Mechanistic study of the oxidative carbonylation of methanol catalyzed by palladium diphosphane complexes with nitrobenzene as oxidant

Article abstract of DOI:10.1002/ejic.201101354

The reactivity of Pd complexes having bidentate diarylphosphane ligands was studied in the oxidative carbonylation of CH₃OH to dimethyl carbonate/oxalate (DMC/O) with PhNO₂ as the oxidant. Different ligands were employed with variation in backbone length and aryl ring substituent, and the acidity, CO pressure, or the partial pressure of H ₂ was varied. At two different stages in the catalytic cycle, one equivalent of DMC/O may evolve for every equivalent of PhNO₂ reduced, which means that the efficiency with which nitrobenzene can function as the oxidant for the oxidative carbonylation of methanol (E_OC) can potentially be 200a% relative to nitrobenzene conversion. The selectivity for DMC relative to DMO is thought to be determined by a species of the type [P ₂PdC(O)OCH₃(R)]; the DMO/DMC ratio can be increased by increasing the CO pressure, by addition of an acid, or by using a ligand with a relatively large bite angle. On the basis of the collected results, we conclude that an ideal catalyst for oxidative carbonylation would have a relatively acidic palladium center and be sterically undemanding in the axial positions but sterically demanding in the equatorial positions of palladium. The Pd complex of bis(diphenylphosphanyl)ferrocene meets these criteria and was found to function most efficiently with PhNO₂ as oxidant for the oxidative carbonylation of methanol among the series of compounds studied, that is, with about 50a% of the theoretical maximum efficiency E_OC. The catalytic reactivity of palladium complexes having bidentate diarylphosphane ligands was studied in the oxidativecarbonylation of methanol to dimethylcarbonate (DMC) and dimethyl oxalate (DMO) with nitrobenzene as terminal oxidant. Nitrobenzene can be reduced to aniline by the hydrogen atoms liberated, and thus a catalytic coupling between methanol oxidation and nitrobenzene hydrogenation is established.

Full text of DOI:10.1002/ejic.201101354

FULL PAPER
DOI: 10.1002/ejic.201101354
Mechanistic Study of the Oxidative Carbonylation of Methanol Catalyzed by
Palladium Diphosphane Complexes with Nitrobenzene as Oxidant
Tiddo J. Mooibroek,[a] Elisabeth Bouwman,*[a] and Eite Drent[a]
Keywords: Palladium / Phosphanes / Dimethyl carbonate / Dimethyl oxalate / Carbonylation
The reactivity of Pd complexes having bidentate diarylphos-
phane ligands was studied in the oxidative carbonylation of
CH3OH to dimethyl carbonate/oxalate (DMC/O) with PhNO2
as the oxidant. Different ligands were employed with varia-
tion in backbone length and aryl ring substituent, and the
acidity, CO pressure, or the partial pressure of H2 was varied.
At two different stages in the catalytic cycle, one equivalent
[P2PdC(O)OCH3(R)]; the DMO/DMC ratio can be increased
by increasing the CO pressure, by addition of an acid, or by
using a ligand with a relatively large bite angle. On the basis
of the collected results, we conclude that an ideal catalyst for
oxidative carbonylation would have a relatively acidic palla-
dium center and be sterically undemanding in the axial posi-
tions but sterically demanding in the equatorial positions of
of DMC/O may evolve for every equivalent of PhNO2 re- palladium. The Pd complex of bis(diphenylphosphanyl)ferro-
duced, which means that the efficiency with which nitro-
benzene can function as the oxidant for the oxidative carb-
onylation of methanol (EOC) can potentially be 200% relative
to nitrobenzene conversion. The selectivity for DMC relative
to DMO is thought to be determined by a species of the type
cene meets these criteria and was found to function most ef-
ficiently with PhNO2 as oxidant for the oxidative carb-
onylation of methanol among the series of compounds
studied, that is, with about 50% of the theoretical maximum
efficiency EOC
.
Introduction
methanol to form a carbamate, which in turn can be pyro-
lyzed to the isocyanate (e.g. TDI or MDI). Such (aliphatic
but also aromatic) carbonates and oxalates can be prepared
by a palladium-catalyzed oxidative carbonylation of the
alcohol (e.g. methanol to DMC and DMO).[14–31] The ter-
minal oxidant is usually O2, but a metal co-catalyst such as
Cu2+/Cu+[24,26] or Pb4+/Pb2+[25] is often necessary to reoxid-
ize palladium, which is difficult to oxidize with molecular
oxygen. Notably, alkyl nitrites (RON=O) have been used to
replace O2 as the oxidant in the synthesis of DMC.[27]
Working with strong oxidants such as O2 and RON=O
is not too problematic when the palladium catalyst is stabi-
lized by N- or O-donor ligands, but nitrogen and oxygen
are generally poor ligands for palladium, since they cannot
be involved in π back-bonding. P-donor ligands, on the
other hand, are very good ligands for palladium for that
very reason, but they easily react with strong oxidants such
as O2 and RON=O to form phosphane oxides leading to
catalyst degradation. It is therefore not surprising that
carbonate and oxalate syntheses employing P-donor ligands
are mainly stoichiometric with regard to palla-
dium.[14,15,20,22,23] For this reaction to be catalytic in palla-
dium, milder oxidants have been considered, such as 1,4-
benzoquinone[24–26,28,32] and NaNO2/NaNO3,[32] but the co-
products that are then obtained stoichiometrically on the
carbonate/oxalate cannot easily be utilized on a large scale.
Notably, for the oxidative carbonylation of phenols to
diphenyl carbonate (DPC), nitrobenzene was proposed as
the oxidant. Demonstrated yields, however, were less than
One of the challenges in current day catalysis is to re-
place wasteful and dangerous industrial processes by more
environmentally friendly and safer ones. An example of
such a challenge is to replace the highly toxic and corrosive
phosgene,[1,2] which is often used as carbonylating agent.
For example, in the synthesis of aromatic isocyanates[3,4]
such as TDI and MDI,[5,6] phosgene is employed on the
megaton scale[7] to carbonylate a reduced nitroaromatic
compound (Scheme 1).
Scheme 1. Two industrially important aromatic isocyanates (MDI,
TDI) and two carbonylating reagents (phosgene, DMC).
Dimethyl carbonate (DMC),[8–12] and to a lesser extent
dimethyl oxalate (DMO),[13] have been proposed to replace
phosgene as carbonylating agent (Scheme 1), which, by
transesterification with an aromatic amine, will liberate
[a] Leiden Institute of Chemistry, Leiden University,
P. O. Box 9502, 2300 RA Leiden, The Netherlands
Fax: +31-71-527-4671
E-mail: bouwman@chem.leidenuniv.nl
Supporting information for this article is available on the
WWW under http://dx.doi.org/10.1002/ejic.201101354.
Eur. J. Inorg. Chem. 2012, 1403–1412
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1403

T. J. Mooibroek, E. Bouwman, E. Drent
FULL PAPER
Scheme 2. A new synthetic route to prepare aromatic isocyanates (in the carbamate form) from dimethyl carbonate and aniline, formed
by the oxidative carbonylation of methanol using nitrobenzene as the terminal oxidant.
stoichiometric on palladium.[33] To the best of our knowl- anol oxidation,[34] and we have focused on the mechanism
edge, nitrobenzene has never before been used or proposed of formation of the nitrobenzene reduction products else-
as oxidant for the preparation of DMC or DMO. Further- where.[35]
more, nitrobenzene may be reduced to aniline by the hydro-
The focus of this paper is on the genesis of methanol
gen atoms liberated by the formation of DMC/DMO, thus oxidation products, in particular the useful oxidative carb-
establishing a catalytic coupling between methanol oxi- onylation products DMC and DMO, and how the pro-
dation and nitrobenzene hydrogenation chemistry. This po- duction of and selectivity for these products depends on the
tentially unlocks a new procedure to prepare the reactants structure of the catalyst and on reaction conditions.
aniline and DMC, which can be used to make aromatic
isocyanates via methyl phenyl carbamate (Scheme 2).
Results
Furthermore, the product DMO is an industrially interest-
ing intermediate for the sustainable production of mono-
ethylene glycol (MEG) solely from synthesis gas as feed-
stock, which can be produced from practically any carbon
source, including coal, gas, bio-waste or even CO2.
General Considerations
In the catalytic reactions for the oxidation of methanol
and the concomitant reduction of nitrobenzene, the catalyst
The reductive carbonylation of nitrobenzene in methanol was formed in situ from Pd(OAc)2 and a bidentate phos-
forming methyl phenyl carbamate is a reported alternative phane ligand (1:1.5). Complex formation is instantaneous
for the industrial preparation of aromatic isocyanates such in methanol with the ligands used in this study,[36] and iden-
as TDI and MDI, which we are actively studying with pal- tical results were obtained when the activity of selected pre-
ladium–diphosphane catalytic systems. In these studies, we formed complexes was tested. It was ensured that methanol
found that, for some diphosphane-supported palladium cat- and nitrobenzene were anhydrous by thoroughly drying
alysts, apart from the desired nitrobenzene carbonylation these liquids. In the initial screening studies, a large variety
product large amounts of DMC, DMO, and aniline were of diarylphosphane ligands have been used, with variations
formed.[34,35] It became clear that in our catalytic system in the substituents on the phenyl rings as well as in the
nitrobenzene reduction is catalytically coupled with meth- length and flexibility of the backbone spacer.
Scheme 3. Overview of the different products that are formed in the palladium-catalyzed reaction of nitrobenzene with CO in methanol.
Figure 1. Overview of the ligands used in this study.
1404
www.eurjic.org
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2012, 1403–1412

Oxidative Carbonylation of Methanol with Nitrobenzene as Oxidant
In the carbonylation of nitrobenzene with diphosphanyl- The table also gives the conversion of nitrobenzene, the
palladium catalyst systems, the products of the oxidative amount of nitrobenzene that is reduced to a “PhN-contain-
carbonylation of methanol, DMC and DMO, as well as ing” fragment (column ΣPhNO2), and the efficiency with
those of methanol oxidation, methyl formate (MF) and car- which a certain catalytic system can use nitrobenzene as
bon monoxide (CO), are formed (Scheme 3). We could ex- oxidant for the oxidative carbonylation of methanol to
perimentally establish the generation of gaseous CO in an DMC and DMO (column EOC, see discussion for details).
estimated amount that satisfied the overall hydrogen mass
balance between methanol oxidation (“H-liberating”) and the bar diagrams shown in this paper. The mechanism of
nitrobenzene reduction (“H-consuming”) products.[34]
formation of all phenyl-containing reduction products of
The “PhN-containing” reduction products of nitrobenz- nitrobenzene is discussed in a separate publication.[35]
ene are methyl phenyl carbamate (MPC), N,NЈ-diphenyl- Note that DMC, DMO, and MF can be easily quantified
The data reported in this table have been used to generate
urea (DPU), aniline, azobenzene (Azo), and azoxybenzene by using GLC analysis, but the amount of CO produced
(Azoxy). Other reduction products of nitrobenzene are CO2 had to be deduced from the hydrogen mass balance.[34]
and H2O, both containing one oxygen atom from
PhNO2. H2O, aniline, and DPU are derived from methanol
Oxidative Carbonylation of Methanol to DMC or DMO
oxidation, as the (OH, NH) hydrogen atoms in these mole-
cules originate from methanol.[34]
A comparison of the amount of DMC/DMO produced
The selectivity of the catalysts for the various products is
highly dependent on the structure of the ligand. The trends
in reactivity and selectivity will be discussed by using the
ligands shown in Figure 1. These ligands have either a pro-
pylene (L3), butylene (L4), or ferrocene (L5Fc) backbone,
which in some cases is made more rigid by substitution
(indicated by X). The aryl rings of the ligands can be func-
tionalized with methoxy moieties in the ortho or para posi-
tion (o-MeO– or p-MeO–) to allow discrimination of steric
from electronic effects.
by the various catalysts bearing the ligands L3, L4X, and
L5Fc with the available ortho-methoxy and para-methoxy
analogues is shown in Figure 2. The amount of DMC and
DMO produced is highest when employing the unsubsti-
tuted ligands L3, L4X, and L5Fc (left). When the o-MeO
analogues of these ligands were employed, the production
of DMC and DMO was significantly suppressed (middle).
Although this effect does not seem to hold when comparing
the oMeO–L4X and L4X systems, it must be noted that
twice as much nitrobenzene is converted when using
oMeO–L4X (90%) as opposed to L4X (52%) as supporting
ligand. The suppression of DMC and DMO production is
considerably less pronounced when employing ligands with
para-methoxyphenyl rings (right), indicating that the effect
must be predominantly steric in nature. For example, the
Ligand Effects in the Oxidation of Methanol
General Comments
The quantities (in mmol) of the C-containing oxidation catalytic systems employing L4 and pMeO–L4 produce 8.0
products of methanol and H-containing reduction products and 5.4 mmol of DMC plus DMO at a nitrobenzene con-
of nitrobenzene formed in nitrobenzene carbonylation ex- version of 60 and 40%, respectively.
periments are given in Table 1. A full analysis of the reac-
The selectivity for either DMC or DMO appeared to
tion mixtures was always performed, however, and the data vary significantly as a function of the bite angle (β) of the
for the other reaction products are available in Table S1. ligand used, as indicated by the increasing DMO/DMC ra-
Table 1. Results of the reactions of nitrobenzene with CO in methanol, catalyzed by a variety of in situ formed PdII(ligand) complexes.[a]
[b]
Entry
Ligand
Conv. PhNO2
(%)
“C-containing” oxidation products
(mmol)
“H-containing” reduction products
(mmol)
ΣPhNO2 reduced EOC
(mmol)
(%)
MF
DMC
DMO
H2O
PhNH2
DPU
1
2
3
4
5
6
7
8
L3
L3X
oMeO–L3
oMeO–L3X
pMeO–L3
L4
67
67
53
98
54
60
52
90
40
70
100
0.1
0.2
0.6
1.1
0.1
0.3
0.2
0.6
0.2
0.4
1.8
3.6
4.2
0.6
0.4
2.2
2.4
2.3
2.1
1.7
4.3
2.5
3.3
3.1
0.5
0.4
2.1
5.6
5.9
7.3
3.7
13.7
8.0
3.0
3.4
n.d.
0.7
n.d.
3.9
10.0
8.7
n.d.
n.d.
n.d.
5.1
8.3
4.1
5.7
3.9
5.4
1.9
10.5
5.9
7.2
11.2
1.7
0.8
0.9
3.1
1.9
0.8
0.5
1.9
0.0
0.2
2.6
15.4
16.2
12.9
23.9
12.9
14.6
12.6
22.1
9.1
45
45
10
5
30
55
65
45
60
105
45
L4X
oMeO–L4X
pMeO–L4
L5Fc
9
10
11
17.4
24.4
oMeO–L5Fc
[a] Reaction mixtures were heated for four hours at 110 °C in 25.0 mL of dry and degassed methanol under 50 bar CO pressure. The
catalyst was always generated in situ from 0.05 mmol Pd(OAc)2. Mole ratios are: Pd(OAc)2/Ligand/nitrobenzene = 1:1.5:488. Quantities
are reported in mmol and “n.d.” stands for “not determined”. [b] EOC: estimated efficiency of the oxidative carbonylation of methanol
in percentages based on nitrobenzene conversion: EOC = (DMC + DMO)/(PhNO2) fully reduced = PhNH2 + MPC + 2[DPU +
Azo(xy)]ϫ100%.
Eur. J. Inorg. Chem. 2012, 1403–1412
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
1405

T. J. Mooibroek, E. Bouwman, E. Drent
FULL PAPER
Figure 2. The influence of selected ligands on the selectivity and productivity of the catalysts for DMC (᭿) and DMO (ᮀ). The DMO/
DMC ratio is given at the top; the conversion of nitrobenzene is given in parentheses.
tio from approximately 0.75 to about 2.5–3.1 when using
Upon addition of 5 equiv. (on Pd) of TMBA to the cata-
the ligands L3X (β ≈ 90°) and L4X/L5Fc (β ≈ 96°), respec- lytic system comprising L3, the production of DMC and
tively. The DMO/DMC ratio does not significantly change DMO decreased from 6.9 to 2.6 mmol and the conversion
when the supporting ligand in the catalyst is functionalized of nitrobenzene also decreased (from 68 to 50%). For the
with electron-donating methoxy groups in either the ortho system PdII(oMeO–L3X), the effect was similar, although
or para position.
less pronounced as a result of the rather low oxidative carb-
onylation activity. When adding 10 equiv. (on Pd) of the
base DMAN, the opposite occurred: the production of
DMC and DMO increased from 6.9 to 11.9 mmol and the
nitrobenzene conversion increased from 68 to 91% for the
catalytic system based on L3. Again, a similar but less pro-
nounced effect was observed for PdII(oMeO–L3X). The
DMO/DMC ratio decreased when going from the acidic to
the basic system: from 1.2 to 0.8 when using PdII(L3) and
from 1.0 to 0.7 when employing PdII(oMeO–L3X).
The Effects of Reactants and Additives in the Oxidation of
Methanol
Effect of the Acidity of the Reaction Medium
To investigate the effect of acidity on the product forma-
tion, a series of experiments was conducted wherein 5 equiv.
(on Pd) of 2,4,6-trimethylbenzoic acid (TMBA) or 10 equiv.
of Proton-sponge [1,8-bis(dimethylamino)naphthalene,
DMAN] were added. The catalyst of choice for these ex-
periments is PdII(L3), as this catalyst proved to be relatively
active for the formation of the products of the carb-
onylation of methanol (DMC + DMO ≈ 7 mmol), but
Effect of the Concentration of CO and H2
The CO pressure was varied in a series of catalytic runs
rather non-preferential with respect to the DMO/DMC ra- with the complexes PdII(L3) and PdII(L4). The results of
tio (ca. 1). For comparison, the o-MeO-functionalized cata- these experiments are shown in Figure 4. When increasing
lyst PdII(oMeO–L3X) was used. The results are shown in the CO pressure from 25 to 50 to 100 bar, the DMO/DMC
Figure 3.
ratio increased from 0.3 to 0.9 to 2.0 when using PdII(L3)
Figure 3. The selectivity and productivity for DMC (᭿) and DMO (ᮀ) as a function of the acidity (adding acid or base) for catalysts
with the ligand L3 (left) and oMeO–L3X (right). The DMO/DMC ratio is shown at the top; the conversion of nitrobenzene is given in
parentheses.
1406
www.eurjic.org
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2012, 1403–1412

Oxidative Carbonylation of Methanol with Nitrobenzene as Oxidant
Figure 4. The selectivity and productivity of catalysts PdII(L3) and PdII(L4) for DMC (᭿) and DMO (ᮀ) as a function of CO pressure.
The DMO/DMC ratio is given at the top; the conversion of nitrobenzene is given in parentheses.
Figure 5. The selectivity and productivity for MF (᭿) and [DMC + DMO] (ᮀ) as a function of the partial pressure of H2 (total = 50 bar)
for catalysts comprising a ligand with a propylene backbone (left) or a butylene backbone (right). The MF/(DMC + DMO) ratio is given
at the top; the nitrobenzene conversion is given in parentheses.
and from 1.2 to 2.5 to 5.0 when using PdII(L4). The conver-
sion of nitrobenzene roughly doubled for both catalyst sys-
Discussion
tems with increasing CO pressures (see also the Supporting
Information).
Overall Mechanism of PhNO2 Reduction and CH3OH
Oxidation with P2Pd Catalysts
When applying higher CO pressures, the total amount of
DMC and DMO produced significantly increased when
During our studies on the reductive carbonylation of
using L4 (5.7 vs. 9.6 mmol at 25 and 100 bar CO), in line nitrobenzene in methanol, we found that nitrobenzene re-
with the increased nitrobenzene conversion. For L3, the ef- duction chemistry is catalytically linked with methanol oxi-
fect of increasing the pressure on the amount of DMC and dation chemistry by two sets of half-reactions.[34,35] One set
DMO produced is less significant; 6.2 vs. 7.4 mmol at 25 of half-reactions describes the oxidation of Pd0 to
and 100 bar CO, whereas the nitrobenzene conversion is PdII=NPh [Equations (1a)–(1c)], while the complementary
nearly doubled.
set describes the reduction of P2PdII=NPh to Pd0 [Equa-
Several catalytic runs were performed under an atmo- tions (2a)–(2d)] in order to make the reactions catalytic (see
sphere of H2 and CO (15 and 35 bar, respectively; see Fig- Scheme 4; the exact stoichiometries are given in the Sup-
ure 5). As expected, the presence of hydrogen resulted in porting Information). Combining the two sets of half-reac-
most cases in an increase in aniline formation. In particular, tions leads to the overall possible stoichiometries and al-
the catalyst supported with the ligand oMeO–L3X led to a lows the construction of the relatively simple and unifying
high yield of aniline (22.6 mmol, 90% selectivity; see Sup- catalytic scheme shown in Scheme 4.
porting Information).
Note that in this scheme nitrobenzene is always the sub-
Whereas in the absence of H2 all catalysts produce rela- strate, whereas nitrosobenzene (PhNO) may be produced as
tively small amounts of MF (0.2–1.1 mmol), under an at- an intermediate and enter the reaction described by Equa-
mosphere containing 15 bar H2 significantly more MF is tions (1a)–(1c). Traces of nitrosobenzene were indeed fre-
produced (1.8–5.9 mmol) at the expense of DMC and quently observed after a catalytic experiment. It is unlikely
DMO production, as reflected by the higher MF/(DMC + that nitrosobenzene is deoxygenated by methanol by the re-
DMO) ratios (Figure 5). The conversion of nitrobenzene is actions in Equations (1b) and (1c), as it is well-documented
only slightly affected in all cases.
that the first deoxygenation (with CO) of nitrobenzene (to
Eur. J. Inorg. Chem. 2012, 1403–1412
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
1407

T. J. Mooibroek, E. Bouwman, E. Drent
FULL PAPER
Scheme 4. Proposed reaction scheme for the overall catalytic processes.
nitrosobenzene) is considerably slower and energetically less
We have defined the efficiency of oxidative carbonylation
favorable compared to nitrosobenzene deoxygenation with (EOC) as the mol ratio of [DMC + DMO] produced to
CO.[37,38]
nitrobenzene converted {i.e. the sum of products containing
nitrobenzene “PhN” fragments [MPC PhNH2
+
+
2Azo(xy) + 2DPU]ϫ 100%}. It follows from the above that
the maximum value for EOC is 200%; that is, when the se-
quence of reactions in Equations (1b) and (2b) or (1b) and
(2d) [which are equivalent to those in Equations (3a) or
Two Pathways to Oxidative Carbonylation Products DMC
and DMO
In this paper, the focus lies on the genesis of the oxidative (3a)*] is exclusively taking place. Lower efficiencies indicate
carbonylation products of methanol, DMC and DMO. that the other reactions along the periphery of the cycles
Scheme 4 is helpful for this purpose, as it shows that DMC in Scheme 4 [Equations (1a), (1c), (2a), and (2c)] are also
and DMO can be produced at two stages (following the operative. One should note that the amounts of aniline and
bold arrows in Scheme 4).
DMC/DMO produced are not necessarily linked only
DMC/DMO production at the first stage starts with a through Equation (3a)/(3a)*. Aniline can also be formed as
P2Pd0 species with oxidation of methanol and concomitant a consecutive product [e.g. by replacing methanol with H2O
reduction of nitrobenzene according to Equation (1b) to in Equation (2a)] from water that is produced by the reac-
produce one equivalent of DMC/DMO on nitrobenzene. At tions given in Equations (1b) or (1c).
The EOC for catalytic systems based on ligands compris-
ing unsubstituted aryl rings and a C3 backbone is about
45% (Table 1). For catalytic systems based on similar li-
gands with a larger backbone (i.e. a larger bite angle), the
EOC is greater: 65% for L4X and even 105% for L5Fc. This
means that some catalysts can use nitrobenzene more ef-
ficiently as an oxidant than other catalysts for the pro-
duction of DMC/DMO.
To unlock potentially even more efficient or selective
catalytic systems for DMC/DMO production, it will be nec-
essary to comprehend the effects of the structural param-
eters of the catalyst on the efficiency of the catalysis and
thereby gain insight into the mechanistic details underlying
the reactions summarized in Scheme 4. We have discussed
our understanding of some of the molecular details of
nitrobenzene deoxygenation[34] and the formation of Azoxy
and MPC elsewhere.[35,38] The complexity of the parallel re-
actions shown in Scheme 4, which together make up the
overall efficiency with which nitrobenzene can act as the
the second stage, DMC/DMO is produced from a reaction
of P2PdII=NPh with either CO/CH3OH [Equation (2b)] or
PhNO2/CO/2CH3OH [Equation (2d)] while the initial
P2Pd0 species is regenerated. The sum of Equations (1b)
and (2b) or (1b) and (2d) results in the overall stoichiomet-
ries given in Equations (3a) or (3b), respectively. The equa-
tions marked with an asterisk are similar to the unmarked
equations, but they describe the formation of DMO instead
of DMC.
PhNO2 + 3CO + 4CH3OH Ǟ 2DMC + PhNH2 + H2O + CO2
(3a)
PhNO2 + 5CO + 4CH3OH Ǟ 2DMO + PhNH2 + H2O + CO2
(3a)*
2PhNO2 + 3CO + 4CH3OH Ǟ 2DMC + Azoxy + 2H2O + CO2
(3b)
2PhNO2 + 5CO + 4CH3OH Ǟ 2DMO + Azoxy + 2H2O + CO2
(3b)* oxidant in methanol carbonylation, makes it impossible to
1408
www.eurjic.org
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2012, 1403–1412

Oxidative Carbonylation of Methanol with Nitrobenzene as Oxidant
present a fully detailed molecular picture of the catalytic tion (1b)] relative to that of oxidative dehydrogenation of
consequences of the variations in the catalyst structure. In- methanol [Equation (1c)] must be the effective displacement
stead, we will focus on some of the most remarkable cata- of the nitrosobenzene ligand in C1b/c*.
lytic observations relevant for the chemistry of the oxidative
carbonylation of methanol, attempting to rationalize these.
This displacement likely involves an associative substitu-
tion of the nitrosobenzene ligand by CO or methoxide via
a fifth coordination site at the Pd center of C1b/c*. Such a
process will be hampered by the presence of o-methoxy
groups protecting the axial coordination sites of palladium,
thus making the conversion to DMC/DMO precursors less
likely.
Effects of Catalyst Structure
Effects of o-MeO– and p-MeO– Substituents
Conversely, without a rapid displacement of nitrosobenz-
ene from the palladium coordination site, the nitrosobenz-
ene “ligand” becomes more likely involved in H-atom trans-
fer at the Pd center from coordinated methoxide to ni-
trosobenzene,[34] ultimately leading to catalytic oxidative de-
hydrogenation of methanol to give CO (and H2O).
We previously developed arguments for the competition
at the second stage of the reaction forming DMC/DMO
[Equations (2a) and (2b); Scheme 4].[34] At this stage, the
presence of axially protective o-methoxy substituents pre-
vents ready protonation and associative displacement of the
phenylamido (PhNH–) ligand in [P2PdII(OCH3)(NHPh)]
(C2a) by methanol to generate aniline and [P2PdII(OCH3)2]
(C2b/d), a precursor species for the formation of DMC/
DMO (Scheme 4). Instead, displacement of the methoxide
anion in C2a by coordination of the smaller neutral CO can
occur, thus leading to carbonylation of C2a and ultimately
to the formation of MPC. This rationalizes the concomitant
increase in MPC formation [Equation (2a)] with a decrease
in DMC/DMO formation [Equation (2b)], as is schemati-
cally illustrated in Scheme 4. That a decrease in the amount
of Azoxy is not observed with increasing amounts of DMC/
DMO is because the formation of Azoxy can also be cou-
pled with DMC/DMO production in the second stage
[Equation (3b)] by the protonation of the intermediate
P2PdII=O species C2c to dimethoxide species C2b/d
(Scheme 4).[34]
The most prominent effect on the yield of DMC/DMO
is observed with the introduction of o-MeO– substituents
in L3-type ligands (Table 1). The formation of DMC/DMO
is strongly inhibited and its amount is decreased by almost
an order of magnitude relative to that with the unsubsti-
tuted ligands. Irrespective of the stage at which DMC/
DMO is produced [Scheme 4; Equations (1b) and (2b) or
(1b) and (2d)], a serious inhibition of methanol carb-
onylation takes place. The significantly weaker effects ob-
served with p-MeO substitution suggests that steric rather
than electronic effects play a dominant role with o-MeO
substitution.
In the first stage, the formation of DMC/DMO [Equa-
tion (1b)] is in competition with the reduction of nitrobenz-
ene with CO only [Equation (1a)] and oxidative dehydro-
genation of methanol [producing MF and CO, Equa-
tion (1c)]. The first product-determining step is thus the re-
action of the palladium(0) center in C0a through oxidative
coupling of CO and nitrobenzene (to C1a) or oxidative ad-
dition of methanol (to C1b/c) as shown in Scheme 4.
We have shown that oxidative coupling of CO with nitro-
benzene is promoted with the use of more electron-donating
ligands, resulting in higher electron density at the Pd cen-
ter,[34,35] which thus partly explains the lower amount of
DMC/DMO produced with the catalytic systems contain-
ing ortho- or para-methoxy-substituted ligands.
We have observed that more methanol is oxidatively de-
We may thus conclude that it is likely that in both pos-
hydrogenated (to MF or CO) when the aryl rings in the sible stages of DMC/DMO production the introduction of
ligand are functionalized with o-MeO groups or when the o-MeO substituents in C3-backbone bis(diphenylphos-
backbone spacer is enlarged from a C3 to a C4-type back- phanyl) ligands leads to a reduction of the efficiency of
bone.[34] Discrimination of the oxidative carbonylation of nitrobenzene as oxidant in the palladium-catalyzed oxidat-
methanol to form DMC or DMO [Equation (1b)] over its ive carbonylation of methanol. Also with ligands L4X and
oxidative dehydrogenation to form CO or MF [Equa- L5Fc that have a larger bite angle, o-MeO substituents are
tion (1c)] must occur after the first deoxygenation of nitro- expected to play a similar role in reducing the efficiency,
benzene by methanol[34] through the different fates of the although to a lesser extent, of nitrobenzene as the oxidant
methoxide complex [P2PdII(OCH3)(ONPh)]+ (C1b/c*), as in the oxidative carbonylation of methanol. In addition, a
depicted in Scheme 5. Crucially important for the occur- significant decrease in the formation of Azoxy is observed
rence of oxidative carbonylation of methanol [Equa- with these ligands, when either p-methoxy or o-methoxy
Scheme 5. First reaction intermediates in the deoxygenation of nitrobenzene that are common for Equations (1b) and (1c).
Eur. J. Inorg. Chem. 2012, 1403–1412
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
1409

T. J. Mooibroek, E. Bouwman, E. Drent
FULL PAPER
substituents are present (Table S1). Apparently the attack
Thus, CO can displace methoxide from the first coordi-
on P2PdII=NPh intermediates by nitrobenzene become less nation sphere of Pd in C1b-1 (C2b/d in Scheme 4), which
likely in comparison with protonation by methanol (eventu- results in Pd carbonyl complexes such as C1b-3 and/or C1b-
ally leading to relatively more aniline and DMC/DMO or 4. It is thought that a restricted coordination space at Pd,
MPC) for a more basic P2PdII=NPh center.[38]
such as in the complexes of the L4- and L5-type ligands,
will relatively favor coordination of the small neutral CO
molecule over the larger-sized methoxide anion. The latter
can be suitably accommodated by solvation and hydrogen
bonding with methanol solvent molecules in close proxim-
ity of the Pd center.
The competition between the formation of C1b-3 and
C1b-4 (Scheme 6) follows the same steric rules. Thus, the
small CO molecule will be more advantageous for coordi-
nation than methoxide in the acyl complexes, thus relatively
favoring C1b-4 over C1b-3 when the coordination space is
more restricted and hence shifting the equilibrium towards
dimethoxycarbonyl complex C1b-5. This restricted coordi-
nation space argument thus clearly rationalizes the increase
in DMO/DMC ratios (up to ca. 3 for L5Fc) when ligands
with a large bite angle are applied instead of ligands with a
smaller bite angle (e.g. a ratio of ca. 1 for L3).
Effects of Ligand Bite Angle
The catalysts based on ligands with larger bite angles,
such as L4, L4X, and L5Fc, systematically produce signifi-
cantly larger amounts of the products of the oxidative carb-
onylation of methanol, DMC and DMO, than do the sys-
tems with the propylene-bridged ligands (Table 1). A most
remarkable observation is, however, that the DMO/DMC
ratio increases significantly with increasing bite angle of the
ligand.
A higher efficiency of nitrobenzene as the oxidant for the
oxidative carbonylation of methanol could again, in prin-
ciple, occur at both possible DMC/DMO-generating stages
1 and 2 (Scheme 4). At stage 1, the restricted coordination
space at the equatorial positions around Pd, such as that in
L4(X) and L5Fc, could favor oxidative addition of meth-
anol at Pd0, ultimately leading to reactions in Equa-
tions (1b) and (1c), over the space-demanding oxidative
coupling of CO and nitrobenzene [Equation (1a); top left
in Scheme 4]. This hypothesis is in correspondence with the
Effects of Reaction Conditions
Addition of H2 during the oxidative carbonylation of
increased contribution of CH3OH as (co-)reductant of methanol results in the formation of more methyl formate
nitrobenzene [both in Equations (1b) and (1c)], as suggested at the expense of DMC/DMO. This can be rationalized by
by quantitative product simulations comparing ligands L3X the existence of [PdIIC(O)OCH3]+-type intermediate species
and L4X.[34] This may thus at least partly account for an for DMC/DMO formation, such as C1b-3 and C1b-4 (see
increase in the formation of DMC/DMO when using li- Scheme 6). Such species are prone to undergo hydro-
gands with a large bite angle, such as L4X and L5Fc.
genolysis by a reaction with dihydrogen, thus forming MF
In stage 2, one might expect that coordination of the at the expense of DMC and DMO.
bulky phenylamido ligand in the PdII complex C2a
The decreased DMO/DMC ratio upon addition of a base
(Scheme 4) is less favorable at the restricted coordination (and the reverse effect when an acid is added) can also be
space in L4(X) and L5Fc relative to L3(X) ligands. There- rationalized by the process summarized in Scheme 6; a
fore, coordination of a smaller CO or methoxide ligand be- higher methoxide concentration will favor rapid coordina-
comes more likely, thus producing a DMC/DMO precursor tion of the methoxide anion to form C1b-3, as a result
complex (C2b/d) rather than the MPC precursor complex. forming more DMC.
This argument may thus rationalize the lower production
of MPC with concomitantly higher DMC/DMO formation, leads to a higher DMO/DMC ratio can be easily rational-
as indeed observed with L4(X) and L5Fc catalysts. ized. A higher CO concentration makes CO more success-
Likewise, the observed effect that increasing CO pressure
The same concept of restriction in equatorial coordina- ful, relative to methoxide, in the competition for the coordi-
tion space with ligands having a larger bite angle can also nation site in the C1b-3 i C1b-4 equilibrium, producing
rationalize an increase in the DMO/DMC ratio observed more of complex C1b-4 and as a result giving a higher
with these ligands, as depicted in Scheme 6.
DMO/DMC ratio.
Scheme 6. Proposed intermediates for DMC or DMO production in stage 1. Complexes C1b-1 to C1b-5 are also intermediates for DMC
or DMO production in stage 2.
1410
www.eurjic.org
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2012, 1403–1412

Oxidative Carbonylation of Methanol with Nitrobenzene as Oxidant
interface, making it possible to record these parameters throughout
the course of the reaction. Procedures for the catalytic experiments
and analysis of the reaction mixtures are described elsewhere.[34] To
ensure reproducibility, some catalytic reactions were performed in
quadruple, and the relative standard deviation was always less than
5% for all products.
Summary and Conclusions
The catalytic reactivity of palladium complexes sup-
ported by bidentate diarylphosphane ligands has been
studied in the oxidative carbonylation of methanol to di-
methyl carbonate (DMC) and dimethyl oxalate (DMO)
using nitrobenzene as terminal oxidant.
Insight into the molecular mechanism of the oxidative
carbonylation process of methanol has been obtained from
catalytic experiments employing a variety of bis(diarylphos-
phanyl) ligands with variations in the substituents on the
phenyl rings as well as in the length of the backbone spacer,
and from experiments in which the acidity or the CO pres-
sure was varied, or in which an additional partial pressure
of H2 was applied.
Supporting Information (see footnote on the first page of this arti-
cle): A table with full analytical data of the experiments and a list
of all half-reactions relevant for the redox chemistry discussed in
the text.
Acknowledgments
This research has been financially supported by the Council for
Chemical Sciences of the Netherlands, Organisation for Scientific
Research (CW-NWO). This work has been performed under the
joint auspices of the Netherlands Institute for Catalysis Research
(NIOK) Graduate School of Leiden University and six other
It was found that two key intermediate stages exist at
which the oxidative carbonylation process of methanol can
be identified. Identification of these two stages for DMC/
DMO production was shown to be helpful in rationalizing Dutch universities. Shell Global Solutions International B. V. is
kindly acknowledged for generously donating phosphane ligands
and for their hospitality during the initial phase of this project.
the observed influence that the structure of the catalyst and
the reaction conditions can have on the oxidative carb-
onylation process.
On the basis of the mechanistic insights, it is concluded
that an ideal P2Pd catalyst for the oxidative carbonylation
of methanol with nitrobenzene as the oxidant would need
a relatively acidic palladium center, be sterically open in the
axial coordination positions, but have restricted coordina-
tion space at the equatorial coordination positions of palla-
dium. The palladium complex of the ligand 1,1Ј-bis(diphen-
ylphosphanyl)ferrocene (L5Fc) meets these criteria and was
found to use nitrobenzene as oxidant for the oxidative carb-
onylation of methanol most efficiently, with an EOC of
105% of the 200% maximum theoretical efficiency possible.
In view of these initial results and the mechanistic infor-
mation generated by the present work, even more active
and/or selective catalytic systems may reasonably be antici-
pated for the oxidative carbonylation of methanol using
nitrobenzene as oxidant.
[1] Registry of Toxic Effects of Chemical Substances (RTECS, on-
line database), United States Department of Health and Hu-
man Services (National Toxicology Information, National Li-
brary of Medicine), Bethesda, MD, 1993.
[2] R. C. Weast (Ed.), Handbook of Chemistry and Physics, D-110,
58th ed., CRC Press Inc., Cleveland, 1977–1978.
[3] R. H. Richter, R. D. Priester in Kirk-Othmer Encyclopedia of
Chemical Technology (Eds.: J. L. Kroschmitz, M. Howe-
Grand), Wiley, New York, 1995.
[4] H. Ulrich, Ullmann’s Encyclopedia of Industrial Chemistry,
VCH publishers, New York, 1989.
[5] A. J. Ryan, J. L. Stanford in Comprehensive Polymer Science
(Eds.: G. Allen, J. C. Bevington), Pergamon, New York, 1989.
[6] K. Weisermel, H. J. Arpe, Industrielle Organische Chemie, VCH
Verslagsgesellschaft GmbH, Weinheim, Germany, 1988.
[7] www.gem-chem.net/capacity.html, December 2011.
[8] Y. Ono, Pure Appl. Chem. 1996, 68, 367–375.
[9] Y. Ono, Appl. Catal. A 1997, 155, 133–166.
[10] Y. Ono, Catal. Today 1997, 35, 15–25.
[11] F. Rivetti, CR Acad. Sci. II C 2000, 3, 497–503.
[12] P. Tundo, M. Selva, Acc. Chem. Res. 2002, 35, 706–716.
[13] J. L. Gong, X. B. Ma, S. P. Wang, Appl. Catal. A 2007, 316, 1–
21.
[14] R. Bertani, G. Cavinato, L. Toniolo, G. Vasapollo, J. Mol. Ca-
tal. 1993, 84, 165–176.
[15] T. W. Dekleva, D. Forster, Adv. Catal. 1986, 34, 81–130.
[16] D. Delledonne, F. Rivetti, U. Romano, J. Organomet. Chem.
1995, 488, C15–C19.
[17] P. Giannoccaro, J. Organomet. Chem. 1994, 470, 249–252.
[18] W. B. Kim, U. A. Joshi, J. S. Lee, Ind. Eng. Chem. Res. 2004,
43, 1897–1914.
[19] J. Muzart, Tetrahedron 2005, 61, 9423–9463.
[20] F. Rivetti, U. Romano, J. Organomet. Chem. 1978, 154, 323–
326.
[21] D. M. Fenton, P. J. Steinwan, J. Org. Chem. 1974, 39, 701–704.
[22] F. Rivetti, U. Romano, J. Organomet. Chem. 1979, 174, 221–
226.
[23] F. Rivetti, U. Romano, Chim. Ind. 1980, 62, 7–12.
[24] M. Goyal, R. Nagahata, J. Sugiyama, M. Asai, M. Ueda, K.
Takeuchi, Catal. Lett. 1998, 54, 29–31.
Experimental Section
All ligands were generously provided by Shell Global Solutions
Amsterdam B. V., where they were synthesized according to litera-
ture procedures.[39–47] All other solids were purchased from Acros
organics and used as received. Methanol and nitrobenzene were of
analytical reagent purity and were distilled under an argon atmo-
sphere from the appropriate drying agent.[48] After distillation,
these liquids were stored under argon. It was ensured that no water
was present by using an analytical reaction with trimethyl orthofor-
mate according to a literature procedure.[49] Carbon monoxide
(Ͼ 99% pure)[50] was purchased from Linde gas Benelux B. V. and
used as received.
1H-, and 13C-NMR spectra were recorded with a Bruker DPX300
(300 MHz) or a Bruker DMX400 (400 MHz) machine. High-pres-
sure experiments were conducted in stainless steel autoclaves
(100 mL) equipped with two inlet/outlet valves, a burst disc, a pres-
sure sensor, and a thermocouple. The autoclaves were heated by
a HEL polyBLOCK electrical heating system. Temperatures and
pressures were measured with probes connected to a computer
[25] M. Takagi, H. Miyagi, T. Yoneyama, Y. Ohgomori, J. Mol.
Catal. A 1998, 129, L1–L3.
[26] M. Goyal, R. Nagahata, J. Sugiyama, M. Asai, M. Ueda, K.
Takeuchi, J. Mol. Catal. A 1999, 137, 147–154.
Eur. J. Inorg. Chem. 2012, 1403–1412
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
1411

T. J. Mooibroek, E. Bouwman, E. Drent
FULL PAPER
[27] S. Uchiumi, K. Ataka, T. Matsuzaki, J. Organomet. Chem. [38]
1999, 576, 279–289.
T. J. Mooibroek, E. Bouwman, E. Drent, Organometallics
2011, submitted.
[28] S. P. Current, J. Org. Chem. 1983, 48, 1779–1780.
[29] D. Delledonne, F. Rivetti, U. Romano, Appl. Catal. A 2001,
221, 241–251.
[30] H. Ishii, M. Goyal, M. Ueda, K. Takeuchi, M. Asai, Appl.
Catal. A 2000, 201, 101–105.
[31] H. Yasuda, K. Watarai, J. C. Choi, T. Sakakura, J. Mol. Catal.
A 2005, 236, 149–155.
[32] E. Amadio, G. Cavinato, A. Dolmella, L. Toniolo, Inorg.
Chem. 2010, 49, 3721–3729.
[39]
[40]
[41]
[42]
[43]
[44]
F. Bickelhaupt, US 4874897, 1989.
W. Eilenberg, EU 364046, 1991.
J. A. van Doorn, US 4994592, 1991.
R. van Ginkel, US 6548708, 2003.
E. Drent, US 5091587, 1992.
C. F. Hobbs, US 4120901, 1987.
[45] P. W. Clark, B. J. Mulraney, J. Organomet. Chem. 1981, 217,
51–59.
[46] J. M. Brown, B. A. Murrer, Tetrahedron Lett. 1980, 21, 581–
[33] I. I. Moiseev, M. N. Vargaftik, T. V. Chernysheva, T. A. Strom-
nova, A. E. Gekhman, G. A. Tsirkov, A. M. Makhlina, J. Mol.
Catal. A 1996, 108, 77–85.
[34] T. J. Mooibroek, L. Schoon, E. Bouwman, E. Drent, Chem.
Eur. J. 2011, 17, 13318–13333.
[35] T. J. Mooibroek, W. Smit, E. Bouwman, E. Drent, 2012, to be
submitted.
[36] T. J. Mooibroek, E. Bouwman, M. Lutz, A. L. Spek, E. Drent,
Eur. J. Inorg. Chem. 2010, 2, 298–310.
584.
[47] R. L. Wife, A. B. van Oort, J. A. van Doorn, P. W. N. M.
van Leeuwen, Synthesis 1983, 1, 71–73.
[48] W. L. F. Armarego, C. L. L. Chai, Purification of Laboratory
Chemicals, 5th ed., Elsevier, Amsterdam, 2003.
[49] J. Chen, J. S. Fritz, Anal. Chem. 1991, 63, 2016–2020.
[50] customer.linde.com/FIRSTspiritWeb/linde/LGNL/media/
datasheets/NL-PIB-0024.pdf, December 2011.
Received: December 3, 2011
[37] H. Lund in Organic Electrochemistry. An Introduction and a
Guide, (Eds.: H. Lund, M. M. Baizer), Marcel Dekker, New
York, 3rd ed., 1991, p. 411.
Published Online: February 10, 2012
1412
www.eurjic.org
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2012, 1403–1412