Glycerol dehydration to hydroxyacetone in gas phase over copper supported on magnesium oxide (hydroxide) fluoride catalysts

Article abstract of DOI:10.1016/j.apcata.2018.03.022

The dehydration of glycerol to hydroxyacetone was studied over copper-based catalysts using magnesium oxide (hydroxide) fluoride with various F/Mg ratio as support of copper. After calcination at 350 °C, the incorporation of copper, mainly at + II oxidation state, into the support lattice was observed for MgO and MgF(OH) while, copper was stabilized as Cu⁺¹ at the surface of Cu-MgF₂. The reaction of dehydration was performed using a mixture of glycerol and water (80%wt of glycerol), in gas phase at 260 °C. Cu-MgF₂ was the most active catalyst with a yield in hydroxyacetone of 45.5%, while the catalytic activity was very low for Cu-MgF(OH) and Cu-MgO (yield in HA < 10%). Moreover, the performances obtained for Cu-MgF₂ were higher than those obtained with La₂CuO₄, a reference catalyst. After four hours of reaction, Cu-MgF₂ was not significantly modified, while for the two other catalysts, Cu²⁺ initially present was reduced into metallic copper. The results obtained revealed that the basic properties of the catalysts did not govern the reaction of dehydration of glycerol into HA. The best catalyst (Cu-MgF₂) was the one possessing the higher amount of Lewis acid sites, and stabilizing copper at + 1 oxidation state.

Full text of DOI:10.1016/j.apcata.2018.03.022

Accepted Manuscript
Title: Glycerol dehydration to hydroxyacetone in gas phase
over copper supported on magnesium oxide (hydroxide)
fluoride catalysts
´
´
Authors: Stephane Celerier, Sophie Morisset, Isabelle
Batonneau-Gener, Thomas Belin, Khaled Younes, Catherine
Batiot-Dupeyrat
PII:
S0926-860X(18)30141-8
DOI:
Reference:
https://doi.org/10.1016/j.apcata.2018.03.022
APCATA 16596
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
26-1-2018
17-3-2018
21-3-2018
´
Please cite this article as: Celerier S, Morisset S, Batonneau-Gener I, Belin T, Younes K,
Batiot-Dupeyrat C, Glycerol dehydration to hydroxyacetone in gas phase over copper
supported on magnesium oxide (hydroxide) fluoride catalysts, Applied Catalysis A,
General (2010), https://doi.org/10.1016/j.apcata.2018.03.022
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Glycerol dehydration to hydroxyacetone in gas phase over copper supported on
magnesium oxide (hydroxide) fluoride catalysts
Stéphane Célerier, Sophie Morisset, Isabelle Batonneau-Gener, Thomas Belin, Khaled
Younes and Catherine Batiot-Dupeyrat*
IC2MP, UMR CNRS 7285, ENSIP, Université de Poitiers, 1 rue Marcel Doré, 86022 Poitiers,
Graphical abstract
Cu-MgF2
80 wt% glycerol in water
OH
HO
OH
Gas phase
Atmospheric pressure
T= 260°C
Yield =45.5%
Highlights




Highlignts
High yields of hydroxyacetone was achieved in gas phase over CuOx-MgF2 at 260°C
The performances obtained with Cu-MgF2 were higher than those obtained with La2CuO4
Copper is stabilized as Cu+1 in Cu-MgF2 during reaction
*Corresponding author: catherine.batiot.dupeyrat@univ-poitiers.fr, Phone: +33549453898,
Abstract:
The dehydration of glycerol to hydroxyacetone was studied over copper-based catalysts using
magnesium oxide (hydroxide) fluoride with various F/Mg ratio as support of copper. After
calcination at 350°C, the incorporation of copper, mainly at + II oxidation state, into the support
lattice was observed for MgO and MgF(OH) while, copper was stabilized as Cu+1 at the surface
of Cu-MgF2. The reaction of dehydration was performed using a mixture of glycerol and water
(80%wt of glycerol), in gas phase at 260°C. Cu-MgF2 was the most active catalyst with a yield
in hydroxyacetone of 45.5%, while the catalytic activity was very low for Cu-MgF(OH) and
Cu-MgO (yield in HA <10%). Moreover, the performances obtained for Cu-MgF2 were higher
than those obtained with La2CuO4, a reference catalyst. After four hours of reaction, Cu-MgF2
was not significantly modified, while for the two other catalysts, Cu2+ initially present was
reduced into metallic copper. The results obtained revealed that the basic properties of the
catalysts did not govern the reaction of dehydration of glycerol into HA. The best catalyst (Cu-
1

MgF2) was the one possessing the higher amount of Lewis acid sites, and stabilizing copper at
+1 oxidation state.
2

Introduction
Glycerol, the main byproduct obtained from biodiesel industry, is a very promising
platform building blocks for fuels and chemicals production [1]. Many applications of glycerol
valorization were reported in the literature, such as hydrogenolysis of glycerol to propanediols
[2-4], dehydration to acrolein [5-8] and reforming to hydrogen or syngas [9]. Glycerol
conversion into acrolein was successfully performed over solid acid catalysts such as heteropoly
acids [10] or zeolites [11]. The formation of acrolein is often accompanied by the presence of
hydroxyacetone.
Hydroxyacetone (HA) is an interesting chemical product used as flavour in food industry, dyes
or additive in cosmetics. It is also an intermediate in the production of valuable compounds
such as propyleneglycol, acrolein or propionaldehyde. Therefore there are needs to develop
active catalysts to produce selectively HA under mild experimental conditions, in a fixed bed
continuous flow process at atmospheric pressure. Indeed, heterogeneous catalytic processes
allow to avoid commonly encountered drawbacks of homogeneous catalytic processes
including the difficulty of catalysts separation and problems of waste disposal.
Hydroxyacetone can be obtained from glycerol dehydration in gas phase, over catalysts
containing Lewis acid sites such as MOx-Al2O3-PO4 [12] or Zn-Cr oxides [13] or catalysts
containing basic sites such as NiCo2O4 [14] while over strong Bronsted acid sites the formation
of acrolein is favoured [15]. High selectivity to HA was also achieved using the 5%Na doped
CeO2 basic catalyst at 350°C, but a poor stability was observed with a rapid deactivation with
time on stream [16].
Copper-based catalysts were widely used to perform the hydrogenolysis of glycerol to
propanediols due to their lower price and higher resistance to poisoning than noble metals [17].
Copper was also preferred to nickel or cobalt due to its lower activity for C-C bond cleavage.
Layered double hydroxide (LDH) supported Cu catalysts were successfully used in the
3

hydrogenolysis of glycerol in aqueous solution, under hydrogen pressure. The high activity was
attributed to the strong basicity of the catalyst [18]. For the hydrogenolysis of glycerol, copper
is used in its reduced form (Cu°) in order to favour the hydrogenation step which follows the
dehydration one at the surface of the oxide support. Moreover it is also proposed that the Cu
metallic site is involved in the dehydrogenation of glycerol to glyceraldehyde, followed by its
dehydration and subsequent hydrogenation to HA [1]. Copper based catalysts were also
successfully used without reduction for low temperature glycerol conversion to lactic acid in
liquid phase under alkaline conditions [19]. The main drawbacks of the reaction performed in
liquid phase are the use of relatively high temperatures and pressures requiring expensive
equipments.
In the past decade, metal fluorides and oxide (hydroxide) fluorides prepared by soft chemistry,
exhibiting tunable acid-base properties and high specific surface area, have shown promising
results in different heterogeneous catalytic processes as active phase or support [20,21]. More
specifically, magnesium fluoride was used successfully as support for numerous reactions such
as CO oxidation [22], DeNOx [23] synthesis of menthol [24], alkylation of thiophenic
compounds [25], synthesis of furfural [26], hydrogenation [27] among others, due to its high
thermal and chemical stability. MgF2 is considered as an inert support [28] or as an active
support with high density of acid sites with moderate strength [29, 30], depending on the
specific surface area and thus, on the synthesis process. Magnesium oxide (hydroxide) fluorides
MgF2-xOx/2 (or MgF2-x(OH)x) or, more specifically, the intimate mixture with controlled
composition of MgO and MgF2, formed after calcination above 400°C, were also successfully
used as support in several applications [31, 32]. The synthesis and characterization of this new
Mg–O–F system and its application as catalytic support were described [33-35]. As an example,
MgF2–MgO can be considered as a potential support of NiO in NOx reduction by propene [33].
The synthesis of the MgF2–MgO mesoporous material, particularly by sol-gel method [31,36],
4

allows to increase the specific surface area compared to single MgF2, MgO or Mg(OH)2
whatever the temperature of calcination [36]. For example, a specific surface area of 644 m2.g-
1 can be obtained for MgF(OH) before calcination whereas the specific surface area of MgF2 is
of 231 m2.g-1 in the same synthesis conditions. The higher specific surface area favors the high
dispersion of the active phase at the surface of the support, leading to high activity in several
applications [31-34]. Moreover, the easy control of the F/Mg ratio, by sol-gel process, allows
to fine tune the acid-base properties. Indeed, higher the fluorine content, higher the strength of
Lewis acidity and lower the amount and strength of basicity [37]. This is obviously due to the
stronger inductive effect of fluorine atom in comparison with oxygen atom. The interest of such
materials was already demonstrated for Michael addition reactions [38]. Among the different
soft chemistry method used to synthesize metal fluorides, the sol-gel method is very promising
because it is simple, flexible and cheap, offering an easy control of porosity and microstructural
properties [20].
In this work, the selective production of HA from glycerol in gas phase using magnesium oxide
(hydroxide) fluoride as support of copper catalyst is reported. To the best of our knowledge,
there is no report on the use of such materials for glycerol conversion. A mixture of glycerol
and water (80wt % glycerol) was used while the reaction temperature was fixed at 260°C. For
comparison, a La2CuO4 catalyst was also evaluated as a reference catalyst, since we showed in
a previous study that a high catalytic activity can be reached using pure glycerol (99.9%) [39].
A correlation between the properties of the materials (acid-base, oxidation state of copper) and
HA yield is discussed.
5

Experimental
Catalyst synthesis
The oxide (hydroxide) fluoride materials used in this work were prepared by a sol-gel
method, partly based on the work of Scholz et al. [36]. In a first step, magnesium metal (3.23g,
Aldrich, 99,98%) was treated with methanol in excess (100 mL, Sigma-Aldrich, 99,8%) under
reflux conditions for 6 h to form a Mg(OCH3)2 metal alkoxide solution. For MgF2 support,
stoichiometric amount of aqueous HF (11.074 g, 48 wt% HF in water) was added progressively
to the solution under stirring (avoiding the formation of a gel). A highly exothermic reaction
proceeds leading to the formation of a sol. This sol was stirred for 24 h, aged at ambient
temperature for 24 h, and dried at 100°C for 24 h, leading to the formation of the powder named
MgF2. For the support named MgF(OH), the protocol was the same except the added amount
of HF, which was adjusted to obtain an initial F/Mg ratio of 1. Additional water was used for
the hydrolysis of remaining -OCH3 groups, to form hydroxyl group (hydrolysis reaction). For
the support named MgO, no HF was added and the hydrolysis reaction was carried out with an
excess of water (6 g, H2O/Mg = 2.5). After drying at 100°C, the materials were calcined at
350°C for 5 h under dry air.
For the impregnation of copper on the support, 2 g of powder were mixed with a copper acetate
solution: 330.5 mg of (CH3COO)2Cu.1H2O (Aldrich, 98%) dissolved in 17 mL of water. The
amount of copper precursor was adjusted to obtain a theoretical weight content of metallic
copper of 5 %. This mixture was stirred for 24 h, at ambient temperature and dried at 100°C on
a sand bath followed by a final drying in a furnace at 80°C for 12 h. The catalysts were then
calcined at 350°C, for 5 hours, under dry air to form copper oxide, the materials are named Cu-
MgO, Cu-MgF(OH) and Cu-MgF2.
In order to compare the performances of these new catalysts with a reference catalyst, La2CuO4
was also prepared as described in ref [39].
6

Characterization
XRD analysis of samples were carried out with a PANalytical EMPYREAN powder
diffractometer using CuK radiation source (K1 = 1.5406 Å and K2 = 1.5444 Å). XRD
patterns were collected between 15 and 80° with a 0.033° step and 300 s dwell time at each
step. The identification of the phases was performed with the HighScorePlus software
(PANalytical©) and by comparison with the ICDD database reference files.
Nitrogen adsorption was performed at -196°C using a TRISTAR 3000 gas adsorption system.
Prior N2 adsorption, the powder samples were degassed under secondary vacuum for 12 h at
250°C. The BET equation was used to calculate the surface area of the samples (SBET in m2.g-
1). The total pore volume was calculated from the adsorbed volume of nitrogen at P/P0 equal to
0.99. The average mesopore-size distribution was calculated from the desorption isotherm
branch using the Barret-Joyner-Halenda (BJH) method.
The magnesium and copper contents of the samples were determined by Inductively Coupled
Plasma-Optical Emission Spectrometry (ICP OES) using a PerkinElmer Optima 2000DV
instrument.
The amount of carbon deposition, after the catalytic test, was calculated by Thermogravimetric
analysis (TGA) using a Q600TA Instrument apparatus, under dry air, with a heating rate of 5
°C min-1 from room temperature to 900°C.
The acidity of solid materials was measured by adsorption of pyridine followed by FT-IR
spectroscopy, using a ThermoNicolet NEXUS 5700 spectrometer with a resolution of 2 cm-1
and 128 scans per spectrum. The samples were pressed into thin pellets (10-30 mg) with
diameter of 16 mm under a pressure of 1-2 t.cm-2 and activated in situ during one night under
vacuum (10-5 Pa) at 250°C. Pyridine was introduced in excess, at 150°C, after the activation
period. The solid sample was vacuum-packed to eliminate physisorbed pyridine and IR
spectrum was recorded at 150°C. The concentration of Lewis acid sites was determined from
7

the integrated area band located between 1445 and 1455 cm-1, using 1.28 cm.mol-1 as molar
extinction coefficient. Note that no Bronsted acid site was detected by this method on all
samples (no band observed between 1540 and 1550 cm-1).
Temperature programmed reduction analysis (TPR) were carried out in a Micromeritics
Autochem 2910 equipment using 100 mg of catalyst. The experiments were performed using a
5 % H2/Ar mixture, with a flow rate of 100mL.min-1, while the temperature was raised at 5 °C
min-1 from ambient to 900 °C, then maintained at this temperature for 30 min. Prior to the
measurements, the samples were outgassed under helium at 350 °C for 8 h.
CO2 adsorption experiments were performed to determine the basic properties of the catalysts.
Adsorption equilibrium data were measured thermogravimetrically at 298 K using a
symmetrical SETARAM microbalance. A weight of 10 to 15 mg of sample was outgassed under
secondary vacuum at 573 K for 6 hours and then cooled down to 298 K prior to the sorption
measurements. The temperature of the system is kept constant during analysis by a water
circulation in the double wall of the analysis tube. The CO2 pressure was then increased step by
step in order to obtain the entire adsorption isotherm. For each uptake, the equilibrium was
reached when the mass recorded versus time and the pressure were stable.
The XPS analysis were carried out with a Kratos Axis Ultra DLD spectrometer using a
monochromatic Al Kα source (10mA, 15kV). The charge Neutraliser system was operated for
all analysis. Instrument base pressure was 9 x 10-8 Pascal. High-resolution spectra were
recorded using an analysis area of 300 µm x 700 µm and a 40 eV pass energy. These pass
energies correspond to Ag 3d5/2 FWHM of 0.55 eV. Data were acquired with 0.1 eV steps. All
the binding energies were calibrated with the Mg2p binding energy fixed at 50.7eV as an
internal reference.
Catalytic activity
8

The reaction was performed using a mixture of glycerol and water (80%wt of glycerol)
with a liquid flow rate of 0.04 mL.min-1. The reactor was a quartz tube of 13mm i.d. and 400mm
length. The catalytic decomposition of glycerol was carried out at atmospheric pressure by
passing a continuous flow of 20% v/v glycerol solution in N2 as the carrier gas over the catalyst
bed (200mg). The Gas Hourly Space velocity was equal to 13.5 L.h-1.g-1. A two zone reactor
was used, glycerol being vaporized in the first empty zone before arriving to the reaction zone.
The temperature in the first reactor was maintained at 300°C, while the temperature in the
second reactor was fixed at 260°C. Liquid phase products were recovered by condensation in
two cold traps located at the reactor outlet for further analysis, whereas gas phase was analyzed
by gas chromatography during the reaction. We checked that under the experimental conditions
used (relatively low temperature) no significant gaseous products such as carbon monoxide or
carbon dioxide was produced. Separation and quantification of main organic compounds in
liquid phase were performed using a gas chromatograph (Varian 430 GC) equipped with a
capillary column (50 m x 0.25mm x 0.2 µm, CP WAX 58 CB) and a FID detector.
Quantification was performed by using butanol as internal standard. The products of the
glycerol decomposition were identified by GC-MS (Varian 3800, injector 1079) coupled with
a mass spectrometer (Analyser triple quadrupole Varian 1200L with an electric impact source
of 70eV).
Conversion of glycerol and selectivity to hydroxyacetone was calculated according to the
following equations:
mol of glycerol reacted
Glycerol conversion (%) = mol of glycerol in the feed × 100
Selectivity to hydroxyacetone (%) = mol of hydroxyacetone produced × 100
mol of glycerol transformed
Results and discussion
Synthesis and characterization of the catalysts
9

The different magnesium oxide (hydroxide) fluorides were prepared according to a sol-
gel method, from a metal alkoxide precursor, in an aqueous HF solution. Briefly, a competition
between fluorolysis (reaction with HF) and hydrolysis (reaction with water) occurs, resulting
in the formation of MgF2-x(OH)x as described in ref [36]. Kinetically, the fluorolysis rate is
higher than the hydrolysis rate, allowing a partial control of the composition. As reported
previously, with an equivalent method [36], the final F/Mg ratio (2-x) corresponds
approximately to the initial HF/Mg ratio. The remaining -OCH3 groups of the magnesium
alkoxide react with water forming the hydroxyl groups of the hydroxide fluorides. Based on
previous works [36], materials with a composition closed to MgF2 (x = 0), MgF(OH) (x = 1)
and Mg(OH)2 (x = 2) were synthesized with an initial HF/Mg ratios of 2, 1 and 0 respectively.
Thus, the fluorine content in the magnesium hydroxide fluoride is easily tuned thanks to the
amount of HF added [31].
The X-ray diffraction (XRD) patterns of Cu-MgO, Cu-MgF(OH) and Cu-MgF2 obtained after
copper impregnation on the magnesium oxide hydroxide fluoride supports, and after calcination
at 350°C are reported in Figure 1. The calcination temperature was chosen in order to stabilize
thermally the catalysts while the reaction was performed at 260°C. MgO (card n°01-089-7746
of the ICDD database), MgF2 (card n°98-009-4270) and a mixture of Mg(OH)2 (card n°01-075-
1527) and MgF2 are mainly observed on the samples Cu-MgO, Cu-MgF2 and Cu-MgF(OH)
respectively. Interestingly, MgO in Cu-MgO and MgF2 in Cu-MgF2 are well crystallized
whereas a low degree of crystallinity is observed for Cu-MgF(OH). Indeed, the presence of two
phases affects the rate of crystallization of both phases as observed by Wojciechowska et al.
[23]. In our experimental conditions, the method used does not lead to the formation of a
magnesium oxide (hydroxide) fluoride but to an intimate mixture of magnesium oxide (or
hydroxide in our case) and fluoride as already observed in previous works [23]. The formation
of CuO (card n°00-041-0254) with the main diffraction peaks at 35.5 and 38.8° and Cu2O (card
10

n°98-005-2043) with the main diffraction peaks at 36.4 and 42.3° are clearly observed for Cu-
MgO and Cu-MgF2 respectively. Nevertheless, the presence of CuO on Cu-MgF2 and Cu2O on
Cu-MgO cannot be totally excluded due to the overlapping of copper oxide with the peaks of
the support. The attribution of the copper phase is more difficult in the case of Cu-MgF(OH).
Only one peak at 36.1 can be attributed to copper oxide phase. Nevertheless, this broad peak
with low intensity, characteristic of poorly crystallized phase, can be attributed to (111) peak
of CuO as well as (111) peak of Cu2O since the other main peaks of these phases can be
overlapped with the broad peaks of Mg(OH)2 and/or MgF2.
Finally, due to the low intensity of copper phases on XRD patterns, the determination of the
crystallite size of copper oxide is difficult. Nevertheless, the higher intensity and thinner peaks
of Cu2O in Cu-MgF2 is characteristic of larger crystallites in comparison with copper oxides in
Cu-MgO and Cu-MgF(OH).
The copper contents determined by ICP OES analysis (Table 1) are closed to the desired
theoretical content (5%) indicating that the chosen synthesis method is well adapted to control
the copper content.
The specific surface areas of the samples are reported in Table 1. As observed, after calcination
at 350°C and before impregnation of copper, the MgF(OH) support possesses a very high
specific surface area (270 m2.g-1), higher than the SBET of MgO (227 m2.g-1) and especially the
SBET of MgF2 (33 m2.g-1). This result confirms that the synthesis of magnesium oxide
(hydroxide) fluoride (or the intimate mixture of MgF2 and Mg(OH)2 in our case) by sol-gel
method, improves the specific surface area compared to MgF2 and MgO alone. This can be
attributed to the low rate of crystallinity of this sample as discussed above. The adsorption-
desorption isotherms of all solids are shown in Fig. S1 (see supplementary information). A type
IV isotherm according to the IUPAC classification was observed for all samples, whatever the
fluorine content, showing the formation of mesoporous solids with high value of porous volume
11

(Table S1). The porous volume decreases with the increase of fluorine content. The average
pore size is significantly smaller for the MgF(OH) support in comparison with MgF2 and MgO.
After impregnation of copper and calcination at 350°C, the specific surface area decreases for
Cu-MgF(OH) from 270 to 165 m2.g-1 and for Cu-MgO from 227 to 68 m2.g-1, whereas it remains
stable for Cu-MgF2 with 33 m2.g-1 before and 36 m2.g-1 after copper impregnation and
calcination. A type IV isotherm is retained for the three catalysts (Fig S1), the pore volume
decreases for Cu-MgF(OH) and for Cu-MgO in comparison with the supports without copper,
which is in accordance with the changes of specific surface area. The average pore sizes are
retained before and after copper impregnation and calcination (table S1). The decrease of the
specific surface area can be attributed to the impregnation step performed in water, with a
probable hydrolysis of the remaining –OCH3 due to the incomplete fluorination/hydrolysis of
the methanolic Mg(OCH3)2 solution [40]. Indeed, the higher the content of the remaining –OR
in the magnesium oxide (hydroxide) fluoride, the larger the surface. Nevertheless, very high
specific surface area is obtained for Cu-MgF(OH) compared to the two other samples which is
in accordance with the low rate of crystallinity observed by XRD (Fig. 1).
The determination of the acidity of the catalysts was performed by adsorption of pyridine
followed by FT-IR spectroscopy and the results are reported in Table 1 and fig S2. A deeper
discussion is also reported in section 2 of the supplementary information concerning the
experiments. Note that the characterization of the acidity of La2CuO4 catalyst was not possible
due to the too low surface area of the catalyst (SBET= 2.2 m2.g-1), that did not allow the
quantification of acid sites. No significant amount of Lewis acid site is observed on Cu-MgO
and Cu-MgF(OH) by this method. On the contrary, a significant amount of Lewis acid sites (50
mol g-1) is determined on Cu-MgF2. Interestingly, this value is closed to the amount observed
on the support alone prepared at 350°C (48mol g-1) showing that the presence of copper does
not change significantly the acidity of the material (section 2 of supplementary information).
12

This could be explained by the replacement of a Lewis acid site of MgF2 by copper ion (+1),
which acts as an electron acceptor providing Lewis acid centers [40]. However, it is also
possible to propose that the Lewis acidity would result mainly from the support since Cu+1 is
present as large particles at the surface of the support (according to XRD). Moreover, it is also
possible that the Lewis acid sites of Cu+1 are not enough strong to react with pyridine.
Furthermore, it is well known that the strength of Lewis acid sites of MgF2 is moderate [29].
For the supports MgO and MgF(OH), exhibiting no significant amount of Lewis acid sites (but
basic properties, as discussed below), the introduction of copper did not enhance the number of
Lewis acid site. The low acidity of CuOx-MgO catalysts was proved by Popescu et al. [41]. It
could also be explained by the diffusion of copper into the basic support as shown by XPS (see
below). Surprisingly, no Lewis acid sites are observed on Cu-MgF(OH) whereas MgF2 is
present in this sample (as shown by XRD). This can be explained by the formation of an
intimate mixture of MgF2 and Mg(OH)2 leading to strong interactions between both phases.
The presence of hydroxyl groups (Mg(OH)2) in the vicinity of unsaturated magnesium Lewis
acid sites (MgF2) decreases probably the strength of Lewis acidity due to the lower inductive
effect of -OH in comparison with fluorine atom. Consequently, the Lewis acid sites of Cu-
MgF(OH) are not enough strong to react with pyridine, a strong base.
TPR analysis
The TPR profiles of the catalysts Cu-MgO, Cu-MgF(OH) and Cu-MgF2 are reported in
Fig. 2.
The reduction profile of Cu-MgO shows two broad peaks in the 250-500°C temperature range,
which is in accordance with the results reported by Reddy et al. [42]. The first reduction peak
can be attributed to the reduction of well distributed CuO species or small copper oxide clusters
interacting weakly with the support, and the second reduction peak, at higher temperature can
13

be associated to large CuO particles or aggregated CuO clusters [42, 43]. Moreover, the
reduction of CuO in two steps was also proposed (CuO → Cu2O → Cu°) [41].
The reduction of Cu-MgF2 proceeds at low temperature (225°C) as a single peak, however the
non-gaussian profile of the peak suggests that different copper oxide species are present. As
shown in Fig. 1, Cu2O particles are observed in Cu-MgF2. According to previous studies, CuO
is more easily reduced than Cu2O, which seems in contradiction with the present study [44, 45]
by example the reduction peak of CuO/SiC is centered at 287°C against 302°C for Cu2O/SiC
[44]. However the reduction temperature depends strongly on the nature of copper support. The
reduction of Cu2O at a temperature as low as 225°C was reported by different authors over TiO2
[46, 47]. It was also shown that the reduction temperature depends strongly on phase structures:
the Cu2O-rutile TiO2 interaction was much stronger than the Cu2O-anatase TiO2 interaction. In
our case, weak Cu2O interaction was observed with MgF2 possessing the biggest particles and
lower specific surface area.
The amount of H2 consumed is reported in Table 2 for each catalyst. It can be observed that the
hydrogen consumption for the reduction of Cu-MgO is significantly higher than that for the
reduction of Cu-MgF(OH) and Cu-MgF2. Assuming that the hydrogen consumption
corresponds to copper oxide reduction, the amount of copper at the oxidation state of +2 and
+1 can be calculated according to the following reactions:
CuO + H2 = Cu + H2O
Cu2O + H2 = 2Cu + H2O
The results show that Cu2+ is mainly obtained on MgO while Cu1+ is the major copper species
formed over MgF2. This is in agreement with the characterizations by XRD exhibiting the
presence of Cu2O in Cu-MgF2 and CuO in Cu-MgO. Thus, MgF2, the more acidic support,
stabilized copper species at +1 oxidation step, while a mixture of Cu2+ and Cu+ is obtained for
14

MgF(OH). Obtaining a mixture of Cu2+ and Cu1+ for this last catalyst is not surprising since the
support is composed of two intimately mixed phases (MgF2 and Mg(OH)2), each phase
stabilizing Cu+ and Cu2+ respectively.
CO2 adsorption isotherms for Cu-MgO, Cu-MgFOH and Cu-MgF2 are used to characterize the
basicity and are presented in figure 3. From the observed results, it appears that the material
basicity is not linked to the copper content since the materials exhibit very different CO2
adsorption behaviors. The total CO2 adsorption capacity depends on the material porosity
whereas the amount of CO2 adsorbed in the monolayer is directly linked to the basic site number
present at the material surface. So, these isotherms have been analyzed using the BET model.
As expected, the calculated amount of CO2 adsorbed in the monolayer is linked to the material
BET surface area determined by nitrogen physisorption (figure 4). The more important the
surface area is, the higher is the CO2 amount directly in interaction with the solid surface. From
these result, it appears that the Cu-MgFOH sample has the higher basic site amount and Cu-
MgF2 the lowest.
XPS analysis
The surface composition of the catalysts were determined by XPS analysis and the Cu/Mg
atomic ratio was calculated. As shown in Table 3, the Cu/Mg ratio is higher at the surface of
the MgF2 support than in the bulk. In contrast, an important enrichment of Mg at the surface of
MgF(OH) and MgO is observed. The occupation of the catalyst surface by Mg was evidenced
by Liu et al. studying a CuO-MgO-TiO2 catalyst [48]. So, with MgF(OH) and MgO as support,
copper is incorporated into the support lattice leading to a higher Cu-Mg interaction, which
corroborates the TPR profiles. Using MgF2 as support, the diffusion of copper into the bulk is
not favored, the metal-support interaction is lower explaining the lower reduction temperature
(see Fig. 2).
15

Cu 2p3/2 XPS spectra of the catalysts Cu-MgO, Cu-MgFOH and Cu-MgF2 are shown in Figure
5. For the Cu-MgO catalyst, the presence of Cu2+ species is observed thanks to the presence of
the satellites peaks (BE region between 938 eV and 944 eV) which is the fingerprint of Cu2+
oxidation state. However the binding energy corresponding to Cu 2p3/2 is observed at 932.9eV,
a value significantly lower than the expected one: 933.6eV [49] probably due a charge transfer
from the metal ion toward the support matrix.
For Cu-MgFOH, two Cu species seem to be present, one with a Cu 2p3/2 binding energy of
932.8 eV and the other with a Cu 2p3/2 binding energy of 936.1 eV. These are consistent with
Cu(II) species and the latter is consistent with Cu(OH)2 as proposed by Frost et al. [50].
Nevertheless the presence of CuF2 cannot be excluded since fluoride atom are present in the
support and can react with copper species. The formation of such species would confirm the
reaction between copper and the support during the calcination step.
For the Cu-MgF2 sample a single and intense peak centered at 932.1 eV is visible, while no
satellite peaks are observed indicating that there is no Cu2+ but the presence of reduced copper
species. As it is well known, XPS cannot differentiate between Cu0 and Cu+ since the binding
energy is similar. However, the presence of Cu0 is unlikely in the present study since the catalyst
was calcined at 350°C under air and no metallic copper are observed by XRD.
The XPS analysis are in accordance with the results obtained by XRD and TPR, showing the
presence of Cu+ with MgF2 used as support, while Cu2+ is formed on MgO. Moreover,
significant higher amount of copper is observed at the surface of MgF2 compared to both other
catalysts.
Consequently the combination of characterization methods (TPD, XRD, TPR, XPS), proves
that copper is obtained at different oxidation state depending on the nature of the support.
However it is not possible to propose a simple correlation between oxidation state of copper
16

and support physico-chemical properties since the three materials differ strongly in terms of
acid/base properties, fluorine content, size of particles…
Catalytic results
The reaction was investigated in gas phase using a mixture of glycerol and water (80 wt
% glycerol) at 260°C to avoid the production of gaseous products at higher temperatures. The
main product was HA but pyruvaldehyde and glyceraldehyde were also formed. Other products
were analysed but they were present in low amount: propionaldehyde, glycidol, 1,3
propanediol, 1 propanol...
First, using the support alone (without copper) it was confirmed that no glycerol transformation
occurs, indicating the crucial role of copper.
The catalytic activity of the copper supported on oxide (hydroxide) fluoride catalysts is
compared with La2CuO4, which exhibited a high activity toward glycerol dehydration into
hydroxyacetone in the absence of water in the gas feed [39]. The catalytic behavior of the three
oxide (hydroxide) fluoride based catalysts differs strongly. The highest glycerol conversion is
obtained over the Cu-MgF2 catalyst, it reaches 82 % and remains relatively stable during four
hours of reaction (figure 6). Cu-MgO and Cu-MgFOH are little active under our experimental
conditions with a glycerol conversion of 19 % and 33 % after one hour of reaction respectively.
The yield in HA is also maximum with Cu-MgF2, decreasing slightly with time on stream
(figure 7). Moreover, the Cu-MgF2 catalyst is significantly more active than La2CuO4, in the
presence of water.
High glycerol conversions were also reported by Carvalho et al. [51] using Cu-based
hydroyapatites. The authors showed that glycerol conversion increased with the amount of
copper (from 3 to 17 %) highlighting the crucial role of copper as in the present work. However
a catalytic deactivation is observed after few hours on stream. Sato et al. [52] studied the
influence of the support on glycerol conversion over copper-based catalysts in gas phase at
17

250°C, they found that the acid-base property of the support affected the selectivity: basic MgO,
CeO2 and ZnO supports showed low selectivity to hydroxyacetone, while acidic supports such
as Al2O3, ZrO2, Fe2O3 and SiO2 promoted HA selectivity which is in accordance with our results
(Cu-MgF2 being the most acidic catalyst).
Glycerol dehydration into hydroxyacetone implies the removal of one of the two OH groups
from the terminal carbons in the glycerol molecule, while the removal of the OH group from
the central carbon atom leads to the formation of acrolein through the unstable 3-
hydroxypropenal. The predominant route depends mainly on the nature of the acid sites, it has
been suggested that acrolein is formed over Brønsted acid sites and hydroxyacetone over Lewis
acid sites [53]. The role of basic centers was also proposed by Stosic et al. [53]. The authors
showed that the yield in HA was increased with the number of basic sites. However, using
hydroxyapatite catalyst, the main reaction product was acrolein, they concluded that acrolein
formation was not only controlled by the surface acidity but also by hindering the
number/strength/activity of the basic sites, and thus limiting the side reactions which affect the
selectivity in acrolein. An intermediate enol is suggested to be formed at the surface of basic
centers, afterwards the enol is rapidly transformed by rearrangement into 1-hydroxyacetone.
The results obtained in the present study showed that the basic properties of the catalysts does
not governed the dehydration of glycerol into HA but that HA is preferentially produced over
the catalyst possessing Lewis acidity (Cu-MgF2), while no acrolein was formed. The high yield
in HA can be directly correlated with the amount of Cu+1, which also corresponds to the
maximum number of Lewis acid sites (figure 8). Moreover, the Lewis acid sites of the MgF2
support are not able to convert glycerol into hydroxyacetone since no activity was observed
with MgF2 alone. Mitta et al. [40] also showed that a Y zeolite possessing 180µmol g-1 of Lewis
acid sites exhibited a very low glycerol conversion without copper. It is thus possible to
conclude that the dehydration of glycerol requires the presence of copper under the
18

experimental conditions used (vapor phase, relatively low reaction temperature: 260°C in the
present study and 210°C in the work of Mitta et al.).
The reaction route to hydroxyacetone involves the formation of an enol as proposed by different
authors on copper based catalysts [54]. The important role of copper oxidation state was
indicated by Pinheiro et al. [55] and Xiao et al. [56]. The weak acid sites, Cu2+ cannot be
responsible for dehydration [57], as confirmed by the results we obtained in the present paper.
It is also clear that metallic copper is an active site for dehydration of glycerol to HA,
dehydrogenation at the surface of Cu° occurs first and the formation of Cu-alkoxide species is
proposed [53, 58, 59]. Over Cu+, the reaction mechanism is more difficult to establish, different
authors showed that Cu+ is active for dehydration of glycerol [56, 57] which corroborates our
results. Xiao et al. [56] indicate that it is not possible to state if Cu+ is active “per se” or if it is
due to the Lewis acid properties of Cu+.
So according to the studies published and to the results we obtained two different mechanisms
are proposed: one based on Lewis acid properties of Cu+ and the other one based on the
oxidation state of copper: Cu+.
The involvement of the Lewis acid sites of Cu+ can be proposed according to the mechanism
of Alhanash et al. (Fig.9 a). In the second mechanism, we propose a homolytic C-H bond
dissociation (similar to the one proposed on metallic copper, Fig.9 b). The C-H bond rupture at
the surface of Cu+ is supported by the work of Wang et al. [60], who performed DFT
calculations to explain the dehydrogenation of cyclohexanol to cyclohexanone. The authors
proved that the hydrogen bonded to carbon is more easily removed on Cu+ than on Cu°, so a
homolytic rupture of the C-H bond in glycerol by Cu+ is proposed.
In order to support the mechanism, theoretical calculations were performed and compared with
the results of Nimlos [61] who established the more probable intermediate species in glycerol
dehydration to HA. The calculations were performed at the B3LYP/6-31+G(d,p) level.
19

Transition states took into consideration were radical intermediate of glycerol as shown in Fig.
9b. All computed energies of the mechanism compounds are Gibbs free energies at 534.15 K.
Calculations were performed with the Gaussian09 code.
The relative energies of the transition states involved in the two mechanisms are 70.9 and 73.2
kcal/mol for mechanism 9a) and 9b) respectively (Fig. 10). Such high energy barriers indicate
that these reactions are only likely to occur at high temperatures and under pyrolysis conditions.
The values obtained, are in the same order indicating that the formation of intermediate species
in the mechanism we propose are favorable in our experimental conditions (T=260°C).
Shortcomings of the simulation held should be highlighted, in our case we didn’t take
into consideration the “Transition State” theory as we considered each step of the mechanism
as independent reactions. Additionally, the surface of the catalyst wasn’t taken into
consideration. This simplification hindered essential details about sorption and stabilization of
glycerol on the surface.
Further studies on the theoretical and experimental level should be held on the role of
the copper catalyst, and we consider that the discussion on the mechanism is still an open issue.
A low catalytic activity towards HA formation is observed with Cu-MgO and Cu-MgF(OH),
which can result from low amount (or absence) of Cu1+ at the surface of the support and high
amount of basic sites which probably favors the occurrence of side reactions giving by-products
such as acids, aldehydes, aromatic compounds, glycerol oligomers [62]… and also coke as
evidenced by TGA analysis (figure 11), as discussed below.
Characterization of the catalysts after reaction
After catalytic tests, the crystalline structure of the supports is retained (Fig. 1) showing
the good stability of the magnesium oxide (hydroxide) fluoride materials under our
experimental conditions, confirming their potential as support. The peaks corresponding to
copper oxides disappear and the formation of metallic copper is observed on Cu-MgF(OH) and
20

Cu-MgO. This result can be explained by the reduction properties of glycerol according to Jin
et al. [63].
On the contrary, Cu2O is always observed on Cu-MgF2 even if the formation of Cu° cannot be
totally excluded due to the peaks with very low intensity at 43.3° (shoulder) and 50.4° (figure
1). Moreover, the peaks of Cu2O are broadened after reaction reflecting a decrease of the
crystallites size probably due to a partial reduction of the oxide. Nevertheless, in our
experimental conditions, MgF2 support seems to stabilize the copper oxide at a degree of
oxidation +I while metallic copper is obtained over the two other supports.
As observed in Table 1, the specific surface areas of Cu-MgF(OH) and Cu-MgO decrease
drastically (from 165 to 17 m2.g-1 and from 68 to 37 m2.g-1 respectively) after catalytic tests
whereas the specific surface area remains comparatively stable for Cu-MgF2. This is in
agreement with the XRD pattern since no change is observed after the catalytic test for Cu-
MgF2.
Thermal analyses (TGA) were performed on the catalysts before and after catalytic tests until
900°C (figure 11). As observed, a very weak weight loss (2.5%) is observed on Cu-MgF2
corresponding to the removal of physisorbed water and/or remaining organic compounds
derived from the sol-gel synthesis process. This weight loss is higher on Cu-MgO (4.7 %) and
especially on Cu-MgF(OH) (14.6 %). In addition to physisorbed water and remaining organic
compounds, these weight losses between 300 and 400 °C for Cu-MgO and between 400 and
500 °C for Cu-MgF(OH) can be explained by the dehydroxylation of magnesium hydroxide.
This is in agreement with the XRD pattern of Cu-MgF(OH) (figure 1) showing the presence of
magnesium hydroxide. Moreover, the XRD pattern of Cu-MgF(OH) after TGA (not shown)
corresponds to a mixture of MgO and MgF2 confirming this dehydroxylation. Even if only MgO
is observed for Cu-MgO (figure 1), the presence of a small amount of hydroxyl group cannot
be totally excluded. After catalytic tests, the weight losses are higher whatever the catalysts.
21

This is due to the formation of “coke” on the catalyst during the transformation of glycerol,
removed by oxidation during the thermal analysis under dry air. The deposition of coke is
clearly limited on Cu-MgF2 since the difference of weight loss before and after the catalytic test
is only of 2.5 %. By contrast, the amount of coke is clearly higher for the two other catalysts
with a difference of weight loss of 24 % for Cu-MgF(OH) and 12.9 % for Cu-MgO. These
results are consistent with the SBET which decreases after reaction for both catalysts while SBET
of Cu-MgF2 remains stable as discussed above. Based on the assumption that basic sites play a
role on the coke formation by side-reactions as discussed above, the amount of coke is also in
accordance with basic properties, higher the basicity (figure 4), higher the amount of coke.
Thus, Cu-MgF2 is a more suitable catalyst than the two other catalysts to avoid decrease of
specific surface area, formation of coke and deactivation. Long terms experiment should be
performed to confirm the catalytic stability of the catalyst.
Conclusions
The dehydration of glycerol to hydroxyacetone was successfully investigated over copper
supported on magnesium oxide (hydroxide) fluoride catalysts, in gas phase in the presence of
water at 260°C. The catalysts were prepared according to a sol-gel method following by a
copper impregnation step and led to different CuOx-magnesium oxide (hydroxide) fluoride
composites with different crystallinities, copper oxidation state and tunable acid-base
properties. Among the different studied catalysts, Cu-MgF2 is the most interesting since Cu+,
the active site, is stabilized at the surface of the support, whereas no (or little) coke formation
during time on stream is observed contrary to Cu-MgO and Cu-Mg(OH)F catalysts. Moreover,
this catalyst led to the best catalytic activity and stability, with a glycerol conversion reaching
82% and a yield in hydroxyacetone of 45.5% after one hour of reaction. Cu-MgF2 is much more
active than La2CuO4 (26.7% yield in HA) which exhibited a high activity towards HA synthesis
but in the absence of water. The results obtained in the present study showed that the basic
22

properties of the catalysts did not governed the dehydration of glycerol into HA. The best
catalyst (Cu-MgF2) is the one possessing Lewis acidity and stabilizing copper at +1 oxidation
state.
Thanks to their tunable properties (acid/base, specific surface area, metal/support interaction),
this work confirms the potential of magnesium oxide (hydroxide) fluoride as support for
different heterogeneous catalytic processes as already observed for other reactions. In another
way, it offers new opportunities for the production of hydroxyacetone from glycerol with an
environmentally friendly process.
References
[1] A. Behr, J. Eilting, K. Irawadi, J. Leschinski, F. Lindner, Green Chem., 10 (2008) 13-30
[2] D. Sun, Y. Yamada, S. Sato, W. Ueda, Appl. Catal. B: Env., 193 (2016) 75-92
[3] D. Roy, B. Subramaniam, R. V.Chaudhari, Catal. Today, 156 (2010) 31-37
[4] V. Montes, M. Checa, A. Marinas, M. Boutonnet, J.M. Marinas, F.J. Urbano, S. Järas, C.
Pinel, Catal. Today, 223, (2014) 129-137
[5] M. Massa, A. Andersson, E. Finocchio, G. Busca, J Catal. 307 (2013) 170-184
[6] J. Deleplanque, J.-L. Dubois, J.-F. Devaux, W. Ueda, Catal. Today, 157 (2010) 351-358
[7] A. Talebian-Kiakalaieh, N. Aishah Saidina Amin, Renewable Energy, 114 (2017) 794-804
[8] T. Ma, J. Ding, R. Shao, W. Xu, Z. Yun, Chem. Eng. J., 316 (2017) 797-806
[9] M. E Doukkali, A. Iriondo, J.F. Cambra, I. Gandarias, L. Jalowiecki-Duhamel, F.
Dumeignil, P.L. Arias, Appl. Catal. A: Gen. 472 (2014) 80-91
[10] M. Dalil, D. Carnevali, M. Edake, A. Auroux, J.L. Dubois, G.S. Patience, J. Mol. Cat. A:
Chem., 421 (2016) 146-155
[11] L. H. Vieira, K.T.G. Carvalho, E. A. Urquieta-González, S . H. Pulcinelli, C. V. Santilli,
L. Martins, J. Mol. Cat. A: Chem., 422 (2016) 148-157
[12] W. Suprun, M. Lutecki, H. Papp, Chem. Eng. Tech. 33 (2010) 1-7
[13] A. Alhanash, E. F. Kozhevnikova, I. V. Kozhevnikov, Appl. Catal. A: Gen., 378 (2010)
11-18
23

[14] C.L. Lima, S. J.S. Vasconcelos, J.M. Filho, B. C. Neto, M.G.C. Rocha, P. Bargiela, A. C.
Oliveira, Appl. Catal. A: Gen 399 (2011) 50-62
[15] P. Lauriol-Garbey, J.M.M. Millet, S. Loridant, V. Bellière-Baca, P. Rey, J. Catal. 281
(2011) 362-370
[16] A. Kinage, P. Upare, P. Kasinathan, Y. Kyu Hwang, J.S. Chang, Catal. Comm. 11 (2010)
620-623
[17] C. Montassier, J.C. Ménézo, L.C. Hoang, C. Renaud, J. Barbier, J. Mol. Cat. 70 (1991)
99-110
[18] S. Xia, R. Nie, X. Lu , L.Wang, P. Chen, Z. Hou, J. Catal. 296 (2012)1-11
[19] D. Roy, B. Subramaniam, R. V. Chaudhari, ACS Catal. 1 (2011) 548-551
[20] E. Kemnitz, Catal. Sci. Technol. 5 (2015), 786-806
[21] S. Célérier F. Richard, Catal. Commun. 67 (2015) 26-30
[22] M. Wojciechowska, W. Przystajko, M. Zielinski, Catal. Today 119 (2007) 338–341
[23] M. Wojciechowska, M. Zielinski, M. Pietrowski, Catal. Today 90 (2004) 35–38
[24] A. Negoi, S. Wuttke, E. Kemnitz, D. Macovei, V. I. Parvulescu, C. M. Teodorescu, S. M.
Coman, Angew. Chem. Int. Ed. 49 (2010) ) 8134 –8138
[25] F. Richard, S. Célérier, M. Vilette, J. D. Comparot, V. Montouillout, Appl. Catal. B 152-
153 (2014) 241-249
[26] I. Agirrezabal-Telleria, F. Hemmann, C. Jäger, P.L. Arias, E. Kemnitz, J. Catal. 305 (2013)
81–91
[27] M. Zielinski, M. Pietrowski, A. Kiderys, M. Kot, E. Alwin, J. Fluor. Chem. 195 (2017)
18–25
[28] M. Wojciechowska, M. Zielinski, M. Pietrowski, J. Fluor. Chem. 120 (2003) 1-11
[29] S. Wuttke, S. M. Coman, G. Scholz, H. Kirmse, A. Vimont, M. Daturi, S. L. M. Schroeder,
E. Kemnitz , Chem. Eur. J. 14 (2008) 11488 – 11499
[30] E. Kemnitz, S. Wuttke, S. M. Coman, Eur. J. Inor. Chem., (2011) 4773-4794
[31] M. Wojciechowska, A. Wajnert, I. Tomska-Foralewska, M. Zielinski, B. Czajka, Catal
Lett 128 (2009) 77–82
[32] I. Tomska-Foralewska, M. Zielinski, M. Pietrowski, W. Przystajko, M. Wojciechowska,
Catal. Today 176 (2011) 263–266
[33] M. Zielinski, I. Tomska-Foralewska, M. Pietrowski, W. Przystajko, M. Wojciechowska,
Catal. Today 191 (2012) 75–78
24

[34] M. Zielinski, A. Kiderys, M. Pietrowski, I. Tomska-Foralewska, M. Wojciechowska,
Catal. Commun. 76 (2016) 54–57
[35] M. Bonarowska, M. Wojciechowska, M. Zielinski, A. Kiderys, M. Zielinski, P. Winiarek,
Z. Karpinski, Molecules 21 (2016) 1620
[36] G. Scholz, C. Stosiek, M. Feist, E. Kemnitz, Eur. J. of Inorg. Chem. 14 (2012) 2337-2340
[37] S. Wuttke, S. M. Coman, J. Kroehnert, F. C. Jentoft, E. Kemnitz, Catal. Today 152 (2010)
2-10
[38] H. A. Prescott, Z. J. Li, E. Kemnitz, J. Deutsch, H. Lieske, J. Mater. Chem. 15 (2005)
4616–4628
[39] M. Velasquez, A. Santamaria, C. Batiot-Dupeyrat, Appl. Catal. B: Env, 160-161 (2014)
606-613
[40] H. Mitta, P. Kumar Seelam, S. Ojala, R. L. Keiski, P. Balla, Appl. Catal. A, General 550
(2018) 308-319
[41] I. Popescu, N. Tanchoux, D. Tichit, I.C. Marcu, Appl. Catal. A: Gen 538 (2017) 81-90
[42] K.H.P. Reddy, Y.W. Suh, N. Anand, B.D. Raju, K. S. Rama Rao, Catal. Com. 95 (2017)21-
25
[43] M. Jablonska, L. Chmielarz, A. Wegrzyn, K. Guzik, Z. Piwowarska, S. Witowski, R.I.
Walton, P.W. Dunne, F. Kovanda, J. Therm. Anal. Calorim. 114 (2013) 731-747
[44] Y. Wang, X. Guo, M. Lu, Z. Zhai, Y. Wang, X. Guo, Chinese Journal of Catalysis 38
(2017) 658-664
[45] C.S. Polster, H. Nair, C.D. Baertsh, J. Catal. 266 (2009) 308-319
[46] Y. Liu, Z. Wang, W. Huang, Appl. Surf. Sci. 389 (2016) 760-767
[47] C.S. Chen, T.C. Chen, C.C. Chen, Y.T. Lai, J.H. You, T.M. Chou, C.H. Chen, J.F. Lee,
Langmuir, 28 (2012) 9996-10006
[48] C. Liu, X. Guo, Q. Guo, D. Mao, J. Yu, G. Lu, J. Mol. Cat. A: Chemical 425 (2016) 86–
93
[49] M.C. Biesinger, L.W.M. Lau, A.R. Gerson, R.S.C. Smart, Appl. Surf. Sci. 257 (2010) 887–
898
[50] Ray L. Frost, Yunfei Xi, Barry J. Wood, Thermochimica Acta 545 (2012) 157– 162
[51] D. C. Carvalho, L. G. Pinheiroa, A. Campos, E. R.C. Millet, F. F. de Sousac, J. M. Filhoc,
G. D. Saraivac, E. C. da Silva Filhod, M. G. Fonsecae, A. C. Oliveira, Appl. Catal. A: Gen 471
(2014) 39-49
[52] S. Sato, M. Akiyama, R. Takahashi, T. Hara, K. Inui, M. Yokota, Appl. Catal. A : Gen
347 (2008) 186-191
25

[53] D. Stosic, S. Bennici, S. Sirotin, C. Calais, J.L. Couturier, J.L. Dubois, A. Travert, A.
Auroux, Appl. Catal. A: Gen., 447-448 (2012) 124-134
[54] B. Katryniok, S. Paul, V. Bellière-Baca, P. Reye and F. Dumeignil, Green Chem., 12
(2010) 2079-2098
[55] T.Pinheiro Braga, N. essayem, A. Valentini, J. Therm. Anal. Calorim. 129 (2017) 65-74
[56] Z. Xiao, X. Wang, J. Xiu, Y. Wang, C.T. Williams, C. Liang, Catal. Today, 234 (2014)
200-207
[57] P.A. Torresi, V.K. Diez, P.J. Luggren, J.I. Di Cosimo, Appl. Catalysis A: general 458
(2013) 119-129
[58] A.J. Gellman, M.T. Buelow, S.C. Street, T.H. Morton, J. Phys. Chem. A 104 (2000) 2476
[59] R.M. Rioux, M.A. Vannice, J. Catal. 216 (2003) 362-376
[60] Z. Wang, X. Liu, R. P. Hu, Surface Science 640 (2015) 181-189
[61] M.R. Nimlos, S.J. Blanksby, X. Qian, M.E. Himmel, D.K. Johnson, J.Phys. Chem. A 110
(2006) 6145-6156
[62] F.A.A. Barros, H.S.A. de Sousa, A . C. Oliviera, M.C. Junior, J.M. Filho, B.C. Viana, A.C.
Oliveira, Catal. Today, 212 (2013) 127-136
[63] F. Jin, X. Zeng, Z. Jing, H. Enomoto, Ind. Eng. Chem. Res. 51 (2012) 9921-9937
26

Table 1 Textural properties of the catalysts
SBET (m2 g-1)
after
catalytic
test
Lewis
Acidity
SBET
SBET
Weight content
of Cu (%)1
Support
Catalyst
(m2 g-1)
(m2 g-1)
(µmol g-1)
MgO
MgF(OH)
MgF2
227
Cu-MgO
Cu-MgF(OH)
Cu-MgF2
68
165
36
37
4.7
4.6
4.8
< 10
< 10
50
270
17
27
33
La2CuO4
2.2
n.d.
1 determined from ICP OES analysis
Table 2 Hydrogen consumption and Cu2+/Cu+ concentration determined from TPR analysis.
H2 consumption for copper oxide
reduction (mL STP/g) ( 2%)
Cu (%)***
Catalyst
Estimated*
Determined**
Cu2+ Cu+
a) Cu2+ b) Cu+
Cu-MgO
Cu-MgF(OH)
Cu-MgF2
21.5
19.5
20.3
10.7
9.8
20.5
16.3
12.7
91
67
25
9
33
75
10.2
*Estimated: calculated from the nominal composition of the material according to the
following reactions: a) CuO + H2 = Cu + H2O and b) Cu2O + H2 = 2Cu + H2O
**Determined: obtained from H2-TPR profile
*** Cu2+ and Cu+ concentration (%) estimated from the volume of H2 obtained by TPR
27

Table 3 Cu/Mg atomic ratio
Atomic ratio Cu/Mg
Cu-MgO
0.036
Cu-MgF(OH)
0.050
Cu-MgF2
0.054
Bulk*
Surface**
0.008
0.029
0.082
* Calculated from ICP OES
** Calculated from XPS analysis
28

Figure 1. XRD patterns of (I and I’) Cu-MgO, (II) Cu-MgF(OH) and (III) Cu-MgF2 after
synthesis (a) and after catalytic test (b).
29