Palladium catalysed oxidation of glycerol - Effect of catalyst support

Article abstract of DOI:10.1016/j.molcata.2016.05.008

The activity and selectivity of Palladium catalysts on various supports, for the liquid-phase oxidation of glycerol in basic condition, have been studied. Activated carbon (AC), SiO₂, Al₂O₃ and TiO₂ were the supports studied. The catalysts were characterized using a variety of techniques (TEM, BET, Pulse-Chemisorption, TPD, TPR and XPS). The support is shown to have a significant effect on both rate and selectivity of the reaction, with the best rate of glycerol conversion, as well as selectivity to glyceric acid, being obtained for activated carbon (AC) among all the supports studied. The results have been interpreted in terms of known concepts of metal-support interactions. It is concluded that the electronic interactions between the metal and support, as well as acidic/basic properties of catalyst, play a vital role in product selectivity and activity.

Full text of DOI:10.1016/j.molcata.2016.05.008

Accepted Manuscript
Title: Palladium catalysed oxidation of glycerol −Effect of
catalyst support
Author: Ashutosh Namdeo S.M. Mahajani A.K Suresh
PII:
S1381-1169(16)30179-0
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.molcata.2016.05.008
MOLCAA 9883
To appear in:
Journal of Molecular Catalysis A: Chemical
Received date:
Revised date:
Accepted date:
1-12-2015
9-5-2016
10-5-2016
Please cite this article as: Ashutosh Namdeo, S.M.Mahajani, A.K Suresh, Palladium
catalysed oxidation of glycerol −Effect of catalyst support, Journal of Molecular
Catalysis A: Chemical http://dx.doi.org/10.1016/j.molcata.2016.05.008
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.

Palladium catalysed oxidation of glycerol -
Effect of catalyst support
Ashutosh Namdeo, S.M. Mahajani and A.K Suresh*
Department of Chemical Engineering, Indian Institute of Technology Bombay, Mumbai,
India 400076
*Corresponding author : aksuresh@iitb.ac.in
1

Graphical Abstract
2

*Graphical Abstract

Highlights





The effect of support on Pd catalyst has been studied for glycerol oxidation.
Metal-support interaction has been probed through XPS and other techniques.
Carbon-supported Pd was more active and selective than oxide supported Pd.
Support effect was less for the carbon support as compared to oxide supports.
Electronic interaction and support acidity have been identified as key factors.
3

Abstract
The activity and selectivity of Palladium catalysts on various supports, for the liquid-phase
oxidation of glycerol in basic condition, have been studied. Activated carbon (AC), SiO2,
Al2O3 and TiO2 were the supports studied. The catalysts were characterized using a variety
of techniques (TEM, BET, Pulse-Chemisorption, TPD, TPR and XPS). The support is shown
to have a significant effect on both rate and selectivity of the reaction, with the best rate of
glycerol conversion, as well as selectivity to glyceric acid, being obtained for activated
carbon (AC) among all the supports studied. The results have been interpreted in terms of
known concepts of metal-support interactions. It is concluded that the electronic interactions
between the metal and support, as well as acidic/basic properties of catalyst, play a vital role
in product selectivity and activity.
Keywords
Oxidation; Glycerol; Catalyst Support; Selectivity; Palladium
4

1. Introduction
Concomitantly with the interest in biodiesel in the last two decades, studies on ways to upgrade
glycerol, the major byproduct of biodiesel production, have also intensified. The yield of
glycerol in biodiesel production is about 10% by weight. A surplus of glycerol is therefore
expected to hit the market, as biodiesel gains ground as a transportation fuel, because of its
renewable, and relatively environment-friendly nature. While a number of target products from
glycerol can be envisioned, given the price volatility of glycerol, there is a view that targeting
high-value, low-volume chemicals through oxidative routes may be the more attractive option[1–
3].Heterogeneous catalytic oxidation of glycerol has therefore attracted considerable attention in
recent years. Such oxidations are usually non selective, and this has always been a challenge for
reaction engineers when it comes to selectivity engineering. Interventions at the micro level
(mainly through the use of selective catalysts) and the macro level (via reactor design) have been
the usual approaches to tackle this problem. Of these, catalysis research is the one that holds by
far the greater potential. The search for a catalyst should not confine itself to catalytically active
species alone, but should also consider issues such as the support and its possible role in
modifying the effect of the catalyst, how amenable the supported catalyst is for use in industrial
reactors and so on.
Glycerol oxidation proceeds by a complex mechanism and leads to a multitude of products, and
different possible schemes have been reported in the literature[2–7]. It has been
shown[2,8,9]that, by tuning the reaction conditions (especially pH), and with appropriate
catalysts, one can direct the oxidation to different product profiles. Thus, under acidic conditions
5

and with bismuth catalyst (either independently or in combination), the secondary alcoholic
group can be preferentially oxidized to obtain dihydroxyacetone (and hydroxy-pyruvic acid),
whereas under basic conditions and with Platinum group catalysts or gold, the reaction yields
glyceric acid and tartronic acid as the main products. In this paper, we concern ourselves with
oxidation under basic conditions. Hirasawa et al.[10] describe some ways of directing the
reaction towards dihydroxyacetone.
The mechanism of glycerol oxidation under basic conditions over heterogeneous catalysts has
been the object of several studies in the literature. A high pH is required for the initial de-
protonation of glycerol and subsequent dehydrogenation to glyceraldehyde, which then
undergoes further oxidation to glyceric acid[3–5,11–13]. According to some theoretical and
experimental studies[3,9], the reaction involves steps on the catalyst surface as well as in
solution. Di-oxygen is only indirectly involved in the oxidation, through its participation in the
recycling of the hydroxyl species. Researchers have also observed peroxide formation during
glycerol oxidation and commented on peroxide formation as the probable reason for C-C
cleavage and formation of lower carbon products such as glycolic and formic acids[7,13–17].
The catalysts, which have mainly attracted attention in the context of glycerol oxidation under
basic conditions, are Pd, Pt, Au and their combinations, with various supports and
promoters[1,4,5,8,12,13,18–25]. There is a view[2,9] that gold is more stable against deactivation
and leaching as compared to Pt and Pd, but some results[9]show a higher incidence of C-C
scission byproducts forming with gold catalysts, possibly due to the high acidity of some of the
catalysts studied[2]. Pt and Pd catalysts also have the advantage of being active under acidic as
6

well as alkaline conditions, while Au is active only under alkaline conditions. The activity and
selectivity of the catalyst are not only influenced by the active phase, but also by the metal
particle size, the support and its interaction with the active phase. Recent reviews by Katryniok
et al.[2] and Davis et al.[3] provide good overviews.
While the importance of the support effect in the catalyst engineering for glycerol oxidation has
been recognized[2], it is only recently that extensive studies are starting to emerge. In general,
interaction of the metal with the support can lead to a better or a worse performance of the
catalyst, and hence is important to understand properly. Furthermore, these effects depend not
only on the nature of the support, but also on the method of catalyst synthesis since the latter
determines the nature of the forces that bind the metal to the support[26–33]. Much of the
literature available on glycerol oxidation focuses on various forms of carbon as the support and
studies on other supports are few. Carbon has the advantage of being stable in both acidic and
basic media[34] and that makes it attractive for glycerol oxidation, which exhibits different
product profiles under the two sets of conditions. Further, recovery of the active metal from the
spent catalyst is easy since the support can be burned off.
The effect of support in glycerol oxidation has been studied mainly for gold-based active metal
catalysts [11,35–38], and studies on palladium catalysts on different supports are few. Most of
the available studies use a carbon support. The studies are at temperatures of 50-600C, and cover
a wide range of oxygen pressures. It is clear from these studies that use of Pd catalyst leads to the
same products as gold, but the selectivities vary. The early studies of Garcia et al.[1] showed that
a high pH increased the activity as well as selectivity to glyceric acid with Pd/C catalysts. These
7

authors conducted their reactions at atmospheric pressure with a continuous flow of air, and the
reaction at pH 11 was fast enough that the process became limited by external mass transfer (as
shown by their measurements of dissolved oxygen). Under such conditions, they report a
maximum selectivity to glyceric acid, of about 70% at 100% conversion of glycerol. Following
upon the findings of Garcia et al.[1], subsequent literature on the palladium catalysis of glycerol
oxidation has tended to focus on high oxygen partial pressures, and high pH. Important examples
are Gallezot et al.[39], Prati et al.[21–23,40], Ketchie et al.[13] and Carrettin et al.[19].
Influence of the metal particle size on the catalyst activity and selectivity has been a focus in
several studies[14,22]. With Pd catalysts, Prati et al.[22] found that the selectivity towards
glyceric acid increases with an increase in particle size while catalytic activity and selectivity
towards tartronic and glycolic acid decrease. These differences were observed for sizes of 10-15
nm, while in the range of 2-10 nm, particle size did not seem to matter. These findings may be
compared with findings of Ketchie et al.[14] on gold catalysts that the smaller the nanoparticles,
the higher the activity, but the lesser the selectivity to C3 products. While considering the
particle size effect, one has to also factor in the possibility that very fine particles are more easily
leached out as compared to the larger particles[2].
While, as noted above, several aspects of supported metal catalysts have attracted attention in the
literature, one aspect that is inadequately dealt with in general, is the issue of transport
limitations[2]. Many inferences linking catalyst characteristics to their performance become
questionable in the absence of a satisfactory demonstration of the absence of transport limitations
in the reaction studies. There is evidence in the literature of a reduction in the rate [10, 35] and a
8

loss in selectivity [41] with an increase in micro-porosity. The importance of choosing the
conditions of catalyst evaluation with an adequate appreciation of possible transport limitations
is therefore underlined.
In the present work, we focus on palladium-catalyzed glycerol oxidation in basic conditions. The
main objective of the study, apart from establishing the kinetics and selectivity of different
catalyst systems, is to understand the influence of the support, and hence several catalyst
supports -activated Carbon, SiO2, Al2O3 and TiO2 - have been employed in the studies. As
mentioned above, activated carbon supported catalysts have been studied more than others for
glycerol oxidation, and therefore provide a benchmark for the performance of other supported
catalysts. Alumina and silica are among the most prevalent supports for metal catalysts. While
the use of alkaline conditions calls into question the physical stability of the silica-supported
catalyst, we nevertheless thought that it would be interesting to include it in our study for the
following reasons: (i) in terms of acidity/basicity, and metal-support interaction (which are likely
to be important in the performance of the catalyst[38]), the inclusion of silica support provides
for a range of conditions to be studied, and (ii) conservative estimates of the dissolution rate (see
Niibori et al.[41] for example) show that the dissolution within the timescales of our experiments
is likely to be very small; in any case, activity comparisons would be based on initial rates. The
choice of TiO2 was because, in general (though not for glycerol oxidation), much of the
information that is available on metal-support interaction for Pd catalysts is for this
support[32,33,42–45]. Characterization techniques such as TEM, Temperature-programmed
techniques (Desorption, reduction and oxidation), Pulse-chemisorption, XPS and BET (Surface
area and porosity) have been used to understand possible support effects. Finally, reactions have
9

been conducted under conditions of kinetic control and attempts have been made to relate
catalyst/support characteristics to activity and selectivity.
2. Experimental Methods
2.1 Catalyst preparation
Pd/SiO2 and Pd/Al2O3 were obtained from Vineeth Chemicals and Pd/AC, from Merck
Chemicals. Pd on TiO2 was synthesized as part of this work, by impregnation method with
chemical reduction[46]. In a typical preparation, an aqueous solution of PdCl2 (99.99% Merck
Chemicals) according to the required weight percentage was added to a slurry of TiO2 (99.9%
Rutile, Wilson Laboratories, India) under agitation, and stirring was continued for 2 hrs. A
solution of KOH+sodium borohydride (99.99%, Merck Chemicals) was then added over 2hr,
small aliquots at a time, to complete the chemical reduction. After chemical reduction, the
catalyst was filtered and washed with milli-Q water. The catalyst was then dried at 1200C for 12
hours followed by calcination at 5000C in an inert atmosphere.
2.2 Catalyst Characterization
The catalysts were characterized using TEM, BET, XPS, temperature programmed
adsorption/reduction and pulse-chemisorption. TEM analyses of catalyst were carried out with
Philips’ model CM200 and Jeol JEM 2100F instruments in order to study the size distribution of
palladium crystallites. Pore size distribution and surface area were measured by BET method
(Micromeritics ASAP 2020). Before subjecting to BET analysis, the catalysts were pretreated by
degassing in an inert atmosphere at 3000C.
10

To determine the binding energies and oxidation state of palladium, X-ray photoelectron
spectroscopy (XPS) was carried out (ThermoVG Scientific A1214). Samples were prepared by
dispersing the catalyst powders on an aluminum foil. The working pressure of Argon in the
analysis chamber was less than 1.5x10-7 Pa. A monochromatic Al K-α source was used. The
kinetic energies obtained which were converted to binding energies using 0.5 eV as spectrometer
correction, using the equation: BE (eV) = (1486.6 - KE) + 0.5.For Pd/SiO2, Pd/Al2O3 and
Pd/TiO2; C1s (284.6 eV) was used as an internal calibration. For Pd/AC, Ag dot was used for
internal calibration (standard Ag3d5/2 peak position is 368.2 eV).
To study the support interaction and dispersion, temperature programmed desorption and
reduction (CO-TPD, TPR) and pulse-chemisorption techniques were used (TPDRO 1100,
Thermo Scientific). In a typical pretreatment procedure, the gas line was first cleaned by inert
gas (Argon), and then the sample was exposed to 5% O2/He followed by 5% H2/Ar to ensure that
the exposed catalyst surface is first oxidized and then reduced completely. Finally, a high
temperature desorption was carried out by the inert gas argon. This ensures that the active metal
surface of the catalyst is at zero oxidation state and the catalyst is free from any impurities.
After pretreatment, Pulse chemisorption and Temperature programmed desorption (TPD) were
carried out on the same sample. Pulse chemisorption studies were carried out by sending a
known number of calibrated pulses of CO. After chemisorption, the sample, now saturated with
CO, was used for TPD analysis. In the case of TPO/TPR, the catalysts were pretreated (as above)
and then TPO followed by TPR were performed.
11

Pyridine-TPD and CO2 pulse chemisorption were performed to determine the acidity and
alkalinity of oxide supports. Further, to know the nature of acidic sites of oxide supports, FTIR
(MAGNA 550, Nicolet Instruments Corporation, USA) spectroscopy was carried out using
pyridine as a probe molecule. All the catalyst samples were first treated with pyridine and then
kept for 12 hours at 1200C under an inert atmosphere to remove all physisorbed pyridine, and the
samples were then used for FTIR and TPD measurements. The alkalinity of the catalysts was
determined by CO2 pulse studies, carried out in a manner similar to CO pulse-chemisorption
studies.
2.3 Reaction studies
Experiments on the oxidation of glycerol were performed in a 100 ml steel autoclave (Sharad
Autoclave). The reactor assembly is shown in Figure 1. The autoclave was equipped with a
pressure gauge, pressure sensor, heater, automatic cooling, temperature control and drive for
agitation. Typically, the autoclave is charged with the required concentration of glycerol in water
(Milli-Q water purged with N2). NaOH (99.9 % Merck) and fresh catalyst (Pd/AC, Pd/SiO2,
Al2O3 and Pd/TiO2) are added according to the required ratio with glycerol for every run.
Initially, the mixture is stirred at low RPM (100 RPM) and heated up to the required temperature
in Nitrogen (99.99% Med-Gas) atmosphere. Once the required temperature is attained, the
nitrogen in the autoclave is flushed out with oxygen, the autoclave is pressurized with oxygen
12

(99.99% Med-Gas), and the speed of agitation is increased to the desired level. A constant
oxygen pressure is maintained in the reactor throughout the experiment.
Figure 1: Oxidation reaction setup
The following conditions were kept the same in all the experiments: Temperature= 600C,
Pressure=6 Bar; Glycerol/Metal (Pd) = 1000 (mol/mol); NaOH/Glycerol=4 (mol/mol); RPM=
800, reaction volume= 60 ml, reactant (glycerol) concentration = 0.6 molar.
The particle size and stirrer speed were chosen so as to eliminate the possibility of any transport
limitations (external or internal), based on preliminary experiments. The conversion-time profiles
as well as product profiles were found to be independent of stirrer speed for stirrer speeds greater
than 800 RPM. Calculations based on Weisz-Prater criterion (for example, as carried out by[47],
see the Supplementary Material (SM) for details) showed that internal transport limitations were
13

absent for catalyst particle sizes of 5 microns (with the estimated W-P parameter being
significantly less than 0.3).
Quantitative analysis of reaction mixtures was carried out on a Waters High Performance Liquid
Chromatograph (HPLC) equipped with UV and RI detectors. Calibration standards such as
glycerol (98%), dihydroxyacetone (98%), acetic acid, formic acid, glycolic acid (70% in water),
and oxalic acid (98%), were obtained from Merck. Tartronic acid (97%), glyceric acid calcium
salt dihydrate (99%) and glyceraldehyde (98%) were obtained from Sigma Aldrich.
Hydroxypyruvic acid sodium salt (99%) and glyoxalic acid (50% in water) were obtained from
Himedia and Spectrochem (India) respectively. An Aminex HPX-87H (Bio-Rad) column was
used for analysis at 600C, with a mobile phase flow rate of 0.6 ml/min. The mobile phase, 5mM
of H2SO4 in water/acetonitrile was made in the following manner. 300 ml of acetonitrile was
taken in a 1000 ml of volumetric flask and 10 ml of 0.5 molar H2SO4 was added. The solution
was made up to 1000 ml with milli-Q water. This mobile phase was degassed in a sonicator for
at least 2 hours before use.
For each component of interest, calibrations were prepared by plotting the area under the peak
(peak height in the case of overlapping peaks) vs. concentration. Reaction samples were filtered
before analysis using a 0.2-micron filter, and neutralized by adding a known amount of H2SO4.
Based on the calibrations, the concentrations of reactant and products were obtained and
selectivity and conversion calculated (See supplementary material for more details).
14

3. Results and discussion
3.1 Pore size distribution and surface area
In view of the fact that intra-particle diffusion limitation can make a significant difference in
activity and selectivity, we have performed BET for pore size distribution and surface area
analysis. Table 1 summarizes the results. An analysis of the results using t-plots and BJH
theory[48] showed that Pd/TiO2, unlike the other catalysts studied (Pd/SiO2, Pd/Al2O3 and
Pd/AC), had very low porosity with most of the measured surface area coming from the external
surface. Pd/SiO2 and Pd/Al2O3 showed a significant contribution from meso-pores with
negligible micro-porosity, while Pd/AC had significant micro-porosity. As discussed above, C-C
cleavage and loss in activity have been observed in microporous catalysts[11,35,49]. Because of
the absence of micro-porosity in oxide supports internal diffusion is unlikely to limit the rate. As
mentioned above, our calculation of Weisz-Prater criteria shows no likelihood of internal
diffusion effects even for Pd/AC although micro-porosity is significant.
15

Table 1: Results of BET characterization of the catalysts used
t-plot micro-
pore volume
(cm3g-1)
Catalyst
BET surface
area m2g-1
Pore diameter (A0,
BJH theory)
Pore width (A0,
BET theory)
Total pore
volume(cm3g-1)
1500
250
280
2.8
42
88
23
94
51
94
0.92
0.59
0.38
0.012
Pd/AC
Pd/SiO2
Pd/Al2O3
Pd/TiO2
47
0.36
0.018
184
0.0065
0.00015
3.2 Crystallite size
Crystallite sizes were determined from transmission electron micrographs as well as inferred
from pulse chemisorption studies. The TEM images of the catalysts are shown in Figures 2-5,
which indicate similar Pd particle sizes dispersed on the support in all cases, with an average
diameter of 4-7 nm.
16

Fig 2: TEM images of Pd on AC showing crystallite size distribution of Pd on the support
Fig 3: TEM images of Pd/SiO2 showing Crystallite size Distribution of Pd on the support
17

Fig 4: TEM images of Pd/Al2O3 showing crystallite size distribution of Pd on the support
Fig 5: TEM images of Pd/TiO2 showing crystallite size distribution of Pd on the support
18

The CO Pulse-chemisorption results are shown in Table 2. While particle sizes derived from
pulse studies agree reasonably with those from TEM for Pd/AC, they are seen to be different for
the other catalysts. These calculations assume a 1:1 stoichiometry for chemisorption of CO on
Pd. Supported Pd catalysts often show evidence of more than one type of adsorbed CO: a
linearly bonded CO as well as multiply co-ordinated (or bridge-bonded) CO[50–52] (see below
under TPD for further discussion on this). In fact, recognizing this, the TPDRO software uses a
default stoichiometry of 1.5 (atoms of Pd per atom of CO), which gives a metal particle size of 7
nm for Pd/AC. CO adsorption can also be reduced because of metal-support interaction, resulting
from a decoration of the metal particle by the oxide and/or from electronic interactions[53]. A
strong interaction between the metal and the support causes a reduction in the extent of CO
adsorption, thus over-estimating the size of metal particle. Literature[50,51] suggests that
chemisorption cannot be depended on for dispersion calculations in such cases, and hence, we
have used the TEM-based dispersion values in our calculations of TOF. On the basis of TEM
results, it is seen that the metal particle sizes for the different preparations used are comparable,
and further, are in a range where significant dependence of performance on metal particle size is
not expected. Any difference in performance between the different catalysts has therefore to be
explained on a basis other than particle size.
19

Table 2: Values of Dispersion based on TEM crystallite size of palladium and CO- Adsorption
values from pulse-chemisorption studies. (Average particle sizes estimated from pulse-
chemisorption are shown for comparison)
Catalysts
CO Adsorption
NCOAd
Ds
Ds*
Percentage
dispersion #
(PDT)%
(µmol/g of Pd)
nm
nm
57.15
9
6
Pd/AC
Pd/SiO2
Pd/Al2O3
Pd/TiO2
19.23
13.67
14.38
17
22.87
21.61
20.07
23
25
26
9
8
7
NCOAd Normalized based on palladium content
Ds  Surface-weighted average Diameter calculated from pulse-chemisorption
Ds*  Calculated from TEM images
#PDT  Percentage dispersion calculated based on the obtained TEM particle size of the palladium
dispersed on the catalysts
20

3.3 TPD, TPO and TPR
TPD results for the catalysts are shown in Fig. 6. According to the literature on CO adsorption on
Pd metal[54–56], a desorption temperature in the range of 100-2000C is expected. For the
supported catalysts of our interest however, we observed significant differences from this range
depending on the support. In the case of Pd/AC, we observe a broad desorption peak, but
desorption starts at a relatively low temperature, as expected of linearly adsorbed CO (PdCO).
For all the oxide supports, the desorption peaks occur at much higher temperatures, with the peak
positions corresponding to bridge bonded CO (Pd2CO), which generally desorbs at temperatures
greater than 2500C[52,57]. For Pd/SiO2, desorption starts at around 4400C with the peak at
6240C. For Pd/Al2O3, desorption starts at about 5300C and peaks at 6500C. For Pd/TiO2 two
peaks are seen; the first peak occurs at a temperature roughly similar to the case of SiO2, whereas
the second occurs at about 7500C. This second peak in the case of TiO2 is probably due to a
spillover of oxygen from the TiO2 lattice, which combines with the adsorbed CO to form
CO2[58]. Alternatively, it may also be due to the decomposition of adsorbed CO on the catalyst
surface into adsorbed carbon and CO2[54,57]. A similar profile for Pd/TiO2 has been reported by
Dropsch et al.[52], who confirmed the formation of CO2 at high temperatures by TPD-MS.
21

Figure 6: TPD profile of Pd/AC, Pd/SiO2, Pd/Al2O3 and Pd/TiO2
The high desorption temperatures in the case of oxide supports are indicative of a high degree of
metal-support interaction. Dropsch et al.[52] concluded from their studies that the strength of
interaction between Pd metal and support increases with increasing ionic character of the support
(SiO2< Al2O3<TiO2). A strong support interaction, in addition to causing a shift in the desorption
temperature, results in a decrease in the extent of carbon monoxide adsorption[53,54].Our results
22

are broadly in support of this; it may also be seen that the discrepancies between TEM and
chemisorption results for metal particle size (discussed earlier) are in the same order as the shift
in desorption peak temperatures for alumina and silica supports. From the TPD results, the TiO2
case seems qualitatively different from the others. Benvenutti et al.[53], who carried out an FTIR
study of hydrogen and carbon monoxide adsorption on Pt/TiO2, Pt/ZrO2, and Pt/Al2O3 found that
the carbonyl band shifts to lower wave numbers, as a consequence of the presence of hydrogen
on the support, possibly at the metal-support interface, and consequent electronic
interactions[53]. They also interpret their results as indicative of strong metal-support electronic
interaction.
TPO studies were carried out on all catalysts except for Pd/AC, in whose case the possibility of
combustion of the support precludes such a study. Oxygen consumption peaks were observed at
3660C, 3310C and 3070C for Pd/Al2O3, Pd/TiO2 and Pd/SiO2 respectively, showing that the
alumina support renders the oxidation of Pd more difficult as compared to the other two oxide
supports. Oxygen consumptions were determined from the peak areas and gave a consumption of
about 1 mol/mol Pd in the case of Pd/Al2O3, and much higher values for the other cases,
suggesting single/multi-layer adsorption in addition to reaction in these cases.
Results from our TPR studies (conducted immediately following TPO; results not shown)
showed a strong negative peak (hydrogen evolution) at a low temperature (below 1000C). This
has also been observed by other research groups (Shen et al.[54] and references therein[59–61]),
and is due to decomposition of PdxHy; Palladium has a propensity to form hydrides even at room
temperature[54], and the low temperature hydrogen evolution is due to the decomposition of
23

these species. The occurrence of this peak makes it difficult to detect any consumption peaks
which follow either at similar or at nearby temperatures in the case of oxide supports. Distinct
hydrogen consumption peaks were only observed at much higher temperatures (above 3500C)
which are due to reduction of oxide supports[29,62–64].
Electronic interaction between the metal and support is usually a major factor in such
cases[31]and XPS is a convenient tool to get an insight into this. The support may also influence
the catalytic activity due to other reasons such as its acidic/basic nature (see for example Gross et
al.[28] on CO and CO2 hydrogenation), and this will also be investigated in the following
sections.
3.4 XPS Analysis
XPS results for the four catalysts used in this work are shown in Figure 7. Peaks were observed
for each catalyst corresponding to Pd3d5/2 and Pd3d3/2. Palladium peaks were obtained by Pd3d
spectra fitting of raw data with Gauss-Lorentz curve after subtraction of Shirley background
using XPSPEAK4.1 software[65]. While fitting, a difference of 5.25 between spin orbital multi-
plates and area ratio were maintained (3d5/2:3d3/2:: 3:2). While we observed the same degree of
shift for both Pd 3d5/2 as well as for Pd 3d3/2, in the following, we focus on the shift of Pd 3d5/2
for interpretation. The fitted values of binding energies for different catalysts are presented in
table 3.
24

Table 3: Values of Binding energies for the catalysts (Pd3d5/2)
3d5/2 Peak position
Corresponding %area
Catalysts
(Binding energy,
eV)
335.60
64.33
Pd/AC
336.60
333.66
35.67
11.81
334.74
336.07
46.91
30.88
Pd/SiO2
337.68
333.51
10.40
31.42
334.89
49.87
Pd/Al2O3
Pd/TiO2
336.41
333.12
18.71
20.63
334.43
79.37
25

Figure 7: XPS of Pd3d for the catalysts
The Pd 3d5/2 level binding energy reported for zero valent palladium (Pd0) is about 335.2 eV
[66–69]. The value obtained for Pd on carbon is about 335.6 eV, which agrees with the value
reported by Bertolini et al.[70]. The peak at 336.6 eV is due to PdOx/Pd[68,69]. In the case of Pd
on SiO2 and Pd on Al2O3, we observe peaks showing both positive and negative shifts in the
binding energy, while for Pd on TiO2 only negative shifts were observed. The peaks with the
positive shifts are again due to the oxidized forms PdO (336-337.2) and PdO3 (337.6-337.7)[69].
26

The binding energy values 334.74 eV and 334.89 eV that we find for Pd on SiO2 and Al2O3,
match with those in literature for Pd/SiOx/Si and Pd/Al2O3/Al[71,72]. Sarapatka [72] proposed
that interaction between metal and support is by charge transfer at the metal-support interface.
Creation of an interface dipole (Pd/Al2O3) may attract free electrons from elemental Al to the
sample surface. Looking at the differences in electro-negativity of Pd with Si (0.3), Al (0.6) and
Ti (0.66), we may surmise the possibility of charge transfer at the metal support interface
(support to metal) to be higher for TiO2 than for Al2O3 and SiO2 (and this would be consistent
with the negative shift in the binding energies and their corresponding areas; see Table 3). Bell et
al.[67] suggested that negative shifts may be due to SmPdxO formed during the preparation of
catalyst (Sm; Ti, Si, Al). Palladium present in SmPdxO structure can be more negative than Pd0
and cause a negative shift.
Thus the XPS evidence confirms the absence of any significant metal-support interaction in the
case of the carbon support, and the existence of varying degrees of interaction in the case of
oxide supports, which was suggested by the pulse and TPD results discussed earlier. Electronic
interactions , depend on a number of factors such as reducibility of the support, presence of
defects, and the nature of support materials[30]. The XPS results are strongly suggestive, in the
case of our catalysts, of an interaction involving a shift in the electron cloud towards the metal.
In oxidation, this would be expected to lower the capacity of the catalyst to adsorb the hydroxyl
species, which is essential for the first step of hydrogen abstraction from the glycerol
molecule[9,73]. Metal-support interaction of the kind observed would thus be expected to lower
the catalytic activity in the case of glycerol oxidation.
27

3.5 Acidity and alkalinity of Catalysts
In this work, a CO2-pulse has been used to characterize the basic sites of the oxide-supported
catalysts. We found that Pd/Al2O3 has basic sites, but not Pd/TiO2 and Pd/SiO2.
Table 4: Basicity of the catalyst in terms of CO2 adsorption capacity
S.N.
Catalysts
CO2 –adsorbed
(µmol/gcatalyst
)
Pd /Al2O3
Pd /TiO2
Pd /SiO2
18.6
0.5
1.
2.
3.
1.4
In order to explore the strength and concentration of acidic sites a pyridine-TPD (Fig. 8) was
performed. The peaks were de-convoluted by spectra fitting of raw data with Gaussian curve
using Fityk0.9.8[74] software. From the Pyridine-TPD (see Figure 8) it is clear that the acidity
(per unit weight of catalyst) is more in Pd/Al2O3 catalyst, followed by Pd/SiO2 and then Pd/TiO2.
The desorption peaks (after de-convolution) at temperatures higher than 5000C correspond to
strong acid sites[75]. On this basis, the catalysts may be ordered in the decreasing order of
acidity as; Pd/Al2O3>Pd/TiO2> Pd/SiO2. Table 5 reports the results on the basis of catalyst
28