|
SYNTHESIS AND X-RAY STRUCTURE OF
IRON STABILIZED STRAINED CYCLIC ALLENES.
VALENCE ISOMERIZATION BETWEEN LINEAR PERPENDICULAR
AND BENT PLANAR ALLENE
BY
SU-MIN OON
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1987
To my mother and
in memory of my father
ACKNOWLEDGEMENTS
I would like to thank Dr. W.M. Jones for providing an
atmosphere conducive to my personal and intellectual growth
during my stay. He has kept me entertained with a fun and
exciting problem and his guidance and encouragement are
crucial toward the successful completion of this study. He
has attracted many fine individuals who are a pleasure to
work with. I would also like to thank Dr. Palenik and
Dr. Koziol for the X-ray crystal structure of my iron
stabilized cyclic allene. The witty, sarcastic and some-
times acrimonious remarks of Mr. Nicholas Conti made me
laugh at even the worst of my scientific blunders. The most
special thanks go to Miss Margaret Easley; a soprano, a
pianist, an artist and a dear friend. I have spent many
happy moments with her engaged in intriguing conversations
spanning an extremely wide range of topics. She introduced
me to opera and classical music and therefore made an
everlasting impact on my life. Her friendship is greatly
appreciated and she will always be remembered with fondness.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS . . . . . . . . .
LIST OF TABLES . . . . . . . . . .
LIST OF FIGURES . . . . . . . . . .
ABSTRACT . . . . . . . . . . ..
CHAPTER
Page
ix
x
xi
* xi~
I INTRODUCTION . . . . . . . . . 1
II DICARBONYL(h5-CYCLOPENTADIENYL)IRON(II)
COMPLETED 1,2-CYCLOHEPTADIENE . . . ... 16
III CARBONYL(h5-CYCLOPENTADIENYL)TRIPHENYLPHOSPHINE-
IRON(II) COMPLETED 1,2-CYCLOHEPTADIENE .... .61
IV IRON(II) COMPLETED 1,2-CYCLOHEXADIENE . . .. .98
V EXPERIMENTAL . . . . . . . ... .108
Improved synthesis ?f dicarbonyl(h -cyclo
pentadienyl)(h -(7-methoxy)cyclohepten-1-
yl)iron(II) . . . . . . .. 109
Thermal decomposition ot dicarbonyl(h -1,2-
cycloheptadiene)(h -cyclopentadienyl)
iron(II) trifluoromethanesulfonate .... .109
NMR scale synthesis of dicarbonyl(h -cyclo
pentadienyl)iron(II) trifluoromethane-
sulfonate . . . . . . . . 110
Attempted thermal decomposition of dicarbonyl
(h -1,2-cycloheptadiene)(h -cyclopentadienyl)
iron(II) tetrafluoroborate . . . . .. 110
13
Variable temperature C NMR studies of the
dicarbonyl(h -1,2-cycloheptadiene)(h -
cyclopentadienyl)iron(II) tetrafluoro-
borate . . . . . . . . . 110
Page
2
The reaction between dicarbonyl(h -1,2-cyclo-
heptadiene)(h -cyclopentadienyl)iron(II)
tetrafluoroborate and triphenylmethane . 111
Attempted enantio eric enrichment of race ic
dicarbonyl(h -1,2-cycloheptadiene)(h -
cyclopentadienyl)iron(II) tetrafluoroborate
with (S)-(-)-2-methylbutanol . . ... Ill
Resolution of 7-methoxy-l-cycloheptenecarboxylic
acid . . . . . .... . . . 112
Synthesis of optically active dicarbonyl(h -
cyclopentadienyl)(h -(7-methoxy)cyclo-
hepten-l-yl)iron(II) (N) . . . ... .112
Synthesis of optically active 1-bromo-
7-methoxycycloheptene . . . ... 113
Synthesis of optically aTtive dicarbonyl(h -
cyclopentadienyl)(h -(7-methoxy)cyclohepen-
1-yl)iron(II) (P) . . . . . . 114
Synthesis of dicarbonyl(h -cyclopentadienyl)
(h -(7-methoxy)cycloheptenl1-yl)
iron(II) from5dicarbonyl(h -1,2-cyclo-
heptadiene)(h -cyclopentadienyl)iron(II)
trifluoromethanesulfonate . . . ... 114
Synthesis of dicarbonyl(l -carbonyl-(7-methoxy)
cyclohepten-1-yl) (h -cyclopentadienyl)
iron(II) via carbonylation of dicarbonyl
(h -cyclopentadienyl)(h -(7-methoxy)
cyclohepten-1-yl)iron(II) . . . ... .114
Synthes s of carbonyl(h5-cyclopentadienyl)
(h -(7-methoxy)cyclohepten-l-yl)triphenyl-
phosphine iron(II) . . . . . ... 115
Synthes s of carbonyl(h2-1,2-cycloheptadiene)
(h -cyclopentadienyl)triphenylphosphine-
iron(II) trifluoromethanesulfonate .... .116
Synthesis of l-bromo-7-methoxycycloheptene-7d1 117
Synthesis of carbonyl(h -1,2-cycloheptadiene-
ld ) (h -cyclopentadienyl)triphenylphos-
phineiron(II) trifluoromethanesulfonate 117
Page
The exchange of hexafluorophosphate anion for
tr fluoromethanesulfonate anion in carbonyl
(h -1,2-cycloheptadiene)(h -cyclopenta-
dienyl)triphenylphosphineiron(II) . .. .118
Synthesis of carbonyl(h 5carbonyl-(7-methoxy)
cyclohepten-l-yl)(h -cyclopentadienyl)
triphenylphosphineiron(II) . . . ... .118
Separation of diastereomers of carbonyl(h -
cagbonyl-(7-methoxy)cyclohepten-l-yl)
(h -cydopentadienyl)triphenylphosphine-
iron(II) . . . . . .... . . 119
Attempted decarbonylation of carbonyl(h -
carbonyl-(7-methoxy)cyclohepten-1-yl)
(h -cyclopentadienyl)triphenylphosphine-
iron(II) . . . . . .... . . 120
Separation of diastereomer of carbonyl(h -cyclo-
pentadienyl)(h -(7-methoxy)cyclohepten-1-
yl)triphenylphosphineiron(II) . . ... 121
Synthes s of carbonyl(h2-1,2-cycloheptadiene)
(h -cyclopentadienyl)triphenylphosphine
iron(II) trifluoromethanesulfonate from the
enriched carbonyl(h -cyclopentadienyl)
(h -(7-methoxy)-cycloheptne-1-yl)triphenyl-
phosphineiron(II) (80:20) . . ... 122
Reaction between carbonyl(h5-cyclopentadienyl)
(h -(7-methoxy)cyclohepten-l-yl)triphenyl-
phosphineiron(II) with a half equivalent
of trimethylsilyl trifluoromethane-
sulfonate . . . . . . . . 122
Synthes s of carbonyl(h -1,2-cycloheptadiene)
(h -cyclopentadienyl)triphenylphosphine
iron(II) trifluoromethapesulfonate from
th enriched carbonyl(h -cyclopentadienyl)
(h -(7-methoxy)cyclohepten-1-yl)triphenyl-
phosphineiron(II) (15:85) . . . ... .123
Synthes s of dicarbonyl(h -cyclopentadienyl)
(h -(7-ethoxy)cycloheptadien-l-yl)
iron(II) . . . . . . . ... 123
Synthesis of carbonyl(h -cyclopentadienyl)
(h -(7-ethoxy)cycloheptadien-1-yl)
triphenylphosphineiron(II) . . . ... .123
Page
Separat on of the diastereom r of carbonyl
(h -cyclopentadienyl)(h -(7-ethoxy)cyclo-
heptadien-1-yl)triphenylphosphineiron(II). 124
2
Synthes s of carbonyl(h -1,2-cycloheptadiene)
(h -cyclopentadienyl)triphenylphosphine
iron(II) tgifluoromethanesulfonate from
carbonyl(h -cyclopentadienyl)(h -(7-
ethoxy)cyclohepten-1-yl)triphenylphosphine
iron(II) . . . . . . . . . 125
Epimerization of carbonyl(h2-1,2-cycloheptene)
(h -cyclopentadienyl)triphenylphosphine
iron(II) hexafluorophosphate in the
presence of triphenylphosphite ...... 125
2
Epimerization5of carbonyl(h -1,2-cyclohepta-
diene)(h -cyclopentadienyl)triphenylphos-
phineiron(II) trifluoromethanesulfonate
in the presence of triphenylphosphine . 126
Synthes s of dicarbonyl(h -cyclopentadienyl)
(h -(6-methoxy)cyclohexen-l-yl)iron(II) 126
2
Attempted synthesis of dicarbonyl(h -1,2-
cyclohexadiene)(h -cyclopentadienyl)iron
(II) trifluoromethanesulfonate ...... 127
2
Attempted synthesis of dicarbonyl(h -1,2-
cyclohexadiene)(h -cyclopentadienyl)
iron(II) tetrafluoroborate . . . ... 127
Low temperature NMR scale synthesis of 5
dicarbonyl(h -1,2-cyclohexadiene)(h -
cyclopentadienyl)iron(II) trifluoromethane-
sulfonate . . . . . . . . 128
Synthesis of carbonyl(h -cyclopentadienyl)
(h -(6-methoxy)cyclohexen-1-yl)triphenyl-
phosphineiron(II) . . . . . .. 129
Attempted synthesis of carbonyl(h -1,2-
cyclohexadiene)(h -cyclopentadienyl)
triphenylphosphineiron(II) trifluoro-
methanesulfonate . . . . . ... .129
Reaction of carbonyl(hl-(6-methoxy)cyclohexen-l-
yl)(h -cyclopentadienyl)triphenylphosphine-
iron(II) with trimethylsilyl trifluoro-
methanesulfonate followed by additions of
ethanol . . . . . .... . . 130
Page
APPENDIX . . . . . . . . . . . 132
REFERENCES . . . . . . . . ... . .133
BIOGRAPHICAL SKETCH . . . . . . . . .. .137
viii
LIST OF TABLES
Table Page
1. Geometries of cyclic allenes calculated
by MNDO . . . . . . . . . . 5
2. Methoxy abstraction from hl-(7-methoxy)
cyclohepten-l-yl Fpp (34) to give
h -1-2-cycloheptadiene Fpp (35a) . . . . 82
LIST OF FIGURES
Figure Page
1. Thermal decomposition of (4a) in CD2Cl2 .. 19
2. Fp trifluoromethanesulfonate and 1,3-
cycloheptadiene from the thermal decomposition
of (4a) in CD2C12 .............. 20
3. High temperature 1C NMR of (4b) in CD3NO2 . 31
4. A CDC1 solution of 10 mole % Eu(hfc) and (32)
from ( ) showing a 25:75 diastereomeric
composition . . . . . . . . . 56
5. A CDC1 solution of 10 mole % Eu(hfc) and (32)
from methoxy abstraction and addition of (N)
showing a 50:50 diastereomeric composition 57
6. A CDC1 solution of 10 mole % Eu(hfc) and (32)
from (P) showing a 60:40 diastereomeric
composition . . . . . . . . . 58
7. A CDC1 solution of 10 mole % Eu(hfc) and (32)
from mathoxy abstraction and addition of (P)
showing a 50:50 diastereomeric composition 59
8. 1H NMR of (36) in CDC1 showing a 20:80
diastereomeric composition . . . . .. 69
9. 1H NMR of (34) in CDC1 showing a 85:15
diastereomeric composition . . . ... 72
10. 1H NMR of (34) in CDC1 showing a 10:90
diastereomeric composition . . . ... 73
11. IH NMR of (35a) in CDC1 .......... 78
12. H S.S.T. of (35a) in CDC13 . . . . . 91
13. X-ray structure of h2-1,2-cycloheptadiene Fpp
cation . . . . . . . . . . 97
14. Reaction between h -(6-methoxy)cyclohexen-l-yl
Fp and trimethylsilyl trifluromethanesulfonate
followed by low temperature H NMR . . . 104
Abstract of Dissertation Presented to the Graduate
School of the University of Florida in Partial
Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
SYNTHESIS AND X-RAY STRUCTURE OF
IRON STABILIZED STRAINED CYCLIC ALLENES.
VALENCE ISOMERIZATION BETWEEN LINEAR PERPENDICULAR
AND BENT PLANAR ALLENE.
By
Su-Min Oon
May 1987
Chairman: William M. Jones
Major Department: Chemistry
The synthesis and isolation of strained organic mole-
cules has been an area of active research for over
forty years. Strained cyclic allenes are of no exception
and have been studied both theoretically and experimentally.
Calculations have shown that strained cyclic allenes down to
1,2-cyclopentadiene remained twisted and chiral but with a
2-5 kcal/mole barrier for racemization. Experimentally,
1,2-cyclohexadiene was found to be chiral but with a very
low barrier for racemization.
Extended Huckel molecular orbital (EHMO) calculations
have shown that the racemization barrier of the twisted
1,2-cycloheptadiene via its bent planar allyl cation is
lowered considerably upon complexation with a transition
metal.
We have synthesized and studied the racemization
process of dicarbonyl(h5-cyclopentadienyl)iron(II) [Fp] and
carbonyl(h5-cyclopentadienyl)triphenylphosphineiron(II)
[Fpp] completed 1,2-cycloheptadiene.
Although we have failed to observe the racemization of
Fp completed 1,2-cycloheptadiene directly, we have experi-
mental evidence which indicates that the bent planar allyl
cation was formed either via methoxy abstraction from the
hl-(7-methoxy)cyclohepten-l-yl Fp leading to the formation
of h2-1,2-cycloheptadiene Fp or from the isomerization of
h2-1,2-cycloheptadiene Fp itself. We have also determined
that the allene to allyl cation isomerization must be at
least 14.7 kcal/mole.
The Fpp completed 1,2-cycloheptadiene was synthesized
as enantiomeric pairs of diastereomers. Its fluxional
barrier for 1,2-Fpp shift is between 15 kcal/mole and
18 kcal/mole. An enriched mixture of one diastereomer
slowly equilibrates at room temperature to its thermodynamic
mixture. The iron center here is stereochemically rigid and
epimerization through dissociation and reassociation of the
triphenylphosphine ligand does not occur. We infer from our
observation that the thermal equilibration occurs via the
allyl cation. Methoxy abstraction from the h -(7-methoxy)
cycloheptadien-1-yl Fpp proceeds via the allyl cation as in
the case with h -(7-methoxy)cycloheptadien-1-yl Fp.
We also report the first X-ray crystal structure of the
h -1,2-cycloheptadiene Fpp cation.
Finally, we were able to synthesize and isolate the
h2-1,2-cyclohexadiene Fpp complex as a thermally unstable
solid. This solid reacted with ethanol, in a manner typical
of most metal olefin complexes, to give hl-(6-ethoxy)cyclo-
hexen-l-yl Fpp.
xiii
CHAPTER I
INTRODUCTION
Allenes are a class of organic compounds which contain
two cumulated double bonds arranged in an orthogonal
geometry. Acyclic allene has a linear structure and the
planes defined by R1R2C1 and R3R4C3 are mutually perpendicu-
lar. R3
R2
R4
One of the most fascinating problems both experimental-
ists and theoreticians have been concerned with regarding
the allene structure is the energy gap separating the ground
state linear perpendicular allene (A) from its excited bent
planar valence isomer (B). 9
A B
Equally interesting is the extent to which this energy gap
can be lessened and perhaps be inverted such that the bent
planar allene becomes the ground state.
One can approach this problem by either increasing the
ground state energy of the linear perpendicular allene or
decreasing the excited state energy of the bent planar
allene or both.
Incorporating a linear perpendicular allenic unit into
a small ring will deform this allenic unit in two ways in
order to facilitate ring closure and reduce ring strain.
The allene will bend at the C2 carbon about an axis perpen-
dicular to the R R2C plane as defined by 8. The allene will
also twist about the R1R2C1 and the R3R4C3 planes as defined
by 1.
This combined effect raises the ground state energy of the
allenic unit and is reflected by its kinetic instability.
1,2-Cyclononadiene1 is a distillable liquid while 1,2-cyclo-
octadiene2 dimerizes within hours at room temperature.
1,2-Cycloheptadiene1 and 1,2-cyclohexadiene3 have only a
fleeting existence and can only be trapped chemically or, in
the second case, by trapping in an argon matrix.
Although cyclic allenes lack an asymmetric carbon, they
are chiral molecules with a C2 point group. Their R and S
configuration, using 1,2-cycloheptadiene as an example, are
shown below.4
H .H H H
R S
In order for one enantiomer to convert to the other,
the allenic pi-bonds of that enantiomer must be rotated out
of orthogonality, passing through a planar state to the
other enantiomer. The smaller the ring size of the cyclic
allene the greater the allene is bent and twisted towards
its planar isomer. When ring constraints force the pi-bonds
of an optically active cyclic allene (1) to rotate out of
orthogonality and become planar (2), optical activity is
lost.
H*H, H H H H
CH2)n (CH2)n CH2)n
1 2
Racemization of an optically active cyclic allene is suffi-
cient proof that a planar intermediate is accessible. The
energy barrier for racemization can be measured via the loss
of optical activity and represents the minimum energy
separating the two forms of the allene, (A) and (B).
Numerous calculations have been performed over the last
20 years on 1,2-propadiene and 1,2-cyclohexadiene, the
archetype cyclic allene, in order to estimate this barrier
for racemization and to address the electronic nature of the
bent planar allene. The ground state geometry for
1,2-cyclononadiene, 1,2-cyclooctadiene and 1,2-cyclohepta-
diene are accepted generally as slightly bent and twisted
towards planarity. The results from the latest MNDO calcu-
lations5 are given in Table 1.
Allenic geometries for this series of three cyclic
allenes clearly show the effect of ring strain on the
allenic unit. An X-ray crystal structure is available for a
phenylurethane derivative of 1,2-cyclononadiene6 and agrees
closely with the calculated results.
Historically, there have been considerable doubts as to
the ground state structure of 1,2-cyclohexadiene. Moore and
Moser7 predicted the singlet diradical as the ground state.
INDO calculations by Dillon and Underwood suggested that
the triplet diradical might be lower in energy. Recent SCF
calculations of Johnson favor a chiral allenic ground state
structure for 1,2-cyclohexadiene.5 The calculations further
predict a chiral ground state structure for 1,2-cyclopenta-
diene but with a 2-5 kcal/mole barrier for racemization.
Mechanistic studies of 1,2-cyclohexadiene by Bottini et
al. suggested that 1,2-cyclohexadiene has an allenic
structure but rapidly racemizes to a singlet diradical form.
He also showed some chemistry arising from the zwitterionic
form of 1,2-cyclohexadiene.
Table 1. Geometries of cyclic allenes calculated by
MNDO.
C1-C2-C3 out of plane H benda
1,2-cyclononadiene 170.40 33.70
1,2-cyclooctadiene 161.5 31.00
1,2-cycloheptadiene 153.40 27.60
defined as the angle made by the C-H bond with the C -C2-C3
plane containing the C2 axis, e.g., 450 for 1,2 propadiene.
Optically active 1,2-cyclohexadiene, synthesized via
asymmetric dehydrobromination of resolved 1-bromocyclo-
hexene-6d1 with potassium tert-butoxide, was trapped in situ
with diphenylisobenzofuran.0 The adduct (3) is optically
active when the reaction is done at low temperatures but
optically inactive above +800C suggesting that racemization
of the cyclic allene is competitive with cycloaddition and
the inversion barrier is low.
0
Br
H K t-OBu > 0
at low temperatures 3
H D
The adduct from reaction of 1,2-cycloheptadiene and
diphenylisobenzofuran remained optically active under
similar reaction conditions. The allene subunit in this
larger ring has maintained its orthogonality.
We have now extended our studies from the free strained
cyclic allenes to the transition metal completed strained
cyclic allenes. We are interested in the effect the transi-
tion metal has on the allenic unit, in particular with
regard to the isomerization of the allene unit from its
linear perpendicular form to its bent planar isomer.
EHMO calculations predict that the energy gap
separating the linear perpendicular allene from its bent
planar isomer is lessened considerably when the allene is
completed onto the (Cp)Fe(CO)2 [Fp] cationic fragment. Here
lies the impetus to study Fp completed allenes.
1,2-Propadiene lies about 80 kcal/mole below that of
its bent planar isomer according to EHMO. The relative
energy between 1,2-propadiene and its bent planar isomer is
only 25 kcal/mole when both are bound to the Fp fragment.
Fp
80 kcal/mole 25 kal/mole
Fp
Similarly, Fp bound 1,2-cycloheptadiene (4) is cal-
culated to be only 17 kcal/mole below its bent planar form
(5). Furthermore, EHMO predicts the Fp bound bent planar
1,2-cyclohexadiene (7) is actually 14 kcal/mole more stable
than the allene form (6).
Fp
Fp+
5 6
17 kcal/mole 14 kcal/mole
Fp+ Fp
4 7
Although energies derived from EHMO should not be
accepted as absolute values, one does see a trend favoring a
bent planar allenic ground state as one proceeds towards Fp
completed allene with the allene incorporated into a smaller
ring. This decrease in relative energy between the linear
perpendicular allene and its bent planar isomer is largely
due to the greater stabilizing interaction between the
Fp-LUMO and the bent planar allenic HOMO.
We decided to choose the Fp completed 1,2-cyclohepta-
diene as the starting point of our work.
Transition metal complexes of acyclic allenes are well
known and have been reviewed extensively.12 However cyclic
counterparts are limited to Fp complexes of 1,2-cyclonona-
diene13 (8) and 1,2-cycloheptadienel4 (4), (PPh3)2Pt(O)
complexesl5 of 1,2-cyclononadiene (9), 1,2-cyclooctadiene
(10) and 1,2-cycloheptadiene (11) and (Am)(Cl)2Pt(II)
complex of 1,2-cyclononadienel6 (12).
Fp+ Pt(O)
sFop+'
8 4 9
Pt(lH)
Pt(0) P(0)
10 11 12
The bonding between the metal fragment and the allene
can be described using the Dewar-Chatt-Duncanson model.17
The bond is formed by the interaction of the HOMO of one of
the allene double bonds with an empty acceptor on the metal
(13). The filled metal d-orbital in turn back bonds with
the LUMO of the same double bond (14).
1c
H H
HOMO
13
H H
C
LUMO H H
LUMO
40'
HOMO
HOMO
This back bonding causes substituents on the completed
double bond to bend away from the metal, the degree of which
is directly related to the extent of this back bonding. The
metal to olefin pi-bond is often looked upon as a metallo-
cyclopropane.18 The metal is not positioned symmetrically
about the C1 and C2 carbons of the allene. The metal is
closer to the C2 carbon because of its greater s character
compared to the C1 carbon (15).
/"
H
H C2
I| M
C1
15
Allene complexes of metals, like olefin complexes of
metals, rotate about the metal to olefin pi-bond. In
addition to this rotational motion, allene complexes of most
metals also exhibit fluxional behavior whereby the completed
metal fragment moves from one double bond of the allene to
the adjacent double bond. Such fluxional behavior of
acyclic allenes completed to Fp was studied extensively by
Rosenblum et al.13 and was shown to be nondissociative (i.e.
intramolecular).
Fp+
\=.^'
)Fp+VA
Unlike the Fp completed allenes, not all platinum
completed allenes are fluxional. The Pt(0) complex of
19
1,3-diphenylpropadiene (16) is not fluxional.1 Ha and Hb
remain distinct in the IH NMR even at high temperatures.
Ha
c/
I I
C
-H-H Pt
C
Ph Hb
PPh3
PPh3
The Pt(II) complex of
with the metal moving
double bonds.20
double bonds.
tetramethylallene however is fluxional
back and forth between the two allenic
1,2-Fp shifts do not involve a bent planar intermedi-
ate. Optical rotation of an optically active 1,2-cyclo-
nonadiene completed onto Fp remained unchanged3 up to
1.2 Fp shift
+800C. A racemate would result if a bent planar intermedi-
ate were involved. In contrast, the optical activity of the
Pt(II) complex of 1,2-cyclononadiene (12) diminishes with
time. The 1,2-cyclononadiene is believed to racemize via a
h3-allyl Pt intermediate (18) and not the bent planar
allene16 (17).
Pt(Am)(CI)2 CI
SPt(Am)(CI)2 Pt(Am)(CI)
17 12 18
1,2-Cycloheptadiene completed to Fp has a measured
fluxional barrier of 13.9 kcal/mole and is the lowest
14
barrier observed for Fp allene complexes. Fluxional
barriers for h2-methylallene, h2-1,1-dimethylallene and
h2-tetramethylallene Fp cationic complexes are
23.1 kcal/mole, 18.0 kcal/mole and 16.3 kcal/mole, respec-
tively.13 h2-1,2-Cyclononadiene Fp has a fluxional barrier
of 16.9 kcal/mole. Although the decrease in fluxional
barrier with increasing methyl substitution is probably
steric in origin, one could not help but to wonder if a bent
planar allene intermediate were involved. The bent planar
allene is an allyl cation and should be favored by electron
donating methyl substituents. Moreover, if one assumes that
the steric environment immediate to the Fp center is the
same for 1,3-dimethylpropadiene, 1,2-cyclononadiene and
1,2-cycloheptadiene, bending and twisting the linear perpen-
dicular allenic unit toward its bent planar form by tying it
into a ring lowers the fluxional barrier. Therefore, it was
not known at the time it was reported whether the fluxional
barrier of 13.9 kcal/mole for h2-1,2-cycloheptadiene Fp (4)
represented a simple 1,2-Fp shift or that an allyl cation
intermediate (5) was involved.
Fp
Fp+ Fp+
allene to allyl 1,2-Fp shift
cation 4
5 4
X-ray crystal data is available for many allene com-
plexes of Rh, Pt and Pd.21 In all cases the completed
double bond is longer than the uncomplexed double bond. The
metal is unsymmetrically positioned about the bound double
bond and is further away from the terminal carbon than the
central carbon. The C1-C2-C3 bond angle falls within the
range of 1580 to 1420. The completed allenes are not
planar. Allene complexes of transition metal dimers (19)
where each of the allene double bonds is bound to a
different metal in the dimer are also known. The C1-C2-C3
angles of such complexes are, on the average, much
smaller.22
LnM MLn
19
The only X-ray structure of an iron allene complex is the
h2-tetramethylallene Fp.23 The completed allene is oriented
parallel to the cyclopentadienyl ring as expected. The bond
length of the uncomplexed double bond is 1.335A and the
completed double bond is 1.367A. The completed double bond
is longer as a result of back bonding from the Fp to the pi*
of the allene double bond. The iron is asymmetrically
placed about the double bond and is closer to the central
0 0
carbon (2.063A) than the outer carbon (2.237A). The
C1-C2-C3 bond angle is 145.7 but the allene is still
orthogonal.
There have been no X-ray data reported on any metal
completed cyclic allenes to date although such compounds are
well known. It would be interesting to compare the X-ray
structure of h2-1,2-cycloheptadiene Fp with h2-tetramethyl-
allene Fp. Of particular importance is how the ring affects
the allene C1-C2-C3 bond angle and how far the allene is
away from orthogonality.
The objective of this study was to determine whether a
1,2-Fp shift between the double bond of the cyclic allene
15
involves a bent planar intermediate and, if not, to
determine if the latter is accessible. It was also to
provide the first X-ray structure of a metal completed
cyclic allene.
CHAPTER II
DICARBONYL(h5-CYCLOPENTADIENYL)IRON(II) COMPLETED
1,2-CYCLOHEPTADIENNE
The h1-(7-methoxy)cyclohepten-l-yl Fp was synthesized
in moderate yields by reacting the l-lithio-7-methoxycyclo-
heptene with either Fp chloride or Fp bromide. Treating a
methylene chloride solution of the hl-(7-methoxy)cyclo-
hepten-l-yl Fp with trimethylsilyl trifluoromethanesulfonate
followed by precipitation with diethyl ether yielded the
desired h2-1,2-cycloheptadiene Fp cation complex (4a) as an
air and thermally sensitive yellow solid.
Br Fp
OCH3 OCH3 Fp
1) n-BuLi TMSOTF OTF
2)FpClorFpBr
4a
Attempts to synthesize the h -(7-methoxy)cyclohepten-l-yl Fp
by this route failed with Fp iodide. It is critical that
the l-bromo-7-methoxycycloheptene be in a slight excess
relative to n-butyllithium when generating the l-lithio-7-
methoxycycloheptene in order to ensure a maximum yield of
the hl-(7-methoxy)cyclohepten-l-yl Fp. Synthesis
of h1-(7-methoxy)cyclohepten-l-yl Fp by this method was far
superior to a longer and lower yield route used pre-
viously.14
The yellow colored methylene chloride solution of the
h2-1,2-cycloheptadiene Fp trifluoromethanesulfonate (4a)
turned red within 30 mins at room temperature. In contrast,
the h2-1,2-cyclononadiene Fp tetrafluoroborate (8) is not
only air stable as a solid but a methylene chloride solution
of it is also stable for over 14 hrs at +830C. Although the
extreme thermal instability of the h2-1,2-cycloheptadiene Fp
(4a) in methylene chloride is an inconvenience, it may
signal a different pathway for the fluxional process; a
pathway that is different from that of the h2-1,2-cyclo-
nonadiene Fp (8) and perhaps involves the bent planar allene
(allyl cation) intermediate from which decomposition may
occur. We will now refer to (4) as the allene form and (5)
as the allyl cationic form.
Fp
Fp+
4 5
1H NMR was used to follow the decomposition of the
h2-1,2-cycloheptadiene Fp (4a) in a methylene chloride
solution at +400C. Decomposition was complete within 3 hrs
and Fp trifluoromethanesulfonate and 1,3-cycloheptadiene
were the only major decomposition products observed. The
former was identified by comparing its 1H NMR with that of
Fp trifluoromethanesulfonate synthesized independently from
Fp chloride and silver trifluoromethanesulfonate and the
latter with an authentic sample of 1,3-cycloheptadiene. The
decomposition mixture was red in color. The result of this
experiment is in agreement with that by Manganiello14
Neither cycloheptene nor the 1,2-cycloheptadiene dimer were
observed in the H NMR.
According to the 1H NMR, the decomposition of the
h -1,2-cycloheptadiene Fp (4a) to Fp trifluoromethane-
sulfonate and 1,3-cycloheptadiene proceeded via a short
lived, thermally unstable intermediate. This intermediate
had a singlet at 5.58 ppm, Fig. 1. Its concentration
increased rapidly up to a certain point and then decreased
to the baseline with a concurrent increase in the concen-
tration of both Fp trifluoromethanesulfonate and 1,3-cyclo-
heptadiene, Fig. 2. Based upon this observation, the
thermally labile intermediate was suspected to be the
h2-1,3-cycloheptadiene Fp cation complex. The singlet at
5.58 ppm is certainly within the range for cyclopentadienyl
hydrogen resonances of cationic Fp-olefin complexes. The
h2-1,3-cycloheptadiene Fp could not be synthesized via a
common thermal exchange reaction between 1,3-cycloheptadiene
and the h2-propene Fp cation,24 which lends further evidence
for its thermal liability.
C4-
r.
0
4-)
.rA
ull
0
-4
E-
4-)
0
4
4-4
4-J
0
>1
4-4
1-
QI)
E
04J N
0
r-4 -HJ
-r~
Wa
P4J
=
-dO.
~rd
(N
H
A reasonable mechanism for the decomposition of the
h2-1,2-cycloheptadiene Fp (4a) is as follows.
FpFp Fp
S+ HOTF )
4a 20
Fp Fp+
FpOTF +0
5
We believe that an initial deprotonation of an appro-
priately situated hydrogen occurs from either the allene (4)
or allyl cation intermediate (5) to form the h -1,3-cyclo-
heptadien-2-yl Fp and trifluomethanesulfonate acid. In the
presence of trifluomethanesulfonic acid, this hl-1,3-cyclo-
heptadien-2-yl Fp is then converted via the conjugated
carbene (20) to the h2-1,3-cycloheptadiene Fp. Under the
experimental conditions it dissociates to the Fp cation and
1,3-cycloheptadiene.
The decomposition reaction would have stopped after the
first step, in the absence of trifluomethanesulfonic acid,
giving hl-1,3-cycloheptadien-2-yl Fp as the only product.
In order to test the validity of this mechanism, the h -1,2-
cycloheptadiene Fp (4a) was subjected to the same decomposi-
tion conditions but in a slurry of lithium or sodium
carbonate to neutralize the trifluoromethanesulfonic acid.
The decomposition mixture did not turn red but remained
brown in color. The 1H NMR cyclopentadienyl resonance of
this brown substance falls within the range typical for most
hl-alkyl Fp. Presumably the hl-1,3-cyclopentadien-2-yl Fp
was formed, but we were unable to purify it sufficiently for
a positive identification. 1,3-Cycloheptadiene and Fp
trifluoromethanesulfonate were no longer the decomposition
products.
Counter ions have been known to affect the way an
organometallic species behaves in solution, a recent example
being the intramolecular rearrangement of the h2-3-bromo-
propene Fp complex.25
Fp+ Fp+
D H p H F D H
H rH Y
Br Br
The 1,3 shift of a bromine atom exo to the Fp-olefin bond
was greatly accelerated when the counter ion was changed
from trifluoromethanesulfonate to hexafluorophosphate. Ion
pair association in the salt may have been responsible for
such behavior. Caseyet al.26 has also reported that a
solution of the tetrafluoroborate salt of the (Cp)Fe(CO)
(PPh3) isopropylidene complex is more stable than its
trifluoromethanesulfonate salt.
Fe+ C H3
CC
Ph3P I
CH3
Since we believe that the basicity of the trifluoro-
methanesulfonate anion is responsible for the decomposition
of the h2-1,2-cycloheptadiene Fp (4a), we thought it would
be desirable to replace it with a non-basic anion. The
tetraphenylborate anion was successfully exchanged for the
trifluoromethanesulfonate anion in a cold (-350C) methanol
solution. We were disappointed that a methylene chloride
solution of the tetraphenylborate salt of h2-1,2-cyclo-
heptadiene Fp (4) was also thermally unstable and decomposed
within 30 mins at +500C to several unidentifiable pro-
ducts.27
The h2-1,2-cyclononadiene Fp tetrafluoroborate (8) is
the only known stable cyclic allene of iron. If by chance
the thermal stability of (8) were due to the tetra-
fluoroborate anion, this would suggest that the tetrafluoro-
borate anion may be the counterion of choice for the
h -1,2-cycloheptadiene Fp (4).
The h2-1,2-cycloheptadiene Fp tetrafluoroborate (4b)
was synthesized in good yields by reacting the h -(7-
methoxy)cyclohepten-1-yl Fp with trimethyloxonium tetra-
fluoroborate.
Fp*
OCH3
+ (CH3)30BF4 BF4
4b
The yellow solid h2-1,2-cycloheptadiene Fp (4b) turns red
when exposed to air, but a room temperature methylene
chloride solution of it is stable for over 3 days. The same
solution is also stable for over 2 hrs at +400C. Thus it
would appear that the thermal decomposition of the
h2-1,2-cycloheptadiene Fp (4a) in methylene chloride is due
to the basicity of the trifluoromethanesulfonate anion. The
tetrafluoroborate and the hexafluorophosphate salt of
h2-1,2-cycloheptadiene Fp (4) may also be synthesized by
reacting the h -(7-methoxy)cycloheptadien-l-yl Fp with
triphenylcarbenium tetrafluoroborate and hexafluorophos-
phate, respectively. This procedure has a drawback in that
the product is sometimes contaminated with the unreacted
triphenylcarbenium cation.
Fp
OCH3 Fp+
+ Ph3CBF4/PF6 BF4/PF6
4b/c
We now have on hand the stable h -1, 2-cycloheptadiene
Fp (4b) and (4c). Both h2-1,2-cycloheptadiene Fp (4b) and
(4c) react with alcohols to give the same ether adducts as
are given by the h2-1,2-cycloheptadiene -Fp (4a). It was
interesting to note that their fluxional barriers were
unaffected by the counterions.
The h2-1,2-cycloheptadiene Fp (4) has a fluxional
barrier of 13.9 kcal/mole and is the lowest barrier yet
measured when compared with other h2-Fp completed acyclic
and cyclic allenes. Does this low fluxional barrier signal
a new mechanism for the fluxional process, perhaps via the
bent planar allene (allyl cation) intermediate (5),
Scheme I?
Scheme I
Fp
Fp+ Fp Fp+
0-6S
4 5
Or perhaps the fluxional process merely involves a simple
intramolecular 1,2-Fp shift from one allenic double bond to
the other, Scheme II, passing through the allene intermedi-
ate (21)? The low fluxional barrier merely reflects in some
manner the effect of ring strain in the smaller cyclic
allene. Such a simple intramolecular 1,2-Fp shift was
demonstrated conclusively for the h2-1,2-cyclononadiene Fp
(8).
Scheme II
Fp+
Fp + Fp +
4 21
The allene (4) to allyl cation (5) valence isomeriza-
tion for Fp complexes is unprecedented. The racemization of
optically active 1,2-cyclononadiene completed onto Pt(II)
(12) was initially proposed to proceed via an allyl spe-
cies16 (17).
Pt(Am)(CI)2
Pt(Am)(CI)2 Pt(Am)(CI)2
12 17
This allyl species (17), with two electrons in its
pi-system, would be formally equivalent to an allyl cation.
It is now accepted that a reversible h2-pi-allene (12)
to h3-pi-allyl (18) isomerization, whereby a chloride is
transferred from the Pt(II) to the pi-allyl, is responsible
for the racemization.
Pt(Am)(CI) Pt(Am)(CI) Pt(Am)(CI)2
n- -S
12 18
Such a h2-pi-allene to h3-pi-allyl isomerization in conjunc-
tion with a ligand transfer from the metal to the pi-allyl,
although rare for h2-allene Pt(II) complexes, is quite
common among its cogener, the h -allene Pd complexes.2 In
the case of the h2-1,2-cycloheptadiene Fp (4), no such
ligand is available on the Fp for an analogous transfer to
the 1,2-cycloheptadiene ligand. This precludes such a
mechanism for an allene to allyl isomerization.
Numerous barriers to the fluxional behavior for h2-
allene complexes of Pt(II) and Fe(II) have been measured.
Fluxional barriers for h2-tetramethylallene complexes of
Pt(II) are generally low and lie between 7 to 10 kcal/
mole.20,29 The h2-1,1-dimethylallene complex of Pt(II) is
not fluxional. The Pt(II) remained bonded to the C2-C3
olefinic bond, the less hindered bond. The Fp in the
h2-1,1-dimethylallene Fp also prefers the distal position.13
This preference for the distal position is presumed to be
steric in origin. Fluxional behavior has not been reported
for the h2-allene, h2-tetramethylallene and h2-1,3-diphenyl-
allene Pt(0) complexes.19 30 They are presumed to be
2 16
dynamically rigid. The h -1,2-cyclononadiene Pt(0)1 is
also not fluxional at room temperature.
For comparison with the h2-1,2-cycloheptadiene Fp (4),
the fluxional barriers of h2-1-methylallene, h2-1,3-
dimethylallene, h2-tetramethylallene and h2-1,2-cyclo-
nonadiene (8) Fp complexes3 are 23.1 kcal/mole, 18.0 kcal/
mole, 16.3 kcal/mole and 16.9 kcal/mole, respectively.
It has been well established that the fluxional process
for these complexes occurs by a concerted 1,2-Fp shift
between the two allenic double bonds and not via an allyl
cation. The fluxional barriers decrease gradually with
increasing methyl substitution on the allene and are
attributed to a steric factor rather than the stabilization
of an allyl cationic intermediate by the methyl substit-
uents. The steric bulk of the methyl substituents increases
the liability of the Fp-olefin pi-bond making it easier for
the Fp to move between the two adjacent double bonds.
The h2-1,2-cyclononadiene Fp (8) has a fluxional
barrier of 16.9 kcal/mole. In this case it was conclusively
demonstrated that the fluxional process did not proceed via
an allyl cationic species (22). When a solution of the
optically active h2-1,2-cyclononadiene Fp (8) was heated to
the point of rapid fluxionality, reisolation gave the allene
complex without any loss in optical activity.3 The
optically active h2-1,2-cyclononadiene Fp (8) would have
racemized if the achiral allyl cation (22) were involved in
the fluxional process either as an intermediate or as a
transition state.
FpFp Fp
H. Fp H Fp H HFp H
8 22
At first thought one would not expect the h2-1,2-cyclo-
heptadiene Fp (4) fluxional barrier to be much different
from that of the h2-1,2-cyclononadiene Fp (8) if both
proceeded by the same mechanism. In fact, we would expect
the fluxional barrier for the h2-1,2-cycloheptadiene Fp (4)
to be higher should the Fp-olefin pi-bond strength parallel
that of the Pt(0)-olefin pi-bond. The coordinating ability
of cyclic allenes onto Pt(PPh3)2 was reported to increase
with decreasing ring size.15 This is due to the greater
release of ring strain upon coordination. The low fluxional
barrier for the h2-1,2-cycloheptadiene Fp (4) coupled with
the strained ring therefore causes one to wonder if the
fluxionality occurs via a mechanism involving an allyl
cationic intermediate (5).
To test for this possibility, a temperature dependent
13C NMR study of h2-1,2-cycloheptadiene Fp (4b) was under-
taken. 1,2-Cycloheptadiene is chiral and when completed to
Fp causes the carbonyl ligands on the Fp to become
diastereotopic and be distinguishable. The two carbonyl
[CO] resonances appear at 207.2 ppm and 210.2 ppm. A simple
intramolecular 1,2-Fp shift would render the COa and COb
unchanged (23). In contrast an allene to allyl cation
isomerization would cause the carbonyls to become equivalent
and coalesce to a single resonance (24).
Cp Cp
Fe COb \ e
Fe Ca Fe COb
H 0
23 24
As the temperature of a nitromethane-d3 solution of
h2-1,2-cycloheptadiene Fp (4b) was raised, the two carbonyl
resonances remained unchanged. The allenic carbons C1 and
C3 however became equivalent at +30C. The allenic carbon
C2 remained unchanged, Fig. 3. Taken together, it means
that the rapid fluxional motion of the h2-1,2-cyclohepta-
diene Fp (4) at +30C does not involve an allyl cationic
intermediate because under this condition, C1 and C3 become
equivalent whereas the two carbonyls remain distinct. If an
allyl cationic intermediate (24) were involved in the
fluxional process at +300C, this would correspond to a free
energy of activation of 14.7 kcal/mole. We can safely say
that the fluxional barrier of 13.9 kcal/mole as determined
by 1H NMR corresponds to a simple intramolecular 1,2-Fp
shift, Scheme II, and that the allene to allyl cation
isomerization, Scheme I, requires an energy of greater than
14.7 kcal/mole.
It is reasonable to expect the allene to allyl cation
isomerization to be the higher energy process of the two.
This isomerization requires bonding and structural changes
and a transfer of a positive charge from the iron to the
organic ligand, whereas a 1,2-Fp shift merely involves a
repositioning of the Fp from one double bond to the other.
It is inconvenient to study any dynamic process of
organometallic compounds at high temperatures by 1H NMR due
to their thermal instability. For example, when seeking NMR
evidence for allyl cation formation, a nitromethane-d3
solution of the h2-1,2-cycloheptadiene Fp (4b) was heated to
+600C and was found to decompose rapidly within the time
needed to acquire a 1C spectrum.
A possible way to study the fluxionality of a thermally
sensitive organometallic complex by NMR without raising the
probe temperature is to use the spin saturation transfer
technique, S.S.T.31
The spin saturation technique involves finding a
condition whereby saturating one of the spins of an exchang-
ing two-spin system results in a partial saturation of the
other spin. From this information and the T1 of both spins,
the energy barrier for the exchange can be calculated. The
advantage of this method was exemplified in the case for
h2-1,2-cycloheptadiene Fp (4a). The conditions required for
obtaining a fluxional barrier from spin saturation transfer
described above were met at -20C. It was necessary to heat
(4a) to +290C before its fluxional barrier could be obtained
by the coalescence method.
For the case of the h2-1,2-cycloheptadiene Fp (4), if a
simple 1,2-Fp shift were the only process occurring, irra-
diation of one of the carbonyl resonances would not affect
the intensity of the other because they would remain dis-
tinct at all times. On the other hand, if an allyl cation
were accessible, the carbonyl ligands would become equiva-
lent and if the relaxation of the carbonyl carbon were
slower than the fluxional process, irradiation of one of the
carbonyl resonances would cause the other carbonyl resonance
to diminish in intensity. From this information, it would
be possible to calculate the barrier for the fluxional
process while maintaining a reasonable probe temperature.
Unfortunately, S.S.T. is impractical here for the two car-
bonyl resonances are too close to each other. We could not
selectively irradiate one carbonyl without irradiating the
other.
Another possible way to detect the allyl cationic
intermediate would be to trap it as it is formed. 2-Sub-
stituted allyl cations (25) are known to react with
1,3-dienes.
+ 2- Y -- I K0 Y
25 Y=cO,OR,OM,NR2,R,etc.
Extensive reviews have been written about the cycloaddition
of such allyl cations (25), the most recent one by
Hoffmann.32 These reactions have been exploited in organic
synthesis of bi- and tricyclic compounds. Several possible
products can result from the cycloaddition of allyl cations
(25) to 1,3 dienes depending upon whether the reaction is
concerted or stepwise and upon the nature of Y. The reader
32
is best referred to the review article by Hoffmann32 for all
the possible products from concerted and stepwise cycloaddi-
tions.
The pi-donating strength of a Fp in the homocycloocta-
trienylidene Fp was found to be similar to that of a methoxy
33
group. This would make the 2-Fp substituted allyl cation
a good candidate to be trapped by 1,3-dienes. There is
however no literature precedent that such a reaction will
occur. In fact, allyl cations with Fp substitution at C2
are not even known to exist at this point.
The cycloaddition products expected from cyclopenta-
diene and furan with the h1-allyl Fp cation (5) by either
the concerted or stepwise addition are shown in Scheme III
and IV, respectively.
Scheme III
Concerted addition
I Z +
Z=O,CH2
S- Fp
5
Scheme IV
Stepwise addition
: Fp -
JjH+
a z Fp+
,, 0 H+
jZ +
Z=O,CH2
We do not expect to isolate the carbene adduct (26) from the
concerted cycloaddition of the h1-allyl Fp cation (5) with
the 1,3-diene. A rapid 1,2-alkyl shift would probably occur
to give the corresponding h2-olefin Fp complexes, (27) and
(28).
Fp Fp+
Z a Fpz
27
b
b + Fp Fp+
26
28
Such a 1,2-alkyl shift to a Fp-carbene carbon is well
precedented in the literature.34
A methylene chloride solution of cyclopentadiene or
furan and the h2-1,2-cycloheptadiene Fp (4b) was stirred
for a day. In no case was there any reaction and (4b) was
recovered unchanged. However this does not mean that the
allyl cation (5) is not accessible. The allyl cation (5)
may have formed but did not react or the concentration of
the allyl cation (5) may have been so small that the rate of
reaction was negligible. In fact, the allyl cation (5) may
possibly be a transition state and not an intermediate in
the isomerization pathway.
A methylene chloride solution of tetramethylethylene
was also treated with the h2-1,2-cycloheptadiene Fp (4b) at
room temperature hoping to obtain the h2-olefin Fp complexes
(29). Again (4) was recovered unchanged.
Fp
Fp Fp
5 29
14
In a much earlier study, Manganiello4 discovered that
the thermal decomposition of a methylene chloride solution
of the trifluoromethanesulfonate salt of h2-1,2-cyclo-
heptadiene Fp (4) followed by treatment of the solution with
sodium iodide/acetone yielded 1,3-cycloheptadiene(40%) and a
trace amount of cycloheptene. The mechanism in Scheme V was
proposed to account for the decomposition reaction based on
the products observed.
Scheme V
Fp
y+H-
5
Fp
6H
1 H
H+
Nal
0+ FplI
INal
\ + Fpl
In this scheme, proton loss from the h -allyl Fp cation
(5) is responsible for the formation of the 1,3-cyclohepta-
diene, whereas a hydride abstraction gives the cycloheptene.
It was argued that if a hydride source were available to the
allyl cation (5) then the amount of the cycloheptene formed
relative to 1,3-cycloheptadiene would increase. Indeed,
when triphenylmethane (a hydride source) was added to a
methylene chloride solution of the h2-1,2-cycloheptadiene Fp
(4a), cycloheptene was isolated in amounts up to 3/5 of that
of 1,3-cycloheptadiene after workup.
The above experiment was repeated using the tetra-
fluoroborate salt of the h2-1,2-cycloheptadiene Fp (4).
Neither cycloheptene nor 1,3-cycloheptadiene were observed
but the h2-1,2-cycloheptadiene Fp (4b) was reisolated
unchanged. There was no reaction between (4b) and tri-
phenylmethane. This cast doubts on the previously proposed
mechanism involving the intermediacy of the allyl cation
(5).
In principle, clear evidence for the accessibility of
the allyl cation (5) may be obtained by using an optically
active cyclic allene as the ligand. As can be seen from
Scheme I, if the allyl cation (5) is accessible, whether as
an intermediate or a transition state, an optically active
complex must racemize. One approach to the synthesis of an
optically active h2-1,2-cycloheptadiene Fp (4) is to dis-
place isobutylene from h2-isobutylene Fp with an optically
active 1,2-cycloheptadiene in much the same way as was used
in the preparation of optically active h2-1,2-cyclononadiene
Fp (8). Unfortunately, unlike 1,2-cyclononadiene, 1,2-
cycloheptadiene dimerizes too rapidly to displace the
isobutylene.
Alternative methods were therefore sought. In the
first approach, racemic h2-1,2-cycloheptadiene Fp (11) could
be treated with less than an equivalent of an optically
active alcohol in order to selectively remove one the allene
enantiomers. This procedure would only be effective if the
racemization of (4) via the allyl cation (5) did not occur
or occurred at a very slow rate under the conditions for
enantiomeric enrichment.
2-Methylbutanol was tried first. When the racemic
alcohol was allowed to react with a suspension of h -1,2-
cycloheptadiene Fp (4) in diethyl ether, it gave the ether
adduct (30).
Fp4 Fp
+"OH
+ OH 0
30
Interestingly, although the ether adduct (30) has two chiral
centers as marked, and therefore should exist as a pair of
diastereomers, only one set of signals was observed in its
1H NMR (we have insufficient material for a C NMR
spectrum). It is quite unfortunate that the diastereomers
have coincidental chemical shifts. One would expect that
the reaction between an optically active alcohol with a
racemic h2-1,2-cycloheptadiene Fp (4) should give unequal
amounts of the diastereomeric ether adduct (30) provided the
reaction is incomplete. If the chemical shifts of the
diastereomeric ether adduct (30) were non-coincidental, NMR
measurement of the diastereomeric composition of the ether
adduct (30) would enable us to draw a conclusion as to the
enantiomeric purity of the h2-1,2-cycloheptadiene Fp (4)
left behind.
Half an equivalent of the (S)-(-)-2-methylbutanol was
allowed to react with the racemic h2-1,2-cycloheptadiene Fp
(4b). The unreacted (4b) recovered, which amounted to a
half of the starting material, showed no detectable optical
rotation.
The chiral carbon on (S)-(-)-2-Methylbutanol is one
carbon away from the alcohol group and may be too far
removed for it to effectively induce asymmetry. We decided
to try (-)-menthol instead. Under the same conditions,
(-)-menthol did not give the ether adduct but a mixture of
unidentifiable products and menthol.
Fp
Fp HO Men
31
Less than half of the starting h2-1,2-cycloheptadiene Fp
(4b) was recovered. A solution of the recovered h2-1,2-
cycloheptadiene Fp (4b) showed no detectable optical
rotation. A reaction between h2-1,2-cycloheptadiene Fp (4b)
and potassium (-)-menthoxide in THF at 0C resulted in total
decomposition of the allene (4b).
Zero optical rotation of the recovered h2-1,2-cyclo-
heptadiene Fp (4b) from the above reactions is consistent
with an allene to allyl cation isomerization, Scheme I.
However, absence of rotation can also be due to other
reasons. First, enantiomeric enrichment by (S)-(-)-2-
methylbutanol and (-)-menthol might not be successful. Even
if the enantiomeric enrichment were successful, the
recovered h2-1,2-cycloheptadiene Fp (4b) might have such low
inherent optical rotation that it could not be measured with
certainty. Unfortunately, it is not possible to increase
the concentration of the solution of h2-1,2-cycloheptadiene
Fp (4b) in order to get a measurable optical rotation
because the solution which is highly colored absorbs so much
light that rotation measurements become quite impossible.
In the second approach, we attempted to synthesize the
optically active h2-1,2-cycloheptadiene Fp (4) directly via
methoxy abstraction with trimethylsilyl trifluoromethane-
sulfonate starting from an optically active hl-(7-methoxy)-
cyclohepten-1-yl Fp.
Let us first consider the conformation of the R and S
enantiomers of the h -(7-methoxy)cyclohepten-1-yl Fp as
obtained from CPK space filling models.
H Fp
OCH3
R
H
Fp
OCH3
S
Se Sg
Of the two conformers available to the h -((R)-7-methoxy)-
cycloheptadien-1-yl Fp (R) we would expect the gauche
conformer (Rg) to be the preferred conformer. Here, the
larger methoxy groups is away from the bulky Fp. Similarly,
we expect the (Sg) to be the preferred conformer for the
h -((S)-7-methoxy)cyclohepten-1-yl Fp (S).
We shall limit our discussion on methoxy abstraction to
one enantiomer, h -((R)-7-methoxy)cyclohepten-l-yl Fp (R).
Cationic h2-olefin Fp complexes are typically made by a
B-hydride abstraction from the h -alkyl Fp.35 Such
B-hydride abstractions are conformationally dependant and
widely accepted to proceed via an antiperiplanar transition
state with the Fp assisting from the anti face in concert
with the hydride loss.
M
N/PE
/1
H
Methoxy abstraction, like hydride abstraction, is also
believed to proceed with anti Fp assistance leading to the
h2-olefin Fp complex. A concerted methoxy abstraction with
anti Fp assistance will give an optically active h2-olefin
Fp complex from an optically active hl-alkyl Fp. When the
methoxy group is part of a ring as in the case with h -
(7-methoxy)cyclohepten-l-yl Fp, it is prevented by the ring
from adopting a conformation where it is anti to the Fp. An
anti alignment of the Fp with the methoxy group would
necessarily force the methoxy group inside the cycloheptene
ring, a conformation which is impossibly strained.
Since a concerted anti methoxy abstraction is unlikely,
a concerted syn methoxy abstraction with the Fp moving over
to the same face as the leaving methoxy group will also give
an optically active h2-olefin Fp complex from an optically
active h -alkyl Fp. Thus by a concerted mechanism with syn
1
Fp assistance the h -((R)-7-methoxy)cyclohepten-1-yl Fp
(R) should give only the h2-(S)-1,2-cycloheptadiene Fp (45),
Scheme VI.
Scheme VI
H p HFp
H FpH Fp H
/CH3 Fp -
`Q
/-- OTF
Rg 4S
Although syn Fp assisted 8-hydride and B-methoxy
abstractions have not been substantiated, there are examples
where an h2-olefin Fp complex is formed by an apparent syn
Fp assistance.36
*
The discussion is still valid should for any reason methoxy
abstraction involve anti Fp assistance; thus R would give
4R.
Fp
H Fp+
H Ph3C+ F
H
H Fp Fp+
Ph3C+
SH
H
There are no hydrogens anti to the Fp in h -cyclobutyl and
hl-cyclopentyl Fp, and yet they undergo hydride abstraction
to give their respective h2-cycloalkene Fp complexes. It is
not known whether hydride abstraction occurs via a distorted
transition state, a e-carbonium cation or by some other
mechanism.
If Fp migration is not concerted with methoxy abstrac-
tion, an allyl cation intermediate (5) would be formed and
should give the racemic h2-1,2-cycloheptadiene Fp (4),
Scheme VII.
Scheme VII
-Fpb a
In this case, the h -((R)-7-methoxy)cyclohepten-l-yl Fp (R)
would give both the (R)- and the (S)-h2-1,2-cycloheptadiene
Fp, (4R) and (4S).
In short, it is possible to synthesize an optically
active h2-1,2-cycloheptadiene Fp (4) from syn Fp assisted
methoxy abstraction of optically active h -(7-methoxy)cyclo-
hepten-l-yl Fp. An optically inactive h2-1,2-cyclohepta-
diene Fp (4) from this reaction would necessarily mean that
methoxy abstraction proceeded via an allyl cation or that
the barrier for isomerization between the allene (4) and the
allyl cation (5) is very low.
Optically active h -(7-methoxy)cyclohepten-l-yl Fp was
prepared according to Scheme VIII.
Scheme VIII
,OCH3
+ Qui
0
II
COOQui
OCH3
separate
The racemic 7-methoxycycloheptenecarboxylic acid was
resolved as its quinine salt via two recrystallizations from
absolute ethanol. The acid was released from the quinine
salt readily in aqueous acid and was found to have a nega-
tive optical rotation. The acid was subsequently converted
to the acid chloride which was then reacted with potassium
Fp to yield the hl-carbonyl-(7-methoxy)cyclohepten-l-yl Fp
(32). The 1H NMR of the cyclopentadienyl hydrogen
.OCH3
0CH3
00
II II
CICCCI
lii
resonances are well separated in a 10 mole % chloroform-d1
solution of Eu(hfc)3 and have an integrated ratio of 25 to
75 which indicates a 50% excess of one enantiomer. Decar-
bonylation under photolytic conditions yielded h -(7-
methoxy)cyclohepten-l-yl Fp (N). We presumed that the
h (7-methoxy)cyclohepten-l-yl Fp (N) is optically active
with the same optical purity as the hl-carbonyl-(7-
methoxy)cyclohepten-l-yl Fp (32). Eu(hfc)3 does not
separate the cyclopentadienyl and methoxy resonances of
h -(7-methoxy)cyclohepten-l-yl Fp (N) and a direct measure-
ment is not possible. It is also very difficult to obtain
optical rotations on these Fp complexes because their
solutions are highly colored.
A mixture weighted with the other enantiomer was
prepared according to Scheme IX.
Scheme IX
Br Br Br
0 H OCH3
+ LAH/Qui CNaH3
(+) 20% e.e.
33 CL Fp
Fp OCH3
OCH3
1) n-BuLi CP2Fe+
2) FpC CO 20% e.e.
P 32
The 2-bromo-2-cycloheptenone was reduced to its alcohol (33)
with a preformed LAH/quinine mixture. The alcohol (33) has
a positive optical rotation and was obtained with a 20%
enantiomeric excess as determined by Eu(hfc)3. Treating
this mixture of alcohols (+33) with sodium hydride followed
by methyl iodide gave the l-bromo-7-methoxycycloheptene.
The h -(7-methoxy)cyclohepten-l-yl Fp was obtained by the
well established route. As mentioned above, the h -(7-
methoxy)cycloheptadien-l-yl Fp does not form a complex with
Eu(hfc)3 and must be converted to its h -carbonyl-
(7-methoxy)cyclohepten-l-yl Fp (32) in order to determine
its enantiomeric purity. The h -carbonyl-(7-methoxy)cyclo-
hepten-l-yl Fp (32) was prepared by stirring a methylene
chloride solution of hl-(7-methoxy)cyclohepten-l-yl Fp in
10 mole % of ferrocenium tetrafluoroborate under 55 psi of
CO gas. The reaction was complete within an hour giving a
quantitative yield of the desired Fp-acyl complex (32). The
h -carbonyl-(7-methoxy)cyclohepten-l-yl Fp (32), in a
10 mole % chloroform-d1 solution of Eu(hfc)3 has a cyclo-
pentadienyl hydrogen integrated ratio of 60 to 40 giving us
a 20% excess of the other enantiomer. We shall call this
mixture (P). It was comforting to know that this 20%
enantiomeric excess was carried forward from the (+)-alcohol
(33). We infer from this information that the
h1-(7-methoxy)cyclohepten-1-yl Fp (P) made was optically
active (20% e.e.) and with an optical rotation opposite to
that prepared by Scheme VIII.
The carbonylation of the h -(7-methoxy)cyclohepten-1-yl
Fp to h1-carbonyl-(7-methoxy)cyclohepten-1-yl Fp (32) is of
particular importance here. As far as we know, this is the
first case of an unassisted CO insertion into a h -vinyl Fp
bond to give a h -acyl Fp complex.
All CO insertions into hl-alkyl Fp or h -vinyl Fp
complexes are assisted by phosphines under thermal condi-
tions to give the corresponding hl-acyl (Cp)Fe(CO)(PPh3)
complexes.37
Fe R P3 > Fe -
0/C OC R
CO R'3P
R allyl, vinyl
CO insertions catalyzed by oxidants, eg. Ce(IV), Ag(I),
Cp2Fe or Lewis acids have only been reported for h -alkyl
(Cp)Fe(CO)(L) where L = PPh3 and P(OPh)3 and h1-vinyl
(Cp)Fe(CO)(P(OPh)3) complexes;38 the former complexes
carbonylate faster than the latter. To date, CO insertions
into h -(alkoxy)methylene Fp to give the corresponding acyls
39
have not been successful.3
0
Fp C2 OR C CH2II
Fp CH2 OR CX)> Fp C- CH20R
We have also attempted without success to carbonylate the
h -(7-methoxy)cyclohepten-l-yl Fp with boron trifluoride
etherate under 1 atm. of CO.
The optically active h -(7-methoxy)cyclohepten-l-yl Fp
(N) and (P) were treated with trimethylsilyl trifluoro-
methanesulfonate to yield the h2-1,2-cycloheptadiene Fp
(4a). A dilute solution of the h2-1,2-cycloheptadiene Fp
(4a) has a negligible optical rotation. When the concentra-
tion of (4a) was increased, the solution became strongly
colored and made optical measurement impossible.
Since rotational measurements were experimentally
impossible, we decided that an alternative solution would be
to convert the h2-1,2-cycloheptadiene Fp (4a) to its methyl
ether adduct and carbonylate the ether adduct to the h-
carbonyl-(7-methoxy)cyclohepten-l-yl Fp (32).
FP Fp
Fp+ OCH3 OCH3
CH3OH CO
4 32
The enantiomeric composition of the h carbonyl-(7-
methoxy)cyclohepten-l-yl Fp (32) can then be determined by
H NMR with Eu(hfc)3 and this will give us a clue as to the
enantiomeric composition of the h2-1,2-cycloheptadiene Fp
(4).
~
To simplify the discussion, let us assume that the
fluxional process proceeds only via the 1,2-Fp shift,
Scheme II, and that the addition of methanol is stereo-
specific.
Anti attack of a nucleophile at a h2-olefin Fp bond is
40
well documented in the literature.40 Anti attack of a
methanol onto the h2-(R)-1,2-cycloheptadiene Fp (4R) regard-
less of which double bond the Fp is bonded to should give
the same h -((R)-7-methoxy)cyclohepten-l-yl Fp (R),
Scheme X.
Scheme X
H .H
Fp' H Fp
O H
Fp CH3 / 1,.CH3
4R -<
H R
The above statement is true because only two of the four
sides of the allene double bonds are available for complex-
ation with the Fp. If the Fp were bonded to the other
two sides of the allene double bond, an impossible geometry
which puts the Fp inside the ring (see 4R below), the allene
(4R) would add methanol in an anti fashion to give the (S)
methyl ether adduct.
Fp
Fp H+ CH30O.i
CH3OH
4R S
It should be noted that the anti face of the h -(R)-
1,2-cycloheptadiene Fp (4R), where the methanol comes in, is
only partially shrouded by the allene ring. A syn attack,
however, would be hindered by the bulky Fp group.
At first glance, eliminating the possibility of a syn
attack by methanol would seem to violate the principle of
microscopic reversibility;41 especially when we suggest syn
assistance as one of the ways methoxy abstraction can occur.
Furthermore, by the principle, anti attack of methanol would
necessarily require the 7-membered ring to adopt an
impossible conformation of putting the methanol inside the
ring. We must keep in mind that neither the methoxy
abstraction nor the methanol addition reactions are
reversible and that the methoxy group is leaving the
h -(7-methoxy)cyclohepten-1-yl Fp as a methyltrimethylsilyl
ether and not as a methanol. Thus, methoxy abstraction and
addition is not bound by microscopic reversibility. Regard-
less of the pathway methoxy abstraction follows (syn or
anti), we should be able to arrive at the enantiomeric
composition of the h2-1,2cycloheptadiene Fp (4) from the
enantiomeric composition of the hl-carbonyl-(7-methoxy)-
cyclohepten-1-yl Fp (32) provided methanol addition is
stereospecific.
The h2-1,2-cycloheptadiene Fp (4a), from the reaction
of the optically active hl-(7-methoxy)cyclohepten-1-yl Fp
(N) with trimethylsilyl trifluoromethanesulfonate, was
reacted with methanol and then carbonylated to give the
hl-carbonyl-(7-methoxy)cyclohepten-1-yl Fp (32). The
integrated cyclopentadienyl hydrogen intensities of h-
carbonyl-(7-methoxy)cyclohepten-1-yl Fp (32) in a chloro-
form-d1 solution containing Eu(hfc)3 were equal, Fig. 4 and
5. At some point complete racemization had occurred. The
optically active h -(7-methoxy)cyclohepten-1-yl Fp (P) was
put through the same series of reactions with the same
results, Fig. 6 and 7. Again complete racemization had
occurred.
Based upon the experimental results it appeared that an
allyl cation (5) had been formed at some point in the
sequence of reactions. One possibility is the concerted
formation of the optically active h2-1,2-cycloheptadiene Fp
(4) which then racemizes via the allyl cation (5). An
0
O4J
U(
o
O-
0
43
10
fa
>1
x>
0 0
. 0
4J -H
a, *
e --
Uo
Eo
o ao
u
0)
OU
r 0
$4
orn
-43
0 m
So
o0
-4l
0 m
C00
0U
mo
-i Z
r4-1
*-H
F:
58
0
0
dp
0'0
H
0
(N
a
0 r
m
0
-H
4
0)*
00E
(n 0
QH
U4J *
w
KW
rlUi
F4 r
'0
.H
4-)-
C.
-4
4-)
X
.0z
1900
4-) -1
W4-
0 04
3-O
u
C4-4
r.0
4-)-
'U
r-q
00
E zr
Eo
4-4 U
.0 H
rco
ul
0
4-r)
Crcr
0 0
mo
4-)0
0-4
-H
mOL
u0-
'0
Nr
60
alternative is allyl cation formation as methoxy is
abstracted followed by collapse to the h2-1,2-cyclo-
heptadiene Fp (4). In either case we would have to invoke
an allyl cation intermediate (5).
CHAPTER III
CARBONYL(h5-CYCLOPENTADIENYL)TRIPHENYLPHOSPHINEIRON(II)
COMPLETED 1,2-CYCLOHEPTADIENE
Methoxy abstraction from optically active hl-(7-
methoxy)cyclohepten-l-yl Fp led to the racemic h2-1,2-cyclo-
heptadiene Fp (4). Two mechanistic pathways were con-
sidered; one involving a concerted stereospecific methoxy
abstraction to the optically active h2-1,2-cycloheptadiene
Fp (4) followed by racemization via the achiral allyl cation
intermediate (5) and the other involving the achiral allyl
cation intermediate (5) from methoxy abstraction and prior
to complexation by the Fp.
In order to complete the stereochemical picture and to
shed light on the methoxy abstraction and the racemization
process, we shall turn our attention to the h and
h2- carbonyl(h -cyclopentadienyl)triphenylphosphineiron(II)
[Fpp] complexes of 7-methoxycycloheptadien-l-yl (34) and
1,2-cycloheptadiene (35), respectively.
L 0 3OCH3
Fe Fe
o0// 0o
03P 03P
34 35
There are quite a few advantages to moving from the Fp
system to the Fpp system. The most obvious is that the
triphenylphosphine ligand generally imparts additional
stability to these type of complexes against exposure to air
and therefore makes them easier to manipulate.
More important to us is that the triphenylphosphine
ligand introduces another chiral center, at the iron, in
both the hl-(7-methoxy)cyclohepten-l-yl Fpp (34) and the
h2-1,2-cycloheptadiene Fpp (35) complexes. The consequence
of this is that both complexes exist as enantiomeric pairs
of diastereomers. Diastereomers, unlike enantiomers, can
often be separated by physical methods and their diastere-
omeric compositions determined directly by NMR spectroscopic
measurements.
Phosphinylation therefore provides us a means to
separate hl-(7-methoxy)cyclohepten-l-yl Fpp (34) into its
diastereomers. We felt that examination of the diastere-
omeric composition of h2-1,2-cycloheptadiene Fpp (35) formed
from different diastereomers of hl-(7-methoxy)cycloheptene-
l-yl Fpp (34) should enable us to elucidate the pathway by
which methoxy abstraction occurs. Also, if h2-1,2-cyclo-
heptadiene Fpp (35) can be separated into its diastereomers,
we can follow directly by NMR the allene to allyl cation
isomerization because such isomerization necessarily con-
verts one diastereomer to a mixture of both diastereomers.
Lastly, since we were unable to grow crystals of the
h2-1,2-cycloheptadiene Fp (4) suitable for X-ray studies, we
would like to try to grow crystals of its Fpp analogue.
Before we proceed any further, the iron center of the
Fpp can be designated as either R or S according to the
Cahn-Ingold-Prelog (CIP) system adapted to organotransition
metal complexes.42 The configurations of the isomers of
hl-(7-methoxy)cycloheptadien-l-yl Fpp (34) and h2-1,2-cyclo-
heptadiene Fpp (35) according to the modified CIP system are
shown below.
H OCH3 Fe H CH3
0C.0 03P j
03P CO
R R S R
and enantiomer RS
and enantiorner SS
FeN Fe+
0H H HN-
OC O3P'*
03P OC
S R R R
and enantiomer RS and enantiomer SS
note that the iron center with the R configuration in
h -(7-methoxy)cyclohepten-l-yl Fpp (34) and h -1,2-cyclo-
heptadiene Fpp (35) have opposite absolute configuration.
h2-(7-Methoxy)cyclohepten-1-yl Fpp (34) was synthesized
by photolyzing a mixture of the hl-(7-methoxy)cyclohepten-
1-yl Fp with excess triphenylphosphine in an equal mixture
of n-pentane and benzene.
/ OCH3 Z OCH3
Fe Fe
/// \
OC CO + P03 hv 03
34
Photolysis was conducted in a quartz photochemical reactor
at room temperature under a stream of N2 to remove the CO
gas evolved. The reaction proceeded rapidly and was com-
plete within 20 mins without forming the bistriphenyl-
phosphine adduct. hl-(7-Methoxy)cycloheptadien-1-yl Fpp
(34) was purified via column chromatography and was obtained
as an air stable red paste. Residual triphenylphosphine may
be removed by filtration as its insoluble methyltriphenyl-
phosphonium iodide salt after warming a n-pentane solution
of the crude h1-(7-methoxy)cyclohepten-1-yl Fpp (34) with
methyl iodide.
The H NMR for h -(7-methoxy)cyclohepten-1-yl Fpp (34)
showed two sets of resonances, one for each diastereomer.
The hydrogen resonances for the methoxy hydrogens and the
ring hydrogens on carbons 2 and 7 are well separated. The
cyclopentadienyl hydrogen resonances are barely separated
from each other at 100 MHz and each is coupled to the
phosphorus.
Let us now consider first the methoxy abstraction of
hl-(7-methoxy)cyclohepten-l-yl Fpp (34) within the context
of a concerted stereospecific Fpp assisted pathway.
Following the same lines of analysis as the Fp ana-
logue, let us assume that the ring conformation necessary to
place the methoxy group anti to the Fpp cannot be achieved
and that any concerted stereospecific pathway for methoxy
abstraction must necessarily proceed with syn Fpp
assistance. Therefore, concerted stereospecific methoxy
abstraction of a single diastereomer of hl-(7-methoxy)
cyclohepten-l-yl Fpp (34) should lead to a single
diastereomer of h2-1,2-cycloheptadiene Fpp (35), Scheme XI.
Scheme XI.
OC_ OC'
03P3 Fe H 03P FeH
H\ H H
O III'OCH3
RR SS
This argument still holds should for any reason methoxy
abstraction proceed via anti Fpp assistance.
The other alternative is for methoxy abstraction to
proceed via a free allyl cation. The Fpp then collapses
onto the free allyl cation to give the h2-1,2-cyclo-
heptadiene Fpp (35). In this case, methoxy abstraction from
one diastereomer of h -(7-methoxy)cyclohepten-l-yl Fpp (34)
should lead to a mixture of diastereomers of h2-1,2-cyclo-
heptadiene Fpp (35), Scheme XII.
Scheme XII.
03P Fe
03P I -Fe
OC
03P Fe
H I H
SOCH3
It is clear that in order to establish unequivocally
the mechanism of methoxy abstraction, we need to separate
h -(7-methoxy)cyclohepten-1-yl Fpp (34) into its
two diastereomers.
hl-(7-Methoxy)cyclohepten-l-yl Fpp is isolated as a red
oil and its diastereomeric composition typically falls
around 60:40. The diastereomeric composition was deter-
mined by 1H NMR integration of the methoxy hydrogen reso-
nances because the cyclopentadienyl hydrogen resonances were
not well resolved. We first attempted the separation by
recrystalization at -350C from a variety of solvents; the
Diastereomeric composition will always be presented as such
a : b where (a) corresponds to the composition of the low
field cyclopentadienyl hydrogen resonance and (b) corre-
sponds to the high field resonance.
solvents tried include pentane, hexane, hexane/benzene
mixture and acetone/water mixture. In each case the
diastereomeric composition of the crystals of hl-(7-
methoxy)cyclohepten-1-yl Fpp (34) remained unchanged. The
diastereomers were also not separated by TLC.
An alternative approach to this problem was to synthe-
size the h -carbonyl-(7-methoxy)cyclohepten-l-yl Fpp (36),
which also exists as enantiomeric pairs of diastereomers,
separate this complex into its diastereomers and then
convert them back to the corresponding hl-(7-methoxy)cyclo-
hepten-l-yl Fpp (34).
The hl-(carbonyl-(7-methoxy)cyclohepten-1-yl Fpp (36)
was readily synthesized by warming h -(7-methoxy)cyclo-
hepten-l-yl Fp (34) with triphenylphosphine in acetonitrile.
6OCH3 0 OCH3
Fe Fe
OC 0 + P3 //
03P
36
The acyl complex (36) is an air stable orange solid and is
formed as a mixture of equal amounts of the two diastere-
omers. After numerous recrystalization attempts, we dis-
covered that a 20:80 ratio of diastereomers of (36) could be
obtained in one recrystalization from a 1:2 v/v mixture of
ethyl acetate and n-pentane at +100C with excellent recov-
ery, Fig. 8. The mother liquor was concentrated to give the
other diastereomer in a 65:35 ratio. A second recrystali-
zation led to further enrichment giving a 10:90 ratio of
diastereomers of (36).
Attempts to decarbonylate the hl-carbonyl-(7-
methoxycyclohepten-l-yl Fpp by either of the standard
photolytic or chemical methods were unsuccessful. Surpris-
ingly, CO was not evolved when a benzene solution of the
acyl complex (36) was photolyzed with a low pressure Hg lamp
even though other phosphinylated iron acyl alkyl complexes
have been reported not only to decarbonylate under photo-
lytic conditions but also to retain high
stereospecificity at the iron.43 The acyl complex (36) was
isolated with its diastereomeric composition unchanged. No
reaction was observed when the acyl complex (36) was allowed
to react with trimethylamine-N-oxide, chlorobis(tri-
phenylphosphine)rhodium dimer or iodosobenzene. In all
cases the acyl complex (36) was reisolated with its
diastereomeric composition unchanged. One should note the
tenacity of this acyl complex (36) against epimerization at
the iron center. The acyl complex (36) gave predominantly
decomposition products when refluxed in dioxane although a
small amount of hl-(7-methoxy)cyclohepten-1-yl Fp was formed
presumably via phosphine dissociation followed by migration
of the organic ligand onto the iron.
We finally succeeded in inducing decarbonylation in the
acyl complex (36) by a combined photolysis and ultrasoni-
cation process. Unfortunately, the h -(7-methoxy)cyclo-
hepten-l-yl Fpp (34) isolated had epimerized to a 55:45
diastereomeric mixture.
It was later discovered that eluting h1-(7-methoxy)
cyclohepten-l-yl Fpp (34) from a 8" x 1" alumina column
(neutral, grade II) with a 1:1 v/v n-pentane/benzene as the
eluant afforded an enriched diastereomeric mixture of the
complex (34). The diastereomeric mixture of complexes (34)
appeared on the column as a broad red band and was collected
in two halves. Surprisingly, the first half and the second
half of this red band both gave mixtures of hl-(7-methoxy)-
cyclohepten-1-yl Fpp (34) enriched in the same diastereomer
(75:25 and 85:15, respectively), Fig. 9. The alumina
apparently did not separate the mixture of diastereomers but
selectively decomposed one of them. Selective decomposition
of h -(7-methoxy)cyclohepten-1-yl Fpp (34) with trimethyl-
silyl trifluoromethanesulfonate also afforded an enriched
diastereomeric mixture of the complex but this time in a
10:90 ratio, Fig. 10. We shall postpone the discussion of
the latter selective decomposition to an appropriate time.
We have also attempted to synthesize other phos-
phinylated h1-(7-methoxy)cyclohepten-1-yl iron complexes but
without much success. The respective phosphinylated com-
plexes were not isolated when h -(7-methoxy)cyclohepten-1-yl
Fp was photolyzed in the presence of either trimethyl-
phosphine or triphenylphosphite. When a cold (-780C)
solution of l-lithio-7-methoxycycloheptene was treated with
carbonyl(h5-cyclopentadienyl)tri-n-butylphosphineiron(II)
73
0
.H
4J
.rl
U3
0
0
(
U
.14
ro
m
0
U)
-4
0
I )
r.
0
z
m
U
U
0
-4
0r
z
0r
-4
[Fppbu] iodide, the major product isolated was unreacted
Fppbu iodide and a minor product perhaps resulting from the
attack of the vinyl lithium reagent on the cyclopentadienyl
ring.
Triphenylphosphite substituted hl-(7-methoxy)cyclo-
hepten-l-yl iron complex was synthesized via a thermal
exchange for triphenylphosphine in h -(7-methoxy)cyclo-
hepten-l-yl Fpp with an excess triphenylphosphite in
refluxing THF.37a
Z OCH3 0 OCH3
Fe j /<
Fe
OC 03P I + P(O0)3 THF> o/ /
(00)3P
The equilibrium of this reaction lies strongly to the right
despite the stronger nucleophilicity of the triphenylphos-
phine and is attributed to a steric effect. The triphenyl-
phosphine ligand being very bulky is unable to displace the
triphenylphosphite ligand once the latter is coordinated at
the iron. We could only isolate this triphenylphosphite
iron complex as a crude mixture containing free triphenyl-
phosphine and triphenylphosphite.
At this point, we have not succeeded in isolating
h -(7-methoxy)cyclohepten-l-yl Fpp (34) as a single
diastereomer but we have developed the methodology which
enables us to separate hl-(7-methoxy)cyclohepten-l-yl Fpp
(34) from a 60:40 diastereomeric mixture to a mixture with a
diastereomeric ratio of 85:15 and 10:90.
Before we look at the methoxy abstraction process, it
is imperative to examine the stereochemical integrity of
hl-(7-methoxy)cyclohepten-1-yl Fpp (34). We discovered that
h -(7-methoxy)cyclohepten-1-yl Fpp (34) with diastereomeric
compositions of 85:15 or 60:40 maintained their respective
compositions indefinitely when frozen in the refrigerator.
A chloroform-dl solution of hl-(7-methoxy)cyclohepten-l-yl
Fpp (34) with a 85:15 diastereomeric composition however
epimerizes very slowly as measured by 1H NMR but only at
elevated temperatures. The diastereomeric composition of a
60:40 mixture of hl-(7-methoxy)cyclohepten-l-yl Fpp (34)
remained unchanged by 1H NMR under the same conditions.
Therefore, we feel it is safe to assume that h -(7-
methoxy)cyclohepten-l-yl Fpp (34) does not epimerize under
the conditions for methoxy abstraction, i.e., short duration
and low temperature.
We tried a number of methods to effect methoxy abstrac-
tion. Trimethyloxonium tetrafluoroborate was used first
instead of the trimethylsilyl trifluoromethanesulfonate
because we anticipated that the trifluoromethanesulfonate
anion might affect the h2-1,2-cycloheptadiene Fpp (35) in an
adverse manner similar to what was observed with the
h2-1,2-cycloheptadiene Fp (4). Treating h -(7-
methoxy)cyclohepten-l-yl Fpp (34) with trimethyloxonium
tetrafluoroborate in methylene chloride at -100C for 7 hrs
gave an impure green substance after the reaction mixture
was quenched with diethyl ether.
Discouraged by the ineffectiveness of trimethyloxonium
tetrafluoroborate as a methoxy abstracting reagent, we went
back to trimethysilyl trifluoromethanesulfonate. The
reaction did not fair any better and a black intractable
paste was obtained. The same reaction in pentane at 0C
gave a greenish yellow precipitate which dissolved when
washed with diethyl ether. If the reaction in pentane was
quenched with diethyl ether, again a black pasty residue was
formed. We discovered, quite by accident, that if this
black paste is dissolved in ethyl acetate, the trifluoro-
methanesulfonate salt of h2-1,2-cycloheptadiene Fpp (35a)
precipitates from the solution as a bright orange air stable
solid. In fact, prior quenching with diethyl ether is
unnecessary. Pentane was removed from the reaction mixture
in vacuo to leave a dark greenish brown paste from which
h -1,2-cycloheptadiene Fpp (35a) was precipitated with ethyl
acetate in the open air. The reaction was also run in an
equal mixture of n-pentane and benzene at 0C with similar
results; the solvent mixture was removed in vacuo to leave a
residue from which h2-1,2-cycloheptadiene Fpp (35) was
precipitated with ethyl acetate.
The 1H NMR of the h2-1,2-cycloheptadiene Fpp (35a) is
consistent with a pair of diastereomers, Fig. 11. The
two cyclopentadienyl hydrogen resonances, each coupled to
the phosphorus, and three of the four allenic hydrogen reso-
nances are well separated. One allenic hydrogen resonance
is obscured by the ring methylene resonances and is
confirmed by 2H NMR of the appropriately deuterated complex.
The assigned structure was confirmed by X-ray videe infra).
The trifluoromethanesulfonate anion was readily
exchanged for the hexafluorophosphate anion by adding water
to a methanol solution of the trifluoromethansulfonate salt
of the h2-1,2-cycloheptadiene Fpp (35) containing an excess
of ammonium hexafluorophosphate. This gave a yellow pre-
cipitate which was filtered to give the pure hexafluoro-
phosphate salt (35b) in quantitative yield. The procedure
was performed in open air which demonstrated the stability
of h -1,2-cycloheptadiene Fpp (35). The h -cycloheptadiene
Fp (4a,b) would have decomposed in the presence of air,
methanol or moisture.
Unfortunately, treating carbonyl(h5-cyclopentadienyl)
(hl-(7-methoxy)cyclohepten-l-yl)triphenylphosphiteiron(II)
with trimethylsilyl trifluoromethanesulfonate failed to give
the h2-1,2-cycloheptadiene complex. we believe that
trimethysilyl trifluoromethanesulfonate is incompatible with
the triphenylphosphite ligand and results in complete
decomposition of the iron complex.
We have thus developed the methodology to synthesize
and purify both the trifluoromethanesulfonate and hexa-
fluorophosphate salt of the h2-1,2-cycloheptadiene Fpp (35).
Both (35a) and (35b) are air stable and moderately stable in
solution.
We are finally ready to look into the process of
methoxy abstraction, paying particular attention to the
h2-1,2-cycloheptadiene Fpp (35) diastereomeric composition
obtained starting with a majority of one of the
diastereomers of h -(7-methoxy)cyclohepten-l-yl Fpp (34).
Methoxy abstraction from a 60:40 mixture of h -(7-
methoxy)cycloheptadien-l-yl Fpp (34) in a 1:1 v/v
n-pentane/benzene solvent mixture at 0C typically gives a
65:35 to 60:40 diastereomeric mixtures of h2-1,2-cyclo-
heptadiene Fpp (35) after precipitation from ethyl acetate.
The diastereomeric composition was determined by 1H NMR
based on the integrated cyclopentadienyl hydrogen
resonances. Repeated precipitation from ethyl acetate does
not change this diastereomeric composition.
The diastereomeric composition of h2-1,2-cyclo-
heptadiene Fpp (35) was found to be dependent upon the
solvent in which methoxy abstraction was performed. When
the reaction was repeated under identical conditions but in
n-pentane, the crude product (35a) was found to have a
diastereomeric composition of around 65:35. The 1H NMR of
the crude product (35a) is broad and poorly resolved and one
should consider the diastereomeric ratio of 65:35 a very
rough value. A single precipitation from ethyl acetate
gave (35b) in a 80:20 diasteromeric composition. Just in
case the discrepancy in the crude and pure diastereomeric
compositions of h2-1,2-cycloheptadiene Fpp (35a) was in some
way caused by the precipitation process, the filtrate was
collected and was examined for h2-1,2-cycloheptadiene Fpp
(35a). We were unable to detect by 1H NMR any trace of
h -1,2-cycloheptadiene Fpp (35a). This 80:20 composition
also remained invariant after repeated precipitations.
Regardless of the solvent used, the h2-1,2-cycloheptadiene
Fpp was obtained at 45% to 55% yields.
1
Methoxy abstraction of a 80:20 mixture of h -(7-
methoxy)cyclohepten-1-yl Fpp (34) in the n-pentane/benzene
solvent mixture at 0C gives a 60:40 diastereomeric mixture
of the h2-1,2-cycloheptadiene Fpp (35a) after precipitation.
The yields are slightly higher; over 55%.
The most interesting observation regarding methoxy
abstraction is from a 60:40 mixture of h -(7-methoxy)cyclo-
hepten-1-yl Fpp (34) with half an equivalent of trimethyl-
silyl trifluoromethanesulfonate at 0C in n-pentane/benzene
solvent mixture. We were able to isolate a 60:40
diastereomeric mixture of h2-1,2-cycloheptadiene Fpp (35a)
in over 90% yield (based on trimethylsilyl trifluoromethane-
sulfonate used). Recovered from the reaction mixture in
The 85:15 mixture of (34) has epimerized to a 80:20
mixture.
almost a quantitative amount was hl-(7-methoxy)cyclo-
heptene-l-yl Fpp (34) with a 10:90 diastereomeric ratio!
Note that the major diastereomer now is the one with a
cyclopentadienyl hydrogen resonance at a higher field.
Methoxy abstraction of the 10:90 mixture of h -(7-
methoxy)cyclohepten-l-yl Fpp with trimethylsilyl trifluoro-
methanesulfonate at 0C in the n-pentane/benzene solvent
gave h2-1,2-cycloheptadiene Fpp (35a) in less than a 10%
yield but with a 80:20 diastereomeric ratio!
The results of the methoxy abstractions were summarized
in Table 2.
Ethoxy abstraction from h -(7-ethoxy)cyclohepten-1-yl
Fpp by trimethylsilyl trifluoromethanesulfonate under the
conditions developed for methoxy abstraction also gave
h2-1,2-cycloheptadiene Fpp (35a). The results of ethoxy
abstraction from enriched diastereomeric mixtures of h -(7-
ethoxy)cycloheptene-1-yl Fpp parallel that of its methoxy
analogue in terms of diastereomeric compositions and percent
yields of h2-1,2-cycloheptadiene Fpp (35a).
We have a few reasons to believe that the diastere-
omeric composition of h2-1,2-cycloheptadiene Fpp (35a) from
methoxy abstraction of various diastereomeric mixtures of
h -(7-methoxy)cyclohepten-1-yl Fpp (34) represents the
initial diastereomeric ratio immediately upon methoxy
abstraction and does not reflect the different solubilities
of the h2-1,2-cycloheptadiene Fpp (35a) diastereomers in
ethyl acetate.
o 0 0 in in
in in m in
r-l
N
4I
0
1-
Cl
0
4-
r.
o
+1
a
a,
0
0
0
0
,C
4-,
I
0o
t-
4r-
0.
.OO
H,1
N
c a
U
H
4r-4
0
41
n
0a m
v ()
00
44
) o
*-V
4-4
mu in
0
*-l --
0
m
4-1
En
cl
4,
O-
Q
N N N N
a) 0
3 .0 .0 .0 .0
0 C) C) C) a)
o o o 0 0
U3C H 0
o o
N 93
0 0
SCo
0 0I i
o 0 0
C 0 0
o o o
0) 0
l- w
V
ra
in
+1
4,
- -,
0
0
0" 0
Oj
(0 .
04
0 >
4-,
cr
0
0
a)
4
.t
4J
0
'I
4--
It
0
4
-1
-)
pB
i4
o
4-,
-l
44
4)
First of all, the ethyl acetate filtrate from the
initial precipitation of h2-1,2-cycloheptadiene Fpp (35a)
was collected and examined for any residual h2-1,2-cyclo-
heptadiene Fpp (35a). We did not detect any trace of
h2-1,2-cycloheptadiene Fpp (35a) by 1H NMR from this
filtrate. Also, repeated precipitations of h2-1,2-cyclo-
heptadiene Fpp (35a) from ethyl acetate did not affect its
diastereomeric compositions.
Lastly, we also tried a different procedure to precipi-
tate h2-1,2-cycloheptadiene Fpp (35) from the reaction
mixture. Instead of precipitating h2-1,2-cycloheptadiene
Fpp (35a) from the crude reaction mixture with ethyl
acetate, we dissolved the crude reaction mixture in methanol
saturated with ammonium hexafluorophosphate and precipitated
the h2-1,2-cycloheptadiene Fpp as its hexafluorophosphate
salt (35b) with water. We found by 1H NMR measurements that
the diastereomeric compositions of h2-1,2-cycloheptadiene
Fpp (35b) obtained by this method are similar to those
obtained by the previous method.
We conclude that methoxy abstraction of h -(7-methoxy)
cyclohepten-l-yl Fpp (34) by trimethylsilyl trifluoro-
methanesulfonate occurs from only one diastereomer. The
other diastereomer merely decomposes in the presence of
trimethylsilyl trifluoromethanesulfonate. Therefore, when a
60:40 diastereomeric ratio of h -(7-methoxy)cyclohepten-1-yl
Fpp (34) was treated with half an equivalent of trimethyl-
silyl trifluoromethanesulfonate, only one diastereomer
reacted to give h2-1,2-cycloheptadiene Fpp (35a) in a 60:40
diastereomeric ratio in almost quantitative yields while the
other diastereomer was recovered from the reaction mixture.
This other diastereomer gave negligible amounts of h2-1,2-
cycloheptadiene Fpp (35a) when treated with trimethylsilyl
trifluoromethanesulfonate.
From the above observation, it becomes clear why
methoxy abstraction in a pentane/benzene solvent from either
a 80:20 or a 60:40 mixture of hl-(7-methoxy)cyclohepten-1-yl
Fpp (34) results in the same 60:40 mixture of h2-1,2-cyclo-
heptadiene Fpp (35a). In each case only one diastereomer,
the major diastereomer, reacted with the trimethylsilyl
trifluoromethanesulfonate.
The choice of solvent for methoxy abstraction affects
the diastereomeric ratio of h2-1,2-cycloheptadiene Fpp
(35a). In a non-polar pentane solvent, we are probably
seeing methoxy loss through Fpp assistance (probably syn Fpp
assistance) leading to an enriched diastereomeric mixture of
h -1,2-cycloheptadiene Fpp (35a). Methoxy loss in a more
polar pentane/benzene solvent probably proceeds via the
allyl cation to give an almost equal diastereomeric mixture
of the h2-1,2-cycloheptadiene Fpp (35a).
At the present moment we are unable to guess as to the
significance of the 80:20 diastereomeric ratio of h2-1,2-
cycloheptadiene Fpp (35a) obtained from the methoxy
abstraction of a 10:90 diastereomeric mixture of h -(7-
methoxy)cyclohepten-1-yl Fpp (34).
Our conjecture as to why only one h1-(7-methoxy)-
cyclohepten-1-yl Fpp (34) diastereomer undergoes methoxy
abstraction to give the desired complex (35a) is that only
this diastereomer has its methoxy group accessible to
trimethylsilyl trifluoromethanesulfonate. The methoxy group
of the other diastereomer is occluded from trimethylsilyl
trifluoromethanesulfonate.
Based upon extended Huckel calculations of Seeman and
Davies44 and X-ray crystal structures45 of similar types of
complexes, we arrived at what we believe to be the most
stable conformers of each of the two hl-(7-methoxy)cyclo-
hepten-l-yl Fpp (34) diastereomers shown below.
OCH3 H
Fe --*1-- Fe
OC H C OCH3
02P 02?--<>P
SR SS
The phosphorus to phenyl bond here is eclipsed with the iron
to (7-methoxy)cyclohepten-1-yl bond forcing the planes of
both rings to lie parallel to each other. This conformation
places the methoxy group of one diastereomer (SR) above the
plane of the cycloheptene ring and places the methoxy group
of the other diastereomer (SS) below the plane of the
cycloheptene ring. The methoxy group of the diastereomer
(SR) is therefore exposed to trimethylsilyl trifluoro-
methanesulfonate whereas the methoxy group of the other
diastereomer (SS) is shielded from abstraction by the phenyl
ring below. MMX calculations for the two diastereomers of
hl-(7-methoxy)cyclohepten-l-yl Fpp (34) confirmed the
conformation of the structures drawn above.
Let us now consider the consequences of an Fpp cation
alternating between the adjacent double bonds of the 1,2-
cycloheptadiene in terms of its IH NMR spectrum and
diastereomeric ratio based upon the two schemes described
below.
In one case, the Fpp fragment moves between the
two allenic double bonds as shown in Scheme XIII. The allyl
cation intermediate is not involved in the 1,2-Fpp shift and
the stereochemical integrity of each diastereomer is
retained; ie. the SR diastereomer (and its RS enantiomer)
does not interconvert with its RR (or SS) diastereomer.
Scheme XIII.
OC 03P-- Fe -CO
03P-e Fe+ Fe+ .09 4CO
H1 \ H3 H 3
H H H 1 H
SR SR SR
A rapid 1,2-Fpp shift in the NMR time scale will render the
two allenic hydrogens of each diastereomer equal. This
effect should be similar to that observed in the case of
h2-1,2-cycloheptadiene Fp (4) when the allenic hydrogen
resonances coalesce. In this case both diastereomers would
still be observed and their diastereomeric composition after
coalescence would remain unchanged. In other words, the
two sets of allenic hydrogen resonances will coalesce and at
the same time the two cyclopentadienyl hydrogen resonances
will remain distinct and separated.
In the other case involving an allyl cation inter-
mediate as shown in Scheme XIV, both cyclopentadienyl
hydrogen resonances and the two pairs of the allenic hydro-
gen resonances would coalesce.
|
