This document is the Accepted Manuscript version of a Published Work that
appeared in final form in ACS Catalysis, copyright © American Chemical Society
after peer review and technical editing by the publisher. To access the final edited
and published work see https://pubs.acs.org/doi/10.1021/acscatal.9b02086.

Mechanistic Insights into the Ceria-Catalyzed
Synthesis of Carbamates as Polyurethane Precursors
Begoña Puértolas,#,† Marcos Rellán-Piñeiro,#,‡ José Luis Núñez-Rico,#,‡ Amol P. Amrute,†
Anton Vidal-Ferran,*,‡,& Núria López,*,‡ Javier Pérez-Ramírez,*,† and Stefan Wershofen§
†

Institute for Chemical and Bioengineering, ETH Zurich, Vladimir-Prelog-Weg 1, 8093 Zurich,

Switzerland.
‡

Institute of Chemical Research of Catalonia (ICIQ) & The Barcelona Institute of Science and

Technology, Av. Països Catalans 16, 43007 Tarragona, Spain.
&

ICREA, Pg. Lluís Companys 23, 08010 Barcelona, Spain.

§

Covestro Deutschland AG, Kaiser-Wilhelm-Allee 60, 51373 Leverkusen, Germany.

1

ABSTRACT
The methoxycarbonylation of toluenediamines with dialkyl carbonates constitutes an alternative
route for the phosgene-free production of isocyanate precursors. Despite the remarkable catalytic
activity of ceria in the reaction, achieving full selectivity and long-term stability still represent
major challenges. Here, the mechanism of the methoxycarbonylation of the industrially-relevant
2,4-diaminotoluene (2,4-TDA) with dimethylcarbonate (DMC) along with the evolution of the
property-performance interplay upon consecutive cycles are rationalized via the structural
identification of reaction products, characterization tools, and density functional theory (DFT).
The formation of the desired carbamates (7% mono- and 83% biscarbamate) is favored over the
(111) facet, the most-abundant in the as-prepared material, and proceeds via a complex reaction
mechanism that involves a broad number of isomers and multiple reaction paths. A consecutive
reaction in which 2,4-TDA is converted into a monocarbamate that further reacts to the
biscarbamate drives the selective path. Part of these carbamates reacts to form productive ureas,
unprecedented intermediates that reversely transform into carbamates. A full product analysis
enables to identify a number of side products that mostly comprise N-methylated carbamates and
N-methylated ureas. Evaluation in subsequent cycles evidences the catalyst deactivation and the
concomitant increase in the formation of by-products, which is linked to the increasing amount of
carbon deposits along with the DMC-induced partial surface restructuring into an
oxygen-defective (100) facet after six cycles. These findings highlight the challenges in the
rational design of robust heterogeneous catalysts for the production of isocyanate precursors.

KEYWORDS: carbamates, ceria, deactivation, isocyanate precursors, methoxycarbonylation
mechanism, structure sensitivity, ureas.

2

1. INTRODUCTION

Aromatic di- and polyisocyanates (e.g. toluene diisocyanates, TDI) are of great commercial
relevance in the manufacture of polyurethanes.1 Their market size was ca. USD 54 billion in 2015
and is projected to grow at an annual rate of 7% to 2025 driven by the increasing demand for
lightweight and durable materials in the furniture, construction, electronics, and automotive
industries.2 The current manufacture of TDI relies on the use of phosgene as building block to
transform both primary amino groups of 2,4-diaminotoluene (2,4-TDA) into isocyanate motifs.3
The process is highly selective and facile, and so far, there are no alternative technologies available
to produce aromatic isocyanates economically on an industrial scale. However, the high toxicity
of phosgene and extraordinary safety precautions associated to its use prompt the search of greener
chemical routes.1 Carbamates, which can be readily transformed into isocyanates by thermal
decomposition, have become an appealing substitute for the synthesis of polyurethanes. They can
be produced by the catalytic alkoxycarbonylation of diamines with dialkyl carbonates, such as
dimethylcarbonate (DMC), with the corresponding formation of an alcohol as the only by-product,
which is easily recyclable to yield the dialkyl carbonate in a second stage. Zinc acetate and zinc
aggregates have been reported as efficient catalysts (ca. 99% yield of biscarbamates),4,5 but their
limited reusability prevents their implementation at the technical scale.
To overcome these limitations, the quest of efficient heterogeneous catalysts has been
pursued.6-13 Several metal oxides such as CeO2, ZrO2, ZnO, or TiO2, supported metal oxides over
aluminophosphates, mesoporous silicas, or activated carbons, heterogenization of the
homogeneous zinc acetate moiety, and MOF-based systems have been explored. Among them,
nanocrystalline ceria emerged as a unique catalyst for the methoxycarbonylation of 2,4-TDA
(Scheme 1). Assessment in catalysis evidenced that the exposure of (111) or (110) facets of CeO2
led to ca. 90% yield of carbamates and 10% of side products, whereas the latter doubled over the
3

(100) facet. Corma et al. studied the reaction mechanism by first principles without considering
the van der Waals contributions and using aniline as a surrogate of 2,4-TDA.14 The formation of
the desired carbamates was driven by the selective DMC dissociation to yield Olatt–COOCH3
(Olatt: lattice oxygen) over the (111), (110), and (100) facets whereas the formation of methylated
products occurred over the (100) facet. Only identification of the selective reaction products was
conducted. Stability studies addressing the potential impact of successive reaction cycles on the
catalytic performance and on the properties of the catalyst were not reported.
In order to advance the understanding of this system, the present study analyzes the reaction
mechanism of CeO2 in the methoxycarbonylation of the industrially-relevant 2,4-TDA with DMC.
Characterization results along with a full product analysis at intermediate and final stages of the
reaction, and atomistic modeling are applied to identify the paths leading to the desired and side
products. Evaluation of the impact of catalyst restructuring on the evolution of the performance
with the number of cycles enables identification of the mechanistic routes resulting in the observed
catalyst deactivation, which remains as the key challenge for making this technology of
commercial appeal.

2. EXPERIMENTAL SECTION
2.1. Catalyst preparation and characterization. CeO2 was prepared by precipitation
following a previously reported protocol.15 Briefly, Ce(NO3)3·6H2O (Acros, 99.5%) was dissolved
in deionized water in a weight ratio of 1:10 under stirring and H2O2 (Aldrich, 35%) was added to
the solution to obtain a molar H2O2:Ce ratio of 3. The precipitation was achieved by the addition
of aqueous NH4OH (Aldrich, 30%) until reaching a pH of 10.5. The slurry was stirred for 4 h and

4

the solid was separated by filtration, washed with deionized water, dried overnight at 373 K, and
calcined in static air at 773 K for 5 h using a heating rate of 5 K min−1.
Powder X-ray diffraction (XRD) was measured using a PANalytical X’Pert PRO-MPD
diffractometer and Cu-K radiation ( = 0.15418 nm). The diffraction data was recorded in the
10-70° 2 range with an angular step size of 0.017° and a counting time of 0.26 s per step.
N2 sorption at 77 K was measured in a Micromeritics TriStar II analyzer. Prior to the measurement,
the solid was outgassed at 573 K for 3 h. The Brunauer-Emmett-Teller (BET) method was applied
to calculate the total surface area, SBET, in m2 g−1. Samples for high-resolution transmission
electron microscopy (HRTEM) studies were prepared by dusting respective powders onto
carbon-coated nickel grids. Imaging was performed on a Talos F200X instrument operated at
200 kV and equipped with a FEI SuperX detector. Temperature-programmed reduction with
hydrogen (H2-TPR) was performed in a Micromeritics Autochem II 2920 analyzer. First, the solid
was loaded into a U-shaped quartz micro-reactor, pretreated in air (20 cm3 min−1) at 773 K for 1 h,
and cooled to 323 K. The H2-TPR analysis was carried out in 5 vol.% H2/N2 (20 cm3 min−1),
ramping the temperature from 323 K to 1073 K at 10 K min−1. The content of carbonaceous
deposits in the used catalysts was determined by thermogravimetric analysis (TGA) using a Linseis
TGA PT1600 thermobalance. The solid (ca. 25 mg) was placed in an alumina crucible and heated
in 5 vol.% O2/Ar (300 cm3 STP min−1) from ambient temperature to 1173 K at 10 K min−1. Raman
spectroscopy was carried out in a confocal Raman microscope (WITec CRM 200) using a 532 nm
diode laser. The microscope was operated in the backscattering mode with a 100× objective lens
and 6 mW power.
2.2. Evaluation of catalyst performance. Methoxycarbonylation reactions were performed in
a 25-cm3 stainless steel autoclave (Berghof high-pressure reactor BR-25). In a typical experiment,

5

the catalyst was placed in the Teflon vessel together with 2,4-TDA (396 mg, 3.2 mmol, Aldrich,
98%) and the required amount of DMC (8.2 cm3, 97.2 mmol, Aldrich, 99%). Prior to the catalytic
test, DMC was dried with a 4Å molecular sieve to ensure a final water content <30 ppm, as
determined by the Karl Fischer method. The weight of catalyst added to the reaction mixture was
varied in the range of 0.95-1.18 g, in order to keep a fixed ratio of 20 m2 per mmol of 2,4-TDA.
The total surface area was measured prior to the test as described in the previous section.
Afterwards, the autoclave was closed, placed in an aluminum heating block, and purged with N2.
Stirring was then started (810 rpm) and reaction times up to 10.5 h were studied. The temperature
was controlled by a thermocouple connected to the heating plate. At the end of each experiment,
the autoclave was removed from the heating block and cooled down in an ice-water bath for
30 min. The autoclave was then opened, the catalyst recovered by filtration, the Teflon vessel and
the solid residue were washed with acetone (5 times, 5 cm3), and the combined organic fractions
concentrated under vacuum to give the reaction mixture, whose composition was subsequently
analyzed as detailed in the next section. The recovered catalyst was placed in an open glass vial
and heated in air at 393 K for 1 h prior to its storage or further reuse. Six consecutive catalytic tests
(reaction time per cycle = 7.5 h) were conducted under the same conditions as those described
above. After each run, the catalyst was recovered by filtration and washed with acetone as
previously detailed. In order to evaluate the impact of the solvent on the properties of the catalyst,
the as-prepared CeO2 was subjected to a 42-h treatment in DMC under the same conditions to
those applied in the catalytic tests. The resulting material was filtered and washed with acetone
prior to characterization and further catalytic studies in methoxycarbonylation reactions.
2.3. Product analysis. An Agilent 1100 series HPLC system was used to identify all products
and to quantify components of the reaction mixture. The analytical conditions were as follows:

6

Kromasil

100

C18

column

(5 m

4.6×150 mm),

T = 298 K,

flow = 1.0 cm3 min−1,

injection = 5×10−3 cm3, UV detection = 225 nm, eluent A: 100 cm3 CH3CN, 900 cm3 H2O,
0.01 M NH4COOCH3, eluent B: 900 cm3 CH3CN, 100 cm3 H2O, 0.01 M NH4COOCH3, gradient:
0 min 100% A, 22 min 100% A; 48 min 80% A, 20% B; 60 min 55% A, 45% B; 80 min 55% A,
45% B; 82 min stop. Up to 24 different peaks in the chromatograms of the methoxycarbonylation
reaction mixtures were identified. Analyses in a HPLC system equipped with an atmospheric
pressure chemical ionization detector in positive mode (APCI+) using the above-mentioned
chromatographic method allowed recording the mass spectra (MS) of all chromatographic peaks.
The list of compounds, retention times, and diagnostic mass-to-charge (m/z) ratios of the
APCI+-MS spectra are summarized in Section S2. Yields of 2,4-TDA, MC1 and MC2
monocarbamates, and biscarbamate (BC) were determined employing external standards either
purchased or prepared in house and quantification methods previously published by some of us.4
The remaining obtained byproducts are not available commercially and thus, their concentrations
are calculated based on the relative area (area%) of each component determined by HPLC analysis.
The assignment of the chromatographic peaks to the corresponding products is based on the mass
spectra of each compound that is collected by using an atmospheric pressure chemical ionization
detector in positive mode (APCI+). Productive urea PU1 (Scheme 1) was detected in real
methoxycarbonylation reaction mixtures by HPLC analyses at short reaction times, whilst PU2
and PU3 were not detected at an appreciable extent at any time in any reaction mixture. From all
productive monomethoxycarbonylated ureas derived from PU1-3, 4 isomers out of 4 (i.e. 4 peaks
with molecular ions in HPLC-APCI+ being in agreement with the corresponding [M+H]+ ion) were
detected in the reaction mixtures (Section S3). A minimum of 2 isomers out of 3 of productive
bis(methoxycarbonylated) ureas derived from PU1-3 were identified. N-methylated compounds

7

derived from 2,4-TDA (MeTDA1-8) were only detected in minor amounts after four consecutive
cycles with the CeO2 catalyst and in the absence of catalyst. With respect to N-methylated
monocarbamates derived from MC1 and MC2 (MeMC1-10), a minimum of 5 isomers were
detected. The 3 isomers of N-methylated biscarbamates derived from BC (MeBC1-3) were
identified. Finally, a minimum of 5 isomers of N-methylated ureas UU1-32 were detected.
2.4. Computational details. Density functional theory (DFT) simulations were performed
with

the

Vienna

Ab

Perdew-Burke-Ernzerhof

initio
(PBE)

Simulation
functional18

Package
coupled

(VASP,
to

version

vdW-D319

5.4.4).16,17 The
was

employed.

Projector-Augmented Wave (PAW) pseudopotentials were used to describe the inner electrons,
whereas valence electronic states were expanded in plane waves with an energy cutoff of 500 eV.20
The Hubbard correction21 of 4.5 eV was applied to the Ce (4f) states, which was optimized through
a linear response method22 and benchmarked against hybrid methods by some of us.23
Spin-polarized calculations were performed when required. The criteria for electronic and
geometric optimization convergence were set to 10−6 eV and 0.025 eV Å−1, respectively. The
simulations were performed using the fluorite phase of CeO2, as observed experimentally by X-ray
analysis. The bulk was optimized with a dense 7×7×7 -centered k-point mesh obtaining a lattice
parameter (acalc = 5.470 Å) in good agreement (1.1% deviation) with the experimental value
(aexp = 5.411 Å).24 The reaction mechanism was investigated on the three stoichiometric CeO2
facets, i.e., (111), (110), and (100), using slab models with three, five, and four O–Ce–O trilayers,
respectively. These slabs thicknesses were successfully used in previous works.25-28 For the (111)
and (110) there is only one surface termination that is stable. On the other hand, due to the polar
character of (100) facet, a surface reconstruction is needed as several terminations are possible.
The most stable reconstruction, in which a half monolayer of oxygen atoms is removed from the

8

top of the slab and placed on the bottom, was chosen.29 A vacuum gap of 15 Å between slabs was
added and the dipole correction along z-axis was utilized to prevent the spurious terms arising
from the asymmetry of the slabs.30 The supercells were large enough to avoid the interaction
between the adsorbates of neighboring cells and the k-points mesh was always denser than 0.3 Å−1.
In addition to the adsorbates, the topmost layers of the slabs, i.e., five, three, and five atom layers
for the (111), (110), and (100) facets, respectively, were allowed to relax. All the molecules were
computed inside a box of 20×21×22 Å. For the reactivity in solution, the DMC solvation
contributions were included through the VASP-MGCM continuous model.31,32 The climbing
image nudged elastic band (CI-NEB) method33,34 refined with the Improve Dimer Method
(IDM)35,36 were used to allocate the transition states. The corresponding vibrational analysis was
used to characterize the intermediates and transition states in the reaction network. All the relevant
structures are published in the ioChem-BD database.37

3. RESULTS AND DISCUSSION
3.1. Characterization of CeO2 and evaluation in the methoxycarbonylation of 2,4-TDA.
Characterization of the as-prepared material by XRD (Figure S1a) revealed the characteristic
pattern of the cubic fluorite structure of CeO2 and application of the Scherrer equation to the most
prominent (111) reflection led to an average crystallite size of 11 nm. The sample exhibited a total
surface area of SBET = 74 m2 g−1. The H2-TPR results (Figure S1b) evidenced the characteristic
low-temperature peak at around 600-800 K, attributed to the reduction of surface oxygen species,
and a high-temperature peak >900 K, associated with the bulk reduction of CeO2.38 HRTEM
analysis (Figure 1a) revealed the presence of octahedron-like geometry CeO2 particles of variable

9

diameter (ca. 10 nm) in which the dominant lattice fringes (0.31 nm) correspond to the (111)
family.
The CeO2 catalyst was evaluated in the methoxycarbonylation of 2,4-TDA with DMC
(Table 1) under the standard reaction conditions, i.e., 20 m2 per mmol of 2,4-TDA, 413 K, 7.5 h.
The substrate was fully converted to yield 7% of monocarbamates (mixture of the MC1 and MC2
isomers) and 83% of biscarbamate (BC). The results were reproducible and the standard deviation
values for the yields of MC1, MC2, and BC after 6 independent runs were 0.6% for the sum of
MC1 and MC2, and 1.5% for BC. Additionally, formation of side products, i.e., 3 area% of
N-methylated carbamates (MeC1-13), 1 area% of productive ureas (PU1-7) and 3 area% of
N-methylated ureas (UU1-32), was observed under these conditions along with 1 area% of
unidentified compounds.
The concentration profiles of 2,4-TDA, productive and unproductive intermediates, and the
target compounds as a function of the reaction time are displayed in Figure 2. The evolution of the
carbamates (i.e. MC1, MC2, and BC) corresponds to that typically observed in competitive
sequential processes: the starting amine was first converted into monocarbamates (MC1 and MC2),
which were then consumed to form the final BC product at longer reaction times (Figure 2a).
Interestingly, an additional compound with a similar concentration profile to those of MC1 or MC2
was observed at short reaction times. HPLC-APCI+ analyses and chemical synthesis allowed us
identifying

the

structure

of

this

unprecedented

intermediate

as

1,3-bis(3-amino-4-

methylphenyl)urea (structure PU1 in Section S2), for which a maximal 10 area% content at
ca. 15 min

was

determined.

Whilst

other

isomers

of

PU1

were

not

detected,

monomethoxycarbonylated (compounds PU4-7 with a maximal 7 area% content at ca. 40 min)
and bis(methoxycarbonylated) (compounds PU8-10 with a maximal 6 area% content at

10

ca. 110 min) ureas were also identified (Figure 2b). Following an analogous trend to that observed
for PU1, mono- and bis(methoxycarbonylated) PU4-10 ureas were also consumed at extended
reaction times, which prompted us to investigate if the final carbamates could arise from these
intermediates. Therefore, PU1 was synthesized and allowed to react under the standard reaction
conditions yielding a reaction mixture that was mainly comprised by carbamates (Table S1), thus
suggesting the key role of CeO2 in the formation of the carbamates, as no methoxycarbonylation
of the starting material or cleavage of the productive ureas into MC or BC occurred to an
appreciable extent in the absence of the catalyst. These results point to a number of competitive
sequential processes involving MC1, MC2, and the productive ureas (Scheme 1), which are
converted to the target BC at late stages in the CeO2-catalyzed process. The complete reactivity of
ureas will be studied in detail in an upcoming work. Regarding the formation of side products
(Figure 2c), the concentration of N-methylated carbamates and ureas (i.e., MeMC, MeBC, and UU
derivatives) increases with time, reaching a plateau in the case of the N-methylated ureas, thus
highlighting the significance of shortened reaction times.
In order to assess the reactivity of the 2,4-TDA, reaction intermediates, and the target
biscarbamate product in the absence or presence of CeO2 and/or using different reaction
temperatures or times, additional reference experiments were conducted (Table S1). 2,4-TDA was
subjected to methoxycarbonylation conditions in the absence of catalyst (DMC, 413 K, 7.5 h)
yielding 10% MeTDA derivatives. In this case, carbamates MC1, MC2, or BC were not detected.
The comparison of these results with those attained under catalytic conditions (Table 1) suggests
that (i) CeO2 is key in the formation of the selective and side products and (ii) any trace of MeTDA
derivatives formed in solution can evolve into the different side products, i.e., N-methylated
carbamates and ureas, detected under catalytic conditions. Indeed, MeTDA derivatives were not

11

identified during the first cycle with CeO2 (Table 1). Additionally, the impact of the reaction
temperature on the product distribution in the methoxycarbonylation of 2,4-TDA with DMC over
CeO2 was assessed after 75 min. By contrasting the results obtained at 413 K (Figure 2, Table S1)
with those at 383 K (Table S1), the accumulation of productive ureas at lower reaction
temperatures was evidenced. In this case, 41 area% of productive ureas was identified, from which
33 area% corresponded to PU1 at reaction temperature of 383 K, whereas 10 area% (1 area%
corresponds to PU1 and the rest to mono- and bis(methoxycarbonylated) ureas) was detected at
413 K. Finally, assessment of the N-methylation of the carbamate groups of BC with DMC was
conducted under thermal and catalytic conditions (413 K, 7.5 h). 8% of the substrate was converted
in the presence of CeO2 leading to the formation of mono-(N-methyl) and bis-(N-methyl) MeBC
derivatives whereas no N-methylated products were detected without catalyst.
3.2. Reaction mechanism. The mechanism for the direct synthesis of biscarbamates and side
reactions in the methoxycarbonylation of 2,4-TDA with DMC was studied via DFT over CeO2
(111) as this is the most abundant facet in the fresh material (Figure 1a). For completeness, the
reaction mechanism was also derived on the other two low-energy (110) and (100) surfaces and
the main mechanistic differences are elaborated in Section S5. The complex reaction network on
CeO2 is summarized in Figure 3 and the reactivity of 2,4-TDA, carbamates, and N-methylated
carbamates and ureas in solution is exemplified in Figure 4. The details of all the steps involved
in the reaction network are explained in the following sections.
DMC activation. The first step in the reaction mechanism is the adsorption of DMC. It adsorbs
in a ɳ2,C,O configuration to a Ce–Olatt bond by -0.89 eV (Table S2). The subsequent DMC
dissociation to a methoxycarbonyl group (Olatt–COOCH3), i.e., the reactant needed for the
methoxycarbonylation reaction, and a methoxy group (Ce–OCH3) is exothermic (0.09 eV) with a

12

small activation energy (0.20 eV). Alternatively, DMC can dissociate to form monomethyl
carbonate (Ce–OCOOCH3) and a surface methyl group (Olatt–CH3) via a SN2 mechanism with the
Olatt acting as a nucleophile, both mechanisms are in line with previous studies that have tackled
the dissociative adsorption of DMC over CeO2 catalysts using in situ FTIR spectroscopy.7 The
Olatt–CH3 groups are potential methylating agents responsible of the formation of side products
(Section S5). This dissociation is exothermic (0.33 eV) although unfeasible attending to its high
activation energy (1.50 eV). As large amounts of DMC are present in the reaction mixture, the
high exothermicity of the adsorption of dissociated DMC (0.98 eV) may result in the full coverage
of the catalyst surface with Olatt–COOCH3 and Ce–OCH3 fragments. Unlike on the pristine surface,
the unselective Olatt–CH3 groups are easily formed if vacancies are present. DMC adsorbs stronger
on an Olatt in contact with a Ce3+ center (-1.16 eV), which results in the dissociation of a methoxy
group of DMC that replenishes the vacancy and leads to the formation of surface methyl groups
(vide infra). This step has a moderate energy barrier (0.40 eV) and is highly exothermic (1.41 eV).
The high energy associated with vacancy formation on this surface (2.11 eV in a (4×4) supercell)
restricts the process to vacancies either present in the fresh catalyst or formed during the catalytic
process in case surface reconstruction occurs.
Mechanism of the selective path. Once DMC is adsorbed, the subsequent adsorption of
2,4-TDA occurs. The substrate adsorbs through one of its two amino groups on a surface cation
(Figure 3) with an adsorption energy for the most accessible para group of -1.43 eV. Since the
adsorption of this molecule would require the desorption of the adsorbed DMC, the replacement
energy (Eads, 2,4-TDA - Eads, DMC = -0.45 eV) (Table S3, Figure S7) is the one considered as a
reference along the manuscript. The Olatt can strip a proton from the –NH2 groups with an
activation barrier of 0.56 eV (Table S4). This step is endothermic by 0.52 eV, and therefore, the

13

equilibrium is shifted towards the reactants (Figure 5, Table S4). In the case of the ortho isomer
(MC1), the steric hindrance due to the presence of the aryl-methyl group hampers both the
adsorption and the –NH2 activation, which explains the higher concentration of the MC2 isomer
observed experimentally (Figure 2a). After the activation of the –NH2 group, the subsequent
formation of the monocarbamate occurs via the nucleophilic attack of the lone electron pair of
nitrogen to the C=O group in Olatt–COOCH3 (Figure 6).39 The formation of MC2 is exothermic
(0.42 eV) with an activation energy of 0.30 eV, and its desorption energy is 0.37 eV. Similar
values were obtained for the formation of MC1 via the reaction on the ortho position of 2,4-TDA
(Table S4). Both MC1 and MC2 yet contain a free amino group that can dissociate on the surface
and react with another methoxycarbonyl group to form the target BC product. Therefore, the
desorbed MC molecules are readsorbed through the second amino group. The subsequent reaction
mechanism and energy values associated to this transformation are analogous to those described
above for 2,4-TDA (Table S4). The low energy requirements are in line with the full conversion
of 2,4-TDA observed experimentally over the fresh material (Table 1).
2,4-TDA condensation and formation of productive ureas. Ureas are formed by condensation
of two aromatic ring-containing molecules that incorporate either an amino or a carbamate
functional group (blue arrows in Figure 3). Productive ureas (PU4-10) constitute unprecedented
intermediates in the formation of the target carbamate and a number of side products which contain
the urea motif (Table 1). The reaction of activated 2,4-TDA and MC2 forms the most stable
productive urea, i.e., 1,3-bis(3-amino-4-methylphenyl)urea (structure PU1 in Scheme 1). The first
step is the dissociation of the –OCH3 moiety of the carbamate group of MC2, which can readily
occur on the catalyst surface after the formation of the monocarbamate. This process is exothermic
(0.28 eV) and has lower energy barrier (0.20 eV) than the competitive monocarbamate desorption

14

(0.37 eV). Once MC2 is dissociated, the condensation product is formed by the nucleophilic attack
of activated 2,4-TDA to the C=O group of the dissociated monocarbamate, which requires an
activation energy of 0.36 eV, thus similar to the values obtained for the methoxycarbonylation
reactions. The desorption energy of the resulting PU1 urea is 0.62 eV. The analogous condensation
between monocarbamates and/or biscarbamates forms the rest of the observed productive ureas
(PU4-10). These compounds (PU1-10) can decompose through the reverse path either into
2,4-TDA and MC, which can evolve to the carbamates as previously discussed. The path through
condensation resembles other routes proposed for the conversion of anilines.40 The computed low
energy barriers favor the rapid formation of these productive ureas at short reaction times in line
with the experimental evidences (Figure 2b). Nevertheless, the formation of ureas is an
endothermic process (e.g., 0.21 eV for the formation of PU1 urea from 2,4-TDA and MC2), and
therefore, as the reaction path is reversible, they are not detected at longer reaction times (Table 1,
Figure 2c). Indeed, when the reaction is conducted at lower temperatures and short reaction times
(383 K, 75 min), a larger fraction of productive ureas is detected (Table S1), consistent with the
lower energy demands for this transformation derived from the DFT simulations.
Mechanism of the side paths. N-methylated 2,4-TDA (MeTDA) derivatives, carbamates
(MeMC1-10 and MeBC1-3), and ureas (UU1-32) constitute the main fraction of side products
(Table 1). MeTDA, MeMC1-10, and MeBC1-3 are formed via a SN2 reaction between either the
activated amine or the carbamate groups with the methyl group of Olatt–COOCH3 (red arrows in
Figure 3).39 On 2,4-TDA, this reaction results in the formation of N-methylated 2,4-TDA
(MeTDA1 and MeTDA2) and CO2 (Figure 3). It is highly exothermic on the most reactive para
amine group (1.20 eV), and has an activation energy of 0.86 eV, which is higher than that
computed for the selective path (methoxycarbonylation of 2,4-TDA, 0.30 eV) due to the different

15

electrophilic character of the two methoxycarbonyl moieties.39 Once MeTDA2 desorbs (0.43 eV),
it is further methoxycarbonylated either on its NH2 or NHCH3 groups to form the experimentally
detected N-methylated carbamates (MeMC1, MeMC4, MeMC5, MeMC6, MeMC7, MeMC9
and/or MeMC10 in Scheme 1 and Section S2), which explains that MeTDA is not detected
experimentally in the catalytic reaction (Table 1). The methoxycarbonylation of the free amino
group (–NH2) of MeTDA2 proceeds similarly to the above-explained reaction for 2,4-TDA and
leads to MeMC1. However, the methoxycarbonylation of the –NHCH3 group in MeTDA2 shows
small energy differences with respect to the activation and methoxycarbonylation of 2,4-TDA: the
activation energy for N–H dissociation is slightly higher (0.64 eV), whereas it is lower in the case
of the methoxycarbonylation (0.21 eV). The desorption energy of the resulting MeMC4 is 0.33 eV.
MeTDA2 could be further methylated in the same nitrogen position towards the bismethylated
compound MeTDA5. This second methylation is also exothermic (1.30 eV) with an activation
energy of 0.80 eV, slightly lower than the energy needed for the first methylation, and the
desorption energy of the bismethylated compound is 0.43 eV. Alternatively, N-methylated
carbamates can also be formed via the methylation on the carbamate group of MC1 or MC2.
Herein, we have only considered the N-methylation in the carbamate group in MC2, which
proceeds via the same reaction mechanism as described above: a proton of the adsorbed carbamate
group is stripped via an endothermic step (0.27 eV) with a lower activation energy (0.34 eV) than
that required for the dissociation of NH2 groups. The methylation reaction to form MeMC4 is
highly exothermic (0.88 eV) and has an activation energy of 1.05 eV, which is higher than that for
the reaction with the amino group. The desorption of the formed MeMC4 requires 0.54 eV
(Table S4). These methylated compounds can react with another aromatic unit following an

16

analogous mechanism to the one described above to form the unproductive ureas (UU1-32)
experimentally detected.
Reactivity in solution. During the catalytic process, the primary and N-methylated amino
groups and carbamates can also react with DMC in the reaction medium, i.e., without interacting
with the catalyst surface, via methoxycarbonylation and N-methylation reactions (black and green
arrows in Figure 4, respectively). The transition states for these reactions were allocated on the
para position of 2,4-TDA, MC2 (carbamate), and MeTDA2 (N-methylated amine). The activation
energies for methoxycarbonylation are 1.34, 1.93, and 1.30 eV, respectively, following the trend
of the nitrogen electronegativity of the reactive group, which hinders the occurrence of these
reactions in the solution. These reactions result in the formation of the methoxycarbonylated
compound and a methanol molecule. The activation energies for N-methylation processes follow
the same trend but exhibit smaller values: 0.77, 0.76, and 1.27 eV on 2,4-TDA, MC2, and
MeTDA2, respectively. Monomethyl carbamate, which quickly decomposes in methanol and CO2,
is concomitantly formed in these reactions. Thus, the activation energies for the
methoxycarbonylation are higher than for the methylation on the three groups, contrasting the
behavior observed for the catalytic process.
3.3. Assessment of catalyst recyclability. Up to now, the performance assessment of CeO2
has been limited to a single catalytic cycle. Therefore, the stability of CeO2 and the possible impact
of the formation of undesired compounds on the performance upon consecutive cycles were
evaluated. After each run, the total surface area of the recovered catalyst was determined and the
amount of catalyst was adjusted to 20 m2 per mmol of 2,4 TDA. Even though the surface area of
the catalyst remained fairly constant upon consecutive use (Table 1) and its performance remained
unaffected during the first three cycles, the yield towards the desired products halved during the

17

fourth cycle (Figure 7a) i.e., the combined yield of MC1, MC2, and BC decreased from ca. 90%
to 40%, and further decreased during the fifth cycle. After that, the used catalyst was treated with
DMC under reaction conditions (413 K, 7.5 h) in an attempt to restore the catalytic activity of the
fresh material. However, the resulting material evidenced further deactivation during the sixth
cycle. The decrease in the yield to the desired products was accompanied by a decrease of the
2,4-TDA conversion and an increase in the formation of undesired N-methylated carbamates
(MeMC and MeBC) and ureas (UU) (Table 1) and other unknown compounds. For instance, in
the case of the N-methylated carbamates, they increased from 3 area% to 33 area% after the first
and sixth cycle, respectively. XRD results of the used material (Figure S1a) revealed that the
fluorite structure and crystallite size are preserved after catalysis. Characterization by TGA showed
an increase in the amount of carbon deposits upon recycling, from 5.1 wt.% after two cycles to
24.1 wt.% after six cycles (Figure 7b), in line with the increase of the heat released via the
combustion of the carbonaceous residuals. Additionally, the maximum of the heat flow profile
shifted towards higher temperatures when increasing the number of cycles, suggesting the presence
of heavier carbonaceous species on the catalyst surface, which might explain the decrease of the
2,4-TDA conversion after several cycles since lower amount of active sites are available for the
reaction. HRTEM micrographs of the CeO2 samples after three and six consecutive cycles
evidenced that the morphology of the CeO2 particles is retained. Nevertheless, the occurrence of
(110) facets after three cycles accompanied by the appearance of the intrinsically defective (100)
facets after six cycles (Figure 1) demonstrated the surface restructuring, which led to the decrease
in the selectivity to the carbamates and the concomitant increase of the formation of side products.
These results are in line with previous studies,14 in which the assessment of CeO2 nanocubes,
which preferentially exhibit the (100) facet, in this reaction led to selectivity values to the desired

18

BC lower than 40% after 10 hours, thus confirming the presence of oxygen vacancies as
detrimental for the selectivity. The presence of oxygen vacancies was further corroborated by
Raman spectroscopy (Figure S1c). The spectrum of the catalyst after 6 cycles evidenced a
frequency downshift of the T2g band of ca. 3 cm-1, which has been previously associated to the
presence of oxygen vacancies.41
As the catalyst is expected to be fully covered by DMC, the possible impact of the solvent on
the deactivation was assessed by subjecting the CeO2 catalyst to a 42-h treatment, i.e., the time
equivalent to six consecutive cycles, with DMC. This treatment led to the occurrence of the (100)
facet, as evidenced on the HRTEM micrographs (Figure 1d), which suggests the potential role of
the solvent in promoting the observed surface reconstruction. The presence of DMC stabilizes the
different facets in an extent proportional to its adsorption energy (per surface area), which was
found very different in the three facets by DFT. This stabilization acts as thermodynamic factor of
shape control that modifies the surface energies changing the surface morphology.42,43 The surface
energies of the most stable CeO2 facets, i.e., (111), (110), and (100), result in the (111) as the only
facet exposed on the fresh catalyst, as shown by the Wulff construction (Figure 7c). The (111)
surface prior to the adsorption of DMC is much more stable (0.95 J m−2) than the (100) surface
(1.77 J m−2) (Table S5). At medium coverage, i.e., DMC = 0.25, the adsorption energies per DMC
molecule are very similar to those obtained on clean surfaces (-1.03 eV and -3.04 eV per molecule
of DMC on (111) and (100), respectively) which results in stronger adsorption energies per surface
area on the (100) surface (-0.020 and -0.051 eV Å−2 on (111) and (100), respectively). These
different adsorption energies per surface area reduce the surface energy in a different extent and
result in closer surface energies of 0.64 J m−2 and 0.94 J m−2 for (111) and (100), respectively,
driving the appearance of the (100) surface as is shown in the Wulff construction in Figure 7c. The

19

presence of intrinsic oxygen vacancies on the (100) facet favors the formation of surface methoxy
groups from the selective dissociation of DMC that are equivalent to Olatt–CH3 (Figure 7d). The
latter are methylation agents that might lead to the increased formation of side products upon
recycling (Section S5). Catalytic assessment of this material revealed lower activity than the nontreated counterpart, as yields of 16% and 72% were obtained for the desired MC and BC products,
respectively, although the selectivity towards methoxycarbonylation products was not
significantly affected and remained similar to that of the as-prepared material. Comparison to the
results obtained after six cycles (Table 1) indicates that the absence of carbonaceous deposits in
this case might also contribute to the observed differences in catalysis.

CONCLUSIONS
The reaction and deactivation mechanisms of CeO2 catalysts in the methoxycarbonylation of
2,4-TDA with DMC for the production of isocyanate precursors has been assessed by a
combination of experimental and theoretical tools. The as-prepared CeO2 material, exhibited 90%
selectivity towards the desired carbamates, i.e., 7% para- and ortho-monocarbamates and 83%
biscarbamate and the production of small amounts of side products. Detailed analyses of the
methoxycarbonylation reaction by HPLC-APCI+ at short reaction times reveal the prominent role
of productive ureas in the formation of the desired carbamates via the condensation of
monocarbamates and 2,4-TDA. Both paths are identified at the molecular level and the derived
energetic and kinetic parameters demonstrate the reversibility of the urea formation. N-methylated
products constitute the major fraction of the side products. Reusability tests indicate that the
conversion of the substrate decreased from 100% to 80% and the selectivity to carbamates dropped
to 16%, which was related to the increased formation of carbon deposits along with the surface

20

restructuring after six consecutive cycles: whereas the fresh material mainly exhibited the (111)
facet, subsequent cycles evidence the occurrence of the oxygen-defective (100) facet. The
dissociation of the DMC solvent over an oxygen vacancy leads to methyl groups on the catalyst
surface that favor the formation of N-methylated products. Our study not only identifies key
intermediates of the reaction and deactivation mechanisms but also sets the fundamental basis
behind the limited recyclability of CeO2 providing guidelines for the design of improved catalysts
for the phosgene-free production of isocyanate precursors. Upcoming studies should be directed
towards the stabilization of the active exposed facet, e,g., by doping strategies, or discovering
alternative active phases.

ASSOCIATED CONTENT
Supporting Information. Additional catalytic and characterization data, detailed list of the
identified molecules, preparation and characterization of productive ureas, and DFT results (PDF).

AUTHOR INFORMATION
Corresponding Authors
*Anton Vidal-Ferran: avidal@iciq.cat, Núria López: nlopez@iciq.es, Javier Pérez-Ramírez:
jpr@chem.ethz.ch.
Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval
to the final version of the manuscript. # These authors contributed equally.

21

ACKNOWLEDGEMENTS
E. Vorobyeva is acknowledged for microscopic analyses and ScopeM at ETH Zurich for access to
its facilities. V. Paunović is thanked for conducting Raman spectroscopy.

22

REFERENCES
(1)

Kreye, O.; Mutlu, H.; Meier, M. A. R. Sustainable Routes to Polyurethane Precursors. Green
Chem. 2013, 15, 1431–1455.

(2)

Global

Polyurethane

(PU)

Market

Size

&

Share,

2025,

Industry

https://www.grandviewresearch.com/industry-analysis/polyurethane-pu-market

Report.
(accessed

May 19, 2019).
(3)

Pérez-Ramírez, J.; Mondelli, C.; Schmidt, T.; Schluter, O. F. K.; Wolf, A.; Mleczko, L.;
Dreier, T. Sustainable Chlorine Recycling via Catalysed HCl Oxidation: from Fundamentals
to Implementation. Energy Environ. Sci. 2011, 4, 4786–4799.

(4)

Reixach, E.; Bonet, N.; Rius-Ruiz, F. X.; Wershofen, S.; Vidal-Ferran, A. Zinc Acetates as
Efficient Catalysts for the Synthesis of Bis-isocyanate Precursors. Ind. Eng. Chem. Res.
2010, 49, 6362–6366.

(5)

Reixach, E.; Haak, R. M.; Wershofen, S.; Vidal-Ferran, A. Alkoxycarbonylation of
Industrially Relevant Anilines Using Zn4O(O2CCH3)6 as Catalyst. Ind. Eng. Chem. Res.
2012, 51, 16165–16170.

(6)

Li, F.; Li, W.; Li, J.; Xue, W.; Wang, Y.; Zhao, X. Investigation of Supported Zn(OAc)2
Catalyst and its Stability in N-Phenyl Carbamate Synthesis. Appl. Catal., A 2014, 475, 355–
362.

(7)

Juárez, R.; Concepción, P.; Corma, A.; Fornés, V.; García, H. Gold‐Catalyzed Phosgene‐
Free Synthesis of Polyurethane Precursors. Angew. Chem. Int. Ed. 2010, 122, 1308–1312.

(8)

Sun, D.-L.; Deng, J.-R.; Chao, Z.-S. Catalysis over Zinc-Incorporated Berlinite (ZnAlPO4)
of the Methoxycarbonylation of 1,6-Hexanediamine with Dimethyl Carbonate to Form
Dimethylhexane-1,6-Dicarbamate. Chem. Cent. J. 2007, 1:27.

23

(9)

Guo, X.; Qin, Z.; Fan, W.; Wang, G.; Zhao, R.; Peng, S.; Wang, J. Zinc Carboxylate
Functionalized Mesoporous SBA-15 Catalyst for Selective Synthesis of Methyl-4,4′di(phenylcarbamate). Catal. Lett. 2009, 128, 405–412.

(10) Wang, Y.; Liu, B. Efficient and Recyclable Heterogeneous Zinc Alkyl Carboxylate Catalyst
for the Synthesis of N-Phenyl Carbamate from Aniline and Dimethylcarbonate. Catal. Sci.
Technol. 2015, 5, 109–113.
(11) Lucas, N.; Amrute, A. P.; Palraj, K.; Shanbhag, G. V.; Vinu, A.; Halligudi, S. B. NonPhosgene Route for the Synthesis of Methyl Phenyl Carbamate using Ordered AlSBA-15
Catalyst. J. Mol. Catal. A: Chem. 2008, 295, 29–33.
(12) Grego, S. Aricò, F. Tundo, P. Phosgene-free Carbamoylation of Aniline via Dimethyl
Carbonate. Pure Appl. Chem. 2012, 84, 695–705.
(13) Rojas-Buzo, S.; Garcia-Garcia, P., Corma, A. Zr-MOF-808@MCM-41 catalyzed phsogenefree synthesis of polyurethane precursors. Catal. Sci. Technol. 2019, 9, 146–156.
(14) Laursen, S.; Combita, D.; Hungría, A. B.; Boronat, M.; Corma, A. First‐Principles Design
of Highly Active and Selective Catalysts for Phosgene‐Free Synthesis of Aromatic
Polyurethanes. Angew. Chem. Int. Ed. 2012, 51, 4190–4193.
(15) Moser, M.; Vilé, G.; Colussi, S.; Krumeich, F.; Teschner, D.; Szentmikló, L.; Trovarelli, A.;
Pérez-Ramírez. J. Structure and Reactivity of Ceria-Zirconia Catalysts for Bromine and
Chlorine Production via the Oxidation of Hydrogen Halides. J. Catal. 2015, 331, 128–137.
(16) Kresse, G.; Furthmueller, J. Efficient Iterative Schemes for ab initio Total-Energy
Calculations using a Plane-Wave Basis Set. Phys. Rev. B: Condens. Matter. 1996, 54,
11169–11186.

24

(17) Kresse, G.; Furthmuller, J. Efficiency of ab-initio Total Energy Calculations for Metals and
Semiconductors using a Plane-Wave Basis Set. Comput. Mater. Sci. 1996, 6, 15–50.
(18) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple.
Phys. Rev. Lett. 1996, 77, 3865–3868.
(19) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and Accurate ab initio
Parametrization of Density Functional Dispersion Correction (DFT-D) for the 94 Elements
H-Pu. J. Chem. Phys. 2010, 132, 154104.
(20) Kresse, G.; Joubert, D. From Ultrasoft Pseudopotentials to the Projector Augmented-Wave
Method. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 1758–1775.
(21) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.; Sutton, A. P. ElectronEnergy-Loss Spectra and the Structural Stability of Nickel Oxide: An LSDA+U study. Phys.
Rev. B 1998, 57, 1505–1509.
(22) Fabris, S.; de Gironcoli, S.; Baroni, S.; Vicario, G.; Balducci, G. Taming Multiple Valency
with Density Functionals: A Case Study of Defective Ceria. Phys. Rev. B 2005, 71, 041102.
(23) Capdevila-Cortada, M.; Łodziana, Z.; López, N. Performance of DFT+U Approaches in the
Study of Catalytic Materials. ACS Catal. 2016, 6, 8370–8379.
(24) Kümmerle, E. A.; Heger, G. The Structures of C–Ce2O3+δ, Ce7O12, and Ce11O20. J. Solid
State Chem. 1999, 147, 485–500.
(25) Capdevila-Cortada, M.; García-Melchor, M.; López, N. Unraveling the Structure Sensitivity
in Methanol Conversion on CeO2: A DFT+U study. J. Catal. 2015, 327, 58–64.
(26) Mei, D.; Deskins, N. A.; Dupuis, M.; Ge, Q. Density Functional Theory Study of Methanol
Decomposition on the CeO2(110) Surface. J. Phys. Chem. C 2008, 112, 4257–4266.

25

(27) Beste, A.; Overbury, S. H. Pathways for Ethanol Dehydrogenation and Dehydration
Catalyzed by Ceria (111) and (100) Surfaces, J. Phys. Chem. C 2015, 119, 2447–2455.
(28) Capdevila-Cortada, M.; Vilé, G.; Teschner, D.; Pérez-Ramírez, J.; López, N. Reactivity
Descriptors for Ceria in Catalysis, App. Catal. B 2016, 197, 299–312.
(29) Capdevila-Cortada, M.; López, N. Entropic Contributions Enhance Polarity Compensation
for CeO2(100) Surfaces, Nat. Mater. 2017, 16, 328–334.
(30) Makov, G.; Payne, M. C. Periodic Boundary Conditions in ab initio calculations. Phys. Rev.
B: Condens. Matter Mater. Phys. 1995, 51, 4014–4022.
(31) Garcia-Ratés M.; López, N. Multigrid-Based Methodology for Implicit Solvation Models in
Periodic DFT. J. Chem. Theory Comput. 2016, 12, 1331–1341.
(32) Garcia-Ratés, M.; García-Muelas, R.; López, N. Solvation Effects on Methanol
Decomposition on Pd(111), Pt(111), and Ru(0001). J. Phys. Chem. C 2017, 121, 13803–
13809.
(33) Henkelman, G.; Jónsson, H. Improved Tangent Estimate in the Nudged Elastic Band Method
for Finding Minimum Energy Paths and Saddle Points. J. Chem. Phys. 2000, 113, 9978–
9985.
(34) Henkelman, G.; Uberuaga, B. P.; Jonsson, H. A. Climbing Image Nudged Elastic Band
Method for Finding Saddle Points and Minimum Energy Paths. J. Chem. Phys. 2000, 113,
9901–9904.
(35) Henkelman, G.; Jónsson, H. A. Dimer Method for Finding Saddle Points on High
Dimensional Potential Surfaces using only First Derivatives. J. Chem. Phys. 1999, 111,
7010–7022.

26

(36) Heyden, A.; Bell, A. T.; Keil, J. Efficient Methods for Finding Transition States in Chemical
Reactions: Comparison of Improved Dimer Method and Partitioned Rational Function
Optimization Method. J. Chem. Phys. 2005, 123, 224101.
(37) ioChem-BD

database

available

at:

https://iochem-bd.iciq.es/browse/review-

collection/100/19621/ad3bb8a8dfcf093a32167f52
(38) Vilé, G.; Colussi, S.; Krumeich, F.; Trovarelli, A.; Pérez-Ramírez, J. Opposite Face
Sensitivity of CeO₂ in Hydrogenation and Oxidation Catalysis. Angew. Chem. Int. Ed. 2014,
53, 12069–12072.
(39) Tundo, P.; Rossi, L.; Loris, A. Dimethyl Carbonate as an Ambident Electrophile. J. Org.
Chem. 2005, 70, 2219–2224.
(40) F. Haber, Z. Z. Gradual Electrolytic Reduction of Nitrobenzene with Limited Cathode
Potential. Elektrochem. Angew. Phys. Chem. 1898, 22, 506–514.
(41) de Castro Silva, I.; Aparecido Sigoli, F.; Odone Mazali, I. Reversible Oxygen Vacancy
Generation on Pure CeO2 Nanorods Evaluated by in Situ Raman Spectroscopy, J. Phys.
Chem. C 2017, 121, 12928–12935.
(42) Wang, Z.; Yang, G.; Zhang, Z.; Jin, M.; Yin, Y. Selectivity on Etching: Creation of HighEnergy Facets on Copper Nanocrystals for CO2 Electrochemical Reduction. ACS Nano 2016,
10, 4559–4564.
(43) Li, Q.; Rellán-Piñeiro, M.; Almora-Barrios, N.; Garcia-Ratés, M.; Remediakis, I. N.; López,
N. Shape Control in Concave Metal Nanoparticles by Etching. Nanoscale 2017, 9, 13089–
13094.

27

Table 1. Recycling results of CeO2 in the methoxycarbonylation of 2,4-TDA with DMC.a Values
are expressed as an average of all independent runs. The structures corresponding to each group
of compounds are shown in Section S2 of the Supporting Information.
SBETb

X2,4-TDAc

YMCd

YBC

YMeTDA

MeCe

PU+UUe

Unknowne

Cbal.f

(m2 g−1)

(%)

(%)

(%)

(%)

(area%)

(area%)

(area%)

(%)

#1
#2

64
55

100
100

7
9

83
83

0
0

3
3

4
2

1
2

98
99

#3
#4
#5

58
62
57

100
100
89

26
23
34

60
43
8

0
<1
4

4
7
15

2
4
4

1
2
8

93
79
84

#6g

68

80

12

4

11

33

2

7

89

Cycle

a Reaction

conditions: DMC:2,4-TDA = 30:1, catalyst loading = 20 m2 per mmol of 2,4-TDA, 413 K, 7.5 h, autogenous
pressure. b Surface area of the catalyst prior to each catalytic cycle. The weight of catalyst added to the reaction mixture was
varied in the range of 0.95-1.18 g, in order to keep a fixed ratio of 20 m2 per mmol of 2,4-TDA. c Conversion of 2,4-TDA.
d Corresponds to the sum of the yields of the para- and ortho- isomers. e Not quantified, results are expressed as area%. f Sum
of all the detected species. Cbal=(100-XTDA)+YMC,BC,MeTDA+area%MeMC, MeMC, PU, UU, unknown. g The values of the sixth cycle
correspond to the results after regeneration of the catalyst with DMC.

28

Scheme 1. Simplified reaction network of the N-methoxycarbonylation of 2,4-TDA with DMC over CeO2 catalysts. The detailed list
of structures is provided in Section S2 of the Supporting Information.
29

Figure 1. HRTEM of CeO2 catalyst in a) fresh form, after b) three and c) six consecutive cycles,
and after d) a 42-h treatment with DMC.
30

Figure 2. Concentration profiles as a function of the reaction time: a) 2,4-TDA, MC1, MC2, and
BC, b) productive ureas (PU), and c) N-methylated carbamates (MeMC and MeBC) and
N-methylated ureas (UU). Solid lines correspond to eye guidelines. The profile corresponding to
N-methylated TDA (MeTDA) is not shown as these compounds were not detected. The detailed
list of structures is provided in Section S2 of the Supporting Information.

31

Figure 3. Detailed reaction network of the methoxycarbonylation of 2,4-TDA with DMC over the
surface of CeO2 catalysts. The selective path is shown with black arrows. The side paths leading
32

to the N-methylated carbamates and N-methylated ureas are shown with red and blue arrows,
respectively.

33

Figure 4. Detailed reaction network of the methoxycarbonylation of 2,4-TDA with DMC in the
reaction medium. The selective path is shown with black arrows. The side paths leading to the
N-methylated compounds are shown with green arrows.

34

Figure 5. Reaction profiles of the selective (black) and side (red and blue for the paths leading to
the N-methylated carbamates and N-methylated ureas, respectively) pathways in the
methoxycarbonylation of 2,4-TDA with DMC over the stoichiometric (111) surface of CeO2. The
reaction on the most favorable para position for both the selective and side paths is considered.
The asterisks denote adsorbed molecules. The position (para or ortho) of the amino group that
reacts is indicated in parenthesis, e.g., MC2(p) refers to MC2 that reacts in the para position to
transform into BC.

35

Figure 6. Top view of the DFT-optimized adsorption configuration of the transition states for the
selective pathways leading to carbamates (middle) and productive ureas (right) and side pathways
leading to the formation of N-methylated carbamates (left) in the methoxycarbonylation of
2,4-TDA with DMC over the stoichiometric (111) surface of CeO2. All the structures can be
retrieved from the ioChem-BD database available at ref. (37).
36

Figure 7. a) Evolution of the yield of BC and combined yields of MC1, MC2, and BC over CeO2
upon consecutive cycles. b) Thermogravimetric analysis (TGA) in air atmosphere of CeO2 after
used in the methoxycarbonylation of 2,4-TDA with DMC in different cycles. The values of the
sixth cycle correspond to the results after regeneration of the catalyst with DMC. c) Wulff
constructions of dissociated DMC over stoichiometric (111) and (110) facets (blue and green,
respectively) for naked, DMC = 0, and partially DMC covered, DMC = 0.25 CeO2 nanoparticle.
37

d) Side view of the DFT-optimized adsorption configuration of dissociated DMC over
stoichiometric (111) and (100) surfaces.

38

TABLE OF CONTENTS GRAPHIC

39