Characterization of manganese(V)-oxo polyoxometalate intermediates and their properties in oxygen-transfer reactions

Article abstract of DOI:10.1021/ja0638455

A manganese(III)-substituted polyoxometalate, [α₂-P ₂Mn^III(L)W₁₇O₆₁]^7- (P₂W₁₇M^III), was studied as an oxidation catalyst using iodopentafluorobenzene bis(tifluoroacetate) (F ₅Phl(TFAc)₂) as a monooxygen donor. Pink P ₂W₁₇Mn^III turns green upon addition of F ₅Phl(TFAc)₂. The ¹⁹F NMR spectrum of F ₅Phl(TFAc)₂ with excess P₂W₁₇Mn ^III at -50°C showed the formation of an intermediate attributed to P₂W₁₇Mn^III-F₅-Phl(TFAc) ₂ that disappeared upon warming. The ³¹P NMR spectra of P₂W₁₇Mn^III with excess F₅Phl(TFAc) ₂ at -50 and -20°C showed a pair of narrow peaks attributed to a diamagnetic, singlet manganese(V)-oxo species, P₂W ₁₇Mn^V=O. An additional broad peak at -10.6 ppm was attributed to both the P₂W₁₇-Mn^III-F ₅Phl(TFAc)₂ complex and a paramagnetic, triplet manganese(V)-oxo species. The electronic structure and reactivity of P ₂W₁₇Mn^V=O were modeled by DFT calculations using the analogous Keggin compound, [PMn^V=OW₁₁O ₃₉]^4-. Calculations with a pure functional, UBLYP, showed singlet and triplet ground states of similar energy. Further calculations using both the UBLYP and UB3LYP functionals for epoxidation and hydroxylation of propene showed lowest lying triplet transition states for both transformations, while singlet and quintet transition states were of higher energy. The calculations especially after corrections for the solvent effect indicate that [PMn^V=OW₁₁O₃₉]^4- should be highly reactive, even more reactive than analogous Mn^V=O porphyrin species. Kinetic measurements of the reaction of P₂W₁₇Mn ^V=O with 1-octene indicated, however, that P₂W ₁₇Mn^V=O was less reactive than a Mn^V=O porphyrin. The experimental enthalpy of activation confirmed that the energy barrier for epoxidation is low, but the highly negative entropy of activation leads to a high free energy of activation. This result originates in our view from the strong solvation of the highly charged polyoxometalate by the polar solvent used and adventitious water. The higher negative charge of the polyoxometalate in the transition versus ground state leads to electrostriction of the solvent molecules and to a loss of degrees of freedom, resulting in a highly negative entropy of activation and slower reactions.

Full text of DOI:10.1021/ja0638455

Published on Web 11/11/2006
Characterization of Manganese(V)-Oxo Polyoxometalate
Intermediates and Their Properties in Oxygen-Transfer
Reactions
Alex M. Khenkin, Devesh Kumar, Sason Shaik,* and Ronny Neumann*,†
†
‡
,‡
Contribution from the Department of Organic Chemistry, Weizmann Institute of Science,
RehoVot, Israel 76100, and Department of Organic Chemistry and the Lise Meitner-MinerVa
Center for Computational Quantum Chemistry, The Hebrew UniVersity, Jerusalem, Israel 91904
Received June 1, 2006; E-mail: sason@yfaat.ch.huji.ac.il; Ronny.Neumann@weizmann.ac.il
III
61]7- (P
III
Abstract: A manganese(III)-substituted polyoxometalate, [R
2
-P
2
Mn (L)W17
O
2
W
17Mn ), was studied
PhI(TFAc) ) as a monooxygen
. The F NMR spectrum of F
17MnIII at -50 °C showed the formation of an intermediate attributed to P
as an oxidation catalyst using iodopentafluorobenzene bis(tifluoroacetate) (F
5
2
III
19
donor. Pink P
with excess P
PhI(TFAc) that disappeared upon warming. The P NMR spectra of P
2
W
17Mn turns green upon addition of F
5
PhI(TFAc)
2
5
PhI(TFAc)
2
III
2
W
2
W
17Mn -F
5
-
31
III
2
2
W
17Mn with excess F
5
PhI(TFAc)
2
at -50 and -20 °C showed a pair of narrow peaks attributed to a diamagnetic, singlet manganese(V)-
V
oxo species, P
2
W
17Mn dO. An additional broad peak at -10.6 ppm was attributed to both the P
2
2
W
17
-
III
Mn -F
5
PhI(TFAc)
complex and a paramagnetic, triplet manganese(V)-oxo species. The electronic
structure and reactivity of P
compound, [PMn dOW11
V
2
W
17Mn dO were modeled by DFT calculations using the analogous Keggin
V
39]4-. Calculations with a pure functional, UBLYP, showed singlet and triplet
O
ground states of similar energy. Further calculations using both the UBLYP and UB3LYP functionals for
epoxidation and hydroxylation of propene showed lowest lying triplet transition states for both transformations,
while singlet and quintet transition states were of higher energy. The calculations especially after corrections
V
4-
for the solvent effect indicate that [PMn dOW11
O
39
]
should be highly reactive, even more reactive than
V
V
analogous Mn dO porphyrin species. Kinetic measurements of the reaction of P
indicated, however, that P W17Mn dO was less reactive than a Mn dO porphyrin. The experimental enthalpy
2
2
W17Mn dO with 1-octene
V
V
of activation confirmed that the energy barrier for epoxidation is low, but the highly negative entropy of
activation leads to a high free energy of activation. This result originates in our view from the strong solvation
of the highly charged polyoxometalate by the polar solvent used and adventitious water. The higher negative
charge of the polyoxometalate in the transition versus ground state leads to electrostriction of the solvent
molecules and to a loss of degrees of freedom, resulting in a highly negative entropy of activation and
slower reactions.
5
-
III
7-
Introduction
W O ] , and Wells-Dawson, [R -P Mn (Br)W O ] ,
1
1
39
2
2
17 61
polyoxometalates, Figure 1, showed good activity and high
selectivity for the epoxidation of alkenes and some activity for
the hydroxylation of alkanes that compared well to the activity
of the manganese(III) tetrakis(2,6-dichlorophenyl)porphyrin,
especially upon factoring in the higher stability of the poly-
oxometalates.
A subclass of polyoxometalate compounds is one where a
transition metal, often in a lower oxidation state, substitutes for
a tungsten- or molybdenum-oxo group at the polyoxometalate
surface (Figure 1). The substituting metal center is thus
pentacoordinated by the “parent” polyoxometalate. The octa-
hedral coordination sphere is completed by an additional sixth
labile ligand, L (usually L ) H2O). The lability of the sixth
ligand allows the interaction of the substituting transition-metal
atom with an oxidant, leading to active oxidizing species. In
analogy to metallorganic chemistry the “pentadentate” poly-
oxometalate can act as an inorganic ligand. As such, transition-
metal-substituted polyoxometalates were early on referred to
as “inorganic metalloporphyrins”, and comparisons of the
catalytic activity of the two systems were made, using
Manganese-substituted polyoxometalates have also been used
as catalysts with other oxidants such as ozone, hydrogen
peroxide, and nitrous oxide, although the preliminary research
appeared to indicate that the active oxidizing intermediate was
2
not likely a manganese-oxo species. If one takes the poly-
(
1) (a) Hill, C. L.; Brown, R. B. J. Am. Chem. Soc. 1986, 108, 536-537. (b)
Mansuy, D.; Bartoli, J. F.; Battioni, P.; Lyon, D. K.; Finke, R. G. J. Am.
Chem. Soc. 1991, 113, 7222-7229.
(
2) (a) Neumann, R.; Khenkin, A. M. Chem. Commun. 1998, 1967-1968. (b)
Neumann, R.; Gara, M. J. Am. Chem. Soc. 1994, 116, 5509-5510. (c)
Neumann, R.; Gara, M. J. Am. Chem. Soc. 1995, 117, 5066-5074. (d)
Neumann, R.; Juwiler, D. Tetrahedron 1996, 47, 8781-8788. (e) Boesing,
M.; Noeh, A.; Loose, I.; Krebs, B. J. Am. Chem. Soc. 1998, 120, 7252-
7259. (f) Ben-Daniel, R.; Weiner, L.; Neumann, R. J. Am. Chem. Soc. 2002,
124, 8788-8789.
1
iodosobenzene and pentafluoroiodosobenzene as oxidants.
II
Especially, the manganese(II/III)-substituted Keggin, [PMn (H2O)-
†
Weizmann Institute of Science.
The Hebrew University.
‡
10.1021/ja0638455 CCC: $33.50 © 2006 American Chemical Society
J. AM. CHEM. SOC. 2006, 128, 15451-15460
9
15451

A R T I C L E S
Khenkin et al.
Figure 1. Ball and stick models of Keggin, [PMn(L)W11O39]5- (left), and Wells-Dawson, [R2-P2Mn(L)W17O61]7- (right), polyoxometalates.
oxometalate- “inorganic metalloporphyrin” analogy a step
further, one should of course discuss also the active intermediate
species for oxygen transfer. In this context, for iron-based
compounds the reactivity of the iron(IV)-oxo porphyrin ca-
as transient intermediates, whereas the less or nonactive
manganese(V)-oxo species with corrole, corrolazine, and cyclic
tetraamido ligands have been isolated.
Although some manganese(IV)-oxo-containing polyoxo-
3
10
tion radical species has been well elucidated. It has been
metalates have been identified, they are inefficient oxygen
computationally shown that a formally analogous iron(V)-oxo
donors and thus highly unlikely to be the reactive intermediates
in manganese-substituted-polyoxometalate-catalyzed reactions.
To date, manganese(V)-oxo polyoxometalate species, reason-
polyoxometalate will also be highly reactive although this
4
intermediate has apparently not yet been accessed. The active
4
intermediate species in the manganese porphyrin system has
been rather extensively studied over the years, and it has been
found that both manganese(IV)-oxo and manganese(V)-oxo
species are competent oxygen-transfer intermediates although
able candidates for such a reactive intermediate, have been
neither isolated nor identified as transient reactive intermediates
in manganese-substituted-polyoxometalate-catalyzed oxidations.
In this paper we provide our combined experimental and
computational results concerning such manganese(V)-oxo
intermediates and their C-H hydroxylation and CdC epoxi-
dation reactivity patterns.
5
the latter is more reactive. Various other high manganese(V)-
6
7
8
oxo species based on salen, corrole, corrolazine, and other
9
nitrogen-based ligands have been identified. Not surprisingly,
the more active manganese(V)-oxo species with porphyrin,
salen, and triazocyclononane ligands have only been observed
Results and Discussion
III
The experimental study was carried out using [R2-P2Mn -
(
3) (a) Ortiz de Montellano, P. R., Ed.; Cytochrome P450: structure,
7-
III
mechanism and biochemistry, 2nd ed.; Plenum Press: New York, 1995.
(H2O)W17O61] (P2W17Mn ) with the Wells-Dawson structure
as the polyoxoanion catalyst and iodopentafluorobenzene bis-
(tifluoroacetate) (F5PhI(TFAc)2) as the oxygen donor. The
(
b) Ortiz de Montellano, P. R., Ed.; Cytochrome P450: structure,
mechanism and biochemistry, 3rd ed.; Kluwer Academic/Plenum Publish-
ers: New York, 2005. (c) Groves, J. T. Proc. Natl. Acad. Sci. U.S.A. 2003,
1
00, 3569-3574. (d) Sono, M.; Roach, M. P.; Coulter, E. D.; Dawson, J.
III
rationale behind these choices was that P2W17Mn had two
H. Chem. ReV. 1996, 96, 2841-2887. (e) Groves, J. T.; Han, Y. Z. Models
and mechanisms of Cytochrome P450 action. In ref 3a, Chapter 1, p 3. (f)
Harris, D. L. Curr. Opin. Chem. Biol. 2001, 5, 724-735. (g) Shaik, S.;
Kumar, D.; de Visser, S. P.; Ahmet, A.; Thiel. W. Chem. ReV. 2005, 105,
phosphorus atoms with different chemical environments that
31
could be effectively utilized in P NMR studies and that the
2
279-2328.
F5PhI(TFAc)2 is freely soluble contrary to its iodosobenzene
counterparts (PhIO and F5PhIO), simplifying mechanistic studies
(
4) (a) Kumar, D.; Derat, E.; Khenkin, A. M.; Neumann, R.; Shaik, S. J. Am.
Chem. Soc. 2005, 127, 17712-17718. (b) de Visser, S. P.; Kumar, D.;
Neumann, R.; Shaik, S. Angew. Chem., Int. Ed. 2004, 43, 5661-5665.
5) (a) Groves, J. T.; Stern, M. K. J. Am. Chem. Soc. 1984, 109, 3812-3814.
19
and also allowing the use of F NMR as a spectroscopic probe.
Since F5PhI(TFAc)2 had not been specifically used as an oxygen
donor in catalyzed reactions, the first step in the research was
the purview of P2MnW11/F5PhI(TFAc)2 reactions. In Table 1
we present the yields of representative reactions at essentially
complete conversion of F5PhI(TFAc)2.
(
(b) Arasasingham, R. D.; He, G. X.; Bruice, T. C. J. Am. Chem. Soc. 1993,
1
1
2
15, 7985-7991. (c) Groves, J. T.; Lee, J.; Marla, S. S. J. Am. Chem. Soc.
997, 119, 6269-6273. (d) Zhang, R.; Newcomb, M. J. Am. Chem. Soc.
003, 125, 12418-12419. (e) Zhang, R.; Horner, J. H.; Newcomb, M. J.
Am. Chem. Soc. 2005, 127, 6573-6582. (f) Zhang, R.; Harischandra, D.
N.; Newcomb, M. Chem.sEur. J. 2005, 11, 5713-5720.
6) (a) Feichtinger, D.; Plattner, D. A. Angew. Chem., Int. Ed. Engl. 1997, 36,
(
1
1
1
718-1719. (b) Feichtinger, D.; Plattner, D. A. Perkin 2 2000, 1023-
028. (d) Li, Z.; Tang, Z. H.; Hu, X. X.; Xia, C. G. Chem.sEur. J. 2005,
1, 1210-1216. (d) Khavrutskii, I. V.; Musaev, D. G.; Morokuma, K. Inorg.
III
The results showed that catalyses by P2W17Mn /F5PhI-
III
(TFAc)2 were largely similar to those described for P2W17Mn /
Chem. 2005, 44, 306-315.
1b
PhIO in the literature with some differences. The following
(
7) (a) Gross, Z.; Golubkov, G.; Simkhovich, L. Angew. Chem., Int. Ed. 2000,
3
9, 4045-4047. (b) Wang, S. H.; Mandimutsira, B. S.; Todd, R.; Randhanie,
conclusions may be drawn: (i) The oxidation of sulfides was
fairly facile; by the use of thianthrene oxide as a probe substrate,
it appears that the active oxidizing species should not be
described as electrophilic or nucleophilic since the disulfoxide
B.; Fox, J. P.; Goldberg, D. P. J. Am. Chem. Soc. 2004, 126, 18-19.
8) (a) Mandimutsira, B. S.; Randhanie, B.; Todd, R. C.; Wang, H.; Zareba,
A. A.; Czernuszewicz, R. S.; Goldberg, D. P. J. Am. Chem. Soc. 2002,
(
1
24, 15170-15171. (b) Lansky, D. E.; Mandimutsira, B.; Ramdhanie, B.;
Clausen, M.; Penner-Hahn, J.; Zvyagin, S. A.; Telser, J.; Krzystek, J.; Zhan,
R.; Ou, Z.; Kadish, K. M.; Zakharov, L.; Rheingold, A. L.; Goldberg, D.
P. Inorg. Chem. 2005, 44, 4485-4498.
9) (a) Collins, T. J.; Powell, R. D.; Slebodnick, C.; Ufellman, E. S. J. Am.
Chem. Soc. 1990, 112, 899-901. (b) Collins, T. J.; Gordon-Wylie, S. W.
J. Am. Chem. Soc. 1989, 111, 4511-4513. (c) Workman, J. M.; Powell,
R. D.; Procyk, A. D.; Collins, T. J.; Bocian, D. F. Inorg. Chem. 1992, 31,
11
and sulfone were formed in equivalent amounts. (ii) The
(
(10) (a) Zhang, X. Y.; Pope, M. T. J. Mol. Catal. A 1996, 114, 201-208. (b)
Zhang, X.-Y.; Jameson, G. B.; O’Connor, C. J.; Pope, M. T. Polyhedron
1996, 15, 917-922. (c) Zhang, X.-Y.; O’Connor, C. J.; Jameson, G. B.;
Pope, Michael T. Inorg. Chem. 1996, 35, 30-34. (d) Zhang, X. Y.; Pope,
M. T.; Chance, M. R.; Jameson, G. B. Polyhedron 1995, 14, 1381-1392.
1
548-1550. (d) Gilbert, B. C.; Kamp, N. W. J.; Lindsay-Smith, J. R.;
Oakes, J. J. Chem. Soc., Perkin. Trans. 2 1998, 1841-1844.
1
5452 J. AM. CHEM. SOC. VOL. 128, NO. 48, 2006
9

Manganese(V)−Oxo Polyoxometalate Characterization
A R T I C L E S
Table 1. Catalytic Oxidation of Organic Compounds by
An important further observation in oxidations catalyzed by
F5PhI(TFAc)2 Catalyzed by P2W17MnIII a
III
P2W17Mn /F5PhI(TFAc)2 was that upon addition of F5PhI-
yield,
III
(
TFAc)2 to a 2 mM pink solution of P2W17Mn , a green solution
substrate
products
mol %
was obtained with or without a substrate such as cyclooctene,
-octene, or thioanisole in 1:1 dichloromethane/acetonitrile. The
solution remained green until the F5PhI(TFAc)2 oxidant was
consumed, whereupon the pink color returned. The related UV-
vis spectra are shown in Figure 2.
cyclooctene
cyclohexene
cyclooctene oxide
80
56
1
cyclohexene oxide (16%),
cyclohexenol (58%),
cyclohexenone (26%)
1-octene oxide
1
-octene
56
53
19
trans-2-octene
cis-stilbene
trans-2-octene oxide
cis-stilbene oxide (74%),
trans-stilbene oxide (26%)
trans-stilbene oxide
One may observe that the original starting compound has a
-
1
-1
clear peak at λmax ) 476 nm (ꢀ ) 2500 M cm ). Upon
trans-stilbene
16
86
59
addition of 1 equiv of F5PhI(TFAc)2 to the 0.04 mM solution
thioanisole
methyl phenyl sulfoxide
III
of P2W17Mn , the solution immediately turns green with a more
thianthrene oxide thianthrene disulfoxide (51%),
intense spectrum throughout the 400-700 nm region with no
thianthrene dioxide (49%)
cyclohexane
cyclohexanol (43%), cyclohexanone (57%)
18
maximum. After 600 s the emerging reappearance of the original
III
P2W17Mn spectrum may be observed (the spectrum is verti-
a
Reaction conditions: 1 M substrate, 0.064 M F5PhI(TFAc)2, 0.002 M
cally offset for improved visualization). If excess substrate, such
III
P2W17Mn , in 1:1 dichloromethane/acetonitrile under argon, T ) room
III
as 1-octene, is added, then the original spectrum of P2W17Mn
temperature, t ) 30 min. The yield is based on F5PhI(TFAc)2 consumed
(
∼100%). The products noted are the only ones observed resulting from
is observable after 600 s. The described UV-vis experiment
indicated the presence of an observable intermediate and led to
further study on the nature of the active species.
First, using 19F NMR, we sought to understand the reaction
the given substrate. No solvent-related products were observed. Control
III
reactions without P2W17Mn on two reactive substrates, cyclooctene and
cis-stilbene, revealed that comparable yields were obtained after 48 and 96
h, respectively, whereas less reactive substrates such as 1-octene and
cyclohexane showed only traces of product after 1 week.
and fate of the F5PhI(TFAc)2 oxidant in the presence of P2W17-
III
Mn , Figure 3. F5PhI(TFAc)2 (bottom spectrum) shows peaks
epoxidation of alkenes showed yields comparable to those
III
1b
at -119.5, -138.7, and -153.8 ppm, marked “s”, attributable
to the m-, p-, and o-fluorine atoms, respectively. An additional
described for P2W17Mn /PhIO; however, the reaction selectiv-
ity was somewhat different. Substrates such as 1-octene, trans-
peak for the trifluoroacetate moiety (not shown) appears at
2-octene, and cyclooctene with stronger allylic C-H bonds gave
III
-
73.0 ppm. Upon addition of P2W17Mn at -50 °C (middle
epoxides as the only observable product. On the other hand,
spectrum) six additional peaks were observed.
cyclohexene with a weaker allylic C-H bond was preferably
III
hydroxylated rather than epoxidized. P2W17Mn /PhIO exhibited
After allowing the solution warm to -20 °C (top spectrum),
only three peaks are observed at -119.23, -152.19, and
-159.34 ppm, marked “p”, attributable also by measurement
of a reference standard to the m-, p-, and o-fluorine atoms,
respectively, of the pentafluoroiodobenzene (F5PhI) product. An
additional peak for the free trifluoracetic acid (not shown)
appears at -75.7 ppm. Thus, the three additional peaks observed
much more selective epoxidation in the oxidation of cyclo-
1
b
hexene. cis- and trans-stilbene were similarly reactive; no
significant isomerization of cis- to trans-stilbene was observed,
and more importantly, the reaction was reasonably stereoselec-
III
tive. P2W17Mn /PhIO showed differences in the reactivity of
1
b
cis- and trans-stilbene and was less stereoselective.
Figure 2. Spectra of P2W17MnIII before and after addition of F5PhI(TFAc)2. Measurement conditions: the UV-vis spectrum of a 0.04 mM solution of
III
P2W17Mn was measured at 25 °C followed by addition of 1 equiv of F5PhI(TFAc)2 and measurement at 5 s intervals.
J. AM. CHEM. SOC.
9
VOL. 128, NO. 48, 2006 15453

A R T I C L E S
Khenkin et al.
and then quenched to -50 °C showed no evidence of the
intermediate species, indicating that the green compound is not
III
P2W17Mn but rather a suggested Mn(V)dO-substituted poly-
V
oxometalate (P2W17Mn dO), Scheme 1. Additionally, the
solution of the species formed at -50 °C remains green at -20
°
C (and even for some time at room temperature), indicating
III
that the disappearance of the intermediate F5PhIO-P2W17Mn
is not associated with re-formation of the initial Mn(III)
compound.
V
To obtain further evidence for P2W17Mn dO, we carried out
3
1
an additional P NMR experiment, Figure 4, utilizing the fact
that the two phosphorus atoms in P2W17Mn are in different
III
chemical environments. The spectrum of the original Mn(III)-
substituted polyoxometalate (Figure 4, left) shows a broad
singlet at -12.8 ppm attributable to the phosphorus atom distal
to the paramagnetic Mn(III) center. The peak for the phosphorus
atom proximal to Mn(III) was expectedly not observed. After
III
addition of F5PhI(TFAc)2 to P2W17Mn at -50 °C (Figure 4,
middle), a considerably different spectrum is observed; broad
peaks are observable at -10.6 and -13.2 ppm, and narrow
peaks are observed at -8.8 and -11.9 ppm. This spectrum
remains mostly unchanged even upon warming to -20 °C
(Figure 4, right).
Considering first the spectrum measured at -50 °C and
according to our hypothesis presented in Scheme 1, based on
1
9
the UV-vis and F NMR experiments, the solution should
Figure 3. 19F NMR spectra of F5PhI(TFAc)2 (bottom), after addition of
contain the proposed P2W17Mn dO species, the F5PhIO-
V
III
P2W17Mn at -50 °C (middle), and after warming to -20 °C (top).
III
III
P2W17Mn intermediate, and some of the initial P2W17Mn
Measurement conditions: F5PhI(TFAc)2 was dissolved in 1:1 CH2Cl2/CD3-
III
polyoxometalate. Taking into account temperature effects and
CN and cooled to -78 °C, after which 5 equiv of P2W17Mn was added.
The spectra were measured first at -50 °C and then at -20 °C.
paramagnetic effects of other species, the broad peak at -13.2
III
ppm can be reasonably assigned to the initial P2W17Mn
III
polyoxometalate. The F5PhIO-P2W17Mn species is necessarily
paramagnetic (d ) and therefore should also display a broad
4
singlet for the phosphorus atom distal to the manganese center.
It is tempting, therefore, to assign the broad peak at -10.6 ppm
III
V
to F5PhIO-P2W17Mn . On the other hand, the P2W17Mn dO
2
species (d ) can be either a low-spin diamagnetic singlet or a
high-spin paramagnetic triplet (see the computational results
31
below). The P NMR spectrum with two narrow peaks at -8.8
and -11.9 ppm can be clearly associated with the two
environmentally different phosphorus atoms in a low-spin
V
P2W17Mn dO species. In fact, diamagnetic species can be
Figure 4. 31P NMR spectra of P2W17MnIII (left), after addition of F5PhI-
12
uniquely associated with a low-spin Mn(V) compound. On
(
TFAc)2 at -50 °C (middle), and after warming to -20 °C (right).
V
III
the other hand, the high-spin triplet P2W17Mn dO species may
Measurement conditions: P2W17Mn was dissolved in 1:1 CH2Cl2/CD3-
CN and cooled to -78 °C, after which 5 equiv of F5PhI(TFAc)2 was added.
The spectra were measured first at -50 °C and then at -20 °C.
also be associated with the broad peak at -10.6 ppm, perhaps
overlapping with the peak associated with the F5PhIO-P2W17-
III
Mn . In fact, by heating the solution to -20 ˚C (Figure 4, right),
at -50 °C, not associated with either F5PhI(TFAc)2 or F5PhI,
and appearing at -121.7, -145.0, and -156.6 ppm, marked
19
conditions under which the F NMR showed the disappearance
III
of F5PhIO-P2W17Mn , yet which remained green, support is
“
i”, are attributable to an intermediate complex, F5PhIO-P2W17-
lent to the conclusion that indeed the peak at -10.6 ppm is
III
Mn . It should be noted that the peaks of F5PhI(TFAc)2 and
F5PhI appear at slightly different chemical shifts in the various
spectra. This is due to the different measurement temperatures
and the different paramagnetism of each solution. Importantly,
III
probably attributable to both F5PhIO-P2W17Mn and high-
V
spin triplet P2W17Mn dO.
(
11) Adam, W.; Haas, W.; Lohray, B. B. J. Am. Chem. Soc. 1991, 113, 6202-
208.
6
III
already at -50 °C the F5PhIO-P2W17Mn intermediate de-
(
12) As a reviewer noted, a diamagnetic species could conceivably also arise
from an antiferromagnetically coupled Mn(IV)-O-Mn(IV) dimer. Al-
though this is possible, we observed no EPR signal even down to 6 K.
Also, we feel that we can exclude this assignment of the narrow peaks
because the diamagnetic species is not persistent in the presence of oxygen
acceptors (substrates), although one would expect the opposite for a Mn-
(IV)-O-Mn(IV) dimer; it should be quite stable in solution. Furthermore,
one would expect the Mn(IV)-O-Mn(IV) dimer to be a poor oxygen donor
especially since it has already been shown that even monomeric Mn(IV)-O
species are not reactive (ref 10).
composes significantly; this is in contrast to the situation using
an iron(III)-substituted polyoxometalate where such an inter-
mediate remained stable for a lengthy period of time and very
4
a
little free F5PhI was formed even at room temperature.
It is also very notable that the green compound formed by
addition of F5PhI(TFAc)2 to P2W17MnIII at room temperature
15454 J. AM. CHEM. SOC.
9
VOL. 128, NO. 48, 2006

Manganese(V)−Oxo Polyoxometalate Characterization
A R T I C L E S
Scheme 1. Reaction of P2W17MnIII with F5PhI(TFAc)2
The combined experimental results using UV-vis and 19F
and P NMR indicate that the addition of F5PhI(TFAc)2 to
to 68% incorporation of 18O into trans-stilbene oxide, while
addition of H2 O to preformed trans-stilbene oxide, also in the
3
1
18
III
V
III
18
P2W17Mn leads to formation of an active P2W17Mn dO
species responsible for the catalytic activity of the manganese-
substituted polyoxometalate; this active species should have a
diamagnetic ground state with a closely lying triplet state. The
presence of P2W17Mn , did not lead to O incorporation. This
1
8
high O incorporation indicates exchange of the active oxygen
facilitated by its coordination to the manganese center and
supports the reaction pathway presented in Scheme 1.13
To gain insight into the electronic structure and reactivity of
V
competency of the proposed P2W17Mn dO species as an
V
V
oxidant was also tested in stoichiometric reactions. Thus, P2W17-
the proposed P2W17Mn dO species, we further studied a Mn d
O-containing polyoxometalate by means of density functional
theory (DFT) using the similar Keggin analogue as a model
V
Mn dO was prepared by addition of 15 µmol of F5PhI(TFAc)2
in 0.5 mL of 1:1 CH2Cl2/CH3CN to 10 µmol of P2W17MnIII
dissolved in 0.5 mL of 1:1 CH2Cl2/CH3CN at -45 °C. The
solution was heated to -20 °C, yielding the “green compound”,
and 50 mmol of substrate was added. The reaction mixture was
kept at -20 °C until the solution turned pink and then was
analyzed by GC and GC-MS. The results for the oxidation of
V
4-
(Figure 1, left). The relevant [PW11O39Mn dO] species was
generated from the parent [PW12O40] , for which there is
excellent neutron diffraction structural data, by replacing a
single W dO moiety with Mn dO. This was followed by full
and unconstrained optimization of the structure and verification
that it is a minimum by means of frequency calculations. Thus,
3
-
1
4
VI
V
III
several substrates are as follows (yield based on P2W17Mn ):
V
4-
(
i) cyclohexene (yield 48%), cyclohexene oxide (13%), cyclo-
hexenol (87%), (ii) cis-stilbene (yield 23%), cis-stilbene oxide
78%), trans-stibene oxide (22%), (iii) thioanisole (yield 85%),
the resulting optimized structure for the [PW11O39Mn dO]
V
complex exhibits a Mn dO moiety coordinated to four oxygen
atoms (Ob) on the surface of the polyoxometalate and to one
axial oxygen atom (Oi), which is part of the central phosphate
(see Figure 1, left). With this distorted hexacoordination, the
five d-block orbitals spread into δ + 2π* + 2σ* orbitals, where
the asterisk signifies antibonding interaction with the ligands,
Figure 5. Importantly, in the case of π* the antibonding
(
methyl phenyl sulfoxide (100%). The results show that the
V
proposed P2W17Mn dO species is a competent oxygen donor.
Furthermore, the reaction selectivities observed in stoichiometric
reactions for cyclohexene oxidation (epoxidation versus allylic
oxidation), cis-stilbene epoxidation (ratio of cis-epoxide to trans-
epoxide), and thioanisole oxidation (exclusive formation of
sulfoxide) are very similar to the selectivities observed in the
catalytic reactions (Table 1). This also supports the notion that
4
interaction is mostly across the MndO linkage. In addition,
there is a high-lying lone-pair orbital (lp) that lies close to these
4
d orbitals. We have calculated, therefore, a closed shell singlet
V
2
1
1
P2W17Mn dO is the reactive species and that the F5PhIO-
with a δ configuration and a triplet with a δ π* configuration,
III
1
1
P2W17Mn intermediate is not involved in oxygen transfer.
along with its corresponding open-shell δ π* singlet (for other
It is also worth noting that the epoxidation of trans-stilbene
states, see the Supporting Information, SI). The key optimized
18
18
V
4-
(
0.5 M substrate, 0.9 M H2 O (94.3% O labeled), 0.064 M
bond lengths for [PW11O39Mn dO] are summarized in Table
2. The most notable difference is along the Ot-Mn-O axis.
The MndO bond is particularly shorter and the Ot-Mn bond
III
F5PhI(TFAc)2, 0.002 M P2W17Mn in 1:1 dichloromethane/
acetonitrile under argon, T ) room temperature, t ) 30 m) led
Figure 5. Molecular orbitals of [PW11O39Mn dO]4-
V
.
J. AM. CHEM. SOC.
9
VOL. 128, NO. 48, 2006 15455

A R T I C L E S
Khenkin et al.
Table 2. UB3LYP/B1-Optimized Key Bond Lengths of
above, one can conclude that the pure UBLYP functional more
appropriately defines the relative energies of the two lowest
spin states for [PW11O39Mn dO] .
V
4-
2
[
PW11O39Mn dO] in the Singlet (δ Configuration) and Triplet
States
V
4- 17
singlet bond
length, Å
triplet bond
length, Å
singlet bond
length, Å
triplet bond
length, Å
After having probed the geometry and the electronic structure
bond
bond
V
4-
of [PW11O39Mn dO] , we turned to a study of the reaction
pathway for epoxidation and hydroxylation using propene as
the substrate, Scheme 2.
Mn-O
Mn-Ob
Mn-Oi
1.571
1.871
2.350
1.751
1.883
2.138
W-Ob
W-Ot
1.857
1.743
1.862
1.733
To probe the nature of the lowest energy transition state, we
calculated the relative energies of the six transition states for
bond activation for the different spin states for epoxidation and
hydroxylation. The results using the pure and hybrid functionals
are shown in Figure 6. It can be seen that for both functionals
the singlet and also quintet states are in all likelihood not the
reactive states; their transition states are significantly higher in
energy than those of the triplet state. The high energy of the
singlet transition states can be traced to the short MndO bond
in the ground state of [PW11O39Mn dO] , Table 2. Thus,
during the bond activation in propene, the MndO bond has to
stretch and will thereby contribute to the barrier; the shorter
the MndO bond, the more energy is expended in its stretching
and the larger will be the barrier. It follows, therefore, that the
high barrier of the singlet state has a clear chemical rationale
and is also independent of the functional used for the calcula-
tions.
Table 3. Relative Energies (kcal/mol) for the Three Spin States of
V
4-
[
PW11O39Mn dO]
UB3LYP
UBLYP
gas
gas
spin state
phase
ꢀ ) 5.7
ꢀ ) 37.5
phase
ꢀ ) 5.7
ꢀ ) 37.5
δ π*1 11′
1
0.4
0.0
-16.0
0.7
0.0
-14.8
0.8
0.0
-15.0
1.4
0.0
0.1
1.5
0.0
1.1
1.5
0.0
1.1
2
1
1
δ
1
δ π*1 31
V
4-
is thus longer for the closed shell singlet, in which the
antibonding Mn-O orbitals are vacant. The MndO bond gets
longer as more of these antibonding orbitals are populated; note
1
7b
that a similar observation was made by Groves et al. This
short MndO bond in the singlet state results in large barriers
for oxidations through this state (see below) and hence in
reactivity via different spin states. For bond lengths not presented
in Table 2 there were no significant differences between the
singlet and triplet states.
The UBLYP energy profiles for double bond epoxidation and
allylic hydroxylation are shown in Figure 7, and the analogous
UB3LYP energy profiles can be found in the SI. A few trends
are notable: (i) As noted above, the gas-phase barriers on the
triplet state are small. (ii) At the B1 level using UBLYP and at
the higher B2 level with UB3LYP with solvent correction, the
epoxidation process is calculated to be downhill, while the
hydroxylation has a tiny barrier for bond activation. The large
solvent correction effect is due to the high negative charge of
the polyoxometalate that magnifies Born solvation energies since
Previously we have shown that for the corresponding oxoiron
4-
species [PW11O39FedO] , both UB3LYP and the pure func-
4
tional BP86 gave the same spin-state ordering. However, for
manganese complexes,15 there is still no consensus on the best
DFT method for reproducing the correct spin-state ordering.
Therefore, we tested alongside UB3LYP a few pure GGA
functionals: UBP86,1
6a,b
UBLYP,
16b,c
UBPW91,
16a,d
and the
1
6e
modified hybrid functional UB3LYP*. The UB3LYP* results
are similar to those for UB3LYP, while all the pure functionals
gave the same ordering of the spin states, which is different
from that of B3LYP. Therefore, we show in Table 3 only the
results using UB3LYP and UBLYP; the rest of the data can be
found in the SI.
It is apparent that using UB3LYP the triplet δ π* state is
the ground state with significantly higher lying singlet states,
whereas using UBLYP the triplet and singlet states have similar
energies. On the basis of the NMR and other data presented
2
4a
solvation effects are proportional to Q (Q ) charge). Thus,
for the reaction complex Q ) 4-, while at the TS the
polyoxometalate cluster acquires a higher negative charge (Q
≈ 4.3-; see the SI) due to charge transfer from the propene.
2
Since the TS solvation will be proportional to [Q + ∆Q], it
1
1
can be shown that, due to this increment of negative charge,
the solvation of the TS will be magnified (in proportion to
2(∆Q)Q) relative to the ground state, thereby lowering the
reaction barrier significantly. (iii) In both oxidation processes,
Scheme 2. Reaction Pathways Studied for Epoxidation and Hydroxylation of Propene by [PW11O39Mn dO]4- a
V
a
RC ) reactive complex, and TS ) transition state.
15456 J. AM. CHEM. SOC.
9
VOL. 128, NO. 48, 2006

Manganese(V)−Oxo Polyoxometalate Characterization
A R T I C L E S
Figure 6. Relative energies (kcal/mol) of bond activation transition states for epoxidation (1,3,5TS1) and hydroxylation (1,3,5TS3).
Figure 7. UBLYP energy profile for epoxidation (left) and hydroxylation (right). The energies in each line follow the order indicated in the box.
if the reaction starts at the singlet state, it will have to cross
over to the triplet surface; hence, it will show two-state reactivity
The calculations show that, for propene, epoxidation is the
preferred reaction pathway compared to allylic hydroxylation,
TS3 > TS1. This is in concert with the experimental results
18
3
3
(TSR). (iv) At the intermediate stage, the quintet intermediate
may be sufficiently low lying to participate in the product
formation. (v) If the reaction begins at the triplet state, no
crossover will be necessary. This scenario of many spin states
is somewhat reminiscent of the reactivity of iron-oxo com-
plexes developed by Que and co-workers.19
observed for 1-octene, a reasonable model for propene, which
showed only epoxidation and no evidence of allylic hydroxy-
lation. Cyclohexene with weaker allylic C-H bonds (BDE-
2
0a
(cyclohexene) ) 81.6 kcal/mol, BDE(propene) ) 86.4 kcal/
20b
mol ) preferably undergoes allylic hydroxylation; this is also
The computations show that [PW11O39MndO]4- is a very
powerful oxidant, a more powerful oxidant than [PW11O39-
in reasonable agreement with the calculated results that show
3
3
only a difference of ∼3 kcal/mol for TS3 versus TS1 in
4
-
4a
FedO] , which we recently computationally investigated. The
rationale for this can be traced to the relative strengths of the
FedO versus MndO bonds that has also been observed in
metalloporphyrin species.1 As seen in Table 2, the MndO
bond length in the reactive triplet state is 1.751 Å, while the
propene.
Furthermore, the calculations show that both POMsTMdO
reagents (TM ) transition metal) are expected to be more
powerful than compound I of cytochrome P450 or any other
7b
(
14) Brown, G. M.; Noe-Spirlet, M. R.; Busing, W. R.; Levy, H. A. Acta
Crystallogr., B 1977, 34, 1038-1046.
corresponding FedO bond length is 1.667 Å.4 Since the TMd
O bond itself undergoes activation in the transition state, the
stronger bond in [PW11O39FedO]4 will lead to higher barriers
a
(
15) (a) Ghosh, A.; Persson, B. J.; Taylor, P. R. J. Biol. Inorg. Chem. 2003, 8,
507-511. (b) Ghosh, A.; Taylor, P. R. Curr. Opin. Chem. Biol. 2003, 7,
113-124.
-
(
16) (a) Becke, A. D. Phys. ReV. A 1988, 36, 3098-3100. (b) Perdew, J. P.
Phys. ReV. B 1986, 33, 8822-8824. (c) Li, C.; Yang. W.; Parr, R. G. Phys.
ReV. B 1998-I, 37, 785-789. (d) Perdew, J. P.; Wang, Y. Phys. ReV. B
4-
compared with those of [PW11O39MndO] , which has a
significantly weaker bond. The weaker MndO bond must
originate in the higher antibonding character of the π* orbitals
of the MndO moiety. This bond weakening effect in [PW11O39-
1
992, 45, 13245-13249. (e) Reiher, M.; Salomon, O.; Hess, B. A. Theor.
Chem. Acc. 2001, 107, 48-55.
(
17) (a) It has been noted in the past that while hybrid functionals are usually
better than pure functionals, they can show occasionally increased stabiliza-
tion of high spin states; cf. Salomon, O.; Reiher, M.; Hess, B.A. J. Chem.
Phys. 2002, 117, 4729-4737. Harvey, J. N. Struct. Bonding 2004, 112,
MndO]4 versus [PW11O39FedO] is apparent by comparing
-
4-
the exothermicity of the bond activation step for the two
1
51-183. Smith, D. M.; Dupuis, M.; Straasma, T. P. Mol. Phys. 2005,
4
a
reagents. In each case, hydroxylation or epoxidation, the
103, 273-278. (b) The use of a pure functional has also been recently
shown to better fit the experimentally observed spin states for a Mn(V)d
O porphyrin; cf. De Angelis, F.; Jin, N.; Car, R. Groves, J. T. Inorg. Chem.
2006, 45, 4268-4276.
-
[
PW11O39MndO]4 process is more exothermic by ca. 10 kcal/
mol.
(
18) (a) Shaik, S; Cohen, S.; de Visser, S. P.; Sharma, P. K.; Kumar, D.; Kozuch,
S.; Ogliaro, F; Danovich, D. Eur. J. Inorg. Chem. 2004, 35, 207-226. (b)
Shaik, S.; de Visser, S. P.; Ogliaro, F.; Schwarz, H.; Schr o¨ der, D. Curr.
Opin. Chem. Biol. 2002, 6, 556-567. (c) Shaik, S.; Filatov, M.; Schr o¨ der,
D.; Schwarz, H. Chem.sEur. J. 1998, 4, 193. (d) Danovich, D.; Shaik, S.
J. Am. Chem. Soc. 1997, 119, 1773-1786.
(
13) (a) Nam, W.; Valentine, J. S. J. Am. Chem. Soc. 1993, 115, 1772-1778.
(
b) Groves, J. T.; Kruper, W. J. J. Am. Chem. Soc. 1979, 101, 7613-7615.
(
c) Leising, R. A.; Brennan, B. A.; Que, L.; Munck, E. J. Am. Chem. Soc.
1
991, 113, 3988-3990.
J. AM. CHEM. SOC.
9
VOL. 128, NO. 48, 2006 15457

A R T I C L E S
Khenkin et al.
oxoiron species computed by us to date.21 As noted in the past,4a
the reasons for the higher calculated potency of the POMs
TMdO reagents can be ascribed to the large anionic charge
(4-) of the POMsTMdO species. Since at the bond activation
transition states there is some charge transfer from the reagent
to the POMsTMdO reagent, the Coulomb interaction between
the positively charged substrate (Q ) 0.24-0.29; see the SI)
and the high negative charge of the reagent lowers the barriers
already in the gas phase by a considerable amount compared
to those of compound I of cytochrome P450, which is a neutral
reagent. Furthermore, as noted above, the transition-state
solvation energy is amplified relative to that of the ground state,
and the barrier is further reduced after solvent correction. In
4-
the case of [PW11O39FedO] , the barriers remain significant,
but in [PW11O39MndO] , the barriers virtually vanish.
4-
Figure 8. log/log plot of the observed rate of disappearance of P2W17-
V
Mn dO as a function of the initial concentration of 1-octene (0.5 mM
Of course, the foregoing argument is based on energy only
and does not include entropic factors, which may be very
significant for the highly charged polyoxometalate reagent.
Indeed, the relative barriers obtained in the computational
V
III
P2W17Mn dO from 1:1 P2W17Mn /F5PhI(TFAc)2, 5-50 mM 1-octene in
1:1 CH2Cl2/CH3CN at 25 °C).
4
results, showing that [PW11O39MndO] is a stronger oxidant
V
compared to the analogous Mn dO porphyrin, has not been
substantiated by experiment. In fact, prior to this work, two
V
lines of evidence pointed to the opposite: (i) P2W17Mn dO
can be trapped and observed quite easily as described herein,
V
whereas the Mn dO porphyrin cannot normally be observed
since it is too reactive; substitution of the porphine moiety by
V
four cationic pyridyl moieties allowed the observation of Mn d
22
O species. (ii) Kinetic measurements carried out under identical
conditions showed that manganese(III) tetraphenylporphyrin
III
(
Mn TPP) was a considerably more active catalyst compared
III
to P2W17Mn although the actual reactivity of neither the
proposed active P2W17Mn dO nor Mn dOTPP has been
measured. Therefore, using UV-vis spectrometry, we evalu-
ated the kinetic behavior of P2W17Mn dO with 1-octene as a
V
V
Figure 9. Arrhenius plot for the epoxidation of 1-octene (0.5 mM P2W17-
V
III
Mn dO from 1:1 P2W17Mn /F5PhI(TFAc)2, 50 mM 1-octene in 1:1 CH2-
1b
Cl2/CH3CN).
V
activation (measured between 7 and 27 °C) was found to be Ea
substrate. First, the linear dependence of the observed rate
constant as a function of the 1-octene concentration was
demonstrated by following the disappearance of 0.5 mM green
)
4.32 ( 0.07 kcal/mol, Figure 9.
From the plot presented in Figure 9 the computed values for
the enthalpy, entropy, and free energy of activation were
V
P2W17Mn dO (prepared by adding 1 equiv of F5PhI(TFAc)2
q
-1
q
calculated: ∆H 298 ) 3.72 ( 0.07 kcal mol , ∆S 298 ) -52.6
(
III
to P2W17Mn ) at 420 nm with various amounts of 1-octene
-
1
-1
q
-1
0.4 cal mol K , and ∆G 298 ) 19.4 ( 0.1 kcal mol .
(
5-100 mM), Figure 8. The log kobs versus log [1-octene] plot
These measurements indicate that experimentally the proposed
2
yielded a slope of 1.06 (r ) 0.99) showing that the disappear-
ance of green P2W17Mn dO is first order in 1-octene.
V
q
P2W17Mn dO complex indeed has a low barrier (∆H 298 ) 3.72
V
-
1
kcal mol ) for oxygen transfer in an epoxidation reaction that
is in general qualitative agreement with the calculated barrier.
Since there are as yet no reports for such measurements for an
It should be noted that at 1-octene concentrations lower than
V
5
mM (10:1 1-octene/P2W17Mn dO) the rate of self-decay of
V
P2W17Mn dO is too fast relative to that with 1-octene to derive
an exact second-order rate equation. However, in the presence
of 100 equiv of 1-octene, a pseudo-first-order rate equation,
V
analogous Mn dO porphyrin, a direct comparison with P2W17-
Mn dO is not possible at this stage. However, it is clear
considering the rate constants measured herein for P2W17Mn d
V
V
V
V
-
d[P2W17Mn dO]/dt ) k[P2W17Mn dO], gave at 25 °C a
O and the literature reports on the reactivity of oxomanganese-
-
2 -1
pseudo-first-order rate constant, k ) 1.082 ( 0.026 × 10
s
22a
V
(
V) tetrakis(N-methylpyrid-4-yl)porphyrin that P2W17Mn d
with τ1/2 ) 48 s. By comparison the reported second-order rate
constant for oxomanganese(V) tetrakis(N-methylpyrid-4-yl)-
O experimentally is in fact a less effective oxygen-transfer agent.
The key to understanding the relatively sluggish reactivity of
5
porphyrin with an active alkene, carbamazepine, was 6.5 × 10
V
P2W17Mn dO probably is connected to the highly negative
-1
-1 22a
M
s . Under pseudo-first-order conditions, the energy of
entropy of activation that was obtained from the experiment in
q
-1
-1
our work (∆S 298 ) -52.6 cal mol K ) and the resulting
(
19) Kumar, D.; Hirao, H.; Que, L.; Shaik, S. J. Am. Chem. Soc. 2005, 127,
026-8027.
20) (a) Bryant, J. R.; Mayer, J. M. J. Am. Chem. Soc. 2003, 125, 10351-
q
-1
high free energy of activation (∆G 298 ) 19.4 kcal mol ). It
appears reasonable that the higher charge of TS1 due to charge
transfer from the substrate to the polyoxometalate versus the
initial polyoxometalate leads to increased solvation of TS1 by
the polar solvent molecules (acetonitrile) used in the experiment
and adventitious water that is surely present in solution. This
8
(
1
0361. (b) Lide, D. R., Ed. Handbook of Chemistry and Physics; CRC
Press: Boca Raton, FL, 1997; pp 9-64.
21) (a) de Visser, S. P.; Ogliaro, F.; Sharma, P. K.; Shaik, S. J. Am. Chem.
Soc. 2002, 124, 11809-11826. (b) See ref 3g.
(
(22) (a) Groves, J. T.; Lee, J.; Marla, S. S. J. Am. Chem. Soc. 1997, 119, 6269-
6
270. (b) Jin, N.; Groves, J. T. J. Am Chem. Soc. 1999, 121, 2923-2924.
15458 J. AM. CHEM. SOC.
9
VOL. 128, NO. 48, 2006

Manganese(V)−Oxo Polyoxometalate Characterization
A R T I C L E S
causes electrostriction of the solvent and a loss of many degrees
of freedom, thereby leading to a high negative entropy of
activation and an experimentally high free energy of activation.
Although the computational model takes into account solvent
effects, it does so by simulating the dielectric constant without
actually including solvent shells in the model reaction pathway
1:1 CH
2
Cl
2
/CD
3
CN and cooled to -78 °C, after which 5 equiv of P
2 17
W -
III
Mn was added. The spectra were measured first at -50 °C and then
at -20 °C. F NMR spectra (376.47 MHz, C
19
6 6
F external standard)
were taken on a Bruker Avance 400 instrument.
31
P NMR Measurements. The samples for low-temperature mea-
III
surements were prepared as follows: P
CH Cl /CD CN and cooled to -78 °C, after which 5 equiv of F
TFAc) was added. The spectra were measured first at -50 °C and
then at -20 °C. P NMR spectra (161.97 MHz, 85% H
standard) were taken on a Bruker Avance 400 instrument.
2
W
17Mn was dissolved in 1:1
2
2
3
5
PhI-
(
Scheme 2). Ideally and especially when considering highly
(
2
charged species such as polyoxometalates, such solvent shells
should be included in the computational model to attain realistic
results in terms of free energies; this will still require a
considerable increase in the power of computational techniques
for molecules such as large polyoxometalates.
31
3
4
PO external
Rate Measurements. UV-vis spectra were taken at 7-27 °C on
an HP 8453 diode array spectrometer equipped with a Unisoko UP-
1000 cooling system for temperature equilibration. Temperature-
dependent measurements were carried out by mixing 2 mL of a 0.5
Conclusions
III
mM solution of P
2
W
17Mn in 1:1 CH
2
Cl
2
/CH
3
CN with 100 equiv of
Using UV-vis, 19F NMR, and 31P NMR spectroscopy, we
have identified a “green” manganese(V)-oxo intermediate
1-octene that was temperature equilibrated for 10 min in the UV-vis
spectrometer with magnetic stirring at 700 rpm. The reaction was
initiated by addition of 1 equiv of F
of 1:1 CH Cl /CH CN, and the UV-vis spectra were then collected
every 10 s. Each rate measurement was repeated two or three times.
5 2
PhI(TFAc) dissolved in 10 mL
V
active species, P2W17Mn dO, formed by the addition of F5-
2
2
3
III
31
PhI(TFAc)2 to P2W17Mn . The P NMR spectral assignments
V
indicate that the P2W17Mn dO species may have both singlet
V
The rate of disappearance of P
where the absorbance of P W
2
2
W17Mn dO was measured at 420 nm,
and triplet ground states. Calculation of the electronic structure
17MnIII is at a minimum (see Figure 2).
V
of P2W17Mn dO by DFT using the analogous Keggin substi-
The pseudo-first-order rate constants (kobs) were obtained by fitting a
V
4-
-kt
tuted [PMn dOW11O39] confirmed the spectral assignments
when using the UBLYP pure functional, while the UB3LYP
hybrid functional gave an apparently incorrect ordering of the
electronic structures of the ground state. Calculation of energy
profiles for the epoxidation and allylic hydroxylation of propene
plot of A420 nm versus time to the equation A ) A e , where A is the
t
0
t
absorbance at any given time and A ) initial absorbance, both at 420
0
nm. The decrease in absorbance followed the exponential law for at
least 3 half-lives in all cases. Compared to the rate of 100:1 1-octene/
V
V
P
2
W
17Mn dO at 25 °C, in the absence of 1-octene, P
2
W17Mn dO
V
4-
disappeared at a rate that was about 20 times slower.
with [PMn dOW11O39] showed lowest triplet transition states
for both transformations independent of the functional used in
the calculation. The calculations predict very low energy barriers
for epoxidation of propene, which are expected to be even lower
than those for analogous manganese(V)-oxo porphyrins. How-
ever, an experimental reaction, the epoxidation of 1-octene by
Computational Methods. The starting geometry was the neutron
diffraction structure14 of [PW12
O
40
]
3-
into which the MndO moiety
was inserted. The resulting structure was optimized by DFT using the
24
unrestricted hybrid functional UB3LYP, combined with the double-ú
25
effective core potential basis set, LACVP, implemented in JAGUAR
26
5
.5 and referred to as B1. All the geometries were optimized with
V
P2W17Mn dO, measured by UV-vis only partially supports
JAGUAR 5.5 without any geometry constraints. The spin states of
-
this prediction. Thus, while the enthalpy of activation was indeed
low, the very high negative entropy of activation generates a
high free energy of activation. The unusually negative entropy
of activation that was measured is explained by the increased
solvation of the transition state that has a higher polyoxometalate
charge than the ground state. The high negative charge of the
polyoxometalates renders them very sensitive to solvent effects.
[PW11O
39MndO]4 and the transition-state structures were verified by
frequency calculations, using GAUSSIAN03 (at the UB3LYP/LACVP
level),27 which has a more robust and faster frequency calculation
routine. Since frequency calculations for these systems take an
extremely long time, the rest of the frequency calculation had to be
waived. Reaction pathways, of epoxidation and hydroxylation, as
depicted in Scheme 2, were verified by a scan along a given coordinate,
while optimizing freely all other coordinates. To test the predictions
of UB3LYP, we carried out single-point calculations with a variety of
functionals, such as UB3LYP*, and pure functionals,16 such as UBP86,
Experimental and Theoretical Methods
Oxidation Reactions. Reactions were carried out in 5 mL vials
UBLYP, etc; all these tests were done with B1. Since only the pure
III
equipped with a septum and a stirring bar. The catalyst P
2
W
17Mn ,
4-
functionals predicted the correct ground state for [PW11O39MndO] ,
2
3
prepared using a known literature method, and substrates were
dissolved in a CH Cl /CH CN (1:1) mixture, and the solution was
saturated with argon. F PhI(TFAc) dissolved in solvent was added to
initiate the reaction. Typical reaction conditions were 1 M substrate,
we describe in the text mostly the results of UBLYP. Nevertheless, as
we noted in the Results and Discussion, both UB3LYP and UBLYP
show that the singlet state of the reagent is not the reactive state. Indeed,
2
2
3
5
2
4,28
B3LYP is generally useful for polyoxometalate calculations, but for
III
0
.064 M F
5
PhI(TFAc)
2
, 0.002 M P
2
W
17Mn , in 1:1 dichloromethane/
16b
some reason is less good for high-valent manganese complexes.
acetonitrile under argon, T ) room temperature, and t ) 30 min. GLC
analysis was performed on an aliquot withdrawn directly from the
reaction mixture after addition of a reference standard to the mixture
(24) (a) Becke, A. D. J. Chem. Phys. 1992, 96, 2155-2160. (b) Becke, A. D.
J. Chem. Phys. 1992, 97, 9173-9177. (c) Becke, A. D. J. Chem. Phys.
1
993, 98, 5648-5652. (d) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B
988, 37, 785-789.
(usually dodecane). The products were identified using reference
1
standards and or GC-MS measurements. GC and GC-MS measure-
ments were recorded on HP 6890 (FID detector) and HP 5973 (MS
detector) instruments equipped with 5% phenyl methyl silicone, 0.32
mm i.d., 0.25 µm coating, 30 m columns (Restek 5MS) using helium
as the eluant.
19F NMR Measurements. The samples for low-temperature mea-
5 2
surements were prepared as follows: F PhI(TFAc) was dissolved in
(25) LACVP is derived from LANL2DZ; see: Dunning, T. H., Jr.; Hay, P. J.
In Modern Theoretical Chemistry; Schaefer, H. F., III, Ed.; Plenum: New
York, 1976; pp 1-28.
(26) Jaguar 5.5; Schr o¨ dinger, Inc.: Portland, OR, 2003.
(
27) Frisch, M. J.; et al. Gaussian 03, Revision C.02; Gaussian, Inc.: Walling-
ford, CT, 2004.
(
28) See for example: (a) Quinonero, D.; Wang, Y.; Morokuma, K.; Khavrutskii,
L. A.; Botar, B.; Geletti, Y. V.; Hill, C. L.; Musaev, D. G. J. Phys. Chem.
B 2006, 110, 170-173. (b) Musaev, D. G.; Morokuma, K.; Geletii, Y. V.;
Hill, C. L Inorg. Chem. 2004, 43, 7702-7708. (c) For, a recent interplay
of DFT and experiment, see: Anderson, T. M.; Neiwert, W. A.; Kirk, M.
L.; Piccoli, P. M. B.; Schultz, A. J.; Koetzle, T. E.; Musaev, D. G.;
Morokuma, K.; Cao, R.; Hill, C. L. Science 2004, 306, 2074-2077.
(
23) Lyon, D. K.; Miller, W. K.; Novet, T.; Domaille, P. J.; Evitt, E.; Johnson,
D. C.; Finke, R. G. J. Am. Chem. Soc. 1991, 113, 7209-7221.
J. AM. CHEM. SOC.
9
VOL. 128, NO. 48, 2006 15459

A R T I C L E S
Khenkin et al.
The energies of all optimized structures were further corrected using
single-point calculations with the larger basis set, LACV3P+*, B2.
The effect of solvent was systematically estimated using the SCRF
model, based on the Poisson-Boltzmann equation, as implemented in
JAGUAR5.5. This model defines the solvent by two parameters, a
dielectric constant (ꢀ) and a probe radius (r); the program calculates
the cavity of the solute. We used to solvents a, ꢀ ) 5.7, r ) 2.7 Å, and
b, ꢀ ) 37.5 (as for acetonitrile), r ) 2.2 Å.
within the framework of the German-Israeli Project Cooperation
(DIP-G7.1) and the Helen and Martin Kimmel Center for
Molecular Design. R.N. is the Rebbeca and Israel Sieff Professor
of Organic Chemistry.
2
5
Supporting Information Available: Figures (21) and tables
(36) of the structure, energy profiles, charges, and spin densities
calculated for this study. This material is available free of charge
via the Internet at http://pubs.acs.org.
Acknowledgment. This research was supported by the
German Federal Ministry of Education and Research (BMBF)
JA0638455
15460 J. AM. CHEM. SOC.
9
VOL. 128, NO. 48, 2006