A cascade mechanism for a simple reaction: The gas-phase methylation of phenol with methanol

Article abstract of DOI:10.1016/j.jcat.2019.01.014

The gas-phase alkylation of phenol with methanol, a reaction triggered for the production of o-cresol and 2,6-xylenol, is catalysed by MgO-based catalysts. Despite the industrial use of this process, the mechanism of the reaction – which is commonly believed to be based on a classical electrophilic attack of activated methanol onto the aromatic ring – is far from being fully understood. In some previous studies we reported that the reaction intermediate is salicylic alcohol, which is formed by the reaction between the adsorbed phenolate and formaldehyde, the latter being formed in-situ by methanol dehydrogenation. Here we elucidate the following steps of the reaction mechanism, by combining reactivity experiments and DFT calculation, with MgO as a model catalyst. It was found that salicylic alcohol dehydrates into quinone methide, which is then reduced via H-transfer by methanol to o-cresol. Moreover, a dehydrogenation/hydrogenation equilibrium is established between salicylic alcohol and salicylic aldehyde. The methide can also react with methanol to form 2-methoxymethylphenol, which may decompose into o-cresol, thus providing an alternative pathway for the formation of the alkylated compound.

Full text of DOI:10.1016/j.jcat.2019.01.014

Journal of Catalysis 370 (2019) 447–460
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
A cascade mechanism for a simple reaction: The gas-phase methylation
of phenol with methanol
Tommaso Tabanelli , Sauro Passeri , Stefania Guidetti , Fabrizio Cavani a,b, Carlo Lucarelli d,
a,
⇑
a
a
c
d,
⇑
Fausto Cargnoni , Massimo Mella
a
Dipartimento di Chimica Industriale ‘‘Toso Montanari’’, ALMA MATER STUDIORUM Università di Bologna, Viale Risorgimento 4, 40136 Bologna, Italy
Consorzio INSTM, Research Unit of Bologna, Florence, Italy
Istituto di Scienze e Tecnologie Molecolari (ISTM), Consiglio Nazionale delle Ricerche, Via Golgi 19, 20133 Milan, Italy
Dipartimento di Scienza ed Alta Tecnologia, Università degli Studi dell’Insubria, Via Valleggio 11, 22100 Como, Italy
b
c
d
a r t i c l e i n f o
a b s t r a c t
Article history:
The gas-phase alkylation of phenol with methanol, a reaction triggered for the production of o-cresol and
Received 25 October 2018
Revised 12 December 2018
Accepted 9 January 2019
2,6-xylenol, is catalysed by MgO-based catalysts. Despite the industrial use of this process, the mecha-
nism of the reaction – which is commonly believed to be based on a classical electrophilic attack of acti-
vated methanol onto the aromatic ring – is far from being fully understood. In some previous studies we
reported that the reaction intermediate is salicylic alcohol, which is formed by the reaction between the
adsorbed phenolate and formaldehyde, the latter being formed in-situ by methanol dehydrogenation.
Here we elucidate the following steps of the reaction mechanism, by combining reactivity experiments
and DFT calculation, with MgO as a model catalyst. It was found that salicylic alcohol dehydrates into qui-
none methide, which is then reduced via H-transfer by methanol to o-cresol. Moreover, a dehydrogena-
tion/hydrogenation equilibrium is established between salicylic alcohol and salicylic aldehyde. The
methide can also react with methanol to form 2-methoxymethylphenol, which may decompose into o-
cresol, thus providing an alternative pathway for the formation of the alkylated compound.
Ó 2019 Elsevier Inc. All rights reserved.
Keywords:
Phenol
Methanol
MgO
Alkylation
o-cresol
Reaction mechanism
1
. Introduction
The gas-phase methylation of phenol with methanol is one of
Therefore, catalyst activity in phenol methylation is governed
by the ability to both: (i) activate phenol and generate the pheno-
late species, this reaction occurring on basic oxides, e.g. alkali and
alkaline earth oxides; and (ii) dehydrogenate methanol into
formaldehyde, which is the rate-determining step of the overall
process. Transition metal oxides, especially those based on Fe,
Mo, and V oxides, show high activity in the latter reaction, but
unfortunately are also active in total formaldehyde decomposition
the first examples of a ‘‘green reaction” applied to industry, in
which an electrophilic substitution on an aromatic ring was carried
out by using an alcohol. In fact, these types of reactions are typi-
cally triggered using reactants which co-produce inorganic salts,
such as dimethylsulphate or methylchloride [1–15]. Even though
this reaction is described in encyclopaedias as a classical S
E
Ar alky-
2
into CO and H , which is an undesired side reaction. Therefore, the
d+
lating Friedel-Crafts reaction in which the reactive species is a CH
3
optimal ‘‘bifunctional” catalyst should be that which combines the
features of both types of elements, while at the same time showing
moderate dehydrogenation properties. For instance, Mg/Fe and
Mg/Cr mixed oxides offer mildly basic properties, deriving from
both the electron-withdrawing properties of the more electroneg-
+
species, formed by the protonation of methanol by the H released
from phenol [16], we demonstrated in several previous papers that
the true reactive species is formaldehyde, generated by the in-situ
dehydrogenation of methanol [17–23]. This aldehyde attacks the
phenolate compound, generating a methylol moiety. This reaction
is facilitated both by the high electrophilicity of formaldehyde, and
by the negative charge delocalisation on the phenolate ring.
Indeed, the reaction between formaldehyde and phenolate may
take place even without the presence of any catalyst.
3
+
3+
ative hosted cation (either Fe or Cr ) that makes the O anion less
basic, and from mild dehydrogenating properties. Indeed, an exces-
sive basic strength is undesired, e.g. in MgO, because of the very
strong interaction which may develop between the phenolic com-
pound and the catalyst: this phenomenon may be responsible for
catalyst deactivation.
In our previous work we described the first steps of the mech-
anism, i.e. methanol dehydrogenation and the attack of formalde-
⇑
Corresponding authors.
https://doi.org/10.1016/j.jcat.2019.01.014
021-9517/Ó 2019 Elsevier Inc. All rights reserved.
0

4
48
T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
hyde onto the phenolate species; we were not, however, able to
define the successive steps, i.e. those from the intermediately
formed salicylic alcohol up to the final products of phenol methy-
lation, o-cresol and 2,6-diphenols. A few different hypotheses,
summarised in Scheme 1, were proposed, but no clear experimen-
tal evidence enabled us to definitively settle on one of them.
The aim of the present work is precisely to confirm the sug-
gested mechanism for phenol methylation with methanol on basic
catalysts by elucidating all the steps of this apparently simple, but
never fully understood, industrial reaction.
2 2 4
Incondensable gases (CO, CO , H , CH ) were analysed by sam-
pling the gaseous stream at the reactor exit before condensation in
acetone, and by injecting the sample into a GC 4300 Carlo Erba gas
chromatograph, equipped with a TCD and a Carbosieve SII column.
The GC oven temperature was programmed from 40 °C to 220 °C,
with a heating rate of 10 °C/min. The products condensed in ace-
tone were analysed by gas chromatography, using a GC6000 Bega
Series 2 Carlo Erba instrument equipped with a FID and a HP-5 col-
umn. The GC oven temperature was programmed from 50 °C to
250 °C. Before carrying out the reactivity tests, Mg/Fe/O catalysts
were pre-treated in a N
2
flow at 450 °C for 3 h, in order to desorb
water and carbon dioxide.
Diffusional limitations were ruled out on the grounds of exper-
iments reported in previous studies [20–22].
2
. Experimental section
2.1. Catalyst preparation and characterisation
Mg/Fe mixed oxide with an Mg/Fe atomic ratio of 0.25 was pre-
2.3. Modelling approach
pared by co-precipitation from an aqueous solution. First, Mg
NO O and Fe(NO O were dissolved separately in
water, to prepare 1 M solutions. The latter were mixed in the nec-
essary quantities to obtain the desired Mg/Fe ratio. The obtained
(
3
)
2
ꢀ6H
2
3
)
3
ꢀ9H
2
Gas-phase electronic structure calculations were performed
using the Gaussian09 suites of codes and the B3LYP [24,25] density
functional theory (DFT), including all the electrons in the calcula-
tions. The basis set used was 6-31++G(d,p). MgO nano-crystals
were modelled using a cluster approach that was previously
benchmarked for structure, energetics, and spectral properties
[26–29], by evaluating any possible size effects on adsorption ener-
gies and structures by increasing the number of cluster atoms,
while maintaining the overall species neutrality. This investigation
solution was added dropwise to an aqueous 2 M solution of Na
pH kept at 10), maintained at 50 °C, while continuing vigorous
stirring. The precipitate was aged for 40 min, then filtered and
washed with H O (40 °C, 0.4 l/g precipitate). The precipitate was
then dried at 90 °C overnight, and finally calcined at 450 °C for
h in air. The crystalline structure was identified by X-ray diffrac-
tion (XRD) using a Phillips X’pert powder diffractometer, with Cu
radiation. The BET specific surface area was measured by N
adsorption–desorption at liquid nitrogen temperature on a Sorpty
750 Carlo Erba instrument. The Mg/Fe/O catalyst showed a speci-
2 3
CO
(
2
8
led us to use Mg10O10; only occasionally was the Mg16O16 cluster
Ka
2
used to limit polarisation effects due to vertexes and edges. The
geometrical parameters of the MgO (cubic, dMgO = 2.1084 Å) clus-
ters were kept frozen in all calculations, as minor effects would
be expected for them [26], as well as for energy quantities. Geome-
tries for the adsorbed species (i.e. phenols, phenates, water) were
fully optimised, while the smallest MgO cluster made it possible
to fully characterise structures by means of frequency calculations;
1
2
fic surface area of 108 m /g; the XRD pattern showed the reflec-
tions of magnesium ferrite (84-0307) and iron oxide (75-0033).
2.2. Reactivity test
in the case of Mg16O16 this was, unfortunately, impossible due to
its size. Putative structures for energy minima were built using
data from our previous studies, followed by a complete geometri-
cal relaxation, while keeping the MgO cluster constrained. In many
cases, putative TS structures were generated by means of relaxed
scans along chosen reaction coordinates (e.g. atom-atom dis-
tances) and subsequently fully optimised by using analytical sec-
ond derivatives of the energy surface. As for the description of
gas-phase thermodynamics, we used the same theory level used
for calculations on clusters. Putative structures for reactants and
products were built on the basis of chemical intuition, and then
fully optimised, validating their nature by computing vibrational
frequencies. The latter were also used to estimate enthalpies and
Gibbs energies via the harmonic oscillator approximation.
The reactivity test was carried out in a continuous-downflow
reactor at different temperatures, from 250 to 450 °C, at atmo-
spheric pressure. The reactor was loaded with variable volumes
of catalyst, shaped in particles with diameters ranging from 0.25
to 0.6 mm. The reagent solution was fed by a Precidor Type 5003
2
diffusion pump together with N as the carrier gas. The flow rate
of the carrier and reagent solution depended on the desired contact
time. The reactor outlet was condensed in 25 ml of HPLC-grade
acetone.
3
. Results and discussion
3.1. The reactivity of MgO in phenol methylation
In a few previous studies, the reactivity of MgO was thoroughly
examined with the aim of determining the reaction mechanism for
the gas-phase methylation of phenol [18–21,23]. In general, MgO is
not a good catalyst for this reaction, since it shows poor activity.
This is the reason why a co-element, typically an element showing
3
+
3+
enhanced dehydrogenating properties (Fe , Cr ), is usually added
to enhance the catalytic activity. Our previous results clearly
demonstrated that the role of these elements is not the generation
of a Lewis-type acid cation that facilitates the activation of metha-
nol, as is often reported in literature, but rather that of enhancing
the rate of methanol dehydrogenation to formaldehyde, the rate-
determining step of the cascade process. On the other hand, the
Scheme 1. A summary of the hypotheses previously formulated for the transfor-
mation of the intermediately formed salicylic alcohol into o-cresol during phenol
methylation with basic-dehydrogenating catalysts [20–22].

T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
449
introduction of these elements increased the extent of methanol
and formaldehyde decomposition, with an enhanced formation of
light compounds (CO, CH , CO , H ). Therefore, we first investi-
4
2
2
gated the effect of the surface area with a pure MgO catalyst, with
the aim of discovering whether this parameter can be an effective
tool for controlling the catalytic performance of MgO.
Fig. 1 compares the catalytic performance based on tempera-
2
ture for both a commercial MgO (surface area 12 m /g), and a syn-
2
thesised MgO (surface area 68 m /g [22]). The two samples
behaved quite similarly, despite their different surface areas. In a
previous paper, we made the hypothesis that active sites in MgO
are corner and steps at crystallite surface.[20] Based on this, it
might be expected that the high-surface-area MgO has both a
greater overall number and a greater density of defect sites, due
to its lower crystallinity. This is in contrast with the experimental
evidence. One possible explanation is that particle efficiency in the
high-surface-area MgO is low, due to internal diffusional limita-
tions. The implication for this is that only sites located at the exter-
nal surface of catalyst particle contribute to reactant conversion. If
this were the case, the reactivity of MgO would not differ too much,
regardless of its specific surface area.
The high-surface-area MgO showed the formation of anisole at
low temperatures – with a corresponding lower selectivity to o-
cresol – which instead was not the case for commercial MgO. In
both cases, the apparent activation energy was 22 ± 2 kcal/mole.
Since external and internal diffusional limitations were ruled out
on the basis of preliminary reactivity experiments, these results
suggest that the rate-determining step of the reaction, the dehy-
drogenation of methanol to formaldehyde, occurs on few active
sites, the amount of which is not proportional to the surface area
of MgO. Moreover, the acidity of phenol is likely to cause a satura-
tion of the surface because of the strong interaction which devel-
ops with MgO basic sites, thus limiting the possibility of
methanol to adsorb and activate. In fact, some experiments carried
out based on the methanol/phenol feed ratio (Fig. 2) showed that,
indeed, this parameter greatly affects catalytic performances; an
increased feed ratio led to a remarkably increased phenol conver-
sion, which was also the reason for the enhanced selectivity to
Fig. 2. Solid symbols and lines: methanol/phenol molar ratio 5/1; empty symbols,
broken lines: 10/1; solid symbols, dotted lines: 20/1. Symbols: (}r) Phenol
conversion; (4N) Selectivity of o-cresol; (sd) Selectivity of 2,6-xylenol. Catalyst:
commercial MgO.
the same as that resulting from methanol: o-cresol and 2,6-
xylenol. With both catalysts there was no formation of anisole;
by-products formed in smaller amounts were p-cresol (also
observed with methanol) and other dimethylated compounds. At
very low temperatures (e.g. at 250 °C) the prevailing product was
salicylic aldehyde, obtained by the ortho-hydroxymethylation of
phenol to salicylic alcohol and dehydrogenation to the aldehyde.
These experiments would suggest that steps (a) and (b) in Scheme 1
2 3
occur because of the reducing action of CH O, and not of CH OH.
However, some in-situ disproportionation of formaldehyde with
generation of methanol cannot be ruled out. On the other hand, a
salicylic alcohol disproportionation is not impossible either (see
experiments on salicylic alcohol reactivity below).
As expected, MgO and Mg/Fe/O showed similar performances in
the reaction starting from formaldehyde (whereas they performed
differently from methanol); this was because the role of Fe in Mg/
Fe/O is that of enhancing the rate of the rate-determining step, i.e.
methanol dehydrogenation to formaldehyde. The apparent activa-
tion energy of the reaction turned out to be very low, equal to
7
± 1 kcal/mole for both catalysts [21]. This was due to the very
2
,6-xylenol and correspondingly decreased selectivity to o-cresol.
high reactivity of formaldehyde in relation to the attack by the aro-
matic ring of the phenolate species.
When a large excess of methanol was used, it was possible to
achieve a high phenol conversion, close to 70% at 450 °C.
In conclusion, these data demonstrated that even though MgO
is a good catalyst for the gas-phase methylation of phenol, espe-
cially in terms of regio- and chemo-selectivity, it is necessary to
have the presence of a second element, more active in methanol
dehydrogenation, in order to enhance the catalytic activity without
the need for a very large surplus of methanol in the feed. In the fol-
lowing chapter, we examine the reactivity of a Mg/Fe mixed oxide,
with an atomic ratio between the two elements of 0.25. This com-
position was chosen based on the results previously reported. A
catalyst based on iron oxide only is active and selective in the title
reaction, but is also very active in the deoxygenation of the pheno-
lic compounds, forming benzene, toluene, and xylenes; moreover,
it is very active in methanol decomposition. The presence of Mg
made it possible to obtain a catalyst considerably more active than
either MgO or FeOx, while showing a moderate amount of metha-
nol decomposition [21,22].
When the reaction was triggered by feeding formaldehyde (an
aqueous solution containing 30% formaldehyde and traces of
methanol) and phenol, the resulting phenol conversion was much
greater than with methanol, thus confirming that the rate-
determining step of the process from methanol is the generation
of the aldehyde [21]. The nature of the products observed was
3.2. The reactivity of Mg/Fe/O catalyst in phenol methylation
Fig. 3(top) shows the conversion of phenol and the selectivity to
the reaction products achieved with the Mg/Fe mixed oxide cata-
lyst. This catalyst shows greater reactivity as compared to MgO,
because of the enhanced dehydrogenation properties induced by
the presence of Fe, which catalyses the transformation of methanol
into formaldehyde. Almost total phenol conversion resulted
Fig. 1. Phenol methylation based on temperature over commercial MgO (solid lines,
solid symbols), and synthesised MgO (broken lines, empty symbols). Feed
methanol/phenol ratio 10/1 (molar). Symbols: (}r) Phenol conversion; (4N)
Selectivity of o-cresol; (sd) Selectivity of 2,6-xylenol; (hj) Selectivity of anisole.

4
50
T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
Fig. 4. Conversion of phenol, molar selectivity to o-cresol, 2,6-xylenol, and
deoxygenated compounds based on temperature. Feed: formalin/phenol
formaldehyde/phenol molar ratio 10/1). The formalin solution contained 30 wt%
formaldehyde, and traces of methanol. Symbols: (r) Phenol conversion;
(
(
N) Selectivity of o-cresol; (d) Selectivity of 2,6-xylenol; (j) Selectivity of
deoxygenated compounds; (}) Selectivity of ”Others”.
In order to confirm the hypothesis that salicylic alcohol is the
first intermediate compound of the reaction, we conducted exper-
iments at very low contact times, to obtain very low phenol con-
version. Fig. 5 shows the selectivity to the products observed by
feeding the phenol/methanol solution, under conditions in which
the registered conversion of phenol was lower than 1%.
The products observed were salicylic alcohol, salicylic aldehyde,
Fig. 3. Above: Conversion of phenol, molar selectivity to o-cresol, 2,6-xylenol, and
benzene + toluene, based on temperature. Symbols: (r) Phenol conversion; (N)
Selectivity of o-cresol; (d) Selectivity of 2,6-xylenol; (j) Selectivity of deoxy-
2
-methoxymethylphenol, o-cresol, and 2,6-xylenol. These results
are in agreement with those previously obtained with MgO
21,22], thus confirming that salicylic alcohol is the primary pro-
duct of the multistep process. Other products, salicylic aldehyde
and 2-methoxymethylphenol, can form by consecutive reactions
occurring on the alcohol: (i) via dehydrogenation, for the formation
of the aldehyde, and (ii) via reaction with methanol for the ether;
these routes, however, will be discussed more in detail later on.
[
genated compounds; (}) Selectivity of ”Others”. Below: Molar yield of CO, CO
and H based on temperature. Methanol/phenol feed ratio 10/1. Symbols: (r) Yield
of methane; (N) Yield of CO ; (d) Yield of H ; (j) Yield of CO. Catalyst: Mg/Fe/O.
2 4
, CH ,
2
2
2
already at 350 °C, with only the formation of ortho-C-alkylated
compounds (o-cresol and 2,6-xylenol), and no anisole or p-cresol
at all; hence the reaction was very chemo- and regio-selective. It
is interesting to note, however, that total phenol conversion was
never achieved, with the maximum conversion being close to
3
.3. The reactivity of the key intermediate: salicylic alcohol
We conducted reactivity tests by directly feeding salicylic alco-
9
0% already at 350 °C, despite the great availability of methanol;
hol and methanol. The results are shown in Fig. 6, plotting the con-
version of salicylic alcohol and the selectivity to products
depending on contact time, at a temperature of 300 °C. It can be
seen that at a relatively high contact time the products were phe-
nol, o-cresol, and 2,6-xylenol. At a low contact time, however, the
main products were salicylic aldehyde, 2-methoxymethylphenol,
and o-cresol.
The selectivity extrapolated to nil contact time was likewise
higher than zero for o-cresol, and then increased along with the
increase in contact time. This means that various routes contribute
this detail will be discussed later. At temperatures over 350 °C,
the formation of deoxygenated aromatics – i.e. benzene, toluene,
and o-xylene – was also observed, likely obtained by the deoxy-
genation of phenol, o-cresol, and 2,6-xylenol, respectively (this
detail will also be discussed later); at 450 °C these compounds
were the prevailing ones. Deoxygenation was likely facilitated by
the presence of considerable amounts of hydrogen, formed by
methanol dehydrogenation. Fig. 3 (below) shows the yields of
2 4 2
methanol decomposition products: CO , CO, CH , and H , the for-
mation of which increased when the temperature was raised.
Fig. 3 clearly shows that the formation of ring-methylated com-
pounds began when methanol started to decompose, which indi-
cates that the dehydrogenation of methanol generates both the
species responsible for the attack onto the aromatic ring, and the
precursor of the light decomposition compounds.
When formaldehyde was used as the methylating agent (Fig. 4),
the conversion of phenol was very significant at low temperatures
also. The products obtained were the same as those obtained with
methanol, but with an higher selectivity to 2,6-xylenol. At a lower
conversion
salicylic
aldehyde
and
2-hydroxy-3-
methylbenzaldehyde also formed, with selectivity lower than 3%;
the presence of these compounds suggests that the formation of
both o-cresol and 2,6-xylenol occurs via a multistep cascade mech-
anism, as already suggested previously. Also in this case, deoxy-
genated compounds were the predominant products at high
temperatures.
Fig. 5. Selectivity to o-cresol, salicylic alcohol, salicylic aldehyde, 2-
methoxymethylphenol, and 2,6-xylenol observed at very low phenol conversion
(contact time < 0.01 s, T 300 °C).

T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
451
Fig. 7. Conversion of salicylic alcohol, molar selectivity of o-cresol, salicylic
aldehyde, salicylic alcohol, and 2-methoxymethylphenol.
Fig. 6. Conversion of salicylic alcohol and methanol based on contact time.
Temperature 300 °C. Symbols: (r) Salicylic alcohol conversion; (N) Selectivity of
o-cresol; (d) Selectivity of 2,6-xylenol; (j) Selectivity of salicylic aldehyde; (})
Selectivity of 2-methoxymethylphenol; ( ) Selectivity of phenol.
to the formation of this compound: (i) a direct route from salicylic
alcohol (or a route involving a very reactive intermediate which
reacts immediately before desorbing into the gas-phase), and (ii)
consecutive reactions occurring on salicylic aldehyde and 2-
methyoxymethylphenol; as a matter of fact, the selectivity of both
compounds decreased when the conversion was raised, with a con-
comitant increase in selectivity to o-cresol. The initial selectivity to
Scheme 3. Direct disproportionation of salicylic alcohol.
In order to better understand the influence of methanol on the
transformation of salicylic alcohol, we conducted a test that con-
sisted of feeding a solution of salicylic alcohol in water, without
methanol. The results (Fig. 7) show that salicylic alcohol can be
converted into salicylic aldehyde and o-cresol even without
methanol. For comparison, a solution containing salicylic alcohol,
water, and methanol was also fed.
These results suggest that the formation of o-cresol might occur
by a direct disproportionation reaction between two molecules of
salicylic alcohol (Scheme 3). In fact, the formation of the two com-
pounds was almost equimolar. A slightly higher formation of sali-
cylic aldehyde can be attributed to the dehydrogenation of the
2
-methoxymethylphenol and salicylic aldehyde was clearly higher
than zero, which indicates that both compounds form directly from
salicylic alcohol. Therefore, these experiments demonstrate that in
addition to the reactions shown in Scheme 1, another reaction may
occur starting from salicylic alcohol, i.e. the formation of ether by a
reaction with methanol, followed by ether decomposition; the
complete set of possible reactions is shown in Scheme 2. On the
other hand, this does not mean that all of these reactions con-
tribute equally to o-cresol formation during phenol methylation.
Scheme 2. Updated summary of the hypothesis for the transformation of the intermediately formed salicylic alcohol into o-cresol during phenol methylation with MgO-
based catalysts.

4
52
T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
At this point, it is important to note the following:
1
2
. The transformation of salicylic aldehyde into phenol cannot
occur by simple decarbonylation; in fact, if this were the case,
the same reaction should occur even without methanol in the
feed. Instead, no reaction was observed in this case. For the
same reason, a Cannizzaro-type disproportionation of the alde-
hyde into salicylic alcohol and salicylic acid, and decarboxyla-
tion of the latter into phenol, is not possible.
. The reduction of salicylic aldehyde might occur either by reac-
tion with the reducing agent (Scheme 2), or by formation of
another intermediate, 2-methoxymethylphenol, as shown in
Scheme 5. The latter might decompose into phenol or o-
cresol, which are the main products obtained from salicylic
aldehyde. An objection might be raised that 2-
methoxymethylphenol was not observed in experiments con-
ducted with salicylic aldehyde (Fig. 8). As will be shown later,
however, under the conditions used, 2-methoxymethylphenol,
even if formed, would be very quickly transformed into consec-
utive compounds.
Scheme 4. H-transfer between methanol and salicylic alcohol.
alcohol. The same type of reaction was reported to occur on etha-
nol with disproportionation to ethane and acetaldehyde, catalysed
by reduced Vanadium oxide [30]. Also, benzyl alcohol dispropor-
tionates into toluene and benzoic aldehyde [31].
It is also worth noting that when methanol is not present, 2-
methoxymethylphenol is not formed; that confirms that the latter
is originated by reaction between methanol and salicylic alcohol
In order to confirm the possible role of 2-methoxymethylphenol
(Scheme 2).
as a key intermediate, we conducted experiments feeding this
compound, in the absence of methanol. The results of these exper-
iments are shown in Fig. 9, plotting the effect of temperature on
The same disproportionation might occur more easily between
salicylic alcohol and methanol (Scheme 4), due to both the large
surplus of the latter, and the unlikely encounter between two
adsorbed molecules of salicylic alcohol; this reaction might also
explain the remarkable formation of methane (Fig. 3, below). In
alternative, the H-transfer might occur with the hydrogenolysis
of benzyl alcohol and the generation of formaldehyde.
conversion
and
yields
obtained
by
feeding
2-
methoxymethylphenol vapours, in a N
2
stream, with the same
molar fraction used for the other experiments described previ-
ously, but without methanol. 2-Methoxymethyphenol proved to
be much more reactive than phenol, salicylic alcohol, and salicylic
aldehyde, and was transformed into both o-cresol (with a co-
3.4. The reactivity of secondary intermediates
2
production of CO and H ) and phenol (with a co-production of
methane and CO) as major products, with small amounts of 2,6-
xylenol (probably formed by transmethylation). Therefore, these
experiments confirmed that the ether might indeed be one reac-
tion intermediate.
In order to further clarify the reaction steps, we conducted some
experiments consisting of feeding salicylic aldehyde, either alone
in which case, no reactivity was observed; therefore, the aldehyde
(
was a stable compound, and did not undergo decarbonylation), or
in the presence of methanol (Fig. 8).
In conclusion, reactivity experiments confirmed that the reac-
tion routes previously hypothesized, and summarised in Scheme 2,
are chemically possible with Mg/Fe/O catalysts. The first interme-
diate, salicylic alcohol, can be transformed into salicylic aldehyde,
which can either be reduced directly to o-cresol, or react with
methanol to generate 2-methoxymethylphenol; the latter can form
even by direct reaction between methanol and salicylic alcohol.
The ether may decompose into o-cresol and phenol. Other possible
routes are the direct hydrogenolysis of salicylic alcohol to o-cresol,
or its dehydration to the o-quinone methide, which is then reduced
to o-cresol.
The main products obtained were o-cresol and phenol, whose
selectivity was not much affected by temperature; minor com-
pounds were 2,6-xylenol and other polymethylated compounds.
These experiments confirm that the reaction between methanol
and salicylic aldehyde generates o-cresol; furthermore, they may
also explain why, in the reaction of phenol methylation, it is
impossible to achieve total phenol conversion (see Figs. 3 and 4);
in fact, phenol is generated in a relatively large amount from sali-
cylic aldehyde.
It was not possible to perform reactivity experiments starting
from the o-quinone methide. Indeed, this compound is extremely
reactive. It can be generated in-situ and made to immediately react
further for the synthesis of various chemicals, but it cannot be iso-
lated [32,33]; in fact, it is not a commercial compound.
For what concerns the catalyst stability, lifetime experiments
reported in a previous work from the same authors [17] showed
that MgO-based catalysts show a slow deactivation during time-
on-stream. Reasons for this were studied in detail, and it was
demonstrated that it was mainly due to the deposition of heavy
compounds, especially in the presence of high formaldehyde con-
centration. The co-feeding of steam helped to limit deactivation.
3.5. The deoxygenation of phenolics
At high temperatures, the predominant products were still
Fig. 8. Conversion of salicylic aldehyde based on temperature. Feed: salicylic
aldehyde/methanol, 1/10 M ratio. Symbols: (r) Salicylic aldehyde conversion; (N)
Selectivity of o-cresol; (d) Selectivity of 2,6-xylenol; (j) Selectivity of deoxy-
genated compounds; (}) Selectivity of ”Others”; ( ) Selectivity of phenol.
deoxygenated compounds (Fig. 8). As for the formation of these
molecules, it might be expected for them to be produced by the
hydrogenolysis of the C-O bond in phenolics, due to the large

T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
453
Scheme 5. The sequence of steps for the transformation of salicylic aldehyde into phenol and o-cresol.
by the electrophilic attach of formaldehyde on the phenolate ring,
we avoid discussing at length the theoretical results on such steps
and the dehydrogenation of methanol to produce formaldehyde,
confining the latter into the Supporting Information for the sake
of future reference. Keeping this in mind, it is straightforward to
select as zero of the energy scale the thermodynamics state
with formaldehyde, phenol and model catalysts non-interacting
among themselves. In this way, it is possible to compare directly
relative energies employing the same energy scale with previous
published works once the energetics involved in the dehydrogena-
tion of methanol into formaldehyde is taken into account
[
+27.0 kcal/mol]. We thus start our discussion focusing on what
happens following the formation of the salicylate anion adsorbed
in the vicinity of a Mg3C site and its dissociated phenolic proton.
In our report on the theoretical results, we shall also discuss the
production of other oxygenated species emerged from the several
experimental tests carried out to investigate specific species as
possible intermediates.
Fig. 9. Conversion of 2-methoxymethylphenol (without methanol) based on
temperature. Symbols: (r) 2-Methoxymethylphenol conversion; (N) Selectivity of
o-cresol; (d) Selectivity of 2,6-xylenol; (j) Selectivity of deoxygenated compounds;
(}) Selectivity of ”Others”; ( ) Selectivity of phenol.
amount of hydrogen available; in alternative, the hydrogenolysis
might occur via H-transfer from methanol. In fact, when phenol
3.7. From salicylic alcohol to o-methyl phenate via o-quinone methide
(
dissolved in water) was fed onto the catalyst, without methanol
at 450 °C, it transformed to benzene with 100% selectivity, but with
only a 7% conversion. An alternative mechanism, however, might
involve the decomposition of phenylformate into either phenol
The anion of the salicylic alcohol, produced as a consequence of
the FA attack onto the phenol ring and the subsequent two-step
proton transfer (see Supporting Information for further details),
may undergo the loss of a hydroxide ion, which remains coordi-
nated onto the MgO cluster, forming an o-quinone methide. This
species is highly electrophilic and, apart from reforming a covalent
bond with the vicinal hydroxide, may sustain an attack from a
nucleophile such as methanol, which can also react as reducing
agent. This may happen via a Meerwein-Ponndorf-Verley-type
(MPV) reaction as previously discussed in [28]; this case, however,
would involve a hydride transfer from the light alcohol to the exo
C@C double bond à la Michael style.
The top panel of Fig. 10 shows the structures and energetics for
the TS leading to the detachment of the hydroxide from the sali-
cylic alcohol, the hydroxide and methide produced, and the latter
species after the desorption of water from the MgO cluster.
The methide species co-adsorbed with methanol, the TS and
products of the MPV process, and the system after the desorption
of o-cresol are instead depicted in the bottom panel of Fig. 10. It
also provides the TS barrier for the hydride transfer, which is only
11.0 kcal/mol. This estimate is slightly lower than the one previ-
ously predicted for the MPV reduction of aliphatic and aromatic
aldehydes, which was suggested to lie in the range 14–18 kcal/mol.
With the estimates for the energetic location of the salicylate
anion (species B provided in the Supporting Information as product
of the formaldehyde attack onto the phenolate), the first of the two
TS is located at ꢁ46.1 kcal/mol below the global zero; the TS for the
hydride transfer is instead located at ꢁ36.9 kcal/mol. Both first
order saddle points are thus located below (hydroxide loss) or at
the same energy (H-transfer) of what predicted for the SN2 process
(ꢁ10.1 kcal/mol from desorbed methanol and phenol, see SI), if one
takes into account the total energy requirement for the formation
(
by decarbonylation) or benzene (by decarboxylation). Phenylfor-
mate might be formed via an attack of formaldehyde onto the O
atom of phenol to produce the hemiacetal, phenoxymethanol,
and dehydrogenation of the latter compound. The experiments
conducted by feeding phenylformate showed that this compound
decomposed completely at 350 °C with a 100% selectivity to phe-
nol, while it was transformed into benzene (55% yield) and phenol
(
45% yield) at 450 °C. The formation of the hemiacetal might also
explain the very high chemo-selectivity of the phenol methylation
process; indeed, the hemiacetal, being very unstable, instead of
being reduced to anisole is immediately deydrogenated to phenyl-
formate and decomposed, while yielding back phenol. Phenylfor-
mate was observed as a product in the reaction between phenol
and methylformate at 250 °C (data not shown), under conditions
in which phenol conversion was less than 10%, and the products
obtained were phenylformate (25% selectivity), anisole (25%), o-
cresol (48%), and 2,6-xylenol (2%). Both anisole and phenylformate,
however, showed a selectivity close to 0% at 350 °C. Nevertheless,
further studies are needed to clarify whether the mechanism of
deoxygenation occurs by direct hydrogenolysis with H
2
or via H-
transfer), or by the intermediate formation of phenylformate.
3.6. Theoretical investigation of methanol and phenol reactivity on
MgO models
Given that experiments and calculations on adsorbed species
20] strongly support the view that phenol methylation proceeds
via the formation of salicylic alcohol, which is in turn produced
[

4
54
T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
Fig. 10. (Top) dehydration of the anion of the salicylic alcohol to generate the o-methide species including water desorption from the MgO crystal; (bottom) MPV-style
hydrogen transfer from methanol to the methide producing the anion of o-cresol, which is subsequently desorbed. Energies are in kcal/mol and refer to the common energy
zero; energies relative to the leftmost species are indicated between square brackets.
of formaldehyde from methanol (+27.0 kcal/mol). These results,
clearly, support the electrophilic attach of formaldehyde and its
consequent reactions as the preferred pathway, at least on ener-
getic grounds.
3.8. Pathway of formation for other possible reaction intermediates:
salicylic aldehyde and 2-methoxymethyl phenol (and other minor
products)
Notice that, beside the anion of o-cresol, also formaldehyde is
produced by the MPV reduction, a key fact as the latter could sus-
tain the alkylation process on its own without requiring the
catalyst to dehydrogenate more methanol. This observation has
an additional important consequence on the energy profile of the
alkylation process involving formaldehyde as, de facto, places
the energy profile of the catalytic cycle even more underneath
the one for the SN2-type process, whose TS is also located
below the energy zero chosen in this work (ꢁ37.1 kcal/mol, see
Fig. 10).
The last system shown in the bottom panel of Fig. 10 provides
an appropriate entry point for a new phenol molecule into the cat-
alytic cycle; the latter species could, in fact, co-adsorb on the Mg3C
site together with formaldehyde from the gas phase while releas-
ing 30.1 kcal/mol of energy and, contemporarily donating a proton
to the crystal. From such state, the two molecules can easily react
via the pathway described above and involving the electrophilic
attack of formaldehyde onto the 2-position of the phenolate anion.
As a final comment with respect to possible alternative mecha-
When only a short contact time (see Fig. 5) is allowed between
the phenol/methanol feed and the catalyst, or salicylic alcohol is
directly fed on the latter with or without methanol (Figs. 6 and
7), oxygenated compounds appear also in the form of salicylic alde-
hyde and 2-methoxymethylphenol, which can also be reaction
intermediates on the way to o-cresol. Their presence provides,
indeed, a robust support to the pathway shown in Fig. 10, as it
can be reconciled easily with the latter. For instance, the formation
of 2-methoxymethyl phenol can be rationalized via the tendency
to undergo a nucleophilic addition à la Michael of the methide spe-
cies, this time with a methoxide generated from the dissociative
adsorption of another methanol molecules (see Fig. 11). The latter
reaction has a TS barrier of only 1.9 kcal/mol and it releases
29.2 kcal/mol of energy; hence, the process is quite facile once
the two molecular species sit close together on the MgO surface,
and it is also substantially biased toward the ether-like species.
This rationalizes also the production in large quantity of 2-
methoxymethylphenol shown in Figs. 5 and 6, which appears fol-
lowing the dehydration of salicylic alcohol to the methide and is
facilitated by the large excess of methanol present in the feed com-
pared to the hydroxide (or water) released as product.
nisms, we notice that one might conceive a direct (S
transfer from gas phase methanol onto the salicylic CH
N
2-style) H-
-OH group,
2
which loses OHA while a CAH bond in methanol is cleaved produc-
ing formaldehyde. Indeed, we were unable to locate the TS
involved in such process despite multiple attempts involving con-
strained optimization conducted freezing the distance between the
hydrogen on methanol and the carbon bearing the OH group in the
salicylic alcohol. De facto, the closest structure to a putative TS we
obtained with this approach (with a constraining force of 0.095
hartree/bohr along the frozen Cꢀ ꢀ ꢀH distance) lied 66.8 kcal/mol
above the non-interacting tautomer B/methanol pair, a quantita-
tive result that strongly suggests the non-competitive nature of
this possible mechanistic step.
Despite this, we haste to point out that equilibrium should be
easily established between the ether and the two dissociated coun-
terparts in the temperature range explored experimentally; this is
quite important as the ether de facto represent a possible sinkhole
for the methide, which, however, could be steadily released back
into the pool of intermediates via the mentioned equilibrium if
the unsaturated ketone is consumed by other processes such as
the MPV reduction discussed above. Interestingly, the latter could
directly involve the just separated moieties if co-adsorbed at short
distance, a process that provides a rationale for the last step pro-
posed in Schemes 4 and 5.

T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
455
Fig. 11. Reactants (left), TS (center) and product (right) of the Michael-style nucleophilic attack by methanol onto the adsorbed methide. Energies are in kcal/mol and refer to
the common energy zero; energies relative to the process reactants species are indicated between square brackets.
In the latter scheme, it was also suggested the possibility of
forming 2-methoxymethylphenol via the emiacetal formed by sal-
icylic aldehyde and methanol, which loses a hydroxide forming a
methoxy-substituted methide; the latter can also be reduced by
methanol via a MPV reaction. Such sequence may, in fact, be com-
petitive with what suggested in Fig. 11, at least due to the assis-
tance provided by the AOAMe group during the loss of OHA.
Turning to the generation of salicylic aldehyde, this appears to
be produced either directly from the phenol/methanol feed
the pure MgO catalyst, the second is worth of further investigation.
Thus, reactants, TS and products of the dehydrogenation of salicylic
alcohol mediated by formaldehyde are shown in Fig. 12, which
indicates that a very low barrier (4.4 kcal/mol) must be sur-
mounted for the process to take place; the transformation is also
quite exoenergetic, releasing 25.7 kcal/mol. This, in turn, suggests
that the oxidation of salicylic alcohol by formaldehyde is made
facile by the conjugation between the forming carbonyl and the
phenyl ring.
(
(
Fig. 5), or by feeding salicylic alcohol with (see Fig. 6) or without
see Fig. 7) methanol on the catalyst. Having already shown how
With specific reference to the fractional amount of salicylic
aldehyde produced as a function of contact time shown in Fig. 6,
it is important to notice that the selectivity with respect to the
aldehyde decreases upon increasing the time, thus suggesting that
the aldehyde is a primary product of salicylic alcohol, and that it
comes primarily from the direct dehydrogenation on the Fe sites
at short times. This is justified by the expectedly low barrier path-
way afforded by the presence of iron ions as dopant, a concept well
supported by the large fractional amount of salicylic aldehyde pro-
duced while feeding salicylic alcohol alone on the catalysts (see
Fig. 7).
salicylic alcohol can be produced from the reaction mixture, we
shall concentrate on the process implicated in Figs. 6 and 7. In
the latter cases, it is possible that either salicylic alcohol dehydro-
genates directly onto the active Fe-doped sites releasing a hydro-
gen molecule (as it is apparent from Fig. 7), or reacting with the
FA produced by the Fe sites from methanol via a MPV reaction
when the latter is present (Fig. 6). While it is clear that the former
process may take place similarly to the production of FA, hence
with a reduced barrier height compared to what predictable for
Fig. 12. MPV-style oxidation of the anion of the salicylic alcohol by FA. Energies are in kcal/mol and refer to the common energy zero; energies relative to the leftmost species
are indicated between square brackets.

4
56
T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
The MPV oxidation of salicylic alcohol by formaldehyde is nev-
what obtained assuming that the reducing agent is methanol (i.e. a
gas phase barrier of 18.5 kcal/mol, see Fig. 13), it clearly emerges
that the disproportionation of salicylic alcohol via methide is, first,
a more facile process than the one involving the formation of the
latter and its reduction by methanol (hence they can compete),
and, second, that its energy requirements are indeed quite low.
This conclusion easily obtained by adding directly the change in
barrier height (ꢁ7.5 kcal/mol) induced by running the hydrogen
transfer between methide and methanol onto MgO (see Fig. 10)
compared to the gas phase situation to the barrier for the MPV
reaction involving methide and salicylic alcohol; this leads to an
estimated barrier of 7.8 kcal/mol. Whether or not the reaction in
Scheme 1 is, in practice, a relevant source of o-cresol may however
depend on the amount of methanol in the feed, which may
increase the reaction rate of the process exploiting it as hydrogen
source.
ertheless important as it may provide a contribution to quantita-
tively justify the low contact time selectivity toward salicylic
aldehyde and 2-hydroxy-3-methylbenzaldehyde highlighted while
discussing the results in Fig. 4; these were obtained by feeding a
phenol/ formaldehyde mixture on the catalyst. The former may
emerge as secondary product following the formation of 2-
methoxymethylphenol via both MPV or Fe-mediated oxidation;
the second, instead, requires, first, the formation of o-cresol, which
successively reacts with formaldehyde to produce, initially, 2-
methyl-6-hydroxymethyl phenol and, successively, the aldehydic
counterpart of the latter. With the large amount of formaldehyde
available to react with phenol and its alkylated/hydroxyalkylated
product in the process shown in Fig. 4, it can be envisaged that a
predominant role in oxidizing compounds may be played by the
MPV reaction rather than by the catalyst directly.
In principle, the disproportionation of salicylic alcohol might
take place also via a mechanism involving radicals [34,35], which
requires the initial dissociation either of the C-OH bond or the C-
3
.9. Alternative pathways for the production of o-cresol involving
established intermediates
2
H one in the ACH OH moiety of salicylic alcohol. In gas phase,
The appearance of o-cresol as a product obtained while feeding
only salicylic alcohol and water onto the catalyst (Fig. 7) suggest
that alternative paths may be available to obtain the reaction prod-
uct. The simplest mechanism we can suggest involves the loss of
water from the alcohol to form the methide, which subsequently
reacts with a second alcohol molecule via a MPV-style hydride
transfer to form the methylated phenol and salicylic aldehyde; in
other words, salicylic alcohol substitutes methanol as reducing
agent in what can be, de facto, considered its disproportionation.
Notice that this mechanism rationalizes the relative selectivity
these require, respectively, 83.3 and 88.5 kcal/mol to dissociate
homolytically at the B3LYP/6-31++g(d,p) level, values that are
much higher than what requested to dissociate heterolytically
(47.9 kcal/mol) and form methide and water in the same condi-
tions. Albeit one would expect the adsorption on MgO to some-
what reduce the quoted radical barriers, as it does for the
formation of methide, it seems highly unlikely that the eventual
decrease would make the barriers comparable with the one for
the heterolytic dissociation of the C-OH bond on the crystal. Con-
tributing to this conclusion, the products of the radical dissociation
of the C-OH bond adsorbed on MgO would represent an electroni-
cally excited state for the system, which might exist in both singlet
and triplet states. To test the latter hypothesis, a TD-DFT calcula-
tion was carried out on the equilibrium geometry the product of
the alcoholic hydroxide loss from salicylic alcohol to gauge the
energy requested for transferring the excess electron borne by
(
1.5) between salicylic aldehyde and o-cresol seen in the left part
of Fig. 7, a third of which descends from the direct dehydrogena-
tion of the alcohol. The relative selectivity is reduced in presence
of a large excess of methanol, which can both lead to the formation
of 2-methoxymethylphenol and compete with the aromatic alco-
hol for the dehydrogenation sites.
Interestingly, the TS for the hydride transfer between methide
and salicylic alcohol requires that a barrier of 15.3 kcal/mol is sur-
mounted when the process takes place in gas phase (see Fig. 13 for
the structure and energetics of the TS); comparing this result with
ꢁ
OH , which is partially stabilized by the MgO ions and the proton
dissociated from the phenolic hydroxyl, back onto the exo methyli-
dene of the methide species. From such calculation, it emerged that
mono-electronic excitations from configurations involving orbitals
Fig. 13. TS’s for the MPV-style reduction of methide by (left) salicylic alcohol and (right) methanol. Energies are in kcal/mol and are relative to the separated reactants in both
cases.

T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
457
Fig. 14. Occupied and virtual molecular orbitals involved in the electron-transferring transition from the ground state that generates a diradical (open shell singlet) state
borne by the methide and oxidryl moieties. The occupied molecular orbitals shown are the ones that bear the negative charge of the hydroxide.
bearing the negative charge on the hydroxide to describe the
ground state and virtual orbitals with a substantial overlap with
the methylidene group requires at least 48.8 kcal/mol to take place
competitive with respect to the one involving the methide species.
De facto, this conclusions rule out also the possibility that 2-
2 3
methoxymethylphenol could rearrange its Ph-CH -OCH moiety
(
see Fig. 14). The direct consequences of such results should thus
producing formaldehyde and Ph-CH (see Scheme 3) following
3
be that the products of the homolytic dissociation of the CH OH
2
the concerted four-center mechanism taking place in the dimethyl
ether case, and leave the loss of methoxide followed by MPV as the
only viable path to generate the methylated phenol from 2-
methoxymethylphenol.
In conclusion, the formation of o-cresol following the feeding of
only salicylic alcohol and water on the catalyst is suggested to pro-
ceed via the heterolytic cleavage of the CAOH bond leading to
methide, which is successively reduced by another salicylic alcohol
molecule to produce the alkylated compound plus salicylic
aldehyde.
on the salicylic alcohol are highly more energetic that ionic coun-
terparts, and that the TS barrier to be surmounted for their produc-
tion would be vastly higher than the 29.2 kcal/mol required by the
heterolytic dissociation.
As a final alternative pathway for the disproportionation of sal-
icylic alcohol, one may also conceive the formation of a symmetric
2
ether following the condensation of the CH OH group of two mole-
cules; de facto, this would pass through the formation of a methide,
which successively adds a molecule of salicylic alcohol with barri-
ers similar to what discussed for the formation of methoxymethyl
phenol (vide supra). The ether could successively be pyrolyzed
leading to o-cresol and salicylic aldehyde via a concerted radical
mechanism similar to the unimolecular pathway exploited by
dimethyl ether [34,36], diethyl ether [37], and ethyl vinyl ether
Scheme 6 summarises the reaction network, while scheme 7
provides the energy details of the reaction mechanism.
3.10. The nature of the active sites, and the relationship between active
sites and mechanism steps
[
38] to produce an aldehyde molecule and an alkane/alkene. In
the gas phase, dimethyl ether requires 93.8 kcal/mol to surmount
the TS barrier and produce methane and formaldehyde [34]; the
same reaction leading to acetaldehyde and ethane from diethyl
ether in gas phase seems to be only five times faster than for
dimethyl ether [36,37] at 400 C. This happens even in supercritical
water, which would be expected to stabilize polar transition states
but it is found without impact on the absolute or relative values of
the kinetic constants; we take this as suggesting that (i) the barrier
for the suggested ether decomposition remains high even when
more substituted ethers are involved, and (ii) that even markedly
polar environments do not provide a stabilization to the TS
involved sufficiently strong to make such decomposition channel
Catalysts based on MgO and Mg/Fe/O were characterised in pre-
vious works [19–22]; their main features can be summarised as
follows:
(a) Before the reaction took place, Mg oxide was strongly car-
bonated and hydrated (hydroxymagnesite, Mg
5 3 4
(CO )
(OH) 4H O, JCPDS file 25-0513, and brucite Mg(OH)
2
2
2
, 44-
1482), whereas after reaction the pattern did correspond
to that of MgO (periclase, 04-0829) [22]. The IR spectra
recorded after co-adsorption of phenol and formaldehyde
demonstrated that phenol develops a much stronger interac-
tion with the MgO surface than the in-situ generated

4
58
T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
Also worth a note, it is the fact that the reaction between
formaldehyde and phenol always generated the same kinds
of products regardless of the catalyst type (acidic or basic),
whereas the nature of products was very different depend-
ing on the catalyst used when feeding methanol instead of
the aldehyde. In addition, we recall that in situ IR spec-
troscopy, when supported by computational studies of
MgO-sorbed molecules, revealed that adsorbed and acti-
vated phenol (i.e. the phenolate species) and formaldehyde
locate preferentially at the corner sites of MgO (three-co-
ordinated Mg atoms, Mg3C). It was also demonstrated that
o-cresol formation starts with the reaction between phenol
and formaldehyde to give salicylic alcohol [20].
(
b) In-situ DRIFT spectroscopy revealed that any adsorbed phe-
nolic intermediate can be hardly isolated [20]. In fact, when
the experiment was carried out by first adsorbing phenol
and then formaldehyde, both at room temperature, and then
by increasing temperature, only the bands attributable to
the stable o-cresol were shown. When instead formaldehyde
was adsorbed first, bands attributed to formaldehyde disap-
peared when the temperature was increased, with the con-
comitant development of bands attributable to adsorbed
salicylaldehyde; these features disappeared at 300 ◦C, while
bands attributed to o-cresol developed. This was probably
due to the high reactivity of salicylic alcohol and of the cor-
responding methide formed by dehydration.
(
c) The XRD pattern of the fresh Mg/Fe/O catalyst was not well
resolved, indicating the presence of a large fraction of amor-
Scheme 6. The overall reaction network in phenol methylation with methanol over
MgO-based catalysts.
phous compound; reflections corresponding to both MgFe
2
-
O
4
(36-0398) and Fe hematite (13-0534) were present.
2 3
O
After reaction, the reflections attributable to hematite were
no longer detectable; instead, those corresponding to either
formaldehyde. Moreover, the reaction between phenol and
formaldehyde required the preadsorption and activation of
phenols with the development of phenolate species; as to
formaldehyde, the latter needs not to be surface adsorbed
to electrophilically attack the activated aromatic ring [21].
3 4 2 4
magnetite Fe O (19-0629) or the spinel phase MgFe O (the
two patterns are almost coincident) were clearly detectable
[22]. Moreover, catalysts characterization carried out with
CO
2
-TPD indicated that the hosting of a cation having a high
2
ꢁ
electronegativity decreases the basicity of bridging O ions
Scheme 7. The reaction mechanism in phenol methylation with methanol over MgO catalysts (details are given starting from the first intermediate, salicylic alcohol; for
details on the first step of phenol hydroxyalkylation with the in-situ generated formaldehyde, see [20]).

T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
459
in mixed systems with respect to pure MgO, in agreement
with previous literature findings [40–42]. Thus, the Mg/Fe
mixed oxides showed a high population of medium strength
sites, the latter being considerably less basic than in the MgO
case, where a predominance of strongly basic sites is
recorded. Focusing on Fe3 ions, the latter were easily
reduced by CO even in mild conditions. Indeed, the main fea-
ture Mg/Fe/O catalysts is the pronounced redox behaviour,
in the more active and more stable Mg/Fe/O system, Fe ions are
extremely active in methanol dehydrogenation, and the in-situ
generated formaldehyde attacks the phenolate species. The latter
have been formed by adsorption of phenol over basic sites (in
the case of MgO, most likely corner sites). Worth a note, the
rate-determining step of the process is methanol dehydrogenation,
after which formaldehyde reacts very rapidly even without the
need of further activation thanks to its very strong electrophilicity.
After the formation as intermediate, salicylic alcohol quickly dehy-
drates (this step probably occurring due to thermal activation and,
without the need of a catalyst); the final step is the H-transfer to
methide that hydrogenates the conjugated electronic system.
Again, basic sites (either in MgO or in Mg/Fe/O) catalyse this last
step.
+
associated with the couple Fe3 /Fe . Besides, Fe
+
2+
O presents
2 3
a typical basic behaviour with respect to phenolic com-
pounds that facilitates the formation of phenolate species
and leads to high chemo- and regio-selectivity in phenol
methylation [22]. At the same time, it has been shown that,
2 3
the pronounced redox activity in Fe O and in Fe-rich Mg/Fe-
mixed oxides enhances methanol dehydrogenation to
formaldehyde and its further decomposition into light com-
2 4
pounds (CO , CO and CH ) [19]. Finally, Mg/Fe/O had a med-
References
ium strength surface acidity, with the predominance of
Lewis sites of moderate strength. However, the dehydro-
genation of methanol largely prevailed over its acid-type
activation despite the Lewis acidity experimentally
observed, and the behaviour observed was that typical of a
basic/dehydrogenating reactivity [39]. Overall, the catalytic
[
1] C.S. Gopinath, T. Mathew, N.R. Shiju, K. Sreekumar, B.S. Rao, Cu-Co synergism
in Cu<inf>1-x</inf>Co<inf>x</inf>Fe<inf>2</inf>O<inf>4</inf>-catalysis and
XPS aspects, J. Catal. 210 (2002) 405–417.
[
2] K.-T. Li, I. Wang, K.-R. Chang, Methylation of phenol to 2,6-dimethylphenol on
a manganese oxide catalyst, Ind. Eng. Chem. Res. 32 (1993) 1007–1011.
3] J. Wolska, K. Przepiera, H. Grabowska, A. Przepiera, M. Jabło n´ ski, R.
[
performance of Fe
2
O
3
was not much different from that
Klimkiewicz, ZnFe<inf>2</inf>O<inf>4</inf>as a new catalyst in the C-
one of the Mg/Fe/O catalyst; the latter, however, was charac-
terised by a greater and more stable surface area.
methylation of phenol, Res. Chem. Intermed. 34 (2008) 43–51.
4] A.R. Gandhe, J.B. Fernandes, S. Varma, N.M. Gupta, TiO<inf>2</inf>: as a
versatile catalyst for the ortho-selective methylation of phenol, J. Mol. Catal. A
Chem. 238 (2005) 63–71.
[
3
.11. Conclusions: the mechanism for the formation of o-cresol in
[5] K.M. Malshe, P.T. Patil, S.B. Umbarkar, M.K. Dongare, Selective C-methylation of
phenol with methanol over borate zirconia solid catalyst, J. Mol. Catal. A Chem.
phenol methylation with Mg/Fe/O catalyst
212 (2004) 337–344.
[
6] A.S. Reddy, C.S. Gopinath, S. Chilukuri, Selective ortho-methylation of phenol
with methanol over copper manganese mixed-oxide spinel catalysts, J. Catal.
In summary, by combining information on reactivity experi-
ments and DFT calculations, we can draw the following
conclusions:
2
43 (2006) 278–291.
[
7] M.E. Sad, C.L. Padró, C.R. Apesteguía, Selective synthesis of p-cresol by
methylation of phenol, Appl. Catal. A Gen. 342 (2008) 40–48.
[
8] J. Kaspi, G.A. Olah, Heterogeneous catalysis by solid superacids. 4. Methylation
of phenols with methyl alcohol and the rearrangement of anisole and
methylanisoles over a perfluorinated resinsulfonic acid (Nafion-H) catalyst, J.
Org. Chem. 43 (1978) 3142–3147.
–
–
–
The very first intermediate in phenol methylation is salicylic
alcohol, in its deprotonated form, which is formed by the reac-
tion between the adsorbed phenolate and the in-situ generated
formaldehyde;
The direct hydrogenolysis of salicylic alcohol via H-transfer is
much less facilitated energy-wise than its dehydration to o-
quinone methide; indeed, the latter is the most preferred route,
energy-wise.
Alcohol may also dehydrogenate to salicylic aldehyde which,
however, due to the large surplus of methanol, is easily reduced
back to alcohol. A dehydrogenation-hydrogenation equilibrium
is rapidly established which, however, shifts towards alcohol
due to its consecutive dehydration to the methide.
Methide is the second key intermediate in the reaction. It may
undergo different transformations: (a) MPV reduction by
methanol to o-cresol; (b) MPV reduction by salicylic alcohol to
o-cresol; however, this reaction probably occurs only in the
absence of methanol; (c) Reaction with methanol to produce
[9] S. Velu, C.S. Swamy, Effect of substitution of Fe3+/Cr3+on the alkylation of
phenol with methanol over magnesium-aluminium calcined hydrotalcite,
Appl. Catal. A Gen. 162 (1997) 81–91.
[
10] K. Sreekumar, S. Sugunan, Ferrospinels based on Co and Ni prepared via a low
temperature route as efficient catalysts for the selective synthesis of o-cresol
and 2,6-xylenol from phenol and methanol, J. Mol. Catal. A Chem. 185 (2002)
259–268.
[
11] K.G. Bhattacharyya, A.K. Talukdar, P. Das, S. Sivasanker, Al-MCM-41 catalysed
alkylation of phenol with methanol, J. Mol. Catal. A Chem. 197 (2003) 255–
262.
[
12] S. Sato, K. Koizumi, F. Nozaki, Ortho-selective methylation of phenol catalyzed
by CeO<inf>2</inf>-MgO prepared by citrate process, J. Catal. 178 (1998) 264–
274.
[13] T. Mathew, M. Vijayaraj, S. Pai, B.B. Tope, S.G. Hegde, B.S. Rao, et al., A
mechanistic approach to phenol methylation on Cu<inf>1-x</inf>Co<inf>x</
inf>Fe<inf>2</inf>O<inf>4</inf>: FTIR study, J. Catal. 227 (2004) 175–185.
–
[
[
[
[
14] M.E. Sad, C.L. Padró, C.R. Apesteguía, Synthesis of cresols by alkylation of
phenol with methanol on solid acids, Catal. Today 133–135 (2008) 720–728.
15] A.R. Gandhe, J.B. Fernandes, Methylation of phenol over Degussa P25 TiO 2, J.
Mol. Catal. A Chem. 226 (2005) 171–177.
16] B.V. Vora, J.A. Kocal, P.T. Barger, R.J. Schmidt, J.A. Johnson, Alkylation, Kirk-
Othmer Encycl. Chem Technol. (2003).
2
-methoxymethylphenol; the latter rapidly decomposes into
17] T. Tabanelli, S. Cocchi, B. Gumina, L. Izzo, M. Mella, S. Passeri, et al., Mg/Ga
mixed-oxide catalysts for phenol methylation: outstanding performance in
o-cresol. This step is both thermodynamically and kinetically
facilitated, due to the large availability of methanol. It should
be noted, however, that in this case, also, an equilibrium
between the forward reaction of 2-methoxymethylphenol for-
mation and the retrodissociation into the methide and methoxy
species may be established, where the two latter compounds
react via MPV to yield o-cresol. Therefore, it may be concluded
that the reaction 4a is most likely the one that offers the great-
est contribution to o-cresol formation.
2,4,6-trimethylphenol synthesis with co-feeding of water, Appl. Catal. A Gen.
552 (2018).
[
18] N. Ballarini, F. Cavani, S. Passeri, L. Pesaresi, A.F. Lee, K. Wilson, Phenol
methylation over nanoparticulate CoFe2O4 inverse spinel catalysts: the effect
of morphology on catalytic performance, Appl. Catal. A Gen. 366 (2009) 184–
192.
[19] V. Crocellà, G. Cerrato, G. Magnacca, C. Morterra, F. Cavani, S. Cocchi, et al., 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, J.
Catal. 270 (2010) 125–135.
[
20] F. Cavani, L. Maselli, S. Passeri, J.A. Lercher, Catalytic methylation of phenol on
MgO – surface chemistry and mechanism, J. Catal. 269 (2010) 340–350.
21] N. Ballarini, F. Cavani, L. Maselli, S. Passeri, S. Rovinetti, Mechanistic studies of
the role of formaldehyde in the gas-phase methylation of phenol, J. Catal. 256
(2008) 215–225.
Regarding the nature of the active sites involved, the reaction
[
2 3
occurs on either MgO or Fe O (which are the basic components
of the two different industrial catalysts for phenol methylation);

4
60
T. Tabanelli et al. / Journal of Catalysis 370 (2019) 447–460
[
22] N. Ballarini, F. Cavani, L. Maselli, A. Montaletti, S. Passeri, D. Scagliarini, et al.,
The transformations involving methanol in the acid- and base-catalyzed gas-
phase methylation of phenol, J. Catal. 251 (2007) 423–436.
23] M. Bolognini, F. Cavani, D. Scagliarini, C. Flego, C. Perego, M. Saba,
Heterogeneous basic catalysts as alternatives to homogeneous catalysts:
reactivity of Mg/Al mixed oxides in the alkylation of m-cresol with methanol,
Catal. Today 75 (2002) 103–111.
[34] J.J. Nash, J.S. Francisco, Unimolecular decomposition pathways of dimethyl
ether: an ab initio study, J. Phys. Chem. A 102 (1998) 236–241.
[35] C.Y. Lin, M.C. Lin, Thermal decomposition of methyl phenyl ether in shock
waves: The kinetics of phenoxy radical reactions, J. Phys. Chem. 90 (1986)
425–431.
[36] Y. Nagai, N. Matubayasi, M. Nakahara, Mechanisms and kinetics of
noncatalytic ether reaction in supercritical water. 2. Proton-transferred
fragmentation of dimethyl ether to formaldehyde in competition with
hydrolysis, J. Phys. Chem. A 109 (2005) 3558–3564.
[
[
[
24] A.D. Becke, J. Chem. Phys. 98 (1993) 5648–5652.
25] P.J. Stephens, F.J. Devlin, C.F. Chabalowski, M.J. Frisch, J. Phys. Chem. 98 (1994)
1
1623–11627.
[37] Y. Nagal, N. Matubayasi, M. Nakahara, Mechanisms and kinetics of
noncatalytic ether reaction in supercritical water. 1. Proton-transferred
fragmentation of diethyl ether to acetaldehyde in competition with
hydrolysis, J. Phys. Chem. A 109 (2005) 3550–3557.
[38] K. Shimofuji, K. Saito, A. Imamura, Unimolecular thermal decomposition of
ethyl vinyl ether and the consecutive thermal reaction of the intermediary
product acetaldehyde: shock wave experiment and ab initio calculation, J.
Phys. Chem. 95 (1991) 155–165.
[
26] A.G. Pelmenschikov, G. Morosi, A. Gamba, S. Coluccia, G. Martra, E.A. Paukshtis,
J. Phys. Chem. 100 (1996) 5011.
27] A. Chieregato, J. Velasquez Ochoa, C. Bandinelli, G. Fornasari, F. Cavani, M.
Mella, ChemSusChem 8 (2015).
28] T. Pasini, A. Lolli, S. Albonetti, F. Cavani, M. Mella, J. Catal. 317 (2014) 206–219.
29] A.G. Pelmenschikov, G. Morosi, A. Gamba, S. Coluccia, J. Phys. Chem. 99 (1995)
[
[
[
1
5018–15022.
[
30] Y. Nakamura, T. Murayama, W. Ueda, Reduced vanadium and molybdenum
oxides catalyze the equivalent formation of ethane and acetaldehyde from
ethanol, ChemCatChem 6 (2014) 741–744.
31] E. Nowicka, J.P. Hofmann, S.F. Parker, M. Sankar, G.M. Lari, S.A. Kondrat, et al.,
In situ spectroscopic investigation of oxidative dehydrogenation and
disproportionation of benzyl alcohol, PCCP 15 (2013) 12147.
[39] V. Crocellà, G. Cerrato, G. Magnacca, C. Morterra, F. Cavani, L. Maselli, S. Passeri,
Dalton Trans. 39 (2010) 8527–8537.
[40] J. Sanchez Valente, F. Figueras, M. Gravelle, P. Kumbhar, J. Lopez, J.P. Besse, J.
Catal. 189 (2000) 370.
[41] T. Sato, T. Wakahayash, M. Shimada, Ind. Eng. Chem. Prod. Res. Devel. 25
(1986) 89.
[
[
32] L. Caruana, M. Fochi, L. Bernardi, The emergence of quinone methides in
asymmetric organocatalysis, Molecules 20 (2015) 11733–11764.
[42] D. Tichit, M.H. Lhouty, A. Guida, B.H. Chiche, F. Figueras, A. Auroux, D. Bartalini,
E. Garrone, J. Catal. 151 (1995) 50.
[
33] T.P. Pathak, M.S. Sigman, Applications of ortho-Quinone methide
intermediates in catalysis and asymmetric synthesis, J. Org. Chem. 76 (2011)
9
210–9215.