Mechanistic and computational study of a palladacycle-catalyzed decomposition of a series of neutral phosphorothioate triesters in methanol

Article abstract of DOI:10.1021/ja107144c

The methanolytic cleavage of a series of O,O-dimethyl O-aryl phosphorothioates (1a-g) catalyzed by a C,N-palladacycle, (2-[N,N- dimethylamino(methyl)phenyl]-C¹,N)(pyridine) palladium(II) triflate (3), at 25 °C and s^spH 11.7 in methanol is reported, along with data for the methanolytic cleavage of 1a-g. The methoxide reaction gives a linear log k₂^-OMe vs s^spK_a (phenol leaving group) Bronsted plot having a gradient of β_lg = -0.47 ± 0.03, suggesting about 34% cleavage of the P-OAr bond in the transition state. On the other hand, the 3-catalyzed cleavage of 1 gives a Bronsted plot with a downward break at s^spK_a (phenol) ～ 13, signifying a change in the rate-limiting step in the catalyzed reaction, with the two wings having β_lg values of 0.0 ± 0.03 and -1.93 ± 0.06. The rate-limiting step for good substrates with low leaving group s^spK_a values is proposed to be substrate/pyridine exchange on the palladacycle, while for substrates with poor leaving groups, the rate-limiting step is a chemical one with extensive cleavage of the P-OAr bond. DFT calculations support this process and also identify two intermediates, namely, one where substrate/pyridine interchange has occurred to give the palladacycle coordinated to substrate through the S - P linkage and to methoxide (6) and another where intramolecular methoxide attack has occurred on the P - S unit to give a five-coordinate phosphorane (7) doubly coordinated to Pd via the S^- and through a bridging methoxide linked to P and Pd. Attempts to identify the existence of the phosphorane by ³¹P NMR in a d₄-methanol solution containing 10 mM each of 3, trimethyl phosphorothioate (a very slow cleaving substrate), and methoxide proved unsuccessful, instead showing that the phosphorothioate was slowly converted to trimethyl phosphate, with the palladacycle decomposing to Pd⁰ and free pyridine. These results provide the first reported example where a palladacycle-promoted solvolysis reaction exhibits a break in the Bronsted plot signifying at least one intermediate, while the DFT calculations provide further insight into a more complex mechanism involving two intermediates.

Full text of DOI:10.1021/ja107144c

Published on Web 10/29/2010
Mechanistic and Computational Study of a
Palladacycle-Catalyzed Decomposition of a Series of Neutral
Phosphorothioate Triesters in Methanol
C. Tony Liu, Christopher I. Maxwell, David R. Edwards, Alexei A. Neverov,
Nicholas J. Mosey,* and R. Stan Brown*
Department of Chemistry, Queen’s UniVersity, Kingston, Ontario, Canada K7L 3N6
Received August 9, 2010; E-mail: nicholas.mosey@chem.queensu.ca; rsbrown@chem.queensu.ca
Abstract: The methanolytic cleavage of a series of O,O-dimethyl O-aryl phosphorothioates (1a-g) catalyzed
by a C,N-palladacycle, (2-[N,N-dimethylamino(methyl)phenyl]-C1,N)(pyridine) palladium(II) triflate (3), at
s
25 °C and pH 11.7 in methanol is reported, along with data for the methanolytic cleavage of 1a-g. The
s
methoxide reaction gives a linear log k2-OMe vs sspKa (phenol leaving group) Brønsted plot having a gradient
of ꢀlg ) -0.47 ( 0.03, suggesting about 34% cleavage of the P-OAr bond in the transition state. On the
s
other hand, the 3-catalyzed cleavage of 1 gives a Brønsted plot with a downward break at pKa (phenol)
s
∼ 13, signifying a change in the rate-limiting step in the catalyzed reaction, with the two wings having ꢀlg
values of 0.0 ( 0.03 and -1.93 ( 0.06. The rate-limiting step for good substrates with low leaving group
sspKa values is proposed to be substrate/pyridine exchange on the palladacycle, while for substrates with
poor leaving groups, the rate-limiting step is a chemical one with extensive cleavage of the P-OAr bond.
DFT calculations support this process and also identify two intermediates, namely, one where substrate/
pyridine interchange has occurred to give the palladacycle coordinated to substrate through the SdP linkage
and to methoxide (6) and another where intramolecular methoxide attack has occurred on the PdS unit to
give a five-coordinate phosphorane (7) doubly coordinated to Pd via the S- and through a bridging methoxide
linked to P and Pd. Attempts to identify the existence of the phosphorane by 31P NMR in a d4-methanol
solution containing 10 mM each of 3, trimethyl phosphorothioate (a very slow cleaving substrate), and
methoxide proved unsuccessful, instead showing that the phosphorothioate was slowly converted to trimethyl
phosphate, with the palladacycle decomposing to Pd0 and free pyridine. These results provide the first
reported example where a palladacycle-promoted solvolysis reaction exhibits a break in the Brønsted plot
signifying at least one intermediate, while the DFT calculations provide further insight into a more complex
mechanism involving two intermediates.
1. Introduction
contrast, catalysts containing softer metal ions (in the Pearson
“Hard/Soft Acid/Base” sense)3 such as Cu(II) can effectively
promote the cleavage of PdS triesters in methanol.4 Pallada-
cycles bearing a formally Pd(I) soft metal ion, such as 3 and
various structural variants, were demonstrated to be efficient
catalysts for the cleavage of phosphorothioate triesters in H2O5,6
and in MeOH.7 The process in water, while effective, suffers
from problems of poor catalyst solubility and dimerization under
basic conditions, as well as product inhibition since the products
are phosphorothioate diesters (RO)2P(dS)OH that bind strongly
to the catalyst in their anionic forms.5,6 Gabba¨ı has provided
important examples whereby the C,N-palladacycle of pheny-
loxazoline and a C,N-2-(2-pyridyl)palladacycle can promote the
hydrolysis of methyl parathion (O,O-dimethyl O-p-nitrophenyl
phosphorothioate, 1, X ) p-NO2) via an intramolecularly
Sulfur-containing organophosphate triesters containing the
PdS group such as fenitrothion (1c) and diazinon (2) are potent
insecticides that inhibit cholinesterases that are essential for
neurotransmission. Due to their relatively lower mammalian
toxicity, these organophosphates have been widely produced
and used,1 but the possible health and ecological threats due to
environmental accumulation necessitate development of effec-
tive methods for their bulk destruction and decontamination
using efficient chemical means.
We previously reported that (La3+)2(-OMe)x (x ) 2-5) and
some Zn(II):(-OCH3) complexes are very efficient at catalyzing
the methanolysis of PdO phosphate triesters,2 but the former
does not decompose PdS triesters in methanol solvent. By
(1) (a) Toy, A. D. F.; Walsh, E. N. In Phosphorus Chemistry in EVeryday
LiVing, 2nd ed.; American Chemical Society: Washington, DC, 1987;
pp 18-20. (b) Quin, L. D. In A Guide to Organophosphorus
Chemistry; Wiley: New York; 2000; pp 367-371.
(2) (a) Tsang, J. S.; Neverov, A. A.; Brown, R. S. J. Am. Chem. Soc.
2003, 125, 7602. (b) Tsang, J. S. W.; Neverov, A. A.; Brown, R. S.
Org. Biomol. Chem. 2004, 2, 3457. (c) Desloges, W.; Neverov, A. A.;
Brown, R. S. Inorg. Chem. 2004, 43, 6752. (d) Liu, T.; Neverov, A. A.;
Tsang, J. S. W.; Brown, R. S. Org. Biomol. Chem. 2005, 3, 1525.
(3) Smith, M. B.; March, J. In AdVanced Organic Chemistry, 5th ed.;
Wiley Interscience: New York, 2001; pp 338 and refs cited therein.
(4) Neverov, A. A.; Brown, R. S. Org. Biomol. Chem. 2004, 2, 2245.
(5) Ryabov, A. D. In Palladacycles; Dupont, J., Pfeffer, M., Wiley-VCH:
Weinheim; 2008; pp307-339 and references within.
(6) (a) Kim, M.; Liu, Q.; Gabba¨ı, F. P. Organometallics 2004, 23, 5560.
(b) Kim, M.; Gabba¨ı, F. P. Dalton Trans. 2004, 3403. (c) Kim, M.;
Picot, A.; Gabba¨ı, F. P. Inorg. Chem. 2006, 45, 5600.
9
10.1021/ja107144c  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 16599–16609 16599

A R T I C L E S
Liu et al.
Scheme 1. Palladacycle (3)-Promoted Methanolysis of
Phosphorothioates 1
2. Experimental Section
2.1. Materials and Methods. Sodium methoxide (0.5 M solution
in methanol, titrated against N/50 Fisher Certified standard aqueous
HCl solution and found to be 0.49 M), Ag(CF3SO3), PdCl2
(g99.9%), 2,2,6,6-tetramethylpiperidine (TMPP) (99+%), dimethyl
chlorothiophosphate (98%), 2-chloro-4-nitrophenol (97%), 2,4,5-
trichlorophenol (99%), 4-nitrophenol (98%), 4-chlorophenol (99+%),
3-nitrophenol (99%), phenol (99%), 4-methoxyphenol (99%), and
1,8-diazabicyclo[5,4,0]undec-7-ene (98%) were purchased from
Aldrich and used without further purification. 1-Methylpiperidine
(99%) and 1-ethylpiperidine (99%) were obtained from Alfa Aesar
and TCI America Laboratory Chemicals, respectively. HClO4 (70%
aqueous solution, titrated to be 12.1 M) and N,N-dimethylbenzy-
lamine (99%) were purchased from Acros Organics and used as
supplied. Anhydrous methanol was acquired from EMD chemicals.
Phosphorothioates 1a-g were prepared previously9 but for this
study were synthesized following a published procedure2d using
dimethyl chlorothiophosphate and the corresponding phenols as the
starting materials. The 1H NMR, 31P NMR, and mass spectra
obtained are consistent with the structures. Complex 3 was prepared
according to a published methodology.7 Caution: All the phospho-
rothioate substrates are acetylcholinesterase inhibitors and should
be handled with great care. All glassware and equipment exposed
to phosphorothioates should be handled with care using gloVes and
fume-hood protocols.
delivered Pd--OH nucleophile within an Pd-S-coordinated
complex.6 For the catalysts studied, a rich chemistry of
palladacycle was noted including product bound dimeric and
monomeric Pd forms. The methanolytic processes are somewhat
simplified because palladacycle 3 is more soluble in alcohol,
and catalyst dimerization does not occur. Moreover, since the
products are neutral triesters, they bind to the metal ion no more
tightly than the substrates and so do not inhibit the catalyst under
conditions far from saturation (Scheme 1). The catalysis is very
1H NMR and 31P NMR spectra were determined at 400.3 and
162.0 MHz. The CH3OH2+ concentrations were determined poten-
tiometrically using a combination glass electrode (Radiometer model
XC100-111-120-161) calibrated with certified standard aqueous
buffers (pH ) 4.00 and 10.00) as described in a previous paper.10
The sspH values in methanol were obtained by subtracting a
correction constant of -2.2410 from the electrode readings, while
the autoprotolysis constant for methanol was taken to be 10-16.77
8
s
effective: at near neutral pH ) 8.75 in MeOH and 25 °C, 1
s
mM 3 accelerates the methanolytic cleavage of fenitrothion (1c)
by 4.9 × 109-fold relative to the background -OCH3-promoted
reaction.7d
M2. The pKa values of different substituted phenols in methanol
s
Herein, we expand upon the initial investigation7a by reporting
a comprehensive structure-reactivity study of the catalytic
cleavage of a homologous series of dimethyl aryl phospho-
s
can be found in a previous report.11
2.2. Kinetics. The kinetic data for the methanolyses of 1 were
obtained using either a stopped-flow reaction analyzer (10 mm light
path) or conventional UV-vis spectrophotometry at 25 °C. For
stopped-flow kinetics, 3 mL stock solutions of catalyst 3 in
anhydrous methanol of 0.04 mM < [3] < 0.4 mM were prepared in
oven-dried vials. The catalyst stock solutions were loaded into one
syringe of the stopped-flow analyzer while the other syringe was
charged with 0.1 mM 1 and 2 mM 2,2,6,6-tetramethylpiperidine
(TMPP) buffer in methanol (sspKa ) 11.867a). Upon mixing, the
final [3] ranged from 0.02 to 0.2 mM, [1] ) 0.05 mM, and [TMPP
rothioate triesters 1a-g promoted by 3 in methanol. The results
s
show that, depending on the pKa of the leaving group phenol,
s
the rate of the palladacycle-promoted reaction is limited by
different steps in the catalytic cycle, with rate-limiting substrate
replacement of pyridine for substrates bearing good leaving
groups and rate-limiting cleavage of the P-OAr bond within a
palladacycle:substrate complex for substrates with poor leaving
groups. This is corroborated by density functional theory (DFT)
calculations that also locate an unusually stable five-coordinate
phosphorane intermediate bound to the palladacycle along the
reaction pathway. All this new information provides valuable
insights on how to design more proficient catalysts for the
solvolytic cleavage of PdS organophosphate esters.
buffer] ) 1 mM. The TMPP buffer was prepared by adding 1/2-
s
equivalent of HClO4 to set the pH at experimentally measured
s
values of 11.7 ( 0.2. At each catalyst concentration, five consecu-
tive kinetic runs were performed, and the average values of the
observed pseudo-first-order rate constants (kobs, obtained by fitting
the Abs vs time profiles to a standard exponential equation) were
plotted against the [catalyst]. For reactions monitored with con-
ventional UV-vis spectrophotometry, the UV-cells were first
charged with 0.05 mM 1 and 1.0 mM TMPP buffer in methanol.
The injection of the catalyst initiates the reaction, and the final [3]
(7) (a) Lu, Z.-L.; Neverov, A. A.; Brown, R. S. Org. Biomol. Chem. 2005,
3, 3379. (b) Yang, X.-S.; Long, D.-L.; Li, H.-M.; Lu, Z.-L. Inorg.
Chem. Commun. 2009, 12, 572. (c) Lu, Z.-L.; Yang, X.-S.; Wang,
R.-Y.; Fun, H.-K.; Chantrapromma, S. Polyhedron. 2009, 28, 2565.
(d) The acceleration from the cleavage of fenitrothion (ref 7a) is
varied from 0.02 to 0.2 mM. The catalyzed reactions were
s
performed in duplicate, and the pH was controlled at 11.7 ( 0.2
s
determined as the ratio of the observed pseudo-first-order rate constant
with TMPP buffer as described above. In all cases, the reaction
progress was followed by the appearance of the phenolic products
of 1a (393 nm; ε ) (1.50 ( 0.03) × 104 M-1 cm-1), 1b (314 nm;
ε ) (2.6 ( 0.1) × 103 M-1 cm-1), 1c (320 nm; ε ) (4.88 ( 0.07)
× 103 M-1 cm-1), 1d (330 nm; ε ) (2.10 ( 0.06) × 103 M-1
s
for cleavage of fenitrothion by a solution of 1 mM 3 at pH 8.75 of
s
3
(kcat ) 35 M-1 s-1 × (1 × 10-3 M 3) ) 3.5 × 10-2 s-1) compared
with the pseudo-first-order rate constant for the methoxide reaction
(7.2 × 10-12 s-1) computed from the second-order rate constant for
the methoxide reaction (7.2 × 10-4 M-1s-1) and the [methoxide )
10-8 M] at pH 8.75.
s
s
(8) (a) For the designation of pH in nonaqueous solvents, we use the forms
recommended by the IUPAC: Compendium of Analytical Nomencla-
ture. DefinitiVe Rules 1997, 3rd ed.: Blackwell: Oxford, U. K., 1998.
Since the autoprotolysis constant of MeOH is 10-16.77, neutral sspH is
8.4. (b) For the description and treatment of “pH” in MeOH, see:
Gibson, G.; Neverov, A. A.; Brown, R. S. Can. J. Chem. 2003, 81,
495.
(9) Kuivalainen, T.; El-Bahraoui, J.; Uggla, R.; Kostiainen, R.; Sundberg,
M. R. J. Am. Chem. Soc. 2000, 122, 8073.
(10) Gibson, G.; Neverov, A. A.; Brown, R. S. Can. J. Chem. 2003, 81,
495.
(11) Neverov, A. A.; Liu, C. T.; Bunn, S. E.; Edwards, D. R.; White, C. J.;
Melnychuk, S. A.; Brown, R. S. J. Am. Chem. Soc. 2008, 130, 16711.
9
16600 J. AM. CHEM. SOC. VOL. 132, NO. 46, 2010

Palladacycle-Catalyzed Decomposition of Triesters
A R T I C L E S
Figure 1. Placement of explicit solvent methanol molecules on nonmetal binding ligands. (a) 1c, (b) 1g, (c) P(S)(OMe)3, (d) pyridine, (e) 4-nitro-3-
methylphenoxide, (f) 4-methoxyphenoxide, (g) 4-nitro-3-methylphenol, and (h) 4-methoxyphenol.
cm-1), 1e (284 nm; ε ) (8.78 ( 0.02) × 103 M-1 cm-1), 1f (273
nm; ε ) (1.76 ( 0.03) × 103 M-1 cm-1), and 1g (292 nm; ε )
(3.28 ( 0.01) × 103 M-1 cm-1). The absorbance vs time traces
exhibit good first-order behavior throughout the full time-course
of the reaction in all cases, including when the [catalyst] is less
than that of the substrate.12 It is important to note that even when
[catalyst] < [1] full release of expected phenolic product was
observed so that turnover occurs and there is no product inhibition.
The -OCH3-promoted reactions were performed by adding varying
amounts of a 1 M stock solution of tetrabutylammonium hydroxide
in MeOH so that the [base] ranged from 0.1 to 0.3 M while the [1]
) 0.1 mM.
2.5. Density Functional Theory Calculations. Geometry op-
timizations and energy calculations of ground and transition states
were performed at the B3LYP13 level of theory using Gaussian
0914 with the application of the IEFPCM15 MeOH solvent model.
In all calculations, the 6-31G(d,p) basis set was used for C and H
atoms, while O, N, P, and S atoms were assigned the 6-31++G(d,p)
basis set. The effective core potential of Hay and Wadt with the
double-ꢁ valence basis set (LanL2DZ) was employed to describe
Pd.16 This combination of basis sets was selected after testing
several lower and higher levels of theory and finding a good balance
of accuracy and speed. Frequency calculations were conducted to
characterize transition states and intermediates and to use as a basis
for determining free energy values at 298 K.
2.3. UV-vis Experiments to Determine the Interaction of
As ligands dissociate from the complex in solution, they are
stabilized by hydrogen bond donors of the solvent. These specific
interactions are not included by the IEFPCM solvation model. To
account for this stabilization, single explicit solvent molecules were
included in some calculations as appropriate, in the form of
hydrogen bonding interactions with ligands not included in the
complex (Figure 1). Due to the disturbed atom balance in the cases
in which the substrate is bidentate-bound, the energy contribution
of extra methanol molecules must be included throughout the
calculations. When not involved with either the complex or free
ligands, these methanol molecules are part of the hydrogen bonded
network of the bulk solvent. When free from H-bonding interactions
with ligands, methanol is represented by a fractional value of the
energy of a PCM solvated methanol hexamer cluster (Figure 2),
mimicking the protic environment of a methanol molecule in
solution. For instance, one methanol molecule released from
Palladacycle
3 with Trimethyl Phosphorothioate under
Buffered Conditions. A UV-vis cell containing 0.05 mM 3 and
1 mM TMPP buffer in 2.5 mL of methanol was prepared, sspH 11.6
( 0.1, 25.0 ( 0.1 °C. To the cell were added aliquots of a 0.5 M
stock solution of trimethyl phosphorothioate in methanol, and the
UV-vis spectrum (350-200 nm) of the mixture was recorded after
each addition. The absorbencies at 255 and 282 nm were corrected
for concentration changes and the intrinsic absorbance of trimethyl
phosphorothioate in methanol (see Figure 10S, Supporting Informa-
tion) prior to being plotted against the [trimethyl phosphorothioate]
as shown in Figure 3 of the Results section.
2.4. NMR Experiments. 1H and 31P NMR spectra were recorded
at 400.3 and 162.0 MHz, respectively. Stock solutions of catalyst
3 (20 mM), trimethyl phosphorothioate (0.20 M), and NaOCD3
(0.10 M) in CD3OD were prepared and used to prepare the NMR
samples, with the final volume being set at 800 µL with CD3OD.
Phosphoric acid (70% phosphoric acid in H2O) was used as the
reference for 31P NMR.
(13) (a) Becke, A. D. Phys. ReV. A 1988, 38, 3098. (b) Lee, C.; Yang, W.;
Parr, R. G. Phys. ReV. B 1988, 37, 785.
(14) Frisch, M. J.; et al. Gaussian 09, Revision A.02; Gaussian, Inc.:
Wallingford, CT, 2009.
(15) (a) Tomasi, J.; Mennuccia, B.; Cance´s, E. THEOCHEM 1999, 464,
211. (b) Tomasi, J.; Mennuccia, B.; Cammi, R. Chem. ReV. 2005,
105, 2999.
(16) (a) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270. (b) Wadt,
W. R.; Hay, P. J. J. Chem. Phys. 1985, 82, 284.
(12) Pseudofirst-order conditions require either that the [catalyst] is at least
10 times larger than substrate or that the concentration of the catalyst
[3] is constant during the reaction. Since the equilibrium binding of 3
and substrate and its methanolytic product is weak and there is no
product inhibition, pseudo-first-order conditions are satisfied since [3]
remains constant.
9
J. AM. CHEM. SOC. VOL. 132, NO. 46, 2010 16601

A R T I C L E S
Liu et al.
Figure 3. Plot of absorbance at 255 nm vs [trimethyl phosphorothioate]
added to a methanol solution of 0.05 mM 3 and 1 mM TMPP buffer at sspH
11.6 ( 0.1 and 25.0 ( 0.1 °C. The line through the data is calculated for
the process given in Scheme 2 for which the equilibrium expression given
in eq 3 gives K ) 7.9 × 10-3; Abs(int) ) 0.13 ( 0.01; and r2 ) 0.9801.
Figure 2. Cyclic hydrogen-bonded methanol hexamer.
H-bonding would be represented as 1/6 of the total energy of the
hexamer. A hexamer was used, as there is an increase in dissociation
energy per mole of methanol as the size of the cluster increases
until it includes six molecules. For clusters of six molecules or
more, the dissociation energy per mole of methanol is constant.17
The frequency calculations provide access to the free energies
evaluated within the ideal gas, rigid rotor, and harmonic oscillator
approximations. Such approximations can lead to relative free
energies that differ significantly from values obtained in experiments
where the systems are solvated. This is particularly true for the
bimolecular reactions considered here. These deviations can be
traced to an overestimation of the entropic contributions arising
from translation and rotational motions. Essentially, the large
suppression of these motions in solvent is not captured in the
calculations. To correct for this behavior, we follow Sakaki and
co-workers18 and calculate the free energies without including any
entropic contributions from translation and rotational motions.
Equations 1 and 2 provide a comparison of the standard relative
free energy (∆G) and the definition employed in this work (∆Gcorr).
The true free energy value would lie between these values, although
there are convincing arguments18 that ∆Gcorr is more accurate when
describing bimolecular processes occurring in solvent.
Scheme 2. Equilibrium between Palladacycle 3(-OCH3) and
Trimethyl Phosphorothioate (TMPT) and the Palladacycle-Bound
Form of 3(-OCH3)(TMPT) Plus Free Pyridinea
a The species labeled as 3(-OCH3)(TMPT) is not meant to imply
structure, only stoichiometry from the data presented here.
Abs )
(-K[TMPT] + K2[TMPT]2 + 4K[TMPT][3(-OCH )])
√
3
ExtPyr
+
2
Abs(init) (3)
In eq 3, ExtPyr refers to the extinction coefficient of pyridine
corresponding to the change in absorbance observed at 255 nm,
and Abs(init) refers to the absorbance in the absence of added
TMPT (see Supporting Information, Figure 10S and footnote 1
on page S6). The so-computed equilibrium constant for the
∆G ) ∆H - T(∆Selec + ∆Strans + ∆Srot + ∆Svib
∆Gcorr ) ∆H - T(∆Selec + ∆Svib
)
(1)
(2)
process given in Scheme 2 is K ) 7.9 × 10-3
.
)
3.2. Methoxide-Promoted Cleavage of 1a-g. The rates of the
2.6. Theoretical 31P NMR Predictions. All molecular geom-
etries were determined using B3LYP and the basis set combination
discussed above. Due to inconsistencies19 in differentiating four-
and five-coordinate phosphorus using B3LYP, 31P NMR chemical
shifts were predicted at the HF/6-311++G(2d,2p) level of theory
using the GIAO formalism.20 The computed chemical shift of
trimethyl phosphorothioate was used as a reference, assigning its
chemical shift at the experimental number of 73.26 ppm.
base-promoted methanolyses of 1a-g are linear between 0.1 <
-OMe
[methoxide] < 0.3 M. The second-order rate constants (k2
)
in Table 1 were determined from fitting the kobs vs [methoxide]
plots to a standard linear regression, forced through the origin.
The Brønsted plot given in Figure 4, constructed from the log
11
-OMe
s
k2
vs pKa values for the parent phenols of the leaving
s
group, has a slope, ꢀlg, of -0.47 ( 0.03.
3.3. Palladacycle 3 Catalyzed Methanolyses of 1a-g. The
catalysis exhibited by 3 in methanol is insensitive to the presence
of [triflate anion]total up to 2 mM in methanol. We have
previously found that 3 tends to decompose slowly over time
in basic methanol solutions, forming Pd black or nanoparticles.2d
Thus, the kinetic experiments here were conducted by preparing
a solution of substrate and buffer in methanol, after which the
chemical reaction was initiated by adding an appropriate amount
of the stock solution of 3. Under these conditions, buffer
3. Results
3.1. UV-vis Experiments. Shown in Figure 3 is a plot of
the change in absorbance at 255 nm that accompanies the
addition of trimethyl phosphorothioate to a buffered methanol
solution of 3. The data were treated according to the expression
given in eq 321 formulated on the basis of the presumed
equilibrium given in Scheme 2.
(17) Boyd, S. L.; Boyd, R. J. J. Chem. Theory Comput. 2007, 3, 54.
(18) (a) Sumimoto, M.; Iwane, N.; Takahama, T.; Sakaki, S. J. Am. Chem.
Soc. 2004, 126, 10457. (b) Tamura, H.; Yamasaki, H.; Sato, H.; Sakaki,
S. J. Am. Chem. Soc. 2003, 125, 16114.
(19) Zhang, Y.; Oldfield, E. J. Phys. Chem. B. 2006, 110, 579.
(20) Ditchfield, R.; Hehre, W. J.; Pople, J. A. J. Chem. Phys. 1972, 54,
724.
(21) The equilibrium expression was derived based on the equilibrium
process in Scheme 2 while considering the mass balance on Pd catalyst
using the commercially available MAPLE software, Maple V Release
5, Waterloo Maple, Inc., Waterloo, Ontario, Canada. Abs(init) is the
initial absorbance without any added trimethyl phosphorothioate, and
K is the equilibrium constant described in Scheme 2. The path length
of the UV-cell used for the spectrophotometric titration is 1 cm.
9
16602 J. AM. CHEM. SOC. VOL. 132, NO. 46, 2010

Palladacycle-Catalyzed Decomposition of Triesters
A R T I C L E S
Table 1. Second-Order Rate Constants for the Base-Promoted (k2-OMe) and 3-Catalyzed (k2cat) Methanolysis of Substrates 1a-g in
Methanol, at T ) 25 °Ca
Substrate
sspKa HOAr
k2-OMe (M-1s-1
)
k2cat (M-1s-1
)
k2cat/k2-OMe
1a
1b
1c
1d
1e
1f
9.35
10.73
11.50
12.41
13.59
14.33
14.77
(9.3 ( 0.2) × 10-3
(3.9 ( 0.2) × 10-3
(1.05 ( 0.04) × 10-3
(9.1 ( 0.2) × 10-4
(1.22 ( 0.03) × 10-4
(5.1 ( 0.2) × 10-5
(3.5 ( 0.3) × 10-5
1900 ( 50
1620 ( 40
2020 ( 60
1770 ( 20
565 ( 5
2.0 × 105
4.2 × 105
1.9 × 106
1.9 × 106
4.7 × 106
6.0 × 105
1.3 × 105
30.0 ( 0.5
1g
4.42 ( 0.07
a
11
s
-OMe
The pKa values of the leaving group phenols and the ratios of k2cat/k2
are also included.
s
Figure 6. Plot of kobs vs [3] for the catalyzed methanolysis of 1c (5 ×
10-5 M) in the presence of 1 mM TMPP buffer and 1.0 mM additional
pyridine at sspH 11.6 ( 0.1 and 25.0 ( 0.1 °C in anhydrous methanol. The
data are fit to a standard linear regression forced through the origin giving
k2cat ) (28.8 ( 0.3) M-1 s-1. The ∆Abs (320 nm) values were found to be
0.28 ( 0.01 at the end of the reactions.
Figure 4. Brønsted plot for the methoxide-promoted cleavages of phos-
phorothioates 1a-g in methanol at 25.0 ( 0.1 °C. The data points fit a
standard linear regression of log(k2-OMe) ) (-0.47 ( 0.03)sspKaphenol + (2.5
( 0.4); r2 ) 0.9755.
M at sspH 11.7, which generates a pseudo-first-order rate constant
of approximately 1 × 10-8 s-1 for the methoxide background
reaction. The slight downward curvature observed at higher [3]
is most likely due to a common species rate depression caused
by a small amount of free pyridine, driving a portion of the
reactive catalyst:substrate complex back to the free components.
As expected, when the reaction is carried out in the presence
of excess free pyridine, the downward curvature of the kobs vs
[3] plots disappears at the expense of reduced reaction rate
(Figure 6). In a similar experiment for the cleavage of 1c (5 ×
10-5 M) at a constant [3] of 3 × 10-5 M, increasing the added
pyridine concentration results in a decrease in the rate of the
cleavage of 1c, and the data can be fit to a common species
rate depression equation described in the Supporting Information
(Scheme S1 and Figure 7S). These results are discussed in
section 4.3.
Figure 5. Plot of kobs vs [3] for the catalyzed methanolysis of 1c (5 ×
10-5 M) in the presence of 1 mM TMPP at sspH 11.7 ( 0.1 and 25.0 ( 0.1
°C in anhydrous methanol. The linear portion of the data is fitted to a
standard linear regression forced through the origin giving k2cat ) (2020 (
60) M-1 s-1. The ∆Abs (320 nm) values were found to be 0.28 ( 0.01 at
the end of the reactions.
cat
s
The Brønsted plot of log k2 vs pKa (phenol) for the
s
3-catalyzed cleavages of 1 shows a clear downward break at
sspKa ∼ 13 (Figure 7). The data were treated by nonlinear least-
squares (NLLSQ) fitting to an expression22 for a process where
there is a change in rate-limiting step with two ꢀlg values to
catalysis is not observed since the kobs value for the methanolysis
of 1c (0.05 mM) stayed constant at [3] ) 0.08 mM and 0.3
mM < [TMPP buffer] < 5.0 mM.
The 3-catalyzed cleavage reactions of 1 were buffered with
s
s
obtain ꢀ1(lg) ) 0.00 ( 0.02 for the plateau region and ꢀ2(lg)
-1.93 ( 0.06 for the descending portion.
)
TMPP at pH 11.7 ( 0.2, about one pH unit higher than the
s
s
reported sspKa of 10.8 for the ionization of the Pd-bound methanol
7
to form the Pd-(-OCH3) active form of the catalyst. At pH
On the basis of the various kinetic data determined above,
we present, in Scheme 3, a working mechanism for the reaction
that serves as a model for the DFT calculations that follow.
3.4. DFT Calculations. Calculations were performed for the
reaction pathways in Figure 8, which contains a series of
structures that are consistent with the mechanism shown in
Scheme 3. To explore different domains of the Brønsted plot
shown in Figure 7, the calculations were performed for
substrates 1c and 1g. Figure 8 shows the calculated structures
and relevant structural information for the catalytic cleavage of
s
s
11.7, the plots of the first-order rate constant (kobs) for the
appearance of the phenolic product vs [3] for all substrates were
linear to 0.07 mM, and the gradients of each linear portion (see
Figure 5 for substrate 1c and Supporting Information for
cat
1a,b,d-g) were taken as the second-order rate constants (k2
)
for the catalyzed reactions; these are given in Table 1. The kobs
vs [3] data were forced through the origin since the contribution
from the methoxide reaction is insignificant in comparison to
the catalyzed process. For example, for 1c, the [-OMe] ∼ 10-5
9
J. AM. CHEM. SOC. VOL. 132, NO. 46, 2010 16603

A R T I C L E S
Liu et al.
relative free energy similar to that of the ligand exchange
process. Interestingly, the bidentate five-coordinate phosphorane
intermediates INT2 (7) are found to be stable species.
Shown in Figure 9 are computed structures, according to the
pathway followed in Figure 8, for the methanolysis of 1c and
1g. The departure of OAr occurs through a lengthening of the
P-OAr bond of INT2. This barrier is low for 1c (∆Gcorr ) 5.2
kcal mol-1). However, a barrier of ∆Gq ) 15.2 kcal mol-1
corr
is required to expel the poorer leaving group of 1g. This
substrate also has a larger calculated P-OAr distance in the
transition state (2.87 Å) vs 1.72 Å in INT2. In both cases, as
the P-OAr bond lengthens, the Pd-OMe bond length increases
somewhat in the transition state (for 1g, 2.49 vs 2.26 Å in INT2)
to accommodate the increasingly tetrahedral geometry of the
phosphorothioate center (see Table 4S, Supporting Information).
This weakened Pd-OMe coordination likely contributes to the
eventual dissociation of the trimethyl phosphorothioate product
during the turnover process. Upon examination of the overall
energy profile in Figure 8 for 1c, the rate-determining process
involves TSLE, and subsequent chemical steps are energetically
downhill. With 1g, the largest energy barrier is associated with
the expulsion of the leaving group (TSLG).
Figure 7. Brønsted plot of log(k2cat) vs the sspKa (phenol) for the 3-catalyzed
cleavage of 1a-g (5 × 10-5 M) in methanol at 25 °C and sspH 11.7 ( 0.2
(1 mM TMPP buffer). The line through the data is generated from a NLLSQ
fit to the expression given in ref 22 to provide two ꢀlg values of 0.00 (
0.02 and -1.93 ( 0.06; r2 ) 0.9993.
Scheme 3. Proposed Catalytic Cycle for the Stepwise 3-Catalyzed
Cleavage of 1
3.5. NMR Experiments. The 31P NMR spectra for a sample
containing 10 mM 3 and trimethyl phosphorothioate in the
absence of added base show only the presence of free trimethyl
phosphorothioate with a multiplet for the phosphorus at δ )
73 ppm (Figure 13S Supporting Information). For this sample,
the proton NMR reveals peaks associated with the free palla-
dacycle 3 (spectrum not shown). With 5 mM added base, a new
phosphorus multiplet at ∼2 ppm was also observed as well as
the free phosphorothioate peak at 73 ppm (Figure 19S Sup-
porting Information), the integrated ratio of these being 0.29/
1.0. The new 31P peak at ∼2 ppm can be linked to the new
1
P-(OCH3) group at H δ ) 3.78 ppm through an HMBC
1c. The corrected free energies relative to ground state (∆Gcorr),
as described in the Experimental Section, do not include the
translational and rotational components of the entropy. The
ground state (GS) geometry has the pyridine in the N-trans
position as shown by the X-ray crystal structure of complex
3.7a The ligand exchange process leading from GS to INT1 (6
in Scheme 3) was modeled as an associative axial attack of the
sulfur atom on Pd concurrent with displacement of Pd-bound
pyridine through a distorted trigonal-bipyramidal transition state
(TSLE). The geometry and bond lengths about the Pd center
(see Supporting Information) are similar for both substrates.
Stable five-coordinate Pd intermediates including incoming and
outgoing ligands could not be located, consistent with what
would be expected for a concerted associative23 ligand exchange
on the Pd center. The ∆Gqcorr values for 1c and 1g are 10.6 and
8.4 kcal mol-1, respectively, in passing from GS to INT1 which
lies in a shallow minimum for both 1c and 1g. For both
substrates, intramolecular nucleophilic attack of the Pd-bound
methoxide proceeds through a transition state TSNu with a
experiment. As the concentration of base increases to 10 mM
while keeping the [3] and [trimethyl phosphorothioate] both at
10 mM, the ratio of the new species (31P δ 2.25 ppm) to the
unbound trimethyl phosphorothioate (31P δ 73 ppm) also
increases to 1.39/1.0 (see Figure 10 and Figures 14S-18S,
Supporting Information). In the Figure 10 spectrum taken after
3 h, there is also a small peak (∼4%) at δ 37.63 which can be
shown by spiking to be the Pd-bound form of the dimethyl
phosphorothioate anion (compare Figure 14S and 17S, Sup-
porting Information). Addition of up to 3 equiv of methoxide
caused the phosphorothioate peak at 73 ppm to be completely
transformed into the peak at ∼2 ppm. This peak proved to
belong to trimethyl phosphate as proven by spiking the sample
with authentic material (Figure 16S Supporting Information).
The 31P chemical shifts of trimethyl phosphorothioate and
palladium-bound tetramethyl phosphorane 7 computed using the
approach described in section 2.6 above were 73 and -22 ppm,
respectively, referenced as described in section 2.6.
4. Discussion
(22) The NLLSQ equation defined for two ꢀ values and a ꢀ for equilibrium
binding of the catalyst and substrate has the following form
4.1. Methoxide-Promoted Cleavage of 1a-g. The methoxide-
promoted methanolyses of phosphorothioates 1a-g show a
moderate sensitivity to the nature of the leaving group on the
substrate, giving a Brønsted ꢀlg value of -0.47 ( 0.03. This
can be compared with the linear Brønsted correlations for the
-OH-promoted hydrolyses of diethyl aryl phosphate triesters
(PdO; ꢀlg ) -0.43)24 and diethyl aryl phosphorothioate triesters
(PdS; ꢀlg ) -0.35)25 and the methoxide-promoted methanolysis
k2obs ) Kbk1k2/(k-1 + k2) )
-1pKa
CbC1C210(ꢀ +ꢀ +ꢀ )pK /(C-110ꢀ
+ C210ꢀ pK
)
b
1
2
a
2
a
presented as eq 3 in: Neverov, A. A.; Sunderland, N. E.; Brown, R. S.
Org. Biomol. Chem. 2005, 3, 65.
(23) (a) Kru¨ger, H.; van Eldik, R. Chem. Commun. 1990, 4, 330. (b)
Holmann, H.; Suvachittanont, S.; van Eldik, R. Inorg. Chim. Acta 1990,
177, 151.
9
16604 J. AM. CHEM. SOC. VOL. 132, NO. 46, 2010

Palladacycle-Catalyzed Decomposition of Triesters
A R T I C L E S
Figure 8. DFT computed reaction pathways (including ∆Gcorr and ∆E values) for the 3-catalyzed methanolysis of 1c (red) and 1g (blue). Dotted lines
corresponding to product dissociation steps are not expected to be kinetically relevant and are currently being investigated. The catalytic products include
protonated leaving groups (PrHOAr, not shown) formed subsequent to the P-OAr cleavage: at the experimental sspH of 11.7, 1c exists ∼50% in its deprotonated
form while 1g exists as the phenol. Plot is to scale.
of diethyl aryl phosphate triesters (ꢀlg ) -0.70).2d The lyoxide-
dimethyl aryl phosphorothioate esters such as fenitrothion (1c)27
are considered to be concerted. The ꢀeq value for the equilibrium
catalyzed cleavages of acyclic neutral phosphate triesters26 and
Figure 9. Computed structures for the methanolysis of 1c. Selected bond lengths are shown with the corresponding value for 1g in parentheses.
9
J. AM. CHEM. SOC. VOL. 132, NO. 46, 2010 16605

A R T I C L E S
Liu et al.
Figure 10. Proton decoupled 31P NMR (162.0 MHz) of a solution containing 10 mM palladacycle 3, trimethyl phosphorothioate, and NaOCD3 in CD3OD
3 h after mixing. Integrated ratio of the two peaks at 2.23 ppm and 73.26 ppm is 6:4, while intensity of the peak at 37.63 ppm is ∼4% of the total.
transfer of the diethoxy phosphoryl group between oxyanion
nucleophiles is reported to be -1.87 in water28 and results from
the change in effective charge on the aryloxy oxygen passing
from +0.87 in the starting phosphate to -1 in the phenoxide
product. Under the assumption that the ꢀeq value in methanol
is similar, the water value was used to calculate the Leffler
parameter29 (R ) ꢀlg/ꢀeq ) -0.70/-1.87 ) 37%) that is a
measure of the extent of P-OAr cleavage for the CH3O--
promoted methanolysis of PdO triesters. In a study concurrent
with this one, we have determined that the ꢀeq value for transfer
of the (MeO)2PdS group between oxyanions in water is ∼
-1.40,30 which replaces an earlier value of -0.88.31 Assuming
that the transfer of the reaction from water to methanol does
not change the ꢀeq value,32 the Leffler parameter for cleavage
of phosphorothioates is R ) ꢀlg/ꢀeq ) 34%, meaning that extent
of cleavage of the OAr group from the PdS esters in the
transition state is slightly less than that for PdO esters (37%),
although consideration of the experimental errors makes the two
numbers essentially the same. This finding is complementary
to data obtained from 18O and 15N kinetic isotope effect studies
indicating that leaving group departure during alkaline hydroly-
sis of O,O-dimethyl O-p-nitrophenyl phosphorothioate is some-
what less advanced than is the case for phosphate triesters,
meaning the transition state is tighter for the former, although
the difference between the two types of triesters is not large.33
4.2. Palladacycle 3-Catalyzed Cleavage of 1a-g. There are
several points of note observed for the 3-catalyzed cleavage of
1a-g that are consistent with the proposed minimal mechanism
s
presented in Scheme 3. First, the pH/rate profile for the
s
s
3-catalyzed methanolysis of 1c has an apparent pKa of 10.8,
s
attributable to the generation of the internal Pd-(-OCH3)
nucleophile (complex 4 f 5 in Scheme 3) as proposed earlier.7a
The kobs vs [3] plots for the cleavage of phosphates 1 (as in
Figure 4) exhibit downward curvature at higher concentrations
due to a common species rate depression induced by free
pyridine in solution, driving complex 6 back to 5 via k-1 as in
Scheme 3 (Figure 7S, Supporting Information). These results
are consistent with an associative reaction pathway34 involving
displacement of pyridine from 5 by the phosphorothioate to
generate an active complex suggested to be 6 or a stoichiometric
equivalent (possibly 7). If ligand exchange on Pd were to
proceed via a dissociative mechanism, the kobs should exhibit a
square root dependence on [3], but this is not observed.
Included in Table 1 are all the second-order rate constants
for the methoxide- and palladacycle-catalyzed reactions as well
-OMe
as the k2cat/k2
ratios of these which are between 105 and
106, showing that catalyst 3 is very efficient at decomposing
series 1 having a wide range of leaving group ability.
As far as we know, this is the first study of a palladacycle-
catalyzed solvolysis of a contiguous series of substrates with
(31) A ꢀeq value for the transfer of the (MeO)2PdS group between
oxyanions was reported27 to be-0.88 as determined from a nonsym-
metrical reaction where the incoming nucleophile (phenoxide) has a
higher pKa than any of the substrate electrophiles. This number serves
as a lower limit for the actual absolute value of the ꢀeq.
(24) (a) Khan, S. A.; Kirby, A. J. J. Chem. Soc. (B) 1970, 1172. (b) Rowell,
R.; Gorenstein, D. G. J. Am. Chem. Soc. 1981, 103, 5894.
(25) Hong, S.-B.; Raushel, F. M. Biochemistry 1996, 35, 10904.
(26) Ba-Saif, S. A.; Williams, A. J. Org. Chem. 1988, 50, 2204.
(27) Omakor, J. E.; Onyido, I.; vanLoon, G. W.; Buncel, E. J. Chem. Soc.,
Perkin Trans. 2001, 2, 324.
(32) This assumption is reasonable if one considers that the linear free
s
energy plot is log k2 vs pKaphenol and has the x-axis corrected for the
s
pKa difference of the phenol in water vs methanol. If we were simply
plotting the LFE relationship vs σ--values obtained in water, the the
relationship would not be corrected for the effect of solvent (the r-value
s
obtained on that solvent). Fortunately, in the case of the pKa values
s
(28) (a) Ba-Saif, S. A.; Waring., M. A.; Williams, A. J. Am. Chem. Soc.
1990, 112, 8115. (b) Ba-Saif, S. A.; Waring., M. A.; Williams, A.
J. Am. Chem. Soc. 1991, 112, 1653.
we have used, the x-axis incorporates the solvent effect on the
ionization of phenol in methanol vs water. We have discussed the
applicability of this assumption to the case of phosphotriesters cleaved
by a dinuclear phosphotriesterase where very large ꢀlg values are
found: Liu, T.; Neverov, A. A.; Tsang, J. S. W.; Brown, R. S. Org.
Biomol. Chem. 2005, 3, 1525.
(29) (a) Leffler, J. E.; Grunwald, E. In Rates and Equilibria of Organic
Reactions; Wiley: New York, 1963. (b) Williams, A. Acc. Chem. Res.
1984, 17, 425.
(30) A preliminary concurrent study of ꢀNuc and ꢀlg values for the cleavage
of substrates 1 gives a ꢀeq value for the transfer of the (MeO)2PdS
group between oxyanion nucleophiles in water of -1.40. Harkness,
R.; Edwards, D. E.; Brown, R. S., Unpublished results.
(33) Catrina, I. E.; Hengge, A. C. J. Am. Chem. Soc. 2003, 125, 7546.
(34) (a) Kru¨ger, H.; van Eldik, R. Chem. Commun. 1990, 4, 330. (b)
Holmann, H.; Suvachittanont, S.; van Eldik, R. Inorg. Chim. Acta 1990,
177, 151.
9
16606 J. AM. CHEM. SOC. VOL. 132, NO. 46, 2010

Palladacycle-Catalyzed Decomposition of Triesters
A R T I C L E S
Scheme 4. Proposed Equilibrium Formation of a Stable
Palladacycle-Coordinated Phosphorane (7OMe) from the Reaction
of 5 and Trimethyl Phosphorothioate via the Intermediacy of 6OMe
controlled leaving group abilities. As such, it shows a heretofore
unexpected feature where the Brønsted plot for 3-catalyzed
cleavage of 1 has a downward break (Figure 7). This sort of
break in a linear free-energy plot is consistent with a change in
rate-limiting step for the catalyzed reaction which can be
described by the minimal form of eq 4 that gives rise (with
steady state treatment in 6) to the expression in eq 5. With good
substrates, 1a-d, the process is insensitive to the nature of OAr
(ꢀlg ) 0.00 ( 0.02), while with poor leaving groups (1e-g)
the observed ꢀlg ) -1.93 ( 0.06, suggestive of a rate-limiting
transition state with substantial P-OAr bond scission, is
consistent with the DFT calculation of the TS structure for the
cleavage of Pd-bound 1g. The change in
rate constants for substrates with HOAr pKa < 8 are insensitive
to the leaving group ability, while a large leaving group
dependency (ꢀ ) -2.2 for PdO and ꢀ ) -2.0 for PdS) was
found for substrates containing OAr groups where the phenols
have pKa values >8.
4.4. Density Functional Theory Study. The DFT calculations
provide further validation of, and insights into, the proposed
mechanism. The reaction pathway suggested in Scheme 3 was
computed for a representative substrate from each domain of
the Brønsted plot in Figure 7 (1c and 1g). What becomes
apparent from Figure 8 is that the highest energy barriers along
the reaction pathway for substrates 1c and 1g are those for ligand
exchange (TSLE) and leaving group departure (TSLG), respec-
tively. In other words, the DFT study supports the Brønsted
plot in Figure 7 where the computed data also suggest a change
in the rate-limiting step of the catalyzed reaction as the
substrate’s leaving group becomes poorer.
k1
kcat
5 + 1h6 + Pyr 8 P
(4)
(5)
k-1
kobs ) k1[1]/(1 + (k-1/kcat)[Pyr])
the rate-limiting step can be accommodated by the process in
Scheme 3 where a step involving substrate exchange for Pd-
bound pyridine precedes the chemical step of intramolecular
attack of Pd-(-OCH3) on the Pd-bound substrate to yield a
possible five-coordinate phosphorane intermediate bound to the
palladacycle (7), followed by P-OAr cleavage. The insensitivity
of the Brønsted ꢀlg to the nature of 1a-d with good leaving
groups suggests that the substrate-pyridine interchange (k1) is
rate-limiting for reactions of these substrates, and P-OAr
scission (k3) is rate-limiting for 1e-g. We point out that the
apparent ꢀlg value of -1.93 for the latter substrates seems large
for a simple cleavage of the P-OAr bond and is larger than
the limit imposed by the ꢀeq for simple transfer of the
(MeO)2PdS group between oxyanion nucleophiles.30 While
more work is underway to investigate the origin of the large
ꢀlg, its magnitude is such that considerable departure of the
leaving group must occur in the TS. Large values of ꢀlg of -1.43
and -1.12 are also seen for La3+ and Zn(II):(1,5,9-triazacy-
clododecane)-catalyzed methanolyses of neutral phosphate
esters, suggesting the rate-limiting TSs for the catalytic metha-
nolysis reactions also have considerable P-OAr cleavage.
4.3. Other Catalytic Mechanistic Considerations. The rather
small k2cat values of ∼2 × 103 M-1 s-1 for the methanolyses of
the fast reacting substrates 1a-c (Table 1) might seem slow
for a proposed process that involves ligand exchange on metal
centers, as these can be extremely rapid ((2-20) × 108 M-1
For substrate 1c, the ligand exchange (TSLE) on the Pd center
requires 10.6 kcal mol-1 of ∆Gq and is the step with the
corr
highest barrier (Figure 8). The subsequent steps are all energeti-
cally downhill. Importantly, the DFT calculations located a
bidentate five-coordinate phosphorane intermediate (INT2) as
a stable entity along the reaction coordinate, although the
immediate fate of this in the case of 1c is product formation.
Following the energy pathway for substrate 1g, a similar
bidentate five-coordinated phosphorane intermediate was also
found prior to the rate-limiting P-OAr bond fission with ∆Gq
corr
of 15.2 kcal mol-1. In that case, INT2 is surprisingly stable with
9.8 kcal mol-1 required to revert to INT1 and 15.2 kcal mol-1
to expel the leaving group. Experimentally, the break in the
Brønsted plot (Figure 7) signals the existence of at least one
intermediate; however, it would be very difficult to propose the
presence of two intermediates (INT1 and INT2) without
computational data.
s-1 35 in numerous cases involving transition metals. However,
)
What became apparent during this study is the importance
of correctly modeling the solvating medium in studying these
chemical reactions in solution. As discussed in the Experimental
Section, the rotational and translational contributions to entropy
are excluded18 to determine the corrected free energies of
activation (∆Gqcorr). It is reasonable to assume that the molecules
are solvated by solvent through weak interactions, restricting
some, but definitely not all, of their freedom to translocate or
rotate freely. The actual free energies most likely lie between
those calculated from eq 1, which includes all components of
entropy and the entropically corrected ones from eq 2. Sakaki
et al. have argued that the values determined from eq 2 are closer
to the true values.18 The data presented here are consistent with
ligand exchange on a Pd(II) center can be much slower.34,36
The overall process that emerges for the complex-catalyzed
cleavage of these phosphorothioates is reminiscent of the
enzymatic hydrolysis of phosphate triesters and phosphorothioate
triesters by a metallo-phosphotriesterase extracted from the
bacterium Pseudomonas diminuta.25,37,38 There, similar biphasic
Brønsted plots were obtained where the enzymatic hydrolysis
(35) Tobe, M. L.; Burgess, J. In Inorganic Reaction Mechanisms; Addison
Wesley; Longman Ltd.: New York, 1999; pp 271-333.
(36) (a) Rau, T.; Shoukry, M.; van Eldik, R. Inorg. Chem. 1997, 36, 1454.
(b) Shoukry, A.; Rau, T.; Shoukry, M.; van Eldik, R. J. Chem. Soc.,
Dalton Trans. 1998, 3105. (c) Salas, G.; Casares, J. A.; Espinet, P.
Dalton Trans. 2009, 8413. (d) Shi, T.; Elding, L. I. Inorg. Chem. 1997,
36, 528.
that view, and within experimental error, the ∆Gq values
corr
determined for the highest barriers from the DFT calculations
(37) Donarski, W. J.; Dumas, D. P.; Heitmeyer, D. H.; Lewis, V. E.;
Raushel, F. M. Biochemistry 1989, 28, 4650.
39
(1c, 10.6 kcal mol-1; 1g, 14.5 kcal mol-1
)
are surprisingly
(38) (a) Caldwell, S. R.; Newcomb, J. R.; Schlecht, K. A.; Raushel, F. M.
Biochemistry 1991, 30, 7438. (b) Caldwell, S. R.; Raushel, F. M.;
Weiss, P. M.; Cleland, W. W. Biochemistry 1991, 30, 7444.
(39) Ideal Gas at 1 atm and 298 K.
9
J. AM. CHEM. SOC. VOL. 132, NO. 46, 2010 16607

A R T I C L E S
Liu et al.
Scheme 5. Minimal Process Describing the Transformations Occurring during Acquisition of the 31P Spectrum of the Reaction Products
Formed from a CD3OD Solution Containing 10 mM Each of Trimethyl Phosphorothioate, Palladacycle 3, and -OCH3
close to those determined experimentally (1c, 12.9 kcal mol-1
1g, 16.6 kcal mol-1).40
;
cycle with the formation of what appears to be Pd0 nanoparticles
as evidenced by a golden brown colored reaction mixture. We
do not know the fate of the S once cleaved from trimethyl
phosphorothioate, but its loss and the formation of Pd0 suggest
these are involved in redox processes. A minimal process
depicting the overall transformation is given in Scheme 5.
It is an apparently surprising consequence that the DFT
calculation for the catalyzed cleavage of 1c (Figure 8) places
the Pd-coordinated phosphorane (7 in Scheme 3) about 8 kcal
mol-1 lower in energy than its immediate reactant (6). It is not
immediately clear from the calculation why this is so, but
chemical intuition suggests that the stabilization arises from an
overall reduction of the magnitude of charges on the S, Pd, and
Pd-coordinated methoxy group, transfer of negative charge from
the methoxy group to the Pd via increased interaction with the
negatively charged thiolate in 7, and the formation of stronger
bonds at the expense of breaking weaker ones. Thus, we
speculate that the “hard” methoxy group has a large amount of
(-)-charge in 6 which is lessened by the formation of a stronger
P-OCH3 bond in 7 concurrent with the breaking of the PdS
bond with transfer of the negative charge to the Pd by way of
a prominent “soft-soft” interaction with the S-. Inspection of
the electrostatic potential-derived charges41 of the Pd, OCH3,
P, and S in the Gaussian optimized INT1 and INT2 structures
from Figure 8 (see Supporting Information) is consistent with
this view.
4.5. Evidence for the Existence of a Palladacycle-Bound
Phosphorane 7. An interesting finding from the computational
data is the predicted existence of the stable phosphorane 7 along
the reaction pathway. This invited an experimental study treating
a relatively nonreactive substrate like trimethyl phosphorothioate
(TMPT, the initial product of cleavage of 1) with the pallada-
cycle where the solvolytic breakdown of a putative Pd-
coordinated tetramethyl-phosphorane similar to 7 (7OMe in
Scheme 4) would be extremely slow based on the Brønsted
correlation and the acidity of methanol in methanol (sspKa )
18.2).42 This may allow one to visualize the putative intermedi-
ate by NMR techniques.
5. Conclusions
The results presented here reveal a somewhat different picture
for the 3-catalyzed cleavage of these PdS pesticide models than
had been suggested from our previous studies of the single
substrate 1c in methanol7 and one which might also be operative
for the various palladacycle-promoted hydrolyses of these in
water.5,6 This is the first example of a Pd-complex-catalyzed
solvolysis of a contiguous series of substrates where a prominent
downward break in the Brønsted plot reveals a change in the
rate-limiting step of the reaction from substrate binding to
breakdown of an intermediate. The latter is suggested to be
phosphorane intermediate 7 on the basis of DFT computations
which indicates it is formed during the 3-catalyzed P-OAr
cleavage of substrates with both good and poor leaving groups.
The experimental and DFT studies provide a complementary
picture for the 3-catalyzed cleavage of phosphorothioates, and
we emphasize that the main conclusion that the overall reaction
has a change in rate-limiting step involving intermediates 6 and
7 would be difficult to make on experimental or theoretical
grounds alone. We note that the catalytic rate constant provided
to cleavages of 1 by the basic form of 3 are some (1.3-47) ×
105 greater than the methoxide-promoted reactions (in terms of
k2cat/k2-OMe), which points out multiple roles of the metal ion
involving substrate activation, perhaps as a Lewis acid, and
intramolecular delivery of the coordinated methoxide nucleophile.
The fact that the NMR study did not allow us to provide
evidence for the putative palladium-bound phosphorane inter-
mediate is disappointing but does not rule out its existence as
an intermediate during the catalytic cleavage of substrates 1.
The test substrate chosen for investigation, trimethyl phospho-
rothioate, is the product of the very fast cleavage reaction of
series 1 and would react extremely slowly if it followed the
same pathway as the other substrates 1a-g. From the fit to the
At first glance, the -71 ppm upfield chemical shift from δ
) 73 ppm for the unbound TMPT to ∼2 ppm (Figure 10) is
consistent with the reported chemical shift range (-55 to -85
ppm)43-45 for neutral and cyclic pentaoxy-phosphoranes and
tetraoxy-monothiophosphoranes relative to the neutral four-
coordinate phosphorus-containing counterparts. However, the
observed 2 ppm peak is without doubt attributable to trimethyl
phosphate, as proven by spiking the mixture with authentic
material. The exact process by which this is formed is not certain
but clearly involves methoxide in transforming the PdS starting
material to PdO product, as well as decomposing the pallada-
Brønsted plot in Figure 7, it would react about 1 million times
cat
slower than 1g with a predicted k2 of 1.5 × 10-6 M-1 s-1
.
The results of the NMR study indicate that an alternative
reaction occurs whereby methoxide and the palladacycle are
involved in converting the TMPT to trimethyl phosphate, with
the concomitant destruction of the palladacycle to form free
pyridine and Pd0.
(40) Standard State at 1 M and 298 K.
(41) (a) Breneman, C. M.; Wiberg, K. B. J. Comput. Chem. 1990, 11, 361.
(b) Leach, A. R., Molecular Modelling Principles and Applications,
2nd ed.; Pearson Education Limited: Harlow, England, 2001; pp 189-
191. (c) Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A. J.
Phys. Chem. 1993, 97, 10269.
The results of this study also suggest that more proficient
catalysts might be prepared through structural tuning of complex
3 by replacing the pyridine with a weaker bound ligand to lower
the energetic barrier for substrate-ligand exchange on the Pd.
This seems to be the only practical method to increase the
(42) Based on the autoprotolysis constant for methanol (10-16.77 M2) and
[methanol] in pure solvent.
9
16608 J. AM. CHEM. SOC. VOL. 132, NO. 46, 2010

Palladacycle-Catalyzed Decomposition of Triesters
A R T I C L E S
Acknowledgment. The authors gratefully acknowledge the
financial support of NSERC, Canada, and the U.S. Defense Threat
Reduction Agency-Joint Science and Technology Office, Basic and
Supporting Sciences Division, through the award of grant #
HDTRA-08-1-0046. In addition, CTL thanks NSERC for a PGS-2
postgraduate scholarship, and CIM thanks the Ontario Graduate
Scholarship. We are also indebted to the High Performance
Computing Virtual Laboratory (HPCVL) for their facilities.
catalytic decomposition of substrates with good leaving groups,
bearing in mind that these are the ones which are commercially
important as pesticides.46 Current work in this lab is focused
on the influence of the C,N-binding ligand and other pyridines
in the prepared Pd complex to develop more efficient catalysts
in a rational manner. Other aspects of the mechanism, such as
the steps leading to catalyst regeneration and the nature of the
large negative ꢀlg seen for substrates with poor leaving groups,
are also under investigation and will be reported in due course.
Supporting Information Available: Descriptions of experi-
ments and results obtained for various kinetics and for identify-
ing intermediates, 31P NMR spectra for attempts to identify
phosphorane, DFT structures and tables of data, CHelpG charges
on various atoms in INT1 and INT2 structures for 1a and 1g,
and optimized structure coordinates. Full authorship for ref 14
can be found on page S12. This material is available free of
charge via the Internet at http://pubs.acs.org.
(43) (a) Quin, L. D. In A Guide to Organophosphorus Chemistry; Wiley:
New York, 2000; pp 341-342. (b) Timosheva, N. V.; Chandrasekaran,
A.; Holmes, R. R. Inorg. Chem. 2006, 45, 10836. (c) Holmes, R. R.
Acc. Chem. Res. 2004, 37, 746. (d) Zhang, Y.; Oldfield, E. J. Phys.
Chem. B 2006, 110, 579.
(44) Sherlock, D. J.; Chandrasekaran, A.; Day, R. O.; Holmes, R. R. Inorg.
Chem. 1997, 36, 5082.
(45) Garrossian, M.; Bentrude, W. G.; Ro¨schenthaler, G.-V. J. Org. Chem.
2001, 66, 6181.
(46) Main, R. A.; Iverson., F. Biochem. J. 1966, 100, 525.
JA107144C
9
J. AM. CHEM. SOC. VOL. 132, NO. 46, 2010 16609