Selective oxidation of glycerol catalyzed by Rh/activated carbon: Importance of support surface chemistry

Article abstract of DOI:10.1007/s10562-010-0515-9

Noble metal catalysts (Pt, Ir, Pd, Rh, Au) supported on activated carbon were assessed for glycerol oxidation. Rhodium is a highly efficient catalyst when the support has neutral or basic properties. The surface chemistry of activated carbon plays a key role in the performance.

Full text of DOI:10.1007/s10562-010-0515-9

Catal Lett (2011) 141:420–431
DOI 10.1007/s10562-010-0515-9
Selective Oxidation of Glycerol Catalyzed by Rh/Activated
Carbon: Importance of Support Surface Chemistry
Elodie G. Rodrigues
Xiaowei Chen Juan J. Delgado
Manuel F. R. Pereira
•
S o´ nia A. C. Carabineiro
•
•
•
Jos e´ L. Figueiredo
•
•
Jos e´ J. M. Orf a˜ o
´
Received: 9 October 2010 / Accepted: 22 November 2010 / Published online: 14 December 2010
Ó Springer Science+Business Media, LLC 2010
Abstract Noble metal catalysts (Pt, Ir, Pd, Rh, Au) sup-
ported on activated carbon were assessed for glycerol
oxidation. Rhodium is a highly efficient catalyst when the
support has neutral or basic properties. The surface
chemistry of activated carbon plays a key role in the
performance.
1 Introduction
The growing concern about global warming leads to
numerous discussions about sources of energy. The use of
renewable feedstocks is essential for the sustainable
development of society, since fossil energy resources are
limited. Over the last decade, biodiesel has emerged as a
viable clean fuel. It is obtained by transesterification with
methanol of triglycerides extracted from seed oils, leaving
glycerol as the main by-product. In this process, about
Keywords Glycerol ꢀ Oxidation ꢀ Noble metals ꢀ
Rhodium ꢀ Activated carbon ꢀ Surface chemistry
100 kg of glycerol are produced for every ton of biodiesel
[
1]. During the last years the market for biodiesel was
continuously increasing, and it is expected that this trend
remains [2]. Consumption of the surplus of glycerol is a
necessary requisite for the commercial viability of bio-
diesel production [3]. The conversion of glycerol into
high-value chemicals is a research area that has received
tremendous attention in recent years, as a result of glyc-
erol’s unique structure, properties, bioavailability and
renewability [2, 4, 5]. Liquid phase catalytic oxidation is a
promising route to convert glycerol into useful compounds,
provided that the catalyst used is sufficiently active and
selective for the formation of chemicals such as glyceric
acid (GLYCEA) and/or dihydroxyacetone (DIHA), poten-
tially useful as chemical intermediates in the fine chemicals
industry, particularly in pharmaceuticals [2, 6]. Until now,
these products are being produced by expensive and non-
environmentally safe oxidation processes, or by low pro-
ductivity microbial fermentation processes [7]. However,
the extensive functionalization of this molecule, with
similar reactive hydroxyl groups, renders its selective
oxidation particularly difficult [8].
E. G. Rodrigues ꢀ S. A. C. Carabineiro ꢀ
´
J. L. Figueiredo ꢀ M. F. R. Pereira ꢀ J. J. M. Orf a˜ o (&)
Laborat o´ rio de Cat a´ lise e Materiais (LCM), Laborat o´ rio
Associado LSRE/LCM, Departamento de Engenharia Qu ´ı mica,
Faculdade de Engenharia, Universidade do Porto,
Rua Dr. Roberto Frias, 4200-465 Porto, Portugal
e-mail: jjmo@fe.up.pt
E. G. Rodrigues
e-mail: elodie.rodrigues@fe.up.pt
S. A. C. Carabineiro
e-mail: scarabin@fe.up.pt
J. L. Figueiredo
e-mail: jlfig@fe.up.pt
M. F. R. Pereira
e-mail: fpereira@fe.up.pt
X. Chen ꢀ J. J. Delgado
Departamento de Ci eˆ ncia de los Materiales e Ingenier ´ı a
Metal u´ rgica y Qu ´ı mica Inorg a´ nica, Facultad de Ciencias,
Universidade de Cadiz, Campus Rio San Pedro Puerto Real,
11510 Cadiz, Spain
e-mail: xiaowei.chen@uca.es
In recent years, there have been many studies dealing
with the oxidation of glycerol to chemical intermediates
using supported catalysts with noble metal nanoparticles,
J. J. Delgado
e-mail: juanjose.delgado@uca.es
1
23

Selective Oxidation of Glycerol Catalyzed by Rh/Activated Carbon
421
such as Pd, Pt and mainly Au on diverse supports [9–15].
The main product of glycerol oxidation using platinum
supported on activated carbon is GLYCEA (selectivity
NORIT ROX 0.8 was selected as starting material for the
preparation of supports with different surface chemistries.
5
5% at 90% conversion) with a small amount of DIHA
2.1.2 Preparation Procedures
(
selectivity 12%) [11]. Pd catalysts supported on activated
carbon under basic conditions have a high selectivity
towards GLYCEA (77 at 90% conversion), whereas the
selectivities to the other reaction products (including
DIHA) are always lower than 10% [11]. The main disad-
vantage of catalysts that are based on these metals is their
deactivation at increasing reaction time due to oxygen
poisoning [12, 16]. On the other hand, gold has a better
resistance to oxygen poisoning compared to platinum,
allowing for the use of high oxygen partial pressures [17].
However, the catalytic activity of gold depends mainly on
nanoparticle sizes, which in turn depend on the preparation
method used [12, 18].
Gold, iridium, palladium, platinum and rhodium catalysts
were prepared by incipient wetness impregnation on the
commercial activated carbon. In addition, a gold catalyst
was also prepared via the sol immobilization method [19].
The NORIT ROX 0.8 activated carbon has a small
amount of surface oxygenated groups. In order to evaluate
if the activity of the catalysts correlates with the concen-
tration of oxygenated surface functional groups, the origi-
nal support was modified by chemical and thermal
treatments to obtain supports with different surface chem-
ical properties. Rhodium was supported on these modified
activated carbon materials.
Under alkaline conditions, catalyst activity and selec-
tivity seem to be influenced by the nature of the metal,
nanoparticle sizes and support. However, catalytic oxida-
tion of glycerol has been investigated by several research
groups under different reaction conditions (pH, tempera-
ture, oxidants, etc.) with different metals and different
supports, leading to a difficult catalyst evaluation and
comparison, since many parameters influence catalyst
performance. Therefore, this work begins with a compar-
ative study among catalysts under the same reaction con-
ditions and preparation method. Additionally to metals
already studied in the literature (platinum, palladium and
gold), two other metals (iridium and rhodium) were tested
in the glycerol oxidation reaction. Moreover, the influence
of the activated carbon surface chemical groups on the
rhodium catalyst performance was evaluated. For this
purpose, the support surface was modified by appropriate
chemical and thermal treatments. The possible role of the
surface functional groups of the support in the reaction
mechanism was also considered.
2.1.2.1 Preparation of Supports A NORIT ROX 0.8
activated carbon (pellets of 0.8 mm diameter and 5 mm
length) was used as starting material for this study (sample
AC ). This material was submitted to different chemical
0
and thermal treatments. Both liquid phase oxidations and
thermal treatments have been reported not to significantly
change the textural properties of activated carbons [20, 21].
An adequate amount of sample AC was introduced in a
0
Soxhlet extraction apparatus connected to a boiling flask,
with a nitric acid 6 M solution, and to a condenser. The
reflux was stopped after 3 h. The oxidized material was
subsequently washed with distilled water until neutral pH,
and then dried at 110 °C for 24 h (sample AC ). The pur-
1
pose of this liquid phase oxidation is to increase the amount
of oxygen-containing surface groups on the support.
All the thermal treatments were performed on sample
AC , since it is important that the starting material presents
1
a large amount of surface groups in order to produce
activated carbons with a high basic character by heat
treatment at high temperatures [22]. Support AC1 was
treated under nitrogen flow for 30 min at different tem-
peratures (400, 600 or 900 °C) followed by treatment under
dry air flow at room temperature for 1 h. These thermal
treatments selectively remove the oxygen-containing sur-
face groups previously introduced [20, 21]. The final
treatment in air is intended to stabilize the surface chem-
istry of the samples. The following samples were obtained:
2
2
2
Experimental
.1 Catalysts Preparation
.1.1 Materials
The metal precursors H PtCl (99.9% ACS, Pt min 37.5%),
2
AC1tt400—AC thermally treated under nitrogen flow at
1
6
PdCl (99.9% ACS, Pd min 59.5%), RhCl (99.9% ACS,
2
400 °C for 30 min plus 1 h under dry air flow at room
temperature; AC1tt600—similar to AC1tt400 but with treat-
ment at 600 °C; AC1tt900—similar to the others but treated
at 900 °C. These treatments were carried out in order to
remove selectively the oxygen-containing groups on the
surface, while maintaining the original textural properties
as far as possible.
3
Rh 38.5–45.5%), (NH ) IrCl (99.9% ACS, Ir 38–42%)
6
4
3
and HAuCl ꢀ3H O (99.99% ACS, Au min 49.5%) were
4
2
supplied by Alfa Aesar.
NaBH of purity [95% from Riedel-de Ha e¨ n, NaOH
4
([97%) and polyvinylalcohol (PVA) (purity [ 99%) from
Aldrich were also used. The commercial activated carbon
123

4
22
E. G. Rodrigues et al.
Before use, all samples were ground to a particle
characterized by microscopy and inductively coupled
plasma (ICP) analysis.
diameter between 0.1 and 0.3 mm in order to minimize
mass transfer resistance effects.
2.2.1 N Adsorption
2
2
.1.2.2 Preparation of Catalysts Au, Pt, Pd, Rh and Ir
catalysts were prepared by incipient wetness impregnation
of activated carbon with aqueous solutions of the corre-
sponding metallic precursors (HAuCl , H PtCl , PdCl ,
Textural characterization of the samples was based on
the analysis of nitrogen adsorption isotherms, measured at
-196 °C in a NOVA 4200e Quantachrome Instruments
apparatus. BET surface areas (SBET) were calculated, as
well as micropore volumes (Vmicro) and mesopore surface
areas (Smeso) according to the t-method.
4
2
6
2
RhCl , (NH ) IrCl ). The amount of noble metal was cal-
3
4 3
6
culated to obtain a nominal metal loading of 1 wt%. In a
complementary study, Au catalyst was also prepared by the
sol immobilization method with NaBH as reducing agent
4
(
Auc). Briefly, HAuCl ꢀ3H O (35.1 mg) was dissolved in
2.2.2 Temperature Programmed Reduction (TPR)
4
2
6
90 mL of H O, and polyvinyl alcohol was added (1.6 mL,
2
0
.2 wt%) under stirring. NaBH (4 mL, 0.1 M) was added
4
Temperature programmed reduction allows to find the
most adequate reduction temperature for each metal. TPR
profiles were obtained with a fully automated AMI-200
(Altamira Instruments). The catalyst (150 mg) was placed
in a U-shaped quartz tube inside an electrical furnace and
to the yellow solution under vigorous magnetic stirring.
The resulting sol was ruby-red in color. Within a few
minutes of sol generation, the colloid was immobilized by
adding the support under vigorous stirring. The amount of
support was calculated to reach a final metal loading of
-
1
submitted to a 5 °C min linear temperature increase up
to 600 °C under 5% (v/v) hydrogen flow diluted in helium
1
wt%. After 3–4 days, the colorless solution was filtered,
3
-1
the catalyst washed thoroughly with distilled water until
(total flow rate = 30 Ncm min ). A thermal conductiv-
ity detector and a quadrupole mass spectrometer (Dymax-
ion 200, Ametek) were used to monitor the change in
the filtrate was free of chloride (AgNO test) and dried at
3
1
10 °C for 24 h [19].
After heat treatment under nitrogen flow for 3 h, the
composition of the reactor effluent. H consumption peaks
2
catalysts were activated by reduction under hydrogen flow
for 3 h. The appropriate reduction temperature for each
catalyst was determined by temperature programmed
reduction (TPR), namely 250 °C for the Pd catalyst,
indicate directly the temperature ranges where reduction
occurs. The selected reduction temperatures are indicated
in Sect. 2.1.2.2.
3
50 °C for Ir, Pt and Au catalysts and 200 °C for Rh cat-
2.2.3 Surface Chemistry Characterization
alysts (see Sect. 2.2.2). The treatment under nitrogen flow
was carried out at the same temperature used for the
reduction.
The qualitative and quantitative determination of oxygenated
surface functional groups was performed by temperature pro-
grammed desorption—mass spectrometry (TPD-MS) [20, 21].
A supplementary catalyst, called Rh/PTAC1tt600, was
also prepared. It was obtained similarly to Rh/AC as fol-
1
CO and CO TPD spectra were obtained with the fully auto-
2
lows: while Rh/AC was heat-treated under nitrogen at
1
mated AMI-200 (Altamira Instruments) already mentioned.
The samples (150 mg) were placed in a U-shaped quartz
2
00 °C, Rh/PTAC1tt600 was heat-treated at 600 °C before
-1
reduction at 200 °C.
tube inside an electrical furnace and subjected to a 5 °C min
An additional catalyst (Au/C), provided by the World
Gold Council, was used as reference. Data sheets accom-
panying this sample indicate that it was prepared by the sol
method and has a metal load of 0.8 wt%; the support is a
carbon black. This catalyst was used without any additional
treatment.
linear temperature increase up to 1100 °C under helium flow
(25 Ncm min ). A quadrupole mass spectrometer (Dymax-
3 -1
ion 200, Ametek) was used to monitor CO and CO signals. For
2
quantification of the CO and CO released, calibration of these
2
gases was carried out at the end of each analysis.
The determination of the pH at the point of zero charge
(pHpzc) of the samples was carried out as follows [23].
2
.2 Characterization of Supports and Catalysts
50 mL of NaCl 0.01 M solution was introduced in closed
Erlenmeyer flasks. The pH was adjusted to a value between
2 and 12 by adding HCl 0.1 M or NaOH 0.1 M solutions.
Then, 0.15 g of the carbon sample was added and the final
pH measured after 48 h under agitation at room tempera-
ture. Blank experiments (without addition of support) were
also prepared for each pH, and the values measured after
48 h correspond to pHinitial. The pHpzc is the point were the
The supports were characterized by N2 adsorption at
196 °C, temperature programmed desorption (TPD) and
by determination of the pH at the point of zero charge
pHpzc). The catalysts were also characterized by tempera-
-
(
ture programmed reduction (TPR). Samples that showed
the most interesting catalytic performances were further
1
23

Selective Oxidation of Glycerol Catalyzed by Rh/Activated Carbon
423
curve pHfinal versus pHinital crosses the line pHfinal
pHinitial
=
yields by other authors) into the different products i at time
t were calculated as:
.
C =ðm ꢀ C Þ
i
0
Si ¼
ð1Þ
2
.2.4 Microscopy
X
-
where C is the concentration of product i (mol L ), C is
1
i
0
-
1
Transmission electron microscopy (TEM) measurements
were performed for the most promising catalysts in a
JEOL2010F instrument, with 0.19 nm spatial resolution
at Scherzer defocus conditions. High Angel Annular
Dark Field—Scanning Transmission Electron Microscopy
the initial concentration of glycerol (mol L ), X is the
glycerol conversion and m corresponds to the moles of
i produced per mol of glycerol consumed, according to the
stoichiometry.
(
HAADF-STEM) images were obtained with the same
microscope. An electron probe with diameter of 0.5 nm at
a diffraction camera length of 10 cm was used for the
HAADF mode. The samples were deposited on the Cu
grids coated holey-carbon for TEM analysis.
3
3
3
Results and Discussion
.1 Characterization of Supports and Catalysts
.1.1 Textural Properties
2
.2.5 Inductively Coupled Plasma (ICP)
The textural characterization of the different supports and
The metal loadings of the prepared catalysts were deter-
mined in duplicate by ICP, using an Iris Intrepid Spec-
trometer (Thermo Elemental). An Ethos 1600 microwave
labstation (Milestone) was used to prepare the acid solution
of all the samples for ICP analysis.
of the Pd/AC catalyst is presented in Table 1. As expec-
0
ted, no major differences in the textural properties of
the materials were observed. The slight decrease in AC1
surface area may be due to the abundant presence of
oxygen-containing groups introduced on the surface of the
activated carbon by treatment with HNO , which possibly
3
block the entry of N inside the small pores. After treat-
2
2
.3 Catalysts Evaluation
ment at 900 °C (sample AC1tt900), the mentioned func-
tional groups were removed by thermal decomposition and
the area of the corresponding support increased slightly
comparatively to the original carbon [24, 25].
The reaction tests were performed in a 350 mL stainless
steel reactor equipped with manometer, temperature sen-
sor, magnetic stirrer and sample outlet. The oxidation
reactions were carried out with oxygen under pressure up
to 10 bar and at temperatures from 40 to 100 °C. NaOH
solution (NaOH/glycerol molar ratio = 2) and catalyst
The textural parameters of the Pd/AC0 catalyst
decreased only slightly compared to the unloaded carbon.
Therefore, it was assumed that the textural properties of the
modified supports and of the supported metal catalysts are
not significantly different from those of the original acti-
vated carbon.
(
700 mg) were added to a 0.3 M aqueous solution of
glycerol (V = 150 mL) under stirring at 1000 rpm. Pre-
liminary experiments show that external mass transfer
limitations were overcome at this stirring speed. After
heating under nitrogen to the desired temperature, the
reaction was initiated by switching from inert gas to oxy-
gen. The reaction was monitored by taking samples
3.1.2 Surface Chemistry Characterization
The total amounts of the various surface groups (acidic
carboxylic acids, carboxylic anhydrides, lactones, phenols,
(
0.5 mL) for analysis at regular time intervals.
The quantitative analysis of the reaction mixtures was
carried out by high performance liquid chromatography
HPLC). The chromatograph (Elite LaChrom HITACHI)
was equipped with a refractive index and an ultraviolet
210 nm) detector. Reactant and products were separated in
Table 1 Textural properties and pHpzc of original and modified
activated carbons
(
2
-1
)
2
-1
3
-1
)
Sample
S
BET (m g
S
meso (m g
)
V
micro (cm g
pHpzc
(
an ion exclusion column (Alltech OA 1000). The eluent
was a solution of H SO (0.01 M). An injection volume of
AC
AC
0
1
1025
965
180
156
200
150
184
140
0.361
0.369
0.350
0.391
0.399
0.366
7.9
2
4
3.2
2
0
0 lL, a measuring time of 15 min and a flow rate of
.5 mL min were selected. Products were identified by
AC1tt400
952
n.d.
n.d.
8.2
-
1
AC1tt600 1042
AC1tt900 1106
comparison with standard samples. Glycerol ([99.5%) was
purchased from Fluka and all reaction products from
Pd/AC
0
887
n.d.
Sigma-Aldrich. The selectivities (S ) (also named relative
i
123

4
24
E. G. Rodrigues et al.
and carbonyls or quinones) on carbon materials can be
determined from TPD spectra, since these groups are
expected, the oxidized sample AC has a low pH (see
1 pzc
Table 1), due to the introduction of the mentioned oxygen-
containing functional groups with acidic properties, namely
carboxylic acids. Thermal treatments selectively remove
those surface groups, originating materials with progres-
sively lower oxygen content. Practically all groups were
removed from sample AC1tt900, which has basic properties
[27]. According to Leon y Leon et al. [28], this basic
character is mainly due to the electron rich oxygen-free
sites located on the carbon basal planes.
decomposed upon heating by releasing CO and/or CO in
2
different temperature ranges. Accordingly, it is possible to
identify the oxygenated groups on the surface of a given
material by TPD experiments. It was reported that CO2
results from the decomposition of carboxylic acids at low
temperatures (150–450 °C) and from lactones at high
temperatures (600–800 °C); carboxylic anhydrides origi-
nate both CO and CO2 (400–650 °C); groups such as
phenols (600–800 °C), carbonyls and
750–1000 °C) originate CO [20, 21].
Figure 1a and b shows, respectively, the CO and CO2
quinones
(
3.1.3 Microscopy and ICP Analyses
TPD spectra of the commercial activated carbon as
TEM micrographs were collected in order to get informa-
tion about the metal particle size distributions (Fig. 2). The
average crystallite sizes of the studied catalysts are pre-
sented in Table 2. It should be noticed than the value
corresponding to Ir/AC0 was obtained by hydrogen
chemisorption; details of the experimental technique used
can be found elsewhere [29]. When using the incipient
wetness method it was possible to obtain relatively small
metal crystallite sizes for the catalysts prepared. The only
exception was the gold catalyst, which required a different
method for promoting a high dispersion. It is well-known
received (AC ) and the modified samples. An increase in
0
the amount of surface oxygenated groups is evidenced by
the increase of CO and CO2 released. Analyzing the
results, it is clear that the liquid phase oxidation with HNO3
significantly increased the amount of CO evolved at low
2
temperatures (carboxylic acid groups) and introduced a
large amount of surface oxygen, which decomposes at high
temperatures by releasing CO (phenol and carbonyl/qui-
none groups), originating a support with acidic and
hydrophilic characteristics (sample AC1) [26]. As
[
12] that sizes of gold nanoparticles mainly depend on the
synthesis method used. For instance, the conventional
impregnation technique (incipient wetness) results in much
larger gold particles than the sol immobilization method
(50 nm vs. 6.8 nm). This is in part due to the negative
effect of the chloride involved in the preparation, which
results in poisoning and excessive sintering of the Au
particles [30]. This does not occur in the case of the other
noble metals, for which the incipient wetness method
allows to produce well-dispersed nanoparticles.
A large amount of oxygen-containing groups on the
activated carbon surface can influence the metal dispersion.
The surface oxygen groups were considered to act as
anchoring sites that interact with metallic precursors [31].
However, the functional groups present on activated carbon
are easily decomposed upon heating even at low temper-
atures. This occurs with some carboxylic acid groups,
which already decompose at temperatures below 200 °C,
as it can be seen in the TPD spectra (Fig. 1, sample AC1).
When the catalysts are submitted to heat treatment and
reduction processes, these oxygenated groups are thermally
decomposed and the rhodium particles that were anchored
to them may have low stability and could be prone to
sintering. This can possibly explain why Rh/AC1 and
Rh/PTAC1tt600 catalysts have larger metallic particles than
Rh/AC , Rh/AC
0
and Rh/AC1tt900, because on this
1tt600
latter group of samples, contrarily to the former, there was
no oxygenated groups that decompose at low temperatures
during the preparation step.
Fig. 1 TPD spectra of the different activated carbons: a CO. b CO
2
1
23

Selective Oxidation of Glycerol Catalyzed by Rh/Activated Carbon
425
Fig. 2 Particle sizes
(
a) 30
(b) 10
distribution of a Pd/AC
b Pt/AC . c Auc/AC
d Rh/AC . e Rh/AC
0
.
0
0
.
25
20
15
10
5
0
8
6
4
2
0
0
1
.
f Rh/AC1tt600. g Rh/PTAC1tt600
and h Rh/AC1tt900. In some
cases the HAADF-STEM
images are inserted
Particle Size (nm)
Particle Size (nm)
Particle Size (nm)
Particle Size (nm)
Particle Size (nm)
(
c) 10
(d) 50
8
6
4
2
0
40
30
20
10
0
Particle Size (nm)
(
e) 35
(f) 35
3
2
2
1
1
0
5
0
5
0
5
0
30
25
20
15
10
5
0
Particle Size (nm)
(
g) 35
(h) 35
3
2
2
1
1
0
5
0
5
0
5
0
30
25
20
15
10
5
0
Particle Size (nm)
123

4
26
E. G. Rodrigues et al.
Table 2 Average crystallite size and metal content of the studied
catalysts
lower than palladium and rhodium, the low performance
observed is probably due to oxygen poisoning, which is
directly related to the oxygen partial pressure, as mentioned
in the literature [11]. This can also explain the inactivity of
the iridium catalyst. However, according to van Dam et al.
Catalyst
Metal loading (%)
M
d (nm)
Pd/AC
Pt/AC
Ir/AC
Au/AC
Auc/AC
0
0.65 ± 0.03
0.50 ± 0.02
n.d.
2.0
3.8
1.7
50
0
[
33], metals with a higher oxidation potential are less prone
a
0
to oxidation. Thus, among the metals tested, platinum
would be the less easily poisoned by over-oxidation, fol-
lowed by iridium, palladium and rhodium, which was not
verified. Actually, incipient wetness allows obtaining very
active Pd and Rh catalysts in a simple, easy and quick way;
these catalysts promote a high glycerol conversion even at a
working pressure of 3 bar. The absence of deactivation by
oxygen on various platinum-group metals due to strong
metal-substrate interactions has already been reported [34].
So, the high activity obtained can possibly be attributed to
the high affinity of glycerol for rhodium and palladium,
which limits oxygen adsorption. This is particularly inter-
esting, since no study concerning rhodium catalysts for
glycerol oxidation has been published previously.
0
n.d.
0
0.32 ± 0.05
0.74 ± 0.05
0.74 ± 0.01
0.74 ± 0.01
0.78 ± 0.01
0.70 ± 0.03
0.8
6.8
1.3
2.0
1.3
1.3
1.8
11
Rh/AC
Rh/AC
0
1
Rh/AC1tt600
Rh/AC1tt900
Rh/PTAC1tt600
Au/C
a
Obtained by hydrogen chemisorption
Table 2 summarizes the metal loading of some of the
catalysts. Regarding to the sol immobilization method, the
electrostatic interaction between the stabilized Au particles
and the support play an important role during the prepa-
ration [32]. The pH of the gold sols was about six. At this
pH the zeta potential of gold sols is negative [32] while the
surface charge of activated carbon is only slightly positive,
considering that their pHpzc is 7.9. Thus, the interaction
between the support and the gold sol was relatively weak,
Contrarily to the other catalysts prepared by the incipi-
ent wetness method, the gold catalyst was the only that
presented low particle dispersion. This can explain its
inactivity once, as already mentioned, well-dispersed gold
nanoparticles are required in order to obtain an active
catalyst. Therefore, to study the influence of the prepara-
tion method on the gold catalyst performance, a sol
immobilization method was applied. From TEM analysis
(Table 2), it was concluded that this method appears to be
suitable for generating small crystallite sizes supported on
active carbon, originating an active catalyst as it can be
observed in Fig. 3. When comparing with the reference
which can explain the low loading of the Auc/AC catalyst.
0
Chemical and heat-treatments of activated carbon did
not influence substantially the metal content of rhodium
catalysts.
3
.2 Oxidation Experiments
Au/C catalyst, Auc/AC exhibits a much higher activity. In
0
3
.2.1 Catalyst Evaluation
1
0
.0
.8
Auc
Pd
Rh
Au/C
Pt
Many parameters play a role in determining the activity of
catalysts in the liquid phase oxidation of glycerol. It is not
easy to compare the performances of catalysts that are
tested under widely different conditions, as reported in the
literature. Therefore, in the first part of this work, several
Ir
Au
0.6
metals were supported on activated carbon AC , using the
0
0
0
0
.4
.2
.0
same preparation method (incipient wetness), and the same
operational conditions were maintained in all catalytic
tests.
The conversion of glycerol on AC supported catalysts
0
was studied as a function of time for different metals. Using
typical conditions (60 °C, 3 bar, 150 mL of glycerol 0.3 M,
NaOH/glycerol = 2 mol/mol, catalyst amount = 700 mg),
the performances presented in Fig. 3 were obtained, which
decrease in the order: Pd [ Rh ꢁ Pt. Under the conditions
tested, Ir and Au prepared by incipient wetness are not
active. Even though platinum has a dispersion slightly
0
50
100
150
200
250
300
Time [min]
Fig. 3 Oxidation of glycerol over noble metal catalysts supported on
activated carbon AC
. Reaction conditions: 60 °C, pO2 = 3 bar,
50 mL of glycerol 0.3 M, NaOH/glycerol = 2 mol/mol, catalyst
amount = 700 mg
0
1
1
23

Selective Oxidation of Glycerol Catalyzed by Rh/Activated Carbon
427
fact, after 3 h, this gold catalyst supported on activated
carbon is able to reach 100% glycerol conversion, whereas
only 44% is achieved with Au/C. This can be explained by
the larger gold particles of the latter catalyst (11 nm vs.
0.7
.6
0.5
1
0
0
0
0
0
.0
.8
.6
.4
.2
.0
0
GLYCEA
DIHA
6
.8 nm), but a support effect can also influence its per-
0
0
0
0
.4
.3
.2
.1
formance, since these gold particles are supported on car-
bon black. It should be noted that the prepared Auc/AC0
has a remarkable catalytic activity, particularly considering
its low metal loading (0.32%).
The selectivities to products obtained in the presence of
the different catalysts are compared in Table 3 at constant
reaction time (3 h), since the selectivities to the main
products changed only slightly during the reaction. How-
ever, for comparison purposes, Table 3 also shows the
selectivities at the same glycerol conversion (25%). The
evolution of DIHA and GLYCEA selectivities as a function
0.0
0
50
100
150
200
250
300
Time [min]
Fig. 4 Evolution of DIHA and GLYCEA selectivities and glycerol
conversion during the reaction in the presence of the Rh/AC catalyst
0
of time in the presence of Rh/AC can be seen in Fig. 4.
0
was examined. The aim of this investigation was to
establish the dependency of the reaction rate and catalyst
selectivity on the oxygen pressure and reaction temperature
The prepared catalysts lead to a high total selectivity of
about 75–80% to products of commercial interest (GLY-
CEA ? DIHA), and a very similar distribution of the
products was observed independently of the nature of the
metal. Besides the desired GLYCEA and DIHA, glycolic
acid (GLYCOA), glyceraldehyde (GLYCER) and tartronic
acid (TARTA) were also measured; oxalic acid was
detected in some cases, but always with very low con-
centrations. The selectivity to GLYCEA is higher than to
DIHA for all the catalysts prepared (around 60% after 3 h).
The selectivities to GLYCOA and to DIHA are similar
for the Rh/AC catalyst.
0
Experiments were carried out with 150 mL of glycerol
(0.3 M), NaOH/glycerol ratio = 2 mol/mol and 700 mg of
catalysts. The oxygen pressure was varied between 3 and
10 bar and the temperature between 40 and 100 °C.
Under the tested conditions, it was found that the oxy-
gen pressure influences significantly the reaction rates of
glycerol oxidation (Fig. 5a). The initial reaction rate
increases with the oxygen pressure but the catalyst deac-
tivates more readily. Increasing the oxygen pressure seems
to lead to a higher oxygen coverage of the surface, and thus
to a premature deactivation of the rhodium catalyst due to
oxygen poisoning of the surface. This metal appears to be
very sensitive to the amount of oxygen dissolved in the
reaction medium. This situation is similar to that described
by Garcia et al. [11] for Pt catalysts.
(
between 10 and 20%). The performance of the reference
Au/C catalyst differs, as it seems to promote almost in the
same way the oxidation of primary and secondary alcohol
functions, since the observed selectivities to DIHA and
GLYCEA after 3 h are 38 and 42%, respectively.
3
.2.2 Influence of Pressure and Temperature
Selectivity to DIHA is slightly increased with the
increase of oxygen pressure while selectivity to GLYCOA
decreases (Table 4).
A more detailed study was carried out with the rhodium
catalyst, which as far as we are aware is reported here for
the first time as active in glycerol oxidation. The influence
of oxygen pressure and temperature on the reaction kinetics
The temperature increase leads to an increase of cata-
lytic activity in the system studied (Fig. 5b), but seems to
0
Table 3 Product selectivities at t = 3 h and X = 25% for glycerol oxidation over noble metal catalysts supported on activated carbon AC and
over the Au/C reference catalyst
t = 180 min
XGLY = 25%
X
GLY
S
GLYCEA
S
GLYCOA
S
DIHA
S
TARTA
S
GLYCEA
S
GLYCOA
S
DIHA
S
TARTA
Pd
0.95
0.88
0.20
1.0
0.64
0.62
0.60
0.60
0.42
0.11
0.15
0.15
0.14
0.20
0.16
0.17
0.20
0.18
0.38
0.09
0.06
0.04
0.05
0
0.62
0.61
0.61
0.61
0.42
0.12
0.16
0.15
0.16
0.21
0.17
0.17
0.21
0.17
0.37
0.09
0.06
0.03
0.05
0
Rh
Pt
Auc
Au/C
0.44
Reaction conditions: 60 °C, pO2 = 3 bar, 150 mL of glycerol 0.3 M, NaOH/glycerol = 2 mol/mol, catalyst amount = 700 mg
123

4
28
E. G. Rodrigues et al.
be very detrimental to the selectivity of the most interesting
products, especially glyceric acid, which oxidizes very
easily to glycolic acid and further to oxalic acid at high
temperatures (Table 4). Once again, the results obtained at
for the oxalic acid formed (selectivity = 0.05), the sum of
selectivities is still lower than one. As no additional
product was detected by HPLC, we supposed that, in this
particular case, some complete oxidation of the partial
oxidation products to CO occurs. The formation of CO as
4
0 and 50 °C show that rhodium suffers deactivation
2
2
resulting from the high level of oxygen present in the
reaction medium at these temperatures. It is a common
assumption that the mechanism of glycerol oxidation
involves an oxidative dehydrogenation step. The last step is
the elimination of hydrogen by reaction with dissociatively
adsorbed oxygen [35]. Therefore, an increase of the oxygen
coverage on the metal surface progressively eliminates
glycerol adsorption. Over-oxidation leads to a loss of
a by-product in the glycerol oxidation was also observed by
Carrettin et al. [9].
It can be concluded that, for rhodium catalyst, a com-
promise between activity and selectivity has to be achieved
using suitable temperature and oxygen pressure in order to
avoid problems of over-oxidation.
3.2.3 Investigation of the Support Effects
in the Glycerol Oxidation
n?
activity due to the transformation of metal M to M , the
conversion reaching a plateau (Fig. 5a, b). However, this
deactivation can be eliminated by decreasing the rate of
oxygen supply (e.g. by working at higher temperatures
The surface oxygenated functional groups of activated
carbon can be controlled by chemical and heat treat-
ments, and can change the catalytic properties. Figure 6
reveals a strong influence of the surface chemical prop-
erties of the support on the catalytic performance in
glycerol oxidation. It was observed by ICP analysis that
the metal loading did not change significantly from
support to support. As the textural properties of the
support also do not differ substantially, the metal particle
sizes were analyzed by TEM to establish differences in
metal dispersion (Fig. 2; Table 2). Rh/AC , Rh/AC
[
16], as can be seen in Fig. 5b).
In some cases, the sum of the calculated selectivities in
Table 4 is lower than one, since some oxalic acid was also
detected. It is important to notice that in the experiment
with the Rh/AC catalyst at 100 °C, even after accounting
0
(
a) 1.0
0
1tt600
0
0
0
0
0
.8
.6
.4
.2
.0
and Rh/AC1tt900 catalysts have similar crystallite sizes,
which are on average smaller than those of Rh/AC and
1
1
3
6
1
.5 bar
bar
bar
Rh/PTAC1tt600. The differences of metal dispersions in
these activated carbon supported rhodium catalysts are
mainly related to the oxygen-containing groups anchored
on the surface of the support, as already mentioned (see
Sect. 3.1.3). However, even though Rh/AC , Rh/AC
0 bar
0
1tt600
and Rh/AC1tt900 catalysts have similar particles sizes,
rhodium particles supported on AC0 exhibit a much
higher activity. The histograms of the rhodium particle
0
50
100
150
200
250
300
Time [min]
size distribution for AC , AC
0
and AC1tt900 supported
1tt600
(
b) 1.0
catalysts are shown in Fig. 2d, f, h, respectively. It can
be seen that the distribution of rhodium particle diame-
ters in the Rh/AC0 catalyst is less homogeneous than
those of Rh/AC1tt600 or Rh/AC1tt900. Although most
particles have about 1.0–1.5 nm, there are a few particles
larger than 2.25 nm that are not present in the other two
catalysts. Then, it seems that the deactivation of the
Rh/AC1tt600 and Rh/AC1tt900 catalysts can be explained
by over-oxidation, since small metal nanoparticles
0
0
0
0
0
.8
.6
.4
.2
.0
4
0 ºC
50 ºC
6
0 ºC
80 ºC
00 ºC
1
(\2 nm) deactivate more readily because of their stronger
0
50
100
150
200
250
300
affinity to oxygen [35]. In fact, in Sect. 3.2.2., it was
already observed that rhodium is very sensitive to
oxygen. Moreover, an experiment carried out with
Rh/AC1tt900 shows that over-oxidation can be delayed
when working at an oxygen pressure of 0.5 bar (see
Fig. 6), which confirms that the deactivation observed is
Time [min]
Fig. 5 Dependency of glycerol conversion on the oxygen pressure at
0 °C (a) and on reaction temperature at pO2 = 3 bar (b) with Rh/
AC . Other reaction conditions: 150 mL of glycerol 0.3 M, NaOH/
glycerol = 2 mol/mol, catalyst amount = 700 mg
6
0
1
23

Selective Oxidation of Glycerol Catalyzed by Rh/Activated Carbon
429
Table 4 GLYCEA, GLYCOA, DIHA and TARTA selectivities (SGLYCEA, SGLYCOA,
temperatures, for 50% glycerol conversions (XGLY) and after 2 h of reaction in the presence of the Rh/AC
S
DIHA, STARTA) at different oxygen pressures and
0
catalyst
t = 120 min XGLY = 50%
X
GLY
S
GLYCEA
S
GLYCOA
S
DIHA
S
TARTA
S
GLYCEA
S
GLYCOA
S
DIHA
S
TARTA
Rh/AC
0
a
1
3
6
1
5
6
8
1
.5 bar
a
bar
0.52
0.74
0.65
0.50
0.43
0.74
1.0
0.58
0.61
0.61
0.58
0.57
0.61
0.41
0.30
0.21
0.16
0.12
0.13
0.14
0.16
0.20
0.25
0.16
0.17
0.20
0.22
0.23
0.17
0.27
0.28
0.05
0.06
0.07
0.07
0.06
0.06
0.07
0.06
0.58
0.61
0.61
0.58
0.58
0.61
0.43
0.30
0.21
0.16
0.12
0.13
0.14
0.16
0.22
0.25
0.16
0.17
0.20
0.22
0.23
0.17
0.29
0.30
0.05
0.06
0.07
0.07
0.05
0.06
0.06
0.04
a
bar
a
0 bar
b
0 8C
0 8C
0 8C
b
b
b
00 8C
1.0
a
Reaction conditions: 60 °C, 150 mL of glycerol 0.3 M, NaOH/glycerol = 2 mol/mol, catalyst amount = 700 mg
Reaction conditions: pO2 = 3 bar, 150 mL of glycerol 0.3 M, NaOH/glycerol = 2 mol/mol, catalyst amount = 700 mg
b
1
0
0
0
0
0
.0
.8
.6
.4
.2
.0
AC0
due to oxygen poisoning; however, as expected, the ini-
tial reaction rate is much lower. In conclusion, the
obtained results suggest that rhodium supported on acti-
vated carbon seems to be a suitable catalyst for glycerol
oxidation provided that its particles are large enough to
avoid over-oxidation. Nevertheless, when considering the
Rh/AC1 catalyst, which has a significant fraction of
particles larger than 2 nm, only a residual activity is
observed. This indicates that surface acid groups are
prejudicial for rhodium catalyst activity. Indeed, when a
similar catalyst is prepared and post-heat-treated at
AC1
AC1tt400
AC1tt600
AC1tt900 (P = 0.5 bar)
AC1tt900
PTAC1tt600
600 °C (instead of 200 °C) before reduction, the more
acidic groups are eliminated, and activity increases dra-
matically, although the average particle size remains
0
50
100
150
200
250
300
approximately the same. This catalyst (Rh/PTAC1tt600
)
Time [min]
does not have carboxylic acid or carboxylic anhydride
groups. Additionally, a Rh/PTAC1tt400 catalyst (heat-
treated at 400 °C) was prepared in order to remove only
carboxylic acids from the surface, but no enhancement of
the activity was observed when compared with Rh/AC1
Fig. 6 Influence of modified activated carbon supports on the
activity of rhodium catalysts. Reaction conditions: 60 °C,
pO2 = 3 bar, 150 mL of glycerol 0.3 M, NaOH/glycerol = 2 mol/
mol, catalyst amount = 700 mg
(
(
data not shown). So, not only carboxylic acid groups
catalyst (0.31 vs. 0.17 at t = 3 h and 0.28 vs. 0.18 at
X = 50%) and a decrease in the selectivity to GLYCEA
(0.53 vs. 0.62 at t = 3 h and 0.56 vs. 0.61 at X = 50%).
The presence of surface groups may cause differences in
adsorption performance [36], i.e. the functional groups
located near the metal particles can influence glycerol
released by thermal treatment up to 400 °C) but also the
carboxylic anhydrides (released between 400 and 600 °C)
appear to lead to almost inactive catalysts.
Two different deactivation procedures were distin-
guished: over-oxidation (oxygen poisoning) and the pres-
ence of a large amount of surface oxygenated acidic
groups. Therefore, rhodium catalysts appear to be suitable
for glycerol oxidation, provided that their crystallites are
not too small and the support has neutral or basic proper-
ties. These two parameters can easily be controlled by
thermal treatments.
adsorption. The AC
support has the largest amount of
catalyst has the most
1
surface groups, so the Rh/AC
1
hydrophilic character, and consequently the highest affinity
for the aqueous phase. This functionalized carbon surface
facilitates adsorption of glycerol from water. So, glycerol
may adsorb preferentially on the support and/or functional
groups can impede the access of the glycerol molecule to
the metal. However, the Rh/PTAC1tt600 catalyst shows an
increase of activity, in spite of the relatively high content of
In terms of selectivity, the presence of oxygenated sur-
face groups on Rh/PTAC1tt600 leads to an increase in the
selectivity to DIHA when compared with the Rh/AC0
123

4
30
E. G. Rodrigues et al.
oxygenated groups on the surface (despite its less acidic
high total selectivity to products of commercial interest
(about 80%).
nature). Moreover, a study carried out on AC , without any
0
additional treatment, allowed to conclude that glycerol also
adsorbs on this support, and its adsorption is well-repre-
sented by the Langmuir isotherm [37]. Then, the adsorption
of glycerol on the support does not seem to explain, at least
completely, the largely different activities.
Acknowledgments This work was carried out with support of
Funda c¸ a˜ o para a Ci eˆ ncia e a Tecnologia (FCT) under research fel-
lowship BD/45280/2008 (for EGR), CIENCIA 2007 program (for
SACC) and by FCT/FEDER in the framework of Program COM-
PETE (project PTDC/EQU-ERQ/101456/2008).
Basic carbons are characterized by a high content of
electron rich sites on their basal planes and a low con-
centration of electron withdrawing groups; the presence of
oxygen-containing functional groups with electron-with-
drawing properties has a negative effect on the adsorption
of anionic species [28]. So, assuming that the glycerol
oxidation mechanism on rhodium involves the spillover of
References
1
2
. Ma F, Hanna MA (1999) Bioresour Technol 70:1
. Zhou CHC, Beltramini JN, Fan YX, Lu GQM (2008) Chem Soc
Rev 37:527
3
4
. Haas MJ, McAloon AJ, Yee WC, Foglia TA (2006) Bioresour
Technol 97:671
. Behr A, Eilting J, Irawadi K, Leschinski J, Lindner F (2008)
Green Chem 10:13
activated oxygen from the metal to the support, Rh/AC is
1
the catalyst with the lower capacity to accept these species.
Therefore, Rh/AC cannot store oxygen, as the support is
1
5. Pagliaro M, Ciriminna R, Kimura H, Rossi M, Della Pina C
2007) Angew Chem Int Ed 46:4434
(
already oxygenated, and the reaction will stop; on the
contrary, Rh/PTAC1tt600 has some oxygen-free carbon
sites, which could provide for an enhancement of activity.
Although it is assumed that alcohol oxidations on metal
surfaces proceed via a dehydrogenation mechanism, the
precise reaction pathway is still not clear, which further
complicates the interpretation of these results. Neverthe-
less, these experiments clearly demonstrate that the support
surface chemistry is a key-factor in the catalytic oxidation
of glycerol, at least when using rhodium catalysts.
6
. Golz-Berner K, Zastrow L (2007) United States Patent Applica-
tion # 10/570,031
7
. Brandner A, Lehnert K, Bienholz A, Lucas M, Claus P (2009)
Top Catal 52:278
8
9
. McMorn P, Roberts G, Hutchings GJ (1999) Catal Lett 63:193
. Carrettin S, McMorn P, Johnston P, Griffin K, Kiely CJ,
Hutchings GJ (2003) Phys Chem Chem Phys 5:1329
1
0. Demirel-G u¨ len S, Lucas M, Claus P (2005) Catal Today
102–103:166
1. Garcia R, Besson M, Gallezot P (1995) Appl Catal A 127:165
2. Porta F, Prati L (2004) J Catal 224:397
3. Gao J, Liang D, Chen P, Hou ZY, Zheng XM (2009) Catal Lett
1
1
1
1
30:185
1
1
1
4. Bianchi CL, Canton P, Dimitratos N, Porta F, Prati L (2005)
Catal Today 102:203
5. Musialska K, Finocchio E, Sobczak I, Busca G, Wojcieszak R,
Gaigneaux E, Ziolek M (2010) Appl Catal A 384:70
6. Besson M, Gallezot P (2000) Catal Today 57:127
4
Conclusions
Under the same reactions conditions and for the same
support (activated carbon) and preparation method (incip-
ient wetness), the catalytic performance in the selec-
tive oxidation of glycerol decreases in the order:
Pd [ Rh ꢁ Pt; on the other hand, Ir and Au are not active.
Contrarily to the other catalysts prepared by the conven-
tional method, the gold catalyst presented low particle
dispersion. This can explain its inactivity once well-dis-
persed gold nanoparticles are required in order to obtain an
active catalyst. This can be achieved by using a sol
immobilisation method of preparation. Palladium and
rhodium showed high activities, whereas platinum was
rapidly deactivated due to oxygen poisoning. The rhodium
catalyst, as far as we know reported for the first time in the
present work as active for the reaction, was found to be
very promising, namely taking into account its simple
preparation method. Additionally, the influence of the
surface chemical properties of the support on the activity of
this catalyst was clearly demonstrated, being concluded
that the presence of acid groups are prejudicial. Particu-
larly, carboxylic acids and anhydrides appeared to be the
most detrimental groups. The prepared catalysts lead to a
17. Prati L, Rossi M (1998) J Catal 176:552
1
1
2
8. Bond GC, Thompson DT (1999) Catal Rev Sci Eng 41:319
9. Onal Y, Schimpf S, Claus P (2004) J Catal 223:122
´
0. Figueiredo JL, Pereira MFR, Freitas MMA, Orf a˜ o JJM (1999)
Carbon 37:1379
´
21. Figueiredo JL, Pereira MFR, Freitas MMA, Orf a˜ o JJM (2007) Ind
Eng Chem Res 46:4110
2
2
2. Papirer E, Li S, Donnet JB (1987) Carbon 25:243
3. Rivera-Utrilla J, Bautista-Toledo I, Ferro-Garcia MA, Moreno-
Castilla C (2001) J Chem Technol Biotechnol 76:1209
24. G o´ mez-Serrano V, Acedo-Ramos M, L o´ pez-Peinado AJ, Valen-
zuela-Calahorro C (1997) Thermochimi Acta 291:109
2
2
2
2
5. Tangsathitkulchai C, Ngernyen Y, Tangsathitkulchai M (2009)
Korean J Chem Eng 26:1341
6. Moreno-Castilla C, Lopez-Ramon MV, Carrasco-Marin F (2000)
Carbon 38:1995
7. Menendez JA, Phillips J, Xia B, Radovic LR (1996) Langmuir
1
8. Leon y Leon CA, Solar JM, Calemma V, Radovic LR (1992)
2:4404
Carbon 30:797
´
29. Soares OSGP, Orf a˜ o JJM, Pereira MFR (2009) Appl Catal B
9
1:441
0. Carabineiro SAC, Thompson DT, (2007) In: Heiz U, Landman U
ed) Nanocatalysis, Springer-Verlag, Berlin Heidelberg, ch.6
1. Ryndin YA, Alekseev OS, Simonov PA, Likholobov VA (1989) J
Mol Catal 55:109
3
3
(
1
23

Selective Oxidation of Glycerol Catalyzed by Rh/Activated Carbon
431
3
3
3
2. Beck A, Horvath A, Stefler G, Scurrell MS, Guczi L (2009) Top
Catal 52:912
3. van Dam HE, Wisse LJ, Van Bekkum H (1990) Appl Catal
35. Besson M, Gallezot P (2003) Catal Today 81:547
36. Vleeming JH, Kuster BFM, Marin GB (1997) Catal Lett 46:187
37. Peereboom L, Koenigsknecht B, Hunter M, Jackson JE, Miller DJ
(2007) Carbon 45:579
6
1:187
4. Vinke P, van Dam HE, Van Bekkum H (1990) In: Centi G, Trifiro
F (ed) New developments in selective oxidation, vol 55. Elsevier,
Amsterdam
123