Superbase catalysts from thermally decomposed sodium azide supported on mesoporous γ-alumina

Article abstract of DOI:10.1016/j.cattod.2010.01.002

Mesoporous g-alumina because of its homogeneous pore size distribution, represents a good support for alkali metals. Controlled thermal decomposition of impregnated sodium azide on such support yields a superbasic catalyst for the double bond migration of

Full text of DOI:10.1016/j.cattod.2010.01.002

Catalysis Today 152 (2010) 99–103
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Superbase catalysts from thermally decomposed sodium azide supported on
mesoporous g-alumina
*
Roxana M. Bota , Kristof Houthoofd, Piet J. Grobet, Pierre A. Jacobs
K.U. Leuven, Centre for Surface Science and Catalysis, Kasteelpark Arenberg 23, 3001 Leuven, Belgium
A R T I C L E I N F O
A B S T R A C T
Mesoporous -alumina because of its homogeneous pore size distribution, represents a good support for
Article history:
g
Available online 10 February 2010
alkali metals. Controlled thermal decomposition of impregnated sodium azide on such support yields a
superbasic catalyst for the double bond migration of vinylcyclohexane to ethylidene cyclohexane in
continuous liquid flow operation. After slurry impregnation of the azide in methanol, 23Na MAS NMR
shows the presence of resonance lines corresponding to Na metal particles and sodium oxide on the
support. When dry mixing of the catalyst components is done, only supported sodium oxide is found, in
association with decreased catalytic activity. It is concluded that both species are necessary components
of the superbasic sites required for the isomerization reaction mentioned. In the transesterification of
soybean oil with methanol in a batch reactor, the same differences in activity are encountered. The 23Na
MAS NMR spectrum of the former catalyst remained unchanged after the transesterification reaction.
ß 2010 Elsevier B.V. All rights reserved.
Keywords:
Base catalysis
Sodium azide
Mesoporous g-alumina
Double bond shift
Transesterification
Biodiesel
1. Introduction
catalysts, which is often very difficult using physico-chemical
techniques.
The first successful preparation of mesoporous silica, viz. MCM-
41 [1,2], was at the basis of a whole pleiade of catalyst supports
with high surface area, uniform pore-size distribution in the
mesoporous region and high thermostability. Given the potential
The double bond isomerization of VCH is considered to be
initiated by abstraction of an allylic H by a strongly basic site of the
catalyst [7–11]. The allylic H of VCH is attached to a tertiary carbon
and is more difficult to abstract in the form of H+ than the H atoms
attached to a secondary or primary carbon atom. Therefore, it is
suggested that superbasic sites are required for VCH double bond
isomerization (Scheme 1).
There have been attempts to prepare superbasic catalysts by
addition of both alkali hydroxides and alkali metals to alumina
supports [12]. Such catalysts show high activity and selectivity in
the reactions mentioned above to probe superbasicity. Controlled
decomposition of sodium azide impregnated in zeolite NaY has
been successfully used first to poison the acid sites and then to
create strong basic sites in zeolites [6,13–24]. Martens et al. [16–
18] have investigated the generation of both metallic particles and
ionic alkali clusters in faujasite-type zeolites during controlled
thermal decomposition of impregnated sodium azide. The ionic
alkali metal clusters were accommodated in the sodalite cages of
the faujasite structure, while the neutral metal particles were
located in the zeolite large cages. The latter were found to be
associated with the development of superbasic sites, able to
activate some of the specific substrates mentioned.
of traditional
petrochemical industries [3], it is not unusual to encounter recent
efforts studying the preparation of mesoporous -alumina.
g-alumina as catalyst support in the chemical and
g
Materials with specific surface areas up to 800 m2/g and pore
sizes ranging from 2.0 to more than 10 nm have been reported to
be characteristic for organized mesoporous aluminas prepared by
neutral, anionic and cationic synthesis routes [4].
Although solid base catalysts in general have been received
much attention especially due to their advantages for easy
separation and recovery, reduced corrosion and environmental
acceptance [5], there is still a need to prepare solid superbasic
catalysts. Such materials are able to catalyze specific reactions such
as double bond shift in olefins such as vinylcyclohexane (VCH),
vinylnorbornene and 2,3-dimethyl-1-butene, alkene dimerization
and side chain alkylation of alkylaromatics. In particular, strong
basic sites with H_ > 34 are needed to catalyze such reactions [6].
The reactions are therefore often used to probe the presence of
superbasic sites, as they allow in situ inspection of the working
Recently, potassium functionalized mesoporous
g-alumina
containing impregnated K2O as active compound, was claimed
to act as a solid superbase [24]. Unfortunately, the low activity
reported for the isomerization of 1-hexene, does not allow
establishing unequivocally the presence of superbasic sites. A
* Corresponding author.
E-mail address: roxanamelania.bota@biw.kuleuven.be (R.M. Bota).
0920-5861/$ – see front matter ß 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.cattod.2010.01.002

100
R.M. Bota et al. / Catalysis Today 152 (2010) 99–103
Scheme 1. Mechanism for isomerization of vinylcyclohexane (VCH) to ethylidencyclohexane (ECH) with superbase catalyst B.
Scheme 2. Transesterification reaction of triglycerides with methanol. RiCOOCH3 corresponds to the respective methyl fatty acid esters derived from the triglycerides.
solid mixture of dry mesoporous alumina and sodium azide
calcined at 400 8C has been shown to have moderate activity in the
double bond shift in 2,3-dimethyl-1-butene [25]. Unfortunately no
characterization of the sodium metal was provided.
mirror was observed on the glass reactor, corresponding to less
than 10% of the Na metal as determined by its dissolution in water
and back titration of the basicity. Mirror formation was not
observed during activation of the NaN3/MSU-g 84 catalyst. When
In this paper we report the preparation and catalytic use of
comparing both samples catalytically, it was assured that
comparable amounts of sodium were present in the reactor
thermally decomposed sodium azide-on-mesoporous
g-alumina
as a superbasic catalyst in the room temperature double bond
migration of VCH, a reaction requiring superbasic catalytic sites to
occur [6]. The potential of such catalyst in the transesterification of
soybean oil with methanol, a reaction also known as biodiesel
synthesis (Scheme 2), was examined as well. It is not implied that
this reaction requires the presence of superbasic sites.
(
ꢀ5 wt.%).
It should be stressed that during activation of supported azide,
in particular around its decomposition point (300–375 8C) much
heat has to be removed from the system in order to avoid extreme
temperature rise in the catalyst bed (see below). The volume of the
vessel should also allow confining the sudden pressure increase
due to N2 release (3 mmol of N2 per 2 mmol of Na generated). It is
of extreme importance to avoid formation of HN3 (hydrazoic acid)
in particular when high amounts of water are present at the
moment of azide decomposition. HN3 might undergo uncontrolla-
ble detonations. In our hands the activation of the catalyst was safe,
provided the procedures and amounts described are respected.
Upscaling should only be attempted by skilled persons.
The catalysts were also characterized by means of 23Na MAS
NMR spectroscopy. The samples were packed under inert
atmosphere in 2.5 mm zirconia rotors and locked. The closed
rotors were air-tight as shown by the constant signal intensity over
time. The 23Na MAS NMR spectra were recorded on a Bruker
Avance 400 spectrometer, at a magnetic field strength of 9.4 T.
1700 scans were accumulated with a recycle delay of 1 s. The
spinning frequency of the rotor was 20 kHz. Solid NaCl was used
for the calibration of the 23Na chemical shift scale.
2. Experimental
2.1. Catalysts preparation
2.1.1. Synthesis of mesoporous
Mesoporous -alumina, denoted as MSU-
according to the method of Zhang et al. [26]. An aqueous mixture of
aluminum tributoxide, 2-butanol and the tri-block
g-alumina
g
g, was prepared
(EO)19(PO)39(EO)19 surfactant, Pluronic P84, was first aged at
65 8C for 8 h, followed by hydrolysis with excess water. The
precipitate was aged again at 80 8C for 6 h and then at 100 8C for
24 h. The powder air-dried at 60 8C was then calcined at 350 8C for
3 h and then at 550 8C for 4 h. The powder obtained is denoted as
MSU-
g 84.
Nitrogen adsorption isotherms of the mesoporous
g-alumina
were measured at 77 K on a TriStar 3000 V6.03A instrument.
2.2. Catalytic tests
2.1.2. Synthesis of supported sodium azide
A methanol slurry of NaN3 was impregnated on pellets of MSU-
84 with a diameter varying between 0.5 and 1 mm. The resulting
2.2.1. Double bond migration of vinylcyclohexane in fixed bed
continuous liquid flow reactor
g
solid was dried at 60 8C in air and then heated slowly under pure
nitrogen at a rate of 1 8C/min till a final temperature of about 370–
375 8C was reached. After keeping the solid for 1 h at this
temperature, it was cooled down to room temperature and stored
under nitrogen. For a typical catalyst, the final Na loading on
Experiments were carried out at room temperature in a fixed
bed continuous liquid flow reactor connected to a syringe pump
(Perfusor Compact, B. Braun Melsungen AG type 871482/7) with a
bed of 0.5 g of catalyst pellets at a hourly space velocity of 8.4 hꢁ1
.
The reactor flow sheet is shown in Fig. 1. Separation of substrate
and reaction products was done using a GC (HP5890) equipped
with a CP-SIL 5CB WCOT Fused Silica column of 50 m, with an
internal diameter of 0.32 mm and FID detector at 280 8C. The
column temperature was raised from 60 8C at a rate of 2 8C/min till
250 8C.
support was 15 wt.%. This catalyst is denoted as NaN3/MSU-
Alternatively, a solid catalyst denoted as NaN3/MSU- 84DM
was prepared by dry mixing under argon atmosphere of air-dry
MSU- 84 powder with NaN3, corresponding to a 15 wt.% loading
g 84.
g
g
with Na. The two components were ground together in a mortar
and the powder obtained compressed and sieved under argon. The
pellets with a diameter ranging from 0.5 to 1 mm were activated
2.2.2. Transesterification of soybean oil with methanol
and stored in the same way as the NaN3/MSU-
activation this sample turned from white to black, while bed
temperatures up to 550 8C were detected. In parallel a metallic
g
84 catalyst. During
The catalytic transesterification of soybean oil with methanol to
yield methyl esters of the fatty acids contained in the oil, also
known as biodiesel ex soybean oil (Scheme 2) was carried out in a

R.M. Bota et al. / Catalysis Today 152 (2010) 99–103
101
3. Results and discussion
3.1. Characterization of support and catalysts
The nitrogen adsorption isotherm measured at 77.3 K for the
calcined mesoporous -alumina support, MSU- 84, is shown in
g
g
Fig. 2. The data point to successful synthesis of the alumina
mesoporous support, with a BET surface area of 298 m2/g, a pore
volume of 0.97 mL/g and an average pore size of 8.8 nm.
The sodium species present in the catalysts after thermal
activation were determined with solid state 23Na NMR. Several
sodium species can be present in such catalysts, namely metallic
sodium particles, ionic sodium clusters as observed in zeolite NaY,
sodium oxide, and residual sodium azide. Metallic sodium particles
are known to give a signal between 900 and 1200 ppm. In
particular, large metallic sodium particles yield a signal around
1125 ppm, corresponding to the Knight shift of sodium metal [27].
On the other hand, sodium oxide gives a broad signal around
ꢁ10 ppm, while residual sodium azide shows a easily distinguish-
able very sharp signal at ꢁ11 ppm.
From Fig. 3 it follows that the 23Na MAS NMR spectrum of the
Fig. 1. Flow sheet of the fixed bed continuous liquid flow reactor. (A) Syringe pump;
(B) removable glass reactor with (C) catalyst bed; (D) thermocouple; (E) product
collection device.
NaN3/MSU-
oxide (ꢁ10 ppm),
= 1132 ppm, while for the NaN3/MSU-
g
84 catalyst shows next to the presence of sodium
high amount of metallic sodium with
84DM catalyst, only
a
d
g
batch reactor equipped with a sampling device at 60 8C with 1 g of
catalyst. The reaction was done under nitrogen for 2 h at a stirring
rate of 300 rpm. The molar ratio of methanol to soybean oil used
was 9:1 and 27:1, respectively.
The analysis of each sample withdrawn from the batch reactor
was performed by dissolving 0.1 g into 1 ml of 0.1% n-C17H36 in n-
hexane solution. The latter was injected into a gas chromatograph
(HP 6890) with injection port at 250 8C, equipped with a polar
column (BPX-70 SGE) of 60.0 m with an internal diameter of
one resonance line at ꢁ10 ppm is observed, that can be attributed
to sodium oxide particles. Metallic species or residual sodium
azide with sharp resonance line at ꢁ11 ppm, were not detected on
the latter catalyst obtained by dry mixing of the components.
The presence of only sodium oxide particles on the activated
NaN3/MSU-g 84DM catalyst can be correlated to the presence of
residual moisture from the mesoporous support, which is still
present when the sodium azide starts to decompose. Moreover,
due to the exotherm generated by azide decomposition, the
residual surface hydroxyls will be dehydroxylated, yielding more
water.
The same air-dry catalyst support impregnated with methanol
slurry of the azide will contain much less water, while most of the
methanol will be released before azide decomposition. As the azide
decomposition occurs also in a more controlled way with a much
less pronounced exotherm, the presence of lower amounts of
sodium oxide and significant amounts of sodium metal are not
unexpected. It cannot be decided whether both species are discrete
entities or should be visualized as sodium metal particles partially
oxidized at their external surface rim. It can be anticipated that the
latter configuration should exhibit higher basic strength compared
320
mm and a film thickness of 0.25 mm. Detection was with a FID-
detector at 280 8C. The column temperature was held initially at
180 8C for 40 min, heated at a rate of 10 8C/min to 260 8C and held
there for 5 minutes. The biodiesel or methyl ester yield, Y, was
calculated according to the following formula:
mIS ꢂ AFAMEs ꢂ fIS
Y ð%Þ ¼
ꢂ 100
AIS ꢂ msample
with IS, n-heptadecane (n-C17H36) used as a chromatographic
internal standard; f, the response factor; A, the peak area; m, the
mass (g) of sample or of IS.
Fig. 2. Nitrogen adsorption isotherms (left) and BJH pore-size distribution (right) for calcined MSU-
g 84.

102
R.M. Bota et al. / Catalysis Today 152 (2010) 99–103
Fig. 3. 23Na MAS NMR spectrum of (a) NaN3/MSU-
g 84 (top spectra) and (b) NaN3/MSU-g 84DM catalysts (bottom spectra), activated up to 375 8C.
to that of bulk sodium oxide by electron donation from the
subsurface metal to the O ions of the surface oxide.
oxide (see above). Though this picture is to a certain degree
speculative, it is in line with current views on superbasicity of Na-
on-support catalysts [6,8,25].
3.2. Catalytic performance of sodium azide supported on mesoporous
23Na MAS NMR on a fresh and used NaN3/MSU-
g 84 catalyst
g
-alumina for the double bond migration of vinylcyclohexane
shows a decrease of the Na metal particles (1132 ppm) and a
parallel increase in the sodium oxide species (ꢁ10 ppm). The
deactivation of the catalyst can therefore be associated with this
transformation, viz. the deeper oxidation of surface oxidized
sodium metal particles and is an a posteriori confirmation of our
interpretation of the NMR lines. The slight deactivation of this
catalyst is due to the presence of impurities in the feed, viz. water,
oxygen and carbon dioxide. Indeed, when ultrapure VCH is used
obtained by thorough de-aeration and drying, the catalyst stability
is extended significantly (results not shown).
As illustrated in Figs. 4 and 5, the catalytic activity of the sodium
azide-on-mesoporous
tion depends on the preparation method. The NaN3/MSU-
catalyst shows the highest activity and stability for comparable
amounts of Na in the reactor. The conversion of VCH decreases only
slowly over time from almost 100 to 80%, while in case of the NaN3/
g
-alumina catalysts for the VCH isomeriza-
84
g
MSU-g 84DM catalyst the conversion decline of VCH is faster from
almost 70 to 38%. The selectivity for ECH remained above 95% in
both cases. The difference in initial activity should be assigned to
the different nature of the active site, viz. oxide ions of sodium
oxide in surface oxidized Na metal particles or in bulk sodium
Dry mixing of the catalyst components followed by thermal
activation, as in NaN3/MSU-
g 84DM, invariably shows after
thermal activation the appearance of sodium mirrors at the
reactor outlet. The extent of Na loss is below 10% and cannot be
responsible alone for the decreased activity. The enhanced
deactivation is for a major part due to sintering of the supported
sodium oxide particles as it occurs even with pure feeds (results
not shown).
3.3. Catalytic performance of sodium azide supported on mesoporous
g
-alumina for the transesterification of soybean oil with methanol
The NaN3/MSU- 84 and NaN3/MSU- 84DM catalysts were
g
g
also tested in transesterification of soybean oil with methanol in a
batch reactor at 60 8C which represent mild reaction conditions. As
shown in Fig. 6, the NaN3/MSU-g 84 shows again higher activity
than NaN3/MSU- 84DM. After 2 h the biodiesel yield was 60 and
g
Fig. 4. Room temperature catalytic activity with a NaN3/MSU-g 84 catalyst.
Fig. 6. Activity of NaN3/MSU-g 84 and NaN3/MSU-g 84DM catalysts in
transesterification of soybean oil with methanol. Reaction conditions: 60 8C;
300 rpm; 1 g catalyst; reaction time, 2 h; MeOH: soybean oil molar ratio of 9:1,
batch reactor.
Fig. 5. Room temperature catalytic activity with a NaN3/MSU-g 84DM catalyst.

R.M. Bota et al. / Catalysis Today 152 (2010) 99–103
103
of azide decomposition and/or the interaction of sodium metal
particles with water from surface OH dehydroxylation. This solid
was significantly less active in the double shift catalysis of VCH,
probably due to the lack of superbasic sites stemming from the
combined action of supported sodium metal and sodium oxide. It
deactivated significantly assumed to be due to sintering of the
supported sodium oxide.
In the transesterification of soybean oil with methanol the first
catalyst showed very high biodiesel yields in mild conditions,
while the 23Na MAS NMR picture of the catalyst remained virtually
unchanged after reaction. The second catalyst devoid of superbasic
sites showed less activity.
Acknowledgements
Part of the work was done in the frame of a STWW project
sponsored by IWT. R.M. Bota acknowledges a fellowship from KU
Leuven R&D and subsequently (from 2008 onwards) in the frame of
a GOA concerted action on catalysis, sponsored by the Flemish
Government.
Fig. 7. Biodiesel yield obtained with NaN3/MSU-g 84 using a methanol/soybean oil
molar ratio of 27. Reaction conditions: 60 8C; 300 rpm; 1 g catalyst; reaction time,
15 min, batch reactor.
45% in case of the NaN3/MSU-g 84 and NaN3/MSU-g 84DM
catalyst, respectively.
References
The stoichiometric ratio of the transesterification reaction
requires three moles of alcohol for one mole of triglyceride to yield
three moles of fatty acid ester and one mole of glycerol. It is known
that enhanced molar ratios yield higher ester yields [28,29]. For the
[1] C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.S. Beck, Nature 359 (1992)
710.
[2] J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, C.T.-W.
Chu, D.H. Olson, E.W. Sheppard, S.B. McCullen, J.B. Higgins, S.J.L. Chlenker, J. Am.
Chem. Soc. 114 (1992) 10834.
NaN3/MSU-g 84 catalyst it was found (Fig. 7) with high methanol/
[3] F. Schu¨ th, K. Unger, in: G. Ertl, H. Kno¨zinger, J. Weitkamp (Eds.), Preparation of
oil ratios that the reaction is extremely rapid as after 15 min
already the thermodynamic equilibrium of about 60% is reached.
The 23Na MAS NMR of the used catalyst shows a spectrum that is
identical to that of the starting catalyst shown in Fig. 3(a). Thus it
seems that the superbasic sites evidenced by the presence of the Na
metal and Na oxide lines are not affected by this reaction medium.
Solid Catalysts, Wiley–VCH, Weinheim, 1999, p. 77.
ˇ
[4] J. Cejka, Appl. Catal. A 254 (2003) 327.
[5] H. Hattori, Mater. Chem. Phys. 18 (1988) 533.
[6] H. Hattori, Chem. Rev. 95 (1995) 537.
[7] H. Hattori, Stud. Surf. Sci. Catal. 78 (1993) 35.
[8] S. Tsuchiya, Acid–Base Catalysis, Kodansha (Tokyo)–VCH, Basel, Cambridge, New
York, Weinheim, 1989, p. 169.
[9] J.W. Hightower, K.W. Hall, J. Am. Chem. Soc. 89 (1967) 778.
[10] A. Satoh, H. Hattori, J. Catal. 45 (1976) 36.
[11] H. Hattori, M. Itoh, K. Tanabe, J. Catal. 41 (1976) 46.
[12] G. Suzukamo, M. Fukao, T. Hibi, K. Chikaishi, Acid–Base Catalysis, Kodansha–VCH,
1989, p. 405.
4. Conclusion
A sodium azide methanol slurry supported on mesoporous
alumina was decomposed in a controlled way, yielding NMR
resonance lines that could be assigned to Na metal and Na oxide
species. This solid showed superbasicity as evidenced by its
capability to catalyze the double bond shift in vinylcyclohexane.
The minor deactivation of this catalyst in time using a continuous
liquid flow reactor could be attributed to feed impurities. The NMR
lines are assumed to correspond to sodium metal particles contain-
ing oxide at the external rim. This is confirmed by the slight decrease
of the Na metal line and the corresponding slight increase of the
sodium oxide line during subsequent runs with unpurified VCH.
When dry sodium azide was grinded in presence of mesoporous
[13] I. Hannus, I. Kiricsi, K. Varga, P. Fejes, React. Kinet. Catal. Lett. 12 (1979) 309.
[14] P. Fejes, I. Kiricsi, I. Hannus, T. Tihanyi, A. Kiss, Stud. Surf. Sci. Catal. 5 (1980) 135.
[15] I. Kiricsi, I. Hannus, A. Kiss, P. Fejes, Zeolites 2 (1982) 247.
[16] L.R.M. Martens, P.J. Grobet, P.A. Jacobs, Nature 315 (1985) 568.
[17] L.R.M. Martens, P.J. Grobet, W.J.M. Vermeiren, P.A. Jacobs, Stud. Surf. Sci. Catal. 28
(1986) 935.
[18] L.R.M. Martens, W.J.M. Vermeiren, P.J. Grobet, W.J.M. Vermeiren, P.A. Jacobs, Stud.
Surf. Sci. Catal. 31 (1987) 531.
[19] B. Xu, L. Kevan, J. Chem. Soc. Faraday Trans 87 (1991) 2843.
[20] B. Xu, X. Chen, L. Kevan, J. Chem. Soc., Faraday Trans. 87 (1991) 3157.
[21] B. Xu, L. Kevan, J. Phys. Chem. 96 (1992) 2642.
[22] M. Brock, C. Edwards, H. Fo¨rster, M. Schro¨der, Stud. Surf. Sci. Catal. 84 (1994) 1515.
[23] I. Hannus, I. Kiricsi, A. Beres, J.B. Nagi, H. Fo¨rster, Stud. Surf. Sci. Catal. 98 (1995) 81.
[24] L.B. Sun, J. Yang, J.H. Kou, F.N. Gu, Y. Chun, Y. Wang, J.H. Zhu, Angew. Chem. IE 47
(2008) 3418.
[25] T. Seki, S. Ikeda, M. Onaka, Micropor. Mesopor. Mater. 96 (2006) 121.
[26] Z. Zhang, R.W. Hicks, T.R. Pauly, T.J. Pinnavaia, J. Am. Chem. Soc. 124 (2002) 1592.
[27] P.A. Anderson, P.P. Edwards, J. Am. Chem. Soc. 114 (1992) 10608.
[28] A. Dermibas, Energy Convers. Manage. 44 (2003) 2093.
[29] S. Saka, D. Kusdiana, Fuel 80 (2001) 225.
g
-alumina and subsequently thermally activated, only Na oxide
species were present on the support, next to Na mirror formation
(less than 10% of Na) on the glass walls of the vessel. This was
ascribed to the presence of some residual water at the temperature