Formation of lactic acid from glycolaldehyde by alkaline hydrothermal reaction

Article abstract of DOI:10.1016/j.carres.2006.06.013

Alkali hydrothermal experiments with glycolaldehyde were carried out at 300 °C. Glycolaldehyde was converted into lactic acid in a yield of 28% based on the starting carbon mass of glycolaldehyde. A conversion pathway for glycolaldehyde into lactic acid is proposed and our results suggest that the pathway via glycolaldehyde is also important in the conversion of glucose into lactic acid.

Full text of DOI:10.1016/j.carres.2006.06.013

Carbohydrate Research 341 (2006) 2619–2623
Note
Formation of lactic acid from glycolaldehyde by
alkaline hydrothermal reaction
a
b,
b
c
b
*
Hisanori Kishida, Fangming Jin, Xiuyi Yan, Takehiko Moriya and Heiji Enomoto
a
Environmental Systems Headquarters, Environmental Research and Development Center,
Hitachi Zosen Corporation, Kyoto 625-8501, Japan
b
Department of Environmental Science and Technology, Graduate School of Environmental Studies, Tohoku University,
Sendai 980-8579, Japan
c
Research and Development Center, Tohoku Electric Power Co., Inc., Sendai 981-0952, Japan
Received 5 March 2006; received in revised form 28 May 2006; accepted 16 June 2006
Available online 6 September 2006
Abstract—Alkali hydrothermal experiments with glycolaldehyde were carried out at 300 °C. Glycolaldehyde was converted into
lactic acid in a yield of 28% based on the starting carbon mass of glycolaldehyde. A conversion pathway for glycolaldehyde into
lactic acid is proposed and our results suggest that the pathway via glycolaldehyde is also important in the conversion of glucose
into lactic acid.
Ó 2006 Elsevier Ltd. All rights reserved.
Keywords: Hydrothermal reaction; Carbohydrates; Biomass; Lactic acid; Alkaline catalyst
The hydrothermal process is one of the most promising
processes for the conversion of biomass into chemicals,
because at high-temperatures and high-pressures water
behaves as a reaction medium having unusual proper-
ties. There has been, therefore, extensive research on
the conversion of biomass into chemicals by hydrother-
mal treatment. Our past studies on the conversion of
carbohydrate biomass into useful substances showed
that lactic acid can be formed by this process without
the addition of any catalyst, probably due to a base
of the three-carbon atom aldose, glyceraldehyde, which
is formed by the reverse aldol condensation of hexoses.
Our previous study on the conversion mechanism of
glucose to lactic acid in hydrothermal reactions also
showed that the transformation appeared to follow the
pathway that was elucidated for the conversion of sugar
3
to lactic acid in an alkaline solution. However, few
studies report on whether glycoaldehyde, an aldose
containing two carbon atoms, can produce lactic acid,
although glycoaldehyde is also generated by the reverse
aldol condensation of hexoses. The purpose of this study
was to examine the formation of lactic acid from glyco-
aldehyde in an alkaline hydrothermal reaction.
1
–3
catalytic effect of high-temperature water.
We have
also shown that lactic acid can be selectively formed
4
by adding a basic catalyst. Lactic acid has recently
been receiving attention as a material for the prepara-
tion of biodegradable polymers with limited environ-
mental impact. To increase lactic acid production, an
understanding of its formation from carbohydrates by
hydrothermal processes is needed.
To examine if lactic acid can be formed from glycol-
aldehyde, an experiment was done with this aldose at
300 °C for 10 min in 0.75 M NaOH. HPLC analyses of
the solution after the reaction (Fig. 1) showed a large
lactic acid peak, indicating that glycolaldehyde also
produces lactic acid, in addition to formic acid, acetic
acid, and acrylic acid. Acrylic acid is most likely the
dehydration product of lactic acid while acetic acid is
the decarbonylation product. A confirmatory experi-
ment was carried out under the same conditions. Mok
It is generally known in carbohydrate chemistry that
lactic acid is generated via base-catalyzed decomposition
*
Corresponding author. Tel./fax: +81 22 795 7385; e-mail: jin@
mail.kankyo.hotoku.ac.jp
0
008-6215/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carres.2006.06.013

2620
H. Kishida et al. / Carbohydrate Research 341 (2006) 2619–2623
Lactic acid
Formic acid
Acetic acid
Acryic acid
1 Glyceraldehyde
2 Erythrose
3 Pyruvaldehyde
4 Lactic acid
7
Solvent
3
4
2
Solvent
5 Formic acid
1
5
6
6
7
Acetic acid
Acrylic acid
0
10
20
30
40
Retention time (min)
0
5
10
15
20
25
30
35
40
Retention time (min)
Figure 1. HPLC chromatogram of the sample after alkali hydrother-
mal treatment of glycolaldehyde (temperature: 300 °C; reaction time:
Figure 3. HPLC chromatogram of the sample after alkali hydrother-
mal treatment of glycolaldehyde (temperature: 300 °C; reaction time:
10 min, 0.75 M NaOH).
1
min; 0.75 M NaOH).
5
6
et al. and Lira et al. also reported that in sub- and
supercritical water, lactic acid underwent dehydration
to give acrylic acid, and that decarbonylation to acetic
acid also occurred. Therefore, formic acid would also
be a by-product in the formation of lactic acid from
glycolaldehyde. Furthermore, as shown in Figure 2, c-
butyrolactone, which contains four-carbon atoms, was
also detected by GC–MS analyses. For compounds 1
and 2 (Fig. 2), a spectral library provided no good
matches. The mass spectra of the two compounds have
a strong peak at m/z = 56, which is a characteristic peak
for lactones, suggesting that these compounds most
likely had this structural motif. These observations sug-
gest that a condensation reaction occurs in the alkali
hydrothermal reaction of glycolaldehyde. In addition,
an experiment with glycolaldehyde was performed at a
short reaction time (1 min), which led to the identifica-
tion of more intermediate products. As shown in Figure
may lose water to form a carbon–carbon double bond
between the a- and b-carbon atoms, yielding 2,4-dihydr-
oxybutanal (3). At the same time, 3 may undergo the
keto-enol tautomerization to 4, which may subsequently
decompose to formaldehyde and pyruvaldehyde (5) by a
reverse aldol condensation. Finally, 5 may undergo
benzilic acid rearrangement to lactic acid (7). A possible
second pathway (ii) is based not only on the traditional
theory of the conversion of carbohydrates to lactic acid
in alkaline solution but also on recent studies on hydro-
thermal reaction pathways of monosaccharide model
7
compounds. In pathway ii, erythrose may rearrange
into 8 by a Lobry de Bruyn–Albeda van Ekenstein
transformation (LBAE). Subsequent cleavage via a
reverse aldol condensation results in formic acid and
glyceraldehyde (9), which then undergoes isomerization
to dihydroxyacetone (10). Both glyceraldehyde and
dihydroxyacetone can undergo dehydration to give
pyruvaldehyde (6), which subsequently undergoes the
benzilic acid rearrangement to lactic acid. The identifica-
tion of glyceraldehyde may provide support for the pres-
ence of pathway ii. A possible co-product with 8, if the
LBAE transformation occurs, would be 1,2-enediol, 11.
Glyceraldehyde could also be formed by cleavage of 11
(pathway iii), because the formation of the double bond
between carbon 1 and carbon 2 in erythrose would be
expected to weaken the single bond between carbon 3
and carbon 4.
3
, erythrose, pyruvaldehyde, and glyceraldehyde were
also produced.
On the basis of these experimental findings and basic
theories in alkaline degradation of monosaccharides, we
propose a pathway for the formation of lactic acid from
glycolaldehyde under these conditions (Fig. 4). Because
erythrose, 2, was identified, glycolaldehyde 1, may un-
dergo aldol condensation to give erythrose. Lactic acid
may be produced from 2 by one of the three pathways.
First (pathway i), considering that the dehydration
easily occurs under hydrothermal conditions, erythrose
In all three pathways, formaldehyde would be formed
and, given the basic reaction medium, this by-product
may be converted into formic acid and methanol by
the Cannizzaro reaction, a process in which aldehydes
containing no a-hydrogens undergo self-oxidation and
reduction to yield a mixture of a carboxylic acid salt
O
O
1
2
8
and an alcohol. To test this, an experiment with form-
aldehyde (a 40% aqueous solution of formaldehyde with
methanol as a stabilizer) was performed at 300 °C in
0.75 M NaOH. Under these conditions, a considerable
yield (15.7%) of formic acid was obtained. Furthermore,
H NMR analyses of the reaction mixtures for both
glycolaldehyde and formaldehyde showed a strong
signal for the methyl protons of methanol at 3.5 ppm,
1
0
12
14
16
18
20
Retention time (min)
1
Figure 2. GC–MS chromatogram of the sample after alkali hydro-
thermal treatment of glycolaldehyde (temperature: 30 °C, reaction
time: 10 min; 0.75 M NaOH).

H. Kishida et al. / Carbohydrate Research 341 (2006) 2619–2623
2621
Formic acid
H
C
C
C
C
H
C
C
C
C
CH3
H
H
OH
H
H
OH
H
R
4
H
C
5
C
C
O
R
3
H
O
+
O
OH
O
O
H
6
O
R5
R
2
H
H
3
4
R2
i
H
H
C
C
C
C
H
2
H
H
H
C
C
C
C
OH
OH
O
H
C
C
C
OH O
H
H
H
OH
OH
OH
O
R
1
R4
H
R
6
H C OH
H C O
H C O
H
C
C
C
H
OH
R6
OH
H
H
H
H
H
H
C
H
C H
H
C
O
+
ii
H
OH
O
5
O
OH
1
H
H
iii
Formic acid
H
8
OH
7
(
× 2)
9
10
H
H
C
C
C
C
H
OH
Enolization
H
H
OH
OH
OH
11
Figure 4. Proposed pathways for the conversion of glycolaldehyde into lactic acid. R
tautomerization; R : reverse aldol condensation; R : benzilic acid rearrangement; R
LBAE).
1
: Aldol condensation; R
6
: Lobry de Bruyn–Albeda van Ekenstein transformation
2
: elimination of H
2
O; R
3
: keto-enol
4
5
(
indicating that a considerable amount of methanol was
also formed. Moreover, because five-membered ring lac-
tones were also found, 2 and/or 3, 8, 10 may also cyclize
into lactones via an aldol condensation.
formic acid. However, the yield of formic acid obtained
using 0.75 M NaOH was lower than that with 0.25 M
NaOH (see Table 1). This does not necessarily indicate
that these results are inconsistent with the mechanism
proposed in Figure 4. In general, the Cannizzaro reac-
tion occurs in the presence of concentrated alkali at a
low temperature. The decrease in formic acid with an in-
crease in NaOH may be due to the high temperature and
pressure, which may lead to other reactions. The results
of the experiments with formaldehyde showed that the
yield of formic acid using 0.75 M NaOH was also lower
than that using 0.25 M NaOH (Table 1). Furthermore,
from Figure 4, it can be seen that the formation of lactic
acid from glycolaldehyde proceeds via erythrose. This
means that a four-carbon aldose can also form lactic
acid, probably with a yield similar to or less than that
from glycolaldehyde. However, experiments to confirm
this cannot be completed because pure erythrose is not
available. From the discussion above, glycolaldehyde
and erythrose can produce lactic acid by the alkali
hydrothermal treatment. That is, the pathway via gly-
colaldehyde is also an important pathway for the con-
version of glucose into lactic acid.
Although aldol and reverse aldol condensations as well
as keto-enol tautomerization occur in both basic and
acidic solutions, the reactions in basic solution are gener-
ally much faster, because of the greater catalytic effect of
9
hydroxide ions. In addition, the benzilic acid rearrange-
ment occurs only in a basic solution. Thus, the formation
of lactic acid from glycolaldehyde is critically dependent
upon the concentration of the base. Experiments were
performed by changing the concentration of NaOH at
3
00 and 200 °C. As expected (Table 1), at both tempera-
tures, the yield of lactic acid decreased with the decreas-
ing concentration of NaOH. These observations further
support the conversion mechanisms shown in Figure 4.
Furthermore, from the data in Table 1, it can be seen that
an increase in temperature leads to an increase in the yield
of lactic acid. For glycolaldehyde, the acidity of its a-
hydrogen is lower than that of aldehydes without the
OH group at the a-carbon, due to the inductive effect
of the oxygen. Thus, the aldol condensation for glycol-
aldehyde may be more difficult than for other aldehydes.
This may be a reason why glycolaldehyde produces little
lactic acid at a lower temperature.
Finally, to examine the contribution of these aldoses
to lactic acid production from glucose, the experiments
with glyceraldehyde and glucose were carried out at
300 °C with 0.75 M NaOH. As shown in Table 1,
the yields of lactic acid from glyceraldehyde and glu-
cose were 39.9% and 23.6%, respectively. The higher
Moreover, it should be noted that according to the
proposed mechanism shown in Figure 4, an increase in
NaOH should lead to an increase in the formation of

2
622
H. Kishida et al. / Carbohydrate Research 341 (2006) 2619–2623
O) (n = 1,2,3,6) by alkali hydrothermal treatment
Table 1. Yields of products from C
n
(H
2
n
a
Hydrothermal conditions
Yield (%)
Initial material
Temp (°C)
NaOH (M)
0.25
Lactic acid
Formic acid
Acetic acid
Acrylic acid
Glycolaldehyde C
2
3
(H
2
O)
2
200
11.0
17.9
17.8
28.1
10.1
5.4
13.7
8.9
1.9
1.6
8.8
3.9
1.0
1.3
5.0
3.8
0
.75
0.25
.75
3
00
0
Glyceraldehyde C
(H
2
O)
3
200
00
0.25
.75
0.25
.75
20.6
30.8
27.8
39.9
5.8
4.3
5.2
4.8
4.3
1.3
4.9
3.1
0
3
0
Acetaldehyde C(H
2
O)
200
00
0.25
.75
0.25
.75
2.9
5.8
3.7
0.0
35.2
26.9
29.9
15.7
8.3
3.6
9.8
0
0
3
0
Glucose C
6
(H
2
O)
6
300
0.75
23.6
4.3
3.1
Reaction time: 10 min.
a
The percentage of products compared to the initial material based on carbon mass.
yield from glyceraldehyde suggests that the major
pathway for the conversion of glucose into lactic acid
proceeds via this aldose. The lower yield of lactic acid
from glucose suggests that degradation of glucose in
alkali hydrothermal reaction is very complex and many
significant side-reactions occur.
carried out. For GC–MS analyses, a Hewlett–Packard
model 5890 Series II Gas Chromatograph equipped with
a model 5890B Mass Selective Detector was used. The
samples were separated on a HP-INNOWAX capillary
column (Cross-Linked Polyethylene Glycol) using he-
lium as the carrier. HPLC analysis was performed with
a Waters HPLC system equipped with a tunable absor-
bance detector (UV–vis detector) and a differential
refractometer (RI detector). Samples were separated
through a series of two identical columns of RSpak
1. Experimental
1
1
.1. Materials
KC-811 to obtain a better separation. H NMR spectra
were recorded on a JEOL JNM-LA300 spectrometer
operating at 300 MHz. Details on the conditions for
Glycolaldehyde (Aldrich, #10503TB) was used; NaOH
99%) was used as the alkaline catalyst. Other chemicals,
such as acetaldehyde (36–38% solution), DL-glyceralde-
hyde (99%), and glucose (99.9%) were obtained from
Wako Pure Chemicals Industries Ltd.
1
(
GC–MS, HPLC, and H NMR analyses are available
2
,11
elsewhere.
Quantitative analyses for major products
were carried out by HPLC. In general, quantitative data
were obtained by comparing the area of the peak. When
a compound peak was not well separated, quantitative
data were obtained by the height of the peak.
1
.2. General procedures
All experiments were conducted in a batch reactor sys-
1
0,11
References
tem described elsewhere.
experimental procedure is given below. Water (4 mL,
0% water fill) containing the desired amount of NaOH
A brief description of the
1
. Jin, F.; Zheng, J.; Enomoto, H.; Moriya, T.; Sato, N.;
Higashijima, H. Chem. Lett. 2002, 31, 504–505.
4
and the test material was added to the reactor, which
was immersed in a salt bath preheated to the desired
temperature. After the reaction, the reactor was quickly
cooled in cold water. All experiments were performed
with degassed water and by purging the reactor with
nitrogen. A liquid sample was filtered through a
2
. Jin, F.; Zhou, Z.; Moriya, T.; Kishda, H.; Higashijima, H.;
Enomoto, H. Environ. Sci. Technol. 2005, 39, 1893–1902.
3. Jin, F.; Zhou, Z.; Enomoto, H.; Moriya, T.; Higashijima,
H. Chem. Lett. 2004, 33, 126–127.
4
. Yan, X.; Jin, F.; Enomoto, H.; Moriya, T.; Hisanori, K.;
Higashijima, H. Water, Steam, and Aqueous Solutions for
Electric Power—Advance in Science and Technology;
Maruzen Ltd.: Tokyo, 2005; pp 724–728.
0
.45 lm filter, and then the pH of the solution was
adjusted to 6–7 with sulfuric acid.
5. Mok, W. S. L.; Antal, M. J., Jr. J. Org. Chem. 1989, 54,
596–4602.
. Lira, C. T.; McCrackin, P. J. Ind. Eng. Chem. Res. 1993,
2, 2608–2613.
. Srokol, Z.; Bouche, A.-G.; van Estrik, A.; Strik, R. C. J.;
Maschmeyer, T.; Peters, J. A. Carbohydr. Res. 2004, 339,
1717–1726.
4
6
7
1
.3. Products analyses
3
After the reaction, the liquid sample was analyzed by
1
HPLC and GC–MS. H NMR analyses were also

H. Kishida et al. / Carbohydrate Research 341 (2006) 2619–2623
2623
8
9
. Morrison, R. T.; Boyd, R. N. Organic Chemistry, 6th ed.;
Prentice-Hall: New Jersey, 1992.
. Knill, C. J.; Kennedy, J. F. Carbohydr. Polym. 2003, 51,
10. Kishida, H.; Jin, F.; Moriya, T.; Enomoto, H. Chem. Lett.
2005, 34, 1560–1562.
11. Jin, F.; Moriya, T.; Enomoto, H. Environ. Sci. Technol.
2003, 37, 3220–3231.
2
81–300.