Sulfated mesoporous alumina: A highly effective solid strong base catalyst for the tishchenko reaction in supercritical carbon dioxide

Article abstract of DOI:10.1021/jp055895m

Heterogeneous strong base catalysis for the intramolecular Tishchenko reaction of aromatic 1,2-dicarbaldehydes to the corresponding phthalides in supercritical CO₂CscCO₂ has been realized with mesoporous alumina containing SO₄^2- ions in the alumina framework (mesoAl₂O₃/SO₄^2-). Infrared spectroscopy of pyrrole adsorbed on the alumina and strong poisoning by a weak Broensted acid of methanol revealed that the SO₄^2- ions in the framework slightly suppressed the average strength of base sites (O^2-) on mesoAl₂O₃/SO₄^2-, but there exists a small number of strong base sites that promote the Tishchenko reaction in scCO₂. Although the intramolecular Tishchenko reaction of phthalaldehyde to phthalide in scCO₂ was somewhat slower than those in organic solvents such as tetrahydrofuran (THF) and benzene, the addition of a small amount of THF as a cosolvent remarkably increased the reaction rate; the reaction in the scCO₂ - THF system proceeded 1.5-fold faster than those in pure benzene and THF solvents.

Full text of DOI:10.1021/jp055895m

1240
J. Phys. Chem. B 2006, 110, 1240-1248
Sulfated Mesoporous Alumina: A Highly Effective Solid Strong Base Catalyst for the
Tishchenko Reaction in Supercritical Carbon Dioxide
Tsunetake Seki† and Makoto Onaka*
Department of Chemistry, Graduate School of Arts and Sciences, The UniVersity of Tokyo,
Komaba, Meguro-ku, Tokyo 153-8902, Japan
ReceiVed: October 14, 2005; In Final Form: NoVember 22, 2005
Heterogeneous strong base catalysis for the intramolecular Tishchenko reaction of aromatic 1,2-dicarbaldehydes
to the corresponding phthalides in supercritical CO2CscCO2 has been realized with mesoporous alumina
containing SO42- ions in the alumina framework (mesoAl2O3/SO42-). Infrared spectroscopy of pyrrole adsorbed
on the alumina and strong poisoning by a weak Bro¨nsted acid of methanol revealed that the SO42- ions in the
framework slightly suppressed the average strength of base sites (O2-) on mesoAl2O3/SO42-, but there exists
a small number of strong base sites that promote the Tishchenko reaction in scCO2. Although the intramolecular
Tishchenko reaction of phthalaldehyde to phthalide in scCO2 was somewhat slower than those in organic
solvents such as tetrahydrofuran (THF) and benzene, the addition of a small amount of THF as a cosolvent
remarkably increased the reaction rate; the reaction in the scCO2-THF system proceeded 1.5-fold faster than
those in pure benzene and THF solvents.
Introduction
in scCO2.4 In this article, we report a full account on the strong
base catalysis of mesoAl2O3/SO42- for the Tishchenko reaction
Supercritical fluids (SCFs) have been recognized as a unique
reaction medium possessing the merits of both gases and liquids,
because their physical properties can be continuously adjusted
from those analogous to gases to those analogous to liquids by
manipulating the pressure and temperature.1 The high density
required to dissolve and extract many organic compounds, the
high diffusivity and the low viscosity enhancing mass transfer,
and the high thermal conductivity eliminating reaction heat
efficiently are suitable especially for heterogeneous catalytic
reactions that take place in the pores of solids. Among the SCFs,
supercritical carbon dioxide (scCO2) is the most attractive
because of its low cost, nontoxic and nonflammable nature, and
mild critical temperature (304 K) and pressure (7.4 MPa). The
Lewis acidity of CO2, however, has restricted the use of scCO2
medium in several homogeneous and heterogeneous acid- or
transition metal-catalyzed reactions;1 to our knowledge, no
attempts have been carried out on strong base-catalyzed reac-
tions in scCO2.
in scCO2 (Scheme 1) and on unique properties of the surface
2-
and bulk of mesoAl2O3/SO4
.
Experimental Section
Preparation of mesoAl2O3. Pure mesoporous alumina
(mesoAl2O3) was synthesized according to Davis’ report:5 into
a polypropylene vessel containing a stirring bar were added Al-
(O-sec-Bu)3 (21.9 g, 89.0 mmol), 1-propanol (120 g, 2.00 mol),
and deionized water (5.15 g, 286 mmol), followed by vigorous
stirring at room temperature. After 1 h, a solution of lauric acid
(5.40 g, 27.0 mmol) in 1-propanol (17.5 g, 290 mmol) was added
under vigorous stirring, and the mixture was further stirred
vigorously at room temperature for 24 h. The resulting white
liquid was placed in a 300-mL autoclave and heated at 383 K
for 48 h without stirring to afford a white precipitate. The
precipitate was washed on a filter paper with a large amount of
ethanol and dried at room temperature under an N2 flow. The
white solid thus obtained was heated from room temperature
to 873 K at a ramping rate of 10 K min-1 under an N2 flow,
and then calcined under an air flow at 873 K for 5 h to burn
out the organic species to give pure mesoAl2O3. After several
repetitions of the synthesis, it was found that the synthesized
Infrared spectroscopic studies revealed that on contact of CO2
with solid bases, carbonate species, such as monodentate or
bidentate carbonate ions, and bicarbonate ions were readily
formed on the surface, which inevitably led to strong inhibition
of solid base catalysis.2 Therefore, it is a challenging task to
perform solid base-catalyzed organic reactions in the scCO2
medium. Development of solid strong base catalysts utilizable
in scCO2 would expand the potential of scCO2 as a safe and
environmentally benign reaction medium for chemical synthesis.
Recently, we have successfully synthesized a new solid strong
mesoAl2O3 had surface areas in the range of 474-523 m2 g-1
.
In the present study, we employed two lots of mesoAl2O3: the
one with the largest surface area of 523 m2 g-1 was designated
as mesoAl2O3(#1) and the other with the smallest surface area
of 474 m2 g-1 as mesoAl2O3(#2).
Preparation of mesoAl2O3/SO42--I and -II. To a mixture
of Al(O-sec-Bu)3 (21.9 g, 89.0 mmol) and 1-propanol (120 g,
2.00 mol) was added deionized water (5.15 g, 286 mmol)
containing 96% H2SO4 (0.1 g, 1 mmol and 0.2 g, 2 mmol for
mesoAl2O3/SO42--I and -II, respectively). The resultant solution
was vigorously stirred for 1 h at room temperature. Lauric acid
(5.40 g, 27.0 mmol) in 1-propanol (17.5 g, 290 mmol) was then
added, and the mixture was further stirred for 24 h at room
2-
base material of mesoporous alumina containing SO4 in the
framework (designated as mesoAl2O3/SO42-), and preliminarily
reported its strong base catalysis for the Tishchenko reaction3
* Address correspondence to this author. E-mail: conaka@mail.ecc.u-
tokyo.ac.jp.
† T.S. is a Research Fellow of the Japan Society for the Promotion of
Science.
10.1021/jp055895m CCC: $33.50 © 2006 American Chemical Society
Published on Web 12/24/2005

2-
Strong Base Catalysis of mesoAl2O3/SO4 in scCO2
J. Phys. Chem. B, Vol. 110, No. 3, 2006 1241
SCHEME 1: Intramolecular Tishchenko Reaction of
Aromatic 1,2-Dicarbaldehydes to Phthalides in ScCO2
Benzene, tetrahydrofuran (THF), acetic acid, and methanol
were purified by distillation, and stored in an N2-filled vial with
MS 4A, which had been activated at 773 K for 2 h under
vacuum (10-4 Torr).
Pyrrole for the infrared spectroscopy was distilled under
vacuum, then passed through a column of MS 4A, which had
been activated at 773 K for 2 h under vacuum (10-4 Torr), and
stored in vacuo, all the procedures of which were performed in
a vacuum-line apparatus.
Measurement of Surface Area and Pore Diameter. N2
adsorption-desorption measurements were performed at 77 K
with a BELSORP 28SA (BEL Japan), using static adsorption
procedures. The samples were pretreated under the same
conditions as employed for the reactions (Ca(OH)2: 873 K, 10-4
Torr, 2 h; aluminas: 773 K, 10-4 Torr, 2 h) prior to analysis.
Adsorption data were analyzed by the Brunauer-Emmett-
Teller (BET) method8a for surface areas, and the Dollimore-
Heal (DH) method8b to obtain pore-size distribution curves.
temperature. The resulting white liquid was placed in a 300-
mL autoclave and heated at 383 K for 48 h without stirring to
afford a white precipitate. The precipitate was washed on a filter
paper with a large amount of ethanol and dried at room
temperature under an N2 flow. The white solid thus obtained
was heated from room temperature to 873 K at a ramping rate
of 10 K min-1 under an N2 flow and calcined under an air flow
at 873 K for 5 h to burn out the organic species to yield
mesoAl2O3/SO42--I or -II. Bedilo and Klabunde employed this
method to synthesize sulfated zirconia: the use of a small
2-
amount of H2SO4 resulted in the incorporation of SO4 ions
into the zirconia framework.6
Infrared Spectroscopy of Adsorbed Pyrrole and Phthala-
Preparation of mesoAl2O3/SO42--III. A mixture of Al(O-
sec-Bu)3 (19.5 g, 79.0 mmol), Al2(SO4)3 (1.7 g, 5.0 mmol),
1-propanol (120 g, 2.00 mol), and deionized water (5.15 g, 286
mmol) was vigorously stirred for 1 h at room temperature.
Lauric acid (5.40 g, 27.0 mmol) in 1-propanol (17.5 g, 290
mmol) was then added, and the resulting mixture was further
stirred for 24 h at room temperature. The resulting white liquid
was placed in a 300-mL autoclave and heated at 383 K for 48
h without stirring to afford a white precipitate. The precipitate
was washed on a filter paper with a large amount of ethanol
and dried at room temperature under an N2 flow. The white
solid thus obtained was heated from room temperature to 873
K at a ramping rate of 10 K min-1 under an N2 flow, and
calcined under an air flow at 873 K for 5 h to burn out the
organic species to give mesoAl2O3/SO42--III.
ldehyde. Infrared (IR) spectra were recorded at room temper-
ature on an FT/IR-550 (JASCO) with a resolution of 4 cm-1
.
The catalyst pressurized into a thin disk at 200 kg cm-2 was
pretreated in a quartz-made IR cell equipped with an optical
path window made of KBr under the same conditions as
employed for the reaction (Ca(OH)2: 873 K, 10-4 Torr, 2 h;
aluminas: 773 K, 10-4 Torr, 2 h). Gaseous pyrrole at 5 Torr or
1a at its vapor pressure at room temperature was then introduced
into the IR cell at room temperature for 30 min and 4 h,
respectively, followed by evacuation at 10-4 Torr at room
temperature for 30 min. The sample thus obtained was
transferred to the optical path part and infrared spectra were
recorded.
X-ray Diffraction Analysis. X-ray powder diffraction mea-
surement was performed with a MultiFlex (Rigaku) under
ambient atmosphere. The sample was pretreated at 773 K at
10-4 Torr for 2 h, mounted on an XRD sample cell as fine
powder under ambient atmosphere, and analyzed at a current
of 40 mA and a voltage of 40 kV in the 2θ range 1° to 80°
Sulfur Content. The sulfur content in mesoAl2O3/SO42- was
determined by the infrared absorption method after combustion
in an induction furnace with an EMIA-520 (HORIBA). A
sample was burned under oxygen atmosphere with a mixture
of metallic iron, tungsten, and tin that served as a combustion
assistant to generate SO2, the amount of which was determined
by infrared spectroscopy. The sulfur contents in mesoAl2O3/
SO42--I, -II, and -III were 0.23, 0.48, and 1.30 wt %,
respectively.
with a scanning rate of 1.0 deg min-1
.
Reaction Procedures. SAFETY WARNING! Operators
handling high-pressure equipment for reactions in scCO2 should
take proper precautions to minimize the risk of personal injury.
The Other Catalysts. CaO was obtained from thermal
decomposition of Ca(OH)2 at 873 K under vacuum (10-4 Torr)
for 2 h. ALO-2, -3, and -4 are the reference alumina catalysts
supplied from the Catalysis Society of Japan, and contain the
following impurities: ALO-2: 0.03% Fe2O3, 0.22% SiO2,
0.04% Na2O, 1.72% SO42-; ALO-3: 0.01% Fe2O3, 0.01% SiO2,
0.3% Na2O, 0.01% TiO2; and ALO-4: 0.01% Fe2O3, 0.01%
SiO2, 0.01% Na2O. mesoAl2O3/SO42--IV was prepared by the
same procedures as that employed for the synthesis of mesoAl2O3/
SO42--I and -II, but (NH4)2SO4 (0.35 g, 2.6 mmol) was used as
a sulfate ion source in place of H2SO4.
Organic Reagents. Phthalaldehyde (1a) (Aldrich) and 2,3-
naphthalenedicarbaldehyde (2a) (Aldrich) were distilled under
vacuum before use. 4,5-Dimethylbenzene-1,2-dicarbaldehyde
(3a) and 4,5-dichlorobenzene-1,2-dicarbaldehyde (4a) were syn-
thesized by Farooq’s procedure:7 the reduction of 4,5-dichlo-
robenzene-1,2-dicarboxylic acid (Aldrich) to the corresponding
aromatic 1,2-dimethanol with LiAlH4, followed by Swern
oxidation. 4,5-Dimethylbenzene-1,2-dimethanol (Wako) was
oxidized by the Swern method. All the aromatic 1,2-dicarbal-
dehydes were well dried under vacuum at room temperature
for several hours and stored in a vial under N2 atmosphere.
The Tishchenko reaction was performed with an scCO2
reactor system of SCF-Get, SCF-Bpg, and SCF-Sro (JASCO).
A typical reaction procedure for the experiment of entry 9 in
Table 1 is as follows: into a 10-mL stainless steel autoclave
(SUS316, TAIATSU TECHNO) containing a stirring bar and
freshly distilled phthalaldehyde (1a, 1.00 mmol) was added
mesoAl2O3/SO42--III (50 mg) preactivated at 773 K under
vacuum (10-4 Torr) for 2 h in a nitrogen-filled glovebag. Liquid
CO2 cooled to 263 K was introduced into the autoclave, which
had been heated to 313 K, using an HPLC pump, and the
pressure was adjusted to 8.0 ( 0.2 MPa. The resultant mixture
was stirred for 2 h at 313 K with the CO2 pressure constantly
maintained at 8.0 ( 0.2 MPa. The autoclave was cooled in an
ice bath, and the pressurized CO2 was gradually released. The
resulting solid mixture was washed from the autoclave with
THF. The THF slurry was filtered through a Celite pad.
Evaporation of the THF filtrate gave a crude solid mixture, and
the solid was dissolved in CDCl3 with Ph3CH (internal standard
substance) and applied to NMR analysis. The yield of phthalide
1
(1b, 73%) was determined with H NMR (270 or 500 MHz)
by the comparison between the peak area of Ph3CH and that of
-CH2- in phthalide. The products, phthalide (1b), naphthalide

1242 J. Phys. Chem. B, Vol. 110, No. 3, 2006
Seki and Onaka
for the reaction in benzene solvent.3c,d However the activity of
CaO was quite low in scCO2, presumably owing to the
instantaneous formation of inactive CaCO3 on the CaO surface
in the dense CO2. This result clearly indicates that representative
solid strong base catalysts of alkaline earth oxides are inadequate
for base-catalyzed reactions in acidic scCO2 medium.
TABLE 1: Catalytic Activities of CaO, Aluminas, and
Sulfated Aluminas for the Intramolecular Tishchenko
Reaction of Phthalaldehyde (1a) to Phthalide (1b)a
surface area pore diameter
yield
(%)e
c
entry
catalystb
(m2 g-1
)
(nm)d
1
2
3
4
5
6
7
8
9
CaOf
ALO-2
ALO-3
ALO-4
48
312
129
173
523
474
532
567
570
571
1
62 (64g)
<1
In our previous studies γ-alumina was found to be able to
promote the intramolecular Tishchenko reaction in benzene
solvent.3d Therefore, we applied three types of γ-aluminas,
ALO-2, -3, and -4, to the reaction. ALO-2 contains a few weight
12 (15g)
62
mesoAl2O3(#1)
mesoAl2O3(#2)
mesoAl2O3/SO42--I
mesoAl2O3/SO42--II
mesoAl2O3/SO42--III
2.7
2.7
2.9
2.4
3.4
2.9
51 (59g)
66
2-
percent of SO4 ions, ALO-3 includes Na+ ions, and ALO-4
62
is almost pure γ-alumina. The three aluminas exhibited remark-
able differences in the activity depending on the kind of
impurities: ALO-2 showed quite high activity (Table 1, entry
2), while more basic aluminas of ALO-3 and -4 gave poor
catalysis (entries 3 and 4).
73 (81g)
56
10 mesoAl2O3/SO42--IV
a Reaction conditions are as follows: reactor, a 10-mL stainless steel
autoclave; catalyst, 50 mg; 1a, 1.00 mmol; CO2 pressure, 8.0 ( 0.2
MPa; reaction temperature, 313 K; reaction time, 2 h. b Each catalyst
was pretreated at 773 K for 2 h under vacuum (10-4 Torr). c Determined
by the BET method. d Determined by the DH method. e Determined
by 1H NMR, using Ph3CH as an internal standard. f Prepared from
Ca(OH)2 at 873 K for 2 h under vacuum (10-4 Torr). g Reaction time
was 4 h.
It is assumed that the surface basicity of ALO-2 is weaker
than that of ALO-3 and -4 through the electron-withdrawing
inductive effects of included SO42- ions,2a,10 which leads to less
adsorption of acidic CO2 onto the surface. The nature of alumina
impeding the formation of the corresponding carbonate and our
2-
capability to adjust the surface basicity by incorporated SO4
(2b), 5,6-dimethylphthalide (3b), and 5,6-dichlorophthalide (4b),
ions gave us a fruitful guide to develop a new type of solid
strong base catalyst that would efficiently function even in acidic
scCO2.
1
were confirmed by comparison with H NMR data in refs 9a,
9b, 9c, and 9d, respectively.
For reaction in an scCO2-cosolvent system, cosolvent (0.1
mL) was added with a syringe into the autoclave containing 1a
and mesoAl2O3/SO42--III in a nitrogen-filled glovebag, followed
by the introduction of liquid CO2.
The reaction procedures described above gave experimentally
reproducible data for several runs.
Observation of Phase Behaviors. Into a 10-mL autoclave
equipped with sapphire windows (TAIATSU TECHNO) was
added 1.00 mmol of freshly distilled 1a or 1b and/or 0.1 mL of
THF with a stirring bar, followed by the introduction of liquid
CO2. The mixture was stirred at 313 K for 15 min, and then
the phase behavior was visually inspected through the windows.
1a-scCO2 system: at the CO2 pressure of 8.0 ( 0.2 MPa, 1a
dissolved almost completely in scCO2, but a small amount of
1a existed at the bottom of the vessel as a yellow liquid.
Increasing the CO2 pressure led to the increase in the solubility
of 1a, and 1a completely dissolved at 10.0 ( 0.2 MPa. The
scCO2 phase was colorless and transparent under all the above
conditions.
Interestingly, we found that the reaction was promoted by
mesoporous alumina (mesoAl2O3) possessing uniform nanosized
pores 2.7 nm in diameter prepared from Al(O-sec-Bu)3 through
the sol-gel process5 (entries 5 and 6), even though the
mesoAl2O3 contained no sulfate ions. The yield of 62% with
mesoAl2O3(#1) was much higher than that of 12% with pure
γ-alumina, ALO-4. In this study we tested two lots of meso-
porous alumina, mesoAl2O3(#1) and mesoAl2O3(#2), which had
been prepared separately. Any difference in the catalytic activity
between mesoAl2O3(#1) and mesoAl2O3(#2) would arise from
a difference in the surface area.
This finding encouraged us to further develop mesoAl2O3-
type solid base intentionally modified with SO42- ions to achieve
higher catalytic performance. We introduced sulfate ions into
the framework of mesoAl2O3, not on its surface, in a similar
way as in the preparation of sulfated zirconia aerogels,6 because
1a requires close contact with a metal oxide surface to initiate
the Tishchenko reaction.2,3 A mesoporous alumina of mesoAl2O3/
2-
SO42--I incorporating SO4 with a sulfur content of 0.23 wt
1a-THF-scCO2 system: the addition of THF enhanced the
solubility of 1a, and 1a completely dissolved even at the CO2
pressure of 8.0 ( 0.2 MPa. The light yellow homogeneous fluid
was then turbid, implying that local density enhancement
occurred by the formation of 1a-THF-CO2 clusters.
1b-scCO2 system: compared with 1a, 1b was far less soluble
in scCO2. Even at the CO2 pressure of 13.0 ( 0.2 MPa, the
white solids of 1b remained at the bottom of the vessel. The
scCO2 phase was colorless and transparent in this system.
1b-THF-scCO2 system: no drastic increase in the solubility
of 1b was observed by the addition of THF at the CO2 pressure
of 8.0 ( 0.2 MPa. The colorless fluid phase was somewhat
turbid, indicative of the formation of 1b-THF-CO2 clusters
in a slight amount.
% was successfully synthesized by the hydrolysis of Al(O-sec-
Bu)3 with deionized water containing H2SO4, and it exhibited
a higher yield of 66% than the nonsulfated mesoAl2O3 (entries
5 and 6 vs entry 7). It is noteworthy that an increase in the
amount of SO42- in mesoAl2O3 with a sulfur content from 0.23
to 0.48 wt % in mesoAl2O3/SO42--II brought about an increase
in the surface area, but a decrease in the yield, presumably owing
to some reduction of the surface basicity (entry 8). mesoAl2O3/
SO42--III with a sulfur content of 1.30 wt % was synthesized
with a relatively large amount of Al2(SO4)3 in place of H2SO4
as a sulfate ion source. mesoAl2O3/SO42--III had a higher
surface area of 570 m2 g-1 than the other aluminas. With
mesoAl2O3/SO42--III, an introduction of sulfate ions into the
alumina framework led to not only an expansion of the surface
area of mesoporous alumina, but also the highest yield of 1b
(entry 9), indicating that mesoAl2O3/SO42--III had the most
suitable basic character for the Tishchenko reaction of 1a in
acidic scCO2.
Results and Discussion
Catalytic Activities of Various Solid Strong Base Catalysts.
We first tested CaO, which had been prepared by thermal
treatment of Ca(OH)2, in the reaction of 1a to 1b in scCO2
(Table 1, entry 1), because CaO exhibited the highest activity
mesoAl2O3/SO42--IV, in which sulfate ions were introduced
by (NH4)2SO4, possessed a high surface area of 571 m2 g-1
,

2-
Strong Base Catalysis of mesoAl2O3/SO4 in scCO2
J. Phys. Chem. B, Vol. 110, No. 3, 2006 1243
TABLE 2: Comparison of the Reaction Rates of the
mesoAl2O3/SO42--III-Catalyzed Intramolecular Tishchenko
Reaction of Phthalaldehyde (1a) to Phthalide (1b) in Various
Reaction Mediaa
pKa of the
yield
(%)c
entry
reaction medium
cosolventb
1
2
3
4
5
6
7
benzened
38
40
31
48
58
<1
6
tetrahydrofurand
e
scCO2
scCO2-benzenef
g
g
4.8
15.5
scCO2-tetrahydrofuranf
scCO2-acetic acidf
scCO2-methanolf
a Reaction conditions are as follows: reactor, a 10-mL flask for
entries 1 and 2 and a 10-mL stainless steel autoclave for entries 3-7;
catalyst (pretreated at 773 K for 2 h under vacuum (10-4 Torr)), 50
mg; 1a, 1.00 mmol; reaction temperature, 313 K; reaction time, 15
min. b Data taken from the following: Ionization Constants of Organic
Acids in Aqueous Solution; Serjeant, E. P., Dempsey, B., Eds.; IUPAC
Chemical Data Series No. 23; Pergamon Press: Oxford, UK, 1979.
c Determined by 1H NMR, using Ph3CH as an internal standard. d The
reaction was carried out with 1 mL of benzene or THF under N2
atmosphere. e CO2 pressure, 8.0 ( 0.2 MPa. f CO2 pressure, 8.0 ( 0.2
MPa; the amount of the cosolvent, 0.1 mL. g Not discussed here.
Figure 1. Variation of the yield of phthalide (1b) as a function of
reaction time. Reaction conditions are as follows: reactor, a 10-mL
stainless steel autoclave; catalyst (activated at 773 K for 2 h under
vacuum (10-4 Torr)), 50 mg; phthalaldehyde (1a), 1.00 mmol; scCO2,
8.0 ( 0.2 MPa; reaction temperature, 313 K; cosolvent, 0.1 mL (0)
ALO-4, (9) ALO-2, (O) mesoAl2O3(#2), (b) mesoAl2O3/SO42--III, and
([) mesoAl2O3/SO42--III in scCO2-THF medium.
but exhibited relatively low activity, indicating that (NH4)2SO4
2-
is not a suitable SO4 ion source for the present purpose
(entry 10).
1
The H NMR spectra of the THF-extracted crude material
1b. The 1a-, 1b-, and CO2-philic natures of THF might lead to
the formation of 1a-THF-CO2 and 1b-THF-CO2 clusters
in the reactor, which should enhance the mass transfer of 1a
and 1b inside the mesopores. It also should be noted that THF
solvent does not essentially cause the rate retardation of the
reaction catalyzed by mesoAl2O3/SO42--III (entries 1 and 2 in
Table 2). This implies that THF molecules would not interact
with Al3+ sites on mesoAl2O3/SO42--III more strongly than with
those on ALO-4, so that 1a would be able to replace the THF
on the Lewis acid site smoothly, thereby initiating the reaction.
In contrast to the favorable effects of benzene and THF, the
addition of a protic cosolvent such as acetic acid (CH3CO2H)
and methanol (CH3OH) gave remarkably negative effects on
the reaction rates (entries 6 and 7). The poisoning by a Very
weak acid, methanol (pKa ) 15.5), clearly demonstrated that
the Tishchenko reaction in scCO2 was promoted by so strong
base sites on mesoAl2O3/SO42--III as to be able to abstract a
proton from methanol. It can be estimated that the strength of
the base sites on the alumina active for the Tishchenko reaction
should be stronger than H- ) 15.5, where H- is “acidity
function” proposed by Paul and Long.2,11
after depressurization of scCO2 in entries 1-10 only showed
those of reactant 1a and product 1b.
Comparison of Reaction Rates in Various Reaction Media.
To compare a pure scCO2 medium with other reaction media,
we carried out the mesoAl2O3/SO42--III-catalyzed reaction of
1a to 1b not only in a pure organic solvent of benzene or THF,
but also in scCO2-cosolvent systems (Table 2).
Although the reaction in pure scCO2 was somewhat slower
than those in benzene and THF solvents (entries 1 and 2 vs
entry 3), a drastic increase in the reaction rate was observed
when a small amount of benzene or THF was added as a
cosolvent into scCO2 (entries 1 and 2 vs entries 4 and 5). It
should be noted that the reaction in the scCO2-THF system
proceeded 1.5- and 2-fold faster than those in pure benzene and
THF solvents and that in pure scCO2 medium, respectively. The
former results clearly indicated that performing the mesoAl2O3/
SO42--III-catalyzed reaction in scCO2 offers distinct improve-
ment in the catalysis as well as green-chemical benefits in using
nonpoisonous inorganic medium. The higher diffusivity and
lower viscosity of the scCO2-THF medium than liquid benzene
and THF along with high dissolving power for 1a and 1b are
likely to bring about the remarkable rate enhancement.
The fact that THF was the best cosolvent was somewhat
surprising to us, because we had observed noticeable inhibition
effects with THF solvent in comparison with benzene solvent
upon the Tishchenko reaction catalyzed by γ-alumina (ALO-
4).3d Since promoting the Tishchenko reaction requires Lewis
acid sites such as surface metal cations as well as strong base
sites, Lewis basic molecules such as THF are considered to
retard the reaction through the adsorption onto Lewis acid sites.
Tanabe and Saito also reported the poisoning effect with Lewis
basic pyridine on the Tishchenko reaction of benzaldehyde over
CaO.3a It can be speculated that THF added into scCO2
predominantly reacts with CO2 to give a Lewis acid-base
complex, and thus the tight adsorption of THF onto the Lewis
acid sites (Al3+) on mesoAl2O3/SO42--III could be suppressed.
In addition, the presence of THF cosolvent in scCO2 medium
could improve the solubilizing power of scCO2 toward 1a and
Kinetic Study. Figure 1 shows changes in the yield of 1b
with reaction time over four alumina catalysts. The pure
γ-alumina catalyst, ALO-4, exhibited the slowest reaction rate,
and the reaction stopped after 30 min. By contrast, ALO-2
2-
containing a few weight percent of SO4 ions that act as a
“surface basicity controller” showed much higher activity and
a longer lifetime of the catalysis compared with ALO-4. It was
noted that the lifetime as well as the catalytic activity of
mesoAl2O3/SO42--III was the highest among the catalysts tested.
The deactivation in catalysis of ALO-4 in a short period was
only observed in the reaction operated in scCO2, while the same
reaction in benzene solvent proceeded until 1a was completely
consumed.3d The deactivation mechanism will be discussed in
detail in the section on Mechanistic Consideration.
Another interesting feature of the Tishchenko reaction
catalyzed by mesoAl2O3/SO42--III and mesoAl2O3(#2) is a two-
stepwise increment in the yield of 1b in Figure 1. Considering
that the relatively low solubility of 1a in scCO2 at 313 K and

1244 J. Phys. Chem. B, Vol. 110, No. 3, 2006
Seki and Onaka
the CO2 density on the other alumina catalysts, we can simply
conclude that a dilution effect on 1a with an increased amount
of CO2 mainly should be responsible for the reduction of the
yield.
Variation in Reactants. Table 3 shows the results of the
Tishchenko reactions of various aromatic 1,2-dicarbaldehydes
(0.10 mmol) to the corresponding lactones with mesoAl2O3/
SO42--III (5 mg) in scCO2 (353 K, 10.0 ( 0.2 MPa) for 2 or
24 h. Under this condition, the conversion of 1a into 1b was
very rapid, and 1b was formed in 86% yield in a short period
of 2 h (entry 1).
Although the reaction of 2,3-naphthalenedicarbaldehyde (2a)
was sluggish compared with that of 1a, the corresponding
lactone (2b) was obtained in 31% yield in 24 h (entry 2). The
reaction of 2a also proceeded much slower than that of 1a in
benzene solvent when γ-alumina (ALO-4) was used as a
catalyst.3d The adsorption of 2a on the alumina surface is a
prerequisite process for the initiation of the reaction. The low
reactivity is ascribed to the steric hindrance of a large
naphthalene ring of the adsorbed species as well as the less CO2-
philic and less soluble nature of the naphthalene derivative.13
The yield, however, could be raised to 60% by using THF (0.1
mL) as a cosolvent (entry 2) with the improvement of the mass
transfer of 2a and 2b in the mesoporous network through the
enhancement of the solubility of them.
Figure 2. Variation of the yield of phthalide (1b) as a function of
CO2 pressure. Reaction conditions are as follows: reactor, a 10-mL
stainless steel autoclave; catalyst (activated at 773 K for 2 h under
vacuum (10-4 Torr)), 50 mg; phthalaldehyde (1a), 1.00 mmol; reaction
temperature, 313 K; reaction time, 2 h; (0) ALO-4, (9) ALO-2, (O)
mesoAl2O3(#2), and (b) mesoAl2O3/SO42--III.
The reaction of 4,5-dimethylbenzene-1,2-dicarbaldehyde (3a)
progressed much faster than that of 2a, but slower than that of
1a (entry 3). This reactivity order suggests that the reaction rate
was influenced mainly by the steric bulkiness of a reactant
dialdehyde.
8 MPa is responsible for the poor contact between 1a and the
alumina surface, we assume that the reaction on the mesoporous
alumina proceeds in the following way: at the early stage within
30 min, the reaction is promoted mainly on the external surface
of the mesoporous alumina. After the external active base sites
are deactivated in 30 to 60 min, 1a is gradually delivered into
the cavity and the reactant can be contacted with the base sites
on the internal alumina surface until the active base sites are
fully deactivated. In Figure 1 is also shown the excellent
acceleration effect of THF as cosolvent on the reaction with
mesoAl2O3/SO42--III. The addition of a slight amount of THF
brought about a monotonic increase in the yield of 1b instead
of a stepwise one. This drastic change in the yield can be
interpreted as follows: the concentration of 1a in the vicinity
of active base sites on the external surface and those inside the
pores would always be almost equal owing to the high
diffusivity and high solubilizing power of the scCO2-THF
medium for 1a. In other words, the stepwise increment in the
yield of 1b would be attributed to the low solubility of 1a in
pure scCO2.
The present catalytic system was also successfully applicable
to the chlorine-containing aromatic 1,2-dicarbaldehyde, 4a (entry
4). The production of 4,5-dichlorophthalide (4b) even in a
moderate yield was attractive, because 4b has been employed
as a precursor for the synthesis of pharmaceutically promising
compounds.9d,14 The result that the reaction of 4a was much
slower than that of 1a might come from the Lewis acid-base
interactions between the Lewis acid sites (Al3+) and the two
chlorine atoms of 4a. Such interactions should prevent 4a from
being adsorbed in the appropriate arrangement for the reaction
to take place. In addition, the highly volatile nature of 4a would
cause poor contact of 4a with the solid catalyst that mainly
stayed around the bottom of the autoclave. 4a sublimes at 303-
313 K at 760 Torr.7
The 1H NMR spectra of the THF-extracted crude material in
entries 1-4 included those of products 1b-4b and reactants
1a-4a exclusively.
Effect of CO2 Pressure. Generally, the solubilizing power
of a substance increases with an increase in its density.1 Thus
if the poor solubility of 1a in scCO2 retards the reaction, an
increase in CO2 density should enhance the reaction rate.
Surface Basicity Measurement by Infrared Spectroscopy
of Adsorbed Pyrrole. The noticeable difference in the catalytic
activity for the Tishchenko reaction in scCO2 between conven-
tional γ-alumina (ALO-4) and mesoAl2O3/SO42- indicates that
As is obvious from Figure 2, however, increasing the CO2
density through an increase in the CO2 pressure gradually
decreased the activities of the catalysts, although some excep-
tional results were also observed. For instance, changing the
CO2 pressure from 9 to 10 MPa in the ALO-4-catalyzed reaction
led to a remarkable increase in the yield of 1b. It is well-known
that the CO2 density at 313 K greatly leaps from 493 kg m-3 at
9.0 MPa to 633 kg m-3 at 10.0 MPa.12 Therefore, the
improvement of the solubility of 1a might account for the large
increase in the yield in the CO2 pressure range. In fact, using
an autoclave with sapphire windows we observed that 1.00
mmol of 1a completely dissolved in scCO2 at 10.0 ( 0.2 MPa,
but that a few oily drops of 1a remained on the wall of the
vessel at 9.0 ( 0.2 MPa. However, as we obtained a general
tendency that the yields of 1b roughly decreased with increasing
2-
mesoAl2O3/SO4 possessed a less CO2-philic surface with
2-
strongly basic sites: the SO4 ions incorporated into the
mesoporous alumina framework should render the surface less
2-
CO2-philic by electron-withdrawing inductive effects of SO4
ions upon lowering the average base strength of the oxide
ions.2a,10 In addition, considering that pure mesoAl2O3 also
exhibited a much higher activity than conventional γ-alumina
(ALO-4), the mesoporous structure itself is associated with some
reduction in the average base strength of oxide ions.
To compare the surface basicity between conventional
γ-alumina (ALO-4) and mesoAl2O3/SO42-, we employed in-
frared spectroscopy using pyrrole as a probe molecule.15 For
less basic oxides such as zeolites, their base strength can be

2-
Strong Base Catalysis of mesoAl2O3/SO4 in scCO2
J. Phys. Chem. B, Vol. 110, No. 3, 2006 1245
TABLE 3: Intramolecular Tishchenko Reaction of Various Aromatic 1,2-Dicarbaldehydes to Phthalides with mesoAl2O3/
SO42--III in Supercritical Carbon Dioxidea
a Reaction conditions are as follows: reactor, a 10-mL stainless steel autoclave; catalyst (mesoAl2O3/SO42--III activated at 773 K for 2 h under
1
vacuum (10-4 Torr)), 5 mg; reactant, 0.10 mmol; CO2 pressure, 10.0 ( 0.2 MPa; reaction temperature, 353 K. b Determined by H NMR, using
Ph3CH as an internal standard. c Tetrahydrofuran (0.1 mL) was used as a cosolvent.
Figure 3. Various types of adsorbed species between metal oxide
surface and a pyrrole molecule.15c
estimated by the shift of the NH stretching vibration (ν(NH))
band of the pyrrole hydrogen bonding with the base sites. This
method, however, cannot be applied to such stronger basic
oxides as alkaline earth oxides and alumina, because their base
sites, O2-, are so strong that they completely abstract an NH
proton of a pyrrole to give a pyrrolate ion, C4H4N-, and a
surface OH group. In such cases, the empirical rule found out
by Binet et al. on the shift of the ring-stretching (νR(ring)) bands
of adsorbed pyrrole is instructive.15c According to this rule, the
Figure 4. Infrared difference spectra between the catalysts before and
after the adsorption of pyrrole: (a) ring-stretching of a pyrrole ring in
two νR(ring) bands of B1,14 and B1,15 for the pyrrolate ion coor-
dinated on a Lewis acid site of a surface metal cation (designated
as PY-/M) appear at the lowest wavenumbers among the νR-
(ring) bands of adsorbed pyrrole species shown in Figure 3.
Thus, one can evaluate the number of strong base sites on a
basic oxide by the intensity of the νR(ring) bands of PY-/M.
pyrrolate ions, νR(ring); (b) CH in-plane deformation of pyrrole, δ-
(CH). Spectra A, B, and C correspond to CaO, ALO-4, and mesoAl2O3/
SO42--III, respectively.
employed only the B1,14 peaks for the discussion on the surface
basicity.
It should be noted that the addition of a small amount of
methanol obstructed the catalysis of mesoAl2O3/SO42--III (see,
Table 2, entry 7). This clearly demonstrates that on the surface
of mesoAl2O3/SO42--III exist a small number of strong base sites
that can promote the Tishchenko reaction even in the acid
medium of scCO2. All the previous reports on the Tishchenko
reaction have also demonstrated that the reaction requires strong
base sites on a solid catalyst to proceed.2,3
Analysis of the Catalyst Structures. The repeated measure-
ment by the infrared absorption method after combustion of
sulfated alumina samples in an induction furnace showed a
constant sulfur content: mesoAl2O3/SO42--I (0.23 wt %),
mesoAl2O3/SO42--II (0.48 wt %), and mesoAl2O3/SO42--III
(1.30 wt %). This proves that these catalysts are made of
aluminum oxides with SO42- ions homogeneously dispersed in
their frameworks.
Panels a and b in Figure 4 show the absorption bands due to
νR(ring), B1,14 of PY-/M species and the bands due to the CH
in-plane deformation, δ(CH), of adsorbed pyrrole, respectively.
The strong absorptions at 1369 and 1373 cm-1 can be attributed
to the νR(ring) of PY-/M species on CaO and ALO-4,
respectively,15c while no distinctive peak appears on mesoAl2O3/
SO42--III in the same range. The noticeable difference in the
intensity reveals that the aVerage base strength of O2- ions of
mesoAl2O3/SO42--III is weaker than that of ALO-4, implying
that not only the oxide framework arranged in the concave
2-
mesopores but the incorporated SO4 ions would cause some
reduction in the surface basicity of mesoAl2O3/SO42--III.
Compared with the B1,14 peaks, the corresponding B1,15 peaks
that would appear near 1450 cm-1 could not be clearly
confirmed because of the confusion with other peaks. We thus

1246 J. Phys. Chem. B, Vol. 110, No. 3, 2006
Seki and Onaka
Figure 5. XRD spectra of ALO-4 (A), ALO-2 (B), mesoAl2O3(#1)
(C), mesoAl2O3/SO42--I (D), and mesoAl2O3/SO42--III (E) in the 2θ
region of 1-10°.
Figure 7. Pore-size distribution curves calculated by Dollimore-Heal
theory.8b Solid line, mesoAl2O3(#1); dashed line, mesoAl2O3/SO42--I;
bold line, mesoAl2O3/SO42--III.
structure. Although the similar shift to a higher d spacing was
not observed with mesoAl2O3/SO42--I, the intensity of the peak
was weak and the peak was split compared with mesoAl2O3-
(#1), indicative of the partial collapse of mesoporous structure
by the bulk sulfur components.
The XRD patterns reflecting the bulk structures appeared in
the higher 2θ region of 10-70° (Figure 6). The peaks attributed
to the cubic spinel structure (400 and 440) were confirmed with
ALO-2 and ALO-4, indicating that the two aluminas are com-
posed of γ- and η-alumina (spectra A and B).16 It should be
noted that mesoAl2O3(#1), mesoAl2O3/SO42--I, and mesoAl2O3/
SO42--III also exhibited the 400 and 440 peaks (spectra C, D,
and E), although their intensities are quite low. It can be specu-
lated that mesoAl2O3(#1), mesoAl2O3/SO42--I, and mesoAl2O3/
SO42--III possess the cubic spinel structures considerably
distorted by the formation of regularly mesoporous frameworks.
The mesoporous structures were also characterized by the
N2 adsorption-desorption measurement employing the Dolli-
more-Heal (DH) theory,8b and the pore-size distribution curves
are given in Figure 7 for mesoAl2O3(#1), mesoAl2O3/SO42--I,
and mesoAl2O3/SO42--III. Although the pore-size distribution
of mesoAl2O3/SO42--III was somewhat broader than those of
mesoAl2O3(#1) and mesoAl2O3/SO42--I, the presence of sulfur
components hardly affected the pore-size distribution.
Figure 6. XRD spectra of ALO-4 (A), ALO-2 (B), mesoAl2O3(#1)
(C), mesoAl2O3/SO42--I (D), and mesoAl2O3/SO42--III (E) in the 2θ
region of 10-70°.
X-ray diffraction analyses of ALO-2, ALO-4, mesoAl2O3-
(#1), mesoAl2O3/SO42--I, and mesoAl2O3/SO42--III were per-
formed for elucidation of their porous and bulk structures, and
the results are shown in Figures 5 and 6.
2-
To further confirm whether the SO4 ions in mesoAl2O3/
For ALO-2 and ALO-4 possessing irregular pores, no intense
peak was observed in the low 2θ region of 1-10° (spectra A
and B in Figure 5), while mesoAl2O3(#1), mesoAl2O3/SO42--I,
and mesoAl2O3/SO42--III exhibited an intense peak at 1-2° that
is assignable to the mesoporous structure (spectra C, D, and E
in Figure 5).5 Davis et al. reported that the use of sodium
dodecylbenzenesulfonate was not appropriate as an anionic
surfactant for the synthesis of mesoporous alumina, because
traces of sulfur components remained in the material even after
calcination treatment and caused some collapse in the mesopore
structure.5 They showed that the partially collapsed mesostruc-
ture brought about a higher d spacing in XRD patterns, which
is peculiar to the mesoporous materials with randomly ordered
pores. The characteristic peak of mesoAl2O3/SO42--III appeared
at a higher d spacing than that of pure mesoAl2O3(#1), indicating
that the sulfur components in mesoAl2O3/SO42--III were
incorporated into the alumina framework to disturb the meso-
SO42--I and mesoAl2O3/SO42--III exist in the alumina frame-
works or on the surface, we measured the infrared spectra of
these catalysts that had been pressurized into thin disks and
activated at 773 K for 2 h under vacuum (10-4 Torr).
Considering that no distinctive absorption band due to the
covalent SdO bonds of SO42- species on the surface2a,10,17 was
observed (not shown here), we can deduce that the SO42- ions
are not on the surface but are incorporated into the alumina
framework, leading to weakening characteristic SdO stretching
vibration. It has been well accepted that surface SO42- species
with SdO bonds possessing high double bond character
remarkably increase the surface acidity of metal oxides and in
some cases convert the oxides into solid superacids, while bulk
SO42- ions with partial double bond character hardly influence
the surface acidity.2a,17 For instance, Bedilo and Klabunde
reported that surface SO42- species on ZrO2 remarkably increase
the surface acidity of ZrO2 and make the oxide into a solid

2-
Strong Base Catalysis of mesoAl2O3/SO4 in scCO2
J. Phys. Chem. B, Vol. 110, No. 3, 2006 1247
2-
Figure 8. Proposed framework structure of mesoAl2O3/SO4
.
SCHEME 2: Proposed Mechanism for the
Intramolecular Tishchenko Reaction of Phthalaldehyde
(1a) to Phthalide (1b) in ScCO2
Figure 9. Infrared difference spectra of mesoAl2O3/SO42--III between
before and after the adsoprtion of phthalaldehyde (1a).
can be assigned to the aryl-H stretching vibration of the adsorbed
species involving V and VI in Scheme 2. The bands at 2831
cm-1 and 2775 cm-1 and the band at 1693 cm-1 are assignable
to the C-H and CdO stretching vibration of the aldehyde
groups of V and VI, respectively. The bands at 2945 and 2883
cm-1 should be due to the C-H stretching vibration of the
-CH2- group of VI, while the band at 1381 cm-1 may be
attributed to the vibration of the -CH2-O- moiety.3d,19a The
band at 1520 cm-1 and its shoulder part around 1450 cm-1 are
assignable to the asymmetric and symmetric stretching vibration
of a carboxylate group of V.
Intermediate VI is playing a key role as an active species in
the catalytic Tishchenko reaction of phthalaldehyde as described
previously (Scheme 2).3d In scCO2 medium, the interaction of
dense CO2 molecules with the alumina surface should also affect
the catalysis. Apparently the prompt deactivation of a catalyst
observed especially for more basic ALO-3 and -4 is caused by
the strong adsorption of an acidic CO2 molecule onto an active
base site of O2-. For the Tishchenko reaction to take place,
tight adsorption of CO2 on alumina surface must be avoided
and the adsorbate-exchange reaction of II to IV on the base
site must proceed smoothly. This prospect has been realized
on ALO-2 and mesoAl2O3/SO42--III, whose average surface
2-
basicities were weakened by the contained SO4 ions for
superacid, while SO42- ions inside the ZrO2 framework scarcely
raise the surface acidity of ZrO2.6
2-
ALO-2 and by both SO4 and the mesoporous structure for
mesoAl2O3/SO42--III.
We suppose that the SO4 ions in mesoAl2O3/SO42--I and
2-
As described above, the deactivation of a catalyst in the
Tishchenko reaction is likely to depend on the relative stability
between adsorbed species of II and IV in Scheme 2. Over the
more basic alumina, stronger adsorption of CO2 at a base site
is realized and the state of II is more stabilized with the base
site occupied. In contrast, on less basic and more acidic
mesoAl2O3/SO42--III due to the electron-withdrawing effect of
SO42- ions, an adsorbed CO2 molecule should be less stabilized
and a strong acid site of an adjacent Al cation tends to coordinate
with an oxygen atom of the adsorbed CO2 to displace the CO2
molecule from the base site to the aluminum cation as shown
in Scheme 3. Therefore, the base site is regenerated. Such a
dynamic migration of CO2 on an oxide surface has been
observed even over a strong solid base, CaO, whose acid
strength is far lower than that of alumina.20 Moreover, Parkyns
confirmed the formation of X on γ-alumina in which only an
oxygen atom of CO2 participates in the coordination based on
the bands at 1820 and 1780 cm-1 with infrared spectroscopy.21
mesoAl2O3/SO42--III slightly suppress the surface basicity of
the mesoporous aluminas.
On the basis of the above XRD, pore-size distribution, and
2-
IR data, we imagine the surface structures of mesoAl2O3/SO4
-
I and -III as depicted in Figure 8.
Mechanistic Consideration. It has been established by
previous studies with IR, XPS, and UPS techniques that the
Cannizzaro-type disproportionation of nonenolizable aldehydes
such as benzaldehyde and formaldehyde takes place over basic
metal oxides such as MgO and alumina.18,19 We have also
proved through an IR study3d that the adsorption of 1a onto an
acid site (Al3+) and a base site (O2-) on γ-alumina produces
Cannizzaro-type products, i.e., a reduced product, o-AlOCH2C6H4-
CHO (VI) and an oxidized product, o-AlOCOC6H4CHO (V),
where Al is a surface aluminum cation in Scheme 2.
Figure 9 shows the infrared spectrum of 1a adsorbed on
mesoAl2O3/SO42--III. The strong absorption band at 3005 cm-1

1248 J. Phys. Chem. B, Vol. 110, No. 3, 2006
Seki and Onaka
SCHEME 3: A Migration of Adsorbed CO2 from a
Strong Base Site (O2-) to an Adjacent Strong Acid Site
(Al3+) through an Adsorbate-Exchange Reaction
The following two roles of THF as cosolvent are considered
for the improvement in the reaction rate: (i) enhancement of
the mass transfer of the reactant and the product into mesopores
by improving their solubilities in scCO2, and (ii) prevention of
strong CO2 adsorption onto an active base site by the formation
of THF-CO2 interaction (Lewis base and acid interaction).
References and Notes
(1) (a) Noyori, R., Ed. Supercritical Fluids. Chem. ReV. 1999, 99, 353.
(b) Chemical Synthesis Using Supercritical Fluids; Jessop, P. G., Leitner,
W., Eds.; Wiley-VCH: Weinheim, Germany, 1999. (c) Noyori, R.; Ikariya,
T. Supercritical Fluids for Organic Synthesis. In Stimulating Concepts in
Chemistry; Vo¨gtle, F., Stoddart, J. F., Shibasaki, M., Eds.; Wiley-VCH:
Weinheim, Germany, 2000; p 13. (d) Tanchoux, N.; Leitner, W. In
Handbook of Green Chemistry and Technology; Clark, J., Macquarrie, D.,
Eds.; Blackwell Science: Oxford, UK, 2002; p 482. (e) Savage, P. E.;
Gopalan, S.; Mizan, T. I.; Martino, C. J.; Brock, E. E. AIChE J. 1995, 41,
1723. (f) Grunwaldt, J.-D.; Wandeler, R.; Baiker, A. Catal. ReV.-Sci. Eng.
2003, 45, 1.
(2) (a) Tanabe, K.; Misono, M.; Ono, Y.; Hattori, H. New Solid Acids
and Bases; Kodansha-Elsevier: Tokyo, Japan, 1989. (b) Hattori, H. Chem.
ReV. 1995, 95, 537.
(3) (a) Tanabe, K.; Saito, K. J. Catal. 1974, 35, 247. (b) Seki, T.;
Akutsu, K.; Hattori, H. Chem. Commun. 2001, 1000. (c) Seki, T.; Hattori,
H. Chem. Commun. 2001, 2510. (d) Seki, T.; Tachikawa, H.; Yamada, T.;
Hattori, H. J. Catal. 2003, 217, 117. (e) Tsuji, H.; Hattori, H. ChemPhy-
sChem 2004, 5, 733.
In addition to the CO2 adsorption on the active base sites,
another cause should be taken into account to explain why the
conversion of 1a to 1b was incomplete and why the two-
stepwise increment was observed in the yield of 1b in the
mesoAl2O3(#2)- and mesoAl2O3/SO42--III-catalyzed reactions
in Figure 1 (see, Kinetic Study). It can be speculated that a
catalytically active species of VI would be oxidized to an
inactive species of V along with the conversion of VI to VII as
shown in Scheme 2. In fact, Kno¨zinger et al.,22 Tanabe and
Saito,3a and we3d have previously observed by infrared spec-
troscopy that the oxidation of surface alkoxide species to the
corresponding carboxylate takes place smoothly over alumina.
Tanabe and Saito concluded that this oxidation, i.e., the
oxidation of aluminum benzylate to aluminum benzoate, is
responsible for the low activity of alumina catalyst for the
intermolecular Tishchenko reaction of benzaldehyde.3a
(4) Seki, T.; Onaka, M. Chem. Lett. 2005, 34, 262.
(5) Vaudry, F.; Khodabandeh, S.; Davis, M. E. Chem. Mater. 1996, 8,
1451.
(6) Bedilo, A. F.; Klabunde, K. J. J. Catal. 1998, 176, 448.
(7) Farooq, O. Synthesis 1994, 1035.
(8) (a) Brunauer, S.; Emmett, P. H.; Teller, E. J. Am. Chem. Soc. 1938,
60, 309. (b) Dollimore, D.; Heal, G. R. J. Appl. Chem. 1964, 14, 109.
(9) (a) Bergens, S. H.; Fairlie, D. P.; Bosnich, B. Organometallics 1990,
9, 566. (b) Kraus, G. A.; Pezzanite, J. O.; Sugimmoto, H. Tetrahedron Lett.
1979, 20, 853. See also ref 9a. (c) Padwa, A.; Bullock, W. H.; Norman, B.
H.; Perumattam, J. J. Org. Chem. 1991, 56, 4252. (d) Donati, C.; Prager,
R. H.; Weber, B. Aust. J. Chem. 1989, 42, 787.
Conclusions
It was demonstrated for the first time that scCO2, an ideal
reaction medium in view of green chemistry, can be applied to
solid strong base-catalyzed reactions. The mesoporous alumina
with SO42- incorporated in the alumina framework (mesoAl2O3/
SO42-) exhibited highly effective strong base catalysis for the
Tishchenko reaction in scCO2. Especially, when scCO2 was used
together with a small amount of THF as a cosolvent, remarkable
acceleration in the reaction rate was achieved; the reaction in
the scCO2-THF medium proceeded 1.5-fold faster than those
in conventional organic solvents such as benzene and THF.
(10) Arata, K. AdV. Catal. 1990, 37, 165.
(11) Paul, M. A.; Long, F. A. Chem. ReV. 1957, 57, 1.
(12) Gilgen, R.; Kleinrahm, R.; Wagner, W. J. Chem. Thermodyn. 1992,
24, 1243.
(13) Xiao, J.-L.; Nefkens, S. C. A.; Jessop, P. G.; Ikariya, T.; Noyori,
R. Tetrahedron Lett. 1996, 37, 2813.
(14) Carroll, W. A.; Altenbach, R. J.; Bai, H.; Brioni, J. D.; Brune, M.
E.; Buckner, S. A.; Cassidy, C.; Chen, Y.; Coghlan, M. J.; Daza, A. V.;
Drizin, I.; Fey, T. A.; Fitzgerald, M.; Gopalakrishnan, M.; Gregg, R. J.;
Henry, R. F.; Holladay, M. W.; King, L. L.; Kort, M. E.; Kym, P. R.;
Milicic, I.; Tang, R.; Turner, S. C.; Whiteaker, K. L.; Yi, L.; Zhang, H.;
Sullivan, J. P. J. Med. Chem. 2004, 47, 3163.
(15) (a) Scokart, P. O.; Rouxhet, P. G. J. Chem. Soc., Faraday Trans.
I 1980, 76, 1476. (b) Barthomeuf, D. J. Phys. Chem. 1984, 88, 42. (c) Binet,
C.; Jadi, A.; Lamotte, J.; Lavalley, J. C. J. Chem. Soc., Faraday Trans.
1996, 92, 123.
The high activity of mesoAl2O3/SO42--III for the Tishchenko
reaction in scCO2 medium is attributed to the following three
key factors: (i) aluminum atoms intrinsically do not form stable
2-
carbonate salts; (ii) SO4 ions incorporated in the alumina
framework and the specific mesoporous structure of mesoAl2O3/
SO42--III decrease the aVerage surface basicity to a certain
extent, but there exists a small number of strong base sites on
its surface that can abstract a proton from a very weak Bro¨nsted
acid of methanol and promote the Tishchenko reaction in scCO2;
and (iii) a Lewis acid site (Al3+) adjacent to an active strong
base site (O2-) enhances the adsorbate-exchange reaction
between CO2 and a reactant aldehyde on the base site: the Al3+
site is coordinated with an oxygen atom of CO2 adsorbed on
the base site as shown in IX in Scheme 3, and the resulting
coordination state of IX becomes unstable to give X with a
free base site.
(16) (a) Yamaguchi, G.; Yanagida, H. Bull. Chem. Soc. Jpn. 1962, 35,
1896. (b) Yanagida, H.; Yamaguchi, G. Bull. Chem. Soc. Jpn. 1964, 37,
1229. (c) Yanagida, H.; Yamaguchi, H.; Kubota, J. Bull. Chem. Soc. Jpn.
1965, 38, 2194.
(17) Jin, T.; Yamaguchi, T.; Tanabe, K. J. Phys. Chem. 1986, 90, 4794.
(18) Cannizzaro reactions over MgO, see: Peng, X. D.; Barteau, M. A.
Langmuir 1989, 5, 1051. See also refs 3a and 3d.
(19) Cannizzaro reactions over alumina, see: (a) Kuiper, A. E. T.;
Medema, J.; Van Bokhoven, J. J. G. M. J. Catal. 1973, 29, 40. (b) Niwa,
M.; Inagaki, S.; Murakami, Y. J. Phys. Chem. 1985, 89, 2550. See also ref
3d.
(20) Tsuji, H.; Okamura-Yoshida, A.; Shishido, T.; Hattori, H. Langmuir
2003, 19, 8793.
(21) Parkyns, N. D. J. Chem. Soc. A 1969, 410.
(22) Kno¨zinger, H.; Bu¨hl, H.; Ress, E. J. Catal. 1968, 12, 121.