Comparative studies of thermally induced homolytic carbon-carbon bond cleavage reactions of strained dicarba[2]ferrocenophanes and their ring-opened oligomers and polymers

Article abstract of DOI:10.1002/chem.201304396

Reactivity studies of dicarba[2]ferrocenophanes and also their corresponding ring-opened oligomers and polymers have been conducted in order to provide mechanistic insight into the processes that occur under the conditions of their thermal ring-opening polymerisation (ROP) (300°C). Thermolysis of dicarba[2]ferrocenophane rac-[Fe(η⁵-C₅H ₄)₂(CHPh)₂] (rac-14; 300°C, 1 h) does not lead to thermal ROP. To investigate this system further, rac-14 was heated in the presence of an excess of cyclopentadienyl anion, to mimic the postulated propagating sites for thermally polymerisable analogues. This afforded acyclic [(η⁵-C₅H₅)Fe(η⁵-C ₅H₄)-CH₂Ph] (17) through cleavage of both a Fe-Cp bond and also the C-C bond derived from the dicarba bridge. Evidence supporting a potential homolytic C-C bond cleavage pathway that occurs in the absence of ring-strain was provided through thermolysis of an acyclic analogue of rac-14, namely [(η⁵-C₅H₅) Fe(η⁵-C₅H₄)(CHPh)₂-C ₅H₅] (15; 300°C, 1 h), which also afforded ferrocene derivative 17. This reactivity pathway appears general for post-ROP species bearing phenyl substituents on adjacent carbons, and consequently was also observed during the thermolysis of linear polyferrocenylethylene [Fe(η⁵-C₅H₄)₂(CHPh) ₂]_n (16; 300°C, 1 h), which was prepared by photocontrolled ROP of rac-14 at 5°C. This afforded ferrocene derivative [Fe(η⁵-C₅H₄CH₂Ph)₂] (23) through selective cleavage of the -H(Ph)C-C(Ph)H- bonds in the dicarba linkers. These processes appear to be facilitated by the presence of bulky, radical-stabilising phenyl substituents on each carbon of the linker, as demonstrated through the contrasting thermal properties of unsubstituted linear trimer [(η⁵-C₅H₅)Fe(η⁵- C₅H₄)(CH₂)₂(η⁵-C ₅H₄)Fe(η⁵-C₅H ₄)(CH₂)₂(η⁵-C₅H ₄)Fe(η⁵-C₅H₅)] (29) with a -H₂C-CH₂- spacer, which proved significantly more stable under analogous conditions. Evidence for the radical intermediates formed through C-C bond cleavage was detected through high-resolution mass spectrometric analysis of co-thermolysis reactions involving rac-14 and 15 (300°C, 1 h), which indicated the presence of higher molecular weight species, postulated to be formed through cross-coupling of these intermediates. Getting ROPed in: Studies of the mechanistic processes that occur under the conditions of thermal ring-opening polymerization (ROP) of dicarba[2] ferrocenophanes have revealed a homolytic C-C bond cleavage pathway that also proceeds in the absence of ring-strain (see scheme).

Full text of DOI:10.1002/chem.201304396

DOI: 10.1002/chem.201304396
Full Paper
&
CÀC Bond Cleavage
Comparative Studies of Thermally Induced Homolytic CarbonÀ
Carbon Bond Cleavage Reactions of Strained
Dicarba[2]ferrocenophanes and Their Ring-Opened Oligomers and
Polymers
Andrew D. Russell,[a] Joe B. Gilroy,[a, b] and Ian Manners*[a]
Abstract: Reactivity studies of dicarba[2]ferrocenophanes
and also their corresponding ring-opened oligomers and
polymers have been conducted in order to provide mecha-
nistic insight into the processes that occur under the condi-
tions of their thermal ring-opening polymerisation (ROP)
(3008C). Thermolysis of dicarba[2]ferrocenophane rac-[Fe(h5-
C5H4)2(CHPh)2] (rac-14; 3008C, 1 h) does not lead to thermal
ROP. To investigate this system further, rac-14 was heated in
the presence of an excess of cyclopentadienyl anion, to
mimic the postulated propagating sites for thermally poly-
merisable analogues. This afforded acyclic [(h5-C5H5)Fe(h5-
C5H4)-CH2Ph] (17) through cleavage of both a FeÀCp bond
and also the CÀC bond derived from the dicarba bridge. Evi-
dence supporting a potential homolytic CÀC bond cleavage
pathway that occurs in the absence of ring-strain was pro-
vided through thermolysis of an acyclic analogue of rac-14,
namely [(h5-C5H5)Fe(h5-C5H4)(CHPh)2-C5H5] (15; 3008C, 1 h),
which also afforded ferrocene derivative 17. This reactivity
pathway appears general for post-ROP species bearing
phenyl substituents on adjacent carbons, and consequently
was also observed during the thermolysis of linear polyferro-
cenylethylene [Fe(h5-C5H4)2(CHPh)2]n (16; 3008C, 1 h), which
was prepared by photocontrolled ROP of rac-14 at 58C. This
afforded ferrocene derivative [Fe(h5-C5H4CH2Ph)2] (23)
through selective cleavage of the ÀH(Ph)CÀC(Ph)HÀ bonds
in the dicarba linkers. These processes appear to be facilitat-
ed by the presence of bulky, radical-stabilising phenyl sub-
stituents on each carbon of the linker, as demonstrated
through the contrasting thermal properties of unsubsti-
tuted linear trimer [(h5-C5H5)Fe(h5-C5H4)(CH2)2(h5-C5H4)Fe(h5-
C5H4)(CH2)2(h5-C5H4)Fe(h5-C5H5)] (29) with
a
ÀH2CÀCH2À
spacer, which proved significantly more stable under analo-
gous conditions. Evidence for the radical intermediates
formed through CÀC bond cleavage was detected through
high-resolution mass spectrometric analysis of co-thermoly-
sis reactions involving rac-14 and 15 (3008C, 1 h), which in-
dicated the presence of higher molecular weight species,
postulated to be formed through cross-coupling of these
intermediates.
ligands.[8] Computational studies indicate that [n]metalloceno-
phanes containing more than four d electrons possess substan-
tial ring-strain upon tilting of the cyclopentadienyl (Cp) ligands
away from their preferred parallel conformation,[9] as defined
quantitatively by the angle a, along with the other structural
distortions illustrated (Figure 1).
Introduction
Strained [n]metallocenophanes 1 are a broad class of cyclic or-
ganometallic species that have received significant attention
since the first example was reported in 1960.[1] The expanding
field of [n]metallocenophane chemistry has been facilitated
through the ability to vary both the central metal atom and
bridging moiety, allowing a wide variety of these species to be
realised,[2] including those of Fe,[3] Ru,[4] Co,[5] Ni[6] and Mo,[7] in
addition to analogous species containing other p-hydrocarbon
[a] A. D. Russell, Prof. J. B. Gilroy, Prof. I. Manners
School of Chemistry, University of Bristol
Cantocks’ Close, Bristol BS8 1TS (UK)
E-mail: ian.manners@bristol.ac.uk
[b] Prof. J. B. Gilroy
Present address: Department of Chemistry
University of Western Ontario, 1151 Richmond Street
London, ON, N6A 5B7 (Canada)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201304396.
Figure 1. Structural metrics for [n]metallocenophanes.
Chem. Eur. J. 2014, 20, 4077 – 4085
4077
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Consequently, when the ansa bridge is suitably short (nꢀ2)
affording a substantial tilt angle (a=14–338), the inherent
ring-strain can be exploited in ring-opening reactions includ-
ing, most significantly, ring-opening polymerisation (ROP).[10]
ROP provides access to metallopolymers 2 that incorporate
metallocene (or related) units in the main chain. To date, these
materials have attracted attention for a variety of applications,
including uses as etch resists,[11] catalytic and magnetic ceramic
precursors,[12] redox-active microcapsules and microspheres,[13]
the redox-active component of photonic crystal displays,[14]
and as self-assembled, nanostructured materials.[15]
excess of MgCp2, confirmed that the FeÀCp bond could be
cleaved heterolytically under the conditions employed in ther-
mal ROP, suggesting that propagation likely proceeds through
this mechanism (Scheme 3).
Dicarba[2]metallocenophanes 1 (M=Fe, Ru, Co; EnRx =C2R4)
are a subclass of [n]metallocenophanes that display diverse re-
activity including thermal ROP,[16] photolytic ROP through
cleavage of the MÀCp bond,[17] and (for species 3) metalÀmetal
bond formation upon oxidation, affording a dicarba[2]rutheno-
cenophanium dimer.[18]
Scheme 3. Thermal ring-opening reactions of dicarba[2]ferrocenophanes 4
and 5 in the presence of an excess of MgCp2.
In contrast to 4 and 5, species 11 and meso/rac-13 undergo
either radical disproportionation, to afford substituted ferro-
cene 12, or isomerisation, to yield the thermodynamically
more stable rac-13 isomer, respectively (Scheme 4 and
Scheme 5).[20] The preferred reactivity pathway is dependent
on the presence or absence of hydrogen substituents at the
carbon in the a position respective to the dicarba bridge.
Studies on the thermal ROP of dicarba[2]ferrocenophanes 4
and 5 have indicated that the resulting cyclic polyferrocenyl-
ethylenes have moderate molecular weight (Mw =4800–
96000 Da) and broad distributions (PDI=1.1–3.6; Scheme 1).
Scheme 4. Thermally induced radical disproportionation of dicarba[2]metal-
locenophanes.
Scheme 1. Thermal ROP of dicarba[2]metallocenophanes.
Insight into the mechanism was achieved through comparative
NMR spectroscopic analysis of random copolymer 7 with the
deuterated analogue 8 (Scheme 2).[19] The absence of NMR sig-
nals which could be assigned to a ÀCH(Ph)CD2À linker in 8 in-
dicated that a mechanism of thermal ROP based upon homo-
lytic CÀC bridge bond cleavage was improbable. In addition,
trapping of the ring-opened monomeric species 9 and 10 by
a thermolysis reaction (3008C, 1 h) in the presence of an
Scheme 5. Thermally induced isomerisation of dicarba[2]metallocenophanes.
These results indicate that an interesting structure–thermal
reactivity relationship exists for dicarba[2]ferrocenophanes.
Thus, species bearing either zero or one non-hydrogen sub-
stituents on the dicarba bridge undergo thermal ROP by FeÀ
Cp bond cleavage. Alternatively, species bearing an equal
number of non-hydrogen substituents undergo homolytic CÀC
bridge bond cleavage, followed by radical disproportionation
or isomerisation.
In order to provide further insight into the reaction path-
ways that take place under thermal polymerisation conditions
for dicarba[2]ferrocenophanes, we now report thermal reactivi-
ty studies of a diphenyl-substituted dicarba[2]ferrocenophane
rac-14 [M=Fe, EnRx =rac-(CHPh)2], its ring-opened analogue
(15) and linear polyferrocenylethylene [Fe(h5-C5H4)2(CHPh)2]n
Scheme 2. Random thermal copolymerisations of dicarba[2]ferrocenophane
5 with either 4 and 6, respectively.
Chem. Eur. J. 2014, 20, 4077 – 4085
www.chemeurj.org
4078
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Scheme 6. Thermolysis of dicarba[2]ferrocenophane rac-14 in the presence
of excess MgCp2.[24]
Figure 2. Dicarba[2]ferrocenophane rac-14, acyclic analogue 15 and polyfer-
rocenylethylene 16.
occurs, to afford the rac-14 isomer exclusively. The distorted
structure of rac-14 [a=22.7(3)8, CÀC bridge=1.572(3) ꢁ],[17b]
which has both an increased a angle and CÀC bridge bond
length relative to those reported for both 4 [a=21.6(4)8, CÀC
bridge=1.539(12) ꢁ][16b] and 5 [a=21.18(9)8, CÀC bridge=
1.545(3) ꢁ],[19] has previously been used as a rationale for the
observed differences in reactivity. Thus, homolytic CÀC bond
cleavage of the C2-bridge is detected for the former species
and heterolytic FeÀCp bond cleavage for the latter two ferro-
cenophanes.[19] However, in contrast to these previous studies,
formation of 17 from rac-14 and MgCp2 at 3008C is apparently
facilitated through cleavage of both the CÀC bond derived
from the bridge and the FeÀCp bond.
(16; Figure 2). The potential for coupling reactions of the radi-
cal intermediates formed during thermally induced, homolytic
cleavage of the dicarba bridges and spacers has also been
explored.
Results and Discussion
Investigations into the thermal lability of the ÀH(Ph)CÀ
C(Ph)HÀ bond in the presence and absence of ring-strain
Thermal reactivity of rac-14, and ring-opened species 15a
and 15b: Our first experiments aimed to extend the studies
used to elucidate the mechanism of the thermal ROP of dicar-
ba[2]ferrocenophanes to other strained dicarba[2]ferroceno-
phanes that do not polymerise.[21] As noted above, we have
previously shown that thermolysis (3008C, 1 h) of thermally
polymerisable metallocenophanes 4 and 5 in the presence of
a large excess of MgCp2 yields the ring-opened monomeric an-
alogues 9 and 10 by cleavage of the FeÀCp bond (Scheme 3).
In contrast, when we heated thermal ROP-resistant rac-14
under identical reaction conditions, the only product success-
fully isolated was ferrocene de-
Two of the most likely mechanisms for the formation of 17
are shown in Scheme 7. Either of these processes could oper-
ate, or both may occur together. Homolytic cleavage of the CÀ
C bridge bond could occur to afford diradical intermediate 19
(mechanism A), a process analogous to that postulated during
the thermolysis of species similar to rac-14, such as 11 and
meso/rac-13.[20] Computational studies of the diradical inter-
mediate 19, which results from homolytic CÀC bridge bond
cleavage,[25] suggest it exists as a resonance hybrid with an al-
ternative resonance form consisting of Fe0 bis-fulvenyl species
rivative 17 in a moderate yield
(50%; Scheme 6). NMR spectro-
scopic analysis of the reaction
mixture indicated that no start-
ing material remained following
the reaction.[22]
The microcrystalline yellow
product 17 was characterised by
1
both H and 13C NMR spectrosco-
py, which gave data consistent
with previous reports describing
the synthesis of 17 by reduction
of a ferrocenyl benzyl alcohol
precursor.[23] High and low-reso-
lution MS data, in addition to el-
emental analysis, were obtained
and were also found to be
consistent with the assigned
structure.
The observed reactivity differs
from the thermolysis reaction of
meso/rac-14 alone at 3008C, for
which isomerisation by homolyt-
ic CÀC bridge bond cleavage Scheme 7. Potential mechanisms of formation of monosubstituted ferrocene 17.
Chem. Eur. J. 2014, 20, 4077 – 4085 www.chemeurj.org ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4079

Full Paper
20.[26] Nucleophilic attack on intermediate 19 could then afford
could be assigned to a product derived from the (C5H5)CH(Ph)
fragment, expected to also form following cleavage of the
ÀH(Ph)CÀC(Ph)HÀ bond. Although the fate of this organic frag-
ment is unknown, it is postulated that it is not stable at the
elevated temperatures employed for the thermal bond cleav-
age reaction.[27] Indeed, a thermolysis reaction of 22, synthes-
ised through a previously published procedure,[28] confirmed
its thermal instability with significant degradation occurring.[29]
Further insight was obtained, however, through low-resolution
MS studies of starting material 15 for which, in addition to
a peak for 21 (275.2 m/z, 100%), a peak at 153.2 m/z (20%) as-
signed to the fulvene 22ÀH, was observed (Figure S1 in the
Supporting Information). Fragment 22 presumably loses H
during the MS analysis. Thus, a mechanism for the thermal
degradation of 15a and 15b was postulated whereby, upon
thermolysis, the CÀC spacer cleaves, affording radical inter-
C
21, which might abstract H to afford 17. The alternative mech-
anism B involves heterolytic cleavage of the FeÀCp bond in ac-
cordance with the observed reactivity discussed previously,[19]
affording the ring-opened intermediate 18. This intermediate
could then undergo thermally induced CÀC cleavage to also
yield intermediate 21.
To investigate the likelihood of the CÀC bond cleavage pro-
cess that occurs after ring-opening, in the absence of ring-
strain (mechanism B), thermolysis of ring-opened species 15
was examined. To prepare this species, irradiation of dicarba[2]-
ferrocenophane rac-14 with Pyrex filtered UV light in the pres-
ence of five equivalents of NaCp (at 208C) was performed. This
afforded the ring-opened product as a mixture of isomers (15a
and 15b) in good yield (69%; Scheme 8).
C
mediate 21, which abstracts H from the fragmented intermedi-
ate to afford species 17 and 22, respectively (Scheme 10).[30]
Scheme 8. Photolytic ring-opening of rac-14 in the presence of excess
NaCp.
The two isomers (15a and 15b) were isolated in a 1:1.3
ratio, and were confirmed through the identification of distinc-
1
tive resonances in both the H NMR spectrum (singlets at d=
3.74 and 3.72 ppm assigned to the h5-C5H5 groups of each
isomer) and 13C{1H} NMR spectrum (signals at d=93.5 and
93.1 ppm assigned to the Cpipso carbons of each isomer). Un-
fortunately, all attempts to separate the isomers proved unsuc-
cessful. Confirmation of the structure of 15 was also provided
through elemental analysis and high-resolution MS. Low-reso-
lution MS also identified the parent ion at 430.4 m/z
(12%, M+).
Scheme 10. Potential mechanism of formation for 17 and 22 (not observed).
C
The proposed mechanism of H transfer was investigated
through deuterium labelling of the pendant Cp ring of 15a
and 15b. NaOD in D2O was added to a dioxane solution of
15a and 15b (for details see the Supporting Information).
Upon work-up, deuteration to form [D5]15a and [D5]15b was
confirmed through NMR spectroscopic analysis. Thus, charac-
teristic signals between 6.65 and 5.90 ppm, assigned to the
olefinic protons, and between 3.00 to 2.50 ppm, assigned to
the CH2 group of the pendant Cp ring, were observed in the
With compounds 15a and 15b in hand, the thermal stability
of the dicarba spacer was then investigated. After 5 min at
3008C 15a and 15b melted into a dark red, free-flowing fluid
and remained in this state for the duration of the thermolysis
experiment (1 h). Following work-up, 17 was isolated in high
yield (91%; Scheme 9). The formation of 17 confirmed that the
dicarba spacer, at least in species bearing phenyl substituents
on both carbons, can be cleaved in the absence of ring-strain.
1
2H NMR spectrum, and were absent from the H NMR spectrum
(Figure S2 in the Supporting Information).
1
Upon thermolysis of [D5]15a and [D5]15b (3008C, 1 h), H
2
and H NMR analysis revealed 38% incorporation of D onto the
CH(Ph) group of the product (17; Figure S3 in the Supporting
Information).[31] Confirmation was also provided through
13C NMR spectroscopic analysis, which displayed a triplet at
36.2 ppm, with characteristic deuterium splitting, assigned to
-CHD(Ph) carbon. These observations suggest that although
the proposed mechanism does occur in the melt, other reactiv-
ity pathways are also operative. However, given the high reac-
tivity of radical species, the reaction temperatures employed
and the formation of other uncharacterised by-products, it is
perhaps unsurprising that non-deuterated 17 is also a product
from the thermolysis reaction.
1
It is important to note that H NMR spectroscopic analysis of
the crude reaction mixture did not display any signals that
Thermal reactivity of polyferrocenylethylene 16: To probe
the potential of thermally induced carbonÀcarbon bond cleav-
Scheme 9. Thermolysis of 15a and 15b (only one isomer is shown) at 3008C
to afford 17.
Chem. Eur. J. 2014, 20, 4077 – 4085
www.chemeurj.org
4080
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
age as a possible depolymerisation pathway, the thermal sta-
bility of polyferrocenylethylenes bearing phenyl substituents
on adjacent carbons of the dicarba spacer was investigated. To
this end, samples of linear polyferrocenylethylene 16 with
a narrow molecular weight distribution (Mw =21800, PDI=
1.21) were prepared by photocontrolled ROP (58C, one
week).[17b] These were then thermolysed at 3008C in evacuated
thermolysis tubes for either 1 h (to give 16a), 5 h (to give
16b), or 16 h (to give 16c). During each thermolysis reaction
the mobility of the sample noticeably increased, consistent
with depolymerisation of 16. After each reaction the sample
was dissolved in THF and filtered, upon which a small amount
of black material was separated.[32]
thermore, the intermediate formed from this process will pre-
sumably be in a conformation in which the two radical sites
are approximately trans to one another. Given the high reactiv-
ity of radical species, coupled with the statistical improbability
of two adjacent ÀH(Ph)CÀC(Ph)HÀ bonds breaking simultane-
ously and combined with the need for rotation around the
FeÀCp bond to enable radical site proximity, it is not surprising
that the thermolysis of 16 results in the formation of 23, and
not rac-14.
NMR spectroscopic and gel permeation chromatographic
(GPC) analysis of polyferrocenylethyene samples 16a (3008C,
1 h) and 16b (3008C, 5 h) indicated that the product was of re-
duced molecular weight and increased polydispersity (PDI) rel-
ative to 16 (Mw =21800 gmolÀ1, PDI=1.21; Figure S4 in the
Supporting Information). Polymer 16b (Mw =3600 gmolÀ1,
PDI=1.32) underwent a greater reduction in molecular weight
Disubstituted ferrocene 23 was isolated from the mother liq-
uors following precipitation of the residual polyferrocenylethy-
lene (16a–b) or unidentified (16c) by-products from the crude
reaction mixtures. The yield of 23 varied depending on the re-
action time and was highest (68%) after a 5 h thermolysis ex-
periment (Scheme 11).[33] 1H NMR spectroscopy and high-reso-
lution MS data for 23 were consistent with that reported when
this species was previously synthesised by reduction of 1,1-di-
benzoylferrocene.[34]
than 16a (Mw =6300 gmolÀ1
, PDI=1.76), suggesting that
longer thermolysis times result in a greater reduction in the
final molecular weight of the polymeric by-products.[35] Inter-
1
estingly, H NMR spectroscopic analysis of both 16a and 16b
did not display any signals that could be assigned to the diene
end group, which suggested that these moieties potentially
participate in additional reactivity, or alternatively, are cleaved
from the polymer backbone upon thermolysis. NMR spectro-
scopic characterisation of the residual species 16c, isolated
from the 16 h thermolysis experiment, was impeded by the
presence of paramagnetic impurities.[36]
Cleavage of the ÀH(Ph)CÀC(Ph)HÀ bonds in 16 results in sig-
nificant depolymerisation, a result which is in contrast to previ-
ously reported thermolysis reactions (3008C, 1 h) on cyclic
polyferrocenylethylene 24,[16b] bearing no non-hydrogen substi-
tutents on the dicarba linker, and linear 25,[19] bearing a phenyl
substituent on only one carbon of the linker (Figure 3). In
Scheme 11. Thermolysis of 16 (Mw =21800 Da, PDI=1.21) at 3008C.
The radical intermediates formed during the thermal depoly-
C
merisation of 16 apparently abstract H to afford 23. Work-up
of the thermolysis reactions of 16 (and 15a/b) with deuterated
C
solvent (CDCl3) confirmed that H is not derived from this
source. Identical products were isolated, as confirmed by both
NMR spectroscopic and mass spectrometric analysis of 23 (and
17), respectively. For both 15a/b and 16, which are thermo-
C
lysed alone, it can therefore be concluded that H is likely pro-
Figure 3. Polyferrocenylethylenes 24 and 25.
vided by a CÀC bond cleavage product. Radical coupling to
reform the dicarba CÀC bond to produce monomer rac-14, as
observed during the previously reported thermal isomerisation
of meso/rac-14 to rac-14,[19] does not result from thermolysis of
16. The differences in the products formed in each case can be
rationalised through examination of the reactivity pathways. In
the case of the thermolysis of meso/rac-14, cleavage of the di-
carba bridge affords intermediate 19 (Scheme 7), with two rad-
ical CH groups in close proximity to each other which, follow-
ing CpCÀCH(Ph) bond rotation, can couple rapidly to reform
the dicarba bridge and thus convert to the thermodynamically
preferred rac-14 isomer. However, for the depolymerisation of
these examples, analysis of the thermolysed polymers indicat-
ed no significant change in the molecular weight for the
former, and a slight increase in both the molecular weight and
PDI in the case of the latter. The change in the observed Mw
and PDI for 25 was postulated to result from either thermally
induced branching, crosslinking, or reactivity of the diene end-
group.[19]
Few examples exist in the literature of polymers that contain
bulky phenyl substituents on adjacent carbons of the back-
16 to also afford rac-14 as a product, two adjacent ÀH(Ph)CÀ bone. Consistent with our observations, head-to-head polystyr-
C(Ph)HÀ bonds of the polymer backbone must break simulta-
neously to produce the same diradical intermediate (19). Fur-
ene 28, synthesised through the radical polymerisation of di-
phenyl butadiene,[37] undergoes thermally induced cleavage of
Chem. Eur. J. 2014, 20, 4077 – 4085
www.chemeurj.org
4081
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
ucts that would be expected to
result from cleavage of the dicar-
ba spacer were detected, al-
though recovered starting mate-
rial was obtained in a substantial-
ly decreased yield (26%) as
a result of the formation of met-
Scheme 12. Synthesis of head-to-head polystyrene 28.[37]
allic iron by-products. It appears
that competing reactivity path-
the ÀH(Ph)CÀC(Ph)HÀ bond at 2808C, to afford a variety of
products (Scheme 12).[38]
ways are in operation under the thermolysis conditions em-
ployed, including thermal decomposition to afford metallic
iron, which appears to be operative in all cases studied.[43] Sig-
nificantly, under standard thermal ROP conditions the ligand
framework of trimer 29 does not appear to undergo carbonÀ
carbon bond cleavage to afford ferrocene based products.
Interestingly, low-resolution MS analysis of 29, which
showed the presence of the trimer parent ion, also displayed
additional peaks at m/z 333.1 (42%, M+ÀFeC17H17) and 199.1
(100%, M+ÀFe2C23H23). The peak at 333.1 was assigned to the
[(h5-C5H5)Fe(h5-C5H4)(CH2)2(h5-C5H4)Fe+] fragment, which results
from cleavage of an FeÀCp bond. Furthermore, the peak at
199.1 m/z was assigned to the (h5-C5H5)Fe(h5-C5H4)CH fragment
believed to result from homolytic cleavage of the CÀC spacer
in trimer 29. This latter observation suggests that MS analysis
can induce homolytic cleavage of this bond although the con-
ditions employed in the MS do not exceed temperatures of
2008C (see the Supporting Information). It is possible that,
once ionised under MS conditions, 29 can undergo more facile
homolytic cleavage of the dicarba spacer than for the neutral
species.
Thermolysis of unsubstituted linear trimer 29: To further
explore the variation in the thermal stability of acyclic ferroce-
nylethylenes bearing different substituents on the dicarba
linker, the linear trimer 29, bearing no non-hydrogen substitu-
ents was prepared. This was achieved through reaction of 9,
prepared using a previously published protocol,[17a] with nBuLi
and FeCl2, affording the linear trimer 29 as a yellow powder in
a good yield (78%; Scheme 13).[39] Formation of trimer 29
Scheme 13. Synthesis of linear trimer 29.
allows the thermal stability of the dicarba bridge in the ab-
sence of non-hydrogen substituents to be ascertained, without
the potential for additional reactivity pathways arising from cy-
clopentadiene end-group loss and CÀC bond labilisation, as
was observed during the thermolysis reactions of both 9,[40]
and polyferrocenylethylene 16.
The trimer 29 was characterised by NMR spectroscopy. The
1H NMR spectrum displayed a distinctive singlet resonance at
d=4.11 ppm, assigned to the h5-C5H5 groups, a multiplet at
4.07 and two broad singlets at 4.03 and 4.01 ppm assigned to
the h5-C5H4 groups and a singlet at 2.53 ppm assigned to the
protons of the dicarba bridge. In addition, the trimer was ana-
lysed by high-resolution MS, which confirmed the presence of
the trimer parent ion at 610.0707 m/z. The electrochemical
properties of the trimer were also investigated, and found to
be consistent with previous studies on analogous species (Fig-
ure S5 in the Supporting Information).[41a,16b,41b] Attempts to
produce crystals suitable for X-ray crystallographic analysis,
however, were unsuccessful.
The relative thermal stability of 29 under the identical reac-
tion conditions (3008C, 1 h) employed to induce the carbonÀ
carbon bond cleavage in 15 and 16 indicates that the pres-
ence of the bulky phenyl substituents on adjacent carbons of
the dicarba bridge facilitates the selective cleavage of the
ÀH(Ph)CÀC(Ph)HÀ bond. Consistent with well-established prec-
edent, the steric-strain, provided through the presence of
phenyl substituents on adjacent carbons, acts to weaken the
carbonÀcarbon bond.[44] In addition, the phenyl substituents
can stabilise the radical intermediates that result from homo-
lytic CÀC bond cleavage through resonance effects. These con-
tributing factors presumably facilitate the observed carbonÀ
carbon bond cleavage under the thermolysis conditions em-
ployed and, importantly, allow the conclusion to be made that
in these and the previous studies discussed, ring-strain is not
a requirement for this process.
Cross-coupling reactions of radical intermediates
The trimer 29 was thermolysed at 3008C for 1 h. The sample
melted to afford a viscous red material after 5 min, with no fur-
ther change being observed for the remaining duration of the
attempted thermolysis. Upon being cooled, the tube was
opened in air and subsequent work-up gave an almost quanti-
tative recovery (yield=94%) of unreacted 29 as identified by
NMR spectroscopy and MS.[42] To observe the extent of the
thermal stability of trimer 29, a thermolysis experiment was
conducted over an increased time period (5 h). Again, no prod-
Trapping of the postulated radical intermediates in the afore-
mentioned thermally induced carbonÀcarbon bond cleavage
reactions was likely to be extremely difficult as the bonds
formed to radical spin traps such as nitroxides are also expect-
ed to be labile at 3008C.[45] Furthermore, the extreme condi-
tions of thermal ROP precluded investigation into a radical
mechanism by EPR spectroscopy. However, we anticipated that
the proposed radicals were likely to participate in random radi-
Chem. Eur. J. 2014, 20, 4077 – 4085
www.chemeurj.org
4082
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
cal coupling reactions with each other under thermolysis con-
ditions to afford mixtures of cross-products. We, therefore, per-
formed a series of co-thermolysis experiments to explore this
possibility.
ty of the radical sites in short-lived intermediate 19
(Scheme 7), prohibits intermolecular coupling reactions in the
absence of other radical species. However, contrasting reactivi-
ty was observed from the thermolysis of 15 alone (Scheme 9),
where analysis of the crude product by low-resolution MS con-
tained a peak at 550.1 m/z (10%) assigned to species 30 (Fig-
ure S7 in the Supporting Information). Identification of 30 indi-
cated that, in contrast to intermediate 19, the lifetime of radi-
cal intermediate 21 may be sufficiently long to facilitate inter-
molecular coupling reactions to give a dimeric product.
Importantly, low-resolution MS analysis of the pure starting
materials rac-14 and 15 did not show any peaks associated
with any higher molecular weight species. This allows the con-
clusion to be made that products 30, 31 and 32 directly result
from the thermolysis reactions and are not formed simply from
rac-14 and 15 during the analysis.
Our initial study involved a co-thermolysis of equimolar
amounts of the dicarba[2]ferrocenophane rac-14 and the ring-
opened monomer 15 at 3008C for 1 h. After work-up, two
products were isolated, 17 (yield=44%) and 23 (yield=
32%),[46] and their identities were confirmed by 1H NMR and
MS analysis (Scheme 14). High-resolution MS analysis of the
The reported observations allow conclusions to be made
concerning the potential reactivity pathways of rac-14 in the
presence of an excess of the Cp anion at 3008C (Scheme 7).
Given the short life-time of intermediate 19, as indicated in the
previously reported isomerisation of meso/rac-14 to rac-14,[49]
and the result that this species does not undergo intermolecu-
lar radical coupling reactions with itself to give dimeric prod-
ucts, it is unlikely that the reaction of 19 with NaCp could
occur sufficiently fast to facilitate complete conversion of rac-
14 to 17 in the 1 h reaction time.
Furthermore, the observation that ring-strain is a requirement
for FeÀCp bond cleavage in dicarba[2]ferrocenophanes indi-
cates that the use of a large excess of CpÀ could induce heter-
olytic cleavage of the FeÀCp bond in rac-14, and that FeÀCp
bond cleavage is disfavoured following dicarba CÀC bridge
cleavage (mechanism A). From the thermal reactivity studies of
species 15 and 16, the dicarba
Scheme 14. Co-thermolysis of rac-14 and 15a and 15b (only one isomer is
shown).
residue (Figures S8–S12 in the Supporting Information) directly
after the thermolysis showed, in addition to 17 and 23, the
presence of species 30, 31 and 32,[47] which were formed in
yields too low to permit subsequent isolation (Figure 4).[48]
For comparison, an analogous thermolysis of rac-14 alone
was performed at 3008C for 1 h and the crude product was an-
alysed by low-resolution MS. Interestingly no peaks were ob-
served that could be assigned to the radical coupled product
32, and rac-14 was the sole species detected (Figure S6 in the
Supporting Information). This result suggests that the proximi-
spacer in the resulting inter-
mediate (18) can be expected to
undergo rapid homolytic cleav-
age under the thermolysis condi-
tions employed, thus implicating
mechanism B (Scheme 7).
Conclusion
New insight into the mechanistic
processes that occur under the
conditions of the thermal ROP of
dicarba[2]ferrocenophanes has
been revealed through studies
of the thermal reactivity of rac-
14, and also the corresponding
ring-opened oligomers 15 and
polyferrocenylethylene 16. Our
results indicated the presence
of thermally induced homolytic
cleavage of CÀC bonds, with
bulky Ph groups as substituents
in the C2 bridge or spacer,
occurs selectively to afford the
Figure 4. Low-resolution MS analysis of the co-thermolysis reaction of rac-14 and 15. Peak for 32 is overlapping
with a peak at 730.1 m/z assigned to a linear dimer 35 (for details see Figure S10 in the Supporting Information).
Chem. Eur. J. 2014, 20, 4077 – 4085
www.chemeurj.org
4083
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Int. Ed. 2004, 43, 3321–3325; c) J. A. Schachner, S. Tockner, C. L. Lund,
J. W. Quail, M. Rehahn, J. Mꢂller, Organometallics 2007, 26, 4658–4662.
[5] a) M. J. Drewitt, S. Barlow, D. O’Hare, J. M. Nelson, P. Nguyen, I. Manners,
Chem. Commun. 1996, 2153–2154; b) S. Fox, J. P. Dunne, M. Tacke, D.
Schmitz, R. Dronskowski, Eur. J. Inorg. Chem. 2002, 3039–3046; c) H.
Braunschweig, F. Breher, M. Kaupp, M. Gross, T. Kupfer, D. Nied, K. Ra-
dacki, S. Schinzel, Organometallics 2008, 27, 6427–6433; d) U. F. J.
Mayer, J. P. H. Charmant, J. Rae, I. Manners, Organometallics 2008, 27,
1524–1533.
acyclic ferrocene derivatives 17 or 23. Comparison with the
corresponding thermal behaviour of the linear trimer 29, a spe-
cies that lacks phenyl substituents and which is significantly
more stable under identical reaction conditions, indicated that
the presence of the bulky aryl substituents on adjacent car-
bons facilitates the selective CÀC bond cleavage reactions.
Furthermore, the observation that these cleavage reactions
occur for both the ring-opened species 15 and 16 as well as
for the dicarba[2]ferrocenophane rac-14 indicated that the
presence of ring-strain is not a prerequisite for this type of pro-
cess. Consistent with theory, the steric strain arising from the
presence of the bulky Ph substituents on adjacent carbons
presumably provides a key thermodynamic driving force for
bond cleavage, and the resonance stabilisation of the radical
intermediates thereby generated likely also plays an additional
role.
[6] H. Braunschweig, M. Gross, K. Radacki, Organometallics 2007, 26, 6688–
6690.
[7] H. Braunschweig, M. Gross, K. Radacki, C. Rothgaengel, Angew. Chem.
2008, 120, 10127–10129; Angew. Chem. Int. Ed. 2008, 47, 9979–9981.
[8] a) K. C. Hultzsch, J. M. Nelson, A. J. Lough, I. Manners, Organometallics
1995, 14, 5496–5502; b) C. Elschenbroich, F. Paganelli, M. Nowotny, B.
Neumꢂller, O. Burghaus, Z. anorg. allg. Chem. 2004, 630, 1599–1606;
c) M. Tamm, A. Kunst, T. Bannenberg, E. Herdtweck, P. Sirsch, C. J. Elsevi-
er, J. M. Ernsting, Angew. Chem. 2004, 116, 5646–5650; Angew. Chem.
Int. Ed. 2004, 43, 5530–5534; d) P. Chadha, J. L. Dutton, M. J. Sgro, P. J.
Ragogna, Organometallics 2007, 26, 6063–6065; e) C. L. Lund, J. A.
Schachner, J. W. Quail, J. Mꢂller, J. Am. Chem. Soc. 2007, 129, 9313–
9320; f) H. Braunschweig, T. Kupfer, Acc. Chem. Res. 2010, 43, 455–465;
g) H. Braunschweig, M. Fuss, T. Kupfer, K. Radacki, J. Am. Chem. Soc.
2011, 133, 5780–5783.
Interestingly, the observation of higher molecular weight
species, such as 30, 31 and 32 from either the co-thermolysis
of rac-14 and 15 or the thermolysis of 15 alone, provided evi-
dence for the coupling of radical intermediates formed from
the thermally induced homolytic cleavage of the dicarba
bridge or spacer. Further studies will aim to explore the role of
analogous radical intermediates in the thermal reactivity of
other [n]metallocenophanes.
[9] J. C. Green, Chem. Soc. Rev. 1998, 27, 263–272.
[10] a) D. A. Foucher, B. Z. Tang, I. Manners, J. Am. Chem. Soc. 1992, 114,
6246–6248; b) A. Bartole-Scott, H. Braunschweig, T. Kupfer, M. Lutz, I.
Manners, T.-I. Nguyen, K. Radacki, F. Seeler, Chem. Eur. J. 2006, 12,
1266–1273; c) V. Bellas, M. Rehahn, Angew. Chem. 2007, 119, 5174–
5197; Angew. Chem. Int. Ed. 2007, 46, 5082–5104; d) U. F. J. Mayer, J. B.
Gilroy, D. O’Hare, I. Manners, J. Am. Chem. Soc. 2009, 131, 10382–
10383; e) B. Bagh, G. Schatte, J. C. Green, J. Mꢂller, J. Am. Chem. Soc.
2012, 134, 7924–7936.
[11] a) J. Y. Cheng, C. A. Ross, V. Z.-H. Chan, E. L. Thomas, R. G. H. Lammertink,
G. J. Vancso, Adv. Mater. 2001, 13, 1174–1178; b) I. Korczagin, R. G. H.
Lammertink, M. A. Hempenius, S. Golze, G. J. Vancso, Adv. Polym. Sci.
2006, 200, 91–117; c) V. P. Chuang, J. Gwyther, R. A. Mickiewicz, I. Man-
ners, C. A. Ross, Nano Lett. 2009, 9, 4364–4369.
[12] a) M. Ginzburg, M. J. MacLachlan, S. M. Yang, N. Coombs, T. W. Coyle,
N. P. Raju, J. E. Greedan, R. H. Herber, G. A. Ozin, I. Manners, J. Am.
Chem. Soc. 2002, 124, 2625–2639; b) J. Q. Lu, T. E. Kopley, N. Moll, D.
Roitman, D. Chamberlin, Q. Fu, J. Liu, T. P. Russell, D. A. Rider, I. Manners,
M. A. Winnik, Chem. Mater. 2005, 17, 2227–2231.
[13] a) K. Kulbaba, A. Cheng, A. Bartole, S. Greenberg, R. Resendes, N.
Coombs, A. Safa-Sefat, J. E. Greedan, H. D. H. Stçver, G. A. Ozin, I. Man-
ners, J. Am. Chem. Soc. 2002, 124, 12522–12534; b) Y. Ma, W.-F. Dong,
M. A. Hempenius, H. Mçhwald, G. J. Vancso, Nat. Mater. 2006, 5, 724–
729.
Acknowledgements
We thank the EPSRC for financial support (A.D.R), the Natural
Sciences and Engineering Research Council of Canada (NSERC)
and the European Union Marie Curie Actions Program for post-
doctoral fellowships (J.B.G.).
Keywords: carbonÀcarbon
dicarba[2]ferrocenophanes
polymerization
bond
cleavage
·
·
radicals
·
ring-opening
[1] K. L. Rinehart, Jr., A. K. Frerichs, P. A. Kittle, L. F. Westman, D. H. Gustaf-
son, R. L. Pruett, J. E. McMahon, J. Am. Chem. Soc. 1960, 82, 4111–4112.
[2] a) D. E. Herbert, U. F. J. Mayer, I. Manners, Angew. Chem. 2007, 119,
5152–5173; Angew. Chem. Int. Ed. 2007, 46, 5060–5081; b) R. A. Mus-
grave, A. D. Russell, I. Manners, Organometallics 2014, 32, 5654–5667.
[3] a) A. G. Osborne, R. H. Whiteley, J. Organomet. Chem. 1975, 101, C27–
C28; b) D. Seyferth, H. P. Withers, Jr., Organometallics 1982, 1, 1275–
1282; c) W. Finckh, B. Z. Tang, A. Lough, I. Manners, Organometallics
1992, 11, 2904–2911; d) D. A. Foucher, M. Edwards, R. A. Burrow, A. J.
Lough, I. Manners, Organometallics 1994, 13, 4959–4966; e) K. H. Pan-
nell, V. V. Dementiev, H. Li, F. Cervantes-Lee, M. T. Nguyen, A. F. Diaz, Or-
ganometallics 1994, 13, 3644–3650; f) S. Sharma, N. Caballero, H. Li,
K. H. Pannell, Organometallics 1999, 18, 2855–2860; g) T. Mizuta, Y. Ima-
mura, K. Miyoshi, J. Am. Chem. Soc. 2003, 125, 2068–2069; h) D. M.
Khramov, E. L. Rosen, V. M. Lynch, C. W. Bielawski, Angew. Chem. 2008,
120, 2299–2302; Angew. Chem. Int. Ed. 2008, 47, 2267–2270; i) C.
Moser, F. Belaj, R. Pietschnig, Chem. Eur. J. 2009, 15, 12589–12591; j) I.
Matas, G. R. Whittell, B. M. Partridge, J. P. Holland, M. F. Haddow, J. C.
Green, I. Manners, J. Am. Chem. Soc. 2010, 132, 13279–13289; k) N. C.
Breit, T. Ancelet, J. W. Quail, G. Schatte, J. Mꢂller, Organometallics 2011,
30, 6150–6158.
[14] D. P. Puzzo, A. C. Arsenault, I. Manners, G. A. Ozin, Angew. Chem. 2009,
121, 961–965; Angew. Chem. Int. Ed. 2009, 48, 943–947.
[15] a) Y. Ma, W.-F. Dong, M. A. Hempenius, H. Mçhwald, G. J. Vancso, Angew.
Chem. 2007, 119, 1732–1735; Angew. Chem. Int. Ed. 2007, 46, 1702–
1705; b) D. A. Rider, K. Liu, J. C. Eloi, L. Vanderark, L. Yang, J.-Y. Wang, D.
Grozea, Z.-H. Lu, T. P. Russell, I. Manners, ACS Nano 2008, 2, 263–270;
c) Z. Cheng, B. Ren, D. Zhao, X. Liu, Z. Tong, Macromolecules 2009, 42,
2762–2766; d) P. A. Rupar, L. Chabanne, I. Manners, Science 2012, 337,
559–562.
[16] a) J. M. Nelson, H. Rengel, I. Manners, J. Am. Chem. Soc. 1993, 115,
7035–7036; b) J. M. Nelson, P. Nguyen, R. Petersen, H. Rengel, P. M.
Macdonald, A. J. Lough, I. Manners, N. P. Raju, J. E. Greedan, S. Barlow,
D. O’Hare, Chem. Eur. J. 1997, 3, 573–584.
[17] a) D. E. Herbert, M. Tanabe, S. C. Bourke, A. J. Lough, I. Manners, J. Am.
Chem. Soc. 2008, 130, 4166–4176; b) D. E. Herbert, U. F. J. Mayer, J. B.
Gilroy, M. J. Lꢃpez-Gꢃmez, A. J. Lough, J. P. H. Charmant, I. Manners,
Chem. Eur. J. 2009, 15, 12234–12246.
[18] A. D. Russell, J. B. Gilroy, K. Lam, M. F. Haddow, J. N. Harvey, W. E. Geiger,
I. Manners, Chem. Eur. J. 2012, 18, 8000–8003.
[19] J. B. Gilroy, A. D. Russell, A. J. Stonor, L. Chabanne, S. Baljak, M. F.
Haddow, I. Manners, Chem. Sci. 2012, 3, 830–841.
[20] D. E. Herbert, J. B. Gilroy, A. Staubitz, M. F. Haddow, J. N. Harvey, I. Man-
ners, J. Am. Chem. Soc. 2010, 132, 1988–1998.
[4] a) J. M. Nelson, A. J. Lough, I. Manners, Angew. Chem. 1994, 106, 1019–
1021; Angew. Chem. Int. Ed. Engl. 1994, 33, 989–991; b) U. Vogel, A. J.
Lough, I. Manners, Angew. Chem. 2004, 116, 3383–3387; Angew. Chem.
Chem. Eur. J. 2014, 20, 4077 – 4085
www.chemeurj.org
4084
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[21] A typical thermolysis experiment involved charging a Pyrex thermolysis
tube with the reactants, evacuating the tube over 30 min, then sealing
with an acetylene flame. The tube is then place in an oven for the dura-
tion of the thermolysis experiment. See the Experimental Section in the
Supporting Information for further details.
[22] The moderate yield obtained for rac-14 results from the formation of
black material, postulated to be metallic Fe-based species, which ab-
sorbed to the chromatography column during purification, in addition
to other products that could not be successfully separated and identi-
fied.
[23] L. Routaboul, J. Chiffre, G. G. A. Balavoine, J.-C. Daran, E. Manoury, J. Or-
ganomet. Chem. 2001, 637–639, 364–371.
[24] Excess magnesocene was quenched through the addition of H2O, with
extraction into DCM separating 17 from the by-products of the reac-
tion.
[38] B. A. Howell, Y. Cui, D. B. Priddy, Thermochim. Acta 2003, 396, 167–177.
[39] Attempts to synthesise the diphenyl substituted trimer species by cou-
pling of 15 were unsuccessful. Unreacted starting material was ob-
tained in all cases.
[40] Thermolysis of 9 at 3008C for 1 h resulted in further reactivity and de-
composition. Upon purification multiple products were isolated, but
due to the complexity of the 1H NMR spectroscopic analysis, assign-
ment of the chemical structure of each species could not be made (see
the Supporting Information). No peaks were observed that could be as-
signed to the cyclopentadiene moiety in any of the isolated products,
suggesting that this fragment is cleaved off during the thermolysis. The
formation of multiple products is in contrast to the case of species 15a
and 15b, in which selective labilisation of the central bond of the CÀC
spacer results in clean conversion to a new product (17) upon thermol-
ysis.
[25] The intermediate in mechanism A, which can be described as a reso-
nance hybrid of species 19 and 20, has not been isolated, but is postu-
lated to be generated in reactions of dicarba[2]ferrocenophanes involv-
ing homolytic cleavage of the dicarba bridge.
[41] a) W. H. Morrison, Jr., S. Krogsrud, D. N. Hendrickson, Inorg. Chem. 1973,
12, 1998–2004; b) P. Aguirre-Etcheverry, D. O’Hare, Chem. Rev. 2010,
110, 4839–4864.
[42] A small amount (2 mg) of black material was isolated. The material was
attracted to a bar magnet and was thus likely to contain metallic Fe.
[43] Thermolysis of trimer 29 at 3008C for 16 h resulted in thermal decom-
position with a negligible amount of trimer 29 and mainly other un-
identified products isolated.
[44] The weakening of the CÀC bond in dicarba[2]ferrocenophanes through
the introduction of bulky substituents has been experimentally ob-
served through X-ray crystallographic analysis, in which the dicarba
bridge bond length increases from 1.545(3) ꢁ in species bearing all hy-
drogen substituents, to 1.612(3) ꢁ in a dicarba[2]ferrocenophane bear-
ing one bulky tBu substituents on each carbon of the dicarba bridge
(see ref. [20] for details).
[45] The thermal stability of common radical traps was investigated in order
to explore whether the radical mechanism proposed could be verified
through trapping of the intermediates. However, control thermolysis
experiments under our standard experimental conditions (3008C, 1 h)
for both TEMPO and the galvinoxyl radical resulted in thermal decom-
position. Thermolysis of 15a/b in the presence of an excess of thermal-
ly stable trityl deuteride (Ph3CD) did not result in any deuterium incor-
poration in 17, perhaps due to the steric restrictions of the trityl spe-
cies.
[46] The yields for 17 and 23 are based on the amount of 15 and rac-14
thermolysed, respectively. Due to the inability to completely separate
species 17 and 23 by column chromatography, yields were calculated
through integration of characteristic signals in the 1H NMR spectra. For
details see the Supporting Information.
[47] Additional high molecular weight peaks were observed in the low-reso-
lution mass spectrometric analysis. These species could not be assigned
to logical products from the coupling of radical intermediates, and are
most likely fragments of the higher molecular weight oligomeric spe-
cies that result through coupling.
[48] Due to overlapping signals in the 1H NMR spectrum, the radically cou-
pled species could not be identified. Attempted isolation using column
chromatography was unsuccessful. Their presence was identified, how-
ever, through high-resolution mass spectrometric analysis. The above
observations suggested that these species were formed in very low
yield.
[49] Cleavage of the dicarba bridge in species rac-14 is likely to occur rapid-
ly, however, the rate of isomerisation of meso/rac-14 to rac-14 is rela-
tively slow (3008C, 16 h; see ref. [19] for details). This suggests that al-
though the equilibration between meso/rac-14 and diradical intermedi-
ate 19 at thermolysis temperatures (3008C) is rapid, the short lifetime
of intermediate 19 disfavours CpispoÀCbridge bond rotation, which must
occur to facilitate isomerisation, as the time required for this operation
is significantly longer.
[26] Computational studies provided evidence for a contribution of the bis-
fulvenyl resonance form to the optimised structure of diradical inter-
mediate 19, predicting a CipsoÀCbridge bond length between that of typi-
cal single and double CÀC bonds.
[27] Several attempts to observe the fragmented species 22 by NMR spec-
troscopy were conducted, including cooling the thermolysis tube in
liquid nitrogen prior to opening to trap any volatile species formed,
however, none was successful.
[28] K. J. Stone, R. D. Little, J. Org. Chem. 1984, 49, 1849–1853.
[29] After heating at 3008C for 1 h, although some signals corresponding to
22 were observed in 1H NMR spectroscopic analysis of the crude prod-
uct, the majority of the material consisted of a black solid, insoluble in
common solvents which precluded spectroscopic analysis.
[30] Control experiments were conducted to assess the effect of Cp anion
sources with both 15 and 17 at 3008C for 1 h. In each case, the starting
material was thermolysed with 5 equiv of MgCp2. For 15, identical reac-
tivity was observed to that reported for the thermolysis of 15 alone,
supporting the conclusion that 17 is formed through a radical mecha-
nism. For 17, no further reactivity was observed, allowing the conclu-
sion to be made that ring-strain is a requirement for FeÀCp bond cleav-
age reactions. For details see the Supporting Information.
[31] A peak at 4.14 ppm was observed in both the 1H and 2H NMR spectra
and assigned to the h5-C5H4D and/or h5-C5H3RD groups through com-
parison with [Fe(h5-C5H4D)2] (see the Supporting Information). This ob-
servation suggests that isotopic redistribution occurs in the melt, and
indicates that initial deuterium incorporation into the product 17
maybe greater than 38%.
[32] The isolated black material was not attracted to a bar magnet, suggest-
ing no metallic iron-based species were present.
[33] The yield of 23 was calculated utilising the molecular weight of a single
monomer unit of polyferrocenylethylene 16 to calculate the number of
moles of polymer thermolysed (see the Supporting Information).
[34] S. Bhattacharyya, J. Org. Chem. 1998, 63, 7101–7102.
[35] MALDI-TOF analysis was conducted upon all materials in an attempt to
observe the potential polymeric thermolysis products that occur upon
carbonÀcarbon bond cleavage of the dicarba backbone. However, it
was concluded that the MS data collected were not representative of
the samples, as no peaks were observable above 4000gmolÀ1. Our find-
ings are consistent with the literature (see ref. [17b]), where no MALDI-
TOF MS data were reported.
[36] Further characterisation was conducted with 16c. Analysis by UV/Vis
spectroscopy was conducted and the spectra obtained did not display
absorbances consistent with the presence of ferrocenyl units (lmax =ca.
440 nm). Furthermore, ESI-MS was conducted and no signals corre-
sponding to oligomeric species were observed. A small peak at 366.11
m/z was assigned to residual 23. The characterisation conducted sug-
gests that by-product 16c is not low molecular weight polyferrocenyl-
ethylene but another paramagnetic by-product, perhaps formed from
the thermal degradation of species 23.
Received: November 9, 2013
Revised: December 12, 2013
Published online on March 3, 2014
[37] E. Fçldes, G. Deak, F. Tꢂdçs, O. Vogl, Eur. Polym. J. 1993, 29, 321–330.
Chem. Eur. J. 2014, 20, 4077 – 4085
www.chemeurj.org
4085
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim