Oil formation from glucose with formic acid and cobalt catalyst in hot-compressed water

Article abstract of DOI:10.1016/j.carres.2006.10.011

Liquefaction of glucose into oil was examined in hot-compressed water at 300 °C and 30 or 60 min in a tumbling batch reactor. The effects of alkali (KHCO₃), a hydrogenating agent (HCO₂H), and a cobalt catalyst (Co₃O₄) were studied. Also the combinations of these additives were investigated. HCO₂H and KHCO₃ showed a positive effect on oil formation. Co₃O₄ was found to be an advantageous additive as well, increasing the oil formation from glucose, but the stability of this catalyst under reaction conditions was quite low.

Full text of DOI:10.1016/j.carres.2006.10.011

Carbohydrate Research 341 (2006) 2891–2900
Oil formation from glucose with formic acid and cobalt catalyst
in hot-compressed water
Masaru Watanabe, Florian Bayer and Andrea Kruse*
ITC-CPV, Forschungszentrum Karlsruhe (FZK), PO Box 3640, 76021 Karlsruhe, Germany
Received 24 April 2006; received in revised form 5 October 2006; accepted 9 October 2006
Available online 14 October 2006
Abstract—Liquefaction of glucose into oil was examined in hot-compressed water at 300 °C and 30 or 60 min in a tumbling batch
reactor. The effects of alkali (KHCO
binations of these additives were investigated. HCO
3
), a hydrogenating agent (HCO
H and KHCO
2
H), and a cobalt catalyst (Co
3 4
showed a positive effect on oil formation. Co O
3
O
4
) were studied. Also the com-
2
3
was found to
be an advantageous additive as well, increasing the oil formation from glucose, but the stability of this catalyst under reaction
conditions was quite low.
Ó 2006 Elsevier Ltd. All rights reserved.
Keywords: Liquefaction; Biomass; Glucose; Formic acid; Cobalt; Co
3
O
4
; Alkali; Potassium hydrogencarbonate; Hot-compressed water
1
. Introduction
tion has been widely studied, and several processes for
the gasification of biomass of different water content
have been reported.
1
Biomass refinery processes must be developed in the
next few decades because only biomass is a sustainable
carbon resource, and its use is required to replace petro-
leum and other fossil fuels. These changes are necessary
to develop a sustainable society and to reduce global
Although biomass liquefaction has several advantages
such as less transport effort, and thus, in effect, the pres-
ervation of biomass, the state-of-the-art of liquefaction
processes for biomass, in particular for biomass with
high water content, are still insufficiently well developed.
Living plants contain more than 80 wt % of water. Bio-
mass feedstocks with such high water content are usu-
ally called ‘wet biomass’ and must be dried for further
processing. Dry biomass that includes only few percent
of water can be converted into a bio-oil by flash pyroly-
sis; however, flash pyrolysis is not useful for wet biomass
with water content of several tens of percent because the
vaporization of water requires much energy. The most
popular process of wet biomass liquefaction (biomass
conversion into liquid fuel) is ethanol production by fer-
mentation, which has a separation step in which water is
removed from the products (i.e., an ethanol–water mix-
ture). This separation demands much energy and leads
to a loss of total energy efficiency, much like that
encountered in the drying of biomass.
warming caused by CO emissions. Possible processes
2
for getting energy from biomass are both direct and
indirect gasification. The latter is a two-step process of
liquefaction of biomass following gasification of the lique-
faction products, as, for example, ethanol or bio-oil.
Advantages of indirect gasification are the avoiding of
the transport of biomass along large distances to large
gasification facilities as well as fewer problems because
of using ‘clean fuels’ for gasification. Salts, especially,
can cause several problems during gasification, and
bio-oil contains almost no salt. Direct biomass gasifica-
*
Corresponding author. Tel.: +49 7247 823388; fax: +49 7247
22244; e-mail: andrea.kruse@itc-cpv.fzk.de
Permanent address: Research Center of Supercritical Fluid Techno-
logy, Tohoku University, 4-4-11-404, Aoba, Aramaki, Aoba-ku,
Sendai 980-8579, Japan.
8
An alternative process for wet biomass liquefaction is
to convert biomass into oil in the presence of water. To
0
008-6215/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carres.2006.10.011

2892
M. Watanabe et al. / Carbohydrate Research 341 (2006) 2891–2900
avoid evaporation an increased pressure is necessary.
The oil should have a low oxygen content and should
be liquid under ambient conditions, because in this case,
the hydrophobic oil spontaneously separates from
water. The best-known method to concentrate wet bio-
mass into oil is the so-called ‘hydrothermal upgrading’
or HTU process originally developed at Shell Research
B.V. (Amsterdam). At 300–350 °C and 15–20 MPa, the
However, we considered that the effective progress of
the D/H process has an inconsistency. That is, to pro-
gress, the D/H process, both the dehydration and the
retro-aldol condensation of sugar must occur, while
the control of these two reactions is difficult because
the best conditions for one of these reactions are detri-
1
4
mental to the other. Sasaki et al. reported that the
retro-aldol condensation of glucose is favored at higher
temperature and lower pressure, and the dehydration of
glucose is preferred at lower temperature and higher
pressure. The other reports show that the retro-aldol
biomass reacts into gaseous products and oil within
2
5
–15 min.
To get oil from biomass, elimination of oxygen atoms
1
5,16
from biomass is required because biomass such as cellu-
lose (C H O ) has a high oxygen content. One way to
condensation is increased under basic conditions,
and that the dehydration is promoted under acidic
6
10 5 n
1
6–19
remove oxygen from an organic compound is dehydra-
conditions.
tion (–H O), with the resultant formation of double
We herein suggest a new concept for oil liquefaction
of cellulosic biomass (C H O ) to solve these inconsis-
2
bonds. However, the dehydration process also causes a
depletion of hydrogen in the organic compound (H/
C < 1) by formation of unsaturated compounds. The
double bonds may polymerize leading to high-molecu-
lar-weight compounds and coke formation. The other
ways of the oxygen elimination are decarbonylation
6
10 5 n
tencies: the cellulosic material decomposes into a mono-
mer unit of biomass such as glucose and then further
converts into small fragments without loss of water via
2
0–22
the retro-aldol condensation.
Finally the small
pieces having C–O or C@O bonds such as alcohols,
acids, aldehydes, and ketones must be connected by los-
ing an oxygen atom to form hydrocarbons. So far, it has
been revealed that (1) the hydrolysis of cellulosic bio-
mass to a monomer unit such as glucose rapidly occurs
without catalyst in hot-compressed water, and (2) glu-
cose further decomposes into a small aldehyde and acids
under alkali conditions. What must be clarified is how to
couple these small C–O or C@O compounds to alkane-
(
–CO) and decarboxylation (–CO ). Under decarboxyl-
2
ation two oxygen atoms with one carbon are removed
from the feedstock, and under decarbonylation only
one oxygen atom per carbon atom is removed. Previ-
ously, other researchers had reported that (1) the oil liq-
2
–11
uefaction of biomass was examined at 200–350 °C,
(
2) the oil liquefaction is sensitive for the concentration
9
,11
of biomass,
and (3) it is enhanced by alkali cata-
The main reason for the enhancement of
2
–8,10,11
lysts.
like hydrocarbons (–CH –). One of the possibilities is to
2
the oil liquefaction by alkali was assumed to be increas-
ing decarboxylation. However, the crude oil from bio-
use a Fischer–Tropsch (FT) type reaction, because the
FT reaction enables C„O, C@O, and C–O compounds
to combine into the alkane-like hydrocarbons. To en-
hance the FT type reaction, a catalyst is needed. Effec-
tive catalysts for the FT type reaction are well known
to be Fe, Ni, Co, and Ru catalysts. An Fe catalyst is inex-
pensive and has a high activity for the water–gas shift
reaction. The active species on Fe catalyst for the FT
1
,7,8
mass is mainly aromatic (C/H ꢀ 1),
and thus it has
high viscosity.
1
2
Recently, Huber et al. reported that a biomass-
derived compound was converted into alkane-like oil
through a new oxygen atom removal reaction, a dehy-
dration/hydrogenation process (D/H process) in hot-
compressed water around 250 °C. Here, the dehydration
of a sugar such as glucose must first be promoted in the
direction of furan compound formation. The furan com-
pounds then couple with small carbonyl compounds to
produce, via retro-aldol condensation of the sugar, a
furan with a side chain. Thus, to form the side chain,
the retro-aldol condensation must also occur. Finally,
the double bonds of the furan with the side chain react
with hydrogen to form single bonds. By this way the
aromatic furan ring is opened to form a straight-chain
alkane. This new concept is quite reasonable because
type reaction, FeC , is well known to be deactivated
x
by steam. A Ni catalyst has been used to gasify biomass
in hot-compressed water around 200–400 °C, and it was
2
3–25
found to be useful to produce CH or H .
Thus, the
4
2
candidates for the FT type reaction for making oil con-
sisting of long-chain alkanes in hot-compressed water at
200–400 °C are Co- and Ru-based catalysts. A Ru cata-
lyst was employed to gasify biomass in hot-compressed
2
6–29
water and was shown to be effective above 350 °C.
Some researchers have studied the effect of water on
the CO reaction in the presence of Co catalyst (the FT
3
0
31
both reactions, the dehydration (–H O) and the hydro-
type reaction and CO oxidation ). From these studies
it was revealed that the water reduces the catalytic activ-
2
genation (+H ), are employed to get alkane-like (H/
2
3
0
31
C > 2) oil from carbohydrate (C:H:O=1:2:1). Thus, to
get good quality oil, hydrogenation is needed. Formic
acid is known to be an active hydrogenation agent in
ity (methanation and CO oxidation ), while water
3
0
improves C–C propagation on the FT type reaction.
3
2
Ding et al. reported that a metal oxide of Co might
be used as a catalyst in hot-compressed water media
because of its high stability in this medium. Thus, Co
1
3
hot-compressed water, and it is applied to improve
the oil formation in this study.

M. Watanabe et al. / Carbohydrate Research 341 (2006) 2891–2900
2893
is expected to support the formation of C–C bonds from
C@O and C–O compounds in hot-compressed water.
The FT type reaction also requires hydrogen. As stated
above, formic acid is an active hydrogenating compound
in hot-compressed water, and it is expected to improve
oil formation from biomass.
the experiment was continued for 60 min at the reaction
temperature. Table 1 shows the experimental conditions.
The symbol Y in Table 1 indicates the additive (HCO H,
2
Co O , and KHCO ) that was added in the experiment.
3
4
3
1
3
The compositions 1, 2, and 5–7 are the experiments per-
formed without the hydrogenating agent (HCO H).
2
In this study, cellulosic biomass liquefaction to oil was
examined. Cellulosic material was found to rapidly
decompose into its monomer and oligomers in hot-
compressed water. Thus, to gather basic knowledge,
glucose was employed as a model compound to study the
These experiments were conducted to study the effect
of KHCO and Co O on the oil formation. The compo-
sitions 3, 4, and 8–10 are the experiments performed in
3
3
4
3
3
the presence of HCO H to understand the effect of
2
HCO H on the oil formation. The compositions 1 and
2
effect of reaction time, formic acid (HCO H), a co-cata-
3 are experiments performed at 30 min reaction time
at the final temperature. Table 1 also lists the values
of pressure inside the reactor at the reaction tem-
perature (300 °C) and the final reaction time (30 or
60 min).
2
lyst (Co O ), and alkali (KHCO ) on oil formation.
3
4
3
2. Experimental
After cooling, the amount of gases formed was mea-
sured by a gas meter. The water solution was forced
out of the reactor by compressed air. Some researchers
have used acetone for recovery of the oil because the
oil from biomass is sometimes defined as acetone-solu-
ble. Therefore, we also used acetone as the solvent
to recover the produced oil in this study. Acetone was
loaded into the reactor, and the reactor was tumbled
for a few minutes to dissolve the formed oil. The ace-
tone–oil solution was also forced out of the reactor by
compressed air. The aqueous solution was filtered, and
samples of the gaseous, liquid, and solid phases were
analyzed. The experimental error, here mainly deter-
mined by reproducibility, was roughly 1% of the given
values.
All reagents and catalysts were obtained from E. Merck:
glucose monohydrate (starting material); formic acid
(
used without further purification); potassium hydrogen-
carbonate (used as alkali); and Co O (the metal oxide
3
4
7
,9
catalyst).
The experiments were conducted in an Inconel 625-
3
4
lined, tumbling batch autoclave. Glucose (20 g) and
a catalyst [2 g of KHCO (2 g) and/or a solid catalyst
3
(
1 g)] were carefully mixed, and the feedstock was
adjusted to a total of 200 g by adding pure water. For
HCO H addition, HCO H (2 g) was poured into the
2
2
glucose solution. The internal volume of the autoclave
was 1 L. The maximum pressure and temperature of
the reactor were 50 MPa at 500 °C. The reactor was
heated by a heating cartridge, and the control of tem-
perature was within ± 5 °C. A tube in which three ther-
mocouples were fastened for temperature measurement
was located inside the reactor. Loading of the solution
in the reactor was achieved by another tube. The heat-
ing rates and the final temperature were adjusted by a
temperature controller connected with the thermocou-
ples, and the pressure was measured by an analog
manometer. The signals were continuously recorded
by a six-channel recorder (Logoprint 100, Jumo Mess-
und Regeltechnik M. K. Juchheim GmbH & Co.). A
disadvantage of this large reactor is the long time
needed for heating up and cooling down. Advantages
are the relative large product samples, enabling a large
number of analytical samples to be taken and the possi-
bility of mixing. The latter two properties are the rea-
sons for why this reactor was chosen to explore these
new research fields.
Gas analysis was conducted by two gas chromato-
graphs. To measure the H content, an HP 5889A gas
2
chromatograph was used. The carrier gas was nitrogen,
the column was a Porapak Q column (80/100 Porapak Q
by Supelco, 1.83 m in length), and the detector was a
thermal conductivity detector. Analysis of all other
gases was carried out by an HP 6890 instrument with
column switching: the first column was an 80/100 Haye-
sep Q (2 m long, Resteck), and the second column was a
Table 1. Experimental conditions
b
Composition t (min) HCO
2
H
KHCO
3
Co
3
O
4
p (MPa)
1
2
3
4
5
6
7
8
9
30
60
30
60
60
60
60
60
60
60
—
—
Y
—
—
—
—
Y
—
Y
Y
—
—
—
—
—
Y
Y
—
Y
8.47
8.53
8.74
8.89
8.48
8.61
8.6
8.74
8.8
8.89
a
Y
The glucose or glucose–catalyst solution or suspen-
sion was loaded in the evacuated autoclave through a
valve. After the mixture was loaded, the autoclave was
—
—
—
Y
Y
Y
purged by N to remove air in the reactor. Then the tem-
2
—
Y
perature controller was adjusted to the desired heating
rate (3 °C/min) and to the final temperature (300 °C).
After the reactor had reached the final temperature,
10
Y
a
Y = additive present.
p = pressure at the reaction temperature and the final reaction time.
b

2894
M. Watanabe et al. / Carbohydrate Research 341 (2006) 2891–2900
˚
6
0/80 Molesieve 5 A (4 m long, Resteck). The two
Loss of Co O ; wt %
3 4
columns, as well as a thermal conductivity and a flame
ionization detector, were connected in series. The second
column was bypassed by a six port valve for analysis
Co in the recovered solution; g
¼
ꢁ 100
ð6Þ
4
Co in the starting Co O ; g
3
of CO and hydrocarbons. The carrier gas of the HP
2
The products analyzed in the aqueous solution were
phenols, 5-hydroxymethyl-2-furaldehyde (HMF), fural-
dehyde (FU), 2-methylfuraldehyde (MF), acetic acid,
and glycolic acid.
6
890 was helium. Determination of the residual total
organic carbon (TOC) content in the liquid effluent
was performed by a commercial TOC analyzing method
(
Hach Lange). The total amounts of phenolic com-
pounds (which are called ‘phenols’ in this study) were
quantified by a UV–vis spectrometer (Cadas 200 spec-
trophotometer, Hach Lange). Two different SPME-GC
applications were used for the qualitative and quantita-
3. Results and discussion
3.1. General observations
3
4
tive determination of organic compounds.
The
amounts of organic acids and furfurals were measured
by LC with a UV–vis detector (L-4250, E. Merck).
The columns attached to the LC were an Aminex
TMHPX-87 H column (Bio-Rad) for organic acids
and an RP-18 column (E. Merck) for furfurals. The
weight of oil was measured after evaporating the ace-
tone and water. The residue was analyzed by a commer-
cial CHNOS elemental analyzer (Vario EL III,
Elementar). Although the precision of the analyzer
detector is 0.1% (max.) relative, we learned by experi-
ence that the accuracy of the results is only 15%
The experimental error, here mainly determined by
reproducibility, is roughly 5% of the given values (except
the oil composition). The pH values shown in Table 2
were measured after the reaction at room temperature.
The gaseous products were mainly CO, CO2 and
H . The quantified products in the recovered water solu-
2
tion were phenols, HMF, FU, methanol, ethanol, acetic
acid, and glycolic acid. The yields of FU, methanol, and
ethanol were always low and are therefore not reported
here. In the presence of HCO H, CO , and H were
2
2
2
(
max.) relative. The reasons are the small sample weight
produced from the HCO H. A small amount of CO
2
of 2 mg, problems faced during sampling, inhomogeni-
ties of solid materials, etc.
may be formed; however, the decarboxylation of
3
5,36
HCO H is predominant in hot-compressed water.
2
During evaporation, polymerization usually occurs,
changing the oil to a black, oily-looking solid. This poly-
merization should not change the CHO ratio, but a fur-
ther identification in view of the identification of single
compounds is not possible.
The conversion of HCO H was always >90%. The yields
2
of H (Eq. 1) and CO (Eq. 2) from HCO H were both
2
2
2
about 7% at 100% conversion of HCO H. The differ-
2
ences in the yield of CO with HCO H compared to that
2
2
without HCO H under the same experimental condi-
2
The definition of each yield is as follows:
tions (compositions 1 and 3, compositions 2 and 4, com-
positions 5 and 8, compositions 6 and 9, compositions 7
and 10) were always 5–6% and reasonable. However, the
differences of the H2 yield were always <4%. Thus,
consumption of H produced from HCO H occurred
H
2
yield; mol %
ꢁ ðProduced H
H atom in the starting glucose; mole
2
2
; moleÞ
2
2
¼
ꢁ 100
ð1Þ
ð2Þ
because H and CO should be formed in the same
2
2
amount from HCO H. So this is a clear indication that
2
CO and CO
2
yield; mol %
CO or CO ; mole
C atom in the starting glucose; mole
HCO H is a hydrogenating agent.
2
2
¼
ꢁ 100
Table 2. Additional experimental results
TOC yield; mol %
a
Composition
pH
Phenols (wt %)
HMF (wt %)
C in the water solution; mole
1
2
3
4
5
6
7
8
9
3.20
3.70
3.60
3.70
4.50
4.10
5.00
4.80
4.40
5.10
0.13
0.15
0.12
0.14
0.20
0.16
0.17
0.17
0.18
0.18
0.00
0.08
0.35
0.16
0.03
0.07
0.03
0.11
0.10
0.07
¼
ꢁ 100
ð3Þ
ð4Þ
C atom in the starting glucose; mole
oil; g
Oil yield; wt % ¼
ꢁ 100
the starting glucose; g
Products yield in water solutions; wt %
Products; g
1
0
¼
ꢁ 100
ð5Þ
a
pH at room temperature after the reaction.
the starting glucose; g

M. Watanabe et al. / Carbohydrate Research 341 (2006) 2891–2900
.2. Time profile of the reaction with and without HCO H
Small fragments
2
2895
3
Gas
C-C splitting
(
aldehyde, acid)
To study the reaction course briefly, the time profile of
O
O
the reaction was examined with and without HCO H.
2
O
O
O
Figure 1 shows the time profile of the yields of CO2,
TOC, and oil without HCO H. The yields of CO
O
Oil
Tar/Char
2
2
increased with the increase in reaction time. In contrast,
the oil yield decreased with time. Figure 2 displays the
time profile about the yields of acetic acid and glycolic
acid because the yields of these organic acids are
remarkably high compared with the other products.
The yield of acetic acid, as well as the yields of oil
Dehydration
Ring compounds
Furans, Phenols)
(
Figure 3. Schematic diagram of glucose reaction in hot-compressed
water.
(
Fig. 1) also decreased with an increase in time. Acetic
bined with a small fragment (aldehyde and acid)
2
2
acid is well known to be stable even in supercritical
through condensation to form oil. The oil produced
polymerizes easily to form an acetone-insoluble tar
and charred mass.
2
9
water. Thus, the conversion of acetic acid must be in-
creased because of the coupling reaction with a reactive
species, not by acceleration of the self-decomposition
into CO and CH . Thus, without an additive, a long-
Figure 4 shows the yield of CO , H , TOC, and oil in
2
2
2
4
the presence of HCO H. As mentioned above, by adding
2
time reaction is not suitable for oil formation. The for-
mation of phenols is not significantly influenced by the
presence of formic acid or the other additives. The
HMF content is increased in the presence of formic acid
HCO H, the yield of H increased (see Fig. 7 compared
2
2
with Fig. 11); however, the increase of H does not agree
2
with the amount of H formed by the conversion of
2
HCO H. This means that hydrogenation might progress
2
(
Table 2). A brief reaction scheme for glucose is shown
in the presence of HCO H. The oil yield was slightly
2
in Figure 3. Glucose decomposes into small fragments
through C–C splitting and into a ring compound
higher than that without HCO H, and the oil yield
was not reduced as a function of time by adding
2
2
0,21
through dehydration.
A ring compound is then com-
HCO H. Figure 5 shows the yields of acetic acid and
2
glycolic acid. The yield of the acids (acetic acid and
3
2
2
1
1
0
5
0
5
0
3
5
TOC
TOC
Oil
30
25
20
Oil
1
1
5
0
5
0
5
0
CO2
CO2
0
20
40
60
80
0
20
40
60
80
Time (min)
Time (min)
2
Figure 1. Time profile of the yields of CO (mol%), TOC (mol%), and
oil (wt%) in the absence of additive.
2
Figure 4. Time profile of the yields of CO (mol%), TOC (mol%), and
2
oil (wt%) in the presence of HCO H.
5
5
4
4
Glycolic acid
Glycolic acid
Acetic acid
3
2
3
2
1
0
1
0
Acetic acid
0
20
40
Time (min)
60
80
0
20
40
60
80
Time (min)
Figure 2. Time profile of the yield of the organic acids in the absence of
additive.
Figure 5. Time profile of the yield of the organic acids in the presence
of HCO H.
2

2896
M. Watanabe et al. / Carbohydrate Research 341 (2006) 2891–2900
Small fragments
40
30
Gas
Hydrogenation
(
aldehyde, acid)
O
O
(
HCO H)
2
O
O
O
O
Oil
Tar/Char
2
1
0
Ring compounds
Furans, Phenols)
Hydrogenation
(
0
0
(
HCO H)
2
2
Figure 6. Effect of HCO H on glucose reaction in hot-compressed
water.
2
5
6
7
Composition
Figure 8. Yield of TOC in the absence of HCO H.
glycolic acid) decreased with time in the presence of
2
HCO H. By the analogy of the above discussion, the
2
decrease of the acid yields shows that the reactions
between the acids and the oil produced (or the ring
compounds), such as polymerization, are still progress-
1
2
Glycolic acid
Acetic acid
ing in the presence of HCO H. The rate of oil formation
10
8
2
possibly compensates for that of oil polymerization, and
thus the yield of oil was constant regardless of the reac-
tion time. The glucose reaction in the presence of
6
HCO H might be considered as shown in Figure 6. Be-
2
cause of the scavenging of active species with hydrogen
4
and the inhibition of polymerization, adding HCO H
2
2
was slightly effective in making oil from glucose in
hot-compressed water. However, the yield of oil was still
not high enough, because the oil consumption to form
tar or char would still occur.
0
2
5
6
7
Composition
Figure 9. Yield of liquid products in the absence of HCO
2
H.
3
absence of HCO H
.3. Effect of KHCO and Co O on the reaction in the
3 3 4
2
2
1
0
5
We examined the effect of KHCO and Co O on the
3
3
4
glucose reaction without HCO H. Figures 7–10 show
2
the yields of gaseous products, of TOC, of some liquid
products in the recovered water solution, and of oil,
respectively. As shown in Figure 7, the CO yield was
2
10
5
enhanced by adding KHCO (composition 5). In the
3
presence of Co O , both CO and H yields were also
3
4
2
2
increased (composition 6, Fig. 7). By adding both,
KHCO and Co O (composition 7), a further increase
3
3
4
0
2
5
6
7
Composition
1
1
5
H
CO
CO
2
2
Figure 10. Yield of oil in the absence of HCO H.
2
0
5
0
of CO and H yields was observed (Fig. 7). The trend
of TOC yield depending on the additives (Fig. 8) was
2
2
similar to that of the CO yield (Fig. 7). From Figure
2
9
, it can be revealed that KHCO promoted the forma-
3
tion of acetic acid, and Co O supported forming both
3
4
acetic and glycolic acid. By addition of both, KHCO3
and Co O , the yields of acetic acid and glycolic acid
were further enhanced. The effect of additives on the
oil formation (Fig. 10) is similar to that on the TOC
2
5
6
7
3
4
Composition
Figure 7. Yield of gaseous compounds in the absence of HCO
2
H.

M. Watanabe et al. / Carbohydrate Research 341 (2006) 2891–2900
2897
yield: each addition of KHCO or Co O had a positive
50
40
3
3
4
impact on the TOC (Fig. 8) and oil (Fig. 10) yields. Both
KHCO and Co O addition lead to a further increase
3
3
4
of the yield of TOC and oil, as shown in Figures 8
and 10. In this case, the maximum yield of oil was
3
2
1
0
0
0
0
1
5.4 g/g% (composition 7).
In the presence of KHCO , the CO and the acid
3
2
formation improved and the yields of TOC and oil were
enhanced. It was proposed that the addition of alkali
promotes the formation as well as the decomposition
of the small fragment and results in inhibition of tar/
4
8
9
10
3
char formation and promotion of oil formation. In
Composition
the presence of Co O , the formation of acid and oil
3
4
Figure 12. Yield of TOC in the presence of HCO
2
H.
was increased as well as in the presence of KHCO3,
and thus Co O was considered to be a basic catalyst
3
4
3
7–45
in the glucose reaction. In our previous studies,
1
2
we have reported that the acidity and basicity of metal
oxides can be roughly estimated from the electronegativ-
Glycolic acid
Acetic acid
10
8
ity of the metal ion. For example, ZrO is a basic mate-
2
rial in some organic reactions, on the other hand, MoO
3
4+
4
3–46
is an acidic material.
The electronegativities of Zr
6
6
+
46
and Mo are 12.6 and 23.4, respectively. The cobalt
ions in Co O are Co and Co . The electronegativi-
ties of Co and Co are 9 and 12.6, respectively. From
the viewpoint of the electronegativity, Co O is possibly
4
2
+
3+
3
4
2
+
3+
2
3
4
0
a solid base material in regard to some organic reactions
4
8
9
10
Composition
in hot-compressed water, as well as ZrO . Thus the
2
results of the present study support the relationship
between the electronegativity and the reactivity (acidity–
basicity of metal oxides).
Figure 13. Yield of liquid products in the presence of HCO
2
H.
2
5
3
presence of HCO H
.4. Effect of KHCO and Co O on the reaction in the
3 3 4
2
20
15
Next, we carried out the experiments to understand the
effect of a combination of HCO H, KHCO , and Co O
4
2
3
3
1
0
5
0
on the glucose reaction under the same conditions. Fig-
ures 11–14 show the yields of gaseous products, TOC,
and some liquid products in the recovered water solu-
tion and in oil, respectively. Figure 11 reveals that the
H yield was reduced by adding KHCO (composition
2
3
4
8
9
10
Composition
2
2
1
1
5
0
5
0
5
0
Figure 14. Yield of oil in the presence of HCO H.
2
H2
CO2
CO
6
). In the presence of Co O , both CO and H yields
3 4 2 2
were slightly increased. By adding both, KHCO and
3
Co O (composition 8), a further increase of CO and
3
4
2
H yields was observed, similar as in the case of the
2
experiments without HCO H (Fig. 7). The trend of the
2
yield of TOC is shown to depend on the additives
(
Fig. 12): in the presence of HCO H the tendency in
2
4
8
9
10
TOC yields was similar to that without HCO H
2
Composition
(
Fig. 8). The effect of KHCO on TOC formation was
3
Figure 11. Yield of gaseous compounds in the presence of HCO
2
H.
not so high, and Co O increases the TOC yield; the
3 4

2898
M. Watanabe et al. / Carbohydrate Research 341 (2006) 2891–2900
presence of both KHCO and Co O was shown to fur-
glucose, the oil shows a lower content of hydrogen
and oxygen. The loss of oxygen is higher than the loss
of hydrogen, but this effect is not high enough in view
of a technical utilization (see Introduction).
3
3
4
ther promote TOC formation. From the data in Figure
3, we confirmed that KHCO promoted the formation
1
3
of acetic acid; Co O supported forming of both acetic
3
4
and glycolic acid. The yields of acetic acid and glycolic
acid were further enhanced by addition of both KHCO3
and Co O . Everything is similar to the results without
3
.5. Dissolution of Co O in water after the reactions
3 4
3
4
HCO H (Fig. 9). As shown in Figure 14, the effect of
2
We measured the amount of cobalt in the recovered
water solution in each experiment. Figure 16 shows
the loss of Co after reaction, that is, the percentage of
Co that was dissolved in the water solution. Under the
alkaline conditions (compositions 7 and 10), the dissolu-
tion was slightly suppressed; however, over half of the
loaded Co O was dissolved in the water under the
additives on oil formation is similar to that on the
TOC yield: KHCO3 as well as Co O increased the
3
4
TOC (Fig. 12) and oil (Fig. 14) yields. The oil yield fur-
ther increased by the addition of both KHCO3 and
Co O . The maximum yield of oil was 20.9 wt % (com-
3
4
position 8) under the experimental conditions (nearly
0 wt % of glucose, 300 °C, 60 min, 1 g of KHCO , 1 g
3
4
1
3
3
2
experimental conditions. Ding et al. stated that oxides
of Co are thermally and hydrothermally stable, and
these compounds could be tried in hot-compressed
water. However, our experimental results show that
Co O was not stable under the experimental conditions
of Co O , and 2 g of HCO H).
3
4
2
In comparison of the results with HCO H (Figs. 11–
2
1
4) to those without HCO H (Figs. 7–10), the yields of
2
TOC and oil were enhanced by adding HCO H; on
2
3
4
the other hand, the yield of the organic acids (acetic acid
and glycolic acid) were decreased. As explained in Intro-
duction, our new concept of oil formation in hot-com-
pressed water is as follows: glucose is decomposed into
small fragments, and the fragments are combined with
each other by a FT type reaction and by inhibiting poly-
3
2
of this study. Ding et al. also suggested that some
additives (such as Bi, Cd, Ga, Ir, K, Mo, Ta, or W)
should be effective for increasing of physical strength
or stability of metal oxides. In the above results, Co
was found to be useful for making oil from glucose in
hot-compressed water, and thus a modified Co with an
additive would be examined for the further studies on
a Co catalyst.
merization to char and tar. In the absence of HCO H,
2
the acid and oil formation was improved by KHCO3
and Co O . As explained just above, these alkali sub-
3
4
stances promote the reaction to small fragments and
inhibit char and tar formation. Further, the FT type
reaction and inhibition of polymerization typically
3
.6. Outlook
3
0
demands hydrogen to make C–H bonds. The further
Further studies planned for the investigation of a cata-
lyzed liquefaction can be divided in to three parts:
increase of oil formation in the presence of HCO H sug-
2
gests that hydrogenation of the intermediate of the FT
type reaction or a precursor of char/tar formation takes
place to produce oil as shown in Figure 15.
1
. Catalyst screening.
In very small autoclaves different catalysts will be
The composition of the oil corresponds to the chemi-
cal composition CH1.2O0.5. It was the same in every
experiment with the relative high experimental error of
the determination of the chemical composition. Here
an improvement of the analysis procedure is necessary.
tested for activity and stability. In these small autoclaves
heating up and cooling down is faster than in the large
reactor used here, which is an advantage. As the sample
Compared to the original composition of CH O for
2
70
6
5
4
3
2
1
0
0
0
0
0
0
0
Alkali / Co O
Alkali / Co O
3
4
3
4
Small fragments
Gas
(
aldehyde, acid)
O
O
FT reaction
(
Co O / HCO H)
3 4 2
O
O
O
O
Oil
Tar/Char
Ring compounds
Furans, Phenols)
Hydrogenation
HCO H)
(
(
2
6
7
9
10
Composition
3 4 2 3
Figure 15. Effect of additives (Co O , HCO H, and KHCO ) on the
glucose reaction in hot-compressed water.
Figure 16. Loss of Co
3 4
O .

M. Watanabe et al. / Carbohydrate Research 341 (2006) 2891–2900
2899
volumes are necessarily small, the analyses must be
restricted to a few parameters.
authors thank A. B o¨ hm and K. Hedwig for the analysis
of the liquid and gaseous sample. Some experimental
work was conducted by T. Tietz, who is gratefully
acknowledged. Many thanks also to Mr. Sch u¨ bel and
Mr. Hager for the technical support.
2
. Fundamental understanding.
After a catalyst is selected, experiments with glucose
will be performed in a tubular reactor with a fixed bed
of catalyst. This reactor shows very short heating up
and cooling down times and therefore gives very precise
results. Additionally the volumes of the different sam-
ples are, in principle, unlimited; therefore, sampling
problems can be minimized. The disadvantage of this
type of reactor is that the residence time range is rather
small. Therefore, the reactor must be designed for a
special purpose, which means, in this case, for a special
catalyst. Therefore, these studies can only be done after
the catalyst screening studies.
References
1. Demirbas, A. Energy Convers. Manage. 2000, 41, 633–646.
2
3
4
. Goudriaan, F.; Peferoen, D. G. R. Chem. Eng. Sci. 1990,
5, 2729–2734.
. Minowa, T.; Murakami, M.; Dote, Y.; Ogi, T.; Yoko-
yama, S. Biomass Bioenergy 1995, 8, 117–120.
4
. Demirbas, A. Energy Convers. Manage. 1998, 39, 685–690.
3
. Experiments with real biomass.
Experiments in a fixed bed tubular reactor are difficult
5. Minowa, T.; Kondo, T.; Sudirjo, S. T. Biomass Bioenergy
998, 14, 517–524.
1
6
. Inoue, S.; Okigawa, K.; Minowa, T.; Ogi, T. Biomass
Bioenergy 1999, 16, 377–383.
to perform because of the risk of plugging. So first
experiments will be carried out in the large autoclave
described here. The disadvantage of this reactor is the
high heating and cooling time. One the other hand, be-
cause of the large sample volume natural inhomogenities
in a piece of biomass can be neglected. The sample
volumes of the products are also relatively large, large
enough to perform a number of analytical measurements.
7
. Demirbas, A. Energy Convers. Manage. 2000, 41, 1601–
1
607.
8. Demirbas, A. Energy Convers. Manage. 2001, 42, 1357–
378.
. Qu, Y.; Wei, X.; Zhong, C. Energy 2003, 28, 597–606.
1
9
1
1
1
1
1
1
1
0. Yang, Y. F.; Feng, C. P.; Inamori, Y.; Maekawa, T.
Resour. Conserv. Recycl. 2004, 43, 21–33.
1. Karag o¨ z, S.; Bhaskar, T.; Muto, A.; Sakata, Y. Bioresour.
Technol. 2006, 97, 90–98.
2. Huber, G. W.; Chheda, J. N.; Barrett, C. J.; Dumesic, J.
A. Science 2005, 308, 1446–1450.
3. Adschiri, T.; Sato, T.; Shibuichi, H.; Fang, Z.; Okazaki,
S.; Arai, K. Fuel 2000, 79, 243–248.
4. Conclusions
4. Sasaki, M.; Goto, K.; Tajima, K.; Adschiri, T.; Arai, K.
Green Chem. 2002, 4, 285–287.
5. Yang, B. Y.; Montgomery, R. Carbohydr. Res. 1996, 280,
It is found that HCO H and Co O show a positive
effect on oil formation from glucose in hot-compressed
water. The role of HCO H is probably that of a hydro-
genating agent that inhibits further polymerization of
2
3
4
2
7–45.
2
6. Srokol, Z.; Bouche, A.-G.; van Estrik, A.; Strik, R. C. J.;
Meschmeyer, T.; Peters, J. A. Carbohydr. Res. 2004, 339,
1717–1726.
the produced oil. The catalytic activity of Co O was
3
4
sometimes similar to alkali (KHCO ), and thus Co O
4
worked as a solid base material under these conditions.
17. Antal, M. J., Jr.; Mok, W. S. L.; Richrads, G. N.
3
3
Carbohydr. Res. 1990, 199, 91–109.
1
1
8. Bicker, M.; Hirth, J.; Vogel, H. Green Chem. 2003, 5, 280–
84.
9. Watanabe, M.; Aizawa, Y.; Iida, T.; Aida, T. M.; Levy,
C.; Sue, K.; Inomata, H. Carbohydr. Res. 2005, 340, 1925–
1930.
In the presence of HCO H, oil formation was further in-
2
2
creased by Co O . We assume the process is that of an
3
4
FT-type reaction in the presence of HCO H and
2
Co O . However, the stability of Co O was low.
3
4
3
4
2
2
2
0. Sasaki, M.; Gota, K.; Tajima, K.; Adschiri, T.; Arai, K.
Green Chem. 2005, 4, 285–287.
1. Kabyemela, B. M.; Adschiri, T.; Malaluan, R. M.; Arai,
K. Ind. Eng. Chem. Res. 1999, 38, 2888–2895.
2. Minowa, T.; Fang, Z.; Ogi, T.; Varhegyi, G. J. Chem. Eng.
Jpn. 1998, 31, 131–134.
The research for the best conditions for producing oil
from cellulosic material is still lacking, and further
experimental studies will be conducted. A bimetallic cat-
alyst including Co with a stabilization metal will be
studied. In that case, the activity of the mixed catalyst
is sensitive for a method of preparation, and thus we
must examine the catalyst by taking the preparation into
account.
2
2
2
3. Minowa, T.; Ogi, T. Renewable Energy 1994, 5, 813–815.
4. Minowa, T.; Ogi, T. Catal. Today 1998, 45, 411–416.
5. Minowa, T.; Inoue, S. Renewable Energy 1999, 16, 1114–
1
117.
6. Park, K. C.; Tomiyasu, H. Chem. Commun. (Cambridge)
003, 694–695.
7. Sato, T.; Osada, M.; Watanabe, M.; Shirai, M.; Arai, K.
Ind. Eng. Chem. Res. 2003, 42, 4277–4282.
8. Osada, M.; Sato, T.; Watanabe, M.; Adschiri, T.; Arai, K.
Energy Fuels 2004, 18, 327–333.
9. Matsumura, Y.; Minowa, T.; Potic, B.; Kersten, S. R. A.;
Prins, W.; van Swaaij, W. P. M.; van de Beld, B.; Elliott,
2
2
2
2
2
Acknowledgments
We thank the Ministry of Education, Culture, Sports,
Science, and Technology of Japan and the Forschungs-
zentrum Karlsruhe for the financial support. The

2900
M. Watanabe et al. / Carbohydrate Research 341 (2006) 2891–2900
D. C.; Neuenschwander, G. G.; Kruse, A.; Antal, M. J.,
Jr. Biomass Bioenergy 2005, 29, 269–292.
0. Betrole, C. J.; Mims, C. A.; Kiss, G. J. Catal. 2002, 210,
4–96.
1. Grillo, F.; Natile, M. M.; Glisenti, A. Appl. Catal. B 2004,
8, 267–274.
2. Ding, Z.; Frisch, M. A.; Li, L.; Gloyna, E. F. Ind. Eng.
Chem. Res. 1996, 35, 3257–3279.
3. Sasaki, M.; Kabyemela, B.; Malaluan, R.; Horose, S.;
Takeda, N.; Adschiri, T.; Arai, K. J. Supercrit. Fluids
38. Watanabe, M.; Aizawa, Y.; Iida, T.; Levy, C.; Aida, T.
M.; Inomata, H. Carbohydr. Res. 2005, 340, 1931–1939.
39. Kruse, A.; Gawlik, A. Ind. Eng. Chem. Res. 2003, 42, 267–
279.
40. Saisu, M.; Sato, T.; Watanabe, M.; Adschiri, T.; Arai, K.
Energy Fuels 2003, 17, 922–928.
41. Watanabe, M.; Inomata, H.; Arai, K. Biomass Bioenergy
2002, 22, 405–410.
42. Watanabe, M.; Inomata, H.; Osada, M.; Sato, T.; Ads-
chiri, T.; Arai, K. Fuel 2003, 82, 545–552.
3
3
3
3
8
4
1
998, 13, 261–268.
4. Sinag, A.; Kruse, A.; Rathert, J. Ind. Eng. Chem. Res.
004, 43, 502–508.
43. Watanabe, M.; Osada, M.; Inomata, H.; Arai, K.; Kruse,
A. Appl. Catal. A 2003, 245/2, 333–341.
44. Watanabe, M.; Iida, T.; Aizawa, Y.; Ura, H.; Inomata,
H.; Arai, K. Green Chem. 2003, 5, 539–544.
45. Watanabe, M.; Aizawa, Y.; Iida, T.; Nishimura, R.;
Inomata, H. Appl. Catal. A 2005, 295, 150–156.
46. Tanaka, K.; Ozakil, A. J. Catal. 1967, 8, 1–7.
3
2
3
3
3
5. Akiya, N.; Savage, P. E. AIChE J. 1998, 44, 405–415.
6. Yu, J.; Savage, P. E. Ind. Eng. Chem. Res. 1998, 37, 2–10.
7. Watanabe, M.; Inomata, H.; Smith, R. L., Jr.; Arai, K.
Appl. Catal. A 2001, 219, 149–156.