Catalytic gas phase oxidation of methanol to formaldehyde

Article abstract of DOI:10.1021/ja028839x

Two gas-phase catalytic cycles for the two-electron oxidation of primary and secondary alcohols were detected by multistage mass spectrometry experiments. A binuclear dimolybdate center [Mo₂O₆(OCHR₂)]^- acts as the catalyst in both these cycles. The first cycle proceeds via three steps: (1) reaction of [Mo₂O₆(OH)]^- with alcohol R₂HCOH and elimination of water to form [Mo₂O₆(OCHR₂)]^-; (2) oxidation of the alkoxo ligand and its elimination as aldehyde or ketone in the rate-determining step; and (3) regeneration of the catalyst via oxidation by nitromethane. Step 2 does not occur at room temperature and requires the use of collisional activation to proceed. The second cycle is similar but differs in the order of reaction with alcohol and nitromethane. The nature of each of these reactions was probed by kinetic measurements and by variation of the substrate alcohols (structure and isotope labeling). The role of the binuclear molybdenum center was assessed by examination of the relative reactivities of the mononuclear [MO₃(OH)]^- and binuclear [M₂O₆(OH)]^- ions (M = Cr, Mo, W). The molybdenum and tungsten binuclear centers [M₂O₆(OH)]^- (M = Mo, W) were reactive toward alcohol but the chromium center [Cr₂O₆(OH)]^- was not. This is consistent with the expected order of basicity of the hydroxo ligand in these species. The chromium and molybdenum centers [M₂O₆(OCHR₂)]^- (M = Cr, Mo) oxidized the alkoxo ligand to aldehyde, while the tungsten center [W₂O₆(OCHR₂)]^- did not, instead preferring the non-redox elimination of alkene. This is consistent with the expected order of oxidizing power of the anions. Each of the mononuclear anions [MO₃(OH)]^- (M = Cr, Mo, W) was inert to reaction with methanol, highlighting the importance of the second MoO₃ unit in these catalytic cycles. Only the dimolybdate center has the mix of properties that allow it to participate in each of the three steps of the two catalytic cycles. The three reactions of these cycles are equivalent to the three essential steps proposed to occur in the industrial oxidation of gaseous methanol to formaldehyde at 300-400°C over solid-state catalysts based upon molybdenum(VI)-trioxide. The new gas-phase catalytic data is compared with those for the heterogeneous process.

Full text of DOI:10.1021/ja028839x

Published on Web 02/25/2003
Catalytic Gas Phase Oxidation of Methanol to Formaldehyde
Tom Waters, Richard A. J. O’Hair,* and Anthony G. Wedd
Contribution from the School of Chemistry, The UniVersity of Melbourne,
ParkVille, Victoria, Australia, 3010
Received October 7, 2002; E-mail: rohair@unimelb.edu.au
Abstract: Two gas-phase catalytic cycles for the two-electron oxidation of primary and secondary alcohols
-
2 6 2
were detected by multistage mass spectrometry experiments. A binuclear dimolybdate center [Mo O (OCHR )]
acts as the catalyst in both these cycles. The first cycle proceeds via three steps: (1) reaction of
-
-
[
2 6 2 2 6 2
Mo O (OH)] with alcohol R HCOH and elimination of water to form [Mo O (OCHR )] ; (2) oxidation of
the alkoxo ligand and its elimination as aldehyde or ketone in the rate-determining step; and (3) regeneration
of the catalyst via oxidation by nitromethane. Step 2 does not occur at room temperature and requires the
use of collisional activation to proceed. The second cycle is similar but differs in the order of reaction with
alcohol and nitromethane. The nature of each of these reactions was probed by kinetic measurements
and by variation of the substrate alcohols (structure and isotope labeling). The role of the binuclear
-
3
molybdenum center was assessed by examination of the relative reactivities of the mononuclear [MO (OH)]
-
and binuclear [M
2
O
6
(OH)] ions (M ) Cr, Mo, W). The molybdenum and tungsten binuclear centers
-
-
[
2
M O
6
(OH)] (M ) Mo, W) were reactive toward alcohol but the chromium center [Cr
2
O
6
(OH)] was not.
This is consistent with the expected order of basicity of the hydroxo ligand in these species. The chromium
and molybdenum centers [M (OCHR
)]- (M ) Cr, Mo) oxidized the alkoxo ligand to aldehyde, while the
tungsten center [W (OCHR
)]- did not, instead preferring the non-redox elimination of alkene. This is
2
O
6
2
2
O
6
2
consistent with the expected order of oxidizing power of the anions. Each of the mononuclear anions
-
[
MO
3
(OH)] (M ) Cr, Mo, W) was inert to reaction with methanol, highlighting the importance of the second
MoO
3
unit in these catalytic cycles. Only the dimolybdate center has the mix of properties that allow it to
participate in each of the three steps of the two catalytic cycles. The three reactions of these cycles are
equivalent to the three essential steps proposed to occur in the industrial oxidation of gaseous methanol
to formaldehyde at 300-400 °C over solid-state catalysts based upon molybdenum(VI)-trioxide. The new
gas-phase catalytic data is compared with those for the heterogeneous process.
C-O, and O-H bonds.4,5 However, electrospray ionization
Introduction
(
ESI) permits transfer of metal-oxo ions from condensed phases
Synthetic catalysts contribute to an estimated one-sixth of
the value of all manufactured goods in industrialized countries.
to the gas phase, allowing comparison of reactivity of ions that
exist in both the gaseous and condensed phases. We now report
1
6
Inorganic catalysts are crucial to production of organic chemicals
and petroleum products, and are playing an increasing role in
development of alternative and cleaner industrial processes.
However, the detailed molecular mechanisms of many important
catalytic processes remain obscure. It has become apparent that
gas-phase studies employing multistage mass spectrometry can
be exploited to provide insights into the elementary steps of
catalysis.2
a detailed study of the catalytic oxidation of methanol to
formaldehyde in the gas phase by an anionic dimolybdate(VI)
center using multistage mass spectrometry experiments (Figure
7
1
, reactions 1-5). In a preliminary communication, similarities
were noted between this gas-phase catalysis (eq 6) and that
(
4) (a) Fialko, E. F.; Kikhtenko, A. V.; Goncharov, V. B.; Zamaraev, K. I. J.
Phys. Chem. B 1997, 101, 5772. (b) Fialko, E. F.; Kikhtenko, A. V.;
Goncharov, V. B. Organometallics, 1998, 17, 25.
Coordinatively unsaturated gas-phase transition metal-oxo
(
5) For a selection of other studies on the gas-phase chemistry of molybdenum
oxo species, see: (a) Cassady, C. J.; McElvany, S. W. Organometallics,
1992, 11, 2367. (b) Keesee, R. G.; Chen, B.; Harms, A. C.; Castleman, A.
W., Jr. Int. J. Mass Spectrom. Ion Proc. 1993, 123, 225. (c) Kretzschmar,
I.; Fiedler, A.; Harvey, J. N.; Schr o¨ der, D.; Schwarz, H. J. Phys. Chem. A
3
fragment ions often display little selectivity for bond activation.
For example, the reaction of MoO2+ with methanol gives rise
to four different reaction channels involving activation of C-H,
1997, 101, 6252. (d) Fialko, E. F.; Kikhtenko, A. V.; Goncharov, V. B.;
Zamaraev, K. I.; J. Phys. Chem. A 1997, 101, 8607. (e) Sievers, M. R.;
Armentrout, P. B. J. Phys. Chem. A 1998, 102, 10 754.
(
(
(
1) Shriver, D. F.; Atkins, P. W.; Langford, C. H. Inorganic Chemistry, 2nd
ed.; Oxford University Press: Oxford, 1994; Chapter 17, p 709.
(6) For a selection of studies that have used ESI to transfer metal-oxide anions
to the gas phase, see (a) Lau, T.-C.; Wang, J.; Siu, K. W. M.; Guevremont,
R. Chem. Commun. 1994, 1487. (b) Deery, M. J.; Howarth, O. W.; Jennings,
K. R. J. Chem. Soc., Dalton Trans. 1997, 4783. (c) Walanda, D. K.; Burns,
R. C.; Lawrance, G. A.; von Nagy-Felsobuki, E. I. J. Chem. Soc., Dalton
Trans. 1999, 311. (d) Truebenbach, C. S.; Houalla, M.; Hercules, D. M. J.
Mass Spectrom. 2000, 35, 1121.
2) (a) Plattner, D. A. Int. J. Mass Spectrom. 2001, 207, 125. (b) Zemski, K.
A.; Justes, D. R.; Castleman, A. W., Jr. J. Phys. Chem. B 2002, 106, 6136.
3) (a) Schr o¨ der, D.; Schwarz, H. Angew. Chem., Int. Ed. Engl. 1995, 34, 1973.
(
b) Schr o¨ der, D.; Schwarz, H.; Shaik, S. In Metal-Oxo and Metal-Peroxo
Species in Catalytic Oxidations; Meunier, B., Ed.; Springer: Berlin, 2000;
Struct. Bond.; Vol. 97, p 91.
3384
9
J. AM. CHEM. SOC. 2003, 125, 3384-3396
10.1021/ja028839x CCC: $25.00 © 2003 American Chemical Society

Oxidation of Methanol to Formaldehyde
A R T I C L E S
Figure 1. Gas-phase catalytic cycles for the oxidation of methanol to formaldehyde. Reactions are numbered and described in the text. Reaction 2 links
-
-
[
Mo2O6(OCH3)] (m/z 319) and [Mo2O5(OH)] (m/z 289) and appears in both cycles. Cycles 1 and 2 differ in the sequence of reaction with CH3NO2 and
CH3OH. Quoted m/z values refer to a single peak selected from the dimolybdate isotope manifold and followed through relevant reactions.
occurring in the industrial oxidation of methanol to formalde-
hyde over solid-state molybdenum(VI) oxide catalysts (eq 7)
(ii) The activated methoxo ligands are oxidized and eliminated
as formaldehyde, resulting in the reduction of the catalyst and
the creation of an oxygen vacancy. This step is rate determining
and elevated temperatures are required.
8
CH OH + CH NO f H CO + H O + CH NO (6)
3
3
2
2
2
3
(
iii) Oxidation of the bulk catalyst by dioxygen completes
the cycle.
Although there is qualitative agreement regarding key reac-
CH OH + 1/2 O f H CO + H O
(7)
3
2
2
2
Due to the industrial and economic importance of formalde-
9
hyde, the partial oxidation of methanol over molybdenum(VI)
tions and intermediates in this process, a detailed molecular
understanding of the elementary reactions has been elusive.
Reactions equivalent to steps i-iii above are described in the
gas-phase cycles of Figure 1 (reactions 1-5). Two cycles are
presented, differing in the sequence of reactions. The role of
the binuclear molybdenum center was assessed by examination
10
oxide based catalysts has been the subject of detailed studies
employing a wide variety of synthetic, kinetic, spectroscopic
11-14
and theoretical techniques.
three essential steps:15
The process can be divided into
(i) Gaseous methanol undergoes dissociative adsorption on
-
the surface of the solid-state catalyst to yield surface methoxo-
molybdenum(VI) centers. Water may be desorbed from the
catalyst in this step.
of the relative reactivities of the mononuclear [MO3(OH)] and
-
binuclear [M2O6(OH)] , (M ) Cr, Mo, W) ions.
Experimental Section
(
7) A number of gas-phase catalytic cycles have been established using mass
spectrometry. See for example: oxidation of CO (a) Kappes, M. M.; Staley,
R. H. J. Am. Chem. Soc. 1981, 103, 1286. (b) Baranov, V.; Javahery, G.;
Hopkinson, A. C.; Bohme, D. K. J. Am. Chem. Soc. 1995, 117, 12 801. (c)
Shi, Y.; Ervin, K. M. J. Chem. Phys. 1998, 108, 1757; oxidation of methane.
+
]2-
Synthesis. Tetra-n-butylammonium (Bu
4
N ) salts of [Cr
2
O
7
,
2-
2-
2-
]2- were synthesized using
[Mo
standard literature methods.
2 7 4 4
O ] , [CrO ] , [MoO ] , and [WO
4
16-20
(d) Wesendrup, R.; Schr o¨ der, D.; Schwarz, H. Angew. Chem., Int. Ed. Engl.
Reagents. Methanol, ethanol, n-propanol, i-propanol, and tert-butyl
alcohol (HPLC grade, 99.5%) were obtained from Aldrich and used
1
994, 33, 1174; and formation of benzene from ethene. (e) Schnabel, P.;
Irion, M. P.; Weil, K. G. J. Phys. Chem. 1991, 95, 9688.
without further purification. CF
ICN Biomedicals. CFH CH OH (95%) was obtained from Lancaster.
CH OD (99.5 atom % D), CD OD (99.8 atom % D), CD CH OH (99
atom % D) and CH CD OH (98 atom % D) were obtained from Aldrich.
3 2
CH OH (99.5%) was obtained from
(
(
8) Waters, T.; O’Hair, R. A. J.; Wedd, A. G. Chem. Commun. 2000, 225.
9) The huge demand for formaldehyde-based resins for particleboard wood
products makes formaldehyde one of the top 25 chemicals produced in
industrial countries, with approximately 15 million metric tons of form-
aldehyde (37% aqueous solution) produced annually. See for example:
Gerberich, H. R.; Seaman, G. C. Formaldehyde in Kirk-Othmer Encyclo-
pedia of Chemical Technology, 4th ed.; Howe-Grant, M., Ed.; John Wiley
and Sons: New York, 1994; vol. 11, p 929-950.
2
2
3
3
3
2
3
2
18
18
3 3
CH OH (95 atom % O) was obtained from Isotec. CH CHDOH
21
was synthesized using a standard literature procedure and characterized
by NMR. Acetonitrile (HPLC grade, 99.8%) was obtained from Merck.
Mass Spectrometry. Mass spectrometry experiments were con-
ducted using a modified Finigan LCQ quadrupole ion trap mass
spectrometer equipped with a Finnigan electrospray ionization source.
Tetrabutylammonium salts of metal-oxide anions were dissolved in
acetonitrile (0.1 mg/mL). The solution was pumped into the electrospray
source at approximately 3 µL/min. Typical electrospray source condi-
tions involved needle potentials of 3.5-4.0 kV and heated capillary
temperatures of 150-250 °C. Extensive tuning of electrospray condi-
tions was often required due to the low signal-to-noise ratio and/or
low abundance of some species. Mass selection and collisional
(
10) A molybdenum(VI) trioxide-based catalyst or a metallic silver catalyst is
used. About 70% of newly installed production capacity uses the former.
The heterogeneous catalysis is typically carried out at atmospheric pressure
and 300-400 °C with atmospheric dioxygen as the oxidant. High methanol
conversion (>99%) and selectivity for formaldehyde (>95%) can be
achieved. ibid.
(
(
(
11) (a) McCarron, E. M., III.; Staley, R. H.; Sleight, A. W. Inorg. Chem. 1984,
2
3, 1043. (b) McCarron, E. M., III.; Sleight, A. W. Polyhedron, 1986, 5,
1
29. (c) McCarron, E. M., III.; Harlow, R. L.; Li, Z. G.; Suto, C.; Yuen,
Y. J. Solid State Chem. 1998, 136, 247.
12) (a) Holstein, W. L.; Machiels, C. J. J. Catal. 1996, 162, 118. (b) Ardissone,
D. E.; Valente, N. G.; Cad u´ s, L. E.; Arr u´ a, L. A. Ind. Eng. Chem. Res.
2
000, 39, 2902. (c) Faliks, A.; Yetter, R. A.; Floudas, C. A.; Bernasek, S.
L.; Fransson, M.; Rabitz, H. J. Phys. Chem. A 2001, 105, 2099.
13) See for example: (a) Groff, R. P. J. Catal. 1984, 86, 215. (b) Hu, H.;
Wachs, I. E.; Bare, S. R. J. Phys. Chem. 1995, 99, 10 897. (c) Zhang, W.;
Desikan, A.; Oyama, S. T. J. Phys. Chem. 1995, 99, 14 468. (d)
Radhakrishnan, R.; Reed, C.; Oyama, S. T.; Seman, M.; Knodo, J. N.;
Domen, K.; Ohminami, Y.; Asakura, K. J. Phys. Chem. B 2001, 105, 8519.
(16) Landini, D.; Rolla, F. Chem. Indust., 1979, 213.
(17) Hur, N. H.; Klemperer, W. G.; Wang, R.-C. In Inorganic Synthesis;
Ginsberg, A. P. Ed.; John Wiley & Sons: New York, 1990; Vol. 27, p
79-80.
(e) Burcham, L. J.; Briand, L. E.; Wachs, I. E. Langmuir, 2001, 17, 6164.
(
14) (a) Allison, J. N.; Goddard, W. A., III J. Catal. 1985, 92, 127. (b) Weber,
R. S. J. Phys. Chem. 1994, 98, 2999. (c) Rahmouni, A.; Barbier, C. J.
Mol. Struct. (THEOCHEM), 1995, 330, 359.
15) (a) Farneth, W. E.; Ohuchi, F.; Staley, R. H.; Chowdhry, U.; Sleight, A.
W. J. Phys. Chem. 1985, 89, 2493. (b) Farneth, W. E.; Staley, R. H.; Sleight,
A. W. J. Am. Chem. Soc. 1986, 108, 2327.
(18) Nakayama, H. Bull. Chem. Soc. Jpn. 1983, 56, 877.
(19) Klemperer, W. G.; Liu, R.-S. Inorg. Chem. 1980, 19, 3863.
(20) Che, T. M.; Day, V. W.; Francesconi, L. C.; Fredrich, M. F.; Klemperer,
W. G.; Shum, W. Inorg. Chem. 1985, 24, 4055.
(
(21) Baldwin, J. E.; Reddy, V. P. J. Am. Chem. Soc. 1987, 109, 8051.
J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003 3385

A R T I C L E S
Waters et al.
differential equation describing the molar flow of neutral reagents about
the ion trap (i.e., the ‘flux’ of neutral reagents) allows the equilibrium
pressure of neutral reagents, P (Torr), to be calculated (eq 8). Due to
n
the large excess of neutral reagents relative to the reactant ion within
the ion trap, all rates are measured under pseudo first-order conditions
molar flow rate of neutral
molar flow rate of helium
molar mass of neutral
molar mass of helium
P ) P ×
×
n
b
x
(8)
Rates were measured by varying the time delay between isolation of
the reactant ion and its mass analysis (‘Reaction Delay’, RD). Pseudo
first-order rates were estimated by extrapolation of plots of ln(reactant
ion intensity) vs RD. 3-4 peaks from the isotope pattern of the reactant
ion were mass selected and their decay monitored over at least 15 values
of RD, spanning at least 3 reaction half-lives. The absolute intensity
of the reactant ion was calculated by integration of its ion count within
the mass selected window. Absolute rate constants were calculated by
dividing the pseudo first order rate coefficient by the calculated
concentration of neutral reagent in the ion trap. The absolute rate
constants reported are the average of at least seven independent
measurements conducted on at least 4 separate days. All rates are
measured over at least a 5-fold neutral pressure range, using at least 3
different neutral reagent flowrates and 3 different helium flowrates.
Upper bounds on rate constants for unreactive species were determined
given that no reaction was observed after 10 s. Standard deviations in
absolute rate constants were typically <10%. A conservative estimate
of error is (25%, but relative rates are expected to be more accurate
due to cancellation of errors.
-
Figure 2. (a) Theoretical isotope pattern of [Mo2O6(OH)] ; (b) Experi-
-
mental isotope pattern of the ion assigned as [Mo2O6(OH)] ; and (c) Mass
To test this procedure, experimental absolute rate constants deter-
mined for two reactions with the present instrumentation were compared
with literature values obtained via flowing afterglow techniques at 298
K (eq 9,10).25 This close agreement demonstrates that the present
instrument provides ion-molecule rate constants of near thermal ions,
selection of a single peak, m/z 305.
activation were carried out using standard isolation and excitation
procedures using the “advanced scan” function of the LCQ software.
Dimolybdate species display a distinctive isotope pattern (e.g., Figure
2
2
2
3
92
consistent with previous rate measurements in a similar system and
findings that ions within the ion trap are essentially at room temper-
ature24
2
a) due to the seven naturally occurring isotopes of molybdenum ( Mo,
9
4
95
96
97
98
1
4.84; Mo, 9.25; Mo, 15.92; Mo, 16.68; Mo, 9.55; Mo, 24.13;
Mo, 9.68 atom %). This facilitates assignment of ion stoichiometry
100
via comparison of experimental and theoretical isotope patterns (e.g.,
Figure 2a, b). However, the broad isotope pattern can also make small
mass changes difficult to detect (e.g. during isotope labeling experiments
involving both hydrogen and deuterium). To avoid this problem, the
single most intense peak was mass selected from the dimolybdate
isotope manifold (e.g. Figure 2c), and used to follow the course of
reactions. Quoted m/z values in the text refer to this most intense peak.
The instrument has been modified to permit introduction of neutral
reagents into the ion trap, allowing the measurement of ion-molecule
-
-
Br + CH I f CH Br + I
3
3
-
11
cm molecule s-1
3
-1
klit ) 2.89 × 10
;
kexp ) 2.7 × 10 cm molecule s-1 (9)
-11
3
-1
-
-
-
(
CF ) CHO + CH CH Br f (CF ) CHOH + CH CH + Br
3
2
3
2
3 2
2
2
-
2
3,24
f (CF ) CHOCH CH + Br
3 2 2 3
reaction rate constants.
The modifications and experimental pro-
3
cedure are described briefly. A known flow of helium (470-4880 cm /
min) is passed through a stainless steel tube. A measured flow of neutral
reagent (2.5-1000 µL/hour) is injected downstream into this tube from
a gastight syringe using a Harvard syringe drive. The injection port is
heated to about 50 °C to ensure rapid and efficient evaporation of neutral
reagent into the helium flow. The composition of this external mixture
is calculated using the known molar flow rates of helium and neutral
reagent. A small fraction of this mixture (∼0.25 mL/min) is carried
through a restriction capillary into the ion trap, while the remainder is
diverted to an exhaust system. The restriction capillary maintains the
klit ) 7.83 × 10 cm molecule s-1
-
12
3
-1
;
kexp ) 7.9 × 10 cm molecule s-1 (10)
-12
3
-1
Reaction efficiencies (φ) were calculated by dividing the experimentally
determined rate constant (kexp) by theoretical predictions of ion-
molecule collison rate constants26 (kado); i.e., φ ) kexp/kado. Uncertainties
in reaction efficiencies have not been estimated, but would be comprised
of uncertainties in the experimentally determined rate constant and in
the theoretical collision rate constants. A complete list of experimental
rate constants, theoretical collision rate constants and reaction efficien-
cies is included in the Supporting Information (Table S1).
-
3
pressure of the ion trap at P
b
) 1.75 ( 0.2 × 10 Torr. Helium and
neutral reagent molecules are able to exit the ion trap via effusion into
the surrounding vacuum manifold. The lighter helium atoms effuse more
quickly and are lost at a greater rate. The steady-state solution of the
The estimation of absolute collision energies applied during colli-
sional activation experiments in quadrupole ion trap instruments is not
straightforward. Colorado and Brodbelt have demonstrated a linear
(
22) Practical Aspects of Ion Trap Mass Spectrometry; March, R. E., Todd, J.
F. J., Eds.; CRC Press: Florida, 1995; (a) vol. 1 Fundamentals of Ion Trap
Mass Spectrometry; (b) vol. 2 Ion Trap Instrumentation.
(25) Gronert, S.; DePuy, C. H.; Bierbaum, V. M. J. Am. Chem. Soc. 1991, 113,
4009.
(26) Su,T.; Bowers, M. T. In Gas-Phase Ion Chemistry; Bowers, M. T. Ed.;
Academic Press: New York, 1979; p. 83.
(
23) (a) Flores, A. E.; Gronert, S. J. Am. Chem. Soc. 1999, 121, 2627. (b) Gronert,
S.; Pratt, L. M.; Mogali, S. J. Am. Chem. Soc. 2001, 123, 3081.
24) Gronert, S. J. Am. Soc. Mass Spectrom. 1998, 9, 845.
(
3386 J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003

Oxidation of Methanol to Formaldehyde
A R T I C L E S
relationship between the threshold activation voltage required for
dissociation in ion-trap experiments and the established critical energies
for a variety of ions.27 However, separate calibration curves were
required for hydrogen bonded and covalently bound species. The effect
of factors such as ion mass and ion density, resonance excitation time
and trapping well depth on threshold activation voltages in ion trap
2
8
experiments have also been investigated. A detailed investigation of
these factors is beyond the scope of the present study, and we refrain
from attempting quantification of critical energies of collisional
activation reactions. However, to provide a qualitative insight into the
critical energies of such reactions (e.g Figure 1, reaction 2), we have
compared activation voltages required for these reactions with those
required for the dissociation of two species with established critical
+
+
5 5 2 3
H ) (generated via ESI of AgNO
energies, Ag(CH
3
OH) and Fe(C
in CH CN, respectively). These species were
29
in CH OH and Fe(C
3
5
5
H )
2
3
chosen as the activation voltages required for their dissociation
effectively ‘brackets’ the activation voltages used in present experi-
ments.
Threshold voltages were determined by mass selecting a single peak
from the isotope pattern of the species of interest and applying an
2
7
activation voltage for a period of 10 ms. Activation voltages (Vp-p
)
were increased in a stepwise fashion until complete dissociation of the
ion of interest was observed. In line with previous studies, the voltage
at which the fragment ion intensity is 10% of the total ion intensity is
27
defined as the ‘threshold voltage’. Given that relatively low concentra-
-
7
tions of neutral reagents (typically 2-4 × 10 Torr of CH
3
OH or
CH I) are required to generate species of interest, the effect of these
3
neutral reagents on the collisional activation process was examined.
No discernible effect in the measured threshold voltage was evident
+
+
for either Ag(CH OH) or Fe(C H ) , suggesting the influence of low
3 5 5 2
concentrations of neutral reagents on the measured ‘threshold voltages’
is negligible. Given important factors such as lifetime effects, kinetic
shifts, differing excitation efficiencies and mass effects have been
ignored in these experiments, the data presented is only intended as a
rough approximation. Further, reactions described in the present work
may exhibit reverse activation barriers, presenting a further complication
to interpretation of data.
Theoretical Calculations. B3LYP Density Functional Theory (DFT)
3
0
calculations were performed using the GAMESS electronic structure
31
program. Local minima were characterized using numerical Hessians.
The Effective Core Potential (ECP) basis sets of Stevens et al. were
3
2
++
used for Cr, Mo, and W, and the 6-311 G** basis set of Pople
33
Figure 3. Mass spectra for reactions 1-5 of Cycles 1 and 2 (Figure 1). A
single peak from the dimolybdate isotope manifold was mass selected and
followed through each reaction. CH3OH and CH3NO2 are present in the
ion-trap in equal concentrations. (a) reaction 1 of Cycle 1; (b) reaction 2 of
Cycles 1 and 2. The small peaks at m/z 303 and 305 are result of reaction
and co-workers for O and H. Unscaled zero point energies are included
34
for all species. Structures were visualized using MacMolPlt. Cartesian
coordinates and energies of optimized geometries are included in the
Supporting Information (Figures S12, S14, S15).
3
of Cycle 1 and reaction 4 of cycle 2 (see (c) below); (c) reaction 3 of
Results and Discussion
Cycle 1 and reaction 4 of Cycle 2. This is the divergent point of the two
cycles. The peak at m/z 319 is the result of the ions of m/z 303 and 305
reacting further (reactions 5, Cycle 2 and reaction 1, Cycle 1, respectively);
and (d) reaction 5 of Cycle 2.
Negative ion electrospray mass spectrometry of (Bu4N)2-
[Mo2O7] in acetonitrile generated three major products: the
2
-
dimolybdate dianion, [Mo2O7] (m/z 152), protonated di-
-
molybdate, [Mo2O6(OH)] (m/z 305; Figure 2), and the ion pair,
{
vided access to important intermediates in the catalytic cycles
of Figure 1.
(
27) Colorado, A.; Brodbelt, J. J. Am. Soc. Mass Spectrom. 1996, 7, 1116.
28) (a) Jackson, G. P.; King, F. L.; Goeringer, D. E.; Duckworth, D. C. Int. J.
Mass Spectrom. 2002, 216, 85. (b) Hart, K. J.; McLuckey, S. A. J. Am.
Soc. Mass Spectrom. 1993, 5, 250.
+
2-
-
35
Bu4N [Mo2O7] } (m/z 546). Each of these species pro-
(
+
+
(29) (a) Ag(CH
3
OH) f Ag + CH
3
OH see: El Aribi, H.; Shoeib, T.; Ling,
-
Y.; Rodriquez, C. F.; Hopkinson, A. C.; Siu, K. W. M. J. Phys. Chem. A
Catalytic Oxidation Of Alcohols by [Mo2O6(OH)] . The
+
+
•
5
2
002, 106, 2908; and (b) Fe(C
H
5 5
)
2
f Fe(C
5 5 5
H ) + C H see: Faulk, J.
two catalytic cycles 1 and 2 of Figure 1 differ in the order of
reaction of [Mo2O5(OH)] with CH3OH and CH3NO2. Repre-
sentative mass spectra for reactions 1-5 of these cycles are
included in Figure 3.
D.; Dunbar, R. C. J. Am. Chem. Soc. 1992, 114, 8596.
-
(
30) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 1372. (b) Stephens, P. J.; Devlin,
F. J.; Chabalowski, C. F.; Frisch, M. J. J. Phys. Chem. 1994, 98, 11 623.
31) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.; Gordon, M.
S.; Jensen, J. H.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.; Su, S. J.;
Windus, T. L.; Dupuis, M.; Montgomery, J. A. J. Comput. Chem. 1993,
(
1
4, 1347.
32) Stevens, W. J.; Krauss, M.; Basch, H.; Jasien, P. G.; Can. J. Chem. 1992,
0, 612.
33) Krishnan, R.; Binkley, J. S.; Seegar, R.; Pople, J. A. J. Chem. Phys. 1980,
2, 650.
34) Bode, B. M.; Gordon, M. S. J. Mol. Graphics Mod. 1998, 16, 133.
(
(
(
(35) ESI of (Bu
4
N)
2
[Mo
2
O
7
] has been reported previously: from CH
3
CN (a)
7
Lau, T.-C.; Wang, J.; Guevremont, R.; Siu, K. W. M. Chem. Commun.
1995, 877; and from CH
Nishikawa, K.; Kinoshita, I.; Isobe, K. Angew. Chem., Int. Ed. Engl. 1999,
38, 3051.
3
OH. (b) Takara, S.; Ogo, S.; Watanabe, Y.;
7
J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003 3387

A R T I C L E S
Waters et al.
Figure 5. First-order plot for reaction of [Mo2O6(OH)]- with CH3OH
(Figure 1, reaction 1).
then reacted with CH3NO2 to generate [MoVI2O6(OCH3)] (m/z
-
3
19) and complete cycle 2 (reaction 5, Figure 3d).
The multistage trapping capabilities of the ion-trap instrument
allowed cycles 1 and 2 to be completed thrice and twice,
respectively, on the same population of ions. This confirmed
that they were truly catalytic for the oxidation of primary and
secondary alcohols to aldehydes (Figures S2, S3). Individual
reactions are now discussed in more detail using the numbering
system shown in Figure 1.
-
Cycle 1, Reaction 1: Interaction of [Mo2O6(OH)] with
-
Alcohols. Mass selection of [Mo2O6(OH)] (m/z 305) in the
presence of methanol solely led to the formation of a new ion,
-
m/z 319, assigned as [Mo2O6(OCH3)] (eq 1a). Isotope labeling
experiments were consistent with the binding of methoxo ligand
from methanol and the elimination of hydroxo ligand as neutral
water (eq 1a-d, Figure 4). Competitive H/D exchange was also
observed for reaction with CH3OD and CD3OD (Figure 4).
Reaction with methanol followed smooth pseudo first-order
kinetics (Figure 5) with a reaction efficiency of approximately
5
% (φ ) 0.049, Table 1)
-
[
Mo O (OH)] (m/z 305) + CH OH f
2
6
3
-
[
Mo O (OCH )] (319) + H O (1a)
2 6 3 2
-
[
Mo O (OH)] (305) + CH OD f
2
6
3
-
-
Figure 4. Mass spectra showing the reaction of [Mo2O6(OH)] with
[
Mo O (OCH )] (319) + DOH (1b)
2 6 3
-
isotopically labeled methanol. Reaction of [Mo2O6(OH)] (m/z 305) with
the following: (a) CH3OH to form [Mo2O6(OCH3)] (m/z 319); (b) CH3OD
to form [Mo2O6(OCH3)] (m/z 319). Note that H/D exchange with CH3OD
also occurs to form [Mo2O6(OD)] (m/z 306); (c) CD3OD to form
Mo2O6(OCD3)] (m/z 322). Note that H/D exchange occurrs again; and
d) CH3 OH to form [Mo2O6( OCH3)] (m/z 321). The peak at m/z 319
is a result of reaction with CH3 OH impurity in the CH3 OH sample.
-
-
-
[Mo O (OH)] (305) + CD OD f
2 6 3
-
-
-
[Mo O (OCD )] (322) + DOH (1c)
2 6 3
[
(
1
8
18
-
1
6
18
16
16
-
18
[
Mo2 O ( OH)] (305) + CH OH f
6 3
1
6
18
-
16
[
Mo2 O ( OCH )] (321) + H O (1d)
6 3 2
In cycle 1, [MVI2O6(OH)] (m/z 305) reacted with methanol
to form [MoVI2O6(OCH3)] (m/z 319) and eliminate water
Figure 1, reaction 1; Figure 3a). Collisional activation of
Mo 2O6(OCH3)] eliminated formaldehyde with the formation
of 2e -reduced [Mo 2O5(OH)] (m/z 289) (reaction 2; Figure
b). This ion then underwent an oxygen atom transfer reaction
with nitromethane to regenerate [Mo 2O6(OH)] and complete
-
-
2-
A reaction between [Mo2O7] and CH3OH to produce
Mo2O6(OCH3)] was reported incorrectly in our previous
is unreactive toward
CH3OH. In those experiments, an acetic acid impurity in the
gas handling system resulted in protonation of the dianion to
form [Mo2O6(OH)] , which then reacted with CH3OH to form
Mo2O6(OCH3)] (equivalent to reaction 1).
[Mo2O6(OH)] also reacted with other alcohols to from
alkylated dimolybdate anions [Mo2O6(OR)] (Table 1, Fig.
(
[
-
[
VI
-
8
2-
communication. In fact, [Mo2O7]
-
V
-
3
VI
-
-
cycle 1 (reaction 3; Figure 3c).
-
[
V
-
Reaction 2 is common to both cycles. Its product, [Mo 2O5-
-
-
(
OH)] (m/z 289), reacted with either CH3NO2 (cycle 1, reaction
) or with CH3OH (cycle 2, reaction 4). This is the divergent
point of the two cycles. Reaction with CH3OH (cycle 2, reaction
; Figure 3c) produced [Mo 2O5(OCH3)] (m/z 303). This ion
3
6
3
S4). A comparison of reaction efficiency and alcohol acidity
for primary alcohols suggests an approximate direct dependence,
consistent with a mechanism involving deprotonation of the
V
-
4
3388 J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003

Oxidation of Methanol to Formaldehyde
A R T I C L E S
Table 1. Reaction Rates for the Reaction of [Mo2O6(OH)]- with
Alcohols
∆
G° a
reaction
efficiency, φ c
acid
b
alcohol
CH3OH
CH3CH2OH
CH3CH2CH2OH
(kcal/mol)
k
exp
375.1
371.7
369.6
368.5
368.1
364.6
354.1
6.9
9.1
14
8.9
2.8
25
66
0.049
0.070
0.11
0.069
0.023
n/a
(
(
CH3)2CHOH
CH3)3COH
d
CFH2CH2OH
CF3CH2OH
d
n/a
a
Gas-phase acidity is defined as the Gibbs energy change for the reaction
-
+ 37
b
-11
3
-1 -1 c
ROH f RO + H .
× 10
cm molecules
s . φ ) kexp/kADO.
d
Theoretical collision rate constants and thus reaction efficiencies for these
alcohols have not been calculated as their dipole moments have not been
determined experimentally.
Figure 6. Plot of Reaction Extent (∑product ions/total ion count) vs
-
activation voltage (Vp-p) for elimination of H2CO from [M2O6(OCH3)]
+
(
M ) Cr, Mo, W). The critical energies for dissociation of Ag(CH3OH)
alcohol and binding of the alkoxo ligand, as indicated by the
isotope labeling experiments (eq 1a-d). The branched alcohols
reacted more slowly than expected based on their acidity,
presumably due to steric constraints. Because the presence of
alcohol in the ion-trap can interfere with studies of subsequent
steps of the catalytic cycles (e.g., Figure 1, Figure 3c; reaction
+
29
(
33.0 ( 3.7 kcal/mol) and Fe(C5H5)2 (85 ( 7 kcal/mol) are known. The
dashed line corresponds to 10% reaction extent, the point at which threshold
activation voltages are measured.
-
about one-third of that observed for reaction of [Mo2O6(OH)]
-
with methanol (Table 1). [Mo2O6(OBu)] also underwent
analogous ligand switching reactions with other alcohols (Figure
S6)
4
of cycle 2 competes with reaction 3 of cycle 1), an independent
-
38
method of generating [Mo2O6(OR)] is now discussed.
-
+
Formation of [Mo2O6(OBu)] from the Ion-Pair, {Bu4N -
-
2
-
-
+
2-
-
[Mo O (OBu)] (m/z 361) + CH OH f
[
Mo2O7] } . Collisional activation of {Bu4N [Mo2O7]
}
2
6
3
-
resulted in elimination of neutral tri-n-butylamine, NBu3, and
[
Mo O (OCH )] (319) + BuOH (14)
2 6 3
-
formation of [Mo2O6(OBu)] (m/z 361) (eq 11). This reaction
is proposed to occur via an SN2 pathway: the nucleophilic
dimolybdate dianion attacks the R-carbon of the tetrabutyl-
ammonium cation. Related SN2 reactions within ion-pairs
containing organo dianions are known.39 Interestingly, SN2
O-alkylation of larger polyoxometalate anions via reaction with
strongly electrophilic trialkyloxonium ions have also been
observed in condensed phases (eqs 12,13)40
Cycles 1 and 2, Reaction 2: Elimination of Formaldehyde
VI
-
VI
from [Mo 2O6(OCH3)] . Collisional activation of [Mo 2O6-
-
(
OCH3)] (m/z 319) resulted in a mass loss of 30, consistent
with elimination of neutral formaldehyde (Figure 1, reaction 2;
Figure 3b). The product ion of this reaction, of stoichiometry
-
V
[
(
Mo2, O6, H] (m/z 289), is tentatively assigned as [Mo 2O5-
-
OH)] . The mechanism of reaction was investigated via
isotopically labeled methoxo ligands generated via eq 1a-d (eq
2a-c, Figure S7). These experiments demonstrated that form-
aldehyde originated from the methoxo ligand. The dimolybdate
center retained the two electrons and proton produced in the
+
2-
-
-
{
Bu N [Mo O ] } f [Mo O (OBu)] + NBu
4 2 7 2 6
3
gas phase (11)
3
-
+
[
PMo O ] + (CH ) O f
-
-
+
1
2
40
3 3
formal redox half-equation: CH3O f H2CO + 2e + H
2-
[
PMo O (OCH )] + (CH ) O solution (12)
12 39 3 3 2
VI
-
[
Mo O (OCH )] (m/z 319) f
2 6 3
2-
V
-
(
R O) [PW O ] f (R O) [PW O (OR)] + R O
[Mo O (OH)] (289) + H CO (2a)
2 5 2
3
3
12 40
3
2
12 39
2
solid state (13)
VI
2
-
[
Mo O (OCD )] (322) f
6
3
The presence of a butoxo ligand in the product of eq 11 was
indicated by the ligand switching reaction with methanol (eq
V
-
[
Mo O (OD)] (290) + D CO (2b)
2 5 2
1
4). Results from reaction of [Mo2O6(OBu)]- with labeled
-
VI 16
O ( OCH )] (321) f
2 6 3
18
-
methanol (Figure S5) were equivalent to those for [Mo2O6(OH)]
[Mo
(
eq 1a-d), suggesting a similar mechanism of reaction and
V 16
16
-
18
[
Mo 2 O ( OH)] (289) + H C O (2c)
5
2
consistent with elimination of the butoxo ligand as butanol.
Reaction efficiency for this reaction was 1.7% (φ ) 0.017),
An estimate of the critical energy of reaction 2 was derived via
comparison with species of known critical energies (Figure 6).
The critical energy for dissociation of Ag(CH3OH) is 33.0 (
(
36) For examples of condensation of alcohols at M-OH bonds in the gas-phase,
see: (a) Fiedler, A.; Schr o¨ der, D.; Schwarz, H.; Tjelta, B. L.; Armentrout,
P. B. J. Am. Chem. Soc. 1996, 118, 5047. (b) Richardson, D. E.; Lang, G.
H. L.; Crestoni, E.; Ryan, M. F.; Eyler, J. E. Int. J. Mass Spectrom. 2001,
+
3
.7 kcal/mol (threshold voltage ) 0.38 Vp-p), and the critical
2
04, 255. (c) Jackson, P.; Fisher, K. J.; Willett, G. D. Chem. Phys. 2000,
62, 179.
+
energy for dissociation of Fe(C5H5)2 is 85 ( 7 kcal/mol
(threshold voltage ) 0.94 Vp-p). The threshold voltages for
2
2
9
(
37) Alcohol acidities are from: http://webbook.nist.gov/chemistry/.
38) Alcohol is present in the ion trap during all stages of a multistage
experiment. If alcohol is used in initial reactions of the catalytic cycle (e.g.,
reaction 1) it remains present during later reactions.
(
these reactions bracket the threshold voltage observed for
-
[Mo2O6(OCH3)] (threshold voltage ) 0.61 Vp-p, Figure 6).
(39) Gronert, S.; Azebu, J. Org. Lett. 1999, 1, 503.
This suggests a crude estimate of 55 ( 20 kcal/mol for the
critical energy of reaction 2.
(40) (a) Knoth, W. H.; Harlow, R. L. J. Am. Chem. Soc. 1981, 103, 4265. (b)
Knoth, W. H.; Farlee, R. D. Inorg. Chem. 1984, 23, 4765.
J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003 3389

A R T I C L E S
Waters et al.
Dimolybdate anions with a range of deuterated ethoxo ligands
were synthesized by reaction of [Mo2O6(OH)]- with the
appropriately labeled ethanol. Collisional activation of these
species resulted in elimination of the ethoxo ligand as acet-
aldehyde (eq 15a-c, Figure 7).41 The hydrogen retained on the
dimolybdate center originated from the R-position of the alkoxo
ligand only (Figure 7a,b). These results are consistent with a
mechanism involving transfer of an hydrogen atom from the
R-position, reduction of the binuclear metal center and elimina-
tion of aldehyde, i.e., consistent with the Westheimer mecha-
eliminated 2-methyl-propene (eq 19). The redox mechanism has
been switched off by the absence of R-hydrogen in the tertiary
alkoxo ligand. The observed products are consistent with transfer
t
of a â-hydrogen atom from the Bu substituent to an oxo ligand
of the dimolybdate center and elimination of 2-methyl-propene.
-
Collisional activation of [Mo2O6(OCH(CH3)2] resulted in
elimination of both acetone and propene (eq 17a,b). This is
consistent with an increased barrier to aldehyde elimination due
to increased steric bulk at the R-carbon of the secondary alkyl
group
42
nism. This mechanism was originally proposed to account for
the elimination of aldehyde from chromate-ester intermediates
in the oxidation of alcohols by high valent chromium(VI)-oxo
reagents
VI
2
-
[Mo O (OCH CH CH )] f
6
2
2
3
V
-
[
Mo O (OH)] + CH CH CHO (16)
2 5 3 2
VI
-
VI
-
[Mo O (OCH(CH )) ] f
2 6 3 2
[
[
[
Mo O (OCH CH )] (m/z 333) f
2 6 2 3
V
2
-
V
-
[Mo O (OH)] + (CH ) CO≈ 75% (17a)
5
3 2
[
Mo O (OH)] (289) + CH CHO (15a)
2 5 3
VI
2
-
VI
2
-
f [Mo O (OH)] + CH CHdCH ≈ 25% (17b)
Mo O (OCD CH )] (335) f
6
3
2
6
2
3
V
2
-
[
Mo O (OD)] (290) + CH CDO (15b)
VI
-
5
3
[
Mo O (OCH (CH ) CH )] f
2 6 2 2 2 3
V
-
VI
2
-
[
Mo O (OH)] + CH (CH ) CHO (18)
Mo O (OCH CD )] (336) f
2 5 3 2 2
6
2
3
V
2
-
[
Mo O (OH)] (289) + CD CHO (15c)
VI
2
-
5
3
[
Mo O (OC(CH ) )] f
6
3 3
Acetaldehyde elimination from [Mo2O6(OCHDCH3)]- was
examined using the ion synthesized via reaction of [Mo2O6(OH)]
VI
2
-
[Mo O (OH)] + (CH ) CdCH (19)
6
3 2
2
-
The elimination of aldehyde from [Mo2O6(OCHR2)]- species
(reaction 2, eq 2, 15-18) does not proceed without the use of
collisional activation. This reaction represents the rate-limiting
step in the gas-phase catalytic cycles of Figure 1.
Cycle 1, Reaction 3: Interaction of [Mo 2O5(OH)] with
CH3NO2. The product of reaction 2 was assigned to [Mo 2O5-
OH)] whose stoichiometry differs from [Mo 2O6(OH)] by
an oxygen atom. This suggested closing cycle 1 by an oxygen
atom transfer reaction. The reaction of [Mo 2O5(OH)] with
potential oxygen atom transfer reagents CH3NO2, (CH3)2SO,
N2O, O2, and 1,2-epoxybutane was investigated. To avoid
with CH3CHDOH. Collisional activation resulted in the loss of
a mixture of CH3CDO and CH3CHO (eq 15d,e; Figure 7c). The
product ratio of CH3CDO to CH3CHO (inferred from the relative
ion intensities of the product anions) suggeted a kinetic isotope
effect of approximately 1.9 ( 0.4 for this reaction under the
V
-
4
3
V
collisional activation conditions of the present experiments
-
VI
-
(
VI
-
[
Mo O (OCHDCH )] (m/z 334) f
2 6 3
V
-
V
2
-
[
Mo O (OH)] (289) + CH CDO ≈ 65% (15d)
5
3
V
2
-
f [Mo O (OD)] (290) + CH CHO ≈ 35% (15e)
5
3
38
-
competing reactions with alcohols, [Mo2O5(OH)] was gener-
ated in a sequence of reactions from the ion-pair (eqs 11 and
Collisional activation of dimolybdate centers with branched
alkoxo ligands was also examined (eq 17, 19) and compared
with their straight chain counterparts (eq 16, 18, Figure S8).
-
1
8). Only CH3NO2 reacted with [Mo2O5(OH)] via oxygen atom
4
4,45
transfer.
V
-
-
Mass selection of [Mo 2O5(OH)] (m/z 289) and reaction with
These species were synthesized via reaction of [Mo2O6(OH)]
n
-
nitromethane solely resulted in the formation of a new ion at
with the appropriate alcohol, whereas [Mo2O6(O Bu)] was
VI
-
+
2-
-
m/z 305, assigned to [Mo 2O6(OH)] (eq 3; Figure 1, reaction
; Figure 3c). This reaction deviated from smooth first order
kinetics (Figure 8a), with approximately 20% of the sample
generated from {Bu4N [Mo2O7] } (eq 11). Collisional activa-
tion of dimolybdate centers with straight chain alkoxo ligands
results in elimination of the corresponding aldehyde (eq 16,
3
4
6
41
-
being unreactive. A theoretical decay curve could be fitted
using an experimental estimate of the amount of unreactive
1
8). In contrast, collisional activation of [Mo2O6(OC(CH3)3]
(
41) A very small amount of alkene loss is also observed from dimolybdate
centres with straight chain alkoxo ligands. This typically represents less
than 2% of the amount of aldehyde loss.
(44) Upper bounds can be placed on the rate constants for reaction of
V
-
-14
-1
[Mo
2
O
5
(OH)] with these unreactive neutrals: (CH
3
)
2
SO, kexp < 5 × 10
3
-1 -1
-14
3
(
42) Westheimer, F. H. Chem. ReV. 1949, 45, 419. We adopt the notation
whereby the first carbon of the alkoxo ligand is denoted as the R-carbon.
43) This kinetic isotope effect (KIE) is estimated based on experiments where
the activation voltage, activation time and, thus, reaction extent were
systematically varied. The estimate of 1.9 ( 0.4 represents an average value
under conditions where greater than 20% reaction occurred, and the error
in this estimate is based upon the upper and lower experimental values
obtained under these conditions. Given that this reaction only occurs upon
collisional activation, and that the ions in these experiments are significantly
above room temperature, the KIE at 298 K may be larger. A crude estimate
of the KIE effect for â-hydrogen transfer and ethene elimination can also
be inferred from these experiments: this channel represents 2.3% of
cm molecule
; N
s
; 1,2-epoxybutane, kexp < 5 × 10
cm molecule
-
1
-13
3
-1 -1
s
2
O, kexp < 5 × 10
cm molecule
s
. There appears to be no
(
correlation between the enthalpy of oxo transfer for these neutrals (Holm,
R. H.; Donahue, J. P. Polyhedron, 1993, 12, 571) and their reactivity
V
-
towards [Mo
2 5
O (OH)] . This suggests kinetic factors are important.
V
-
(45) The reactivity of O
2
2 5
towards [Mo O (OH)] is of particular interest given
that this is the oxidant used in the industrial process (eq 7). Reaction of
V
-
[Mo
2
O
5
(OH)] (m/z 289) with dioxygen proceeded slowly (φ ≈ 0.004),
forming a product at m/z 321, assigned to the ion-molecule complex
V
-
[Mo O (OH)sO ] . This ion could not be mass selected in the ion trap
2 5 2
instrument, presumably due to its ‘fragile’ nature (consistent with the
observation of ‘peak fronting’ for this species, see: McClellan, J. E.;
Murphy, J. P., III.; Mulholland, J. J.; Yost, R. A. Anal. Chem. 2002, 74,
402).
- -
2 6 2 3 2 6 2 3
O (OCH CH )] and 0.4% for [Mo O (OCH CD )] ,
products for [Mo
suggesting a crude estimate for the KIE of this reaction of 6.
3390 J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003

Oxidation of Methanol to Formaldehyde
A R T I C L E S
Figure 8. Reaction of [Mo2O5(OH)]- with (a) CH3NO2 (filled circles) and
(b) CH3OH (filled circles). Note the deviation from pesudo first order
kinetics in both plots: approximately 20% of the total ion count is unreactive
toward both these neutral reagents. The solid lines represent theoretical fits
based on the experimentally determined rate constant and amount of
unreactive component.
methanol leads to the formation of a new ion at m/z 303,
V
-
assigned to [Mo 2O5(OCH3)] (Figure 1, cycle 2, reaction 4;
Figure 3c). Isotope labeling experiments were consistent with
binding of methoxo ligand from methanol and elimination of
hydroxo ligand as neutral water (eq 4a-d, Fig. S9). These results
mirror those observed for reaction of [Mo 2O6(OH)] with
methanol (eq 1a-d), suggesting similar mechanisms of reaction
and supporting the presence of ligand hydroxo in both species
Figure 7. Collisional activation of selectively deuterated [Mo2O6-
-
(
OCH2CH3)] centers. Identical collisional activation conditions were used
-
in each experiment. (a) Collisional activation of [Mo2O6(OCH2CH3)] (m/z
3
at m/z 305 is the product of elimination of CH2CH2 to form [Mo2O6(OH)] ,
whereas the peak at m/z 317 is the product of reaction of m/z 289 with
CH3CH2OH. (b) Collisional activation of [Mo2O6(OCH2CD3)] (m/z 336)
-
VI
-
33) to eliminate CH3CHO and form [Mo2O5(OH)] (m/z 289). The peak
-
-
-
to eliminate CD3CHO and form [Mo2O5(OH)] (m/z 289). The peak at m/z
V
-
3
20 is the product of reaction of m/z 289 with CD3CH2OH. (c) Collisional
[Mo O (OH)] (m/z 289) + CH OHf
2 5 3
-
activation of [Mo2O6(OCHDCH3)] (m/z 334) to eliminate CH3CDO or
CH3CHO and form [Mo2O5(OH)] (m/z 289) or [Mo2O5(OD)] (m/z 290),
respectively. The peak at m/z 305 is the product of elimination of CHDCH2
V
-
-
-
[Mo O (OCH )] (303) + H O (4a)
2 5 3 2
-
to form [Mo2O6(OH)] , whereas the peak at m/z 318 the product of reaction
V
-
[
Mo O (OH)] (289) + CH OD f
2 5 3
of m/z 289 and m/z 290 with CH3CHDOH.
V
-
[
Mo O (OCH )] (303) + DOH (4b)
2 5 3
component, together with the experimentally determined rate
constant, kexp (Figure 8a). This reaction occurred at 70% of the
collision rate (φ ) 0.70). The apparent presence of an unreactive
form is discussed in more detail later
V
2
-
[
Mo O (OH)] (289) + CD OD f
5
3
V
-
[Mo O (OCD )] (306) + DOH (4c)
2
5
3
V
-
VI
-
V 16
16
-
18
[
Mo O (OH)] + CH NO f [Mo O (OH)] + CH NO
2 5 3 2 2 6 3
[Mo 2 O ( OH)] (289) + CH OH f
5 3
(3)
V 16
18
-
16
[
Mo 2 O ( OCH )] (305) + H O (4d)
5 3 2
Reaction 3 closes cycle 1 (Figure 1). Using the multistage
trapping capabilities of the ion trap instrument it has been
possible to carry out the complete cycle 1 three times on the
same population of ions (Fig. S2). This establishes cycle 1 as
truly catalytic for the oxidation of primary and secondary
alcohols to aldehydes and ketones, respectively.
-
[Mo O (OH)] (generated via the sequence eqs 11 and 18)
2
5
reacted rapidly with methanol (Figure 8b). Deviation from
smooth first-order kinetics was again evident and the qualitative
observations are similar to those for reaction of this ion with
nitromethane, i.e., 20% was unreactive (Figure 8a).46 After
correction for this unreactive component, smooth first-order
-
Cycle 2, Reaction 4: Interaction of [Mo2O5(OH)] with
V
-
Methanol. [Mo 2O5(OH)] is formulated to contain an hydroxo
(
46) An upper bound can be placed on the rate constant for reaction of the
ligand, and so its reactivity toward alcohol was of interest. Mass
-
2 6
unreactive component of [Mo , O , H] with nitromethane and methanol:
V
-
-14
3
-1 -1
selection of [Mo 2O5(OH)] (m/z 289) and reaction with
k
exp < 5 × 10
cm molecule
s .
J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003 3391

A R T I C L E S
Waters et al.
kinetics were observed for at least 4 half-lives, with the reaction
occurring at the collision rate (φ ) 1.0).
scale and concentrations of the experiments (Figure 8a,b;
reactions 3, 4). This could be due to the presence of (i) excited
‘hot’ ions, of greater reactivity relative to ground-state ions, or
(ii) isomers of differing reactivity.
While the mechanisms of reaction of [MoVI2O6(OH)]- and
V
-
[
Mo 2O5(OH)] with methanol seemed similar, the rates of
VI
-
reaction differed markedly: [Mo 2O6(OH)] reacted at only
V
-
Given that the ion assigned to [Mo 2O5(OH)] was synthe-
5
% of the collision rate (φ ) 0.049), whereas the reactive
VI
-
sized via collisional activation of [Mo 2O6(OBu)] (eq 11, 18),
the generation of ‘hot’ ions was a possibility. However, the high
background pressure of helium within the ion trap relative to
the pressure of reactive neutrals ensured that ions underwent a
large number of thermalizing collisions with helium before
V
-
component of [Mo 2O5(OH)] reacted at the collision rate (φ
)
V
-
1.0). The increased reactivity of [Mo 2O5(OH)] can be
attributed to (i) reduced coordinative saturation and/or (ii)
increased basicity of the hydroxo ligand due to the lower oxi-
V
-
dation level of the binuclear center (weaker acid). [Mo 2O5(OH)]
49
collision with reactive neutrals. This suggested that the
also reacted with other alcohols ROH in an analogous reaction
presence of ‘hot’ ions is an unlikely explanation for the unusual
-
to form [Mo2O5(OR)] and eliminate neutral water (Figure S10).
V
-
reactivity of the [Mo 2O5(OH)] sample. The presence of
isomeric forms generated by competitive mechanisms of alde-
V
-
The product ion from reaction of [Mo 2O5(OH)] with
V
-
VI
alcohol, [Mo 2O5(OR)] , differs in stoichiometry from [Mo 2O6-
OR)] by an oxygen atom. This suggested comparison of the
VI
-
hyde elimination from [Mo 2O6(OBu)] or by subsequent
isomerization seemed a more plausible interpretation. The ion
-
(
behavior of these ions upon collisional activation. The com-
petition between aldehyde and alkene elimination from
V
-
previously assigned as [Mo 2O5(OH)] (m/z 289) is relabeled
Mo2, O6, H] to emphasize its stoichiometry and the uncertainty
-
[
VI
-
[
Mo 2O6(OR)] was described above (eqs 2, 15-19). A
50
of its structure.
V
-
dramatic difference in reactivity is observed for [Mo 2O5(OR)] ,
which undergoes heterolytic cleavage of the O-R bond of the
alkoxo ligand and elimination of an alkyl radical (eq 20, Figure
S11).47 The inability of [Mo 2O5(OR)] to eliminate aldehyde
reinforces the importance of the stronger oxidizing power of
the dimolybdate(VI) center
Previous theoretical studies suggest possible mechanisms for
-
the formation of [Mo2, O6, H] isomers. Allison and Goddard
used generalized valence bond theory to study the interaction
of methanol with a range of oxomolybdenum(VI) species.1
They concluded that the most likely mechanism for elimination
of formaldehyde from binuclear oxomolybdenum (VI) species
containing a methoxo ligand was transfer of hydrogen from the
methoxo ligand to an oxo ligand of the neighboring metal center.
However, they also suggested that direct abstraction of hydride
by the molybdenum(VI) center was only slightly less favored
at 300 °C, and could not be excluded as a possibility. Weber
examined theoretically the elimination of formaldehyde from
idealized fragments of metal-oxo clusters using extended H u¨ ckel
V
-
4a
-
-
•
[
Mo O (OR)] f [Mo O ] + R
(20)
2
5
2
6
V
-
Cycle 2, Reaction 5: Reaction of [Mo 2O5(OCH3)] with
V
-
CH3NO2. Mass selection of [Mo 2O5(OCH3)] (m/z 303) and
reaction with nitromethane resulted in formation of a new
product ion with a mass increase of 16 Da, assigned as
Mo 2O6(OCH3)] (m/z 319) (eq 5, Figure 1, Cycle 2, reaction
; Figure 3d). This reaction proceeded at 61% of collision rate
φ ) 0.61). This rate can be compared with that for reaction of
VI
-
[
5
(
[
1
4b
theory. It was concluded that hydride transfer to the metal
center was the first step in the elimination of formaldeyde and
was followed by transfer of this hydrogen to an oxo ligand to
form an hydroxo ligand.
V
-
Mo 2O5(OH)] with CH3NO2 (eq 4; φ ) 0.70). The slight
decrease in rate for [Mo 2O5(OCH3)] may arise from the
V
-
These studies suggested possible mechanisms for elimination
greater steric demands of the methoxo ligand relative to an
-
of aldehyde from [Mo2O6(OCHR2)] and formation of different
V
-
hydroxo ligand. In contrast with [Mo 2O5(OH)] (Figure 8a),
[
-
isomers of [Mo2, O6, H] : (i) transfer of hydrogen from the
V
-
Mo 2O5(OCH3)] reacted with nitromethane under smooth
-
R-position to an oxo group and 2e reduction of the dimolybdate
pseudo first-order kinetics. This is consistent with the reactive
center, (ii) transfer of hydride from the R-position directly onto
a metal center, or (iii) transfer of hydride from the R-position
directly onto a metal center followed by subsequent transfer of
V
-
component only of [Mo 2O5(OH)] reacting with methanol to
V
-
form [Mo 2O5(OCH3)] in reaction 4 of cycle 2
-
V
-
this hydrogen to an oxo unit and 2e reduction of the
[
Mo O (OCH )] (m/z 303) + CH NO f
2 5 3 3 2
dimolybdate center. Although the isotope labeling experiments
of eqs 2 and 15 did not distinguish between these mechanistic
pathways, the kinetic data (Figure 8) suggested the formation
VI
-
[
Mo O (OCH )] (319) + CH NO (5)
2 6 3 3
Reaction 5 closes cycle 2 (Figure 1). Using the multistage
trapping capabilities of the ion trap instrument it has been
possible to carry out the complete cycle 2 twice on the same
population of ions (Figure S3). This establishes cycle 2 as truly
catalytic for the oxidation of primary and secondary alcohols
to aldehydes and ketones, respectively.4
-3
(
49) The pressure of helium within the ion trap (PHe ≈ 1.75 × 10 Torr) is
about 4 orders of magnitude higher than the pressures of neutral reactants
-7
used in these experiments (typically, PNeut ≈ 1.5 × 10 Torr). Thus each
4
ion undergoes an average of approximately 10
collisions with helium before
collision with reactive neutrals. This would appear to be sufficient to cool
-
2 6
the ions and provide a thermalised population of [Mo , O , H] ions. To
8
test this proposition, the rate of reaction was measured over a 10-fold range
of methanol pressure, effectively varying the average number of collisions
with helium before collision with methanol. Reaction rate was found to be
independent of methanol pressure.
-
Nature of the Unreactive Component of [Mo2O5(OH)] .
V
-
Approximately 20% of the sample assigned to [Mo 2O5(OH)]
did not react with either nitromethane or methanol on the time
-
(O Bu)]-
n
(
50) In these experiments, [Mo
eq 11, 18). Alternatively, [Mo
and 2 of cycle 1 (Figure 1). An unreactive component of [Mo
2
, O
6
, H] is generated via CID of [Mo
2
O
6
-
(
2
, O
6
, H] can be generated via reactions 1
-
2
6
, O , H] is
(
47) The threshold voltage for this reaction is approximately 0.54 Vp-p. This
also observed using this method of synthesis, accounting for between 10
and 20% of the total ion current. The proportion of this unreactive
-
3
)]
can be compared with that required for dissociation of [Mo
Figure 6).
48) The 20% of [Mo
and nitromethane represents a 'sink’ in the catalytic cycles of Figure 1.
2
O
6
(OCH
(
component is expected to be sensitive to the nature of the alkoxo ligand of
, H]- sample that is unreactive towards methanol
[Mo
)] and the degree of collisional activation used, but no
-
(
2
, O
6
2
6
O (OCHR
2
attempts have been made to quantify the influence of these factors.
3392 J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003

Oxidation of Methanol to Formaldehyde
A R T I C L E S
Table 2. Reaction Rates for Reaction of [MVI2O6(OH)]- (M ) Cr,
Mo, W) with CH3OH
a
reaction efficiency, φb
ion
k
exp
[
[
[
Cr2O6(OH)]-
< 0.005
6.9
78
< 0.00003
0.049
0.57
-
Mo2O6(OH)]
-
W2O6(OH)]
a
10-11 cm molecules s-1. b φ ) kexp/kado.
3
-1
×
discussed in more detail in the next section in the context of
comparisons with the analogous chromium and tungsten centers.
Comparison of the Reactivity of Dimolybdate Ions with
Dichromate and Ditungstate Analogs. Dichromate can be
generated and trapped in the gas phase via ESI of CH3CN
solutions of (Bu4N)2[Cr2O7]. However, the ditungstate dianion
5
5
has not been isolated in the condensed phase. It can be
generated and trapped in the gas phase via ESI of CH3CN
solutions of (Bu4N)2[WO4] and is presumably formed via
56
condensation of two monotungstate centers (eq 21). Sufficient
-
quantities of protonated ditungstate [W2O6(OH)] and the ion-
Figure 9. Predicted structural isomers for the ion of stoichiometry [Mo2,
+
2- -
pair {Bu4N [W2O7] } could be generated for subsequent gas-
phase experiments. Consequently, comparison of reactivities of
the chromium, molybdenum and tungsten congeners was
possible for reactions relevant to the catalytic cycles of Figure
1
-
V
O6, H] (a) a di-µ-oxo-Mo 2 species with a terminal hydroxo ligand; and
VI
(
b) a µ-oxo-Mo 2 species with a terminal hydride ligand.
-
of two distinct [Mo2, O6, H] isomers. Density functional theory
(
DFT) calculations have been used to investigate the relative
-
stabilities of possible isomers of [Mo2, O6, H] generated via
these mechanisms.
5
1
2 [WO4]2- + 2 H f [W O ] + H O
+
2-
(21)
2
7
2
-
The most stable isomer of [Mo2, O6, H] is predicted to be
V
a di-µ-oxo-Mo 2 species with a terminal hydroxo ligand and a
formal Mo-Mo bond, [Mo2 O5(OH)] (Figure 9a). However,
a µ-oxo-Mo 2 species featuring a terminal hydrido ligand
[
energy (Figure 9b). Mixed valence Mo Mo species are all
predicted to be at least a further 12 kcal/mol higher in energy
Reaction 1. This first step in cycle 1 involves reaction of
V
-
-
[Mo O (OH)] with methanol (Figure 1). The rate of this
2
6
VI
-
2 6
reaction was compared with those for [Cr O (OH)] and
VI
-
-
HMo 2O6] is predicted to be only 0.3 kcal/mol higher in
2 6
[W O (OH)] (Table 2).
VI
IV
-
2 6
[Cr O (OH)] did not react with methanol on the time scale
and methanol concentrations used in these experiments, allowing
5
2
V
-
(Fig. S12). The most stable isomer, [Mo 2O5(OH)] (Figure
an upper bound to be placed on its rate constant, i.e.,
9
a), possesses the Mo-OH unit implicated in reaction with
[Cr O (OH)]- reacted with methanol less than once in every
2
6
methanol and is reduced, suggesting reduction of nitromethane
is possible. Thus, in line with its previous assignment,
30 000 collisions (φ < 0.000 03). In comparison, [Mo O (OH)]-
2
6
reacted about once in every 20 collisions (φ ) 0.049),
representing a minimal 3 orders of magnitude increase in rate.
V
-
[
[
Mo 2O5(OH)] is proposed to be the reactive component of
-
-
Mo2, O6, H] , formed via abstraction of an R-hydrogen of the
[W O (OH)] is a further order of magnitude more reactive than
2
6
alkoxo ligand by an adjacent oxo ligand. The second most stable
[Mo O (OH)]-. Consequently, the first step of cycle 1 is not
2 6
VI
-
-
isomer, [HMo 2O6] (Figure 9b), does not contain a Mo-OH
unit and the metal atoms are in their highest oxidation state,
suggesting reduction of nitromethane is less likely. This isomer
observed for [Cr
2
O6(OH)] and proceeds most rapidly for
[W O (OH)]- (Table 2).
2
6
The isotope labeling experiments for [Mo O (OH)]- with
2
6
-
is proposed as the unreactive component of [Mo2, O6, H] ,
methanol (eq 1a-d) and reactions with alcohols of varying
acidity (Table 1) were consistent with a mechanism involving
deprotonation of the alcohol and elimination of hydroxo ligand
as neutral water. This suggested that changes in basicity of the
formed via hydride abstraction by the metal center of an
R-hydrogen of the alkoxo ligand.5
3,54
These conclusions are
(
(
(
51) DFT has been used previously to study the related [M
2
O
7
]2 anions (M )
-
-
M-OH units in [M2O6(OH)] (M ) Cr, Mo, W) would affect
Cr, Mo, W): (a) Amado, A. M.; Ribeiro-Claro, P. J. A. J. Mol. Struct.
THEOCHEM), 1999, 469, 191 (b) Bridgeman, A. J.; Cavigliasso, G. J.
(
reaction rate. The rate of reaction 1 increased down the group:
Phys. Chem. A 2001, 105, 7111.
52) These calculations address only the thermodynamic stability of possible
Cr , Mo < W (Table 2). This correlates with a decrease in the
-
isomeric forms of [Mo
2
, O
6
, H] . Given that these species are generated
VI
VI
VI
Sanderson electronegativity Cr (3.37) > Mo (2.20) > W
via collisional activation, kinetic considerations are also likely to be
important in determining the relative populations of these isomers.
(1.67) and a consequent increase in M-OH basicity.57
-
VI
-
53) The component of [Mo
2 6 2
, O , H] assigned as [HMo O ] is also unreactive
14 3
6
Reaction 2. This reaction is common to cycles 1 and 2
-
-1 -1
towards CH CO
3
2
H (kexp < 5 × 10
cm molecule
s
). This isomer
might be expected to react with acids via loss of H
2
. For example,
(Figure 1) and involves elimination of aldehyde from [Mo2O6-
-
-
[
HCr(CO)
3
]
reacts with alcohols ROH to form [ROCr(CO)
(Lane, K. R.; Squires, R. R. J. Am. Chem. Soc. 1985, 107, 6403).
3
]
with loss
-
+
2- -
(
OCHR2)] . Collisional activation of {Bu4N [M2O7] } ion-
of H
2
-
However, other metal hydrides are less reactive: [(CH
3
)
3
AlH] only
OH (Damrauer, R.; Krempp, M.;
Damrauer, N. H.; Schmidt, M. W.; Gordon, M. S. J. Am. Chem. Soc. 1993,
15, 5218).
54) Given the possibility of competing mechanisms of aldehyde elimination
undergoes adduct formation with CF
3
CH
2
(55) (a) Errington, R. J.; Kerlogue, M. D. Richards, D. G. Chem. Commun.
1993, 649 (b) Clegg, W.; Errington, R. J.; Fraser, K. A.; Richards, D. G.
Chem. Commun. 1993, 1105.
(56) Ditungstate species have been observed before using electrospray: see refs
6b,d.
(57) Huhey, J. E.; Keiter, E. A.; Keiter, R. L. Inorganic Chemistry, 4th ed.;
Harper-Collins: New York, 1993; pp 188-190.
1
(
-
2 6 2
O (OCHR )] , the kinetic isotope effect for elimination of
from [Mo
CH CHO from [Mo
aVerage isotope effect for these two processes.
-
3
2 6
O (OCH
2
CH
3
)] described earlier may in fact be an
J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003 3393

A R T I C L E S
Waters et al.
Table 3. Predicted Stability of [HMVI2O6]- Isomer Relative to
pairs (M ) Cr, Mo, W; cf, eq 11) allowed generation of the
respective [M2O6(OBu)]- anions and allowed exploration of
their relative reactivities (eqs 22 and 23). Both [Cr2O6(OBu)]
and [Mo2O6(OBu)] eliminated butanal upon collisional activa-
tion (eq 22) whereas [W2O6(OBu)]- underwent non-redox
elimination of butene (eq 23, Fig S13). The clean elimination
of alkene from [Mo2O6(OC(CH3)3)] was observed previously
in the absence of an R-hydrogen which prevented elimination
of aldehyde (eq 19). In contrast, the presence of an R-hydrogen
in [W2O6(OBu)] suggested that alkene elimination is favored
[M 2O5(OH)]- Isomer (kcal/mol) for [M2, O6, H] (M ) Cr, Mo, W)
V
-
-
metal
centre
reactive
proportion
[MV
O
(OH)]- a
[HMVI
O
]
- b
-
2
5
2
6
Cr
Mo
W
0.0
0.0
0.0
41.2
0.3
-21.3
100
80
0
-
a
Corresponds to structure of Figure 9a. b Corresponds to structure of
Figure 9b.
-
IV VI
Cr Cr centers were predicted to be at least 16 kcal/mol higher
over aldehyde elimination for the tungsten congener. This
contrasting behavior is consistent with the expected weaker
VI
in energy, while the µ-oxo Cr 2 center with a terminal hydrido
VI
-
ligand [HCr 2O6] was predicted to be 41 kcal/mol higher in
oxidizing power of tungsten species compared to those of
V
-
chromium and molybdenum58
energy (Figure S14). The most stable isomer, [Cr 2O5(OH)] ,
has the equivalent structure to that assigned as the reactive
component of [Mo2, O6, H] . This prediction is consistent with
VI
-
-
-
[M
O (OBu)] f [M , O , H] + CH CH CH CHO
2 6 2 6 3 2 2
the experimental observation that the chromium system reacts
completely with methanol, with no evidence of an unreactive
(
M ) Cr, Mo) (22)
-
VI
-
VI
-
component. The ion of stoichiometry [Cr2, O6, H] is assigned
[W
O (OBu)] f [W O (OH)] + CH CH CHdCH
2 6 2 6 3 2
2
V
-
tentatively to [Cr 2O5(OH)] .
(
23)
V
-
The reactivity of [Cr 2O5(OH)] can be compared with that
VI
-
V
-
V
-
-
2 6 2 5 2 5
of [Cr O (OH)] and [Mo O (OH)] . Reaction of [Cr O -
[W2O6(OCH3)] (generated via reaction of [W2O6(OH)] with
-
(
OH)] with methanol (φ ) 0.000 30) is at least an order of
magnitude faster than that of [Cr 2O6(OH)] (φ < 0.000 03).
CH3OH) does not contain a â-hydrogen. Consequently, alkene
elimination is not possible and formaldehyde elimination was
observed instead (eq 24). This reaction is analogous to reaction
VI
-
This correlates with the expected increased basicity of hydroxo
V
-
ligand and the lower steric barriers for [Cr 2O5(OH)] . The
2
of cycles 1 and 2 for the dimolybdate center (Figure 1). This
-
-
reactive component of [Mo2, O6, H] , assigned as [Mo2O5(OH)] ,
confirms that elimination of aldehyde is possible for the
ditungstate congener, but that the lower oxidizing power favors
elimination of alkene where possible
is approximately 3000 times more reactive toward methanol than
-
[
Cr2O5(OH)] . This is presumably for similar reasons to those
VI
-
which cause the oxidized [Mo 2O6(OH)] center to be at least
orders of magnitude more reactive than [CrVI2O6(OH)] (Table
VI
-
-
-
3
2
[W
O (OCH )] f [W , O , H] + H CO (24)
2 6 3 2 6 2
).
While [W2, O6, H]- cannot be generated via collisional
activation of [W2O6(OBu)] (eq 23), it can be generated via
collisional activation of [W2O6(OCH3)] (eq 24). [W2, O6, H]
The threshold voltage for elimination of formaldehyde from
-
-
[M2O6(OCH3)] (M ) Cr, W) can be compared with that
-
-
-
described earlier for [Mo2O6(OCH3)] i.e., M ) Cr (Vp-p )
0
.38 V) < M ) Mo (Vp-p ) 0.61 V) < M ) W (Vp-p ) 0.77
was unreactive toward methanol on the time scale and concen-
trations used in these experiments, allowing an upper bound to
5
9
V) (Figure 6). These experiments suggest the general trend
Cr < Mo < W, consistent with that based on the expected
oxidizing power of the respective metal centers.
-
14
3
be placed upon the rate constant (kexp < 5.0 × 10
cm
5
8
-1 -1
molecule s ). This contrasted with the rapid reaction of
[W2O6(OH)] with methanol (φ ) 0.57, Table 2). Further, the
reactive component of [Mo2, O6, H] , assigned to [Mo 2O5-
(OH)] , reacted at the collision rate with methanol. DFT
calculations predicted the most stable isomer of [W2, O6, H]
to be the µ-oxo-W 2 center with a terminal hydride ligand,
[HW 2O6] (Table 3). Mixed valence species are predicted to
be at least 19 kcal/mol higher in energy, and [W 2O5(OH)] is
predicted to be approximately 21 kcal/mol higher in energy
(Figure S15). This most stable hydride isomer has a structure
equivalent to that assigned to the unreactive component of [Mo2,
O6, H] (Figure 9b). This prediction is consistent with the
experimental observation that [W2, O6, H] is unreactive toward
methanol. This ion is now assigned as [HW 2O6] . Its stability
correlates with the weaker oxidizing power of tungsten centers.
The reactivities of the chromium and tungsten systems
-
Reactions 3 and 4. The final step of cycle 1 is the oxidation
V
-
V
-
-
V
of [Mo 2O5(OH)] with nitromethane (Figure 1). [Mo 2O5(OH)]
V
-
was also observed to react with methanol to form [Mo 2O5-
-
-
(
OCH3)] as part of cycle 2. In addition, samples of this ion
VI
contained an unreactive component (Figure 8). On the basis of
DFT calculations, the reactive and unreactive components were
assigned tentatively to [Mo 2O5(OH)] and [HMo 2O6] ,
respectively (Figure 9). Equivalent DFT calculations were
carried out for the chromium and tungsten congeners (Table
VI
-
V
-
VI
-
V
-
3
).
Reaction of [Cr2, O6, H] (formed via eq 22) with methanol
-
-
-
proceeded extremely slowly (φ ) 0.000 30). In contrast to the
molybdenum system, the reaction proceeded under smooth
pseudo first-order kinetics with no evidence of an unreactive
component. DFT calculations predicted the most stable isomer
of [Cr2, O6, H] to be the di-µ-oxo Cr 2 species with a terminal
hydroxo ligand, [Cr 2O5(OH)] (Table 3). Mixed valence
VI
-
-
V
V
-
VI
-
(assigned as [Cr 2O5(OH)] and [HW 2O6] ) toward nitro-
methane were also explored. Neither reduced nitromethane. An
upper bound can be placed on their rate constant (kexp < 5.0 ×
V
-
(
58) (a) Brownstein, S.; Heath, G. A.; Sengupta, A.; Sharp, D. W. A. Chem.
Commun. 1983, 669 (b) Heath, G. A.; Moock, K. A.; Sharp, D. W. A.;
-14
3
-1 -1
V
-
1
0
cm molecule s ). The inertness of [Cr 2O5(OH)] is
V
-
Yellowlees, L. J. Chem. Commun. 1985, 1503.
consistent with it being a poorer reductant than [Mo2 O5(OH)] ,
while that of [HW 2O6] is consistent with its maximum
oxidation state.
-
(OH)]- with
(
59) [Cr
CH
with CH
2
O
6
(OCH
3
)] cannot be generated via reaction of [Cr
2
O
6
VI
-
2-
3
OH (Table 2). Instead, this ion was generated via reaction of [Cr
I.
O ]
2 7
3
3394 J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003

Oxidation of Methanol to Formaldehyde
A R T I C L E S
for the binuclear ditungstate center (e.g., eq 23)
Comparison of Reactivity of Mononuclear Species,
-
[
MO3(OH)] (M ) Cr, Mo, W). To evaluate the importance
VI
-
-
[
Cr O (OCH CH )] f [Cr, O , H] + CH CHO (26)
of the binuclear nature of these group VI oxoanions in reactions
3
2
3
3
3
relevant to catalytic cycles 1 and 2 of Figure 1, the reactivity
VI
-
VI
-
VI
-
of the analogous mononuclear systems, [M O3(OH)] (M )
Cr, Mo, W) was examined. These ions were generated by ESI
of (Bu4N)2[MO4] (M ) Cr, Mo, W) in CH3CN.
[M O (OCH CH )] f [M O (OH)] + CH dCH
3
2
3
3
2
2
(M ) Mo, W) (27)
Neither of the first two steps of cycle 1 (Figure 1) proceeded
with the mononuclear molybdate center. The difference in
reactivity between mononuclear molybdate and binuclear di-
molybdate reinforces the importance of the second MoO3 unit
in the sequence of reactions observed in the catalytic cycles of
Figure 1.
Reaction 1. The first reaction of cycle 1 involves reaction
-
-
of [Mo2O6(OH)] with methanol. Each of the [MO3(OH)] ions
was inert to reaction with methanol on the time scale and
methanol concentrations used in these experiments. An upper
bound could be placed on the rate constants for their reaction
-
14
3
-1 -1 60
(
kexp < 5 × 10
cm molecule s ). This represents at
least a 3 orders of magnitude decrease over the binuclear
dimolybdate and ditungstate systems (Table 3), reinforcing the
importance of the second metal center in reaction 1 of cycle 1.
Reaction 2. Reaction 2 is the elimination of formaldehyde
Conclusions
Two gas-phase catalytic cycles are presented for the oxidation
of primary and secondary alcohols to aldehydes and ketones,
respectively, with a binuclear dimolybdate center acting as the
catalyst (Figure 1). Cycle 1 proceeds via three steps: (1)
activation of the alcohol to form an alkoxo-molybdenum bond
with elimination of water; (2) oxidation of the alkoxo ligand
and its elimination as aldehyde upon collisional activation in
the rate-determining step; and (3) regeneration of the catalyst
via oxidation by nitromethane. Cycle 2 shares reaction 2 in
common with cycle 1, but differs in the sequence of reaction
with alcohol and nitromethane.
-
from [Mo2O6(OCH3)] . Although the equivalent methoxo
mononuclear centers cannot be generated via reaction of
-
[
MO3(OH)] with methanol, they can be generated and trapped
in the gas phase via ESI of CH3CN solutions of (Bu4N)2[MO4]
M ) Cr, Mo, W) containing methanol (1% v/v). The reactivities
(
of these species upon collisional activation were explored.
-
[
CrO3(OCH3)] eliminated formaldehyde (eq 25), as observed
-
for the analogous binuclear species, [Cr2O6(OCH3)] , and is
consistent with the known strong oxidizing power of mono-
nuclear Cr(VI) centers. No reaction was observed for the
mononuclear molybdenum and tungsten centers. Attempts to
use higher collisional activation energies resulted in loss of ion
signal. This may be due to electron detachment, formation of
an ionic species whose mass is too low to be detected, or simply
loss of ions from the ion trap. Although the lack of an observed
reaction does not strictly rule out the possibility for aldehyde
elimination, it demonstrates that a higher activation energy is
The nature of these reactions was probed by kinetic measure-
ments, by variation of the substrate alcohols (structure and
isotope labeling) and by variation of the reactive center (metal
and nuclearity). The molybdenum and tungsten binuclear centers
-
[
M2O6(OH)] (M ) Mo, W) promoted reaction 1 but the
-
chromium center [Cr2O6(OH)] did not (Table 2). This is
consistent with the expected order of basicity of the hydroxo
ligand in these species. In general, the chromium and molyb-
denum centers [M2O6(OCHR2)] (M ) Cr, Mo) promoted
reaction 2, but the tungsten centers [W2O6(OCHR2)] did not.
-
-
required than that for aldehyde elimination from [CrO3(OCH3)]
-
and the relevant binuclear centers
This is consistent with the expected order of oxidizing power
of the anions.
-
-
[
CrO (OCH )] f [Cr, O , H] + H CO
(25)
3
3
3
2
Evidence was presented that the product of reaction 2 involves
-
structural isomers of stoichiometry [Mo2, O6, H] . Approxi-
The elimination of alkenes is possible for alkoxo ligands other
than methoxo. This was observed previously for the molyb-
denum binuclear centers with alkoxo ligands lacking an
R-hydrogen (eq 19) or in the case of ditungstate where alkene
elimination was favored over aldehyde elimination (eq 23). This
possibility was examined for the mononuclear centers via the
ethoxo ligands (eqs 26 and 27). The chromium center eliminated
acetaldehyde (eq 26), consistent with elimination of formalde-
-
mately 80% of the sample of [Mo2, O6, H] reacted with
nitromethane in reaction 3 (cycle 1) and with methanol in
reaction 4 (cycle 2), whereas the remaining 20% was unreactive
to both neutrals. DFT calculations suggest that at least two
structural isomers may be accessible (Figure 9). The reactive
-
V
-
form of [Mo2, O6, H] is assigned to [Mo 2O5(OH)] and the
VI
-
unreactive form to [HMo 2O6] . The experimental observa-
tions are consistent with the expected relative reactivities of
these isomers. Calculations also predicted that the reduced
-
hyde from [CrO3(OCH3)] (eq 25). The mononuclear molyb-
denum center eliminated ethene in a nonredox reaction (eq 27),
contrasting with the elimination of acetaldehyde by binuclear
Mo2O6(OCH2CH3)]- (eq 15). These differing reactivities
reinforce the importance of the second metal center in reaction
of cycles 1 and 2. The mononuclear tungsten center also
eliminated the ethoxide ligand as ethene (eq 27), as observed
V
-
[
Cr 2O5(OH)] isomer is favored for chromium, whereas the
VI
-
hydride form [HW 2O6] is favored for tungsten. These
assignments are consistent with the greater oxidizing power
expected for chromium(VI) relative to tungsten(VI), and are
[
2
V
-
consistent with the observed reactivities of [Cr 2O5(OH)] and
HW 2O6] toward methanol and nitromethane.
VI
-
[
(
60) Given the order of magnitude rate increase of CF
3
CH
2
OH over methanol
The experiments revealed that only the dimolybdate center
-
towards [Mo
of these mononuclear centres towards CF
active towards CF CH OH under our experimental conditions, while
3
(OH)] and [WO (OH)] each undergo extremely slow reaction to
2 6
O (OH)] (Table 1), we have also investigated the reactivity
-
has the mix of properties that allow it to participate in each of
the three steps required by cycles 1 and 2. Although the
dichromate congener is the strongest oxidant, it is the weakest
base, making it unreactive toward alcohols. By contrast, the
ditungstate congener is the strongest base, but its weaker
3
2 3
CH OH. [CrO (OH)] is unre-
3
2
-
-
[
MoO
form [MO CF
to be measured with the present instrument (kexp < 5 × 10
3 3 2
). Clustering reactions to produce [MO (OH)sCF CH OH]
3
-
3
)] (M ) Mo, W). The rate of reaction was too slow
-14 3
3
(OCH
2
cm
molecule-
were also observed (M ) Mo, W).
1
s
-1
-
J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003 3395

A R T I C L E S
Waters et al.
oxidizing power means it prefers the non-redox elimination of
alkene rather than aldehyde.
(v) Oxo-MoVI binuclear units on SiO2 supports were
3
prepared by reaction of Mo2(η -C3H5)4 with SiO2, followed by
-
64
Each of the mononuclear anions [MO3(OH)] (M ) Cr, Mo,
successive reaction with H and O . This method allowed
2
2
-
W) was inert to reaction with methanol. [CrO3(OCH2CH3)]
assembly of well-defined adjacent MoO units on the SiO2
4
eliminated acetaldehyde, whereas the molybdenum and tungsten
equivalents eliminated ethene in a non-redox reaction. This
behavior contrasted with the elimination of acetaldehyde from
surface. This catalyst showed an order of magnitude higher
activity than molybdate catalysts prepared via conventional
impregnation methods. The observation was rationalized in
terms of a “cooperative effect” between the two units of a
binuclear molybdate center. An ethanol molecule was proposed
to condense onto a single molybdate unit to form an ethoxo-
Mo(VI) center, followed by elimination of acetaldehyde via
abstraction of an R-hydrogen of the ethoxo ligand by the
neighboring molybdate center. Results from the present study
indicating that mononuclear molybdate centers are ineffective
but that binuclear centers are effective for the elimination of
aldehyde are consistent with this proposal.
-
binuclear [Mo2O6(OCH2CH3)] . The difference in reactivity
between the mononuclear and binuclear molybdenum anions
reinforces the importance of the second MoO3 unit in the
sequence of reactions of catalytic cycles 1 and 2.
The three reactions which constitute cycle 1 are equivalent
to the three essential steps proposed to occur in the industrial
oxidation of gaseous methanol to formaldehyde over molyb-
denum(VI)-trioxide solid-state catalysts at 300-400 °C (see
Introduction above). A number of comparisons can be made
between the catalytic cycles of the present gas phase work and
that of the industrial process:
vi) The compound MoVI2O5(OCH3)2 has been proposed as
(
11c
a model heterogeneous catalyst. Its layered structure features
(i) The rate-limiting step is elimination of aldehyde in each
VI
trans-Mo O(OCH3) units which are perpendicular to the layer
system. In the present gas phase study, reaction 2 does not
proceed at room temperature, and the use of collisional
activation is required. Similarly, no reactivity is observed at
room temperature in the industrial process, and the use of
elevated temperatures is required.
planes and are separated by in-plane bridging oxo ligands.
Neighboring units alternate in their orientation relative to the
plane. A mechanism of formaldehyde elimination similar to that
proposed for the binuclear centers in the present work is easily
VI
envisaged, involving adjacent Mo centers within one layer
(ii) The oxidation of ethanol to acetaldehyde over a variety
plane or between adjacent planes. Oxygen vacancies created
of supported molybdenum(VI)-trioxide catalysts at temperatures
VI
by formaldehyde elimination from adjacent Mo centers would
6
1
between 410 °C and 510 °C has been investigated. From a
comparison of turnover rates for CH3CD2OH and CH3CH2OH,
an isotope effect in the range of 1.7-2.2 was estimated. The
kinetic isotope effect of 1.9 ( 0.4 for the elimination of
be amenable to regeneration by the two O atoms available from
O2. The presence of adjacent oxygen-deficient units in the solid-
state catalyst should expedite reaction with O2.
-
(vii) The observation of a second catalytic cycle in the present
work differing in the sequence of reaction with methanol and
nitromethane suggests related reactions may also occur in
reverse order in the industrial process. This proposal could be
tested using temperature programmed desorption experiments.15
acetaldehyde from [Mo2O6(OCHDCH3)] in the present study
matches that estimate.
(iii) Temperature programmed desorption has been used to
study the effect of alcohol structure on the products of reaction
15
over molybdate solid-state catalysts. The only products
detected for t-BuOH were 2-methyl-propene and water from
the nonredox dehydration reaction. The elimination of 2-methyl-
Acknowledgment. T.W. acknowledges the support of an
Australian Postgraduate Award and a Melbourne Abroad Travel
Scholarship, and thanks Professor Mark Gordon and Dr. Mike
Schmidt for hosting a visit and assistance with the use of
GAMESS. T.W. and R.A.J.O. thank Professor Scott Gronert
for helpful discussions concerning rate measurements. R.A.J.O.
thanks the Australian Research Council for financial support
-
propene from [Mo2O6(OC(CH3)] observed in the present work
(eq 19) is consistent with these observations. Further, temper-
ature programmed desorption experiments suggest branched
alcohols such as i-propanol result in increased yield of the alkene
versus the aldehyde, also consistent with observations from the
present study (eq 17).
(Grant No. A00103008) and the University of Melbourne for
(iv) The catalytic activities of ferric tungstate Fe2(WO4)3 and
funds to purchase the LCQ. R.A.J.O. also thanks Professors
Peter Armentrout and Michael Bruce for interesting discussions
and helpful suggestions. Reviewers are thanked for their helpful
comments and suggestions.
ferric molybdate Fe2(MoO4)3 have been compared for the
oxidation of methanol.62 Although the reaction rates were
similar, the tungsten system produced mainly dimethyl ether
via a nonredox process, whereas the molybdenum system
evolved mainly formaldehyde via the redox process. These
observations are consistent with those from the present work
where the weaker oxidizing power of tungsten led to a
preference for alkene elimination compared to aldehyde elimi-
nation in the molydenum analogue (eq 22, 23). Polyoxotungstate
systems have also been shown to be effective catalysts for the
Supporting Information Available: Relevant mass spectra
-
(see text), Cartesian coordinates and energies of [M2, O6, H]
isomers (M ) Cr, Mo, W). This material is available free of
charge via the Internet at http://pubs.acs.org.
JA028839X
6
3
dehydration of ethanol to ethylene in a pseudo-liquid phase.
(
63) Lee, K. Y.; Arai, T.; Nakata, S.; Asaoka, S.; Okuhara, T.; Misono, M. J.
(
61) Zhang, W.; Oyama, S. T. J. Phys. Chem. 1996, 100, 10 759.
Am. Chem. Soc. 1992, 114, 2836.
(
62) Harrison, W. T. A.; Chowdhry, U.; Machiels, C. J.; Sleight, A. W.;
(64) Iwasawa, Y. In Tailored Metal Catalysts; Iwasawa, Y. Eds.; D. Reidel
Publishing Company: Holland, 1986, p 54-60 and references therein.
Cheetham, A. K. J. Solid State. Chem. 1985, 60, 101.
3396 J. AM. CHEM. SOC.
9
VOL. 125, NO. 11, 2003