A highly active zinc catalyst for the controlled polymerization of lactide

Article abstract of DOI:10.1021/ja0359512

We report the preparation, structural characterization, and detailed lactide polymerization behavior of a new ZN(II) alkoxide complex, (L ₁ZnOEt)₂ (L₁ = 2,4-di-tert-butyl-6-{[(2′ -dimethylaminoethyl)methylamino]-methyl}phenolate). While an X-ray crystal structure revealed the complex to be dimeric in the solid state, nuclear magnetic resonance and mass spectrometric analyses showed that the monomeric form L₁ZnOEt predominates in solution. The polymerization of lactide using this complex proceeded with good molecular weight control and gave relatively narrow molecular weight distribution polylactide, even at catalyst loadings of <0.1% that yielded M_n as high as 130 kg mol ^-1. The effect of impurities on the molecular weight of the product polymers was accounted for using a simple model. Detailed kinetic studies of the polymerization reaction enabled integral and nonintegral orders in L ₁ZnOEt to be distinguished and the empirical rate law to be elucidated, -d[LA]/dt = K_p[L₁ZnOEt][LA]. These studies also showed that L₁ZnOEt polymerizes lactide at a rate faster than any other Zn-containing system reported previously. This work provides important mechanistic information pertaining to the polymerization of lactide and other cyclic esters by discrete metal alkoxide complexes.

Full text of DOI:10.1021/ja0359512

Published on Web 08/21/2003
A Highly Active Zinc Catalyst for the Controlled
Polymerization of Lactide
Charlotte K. Williams,† Laurie E. Breyfogle,† Sun Kyung Choi,‡ Wonwoo Nam,‡
Victor G. Young, Jr.,† Marc A. Hillmyer,*,† and William B. Tolman*,†
Contribution from the Department of Chemistry, UniVersity of Minnesota, 207 Pleasant Street
SE, Minneapolis, Minnesota, 55455 and Department of Chemistry and DiVision of Nano
Sciences, Ewha Womans UniVersity, Seoul 120-750, Korea
Received May 5, 2003; E-mail: hillmyer@chem.umn.edu; tolman@chem.umn.edu
Abstract: We report the preparation, structural characterization, and detailed lactide polymerization behavior
of a new Zn(II) alkoxide complex, (L1ZnOEt)2 (L1 ) 2,4-di-tert-butyl-6-{[(2′-dimethylaminoethyl)methylamino]-
methyl}phenolate). While an X-ray crystal structure revealed the complex to be dimeric in the solid state,
nuclear magnetic resonance and mass spectrometric analyses showed that the monomeric form L1ZnOEt
predominates in solution. The polymerization of lactide using this complex proceeded with good molecular
weight control and gave relatively narrow molecular weight distribution polylactide, even at catalyst loadings
of <0.1% that yielded Mn as high as 130 kg mol-1. The effect of impurities on the molecular weight of the
product polymers was accounted for using a simple model. Detailed kinetic studies of the polymerization
reaction enabled integral and nonintegral orders in L1ZnOEt to be distinguished and the empirical rate law
to be elucidated, -d[LA]/dt ) kp[L1ZnOEt][LA]. These studies also showed that L1ZnOEt polymerizes lactide
at a rate faster than any other Zn-containing system reported previously. This work provides important
mechanistic information pertaining to the polymerization of lactide and other cyclic esters by discrete metal
alkoxide complexes.
Introduction
analysis (e.g., non first-order kinetic dependencies on catalyst
concentration).6,7 Alternatively, mononuclear “single-site” Zn(II)
The metal catalyzed ring-opening polymerization of lactide
(LA), a renewable resource, is an efficient way to produce a
biodegradable and recyclable plastic, polylactide (PLA).1 The
commercial production of PLA for applications in medicine,
packaging, and fiber technology relies on this important catalytic
transformation.2 Diverse metal alkoxide complexes exhibit
varying degrees of effectiveness for the polymerization of LA
(and related cyclic esters), as reflected by differing rates and
levels of polymerization control and stereoselectivity.3 Notably
lacking is fundamental mechanistic knowledge to help under-
stand these differences and facilitate next-generation catalyst
design. Thus, we aim to develop, structurally characterize, and
carefully delineate the mechanistic behavior of new, discrete
metal-alkoxide LA polymerization catalysts comprising non-
toxic, inexpensive metals such as zinc4 or iron.5 In the case of
Zn(II), simple salts suffer from low reactivity, limited solubility,
and solution aggregation phenomena that complicate mechanistic
alkoxides supported by tris(pyrazolyl)hydroborate8 or â-diketim-
inate9,10 ligands, for example, have been shown to be quite
active, controlled, and, in some cases,11,12 stereoelective LA
polymerization catalysts.
In a new strategy inspired by the cooperativity of dizinc sites
in hydrolytic enzymes,13 we recently reported the LA poly-
merization behavior for a dizinc complex of a dinucleating
ligand HL.4 For the purpose of comparison, we chose to examine
a Zn(II) alkoxide supported by a closely related mononucleating
† University of Minnesota.
‡ Ewha Womans University.
(1) (a) Drumright, R. W.; Gruber, P. R.; Henton, D. E. AdV. Mater. 2000, 12,
1841. (b) Amass, W.; Amass, A.; Tighe, B. Polym. Int. 1998, 47, 89.
(2) http://www.cdpoly.com/.
(3) (a) O’Keefe, B. J.; Hillmyer, M. A.; Tolman, W. B. J. Chem. Soc., Dalton
Trans. 2001, 2215. (b) Kuran, W. Prog. Polym. Sci. 1998, 23, 919. (c)
Lo¨fgren, A.; Albertsson, A.-C.; Dubois, P.; Je´roˆme, R. J. Macromol. Sci.,
ReV. Macromol. Chem. Phys. 1995, C35, 379.
(4) Williams, C. K.; Brooks, N. R.; Hillmyer, M. A.; Tolman, W. B. Chem.
Commun. 2002, 2132.
(6) (a) Noltes, J. G.; Verbeek, F.; Overmars, H. G. J.; Boersma, J. J. Organomet.
Chem. 1970, 24, 257. (b) Nijenhuis, A. J.; Grijpma, D. W.; Pennings, A.
J. Polym. Bull. 1991, 26, 71. (c) Barakat, I.; Dubois, Ph., Je´roˆme, R.;
Teyssie, Ph. Macromolecules 1991, 24, 6542. (d) Schwach, G.; Coudane,
J.; Engel, R.; Vert, M. Polym. Int. 1998, 46, 177. (e) Kricheldorf, H. R.;
Damrau, D. O. Macromol. Chem. Phys. 1998, 199, 1747.
(7) Kowalski, A.; Duda, A.; Penczek, S. Macromolecules 1998, 31, 2114 and
references therein.
(8) Chisholm, M. H.; Eilerts, N. W.; Huffman, J. C.; Iyer, S. S.; Pacold, M.;
Phomphrai, K. J. Am. Chem. Soc. 2000, 122, 11845.
(9) Chamberlain, B. M.; Cheng, M.; Moore, D. R.; Ovitt, T. M.; Lobkovsky,
E. B.; Coates, G. W. J. Am. Chem. Soc. 2001, 123, 3229.
(5) (a) O’Keefe, B. J.; Monnier, S. M.; Hillmyer, M. A.; Tolman, W. B. J.
Am. Chem. Soc. 2001, 123, 339. (b) O’Keefe, B. J.; Breyfogle, L. E.;
Hillmyer, M. A.; Tolman, W. B. J. Am. Chem. Soc. 2002, 124, 4384.
9
11350
J. AM. CHEM. SOC. 2003, 125, 11350-11359
10.1021/ja0359512 CCC: $25.00 © 2003 American Chemical Society

Zinc Catalyst for Controlled Lactide Polymerization
A R T I C L E S
Scheme 1
Figure 1. Representation of the X-ray crystal structure of 1a, with selected
nonhydrogen atoms labeled and hydrogen atoms omitted for clarity. Selected
bond distances (Å) and angles (deg): Zn1-O1, 1.956(2); Zn1-C21,
1.997(4); Zn1-N2, 2.128(3); Zn1-N1, 2.147(3); O1-Zn1-C21, 128.77(14);
O1-Zn1-N2, 99.76(12); C21-Zn1-N2, 121.23(15); O1-Zn1-N1,
93.32(12); C21-Zn1-N1, 117.82(16); N2-Zn1-N1, 84.85(13).
Table 1. Summary of X-ray Crystallography Data
1a
2a
empirical foumula
formula weight
crystal system
space group
a (Å)
b (Å)
c (Å)
C
22H40N2OZn
C44H80N4O4Zn2
859.86
orthorhombic
Pbcn
38.339(3)
17.0503(15)
17.3185(15)
90
413.93
monoclinic
P21/c
11.9224(16)
12.0853(16)
16.687(2)
90
103.489(2)
90
2338.1(5)
4
1.176
colorless, plate
0.22 × 0.19 × 0.08
1.062
50.06
16067
4121
2894
262
0.0467
0.1233
0.990
ligand (HL1). We report herein the synthesis of such a complex
and our finding that it polymerizes LA with good control and
at a rate faster than any other Zn-containing system reported
previously. Considerable effort has been expended to character-
ize the structure of the new complex, both in the solid state
(X-ray crystallography, mass spectrometry) and in solution
(NMR, pulsed gradient spin-echo (PGSE)14) under catalytically
relevant conditions. Moreover, we disclose detailed studies of
the polymerization reaction that elucidate the empirical rate law
and provide important mechanistic information.
R (deg)
â (deg)
90
90
γ (deg)
V (Å3)
11321.0(17)
8
1.009
colorless, plate
0.20 × 0.20 × 0.06
0.882
50.06
30924
9979
6142
485
0.0514
0.1318
0.947
Z
density (calcd) (g cm-3
)
crystal color and morphology
crystal size (mm3)
absorption coefficient (mm-1
2θmax (deg)
)
Results
no. of reflcns collected
no. of ind reflcns
no. of obsd reflcns [I > 2σ(I)]
parameters
Synthesis and Structures of Complexes. The ligand HL1
was prepared by refluxing N,N,N′-trimethylenediamine, para-
formaldehyde, and 2,4-di-tert-butylphenol. A related ligand HL2
(R ) Me) was prepared in a similar manner and was used to
synthesize an analogous complex to that made with HL1 for
crossover experiments (see below).15 Reaction of HL1 or HL2
with Et2Zn yielded rare examples of well-defined NNO-ligated
zinc alkyls,16 L1ZnEt (1a) and L2ZnEt (1b) (Scheme 1). The
R1a [I > 2σ(I)]
wR2b
goodness-of-fit
largest diff peak, hole (eÅ-3
)
1.796, -0.472
0.461, -0.458
2
2
2
a R1 ) Σ|Fo| - |Fc|/Σ|Fo|. b wR2 ) [Σ[w(Fo - Fc )2]/[Σ[w(Fo )2]]1/2
,
2
where w ) 1/σ2(Fo ) + (aP)2 + bP.
ZnOEt]n (2b), which were isolated as colorless solids in >80%
yield. The X-ray crystal structure of 2a (Figure 2, Table 1)
shows that in the solid state the complex exists as a dimer
bridged by two ethoxides, the intermetal separation being
3.0835(6) Å. Each zinc atom is pentacoordinate with a distorted
square pyramidal geometry.18 The bridging oxygen atoms are
asymmetrically placed between the two zinc atoms (e.g., O3 is
0.036 Å closer to Zn1 than to Zn2), as was also found for the
dizinc complex of HL.4
The 1H NMR spectra of 1a, 1b, 2a, and 2b in CD2Cl2 contain
single sets of sharp peaks at ambient temperature that are
consistent with structures in solution that are monomeric,
dimeric, or rapidly equilibrating mixtures. A representative
spectrum of 2a is shown in Figure S1 (Supporting Information).
The features in this spectrum broaden and become more
complex at low temperatures, indicating that a fluxional process-
(es) is(are) operative. This process(es) could not be slowed
sufficiently for study, even at -90 °C, so we have been unable
to discern by simple variable temperature NMR spectral analysis
whether intramolecular rearrangements or monomer/dimer
equilibration (or both) are involved. Because the nuclearity of
1
complexes were formulated on the basis of H and 13C{1H}
NMR spectroscopy, elemental analysis, and, for 1a, an X-ray
crystal structure (Figure 1, Table 1). The X-ray structure shows
that 1a is monomeric, with the four-coordinate Zn(II) ion
adopting a distorted tetrahedral geometry. The differences
between the angles C21-Zn-O1 (128.77(14)°) and N1-Zn-
N2 (84.85(13)°) are illustrative of the distortion from an ideal
tetrahedral topology. The Zn-C21 distance of 1.997(4) Å is
slightly longer than usually seen for Zn(II)-alkyl bonds (1.93-
1.98 Å).16,17
Treatment of these complexes with EtOH (∼1.2 equiv)
generated the ethoxide complexes [L1ZnOEt]n (2a) and [L2-
(10) Chisholm, M. H.; Gallucci, J.; Phomphrai, K. Inorg. Chem. 2002, 41, 2785.
(11) Coates, G. W. J. Chem. Soc., Dalton Trans. 2002, 467.
(12) (a) Cheng, M.; Attygalle, A. B.; Lobkovsky, E. B.; Coates, G. W. J. Am.
Chem. Soc. 1999, 121, 11583. (b) Ovitt, T. M.; Coates, G. W. J. Am. Chem.
Soc. 1999, 121, 4072. (c) Ovitt, T. M.; Coates, G. W. J. Polym. Sci., Part
A: Polym. Chem. 2000, 38, 4686.
(13) Selected reviews: (a) Stra¨ter, N.; Lipscomb, W. N.; Klabunde, T.; Krebs,
B. Angew Chem., Int. Ed. Engl. 1996, 35, 2024. (b) Wilcox, D. E. Chem.
ReV. 1996, 96, 2435. (c) Cowan, J. A. Chem. ReV. 1998, 98, 1067.
(14) Valentini, M.; Pregosin, P. S.; Ru¨egger, H. Organometallics 2000, 19, 2551.
(15) The polymerization activity of this complex was not explored.
(16) (a) Corey, E. J.; Yuen, P.-W.; Hannon, F. J.; Wierda, D. A. J. Org. Chem.
1990, 55, 784. (b) Dowling, C.; Parkin, G. Polyhedron 1996, 15, 2463. (c)
Hammes, B. S.; Carrano, C. J. Inorg. Chem. 1999, 38, 4593.
(18) τ ) 0.43 for Zn(1), with O(1) axial, and τ ) 0.40 for Zn(2), with O(2)
axial. See: Addison, A. W.; Rao, T. N.; Reedjik, J.; von Rijn, J.; Verschoor,
G. C. J. Chem. Soc., Dalton Trans. 1984, 1349.
(17) Looney, A.; Han, R.; Gorrell, I. B.; Cornebise, M.; Yoon, K.; Parkin, G.;
Rheingold, A. L. Organometallics 1995, 14, 274.
9
J. AM. CHEM. SOC. VOL. 125, NO. 37, 2003 11351

A R T I C L E S
Williams et al.
Figure 3. LDMS data obtained on the solids resulting from evaporation
of CH2Cl2 solutions of (a) 2a, (b) 2b, and (c) a 1:1 mixture of 2a and 2b.
Simulations of the indicated ions are shown in part d.
Figure 2. Representation of the X-ray crystal structure of 2a, with selected
nonhydrogen atoms labeled and hydrogen atoms omitted for clarity. Selected
bond distances (Å): Zn1-Zn2, 3.0835(6); Zn1-O1, 1.945(3); Zn1-O3,
2.037(2); Zn1-O4, 2.009(2); Zn1-N1, 2.214(3); Zn1-N2, 2.268(3); Zn2-
O2, 1.942(2); Zn2-O3, 2.001(2); Zn2-O4, 2.041(3); Zn2-N3, 2.233(3);
Zn2-N4, 2.266(3).
2a in solution is of particular importance for understanding its
catalytic behavior (see below), we performed additional experi-
ments aimed specifically at determining whether 2a, which by
X-ray crystallography is dimeric with five-coordinate Zn(II)
centers in the solid state, remains dimeric or cleaves into four-
coordinate monomers in solution.
Figure 4. Plot of ln(I/I0) vs γ2δ2G2[∆-(δ/3)] × 109 (m-2 s) from PGSE
experiments for 1a (2), 2a (9), and 3 (b). The slopes (D) of the indicated
linear fits were used to calculate r values (Table 2) as described in the text.
First, we performed crossover experiments whereby 2a and
2b were mixed in CH2Cl2 or CD2Cl2. Analysis of 1H NMR spec-
tra of mixtures with 2:1, 1:1, and 1:2 molar ratios of 2a and 2b
in CD2Cl2 showed peaks identical to those of the indepen-
dent complexes, with relative intensities that corresponded to
their molar ratios. Although these data are consistent with 2a
and 2b existing as monomers, as dimers or as either or both in
rapid equilibrium, the absence of additional resonances rules
out the exclusive formation of a heterodimeric species such as
[L1Zn(µ-OEt)2ZnL2] (unless such a species has NMR properties
fortuitously identical to those of the separate compounds 2a
and 2b).
To assess the possible equilibration between monomeric and
dimeric structures, the solvent was removed from CH2Cl2 solu-
tions of 2a, 2b, and mixtures of the two, and the resulting solids
were analyzed by laser desorption mass spectrometry (LDMS).
While parent ion envelopes were not observed for the separate
complexes 2a and 2b, each exhibited a peak envelope corre-
sponding to the respective dimeric ion [(L1)2Zn2(OH)]+ and
[(L2)2Zn2(OH)]+ (Figure 3a and b), corroborated by spectral
simulation (Figure 3d).19 Laser ablation of the solid resulting
from removal of CH2Cl2 from a 1:1 solution mixture of 2a and
2b after 30 min of equilibration time yielded the mass spectrum
in Figure 3c. In addition to the same peaks as those seen for
the separate samples of 2a and 2b, a new peak envelope appears
at m/z ≈ 745 that corresponds to [L1Zn(OH)ZnL2]+. The approx-
imate 1:2:1 intensity ratio of the three signals corresponding to
the homo- and heterodimeric ions is consistent with the gener-
ation of monomers in solution that statistically dimerize in the
solid state.
The question of whether 2a exists primarily in solution as a
monomer or dimer was also addressed through pulsed gradient
spin-echo (PGSE) NMR measurements. The PGSE method
assesses translational motion of the analyte in solution through
determination of the diffusion coefficient, a quantity inversely
related to its effective hydrodynamic volume.20 The method has
been used successfully for characterizing various organometallic
systems,14 including ones involved in monomer/dimer equilib-
ria.21 In addition to performing PGSE experiments on CD2Cl2
solutions of 2a, we examined two reference compounds. These
are 1a, which is similar in size to the monomeric form of 2a,
and a larger Al complex [L”Al(OiPr)] (3) prepared in a separate
work.22 The results are displayed in Figure 4, plotted as ln(I/I0)
(where I ) observed spin-echo intensity and I0 ) intensity in
(20) Price, W. S. Concepts Magn. Reson. 1997, 9, 299.
(21) (a) Geldbach, T. J.; Pregosin, P. S.; Albinati, A.; Rominger, F. Organo-
metallics, 2001, 20, 1932. (b) Pichota, A.; Pregosin, P. S.; Valentini, M.;
Wo¨rle, M.; Seebach, D. Angew. Chem., Int. Ed. 2000, 39, 153. (c) Jiang,
Q.; Ru¨egger, H.; Venanzi, L. M. Inorg. Chim. Acta 1999, 290, 64.
(22) L” ) N,N-bis[methyl(2-hydroxy-3,5-di-tert-butylphenyl)]-N′,N′-dimethyl-
ethylenediamine. Alcazar-Roman, L. M.; O′Keefe, B.; Hillmyer, M. A.;
Tolman, W. B. J. Chem. Soc., Dalton Trans. 2003, 3082.
(19) Rockwood, A. L.; VanOrden, S. L.; Smith, R. D. Anal. Chem. 1995, 67,
2699.
9
11352 J. AM. CHEM. SOC. VOL. 125, NO. 37, 2003

Zinc Catalyst for Controlled Lactide Polymerization
A R T I C L E S
Table 2. PGSE Experimental Results Compared to Estimated
Table 3. Representative Results for the Polymerization of LA
by 2aa
Data from X-ray Crystal Structuresa
b
D
D′
[LA]0/[L1ZnOEt]0
time (min)
conversion (%)b
M (kg mol-1 c
)
PDIc
n
2
2
complex
(× 10-10 m s-1
)
rc (Å) ad (Å) bd (Å) r′ e (Å) (× 10-10 m s-1
)
650
1000
1500
5
13
18
96
96
93
67
99
130
1.42
1.40
1.34
L1ZnEt (1a)
[L”Al(OiPr)] (3)
2a
12.49(1)
9.650(9)
10.55(2)
4.2(3) 5.6(4) 4.2(4) 4.6(4)
5.4(3) 6.8(5) 5.5(4) 5.9(5)
4.9(3) 9.9(7) 5.6(4) 6.8(6)
11.4(2)
9.0(2)
7.6(2)
a Conditions: CH2Cl2, 25 °C, [LA]0 ) 1 M, in a glovebox. b Determined
by 1H NMR spectroscopy. c Determined by SEC using light scattering
detection (see Experimental Section).
a Estimated errors are indicated in parentheses, with terms defined in
the text. b From slopes of the lines in Figure 4, eq 1, and eq 2. c The errors
are estimated to be ∼5% primarily on the basis of the accuracy of η used
in the calculation. d Estimated values from the X-ray crystal structure by
measuring the lengths between the centers of distant hydrogen atoms on
the periphery corresponding to each arbitrary axis and by calculating the
minimum ellipsoid into which the molecule fits. e Calculated using eq 3.
for 1a (4.6 Å). This estimated r′ was then used to calculate D′
) 10.0 × 10-10 m2 s-1 (assuming a spherical shape for the
molecule). Both the estimated r′ and D′ values for the putative
monomeric form of L1ZnOEt closely agree with those that were
experimentally determined for 2a, consistent with assignment
of the monomeric structure L1ZnOEt in solution. As such, we
refer to 2a in solution as L1ZnOEt in the following discussion
of its polymerization behavior.
the absence of gradients), versus a function with the square of
the gradient strength over a constant diffusion time. Each data
point represents an average of the values obtained for the dif-
ferent resonances in the NMR spectrum of each compound at
varying gradient strengths. The slopes of the lines in Figure 4
are equal to the diffusion coefficients (D) according to eq 1
(where G ) gradient strength, γ ) gyromagnetic ratio, δ )
length of gradient pulse, and ∆ ) delay between gradient mid-
points). The absolute value of the slope decreases with increas-
ing effective hydrodynamic radii according to the Stokes-
Einstein equation (eq 2, where r ) effective hydrodynamic rad-
ius and η ) viscosity). The experimentally determined values
of D and r were calculated and are listed in Table 2. From these
data for solutions of the three complexes analyzed, the following
ranking of effective hydrodynamic volumes is given: 3 > 2a
> 1a.
For comparison, we estimated values of the molecular radii
(r′) for 1a, 2a, and 3 from their X-ray crystal structures by
measuring the lengths between centers of distant hydrogen atoms
on the molecular periphery.23 All of the complexes were
considered as prolate ellipsoids with major and minor axes a
and b, respectively (Table 2), and r′ was determined by using
eq 3.24 Importantly, the values of r (PGSE) and r′ (X-ray) for
1a and 3 closely agree. In contrast, for 2a, the radius determined
by the PGSE measurement (r) is ca. 25% smaller than the radius
estimated from the X-ray crystal structure (r′) of the dinuclear
form. These data strongly suggest that 2a does not retain its
dimeric form in solution.
We also calculated diffusion coefficients (D′) from the values
of a, b, and r′, which required the use of a variant of eq 2
appropriate for prolate ellipsoids.24 As expected, the agreement
between the PGSE measured (D) and predicted (D′) diffusion
coefficients is excellent for 1a and 3 (consistent with the
agreement of r and r′), but D′ for 2a is much less than the
measured value, again consistent with the form of 2a in solution
being significantly smaller than the dinuclear structure adopted
in the crystal. On this basis, therefore, we postulate that 2a exists
predominantly as a monomer in CD2Cl2. As a check on this
hypothesis, we estimated an upper bound for the radius expected
for monomeric L1ZnOEt, r′ ) 5.3 Å, by adding approximately
one-half of the distance of an O-Et bond (0.7 Å) to the r′ value
I
δ
3
ln
) -γ2δ2G2 ∆ -
D
(1)
( )
(
)
I0
kT
D )
(2)
(3)
6πηr
r′ ) (ab2)1/3
Polymerization Activity. While 1a does not polymerize D,L-
LA, consistent with a coordination insertion mechanism requir-
ing an initiating and propagating metal alkoxide,25 2a is highly
active for the polymerization of D,L-LA in CH2Cl2 at ambient
temperature (Table 3). Polymerizations were performed under
strictly dry and anaerobic conditions in a glovebox with initial
monomer concentrations, [LA]0, of 1 M. Conversions were
assayed by 1H NMR spectroscopy, and accurate polymer
molecular weights and polydispersities were determined in all
cases by size exclusion chromatography (SEC) using multi-
wavelength light scattering detection. We observed rapid
reactions over a wide range of initial monomer-to-catalyst ratios,
with a high conversion of LA possible even up to [LA]0/[L1-
ZnOEt]0 ratios of 1500, yielding PLA with molecular weights
as large as 130 kg mol-1. We have found no other report of a
Zn(II) complex that is effective at such low levels of catalyst
loading ([L1ZnOEt]0 ≈ 0.7 mM). Good molecular weight control
is demonstrated by a linear increase in Mn with LA conversion
at [LA]0/[L1ZnOEt]0 ) 1000 (Figure 5) and relatively narrow
molecular weight distributions of the products (polydispersity
index, PDI ≈ 1.4). Despite the linear dependence of Mn with
LA conversion, the polylactide molecular weights are lower than
expected (dotted line in Figure 5) based on the initial concentra-
tions of catalyst and monomer, particularly at high conversion.
These features also are evident in a plot of Mn versus ([LA]0 -
[LA]t)/[L1ZnOEt]0, measured over a wide range of initial
monomer-to-catalyst ratios (filled circles, Figure 6). For ex-
ample, at [LA]0/[L1ZnOEt]0 ) 1500 at 98.5% conversion, the
PLA molecular weight should be 214 kg mol-1 compared to
the experimental value of 130 kg mol-1
.
Such divergences from anticipated molecular weights may
be explained by an effective concentration of catalyst, [L1-
(23) Hydrodynamic volumes may be estimated via an alternative method of
PGSE data analysis that uses summations of van der Waals radii, but this
method was not utilized for the complexes studied herein because of
possible inaccuracies resulting from void spaces in the structures. Stahl,
N. G.; Zuccaccia, C.; Jensen, T. R.; Marks, T. J. J. Am. Chem. Soc. 2003,
125, 5256.
(25) Duda, A.; Penczek, S. Polymers from Renewable Resources: Biopolyesters
and Biocatalysis; Scholz, C., Gross, R. A., Eds.; ACS Symposium Series
Vol. 764; American Chemical Society: Washington, DC, 2000; pp 160-
198.
(24) Price, W. S. In New AdVances in Analytical Chemistry; Rahman, A.-ur-.
Ed.; Hardwood Academic Publishers: Singapore, 2000; pp 31-72.
9
J. AM. CHEM. SOC. VOL. 125, NO. 37, 2003 11353

A R T I C L E S
Williams et al.
Accordingly, since [CD] can conceivably range from 0 (no
deactivating impurity) to just less than 1 mM () [L1ZnOEt]0,
thus virtually all catalyst destroyed), the range of possible values
for [EA] in this analysis is 0.34 mM < [EA] < 1.34 mM.
([LA]0 - [LA]t)(144 g mol-1)
Mn )
(4)
[L1ZnOEt]0 + C
To further test the feasibility of rapid chain exchange during
catalysis by 2a, we added various amounts of benzyl alcohol
(BnOH) to polymerization reactions performed with a low
concentration of catalyst ([L1ZnOEt]0 ≈ 0.7 mM, [LA]0 ) 1
M). No significant changes in the rate of PLA production or
the polylactide PDI with added BnOH were obvious, indicating
that 2a retains its activity in the presence of added alcohol (even
at [ROH]0/[L1ZnOEt]0 ) 150, for R ) Et). Importantly, added
BnOH caused Mn to decrease, as shown by the data plotted as
squares in Figure 6 (for these points, we used [L1ZnOEt]0 )
0.67 mM + [BnOH]0, where [BnOH]0 ) 0.17-1.34 mM). As
predicted for BnOH acting as a rapid EA, the data (squares)
closely parallel those obtained when ([LA]0 - [LA]t)/[L1-
ZnOEt]0 was varied in the absence of BnOH (circles). Indeed,
in an analysis analogous to that applied to Figure 5, the
combined data in Figure 6 may be fit (solid curve) to eq 4,
where EA is now the “intrinsic” impurity present in the absence
of added BnOH. The resulting value C ) 0.46(3) mM yields a
range 0.46 mM < [EA] < 1.13 mM (0 < [CD] < 0.67 mM),
in satisfyingly close agreement with the results from the analysis
of the independently acquired data in Figure 5. In sum, the
results confirm that LA polymerizations by 2a are sensitive to
rapid chain exchange both by purposefully added BnOH and
by an impurity EA. The amount of this EA has been estimated
by considering possible limiting values of a CD, independent
kinetic evidence for which is presented in the next section.
Figure 5. Plot of Mn and PDI vs conversion for the polymerization of LA
by 2a. Conditions: [LA]0 ) 1 M, [LA]0/[L1ZnOEt]0 ) 1000, CH2Cl2, 25
°C. The dashed line represents the expected relationship of Mn with %
conversion based on eq 4 with C ) 0 mM, while the solid line is the fit of
the data to eq 4, yielding C ) 0.34(3) mM.
We determined that the ethoxide group of 2a initiates the
polymerization of LA through acyl cleavage by performing
polymerizations at high catalyst loadings. A reaction performed
with [LA]0/[L1ZnOEt]0 ) 50 was quenched after 10 s (95%
conversion) and analyzed by 1H NMR spectroscopy. Both
ethoxy ester and hydroxy end groups were visible in the
spectrum in a 1:1 ratio. However, by integration there were 70
LA units per endgroup, rather than 50 as predicted from the
[LA]0/[L1ZnOEt]0 ratio used. In addition, the PDI of the
resulting polymer was broad (2.31). We hypothesize that the
low level of ethoxy end groups and high PDI under these
conditions (high catalyst loading) may be due to initiation being
somewhat slower than propagation. In separate experiments
using [LA]0/[L1ZnOEt]0 ) 1075 and 20 equiv of added EtOH
exchange reagent, the discrepancy between experimental ob-
servations and theory narrowed. At 92% conversion, 49 LA units
Figure 6. Plot of Mn (b in the absence of added BnOH, 9 with added
BnOH) and PDI vs ([LA]0 - [LA]t)/[L1ZnOEt]0 for the polymerization of
LA by 2a. All LA conversions were > 83%. Conditions: [LA]0 ) 1 M,
CH2Cl2, 25 °C. For the points corresponding to runs with added BnOH
(9), [L1ZnOEt]0 ) 0.67 mM + [BnOH]0. The dashed line represents the
expected relationship of Mn with ([LA]0 - [LA]t)/[L1ZnOEt]0 based on eq
4 with C ) 0 mM, while the solid curve is the fit of the data to eq 4,
yielding C ) 0.46(3) mM.
ZnOEt]eff, that differs from the amount added, [L1ZnOEt]0, due
to the presence of impurities that may destroy some of the
catalyst (catalyst deactivator, CD) and/or promote rapid ex-
change between a propagating chain and an exchange agent
(EA)26 or a polymeric alcohol. An EA would cause [L1ZnOEt]eff
to be greater than [L1ZnOEt]0, increase the total number of
polymer chains, and thus decrease Mn at a fixed conversion,
while a CD would have the opposite effect. These relationships
are described by eq 4, where C ) [EA] - [CD]. This analysis
reasonably assumes that chain exchange equilibria are more
rapid than propagation.27 A fit of the data in Figure 5 to eq 4
(where [LA]0 - [LA]t ) % conversion/100; [LA]0 ) 1 M) is
shown as the solid line and yields [L1ZnOEt]eff ) 1.34(3) mM.
1
were expected per endgroup; 46 were observed via H NMR
integration. The combined data firmly establish that ethoxy
groups initiate polymer chain growth, presumably through a
coordinative insertion pathway.
1
Finally, H NMR spectroscopic analysis of PLA produced
from D,L-LA indicates an atactic microstructure, even when the
polymerization was performed at low temperature (0 °C).28
A
(26) Odian, G. Principles of Polymerization; John Wiley & Sons: New York,
1991; pp 538-540.
(27) Ovitt, T. M.; Coates, G. W. J. Polym. Sci., Part A: Polym. Chem. 2000,
38, 4686.
(28) Thakur, K. A. M.; Kean, R. T.; Hall, E. S.; Kolstad, J. J.; Munson, E. J.
Macromolecules 1998, 31, 1487.
9
11354 J. AM. CHEM. SOC. VOL. 125, NO. 37, 2003

Zinc Catalyst for Controlled Lactide Polymerization
A R T I C L E S
Table 4. Observed Rate Constants for LA Polymerization by 2a
Relative to Selected Complexesa
b
complex
k
rel
[complex]0 (mM)
ref
(BDI)ZnOiPr
LZn2Cl2OEt
TpMgOEt
5.1
8.2
36
1220
2.1
3.3
11
9
4
8
8
TpZnOEt
11
a Conditions: CH2Cl2 or CD2Cl2 at 25 °C. b Equal to the ratio kobs(2a)/
obs(complex) at [complex]0 ) [L1ZnOEt]0. Values of kobs(2a) were
k
calculated using kobs(2a) ) kprop[L1ZnOEt]0, with kp ) 2.2 M-1 s-1
.
sample of PLA derived from L-LA was isotactic, indicating no
discernible epimerization during the polymerization.
Polymerization Kinetics. Insight into the mechanism of
polymerization of LA by 2a was obtained through kinetics
studies. Polymerizations were performed at either of two
temperatures at a fixed initial lactide concentration ([LA]0 )
1.0 M at 25 °C or [LA]0 ) 0.5 M at 0 °C), and catalyst
concentrations were varied over a range [L1ZnOEt]0 ) 1-20
mM. The polymerizations were continually monitored via in
situ FTIR spectroscopy (ReactIR). All polymerizations showed
a first order dependency on the concentration of LA over at
least four half-lives. The observed rate constants (kobs) for each
run were extracted from the exponential polymer growth/
monomer decay curves (Figure S2, Table S1). At ambient
temperature, the t1/2 is ≈ 2.5 min for [L1ZnOEt]0 ) 2.2 mM
(0.2 mol % Zn relative to lactide). To draw comparisons to
previously reported kinetic data for other catalysts, we list
relative rate constants krel, defined as the ratio of kobs for L1-
ZnOEt vs kobs for each complex (at equivalent concentrations
of complex) in Table 4.4,8,9 The data show that 2a is the fastest
Zn-containing LA polymerization catalyst discovered.29
To determine the kinetic order in [Zn], we analyzed the
dependence of kobs on [L1ZnOEt]0. Plots of ln(kobs) vs ln([L1-
ZnOEt]0) are linear (Figure 7a), but with noninteger slopes of
1.33 (at 0 °C) and 1.75 (at 25 °C). The plots suggest the
intriguing proportionality -d[LA]/dt [L1ZnOEt]n with n )
1.33 (0 °C) or 1.75 (25 °C). Such noninteger orders in catalyst
are difficult to interpret mechanistically, although models that
incorporate complicated aggregation phenomena have been
postulated for such cases.9,30 In an alternative interpretation of
the data, plots of kobs vs [L1ZnOEt]0 are linear, indicating a first
order dependency of the rate on [L1ZnOEt] (Figure 7b) and
giving the rate law -d[LA]/dt ) kp[L1ZnOEt][LA], with kp )
propagation rate constant ) 2.2 M-1 s-1 at 25 °C and 0.30 M-1
s-1 at 0 °C. This rate law is consistent with a mechanism
involving coordinative insertion at a single Zn site.
Figure 7. Kinetic plots for the polymerization of LA by 2a at 25 °C (b)
and 0 °C (2). Conditions: CH2Cl2, [LA]0 ) 1.0 M (25 °C) or 0.5 M (0
°C). The open triangle in part b is the data point for the experiment (0 °C)
in which [L1ZnOEt]0 was initially 1.0 mM (no polymerization) and then
was increased to 5.0 mM (polymerization initiated); see text.
the entire procedure on the benchtop by addition of catalyst
via syringe injection through a septum to a precooled LA
solution immersed in an ice bath with the ReactIR probe
inserted. Introduction of a trace impurity is more difficult to
avoid in the latter case. The 25 °C threshold value also is
essentially the same catalyst concentration used in successful
batch polymerizations using [LA]0/[L1ZnOEt]0 ) 1500 (Table
3) that were kept in the glovebox for the entire reaction time.
Taken together, these results imply that there is a low level of
impurity relative to [LA]0 ) 1 M (or [LA]0 ) 0.5 M at 0 °C)
that deactivates the catalyst and that the amount of impurity
(CD) varies with experimental protocol, suggesting that it may
be introduced inadvertently. Several additional experiments were
performed to test these hypotheses.
First, a kinetic experiment was run at 0 °C using syringe
transfer conditions and a catalyst concentration below the
threshold ([L1ZnOEt]0 ) 1.0 mM). ReactIR monitoring showed
polymerization neither after ∼30 min nor after warming to 25
°C and standing for an additional 30 min. These results show
that slow initiation at 0 °C is not responsible for the absence of
polymerization at low [L1ZnOEt]0 in the syringe transfer kinetic
runs. In another kinetic experiment run under syringe transfer
conditions at 0 °C using [L1ZnOEt]0 ) 1.0 mM, monitoring
for ∼1.5 h again showed no PLA production, but subsequent
addition of 2a to bring the total catalyst concentration to 5.0
mM (i.e., above the threshold) triggered the reaction. Poly-
merization proceeded with a kobs value of 9.2 × 10-4 s-1 (open
triangle in Figure 7b), in line with values obtained in kinetic
runs that began with this amount of catalyst. This result is
consistent with the hypothesis of an impurity (CD) that prohibits
polymerization catalysis when [L1ZnOEt]0 < 2.4 mM at 0 °C.
The x-intercepts for the linear fits shown in Figure 7b are
nonzero, suggesting that there is a threshold catalyst concentra-
tion below which polymerization will not occur.5 For the kinetic
runs performed at 25 °C, the threshold value is 0.7 mM, but at
0 °C it is 2.4 mM. Note that the experimental procedures differed
at the two temperatures; at 25 °C polymerization was initiated
by adding the catalyst stock solution to a solution of LA in the
glovebox and then removing the flask from the glovebox for
monitoring by ReactIR, but at 0 °C it was necessary to perform
(29) While no detailed kinetic data are available for some recently reported highly
active Ca29a and Mg29b catalysts, to a first approximation, 2a is comparably
effective. (a) Chisholm, M. H.; Gallucci, J.; Phomphrai, K. Chem. Commun.
2003, 48. (b) Chisholm, M. H.; Huffman, J. C.; Phomphrai, K. J. Chem.
Soc., Dalton Trans. 2001, 222.
(30) Kowalski, A.; Libiszowski, J.; Duda, A.; Penczek, S. Macromolecules 2000,
33, 1964.
9
J. AM. CHEM. SOC. VOL. 125, NO. 37, 2003 11355

A R T I C L E S
Williams et al.
Scheme 2
Working under the notion that this impurity was introduced
during the syringe transfer protocol necessary for ReactIR
monitoring in the 0 °C kinetic runs, we modified the procedure
to see if the impurity level could be lowered sufficiently to
enable polymerization below the [CD] ) 2.4 mM threshold at
0 °C. We carefully vacuum transferred CH2Cl2 onto a solid
mixture of LA (1 M) and 2a ([L1ZnOEt]0 ) 2.0 mM) at -80
°C, warmed the mixture to 0 °C, and allowed the reaction to
run for 2 h at 0 °C. PLA was formed to greater than 90%
conversion, confirming that if appropriate care is taken, the
impurity level can be reduced sufficiently to enable polymer-
ization at low [L1ZnOEt]0 at 0 °C. Finally, SEC data for
polymers from kinetic runs at both temperatures at equal [LA]0/
[L1ZnOEt]0 ()500) revealed different Mn values: Mn (0 °C) )
14.7 kg mol-1 versus Mn (25 °C) ) 46.2 kg mol-1. The finding
of lower polymer molecular weight with higher levels of
deactivating impurity suggests that the product of the catalyst
deactivation is itself a chain exchange agent.
order rate law, with first order dependencies on [LA] and on
[L1ZnOR] but with the presence of a low level (0.07-0.48%
based on [LA]0) of impurity that prohibits polymerization below
a threshold catalyst concentration. In addition to being indicated
by the nonzero intercepts in plots of kobs versus [L1ZnOEt]0
(Figure 7b), the presence of such an impurity was supported
by several independent experiments, perhaps the most critical
being commencement of polymerization upon addition of
catalyst to a nonactive reaction using [L1ZnOEt]0 below the
threshold value. Similar levels of deactivating impurity also have
been observed for several other reported LA polymerization
catalysts.4,5b With 2a, when special care was taken to decrease
the impurity level below the threshold found in the kinetic
experiments, use of high [LA]0/[L1ZnOEt]0 ratios led success-
fully to very high molecular weight PLA (up to 130 kg mol-1).
An alternative analysis of the rate data using plots of ln(kobs
)
versus ln([L1ZnOEt]0) yielded nonintegral orders in catalyst
(Figure 7a), similar to previous reports that invoked aggregation
phenomena as a rationale.9,30 On the basis of the combined
experimental data, we deem this interpretation less likely to be
correct in our system. Indeed, we caution against only evaluating
order in catalyst using the logarithmic analysis because the
possibility of a deactivating impurity is effectively obscured if
the nonintegral order is accepted without further scrutiny.
A candidate for the adventitious/introduced impurity is water,
since addition of 1 equiv (relative to L1ZnOEt) of water
completely terminated a kinetic experiment at 25 °C. In a second
experiment ([LA]0/[L1ZnOEt]0 ) 20), polymerization proceeded
to >95% conversion upon addition of 0.5 equiv of water. A
possible explanation for the effects of water on the polymeri-
zation reaction is presented below.
The coordination-insertion pathway shown in Scheme 2 is
consistent with the second order rate law and with the analysis
of polymer end groups. Additional chain exchange processes
also are illustrated, which are postulated in order to explain the
lower than predicted polymer molecular weights that we
measured accurately using SEC with light scattering detection.
Fits to the data in the plots of Mn versus % conversion and
([LA]0 - [LA]t)/[L1ZnOEt]0 (Figures 5 and 6) enabled [L1-
ZnOEt]eff to be determined, but in the absence of definitive
values for deactivating impurity concentration ([CD]), only a
possible range for the concentration of the exchange agent
([EA]) could be calculated: ∼0.4 mM < [EA] < ∼1.3 mM.
Although relatively low, these concentrations are significant at
low catalyst loadings (high [LA]0/[L1ZnOEt]0) and can consid-
erably decrease the observed polymer molecular weight. In
further support of the feasibility of chain exchange processes,
we found that added BnOH lowered the polymer molecular
weight without affecting the polymerization rate or polydisper-
Discussion
We have characterized a new Zn(II) alkoxide 2a that exhibits
good control in the polymerization of LA to yield high molecular
weight PLA at rates faster than other Zn systems reported to
date. Scheme 2 shows a proposed mechanism for the poly-
merization of LA catalyzed by 2a consistent with the combined
structural, reactivity, and kinetic data. Although dimeric in the
solid state, as shown by X-ray crystallography and LDMS, the
results of crossover experiments with analogue 2b and, most
convincingly, the PGSE data indicate that in CH2Cl2 solution
at ambient temperature 2a exists predominantly as a monomer.
Since the LA polymerizations were performed under the same
conditions (25 °C, CH2Cl2), we thus conclude that the resting
and, most likely, the catalytically active form of 2a is the
monomeric species L1ZnOR.
This conclusion is further supported by the results of the
kinetics experiments, from which we deduce an overall second
9
11356 J. AM. CHEM. SOC. VOL. 125, NO. 37, 2003

Zinc Catalyst for Controlled Lactide Polymerization
A R T I C L E S
or a glovebox. All solvents and reagents were obtained from commercial
sources and used as received unless otherwise stated. Pentane, dichlo-
romethane, and toluene were passed through purification columns (Glass
Contour, Laguna, CA). Deuterated solvents were dried over calcium
hydride, distilled under vacuum, and stored under nitrogen. Ethanol
and benzyl alcohol were distilled from calcium hydride. Diethyl zinc
(1 M in hexanes) was distilled immediately prior to use. D,L-Lactide
and L-lactide were purified by recrystallization from toluene, followed
by repeated (2-3×) vacuum sublimation. NMR Spectra were collected
on a Varian INOVA-300 or Varian INOVA-500 spectrometer. Elemen-
tal analyses were determined by Atlantic Microlab, Inc., Norcross, GA
and Robertson Microlit Laboratories, Inc., Madison, NJ. The molecular
weights and molecular weight distributions of the high molecular weight
polymers were determined by size exclusion chromotography (SEC)
in THF. Separation was achieved by three Phenomenex Phenogel
columns with pore sizes of 103, 104, and 105 Å. An interferometric
refractometer (Wyatt Technology OPTILAB) and a multiangle light
scattering detector equipped with a 632.8 nm laser (Wyatt Technology
DAWN) positioned downstream of the columns enabled the determi-
nation of polymer molecular weights based on a dn/dc value of 0.0576
mL g-1. High resolution electrospray mass spectra were recorded on a
Bruker Reflex III system. Direct laser desorption mass spectrometry
was performed using a Bruker Reflex III MALDI-TOF mass spec-
trometer (Bruker Daltonics, Billerica, MA). Samples were spotted onto
a stainless steel target in a glovebox and transported to the instrument
in a sealed container. The plate was quickly taken from the container
and inserted into the vacuum lock of the instrument. Exposure to air
in each case was <10 s. The laser employed was a pulsed nitrogen
laser operating at 337 nm with a maximum average power of 6 mW.
Typically, the laser power was attenuated by 10-20 dB in order to
obtain satisfactory spectra for the samples analyzed. The instrument
was operated in reflectron mode using delayed extraction techniques
sity. These results are consistent with BnOH being an effective
chain exchange reagent and with exchange being more rapid
than monomer enchainment.26
The observed effects of CD and EA on the polymerization
kinetics and polymer molecular weights may be rationalized
by invoking water as an adventitious impurity that reacts with
2a according to eq 5, assuming that L1ZnOH (or an oligomerized
form) is capable of rapid chain exchange but is inactive for chain
propagation. According to this hypothesis, substoichiometric
water decreases the amount of active catalyst (L1ZnOR) yet
produces twice the amount of chain exchange reagents (L1ZnOH
and EtOH) compared to the amount of catalyst deactivated. An
alternative scenario suggested by a reviewer may also be
envisioned wherein water acts to hydrolyze LA to lactic acid,
which exchanges with the more basic phenolate ligand to yield
an inactive carboxylate/alkoxide species. Decomposition of this
species would liberate 2 equiv of chain exchange agent for each
catalyst molecule deactivated. We recognize that although eq 5
or the above alternative adequately explains the experimental
results, we lack conclusive data with which to unambiguously
identify the nature of the trace reagents CD and EA.
2L1ZnOEt + H2O f L1ZnOEt + L1ZnOH + EtOH (5)
The remarkably rapid rate of LA polymerization by 2a is
worthy of note. Comparisons of kobs values at identical catalyst
concentrations were used to rank 2a within the set of most effi-
cient LA polymerization catalysts for which kinetic data are
available (Table 4), placing 2a faster than other Zn(II)-con-
taining catalysts. More insight is possible through comparison
of propagation rate constants measured under identical condi-
tions for reactions with identical rate laws. This is possible for
2a and its dinuclear counterpart LZn2(Cl)2OEt,4 for which kprop
(0.37 M-1 s-1) is almost 6 times smaller at 25 °C in CH2Cl2.
We speculate that the possible rate enhancing effect of the
nearby Zn(II) in the dinuclear catalyst is inhibited by the chloride
ligands, which yield metal sites with decreased Lewis acidity
and a higher coordination number (CN ) 4, not including the
alkoxide) than those in “L1Zn” (CN ) 3). The ring opening
polymerization activity of (â-diketiminate)Zn(II)-alkoxide cata-
lysts has also been observed to correlate with the tendency to
dimerize; bulkier ligands that favor the lower coordination num-
ber monomeric form induce greater polymerization activity.31
Finally, we note that the very fast rate of LA polymerization
by 2a is a potentially quite useful characteristic, as it indicates
a degree of kinetic lability that should enable ring opening poly-
merization of less reactive (i.e., low ring strain) monomers at
low temperatures to generate new types of useful materials (e.g.,
perfectly alternating copolymers of hydroxy acids and ep-
oxides).32
1
to improve resolution. The H PGSE experiments were performed on
a Varian INOVA 600 MHz spectrometer equipped with a triple axis
gradient unit and a BioSelect HCN triple axis probe with actively
shielded gradients. Samples were prepared by dissolving 5 mg of 1a,
2a, or 3 in 0.4 mL of CD2Cl2, and data were collected at 298 K without
spinning of the sample. A published pulse sequence was used to
suppress convection artifacts33 with a gradient length of 4 ms and a
diffusion time of 30 s. The data were analyzed as described in the text.
2,4-Di-tert-butyl-6-{[(2′-dimethylaminoethyl)methylamino]methyl}-
phenol (HL1). N,N,N′-Trimethylethylenediamine (7.90 g, 75.0 mmol),
para-formaldehyde (3.08 g, 97.5 mmol), and 2,4-di-tert-butylphenol
(15.63 g, 75.0 mmol) were dissolved in ethanol (50 mL). The solution
was heated at reflux for 14 h under nitrogen and then cooled to room
temperature. Hydrobromic acid (12.73 mL, 112.5 mmol, 48% solution)
was then added, and the liquid solution began to gel. This gel was
dried thoroughly under vacuum to yield a solid, which was washed
with ethanol to remove all of the yellow color. The white solid was
dissolved in water (300 mL), neutralized with NaHCO3, and extracted
into CHCl3 (3 × 100 mL). The combined organic layers were dried,
and the solvent was removed in vacuo to give a colorless oil (18.26 g,
1
4
56.97 mmol, 76%). H NMR (CDCl3, 300 MHz) δ 7.21 (1H, d, JH-H
) 2.7 Hz, ArH), 6.83 (1H, d, 4JH-H ) 2.4 Hz, ArH), 3.69 (2H, s, NCH2-
Ar), 2.54 (4H, m, NCH2CH2N), 2.33 (3H, s, NCH3), 2.23 (6H, s,
N(CH3)2), 1.42 (9H, s, C(CH3)3), 1.29 (9H, s, C(CH3)3) ppm; 13C {1H}
NMR (CDCl3, 75 MHz) δ 154.26, 140.15, 135.36, 123.33, 122.69,
121.28, 62.03, 57.00, 54.21, 45.68, 41.80, 34.78, 34.03, 31.67, 29.55
ppm. Anal. Calcd for C20H36N2O: C, 74.95%; H, 11.32%; N, 8.74%.
Found: C, 74.25%; H, 11.07%; N, 8.44%.
Experimental Section
General Methods and Materials. All reactions were performed
under a nitrogen atmosphere, using either standard Schlenk techniques
2-tert-Butyl-4-methyl-6-{[(2′-dimethylaminoethyl)methylamino]-
methyl}phenol (HL2). N,N,N′-Trimethylethylenediamine (3.10 g, 30.5
mmol), para-formaldehyde (1.20 g, 39.5 mmol), and 2-tert-butyl-4-
methylphenol (5.00 g, 30.5 mmol) were dissolved in ethanol (25 mL).
(31) Rieth, L. R.; Moore, D. R.; Lobkovsky, E. B.; Coates, G. W. J. Am. Chem.
Soc. 2002, 124, 15239.
(32) Bechtold, K.; Hillmyer, M. A.; Tolman, W. B. Macromolecules 2001, 34,
8641.
(33) Jerschow, A.; Mu¨ller, N. J. Magn. Reson. 1997, 125, 372.
9
J. AM. CHEM. SOC. VOL. 125, NO. 37, 2003 11357

A R T I C L E S
Williams et al.
The solution was heated at reflux for 14 h under nitrogen and cooled
to room temperature. Hydrobromic acid (5.00 mL, 45.8 mmol, 48%
solution) was then added, and the liquid solution began to gel. This
gel was dried thoroughly under vacuum to yield a solid, which was
washed with ethanol to remove all of the yellow color. The white solid
was dissolved in water (300 mL) and extracted with CH2Cl2 (3 × 100
mL). The aqueous layer was neutralized with NaHCO3 and further
extracted into CH2Cl2 (3 × 100 mL). The combined organic layers
were dried over Na2SO4, and the solvent was removed in vacuo to
(L2ZnOEt)2 (2b). Complex 1b (0.50 g, 1.34 mmol) was dissolved
in toluene (10 mL) and cooled to -35 °C. Ethanol (0.14 mL, 2.42
mmol) was added to this solution slowly. The mixture was stirred for
12 h at room temperature and then filtered through Celite. The filtrate
was dried in vacuo to give a white solid (0.47 g, 1.21 mmol, 90%).
This compound was recrystallized from toluene and pentane at -35
1
°C prior to use in LDMS and NMR crossover experiments. H NMR
(CD2Cl2, 300 MHz) δ 6.96 (1/2H, d, 4JH-H ) 4.0 Hz, ArH), 6.93 (1/2H,
4
4
d, JH-H ) 4.0 Hz, ArH), 6.61 (1H, d, JH-H ) 3.0 Hz, ArH), 3.88
(2H, q, 3JH-H ) 12.0 Hz, OCH2CH3), 3.77 (1H, s, NCH2Ar), 3.37 (1H,
d, 2JH-H ) 20.0 Hz, NCH2Ar), 2.77 (4H, m, NCH2CH2N), 2.59 (3H, s,
NCH3), 2.22 (3H, s, CH3Ar), 2.17 (6H, s, N(CH3)2), 1.40 (9H, s,
1
give a colorless oil (7.50 g, 23.2 mmol, 76%). H NMR (CDCl3, 300
MHz) δ 6.95 (1H, s, ArH), 6.62 (1H, s, ArH), 3.62 (2H, s, NCH2Ar),
2.51 (4H, m, NCH2CH2N), 2.29 (3H, s, NCH3), 2.21 (3H, s, NCH3),
2.11 (3H, s, NCH3), 1.38 (9H, s, C(CH3)3) ppm; 13C {1H} NMR (CDCl3,
75 MHz) δ 154.43, 136.01, 127.19, 126.42, 122.08, 61.51, 56.96, 54.10,
53.35, 45.54, 41.60, 34.48, 29.52, 20.72 ppm. HRMS calcd for
C17H30N2O1 (M+ + 1) 279.2358, found 279.2409.
3
C(CH3)3), 1.21 (3H, t, JH-H ) 11.5 Hz, OCH2CH3) ppm; 13C {1H}
NMR (CD2Cl2, 75 MHz) δ 164.09, 138.57, 130.01, 128.09, 121.28,
65.18, 57.07, 46.92, 45.18, 44.05, 35.41, 35.24, 29.99, 20.86 ppm. Anal.
Calcd for C19H34N2O2Zn: C, 58.81%; H, 8.83%; N, 7.25%. Found C,
58.61%; H, 8.97%; N, 7.91%.
L1ZnEt (1a). HL1 (2.00 g, 6.24 mmol) was dissolved in pentane (5
mL), and the solution was cooled to -35 °C. Diethyl zinc (6.24 mL of
a 1 M solution in hexane, 6.24 mmol) was added to this solution slowly,
and the mixture was stirred overnight at room temperature. The resulting
precipitate was collected by filtration, washed with cold pentane (2 ×
10 mL), and dried in vacuo (1.64 g, 3.96 mmol, 63%). Crystals suitable
for X-ray crystallographic analysis were grown from pentane (-35 °C).
Typical Polymerization Procedure. All glassware used for poly-
merizations was oven dried, treated with a 1.0 M solution of Me2SiCl2
in CH2Cl2, and dried for a minimum of an additional 3 h at 200 °C.
The solvent, CH2Cl2, was passed through a plug of alumina immediately
prior to use. A stock solution of catalyst, 2a, was prepared such that
[L1ZnOEt]0 ) 0.0023 M in CH2Cl2 and was stored under nitrogen at
-35 °C. A representative procedure is given below. In the glovebox,
a glass vial was charged with LA (0.180 g, 1.25 mmol, 1500 equiv)
and CH2Cl2 (1.21 mL). A volume (0.036 mL, 8.32 × 10-4 mmol, 1
equiv) of the catalyst stock solution was injected into the monomer
solution and stirred at room temperature. After stirring for an appropriate
time, an aliquot was withdrawn from the reaction and the product
polymer was precipitated by adding excess (2 mL) pentane, followed
by removal of the solvents in vacuo at RT.
4
1H NMR (CD2Cl2, 300 MHz) δ 7.16 (1H, d, JH-H ) 2.7 Hz, ArH),
4
2
6.78 (1H, d, JH-H ) 2.4 Hz, ArH), 3.66 (1H, d, JH-H ) 12 Hz,
2
NCH2C), 3.43 (1H, d, JH-H ) 12 Hz, NCH2C), 2.62 (4H, m,
NCH2CH2N), 2.50 (3H, s, NCH3), 2.37 (3H, s, NCH3), 2.08 (3H, s,
NCH3), 1.41 (9H, s, C(CH3)3), 1.25 (9H, s, C(CH3)3), 1.22 (3H, t, 3JH-H
3
)8.1 Hz, CH2CH3), 0.04 (2H, q, JH-H )8.1 Hz, CH2CH3) ppm; 13C
{1H} NMR (CD2Cl2, 75 MHz) δ 165.06, 137.80, 134.93, 126.27,
124.36, 122.74, 62.56, 57.96, 52.98, 47.78, 46.10, 45.97, 35.69, 34.24,
32.20, 30.05, 13.64, 4.00 ppm. Anal. Calcd for C22H40N2OZn: C,
63.83%; H, 9.74%; N, 6.77%. Found C, 63.50%; H, 9.59%; N, 6.71%.
Kinetics. Data for the polymerization of LA by 2a were collected
by an in situ monitoring of the changes in IR spectra using an ASI
Applied Systems ReactIR 4000 spectrometer. For the 25 °C runs, a
reaction flask was charged with LA and a solution of 2a in CH2Cl2 in
the glovebox (the total volume varied, but the average volume was
approximately 3 mL). The ReactIR probe was then attached to the flask
via a ground glass joint, and the apparatus was removed from the
glovebox and attached to the IR spectrometer so that the reaction flask
was immersed in a temperature controlled bath at 25 °C. For the 0 °C
runs, catalyst was not added in the glovebox, and the apparatus was
subsequently immersed in an ice-water bath. The solution temperature
was allowed to equilibrate over 5 min, and 2a (as a CH2Cl2 stock
solution prepared in the glovebox) was injected through a septum via
syringe. Peaks due to LA at 1345, 1240, and 930 cm-1 and peaks due
to PLA at 1184, 1147, 1131, and 863 cm-1 were monitored; analysis
of LA decay and PLA growth gave very similar results. The observed
rate constants (kobs) were extracted by fitting an exponential curve to
the plot of absorbance versus time using At ) (A0 - A∞)e-kt + A∞,
allowing A0, At, and k to vary freely and taking the average value from
the seven peaks monitored. The data were analyzed over at least four
half-lives. All linear and nonlinear curve fits were performed with
KaleidaGraph.
L2ZnEt (1b). HL2 (2.00 g, 6.24 mmol) was dissolved in pentane
(10 mL), and the solution was cooled to -35 °C. Diethyl zinc (7.10
mL of a 0.88 M solution in hexane, 6.24 mmol) was added to this
solution slowly, and the mixture was stirred overnight at room
temperature. The resulting precipitate was collected by filtration, washed
with cold pentane (2 × 10 mL), and dried in vacuo (1.51 g, 4.06 mmol,
1
4
65%). H NMR (CD2Cl2, 300 MHz) δ 6.94 (1H, d, JH-H ) 4.0 Hz,
4
2
ArH), 6.61 (1H, d, JH-H ) 3.5 Hz, ArH), 3.65 (1H, d, JH-H ) 20.0
Hz, NCH2C), 3.39 (1H, d, JH-H ) 20.0 Hz, NCH2C), 2.78 (4H, m,
2
NCH2CH2N), 2.49 (3H, s, NCH3), 2.37 (3H, s, ArCH3), 2.17 (3H, s,
NCH3), 2.12 (3H, s, NCH3), 1.40 (9H, s, C(CH3)3), 1.22 (3H, t, 3JH-H
) 14.0 Hz, CH2CH3), 0.05 (2H, q, 3JH-H )8.1 Hz, CH2CH3) ppm; 13
C
{1H} NMR (CD2Cl2, 75 MHz) δ 165.16, 138.632, 130.08, 128.27,
121.08, 62.25, 57.99, 52.95, 47.77, 46.21, 45.95, 35.37, 30.05, 20.95,
13.61, -4.06 ppm. Anal. Calcd for C19H34N2OZn: C, 61.35%; H,
9.21%; N, 7.56%. Found C, 59.90%; H, 8.90%; N, 7.31%.
(L1ZnOEt)2 (2a). Complex 1a (1.00 g, 2.42 mmol) was dissolved
in toluene (13 mL) and cooled to -35 °C. Ethanol (0.14 mL, 2.42
mmol) was added to this solution slowly. The mixture was stirred for
1 d at room temperature and then filtered through Celite. The filtrate
was dried in vacuo to give a white solid (0.85 g, 1.98 mmol, 82%).
Crystals suitable for X-ray diffraction were grown from a mixture of
toluene and pentane at -35 °C. 1H NMR (CD2Cl2, 300 MHz, see Figure
S1) δ 7.19 (1H, d, 4JH-H ) 2.4 Hz, ArH), 6.79 (1H, d, 4JH-H ) 2.7 Hz,
X-ray Crystallography. Single crystals of 1a and 2a were attached
to glass fibers and mounted on a Siemens or Bruker SMART system
for data collection at 173(2) K. An initial set of cell constants was
calculated from three sets of 20 frames. These initial sets of frames
were oriented such that orthogonal wedges of reciprocal space were
surveyed; orientation matrixes were calculated from 90 to 150 reflec-
tions. The data collection was performed using Mo KR radiation with
a frame time of 20 and 45 s, respectively, and a detector distance of
4.9 cm. A randomly oriented region of reciprocal space was surveyed
to the extent of 1.5 hemispheres and to a resolution of 0.84 Å. Three
major sections of frames were collected with 0.30° steps in ω at three
different φ settings and a detector position of -28° in 2θ. The intesity
data were corrected for absorption and decay (SADABS).34 Final cell
3
2
ArH), 3.92 (2H, q, JH-H ) 6.8 Hz, OCH2CH3), 3.80 (1H, d, JH-H
)
2
10.2 Hz, NCH2C), 3.53 (1H, d, JH-H ) 12.0 Hz, NCH2C), 2.76 (4H,
m, NCH2CH2N), 2.55 (3H, s, NCH3), 2.32 (6H, m, N(CH3)2), 1.43 (9H,
3
s, C(CH3)3), 1.26 (9H, s, C(CH3)3), 1.21 (3H, t, JH-H ) 6.8 Hz,
OCH2CH3) ppm; 13C {1H} NMR (CD2Cl2, 75 MHz) δ 164.43, 138.13,
135.59, 126.48, 124.57, 121.63, 63.29, 62.36, 57.24, 52.98, 47.27, 45.17,
35.72, 34.25, 32.18, 30.07, 22.83 ppm. Anal. Calcd for C22H40N2O2-
Zn: C, 61.46%; H, 9.38%; N, 6.52%. Found C, 61.58%; H, 9.32%; N,
6.16%.
9
11358 J. AM. CHEM. SOC. VOL. 125, NO. 37, 2003

Zinc Catalyst for Controlled Lactide Polymerization
A R T I C L E S
constants were calculated from the xyz centroids of strong reflections
from the actual data collection after integration (SAINT).35 The
structures were solved by direct methods using SIR9236 and SIR97,37
which located most of the non-hydrogen atoms from the E-map. Full-
matrix least squares/difference Fourier cycles were performed using
SHELXL-97.38 Space groups were determined based on systematic
absences and intensity statistics. All non-hydrogen atoms were refined
with anisotropic displacement parameters. All hydrogen atoms were
placed in ideal positions and refined as riding atoms with relative
isotropic displacement parameters. For 2a, solvent molecules (toluene,
pentane) were found in channels throughout the crystal and could not
be individually modeled. To obtain a better model, the solvent molecules
were removed using program PLATON, function SQUEEZE.36 Full
crystallographic information for each complex is available in its
CIF.
Acknowledgment. Financial support for this research was
provided by the NSF (CHE9975357) and the Korean Ministry
of Science and Technology (Creative Research Initiative
Program). In addition, NMR instrumentation for PGSE mea-
surements was provided with funds from the NSF (BIR-961477),
the University of Minnesota Medical School, and the Minnesota
Medical Foundation. We thank Dr. Luis Alcazar Roman and
Prof. Clark Landis (U. Wisconsin, Madison) for helpful dis-
cussions, Dr. Dana Reed for technical assistance with the laser
ablation experiments, and Dr. Letitia Yao and Dr. Irina
Nesmelova for technical assistance with PGSE measurements.
Supporting Information Available: NMR spectrum of 2a
(Figure S1), representative kinetic trace (Figure S2), listing of
kobs values (Table S1), and X-ray crystallographic data (CIF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
(34) An empirical correction for absorption anisotropy. Blessing, R. Acta
Crystallogr. 1995, A51, 33.
(35) SAINT V6.2; Bruker Analytical X-ray Systems: Madison, WI, 2001.
(36) (a) Spek, A. L. Acta Crystallogr. 1990, A46, C34. (b) Spek, A. L. PLATON,
A Multipurpose Crystallographic Tool; Utrecht University: Utrecht, The
Netherlands, 2000.
(37) Altomare, A.; Burla, M. C.; Camalli, M.; Cascarano, G.; Giacovazzo, C.;
Guargliardi, A.; Moliterni, A. G. G.; Polidori, G.; Spagna, R. SIR97: a
new tool for crystal structure determination and refinement. J. Appl.
Crystallogr. 1998, 32, 115.
JA0359512
(38) SHELXTL V6.10; Bruker Analytical X-ray Systems: Madison, WI, 2000.
9
J. AM. CHEM. SOC. VOL. 125, NO. 37, 2003 11359