Autocatalytic Cycles in a Copper-Catalyzed Azide-Alkyne Cycloaddition Reaction

Article abstract of DOI:10.1021/jacs.8b05048

This work describes the autocatalytic copper-catalyzed azide-alkyne cycloaddition (CuAAC) reaction between tripropargylamine and 2-azidoethanol in the presence of Cu(II) salts. The product of this reaction, tris-(hydroxyethyltriazolylmethyl)amine (N(C

Full text of DOI:10.1021/jacs.8b05048

Subscriber access provided by READING UNIV
Article
Autocatalytic Cycles in a Copper-Catalyzed
Azide-Alkyne Cycloaddition Reaction
Sergey N Semenov, Lee Belding, Brian J. Cafferty, Maral P. S. Mousavi, Anastasiia
M. Finogenova, Ricardo S. Cruz, Ekaterina V. Skorb, and George M. Whitesides
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b05048 • Publication Date (Web): 23 Jul 2018
Downloaded from http://pubs.acs.org on July 23, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the
course of their duties.

Page 1 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Autocatalytic Cycles in a Copper-Catalyzed Azide-Alkyne Cycloaddition Reaction
a‡
a
a
a
Sergey N. Semenov, Lee Belding, Brian J. Cafferty, Maral P.S. Mousavi, Anastasiia M.
Finogenova, Ricardo S. Cruz, Ekaterina V. Skorb, and George M. Whitesidesa,b,c*
a
a
a†
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Department of Chemistry and Chemical Biology, Harvard University
1
2 Oxford Street, MA 02138
b
Kavli Institute for Bionano Inspired Science and Technology, School of Engineering and
Applied Sciences, Harvard University
2
9 Oxford Street, MA 02138
Wyss Institute for Biologically Inspired Engineering, Harvard University
0 Oxford Street, MA 02138
c
6
*
Author to whom correspondence should be addressed.
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Abstract
This work describes the autocatalytic copper-catalyzed azide-alkyne cycloaddition (CuAAC)
reaction between tripropargylamine and 2-azidoethanol, in the presence of Cu(II) salts. The
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
product of this reaction, tris-(hydroxyethyltriazolylmethyl)amine (N(C
3
N
3
) ), accelerates the
3
cycloaddition reaction (and thus its own production) by two mechanisms: i) by coordinating
Cu(II) and promoting its reduction to Cu(I), and ii) by enhancing the catalytic reactivity of
Cu(I) in the cycloaddition step. Because of the cooperation of these two processes, a rate
enhancement of > 400x is observed over the course of the reaction. The kinetic profile of the
autocatalysis can be controlled by using different azides and alkynes, or ligands (e.g.,
ammonia) for Cu(II). When carried out in a layer of 1% agarose gel, and initiated by ascorbic
acid, this autocatalytic reaction generates an autocatalytic front. This system is prototypical
of autocatalytic reactions where the formation of a product, which acts as a ligand for a
catalytic metal ion, enhances the production and activity of the catalyst.
Introduction
Autoamplification and autocatalysis are important – although surprisingly uncommon – types
1
of processes in chemistry. Biological cellular division is – in a sense – a type of
autoamplification. Flames and explosions are autocatalytic, as is the formose reaction,2-3
4
5-7
silver-halide photography, photolithography using chemically amplified photoresists,
8
9-12
crystallization, electroless deposition of metals, the Soai reaction,
the formaldehyde–
sulfite reaction,13-14 and the removal of the 9-fluorenylmethoxycarbonyl (Fmoc) protecting
15
group. The Belousov–Zhabotinsky (BZ) reaction (the best known oscillating chemical
reaction) has autocatalysis as a core element,13, 16-17 as does a reaction based on the Kent
ligation – a reaction that we have designed to oscillate.18
This work describes an autocatalytic, copper-catalyzed, azide-alkyne cycloaddition
(CuAAC) reaction that uses the designed reduction of Cu(II) to Cu(I) to generate
2
ACS Paragon Plus Environment

Page 3 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
autocatalysis. We can view the reaction as an autocatalytic cycle driven by the formation of a
ligand that promotes the reduction of Cu(II) to Cu(I) – where Cu(I) is the catalytic metal ion.
This autocatalytic organic reaction has the potential to be applied to a broad range of
substrates, and represents a potentially general mechanism to use in the design of
autocatalytic cycles.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Autoamplification and autocatalysis have been suggested as processes that contribute
to the solution of two core problems in considerations of the origin of life – that is “dilution”
and “mixtures”.19-20 Although Eschenmoser, Sutherland, De Duve, Breslow, Wächtershäuser,
Morowitz, and many others have famously demonstrated how simple, plausible prebiotic
molecules (e.g., cyanide, formaldehyde, formamide, sulfur dioxide, hydrogen sulfide, carbon
dioxide, others) can convert, usually, under carefully controlled laboratory conditions, into
the more complex molecules that make up metabolism (or fragments of them),2, 21-28 it
remains unclear how, or if, dilute solutions containing complex mixtures of these, and other,
compounds would do so. One possible solution to these problems is for reactions to occur in
enclosed or dimensionally constrained spaces (including, but not restricted to, liposomes or
vesicles, water droplets in oil, cracks in rocks, evaporating ponds, freezing water) or adsorbed
on surfaces.29-32 A second solution to the problem of dilution/mixtures is autocatalysis and
autoamplification. Autoamplifying reactions – by providing very efficient conversion of
specific reactants to specific products – might provide one mechanism for generating high
local concentrations of these products. Autocatalysis, thus, might provide a route to increase
the availability of particular molecules (or sets of molecules) important for the emergence of
life.1, 33-37
3
8
Multi-reaction systems that make up metabolism do not ordinarily use direct
autocatalysis – that is, processes in which a catalytic entity catalyzes its own production.
Instead, complex autocatalytic cycles usually require multiple reactions to support
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
autoamplification.1-2, 18, 39-40 Acid-catalyzed hydrolysis of esters, formation of trypsin from
41
4
2
43
trypsinogen, autophosphorylation of protein kinase CK2, and oxidation of oxalic acid by
4
4
40, 45
permanganate are examples of direct autocatalysis. The reverse Krebs cycle,
blood
4
6
18
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
coagulation cascade, thiol autocatalytic reaction, and formose reaction are examples of
autocatalytic cycles.2
Although the subject of autoamplification/catalysis has been a subject of core interest
in chemistry, it has proved very difficult to design new autocatalytic cycles from organic
reactions. Despite the extraordinary versatility of organic chemistry, autocatalytic reactions
are surprisingly rare, and almost all have been discovered by accident.2-3, 9, 47 The literature on
autocatalytic reactions directly relevant to the one we have developed here is large, but not
1
predictive (at least so far) of new reactions. Template-directed reactions, which were
pioneered by von Kiedrowski and Rebek,48-51 are an exception. These reactions are designed
largely based on the rules of molecular recognition. They suffer, however, from product
inhibition and small (usually less than an order of magnitude) difference in rates of templated
and random reaction pathways, and from the structural complexity of the starting material.52
Zubarev et. al, in search for prebiotic precursors to the citric acid cycle, used computational
approaches to propose plausible autocatalytic cycles in the chemistry of carboxylic acids. 40,
53
Our group recently designed a simple autocatalytic cycle based on chemistry of organic
18
thiols, and Otto and coworkers developed, after initial incidental discovery,
mechanochemical autocatalysis in assemblies of cyclic disulfides.54
55
56
57
58
Early work by Finn , Fokin , Binder , and Hardy suggested that the cycloaddition
5
5
56
step of Cu(I)-catalyzed click reactions can be autocatalytic. Finn and Fokin noticed that
tris-(triazolylmethyl)amines form Cu(I) complexes that are more reactive catalysts for
cycloaddition, and, therefore, suggested that the formation of tris-(triazolylmethyl)amines
from tris-(alkynylmethyl)amines proceeds autocatalytically. Although the kinetics of this
4
ACS Paragon Plus Environment

Page 5 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
autocatalysis has not been characterized, Binder reported a CuAAC-based polymerization
that might also have proceeded autocatalytically, and Hardy demonstrated that a CuAAC
5
8
reaction can promote the self-replication of vesicles. These examples, which begin from
catalytically active Cu(I) compounds, however, describe only modest rate enhancements (less
than an order of magnitude) over the course of the reactions.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Our motivation for examining an autocatalytic copper-catalyzed click reaction, based
on the reduction of an inactive Cu(II) starting material to a catalytically active Cu(I) species,
56
was: i) Fokin noted that tris-(triazolylmethyl)amine ligands appeared to stabilize Cu(I) from
disproportionation and increased the redox potential of Cu(I)/Cu(II) by nearly 300 mV. ii)
5
9
Zhu observed that the CuAAC reaction proceeds with Cu(OAc) in the absence of any
2
added reducing agent, and that the addition of 2 mol % of tris-(triazolylmethyl)amine ligands
increased the rate of the reaction. He suggested that: “tris-(triazolylmethyl)amine ligands may
increase the thermodynamic driving force for the reduction of Cu(II) during the induction
period to rapidly produce a highly catalytic Cu(I) species for the AAC reactions.”59
The focus of this manuscript is on the participation of multiple reactions (reaction
networks) to generate a strong autocatalytic rate enhancement, which is an important kinetic
parameter for generating dynamic behaviours, such as oscillations and multi-stability, and for
creating conditions for chemical evolution.
Results and Discussion
We hypothesized that we could design an autocatalytic reaction with an initial reaction rate
that is negligible, thereby creating a larger difference between the initial and final rates of the
reaction by using – as a starting material – a water-soluble and catalytically inactive Cu(II)
salt (CuSO ). To increase the concentration of the catalytic species, the triazole formed in this
4
reaction must be a ligand that promotes the reduction of Cu(II) to Cu(I), where Cu(I) is
required to form the active catalyst, which is likely a dynamic ensemble of multi-nuclear
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Cu(I) species. Scheme 1 outlines the major features of the system of reactions we have
examined.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
ACS Paragon Plus Environment

Page 7 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Scheme 1. A simplified scheme describing the reactions that are involved in the autocatalytic
formation of tris-(hydroxyethyltrizolylmethyl)amine (N(C
(hydroxyethyltrizolylmethyl)propargylamine (N(C )(N
hydroxyethyltrizolylmethyl)dipropargylamine (N(C )(N
. The scheme uses the conversion of 1 to 3
3
N
3
)
3
) bis-
) and
) from tripropargylamine (1)
3
3
C
)
3 2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
3
3
C )
3 2
and 2-azidoethanol (2), in the presence of CuSO
4
to illustrate one plausible route for the initial reduction of Cu(II) to Cu(I), and does not
consider alternative products from the oxidative coupling of 1, or the nature of the Cu(I)
species in the initiation step. The abbreviations we use for the compounds (e.g., N(C
3
N
3
3
) )
are indicated in bold-face text on the figure.
Kinetic Studies of the Reaction of Tripropargylamine with 2-Azidoethanol in the Presence
of CuSO
We tested our hypothesis by allowing tripropargylamine (1) to react with 2-
4
.
azidoethanol (2) and CuSO , in a water:methanol mixture (9:4; v:v) and monitored the
4
1
reaction by H-NMR spectroscopy. We performed this reaction by adding a solution of 1
(109 mM) in CD
3
OD to a solution of 2 (309 mM) and CuSO
4
(43 mM) in D O, at room
2
temperature (Fig. 1a). The low concentrations of reactants, compared to previous studies55-56
,
62
allowed us to overcome issues with product inhibition and to characterize the kinetics of the
reaction in detail. Simple visual observation of the reaction showed an initially pale blue,
almost clear, solution containing hydrated Cu(II) ions, which remained unchanged for about
2
0 minutes, before the solution became more opaque and – after approximately 50 minutes –
changed to a dark blue color – a color typical of Cu(II) triazole complexes (Fig. 1c). This
apparent incubation period, followed by a relatively sudden change of color (associated with
the formation of Cu(II) triazole complexes), suggested that the reaction between 1, 2, and
CuSO has an autocatalytic character.
4
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. The time course of the reaction between tripropargylamine (1), 2-azidoethanol (2),
and CuSO
4
. Concentrations were estimated by integrating the alkyne proton against a tert-
1
butanol internal standard. a) H NMR spectra showing the disappearance of the proton
signals of 1 and 2 over time. b) Plot of the chemical shifts of 1 during the first 3300 s of the
reaction. After 3300 s, the alkyne protons (~ 2.6 ppm) disappear, and the propargylic protons
(~ 3.4 ppm) change (bracketed region); this change indicates the formation of a small amount
of a new species (□, whose structure we have not defined). c) Images of an NMR tube
containing the reaction mixture at different times. Standard reaction conditions were 1 (109
mM), 2 (309 mM), and CuSO
4
(43 mM) in a mixture of D
2
O/CD OD (9:4, v:v) at 25 °C.
3
Monitoring a reaction by NMR is often impractical in the presence of paramagnetic
Cu(II) ions. Fortunately, however, the NMR signals of 1 and 2, though slightly broad, were
8
ACS Paragon Plus Environment

Page 9 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
sufficiently sharp for quantitative spectroscopy, and could be accurately integrated against an
internal standard of tert-butanol. The reaction products, mono-, bis- and tris-
(triazolylmethyl)amines, however, were not visible in the NMR spectrum when Cu(II) ions
were present.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
To examine the kinetics of the reaction, we followed the disappearance of the alkyne
proton of 1, at 2.6 ppm (Fig. 1a). We used this proton to monitor the progress of the reaction,
because it appears in a clear region of the NMR spectrum and H-D exchange was negligible
during an hour at pH 4.7 (which corresponds to the pH of the initial reaction mixture, see
supporting information for details). The kinetic profile of the reaction resembled that of a
typical autocatalytic reaction, with a lag phase, an exponential phase, and a saturation phase
(Fig. 2a). The exponential phase was accompanied by a shift (of only partly identified origin)
in the resonance frequency of the protons of 1 (Fig. 1b), which correlated with the change in
color of the solution to dark blue. We determined the final composition of the reaction
mixture by reducing all remaining Cu(II) to [Cu(I)(CN)
](x-1)- with an excess of potassium
x
cyanide63-64, and analyzing the mixture by H NMR spectroscopy. The final product of the
1
reaction was the tripodal ligand tris-(2-hydroxyethyltriazolylmethyl)amine (which we
1
abbreviate as N(C
3
N
3
)
3
), which formed in 85% yield (as determined by H-NMR); the
methylene signal adjacent to the amine was integrated relative to an internal tert-butanol
standard.
If a reaction is autocatalytic, then addition of the autocatalyst to the reaction will
shorten its lag phase. We performed an NMR kinetics experiment, identical in form to the
one described above, but with the addition of N(C
3
N
3
3
) (1 mol % relative to 1) and observed
a decrease in the duration of the lag phase, by a factor of three (Fig. 2a).
9
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. Experiments showing elimination of the lag period with the addition of an
autocatalyst in the reaction between 1 (109 mM), 2 (309 mM), and CuSO (43 mM). a) Plot
4
showing the disappearance of the alkyne proton of 1 (at 2.6 ppm) over time, as determined by
1H NMR. The numbers above the traces show the mol% of tris-(2-
hydroxyethyltriazolylmethyl)-amine (N(C
3
N
3
3
) ) added relative to 1. All reactions were
performed in a mixture of D O/CD OD (9:4, v:v) at 25 °C, in an NMR tube, and the
2
3
concentration of tripropargylamine was calculated by integrating the alkyne proton against a
tert-butanol internal standard. b) UV-Vis absorption spectra at various time points during the
reaction with 1 mol% of autocatalyst (the mixture of complexes of mono-, bis- and tris-
1
0
ACS Paragon Plus Environment

Page 11 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
(
triazolylmethyl)amines with copper) added relative to 1, in a H
2
O/CH OH (9:4, v:v) mixture
3
at 25 °C. Copper complexes of triazolylmethylamines absorb at 650 nm. c) UV-Vis analysis
of the reaction, using the same conditions as in b, performed by measuring the absorption at
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
50 nm. The numbers above the traces show the approximate mol% of the autocatalyst (the
mixture of complexes of mono-, bis- and tris-(triazolylmethyl)amines with copper from a
reaction that had previously reached completion) relative to 1.
We also tested the reaction in a H
2
O:CH OH (9:4, v:v) mixture by monitoring the
3
change in absorption at 650 nm (Fig. 2b), because Cu(II)-triazolylmethylamines complexes
absorb light more strongly at this wavelength than unbound Cu(II) (i.e., the aqua complex)
(Fig. S2). Unexpectedly, in the reaction without any added autocatalyst, there was no
detectable reaction within the first 6000 s, and autocatalysis began only after 7000 s (~2
hours) (Fig. 2c; details of this difference in rate are discussed in a following section). The
addition of 1 mol % (relative to 1) of the autocatalyst – the mixture of complexes of mono-,
bis- and tris-(triazolylmethyl)amines with copper from the previously complete reaction –
shortened the lag phase to 1800 s, and the addition of 5 mol % or 10 mol % of the
autocatalyst completely eliminated the lag phase (Fig. 2c).
During the reaction of 1 (109 mM), 2 (309 mM), and CuSO
4
(43 mM) in a
H
2
O:CH OH (9:4, v:v) mixture, the pH of the solution increased from 4.7 to 6.2. To test
3
whether this increase of 1.5 pH units contributed to autocatalysis we ran the reaction in
acetate buffer (340 mM), but under otherwise identical reaction conditions. The buffered
reaction gave similar kinetics to that of the unbuffered reaction, suggesting that the change in
pH does not contribute strongly to autocatalysis (Fig. S3).
Propagation of an Autocatalytic Reaction Front. Autocatalytic reactions form autocatalytic
65
fronts when they take place without mixing. The observation of an autocatalytic front
1
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
provides additional support for autocatalysis, as opposed to other mechanisms for delayed
activation. For instance, simple CuAAC reactions accelerated by tris-triazolyl ligands can
6
6
have observable lag phases . Because the catalytic species in CuAAC reactions are multi-
nuclear, and under most circumstances only a fraction of the total copper present is a part of
the operational catalyst, the required evolution of catalyst speciation may result in an
observable lag phase. We demonstrated that the autocatalytic CuAAC reaction formed an
autocatalytic reaction front by performing the reaction in a layer of 1 % agarose gel (1 mm
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
thick) in H
2
O:CH
3
OH (9:4, v:v), loaded with 1 (125 mM), 2 (309 mM), and CuSO (84 mM).
4
We initiated autocatalysis by adding a small (~0.1 mm) crystal of ascorbic acid (which
rapidly reduces Cu(II) to Cu(I)) (Fig. 3a and supplementary video). Initially, the agarose gel
appeared clear, with weak blue coloring When the ascorbic acid was added, the area in
.
contact with the crystal turned yellow, because Cu(II) was reduced to Cu(I) (which, in the
presence of alkynes, forms polynuclear Cu(I) acetylide complexes that are yellow). The area
in contact with Cu(I) subsequently underwent the CuAAC reaction, and as triazolyl ligands
were produced, the gel turned to the dark blue color associated with Cu(II)/triazolyl
complexes. The area closest to the ascorbic acid crystal used to initiate the reaction remained
yellow because Cu(II) was being continuously reduced to Cu(I). The autocatalytic front
propagated radially with constant velocity (as illustrated by the time/space plot, Fig. 3b) at a
rate of 0.0325 ± 0.0010 mm/min. Propagation of the reaction front continued for 4 hours,
with a final radius of 10 mm.
Two characteristics of the autocatalytic CuAAC reaction described here make it
suitable for the study of dynamic phenomena in reaction-diffusion systems: (i) a low rate of
spontaneous activation; and (ii) an easy detection by color change. We note that organic
autocatalytic reactions (i.e. autocatalytic reaction of thiols and template-directed reactions)18,
5
1
usually have rates of spontaneous activation that prevent prolonged observation of an
1
2
ACS Paragon Plus Environment

Page 13 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
autocatalytic front. For example, an autocatalytic reaction front driven by the template-
directed cycloaddition of a nitrone to an alkene propagated only for about 20 min before the
5
2
reaction spontaneously activated in bulk. By contrast, for the system described in this paper,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
spontaneous activation beyond the propagating front was only observed after 300 min.
Figure 3. Reaction front driven by the autocatalytic copper catalyzed azide-alkyne
cycloaddition. a) Photographs of the reaction propagating in 1 mm thick agarose gel loaded
1
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
with 1 (125 mM), azidoethanol (320 mM), and CuSO
4
(84 mM). We initiated the reaction at
the central point in the gel using a crystal of ascorbic acid. The yellow color comes from the
reduced Cu(I) species, the blue color comes from the Cu(II) complex with N(C
N(C ) and indicates progress of the reaction. b) Graph that shows that the reaction front
propagates with constant velocity.
3
N
3
)
3
(Cu(II)
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
N )
3 3
Mechanism of the Reaction of Tripropargylamine With 2-Azidoethanol in the Presence of
CuSO
4
.
Initiation of the Reaction. Our initial hypothesis was that autocatalysis would require the
addition of a reducing reagent to convert Cu(II) to Cu(I). In fact, this reduction proceeded in
the presence of only 1 and 2: no additional reducing agent was required. Because the
reduction of Cu(II) to Cu(I) by alkynes is a well-known reaction, and is the basis for the
6
7
Eglinton coupling, we propose that 1 (either as an alkyne or a tertiary amine) acts as a
reducing agent in the reaction. To test this hypothesis, we mixed 1 (109 mM) and CuSO
4
(43
mM) in D
2
O:CD OD in the absence of azide 2. The yellow precipitate expected for a Cu(I)
3
acetylide formed within an hour. X-ray photoelectron spectroscopic (XPS) data confirmed
the presence of Cu(I), carbon, and nitrogen in this precipitate (Fig. 4a; Fig. S4).
To determine which functional group of 1 (the alkyne or the amine) acts as the
reducing agent, we examined two model reactions: (i) We allowed propargyl alcohol (500
o
mM) to react with CuSO
4
(43 mM) in acetate buffer (200 mM, pH 4.7) at 60 C for 2 min,
and (ii) we allowed triethylamine (110 mM) to react with CuSO (43 mM) in acetate buffer
4
o
(
200 mM, pH 4.7) at 60 C for 2 min. The reaction with propargyl alcohol resulted in the
reduction of Cu(II) to Cu(I), and formation of a yellow precipitate of Cu(I) acetylide, while
no reaction was observed with triethylamine. ESI-MS data from the reaction of 1, 2, and
CuSO
4
in H
2
O:CH
OH showed the presence of butadiyne 3 in the reaction mixture (M+Na+
3
1
4
ACS Paragon Plus Environment

Page 15 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
2
83.1). We therefore infer that the reduction of Cu(II) to Cu(I) by the alkyne functionality of
is likely the initiation step for the cycloaddition between the azide and alkyne. To support
1
this proposal, we demonstrated that increasing the starting concentration of 1 decreased the
duration of the lag phase (Fig. 4b). We note, however, that the reduction in the lag phase may
be partially influenced by the increased concentration of the tertiary amine, which could be
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
2
functioning to depolymerize unreactive and highly-aggregated Cu(I) acetylides .
Catalytic Properties of Cu(I) Complexes with Tris-triazolylmethylamines
To investigate the contribution of tris-triazolylmethylamine ligands on the
acceleration of the Cu(I)-catalyzed cycloaddition reaction, we performed a control
experiment in which Cu(I) was added at the start of the reaction, and was maintained in the
reduced state by the presence of a 2x excess (relative to the concentration of CuSO ) of
4
ascorbic acid (supporting information Fig. S5). Reactions initiated with Cu(I) at 43 mM
proceeded at rates that were too large to be monitored by NMR. To decrease the rate of the
reaction to a rate that is compatible with NMR analysis, and especially to monitor the initial
stages of the reaction, we decreased the concentration of copper to 2 mM. Because Cu(I) was
present at the beginning of the reaction, we saw no lag phase. We did, however, observe a
slight (approximately 2x) increase in rate during the initial stages of the reaction; the
observation is compatible with autocatalysis. Because the initial concentration of Cu(I) was
lower, the speciation of Cu(I) (which may have a significant impact on the rate of the
6
8
cycloaddition ) will have been different, and thus the rate (and change in rate over time) is
not necessarily directly comparable with our other experiments. Nevertheless, this increase in
rate, although small compared to our systems that use Cu(II) as a precursor, is probably
analogous to the rate enhancement reported by Fokin,56, 62 and is comparable to that reported
by Binder.57
1
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 16 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
6
ACS Paragon Plus Environment

Page 17 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Figure 4. Mechanistic studies of the reaction between 1, 2, and CuSO . a) XPS data showing
4
the presence of Cu(I) in the precipitate formed in the reaction of 1 (109 mM) and CuSO (43
4
1
mM) in D
2
O/CD
3
OD (9:4, v:v) mixture. b) H NMR kinetic experiments for the reaction
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
between 1, 2 (309 mM), and CuSO
4
(43 mM) in a D
2
O/CD OD (9:4 v:v) mixture at 25 °C,
3
starting from different amounts of 1. The concentration of tripropargylamine was calculated
by integrating the alkyne proton against a tert-butanol internal standard. c) Changes in
intensity of ESI-MS signals of some triazole species during the autocatalytic CuAAC
reaction. The reaction was carried out under the same conditions as the experiment shown in
panel d. d) Change in concentrations of N(C
3
)
2
(C
3
N
3
), N(C
3
)(C
3
N
3
)
2
, and N(C
3
3
N ) in the
reaction of 1 (109 mM), 2 (309 mM), and CuSO
4
(43 mM) determined by NMR
measurements. Samples were removed from the reaction and quenched by addition to 2
weight% aqueous solutions of KCN. e) UV-vis analysis of reactions with different starting
concentrations of N(C
43 mM) in H O/CH OH (9:4 v:v) mixture at 25 °C. f) Cyclic voltammogram (scan rate 100
mV/s) of CuSO (5 mM), Na SO (50 mM), and N(C (10 mM) in H O/CH OH (9:4
v:v). g) The change in potential of a Pt wire electrode vs a Ag|AgCl reference electrode (1.0
M KCl as reference solution) during the reaction of 1 (109 mM), 2 (309 mM), and CuSO (43
mM) in a H O/CH OH (9:4 v:v) mixture at 25 °C. The reaction was initiated by 1 mol % of
triazolylmethylamines.
3
N
3
)
3
. The reaction contained 1 (109 mM), 2 (309 mM), and CuSO
4
(
2
3
4
2
4
3
N
3
)
3
2
3
4
2
3
The Role of Intermediate Cycloaddition Products. The simplified sequence of
reactions summarized in Scheme 1 proposes sequential formation of mono-, bis-, and tris-(2-
hydroxyethyltriazolylmethyl)amines. We investigated the roles of these species in
autocatalysis. First, we used ESI-MS to monitor the reaction (see experimental section for
details), and observed that (2-hydroxyethyltriazolylmethyl)dipropargylamine (N(C
3
)
2
(C
3
3
N )),
1
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 18 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
and bis-(2-hydroxyethyltriazolylmethyl)propargylamine (N(C
species formed during the initial stages of the reaction (i.e., during the lag phase),
N(C )(C was the major species formed during the exponential phase, and tris-(2-
hydroxyethyltriazolylmethyl)amine (N(C ) was formed in significant quantities only near
the end of the reaction (once almost all of the tripropargylamine had been consumed; Fig.
3
)(C
3
N
3
)
2
) were the major
3
3
N )
3 2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
N )
3 3
4
1
c). Second, we measured the kinetics of the reaction by NMR spectroscopy by collecting
00-μL samples, quenching them in KCN solution (2 weight% in D O:CD OD), and
2
3
measuring their NMR spectra (Fig. S6). KCN quenches the reaction by converting all Cu(II)
to [Cu(I)(CN)
x
](x-1)-, which is not an active catalyst for cycloaddition. This system also
1
permits recording of H NMR spectra, where N(C
3
)
2
(C
3
N
3
), N(C
3
)(C
3
N
3
)
2
, and N(C
3
N
3
3
) are
visible and resolvable. The results show that no triazole compounds are formed (above the
.
-3
detection limit of NMR spectroscopy: about 1 mM) until 800 s (rate < 1.25 10 mM/s), and
that the maximum rate of formation of triazoles, at around 2000 s, is about 0.5 mM/s (Fig. 4d,
Fig S9). Thus, we observed a rate enhancement of more than 400x, which explains the
prolonged propagation of the autocatalytic front without spontaneous reaction outside of the
reaction front. Consistent with the MS data, N(C
3
)(C
3
N
3
)
2
was the major species formed
during the exponential phase (Fig. 4d). This result might be, at least partially, a consequence
ligands to Cu(I),62
of product inhibition by bidentate chelation of two N(C
3
)(C
3
N )
3 2
effectively trapping the active Cu(I) catalyst in a stable, inactive, form and briefly isolating
N(C
3
)(C
Both the MS and NMR experiments suggest that the formation of N(C
)(C is not cooperative, since N(C is not formed in the earlier stages of the
)(C from
) did not accumulate in the
3
N
3
)
2
from further reaction.
3
N
3
3
) from
N(C
3
3
N
3
)
2
3
N )
3 3
reaction. The NMR data, however, suggested that the formation of N(C
N(C (C ), is, to some extent, cooperative, since N(C (C
mixture and was quickly converted to N(C )(C
3
3
N
)
3 2
3
)
2
3
N
3
3
)
2
3
N
3
3
3
N
3
2
) .
1
8
ACS Paragon Plus Environment

Page 19 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
To understand the roles of the N(C
3
)
2
(C
3
N
3
), N(C
3
)(C
3
N
3
)
2
, and N(C
3
N
3
) in the
3
autocatalytic process, we studied the effect of adding them to the initial reaction mixture on
the kinetics of this reaction (Fig. 4e and supporting information Fig. S7). Adding a small
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
amount of N(C
indistinguishable from that obtained by adding 1 mol % (relative to 1) of the mixture from
the completed reaction (i.e., a mixture of N(C (C ), N(C )(C , N(C , and their
copper complexes). Adding either 5 mol % or 10 mol % of N(C eliminated the lag
phase, but also decreased the maximum slope of the kinetic curve. When 10 mol % of
N(C (C ) or N(C )(C was added to the reaction, the lag phase (which included the
interval from 0 - 4000 s for N(C (C ), and from 0 - 1000 s for N(C )(C ; Fig. S7)
was not completely eliminated, although the slopes of the kinetic curves were higher than in
the experiment with 1 mol % of N(C . This observation suggested that N(C is the
most active of these three species in accelerating the reduction of Cu(II) to Cu(I), although
N(C )(C might play a more important role in catalyzing the CuAAC reaction. We note
3
N
3
)
3
(1 mol % relative to 1) resulted in a kinetic curve that is effectively
3
)
2
3
N
3
3
3
N
3
)
2
3
N
)
3 3
3
N
)
3 3
3
)
2
3
N
3
3
3
N )
3 2
3
)
2
3
N
3
3
3
N
)
3 2
3
N
3
)
3
3
N )
3 3
3
3
N )
3 2
here, however, that the exact mechanism for the reduction of Cu(II) to Cu(I), and the nature
of the species involved, are not known.
Electrochemical studies. We hypothesized that triazolylmethylamines stabilize Cu(I)
against disproportionation in water/methanol mixtures. Cu(I) ions disproportionate in water,
6
9
or water/methanol mixtures, to Cu(II) and Cu(0) . As a consequence of the tendency for
Cu(I) to disproportionate, the cyclic voltammogram (CV) obtained from CuSO (5 mM) in a
mixture of H O:CH OH (9:4, v:v) gave two oxidation and reduction peaks (Fig. S8). The CV
of CuSO (5 mM) and N(C (10 mM), in a mixture of H O:CH OH (9:4, v:v), however,
gave only one peak (Fig. 4f), corresponding to the reduction of Cu(II) to Cu(I). The ligand,
N(C , pushes the redox potential of the reduction of Cu(I)/Cu(0) to negative values, to
4
2
3
4
3
N
3
)
3
2
3
3
N
)
3 3
the extent that we do not observe this peak within the 2 V potential window. This shift in the
1
9
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 20 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
E°Cu(I)/Cu(0) makes the disproportionation of Cu(I) unfavorable (Edisproportionation=ECu(I)/Cu(0)
-
E
Cu(II)/Cu(I)), and stabilizes Cu(I) in the complex with N(C
3
N
3
3
) . This stabilization of the
catalytically active Cu(I) ions in solution facilitates the cycloaddition reaction.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
To monitor the redox reactions taking place during the autocatalytic reaction, we
recorded the open-circuit potential of the solution. We monitored the potential of a Pt wire
(relative to a Ag/AgCl reference electrode) during the reaction between 1 (109 mM), 2 (309
mM), and CuSO (43 mM), in a H O:CH OH mixture. Figure 4g shows the resulting
4
2
3
potential curve, which has four characteristic features: (i) an initial spike in potential,
immediately after the addition of 1; (ii) an 80 mV drop in potential after the addition of 1%
triazolylmethylamines (500 – 2000 s); (iii) a period of approximately constant potential (2000
–
4000 s); and (iv) a 100 mV drop in potential starting at 4000 s. The potential drop at 4000 s
correlated with a color change from pale to dark blue. Although unambiguous interpretation
of open circuit potential measurements is difficult, the second drop in potential (4500 s)
might plausibly originate from an increase in the concentration of Cu(I), caused by the
chemical reduction of Cu(II) during the autocatalytic process.
Inverse Solvent Kinetic Isotope Effect. We observed (based on the duration of the lag
phase) an apparent inverse solvent kinetic isotope effect (KIE) in the reaction between 1, 2,
and CuSO
9:4, v:v)). The lag phase ends at around 1500 s (~25 min) in D
7000 s (~116 min) in H O/MeOH, under otherwise identical reaction conditions (Fig. 2b).
4
(that is, the lag phase ended earlier in D
2
O:MeOD (9:4, v:v) than in H
2
O:MeOH
(
2
O/MeOD (Fig. 2b), and after
2
Because we believe that the lag phase is a consequence of the slow reduction of Cu(II) to
Cu(I), the observed solvent kinetic isotope effect likely involves the alkyne-mediated
reduction of Cu(II) to Cu(I). The details of the mechanism and intermediate species of the
reduction of Cu(II) to Cu(I) by terminal alkynes are complex, and are still under considerable
7
0
debate (as are the details of the mechanism and intermediate species of the CuAAC
2
0
ACS Paragon Plus Environment

Page 21 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
71
reaction ). The processes that are believed to be involved, however, (hybridization changes,
reductive elimination, and/or transition metal C-H activation) are chemical processes often
7
2
associated with KIEs .
We hypothesized that, in deuterated solvent and in the presence of copper, the alkyne
protons of 1 may exchange with deuterium from D O and/or MeOD, and that the deuterated
product 1-d (i.e., tripropargylamine with its three alkyne protons replaced with deuterium)
may be the origin of the observed inverse KIE. We thus ran the reaction between 1-d , 2, and
CuSO in a mixture of H O:MeOH (9:4, v:v), and monitored the reaction by UV/Vis
spectroscopy, at 650 nm (Fig. 5). As a control, we also ran the reaction between 1-d
CuSO in a mixture of D O:MeOD (9:4, v:v).
Figure 5 shows the reaction progress of four different reactions, run under the same
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
3
4
2
3
, 2, and
4
2
reaction conditions: 1) 1-d
and CuSO in a mixture of D
O:MeOH (9:4, v:v); 4) 1, 2, and CuSO
If 1-d
reaction involving 1-d
and D O/MeOD. Figure 5, however, shows that the lag phase for the reaction with 1-d
O/MeOH was even longer than that using 1 in H O/MeOH, ending after ~11000s (183
3
, 2, and CuSO
4
in a mixture of H
2
O:MeOH (9:4, v:v); 2) 1-d
in a mixture of
O:MeOD (9:4, v:v).
3
, 2,
4
2
O:MeOD (9:4, v:v); 3) 1, 2, and CuSO
4
H
2
4
in a mixture of D
2
3
were causing the observed inverse KIE, the duration of the lag phase of the
and H O/MeOH would resemble that observed in the reaction of 1
in
3
2
2
3
H
2
2
min). This observed normal KIE supports the involvement of the alkyne proton in the lag
phase (reduction of Cu(II) to Cu(I)), but also indicates that it is not the origin of the observed
inverse KIE. Furthermore, the control reaction between 1-d
3
, 2, and CuSO
4
in a mixture of
D
2
O:MeOD (9:4, v:v), had a longer lag phase than that of 1 in D
2
O/MeOD. Thus, while the
alkyne displays a normal KIE, and is involved in the lag phase, the observed inverse solvent
KIE is not affected by the alkyne proton. The idea that these two KIEs act independently is
2
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 22 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
supported by the effect of isotopic substitution of the alkyne (1-d ) on the duration of the lag
3
phase, which was roughly the same for both solvent systems.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. Top: reaction progress monitored by UV/Vis spectroscopy, at 650 nm, of four
different reactions. In all forms, the starting concentrations were: 1 or 1-d (109 mM), 2 (309
mM), and CuSO (43 mM) in a mixture of H O:MeOH or D O; MeOD (9:4, v:v). The dashed
3
4
2
2
lines are sigmoidal fits to the data, omitting the region containing the artifact of precipitation
and light scattering. Bottom: Representation of how the two kinetic isotope effects (KIEs)
plausibly and approximately independently influence the lag phase of these four reactions.
The position of the colored bars corresponds to the approximate duration of the lag phase on
the x-axis.
2
2
ACS Paragon Plus Environment

Page 23 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Figure 5 indicates that these reactions show a spike in absorbance as the lag phase
ends. This absorbance peak corresponds to the formation of precipitates, which we expect are
insoluble Cu(I) intermediates. The intensity of this absorbance peak (and thus the degree of
precipitation) also correlates with the duration of the lag phase (reactions with shorter lag
phases have larger absorbance peaks). We attribute this observation to the fact that shorter lag
phases have more rapid formation of Cu(I) intermediates, which thus accumulate in larger
concentrations (and thus precipitate to larger extents).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Given that the inverse solvent KIE is not affected by the alkyne proton but is still
involved in the reduction of Cu(II), we suspected that D
reduction potential of the Cu(II)/N(C complex. We did not, however, see a change in the
cyclic voltammogram (scan rate 100 mV/s) of CuSO (5 mM), Na SO (50 mM), and
N(C (10 mM) in D O/CH OD (9:4; v:v), as compared to that in H O/CH OH (9:4; v:v)
2
O and/or MeOD may influence the
3
N )
3 3
4
2
4
3
N
3
)
3
2
3
2
3
(Fig. 4f). We can thus only speculate that these deuterated solvents influence the rate of
reduction of Cu(II) to Cu(I) through an isotope-dependent solvation effect that reduces the
7
3
activation free energy of electron transfer. We have thus not identified the origin of the
negative KIE at this time. Since this mechanistic feature―while interesting―is secondary to
the focus of the work, we leave it unresolved.
Summary of the Mechanism. We summarize our current inferences concerning the
mechanism of the autocatalytic CuAAC reaction as follows: The reaction starts with an
initial, slow, reduction of hydrated Cu(II) to Cu(I), where an alkyne serves as the reducing
agent. The reduction of Cu(II) to Cu(I) by the acetylenic group of N(CH
the initial Cu(I) complexes that are catalytically active in the cycloaddition. The products of
the initial and subsequent cycloadditions—N(C (C ), N(C )(C , and N(C
Scheme I)—form coordination complexes with Cu(I) and Cu(II). Uncoordinated Cu(I) is
unstable in water/methanol solutions and disproportionates. Here, the triazolyl amine ligands
2
C≡CH) (1), leads to
3
3
)
2
3
N
3
3
3
N
3
)
2
3
N )
3 3
(
2
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 24 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
form stable and soluble complexes with Cu(I), which maintain copper in its catalytically
active oxidation state, Cu(I), in solution. The formation of N(C
N(C also accelerate the reduction of Cu(II) to Cu(I), although the exact reasons for this
acceleration are unclear, and might involve intermediates in the CuAAC reaction.
3
)
2
(C
3
N
3
), N(C
3
)(C
3
N
3
2
) , and
3
N )
3 3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Thus, formation of the Cu(I) species—the catalytically active species in the click
(cycloaddition) reaction—is promoted by the formation of ligands that are the product of that
reaction. The reaction cycle is autocatalytic because the production, and stability in solution,
of Cu(I) is promoted by the aminotriazolyl ligands, and production of the aminotriazolyl
ligands is accelerated by Cu(I) (Scheme 2). The Cu(I) species that are formed in the reduction
process might, however, be initially catalytically inactive and require extra steps to rearrange
into catalytically active complexes. An additional contribution to autocatalysis, although
probably a less important one, comes from the increased activity of Cu(I) in the CuAAC
reaction when it is complexed with an aminotriazolyl ligand. As the CuAAC reaction
(catalyzed by Cu(I)) progresses, more aminotriazolyl ligands are produced. The
aminotriazolyl ligands coordinate Cu(I) (in addition to Cu(II)) to form a more reactive Cu(I)
catalyst, which in turn accelerates the rate of formation of the aminotriazolyl ligands.
Based on this reaction profile, we have developed a numerical model, involving six
simplified reactions, to describe the proposed mechanism (see supporting information for
details). The numerical solution of these equations shows kinetics that resemble the
experimental data (supporting information, Fig. S9). This type of modeling shows that a
plausible kinetic scheme (with adjustable rate constants) can model the observed data
adequately. As with all similar weakly constrained models, “compatibility” is not “proof”,
but the goodness of fit of the simulated data—using physically plausible values of rate
constants—provides further support for the general scheme proposed.
2
4
ACS Paragon Plus Environment

Page 25 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 2. Proposed important steps in the autocatalytic reaction between propargylamines
and azides in water or water/methanol in the presence of Cu(II) salts.
Substrate Scope.
The reaction mechanism outlined in scheme 2 suggests that autocatalysis is not dependent on
the structure of the azide. To test the dependence of the structure of the substrate on
autocatalysis, we ran the reaction with two additional azides–tetraethylene glycol diazide (4)
and benzyl azide (5). In the first experiment we allowed 1 (309 mM), 4 (150 mM), and
Cu(SO
4
) (43 mM) to react in a mixture of D
2
O/CD OD (9:4, v:v). The concentration of 4 was
3
reduced (relative to the reactions with 2) to maintain the same relative concentration of azide.
.3H
O (43
In the second experiment, we allowed 1 (309 mM), 5 (309 mM), and Cu(NO
mM) to react in pure CD
azide is insoluble in the water/methanol (9:4, v:v) mixture, and we used a different source of
Cu(II) to increase its solubility in CD OD. Both the reaction with azide 4, and that with azide
, gave sigmoidal kinetics, with lag phases and exponential growth phases that were similar
3
)
2
2
3
OD. We used a different solvent in this experiment because benzyl
3
5
to those observed with 2 (Fig. 6a). We therefore conclude that the structures of the azide
2
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 26 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
component has only a weak influence on the kinetics of the reaction, and that the reaction can
tolerate a variety of substituted azides.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
Figure 6. Scope of the autocatalytic CuAAC reaction. H NMR kinetics experiments for the
reaction between 1 (109 mM), CuSO
or benzylazide (5, 260 mM). The experiment with 4 was conducted in a D
v:v) mixture at 25 °C. The experiment with 5 was conducted in pure CD OD. The
4
(43 mM), and tetraethylene glycol diazide (4, 150 mM)
2
O/CD OD (9:4,
3
3
concentration of alkyne was calculated by integrating the alkyne proton against a tert-butanol
internal standard.
We also tested the reaction of 2 (327 mM) with propargylamine (309 mM) and
CuSO (43 mM) in a water/methanol (9:4, v:v) mixture. The reaction displays sigmoidal
4
kinetics, but the formation of precipitates, and the combination of copper speciation,
disproportionation of Cu(I) complexes, and depolymerization of insoluble Cu poly-
acetylides, makes an unambiguous interpretation of this sigmoidal kinetic curve challenging
(Fig S10 and supplementary discussion).
Displacing Ammonia From Cu(II) ions.
A possible extension of the autocatalytic cycle (Scheme 2) is the displacement of a ligand
that binds to Cu(II) (such as ammonia) by the triazolylmethylamines formed in the reaction
(Scheme 3). The release of a free ligand opens a new path to couple autocatalysis to
independent chemical reactions.
2
6
ACS Paragon Plus Environment

Page 27 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 3. Substitution of ammonia from Cu(II) ammonia complex by N(C
3
N
3
3
) ligand.
We ran the reaction of 1, 2, and CuSO
4
in the presence of ammonia (240 mM) and
1
ammonium chloride (430 mM), and monitored the reaction by H NMR. The disappearance
of 1 followed an approximately sigmoidal curve, characteristic of an autocatalytic reaction
(Fig. 7). The formation of precipitates during intermediate stages of the reaction may be the
cause of the deviation of the course of the reaction from the expected sigmoid. When the
reaction was complete, the solution was pale yellow, which is in contrast to the bright blue
color of reactions without ammonia. The most plausible explanation for this difference in
color is a faster reduction of Cu(II) aminotriazolyl complexes in the presence of ammonia,
perhaps as a result of the increased pH of the solution. Reduction of Cu(II), thus, happens
faster than cycloaddition, and all copper is reduced to yellow Cu(I) complexes. When
exposed to air, the color of the complete reaction mixture changes back to blue. This
experiment demonstrated that we can extend the scope of the autocatalytic CuAAC reaction
to reactions that involve complexes of Cu(II). This experiment also provided further evidence
that autocatalysis is not a consequence of an increase in pH during the reaction, because the
reaction remains autocatalytic when performed in an ammonia/ammonium chloride buffer.
2
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 28 of 32
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
Figure 7. H NMR kinetic experiments for the reaction between propargylamine (109 mM),
azidoethanol (309 mM), CuSO (43 mM), NH (240 mM), and NH Cl (430 mM).
Experiments were conducted in D O/CD OD (9:4 v:v) mixture at 25 °C. The concentration of
4
3
4
2
3
tripropargylamine was calculated by integrating the alkyne proton against a tert-butanol
internal standard.
Conclusions
This work describes an autocatalytic system where coupling the CuAAC reaction and
the reduction of Cu(II) to Cu(I) affords a large rate enhancements over the course of the
reaction. We consider this system of reactions as prototypical of autocatalytic cycles. In this
example, a classical catalytic cycle (the CuAAC reaction) is coupled to a process (the
reduction of Cu(II) to Cu(I)) that generates an extra molecule of the catalyst – a process that
“amplifies” the number of molecules of catalyst (in principle, exponentially) and that
underlies the mechanism of all autocatalytic reactions. This system is driven by the catalytic
formation of a product that, by acting as a ligand, enhances the production and activity of the
catalyst. This characteristic of the product(s) is achieved by (i) the formation of a
nucleophilic triazole ring from a non-nucleophilic azido group, and (ii) the formation of a
chelate ligand, from a monodentate ligand. Specifically, the organic azide group from
azidoethanol (which does not bind strongly to copper ions) converts to a triazole (which does
2
8
ACS Paragon Plus Environment

Page 29 of 32
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
coordinate strongly to copper ions) and a monodentate tripropargylamine converts to a
tetradentate triazolylmethylamine (which bind tightly to Cu(I) and Cu(II) ions).
The autocatalytic CuAAC reaction is compatible with a range of substrates, and can,
in principle, generate polymeric/oligomeric products. We illustrated two subtypes of the
autocatalytic cycle (See Scheme S1, supporting information): (i) the product ligand forms the
active catalyst from a solvated metal ion; (ii) the product ligand forms a complex from a
metal ion containing an ancillary ligand which is released upon complexation.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
This reaction will aid in the development and understanding of chemical reaction
networks. This, and other work examining mechanisms of autocatalysis, may also help to
form a better picture of the processes that led to the emergence of life on earth, because
similar processes (kinetically, although certainly—in this case—not in molecular detail)
might generate autocatalysis in mixtures of molecules (for example, alkynes and nitriles, or
metal ions bound to peptides) that may have been important for the origin of life.74 75-76
Associated Content
Supporting Information
Experimental details of syntheses and kinetics experiments.
Details of the kinetic model describing autocatalysis.
Supplementary video of the propagating autocatalytic front.
The Supporting Information is available free of charge on the ACS Publications website.
Author Information
Corresponding Author
* E-mail: gwhitesides@gmwgroup.harvard.edu
Present Addresses
‡
Department of Organic Chemistry, Weizmann Institute of Science, Rehovot 76100, Israel.
†
SCAMT Laboratory, ITMO University, St. Petersburg 197101, Russian Federation.
2
9
ACS Paragon Plus Environment