Dioxygen activation at a single copper site: Structure, bonding, and mechanism of formation of 1:1 Cu-O2 adducts

Article abstract of DOI:10.1021/ja045678j

To evaluate the fundamental process of O₂ activation at a single copper site that occurs in biological and catalytic systems, a detailed study of O₂ binding to Cu(I) complexes of β-diketiminate ligands L (L¹ = backbone Me; L² = backbone tBu) by X-ray crystallography, X-ray absorption spectroscopy (XAS), cryogenic stopped-flow kinetics, and theoretical calculations was performed. Using synchrotron radiation, an X-ray diffraction data set for L²CuO₂ was acquired, which led to structural parameters in close agreement to theoretical predictions. Significant Cu(III) - peroxo character for the complex was corroborated by XAS. On the basis of stopped-flow kinetics data and theoretical calculations for the oxygenation of L¹Cu(RCN) (R = alkyl, aryl) in THF and THF/RCN mixtures between 193 and 233 K, a dual pathway mechanism is proposed involving (a) rate-determining solvolysis of RCN by THF followed by rapid oxygenation of L¹Cu(THF) and (b) direct, bimolecular oxygenation of L¹Cu(RCN) via an associative process.

Full text of DOI:10.1021/ja045678j

Published on Web 12/04/2004
Dioxygen Activation at a Single Copper Site: Structure,
Bonding, and Mechanism of Formation of 1:1 Cu-O2 Adducts
Nermeen W. Aboelella,† Sergey V. Kryatov,‡ Benjamin F. Gherman,†
William W. Brennessel,† Victor G. Young, Jr.,† Ritimukta Sarangi,§
Elena V. Rybak-Akimova,‡ Keith O. Hodgson,§,¶ Britt Hedman,¶ Edward I. Solomon,§
Christopher J. Cramer,† and William B. Tolman*,†
Contribution from the Department of Chemistry, Center for Metals in Biocatalysis,
and Supercomputer Institute, UniVersity of Minnesota, 207 Pleasant Street SE,
Minneapolis, Minnesota 55455, Department of Chemistry, Tufts UniVersity, 62 Talbot AVenue,
Medford, Massachusetts, 02155, Department of Chemistry, Stanford UniVersity,
Stanford, California 94305, and Stanford Synchrotron Radiation Laboratory,
Stanford UniVersity, Stanford, California 94305
Received July 19, 2004; Revised Manuscript Received October 3, 2004; E-mail: tolman@chem.umn.edu
Abstract: To evaluate the fundamental process of O2 activation at a single copper site that occurs in
biological and catalytic systems, a detailed study of O2 binding to Cu(I) complexes of â-diketiminate ligands
L (L1 ) backbone Me; L2 ) backbone tBu) by X-ray crystallography, X-ray absorption spectroscopy (XAS),
cryogenic stopped-flow kinetics, and theoretical calculations was performed. Using synchrotron radiation,
an X-ray diffraction data set for L2CuO2 was acquired, which led to structural parameters in close agreement
to theoretical predictions. Significant Cu(III)-peroxo character for the complex was corroborated by XAS.
On the basis of stopped-flow kinetics data and theoretical calculations for the oxygenation of L1Cu(RCN)
(R ) alkyl, aryl) in THF and THF/RCN mixtures between 193 and 233 K, a dual pathway mechanism is
proposed involving (a) rate-determining solvolysis of RCN by THF followed by rapid oxygenation of
L1Cu(THF) and (b) direct, bimolecular oxygenation of L1Cu(RCN) via an associative process.
Introduction
with two His and 1 Met (CuM) provided by the protein. Kinetic
data support a mechanism for substrate attack by DâM and PHM
A critical function of copper centers in enzymes1,2 and other
catalysts3 is to bind and/or activate dioxygen. In particular, O2
activation by monocopper active sites is a central part of catalytic
cycles traversed by enzymes involved in diverse metabolic
processes.4 Notable examples are dopamine â-monooxygenase
(DâM)2,5 and peptidylglycine R-hydroxylating monooxygenase
(PHM),2,6 which hydroxylate aliphatic C-H bonds of their
respective substrates, dopamine or peptide hormones, to yield
the neurotransmitter norepinephrine (DâM) or, after subsequent
chemistry, fully bioactive C-terminal amidated peptides (PHM).
Extensive similarities between these enzymes implicate analo-
gous active site structures and catalytic mechanisms. X-ray
crystallographic6,7 and EXAFS8 data for PHM have revealed
two copper sites, one with three His ligands (CuH) and the other
that involves abstraction of an H-atom with quantum mechanical
tunneling.9 While there appears to be consensus that a Cu-O2
intermediate is involved in this process, the nature of the inter-
mediate(s) and the pathways for its(their) formation and sub-
sequent reactions are not known. Recent kinetic,9e theoretical,10
and X-ray crystallographic7c results suggest that a Cu(II)-O2
-
moiety at CuM is responsible for the H-atom abstraction. Many
aspects of the properties of this species remain obscure, however,
and the possible role of alternative intermediates continues to
be debated (e.g., Cu(II)-OOH).11 Similar Cu-O2 species have
been suggested as intermediates in H2O2 production by amine
(7) Prigge, S. T.; Kolhekar, A. S.; Eipper, B. A.; Mains, R. E.; Amzel, L. M.
Science 1997, 278, 1300. (b) Prigge, S. T.; Kolhekar, A. S.; Eipper, B. A.;
Mains, R. E.; Amzel, L. M. Nat. Struct. Biol. 1999, 6, 976. (c) Prigge, S.
T.; Eipper, B. A.; Mains, R. E.; Amzel, L. M. Science 2004, 304, 864.
(8) Boswell, J. S.; Reddy, B. J.; Kulathila, R.; Merkler, D.; Blackburn, N. J.
Biochemistry 1996, 35, 12241. (b) Blackburn, N. J.; Rhames, F. C.; Ralle,
M.; Jaron, S. J. Biol. Inorg. Chem. 2000, 5, 341. (c) Jaron, S.; Blackburn,
N. J. Biochemistry 2001, 40, 6867. (d) Rhames, F. C.; Murthy, N. N.; Karlin,
K. D.; Blackburn, N. J. J. Biol. Inorg. Chem. 2001, 6, 567. (e) Jaron, S.;
Mains, R. E.; Eipper, B. A.; Blackburn, N. J. Biochemistry 2002, 41, 13274.
(9) Tian, G.; Berry, J. A.; Klinman, J. P. Biochemistry 1994, 33, 226. (b)
Francisco, W. A.; Merkler, D. J.; Blackburn, N. J.; Klinman, J. P.
Biochemistry 1998, 37, 8244. (c) Francisco, W. A.; Knapp, M. J.; Blackburn,
N. J.; Klinman, J. P. J. Am. Chem. Soc. 2002, 124, 8194. (d) Francisco,
W. A.; Blackburn, N. J.; Klinman, J. P. Biochemistry 2003, 42, 1813. (e)
Evans, J. P.; Ahn, K.; Klinman, J. P. J. Biol. Chem. 2003, 278, 49691.
(10) Chen, P.; Solomon, E. I. J. Am. Chem. Soc. 2004, 126, 4991.
(11) Chen, P.; Bell, J.; Eipper, B. A.; Solomon, E. I. Biochemistry 2004, 43,
5735.
† University of Minnesota.
‡ Tufts University.
§ Department of Chemistry, Stanford University.
¶ Stanford Synchrotron Radiation Laboratory, Stanford University.
(1) Solomon, E. I.; Chen, P.; Metz, M.; Lee, S.-K.; Palmer, A. E. Angew. Chem.,
Int. Ed. 2001, 40, 4570.
(2) Klinman, J. P. Chem. ReV. 1996, 96, 2541.
(3) For a recent example, see: Marko´, I. E.; Gautier, A.; Dumeunier, R.; Doda,
K.; Philippart, F.; Brown, S. M.; Urch, C. J. Angew. Chem., Int. Ed. 2004,
43, 1588.
(4) Halcrow, M. A. In ComprehensiVe Coordination Chemistry II; McCleverty,
J. A., Meyer, T. J., Eds.; Elsevier: Amsterdam, 2004; Vol. 8, pp 395-436.
(5) Stewart, L. C.; Klinman, J. P. Annu. ReV. Biochem. 1988, 57, 551.
(6) Prigge, S. T.; Mains, R. E.; Eipper, B. A.; Amzel, L. M. Cell. Mol. Life
Sci. 2000, 57, 1236.
9
16896
J. AM. CHEM. SOC. 2004, 126, 16896-16911
10.1021/ja045678j CCC: $27.50 © 2004 American Chemical Society

Dioxygen Activation at a Single Copper Site
A R T I C L E S
oxidases (AOs) and galactose oxidase (GAO),12-14 as well as
in the self-processing biogenesis of their respective cofactors
(topaquinone and S-Cys-modified tyrosine).15 However, direct
structural or spectroscopic information is not available for any
such species in AO or GAO, with the exception of X-ray
structural evidence for product H2O2 near the Cu site in
Escherichia coli AO.16
has been defined, wherein for S ) THF, faster reactions and
accentuated supporting ligand electronic influences are observed
relative to those when S ) RCN.27,28 Activation parameters for
these and other related systems22b have been interpreted to
indicate dissociative or associative processes for the oxygenation
process, depending on the nature of S or the supporting ligand
L. Notwithstanding these advances, there is a general lack of
knowledge of the elementary reaction steps involved in the
displacement of S from Cu(I) by O2 and of the structure(s) of
the key transition state(s) and/or intermediates involved in
oxygenations that yield well-defined 1:1 Cu-O2 adducts.
Significant insights into dioxygen activation by copper
proteins have been obtained through synthetic modeling studies,
with peroxo- and bis(µ-oxo)dicopper complexes having been
especially thoroughly examined.17 Detailed understanding of
monocopper-dioxygen species is less developed, in part because
in reactions of Cu(I) complexes with O2, 1:1 Cu-O2 intermedi-
ates are often no more than transient species that are rapidly
trapped by a second Cu(I) ion to yield relatively stable dicopper
complexes.4,17b Thus, only a few examples of isolable and well-
defined synthetic monocopper-dioxygen species have been
reported (Table S1 in the Supporting Information). A side-on
(η2) Cu-O2 complex supported by a hindered tris(pyrazolyl)-
hydroborate (Tp) ligand has been characterized by X-ray
crystallography18 and identified (along with an analog) as a
Cu(II)-O2- species on the basis of spectroscopy and theory.19
An end-on (η1) complex has been identified by FTIR spec-
troscopy and shown to evolve H2O2 upon protonation.20 Ki-
netic, UV-vis, and resonance Raman evidence for a stable
In recent explorations of the O2 chemistry of Cu(I) complexes
of â-diketiminate ligands, we found that bis(µ-oxo)dicopper
complex formation29 could be inhibited by using the sterically
hindered ligands L1 and L2.30,31 Oxygenation of LCu(MeCN)
(L ) L1 or L2) at low temperature yielded stable η2 1:1 Cu-O2
adducts that were characterized by NMR, UV-vis, EPR, and
resonance Raman spectroscopy, DFT calculations, and a pre-
liminary X-ray crystal structure for L ) L2 (Table S1 in the
Supporting Information). Notably, low values for the O-O
stretching frequency (e.g., for L ) L2, ν(16O2) ) 961 cm-1 and
∆ν(18O2) ) 49 cm-1) suggested significant contribution of a
Cu(III)-O22- resonance form. This conclusion was corroborated
by the relatively long O-O distance calculated (1.376 Å) and
observed by crystallography (1.44 Å),31,32 although the reliability
of the latter was mitigated by the relatively poor quality of the
X-ray structural data coupled with extreme disorder of the
molecule in the crystal.
-
Cu(II)-O2 complex of a tetradentate N3O donor ligand also
have been presented;21 similar data have appeared for ana-
logous yet less thermodynamically stable examples.22 Finally,
Cu(II)-OOR (R ) acyl, alkyl, or H) complexes have been
isolated that are relevant in the current context, although they
are not obtained from reactions of Cu(I) complexes with O2.4,23
Extensive kinetics studies of the oxygenation of Cu(I) com-
plexes have provided some benchmark activation and thermo-
dynamic parameters for 1:1 adduct formation, although the level
of mechanistic detail available is limited.17b,24-26 Recently, the
importance of the solvent ligand (S) in (TMPA)Cu(I) complexes
(TMPA ) tris(pyridylmethyl)amine) on the rate of oxygenation
Herein, we report revised X-ray crystallographic results for
L2CuO2 based on a better quality data set acquired using a
synchrotron radiation source. Further insight into the geometry
(12) Halcrow, M.; Phillips, S.; Knowles, P. In Subcellular Biochemistry, Vol.
35: Enzyme-Catalyzed Electron and Radical Transfer; Holzenburg, A.,
Scrutton, N. S., Eds.; Plenum: New York, 2000; pp 183-231.
(13) Klinman, J. P. J. Biol. Inorg. Chem. 2001, 6, 1.
(23) Fujisawa, K.; Kobayashi, T.; Kitajima, N.; Moro-oka, Y. J. Am. Chem.
Soc. 1996, 118, 1. (b) Kitajima, N.; Katayama, T.; Fujisawa, K.; Iwata,
Y.; Moro-oka, Y. J. Am. Chem. Soc. 1993, 115, 7872. (c) Sanyal, I.; Ghosh,
P.; Karlin, K. Inorg. Chem. 1995, 34, 3050. (d) Chen, P.; Fujisawa, K.;
Solomon, E. I. J. Am. Chem. Soc. 2000, 122, 10177. (e) Wada, A.; Harata,
M.; Hasegawa, K.; Jitsukawa, K.; Masuda, H.; Mukai, M.; Kitagawa, T.;
Einaga, H. Angew. Chem., Int. Ed. 1998, 37, 798. (f) Kodera, M.; Kita, T.;
Miura, I.; Nakayama, N.; Kawata, T.; Kano, K.; Hirota, S. J. Am. Chem.
Soc. 2001, 123, 7715. (g) Yamaguchi, S.; Nagatomo, S.; Kitagawa, T.;
Funahashi, Y.; Ozawa, T.; Jitsukawa, K.; Masuda, H. Inorg. Chem. 2003,
42, 6968. (h) Fujii, T.; Naito, A.; Yamaguchi, S.; Wada, A.; Funahashi,
Y.; Jitsukawa, K.; Nagatomo, S.; Kitagawa, T.; Masuda, H. Chem. Commun.
2003, 2700.
(14) Whittaker, J. W. Chem. ReV. 2003, 103, 2347.
(15) Dooley, D. M. J. Biol. Inorg. Chem. 1999, 4, 1. (b) Rogers, M. S.; Baron,
A. J.; McPherson, M. J.; Knowles, P. F.; Dooley, D. M. J. Am. Chem. Soc.
2000, 122, 990. (c) Firbank, S. J.; Rogers, M. S.; Wilmot, C. M.; Dooley,
D. M.; Halcrow, M. A.; Knowles, P. F.; McPherson, M. J. Proc. Natl.
Acad. Sci. U.S.A. 2001, 98, 12932.
(16) Wilmot, C. M.; Hajdu, J.; Mcpherson, M. J.; Knowles, P. F.; Phillips, S.
E. V. Science 1999, 286, 1724.
(17) For selected recent reviews, see: (a) Mirica, L. M.; Ottensaelder, X.; Stack,
T. D. P. Chem. ReV. 2004, 104, 1013. (b) Lewis, E. A.; Tolman, W. B.
Chem. ReV. 2004, 104, 1047. (c) Itoh, S. In ComprehensiVe Coordination
Chemistry II; McCleverty, J. A., Meyer, T. J., Eds.; Elsevier: Amsterdam,
2004; Vol. 8, pp 369-393. (d) Stack, T. D. P. Dalton Trans. 2003, 1881.
(e) Que, L., Jr.; Tolman, W. B. Angew. Chem., Int. Ed. 2002, 41, 1114. (f)
Kopf, M.-A.; Karlin, K. D. In Biomimetic Oxidations Catalyzed by
Transition Metal Complexes; Meunier, B., Ed.; Imperial College Press:
London, 2000; pp 309-362.
(24) Karlin, K. D.; Tolman, W. B.; Kaderli, S.; Zuberbu¨hler, A. D. J. Mol. Catal.
A 1997, 117, 215.
(25) Schindler, S. Eur. J. Inorg. Chem. 2000, 2311.
(26) For a previous report on the kinetics of the formation of a 1:1 Cu/O2 adduct
supported by BQPA (a bis(quinolyl) derivative of TMPA) that has been
revised (2:1 species are involved instead), see: (a) Karlin, K. D.; Wei, N.;
Jung, B.; Kaderli, S.; Niklaus, P.; Zuberbu¨hler, A. D. J. Am. Chem. Soc.
1993, 115, 9506. (b) Wick, P. K.; Karlin, K. D.; Suzuki, M.; Zuberbu¨hler,
A. D. Micron 2004, 35, 137.
(18) Fujisawa, K.; Tanaka, M.; Moro-oka, Y.; Kitajima, N. J. Am. Chem. Soc.
1994, 116, 12079.
(19) Chen, P.; Root, D. E.; Campochiaro, C.; Fujisawa, K.; Solomon, E. I. J.
Am. Chem. Soc. 2003, 125, 466.
(27) Zhang, C. X.; Kaderli, S.; Costas, M.; Kim, E.; Neuhold, Y.-M.; Karlin,
K. D.; Zuberbu¨hler, A. D. Inorg. Chem. 2003, 42, 1807.
(28) Fry, H. C.; Scaltrito, D. V.; Karlin, K. D.; Meyer, G. J. J. Am. Chem. Soc.
2003, 125, 11866.
(20) Chaudhuri, P.; Hess, M.; Weyhermu¨ller, T.; Wieghardt, K. Angew. Chem.,
Int. Ed. 1999, 38, 1095.
(21) Jazdzewski, B. A.; Reynolds, A. M.; Holland, P. L.; Young, V. G., Jr.;
Kaderli, S.; Zuberbu¨hler, A. D.; Tolman, W. B. J. Biol. Inorg. Chem. 2003,
8, 381.
(22) Komiyama, K.; Furutachi, H.; Nagatomo, S.; Hashimoto, A.; Hayashi, H.;
Fujinami, S.; Suzuki, M.; Kitagawa, T. Bull. Chem. Soc. Jpn. 2004, 77,
59. (b) Weitzer, M.; Schindler, S.; Brehm, G.; Schneider, S.; Ho¨rmann,
E.; Jung, B.; Kaderli, S.; Zuberbu¨hler, A. D. Inorg. Chem. 2003, 42, 1800.
(29) Spencer, D. J. E.; Reynolds, A. M.; Holland, P. L.; Jazdzewski, B. A.;
Duboc-Toia, C.; Pape, L. L.; Yokota, S.; Tachi, Y.; Itoh, S.; Tolman, W.
B. Inorg. Chem. 2002, 41, 6307.
(30) Spencer, D. J. E.; Aboelella, N. W.; Reynolds, A. M.; Holland, P. L.;
Tolman, W. B. J. Am. Chem. Soc. 2002, 124, 2108.
(31) Aboelella, N. W.; Lewis, E. A.; Reynolds, A. M.; Brennessel, W. W.;
Cramer, C. J.; Tolman, W. B. J. Am. Chem. Soc. 2002, 124, 10660.
9
J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004 16897

A R T I C L E S
Aboelella et al.
K, and dry O2 was bubbled through it for approximately 30 min. The
solution became bright green, and a green precipitate began to form.
The mixture was then stored at 193 K overnight. The acetone mother
liquor was removed via cannula to leave behind a green solid. This
solid was washed with pentane at 193 K several times and then dried
under vacuum while maintaining the temperature at 193 K (∼0.135 g,
69%).
X-ray Crystallography. (A) [(L1Cu)2(p-NC-C6H4-CN)]. Data col-
lection details are provided in Table S2 and in the Supporting
Information as a CIF. The structure was solved using Sir9733 and refined
using SHELXL-97.34 A highly disordered solvent molecule (on a 4h
position) could not be modeled appropriately and was removed using
the PLATON program, SQUEEZE function.35 A total of 53 electrons
in a volume of 531.2 Å3 in the unit cell were determined. These
electrons were distributed among two main voids with volumes of 167
and 168 Å3 containing 26 electrons each, corresponding to two solvent
molecules per cell. However, the identity of the solvent based on the
volume of the void space and number of electrons could not be
determined. As a result, various fields in Table S2 are incorrect (e.g.,
formula, formula weight, and density). One isopropyl group (C24, C25,
C26) contains a slight disorder, but attempts to model this disorder
were unsuccessful.
(B) [L1Cu(p-NC-C6H4-OMe). Data collection details are provided
in the Supporting Information (Table S2 and a CIF). The structure was
solved using Sir9733 and refined using SHELXL-97.34 The asymmetric
unit contains two independent molecules, and in one molecule (A),
the methoxy group is disordered over two positions (72:28). This
disorder was modeled by splitting the OMe unit into two parts; O1A
was split into O1A and O1A′ using identical coordinates and displace-
ment parameters (applying EADP and EXYZ commands), and C37A
was split into C37A and C37A′.
(C) L2CuO2. Crystals suitable for X-ray crystallography were grown
at 193 K from a 50:50 THF/pentane solution of L2Cu(MeCN), which
was treated with O2 at 193 K. A crystal (approximate dimensions of
2.0 × 0.02 × 0.01 mm) was placed onto the tip of a 0.1 mm diameter
glass capillary and mounted on a Bruker Kappa SMART 6000 system
for data collection at 100(1) K. The data collection was carried out
using 0.5500 Å radiation (double-diamond monochromator) with a
frame time of 1 s and a detector distance of 4.9 cm at APS
ChemMatCARS 15-ID-C. A randomly oriented region of reciprocal
space was surveyed to the extent of 2.0 hemispheres and to a resolution
of 0.76 Å. The beam was 100 × 100 µ2 and was centered on a section
of the 2 mm long needle. One set of frames was collected with 0.30°
steps in a complete rotation of phi. The intensity data were corrected
for absorption and decay (SADABS).36 Final cell constants were
calculated from 3281 strong reflections from the actual data collection
after integration (SAINT 6.35A, 2002).37 Additional crystallographic
information is presented in Table S2.
and electronic structure of the 1:1 Cu-O2 adduct has been
provided by X-ray absorption spectroscopy. In addition, we have
completed a cryogenic stopped-flow kinetic study of the
formation of L1CuO2 from the reaction of O2 with L1Cu(NCR)
complexes and have analyzed the oxygenation reaction
pathway(s) by theoretical methods. On the basis of the com-
bined and cross-validated experimental and theoretical results,
a detailed mechanism for the oxygenation reaction is proposed.
Experimental Section
General Considerations. All reagents were obtained from com-
mercial sources and used as received unless stated otherwise. The
solvents, tetrahydrofuran (THF), heptane, and pentane, were distilled
from Na/benzophenone or passed through solvent purification columns
(Glass Contour, Laguna, CA). Acetonitrile (CH3CN) was distilled from
calcium hydride. Acetone was distilled from anhydrous calcium sulfate.
All metal complexes were prepared and stored in a vacuum atmospheres
inert atmosphere glovebox under a dry nitrogen atmosphere or were
manipulated using standard Schlenk techniques. The Cu(I) compounds,
LCu(MeCN) (L ) L1 or L2), and their O2 adducts, LCuO2, were
prepared as reported previously.29-31 NMR spectra were recorded on a
1
Varian VI-300 or VXR-300 spectrometer. Chemical shifts (δ) for H
or 13C NMR spectra were referenced to residual protium in the
deuterated solvent. UV-vis spectra were recorded on an HP8453
(190-1100 nm) diode array spectrophotometer. Low-temperature
spectra were acquired using a custom-manufactured vacuum dewar
equipped with quartz windows, with low temperatures achieved with
the use of a low-temperature MeOH bath circulator.
[(L1Cu)2(p-NC-C6H4-CN)]. A solution of L1Cu(MeCN) (0.121 g,
0.232 mmol) in THF (2 mL) was added to a stirring solution of 1,4-
dicyanobenzene (0.015 g, 0.117 mmol) in THF (2 mL). A deep purple
color developed upon complete addition of L1Cu(MeCN). After the
mixture was stirred for 30 min, solvent was removed under reduced
pressure, and the resulting residue was washed with 2-3 mL of heptane,
collected by filtration, and then dried in vacuo to yield a purple solid
(67 mg, 53%). X-ray quality crystals were obtained by cooling a
1
concentrated pentane solution to 253 K. H NMR (300 MHz, C6D6):
δ 7.06-7.19 (m, 6H), 6.03 (br s, 2H), 5.00 (s, 1H), 3.54 (heptet, J )
6.9 Hz, 4H), 1.85 (s, 6H), 1.39 (d, J ) 6.9 Hz, 12H), 1.27 (d, J ) 6.9
Hz, 12H). 13C{1H} NMR (75 MHz, C6D6): δ 163.63, 148.94, 141.04,
131.92, 124.04, 123.73, 116.78, 115.25, 94.00, 28.56, 24.91, 23.86,
23.75. UV-vis (THF) [λmax, nm (ꢀ, M-1 cm-1)]: 281 (31100), 349
(35700), 525 (5300). Anal. Calcd for C66H86N6Cu2: C, 72.69; H, 7.95;
N, 7.71. Found: C, 72.05; H, 8.45; N, 7.76.
L1Cu(p-NC-C6H4-OMe). A solution of 4-methoxybenzonitrile
(0.031 g, 0.233 mmol) in THF (2 mL) was added to a stirring solution
of L1Cu(MeCN) (0.096 g, 0.184 mmol) in THF (2 mL). After the
mixture was stirred for 45 min, solvent was removed under reduced
pressure to yield a yellow residue. The residue was extracted with
approximately 10 mL of pentane, and the extract was filtered through
Celite. The volume of the filtrate was reduced to approximately 1 mL
and placed in a 253 K freezer. After several days, large orange crystals
of the product formed, which were isolated, washed with a minimal
amount of cold pentane, and dried in vacuo (0.071 g, 63%). 1H NMR
(300 MHz, C6D6): δ 7.10-7.23 (m, 6H), 6.52 (m, 2H), 6.15 (m, 2H),
5.03 (s, 1H), 3.63 (heptet, J ) 6.9 Hz, 4H), 2.90 (s, 3H), 1.89 (s, 6H),
1.47 (d, J ) 6.9 Hz, 12H), 1.30 (d, J ) 6.9 Hz, 12H). 13C{1H} NMR
(75 MHz, C6D6): δ 163.45, 163.28, 149.24, 141.08, 134.38, 123.78,
123.67, 118.74, 114.96, 93.76, 55.11, 28.61, 24.94, 23.89, 23.80. UV-
vis (THF) [λmax, nm (ꢀ, M-1 cm-1)]: 249 (27800), 273 (26800), 352
(24800), 413 (sh, 2700). Anal. Calcd for C66H86N6Cu2: C, 72.34; H,
7.88; N, 6.84. Found: C, 72.17; H, 8.50; N, 6.82.
The structure was solved using SHELXS-86 and refined using
SHELXL-97.34 The space group Imm2 was determined based on
systematic absences and intensity statistics. A direct-methods solution
was calculated, which provided most non-hydrogen atoms from the
E-map. Full-matrix least-squares difference Fourier cycles were
performed, which located the remaining non-hydrogen atoms. All non-
hydrogen atoms were refined with anisotropic displacement parameters
unless stated otherwise. All hydrogen atoms were placed in ideal
(32) Cramer, C. J.; Tolman, W. B.; Theopold, K. H.; Rheingold, A. L. Proc.
Natl. Acad. Sci. U.S.A. 2003, 100, 3635.
(33) Altomare, A.; Burla, M. C.; Camalli, M.; Cascarano, G.; Giacovazzo, C.;
Guagliardi, A.; Moliterni, G. G.; Polidori, R.; Spagna, R. J. Appl.
Crystallogr. 1998, 119, 115.
(34) SHELXTL-Plus, version 6.10; Bruker Analytical X-ray Systems: Madison,
WI.
(35) Spek, A. L. Acta Crystallogr. 1990, A46, C34. (b) Spek, A. L. PLATON,
A Multipurpose Crystallographic Tool; Utrecht University: Utrecht, The
Netherlands, 2000.
(36) For an empirical correction for absorption anisotropy, see: Blessing, R.
Acta Crystallogr. 1995, A51, 33-38.
(37) SAINT, version 6.35A; Bruker Analytical X-ray Systems: Madison, WI.
Isolation of Solid L1CuO2. A solid sample for XAS measurements
was obtained as follows. Solid L1Cu(MeCN) (0.198 g, 0.379 mmol)
was dissolved in acetone (∼20 mL). The solution was cooled to 193
9
16898 J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004

Dioxygen Activation at a Single Copper Site
A R T I C L E S
positions and refined as riding atoms with relative isotropic displacement
parameters.
taken in large excess to the copper(I) complexes (0.1-0.4 mM). In all
kinetic experiments, a series of 5-7 shots gave standard deviations
within 5%, with overall reproducibility within 10%. Data analysis was
performed with the IS-2 Rapid Kinetics Software (Hi-Tech Scientific)
for kinetic traces at a single wavelength or with the Specfit program
(BioLogic Science Instruments, Grenoble, France) for global fitting of
the spectral changes acquired in a diode array mode. Standard deviations
of the rate constants and activation parameters were either calculated
by the Specfit program or determined from linear fits using the method
of least squares.41
Computational Methods. An earlier study42 modeled η2 1:1
Cu-O2 adducts supported by a simplified version of the full â-diketim-
inate ligands L1 and L2. Geometries obtained from restricted and
unrestricted density functional theory (DFT) methods were found to
be optimal for the singlet and triplet states of the Cu-O2 adducts,
respectively, while the multideterminantal character of the singlet
adducts mandated the use of multireference second-order perturbation
theory for the prediction of system energies. The computational protocol
employed herein is based closely upon these findings.
X-ray Absorption Spectroscopy. X-ray absorption spectra of
L1CuO2 were measured at the Stanford Synchrotron Radiation Labora-
tory on the focused 16-pole 2.0 T wiggler beam line 9-3 and the
unfocused 8-pole 1.8 T wiggler beam line 7-3 under standard ring
conditions of 3 GeV and 60-100 mA. A Si(220) double-crystal
monochromator was used for energy selection. A Rh-coated harmonic
rejection mirror and a cylindrical Rh-coated bent focusing mirror were
used for beam line 9-3, whereas the monochromator was detuned 50%
at 9987 eV on beam line 7-3 to reject components of higher harmonics.
The solid sample was finely ground with BN into a homogeneous
mixture and pressed into a 1 mm aluminum spacer between X-ray
transparent Kapton tape at dry ice temperature and under a N2
atmosphere. The sample was immediately frozen thereafter and stored
under liquid N2. During data collection, it was maintained at a constant
temperature of 10 K using an Oxford Instruments CF 1208 liquid helium
cryostat. Transmission mode was used to measure data to k ) 16 Å-1
,
which was possible in the absence of any Zn contamination. Internal
energy calibration was accomplished by simultaneous measurement of
the absorption of a Cu foil placed between two ionization chambers
situated after the sample. The first inflection point of the foil spectrum
was assigned to 8980.3 eV. Data represented here is a three-scan average
spectrum, which was processed by fitting a second-order polynomial
to the pre-edge region and subtracting this from the entire spectrum as
background. A three-region spline of orders 2, 3, and 3 was used to
model the smoothly decaying post-edge region. The data were
normalized by subtracting the cubic spline and by assigning the edge
jump to 1.0 at 9000 eV using the SPLINE program in the XFIT suite
of programs (Dr. Paul Ellis, SSRL). Theoretical EXAFS signals, ø(k),
were calculated using FEFF (version 7.0)38,39 and fit to the data using
EXAFSPAK (G. N. George, SSRL). The structural parameters that
varied during the fitting process were the bond distance (R) and the
bond variance (σ2), which is related to the Debye-Waller factor
resulting from thermal motion and static disorder. The nonstructural
parameter, E0 (the energy at which k ) 0), was also allowed to vary
but was restricted to a common value for every component in a given
fit. Coordination numbers were systematically varied in the course of
the fit but were fixed within a given fit.
Stopped-Flow Kinetics. All manipulations of the copper(I) com-
plexes and their solutions were done inside a Vacuum Atmospheres
glovebox with argon or using Hamilton airtight syringes. Tetrahydro-
furan (THF) was purified by purging with ultrapure argon and passing
through a column of activated anhydrous alumina under an argon
atmosphere. Nitriles were purchased from Aldrich or Acros in the
highest purity grade available and degassed before taking into the
glovebox. Saturated solutions of O2 in THF (10 mM) were prepared
by bubbling dry O2 gas through argon-saturated THF in a syringe at
293 K for 15 min.40 Solutions of O2 with smaller concentrations were
prepared by diluting the 10 mM O2 solution with argon-saturated THF
using graduated gastight syringes. Kinetic measurements were per-
formed using a Hi-Tech Scientific (Salisbury, Wiltshire, U.K.) SF-43
multimixing anaerobic cryogenic stopped-flow instrument combined
with a Hi-Tech Scientific Kinetascan diode array rapid scanning unit.
The solutions of L1Cu(RCN) in THF or THF/RCN and O2 in THF
were separately cooled to a low temperature (193-233 K) and mixed
in a 1:1 volume ratio. The mixing cell (1 cm) was maintained at (0.1
K, and the mixing time was ∼10 ms. Concentrations of all reagents
are reported at the onset of the reaction (after mixing) and corrected
for the 1:1 dilution. No correction for the temperature contraction of
solvent was applied in this work. Dioxygen (1.25-5.0 mM) was always
DFT Methods. Geometry optimizations were performed with DFT
using the B3LYP functional43-45 as implemented in the Jaguar, version
5.0, suite of ab initio quantum chemistry programs.46 A restricted
(RDFT) methodology was used for all species where dioxygen is not
present and for the singlet Cu-O2 adducts. Unrestricted (UDFT) wave
functions were obtained for molecular O2 and the triplet Cu-O2 adducts.
The 6-31G** basis set was used for all atoms, with the exception of
the Los Alamos lacvp** basis set47-49 (including an effective core
potential and polarization functions) for the one Cu atom. Mappings
along reaction coordinates were performed by freezing at least one
interatomic distance and by optimizing all other degrees of freedom.
Maxima along these coordinates were used as starting structures for
quadratic synchronous transit (QST) transition-state searches.50
Vibrational frequencies were determined analytically using a trun-
cated version of L1 obtained by removing the four isopropyl groups,
replacing them with hydrogen atoms, and optimizing the positions of
those hydrogen atoms while fixing the rest of the structure. Calculation
of vibrational frequencies not only allowed for verification of the
computed intermediates and transition states as bona fide stationary
points but also enabled zero-point energy, enthalpy, and entropy
corrections to be made and, therefore, free energies for all species to
be obtained. To calculate O-O stretching frequencies, a triple-ú quality
basis set (6-311G** for all atoms, except lacv3p** for Cu47-49) and a
scaling factor of 0.9751 were employed, as these steps have been shown
to lead to increased reliability in the computational result.32
Single-point solvation energies were calculated with a self-consistent
reaction field method using the Poisson-Boltzmann solver implemented
in Jaguar.52,53 The dielectric constant, ꢀ, for tetrahydrofuran (THF), used
as the solvent in order to compare with available experimental kinetics
data, was computed from a quadratic fitting of ꢀ to the absolute
temperature T (eq 1), where a, b, and c are 30.739, -0.12946, and
0.000 171 95, respectively.54 The dielectric constant for THF is then
7.43 at 298 K and 10.42 at 223 K (the experimental reaction
(41) Skoog, D. A.; Leary, J. J. Principles of Instrumental Analysis, 4th ed.;
Saunders: Fort Worth, TX, 1992.
(42) Gherman, B. F.; Cramer, C. J. Inorg. Chem. 2004, 43, 7281.
(43) Johnson, B. G.; Gill, P. M. W.; Pople, J. A. J. Chem. Phys. 1993, 98,
5612.
(44) Becke, A. D. J. Chem. Phys. 1993, 98, 1372.
(45) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(46) Jaguar 5.0; Schrodinger, LLC: Portland, OR, 2002.
(47) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270.
(48) Wadt, W. R.; Hay, P. J. J. Chem. Phys. 1985, 82, 284.
(49) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299.
(50) Peng, C. Y.; Schlegel, H. B. Isr. J. Chem. 1993, 33, 449.
(51) Bauschlicher, C. W.; Partridge, H. J. Chem. Phys. 1995, 103, 1788.
(52) Marten, B.; Kim, K.; Cortis, C.; Friesner, R. A.; Murphy, R. B.; Ringnalda,
M. N.; Sitkoff, D.; Honig, B. J. Phys. Chem. 1996, 100, 11775.
(53) Tannor, D. J.; Marten, B.; Murphy, R. B.; Friesner, R. A.; Sitkoff, D.;
Nicholls, A.; Ringnalda, M. N.; Goddard, W. A., III.; Honig, B. J. Am.
Chem. Soc. 1994, 116, 11875.
(38) Mustre de Leon, J.; Rehr, J. J.; Zabinsky, S. I.; Albers, R. C. Phys. ReV. B
1991, 44, 4146.
(39) Rehr, J. J.; Mustre de Leon, J.; Zabinsky, S. I.; Albers, R. C. J. Am. Chem.
Soc. 1991, 113, 5135.
(40) Battino, R. Oxygen and Ozone; Battino, R., Ed.; Pergamon Press: New
York, 1981; Vol. 7.
9
J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004 16899

A R T I C L E S
Aboelella et al.
temperature).55 When computing free energy changes in solution, a
correction for translational entropy due to a change in concentration
from gas phase (i.e., the concentration at 1 atm, which may be computed
from the ideal gas law) to the 1 M standard-state solution concentration
was included.56 In the case that the THF solvent plays an explicit role
in a chemical reaction, the concentration of the solvent (determined
from its density57,58 and molecular weight) was factored into the entropy
correction.
CASPT2 correction and Cu-O bond length from 1.83 to 2.33 Å can
be obtained for singlet end-on Cu-O2 adducts (Figure S1b in the
Supporting Information). In all cases, the corrections have negative
values, indicating that DFT systematically predicts the energies for the
significantly multideterminantal singlet states to be too high.
∆ ) (1A1 - 3B1)CASPT2 - (1A1 - 3B1)DFT ) [(1A1)CASPT2
-
(1A1)DFT] - [(3B1)CASPT2 - (3B1)DFT] (2)
ꢀ(T) ) a + bT + cT2
(1)
At infinite Cu-O distances, any CASPT2 correction refers only to
molecular dioxygen. Calculations show the correction for singlet
dioxygen to be -62.7 kJ mol-1. Corrections at long, yet finite, Cu-O
distances were obtained by extrapolating between the dioxygen
correction and the maximum CASPT2 correction from the linear
correlations. Specifically, such longer range corrections are given by
eq 3, where “max correction” is the CASPT2 correction at the last
point in the correlation; O-Omax and O-O are the oxygen-oxygen
bond lengths at that last point in the correlation and in the particular
long-range Cu-O structure, respectively, and 1.215 Å is the equilibrium
O2 bond length.
Multireference Methods. In general, a closed-shell singlet state and
a high-spin triplet state can each be described within a single-
determinantal formalism. An open-shell singlet, however, cannot be
formally expressed in a single-determinantal framework (such as Kohn-
Sham DFT) and requires at least two determinants. With regard to
Cu-O2 adducts and in particular those possessing Cu(II)-superoxide
character, the open-shell singlet state becomes significant since
Cu(II)-superoxide is formally a biradical with one electron localized
to Cu and another to the O2 moiety. The multideterminantal character
of the singlet Cu-O2 adducts must then be accounted for when these
species’ energies are calculated. Thus, to obtain accurate energy differ-
ences between singlet and triplet states, single-point calculations were
performed using multireference second-order perturbation theory
(CASPT2).59 The initial complete active space (CAS) for the reference
wave functions consisted of 18 electrons in 12 orbitals, specifically,
the valence orbitals/electrons of Cu and the O2 σ2p, σ2p*, π2p, and π2p*
orbitals/electrons. A final (12,9) subspace was obtained by removing
orbitals with occupation numbers greater than 1.999 for each geometry
under consideration. All CAS and CASPT2 calculations were done with
MOLCAS,60 using a polarized valence double-ú atomic natural orbital
basis set and 17-electron relativistic effective core potential for Cu.
Corrections to the DFT Singlet Energies. CASPT2 calculations
were carried out using a simplified version (sans the 2,6-diisopropyl-
phenyl flanking groups and methyl groups on the L1 backbone) of the
full â-diketiminate ligand system in order to make calculations at this
level of theory tractable. CASPT2 energies were determined for the
singlet (1A1) and triplet (3B1) states for the C2V side-on (η2) adducts
over a range of distances between Cu and the O-O midpoint.
Differences between the singlet/triplet splittings at the CASPT2 level
of theory and from the DFT calculations were then calculated according
to eq 2. Assuming the triplet state to be well-described by a single
determinant, as in DFT,61 the relative energy difference between the
triplet at the two levels of theory is zero. The ∆ quantity in eq 2 is
then equal to the relative energy difference between the singlet at the
DFT and CASPT2 levels and serves as a correction to the singlet
energies produced by DFT, while the triplet correction to the DFT
energies is zero. This “CASPT2 correction” to the singlet energy
correlates linearly with the Cu-O bond length over a range from 1.85
to 2.41 Å (Figure S1a in the Supporting Information), allowing CASPT2
corrections to be readily determined for any singlet side-on adduct
between these two endpoints. A similar linear correlation between the
O-O - 1.215
O-Omax - 1.215
(max correction + 62.7)
- 62.7
(3)
(
)
The CASPT2 corrections are largest at intermediate Cu-O distances,
where Cu(II)-superoxo and, hence, biradical and multideterminantal
character is the greatest. At short Cu-O distances, the structures
increasingly resemble Cu(III)-peroxo, and at long Cu-O distances,
the interaction between the Cu(I) complex and dioxygen is relatively
minimal. Both latter cases lead to a decrease in the magnitude of the
CASPT2 correction to the singlet energy.
A complete listing of individual energies and the precise protocol
for computing composite energies, enthalpies, entropies, and free
energies in the gas phase and in THF solution are provided in the
Supporting Information.
Results
Synthesis. The synthesis, spectroscopic data, and X-ray
crystal structures of the complexes LCu(MeCN) (L ) L1 or
L2) were reported previously.30,31 As described below, substitu-
tion of the MeCN ligand by various para-substituted benzoni-
triles was implicated by UV-vis spectroscopic monitoring of
stopped-flow kinetic experiments. To confirm that benzonitrile
adduct formation occurred, complexes of two members of the
series were independently synthesized and characterized. Thus,
treatment of L1Cu(MeCN) in THF with p-NC-C6H4-CN (0.5
equiv) or p-NC-C6H4-OMe (1 equiv) yielded [(L1Cu)2-
(p-NC-C6H4-CN)] or L1Cu(p-NC-C6H4-OMe), respectively, as
solids amenable to analysis by NMR and UV-vis spectroscopy,
elemental analysis, and X-ray crystallography (Figure S2 in the
Supporting Information).
Solid samples of the O2 adduct, L1CuO2, were prepared by
bubbling O2 through solutions of L1Cu(MeCN) in acetone at
193 K. The complex L1CuO2 precipitated as a green, temper-
ature-sensitive powder. The adduct, L2CuO2, was prepared
similarly, except it was isolated as crystals suitable for analysis
by X-ray diffraction from solutions of THF/pentane (1:1).
X-ray Crystallography. (A) [(L1Cu)2(p-NC-C6H4-CN)] and
L1Cu(p-NC-C6H4-OMe). Representations of the X-ray crystal
structures of these molecules are shown in Figure S2 (Supporting
Information), with selected crystallographic data listed in Table
S2 (Supporting Information). Overall, the structures resemble
those reported previously for (â-diketiminate)Cu(MeCN) com-
(54) Lide, D. R. Handbook of Chemistry and Physics, 84th ed.; CRC Press:
Boca Raton, FL, 2003.
(55) These dielectric constant values will not be appreciably altered by the
addition of the small amounts of MeCN to the THF solvent used during
the synthesis of L1CuO2.
(56) Cramer, C. J. Essentials of Computational Chemistry, 2nd ed.; John Wiley
& Sons: Chichester, 2004.
(57) Carvajal, C.; Tolle, K. J.; Smid, J.; Szwarc, M. J. Am. Chem. Soc. 1965,
87, 5548.
(58) Metz, D. J.; Glines, A. J. Phys. Chem. 1967, 71, 1158.
(59) Andersson, K.; Malmqvist, P. A.; Roos, B. O. J. Chem. Phys. 1992, 96,
1218.
(60) Karlstrom, G.; Lindh, R.; Malmqvist, P. A.; Roos, B. O.; Ryde, U.;
Veryazov, V.; Widmark, P. O.; Cossi, M.; Schimmelpfennig, B.; Neogrady,
P.; Seijo, L. Comput. Mater. Sci. 2003, 28, 222.
(61) This assumption is supported by the CASPT2 calculations on the triplets,
which show that their multideterminantal wave functions are dominated
by a single configuration with a weight of approximately 0.95, with any
other contributors having weights <0.015.
9
16900 J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004

Dioxygen Activation at a Single Copper Site
A R T I C L E S
was solved in space group Imm2, although several others,
including I222 and I212121, were evaluated. Only one-fourth of
the molecule is unique, with the rest generated by symmetry
operations. Upon collection of the new data set, other space
groups were considered but were again all rejected in favor of
Imm2. The entire molecule is disordered over two positions in
a 66:34 ratio (Figure 1b). Several possible twinning modes were
evaluated, and none led to a model removing the disorder. While
some THF solvent molecules could be located by the difference
Fourier map, all solvent atoms were removed using the
PLATON/SQUEEZE program,35 which resulted in a better
model. The void space suggests approximately six THF
molecules, or three per copper complex. The structure contains
solvent channels that minimize direct van der Waals contacts
between copper complexes in adjacent columns. This loose
packing allows complexes to fill the space in both directions,
and it is the averaging of the complexes in opposing directions
that leads to the static whole-molecule disorder.
X-ray Absorption Spectroscopy. (A) Cu K-Edges. Experi-
mental insight into the oxidation state of the copper ion in the
1:1 Cu-O2 adducts was obtained through Cu K-edge XAS.
Previous studies of the copper K-edge XAS data for series of
Cu(II) and Cu(III) complexes showed that this method can
clearly distinguish these oxidation levels.62 Notably, a prominent
pre-edge 1s f 3d transition is centered at ∼8981 eV for
Cu(III), approximately 2 eV higher than the energy of this
transition in Cu(II) complexes. This distinction has been found
to hold irrespective of the ligands, and thus is analytically useful
for distinguishing these oxidation levels, even when counterparts
with the same ligand set but a different oxidation state are
unavailable for comparison.
Figure 1. (a) Representation of the X-ray crystal structure of L2CuO2, with
all non-hydrogen atoms shown as 50% thermal ellipsoids. Selected bond
distances (angstroms) and angles (degrees): Cu1-N1, 1.856(6); Cu1-O1,
1.821(5); O1-O1′, 1.392(12); O1-Cu1-O1′, 44.9(4). (b) Illustration of
the disorder present in the crystals of L2CuO2.
plexes.29,30 The Cu(I) ions exhibit similar three-coordinate
geometries distorted by virtue of the benzonitrile ligand being
displaced toward one side of the â-diketiminate ligand L1. This
displacement is characterized by N(L1)-Cu-N(nitrile) angles
for each Cu site that differ by 34-40°. Asymmetry in the
coordination of L1 is evidenced by Cu-N(L1) distances that
differ by 0.07-0.09 Å.
The normalized X-ray absorption K-edge spectrum of solid
L1CuO2 is shown in Figure 2a. The pre-edge feature observed
at ∼8981 eV is a low-intensity (dipole-forbidden) quadrupole-
allowed transition, which gains intensity due to 4p orbital mixing
into the d-manifold as the molecule deviates toward lower
symmetry relative to a centrosymmetric species. The maximum
of the pre-edge appears at ∼8980.7 eV (inset). A characteristic
1s f 4p + LMCT shakedown transition62b,63 is observed on
the rising edge at ∼8986.7 eV. In Figure 2b, the Cu-edge
spectrum of L1CuO2 is compared to previously reported62a
(B) L2CuO2. We previously noted that the X-ray crystal
structure of L2CuO2 contains a whole-molecule disorder, and
that the large number of restraints needed to model this disorder
affects the precision of the bond lengths and angles.31 As such,
though the crystal structure confirms the η2-coordination of O2
to copper, the O-O bond distance of 1.440(16) Å, while
consistent with a formulation with significant Cu(III)-peroxo
character,32 can only be deemed approximate. Given the
importance of the O-O bond distance in evaluating the
electronic structure of the 1:1 adduct, the crystal structure was
redetermined using a more powerful synchrotron beam (Figure
1). Selected crystallographic data for the new structure of
L2CuO2 is given in Table S2 (Supporting Information). With
the synchrotron radiation, a higher quality (i.e., more intense)
data set was collected; however, the whole-molecule disorder
is still evident and the original structure solution still applicable.
The Cu-N/O ligand distances are reasonable (i.e., shorter than
analogous distances in Cu(I) complexes) and agree well with
those determined independently for L1CuO2 by EXAFS (see
below). Importantly, the improved data set gives a shorter
O-O bond distance of 1.392(12) Å, which more closely re-
sembles the value determined previously using DFT calculations
(1.376 Å).31
spectra for the dicopper(II,II) complex, [L3 Cu2(µ-OH)2]2+, and
2
the dicopper(III,III) compound, [L32Cu2(µ-O)2]2+ (L3 ) N,N,N′,N′-
tetraethylethylenediamine). The latter compounds exhibit data
typical for their assigned oxidation states, with 1s f 3d
transitions at ∼8979 ( 0.3 eV for Cu(II) and at ∼8981 ( 0.5
eV for Cu(III). It is evident (inset) that the pre-edge feature for
L1CuO2 at ∼8980.7 eV falls in the Cu(III) region, approximately
1.5-2 eV higher than that for Cu(II) complexes, and is quite
similar to that of [L3 Cu2(µ-O)2]2+
.
2
(B) EXAFS. In lieu of X-ray crystallographic data (only
available for L2CuO2), structural information for L1CuO2 was
obtained by EXAFS analysis. The k3-weighted EXAFS data and
their Fourier transform are shown in Figure 3. Theoretical phase
and amplitude parameters for the fit were generated by
(62) DuBois, J. L.; Mukherjee, P.; Collier, A. M.; Mayer, J. M.; Solomon, E.
I.; Hedman, B.; Stack, T. D. P.; Hodgson, K. O. J. Am. Chem. Soc. 1997,
119, 8578. (b) DuBois, J. L.; Mukherjee, P.; Stack, T. D. P.; Hedman, B.;
Solomon, E. I.; Hodgson, K. O. J. Am. Chem. Soc. 2000, 122, 5775.
(63) Kau, L.-S.; Spira-Solomon, D. J.; Penner-Hahn, J. E.; Hodgson, K. O.;
Solomon, E. I. J. Am. Chem. Soc. 1987, 109, 6433.
Nonetheless, this distance should still be considered an
approximation due to the number of restraints needed to model
the disorder, which warrants explanation. The original structure
9
J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004 16901

A R T I C L E S
Aboelella et al.
Table 1. EXAFS Curve-Fitting Results
fit no.
1
R (Å)a
σ
2 (Å2)b
E
0 (eV)
Fc
4 Cu-N/O
1.83
33
-7.7
-7.05
0.31
0.20
2
4 Cu-N/O
4 Cu-C
1.83
2.80
37
34
3
4 Cu-N/O
4 Cu-C
8 Cu-N-C
1 Cu-C
1.83
2.80
2.98
3.17
34
42
42d
14
-6.90
0.18
a Estimated standard deviation for the distances are on the order of (0.02
Å. b The σ2 values are multiplied by 1 × 105. c Error is given by ∑[(øobs
-
ø
calcd)2k6]/∑[(øobs)2k6]. d The σ2 factor of the MS path is linked to that of
the corresponding SS path.
Figure 4. Stopped-flow time-resolved UV-vis spectra of the reaction of
L1Cu(MeCN) (0.25 mM) with O2 (∼5 mM) in THF at 203 K recorded
every 0.02 s (approximately every 4th spectrum shown). Inset: kinetic trace
at 410 nm with single-exponential fit to At ) A∞ - (A∞ - A0) exp(-kt), k
) 3.73 s-1. A similar plot at 600 nm is shown in Figure S3 (Supporting
Information).
Figure 2. (a) Normalized Cu K-edge X-ray absorption spectrum of
L1CuO2. The inset shows the pre-edge region (;) and its second deriva-
tive (- - -). (b) Overlay of the normalized Cu K-edge X-ray absorption spec-
tra of L1CuO2 (black), the previously published62 data of the dicopper(II,-
II) complex [L3 Cu2(µ-OH)2]2+ (green), and the dicopper(III,III) complex
2
[L3 Cu2(µ-O)2]2+ (red), where L3 ) N,N,N′,N′-tetraethylethylenediamine.
2
The pre-edge region is expanded in the inset.
scattering (SS), and/or multiple scattering (MS) pathways from
the rigid â-diketiminate ligand system. Figure 3a shows EXAFS
fits and the corresponding Fourier transforms, including only
SS pathways, while Figure 3b shows the complete fit involving
both SS and MS paths. The best fit is obtained with four
Cu-N/O distances at 1.83 Å, four Cu-C at 2.79 Å, one Cu-C
at 3.17 Å, and a MS pathway Cu-N-C at 2.98 Å (Table 1).
The σ2 value for the MS pathway is linked to that of the
corresponding Cu-C SS pathway, and its inclusion improved
the fit significantly, as shown in the comparison of the Fourier
transforms (Figure 3b). Each atom in the â-diketiminate-Cu
ring contributes to the EXAFS signal, whereas the aryl substit-
uents (except for the carbon directly attached to the N atom)
do not contribute significantly, most likely due to the relative
rigidity of the six-membered â-diketiminate-Cu ring. In sum-
mary, the good fits and resulting interatomic distances ob-
tained from the EXAFS data for L1CuO2 corroborate the
distances obtained by X-ray crystallography for L2CuO2. Thus,
the EXAFS results support the crystallographic conclusions,
notwithstanding the disorder problems that adversely influence
the precision of the X-ray diffraction results.
Oxygenation Kinetics. The kinetics of the reaction of
L1Cu(MeCN) with O2 in THF or THF/MeCN mixtures
(0.3-5% MeCN by volume) at low temperatures (193-233 K)
were studied using the stopped-flow technique with time-
resolved UV-vis spectroscopic monitoring. As shown in Figure
4, the spectrum of yellow L1Cu(MeCN) converts upon oxy-
genation to that of green L1CuO2 with no accumulation of inter-
mediates. The yield of L1CuO2, as indicated by final absorbance
values (approaching 100%), was the same with or without added
MeCN, consistent with an irreversible oxygenation reaction and
Figure 3. EXAFS data (black), fits (red), and the corresponding Fourier
transforms for L1CuO2, including (a) only single scattering (SS) and (b)
both SS and multiple scattering (MS) pathways.
FEFF,38,39 using the crystallographic parameters for L2CuO2.
Good fits to the EXAFS were obtained by including a first shell
of four O/N scatterers and a second shell including single
9
16902 J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004

Dioxygen Activation at a Single Copper Site
A R T I C L E S
Figure 6. Plots of observed pseudo-first-order rate constants for the
oxygenation of L1Cu(MeCN) (0.25 mM) at various temperatures as a
function of [O2] in (a) THF/MeCN (v/v 160:1; [MeCN] ) 0.12 M) and (b)
neat THF. The data presented in these plots are listed in Table S4
(Supporting Information).
Scheme 1
Figure 5. Plots of the observed pseudo-first-order rate constant for the
oxygenation of L1Cu(MeCN) (0.25 mM) at T ) 203 K, (a) at variable [O2]
in neat THF (b) and THF/MeCN (9) ([MeCN] ) 60 mM) and (b) at
variable [MeCN] and constant [O2] ) 5 mM.
in agreement with batch experiments wherein the amount of
L1CuO2 was found to be unperturbed by removal of O2 from
solution. Under excess O2, the spectral changes with time are
fit well to a single exponential (Figure 4 inset and Figure S3 in
the Supporting Information) to yield a pseudo-first-order rate
constant (kobs) that is independent of the initial concentration
of L1Cu(MeCN) (Table S3 in the Supporting Information).
These data indicate a first-order dependence on [L1Cu(MeCN)],
eq 4.
∂[L1CuO2]
∂t
∂[L1Cu(MeCN)]
∂t
) -
) kobs[L1Cu(MeCN)] (4)
addition, the rate constants in neat THF are systematically higher
than those in THF/MeCN (Figure S4 and Table S4 in the
Supporting Information). These results indicate that there is an
additional oxygenation pathway operative in neat THF that
exhibits a zero-order dependence on [O2], such that the overall
rate law has two terms (eqs 7 and 8).
The dependence on [O2] is more complicated and is affected
by the presence of MeCN, which moderately decreases kobs
(Figure 5 and Figure S4 and Table S4 in the Supporting
Information). For instance, under the conditions T ) 203 K,
[L1Cu(MeCN)]0 ) 0.25 mM, and maximum available excess
[O2] (5 mM), the kobs value in neat THF of 3.7 ( 0.1 s-1 changes
to 2.66 ( 0.06 s-1 in the presence of 120 mM MeCN. The
latter value of kobs is unchanged by further increases in [MeCN]0
(up to 950 mM, 5 vol %), indicating saturation behavior and a
zero-order dependence in [MeCN] when in excess. In THF/
MeCN mixtures, plots of kobs versus [O2] are linear with zero
intercepts at all temperatures (193-233 K, Figure 6a). Thus,
in the presence of MeCN, the oxygenation displays a first-order
dependence on [O2] with an overall second-order rate law (eqs
5 and 6).
kTobHsF ) kA[O2] + kB
(7)
∂[L1CuO2]
∂t
∂[L1Cu(MeCN)]
) -
)
∂t
kA[L1Cu(MeCN)][O2] + kB[L1Cu(MeCN)] (8)
By analogy to the results of classic studies of ligand sub-
stitution reactions of square planar complexes that yielded rate
laws of identical form to eq 8 (see Discussion),64 we propose a
dual-pathway mechanism (Scheme 1). The first and rate-limiting
step of pathway A is a direct bimolecular reaction of L1Cu-
(MeCN) with O2, where adduct L1Cu(MeCN)(O2) may be
considered as a transition state or an unstable intermediate that
releases MeCN and converts to L1CuO2 in a post-rate-limiting
step(s). In pathway B, the first and rate-limiting step is sol-
volysis of L1Cu(MeCN) to yield a highly reactive intermediate
kTobHsF/MeCN ) kA[O2]
∂[L1Cu(MeCN)]
(5)
∂[L1CuO2]
∂t
) -
) kA[L1Cu(MeCN)][O2]
∂t
(6)
(64) Wilkins, R. G. Kinetics and Mechanism of Reactions of Transition Metal
Complexes, 2nd ed.; VCH: Weinheim, Germany, 1991; pp 232-243 and
references therein.
In neat THF, however, plots of kobs versus [O2] also are linear,
but with significant positive y-intercepts (Figures 5 and 6b). In
9
J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004 16903

A R T I C L E S
Aboelella et al.
Table 2. Selected Kinetic Parameters for Oxygenation of Mononuclear Cu(I) Complexes Leading to 1:1 Cu-O2 Adductsa
1
1
Cu(I) complex
solvent
k
on (223 K)b
∆
Hq (kJ mol-
)
∆
Sq (J K -1 mol-
)
ref
L1Cu(MeCN), pathway Ac
THF-MeCN or THF
1560 ( 19
18 ( 2
14.9 (t)
30 ( 2 2
7.2 (t)
37.2 ( 0.5
17.1 ( 0.6
31.6 ( 0.5
7.6
-100 ( 10
-108 (t)
-98 ( 10
-101.0 (t)
-62 ( 2
-52 ( 3
10 ( 3
this work
L1Cu(MeCN), pathway Bd
THF
3.95 ( 0.59e
this work
LiPr3Cu(MeCN)
(Me6tren)Cu(EtCN)
(TMPA)Cu(EtCN)
(TMPA)Cu(THF)
Me2CO
EtCN
EtCN
THF
5.2f
74
22b
27
(8.7 ( 0.4) × 105
(5.0 ( 0.3) × 105
3.4 × 108 f
-45
28
a All are experimental values, except those denoted by “t”, which are derived from theoretical calculations. Abbreviations: LiPr3 ) 1,4,7-triisopropyl-
1,4,7-triazacyclononane, Me6tren ) hexamethyltriaminoethylamine, TMPA ) tris(pyridylmethyl)amine. b Unless noted otherwise, units are in M-1 s-1
.
c Parameters determined from experiments in THF/MeCN (v/v 160:1; [MeCN] ) 0.12 M). d Parameters refer to the solvolysis reaction (exchange of THF
for MeCN), not the oxygenation. Thus, kon refers to kB ) kB1 (eqs 8 and 9). e Units are in s-1 f Calculated from the published activation parameters.
.
L1Cu(THF) that is subsequently scavenged rapidly by O2; thus,
pathway B is kinetically independent of [O2]. Assuming that
L1Cu(THF) is a steady-state intermediate, the mechanism in
Scheme 1 is described by the rate law in eq 9, which can be
shown to be consistent with the rate laws determined experi-
mentally under the different reaction conditions used. Thus, with
excess O2 and in neat THF (no added MeCN), k-B1[MeCN] ,
kB2[O2], and eq 9 reduces to the experimentally determined eq
8, where kB ) kB1. When a large excess of MeCN is present,
k-B1[MeCN] . kB1kB2[L1Cu(MeCN)][O2], and the second term
of eq 9 approaches zero, such that eq 9 reduces to eq 6. In
other words, in the presence of excess MeCN, pathway B is
effectively shut down because nearly all L1Cu(THF) is converted
to L1Cu(MeCN), and the oxygenation process proceeds entirely
through pathway A.
Figure 7. UV-vis spectra at 203 K of L1Cu(MeCN) (0.2-0.5 mM) in
the presence of excess RCN (100 mM) in THF, where R ) Me, Et, and
p-X-C6H4 (X ) OMe, Me, F, H, Cl, CN).
∂[L1CuO2]
∂t
∂[L1Cu(MeCN)]
∂t
) -
)
kB1kB2[L1Cu(MeCN)][O2]
k-B1[MeCN] + kB2[O2]
kA[L1Cu(MeCN)][O2] +
(9)
Activation parameters for both pathways A (kA) and B (kB
) kB1) were obtained from Eyring plots for the respective rate
constants measured over the temperature range T ) 193-233
K (Figure S5 and Table S5 in the Supporting Information, and
Table 2). Associative mechanisms for the rate-determining steps
of both pathways are supported by the low ∆Hq and large
negative ∆Sq values. Further interpretations of the activation
parameters are presented in the Discussion.
To obtain deeper insight into the nature of the transition state
for the direct oxygenation pathway A, we sought to examine
how the kinetic behavior is influenced by variation of ligand
electronic properties. Rather than varying the attributes of L1,
which would require extensive synthetic effort, we focused on
replacing the readily exchanged MeCN ligand with a series of
para-substituted benzonitriles (p-X-C6H4-CN, X ) OMe, Me,
H, F, Cl, CN). On the basis of the kinetic results for L1Cu-
(MeCN), we reasoned that addition of excess p-X-C6H4-CN to
L1Cu(MeCN) in THF would convert most of L1Cu(MeCN) to
the adduct L1Cu(NCC6H6X) and suppress a solvolytic oxygen-
ation mechanism (pathway B), thus leaving direct reaction of
O2 (pathway A) as the only route to L1CuO2. Importantly, in
pathway A, the p-X-C6H4-CN ligand would be lost only after
the rate-limiting step of the oxygenation reaction, so that its
electronic effects would be manifested by differences in
measurable rate constants.
Figure 8. Correlations of the Hammett substituent constants (σp) and the
energies of the MLCT bands (shoulders, 9) and the pseudo-first-order rate
constants for oxygenation of solutions of L1Cu(MeCN) in the presence of
excess p-X-C6H4-CN (X ) OMe, Me, F, H, Cl, CN; 100 mM) (b).
Addition of excess p-X-C6H4-CN (200 equiv) to solutions
of L1Cu(MeCN) in THF (0.2-0.5 mM) at 203 K resulted in
dramatic color and UV-vis spectral changes indicative of the
formation of complexes L1Cu(NCC6H6X) (Figure 7). An intense
absorbance (ꢀ ) 1000-3000 M-1 cm-1) above 400 nm
developed, which shifted hypsochromically with increasing
electron-withdrawing capabilities of the substituent X. Indeed,
the energy of the absorption feature is linearly correlated with
the substituent Hammett constants (σp, Figure 8, blue squares),
with a negative slope consistent with assignment of the
9
16904 J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004

Dioxygen Activation at a Single Copper Site
A R T I C L E S
Table 3. Values of Pseudo-First-Order Rate Constants for
Table 4. Relative Free Energies (kJ mol-1) Calculated for the 1:1
Cu-O2 Adducts, L1CuO2 (Within Each Column, Energies Are
Relative to the Side-On Singlet)
Oxygenation of [L1Cu(RCN)] Complexesa
1
solvent
copper(I) species
k )
obs (s-
gas phase
298 K
gas phase
223 K
THF
THF
THF only
[L1Cu (MeCN)] + [L1Cu(THF)] 3.7 ( 0.1
298 K
223 K
MeCN in THF
EtCN in THF
[L1Cu(MeCN)]
[L1Cu(EtCN)]
2.66 ( 0.06
2.25 ( 0.05
1.24 ( 0.05
1.14 ( 0.05
1.05 ( 0.05
1.02 ( 0.03
0.87 ( 0.03
0.6 ( 0.2
end-on singlet
end-on triplet
side-on singlet
side-on triplet
16.9
48.4
0.0
18.3
51.8
0.0
18.0
56.8
0.0
19.8
61.5
0.0
p-MeO-C6H4-CN in THF [L1Cu(p-MeO-C6H4-CN)]
p-Me-C6H4-CN in THF
PhCN in THF
p-F-C6H4-CN in THF
p-Cl-C6H4-CN in THF
p-NC-C6H4-CN in THF
[L1Cu(p-MeC6H4-CN)]
[L1Cu(PhCN)]
22.7
24.0
26.5
28.8
[L1Cu(p-F-C6H4-CN)]
[L1Cu(p-Cl-C6H4-CN)]
[L1Cu(p-NC-C6H4-CN)]
constitutes a mechanistically homogeneous reaction series,
supporting the assumption that the same mechanism (Scheme
1) applies to all of the L1Cu(RCN) complexes studied.
Theoretical Calculations. (A) Structure of the 1:1
Cu-O2 Adducts. Both side-on (η2) and end-on (η1) 1:1
Cu-O2 adducts supported by L1 were considered in calculations
using DFT and multireference second-order perturbation theory
(CASPT2), with either singlet or triplet multiplicities (Experi-
mental Section). Comparing the energies of these adducts (Table
4) indicates that their relative stabilities are affected only
minimally by the temperature difference or the use of THF
as solvent. Under all conditions, the side-on adducts of a
given multiplicity are lower in energy than their end-on
counterparts, and of the side-on adducts, the singlet is most
stable. These results, as well as key computed properties of the
side-on singlet structure for L1CuO2, agree well with the
experimental findings and differ only slightly from previously
reported calculations on L2CuO2 (Table S1).31,32 As discussed
more fully below, the O-O bond distance and ν(O-O) values,
in conjunction with the degree of multideterminantal character
in the CASPT2 wave function and the XAS results, suggest
significant Cu(III)-peroxo character for the singlet side-on
adduct.
a Solutions were prepared by dissolving solid complex 1 in THF, with
[RCN]0 ) 50 mM. Conditions: T ) 203 K, [1]0 ) 0.25 mM, [O2]0 ) 5
mM.
absorption as a metal-to-ligand charge-transfer transition (MLCT,
Cu(I) f p-X-C6H4-CN). No discernible spectral changes
occurred when excess EtCN was added to L1Cu(MeCN) in THF,
consistent with indistinguishable UV-vis spectra for L1Cu-
(MeCN) and L1Cu(EtCN), as noted previously for other Cu(I)
adducts of aliphatic nitriles.26a,27
Final corroboration of the formation of L1Cu(NCC6H6X) in
the L1Cu(MeCN)/p-X-C6H4-CN mixtures came from the inde-
pendent isolation and structural characterization of L1Cu(p-CN-
C6H4-OMe), the UV-vis spectrum of which matched that
obtained in the mixing experiments. We note that the spectrum
of the isolated dinuclear complex [(L1Cu)2(p-NC-C6H4-CN)]
was perturbed relative to that seen in the kinetics experiments,
wherein p-NC-C6H4-CN was in large excess relative to L1Cu-
(MeCN). However, addition of excess p-NC-C6H4-CN to the
dicopper complex [(L1Cu)2(p-NC-C6H4-CN)] resulted in a shift
of the intense absorbance feature to approximately the same
position as in the kinetics experiments, indicating conversion
of the isolated dicopper adduct to the monocopper species
L1Cu(p-NC-C6H4-CN) in THF in the presence of an excess of
the benzonitrile.
(B) Oxygenation Pathway A. Pathways A and B (Scheme
1), postulated on the basis of the kinetics results, were considered
as starting points for theoretical assessment of the mechanism
of formation of L1CuO2. Turning first to pathway A, we
considered the possibility of a rapid pre-equilibrium dissociation
of MeCN from L1CuO2 to yield “L1Cu” prior to the rate-
controlling O2-binding event, which could result in either an
end-on (η1) or a side-on (η2) adduct. The η1 adduct, if formed,
could then isomerize to the final η2 species. Although incon-
sistent with the observed kinetics,65 such a mechanism was
analyzed by theory for comparative purposes. Combining the
reaction coordinates calculated for the individual steps, we
obtain the reaction profile shown in Figure S7 in the Supporting
Information. The initial activation barrier corresponds to the
loss of MeCN from L1Cu(MeCN). This step is rate-determining
Exposure of the solutions comprising L1Cu(MeCN) and 200
equiv of p-X-C6H4-CN in THF to O2 resulted in rapid bleaching
of the intense low-energy absorption features to yield a final
spectrum identical to that previously identified for L1CuO2
(Figure S6 in the Supporting Information). Stopped-flow kinetic
studies of the oxygenation reaction in THF at T ) 203 K with
[L1Cu(MeCN)]0 ) 0.25 mM, [O2]0 ) 5.0 mM (excess), and
[p-X-C6H4-CN]0 ) 50 mM, yielded single-exponential absor-
bance versus time curves, for which fits provided pseudo-first-
order rate constants, kobs (eqs 5 and 6, Table 3). The kobs values
for the adducts of p-X-C6H4-CN are all smaller than those for
L1Cu(MeCN) itself and for L1Cu(MeCN) in the presence of
MeCN
obs
EtCN
obs
p-X-C6H4-CN
obs
EtCN, with the trend k
> k
> k
being
in both the gas phase (298 K, ∆Gq ) 91.1 kJ mol-1 66
and in
)
consistent with steric influences on the rate-determining attack
of O2 at the copper center. The trends within the series p-X-
C6H4-CN were most illuminating, however. The rate constants
decreased with increasing electron-withdrawing capabilities of
the X substituent, yielding a linear Hammett correlation with F
) - 0.34 (Figure 8, black circles). These data show that positive
charge develops at the copper site in the activated complex
during the rate-limiting step of the oxygenation reaction,
consistent with the direct associative pathway A (Scheme 1)
involving coordination of the nitrile in the activated complex
and some degree of charge transfer from Cu(I) to O2. The high
quality of the Hammett correlation in Figure 8 also indicates
that the oxygenation of the series of p-X-C6H4-CN adducts
THF solution (223 K, ∆Gq ) 79.2 kJ mol-1). In addition to
yielding a first-order rate law inconsistent with the bimolecular
kinetics observed, the free energy of activation is computed to
be higher than that calculated for an associative bimolecular
mechanism (see below). The findings thus confirm that a
pathway involving prior dissociation of MeCN to yield a
“L1Cu” intermediate is untenable.
(65) The rate law for such a mechanism exhibits an inverse order in [MeCN],
which was not observed: ∂[L1CuO2]/∂t
)
k2Keq[L1Cu(MeCN)][O2]/
[MeCN], where the rapid pre-equilibrium Keq ) k1/k-1 ) [L1Cu(MeCN)]/
[MeCN][“L1Cu”] and k2 refers to the rate-determining oxygenation step.
(66) Unless otherwise indicated, enthalpies, entropies, and free energies pertain
to 298 K and gas-phase values.
9
J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004 16905

A R T I C L E S
Aboelella et al.
Figure 9. Calculated reaction energy profiles (black line, 298 K, gas phase; red line, 223 K, THF) for (left) the associative bimolecular oxygenation
mechanism corresponding to pathway A (Scheme 1) and for (right) the solvolytic oxygenation mechanism corresponding to pathway B (Scheme 1). Cores
of the minimized structures are shown, with the substituents on L1 omitted for clarity and selected interatomic distances indicated (Å). The colors are gray
for C, blue for N, red for O, white for H, and pink for Cu. Full details on the protocol used to compute relative energies for all stationary points are provided
in the Supporting Information.
Table 5. Calculated Relative Enthalpies, Entropies, and Free
A second possibility corresponding to pathway A is an
Energies for Species along the Associative Bimolecular Pathway
associative mechanism that involves direct binding of O2 to
L1Cu(MeCN) (Figure 9a). The reaction begins along the triplet
surface at long Cu-O2 distances (Figure S8 and Table S6 in
the Supporting Information). The rate-determining transition
state (1) occurs on the triplet surface at Cu-O1/O267 distances
of 2.40/2.97 Å (Cu-O2-midpoint distance of 2.62 Å) and
resembles a weak end-on coordination of dioxygen to the Cu
center. At this relatively early transition state, the O-O bond
length has increased by only 0.023 Å versus that in free
Aa
speciesb
∆
Hc
∆
Sd
∆
Gc
∆
Gc
∆
Gc
∆
Gc
298 Κe
223 Ke 298 K,
THF
223 K,
THF
0.0
39.0
16.2
8.2
L1Cu(MeCN)
1
0.0
0.0
0.0
0.0
41.6
22.8
24.4
0.0
49.1
28.2
10.1 -141.5
-13.4 -161.9
-11.4 -160.6
52.2
34.9
36.4
2
3
20.7
L1CuO2
-22.9
-22.8 -16.2 -17.7
-37.3
-41.4
dioxygen. The gas-phase activation free energy is 52.2 kJ mol-1
,
a Details pertaining to the calculation of these quantities can be found
in Table S5 in the Supporting Information. b Gas phase. c Units are in kJ
with entropy loss due to O2 binding being a large contributor
(Table 5). Crossing from the triplet to singlet surfaces then
occurs at Cu-O1/O2 distances of 2.27/2.52 Å (Cu-O2-midpoint
distance of 2.36 Å), after which an unstable L1Cu(MeCN)(O2)
intermediate (2) is formed. The reaction taking 1 to 2 involves
formation of the Cu-O1 bond, with concomitant rotation of
O2 toward the Cu center to yield an asymmetric, side-on-bound
O2 moiety in 2. The MeCN ligand is also only weakly bound
at this point, with the Cu-MeCN distance 0.37 Å longer than
that in L1Cu(MeCN). From the L1Cu(MeCN)(O2) intermediate
(2), MeCN is lost in a low-barrier process. At the transition
state (3), O2 has completed its rotation toward Cu and adopted
a position symmetric with respect to O1. The overall reaction
is exergonic as expected.
mol-1 d Units are in J K-1 mol-1 e All energy comparisons are for
. .
stoichiometrically equivalent species.
Smaller activation barriers and smaller free energy changes
for individual reaction segments were calculated at 223 K
compared to 298 K, due to the decreased influence of the
entropy changes on the free energy values (Table 5). When
solvation by THF was included in the calculations, 1 remains
the rate-determining transition-state structure, but 2, which is
an intermediate in the gas phase at 298 K by virtue of being
marginally lower in energy than 3, is apparently no longer a
stationary point. In addition, due to the large solvation energy
for MeCN in THF (approximately -29 kJ mol-1), the overall
reaction becomes more exergonic when solvent effects are
included (as reflected by the significantly decreased energy of
L1CuO2, Table 5). Notably, the activation parameters calculated
(67) O1 and O2 refer to the oxygen atoms of the O2 moiety closer to and farther
from the Cu center, respectively.
9
16906 J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004

Dioxygen Activation at a Single Copper Site
A R T I C L E S
for the bimolecular pathway A in THF (∆Hq ) 14.9 kJ mol-1
,
Table 6. Calculated Relative Enthalpies, Entropies, and Free
Energies for Species along the Solvolytic Pathway Ba
∆Sq ) -108.0 J K-1 mol-1) are in excellent agreement with
those measured experimentally (∆Hq ) 18 kJ mol-1, ∆Sq )
-100 J K-1 mol-1; see Tables 2 and S7).
speciesb
∆
Hc
∆
Sd
∆
Gc
∆
Gc
∆
Gc
∆
Gc
298 Κe
223 Ke 298 K, 223 K,
THF
0.0
THF
0.0
49.7
15.3
33.6
3.2
15.0
-21.6
-7.5
-41.4
The flow of electrons as the pathway in Figure 9a is traversed
may be assessed by considering simple Mulliken charge
distributions for the O and Cu atoms, as well as L1 and MeCN,
calculated at 223 K in THF (Table S8 in the Supporting
Information). In the reactant L1Cu(MeCN), the positive charge
on Cu is balanced by the negatively charged L1. At the transition
state 1, the O2 fragment begins to accept electron density from
Cu into its π* orbitals, but L1 stabilizes the oxidized Cu center,
giving a net effect of electron density flowing from L1 to the
O2 moiety and minimal change in the charge on Cu (0.02 units).
The degree of electron transfer to the O2 unit in 1 is 20% relative
to the product L1CuO2, according to a comparison of the total
Mulliken charges for the O atoms. Upon formation of 2, the
reduction of O2 is nearly complete, with the additional electron
density on the O atoms having come in almost equal amounts
from the Cu center (now ∼50% oxidized) and L1. Release of
MeCN from 2 to give L1CuO2 results in the last significant
oxidation of the L1Cu moiety, with the increases in the Mulliken
charges of L1 and Cu again being nearly equal. Comparing
L1Cu(MeCN) to L1CuO2, the overall process of O2 reduction
involves transfer of electron density to the dioxygen fragment
in nearly equal amounts from both the Cu and the â-diketiminate
ligand. Note that we do not suggest that there is any equality
between a computed Mulliken charge and a formally assigned
metal oxidation state; rather, our analysis here is simply to
provide insight into how electronic charge is internally distrib-
uted along the reaction path, particularly insofar as that
redistribution is expected to influence the Hammett plot for
substituted nitrile ligands.
L1Cu(MeCN)
0.0
15.9
19.0
0.0
-158.9
-26.0
-164.8
-164.8
-164.8
-3.4
0.0
0.0
51.2
24.9
59.3
28.9
38.7
4
63.2
26.8
68.9
41.3
51.1
0.8
60.6
18.1
46.5
16.1
16.9
L1Cu(THF)
t.s. A
19.0
5
-7.8
2.0
t.s. B
6
7
-0.2
14.1
-22.9
0.6 -19.3
-19.4
19.8
18.5 0.1
L1CuO2
-22.8 -16.2 -17.7 -37.3
a Details pertaining to the calculation of these quantities can be found
in Table S5 in the Supporting Information. b All energy comparisons are
for stoichiometrically equivalent species. c Units are in kJ mol-1 d Units
.
are in J K-1 mol-1 e Gas phase.
.
long Cu-O1 distances during the process by which L1Cu(THF)
converts to 5, the reaction takes place along the triplet surface.
Crossover to the singlet surface occurs at a Cu-O1 distance of
2.46 Å, thereafter yielding the singlet product 5 in which the
O2 fragment has an O-O distance 0.072 Å longer than that in
dioxygen. Loss of THF from 5 yields an end-on Cu-O2 adduct
(6), which then isomerizes to the final product L1CuO2. At the
early transition state (7) for this reaction, the O-O bond length
increases by a minimal 0.021 Å. The product L1CuO2 is formed
as O2 rotates closer to Cu to give symmetric side-on binding
of dioxygen to Cu, and the O-O bond length grows by another
0.056 Å as the reduction of O2 is completed.
The effects of changing temperature from 298 to 223 K and
from being in the gas phase to in THF solvent are signifi-
cant along the solvolytic pathway. The most important conse-
quence of these temperature and solvent effects is a change in
the rate-determining transition state from t.s. A at 298 K in the
gas phase to 4 at 223 K in THF. That is, binding of dioxygen
to L1Cu(THF) is rate-determining in the gas phase at the higher
temperature, whereas the exchange of THF for MeCN becomes
rate-determining in solution at the lower temperature, in
agreement with the experimental conclusions. In addition, the
activation parameters calculated for pathway B in THF
(L1Cu(MeCN) f L1Cu(THF), ∆Hq ) 27.2 kJ mol-1, ∆Sq )
-100.1 J K-1 mol-1) are almost identical to those measured
(C) Oxygenation Pathway B. The solvolytic pathway B
(Scheme 1) may be divided into two segments; first, the
exchange of MeCN in L1Cu(MeCN) for THF to yield
L1Cu(THF), and second, the reaction of dioxygen with
L1Cu(THF) to give L1CuO2. As already noted above, a purely
dissociative process that involves loss of MeCN to yield the
intermediate “L1Cu” that would then be trapped by THF has
an activation energy too high to be considered to be reasonable
(Figures S7 and S9 in the Supporting Information). An energeti-
cally more favorable alternative is a concerted interchange
mechanism (Figure 9b) involving a transition-state structure 4
with MeCN and THF both weakly bound. The reaction is
endergonic, ∆G ) +26.8 kJ mol-1 (Table 6), consistent with
the relative strengths of MeCN and THF coordination to Cu(I)
(MeCN > THF).
As in the calculations for oxygenation of L1Cu(MeCN) via
pathway A, a mechanism for reaction of L1Cu(THF) with O2
involving prior dissociation of THF and trapping of the resultant
“L1Cu” species by O2 has an activation free energy (from
L1Cu(THF)) that is greater than an associative alternative by
25.6 kJ mol-1 (Figures 9b and S9). The latter, kinetically pre-
ferred route involves the formation of an unstable intermediate
L1Cu(THF)(O2) (5), in which O2 is bound end-on to Cu, via
t.s. A. Note that the end-on coordination of O2 in 5 contrasts
with the side-on binding in its counterpart in the associative
pathway A, compound 2. This difference may be attributed to
the greater steric bulk of THF compared to that of MeCN. At
experimentally (∆Hq ) 30 kJ mol-1, ∆Sq ) -98 J K-1 mol-1
Tables 2 and S7).
,
These results including solvation and lowering the temper-
ature may be traced to several effects. As in the bimolecular
pathway A, smaller contributions from entropy changes at the
lower temperature are reflected in the ∆G values. Solvent effects
are also noticeable in reaction steps involving the binding or
release of THF (solvation energy of approximately -10.5 kJ
mol-1) or MeCN, both of which, in contrast to dioxygen, are
well solvated in THF. In addition, entropic considerations tied
to the explicit role that THF solvent molecules play in reactions
along the solvolytic pathway contribute to lower ∆G values for
species incorporating THF (see Table S7).
Of these three effects, only the desolvation of THF leads to
a destabilization of transition state 4 versus L1Cu(MeCN) at
223 K in THF compared to 298 K in the gas phase. Inclusion
of the two entropic effects more than overcomes the desolvation
penalty, yielding an activation energy for solvolysis which is
13.5 kJ mol-1 lower than that at 298 K in the gas phase (Table
9
J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004 16907

A R T I C L E S
Aboelella et al.
6). In contrast, the energy of t.s. A (which, to be stoichiomet-
rically equivalent to the reactants, i.e., L1Cu(MeCN), O2, and
THF, includes the energy of MeCN) is lowered by 35.3 kJ mol-1
in solution at the lower temperature. The effect of the negative
∆Sq upon ∆Gq is decreased by ∼8.5-12.5 kJ mol-1, while
line with a trend noted previously in a plot of O-O vibrational
frequency versus O-O bond distance for η2 1:1 metal-O2
adducts.32 In this plot, L2CuO2 lies in the center of a continuum
between metal-peroxide and -superoxide complexes,68 sug-
gesting that it has a novel electronic structure insofar as it has
significant Cu(III)-peroxide character.
binding of solvent THF in t.s. A lowers ∆Gq by ∼6.3 kJ mol-1
.
Moreover, the exchange of solvent-derived ligands in going from
L1Cu(MeCN) to L1Cu(THF), leading to the release of the more
favorably solvated MeCN into solution, leads to an overall lower
solvation energy for t.s. A compared to that for L1Cu(MeCN).
The energy of t.s. B is also analogously lowest at 223 K in
THF. The inclusion of solvation energy for both MeCN and
THF at points 6, 7, and product L1CuO2 accounts for their
noticeably lower total free energies in THF as opposed to those
in the gas phase. The result of all of these effects is a change
in the rate-determining step, from dioxygen coordination to L1-
Cu(THF) in the gas phase at the higher temperature to exchange
of THF for MeCN in solution at the lower temperature, as found
by experiment.
To further assess this notion experimentally, Cu K-edge XAS
data were acquired and interpreted within the context of a
previous study,62 in which it was demonstrated that the pre-
edge 1s f 3d transition energy may be used as a diagnostic
tool for distinguishing Cu(II) and Cu(III) oxidation states. The
pre-edge position is affected by two factors, the ligand field
strength and the effective nuclear charge (Zeff) on the metal.
An increase in the oxidation state (Zeff) causes a shift of the 1s
core orbital to deeper binding energy and a larger repulsive
2
2
interaction between the dx 9y orbital (LUMO for Cu(III)) and
the ligand donor orbitals, inducing a shift of the LUMO to higher
energy. As a result, the 1s f 3d energy difference becomes
larger, which is manifested in the pre-edge shifting to higher
energies for Cu(III) compared to Cu(II). The observed congru-
ence between the energy and intensity of the pre-edge feature
for L1CuO2 and a bis(µ-oxo)dicopper(III,III) complex attests
to similar oxidation levels for the metal ions in these complexes
and supports attribution of significant Cu(III) character to
L1CuO2. Nonetheless, there are differences between the XAS
spectra for these compounds, in particular, in the energies and
intensities of the two-electron 1s f 4 p + LMCT shakedown
transitions. The 1s f 4p + LMCT shakedown transition at
∼8987 eV is less intense in L1CuO2 relative to that of the bis-
(µ-oxo) species, which indicates a more covalent interaction in
the bis(µ-oxo) compared to that in the side-on peroxo L1CuO2
system. This is consistent with the differences in ligand-to-metal
charge-transfer (LMCT) intensity between the complexes in
absorption spectra, where the bis(µ-oxo) features have an
intensity an order of magnitude higher than those of L1CuO2.
This difference in peroxo- relative to oxo-Cu covalency
reflects differences both in the energies and the natures of the
valence orbitals of peroxide and oxide and the strong donor
interaction of the â-diketiminate ligand with the Cu center,
which impacts the peroxide-Cu bond. Also, both of the 1s f
4p + LMCT shakedown and pre-edge transitions are to slightly
lower energy in L1CuO2, which indicates a slightly lower Zeff
on Cu, again reflecting the high donor strength of the â-diketim-
inate.
Mulliken charge distributions for the species involved in the
solvolytic pathway B (Figure 9b) are listed in Table S8
(Supporting Information). In this pathway, the process of
dioxygen activation begins with the formation of 5 from
L1Cu(THF) and O2. Similar to the results for the bimolecular
pathway, the negative charge borne by the O2 moiety in 5
originates in nearly equal amounts from both Cu and the ligand
L1, with the oxidation of Cu ∼50% complete and the reduction
of dioxygen about two-thirds complete relative to the final
product L1CuO2. As THF, which bears a partial positive total
Mulliken charge in 5, is lost to give 6, the charge assignable to
the ligand L1 decreases further while the Cu is oxidized only
minimally. This can be explained through simultaneous oxida-
tion of Cu and donation of electron density from the still
relatively electron-rich ligand L1 to Cu to create the net effect
of a decreased Mulliken charge on L1. Notably, at 6, the charge
on the ligand is nearly equal to its final value in the product
L1CuO2, indicating that further charge donation from the ligand
to Cu and/or O2 does not occur beyond this point. Conversion
from the end-on to side-on 1:1 Cu-O2 adduct brings about the
remaining oxidation and reduction of Cu and O2, respectively,
as the O2 group becomes more strongly bound to the Cu center.
Discussion
Structure of 1:1 Cu-O2 Adducts. The better quality X-ray
crystallographic data for L2CuO2 in conjunction with the Cu
K-edge XAS and EXAFS data on L1CuO2 provide important
corroboration for the previously proposed electronic structure
for these side-on (η2) 1:1 Cu-O2 adducts.31,32,42 Although
disorder problems continue to complicate the X-ray crystal-
lographic analysis and mitigate the significance of the observed
interatomic distances, the new data yield an O-O distance of
1.392(12) Å that is significantly shorter than that determined
previously (1.44 Å) and in much closer agreement to that of a
theoretical prediction (1.376 Å)31 made at a level of theory that
has already been used to correct another inaccurately reported
O-O bond length in a side-on metal-dioxygen complex crystal
structure.32 This revised distance brings the complex more in
The Cu-O/N bond distances from the X-ray crystallographic
and EXAFS data also support significant Cu(III) character in
the LCuO2 complexes. To draw this conclusion, we compared
the crystallographically determined distances between Cu and
the N atoms of â-diketiminate ligands for 36 reported complexes
(Table S9 in the Supporting Information); other metal-ligand
bond distances were ignored due to complicating differences
in ligand donor atom types.69 The results are summarized in
Table 7, wherein the compounds are grouped according to
coordination number and Cu oxidation state. When the average
Cu-N distances for each group are compared, caveats are the
considerable overlap in the ranges of Cu-N distances between
the groups (also reflected in the standard deviations) and the
small data set for Cu(III) that is restricted to two bis(µ-oxo)-
dicopper structures. Nonetheless, and as expected, at parity of
coordination number, an increase in oxidation level is ac-
(68) The validity of this continuum has been questioned recently, but in more
recent work, the arguments presented in this critique have been refuted
(ref 42): Pantazis, D. A.; McGrady, J. E. Inorg. Chem. 2003, 42, 7734.
9
16908 J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004

Dioxygen Activation at a Single Copper Site
A R T I C L E S
Table 7. Summary of Cu-N(â-Diketiminate) Distance Data from
Available X-ray Crystal Structures (Full List Provided in Table S9
in the Supporting Information)
appear to be gross similarities between the bonding descriptions
for the Cu and Ni species, differences in placement along the
superoxo-peroxo continuum are likely, particularly in view of
the differences between the complex geometries and the
electron-donating capabilities of their supporting ligands.
Mechanism of Formation of L1CuO2. The data acquired
from cryogenic stopped-flow kinetic experiments for the
oxygenation of L1Cu(NCR) complexes support a mechanism
(Scheme 1) involving two paths that differ kinetically with
respect to the order in O2 (eq 8). According to this mechanism,
direct bimolecular attack of O2 onto the Cu(I) complex (pathway
A, first-order dependence on O2) occurs in parallel with
reversible replacement of nitrile by THF followed by oxygen-
ation (pathway B, no dependence on O2). Precedence for this
model is amply provided by numerous studies of ligand-
substitution processes in square planar d8 complexes that proceed
by dual paths, one that is dependent on incoming ligand and
the other involving a solvent-assisted process that is not.64
Solvent interference in oxygenations of Cu(I) complexes also
has been reported, wherein the reactions are retarded in the
presence of free nitrile.27,71-73 Similarly, in the presence of
excess MeCN (or other nitriles), the reaction of L1Cu(MeCN)
with O2 is slowed. Under these conditions, pathway B is
effectively shut down, enabling detailed assessment of the direct
bimolecular pathway A.
Activation parameters obtained for pathway A for the
oxygenation of L1Cu(MeCN) are compared to illustrative values
reported previously for analogous O2-dependent reactions of
other mononuclear Cu(I) complexes, as shown in Table
2.22b,27,28,74 The low activation enthalpy observed for L1Cu-
(MeCN) is similar to that observed for other compounds, but
the ∆Sq value sets a record low for such reactions. We suggest
that this highly unfavorable (negative) entropy of activation is
due to the high degree of steric hindrance around the metal ion
created by L1. This hypothesis is supported by previous work
that established a connection between ∆Sq and metal complex
steric properties in oxygenation reactions.72,75,76
coordination
number
oxidation
state
standard
deviationa
averagea
rangea
entriesb
3
3
4
4
I
1.940
1.887
1.926
1.892
1.90-1.992
1.858-1.923
1.88-1.987
1.881-1.902
0.025
0.020
0.021
0.009
26
18
30
4
II
II
III
a Units are in Å. b Number of independent Cu-N(â-diketiminate)
distances considered.
companied by a decrease in the average Cu-N distance. The
near identity of the average Cu-N distances for three-coordinate
Cu(II) and four-coordinate Cu(III) reflect the importance of
coordination number in the overall analysis. Notably, the Cu-N
distances in L2CuO2 (1.856(6) Å) and L1CuO2 (avg Cu-N/O
) 1.83 Å) are less than any of the average values listed in Table
7. Moreover, while closely congruent with the shortest reported
distance for a three-coordinate Cu(II) complex (1.858(2) Å),
the distances for the 1:1 adducts are less than any of those
reported for four-coordinate compounds at Cu(II) or Cu(III)
levels. Taken together, the data are consistent with a formulation
for LCuO2 that features considerable Cu(III)-peroxo character.
It is of interest to compare the bonding description for LCuO2
and that for the related complex [PhTtAd]NiO2 reported recently
(PhTtAd ) phenyltris((1-adamantylthio)methyl)borate).70 The Ni
adduct was assigned as a side-on (η2) O2 complex on the basis
of EXAFS spectroscopy and density functional calculations. The
authors concluded that a Ni(II)-superoxo formalism is most
consistent with (a) its XAS edge energy ∼1 eV below that of
a bis(µ-oxo)dinickel(III,III) complex, (b) its EPR spectrum,
which contains a signal indicative of a single unpaired electron
2
highly localized in a Ni dz orbital, and (c) a calculated O-O
distance of 1.38 Å. In addition, DFT calculations showed that
the SOMO is indeed highly localized on Ni, and that the doubly
occupied HOMO below the SOMO covalently mixes the Ni
2
2
dx 9y orbital with the in-plane O2 π*. The authors interpret
this doubly occupied orbital to represent a strong antiferromag-
netic coupling between a superoxide-like O2 and Ni(II).
Our own analysis of this situation suggests that the Ni-O2
interaction is somewhat analogous to that which we have defined
for LCuO2 and should be thought of as lying roughly midway
along a continuum between Ni(II)-superoxide and Ni(III)-
peroxide character,32 where the connection between the two is
dictated by the degree of charge transfer from Ni to O2 in the
covalent orbital in question. The published computational results
support this analysis from a structural standpoint.70 Thus, the
computed O-O bond length of 1.38 Å is shorter than those
typically found in peroxides (ca. 1.42 Å or greater) but is
substantially longer than a typical superoxide bond length, in
particular the 1.29 Å distance computed by the authors for an
isomer of [PhTtAd]NiO2, in which the O2 fragment is bound
end-on to Ni and is much more accurately described as a pure
Ni(II)-superoxide species. Finally, we note that while there
Further insight into pathway A for the reaction of L1Cu-
(MeCN) with O2 was provided by a Hammett study of
oxygenations performed in the presence of excess quantities of
various para-substituted benzonitriles. The excellent correlation
of kA with substituent σp values (Figure 8) represents a
quantitative measure of the sensitivity of oxygenation rate
constants to electronic effects across a mechanistically homo-
geneous series of metal complexes. The resulting F of -0.34
reflects a modest build-up of positive charge at the metal center
in the rate-controlling transition state or, in other words,
nucleophilic character for the metal complex in the oxygenation
process. This result is not surprising considering the degree of
reduction of the O2 molecule and corresponding degree of
oxidation of the L1Cu fragment in the L1CuO2 product, shown
above to have considerable Cu(III)-peroxide character. A
(71) Becker, M.; Schindler, S.; van Eldik, R. Inorg. Chem. 1994, 33, 5370.
(72) Karlin, K. D.; Kaderli, S.; Zuberbu¨hler, A. D. Acc. Chem. Res. 1997, 30,
1399.
(73) Liang, H.-C.; Karlin, K. D.; Dyson, R.; Kaderli, S.; Jung, B.; Zuberbu¨hler,
A. D. Inorg. Chem. 2000, 39, 5884.
(74) Halfen, J. A.; Mahapatra, S.; Wilkinson, E. C.; Kaderli, S.; Young, V. G.,
Jr.; Que, L., Jr.; Zuberbu¨hler, A. D.; Tolman, W. B. Science 1996, 271,
1397.
(75) Rybak-Akimova, E. V.; Marek, K.; Masarwa, M.; Busch, D. H. Inorg. Chim.
Acta 1998, 270, 151.
(76) Kryatov, S. V.; Rybak-Akimova, E. V.; MacMurdo, V. L.; Que, L., Jr.
Inorg. Chem. 2001, 40, 2220.
(69) We considered an alternative bond valence sum analysis but chose the
present simpler method due to complications associated with an appropriate
choice of reference (r0) values: (a) Brown, I. D.; Altermatt, D. Acta
Crystallogr. 1985, B41, 240 and 244. (b) Thorp, H. H. Inorg. Chem. 1992,
31, 1585. (c) Hati, S.; Datta, D. J. Chem. Soc., Dalton Trans. 1995, 1177.
(d) Liu, W.; Thorp, H. H. Inorg. Chem. 1993, 32, 4102.
(70) Fujita, K.; Schenker, R.; Gu, W.; Brunold, T. C.; Cramer, S. P.; Riordan,
C. G. Inorg. Chem. 2004, 43, 3324.
9
J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004 16909

A R T I C L E S
Aboelella et al.
similar trend of greater O2 reactivity with increased ligand
electron donation was reported previously for some other
Cu(I) complexes,27,72,77,78 but quantitative Hammett-type cor-
relations between O2 reactivity and substituent electronic prop-
erties have only been drawn in one instance.79 In this study of
the oxygenation of Cu(I) complexes of substituted phenanthro-
lines in which O2 adducts were not isolated, a plot of the third-
order rate constants (rate ) k[Cu(I)]2[O2]) yielded F ) -0.71.
Thermodynamic stabilization of O2 binding to metal sites by
electron-donating ligands has been demonstrated for numerous
systems (e.g., hemes and analogues),80 but this effect has been
attributed primarily to decreased rate constants for O2 dissocia-
tion, with little contribution from changes in rate constants for
the O2 association step.75,81,82
Theoretical calculations corroborate the mechanistic picture
deduced from the experimental results and reveal further details
of the reaction trajectory for pathway A. Importantly, excellent
agreement between experimentally and theoretically determined
activation parameters for the oxygenation of L1Cu(MeCN) via
pathway A when THF solvation is appropriately considered
lends support to the calculated reaction profile. According to
these calculations, the transition state (1) features an O2 fragment
that is weakly bound and an MeCN that is significantly
displaced, yet still strongly bound; this is consistent with the
observed Hammett rate constant correlation when substituted
benzonitriles were used. The calculations also showed that loss
of MeCN from 1 and movement of the coordinated O2 moiety
to its final symmetric η2 position is strongly exergonic, in
agreement with the observed irreversibility of the oxygenation
reaction. Such irreversibility is unique among Cu(I)-O2 reac-
tions and, in combination with the absence of dicopper complex
formation for this system that is more typical in Cu(I)-O2
chemistry,17 enables the use of the L1CuO2 as a synthon for the
construction of novel asymmetric bimetallic31 and heterobime-
tallic complexes.83
The measured activation parameters for pathway B (Table
2) reflect the process by which THF substitutes for coordinated
MeCN (kB, kB1) since this step is slow and rate-controlling
relative to the subsequent oxygenation of L1Cu(THF). It is
probably coincidental that the entropy of activation for the ligand
substitution step in pathway B is identical within experimental
error to that measured for the oxygenation step of pathway A.
Nonetheless, one may argue that steric hindrance for addition
of a ligand (THF or O2) to L1Cu(MeCN) underlies the large
negative values found for both steps, particularly in view of
the mechanistic similarities between them (associative processes;
compare 1 to 4 in Figure 9). In contrast, the value of ∆Hq
is significantly larger for the addition of THF (30 kJ
mol-1) than for the addition of O2 (18 kJ mol-1) to L1Cu-
(MeCN). This difference may be rationalized by considering
that the nascent Cu-O2 bond(s) are stronger than the nascent
Cu-THF interaction. The weak Cu(I)-THF bond also is
reflected in the endergonic nature of the conversion of L1Cu-
(MeCN) to L1Cu(THF) (∆G ) 15.3 kJ mol-1 at 223 K in THF;
Table 6).
In addition, the weak Cu(I)-THF interaction contributes to
the lower barrier for oxygenation of L1Cu(THF) relative to that
of L1Cu(MeCN), as reflected by the calculated (223 K, THF)
free energy differences between L1Cu(MeCN) and 1 (39.0 kJ
mol-1) versus between L1Cu(THF) and t.s. A (18.3 kJ mol-1).
The reaction trajectories for these oxygenations differ as well,
such that O2 adopts a side-on position essentially directly upon
binding to the copper ion in pathway A (without an end-on
intermediate) but first binds end-on in its reaction with L1Cu-
(THF) prior to rapid isomerization to its final side-on position
in L1CuO2 in pathway B (Figure 9). We postulate that this
difference derives from steric factors; direct side-on binding of
O2 is inhibited by the greater bulk of THF relative to MeCN.
Our finding of greater oxygenation reactivity for L1Cu(THF)
compared to that for L1Cu(MeCN) is consistent with pre-
vious reports for the (TMPA)Cu(I) system, where (a) oxygen-
ation of (TMPA)Cu(MeCN) in EtCN was slower than in THF
and (b) (TMPA)Cu(THF) generated by photolysis of (TMPA)-
CuCO+ in THF reacted with O2 at a rate close to the diffusion
limit.27
Analysis of the Mulliken charge changes, which take place
in the course of the reaction (Table S8), reveals that the electron
density responsible for the reduction of dioxygen originates not
entirely (or even largely) from the Cu(I) center. Rather, the
Cu(I) center and the formally negatively charged â-diketiminate
ligand contribute equally to the reduction of dioxygen. The
electron-rich ligand thus plays a central role not only in
stabilizing the oxidized Cu center but also in allowing for
efficient reduction of the O2 moiety. The importance of the
ligand is especially noticeable in the transition state (1) in the
bimolecular pathway, in which charge essentially flows from
the â-diketiminate group, through the Cu, and into dioxygen.
Notably, the electronics at this transition state, in which the
L1Cu unit has become partially oxidized, are consistent with
the negative Hammett F value, implicating positive charge build-
up on LCu at the transition state of the bimolecular pathway.
Finally, one may also draw a conceptual connection between
the metal-ligand interactions during O2 binding by LCu to those
postulated during dioxygen activation by heme-iron centers,
wherein the metal has been suggested to “relay the electron flux”
from the porphyrin to a bound dioxygen moiety.84
Conclusion
Detailed insight into the process of O2 activation at a single
copper center supported by â-diketiminate ligands has been
obtained through a synergistic combination of experiment and
theory. The degree of O2 reduction upon binding to the LCu
system is extensive, as indicated by previous theoretical
calculations and vibrational data,30-32 as well as the new X-ray
crystallographic, Cu K-edge XAS, EXAFS, and theoretical
results reported herein, all of which are consistent with
significant Cu(III)-peroxo character in LCuO2. A dual pathway
(77) Karlin, K. D.; Nasir, M. S.; Cohen, B. I.; Cruse, R. W.; Kaderli, S.;
Zuberbu¨hler, A. D. J. Am. Chem. Soc. 1994, 116, 1324.
(78) Lee, D.-H.; Wei, N.; Murthy, N. N.; Tyeklar, Z.; Karlin, K. D.; Kaderli,
S.; Jung, B.; Zuberbu¨hler, A. D. J. Am. Chem. Soc. 1995, 117, 12498.
(79) Arce, J. A.; Spodine, E.; Zamudio, W. J. Inorg. Nucl. Chem. 1976, 38,
2029.
(80) Momenteau, M.; Reed, C. A. Chem. ReV. 1994, 94, 659.
(81) Momenteau, M.; Loock, B.; Lavalette, D.; Te´treau, C.; Mispelter, J. J. Chem.
Soc., Chem. Commun. 1983, 962. (b) Traylor, T. G.; Traylor, P. S. Annu.
ReV. Biophys. Bioeng. 1982, 11, 105.
(82) Collman, J. P.; Brauman, J. I.; Doxsee, K. M.; Sessler, J. L.; Morris, R.
M.; Gibson, Q. H. Inorg. Chem. 1983, 22, 1427.
(83) Aboelella, N. W.; York, J. T.; Reynolds, A. M.; Fujita, K.; Kinsinger, C.
R.; Cramer, C. J.; Riordan, C. G.; Tolman, W. B. Chem. Commun. 2004,
1716.
(84) (a) Silaghi-Dumitrescu, R. J. Biol. Inorg. Chem. 2004, 9, 471. (b) Johansson,
M. P.; Blomberg, M. R. A.; Sundholm, D.; Wikstro¨m, M. Biochim. Biophys.
Acta 2003, 1553, 183.
9
16910 J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004

Dioxygen Activation at a Single Copper Site
A R T I C L E S
mechanism for the oxygenation of L1Cu(RCN) in THF and
THF/RCN mixtures is proposed on the basis of cryogenic
stopped-flow kinetics data and theoretical calculations. One
pathway (B in Scheme 1) involves rate-controlling solvolysis
via an associative interchange mechanism, followed by rapid
substitution of the weakly bound THF ligand by O2. Calculations
indicate that in this path, end-on O2 coordination occurs prior
to rapid isomerization to the final side-on structure of the
product. In the presence of excess RCN, this solvolytic pathway
is suppressed and a direct bimolecular, associative substitution
by O2 occurs instead (pathway A). A detailed picture of this
oxygenation process has been provided by the experimental
kinetic data, in particular the ∆Hq, ∆Sq, and Hammett F values,
in conjunction with a theoretical assessment of the reaction
trajectory (both in the gas phase and in THF solution) that
yielded activation parameters highly congruent with those
obtained by experiment. The results suggest that electron density
flows from both the â-diketiminate and the Cu ion to the O2
moiety, with a transition-state structure that features both
coordinated nitrile and dioxygen (weakly bound, in asymmetric
fashion). Taken together, the results of this study yield a
comprehensive understanding of O2 binding in a synthetic
copper system, a fundamentally important process that represents
a model for how dioxygen activation may occur at single copper
centers in biology and catalysis.
Acknowledgment. We thank the NIH (GM47365 to
W.B.T., postdoctoral fellowship to B.F.G., DK-31450 to
E.I.S., RR-01209 to K.O.H.), NSF (CHE-0203346 to C.J.C.,
CHE-0111202 to E.V.R.-A., predoctoral fellowship to N.W.A.),
and the Research Corporation (RI0223 to E.V.R.-A.). Chem-
MatCARS Sector 15 is principally supported by the National
Science Foundation/Department of Energy under Grant CHE-
0087817, and by the Illinois board of higher education. The
Advanced Photon Source is supported by the U.S. Department
of Energy, Basic Energy Sciences, Office of Science, under
Contract W-31-109-Eng-38. The XAS data were measured at
SSRL, which is supported by the U.S. Department of Energy,
Basic Energy Sciences. The SSRL Structural Molecular Biology
Program is supported by the U.S. Department of Energy, Office
of Biological and Environmental Research, and by the National
Institutes of Health, National Center for Research Resources,
and Biomedical Technology Program.
Supporting Information Available: Calculation details, ki-
netic data, X-ray structure representations, and tabulation of
bond distances for â-diketiminate complexes of copper (PDF),
and X-ray crystallographic data (CIF). This material is available
free of charge via the Internet at http://pubs.acs.org.
JA045678J
9
J. AM. CHEM. SOC. VOL. 126, NO. 51, 2004 16911