Palladium/iodide catalyzed oxidative carbonylation of aniline to diphenylurea: Effect of ppm amounts of iron salts

Article abstract of DOI:10.1016/j.jcat.2018.11.010

The palladium/iodide couple is the most investigated catalytic system for the oxidative carbonylation of amines to give ureas or carbamates. In reinvestigating it, we found that the most prominent role of iodide is to etch the stainless steel of the autoclave employed in most of previous works, releasing in solution small amounts of iron salts. The latter are much better promoters than iodide itself. Iron and iodide have a complex interplay and, depending on relative ratios, can even deactivate each other. The presence of a halide is beneficial, but chloride is better than iodide in this respect. The ideal Fe/Pd ratio is around 10, but even an equimolar amount of iron with respect to palladium (0.02 mol% with respect to aniline, corresponding to 12 ppm Fe with respect to the whole solution) is sufficient to boost the activity of the catalytic system. Such small amount may also come from Fe(CO)₅ impurities present in the CO gas when stored in steel tanks. The role of the solvent has also been investigated. It was found that the reason for the better selectivity in some cases is at least in part due to a hydrolysis of the solvent itself, which removes the coproduced water.

Full text of DOI:10.1016/j.jcat.2018.11.010

Journal of Catalysis 369 (2019) 257–266
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Palladium/iodide catalyzed oxidative carbonylation of aniline to
diphenylurea: Effect of ppm amounts of iron salts
⇑
⇑
Francesco Ferretti , Edoardo Barraco, Claudia Gatti, Doaa R. Ramadan, Fabio Ragaini
Dipartimento di Chimica, Università degli Studi di Milano, via C. Golgi 19, 20133 Milano, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
The palladium/iodide couple is the most investigated catalytic system for the oxidative carbonylation of
amines to give ureas or carbamates. In reinvestigating it, we found that the most prominent role of iodide
is to etch the stainless steel of the autoclave employed in most of previous works, releasing in solution
small amounts of iron salts. The latter are much better promoters than iodide itself. Iron and iodide have
a complex interplay and, depending on relative ratios, can even deactivate each other. The presence of a
halide is beneficial, but chloride is better than iodide in this respect. The ideal Fe/Pd ratio is around 10,
but even an equimolar amount of iron with respect to palladium (0.02 mol% with respect to aniline, cor-
responding to 12 ppm Fe with respect to the whole solution) is sufficient to boost the activity of the cat-
alytic system. Such small amount may also come from Fe(CO)5 impurities present in the CO gas when
stored in steel tanks. The role of the solvent has also been investigated. It was found that the reason
for the better selectivity in some cases is at least in part due to a hydrolysis of the solvent itself, which
removes the coproduced water.
Received 13 May 2018
Revised 27 October 2018
Accepted 5 November 2018
Keywords:
Oxidative carbonylation
Aromatic amine
Diphenylurea
Palladium
Iodide
Iron chloride
Dimethoxyethane
Carbon monoxide
Ó 2018 Elsevier Inc. All rights reserved.
1. Introduction
The two alternatives that have been most heavily investigated both
in industrial and academic laboratories are the reductive carbony-
Isocyanates are commodity chemicals mostly employed in
polyurethane synthesis. Polyurethanes are widely applied in
almost every part of modern life in the form of plastic foams, coat-
ings, adhesives, sealants, elastomers and binders [1,2]. The annual
world production of isocyanates is several millions metric tons and
it is steadily increasing. The industrial interest for these com-
pounds is evident from the recent investments on new plants by
the major producers in both Europe and Asia. The industrial syn-
thesis of isocyanates involves the reaction of an amine with phos-
gene. From an economical point of view, the phosgene-based route
seems to be the most effective strategy and the well-established
phosgene technology appears to be difficult to replace. However,
phosgene is a very toxic and corrosive material that is listed as a
potential chemical weapon by the Chemical Weapon Convention.
In 2006, a large plant producing toluene diisocyanate (TDI) was
shut in Italy mainly because of the pressure from the local popula-
tion and authorities against the use of phosgene close to a densely
populated area. Thus, it is not surprising that an enormous effort
has been made to develop a phosgene-free routes to isocyanates.
lation of nitroarenes [3–6] and the oxidative carbonylation of ami-
nes [6–9]. The former can afford isocyanates in one step, but the
direct synthesis of these products requires forcing conditions and
the formed isocyanates tend to oligomerize under these condi-
tions. Thus, research in the last few decades have mostly been
focused on reactions performed in the presence of an excess of
an alcohol or amine. Under these conditions, the obtained products
are carbamates and ureas respectively. Both of them can be ther-
mally cracked in a separate step to give the desired isocyanate
and regenerate an equivalent of alcohol or amine [10–16]. Carba-
mates and ureas are also the products derived from the oxidative
carbonylation of amines (Scheme 1).
It should also be recalled that carbamates and urea are not only
possible intermediates in the production of isocyanate, but are
important chemicals themselves. Carbamates are final products
and synthetic intermediates for the pharmaceutical [17,18] and
agrochemical industries [19] and ureas have a variety of applica-
tion [20], from traditional ones such as fertilizers [21] and pesti-
cides [22] to more recent and sophisticated ones such as
receptors for anion recognition [23], biosensors [24] and pharma-
ceutically active agents [25,26]. Thus, the interest in the synthesis
of carbamates and ureas goes well beyond their use as intermedi-
ates for isocyanate production.
⇑
Corresponding authors.
E-mail addresses: francesco.ferretti@unimi.it (F. Ferretti), fabio.ragaini@unimi.it
(F. Ragaini).
https://doi.org/10.1016/j.jcat.2018.11.010
0021-9517/Ó 2018 Elsevier Inc. All rights reserved.

258
F. Ferretti et al. / Journal of Catalysis 369 (2019) 257–266
highly polar solvents (e.g. DMF) give faster reactions, but when ali-
phatic amines are employed as substrates, substantial amounts of
oxamides are formed as byproducts. Gabriele and coworkers [59]
obtained the highest selectivities in ureas by using dimethox-
yethane (DME) as solvent and this was our starting point.
2. Results and discussion
2.1. Role of iodide
Scheme 1. Reductive carbonylation of nitroarenes and oxidative carbonylation of
amines.
We first tried to reproduce some of the best results reported in
Ref. [59] (Table 1, entries 1 and 2) employing PdI2 as catalyst and KI
as promoter, working in a Teflon-coated autoclave (entry 3). To our
surprise, the activity of the catalytic system was quite lower than
expected.
The reductive carbonylation approach allows to save one step
on the route to carbamates and ureas, but can only be applied to
nitroarenes, since nitroalkanes are much more difficult to reduce
and usually displace a completely different reactivity with respect
to their aromatic counterparts. On the other hand, the oxidative
carbonylation route can be applied to both aromatic and aliphatic
amines. Our group has worked for many years on the reductive
carbonylation of nitroarenes [27–36]. Although this reaction and
the oxidative carbonylation reactions of amines may seem at first
sight very different, it has been shown that all the most active cat-
alytic systems for the former reaction proceed through a reaction
pathway in which the nitroarene is intermediately reduced to
the corresponding aromatic amine, which is in turn carbonylated
at a later stage [37–43]. During the latter stage, the metal is
reduced. The carbonylation step is thus the same in both processes
and the main difference between them is that the oxidant is the
nitroarene in one case and a separately added oxidant (in most
cases dioxygen, possibly in the form of air) in the other. We thus
engaged in a study of the oxidative carbonylation of amines, to
see if we could improve the current state of the art taking advan-
tage of our previous expertize. Among the different catalytic sys-
tems that have been described in the literature for this reaction,
the one that has been most investigated and appears to be closer
to fulfilling the requirements for an industrial application is that
based on the use of a palladium catalyst with an iodide salt or
iodine as a promoter [44–73]. The promoting effect of iodide/
iodine was first reported by Fukuoka while employing a heteroge-
neous catalyst [44–47], but the activity of this system was quite
low. The activity of heterogeneous systems has been later
improved by others [48,51,52,54,63,69,71], but the turnover fre-
quencies (TOF) that can be obtained are nevertheless one or two
orders of magnitude lower than those achievable with the use of
homogeneous palladium catalysts. We thus decided to focus on
homogeneous systems. The products of the reaction can be ureas,
oxamides (double carbonylation products) or carbamates. The lat-
ter are obtained if an alcohol is employed as solvent. However, at
least in the case of aromatic amines and analogously to what it
occurs during the reductive carbonylation of nitroarenes, carba-
mates are only formed at a later stage of the reaction by alcoholysis
of the primarily formed urea. Thus, we decided to skip the addition
of alcohols in a first investigation to focus on the carbonylation
step. Aniline was employed as substrate because aromatic iso-
cyanates are the most important and because oxidative carbonyla-
tion of aromatic amines is in general more difficult than that of the
more nucleophilic aliphatic amines and finding good conditions for
the carbonylation of the former may lead to a system more easily
extendable to the latter than the reverse. However, the possibility
to extend the obtained results to aliphatic amines was clear to our
mind. Thus, as a starting point we selected a system in the litera-
ture that would be not only highly active, but that would also give
high selectivity for both aromatic and aliphatic amines. Besides
some outstanding results obtained in non-conventional solvents
such as supercritical CO2 and ionic liquids [64,68,70], in general
Taking into account that Gabriele used a stainless-steel auto-
clave without Teflon coating, we considered that iodide ion and
the hydriodic acid formed during the reaction could partly dissolve
the metal of which the autoclave is composed. Thus, we performed
a reaction in a non-coated stainless-steel autoclave. In this way, we
could indeed reproduce the literature results (entry 4, Table 1).
The role of iodide has been variably interpreted in the previous
literature as being due to its adsorption onto heterogeneous palla-
dium nanoparticles [47,48], to the formation of N-iodoaniline [55],
to that of carbamoyl iodides (RNHC(O)I) [74], to the generation of
anionic low valent palladium complexes such as [Pd(CO)3I]À, con-
sidered to be the catalytically active species [50,57], and especially
to the ability of iodine, formed by iodide anion by the action of
dioxygen, to reoxidize palladium(0) complexes formed at the end
of the catalytic cycle back to palladium(II) [58–60,64,66,75]. Direct
oxidation of palladium(0) complexes by oxygen is possible [76],
but is generally too slow to avoid precipitation of metallic palla-
dium under carbonylation conditions. Indeed a co-catalyst, that
acts as an oxidation catalyst for palladium, must be present and
the I2+KI system is able to play this role by oxidizing even metallic
palladium and being quickly reoxidized by dioxygen [75]. How-
ever, most of the published oxidative carbonylation works where
performed in stainless steel autoclaves and only in a few cases
were Hastelloy autoclaves [48,51,52,54] or even less frequently
glass reactors [50,57,60] employed and these are not the cases in
which the highest activities were obtained.
2.2. Effect of iron-containing promoters
Iron is the most abundant element in stainless steel, so we
tested the use of iron salts as promoters for the catalytic reactions
[77]. In the literature some examples have been reported of palla-
Table 1
Comparison with previous results and effect of the autoclave internal coating in the
PdI2 catalyzed oxidative carbonylation of aniline.a
Entry
KI/Pd mol ratio
t (h)
PhNH2 conv. (%)b
Urea sel. (%)c
1d
2d
3e
4f
10
100
10
15
16
16
16
89.0
96.0
33.7
100
84.3
90.6
60.6
94.6
10
a
Experimental conditions: the reactions were carried out using 10 mmol of
aniline at 100 °C under 16 bar of CO and 4 bar of air, in DME (10 mL).
PdI2 = 0.01 mmol.
b
Based on starting aniline and on quantitative GC analysis using benzophenone
as an internal standard.
c
Based on reacted aniline and on quantitative HPLC analysis using benzophe-
none as an internal standard.
d
Data from Ref. [59].
Performed in a Teflon-lined stainless steel autoclave.
Performed in a stainless-steel autoclave.
e
f

F. Ferretti et al. / Journal of Catalysis 369 (2019) 257–266
259
dium catalyzed carbonylation reactions of nitroarenes that use iron
as co-catalyst, in which aniline appears to be an intermediate [6].
Moreover, recently Krogul and Litwinienko [67] have studied the
effect of iron on the PdPy2Cl2/I2 catalyzed oxidative carbonylation
of amines to carbamates and found an effect of this metal mostly
on selectivity. However, the effect of iron in the absence of iodine
was not investigated.
Since the reaction at 100 °C is relatively slow (less than 100
turnovers/hour on average) and having as a goal an industrial
application, we performed our reactions under harsher conditions,
150 °C, 40 bar of carbon monoxide and 10 bar of air. Note that the
ratio between the gases is such that the system is outside the CO/
O2 flammability limits [78]. This is not always the case for the sys-
tems reported in the literature. In general, increasing the O2 rela-
tive amount over the lower limit allows getting faster reactions,
but the associated risk would not be acceptable at an industrial
level. The catalyst loading was also decreased from 0.1 to
0.02 mol%.
and 15). Further increasing the iron amount over a Fe/Pd = 50 ratio
causes a decrease in selectivity (compare entries 11 and 12).
From the available data on the interplay between iron and
iodide, it emerges that the inhibitory role of iodide on the iron pro-
moting ability is stronger when the iodide amount is large with
respect to that of iron and the effect is larger if chloride is also pre-
sent, despite the presence of chloride being beneficial when no
iodide is added. Considering that both iodide and chloride are good
ligands for iron(II/III) complexes, the best explanation for the
trends observed is that when the total halide concentration greatly
exceeds that of iron, stable iron complexes (e.g. [FeIIX4]2À or [FeIII-
X6]3À) are formed, which are unable to enter the catalytic cycle.
Indeed, aniline can coordinate both to palladium and iron and
the formation of these complexes may be relevant to the outcome
of the catalytic reaction. Conversely, formation of stable halide
complexes also decreases the iodide concentration in solution, so
that even the promoting effect of iodide is compromised. We will
come back to discuss this aspect later in the paper.
We started to test different iron promoters, FeCl3, FeCl2, FeSO4-
Á2H2O, Fe(CO)5 and Fe(OAc)2, using PdI2 as the catalyst.
Since iodine is toxic and difficult to remove from the reaction
products, we tried to completely avoid its presence and tested both
Pd(OAc)2 and PdCl2 as catalysts (Table 3).
As a first control experiment, we run our reactions without the
catalyst, both with and without potassium iodide, and we checked
that no product was formed (entries 1 and 2, Table 2).
Palladium acetate alone gave a low conversion and a poor selec-
tivity into diphenylurea (entry 1, Table 3). Conversion was compa-
rable, but selectivity was quite low with respect to the use of PdI2
under the same conditions (compare entry 1, Table 3, with entry 3
in Table 2). The addition of iron salts not containing halides, FeSO4
and Fe(OAc)2, had a moderate effect on conversion, which may be
positive or even negative depending on the relative amounts, and a
positive effect on selectivity (entries 2–4). Notably, in the absence
of any halide, the addition of Fe(CO)5 inhibited the reaction (entry
5). This is an important observation because Fe(CO)5 is a typical
contaminant of carbon monoxide, especially when the latter is
stored in steel tanks. We employed aluminum alloy tanks during
all this work, which should minimize the Fe(CO)5 content in our
CO. The tank material was never declared in the works we cited,
but it is obvious that steel tanks were employed at least in the
older works, since aluminum alloy tanks have become of common
usage only more recently. Thus, it is clear that iron itself is not a
promoter and only when a suitable anion is present it efficiently
speeds up the reaction.
Comparing the results of the reactions performed with and
without an iron promoter, it clearly emerges that iron has a posi-
tive effect on the catalytic cycle, both on the conversion and on
the selectivity. The effect of iodide is on the other hand ambiguous.
We confirm that it accelerates the reaction when no iron is present
(entries 3 and 4, Table 2), but it inhibits it when an iron co-catalyst
is present. Iron acetate and iron sulfate give similar results in the
catalytic reactions without KI. An addition of the latter has a neg-
ative influence mainly on the conversion (entries 8 and 9). This
inhibitory effect can be noticed also in the reactions carried out
with FeCl3. Comparing the results in entries 15–17, in which the
promoter amount is the same, the conversion decreases with an
increase in the KI amount. Additionally, the negative influence of
KI is more evident in the presence of FeCl2 and FeCl3 (compare
entries 13 with 11 and 17 with 15). The best results were obtained
using FeCl2 and FeCl3 as co-catalysts without potassium iodide,
affording a relatively high conversion and selectivity (entries 11
Table 2
PdI2 catalyzed reactions using different iron promoters.a
Entry
KI/Pd mol ratio
Promoter
Promoter/Pd mol ratio
PhNH2 conv. (%)b
Urea sel. (%)c
1d
2d
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
–
100
–
100
100
100
100
–
–
–
–
–
<1
<1
–
–
20.3
31.5
40.2
46.8
19.0
61.7
20.0
64.7
95.8
95.0
38.4
81.6
100
91.8
89.4
95.7
30.4
88.7
51.2
78.1
45.1
79.0
80.5
92.0
85.8
79.8
70.3
74.0
66.7
70.4
72.5
68.6
63.6
68.9
Fe(CO)5
Fe(CO)5
FeSO4Á2H2O
FeSO4Á2H2O
FeSO4Á2H2O
Fe(OAc)2
FeCl2
FeCl2
FeCl2
FeCl2
FeCl3
50
100
30
50
50
50
50
100
50
100
50
100
–
–
–
100
100
–
10
FeCl3
FeCl3
FeCl3
FeCl3
50
50
100
10
100
100
100
a
Experimental conditions: the reactions were carried out in a Teflon-lined stainless-steel autoclave, using 10 mmol of aniline at 150 °C, for 3 h, under 40 bar of CO and
10 bar of air, in DME (10 mL). Catalyst amount 2.0 Â 10À3 mmol.
b
Based on starting aniline and measured by GC analysis using benzophenone as an internal standard.
c
Based on converted aniline and measured by HPLC analysis using benzophenone as an internal standard.
d
No palladium was added. The KI/Pd ratio reported is calculated based on a hypothetical 2.0 Â 10À3 mmol amount of palladium.

260
F. Ferretti et al. / Journal of Catalysis 369 (2019) 257–266
Table 3
Pd(OAc)2 and PdCl2 catalyzed reactions using different iron promoters and in the absence of any iodide source.a
Entry
Catalyst
Promoter
Promoter/Pd mol ratio
PhNH2 conv. (%)b
Urea sel. (%)c
1
2
3
4
5
6
7
8
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
PdCl2
–
22.2
30.7
19.0
8.3
50.0
76.7
75.3
55.4
55.1
82.3
79.4
74.3
81.2
76.9
72.9
76.6
70.5
86.9
85.0
14.3
FeSO4Á2H2O
Fe(OAc)2
Fe(OAc)2
Fe(CO)5
FeCl2
FeCl2
FeCl2
FeCl3
FeCl3
50
50
100
50
10
50
100
10
50
100
50
6.5
95.0
97.0
97.2
95.3
99.6
98.9
99.2
63.5
68.0
41.8
3.1
9
10
11
12
13
14d
15e
16
FeCl3
FeCl3
PdCl2
Pd(OAc)2
Pd(OAc)2
FeSO4Á2H2O
50
10
FeCl2
FeCl2
FeCl3
10
50f
a
Experimental conditions: the reactions were carried out in a Teflon-lined stainless-steel autoclave, using 10 mmol of aniline at 150 °C, for 3 h, under 40 bar of CO and
10 bar of air, in DME (10 mL). Catalyst amount 2.0 Â 10À3 mmol.
b
Based on starting aniline and measured by GC analysis using benzophenone as an internal standard.
Based on converted aniline and measured by HPLC analysis using benzophenone as an internal standard.
Reaction time 1 h.
c
d
e
Reaction time 30 min (see text for a discussion of this reaction).
f
The promoter/Pd ratio is calculated based on the standard 2.0 Â 10À3 mmol palladium amount, even if the reaction was run in the absence of any palladium compound.
Much better conversions and selectivities were achieved when
palladium acetate was combined with FeCl2 or FeCl3 (Table 3,
entries 6–11). A graphical representation of the obtained results
(Fig. 1) shows that the effect of FeCl2 and FeCl3 is very similar.
Clearly, the initial oxidation state of iron is not important, both
precursors will equilibrate to the same complex under the reaction
conditions, and the small observed difference is due to the slightly
different chloride content of the catalytic mixture.
Noteworthy, just 10 equiv of iron with respect to palladium
(0.2 mol% with respect to aniline) is sufficient to boost the activity
almost to its highest level. A further increase in the iron amount
has only a small positive effect on conversion, but lowers the selec-
tivity of the reaction.
Palladium chloride was also tested as catalyst in the presence of
either FeCl3 or FeSO4Á2H2O as promoters. The results were compa-
rable to those obtained employing palladium iodide as catalyst
under the same conditions (compare entry 12 in Table 3, with 15
in Table 2 and entry 13 in Table 3 with entry 8 in Table 2). On
the other hand, it is worth noting that reactions performed with
palladium acetate using Fe(CO)5, FeSO4Á2 H2O, and Fe(OAc)2 as pro-
moters gave worse results with respect to the use of the same pro-
moters with palladium iodide.
As a final test, a reaction was performed in the presence of
FeCl3, but in the absence of any palladium compound (entry 16).
A very low conversion and trace amounts of diphenylurea were
obtained, proving that iron itself has a negligible catalytic activity
in this reaction and also that no or negligible palladium contamina-
tion of the Teflon coating of the autoclave occurred.
Overall, the trends discussed above indicate that the best
results are obtained when palladium and iron are employed
together and in the presence of a halide. Iodide is no better than
chloride and it does not appear to be relevant whether the halide
is initially bound to either palladium or iron or both.
100
90
80
70
60
The results obtained when Fe(CO)5 is employed as promoter are
worth an additional comment. This compound promotes the reac-
tion when PdI2 is employed as catalyst, but even inhibits it when
Pd(OAc)2 is used. This is a confirmation that the presence of a
halide is necessary for iron to act as a promoter and a very small
amount of it may already have a measurable effect, even if a larger
amount is more effective (see later, paragraph 2.3). The result is
important in understanding the data reported in the previous liter-
ature. As previously mentioned, some Fe(CO)5 was undoubtedly
present in the CO gas of most earlier reports this has likely caused
the presence of some iron even when Hastelly or glass reactors
were employed. So the role of iodide may have been at least in part
linked to its interaction with iron even in those cases in which it
did not actively provide the required iron by etching the autoclave.
The possible reason for a negative effect of an excess of halide
has been discussed above. The positive effect of a small amount
of halide may be explained by the formation of bimetallic Pd-Fe
complexes with an halogen bridge (halides are much better bridg-
ing ligands than all other anions we tested), but a simpler effect of
halides on the redox potentials of the palladium and iron species
involved cannot be excluded at this stage.
Pd(OAc)2/FeCl2 conv
50
Pd(OAc)2/FeCl2 sel
40
Pd(OAc)2/FeCl3 conv
30
Pd(OAc)2/FeCl3 sel
20
PdI2/KI (100)/FeCl3 conv
10
PdI2/KI (100)/FeCl3 sel
0
0
10
20
30
40
50
60
70
80
90 100
FeCl2-FeCl3/Pd mol ratio
Fig. 1. Effect of FeCl2 and FeCl3 as co-catalysts in the oxidative carbonylation
reactions of aniline catalyzed by Pd(OAc)2 and PdI2+KI (mol ratio KI/Pd = 100). Data
from Tables 2 and 3.

F. Ferretti et al. / Journal of Catalysis 369 (2019) 257–266
261
Scheme 2. Diphenylurea hydrolysis.
Finally, it should be considered that at 150 °C, in 3 h, dipheny-
Table 4
Pd(OAc)2 catalyzed reactions using Fe(OAc)2 as promoter in the presence of either
Bu4N+ClÀ or Bu4N+IÀ a
lurea can be hydrolyzed by the water formed during the reaction
(Scheme 2). The so formed aniline may be carbonylated again,
but may also originates byproducts. Thus prolonging the reaction
time once the reaction is almost complete can only have negative
effects. In order to check this possibility, we repeated the reaction
corresponding to entry 6 in Table 3, stopping it after one hour or
after just 30 min. The results (entry 14 and 15 in Table 3) confirm
that a higher selectivity is obtained in a shorter reaction time.
We have independently tested diphenylurea hydrolysis under
the reaction conditions in Table 3, starting from the urea and water
concentrations that may be present at the end of the reaction if a
complete conversion of aniline into diphenylurea and water had
occurred, with no side reaction. In the absence of any other com-
pound and in the Teflon-lined autoclave, the hydrolysis was negli-
gible, but the presence of a FeCl3 amount equal to that used in run
12 in Table 3 caused 16% of the urea to hydrolyze. Palladium salts
should also be catalysts for this hydrolysis, but we could not add
them because they may have caused the back carbonylation of
some of the formed aniline.
.
Entry
Halide
Halide/Pd mol ratio
PhNH2 conv. (%)b
Urea sel. (%)c
1
2
3
4
5
6
7
8
–
–
50
19.0
54.3
79.9
85.2
93.5
94.7
73.8
60.9
36.7
40.0
27.4
75.3
81.2
79.0
83.9
89.0
89.7
85.6
82.0
85.5
83.3
74.5
ClÀ
ClÀ
ClÀ
ClÀ
ClÀ
IÀ
100
150
200
250
50
100
150
200
250
IÀ
9
10
11
IÀ
IÀ
IÀ
a
Experimental conditions: the reactions were carried out in
a Teflon-lined
stainless-steel autoclave, using 10 mmol of aniline at 150 °C, for 3 h, under 40 bar of
CO and 10 bar of air, in DME (10 mL). Catalyst amount 2.0 Â 10À3 mmol, mol ratio
Fe(OAc)2/Pd = 50.
b
Based on starting aniline and measured by GC analysis using benzophenone as
an internal standard.
c
Based on converted aniline and measured by HPLC analysis using benzophe-
none as an internal standard.
Notably, already a 41.8% conversion was observed after 30 min,
corresponding to a TOF of 4190/h (or 1780/h with respect to the
formed urea). However, note that the TOF values measured at a
short reaction time are strongly underestimated in our case.
Indeed, we placed our autoclaves in a preheated oil bath and con-
sidered that moment as the start of the reaction. However, the
insertion of the autoclave into the bath caused a temperature drop
of the oil temperature, which took 15 min to reach again the set
value. The content of the autoclave surely took even longer to equi-
librate due to the insulating power of the Teflon coating (2 mm
thickness). Thus the reaction was stopped very close to the
moment in which the reaction solution reached the set tempera-
ture or even earlier than it did [79].
Fig. 2 shows a clear trend with respect to chloride amount for
the catalytic reactions: both conversion and selectivity increase
with increasing tetrabutylammonium chloride content. The effect
is more remarkable on the conversion, but tends to flatten at the
highest ratios. The effect on selectivity is more limited, but still sig-
nificant. The ideal amount of Bu4N+ClÀ is 250 to 1 compared to the
catalyst (at a Fe/Pd ratio of 50), that is a 5:1 ratio with respect to
iron (or about 5:1 with respect to the sum Fe + Pd).
The effect of iodide is quite different, though not unexpected,
given the previous results. An amount of iodide equimolar to that
of iron promotes the reaction even more efficiently than the same
amount of chloride, but a further increase leads to an inhibition of
the reaction. At a 5:1 I/Fe mol ratio the results both in term of con-
version and selectivity are close to those obtained in the absence of
iodide, which are in turn close, at least for the conversion, to those
obtained with Pd(OAc)2 in the absence of both iron and iodide
(Table 3, entry 1).
It may appear surprising that the addition of increasing
amounts of chloride increases the conversion whereas that of
iodide has an inhibiting effect. However, it should be recalled that
the reaction produces water and the latter gives strong hydrogen
bonds to chloride, markedly decreasing its nucleophilicity and ten-
dency to coordinate to metals [80,81]. The same occurs with iodide
to a much lower extent. Indeed, it is well known that the order of
nucleophilicity is F > Cl > Br > I in non-protic solvents, but exactly
the reverse in water and alcohols [80,81]. Thus, the amount of
halide necessary to give the same number of coordinated halide
anions is larger for chloride than for iodide. If we naively suppose
than the coordinating ability of iodide is not affected at all by
water, then we reach the conclusion that the ideal ‘‘coordinated
halide”/(Fe+Pd) mol ratio is close to one. This is clearly a simplifi-
cation, but the actual number cannot be too far from it. Earlier in
this paper we have said that the easiest explanation for the inhibit-
ing effect of halides is the possible formation of stable iron com-
plexes of the kind [FeIIX4]2À or [FeIIIX6]3À, where coordination of
aniline would be inhibited. However, the results now obtained
2.3. The effect of different I/Fe and Cl/Fe ratios
Comparing the reactions performed with the Teflon lined
stainless-steel autoclave using Pd(OAc)2 (without KI) with those
made using PdI2 (with KI), in the presence of FeCl3 as co-catalyst
(Fig. 1), it is clear that the reactions lead to better results when
iodide is absent. Moreover, there is a very negative effect on con-
version when a low amount of FeCl3 is used, which further con-
firms the hypotheses made earlier that iodine coordinates to
iron. Indeed, the worse result is obtained when a low amount of
FeCl3 (Fe/Pd = 10) is added to a solution containing a much larger
amount of iodide (I/Pd = 100). Under these conditions, iron is prob-
ably completely deactivated and its only role is to withdraw iodide
from the solution, so that a mutual deactivation occurs.
To better investigate the effect of the amount of chloride and
iodide independently from the amount of iron, we decided to test
the effect of different amounts of either tetrabutylammonium
chloride or iodide on the activity of a catalytic system, Pd(OAc)2 +
Fe(OAc)2 (mol ratio Fe/Pd = 50), that does not initially contain any
halide. Results are reported in Table 4 and Fig. 2. Tetrabutylammo-
nium was chosen as a countercation to ensure complete solubility
of the salt under the reaction conditions and minimize electrostatic
interactions between the cation and the halide, which may alter
the reactivity of the latter.

262
F. Ferretti et al. / Journal of Catalysis 369 (2019) 257–266
100
90
80
70
60
50
40
30
20
10
0
idize palladium. However, it was also considered that a bimetallic
Pd-Fe complex with bridging chlorides might be the real catalytic
species [3,39]. At the time, the comparison between the reductive
carbonylation of nitroarenes and the oxidative carbonylation of
amines was not stressed for the palladium catalysts because it
seemed that the oxidative carbonylation reaction did not need a
second metal. However, the results just discussed strengthen the
similarity between the two reactions and led support to the
hypothesis that bimetallic species may be involved in both cases.
conv Cl-
sel Cl-
conv I-
sel I-
2.4. Reactions under milder conditions
The results shown up to now prove that a high conversion can
be reached in a short time at 150 °C, but the selectivity of the reac-
tion reaches a maximum around 90%, that would not be enough for
an industrial application. In an attempt to increase the selectivity
over this limit, we decided to investigate lower reactions temper-
atures, even if this means to slow down the reaction. Thus, we low-
ered the temperature to 100 °C and the pressure to a total of
20 bar, and increased the reaction time to 15 h (Table 5).
The effect of small amounts of iron promoter were first investi-
gated (Table 5, entries 1–5). The ideal Fe/Pd ratio is confirmed to be
around 10 even under these milder conditions, but it is noteworthy
that just a 1:1 ratio Fe/Pd is sufficient to double the conversion
with respect to the palladium-only system. This ratio corresponds
0
50
100
150
200
250
Bu4N+Cl- or Bu4N+I- /Pd mol ratio
Fig. 2. The effect of Bu4N+ClÀ and Bu4N+IÀ on the oxidative carbonylation of aniline
catalyzed by Pd(OAc)2 + Fe(OAc)2 (mol ratio Fe/Pd = 50). Data from Table 4.
to just two lmol of iron!
show that the amount of iodide sufficient to inhibit the reaction is
much lower than that required to generate such complexes. An
ideal 1:1 ‘‘coordinated halide”/(Fe+Pd) ratio is only consistent with
the formation of halogen-bridged Pd-X-Fe or Fe-X-Fe species as
catalytically active species [82–85]. Higher halide concentrations
would be sufficient to break such complexes even if considerably
lower than those required to completely saturate the coordination
sphere of the isolated palladium and iron ions.
As previously mentioned, iron compounds have been employed
in several cases as promoters for palladium-catalyzed reductive
carbonylation reactions of nitroarenes. By analyzing the available
information, mostly contained in the patent literature, one of us
had noted more than twenty years ago that all active systems also
contained chlorides [3]. Since much evidence also suggested that
amines were intermediates in the process, it was proposed for
the first time that the role of palladium might be that of carbony-
lating the amine and that of iron to reduce the nitroarene and reox-
It was also shown that 15 h are not required to reach complete
conversion and 6 h are enough (entry 6). A shorter time, 3 h, is
however not enough at this temperature, even if the pressure
had been increased back to the previously employed values (Entry
7).
Unfortunately, no benefit was observed on selectivity by lower-
ing the reaction temperature. The effect of the catalyst amount was
thus investigated by increasing it fivefold at a constant aniline con-
centration. Both PdI2 (entries 8–13) and Pd(OAc)2 (entries 14 and
15) were tested at this catalytic ratio, in the presence of either KI
or FeCl3 as promoters. The results obtained confirm the generality
of the observation made under more forcing conditions: KI is a less
effective promoter than FeCl3 for PdI2 and the catalytic system
lacking any iodide is better than those containing it are. However,
again no better selectivities were obtained by lowering the cat-
alytic ratio. On the contrary, a direct comparison between entries
6 and 15 evidences that better selectivities are obtained with less
Table 5
Synthesis of diphenylurea under milder conditions.a
Entry
Catalyst
Promoter
PhNH2/Pd mol ratio
Promoter/Pd mol ratio
t (h)
PhNH2 conv. (%)b
Urea sel. (%)c
1
2
3
4
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
Pd(OAc)2
PdI2
PdI2
PdI2
PdI2
PdI2
–
5000
5000
5000
5000
5000
5000
5000
1000
1000
1000
1000
1000
1000
1000
1000
–
1
3
15
15
15
15
15
6
3
3
6
16
6
37.2
71.9
90.5
100
100
100
75.5
9.0
19.4
33.7
91.5
96.0
89.9
100
100
86.8
83.2
85.5
83.4
82.1
80.8
87.9
28.7
26.2
60.6
80.4
67.7
66.7
81.9
74.6
FeCl3
FeCl3
FeCl3
FeCl3
FeCl3
FeCl2
KI
10
50
50
10
10
10
10
10
10
50
50
50
5
6
7d
8
9
KI
KI
10
11
12
13
14
15
FeCl3
FeCl3
FeCl3
FeCl3
FeCl3
15
15
15
6
PdI2
Pd(OAc)2
Pd(OAc)2
a
Experimental conditions: the reactions were carried out in a Teflon lined stainless-steel autoclave, using 10 mmol of aniline at 100 °C under 16 bar of CO and 4 bar of air,
in DME (10 mL).
b
Based on starting aniline and measured by GC analysis using benzophenone as an internal standard.
Based on converted aniline and measured by HPLC analysis using benzophenone as an internal standard.
Under 40 bar of CO and 10 bar of air.
c
d

F. Ferretti et al. / Journal of Catalysis 369 (2019) 257–266
263
catalyst, in agreement with the previously reached conclusion that
a higher absolute amount of FeCl3 causes a decrease in selectivity.
Scheme 3). It should be noted that the reaction between dipheny-
lurea and an alcohol to give a carbamate and an aniline equivalent
is well known. It typically occurs to a large extent when oxidative
carbonylation reactions of amines or the reductive carbonylation of
nitroarenes are performed in an alcohol as solvent. However, it is
an equilibrium reaction and only becomes significant in the pres-
ence of a large excess of alcohol with respect to the free amine
[37]. Thus, the observed carbamates cannot result from the pres-
ence of trace amounts of the two alcohols and their formation
implies that DME hydrolysis occurs to a significant extent during
the reaction.
Indole (d), on the other hand, can derive from a reaction of ani-
line with ethylene glycol, the doubly hydrolyzed DME. Such reac-
tion is known to occur under oxidizing conditions [87,88].
Two more byproducts could be identified, which were present
in trace amounts. Benzoxazolone (e) clearly derives from the oxi-
dation of aniline in the ortho position to give o-aminophenol. The
latter easily gives the cyclic carbamate under the reaction condi-
tions [89]. The formation of 3-methoxyquinoline (f) is less obvious.
The presence of the methoxy group anyway suggests that DME is
again involved in its formation.
In order to quantify at least the main byproducts, pure methyl
phenylcarbamate, 2-methoxyethyl phenylcarbamate, and indole
were used to set a calibration curve at the gas-chromatograph
and the reaction was repeated. The selectivities into these three
products were respectively 2.4, 1.1 and 1.0%.
The formation of these products raises some concern on the use
of DME as the solvent and suggests that one of the reasons for
which it seems to be more effective than others may simply be
the fact that it acts as internal drying agent, preventing the formed
water to hydrolyze the produced urea.
Another solvent that is often employed in oxidative carbonyla-
tion reactions is dioxane. Though we did not investigate the use of
dioxane in detail, the GC-MS analysis of the solution after a reac-
tion run in this solvent instead of DME evidenced again the forma-
tion of carbamates deriving from its hydrolytical ring opening. In
this context, it should also be mentioned that the solvent that often
gives the best results in most of the cited references, although with
low urea/oxalamide selectivity, is dimethylformamide. The
declared, and surely at least partly correct, rationale for this is that
it has a high dielectric moment, thus stabilizing the likely very
polar transition state of the reaction. However, it should not be for-
gotten that DMF is far from being an inert solvent [90]. Its hydrol-
ysis reaction has been studied in detail [91,92] and DMF has even
been used as a drying agent [93]! Clearly, this would not be an
acceptable procedure for an industrial production and a different
strategy must be pursued to reach complete conversion. This will
be the object of future work.
2.5. Reaction byproducts
Although many studies have been devoted to the oxidative car-
bonylation of amines with different catalysts, we are not aware of
any paper in which the identity of the byproducts (except for oxa-
lamides) has been investigated in detail. The observation of for-
manilide, derived from the reductive elimination from a PhNHC
(O)-Pd-H complex, has occasionally been mentioned [86].
In order to shed light on the lack of full selectivity for the oxida-
tive carbonylation reaction we subjected the reaction mixture of
several of the catalytic reactions showing high conversion, but
poor selectivity to GC-MS and HPLC-MS analyses.
From the GC-MS analyses (see the Supplementary Material), it
immediately emerges that apart from the aniline and phenyl iso-
cyanate peaks (which derive from a thermal cracking of the pro-
duct diphenylurea, also observable as a weak peak, under the
injector conditions) and trace amount of formanilide (a), some
smaller peaks can be observed. These are clearly identified in the
NIRST database as methyl phenylcarbamate (b), 2-methoxyethyl
phenylcarbamate (c) and indole (d). Some even smaller peaks
could be identified as benzoxazolone (e) and 3-methoxyquinoline
(f) (Scheme 3).
The HPLC chromatogram, on the other hand, was very clean and
only the diphenylurea peak could be clearly observed. Specifically,
no oligomeric polyanilines could be detected by this technique.
The latter would result from the direct oxidation of aniline by
dioxygen and would easily escape GC analysis.
The observation of phenyl isocyanate by gas-chromatography
should not be taken as a proof of its presence in solution without
other supporting evidence. Indeed, diphenylurea can extensively
thermally crack in the injector of the instrument. For example,
no aniline had remained at the end of the reactions investigated
by GC-MS in this work and yet the aniline peak is among the most
intense in the GC-MS spectrum (see the chromatogram reported in
the Supplementary Material). Gas-chromatographic analysis of
aniline and phenyl isocyanate should only been performed by
employing a programmed variable temperature injector, so that
the temperature can be increased to the final value only after the
more volatile (including aniline) compounds of the mixture have
left the injector. We routinely used such a procedure in the quan-
titative analyses of our reaction mixtures and never observed the
presence of phenyl isocyanate. Thus, we are confident that the iso-
cyanate observed during the GC-MS analysis is an artefact of the
analytical technique.
Although the formation of the products in Scheme 3 was unex-
pected, it is not difficult to trace back their origin. They are all
clearly originating from the hydrolysis of the dimethoxyethane sol-
vent by the water formed during the reaction. The two alcohols so
formed originate the two observed carbamates (b and c in
3. Conclusions
In this work we have investigated in more depth the classical
Pd/IÀ catalytic system for the oxidative carbonylation reaction of
amines to ureas, which has been previously the topic of at least
30 papers. Some surprising results have been found that lead to
extensively reconsider what is known of this catalytic system:
1) We confirmed that iodide alone promotes the activity of pal-
ladium in the oxidative carbonylation reaction of amines,
but its promoting ability is much lower than previously
estimated.
2) Under most of the previously employed experimental condi-
tions, the main role of iodide is to etch the autoclave walls to
provide small amounts of iron in solution.
Scheme 3. Byproducts observed by GC-MS.

264
F. Ferretti et al. / Journal of Catalysis 369 (2019) 257–266
3) When iron etching from the autoclave walls cannot occur,
Methoxyethyl phenylcarbamate was prepared as described in the
Supplementary Material. All the other reagents and solvents were
purchased from Sigma-Aldrich or Alfa-Aesar and used without any
further purification.
small, but sufficient, amounts of iron may still come from
the Fe(CO)5 present as a contaminant of pressurized CO in
all those cases in which iron tanks were used to store it.
4) The ideal absolute iron amount under the presently
employed conditions (930 mg aniline) is just 1.1 mg, but
even a tenfold lower amount (around 12 ppm with respect
to the whole solution weight) is sufficient to double the
activity of the palladium catalyst. It is worth to note that
contamination of iron catalysts by noble metals is well
known to be responsible for the catalytic activity observed
in some cases. However, the present work is a rare, if not
unique, case in which the contamination of a palladium cat-
alysts by trace amounts of iron has such a large promoting
effect.
5) The presence of a halide is anyway required, but chloride can
be equally or even more effective than iodide in this respect.
The ideal amounts of chloride and iodide differ, likely
because chloride coordination to metal ions can be ham-
pered by the formation of strong hydrogen bonds with the
water formed during the reaction, whereas iodide is less
affected by the presence of water. A molar amount of iodide
larger than that of iron causes a deactivation of the system,
suggesting that the active catalytic species may be a Pd/Fe
dinuclear or higher aggregate hold together by bridging
halide ligands.
6) The reason for the good efficiency of DME as a reaction sol-
vent is at least partly due to its ability to act as a dehydrating
agent, but the alcohols formed by its hydrolysis can reenter
the reaction and give small amounts of the corresponding
carbamates and even some indole. Hydrolysis of the solvent
is likely involved even in other cases.
7) Though usually not mentioned, iodide is toxic and difficult to
remove completely from the reaction products. This is espe-
cially a problem if the produced urea is employed as an
intermediate in the synthesis of pharmaceuticals. The possi-
bility of substituting it with a very low amount of non-toxic
iron chloride should be viewed favorably from an industrial
point of view.
4.2. Catalytic reactions
All the catalytic reactions were performed in a stainless steel
autoclave lined with Teflon (c.a. 140 mL free volume), or, for the
control experiments, directly in a stainless-steel autoclave, both
equipped with a stirring bar. Since small amounts of catalyst were
used, a stock solution of the catalyst (PdI2 or PdCl2, the latter added
as PdCl2(CH3CN)2) in aniline was prepared and stored under dini-
trogen atmosphere. The solution already contained the correct
Pd/aniline ratio, so that the catalyst was weighed together with
aniline. When Pd(OAc)2 was used as the catalyst it was freshly dis-
solved in DME and the appropriate amount of the obtained solu-
tion was added by volume. In a typical catalytic reaction, the
reagents were transferred to the autoclave following always the
same order. First KI, weighted under air, was added. Then the ani-
line solution, also containing PdI2 when appropriate, was rapidly
weighted under air in a test tube and transferred to the autoclave.
To ensure a complete transfer of the reagents to the autoclave, the
test tube was washed with the reaction solvent. When Pd(OAc)2
was used as the catalyst, its solution in DME was added by volume
and aniline was weighted in a test tube separately. The remaining
amount of solvent was then added by volume, using it to wash the
test tube used to weigh aniline. Finally, the iron promoter was
added. When the amount of the iron promoter was lower than
10 mg, a freshly prepared solution of it in DME was prepared under
dinitrogen atmosphere and then it was added by volume to the
reaction mixture. Reagent amounts or their molar ratios are
reported in the tables or in their footnotes. In all cases, the amine
concentration was 1 M. The autoclave was then closed and CO and
air were charged in this order at room temperature. The value of
pressure after each gas addition was read after complete stabiliza-
tion of the system (i.e. when it remained constant for at least
5 min). Finally, the autoclave was immersed in an oil bath pre-
heated at the required temperature. This moment was taken as
the start of the reaction. At the end of the reaction, the autoclave
was quickly cooled in an ice bath and vented. The internal stan-
dard, benzophenone, was added to the reaction mixture (1/4 mass
ratio with respect to the starting aniline) and 25 mL of THF were
also added to completely dissolve both benzophenone and any pre-
cipitated diphenylurea. The solution was kept under stirring for
half an hour. An aliquot of the obtained solution was then diluted
with CH2Cl2 in order to perform the GC and HPLC analyses.
The present work has clarified some previously overlooked
aspects of the Pd/IÀ catalytic system, but at the same time has
raised new questions both of industrial and scientific nature: Stain-
less steel contains other metals apart from iron: do any of them
also show any promoting activity? Which is the best way to
remove the formed water without resorting to solvent hydrolysis
or to the use of any reagent that is stoichiometrically consumed?
Which is the exact interplay between palladium, iron and the
halide? May a better catalyst be prepared by knowing this? Work
is in progress in our laboratories to answer at least some of them.
4.3. Quantitative analysis methods for catalytic reactions
For catalytic reactions, quantitative analyses were performed on
a DANI 86.10 HT gas chromatograph equipped with a SUPELCO
Analytical SLBTM-5 ms column (Fused Silica Capillary Column
4. Experimental
4.1. Materials and general procedures
30 m  0.32 mm  0.5
lm film thickness), on a Shimadzu GC-
2010 equipped with a Supelco SLBTM-5 ms capillary column (GC-
FAST technique), or on a HP 1050 series modular HPLC system
equipped with a MERCK LiChroCARTÒ 125-4 HPLC-Cartridge Puro-
spherÒ RP-18e (5 mm). HPLC grade solvents – CHROMASOLVÒ were
used for HPLC analyses, which were degassed by sonication in an
ultrasonic bath for half an hour before use. Quantitative analyses
were carried out using the internal standard method. Benzophe-
none was used as the internal standard. The wavelength of the
UV detector was set at 265 nm because this frequency is close to
the maximum absorption of both diphenylurea and benzophenone,
thus maximizing the accuracy of the analysis. Unfortunately,
aniline absorbs weakly at this wavelength and its analysis is less
1,2-Dimethoxyethane used in the catalytic reactions was dried
by distillation over Na/benzophenone and stored under a dinitro-
gen atmosphere. Aniline was distilled over KOH under reduced
pressure and stored under dinitrogen. It can be weighed under
air without problems, but must be stored under an inert atmo-
sphere to avoid oxidation, carbonation and water uptake. Pd
(OAc)2 [94], and PdI2 [95] were prepared by literature methods.
PdCl2(CH3CN)2 was prepared by refluxing PdCl2 in acetonitrile until
the brown solid had completely dissolved and a yellow solution
had formed. The solution was filtered while hot to remove any
traces of undissolved material and evaporated to dryness. 2-

F. Ferretti et al. / Journal of Catalysis 369 (2019) 257–266
265
[24] G. Dhawan, G. Sumana, B.D. Malhotra, Biochem. Eng. J. 44 (2009) 42–52,
https://doi.org/10.1016/j.bej.2008.07.004.
[25] H.Q. Li, P.C. Lv, T. Yan, H.L. Zhu, Anti-Cancer Agents Med. Chem. 9 (2009) 471–
480, https://doi.org/10.2174/1871520610909040471.
[26] J. Dumas, R.A. Smith, T.B. Lowinger, Curr. Opin. Drug Discov. Dev. 7 (2004)
600–616.
[27] S. Cenini, M. Pizzotti, C. Crotti, F. Ragaini, F. Porta, J. Mol. Catal. 49 (1988) 59–
69.
[28] F. Ragaini, S. Cenini, A. Fumagalli, C. Crotti, J. Organomet. Chem. 428 (1992)
401–408.
[29] F. Ragaini, E. Gallo, S. Cenini, J. Organomet. Chem. 594 (2000) 109–118.
[30] F. Ragaini, C. Cognolato, M. Gasperini, S. Cenini, Angew. Chem., Int. Ed. 42
(2003) 2886–2889.
[31] F. Ragaini, M. Gasperini, S. Cenini, Adv. Synth. Catal. 346 (2004) 63–71.
[32] M. Gasperini, F. Ragaini, C. Cazzaniga, S. Cenini, Adv. Synth. Catal. 347 (2005)
105–120.
[33] M. Gasperini, F. Ragaini, S. Cenini, E. Gallo, J. Mol. Catal. A: Chem. 204–205
(2003) 107–114.
[34] F. Ferretti, F. Ragaini, R. Lariccia, E. Gallo, S. Cenini, Organometallics 29 (2010)
1465–1471.
[35] F. Ferretti, E. Gallo, F. Ragaini, J. Organomet. Chem. 771 (2014) 59–67, https://
doi.org/10.1016/j.jorganchem.2014.04.010.
[36] F. Ferretti, E. Gallo, F. Ragaini, ChemCatChem 7 (2015) 2241–2247, https://doi.
org/10.1002/cctc.201500452.
[37] J.D. Gargulak, W.L. Gladfelter, J. Am. Chem. Soc. 116 (1994) 3792–3800.
[38] F. Ragaini, S. Cenini, F. Demartin, Organometallics 13 (1994) 1178–1189.
[39] F. Ragaini, S. Cenini, J. Mol. Catal. A: Chem. 109 (1996) 1–25.
[40] F. Ragaini, M. Macchi, S. Cenini, J. Mol. Catal. A: Chem. 127 (1997) 33–42.
[41] F. Ragaini, A. Ghitti, S. Cenini, Organometallics 18 (1999) 4925–4933.
[42] F. Ragaini, S. Cenini, J. Mol. Catal. A: Chem. 161 (2000) 31–38.
[43] F. Ragaini, M. Gasperini, S. Cenini, L. Arnera, A. Caselli, P. Macchi, N. Casati,
Chem. Eur. J. 15 (2009) 8064–8077.
reliable. This is the reason for which it was routinely quantified by
GC.
A standard GC analysis involves the preparation of a sample
solution in CH2Cl2. An aliquot of the reaction solution, treated as
mentioned above, was diluted in CH2Cl2 in order to obtain a final
standard concentration of 0.05 mg/mL for the analysis made with
DANI 86.10 HT and 0.1 mg/mL for the analysis made with Shi-
madzu GC-2010. Then 1 mL was injected into the GC. This analysis
method was used to determine the residual amount of aniline after
the catalytic reaction because it showed a higher reproducibility
with respect to HPLC analysis. A complete list of the instrumental
parameters employed is present in the Supplementary Material.
A standard HPLC analysis involves the preparation of a sample
solution in CH3OH/H2O 55:45. An aliquot of the previously pre-
pared reaction solution was diluted with CH3OH/H2O 55:45 in
order to obtain a final standard concentration of 0.01 mg/mL of
benzophenone. Then 100 mL of this sample solution were injected
into the HPLC (injection loop 20 mL). This analysis method was used
to determine the amount of diphenylurea.
Acknowledgements
This work was supported by the Italian Ministero dell’Univer-
sità e della Ricerca (MiUR) (PRIN 20154X9ATP). F.F. thanks the
Università degli Studi di Milano for a postdoctoral fellowship.
[44] S. Fukuoka, M. Chono, Eur. Pat. Appl. EP83096A2, to Asahi Chemical Industry
Co., Ltd., Japan, 1983.
[45] S. Fukuoka, M. Chono, M. Kohno, CHEMTECH 14 (1984) 670–676.
[46] S. Fukuoka, M. Chono, M. Kohno, J. Chem. Soc., Chem. Commun. (1984) 399–
400.
Appendix A. Supplementary material
[47] S. Fukuoka, M. Chono, M. Kohno, J. Org. Chem. 49 (1984) 1458–1460, https://
doi.org/10.1021/jo00182a035.
[48] S.P. Gupte, R.V. Chaudhari, J. Catal. 114 (1988) 246–258, https://doi.org/
10.1016/0021-9517(88)90028-0.
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.jcat.2018.11.010.
[49] I. Pri-Bar, H. Alper, Can. J. Chem. 68 (1990) 1544–1547.
[50] S.A.R. Mulla, S.P. Gupte, R.V. Chaudhari, J. Mol. Catal. 67 (1991) L7–L10, https://
doi.org/10.1016/0304-5102(91)80038-5.
References
[51] S.P. Gupte, R.V. Chaudhari, Ind. Eng. Chem. Res. 31 (1992) 2069–2074.
[52] A.A. Kelkar, D.S. Kolhe, S. Kanagasabapathy, R.V. Chaudhari, Ind. Eng. Chem.
Res. 31 (1992) 172–176.
[53] K.L. Loh, P. Shieh, J.L. Chen, T.K. Chuang, US Pat. US5101063A, to Industrial
Technology Research Institute, Taiwan, 1992.
[54] S. Kanagasabapathy, S.P. Gupte, R.V. Chaudhari, Ind. Eng. Chem. Res. 33 (1994)
1–6.
[55] I. Pri-Bar, J. Schwartz, J. Org. Chem. 60 (1995) 8124–8125.
[56] V.L.K. Valli, H. Alper, Organometallics 14 (1995) 80–82.
[57] S.A.R. Mulla, C.V. Rode, A.A. Kelkar, S.P. Gupte, J. Mol. Catal. A: Chem. 122
(1997) 103–109.
[58] B. Gabriele, R. Mancuso, G. Salerno, M. Costa, Chem. Commun. (2003) 486–487.
[59] B. Gabriele, G. Salerno, R. Mancuso, M. Costa, J. Org. Chem. 69 (2004) 4741–
4750.
[60] K. Hiwatari, Y. Kayaki, K. Okita, T. Ukai, I. Shimizu, A. Yamamoto, Bull. Chem.
Soc. Jpn. 77 (2004) 2237–2250, https://doi.org/10.1246/bcsj.77.2237.
[61] S. Zheng, X. Peng, J. Liu, W. Sun, C. Xia, Helv. Chim. Acta 90 (2007) 1471–1476.
[62] M.R. Didgikar, S.S. Joshi, S.P. Gupte, M.M. Diwakar, R.M. Deshpande, R.V.
Chaudhari, J. Mol. Catal. A: Chem. 334 (2010) 20–28.
[63] M.R. Didgikar, D. Roy, S.P. Gupte, S.S. Joshi, R.V. Chaudhari, Ind. Eng. Chem. Res.
49 (2010) 1027–1032.
[64] N. Della Ca, P. Bottarelli, A. Dibenedetto, M. Aresta, B. Gabriele, G. Salerno, M.
Costa, J. Catal. 282 (2011) 120–127, https://doi.org/10.1016/j.jcat.2011.06.003.
[65] M.R. Didgikar, S.S. Joshi, S.P. Gupte, M.M. Diwakar, R.M. Deshpande, R.V.
Chaudhari, J. Mol. Catal. A: Chem. 334 (2011) 20–28.
[1] H.-W. Engels, H.-G. Pirkl, R. Albers, R.W. Albach, J. Krause, A. Hoffmann, H.
Casselmann, J. Dormish, Angew. Chem., Int. Ed. 52 (2013) 9422–9441.
[2] See the American Chemistry Councyl site at <http://polyurethane.
americanchemistry.com/> (accessed May 10, 2018).
[3] S. Cenini, F. Ragaini, Catalytic Reductive Carbonylation of Organic Nitro
Compounds, Kluwer Academic Publishers, Dordrecht, The Netherlands, 1996.
[4] A.M. Tafesh, J. Weiguny, Chem. Rev. 96 (1996) 2035–2052.
[5] F. Paul, Coord. Chem. Rev. 203 (2000) 269–323.
[6] F. Ragaini, Dalton Trans. (2009) 6251–6266, https://doi.org/10.1039/
b902425p.
[7] B. Gabriele, G. Salerno, M. Costa, Top. Organomet. Chem. 18 (2006) 239–272.
[8] D.J. Diaz, A.K. Darko, L. McElwee-White, Eur. J. Org. Chem. (2007) 4453–4465.
[9] X.-F. Wu, H. Neumann, M. Beller, Chemsuschem 6 (2013) 229–241.
[10] Y. Dai, Y. Wang, J. Yao, Q. Wang, L. Liu, W. Chu, G. Wang, Catal. Lett. 123 (2008)
307–316, https://doi.org/10.1007/s10562-008-9424-6.
[11] M. Kloetzer, H.-j. Blankertz, G. Georgi, E. Stroefer, V. Krase, G. Schulz, A.
Warzecha, United States US 8193385 B2, to BASF SE (Ludwigshafen, DE), 2012.
[12] A.Y. Samuilov, F.B. Balabanova, Y.D. Samuilov, A.I. Konovalov, Russ. J. Gen.
Chem. 82 (2012) 1110–1114, https://doi.org/10.1134/s1070363212060126.
[13] R. Rosenthal, J.G. Zajacek, United States US 3936484, to Atlantic Richfield
Company (Los Angeles, CA), 1976.
[14] E.T. Shawl, H.S. Kesling, Jr., US Patent US 4871871, to (Arco Chemical
Technology, Inc., USA), 1989.
[15] E. Stroefer, W. Mackenroth, M. Sohn, C. Knoesche, O. Schweers, PCT Int. Appl.
WO 2008031755 A1 20080320, to (BASF Aktiengesellschaft, Germany), 2008.
[16] F.L. Serrano Fernandez, B. Almena Munoz, A. Padilla Polo, A. Orejon Alvarez, C.
Claver Cabrero, S. Castillon Miranda, P. Salagre Carnero, A. Aghmiz, U.S. Pat.
Appl. Publ. US 2007293696 A1 20071220, to (Repsol-Ypf,S.A., Spain), 2007.
[17] S. Ray, D. Chaturvedi, Drug Future 29 (2004) 343–357, https://doi.org/10.1358/
dof.2004.029.04.787236.
[18] S. Ray, S.A. Pathak, D. Chaturvedi, Drug Future 30 (2005) 161–180, https://doi.
org/10.1358/dof.2005.030.02.869228.
[19] R.L. Metcalf, Insect Control, Ullmann’s Encyclopedia of Industrial Chemistry,
Wiley-VCH Verlag GmbH & Co. KGaA, 2000.
[20] N. Volz, J. Clayden, Angew. Chem., Int. Ed. 50 (2011) 12148–12155.
[21] T. Li, S.Y. Lv, J. Chen, C.M. Gao, S.F. Zhang, M.Z. Liu, Acta Polym. Sin. (2018) 336–
348, https://doi.org/10.11777/j.issn1000-3304.2017.17152.
[22] S.R. Sorensen, G.D. Bending, C.S. Jacobsen, A. Walker, J. Aamand, FEMS
Microbiol Ecol. 45 (2003) 1–11, https://doi.org/10.1016/s0168-6496(03)
00127-2.
[66] R. Mancuso, D.S. Raut, N. Della, Ca, F. Fini, C. Carfagna, B. Gabriele,
Chemsuschem 8 (2015) 2204–2211, https://doi.org/10.1002/cssc.201500343.
[67] A. Krogul, G. Litwinienko, J. Mol. Catal. A: Chem. 407 (2015) 204–211, https://
doi.org/10.1016/j.molcata.2015.06.027.
[68] N. Zahrtmann, C. Claver, C. Godard, A. Riisager, E.J. Garcia-Suarez,
ChemCatChem
cctc.201800004.
10
(2018)
2450–2457,
https://doi.org/10.1002/
[69] P. Toochinda, S.S.C. Chuang, Ind. Eng. Chem. Res. 43 (2004) 1192–1199.
[70] X. Peng, F. Li, X. Hu, C. Xia, C.A. Sandoval, Chin. J. Catal. 29 (2008) 638–642,
https://doi.org/10.1016/S1872-2067(08)60062-5.
[71] M. Liang, T.J. Lee, C.C. Huang, K.Y. Lin, J. Chin. Chem. Soc. 54 (2007) 885–892,
https://doi.org/10.1002/jccs.200700128.
[72] S.T. Gadge, E.N. Kusumawati, K. Harada, T. Sasaki, D. Nishio-Hamane, B.M.
Bhanage, J. Mol. Catal. A: Chem. 400 (2015) 170–178, https://doi.org/10.1016/
j.molcata.2015.01.009.
[73] Z. Shu-Zhan, P. Xin-Gao, L. Jian-Ming, S. Wei, X. Chun-Gu, Chin. J. Chem. 25
(2007) 1065–1068, https://doi.org/10.1002/cjoc.200790196.
[23] V. Amendola, L. Fabbrizzi, L. Mosca, Chem. Soc. Rev. 39 (2010) 3889–3915,
https://doi.org/10.1039/b822552b.

266
F. Ferretti et al. / Journal of Catalysis 369 (2019) 257–266
[74] M. Aresta, P. Giannoccaro, I. Tommasi, A. Dibenedetto, A.M. Manotti Lanfredi, F.
Ugozzoli, Organometallics 19 (2000) 3879–3889.
they may be involved at some stage of the reaction, following or preceeding a
palladium-catalyzed reaction.
[75] B. Gabriele, M. Costa, G. Salerno, G.P. Chiusoli, J. Chem. Soc., Perkin Trans. 1
(1994) 83–87, https://doi.org/10.1039/p19940000083.
[76] S.S. Stahl, Angew. Chem., Int. Ed. 43 (2004) 3400–3420.
[83] P. Braunstein, E. Colomer, M. Knorr, A. Tiripicchio, M. Tiripicchio Camellini, J.
Chem. Soc., Dalton Trans. (1992) 903–909, https://doi.org/10.1039/
DT9920000903.
[77] Reaction 4 in Table 1 was repeated to check the presence of iron in solution
and the result was compared with that obtained when a reaction is performed
under exactly the same conditions, but with palladium acetate as a catalyst
and without any source of iodine. The results showed indeed a higher iron
content in the reaction solution after the reaction when iodide was present,
but technical problems in completely mineralizing the organic compounds
prevented us to get reliable quantitative data by ICP.
[78] The flammability range of CO in O2 is 16.7–93.5% at RT. Kirk-Othmer
Encyclopedia of Chemical Technology, third ed, vol 4, John Wiley and Sons,
pp. 774.
[84] F. Balegroune, P. Braunstein, J. Durand, T. Faure, D. Grandjean, M. Knorr, M.
Lanfranchi, C. Massera, X. Morise, A. Tiripicchio, Monatsh. Chem. 132 (2001)
885–896, https://doi.org/10.1007/s007060170052.
[85] D. Serra, M.C. Correia, L. McElwee-White, Organometallics 30 (2011) 5568–
5577, https://doi.org/10.1021/om101070z.
[86] Alkylated amines and alkylquinolines have been detected as byproducts when
the synthesis of carbamates is performed and an alcohol is employed as
solvent [67]. However, these byproducts derive from secondary reactions
involving the alcohol.
[87] M. Campanati, S. Franceschini, O. Piccolo, A. Vaccari, J. Catal. 232 (2005) 1–9,
https://doi.org/10.1016/j.jcat.2005.02.014.
[79] Higher TOF values have been reported in
a few cases among the cited
literature. While it is difficult to compare rate values obtained by using
different pressure equipment and we claim no record activity for our system,
we want to emphasize that the highest TOF values have been obtained by
running reactions for a short time (15-20 min). Moreover, it appears that the
reaction time was measured from the onset of the reaction temperature,
which may take up to one hour to be reached. Clearly, the reaction starts
before and the heating time contributes to the proceeding of the reaction in a
non-negligible way. On the contrary, we placed our autoclaves in a preheated
oil bath and considered that moment as the start of the reaction.
[88] T. Nishida, Y. Tokuda, M. Tsuchiya, J. Chem. Soc., Perkin Trans. 2 (1995) 823–
830, https://doi.org/10.1039/P29950000823.
[89] B. Gabriele, R. Mancuso, G. Salerno, Eur. J. Org. Chem. (2012) 6825–6839,
https://doi.org/10.1002/ejoc.201200794.
[90] J. Muzart, Tetrahedron 65 (2009) 8313–8323.
[91] A.D. Burrows, K. Cassar, R.M.W. Friend, M.F. Mahon, S.P. Rigby, J.E. Warren,
Crystengcomm 7 (2005) 548–550, https://doi.org/10.1039/b509460g.
[92] W.C. Cooper, A. Chilukoorie, S. Polam, D. Scott, F. Wiseman, J. Phys. Org. Chem.
30 (2017), https://doi.org/10.1002/poc.3701.
[80] D.M. Branan, N.W. Hoffman, E.A. McElroy, N.C. Miller, D.L. Ramage, A.F. Schott,
S.H. Young, Inorg. Chem. 26 (1987) 2915–2917.
[93] P. Supsana, T. Liaskopoulosa, P.G. Tsoungas, G. Varvounis, Synlett (2007) 2671–
2674, https://doi.org/10.1055/s-2007-991067.
[81] D. Forster, Inorg. Chem. 11 (1972) 1686–1687.
[94] V.I. Bakhmutov, J.F. Berry, F.A. Cotton, S. Ibragimov, C.A. Murillo, Dalton Trans.
(2005) 1989–1992.
[95] U. Nagel, B. Rieger, A. Bublewitz, J. Organomet. Chem. 370 (1989) 223–239,
https://doi.org/10.1016/0022-328X(89)87286-9.
[82] A few dimeric Pd-Fe complexes with bridging halide ligands are known [83-
85]. Countless di- and oligo-nuclear iron complexes with bridging halides are
known. The latter complexes cannot be active catalysts by themselves, but