Mechanism of Pd(OAc)2/DMSO-catalyzed aerobic alcohol oxidation: Mass-transfer-limitation effects and catalyst decomposition pathways

Article abstract of DOI:10.1021/ja057914b

Pd(OAc)₂ in DMSO is an effective catalyst for the aerobic oxidation of alcohols and numerous other organic substrates. Kinetic studies of the catalytic oxidation of primary and secondary benzylic alcohol substrates provide fundamental insights into the catalytic mechanism. In contrast to the conclusion reached in our earlier study (J. Am. Chem. Soc. 2002, 124, 766-767), we find that Pd(II)-mediated alcohol oxidation is the turnover-limiting step of the catalytic reaction. At elevated catalyst loading, however, the rate of catalytic turnover is limited by the dissolution of oxygen gas into solution. This mass-transfer rate is measured directly by using gas-uptake methods, and it correlates with the maximum rate observed during catalysis. Initial-rate studies were complemented by kinetic analysis of the full-reaction timecourses at different catalyst concentrations. Kinetic fits of these traces reveal the presence of unimolecular and bimolecular catalyst decomposition pathways that compete with productive catalytic turnover.

Full text of DOI:10.1021/ja057914b

Published on Web 03/10/2006
Mechanism of Pd(OAc)2/DMSO-Catalyzed Aerobic Alcohol
Oxidation: Mass-Transfer-Limitation Effects and Catalyst
Decomposition Pathways
Bradley A. Steinhoff and Shannon S. Stahl*
Contribution from the Department of Chemistry, UniVersity of WisconsinsMadison,
1101 UniVersity AVenue, Madison, Wisconsin 53706
Received November 22, 2005; E-mail: stahl@chem.wisc.edu
Abstract: Pd(OAc)2 in DMSO is an effective catalyst for the aerobic oxidation of alcohols and numerous
other organic substrates. Kinetic studies of the catalytic oxidation of primary and secondary benzylic alcohol
substrates provide fundamental insights into the catalytic mechanism. In contrast to the conclusion reached
in our earlier study (J. Am. Chem. Soc. 2002, 124, 766-767), we find that Pd(II)-mediated alcohol oxidation
is the turnover-limiting step of the catalytic reaction. At elevated catalyst loading, however, the rate of
catalytic turnover is limited by the dissolution of oxygen gas into solution. This mass-transfer rate is measured
directly by using gas-uptake methods, and it correlates with the maximum rate observed during catalysis.
Initial-rate studies were complemented by kinetic analysis of the full-reaction timecourses at different catalyst
concentrations. Kinetic fits of these traces reveal the presence of unimolecular and bimolecular catalyst
decomposition pathways that compete with productive catalytic turnover.
Introduction
Palladium-catalyzed oxidation reactions rank among the most
versatile strategies for the selective oxidation of organic
molecules with molecular oxygen. Effective transformations
range from alcohol oxidation and oxidative carbonylation
reactions to oxidative C-C, C-O, and C-N coupling reactions
with alkenes.1 The Wacker process (eq 1), discovered in the
late 1950s,2 represents the most prominent example of this
reaction class; however, substantial advances have occurred over
the past decade. The recent developments can be linked to the
discovery that coordinating solvents and oxidatively stable
ligands enable dioxygen-coupled catalytic turnover to be
achieved in the absence of cocatalytic and/or stoichiometric
oxidants such as benzoquinone and copper(II) chloride. The Pd-
(OAc)2/DMSO catalyst system, discovered independently by the
groups of Larock and Hiemstra in the mid-1990s, was among
the first breakthroughs in this area.3,4 Although earlier examples
of cocatalyst-free aerobic oxidation reactions were known,5 this
catalyst system was the first to find broad applicability to a
wide range of organic oxidation reactions (e.g., eqs 2-5).
reactions: substrate oxidation by palladium(II) followed by
aerobic oxidation of the reduced catalyst (steps i and ii, Scheme
1).1a Mechanistic studies have illuminated important fundamen-
tal steps associated with these half reactions,7 and it is evident
Numerous additional catalyst systems have been reported
subsequently (Chart 1), particularly for application to aerobic
alcohol oxidation.6 The catalytic mechanism of these reactions
features an oxidase-style sequence consisting of two half
(3) For leading references, see: (a) Larock, R. C.; Hightower, T. R. J. Org.
Chem. 1993, 58, 5298-5300. (b) van Benthem, R. A. T. M.; Hiemstra,
H.; Michels, J. J.; Speckamp, W. N. J. Chem. Soc., Chem. Commun. 1994,
357-359. (c) van Benthem, R. A. T. M.; Hiemstra, H.; Longarela, G. R.;
Speckamp, W. N. Tetrahedron Lett. 1994, 35, 9281-9284. (d) Larock, R.
C.; Hightower, T. R.; Kraus, G. A.; Hahn, P.; Zheng, D. Tetrahedron Lett.
1995, 36, 2423-2426. (e) van Benthem, R. A. T. M.; Hiemstra, H.; van
Leeuwen, P. W. N. M.; Geus, J. W.; Speckamp, W. N. Angew. Chem., Int.
Ed. Engl. 1995, 34, 457-460. (f) Larock, R. C.; Hightower, T. R.; Hasvold,
L. A.; Peterson, K. P. J. Org. Chem. 1996, 61, 3584-3585. (g) Peterson,
K. P.; Larock, R. C. J. Org. Chem. 1998, 63, 3185-3189.
(1) For recent review articles and accounts, see the following: (a) Stahl, S. S.
Science 2005, 309, 1824-1826. (b) Stahl, S. S. Angew. Chem., Int. Ed.
2004, 43, 3400-3420. (c) Sheldon, R. A.; Arends, I. W. C. E.; ten Brink,
G.-J.; Dijksman, A. Acc. Chem. Res. 2002, 35, 774-781. (d) Stoltz, B. M.
Chem. Lett. 2004, 33, 362-367. (e) Sigman, M. S.; Schultz, M. J. Org.
Biomol. Chem. 2004, 2, 2551-2554. (f) Toyota, M.; Ihara, M. Synlett 2002,
1211-1222. (g) Nishimura, T.; Uemura, S. Synlett 2004, 201-216.
(2) Smidt, J.; Hafner, W.; Jira, R.; Sedlmeier, J.; Sieber, R.; Ru¨ttinger, R.;
Kojer, H. Angew. Chem. 1959, 71, 176-182.
9
4348
J. AM. CHEM. SOC. 2006, 128, 4348-4355
10.1021/ja057914b CCC: $33.50 © 2006 American Chemical Society

Pd-Catalyzed Aerobic Alcohol Oxidation Mechanism
A R T I C L E S
Chart 1. Catalyst Systems that Promote Direct Dioxygen-Coupled
Scheme 1. General Mechanism of Palladium-Catalyzed Aerobic
Alcohol Oxidation
Oxidation of Organic Substrates
conditions commonly employed for synthetic reactions. Namely,
the catalytic rate can be controlled by the rate of oxygen gas
dissolution into DMSO. If an adequate supply of dioxygen is
available in solution, the Pd(OAc)2/DMSO system also exhibits
turnover-limiting alcohol oxidation by palladium(II). Because
palladium(0) is an intrinsically unstable form of the catalyst
(cf. Scheme 1), slow dissolution of oxygen into the reaction
mixture not only limits the reaction rate but also leads to
enhanced catalyst decomposition. Mechanistic insights into
catalyst decomposition pathways are obtained from analysis of
the full kinetic timecourses of catalytic reactions. The results
of this study and the methods described herein are applicable
to the entire scope of palladium-catalyzed aerobic oxidation
reactions.
from these studies that the ligands coordinated to palladium exert
a significant influence on the reaction, including the rate
(turnover frequency), the nature of the catalyst resting state, and
the identity of the turnover-limiting step. Ligands also influence
the catalyst stability by affecting the rate of catalyst reoxidation
relative to decomposition (steps ii and iii, Scheme 1).
Pd(OAc)2/DMSO is the only catalyst system for which
aerobic oxidation of the catalyst (step ii, Scheme 1) has been
reported to be the turnover-limiting step.7a,8 In all other systems,
steps associated with palladium(II)-mediated oxidation of the
alcohol are turnover limiting (step i, Scheme 1).7 In the current
study, we present a more thorough kinetic analysis of Pd(OAc)2/
DMSO-catalyzed aerobic alcohol oxidation that reveals this
catalyst system is susceptible to mass-transfer-limited rates under
Results and Discussion
Kinetic Studies of Alcohol Oxidation. Our initial mecha-
nistic studies of the Pd(OAc)2/DMSO system focused on the
oxidation of 2,5-dimethoxybenzyl alcohol, 1.7a This reaction,
originally reported by Peterson and Larock,3g proceeds to nearly
complete conversion within 12 h at 80 °C (eq 6). In our earlier
(4) Other groups have also employed the Pd(OAc)2/DMSO catalyst system in
aerobic oxidation reactions. See, for example: (a) Ref 1f. (b) Ro¨nn, M.;
Ba¨ckvall, J.-E.; Andersson, P. G. Tetrahedron Lett. 1995, 36, 7749-7752.
(c) Ro¨nn, M.; Andersson, P. G.; Ba¨ckvall, J.-E. Acta Chem. Scand. 1997,
51, 773-777. (d) Bee, C.; Leclerc, E.; Tius, M. A. Org. Lett. 2003, 5,
4927-4930.
(5) See, for example: (a) Davidson, J. M.; Triggs, C. Chem. Ind. 1967, 1361.
(b) Itatani, H.; Yoshimoto, H. Chem. Ind. 1971, 674-675. (c) Blackburn,
T. F.; Schwartz, J. J. Chem. Soc., Chem. Commun. 1977, 157-158. (d)
Hosokawa, T.; Miyagi, S.; Murahashi, S.; Sonoda, A. J. Org. Chem. 1978,
43, 2752-2757.
(6) See review articles in ref 1 and the following leading references: Pd(OAc)2/
pyridine: (a) Nishimura, T.; Onoue, T.; Ohe, K.; Uemura, S. J. Org. Chem.
1999, 64, 6750-6755. (b) Iwasawa, T.; Tokunaga, M.; Obora, Y.; Tsuji,
Y. J. Am. Chem. Soc. 2004, 126, 6554-6555. Pd(OAc)2/phenanthroline
derivatives: (c) ten Brink, G.-J.; Arends, I. W. C. E.; Sheldon, R. A. Science
2000, 287, 1636-1639. (d) Bianchi, D.; Bortolo, R.; D’Aloisio, R.; Ricci,
M. Angew. Chem., Int. Ed. 1999, 38, 706-708. PdCl2/(-)-sparteine: (e)
Ferreira, E. M.; Stoltz, B. M. J. Am. Chem. Soc. 2001, 123, 7725-7726.
(f) Jensen, D. R.; Pugsley, J. S.; Sigman, M. S. J. Am. Chem. Soc. 2001,
123, 7475-7476. PdX2/NEt3: (g) Schultz, M. J.; Park, C. C.; Sigman, M.
S. Chem. Commun. 2002, 3034-3035. (h) Timokhin, V. I.; Anastasi, N.
R.; Stahl, S. S. J. Am. Chem. Soc. 2003, 125, 12996-12997. (NHC)Pd-
(OAc)2(OH2): (i) Jensen, D. R.; Schultz, M. J.; Mueller, J. A.; Sigman,
M. S. Angew. Chem., Int. Ed. 2003, 42, 3810-3813. Palladacycles: (j)
Hallman, K.; Moberg, C. AdV. Synth. Catal. 2001, 343, 260-263.
(7) For mechanistic studies of catalytic aerobic alcohol oxidation with different
catalyst systems, see the following. Pd(OAc)2/DMSO: (a) Steinhoff, B.
A.; Fix, S. R.; Stahl, S. S. J. Am. Chem. Soc. 2002, 124, 766-767. Pd-
(OAc)2/pyridine: (b) Steinhoff, B. A.; Stahl, S. S. Org. Lett. 2002, 4, 4179-
4181. (c) Steinhoff, B. A.; Guzei, I. A.; Stahl, S. S. J. Am. Chem. Soc.
2004, 126, 11268-11278. Pd(OAc)2/phenanthroline derivatives: (d) ten
Brink, G.-J.; Arends, I. W. C. E.; Sheldon, R. A. AdV. Synth. Catal. 2002,
344, 355-369. (e) ten Brink, G.-J.; Arends, I. W. C. E.; Hoogenraad, M.;
Verspui, G.; Sheldon, R. A. AdV. Synth. Catal. 2003, 345, 497-505. (f)
ten Brink, G.-J.; Arends, I. W. C. E.; Hoogenraad, M.; Verspui, G.; Sheldon,
R. A. AdV. Synth. Catal. 2003, 345, 1341-1352. (g) Bortolo, R.; Bianchi,
D.; D’Aloisio, R.; Querci, C.; Ricci, M. J. Mol. Catal. A 2000, 153, 25-
29. PdCl2/(-)-sparteine: (h) Mueller, J. A.; Jensen, D. R.; Sigman, M. S.
J. Am. Chem. Soc. 2002, 124, 8202-8203. (i) Mueller, J. A.; Sigman, M.
S. J. Am. Chem. Soc. 2003, 125, 7005-7013. (j) Trend, R. M.; Stoltz, B.
M. J. Am. Chem. Soc. 2004, 126, 4482-4483. Pd(OAc)2/NEt3: (k) Schultz,
M. J.; Adler, R. S.; Zierkiewicz, W.; Privalov, T.; Sigman, M. S. J. Am.
Chem. Soc. 2005, 127, 8499-8507. (NHC)Pd(OAc)2(OH2): (l) Mueller,
J. A.; Goller, C. P.; Sigman, M. S. J. Am. Chem. Soc. 2004, 126, 9724-
9734. Palladacycles: (m) Paavola, S.; Zetterberg, K.; Privalov, T.; Cso¨regh,
I.; Moberg, C. AdV. Synth. Catal. 2004, 346, 237-244.
study, we demonstrated that the reaction stoichiometry corre-
sponds to an O2/substrate ratio of 0.50(3). Control experiments
reveal that hydrogen peroxide, if it forms in the reaction, is not
competent to serve as an oxidant because it undergoes rapid
disproportionation in the presence of the palladium catalyst.9
Kinetic studies of the catalytic reaction were performed by
using a computer-interfaced gas-uptake apparatus to monitor
the change in oxygen pressure within in a sealed, temperature-
controlled reaction vessel. The lack of an induction period
enabled us to obtain much of our kinetics data via initial-rate
methods. Analysis of the full-reaction timecourse provides
additional insights into the catalytic mechanism (see below).
The dependence of the rate on [Pd(OAc)2] displays a
nonlinear correlation that fits well to a hyperbolic function
(saturation-like kinetics). The curve-fit asymptotically ap-
proaches a maximum rate of 126 µmol/min (Figure 1A). The
(9) For a detailed description of the control experiments used to establish this
fact, see the Supporting Information of ref 7a. The precise mechanism of
disproportionation is not known; however, we have made the following
qualitative observations. Addition of aqueous hydrogen peroxide (30%) to
a fresh solution of Pd(OAc)2 (recrystallized from toluene) in DMSO results
in rapid disproportionation. In contrast, if hydrogen peroxide is added to a
suspension of palladium black obtained from a completed catalytic reaction,
disproportionation is slow. These observations suggest that the Pd-catalyzed
disproportionation mechanism may proceed via a Pd(II)/Pd(IV) cycle.
(8) For a recent computational study of the Pd(OAc)2/DMSO system, see:
Zierkiewicz, W.; Privalov, T. Organometallics 2005, 24, 6019-6028.
9
J. AM. CHEM. SOC. VOL. 128, NO. 13, 2006 4349

A R T I C L E S
Steinhoff and Stahl
Figure 1. Dependence of the initial rate on [Pd(OAc)2] for the oxidation of primary alcohol, 1, in two different reaction vessels: (A) 25 mL round-bottom
flask and (B) 25.4 mm OD high-pressure glass reaction tube. Conditions: (A) 2.2-100 mM Pd(OAc)2, 745 Torr O2, 0.52 M 2,5-dimethoxybenzyl alcohol,
2 mL of DMSO, 80 °C, data fit to a hyperbolic function; (B) 0.48-98.9 mM Pd(OAc)2, 725 Torr O2, 520 mM 2,5-dimethoxybenzyl alcohol, 2 mL of
DMSO, 80 °C, data fit to a hyperbolic function.
that yields the maximum rate. Under these conditions, the
reaction rate exhibits a first-order dependence on the oxygen
pressure and is invariant with respect to [alcohol] (Figure 3C,D).
Similar kinetic studies were performed for the oxidation of
secondary alcohol 2 at low and high catalyst loading (2.5 and
25 mM, respectively), and identical trends emerged (Supporting
Information, Figure S1). At low [Pd(OAc)2], the reaction rate
exhibits an approximately zero-order dependence on the oxygen
pressure and a saturation dependence on [alcohol], and at high
[Pd(OAc)2], the rate exhibits a first-order dependence on oxygen
pressure and a zero-order dependence on [alcohol].
Figure 2. Dependence of the initial rate on [Pd(OAc)2] for the oxidation
of secondary alcohol, 2. Inset: region in which the rate dependence is first
order on [Pd]. Conditions: 1.3-125 mM Pd(OAc)2, 735 Torr O2, 0.52 M
1-phenylethanol, 2 mL of DMSO, 80 °C, in 25.4 mm OD high-pressure
tube.
Characterization of Mass-Transfer-Limited Conditions.
The data described above suggest that reactions conducted at
high catalyst loading might experience rate-limiting mass-
transfer of oxygen gas into solution. Data consistent with mass-
transfer-limitation effects include (1) the invariance of the rate
on [Pd(OAc)2] above a threshold concentration, (2) the first-
order dependence of the rate on oxygen pressure, and (3) the
dependence of the maximum rate on the identity of the reaction
vessel under otherwise identical conditions (Figure 1 and Table
1).10 Additional studies provided further support for this
conclusion. At high catalyst loading (25 mM Pd(OAc)2), the
initial rate of the catalytic reaction exhibits a linear dependence
on the stirring frequency (closed circles, Figure 4). At low
catalyst loading (2.5 mM Pd(OAc)2), no stir-rate dependence
is observed if stirring is maintained at g2000 rpm. Below this
value, however, a linear dependence of the rate on stirring
frequency is observed. Thus, even at low catalyst concentration,
the reaction rate is susceptible to mass-transfer limitations if
inadequate agitation is provided.
[Pd(OAc)] dependence differs somewhat if the identical reac-
tions are conducted in a different reaction vessel, namely, a 25.4
mm OD high-pressure glass reaction tube instead of a 25 mL
round-bottom flask (Figure 1B). Again, a hyperbolic function
fits the data reasonably well; however, some scatter is evident
in the data above ∼18 mM Pd(OAc)2, and in this case, the fit
predicts a maximum rate of only 76 µmol/min.
Subsequent kinetic studies of Pd(OAc)2/DMSO-catalyzed
oxidation of 1-phenylethanol, 2 (eq 7), provided valuable
additional insights. For this reaction, the [Pd(OAc)2]-dependence
plot reveals a sharp discontinuity at 12 mM Pd(OAc)2 (Figure
2). Above this concentration, the reaction rate is invariant with
respect to [Pd(OAc)2)]. The rate plateau observed in these
studies (70 µmol/min) is comparable to the maximum rate
observed for the oxidation of the primary alcohol 1 in the same
reaction vessel (high-pressure reaction tube; Figure 1B). In
contrast to the data obtained for the oxidation of 1 (Figure 1),
no curvature is evident in the data for the oxidation of 2 at low
[Pd] (see inset, Figure 2).
On the basis of these results, we performed additional kinetic
studies on the oxidation of 1 and 2 at two different concentra-
tions of Pd(OAc)2. In the oxidation of 1, reactions conducted
at low catalyst loading (0.5 mol %, or 2.5 mM, Pd(OAc)2) reveal
that the rate is virtually independent of the oxygen pressure and
exhibits a saturation dependence on [alcohol] (Figure 3A,B).
Similar measurements were obtained at higher catalyst loading
(5 mol %, or 25 mM, Pd(OAc)2), conditions commonly
employed for synthetic reactions and in the range of [Pd(OAc)2]
Under mass-transfer-limiting conditions, the dissolved gas
concentration approaches zero because the chemical reaction
depletes the gaseous substrate from solution faster than it can
be replenished (eq 8).
kL(area)
krxn[substrate(s)]
Gas (g) y
z Gas (soln) y
z Product (8)
-kL(area)
The rate of mass transfer is influenced by the surface area of
the liquid exposed to the gas and the gas pressure (eq 9).
d[gas]
dt
) kL(area)‚pgas
(9)
Several factors influence the surface area of the liquid exposed
(10) For a discussion of mass-transfer-limitation effects in catalysis, see: Roberts,
G. W. In Catalysis in Organic Syntheses; Rylander, P. N., Greenfield, H.,
Eds.; Academic Press: New York, 1976; pp 1-48.
9
4350 J. AM. CHEM. SOC. VOL. 128, NO. 13, 2006

Pd-Catalyzed Aerobic Alcohol Oxidation Mechanism
A R T I C L E S
Figure 3. Dependence of the initial rate on oxygen pressure and alcohol concentration at both low and high catalyst loadings in the Pd(OAc)2/DMSO-
catalyzed aerobic oxidation of primary alcohol 1. Conditions: (A) 2.5 mM Pd(OAc)2, 399-913 Torr O2, 0.52 M 2,5-dimethoxybenzyl alcohol, 2 mL of
DMSO, 80 °C, in 25.4 mm OD tube; (B) 2.5 mM Pd(OAc)2, 730 Torr O2, 0.105-1.04 M 2,5-dimethoxybenzyl alcohol, 2 mL of DMSO, 80 °C, in 25.4 mm
OD tube, data fit to a hyperbolic curve; (C) 25.2 mM Pd(OAc)2, 179-735 Torr O2, 520 mM 2,5-dimethoxybenzyl alcohol, 2 mL of DMSO, 80 °C, in 25.4
mm OD tube; (D) 24.9 mM Pd(OAc)2, 725 Torr O2, 0.349-1.05 M 2,5-dimethoxybenzyl alcohol, 2 mL of DMSO, 80 °C, in 25.4 mm OD tube, rates
normalized to 1 atm.
Table 1. Measured Rates of O2 Mass Transfer into DMSO at 80
°C Depending on Vessel Shape and Solution Volumea
solution
volume
(mL)
maximum catalytic
rate
reaction vessel
(µmol/min)
25 mL round-bottom flask
25 mL round-bottom flask
25.4 mm OD high-pressure tube
2.00
5.00
2.00
102.7 ( 9.1
98.5 ( 5.6
69.8 ( 7.4
a Reaction Conditions: 0.52 M 2, 730 Torr O2, 25 mM Pd(OAc)2,
DMSO, 80 °C.
Figure 5. Measurement of the rate of oxygen gas dissolution into 5 mL of
DMSO at 80 ° C. The curve-fit reflects a nonlinear least-squares fit to eq
10.
mM measured at lower temperature (25 °C).11 The exponential
fit of the data (eq 10, Figure 5)12 provides a direct measurement
of the rate of oxygen gas transfer into DMSO (85.4 ( 5.6 µmol/
min).
Pt ) Pfinal + (Pinitial - Pfinal) ×
-(Pinitial - Pbackground)‚kL(area)‚t
exp
(10)
[
]
Pfinal - Pbackground
Figure 4. Dependence of the initial rate of Pd(OAc)2/DMSO-catalyzed
oxidation of 2 on the stirring speed. Conditions: (filled circles) 25.2 mM
Pd(OAc)2, 730 Torr O2, 0.52 M 1-phenylethanol, 400-2950 rpm, 2 mL of
DMSO, 80 °C, in 25.4 mm OD tube; (open circles) 2.5 mM Pd(OAc)2,
730 Torr O2, 0.52 M 1-phenylethanol, 400-2930 rpm, 2 mL of DMSO, 80
°C, in 25.4 mm OD tube.
To correlate these data with a mass-transfer-limited catalytic
reaction, this oxygen dissolution rate was compared to the
maximum rate measured for catalytic reactions carried out under
the same conditions (5.00 mL DMSO, 80 °C, 25 mL round-
bottom flask). The maximum catalytic rate (98.5 ( 5.6 µmol/
min, Table 1) is within two standard deviations of the
independently measured gas dissolution rate (Figure 6).
to the gas-phase substrate, including the reactor design, solvent
volume, and agitation rate. Use of the gas-uptake apparatus
enabled us to obtain a direct measurement of the rate of oxygen
gas dissolution into DMSO under catalytically relevant condi-
tions. The data shown in Figure 5 reveal that 16 ( 1 µmol of
dioxygen dissolves in 5.00 mL of DMSO in a 25 mL round-
bottom flask. This solubility value of 3.2 ( 0.2 mM at 80 °C
compares favorably to literature values ranging from 1.5 to 2.8
(11) (a) Gerrard, W. Solubility of Gases and Liquids: A Graphic Approach.
Data-Causes-Prediction; Plenum Press: New York, 1976; p 24. (b) Achord,
J. M.; Hussey, C. L. Anal. Chem. 1980, 52, 601-602. (c) Battino, R.;
Rettich, T. R.; Tominaga, T. J. Phys. Chem. Ref. Data 1983, 12, 163-
178.
(12) Deimling, A.; Karandikar, B. M.; Shah, Y. T.; Carr, N. L. Chem. Eng. J.
1984, 29, 127-140.
9
J. AM. CHEM. SOC. VOL. 128, NO. 13, 2006 4351

A R T I C L E S
Steinhoff and Stahl
By analogy, catalyst decomposition in the present reactions
could occur via both unimolecular and bimolecular pathways.
Decomposition pathways that might exhibit a unimolecular
dependence on [Pd(OAc)2] include nucleation events associated
with interactions between soluble Pd(0) and impurities in the
solvent, the solvent itself, and/or the wall of the reaction vessel,
for example. The most straightforward bimolecular decomposi-
tion mechanism is direct interaction of two Pd centers in a
nucleation or particle-growth process. The kinetics of palladium
decomposition (i.e., aggregation) have direct implications for
catalyst performance.
Figure 6. Dependence of the initial rate on [Pd(OAc)2]. Conditions: 1.5-
50.2 mM Pd(OAc)2, 800 Torr O2, 0.50 M 1-phenylethanol, 5 mL of DMSO,
80 °C, in round-bottom flask, rates normalized to 1 atm.
Aerobic oxidation of the catalyst should exhibit a first-order
dependence on [Pd] (e.g., rate ) kii[Pd][O2]).14 If the catalyst
decomposes via a unimolecular pathway, the first-order depen-
dence of both catalytic turnover and catalyst decomposition
(steps ii and iii, Scheme 1) will result in a linear dependence
of the turnover rate on [Pd]. Such a trend is observed in the
oxidation of the secondary alcohol 2 (inset, Figure 2). If
decomposition proceeds via a bimolecular pathway, the decom-
position rate will increase more rapidly than the catalytic rate
at elevated [Pd] (i.e., Vcat [Pd]; Vdecomp [Pd]2). Therefore,
the rate will exhibit a nonlinear dependence on the [Pd]. Such
a trend is observed in the oxidation of the primary alcohol 1
(Figure 1A15 and inset of Figure 1B).
The origin of the differences between the catalyst decomposi-
tion pathways in reactions involving primary and secondary
alcohols cannot be discerned from the present study. Neverthe-
less, it is noteworthy that the catalyst decomposition pathway
is substrate dependent. This observation may be related to the
observation by Hiemstra and co-workers that colloidal palladium
formed in Pd(OAc)2/DMSO-catalyzed oxidative heterocycliza-
tion of alkenes (eq 11) retains catalytic activity.3e In contrast,
Figure 7. Reaction timecourses for the Pd(OAc)2/DMSO-catalyzed oxida-
tion of 2 at three different oxygen pressures. Conditions: 2.5 mM Pd(OAc)2,
542-992 Torr O2, 0.52 M 1-phenylethanol, 2 mL of DMSO, 80 °C, in
25.4 mm OD tube.
Oxygen Pressure Effects on Catalyst Decomposition. Mass-
transfer effects are especially significant for palladium-catalyzed
aerobic oxidation reactions because of the susceptibility of the
reduced catalyst to decompose into inactive palladium metal.
Thus, slow transfer of oxygen gas into solution not only affects
the reaction rate but also contributes to catalyst decomposition.
Catalyst decomposition also can be problematic under conditions
that are not mass-transfer limited. At low [Pd(OAc)2], the rate
of alcohol oxidation exhibits virtually no dependence on the
oxygen pressure (Figure 3A and Supporting Information Figure
S1A). In Figure 7, the full-reaction timecourse is displayed for
the Pd(OAc)2/DMSO-catalyzed oxidation of the secondary
alcohol 2 at three different oxygen gas pressures. Although the
initial rates of the reactions are essentially identical (Figure 7
and Supporting Information Table S1), higher turnover numbers
are evident at higher oxygen pressures (58, 72, and 84 turnovers
are observed at initial oxygen pressures of 542, 728, and 992
Torr, respectively). This trend is readily rationalized by the
mechanistic model in Scheme 1 whereby oxidation of the
reduced catalyst competes directly with catalyst decomposition
(step ii vs iii, Scheme 1). Although the present study has a
mechanistic focus and, therefore, is not focused on optimization
of the reaction, Sheldon et al. have reported high yields for
palladium-catalyzed aerobic alcohol oxidation conducted in
DMSO and H2O/DMSO mixtures by employing 30 bar air
pressure.7f
the aggregated palladium formed during alcohol oxidation is
inactive. The enhanced catalyst stability in eq 11 appears to be
a substrate- (or product-) induced phenomenon.
Catalyst Decomposition Pathways: Analysis of the Cata-
lytic Reaction Timecourse Plots. Insights into the catalyst
decomposition pathways were corroborated by evaluating the
full timecourse of the reactions at different Pd(OAc)2 concentra-
tions. In the absence of mass-transfer effects, the rate of catalytic
alcohol oxidation is approximately first order in [alcohol] at
low to moderate concentrations (e0.5 M) (cf. Figure 3B), and
the rate law associated with this reaction (eqs 12 and 13) predicts
that [alcohol] will exhibit an exponential decay with respect to
time.16 We recently observed such behavior, for example, in
the Pd(OAc)2/pyridine-catalyzed oxidation of 1-phenylethanol.7c
Catalyst Decomposition Pathways. Analysis of the [Pd-
(OAc)2] Dependence Plots. The precise mechanism of catalyst
decomposition is not known; however, the formation of catalyti-
cally inactive palladium black during the reactions suggests
decomposition is associated with catalyst aggregation. Metal
aggregation has been studied extensively in the context of col-
loid formation and nanoparticle synthesis, and both unimolecular
and bimolecular pathways have been implicated in these
processes.13
(13) For extensive references to this literature, see the following: (a) Watzky,
M. A.; Finke, R. G. J. Am. Chem. Soc. 1997, 119, 10382-10400. (b)
Besson, C.; Finney, E. E.; Finke, R. G. J. Am. Chem. Soc. 2005, 127, 8179-
8184.
(14) Such a rate law has been observed for the aerobic oxidation of a well-
defined Pd(0) complex, for example: Stahl, S. S.; Thorman, J. L.; Nelson,
R. C.; Kozee, M. A. J. Am. Chem. Soc. 2001, 123, 7188-7189.
(15) In Figure 1A, only the highest [Pd] approaches the mass-transport-limit
measure for a 25 mL round-bottom flask. The nonlinear [Pd(OAc)2]
dependence is therefore explained by invoking a higher-order (>1)
dependence of the decomposition rate on [Pd].
9
4352 J. AM. CHEM. SOC. VOL. 128, NO. 13, 2006

Pd-Catalyzed Aerobic Alcohol Oxidation Mechanism
A R T I C L E S
[R2CHOH]t ) [R2CHOH]0(1 + kbi[Pd]0‚t)-k /k (21)
i
bi
d[R2CHOH]
) ki[R2CHOH][Pd]
(12)
(13)
dt
The full-reaction timecourses for the oxidation of 1 and 2 at
different [Pd(OAc)2] highlight the detrimental influence of
catalyst decomposition (Figures 8 and 9). None of the timecourse
traces fit the exponential decay curve expected if no catalyst
decomposition occurs, and in many cases, the reactions stop
far short of full conversion.
Analysis of the timecourse traces for the oxidation of the
primary alcohol 1 reveals that unimolecular Pd decomposition
pathway(s) dominate at the lowest [Pd] tested (2.5 mM, Figure
8A); at higher [Pd], the unimolecular decomposition model is
inadequate. At [Pd] ) 5-12.5 mM, the timecourse traces fall
between the simulated curves for unimolecular and bimolecular
decomposition (Figure 8B and Supporting Information Figures
S2-S4). It is reasonable to imagine that both unimolecular and
bimolecular pathways are relevant in this [Pd] range. When [Pd]
g 15 mM, the timecourse fits are consistent with predominant
bimolecular decomposition (Figure 8C,D). These timecourse-
fitting results are consistent with the nonlinear dependence of
the rate on [Pd(OAc)2] observed in the oxidation of 1 (Figure
1; see also the discussion in previous section).
[R2CHOH]t ) [R2CHOH]0 exp(-ki[Pd]‚t)
If the catalyst decomposes during the course of the reaction,
the concentration of active Pd catalyst will decrease and the
rate will decrease proportionately. The effect of catalyst
decomposition on the rate can be modeled by incorporating a
time-dependent catalyst concentration term, [Pd]t, into the
original rate equation (eq 12). The identity of this term depends
on whether decomposition proceeds via unimolecular or bimo-
lecular process(es).17 This treatment is elaborated in eqs 14-
17 for the unimolecular decomposition pathway(s) and in eqs
18-21 for the bimolecular pathway(s). This analysis yields three
separate integrated rate laws (eqs 13, 17, and 21) that can be
used to evaluate the experimental timecourse data.
Rate law for unimolecular catalyst decomposition:
-d[Pd]
) kuni[Pd]
(14)
dt
Integrated rate law for unimolecular catalyst decomposition:
The timecourse traces obtained for the oxidation of the
secondary alcohol 2 differ from those obtained for the oxidation
of 1. Unimolecular catalyst decomposition pathway(s) appear
to dominate until the Pd(OAc)2 concentration is high enough
to induce mass-transfer-limitation effects (Figure 9). At a [Pd-
(OAc)2] of 2.5, 5.0, and 10 mM, the rate equation associated
with unimolecular decomposition (eq 17) fits the data well
(Figure 9A-C). Only at [Pd(OAc)2] ) 25 mM, a concentration
at which mass-transfer-limitation effects are observed (Figure
2), does the bimolecular equation (eq 21) provide the appropriate
fit (Figure 9D).18 The unimolecular decomposition pathway that
predominates in the oxidation of 2 is consistent with the linear
dependence of the rate on [Pd(OAc)2] (inset, Figure 2).
[Pd]t ) [Pd]0 exp(-kuni‚t)
(15)
Incorporation of unimolecular catalyst decomposition term (eq
15) into catalytic rate law (eq 12):
d[R2CHOH]
) ki[R2CHOH][Pd]0 exp(-kuni‚t) (16)
dt
Integrated catalytic rate law incorporating unimolecular catalyst
decomposition:
ki[Pd]
[R2CHOH]t ) [R2CHOH]0 exp
0(exp[-kuni‚t] - 1)
[
]
Conclusions
kuni
(17)
Implications for the Palladium-Catalyzed Aerobic Oxida-
tion Reactions. Pd(OAc)2 in DMSO is among the most widely
used catalyst systems for palladium-catalyzed aerobic oxidation
of organic molecules. The present study provides a number of
fundamental insights into this class of catalytic reactions,
particularly associated with the influence of mass-transfer-
limitation effects and catalyst decomposition pathways on the
reactions. The insights gained from these studies and the
methods described herein can also be applied to other palladium-
catalyzed aerobic oxidation reactions.
The significant role of mass-transfer effects on Pd(OAc)2/
DMSO-catalyzed reactions undoubtedly reflects the low solubil-
ity of oxygen gas in DMSO. A number of common solvents
are listed in Table 2, and among these examples, only water
has lower oxygen solubility than DMSO. Although numerous
Rate law for bimolecular catalyst decomposition:
-d[Pd]
) kbi[Pd]2
dt
(18)
Integrated rate law for bimolecular catalyst decomposition:
[Pd]0
[Pd]t )
(19)
1 + kbi[Pd]0‚t
Incorporation of bimolecular catalyst decomposition term (eq
19) into catalytic rate law (eq 12):
d[R2CHOH] ki[R2CHOH][Pd]0
)
(20)
dt
1 + kbi[Pd]0‚t
(17) As we stated previously, the precise identity of these steps is not known,
and it is reasonable to expect that multiple pathways might operate in
parallel. Therefore, we treat all unimolecular (and bimolecular) processes
collectively with a single phenomenological rate constant, kuni (kbi).
(18) It might be expected that a different set of rate equations would apply to
reactions that exhibit mass-transfer-limitation effects because the rate is
formally not dependent on the [alcohol] under these conditions. However,
only the initial rate is limited by mass-transfer effects. Analysis of the
timecourse data reveals that the reaction rate quickly falls below the mass-
transfer limit (i.e., within the first few minutes), and >95% of the data is
acquired under conditions that exhibit the typical [alcohol] dependence.
Integrated catalytic rate law incorporating bimolecular catalyst
decomposition:
(16) Kinetic studies of the Pd(OAc)2/pyridine-catalyzed reaction also reveal
saturation dependence on [alcohol] (cf. Figure 3B); however, a first-order
approximation for the [alcohol] dependence led to a very good exponential
fit of the reaction timecourse.
9
J. AM. CHEM. SOC. VOL. 128, NO. 13, 2006 4353

A R T I C L E S
Steinhoff and Stahl
Figure 8. Representative timecourse of the oxidation of 1 with 2.5-25 mM Pd(OAc)2. Initial and final time points were used to fit the timecourses to either
a unimolecular decomposition equation (eq 17) or a bimolecular decomposition equation (eq 21). Only initial time points were used to fit the timecourse to
an equation without a decomposition term (eq 13). Conditions: 2.5-25 mM Pd(OAc)2, 730 Torr O2, 0.52 M 2,5-dimethoxybenzyl alcohol, 2 mL of DMSO,
80 °C, in 25.4 mm OD tube.
Figure 9. Representative timecourse of the oxidation of 2 with 2.5-25 mM Pd(OAc)2. Initial and final time points were used to fit the timecourses to either
a unimolecular decomposition equation (eq 17) or a bimolecular decomposition equation (eq 21). Only initial time points were used to fit the timecourse to
an equation without a decomposition term (eq 13). Conditions: 2.5-25 mM Pd(OAc)2, 730 Torr O2, 0.52 M 1-phenylethanol, 2 mL of DMSO, 80 °C, in
25.4 mm OD tube.
factors are considered in the screening and development of new
palladium catalyst systems for aerobic oxidation reactions, the
role of oxygen solubility in the solvent is seldom explicitly
discussed. Because of the intrinsic instability of palladium(0),
it is necessary to ensure efficient reaction between dioxygen
and the reduced palladium catalyst, even if this reaction is not
the turnover-limiting step of the catalytic mechanism.19 The
presence of a high concentration of dissolved oxygen will play
an important role in this process, and the use of a solvent with
higher oxygen solubility may provide a convenient alternative
to conducting the reaction at elevated gas pressures.20
Analysis of the catalytic timecourses implicates both unimo-
lecular and bimolecular catalyst decomposition pathways. The
bimolecular pathways can be minimized by reducing the catalyst
(19) This observation underlies our current interest in the effects of the catalyst
coordination environment on the rates of palladium(0) oxygenation. See
ref 14 and the following: Konnick, M. M.; Guzei, I. A.; Stahl, S. S. J.
Am. Chem. Soc. 2004, 126, 10212-10213.
9
4354 J. AM. CHEM. SOC. VOL. 128, NO. 13, 2006

Pd-Catalyzed Aerobic Alcohol Oxidation Mechanism
A R T I C L E S
Table 2. Oxygen Gas Solubility in Common Solvents
syringe through a septum. Data were acquired using custom software
written within LabVIEW (National Instruments). Correlations between
oxygen uptake and conversion were made by analysis by gas chroma-
tography with hexadecane as an internal standard.
Data-Fitting Procedure. Initial rates were determined from a linear
fit to gas-uptake traces within the first 5% of substrate conversion. A
typical data set and fit are shown in Supporting Information Figure
S5, and analogous plots were used to obtain the data points shown in
Figures 1-3 and 6. In some cases, pressure fluctuations arising from
sample injections were evident in the first 10-20 s of the reactions; in
such cases, these data points were not included in the fit.
Nonlinear least-squares fits of the timecourse data (Figures 8 and 9,
Supporting Information Figures S2-S4) to various rate equations (eq
13, 17, or 21) were obtained by using the Solver function within
Microsoft Excel. The exponential fits (perhaps better described as a
simulation based on eq 13) correspond to an exponential extrapolation
of the data acquired during the first 3 min of the reaction; the simulated
data should fit the experimental data if no catalyst decomposition occurs.
In the fits that account for catalyst decompostion (based on eq 17 or
21), the timecourse fit was initially optimized using the initial 3 min
and the final 5 min of the reaction to ensure good starting and ending
points. Subsequently, the entire timecourse was used to obtain a final
fit. Rate constants obtained from the best fits to the data in Figures 8
and 9 and Supporting Information Figure S2 are provided in Table S2.
O2 Dissolution in DMSO. For more precise measurements of the
oxygen solubility, we increased the solution volume from 2 to 5 mL,
which also dictated a change in vessel shape to allow better stirring.
Oxygen dissolution measurements were made by heating DMSO to
80 °C under small amounts of nitrogen (around 30 Torr) in a vessel
with known volume. A separate, but much smaller, vessel was
pressurized to a high pressure of oxygen (around 8000 Torr). These
vessels were connected through a solenoid valve, and when it was
opened, the vessels quickly equilibrated to around 800 Torr. After
equilibration, the solenoid valve was closed and stirring of the DMSO
commenced.
O
2 solubility
(mM)a
solvent
ref
perfluorobenzene
n-hexane
acetone
chloroform
ethyl acetate
toluene
21.0
15.7
11.0
9.1b
8.9c
8.7
11c
11c
11b
11a
11c
11c
acetonitrile
8.1
11b
11b
this work
11c
N,N-dimethylformamide
dimethyl sulfoxide
dimethyl sulfoxide
H2O
4.5
3.2d
2.2
1.3
11c
a 25 °C unless otherwise indicated. b 16 °C. c 20 °C. d 80 °C.
concentration; however, the presence of unimolecular decom-
position pathways places intrinsic limits on the catalyst lifetime.
The analytical methods described above provide the foundation
for future studies to investigate factors that contribute to
improved catalyst stability.
Experimental Section
General Considerations. Palladium acetate (DuPont) was recrystal-
lized prior to use from benzene/acetic acid.21 Oxygen gas (BOC), 2,5-
dimethoxybenzyl alcohol (Aldrich), 1-phenylethanol (Aldrich), benzene
(Fisher), hexadecane (Aldrich), and anhydrous dimethyl sulfoxide
(Aldrich) were used without purification. Gas chromatographic analysis
of reactions was conducted with a Shimadzu GC-17A gas chromato-
graph with either a DB-Wax or a RTX-5 column.
Gas-Uptake Kinetics. A typical reaction was conducted as follows.
Pd(OAc)2 (11.2 mg, 50 µmol) was added to a 25 mL round-bottom
flask with a stirbar. The flask was attached to an apparatus with a
calibrated volume and a pressure transducer designed to measure the
gas pressure within the sealed reaction vessel. The apparatus was
evacuated to 200 Torr and filled with oxygen to 800 Torr, and this
cycle was repeated 10 times. The pressure was established at 675 Torr.
DMSO (1.85 mL) was added via syringe through a septum. The flask
was heated to 80 °C. When the pressure stabilized in the apparatus,
2,5-dimethoxybenzyl alcohol (0.150 mL, 1.42 mmol) was added via
Acknowledgment. We thank A. E. King for assistance in
acquiring data for Figure 3A. This work was supported by the
NIH (RO1 GM67173), the Dreyfus Foundation (Teacher-
Scholar Award), and the Sloan Foundation (Research Fellow-
ship).
Supporting Information Available: Additional kinetics plots
and timecourse figures. This material is available free of charge
via the Internet at http://pubs.acs.org.
(20) Safety is also a critical consideration in these reactions. In many cases,
conditions that employ organic solvents with 1 atm of oxygen pressure
fall within the explosive limits. Specific safety assessments generally must
be conducted on a case-by-case basis.
(21) Stephenson, T. A.; Morehouse, S. M.; Powell, A. R.; Heffer, J. P.;
Wilkinson, G. J. Chem. Soc. 1965, 3632-3640.
JA057914B
9
J. AM. CHEM. SOC. VOL. 128, NO. 13, 2006 4355