Synthesis of ethyl lactate from triose sugars on Sn/Al2O3 catalysts

Article abstract of DOI:10.1016/j.apcata.2016.03.007

Sn-promoted alumina catalysts with different Sn loadings (1-8 wt.%) were prepared by impregnation and characterized by N₂ physisorption, X-ray diffraction, UV-vis-DRS, FTIR of pyridine and TPD of NH₃. Surface Sn species present Lewis acid properties different from those of the alumina support; the number and strength of the surface Lewis centers increase with the Sn loading. These materials exhibit good catalytic performance for conversion of triose sugars, such as dihydroxyacetone, toward ethyl lactates at mild conditions. The catalyst preparation conditions, Sn content and reaction conditions affect the final ethyl lactate yield. Yields of ≈70% were obtained at 353 K after 7 h of reaction. Surface Sn species participate in the kinetically relevant reaction steps.

Full text of DOI:10.1016/j.apcata.2016.03.007

Accepted Manuscript
Title: Synthesis of ethyl lactate from triose sugars on
Sn/Al2O3 catalysts
´
Author: E. Pighin V.K. Dıez J.I.Di Cosimo
PII:
S0926-860X(16)30117-X
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2016.03.007
APCATA 15799
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
5-1-2016
26-2-2016
8-3-2016
´
Please cite this article as: E.Pighin, V.K.Diez, J.I.Di Cosimo, Synthesis of ethyl
lactate from triose sugars on Sn/Al2O3 catalysts, Applied Catalysis A, General
http://dx.doi.org/10.1016/j.apcata.2016.03.007
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.

Submitted to Appl. Catal. A General, January 2016. Revised version, February,
2016.
Synthesis of ethyl lactate from triose sugars on Sn/Al2O3 catalysts
E. Pighin, V.K. Díez and J.I. Di Cosimo*
Catalysis Science and Engineering Research Group (GICIC), INCAPE, UNL-
CONICET, CCT CONICET Santa Fe, Colectora Ruta Nac. 168, km 0, Paraje "El Pozo",
(3000) Santa Fe, Argentina
*Corresponding author. Tel.: +54 342 4511546 , Ex 6118; Fax: +54 342 4511170.
E-mail address: dicosimo@fiq.unl.edu.ar (J.I. Di Cosimo).
0

Graphical abstract
Highlights:
 Inexpensive and easily scalable alumina-supported Sn catalysts with different Sn
loadings promote ethyl lactate formation from triose sugars and ethanol.
 Ethyl lactate yields of 70% are obtained on a 7.6 wt.% Sn/Al2O3 catalyst at
353K.
 Catalyst preparation conditions, Sn content and reaction conditions affect the
final ethyl lactate yield.
 Surface Lewis acid sites that increase with Sn loading participate in the
kinetically relevant reaction steps.
Abstract
Sn-promoted alumina catalysts with different Sn loadings (1-8 wt.%) were prepared
by impregnation and characterized by N2 physisorption, X-ray diffraction, UV-vis-DRS,
FTIR of pyridine and TPD of NH3. Surface Sn species present Lewis acid properties
different from those of the alumina support; the number and strength of the surface
1

Lewis centers increase with the Sn loading. These materials exhibit good catalytic
performance for conversion of triose sugars, such as dihydroxyacetone, toward ethyl
lactates at mild conditions. The catalyst preparation conditions, Sn content and reaction
conditions affect the final ethyl lactate yield. Yields of 70% were obtained at 353K
after 7 hours of reaction. Surface Sn species participate in the kinetically relevant
reaction steps.
Keywords: ethyl lactate; dihydroxyacetone; ethanol; Lewis acid; sugars
1. Introduction
In recent years, the synthesis of alkyl esters of lactic acid (alkyl lactates) has
attracted much attention because they can be obtained from renewable resources. They
are currently produced from lactic acid and alcohols (methanol, ethanol, butanol) using
mineral acids as catalysts, with the consequent technological and environmental hazard.
Several works describe the synthesis by heterogeneous catalysis using Amberlyst 15,
Dowex 50W and heteropolyacids, but the process is equilibrium-limited reaching
conversions of just 35% [1,2,3].
Alkyl lactates are high boiling point liquids used as solvents and plasticizers for
cellulose plastics and vinyl resins, among other many applications. They also take part
in cosmetic and herbicidal formulations and are used as skin whitening products [4]. In
particular, ethyl lactate (EL) is widely used as a green solvent to replace chlorinated
hydrocarbons [5]. It presents environmental, technological and economic advantages for
many industrial applications. It is biodegradable, harmless, recyclable and non-corrosive
2

[6]. In addition to the above-described uses of alkyl lactates, EL is used in
pharmaceutical formulations for the treatment of acne and seborrhea [4].
Lately, many works report the synthesis of alkyl esters of lactic acid from triose
sugars such as dihydroxyacetone (DHA) and short chain alcohols using acid catalysts.
This procedure presents the advantage that the extent of reaction is not limited by the
equilibrium. Furthermore, DHA can be obtained in a biorefinery by glycerol oxidation
[7] or by cellulose hydrolysis followed by isomerization and retro-aldol reactions [8].
The pioneering works describe the use of alcohol solutions of tin chlorides as
homogeneous catalysts [9], zeolites [10,11,12,13] and mixed oxides [14]. More
recently, many works have been devoted to study the performance of Sn-containing
solids such as mesoporous materials (MCM-41 [8,15], SBA-15 [16]), clays [17] and
zeolites (MFI [18], BEA [19,20,21,22,23]) during this synthesis. Some authors claim
that Brønsted acid sites [11,17] or a combination of mild Brønsted and Lewis acid sites
[8,15,19] promote the alkyl lactate formation from DHA. In contrast, other researchers
postulate that the Lewis acid sites of Sn-substituted BEA zeolites are responsible for the
catalytic activity [20]. In any case, the use of Sn-modified materials such as MCM-41 or
BEA zeolite for the production of alkyl lactates may be limited not only by the cost and
long synthesis time of these solids but also by the large solid quantities needed for the
scaling up of a commodity production.
Here, we present our investigations on the synthesis of ethyl lactate from DHA and
ethanol on inexpensive and easily scalable alumina-supported Sn catalysts prepared by
impregnation. For this purpose, a series of tin containing catalysts coded as ZSnAl (Z
=1.4, 4.2 and 7.6wt.% Sn) was prepared. These materials were characterized by several
techniques in order to determine their structural, textural, and optical properties, with
special emphasis on the knowledge of acid site nature and number. The role of the tin
3

loading on the generation of surface acid sites was investigated as well as the
participation of Sn species in the kinetically relevant reaction steps. We found that the
ZSnAl catalysts convert DHA at mild conditions through a series of reaction steps
leading to formation of ethyl lactate and other valuable oxygenates, Scheme 1; these
materials present a catalytic performance comparable to those of other catalytic solids
reported in the literature.
1. Experimental
2.1 Catalyst synthesis
Alumina-supported tin catalysts containing 1.4, 4.2 and 7.6 wt.% Sn were prepared
by incipient wetness impregnation of commercial -Al2O3 Cyanamid Ketjen CK 300
with aqueous solutions of SnCl4.5H2O (Aldrich, 98%). Catalysts were denoted as
ZSnAl, where Z is the Sn content expressed in wt.%. The ZSnAl precursors were
thermally treated overnight (18 h) in flowing air at 573K. Another set of catalysts was
prepared using aqueous solutions of anhydrous SnCl2 (Riedel-de Haën, 98%) and
precursor calcination in air at 573K and 773K; these catalysts were coded as 4.8SnAl-
D-573 and 4.0SnAl-D-773, respectively.
Pure SnO2 was prepared by precipitation of SnCl4.5H2O with urea at 363K,
according to the procedure of Acarbas et al [24]. After washing and drying at 363K, the
precipitate was calcined at 573K in flowing air for 3h.
The commercial -Al2O3 was treated at 773K in flowing air for 3h to remove
adsorbed water before the impregnation procedures.
2.2 Catalyst characterization.
BET surface areas (SA) were measured by N2 physisorption at 77K using an
4

Autosorb Quantachrome 1-C sorptometer. After calcination, the chemical content of Sn
and of residual Cl was analyzed by inductively coupled plasma (ICP-OES). The catalyst
structural properties were determined by X-Ray Diffraction (XRD) between 20 and 80°
using a Shimadzu XD-D1 instrument with nickel filtered Cu K radiations. Crystallite
size was calculated employing the Debye-Scherrer equation: L = 0.9/β cos , where L
is the crystallite size,  is the wavelength of the used radiation, β is the full width at
half-maximum and  is the Bragg diffraction angle.
UV-vis-DRS experiments were carried out using a Perkin Elmer Lambda 40
spectrophotometer equipped with a diffuse reflectance chamber and integrating sphere
(Labsphere RSA-PE-20) coated internally with polytetrafluoroethylene (PTFE). Prior
to each experiment, samples were compacted in a sample holder to obtain a sample
thickness of 2 mm. Spectra were recorded at room temperature in reflectance mode (R)
ꢀ
ꢃ
ꢄ
between 200 and 800 nm and converted to the Kubelka-Munk function: F ꢁꢂ
ꢀ
ꢃꢆ
1 ꢅ ꢁꢂ /2 ꢁꢂ ꢄ ꢇ ꢈ, where K is a constant including the sample scattering
coefficient and absorptivity, and C is the analyte concentration [25]. The material used
as reference was PTFE.
Total acid site number (na, mol/g) and acid site density (Na, mol/m2) were
determined by TPD of NH3. Samples were thermally treated in He at the corresponding
calcination temperature and then exposed to a 1.01% NH3/He flow at 373 K during 30
min to enable surface saturation. Weakly adsorbed NH3 was removed by flushing with
He. The temperature was then increased at a rate of 10 K/min from 373K to 593K and
kept constant for 1h. NH3 concentration in the reactor effluent was monitored by a mass
spectrometer (MS) detector in a Baltzers Omnistar unit.
The chemical nature of the acid sites present on the catalyst surface was determined
by Infrared Spectroscopy (IR) of pyridine adsorbed at room temperature and evacuated
5

at increasing temperatures. Experiments were carried out using a Shimadzu FTIR
Prestige-21 spectrophotometer. An inverted T-shaped cell containing the sample wafer
and fitted with CaF2 windows was used. In a typical experiment, the sample wafer was
evacuated at 573K and then cooled down to room temperature to take the catalyst
spectrum. After that, the sample was exposed to 0.12 kPa of pyridine at room
temperature and then evacuated consecutively at 298, 373, 423 and 473K. The resulting
spectrum at each evacuation temperature was recorded at room temperature. Spectra of
the adsorbed species were obtained by subtracting the catalyst spectrum. The
absorbance scales of the absorbed species spectra were normalized using the wafer
density ( = W/S; W is the wafer weight in grams and S the wafer area in cm2). An
indication of the number of Lewis acid sites (nL, mol/g) was determined based on the
Lambert-Beer equation,
ꢋꢌ
ꢉꢊ ꢄ
ꢊ 
where LA (Lewis area in cm-1) stands for the integrated absorption of the band at
1450cm-1 corresponding to pyridine coordinated to Lewis acid sites, and ꢊ is the molar
extinction coefficient taken as 0.64 cm/mol [26,27].
2.3 Catalytic testing
The liquid-phase reaction of dihydroxyacetone, DHA (Aldrich, 97%) with ethanol
(Merck, 99.8%) was carried out at 353K and at an autogenous pressure of 250 kPa in a
batch PARR reactor. In a typical experiment, a solution of DHA in ethanol with
ethanol/DHA=43 (molar ratio) was loaded in the reactor and a catalyst/DHA weight
ratio of 43% was used. Catalysts were thermally treated ex-situ in air flow at the
corresponding calcination temperature to remove adsorbed water. After introducing the
reactant mixture the reactor was sealed and flushed with N2 and then the mixture was
6

heated up to the reaction temperature under stirring (400 rpm). Then, the catalyst as
ground powder was added to the reaction mixture to start the reaction. Inter- and intra-
particle diffusional limitations were verified to be negligible. Quantification of reaction
products was carried out after proper product identification using a Thermo Scientific
Trace 1300 gas chromatograph (GC) with a Thermo Scientific TR-5MS capillary
column coupled to a Thermo Scientific ISQ QD MS unit. Main reaction products were
glyceraldehyde (GLA), pyruvic aldehyde (PA), ethyl lactate (EL), pyruvic aldehyde
hemiacetal (PAHA) and pyruvic aldehyde diethyl acetal (PADA). Minor products
include glyceraldehyde diethyl acetal. Solutions containing weighted amounts of the
different reactants and products and the standard (n-octanol, BDH, 99%) were injected
in the GC; the compound concentration was varied in each solution. Molar response
factors relative to the standard were calculated using the GC peak area of each
compound and the corresponding molecular weight. All reagents were analytical grade.
Yields (Yi, mol of product i/mol of DHA at t=0) were calculated as Yi = Si XDHA
,
where XDHA is the conversion of DHA and Si is the selectivity of product i (Si, mol of
product i/mol of DHA reacted). Selectivity was calculated as Si = Ci/Ci, where Ci is the
concentration of product i.
The initial DHA conversion rate (r0DHA, mmol/h g cat) was calculated from the
0
0
n
n
initial slope of the XDHA vs. time curve after multiplication by DHA /W, where
is the
DHA
number of moles of dihydroxyacetone in the reactor at t=0 and W the catalyst load.
Initial turnover rate (TOR0, min-1) is defined as the moles of DHA converted per minute
and per mole of acid site measured by NH3 TPD. Ethyl lactate productivity (PEL) was
calculated as grams of EL produced per kilogram of catalyst and per hour at the end of
the 7h-run.
7

3. Results and discussion
3.1 Catalyst characterization
The ZSnAl catalysts were prepared by incipient wetness impregnation of
commercial -Al2O3 with SnCl4.5H2O followed by calcination. They were analyzed by
several techniques in order to study their chemical, structural, textural and acid
properties. Results are summarized in Table 1.
In all the alumina-supported Sn catalysts the chlorine content remaining after
calcination in air was below 0.3 wt. % irrespective of the Sn loading and preparation
procedure. Catalysts presented slightly lower SA values than Al2O3. This is probably
due to pore blockage by tin species during preparation procedures.
The structural analysis by XRD, Figure 1, shows that regardless of the preparation
conditions, no crystalline Sn species could be detected in spite of the wide Sn
compositional range used. Thus, Sn species are well dispersed on the alumina surface
forming small domains not detectable by XRD even at loadings close to 8wt. %. This is
striking considering that calculations of the SnO2 monolayer coverage fraction indicate
that 25% of the alumina surface would be covered by SnO2 in sample 7.6SnAl.
Alumina is a solid with a defect spinel structure in which the Al3+ cations occupy
tetrahedral (Td) and octahedral (Oh) positions so that about 30% of the cations are in
tetrahedral coordination (AlOh/AlTd = 2.36) [28]. Thus, in the ZSnAl samples, the
possible presence of Sn species with different coordination and degree of interaction
with the alumina support was investigated by UV-vis-DRS. In Figure 2A the spectra of
reference materials, a commercial -Al2O3 and a mechanical mixture of 7.8 wt.% SnO2-
alumina, were plotted as Kubelka-Munk function vs. wavelength. The latter shows
signals similar to those reported in the literature for bulk SnO2 with a band at ≈220 nm
and a broad double band at about 275-300 nm assigned to hexacoordinated polymeric
8

Sn-O-Sn type of species [19,29,30]. On the other hand, alumina does not show any
relevant signal.
The spectra of the ZSnAl samples plotted in Figure 2A present a main band at ≈205
nm and very weak bands at ≈265 and ≈298nm (inset in Figure 2A) which are typical
features of SnO2 nanoparticles [31,32]. For semiconducting nanoparticles, the UV-vis-
DRS technique is very sensitive not only to cation coordination but to SnO2 particle size
since the absorption edge is expected to shift to a higher energy value when the particle
size decreases [31,33]. Thus, the blue-shift (lower wavelength) of the band at 205nm
respect to the spectrum of bulk SnO2 (275-300 nm) is attributed to dispersed Sn4+
species forming amorphous domains in ZSnAl samples that strongly interact with the
alumina support. In these species, the Sn4+ cations would be probably surrounded by
less oxygen anions than in the octahedral coordination [34].
In order to investigate in a comparative fashion the particle size of Sn species in
ZSnAl samples with different Sn loadings, their band gap energies were studied by UV-
vis-DRS following the procedure of Roy et al. [35]. The Kubelka-Munk function
ꢀ
ꢃ
F ꢁꢂ was used to determine the band gap energy (Eg). Figure 2B shows the plots of
[F ꢁꢂ × (h vs h. The Eg values were obtained by extrapolation of the linear part
of the plots to the X-axis (straight lines in Figure 2B) and the results are summarized in
Table 1. The Eg of bulk SnO2 (3.52 eV) was similar to values reported before (3.65-
3.95 eV) for nano-sized oxides [31,33]. In fact, the crystallite size of our home-made
SnO2 calcined at 573K was 4nm (calculated from the XRD pattern of Figure 1), a value
similar to that reported by Pang et al [31] for an oxide obtained at 673K.
ꢀ
ꢃ
The calculated Eg values (4.8-5.1 eV) for the Sn-containing samples (Table 1) are in
agreement with the determinations of Dvininov et al [34] for SnO2 particles prepared by
wet impregnation of SnCl4.5H2O on a Mg-Al-layered double hydroxide followed by
9

hydrolysis and calcination. As stated above, the larger the SnO2 particle size, the lower
the energy. Thus, for the ZSnAl samples, the shift of Eg to higher values compared to
bulk SnO2 indicates smaller SnO2 particles than in the reference sample. Furthermore,
the gradual decrease of the Eg values (Table 1) as Z increases from 1.4 to 7.6 wt. % Sn,
results in a concomitant increase of the SnO2 cluster size. Thus, agglomeration of SnO2
species occurs in ZSnAl samples with higher Sn loadings giving rise to bigger particles.
In Figure 2A the lower signal intensity of sample 7.6SnAl compared to 4.2SnAl is in
line with this finding.
The surface acid properties of the ZSnAl samples were investigated by combining
NH3 TPD and FTIR of pyridine. The desorption profiles of Figure 3 present similar
desorption features as Z increases. A broad desorption peak centered at 580-590K
indicates the presence of acid species that adsorb NH3 with different binding energies.
The desorption temperature range and the na values of the ZSnAl samples are much
higher than those of pure -Al2O3. These results suggest that in the former, surface Sn
species are responsible for the generation of new and stronger acid sites compared to
those of -Al2O3. Furthermore, the number of acid sites na (measured as the area under
the TPD curve, Table 1) increases linearly with Z as shown in open circles in Figure 4.
The chemical nature of the surface acid sites present on ZSnAl and -Al2O3 catalysts
was studied by FTIR of pyridine preadsorbed at room temperature and evacuated at
increasing temperatures. Figure 5A and 5B present the spectra of the νCCN and νOH
region, respectively, after evacuation at 423K. In the spectra of the νCCN region, Figure
5A, the Lewis acid character of the ZSnAl samples was confirmed by the bands at 1450
and 1618cm-1 assigned to pyridine coordinated on the oxide cations [36]. The shoulder
on the high frequency side of both bands probably indicates the presence of Sn4+ and
Al3+ cations in different chemical environments, as discussed above. The acid strength
10

of the Lewis centers generated on the ZSnAl samples by Sn doping seems to depend on
the Sn loading, so that an acid strength increase can be envisaged by the shift toward
higher frequencies of the band at 1450 cm-1 as Z increases [37].
The spectra of Figure 5A do not show any band at 1640 and 1550 cm-1 which is an
indication of the absence of Brønsted acid sites strong enough to protonate pyridine.
However, the bands in the νOH region of the spectra, Figure 5B, suggest the presence of
OH species weakly interacting with pyridine at 423K. In fact, the broad bands in the
νOH region corresponding to the ZSnAl samples before pyridine adsorption, Figure 5C,
confirm the presence of isolated surface OH groups with different coordination
numbers, i.e., bridging and multi-centered OH (3750-3650 cm-1) [38]. The OH species
remain on the surface after calcination of the ZSnAl oxides owing to the low treatment
temperature used in the preparation procedures (573K). These OH bands are more
evident on -Al2O3 and decrease with increasing the Sn loading. Therefore, sample
7.6SnAl in Figure 5C presents mainly one broad band centered at 3567 cm-1 attributed
to H-bonded OH groups. In agreement with that, the number of OH species interacting
with pyridine decreases as Z increases, as indicated by the intensity of the negative
peaks at 3800-3600cm-1 shown in Figure 5B.
In an attempt to use in a quantitative fashion the FTIR results of Figure 5A, the
number of Lewis acid sites (ꢉꢊ) was calculated for the ZSnAl samples and plotted as
closed circles in Figure 4. A good agreement was found between ꢉꢊ and the na values
determined by TPD of NH3.
3.2 Effect of preparation conditions
The influence of the preparation conditions on the physicochemical properties of the
alumina-supported Sn catalysts was studied. At similar Sn loadings, no effect of the Sn
11

oxidation state (Sn4+ or Sn2+) in the impregnating salt solution was found; when SnCl2
was used (sample 4.8SnAl-D-573), the SA value, the structural, optical and acid
properties were similar to those of sample 4.2SnAl prepared from SnCl4.5H2O (Table 1
and Figures 1, 3, 4 and 5A). Likewise, no effect was observed on the residual chlorine
content. Thus, the use of SnCl4.5H2O was preferred due to its higher solubility in water
and easier handling.
The effect of the calcination temperature was investigated with two samples
prepared from SnCl2 (samples 4.8SnAl-D-573 and 4.0SnAl-D-773, Table 1). The
alumina support was stabilized at 773K before impregnation and therefore, the slight
decrease of the SA upon increasing the calcination temperature from 573K to 773K is
likely due to agglomeration of the Sn phase and pore blockage rather than to a collapse
of the porous structure of alumina. This SnO2 particle size increase is not detected by
XRD in Figure 1 but is evident in the Eg values of Table 1; in the sample calcined at
higher temperature the Eg shifted to lower values reflecting the presence of larger SnO2
domains. In addition, the na value also decreased as a consequence of the particle
agglomeration (open down triangle in Figure 4).
3.3 Sn-catalyzed DHA conversion reactions under homogeneous and heterogeneous
conditions
Preliminary blank tests were carried with DHA and ethanol at 353K without any
catalyst. Thermal reactions were ruled out since no measurable conversion was found
after 7h under those conditions.
The conversion of DHA with ethanol was studied under homogeneous catalysis
conditions using a DHA/Sn74 (molar ratio) and ethanol solutions of SnCl4.5H2O or
SnCl2 (entries 7 and 8 in Table 2). In both cases, total conversion was attained at 3h. In
12

both tests, after 3h the ethyl lactate (EL) yield increased at the expense of that of
pyruvic aldehyde (PA) reaching values of 78-83% at the end of the 7h-run. Tin
chlorides have been shown by Hayashi et al [9] to be excellent catalysts for conversion
of trioses (DHA and GLA) into alkyl lactates with a superior performance than other
metal halides. They assigned this unique property of tin chlorides to the way they
interact with pyruvic aldehyde (PA); rather than being just a Lewis acid center, Sn in
SnClx species might strongly coordinate to the substrate to form a specific intermediate
that would give rise to alkyl lactate formation by a 1,2-hydride shift mechanism. In
agreement with the results of these researchers, the comparison of the two chlorides
after 7h of reaction, Table 2, shows that slightly higher EL yields were obtained using
SnCl4.5H2O.
The catalytic performance of our Sn-promoted solid catalysts prepared by different
procedures (Table 2, entries 2, 4 and 5) was compared under similar reaction conditions
(353K and DHA/Sn74) with the homogeneously catalyzed tests (entries 7 and 8).
After 7h of reaction, the EL yields of the alumina-supported Sn samples calcined at
573K and prepared from SnCl4.5H2O (entry 2) and SnCl2 (entry 4) were lower than in
the homogeneous tests and not much dependent on the preparation technique. But the
XDHA values of the heterogeneous tests at the end of the 7h-run were close to those
obtained by homogeneous catalysis in 3h. Although the time difference is not large, this
result indicates that the SnO2 particles present on the solids with similar Sn loadings are
less efficient than the SnClx solutions to convert DHA regardless of the tin oxidation
state in the impregnating salt. On the other hand, sample 4.0SnAl-D-773 (entry 5) that
because of calcination at 773K contains larger SnO2 particles and a lower number of
acid sites (na), gave the lowest XDHA and ethyl lactate productivity (PEL, g EL/kg cat h)
after 7h.
13

The PEL values (200 g EL/kg cat h) obtained at 353K on our alumina-supported Sn
samples with 4wt.% Sn and calcined at 573K can be more than twofold enhanced (528
g EL/kg cat h) by increasing the reaction temperature to 373K and the Sn loading to
7.6wt.% (Table 2, entry 9). Our PEL values compare favorably with those of other solid
catalysts tested at higher reaction temperatures (363-433K), such as zeolites [10,11], Zr-
Ti mixed oxides [14], Zn-MCM-41 [39] and Sn-MCM41[8,15] but they are lower than
those obtained on Sn-dealuminated BEA zeolites tested at 363K [19,20].
The heterogeneous nature of the DHA and ethanol conversion on Sn-promoted
alumina catalysts was confirmed by removing the solid catalyst after XDHA100% was
reached at 353K and then adding a fresh DHA load. The reaction was continued for 4
more hours at 353K but no conversion was observed.
3.4 Reaction pathways and active sites on ZSnAl catalysts
The ZSnAl samples of Table 1 prepared by impregnation of alumina with aqueous
solutions of SnCl4.5H2O were tested in the conversion of DHA and ethanol to ethyl
lactate. Figure 6 shows the time evolution of the DHA conversion (XDHA) and of the
major product yields on the ZSnAl catalysts containing different Sn loadings. Main
products were ethyl lactate (EL), pyruvic aldehyde (PA), pyruvic aldehyde diethyl
acetal (PADA) and pyruvic aldehyde hemiacetal (PAHA). The shape of the curves
clearly indicates that PA and PAHA are reaction intermediates whereas PADA and EL
are final products as postulated in Scheme 1. Thus, longer reaction times than those of
Figure 6 would increase the final EL yield because of the additional PA transformation
into EL.
The reaction steps conducting to EL and other products are depicted in Scheme 1.
The reaction starts by DHA isomerization to glyceraldehyde (GLA) [40], step 1.
14

Formation of GLA cannot be avoided but under our reaction conditions the equilibrium
of step 1 is expected to be shifted toward DHA [13,41]. Since both trioses give similar
product distribution when react with an alcohol [10,11], in our calculations the overall
conversion of both trioses (coded as XDHA) was quantified. The pathway to reaction
products bears the initial triose (DHA and GLA) dehydration to PA, step 2.
From PA the reaction pathway might proceed by re-arrangement with incorporation
of an alcohol molecule toward the desired ethyl lactate product, step 3, or toward PAHA
(step 4) and PADA (step 5) by addition of an alcohol molecule in each of these reaction
steps. Step 6 is the isomerization of PAHA to EL. The fact that in Figure 6A the PAHA
curve goes above that of PA at short reaction times suggests that step 5 is reversible. In
agreement with that, West et al. have observed a decrease in the PADA yield on zeolites
but at reaction times much longer than the time frame of our experiments [11].
In previous works, the role of Brønsted and Lewis acid sites on alkyl lactate and
other oxygenate formation from DHA and alcohols has been discussed [10,11,12]. It is
generally accepted that PA formation is promoted by mild Brønsted acid sites but some
authors claim the participation of Lewis acid sites [8,13]. The subsequent pathways
toward formation of EL or PADA involve sites with different acid nature; the former
occurs on Lewis acid sites whereas the latter takes place on Brønsted acid sites.
Therefore, the presence of strong Brønsted acid sites should be avoided, because in the
overall process they convert DHA into PADA. Thus, weakly acidic Brønsted sites and
Lewis acid sites seems to be a suitable combination to promote EL formation. The role
of the Lewis sites has been shown to be the 1,2-hydride shift during re-arrangement of
PA, step 3, or PAHA, step 6 [15,23].
The Lewis acid nature of the surface species generated on alumina by Sn doping
was verified by the FTIR experiments of Figure 5. Since alumina presented low activity
15

for the reaction (Table 2, entry 6), the better catalytic performance of the Sn-containing
oxides must be attributed to Sn species. Thus, based on the mechanism reported by
Hayashi et al [9] for the homogenous synthesis of alkyl lactates from trioses using Sn
chlorides, we postulate here in Scheme 2 the possible surface pathway for step 3 on the
ZSnAl oxides, i.e., the Sn-mediated conversion of PA into EL. PA is activated on tin
species forming a cyclic intermediate in which Sn is coordinated to both C=O groups of
PA. On the other hand, ethyl alcohol can be dissociatively adsorbed through the O-H
bond on Sn-O or Al-O pair sites giving rise to a surface ethoxide intermediate and a
surface hydrogen atom. The next step is the nucleophilic addition of the ethoxide to the
terminal carbonyl function of PA, followed by an intramolecular rearrangement with a
1,2-hydride shift and formation of an anionic cyclic intermediate. Then, the latter
interacts with the adjacently adsorbed hydrogen atom to yield ethyl lactate.
The other possible surface pathway conducting to EL on ZSnAl samples, i.e., the
isomerization of PAHA, step 6 of Scheme 1, is also depicted in Scheme 2. It involves a
bidentate PAHA coordination on a Sn atom and the O-H bond breaking on a surface
acid-base pair site. The following step is the formation of an intermediate similar to that
of the pathway from PA, and intramolecular rearrangement with a shift of the hydride.
Notwithstanding we have included step 3 in the reaction Scheme 1, a preliminary
kinetic modeling (not shown here) indicated a better fit of the results of Figure 6 when
the rate equations were regressed setting k3 (the kinetic constant of step 3) to zero.
Therefore, formation of EL mainly proceeds through the PAHA intermediate in
agreement with the results of Li et al. on Sn-MCM-41 [15] and of van der Graaff et al.
on Sn-BEA [23]. Thus, the thickness of the arrow lines in Scheme 2 indicates the
relative importance of the two mechanisms toward EL emphasizing that most of the EL
product arises from PAHA and to a lesser extent directly from PA.
16

Formation of PADA among the reaction products suggests that the isolated OH
groups present on the surface of the Sn-promoted oxides, as shown in Figure 5C, might
behave as weak Brønsted acid sites that participate in steps 4 and 5 of Scheme 1. In
fact, the sample prepared from SnCl2 and calcined at 573K (4.8SnAl-D-573), that shows
a higher degree of surface hydroxylation than that treated at 773K (4.0SnAl-D-773),
gives a lower EL/PADA ratio when compared at 85% DHA conversion (Table 2). In
Figure 5C a gradual decrease of the surface hydroxylation as Z increases can be
observed. This suggests that upon impregnation, Sn species displace the OH groups
from the surface. In line with that, on the ZSnAl catalysts the yield to PADA at 85%
conversion decreases in the order of 9.4, 8.8, 6.2% as Z increases from 1.4, 4.2 to
7.6wt.% Sn.
Another explanation for PADA formation is the in situ generation of weak Brønsted
acid sites during reaction due to water formation in step 2. However, the amount of
these new OH species should be low since ethanol competes favorably with water for
the adsorption sites and probably displaces it from the surface [42].
Scheme 3 shows the possible formation of PADA by esterification of PAHA with
ethanol on weak Brønsted sites.
3.5 Effect of the Sn content and acid properties on the catalytic performance of ZSnAl
catalysts
As we stated above, alumina was poorly active under our reaction conditions.
Although alumina is a Lewis acid, it seems that the Al3+ cations are not acidic enough to
transform DHA to a significant extent. The main product was PAHA owing to the low
DHA conversion and in agreement with the weak acid properties probed in Figure 3. On
the other hand, our alumina-supported Sn catalysts are Lewis acid solids in spite of the
17

fact that the alumina surface is partially hydroxylated after calcination at 573K. In fact,
it has been shown that highly hydroxylated alumina surfaces retain their Lewis acid
character even in the presence of liquid water [42,43]. Thus, we assume that the
measured number of acid sites (na) essentially corresponds to titration of Lewis acid
centers provided by Al3+ and Sn4+ cations in different coordination numbers, as
suggested by the results of Figure 4.
Figure 6 shows that both, the final DHA conversion and EL yield reached by the
ZSnAl samples increase with the Sn loading. The higher number of acid sites (na,
Figure 4) and the enhanced Lewis acid strength (Figure 5A) as Z increases explain the
improved EL yields of Figure 6. It seems that the bimolecular nature of the reactions
involved in DHA conversion requires not only a high concentration of Sn species
available for reaction but also the close proximity of those species in order to allow the
vicinal chemisorption of the different reaction intermediates.
To elucidate the participation of Sn species in the promotion of the reaction steps
involved in the conversion of DHA with ethanol, the results obtained with the ZSnAl
samples containing 1.4, 4.2 and 7.6 wt. % Sn were further investigated. From the initial
slopes of the XDHA vs time curves of Figure 6 the initial DHA conversion rates (r0
,
DHA
mmol/hg) were calculated and the results are presented in Figure 7 (open symbols) as a
function of na; commercial alumina and the samples prepared from SnCl2 (4.8SnAl-D-
573 and 4.0SnAl-D-773) were also included. The linear correlation suggests that Lewis
acid sites are involved in the kinetically relevant steps of the DHA conversion. The
initial turnover rate (TOR0, min-1) values of Table 2 are also plotted in Figure 7 (closed
symbols). Similar TOR0 values (6 min-1) were obtained for the Sn-promoted samples
regardless of the Sn loading or preparation procedure, which more than triple that of
alumina (1.7 min-1). This clearly confirms that highly dispersed Sn4+ species with
18

Lewis acid properties are responsible for most of the catalytic activity of the Sn-
containing samples.
4. Conclusions
Alumina-supported Sn catalysts, which are easily prepared by impregnation,
contain Lewis acid centers that efficiently promote conversion of trioses and ethanol
into ethyl lactate at mild reaction conditions. Promotion with Sn generates more and
stronger Lewis acid sites than those present on pure alumina. These sites participate in
rate-limiting steps and therefore, the ethyl lactate yield is enhanced as the Sn loading
increases.
Acknowledgements
Authors thank the Agencia Nacional de Promoción Científica y Tecnológica
(ANPCyT), Argentina (grant PICT 1888/10), CONICET, Argentina (grant PIP
11220090100203/10) and Universidad Nacional del Litoral, Santa Fe, Argentina (grant
CAID PI 64-103 /11) for financial support of this work.
References
[1] A. Corma, S. Iborra, A Velty, Chem. Rev. 107 (2007) 2411.
[2] A. Engin, H. Haluk, K. Gurkan, Green Chem. 5 (2003) 460.
[3] M.T. Sanz, R. Murga, S. Beltrán, J.L. Cabezas, J. Coca, Ind. Eng. Chem. Res. 41
(2002) 512.
[4] T.N.B. Kaimal, P. Vijayalakshmi, B. Ramalinga, A. A. Laxmi, US Patent N°
6342626, 2002.
[5] C.S.M. Pereira, V.M.T.M. Silva, A.E. Rodrigues, Green Chem. 13 (2011) 2658-
19

2671.
[6] S. Aparicio and R. Alcalde, J. Phys. Chem. B 113 (2009) 14257-14269.
[7] C.L. Bianchi, P. Canton, N. Dimitratos, F. Porta, L. Prati., Cat. Today, 102-103
(2005) 203-212.
[8] F. de Clippel, M. Dusselier, R. Van Rompaey, P. Vanelderen, J. Dijkmans, E.
Makshina, L. Giebeler, S. Oswald, G.V. Baron, J. F. M. Denayer, P. P. Pescarmona,
P.A. Jacobs, and B.F. Sels, J. Am. Chem. Soc. 134 (2012) 10089-10101.
[9] Y. Hayashi, Y. Sasaki, Chem. Comm. (2005) 2716-2718.
[10] P.P. Pescarmona, K.P.F. Janssen, C. Delaet, C. Stroobants, K. Houthoofd, A.
Philippaerts, C. De Jonghe, J.S. Paul, P.A. Jacobs and B.F. Sels, Green Chem. 12
(2010) 1083-1089.
[11] R.M. West, M.S. Holm, S. Saravanamurugan, J. Xiong, Z. Beversdorf, E. Taarning,
C. H. Christensen, J. Catal. 269 (2010) 122-130.
[12] K.P.F. Janssen, J.S. Paul, B.F. Sels, P.A. Jacobs, Stud. Surf. Sc. Catal. 70 (B)
(2007) 1222-1227.
[13] P.Y. Dapsens, B.T. Kusema, C. Mondelli, J. Pérez-Ramírez, J. Molec. Catal. A:
Chem. 388-389 (2014) 141-147.
[14] A.M. Mylin, S.I. Levytska, M.E. Sharanda, V.V. Brei, Catal. Comm. 47 (2014) 36-
39.
[15] L. Li, C. Stroobants, K. Lin, P.A. Jacobs, B.F. Sels and P.P. Pescarmona, Green
Chem. 13 (2011) 1175-1181.
[16] C. M. Osmundsen, M. S. Holm, S. D. and E. Taarning, Proc. R. Soc. A (2012) 1-
17.
[17] J. Wang, Y. Masui, M. Onaka, Appl. Catal. B: Environm. 107 (2011) 135- 139.
[18] H.J. Cho, P. Dornath, and W. Fan, ACS Catal. 4 (2014) 2029-2037.
20

[19] J. Dijkmans, M. Dusselier, D. Gabriels, K. Houthoofd, P. C. M. M. Magusin, S.
Huang, Y. Pontikes, M. Trekels, A. Vantomme, L. Giebeler, S. Oswald, and B. F.
Sels, ACS Catal. 5 (2015) 928-940.
[20] C. Hammond, S. Conrad, and I. Hermans, Angew. Chem. Int. Ed., 51 (2012) 11736
-11739.
[21] E. Taarning, S. Saravanamurugan, M.S. Holm, J. Xiong, R.M. West and C.H.
Christensen, ChemSusChem, 2 (2009) 625-627.
[22] S. Tolborg, A. Katerinopoulou, D.D. Falcone, I. Sádaba, C.M. Osmundsen, R.J.
Davis, E. Taarning, P. Fristrup and M.S. Holm, J. Mater. Chem. A, 2 (2014) 20252-
20262.
[23] W.N.P. van der Graaff, G. Li, B. Mezari, E.A. Pidko, and E.J.M. Hensen,
ChemCatChem, 7 (7) (2015) 1152-1160.
[24] O. Acarbas¸ E. Suvaci, A. Dogan, Ceram. Intern. 33 (2007) 537-542.
[25] F. M. Mirabella, Modern techniques in applied molecular spectroscopy, first ed.,
Wiley-Interscience, New York, 1998.
[26] E. Selli, L. Forni, Microp. Mesop. Mat. 31 (1999) 129-140.
[27] M. Tamura, K. Shimizu, A. Satsuma, Appl. Catal. A: General 433-434 (2012) 135-
145.
[28] V.K. Díez, C.R. Apesteguía, and J.I. Di Cosimo, J. Catal. 215 (2003) 220-233.
[29] N. K. Mal, A. V. Ramaswamy, J. Molec. Catal. A: Chem. 105 (1996) 149-158.
[30] R. Bermejo-Deval, R. Gounder, M. E. Davis, ACS Catal. 2 (2012) 2705-2713
[31] G. Pang, S. Chen, Y. Koltypin, A. Zaban, S. Feng, and A. Gedanken, Nano
Lett.1(12) (2001) 723-726.
[32] H. Deng and J. M. Hossenlopp, J. Phys. Chem. B, 109 (2005) 66-73.
[33] M. Bhagwat, P. Shah, V. Ramaswamy, Mat. Lett. 57 (2003) 1604-1611.
21

[34] E. Dvininov, M. Ignat, P. Barvinschi, M.A. Smithers, E. Popovici, J. Hazard. Mat.
177 (2010) 150-158.
[35] S. Roy, K. Bakhmutsky, E. Mahmoud, R.F. Lobo and R. J. Gorte, ACS Catal. 3
(2013), 573-580.
[36] M. I. Zaki, M.A. Hasan, F.A. Al-Sagheer, L. Pasupulety, Colloids Surf. A: Phys.
Eng. Asp. 190 (2001) 261-274.
[37] E.P. Parry, J. Catal. 2 (1963) 371-379.
[38] H. Knozinger, P. Ratnasamy, Catal. Rev. Sci. Eng. 17 (1978) 31-70.
[39] X. Collard, P. Louette, S. Fiorilli and C. Aprile, Phys. Chem. Chem. Phys. 17 (40)
(2015) 26756-26765.
[40] V.A. Yaylayan, S. Harty-Majors, A.A. Ismail, Carbohyd. Res. 318 (1999) 20-25.
[41] R.S. Assary and L.A. Curtiss, J. Phys. Chem. A 115 (2011) 8754-8760.
[42] G. Busca, Catal. Today 226 (2014) 2-13.
[43] B. Kasprzyk-Hordern, Adv. Coll. Interf. Sc. 110 (2004) 19-48.
22

Caption to Figures
Figure 1: XRD patterns of ZSnAl catalysts (Z: wt.% Sn) prepared by different
procedures, commercial -Al2O3 and home-made SnO2.
Figure 2: UV-vis-DRS spectra of ZSnAl catalysts (Z: wt.% Sn) with different Sn
loading. -Al2O3 and 7.8wt.% SnO2-alumina included as references. (A) Kubelka-Munk
function, F ꢁꢂ , as a function of wavelength; (B) [F ꢁꢂ × (h vs h plot for band
ꢀ
ꢃ
ꢀ
ꢃ
gap (Eg) energy determination.
Figure 3: TPD of NH3 on ZSnAl catalysts (Z: wt.% Sn) prepared by different
procedures. -Al2O3 included as reference.
Figure 4: Number of acid sites on ZSnAl catalysts prepared by different procedures as a
function of the Sn loading. -Al2O3 included as reference [Open symbols: na values
from TPD of NH3, Figure 3; Closed symbols: Lewis sites (nL) from FTIR of pyridine,
Figure 5A].
Figure 5: FTIR of pyridine adsorbed at 298K and evacuated at 423K on the ZSnAl
catalysts. -Al2O3 included as reference. (A) νCCN region; (B) νOH region; (C) νOH region
before pyridine adsorption.
Figure 6: DHA conversion (XDHA) and product yields as a function of reaction time on
ZSnAl catalysts (A) Z=1.4wt.% Sn; (B) Z=4.2wt.% Sn; (C) Z=7.6wt.% Sn [Z: Sn
loading; 353K; ethanol/DHA=43 (molar); 1g of catalyst].
23

Figure 7: Initial DHA conversion rate (r0DHA, open symbols) and turnover rate (TOR0,
closed symbols) as a function of the number of acid sites (na) of ZSnAl catalysts
prepared by different procedures and -Al2O3 [353K; ethanol/DHA=43 (molar); 1g of
catalyst].
Scheme 1: Reaction steps toward formation of ethyl lactate and other oxygenates from
DHA and ethanol on ZSnAl catalysts.
Scheme 2: Postulated surface reaction mechanism on Sn sites for EL formation by steps
3 and 6 of Scheme 1.
Scheme 3: Postulated surface reaction mechanism on weak Brønsted acid sites for
PADA formation by steps 4 and 5 of Scheme 1.
24

SnO2
Al O
*
0.2
o
2
3
o
o
o
o
o
7.6SnAl
4.2SnAl
o
o
o
o
o
o
1.4SnAl
o
o
o
o
4.0SnAl-D-773
o
4.8SnAl-D-573
o
o
o
o
o
o
o
o
o
-Al2O3
*
*
*
*
SnO2
*
20 30 40 50 60 70 80
2(°)
Figure 1: XRD patterns of ZSnAl catalysts (Z: wt.% Sn) prepared by different
procedures, commercial -Al2O3 and home-made SnO2.
25

50
40
30
20
10
0
10
8
0.6
0.4
0.2
(B)
(a)
(A)
(b)
4.2SnAl (b)
8
(d)
(c)
250 275 300 325
F
4.2SnAl

6
7.6SnAl (c)
8
4
7.8wt.% SnO2-alumina x 10
7.6SnAl
F
2
1.4SnAl (a)
-Al2O3 (d)
1.4SnAl
-Al2O3
0
3.0
3.5
4.0
4.5
5.0
5.5
6.0
200 225 250 275 300 325 350 375 400
Energy, h (eV)
Wavelength (nm)
Figure 2: UV-vis-DRS spectra of ZSnAl catalysts (Z: wt.% Sn) with different Sn loading. -Al2O3 and 7.8wt.% SnO2-alumina included as
references. (A) Kubelka-Munk function, F ꢁꢂ , as a function of wavelength; (B) [F ꢁꢂ × (h vs h plot for band gap (Eg) energy
ꢀ
ꢃ
ꢀ
ꢃ
determination.
26

650
600
550
500
450
400
350
125
7.6SnAl
593K
4.2SnAl
1.4SnAl
4.0SnAl-D-773
4.8SnAl-D-573
-Al2O3
0
10 20 30 40 50 60 70
Time (min)
Figure 3: TPD of NH3 on ZSnAl catalysts (Z: wt.% Sn) prepared by different procedures. -Al2O3 included as reference.
27

,
,
ZSnAl (Z=1.4; 4.2; 7.6wt.% Sn)
-Al O
200
150
100
50
2
3
,
,
4.8SnAl-D-573
4.0SnAl-D-773
0
0
1
2
3
4
5
6
7
8
Z, Sn loading (wt. %)
Figure 4: Number of acid sites on ZSnAl catalysts prepared by different procedures as a function of the Sn loading. -Al2O3 included as
reference [Open symbols: na values from TPD of NH3, Figure 3; Closed symbols: Lewis sites (nL) from FTIR of pyridine, Figure 5A].
28