Continuous catalytic process for the selective dehydration of glycerol over Cu-based mixed oxide

Journal of Catalysis 385 (2020) 160–175

Contents lists available at ScienceDirect

Journal of Catalysis

journal homepage: www.elsevier.com/locate/jcat

Continuous catalytic process for the selective dehydration of glycerol

over Cu-based mixed oxide

⇑

Jaime Mazarío, Patricia Concepción, María Ventura, Marcelo E. Domine

Instituto de Tecnología Química (UPV-CSIC), Universitat Politècnica de València, Consejo Superior de Investigaciones Cientíﬁcas, Avda. de los Naranjos s/n, 46022 Valencia, Spain

a r t i c l e i n f o

a b s t r a c t

Article history:

The selective dehydration of glycerol to hydroxyacetone (acetol) was studied with Cu-based mixed oxi-

des derived from hydrotalcite as catalysts in a continuous ﬂow ﬁx-bed reactor. Catalysts were prepared

Received 9 September 2019

Revised 9 March 2020

Accepted 15 March 2020

by co-precipitation and characterized by ICP, N

Different parameters were investigated to develop the most appropriate material as well as to determine

the function of every metallic species. The optimized Cu-Mg-AlO

2 3 2

adsorption, XRD, NH -TPD, CO -TPD, TPR and TEM.

x

offered ꢀ60% acetol selectivity at >90%

glycerol conversion (ꢀ80% liquid yield, up to TOS = 9 h). The catalyst could be regenerated by calcination,

achieving full activity recovery after ﬁve re-cycles. ‘‘In-situ” FTIR and XPS measurements evidenced that

the presence of Cu, specially the most active Cu¹species, was essential to carry out the dehydration to

acetol with high reaction rates, and to form the preferred intermediate (with C@O group); although a

minor contribution from Cu⁰and Cu species could not be discarded.

Keywords:

Acetol

Glycerol

Glycerol dehydration

Copper catalyst

Mixed oxides

+

2+

1

. Introduction

Nowadays, fossil fuels such as coal, petroleum and natural gas

esters. The raw materials being the source of triglycerides used

to be colza, sunﬂower, soy or palm oil, whereas the biodiesel

obtained can be used directly as a fuel or mixed with petro-

diesel in different ratios [3]. However, in the last years, and due

to the constrains derived from the use of agricultural suitable land

and vegetables for fuel industry against food applications, biodiesel

production coming from the transformation and upgrading of used

oils derived from food industry, as well as from other residues and

unconventional raw materials (i.e. animal fats, agricultural and

domestic wastes, oil wastes, etc.) [7–9] started to be developed.

Even in these novel cases, huge amounts of glycerol are also

obtained as a by-product, which is supposed to be somehow used

if the bioreﬁnery wants the process to be proﬁtable [10]. Conse-

quently, the transformation and valorization of glycerol has been

broadly studied by many research groups working in catalysis,

and compiled in some interesting reviews [11–13]. This academic

effort has resulted in glycerol integrating the list of the twelve

more important platform molecules derived from biomass [14].

Many possibilities of valorization for this molecule by using

heterogenous catalysis have been developed in the last years. Pro-

cesses such as the synthesis of glycerol carbonate for solvent and

polymer production with hydrotalcite-derived oxides [15] or zeo-

lites [16] as catalysts, the glycerol steam reforming to produce

syn-gas using catalysts containing Pt, Ru or Re supported on differ-

ent inorganic matrix [17], as well as the manufacture of fuel addi-

tives by means of glycerol acetalization, etheriﬁcation or

provide more than 75% of the total energy consumed in our planet

and represent an essential source of raw materials for the chemical

industry [1]. Nonetheless, the growing use of these resources has

led to their depletion along with an increment in the levels of

2

CO emissions to the atmosphere, which is considered as the main

cause of the greenhouse gases (GHG) effect in the planet. Thus, the

conversion of alternative, clean and renewable raw materials such

as biomass into fuels and chemicals has become an interesting way

of dealing with global warming and contributing to the diversity of

energy sources [2–5].

In the last years, a wide variety of the so-called biofuels has

appeared in the market: bioethanol, biobutanol, biodiesel, biogas

or pyrolysis gases as promising renewable alternatives [6]. The

common feature of these biofuels is the reduced amount of green-

house gases generated while using them in different applications,

thereby allowing to design processes without CO net emission

2

provided that efﬁcient crop methods are developed to produce

the biomass of interest [3,5].

Regarding the biodiesel, it is synthesized by means of a transes-

teriﬁcation reaction using triglycerides (vegetable oils) and metha-

nol or ethanol to produce the corresponding methyl or ethyl-

⇑

Corresponding author.

E-mail address: mdomine@itq.upv.es (M.E. Domine).

https://doi.org/10.1016/j.jcat.2020.03.010

0

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

161

esteriﬁcation reactions [18] are among the most relevant catalytic

strategies up to now reported.

either during the synthesis or in post-synthesis steps, thus extend-

ing the versatility of these materials.

The selective dehydration of glycerol to hydroxyacetone (or

acetol) is of interest if the reactivity of this molecule, having a ter-

minal hydroxyl group and a carbonyl group in the second carbon, is

considered. These functional features give this molecule the possi-

bility of taking part in different organic reactions such as the Man-

nich reaction [19] or some aldolic condensations [20]. Besides,

acetol is a key intermediate in the production of 1,2-propanediol

or propylene glycol [21]. Therefore, a better understanding of this

glycerol dehydration could also give an improvement on the yield

to propylene glycol and on any product derived from it.

With respect to acetol production, Chiu et al. have studied the

dehydration of glycerol to acetol with different heterogeneous cat-

alysts in a reactive distillation system, the highest values of con-

version and selectivity being reached with copper-chromite

catalyst [22]. Nevertheless, copper-chromite cannot be considered

as an environmentally friendly catalyst due to its toxicity and,

additionally, the scale-up of a reactive distillation system is not

straightforward from an industrial viewpoint. In this sense, contin-

uous catalytic processes would be more suitable, although just a

few articles can be found in the literature [23–25], most of them

involving the use of hydrogen. In this regard, it would be very

interesting to develop a continuous and sustainable process for

the selective dehydration of glycerol to acetol by using an efﬁcient

heterogeneous catalyst.

In light of all the above-mentioned, in this work an efﬁcient

continuous catalytic process for the selective dehydration of glyc-

erol to acetol (see Scheme 1) in a ﬁxed bed reactor by using tran-

sition metal hydrotalcite-derived materials as catalysts and

without introducing hydrogen in the reactor is designed and opti-

mized. Concretely, Mg/Al hydrotalcite-type materials with differ-

ent amounts of Cu, Ni, or Co were prepared, since these

transition metals have been reported to help to produce acetol as

an intermediate during glycerol reforming [35,36]. Their

hydrotalcite-derived metallic mixed oxides were evaluated in the

production of the desired acetol from glycerol. Special attention

is paid to the role of copper in the formation of acetol and the inﬂu-

ence of its oxidation state in that process, as it appears to be a key

point to generate the dehydration product, as it was observed dur-

ing glycerol hydrogenolysis process [37,38].

2

. Experimental

2.1. Materials

Glycerol (Sigma-Aldrich, 99.5%) was used as reagent in the

dehydration reaction. For the analysis of the reaction products ace-

tol (90%), methyl glycolate (99%), methyl lactate (98%), acetoin

(

98%), methyl pyruvate (98%), acetic acid (99%), propionic acid

99.5%), 1,2-propanediol 99.5%), 1,3-propanediol (98%), methyl

Due to the considerable number of studies carried out to pro-

duce acrolein, the other possible dehydration product from glyc-

erol, Brönsted acid centers are believed to present a higher

selectivity to its formation. On the contrary, Lewis acid centers

would in principle provide more selectivity towards acetol

acetate (98%), methyl formiate (99%) and chlorobenzene (internal

standard, 99%) were purchased from Sigma-Aldrich and used as

received. Acrolein (Merck, 95%) and dihydroxyacetone (Merck,

9

8%) were also used. Methanol (reagent grade 99.9%) supplied by

[

26,27]. Nonetheless, it is worth noting that some contradictory

Scharlau and water (Milli-Q quality, Millipore) were used as sol-

vents. Hydrotalcite-type materials were prepared by using in each

case the suitable metallic precursors bought from Sigma-Aldrich:

opinions can be found in the literature about the reaction mecha-

nism of this system. Some authors argue that the acetol formation

could be favored by a certain ratio between acid and basic Lewis

centers [28,29]. Following this idea, and to effectively carry out

the selective dehydration of glycerol to acetol, it is reasonable to

think that the combination of both acid and basic Lewis centers

contained in metallic mixed oxides derived from hydrotalcites con-

stitutes an excellent alternative to accomplish this target.

Mg(NO

98%), Ni(NO

tion, Na CO

8.0%) and H

sis. For comparative purposes MgO and Al

3

)

2

ꢂ6H

2

O

(99%), Al(NO

ꢂ6H O (98.5%) and Co(NO

(Fisher Chemical, 99.7%), NaOH (Sigma-Aldrich,

O (milliQ, Millipore) were also used for their synthe-

(Sigma-Aldrich) were

3

)

3

ꢂ9H

2

O

(98%), Cu(NO

3

)

2

ꢂ2.5H

2

O

(

3

)

2

3

)

2

ꢂ6H O (98%). In addi-

2

3

9

2

2 3

O

also tested in the reaction. Hydrogen (99.999%) was purchased

from Abelló Linde S.A.

Hydrotalcite-type materials are Mg-Al lamellar double hydrox-

ides with a structure derived from brucite [Mg(OH) ]. This struc-

2

ture is constituted by two dioctahedral sheets of magnesium

hydroxide with water molecules in the interlaminar space. Hydro-

2

.2. Catalyst preparation

2

+

talcites are the result of an isomorphic substitution of some Mg

_b_y_A_l₃+ thereby the sheet acquires a positive charge which allows

several anions to be located beside the water molecules in the

Hydrotalcite-type materials were prepared by the co-

precipitation method under well optimized synthesis (and aging)

conditions, following the procedure described in refs. [15,39] and

starting from two different solutions. Solution A, containing the

metallic species (Al, Mg and Cu/Ni/Co) in the desired molar ratios

as nitrates and not exceeding 1.5 of total cationic concentration;

and solution B, containing sodium hydroxide and sodium carbon-

ꢁ

2ꢁ

3

interlaminar space (in its natural state mainly OH and CO ). This

structure is common to many natural and synthetic materials hav-

2

+

3+

ꢁ

ing the same empirical formula [30–32]: M

a

M

b

(OH)2a+2b(X )2b

-

ꢂxH

2

O. During the calcination, the structure collapses and yields a

low-crystalline mixed oxide with cationic vacancies produced

due to the need of equilibrating the excess of positive charge intro-

2

ꢁ

ate in adequate amounts to accomplish a CO /(total cationic num-

3

_d_u_c_e_d_b_y_r_e_p_l_a_c_i_n_g_s_o_m_e_M_g2 by Al . Therefore, with respect to

+

3+

ber) ratio always equal to 0.66. These solutions were added at the

same rate (20 mL/h) to an empty plastic beaker with continuous

stirring at 200 rpm. The precipitates were aged at 60 °C overnight

in a thermostatic bath. The resulting products were ﬁltered and

washed thoroughly with deionized water until pH = 7. Depending

on the transition metal to be incorporated in the hydrotalcite, the

MgO, the material exhibits some stronger Lewis basic sites, corre-

2ꢁ

sponding to unsaturated O , and some weak acid sites, corre-

3

+

sponding to Al [32–34].

Among their many advantages, hydrotalcite-type materials

have an easy and low-cost synthesis procedure; they are non-

toxic and possess the capacity of easily modifying their physico-

chemical properties by means of controlling their composition.

Thus, amount and strength of both basic and acid sites can be

2

+

adapted to the catalytic process needs just by changing the M

M

/

3

+

molar ratio. In addition, different metals (others than Mg

and Al) can also be incorporated in the hydrotalcite precursor

Scheme 1. Selective dehydration of glycerol to acetol over Cu-Mg-Al mixed oxide.

1

62

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

corresponding Cu [Cu(NO

3

)

2

ꢂ2.5H

2

O], Ni [Ni(NO

3

)

2

ꢂ6H

2

O] or Co [Co

samples were pressed into self-supported wafers and treated at

ꢁ1

(

NO O] precursors were used. The hydrotalcites were cal-

3

)

2

ꢂ6H

2

300 °C in Nitrogen ﬂow (20 mLꢂmin ) during 1 h followed by

ꢁ4

cined at 550 °C during 6 h in air to yield the corresponding

metal-based mixed oxides used in the catalytic tests. Additionally,

and in the case of Cu-based samples, one material was reduced

evacuation at 10 mbar at 100 °C during 1 h, except in the ex situ

reduced sample which was activated in vacuum at 100 °C for

H

2

1 h. After activation the samples were cooled down to 25 °C under

dynamic vacuum conditions and exposed to 1.5 mbar of 1,2-

propanediol and/or hydroxyacetone followed by evacuation at

25 °C during 5 min. In the case of 1,2-propanediol the temperature

was risen under static vacuum conditions to 80 °C and 160 °C

acquiring IR spectra at 45 min of each temperature. Once achieved

2

under H ﬂow (100 mL/min) at 450 °C during 3 h, and another

amount of the same material was partially reduced at 200 °C dur-

ing 3 h, prior to their use in the catalytic experiments.

2.3. Catalyst characterization

4

5 min at 160 °C, the sample was evacuated and cooled down to

Hydrotalcite-type and metal mixed oxides materials were char-

acterized by ICP (inductively coupled plasma), with a Varian 715-

ES ICP-Optical Emission spectrometer, after solid dissolution in

25 °C where 10 mbar CO has been adsorbed for surface titration.

In order to avoid surface reduction by CO, IR spectra was immedi-

ately collected. IR spectra were processed using the Origin soft-

ware. For NO adsorption studies, after activation the samples

were cooled down to ꢁ156 °C under dynamic vacuum conditions

followed by NO dosing at increasing pressure (0.05–0.60 mbar).

IR spectra were recorded after each dosage. After maximum

NO dosing, the samples were evacuated under dynamic vacuum

HNO

determined by X-ray diffraction (XRD) in a Philips X’Pert MPD

diffractometer equipped with a PW3050 goniometer (CuK radia-

3

/HCl aqueous solution. Phase purity of the catalysts was

a

tion, graphite monochromator), provided with a variable diver-

gence slit and working in the ﬁxed irradiated area mode. The

obtained diffractograms were compared with those found at the

PDF2 database (codes in parenthesis). Surface areas of the solid

samples (200 mg) were calculated by applying the BET method

to the N adsorption isotherms obtained by carrying out liquid

2

nitrogen adsorption experiments at 77 K, in a Micromeritics ﬂow-

sorb apparatus.

ꢁ5

conditions at 10

times.

mbar and IR spectra acquired at controlled

High-resolution transmission electron microscopy (HR-TEM)

was done using a Jeol JEM-2100F operating at 200 kV. The micro-

scope was also equipped with a high-angle annular dark-ﬁeld

(HAADF) detector and an EDS X-Max 80 detector. HR micrograph

analysis, lattice spacing, First Fourier Transform (FFT) and phase

interpretation, were done with the Gatan Digital Micrograph soft-

ware (Gatan Inc.). Besides, thanks to the EDS detector, with a res-

olution of 127 eV, maps with different colours depending on the

element were obtained. Jeol JSM6300 operating at 20 kV and

equipped with an energy-dispersive X-ray spectrometer (EDX)

was used to obtain Screen Electron Microscopy (SEM) images and

compositional mappings.

The reduction behavior of copper in these materials was studied

by temperature-programmed reduction (TPR) analysis carried out

in a Micromeritics Autochem 2910 equipment. About 30 mg of

3

sample were initially ﬂushed with 30 cm /min of Ar at room tem-

2

perature for 30 min, and then a mixture of 10 vol% of H in Ar was

3

passed through the catalyst at a total ﬂow rate of 50 cm /min,

while the temperature was increased up to 1273 K at a heating rate

2

of 10 K/min. The H consumption rate was monitored in a thermal

conductivity detector (TCD) previously calibrated using the reduc-

tion of CuO as reference. In addition, X-ray photoelectron spec-

troscopy (XPS) was used to determine the oxidation states of

copper species at the catalyst surface. XPS data were collected on

a SPECS spectrometer equipped with a 150-MCD-9 detector and

Thermogravimetric analyses (TG) were carried out in a Mettler

Toledo TGA/SDTA 851 apparatus, using a heating rate of 10 °C/min

in an air stream until 800 °C was reached. Elemental analysis (EA)

was carried out in a Fisons EA1108CHN-S apparatus. These two

2

techniques (TG and EA) together with ICP, N adsorption and

using a non-monochromatic Mg K

Spectra were recorded at 25 °C, using an analyzer pass energy of

0 eV, an X-ray power of 50 W (to avoid photo-reduction) and

under an operating pressure of 10 mbar. During data processing

of the XPS spectra, binding energy (BE) values were referenced to

the Al 2p peak settled at 73.5 eV. Spectra treatment was performed

using the CASA software.

a

(1253.6 eV) X-ray source.

XRD measurements were applied to determine and evaluate the

organic content, the metal leaching, and the stability of metal

mixed oxide structures, respectively, after several consecutive uses

of the catalysts under reaction conditions.

3

ꢁ9

2.4. Catalytic tests

Acid and basic centers present in our materials were deter-

mined by temperature programmed desorption of ammonia

Catalytic experiments were performed in a stainless-steel tubu-

lar ﬁxed-bed reactor (length = 25 cm and diameter = 0.5 cm), with

the catalyst pelletized in particles 0.4–0.6 mm in size and diluted

with SiC (0.6–0.8 mm). Typically, catalytic tests were carried out

by feeding the reactor with a liquid mixture of glycerol (GLY)

and methanol (MeOH) (10:90 wt ratio) at temperatures between

220 and 280 °C during 9 h. For each experiment, cumulative frac-

tions corresponding to 0–1 h, 1–2 h, 2–3 h, etc. were collected in

a glass recipient submerged in an ice bath, and then analyzed by

HPLC (Shimadzu Nexera LC-20 AD XR, reverse phase, Hi-Plex H

Column 300 ꢃ 7.7 mm, and PL Hi-Plex Guard Column 50 ꢃ 7.7 m

m), and also by GC (Varian CP-3800, CARBOWAX Column

15 m ꢃ 3.2 mm). Product identiﬁcation was done by GC–MS (Agi-

lent 6890N GC System coupled with an Agilent 5973N mass

detector).

(

NH

out on a TPD/2900 apparatus from Micromeritics. Approximately,

.100 g of sample were pre-treated in an Ar stream at 300 °C dur-

3 2

-TPD) and carbon dioxide (CO -TPD), respectively, carried

0

ing 1 h. Ammonia was chemisorbed by pulses at 100 °C until equi-

librium was reached, while carbon dioxide was chemisorbed in the

same way at room temperature. Then, the sample was ﬂuxed with

helium stream during 15 min, prior to increase the temperature up

to either 500 °C (NH

00 mL/min and using a heating rate of 10 °C/min. Gas adsorption

was monitored with a calibrated thermal conductivity detector

TCD), representing a quantitative value. On the other hand, a

3 2

) or 650 °C (CO ) in a helium stream of

1

(

non-calibrated mass-spectrometer (MS) was used to follow ammo-

nia desorption, providing us with a qualitative information about

the strength distribution of acid sites.

With the aim of calculating conversions and, taking into

account that some few amounts of glycerol are retained on the cat-

alytic surface and part of it could be later transformed into coke,

the total yield to products was considered as the actual conversion,

as it gives more information about the ‘‘effective activity” of our

materials:

FT-IR spectrometry studies were recorded with a Bruker 70 V

ꢁ1

spectrometer using a DTGS detector and acquiring at 4 cm reso-

lution. An IR cell allowing ‘‘in situ” treatments in controlled atmo-

spheres and temperatures from 25 °C to 500 °C was connected to a

vacuum system with gas/liquid dosing facility. For IR studies the

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

163

P

f

n

a

i

in practically all samples. Only in the case of 5.0%Cu-HT-1 sample,

several diffraction peaks from aluminum hydroxide are observed.

This is probably due to the large amount of Al present in the ini-

product

i

X

glycerolð%Þ ¼

ꢂ 100

ð1Þ

0

glycerol

3+

Being a

i

the stoichiometric correction factor for the product_i.

tial solution, precipitated apart from the Cu-Mg-Al hydrotalcite

and unable to enter inside the hydrotalcite structure during the

co-precipitation process.

Selectivity to acetol (and to the different products) was also cal-

culated as a function of the total amount of products as follows:

After the calcination of the Cu-based HT-precursors, the corre-

sponding Cu-Mg-Al mixed oxides were attained, and their XRD

patterns are drawn in Fig. 1b. For Cu-based mixed oxides the

desired phase with poorly crystalline MgO structure containing

the cationic species in the appropriate proportions has been

obtained in practically all the measured materials, being necessary

to emphasize the greater level of crystallization when increasing

the Mg content. In the case of 5.0%Cu-HT-1 calcined sample, its

X-ray diffractogram exhibits some peaks corresponding to alu-

minum oxide, as expected after analyzing the diffractogram

belonging to the hydrotalcite precursor. It seems that after the cal-

f

n

acetol

S

acetolð%Þ ¼

ꢂ 100

ð2Þ

f

n

total products

Finally, gaseous products mainly derived from methanol con-

version in some speciﬁc pre-treatment experiments were analyzed

at the exit of the reactor. The gas composition was determined

using a GC ﬁtted with 3 detection lines: (i) one for H

on a 5A molecular sieve column (with Ar as carrier) and quantiﬁed

on a TCD, (ii) a second one for N separated on a 13X molecular

sieve column (with He as carrier) and quantiﬁed on a second

TCD, and iii) a third one equipped with Al-plot column (with He

as carrier) and a FID detector for hydrocarbons quantiﬁcation. This,

together, with the TG analysis (see Section 2.3) allowed us to cal-

culate the carbon balance of the reaction with respect to the total

amount of glycerol fed into the reactor.

2

separated

2

3+

cination process some Al from bayerite became part of the crystal

lattice but other remained apart, giving rise to this aluminum

oxide.

A second series of hydrotalcite-type materials having the same

proportion (5.0 wt% with respect to the calcined material) of differ-

ent transition metals, such as Cu, Co and Ni, and by ﬁxing the molar

P

f

n

ꢂ x

i

C atoms

product

i

2+

3+

CBglycerolð%Þ ¼

ꢂ 100

ð3Þ

ratio M /M equal to 4 were prepared and characterized. The

main physicochemical and textural properties of 5.0%Cu-HT-4,

0

n

ꢂ 3C atoms

glycerol

5

.0%Co-HT-4, and 5.0%Ni-HT-4 calcined samples are summarized

Being x

from glycerol reaction.

i

the number of carbon atoms in the product coming

in Table 2. Chemical composition of this series of materials is again

very similar to the theoretical one, while the attained surface areas

are close to 200 m /g in the three cases.

As can be seen in Figure S1a (see SI), the XRD patterns of the

three Cu-HT-4, Co-HT-4 and Ni-HT-4 as-prepared samples showed

the characteristic peaks corresponding to the formation of well-

crystallized layered double hydroxides of the hydrotalcite struc-

ture. Therefore, both the 5.0%Ni-HT-4 and the 5.0%Co-HT-4 sam-

2

.5. Re-usability tests

After the reaction, the solid catalyst was washed with 20 mL of

methanol (2 mL/min) in the same reactor, and then calcined at

50 °C during 6 h in air. The catalytic experiments together with

the analytics corresponding to the re-usability tests were per-

formed in the same way as the common experiments already

described at the beginning of Section 2.4.

5

2

+

3+

ples contain the majority of either Ni or Co ions in the

2

+

3+

hydrotalcite structure partially substituting Mg and Al ions,

respectively. In the case of Ni- and Co-based mixed oxides obtained

after calcination of hydrotalcite precursors, the X-ray diffrac-

tograms of Figure S1b (see SI) only show the peaks corresponding

to the formation of the MgO structure. This meaning that Ni and

Co ions are well dispersed on the Mg-Al-O structure in both

3

. Results and discussions

2+

3

.1. Catalysts characterization

3+

cases.

A series of hydrotalcite-type (HTs) materials with different

II

III

Finally, different hydrotalcite-type materials with a M /M

molar ratio of ꢀ4 possessing different Cu loadings (from 1.0 to

2.8 wt%) were synthesized and characterized. The main physico-

2

+

3+

molar ratio M /M , copper loading close to 5 wt%, and surface

2

areas of around 200 m /g were ﬁrstly synthesized and well charac-

1

terized. Table 1 shows the main physicochemical and textural

properties of Cu-based hydrotalcite-derived materials after

calcination.

The structural characteristics of the Cu-based hydrotalcite-type

materials were analyzed by XRD measurements. For the non-

calcined Cu-HTs samples, the X-ray diffractograms of Fig. 1 show

that a common Mg/Al layered hydrotalcite structure is obtained

chemical and textural properties of these Cu-based HT-4 calcined

samples are detailed in Table 3, showing that the Cu was incorpo-

rated in the expected amounts into the hydrotalcite precursors,

which maintain a M /M molar ratio around 4 and surface areas

close to 200 m /g.

The XRD patterns of the Cu-based materials containing different

Cu loadings before and after calcination treatment are depicted in

Figure S2a and S2b (see SI), respectively. For these Cu-based series

of samples, both the desired hydrotalcite phase and the mixed

oxide phase with MgO structure are observed for the non-

calcined and calcined material, respectively. Interestingly, even

for the materials containing quantities of copper above 5.0 wt%

non-isolated copper species are noticed.

II

III

2

Table 1

Main physicochemical and textural properties of Cu-based HT-derived calcined

II

III

materials with different M /M ratios.

II

III

Catalyst Cu content

M /M

Surface area

(m /g) [BET method]

a

2

b

(wt%)

molar ratio

In good agreement with what has been observed by XRD,

microscopy analyses (SEM and HR-TEM) of the materials 5.0%Cu-

HT-4, 10.0%Cu-HT-4 and 12.0%Cu-HT-4 (Fig. 2, and Figs. S6–8 in

SI) conﬁrm that the materials were, mostly, homogeneous, with a

MgO structure, also showing a very high copper dispersion inside

this crystalline structure. However, a very careful analysis by using

5

.0%Cu-HT-1

.0%Cu-HT-2

.0%Cu-HT-3

.0%Cu-HT-4

.0%Cu-HT-5

.0%Cu-HT-6

4.9

4.6

4.9

5.1

4.8

1.0

2.3

3.3

4.1

5.5

6.3

218

245

212

205

224

177

a

Cu content and chemical composition measured by ICP.-

Values calculated from N adsorption isotherms by applying the BET method.

2

HR-TEM allowed us to detect CuO

x

nanoparticles in 12.0%Cu-HT-4

b

(Figure S16b). For materials containing lower copper loadings,

1

64

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

II

III

Fig. 1. (a) XRD patterns of as-synthesized Cu-based hydrotalcites with different M /M ratios (the ‘‘s” at the end of the name indicates a material analyzed just after the

II

III

synthesis). (b) XRD patterns of Cu-based hydrotalcite-derived mixed oxides with different M /M ratios.

3.2. Catalytic results in continuous ﬂow ﬁxed-bed reactor

Table 2

Main physicochemical and textural properties of HT-derived calcined materials

containing different transition metals.

Catalytic experiments for selective dehydration of glycerol

GLY) to acetol were carried out as described in Experimental Sec-

tion 2.4. In a ﬁrst attempt aimed at optimizing the most important

reaction conditions, different operational parameters, such as glyc-

erol dilution and temperature, were varied by working always with

(

II

III

Catalyst

Metal content

wt%)^a

M /M

Surface area

(m /g) [BET method]

a

2

b

(

molar ratio

5

.0%Cu-HT-4

.0%Co-HT-4

.0%Ni-HT-4

4.9

5.1

4.1

4.2

205

198

190

5

.0%Cu-HT-4 calcined material as catalyst. The attained results in

terms of catalytic activity (i.e. glycerol conversion), stability (main-

tenance of conversion with TOS), and more importantly, selectivity

to acetol, were evaluated, main data being summarized in Table S1

a

Metal (Cu, Co or Ni) content and chemical composition measured by ICP.

Values calculated from N adsorption isotherms by applying the BET method.-

2

b

(

see SI) together with its more relevant discussion. From these

results, the mixture MeOH/GLY = 90:10 (weight ratio) was selected

as the optimal for further studies because it allows the system to

work up to longer TOS with higher productivities of acetol, also

maintaining a good catalyst stability. Once the optimum dilution

degree of glycerol in methanol had been stablished, a series of

experiments were done to ﬁgure out which temperature could

improve both, glycerol conversion and selectivity to acetol. Tests

of selective dehydration of glycerol over 5.0%Cu-HT-4 catalyst

were carried out by varying temperature from 220 to 280 °C, keep-

ing constant all the other reaction parameters (feed: MeOH/

GLY = 90/10, ﬂow = 2 mL/h, catalyst = 0.5 g, TOS = 9 h). The attained

results are depicted in Figure S3 in terms of average conversion

Table 3

Main physicochemical and textural properties of Cu-based HT-derived calcined

materials with different copper loadings.

II

III

Catalyst

Cu content

wt%)^a

M /M

Surface area

(m /g) [BET method]

a

2

b

(

molar ratio

HT-4

–

4.3

4.4

4.1

4.2

4.1

249

196

199

205

193

196

186

1

2

5

7

1

.0%Cu-HT-4

.5%Cu-HT-4

.0%Cu-HT-4

0.0%Cu-HT-4

2.0%Cu-HT-4

1.0

2.4

4.9

6.7

9.9

12.8

(

cumulative conversion of glycerol during all the experiment,

a

Cu content and chemical composition measured by ICP.

Values calculated from N adsorption isotherms by applying the BET method.

2

TOS = 1–9 h) and average selectivity to acetol (cumulative acetol

selectivity during all reaction, TOS = 1–9 h). The temperature of

b

2

40 °C turned out to be the best by a small difference in terms of

glycerol conversion and clearly better as far as selectivity to acetol

is concerned during most part of the reaction. No great differences

were noticed with respect to the catalyst stability.

Other important parameter to be considered is the particle size

of the catalyst, which will determine whether we have our reaction

being dominated by internal diffusion processes across the parti-

cle. We found that the internal diffusion controls the reaction

when working with particle sizes above 0.600 mm, being the value

those nanoparticles were not evident with this technique but,

when using a more sensitive bulk technique such as NO adsorption

(

x

see Figure S9), again isolated CuO species became visible.

Nonetheless, the XRD data together with the microscopic measure-

ments evidence that most of the copper integrates the crystalline

lattice in materials containing less than 12.0 wt% of Cu.

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

165

reactor. Catalytic results in terms of glycerol conversion and selec-

tivity to acetol (at 240 °C and TOS = 4 h) attained for Cu-Mg-Al

mixed oxides (5.0%-HTs) with different (Cu + Mg)/Al ratios are

2 3

depicted in Fig. 4 and compared with CuO/Al O and CuO/MgO

(

with 5 wt% Cu in both cases), as reference materials. In addition,

the average both, glycerol conversion and acetol selectivity

obtained for the different Cu-Mg-Al mixed oxides here studied dur-

ing 9 h of reactor operation at 240 °C are also shown in Fig. 5. As

II

III

can be seen, M /M molar ratios around 4 in the catalysts offered

the highest both glycerol conversion and acetol selectivity (highest

yield of acetol) achieved. The glycerol conversion and, more impor-

tantly, the acetol selectivity reached with 5.0%Cu-HTs catalysts

with Mg/Al ratios from 3 to 5 are higher than the values achieved

2 3

with CuO/Al O reference material, while CuO/MgO material

offered the worse results. However, the catalyst stability makes

the difference between these values since Cu-Mg-Al materials with

II

III

higher M /M molar ratios ((Cu + Mg)/Al ratios ꢄ4) are slightly

more capable to keep the catalytic performance over at least 9 h.

Regarding the acid properties of the Cu-Mg-Al samples, looking

3

at the NH -TPD proﬁles (Fig. 6a) two different peaks can be distin-

guished, the ﬁrst one appearing at temperatures around 200 °C,

and the second at temperatures above 350 °C. As it has been

reported for calcined hydrotalcites, the acidity in-between this

range can be ascribed to weak and moderate Lewis acid sites cor-

3

+

responding mainly to different Al species [40,41], although a con-

tribution of copper to the moderate acid sites cannot be discarded

[

42]. As far as the basic sites is concerned, the two peaks at lower

temperatures appearing in Fig. 6b could be assigned to MgO (weak,

2ꢁ

physisorbed CO

CO ). Similar sites can be found in MgO [43].

Therefore, an explanation to the higher performance of these

2

) and O unsaturated (moderate, chemisorbed

2

II

III

mixed oxides having M /M molar ratios between 3 and 4 could

be found in the different distribution of active sites exhibited by

each material, being more beneﬁcial a material containing mainly

weak Lewis acid sites and moderate Lewis basic sites. This ade-

quate combination of Lewis acid and basic sites can be encoun-

Fig. 2. HR-TEM (1) and STEM (2) images of (a) 5.0%Cu-HT-4, (b) 10.0%Cu-HT-4, (c)

&

1

2.0%Cu-HT-4. ( ) Copper detected by EDS mapping.

II

III

tered in the aforementioned samples with M /M molar ratios

3-4 (see Fig. 6 and Table 4). On the contrary, higher ratios make

ꢀ

the material lose both total acidity (concretely weak centers) and

basicity, the former because of the lower aluminum loading, the

latter due to the decrease of the vacancies in the mixed oxide

structure produced by the presence of aluminum [34]. Neverthe-

less, the MgAl mixed oxides possessing a combination of Lewis

both basic and acid sites but without Cu in the structure are not

capable to perform the selective dehydration of glycerol to acetol.

Thus, the presence of Cu is essential to carry out this process.

The presence of Cu in the hydrotalcite-derived mixed oxide pro-

vokes a decrease in the acidity and also in the basicity of the sam-

ples compared to the MgAl mixed oxides (see Table 4). This

behavior can be explained by different effects occurring simultane-

0

.425–0.600 the smallest one to be chosen to avoid load losses and

reactor seals (see Figure S4 in SI). In addition, some experiments

varying the liquid feed ﬂow (q) along with the catalyst loading

(

W) were accomplished to prove that the liquid ﬂow correspond-

ing to 2 mL/h is in the area free of mass transfer limitations (see

Figure S5 in SI). Thus, the set of experiments aimed to evaluate

the catalytic performance of different hydrotalcite-derived mixed

oxides can be carried out under these conditions (240 °C, MeOH/

GLY = 90:10 wt, particle size = 0.425–0.600 mm, 2 mL/h). More

information corresponding to these experiments can be found in

the supporting information (SI).

The product distribution for the glycerol dehydration reaction

under the aforementioned conditions and the reaction network

proposed can be found in Fig. 3 and Scheme 2 (appearing in color

all the identiﬁed compounds), respectively.

2

+

2+

ously. When Cu replacing Mg in the mixed oxide structure is

3

+

close to Al sites, Cu acts as the base but the acid site related to

the Al is less acid with respect to the corresponding site in the only

presence of Mg. Thus, the acidity (mainly week acidity) of the Cu-

HT material is lower with respect to the acidity observed in the

analogous HT sample without Cu (see also Table 5). In addition,

II

III

3.3. Optimizing the M /M molar ratio in Cu-based catalysts

2

+

2+

by replacing Mg by Cu in the mixed oxide structure the amount

and strength of basic sites are diminished in the material, mainly

because CuO species have lower basicity than MgO species (see

Table 5). Summarizing, the presence of Cu in MgAl mixed oxides

lead to a decrease in the acidity and the basicity of the materials,

being this effect more pronounced in the case of the basic proper-

As it has been already explained, thermal decomposition of Mg-

Al hydrotalcite precursor yields a high surface area Mg-Al mixed

oxide, which mainly exposes strong Lewis basic sites and some

weak acid sites. The acid-base properties of these sites depend

II

III

on the M /M molar ratio, i.e. mainly the Mg-Al ratio, in the hydro-

II

III

2+

talcite precursor [32–34]. With the aim of optimizing this M /M

ties. It is also possible that the Lewis acidity of the Cu species

ratios in Cu-Mg-Al mixed oxides, hydrotalcites with (Cu + Mg)/Al

ratios from 1 to 6 were prepared, calcined and tested in the selec-

tive dehydration of glycerol in continuous ﬂow ﬁxed-bed catalytic

could contribute to the total acidity presented by the Cu-HT sam-

ples, but this contribution should be quite minor in comparison to

the amount of Al contained in the samples. In that case, the effect

1

66

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

Fig. 3. (a) Glycerol conversion and selectivity to acetol with TOS under the optimal reaction conditions over the 5.0%Cu-HT-4 catalyst (b) Product distribution and carbon

balance: ^aCarbon balance calculated with the total amount of products quantiﬁed during reaction, and ^bCarbon balance calculated including also the carbonaceous matter

deposited on catalytic surface and the gas products coming from glycerol. (Reaction conditions: feed: MeOH/GLY = 90/10 wt, ﬂow = 2 mL/h, temperature = 240 °C,

catalyst = 0.5 g).

Scheme 2. Reaction network describing the different pathways when using 0.5 g of 5.0% Cu-HT-4 as catalyst under the following reaction conditions; T = 240C, Feed MeOH/

Gly (9.1 wt), F = 2 mL/h, p.size = 0.425–0.600 mm. Colors correspond with detected products and match with colors of Fig. 3. (For interpretation of the references to colour in

this ﬁgure legend, the reader is referred to the web version of this article.)

could be more relevant only in the case of Cu-HT samples with very

low amount of Al (i.e. Cu-HT-6).

ate basic sites presents a medium activity (in terms of glycerol con-

version) between Al -based (mainly acid) and MgO-based

2 3

O

In summary, taking also into account the differences observed

in Figs. 4–6, where the comparison between our Cu-based materi-

als and CuO supported on Al O and MgO, as well as with the pure

2 3

MgAl mixed oxide (HT-4, Mg/Al = 4) is shown, a 5.0%Cu-HT-4

material having an adequate combination of weak acid and moder-

(basic) materials. With this 5.0%Cu-HT-4 material the best catalytic

results (maximum values of selectivity and yield to acetol) can be

reached. Therefore, weak acid sites corresponding to Al³⁺can be

concluded as the main responsible of the glycerol conversion as

they are almost the only one present on the most active material

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

167

(

2 3

CuO/Al O ). This fact can be conﬁrmed by the activity reached for

MgO, without any acid site that was the worst among all the mate-

rials tested. In this sense, a compromise between acid and basic

centers present in the solid seems to be needed for obtaining a

good yield to acetol. On one hand, low Mg/Al molar ratios in the

catalyst do not offer the necessary basicity to perform the reaction

with good selectivity. On the other hand, higher Mg/Al ratios as

well as pure MgO lack the acid site density needed to carry out

the reaction with the desired acetol production.

Last but not least, and speaking about catalytic deactivation

detected in these Cu-based catalysts, one possible explanation that

ﬁts in the tendency observed for these HT-derived materials would

be the deactivation occurring on the weak acid sites, which ratio

increases with decreasing of (Cu + Mg)/Al in the solids. However,

with the handled data, the deactivation on the weak basic sites

cannot be discarded either. Actually, it is impossible neither to dis-

tinguish between both options nor to discard any of them. Besides,

there are literature reports making both acid and basic sites

responsible for the catalytic deactivation in similar reactions, as

coke formation would occur on acid sites, and polyglycerols or

acetalization products of glycerol would be formed and remain

adsorbed on basic sites [45].

Fig. 4. Glycerol conversion, selectivity and yield to acetol in the selective

dehydration of glycerol over 5.0%Cu-HT catalysts with different (Cu + Mg)/Al molar

ratios at TOS = 4 h. (Reaction conditions: feed: MeOH/GLY = 90/10 wt, ﬂow = 2 mL/h,

temperature = 240 °C, catalyst = 0.5 g).

Finally, as it was previously mentioned in this section, it is

worth noting that the MgAl mixed oxide (HT-4), despite offering

both weak acid and moderate basic centers, is not able to carry

out the reaction in almost any extent (see Table 5), which unveils

the need of having copper to interplay with glycerol through some

interaction beyond a simple acid–base effect. This fact will be the

subject of further discussions in next sections.

3.4. The role of Cu in Cu-Mg-Al mixed oxides

Apart from the results above-mentioned, when using other

divalent transition metals with redox properties and a proven

capability to produce acetol in the glycerol hydrogenolysis, such

as nickel or cobalt [35,36,46,47], the results also revealed the need

of using copper in this system (see SI, Table S3 and the explanatory

text). This is in good agreement with what the majority of litera-

ture reports about glycerol transformation to propane-diols on

the fact that copper is the essential metal when working with

Fig. 5. Average glycerol conversion and selectivity to acetol in the selective

dehydration of glycerol over 5.0%Cu-HT catalysts with different (Cu + Mg)/Al molar

ratios during TOS = 1–9 h. (Reaction conditions: feed: MeOH/GLY = 90/10 wt,

ﬂow = 2 mL/h, temperature = 240 °C, catalyst = 0.5 g).

II

III

Fig. 6. NH

3 2 x

-TPD (a) and CO -TPD (b) proﬁles of Cu-Mg-Al hydrotalcite-derived mixed oxides with different M /M ratios. Note: Baseline drift in HT-4 sample due to NO

formation (T ꢄ 450 °C).

1

68

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

Table 4

Quantitative data for temperature programmed desorption (NH

II

III

3

2

-TPD and CO -TPD) of Cu-Mg-Al hydrotalcite-derived materials with different M /M molar ratios.

Catalyst

CuO/Al

Acid site density (mmol/g)

Basic site density (mmol/g)

Acid sites (weak/medium)

Basic sites (weak/medium)

2

O

3

102

122

87

63

–

33

100

77:23

66:34

21:79

44:56

39:61

5

.0%Cu-HT-1

.0%Cu-HT-4

.0%Cu-HT-6

111

159

153

111

68:32

42:58

37:63

a

CuO/MgO

-

a

MgO can be considered as having almost zero acidity [44].

Table 5

3 2

Comparison of NH -TPD and CO -TPD quantitative data and catalytic activity of hydrotalcite-derived Cu-Mg-Al and Mg-Al mixed oxides with the same Mg/Al ratio.

Catalyst

Acid sites (mmol/g)

weak:medium)

Basic sites (mmol/g)

(weak:medium)

GLY conversion^a

Selectivity to acetol^a

(

5

HT-4

.0%Cu-HT-4

87 (42:58)

124 (56:44)

159 (21:79)

245 (31:69)

64.1

6.9

52.2

39.0

a

Average glycerol conversion and selectivity to acetol during TOS = 1–9 h. (Reaction conditions: feed: MeOH/GLY = 90/10 wt, ﬂow = 2 mL/h, temperature = 240 °C,

catalyst = 0.5 g).

metallic mixed oxides by using a ﬁxed-bed reactor under close

related reaction conditions [38,47]. Thus, the study of the different

Cu species present in the Cu-based mixed oxide materials and how

they speciﬁcally interact with the glycerol to selectively produce

acetol becomes crucial to understand the catalytic process and

the role of Cu in the solid material.

Cu²⁺and to reduced Cu¹⁺, respectively. The presence of reduced

copper is conﬁrmed by the value of the modiﬁed Auger Parameter

at 1849.6 eV. Finally, in the ‘‘in situ” H reduced sample (5.0%Cu-

2

HT-4R-INSITU) two main peaks corresponding to Cu reduced spe-

cies can be distinguished at 928.9 and 932.2 eV, and also one

2

+

shoulder at 935.5 (Cu ). The value of the modiﬁed Auger Parame-

0

1+

ter at 1850.2 eV unveils the majority presence of Cu versus Cu

.

In addition, the splitting of the Cu2p3/2 XPS peak into two compo-

nents (928.9 and 932.2 eV) is related to differential charging of the

sample because of the presence of Cu metal species in different

environments and/or particle sizes. Small amounts of Cu²⁺are also

detected, based on the XPS component at BE of 935.5 eV and the

shake-up peak.

3

.4.1. Effect of the oxidation state of Cu in Cu-Mg-Al mixed oxides

It has been reported that the coexistence of Cu , Cu and Cu

0

1+

2+

species in a partially reduced Cu-Al oxide catalyst is responsible for

its multifunctional role, thus catalyzing, for example, the simulta-

neous dehydration to acetol and the hydrogenolysis (hydrogena-

tion) to 1,2-propanediol [38]. Therefore, the effects of having

different copper species in our Cu-based mixed oxides were

assessed by means of a detailed XPS study.

XPS experiments were performed by using: a) 5.0%Cu-HT-4, a

fresh calcined material; b) 5.0%Cu-HT-4U, the same material recov-

ered after reaction; c) 5.0%Cu-HT-4R, again the same material but

1

+

The presence of Cu in the 5.0%Cu-HT-4U used catalyst evi-

2

+

dences that the reduction of Cu species present in the fresh (cal-

cined) sample is taking place during the reaction. However, since

the used catalyst has been analysed after being exposed to atmo-

0

spheric air and metallic copper (Cu ) is easily oxidized into either

1

+

2+

Cu or Cu [50], the extent of copper species reduction through

this time reduced ‘‘ex situ” at 450 °C under H

without having been tested in reaction; and ﬁnally d) 5.0%Cu-

HT-4R-INSITU, the same material but reduced at 450 °C under H

2

atmosphere and

this ‘‘on reaction” reduction is difﬁcult to deﬁne. The TPR analysis

of the fresh (calcined) material and the ‘‘ex situ” H

rial (see Figure S12) also conﬁrms the presence of Cu species

peak at 285 °C) [51,52] in the former (5.0%Cu-HT-4) and segre-

2

reduced mate-

2

2+

atmosphere in the instrument’s own cell, to accurate that sample

has not had contact with atmospheric air before being analysed.

The attained results of XPS measurements are shown in Fig. 7

and summarized in Table 6. In addition, Figure S11 in the SI pre-

sents the XPS data attained by studying the stability (reduction)

of the oxidation state of Cu on solid samples under measurement

conditions. The results allow discarding the reduction of Cu species

during these experiments.

(

2

+

1+

gated Cu and/or Cu species (peak at 190 °C) [53,54] in the later

1+

(

5.0%Cu-HT-4R). Probably, both Cu⁰and Cu are present in the

0

reduced material 5.0%Cu-HT-4R. Cu would be formed during the

2

reduction with H at 450 °C (see XPS Cu-HT4-R-INSITU, Fig. 7d),

1

+

0

1+

whereas Cu would result from the oxidation of Cu to Cu in

the presence of air (see XPS Cu-HT4-R, Fig. 7c). And, if the used

material (5.0%Cu-HT4U) also suffers a reduction during the glyc-

erol dehydration process, as it seems, this could also be applied

to it.

In the fresh sample the Cu2p3/2 main peak appears at 935.7 eV

together with a shoulder at lower BE (932.6 eV). The presence of

the shake-up satellite (s/m = 0.45) and the corresponding modiﬁed

2

+

With the aim of setting up if these reduced copper species

detected in Cu-Mg-Al mixed oxides are active or not in the selec-

Auger Parameter of 1850.9 eV conﬁrm the assignation as Cu pre-

dominately [48], where the lower BE is related to a minor amount

_o_f_r_e_d_u_c_e_d_c_o_p_p_e_r_s_p_e_c_i_e_s₍_C_u1 /Cu ). The high BE shift of the Cu

+

0

2+

tive dehydration of glycerol, the 5.0%Cu-HT-4R (‘‘ex situ” H

2

1

+

reduced) material having mainly Cu species (see XPS data,

Fig. 7 and Table 6) was tested in reaction, ﬁnding very similar

activity (glycerol conversion) and selectivity to acetol values to

those obtained with the original 5.0%Cu-HT-4 calcined or un-

reduced material (see Fig. 8). In addition, no signiﬁcantly faster

deactivation occurred at larger TOS for the reduced material com-

pared to the calcined one. Another experiment with the 5.0%Cu-

HT-4R-INSITU material (reduced in the same ﬁx-bed catalytic reac-

ion (935.7 eV) is due to ﬁnal state effects related to their high dis-

persion in the HT matrix [49]. An opposite situation is observed in

the case of the used material (5.0%Cu-HT-4U), where the main

peak appears at low BE (932.9 eV) together with a shoulder at high

BE (935.6 eV). The lower shake-up intensity (s/m = 0.30) and the

value of the modiﬁed Auger Parameter of 1849.5 eV indicate the

1+

presence of reduced copper species Cu predominately, whereas

_t_h_e_s_h_o_u_l_d_e_r_a_t_h_i_g_h_-_B_E_c_o_r_r_e_s_p_o_n_d_s_t_o_C_u2+. On the other hand,

tor at 450 °C under H

2

atmosphere during 4 h, prior to the reac-

2

in the case of the ‘‘ex situ” H reduced sample (5.0%Cu-HT-4R)

0

tion), having mainly Cu (see XPS data, Fig. 7 and Table 6), was

two peaks at 935.6 eV and 933.5 eV are observed, associated to

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

169

Fig. 7. Cu2p3/2 XPS peak of (a) 5.0%Cu-HT-4 (fresh, calcined), (b) 5.0%Cu-HT-4U (used), (c) 5.0%Cu-HT-4R (H

‘in situ”), (e) L 45 Cu Auger peak for the studied Cu-Mg-Al samples.

2

reduced ‘‘ex situ”), (d) 5.0%Cu-HT-4R-INSITU (H

2

reduced

‘

3 45

M M

Table 6

XPS data of the Cu2p3/2 core level and the surface composition.

a

Catalyst

Cu (2p3/2

B.E. (eV

Cu²⁺

)

a

Molar ratio

Cu²⁺

Molar ratio

Cu: Al: Mg

(eV)

Cu¹⁺^/⁰

s/m

5

.0%Cu-HT-4

.0%Cu-HT-4U

.0%Cu-HT-4R

.0%Cu-HT-4R-INSITU

935.7

935.6

935.5

932.6

932.9

933.5

932.3/928.9

0.45

0.30

0.42

0.12

1850.9

1849.5

1849.6

1850.2

84

47

20

16

53

80

2.4: 58.3: 39.3

2.1: 56.2: 41.7

1.6: 56.5: 41.8

1.2: 57.0: 41.7

a

Cu2p3/2 XPS (BE) + L

3

M

45

M45 CuAES (KE).

carried out. In this case, a signiﬁcant decrease in glycerol conver-

sion was observed, falling the average value from ꢀ65% to ꢀ45%

with respect to the other two samples discussed above. As for

the acetol selectivity, also a lower value was obtained (from

carbons, while liquid composition was determined by the standard

GC-FID system (see Experimental section).

For the ﬁrst experience, the reactor previously charged with

5.0%Cu-HT-4 catalyst was fed for one hour only with absolute

methanol (MeOH pre-treatment). The thus obtained results are

shown in Fig. 9a. As can be seen, ꢀ30% of methanol fed is converted

in the presence of Cu-Mg-Al material (mainly containing Cu²⁺spe-

ꢀ

50% to ꢀ45% in average) over this reduced sample mainly pos-

0

sessing Cu species.

From the results obtained up to now, it is reasonable to think

2

+

that the Cu species originally present in the 5.0%Cu-HT-4 (cal-

cined) catalyst at the beginning of the reaction are reduced

cies) into gaseous products, such as CO, CO

very low amounts of methyl-formate.

2 2

, and H , together with

1

+

0

‘

‘in situ” to Cu and/or Cu species throughout the glycerol dehy-

Then, the same 5.0%Cu-HT-4 catalyst (already pre-treated with

MeOH at 200 °C) was fed with a mixture of GLY/Water (10/90 in

weight) during 6 h, and the catalytic results were compared with

another experiment performed by feeding a fresh 5.0%Cu-HT-4C

catalyst (without the previous pre-treatment with methanol) with

the same GLY/Water (10/90 wt) mixture during 6 h as well. The

attained results are depicted in Fig. 9b, and they make clear the fact

that the pre-treatment had beneﬁcial effects on the catalytic per-

formance. This could be attributed to the presence of either Cu(I)

or Cu(0) instead of Cu(II) species, thanks to the hydrogen formation

when feeding with MeOH during the catalyst’s pre-treatment (see

Fig. 9a). Thus, these copper reduced species would be more active

1

+

dration process. The experiments herein described have set Cu

2

+

and Cu appearing to be more active for this reaction than metallic

copper, being the catalyst when the Cu is found mainly as Cu⁰

much less active and slightly less selective towards acetol com-

pared to the solids having majorly cationic Cu species.

3.4.2. Additional tests and measurements

In order to provide further insights into the role of copper spe-

cies in the Cu-Mg-Al mixed oxides during the catalytic selective

dehydration of glycerol in continuous ﬂow ﬁx-bed reactor, three

additional experiments were carried out and the liquid and gas-

eous products were analyzed at the exit of the reactor. Gas compo-

sition was determined by using a speciﬁc GC-TCD system ﬁtted

2

+

1+

than Cu species for this reaction and, since Cu has already been

0

1+

found to be more active than Cu (Section 3.4.1, Fig. 8), Cu can be

concluded as the most active species for this reaction. These evi-

2 2

with 3 detection lines for determination of H , N and light hydro-

1

70

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

Fig. 8. Glycerol conversion (a) and selectivity to acetol (b) for 5.0%Cu-HT-4, 5.0%Cu-HT-4R and 5.0%Cu-HT-4R-INSITU materials. (Reaction conditions: Temperature = 240 °C,

feed: MeOH/GLY (90/10 w.), ﬂow = 2 mL/h, catalyst: 0.5 g).

Fig. 9. (a) Pre-treatment test of 5.0%Cu-HT-4C catalyst with methanol (Reaction conditions: Feed: methanol, temperature = 240 °C, ﬂow = 2 mL/h, catalyst: 0.5 g). (b) Catalytic

results of selective glycerol dehydration over 5.0%Cu-HT-4 with and without methanol pre-treatment (Reaction conditions: Feed: water/glycerol (90/10 wt),

temperature = 240 °C, ﬂow = 2 mL/h, catalyst: 0.5 g).

dences are in good agreement with some recent results reported by

Batiot-Dupeyrat and co-workers based on catalytic results together

with DFT calculations [55].

TPR measurements of the 5.0%Cu-HT-4 calcined sample before

and after pre-treatment with methanol (at 200 °C) unveils the oxi-

dation of Cu(II) when comparing with the original calcined sample

(Fig. 10). Therefore, it is reasonable to think that the methanol

feeding used throughout this work is able to make the correspond-

ing Cu(II) ? Cu(I), Cu(0) transformation. In fact, during the pre-

treatment with MeOH, approximately 25 mmol of H were pro-

2

duced, as 0.4 mmol of copper are present in our catalyst, this quan-

tity could be enough to reduce most of the Cu²species inside the

catalyst.

+

In summary, under the reaction conditions here employed, the

Cu²⁺species present in the catalyst undergo the corresponding

1

+

0

reduction to Cu (and Cu ) species, thanks to the presence of

1

+

MeOH, favoring Cu species the catalytic selective dehydration

of glycerol to acetol. Although it is hard to distinguish between

the presence of Cu and Cu due to the oxidation behavior of cop-

per once exposed to the air, results shown in Fig. 8 point out Cu

as more active species to carry out the reaction than Cu .

1

+

0

1

+

0

Fig. 10. TPR traces for 5.0%Cu-HT-4 (calcined, un-reduced) and 5.0%Cu-HT-4R

calcined + reduced at 200 °C with MeOH).

(

3.4.3. FT-IR measurements

In order to explain the catalytic behavior of the 5.0%Cu-HT-4

sample, and to provide further information about the reaction

mechanism and the role of copper in the reaction pathways, IR

spectroscopic studies have been performed using 1,2-propanodiol

as probe molecule instead of glycerol. The selection of 1,2-

propanediol (boiling point = 214 °C) as probe molecule respond

only to an experimental reason, since the extremely high boiling

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

171

point of glycerol (290 °C) made impossible to properly work with it

in gas phase in our FTIR system. In addition, the HT-4 (Mg/Al = 4),

acid sites takes place on Ni and Co catalysts. Moreover, using CO as

a probe molecule, Cu species have been evidenced in the IR stud-

1

+

5

.0%Ni-HT-4 and 5.0%Co-HT-4 samples displaying much lower cat-

ies after 1,2-propanediol adsorption at 160 °C on the 5.0%Cu-HT-4

sample (IR band at 2098 cm )[56], which may be related to Cu

ꢁ1

2+

alytic activity have also been analyzed. 1,2-propanodiol has been

adsorbed at 25 °C on all samples until saturation coverage, and

then the temperature has been stepwise increased from 25 °C to

reduction by the aforementioned Cu-hydride interaction (Fig. 13).

1

+

2+/1+

In opposite, no reduced species Ni nor Co

have been observed

1

60 °C, acquiring IR spectra at each temperature.

Gas phase 1,2 propanediol shows a complex set of IR bands in

in the 5.0%Ni-HT-4 and 5.0%Co-HT-4 samples under similar condi-

tions. This fact was fully expected considering the very low

2+

the low frequency region with maxima at 1654, 1137, 1079,

reducibility of these samples, as very low amounts of Ni and

ꢁ1

3+

1

043, 989, 924 and 839 cm . After adsorption at 25 °C, the IR

bands corresponding to the CAO vibration (1137, 1079 and

043 cm ) are shifted to higher frequencies (1143, 1085 and

Co are available to be reduced at low temperatures (see TPR, Fig-

m

ure S13 in SI). Thus, it can be concluded that weak acid centers are

important because they are 1,2-propanediol adsorption centers

ꢁ1

1

ꢁ1

058 cm ) on all samples, in addition to a corresponding shift

3

(see NH -TPD, Figure S10 in SI), and that the additional presence

ꢁ1

2+

3+

in the OH band of all samples (3730 cm ) to lower frequency

of reducible centers, such as Cu (but not Ni and Co ), could

ꢁ

(Fig. 11). Since the interaction of 1,2 propanediol is relatively

favor the abstraction of H , thereby accelerating the carbonyl

ꢁ6

strong in all cases (i.e. stable toward evacuation at 10

mbar

group formation pathway.

2

+

1+

and 25 °C) the red shift of the

an interaction with Lewis acid sites, besides that with OH groups.

In addition, the shift in the CAO IR bands is the same in all sam-

ples, indicating similar adsorption sites in all samples. This ﬁts

with what has been already pointed out about the weak acid sites

mCAO IR bands may correspond to

Once Cu and Cu are identiﬁed as the two main species pre-

sent in the catalyst when the reaction is going on, and to assess if

2

+

1+

m

Cu or Cu are involved in the possible intermediate formation

step, the 5.0%Cu-HT-4 sample has been ‘‘ex situ” reduced in H ﬂow

2

1

+

(containing mainly Cu , see XPS data of 5.0%Cu-HT-4R in Fig. 7)

and analyzed in the IR study. As seen in the Cu-based non-

3

+

corresponding to Al as the main responsible of the catalytic activ-

ity. Since these acid sites are present in all the studied materials

ꢁ1

reduced sample, a band at 1705 cm due to the presence of car-

bonyl group has been observed in the 1,2-propanediol temperature

dependent reaction proﬁle. Nevertheless, the onset of carbonyl

group formation appears at a slightly lower temperature, i.e.

120 °C, in the ‘‘ex situ” reduced sample versus 160 °C in the calcined

(

3

see NH -TPD measurements, Figure S10 in SI), they could be the

main responsible of the reactant adsorption.

Despite of this, by analyzing the evolution of the adsorbed 1,2-

propanediol compound at increasing temperatures, a different

behavior could be observed depending on the nature of the transi-

tion metal site (Cu, Ni, and Co), envisaging different reaction mech-

anisms and, accordingly, reaction intermediate species and

reaction rates. Thus, in the 5.0%Cu-HT-4 sample increasing reaction

temperature up to 160 °C leads to the appearance of a band at

2

+

1+

sample (see Figure S14 in SI). Thus, while both Cu and Cu can be

considered as active sites in the 5.0%Cu-HT-4 sample, a slightly

1

+

higher reactivity could be observed for Cu , which agrees on the

improvement already seen when reducing the Cu-containing cata-

lyst with MeOH at 240 °C (see Section 3.5.2). It is important to

remark based on the IR data that the enhanced catalytic activity

in the glycerol dehydration observed with 5.0%Cu-HT-4 sample

can be ascribed to the speciﬁc role of copper versus other transition

metals to promote the carbonyl group formation pathway at the

beginning of the process.

ꢁ1

1

705 cm corresponding to a carbonyl functional group, whereas

under the same conditions in the Ni and Co based catalyst a band at

ꢁ1

1

596 cm is observed corresponding to C@C bond, while no addi-

tional band could be observed in the HT-4 sample (Fig. 12). This

may correspond to a different activation mode of the 1,2-

propanediol, where a hydride abstraction could take place pro-

However, despite displaying different reaction mechanisms,

this cannot explain the markedly lower activity of 5.0%Co-HT-4

and 5.0%Ni-HT-4 catalysts. Another aspect which could inﬂuence

the catalytic activity is the interaction strength of the reaction

products with the catalyst surface, in a way that a stronger interac-

tion would result in a fast catalyst deactivation. Thus, and in order

to analyze the interaction strength of the reaction products,

hydroxyacetone (or acetol) has been adsorbed on the 5.0%Cu-HT-

ꢁ

moted by Cu species, whereas OH abstraction catalyzed by Lewis

4

and 5.0%Ni-HT-4 samples and followed by FT-IR spectroscopic

measurements. As shown in Fig. 14, hydroxyacetone interacts on

both catalysts, although the interaction is lower on the 5.0%Cu-

HT-4 sample and stronger on the 5.0%Ni-HT-4 sample, probably

because of the higher density of acid sites in this sample (see

3

NH -TPD, Table S2 in SI).

Summarizing, it can be inferred that weak acid sites are relevant

for adsorption of 1,2-propanediol, and the presence of speciﬁc cop-

per sites in the Cu-HT sample could favor hydride abstraction in

1

+

the ﬁrst step of reaction. Moreover, Cu species show a higher

2

+

2+

reactivity compared to Cu , and since Cu species are reduced

in the ﬁrst state of the reaction due to the presence of methanol,

1

+

we can assess and conﬁrm Cu as active sites in the glycerol dehy-

dration reaction over 5.0%Cu-HT-4 catalyst. Moreover, the desorp-

tion of reaction products is enhanced in the 5.0%Cu-HT-4 sample

compared to the 5.0%Ni-HT-4 and 5.0%Co-HT-4 materials. This

could be related to the different distribution of acid-base sites in

Fig. 11. IR spectra of 1,2-propanediol adsorbed at 25 °C on 5.0%Cu-HT-4 (blue), 5.0%

3

both Co- and Ni-based samples (see NH -TPD measurements,

Ni-HT-4 (red), 5.0%Co-HT-4 (green), HT-4 (cyan) and 5.0%Cu-HT-4R (ex situ H

2

Table S10 in SI) that could have too acid density, thus making prob-

lematic acetol to leave the catalytic surface. Thus, the essential role

of copper species together with the higher interaction of the prod-

reduced, black). In magenta gas phase IR spectra of 1,2-propanediol as reference

spectra. (For interpretation of the references to colour in this ﬁgure legend, the

reader is referred to the web version of this article.)

1

72

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

2

Fig. 12. IR spectra of 1,2-propanediol at 160 °C on 5.0%Cu-HT-4 (blue), 5.0%Ni-HT-4 (red), 5.0%Co-HT-4 (green), HT-4 (cyan) and 5.0%Cu-HT-4 ‘‘ex situ” H reduced (black)1,2-

PDO: IR peak related to 1,2-propanediol. (For interpretation of the references to colour in this ﬁgure legend, the reader is referred to the web version of this article.)

Fig. 13. IR spectra of CO adsorption at 25 °C as probe molecule on the 5.0%Cu-HT-4

sample after 1,2-propanediol adsorption and increasing temperature to 160 °C.

Fig. 14. IR spectra of hydroxyacetone adsorbed at 25 °C on 5.0%Cu-HT-4 (blue) and

.0%Ni-HT-4 (red) (a), and after evacuation at 25 °C for 5 min (b). (For interpretation

5

of the references to colour in this ﬁgure legend, the reader is referred to the web

version of this article.)

ucts on the Co- and Ni-based catalysts may explain the markedly

higher catalytic activity observed with the Cu-containing catalyst.

low percent of glyceraldehyde into the starting feed were carried

out.

3.4.4. Proposed mechanism

This reaction was performed over the most active material that

was 5.0%Cu-HT-4 so far, under similar conditions used before. They

The foregoing results indicated that Cu is essential to achieve

good conversion and selectivity to acetol, mainly due to the forma-

tion of an intermediate possessing a carbonyl group at the begin-

ning of the process. In this sense, glyceraldehyde should be the

intermediate reaction, following the results getting by the FT-IR

study. In Scheme 3 a possible reaction pathway is shown. In order

to clarify, even more, the role of Cu and to elucidate some insights

of the mechanism, catalytic activity tests of acetol formation with a

1

3

were followed by C NMR at short reaction time, 45 min, observ-

ing the presence of a low amount of glyceraldehyde (see Figure S21

in SI). Because of the glyceraldehyde nature that gives a very high

reactivity in that reaction medium, it was impossible to isolate or

accumulate more quantity of glyceraldehyde during reaction;

1

therefore, H NMR data were unclear.

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

173

Scheme 3. Proposed reaction pathway for acetol formation over Cu-HT catalyst.

Experiments in continuous ﬂow carried out with the less active

catalyst Co-HT showed an improvement in the yield towards acetol

when the ﬁxed bed reactor was fed with the mixture of glycerol +

0

.45 wt% glyceraldehyde (Fig. 15). This improvement has the same

percentage value as the amount introduced on the feed: 0.45%. In

these cases, it was necessary to use a small amount of water to dis-

solve the glyceraldehyde in the initial mixture.

These results gave us a useful proof about the mechanism of the

reaction: Co is unable to activate the C-OH bond of glycerol

towards the formation of glyceraldehyde, the intermediate to ace-

tol. However, skipping the ﬁrst step, Co-HT was able to transform

glyceraldehyde to acetol.

Fig. 15. Catalytic results of selective glycerol dehydration over 5.0%Co-HT-4 with

(

orange) and without (blue) glyceraldehyde co-feed (Reaction conditions: Feed:

methanol/glycerol/glyceraldehyde/water (87.8/9.5/0.45/2.2 wt), tempera-

ture = 240 °C, ﬂow = 2 mL/h, catalyst: 0.5 g). (For interpretation of the references

to colour in this ﬁgure legend, the reader is referred to the web version of this

article.)

3.5. Effect of the amount of Cu in Cu-based catalysts

Once the adequate Cu-Mg-Al mixed oxide with the optimal Mg/

Al molar ratio was stablished as the best catalyst for the selective

dehydration of glycerol to acetol, the effect of the amount of Cu

in these materials was studied. Cu loading was varied from 0.0 to

1

2.0 wt%, ﬁnding out that, although the selectivity to acetol is

barely affected when increasing the quantity of copper, the more

copper the solid contains, the higher both conversion and yield

to acetol are attained. Nonetheless, this is only true until reaching

Cu loading of 10 wt% (see Fig. 16, and Table S4 in SI). From this

amount of copper onwards, no further increase is observed for

the glycerol conversion and the selectivity to acetol. Thus, a mixed

oxide containing 10.0 wt% of copper would correspond to the opti-

mal composition. The 10.0%Cu-HT-4 catalyst kept a good compro-

mise between glycerol conversion and acetol selectivity and, more

importantly, it provided the best catalytic stability during 9 h of

reaction with practically no variation in the selectivity to acetol

(

ꢀ55%, see Figure S15 in SI).

Comparison of the acidity and basicity data obtained by means

Fig. 16. Average glycerol conversion and average acetol selectivity for Cu-Mg-Al

of TPD analysis of both 5.0%Cu-HT-4 and 10.0%Cu-HT-4 calcined

samples (see Table S5, in SI) show that the introduction of more

copper into the structure does not affect signiﬁcantly the amount

and force of the active acid-basic centers. In fact, the two Cu-

based materials exhibit almost the same acid/basic sites distribu-

tion, being slightly more acid and less basic the one containing

lower amounts of copper (5.0%Cu-HT-4 sample). Therefore, we

can associate higher activity observed for 10.0%Cu-HT-4 catalyst

with increasing quantities of copper, as no effect of this greater

amount of copper in the acid-basic properties of our materials

was observed. In this sense, it was noticed that the more copper

the material has, the more reducible it is and, therefore, a higher

amount of copper is available to carry out the reaction. Neverthe-

less, once the 10 wt% of Cu is surpassed, part of the copper needs

higher temperatures to be reduced (see Figure S16a, SI), thereby

showing more difﬁculties to take part in the reaction. The position

calcined materials with different Cu contents during TOS

= 1–9 h (Reaction

conditions: feed = methanol/glycerol (90/10 in weight), ﬂow = 2 mL/h, with 0.5 g of

catalyst at 240 °C).

to Cu¹⁺, which has already been pointed out as the most active spe-

cies in this process (see Sections 3.4.1 and 3.4.2). Thus, the pres-

ence of methanol facilitates the reaction occurrence. In this

sense, several reactions for the selective dehydration of glycerol

to acetol were performed to study the effects of the inclusion of

water in the feed. Water was added to the GLY/MeOH mixture to

evaluate if it could eventually replace, at least partially, the use

of MeOH as solvent. Figure S18 summarizes the glycerol conver-

sion results for different water additions to the system. The pres-

ence of added water resulted in a decrease in the amount of

glycerol converted. It was also observed that the introduction of

water had a signiﬁcant effect on the catalyst stability. In conclu-

sion, when the activity of the catalyst reaches a maximum in

almost all cases, the overall yield of the reaction signiﬁcantly

decreases as soon as there is even a small quantity of water in

the reaction medium.

of this TPR peak, together with the CuO

x

nanoparticles seen for the

material 12.0%Cu-HT-4 in the HR-TEM measurements (Figure S16b,

SI) made us thing about large aggregates of CuO

of this behavior [54].

x

as the responsible

3

.6. Importance of the use of MeOH as solvent

This phenomenon could be explained not only because of the

total or partial absence of methanol, but also taking into account

the rehydration of the hydrotalcite-type material when water is

introduced into the reaction media, water molecules could block

acid centers needed for the ﬁrst glycerol adsorption [57]. Neverthe-

Results up to now attained make necessary to remark that the

choice and use of methanol as solvent for the selective dehydration

of glycerol is a key parameter. Methanol is helping to reduce Cu

2

+

1

74

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

less, the formation of the corresponding Cu-Mg-Al hydrotalcite

phase was not detected, even for the reaction carried out in pure

water (see Figure S19 in SI). Then, other possibility could be the

effect that water could exert on the thermodynamic equilibrium.

tor with a mixture containing small amount of water (see Fig-

ure S19) was not noticed, even though when water is generated

throughout the reaction. The only structural difference observed

after the reaction is the formation of a very small amount of

CuO, probably growing from the mixed oxide phase and sintering

during the reaction (see Figure S20). Nevertheless, this effect does

not affect the re-uses of the material. All these observations let us

to conclude that the carbonaceous compounds deposition on the

solid surface could be considered, at this stage, as the main cause

of catalyst deactivation in this process.

Summarizing, Cu-Mg-Al materials have demonstrated to be

highly active, selective, and also re-usable in the selective dehydra-

tion of glycerol in a continuous ﬂow ﬁxed bed reactor. Besides that,

the catalysts remained active after several re-uses by introducing

regeneration cycles in-between reactions and preserved their

chemical and structural properties at least after ﬁve consecutive

re-uses.

3.7. Re-usability

The 10.0%Cu-HT-4 catalyst showed high catalytic activity

accompanied with good selectivity to acetol in the selective dehy-

dration of glycerol in continuous ﬂow ﬁxed bed reactor, although

some deactivation was observed with time on stream. At this point,

the question is if the Cu-Mg-Al material could be used several

times in the reaction, for example by introducing regeneration

cycles in-between reactions. To address this issue, catalyst re-

cycles were carried out by washing the catalyst used in the reac-

tion with MeOH, and then regenerating it under the same calcina-

tion conditions used for the fresh material (before its 1st use). Five

consecutive re-uses of the 10%Cu-HT-4C catalyst with the preced-

ing catalyst regenerations in each case were performed under stan-

dard reactions conditions for the selective glycerol dehydration

and the attained results are given in Fig. 17. In general, both glyc-

erol conversion and selectivity to acetol values remain practically

unaltered during the ﬁve consecutive catalytic re-uses.

4

. Conclusions

This research indicates that waste by-product glycerol from bio-

diesel production can be transformed into acetol by using environ-

mentally friendly Cu-based mixed oxides derived from

hydrotalcites as catalysts. Additionally, the reaction is carried out

in a continuous ﬂow ﬁx-bed reactor, which could be scaled up from

an industrial point of view. Concretely, the 10.0%Cu-HT-4 catalyst

offered the best results for glycerol conversion (ꢀ80%) and acetol

selectivity (ꢀ60%), being stable for 9 h. This type of catalysts

showed similar results (yield ꢀ40%) when comparing with other

materials with similar copper loadings, such as the recently

The issue of possible leaching of metallic species (i.e. Cu, Mg and

Al) together with the deposition of carbonaceous compounds dur-

ing the catalyst re-cycling tests was assessed by means of different

comparative analysis of the catalysts before and after re-uses.

Thus, thermogravimetry (TGA) and elemental (EA) analysis,

together with ICP and surface area measurements of 5.0%Cu-HT-

4

and 10.0%Cu-HT-4 re-used materials reveal that our Cu-Mg-Al

materials are mainly deactivated by carbonaceous matter deposi-

tion, this making possible the recovering of the initial catalytic per-

formance by means of a calcination process (see Tables S6 and S7

in SI). Any signiﬁcant metal leaching was detected for both 5.0%Cu-

HT-4 and 10.0%Cu-HT-4 catalysts after several re-uses, while after

a primary surface area loss during the ﬁrst use the surface area val-

2

described Cu-MgF . However, more importantly, the here opti-

mized catalyst successfully addressed the relevant reusability

issue. Thus, the catalyst remains active after several re-uses by

introducing regeneration cycles in-between reactions and preserv-

ing their chemical and structural properties at least after two con-

secutive re-uses. Together with the development of

a new

2

ues stabilize around 150 m /g (Table S7 in SI). Besides that, it is

alternative system to produce acetol from glycerol, interesting con-

tributions to understand the reaction mechanism have been done.

The experiments carried out to evaluate the effect on the catalytic

performance of copper have pointed out this metal is essential to

perform the glycerol dehydration to acetol with a high reaction

rate. Probably, the preferred formation of a C@O containing inter-

mediate makes the difference between Cu and Co or Ni. In this

necessary to point out that the reduction Cu²⁺? Cu /Cu already

1+

0

observed during the reaction does not affect the catalytic perfor-

1

+

mance as Cu are the most active species, as it has been already

discussed in previous sections. Furthermore, the rehydration of

the calcined hydrotalcite, due to the extensively described ‘‘mem-

ory effect” [57], could also have a role in the catalytic deactivation.

However, a retrotopotactic transformation when feeding the reac-

1+

sense, Cu has been concluded as the most active Cu species pre-

sent in the catalyst, these species being formed at the beginning of

2

+

the reaction by reduction of Cu species. However, minor contri-

0

butions to the reaction of the other two Cu-species (Cu and

2

+

Cu ) cannot be discarded. As far as the acid-base sites role is con-

cerned, glycerol adsorption takes place on weak acid centers and

then, once copper has interacted with the molecule and the inter-

mediate is formed, and acid-base pair interaction would be respon-

sible for the ﬁnal generation of acetol.

Declaration of Competing Interest

The authors declare that they have no known competing ﬁnan-

cial interests or personal relationships that could have appeared

to inﬂuence the work reported in this paper.

Acknowledgments

Fig. 17. Glycerol conversion and acetol selectivity accumulative data during

reusability tests of 10.0%Cu-HT-4 catalyst (Average values of four repetitions).

Reaction conditions: feed = MeOH/GLY (90/10 in weight), ﬂow = 2 mL/h, with 0.5 g

of catalyst at 240 °C. * For uses 5th and 6th only one experiment was carried out,

therefore no error bars are presented.

Financial support by Spanish Government (CTQ-2015-67592,

PGC2018-097277-B-100 and SEV-2016-0683) and PhosAgro/

UNESCO/IUPAC Partnership (Proj. 139) is gratefully acknowledged.

J. Mazarío et al. / Journal of Catalysis 385 (2020) 160–175

175

J.M. also thanks Spanish Government (CTQ-2015-67592) for his FPI

fellowship.

[26] A. Talebian-Kiakalaieh, N.A.S. Amin, H. Hezaveh, Renew. Sust. Energy Rev. 40

2014) 28–59 .

(

[

27] M. Pagliaro, M. Rossi, Chem. Sus. Chem. 1 (2008), 653-653 .

[

28] A.K. Kinage, P.P. Upare, P. Kasinathan, Y.K. Hwang, J. Chang, Catal. Comm. 11

(

2010) 620–623 .

Appendix A. Supplementary material

[

29] D.C. Carvalho, L.G. Pinheiro, A. Campos, E.R.C. Millet, F.F. de Sousa, J.M. Filho, G.

D. Saraiva, E.C. da Silva Filho, M.G. Fonseca, A.C. Oliveira, Appl. Catal. 471 (A:

Gen. 2014,) 39–49 .

30] W.T. Reichle, Solid State Ion. 22 (1986) 135–141 .

31] D. Tichit, B. Coq, Cattech. 7 (2003) 206–217 .

Supplementary data to this article can be found online at

https://doi.org/10.1016/j.jcat.2020.03.010.

[

32] F. Cavani, F. Triﬁrò, A. Vaccari, Catal. Today. 11 (1991) 173–301 .

References

[33] J. Di Cosimo, V. Díez, M. Xu, E. Iglesia, C. Apesteguía, J. Catal. 178 (1998) 499–

10 .

5

[

1] BP Statistical Review of World Energy, #BPstats. 40 (2015). bp.com/

statisticalreview.

2] M. Stöcker, Angew. Chem. Int. Ed. 47 (48) (2008) 9200–9211 .

3] G.W. Huber, S. Iborra, A. Corma, Chem Rev. 106 (2006) 4044–4098 .

4] G.W. Huber, A. Corma, Angew. Chem. Int. Ed. 46 (2007) 7184–7201 .

5] J.N. Chheda, G.W. Huber, J.A. Dumesic, Angew. Chem. Int. Ed . 46 (2007) 7164–

[34] B.F. Sels, D.E. De Vos, P.A. Jacobs, Cat. Rev. 43 (2001) 443–488 .

[35] R.L. Manfro, T.P. Pires, N.F. Ribeiro, M.M. Souza, Catal. Sci. Technol. 3 (2013)

1278–1287 .

[36] M. Araque, L.M. Martínez, J.C. Vargas, M.A. Centeno, A.C. Roger, Appl. Catal. B:

Environ. 125 (2012) 556–566 .

[37] A. Bienholz, H. Hofmann, P. Claus, Appl. Catal. A:Gen. 391 (2011) 153–157 .

[38] R.B. Mane, C.V. Rode, Green. Chem. 14 (10) (2012) 2780–2789 .

[39] N. Blanch-Raga, A.E. Palomares, J. Martínez-Triguero, G. Fetter, P. Bosch, Ind.

Eng. Chem. Res. 52 (2013) 15772–15779 .

[

7

183 .

[

6] A. Demirbas, Energy Convers. Manage. 50 (2009) 2239–2249 .

7] A.L. Moreira, J.M. Dias, M.F. Almeida, M.C. Alvim-Ferraz, Energy Fuels 24 (2010)

5

717–5721 .

8] J. Costa, M. Almeida, M. Alvim-Ferraz, J. Dias, Energy Convers Manage. 74

2013) 17–23 .

9] J.M. Dias, M.C. Alvim-Ferraz, M.F. Almeida, Bioresour. Technol. 100 (2009)

355–6361.

[40] J.Y. Shen, M. Tu, C. Hu, J. Solid State Chem. 137 (1998) 295–301 .

[41] D. Yuan, X. Li, Q. Zhao, J. Zhao, S. Liu, M. Tadé, Appl. Catal. A:Gen. 451 (2013)

176–183 .

[42] M. Trombetta, G. Ramis, G. Busca, B. Montanari, A. Vaccari, Langmuir 13 (1997)

4628–4637 .

[43] M.D. Romero, J.A. Calles, M.A. Ocaña, J.M. Gómez, Microp. Mesop. Mater. 111

(2008) 243–253 .

[44] I. Popescu, N. Tanchoux, D. Tichit, I.C. Marcu, Appl. Catal. A: Gen. 538 (2017)

81–90 .

[45] P. Lauriol-Garbay, J. Millet, S. Loridant, V. Bellière-Baca, P. Rey, J. Catal. 280

(2011) 68–76 .

[46] S. Mészáros, J. Halász, Z. Kónya, P. Sipos, I. Pálinkó, Appl. Clay Sci. 80 (2013)

245–248 .

[47] X. Guo, Y. Li, W. Song, W. Shen, Catal. Lett. 141 (2011) 1458–1463 .

[48] V.A. Naumkin, A. Kraut-Vass, S.W. Gaarenstroom, C.J. Powell, NIST standard

reference database 20, version 4.1. https://srdata.nist.gov/xps/WagnerPlot.

aspx .

[49] W. Grünert, N.W. Hayes, R.W. Joyner, E.S. Shpiro, M.R.H. Siddiqui, G.N. Baeva, J.

Phys. Chem. 98 (1994) 10832–10846 .

[

(

6

[

10] S.C. D’Angelo, A. Dall’Ara, C. Mondelli, J. Pérez-Ramírez, S. Papadokonstantakis,

ACS Sust. Chem. Eng. 6 (2018) 16563–16572.

11] G.M. Lari, G. Pastore, M. Haus, Y. Ding, S. Papadokonstantakis, C. Mondelli, J.

Pérez-Ramírez, Energy Environ. Sci. 11 (2018) 1012–1029 .

12] D.T. Johnson, K.A. Taconi, Environ. Prog. 26 (2007) 338–348 .

13] C.C. Zhou, J.N. Beltramini, Y. Fan, G.M. Lu, Chem. Soc. Rev. 37 (2008) 527–549 .

14] T. Werpy, G. Petersen, A. Aden, J. Bozell, J. Holladay, J. White, A. Manheim, Top

value-added chemicals from biomass, Volume 1-Results of screening for

potential candidates from sugars and synthesis gas, Report of U.S. DOE, Energy

Efﬁciency and Renewable Energy, Biomass, 2004 .

[

15] M.J. Climent, A. Corma, P. De Frutos, S. Iborra, M. Noy, A. Velty, P. Concepción, J.

Catal. 269 (2010) 140–149 .

16] C. Vieville, J. Yoo, S. Pelet, Z. Mouloungui, Catal. Lett . 56 (1998) 245–247 .

17] R.R. Soares, D.A. Simonetti, J.A. Dumesic, Angew. Chem. 118 (2006) 4086–

[

4

089 .

18] N. Rahmat, A.Z. Abdullah, A.R. Mohamed, Renew. Sust. Energy Rev. 14 (2010)

87–1000 .

[50] P. Keil, D. Lützenkirchen-Hecht, R. Frahm, AIP Conf. Proc. 882 (2007) 490–492 .

[51] L. Chmielarz, P. Ku s´ trowski, A. Rafalska-Łasocha, R. Dziembaj, Thermochim.

Acta 395 (2002) 225–236 .

[

9

[

19] B. List, J. Am. Chem. Soc. 122 (2000) 9336–9337 .

20] W. Notz, B. List, J. Am. Chem. Soc. 122 (2000) 7386–7387 .

21] M.A. Dasari, P. Kiatsimkul, W.R. Sutterlin, G.J. Suppes, Appl. Catal. A:Gen. 281

[52] S. Kannan, A. Dubey, H. Knozinger, J. Catal. 231 (2005) 381–392 .

[53] G. Fierro, M. Lojacono, M. Inversi, P. Porta, R. Lavecchia, F. Cioci, J. Catal. 148

(1994) 709–721 .

[54] A. Alejandre, F. Medina, P. Salagre, X. Correig, J.E. Sueiras, Chem. Mater. 11

(1999) 939 .

[55] S. Célerier, S. Morisset, I. Batonneau-Gener, T. Belin, K. Younes, C. Batiot-

Dupeyrat, Appl. Catal. A:Gen. 557 (2018) 135–144 .

[56] K. Hadjiivanov, G.N. Vayssilov, Adv. Catal. 47 (2002) 307–511 .

[57] S. Miyata, Clays Clay Miner. 28 (1980) 50–56 .

(

2005) 225–231 .

[

22] C. Chiu, M.A. Dasari, G.J. Suppes, W.R. Sutterlin, AIChE J. 52 (2006) 3543–3548 .

23] T.P. Braga, N. Essayem, S. Prakash, A. Valentini, J. Braz. Chem. Soc. 27 (2016)

2

361–2371 .

[

24] S. Sato, M. Akiyama, R. Takahashi, T. Hara, K. Inui, M. Yokota, Appl. Catal. A:

Gen. 347 (2008) 186–191 .

[

25] M. Velasquez, A. Santamaria, C. Batiot-Dupeyrat, Appl. Catal. B:Environ. 160

(

2014) 606–613 .

Article Doi

Continuous catalytic process for the selective dehydration of glycerol over Cu-based mixed oxide

DOI: 10.1016/j.jcat.2020.03.010

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.jcat.2020.03.010

Full text of DOI:10.1016/j.jcat.2020.03.010

Products guided by the article

R&D Labs maybe for 116-09-6

Relevant to this article

Hot Product