The balance of acid, basic and redox sites in Mg/Me-mixed oxides: The effect on catalytic performance in the gas-phase alkylation of m-cresol with methanol

Article abstract of DOI:10.1016/j.jcat.2009.12.011

The reactivity of Mg/Me-mixed oxides (Me = Fe, Cr, Al) in m-cresol gas-phase methylation with methanol was studied, with the aim of finding relationships between catalytic performance and surface/bulk features. All systems exhibit similar surface properties (basically due to the presence of basic and Lewis-type acid sites), but the presence of extra cations in the MgO framework affects greatly the products obtained in the mentioned catalytic reaction, depending on the possible co-presence of redox capacity. In particular, (i) the Mg/Al/O system, characterized by no redox capacity, shows catalytic behaviour quite similar to that of conventional Bronsted-type acid catalysts, though with the advantage of exhibiting almost no deactivation during use (a rather common problem for acid catalysts activating methanol for phenol attack); (ii) the Mg/Cr/O catalyst shows a reactivity typical of a basic catalyst, with high chemo- and regio-selectivity and the favoured formation of 2,5-dimethylphenol; (iii) the Mg/Fe/O system presents a catalytic behaviour closer to that of Mg/Cr/O, if referred to the distribution of phenolic products, but it also exhibits a better activity in methanol dehydrogenation to formaldehyde (that is the true electrophilic reactant for m-cresol ring methylation), and in the decomposition of formaldehyde into light compounds. This behaviour causes a poor selectivity to dimethylphenols with respect to converted methanol, whereas the selectivity with respect to converted m-cresol is not different from that shown by Mg/Cr/O system.

Full text of DOI:10.1016/j.jcat.2009.12.011

Journal of Catalysis 270 (2010) 125–135
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
The balance of acid, basic and redox sites in Mg/Me-mixed oxides: The effect
on catalytic performance in the gas-phase alkylation of m-cresol with methanol
V. Crocellà a, G. Cerrato a, G. Magnacca a, C. Morterra a, F. Cavani b,1, , S. Cocchi b,1, S. Passeri b,1
,
*
D. Scagliarini b,1, C. Flego c, C. Perego d
a Dept. of Chemistry IFM and NIS Centre of Excellence, University of Turin, Consortium INSTM (Florence), Research Unit of Turin University, Via P. Giuria 7, 10125 Torino, Italy
b Dipartimento di Chimica Industriale e dei Materiali, ALMA MATER STUDIORUM Università di Bologna, Viale Risorgimento 4, 40136 Bologna, Italy
c eni SpA, Div. R&M, CR SDM, Via Maritano 26, S. Donato Milanese, MI, Italy
d eni SpA, CR Energie Non Convenzionali, Istituto Eni Donegani, Via Fauser 4, Novara, NO, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 30 July 2009
Revised 9 December 2009
Accepted 13 December 2009
Available online 21 January 2010
The reactivity of Mg/Me-mixed oxides (Me = Fe, Cr, Al) in m-cresol gas-phase methylation with methanol
was studied, with the aim of finding relationships between catalytic performance and surface/bulk fea-
tures. All systems exhibit similar surface properties (basically due to the presence of basic and Lewis-type
acid sites), but the presence of extra cations in the MgO framework affects greatly the products obtained
in the mentioned catalytic reaction, depending on the possible co-presence of redox capacity.
In particular, (i) the Mg/Al/O system, characterized by no redox capacity, shows catalytic behaviour
quite similar to that of conventional Brønsted-type acid catalysts, though with the advantage of exhibit-
ing almost no deactivation during use (a rather common problem for acid catalysts activating methanol for
phenol attack); (ii) the Mg/Cr/O catalyst shows a reactivity typical of a basic catalyst, with high chemo-
and regio-selectivity and the favoured formation of 2,5-dimethylphenol; (iii) the Mg/Fe/O system pre-
sents a catalytic behaviour closer to that of Mg/Cr/O, if referred to the distribution of phenolic products,
but it also exhibits a better activity in methanol dehydrogenation to formaldehyde (that is the true elec-
trophilic reactant for m-cresol ring methylation), and in the decomposition of formaldehyde into light
compounds. This behaviour causes a poor selectivity to dimethylphenols with respect to converted meth-
anol, whereas the selectivity with respect to converted m-cresol is not different from that shown by Mg/
Cr/O system.
Keywords:
Basic catalysis
Bifunctional catalysis
m-Cresol methylation
Methanol decomposition
Mg/Al/O
Mg/Cr/O
Mg/Fe/O
Magnesium oxide
Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction
The C-alkylation of phenol derivatives is carried out either in
the liquid-phase, using alkyl halides as alkylating agents, or in
The methylation of phenol and phenol derivatives has a high
industrial relevance. For instance, 2,6-xylenol is the monomer for
the production of poly-(2,6-dimethyl)phenylene oxide resins, 2-
methylphenol (o-cresol) is the monomer for the synthesis of
epoxycresol paints, 2,5-dimethylphenol is the intermediate for
the synthesis of dyes, antiseptics and antioxidants, and 2,3,6-tri-
methylphenol is the starting compound for the synthesis of vita-
min E [1]. The products of phenol or anisole and diphenols
methylation reaction are intermediates in the production of skin-
protecting agents and food additives. Starting from m-cresol, and
using 2-propanol as the alkylating agent, thymol is produced,
which is a commercially interesting intermediate for the synthesis
of menthol.
the gas-phase, with alcohols over acid or basic catalysts. The latter
may be (supported) alkali and alkaline-earth metal oxides, (mixed)
transition metal oxides, or combinations of both [2–19].
Previous reactivity experiments, carried out in liquid-phase in
mild conditions, showed that this reaction may provide useful
information about chemo- and regio-selectivity; they are driven
by catalyst properties [35–39]. However, these kinds of tests
achieved very low m-cresol conversion, so limiting the possibility
of examining catalyst deactivation and the role of methanol
decomposition. And in this respect, we decided to study the gas-
phase methylation of m-cresol.
When methanol is the co-reactant, the ring methylation of phe-
nolic compounds is carried out in the gas-phase, typically at tem-
peratures higher than 300 °C, and one of the major problems of the
industrial process is the decomposition of methanol occurring in
these conditions. Consequently, a large excess of methanol is usu-
ally fed in order to reach an acceptable per-pass conversion of
phenol.
* Corresponding author. Fax: +39 051 209 3680.
E-mail address: fabrizio.cavani@unibo.it (F. Cavani).
1
Also: Consortium INSTM (Florence), Research Unit of Bologna (a partner of NoE
Idecat, FP6 of the EU).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.12.011

126
V. Crocellà et al. / Journal of Catalysis 270 (2010) 125–135
When the ring alkylation is carried out with solid acid catalysts,
areas (SSA) were measured by means of the BET single-point meth-
od (N2 adsorption at the temperature of liquid N2), using a Sorpty
1750 Fisons Instrument.
Laser Raman spectra were recorded at room temperature using
a Renishaw 1000 spectrometer equipped with a Leica DMLM
microscope. Samples were excited with an Argon laser beam
(514 nm).
the main drawback is a rapid catalyst deactivation, due to the par-
allel transformation of methanol into olefins and aromatics, both
precursors of coke formation. Moreover, the use of basic or acid
catalysts can select the type of products obtained [22–26]. With
acid catalysts, the reaction is scarcely regio- and chemo-selective,
as the products of both O- and C-methylation are formed (O/C-
selectivity ratio is a function of the reaction temperature and of
the conversion degree). Furthermore, the regio-selectivity is low
since all possible isomers for C-methylation are formed, although
a predominance of the ortho-isomer can be observed. On the con-
trary, basic catalysts show very high chemo-selectivity, producing
only minor amounts of the O-methylation product, as well as high
regio-selectivity, being the ortho-isomer by far favoured.
In previous works [23,24], we reported that basic catalysts of
moderate strength, e.g., MgO-based systems, act first on the dehy-
drogenation of methanol to formaldehyde. Then, the latter acts as
the active species in C-alkylation of the aromatic ring. A good cata-
lyst for this kind of reactions needs to reach a proper balance be-
tween (i) the generation of adsorbed phenolate, via deprotonation
of phenolic –OH groups; (ii) methanol dehydrogenation, to generate
the highly reactive electrophilic formaldehyde; (iii) the decomposi-
tion of formaldehyde into waste light compounds [23,24]. Basic cat-
alysts can guarantee all these features. Among them, MgO-based
systems, and specifically Mg/Me-mixed oxides prepared by incorpo-
ration of guest ions into the MgO lattice, can provide a flexible tool
for a proper tuning of the material reactivity properties.
Mixed oxides of the cited family can be prepared by various
procedures, including decomposition of hydrotalcite precursors
and sol–gel methods [27–34]. In the present work, we will com-
pare the physico-chemical features and the catalytic properties of
Mg2+/Me3+-mixed oxides (Me = Al, Fe, Cr), with the aim of finding
general rules for the control of catalytic performance in the gas-
phase methylation of phenol derivatives.
A fixed Mg/Me atomic ratio equal to 2 was chosen in order to (i)
allow the synthesis of the proper mixed oxides starting from
hydrotalcite-like precursors; (ii) guarantee a good dispersion of
the elements; (iii) tune the strong basic properties of MgO by add-
ing in the MgO structure a small amount of the hetero-element.
The role of methanol during the catalytic reaction and its
decomposition during alkylation of phenolic compounds have not
been sufficiently studied [5,17,20,21] so far. For this reason, final
aim of the present contribution is to examine all these factors in
order to explain the alkylation reaction mechanism.
In situ FTIR spectra were obtained at 4 cmꢀ1 resolution on a Bru-
ker IFS113v spectrophotometer equipped with a MCT cryodetector.
All samples were examined in the form of self-supporting pellets
(ꢁ20 mg cmꢀ2). The spectra of adsorbed probe molecules were ob-
tained with the solid sample contained in a homemade quartz cell,
equipped with KBr windows, connected to a conventional high vac-
uum line (UHV) that allows samples activation and vibrational spec-
tral analysis of the materials in a strictly in situ situation. Prior to
probe-moleculesadsorption/desorptionandspectroscopic analyses,
all catalysts were preliminarily thermally activated in vacuo at
500 °C. In the case of the Mg/Fe/O system, thermal activations were
carried out either in vacuo (reduced ‘‘red” samples) or in oxidizing
conditions in the presence of O2 (oxidized ‘‘ox” samples). The surface
characterization has been performed using: (i) carbon monoxide to
check strong Lewis acidity and redox capacity; (ii) 2,6-dimethylpyr-
idine(familyof lutidine,Lu inthetext)to testbothtotalandBrønsted
acidity. Lu was first adsorbed/desorbed at BT (the temperature of the
sample in the IR beam), and then desorbed at 100 °C.
Catalytic tests were carried out by vapourization of a methanol/
m-cresol liquid mixture (methanol/m-cresol molar ratio 5/1; m-
cresol supplied by Sigma Aldrich; methanol supplied by Carlo Erba
Reagenti) in a N2 stream (gas flow 60 N mL/min, with a N2/reac-
tants molar ratio equal to 5). The composition of the feed gas
was the following (in molar fractions): methanol 0.139, m-cresol
0.028, nitrogen 0.833. Overall gas reciprocal flow was
50 gcat h molꢀ1. Total pressure was atmospheric. For tests made
in the absence of m-cresol, the inlet stream of methanol and carrier
was maintained the same as the one used in m-cresol methylation
tests. The gas/vapours stream was fed to a stainless steel reactor
(length 30 cm, internal diameter 3/400), containing 1.5 g of catalyst
shaped in 30–60 mesh particles. Catalyst particles were prepared
by pressing the calcined powder to obtain pellets that were then
broken into smaller granules. During catalytic measurements, the
reactor exit stream was condensed in 25 mL of HPLC-grade acetone
for 1 h, maintained at 6 °C. Products condensed in acetone were
analysed by gas chromatography, using a GC6000 Carlo Erba
instrument equipped with a FID and a HP-5 column. The GC oven
temperature was programmed from 50 °C to 250 °C, with a heating
rate of 10 °C/min. Non-condensable gases (CO, CO2, H2, CH4) were
analysed by sampling the gaseous stream with a syringe at the
reactor exit, before condensation in acetone, and by injecting the
sample in a GC 4300 Carlo Erba gas chromatograph equipped with
a TCD and a Carbosieve SII column. The GC oven temperature was
programmed from 55 °C to 220 °C, with a heating rate of 10 °C/min.
2. Experimental
Mg/Me-mixed oxides were prepared by thermal treatment of
hydrotalcite-like precursors. The latter were synthesized by a con-
ventional procedure that was described elsewhere [35,40–42]. Pre-
cursors were synthesized using a Mg2+/Me3+ ratio equal to 2, which
is inside the interval of values at which the corresponding hydro-
talcite-like material is formed. X-ray diffraction (XRD) pattern of
precursors showed the presence of crystalline-layered mixed hy-
droxyl-carbonates in the case of Mg/Fe/O and Mg/Al/O systems,
whereas an amorphous pattern was observed in the case of the
Mg/Cr/O precursor. All samples were calcined in air at 450 °C for
8 h in a muffle, and then in N2 for 3 h inside the reactor. Reference
3. Results
3.1. Bulk features of the catalysts
Table 1 reports some results concerning morphology and struc-
ture (relevant XRD patterns are shown in Fig. 1) of the calcined Mg/
Me samples. XRD patterns well agree with those reported in the lit-
erature [34,43,44] for the formation of microcrystalline-mixed Mg/
Me oxides, in which the incorporation of the trivalent Me3+ cation
in the periclase lattice generates cationic defects and produces a
low crystallinity degree. In the case of the Mg/Cr/O sample, reflexes
consistent with the presence of MgCrO4 (JCPDS file 21-1255)
are also observed (in the forthcoming section, the presence of
oxides, such as MgO and
c-Al2O3, were prepared following the
same procedure as for the mixed oxides. The USY zeolite is a com-
mercial sample, supplied by Tosoh (code: HSZ-330HUA).
XRD powder patterns of the catalysts were recorded with Ni-fil-
tered Cu Ka radiation (k = 1.54178 Å) on a Philips X’Pert vertical
diffractometer equipped with a pulse height analyser and a sec-
ondary curved graphite-crystal monochromator. Specific surface

V. Crocellà et al. / Journal of Catalysis 270 (2010) 125–135
127
Table 1
shifts of XRD reflexes were observed, because the ionic radii of
guest cations (Fe3+ 0.69 Å; Cr3+ 0.615 Å) are very similar to the ra-
dius of Mg2+ (0.65 Å).
Nature and specific surface area of the catalysts.
Catalyst Surface
Crystalline phases
Composition
(atomic
area (m2/g) (XRD)
In the case of Mg/Al/O, the pattern of the used sample is quite
similar to that of the corresponding fresh catalyst. Conversely,
the patterns of both used Mg/Fe/O and Mg/Cr/O systems exhibit
appreciable differences with respect to the corresponding calcined
samples. Used Mg/Fe/O is more crystalline than the fresh one, as
indicated by the full width at half maximum (FWHM) of the XRD
signals, and components due to the most intense reflexes of the
MgFe2O4 spinel phase (JCPDS 88-1943) appear. (Still, we cannot
exclude the partial reduction of Fe3+ species during the reaction,
with formation of magnetite (Fe3O4), because MgFe2O4 and Fe3O4
spinel phases possess similar XRD patterns.) In the case of the used
Mg/Cr/O system, reflexes attributed to Mg chromate disappear.
This suggests that the catalytic reaction generates a strongly reduc-
ing environment [23–25], which may cause the reduction of Cr6+
and either the segregation of amorphous Cr2O3 or the incorpora-
tion of Cr3+ within the MgO lattice. FTIR spectroscopic measure-
ments are expected to better clarify this point.
Laser Raman spectra relative to the calcined samples (not re-
ported for the sake of brevity) confirmed the presence of MgCrO4
in the as-calcined (fresh) Mg/Cr/O system, whereas no bands
ascribable to this compound were present anymore in the spec-
trum of the used sample. No Raman spectroscopic evidence was
either found for the formation of Cr2O3 in the used Mg/Cr/O cata-
lyst, and this seems to support the hypothesis of an incorporation
of Cr3+ into the MgO lattice. In the case of used Mg/Fe/O, no Raman
bands ascribable to magnetite were observed, and this confirms
the segregation of the MgFe2O4 spinel compound. However, a pre-
liminary calcination of the used sample was carried out before Ra-
man investigation in order to remove organic contaminants
responsible for a strong fluorescence, and this treatment could
have modified the used catalyst, causing the oxidation of reduced
Fen+-containing species, if present. Bands ascribable to coke were
evident in the spectra of used Mg/Fe/O, but they did not appear
in the spectra of both used Mg/Cr/O and Mg/Al/O catalysts.
ratio, by AAS)
MgO
Fe2O3
Al2O3
Mg/Al/O 185
Mg/Fe/
O
206
32
203
Periclase
Haematite
–
–
–
c
-Alumina
MgO-like mixed oxide
MgO-like mixed oxide
Mg/(Mg + Al) = 0.64
Mg/(Mg + Fe) = 0.67
149
62
Mg/Cr/
O
MgO-like mixed
oxide + MgCrO4
Mg/(Mg + Cr) = 0.65
x
x
Mg/Al
x
x
x
used
fresh
x
x
Mg/Fe
x
x
used
fresh
x
x
x
x
x
x
x
x
Mg/Cr
x
x
x
used
3.2. Surface features of the catalysts
x
The surface basicity of our Mg/Me/O samples, determined by
means of CO2-TPD and reported in previous papers [37,39], showed
that in mixed systems the hosting of a cation having high electro-
negativity decreases the basicity of the bridging O2ꢀ ion when
compared to pure MgO, in agreement with literature findings
[34,43,44]. On the other hand, O2ꢀ ions are sites characterized by
strong basic strength, in particular when O2ꢀ ions are paired with
cationic vacancies. Mg/Me-mixed oxides showed a predominance
of medium strength sites, in that all of them were considerably less
basic than MgO, which showed a predominance of strongly basic
sites [37,39].
fresh
20
30
40
50
60
70
80
°2θ
Fig. 1. XRD patterns of calcined (fresh) and used mixed oxides. All fresh samples
were calcined at 450 °C. ꢂ, MgO-like oxide; ꢃ, MgCrO4;
r
, MgFe2O4. Dotted lines:
main reflections of reference MgO (periclase).
In order to integrate the data obtained with different techniques
and to correlate the surface behaviour of the materials with cata-
lytic reactivity, FTIR spectroscopy of adsorbed probe molecules
has been resorted to.
First of all, the ambient temperature adsorption of CO2 on the
various Mg/Me-mixed oxides (no IR spectra are reported here for
brevity) has been examined in a recent study [45] and indicated
that redox systems, like Mg/Fe/O and Mg/Cr/O, result less basic
than the non-redox one (Mg/Al/O).
As for strong Lewis acidity, carbon monoxide (CO) adsorption/
desorption at BT has been carried out. In fact: (i) CO acts as a weak
Lewis base and, in these experimental conditions, can reveal only
the strongest fraction of surface Lewis acid sites [46]. Moreover,
(ii) CO adsorption is expected to evidence the redox capacity of
the samples, when present.
chromates will be confirmed by IR spectroscopy). The formation of
the latter compound, which crystallizes in larger particles (as sug-
gested by the sharp peaks present in the XRD curves), may be con-
sidered to explain the low surface area obtained for this sample
(actually, much smaller than that of the other mixed oxides).
Fig. 1 shows the XRD patterns of both as-calcined and used
samples. All samples present low crystallinity (broad peaks). For
all of them, the observed patterns are fairly similar to that of the
starting MgO phase, but for Mg/Al/O where a shift of the reflexes
towards greater 2 theta values (with respect to those of the refer-
ence MgO) is indicative of the partial substitution in the lattice of
Mg2+ cations with Al3+ ones (ionic radii: Mg2+ 0.65 Å vs. Al3+
0.50 Å). In the case of either Mg/Fe/O or Mg/Cr/O, no appreciable

128
V. Crocellà et al. / Journal of Catalysis 270 (2010) 125–135
Section B
(d)
Section A
0.05
0.04
(d)
ox
(c)
2186
2192
0.5
ox
(c)
(b)
1200
1100
1000
900
800
Wavenumber cm-1
(b)
(a)
(0.05x)
2183
(a)
2300
2200
2100
2000
1800
1600
Wavenumber cm-1
1400
1200
Wavenumber cm-1
Fig. 2. Absorbance differential spectra, normalized against the background spectrum of the starting samples activated at 500 °C of Mg/Al/O (spectral set a), Mg/Cr/O (spectral
set b), and ox and red Mg/Fe/O (spectral set c and d), obtained in the CO (Section A) and in the OCO (Section B) vibrational regions, after adsorption of 100 Torr CO at BT (solid
m
m
lines), after up to 150 min of contact with CO (broken lines), and after prolonged contact with CO at 350 °C (dotted lines). Where the symbol ox is reported, within the
activation process also a contact (for 30 min) with O2 at the activation temperature was performed, to guarantee the oxidation of the sample. In Section B, spectra relative to
the Mg/Cr/O sample (curves b) are 0.05-fold magnified. Inset to Section A: absorbance spectra obtained in the 1200–900 cmꢀ1 region of Mg/Cr/O sample after CO adsorption
at BT (solid lines) and after contact with CO at 350 °C (dotted lines).
Fig. 2 reports the spectra of CO contacted with Mg/Al/O (curves
a), Mg/Cr/O (curves b), (reduced) ‘‘red” Mg/Fe/O and (oxidized) ‘‘ox”
Mg/Fe/O samples (curves c and d, respectively), vacuum-activated
at 500 °C. Two interesting ‘‘analytical” spectral regions are consid-
ered: Section A is relative to mCO signals (carbonyl-like species),
whereas Section B is relative to mOCO vibrations (carbonate-like spe-
cies). Upon BT contact with CO, all samples present Lewis acidity,
characterized by the formation of Lewis acid/base surface CO-com-
face carbonate-like species, characterized by typical absorptions
in the 2000–1000 cmꢀ1 spectral range [49]. Section B of Fig. 2 sche-
matizes the evolution of low-m spectra with CO contact. It is clear
that virtually no reactivity towards BT–CO adsorbed is evidenced
for both Mg/Al/O and Mg/Cr/O samples (very weak solid line com-
ponents and no increase upon prolonged contact are observed in
curves a and b, respectively), whereas strong bands due to biden-
tate carbonates (
m
OCO bands at ꢁ1600 and ꢁ1350 cmꢀ1) and bicar-
plexes, absorbing in the high-
m
spectral range of Section A.
bonates (mOCO bands at ꢁ1650, ꢁ1400 cmꢀ1, and a dOH signal at
1230 cmꢀ1) are formed on both ‘‘ox” and ‘‘red” Mg/Fe/O samples
in the presence of CO (solid line curves c and d). All these compo-
nents increase upon BT–CO contact, confirming the capacity of CO
in increasing the reduction state of the samples (broken-line
curves c and d). The redox behaviour of Fen+ species is well known
and explains well the experimental results just reported, but one
would expect that Crn+ species could exert the same redox behav-
iour. To ascertain this possibility, some IR spectroscopic measure-
ments have been carried out after heating a just activated Cr-
containing sample at 350 °C in the presence of CO. The results
In the case of Mg/Al/O, CO adsorption indicates the presence of
only
r-coordinated adspecies (solid line a). In fact, a single broad
band is observed, located at ꢁ2183 cmꢀ1 and ascribable to coord-
inatively unsaturated (cus) Al3+ ions in tetrahedral coordination
[47]. This signal is not modified for prolonged contact with CO at
BT.
Curve b (solid line) in the same figure reports the spectrum rel-
ative to the interaction of CO with Mg/Cr/O. A single band located
at ꢁ2192 cmꢀ1, sharp and symmetrical (FWHM ꢄ 25 cmꢀ1), wit-
nesses the homogeneity of the adsorbing sites, which can be iden-
tified as cus Cr3+ cations, on the basis of the spectral behaviour and
what reported in the literature [48].
The behaviour of Mg/Fe/O is completely different in that CO
admitted on ‘‘ox” and ‘‘red” samples produces very different spec-
tra: CO adsorbed on ‘‘ox” sample yields a principal absorption at
ꢁ2190 cmꢀ1 (ascribable to the interaction with cus Fe3+ ions
[45]) and (almost) negligible absorptions at lower wavenumbers,
whereas CO contacted with ‘‘red” sample yields a very strong and
heterogeneous absorption in the 2190–2000 cmꢀ1 range (indicat-
are reported in Fig. 2A, dotted line curve b (for the high-m region),
and in the inset (relative to the spectral region below 1200 cmꢀ1).
A very large signal centred at ꢁ900 cmꢀ1 is present in the back-
ground spectrum of the Mg/Cr/O sample and is due to the presence
of chromate groups [50,51], as already evidenced by XRD data
(Fig. 1). The strong absorption disappears upon heating the sample
in the presence of CO (dotted line spectrum in the inset), and
simultaneously the sharp ꢁ2190 cmꢀ1 band, ascribed above to
CO/Cr3+ carbonyl-like species, shifts down to 2186 cmꢀ1, indicating
the presence of Cr2+ species, that are known from literature data
[48] to be formed only if very dry conditions are reached. This re-
ing carbonyl complexes characterized by the presence of
p-back-
donation), most likely due to the presence of Fen+ species in which
n 6 3. Prolonged contact with CO at BT (up to 2 h) with both sam-
ples produces an increase in the spectral components at low wave-
numbers (broken-line curves c and d), indicating the ability of BT–
CO adsorbed in increasing the reduction state of the samples. As a
consequence of samples reduction, CO is oxidized and forms sur-
dox mechanism is confirmed by spectra in the low-m region (Sec-
tion B, dotted line curve b), where very large absorptions appear,
most likely due to the formation of a huge amount of carbonate
species of the bidentate type and representing the oxidation prod-
ucts. This is a clear indication that a redox activity can indeed be

V. Crocellà et al. / Journal of Catalysis 270 (2010) 125–135
129
exerted by chromium species, although this activity needs to be
activated by a treatment at relatively high temperatures. The over-
all redox behaviour of the Cr-containing system is reversible, as the
back-formation of Cr3+ species occurs when some water is present
in the reactor, and the re-formation of chromate species is easily
induced by oxidation in air.
It is so preliminarily concluded that an order of redox capability
can be set for the MgO-based samples under study, in that: (i) as
expected, the Mg/Al/O system shows no redox capacity; (ii) the
Mg/Cr/O system exhibits weak redox capacity, because chromates
reduction is an activated process; (iii) the Mg/Fe/O system is char-
acterized by a very vigorous redox capacity, evidenced even at low
temperatures and with mild reducing agents (like vacuum activa-
tion and BT contact with CO).
similar behaviour, as witnessed by the common presence of an
absorption of variable complexity at 1600 cmꢀ1 < 1615 cmꢀ1
When the base excess is evacuated at 100 °C (dotted line curves
in Fig. 3), on all systems only part of the higher- Lewis coordi-
nated component remains, whereas for the sole non-reduced Mg/
Cr/O sample (dotted line curve c) complex absorption at
> 1620 cmꢀ1 evidences, far more clearly than in the presence of
Lu excess, the presence of Brønsted acidity. The presence of
Brønsted acid sites turns out to be related to the surface presence
of chromate groups, as Brønsted Lu activity totally disappears
when chromates are reduced by carbon monoxide (see curves set
d). The phenomenon is reversible also in this case, and when chro-
mates are formed again by oxidation in air, Brønsted Lu activity
shows up again and the spectral set c is reproduced.
<
m
.
m
a
m
In order to find out whether the samples under study exhibit
also weak Lewis and/or Brønsted acidity, a strong base like Lu
has been employed. Fig. 3 reports the spectra relative to the
adsorption/desorption of Lu at BT (solid line curves), and Lu
desorption at 100 °C (dotted line curves), on the various Mg/Me-
mixed oxides. It is known from the literature [52] that in the
1680–1540 cmꢀ1 spectral range are located the ‘‘analytical” 8a
and 8b vibrational modes of both neutral or ionic (lutidinium) ad-
sorbed species. For this reason, we focussed our attention only on
this spectral region.
In summary, characterization measurements indicate the
following:
(1) All samples show basic features [45], but also Lewis-type
acidity associated to the presence of guest trivalent metal
cations.
(2) In Mg/Al/O, Al3+ species are exclusively present in the form
of cus Al3+ cations in either octahedral and/or tetrahedral
coordination, but only the latter species exhibit strong Lewis
acidity.
Lu vapour pressure admitted/evacuated at BT on the mixed oxi-
des leads, on all systems, to the formation of three spectral compo-
(3) In Mg/Cr/O, part of Cr is present as Cr3+ species incorporated
in the MgO lattice, and part of it forms crystalline Mg chro-
mates. Brønsted acid sites are attributed to OH groups linked
nents at
m
< 1615 cmꢀ1: (i) the 8b mode band of all species, and
thus related to the total amount of Lu adsorbed on the samples,
is located at ꢁ1580 cmꢀ1; (ii) the 8a mode relative to very weak
interactions (i.e., H-bonding between Lu and surface OH groups
to chromate groups. The latter species are reduced to Cr2+
/
Cr3+ species by CO at high temperature (and also during
the catalytic reaction), becoming (reversibly) incorporated
in the MgO lattice.
and/or non-specific liquid-like physisorption) lies at ꢁ1595 cmꢀ1
;
(iii) the 8a mode relative to stronger interactions (i.e., specific coor-
dinative chemisorption onto surface cus Men+ and/or Mg2+ cations)
is located in the ꢁ1600–1610 cmꢀ1 range. It is so deduced that, as
far as Lewis acidity is concerned, all mixed oxide systems exhibit a
(4) In Mg/Fe/O, Fe3+ ions are incorporated in the MgO frame-
work and are easily reduced by CO, also in mild conditions.
Under catalytic conditions, a partial segregation of either
Fe3O4 or MgFe2O4 occurs. The main feature of this sample
is the pronounced redox behaviour, associated with the cou-
ple Fe3+/Fen+ with n < 3.
1595
1580
3.3. The reactivity in gas-phase methylation of m-cresol: acid and
basic behaviour
0.5
1603
1609
Fig. 4 reports, for the Mg/Al/O catalyst, the conversion of m-cre-
sol and the selectivity to the products, as a function of the reaction
temperature. Products were 3-methylanisole (3-MA, the product of
O-methylation), 2,3-dimethylphenol and 2,5-dimethylphenol (2,3-
DMP and 2,5-DMP, the products of ortho-C-methylation), 3,4-
dimethylphenol (3,4-DMP, the product of para-C-methylation),
and polyalkylated phenols. Products formed in minor amount were
dimethylanisoles (DMA), toluene and benzene. At low tempera-
ture, the prevailing product was 3-MA. Its selectivity rapidly de-
clined when the temperature was increased, with a concomitant
increase in the selectivity first to DMPs and then to polyalkylated
compounds. Above 350 °C, the selectivity to 3-MA became nil,
and also that to DMPs decreased in favour of the formation of
polyalkylates.
These data can be interpreted by considering the following: (i)
the different activation energy for the reactions of O- and C-meth-
ylation, the former being favoured at low temperature and the lat-
ter at high temperature; (ii) the effect of m-cresol conversion on
the distribution of products, since an increasing conversion favours
consecutive reactions occurring on 3-MA, that may act as an alkyl-
ating agent. This is shown in Fig. 5, which reports the results of
tests carried out by feeding 3-MA (i.e., without methanol) at
increasing temperatures over the Mg/Al/O catalyst. The almost
equimolar amount of m-cresol and DMAs produced at low temper-
(d)
(c)
(b)
(a)
1650
1600
Wavenumber cm-1
1550
Fig. 3. Absorbance differential spectra, normalized against the background spec-
trum of the starting samples activated at 500 °C of Mg/Al/O (curves a), Mg/Fe/O
(curves b) and Mg/Cr/O before and after the reduction with CO (curves c and d,
respectively), obtained in the vibrational region of Lu 8a and 8b modes after (i)
contact at BT with the vapour pressure of Lu, and subsequent BT vacuum removal of
the Lu excess (solid lines); (ii) outgassing at 100 °C for 10 min (dotted lines).

130
V. Crocellà et al. / Journal of Catalysis 270 (2010) 125–135
100
80
60
40
20
0
OCH3
MA
+ CH OH
3
+ CH OH
3
OH
mC
OCH3
DMA
CH3
+ mC
- mC
CH3
CH3
H3C
OH
+ mC
- DMP
+ CH OH
3
CH3
H3C
250
300
350
400
450
Temperature (°C)
2,3-DMP
2,5-DMP
4,3-DMP
Fig. 4. m-Cresol conversion (ꢀ) and selectivity to: 3-MA (j), 2,3-DMP (ꢃ), 2,5-DMP
function of the reaction
(N), 3,4-DMP (d), DMA (s), polyalkylates (ꢂ), as
a
temperature. Catalyst: Mg/Al/O. Feed composition (molar fractions): methanol,
Scheme 1. Reaction network for the gas-phase methylation of m-cresol with
methanol, catalysed by Mg/Al/O.
0.139; m-cresol, 0.028; nitrogen, 0.833.
ature indicates an intermolecular reaction between two 3-MA mol-
ecules. Also, minor amounts of DMPs formed. When the reaction
temperature was increased, the selectivity to DMAs decreased,
with a concomitant increase in the selectivity to DMPs and poly-
alkylates. DMPs formed either by reaction between 3-MA and m-
cresol, with the co-formation again of m-cresol (this explains
why the selectivity to m-cresol was substantially not affected by
temperature), or by reaction between DMAs and m-cresol (in this
case two molecules of DMPs are produced). These data suggest that
consecutive reactions occurring on 3-MA may contribute to the de-
crease in selectivity to the latter compound during m-cresol meth-
ylation, when the temperature is increased. However, the very high
selectivity to polyalkylated phenols observed at high temperature
during m-cresol methylation suggests that the direct attack on
the aromatic ring by the electrophilic species is the preferred
reaction.
USY zeolite is shown in Fig. 6. It is noted that, at low temperature,
3-MA was the major product, with also an appreciable formation of
DMAs. The selectivity to the latter compounds was similar to the
overall selectivity to monoalkylated cresols (DMPs) plus polyalky-
lated phenols, showing that the consecutive methylation on 3-MA
(to yield DMAs) was kinetically favoured as much as the direct
methylation on m-cresol, likely because the counter-diffusion path
of 3-MA inside the zeolite porous network increased the effective
residence time of the molecule and the probability of consecutive
reactions. Another difference of USY with respect to Mg/Al/O was
a remarkable deactivation effect, which was the reason for the con-
version decrease experimentally observed when the temperature
was raised. The deactivation was due to the formation of coke
[23,26]. As a consequence of the decreased conversion, the selec-
tivity to 3-MA did not change even when reaction temperatures
higher than 350 °C were used. But, at this point, it is important
to note that, despite the similar conversion exhibited at low tem-
perature and the similar distribution of products obtained with
Mg/Al/O and USY catalysts, the former system did not show any
Scheme 1 summarizes the reaction network, showing the main
reactions occurring with the Mg/Al/O catalyst. It is worth noting
that the nature of products obtained with this catalytic system is
similar to that obtained with acid catalysts and different from that
obtained with MgO. For comparison purposes, the behaviour of a
100
80
60
40
20
0
100
80
60
40
20
0
250
300
350
400
450
250
300
350
400
Temperature (°C)
Temperature (°C)
Fig. 5. 3-Methylanisole (3-MA) conversion (j) and selectivity to: m-cresol (ꢀ), 2,3-
DMP (ꢃ), 2,5-DMP (N), 3,4-DMP (d), DMA (s), polyalkylates (ꢂ), as a function of the
reaction temperature. Catalyst: Mg/Al/O. Feed composition (molar fractions): 3-MA,
0.01; nitrogen, 0.99.
Fig. 6. m-Cresol conversion (ꢀ) and selectivity to: 3-MA (j), 2,3-DMP (ꢃ), 2,5-DMP
function of the reaction
temperature. Catalyst: USY zeolite. Feed composition (molar fractions): methanol,
(N), 3,4-DMP (d), DMA (s), polyalkylates (ꢂ), as
a
0.139; m-cresol, 0.028; nitrogen, 0.833.

V. Crocellà et al. / Journal of Catalysis 270 (2010) 125–135
131
noticeable deactivation effect (as also confirmed by repeated tests
on the used catalysts).
70
60
50
40
30
20
10
0
MgO was found to be much less active than Mg/Al/O [23]. This
is because strongly basic sites, typical of MgO, are poisoned by the
almost irreversible adsorption of acid molecules like, for instance,
of carbon dioxide or even m-cresol itself. Also with the MgO used
in the present work, the conversion at 400 °C was only 6%, not-
withstanding the high surface area of the sample used. The prevail-
ing products were 2,3- and 2,5-DMP (overall selectivity 75%), with
a lesser formation of the product of para-alkylation (3,4-DMP,
selectivity 20%) and of 3-MA (selectivity 5%).
3.4. The reactivity in gas-phase methylation of m-cresol: the role of Me
in Mg/Me-mixed oxides
300
350
400
450
Figs. 7 and 8 show the results obtained with Mg/Fe/O and Mg/
Cr/O catalysts, respectively. It is evident that the two samples be-
haved quite similarly: the conversion was very similar, though
Temperature (°C)
Fig. 9. Yield to: CO (j), CO2 (N), CH4 (d), H2 (ꢀ) as a function of temperature.
Catalyst: Mg/Fe/O. Feed composition (molar fractions): methanol, 0.14; nitrogen,
0.86.
100
80
60
40
20
0
the surface area of Mg/Cr/O is half that of Mg/Fe/O. Moreover, with
the latter catalyst the conversion decreased at 450 °C, because of a
deactivation effect. In fact, when catalytic measurements were car-
ried out again at 400 and 350 °C, the conversion measured was
lower than that originally recorded for these temperatures. Deacti-
vation arose because of the reduction of Fe3+ to Fe2+, the latter spe-
cies being produced by the dehydrogenation of methanol and
oxidation of the hydrogen formed to water [23,24,53,54] (see the
description on the transformations occurring on methanol). Unlike
that, no high-temperature deactivation occurred in the case of the
Mg/Cr/O catalyst.
For what concerns the distribution of products, both samples
exhibited a behaviour which should be that typical of basic sys-
tems [23–26], with predominance of the products of ortho-methyl-
ation (2,3- and 2,5-DMP), and very low selectivity to 3,4-DMP. The
prevailing polyalkylated phenol was the 2,3,6-trimethylphenol, the
product of m-cresol ortho-dialkylation. The selectivity to 3-MA was
very low with Mg/Cr/O and nil with Mg/Fe/O.
250
300
350
400
450
Temperature (°C)
Fig. 7. m-Cresol conversion (ꢀ) and selectivity to: 3-MA (j), 2,3-DMP (ꢃ), 2,5-DMP
function of the reaction
(N), 3,4-DMP (d), DMA (s), polyalkylates (ꢂ), as
a
temperature. Catalyst: Mg/Fe/O. Feed composition (molar fractions): methanol,
0.139; m-cresol, 0.028; nitrogen 0.833.
3.5. The reactivity in gas-phase methylation of m-cresol: the
transformations occurring on methanol
100
80
60
40
20
0
In previous works, we demonstrated that catalysts possessing
basic reactivity cause the dehydrogenation of methanol to formal-
dehyde [23,24]. The latter is a highly electrophilic reagent and acts
as the real methylating agent for ring substitution in phenol, likely
via the intermediate hydroxymethyl group. On the other hand,
formaldehyde may rapidly dehydrogenate to CO or dimerize to
yield methylformate (Tishchenko reaction), which also may even-
tually decompose to CO2 and CH4. In order to highlight a possible
different behaviour of the catalysts in regard to methanol dehydro-
genation and decomposition, tests were carried out by feeding only
methanol.
Mg/Al/O showed a negligible activity in methanol dehydrogena-
tion to formaldehyde and decomposition into light compounds. In
fact, the formation of CO, CO2 and CH4 was very low up to 400 °C
(yield to light compounds less than 5%). The main product of meth-
anol transformation was dimethylether (with the co-production of
water), yield to this latter compound being comprised between
20% and 35% below 400 °C. Noteworthy, no aliphatic and aromatic
compounds were formed, whereas they are produced in the zeo-
lite-catalysed transformation of methanol and are precursors of
coke formation. However, at 450 °C, the dehydrogenation of meth-
250
300
350
400
450
Temperature (°C)
Fig. 8. m-Cresol conversion (ꢀ) and selectivity to: 3-MA (j), 2,3-DMP (ꢃ), 2,5-DMP
function of the reaction
(N), 3,4-DMP (d), DMA (s), polyalkylates (ꢂ), as
a
temperature. Catalyst: Mg/Cr/O. Feed composition (molar fractions): methanol,
0.139; m-cresol, 0.028; nitrogen, 0.833.

132
V. Crocellà et al. / Journal of Catalysis 270 (2010) 125–135
the reason for the decreased activity in m-cresol methylation
70
60
50
40
30
20
10
0
(Fig. 7); a lower rate of generation of formaldehyde and a de-
creased rate of m-cresol activation (due to the removal of basic
O2ꢀ) cause the observed deactivation phenomenon. This did not
occur with Mg/Cr/O and no deactivation effect was observed at
high temperature in neither m-cresol or methanol transformation.
4. Discussion
The peculiar reactivity of Mg/Fe-mixed oxides in the gas-phase
methylation of phenol was previously discussed by Velu and
Swamy [33]. The authors hypothesized that the Lewis-type coordi-
native properties of Fe cations play an important role by orientat-
ing the introduction of the methyl group in the ortho position of
adsorbed phenolate. This was assumed to be the reason for the
high regio-selectivity observed with this catalyst when compared
to Mg/Al/O. Still, this interpretation does not fit our experimental
data. In fact, the same Lewis acidity is present in our Mg/Al/O
and Mg/Fe/O systems that, however, gave quite a different catalytic
performance in m-cresol methylation. Moreover, according to the
widely accepted Tanabe’s model for the adsorption of phenolic
compounds on basic ionic oxides (the model predicts an orthogo-
nal orientation of phenolate species with respect to the catalyst
surface, due to the repulsion between the aromatic ring and nucle-
ophilic surface (lattice) O2ꢀ ions) [55], the presence of electrophilic
surface sites should favour the interaction with the aromatic ring
and bring the para position close to the surface, thus making easier
the methylation at that position.
300
350
400
450
Temperature (°C)
Fig. 10. Yield to: CO (j), CO2 (N), CH4 (d), H2 (ꢀ) as a function of temperature.
Catalyst: Mg/Cr/O. Feed composition (molar fractions): methanol, 0.14; nitrogen
0.86.
anol to formaldehyde and to CO became appreciable: the yield to
CO was 15%, whereas the yield to CO2 was even lower than at
400 °C.
The behaviour of Mg/Fe/O (Fig. 9) and Mg/Cr/O (Fig. 10) was
quite different from that exhibited by Mg/Al/O. Both catalysts were
very active in the dehydrogenation of methanol to formaldehyde
and in the subsequent dehydrogenation of the aldehyde to CO.
However, Mg/Fe/O was much more active than Mg/Cr/O. If we as-
sume that there is no additional source of H2, besides the H2 de-
rived from methanol (and, in fact, with both catalysts, there was
no formation of dimethylether and water), the amount of H2 pro-
duced should correspond to the sum of 2CO + CH4 + CO2, and the
yield of H2 to the sum of yields of CO + 1/2CH4 + 1/2CO2. In fact,
for mere calculation purposes, the following stoichiometries can
be assumed:
Characteristics and reactivity of Mg/Me/O systems described in
the present work turned out to be greatly affected by the nature of
host cation and by the type of mixed oxide formed. It is worth not-
ing that all mixed oxide samples showed reactivity properties
quite different from those typically observed with the correspond-
ing single oxides (i.e., in MgO and Me2O3). In general, MgO presents
strongly basic features, arising from the high nucleophilicity of O2ꢀ
surface ions and from the ionicity of the MgO lattice [56,57].
c-
Al2O3 exhibits amphoteric behaviour [58–60], due to the presence
of acidic cus Al3+ sites (besides, possibly, some Al–OH groups)
going along with Al–O–Al moieties of pronounced basic character.
However, the catalytic behaviour exhibited by alumina is a func-
tion of the characteristics of the reactants used; fairly acidic phenol
and cresols undergo deprotonation by O2ꢀ sites, whereas the elec-
trophylic O atom in methanol undergoes protonation by Al–OH
sites (possibly via methoxy or methylenoxy species). Fe2O3 shows
a typically basic behaviour in regard to phenolic compounds,
favouring the formation of phenolate species and leading to high
chemo- and regio-selectivity in phenol [23] and 1-naphthol meth-
ylation [12]. On the other hand, it has been shown that in Fe2O3
and in Fe-rich Mg/Fe-mixed oxides, the pronounced redox activity
favours methanol dehydrogenation to formaldehyde and its fur-
ther decomposition into light compounds (CO2, CO and CH4) [24].
Acid catalysts drive the reaction preferentially towards ortho-
isomer formation [23, and references therein], but they form also
a not negligible amount of para-isomer. Moreover, the para-isomer
becomes the prevailing species in the presence of some zeolitic
catalysts [61]. On the contrary, heterogeneous systems showing
some basic behaviour give rise only to ortho-mono and dimethy-
lated phenol [23–24].
CH3OH ! CO þ 2 H2
CH3OH ! 1=2CO2 þ 1=2CH4 þ H2
Where the first reaction corresponds to methanol decomposition
and lumps together the dehydrogenation to formaldehyde and then
to CO; the second reaction represents the overall stoichiometry for
the transformation of methanol to formaldehyde, followed by form-
aldehyde dimerization to yield methylformate, and eventual
decomposition of the latter into CH4 and CO2. Therefore, from the
mass-balance point of view, all reactions that lead to the formation
of H2 starting from CH3OH (and in the absence of any additional co-
reactant) are actually included in these two reactions, when the
assumption is made that there is no coke formation on the catalysts
(the accumulation of which would lead to an additional formation
of H2). Indeed, used Mg/Fe/O contained approximately 5 wt.% of
coke, whereas in used Mg/Cr/O the amount of coke was less than
1 wt.%.
Mg/Fe/O was much more active than Mg/Cr/O in the formation
of light compounds. With the former catalyst, the conversion of
methanol was almost total at 400 °C, whereas with Mg/Cr/O it
was less than 40%. This confirms the result of spectroscopic mea-
surements concerning the superior redox properties of Mg/Fe/O
when compared to Mg/Cr/O. With both catalysts, the yield to H2
did effectively correspond to the aforementioned sum of yields to
light compounds. However, at temperatures higher than 400 °C,
Mg/Fe/O did show a decline of light compounds. This is due to
the reduction of Fe3+ and to the depletion of lattice O2ꢀ, brought
about by the oxidation of the H2 generated to H2O. This is also
The reactivity behaviour of the Mg/Me-mixed oxides presented
here can be summarized as follows.
Mg/Al/O does not present the basic reactivity typical of MgO.
Still the interaction of CO2 on this sample indicates a good basicity,
if compared with that exhibited by the Mg/Fe/O and Mg/Cr/O sys-
tems [45]. Moreover, the Mg/Al/O catalyst exhibits a moderate Le-
wis acidity, deriving from cus Al3+ sites, whereas it has no surface

V. Crocellà et al. / Journal of Catalysis 270 (2010) 125–135
133
Brønsted acidity. This is likely because surface Mg–O–Al(OH)–O–
Mg moieties are unstable at high temperatures, so that dehydra-
tion occurs easily during thermal treatments, with water release
and generation of low coordinated Al3+ sites. In the reaction envi-
ronment, the presence of water (derived from the transformation
of methanol into dimethylether) may lead to the regeneration of
Al–OH bonds and, possibly, to the hydrolysis of Mg–O–Al bonds.
On the other hand, the presence of Al makes the bridging O less
nucleophilic than in MgO. MgO is known to catalyse, though with
very low activity, the dehydrogenation of methanol to formalde-
hyde [23]. But a lower nucleophilicity of the basic O site in Mg/
Al/O, when compared to pure MgO, may limit the dehydrogenative
property typical of high surface area MgO and, in the end, this can
be the reason for the acid-type reactivity exhibited by Mg/Al/O
(whose behaviour is, in fact, similar to that of conventional acid
catalysts). However, one remarkable advantage of this system with
respect to conventional solid acid materials is the large predomi-
nance of sites of medium acidic strength. Though strong enough
to catalyse ring methylation, a moderate acidity is useful to limit
the production of heavy compounds and the formation of coke,
responsible for the rapid deactivation phenomena typically met
with both Brønsted (e.g., zeolites [26]) and Lewis (e.g., AlF3 [62])
strong acids during reactions in which methanol is one of the
reactants.
Also in the case of Mg/Cr/O, the main feature was a very homo-
geneous distribution of Lewis acid sites of medium strength, asso-
ciated with Cr3+ ions. The presence of Brønsted acid sites resulted
not so important in assessing the mechanism of catalytic reaction,
as this feature is entirely due to chromate groups that reduce to
Cr2+ (in reducing and very dry conditions) and/or to Cr3+ in the first
reaction steps. The redox capability of the Mg/Cr/O system is evi-
denced only when the redox process is forced to occur, for instance
by treatment at high temperatures. Mg/Cr/O was as active as Mg/
Fe/O in the conversion of m-cresol, but it was remarkably less ac-
tive in methanol transformation (as it should be, considering the
different redox activity exhibited by the two systems). If the differ-
ent SSA of the systems is taken into account, the aeral TOF for m-
cresol methylation at 350 °C is, for Mg/Cr/O, twice as large as that
of Mg/Fe/O, whereas the aeral TOF relative to methanol decompo-
sition is approximately 1/10th of that of Mg/Fe/O. This means that
the efficiency of Mg/Cr/O in the activation of m-cresol was similar
to that of Mg/Fe/O, whereas the latter was far more efficient in the
activation of methanol (i.e., in the dehydrogenation of methanol to
formaldehyde, which is the actual electrophilic reactant for the
activated aromatic ring). However, Mg/Cr/O was much more selec-
tive than Mg/Fe/O in regard to transformations occurring on form-
aldehyde, as a major fraction of the aldehyde was incorporated in
the aromatic ring, and only a minor part of it decomposed to light
compounds. The opposite holds for Mg/Fe/O. Most of the aldehyde
eventually decomposed, and only a very minor fraction of it re-
acted with the aromatics. Finally, Mg/Cr/O catalyst is expected to
show also a limited reactivity due to its Brønsted acidity, at least
in the first steps of the catalytic reaction, i.e., when some surface
chromate groups (responsible for Brønsted acidity) are not reduced
yet. Actually, the formation of 3-MA, albeit very low, was not neg-
ligible. All these explain the catalytic behaviour experimentally ob-
was very active in the dehydrogenation of methanol to formalde-
hyde and in its decomposition to light compounds, in that formal-
dehyde is responsible for the electrophilic attack to the aromatic
ring [24], but is also the precursor for the formation of light com-
pounds. Therefore, despite the Lewis acidity experimentally ob-
served, the dehydrogenation of methanol largely prevailed over
its acid-type activation, and the behaviour observed was that typ-
ical of a basic/dehydrogenating reactivity, with both high regio-
and chemo-selectivity. The selectivity to 3-MA was nil. This behav-
iour was not much different from that observed in the case of hae-
matite [23], but in that case the dissolution of Fe inside MgO leads
to a material of much higher SSA and, therefore, of much higher
activity than bulky Fe2O3.
Our results indicate that the performance of the catalysts is
mainly affected by the oxide capacity of generating the electro-
philic reactant (formaldehyde) and of activating methanol with
an acid-type mechanism. In the case of Mg/Cr/O and Mg/Fe/O sys-
tems, the dehydrogenating properties largely prevail over the acid/
basic properties. Otherwise Mg/Al/O, which is a material character-
ized by good basicity (as revealed by the use of the probe molecule
CO2 [37,39,45]), shows a catalytic behaviour typical of acid/basic
systems. And, in fact, it does not possess sites with dehydrogenat-
ing capacity.
Another peculiarity of the model proposed here is that metha-
nol and formaldehyde give a different range of products in the
reaction with m-cresol, regardless of the catalyst used. In other
words, the catalyst affects the nature of the true electrophilic spe-
cies, but the final type of phenol derivatives obtained is no longer a
function of the catalyst type. This was previously demonstrated by
tests carried out by feeding phenol and formaldehyde over a zeo-
litic material; in this case, the distribution of products obtained
was quite different from that obtained by reacting phenol and
methanol over the same zeolite, and similar to that obtained by
reacting phenol and formaldehyde (or phenol and methanol) over
MgO or Mg/Fe/O catalysts [24]. This also explains why no products
of O-methylation were obtained using Mg/Fe/O or Mg/Cr/O as cat-
alysts for the reaction between m-cresol and methanol: it is prob-
able that the species /-O–CH2OH, which is the intermediate
eventually formed by O-hydroxymethylation, is not stable enough
to further react and give 3-MA. This will appear clearly in a forth-
coming paper (still in preparation), where the whole reaction
mechanism will be described in detail.
5. Conclusions
Mg/Al/O, Mg/Fe/O and Mg/Cr/O-mixed oxides have been thor-
oughly characterized and tested as catalysts for the gas-phase
methylation of m-cresol with methanol. The incorporation of triva-
lent metal ions into the MgO lattice led to the generation of cata-
lysts having pronounced basic features of medium strength as
well as Lewis-type acidity. Despite these analogies, the catalytic
behaviour (in m-cresol methylation) turned out to be completely
different for the three systems under investigation. Mg/Cr/O
showed a typical basic/dehydrogenating catalytic reactivity, with
high chemo- and regio-selectivity and the favoured formation of
2,5-dimethylphenol. Mg/Al/O exhibited a catalytic behaviour quite
similar to that shown by conventional Brønsted-type acid catalysts
(e.g., zeolites), but one remarkable advantage of Mg/Al/O systems
consists in the absence of deactivation phenomena (typically ob-
served with conventional acid catalysts in reactions employing
methanol as reactant). With respect to the distribution of phenolic
products, Mg/Fe/O presented a catalytic behaviour close to that of
Mg/Cr/O, but showed a pronounced activity in both methanol-to-
formaldehyde dehydrogenation and the decomposition of formal-
dehyde into light compounds.
served, with
a high regio-selectivity (typical of truly basic
systems), and a very high (but not total) chemo-selectivity.
Mg/Fe/O has a surface acidity close to that of Mg/Al/O, with the
predominance of Lewis sites of moderate strength. Therefore, one
might expect also a similar catalytic behaviour. However, the main
feature of the system was its pronounced redox behaviour, as
shown by the reduction of Fe3+ cations by CO. During the reaction
with methanol and m-cresol, reduction of Fe3+ by either H2 or CO
probably occurred in the entire range of temperatures examined,
but reduction became relevant only above 400 °C. The catalyst

134
V. Crocellà et al. / Journal of Catalysis 270 (2010) 125–135
Favoured by acidic properties (Mg/Al/O)
OCH3
OH
CH3
Favoured by basic-dehydrogenating
H3C
CH3
properties (Mg/Cr/O)
OH
OH
OH
CH3
H3C
CH3
CH2O + H2
CH3OH
CH3
CO, CH4, CO2, H2
Favoured by redox properties (Mg/Fe/O)
Scheme 2. Summary of reactions occurring with Mg/Me/O systems.
[16] T. Mathew, M. Vijayaraj, S. Pai, B.B. Tope, S.G. Hegde, B.S. Rao, C.S. Gopinath, J.
Catal. 227 (2004) 175.
[17] S. Sato, K. Koizumi, F. Nozaki, J. Catal. 178 (1998) 264.
[18] S. Velu, C.S. Swamy, Appl. Catal. A 145 (1996) 225.
[19] A.H. Padmasri, A. Venugopal, V. Durgakumari, K.S. Rama Rao, P. Kanta Rao, J.
Mol. Catal. A 188 (2002) 255.
[20] K.-T. Li, I. Wang, K.-R. Chang, Ind. Eng. Chem. Res. 32 (1993) 1007.
[21] P.D. Chantal, S. Kaliaguine, J.L. Grandmaison, Appl. Catal. 18 (1985) 133.
[22] R.F. Parton, J.M. Jacobs, D.R. Huybrechts, P.A. Jacobs, Stud. Surf. Sci. Catal. 46
(1989) 163.
[23] N. Ballarini, F. Cavani, L. Maselli, A. Montaletti, S. Passeri, D. Scagliarini, C.
Flego, C. Perego, J. Catal. 251 (2007) 423.
[24] N. Ballarini, F. Cavani, L. Maselli, S. Passeri, S. Rovinetti, J. Catal. 256 (2008) 215.
[25] N. Ballarini, F. Cavani, S. Guidetti, L. Maselli, A. Montaletti, S. Passeri, S.
Rovinetti, Stud. Surf. Sci. Catal. 174 (2008) 1243.
[26] M. Bregolato, V. Bolis, C. Busco, P. Ugliengo, S. Bordiga, F. Cavani, N. Ballarini, L.
Maselli, S. Passeri, I. Rossetti, L. Forni, J. Catal. 245 (2007) 283.
[27] T. Lopez, P. Bosch, E. Ramos, R. Gomez, O. Novaro, D. Acosta, F. Figueras,
Langmuir 12 (1996) 189.
The catalytic behaviour of the three mixed oxide systems fol-
lowed the redox features of the hosted Me3+ species, as repre-
sented in Scheme 2. Only with Mg/Al/O, showing no capacity for
methanol dehydrogenation, the performance was that typical of
conventional acid-catalysed methylation of phenolic compounds.
With Mg/Cr/O and Mg/Fe/O, methanol was activated and dehydro-
genated to formaldehyde. Formaldehyde can either react with m-
cresol or decompose into light compounds. The former reaction
was the quickest one with Mg/Cr/O, whereas with Mg/Fe/O most
of the formaldehyde generated kept reacting with continuity to
yield light compounds. In the latter case, the selectivity to methyl-
ated phenolic compounds (with respect to converted methanol)
was very low.
Acknowledgments
[28] F. Prinetto, G. Ghiotti, P. Graffin, D. Tichit, Micropor. Mesopor. Mater. 39 (2000)
229.
[29] A. Guida, M.H. Lhouty, D. Tichit, F. Figueras, P. Geneste, Appl. Catal. A 64 (1997)
251.
INSTM is gratefully acknowledged for financial support (PRIS-
MA project) and for co-financing the PhD grant of S.P. ENI is
acknowledged for the PhD grant of D.S.
[30] A. Corma, V. Fornés, R.M. Martina-Aranda, F. Rey, J. Catal. 134 (1992) 58.
[31] A. Corma, V. Fornés, F. Rey, J. Catal. 148 (1994) 205.
[32] R.J. Davis, E.G. Derouane, J. Catal. 132 (1991) 269.
[33] S. Velu, C.S. Swamy, Appl. Catal. A 162 (1997) 81.
[34] J. Sanchez Valente, F. Figueras, M. Gravelle, P. Kumbhar, J. Lopez, J.P. Besse, J.
Catal. 189 (2000) 370.
References
[35] M. Bolognini, F. Cavani, D. Scagliarini, C. Flego, C. Perego, M. Saba, Catal. Today
75 (2002) 103.
[36] F. Cavani, C. Felloni, D. Scagliarini, A. Tubertini, C. Flego, C. Perego, Stud. Surf.
Sci. Catal. 143 (2002) 953.
[37] M. Bolognini, F. Cavani, C. Felloni, D. Scagliarini, C. Flego, C. Perego, in: D.G.
Morrell (Ed.), Catalysis of Organic Reactions, Marcel Dekker, Inc., New York,
2002, p. 115.
[38] M. Bolognini, F. Cavani, D. Scagliarini, C. Flego, C. Perego, M. Saba, Micropor.
Mesopor. Mater. 66 (2003) 77.
[1] H. Fiege, ‘‘Cresols and Xylenols” in Ullmann’s Encyclopedia of Industrial
Chemistry, Wiley VCH, Weinheim, 2005, p. 25.
[2] S. Velu, C.S. Swamy, Appl. Catal. A 119 (1994) 241.
[3] S. Sato, R. Takahashi, T. Sodesawa, K. Matsumoto, Y. Kamimumra, J. Catal. 184
(1999) 180.
[4] T.M. Jyothi, T. Raja, M.B. Talawar, B.S. Rao, Appl. Catal. A 211 (2001) 41.
[5] V. Durgakumari, S. Narayanan, J. Mol. Catal. 65 (1991) 385.
[6] S. Sato, K. Koizumi, F. Nozaki, Appl. Catal. A 133 (1995) L7.
[7] R. Bal, S. Mayadevi, S. Sivasanker, Org. Process Res. Devel. 7 (2003) 17.
[8] R. Bal, B.B. Tope, S. Sivasanker, J. Mol. Catal. A 181 (2002) 161.
[9] A.S. Reddy, C.S. Gopinath, S. Chilukuri, J. Catal. 243 (2006) 278.
[10] A.R. Gandhe, J.B. Fernandes, S. Varma, N.M. Gupta, J. Mol. Catal. A 238 (2005)
63.
[11] V.V. Rao, K.V.R. Chary, V. Durgakumari, S. Narayanan, Appl. Catal. 61 (1990) 89.
[12] H. Grabowska, W. Mista, J. Trawczynski, J. Wrzyszcz, M. Zawadzki, Appl. Catal.
A 220 (2001) 207.
[13] T. Mathew, N.R. Shiju, K. Sreekumar, B.S. Rao, C.S. Gopinath, J. Catal. 210 (2002)
405.
[14] T. Mathew, S. Shylesh, B.M. Devassy, C.V.V. Satyanarayana, B.S. Rao, C.S.
Gopinath, Appl. Catal. A 261 (2004) 292.
[15] T. Mathew, N.R. Shiju, B.B. Tope, S.G. Hegde, B.S. Rao, C.S. Gopinath, Phys.
Chem. Chem. Phys. 4 (2002) 4260.
[39] F. Cavani, L. Maselli, D. Scagliarini, C. Flego, C. Perego, Stud. Surf. Sci. Catal. 155
(2005) 167.
[40] F. Cavani, F. Trifirò, A. Vaccari, Catal. Today 11 (1991) 173.
[41] J.I. Di Cosimo, V.K. Diez, M. Xu, E. Iglesia, C.R. Apesteguia, J. Catal. 178 (1998) 499.
[42] J.I. Di Cosimo, C.R. Apesteguia, M.J.L. Gines, E. Iglesia, J. Catal. 190 (2000) 261.
[43] T. Sato, T. Wakahayash, M. Shimada, Ind. Eng. Chem. Prod. Res. Devel. 25
(1986) 89.
[44] D. Tichit, M.H. Lhouty, A. Guida, B.H. Chiche, F. Figueras, A. Auroux, D. Bartalini,
E. Garrone, J. Catal. 151 (1995) 50.
[45] G. Magnacca, G. Cerrato, V. Crocellà, C. Morterra Il Nuovo Cimento 123B (10–
11) (2008) 1505.
[46] G. Magnacca, G. Cerrato, C. Morterra, M. Signoretto, F. Somma, F. Pinna, Chem.
Mater. 15 (2003) 675.

V. Crocellà et al. / Journal of Catalysis 270 (2010) 125–135
135
[47] C. Morterra, V. Bolis, G. Magnacca, Langmuir 10 (1994) 1812.
[48] A. Zecchina, E. Garrone, G. Ghiotti, S. Coluccia, J. Phys. Chem. 79 (1975) 972.
[49] G. Busca, V. Lorenzelli, Mater. Chem. 7 (1982) 89.
[50] R.A. Nyquist, R.D. Kagel, Infrared Spectra of Inorganic Compounds, Academic
Press, 1971.
[56] C. Di Valentin, A. Del Vitto, G. Pacchioni, S. Abbet, A.S. Wörz, K. Judai, U. Heiz, J.
Phys. Chem. B 106 (2002) 11961.
[57] M. Ziolek, J. Kujawa, O. Saur, J.C. Lavalley, J. Phys. Chem. 97 (1992) 9761.
[58] E. Santacesaria, D. Grasso, D. Gelosa, S. Carrà, Appl. Catal. 64 (1990). 83 and
101.
[51] N.E. Fouad, H. Knözinger, M.I. Zaki, S.A. Mansour, Z. Phys. Chem. 171 (1991) 75.
[52] C. Morterra, G. Cerrato, G. Meligrana, Langmuir 17 (2001) 7053.
[53] H. Grabowska, W. Mista, L. Syper, J. Wrzyszcz, M. Zawadzki, Appl. Catal. A 144
(1996) L1.
[54] H. Grabowska, W. Mista, L. Syper, J. Wrzyszcz, M. Zawadzki, J. Catal. 160 (1996)
134.
[59] Y. Fu, T. Baba, Y. Ono, Appl. Catal. A 166 (1998). 419 and 425.
[60] S. Porchet, L. Kiwi-Minsker, R. Doepper, A. Renken, Chem. Eng. Sci. 51 (11)
(1996) 2933.
[61] M.E. Sad, C.L. Padró, C.R. Apesteguía, Appl. Catal. A 342 (2008) 40.
[62] M. Ardizzi, N. Ballarini, F. Cavani, E. Chiappini, L. Dal Pozzo, L. Maselli, T. Monti,
Appl. Catal. B 70 (2007) 597.
[55] K. Tanabe, Stud. Surf. Sci. Catal. 20 (1985) 1.