Benzimidazole and related ligands for Cu-catalyzed azide-alkyne cycloaddition

Article abstract of DOI:10.1021/ja072678l

Tris(2-benzimidazolylmethyl)amines have been found to be superior accelerating ligands for the copper(I)-catalyzed azide-alkyne cycloaddition reaction. Candidates bearing different benzimidazole N-substituents as well as benzothiazole and pyridyl ligand a

Full text of DOI:10.1021/ja072678l

Published on Web 10/03/2007
Benzimidazole and Related Ligands for Cu-Catalyzed
Azide-Alkyne Cycloaddition
Valentin O. Rodionov, Stanislav I. Presolski, Sean Gardinier, Yeon-Hee Lim, 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: Tris(2-benzimidazolylmethyl)amines have been found to be superior accelerating ligands for the
copper(I)-catalyzed azide-alkyne cycloaddition reaction. Candidates bearing different benzimidazole
N-substituents as well as benzothiazole and pyridyl ligand arms were evaluated by absolute rate
measurements under relatively dilute conditions by aliquot quenching kinetics and by relative rate
measurements under concentrated conditions by reaction calorimetry. Benzimidazole-based ligands with
pendant alkylcarboxylate arms proved to be advantageous in the latter case. The catalyst system was
shown to involve more than one active species, providing a complex response to changes in pH and buffer
salts and the persistence of high catalytic rate in the presence of high concentrations of coordinating ligands.
The water-soluble ligand (BimC
4 3
A) was found to be especially convenient for the rapid and high-yielding
synthesis of several functionalized triazoles with 0.01-0.5 mol % Cu.
Introduction
Unlike the well-established thermal Huisgen cycloaddition
2
4
Since its discovery in 2002,1,2 the copper(I)-catalyzed azide-
reaction, the Cu-accelerated version offers consistent 1,4-
stereoselectivity, is not limited to highly activated alkynes, and
proceeds efficiently even at micromolar concentrations of
reactants in aqueous media. Many investigators have reported
the use of amines as copper-binding ligands and/or protic bases
to aid the process, including 2,6-lutidine, triethylamine, N,N,N′-
trimethylenediamine, diisopropylethylamine, proline, Amberlyst
A21 amine resin, and other aliphatic amines (see Supporting
Information for a list of references). The most commonly used
accelerating ligand has been tris((1-benzyl-1H-1,2,3-triazol-4-
yl)methyl)amine (TBTA, 1) discovered by the Sharpless labora-
alkyne cycloaddition (CuAAC) reaction has received a great
3
4,5
deal of use in such diverse fields as bioconjugation in Vitro
6-9
10,11
12-16
and in ViVo, dendrimer synthesis
and polymer ligation,
combinatorial organic synthesis,17,18 and surface science.
19-23
(
(
(
1) Tornøe, C. W.; Christensen, C.; Meldal, M. J. Org. Chem. 2002, 67, 3057-
3
062.
2) Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless, K. B. Angew.
Chem., Int. Ed. 2002, 41, 2596-2599.
3) More than 600 articles have been published describing the use of the
CuAAC process since its discovery; the particular examples cited here are
therefore intended to be illustrative rather than comprehensive.
4) Gupta, S. S.; Kuzelka, J.; Singh, P.; Lewis, W. G.; Manchester, M.; Finn,
M. G. Bioconjugate Chem. 2005, 16, 1572-1579.
(
(
2
5
tory. Here we describe the preparation, comparative kinetic
evaluation, and practical use of a related family of ligands based
on the motif of a central tertiary amine surrounded by three
benzimidazole heterocycles. The accompanying article describes
kinetic measurements of mechanistic relevance and reports on
our current understanding of the reaction mechanism.
5) Dirks, A. J.; Van Berkel, S. S.; Hatzakis, N. S.; Opsteen, J. A.; van Delft,
F. L.; Cornelissen, J. J. L. M.; Rowan, A. E.; van Hest, J. C. M.; Rutjes,
F. P. J. T.; Nolte, R. J. M. Chem. Commun. 2005, 4172-4174.
6) Link, A. J.; Tirrell, D. A. J. Am. Chem. Soc. 2003, 125, 11164-11165.
7) Speers, A. E.; Cravatt, B. F. Chem. Biol. 2004, 11, 535-546.
8) Beatty, K. E.; Xie, F.; Wang, Q.; Tirrell, D. A. J. Am. Chem. Soc. 2005,
(
(
(
1
27, 14150-14151.
(9) Sieber, S. A.; Niessen, S.; Hoover, H. S.; Cravatt, B. F. Nat. Chem. Biol.
2
006, 2, 274-281.
(
10) Wu, P.; Malkoch, M.; Hunt, J.; Vestberg, R.; Kaltgrad, E.; Finn, M. G.;
Results and Discussion
Fokin, V. V.; Sharpless, K. B.; Hawker, C. J. Chem. Commun. 2005, 5775-
5
777.
Ligand Synthesis. The parent structure of the new ligand
class described here [tris(2-benzimidazolylmethyl)amine, des-
ignated (BimH)3 as explained in the caption to Figure 1] has
(
(
(
11) Malkoch, M.; Schleicher, K.; Drockenmuller, E.; Hawker, C. J.; Russell,
T. P.; Wu, P.; Fokin, V. V. Macromolecules 2005, 38, 3663-3678.
12) Gupta, S. S.; Raja, K. S.; Kaltgrad, E.; Strable, E.; Finn, M. G. Chem.
Commun. 2005, 4315-4317.
13) Parrish, B.; Breitenkamp, R. B.; Emrick, T. J. Am. Chem. Soc. 2005, 127,
7
404-7410.
(20) Meng, J.-C.; Averbuj, C.; Lewis, W. G.; Siuzdak, G.; Finn, M. G. Angew.
Chem., Int. Ed. 2004, 43, 1255-1260.
(14) Opsteen, J. A.; van Hest, J. C. M. Chem. Commun. 2005, 57-59.
(15) Binder, W. H.; Kluger, C. Macromolecules 2004, 37, 9321-9330.
(16) Sumerlin, B. S.; Tsarevsky, N. V.; Louche, G.; Lee, R. Y.; Matyjaszewski,
K. Macromolecules 2005, 38, 7540-7545.
(21) Collman, J. P.; Devaraj, N. K.; Chidsey, C. E. D. Langmuir 2004, 20, 1051-
1053.
(22) Lummerstorfer, T.; Hoffmann, H. J. Phys. Chem. B 2004, 108, 39663-
33966.
(
(
(
17) Fazio, F.; Bryan, M. C.; Blixt, O.; Paulson, J. C.; Wong, C.-H. J. Am.
Chem. Soc. 2002, 124, 14397-14402.
(23) Devaraj, N. K.; Miller, G. P.; Ebina, W.; Kakaradov, B.; Collman, J. P.;
Kool, E. T.; Chidsey, C. E. D. J. Am. Chem. Soc. 2005, 127, 8600-8601.
(24) Huisgen, R. In 1,3-Dipolar Cycloaddition Chemistry; Padwa, A., Ed.;
Wiley: New York, 1984; Vol. 1, pp 1-176.
18) Goess, B. C.; Hannoush, R. N.; Chan, L. K.; Kirchhausen, T.; Shair, M.
D. J. Am. Chem. Soc. 2006, 128, 5391-5403.
19) D ´ı az, 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,
(25) Chan, T. R.; Hilgraf, R.; Sharpless, K. B.; Fokin, V. V. Org. Lett. 2004, 6,
2853-2855.
4
392-4403.
1
2696 J. AM. CHEM. SOC. 2007, 129, 12696-12704
9
10.1021/ja072678l CCC: $37.00 © 2007 American Chemical Society

Benzimidazole Ligands for the CuAAC Reaction
A R T I C L E S
Figure 1. (A) Synthesis of tris(benzimidazole) and tris(benzothiazole) ligands. (B) Synthesis of mixed heterocycle and alkylated benzimidazole variants.
C) Structures of ligands, with changes made in the substitution pattern of each succeeding ligand highlighted in red. 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 .
long been known26 and was conveniently prepared by the
conventional condensation of 1,2-phenylenediamine with ni-
trilotriacetic acid. A variety of X-ray crystal structures involving
this type of ligand have been reported,27 including Cu(I) and
Cu(II) complexes of the N-propyl derivative (BimC3H)3.
However, no known structures of this ligand class incorporate
acetylide or π-alkynyl ligands, and thus their relevance to the
present situation is not clear. The analogues of (BimH) prepared
3
to explore various structure-activity relationships are shown
in Figure 1. Tris- and bis(benzothiazole)amines (Bth)3 and H-
2
8
(Bth)2 were obtained by the condensation of 1,2-aminothiophe-
nol with nitrilotriacetonitrile and iminodiacetonitrile, respec-
tively. Mixed benzimidazole/benzothiazole ligands (BimH)2(Bth)
and (BimH)(Bth)2 were prepared by reductive amination
reactions between secondary amines H(BimH)2 and H(Bth)2
and the corresponding heterocyclic aldehydes, as shown in
(
26) Thompson, L. K.; Ramaswamy, B. S.; Seymour, E. A. Can. J. Chem. 1977,
5
5, 878-888.
(
27) Blackman, A. G. Polyhedron 2004, 24, 1-39.
(
28) 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.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 42, 2007 12697

A R T I C L E S
Rodionov et al.
Figure 1. Among the tripodal structures, (Bth)3 and (BimC1H)3
have unusually low solubility in common organic solvents,
including DMSO.
catalyst development. The issue of copper speciation among
Cu(0), Cu(I), and Cu(II) oxidation states, accessible under inert
atmosphere by disproportionation of Cu(I), is also not directly
addressed here. The preservation of large amounts of Cu(I) under
our kinetic conditions (4:1 DMSO/water buffer, >20-fold excess
of ascorbate relative to Cu) is indicated by the lack of
characteristic copper complex color changes for appropriate
ligands (colorless Cu(I) to bright yellow Cu(II) in the presence
of (BimH)3).
Absolute rates were determined under relatively dilute
conditions (1 to 2 mM in substrates), which were required to
slow the reactions enough so that aliquots could be reliably taken
over a substantial fraction of the completion curve. Each aliquot
was immediately quenched and the amount of product formed
monitored by quantitative LC-MS against an internal standard.
In this way, the clean conversion to triazole was verified in
each case, information that is not available by following the
disappearance of azide or alkyne by infrared spectroscopy. Plots
of product concentrations Vs time fitted very well to second-
order kinetics over at least 70% of the reaction (Figure 2A). A
ligand/metal ratio of 2:1 was found to be optimal for TBTA
(BimR)3 derivatives were prepared to test the effects of
heterocyclic N-substitution on catalytic activity and water
solubility. Alkylation with methyl, ethylacetyl, and tert-buty-
lacetyl groups gave the neutral ligands (BimC1H)3, (BimC1E)3,
and (BimC1E′)3, respectively; ester hydrolysis provided the
water-soluble analogue (BimC1A)3. (A derivative bearing one
acetic acid chain and its supramolecular complex with Cu(II)
29
have been previously described. ) Longer-chain analogues were
made similarly, and (BimCnA)3 (n ) 3,4,5) ligands were
somewhat water soluble and detergent-like at higher concentra-
tions. Ligands (BimH/S)3 and (BimH/Me2)3 provided alterna-
tives that were much more hydrophilic and mildly more
hydrophobic and electron-rich, respectively.
Kinetic Comparisons. The relative effectiveness of Cu‚ligand
complexes was screened on the reaction between phenylacety-
lene and benzyl azide in 4:1 DMSO/aqueous buffer containing
sodium ascorbate as reducing agent, which served as the
standard reaction in our previous mechanistic work.30 We refer
to ligand/Cu ratios to describe the amounts of each component
added and designate candidate mixtures according to those ratios
for convenience. However, it must be emphasized that the active
catalysts are not necessarily composed of those same ratios of
ligand and metal. The equilibria that govern the composition
of these catalysts are discussed below, and their study will be
the subject of a separate publication.
(
1) by this method (Supporting Information), which matches
32
the results from previous work on other ligands. For this
reason, comparisons were made between second-order specific
activities (rate constants determined at a standard Cu concentra-
tion of 0.1 mM) for catalysts composed of 2:1 mixtures of
ligands and Cu(I), in the presence of excess ascorbate relative
to Cu.
In general, benzimidazole-based ligands provided for reactions
faster than those of the parent tris(triazolyl) structure 1 and in
the same range as the previously identified sulfonated batho-
phenanthroline 2 (data not shown). The magnitude of ligand-
accelerated catalysis (ratio of the observed rates in the presence
Vs the absence of added ligand) was substantial (more than an
The utility of the benzimidazole motif was immediately
apparent in reactions at 50 mM in azide and alkyne, 5 mM Cu,
and 10 mM in sodium ascorbate. In the absence of ligand, the
reaction reached only approximately 10% completion in 1 h at
room temperature, whereas the use of 1 equiv of 1 or (BimH)3
with respect to metal gave 75 and 100% completion, respec-
tively, in the same time, providing strong evidence of ligand-
accelerated catalysis. Our previous work demonstrated that
CuAAC reactions in the absence of accelerating ligands
exhibited a second-order rate dependence on copper concentra-
tion. Accordingly, a “dimerized” version of the (BimH)3 ligand,
3
2
order of magnitude), but not as great as the factors in excess of
3
1
0 observed previously for reactions of Cu‚2 under the highly
dilute conditions (micromolar concentrations of substrates and
3
2
catalysts) required for bioconjugation. As discussed in the
accompanying article, we believe this reflects the relative
binding affinities of 2 Vs the tris(heterocyclic methyl)amines
for Cu(I). It is also apparent that reaction rate is sensitive to
both the nature of the heterocycle and to substitution on the
benzimidazole ring. The replacement of benzimidazoles with
benzothiazoles in the N(CH2heterocycle) format gave no ap-
preciable loss in catalyst efficiency, as long as at least one
benzimidazole side arm was present. (Bth)2(BimH) was found
to be the best ligand in the series, with rates following the order
[
(BimH)2C2(BimH)2], was prepared in the hope that its bi-
31
nuclear Cu(I) complex would make a better catalyst. The
ligand proved to be a very potent inhibitor in early testing and
was not further examined, but other binucleating systems remain
to be explored.
The obvious accelerating abilities the parent tris(benzimida-
zole) ligand prompted us to measure more accurately the
absolute and relative activities of candidate catalysts in two
different concentration regimes. We chose to focus on catalyst
activity, as indicated by observed pseudo-first- or second-order
rate constants (specific activities) in the presence of large
amounts of ascorbate reducing agent, thereby ignoring the issue
of resistance of Cu(I) catalysts to oxidative deactivation in air.
While the latter is an important parameter for the practical
applicability of candidate systems, we believe that these issues
are better tackled separately, at least in the early stages of
(Bth)3 , (Bth)(BimH)2 < (BimH)3 , (Bth)2(BimH). N-
Substitution of the tris(benzimidazole) ligand was extensively
investigated, with small pendant alkyl and ester groups
[(BimC1E)3, (BimC1E′)3, (BimC1H)3] performing as well as
the parent structure (BimH)3 (data for the latter two ligands
not shown). The installation of pendant carboxylic acid groups
gave rise to noticeably better catalyst performance in the case
of longer chain lengths [(BimC4A)3, (BimC4A/Me2)3, (BimH)-
(BimC5A)2, (BimC3A)3, (BimC5A)3]. The ester (BimC4E)3 was
just as active as its hydrolyzed (carboxylate) analogue. In
contrast, ligand (BimC1A)3, which features carboxylic acid
(
(
(
29) Su, C.-Y.; Yang, X.-P.; Kang, B.-S.; Mak, T. C. W. Angew. Chem., Int.
Ed. 2001, 40, 1725-1728.
30) Rodionov, V. O.; Fokin, V. V.; Finn, M. G. Angew. Chem., Int. Ed. 2005,
4
4, 2210-2215.
31) Hendriks, H. M. J.; Birker, P. J. M. W. L.; van Rijn, J.; Verschoor, G. C.;
(32) Lewis, W. G.; Magallon, F. G.; Fokin, V. V.; Finn, M. G. J. Am. Chem.
Soc. 2004, 126, 9152-9153.
Reedijk, J. J. Am. Chem. Soc. 1982, 104, 3607-3617.
12698 J. AM. CHEM. SOC.
9
VOL. 129, NO. 42, 2007

Benzimidazole Ligands for the CuAAC Reaction
A R T I C L E S
Figure 2. (A) Example of second-order rate treatment for the reaction of benzyl azide with phenylacetylene involving ligand (BimH)3. The circles mark
observed data points; the curve is calculated from the integrated rate equation with the indicated specific activity kobs. Experimental errors in rate constants
based on repeat measurements are within 15% of the indicated values in the vast majority of cases. (B) Relative magnitudes of second-order specific
activities at pH 8, with the rate in the absence of added ligand set to 1.0. Single asterisks (/) mark catalysts that exhibit a burst of faster activity at the
beginning of the reaction (2-4% reaction, representing less than one catalyst turnover) and then settle down to the slower rate indicated. The double asterisk
(//) marks a ligand for which the initial rate is reported; the reaction slows after approximately 30% completion. (C) Observed second-order specific
activities for a selection of ligands in 4:1 DMSO/buffer, as a function of pH of the HEPES buffer. The ratio of rates (pH 8/pH 7) for each ligand is shown
in italics. (D) Observed second-order specific activities for a selection of ligands at pH 8.0 in 4:1 DMSO/aqueous, where “aqueous” is either HEPES buffer
or water. In the latter case, sodium ascorbate provides weak buffering capacity; the pH of the mixture was found to be the same at the beginning and end
of the reaction. The ratio of rates (with HEPES/without HEPES) for each ligand is shown in italics. In all cases, [BnN3] ) 1 mM; [PhCCH] ) 2 mM;
[CuSO4] ) 0.1 mM; [ligand] ) 2 mM; temp ) 24 ( 2 °C; solvent ) 4:1 DMSO/aqueous.
groups separated from the benzimidazole core by just one carbon
atom, produced one of the poorest catalysts in the series, much
less effective than its ester (BimC1E)3.
significantly in the presence of Tris buffer over a standard range
of concentrations, with the inhibitory effect being significantly
more pronounced for 1 (data not shown). Another popular buffer
system is based on 4-(2-hydroxyethyl)-1-piperazineethane-
sulfonic acid (HEPES), which is known to have moderate
Some indications of the response of the process to changes
in pH and to the presence of added buffer salts are provided by
Figure 2C,D. In absence of catalytic ligands, preparative CuAAC
reactions at the bench are remarkably tolerant of large changes
in pH: only highly acidic conditions (pH 2 or lower) can halt
the reaction, possibly by affecting Cu speciation or inhibiting
the formation of Cu‚acetylides (data not shown). In general,
more basic conditions promote the reaction, and this trend is
also evident in the absolute rates measured for several added
ligands, as shown in Figure 2C. However, different Cu‚ligand
catalysts respond differently. For example, the mixed benzimi-
dazole/benzothiazole system (Bth)2(BimH) showed very little
change in rate when the pH of the aqueous component of the
solvent mixture was raised from 7 to 8, whereas the parent
tripodal ligands (Bth)3 and (BimH)3 were more sensitive.
Concern for pH effects prompted us to test added buffer salts,
even though excess ascorbate functions as a low-capacity buffer
under our standard conditions. The nature of the buffer was
also found to have an impact on the CuAAC reaction. For
example, while we and others have successfully used pH 8 Tris-
HCl buffer for bioconjugation applications, we found in the
course of our kinetic studies that reactions catalyzed by Cu‚1
and Cu‚(BimH)3 mixtures in 4:1 DMSO/buffer are slowed
3
3,34
34
affinity for Cu(II)
and can slowly reduce Cu(II) to Cu(I).
We found HEPES buffer to have little effect on the catalytic
rate of a wide variety of Cu‚ligand complexes at moderate
concentrations (<100 mM, data not shown). We therefore
recommend the use of HEPES (at 100 mM or less) whenever
CuAAC reactions need to be performed at controlled moderate
pH values in an aqueous environment. However, the use of a
higher buffer concentration (180 mM) gave rise to the varying
results shown in Figure 2D. Three catalysts were found to give
faster turnover in the presence of HEPES, with the effect being
especially dramatic for Cu‚(Bth)2(BimH). Other catalysts bear-
ing pendant carboxylate arms were found to be inhibited by
180 mM HEPES.
Figure 2B also shows a small number of ligands that are either
inhibitory or inconsequential relative to the rate of the “ligand-
free” case. Thus, tris(pyridylmethyl)amine (Py)3, secondary
amine H(BimH)2, and a ligand having heterocycles flanking a
central pyridine motif [Py(BimC4A)2]] each give rates slower
than that observed for the “ligand-free” case. Pybox derivative
(
33) Sokolowska, M.; Bal, W. J. Inorg. Biochem. 2005, 99, 1653-1660.
(34) Hegetschweiler, K.; Saltman, P. Inorg. Chem. 1986, 25, 107-109.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 42, 2007 12699

A R T I C L E S
Rodionov et al.
3
, previously used for kinetic resolution of racemic diazides,35
is also not acceleratory. We believe that these systems do not
allow access to the correct coordination geometry to promote
the reaction. The spectrum of results shown in Figure 2 and
the Supporting Information suggest that the response of various
Cu‚ligand catalysts to changes in pH and added coordinating
ions is complex, likely to involve ligand-dependent changes in
speciation, structure, and, perhaps, rate-limiting step of the
catalytic cycle. From a practical perspective, (BimC4A) and
(BH/S)3 are attractive because they provide strong rate accelera-
tion through a wide pH range and are soluble in water to at
least 50 mM for the potassium salts, thus allowing for easy
workup (see below).
Reaction Calorimetry. To test the relative merits of candi-
date ligands under practical conditions of organic synthesis
(
involving high concentrations of azide and alkyne reagents and
36-39
low loadings of catalyst), we turned to reaction calorimetry.
After correction for instrumental heat transfer parameters,
thermograms provide a direct real-time measure of reaction rate
(power output); integration gives the overall time Vs completion
curve. As discussed fully in the Supporting Information, an
instrumental limitation in sensitivity foiled attempts to obtain
detailed mechanistic information3 on reactions that were less
than optimal, but the peak reaction exotherm proved to be a
reproducible and convenient measurement of relative catalytic
activity. All such measurements were made in the same 4:1
DMSO/aqueous solvent as before, but at 65 °C rather than room
temperature to keep the product triazole in solution at the higher
concentrations used. The technique also allowed us to ascertain
at a glance if candidate catalysts were “well-behaved”, producing
a rapid burst of heat immediately upon initiation of the reaction
followed by smooth exponential decay, or if they exhibited more
complex behavior such as induction times or bimodal power
Vs time curves.
Most systems gave a clean profile characteristic of a second-
order reaction involving a single catalyst species, as shown in
the example of Figure 3A. TBTA (1) was an important
exception, showing a markedly more gradual approach to a peak
rate followed by a rapid drop to give the “sawtooth” thermogram
shown in Figure 3B. This reaction profile made direct calori-
metric comparison between 1 and other ligands impossible;
however, the peak power and integrated area showed 1 to give
a sluggish catalyst compared to many of the others tested here
8,39
Figure 3. (A) Calorimetry data for reactions of varying concentrations of
phenylacetylene with 0.45 M benzyl azide in the presence of CuSO4 (2
mM), (BimH)3 (4 mM), and sodium ascorbate (25 mM), in 4:1 DMSO/
water (pH 8), at 65 ( 2 °C. The azide was added last to initiate the reaction
at the 300-s mark. The maximum power produced by the reaction,
represented by the peak of the signal produced immediately after azide
addition, was used as a relative measure of catalyst activity. Insets: (left)
expanded view; (right) corresponding data for varying concentrations of
BnN3 in the presence of 0.45 M PhCCH. (B) Thermogram for the same
reaction as in part A (0.45 M alkyne, 0.25 M azide), performed with 1
instead of (BimH)3.
was inactive; the other performed approximately as well as 1,
showing the same type of unusual rate Vs time profile (Sup-
porting Information).
Approximately 100 ligands were tested by calorimetry at a
ligand/Cu ratio of 2:1, and a subset of these were also evaluated
at a 1:1 ratio. The results from the best systems are shown in
Figure 4; a summary of the remaining measurements is given
in the Supporting Information. All the values in Figure 4
represent substantial levels of ligand-accelerated catalysis, since
the reaction without added ligand does not release enough heat
to register on the calorimeter at all (although it does go to
completion within 12 h at 65 °C). All of the ligands in Figure
(although still far more active than without ligand). Its unique
thermogram suggests that changes in the Cu‚1 catalyst composi-
tion occur as the reaction proceeds, since its profile does not
correspond to typical kinetic expressions involving a single
catalyst species. Two N-heterocyclic carbene complexes of Cu-
4
show single-component “well-behaved” kinetic characteristics
(Figure 3A). The ligand/metal ratio made little difference in
the overall trends.
(
I) were also tested, because a recent article describing the use
The results from calorimetry screening are largely consistent
with the results obtained by analysis of quenched aliquots by
LC-MS (Figure 2). Both methods show tris(benzimidazole)
derivatives to be excellent catalytic ligands, and those bearing
longer alkylcarboxylate groups [(BimC4A)3, (BimH)(BimC5A)2,
40
of this class of ligands has appeared. One of these ligands
(
35) Meng, J.-C.; Fokin, V. V.; Finn, M. G. Tetrahedron Lett. 2005, 46, 4543-
546.
4
(
36) LeBlond, C.; Wang, J.; Larsen, R. D.; Orella, C. J.; Forman, A. L.; Landau,
R. N.; Laquidara, J.; Sowa, J. R.; Blackmond, D. G.; Sun, Y. K.
Thermochim. Acta 1996, 289, 189-207.
(
(
BimC3A)3, (BimC5A)3] were the best in the class. (The ester
BimC4E)3 was not tested because ester hydrolysis can occur
(
(
(
37) Blackmond, D. G.; Rosner, T.; Pfaltz, A. Org. Process Res. DeV. 1999, 3,
2
75-280.
38) Singh, U. K.; Strieter, E. R.; Blackmond, D. G.; Buchwald, S. L. J. Am.
Chem. Soc. 2002, 124, 14104-14114.
39) Mathew, J. S.; Klussmann, M.; Iwamura, H.; Valera, F.; Futran, A.;
Emanuelsson, E. A. C.; Blackmond, D. G. J. Org. Chem. 2006, 71, 4711-
under the reaction conditions.) The important exceptions are
benzothiazole-containing ligands [(Bth)3, (Bth)2(BimH), (Bth)-
(BimH)2] and the tris(pyridylmethyl)amine ligand (Py)3. The
former showed significant activity in the aliquot quenching
screen (low reactant concentrations, 10% Cu) yet were largely
4
722.
(40) D ´ı ez-Gonz a´ lez, S.; Correa, A.; Cavallo, L.; Nolan, S. P. Chem.-Eur. J.
2
006, 12, 7558-7564.
12700 J. AM. CHEM. SOC.
9
VOL. 129, NO. 42, 2007

Benzimidazole Ligands for the CuAAC Reaction
A R T I C L E S
Figure 4. Dynamically corrected peak power values for the best ligands identified by reaction calorimetry under the conditions described in Figure 3 (0.45
M PhCCH, 0.25 M BnN3), at the indicated ligand/Cu ratios. The experimental error for these measurements was established as (160 mW from more than
five independent trials for a representative selection of ligands spanning the activity range.
Figure 5. Dependence of the peak reaction power (peak rate) on ligand/Cu ratio for a subset of the best ligands identified in the initial screen. Conditions:
benzyl azide (0.45 M); phenylacetylene (0.25 M); CuSO4 (2 mM); ligand (variable); sodium ascorbate (25 mM); 4:1 DMSO/water; 65 ( 2 °C.
inactive under calorimetry conditions (high reactant concentra-
tions, 1% Cu). The opposite trend was observed for (Py)3, which
was one of the best ligands in calorimetry screening while
performing poorly in aliquot quenching kinetics. We attribute
these differences to changes in copper speciation that are likely
to accompany changes in overall reaction concentration and
catalyst loading.
From a practical perspective, the tris(benzimidazole) family
emerged from both types of rate measurements as an effective
platform for ligand development. Bipyridine and phenanthroline
are accelerating additives that we recommend testing as readily
available options when the reaction must be optimized for a
particular application. Histidine provided for the fastest CuAAC
reaction among all the amino acids, as has been recently noted
Figure 6. Dependence of peak reaction power (peak rate) on added 1-n-
octyl-4-phenyl-1,2,3-triazole for the reaction of n-octylazide (350 mM) +
phenylacetylene (450 mM), in the presence of CuSO4 (2 mM); (BimH)3 (4
mM); sodium ascorbate (25 mM); 4:1 DMSO/water; 65 ( 2 °C. The
thermogram shape for each reaction was bimodal as shown in Figure 7A;
this system was chosen to allow the addition of triazole without precipitation.
4
1
elsewhere, defining the lower boundary of our “effective”
ligand category. While not directly comparable because of the
differences in rate profile described above, 1 ranks in the lower
half of this group.
We also explored a wider range of ligand/Cu ratios for some
of the best ligands under concentrated conditions by calorimetry.
These results, summarized in Figure 5, show that the ligands
may be divided into two categories: those that perform
optimally at a 1:1 ratio [(BimH)3, (BimC1A)3, (BH/S)3], and
those that reach either a peak or a plateau at a 2:1 ratio
[
(BimC1H)3, (BimC1E)3, (BimC1E′)3, (BimC4A)3, (BH/Me2)3,
(
41) Tanaka, K.; Kageyama, C.; Fukase, K. Tetrahedron Lett. 2007, 48, 6475-
6
479.
(Py)3, Py(Tet)2]. In most of the latter cases, the reaction was
J. AM. CHEM. SOC.
9
VOL. 129, NO. 42, 2007 12701

A R T I C L E S
Rodionov et al.
Figure 7. Corrected calorimetry data for reactions of the indicated azides and additives. (A-C) Azide (250 mM); phenylacetylene (450 mM); CuSO4 (2
mM); (BimH)3 (4 mM); sodium ascorbate (25 mM); 4:1 DMSO/water; 65 ( 2 °C. (D) In DMSO with [Cu(MeCN)4]OTf (2 mM) and no added reducing
agent. The asterisk in plot A marks an exothermic crystallization of product triazole.
Table 1. Tests of Reduced Catalyst Loading for the Reaction of
not dramatically slowed by the addition of excess ligand (4 or
Phenylacetylene (0.75 M) with Benzyl Azide (0.75 M), 3:1 MeOH/
7
equiv). Even when a pronounced peak in activity was observed
Water, Sodium Ascorbate (30 mM), 24 °Ca
at an equimolar ligand/Cu ratio [(BimH)3, (BimC1A)3], the
reaction proceeded at a significant rate in the presence of excess
ligand. This type of behavior is not normally exhibited by
systems having a single active catalyst, the concentration of
which is governed by a simple set of equilibria such as that
shown in eq 1.
mol %
Cu
mol %
ligand
time
(h)
yield
(%)b
entry
ligand
Cu(I) source
1
2
3
4
5
6
7
8
9
1
1
CuSO4 ascorbate
CuSO4 ascorbate
CuSO4 ascorbate
CuSO4 ascorbate
[Cu(MeCN)4]OTf
CuI
0.1
0.5
1
0.5
1
1
0.1
0.5
0.05
0.01
0.2
0.5
1
0.5
1
24
4.5
5
27
47
94
92
93
6
(BimH)3
(BimH)3
(BimH)3
(BimH)3
(BimH)3
(BimC4A)3 CuSO4 ascorbate
(BimC4A)3 CuSO4 ascorbate
(BimC4A)3 CuSO4 ascorbate
(BimC4A)3 CuSO4 ascorbate
5
24
24
24
1
K1
K2
CuSO4 ascorbate
0.1
0.5
0.05
0.01
73
+
L
+L
MnLm y z M L
y z M L
(1)
0.2 100
5
24
n
m+1
n
m+2
-
L
-L
98
95
35
29
inactiVe
actiVe
inactiVe
1
1
1
0
1
2
0.005 0.005 72
0.5 72
In situations such as eq 1, ligand-accelerated reactivity derives
from the catalytic power of a complex containing a greater
ligand/metal ratio than an inactive (or less active) precursor; in
the simplest case, the first ligand to bind creates the catalyst of
interest (m ) 0). The association of an additional equivalent of
ligand diminishes the rate by blocking access of a substrate
molecule to an open coordination site on the metal or by
diverting the system to a less active structure. A plot of ligand/
metal ratio Vs catalysis rate would therefore appear as a single
peak, approaching a value of zero in rate as the concentration
of ligand is increased. In contrast, the benzimidazole and related
CuAAC systems show persistent and considerable catalytic
activity even at high ligand/Cu. This is likely due to one or
both of two factors: (a) K2 , K1 and (b) at least two active
ligand-containing catalysts or mechanistic pathways exist: one
that depends on ligand/Cu, and another that is largely indepen-
dent of that ratio. (For all of the catalysts shown, the reaction
rate in the absence of ligand was too low to be measured on
our calorimeter. The accelerating ligands therefore either must
be components of the active catalytic species or must enhance
CuIc
iPr2EtN
5
a
Reactions producing >90% yield were completed in substantially
shorter time than the listed values. Analogous reactions using CuSO4 or CuI
b
but without added ligand gave <10% yield after 24 h. Isolated yield of
analytically pure triazole product. c Reaction in absolute MeOH; the use of
cuprous iodide and excess additive reflects the customary practice reported
in the literature.
the concentration of active copper centers.) That a complex set
of equilibria is involved is further indicated by the fact that the
addition of even weakly coordinating additives such as 2,6-
lutidine or 2-amino-2-(hydroxymethyl)propane-1,3-diol (Tris)
to the solvent system for catalyst Cu‚[(BimH)3]n removes the
relatively sharp peak in the plot of rate Vs ligand/Cu, making
the relationship look more like that of Cu‚(BimC4A)3 (data not
shown).
The persistence of high reaction rates in the presence of excess
ligand suggests that these catalysts should also remain active
in the presence of high concentrations of product triazoles, which
can be expected to have some affinity for Cu(I). This has been
demonstrated by the successful use of low catalysts loadings
12702 J. AM. CHEM. SOC.
9
VOL. 129, NO. 42, 2007

Benzimidazole Ligands for the CuAAC Reaction
A R T I C L E S
Figure 8. Triazoles prepared by reaction of equimolar amounts of the appropriate azide and alkyne (each 10 wt %, 0.13-0.35 M, depending on molecular
weight) in the presence of CuSO4 (0.1 mol %); (BimC4A)3 (0.2 mol %); and sodium ascorbate (4 mol %) in 95:5 DMF/water at room temperature. Isolated
yields of purified products are indicated in parentheses; compound 10 was obtained in relatively low yield because of competitive reaction of the starting
material with water.
(
discussed below) and by calorimetric measurements sum-
methyl)amine ligands such as 1 and (BimH)3 (entry 12). The
successful application of (BimC4A)3 to the synthesis of other
triazoles, containing steroid, carbohydrate, boronic acid, amine,
alcohol, and electrophilic chlorotriazine moieties, is shown in
Figure 8. The selection of substrates is meant to illustrate the
reliability and convenience of the Cu‚[(BimC4A)3] system for
synthetic purposes. In addition, the water solubility of (BimC4A)3
makes the preparation of relatively hydrophobic triazoles more
convenient, since a simple aqueous wash removes both the
ligand and metal, leaving the product in pure form.
marized in Figure 6. Peak catalytic activity was found to be
unchanged in the presence of triazole in an amount up to 400
times the concentration of catalyst.
Effects of Aggregation of Long-Chain Aliphatic Azides.
Under the typical reaction conditions in 20% aqueous DMSO,
the thermograms of reactions involving long-chain aliphatic
azides (accelerated by ligand (BimH)3) exhibited a distinctive
two-stage reaction profile as shown in Figure 7A. The reactions
of those azides became slower, and the bimodal nature of
thermograms more distinct, with elongation of the aliphatic
chain. The addition of n-tetradecane to the well-behaved
n-hexylazide also slowed the reaction dramatically, even though
the overall concentration of azide was not changed (Figure 7B).
When the reaction was performed in the presence of the non-
ionic detergent Triton-X100 (27 mM), the thermogram reverted
to the normal single-peak appearance (Figure 7C). Single-peak
behavior was also observed in pure DMSO, although the
thermogram peaks were not as sharp as those in the standard
DMSO/water solvent system (Figure 7D). Last, the use of
amphiphilic ligand (BimC4A)3 in place of (BimH)3 largely
eliminated the bimodal behavior of aliphatic azides. All of these
observations support the hypothesis that long-chain aliphatic
azides partition between solvated single molecules and detergent-
sensitive aggregates in the aqueous DMSO medium, and in the
latter case are partially sequestered from the catalyst and alkyne.
Synthetic Tests. The ligands (BimH)3 and (BimC4A)3,
identified in the above studies as being superior under conditions
of high substrate concentration, were tested along with 1 for
their ability to support reactions at low catalyst loading (Table
Conclusions
Tris(2-benzimidazolylmethyl)amines have been found to be
the most promising family of accelerating ligands for the Cu-
catalyzed azide-alkyne cycloaddition reaction from among
more than 100 mono-, bi-, and polydentate candidates. The
benzimidazole compounds are easy to prepare and compare
favorably to tris(triazolylmethyl)amines and other ligands, such
as diisopropylethylamine and sulfonated bathophenanthroline
that have achieved wide use. Under both preparative (high
concentration, low catalyst loading) and dilute (lower substrate
concentration, higher catalyst loading) conditions, tripodal
benzimidazole derivatives give substantial improvements in rate
and yields, with convenient workup to remove residual Cu and
4
2
ligand. The relative effectiveness of ligands with different
substitution patterns changes when examined under different
reaction conditions, pointing to interesting mechanistic features
that are discussed in the accompanying article.
Experimental Section
1
). The results mirror the findings by calorimetry, and yields
followed the trend Cu‚[(BimC4A)3] > Cu‚[(BimH)3] . Cu‚
1] (entries 2, 4, and 8). A maximum value of approximately
Experimental procedures used for kinetics measurements and reaction
calorimetry are fully described in the Supporting Information.
4
3
Ligand Synthesis. Tris-benzimidazole ligand (BimH)
3
and ana-
[
44
4
2 2 2 2
logues H(BimH) and (BimH) C (BimH)
were obtained by con-
10 turnovers per metal center was achieved (0.01 mol %
densation of 1,2-phenylenediamine with the corresponding tri-, di-, and
tetracarboxylic acid in boiling ethylene glycol. A similar condensation
reaction between 1,2-aminothiophenol and nitrilotriacetonitrile or
catalyst, entry 10), allowing conveniently rapid reactions at 0.05
mol % or greater. The use of 0.001 mol % catalyst provided a
3
0% yield after 3 weeks in an inert atmosphere glovebox, but
iminodiacetonitrile was used to synthesize benzothiazole ligands (Bth)
and (H)(Bth) . The crude products usually crystallized when water was
added to the cooled reaction mixture (it is important not to add water
3
no product on the benchtop, suggesting that oxidation of the
Cu‚ascorbate system was competitive with the cycloaddition
process at that point. The preformed soluble Cu(I) acetonitrile
complex made a competent catalyst in the absence of ascorbate
2
(
42) Not discussed here are related series of tris(oxazolinylmethyl)amine ligands,
which we have found in preliminary studies to give catalysts of comparable
efficiency to tris(triazolylmethyl)amines. The attempted use of these systems
in chiral kinetic resolution and desymmetrization reactions will be described
elsewhere.
(entry 5), but cuprous iodide did not (entry 6), at least in a mixed
MeOH/water solvent system. The use of diisopropylethylamine
as an additive with CuI has gained some popularity (Supporting
Information), but we have found such reactions to be much
slower than those performed in the presence of tris(heterocyclic
(43) Thompson, L. K.; Ramaswamy, B. S.; Seymour, E. A. Can. J. Chem. 1977,
55, 878-888.
(
44) Birker, P. J. M. W. L.; Hendriks, H. M. J.; Reedijk, J.; Verschoor, G. C.
Inorg. Chem. 1981, 20, 2408-2414.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 42, 2007 12703

A R T I C L E S
Rodionov et al.
to the ethylene glycol solution while it is still hot). For the purposes of
this study, the ligands were subsequently recrystallized from absolute
ethanol and were found to be pure by elemental analysis. The procedure
scales up without difficulty. The preparation of 1 from tripropargylamine
and benzyl azide under the influence of Cu(I) sometimes suffers from
incomplete reaction, the presence of colored impurities, and/or the
presence of trapped copper ions in the solid product.
the ligand, methanol (or another appropriate solvent), azide, and alkyne
in that order. A solution of CuSO in water was added, the flask was
4
capped, and the mixture was stirred vigorously or sonicated for 5-10
min until all of the ascorbate was dissolved. The resulting homogeneous
solution was left standing at room temperature and periodically
monitored by TLC. When reagent conversion was complete, several
volumes of deionized water were added to the reaction and the
precipitated triazole product was recovered by filtration. For the most
part, these products were found to be >98% pure by NMR; when
necessary, purification to reach this level was performed by flash
chromatography, and yields are reported for the final purified products.
The mixed ligands (Bth)(BimH)
by reductive amination of aldehydes 4 and 5 (Figure 1) with amines
Bth) and (H)(BimH) , respectively. N-Alkylated derivatives (BR)
were obtained by deprotonation of (BimH) with t-BuOK and reaction
2 2
and (Bth) (BimH) were obtained
45
46
(
2
2
3
3
with the corresponding alkylating agent. Poor results were obtained
when NaH (either dry or oil suspension) was used as the base. To
Acknowledgment. We are grateful to Professors Valery V.
Fokin and K. Barry Sharpless for helpful discussions, Dr. Timo
Weide for the N-heterocyclic carbene complexes mentioned in
the Supporting Information, and Ms. Hena Din for experimental
assistance. This work was supported by The Skaggs Institute
for Chemical Biology, the TSRI Summer Internship Program
(S.G.), Boehringer Ingelheim, and the Organic Division of the
American Chemical Society (graduate fellowship to V.R.).
produce the free carboxylate ligand (BimC
corresponding ester was reacted with exactly 3 equiv of KOH in boiling
aqueous ethanol. Tris(sulfonic acid) (BH/S) was obtained by a
4 3
A) and its analogues, the
3
47
published procedure. Most ligands with free benzimidazole NH groups
were purified by flash chromatography on a short Florisil column,
2 2
eluting with 0.5% MeOH in CH Cl . Ordinary silica gel is not well
suited to this separation because of the strong tendency of these
compounds to “streak”, resulting in poor recovery of the desired
products.
Typical Procedure for Preparative Scale CuAAC Reactions. The
reaction flask was charged with solid sodium ascorbate, followed by
Supporting Information Available: Experimental procedures,
detailed kinetic data, and characterization data for ligands. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(45) Zakrzewski, A.; Janota, H. Pestycydy (Warsaw) 2004, 31-38.
(46) Vetelino, M. G.; Coe, J. W. Tetrahedron Lett. 1994, 35, 219-222.
(47) Ichikawa, K.; Nakata, K.; Ibrahim, M. M. Chem. Lett. 2000, 796-797.
JA072678L
12704 J. AM. CHEM. SOC.
9
VOL. 129, NO. 42, 2007