Ligand-accelerated Cu-catalyzed azide-alkyne cycloaddition: A mechanistic report

Article abstract of DOI:10.1021/ja072679d

The experimental rate law for the Cu(I)-catalyzed azide-alkyne cycloaddition (CuAAC) reaction was found to vary in complex ways with concentration, the presence of chloride ion, and the presence of accelerating ligands. Several examples of discontinuous (

Full text of DOI:10.1021/ja072679d

Published on Web 10/03/2007
Ligand-Accelerated Cu-Catalyzed Azide-Alkyne
Cycloaddition: A Mechanistic Report
Valentin O. Rodionov, Stanislav I. Presolski, David D´ıaz D´ıaz, Valery V. Fokin, and
M. G. Finn*
Contribution from the Department of Chemistry and The Skaggs Institute of Chemical Biology,
The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037
Received April 17, 2007; E-mail: mgfinn@scripps.edu
Abstract: The experimental rate law for the Cu(I)-catalyzed azide-alkyne cycloaddition (CuAAC) reaction
was found to vary in complex ways with concentration, the presence of chloride ion, and the presence of
accelerating ligands. Several examples of discontinuous (“threshold behavior”) kinetics were observed,
along with a decidedly nonlinear correlation of electronic substituent parameter with the rate of CuAAC
reaction with p-substituted arylazides. The previously observed tendency of the CuAAC reaction to provide
ditriazoles from a conformationally constrained 1,3-diazide was found to be affected by a class of
polybenzimidazole ligands introduced in the accompanying article. Various lines of evidence suggest that
the standard tris(triazolylmethyl)amine ligand binds less strongly to Cu(I) than its benzimidazole analogues.
On the basis of these observations, it is proposed that (a) a central nitrogen donor provides electron density
at Cu(I) that assists the cycloaddition reaction, (b) the three-armed motif bearing relatively weakly
coordinating heterocyclic ligands serves to bind the metal with sufficient strength while providing access to
necessary coordination site(s), (c) at least two active catalysts or mechanisms are operative under the
conditions studied, and (d) pendant acid or ester arms in the proper position can assist the reaction by
speeding the protiolysis step that cleaves the Cu-C bond of a Cu‚triazolyl intermediate.
Introduction
In the presence of Tris buffer, we observed for the first time
an experimental rate law in which catalyst, azide, and alkyne
appear together with integer representation corresponding to the
values observed in the absence of ligand (Table 1, entries 1-4):
The preceding article in this issue describes the synthesis and
utility of benzimidazole- and benzothiazole-based ligands for
the Cu(I)-catalyzed azide-alkyne cycloaddition (CuAAC) reac-
tion. Our previous study of the reaction mechanism1 did not
involve the use of accelerating ligands. Here we describe the
results of kinetic studies involving representative examples of
such ligands, along with an interpretation of the results that
allows for predictive hypotheses to be made about the reaction
mechanism.
rate ) k[BnN3]1[PhCCH]1[Cu‚(BimH)3]2
In the absence of Tris buffer, however, the picture was more
complicated. The copper complexes of benzimidazole deriva-
tives (BimH)3 and (BimC4A)3 exhibited a threshold behavior:
below 50 µM in Cu and 100 µM in ligand, the catalysts gave
only a few very rapid turnovers and then the reactions stopped
(Supporting Information). At higher concentrations (100-250
µM), sustained and rapid reactivity was observed, but the rate
for both Cu‚[(BimH)3]2 and Cu‚[(BimC4A)3]2 systems increased
only marginally with increasing catalyst concentration. (We have
observed similar behavior for ligands (BimC1A)3 and (Py)3 in
the presence of HEPES buffer; data not shown.) Discontinuous
kinetics was also observed for azide in the absence of Tris-HCl
buffer, with a modest rate order dependence on azide concentra-
tion of approximately 0.4 at 10-20 mM, changing to concentra-
tion-independent catalysis above 40-50 mM (Table 1, entry
7). At lower catalyst loading, the reaction was zero-order in
azide throughout the range tested (entry 11). The experimental
rate order in alkyne was found to be approximately 0.6 (entries
8 and 12) at both 200 and 50 µM concentrations of Cu‚
[(BimH)3]2.
Results and Discussion
Kinetic Rate Law in the Presence of Accelerating Ligands.
We reported earlier that in the absence of added accelerating
ligands the rate of CuAAC reaction displays second-order
dependence on Cu (with rate independent of azide and alkyne
concentration) under dilute catalytic conditions. Under single-
turnover conditions, the rate had approximately first-order
sensitivities to azide and alkyne concentration and was inde-
pendent of [Cu].1 These results are consistent with changes in
the turnover-limiting step of a catalytic cycle in response to
changes in the relative concentrations of the participating
species. Table 1 shows the results of similar kinetic measure-
ments performed in the presence of accelerating ligands (Figure
1).
Since the pH of the reactions in the presence and absence of
Tris-HCl buffer was made the same, the excess ascorbate serving
(1) Rodionov, V. O.; Fokin, V. V.; Finn, M. G. Angew. Chem., Int. Ed. 2005,
44, 2210-2215.
9
10.1021/ja072679d CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 12705-12712
12705

A R T I C L E S
Rodionov et al.
Table 1. Experimental Rate Orders for Reactions in 4:1 DMSO/Aqueous Buffer, 25 mM Na Ascorbate, 24 ( 1 °Ca
aqueous
[PhCCH]
(mM)
[BnN3]
(mM)
[Cu+]
(mM)
varied
rate
entry
component
ligand
reactant
order
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Trisb
Trisb
Trisb
Trisb
water
water
water
water
KClc
KClc
water
water
water
water
water
1
1
1
0.050-0.30
0.10
0.50
(BimH)3
(BimH)3
(BimH)3
(BimH)3
(BimH)3
(BimC4A)3
(BimH)3
(BimH)3
(BimH)3
(BimH)3
(BimH)3
(BimH)3
TBTA
Cu+/L2
BnN3
PhCCH
PhCCH
Cu+/L2
Cu+/L2
BnN3
PhCCH
BnN3
PhCCH
BnN3
2.32 ( 0.3
10-25
1
1
1
0.93 ( 0.2
1.01 ( 0.2
1.07 ( 0.2
see text
10-25
10-25
1
1
1
10-50
1
10-50
1
10-50
1
10-50
1
0.10
0.01-0.25
0.01-0.25
0.20
0.20
0.20
0.20
0.050
0.050
1
see text
10-50
1
10-50
1
10-50
1
1
1
<0.4, see text
0.64 ( 0.2
0.84 ( 0.2
0.60 ( 0.2
0
0.58 ( 0.2
1.01 ( 0.2
-0.28 ( 0.2
0
PhCCH
Cu+/L2
PhCCH
BnN3
0.020-0.25
0.050
0.050
TBTA
TBTA
10-50
a Reactions were started by the addition of CuSO4 solution, except in reactions employing varying amounts of Cu, which were started by adding ascorbate
solution. Ligand abbreviations are explained in the caption to Figure 1. b Tris buffer containing 133 mM Tris base neutralized with 78 mM HCl, along with
25 mM sodium ascorbate; pH 8.0 ( 0.05 measured at room temperature. c 133 mM aqueous KCl, along with 25 mM sodium ascorbate; pH 8.0 ( 0.1
measured at room temperature.
Figure 1. CuAAC reaction used in kinetics studies, and the structures of ligands mentioned in the text. For convenience, most ligands are designated by
an abbreviation to represent structure, rather than by number, with the following conventions: BimR ) benzimidazoylmethyl with N-substituent R, attached
to amine N; Bth ) benzothiazoylmethyl; CnR ) chain of n methylene groups terminated with R; E ) CO2Et; E′ ) CO2t-Bu; A ) CO2-K+.
as a low capacity buffer, the difference in kinetic profile under
these conditions is most probably due to the presence of either
the Tris base or the chloride counterion. The hindered primary
amine group of Tris was not expected to be a strong ligand for
Cu(I). Indeed, the addition of KCl inhibited the overall rate by
approximately a factor of 3 and removed the nonlinear response
of rate to azide concentration, giving a nearly unimolecular rate
order in this reagent (Table 1, entry 9). The observed rate order
in alkyne was unchanged in the presence and absence of the
added salt (entry 10 Vs 8). It therefore appears that the presence
of high concentrations of chloride influences the distribution
of catalytic species in solution. The sensitivity of rate order in
azide, but not in alkyne, to chloride concentration suggests that
Cl- and RN3 interact with copper(I) with similar affinities,
whereas alkyne/acetylide is a much stronger ligand than chloride
and is therefore insensitive to its presence.
dependent on the amount of added ligand (Supporting Informa-
tion). In other words, at a ligand/Cu ratio of 0.4:1, a faster initial
reaction was observed, amounting to approximately 30 turnovers
per metal compared to five turnovers at a 5:1 ligand/Cu ratio,
before settling down to a constant rate. These observations echo
the finding by reaction calorimetry described in the ac-
companying article of two catalytic modes: one that is sup-
pressed by extra ligand and one that is not. At lower concen-
trations of substrate, the former appears to be only transiently
available at the beginning of the reaction. The observation of
nonlinear “threshold” behavior shows that the catalysts change
their nature depending on overall concentration and suggests
that aggregates are likely to be important.
A comparison of tris(triazolylmethyl)amine (TBTA) to tris-
(benzimidazolylmethyl)amine ligands under otherwise identical
conditions (Table 1, entries 13 Vs 5 and 6; 14 Vs 12; 15 Vs 11)
showed the same zero-order insensitivity to [BnN3], but different
kinetic response to the other components. TBTA elicits none
of the “threshold” performance in [CuL2] observed with
(BimH)3 or (BimC4A)3, and an apparent first-order dependence
was observed in Cu for the first time (Table 1, entry 13). The
rate dependence on Cu‚TBTA in Tris buffer could not be
The presence of more than one active species was also
indicated by experiments in which [Cu] was held constant at
50 µM and [(BimH)3] was increased from 0.4 to 5 equiv with
respect to metal. While the steady-state second-order rate of
these reactions did not change, an initial amount of rapid
catalytic activity was again observed, which was inversely
9
12706 J. AM. CHEM. SOC. VOL. 129, NO. 42, 2007

Ligand Acceleration of the CuAAC Reaction
A R T I C L E S
Figure 2. (A) Relative product concentrations for the reaction of 1 (2 mM) with phenylacetylene (2 mM) in the presence of 0.1 mM CuSO4, 0.2 mM of
the indicated ligands, and 20 mM Na ascorbate (4:1 DMSO/H2O, room temperature) after approximately 30 h. Entries are arranged in order of relative
consumption of alkyne. (B) Comparison of apparent second-order initial specific activities (rate constants at standard Cu concentration of 0.1 mM) for the
production of 2 and 3 for the indicated Cu‚ligand catalysts. Above each bar are the relative rates of overall consumption of alkyne, with the ligand-free
reaction set at 1.0. (C) The same as part A, for different ligand/Cu ratios using the indicated four ligands, reaction time approximately 20 h. The three
categories of ligands described in the text (classes I, II, and III) are shown in black, red, and blue, respectively. Experimental error for the great majority of
values plotted here is (15% or less; complete details are given in Supporting Information.
measured because the reaction was very slow. Furthermore, in
the presence of TBTA, the CuAAC reaction was slightly
inhibited by the addition of alkyne (rate order ≈ -0.3),2 whereas
the Cu‚[(BimH)3]-catalyzed process was accelerated (rate order
in alkyne ≈ 0.6). These data are consistent with weaker
coordination of the active metal center by the triazole ligand
relative to the benzimidazole structure (see below). We therefore
expect Cu‚TBTA complexes to exchange more rapidly and to
not get trapped in the kind of dead end that leads to the
requirement for a threshold concentration. By the same token,
we suggest that alkyne and acetylide may be able to displace
TBTA from the metal center, thus resulting in the observed
inhibitory behavior by loss of ligand-accelerated catalysis.
Ligand Effects on the Reactivity of 1,3-Diazides. An
unusual phenomenon described in our previous report was the
reactivity of 1,2- and conformationally constrained 1,3-diazides,
most notably 2,2-bis(azidomethyl)propane-1,3-diol 1 (Figure 2).1
The bistriazole 3 was found to be formed in preference to
monotriazole 2, even in the presence of excess diazide. The
monotriazole was shown not to be an intermediate in the
formation of the final product, and a highly reactive Cu‚
C(triazolyl) species was proposed. We also reported that TBTA,
the only accelerating ligand in wide use at the time, did not
have a significant effect on the outcome of the 1,3-diazide
reaction.
(2) In our original article (ref 1), we noted a weak inhibitory property of benzyl
azide. We have subsequently found that commercial benzyl azide frequently
contains an unidentified trace impurity that either inhibits the CuAAC
reaction or poisons the copper catalyst. No inhibitory effect was observed
when carefully distilled benzyl azide was employed, and therefore our prior
report is in error in this respect. Such distilled material was used in all
experiments described in these articles.
We have observed that (BimH)3, in contrast to TBTA, has a
profound influence on the reaction of 1 with phenylacetylene,
9
J. AM. CHEM. SOC. VOL. 129, NO. 42, 2007 12707

A R T I C L E S
Rodionov et al.
giving rise to monotriazole as the major product (3.6:1 ratio of
2/3 for the starting 1:1 ratio of 1/phenylacetylene). The results
of a survey of 21 ligands in this process are summarized in
Figure 2A, revealing the existence of three major categories of
performance: (I) inhibitory ligands that provided an excess of
monotriazole [(Bth)(BimH)2, (BimC1A)3, H(BimH)2], (II)
ligands that affected neither the rate of the reaction nor the
preference of the ligand-free process to give predominantly
bistriazole [1,10-phenanthroline, 2,2′-bipyridine, H(Bth)2, TBTA,
(Bth)3, 4, and probably (Py)3], and (III) accelerating ligands
that gave rise to comparable amounts of mono- and bistriazole
[(BimH/Me)3, (BimH)3, Py(BimH)2, Py(Tet)2, (BimC1H)3,
(Bth)2BH, (BimC1E)3, (BimC1E′)3, (BimC4A)3].
Because Figure 2A reports on the product distribution after
a relatively long reaction time, we also measured the apparent
initial rate constants for production of 2 and 3 involving a subset
of class II ligands and additional examples of pendant alkyl-
carboxylate analogues of the benzimidazole family, all of which
proved to be in class III (Figure 2B). With the exception of
(BimH)3, which initially provided almost exclusively monot-
riazole, the product selectivities for each catalyst did not vary
a great deal over the course of the reaction. Figure 2B further
shows that the overall rate of reaction was directly proportional
to the length of the carboxylate arm: (BimC5A)3 > BimC4A
derivatives > (BimC3A)3.
A single pathway process, in which monotriazole is an
intermediate on the way to bistriazole, would give a 1.2:1 ratio
of 2/3 if the rates of the two triazole-forming steps were equal
(Supporting Information). The results from class III ligands are
consistent with this model, suggesting that these ligands subvert
the ditriazole-selective reaction in the absence of ligand by
defeating an alternative reaction pathway that gives ditriazole
selectively. Four ligands of the last class [Py(BimH)2, Py(Tet)2,
Np(BimH)2, and NpCH2(BimH)2] were tested in varying
ligand/Cu ratios (Figure 2C). These were investigated because
of earlier observations that 2,6-bis(heterocyclic)pyridyl and some
bis(heterocyclic methyl)amines were inhibitors of the reaction.
The addition of increasing amounts of each ligand slowed the
reaction and produced less ditriazole, as would be expected from
a stepwise process. It must be emphasized, however, that the
reactions with small amounts of ligand (0.5 equiv relative to
Cu) were significantly faster than those with no added ligand,
even while the ditriazole selectivity was diminished. Ligand
[NpCH2(BimH)2] was an exception, proving to be a potent and
dose-dependent inhibitor of the CuAAC reaction of 1. This
dramatic difference upon the addition of a single methylene unit
at the central nitrogen atom is discussed below.
Figure 3. Rate Vs substituent σ and σ+ parameters for the CuAAC reaction
of PhCCH (50 mM) with the indicated aromatic azides (2 mM) in the
presence of ligand (BimH)3 (0.2 mM), CuSO4 (0.1 mM), sodium ascorbate
(25 mM), and 1-d7-benzyl-4-phenyl-1,2,3-triazole as internal standard, in
4:1 t-BuOH/H2O, at 25 °C. Aromatic azide was added last to start the
reaction.
phenyl azide derivatives were prepared from p-substituted
anilines by diazotization and azide trapping. They were found
to be prone to decomposition upon standing under ambient light
and even to some extent in the dark. In the absence of
accelerating ligands, such decomposition is competitive with
the CuAAC reactions. We therefore employed the parent tris-
(benzimidazole) (BimH)3 to speed up the desired cycloaddition
reaction to obtain clean kinetics data for a substituent effect
study, the results of which are shown in Figure 3.
Second-order rate constants were determined in the presence
of 5 mol % of the Cu‚[(BimH)3]2 precatalyst mixture relative
to azide, with minimal exposure of azide stock solutions and
the reaction mixtures to light. The solvent used was t-BuOH/
water, 4:1, instead of the usual DMSO/water mixture, to avoid
the formation of unwanted photolysis products.6 To obtain
absolute concentrations, calibration curves were constructed
using purified samples of each 1,4-triazole product. The kinetics
reveal a bifurcation in the electronic dependence, with rate
increasing for substituents both more electron-withdrawing and
electron-donating than H or CH3. A full exploration of the
questions raised by this observation must await more complete
kinetic measurements, such as extension to other substituents
and determination of the experimental rate law at each extreme.
But it is clear that either the mechanism or the turnover-limiting
step of the catalytic cycle changes to some degree with variation
in the electronic properties of the azide, perhaps unsurprising
considering the sensitivity of the rate order in azide to changes
in reaction conditions in the presence of the Cu‚(BimH)3 catalyst
described above. We believe that pre-equilibrium coordination
of the organic azide to Cu may be important,7 as is the
committed kinetic step to form a Cu‚triazolyl intermediate, and
these may have divergent electronic demands.
Electronic Effects in Aromatic Azides. The participation
of azide in the kinetic rate law and the kinetic requirement for
two copper centers under most conditions suggest that a Cu-
N3R interaction may be significant. Indeed, the activation of
organic azide by Cu(I) has been proposed to be a key feature
of the favored pathway identified by ab initio calculations.3,4
We therefore explored the sensitivity of the CuAAC reaction
to changes in the electronic nature of aromatic azides, which
are well-known participants in the process.5 The p-substituted
Mechanistic Considerations. The CuAAC reaction catalytic
cycle comprises the three general steps shown in Figure 4:
(6) We have observed that certain minor products of the photochemical
decomposition of aromatic azides in DMSO act as potent accelerating
ligands for CuAAC reactions of other aromatic azides, increasing rates by
several orders of magnitude and producing unprecedented amounts of 1,5-
triazoles for a copper-mediated process. These phenomena are currently
being investigated and will be described in detail elsewhere. The catalytic
photoproducts do not form in the presence of tert-butanol.
(3) Ahlquist, M.; Fokin, V. V. Organometallics 2007, 26, 4389-4391.
(4) Himo, F.; Lovell, T.; Hilgraf, R.; Rostovtsev, V. V.; Fokin, V. V.;
Noodleman, L.; Sharpless, K. B. J. Am. Chem. Soc. 2005, 127, 210-216.
(5) Feldman, A. K.; Colasson, B.; Fokin, V. V. Org. Lett. 2004, 6, 3897-
3899.
(7) Dias, H. V. R.; Polach, S. A.; Goh, S.-K.; Archibong, E. F.; Marynick, D.
S. Inorg. Chem. 2000, 39, 3894-3901.
9
12708 J. AM. CHEM. SOC. VOL. 129, NO. 42, 2007

Ligand Acceleration of the CuAAC Reaction
A R T I C L E S
oxidation or disproportionation, preventing Cu(II)-mediated
degradation of protein or nucleic acid.
The conformationally flexible tris(benzimidazolylmethyl)-
amines are less potent binders of Cu(I) than phenanthroline-
based ligands, as shown by a competition between (BimH)3
and neocuproine 8, a close analogue of the effective CuAAC
ligand 7. The addition of a large excess of (BimH)3 to a Cu‚8
mixture did not change the characteristic yellow color or
electronic spectrum of the neocuproine complex; the Cu(I)‚
(BimH)3 complex is colorless. To the extent that protic basicity
is correlated with binding affinity (and for Cu(I), such a
correlation is likely to be only partially valid10), triazoles (pKa
of conjugate acid ≈ 0.1-1.0)11 are expected to be the poorest
binders, followed closely by benzothiazoles (pKa ≈ 1.2),12 and
then by pyridines (pKa ≈ 5.2) and benzimidazoles (pKa ≈
6.3).13,14 All of these heterocycles are kinetically labile on Cu-
(I).
Figure 4. Major steps in the proposed catalytic cycle. Each Cu species
can participate in equilibria with larger aggregates (not shown).
activation of terminal alkyne as Cu‚acetylide 5, formal cycload-
dition to give a Cu‚C(triazole) intermediate 6,8 and protiolysis
of the Cu-C bond to give the triazole product and regenerate
the catalyst. Each stage can involve multinuclear Cu species, a
possibility made likely by the rich bridging coordination
chemistry of Cu‚acetylides.9 Copper-binding ligands can con-
ceivably affect the rates of each of these steps. The following
discussion comprises our present informed speculation about
the reaction mechanism that touches on the most interesting
data obtained thus far and provide testable hypotheses for future
work.
In the absence of chelating ligand, the rate of the CuAAC
reaction of phenylacetylene with benzyl azide under catalytic
conditions was found to be second order in copper and
independent of azide and alkyne concentrations.1 The observa-
tion is consistent with saturation of the metal center with azide
and alkyne reactants, as might occur if the protiolysis of Cu‚
triazolyl 6, rather than its formation, were turnover-limiting. In
contrast, the presence of (BimH)3 gave rise to an experimental
rate law involving both substrates and the Cu‚ligand complex,
suggesting that the expected competition by azide and alkyne
for metal-binding sites becomes kinetically significant. This
would be consistent with the formation of Cu‚acetylide 5 and
its cycloaddition to 6 being turnover-limiting. Therefore, the
(BimH)3 ligand, among other effects, may make Cu‚triazolyl
hydrolysis faster, relieving a catalytic bottleneck by suppressing
the buildup of 6.
An important conceptual theme to emerge from these studies
concerns the potential importance of binding affinities of ligand
classes for the Cu(I) center, and the idea that weak binding can
be advantageous. Rigid chelating ligands such as 2,2′-bipyridine
and sulfonated bathophenanthroline are certainly effective for
the CuAAC reaction, but are inhibitory when used in a greater
than 2:1 ligand/Cu excess at low overall concentration, or at
greater than 1:1 ligand/Cu at higher concentration. For biocon-
jugation applications, we have found a 2:1 ratio to be optimal,
but the active catalyst is likely to involve a 1:1 ligand/Cu
combination. As illustrated in Figure 5, the coordination of a
second equivalent of ligand to Cu(I) diverts the metal into a
catalytically inactive form, but a sufficient amount of the active
1:1 system can be accessible in the equilibrium mixture. The
small amount of excess ligand serves the additional purpose of
tying up Cu(II) ions (for which these ligands have a higher
affinity than those for Cu(I)) that may form by adventitious
Multidentate interactions are therefore required for the
imidazole, thiazole, and triazole ligands to have a high overall
affinity for the metal, but open coordination sites are still
available in such complexes because of two factors. First, the
preferred tetrahedral coordination geometry of Cu(I), unlike that
of Cu(II),15-17 makes it difficult for all four donor atoms of
ligands such as TBTA and (BimR)3 to bind the same metal
ion, especially when an additional monodentate ligand is
present.18 In terms of the candidate mononuclear structures
shown in Figure 6, this means that structures 9 and 11, both of
which have crystallographic precedent,15 are much more likely
for Cu(I) than structure 10 (K2 << K1). We further suggest
that electron richness at the metal is helpful to the triazole-
forming step, since analogous ligands lacking the central donor
nitrogen atom do not provide for accelerated CuAAC reactions,
even in the presence of added base. Therefore, structures with
the central tertiary nitrogen bound (derived from 9) should be
much more catalytically active than those in which it is
dissociated (derived from 11). Second, monodentate binding by
an arm of a second ligand is, like any monodentate version of
these heterocycles, of relatively low affinity; in other words,
structure 12 is not favored, and K3 << K1. Thus, many of the
tris(heterocyclic methyl) amine ligands described here do not
shut down catalysis even when used in large excess relative to
the concentration of Cu (Figure 5 of the preceding article), in
contrast to stronger-binding ligands such as phenanthroline.
Coordinatively unsaturated intermediates such as 13 are rela-
tively easily accessible in a kinetic if not a thermodynamic sense
and are crucial to the success of the reaction.
The experimental rate law varies widely depending on the
reaction conditions, suggesting both that the coordination
(10) Sivasankar, C.; Sadhukhan, N.; Bera, J. K.; Samuelson, A. G. New J. Chem.
2007, 31, 385-393.
(11) Abboud, J.-L. M.; Foces-Foces, C.; Notario, R.; Trifonov, R. E.; Volo-
vodenko, A. P.; Ostrovskii, V. A.; Alkorta, I.; Elguero, J. Eur. J. Org.
Chem. 2001, 3013-3024.
(12) Albert, A.; Goldacre, R.; Phillips, J. J. Chem. Soc. 1948, 2240-2249.
(13) Langner, R.; Zundel, G. Can. J. Chem. 2001, 79, 1376-1380.
(14) Huyskens, P. L.; Cleuren, W.; Franz, M.; Vuylsteke, M. A. J. Phys. Chem.
1980, 84, 2748-2751.
(15) Su, C.-Y.; Kang, B.-S.; Wen, T.-B.; Tong, Y.-X.; Yang, X.-P.; Zhang, C.;
Liu, H.-Q.; Sun, J. Polyhedron 1999, 18, 1577-1585.
(16) Osako, T.; Tachi, Y.; Taki, M.; Fukuzumi, S.; Itoh, S. Inorg. Chem. 2001,
40, 6604-6609.
(8) An example of a Cu‚C(triazole) intermediate obtained from a CuAAC
reaction has been recently characterized by X-ray crystallography: Nolte,
C.; Mayer, P.; Straub, B. F. Angew. Chem., Int. Ed. 2007, 46, 2101-2103.
(9) Mykhalichko, B. M.; Temkin, O. N.; Mys’kiv, M. G. Russ. Chem. ReV.
2001, 69, 957-984.
(17) Hendriks, H. M. J.; Birker, P. J. M. W. L.; van Rijn, J.; Verschoor, G. C.;
Reedijk, J. J. Am. Chem. Soc. 1982, 104, 3607-3617.
(18) Wei, N.; Murthy, N. N.; Chen, Q.; Zubieta, J.; Karlin, K. D. Inorg. Chem.
1994, 33, 1953-1965.
9
J. AM. CHEM. SOC. VOL. 129, NO. 42, 2007 12709

A R T I C L E S
Rodionov et al.
Figure 5. Simple coordination equilibria of bidentate ligands such as bipyridine and phenanthroline derivatives.
Figure 6. Potential coordination equilibria of tripodal ligands such as TBTA and (BimH)3 (S ) solvent; X ) halide, acetylide, hydroxide, triflate, or a
neutral donor that leaves Cu with a positive charge).
chemistry of effective Cu‚ligand complexes is diverse and that
the rate-limiting step of the catalytic cycle may easily change.
It may well be possible for a single copper center to activate
alkyne and azide in an accelerated reaction; indeed, we report
here one set of measurements that gives a first-order dependence
in [Cu‚TBTA]. However, we continue to find a bimolecular
dependence on [Cu‚ligand] under the most demanding condi-
tions of low catalyst concentration, and we believe that this is
the mode in which the most active systems operate. Such
binuclear kinetics are consistent with the need for structures
such as 14 (Figure 6), in which one Cu center assists the
cycloaddition of acetylide and azide fragments bound to the
other Cu center. The simultaneous σ,π-coordination of the
acetylide is well-precedented,9 and recent density functional
theory calculations show this type of arrangement to be
advantageous for the CuAAC process.3 But whatever the
particular arrangement of atoms in the key triazole-forming step,
the liberation of a coordination site on each Cu atom by
dissociation of two ligand arms (structure 13) would appear to
be necessary. We further suggest that the tripodal nature of such
ligands as TBTA and (BimR)3 may keep the metal coordination
chemistry “cleaner” by providing a high local concentration of
weakly binding arms, while at the same time allowing access
to open coordination sites. A similar situation was recently
engineered for rhodium hydroformylation catalysis in a remark-
ably elegant design of a tetraphosphine ligand by Zhang and
co-workers.19
It is noted in the accompanying article that benzothiazole-
based ligands were highly active at room temperature in the
presence of low substrate concentrations (1 to 2 mM) and
relatively high catalyst loadings (0.1 mM Cu, 10 mol %), but
not at 65 °C when substrate concentrations were increased
(200-400 mM) much more than the catalyst (2 mM Cu, 0.5-1
mol %). Conversely, tris(2-pyridylmethyl)amine ligand (Py)3
gave a catalyst showing strong rate acceleration under the latter
conditions, but rapid decomposition (after only a few turnovers)
in the former situation. It seems likely that both thermodynamic
(distribution of Cu complexes among active and inactive forms)
and kinetic (ligand substitution rates) factors can contribute to
such phenomena.
The sensitivity of the reaction to changes in the ligand that
may affect the copper coordination sphere was unexpectedly
(19) Yan, Y.; Zhang, X.; Zhang, X. J. Am. Chem. Soc. 2006, 128, 16058-
16061.
9
12710 J. AM. CHEM. SOC. VOL. 129, NO. 42, 2007

Ligand Acceleration of the CuAAC Reaction
A R T I C L E S
Figure 7. Ligand conformation that brings pendant alkylcarboxylate arms in the vicinity of the copper center.
illustrated by the comparison of Np(BimH)2 and NpCH2-
(BimH)2 ligands in Figure 2. Here, the addition of a single
methylene unit dramatically diminished the reactivity of the
catalyst. This is somewhat counterintuitive, since the neopentyl
substituent of Np(BimH)2 would be expected to provide more
steric hindrance than its neopentylmethyl homologue. Instead,
we suspect that the extra length of the latter puts the tert-butyl
group closer to the adjacent binding sites on the metal and there-
fore in a more likely position to disrupt an important interaction
(such as with organic azide) or coordination geometry.
Variations on the (BimR)3 theme showed that ligands bearing
N-alkylcarboxylic acid or ester groups with propyl, butyl, or
pentyl tethers were superior under most conditions. These ligand
arms have the capability to present polar functional groups close
to the metal center, as shown in Figure 7 for structures such as
acetylide complex 9. Pendant groups such as acids20-26 and
esters may participate in the coordination chemistry of CuI,
perhaps to the advantage of the CuAAC reaction.27 Alternatively,
the pendant group, by virtue of an active proton or hydrogen
bonding to solvent water, may aid in the protiolysis of the Cu‚
monotriazolyl intermediate (15),8 identified earlier as the
putative key species in the rapid formation of bistriazole.1 For
each possibility, inspection of molecular models suggests that
the pendant arms of butanoic, pentanoic, and hexanoic acid
chains, but not propanoic acid, are long enough to reach the
site of action, consistent with our observed structure-activity
data.
azide to Cu(I) may be more important than previously real-
ized: the “dog-legged” Hammett plot of Figure 3 suggests as
much. Still, the apparent insensitivity of the [1 + phenylacety-
lene] reaction to the presence of class II ligands will require
more study before we can propose even a tentative rationale
for it.
Finally, a few remarks on the presence of sodium ascorbate
are in order. We have chosen thus far to explore only those
systems containing aqueous ascorbate, both for experimental
convenience and because such conditions constitute by far the
most common way in which the reaction is performed by us
and others. We have ignored here the potential role of ascorbate
as anything except a reducing agent to maintain the bulk of the
Cu species in the +1 oxidation state. In general, we have found
changes in the concentration of ascorbate to make little
difference in absolute or relative rates obtained in our kinetics
measurements. However, we have not exhaustively compared
the reactivities of preformed Cu(I) complexes in the absence
of both oxygen and ascorbate to the same Cu‚ligand combina-
tions in the presence of ascorbate, and thus the potential
participation of ascorbate or dehydroascorbate in the coordina-
tion and catalytic chemistry of these systems is not ruled out.
Conclusions
The CuAAC reaction responds to changes in ligands, buffer
salts, and substrates in a complex manner. The presence of more
than one type of active Cu complex or mechanistic pathway is
indicated, since simple equilibria cannot explain many of the
observed rate Vs concentration profiles exhibited by this
system.28 From a practical perspective, it is remarkable that Cu-
(I) complexes are able to catalyze the reaction at high rates under
highly diverse conditions of solvent, temperature, nature of
catalyst precursor, and substrate structure. Perhaps it is the ability
of the system to bring different effective catalytic species to
bear that allows it to “adapt” to such challenges. Certainly the
flexibility of the reaction makes it both difficult and highly
worthwhile to study. The relationships of various aspects of
ligand structure and binding to the catalytic mechanism proposed
here suggest new modifications that can be made for improved-
performance. Such experiments are currently underway in our
laboratories.
Certain ligands were grouped into a category (“class II”) that
gave rise to no apparent rate acceleration and very little
production of monotriazole 2 from diazide 1; in other words,
that made little difference relative to the “ligand-free” reaction.
This is very curious, since several of these ligands (most notably
TBTA, 1,10-phenanthroline, and 2,2′-bipyridine) are well-
established as acceleratory ligands for the reaction of a variety
of monoazides with alkynes. We find it difficult to believe that
the 1,3-diazide 1 can compete effectively with all of the ligands
in this class for binding to Cu, but the coordination of organic
(20) Germanas, J. P.; Di Bilio, A. J.; Gray, H. B.; Richards, J. H. Biochemistry
1993, 32, 7698-7702.
(21) Faham, S.; Mizoguchi, T. J.; Adman, E. T.; Gray, H. B.; Richards, J. H.;
Rees, D. C. J. Biol. Inorg. Chem. 1997, 2, 464-469.
(22) Mizoguchi, T. J.; Di Bilio, A. J.; Gray, H. B.; Richards, J. H. J. Am. Chem.
Soc. 1992, 114, 10076-10078.
(23) Kataoka, K.; Yamaguchi, K.; Sakai, S.; Takagi, K.; Suzuki, S. Biochem.
Biophys. Res. Commun. 2003, 303, 519-524.
Experimental Section
(24) Romero, A.; Hoitink, C. W. G.; Nar, H.; Huber, R.; Messerschmidt, A.;
Canters, G. W. J. Mol. Biol. 1993, 229, 1007-1021.
The ligands, experimental procedures, and data analyses were the
same as those employed for the accompanying article. Complete details
can be found in the Supporting Information.
(25) Li, H.; Webb, S. P.; Ivanic, J.; Jensen, J. H. J. Am. Chem. Soc. 2004, 126,
8010-8019.
(26) Arnesano, F.; Banci, L.; Bertini, I.; Mangani, S.; Thompsett, A. R. Proc.
Natl. Acad. Sci. U.S.A. 2003, 100, 3814-3819.
(27) Multinuclear assemblies of Cu(II) complexes have been characterized for
the ligand that would be designated (BimH)2(BimC1A) in our parlance,
with the pendant carboxylate arm serving a coordinating role: Su, C.-Y.;
Yang, X.-P.; Kang, B.-S.; Mak, T. C. W. Angew. Chem., Int. Ed. 2001,
40, 1725-1728.
(28) A similar situation of interconverting catalysts giving rise to unusual rate
profiles was sorted out nicely in Nielsen, L. P. C.; Stevenson, C. P.;
Blackmond, D. G.; Jacobsen, E. N. J. Am. Chem. Soc. 2004, 126, 1360-
1362.
9
J. AM. CHEM. SOC. VOL. 129, NO. 42, 2007 12711

A R T I C L E S
Rodionov et al.
Acknowledgment. We are grateful to Professor K. Barry
Sharpless for helpful discussions. This work was supported by
The Skaggs Institute for Chemical Biology, Boehringer Ingel-
heim, and the Organic Division of the American Chemical
Society (graduate fellowship to V.R.).
Supporting Information Available: Experimental procedures
and kinetic data. This material is available free of charge via
the Internet at http://pubs.acs.org.
JA072679D
9
12712 J. AM. CHEM. SOC. VOL. 129, NO. 42, 2007