Understanding atom transfer radical polymerization: Effect of ligand and initiator structures on the equilibrium constants

Article abstract of DOI:10.1021/ja802290a

Equilibrium constants in Cu-based atom transfer radical polymerization (ATRP) were determined for a wide range of ligands and initiators in acetonitrile at 22°C. The ATRP equilibrium constants obtained vary over 7 orders of magnitude and strongly depend o

Full text of DOI:10.1021/ja802290a

Published on Web 07/19/2008
Understanding Atom Transfer Radical Polymerization: Effect
of Ligand and Initiator Structures on the Equilibrium
Constants
Wei Tang,† Yungwan Kwak,† Wade Braunecker,† Nicolay V. Tsarevsky,†
Michelle L. Coote,‡ and Krzysztof Matyjaszewski*,†
Center for Macromolecular Engineering, Department of Chemistry, Carnegie Mellon UniVersity,
4400 Fifth AVenue, Pittsburgh, PennsylVania 15213, and ARC Centre of Excellence for
Free-Radical Chemistry and Biotechnology, Research School of Chemistry, Australian National
UniVersity, Canberra ACT 0200, Australia
Received March 28, 2008; E-mail: km3b@andrew.cmu.edu
Abstract: Equilibrium constants in Cu-based atom transfer radical polymerization (ATRP) were determined
for a wide range of ligands and initiators in acetonitrile at 22 °C. The ATRP equilibrium constants obtained
vary over 7 orders of magnitude and strongly depend on the ligand and initiator structures. The activities
of the CuI/ligand complexes are highest for tetradentate ligands, lower for tridentate ligands, and lowest
for bidentate ligands. Complexes with tripodal and bridged ligands (Me6TREN and bridged cyclam) tend to
be more active than those with the corresponding linear ligands. The equilibrium constants are largest for
tertiary alkyl halides and smallest for primary alkyl halides. The activities of alkyl bromides are several
times larger than those of the analogous alkyl chlorides. The equilibrium constants are largest for the nitrile
derivatives, followed by those for the benzyl derivatives and the corresponding esters. Other equilibrium
constants that are not readily measurable were extrapolated from the values for the reference ligands and
initiators. Excellent correlations of the equilibrium constants with the CuII/I redox potentials and the
carbon–halogen bond dissociation energies were observed.
Introduction
kdeact ) kact/KATRP if values of both kact and KATRP are known.
Therefore, determination of KATRP values is very important. The
Atom transfer radical polymerization (ATRP, Scheme 1) is
one of the most powerful and robust controlled radical polym-
erization (CRP) techniques.1–5 The rate of ATRP depends on
the value of the ATRP equilibrium constant (KATRP), i.e., the
ratio of the rate constants for activation and deactivation (eq
1). The polydispersity (Mw/Mn) of the polymers obtained
depends on the ratio of the propagation rate constant (kp) to the
deactivation rate constant (kdeact), the concentration of the
deactivator X-CuIIY/Ln (denoted as [CuII]), the concentration
of the initiator ([RX]), monomer conversion (p), and the targeted
degree of polymerization (DPn) (eq 2). The activation rate
constant (kact) has been extensively examined in the literature.2,6–23
Direct determination of kdeact is more difficult.6,7,24,25 On the
other hand, values of kdeact can be calculated from the equation
rate of polymerization of a given monomer (M) depends on
the value of kp and on the radical concentration ([Pm ·]), which
is determined by KATRP (eq 1). Thus, the evaluation of KATRP is
(10) Pintauer, T.; Zhou, P.; Matyjaszewski, K. J. Am. Chem. Soc. 2002,
124, 8196–8197.
(11) Nanda, A. K.; Matyjaszewski, K. Macromolecules 2003, 36, 599–
604.
(12) Nanda, A. K.; Matyjaszewski, K. Macromolecules 2003, 36, 1487–
1493.
(13) Nanda, A. K.; Matyjaszewski, K. Macromolecules 2003, 36, 8222–
8224.
(14) Pintauer, T.; Braunecker, W.; Collange, E.; Poli, R.; Matyjaszewski,
K. Macromolecules 2004, 37, 2679–2682.
(15) Matyjaszewski, K.; Nanda, A. K.; Tang, W. Macromolecules 2005,
38, 2015–2018.
(16) Tang, W.; Nanda, A. K.; Matyjaszewski, K. Macromol. Chem. Phys.
2005, 206, 1171–1177.
(17) Tang, W.; Matyjaszewski, K. Macromolecules 2006, 39, 4953–4959.
(18) Tang, W.; Matyjaszewski, K. Macromolecules 2007, 40, 1858–1863.
(19) Chambard, G.; Klumperman, B.; German, A. L. Macromolecules 2000,
33, 4417–4421.
(20) Goto, A.; Fukuda, T. Macromol. Rapid Commun. 1999, 20, 633–636.
(21) Ohno, K.; Goto, A.; Fukuda, T.; Xia, J.; Matyjaszewski, K. Macro-
molecules 1998, 31, 2699–2701.
(22) Schellekens, M. A. J.; de Wit, F.; Klumperman, B. Macromolecules
2001, 34, 7961–7966.
(23) Venkatesh, R.; Vergouwen, F.; Klumperman, B. Macromol. Chem.
Phys. 2005, 206, 547–552.
(24) Chambard, G.; Klumperman, B.; German, A. L. Macromolecules 2002,
35, 3420–3425.
(25) Gromada, J.; Matyjaszewski, K. Macromolecules 2002, 35, 6167–
6173.
† Carnegie Mellon University.
‡ Australian National University.
(1) Wang, J.-S.; Matyjaszewski, K. J. Am. Chem. Soc. 1995, 117, 5614–
5615.
(2) Matyjaszewski, K. J. Macromol. Sci., Pure Appl. Chem. 1997, A34,
1785–1801.
(3) Matyjaszewski, K.; Xia, J. Chem. ReV. 2001, 101, 2921–2990.
(4) Goto, A.; Fukuda, T. Prog. Polym. Sci. 2004, 29, 329–385.
(5) Matyjaszewski, K. Prog. Polym. Sci. 2005, 30, 858–875.
(6) Matyjaszewski, K.; Paik, H.-j.; Zhou, P.; Diamanti, S. J. Macromol-
ecules 2001, 34, 5125–5131.
(7) Matyjaszewski, K.; Goebelt, B.; Paik, H.-j.; Horwitz, C. P. Macro-
molecules 2001, 34, 430–440.
(8) Matyjaszewski, K.; Paik, H.-j.; Shipp, D. A.; Isobe, Y.; Okamoto, Y.
Macromolecules 2001, 34, 3127–3129.
(9) Matyjaszewski, K. Macromolecules 2002, 35, 6773–6781.
9
10702 J. AM. CHEM. SOC. 2008, 130, 10702–10713
10.1021/ja802290a CCC: $40.75
2008 American Chemical Society

Understanding Atom Transfer Radical Polymerization
A R T I C L E S
Scheme 1. Proposed Mechanism for Atom Transfer Radical
Polymerization
Equation 4 must be used when C0 ) I0:
2
C0
C0
1
F(Y) )
+
+
)
3(C0 - Y)3 (C0 - Y)2
C0 - Y
1
3C0
2ktKATRP2t +
(4)
crucial for deeper understanding of this catalytic system and for
optimal catalyst selection, especially for newly developed ATRP
systems that use low concentrations of the catalyst CuIY/Ln (i.e.,
[CuI] on the order of ppm).26,27 Small values of KATRP (∼10-9 to
∼10-4) are required in order to maintain a low radical concentra-
In either case, a plot of F(Y) versus t should give a straight
line, and the equilibrium constant for the reaction can be
calculated from the slope (m) using the formula KATRP ) (m/
2kt)1/2. In our calculations, a kt value of ) 2.5 × 109 M-1 s-1
(obtained for model small radicals in acetonitrile at 22 °C) was
used.33 UV-vis spectrometry and gas chromatography were
used to follow the evolution of the CuII species and the initiator
concentration, respectively, for several catalysts and alkyl halide
initiators, from which the corresponding values of KATRP were
determined.30
In this work, we present a large set of KATRP values
determined for ATRP using various alkyl halide initiators and
Cu catalysts with nitrogen-based ligands, and we discuss how
the structures of the ligands and initiators affect the KATRP
values. Values of KATRP that were not readily available were
extrapolated by assuming similar selectivities of alkyl halides
for various complexes.
tion and minimize termination reactions. The apparent KATRP
app
(K
) KATRP/[CuII]) can be estimated from the kinetic plot
ATRP
of ln([M]0/[M]) versus time for an ATRP experiment;28
however, precise values for KATRP are difficult to measure during
polymerization because CuII concentrations continuously change
as a result of the persistent radical effect (PRE), which is
described in the Results and Discussion.29
[M]0
[M]t
kpKATRP[PmX][CuI]
[CuII]
ln
)
t
(1)
(2)
(
)
M
kp([RX]0 - [RX])
w ) 1 +
+
- 1
1
DPn
2
p
(
)
kdeact[CuII]
(
)
Mn
Our group recently modified Fischer’s original equations for
the PRE and successfully applied the new equations to determine
KATRP values for various ATRP catalytic systems by monitoring
the amount of accumulated deactivator as a function of time.30,31
Fischer’s original equations, which assume constant values of
the catalyst and initiator concentrations, correctly describe the
PRE up to only ∼10% conversion of CuI to CuII, and
Experimental Section
Materials. N-(n-Propyl)pyridylmethanimine (NPPMI) and N-(n-
octyl)pyridylmethanimine (NOPMI) were synthesized by condensa-
tion of n-propylamine and n-octylamine, respectively, with pyridine-
2-carboxaldehyde.34 Tris(2-(dimethylamino)ethyl)amine (Me6TREN)
was synthesized by methylation of tris(2-aminoethyl)amine (TREN).35
Tris(2-(diethylamino)ethyl)amine (Et6TREN) was synthesized accord-
ing to the literature.36 Tris(2-(di-(2-cyanoethyl)amino)ethyl)amine
(AN6TREN), tris(2-(di-(2-(methoxycarbonyl)ethyl)amino)ethyl)amine
(MA6TREN), and tris(2-(di-(2-(n-butoxycarbonyl)ethyl)amino)ethyl)-
amine (BA6TREN) were synthesized by Michael addition of TREN
to acrylonitrile, methyl acrylate, and n-butyl acrylate, respectively.37
2,5,9,12-Tetramethyl-2,5,9,12-tetraazatridecane (N4[2,3,2]),38 2,6,9,13-
tetramethyl-2,6,9,13-tetraazatetradecane (N4[3,2,3]),38 4,11-dimethyl-
1,4,8,11-tetraazabicyclo[6.6.2]hexadecane (Cyclam-B),39,40 N′,N′′-
dimethyl-N′,N′′-bis((pyridin-2-yl)methyl)ethane-1,2-diamine (BPED),41
4,4′-di-(5-nonyl)-2,2′-bipyridine (dNbpy),42 and tris[(2-pyridyl)methyl]-
amine (TPMA)43 were synthesized according to the literature. 2,5,8,12-
Tetramethyl-2,5,8,12-tetraazatridecane (N4[2,2,3]) was synthesized by
a similar methylation of 2,5,8,12-tetraazatridecane tetrahydrochloride
salt with formaldehyde and formic acid and distilled under reduced
consequently, they can only be used in the determination of
32
relatively small values of KATRP
.
For greater conversion of
CuI to CuII (>10%) and larger values of KATRP (>10-7),
modified equations that take into account the changes in catalyst
and initiator concentrations should be used, as shown in eqs 3
and 4, in which I0 ) [RX]0, C0 ) [CuI]0, and Y ) [CuII].30
Equation 3 applies when C0 * I0:
2
I0C0
I0 - Y
C0 - Y
1
2
F(Y) )
+
ln
+
(
)
2
(
)
C0 - I0
I0C0(C0 - I0)
[
C0 (I0 - Y)
1
) 2ktKATRP2t + c′ (3)
2
]
I0 (C0 - Y)
where kt is the termination rate constant and c′ is given by the
(33) Fischer, H.; Radom, L. Angew. Chem., Int. Ed. 2001, 40, 1340–1371.
(34) Haddleton, D. M.; Clark, A. J.; Crossman, M. C.; Duncalf, D. J.;
Heming, A. M.; Morsley, S. R.; Shooter, A. J. Chem. Commun 1997,
1173.
(35) Queffelec, J.; Gaynor, S. G.; Matyjaszewski, K. Macromolecules 2000,
33, 8629–8639.
(36) Inoue, Y.; Matyjaszewski, K. Macromolecules 2004, 37, 4014–4021.
(37) Zeng, F.; Shen, Y.; Zhu, S.; Pelton, R. Macromolecules 2000, 33,
1628–1635.
expression
2
I0C0
I0
1
2
1
c′ )
+
ln
I0C0(C0 - I0) C0
+
(
)
2
2
C0 - I0
[
]
C0 I0
I0 C0
(26) Matyjaszewski, K.; Jakubowski, W.; Min, K.; Tang, W.; Huang, J.;
Braunecker, W. A.; Tsarevsky, N. V. Proc. Natl. Acad. Sci. U.S.A.
2006, 103, 15309–15314.
(38) Golub, G.; Cohen, H.; Paoletti, P.; Bencini, A.; Messori, L.; Bertini,
I.; Meyerstein, D. J. Am. Chem. Soc. 1995, 117, 8353–8361.
(39) Wong, E. H.; Weisman, G. R.; Hill, D. C.; Reed, D. P.; Rogers, M. E.;
Condon, J. S.; Fagan, M. A.; Calabrese, J. C.; Lam, K.-C.; Guzei,
I. A.; Rheingold, A. L. J. Am. Chem. Soc. 2000, 122, 10561–10572.
(40) Tsarevsky, N. V.; Braunecker, W. A.; Tang, W.; Brooks, S. J.;
Matyjaszewski, K.; Weisman, G. R.; Wong, E. H. J. Mol. Catal. A:
Chem. 2006, 257, 132–140.
(27) Jakubowski, W.; Matyjaszewski, K. Angew. Chem., Int. Ed. 2006, 45,
4482–4486.
(28) Qiu, J.; Matyjaszewski, K.; Thouin, L.; Amatore, C. Macromol. Chem.
Phys. 2000, 201, 1625–1631.
(29) Fischer, H. Chem. ReV. 2001, 101, 3581–3610.
(30) Tang, W.; Tsarevsky, N. V.; Matyjaszewski, K. J. Am. Chem. Soc.
2006, 128, 1598–1604.
(41) Xia, J.; Zhang, X.; Matyjaszewski, K. ACS Symp. Ser. 2000, 760, 207–
223.
(31) Tang, W.; Fukuda, T.; Matyjaszewski, K. Macromolecules 2006, 39,
4332–4337.
(42) Matyjaszewski, K.; Patten, T. E.; Xia, J. J. Am. Chem. Soc. 1997,
119, 674–680.
(32) Fischer, H. J. Polym. Sci., Part A: Polym. Chem. 1999, 37, 1885–
1901.
(43) Xia, J.; Matyjaszewski, K. Macromolecules 1999, 32, 2434–2437.
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10703

A R T I C L E S
Tang et al.
Figure 1. ATRP equilibrium constants KATRP for various N-based ligands with the initiator EtBriB in the presence of CuIBr in MeCN at 22 °C. Color key:
(red) N2; (black) N3 and N6; (blue) N4. Symbol key: (solid) amine/imine; (open) pyridine; (left-half-solid) mixed; (0) linear; (4) branched; (O) cyclic.
was estimated from the value for CuICl/Cyclam-B with MClAc as
follows: KATRP(CuIBr/Cyclam-B with EtBriB) ) KATRP(CuICl/
Cyclam-B with MClAc) × KATRP (CuIBr/TPMA with EtBriB)/
KATRP(CuICl/TPMA with MClAc) ) (9.9 × 10-5)(9.6 × 10-6)/
(4.2 × 10-7) ) 2.3 × 10-3
.
Cyclic Voltammetry. All of the voltammograms were recorded
at 25 °C with a Gamry Reference 600 potentiostat. Solutions of
CuIIBr2/L (1.0 mM) were prepared in dry solvent containing 0.1
M Bu4NPF6 as the supporting electrolyte. Measurements were
carried out under nitrogen at a scan rate of 0.1 V s-1 using a glassy
carbon disk as the working electrode and a platinum wire as the
counter electrode. Potentials were recorded versus SCE using a 0.1
M Bu4NPF6 salt bridge to minimize contamination of the analyte
with Cl- ions.
Calculation of Bond Dissociation Energies. R-X bond dis-
sociation energies (BDEs) for the alkyl halides in acetonitrile at
22 °C were taken directly from a recent study.45 The data in this
study were obtained via standard ab initio molecular orbital theory
calculations performed at a high level of theory chosen on the basis
of recent assessment studies for alkyl halide homolytic BDEs.46,47
Geometries and frequencies were obtained at the B3LYP/6-31+G(d)
level of theory, and improved energies were obtained using the
high-level composite method G3(MP2)-RAD(+). Partition functions
and associated thermodynamic functions (i.e., entropies and tem-
perature corrections) were evaluated using the TES 60 °C hindered-
Figure 2. Plot of KATRP (measured with EtBriB) vs E1/2 for 12 CuIIBr2/L
complexes.
pressure. N,N,N′,N′-Tetra[(2-pyridyl)methyl]ethylenediamine (TPEDA)
was synthesized by a reaction of benzyl chloride and ethylenedi-
amine.44 N-[2-(Dimethylamino)ethyl]-N,N′,N′-trimethyl-1,3-propanedi-
amine (N3[2,3]) was synthesized by methylation of N-(2-aminoethyl)-
1,3-propanediamine. All of the other ligands, initiators, and solvents
were used as received. Abbreviations for the ligands17 and initiators18
are listed in Figures 1 and 3.
Determination of Equilibrium Constants. Typical procedures
for determination of the equilibrium constants can be found in our
previous report.30 The KATRP values for exceptionally fast or slow
systems were estimated from those for slower or faster systems,
respectively. For example, KATRP for CuIBr/Cyclam-B with EtBriB
(45) Lin, C. Y.; Coote, M. L.; Gennaro, A.; Matyjaszewski, K. J. Am. Chem.
Soc., submitted for publication, 2008.
(46) Izgorodina, E. I.; Coote, M. L.; Radom, L. J. Phys. Chem. A 2005,
109, 7558–7566.
(47) Izgorodina, E. I.; Brittain, D. R. B.; Hodgson, J. L.; Krenske, E. H.;
Lin, C. Y.; Namazian, M.; Coote, M. L. J. Phys. Chem. A 2007, 111,
10754–10768.
(44) Tang, H.; Arulsamy, N.; Radosz, M.; Shen, Y.; Tsarevsky, N. V.;
Braunecker, W. A.; Tang, W.; Matyjaszewski, K. J. Am. Chem. Soc.
2006, 128, 16277–16285.
9
10704 J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008

Understanding Atom Transfer Radical Polymerization
A R T I C L E S
Table 1. CV Data for CuIIBr2/L Complexes in Acetonitrilea
rotor model,48 and free energies of solvation were evaluated using
the CPCM continuum model at the HF/6-31+G(d) level of theory
in conjunction with UAHF radii. All of the calculations were
performed using Gaussian 03.49
d
ligand
Ep,a (V)b
Ep,c (V)c
∆Ep (mV)
ib/if
E1/2 (V)e
Cyclam
-0.890
-
-1.000
-0.446
-0.361
-0.339
-0.276
-0.264
-0.210
-0.128
-0.119
-0.095
-0.097
-0.065
-0.037
-0.041
0.041
110
-
0.87
<0.01
0.89
0.89
0.95
0.94
0.90
0.90
0.92
0.99
0.84
0.67
0.97
0.76
0.91
1.00
-0.95
-
Me4Cyclam
Cyclam-B
Me6TREN
TPMA
-0.270
-0.258
-0.206
-0.160
-0.137
-0.026
-0.012
-0.006
0.058
0.083
0.084
0.191
0.129
91
-0.32
-0.30
-0.24
-0.21
-0.17
-0.08
-0.07
-0.05
-0.02
0.01
Results and Discussion
81
70
Effect of Ligand Structure on Equilibrium Constants. To
compare the activities of CuI complexes containing various
nitrogen-based ligands, most of the experiments were carried
out under the same conditions, i.e., in the same solvent
(acetonitrile), at the same temperature (22 °C), and with the
same initiator (EtBriB). Figure 1 shows the measured KATRP
values on a logarithmic scale. The activities of the catalysts
formed with these ligands range over 7 orders of magnitude.
It was previously reported that the catalysts become more
active when the CuII state of the catalyst is better stabilized by
the ligand.7,41,50 In general, the activities of complexes with
various ligands decrease in the following order: alkyl amine ≈
pyridine > alkyl imine . aryl imine > aryl amine. They also
decrease as the number of nitrogens in the ligand decreases.7,41
The present results generally confirm such observations, but
some other factors, including steric and electronic effects and
bite angle, are also important and are discussed in the next
section.51,52
BPED
104
73
102
107
89
155
148
121
232
88
TPEDA
PMDETA
BPMPA
dNbpy
HMTETA
N4[2,2,3]
bpy
N4[2,3,2]
N4[3,2,3]
ferrocene
0.02
0.08
0.09
0.42
0.458
0.388
70
a Conditions: 0.1 M NBu4PF6, 1.0 mM CuIIBr2/L complex, scan rate
0.10 V s-1. Potentials are reported vs SCE. b Peak potential of the
oxidation wave. c Peak potential of the reduction wave. d back-
ward-to-forward peak current ratio. e E1/2 ) (Ep,a + Ep,c)/2.
is therefore a more accurate measure of catalyst activity than
the rate of catalyzed polymerization.
Table 1 summarizes cyclic voltammetry (CV) data recorded
for 15 ATRP catalysts in acetonitrile. The majority of the
complexes investigated showed good (quasi)reversibility under
these conditions, as judged by their relatively small ∆Ep values
and their backward-to-forward peak current ratios (ib/if), which
had values close to unity. Figure 2 shows the correlation of the
logarithmic values of KATRP with measured E1/2 values from
the more-reversible voltammograms of 12 catalysts (those for
which ib/if > 0.75).
The excellent correlation between ATRP activities and values
of E1/2 for a wide range of complexes, as observed in Figure 2,
warrants an in-depth discussion of how various ligand mor-
phologies and substituents affect the CuII/I redox potentials and
thereby govern catalyst activities. While parametrization meth-
ods have been successfully employed to predict redox potentials
among complexes having the same metal, stereochemistry, and
oxidation and spin states,58 such parametrization techniques are
inadequate for predicting the activity of most Cu-containing
ATRP catalysts because the application of multidentate ligands
introduces varying degrees of distortion from ideal or preferred
geometries. Nevertheless, some general rules still apply.50 For
example, the introduction of electron-withdrawing groups on
the ligand results in less-reducing complexes while electron-
donating groups increase the reducing power of a complex.
Specific examples and ligand morphologies illustrated in Figure
2 are discussed in the following paragraphs.
Linear Tetradentate Ligands. Four linear aliphatic tetradentate
ligands having various ethylene and propylene linking units were
studied. N4[2,2,2] and N4[2,2,3] form relatively active catalysts
having KATRP values of 1.1 × 10-8 and 1.5 × 10-7, respectively,
but N4[2,3,2] and N4[3,2,3] form less-active catalysts having KATRP
values of 4.2 × 10-10 and 3.2 × 10-9, respectively. The order of
reducing power measured in acetonitrile is consistent with literature
values recorded in water, i.e., the Cu complex with N4[2,2,2]
(HMTETA) is more reducing (and more active) than the complex
with N4[2,3,2], which is slightly more reducing than the one with
N4[3,2,3].38 While these ligands all have essentially the same
basicity and electronic properties, their redox potentials vary by
Correlation of Catalyst Structure with ATRP Activity and
Electrochemistry. As the ATRP equilibrium is governed by a
redox process between the higher and lower oxidation states of
a transition-metal catalyst, electrochemistry can be a useful tool
for understanding and predicting catalytic activity.50,53 The
general trend that more reducing complexes are better catalysts
(i.e., yield greater rates of polymerization) has been observed
among complexes of the same metal for a wide range of
Cu,6,28,30,40,54 Fe,55,56 and Ru-based catalysts.57 In fact, a linear
app
ATRP
correlation between the logarithm of K
(determined from
rates of catalyzed polymerization) and measured E1/2 values for
Cu complexes of Me6TREN, TPMA, PMDETA, and derivatives
of bpy and BPMODA (complexes also used in this study) has
been observed.28 However, exceptionally active catalysts can
generate a large number of radicals at the early stages of
polymerization that terminate before the system reaches equi-
librium. Since these termination reactions result in the irrevers-
ible accumulation of the ATRP deactivator (a phenomenon
termed the persistent radical effect), it is possible that more
active ATRP catalysts actually give rise to slower polymeriza-
tions and/or reactions that stop after relatively low monomer
conversion.30 The KATRP value, as quantified via model reactions,
(48) Lin, C. Y.; Izgorodina, E. I.; Coote, M. L. J. Phys. Chem. A 2008,
112, 1956–1964.
(49) Frisch, M. J.; et al. Gaussian 03, revision B.03; Gaussian, Inc.:
Wallingford, CT, 2003.
(50) Tsarevsky, N. V.; Matyjaszewski, K. Chem. ReV. 2007, 107, 2270–
2299.
(51) Levy, A. T.; Patten, T. E. Polym. Mater. Sci. Eng. 1999, 80, 467–
468.
(52) Pintauer, T.; Matyjaszewski, K. Coord. Chem. ReV. 2005, 249, 1155–
1184.
(53) Tsarevsky, N. V.; Tang, W.; Brooks, S. J.; Matyjaszewski, K. ACS
Symp. Ser. 2006, 944, 56.
(54) Tsarevsky, N. V.; Braunecker, W. A.; Brooks, S. J.; Matyjaszewski,
K. Macromolecules 2006, 39, 6817–6824.
(55) O’Reilly, R. K.; Gibson, V. C.; White, A. J. P.; Williams, D. J. J. Am.
Chem. Soc. 2003, 125, 8450–8451.
(56) O’Reilly, R. K.; Gibson, V. C.; White, A. J. P.; Williams, D. J.
Polyhedron 2004, 23, 2921–2928.
(57) Ando, T.; Kamigaito, M.; Sawamoto, M. Macromolecules 2000, 33,
5825–5829.
(58) Lever, A. B. P. Inorg. Chem. 1990, 29, 1271–1285.
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10705

A R T I C L E S
Tang et al.
Figure 3. ATRP equilibrium constants for various initiators with CuIX/TPMA (X ) Br, Cl) in MeCN at 22 °C. Color key: (red) 3°; (blue) 2°; (black) 1°.
Symbol key: (solid) R-Br; (open) R-Cl; (bottom-half-solid) R-I; (4) phenyl; (0) ester; (O) nitrile; (]) phenyl ester; (g) allyl.
110 mV. The main difference between these ligands is the relative
location of the propylene linker, which increases the preferred
chelating angle from 70 to 109° and, by inducing additional
reducing strength, decreases the stability of the CuII complex that
determines KATRP. This shows how subtle differences in catalyst
structure can have dramatic effects on the relative stabilities of the
CuI and CuII oxidation states.
Replacing the two aliphatic amines on the ends of HMTETA
with aromatic pyridyl groups (to form BPED) results in a ligand
whose copper bromide complex is 190 mV more reducing and
precisely 3 orders of magnitude more active in ATRP (KATRP
) 1.1 × 10-5 for BPED). The relative reducing power of the
BPED complex is surprising, given that substituting aliphatic
nitrogens with aromatic ones typically results in CuI compounds
that are much less reducing.59 However, this anomaly can be
partially rationalized when the structure of the analogous
nonmethylated CuII/BPED complex discussed in the literature
is considered. Spectroscopic evidence indicates that the CuII
structure is quite planar and more rigid than those of some
analogous compounds. This effectively results in a relative
increase in the stability of the CuII oxidation state over the CuI
state, as the latter is less able to accommodate a planar structure,
preferring instead a tetrahedral geometry.60
10-5). Furthermore, it has been observed that by progressively
increasing the electron-donating power of the para substituents
on the pyridyl rings of TPMA (through the use of p-H-, p-t-
Bu-, p-MeO-, and p-Me2N-substituted ligands), one can increase
the reducing power of the Cu complexes by 300 mV (corre-
sponding to an increase of 5 orders of magnitude in the
equilibrium constant for electron transfer).61 In a similar fashion,
the electron-withdrawing power of TREN-based ligands was
progressively increased in the present study by using -Me,
-(CH2)2CO2tBu, -(CH2)2CO2Me, and -(CH2)2CN substituents
on the ethylamine nitrogens. As expected, the values of KATRP
for these ligands decreased (spanning 5 orders of magnitude)
as the electron-withdrawing power of the ligand and steric
effects increased.
Macrocyclic Ligands. The stability constants for CuII com-
plexes of macrocyclic ligands such as Cyclam and its derivatives
are quite large (∼1027 in water).62 In addition, these relatively
rigid ligands form planar complexes that do not stabilize the
tetrahedral CuI oxidation state very well. As a consequence of
these two facts, CuI complexes with Cyclam-type ligands tend
to form exceptionally reducing and active catalysts in ATRP.
Ideally, the effect on catalyst activity of substituting -H with
-Me among complexes of Cyclam and Me4Cyclam could be
studied. Unfortunately, because CuIBr/Cyclam is so exception-
ally reducing, the complex was far too active to accurately
determine KATRP, even when alkyl halide initiators with low
activities were used. Furthermore, the voltammogram for the
complex with Me4Cyclam was not reversible, making a direct
comparison of the redox potentials of the two complexes
impossible. The activities and reducing power of the catalysts
could still be compared, however, by employing the trend line
Tripodal/Branched Tetradentate Ligands. Cu complexes with
Me6TREN form some of the most active (KATRP ) 1.5 × 10-4
)
and most reducing catalysts successfully employed in ATRP.
They were also successfully used in the new ATRP systems
that employ only parts per million levels of Cu catalyst.26 In a
manner that is qualitatively consistent with literature predic-
tions,59 replacing the aliphatic amine nitrogens of ME6TREN
with aromatic ones (to form TPMA) results in a Cu complex
that is slightly less reducing and less active (KATRP ) 1.0 ×
(61) Zhang, C. X.; Kaderli, S.; Costas, M.; Kim, E.-i.; Neuhold, Y.-M.;
Karlin, K. D.; Zuberbu¨hler, A. D. Inorg. Chem. 2003, 42, 1807–1824.
(62) Sun, X.; Wuest, M.; Weisman, G. R.; Wong, E. H.; Reed, D. P.;
Boswell, C. A.; Motekaitis, R.; Martell, A. E.; Welch, M. J.; Anderson,
C. J. J. Med. Chem. 2002, 45, 469–477.
(59) Addison, A. W. Inorg. Chim. Acta 1989, 162, 217–220.
(60) Zanello, P. In Stereochemical Control, Bonding and Steric Rearrange-
ments; Bernal, I., Ed.; Elsevier: Amsterdam, 1990; Vol. 4, pp 181-
366.
9
10706 J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008

Understanding Atom Transfer Radical Polymerization
A R T I C L E S
Scheme 2. Atom-Transfer Equilibrium Represented as a
generated from the data in Figure 2. In this manner, one can
see that Cyclam complex (E1/2 ≈ -0.95 V vs SCE) should be
>10 orders of magnitude more active than the complex with
Me4Cyclam (KATRP ) 4.7 × 10-5), consistent with the trend
observed in the literature that substituting -H with -Me yields
complexes that are significantly less reducing. This has been
seen not only for macrocyclic ligands but also for linear and
branched aliphatic amines. This observation contradicts the
“rule” that increasing the electron-donating power of a sub-
stituent results in a more-reducing complex and cannot be
attributed merely to a distortion of preferred geometries, as
analogous results have been observed for a wide variety of
metals preferring a wide variety of geometries. It was, however,
concluded that N-alkylation effectively induces a strain on the
ligand skeleton that is naturally smaller for the CuI complexes
than for the CuII complexes, i.e., N-alkylation of ligands
increases their binding constants for large cations but decreases
their binding constants for small cations.38
Cyclam-B has a special bridged cyclic structure in which an
ethylene group cross-bridges two diagonal nitrogens. The CuI/
Cyclam-B catalyst is by far the most active ATRP catalyst: it
can successfully mediate ATRP with a KATRP value ∼13 times
larger than that of Me6TREN.40 The ligand TPEDA was the
only hexadentate ligand studied here. The CuIBr complex
formed by this ligand is fairly active (25 times more active than
CuIBr/PMDETA but ∼75 times less active than CuIBr/
Me6TREN).44
Combination of Alkyl Halide Bond Homolysis and CuII-X Bond
Formation (Halogenophilicity)
TPMA. Using CuIX/TPMA as the catalyst afforded a fairly fast
reaction for most of the initiators. As shown in Figure 3, the
activity ratio among these initiators exceeds 105. The ATRP
equilibrium constant KATRP can be defined as the product of
the equilibrium constant for homolytic bond dissociation of
R-X and that for formation of CuII-X from CuI and X• (i.e.,
the halogenophilicity) (see Scheme 2). Thus, KATRP values
should be correlated with the BDEs of the initiators. Unfortu-
nately, there are very few experimental data available on BDEs
of relevant alkyl halides, and the data exhibit significant
scatter.64 Therefore, we will first discuss the phenomenological
dependence of the KATRP values on the most important structural
parameters. Subsequently, we will compare the computationally
accessible free energies of C-X bond dissociation with those
estimated from KATRP measurements assuming constant halo-
genophilicity values for various alkyl halides.
Effect of the Leaving Atom. Alkyl bromides are several times
more active than the corresponding alkyl chlorides: for example,
EtBriB (6) > EtCliB (1), BrPN (5) > ClPN (1), PEBr (5) >
PECl (1), and MBrP (8) > MClP (1), where the numbers in
parentheses are the relative KATRP values. The difference in the
KATRP values for EtBriB and EtCliB facilitates the halogen-
exchange technique required to synthesize well-defined (meth)-
acrylic block copolymers.65 Although kact for MIP is close to
that for MBrP, KATRP for MIP is ∼15 times smaller than that
for MBrP. This is a consequence of the very low stability of
the CuII-I bond.
Tertiary, Secondary, And Primary Alkyl Halides. The kact
values for tertiary, secondary, and primary ester initiators follow
the order 3° > 2° > 1°.18 The order of the KATRP values is
similar, but there are some differences. Values of KATRP are
∼30 times larger for tertiary haloisobutyrate esters than for
secondary halopropionates. However, the difference between
the values for secondary and primary species is smaller. For
example, KATRP for BrPN is ∼2 times larger than that for BrAN,
and the same behavior is observed for MBrP and MBrAc as
well as for PECl and BzCl. The difference for benzyl bromides
is larger, however. It is possible that values of KATRP for alkyl
halides that form very electrophilic primary radicals can be
overestimated. The evolution of CuII species may be due not
Bi- and Tridentate Ligands. CuIBr complexes with two bid-
entate pyridine imine ligands are some of the least active ATRP
catalysts (KATRP ) 5.0 × 10-11 for both NPPMI and NOPMI).
This observation is consistent with the positive redox potentials
reported in the literature for structurally related complexes.60
Catalysts employing bipyridine ligands, the first developed for
Cu-based ATRP, are more active than the pyridine imines
(KATRP ) 3.9 × 10-9 for bpy). Long alkyl groups were
incorporated onto the pyridine rings in order to increase the
solubility of the catalyst in less polar solvents. However, these
alkyl substituents also increased the catalytic activity by an order
of magnitude (KATRP ) 3.0 × 10-8 for dNbpy), as the electron-
donating groups made the catalyst more reducing.
CuIBr complexes with two different tridentate ligands,
PMDETA and BPMPA, were also investigated. CuIBr/PM-
DETA (KATRP ) 7.5 × 10-8) was slightly more active and more
reducing than the BPMPA complex bearing two pyridylmethyl
groups. However, it should be noted that the complexes of these
ligands with CuIX are atypical of traditional CuI-based catalysts
in that the halide anion also acts as a ligand rather than just a
counterion. Under polymerization conditions, where a substantial
amount of vinyl monomer coordinates to CuI/PMDETA by
displacing the halide ligand,63 the redox potentials and hence
the values of KATRP may not accurately reflect the true
polymerization activity of the catalyst. When the two amino
groups in PMDETA were replaced with pyridyl groups, the
only to radical coupling (which is directly correlated with KATRP
)
but also to reduction of the radicals to carbanions by CuI species
via an outer-sphere electron transfer (OSET) process.
resulting BPMPA catalyst had a KATRP value of 6.2 × 10-8
,
Stabilization of Radicals by Substituents. Within a given
initiator type, initiators with a nitrile substituent attached to the
carbon atom of the C-X bond are more active than those with
phenyl groups, which are in turn more active than initiators with
similar to that for PMDETA.
Effect of Initiator Structure on Equilibrium Constants.
Similar methodology was employed to investigate the effect of
initiator structure on KATRP in ATRP. Most of the bromine- and
chlorine-based initiators as well as the iodine-based initiators
for Cu-based ATRP were studied with the same catalyst, CuIX/
(64) Gillies, M. B.; Matyjaszewski, K.; Norrby, P.-O.; Pintauer, T.; Poli,
R.; Richard, P. Macromolecules 2003, 36, 8551–8559.
(65) Matyjaszewski, K.; Shipp, D. A.; McMurtry, G. P.; Gaynor, S. G.;
Pakula, T. J. Polym. Sci., Part A: Polym. Chem. 2000, 38, 2023–
2031.
(63) Braunecker, W. A.; Tsarevsky, N. V.; Pintauer, T.; Gil, R. R.;
Matyjaszewski, K. Macromolecules 2005, 38, 4081–4088.
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10707

A R T I C L E S
Tang et al.
ester groups. The orders of the KATRP values with CuIBr/TPMA
for two types of bromide initiators were BrPN (340) > PEBr
(14) > MBrP (1) and BrAN (175) . BzBr (2) g MBrAc (1).
A similar order was also found for the chlorides. The differences
in the equilibrium constants can be explained in terms of the
competing effects of the substituents on the stability of the alkyl
halide bond, which increases with the electron-donating ability
of the substituent (i.e., cyano < ester < phenyl), and their effects
on the stabilities of the alkyl radical formed (ester < cyano <
phenyl).66,67
MBriB and EtBriB have similar values of KATRP, while tBBrP
is ∼8 times less active than MBrP. This could be attributed to
the fact that the tert-butyl group is more bulky than a methyl
group and introduces more steric congestion to the sp2 hybrid-
ized radical than to the alkyl bromide.
The most active initiator is EBPA (a mixed benzyl-ester
initiator), which is ∼20,000 times more active than PEBr and
300,000 times more active than MBrP. This is due to the
synergistic stabilization effect of the phenyl and carboxyethyl
groups. EBPA is an efficient initiator for monomers with large
equilibrium constants, such as acrylonitrile and methyl methacry-
late.26,27
As illustrated in Figure 3, the ratio of the highest and lowest
equilibrium constants for the alkyl halides in this study exceeds
105. These values were mostly measured by using CuIBr/TPMA
as the catalyst in MeCN at 22 °C. The equilibrium constants
increase with (a) initiator substitution [1° (black) < 2° (blue)
< 3° (red)] and (b) the presence of radical-stabilizing groups for
secondary initiators [-COOR (squares) < -Ph (triangles) ,
-CN (circles) , both -Ph and R-COOEt groups (diamonds)];
KATRP also depends on (c) the leaving atom/group [e.g., for
methyl 2-halopropionates, iodide (bottom-half-solid) < chloride
(open) < bromide (filled)].
nization of the transition-metal complex, including an expansion
of the coordination sphere, a change in the oxidation state, and
any related changes in its solvation state (including any specific
•
solvation as well). On the other hand, KCu,X can be calculated
•
from KATRP and KRX via the expression KCu,X ) KATRP/KRX. It
•
should be noted that the value of KCu,X depends on the type of
ligand (L in CuIY/L) and the type of halogen (X).
•
Values of KCu,X were calculated as the ratio KATRP/KRX for each
catalyst using EtBriB and EtCliB as reference compounds, since
both haloisobutyrates were used in reactions with most catalysts.
•
The resulting values of KCu,X are listed in Table 2. These values
are very large and are comparable to the reciprocal values of KRX
,
since their product gives a much smaller value of KATRP. The
halogenophilicities are larger for more-active catalysts and larger
•
for chlorine than for bromine. Values of log KCu,Br range from
•
27.1 (NPPMI) to 33.6 (Me6TREN) and values of log KCu,Cl from
38.5 (bpy) to 43.1 (Me6TREN). Once the values of the halogeno-
philicities for various catalysts are known, they can be used together
with values of KATRP to calculate bond-dissociation equilibrium
•
constants (KRX )KATRP/KCu,X ). These values can then be compared
with the ab initio computed values of KRX. It should be noted that
while the resulting experimental values of KRX for the alkyl
chlorides and alkyl bromides are, by necessity, effectively scaled
by the respective ab initio values for EtCliB and EtBriB, their
relative values are completely independent of the theoretical
calculations. Hence, we can use the experimental KRX values to
assess the ability of the theoretical data to model the wide variation
in KRX and hence in KATRP
.
•
Figure 4 presents a plot of log KATRP - log KCu,X versus log
KRX, in which the green dashed line is the line with slope 1
that passes through the values for the two standards EtBriB and
EtCliB. In general, there is an excellent correlation between
the equilibrium constants obtained via the calculated BDEs and
those obtained experimentally. In most cases, the scatter is
within or close to the expected order-of-magnitude uncertainty
in the theoretical calculations; however, there are some outliers.
In particular, points above the line represent experimental values
of KATRP that are larger than expected, plausibly as a result of
the concurrent reduction of radicals to carbanions via OSET
(e.g., entries 10 and 16 for MBrAc and 25 and 30 for MClAc).
It should also be noted that Figure 4 was constructed assuming
constant selectivities of initiators and catalysts, as will be
discussed in the next section.
Bond Dissociation Energies and Values of KATRP. As discussed
above, experimental data on BDEs for alkyl halides relevant to
ATRP are limited. However, both enthalpies and free energies
of bond dissociation can be accessed computationally using
density functional theory (DFT)64 or ab initio45 procedures.
BDEs should correlate well with values of KATRP, since the
ATRP equilibrium can be represented formally as a combination
of two hypothetical reactions: (i) homolytic cleavage of the C-X
bond in the alkyl halide and (ii) formation of the X-CuII bond
in the deactivator via the reaction of CuI with the halogen
atom.45
Overview of Different Structural Parameters Affecting the
Equilibrium Constants. The structural effects of ligand and
initiator on KATRP reported herein correlate very well with previ-
ously reported trends for kact.17,18 Most values of KATRP were
obtained using EtBriB as a standard initiator for screening different
catalysts and TPMA as a standard ligand for screening various
alkyl halides. It is tempting to predict values of KATRP for other
systems by assuming constant selectivities of alkyl halides toward
various ligands and also constant selectivities of catalysts toward
various initiators. In this way, one can extrapolate (or interpolate)
the KATRP values to ligands with initiators other than EtBriB and
initiators with ligands other than TPMA. The same strategy was
previously employed for extrapolating kact values for various
catalysts and initiators.18 The constant selectivity principle, i.e., the
assumption that CuIBr/ligand has the same selectivity as CuIBr/
TPMA toward more-active and less-active initiators, was verified
for several systems, as shown in Table 3.
As shown in Scheme 2, the first process is characterized by
the equilibrium constant KRX, which is related to the free energy
of C-X bond dissociation (∆G), which can be accessed
computationally not only in vacuum but also in a solvent.
Computed ∆G values and the corresponding values of log KRX
for alkyl halides at 295.15 K (22 °C) in MeCN are listed in
Table 2. These values were obtained from the high-level ab
initio calculations described in a related study.45 The second
•
process is characterized by the equilibrium constant KCu,X
,
which is called the halogenophilicity and indicates the affinity
of the transition metal in its lower oxidation state for the halogen
•
atom. Unfortunately, the value of KCu,X , which is the reciprocal
of the equilibrium constant for homolytic bond dissociation of
the X-CuII species, cannot be directly measured, and its
computation is complicated since the process involves reorga-
(66) Coote, M. L.; Henry, D. J. Macromolecules 2005, 38, 1415–1433.
(67) Krenske, E. H.; Izgorodina, E. I.; Coote, M. L. ACS Symp. Ser. 2006,
944, 406–420.
(68) Jakubowski, W.; Tsarevsky, N. V.; Higashihara, T.; Faust, R.;
Matyjaszewski, K. Macromolecules 2008, 41, 2318–2323.
9
10708 J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008

Understanding Atom Transfer Radical Polymerization
A R T I C L E S
•
Table 2. Values of Free Energies (∆G) and Equilibrium Constants (KRX) for Bond Homolysis, Halogenophilicities (KCu,X ), and ATRP
Equilibrium Constants (KATRP) in Acetonitrile at 22 °C
a
b
c
•d
∆G (kcal mol-1
)
log KRX
log KATRP
log KCu,X
log KATRP - log KCu,X
•
ligand
initiator
NPPMI
NPPMI
EtBriB
EBPA
50.49
45.21
-37.38
-33.47
-10.3
-6.32
27.08
27.08
-37.38
-33.40
bpy
bpy
EtBriB
PEBr
50.49
52.61
-37.38
-38.95
-8.41
-9
28.97
28.97
-37.38
-37.97
HMTETA
HMTETA
HMTETA
HMTETA
EtBriB
BrAN
BrPN
PEBr
50.49
50.75
48.31
52.61
-37.38
-37.57
-35.77
-38.95
-7.96
-8.54
-7.1
29.42
29.42
29.42
29.42
-37.38
-37.96
-36.52
-37.96
-8.54
PMDETA
PMDETA
PMDETA
PMDETA
PMDETA
EtBriB
BrAN
BrPN
MBrP
PEBr
50.49
50.75
48.31
52.70
52.61
-37.38
-37.57
-35.77
-39.02
-38.95
-7.12
-5.06
-6.23
-8.40
-7.48
30.26
30.26
30.26
30.26
30.26
-37.38
-35.32
-36.49
-38.66
-37.74
TPMA
TPMA
TPMA
TPMA
TPMA
TPMA
TPMA
EtBriB
BrAN
BrPN
MBrAc
MBrP
BzBr
50.49
50.75
48.31
54.86
52.70
52.75
52.61
-37.38
-37.57
-35.77
-40.62
-39.02
-39.06
-38.95
-5.02
-4.25
-3.96
-6.49
-6.49
-6.17
-5.34
32.36
32.36
32.36
32.36
32.36
32.36
32.36
-37.38
-36.61
-36.32
-38.85
-38.85
-38.53
-37.70
PEBr
Me6TREN
Me6TREN
Me6TREN
Me6TREN
Me6TREN
EtBriB
MBrAc
MBrP
BzBr
50.49
54.86
52.70
52.75
52.61
-37.38
-40.62
-39.02
-39.06
-38.95
-3.82
-4.66
-5.54
-6.08
-3.59
33.56
33.56
33.56
33.56
33.56
-37.38
-38.22
-39.10
-39.64
-37.15
PEBr
bpy
bpy
EtCliB
ECPA
64.49
58.76
-47.75
-43.51
-9.22
-6.48
38.53
38.53
-47.75
-45.01
TPMA
TPMA
TPMA
TPMA
TPMA
TPMA
TPMA
EtCliB
ClAN
ClPN
MClAc
MClP
BzCl
64.49
63.60
61.71
66.18
66.44
65.88
65.01
-47.75
-47.09
-45.69
-49.00
-49.19
-48.78
-48.14
-5.83
-4.51
-4.66
-6.38
-7.38
-6.35
-6.07
41.92
41.92
41.92
41.92
41.92
41.92
41.92
-47.75
-46.43
-46.58
-48.30
-49.30
-48.27
-47.99
PECl
Me6TREN
Me6TREN
Me6TREN
Me6TREN
Me6TREN
EtCliB
MClAc
MClP
BzCl
64.49
66.18
66.44
65.88
65.01
-47.75
-49.00
-49.19
-48.78
-48.14
-4.64
-5.46
-5.7
-5.44
-4.85
43.11
43.11
43.11
43.11
43.11
-47.75
-48.57
-48.81
-48.55
-47.96
PECl
a Values taken from ref 45. b Calculated from corresponding calculated values of ∆G. c Experimentally determined. d Calculated for each catalyst
•
using EtBriB and/or EtCliB as the reference compound according to the equation log KCu,X ) log KATRP - log KRX
.
Most of the extrapolated values are close to the measured
values (within a factor of 2). It is possible that some additional
special effects can influence the selectivities of catalysts or
initiators. For example, steric effects may give a special
contribution for tertiary species. Some functional groups (such
as amide or unsaturation) may interact with the catalyst. Alkyl
halides with strong electron affinities may directly undergo
OSET, and the resulting radicals may be reduced to anions.7
An independent project to investigate the selectivities of ligands
and initiators is underway. Nevertheless, the extrapolated values
of KATRP should be useful for a rough estimation of the activity
of a particular catalyst/initiator system.
The rate of ATRP depends strongly on the value of KATRP (eq
1). However, control of polymerization (e.g., polydispersity)
depends on the values of deactivation rate constant kdeact (eq 2).
Direct measurements of kdeact are difficult, since the rate constants
are very large, approaching diffusion-controlled values. They were
previously determined for model compounds using “clock” reac-
tions6 and for macromolecular species from initial degrees of
polymerization25 or evolution of polydispersity with conversion.24
However, the kdeact values could be calculated directly from the
ratio kact/KATRP. We have previously reported kact values for a wide
range of ligands and catalysts at 35 °C in acetonitrile. In this work,
we have determined the ATRP equilibrium constants for a similar
range of initiators and catalysts at 22 °C. Values of kact show a
moderate dependence on temperature. As shown in Table 4, values
of kact increase on average by a factor of 2 as the temperature
increases from 22 to 35 °C.
We realize that activation energies for kact can be affected by
the structure of ligands and initiators; however, for simplicity, we
use the average value of the correcting ratio (0.5) to convert the
values of kact from 35 to 22 °C. Values of KATRP, kact, and kdeact in
acetonitrile at 22 °C calculated in this way are listed in Table 5.
Values in red are extrapolated values, which were discussed in
previous reports17,18 and also in this study. Some of the KATRP and
kdeact values are not yet available because the reactions are either
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10709

A R T I C L E S
Tang et al.
Figure 4. Correlation of KRX values calculated from KATRP using constant halogenophilicities for various complexes with values calculated from free
energies of homolytic bond dissociation for alkyl halides. Values of log KRX correspond to calculated values of ∆G,45 values of log KATRP were experimentally
•
determined, and values of log KCu,X were calculated for each catalyst using EtBriB and/or EtCliB as the reference compound according to the equation log
•
KCu,X ) log KATRP - log KRX
.
Table 3. Comparison of Extrapolated and Measured KATRP Values
KATRP ratio
Table 4. Comparison of Activation Rate Constant Values at 22
and 35 °C
no.
ligand
initiator KATRP (measured)a KATRP (extrap.)b (extrap./measured)
no.
ligand
initiator
kact (22 °C)
kact (35 °C)
kact ratio (35 °C/25 °C)
1
2
3
4
5
6
7
8
9
Me6TREN MClAc
Me6TREN BzCl
Me6TREN PECl
Me6TREN MClP
3.5 × 10-6
3.6 × 10-6
1.4 × 10-5
2.0 × 10-6
6.6 × 10-6
7.0 × 10-6
1.3 × 10-5
5.9 × 10-7
5.0 × 10-6
1.1 × 10-5
6.6 × 10-7
7.2 × 10-5
5.0 × 10-6
1.9 × 10-9
7.0 × 10-9
3.6 × 10-8
3.0 × 10-9
8.6 × 10-7
5.3 × 10-9
9.9 × 10-9
6.4 × 10-8
1.3 × 10-7
1.88
1.95
0.96
0.30
0.23
0.39
0.78
0.28
1.72
1.90
2.33
1.10
0.75
1.46
1.83
1.25
0.84
4.06
1
2
3
4
5
NPPMI
bpy
bpy
HMTETA EtBriB 0.06
PMDETA MBrP 0.17
EtBriB 1.2 × 10-3 2.4 × 10-3
2.0
1.6
2.2
2.3
1.9
EtBriB 0.04
0.066
PEBr
4.5 × 10-3 0.010
0.14
0.33
Me6TREN MBrAc 2.2 × 10-5
Me6TREN BzBr
Me6TREN tBBrP
Me6TREN PEBr
Me6TREN MBrP
2.8 × 10-5
8.4 × 10-7
2.6 × 10-4
2.9 × 10-6
1.0 × 10-9
3.0 × 10-9
3.3 × 10-8
4.0 × 10-9
5.9 × 10-7
2.9 × 10-9
7.9 × 10-10
7.9 × 10-8
3.2 × 10-8
monomers and initiators. Nevertheless, all of the extrapolated values
should be taken with a large degree of caution because of the
assumption of constant selectivity as well as the neglect of
variations in the activation energy. Larger errors might happen for
some very active systems because of the extrapolation over several
orders of magnitude; for example, the extrapolated ratio of KATRP
for EBPA and ECPA with TPMA ligand was ∼102, but direct
measurements of KATRP for several alkyl bromides and chlorides
gave a ratio e10.
10 bpy
11 bpy
PEBr
AllBr
12 PMDETA PEBr
13 PMDETA MBrP
14 PMDETA BrPN
15 HMTETA PEBr
16 HMTETA PECl
17 HMTETA BrAN
18 HMTETA BrPN
a Values were measured at 22 °C in MeCN.30,44,54,68 b Values were
extrapolated using the following calculation: KATRP(CuI/ligand with
initiator) ) KATRP (CuIBr/ligand with EtBriB) × KATRP(CuI/TPMA with
initiator)/(CuIBr/TPMA with EtBriB) (see Table 5).
It should be noted that both thermodynamic and kinetic
parameters depend on temperature and solvent. Detailed investiga-
tions of temperature and solvent effects on kact, kdeact, and KATRP
are currently being carried out in our laboratories.
too slow or too fast to carry out at present. The values extrapolated
from the reference catalyst and initiators are expected to be close
to the experimental data, as demonstrated for some extrapolated
and measured values of kact18 and KATRP. Thus, the table not only
enables a fast comparison of activities for catalysts and initiators
in ATRP but also provides good direction for future development
of new ligands and identifies limitations for the use of new
Summary of Structural Effects in Catalyst and Initiator
on kact, kdeact and KATRP. This final section attempts to demonstrate
how the structure of the ligands complexing to Cu, together with
the initiators used, affects ATRP in terms of kact, kdeact, and KATRP
.
Figure 5 shows that ATRP equilibrium constants increase as a
result of both increases in kact and decreases in kdeact and that kact
9
10710 J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008

Understanding Atom Transfer Radical Polymerization
A R T I C L E S
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10711

A R T I C L E S
Tang et al.
9
10712 J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008

Understanding Atom Transfer Radical Polymerization
A R T I C L E S
Figure 7. Correlation of kact and kdeact with KATRP for CuI/TPMA with
various initiators at 22 °C in MeCN (the broken lines are just to guide the
reader’s eye). The kact values were extrapolated from those at 35 °C.16,18
Initiator number key: (1) MClAc; (2) BzCl; (3) PECl; (4) MClP; (5) EtCliB;
(6) BzBr; (7) ClAN; (8) PEBr; (9) MBrP; (10) ClPN; (11) EtBriB; (12)
BrPN; (13) EBPA.
Figure 5. Correlation of kact and kdeact with KATRP for various CuIBr/ligand
catalysts with EtBriB at 22 °C in MeCN (the broken lines are just to guide
the reader’s eye). The kact values were extrapolated from those at 35 °C.16,17
Ligand number key: (1) Cyclam-B; (2) Me6TREN; (3) TPMA; (4) BPED;
(5) TPEDA; (6) PMDETA; (7) BPMPA; (8) dNbpy; (9) HMTETA; (10)
bpy; (11) N4[3,2,3]; (12) N4[2,3,2].
and N[2,3,2], are included in this report. Although ligands in the
upper-left section of the map (e.g., Me4Cyclam) can form catalysts
that afford very fast polymerization with large KATRP values, the
polymerization is uncontrolled because of the small values of kdeact
.
The trend in Figure 7 for initiators is similar to that in Figure 5.
The more-reactive dormant species (larger kact) form more-stable,
i.e., less-reactive radicals (smaller kdeact). The initiators that have
larger KATRP values also have larger kact and smaller kdeact values.
However, kdeact is much less dependent on KATRP for initiators than
for ligands. This suggests a smaller radical selectivity toward CuII
species.
Conclusions
Figure 6. Correlation of KATRP and kdeact for various CuI/ligand catalysts
with EtBriB at 22 °C in MeCN (see Table 5).
Equilibrium constants for catalysts with various nitrogen-based
ligands and alkyl halide-based initiators have been determined in
acetonitrile at 22 °C. The activities of the CuI/ligands and initiators
span 7 orders of magnitude and are strongly affected by the
structures of the ligand and the initiator. In general, the activity of
the Cu complex decreases as the ligand is changed from Cyclam-B
> N4-branched > N4-linear > N3 > N2. The ethylene group
provides a better linkage than the propylene group for the
coordinating nitrogens in the ligand. The activity of the initiator
decreases as the alkyl group is varied from 3° > 2° > 1° and its
R-substituent varied from -CN < -Ph < -C(O)OR > -CH3.
In general, the activities of alkyl bromide initiators are several times
larger than those of the corresponding chlorides. A full kinetic
picture for ATRP was constructed in order to comprehensively
understand the structure-activity relationships for the ligand and
initiators in ATRP. Along with the excellent correlation between
CuII/I redox potentials and equilibrium constants, it is possible to
design and predict the activities of some new ATRP catalytic and
initiating systems.
contributes to the overall value of KATRP much more than kdeact
.
Figure 5 also shows that stronger activators are generally weaker
deactivators.
However, the “ideal” ATRP catalyst should have a large value
of KATRP (so it can be used at low concentrations) but also should
have a large value of kdeact (to provide good control and generate
polymers with low polydispersity). Such catalysts would be ideal
for activators regenerated by electron transfer (ARGET) and
initiators for continuous activator regeneration (ICAR) methods
of ATRP, which are carried out with parts per million amounts of
Cu.26,27 Figure 6 shows the more-detailed correlation of KATRP and
kdeact for all of the ligands discussed in this report. The ligands in
this “map” for rational ligand selection are divided into four types.
The ligands in the lower-right section of the map are typical ligands
used in ATRP that give moderately fast polymerization with fairly
large KATRP values and good control due to sufficiently large kdeact
values. Most traditional ATRP ligands, such as bpy, dNbpy,
PMDETA and HMTETA, fall within this area. The ligands in the
upper-right area are a new generation of ligands that afford both
very fast polymerization with large KATRP values and good control
with fairly large kdeact values. These ligands have been successfully
used in new ATRP systems such as ICAR ATRP and ARGET
ATRP. Ligands in the lower-left part can form catalysts that give
rise to slow rates of polymerization and poor control due to small
KATRP and kdeact values, respectively. These are inefficient ligands
for ATRP, and therefore, only a few of them, such as N4[3,2,3]
Acknowledgment. Financial support from the NSF (CHE-07-
15494) and the CRP Consortium at CMU is gratefully acknowledged.
M.L.C. gratefully acknowledges financial support from the Australian
Research Council under their Centres of Excellence program.
Supporting Information Available: Complete ref 49. This
material is available free of charge via the Internet at http://
pubs.acs.org.
JA802290A
9
J. AM. CHEM. SOC. VOL. 130, NO. 32, 2008 10713