Catalysis Today 128 (2007) 191–200

www.elsevier.com/locate/cattod

Kinetics of the oxidative dehydrogenation (ODH) of methanol

to formaldehyde by supported vanadium-based nanocatalysts
M. Cozzolino, R. Tesser, M. Di Serio, P. D’Onofrio, E. Santacesaria *
Department of Chemistry, University of Naples FEDERICO II, Complesso di M.te S. Angelo, Via Cinthia (80126) Napoli, Italy

Available online 21 August 2007

Abstract
The aim of the present contribution was to develop a detailed kinetic analysis of the oxidative dehydrogenation (ODH) reaction of methanol to
formaldehyde on a nano-structured supported vanadium oxide catalyst, selected in a preliminary screening. The chosen vanadium catalyst,
supported on TiO2/SiO2, has been prepared by grafting vanadyl alkoxide, dissolved in dioxane, and characterized by BET, XRD, Raman, XPS and
SEM. An exhaustive set of experimental runs has been conducted in an isothermal packed bed tubular reactor by investigating several operative
conditions, such as: temperature, contact time, methanol/oxygen feed molar ratio and water feed concentration. Depending on the operative
conditions adopted, the main products observed were formaldehyde and dimethoxymethane while lower amounts of methyl formate and CO2 were
also found. At low contact time, the main reaction product was dimethoxymethane which was then converted into formaldehyde through the
reverse equilibrium reaction with water. As a confirmation of this observation, a peculiar behaviour was detected consisting in an increase of
selectivity to formaldehyde by increasing methanol conversion. The obtained experimental data of methanol conversion and selectivity towards
products were modelled by means of an integral reactor model and the related kinetic parameters were determined by non-linear regression
analysis. The adopted reaction rate expressions were of the Mars van Krevelen–Langmuir Hinshelwood type and a good agreement was found
between the model theoretical prediction and the experimental data. A reaction mechanism and a detailed reaction scheme (rake-type) were
proposed for methanol ODH on a nano-structured catalyst that were able to interpret correctly the collected experimental observations.
# 2007 Elsevier B.V. All rights reserved.

Keywords: ODH methanol; Formaldehyde; Vanadium-based nanocatalysts

1. Introduction production is based [1]. In particular, the industrial processes

currently used can be divided in: (a) silver catalyst process,
Formaldehyde is one of the most versatile chemicals and is based on partial oxidation and dehydrogenation with air in the
employed by the chemical and other industries to produce a presence of silver crystals, steam and excess methanol at
large number of indispensable products used in daily life [1]. atmospheric pressure and 680–720 8C; (b) Formox process, in
The largest amounts of formaldehyde are mainly used to give a which a metal oxide (Fe-Mo) is used for the conversion of
wide range of condensates, i.e. urea-phenolic and melamine methanol to formaldehyde, by oxidation with excess air,
resins, and, to a small extent, their derivates. The properties of essentially at atmospheric pressure and 250–400 8C. In both
these compounds have been adapted by industry to a very large processes, a conversion of about 99 and 92% of selectivity to
number of applications, such as the manufacture of chipboards formaldehyde are obtained. However, the very low reaction
(compressed wood) and plywoods. Other well-established temperature used in the Formox process, which permits to
applications are in the production of papers, paints, adhesives, achieve high selectivity to formaldehyde, and the very simple
cosmetics, explosives, fertilizers and textiles. method of steam regeneration, which reduces the operating
The selective catalytic oxidation of methanol represents the costs, make this process easily controlled and more attractive
most relevant route on which the industrial formaldehyde from the industrial point of view. For this reason, the most
common actual plants (more than 70%) employ Formox
methanol oxidation to produce formaldehyde. However, iron-
* Corresponding author. Tel.: +39 081674027; fax: +39 081674026. molybdate catalysts become less active in the presence of an
E-mail address: [email protected] (E. Santacesaria). excess of methanol and therefore require a relatively high

0920-5861/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.cattod.2007.06.072
192 M. Cozzolino et al. / Catalysis Today 128 (2007) 191–200

to the surface structure of vanadium species and, generally, a

Nomenclature good surface vanadium dispersion is one of the key-factors to
obtain very active and selective catalysts. In particular, the
bj adsorption parameter of the component jth
results obtained in the ODH of ethanol to acetaldehyde,
(atm1)
investigated in a previous works [15,17], showed that supported
C methanol fractional conversion
vanadium catalysts may be able to promote this process under
Fj molar feed flow rate of component jth (mol/min)
mild conditions of both temperature and pressure, allowing to
ki reaction rate constant of the reaction ith (mol/
achieve high activities and selectivities, to prevent coke
min gcat atm)
formation and to improve the catalyst lifetime.
kox catalytic site re-oxidation constant (mol/min gca-
0.5 More recently, we focused our attention on the study of the
t atm )
catalytic performances of vanadium-based catalysts in the
Keq equilibrium constant for reaction 5.2
ODH of methanol to formaldehyde [18]. At this purpose, a
nc number of carbon atoms in the ith reaction
detailed catalytic screening over vanadium catalysts, supported
product
on both silica and titania/silica by grafting, was performed. The
Nr number of reactions
influence of reaction temperature (140–260 8C), at a fixed
Pj partial pressure of the component jth (atm)
contact time (W/F = 25.3 gcat h/molMeOH), and vanadium
ri reaction rate of the reaction ith (mol/min gcat)
loading, as well as preparation method and type of support,
v j;i stoichiometric coefficient of the component jth in
on the activity and selectivity was thoroughly investigated. For
the reaction ith
a useful comparison, some catalysts were also prepared by
W catalyst weight loaded in the reactor (g)
impregnation. The results obtained showed that the catalysts,
Yi fractional yield to product ith
prepared by grafting vanadyl alkoxide on TiO2/SiO2 support,
are more active than the impregnated ones with similar
oxygen partial pressure to remain active [2]. Multi-component vanadium loadings. Moreover, by comparing the catalytic
unsupported Ni–Fe–Bi–Ti–P–O catalysts [3] or catalyst performances of the catalysts synthesized by grafting (V2O5/
containing Cu, Zn, Ni, etc. have also been proposed [4]. TiO2/SiO2), we observed that the activity increases by
However, the stability and catalytic performances of these increasing the vanadium content, while the selectivity to
catalysts make them less favorable for an industrial application. formaldehyde only depends on the conversion level and seems
Thus, motivated by the perspective to look for catalysts with to be independent of the coverage degree of the starting silica
better physico-chemical properties, the ODH of methanol to support by titania. Thus, we concluded that the vanadium
formaldehyde has been widely studied in the literature [5–7]. loading and the surface dispersion are the key-factors which
Good catalytic performances in terms of both activity and determine the activity. For what concerns the product-
selectivity have also been documented for supported metal distribution, depending on the operative conditions adopted,
oxide catalysts, which consist of two-dimensional surface the main products observed were formaldehyde and dimethox-
metal oxide overlayers on high-surface support oxides (e.g. ymethane, as well as lower amounts of methyl formate and
Al2O3, TiO2, SiO2, etc.) [8]. Among all the catalysts explored, CO2. The presence of CO, verified by using a different
supported vanadium oxides have received much attention in chromatographic column (Carboxen 1006 PLOT), in the
recent years [9–11]. Their application is strongly affected by reactor outlet has never been observed. An expected increase
the specific interaction between vanadium oxide and the of methanol conversion resulted by increasing vanadium
support. Several authors have compared the properties of loading, from 1% up to 5% of V2O5 by weight. On the contrary,
vanadia supported on different carriers (e.g. SiO2, Al2O3, TiO2, an unusual trend consisting in an increase of selectivity to
ZrO2, MgO) and have concluded that, depending on the specific formaldehyde as a function of methanol conversion, was
support, vanadia based catalysts may show different catalytic observed. This trend may be explained by considering that the
activity and selectivity. Apart from the role of the support, the first reaction product was dimethoxymethane which is then
importance of the preparation method on the final catalyst converted to formaldehyde by the reverse equilibrium reaction
behaviour and uniformity of the vanadia species has been with water. This observation has been supported by the
largely discussed in literature. At this purpose, the catalyst experimental screening results, according to which, at low-
preparation by using the grafting technique has gained interest temperature and conversion, the main product was dimethox-
in the recent years as a route for obtaining well-dispersed ymethane, while, by increasing the temperature, an increase in
catalysts. It has been pointed out that grafting techniques lead to formaldehyde formation was detected. At temperature higher
more dispersed catalysts that are stable when an opportune than 220 8C, when methanol conversion is nearly complete,
support is used. In our previous works [12–16] we have used, formaldehyde oxidation occurs and an increase in CO2
for example, vanadyl alkoxide, dissolved in different aprotic production was observed. This scenario, provided by the
solvents, to undertake the grafting operations and we have screening, seems to confirm the structure-sensitivity of
tested their catalytic performances in different reactions, such methanol oxidation, largely discussed in literature, according
as: SCR of nitrogen oxides by ammonia and ODH of propane, to which different products can be obtained by varying the
n-butane, isobutane and ethanol. We have found that the catalyst nature and reaction conditions [19]. Thus, a good
different catalytic properties of vanadia catalysts can be related knowledge of the reaction mechanism is necessary to elaborate
 M. Cozzolino et al. / Catalysis Today 128 (2007) 191–200 193

Table 1
Operative conditions used for the preparation of the catalysts and some related properties
Precursor/solvent/support Acronym Ti(O-iPr)4 (g) Support (g) Volume of % wt of SBET Pore volume
Solvent (cm3) MOx (m2/g) (cm3/g)
Ti(O-Pri)4/dioxane/SiO2 TSm 10.79 24 400 7.29a 278 1.70
VO(O-Pri)3/dioxane/TiO2-SiO2 4.69Vgraf/TSm 0.65 3 50 4.69b 200 1.02
a
Percentage by weight of supported TiO2.
b
Percentage by weight of supported V2O5.

a correct interpretation of variations of the catalytic behaviour. 500 8C for 2 h. More details about the preparation method and
In spite of the large number of papers devoted to investigate the the properties of this support are reported elsewhere [18]. The
influence of the surface properties and the interaction between vanadium-based catalyst, indicated with the acronym 4.69Vgraf/
the active phase and the support on the catalyst behavior, very TSm, was prepared by putting the described support in contact
few works reported a detailed kinetics model able to describe with a solution of vanadyl tri-isopropoxide (VO[O-Pri]3,
the products distribution. According to the literature, the Aldrich 99.999%, d = 0.963 g/ml), dissolved in anhydrous
selective oxidation of methanol to formaldehyde over metal dioxane. The grafting reaction was performed for 5 h in a well-
oxide based catalysts follows a Mars and van Krevelen [20] stirred jacketed glass reactor, under inert helium atmosphere.
type redox mechanism, in which the oxidation proceeds by a The solid obtained was filtered, washed with dioxane, oven-
reduction of an oxidized surface site, which is subsequently re- dried at 120 8C overnight, heated at 200 8C and then calcined at
oxidized with gas-phase molecular oxygen [21]. Several 500 8C. The operative conditions used for preparing the support
authors have studied the kinetics of the methanol partial TSm and the vanadium catalyst 4.69Vgraf/TSm, respectively,
oxidation to formaldehyde over a Fe-Mo catalyst and only a are reported in Table 1 together with the resulted physico-
limited number of papers were addressed towards vanadium chemical properties of the solids obtained.
oxide catalysts. However, most of the previous works was
limited to the formation of formaldehyde only and not much 2.2. Catalyst and support characterization techniques
attention was paid to the formation of side products, such as
carbon dioxide, dimethylether (DME) and dimethoxymethane The titanium loading was determined using a colorimetric
(DMM). Moreover, all the kinetic data reported in the literature method [23], after dissolution of titanium in a concentrated
were measured at low methanol concentrations (maximum 6%) sulfuric acid solution, dilution, and then treatment with H2O2.
[22]. The vanadium load in the prepared catalyst was determined by
Thus, the lack of detailed kinetic data capable of describing atomic absorption, after dissolution in concentrated sulfuric
all the products observed in the ODH of methanol, over acid.
supported vanadium-based catalysts, led us to investigate this Textural analyses were carried out by using a Thermoquest
reaction also from a kinetic point of view. In this work, we wish Sorptomatic 1990 Instrument (Fisons Instrument) and by
to propose a complete kinetic analysis on a supported determining the nitrogen adsorption/desorption isotherms at
vanadium-based catalyst (4.69Vgraf/TSm), resulted from the 77 K. The samples were thermally pre-treated under vacuum
catalytic screening as the best one in terms of both activity and overnight up to 200 8C (heating rate = 1 8C/min). Specific
selectivity. The effect of several operative variables such as: surface area (SBET) and pores size distribution were determined
temperature, contact time, methanol/oxygen feed molar ratio by using the BET and Dollimore–Heal methods [24,25]. Details
and water feed concentration was investigated. about other characterization analyses (XRD, XPS and NH3-
TPD), performed on both the support TSm ad the catalyst
2. Experimental 4.69Vgraf/TSm, as well as the related results are reported
elsewhere [18]. This catalyst has been selected for the kinetic
2.1. Support and catalyst preparation study, because, corresponds to a vanadium load near to the
monolayer and it has shown the highest TON with respect to
The support of silica coated with TiO2, indicated with the other catalysts containing different amounts of vanadium oxide
acronym TSm, was prepared in a single-step grafting on the surface [18]. The activities shown for increasing amounts
procedure, using titanium tetra-isopropoxide (Fluka) dissolved of vanadium on the surface increases almost linearly until to
in anhydrous dioxane and a commercial silica (Grace S-432), reach a maximum value in correspondence of 4.8–5.4% of
calcined at 500 8C for 8 h. The quantity of titanium tetra- V2O5 [18]. The support not containing vanadium has shown
isopropoxide dissolved in dioxane corresponds to the mono- negligible activity in the reaction.
layer coverage, assuming a conventional stoichiometry of one
hydroxyl per alkoxide molecule [16]. After calcination, the 2.3. Methods, techniques and operating conditions used in
silica was put in contact with the titanium alkoxide solution for the catalytic runs
5h at room temperature under constant stirring. The solid
obtained was then filtered, washed with dioxane, dried at Kinetic runs were performed in a stainless steel tubular
120 8C overnight, heated at 200 8C for 2 h and then calcined at reactor with an internal diameter of 1 cm, kept isothermal with
194 M. Cozzolino et al. / Catalysis Today 128 (2007) 191–200

a fluidized bed of sand. Liquid methanol was fed, by a other reactions, as the n-butane oxidation to maleic
syringe pump, into a vaporizer chamber kept at 250 8C and anhydride, the gaseous oxygen seems to be responsible for
was then sent, after the addition of a stream of oxygen and the oxidation process [27]. However, all these reactions occur
helium, into a stainless steel coil kept at the same temperature at a relatively higher temperature if compared with the one
of the reactor. The composition of the gases at the outlet of used in the present study. Very probably dehydrogenation is,
the reactor was gas-cromathographically analyzed by with- in our case, the first reaction step which involves hydride
drawing a sample with an on-line sampling valve kept at formation while the gaseous oxygen is responsible for
170 8C. The GC used was an HP 5890 instrument, with a catalytic site re-oxidation. Another uncertainty regards the
Restek RT-Q-Plot 30 m  0.32 mm column. Helium was used structure of the active site that, in the case of vanadium-based
as the carrier gas. The conditions used for the analyses were as catalysts, depends also on the vanadium load [28], on the
follows: temperature held at 50 8C for 2 min, increased at a hydration condition of the environment and on the used
rate of 5 8C/min to 100 8C and then at a rate of 20 8C/min to support. Despite the relatively high reaction temperature, the
180 8C for 5 min, and finally kept at this temperature for water produced could give place to the presence of a
5 min. A TCD detector kept at 250 8C was used. Samples of vanadium catalytic site in an hydrated form, characterized by
powdered catalyst, generally 0.3 g, were placed inside the a structure that is intermediate between the following
reactor on a bed of glass wool. Two thermocouples located ones:
immediately upon and under the catalytic bed allowed the
validity of the isothermal conditions to be controlled with-
in 1 8C. The operative conditions adopted are listed in
Table 2. (1)
Three different sets of kinetics runs were conducted. The
first set of runs was performed in the temperature range 140–
200 8C by changing the methanol residence time from 10 to
65 gcat h/molMeOH but keeping the molar ratio between Weckhuysen and Keller [21] have proposed a reaction
methanol and oxygen constant (1:1). The second set of runs mechanism for the partial oxidation of methanol to formalde-
was carried out at T = 160 8C by significantly varying both the hyde, on vanadium-based catalysts, that involves a single
methanol residence time (from 10 to 65 gcat h/molMeOH) and vanadium surface site for which the vanadium oxidation state is
the feed molar ratio between methanol and oxygen. The third nearly constant (V) and a limited reduction was observed.
set of kinetics experiments was conducted with the aim to According to this mechanism, the conversion of methanol to
investigate the effect of the presence of water in the feed on the formaldehyde occurs in four steps that can be schematically
catalytic performances. These last runs were carried out by represented as follows:
using the following reaction conditions: 0.3 g of catalyst,
methanol residence time ranging from 10 to 65 gcat h/molMeOH
while keeping at 180 8C the reaction temperature. Water was
added in the feed with an apparatus similar to that of methanol
(syringe pump).
Results are reported in terms of the methanol conversion and
product yields. The methanol conversion is defined as
(2)
Number of moles of methanol reacted
C¼
Number of moles of methanol fed
while the yields of the ith product is defined as
Number of moles of product i formed
Y i ¼ nc
Number of moles of methanol fed
where nc is the number of carbon atoms contained in the
considered ith molecule. As can be seen from (2), the methoxydic intermediate
V–O–CH3, observed in literature through spectroscopy
3. Results and discussion experiments [29], is the precursor to which the formaldehyde
formation (assumed as the rate determining step) can be
The detailed mechanism of alcohols partial oxidation to attributed while the molecular oxygen is responsible for the
aldehydes is still under investigation. It is well known that the catalytic site re-oxidation which is then restored by water
oxygen responsible of the oxidation is present in the lattice elimination.
structure. This is supported by the experimental evidence that It is interesting to observe that the proposed mechanism (2)
oxygenated compounds are formed onto catalyst surface also accounts only for the main reaction to formaldehyde, while in
in the absence of gaseous oxygen, as in the case of acrolein our investigation significant amounts of other products have
synthesis from propylene on mixed Mo/Bi oxides [26]; in been observed, depending on the operative conditions adopted.
 M. Cozzolino et al. / Catalysis Today 128 (2007) 191–200 195

Table 2
List of runs and related operative conditions
Run ID W/F Methanol feed flow O2 feed flow rate Water feed flow Temperature
(gcat h/molMeOH) rate (liquid) (ml/h) (gas) (ml/min) rate (liquid) (ml/h) (8C)
R1-R5 12.91 1.00 10.07 – 140–200
R6-R10 16.13 0.8 8.06 – 140–200
R11-R15 32.37 0.4 4.06 – 140–200
R16-R20 61.46 0.21 2.09 – 140–200
R21 12.91 1.00 5.03 – 160
R22 16.13 0.8 4.04 – 160
R23 32.37 0.4 2.03 – 160
R24 61.46 0.21 1.05 – 160
R25 12.91 1.00 15.1 – 160
R26 16.13 0.8 12.11 – 160
R27 32.37 0.4 6.09 – 160
R28 61.46 0.21 3.14 – 160
R29 12.91 1.00 10.07 0.08 180
R30 16.13 0.8 8.06 0.065 180
R31 32.37 0.4 4.06 0.04 180
R32 61.46 0.21 2.09 0.02 180
Other conditions: catalyst load W = 0.32 g; helium flow rate = 30 ml/min.

Table 3
Experimental data
Run ID W/Fa Temperature Feed molar Methanol Products yieldsb
(8C) ratio CH3OH/O2 conversion
F MF DMM CO2
R1 12.91 140 1 0.19034 0.01006 0.00750 0.15517 0.01760
R6 16.13 140 1 0.20076 0.01193 0.01071 0.15932 0.01879
R11 32.27 140 1 0.23034 0.04762 0.01428 0.13506 0.03339
R16 61.46 140 1 0.25428 0.14727 0.01535 0.03495 0.05671
R2 12.91 160 1 0.22961 0.01674 0.01506 0.18860 0.00921
R7 16.13 160 1 0.27392 0.03294 0.02120 0.20586 0.01391
R12 32.27 160 1 0.31864 0.09665 0.02294 0.16824 0.03079
R17 61.46 160 1 0.38799 0.22947 0.11163 – 0.04688
R3 12.91 180 1 0.26222 0.10816 0.02604 0.11275 0.01742
R8 16.13 180 1 0.33117 0.16248 0.03415 0.11728 0.01726
R13 32.27 180 1 0.35806 0.22538 0.07302 0.02467 0.03499
R18 61.46 180 1 0.57301 0.43106 0.08626 – 0.05569
R4 12.91 190 1 0.38276 0.21186 0.04694 0.11200 0.01196
R9 16.13 190 1 0.41255 0.22471 0.04972 0.12469 0.01343
R14 32.27 190 1 0.49893 0.38390 0.07377 0.01696 0.02429
R19 61.46 190 1 0.82799 0.62684 0.09231 0.00990 0.09894
R5 12.91 200 1 0.37585 0.24133 0.04420 0.07531 0.01501
R10 16.13 200 1 0.48011 0.34829 0.06027 0.04140 0.03016
R15 32.27 200 1 0.68976 0.56172 0.07448 0.00821 0.04535
R20 61.46 200 1 0.91946 0.72141 0.09027 – 0.10778
R21 12.91 160 0.66 0.28527 0.05027 0.01221 0.19741 0.02538
R22 16.13 160 0.66 0.30141 0.06976 0.01529 0.18603 0.03603
R23 32.27 160 0.66 0.32651 0.09424 0.01896 0.17373 0.03958
R24 61.46 160 0.66 0.45710 0.29185 0.03756 0.03135 0.09634
R25 12.91 160 2 0.19429 0.01721 0.00619 0.15924 0.01165
R26 16.13 160 2 0.26533 0.03288 0.01179 0.20050 0.02015
R27 32.27 160 2 0.30027 0.07750 0.01578 0.15293 0.05405
R28 61.46 160 2 0.31312 0.13749 0.00689 0.02220 0.14651
R29 12.91 180 1 0.31477 0.12573 0.02543 0.13139 0.03222
R30 16.13 180 1 0.31684 0.12695 0.02514 0.13566 0.02909
R31 32.27 180 1 0.40715 0.25941 0.04082 0.06218 0.04474
R32 61.46 180 1 0.67661 0.50895 0.06176 0.00828 0.09761
a
(gcat h/molMeOH).
b
F: formaldehyde, MF: methyl formate, DMM: dimethoxymethane.
196 M. Cozzolino et al. / Catalysis Today 128 (2007) 191–200

In particular, at low contact time, the main reaction product presents a vanadium-hydride bond. The dioxymethylenic
was dimethoxymethane while, increasing the contact time, an group can further react with methanol adsorbed from gaseous
increased concentration of formaldehyde was observed (see for phase and lead to the formation of dimethoxymethane in two
example the run at 160 8C reported in Fig. 2). This behavior is successive steps. Formaldehyde is then formed for the
the classical one in kinetics for the consecutive reactions and occurrence of the hydrolysis equilibrium (F) when the
suggests that the formation of dimethoxymethane and methanol conversion is high and water is present at a
formaldehyde occurs in series, that is, dimethoxymethane is sufficiently high concentration in the system. According to
converted into formaldehyde by the reverse equilibrium Wang and Wachs [36] and Tatibouët the formation of
reaction with water. As a further confirmation of this dimethoxymethane is favored by the presence on the surface
observation, an increase in selectivity to formaldehyde was of both redox and acid sites and by the low-temperature. Silica
generally observed in correspondence to an increase of coated by grafting titanium alkoxide contains a great amount of
methanol conversion (see Table 3). The described behavior is Lewis acid sites of medium strength as recently shown by
different from the one observed in the ODH of ethanol, Bonelli et al. [37] and this could explain the obtained results.
performed in similar conditions with the same type of catalyst The catalytic site in the initial form (A) is finally restored from
[17], suggesting a different operating reaction mechanism in the reduced site (E) by the intervention of gaseous oxygen and
the two cases. by water elimination, closing the cycle. A mechanism
However, the preminent formation of dimethoxy methane at substantially similar to (3) could be proposed also if we
low methanol conversions has already been observed also by consider that there the initial vanadium site is in the dehydrated
other authors, in particular by Tatibouët and Germain [30], as form.
reported in the review of Forzatti et al. [31]. According to the proposed mechanism, the presence of
On the basis of the observations reported, we hypothesized a water is not detrimental for the formaldehyde production,
mechanism for the catalytic cycle of the ODH of methanol that because, it favorably promote the equilibrium reaction (F), as it
starts from a vanadium site in hydrated form, (A), and that can has been confirmed by introducing small amounts of water (see
be schematized as follows: Table 2) in the reactor feed obtaining a slight increase in

(3)

The first elementary step in this mechanism, that is, the formaldehyde yield (compare runs R3-8-13-18 with R29-30-
formation of the methoxydic intermediate (B) by methanol 31-32 of Table 3). A further experimental observation is that
dissociative adsorption, is in agreement with the suggestions of CO2 formation appears significant only at high methanol
different authors [21,31,32], in particular by Wachs et al. that conversion and this suggests that methanol is not directly
devoted different papers to this reaction performed in the involved in the total oxidation. No presence of CO traces has
presence of vanadium oxide supported on different supports been detected in our experimental conditions. On the basis of
[33–36]. In the subsequent step this intermediate is rearranged the experimentally observed compounds and of their evolution
into a dioxymethylenic surface species, spectroscopically with contact time, the following rake-type [30] reaction scheme
observed by Busca [38], with the vanadium atom which can be proposed:
 M. Cozzolino et al. / Catalysis Today 128 (2007) 191–200 197

(4)

At higher temperatures, very probably, COx could also be ment, the COx has been identified with CO2 because the
formed from formaldehyde and a contribution of this route to presence of CO has not been detected.
the reaction scheme (4) would be considered although we think On the basis of this approximation, only five kinetic
that oxidation reactions would follow the progressive steps parameters are necessary to completely describe an isothermal
methanol-aldheyde-acid-COx, for this reason we considered run. By adopting the standard form of material balance
adsorbed formic acid the main source of CO2. equation, related to a tubular packed bed reactor (plug flow,
According to the scheme (4), the adsorbed forms of absence of diffusion resistances), for each component j present
methanol, formaldehyde and formic acid give place, through in the gas-phase:
desorptive reactions, to all the observed products. In
particular, methyl formate and dimethoxymethane are formed dF J X Nr

by reaction between an adsorbed oxidized species and  ¼ nJ;i ðr i Þ (6)

dW i¼1
unreacted methanol from gas-phase. In this way, the
formation of both the products mentioned is predominant where F j is the molar flow rate of the component j, W is the total
at low conversion when methanol partial pressure is still amount of catalyst loaded in the reactor, v j;i is the stoichio-
relatively high. On the contrary, when methanol is almost metric coefficient of the component j in the reaction i and ri is
completely converted, the expected predominant products are the ith reaction rate. In order to integrate the ordinary differ-
formaldehyde, methyl formate (from Tischencko pathway ential equations system (6) we have to introduce the expression
between surface adsorbed methanol and gaseous formalde- for the reaction rates ri. By assuming for the redox reactions
hyde) and carbon dioxide. occurring on the surface a Mars and van Krevelen mechanism
A comprehensive interpretation of the reaction scheme (4) in [20], corrected, as suggested by Carrà and Forzatti [39], for the
terms of elementary steps would involve too many parameters depressive effect that could be exerted by water and methanol,
of difficult experimental evaluation. Thus, we adopted a both adsorbing on the redox catalytic sites, we can write:
simplified overall reaction scheme that takes into account all
the products observed, represented as follows: A þ Site-OX ! Product þ Site-RED
k1
ð5:1Þ 3CH3 OH þ 1=2O2 !CH2 ðOCH3 Þ2 þ H2 O Site-RED þ 1=2O2 ! Site-OX
k2
ð5:2Þ CH2 ðOCH3 Þ2 þ H2 O !2CH3 OH þ HCHO
k2 By assuming steady state conditions and considering that
k3
ð5:3Þ CH3 OH þ 3=2O2 !CO2 þ 2H2 O uox + ured + uocc = 1 the following general expression can be
k4 obtained:
ð5:4Þ CH3 OH þ HCHO þ 1=2O2 !HCOOCH3 þ H2 O
(5) k i Pi
ri ¼  ð1  uocc Þ
It must be pointed out that, even if the tendencies of 1 þ ðki PÞ=ðkox P0:5
O2 Þ
formaldehyde to form COx is higher than that of methanol, in
the reaction scheme (5) the route for CO2 formation starts from By applying the general expression to each redox reaction,
methanol. This approximation has been introduced by by assuming the surface oxidation as rate determining step and
considering that in the detailed reaction scheme (4) the actual a depressive effect due to the adsorption of water and methanol
CO2 formation is due to formic acid in an adsorbed form which, on the redox sites we obtain:
in turn, comes from a formaldehyde surface specie that,
k 1 PM 1
ultimately, derives from methanol. This complex route has been r1 ¼  (7)
lumped into the reaction (5.3). Furthermore, in our develop- 1 þ ðk1 PM Þ=ðkox P0:5
O2 Þ 1 þ b M M þ bW PW
P
198 M. Cozzolino et al. / Catalysis Today 128 (2007) 191–200

k 3 PM 1
r3 ¼ 0:5
 (8)
1 þ ðk3 PM Þ=ðkox PO2 Þ 1 þ bM PM þ bW PW

k 4 PM PF 1
r4 ¼  (9)
1 þ ðk4 PM Þ=ðkox P0:5
O2 Þ 1 þ b M P M þ b W PW

The equilibrium reaction of DMM hydrolysis has been

considered as a first approach occurring with a kinetic law of
the type:
k2 2
r 2 ¼ k 2 PA PW  P PF (10)
K eq M
derived from the assumption of a two steps equilibrium invol-
ving also hemiacetal as intermediate.In the relations (7–10), P
are the partial pressures, k are the kinetic parameters and b are
the adsorption parameters. As mentioned, r1, r3 and r4 are a
combination of the Mars–van Krevelen and Langmuir Hinshel- Fig. 2. Agreement between model behavior and experimental data of methanol
wood models (MVK-LH), but assuming that the re-oxidation of conversion and products distribution as a function of methanol contact time W/
F. Run R2-R7-R12-R17 at 160 8C. Legend: F – formaldehyde, MF – methyl
the catalytic sites occurs, in the three cases, with the same rate formate, DMM – dimethoxymethane.
(one only parameter) and that only methanol and water compete
for the adsorption on the redox catalytic sites. Obviously, other
kinetic models could fit well the experimental runs but the Table 4
Kinetic parameters
adopted one is compatible with the postulated reaction scheme
and mechanism. Constant ln(A) or ln(b0) EA or DH (kcal/mol)
The kinetic parameters in the expressions (7–10) have been k1 23.9  2.6 20.4  2.3
determined by non-linear regression for the fitting of all the k2 20.6  1.1 11.1  1.0
experimental data and have been correlated with temperature k3 15.6  4.8 14.8  4.2
by means of the Arrhenius equation. k4 39.0  3.2 27.8  2.8
kox 24.3  3.8 19.0  3.4
In Fig. 1, the results of conversion and yield to formaldehyde bM 28.8  4.9 28.9  4.3
are reported as a function of contact time for all the temperature bW 23.2  8.9 25.3  7.9
range investigated, in comparison with model prediction. In
ki = Aexp(EA/RT) i = 1, 2, 3, 4, OX. bj = b0exp(DH/RT) j = M, W.
Fig. 2 is reported, as example, the agreement between the Keq = exp(17.5  7548/T) experimental. Keq = exp(13.7  6090/T) from ther-
experimental products distribution, obtained for the set of runs modynamic calculations.
R2-7-12-17, and the kinetic model. As an overall model
agreement with respect to the experimental data, in Fig. 3 the
parity plots are reported for both methanol conversion and if not sufficient, and other models could have similar
formaldehyde yield. As can be seen, the adopted model, even if performances in describing the experimental observations.
simplified, resulted in a satisfactory performance in the The values of the kinetic and adsorption parameters, employed
description of the experimental observations collected. The in the model, are reported in Table 4 while in Fig. 4 the main
agreement between the experimental data and the adopted kinetic parameters are plotted against the reciprocal of the
simplified model represents an element supporting the validity of absolute temperature in the Arrhenius plot. From this plot, an
the proposed reaction scheme. This is a necessary condition, even evaluation of activation energies for formaldehyde and

Fig. 1. Kinetic results at various reaction temperatures: comparison between kinetic model (continuous lines) and experimental data for (A) methanol conversion as a
function of methanol contact time W/F and (B) formaldehyde yield as a function of methanol contact time W/F.
 M. Cozzolino et al. / Catalysis Today 128 (2007) 191–200 199

Fig. 3. Parity plots for (A) methanol conversion and (B) formaldehyde yield related to all the data collected. The absolute deviations of 10% between model and
experimental data (dotted lines), are also reported.

dimethoxymethane formation was possible and values of 20.4 to the increased efficiency in the catalytic site re-oxidation.
and 11.1 kcal/mole, respectively, were obtained. These values This phenomenon has also been observed by other authors [30],
are in agreement with those reported in the literature [30] in the according to which iron-molybdate catalysts seem to become
range 10–20 kcal/mole for methanol ODH on supported and less active in the presence of an excess of methanol.
unsupported oxide catalysts. The equilibrium constant in Eq. (8), At last, it is interesting to point out that the adopted kinetic
whose value is reported in Table 4, is in rather good agreement model well reproduce also some apparently strange results
with the value obtained from thermodynamic calculations. appearing in Table 3 such as the very small increase of
The values of the adsorption parameters for both methanol methanol conversion (from 19 to 25%) for a large change in the
and water seem to indicate that adsorption effect is rather contact time (from 12.91 to 61.4) corresponding to a very strong
strong, although no discrimination between methanol and water change in the selectivities passing from a large preminence of
was possible because both these components showed a similar DMM to the one of formaldehyde. This behaviour can be
behavior and no one can be neglected with respect to the other. explained considering a very low rate of DMM hydrolysis at
As mentioned above, the effect of the presence of water in low-temperature.
the feed has no negative influence on catalytic activity, despite
it seems to adsorb on the catalyst surface in competition with 4. Conclusions
methanol. This behavior supports the proposed mechanism and
reactions scheme in which water, at least in the explored The kinetics of the ODH of methanol to formaldehyde on
concentration range, promotes the equilibrium reaction (5.2). vanadium-based grafting catalysts has been extensively
The effect of methanol/oxygen feed molar ratio was also investigated. The effect of various operative conditions
investigated and a general increase of conversion and yield (temperature, contact time, feed composition) has been studied
were observed by decreasing this ratio, in particular at high and a detailed kinetic model has been developed to interpret the
contact time. In other words, the oxygen partial pressure seems collected integral data of a tubular packed bed reactor. A
to have a positive effect on methanol conversion, probably due simplified kinetic model was formulated on the basis of a
reaction mechanism that led us to a rake-type reaction scheme
in which all the observed products derive from adsorbed species
of methanol, formaldehyde and formic acid. The agreement
between the model and the experimental data, in the range of
operating conditions adopted, can be considered satisfactory
and worth to be deepened in the perspective of an industrial
development.

Acknowledgement

Thanks are due to MIUR-PRIN-2005038244 for the

financial support.

References

[1] G. Reuss, W. Disteldorf, O. Grundler, A. Hilt, Formaldehyde, Fifth ed.,

vol. A11, Ullmann’s Enc. of Industrial Chemistry, p. 619.
Fig. 4. Arrhenius plot for the kinetic constants k1 (dimethoxymethane forma- [2] A.R. Chauvel, Ph.R. Country, R. Maux, Cl. Petilpas, Hydrocarbons
tion), k2 (formaldehyde formation) and kox (catalytic site re-oxidation). Process. (1973) 179.
200 M. Cozzolino et al. / Catalysis Today 128 (2007) 191–200

[3] D. Klissurski, Y. Pesheva, N. Abadjieva, D. Filiukova, L. Petrov, Appl. [19] J.M. Tatibouët, Appl. Catal. A: Gen. 148 (1997) 213.
Catal. 77 (1991) 55. [20] P. Mars, D.W. van Krevelen, Chem. Eng. Sci. (Spec. Suppl.) 3 (1954) 41.
[4] Ph. Zaza, Ph.D. Thesis, Departement de Chimie, Ecole Polytechnique [21] B.M. Weckhuysen, D.E. Keller, Catal. Today 78 (2003) 25.
Federale de Lausanne, Lausanne, Switzerland, 1993. [22] S.A.R.K. Deshmukh, et al. Appl. Catal. A: Gen. 289 (2005) 240.
[5] G. Deo, I.E. Wachs, J. Haber, Crit. Rev. Surf. Chem. 4 (1994) 141. [23] F.R.D. Snell, L.S. Ettre, Enc. of Ind. Chem. Anal., vol. 19, Interscience,
[6] G. Deo, I.E. Wachs, J. Catal. 146 (1995) 323. New York, 1974, p. 107.
[7] G. Deo, I.E. Wachs, J. Catal. 146 (1995) 335. [24] S. Brunaber, P.H. Emmet, J. Am. Chem. Soc. 60 (1938) 309.
[8] E. Wachs, G. Deo, M.A. Vuurman, H. Hu, D.S. Kim, J.M. Jehng, J. Mol. [25] D. Dollimore, G.R. Heal, J. Appl. Chem. 14 (1964) 109.
Catal. 82 (1993) 443. [26] B. Schiøtt, K.A. Jørgensen, J.Phys. Chem. 95 (1991) 2297.
[9] G. Bond, S. Tahir, Appl. Catal. A 71 (1991) 1. [27] G.C. Bond, S.F. Tahir, Appl. Catal. 71 (1991) 1.
[10] I. Wachs, B. Weckhuysen, Appl. Catal. A: Gen. 157 (1997) 67. [28] G.C. Bond, J. Catal. 53 (1989) 116.
[11] F. Arena, F. Fusteri, A. Parmaliana, Appl. Catal. A: Gen. 176 (1999) 189. [29] B.M. Weckhuysen, I.E. Wachs, in: H.S. Nalwa (Ed.), Handbook of
[12] A. Comite, A. Sorrentino, G. Capanelli, M. Di Serio, R. Tesser, E. Surfaces and Interfaces of Materials, vol. 1, Academic Press, San Diego,
Santacesaria, J. Mol. Catal. A: Chem. 198 (2003) 151. 2001, p. 613.
[13] R. Monaci, E. Rombi, V. Solinas, A. Sorrentino, E. Santacesaria, G. Colon, [30] J.M. Tatibouët, J.E. Germain, C.R. Acad. Sci., Paris 289 (II) (1979) 305.
Appl. Catal. A: Gen. 214 (2001) 203. [31] P. Forzatti, E. Tronconi, A.S. Elmi, G. Busca, Appl. Catal. A: Gen. 157
[14] V. Iannazzo, G. Neri, S. Galvagno, M. Di Serio, R. Tesser, E. Santacesaria, (1997) 387.
Appl. Catal. A: Gen. 246 (2003) 49. [32] N. Pernicone, F. Lazzerin, G. Liberti, G. Lanzavecchia, J. Catal. 10 (1968)
[15] E. Santacesaria, A. Sorrentino, R. Tesser, M. Di Serio, A. Ruggiero, J. 83.
Mol. Catal. A: Chem. 204–205 (2003) 617. [33] X. Gao, I.E. Wachs, J. Catal. 192 (2000) 18–28.
[16] E. Santacesaria, M. Cozzolino, M. Di Serio, A.M. Venezia, R. Tesser, [34] L.J. Burcham, L.E. Briand, I.E. Wachs, Langmuir 17 (2001) 6164–6174.
Appl. Catal. A: Gen. 270 (2004) 177. [35] X. Gao, I.E. Wachs, Topics Catal. 18 (2002) 243–250.
[17] R. Tesser, V. Maradei, M. Di Serio, E. Santacesaria, Ind. Eng. Chem. Res. [36] X. Wang, I.E. Wachs, Catal. Today 96 (2004) 211–222.
43 (2004) 1623. [37] B. Bonelli, M. Cozzolino, R. Tesser, M. Di Serio, M. Piumetti, E. Garrone,
[18] M. Cozzolino, R. Tesser, M. Di Serio, E. Gaigneaux, P. Eloy, E. Santa- E. Santacesaria, J. Catal. 246 (2007) 293–300.
cesaria, in: E. Gaigneaux, et al. (Eds.), Studies in Surface Science and [38] G. Busca, J. Mol. Catal. 50 (1989) 241.
Catalysis, 162 (2006) 697. [39] S. Carrà, P. Forzatti, Catal. Rev. 15 (1977) 1.