Catalytic "active-metal" template synthesis of [2]rotaxanes, [3]rotaxanes, and molecular shuttles, and some observations on the mechanism of the Cu(I)-catalyzed azide-alkyne 1,3-cycloaddition

Article abstract of DOI:10.1021/ja073513f

A synthetic approach to rotaxane architectures is described in which metal atoms catalyze covalent bond formation while simultaneously acting as the template for the assembly of the mechanically interlocked structure. This "active-metal" template strategy is exemplified using the Huisgen-Meldal-Fokin Cu(I)-catalyzed 1,3-cycloaddition of azides with terminal alkynes (the CuAAC "click" reaction). Coordination of Cu(I) to an endotopic pyridine-containing macrocycle allows the alkyne and azide to bind to metal atoms in such a way that the metal-mediated bond-forming reaction takes place through the cavity of the macrocycle - or macrocycles - forming a rotaxane. A variety of mono- and bidentate macrocyclic ligands are demonstrated to form [2]rotaxanes in this way, and by adding pyridine, the metal can turn over during the reaction, giving a catalytic active-metal template assembly process. Both the stoichiometric and catalytic versions of the reaction were also used to synthesize more complex two-station molecular shuttles. The dynamics of the translocation of the macrocycle by ligand exchange in these two-station shuttles could be controlled by coordination to different metal ions (rapid shuttling is observed with Cu(I), slow shuttling with Pd(II)). Under active-metal template reaction conditions that feature a high macrocycle:copper ratio, [3]-rotaxanes (two macrocycles on a thread containing a single triazole ring) are also produced during the reaction. The latter observation shows that under these conditions the mechanism of the Cu(I)-catalyzed terminal alkyne-azide cycloaddition involves a reactive intermediate that features at least two metal ions.

Full text of DOI:10.1021/ja073513f

Published on Web 09/11/2007
Catalytic “Active-Metal” Template Synthesis of [2]Rotaxanes,
[3]Rotaxanes, and Molecular Shuttles, and Some
Observations on the Mechanism of the Cu(I)-Catalyzed
Azide-Alkyne 1,3-Cycloaddition
Vincent Aucagne,† Jose´ Berna´,† James D. Crowley,† Stephen M. Goldup,†
Kevin D. Ha¨nni,† David A. Leigh,*,† Paul J. Lusby,† Vicki E. Ronaldson,†
Alexandra M. Z. Slawin,‡ Aure´lien Viterisi,† and D. Barney Walker†
Contribution from the Schools of Chemistry, UniVersity of Edinburgh, The King’s Buildings,
West Mains Road, Edinburgh EH9 3JJ, and UniVersity of St. Andrews, Purdie Building,
St. Andrews, Fife KY16 9ST, United Kingdom
Received May 17, 2007; E-mail: David.Leigh@ed.ac.uk
Abstract: A synthetic approach to rotaxane architectures is described in which metal atoms catalyze covalent
bond formation while simultaneously acting as the template for the assembly of the mechanically interlocked
structure. This “active-metal” template strategy is exemplified using the Huisgen-Meldal-Fokin Cu(I)-
catalyzed 1,3-cycloaddition of azides with terminal alkynes (the CuAAC “click” reaction). Coordination of
Cu(I) to an endotopic pyridine-containing macrocycle allows the alkyne and azide to bind to metal atoms
in such a way that the metal-mediated bond-forming reaction takes place through the cavity of the
macrocyclesor macrocyclessforming a rotaxane. A variety of mono- and bidentate macrocyclic ligands
are demonstrated to form [2]rotaxanes in this way, and by adding pyridine, the metal can turn over during
the reaction, giving a catalytic active-metal template assembly process. Both the stoichiometric and catalytic
versions of the reaction were also used to synthesize more complex two-station molecular shuttles. The
dynamics of the translocation of the macrocycle by ligand exchange in these two-station shuttles could be
controlled by coordination to different metal ions (rapid shuttling is observed with Cu(I), slow shuttling with
Pd(II)). Under active-metal template reaction conditions that feature a high macrocycle:copper ratio, [3]-
rotaxanes (two macrocycles on a thread containing a single triazole ring) are also produced during the
reaction. The latter observation shows that under these conditions the mechanism of the Cu(I)-catalyzed
terminal alkyne-azide cycloaddition involves a reactive intermediate that features at least two metal ions.
recognition motifs that “live on” 2 in the mechanically inter-
locked molecule that is formed. Besides holding the reactive
Introduction
Most noncovalent bond directed approaches1 to rotaxanes
developed to date require at least stoichiometric quantities of a
template which often involves pre-established strongly binding
fragments in an orientation that directs interlocking, the template
is generally otherwise passive during the reaction. Building on
the principles of transition-metal catalysis, we recently3 began
to explore a strategy in which template ions could also play an
active role in promoting the crucial final covalent bond forming
reaction that captures the interlocked structure (i.e., the metal
has a dual function, acting as a template for entwining the
precursors and catalyzing covalent bond formation between the
reactants). This “active-metal” 4 template process is shown
schematically in Figure 1 in both stoichiometric (1 equiv of
† University of Edinburgh.
‡ University of St. Andrews.
(1) For reviews which highlight various aspects of template strategies to
rotaxanes, see: (a) Amabilino, D. B.; Stoddart, J. F. Chem. ReV. 1995, 95,
2725-2828. (b) Molecular Catenanes, Rotaxanes and Knots: A Journey
Through the World of Molecular Topology; Sauvage, J.-P., Dietrich-
Buchecker, C., Eds.; Wiley-VCH: Weinheim, Germany, 1999. (c) Breault,
G. A.; Hunter, C. A.; Mayers, P. C. Tetrahedron 1999, 55, 5265-5293.
(d) Templated Organic Synthesis; Diederich, F., Stang, P. J., Eds.; Wiley-
VCH: Weinheim, Germany, 2000. (e) Hubin, T. J.; Busch, D. H. Coord.
Chem. ReV. 2000, 200, 5-52. (f) Raehm, L.; Hamilton, D. G.; Sanders, J.
K. M. Synlett 2002, 1743-1761. (g) Kim, K. Chem. Soc. ReV. 2002, 31,
96-107. (h) Arico´, F.; Badjic´, J. D.; Cantrill, S. J.; Flood, A. H.; Leung,
K. C.-F.; Liu, Y.; Stoddart, J. F. Top. Curr. Chem. 2005, 249, 203-259.
(i) Dietrich-Buchecker, C.; Colasson, B. X.; Sauvage, J.-P. Top. Curr. Chem.
2005, 249, 261-283. (j) Kay, E. R.; Leigh, D. A. Top. Curr. Chem. 2005,
262, 133-177. (k) Loeb, S. J. Chem. Commun. 2005, 1511-1518. (l)
Bogdan, A.; Rudzevich, Y.; Vysotsky, M. O.; Bo¨hmer, V. Chem. Commun.
2006, 2941-2952. (m) Nitschke, J. R. Acc. Chem. Res. 2007, 40, 103-
112. (n) Vickers, M. S.; Beer, P. D. Chem. Soc. ReV. 2007, 36, 211-225.
(o) Loeb, S. J. Chem. Soc. ReV. 2007, 36, 226-235. For reviews on
polyrotaxanes, see: (p) Harada, A. Acta Polym. 1998, 49, 3-17. (q) Takata,
T.; Kihara, N.; Furusho, Y. AdV. Polym. Sci. 2004, 171, 1-75. (r) Huang,
F.; Gibson, H. W. Prog. Polym. Sci. 2005, 30, 982-1018. (s) Wenz, G.;
Han, B.-H.; Mu¨ller, A. Chem. ReV. 2006, 106, 782-817.
(2) (a) Philp, D.; Stoddart, J. F. Angew. Chem., Int. Ed. Engl. 1996, 35, 1154-
1196. For a discussion of how and why this feature can be usefully exploited
in the field of molecular machinery, see: (b) Kay, E. R.; Leigh, D. A.;
Zerbetto, F. Angew. Chem., Int. Ed. 2007, 46, 72-191.
(3) Aucagne, V.; Ha¨nni, K. D.; Leigh, D. A.; Lusby, P. J.; Walker, D. B. J.
Am. Chem. Soc. 2006, 128, 2186-2187.
(4) We use “active template” as a general term to describe a reaction in which
a moiety both catalyzes covalent bond formation and acts as a template
for the assembly of a particular structure. The term “active-metal template”
describes a subset of active-template reactions in which the “active” moiety
is a metal. A “catalytic active-metal template” reaction is one in which the
metal catalyst turns over; a “stoichiometric active-metal template” reaction
is one in which it does not.
9
11950
J. AM. CHEM. SOC. 2007, 129, 11950-11963
10.1021/ja073513f CCC: $37.00 © 2007 American Chemical Society

Catalytic “Active-Metal” Template Strategy to Rotaxanes
A R T I C L E S
strategy could prove applicable to many different types of well-
known transition-metal-catalyzed (and even organocatalytic)
reactions, (v) reactions that only proceed through a threaded
intermediate would allow access to several currently inaccessible
mechanically linked macromolecular architectures, and, finally,
(vi) the coordination requirements during key stages of the
catalytic cycle of active-template reactions could provide insight
into the mechanisms of the catalyzed reactions.
Since the preliminary report3 appeared on the realization of
this strategy for rotaxane formation utilizing the Cu(I)-
catalyzed5-7 1,3-cycloaddition8 of organic azides and terminal
alkynes (the CuAAC “click” 9,10 reaction), we have been
delighted to see the concept be quickly adopted11 to make
rotaxanes12 with other Cu(I)-catalyzed reactions, including
alkyne-homocoupling and C-S bond forming reactions.13,14
Here we expand on our investigation of the original system,
showing that the active-metal template rotaxane-forming CuAAC
reaction works well for both mono- and bidentate pyridine-
containing macrocyclic ligands and can also be used to
synthesize more complex two-station degenerate molecular
shuttles whose interstation shuttling can be controlled by
coordination to different metal ions. Furthermore, using a high
macrocycle:copper ratio, [3]rotaxanes with two macrocycles on
a single thread are somewhat unexpectedly produced during the
active-metal template reaction. Together with some simple
kinetic studies carried out under various rotaxane- and thread-
forming reaction conditions, these experimental results provide
some insight into the mechanism of the CuAAC reaction.
Figure 1. “Active-template” strategy to rotaxane architectures. The
formation of a covalent bond between the green and orange “stoppered”
units to generate the thread is promoted by the catalyst (shown in gray)
and directed through the cavity of the macrocycle (shown in blue) by the
catalyst’s coordination requirements. (a) Stoichiometric active-metal tem-
plate synthesis of a [2]rotaxane: (i) template assembly and covalent bond
forming catalysis, (ii) subsequent demetalation. (b) Catalytic active-metal
template synthesis of a [2]rotaxane.
the active template is required) and catalytic (the active template
turns over during the reaction) forms. There are several
potentially attractive features of such a synthetic approach to
mechanically interlocked architectures, including (i) the inherent
efficiency of a reaction in which the macrocycle-metal complex
performs multiple functions, (ii) the lack of a requirement for
permanent recognition elements in each component of the
interlocked product, which increases the structural diversity
possible in catenanes and rotaxanes and enables their formation
to be “traceless”, (iii) in some cases only substoichiometric
quantities of the active template may be required (i.e., the
catalytic active-metal template variant, Figure 1b), (iv) the
(7) For reviews of the CuAAC reaction, see: (a) Bock, V. D.; Hiemstra, H.;
van Maarseveen, J. H. Eur. J. Org. Chem. 2005, 51-68. (b) Wang, Q.;
Chittaboina, S.; Barnhill, H. N. Lett. Org. Chem. 2005, 2, 293-301. (c)
Wu, P.; Fokin, V. V. Aldrichimica Acta 2007, 40, 7-17.
(8) (a) 1,3-Dipolar Cycloadditions Chemistry; Huisgen, R., Ed.; Wiley: New
York, 1984; Vol. 1, pp 1-176. (b) Huisgen, R. Pure Appl. Chem. 1989,
61, 613-628.
(9) For reviews and discussion of the “click chemistry” concept, see: (a) Kolb,
H. C.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int. Ed. 2001, 40,
2004-2021. (b) Kolb, H. C.; Sharpless, K. B. Drug DiscoVery Today 2003,
8, 1128-1137. (c) Ball, P. Chem. World 2007, 4 (4), 46-51.
(5) (a) Tornøe, C. W.; Christensen, C.; Meldal, M. J. Org. Chem. 2002, 67,
3057-3064. (b) Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless,
K. B. Angew. Chem., Int. Ed. 2002, 41, 2596-2599.
(10) A listing of examples of the use of “click” reactions is available at http://
www.scripps.edu/chem/sharpless/click.html.
(11) Indeed, it was predicted3 that, “Chelation to catalytic centers could lead to
rotaxane- and catenane-forming protocols based on other metal-mediated
reactions, including cross-couplings, condensations, and other cycloaddition
reactions”. For a recent example involving Pd(II)-catalyzed alkyne homo-
couplings, see: Berna´, J.; Crowley, J. D.; Goldup, S. M.; Ha¨nni, K. D.;
Lee, A.-L.; Leigh, D. A. Angew. Chem., Int. Ed. 2007, 46, 5709-5713.
(12) For the synthesis of rotaxanes and catenanes using the CuAAC reaction in
“passive” template stoppering or macrocyclization protocols, see: (a)
Mobian, P.; Collin, J.-P.; Sauvage, J.-P. Tetrahedron Lett. 2006, 47, 4907-
4909. (b) Dichtel, W. R.; Miljanic´, O. Sˇ.; Spruell, J. M.; Heath, J. R.;
Stoddart, J. F. J. Am. Chem. Soc. 2006, 128, 10388-10390. (c) Miljanic´,
O. Sˇ.; Dichtel, W. R.; Mortezaei, S.; Stoddart, J. F. Org. Lett. 2006, 8,
4835-4838. (d) Aprahamian, I.; Dichtel, W. R.; Ikeda, T.; Heath, J. R.;
Stoddart, J. F. Org. Lett. 2007, 9, 1287-1290; (e) Braunschweig, A. B.;
Dichtel, W. R.; Miljanic´, O. Sˇ.; Olson, M. A.; Spruell, J. M.; Khan, S. I.;
Heath, J. R.; Stoddart, J. F. Chem. Asian J. 2007, 2, 634-647.
(6) For some interesting examples of the CuAAC reaction, see: (a) Lee, L.
V.; Mitchell, M. L.; Huang, S.-J.; Fokin, V. V.; Sharpless, K. B.; Wong,
C.-H. J. Am. Chem. Soc. 2003, 125, 9588-9589. (b) Wang, Q.; Chan, T.
R.; Hilgraf, R.; Fokin, V. V.; Sharpless, K. B.; Finn, M. G. J. Am. Chem.
Soc. 2003, 125, 3192-3193. (c) Horne, W. S.; Yadav, M. K.; Stout, C.
D.; Ghadiri, M. R. J. Am. Chem. Soc. 2004, 126, 15366-15367. (d) Diaz,
D. D.; Punna, S.; Holzer, P.; McPherson, A. K.; Sharpless, K. B.; Fokin,
V. V.; Finn, M. G. J. Polym. Sci., Part A: Polym. Chem. 2004, 42, 4392-
4403. (e) Manetsch, R.; Krasin´ski, A.; Radic´, Z.; Raushel, J.; Taylor, P.;
Sharpless, K. B.; Kolb, H. C. J. Am. Chem. Soc. 2004, 126, 12809-12818.
(f) Helms, B.; Mynar, J. L.; Hawker, C. J.; Frechet, J. M. J. J. Am. Chem.
Soc. 2004, 126, 15020-15021. (g) Steffensen, M. B.; Simanek, E. E.
Angew. Chem., Int. Ed. 2004, 43, 5178-5180. (h) Collman, J. P.; Devaraj,
N. K.; Chidsey, C. E. D. Langmuir 2004, 20, 1051-1053. (i) Jang, H.;
Fafarman, A.; Holub, J. M.; Kirshenbaum, K. Org. Lett. 2005, 7, 1951-
1954; (j) Krasin´ski, A.; Radic´, Z.; Manetsch, R.; Raushel, J.; Taylor, P.;
Sharpless, K. B.; Kolb, H. C. J. Am. Chem. Soc. 2005, 127, 6686-6692.
(k) Wu, P.; Malkoch, M.; Hunt, J. N.; Vestberg, R.; Kaltgrad, E.; Finn, M.
G.; Fokin, V. V.; Sharpless, K. B.; Hawker, C. J. Chem. Commun. 2005,
5775-5777. (l) Mocharla, V. P.; Colasson, B.; Lee, L. V.; Roeper, S.;
Sharpless, K. B.; Wong, C.-H.; Kolb, H. C. Angew. Chem., Int. Ed. 2005,
44, 116-120. (m) Srinivasachari, S.; Liu, Y.; Zhang, G.; Prevette, L.;
Reineke, T. M. J. Am. Chem. Soc. 2006, 128, 8176-8184. (n) Slater, M.;
Snauko, M.; Svec, F.; Frechet, J. M. J. Anal. Chem. 2006, 78, 4969-4975.
(o) Ladmiral, V.; Mantovani, G.; Clarkson, G. J.; Cauet, S.; Irwin, J. L.;
Haddleton, D. M. J. Am. Chem. Soc. 2006, 128, 4823-4830. (p) White,
M. A.; Johnson, J. A.; Koberstein, J. T.; Turro, N. J. J. Am. Chem. Soc.
2006, 128, 11356-11357. (q) Collman, J. P.; Devaraj, N. K.; Eberspacher,
T. P. A.; Chidsey, C. E. D. Langmuir 2006, 22, 2457-2464. (r) Devaraj,
N. K.; Dinolfo, P. H.; Chidsey, C. E. D.; Collman, J. P. J. Am. Chem. Soc.
2006, 128, 1794-1795. (s) Detz, R. J.; Arevalo Heras, S.; de Gelder, R.;
van Leeuwen, P. W. N. M.; Hiemstra, H.; Reek, J. N. H.; van Maarseveen,
J. H. Org. Lett. 2006, 8, 3227-3230. (t) Aucagne, V.; Leigh, D. A. Org.
Lett. 2006, 8, 4505-4507. (u) Viguier, R. F. H.; Hulme, A. N. J. Am.
Chem. Soc. 2006, 128, 11370-11371.
(13) Saito, S.; Takahashi, E.; Nakazono, K. Org. Lett. 2006, 8, 5133-5136.
(14) Saito et al. imply13 that the concept of binding of a substrate in a cavity
while simultaneously activating it to catalysis is an extension of Vo¨gtle’s
anion template route to rotaxanes [(a) Seel, C.; Vo¨gtle, F. Chem.sEur. J.
2000, 6, 21-24]. Actually the two strategies are rather fundamentally
different. In the Vo¨gtle reaction the macrocycle does not increase the
reactivity of any of the building blocks for the rotaxane. In fact, the
hydrogen bonding of the macrocycle to the phenoxide anion, combined
with the steric hinderance conferred by the presence of the macrocycle,
greatly decreases its reactivity. The only reason that rotaxane is formed in
the Vo¨gtle system is that the more reactive unthreaded phenoxide anion is
completely insoluble under the reaction conditions. The active-template
strategy described in ref 3 and elaborated upon in this paper has much
more in common with Sauvage’s original (passive) metal template ideas
combined with Mock’s [(b) Mock, W. L.; Irra, T. A.; Wepsiec, J. P.; Adhya,
M. J. Org. Chem. 1989, 54, 5302-5308] and later Steinke’s [(c) Tuncel,
D.; Steinke, J. H. G. Chem. Commun. 1999, 1509-1510. (d) Tuncel, D.;
Steinke, J. H. G. Chem. Commun. 2002, 496-497. (e) Tuncel, D.; Steinke,
J. H. G. Macromolecules 2004, 37, 288-302] use of curcubituril to
accelerate a reaction within a macrocycle cavity to form rotaxanes.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11951

A R T I C L E S
Aucagne et al.
Cu(I)-Catalyzed Terminal Alkyne-Azide Cycloaddition
(CuAAC Reaction)
to copper. The same studies found first-order kinetics with
respect to the azide and alkyne (actually slightly higher than
first-order with respect to the alkyne).25 However, relatively little
is known about the ligand-promoted Cu(I)-catalyzed cycload-
dition in organic solvents. Recent experiments by Straub,26 in
which mononuclear copper(I) acetylides ligated to a sterically
demanding N-heterocyclic carbene react efficiently with bulky
organoazides at room temperature, support the notion that a
single copper atom mechanism (A(i), Scheme 1) is viable for
the reaction, at least when the copper is bound to bulky ligands.
Recent DFT calculations7c,21 suggest that π-activation of the
copper acetylide unit by coordination of a second copper atom
(e.g., A(ii) or, more likely under ligand-free conditions or with
small monodentate ligands, bridged27 intermediates such as B(i)
or B(ii), Scheme 1) greatly enhances the reactivity of the copper
σ-acetylide, accelerating formation of the metallacycle. Alter-
natively, a pathway25 in which the reacting azide and alkyne
are coordinated to different copper(I) atoms (intermediate C(i)
or C(ii)) would also be consistent with the second-order kinetics
observed in DMSO-water. It may well be that several of these
types of intermediates28 can provide viable pathways for the
CuAAC reaction, with the different characteristics of the
intermediates being relatively favored or inhibited by factors
such as the solvent, bulk and coordination number of an added
ligand, the strength of ligand-copper binding, and the amount
of ligand-free Cu(I) present in solution. Given the tendency of
copper(I) acetylides to aggregate, however, doubly bridged
intermediates such as B or C should be abundant species under
most conditions. Accordingly, if coordination of the azide to
the same copper atom significantly increases the reactivity of
the σ-acetylide, type B(ii) intermediates are probably involved
in the dominant pathways in most reported CuAAC reac-
tions.
Recently, there has been a tremendous surge of interest in
so-called click9,10 methodologies for functional molecule syn-
thesis, the most popular of which is the Huisgen-Meldal-
Fokin15 Cu(I)-catalyzed 1,3-cycloaddition of organic azides with
terminal alkynes (the CuAAC reaction).5-7 The most common
catalyst systems for this reaction employ water or alcohol
solvents and use a Cu(II) salt in the presence of a reducing agent
(often sodium ascorbate) to generate the required Cu(I) catalyst
in situ.5b,7a Metallic copper5b,7a or copper clusters16 have also
been employed as precatalysts, and in some cases Cu(I) salts
can be used directly. However, in apolar solvents Cu(I) salts
usually require the presence of nitrogen5b,7a,17 or phosphorus18
ligands, or acetonitrile as a cosolvent, to stabilize the Cu(I)
oxidation state, and undesired alkyne-alkyne homocoupling
products are often observed under such reaction conditions.5b,7a
The basic mechanism of the CuAAC reaction is believed to
be that shown in Scheme 1. A [2+3] cycloadditionsthe
mechanism of the thermal (i.e., uncatalyzed) Huisgen reaction8s
is ruled out19 for the Cu(I)-catalyzed reaction on the basis of
DFT calculations which show that reaction via a (formally Cu-
(III)) metallacycle is a more favorable pathway by up to
11.7 kcal mol-1 (Scheme 1).5b,20,21 The same calculations
suggest that the rate-determining step is the formation of the
Cu metallacycle from a reactive intermediate involving copper-
coordinated alkyne and, presumably, azide (organoazido-metal
complexes are likely intermediates in many transition-metal-
mediated reactions of azides, and Cu(I)-N3R complexes have
been characterized by X-ray crystallography22). However, the
exact nature of this reactive intermediate is unclear (Scheme 1,
types A-C).
In the absence of competing ligands,23 copper(I) acetylides
exist as complex multi metal atom aggregates,24 and kinetic
studies17a,25 by Fokin and Finn on the generic ligand-free23 Cu-
(I)-catalyzed alkyne-azide reaction show that in DMSO-water
mixtures the reaction mechanism is second-order with respect
Despite the uncertainty over the precise nature of the reactive
intermediate, since tertiary amines and pyridines facilitate5b,17
the reaction in organic solvents, we reasoned that a macrocycle,
1, bearing an endotopic ligating nitrogen atom might be able to
direct the CuAAC reaction of a “stoppered” alkyne, 2, and a
“stoppered” azide, 3, through the macrocycle cavity to give a
[2]rotaxane, 4, in an active-metal template synthesis (Scheme
2).
(15) The Cu(I)-catalyzed terminal alkyne-azide cycloaddition is often referred
to as the “Sharpless copper click reaction” or the “Sharpless-Huisgen
alkyne-azide cycloaddition”. However, Sharpless gives Fokin the credit
for the idea and recognition of the copper catalysis of this reaction at Scripps
[Rouhi, A. M. Chem. Eng. News 2004, 82 (7), 63-65]. The initial work at
Scripps was carried out shortly after, and without knowledge of, the first
paper (ref 5a) describing the Cu(I)-catalyzed alkyne-azide cycloaddition
(on solid-supported resins) had been submitted by Meldal and co-workers.
(16) Pachon, L. D.; van Maarseveen, J. H.; Rothenberg, G. AdV. Synth. Catal.
2005, 347, 811-815.
Active-Metal Template CuAAC Rotaxane Synthesis
A 2,6-bis[(alkyloxy)methyl]pyridine macrocycle (1a), previ-
ously used29 as both a mono- and bidentate ligand for various
transition metals in classical “passive” template rotaxane and
catenane syntheses, seemed a suitable candidate for initial
investigations. Pleasingly, stirring of an equimolar mixture of
(17) (a) Chan, T. R.; Hilgraf, R.; Sharpless, K. B.; Fokin, V. V. Org. Lett. 2004,
6, 2853-2855. (b) Lewis, W. G.; Magallon, F. G.; Fokin, V. V.; Finn, M.
G. J. Am. Chem. Soc. 2004, 126, 9152-9153.
(18) Perez-Balderas, F.; Ortega-Munoz, M.; Morales-Sanfrutos, J.; Hernandez-
Mateo, F.; Calvo-Flores, F. G.; Calvo-Asin, J. A.; Isac-Garcia, J.; Santoyo-
Gonzalez, F. Org. Lett. 2003, 5, 1951-1954.
(19) If the [2+3] cycloaddition were the preferred mechanism of the Cu-
catalyzed reaction, the structures A-C shown in Scheme 1 would still be
relevant as reactive intermediates.
(26) Nolte, C.; Mayer, P.; Straub, B. F. Angew. Chem., Int. Ed. 2007, 46, 2101-
2103.
(27) (a) Bohlmann, F.; Scho¨nowsky, H.; Inhoffen, E.; Grau, G. Chem. Ber. 1964,
97, 794-800. (b) Siemsen, P.; Livingston, R. C.; Diederich, F. Angew.
Chem., Int. Ed. 2000, 39, 2632-2657.
(28) Note: Copper(I) acetylides are generally complex extended multiatom
aggregates, at least in the solid state and in the absence of good nitrogen
ligands (see ref 24). The types of reactive intermediates shown in Scheme
1 are not meant to be precise or definitive structuressindeed, some of them
(e.g., A(ii) and B(ii)) are very closely relatedsbut rather are meant to
illustrate different (possible) features of the putative reactive intermediate:
How many copper atoms are needed to play significant structural or
electronic roles during the catalysis? Is the copper σ-acetylide π-activated
by an additional Cu atom? If the intermediate has two or more copper
atoms, is it doubly bridgedsas envisaged for a Glaser coupling, which
can be ruled out with bidentate ligands for Cu if azide is also coordinateds
or singly bridged? Are the reacting azide and alkyne attached to the same
or different Cu atoms?
(20) Himo, F.; Lovell, T.; Hilgraf, R.; Rostovtsev, V. V.; Noodleman, L.;
Sharpless, K. B.; Fokin, V. V. J. Am. Chem. Soc. 2005, 127, 210-216.
(21) Ahlquist, M.; Fokin, V. V. Organometallics, in press, cited in ref 7c.
(22) Dias, H. V. R.; Polach, S. A.; Goh, S.-K.; Archibong, E. F.; Marynick, D.
S. Inorg. Chem. 2000, 39, 3894-3901.
(23) In this paper we use the phrases “ligandless” and “ligand-free” to describe
copper that is not bound to the macrocyclic pyridine ligands added to the
CuAAC reactions to generate the active-template synthesis. As discussed
in the text and ref 28, any Cu(I) that is not coordinated to pyridine units
will be complexed by molecules of acetonitrile, azide, alkyne, water, or
other donor atoms present.
(24) Mykhalichko, B. M.; Temkin, O. N.; Mys’kiv, M. G. Russ. Chem. ReV.
2000, 69, 957-984.
(25) Rodionov, V. O.; Fokin, V. V.; Finn, M. G. Angew. Chem., Int. Ed. 2005,
44, 2210-2215.
9
11952 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Catalytic “Active-Metal” Template Strategy to Rotaxanes
A R T I C L E S
Scheme 1. Proposed Mechanism of the Huisgen-Meldal-Fokin Cu(I)-Catalyzed Alkyne-Azide Cycloaddition (CuAAC) Reaction48
the pyridine macrocycle 1a, alkyne 2, azide 3, and [Cu(CH3-
CN)4](PF6) in CH2Cl2 for 24 h affordedsafter demetalation with
KCNsa mixture of [2]rotaxane 4a (57%) and the noninterlocked
triazole thread 5 (41%), together with some of the unconsumed
starting macrocycle (Scheme 2 and Table 1, entry 1).3 By
varying the reaction conditions and reactant stoichiometry (Table
1), yields of up to 94% of [2]rotaxane with respect to the amount
of macrocycle used were achieved (5 equiv of 2 and 3; Table
1, entry 2) for this stoichiometric active-metal template reaction
(Figure 1a). The use of substoichiometric amounts of copper
was investigated to determine whether the metal would turn
over as both a template and a cycloaddition catalyst (i.e., a
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11953

A R T I C L E S
Aucagne et al.
Scheme 2. Active-Metal Template CuAAC Synthesis of [2]Rotaxanes 4a-o from Alkyne 2, Azide 3, and Macrocycles 1a-oa
a Reagents and conditions: (i) 1a-o (0.01 M), Cu(I) salt (generally [Cu(CH3CN)4](PF6); see the text), poorly coordinating solvent (generally CH2Cl2; see
the text), and, in the catalytic version of the active-metal template reaction (Figure 1b), 3 equiv of pyridine; (ii) KCN, CH2Cl2/CH3OH.3 For the effect of
the reaction conditions and reagent stoichiometry on the yields of 4a and 5, see Table 1. For the relative yields of 4a-o using a standardized set of reaction
conditions see Figure 4
.
catalytic active-metal template synthesis, Figure 1b). When using
20 mol % (with respect to 1a) [Cu(CH3CN)4](PF6) at room
temperature, the reaction appeared to stop after an amount of
[2]rotaxane equivalent to the amount of copper present had been
formed, suggesting that the multidentate rotaxane sequesters the
transition metal during the reaction, inhibiting further catalytic
activity. Addition of pyridine as a competing ligand enabled
the catalyst to turn over, producing a substoichiometric reaction,
but the reaction was extremely slow at 25 °C (Table 1, entry
3). Elevating the temperature (70 °C, ClCH2CH2Cl) gave an
(29) (a) Fuller, A.-M.; Leigh, D. A.; Lusby, P. J.; Oswald, I. D. H.; Parsons, S.;
Walker, D. B. Angew. Chem., Int. Ed. 2004, 43, 3914-3918. (b) Fuller,
A.-M. L.; Leigh, D. A.; Lusby, P. J.; Slawin, A. M. Z.; Walker, D. B. J.
Am. Chem. Soc. 2005, 127, 12612-12619. (c) Leigh, D. A.; Lusby, P. J.;
Slawin, A. M. Z.; Walker, D. B. Chem. Commun. 2005, 4919-4921. (d)
Leigh, D. A.; Lusby, P. J.; Slawin, A. M. Z.; Walker, D. B. Angew. Chem.,
Int. Ed. 2005, 44, 4557-4564. (e) Fuller, A.-M. L.; Leigh, D. A.; Lusby,
P. J. Angew. Chem., Int. Ed. 2007, 46, 5015-5019.
9
11954 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Catalytic “Active-Metal” Template Strategy to Rotaxanes
A R T I C L E S
Table 1. Effect of Reaction Conditions and Reagent Stoichiometry on the Active-Metal Template CuAAC Synthesis of [2]Rotaxane 4a
(Scheme 2)3
amt of
2 and 3
(equiv)
amt of
[Cu(CH3CN)4](PF6)
(equiv)
T
conversion to
triazole 2 4a
yield of
rotaxane 1a f4a (%)
entry
solvent
(
°
C)
+
3
f
+
5 (%)
1a
2a
3b
4b
5c
1
5
5
5
1
1
1
0.2
0.2
0
CH2Cl2
CH2Cl2
CH2Cl2
ClCH2CH2Cl
ClCH2CH2Cl
25d
>95
92
44
94
<5
57
94
59
82
0
25d
25d
25 f 70e
25 f 70f
a All reactions were carried out at 0.01 M concentration with respect to 1a using the procedure shown in Scheme 2. b 3 equiv of pyridine. c Control
experiment with no [Cu(CH3CN)4](PF6) present to demonstrate that the thermal reaction does not occur at these temperatures. d 24 h. e 12 h and then 24 h.
f 12 h and then 72 h.
improved yield (82%) of rotaxane 4a in a reasonable time period
(36 h) using only 4 mol % Cu(I) with respect to both 2 and 3
(Table 1, entry 4).
Effect of the Nature of the Cu(I) Source
A number of different Cu(I) salts were screened for the
reaction shown in Scheme 2 using the conditions shown in Table
1, entry 1, but replacing the [Cu(CH3CN)4](PF6) salt with other
Cu(I) complexes. The use of CuOTf‚benzene gave a 78%
conversion of 2 and 3 into triazole products after 24 h but only
a 46% yield of [2]rotaxane 4a. By letting the reaction continue
for a further 2 days (72 h in total), complete conversion of the
starting azide and alkyne to triazole products was achieved (48%
rotaxane). Substituting CuI for [Cu(CH3CN)4](PF6) led to very
little reaction during the first 24 h (8% conversion to triazole),
presumably due to the low solubility of CuI in CH2Cl2, but
interestingly, what little alkyne and azide had reacted formed
rotaxane with high selectivity (7:1 rotaxane:thread). The reaction
proceeded slowly to completion (>98% conversion) over three
weeks. However, at the end point of the reaction the product
ratio (56:44 rotaxane:thread) was no different from that of the
analogous [Cu(CH3CN)4](PF6)-mediated experiment.
Figure 2. Conversion to triazole (i.e., thread + rotaxane if applicable) vs
time for different pyridine-based ligands for the CuAAC reaction. Condi-
tions: (i) ligand (1 equiv), alkyne 2 (1 equiv), azide 3 (1 equiv), [Cu(CH3-
CN)4](PF6) (1 equiv, 1 or 2 equiv with 1a), CD2Cl2, rt, 0.01 M. The
conversion of 2 and 3 into the triazole products (thread 5, + rotaxane 4a
1
where relevant) was monitored by H NMR.
Kinetic Studies
We also synthesized and isolated various discrete Cu(I)-1a
complexes and tested their efficacy in the active-metal template
reaction. The macrocycle 1a and the different Cu(I) salts were
mixed in CH3CN solution and the resulting Cu(I)-1a complexes
obtained through precipitation by vapor diffusion with diethyl
ether. The preformed Cu(I)-1a complexes were submitted to
the standard rotaxane-forming reaction conditions (Table 1, entry
1). The preformed [Cu(CH3CN)4](PF6)-1a complex gave 76%
conversion to triazole and a 53% yield of [2]rotaxane; the
preformed CuOTf-1a complex gave 53% conversion to triazole
but only a 5% yield of [2]rotaxane; the preformed CuI-1a
complex gave results identical to those of the CuI-1a complex
formed in situ (i.e., >98% triazole conversion over 3 weeks,
56% rotaxane).
Although there is some variation in the efficacy of the various
Cu(I) salts on the rotaxane-forming reaction and some unex-
plained variation in the rotaxane:thread ratio during the course
of the reaction with some sources of Cu(I), simple addition of
[Cu(CH3CN)4](PF6) is the most convenient way of maximizing
the CuAAC yield of rotaxane.30
Some simple kinetic measurements were made to determine
whether, under these reaction conditions, the Cu(I)-catalyzed
alkyne-azide cycloaddition is actually accelerated by pyridine-
based ligands. The rate of formation of triazole products (i.e.,
thread, plus rotaxane where relevant) was compared for the
ligand-free23 reaction and reactions containing pyridine, 2,6-
dimethylpyridine (lutidine), 2,6-bis[(alkyloxy)methyl]pyridine
macrocycle 1a, and 6, a close but acyclic analogue of 1a.
The results (Figure 2) show that both lutidine and 6, the
acyclic analogue of 1a, significantly accelerate the CuAAC
reaction rate (essentially complete conversion to triazole after
3 h) compared to that of the Cu(I)-catalyzed reaction when no
pyridine-based ligand is added (complete conversion after 6 h).
The presence of pyridine also initially accelerates the reaction,
but the conversion tails off after 2 h, suggesting that using
unsubstituted pyridine as a ligand facilitates the oxidation of
Cu(I) to noncatalytic Cu(II) under the reaction conditions. The
role of the pyridine ligands in the rate acceleration is probably
to break up extended copper(I) acetylide aggregates to form
smaller reactive intermediates of the types shown in Scheme 1,
although we cannot rule out an electronic effect of the ligand
on the metal as well.
(30) We also examined the effect of different alkyne substrates on the rotaxane
formation. The alkynes (see the Supporting Information) were submitted
to the standard click reaction conditions (Scheme 2). Using a longer, more
flexible, alkylalkyne or an arylalkyne in the place of 2 led to similar yields
of the corresponding [2]rotaxanes (and similar overall conversions of the
substrates into triazole products), indicating the active-metal template
rotaxane-forming reaction is rather insensitive to substrate modifications.
Interestingly, the Cu(I)-catalyzed formation of triazole prod-
ucts (both rotaxane and thread) in the presence of macrocycle
1a (with 1 equiv of Cu)swhich also presumably breaks up the
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11955

A R T I C L E S
Aucagne et al.
2, we varied the structure of the pyridine-based macrocycle.
The macrocycles used in the study are shown in Figure 4, and
their syntheses are detailed in the Supporting Information. To
compare the relative yields of rotaxane and thread within the
ligand series, the macrocyclic ligands were screened using a
standard, rather than optimized,31 set of stoichiometric active-
metal template reaction conditions in which the original
macrocycle 1a generates appreciable quantities of both rotaxane
and thread. The conditions used were 1 equiv of macrocycles
1a-o, 1 equiv of alkyne 2, 1 equiv of azide 3, and 1 equiv of
[Cu(CH3CN)4](PF6) at a concentration of 0.01 M in CH2Cl2 at
room temperature for 24 h. In most cases the starting alkyne 2
and azide 3 were completely consumed during the course of
the reaction. After 24 h (72 h for 1n and 1o) each reaction was
1
treated with KCN to remove the Cu(I), analyzed by H NMR
to determine the yield (shown in Figure 4) under the standard
set of conditions, and then purified to give authentic samples
of each [2]rotaxane. The yield (the percentage shown in red in
Figure 4) of each rotaxane is the conversion of the macrocycles
1a-o to the corresponding [2]rotaxanes 4a-o. The conversion
to triazole (the percentage shown in parentheses in Figure 4) is
the overall conversion of the alkyne 2 and azide 3 into both
thread 5 and rotaxanes 4a-o.
Mono- and Tridentate Macrocycles. In addition to the
strongly coordinating pyridine nitrogen, macrocycle 1a has two
ether oxygen atoms that could potentially be involved in (albeit
weak) coordination to the metal atom during the active-template
reaction, making 1a a monodentate or possibly weakly bidentate
or very weakly tridentate ligand.32 Not only is macrocycle 1b
missing the ether oxygen atoms, but the pyridine nitrogen atom
lone pair is tied up in strong intramolecular bifurcated hydrogen
bonds to the adjacent amide groups,33 so it was rather surprising
to find that [2]rotaxane 4b was successfully produced in the
active-metal template reaction, albeit in only 6% yield. Coor-
dination of 1b to the copper occurs at the expense of the
intramolecular hydrogen bonds34,35 so that 1b is able to act as
an effective monodentate ligand. In contrast, the use of 1csa
tridentate macrocyclic ligand with three hard donor atomss
gave a similar overall conversion to triazole but with no rotaxane
Figure 3. Formation of triazole products (rotaxane 4a, 9; thread 5, b) vs
time for the CuAAC reaction in the presence of macrocycle 1a. Condi-
tions: (i) 1a (1 equiv), 2 (1 equiv), 3 (1 equiv), [Cu(CH3CN)4](PF6) (1 or
2 equiv), CD2Cl2, rt, 0.01 M. The conversion of 2 and 3 into thread 5 and
1
rotaxane 4a was monitored by H NMR.
copper acetylide aggregatessis actually slower (24 h compared
to 6 h) than the ligand-free Cu(I)-catalyzed reaction. While it
might seem surprising that the active-metal template reaction
is slower than the ligand-free reaction given that excellent yields
of rotaxane are obtained under both stoichiometric and catalytic
active-metal template conditions (Table 1), the rationale for this
is quite straightforward. First, the alkyne-azide cycloaddition
with the macrocyclic ligand must take place through the
macrocycle cavity, a sterically restricted environment compared
to the ligandless Cu(I)-catalyzed reaction, effecting the solvation
of the reactive species as well as hindering any motion required
to achieve bond formation or changes of geometry at the copper
center. Second, the macrocyclic ligand might disfavor certain
types of reactive intermediates (e.g., B) on steric grounds, so
the reaction may proceed through another type of slower
reacting, but still viable, intermediate (e.g., A or C rather than
B). The reason that the ligandless formation of the triazole thread
does not dominate given the slow rate of rotaxane formation is
that the macrocycle is an excellent ligand for the Cu(I) and
sequesters it, preventing the inherently faster ligand-free reaction
(probably via B(ii)) from occurring. Accordingly, addition of a
second equivalent of Cu(I) to the reaction containing 1a should
accelerate the rate of triazole formation to close to the ligand-
free rate. Indeed, this was found to be the case (Figures 2 and
3). However, while one might expect this increase to be totally
due to thread formation, analysis shows that the rate of rotaxane
formation is also increased by adding an extra equivalent of
Cu(I) to this reaction (Figure 3). In fact, the final yield of [2]-
rotaxane only falls from 57% to 22% even though the triazole-
forming reaction is complete in 6 h instead of 24 h. The
acceleration of the rotaxane-forming reaction by excess Cu(I)
strongly suggests that π-activation of the copper acetylide unit
(see Scheme 1) is the dominant process in the CuAAC reaction
mechanism of 1a.
(31) The standard reaction conditions use 1 equiv of each reagent and low
temperatures to minimize the background uncatalyzed thermal cycloaddi-
tion. These conditions were chosen to allow the relative efficacy of the
different macrocycles in promoting rotaxane formation to be assessed. No
attempt was made to optimize the reaction conditions to improve the
rotaxane yields reported in Figure 4. For macrocycles 1a, 1i, 1j, 1f, 1l,
1m, and 1nsall of which produce significant rotaxane and are not degraded
under the reaction conditionssvirtually all the macrocyclic ligand can be
converted to rotaxane by using extended reaction times and an excess of
the azide and alkyne building blocks.
(32) For an X-ray structure of 1a as a monodentate ligand in a rotaxane (Pd(II)
as the metal atom) see ref 29b; for X-ray structures of a related pyridine
diether ligand acting as a bidentate ligand in a rotaxane (Cu(II) or Ni(II)
as the metal atom) see ref 29d.
(33) (a) Hunter, C. A.; Purvis, D. H. Angew. Chem., Int. Ed. Engl. 1992, 31,
792-795. (b) Johnston, A. G.; Leigh, D. A.; Nezhat, L.; Smart, J. P.;
Deegan, M. D. Angew. Chem., Int. Ed. Engl. 1995, 34, 1212-1216. (c)
Leigh, D. A.; Moody, K.; Smart, J. P.; Watson, K. J.; Slawin, A. M. Z.
Angew. Chem., Int. Ed. Engl. 1996, 35, 306-310. (d) Leigh, D. A.; Murphy,
A.; Smart, J. P.; Slawin, A. M. Z. Angew. Chem., Int. Ed. Engl. 1997, 36,
728-732. (e) Leigh, D. A.; Murphy, A.; Smart, J. P.; Deleuze, M. S.;
Zerbetto, F. J. Am. Chem. Soc. 1998, 120, 6458-6467. (f) Deleuze, M. S.;
Leigh, D. A.; Zerbetto, F. J. Am. Chem. Soc. 1999, 121, 2364-2379. (g)
Clegg, W.; Gimenez-Saiz, C.; Leigh, D. A.; Murphy, A.; Slawin, A. M.
Z.; Teat, S. J. J. Am. Chem. Soc. 1999, 121, 4124-4129. (h) Biscarini, F.;
Cavallini, M.; Leigh, D. A.; Leo´n, S.; Teat, S. J.; Wong, J. K. Y.; Zerbetto,
F. J. Am. Chem. Soc. 2002, 124, 225-233. (i) Bottari, G.; Dehez, F.; Leigh,
D. A.; Nash, P. J.; Pe´rez, E. M.; Wong, J. K. Y.; Zerbetto, F. Angew. Chem.,
Int. Ed. 2003, 42, 5886-5889. (j) Leigh, D. A.; Venturini, A.; Wilson, A.
J.; Wong, J. K. Y.; Zerbetto, F. Chem.sEur. J. 2004, 10, 4960-4969.
Effect of the Macrocycle Structure on [2]Rotaxane
Formation
To further investigate both the scope and the mechanism of
the active-template rotaxane-forming reaction shown in Scheme
9
11956 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Catalytic “Active-Metal” Template Strategy to Rotaxanes
A R T I C L E S
Figure 4. Influence of the macrocycle structure on the active-metal template CuAAC rotaxane-forming reaction in Scheme 2 under a standardized set of
reaction conditions. Conditions for Scheme 2: (i) 1a-o, 2, 3, [Cu(CH3CN)4](PF6), CH2Cl2, rt, 24 h (72 h for 4n and 4o), 0.01 M concentration with respect
to each of 1a-o, alkyne 2, azide 3, and [Cu(CH3CN)4](PF6); the reactions were not run under an inert atmosphere, nor using distilled or dried solvents; (ii)
KCN, CH2Cl2/CH3OH. The conversion of each macrocycle to [2]rotaxane is given as a percentage yield in red, and the overall conversion of 2 and 3 into
triazole products (rotaxanes 4a-o and thread 5) is shown for each macrocycle in parentheses. The designation “trace” means that the rotaxane was observed
1
by ESI-MS but could not be detected by H NMR. The colors and lettering correspond to the signals and assignments of the 1H NMR spectra shown in
Figure 5.
formed. Upon addition of 1c to the copper catalyst and other
reagents the solution changed color immediately from colorless
to blue-green. This suggests that most of the Cu(I) is rapidly
oxidized to Cu(II) upon coordination to 1c.36 The resulting Cu-
(II)-1c complex is not catalytically active for the alkyne-azide
cycloaddition, and any tridentate Cu(I)-1c complex lacks the
free coordination sites necessary to bind azide and alkyne in
the same aggregate.37 However, enough of the copper in the
reaction is not tied up as the 1c complex for the ligandless
thread-forming triazole reaction to proceed to completion within
24 h.
Macrocycle 1d has a structure similar to that of 1a except
that the pyridine nitrogen is oxidized. The result, >98%
conversion to triazole but only a trace of rotaxane when using
1d as the macrocycle in the active-template reaction, shows that
the N-oxygen atom greatly affects the ligand’s ability to
coordinate to Cu(I). Macrocycle 1e also has a constitution
(34) Hydrogen bond energies are usually in the range of 1-10 kcal/mol, whereas
coordinate bonds have energies of 20-80 kcal/mol;35 thus, the stabilization
of the copper(I) ion upon coordination should compensate for the loss of
the two hydrogen bond interactions. In fact, all of the pyridine-containing
macrocycles studied exhibit complexation-induced shifts in the 1H NMR
spectra (CDCl3) upon addition of [Cu(CH3CN)4](PF6), indicating that
copper(I) ions interact with the macrocycles in every case.
(35) Goshe, A. J.; Crowley, J. D.; Bosnich, B. HelV. Chim. Acta 2001, 84, 2971-
2985.
(36) This oxidation of Cu(I) to Cu(II) in the presence of 1c occurs even when
great care is taken to exclude both moisture and oxygen from the reaction
mixture. It appears that the macrocycle itself is the oxidant. However, there
is some remaining [Cu(CH3CN)4](PF6) which catalyzes the formation of
the thread 5.
(37) The multidentate ligand tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine
(TBTA) is a highly effective copper ligand for the CuAAC reaction [see
refs 6b and 17a,b and Gupta, S. S.; Kuzelka, J.; Singh, P.; Lewis, W. G.;
Manchester, M.; Finn, M. G. Bioconjugate Chem. 2005, 16, 1572-1579].
However, the individual triazole-functionalized “arms” of TBTA are weakly
coordinating ligands, so TBTA probably functions by the alkyne and azide
displacing two of the triazole groups to form the reactive intermediate.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11957

A R T I C L E S
Aucagne et al.
similar to that of 1b, but the secondary amides are methylated,
eliminating their ability to form hydrogen bonds. AM1 calcula-
tions38,39 using Spartan (see the Supporting Information) show
that the methyl groups distort and partially fill the macrocycle
cavity. The result is that, like that of 1d, little of the CuAAC
reaction is directed through the macrocycle cavity to form
rotaxane. The same effect is seen for macrocycle 1f, the bis-
N-acetylated analogue of 1c.
the electron-donating ability of the heterocyclic nitrogen atom,
macrocycle 1m, had little effect on the rotaxane yield.
Bidentate Macrocycles. Having established that monodentate
ligands of several different macrocycle geometries and donor
atom orientations (1a, 1j, 1l, 1m) efficiently promote active-
template rotaxane formation, whereas strongly tridentate mac-
rocycles (1c and 1h) do not, we investigated the efficacy of
bidentate macrocycles 1n and 1o.17,42 The bipyridyl macrocycle
1n directs the CuAAC reaction through its cavity to form
rotaxane almost as efficiently as the best monodentate macro-
cycles (45% for 1n f 4n compared to 57% for 1a f 4a and
60% for 1j f 4j). However, it severely inhibits the rate of the
Cu(I)-catalyzed reaction, and it took 3 days for the active-metal
template reaction 1n f 4n to go to completion, compared to
24 h for 1a f 4a and 6 h for the ligand-free Cu(I)-catalyzed
control reaction (Figure 3). The inhibition of the Cu(I)-catalyzed
cycloaddition is even more dramatic with the other bidentate
ligand investigated, the 2,9-diphenylphenanthroline macrocycle
1o used extensively to assemble rotaxanes and catenanes by
the Sauvage group.1b,i Macrocycle 1o completely inhibits the
CuAAC reaction. Even after 3 days under the standard sto-
ichiometric active-metal reaction conditions in the presence of
1o, no triazole products were observed. This latter result is
particularly interesting given that phenanthroline ligands have
previously been reported17 to promote the CuAAC reaction. The
lack of reactivity is presumably a result of the steric bulk about
the Cu(I) center in the complex preventing the complex from
undergoing the various structural variations required for reaction
to take place (e.g., tolerating the change in geometry from Cu-
(I) to the formal Cu(III) species, Scheme 1), together with the
macrocycle being very effective in sequestering the Cu(I) in
this unreactive form. A related Cu(I)-macrocycle complex has
recently been reported to promote the formation of [2]rotaxanes
via a Glaser alkyne homocoupling.13 However, in that case,
intermediates of type B (Scheme 1) are possible with a bidentate
ligand for copper because no azide need be coordinated to the
metal atom.
Removing the pyridine nitrogen atom, macrocycle 1g,
completely switches off rotaxane formation, confirming the need
for a ligating donor atom. Replacing the benzylic oxygens of
1a by more strongly coordinating sulfur atoms produces another
tridentate macrocycle, 1h. As with 1c, no rotaxane is formeds
consistent with the premise that tridentate ligand complexes of
Cu(I) lack the vacant coordination sites necessary to bind both
azide and alkyne37sand the yield of triazole is also only 30%,
indicating that 1h is particularly effective in sequestering the
copper. Oxidation of the sulfide groups to sulfones, macrocycle
1i, restores the ability of the macrocycle to bind to the Cu(I) as
a monodentate ligand, and despite the increased bulk around
the coordinated metal ion, a significant amount (13%) of [2]-
rotaxane 4i is formed by the stoichiometric active-metal template
reaction. In fact, substitution of the benzylic ether oxygen atoms
by methylene groups confirms that only a monodentate mac-
rocycle with a single strongly coordinating donor atom is
required for an efficient active-template rotaxane-forming reac-
tion; macrocycle 1j generates just as much [2]rotaxane as the
parent macrocycle 1a. Modeling suggests that the cavity of
macrocycle 1j is rather more flexible than that of 1a, yet this is
not detrimental for rotaxane formation.
To test whether other macrocycle geometries would be
tolerated by the active-template reaction, we prepared 1k and
1l in which the substituted benzylic units were replaced by
arylalkynyl and aryl groups, respectively. The 2,6-bisalkynylpy-
ridine unit proved to be unstable40,41 under the reaction
conditions, suppressing the overall yield of the CuAAC reaction,
with no rotaxane being produced. However, the active-metal
template CuAAC reaction proceeded smoothly with 1l, albeit
generating rather less rotaxane (25%) than the most effective
examples of other macrocycle geometries (1a and 1j). AM1
calculations38 using Spartan (see the Supporting Information)
indicate that 1l has a well-defined persistent cavity suitable for
the threading of a molecular chain. However, the phenyl groups
flanking the pyridine sterically encumber the nitrogen donor
atom of the macrocycle, and a combination of steric and
electronic factors probably reduces the binding strength of this
macrocycle for Cu(I) ions. The weaker binding leads to more
ligandless23 Cu(I) in solution, resulting in higher conversion of
the substrates into the noninterlocked thread 5 and the relatively
modest yield of [2]rotaxane 4l. Increasing the steric bulk at the
4-position of the pyridine group while simultaneously increasing
1H NMR Spectra of the Metal-Free [2]Rotaxanes
The 1H NMR spectra (400 MHz, 300 K, CDCl3) for each of
the rotaxanes formed in >5% yield are shown in Figure 5. The
1H NMR spectra of the rotaxanes (Figure 5b-g) all show upfield
shifts of several signals with respect to the signals of the
noninterlocked components (thread 5, Figure 5a, and the
macrocycles 1a, 1b, 1i, 1j, 1m, and 1n; see the Supporting
Information). Such shielding is typical of interlocked structures
in which the aromatic rings of one component are positioned
face on to with another component and is observed for all the
nonstopper resonances of the axle (Hf-j), indicating that the
macrocycle accesses the full length of the thread. This is as
expected; there should be no strong intercomponent noncovalent
interactions between the thread and the macrocycle in the metal-
free rotaxanes. The one exception is the amide-containing
rotaxane 4b (Figure 5c), which exhibits a significant downfield
shift of the amide resonance with respect to that of the free
macrocycle (see Supporting Information). This can be attributed
to hydrogen bonding between the amides of the macrocycle and
(38) Minimum energy conformations were calculated for each of the metal-
free macrocycles using the Spartan molecular modeling program (Semi-
Empirical, AM1). The results are provided in the Supporting Information.
(39) In the case of 1e no hydrogen bonds are present, thus allowing the Cu(I)
ion better access to the macrocycle’s pyridine nitrogen atom. However, in
each of macrocycles 1e and 1f the nitrogen atom of the pyridine is more
hindered, which may reduce the binding ability of the ligand.
(40) Internal alkynes have been found to react under CuAAC conditions; see:
D´ıez-Gonza´lez, S.; Correa, A.; Cavallo, L.; Nolan, S. P. Chem.sEur. J.
2006, 12, 7558-7564.
(41) Macrocycle 1k is also photosensitive and decomposes slowly in ambient
light.
(42) Bidentate bipyridine and phenanthroline ligands have been shown to
significantly enhance the kinetics of the CuAAC reaction; see ref 17b.
9
11958 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Catalytic “Active-Metal” Template Strategy to Rotaxanes
A R T I C L E S
In an attempt to minimize the amount of ligandless Cu(I)
present in the reaction, we carried out the active-metal template
CuAAC protocol under the standard set of conditions (1 equiv
of 2, 1 equiv of 3, 1 equiv of [Cu(CH3CN)4](PF6), CH2Cl2, rt)
used previously but in the presence of 10 equiv of macrocycle
(Scheme 3). Our initial experiments43 were carried out with
monodentate macrocycle 1l, and we were immediately intrigued
to find that the reaction was much slower with 10 equiv of the
macrocycle than it had been with 1 equiv, the higher macrocycle:
Cu(I) reaction taking more than one week to go to completion.
Upon reaching this end point, the mixture of triazole products
was found to consist of 30% thread 5, 37% [2]rotaxane 4l, and,
to our surprise, 33% [3]rotaxane 7l (yields quoted with respect
to the alkyne and azide reactants). A reaction using macrocycle
1a under the same experimental conditions (1 equiv of 2, 1 equiv
of 3, 1 equiv of [Cu(CH3CN)4](PF6), and 10 equiv of 1a, CH2-
Cl2, rt) was again slower than the same reaction with 1 equiv
of macrocycle (3 days to reach completion compared to 24 h),
generating a product mixture of 5% thread 5, 90% [2]rotaxane
4a, and 5% [3]rotaxane 7a (yields quoted with respect to the
alkyne and azide reactants). A similar but less dramatic trend
was seen with bidentate macrocycle 1n; with 10 equiv of 1n
the reaction took 10 days to complete (cf. 3 days with 1 equiv),
producing 3% thread 5 and 97% [2]rotaxane 4n (in this case
no [3]rotaxane 7n could be detected by 1H NMR spectroscopy).
The remarkable features of these high macrocycle:Cu(I) ratio
active-metal template reactions are the following:
(i) Exceptional combined rotaxane yields: 95% ([2]rotaxane
4a + [3]rotaxane 7a) compared to 57% for 4a with 1 equiv of
1a; 70% ([2]rotaxane 4l + [3]rotaxane 7l) compared to 25%
for 4l with 1 equiv of 1l; 97% ([2]rotaxane 4n) compared to
45% for 4n with 1 equiv of 1n.
Figure 5. 1H NMR spectra (400 MHz, CDCl3, 300 K) of (a) triazole thread
5, (b) [2]rotaxane 4a, (c) [2]rotaxane 4b, (d) [2]rotaxane 4i, (e) [2]rotaxane
4j, (f) [2]rotaxane 4m, and (g) [2]rotaxane 4n. The assignments correspond
to the lettering shown in Scheme 2 and Figure 4.
(ii) Significant slowing of the reaction rates compared to the
low macrocycle:Cu(I) ratio reactions. The largest effect on the
rate is seen with the weakly copper-binding macrocycle 1l; the
smallest effect on the rate occurs for the strongly copper-binding
macrocycle 1n. Again, this is strongly suggestive experimental
evidence that the dominant mechanism of the CuAAC reaction
under these conditions involves activation of the copper
σ-acetylide unit by a second, preferably ligandless for steric
reasons, copper atom (i.e., I or II, Scheme 3).
the triazole unit of the thread in an interaction similar to that
previously observed between interlocked amide and pyridine
components in rotaxanes and catenanes.1j,29
Active-Metal Template CuAAC Reaction at High
Macrocycle:Cu(I) Ratios: Unexpected Formation of
[3]Rotaxanes
The picture of the active-metal template CuAAC reaction that
we can build up that is consistent with the experimental
observations so far is that the ligand-free [Cu(CH3CN)4](PF6)
salt and macrocycles 1a-o are in equilibrium with the corre-
sponding copper(I)-macrocycle complex in which, in most
cases, the metal atom directs the cycloaddition of the azide and
alkyne through the macrocycle cavity. Although the ligand-free
reaction is inherently faster than the active-metal template one,
the coordinating ability of the macrocycle means that the
rotaxane-forming reaction can become competitive with, or even
dominate, the thread-forming reaction. In general, a stronger
binding macrocyclic ligand (e.g., 1a compared to 1b) should
move this equilibrium in favor of rotaxane formation. However,
some changes in coordination geometry about the metal are
required for the CuAAC mechanism to operate (Scheme 1), so
the most strongly binding macrocycle (1o), which apparently
does not tolerate such changes, actually inhibits the rotaxane-
forming reaction. In view of this, we decided to investigate other
conditions in which rotaxane formation might be increased at
the expense of the thread.
(iii) Formation of [3]rotaxane (in 33% yield using macrocycle
1l)si.e., two macrocycles being threaded during the formation
of one triazole ring. Molecular models show that this can most
reasonably occur through the sort of bridged two copper atom
intermediate III shown in Scheme 3. Since such a doubly
bridged intermediate cannot occur with bidentate ligands (and
no [3]rotaxane is observed with 1n), it seems likely that
monodentate pyridine ligand-promoted CuAAC reactions pro-
ceed via doubly bridged intermediates such as I (Scheme 3),
the equivalent of intermediate B(ii) in Scheme 1, whereas
bidentate bipyridyl ligand-promoted CuAAC reactions proceed
(43) We chose macrocycle 1l for this study because it is a rather poor ligand
(indeed, we were unable to generate complexes with it using several other
metals) due to the steric crowding around the pyridine nitrogen and the
π-electron density presented to the low-oxidation-state copper from the
adjacent aromatic rings. The rather weak Cu(I) binding affinitysmeaning
there is more ligandless23 Cu(I) present in solution than many of the other
macrocycles shown in Figure 4sis most likely the reason for the modest
yield of rotaxane using 1 equiv of this macrocycle. We reasoned the yield
should be improved by increasing the macrocycle:copper ratio.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11959

A R T I C L E S
Aucagne et al.
Scheme 3. Effect of the Macrocycle:Cu(I) Ratio on the Active-Metal Template CuAAC Reactiona,b
a Reagents and conditions: (i) 1a, 1l, or 1n (10 equiv), 2 (1 equiv), 3 (1 equiv), [Cu(CH3CN)4](PF6) (1 equiv), CH2Cl2, rt, 24 h (1a), 7 days (1l), 10 days
(1n); (ii) KCN, CH2Cl2/CH3OH. Product yields starting from macrocycle 1a: thread 5 (5%), [2]rotaxane 4a (90%), [3]rotaxane 7a (5%). Product yields
starting from macrocycle 1a: thread 5 (5%), [2]rotaxane 4i (33%), [3]rotaxane 7l (37%). Product yields starting from macrocycle 1n: thread 5 (3%),
[2]rotaxane 4n (97%), [3]rotaxane 7n (0%). bIn the reactive intermediates shown, the copper azide (orange) reacts with any of the copper alkyne units shown
in green. L can be any nonreacting ligand, including other alkyne and azide groups.
Stoichiometric and Catalytic Active-Metal Template
Synthesis of Bistriazole [2]Rotaxanes
via simple π-coordinated complexes such as II (Scheme 3), the
equivalent of intermediate A(ii) in Scheme 1.
1
The H NMR spectra of [2]rotaxane 4l and [3]rotaxane 7l
To see whether the copper(I) would turn over catalytically
during the formation of a single molecule, we investigated
systems which require two cycloaddition reactions to be
catalyzed in the formation of each molecule of [2]rotaxane
(Scheme 4). [2]Rotaxanes 10 and 12 were each obtained in good
yields (81% and 74%, respectively) using the conditions shown
in Scheme 4 (0.5 equiv of Cu(I) for each triazole ring formed
in a rotaxane) with either 1a, 2, and the diazide 8 or 1a, 3, and
the dialkyne 9. Since the rotaxane-Cu(I) complex in 4a is strong
enough to sequester the metal in the absence of pyridine, the
lack of formation of more than trace amounts of [3]rotaxane in
the reaction mixtures from Scheme 4 suggests that one mac-
rocycle-Cu(I) complex may act processively to catalyze the
formation of both triazole rings of an individual rotaxane thread.
Such a hypothesis is consistent with the observation25 by
Rodionov et al. (albeit in a ligand-free Cu(I) system) of a strong
kinetic enhancement in the second cycloaddition reaction when
using diazide reactants. Reducing the amount of the copper salt
to 20 mol % with respect to 1a (10 mol % with respect to each
are shown in parts b and c, respectively, of Figure 6. The
resonances for the axle (Hf-j) of the rotaxanes are shifted upfield
compared to the signal for the free noninterlocked thread 5
(Figure 6a). The effect is more pronounced in the [3]rotaxane
(Figure 6c) than in the [2]rotaxane (Figure 6b). Due to the
asymmetry of the thread, the 1H NMR spectrum of [3]rotaxane
7l (Figure 6c) displays inequivalent but overlapping peaks for
the two macrocycle components, which are double the intensity
of the corresponding macrocycle signals in the [2]rotaxane.
Single crystals of 7l suitable for X-ray crystallography were
grown by slow cooling of a hot saturated solution of 7l in
acetonitrile. The solid-state structure (Figure 7) confirmed the
constitution of 7l as a [3]rotaxane. Although it may superficially
appear that face-to-face π-stacking interactions might aid the
relative orientation of the two macrocycles on the thread, in
the solid state the aromatic rings of adjacent macrocycles are
not coplanar and their separation (>3.8 Å) is somewhat greater
than that typically associated with aromatic stacking interactions.
9
11960 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Catalytic “Active-Metal” Template Strategy to Rotaxanes
A R T I C L E S
Scheme 4. Active-Metal Template CuAAC Synthesis of Bistriazole
Molecular Shuttles 10 and 12a
Figure 6. 1H NMR spectra (400 MHz, CDCl3, 300 K) of (a) triazole thread
5, (b) [2]rotaxane 4l, and (c) [3]rotaxane 7l. The assignments correspond
to the lettering shown in Scheme 2 and Figure 4.
a Reagents and conditions: (i) 2, [Cu(CH3CN)4](PF6), CH2Cl2, 25 °C,
72 h; (ii) KCN, CH2Cl2/CH3OH, 81% ([2]rotaxane 10) over two steps with
0.5 equiv of Cu(I), 63% ([2]rotaxane 10) over two steps with 0.1 equiv of
Cu(I) + 3 equiv of pyridine; (iii) 3, [Cu(CH3CN)4](PF6), CH2Cl2, 25 °C,
72 h; (iv) KCN, CH2Cl2/CH3OH, 74% ([2]rotaxane 12) over two steps.
Figure 7. X-ray structure of [3]rotaxane 7l, from a single crystal obtained
from a saturated acetonitrile solution: (a) viewed side-on; (b) viewed along
the axis of the thread. For clarity, different shades of purple are used for
the carbon atoms of the two rings.
rotaxane triazole ring formed) and introducing 3 equiv of
pyridine to the reaction of 1a, 2, and 8 to allow the processive
catalyst to turn over intermolecularly still generated [2]rotaxane
10 in 63% yield (Scheme 4).
the thread signals (Hd, He, Hf, Hg, Hh), indicating that the
macrocycle randomly explores the full length of the thread.
Reintroducing copper(I) into the bistriazole rotaxane44 to form
[Cu10](PF6) (Scheme 5) gave a complex that exhibited fast
shuttling of the macrocycle between the two triazole units in
the room temperature 1H NMR spectra in CD3CN/CD2Cl2
Transition-Metal-Mediated Control of the Shuttling Rate
in Degenerate Two-Triazole-Station Molecular Shuttles
The shuttling of the ring in the degenerate two-station
bistriazole rotaxane 10 in the presence of different diamagnetic
transition-metal salts was investigated using 1H NMR spectros-
copy. Unlike most previously reported molecular shuttless
where the interactions used to direct the synthesis of the
rotaxanes persist in the interlocked structure, resulting in a well-
defined coconformationsthere are no strong noncovalent in-
teractions between the components of metal-free 10. Comparison
of the 1H NMR spectrum of the free thread 11 (Figure 8a) with
that of the rotaxane 10 (Figure 8b) shows uniform shielding of
(44) For examples of 1,2,3-triazoles as ligands see refs 3, 6s, and 17a,b and (a)
Liu, D.; Gao, W.; Dai, Q.; Zhang, X. Org. Lett. 2005, 7, 4907-4910. (b)
Dai, Q.; Gao, W.; Liu, D.; Kapes, L. M.; Zhang, X. J. Org. Chem. 2006,
71, 3928-3934. (c) Mindt, T. L.; Struthers, H.; Brans, L.; Anguelov, T.;
Schweinsberg, C.; Maes, V.; Tourwe, D.; Schibli, R. J. Am. Chem. Soc.
2006, 128, 15096-15097. (d) Suijkerbuijk, B. M. J. M.; Aerts, B. N. H.;
Dijkstra, H. P.; Lutz, M.; Spek, A. L.; van Koten, G.; Klein Gebbink, R.
J. M. Dalton Trans. 2007, 1273-1276. (e) Huffman, J. C.; Flood, A. H.;
Li, Y. Chem. Commun. 2007, 2692-2694.
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11961

A R T I C L E S
Aucagne et al.
Scheme 5. Controlling the Dynamics of Bistriazole Molecule
Shuttle 10 by Coordination to Cu(I) or Pd(II)a
Figure 8. 1H NMR spectra (400 MHz, CD3CN/CD2Cl2 (9/1), 300 K) of
(a) bistriazole thread 11, (b) bistriazole [2]rotaxane 10, (c) copper-
bistriazole [2]rotaxane complex [Cu10](PF6), (d) palladium dichloride-
bistriazole [2]rotaxane complex Cl2Pd10, and (e) palladium dichloride-
triazole rotaxane complex Cl2Pd4a. The assignments correspond to the
lettering shown in Scheme 5.
(Figure 8c).45 Although fast shuttling between well-defined
macrocycle binding sites is very unusual for metal-coordinated
molecular shuttles,46 Cu(I) complexes can exhibit extremely fast
1
exchange rates.47 In contrast to this dynamic behavior, the H
NMR spectrum (Figure 8d) of rotaxane 10 following coordina-
tion of PdCl2 (Scheme 5, ii) displays two sets of peaks for
protons close to the thread triazole rings (e.g., Hf/f′, Hg/g′). This
observation is indicative of the macrocycle in the Cl2Pd10
complex being unable to shuttle between the triazole stations
on the NMR time scale. This restriction to shuttling also
manifests itself in the geminal splitting of the signals corre-
sponding to HC and HD of the macrocycle, which occurs as a
consequence of the two “faces” of the ring being in different
chemical environments. The chemical shifts of one set of the
duplicated signals (e.g., Hf) correspond closely to those of the
free thread 11 (Figure 8a). The other set of duplicate signals
(e.g., Hf′) correspond closely to a model single-triazole rotax-
ane-PdCl2 complex, Cl2Pd4a (Figure 8e, Scheme 5). Heating
the sample of Cl2Pd10 to 343 K (close to the solvent boiling
point of CD3CN) did not lead to any broadening of the signals
in the 1H NMR spectrum, indicating that the macrocycle
shuttling is very effectively blocked by coordination to Pd.
a Reagents and conditions: (i) [Cu(CH3CN)4](PF6) (1 equiv), CD2Cl2
(90%)/CD3CN (10%), 25 °C, 5 min, >99%; (ii) Pd(CH3CN)2Cl2 (1 equiv),
CD2Cl2 (90%)/CD3CN (10%), 25 °C, 5 min, >99%.
remarkable ease, simply through the judicious choice of different
transition-metal ions.
The notable features of these degenerate two-station bistria-
zole rotaxanes are the following:
(iii) An active-metal template strategy is much more amend-
able to the synthesis of such shuttles than traditional (passive)
metal template strategies, because the number of binding sites
in the products is not necessarily related to the number of
template sites used for the shuttle synthesis.
(i) To our knowledge these are the first examples of molecular
shuttles containing two degenerate transition-metal-binding
stations.
(ii) The dynamics of the macrocycle exchange between the
two well-defined discrete binding sites can be controlled with
Conclusions
The Cu(I)-catalyzed 1,3-cycloaddition of azides with terminal
alkynes is a highly effective reaction for the active-metal
template synthesis of rotaxanes, as illustrated through both
stoichiometric (up to 97% yields) and catalytic (up to 82%
yields) versions of the strategy. The reaction tolerates a wide
(45) We were unable to slow the macrocycle shuttling dynamics in [Cu(10)]-
(PF6) sufficiently, nor isolate that motion from other conformational
processes at low temperatures, to determine an accurate shuttling rate.
(46) Durola, F.; Sauvage, J.-P. Angew. Chem., Int. Ed. 2007, 46, 3537-3540.
(47) For an evaluation of ligand exchange rates at Cu(I) centers, see: Riesgo,
E.; Hu, Y.-Z.; Bouvier, F.; Thummel, R. P. Inorg. Chem. 2001, 40, 2541-
2546.
9
11962 J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007

Catalytic “Active-Metal” Template Strategy to Rotaxanes
A R T I C L E S
variety of both monodentate and bidentate macrocyclic ligands
which, although inherently slower in promoting covalent bond
formation than the ligand-free23 Cu(I)-catalyzed reaction, all
work by sequestering the majority of the Cu(I) present so that
the rotaxane-forming CuAAC reaction becomes competitive
withsand under many conditions dominatessthe ligand-free
CuAAC reaction. The addition of pyridine enables the Cu(I)
active-metal template to turn over without significantly com-
promising the rotaxane yield. Among several potentially attrac-
tive features of this type of synthetic strategy is that it offers an
unusual experimental probe of the reaction mechanism. The
increase in rates of rotaxane formation by excess ligand-free
Cu(I) supports the suggestion that in CH2Cl2 the copper
σ-acetylide is activated by coordination to a second ligandless
copper atom, and the sluggish reaction rates and remarkable
formation of [3]rotaxanes with monodentate macrocycles at high
macrocycle:Cu(I) ratios suggest that, under these conditions,
the CuAAC reaction proceeds via a doubly bridged two copper
atom intermediate (type B(ii), Scheme 1) when using a
monodentate macrocyclic ligand and a simpler π-coordinated
two copper atom intermediate (type A(ii), Scheme 1) when using
a bidentate macrocyclic ligand.48
in which the ring still shuttles rapidly between the two triazole
units of the thread, even at low temperatures in noncoordinating
solvents. In contrast, complexation of the same two-triazole-
station shuttle to PdCl2 gives a rotaxane in which shuttling does
not occur even at 343K in the presence of d3-acetonitrile.
The template-directed assembly of otherwise difficult-to-
access molecular architectures and the catalysis of covalent bond
formation are two of the most useful tasks that transition metals
can perform in organic chemistry. The active-template concept
combines these two apparently disparate functions into one,
which can producesas the present work exemplifiesshigh-
yielding, mild and effective synthetic routes to complex mo-
lecular structures requiring only catalytic quanitities of metal.
The development of other active-template systems is in progress.
Acknowledgment. This work was supported by the European
Union Future and Emerging Technology program Hy3M, the
Ramsay Memorial Fellowships Trust, the Royal Society of
Edinburgh, and the EPSRC. D.A.L. is an EPSRC Senior
Research Fellow and holds a Royal Society-Wolfson Research
Merit Award, P.J.L. is a Royal Society University Research
Fellow, and J.D.C. is a British Ramsay Memorial Fellow. We
thank Patrice Staub and Professor J.-P. Sauvage (Strasbourg)
for generously providing us with a sample of macrocycle 1o,
termed “m-30” by the Strasbourg group in many papers. We
thank Professor B. F. Straub (Munich) for a preprint of ref 26.
We thank Professor V. V. Fokin (Scripps) for a preprint of ref
21 and for sharing with us his thoughts on the mechanism of
the CuAAC reaction.
The experimental results concerning the active-metal template
CuAAC synthesis of bistriazole rotaxanes suggest that the
macrocycle-Cu(I) complex may act processively to catalyze
the formation of more than one triazole ring per thread molecule.
The resulting two-triazole-station molecular shuttles have
interesting and unusual dynamic properties: coordination of the
rotaxane macrocycle to Cu(I) gives a metal-complexed shuttle
Supporting Information Available: Synthetic schemes, ex-
perimental procedures, and spectroscopic data for all new
compounds. This material is available free of charge via the
Internet at http://pubs.acs.org.
(48) Note added in proof: Very recent calculations (Straub, B. F. Chem. Commun.
published online 12th July 2007 DOI: 10.1039/b706926j) suggest that the
metallacycle formed in Scheme 1 consists of a strain-free Cu-C(Cu))C
unit rather than the depicted Cu)C)C unit. The second copper atom in
the type B(ii) and A(ii) reactive intermediates shown in Scheme 1 means
they are perfectly set up to form such a Cu-C(Cu))C metallacycle.
JA073513F
9
J. AM. CHEM. SOC. VOL. 129, NO. 39, 2007 11963