The palladium-phenanthroline catalyzed carbonylation of nitroarenes to diarylureas: Effect of chloride and diphenylphosphinic acid

Article abstract of DOI:10.1016/j.jorganchem.2005.03.059

The application of the palladium-phenanthroline catalytic system to the carbonylation of nitrobenzene in the presence of aniline to afford diphenylurea has been investigated. The reaction is best performed with equimolar amounts of the two reagents. Use of higher concentrations of either aniline or nitrobenzene or an increase in temperature in the range 120-170 °C leads to the formation of higher amounts of azo- and azoxybenzene. The latter were found to contain exclusively the aryl moiety deriving from nitrobenzene, with no inclusion of that derived from aniline. The addition of a small amount of diphenylphosphinic acid doubles the conversion and improves the selectivity in diphenylurea, but the effect is attenuated for larger amounts of acid. Small amounts of chloride, of the order of 10-30 mol% with respect to palladium, improve both rate and selectivity, but only inhibiting effects are detected when chloride is added to the reaction mixture for the carbonylation of 2,4-dinitrotoluene to dimethyl 2,4-toluenedicarbamate. The data obtained and that previously reported in the literature has been analyzed in the context of a unifying mechanism and an explanation for some apparent contradictions has been given.

Full text of DOI:10.1016/j.jorganchem.2005.03.059

Journal of Organometallic Chemistry 690 (2005) 4517–4529
www.elsevier.com/locate/jorganchem
The palladium–phenanthroline catalyzed carbonylation
of nitroarenes to diarylureas: Effect of chloride
and diphenylphosphinic acid
*
Michela Gasperini, Fabio Ragaini , Chiara Remondini, Alessandro Caselli, Sergio Cenini
Dipartimento di Chimica Inorganica, Metallorganica e Analitica, UniversitaÕ degli Studi di Milano, ISTM-CNR, via Venezian 21, 20133 Milano, Italy
Received 27 January 2005; received in revised form 9 March 2005; accepted 9 March 2005
Available online 26 May 2005
Abstract
The application of the palladium–phenanthroline catalytic system to the carbonylation of nitrobenzene in the presence of aniline
to afford diphenylurea has been investigated. The reaction is best performed with equimolar amounts of the two reagents. Use of
higher concentrations of either aniline or nitrobenzene or an increase in temperature in the range 120–170 ꢀC leads to the formation
of higher amounts of azo- and azoxybenzene. The latter were found to contain exclusively the aryl moiety deriving from nitroben-
zene, with no inclusion of that derived from aniline. The addition of a small amount of diphenylphosphinic acid doubles the con-
version and improves the selectivity in diphenylurea, but the effect is attenuated for larger amounts of acid. Small amounts of
chloride, of the order of 10–30 mol% with respect to palladium, improve both rate and selectivity, but only inhibiting effects are
detected when chloride is added to the reaction mixture for the carbonylation of 2,4-dinitrotoluene to dimethyl 2,4-toluenedicarba-
mate. The data obtained and that previously reported in the literature has been analyzed in the context of a unifying mechanism and
an explanation for some apparent contradictions has been given.
ꢁ
2005 Elsevier B.V. All rights reserved.
Keywords: Ureas; Carbonylation; Homogeneous catalysis; Nitroarenes; Palladium
1
. Introduction
compound to amine, followed by reaction with phos-
gene. However, phosgene is a very toxic and corrosive
material and an enormous effort has been applied to
the development of phosgene-free routes to isocyanates.
Among these, the catalytic carbonylation of nitro com-
pounds, particularly of aromatic ones, represents one
of the most interesting alternatives, but the direct car-
bonylation of nitro compounds to the corresponding
isocyanates has proved to be a difficult reaction. How-
ever, in the presence of an alcohol, carbamates can be
obtained more easily and with a high selectivity
(Eq. (1)):
The carbonylation of organic nitro compounds is a
process with a high potential synthetic and industrial
interest, since many products can be obtained from nitro
compounds and CO, including isocyanates, carbamates
and ureas [1–3]. Ureas and carbamates are important fi-
nal products and intermediates in the synthesis of pesti-
cides and fertilizers and mono- and diisocyanates are
important intermediates in the manufacturing of pesti-
cides, polyurethane foams plastics, synthetic leather,
adhesives, and coatings.
The classical method for the production of isocya-
nates requires the intermediate reduction of the nitro
cat
þ ROH þ 3CO ! ArNHCOOR þ 2CO
.
ArNO
2
2
ð1Þ
*
If no alcohol is added, but at least an equimolar amount
of aromatic amine with respect to the nitroarene is pres-
ent, diarylureas are produced (Eq. (2))
Corresponding author. Tel.: +39 2 5031 4373; fax.: +39 2 5031
4
405.
E-mail address: fabio.ragaini@unimi.it (F. Ragaini).
0
022-328X/$ - see front matter ꢁ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jorganchem.2005.03.059

4518
M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
cat
þ 3CO ! ArNHCðOÞNHAr þ 2CO
.
can be obtained by reaction of aromatic amines with
nitrosoarenes [41]. The latter are well established inter-
mediates in the reductive carbonylation of nitroarenes.
To definitely set the problem of the origin of azo
derivatives, we run several reactions between pentadeu-
terated nitrobenzene and undeuterated aniline, in the
presence of palladium trimethylbenzoate/phenanthro-
ArNO
2
þ ArNH
2
2
ð2Þ
Moreover, it has been shown that, at least in the case of
ruthenium-catalyzed reactions, diarylureas are interme-
diately formed during the reaction even when alcohols
are present [4–6]. Both carbamates and ureas are impor-
tant industrial products themselves, but can also be ther-
mally cracked to the corresponding isocyanates, thus
providing a phosgene free route to these important
intermediates [7].
In recent years, the catalytic system based on palla-
dium–phenanthroline complexes has emerged as the
most active and promising for a possible industrial
application [8–40]. We have recently much improved
the activity of this catalytic system in the production
of carbamates by the addition of phosphorus acids
line or [Pd(Phen) ][BF ] as catalysts. A longer series
2 4 2
of these experiments is still in progress, but here it suf-
fices to mention that in all reactions, independent of
the catalyst, the solvent and other promoters, only deca-
deuterated azo- and azoxybenzene could be detected by
GC–MS analysis of the solutions after the catalytic reac-
tions. Only very small amounts of nona- and octadeuter-
ated products derived from incomplete deuteration of
the starting nitrobenzene were also detected, but no pen-
tadeuterated or non-deuterated azo- and azoxybenzene
were ever detected. Thus, these derivatives do not in-
volve aniline in any way. Their formation can be ac-
counted for by the reactions described in Eqs. (4) and
(5), although we will see in the discussion that another
alternative may exists.
[
38–40]. Since diarylureas appeared to be intermediates
in the reaction, we decided to investigate more closely
their formation with the dual aim of improving their
synthesis from nitroarenes and gaining more informa-
tion on the mechanism of the carbonylation reaction
even when carbamates are the final products. Since some
apparently contradictory results had been earlier re-
ported on the effect of chloride anion on the palla-
dium–phenanthroline catalyzed carbonylation of
nitrobenzene to methyl phenylcarbamate, we also
decided to investigate more in depth this aspect and
the results of this study are also reported in this paper.
Before describing our results, it must be mentioned
that, in addition to that reported in Eq. (2), a second stoi-
chiometry has been proposed for the production of dia-
rylureas from nitroarenes and anilines (Eq. (3)) [1–3,23]:
cat
þ 3CO ! ArNðOÞ@NAr þ 3CO
.
2
ArNO
2
2
ð4Þ
ð5Þ
cat
ArNðOÞ@NAr þ CO ! ArN@NAr þ CO
.
2
2.2. Effect of reaction conditions and diphenylphosphinic
acid in the synthesis of diphenylurea
The most active palladium/phenanthroline catalysts
for the carbonylation of nitroarenes to methyl phenylc-
arbamates in methanol as solvent have formulation
cat
.
ꢀ
ArNO
2
þ5ArNH
2
þ3CO!3ArNHCðOÞNHArþ2H
2
O
[
(e.g., BF or PF ). However, this complex is inactive
Pd(Phen) ][X] , where X is a non-coordinating anion
ꢀ
2
ꢀ
2
ð3Þ
4
6
in polar non-protic, coordinating solvents, and is com-
pletely insoluble in medium to low polarity solvents such
as toluene. In order to study the synthesis of ureas, we
thus employed palladium 2,4,6-trimethylbenzoate,
In this process, more aniline is consumed than nitroa-
rene and water is also formed. We have recently shown
[
when the combination Ru (CO) /Cl is employed as
5,6] that both processes in Eqs. (2) and (3) play a role
ꢀ
3
12
Pd(TMB) , as catalyst, in the presence of phenanthro-
2
catalyst. For simplicity, in the following we will refer
to the processes in Eqs. (2) and (3) as the 1:1 and 1:5
process, respectively.
line. This compound has already been shown by Mes-
troni and us to be suitable catalyst precursor even in
toluene as solvent [15]. Note that palladium trim-
ethylbenzoate and phenanthroline immediately react in
solution to afford Pd(Phen)(TMB) [15].
2
2
. Results
The results of a series of reactions run in the absence
of promoters or in the presence of diphenylphosphinic
acid are reported in Table 1. In the same table, we also
reported the ratio, R, between the molar amounts of ani-
line and nitrobenzene reacted. Since azo- and azoxyben-
zene exclusively derive from nitrobenzene, each
consuming two equivalents of nitrobenzene to be pro-
duced, we also calculated a ratio R* in which the nitro-
benzene amount which has been consumed to produce
azo derivatives has been subtracted, that is (Eq. (6))
2
.1. Formation of azo- and azoxybenzene
The main byproducts of the carbonylation reaction of
nitrobenzene and aniline are azo- and azoxybenzene, the
latter being always more abundant than the first. Since it
was an aim of this work to investigate the stoichiometry
of the process, it was important to trace their origin. It is
well known from the organic chemistry that azoarenes

Table 1
Effect of temperature, reagent amounts, and addition of diphenylphosphinic acid
a
f
Run
T
t
PhNO /Pd
2
PhNH
mol ratio
2
/Pd
Acid/Pd
mol ratio
PhNH
conversion
2
PhNO
conversion
2
PhNHC(O)NHPh
d
selectivity (%)
PhN(O)@NPhe
selectivity (%)
PhN@NPh
selectivity (%)
R
R*
TOF
e
(
ꢀC)
(h)
mol ratio
b
c
(
%)
(%)
1
2
3
4
5
120
120
120
120
120
120
120
120
120
120
135
150
170
120
120
120
1
2
3
2
2
2
2
2
2
2
1
1
1
2
2
2
300
300
300
300
300
150
225
300
450
300
300
300
300
600
300
300
300
300
300
600
900
300
300
300
300
225
300
300
300
600
300
300
–
19.8
31.4
39.4
20.9
23.7
25.2
29.4
34.0
32.4
39.7
33.7
35.9
45.0
19.6
63.6
55.8
15.9
27.0
36.8
27.4
31.8
50.1
38.1
31.4
20.7
30.6
31.1
32.7
42.5
15.7
63.5
56.7
70.4
81.6
82.0
64.5
53.1
80.0
83.9
83.0
73.1
83.5
77.3
69.4
59.4
71.2
94.7
96.2
7.9
12.2
14.7
15.5
20.9
11.9
16.1
22.0
36.6
20.2
19.3
31.9
46.8
22.7
1.2
<0.1
<0.1
1.1
1.22
1.16
1.07
1.52
2.24
0.98
1.01
1.05
1.01
0.98
1.08
1.10
1.06
1.25
1.00
0.98
1.34
1.28
1.29
1.79
2.84
1.11
1.20
1.37
1.64
1.22
1.37
1.64
2.08
1.61
1.03
1.03
47.7
40.5
36.8
41.1
47.7
38.3
43.2
47.3
46.9
46.1
31.1
32.7
42.5
47.1
95.3
85.1
–
–
–
<0.1
1.5
–
g
6
–
<0.1
<0.1
1.6
g
7
–
g
8
g
–
9
–
2.0
g
1
1
1
1
1
1
1
0
1
2
3
4
5
6
–
<0.1
1.1
–
–
1.4
2.0
–
h
–
<0.1
1.0
4.4
8.8
2.3
1.9
a
b
c
2
Experimental conditions: Pd(TMB) = 0.030 mmol; mol ratio Phen/Pd = 4, PCO = 40 bar, in toluene (8 ml).
Calculated with respect to the starting aniline.
Calculated with respect to the starting nitrobenzene.
Calculated with respect to the sum of reacted nitrobenzene and aniline.
Calculated with respect to the reacted nitrobenzene.
d
e
f
TOF = turnover frequency = mol PhNO
2
reacted/mol Pd · h.
Reactions 6–10 were performed at a more than one year distance from the others and by a different person. Runs 2 and 8 correspond to the same experimental conditions. Although there is a
g
good qualitative agreement between the two reactions, small differences are present and comparisons should be made within the individual series.
h
2
Pd(TMB) = 0.015 mmol; mol ratio Phen/Pd = 8:1.

4520
M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
ꢁ
R ¼ ðmol PhNH
2
consumedÞ=½mol PhNO
2ðmol PhNðOÞ@NPh þmol PhN@NPh formedÞꢂ
ð6Þ
By examining the stoichiometries in Eqs. (2) and (3) (the
:1 and 1:5 processes), it is immediately evident that if
2
consumed
indistinguishable within the experimental error. This is
a clear evidence of zero order dependence of the rate
on nitrobenzene concentration. When the catalytic ratio
is further lowered, the TOF decreases. However, the
activities observed for runs 6 and 7 are not consistent
with a first-order kinetics in nitrobenzene. Overall, these
experiments indicate that we are operating in a transi-
tion area in which neither a clean zero nor a clean first
order in nitrobenzene concentration are observed. By
increasing the nitrobenzene amount, the rate approaches
zero order dependence, whereas it approaches first order
as the nitrobenzene concentration decreases, as it occurs
while the reaction is proceeding. Run 10 confirms the
independence of the rate on aniline concentration.
(2) By increasing the amount of aniline at constant
temperature (runs 2, 4 and 5) the aniline consumption
and the R* value increase, indicating a shift towards
the 1:5 process. This effect also explains why a higher
R* value is observed in run 1 with respect to runs 2
and 3. Indeed the aniline concentration lowers as the
reaction proceeds and the corresponding lowering of
R* will result in a lower global R* value, although differ-
ences are at the limits of the experimental error. A com-
parison of the R* values for runs 6–9 shows that the R*
value also rises when the nitrobenzene concentration
increases.
ꢀ
1
only the 1:1 process is operating, R* should be 1,
whereas if only the 1:5 is active, R* should be 5. Any
intermediate value indicates that both processes are
operating in competition.
It should be noted that we intentionally operated as
to have low conversions, to avoid that the complete con-
sumption of one of the reagents may alter the kinetics of
the reaction or the relative weigh of the two
stoichiometries.
After all reactions, partial decomposition of the cata-
lytic system to give metallic palladium was observed.
A few trends emerge from the analysis of the data.
1) As the reaction proceeds, its rate decreases (runs
–3). By plotting log([PhNO ] /[PhNO ] ) (where
(
1
[
2
0
2 f
PhNO ] and [PhNO ] are the nitrobenzene concentra-
2 0 2 f
tion, respectively, at the beginning and at the end of
reactions 1–3 in Table 1) against the reaction time, a
2
straight line is obtained (R = 0.997) which is indicative
of a first order dependence of the rate on nitrobenzene
concentration. Since aniline and nitrobenzene consump-
tions run almost parallel in runs 1–3, it should be con-
sidered that an apparent first-order consumption of
nitrobenzene may instead derive from a first order
dependence of the rate on aniline concentration. This
possibility is discounted by the fact that an overall three-
fold increase in the initial aniline amount (runs 2, 4 and
(3) In general, the selectivity in diphenylurea de-
creases if either the nitrobenzene or the aniline concen-
trations increases, while more azo- and azoxybenzene
are formed.
(4) By increasing the reaction temperature up to 170
ꢀC (runs 1 and 11–13), the conversion increases,
although irregularly, but the selectivity in diphenylurea
markedly decreases. The R* value and the azoxybenzene
selectivity rise steadily. It should be noted that an in-
crease in the azoxybenzene selectivity with temperature
had been earlier observed even when the reaction is per-
formed in methanol and in the absence of added aniline
[38,39]. In any case, temperatures higher than 135 ꢀC ap-
pear to be unsuitable for the production of ureas,
although the ideal temperature for the formation of car-
bamates is 170 ꢀC [38,39]. At the end of these reactions,
increasing amounts of metallic palladium were observed
as the temperature was increased. Clearly, the catalytic
system is deactivating faster at higher temperatures
and this explains the irregularity in the rate increase.
(5) When the palladium amount is halved (runs 2 and
14), the conversion decreases, but does not halve, indi-
cating that the catalyst deactivates more slowly when
its concentration is lower, an effect already observed in
several cases, including the synthesis of carbamates by
the same catalytic system [38,39]. Selectivities are differ-
ent in runs 2 and 14, but it is more instructive to com-
pare run 14 with run 1, where the reaction was
performed with a double amount of catalyst and for half
the time. The outcome of the two reactions is very
5
) results only in a small variation in the nitrobenzene
consumption, which is at the limits of experimental error
between runs 2 and 4, despite a doubling of the aniline
amount in the second reaction. Such small variations
may just be due to a solvent effect, since the aniline vol-
ume is not completely negligible and affects both the vol-
ume and the polarity of the reaction medium. However,
it must also been considered that catalyst deactivation is
observed and we are also looking at the overall result of
two different stoichiometries. Thus, a clean first order
dependence on nitrobenzene concentration cannot be
claimed, although it is evident that nitrobenzene concen-
tration has an effect on the reaction rate. To gain more
information on the kinetics of this reaction, a further
series of reactions (runs 6–10) was performed by varying
the initial nitrobenzene amount. Reactions 6–10 were
performed at a more than one year distance from the
others and by a different person. Runs 2 and 8 corre-
spond to the same experimental conditions. Although
there is a good qualitative agreement between the two
reactions, small differences are present and comparisons
should be made within the individual series. The turn-
over frequencies of the reactions run with a nitroben-
zene/Pd ratio 450 or 300 (runs 8 and 9) are

M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
4521
similar, indicating that nitrobenzene and aniline concen-
trations are more important than palladium concentra-
tion and reaction time in determining the reaction rate
and selectivity.
spect to palladium to a catalytic system very similar to
that reported by Mestroni was sufficient to decrease
the catalytic activity.
At this point, it was important to investigate more
quantitatively the promoting effect of chloride and its
possible role in addressing the reaction stoichiometry.
Indeed, it should be noted that the 1:5 process can be
envisioned as an oxidative carbonylation reaction of
aniline, with nitrobenzene acting as the oxidant in place
of oxygen, and chloride is known to have a profound ef-
fect on the latter reaction when palladium is employed
as catalyst [1]. For reasons of solubility and stability,
(
6) The addition of diphenylphosphinic acid (run 15)
in a 4.4 acid/Pd mol ratio more than doubles the nitro-
benzene conversion (from 27.0, run 2, to 63.5%) and
markedly increases the diphenylurea selectivity. Quite
importantly, very little azoxybenzene is formed and
the R* is almost unity, indicating that the contribution
of the 1:5 process becomes negligible. A further increase
in the acid amount results in a slower reaction, but the
selectivity in urea is further enhanced to 96.2%. The
mechanistic implications of these results will be dis-
cussed in a later paragraph, together with the trends in
the R value, but the important result from a synthetic
point of view is that the addition of phosphorus acids
is strongly beneficial not only in the synthesis of carba-
mates, but also of ureas.
+
+
+
an organic countercation, PPN (PPN = (PPh ) N ),
3
2
was employed. This specific salt was chosen because it
is more chemically stable than other quaternary ammo-
nium and phosphonium chlorides and, moreover is not
hygroscopic. It is our experience that small amounts of
water can almost completely inhibit the activity of chlo-
ride in an organic solvent, likely because of the forma-
tion of strong hydrogen bonds.
2
.3. Effect of chloride on the synthesis of diphenylurea
The addition of increasing amounts of [PPN][Cl]
from 10 to 30 mol% with respect to palladium) led to
(
Several years ago, Mestroni and co-workers [11] re-
an increase in conversion from 39% to 50% and to low-
ering of the amount of azoxybenzene formed and of the
R* value (Table 2, runs 1–3, to be compared to run 3 in
ported that the presence of small amounts of chloride
in the palladium–phenanthroline catalytic system in-
creased the rate of the carbonylation reaction, in meth-
anol, of nitrobenzene to methyl phenylcarbamate,
despite the fact that the use of palladium chloride as cat-
alyst was ineffective. Later, Lee and co-workers [42] re-
ꢀ
Table 1). The best results were obtained at a ratio Cl /
ꢀ
Pd = 0.2, but the results at Cl /Pd = 0.3 are essentially
indistinguishable. Higher amounts of chloride (60–100
mol% with respect to palladium) gradually decreased
the conversion to approximately the initial value, but
the selectivity in urea remained higher than in the ab-
sence of chloride (Table 2, runs 4 and 5, and Table 1,
run 3).
To check if the addition of chloride was modifying
the kinetic order with respect to nitrobenzene, two fur-
ther experiments were run, employing different initial
ported
a
positive effect of chloride on the
carbonylation of nitrobenzene to diphenylurea when
palladium catalysts with monodentate phosphines as li-
gands were employed. When chelating phosphines were
employed [43], an inhibiting effect was observed instead.
Finally, van Leeuwen and co-workers [17] reported that
the addition of a 1:1 molar amount of chloride with re-
Table 2
Effect of the addition of chloride in the synthesis of diphenylurea
a
ꢀ
f
Run P (bar) PhNO /Pd Cl /Pd
2
PhNH
2
PhNO
2
PhNHC(O)NHPh PhN(O)@NPh PhN@NPh
R
R*
TOF
mol ratio
mol ratio conversion conversion selectivity
b
selectivity
e
(%)
selectivity
e
(%)
c
d
(%)
(
%)
(%)
1
2
3
4
5
6
7
8
9
40
40
40
40
40
40
40
30
60
300
300
300
300
300
225
600
300
300
0.10
0.20
0.30
0.60
1.00
0.20
0.20
0.20
0.20
48.9
50.2
47.6
41.0
42.3
44.9
51.8
42.3
40.1
44.8
49.4
48.2
38.6
41.7
60.4
29.1
47.0
43.9
89.7
93.6
94.5
91.2
89.8
95.0
93.7
97.1
98.2
12.2
3.7
3.4
4.5
3.7
2.7
8.7
5.8
3.4
1.3
<0.1
<0.1
<0.1
<0.1
<0.1
<0.1
<0.1
<0.1
1.09 1.24 44.8
1.02 1.06 49.4
0.99 1.02 48.2
1.06 1.10 38.6
1.01 1.05 41.7
1.01 1.02 45.7
0.89 0.94 58.2
0.90 0.95 47.0
0.91 0.95 43.9
a
b
c
d
e
f
2 2
Experimental conditions: Pd(TMB) = 0.03 mmol; mol ratios PhNH /Phen/Pd = 300:4:1, T = 120 ꢀC, in toluene (8 ml) for 3 h.
Calculated with respect to the starting aniline.
Calculated with respect to the starting nitrobenzene.
Calculated with respect to the sum of reacted nitrobenzene and aniline.
Calculated with respect to the reacted nitrobenzene.
TOF = turnover frequency = mol PhNO reacted/mol Pd · h.
2

4522
M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
amounts of nitrobenzene (runs 6 and 7). A comparison
of the turnover frequencies observed in these reactions
with that of run 2 shows a situation analogous to that
in the absence of chloride. A dependence of the rate
on nitrobenzene concentration is clearly observable,
but it is lower than that expected for a first-order kinet-
ics. Thus, again an intermediate kinetic regime is
observed.
We finally also investigated the influence of CO pres-
sure (runs 2, 8 and 9). The results of the three reactions
are close, with a bell-shaped effect on nitrobenzene con-
version, at variance with the results previously obtained
for the related carbonylation to give methyl phenylcar-
bamate, where a clean first order dependence of the rate
on CO pressure was observed [38,39].
mates and ureas. The results of two experiments are re-
ported in Table 3.
The results clearly show that with both catalysts only
a trace amount of carbamate is formed, the main prod-
uct being diphenylurea. This confirms that diphenylurea
is indeed intermediately formed even in the presence of
methanol and that the presence of a carboxylate ligand
is not necessary. The tetrafluoborate catalyst is inactive
in toluene for solubility reasons, but can produce ureas
when dissolved in a suitable solvent. The comparison
also confirms other expected features. The presence of
coordinating anions is known to decrease the activity
of the catalytic system [18], in accord with the lower con-
version of run 2. Moreover, the addition of carboxylate
anions has been reported to increase the amount of
azoxybenzene [19], which is also confirmed by the pres-
ent results.
2
.4. Synthesis of diphenylurea in methanol as solvent
A comparison with the results reported in Table 1,
also shows that the catalytic system is much more active
in methanol, but the effect of the presence of a large
amount of aniline is also more negative on the selectivity
than in toluene.
Several pieces of evidence in the literature show that
when the carbonylation of nitroarenes is performed in
the presence of alcohols, diarylureas are intermediately
formed. Only at a later stage, these react with the alco-
hol to produce the finally observed carbamate. An
equivalent of aromatic amine is regenerated at the same
time, which re-enters the catalytic cycle [4–6]. In order to
evidence if this is the case even when palladium–phenan-
throline catalytic systems are employed, we performed a
few reactions. First, the uncatalyzed alcoholysis of
diphenylurea in methanol (Eq. (7)) was investigated.
2.5. Effect of chloride on the carbonylation of 2,4-
dinitrotoluene to dimethyl 2,4-toluenedicarbamate
The first problem which is encountered when dealing
with the carbonylation of 2,4-dinitrotoluene is the large
number of possible intermediates and byproducts. Apart
from the desired dicarbamate (1), there are two nitrocar-
bamates (2a,b), two aminocarbamates (3a,b), two nitro-
toluidines (4a,b), diaminotoluene (5) (Scheme 1) and a
virtually unlimited number of azo- and azoxyarenes,
and ureas. In a very recent work [40], we have indepen-
dently synthesized all the nitro- and aminocarbamates
reported in Scheme 1, so that it is possible to analyze
them quantitatively. The experimental conditions for
the synthesis of the dicarbamate 1 had also been opti-
mized. The experiments in the presence of chloride are
reported in Table 4, together with a reference reaction
under optimized conditions. Note that in this case
PhNHCðOÞNHPhþMeOH!PhNHCOOMeþPhNH2
ð7Þ
The reaction proceeded with a rate increasing with the
temperature at temperatures higher or equal 120 ꢀC,
but was very slow at 110 ꢀC. This temperature was thus
selected for the carbonylation experiments because pos-
sibly formed urea would not be significantly alcoholyzed
and should be detectable. Both Pd(TMB)2 and
[Pd(Phen) ][BF ] were tested in order to further
2 4 2
strengthen the analogy between the synthesis of carba-
Table 3
Synthesis of diphenylurea in methanol as solvent
a
Run
Catalyst
PhNH
conversion
2
PhNO
conversion
2
PhNHC(O)NHPh
selectivity
PhN(O)@NPh
selectivity
PhN@NPh
selectivity
PhNHCOOMe
selectivity
R
R*
b
c
d
e
e
d
(
%)
(%)
(%)
(%)
(%)
(%)
1
2
a
[Pd(Phen)
Pd(TMB)
2
][BF
4
]
2
23.2
17.7
21.0
13.8
67.1
53.5
46.9
64.0
1.3
<0.1
1.9
2.9
1.10
1.28
2.14
3.56
2
2 2
Experimental conditions: Catalyst 0.020 mmol, mol ratios PhNO /PhNH /Phen/Pd = 730:730:6:1 (in the case of run 1, the ratio Phen/Pd = 6
includes the amount of phenanthroline already coordinated to palladium in the starting complex), T = 110 ꢀC, PCO = 60 bar, in methanol (10 ml) for
.5 h.
1
b
Calculated with respect to the starting aniline.
Calculated with respect to the starting nitrobenzene.
c
d
e
Calculated with respect to the sum of reacted nitrobenzene and aniline.
Calculated with respect to the reacted nitrobenzene.

M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
4523
CH3
CH3
CH3
of diphenylphosphinic acid and chloride anion on the
palladium–phenanthroline catalyzed carbonylation of
nitrobenzene to diphenylurea and the effect of chloride
anion on the carbonylation by the same type of catalyst
of 2,4-dinitrotoluene to the corresponding dicarbamate.
The strongly promoting effect of phosphorus acids on
the palladium–phenanthroline catalyzed carbonylation
of nitroarenes to carbamates had already been evi-
denced by us in previous papers [38–40]. The syntheti-
cally relevant conclusions that can be drawn from the
results reported in Tables 1–4 are the following:
NHCOOCH3
NO2
NHCOOCH3
NHCOOCH3
NHCOOCH3
NO2
1
2a
2b
CH3
CH3
NH2
NHCOOCH3
NHCOOCH3
NH2
(
1) The palladium–phenanthroline catalytic system
can be employed to produce diphenylurea with a
high selectivity, but the activity is lower than in
the corresponding carbonylation reaction to afford
carbamates. This is likely mostly due to the fact
that the catalytic system deactivates to give metal-
lic palladium quite rapidly.
3
a
3b
CH3
CH3
NH2
CH3
NO2
NH2
NH2
NO2
NH2
(2) Although aniline is a reagent in the synthesis of
diphenylurea, its amount should not exceed that
of nitrobenzene, otherwise selectivity is shifted
towards the formation of azo derivatives.
4
a
4b
5
Scheme 1. Numbering scheme for the products derived from 2,4-
dinitrotoluene.
(3) An increase in temperature in the range 120–170
ꢀ
C is also detrimental to the selectivity, whereas
the CO pressure is little influential in the range
phenylphosphonic acid was employed in place of diphe-
nylphosphinic acid because the former had been found
to give superior results in the synthesis of carbamates.
From the data in Table 4 it is immediately evident
that, under these conditions, chloride always inhibits
the formation of the dicarbamate, the effect increasing
with an increase in the chloride amount. The effect is
remarkable if we consider that even at the highest
30–60 bar.
(4) The addition of a small amount (4.4-fold that of
palladium) of diphenylphosphinic acid results in
a doubling of the reaction rate and an increase in
selectivity towards diphenylurea, but a further
increase in the acid amount gave an attenuated
effect. Although the positive effect on rate and
selectivity was also observed in the synthesis of
carbamates, in that case much larger amounts of
acid were tolerated before the effect flattened and
only a small decrease in activity was observed at
very high acid loadings. Note that phosphoric or
phenylphosphonic acids, which are the best pro-
moters in the case of the synthesis of carbamates
were not tested in the synthesis of ureas because
of solubility/miscibility problems with the toluene
solvent.
ꢀ4
amount investigated, its concentration is only 5 · 10
M.
3
. Discussion
3
.1. Synthetic aspects
In this paper, we have examined the influence of sev-
eral experimental variables and of the promoting effect
Table 4
Effect of chloride on the carbonylation of 2,4-dinitrotoluene to dimethyl 2,4-toluenedicarbamate (1)
a
ꢀ
b
b
b
b
b
Run
Cl /Pd mol ratio
1 selectivity (%)
2a selectivity (%)
2b selectivity (%)
3a selectivity (%)
3b selectivity (%)
c
1
–
77.6
68.4
63.7
55.7
1.1
1.5
3.8
3.6
–
2.6
2.3
2.5
2.2
0.8
0.7
1.1
0.7
2
3
4
a
0.5
1
1.9
3.5
3.6
2
ꢀ3
Experimental conditions: Catalyst [Pd(Phen)
2
4 2
][BF ] = 3.2 mg, 5.0 · 10 mmol, mol ratios 2,4-dinitrotoluene/4b/phenylphosphonic acid/Phen/
Pd = 2920:400:330:32:1, 2,2-dimethoxypropane = 1.0 ml, PCO = 100 bar, T = 170 ꢀC, in methanol (20 ml), for 3.5 h. Conversion of dinitrotoluene
and 4b was complete in all cases. No 4a,b or 5 was detected at the end of the reaction. Chloride was added as [PPN][Cl].
b
Calculated with respect to the sum of reacted 4b and dinitrotoluene.
Data from [40].
c

4524
M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
(
5) Small amounts of chloride have been confirmed to
and other metals catalysts, that the addition of large
have a positive role on nitrobenzene conversion in
the synthesis of diphenylurea, with a bell-shaped
concentration/activity relationship, but only inhib-
iting effects were detected in the synthesis of carba-
mates. It is important to note that, in the case of
the synthesis of diphenylurea, not only an increase
in rate, but also in selectivity was observed and the
last persisted even at chloride levels where the
activity decreases. A positive effect of chloride on
the selectivity of a palladium–phenanthroline cata-
lyzed carbonylation reaction of a nitroarene had
never been reported before.
amounts of aniline to the reaction mixture for the car-
bonylation of nitroarenes to carbamates much increases
the amount of azo derivatives formed. Since it has also
been reported that bases have an analogous effect [1],
it is now clear that the effect of aniline in the formation
of azo derivatives is due to its basicity and not to a direct
involvement into the product formation.
3.2.2. Role of chloride
As previously mentioned, small amounts of chloride
have been reported to be beneficial to the carbonylation
reactions of nitroarenes under certain conditions [11,42],
but to inhibit it in others [17,43]. Since non-coordinating
anions give better catalytic systems, the origin of the
inhibiting effect can be immediately identified. We also
Other trends are either self-evident or are of more rele-
vance to the reaction mechanism and will be discussed
in the next paragraph.
recall that Pd(Phen)Cl is very stable and insoluble in
2
any solvent that does not displace phenanthroline. The
origin of the positive effect is, on the other hand, less
obvious. In review articles [1,23], we have earlier pro-
posed that the promoting effect may be due to the for-
mation of an anionic palladium(0) complex having a
coordinated chloride anion (Scheme 2).
3
.2. Reaction mechanism
The data collected and a comparison with previously
reported results pose several mechanistic questions and
give some hints as to how solve them.
The idea had been prompted us by a series of papers
by Amatore, Jutand, and Co-workers [44] in which these
authors analyzed the products of the electrochemically
induced reduction of PdX (PPh ) (X = Cl, Br), possibly
3
.2.1. Origin of azo- and azoxyarenes
Azo- and, especially azoxyarenes are the most ubiqui-
tous byproducts during the carbonylation reactions of
nitroarenes with a variety of catalysts [1]. It has long
been known in the literature that reaction of aromatic
amines with nitrosoarenes, catalyzed by both acids and
bases, affords azoarenes even under mild conditions
2
3 2
in the presence of added n-Bu NX, and showed them to
4
ꢀ
be an equilibrium mixture of [Pd(PPh ) X] ,
3
2
2
ꢀ
2ꢀ
½PdðPPh Þ ðl-XÞꢂ , and [Pd(PPh ) X ] with the first
3
2
2
3 2
2
largely predominating in the absence of added halide
and the last becoming increasingly important as the con-
[
41]. Since nitrosoarenes are well recognized intermedi-
ꢀ
ates in the reduction of nitroarenes by CO and aromatic
amines are also often added or generated in the reaction
mixture, it has been proposed that at least part of the
produced azoarenes may derive from such a coupling
reaction. Although such assumption appears to be very
reasonable, our data on the deuteration level of azo- and
azoxybenzene in reactions between deuterated nitroben-
zene and undeuterated aniline shows unequivocally that
reaction of aniline with nitrosobenzene plays no detect-
able role in the production of azobenzene, at least with
the palladium–phenanthroline catalytic system. Azoben-
zene can also derive from a deoxygenation reaction of
azoxybenzene and, since in most experiments its amount
is very low and only increases at long reaction times, it is
likely that this process is indeed the only one relevant to
its production. Several possibilities, on the other hand,
exists for the formation of azoxybenzene, but the data
reported in this paper do not allow to make any choice
among them, so we will postpone the discussion of this
aspect until we can give a clearer answer to this problem.
The absence of aniline-derived aryl moieties in azo-
and azoxybenzene was quite surprising even because it
has been observed in a number of reports [1,20], includ-
ing the present one, both with palladium/phenanthroline
centration of X increases. From the point of view of
the reactions presently discussed, it is particularly
important that the polarographic oxidation wave poten-
0
tial of a solution of Pd (PPh ) decreased by as much as
3
4
0.3 V when n-Bu NCl was also added [45]. Since the first
4
step of the nitro compound reduction is almost surely an
0
electron transfer from a Pd complex to the nitro com-
pound [1,23], a similar shift of the oxidation potential
of the complex should markedly accelerate it. The addi-
tion of halides increased the rate of oxidative addition of
ArI [45] and ArOTf [46] to Pd(PPh ) .
3
4
Cl-
Pd(Phen)(CO)n
Pd(Phen)(CO)n-1(Cl)
CO
ArNO2
ArNO2
slow
fast
products
products
Scheme 2.

M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
4525
The explanation we proposed can be correct only in
3.2.3. Kinetics
case the activation of the nitroarene is the slow step of
the reaction. This implies that the kinetics should be
dependent on the nitroarene concentration. The experi-
mental data reported in this paper shows that this is in-
deed the case when the synthesis of diphenylurea is
concerned, which is also the system in which we detected
a positive effect. On the contrary, van Leeuwen and co-
workers [18] showed the kinetics to be approximately
zero order in nitrobenzene under the same conditions
in which the negative effect of chloride was detected
Before discussing the kinetics of the process, it is use-
ful to recall the possible alternative mechanisms for the
reaction (Scheme 3).
Without entering into the details, two main alterna-
tives exist. In the first, path A in Scheme 3, nitrobenzene
is intermediately reduced to aniline, while a palla-
dium(II) complex is formed (step (a)). Only in the fol-
lowing step (step (b)) does this palladium complex
react with aniline to afford phenylisocyanate. This is in
turn trapped by excess aniline to afford diphenylurea.
When the reaction is run in methanol and at tempera-
tures higher than 110 ꢀC, alcoholysis of the urea gener-
ates methyl phenylcarbamate and an equivalent of
aniline. The identity of the intermediate palladium com-
plex is not relevant at this stage of the description. When
methanol is present, it is likely to be Pd(Phen)(-
[
17]. We have also found a zero-order rate dependence
on nitrobenzene concentration under conditions [38,39]
very similar to the ones in which the experiments in Ta-
ble 4 were performed and the available data on the car-
bonylation of dinitrotoluene [40] strongly suggests that
its initial activation is not the rate determining step of
the reaction, although a detailed kinetic study was not
performed. Thus, the apparent contradiction between
the data obtained in different conditions, both in the lit-
erature and in the present paper, can be easily solved. A
positive effect is to be expected, and is indeed observed,
when the activation of the nitroarene is slow, whereas
should not be present, and is indeed not observed, when
it is fast. Since chloride can also have an inhibiting effect
when coordinated to a palladium(II) intermediate, this
will be the only effect detectable when no positive effect
occurs. In his original experiments, Mestroni did not
measure the kinetics of the reaction, but extension of
the discussion above suggests that the activation of
COOMe) [1,19,23,37,47], but in the present system it
2
may be Pd(Phen)(TMB) . In any case, it is known that
2
many palladium(II) compounds reacts with amines un-
der a CO atmosphere to give ureas. It is important to re-
call that it has been found that the reaction of
Pd(Phen)(COOMe) with amines is accelerated by the
2
presence of acids [48] and we have very recently found
that the reaction is also very sensitive to the concentra-
tion of CO in solution [49]. Moreover, even the reaction
of Pd(Phen)(TMB) with aniline is accelerated by acids.
2
A second alternative (path B) involves the formation
of a metallacyclic intermediate [29–33]. This can decom-
pose to yield phenylisocyanate, which will be trapped by
aniline as in path A. Quite importantly, even the decom-
position of this metallacycle has been reported to be
accelerated by acids (specifically trimethylbenzoic acid)
[29]. This second pathway can provide urea without ani-
line being intermediately formed.
Since both pathways consume equal amounts of
nitrobenzene and aniline, it is obvious that at least a
third pathway should operate in order to justify the dif-
ferent consumption of the two reagents. We will deal
with this in one of the following paragraphs.
In this paper, we have unequivocally shown that
diphenylurea is indeed intermediately formed during
the carbonylation of nitrobenzene even when methanol
is employed as solvent. This allows a comparison of
the results obtained in the synthesis of ureas with those
previously obtained for the synthesis of carbamates. We
have previously provided strong evidence that in the
synthesis of methyl phenylcarbamate catalyzed by palla-
dium–phenanthroline complexes path A in Scheme 3 is
either the only operating or, more likely, the strongly
prevailing pathway for the reaction [38,39]. The kinetics
is zero order in nitrobenzene, first order in CO pressure
and accelerated by an increase in the concentration of
aniline (a clean first order dependence is not observed
because of the different equilibria in which this com-
pound is involved). A strong acceleration by acids was
nitrobenzene was not
conditions.
a fast process under his
It may be noted that carboxylate anions can also
coordinate to palladium(0) complexes [47] and play a
similar role to chloride. However, it has been shown
by van Leeuwen and co-workers [19] that the addition
of excess carboxylate anions to the palladium–phenan-
throline catalytic system not only slows down the reac-
tion, but also results in higher amounts of
azoxybenzene. Although the effect of very small
amounts of carboxylates was not investigated and it is
not possible to say if an accelerating effect would be ob-
served, the effect on selectivity is clearly opposite to that
here observed for chloride. Since carboxylate anions are
basic, whereas chloride is not, and a large body of accu-
mulating evidence [1,19] shows that the presence of
bases results in an increase of azoderivatives formation
in the carbonylation reactions of nitroarenes, it is clear
that the basicity effects of carboxylate prevail on any
possible positive effect. That chloride has also a positive
effect on selectivity and not only on activity is an unprec-
edented and unexpected feature. At the moment, our
knowledge of the detailed reaction mechanism is insuffi-
cient to allow an explanation supported by experimental
data and we prefer not to make any unsupported
speculation.

4526
M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
N
Pd(CO)n
N
PhNO2
CO
Path A
Path B
MeOH or
RCOOH
-
CO2
O
step (a)
PhNH2
CO2
O
O
R
R
N
OMe
OMe
N
O
O
N
Pd
or
Pd
O
N
Pd
N
N
N
O
Ph
O
O
PhNH2
CO
step (b)
(
accelerated by
acids)
N
N
N
Pd(CO)n
Pd(CO)n
N
MeOH or RCOOH
CO2
PhNCO
PhNH2
MeOH
PhNHCOOMe
PhNHC(O)NHPh
+
PhNH2
T > 110 ˚C
Scheme 3.
observed. Since the rate of the step (b) depends on the
aniline concentration, but that of step (a) does not, it
is reasonable to assume that by gradually increasing
the aniline amount we may reach the point at which
the rate of the nitrobenzene activation and aniline car-
bonylation become comparable or the latter is even fas-
ter than the former. The kinetic data reported in this
paper indicate that we have indeed reached this limit.
Under these conditions, either a first order dependence
of the rate on nitroarene concentration (if the rate of
step (a) is markedly slower than step (b)) or a complex
kinetic rate (if the two rates are comparable), is expected
and is indeed observed. If step (a) is rate determining,
the kinetics should also become independent from both
aniline and CO concentrations (CO coordination to a
palladium(0) complex should be easy), which is also in
accord with the kinetic data.
The role of the acid is more difficult to quantify, be-
cause it is apparently involved at several stages of the
reaction and because of the possible involvement of
the process in path B. If we suppose that the main effect
of the acid is to accelerate the reaction between the inter-
mediate palladium(II) complex and aniline (step (b) in
path A), a small acceleration by the acid should still
be observable if the rate of step (a) and (b) are compa-
rable, but this should rapidly flatten. Indeed, as the rate
of step (b) is accelerated, it will become faster than step
(a) and any further acceleration will not be kinetically
observable. This is in agreement with the experimental
data.
Note that the reaction associated with path B can
surely contribute to the overall process because we have
previously reported that when the Pd(TMB) /Phen cata-
2
lytic system is employed under very similar conditions,
but in the absence of aniline, phenylisocyanate is formed
[15]. In the absence of added aniline, path B is likely to
be dominant because the small amounts of aniline pro-
duced during the reaction would be mostly trapped by
phenylisocyanate to give diphenylurea and the rate of
step (b) in path A would be strongly decreased. How-
ever, by the same token, a large amount of aniline, such
as that employed in this work, should increase the
importance of path A at the expense of path B.
Finally, it can be noted that the scenario described
above also explains why catalyst deactivation is much
faster during the synthesis of ureas than during that of
carbamates. Formation of metallic palladium clearly de-
rives from aggregation of insufficiently stabilized palla-
dium(0) complexes. Under the conditions in which the
synthesis of carbamates is performed, at least in our
group and the one of van Leeuwen, the oxidation of
the palladium(0) complex to palladium(II) is a fast reac-
tion (zero-order kinetics with respect to nitrobenzene)
and the resting state of the catalyst is a palladium(II)

M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
4527
complex, which cannot aggregate to metallic palladium.
4. Conclusions
However, as the aniline concentration is increased, the
rate of the reduction of palladium(II) increases and
can even become higher than that of its formation. Un-
der these conditions (first-order kinetics in nitrobenzene)
a palladium(0) will become the resting state of the cata-
lytic cycle, thus explaining the easy deactivation.
Carbonylation of nitroarenes is a complex reaction.
Even when an apparently simple catalytic system is em-
ployed, such as that based on palladium–phenanthroline
complexes, several reaction pathways and even different
reaction stoichiometries are clearly involved. This has
led to apparently contradictory results from several lab-
oratories. We have started several years ago to investi-
gate more deeply this reaction and, although a
complete picture is still far to be reached, several pieces
of information begins to fit a general scheme. In this pa-
per, we have shown that some apparently contradictory
results on the effect of chloride are probably all correct
and an explanation can be given for this. Some new find-
ings were also discussed, which also restrict the range of
possible mechanisms and will constitute the basis for
further investigations.
3
.2.4. Reaction stoichiometry
Based on the literature precedents and on our find-
ing that azoderivatives only originate from nitroben-
zene, in the Results section we have interpreted the
observed reagents consumption as deriving from the
two stoichiometries described in Eqs. (2) and (3),
respectively (the 1:1 and 1:5 processes). However, an
examination of the data reported in Tables 1–3 clearly
shows that with few exceptions, the most outstanding
of which corresponding to reactions in which a very
high concentration of aniline was present, the molar
amounts of aniline and nitrobenzene consumed are al-
ways very similar. In most cases the ratio is so close
to one that the difference is well within the experimen-
tal error. This occurs despite the amount of azoxyben-
zene, only deriving from nitrobenzene (Eq. (4)), varies
widely. The major process, that described by Eq. (2),
consumes equal amounts of the two reagents, so that
the larger inclusion of aniline-derived moieties in the
produced urea should derive from the 1:5 process.
However, that two completely independent reactions,
the synthesis of azoxybenzene (Eq. (4)) and the 1:5
process (Eq. (3)), exactly compensate each other to
give an overall 1:1 consumption of aniline and nitro-
benzene may coincidentally occur in one case, but it
is extremely unlikely to occur in most reactions. The
only reasonable explanation is that under most of
the conditions employed in this study, one only pro-
cess, rather than two independent ones, is operating.
Such process should have the stoichiometry shown
in Eq. (8):
5. Experimental
5.1. General procedure
All solvents were dried by standard procedures, dis-
tilled and stored under dinitrogen before use. Pd(trim-
ethylbenzoate) [15] and [Pd(Phen) ][BF ] [11,50] were
2 2 4 2
prepared as reported in the literature. Compounds
2a,b and 3a,b were prepared as previously reported
[40]. Nitrobenzene was purified by shaking with 10%
H SO , washing with water, and drying with Na SO ,
2 4 2 4
followed by distillation under dinitrogen and storage
under an inert atmosphere. Aniline was distilled and
stored under dinitrogen before use. 2,4-Dinitrotoluene
was recrystallized from methanol to remove water. All
other reagents were commercial products and were em-
ployed as received.
5.2. Catalytic reactions
2
PhNO
2
þ2PhNH
2
þ3CO!PhNðOÞ@NPh
þH
In a typical catalytic reaction, the catalyst, Phen,
PhNO , and, when required, PhNH and the acidic pro-
þPhNHCðOÞNHPhþ2CO
2
2
O
ð8Þ
2
2
To the best of our knowledge, no process of this kind
has ever been described or proposed in the literature
and we are now trying to provide an experimental basis
for its occurrence. The process must clearly be a multi-
stage one and the only information we can give on it
at the moment, based on the selectivity data reported
in the literature and collected here, is that it appears
to be favored by the presence of bases and inhibited
by both acids and chloride anion. Since formation of
azo derivatives as byproducts is at the moment the main
obstacle to commercialization of a technology based on
carbonylation of nitroarenes, understanding the mecha-
nism of their formation is of outmost importance for the
development of this field.
moter or [PPN][Cl] were weighed in a glass liner. The
liner was placed inside a Schlenk tube with a wide
mouth under dinitrogen and was frozen at ꢀ78 ꢀC with
dry ice, evacuated and filled with dinitrogen, after which
the solvent was added. After the solvent was also frozen,
the liner was closed with a screw cap having a glass
wool-filled open mouth which allows gaseous reagents
to exchange and rapidly transferred to a 200 ml stainless
steel autoclave with magnetic stirring. The autoclave
was then evacuated and filled with dinitrogen three
times. CO was then charged at room temperature at
the required pressure and the autoclave was immersed
in an oil bath preheated at the required temperature;
the autoclave took about 15 min to fully equilibrate at

4528
M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
the final temperature. Other experimental conditions are
reported in the captions to the tables and figures. At the
end of the reaction the autoclave was cooled with an ice
bath and vented.
(b) J.H. Grate, D.R. Hamm, D.H. Valentine, U.S. Patent US
,629,804 (1986);
c) J.H. Grate, D.R. Hamm, D.H. Valentine, U.S. Patent US
,600,793 (1986);
d) J.H. Grate, D.R. Hamm, D.H. Valentine, U.S. Patent US
,603,216 (1986);
(e) J.D. Gargulak, W.L. Gladfelter, J. Am. Chem. Soc. 116 (1994)
792.
4
(
4
(
4
5
5
.3. Products analysis
3
[
[
[
5] F. Ragaini, A. Ghitti, S. Cenini, Organometallics 18 (1999) 4925.
6] F. Ragaini, S. Cenini, J. Mol. Catal. A 161 (2000) 31.
7] (a) Chem. Week, March 9 (1977) 43;
.3.1. Synthesis of diphenylurea
At the end of the reaction, most of the produced
diphenylurea precipitated out of the solution. Naphtha-
lene (GC internal standard) and benzophenone (HPLC
internal standard) were added to the reaction mixture
and dissolved. Then the suspension was filtered through
a glass frit and the solid washed with additional toluene.
The solid was dried under vacuum and weighed. It is
pure diphenylurea and no trace of contaminants could
be detected by any spectroscopic or chromatographic
analysis. The solution was analyzed by HPLC (RP-18
column, methanol/water 55:45 as eluent) to quantify
the diphenylurea still remained in solution, and by gas
chromatography (Dani 8610 gas chromatograph,
equipped with a PS 255 column) for nitrobenzene, ani-
line, azo- and azoxybenzene. Aniline and nitrobenzene
were also quantified by HPLC analysis. The obtained
value are in good agreement with the values obtained
by GC analysis, but were affected by a larger standard
deviation and only the GC data have been employed.
(b) R. Rosenthal, J.G. Zajacek, U.S. Patent US 3,962,302 (1976).
Chem. Abstr. 85 (1976) 177053a;
(
(
(
c) R. Tsumura, U. Takaki, A. Takeshi, Ger. Offen. 2,635,490
1977). Chem. Abstr. 87 (1977) 23929v;
d) E.T. Shawl, H.S. Kesling Jr., U.S. Patent US 4,871,871 (1989).
Chem. Abstr. 112 (1990) 138777w.
[8] E. Drent, P.W.N.M. van Leeuwen, U.S. Patent US 4,474,978
(1984) [Eur. Pat. Appl. EP 86,281 (1983). Chem. Abstr. 100 (1984)
6
109x].
9] E. Drent, Eur. Pat. Appl. EP 231,045 (1987). Chem. Abstr. 109
1988) 128605n.
[
(
[
10] J. Stapersma, K. Steernberg, Eur. Pat. Appl. EP 296,686 (1988).
Chem. Abstr. 111 (1989) 133802v.
[11] A. Bontempi, E. Alessio, G. Chanos, G. Mestroni, J. Mol. Catal.
2 (1987) 67.
4
[
[
[
12] E. Alessio, G. Mestroni, J. Mol. Catal. 26 (1984) 337.
13] E. Alessio, G. Mestroni, J. Organomet. Chem. 291 (1985) 117.
14] E. Drent, Pure Appl. Chem. 62 (1990) 661.
[15] S. Cenini, F. Ragaini, M. Pizzotti, F. Porta, G. Mestroni, E.
Alessio, J. Mol. Catal. 64 (1991) 179.
[
16] P. Leconte, F. Metz, Eur. Pat. Appl. EP 330,591 (1989). Chem.
Abstr. 112 (1990) 78160c.
[
17] P. Wehman, G.C. Dol, E.R. Moorman, P.C.J. Kamer,
P.W.N.M. van Leeuwen, J. Fraanje, K. Goubitz, Organometallics
13 (1994) 4856.
5
.3.2. Synthesis of dimethyl 2,4-toluenedicarbamate
At the end of the reaction, part of the dicarbamate (1)
was present as a precipitate. Naphthalene (GC internal
standard) and benzophenone (HPLC internal standard)
were added to the reaction mixture and enough THF
was added to dissolve any solid. The solution was then
analyzed by HPLC (RP-18 column, methanol/water
[18] P. Wehman, V.E. Kaasjager, W.G.J. de Lange, F. Hartl, P.C.J.
Kamer, P.W.N.M. van Leeuwen, J. Fraanje, K. Goubitz,
Organometallics 14 (1995) 3751.
[19] P. Wehman, L. Borst, P.C.J. Kamer, P.W.N.M. van Leeuwen, J.
Mol. Catal. A 112 (1996) 23.
[20] P. Wehman, L. Borst, P.C.J. Kamer, P.W.N.M. van Leeuwen,
Chem. Ber./Recueil 130 (1997) 13.
5
5:45 as eluent; for 2,4-dinitrotoluene, 1, 2a,b, 3a,b)
and gas chromatography (PS 255 column; for 4a,b and
).
[
[
[
[
[
21] P. Wehman, H.M.A. Donge, A. Hagos, P.C.J. Kamer,
P.W.N.M. van Leeuwen, J. Organomet. Chem. 535 (1997) 183.
22] P. Wehman, P.C.J. Kamer, P.W.N.M. van Leeuwen, Chem.
Commun. (1996) 217.
5
23] F. Ragaini, S. Cenini, J. Mol. Catal. A 109 (1996) 1, and
references therein.
Acknowledgements
24] R. Santi, A.M. Romano, P. Panella, G. Mestroni, It. Pat. Appl.
MI96A 002071 (1996).
We thank the MIUR (Programmi di Ricerca Scientif-
ica di Rilevante Interesse Nazionale, PRIN 2003033857)
for financial support.
25] R. Santi, A.M. Romano, P. Panella, G. Mestroni, It. Pat. Appl.
MI96A 002072 (1996).
[26] R. Santi, A.M. Romano, R. Garrone, P. Panella, It. Pat. Appl.
MI97A 000433 (1997).
[
27] R. Santi, A.M. Romano, P. Panella, C. Santini, J. Mol. Catal. A
27 (1997) 95.
1
References
[28] R. Santi, A.M. Romano, P. Panella, G. Mestroni, A. Sessanta o
Santi, J. Mol. Catal. A 144 (1999) 41.
[
1] S. Cenini, F. Ragaini, Catalytic Reductive Carbonylation of
Organic Nitro Compounds, Kluwer Academic Press, Dordrecht,
[29] P. Leconte, F. Metz, A. Mortreux, J.A. Osborn, F. Paul, F. Petit,
A. Pillot, J. Chem. Soc., Chem. Commun. (1990) 1616.
[30] (a) F. Paul, J. Fischer, P. Ochsenbein, J.A. Osborn, Angew.
Chem., Int. Ed. Engl. 32 (1993) 1638;
1997, and references therein.
[
[
2] F. Paul, Coord. Chem. Rev. 20 (2000) 269, and references therein.
3] A.M. Tafesh, J. Weiguny, Chem. Rev. 96 (1996) 2035, and
references therein.
(b) F. Paul, J. Fischer, P. Ochsenbein, J.A. Osborn, C. R. Chimie
5 (2002) 267.
[
4] (a) J.H. Grate, D.R. Hamm, D.H. Valentine, U.S. Patent US
[31] A. Sessanta o Santi, B. Milani, G. Mestroni, E. Zangrando, L.
Randaccio, J. Organomet. Chem. 545–546 (1997) 89.
4,705,883 (1987);

M. Gasperini et al. / Journal of Organometallic Chemistry 690 (2005) 4517–4529
4529
[
[
[
[
[
[
[
32] N. Masciocchi, F. Ragaini, S. Cenini, A. Sironi, Organometallics
7 (1998) 1052.
33] F. Paul, J. Fisher, P. Ochsenbein, J.A. Osborn, Organometallics
7 (1998) 2199.
[41] S.B. Park, R.B. Standaert, Tetrahedron Lett. 40 (1999) 6557, and
references therein.
1
[42] J.S. Oh, S.M. Lee, J.K. Yeo, C.W. Lee, J.S. Lee, Ind. Eng. Chem.
Res. 30 (1991) 1456.
1
34] E. Gallo, F. Ragaini, S. Cenini, F. Demartin, J. Organomet.
Chem. 586 (1999) 190.
[43] C.W. Lee, S.M. Lee, J.S. Oh, J.S. Lee, Catal. Lett. 19 (1993) 217.
[44] C. Amatore, A. Jutand, A. Suarez, J. Am. Chem. Soc. 115 (1993)
9531, and references therein.
35] A. Sessanta o Santi, B. Milani, E. Zangrando, G. Mestroni, Eur.
J. Inorg. Chem. (2000) 2351.
36] M. Gasperini, F. Ragaini, S. Cenini, E. Gallo, J. Mol. Catal. A
[45] C. Amatore, M. Azzabi, A. Jutand, J. Am. Chem. Soc. 113 (1991)
8375.
2
04–205 (2003) 107.
37] F. Ragaini, E. Gallo, S. Cenini, J. Organomet. Chem. 593–594
2000) 109.
[46] A. Jutand, A. Mosleh, Organometallics 14 (1995) 1810.
[47] S. Kozuch, S. Shik, A. Jutand, C. Amatore, Chem. Eur. J. 10
(2004) 3072, and references therein.
(
38] F. Ragaini, C. Cognolato, M. Gasperini, S. Cenini, Angew.
Chem. Int. Ed. 42 (2003) 2886;
Angew. Chem. 115 (2003) 2992.
[48] P. Giannoccaro, J. Organomet. Chem. 470 (1994) 249.
[49] F. Ragaini, M. Gasperini, S. Cenini, in: Proceedings of the 14th
International Symposium on Homogeneous Catalysis (ISHC-14),
Munich, 2004, P0080.
[
[
39] F. Ragaini, M. Gasperini, S. Cenini, Adv. Synth. Catal. 346
(2004) 63.
[50] B. Milani, A. Anzilutti, L. Vicentini, A. Sessanta o Santi, E.
Zangrando, S. Geremia, G. Mestroni, Organometallics 16 (1997)
5064.
40] M. Gasperini, F. Ragaini, C. Cazzaniga, S. Cenini, Adv. Synth.
Catal. 347 (2005) 105.