Académique Documents
Professionnel Documents
Culture Documents
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
a r t i c l e i n f o a b s t r a c t
Article history: Nano-sized (3 nm) alkaline earth metal oxides supported on carbon nanofibers were prepared by a facile
Received 25 January 2013 impregnation and heat treatment method of the corresponding nitrates. These supported catalysts
Revised 26 March 2013 showed a significant improved activity for the aldol reaction and transesterification compared to the
Accepted 24 April 2013
respective bulk materials which did not show any appreciable activity. A linear correlation was found
Available online 24 May 2013
between the initial reaction rate and the strength of the basic sites as determined by CO2 calorimetry
indicating the importance of the base in the rate determining step of the reactions.
Keywords:
Ó 2013 Elsevier Inc. All rights reserved.
Nano-sized alkaline earth metal oxides
Carbon nanofibers
Solid base catalysts
Base strength
Transesterification
Aldol reaction
0021-9517/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2013.04.019
2 A.M. Frey et al. / Journal of Catalysis 305 (2013) 1–6
position in the TPD was located at increasing temperatures when loading. After impregnation, the samples were equilibrated for
going down the group of alkaline earth metal oxides showing that 12 h followed by drying in stagnant air for 12 h at 120 °C. The sam-
the base strength increased down the group. ples (25–150 lm fraction) were heat treated in 30 ml min 1 N2 at
Calorimetry of CO2 adsorption is able to reveal the strength of 600 °C for 3 h (5 °C min 1) immediately prior to the catalytic test.
base sites. For example, Horiuchi et al. used this technique to
investigate the basicity of ceramics based on c-Al2O3 and MO 2.2. Catalyst testing
[25]. The heat of CO2 adsorption was found to increase with
decreasing electronegativity of the investigated metals in the 2.2.1. Self-condensation of acetone to diacetone alcohol
materials, showing that the base strength increased down a group. The self-condensation of acetone was carried out at 0 °C in a
In this study, we have investigated the influence of base 200-ml double-walled thermostatted batch glass reactor using
strength on catalytic activity for supported alkaline earth metal 100 g of acetone as reactant, 6 g of iso-octane as an internal stan-
oxide containing the same number of active sites per unit mass. dard, and 1 g of catalyst. Prior to the experiment, the reactor was
MO were supported on carbon nanofibers (CNF) staring from the flushed with nitrogen and during the entire experiment a nitrogen
cheap and widely available nitrate precursors. Carbon nanofibers atmosphere was kept over the reaction to avoid exposure to air.
have been used as support to suppress support/metal oxide inter- The reaction mixture was mechanically stirred with 1700 rpm.
actions as often seen for traditional oxide support as discussed Aliquots of 1 ml solution were taken at specific times during a total
above [20]. Bulk MOs prepared from the same precursors were of 24 h. The samples were filtrated through hyflo (Sigma Aldrich) in
included for comparison. After preparation, the materials were order to remove traces of catalyst before GC analysis.
characterized by CO2-chemisorption, CO2-calorimetry, XRD, and
TEM. Biomass-related processes as the transesterification and the
2.2.2. Regeneration test
self-condensation reaction of acetone to diacetone alcohol (DAA)
The first catalyst test was performed as described above. After
as example of an aldol reaction were chosen as test reaction
24 h of test, the catalyst was isolated by centrifugation and decan-
(Scheme 1).
tation and washed with 3 30 ml acetone and 2 30 ml heptane
followed by drying at 120 °C for 18 h in stagnant air. After this
2. Materials and methods procedure, the catalyst was heat treated in 30 ml min 1 N2 at
600 °C for 3 h (5 °C min 1) immediately prior to the second catalyst
2.1. Preparation of the catalysts test. The same procedure was carried out before the third activity
experiment.
2.1.1. Bulk metal oxides The concentration of diacetone alcohol in the aliquots was mea-
The bulk metal oxides were prepared from the nitrate-precur- sured using a Chrompack CP 9001 gas chromatograph equipped
sors, magnesium nitrate hexahydrate (Acros, 99+%), calcium with a CP-SIL-5 column and an FID detector in all types of catalytic
nitrate tetrahydrate (Sigma–Aldrich, 99%), strontium nitrate (Sig- experiments. Traces of by-products were identified with a
ma–Aldrich, p.a.), and barium nitrate (Fluka, p.a.) by heat treat- Shimadzu GCMS-QP2010. Since only traces of by-products were
ment at 700 °C for 2 h (10 °C min 1) in stagnate air. observed, the initial catalytic activity was expressed as
A high surface area MgO (230 m2 g 1) from Stream Chemicals 1 1
mmolacetone converted mmolMO h over the first 15 min of the reaction
was stored under N2 and used without any activation procedure. where mmolacetone converted was defined as 2 mmolDAA formed.
2.1.2. Alkaline earth metal oxide supported on carbon nanofibers 2.2.3. Transesterification
Oxidized carbon nanofibers (CNF) (pore volume 0.7 ml g 1) The transesterification reaction was carried out in N2 atmo-
were prepared from syngas at 550 °C and 3 bars overpressure using sphere at 60 °C in a 100-ml round-bottom flask using 10 g triacetin,
a 5 wt% Ni/SiO2 catalyst as described in detail elsewhere [7]. 2.5 g 11.7 ml methanol, 0.3 ml toluene as internal standard, and 100 mg
CNF was dried overnight at 120 °C, followed by evacuation for half of catalyst while stirring at 500 rpm. Aliquots of 0.2 ml solution
an hour. The desired metal was introduced to the support by incip- were taken at specific times during a total of 3 h and diluted with
ient wetness impregnation using an aqueous solution of the metal 1 ml dicholoromethane. The samples were filtrated through a
nitrate salt. The loadings were 3.4 wt% MgO, 4.8 wt% CaO, 8.8 wt% micropore filter in order to remove traces of catalyst before GC
SrO, and 13 wt% BaO on CNF. These loadings result in equimolar analysis. The triacetin concentration was measured on a Shimadzu
amounts of metal oxide in the catalysts. In case of BaO, three GC 2010 with a CP8822 column equipped with an FID detector, and
1 1
impregnations with barium nitrate were needed to obtain the high the activity was expressed as mmoltriacetin converted mmolMO h .
Scheme 1. (a) Self-condensation of acetone to diacetone alcohol and (b) transesterification of triacetin with methanol.
A.M. Frey et al. / Journal of Catalysis 305 (2013) 1–6 3
2.3. Characterization catalyst was also the same, which is in agreement with the CO2
sorption data.
The N2-physisorption experiments were performed at 196 °C In order to study the base strength of the active sites, micro cal-
on a Micrometrics Tristar Surface Area and Porosity analyzer. The orimetry was applied. The curves showing the differential heat of
measurements were used to determine pore volumes and specific CO2 adsorption on the supported alkaline earth metal oxides are
surface areas of the catalysts using BET theory. displayed in Fig. 3.
A number of base sites were evaluated using volumetric CO2 For the supported MO samples, the initial differential heat of
adsorption (Micromeritics ASAP 2010C) at 0 °C. The samples were CO2 adsorption increased from 105 to 193 kJ mol 1 (0–5 lmol of
dried in situ at 450 °C in vacuum for at least 20 h prior to measure- adsorbed CO2 gcat1 , Table 1), when going from MgO to BaO. For bulk
ments. Prior to vacuum drying, all samples had been heated to materials, a similar trend in the initial heat of adsorption has been
600 °C in nitrogen for 3 h under similar condition as before the reported though in a narrower range, i.e., 120–160 kJ mol 1 [29].
catalytic tests. Adsorption isotherms were obtained by plotting The strongest base sites of SrO/CNF and BaO/CNF (Table 1, Fig. 3)
CO2 uptake versus the absolute pressure. The total number of ba- are comparable in strength to the super basic sites present in bulk
sics sites was determined by extrapolation the linear part of the LaMgO, CsY, and NaN3/NaY [27,30]. In our materials, these strong
isotherm (100–250 mbar) to zero bar pressure. base sites are, however, only found for a limited number of sites
The heat of adsorption of CO2 at 30 °C was measured for the while the majority of the base sites displayed a lower differential
supported samples in a Tian–Calvet calorimeter after a pretreat- heat of CO2 adsorption, 100–130 kJ mol 1, at higher CO2 coverages.
ment at 450 °C in vacuum for 2 h [26,27]. The trend in basicity for these sites is the same as for the strong ba-
Powder X-ray diffraction (XRD) was measured using an Enraf– sic site, i.e., BaO/CNF > SrO/CNF > CaO/CNF MgO/CNF.
Nonius CPS 120 apparatus with Co Ka radiation (k = 1.789 Å). The crystalline phases present in the samples were investigated
The supported samples were examined with transmission by XRD analysis of materials which had been exposed to air during
electron microscopy (TEM) using an FEI Tecnai 12 or in an FEI preparation and XRD measurement. The diffractograms after
Technai 20F. The samples were placed on holy carbon grid, and subtraction of the background are displayed in Fig. 4 for the sup-
both bright field and dark field TEM images were recorded. ported materials.
For the supported samples, the XRD patterns were dominated
3. Results and discussion by diffraction peaks originating from CNF. For MgO/CNF and CaO/
CNF, no additional diffraction peaks were observed. For SrO/CNF
3.1. Characterization and BaO/CNF, diffraction peaks with low intensity were observed
as indicated with marks in the figure. These could be ascribed to
Table 1 displays selected properties of the supported alkaline traces of the respective metal carbonates. Carbonate formation is
earth metal oxides under study. due to reaction between the oxide and CO2 in the atmosphere
All bulk samples had very low specific surface areas, 0–2 m2 g 1 before or during the XRD measurement. Since no reflection repre-
in accordance with literature [1], whereas supported catalysts had senting oxides are found, it is likely that the metal oxides on the
surface areas of about 150 m2 g 1 which is close to that of the sup- carbon nanofibers were amorphous and or too small to be
port. The amount of CO2 adsorbed on the supported catalysts (at detected, while the carbonate of the heavier alkaline earth metal
0 °C) was similar for all samples (165 15 lmol CO2 gcat1 ) which oxides are more crystalline. For carbon nanofiber samples, an
shows that the number of accessible base sites per gram of catalyst in situ XRD study was performed following the decomposition of
is comparable. the nitrates and the formation of the oxides. Due to the partial
TEM was used to investigate the particle size of the supported amorphous nature of the oxides (Fig. 4), higher loadings of the
MOs (Fig. 1). alkaline earth metal were in order to increase the likelihood of
Fig. 1a–d displays bright field images in which next to the oxide observing the phases. The results are found in the Supplementary
(white arrows) only short fibers are visible. A more comprehensive material S1. For the sample with higher loadings, it was found that
TEM study (not shown) indicated that also longer fibers 0.5–1 lm exclusively oxide (Mg, Ca, and Sr) or amorphous phase (in the case
are present. The metal oxide particles are more visible in dark field Ba) is present after the heat treatment. No carbonate is thus
mode as shown in Fig. 1e. observed after the activation procedure if the samples are not ex-
Fig. 2 displays the particle size distribution in the different cat- posed to the air. It is, however, interesting to note that the forma-
alyst base on TEM analysis. The average particle size of the oxides tion of the oxides on the CNF goes via a carbonate route for Ca, Sr,
was 2.7 nm for MgO, 3.0 nm for CaO, 3.1 nm for SrO, and 3.6 nm for and Ba.
BaO with a narrow size distribution as shown in the figure.
The increase from 2.7 nm for MgO to 3.6 nm for BaO correlates 3.2. Catalytic activity
with the increase in unit cell size of 30% when going from MgO to
BaO [28]. Therefore, all samples contain the same number of The catalytic activities of the bulk and supported materials in
surface atoms per oxide particle. As the molar loading of the oxides the self-condensation of acetone to DAA are shown in Fig. 5. The
was the same, the number of accessible base sites per gram of thermo dynamical calculated equilibrium concentration of DAA
Table 1
Loading, specific surface area, initial activity, CO2 uptake, and initial heat of adsorption for the investigated samples.
a b
MgO/CNF 138 105 /107 180 16 180
CaO/CNF 143 120a/106b 154 62 211
SrO/CNF 166 162a/121b 174 122 313
BaO/CNF 141 193a/132b 168 232 458
a
Initial heat of CO2 adsorption (0–5 lmol/gcat).
b
Average heat of CO2 adsorption (10–50 lmol/gcat).
c 1
Initial rate is calculated as mmol converted reactant molecules per mmol MO in the sample per h , evaluated over the first 15 min reaction time.
4 A.M. Frey et al. / Journal of Catalysis 305 (2013) 1–6
Fig. 1. TEM images of (a) BaO/CNF, (b) SrO/CNF, (c) CaO/CNF, (d) MgO/CNF, and (e) HAADF of MgO/CNF.
Fig. 2. Particles size distribution and average particle size measured in TEM (a) MgO/CNF, (b) CaO/CNF, and (c) SrO/CNF (d) BaO/CNF.
A.M. Frey et al. / Journal of Catalysis 305 (2013) 1–6 5
Fig. 6. Initial rate versus initial heat of CO2 adsorption for the strongest base sites
for the alkaline earth metal oxides supported on CNF.
Fig. 3. Differential heat of CO2 adsorption for MgO/CNF, CaO/CNF, SrO/CNF, and
BaO/CNF versus the accumulated amount of CO2 adsorbed on the material.
Fig. 8. Use and reuse of high surface area bulk MgO. The sample was washed with Supplementary data associated with this article can be found, in
acetone/heptane and heat-treated in N2 at 600 °C for 3 h (5 °C min 1) between the the online version, at http://dx.doi.org/10.1016/j.jcat.2013.04.019.
tests.
References
Fig. 7 shows that the formation rate of DAA is decreased 7% [1] A. Corma, S. Iborra, Adv. Catal. 49 (2006) 239.
(based on initial rate) from the first to the second test and 30% [2] A. Kawashima, K. Matsubara, K. Honda, Bioresour. Technol. 99 (2008) 3439.
from the second to the third test. This loss in activity is possibly [3] X. Liu, H. He, Y. Wang, S. Zhu, X. Piao, Fuel 87 (2008) 216.
[4] Y. Ono, J. Catal. 216 (2003) 406.
due to blocking of active sites by larger condensation products. [5] H. Hattori, J. Jap. Pet. Inst. 47 (2) (2004) 67.
The bulk MO prepared from nitrate precursors showed very [6] H. Hattori, Chem. Rev. 95 (1995) 537.
little catalytic activity, most likely due to their low specific surface [7] A.M. Frey, J.H. Bitter, K.P. de Jong, ChemCatChem 3 (2010) 1193.
[8] F. Winter, A.J. van Dillen, K.P. de Jong, Chem. Commun. (2005) 3977.
area (Table 1) as discussed above. A high surface area (230 m2 g 1) [9] K. Tanabe, N. Yoshii, H. Hattori, J. Chem. Soc. Chem. Commun. (1971) 464.
commercial MgO was tested for comparison and displayed an ini- [10] F. Figueras, J. Lopez, J. Sanchez-Valente, T.T.H. Vu, J.M. Clacens, J. Palomeque, J.
tial activity which is higher than the CNF-supported MgO (28 ver- Catal. 211 (2002) 144.
[11] M.J. Climent, A. Corma, S. Iborra, J. Primo, J. Catal. 151 (1995) 60.
sus 16 h 1 respectively). The number of base sites (surface area) is
[12] M.T. Drexler, M.D. Amiridis, Catal. Lett. 79 (2002) 175.
thus found to be crucial for the catalytic performance. However, [13] A. Corma, S. Iborra, J. Primo, F. Rey, Appl. Catal. A Gen. 114 (1994) 215.
the high surface area unsupported material deactivated signifi- [14] H. Kabashima, H. Tsuji, H. Hattori, Appl. Catal. A Gen. 165 (1997) 319.
cantly, and after 5 h, no increase in DAA formation was measured [15] T. Seki, H. Kabashima, K. Akutsu, H. Tachikawa, H. Hattori, J. Catal. 204 (2001)
393.
as shown in Fig. 8 (MgO test 1). [16] C. Noda, G.P. Alt, R.M. Werneck, C.A. Henriques, J.L.F. Monteiro, Braz. J. Chem.
Even after regeneration (i.e., washing with acetone/heptane fol- Eng. 15 (1998) 120.
lowed by heat treatment in N2 at 600 °C), almost no activity was [17] D.-W. Lee, Y.-M. Park, K.-Y. Lee, Catal. Surv. Asia 13 (2009) 63.
[18] G. Busca, Chem. Rev 110 (2010) 2217.
observed (initial rate 0.1 h 1). In contrast, only a minor loss in [19] G. Busca, Ind. Eng. Chem. Res. 48 (2009) 6486.
activity was observed in recycle tests of our CNF-supported MO [20] M. Haneda, G. Tsuboi, Y. Nagao, Y. Kintaichi, H. Hamada, Catal. Lett. 97 (2004)
particles (Fig. 7) as discussed above. By using a support small MO 145.
[21] S. Vasefir, M. Parvari, Kor. J. Chem. Eng. 27 (2010) 422.
particles with many active sites can thus be stabilized compared [22] A. Auroux, A. Gervasini, J. Phys. Chem. 94 (1990) 6371.
to using bulk materials, where sintering occurs easily. On the oxi- [23] K. Pokrovski, T.P. Jung, A.T. Bell, Langmuir 17 (2001) 4297.
dized carbon nanofibers, the oxygen containing groups are [24] Y.-W. Chen, H.-Y. Chen, W.-F. Lin, React. Kinet. Catal. Lett. 65 (1998) 83.
[25] T. Horiuchi, H. Hidaka, T. Fukui, Y. Kubo, M. Horio, K. Suzuki, T. Mori, Appl.
believed to be anchoring the nano-sized metal oxide particles lim- Catal. A 167 (1998) 195.
iting particle growth both during the catalytic test in liquid phase [26] P.C. Gravelle, Adv. Catal. 22 (1972) 191.
and by regeneration by heat treatment. [27] A. Desmartin-Chomel, B. Hamad, J. Palomeque, N. Essayem, G. Bergeret, F.
Figueras, Catal. Today 152 (2010) 110.
[28] X.-Y. Liu, B.P. Uberuaga, K.E. Sickafus, J. Phys.: Condens. Matter 21 (2009)
4. Conclusion 045403.
[29] V. Solinas, I. Ferino, Catal. Today 41 (1998) 179.
[30] R.J. Davis, Res. Chem. Intermed. 26 (2000) 21.
We prepared 3 nm alkaline earth metal oxide particles on [31] G.G. Podrebarac, F.T.T. Ng, G.L. Rempel, Chem. Eng. Sci. 52 (1997) 17.
carbon nanofibers containing the same number of active sites per [32] G. Zhang, H. Hattori, K. Tanabe, Appl. Catal. 40 (1988) 183.
gram catalyst. This allowed a study of the independent influence [33] Y. Xiao, L. Gao, G. Xiao, J. Lv, Energy Fuel 24 (2010) 5829.
[34] T.F. Dossin, M.F. Reyniers, G.B. Marin, Appl. Catal. B 62 (2006) 35.
of base strength on catalytic performance for the nano-sized