Synthesis of formaldehyde from dimethyl ether on alumina-supported molybdenum oxide catalyst

Article abstract of DOI:10.1016/j.apcata.2016.09.002

The selective oxidation of DME to formaldehyde over alumina-supported MoO_x catalyst (prepared by dry impregnation) is studied in this work. The activity and stability of the catalyst were evaluated in a fixed-bed continuous reactor at different temperatures and reactant concentrations. The influence of the main operating conditions, (DME, O₂, CO₂ and CO feed concentrations; reaction temperature) on reaction rate and product selectivity was experimentally determined. Thus, DME conversion decreases on increasing DME feed concentration and increases on increasing O₂ feed concentration. Formaldehyde selectivity remained almost unaffected. A reaction mechanism, based on a Mars-van-Krevelen redox cycle representing DME oxidation to formaldehyde was used as a basis to develop a kinetic model for the reaction. The resulting simplified model suggests power law dependences for the reaction rate of 0.2 for the O₂ and 0.5 for the DME.

Full text of DOI:10.1016/j.apcata.2016.09.002

Applied Catalysis A: General 527 (2016) 137–145
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Synthesis of formaldehyde from dimethyl ether on
alumina-supported molybdenum oxide catalyst
Raquel Peláez, Pablo Marín, Salvador Ordón˜ez∗
Department of Chemical and Environmental Engineering, University of Oviedo, Faculty of Chemistry, Julián Clavería 8, 33006 Oviedo, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 29 May 2016
Received in revised form 2 September 2016
Accepted 7 September 2016
Available online 9 September 2016
The selective oxidation of DME to formaldehyde over alumina-supported MoOx catalyst (prepared by
dry impregnation) is studied in this work. The activity and stability of the catalyst were evaluated in a
fixed-bed continuous reactor at different temperatures and reactant concentrations.
The influence of the main operating conditions, (DME, O2, CO2 and CO feed concentrations; reaction
temperature) on reaction rate and product selectivity was experimentally determined. Thus, DME conver-
sion decreases on increasing DME feed concentration and increases on increasing O2 feed concentration.
Formaldehyde selectivity remained almost unaffected. A reaction mechanism, based on a Mars-van-
Krevelen redox cycle representing DME oxidation to formaldehyde was used as a basis to develop a
kinetic model for the reaction. The resulting simplified model suggests power law dependences for the
reaction rate of 0.2 for the O2 and 0.5 for the DME.
Keywords:
Renewable chemicals
Biomass processing
Partial oxidation
Redox catalyst
© 2016 Elsevier B.V. All rights reserved.
Kinetic modelling
1. Introduction
thesis of chemicals from DME is a very promising pathway in the
chemical industry.
The possibility of obtaining dimethyl ether (DME) from biomass
with high yields has substantially increased the interest of this
product as biofuel and a renewable chemical intermediate [1–4].
The process is based on the gasification of biomass into syngas, fol-
lowed by syngas cleaning and its catalytic conversion into DME. In
the traditional manufacture process of DME, the catalytic conver-
sion of the syngas consists of two steps: the conversion to methanol
and then the dehydration of methanol to DME [5].
At this point, formaldehyde is a major product obtained by
partial oxidation of methanol [8,9] (more than half amount of
global methanol is used to produce formaldehyde [10]). The
possibility of producing formaldehyde from DME would be a
promising alternative route. In addition, several new genera-
tion fuels and fuel additives, such as 1,1,-dimetoxymethane and
polyoxymethylene dimethylethers (POMMs), are synthetized from
DME-formaldehyde mixtures [11].
Recent advances have shown that the direct synthesis of DME
from syngas, where both reactions take place in the same reac-
tor using bifunctional catalysts, is an attractive alternative. As an
advantage, the thermodynamic limitations associated with the
classical process are relaxed, leading to higher conversion and
selectivity at lower pressure [1,6]. Thus, at the reaction conditions
at which methanol synthesis process reaches only a 40% of CO con-
version, DME synthesis from syngas can achieve more than 95% of
CO conversion, with a high DME selectivity (67%) [7]. Therefore,
DME production costs are largely lower, making this molecule into
a competitive platform molecule.
A new alternative formaldehyde production method has
emerged, after the new developments in the production of DME.
Formaldehyde can be obtained by partial oxidation of DME over
metal oxide catalysts. Several different catalysts have been pro-
posed and tested for this reaction, but low reaction rates and
selectivities have been reported. Mitsushima et al. proposed
tungsten oxide (with or without additives) as catalyst, with
a formaldehyde yield of 65–80%, although high temperatures,
673–773 K, are required [12]. Manganese nodules [13] are another
material proposed in the literature for this reaction, but low
formaldehyde yields were reported (<4%). The oxidation of DME
over metallic silver catalysts yields a mixture of HCHO, alkanes,
carbon oxides and water [14,15]. In addition, several authors sug-
gest the use of mixtures of bismuth and molybdenum oxides with
iron or copper oxides, or including with phosphorus and silicon
[16,17]. These catalysts achieve values between 24 and 46% of con-
version, with a formaldehyde selectivity of 14–45%, operating at
Although DME can be used as alternative fuel, with limited soot
formation when used in internal combustion engines [1,2], the syn-
∗
Corresponding author.
E-mail address: sordonez@uniovi.es (S. Ordón˜ez).
http://dx.doi.org/10.1016/j.apcata.2016.09.002
0926-860X/© 2016 Elsevier B.V. All rights reserved.

138
R. Peláez et al. / Applied Catalysis A: General 527 (2016) 137–145
high temperatures (673–773 K). Some of these catalysts have sig-
nificant yields but temperatures that are high are required and
the oxidation of formaldehyde to carbon oxides was frequently
reported, so it is necessary to propose alternative catalysts.
Recent studies of this reaction have proposed catalysts based
on Mo and V oxides [18,19]. It has been reported that MoOx and
VOx supported on Al2O3, ZrO2 and SnO2 present a good balance
between reactivity and accessibility of oxide surfaces. The use of
these materials has many advantages over other reported catalysts,
mainly higher reaction rate and formaldehyde selectivity, together
with a lower reaction temperature. In this sense, according to the
reported results, MoOx/Al2O3 is the most selective catalyst [18].
These studies suggest that the reaction proceeds via redox cycles
where DME dissociation was followed by an oxidation using lat-
tice oxygen. This is based on the fact that methanol oxidation to
formaldehyde occurs via Mars-van-Krevelen redox cycles with sur-
face methoxide intermediates and these intermediates can also be
generated via C O bond cleavage of the DME molecule [20–22].
Detailed kinetic studies have been developed for this reaction
using zirconia-supported MoOx catalysts [23]. For the alumina-
supported MoOx catalyst, several published studies report the
reaction orders for DME and O2 partial pressures [24,25].
Although alumina supported molybdenum oxide catalysts pre-
sented the best behavior, there are several aspects that have not
been considered, such as the influence of carbon oxides on catalyst
performance (typical components on DME feedstock derived from
syngas), or the development of a more detailed kinetic model for
this reaction with a MoO3/Al2O3 catalyst.
The scope of this work is to fill these gaps providing the fun-
damental information necessary for design industrial processes
for transforming DME into formaldehyde. To accomplish this goal,
an alumina-supported MoOx catalyst was prepared, characterized
and tested in a fixed-bed continuous reactor. First, the stabil-
ity of the catalyst upon time was studied. Then, a kinetic study
was conducted varying the most important operating conditions
(concentration of DME, O2, CO and CO2, and temperature). The
experimental data was used for proposing a simplified kinetic
model inspired on the mechanism proposed for the reaction.
The total amount of Mo on alumina was measured by Inductively
Coupled Plasma Mass Spectrometer, ICP-MS, with collision cell (HP
7700, Agilent Technologies), after the total digestion of the sample
in aqua regia.
The morphology of the catalytic material was investigated by
Scanning Electron Microscope (SEM). The analysis was conducted
using a JEOL-6610LV SEM-EDX. The samples were deposited on
a standard aluminum holder and gold-coated. The metal surface
concentration was also determined by EDX analysis.
X-Ray Diffraction (XRD) tests were carried out in a Seifert XRD
3000 diffractometer, which is equipped with a temperature con-
trolled chamber.
Used catalysts were analyzed by Temperature-Programmed
Oxidation (TPO) using a Micromeritics TPD/TPR 2900 coupled to
a Pfeiffer Vacuum Omnistar Quadrupole Mass Spectrometer (MS).
The samples were exposed to an oxidant gas (2% vol. O2) while the
temperature was increased (2.5 K min−1) from 293 K to 1273 K. The
evolution of CO and CO2 concentrations was monitored continu-
ously by MS. Temperature-programmed reduction (TPR) was also
performed in the same device to determine the oxidation states of
the catalyst, before and after the reaction. In the case, the samples
were exposed to a reducing gas (10% vol. H2 in Ar). The Origin Pro
8 analysis program was used for the signal processing.
The surface composition and binding energy of Mo, Al and C
in the oxides were measured by X-ray Photoelectron Spectroscopy
(XPS), using a SPECS system equipped with a Hemispherical Phoi-
bos detector operating in a constant pass energy (Mg K␣ radiation,
h·ꢀ = 1253.6 eV). During the deconvolution of the spectra, the full
widths at half maximum of Mo3d3/2 and Mo3d5/2 were taken the
same value, and the peak area ratio between both peaks was equal
to 2:3. The full widths at half maximum of the C1s spectra for the
different species were assumed to be equal, and this procedure was
also applied for the O1s spectra.
2.4. Experimental device
Experiments were carried out in a continuous fixed-bed isother-
mal reactor. The reactor consisted of a stainless steel tube (9 mm
diameter and 600 mm length) with the catalyst sample (2 g,
100–250 ␮m) placed inside. The catalyst was diluted with glass
particles (6 g, 355–710 ␮m) in order to minimize temperature gra-
dients within the catalyst bed. Temperature was measured by
several thermocouples along the tube wall and one thermocou-
ple placed inside the reactor tube. The latter was used to control
the reactor temperature using a PID controller. The reactants were
mixed at the desired proportions using different mass flow con-
trollers (Bronkhorst High-Tech instruments).
2. Materials and methods
2.1. Chemicals and reactants
The reactant mixture consisted of dimethyl ether, O2, CO, CO2
and N2 as balance gas. These reaction gases and chromatographic
gases (He, H2, Air) were supplied by Air Liquide with purities higher
than 99%, and used without further purification.
The reactor effluent was maintained at 423 K using a heating
tape to prevent formaldehyde condensation or oligomerization.
On-line analysis of the reactor feed and effluent streams was car-
ried out using a Gas Chromatograph (GC Agilent HP 6890N). It is
equipped with a HP Plot Q capillary column for CO2, DME, methanol,
water and formaldehyde analysis, and a HP MoleSieve 5A capillary
column for CO, O2 and N2 determination. The HP MoleSieve 5A col-
umn is connected to a valve which allows its connection or isolation
from the system, according to the required analysis. Both columns
are connected to two detectors: thermal conductivity (TCD) and
flame ionization detectors (FID). The temperature program of the
analysis is the following: 70 ◦C for 4.5 min, then a ramp of 10 ◦C/min
up to 160 ◦C and a second ramp of 20 ◦C/min up to 200 ◦C and hold
5 min; finally, cold down to 70 ◦C at 30 ◦C/min and hold 11 min.
2.2. Preparation of the MoOx/Al2O3 catalyst
The catalyst was supported on ␥-Al2O3 particles (BASF, surface
area of 242 m2 g−1), previously ground to 100–250 ␮m. The active
phase was added by incipient wetness, using an aqueous solution of
ammonium heptamolybdate tetrahydrate ((NH4)6Mo7O24·4H2O,
99% Fluka). The impregnated solid was dried at 373 K overnight
and treated in an air flow at 775 K (10 K min−1 until 775 K, holding
for 3 h) [10,18].
2.3. Catalyst characterization
The textural characteristics of the fresh and used catalysts were
determined by nitrogen physisorption at 77 K in a Micromeritics
ASAP 2020 analyzer by the Brunauer–Emmett–Teller (BET) method
for the specific surface area, and the Barrett–Joyner–Halenda (BJH)
approach to determine the pore volume and diameter.
2.5. Reaction experiments
The catalyst stability was studied by operating the reactor at
constant pass conditions for long reaction times (typically more

R. Peláez et al. / Applied Catalysis A: General 527 (2016) 137–145
139
than 20 h). The influence of the different operating variables (e.g.
temperature, DME, oxygen, CO and CO2 concentrations) on catalyst
performance was also studied.
DME conversion (XDME) was calculated from the inlet and outlet
DME molar fractions (yDME) and the total molar flowrates (F) using
the following expression:
15%
10%
5%
ꢀ
ꢁ
yDME
yDME
F
XDME = 1 −
(1)
F0
feed
The selectivity to the different products (Si) was determined as
the total amount of desired product formed per total amount of
reactant consumed.
ꢂ ꢂ
ꢃ
ꢃ
F
F
yi
− yi
/ꢁi
feed
0
0%
ꢂ
ꢃ
Si =
(2)
yDME XDME /ꢁDME
0
5
10
15
20
feed
t (h)
where yi and yifeed were the molar fraction of a compound in the
outlet stream and in the feed respectively. F/F0 corresponded to the
ratio between the total molar flowrates in the outlet stream and in
the feed, and it was obtained by applying the total molar balance to
the reactor. The terms ꢁi and ꢁDME correspond to the stoichiometric
coefficients of the compounds on the reaction, and XDME, as already
mentioned, is the DME conversion.
Fig. 1. Evolution of DME conversion with time on stream.
The error associated to the reaction experiments was estimated
to be less than 5%.
The fixed-bed reactor has been modelled as plug-flow, according
to the DME mass balance:
dXDME
−r
(
)
DME
Scheme 1. Primary and secondary reaction pathways for dimethyl ether oxidation.
=
(3)
dw¯
yDMEfeed SV
Table 1
Where w¯ was the fractional length of the reactor fixed-bed (0 = inlet,
was DME reaction rate per unit weight of cata-
lyst and SV = F/W was the space velocity.
Peak position in XPS spectra of MoO3/Al2O3 system.
1 = outlet), −r
(
)
DME
Eb (eV)
Fresh
Mo 3d5/2
233.3
Mo 3d3/2
236.5
O1s
C1s
531.2
284.6
286.5
288.8
284.3
286.8
289.2
3. Results and discussion
Used
233.3
236.4
531.2
533.5
3.1. Catalyst stability
The first studies were devoted to study the catalyst stability at
reaction conditions in a continuous fixed-bed reactor. The reactor
was operated at constant conditions, 0.2 MPa absolute pressure and
513 K, with a space velocity of 0.67 kmol kg−1 h−1 (feed flow rate
of 0.5 NL min−1). The feed composition was 10% DME and 13% O2
with N2 balance; hence, the reaction was carried out under oxygen-
rich conditions. Reactor outlet concentrations were measured as a
function of time to calculate conversion and selectivity.
The results indicated the existence of an initial stabilization
period of 14 h, where DME conversion decreased as a function of
time from 13% to 6% (Fig. 1). Afterwards, DME conversion was main-
tained constant. At this point, the main reaction products were
formaldehyde, CO, methanol and CO2, remaining the selectivity for
formaldehyde formation higher than 65% during all the experiment.
Concerning the other reaction products, selectivities for the forma-
tion of CO and CO2 were always lower than 12 and 14%, respectively,
whereas selectivity for methanol formation was about 8% during
all the experiment. According to these facts, the reaction scheme
depicted in Scheme 1 was proposed to describe the global process.
In addition to DME oxidation to formaldehyde, side reactions of
DME hydration to methanol and formaldehyde oxidations to form
COx can take place. The possibility of CO oxidation to CO2 over these
catalysts, has been ruled out by additional experiments carried out
with a CO feed at the same conditions.
due to a further oxidation of formaldehyde, with formic acid as
intermediate (though rarely detected due to its low concentration).
3.2. Catalyst characterization
Nitrogen physisorption analysis of the fresh and used cata-
lysts showed a decrease in the surface area after reaction, from
230 to 144 m2 g−1 (37% lower). This may explain the decrease
in DME conversion during the catalyst stabilization. Fig. 2 shows
the XPS spectra in the fresh and the used catalyst. The pres-
ence of two well resolved spectral lines, assigned to the Mo3d5/2
and Mo3d3/2 spin-orbit components were observed. The corre-
sponding electron binding energies (Eb) are reported in Table 1.
The experimental envelope of MoO3/Al2O3 system was fitted to
a single Mo3d3/2–Mo3d5/2 doublet (236.5 eV and 233.3 eV respec-
tively). This indicates the presence of only one type of molybdenum
(VI) oxo species, which strongly interacts with alumina surface.
The XPS analysis conducted to the used catalyst presented the
Mo3d3/2–Mo3d5/2 doublet at the same energies, suggesting that
Mo oxidation state did not change after reaction.
These results are in good agreement with the reducibility of the
surface species determined by TPR. Fig. 3 shows the TPR profiles
of the MoOx/Al2O3 catalyst before (a) and after (b) the reaction.
Two main reduction regions were easily distinguished in both
plots: at 533–733 K and 763–1193 K. The low temperature peak was
attributed to a first reduction of Mo6+ to Mo4+ of octahedrally coor-
A similar reaction scheme was reported in the literature [26]
for formaldehyde synthesis from methanol, which resulted in the
formation of dimethyl ether, methylal, methyl formate or carbon
oxides, as secondary products. The formation of carbon oxides was

140
R. Peláez et al. / Applied Catalysis A: General 527 (2016) 137–145
a)
Mo3d
b)
243
239
235
231
227 243
239
235
231
227
Binding Energy (eV)
Binding Energy (eV)
O1s
540
536
532
528
524
520 540
536
532
528
524
520
Binding Energy (eV)
Binding Energy (eV)
C1s
296
292
288
284
280
276 296
292
288
284
280
276
Binding Energy (eV)
Binding Energy (eV)
Fig. 2. XPS spectra of the fresh (a) and used (b) catalyst (Mo3d, O1s and C1s spectra).
dinated molybdenum species weakly bound to the support (Al2O3),
predominantly as multilayer molybdenum domains, with some
polymolybdates in monolayer patches [27]. The second reduction
peak was associated to the reduction of Mo4+ to Mo0 of molybde-
num species strongly bound to Al2O3 and those previously partially
reduced at low temperature [28]. Both reductions consumed a total
amount of 1.6 mmol H2 gcat−1, with a ratio between H2 consumed
alyst showed a unique peak at binding energy of 531.2 eV. It is
attributable to the overlapping of oxygen from MoO3 and Al2O3,
with energies of 230.6 eV and 231.1–231.4 eV [29]. Used catalyst
O1s spectrum showed an additional peak at higher binding energy
(533.5 eV). According to the reported results, peaks at ∼530 eV
was attributed to oxides whereas the peak at ∼532 eV was due to
organic oxygen species adsorbed on the catalyst surface [30]. This
second analyzed peak can be assigned to methoxy species formed
during the reaction, probably CH3O-species bounded to multiple
Mo ions [10]. This is confirmed by the C1s spectra. The C1s spec-
trum of the fresh catalyst corresponds with adventitious carbon,
typically detected in samples that have been exposed to the atmo-
sphere or generated during the analysis. By comparison with it, the
C1s spectrum of the used catalyst shows higher amount of the com-
ponent that appears at 286.8 eV. These binding energies have been
reported to correspond to the methoxy species formed as reac-
−1
in the two reduction peaks of 1/2 (0.54 mmol H2 gcat in the first
−1
peak and 1.11 mmol H2 gcat in the second). This is the stoichio-
metric relation between the Mo6+ and Mo4+ reduction reactions,
so the reduction of Mo4+ in the second peak corresponds to the
Mo6+ species previously reduced to Mo4+. These findings confirm
that only MoO3 was present in the catalyst before and after the
reaction.
The analysis of O1s and C1s XPS spectra provided new insights
about the catalyst deactivation. The O1s spectra for the fresh cat-

R. Peláez et al. / Applied Catalysis A: General 527 (2016) 137–145
141
Table 2
a
EDX analysis results of fresh and post-reaction catalysts.
% wt.
C
O
Al
Mo
Fresh
Std
–
–
58.4
6.3
31.3
5.3
10.3
1.3
Used
Std
37.0
2.9
51.1
1.7
10.7
1.8
1.2
0.09
The morphology and surface properties of the catalyst, obtained
with a SEM-EDX, were compared before and after the reaction
(Fig. 5). In fresh catalyst, it can be observed a uniform distribution of
MoOx particles on the surface the catalyst, suggesting an adequate
preparation procedure. The elemental composition of the catalyst
surface of the fresh and used catalysts was measured by EDX analy-
sis in a 10–30 ␮m surface (Table 2). The results confirm the presence
of C in the used catalyst, but also O was found in high amount. This
fact suggested the formation of methoxy oligomers (with C and O)
that can explain the decrease in conversion observed during the
first 14 h in the catalyst stabilization test. After this time, the activ-
ity of the catalyst remained constant, suggesting that the formation
of these species reached a steady state. The following kinetic tests
have been carried out at steady state with the catalyst stabilization
finished.
293
433
573
713
853
T (K)
993
1133
1273
b
The X-ray diffraction patterns of the fresh catalyst did not show
any reflections attributable to MoO3, with only the presence of
characteristic peaks of ␥-Al2O3 being observed. At this point, Chen
et al. [33] suggested the absence of Mo phases characteristic peaks
for Mo surface densities lower than 5 Mo atoms nm−2. It can be
concluded that, as indicated by EDX analysis, the MoO3 was well
distributed on the support surface of the fresh catalyst, with a high
dispersion. The chemical composition determined by ICP-MS anal-
ysis showed a total Mo concentration of 8.4% wt. on the catalyst
293
433
573
713
853
T (K)
993
1133
1273
Fig. 3. TPR profiles for fresh (a) and used (b) catalyst.
3.3. Kinetic experiments
The following reaction experiments were devoted to determine
the kinetics of the DME oxidation reaction to formaldehyde. These
experiments have been carried out during the constant activity
period observed after 14 h of stabilization under reaction condi-
tions.
The first set of experiments studied the influence of DME feed
concentration within the range 3% to 14%. All the other operation
conditions were kept unaltered: 13% O2 (and N2 balance), space
velocity 0.94 kmol kg−1 h−1, 2 bar and 518 K. Fig. 6 shows the results
of the experiments in terms of DME conversion and product selec-
tivity. DME conversion decreases from 17% to 7% when DME mole
fraction in the feed increases from 3% to 14%. This fact suggests an
apparent reaction order of DME much lower than one. On the con-
trary, the selectivity for the formation of different products remains
almost constant. The main product obtained is HCHO with 69.0%
selectivity, while the other products have a selectivity of: 20.6% for
CO, 4.0% for CO2 and 6.4% for methanol.
Typical DME feedstock derived from syngas is not composed
only of DME, but can also contain other gases, such as CO and CO2.
The influence of these gases on the catalyst performance should be
also considered. The effect of these gases on reactor performance
was studied separately, by introducing reactor feeds with a 12%
of CO2 or 4% of CO, typical downstream concentrations in DME-
containing streams derived from biomass. The rest components
of the feed consisted of 10% DME and 13% O2 (with N2 balance).
The operating conditions were the same as before: space veloc-
ity 0.94 kmol kg−1 h−1, pressure 0.2 MPa and temperature 518 K.
According to the experimental results, it can be concluded that
both conversion and selectivity are not affected when a CO/CO2-
containing feed is introduced in the reactor.
273
373
473
573
673
773
873
T (K)
Fig. 4. TPO analysis of the catalyst after being used in the reaction.
tion intermediates and CO [10,31]. These methoxy species were
probably bonded to more than two Mo ions and are related to COx
formation [10]. A slight increase on the higher energy peak was also
detected (289.2 eV). This peak can be attributed to CH2O (288.8 eV),
whose spectrum of oxygen (533.1 eV) could overlap with the cor-
responding to methoxy species [32].
The used catalyst was also analyzed by TPOs. The CO2 release
profiles were depicted in Fig. 4. The presence of a large peak in the
range 573–773 K, with maximum at 697 K, indicates the presence
of heavy organic compounds adsorbed on the surface of the catalyst
that are oxidized at this temperature. The total amount of formi−n1g
CO2 on the aged catalyst corresponds with 188.3 ␮mol CO2 gcat
.
The presence of these compounds is in agreement with the XPS
measurements of the C1s spectrum (hydrocarbon and methoxy
species).

142
R. Peláez et al. / Applied Catalysis A: General 527 (2016) 137–145
Fig. 5. SEM micrographs of fresh (a) and used (b) catalysts.
The influence of oxygen feed concentration was also assessed by
limitations for the most unfavorable conditions. To rule out inter-
nal and external heat transfer limitations determination, several
parameters proposed in the literature were used [34].
means of experiments carried out in the range 1% to 13% O2; with
10% of DME and N2 balance, and the same operating conditions con-
sidered before. Fig. 7 shows that the influence of oxygen on DME
conversion was more marked at low concentration; above 7% O2,
DME conversion was nearly constant to 9%. Regarding the selectiv-
ity of the different compounds, they were found to be constant also
during this study.
The results of the kinetic experiments of this section have
been fit to the following power-law empirical rate expression
(R2 = 0.921), using the plug flow reactor model of Eq. (3):
The conditions at which external mass transfer is more
likely to affect the overall kinetics correspond to
a reactor
feed of 14% DME, 13% O2 (N2 balance) at 518 K. For internal
mass transfer, the worst conditions correspond to a reactor
feed of 3% DME, 13% O2 (N2 balance) at the same tempera-
ture. The external mass transfer was discarded, with a Carberry
number, Ca = robs/KGaSCG = 5 × 10−6 < 0.05, and the same was
found with the internal mass transfer, Wheeler–Weisz crite-
ria, ꢂꢃ2 = robsdp2/DeCS = 1 × 10−1 < 0.1. External heat transfer,
|␤ex␥ECa| = |KG(−ꢄH)CGEaCa/hRTG2| = 1 × 10−3 < 0.05, and internal
heattransfer, |␤in␥E ꢂꢃ2| = |De(−ꢄH)CSEaꢂꢃ2/keRTS2| = 9 × 10−3 < 0.05,
were also negligible.
0.50
0.20
−rDME = k CH OCH
O
]
2
(4)
[
]
[
3
3
With k = 8.73 × 10−4 mol0.3 (m3)0.7 kg−1 s−1
This expression can be compared to the analogous one reported
in the literature for methanol oxidation to formaldehyde [20]. The
reported reaction order of methanol was close to unity, 0.94, which
is almost double the value fitted for DME. This fact is explained con-
sidering the reaction mechanism described in the following section.
Regarding oxygen, the reaction order was low, 0.10, but similar to
the value obtained experimentally in the reaction studied in the
present work.
3.4. Reaction mechanism
The reaction mechanism shown in Table 3 is proposed to explain
the oxidation of dimethyl ether to formaldehyde. This mechanism
is based on previously reported studies [10,24] and the mechanism
of methanol oxidation to formaldehyde proposed in the literature
[20,26,35]. A new kinetic equation was developed and experimen-
Carberry number and Wheeler–Weisz criteria were applied in
order to check the absence of external and internal mass transfer

R. Peláez et al. / Applied Catalysis A: General 527 (2016) 137–145
143
15%
25%
20%
15%
10%
5%
10%
5%
0%
0%
0.00
0.05
0.10
0.15
0.02
0.06
0.10
0.14
yO2
yDME
feed
feed
80%
60%
40%
20%
0%
80%
60%
40%
20%
0%
0.02
0.06
0.10
0.14
0.00
0.05
0.10
0.15
yO2
yDME
feed
feed
Fig. 7. Oxygen feed concentration influence over (a) reactant conversion ( ) and
Fig. 6. DME feed concentration influence over (a) reactant conversion ( ) and (b)
(b) selectivities to: formaldehyde ( ), methanol ( ), CO ( ) and CO2
oxidation to formaldehyde (518 K, 2 bar, 21 Nm3 kg−1 h−1).
(
), in DME
selectivities to: formaldehyde ( ), methanol (
), CO ( ) and CO2
(
), in DME
oxidation to formaldehyde (518 K, 2 bar, 21 Nm3 kg−1 h−1).
in the literature for the methanol oxidation mechanism, where
the rate-determining step is the conversion of the methoxy
intermediates to formaldehyde by abstraction of a hydrogen atom
[20]. It is assumed that HCHO adsorption on the surface is weak,
so it can desorb easily. Considering this reaction mechanism and
the aforementioned assumption, the following kinetic expression
was derived. The complete derivation is given as Supplementary
information.
Table 3
Proposed mechanism for the oxidation of DME to formaldehyde catalyzed by
MoOx/Al2O3.
K
1
Adsorption
CH3OCH3 + M∗ + O∗↔M∗ − O∗ − CH3OCH3
K
2
M∗ − O∗ − CH3OCH3↔2CH3O∗
k
3
Formaldehyde formation
Methanol formation
Water formation
CH3O∗ + O∗→HCHO + HO∗ + ∗
K
4
CH3O∗ + HO∗↔M∗ + O∗ + CH3OH
K
5
2HO∗↔M∗ + O∗ + H2O
1/2
k /2 K K 1/2Ct2 CH OCH
(
)
[
]
K
3
1
2
3
3
6
Re-oxidation
2 ∗ +O2↔2O∗
−rDME
=
(5)
ꢀ
ꢁ
2
ꢂ
ꢃ
1/2
1
2 + K K CH OCH
+
(
[
3])1/2
2
1
3
K
O
]
2
[
6
tally verified with the results of the kinetic experiments of the
previous section.
3.5. Kinetic modelling
The first step of the reaction pathway is the non-dissociative
DME chemisorption, which occurs with interaction with lattice
oxygen and metal sites. Then, the adsorbed DME dissociates yield-
ing methoxide intermediates. Both steps are fast and assumed in
equilibrium. Secondly, formaldehyde is generated by irreversible
hydrogen loss of the methoxide species *OCH3 promoted by a
neighboring lattice oxygen atom. In this reaction, a *OH and a
reduced Mo cation are also formed. Laterally, methanol production
by *OCH3 and *OH reversible reaction also can take place. Reversible
reaction involving two *OH groups forms the water product of the
reaction and the dissociative chemisorption of O2 re-oxidizes the
Mo sites.
In this section, the proposed model was validated, and the
parameters fitted, according to the results of the kinetic experi-
ments.
- DME influence
In the experiments carried out varying the feed concentration
of DME, O2 was in excess, so its concentration can be assumed
constant and the previous kinetic expression was simplified:
1/2
CH OCH
[
]
3
3
−rDME
=
(6)
ꢂ
ꢃ
2
1/2
According
to
the
experimental
results,
the
oxy-
Kꢀ + Kꢀꢀ CH OCH
[
]
3
3
dehydrogenation of the methoxide *OCH3 species to generate
formaldehyde is the most likely rate-determining step of the
mechanism. This mechanism is in agreement with that reported
This expression is only dependent on DME concentration. The
parameters were fitted to the experimental data, considering the

144
R. Peláez et al. / Applied Catalysis A: General 527 (2016) 137–145
15%
10%
5%
fixed-bed as a plug flow, by the least-squares method using the
EXCEL Solver. This has allowed us to determine that the term
located in the denominator dependent of dimethyl ether concen-
tration is not significant compared to those who accompany it, so
the rate expression can be further simplified:
1/2
−rDME = kap CH OCH
(7)
[
]
I
3
3
The adequacy of this last model was evaluated by comparison of
experimental and predicted values, as shown in Fig. 6. The regres-
sion coefficient of the fit was R2 = 0.990. It can be concluded that,
in excess oxygen, the DME apparent reaction order proposed by
the mechanistic kinetic model was in agreement with the exper-
iments (0.5). According to the reaction mechanism, DME adsorbs
and dissociates in two methoxide species *OCH3, which result in an
apparent reaction order of 1/2. On the contrary, in the oxidation of
methanol to formaldehyde, one methanol molecule generates only
one methoxide, and for this reason the reaction order reported in
the literature [20] was close to one.
0%
506
508
510
512
514
516
518
520
T (K)
80%
60%
40%
20%
0%
- Oxygen influence
The findings of the previous section allowed the simplification
of the general kinetic expression, Eq. (5):
k3ꢀ CH OCH
1/2
[
]
3
3
−rDME
=
(8)
ꢄ
ꢅ
2
ꢀ
K
6
1 +
1/2
O
[
]
2
The experiments performed at different oxygen feed concentra-
tions were used to evaluate the validity of the denominator term
regarding oxygen and fit the corresponding kinetic parameters.
Fig. 7 shows the least-square fit of the model to the experimental
data (R2 = 0.980). The value of the model parameters is the follow-
ing: the kinetic constant k3ꢀ = 1.51 × 10−3(m3)0,5(mol)0,5 kg−1 s−1
506
508
510
512
514
516
518
520
T (K)
Fig. 8. Temperature effect over DME conversion ( ) and selectivities to: formalde-
hyde ( ), methanol (
), CO ( ) and CO2
(
), in DME oxidation to formaldehyde
(2 bar, 21 Nm3 kg−1 h−1).
and O2 adsorption constant K6ꢀ = 0.29 mol0.5/(m3)0.5
.
Table 4
Summary of the main equations of the mathematical model proposed for the mod-
elling of the oxidation of DME catalyzed by MoOx/Al2O3.
- Influence of the reaction temperature
Mass balance to the reactor
The influence of temperature on the catalyst performance has
been studied in the range 508–518 K, with a feed composed of
10% of DME and 13% of O2 (N2 balance). Fig. 8 shows the effect of
this factor on DME conversion and product selectivity. On increas-
ing temperature, conversion increased (from 5.0% at 508 K to 8.1%
at 518 K), which was attributed to an increase in reaction rate.
Formaldehyde was obtained as the major product at all temper-
atures, with a maximum selectivity of 70% reached at 515 K. This
temperature corresponded to minimum selectivity towards CO
(13.7%). On the contrary, the selectivity of both, methanol and CO2
decreased with temperature.
The reaction was also tested at higher temperatures. However,
when temperature was increased above 523 K, complete oxidation
of formaldehyde was favored, which resulted in fast reaction and
complete oxygen consumption. Therefore, the maximum recom-
mended temperature for this reaction is 518 K.
dF
DME
= −r1 − r2
dꢅ
dF
FAL
r1 = kapI C1/2
= 2r1 − r3 − r4
dꢅ
DME
dF
dF
MeOH
= 2r2
= r3
1/2
dꢅ
r2 = kap
C
II
DME
CO
2
r3 = kap CFAL
dꢅ
dF
III
CO
= r4
r4 = kap CFAL
IV
dꢅ
dF
O
1 r4
2
= −r1 − r3
−
= r1 − r2 + r32 + r4
dꢅ
dF
H
O
2
dꢅ
Initial conditions
FDME| = FDME|
0
feed
FFAL| = FMeOH| = FCO
= FCO| = FH
= 0
O|
|
0
0
2
0
0
2
0
FO
= FO
|
feed
2
2
|
0
98 6 kJ mol−1 [20]. This value is slightly lower than the experi-
mentally observed for formaldehyde synthesis from DME.
The experiments were carried out in excess of oxygen, and
also considering the findings of the previous sections regarding
the kinetics of the reaction, the kinetic expression that can be
used to describe the experiments at these conditions was Eq. (7).
Using the conversion data, the apparent kinetic constant was cal-
culated for each temperature, and fitted to the Arrhenius equation
(R2 = 0.972). The resulting activation energy was 108.5 kJ mol−1
3.6. Reaction model
Once the kinetic model corresponding to the reaction of DME
oxidation to formaldehyde has been determined, the model was
extended to all the reactions of Scheme 1. The compound balances
and kinetic expressions are summarized in Table 4. By solving this
model, both DME conversion and selectivity to formaldehyde can
be predicted.
and the pre-exponential factor 9.09 × 107 (mol)0,5 (m3)0,5 kg−1 s−1
.
The activation energy of the analogous reaction of formalde-
hyde synthesis from methanol was, according to the literature,

R. Peláez et al. / Applied Catalysis A: General 527 (2016) 137–145
145
Table 5
Acknowledgements
Experimental values for the reaction apparent constants of the proposed reaction
scheme at 518 K and 2 bar.
This work has been financed by Research Projects of the Regional
Government of Asturias (project reference GRUPIN14-078) and
Spanish Ministry of Economy and Competitiveness (CTQ2014-
52956-C3-1-R) and by BPP company. Raquel Peláez acknowledges
the Spanish Ministry of Education for the PhD grant that supports
her research.
ki × 103
kI [(m3)0.5(mol)0.5 kg−1 s−1
]
]
1.04 0.2
0.07 0.1
0.60 1.6
3.02 1.6
kII [(m3)0.5(mol)0.5 kg−1 s−1
kIII [(m3) kg−1 s−1
kIV [(m3) kg−1 s−1
]
]
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.apcata.2016.09.
002.
References
[1] J. Sun, G. Yang, Y. Yoneyama, N. Tsubaki, ACS Catal. 4 (2014) 3346–3356.
[2] T.A. Semelsberger, R.L. Borup, H.L. Greene, J. Power Sources 156 (2006)
497–511.
[3] L. Faba, E. Díaz, S. Ordón˜ez, Renew. Sustain. Energy Rev. 51 (2015) 273–287.
[4] K. Takeishi, Y. Akaike, Appl. Catal. A: Gen. 510 (2016) 20–26.
[5] Z. Azizi, M. Rezaeimanesh, T. Tohidian, M.R. Rahimpour, Chem. Eng. Process.
Process Intensif. 82 (2014) 150–172.
[6] V.V. Ordomsky, M. Cai, V. Sushkevich, S. Moldovan, O. Ersen, C. Lancelot, V.
Valtchev, A.Y. Khodakov, Appl. Catal. A: Gen. 486 (2014) 266–275.
[7] J.L. Li, X.G. Zhang, T. Inui, Appl. Catal. A: Gen. 147 (1996) 23–33.
[8] R.E. Kirk, D.R. Othmer, Kirk–Othmer Encyclopedia of Chemical Technology,
4th ed., Wiley, New York, 1996.
[9] G. Reuss, W. Disteldorf, O. Grundler, A. Hilt, I.F. Ullmann, W. Gerhartz, Y.S.
Yamamoto, F.T. Campbell, R. Pfefferkorn, J.F. Rounsaville (Eds.), Ullmann’s
Encyclopedia of Industrial Chemistry, 5th ed., Wiley-VCH, Deerfield Beach, FL,
USA, 1985.
[10] X. Huang, J. Liu, J. Chen, Y. Xu, W. Shen, Catal. Lett. 108 (2006) 79–86.
[11] G.P. Hagen, M.J. Spangler, US Patent 6,160,186, BP Amoco Corporation (2000).
[12] H. Mitsushima, T. Murakami, H. Tadenuma, US Patent 3,655,771, Sumitomo
Chemical CO. LTD (1972).
[13] R.M. Lewis, L.H. Slaugh, US Patent 4,435,602, Shell Oil Company (1984).
[14] G.P. Hagen, M.J. Spangler, US Patent 6,265,528, BP Corporation North America
Inc (2001).
Fig. 9. Experimental data (ꢁ) and model prediction (—) to the study about the influ-
ence of DME initial concentration. DME ( ), formaldehyde ( ), methanol ( ), CO2
(ꢁ) and CO ( ).
In order to get the model constants, the experimental data from
the test performed at different initial DME concentrations were
used. The model was solved in MATLAB using ode15s. The kinetic
constants were fitted by the least square method using lsqcurvefit.
Table 5 shows the obtained values and the corresponding confi-
dence intervals. It can be observed that the reactions of synthesis
of methanol and CO2 did not have much significance in the overall
reaction scheme, because the amount of these compounds was very
low. Fig. 9 shows the experimental and model predicted concen-
trations of the different compounds. Formaldehyde was the main
reaction product, being the oxidation of formaldehyde to CO the
main side reaction.
[15] L.B. Ralph, US Patent 2,246,569, The Solvay Process Company (1941).
[16] R.M. Lewis, R.C. Ryan, L.H. Slaugh, US Patent 4,439,624, Shell Oil Company
(1984).
[17] R.M. Lewis, R.C. Ryan, L.H. Slaugh, US Patent 4,442,307, Shell Oil Company
(1984).
[18] H. Liu, E. Iglesia, J. Catal. 208 (2002) 1–5.
[19] H. Liu, P. Cheung, E. Iglesia, Phys. Chem. Chem. Phys. 5 (2003) 3795–3800.
[20] W.L. Holstein, C.J. Machiels, J. Catal. 162 (1996) 118–124.
[21] S.T. Oyama, R. Radhakrishnan, M. Seman, J.N. Kondo, K. Domen, K. Asakura, J.
Phys. Chem. B 107 (2003) 1845–1852.
[22] N. Pernicone, F. Lazzerin, G. Liberti, G. Lanzavecchia, J. Catal. 14 (1969)
293–302.
4. Conclusions
The partial oxidation of dimethyl ether to formaldehyde was
studied over a MoOx/Al2O3 catalyst in an isothermal fixed-bed reac-
tor. The use of this catalyst has been showed to be an efficient way of
obtaining formaldehyde from DME. Although a slight deactivation
was observed at the initial hours, mainly because of the formation of
formaldehyde-derived oligomers, the catalyst reached steady state
conditions. High (higher than 65%) and almost constant formalde-
hyde selectivity was obtained in the reported experiments.
Experimental results indicated that conversion depends on DME
and O2 concentration. A reaction mechanism based on the for-
mation of methoxy intermediates, followed by the rate-limiting
reaction over the oxidized catalyst centers, has been proposed. The
kinetic model derived from this mechanism was found to repre-
sent the experimental data regarding dimethyl ether oxidation at
different conditions.
[23] H. Liu, P. Cheung, E. Iglesia, J. Phys. Chem. B 107 (2003) 4118–4127.
[24] P. Cheung, Liu, E. Iglesia, J. Phys. Chem. B 108 (2004) 18650–18658.
[25] X. Huang, Y. Li, Y. Xu, W. Shen, Catal. Lett. 97 (2004) 185–190.
[26] J.M. Tatibouët, Appl. Catal. A: Gen. 148 (1997) 213–252.
[27] V.M. Benitez, C.A. Querini, N.S. Fıı´goli, Appl. Catal. A: Gen. 252 (2003) 427–436.
[28] R. López Cordero, F.J. Gil Llambias, A. López Agudo, Applied Catalysis 74
(1991) 125–136.
[29] B.M. Reddy, B. Chowdhury, E.P. Reddy, A. Fernández, Appl. Catal. A: Gen. 213
(2001) 279–288.
[30] J.G. Choi, L.T. Thompson, Appl. Surf. Sci. 93 (1996) 143–149.
[31] O. Rodrıı´guez de la Fuente, M. Borasio, P. Galletto, G. Rupprechter, H.J. Freund,
Surf. Sci. 566–568 (Part 2) (2004) 740–745.
[32] C.T. Au, W. Hirsch, W. Hirschwald, Surf. Sci. 221 (1989) 113–130.
[33] K. Chen, S. Xie, A.T. Bell, E. Iglesia, J. Catal. 198 (2001) 232–242.
[34] J. Fernández, P. Marín, F.V. Díez, S. Ordón˜ez, Fuel Process. Technol. 133 (2015)
202–209.
[35] A.P.V. Soares, M.F. Portela, A. Kiennemann, Catal. Rev. 47 (2005) 125–174.
The overall reaction scheme was also considered and the side
reactions fitted to their corresponding kinetic models. Overall,
formaldehyde formation was the most important reaction, fol-
lowed by formaldehyde oxidation to carbon monoxide.