Reversing the stereochemistry of a Diels-Alder reaction: Use of metalloporphyrin oligomers to control transition state stability

Article abstract of DOI:10.1039/a709199k

A cyclic Zn-porphyrin trimer with ethyne and butadiyne links stabilises the thermodynamically disfavoured endo transition state and product of a reversible Diels-Alder reaction; at 30°C the endo adduct is formed rapidly and almost exclusively. Linear porphyrin dimers containing ethyne or butadiyne links show related stereochemical preferences to the corresponding cyclic trimers; substantial rate accelerations are observed despite rotational freedom around the linkers. A qualitative correlation is observed between rate acceleration (i.e., transition state binding) and product binding, and the results are rationalised in terms of host geometry and flexibility.

Full text of DOI:10.1039/a709199k

View Article Online / Journal Homepage / Table of Contents for this issue
Reversing the stereochemistry of a Diels–Alder reaction: use of
metalloporphyrin oligomers to control transition state stability
Zoe Clyde-Watson, Anton Vidal-Ferran, Lance J. Twyman, Christopher J. Walter,
Duncan W. J. McCallien, Stefano Fanni, Nick Bampos, R. Stephen Wylie and
Jeremy K. M. Sanders*
Cambridge Centre for Molecular Recognition, University Chemical L aboratory, L ensÐeld Road,
Cambridge, UK CB2 1EW
A cyclic Zn-porphyrin trimer with ethyne and butadiyne links stabilises the thermodynamically disfavoured
endo transition state and product of a reversible DielsÈAlder reaction; at 30 ¡C the endo adduct is formed rapidly
and almost exclusively. Linear porphyrin dimers containing ethyne or butadiyne links show related
stereochemical preferences to the corresponding cyclic trimers; substantial rate accelerations are observed
despite rotational freedom around the linkers. A qualitative correlation is observed between rate acceleration
(i.e., transition state binding) and product binding, and the results are rationalised in terms of host geometry and
Ñexibility.
Scheme 1 Reversible DielsÈAlder reaction between diene 3 and dienophile 4 to give exo adduct 5 and endo adduct 6
One of the key aims of supramolecular chemistry is to catalyse
unfavourable reaction pathways. There are several reports of
the use of catalytic antibodies1 and cyclodextrins2 to catalyse
unfavourable stepwise reactions, but at present there are no
examples of the use of designed, speciÐc host-guest binding to
control pericyclic reaction geometry. We now report a small
but signiÐcant step in that direction by demonstrating that the
The DielsÈAlder reaction possesses a number of features
that make it an attractive target for supramolecular manipula-
tion: it is a bimolecular process; it requires no external
reagents; it proceeds through a highly ordered transition state
that resembles the product (Scheme 1);7 and it o†ers the
opportunity for changing the stereochemical and regiochemi-
cal outcome by selective stabilisation of a disfavoured tran-
sition state. All DielsÈAlder reactions are reversible, but often
the reverse process is so slow that it is not observed. The endo
adduct is usually the kinetic product, the transition state being
stabilised by secondary p-orbital interactions; reactions that
are e†ectively irreversible therefore tend to give predomi-
nantly the endo adduct. However, the exo adduct is generally
sterically and electronically favoured, and tends to be the
dominant product in reversible thermodynamically controlled
reactions. Not surprisingly, the DielsÈAlder reaction has been
the subject of many supramolecular studies exploiting systems
as diverse as micelles,8 synthetic hosts in solution,5,9,10 crys-
talline hosts in the solid state11 and catalytic antibodies.6,12
1,1,2-linked3 cyclic porphyrin trimer Zn 1 can reverse the
3
stereochemical outcome of a DielsÈAlder reaction within its
cavity by selectively accelerating the thermodynamically and
kinetically disfavoured endo reaction pathway. Furthermore,
we show that linear porphyrin dimers possessing rotational
freedom between their two binding sites retain similar stereo-
chemical selectivity and surprisingly large rate accelerations.
Previous examples of redirected pericyclic reactions either
involved non-directional binding4,5 or catalytic antibodies
developed using appropriate transition state analogues.6
* Email:jkms=cam.ac.uk
New J. Chem., 1998, Pages 493È502
493

View Article Online
Our approach to biomimetic catalysis has been to create
synthetic systems in which convergent binding sites are posi-
tioned in such a way that substrate molecules can be held in
close proximity.13,14 In principle such hosts should catalyse
reactions simply by virtue of their binding properties: the
unfavourable entropy of activation would be reduced or com-
pensated for by the favourable enthalpy of binding. Porphyrin
hosts such as Zn 1 and Zn 2 can bind pyridine-based ligands
3
3
within the cavity; if these ligands are relatively rigid then the
high e†ective concentration generated by their binding should
be conveyed efficiently to the reactive site. Our long-term
design strategy has been to create a series of receptors using
the same diarylporphyrin building block but exploring di†er-
ent metal recognition properties and a range of cavity shapes
and sizes. In this way we may dissect out the di†erent contri-
butions to binding and catalysis and inÑuence the outcome of
reactions with multiple pathways.
Scheme 2 Synthesis of dienophile 4. (i) THF, 25 ¡C, 12 h; (ii) HOBT,
DCC, CH Cl ; (iii) 160 ¡C, 15 mm Hg
2
2
around 13 Ó and a ZnwZn distance of ca. 15 Ó, allowing
formation of two good ZnwN bonds in the 1 : 1 complex. In
other words, stabilisation of the exo transition state and
adduct leads to selective acceleration of the reaction.
However, the predicted NwN distance in the optimum
binding geometry of endo adduct 6 is only 9 Ó, so the Ðt is
poor. At 60 ¡C in tetrachloroethane solution, the binding
energy for the endo adduct is ca. 1 kJ mol~1 less favourable
than for the diene and dienophile binding independently and 5
kJ mol~1 less favourable than for the exo adduct.18 By con-
trast, the 1,1,2-trimer20,21 Zn 1 encloses a smaller and more
3
rigid cavity characterised by two distinct Zn-porphyrin
environments and distances. The ZnwZn distance across the
butadiyne linkage is expected to be 15 Ó, as in the 2,2,2-
trimer, while across the ethyne bridge it should be only 12 Ó
and therefore more complementary to the endo adduct.
The results reported here conÐrm that indeed the smaller
binding site formed by two ethyne-linked porphyrins is
capable of stabilising both the endo transition state and
product, thereby reversing the “naturalÏ stereochemistry of the
reaction. However, the results also indicate that less tangible
factors of host and substrate Ñexibility also play vital roles
that are much more difficult to predict.
Initially, we chose to study the reaction of the furan-derived
diene 3 with the maleimide-derived dienophile 4 (Scheme 1)
within the large cavity of our Ðrst generation host, the 2,2,2-
Results
trimer Zn 2. Model building suggested that the substrates
Syntheses
3
bind in an orientation that closely approaches the exo tran-
sition state, and precedent7 showed that this DielsÈAlder
process is reversible. This reversibility o†ered the opportunity
to study the kinetics, and therefore the approach to the tran-
sition state, in both the forward and reverse directions. In the
Diene 3 was originally prepared here by a rather indirect
method,22 but we have since found it to be much more reli-
able to couple 3-bromofuran with 4-trimethylstannyl-
pyridine using [PPh ] Pd. As 3 is an oil, it was converted to
3 4
event, the symmetrical 2,2,2-linked Zn 2 is so selective in
its crystalline hydrochloride salt for convenient storage and
3
accelerating the formation of the thermodynamically more
handling of small quantities. Dienophile 4 was prepared by
the route shown in Scheme 2. The adducts 5 and 6 were pre-
pared for binding studies by the DielsÈAlder reaction of 3 and
4, as described previously.17 However, multimilligram quan-
tities of pure endo adduct 6 were difficult to obtain: at low
temperatures the reaction is extremely slow ([10 days) and
produces mostly exo adduct 5, which is nontrivial to separate
from 6 chromatographically. Above room temperature the
endo adduct 6 is unstable in solution, reverting to component
diene and dieneophile, which can then recombine to give 5;
attempts to concentrate solutions of the endo adduct, e.g., by
rotary evaporation at room temperature, therefore tend to
lead to signiÐcant contamination by 5. Isolation of pre-
parative quantities of pure 6 from a selective host-accelerated
reaction was subject to the same difficulties.
stable exo adduct 5 that no endo adduct 6 can be detected at
any stage in the time course.15h18 Under the reaction condi-
tions macroscopic accelerations of several hundred-fold are
observed,19 but no catalytic turnover is obtained or expected
because the product will bind strongly and inhibit further sub-
strate binding. In the Ðrst instance our objective was to
control the stereochemical outcome of the reaction, and in
such circumstances the role of the host is more like that of a
chiral auxiliary than that of a catalyst. Above room tem-
perature, this particular endo adduct is disfavoured both
kinetically and thermodynamically.17,18
The selectivity observed in the 2,2,2-linked trimer originates
in its geometrical complementarity to the exo adduct and
transition state; modelling indicates an NwN distance of
494
New J. Chem., 1998, Pages 493È502

View Article Online
The large 2,2,2-trimer Zn 2 and butadiyne-linked dimer
3
Zn 7 were prepared using published methods,23 while the
2
shorter ethyne-linked dimer Zn 8 was prepared in 68% yield
2
by Pd0-catalysed coupling of monomers Zn9 and Zn10; syn-
catalysed coupling of monomer Zn13 with two equivalents of
Zn9 was employed, but chromatographic puriÐcation of Zn9
on a large scale is far from trivial, and recent experience has
shown that it is more convenient to couple the diiodo
monomer Zn14 with two equivalents of Zn10.
thesis of these monomers has been described previously.21,23
Hydrogenation of Zn 8 gave the Ñexible linear dimer Zn 11.24
2
2
The smaller cyclic 1,1,2-trimer Zn 1 was prepared by templat-
3
ed GlaserÈHay cyclisation of linear trimer Zn 12, which was
3
itself synthesised by two di†erent routes: originally21 the Pd0-
Kinetic studies
Solubility considerations and the slow rate of the reaction at
room temperature dictated that the DielsÈAlder reaction was
carried out in tetrachlorethane at a concentration of 0.9 mM
for each of the substrates and the porphyrin host. FTIR spec-
troscopy was e†ective at measuring overall accelerations but
was unable to distinguish exo from endo adducts. Gas chro-
matography was found to be unsuitable due to thermal
reversal of the products to starting materials in the injector
port. HPLC and UV spectroscopy su†ered from the disadvan-
tage that the host must be removed from the reaction mixture
before assay.
In the event, therefore, 1H NMR spectroscopy was adopted
as the assay technique, although this also proved not to be
ideal: at the concentration at which the reaction is carried out,
the proportion of both monodentate reactants and bidentate
products bound to the host at any one time is substantial;
these proportions also change during the course of the reac-
tion. The combination of exchange broadening and time-
dependent shifts in the presence of binding sites rendered it
necessary to demetallate each aliquot removed from the reac-
tion mixture before the ratio of the various signals could be
obtained by NMR spectroscopy. In order to achieve this, a
standard amount of TFAÈMeOD was added to each aliquot,
followed by one drop of 100% deuteriopyridine to ensure
complete neutralisation. The relative amounts of the two
adducts could be quantiÐed ( ^ 5%) by integration of a char-
acteristic exo signal at 6.92 ppm and endo signals at 6.84 and
6.94 ppm. This procedure, carried out in duplicate on parallel
samples, gave reproducible exo : endo proportions and semi-
quantitative values for reaction rates. However, the rate data
presented below should be taken as indicative rather than
deÐnitive.
Control reactions in the absence of hosts were so slow that
the rate constants summarised in Table 1 were measured with
9 mM substrates; NMR spectroscopy was used for these rate
New J. Chem., 1998, Pages 493È502
495

View Article Online
Table 1 Rate constants of exo and endo reactions in tetrachloroethane in the absence of host
30 ¡C
60 ¡C
Exo
Endo
Exo
Endo
Forward/
3.1(^0.6) ] 10~5
1.5(^0.3) ] 10~5
2.2(^0.4) ] 10~4
5(^2) ] 10~5
mol~1 dm3 s~1
Reverse/s~1
\6 ] 10~9
\6 ] 10~9
5.5(^1.1) ] 10~7
3.1(^0.6) ] 10~5
measurements as there were no complications from host-
induced e†ects. In the absence of host at 30 ¡C, the exo : endo
proportion is ca. 2 : 1; at 60 ¡C the initial exo : endo ratio is ca.
4 : 1, but the much faster reverse reaction of the endo adduct
soon leads to complete loss of this product in favour of the
exo adduct, making measurement of the forward rate very
imprecise.
The reaction in the presence of the various hosts was
studied at 30 and 60 ¡C. Preliminary studies had indicated
that the macroscopic rate of the reaction in the presence of
2,2,2-trimer was relatively temperature independent: raising
the temperature leads to lower binding constants but faster
intrinsic reactions, and for the trimer these e†ects roughly
cancel out in the accessible temperature range.16 As the tem-
perature dependence of the reverse reaction rates and binding
constants to hosts were rather di†erent for the two adducts,
we hoped to see temperature-dependent outcomes for host
accelerations. This hope was indeed realised in practice, and
Table 2 summarises the qualitative aspects of the observed
exo : endo selectivity. Table 3 summarises the observed rate
accelerations for each adduct relative to the rate of formation
of that adduct in the absence of host.
beginning of the reaction: after 10 h the yield of each adduct is
around 20% (Fig. 2). However, the yield of endo adduct then
reaches a peak of around 24% and then as the reaction pro-
ceeds towards equilibrium, the amount of endo adduct drops
below that of the exo adduct, eventually reaching ca. 10%.
Nevertheless, thermodynamic stabilisation provided by the
1,1,2-trimer cavity gives a long lifetime to a product that
otherwise simply decomposes by reverse reaction at this tem-
perature.
The obvious interpretation of this result is that the endo
geometry is stabilised by binding to two porphyrins across the
ethyne linkage. In an attempt to conÐrm this, the e†ect of
Zn Ni1 was brieÑy examined at 60 ¡C. Nickel porphyrins gen-
2
erally bind much more weakly than their Zn analogues to
pyridines, so we predicted that Zn Ni1 would be ine†ective at
2
accelerating the endo reaction. This proved to be only partly
correct: as expected the exo adduct now dominates, but is
formed at only half the rate observed with Zn 1 while some
3
endo adduct is formed, reaching a maximum of 10% after 45 h
before dying away and being replaced by exo adduct. We have
observed unexpectedly strong, spatially-enforced binding of
the tridentate ligand tripyridyltriazine to the Ni centre in
At 30 ¡C, in contrast to any of the other cyclic hosts we
Zn Ni121 and presume that a similar e†ect is operating here,
2
have studied, the 1,1,2-trimer Zn 1 is totally endo-selective: no
enhancing the binding of bidentate adducts across the ethyne
3
trace of any exo adduct was detected by 1H NMR spectros-
linkage to the Ni centre. This result obscures what would
otherwise be a straightforward control experiment. A better
control would be to use the host resulting from cyclisation of
the ZnwfreebasewZn linear trimer 12, but this synthesis has
never succeeded even in the most experienced hands.25 The
reasons for this and consistent failures of the cyclisation step
in related systems are unclear, but they appear to be con-
nected with deformations (or the deformability) of the central
copy at any stage of the reaction. As Fig. 1 shows, the macro-
scopic reaction rate achieved by Zn 1 is somewhat smaller
3
than that achieved by the 2,2,2-trimer Zn 2, but the stereoche-
3
mical outcome of the reaction is completely reversed, despite
the endo reaction being intrinsically slower than the exo.
In the presence of 1,1,2-trimer at 60 ¡C, the endo and exo
adducts are formed in essentially the same proportions at the
Table 2 Summary of host-induced stereoselectivitiesa
Host
Product selectivity at 30 ¡C
Product selectivity at 60 ¡C
None
exo ] endo
endo
exo ] transient trace endo
1,1,2-Trimer Zn 1
exo ] endo
3
2,2,2-Trimer Zn 2
exo
exo
3
Ethyne-linked dimer Zn 8
exo ] endo
exo ] endo (trace)
2
Butadiyne-linked dimer Zn 7
2
exo
exo
exo
wCH CH w linked dimer Zn 11
È
2
2
2
a Initial concentrations of diene, dienophile and host were 0.9 mM in tetrachloroethane.
Table 3 Host-induced macroscopic initial rate accelerationsa
Rate acceleration at 30 ¡C
Rate acceleration at 60 ¡C
Host
Exo
Endo
Exo
Endo
1,1,2-Trimer Zn 1
È
500
È
È
25
È
50
680
200
15
30
3
200
È
È
È
È
È
3
2,2,2-Trimer Zn 2 (ester)
1500
1000
15
3
2,2,2-Trimer Zn 2 (ethyl)
3
Ethyne-linked dimer Zn 8
2
Butadiyne-linked dimer Zn 7
2
80
wCH CH w linked dimer Zn 11
2
2
2
a Relative to rates in the absence of host. Initial concentrations of diene, dienophile and host were 0.9 mM in tetrachloroethane.
496
New J. Chem., 1998, Pages 493È502

View Article Online
Fig. 1 Progress of the DielsÈAlder reaction in Scheme 1 at 30 ¡C in
Fig. 3 Progress of the DielsÈAlder reaction at 30 ¡C in tetra-
tetrachloroethane solution in the presence of Zn 1 (only endo adduct
chloroethane solution in the presence of the linear dimers Zn 7 (only
3
observed) or Zn 2 (only exo adduct observed). Host, diene and die-
3
2
exo observed) and Zn 8 (exo and endo adducts shown separately)
2
nophile were initially present at 0.9 mM concentration. Also shown is
a progress curve for the control reaction (exo plus endo) calculated
from measurements at higher concentration15,16
unsurprising in the light of the 30 ¡C behaviour.
Given the rotational freedom of these linear dimers, the
large rate enhancement observed is remarkable; presumably
the success of these hosts can be partially attributed to the
rigidity of the linker groups restricting the conformational
space available to the porphyrin units. This is conÐrmed by
the observation that in the presence of the more Ñexible dimer
porphyrin and with the lack of Ñexibility of the resulting
macrocycle, as discussed below.
In order to conÐrm more directly the idea that the endo
geometry is stabilised by binding to two porphyrins across the
ethyne linkage while the exo geometry is stabilised by binding
across the butadiyne linkage, we turned to the linear dimers
Zn 7 and Zn 8. At 30 ¡C the product distribution is roughly
Zn 11 at 60 ¡C the reaction was much slower than with the
2
previous linear dimers. It was a much longer period of time
2
2
before any exo adduct could be detected. No endo adduct was
observed at all during the experiment, which is not surprising
given the dominance in previous experiments of the ther-
modynamic exo product after an extended period at this tem-
perature. At 30 ¡C the reaction with the Ñexible dimer was too
slow to monitor in a practical way.
3 : 2 exo : endo in the presence of ethyne-linked Zn 8 (Fig. 3)
2
and signiÐcant rate enhancement is achieved; the butadiyne-
linked Zn 7 shows complete exo selectivity at 30 ¡C and very
2
substantial rate enhancement. At 60 ¡C the ethyne-linked Zn 8
2
does not alter the product distribution signiÐcantly relative to
the 60 ¡C control and the overall acceleration decreases dra-
matically: only a very small proportion of the endo adduct is
detected and this drops after 48 h, whereafter only the exo
adduct can be observed. This resembles the behaviour of the
1,1,2-trimer at 60 ¡C and again is attributed to the large intrin-
sic rate of the endo back-reaction at 60 ¡C. The butadiyne-
Binding studies
The strengths of the various hostÈproduct interactions were
gauged by UV/VIS titrations in the usual way:26 on binding
pyridine, the main Soret absorption band of zinc porphyrins
experiences a 10 nm bathochromic shift, which can be moni-
tored in a spectrophotometric titration to yield association
constants. Two weak solutions of host are made up ( B 10~6
M), one of which contains the appropriate ligand at a much
higher concentration ( B 0.05 M). The latter solution is added
in known portions to the pure host solution and the absorp-
tion intensities are monitored at the wavelengths of maximum
linked dimer Zn 7 at 60¡ C gives only the exo adduct, which is
2
intensity change, i.e., at k
for the Soret bands of the
max
unbound host and of the bound complex. The binding con-
stants can be derived using a Simplex least-squares curve-
Ðtting program and a plausible model for the stoichiometry of
the hostÈguest interaction. At 30 ¡C in tetrachloroethane, the
diene binds to the various hosts with a binding constant of
around 2000 M~1, falling to around 500 M~1 at 60 ¡C;
binding constants for the dienophile are around half as strong
due to the electron-withdrawing maleimide substitution.27
The analysis can become more complicated for the binding
of multidentate ligands to hosts with multiple binding sites. In
the case of an n-dentate ligand binding to a host with n
binding sites, only the macroscopic binding constant may be
measured and not that of any of the individual ligand inter-
actions. A further complication arises when an (n [ x)-dentate
ligand binds to an n-dentate host, for example in the binding
of a bidentate ligand to a tridentate receptor. A bidentate
ligand will Ðrst bind to two sites, providing the geometry is
appropriate, after which a second ligand molecule will bind
Fig. 2 Progress of the DielsÈAlder reaction at 60 ¡C in tetra-
chloroethane solution in the presence of Zn 1 (exo and endo adducts
3
shown separately) or Zn 2 (only exo observed). Also shown is a
3
progress curve for the control reaction (exo plus endo) calculated from
measurements at higher concentration15,16
New J. Chem., 1998, Pages 493È502
497

View Article Online
Table 4 Selected binding constants measured in tetrachloroethane
K/M~1 at 30 ¡C
K/M~1 at 60 ¡C
Host
Exo
Endo
Exo
Endo
1,1,2-Trimer Zn 1
3 ] 105
2 ] 107
9 ] 106
3 ] 105
1 ] 106
1 ] 105
4 ] 106
2 ] 106
6 ] 105
5 ] 105
1 ] 105
1 ] 105
2 ] 105
nda
4 ] 105
9 ] 104
6 ] 105
1 ] 104
1 ] 105
nda
8 ] 104
3 ] 104
1 ] 105
1 ] 104
3
2,2,2-Trimer Zn 2 (ester)
3
2,2,2-Trimer Zn 2 (ethyl)
3
Ethyne-linked dimer Zn 8
2
Butadiyne-linked dimer Zn 7
2
wCH CH w linked dimer Zn 11
2
2
2
a Not determined.
monodentately to the third binding site. In all the cases
described here, sufficiently low concentrations of ligands were
used such that there was no interference from complexes of
stoichiometry other than 1 : 1, but the experiment cannot dis-
tinguish binding at non-equivalent sites within one host.
Table 4 summarises the product binding constants mea-
sured as described above. Note, however, that the problems
associated with obtaining the endo adduct and working with it
at 60 ¡C limit the reliability of its binding constants to order-
of-magnitude estimates.
cannot be separated into K
ethyne
and K
butadiyne
components.
There are too many binding constants and rate constants
(forward and reverse, across ethyne and across butadiyne,
endo and exo) in this system to allow meaningful simulation of
the entire reaction scheme and construction of a free-energy
proÐle. With so many variables it has not been possible to
Ðnd a unique and reliable Ðt to the experimental data using a
simulation program,31 so our discussion is mainly restricted
to the qualitative aspects.
The DielsÈAlder results at 30 ¡C for the two cyclic trimers
can be summarised in the cartoon shown in Fig. 4: the endo
adduct shows a high complementarity to the porphyrin
binding sites across the shorter ethyne-linker while the exo
adduct is more complementary to the longer butadiyne-linked
binding site. At 30 ¡C the 1,1,2-trimer produces only endo
adduct so it must lower the free energy of the endo transition
state to below that of the exo even though intrinsically the
endo transition state is slightly higher in energy. Assuming
that total amount of exo adduct produced is less than B 3%
(the minimum amount detectable by the 1H NMR assay), the
Discussion
The exo selectivity and large rate acceleration induced by the
cyclic 2,2,2-trimer15h18 raised two obvious questions. (i) Can
the stereochemical selectivity be reversed by use of a suitable
host? (ii) Would an acyclic system of the same potential
geometry show any activity? The results presented here
provide emphatically positive answers. The ability, at 30 ¡C, of
the 1,1,2-trimer to completely reverse the stereochemical
outcome of the DielsÈAlder reaction, and of the rotationally-
free linear dimer Zn 7 to induce exo-selective accelerations as
2
large as 80-fold, are satisfying examples of the power of
designed hostÈguest interactions to inÑuence chemical out-
comes.
However, design does not always lead to the expected
result, especially where access to the desired host is frustrated
by synthetic problems: the small 1,1,1-trimer was a prime syn-
thetic target for some time, as it was expected to be endo-
selective by virtue of geometry and amenable to quantitative
kinetic studies by virtue of its symmetry. In the event, substan-
tial quantities of this material proved elusive despite the
exploration of several di†erent synthetic routes.28,29 No
serious binding or kinetic studies were possible, but a single
experiment at 60 ¡C showed that it accelerates the DielsÈAlder
reaction e†ectively and with considerable (but not exclusive)
endo selectivity, as expected.
The 1,1,2-trimer system was, therefore, initially conceived as
an alternative ““smallÏÏ cavity. The convergent synthetic route,
especially incorporating the improvement described in this
paper, has proved to be extremely versatile. It has allowed
access to trimers containing well-deÐned combinations of
metals such as Zn, Ni, Ru and Sn, and to combinations of
building blocks such as dioxo- and nitroporphyrins,30
although as pointed out above the cyclisation step fails consis-
tently for the ZnwfreebasewZn trimer and some other
systems, for reasons that are unclear.
The asymmetry generated by the unique environment of the
““northernÏÏ porphyrin is a rich source of geometrical informa-
tion for ligands that bind extremely strongly within the
cavity,20,21,29 but it makes any quantitative analysis of
binding and kinetics in the present DielsÈAlder investigation
almost impossible: for example, the overall binding constants
Fig. 4 Cartoon representation of the Ðt of the endo and exo tran-
sition states across the ethyne and butadiyne linkages of the cyclic
hosts
498
New J. Chem., 1998, Pages 493È502

View Article Online
endo reaction is faster than the exo reaction in this case by a
factor of at least thirty. The endo acceleration is 500-fold, so
the acceleration of the exo reaction pathway due to the 1,1,2-
trimer is less than around 10-fold (remembering that the exo
reaction is intrinsically twice as fast). This is a dramatic and
unexpected di†erence from the 1500-fold acceleration seen
with 2,2,2-trimer, even allowing for the presence of only one
butadiyne-linked site rather than three.
The complete suppression of exo adduct in the 1,1,2-cavity
in kinetic experiments at 30 ¡C is supported by the measured
binding constants: a high affinity for the endo adduct is
accompanied by remarkably weak binding of the exo adduct.
The binding of exo adduct to the 1,1,2-trimer is around
60-fold weaker than to the 2,2,2-trimer, rather than the factor
of three that one might expect. Model building and molecular
modelling (Cerius2) give no reason to expect any signiÐcant
equilibrium geometry di†erence across the butadiyne linkage
in the two trimers, so one must conclude that the kinetics and
binding are more sensitive probes for geometry in such large
systems. We take up this question again below.
The picture at 60 ¡C for the 1,1,2-trimer is less clear in
detail, but again endo acceleration by the ethyne-linked sites
appears to be more e†ective than exo acceleration by the
butadiyne-linked sites. An equimolar mixture of the endo and
exo products is produced, corresponding to an acceleration of
roughly 200-fold for the endo reaction, compared with the
control endo reaction, and an acceleration of roughly 50-fold
for the exo reaction compared with the exo control. Beyond
twenty-four hours or so, the level of endo adduct drops signiÐ-
cantly and the thermodynamic exo adduct begins to domi-
nate. It is apparent that at 60 ¡C the exo transition state can
take up a conformation that can bind bidentately within the
cavity, so formation of this product is accelerated. It is not
clear from this observation alone whether this binding is
across the ethyne or butadiyne linkage, or a mixture, but
modynamic stereoselectivity of the butadiyne-linked dimer
Zn 7 are broadly as one would have expected from the 2,2,2-
2
trimer: Zn 7 is geometrically complementary to the exo
2
adduct and accelerates exo formation. For the ethyne-linked
dimer Zn 8 the binding preference for the adducts is reversed,
2
as expected: at 30 ¡C the endo adduct is bound marginally
more strongly, but the exo reaction is intrinsically faster so
both adducts are observed, while at 60 ¡C exo is preferred
kinetically and thermodynamically. This behaviour at the
higher temperature reÑects that of the 1,1,2-trimer, and con-
Ðrms that it is indeed possible at higher temperatures for the
exo transition state to take up a conformation that can bind
bidentately across the ethyne linkage; the large product
binding constant supports this observation.
The rigidly linked linear dimers can rotate freely about the
linker group axis but the trajectories followed by the porphy-
rin binding sites are constrained to be essentially circular. In
order to accelerate the DielsÈAlder reaction, the two porphy-
rin units must take up a limited range of conformations to
allow the bound substrates to react. Consequently, the reac-
tion is roughly 20-fold slower at 30 ¡C than for the trimers; on
purely statistical grounds one would expect a three-fold
reduction in the acceleration,33 so the e†ective rate acceler-
ations are only reduced by around seven-fold by rotational
freedom. These large accelerations are consistent with the sur-
prisingly large binding constants seen at 30 ¡C, corresponding
to e†ective molarities of around 1 M for the favoured adducts:
the entropic cost of restricting rotation around the linker must
be e†ectively compensated by the enthalpic beneÐt derived
from achieving ideal ZnwN bonding at both ends of the
adducts.
Hydrogenation of the rigid linker to give Zn 11 greatly
2
increases the amount of conformational space that the
porphyrin units can explore, and may allow intramolecular
self-association,27 so it is not surprising that both binding and
acceleration are so small. The ability of this host to bind
either adduct with equal strength is a reÑection of the confor-
mational Ñexibility that enables it to wrap with equal facility
around bidentate ligands of di†erent geometries in order to
maximise favourable binding interactions. The net energetic
beneÐt of such binding is, however, weak because there is a
very large entropic price to pay for restricting a highly Ñexible
system. The potential of a particular host system to accelerate
such a reaction is thus dependent on a delicate balance
between two opposing factors: Ñexibility results in good
binding but poor rate acceleration whereas rigidity may result
in good rate acceleration (and in some cases stereoselectivity)
but only if the geometry is precisely right.
60 ¡C evidence from linear dimer Zn 8 and the 1,1,1-trimer
2
suggests at least some exo-binding across the ethyne linkage.
The greater thermodynamic stability of the exo adduct (i.e., its
much slower reverse reaction) means that this becomes the
dominant product as the reaction approaches equilibrium.
To summarise the 1,1,2-trimer results, the geometry and size
of the cavity favour the formation of the kinetic endo product
at both temperatures, but at 60 ¡C this e†ect is largely masked
by the increasing dominance of the thermodynamic exo
product. Nevertheless, the persistence of the endo adduct at
60 ¡C can be attributed to stabilisation within the cavity by
binding across the ethyne bridge.
The reversal of stereoselectivity between the two cyclic
trimers at 30 ¡C is the result of two separate e†ects, one pre-
dicted and one not predicted: the large endo acceleration
induced by the small trimer was expected and may be con-
sidered a success for the design approach. However, the fact
that this trimer is so much less e†ective at 30 ¡C than the
larger analogue at accelerating the exo reaction and binding
exo adduct is a surprise. Perhaps the key di†erence lies in the
greater Ñexibility of the larger 2,2,2-system: if a small change
in host geometry is required for optimum binding of the exo
adduct, it will be energetically less costly for the more Ñexible
system. In other words, success requires mutuality: the 2,2,2-
trimer is sufficiently Ñexible that it can respond better than the
1,1,2-trimer to the geometrical needs of the exo adduct. Cer-
tainly the Ñexibility of 2,2,2 is apparent both in its ligand
binding properties26 and in the fact that it is easier to prepare
analogues with distorted dodecasubstituted porphyrins in the
2,2,2-series32 than in the 1,1,2-series.29 The binding constant
between exo adduct and 2,2,2-trimer at 30 ¡C is large (Table 4)
but is at least ten times smaller than for a ligand that is of
optimum geometry.26 This is additional evidence of a mis-
match of ground state geometries.
Conclusion
Qualitatively, it has been possible to obtain a clear picture of
the e†ects of cavity size and host geometry on this particular
DielsÈAlder reaction. We have shown that subtle changes in
host design can lead to dramatic changes in the stereochemi-
cal outcome of a DielsÈAlder reaction where there are two
Ðnely balanced pathways, and have demonstrated a qualit-
ative correlation between rate acceleration (i.e., transition state
binding) and product binding.34 We have explored the
minimal requirements for e†ective acceleration and found that
rotational freedom is surprisingly unimportant if the sites are
constrained to a limited trajectory. Indeed, the accelerations
and stereoselectivites achieved by the rigidly linked linear
dimers would have been exciting had they been obtained
before any cyclic trimer results. Flexibility emerges as a factor
that is as important as equilibrium geometry, but is inevitably
more elusive to characterise; excessive Ñexibility, as achieved
by hydrogenation of the butadiyne linkers,35 abolishes DielsÈ
Alder acceleration altogether.18
We turn now to the linear dimers. The kinetic and ther-
New J. Chem., 1998, Pages 493È502
499

View Article Online
One might claim that the stereochemical reversal at 30 ¡C is
a major success for the design approach to supramolecular
chemistry, but it is salutory to remember that at 60 ¡C the
results are equivocal while at 80 ¡C one sees little
acceleration18,32 and would expect no stereochemical reversal.
As we have pointed out previously in relation to the concen-
tration dependence of templated synthesis, one is often deli-
cately poised between spectacular success and dismal failure.13
Certainly the more difficult tasks of changing regiochemistry
and engineering catalytic turnover remain a challenge.
drate in reÑuxing chloroform (10 min) until no further change
could be detected by UV inspection. Excess of zinc acetate
was removed by washing twice with water before drying
(MgSO ) and removal of the solvent by evaporation. The Ðnal
4
puriÐcation step in the preparation of porphyrins was crys-
tallisation by layered addition of methanol to a chloroform
solution of the compound, (or the layered addition of hexane
to a dichloromethane solution), followed by drying in vacuo.
Dienophile 4
To investigate this reaction more quantitatively is a chal-
lenging objective that is severely constrained by experimental
and theoretical considerations. To add to the purely practical
difficulties of obtaining reliable data for this particular system
is the more profound problem of demonstrating that one non-
trivial kinetic model is uniquely better or worse than any
other: the kinetics of host-accelerated reactions become
increasingly complex when the formation rates of two di†erent
products with di†ering thermodynamic stabilities are acceler-
ated by di†erent amounts. In our system, even demonstrating
whether the substrates bind selectively or randomly inside and
outside is not trivial.36 This is arguably a result of poor design
in our case, but as Schneider et al. have noted,37 the formal
analysis of artiÐcial catalytic hostÈguest systems is often more
complicated than that of biological systems. SimpliÐcations
can be made in the analysis of an enzymatic system because
these catalysts have become highly efficient under the pres-
sures of natural evolution over many millions of years. As a
consequence, non-catalysed and other side reactions are negli-
gible, catalyst saturation can be achieved, and well-deÐned
binary and ternary complexes can be assumed. Current syn-
thetic biomimetic systems have evolved over only a short
period, usually under conscious control rather than under
evolutionary pressure, and are therefore often not amenable to
such simpliÐcations.
3,6-Epoxy-1,2,3,6-tetrahydro-N-(4-pyridylmethyl)phthalamic
acid. 4-Aminomethylpyridine (2 ml, 20 mmol) in anhydrous
THF (30 ml) was added dropwise to a solution of 3,6-epoxy-1,
2,3,6-tetrahydrophthalic anhydride (2.66 g, 15 mmol) in anhy-
drous THF (200 ml). The solution was stirred under a drying
tube at 25 ¡C for 12 h. The resulting precipitate was washed
with tetrahydrofuran and dried in vacuo to a†ord white crys-
tals of the product. Yield 4.22 g (78%). 1H NMR (250 MHz,
CDCl ) d: 2.63 (1H, d, J \ 8, a to CxO), 2.65 (1H, d, J \ 8, a
3
to CxO), 4.25 (1H, dd, J \ 3, 15, benzylic CH ), 4.31 (1H, dd,
2
J \ 3, 15, benzylic CH ), 4.96 (1H, br m, bridgehead), 5.11
2
(1H, br m, bridgehead), 6.46 (2H, br m, alkene protons), 7.28
(2H, d, J \ 6, pyridyl), 8.22 (1H, t, J \ 3, NH), 8.47 (2H, d,
J \ 6 Hz, pyridyl).
3,6-Epoxy-1,2,3,6-tetrahydro-N-(4-pyridylmethyl)phthalimide.
1-Hydroxybenzotriazole hydrate (HOBT, 2.38 g, 16.9 mmol)
and N,N@-dicyclohexylcarbodiimide (DCC, 4.07 g, 18.9 mmol)
were added to a solution of 3,6-epoxy-1,2,3,6-tetrahydro-N-(4-
pyridomethyl)phthalamic acid (4.22 g, 0.015 mmol) in dry
dichloromethane (300 ml) at 0 ¡C. The resulting mixture was
stirred at this temperature under a drying tube for 2 h and
then for 12 h at room temperature. Insoluble by-products
were removed by Ðltration and the solvent subsequently
removed by evaporation to give a pale brown oil. PuriÐcation
by column chromatography on silica, eluting with neat chlo-
roform, yielded the product as a white solid (2.43 g, 62%). 1H
There is now a movement towards developing evolutionary
and selection approaches38 using catalytic antibodies,1,6
imprinted polymers39 and small molecular systems.40,41 It
remains to be seen whether these will provide a shorter route
to e†ective catalysts.
NMR (250 MHz, CDCl ) d: 2.90 (2H, br m, a to CxO), 4.62
3
(2H, s, benzylic CH ), 5.29 (2H, br m, bridge head), 6.51 (2H,
2
br m, alkene protons), 7.16 (2H, d, J \ 6, pyridyl), 8.52 (2H, d,
J \ 6 Hz, pyridyl).
Experimental
1H NMR spectra were recorded on Bruker WM-250 or
DRX-500 spectrometers in deuteriochloroform operating at
250 or 500 MHz, respectively, while 13C NMR spectra were
obtained on a Bruker WM-250 spectrometer operating at 62.5
MHz. Chemical shifts are quoted relative to residual solvent
4-(Maleimidomethyl)pyridine, 4. 3,6-Epoxy-1,2,3,6-tetra-
hydro-N-(4-pyridylmethyl)phthalimide (90 mg, 0.35 mmol)
was heated to 160 ¡C at a pressure of 15 mm Hg and the
product collected on a cold Ðnger (cooled with ice water) held
2 cm above the reactant. Yield 40 mg (61%). 1H NMR (250
MHz, CDCl ) d: 4.66 (2H, s, benzylic CH ), 6.76 (2H, s, alkene
(7.25 ppm for 1H and 77.0 ppm for 13C of CDCl ). Unless
3
otherwise stated, the acquisition temperature of NMR spectra
3
2
protons), 7.19 (2H, d, J \ 6, pyridyl), 8.55 (2H, d, J \ 6 Hz,
pyridyl).
was 25( ^ 3) ¡C. Porphyrin 1H and 13C assignments were
made by comparison with previously assigned, similarly sub-
stituted porphyrin species. Routine UV/VIS spectra were
obtained on a Uvikon 810 spectrometer or a Hewlett Packard
8452A diode array spectrophotometer in 10 mm oven-dried
cuvettes. Fast atom bombardment mass spectra (FAB MS)
Diene 3
4-Trimethylstannylpyridine. The hydrogen chloride salt of 4-
bromopyridine (4.00 g, 2.05 mmol) was suspended in dry
diethyl ether (15 ml) and cooled to 0 ¡C. Sodium bicarbonate
solution ( B 1 M) was added dropwise until the e†ervescence
subsided and the aqueous solution was basic (pH paper). The
two layers were separated and the aqueous phase further
extracted with diethyl ether (2 ] 10 ml). The organic portions
were recorded on
(Cambridge) or on
a
a
Kratos MS-50 mass spectrometer
VG Autospec mass spectrometer
(EPSRC service at Swansea).
All solvents were distilled prior to use and when used dry,
were obtained fresh from solvent stills: triethylamine, toluene
and dichloromethane were distilled from CaH under an
2
atmosphere of argon while tetrahydrofuran and diethyl ether
were combined and left to stand (MgSO ) for 3 h in the fridge
4
were distilled from LiAlH ÈCaH , also under argon. Flash
before Ðltering and transferring to an argon-purged Ñask con-
4
2
chromatography was carried out using dry-packed 60 mesh
taining 4 Ó molecular sieves. n-Butyl lithium (10.3 ml, 0.021
mol, 2 M solution in hexane) was added to an oven-dried,
argon-purged Ñask containing dry diethyl ether (50 ml) at
[78 ¡C. The bromopyridine solution was then transferred via
cannula to this reaction vessel and the resulting pale brown
silica gel and TLC was carried out on Kieselgel 60 PF
(Merck) 0.2 mm plates.
254
Free base porphyrins were converted into zinc complexes in
near quantitative yield by treatment with zinc acetate dihy-
500
New J. Chem., 1998, Pages 493È502

View Article Online
solution stirred at [78 ¡C for 15 min before being allowed to
warm up to 0 ¡C. Trimethyltin chloride (26.4 ml of a 1 M solu-
tion in THF, 0.0264 M) was syringed into the mixture and
stirring continued at this temperature for 90 min before the
addition of aqueous potassium Ñuoride solution (20 ml, 2 M).
After stirring for 20 min at room temperature, the mixture was
Ðltered through celite to remove the white precipitate of tri-
methyltin Ñuoride. The residue was washed with ethyl acetate
TMS), 15.65 (Me), 21.91, 36.94 (CH ), 51.69 (MeO), 89.99,
94.67 (Cy Cwaryl), 97.45 (meso-H), 105.02 (Cy CwTMS),
2
127.61, 127.83, 131.52, 131.66, 131.93, 133.09, 136.14, 136.36
(aryl-H), 118.75, 122.71, 138.37, 139.03, 141.53, 143.33, 143.66,
143.71, 146.03, 147.61 (quaternary pyrrole and aryl carbons),
173.5 (ester CxO); UV/VIS (CH Cl ) k /nm: 334.1, 409.2,
2
2
max
539.7, 574.6; MALDI-TOF MS (C
H N O Si Zn ) cal-
116 118
8
16
2
2
culated: 2067.1832, found: 2069; FAB MS: 2065 [M [ 2]`,
and the Ðltrate dried (MgSO ) before removing the solvent by
2067 [M]`, 2089 [M ] Na]`.
4
evaporation to give a brown oil. The product was puriÐed by
column chromatography on silica, eluting with a 1 : 1 mixture
of 40È60 petroleum ether : ethyl acetate to give 2.80 g of a
yellow oil (56%), which was used without further puriÐcation.
Zn 11. Palladium on charcoal (250 mg, 10%) was added to
2
a solution of Zn 8 (70 mg, 0.034 mmol) in anhydrous THF
2
(150 ml). The mixture was purged with hydrogen and stirred
1H NMR (250 MHz, CDCl ) d: 0.25 (9H, s, Me), 7.30 (2H, d,
3
for 12 h under a slight positive pressure of hydrogen. The
catalyst was removed by Ðltration through celite and the
solvent evaporated under reduced pressure to give a red solid
that was recrystallised from chloroformÈmethanol (40 mg,
J \ 5, Hb), 8.42 (2H, d, J \ 5 Hz, Ha).
4-(3-Furyl)pyridine 3. Tetrakis(triphenylphosphine)-pallad-
ium (0) (125 mg, 0.108 mmol), 4-trimethylstannyl-
pyridine (0.5 g, 2.07 mmol) and 3-bromofuran (220 ml, 2.44
mmol, 1.2 equiv) were added to a suspension of dry lithium
chloride (0.359 g, 8.47 mmol) in anhydrous toluene (30 ml) and
the mixture reÑuxed under argon for 3 days. The reaction
mixture was allowed to cool to room temperature and
quenched with aqueous potassium Ñuoride solution (20 ml, 2
M). After stirring for 20 min, the mixture was Ðltered through
celite and the residue washed with ethyl acetate. The Ðltrate
was washed with water (50 ml) and brine (50 ml) before drying
(Na SO ) and removal of the solvent by evaporation. The
57%). 1H NMR (250 MHz, CDCl ) d: 0.08 (18H, s, TMS),
3
1.08 (4H, m, CH -TMS), 2.30, 2.31, 2.37, 2.47, 2.56 (24H, 5 ] s,
2
Me), 2.78È2.85 (8H, m, CH -aryl), 3.12È3.36 (16H, m,
2
CH CH CO Me), 3.44, 3.61, 3.65, 3.66, 3.67 (24H, 5 ] s,
2
2
2
MeO), 3.97È4.30 (16H, m, CH CH CO Me), 7.54È7.96 (16H,
2
2
2
m, aryl-H), 10.08, 10.10, 10.20 (4H, 3 ] s, meso-H); 13C NMR
(62.5 MHz, APT, CDCl ) d: 1.01 (Me, TMS), 15.25, 15.15
3
(Me), 19.33, 21.70, 21.97, 30.31, 36.63, 36.70, 36.99 (CH ),
51.42, 51.64 (MeO), 97.04 (meso-H), 127.60, 127.74, 128.74,
2
130.31, 130.41, 131.04, 132.83, 133.52 (aryl-H), 112.63, 120.21,
139.17, 139.29, 140.81, 141.00, 141.05, 141.20, 143.12, 143.35,
144.81, 145.85, 145.95, 147.70 (quaternary pyrrole and aryl
carbons), 173.32, 173.56 (ester CxO); FAB MS
2
4
resulting brown oil was puriÐed by column chromatography
on silica, eluting with a 2 : 1 mixture of 40È60 petroleum
ether : ethyl acetate. The desired product was eluted Ðrst and
evaporated to give 0.15 g of a pale brown oil (50%). 4-
Phenylpyridine was eluted immediately afterwards as a by-
(C
H
N O Si Zn ) calculated: 2079.28, found: 2079;
116 130
8
16
2
2
[M]`, 2080 [M ] H]`; UV/VIS (CH Cl ) k /nm: 332.0,
409.2, 502.9, 574.4.
2
2
max
product. 1H NMR (250 MHz, CDCl ) d: 6.73 (1H, m), 7.36
3
(2H, d, J \ 5, Hb, pyridyl), 7.52 (1H, m), 7.87 (1H, s), 8.58 (2H,
Zn 12 (TMS-protected linear trimer). A solution of diiodo
3
d, J \ 5 Hz, Ha, pyridyl).
porphyrin monomer Zn14 (59 mg, 0.050 mmol) and mono-
TMS porphyrin Zn10 (110 mg, 0.105 mmol, 2.1 equiv) in
THFÈtriethylamine (35 ml : 7 ml) was degassed (2 ] freeze-
thaw cycles) and heated to 35 ¡C under argon. Triphenylarsine
(37 mg, 0.12 mmol) and tris(dibenzylideneacetone)dipalladium-
(0) (14 mg, 0.015 mmol) were added and the mixture Ñushed
through with argon for 5 min. The mixture was heated at
35 ¡C for 90 min before removal of the solvent by rotary
evaporation to give a red solid. This was chromatographed on
silica, eluting initially with hexaneÈethyl acetateÈchloroform
(2 : 1 : 1 v/v) to remove any monomers, followed by neat chlo-
roform to remove the product, which developed as a red band.
Recrystallisation from dichloromethaneÈhexane yielded pure
product (103 mg, 68%) identical to authentic material.21
3 Æ Hydrochloride. A stream of dry hydrogen chloride gas
was bubbled through a stirred solution of 4-(3-furyl)pyridine
(150 mg, 1.03 mmol) in diethyl ether (25 ml). After several
minutes a white precipitate of 4-(3-furyl)pyridine hydro-
chloride formed. The crystals were Ðltered o†, washed with 25
ml of diethyl ether and dried in vacuo (165 mg, 88%).
Oligomers
Zn 8. Tetrakis(triphenylphosphine)palladium (0) (7.6 mg,
2
0.0066 mmol) and copper(I) iodide (3 mg, 0.016 mmol) were
added to a solution of mono-TMS porphyrin Zn10 (65 mg,
0.062 mmol) and TMS-iodoporphyrin Zn9 (72 mg, 0.062
mmol) in anhydrous THF (12 ml) and triethylamine (12 ml).
The mixture was degassed (3 freeze-thaw cycles) and reÑuxed
at 85 ¡C under argon for 5 h after which TLC analysis
(hexaneÈchloroformÈethyl acetate, 2 : 1 : 1) showed only a
trace of starting material left. The mixture was poured into
water (50 ml) and extracted with dichloromethane (50 ml). The
organic phase was further washed with water (50 ml) before
1H NMR assay for kinetic analysis of the Diels–Alder reaction
[2H ]-1,1,2,2-Tetrachloroethane was deacidiÐed by standing
2
over potassium carbonate for 24 h. The host and substrates
were weighed out using a six-Ðgure balance and each made up
to a concentration of 9 ] 10~4 M. The hydrochloride salt of
the diene was deacidiÐed by stirring in tetrachloroethane over
potassium carbonate for 2 h before Ðltering through a cotton
wool plug. The host and the reactant solutions were then
combined and transferred to a small Schlenk tube in a Grant
LTD6 constant-temperature water bath. Solutions that were
to be left for a long period of time were Ñushed with argon
prior to immersion.
drying (MgSO ) and evaporating to give a red solid, which
4
was puriÐed by column chromatography on silica. Elution
with hexaneÈchloroformÈethyl acetate (2 : 1 : 1 v/v) removed
any remaining traces of monomer, and the product was then
eluted with neat chloroform. Recrystallisation from
dichloromethaneÈhexane yielded 88 mg of a red solid (68%).
Samples (0.4 ml) of the reaction mixture were removed at
regular intervals using a 1 ml disposable syringe with a 6A
1H NMR (250 MHz, CDCl ) d: 0.25 (18H, s, TMS), 2.49, 2.54
3
(24H, 2 ] s, Me), 3.14 (16H, t, J \ 7.6, CH CH CO Me), 3.66
18SWG needle and transferred to
a
small vial.
2
2
2
and 3.67 (24H, 2 ] s, MeO), 4.32 (16H, t, J \ 7.6 Hz,
CH CH CO Me), 7.65È8.31 (16H, m, aryl-H), 10.19 (4H, s,
meso-H); 13C NMR (62.5 MHz, APT, CDCl ) d: 0.02 (Me,
[2H ]TriÑuoroacetic acidÈ[2H ]methanol (1 : 1, 0.2 ml) was
1
4
added, resulting in a colour change from red to dark green.
After stirring at room temperature for 10 min, the solution
2
2
2
3
New J. Chem., 1998, Pages 493È502
501

View Article Online
13 J. K. M. Sanders, Comprehensive Supramolecular Chemistry, eds. J.
L. Atwood, J. E. D. Davies, D. D. Macnicol and F. Vogtle, Else-
vier, Oxford, 1996, vol. 9, pp. 131È164.
14 H. L. Anderson and J. K. M. Sanders, J. Chem. Soc., Perkin T rans.
1, 1995, 2223.
15 C. J. Walter, H. L. Anderson and J. K. M. Sanders, J. Chem. Soc.,
Chem. Commun., 1993, 458.
was neutralised by the addition of potassium carbonate (3
microspatula tips). On stirring for a further 15 min the solu-
tion turned red once more and the potassium carbonate was
removed by Ðltration through
a
cotton wool plug.
[2H ]Pyridine (100%, 2 drops) was added to ensure complete
5
neutralisation and the samples were then analysed by 500
16 R. P. Bonar-Law, L. G. Mackay, C. J. Walter, V. Marvaud and J.
K. M. Sanders, Pure Appl. Chem., 1994, 66, 803.
MHz 1H NMR spectroscopy.
17 C. J. Walter and J. K. M. Sanders, Angew. Chem., Int. Ed. Engl.,
1995, 34, 217.
Measurement of binding constants
18 C. J. Walter, Ph.D. Thesis, University of Cambridge, 1994.
19 Early DielsÈAlder experiments15h18 were carried out with the
All UV cuvettes, volumetric Ñasks and syringes were dried in
vacuo prior to use. Dichloromethane was freshly distilled (ex.
ethyl sidechain version of Zn 2; recent results are from the more
3
CaH ) and tetrachloroethane was deacidiÐed by passing
soluble propionate ester sidechain version. The latter is slightly
2
through a short column of basic alumina (Brockmann grade
more e†ective in accelerating the DielsÈAlder reaction than the
original trimer by virtue of larger binding constants and a larger
e†ective molarity for the reaction within the cavity but the stereo-
selectivity is unchanged.18
III) and storing over potassium carbonate. All UV titrations
were carried out using a Hewlett-Packard 8452A diode array
spectrometer Ðtted with a heating jacket (estimated error in
temperature ^0.1 ¡C). Other experimental details and data
analysis were as described previously.26
20 A. Vidal-Ferran, C. M. Muller and J. K. M. Sanders, J. Chem.
Soc., Chem. Commun., 1994, 2657; structure correction: ibid., 1996,
1849.
21 A. Vidal-Ferran, N. Bampos and J. K. M. Sanders, Inorg. Chem.,
1997, 36, 6117.
Acknowledgements
22 P. Ribereau and G. Queguiner, Can. J. Chem., 1983, 61, 334.
23 S. Anderson, H. L. Anderson and J. K. M. Sanders, J. Chem. Soc.,
Perkin T rans. 1, 1995, 2247.
We thank P. Ribereau and G. Queguiner (Rouen) for helpful
advice concerning the synthesis of furylpyridine 3, the EPSRC
mass spectrometry service for FAB spectra, and European
Union HCM and TMR Programs, NSERC Canada and
EPSRC for Ðnancial support.
24 Zn 7 was also hydrogenated to give the even more Ñexible
2
wCH CH CH CH w linked linear dimer but this showed no sig-
2
2
2
2
niÐcant DielsÈAlder acceleration and was not explored further.
25 J. C. Hawley, Z. Clyde-Watson and J. K. M. Sanders, unpublished
results.
26 H. L. Anderson, S. Anderson and J. K. M. Sanders, J. Chem. Soc.,
References
Perkin T rans. 1, 1995, 2231.
27 Diene and dienophile binding constants for Zn 11 were approx-
2
1 See for example T. Li, R. A. Lerner and K. Janda, Acc. Chem. Res.,
1997, 30, 115; P. Wentworth, Jr., A. Datta, S. Smith, A. Marshall,
L. J. Partridge and G. M. Blackburn, J. Am. Chem. Soc., 1997, 119,
2315.
2 R. Breslow and C. Schmuck, J. Am. Chem. Soc., 1996, 118, 6601;
R. Breslow, J. Desper and Y. Huang, T etrahedron L ett., 1996, 37,
2541.
imately half the size as for other hosts, presumably because the
Ñexible dimer can partly self-associate through intramolecular pÈp
interactions, thereby reducing the affinity for external ligands: C.
A. Hunter, M. N. Meah and J. K. M. Sanders, J. Am. Chem. Soc.,
1990, 112, 5773.
28 A. Vidal-Ferran, Z. Clyde-Watson, N. Bampos and J. K. M.
Sanders, J. Org. Chem., 1997, 62, 240.
29 Z. Clyde-Watson, Ph.D. Thesis, University of Cambridge, 1997.
30 S. J. Webb, Z. Clyde-Watson, N. Bampos and J. K. M. Sanders, in
preparation.
3 In this shorthand notation, each number signiÐes the number of
acetylene units in the porphyrin-porphyrin link. Thus Zn 1 is the
3
1,1,2-trimer and Zn 2 is the 2,2,2-trimer.
3
4 W. L. Mock, T. A. Irra, J. P. Wepsiec and M. Adhya, J. Org.
31 R. S. Wylie and J. K. M. Sanders, T etrahedron, 1995, 51, 513.
32 D. W. J. McCallien, Ph.D. Thesis, University of Cambridge, 1995.
33 Two substrates can be arranged productively in two ways on two
binding sites, but in six ways on three binding sites.
34 More extensive studies on a range of substituted 2,2,2-trimers have
quantiÐed this linear free energy relationship.18,32
35 N. Bampos, V. Marvaud and J. K. M. Sanders, Chem. Eur. J.,
1998, 4, 335.
Chem., 1989, 54, 5302.
5 W.-S. Chung, N.-J. Wang, Y.-D. Liu, Y.-J. Leu and M. Y. Chiang,
J. Chem. Soc., Perkin T rans. 2, 1995, 307.
6 V. E. Gouverneur, K. N. Houk, B. Depascualteresa, B. Beno, K. D.
Janda and R. A. Lerner, Science, 1993, 262, 204; M. Resmini, A. A.
P. Meekel and U. K. Pandit, Pure Appl. Chem., 1996, 68, 2025.
7 F. Fringuelli and A. Taticchi, Dienes in the DielsÈAlder Reaction,
Wiley, Chichester, 1990.
8 R. Breslow and D. C. Rideout, J. Am. Chem. Soc., 1980, 102, 7816;
N. K. Sangwan and H.-J. Schneider, Angew. Chem., Int. Ed. Engl.,
1987, 26, 896; V. K. Singh, B. N. S. Raju and P. T. Deota, Synth.
Commun., 1988, 18, 567; P. A. Grieco, P. Garner and Z. He, T etra-
hedron L ett., 1983, 24, 1897; D. A. Jaeger and J. Wang, T etra-
hedron L ett., 1992, 33, 6415.
9 A. D. Hamilton and S. C. Hirst, J. Am. Chem. Soc., 1991, 113, 382;
T. R. Kelly, V. S. Ekkundi and P. Meghani, T etrahedron L ett.,
1990, 31, 3381; J. Rebek, Jr. and J. Kang, Nature (L ondon), 1997,
385, 50.
10 For a self-replicating DielsÈAlder reaction see B. Wang and I. O.
Sutherland, Chem. Commun., 1997, 1495.
36 R. S. Wylie, E. G. Levy and J. K. M. Sanders, Chem. Commun.,
1997, 1611.
37 H.-J. Schneider, R. Kramer and J. Rammo, J. Am. Chem. Soc.,
1993, 115, 8980.
38 P. A. Brady and J. K. M. Sanders, Chem. Soc. Rev., 1997, 26, 327.
39 G. Wul†, T. Gross and R. Schonfeld, Angew. Chem., Int. Ed. Engl.,
1997, 36, 1962.
40 S. J. Rowan and J. K. M. Sanders, Chem. Commun., 1997, 1407; P.
A. Brady and J. K. M. Sanders, J. Chem. Soc., Perkin T rans. 1,
1997, 3237.
41 P. G. Swann, R. A. Casanova, A. Desai, M. M. Frauenho†, M.
Urbancic, U. Slomczynka, A. J. HopÐnger, G. C. Le Breton and D.
L. Venton, Biopolymers, 1996, 40, 617.
11 K. Endo, K. Takashi, T. Sawaki, O. Hayashida, H. Masuda and Y.
Aoyama, J. Am. Chem. Soc., 1997, 119, 4117.
12 D. Hilvert, K. W. Hill, K. D. Nared and M.-T. M. Auditor, J. Am.
Chem. Soc., 1989, 111, 9261.
Received in Montpellier, France, 2nd December 1997;
Paper 7/09199K
502
New J. Chem., 1998, Pages 493È502