2)1/2
n
n
% error )
ci(obs)/n (7)
∑
∑
(
)
ν
/
i)1
i)1
concentrations by the average concentration. Finally, the
sensitivity of øν to changes in each rate constant was
2
determined by calculating the change in each rate constant
2
required to give a 1% increase in øν , and this is reported as
the deviations in the rate constants in Figure 3 and Table S-4
in the Supporting Information. The output of CRK2005
includes a user-set number of calculated points per line for
plotting, and 50 points per line was found to be adequate; the
output was plotted using the program pro Fit 6.0.0.
[ci(obs) - ci(calcd)]2
Acknowledgment. Financial support for this work
from Research Corporation, the UCLA Committee on
Research, the Universitywide Energy Research Group,
the donors of the Petroleum Research Fund, adminis-
tered by the American Chemical Society, NSF Grant No.
CHE-9096105, and the City University of New York
PSC-CUNY Research Award program is gratefully
acknowledged.
n
2
øν
)
ν
∑
/
σi2
(
)
i)1
ν ) n - (number of parameters)
(6)
of nonzero data points, the number of parameters is the
number of rate constants plus the number of initially varied
reactant concentrations, ci(obs) and ci(calcd) are the observed
Supporting Information Available: Kinetic data for
runs 1-6, 9, and 10, table of product yields, data for additional
runs using 3a and 3b, kinetic details for the CpFe(CO)Me
mechanism, differential equations for Scheme 6 and the CpFe-
(CO)Me mechanism, and a listing of the Fortran program
CRK2005. This material is available free of charge via the
Internet at http://pubs.acs.org.
(50) Integration of such “stiff” differential equations in principle can
be done using the Gear integration scheme, but our attempts using
the IMSL, Inc. package were not successful. A package of PC programs
written by T. Beukelman, R. J. McKinney, and F. J. Weigert (PRGEAR,
GEAR, and GIT), available through Project SERAPHIM of the ACS,
does implement the Gear scheme, and we have used this set of
programs to check parts of our mechanisms.
(51) Bevington, P.; Robinson, D. K. Data Reduction and Error
Analysis for the Physical Sciences, 3rd ed.; McGraw-Hill: New York,
2003.
OM0502640