Carbonylation of aryl halides: Extending the scope of the reaction

Article abstract of DOI:10.1021/op800069w

Carbonylation reactions are being increasingly favoured in pharmaceutical chemistry for the atom-efficient introduction of carbonyl centres in aldehydes, acids, esters, and amides. Convenient procedures for simple aryl iodides and bromides are well established, and now the need is to develop improved conditions to allow the reactions to be extended to the more unreactive substrates, such as sterically hindered compounds and aryl chlorides. Sterically hindered compounds such as 2-iodo- or 2-bromo-m-xylenes can be converted using alkoxy and aminocarbonylation, while dehalogenation becomes a significant side reaction for reductive carbonylation. Less hindered compounds such as 2-iodo- or bromotoluene can be reacted successfully. Changing the aryl ligands of PdCl₂{Ph₂P(CH₂)₃PPh₂} to alkyl groups improves the rate of oxidative addition but slows the carbonyl insertion step such that rates for the majority of aryl bromides are not improved by this change. Complexes such as PdCl₂{Cy ₂P(CH₂)₃PCy₂} offer better performance for alkoxy and aminocarbonylation of aryl chlorides. However, for reductive carbonylation dehalogenation is a significant side reaction. Increasing CO pressure results in additional CO coordination to the catalytic intermediates and slows the reaction, while the dehalogenation is little affected, so reaction selectivity suffers. Thus, CO pressure is a critical parameter, particularly for reductive carbonylation, in achieving the optimum performance.

Full text of DOI:10.1021/op800069w

Organic Process Research & Development 2008, 12, 566–574
Full Papers
Carbonylation of Aryl Halides: Extending the Scope of the Reaction
Christopher F. J. Barnard*
Johnson Matthey Technology Centre, Blounts Court, Sonning Common, Reading RG4 9NH, U.K.
Abstract:
of aryl iodides and bromides bearing substituents with differing
electronic properties. The results did however show sensitivity
to steric influence, which has now been further investigated.
As part of generalising this study across a variety of catalytic
carbonylation reactions the alkoxy- and aminocarbonylation
reactions have also been studied. The alkoxycarbonylation is
the most robust carbonylation reaction as evidenced by the much
larger number of examples of this conversion,2 and thus it is
the most appropriate choice for catalyst evaluation. An improved
route to amides is one of the most sought-after goals in
pharmaceutical chemistry,3 and aminocarbonylation is poten-
tially a very valuable and simple means of preparing amides.
For each of the above reaction types, the adaptation of the
chemistry to aryl chlorides has been investigated. The outstand-
ing study performed in this area is that of Milstein et al.4 They
reported the use of a range of alkyldiphosphane complexes of
palladium for carbonylation of aryl chlorides. Their study of
the mechanism identified the oxidative addition of aryl chloride
to the 14-electron [L2Pd] species as the rate-determining step,
with the increased electron donation of the alkyl phosphane
substituents, relative to aryl, being necessary to achieve an
effective rate for this step. The amount of the catalytically active
species is reduced by various equilibria involving coordination
of further phosphane or carbonyl ligands, indicating that high
pressures of CO should be avoided. The reductive carbonylation
of aryl chlorides was found to proceed at 150 °C and 5 bar CO
with sodium formate as hydride donor.
Carbonylation reactions are being increasingly favoured in phar-
maceutical chemistry for the atom-efficient introduction of car-
bonyl centres in aldehydes, acids, esters, and amides. Convenient
procedures for simple aryl iodides and bromides are well estab-
lished, and now the need is to develop improved conditions to allow
the reactions to be extended to the more unreactive substrates,
such as sterically hindered compounds and aryl chlorides. Steri-
cally hindered compounds such as 2-iodo- or 2-bromo-m-xylenes
can be converted using alkoxy and aminocarbonylation, while
dehalogenation becomes a significant side reaction for reductive
carbonylation. Less hindered compounds such as 2-iodo- or
bromotoluene can be reacted successfully. Changing the aryl
ligands of PdCl2{Ph2P(CH2)3PPh2} to alkyl groups improves the
rate of oxidative addition but slows the carbonyl insertion step
such that rates for the majority of aryl bromides are not improved
by this change. Complexes such as PdCl2{Cy2P(CH2)3PCy2} offer
better performance for alkoxy and aminocarbonylation of aryl
chlorides. However, for reductive carbonylation dehalogenation
is a significant side reaction. Increasing CO pressure results in
additional CO coordination to the catalytic intermediates and slows
the reaction, while the dehalogenation is little affected, so reaction
selectivity suffers. Thus, CO pressure is a critical parameter,
particularly for reductive carbonylation, in achieving the optimum
performance.
Bis(diphenylphosphanyl)alkane ligands were also the focus
of work by Kim and Sen5 on the aminocarbonylation of
aromatic dichlorides (using 75 psig CO and 175 °C) and a patent
by Nihon Nohyaku6 for the preparation of carboxylic acids,
although the conditions here were much harsher. Nomura et
al. reported good yields of carboxylic acids with 5 bar CO at
180 °C using catalysts containing either mono- or bidentate
cyclohexyl phosphane ligands.7
Introduction
The introduction of functional groups into pharmaceutical
intermediates is a vital part of synthesis and the modification
of these groups allows many structural variations to be obtained.
Aryl halides have been at the forefront of recent developments
in coupling and so they are widely available as key intermedi-
ates. Improving the usefulness of aryl halides through extending
the range of reactions that they can perform under mild
conditions is therefore desirable. Carbonylation chemistry offers
a variety of such reactions, allowing the formation of aldehydes,
acids, esters, and amides.
We previously reported1 on the reductive carbonylation of
aryl halides using silanes as hydride donor. This work showed
that under mild conditions (3 bar CO and 60–120 °C) it was
possible to achieve high yields of the desired aldehydes. The
reaction was demonstrated to perform satisfactorily for a range
(1) Ashfield, L.; Barnard, C. F. J. Org. Process Res. DeV. 2007, 11, 39.
(2) Skoda-Foldes, R.; Kollar, L. Curr. Org. Chem. 2002, 6, 1097.
(3) Constable, D. J. C. Green Chem. 2007, 9, 411.
(4) (a) Milstein, D. J. Chem. Soc., Chem. Commun. 1986, 817. (b) Ben-
David, Y.; Portnoy, M.; Milstein, D. J. Am. Chem. Soc. 1989, 111,
8742. (c) Ben-David, Y.; Portnoy, M.; Milstein, D. J. Chem. Soc.,
Chem. Commun. 1989, 1816. (d) Portnoy, M.; Milstein, D. Organo-
metallics 1993, 12, 1655. (e) Portnoy, M.; Milstein, D. Organometallics
1993, 12, 1665.
(5) Kim, J. S.; Sen, A. J. Mol. Cat. A: Chem. 1999, 143, 197.
(6) EP 0283194, 1992.
* Author to whom correspondence may be sent. E-mail: barnacfj@
matthey.com
(7) Miyawaki, T.; Nomura, K.; Hazama, M.; Suzukamo, G. J. Mol. Cat.
A: Chem. 1997, 120, L9.
566
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development
10.1021/op800069w CCC: $40.75
2008 American Chemical Society
Published on Web 06/14/2008

The use of ferrocenylphosphanes8 was investigated by
Ma¨gerlein, Indolese, and Beller, who identified 1-[2-dicyclo-
hexylphosphanyl)ferrocenyl ethyldicyclohexylphosphane as the
preferred ligand. This was effective for alkoxycarbonylation of
chlorobenzene at ca. 145 °C, 1 bar CO, but requiring a 4-fold
excess of phosphane ligand and the addition of excess base and
molecular sieve. Moderate yields (15-50%) were obtained for
carbonylation reactions (amino and alkoxy) of electron-deficient
aryl chlorides using a Pd-carbene catalyst in DMA at 140 °C
with atmospheric pressure CO gas.9
Some moderation of the conditions for reaction of aryl
chlorides was achieved by complexation of the aryl halide to a
Cr(CO)3 moiety, but for good yields reaction conditions were
still quite harsh (>130 °C and >15 bar CO).10,11 The addition
of sodium iodide was also reported to allow milder conditions,
where 5 psig CO and only 115 °C provided good yields for
activated substrates, but incomplete reaction in other cases.12
Conditions described by Indolese et al., 5 bar CO and 120 °C
with formamide as the amine source, gave moderate yields for
activated aryl chlorides.13
Most recently, the use of sodium phenoxide as an additive
for assisting aminocarbonylation has been described.14 Buch-
wald et al. selected 1,3- bis(dicyclohexylphosphanyl)propane
(dcpp) as the most appropriate ligand for this alternative route
to amides via alkoxycarbonylation. In the presence of sodium
phenoxide the phenyl ester is initially formed and subsequently
converted to the amide by acyl transfer to an amine catalysed
by phenoxide. This allows reactions to be carried out at modest
temperatures (100-120 °C) and pressures (0-1 bar CO).
The amino- and alkoxycarbonylation of aryl chlorides has
been reported by Larhed et al., using Mo(CO)6 rather than CO
gas, under microwave irradiation.15 Although the reaction times
are short, the associated reaction temperatures are high (>170
°C). Conveniently, it was found that such reactions could be
carried out in water with only minor amounts of benzoic acid
byproduct formed.16
reported below represent our efforts to achieve a better
understanding of the interplay of the many reaction variables.
Results
Studies of Milstein4 and others have shown that chelating
ligands often perform better than their monodentate counterparts
in palladium catalysts used for carbonylation reactions. Also,
they identified the bis(phosphanyl)propane ligands as more
effective than the corresponding ethane- or butane-based ligands.
As has been seen in coupling chemistry, replacing the phos-
phane aryl substituents with alkyl groups increases the electron
donation to palladium and promotes the initial oxidative
addition. However, there has been little comparative study of
the catalysts obtained by changing these alkyl groups. Variations
in performance might be expected from both electronic and
steric effects. Therefore, the series of compounds illustrated in
Figure 1 were prepared from commercially available secondary
phosphanes using literature methods. These catalysts were used
to supplement the commercially available materials 1-3.
Our previous work1 on reductive carbonylation studied
electronic and substituent effects using simple groups unlikely
to react under the chosen conditions. To extend this work we
now report reactions illustrating the chemoselectivity of this
method and some limitations imposed by steric hindrance. To
demonstrate the selectivity of the reaction in the presence of a
more reactive functional group, the reductive carbonylation of
4-bromostyrene was studied. The results are shown in Table 1.
The main side products were those expected from reaction of
the vinyl group with hydride, being ethyl benzene and ethyl
benzaldehyde. Small amounts (<1%) of hydroformylation
products were also formed. Thus, even without optimisation,
our standard conditions can deliver very good selectivity for
aldehyde formation versus hydrogenation or hydroformylation
with better than 95% selectivity for the desired product.
Our previous work1 and that of others (for example, refs 8,
14, and 17) has shown that the efficiency of carbonylation
reactions can be markedly influenced by steric hindrance. While
we obtained good yields for both 4-iodotoluene and 2-iodot-
oluene (96 and 97% yields, respectively) conditions giving
complete conversion of 4-bromotoluene gave minimal conver-
sion with 2-bromotoluene. Therefore, to study the conditions
required to overcome this effect, comparisons were made using
2-halotolyl and 2-halo-m-xylyl derivatives. Initial reaction
conditions taken from our previous work with 3 were extended
to include other catalysts, and changes to the temperature and
CO pressure were made. The results are reported in Table 2. It
can be seen that adding the second “blocking” methyl group in
2-iodo-m-xylene makes it difficult to achieve high yields. The
dehalogenation reaction becomes more significant, and poten-
tially dominant. It seems that attack on the complex by the silane
is less susceptible to steric hindrance than attack by CO.
Adjusting the reaction conditions, for example increasing the
CO pressure, decreasing the excess of silane, and changing the
temperature, can improve the selectivity slightly (see Table 3),
but this also increases the likelihood of other side reactions. A
similar story is evident for the aryl bromide (see Table 2), but
in this case the inhibition associated with the two methyl groups
has a more marked effect on reactivity. Even a temperature
This long history of study reflects the desirability of
establishing the most effective conditions for these transforma-
tions, while at the same time emphasising the complexities
imposed by the multistep mechanism of the reaction. The results
(8) Ma¨gerlein, W.; Indolese, A. F.; Beller, M. Angew. Chem., Int. Ed.
2001, 40, 2856.
(9) Calo, V.; Giannoccaro, P.; Nacci, A.; Monopoli, A. J. Organomet.
Chem. 2002, 645, 152.
(10) (a) Mutin, R.; Lucas, C.; Thivolle-Cazat, J.; Dufaud, V.; Dany, F.;
Basset, J. M. J. Chem. Soc., Chem. Commun. 1988, 896. (b) Dufaud,
V.; Thivolle-Cazat, J.; Basset, J.; Mathieu, R.; Jaud, J.; Waissermann,
J. Organometallics 1991, 10, 4005.
(11) Carpentier, J. F.; Petit, F.; Mortreux, A.; Dufaud, V.; Basset, J.;
Thivolle-Cazat, J. J. Mol. Catal. 1993, 81, 1.
(12) (a) Perry, R. J.; Wilson, B. D. J. Org. Chem. 1996, 61, 7482. (b) U.S.
Patent 5,672,750, 1997.
(13) (a) Schnyder, A.; Beller, M.; Mehltretter, G.; Nsenda, T.; Studer, M.;
Indolese, A. J. Org. Chem. 2001, 66, 4311. (b) U.S. Patent 6,441,233,
2002.
(14) Martinelli, J.; Clark, T. P.; Watson, D. A.; Munday, R. H.; Buchwald,
S. L. Angew. Chem., Int. Ed. 2007, 46, 8460.
(15) Lagerlund, O.; Larhed, M. J. Combi. Chem. 2006, 8, 4.
(16) (a) Wu, X.; Larhed, M. Org. Lett. 2005, 7, 3327. (b) Wu, X.; Ekegren,
J. E.; Larhed, M. Organometallics 2006, 25, 1434.
(17) Albaneze-Walker, J.; Bazaral, C.; Leavey, T.; Dormer, P. G.; Murry,
J. A. Org. Lett. 2004, 6, 2097.
(18) Klaus, S.; Neumann, H.; Zapf, A.; Stru¨bing, D.; Hu¨bner, S.; Almena,
J.; Riermeier, T.; Gross, P.; Sarich, M.; Krahnert, W.-R.; Rossen, K.;
Beller, M. Angew. Chem., Int. Ed. 2006, 45, 154.
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
567

Figure 1. Catalyst structures.
Table 1. Chemoselectivity of reductive carbonylation of 4-bromostyrenea
catalyst
bromoethylbenzene (%)
bromostyrene (%)
ethylbenzaldehyde (%)
vinylbenzaldehyde (%)
2
3
8
0.8
1.0
3.7
1.8
2.7
65.8
1.2
6.1
2.1
95.8
89.4
26.3
a Values reported as GC-MS normalised area. Method A: 4-bromostyrene (1 mmol); HSiEt3 (2 mmol); Na2CO3 (1 mmol); DMF (5 mL); 2 mol % catalyst; 3 bar CO; 80
°C; 24 h.
increase of 30 °C is not sufficient to ensure complete reaction
it unlikely that suitable conditions can be found for high yields
from the reaction for severely sterically hindered aryl bromides.
In order to react aryl chlorides the increased bond strength
C-Cl > C-Br > C-I requires harsher conditions to promote
the oxidative addition, so the reaction temperature must be
increased above that required for the bromides to maintain a
reasonable reaction time. Also, increased electron donation from
with most of the catalysts. As for the iodide, dehalogenation
becomes the dominant reaction. It is noteworthy that catalysts
4 and 8 gave lower conversion than 3 (also the case for 8 and
3 for 4-bromostyrene, Table 1), indicating that these catalysts
are significantly slower for this reaction than 3. However, the
high levels of dehalogenation occurring in these reactions make
568
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

Table 2. Reductive carbonylation of sterically hindered aryl halidesa
catalyst
2-iodotolueneb
2-iodo-m-xylenec
2-bromotoluened
2-bromo-m-xylenee
1
-
-
88 (6)
75 (9)
82 (6)
26 (3)
29 (7)
42 (2)
58 (8)
32 (3)
30 (6)
60 (3)
-
-
2
-
-
-
3
97(0)
-
8 (92)
16 (84)
10 (89)
-
12 (88)
30 (70)
13 (86)
-
3 (97)
3 (97)
5 (95)
1 (59)
9 (27)
-
4
5
-
6
-
1.7 (10)
12 (78)
14 (57)
21 (46)
-
7
-
8
-
9
-
14
15
17
18
-
-
6 (32)
-
7 (24)
-
-
-
-
a Values in parentheses indicate formation of arene (dehalogenation). Values reported as GC-MS normalised area. b Method A: 2-iodotoluene (3.6 mmol); HSiEt3 (7.2
mmol) Na2CO3 (3.6 mmol); DMF (5 mL); 2 mol % catalyst; 3 bar CO; 60 °C; 18 h. c Method A: 2-iodo-m-xylene (1 mmol); HSiEt3 (2 mmol); Na2CO3 (1 mmol); DMF (5
mL); 2 mol % catalyst; 3 bar CO; 100 °C; 24 h. d Method A: 2-bromotoluene (1 mmol); HSiEt3 (2 mmol); Na2CO3 (1 mmol); DMF (5 mL); 2 mol % catalyst; 3 bar CO; 100
°C; 24 h. e Method A: 2-bromo-m-xylene (1 mmol); HSiEt3 (2 mmol); Na2CO3 (1 mmol); DMF (5 mL); 2 mol % catalyst; 15 bar CO; 140 °C; 24 h.
Table 3. Variation in yields with reaction conditions for
reductive carbonylation of 2-Iodo-m-xylenea
by the alkyl-substituted ligands compared to the aryl-substituted
ones, and so the reduced rate here leads to more competition
from reaction of the hydride donor with the aryl intermediate.
Increasing the CO pressure in an effort to promote carbonyl
insertion rather than dehalogenation results in a reduction in
conversion (due to increased binding of CO by the catalytic
intermediates) and no improvement in selectivity. Increasing
the temperature results in the appearance of amide side products
formed from dimethylamine, this arising from catalysed decar-
bonylation of the dimethylformamide solvent.
Other reagents have been reported for reductive carbonyla-
tion. Beller et al. have reported18 that n-butyldi(adamantyl)phos-
phane ligand with palladium acetate is effective for the reductive
carbonylation of aryl bromides under CO/H2. The bidentate
ligands dppf and dppp (i.e., precursors to 2 and 3) were reported
to be ineffective under these conditions. This has been verified
in this study, with further results for 4 and 8 (yields <1%) for
2-bromotoluene and 4-tert-butylbromobenzene, while yields
were ca. 50% for 22, under 2 bar 1:1 CO/H2 at 100 °C in toluene
with TMEDA as base. However, catalysts 21 and 22 were
ineffective under the standard conditions for alkoxy- and
aminocarbonylation reported here (for example, yield <1% for
alkoxycarbonylation of phenyl triflate with methanol at 0.5 bar
CO at 80 °C, whereas yields of methyl benzoate for 1, 2, and
3 are all >60%). Good yields with these ligands can be obtained
under different conditions (3:1 ligand/Pd, TMEDA as base and
pressures >5 bar CO at elevated temperature, 115 °C).19 The
implications of this for our understanding of the mechanism of
the reaction are discussed below.
CO
pressure temperature
conversion aldehyde m-xylene
(bar)
(°C)
catalyst
(%)
(%)
(%)
3
100
100
100
90
3
8
9
1
2
3
8
9
1
2
3
8
9
100
3
30
13
35
26
16
36
41
33
54
24
33
56
92
70
86
34
61
64
54
49
67
46
76
66
44
3
100
100
90
3
5b
5b
5b
5b
5b
15
15
15
15
15
90
98
99.5
90
90
98
98
100
100
100
100
100
90
110
110
110
110
110
a Method A. Values reported as GC-MS normalised area. b Only one equiv of
HSiEt3 added.
Table 4. Reductive carbonylation of 4-chloroacetophenonea
CO
pressure temperature
conversion aldehyde acetophenone
(bar)
(°C)
catalyst
(%)
(%)
(%)
5
5
5
3
3
3
3
110
110
110
120
120
145
145
3
7
8
3
7
7
8
0
12
0
0
12
0
5
0
0.3
0
0
0
5
36
42
100
100
22
27
100
100
a Method A: values reported as GC-MS normalised area.
Alkoxycarbonylation is the most robust of this series of
carbonylation reactions due to the weakness of the interaction
of the alcohol (in comparison with an amine or hydride donor)
with the palladium catalyst. Carbonylation reactions are sensitive
to the choice of base, and preference may differ depending on
the nature of the nucleophile, i.e. the alcohol, amine, or hydride
donor. Our previous work identified the benefits of sodium
carbonate in reductive carbonylation in DMF. Screening of a
selection of bases for the alkoxycarbonylation of 4-chloroac-
etophenone (see Table 5) showed that the inorganic bases are
the ligand to palladium enhances the oxidative addition, so
improved performance is expected for the alkyl-substituted
bisphosphanes in comparison to the aryl-substituted derivatives.
However, reaction of the electron-deficient aryl chloride 4-chlo-
roacetophenone (see Table 4) carried out at 110-145 °C with
3-5 bar CO shows increasing conversion with temperature but
with dehalogenation as the dominant reaction. Attempts to react
an unactivated compound such as chlorobenzene were even less
successful with complete formation of the arene. As was seen
for the aryl bromides, the carbonylation steps are less favoured
(19) Neumann, H.; Brennfu¨hrer, A.; Gross, P.; Riermeier, T.; Almena, J.;
Beller, M. AdV. Synth. Catal. 2006, 348, 1255.
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
569

Table 5. Influence of base on alkoxycarbonylation of
4-chloroacetophenone with ethanola
base
4-acetobenzoic acid ethyl ester (%)
Na2CO3
K2CO3
KOAc
K3PO4
KF
98
98
74
53
70
47
34
27
28
29
Cs2CO3
NaOEt
NEt3
NEtiPr2
DMAP
a Method B: 4-chloroacetophenone (3.6 mmol); ethanol (1 mL); base (1 equiv);
NMP (5 mL); 2 mol % 3; 5 bar CO; 120 °C; 24 h. Values reported as GC-MS
normalised area.
also superior to amines for this reaction. However, although
sodium and potassium carbonate perform equally well in this
example, over a series of reactions potassium carbonate was
preferred to sodium carbonate for further exploration of alkoxy-
carbonylation.
As shown above, carbonylation reactions are affected by
steric factors. However, in contrast to the reductive carbony-
lation, more intense conditions do allow reactivity to be obtained
for alkoxycarbonylation without significant dehalogenation.
Thus although the reaction temperature is increased by 30 °C
to complete the reaction of 2-bromo-m-xylene in comparison
with 2-bromotoluene, good selectivity is maintained as shown
in Table 6. It can be seen that even for 8 temperatures of 120
°C or greater and higher catalyst loadings (ca. 5 mol %) are
required to achieve good conversion with electron-neutral or
electron-rich substrates. This compares with 100 °C reported
by Buchwald using dcpp ligand14 in alkoxycarbonylation with
phenoxide (but with rigorous exclusion of water and air). The
yields of esters are also good, in contrast to reductive carbo-
nylation with HSiEt3 where dehalogenation occurred. Optimi-
sation of the conditions of temperature and pressure should
allow the reaction to give high yields for most aryl chlorides.
It should be noted that comparison of NMP, commonly used
here, with DMSO as solvent, as suggested by Buchwald,14
showed no significant difference in rate or selectivity.
Although there are fewer reports of aminocarbonylation than
of alkoxycarbonylation,2 the conditions for the two reactions
are similar, reflecting the similarity in their mechanisms.
Conditions for aryl bromides and iodides have been established,
but there are few reports on the reaction of aryl chlorides. A
notable recent report is the work of Buchwald et al. with the
addition of sodium phenoxide.14 However, in our work we have
concentrated on the simpler, direct procedure and have taken
only basic precautions to minimise exposure to moisture to
better reflect an industrial setting. Initial attempts to carry out
reaction with diethylamine in DMF as solvent resulted in a
mixture of amides due to release of dimethylamine from the
solvent. Therefore, NMP was used as solvent in all further work
on aminocarbonylation. As for the alkoxycarbonylation, a series
of substrates were studied to determine the relative performance
of the catalysts for sterically demanding substrates and aryl
chlorides. The results are shown in Table 7. While steric factors
clearly inhibit the reactivity in aminocarbonylation, increasing
570
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

Table 7. Comparison of catalysts for aminocarbonylation of aryl halidesa
2-iodo-m-xyleneb 2-bromo-m-xylenec 4-chloroacetophenoned
catalyst conversion (%) GC yield (%) conversion (%) GC yield (%) conversion (%) GC yield (%) conversion (%) GC yield (%)
chlorobenzenee
1
2
100
100
100
100
100
100
100
100
100
100
61
60
64
65
62
67
57
54
53
57
37
8
27
5
52
18
38
44
56
38
96
100
25
75
45
8
2
3
12
42
20
9
62
77
42
75
2
2
11
41
19
8
61
76
41
64
3
40
84
44
70
98
100
80
93
35
46
23
37
54
71
44
77
20
40
54
30
95
100
18
64
4
5
6
7
8
9
18
a Main byproduct ketoamide formed by double carbonylation. b Method C: 2-iodo-m-xylene (1 mmol); HNEt2 0.2 mL, 2 equiv); K2CO3 (1 equiv); NMP (5 mL); 2 mol %
catalyst; 3 bar CO; 100 °C; 24 h. c Method C: 2-bromo-m-xylene (1 mmol); HNEt2 (1 mL); K2CO3 (1 equiv); NMP (5 mL); 2 mol % catalyst; 5 bar CO; 140 °C; 24 h.
d Method C: 4-chloroacetophenone (1 mmol); HNEt2 (1 mL); K2CO3 (1 equiv); NMP (5 mL); 2 mol % catalyst; 5 bar CO; 120 °C; 24 h. e Method C: chlorobenzene (1
mmol); HNEt2 (1 mL); K2CO3 (1 equiv); NMP (5 mL); 2 mol % catalyst; 5 bar CO; 140 °C; 24 h.
Table 8. Concentration effects in carbonylation
substrate
concn (M) CO pressure (bar) temperature (°C) catalyst catalyst loading (mol %)
product
yield (%)
4-bromostyrene
4-bromostyrene
Chlorobenzene
Chlorobenzene
0.2
1.0
0.2
0.7
3
3
5
5
80
80
140
140
3
3
8
8
2
2
5
2
vinylbenzaldehyde
vinylbenzaldehyde
ethyl benzoate
89.4
82.3a
80.4b
95.8
ethyl benzoate
a Main byproduct 4-ethylbenzaldehyde. b Main byproducts benzene and benzaldehyde.
the reaction temperature can give satisfactory yields without
significant dehalogenation. However, in some cases the double
carbonylation yielding the ketoamide does become a significant
side reaction, but for the aryl chlorides there was no evidence
for the ketoamide byproduct. Comparison of the catalysts gives
similar results to the alkoxycarbonylation: where oxidative
addition is easy, the arylphosphane derivatives perform well.
However, when improved oxidative addition is required, 8 is
generally the preferred catalyst, although for particular com-
pounds one of the other alkylphosphane catalysts may perform
better.
While the detrimental effect of high CO pressure on Pd-
catalysed carbonylation has been noted previously, there has
been little reporting on the study of this important variable, with
most authors reporting work in a narrow pressure range. The
need to achieve the correct pressure can be shown by the
alkoxycarbonylation of chlorobenzene. With 2 mol % 4 at 120
°C the yield at 5 bar CO is encouraging (13%), but if the
pressure is reduced to 2 bar CO, the reaction slows markedly
(ca. 1% yield), whereas increasing the pressure to 15 bar CO
results in almost complete shut-down of the reaction. Similar
results have been seen in other cases. It seems likely that the
anion bound to palladium is an important factor in determining
the optimum pressure, and a more extensive study of this topic
is warranted. Certainly, the affect of pressure should be included
as part of any initial evaluation of a carbonylation process.
In order to maximise productivity it is generally best to
operate at the maximum convenient concentration. However,
concentration can sometimes influence the selectivity of the
reaction. Examples of differences in the effect of concentration
are shown in Table 8. Therefore, substrate concentration is
another parameter that should be thoroughly investigated during
reaction optimisation.
Discussion
While individual substrates may differ in the preferred choice
of catalyst, base and solvent combinations, the previously
reported results for the reductive carbonylation of aryl bromides
and iodides suggest that PdX2(dppp), PdX2(dppf) or PdX2-
(BINAP) (X ) Cl or Br) with sodium carbonate as base in DMF
as solvent is a reasonable starting point for reductive carbony-
lation. The results presented here offer further insight into the
effects of substrate variation, as well as catalyst choice, on the
reaction conditions.
Increasing electron donation from the phosphane ligand to
promote the oxidative addition step has been well studied in
coupling chemistry.20 However, the results for 4-bromostyrene
and for other aryl iodides and bromides, comparing the results
for 3 and 8, indicate that the carbonylation step is slowed by
this change (for further discussion, see below). For the majority
of aryl bromides reaction of the aryl-palladium intermediate with
carbon monoxide is much faster than reaction with hydride
leading to very good selectivity versus dehalogenation. This
remains true for 4-bromostyrene with regard to carbonylation
versus reduction of the alkene. However, when the carbony-
lation is slowed by steric effects, or by strong electron donation
from the ligand, then the dehalogenation reaction becomes
dominant. The high levels of conversion of the 2-halotoluene
and 2-halo-m-xylene compounds confirm that the oxidative
addition still proceeds satisfactorily with the sterically demand-
ing substrates. However, the low yield of carbonyl products
indicates that attack by CO on the intermediate aryl complex
is strongly inhibited. This may be understood in terms of the
(20) A. F.; Littke, G. C.; Fu, Angew. Chem., Int. Ed. 2002, 41, 4176.
(21) Moser, W. R.; Wang, A. W.; Kildahl, N. K. J. Am. Chem. Soc. 1988,
110, 2816.
(22) Tani, K.; Tanigawa, E.; Tatsuno, Y.; Otsuka, S. J. Organomet. Chem.
1985, 279, 87.
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
571

and iodo- substrates proceeds more readily for 3 than for 8 under
circumstances where the carbonylation step(s) are rate-
determining. However, for aryl chlorides oxidative addition is
limiting for 3, while the situation is less clear for 8, where it
seems likely that the carbonylation is the slowest part of the
catalytic cycle. Because of the conflicting requirements for these
two essential parts of the cycle, it seems unlikely that there will
be the same dramatic improvement in carbonylation catalysts
as has been seen for cross-coupling. Nonetheless, optimisation
of conditions (catalyst, base, solvent, pressure, temperature) still
allows Pd-catalysed carbonylation to be an extremely effective
means for the conversion of aryl halides to other functional
groups.
Comparison of the results for bidentate bis(phosphanyl)pro-
pane complexes with those of monodentate n-butyldi(adaman-
tyl)phosphine suggests significant mechanistic differences be-
tween the reductive carbonylation carried out in the presence
of hydrogen or silane. The former complexes perform well in
silane-based reductive carbonylation, in alkoxy- and aminocar-
bonylation but do not give significant product with hydrogen-
based methodology. The reverse is true for the monodentate
phosphine complexes, which perform very poorly in alkoxy-
and aminocarbonylation under the conditions described in this
paper. This suggests that a monophosphane Pd-P species plays
a much greater role in the hydrogen-based reductive carbony-
lation. For the chelating bis(phosphanyl)propane, ring-opening
is anticipated to allow CO complexation prior to insertion into
the Pd-Aryl bond. However, binding of hydrogen in a similar
manner is not expected. In contrast, for hydrogen coordination
a readily available vacant coordination site, associated with
complete dissociation of one phosphane, is required. It is likely
that the bidentate phosphane catalyst performs better than the
monophosphane one in amino- and alkoxycarbonylation due
to the assistance that the second phosphane offers in promoting
the carbonyl insertion. The improved performance of n-
butyldi(adamantyl)phosphine in amino- and alkoxycarbonyla-
tion with the presence of excess ligand19 also supports this
suggestion.
Comparing the performance of the catalysts indicates that
both alkyl- and aryl-substituted phosphane palladium complexes
are required to give effective reaction across a wide range of
substrates. The simple phenylphosphane derivatives 1-3 are
highly effective for aryl iodides and bromides, while catalysts
such as 8 generally perform better for aryl chlorides. Comparing
the propane- and butane-based ligands, the results are in
accordance with those of Milstein et al., who showed that
catalyst activity varies with chelate ring size 6 > 7 . 5.
However, within the series of phosphanylpropane derivatives
the order of activity is less easily rationalised. Evidently, the
cyclohexyl derivative provides the optimum compromise of
electron donation and steric influence for most substrates,
although occasionally another derivative may perform better.
It is our assumption that it is mainly the steric match to the
substrate that is causing these changes in relative performance.
Chloro- and bromo- forms of the catalysts perform equally well
since the substrate provides a great excess of halide over that
originally present in the catalyst.
Figure 2. Steric influence on binding of carbon monoxide.
standard associative mechanism for CO substitution reactions,
with substitution of one phosphorus donor being required before
the migratory insertion takes place. The initial coordination of
CO must take place along the axial coordinate of the square
planar complex as illustrated in Figure 2. To minimise steric
interactions in the plane, the aryl group will lie vertically and
therefore the methyl substituent will sit over the axial site. With
one methyl group there is still one axial site readily available
and the reaction, though slowed, proceeds well, while with
methyl groups blocking both axial sites the reaction is severely
hindered. Rotation of the methyl groups away from the axial
sites correspondingly increases steric interaction in the plane,
also adversely affecting the reaction. (An alternative proposal
of initial coordination of CO at an axial site has been considered
by Albaneze et al.,17 but does not seem to provide a straight-
forward explanation of these results.) While it might be assumed
that increasing the CO pressure would promote carbonylation
versus reductive dehalogenation, this also leads to further CO
coordination, which depletes the level of active intermediates
in the catalytic cycle, and overall the reaction is slowed.
Compensating through increasing the temperature increases the
rate of side reactions. Determination of the optimum pressure
for these reactions should therefore be a major feature of process
development.
In contrast to coupling reactions, the final step in the alkoxy-
and aminocarbonylation mechanism (see Figure 3) is attack by
the nucleophile on the acyl palladium complex rather than
reductive elimination.4a These results were confirmed by Moser
et al.21 in an IR study of the formation of [PdBr-
(COC6H5)(PPh3)2] from [PdBr(C6H5)(PPh3)2] and CO which
also established that the acyl group in the former is stable to
decarbonylation. The contribution of base in generating the
appropriate nucleophile (e.g., methoxide) for attack on the
benzoyl complex is significant.
The changes to the choice of preferred catalyst and the
conditions for carbonylation of different substrates arise from
the conflicting requirements for the oxidative addition and
carbonylation steps. Milstein reported that for aryl chloride
carbonylation oxidative addition is the rate-determining step.
However, work on cross-coupling reactions has shown that
oxidative addition for aryl bromides and chlorides can occur
readily at room temperature with appropriate catalysts.20
Therefore, for many of these reactions oxidative addition is
clearly not the rate-limiting step. This must be one of the steps
involved in the formation of the acyl ligand (CO binding/
phosphane displacement or CO insertion). Promotion of the
oxidative addition by increasing electron density on palladium
will increase the capability of the palladium for back bonding
to the π-accepting ligands CO and aryl. The barrier to
combining these ligands will rise, slowing this step. For
example, as noted above, the reductive carbonylation of bromo-
572
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development

Figure 3. Outline mechanism for alkoxy- and aminocarbonylation.
Conclusion
Materials were characterised by elemental analysis (University
of Strathclyde) and IR and in some cases by NMR spectroscopy
(see Supporting Information).
The results presented here can be understood using the
conventional model for the mechanism of palladium-catalysed
carbonylation. Consideration of the main steps in the model as
described in Figure 3 indicates that a range of catalysts are
required to deliver optimum performance for the different aryl
halides. It is evident that the rate-determining step in the
mechanism can differ, depending on the substrate, the catalyst,
and the nucleophile. Thus, for aryl iodides and bromides, where
the oxidative addition occurs readily, the less electron-donating
phenyl phosphane derivatives perform well by achieving faster
carbonylation steps, while for aryl chlorides more electron-
donating alkylphosphanes are required to promote the oxidative
addition even though they slow the carbonylation steps. Once
the reagents and catalyst have been selected, to optimise the
reaction for a given conversion the concentration and rela-
tive amounts of reactants and catalyst need to be determined
(in addition to defining the most appropriate combination of
temperature and CO pressure) to achieve the best selectivity
and a suitable rate. If the CO pressure is too high, the reaction
is slowed. Compensating for this by using a higher temperature
will generally lead to reductions in selectivity, for example
through increased dehalogenation. However, with the avail-
ability of parallel screening equipment such optimisation can
now be carried out expediently.
Reactions were performed in a Baskerville 10 × 30 mL
multicell pressure reactor rated to 50 bar and 200 °C. In contrast
to the work of Buchwald et al.,14 no stringent precautions (e.g.,
flame-dried glassware, glovebox, etc.) were taken to avoid
contact with air or moisture. Catalyst, substrate, reagent, base,
and solvent were measured into reaction tubes in air. After
transfer of the tubes to the autoclave, this was sealed and then
purged four times using nitrogen and four times with CO, before
charging to the required pressure at ambient temperature. There
was a slight increase in pressure during initial heating, which
then returned to the set pressure as CO was consumed. This
pressure was then maintained at the reported value using
additional CO for the remainder of the reaction time. Analysis
of the reaction mixture was carried out by GC-MS using a
Perkin-Elmer AutoSystem XL gas chromatograph with Tur-
boMass mass spectrometer. Conditions used: 1.0 µL sample
injection, columns: 30 m × 0.25 mm DF 0.25 µm Elite Series
PE-5MS, 30 m × 0.32 mm DF 0.25 µm Elite Series PE-35MS
or 30 m × 0.32 mm DF 0.25µm PE-Wax, injection port 320
°C, initial temperature 130 °C, hold 4 min, ramp at 30 °C/min
to 300 °C, hold 5 min, split ratio 100/1, MS scan 40.0 to 600.0
EI+ (centroid). A lower initial temperature of 50 °C was used
where dehalogenation was suspected and separation of volatile
arenes was required. For these screening runs no isolation of
the products was attempted, but previous work has shown that
aldehyde,1 ester,23 and amide14 products may be isolated with
excellent recovery by routine methods.
Experimental Section
All solvents and reagents were obtained from commercial
suppliers and used without further purification. Unless indicated,
catalysts and ligands may be obtained from Johnson Matthey
Catalysts or Alfa Aesar. Bisphosphane ligands were prepared
in accordance with literature procedures22 from PH(alkyl)2
(Strem). Complexes [PdX2(bisphosphane)] (X ) Cl or Br) were
prepared by stirring together [PdX2(cod)] and the ligand in
acetone overnight at ambient temperature. The products were
collected by filtration and used without further purification.
General Procedure for the Reductive Carbonylation of
Aryl Halides (Method A). To each reactor tube was added
substrate (1 mmol) catalyst (20 µmol) and Na2CO3 (1-1.1
mmol) DMF (5 mL) and Et3SiH (0.324 mL, 2 mmol). Once
(23) Cai, C.; Rivera, N. R.; Balsells, J.; Sidler, R. R.; McWilliams, J. C.;
Shultz, C. S.; Sun, Y. Org. Lett. 2006, 8, 5161.
Vol. 12, No. 4, 2008 / Organic Process Research & Development
•
573

sealed inside the reactor, the system was purged with nitrogen
(pressurising to 3-4 bar and venting four times) and then
repeating with CO. The autoclave was pressurised with CO and
heated to the desired temperature. The reaction was stirred under
these conditions for 24 h with additional CO added to maintain
the set pressure. The reactor was allowed to cool and the
reaction mixture was filtered and the filtrate analysed by
GC-MS to determine conversion and yield of the aldehyde.
General Procedure for Alkoxycarbonylation of Aryl
Halides (Method B). To each reactor tube was added substrate
(1 mmol) catalyst (20 µmol) K2CO3 (1-1.1 mmol) and NMP
(5 mL), and alcohol (1 mL). Once sealed inside the reactor,
the system was purged with nitrogen (pressurising to 3-4 bar
and venting four times) and then repeating with CO. The
autoclave was pressurised with CO and heated to the desired
temperature. The reaction was stirred under these conditions
for 24 h with additional CO added to maintain the set pressure.
The reactor was allowed to cool and the reaction mixture was
filtered and the filtrate analysed by GC-MS to determine
conversion and yield of the ester.
General Procedure for Aminocarbonylation of Aryl
Halides (Method C). To each reactor tube was added substrate
(1 mmol) catalyst (20 µmol) NMP (5 mL) and diethylamine (1
mL). Additional base (typically K2CO3 1 mmol) was added if
desired. Once sealed inside the reactor, the system was purged
with nitrogen (pressurising to 3-4 bar and venting four times)
and then repeating with CO. The autoclave was pressurised with
CO and heated to the desired temperature. The reaction was
stirred under these conditions for 24 h with additional CO added
to maintain the set pressure. The reactor was allowed to cool
and the reaction mixture was filtered and the filtrate analysed
by GC-MS to determine conversion and yield of the amide.
Supporting Information Available
Details of the elemental analysis for catalysts synthesised
as part of this work; selected 31P NMR data. This information
is available free of charge via the Internet at http://pubs.acs.org.
Received for review March 25, 2008.
OP800069W
574
•
Vol. 12, No. 4, 2008 / Organic Process Research & Development