Gold as active phase of BN-supported catalysts for lactose oxidation

Article abstract of DOI:10.1016/j.apcata.2015.01.009

Au/h-BN catalysts have been prepared by wet impregnation in order to combine the high activity of gold with the promising h-BN support to optimize the catalytic performances in lactose oxidation. After 1 h reaction, the catalysts were more active than all the Pd/h-BN catalysts described in previous works: 100% yield was reached after 2 h for a Au/h-BN catalyst prepared in water and containing only 1 wt.% of gold. The influence of α-Al₂O₃, γ-Al₂O₃ and C_black as supports for Au was compared to h-BN and Au/α-Al₂O₃ was the most active. All of them exhibited 100% selectivity toward lactobionic acid. After a second run, Au/γ-Al₂O₃ presented a loss of selectivity and all were less active than during their first run. Au/h-BN and Au/α-Al₂O₃ have been regenerated and a thermal treatment permits to keep the catalysts active with 100% selectivity. Au/h-BN was the most active after regeneration thanks to the more facile poison removal from its surface and the high stability of boron nitride.

Full text of DOI:10.1016/j.apcata.2015.01.009

G Model
APCATA-15201; No. of Pages10
ARTICLE IN PRESS
Applied Catalysis A: General xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Gold as active phase of BN-supported catalysts for lactose oxidation
∗
Nathalie Meyer, Coralie Renders, Raphaël Lanckman, Michel Devillers, Sophie Hermans
Université catholique de Louvain, Institut de la Matière Condensée et des Nanosciences (IMCN), Place Louis Pasteur 1/3, B-1348 Louvain-la-Neuve, Belgium
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 15 September 2014
Received in revised form
Au/h-BN catalysts have been prepared by wet impregnation in order to combine the high activity of gold
with the promising h-BN support to optimize the catalytic performances in lactose oxidation. After 1 h
reaction, the catalysts were more active than all the Pd/h-BN catalysts described in previous works: 100%
yield was reached after 2 h for a Au/h-BN catalyst prepared in water and containing only 1 wt.% of gold.
The influence of ␣-Al2O3, ␥-Al2O3 and Cblack as supports for Au was compared to h-BN and Au/␣-Al2O3
was the most active. All of them exhibited 100% selectivity toward lactobionic acid. After a second run,
Au/␥-Al2O3 presented a loss of selectivity and all were less active than during their first run. Au/h-BN
and Au/␣-Al2O3 have been regenerated and a thermal treatment permits to keep the catalysts active
with 100% selectivity. Au/h-BN was the most active after regeneration thanks to the more facile poison
removal from its surface and the high stability of boron nitride.
2
9 December 2014
Accepted 5 January 2015
Available online xxx
Keywords:
Gold
Boron nitride
Lactose
Regeneration
Selectivity
© 2015 Elsevier B.V. All rights reserved.
1
. Introduction
Carbohydrates represent an important source of renewable raw
nanoparticles on alumina, titania or ceria allowed obtaining mod-
erate to high conversions and selectivities [13,14,18,22]. Many
studies have been carried out with Au supported catalysts in gen-
eral and they have all demonstrated the promising future of this
metal in heterogeneous catalysis [23–34] which was related to
quantum size effects within very small gold particles [34–37].
Moreover, gold presents in general a high resistance to deactiva-
tion, especially by oxygen poisoning, which makes it a more stable
active phase than other noble metals such as palladium or plat-
inum. For example, Mirescu and Prusse studied the influence of
materials because of their large availability, high framework com-
plexity and independency from fossil energies. In this work, we
focus on lactose, which is an important by-product of the dairy
industry, and whose production is estimated at about 1.2 million
tons per year [1]. Because of its low solubility and sweetness, as
well as certain intolerance of many people in the world, it presents
a limited usage. However, with this surplus and low cost, there
is an interest in the research of innovative processes for trans-
forming lactose for applications in the food and pharmaceutical
industries [2–4]. The valuable products that can be obtained from
it are lactulose, lactitol, lactobionic acid, lactosucrose, and galacto-
oligosaccharides [2]. A recently published review focuses on the
main characteristics, uses and physiological effects of lactobionic
acid (LBA) which has high potential applications as a bioactive
molecule [5]. The main commercial utilization of LBA derives from
its antioxidant, chelating and emulsifying properties [5–11]. It is
an antioxidant compound which can replace formol as an organ
preservation liquid with the advantage of being less toxic [8,12].
In the literature, the heterogeneous catalysts reported to carry
out lactose oxidation are mainly based on noble metals (Au,
several parameters on the activity of Au/Al O3 and Au/TiO2 cata-
2
◦
lysts for lactose oxidation [13]. They have shown that at 80 C, the
activity of the catalysts was much lower than at 40 C because of
◦
the lower solubility of oxygen and that Au/Al O3 were deactivated
2
in acidic medium as well as at strong alkaline pH. The recycling was
also studied with Au/Al O3 and a loss of activity was observed but
2
the catalysts were still able to carry out the reaction with 99% of
selectivity in LBA. In a recent study, Au/mesoporous silica catalysts
have been used and exhibited excellent performances by reaching
100% selectivity and 100% conversion after 100 min in lactose oxi-
dation [2,3]. In our previous works, hexagonal boron nitride (h-BN)
has been evaluated as catalyst support and it was shown that it is
an appropriate candidate to obtain high selectivity in the oxida-
tion of lactose [38]. The best Pd/h-BN catalysts presented a particle
size distribution comprised between 3 and 15 nm with no parti-
cles smaller than 3 nm [39]. In the present work, we will try to
combine the high catalytic activity of gold with the promising sup-
port which is boron nitride, in order to obtain catalysts presenting
optimized activities and selectivites in the oxidation of lactose to
Pd, Bi-Pd, Pt) supported on Al O , SiO , TiO , CeO , carbon or
zeolites [13–21]. Few studies have reported that supporting Au
2
3
2
2
2
∗ Corresponding author. Tel.: +32 10 472810; fax: +32 10 472330.
E-mail address: sophie.hermans@uclouvain.be (S. Hermans).
http://dx.doi.org/10.1016/j.apcata.2015.01.009
0
926-860X/© 2015 Elsevier B.V. All rights reserved.
Please cite this article in press as: N. Meyer, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.01.009

G Model
APCATA-15201; No. of Pages10
ARTICLE IN PRESS
N. Meyer et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
2
lactobionic acid. In addition to the evaluation of Au/h-BN catalysts,
we will compare boron nitride with three other supports, namely
by filtration and the filtrate was analyzed by HPLC. Because selec-
tivity was 100% in all cases, the lactose conversion always equals
the yield in lactobionic acid.
␣
-Al O , ␥-Al O and Cblack. The recycling of Au-based catalysts
2 3 2 3
will be also investigated and the regeneration of some catalysts by
two methods will be described.
2.5. Physico-chemical characterization techniques
2
. Experimental
The catalysts were characterized by X-ray photoelectron spec-
troscopy (XPS), transmission electron microscopy (TEM), X-ray
powder diffraction (XRD) and CO chemisorption.
2.1. Materials
X-ray photoelectron spectroscopy (XPS) was performed on a
SSI-X-probe (SSX-100/206) spectrometer from Fisons. The sam-
ples were pressed on little stainless steel cell-cups and then placed
on an insulating home-made ceramic carousel with a nickel grid
positioned 3 mm above the sample surface, to avoid differential
charging effects: a floodgun set at 8 eV was used for charge stabi-
lization. The C1s binding energy of carbon (C C, H) set at 284.8 eV
was used as internal standard value. Data treatment was performed
with the CasaXPS program (Casa software Ltd.). The analytical peaks
used were C1s, Au4f, Na1s, Cl2p, O1s, N1s, Al2p and B1s. The following
constraints were used for photopeak decomposition in the case of
Au: intensity ratios: I(Au 4f7/2)/I(Au 4f5/2) = 1.33, FWHM ratio = 1,
ꢀ(Au 4f5/2 − Au 4f7/2) = 3.67 eV.
The selected supports were a commercial alumina (␥-Al O ,
2
3
2
−1
Sigma–Aldrich, SBET ≈ 173 m g ), a commercial hexagonal boron
2
−1
nitride (h-BN, Sigma–Aldrich, SBET ≈ 23 m g ) and a commercial
2
−1
carbon black (C
, Timcal, SBET ≈ 63 m g ). A batch of the alu-
black
◦
mina support was calcined in air at 1000 C for 24 h to give ␣-Al O
2
3
2
−1
(
SBET ≈ 16 m g ). The gold precursor was gold (III) chroride trihy-
drate (HAuCl ·3H O, Sigma–Aldrich, 99.9%) in all cases.
4
2
2
.2. Wet impregnation synthesis (WI)
In a typical experiment, 2.5 g of support was suspended in 50 mL
of solvent and stirred for 15 min. The gold precursor, HAuCl ·3H O,
4
2
was dissolved in 25 mL of the same solvent and added dropwise
within 30 min to the support suspension. Then, the suspension was
TEM images were obtained with a LEO 922 omega energy filter
transmission electron microscope operating at 200 kV. The sam-
ples were suspended in hexane under ultrasonic treatment, and
then allowed to settle to discard the biggest particles. A drop of the
supernatant was deposited on a holey carbon film supported on
a copper grid, which was dried overnight under vacuum at room
temperature, before introduction in the microscope. Histograms
of particles size distribution were determined by measuring 100
particles for each catalyst.
XRD analyses were performed with a D5000 Siemens diffrac-
tometer equipped with a copper source (ꢁK␣ = 154.18 pm). The
samples were supported on quartz monocrystals. The crystalline
phases detected were identified by reference to the JCPDS-ICDD
database. The Pd particles sizes were extracted from X-ray diffrac-
tion measurements by using the Debye–Scherrer equation [40]:
stirred for a further 15 min. The catalyst was chemically reduced at
◦
8
0 C by addition of 6 mL formalin (formaldehyde 37 wt.% aqueous
solution, Sigma–Aldrich) or NaBH previously dissolved in the same
4
solvent (0.1 g/mL). Finally, the catalyst was separated by filtration,
washed with the corresponding solvent and dried overnight. The
filtrates containing potentially non-adsorbed Au were analyzed by
atomic absorption spectrometry. The solvents used in this prepa-
ration method were deionized water, acetonitrile (Riedel-de-Haën,
9.5%) or a mixture 50:50 v/v water/methanol (VWR, 99.9%).
The reproducibility of the synthesis method was checked by
preparing two catalysts (Au WI-2) in the same manner and these
were shown to present the same characteristics in terms of metallic
percentages at the support surface as well as Au/B ratio determined
by XPS.
9
kꢁ
B =
2.3. Regeneration of the catalysts
s cos ꢂ
2
.3.1. Chemical regeneration
The used catalyst was dispersed in 50 mL of water and the sus-
where k is a constant taken as 1.00 here, ꢁ is the wavelength of
X-ray radiation (CuK␣ = 0.1541 nm), s is the crystalline size, ꢂ is the
diffraction angle, B is the line width at half maximum height.
The quantification of Au in the synthesis filtrates and the cat-
alytic reaction mixtures was performed by atomic absorption
analysis, using a Perkin-Elmer 3110 spectrometer equipped with
an air-acetylene flame atomizer. The calibration curve (from 1 to
15 mg/L Au) was realized with standard solutions obtained by dilu-
tion of a commercial gold (1 g/mL, Acros) solution.
pension was stirred for 15 min. A solution of NaBH4 0.1 g/mL was
added and the suspension was stirred for 1 h. The catalyst was then
filtered, washed and dried overnight.
2
.3.2. Thermal regeneration
The used catalyst was calcinated in air flow (500 cm /h) at 500 C
3
◦
for 10 h. The catalyst was then reduced in a tubular oven for 2 h at
2
◦
00 C under H (5%)/Ar(95%) flow.
HPLC analysis was performed on a Waters system equipped with
a refractive index 2414 detector and a UV 2998 photodiode array
detector. To quantify the lactobionic acid, an Aminex BioRad HPX-
2
2.4. Oxidation of lactose
8
7C column was used, with CaSO4 1.2 mmol/L as eluent, a flow of
◦
All catalytic tests were performed in a thermostatized double-
0.8 mL/min, a column temperature of 80 C and 10 ␮L of injected
volume. To quantify the remaining lactose and the potential lac-
tulose formed (lactose’s isomer), a Carbohydrate Transgenomic
CarboSep CHO682 column was used, with mQ H2O (18 Mꢃ at 25 C)
as eluent, a flow of 0.4 mL/min, a column temperature of 80 C and
walled glass reactor. The pH of the lactose solution was measured
continuously by a combined AgCl/Ag Beckman electrode calibrated
with buffer solutions of pH 7 and 10 (Fluka). An automatic titra-
tion device Metrohm 842 Titrando was used to neutralize the
acids formed over time with KOH (Riedel-de-Haën, ≥85%). Constant
stirring was ensured by a mechanical stirrer (Heidolph RZR 2051
electronic). Oxygen was introduced into the solution at a constant
flow rate. The detailed experimental conditions used for lactose
oxidation are given in Supplementary information (S1). Aliquots
were taken each hour in order to follow the reaction by HPLC. The
experiments were stopped after 1 or 4 h. The catalyst was recovered
◦
◦
10 ␮L of injected volume.
2.6. Calculation of the specific activity
In order to make comparisons with the literature [13]
the specific activity was calculated for several catalysts, as
follows:Specific activity =
substrate
time×metal
−1 -1
(mmol min
g )where
Please cite this article in press as: N. Meyer, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.01.009

G Model
APCATA-15201; No. of Pages10
ARTICLE IN PRESS
N. Meyer et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
3
Table 1
Au/h-BN catalysts prepared by wet impregnation.
Code
Impregnation solvent
Au (wt.%) engaged
Auc (wt.%) (AAS)
Particle size (XRD)d
Au at.% (XPS)
Au/B × 100 (XPS)
Au WI-1a
Au WI-2a
H2O
H2O
CH3CN
CH3CN
CH3CN
CH3OH–H2O
CH3OH–H2O
4.92
1.02
5.07
3.08
1.00
4.95
0.99
4.9
1.0
5.0
3.1
1.0
3.0
0.99
25
34
18
17
12
47
38
0.14
0.06
0.52
0.38
0.10
0.017
0.0090
0.28
0.15
1.1
0.84
0.21
0.04
0.02
a
Au WI-3
Au WI-4a
Au WI-5a
b
Au WI-6
Au WI-7b
a
Reducing agent = NaBH4.
Reducing agent = formaldehyde.
Au loading determined by difference after atomic absorption analysis of the synthesis filtrates.
Average particle size determined by XRD using Scherrer equation.
b
c
d
Table 2
substrate = amount of converted substrate at 10% conver-
sion, time = time when 10% conversion was reached (in min),
metal = amount of the active metal in the catalyst.
Time to convert lactose entirely with Au/h-BN catalysts.
Code
Au loading
wt.%)
XLactose (%)
after 4 h
Time for complete
conversion
(
Au WI-1
Au WI-2
Au WI-3
Au WI-4
Au WI-5
Au WI-6
Au WI-7
4.9
1.0
5.0
3.1
1.0
3.0
0.99
100
100
100
100
100
0
2h36
2h
1h28
1h23
2h43
n.d.
3
. Results and discussion
Four supports were selected in the present study (h-BN,
␣
-Al O , ␥-Al O and Cblack) and their textural properties are pre-
2 3 2 3
sented in Supplementary information (S2). The nitrogen isotherms
are shown in Fig. S3. h-BN and ␣-Al O can be considered as non-
0
n.d.
2
3
porous, with intergranular macropores, while Cblack presents some
micropores. This latter was chosen because it has a surface area
analyses do not give the same particle sizes than TEM but this is
due to the fact that an average size is obtained by XRD and more-
over the smallest particles are not detected by XRD because they are
amorphous. These images permit to explain the weak percentage
of gold detected for Au WI-6 and Au WI-7 by XPS, which analyzes
only 5–10 nm thickness.
close to h-BN. Only ␥-Al O3 develops some surface area due to the
2
presence of pores. The major difference between the supports is
the amount of oxygenated groups on their surfaces, following the
decreasing order: ␥-Al O > C
> ␣-Al O > h-BN.
2
3
black
2 3
Blank tests have been carried out with the supports alone and no
lactose conversion was observed. The h-BN and ␣-Al O3 supports
2
All these catalysts have been tested in the oxidation of lactose
into lactobionic acid (LBA) for 4 h. At this time, except for the two
catalysts prepared in water/methanol, they all reach 100% yield
with 100% selectivity. However, the time to convert fully lactose
differs from one catalyst to the other (Table 2). The curves showing
the evolution of lactose conversion over time are presented in Sup-
plementary information (S6). Au WI-2 contains only 1% of Au and
converts lactose as fast as Au-WI-1 loaded with 5% of metal. These
results show the high activity of Au/h-BN catalysts, much higher
than the best Pd/h-BN reported previously [38,39] and prepared
using the same method. We can point out that Au(1 wt.%)/h-BN
catalysts are, after 1 h, even more active than their homologues
Pd(1 and 5 wt.%)/h-BN after 4 h [39]. As expected, the catalysts pre-
pared in water/methanol mixture were not able to carry out the
transformation.
alone before and after tests have been analyzed by XPS (see Table
S4). No major change was observed.
3.1. Au/h-BN catalysts prepared by wet impregnation (WI)
The catalysts prepared by WI together with the synthesis sol-
vents used as well as the metal loading determined by atomic
absorption spectrometry are summarized in Table 1. When water or
acetonitrile were used as impregnation solvents, all the metal was
deposited on the h-BN support. For the catalyst prepared in water-
methanol (Au WI-6) with 5 wt.% of Au, it can be observed that the
amount of gold not deposited on the support is non-negligible.
Table 1 presents also the average metallic particle size deter-
mined by XRD using the Scherrer equation and the amount of gold
present at the catalyst surface as analyzed by XPS. No surface poi-
soning by chloride was detected by XPS. The lowest average particle
sizes have been observed for the samples prepared in acetoni-
trile while the highest are present when using water-methanol.
The smallest particles are obtained when the most hydrophobic
solvent is used. Indeed, the hydrophobicity increases following ace-
tonitrile > water–methanol > water [41,42]. Au WI-6 and Au WI-7
present a very small Au at.% at the boron nitride surface, in relation
with their low loadings.
At the end of the catalytic tests, the filtrates were analyzed by
atomic absorption spectroscopy and we observed weak Au losses
for the catalysts prepared in acetonitrile only (4.08%, 7.21% and
6.19% of the total Au amount engaged for Au WI-3, Au WI-4 and
Au WI-5, respectively). For this reason, the next steps will be inves-
tigated with the catalysts prepared in water only.
3.2. Diffusional limitations
All the gold XPS peaks are well aligned and correspond to the
gold metallic phase (S5). A difference in intensity is nevertheless
observed. This could be because a softer reducer would take more
Before studying several parameters that could influence the cat-
alytic results with Au/h-BN materials, the absence of diffusional
limitation had to be checked given their very high activities. Boron
nitride is a “non-porous”/macroporous support and should not
exhibit internal diffusion limitations referring to the reactant dif-
fusion throughout the porous network. In order to verify this, the
catalyst can be ground and the catalytic activity should not change.
Moreover, when the reactants flow is too low, the observed rate
could also be controlled by external diffusion and therefore not
represent the actual rate of the catalyzed reaction [42–44]. This
0
time to produce Au , favoring growth of the gold particles, while
NaBH4 will nucleate more rapidly the particles at the support sur-
face giving smaller particles for a same amount of metal engaged.
Fig. 1 presents respectively TEM images of Au WI-2 and Au WI-
7
. The catalyst prepared in water-methanol mixture exhibits some
huge aggregates with a size of 200 nm (Fig. 1(b)) while the metal-
lic particles in Au WI-2 are of the order of 10 nm (Fig. 1(a)). XRD
Please cite this article in press as: N. Meyer, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.01.009

G Model
APCATA-15201; No. of Pages10
ARTICLE IN PRESS
N. Meyer et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
4
Fig. 1. TEM images of (a) Au WI-2 prepared in water (Au 1 wt.%/h-BN) and (b) Au WI-7 prepared in water-methanol (Au 1 wt.%/h-BN).
Table 3
phenomenon can be detected by increasing the stirring rate. The
last potential problem is the total occupancy of the active sites by
Gold catalysts prepared by WI on different supports.
Code
Support
Au loadinga
wt.%)
Time at 100%
conversion (min)
YLBA after 1 h
reactants and the rate limiting factor is, in this case, the surface
reactivity. By decreasing the catalyst amount, the number of active
sites decreases as well leading in principle, to a lower reaction rate
in the same proportions. The control experiments for diffusional
limitations have been carried out with the Au WI-2 catalyst syn-
thesized in water and containing 1 wt.% Au. Six catalytic tests were
followed over time by taking out samples of the reaction mixture
every 30 min, for 4 h and analyzing them by HPLC (S7).
Lactobionic acid selectivity was 100% in each case. The yields
in lactobionic acid at 1000 or 1500 rpm stirring rate were similar
whatever the amount of catalyst engaged. Grinding had no effect
on the catalytic performances and when the amount of catalyst
was halved, the lactobionic acid yield decreased proportionally. All
these observations confirm that there is no diffusional limitation.
b
reactionc (%)
(
Au WI-2
Au WI-8
Au WI-9
Au WI-10
h-BN
1.0
1.0
1.0
1.0
120
93
106
162
68
87
81
53
␣-Al2O3
␥-Al2O3
Cblack
a
Au loading determined by difference after atomic absorption analysis of the
synthesis filtrates.
b
Determined by automatic titration.
Determined by HPLC.
c
␥-Al O . Table 3 presents catalysts prepared on the four supports
2
3
synthesized by the same wet impregnation method.
The curves presenting the evolution of activity for 4 h are shown
in SI (S9). ␣-Al O3 permits to obtain the most rapidly total conver-
2
sion of lactose. ␥-Al O3 allows also a high conversion of lactose
2
3
.3. Influence of the support nature
followed by h-BN while Cblack gives lower results. In order to make
comparisons, the yield after 1 h reaction was determined by HPLC
(Table 3). The same trend was observed. In all cases, 100% selectiv-
ity was obtained whatever lactose conversion reached, therefore
YLBA equals the conversion. The catalytic filtrates have been ana-
lyzed by atomic absorption spectroscopy and no Au losses were
observed during the oxidation reaction. The four catalysts were
characterized by TEM and XPS before and after 1 h reaction. Table 4
shows the XPS surface atomic percentage and Au/B,C or Al atomic
ratios. The first observation is that the higher the surface atomic
Au percentage, the higher the catalytic activity. Except for ␥-
It has been described in the literature that the support can have
an important effect on the activity. For example, Murzin et al. stud-
ied lactose oxidation with several supports, mainly Fe O and Al O
and they showed that Au(2 wt.%)/Al O3 was the most active with a
3
4
2
3
2
LBA selectivity of 95% at 20% lactose conversion. Au/Fe O catalyst
3
4
was shown to be more prone to deactivation because of oxygen
coverage on the surface [18]. Au/h-BN supported catalysts are very
active in the selective oxidation of lactose, therefore ␣-Al O , ␥-
2
3
Al O and Cblack have been selected for comparison with boron
2
3
nitride given their similar textural properties in the case of ␣-Al O3
Al O where an obvious decrease can be observed, the XPS results
2 3
2
and Cblack, and the presence of more surface hydroxyls in the case of
do not vary much after a catalytic test. TEM images of Au WI-2
Please cite this article in press as: N. Meyer, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.01.009

G Model
APCATA-15201; No. of Pages10
ARTICLE IN PRESS
N. Meyer et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
5
Fig. 2. TEM images of Au WI-2 after a catalytic test.
Table 4
(Au/h-BN) show a relatively homogeneous distribution of the par-
XPS results of Au(1 wt.%) catalysts supported on different supports before use, after
ticles on the support as described previously (Fig. 1(a)), before but
also after the catalytic test (Fig. 2). The particle size distribution
was also established by measuring around 100 particles (S10). The
histograms before and after catalytic test are similar, confirming
the XPS results. The average size of gold particles is comprised
between 5 and 15 nm, with only very few particles below 3 nm.
Fig. 3 shows TEM images of Au WI-8, prepared on alpha alumina,
before and after a catalytic test. It comes out here that the particle
size distribution is more heterogeneous than in the case of hexago-
nal boron nitride as confirmed by the histograms presented in S11.
Moreover, after the catalyst use, small particles seem to be formed,
one and two uses (t = 1 h).
Code
Support Au at.% (XPS)
Au/(B,C,Al) × 100 (XPS)
BTa
BTa
1st use 2nd use
1st use 2nd use
Au WI-2
Au WI-8
Au WI-9
Au WI-10
h-BN
0.064 0.067
0.65
0.25
0.063
0.90
0.15
2.0
0.15
2.4
0.14
3.0
␣-Al2O3
␥-Al2O3
Cblack
0.67
0.079
0.20
0.74
0.042 0.04
0.29
0.72
0.033
0.041 0.040
0.031
a
Before test.
Fig. 3. TEM images of Au WI-8, (a) before a catalytic test and (b) after a catalytic test.
Please cite this article in press as: N. Meyer, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.01.009

G Model
APCATA-15201; No. of Pages10
ARTICLE IN PRESS
N. Meyer et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
6
leading to an increase in the proportion of particles with a size
comprised between 3 and 5 nm. At some places, bigger agglom-
erates are present while smaller particles can be found at other
locations. TEM images have also been taken for the catalysts pre-
pared on ␥-Al O and Cblack. However, for these two samples, due
2
3
to the lack of contrast between support and metal particles in the
TEM images, the size distribution could not be determined pre-
cisely. In the case of ␥-Al O , a mix of small particles and bigger
2
3
aggregates are present before and after being tested in the oxidation
reaction (S12). S13 shows images of Au WI-10 (Au/Cblack), before
test but the gold particles are not easily visualized. The curves
presenting the evolution of the activity with time (S9) permit to
determine the specific activity of the catalysts. For Au WI-8 sup-
−
1
−1
Fig. 4. 1st and 2nd use of Au(1 wt.%) catalysts supported on several supports in
lactose oxidation after 1 h reaction.
ported on ␣-Al O , the specific activity is 12 mmol min
g
. The
2
3
catalyst supported on boron nitride presents a specific activity of
−1
−1
which is higher than the best Pd/h-BN cata-
7
.5 mmol min
g
The catalysts will be characterized by TEM and XPS in order to see
the impact of several uses.
lyst described in a previous work [38] that displayed an activity of
−1
−1
g . The gold catalysts presented here are compet-
2
.5 mmol min
itive with the best gold catalysts for lactose oxidation found in the
literature. Indeed, Mirescu and Prusse obtained a specific activity
of 18 mmol min
3.4.1. Recycling of 1 wt.% Au catalysts prepared on different
supports
−1
−1
g for a Au/TiO2 catalyst [13].
A loss of 48% of yield for Au/h-BN catalyst after a second use is
observed while 45% of loss for Au/␣-Al O3 and 50% for Au/Cblack
2
appeared. After a second use, the catalyst supported on ␥-Al O3
2
3
.4. Recycling of the catalysts
gives the best yield in LBA with only 3% of loss (Fig. 4). However,
this catalyst suffers from a loss in selectivity because of the for-
mation of 2-keto-lactobionic acid. This might be attributable to the
presence of hydroxyl functions at the support surface as shown in
a previous work [39], where the selectivity was shown to be influ-
enced negatively when the number of surface oxygenated groups
increased. Indeed, the XPS Au/Al ratio decreases after one test sug-
gesting a particle agglomeration. Thus, during the second run, the
In the literature, studies about catalyst recyclability in sugars
oxidation are quite scarce. Even though Au supported catalysts have
been shown to be the best for lactose oxidation, they are very often
deactivated at prolonged reaction times and uses [45,46]. In this
section, the recycling of the catalysts prepared by wet impregnation
on boron nitride and on the three other supports will be studied.
Fig. 5. TEM images of Au WI-2 (Au/h-BN): above and Au WI-8 (Au/␣-Al2O3) after two uses in lactose oxidation: below.
Please cite this article in press as: N. Meyer, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.01.009

G Model
APCATA-15201; No. of Pages10
ARTICLE IN PRESS
N. Meyer et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
Table 5
7
XPS characterization results for Au WI-2 and Au WI-8.
Code
% O
% C
O/B
C/B
h-BN
h-BN blank
Au WI-2
Au WI-2 1st run
Au WI-2 2nd run
3.0
3.6
4.7
5.4
4.1
3.9
3.8
4.8
2.7
4.5
0.063
0.075
0.11
0.12
0.091
0.082
0.080
0.11
0.058
0.10
a
%
O
% C
O/Al
C/Al
␣
␣
-Al2O3
-Al2O3 blank
51
54
55
55
56
14
12
12.
16
13
1.5
1.6
1.7
2.0
1.9
0.41
0.34
0.38
0.57
0.42
a
Au WI-8
Au WI-8 1st run
Au WI-8 2nd run
Fig. 6. Monitoring of the oxidation of lactose into lactobionic acid for 4 h with
Au WI-8 (Au/␣-Al2O3) and Au WI-2 (Au/h-BN) during their 2nd run.
a
Supports alone tested in oxidation reaction.
histograms. XPS Au/Al ratios confirm this observation because it
increases after one run corresponding to higher metallic disper-
sion. The fact that during the second run this catalyst is not as active
is explained by the presence of these particles smaller than 3 nm
which play a detrimental role in this reaction.
OH groups present at the surface become more accessible and
are able to influence the selectivity. We can conclude that the OH
functions are deleterious for the selectivity but beneficial for the
stability. ␣-Al O gives better results than h-BN and C black after a
2
3
second use with still 100% selectivity. No Au losses were observed
after two catalytic tests. XPS characterization data of the catalysts
before use, after one test and after a second test are presented in
Table 4. The catalyst prepared on h-BN does not show any change
in Au at.% or in Au/B atomic ratios over the uses. When Cblack is
used, the surface Au% decreases after the second use which might
correspond to an agglomeration of the particles. On the opposite,
3.4.2.3. Poisoning. Poisoning of the metal surface by strongly
adsorbed species such as impurities, products, by-products or oxy-
gen can occur [47]. In order to determine if a surface anchoring
of sugars appears on the supports without metallic phase, two cat-
alytic tests with the supports alone have been carried out and these
were analyzed before and after tests by XPS. The atomic percent-
with ␣-Al O as support, an increase is observed which could be
2
3
ages of C and O on the surface of h-BN or ␣-Al O3 before and after
due to a re-distribution of the metallic particles.
2
test were similar (Table 5). The possible poisons are molecular oxy-
gen, lactose or lactobionic acid. Molecular oxygen can poison the
active phase which might then be oxidized. The oxidation state of
gold does not vary over the catalyst runs for Au WI-2 (S14) but
seems to be slightly different for Au WI-8 after two runs (S15).
However, Au is usually less sensitive to O2 poisoning than other
metals so this cause can be excluded here.
Carbon and oxygen surface percentages as well as O/(B or Al)
and C/(B or Al) have been determined by XPS and compared before,
after one and after a second run for the Au catalysts. In both cases,
the oxygen % as well as O/B or O/Al ratios increase after the first run
TEM images after two catalytic tests (Fig. 5) have been taken
and the particle size distributions before, after one and two runs
were determined for the catalysts prepared on h-BN and ␣-Al O .
2
3
For Au/h-BN catalyst (Au WI-2), the particle size distribution pro-
file remains unchanged while for Au/␣-Al O , some large particles
2
3
are present after two runs in addition to smaller ones. After the
second run, the proportion of particles inferior to 3 nm decreases
significantly in favor of particles with a size superior to 10 nm.
3.4.2. Identification of deactivation causes of Au/h-BN and
Au/˛-Al O3 catalysts
2
(Table 5 and S16). The XPS O1s peaks are similar before and after test
A monitoring over time was realized during the second run of
these two catalysts for 4 h (Fig. 6). Both catalysts give relatively high
activities after 4 h, with an optimal selectivity toward lactobionic
acid. However, compared to their first run, a non-negligible loss
of activity was observed. Here, the possible causes of deactivation
that can be considered are leaching, sintering or poisoning. We will
discuss them in turn in the following sections.
for both catalysts and correspond probably to adsorbed oxygen. In
the case of Au/␣-Al O3 (Au WI-8), C at.% increases after one test,
2
which may indicate sugar or product poisoning (S16). When the
sugar or LBA is adsorbed on Au particles, hydrogen-bridge bonds
can appear between the OH functions of the support and
O atom
of lactose or the C O of LBA. It was already shown in the literature
that LBA can have an inhibiting effect on lactose conversion by cat-
alyst deactivation [21]. Hence we can conclude that poisoning is
probably the cause of activity losses.
3.4.2.1. Leaching. The cause of deactivation of these two catalysts
is not leaching because no Au has been found in the catalytic test
filtrates.
3.5. Regeneration of the catalysts
3.4.2.2. Sintering. In order to verify if the activity loss during the
second run is due to sintering, the characterization data of the cat-
alysts after one run have to be analyzed. For Au/h-BN, no particle
growth was observed by TEM and the particle size distribution
profile remains unchanged after one test indicating that no sin-
tering occurred. The majority of particles have a size comprised
between 5 and 10 nm. The proportion of particles smaller than 3 nm
remains inferior to 10% which is favorable in lactose oxidation as
described in a previous work [39]. In addition, Au/B ratios and Au
at.% determined by XPS do not vary after one test. Concerning Au/␣-
Al O , the metallic particles are not growing either. However, they
The recycling tests presented in the previous section have
shown that the efficiency of our catalysts decreases significantly
along the successive runs. In order to counteract the loss of effi-
ciency of our catalysts, two methods of regeneration have been
tested for Au/␣-Al O3 and Au/h-BN. These methods aim at either
2
cleaning up the poisoned catalyst surface, re-disperse the active
phase or to reduce gold to the metallic oxidation sate. The first
used method is chemical while the second one is thermal.
Considering that the metallic state is the active phase for the oxi-
dation of lactose, NaBH4 as a reducer was used to “re-activate” this
phase if a thin layer of oxide had formed during the reaction. Indeed,
for Au WI-8, a slight shift in Au binding energy was observed after
2
3
tend to decrease after one test giving a higher proportion of par-
ticles smaller than 3 nm, as observed in particle size distribution
Please cite this article in press as: N. Meyer, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.01.009

G Model
APCATA-15201; No. of Pages10
ARTICLE IN PRESS
N. Meyer et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
8
Fig. 7. Monitoring of lactose oxidation into lactobionic acid for 4 h with Au WI-8
Au/␣-Al2O3) and Au WI-2 (Au/h-BN) thermally regenerated.
Fig. 9. Particle size distribution in Au/␣-Al2O3 catalyst before test, after one run and
after two runs, after regeneration and tested after regeneration.
(
two runs (by XPS) which could correspond to Au+1 species. The
activity after chemical regeneration was followed during 4 h and it
came out that the chemical regeneration has a negative impact on
the catalytic performances because about 30% lactose conversion
after 4 h was obtained (S17) while before this regeneration, almost
better activity than Au/␣-Al O . It should be noted that the calci-
2
3
◦
nation temperature could be optimized because above 200 C gold
◦
is spontaneously reduced, and 500 C might not be enough to burn
organic poisons.
These catalysts have been characterized by XPS after the regen-
eration and after use of the regenerated catalysts for both chemical
and thermal treatments. The atomic Au% and the Au/B or Au/Al
ratios obtained for the two catalysts are presented in Table 6.
Concerning the chemical regeneration, it gives lower Au at.% and
lower Au/B or Al ratios after treatment for both catalysts lead-
ing to a lower dispersion of the metallic phase. The use of these
regenerated catalysts causes a re-dispersion of the metallic parti-
cles given the increased Au% at the catalyst surface and the higher
Au/B, Al obtained after test. Concerning the catalysts thermally
regenerated, the catalyst prepared on ␣-Al O displayed a decrease
1
00% was obtained after 4 h as shown previously in Fig. 6.
The thermal regeneration of the catalysts has also been car-
ried out for the catalysts supported on ␣-Al O3 and h-BN only.
2
Indeed, at high temperature, ␥-Al O begins to be transformed in
2
3
␣
-Al O and Cblack is burnt. Au/h-BN and Au/␣-Al O after two
2
3
2
3
◦
◦
runs have been calcined at 500 C and reduced at 200 C for 2 h
with H (5%)/Ar(95%). These catalysts were monitored for 4 h in lac-
2
tose oxidation (Fig. 7). The selectivity in LBA was 100% for both
catalysts. The treatment applied failed to restore the initial cat-
alyst effectiveness. However, the regenerated Au/h-BN showed a
2
3
Fig. 8. TEM images of Au WI-8: (a) and (b) after thermal regeneration and (c) and (d) Au WI-8 after test of regenerated catalyst.
Please cite this article in press as: N. Meyer, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.01.009

G Model
APCATA-15201; No. of Pages10
ARTICLE IN PRESS
N. Meyer et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
9
Table 6
XPS results of Au WI-2 and Au WI-8 after two runs, regeneration and tested regenerated catalysts.
Code
After two runs
Au at.%
Chemically regenerated
Tested after chemical regeneration
Au/(B,Al) × 100
Au at.%
Au/(B,Al) × 100
Au at.%
Au/(B,Al) × 100
Au WI-8 ␣-Al2O3
0.90
3.0
0.33
1.1
0.53
1.9
Au WI-2 h-BN
0.063
0.14
0.035
0.098
0.066
0.14
Thermally regenerated
Testedafterthermalregeneration
Au WI-8 ␣-Al2O3
0.90
3.0
0.37
1.3
0.31
1.12
Au WI-2 h-BN
0.063
0.14
0.075
0.28
0.041
0.090
Fig. 10. TEM images of Au WI-2: (a) and (b) after thermal regeneration and (c) and (d) after test of regenerated catalyst.
that Au/h-BN is more active after thermal regeneration than Au/␣-
Al O could be due to removal of the poisons of the active sites.
2
3
During the regeneration process, the surface of gold is cleaned more
easily in the case of BN thanks to the absence of functional groups on
the support. But XPS analyses of carbon at.% and C/B or C/Al atomic
ratios did not permit to prove this hypothesis. TEM images of the
thermally regenerated and tested catalysts (Fig. 8) show, in the case
of Au WI-8 supported on ␣-Al O3 that the main difference before
2
and after regeneration is the proportion of particles smaller than
3
nm which are absent for this latter. Particle size distributions of
the catalysts after regeneration and after test allow confirming this
observation (Fig. 9). Indeed, no particle inferior to 3 nm is present
anymore for the regenerated catalyst or after its use in lactose
oxidation while many much larger particles appear. The thermal
Fig. 11. Particle size distribution in Au/h-BN catalysts before test, after one run and
after two runs, after regeneration and tested after regeneration.
regeneration implies an increase in gold particle size with ␣-Al O3
2
as support. Concerning the catalyst prepared on boron nitride, the
thermal regeneration seems to be more beneficial on the parti-
cle distribution which is more homogeneous after regeneration as
shown on the TEM images (Fig. 10). The particle size distributions
for Au WI-2 after regeneration and for the tested regenerated cat-
alyst have been measured and as in the case of Au/␣-Al2O3, the
proportion of particles smaller than 3 nm decreased significantly
in the atomic Au% and of Au/Al ratio after regeneration, proba-
bly due to particles agglomeration. In the case of boron nitride
supported catalyst, an increase was observed pointing toward
re-dispersion. Au XPS spectra of both catalysts, after thermal regen-
eration and use, show the presence of metallic gold (S18). The fact
Please cite this article in press as: N. Meyer, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.01.009

G Model
APCATA-15201; No. of Pages10
ARTICLE IN PRESS
N. Meyer et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
1
0
(Fig. 11). However, the amount of very large particles (>20 nm)
References
did not increase as much. This shows that BN is beneficial for the
regeneration step.
[
1] J. Kowalczyk, U., Prusse, H., Berndt, I. Pitsch, (2007). Method for selective carbo-
hydrate oxidation using supported gold catalysts, US Patent 20070112186A1.
2] J.G. Zadow, J. Dairy Sci. 67 (1984) 2654–2679.
[
4
. Conclusion
[3] L.-F. Gutierrez, S. Hamoudi, K. Belkacemi, Appl. Catal. A: Gen. 402 (2011)
4–103.
9
[
[
4] L.-F. Gutierrez, S. Hamoudi, K. Belkacemi, Int. Dairy J. 26 (2012) 103–111.
5] K.J. Foster, S. Lin, C. Turk, Crit. Care Nurs. Clin. N. Am. 22 (2010) 341–350.
Gold is known for its surprisingly high activity in heterogeneous
catalysis, the objective of the present work was to combine the
high activity of gold with the promising h-BN support in order to
optimize the catalytic performances in lactose oxidation. Au/h-BN
catalysts have been prepared by the wet impregnation method.
Some tests have shown that we were not in diffusional regime. After
h reaction, the prepared Au/h-BN catalysts were more active than
all the Pd/h-BN materials described previously and were even com-
petitive with the best Au catalysts on different supports reported
[6] Y. Paik, K. Lee, H. Han, Yonsei Med. J. 46 (2005) 399–407.
[
7] B. Sharma, P. Sharma, A. Agrawal, S. Sarin, Gastroenterology 137 (2009)
85–891.
8] K. Hicks, F. Parrish, Carbohydr. Res. 82 (1980) 393–397.
8
[
[9] M. Kozampel, M. Kurantz, J. Craig, K. Hicks, Biotechnol. Prog. 11 (1995) 592–595.
10] M. Adamczak, D. Charubin, W. Bednarski, Chem. Pap. 6 (2009) 111–116.
11] L. Mayer, B. Kranz, L. Fisher, J. Biotechnol. 145 (2010) 389–393.
12] J.H. Southard, F.O. Belzer, Annu. Rev. Med. 46 (1995) 235–247.
[13] A. Mirescu, U. Prusse, Appl. Catal. B: Environ. 70 (2007) 644–652.
[14] J. Kuusisto, A.V. Tokarev, E.V. Murzina, M.U. Roslund, J.-P. Mikkola, D.Y. Murzin,
T. Salmi, Catal. Today 121 (2007) 92–99.
15] A. Abbadi, K.F. Gotlieb, J.B.M. Meiberg, H. van Bekkum, Appl. Catal. A: Gen. 156
1997) 105–115.
[16] S. Karski, I. Witonka, J. Goluchowska, J. Mol. Catal. A: Chem. 245 (2006)
25–230.
17] A.V. Tokarev, D.Y. Murzin, Microporous Mesoporous Mater. 113 (2008)
22–131.
[18] E.V. Murzina, A.V. Tokarev, K. Kordas, H. Karhu, J.-P. Mikkola, D.Y. Murzin, Catal.
Today 131 (2008) 385–392.
19] L. Prati, M. Rossi, J. Catal. 176 (1998) 552–560.
20] K.B. Kokoh, N. Alonso-Vante, J. Appl. Electrochem. 36 (2006) 147–151.
[21] D.Y. Murzin, E.V. Murzina, A.V. Tokarev, J.-P. Mikkola, Russ. J. Electrochem. 45
2009) 1017–1026.
[
[
[
1
in the literature. The influence of ␣-Al O , ␥-Al O and Cblack as
2
3
2
3
[
supports for Au has been compared to h-BN. It was shown that the
(
catalyst prepared on ␣-Al O3 is the most active and competitive.
2
2
During their first run, they have all demonstrated 100% selectivity
toward LBA. The recycling was also studied with the Au catalysts.
After a 2nd run, a non-negligible loss of activity was observed
except for ␥-Al O . However, the selectivity for this catalyst was
[
1
2
3
[
[
not 100% anymore. The causes of deactivation were studied for
Au/␣-Al O and Au/h-BN catalysts. The XPS characterization anal-
2
3
(
yses have shown that a poisoning of the metallic phase takes place
during the catalytic reaction because the at.% of O as well as O/B or
O/Al ratios increased after one run. The poisoning is more conse-
[
22] A.V. Tokarev, E.V. Murzina, J.P. Mikkola, J. Kuusisto, L.M. Kustov, D.Y. Murzin,
Chem. Eng. J. 134 (2007) 153–161.
[23] A. Montilla, M. Castillo, M. Sanz, A. Olano, Food Chem. 90 (2005) 883–890.
24] G.C. Bond, Molecules 17 (2012) 1716–1743.
25] G.C. Bond, Gold Bull. 5 (1972) 11–13.
26] G.J. Hutchings, J. Catal. 96 (1985) 292–295.
[
[
[
quent for Au/␣-Al O regarding the atomic C% that increased also
2
3
after one run. A regeneration step was carried out for Au/␣-Al O3
2
and Au/h-BN in order to recover the effectiveness of the fresh cat-
alysts. Thermal regeneration permits to keep the catalysts active
with 100% selectivity. After thermal regeneration, the catalyst sup-
ported on boron nitride presented the highest activity and the most
homogeneous particle size distribution. This could be explained by
[27] C.H. Christensen, B. Jorgensen, J. Rass-Hansen, K. Egeblad, R. Madsen, S.K. Klit-
gaard, S.M. Hansen, M.R. Hansen, H.C. Andersen, A. Rijsager, Angew. Chem. Int.
Ed. 45 (2006) 4648–4651.
[
[
28] A.S.L. Hashmi, Chem. Rev. 107 (2007) 3180–3211.
29] A. Corma, M.E. Domine, Chem. Commun. (2005) 4042–4044.
[30] N. Lopez, T.V.J. Janssens, B.S. Clausen, Y. Xu, M. Mavrikakis, T. Bligaard, J.K.
Norskov, J. Catal. 223 (2004) 232–235.
the presence of oxygenated functions on ␣-Al O3 which stabilize
2
[
[
[
[
31] C. Baatz, N. Decker, U. Prüsse, J. Catal. 258 (2008) 165–169.
32] L. Prati, G. Martra, Gold Bull. 32 (1999) 96–101.
33] M.S. Chen, D.W. Goodman, Catal. Today 111 (2006) 22–32.
34] S. Schimpf, M. Lucas, C. Mohr, U. Rodemerck, A. Brückner, J. Radnik, H. Hofmeis-
ter, P. Claus, Catal. Today 72 (2002) 63–78.
surface poisons and might hinder cleaning during regeneration.
Acknowledgements
[
35] M. Haruta, Catal. Today 36 (1997) 153–166.
The authors wish to thank the Fonds de la Recherche Scientifique
FRS-FNRS) with the assistance of the Fédération Wallonie-
[36] G.C. Hutchings, Catal. Today 100 (2005) 55–61.
[
[
37] M. Haruta, Nature 437 (2005) 1098–1099.
38] N. Meyer, K. Bekaert, D. Pirson, M. Devillers, S. Hermans, Catal. Commun. 29
(
Bruxelles and the Belgian National Lottery, as well as the Université
catholique de Louvain for funding. This work was also partially
funded by the Belgian State (Belgian Science Policy, IAP Project
(
2012) 170–174.
[39] N. Meyer, D. Pirson, M. Devillers, S. Hermans, Appl. Catal. A: Gen. 467 (2013)
63–473.
40] N. von Weymarn, M. Hujanen, M. Leisola, Process Biochem. 37 (2002)
207–1213.
4
[
◦
INANOMAT N P6/17). We are grateful as well to Jean-Fran c¸ ois Stat-
1
syns, Pierre Eloy and Pascale Lipnik for technical assistance and
useful discussions.
[41] A.I. Vogel, A.R. Tatchell, B.S. Furnis, A.J. Hannaford, P.W.G. Smith, Vogel’s
Textbook of Practical Organic Chemistry, 5th ed., Pearson Education India,
1
989.
42] S.J. Suresh, V. Naik, J. Chem. Phys. 116 (2002) 4212–4220.
[43] U.K. Singh, M.A. Vannice, Appl. Catal. A: Gen. 213 (2001) 1–24.
[
Appendix A. Supplementary data
[
[
[
44] T.K. Sherwood, Pure Appl. Chem. 10 (1965) 595–610.
45] S.W. Weller, Catal. Rev. Sci. Eng. 34 (1992) 227–280.
46] B.T. Kusema, D.Y. Murzin, Catal. Sci. Technol. 3 (2013) 297–307.
Supplementary data associated with this article can be
found, in the online version, at http://dx.doi.org/10.1016/j.apcata.
[47] M. Besson, P. Gallezot, Catal. Today 57 (2000) 127–141.
2
015.01.009.
Please cite this article in press as: N. Meyer, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.01.009