Zeolite H-USY for the production of lactic acid and methyl lactate from C3-sugars

Article abstract of DOI:10.1016/j.jcat.2009.10.023

Lactic acid is an interesting platform chemical with many promising applications. This includes the use as a building block for the production of biodegradable plastics and environmentally friendly solvents. A study of the liquid-phase conversion of the triose-sugars, glyceraldehyde and dihydroxyacetone directly to methyl lactate and lactic acid catalyzed by inexpensive commercially available zeolites is presented. One particular zeolite, H-USY (Si/Al = 6) is shown to be quite active with near quantitative yields for this isomerization. Deactivation of the H-USY-zeolite was studied by correlating the catalytic activity to data obtained by TPO, XRD, N₂-sorption, and NH₃-TPD on fresh and used catalysts. Coking and irreversible framework damage occurs when lactic acid is produced under aqueous conditions. In methanol, methyl lactate is produced and catalyst deactivation is suppressed. Additionally, reaction rates for the formation of methyl lactate in methanol are almost an order of magnitude higher as compared to the rate of lactic acid formation in water.

Full text of DOI:10.1016/j.jcat.2009.10.023

Journal of Catalysis 269 (2010) 122–130
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Zeolite H-USY for the production of lactic acid and methyl lactate from C3-sugars
Ryan M. West a, Martin Spangsberg Holm b, Shunmugavel Saravanamurugan b, Jianmin Xiong b,
d
Zachary Beversdorf c, Esben Taarning d, , Claus Hviid Christensen
*
a University of Wisconsin-Madison, Department of Chemical and Biological Engineering, Madison, WI 53706, USA
b Center for Sustainable and Green Chemistry, Department of Chemistry, Technical University of Denmark, DK-2800 Lyngby, Denmark
c Iowa State University, Department of Chemical and Biological Engineering, Ames, IA 50011, USA
d Haldor Topsøe A/S, Nymøllevej 55, DK-2800, Denmark
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 4 June 2009
Revised 21 October 2009
Accepted 24 October 2009
Available online 25 November 2009
Lactic acid is an interesting platform chemical with many promising applications. This includes the use as
a building block for the production of biodegradable plastics and environmentally friendly solvents. A
study of the liquid-phase conversion of the triose-sugars, glyceraldehyde and dihydroxyacetone directly
to methyl lactate and lactic acid catalyzed by inexpensive commercially available zeolites is presented.
One particular zeolite, H-USY (Si/Al = 6) is shown to be quite active with near quantitative yields for this
isomerization. Deactivation of the H-USY-zeolite was studied by correlating the catalytic activity to data
obtained by TPO, XRD, N2-sorption, and NH3-TPD on fresh and used catalysts. Coking and irreversible
framework damage occurs when lactic acid is produced under aqueous conditions. In methanol, methyl
lactate is produced and catalyst deactivation is suppressed. Additionally, reaction rates for the formation
of methyl lactate in methanol are almost an order of magnitude higher as compared to the rate of lactic
acid formation in water.
Keywords:
Lactic acid
Methyl lactate
H-USY
Zeolite
Dihydroxyacetone
Glyceraldehyde
Triose sugar
Deactivation
Biomass
Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction
tion of calcium lactate, followed by acidification with sulfuric acid
releases the crude lactic acid and gypsum. Typically, one ton of
Lactic acid is currently emerging as a building block in a new
generation of materials such as biodegradable plastics and solvents
[1–3]. These new materials can be produced from biomass-derived
precursors and have the potential to replace existing petroleum-
based materials by displaying comparable and even superior prop-
erties [4]. Lactic acid also has the potential to become a central
chemical feedstock for the chemical industry in the production of
acrylic acid, propylene glycol, and different useful condensation
products [5–8].
However despite its high potential, the major obstacle in a
wider implementation of lactic acid-based materials and a lactic
acid platform in the chemical industry is the high cost associated
with the expensive and cumbersome manufacturing route of lactic
acid. The large-scale production of lactic acid relies on the batch-
wise fermentation of aqueous glucose under anaerobic conditions
[9,10]. The fermentation reaction typically takes 2–4 days and re-
quires calcium hydroxide to be added continuously to maintain a
neutral pH-level in order for the bacteria to function optimally,
thereby resulting in the formation of calcium lactate. Crystalliza-
gypsum is formed for every ton of lactic acid produced [1]. Further
purification of lactic acid is done by esterification to methyl lactate
followed by distillation and hydrolysis to release pure lactic acid.
Lactic acid is an isomer of the triose sugars dihydroxyacetone
(DHA) and glyceraldehyde (GLA). The relative stability of the three
isomers are in the order of lactic acid ꢀ DHA > GLA. The triose
sugars can be formed by aerobic oxidation of glycerol using
both homogeneous and heterogeneous catalysts [11–14] or by
fermentation of glycerol using the Gluconobactor suboxydans strain
[15–18].
Homogeneous catalysts such as sulfuric acid and sodium
hydroxide are known to catalyze the isomerization of DHA and
GLA in very hot water (250–300 °C) to give low yields of lactic acid
[19,20]. A more effective isomerization catalyst is zinc sulfate,
which is reported to give a 75–86% yield of lactic acid in 300 °C
hot water [21]. However, the use of a homogeneous isomerization
catalyst is problematic from an environmental point of view and
the purification of lactic acid can be problematic. Furthermore,
the use of very hot water as the reaction media requires expensive
pressure resistant equipment which will make
a large-scale
process unattractive. Homogeneous metal chlorides are active in
this isomerization–esterification reaction. In particular, tin
* Corresponding author.
E-mail address: esta@topsoe.dk (E. Taarning).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.10.023

R.M. West et al. / Journal of Catalysis 269 (2010) 122–130
123
chloride has been found to be able to form methyl lactate in a high
89% yield [22]. However, a high catalytic amount (10 mol%) of tin
chloride is used in this case.
tures were heated to 69 °C under stirring for 1–2 h to help dissolve
the substrate.
Effluent concentrations for batch and flow experiments were
characterized by an Agilent 1200 Series HPLC with an R.I. detector
and a Biorad Aminex HPX-87H column and an Agilent 6890N GC
with an HP-5column and an FID detector. Species were quantified
with standards and confirmed with GC-MS, Agilent 6850 GC sys-
tem coupled with Agilent 5975 C MSD.
Currently, very limited work has been done using acidic zeolites
for the conversion of biomass to value added chemicals. Acidic zeo-
lites have been demonstrated to catalyze the conversion of DHA in
ethanol to form ethyl lactate, although in a moderate 65% yield
[23]. Very recently Lewis acidic beta zeolites were shown to be
highly active in the isomerization reaction of DHA and GLA with
excellent selectivity toward methyl lactate/lactic acid [24]. These
hydrophobic zeolites were made by incorporating Ti, Zr, and Sn
into the framework. In particular the Sn-Beta zeolite appears to
be an exceptionally promising catalyst in the DHA/GLA conversion.
However, application may be limited due to long synthesis time
and to the use of tin. Thus an investigation using commercially
available solid acids, mainly zeolites containing only silicon and
aluminum and sulfated zirconia was undertaken in batch reactors.
The most promising of these catalysts, the proton form of an ultra
stable Y-zeolite with a Si/Al = 6 (H-USY-6) showed very good activ-
ity with near quantitative yields at appropriate conditions. This
zeolite was tested under continuous flow within a plug flow reac-
tor to determine the kinetics of the isomerization network and to
study the deactivation of the system. A discussion of the catalytic
behavior is coupled to the results from XRD, NH3-TPD, N2-sorption,
FT-IR, and TPO of the fresh and spent zeolite.
2.3. Zeolite characterization
The degree of carbon deposition on used catalysts was deter-
mined using temperature-programed oxidation (TPO). Prior to
analysis the used zeolites were dried at 110 °C. The TPO was car-
ried out by heating a premeasured amount of used catalyst to
650 °C at 3.5 °C minꢁ1 in a gas flow of 20 mL minꢁ1 consisting of
5% oxygen in helium. The release of CO and CO2 was quantified
using a BINOS detector. The TPO data are shown in Table 4 as the
weight percent of carbon present on the coked catalyst.
NH3-TPD, N2-sorption, and XRD were performed on calcined
catalyst samples. The procedures used were either as described
for the TPO or calcination at 550 °C for 4 h in static air, heated at
a ramp of ꢂ2 °C/min. Nitrogen physisorption was measured on a
Micromeretics ASAP 2020 after the samples had been degassed
in vacuum at 300 °C for 2 h.
Powder X-ray diffraction (XRD) patterns were recorded on a
Bruker AXS powder diffractometer. The XRD data are shown in Ta-
ble 4 as a relative crystallinity, calculated by integrating the peak
area of eight characteristic reflections and comparing to the un-
used zeolite. The reflections used along with the respective 2h val-
ues given in parenthesis were 331 (15.97), 511 (19.01), 440
(20.71), 533 (24.06), 642 (27.52), 822 (31.29), 555 (31.95), and
664 (34.69) [25].
2. Experimental
2.1. Chemicals
Dihydroxyacetone (97%), pyruvic aldehyde dimethyl acetal
(PADA) (97%), glyceraldehyde (95%), anhydrous pyridine (99.8%),
and methanol (99.9%) were purchased from Sigma-Aldrich. Methyl
lactate (97%) was obtained from Fluka. Pyruvic aldehyde (PA) was
obtained from SAFC Supply as a 40 wt.% aqueous solution while
lactic acid was obtained from Fluka and Riedel-de Haën. All the
commercially available zeolites used throughout this study were
kindly provided by Zeolyst International. The zeolites are pure
and do not contain any binder material. Some of the zeolites were
received in the NH4-form. In these cases the zeolites were calcined
at 550 °C in air for 6 h prior to use in order to produce the acidic
form.
NH3-TPD measurements were performed on a Micromeritics
Autochem II equipped with a TCD detector. Dry weights of the
samples were found after evacuation at 300 °C for 1 h. After satu-
ration with ammonia, the weakly bound ammonia was desorbed
prior to measurement at 150 °C for 1 h in a He flow of 25 mL minꢁ1
.
The desorption curve was then attained at a heating ramp of 15 °C
per minute to 550 °C at a He flow rate of 25 mL minꢁ1
.
FT-IR operated in transmission mode was used to analyze the
zeolites on a BioRad FTS 80 spectrometer equipped with a MCT
detector. Self-supporting wafers of the zeolites were pressed and
were prior to analysis dehydrated under evacuation at 400 °C for
4 h. The absorbance spectra were obtained after the samples were
allowed to cool to RT. Pyridine adsorption was done by saturating
the zeolite at RT and subsequently heating the sample to 200 °C for
30 min under evacuation. The sample was once again allowed to
cool to RT before the spectrum was recorded.
2.2. Catalytic tests
Batch experiments were performed in an Ace pressure tube
with magnetic stirring. In these runs, 80 mg of catalyst, 1.25 mmol
of substrate (calculated as monomer), and 4 g of water or methanol
were added and mixed in the pressure tube using magnetic stirrer
ring. The ace pressure tube was then dipped into an oil bath having
a temperature of 140 °C for experiments with water and 120 °C for
experiments with methanol. The internal temperature during reac-
tion as recorded with an internal thermocouple was 125 °C and
115 °C, respectively.
3. Results and discussion
3.1. Reaction pathway
In the flow reactor set-up, the feed was introduced into the sys-
tem by either a Waters 501 HPLC pump, or by a Knauer K-120 HPLC
pump. The catalyst was held in place in a stainless steel reactor
tube by quartz wool on both sides of the catalyst. The reactor tube
was heated by a Carbolite oven while a type E thermocouple at-
tached to the external surface of the reactor tube was used to mon-
itor the reactor temperature. The effluent was collected in a
stainless steel collector tube. The system was pressurized to
20 bar of pressure with Argon before each run. Feeds were created
by placing a measured amount of substrate in a volumetric flask
and filling to the set volume. For concentrated solutions, the mix-
We have investigated the isomerization of the two three-carbon
sugars dihydroxyacetone (DHA) and glyceraldehyde (GLA) into
methyl lactate and lactic acid over an acidic zeolite catalyst. When
the solvent used is methanol, the resulting product becomes
methyl lactate and if water is used, lactic acid is formed. Scheme
1 gives the proposed reaction path of the isomerization reaction
in either alcohol or water [22–24]. Pyruvaldehyde (PA) is believed
to be an initial product formed by the dehydration of DHA/GLA.
After addition of water/methanol the resulting hydrate/hemiacetal
can isomerize into lactic acid/methyl lactate (path A). The reaction

124
R.M. West et al. / Journal of Catalysis 269 (2010) 122–130
OH
OH
O
HO
OR
GLA
O
O
A
OR
+ROH
-ROH
O
-H2O
O
+ROH
-H2O
O
O
OH
B
-ROH
+H2O
HO
OH
OR
DHA
OR
Scheme 1. Reactive pathways of DHA/GLA to lactate/lactic acid (path A) or hydrated/methylated pyruvic aldehyde derivates (path B), in methanol, (R = CH3) and water
(R = H).
path in Scheme 1 will be discussed in detail along with the presen-
tation of the experimental findings.
Product formation in a typical batch experiment was monitored
as a function of reaction time. Concentration profiles, based on the
carbon balance of the initial triose sugars, are shown in Figs. 1 and
2 for the H-USY-6 zeolite. Fig. 1a and b shows the reaction profiles
for DHA and GLA, respectively, in methanol while Fig. 2a and b
DHA
GLA
PA
100
a
100
a
LacticAcid
Other
80
60
40
20
0
80
60
DHA
GLA
PADA
Methyl Lactate
Other
40
20
0
0
2
4
6
8
24
25
0
1
2
3
20
21
Time (Hours)
Time (Hours)
100
b
100
80
60
40
20
0
b
DHA
GLA
PA
Lactic Acid
Other
80
60
40
20
0
DHA
GLA
PADA
Methyl Lactate
Other
0
2
4
6
8
24 36 48
0
1
2
3
4
5
6
22
23
Time (Hours)
Time (Hours)
Fig. 1. Batch concentration profiles for DHA and GLA in methanol. Other includes
Pyruvic Acid PA, MMP, and TMO. (a) 1.25 mmol DHA in 4.0 g methanol at 115 °C
with 80 mg H-USY-6 catalyst; and (b) 1.25 mmol GLA in 4.0 g methanol at 115 °C
with 80 mg H-USY-6 catalyst.
Fig. 2. Batch concentration profiles for DHA and GLA in water. Other includes
pyruvic acid, formic acid, acetic acid, levulinic acid, and acetol. (a) 1.25 mmol DHA
in 4.0 g water at 125 °C with 80 mg H-USY-6 catalyst; and (b) 1.25 mmol GLA in
4.0 g water at 125 °C with 80 mg H-USY-6 catalyst.

R.M. West et al. / Journal of Catalysis 269 (2010) 122–130
125
Table 1
Yields obtained from DHA/GLA conversion to lactic acid and methyl lactate over commercial catalysts; 80 mg catalyst, 1.25 mmol substrate (calculated as monomer), 4 g solvent
(water or methanol), reaction time 24 h (DHA) or 48 h (GLA), temperature 115 °C (methanol) or 125 °C (water).
Zeolite
Si/Al
Substrate
Water solvent
Methanol solvent
Yield of lactic acid (%)
Substrate conversion (%)
Yield of methyl lactate (%)
Substrate conversion (%)
H-USY-6
H-USY-30
H-beta
6
DHA
GLA
DHA
GLA
DHA
GLA
DHA
GLA
DHA
GLA
DHA
GLA
DHA
GLA
DHA
GLA
DHA
GLA
DHA
GLA
71
63
47
41
63
60
37
37
32
30
22
23
39
32
46
44
39
41
5
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
>99
96
98
26
25
42
63
19
8
17
19
7
7
8
10
29
30
17
37
0
99
>99
79
64
88
90
79
64
76
73
61
59
74
28
44
43
96
91
<0.1
<0.1
30
12.5
19
11.5
25
10
–
H-ZSM-5
H-MOR
H-montmorillonite
Sulphated zirconia
No catalyst
–
–
4
0
shows the profiles in water. The yields of methyl lactate/lactic acid
for a variety of solid acids are summarized in Table 1 and are sim-
ply the end data point of the individual experiments.
3.2. Screening of commercial zeolites
A series of different commercially available zeolites were inves-
tigated in the isomerization of the trioses into lactic acid or methyl
lactate. Table 1 provides an overview of the yields observed in
batch experiments for both GLA and DHA in water and methanol.
From Table 1 we conclude that very similar selectivity toward lac-
tic acid/methyl lactate is obtained using either DHA or GLA as the
substrate, with only H-beta and sulphated zirconia showing a sig-
nificant but reproducible selectivity difference with methanol as
the solvent.
Both GLA and DHA are observed when either substrate is used
as a starting reagent as can be seen from Figs. 1a,b and 2a,b. These
observations indicate that isomerization between these two sugars
occurs and PA is formed as an intermediary product, in accordance
with previous reports [26–28]. DHA reacts more quickly than
does GLA. In water (Fig. 2b), it is noted that GLA first isomerizes
to DHA and the DHA then appears to react. Starting form GLA,
the reaction network appears to be dominated by the isomeriza-
tion and dehydration over the first 6 h. After 6 h, the system ap-
proaches the final distribution seen at 22 h. When DHA is instead
used as a starting reagent, the final distribution is reached in less
than 1 h.
The presence of PA as an intermediate can also be seen in Fig. 2.
The concentration of PA is observed to be highest at very short
reaction times for both DHA and GLA, thus supporting the hypoth-
esis that PA is an initial intermediate. Addition of either water or
methanol forms either hydrated PA or the methyl hemiacetal of
PA (PAMA), respectively, both of which can undergo a 1,2-hydride
shift and isomerize into lactic acid or methyl lactate (Scheme 1,
path A), respectively. The hydrated form of PA can react with an
additional water molecule to form the di-hydrated form of PA. It
is noted that in water, pyruvaldehyde exists in three forms; alde-
hyde, hydrated, and di-hydrated with a typical distribution of
trace, 56% and 44%, respectively [29]. In the analytical methods
used in this study, it was not possible to distinguish these three
species, but all are expected to be present. Additionally, the degra-
dation products seen in water consisted of known PA degradation
products as seen in commercial sources of pyruvaldehyde, namely
pyruvic acid, acetol, and formaldehyde [30].
It is clear from Table 1 that the most effective catalysts for this
isomerization–esterification reaction are the proton form of the
highly acidic ultra stable Y-zeolite with a Si/Al ratio of 6 (H-USY-
6). Indeed, the H-USY-6 zeolite affords methyl lactate yields higher
than 95% within the 24 h reaction time which is comparable to the
most selective catalysts reported so far in the literature [21–24]. In
the absence of a catalyst, DHA and GLA are not converted into
methyl lactate/lactic acid in any significant degree. The trioses
are stable in pure methanol, but degrade readily in water to form
a product mixture consisting of carbonaceous products along with
pyruvic aldehyde and some lactic acid. The H-beta zeolite gives
moderate yields (DHA 42%/GLA 62%) of methyl lactate, whereas
the best H-ZSM-5 catalyst affords only a low yield (<20%) of methyl
lactate. In general as the Si/Al ratios are lowered (higher acid den-
sity) a higher triose conversion and moderately higher selectivity
for methyl lactate is achieved. In a previous work zeolite beta
was tested in the isomerization of DHA/GLA to methyl lactate/lac-
tic acid. It was shown that dealumination of zeolite beta by steam-
ing increased the yields of methyl lactate/lactic acid relative to the
non-dealuminated catalyst thus indicating that the presence of ex-
tra framework aluminum was beneficial for the reaction. However,
other factors such as, the effective diffusion rates, acid densities as
well as acid strength could also have important roles in determin-
ing the activity of the various zeolite structures.
The same is not true for the methanol case. The PAMA was ob-
served directly as an intermediate species and it can react with an-
other molecule of methanol to form pyruvaldehyde dimethylacetal
(PADA), (Scheme 1, path B). PADA has additionally been observed
to react further in very small amounts to the methylated form
TMP (1,1,2,2-tetramethoxy propane) at low to intermediate con-
versions. At full conversion, however, PADA, TMP, and PAMA are
converted into methyl lactate (Fig. 1) suggesting the back reaction
shown in Scheme 1.
3.3. Characterization by FT-IR
Due to the very large difference in the yield of methyl lactate of
the two H-USY samples (98% and 25%, from Table 1) we chose to
investigate these two samples by FT-IR using pyridine as a molec-

126
R.M. West et al. / Journal of Catalysis 269 (2010) 122–130
Fig. 3. IR difference spectra after pyridine saturation and subsequent evacuation at 200 °C for 30 min: (a) H-USY-6; and (b) H-USY-30.
100
ular probe while combined the results with acidity measurements
from NH3-TPD. The dehydrated spectra of the two samples (not
shown) reveal the high frequency bridging hydroxyl (HF(OH)) lo-
cated within the supercage of the zeolite at 3632 cmꢁ1, and the
low frequency bridging hydroxyl (LF(OH)) present in the sodalite
a
90
80
70
60
cage at 3566 cmꢁ1 [31]. In addition, contributions at 3602 cmꢁ1
,
3526 cmꢁ1, and 3674 cmꢁ1 are seen in the case of the H-USY-6
sample. The two former bands located at 3602 cmꢁ1 (HF’(OH))
and 3526 cmꢁ1 (LF’(OH)) have been reported to occur due to per-
turbation by extra framework silicoaluminous debris formed upon
dealumination of zeolite Y [32,33]. Further, the 3674 cmꢁ1 band is
said to arise from hydroxyl groups linked to extra framework alu-
minum (EFAL) species [33]. From pyridine adsorption we were able
to differentiate between Lewis and Brønsted acidity while the NH3-
TPD provided a measure of the total acidity. Fig. 3 presents the dif-
ference spectra of the pyridine saturated samples after evacuation
at 200 cmꢁ1 C for 30 min. The relative Brønsted and Lewis acidity
ratio could be obtained from integrating the areas under the bands
at 1545 cmꢁ1 (pyridinium ion) and 1455 cmꢁ1 (pyridine) while
taking the extinction coefficients into account [34,35]. Using values
DHA
GLA
PA
LacticAcid
Other
15
10
5
0
0
5
10
15
20
25
30
35
40
45
Time on Stream (Hours)
b
100
90
of eBrønsted = 1.67 and eLewis = 2.22 as reported in Ref. [35], we ob-
tained values of B/L ratio of 1.8 for H-USY-6 and 5.6 for the H-
USY-30 zeolite. From NH3-TPD we found that the total acidity of
the H-USY-6 and H-USY-30 samples was 563
lmol/gzeolite and
80
199 mol/gzeolite, respectively. Interestingly, these results show
l
that a large fraction of the acidity of the H-USY-6 zeolite actually
consists of Lewis acidic sites. Also the total number of Lewis acid
sites per gram of zeolite is much larger for H-USY-6 as compared
DHA
GLA
PADA
MethylLactate
Other
70
60
to the H-USY-30 zeolite (ꢂ200
lmol/gzeolite and ꢂ30
lmol/gzeolite,
respectively). These findings are in agreement with an earlier re-
port showing that a purely Lewis acidic zeolite beta is highly active
for the formation of methyl lactate from triose sugars [24] and they
support the hypothesis that the Brønsted and Lewis acidic sites can
catalyze two different reaction paths, with only the Lewis acidic
sites leading to the formation of lactic acid/methyl lactate in appre-
ciable amounts from DHA/GLA.
15
10
5
0
0
10
20
30
40
50
3.4. Flow experiments
Time on Stream (Hours)
Fig. 4. Concentration profiles vs. time on stream for 0.31M DHA over H-USY-6
catalyst. Other includes pyruvic acid and TMO. (a) DHA conversion in water at
177 °C, WHSV of 0.18 hꢁ1. Other includes pyruvic acid, formic acid, acetic acid,
levulinic acid, and acetol; and (b) DHA conversion in methanol at 157 °C, WHSV of
The high yields obtained when using the H-USY-6 zeolite as cat-
alyst merited further study in continuous flow mode. The stability
of the H-USY-6 catalyst at high conversion in water and in metha-
nol was investigated for approximately 45 h on stream. Reaction
0.16 hꢁ1
.

R.M. West et al. / Journal of Catalysis 269 (2010) 122–130
127
conditions for the two experiments were chosen to achieve full
conversion initially, and a high yield of products, i.e. in the case
of water, a reaction temperature of 177 °C, concentration of
0.31 M DHA, and a WHSV of 0.18 hꢁ1 was chosen, while for meth-
anol a temperature of 157 °C, concentration of 0.31 M DHA, and
WHSV of 0.16 hꢁ1 was used. Fig. 4 presents the observed product
compositions obtained by sampling the effluent as a function of
time on stream.
In Fig. 4a while the initial yield of lactic acid in water was quite
high, 95%, the sample showed continual deactivation. After around
25 h, the concentration of both unreacted feed and pyruvic alde-
hyde (PA) increased significantly, while the yield of lactic acid de-
creased sharply. This deactivation continued for the duration of the
run reaching a ꢂ70% yield of lactic acid after 44 h. In the case of
methanol as a solvent we observed a much improved stability
within the 47 h as shown in Fig. 4b. The yield of methyl lactate re-
mained around or above 90% for the duration of the run. As the
yield of methyl lactate drops slowly, the yield of pyruvic aldehyde
dimethyl acetal (PADA) increases. Only low concentrations of unre-
acted triose sugars were observed in the effluent.
tries 4 and 7) showed quite different results. The total disappear-
ance rate when GLA is the starting reagent is only 305
mol/
l
*
(min gcat), while this value is 425 when DHA is the starting re-
agent. Likewise, the appearance of lactic acid from GLA is only
120 compared to 355 for DHA at the same conditions. The observa-
tion suggests isomerization of GLA to DHA readily occurs with the
DHA ultimately reacting as discussed later. When the suspected
intermediate species PA is reacted at similar conditions, (entry
8), a higher disappearance of feed of 820 and an appearance of lac-
tic acid of 770 is observed. This observation appears reasonable
since the dehydration step is surpassed and the kinetics of isomer-
ization are quite favorable as discussed in a later section.
When the feed concentration of DHA is increased at any tem-
perature, the disappearance of triose sugars is seen to likewise in-
crease (entries 2, 9, and 10 and entries 4, 11, and 13). However, the
appearance of lactic acid is not observed to increase correspond-
ingly. In Table 1, it is noted that when given enough time, complete
conversion of triose sugars to products other than lactic acid is pos-
sible without catalyst in water. It is believed that similar degrada-
tion reactions are occurring when the concentration of feeds is
increased beyond 0.31 M.
3.5. Kinetic study in the flow system
Switching the solvent to methanol (entries 14–19) caused a
drastic difference in the observed rates. For a 1.25 M feed solution
of DHA, the initial disappearance of triose sugars increased by a
factor of ꢂ7 from 775 in water at 160 °C (entry 11) to 5590 in
methanol at 150 °C (entry 18), while the appearance increased by
a factor of ꢂ8 from 405 for lactic acid to 3315 for methyl lactate.
A similar trend is seen at 130 °C where the rate of disappearance
increases from 225 to 1305, while the rate of appearance increases
from 50 to 1060 (entries 2 and 15, respectively).
Interestingly, using a mixed solvent of methanol and water (en-
try 17) produced disappearance and production rates between that
of pure methanol and water. For a 1.25 M solution of DHA, a 45/55
molar mixture of methanol and water (64.5 wt.% methanol) at
150 °C showed a disappearance rate of 1300, double that of pure
water, 775 (entry 11), but much less than that of pure methanol,
5590 (entry 18). The rate of appearance of product, a mixture of
lactic acid and methyl lactate for the 45/55 mixture was 860, again
about double that of pure water, 405, and much less than that of
pure methanol 3315. A mixture of the solvents is indeed an inter-
esting situation since for every one mole of DHA reacted, one mole
of water is created in an alcohol solvent meaning that minor
In order to better understand the current system, a simulta-
neous investigation of the kinetics and deactivation under various
reactions conditions were conducted. The temperature, feed con-
centration, and feed substrates were systematically varied as
shown in Table 2 while the catalyst loading was adjusted to keep
the initial disappearance of reagent to less than 70% and the yield
of lactic acid/methyl lactate to less than 50%. The initial activity
and calculated initial activity of the H-USY-6 as a function of sys-
tematically altered reaction conditions are shown in Table 2. The
initial measured activities are given for the disappearance of triose
sugars (total disappearance of GLA and DHA), the appearance of
lactic acid/methyl lactate, and for the appearance of intermediate
species PA/PAMA (pyruvic aldehyde methyl acetal), and PADA,
(depending on the solvents). The calculated values were fit with
a simple kinetic model while as described in detail in the modeling
section below.
As one would expect, the rates increased as the temperature
was increased at otherwise constant conditions, (entries 1–6).
The activity of DHA and GLA in water at ꢂ160 °C and 0.31 M (en-
Table 2
Initial rates of reaction under systematically varied conditions using H-USY-6. The total number of acid sites are 563 umol/g based on NH3-deposition. The number of Brønsted
and Lewis acid sites corresponds to 362
lmol/g and 200 lmol/g, respectively (based on pyridine IR-adsorption).
Entry Conditions of run
Measured initial rate (lmol/(min*gcat))
Calculated rate (lmol/(min*gcat))
Temp. (°C) Reagent Solvent
Concentration Triose LA/ML
PA/PAMA
PADA
Triose
LA appearance PA appearance
(M)
dissappearance appearance appearance appearance dissappearance
1
2
3
4
5
6
7
8
116
130
142
165
172
190
160
160
130
130
160
160
160
112
130
130
150
150
165
DHA
DHA
DHA
DHA
DHA
DHA
GLA
PA
DHA
DHA
DHA
PA
DHA
DHA
DHA
GLA
DHA
DHA
DHA
Water
Water
Water
Water
Water
Water
Water
Water
Water
Water
Water
Water
Water
MeOH
MeOH
MeOH
0.31
0.31
0.31
0.31
0.31
0.31
0.31
0.36
0.80
1.25
1.25
1.25
3.44
1.25
1.25
1.25
27
75
17
50
2
8
15
40
32
75
110
450
525
1545
23
40
65
315
380
1075
8
35
35
90
75
230
100
425
550
1530
305
–
180
225
775
–
2475
735
1305
1270
920
5590
10080
100
355
445
1055
120
770
50
25
205
30
820a
30
–
820
90
120
415
855
1495
930a
110
130
240
975a
785
9
210
260
710
–
10
11
12
13
14
15
16
17
18
19
30
45
405
780
340
620
1060
800
865
3315
4600
125
865a
405
30
35
25
95
385
905
2480
45
300
190
150
1400
2850
MeOH/water 1.25
MeOH
MeOH
1.25
1.25
a
Disappearance of PA.

128
R.M. West et al. / Journal of Catalysis 269 (2010) 122–130
amounts of water are inevitably present. Further, this situation is
important if operating in solvents of only technical grade. Impor-
tantly as just outlined above using the mixture of water and meth-
anol did increase the rates with respect to those observed in pure
water. However, reaction rates in the solvent mixture are still
much closer to the rates seen in water than those in pure
methanol.
nia-TPD, porosity measurements, and the estimated crystallinity
all shows that less framework deterioration takes place during
the reaction in methanol compared to water (entry 3A). However,
a modest loss in micropore volume, surface area, and acidity is
seen along with a decrease in the calculated crystallinity of around
40% indicating that the catalyst has suffered some structural dam-
age in excess of the deactivation by carbon deposits.
Entries 5 and 6 shows that the highest deposition of carbon
takes place when PA is used as the feed, (Table 2 entries 8 and
12, respectively). In 9 h at 160 °C, 5.2% of carbon had deposited
onto the H-USY-6 zeolite when using a feed of 0.36 M PA as sub-
strate. Increasing the PA concentration aggravates the deposition
of carbon, since 15.7% is deposited in only 5 h when a feed consist-
ing of 1.25 M PA was used. In contrast, after running for 44 h, a
0.31 M DHA solution in water amounted to 9.0% carbon. In addi-
tion to the excessive carbon deposits, high surface area and micro-
pore volume losses were also observed (entry 5). It is also noted
that the deactivation of the catalyst in the two runs using PA as a
feed was also the most rapid among all feeds and conditions tested.
Entries 7 and 8 in Table 4 were created by flushing the H-USY-6
zeolite with either a 0.31 M lactic acid solution in water or a 0.31 M
methyl lactate solution in methanol overnight at the specified tem-
perature. The lactic acid and methyl lactate were quantitatively
recovered in the effluent indicating that the reverse reactions do
not occur. These experiments show that the desired end products
in methanol and water, methyl lactate and lactic acid, are not the
cause for the excessive carbon deposition onto the zeolite. Impor-
tantly, we observe that the product in water, lactic acid, strongly
damages the catalyst already after 12 h. Again this is in strong con-
trast to what is observed in the case of methyl lactate in methanol
where more modest changes in the structure of the catalyst were
detected for the comparable reaction conditions (entries 4 and
8). Although the XRD show a decrease of 41.2% for this sample
the effect on micropore volume, surface area and total acidity of
the zeolite is much smaller.
3.6. Deactivation
The reasons for deactivation were investigated by characteriz-
ing fresh and used catalysts. Table 4 demonstrates the surface
properties for fresh and selected used catalyst collects from Table
3 as well as H-USY-6 zeolites under modeled deactivating condi-
tions. Table 4 entry 1 represents the fresh, unused H-USY-6 zeolite
which is used for comparison. Entries 2–8 contains data for the
used catalysts and were chosen as to cover the range of reaction
times, temperatures, feeds, concentrations, and solvents used.
Entry 2 was tested at low temperature (116 °C) and short time
(6 h) on stream while entry 3A corresponds to the zeolite used in
Fig. 3a namely a temperature of 177 °C and time on stream 44 h
in water. Entry 3B is the same sample without calcination to show
the influence of carbon on the surface area. Characterizing the
three used zeolites we note that the adsorption characteristic of
entry 2 is very similar to the parent zeolite exemplified by a micro-
pore volume which only decreased from 0.253 ml/g to 0.236 ml/g
indicating that the zeolite structure is mostly preserved. This is
in strong contrast to the prolonged and high temperature reaction
used for entry 3A which resulted in an almost complete loss of
crystal integrity. This can be seen by a reduction in the measured
surface area and micropore volume of around 80%, a loss of acidity
of 46% as well as a final crystallinity calculated to <10%. Further-
more, the catalyst is heavily covered in carbon, containing as much
as 9 wt.% carbon.
Entry 4 shows data for the zeolite after a time on stream of 47 h
at 157 °C in the conversion of DHA in methanol (Fig. 3b). Ammo-
The specific cause of irreversible zeolite destruction was further
systematically tested under controlled conditions. Entries
9
through 15 were created by exposing the fresh H-USY-6 zeolite
to the specified conditions listed in a sealed teflon container ro-
tated at 20 rpm. The zeolite was subsequently recovered by filtra-
tion and washed thoroughly with demineralized water and dried
before analysis.
Stirring the zeolite in demineralized water for 24 or 48 h at
140 °C (entries 9 and 10) slightly decreased the surface areas and
Table 3
Modeled kinetic parameters for Scheme 1 on H-USY-6.
Ea (kJ/mol K)
ln(A) (1/mass)
1
2
3
53 13
61 15
89
17.4 8.9
20 13
26
Table 4
Physical properties of fresh, used, and treated H-USY-6 catalysts.
Entry
Temperature
(°C)
Solvent
Feed
Concentration
(M)
Time (h)
Location
TPO
(% C)
BET
Vmicro
(mL/g)
Vmeso
(mL/g)
Smeso
NH3-TPD
(umol/g)
XRD,
C/C0 (%)
(m2/g)
(m2/g)
1
2
–
–
–
–
–
6
44
44
47
9
–
–
2.0
9.0
725
674
182
115
619
246
0.253
0.236
0.048
0.015
0.217
0.076
0.058
0.076
0.198
0.188
0.060
0.102
134
142
97
71
128
79
563
306
461
100
8.6
59
116
177
177
157
160
160
177
137
140
140
140
140
140
140
140
Water
Water
Water
MeOH
Water
Water
Water
MeOH
Water
Water
Water
Water
Water
Water
MeOH
DHA
DHA
DHA
DHA
PA
PA
LA
ML
–
–
0.31
0.31
0.31
0.31
0.36
1.25
0.31
0.31
–
Reactor
Reactor
Reactor
Reactor
Reactor
Reactor
Reactor
Reactor
Vial
Vial
Vial
Vial
Vial
3A
3B
4
2.7
5.2
15.7
1.7
1.8
5
6
7
8
5
12
11
24
48
2
405
708
555
554
716
670
537
238
693
0.129
0.246
0.211
0.208
0.248
0.232
0.188
0.017
0.251
0.151
0.071
0.089
0.108
0.094
0.106
0.118
0.290
0.057
127
142
108
108
164
164
129
223
131
394
486
561
34
56
80
9
10
11
12
13
14
15
–
LA
LA
LA
LA
0.3
0.3
0.3
1
8
24
24
24
490
158
486
71
ꢃ0
93
Vial
Vial
0.4
–
–
A
B
Calcined
uncalcined.

R.M. West et al. / Journal of Catalysis 269 (2010) 122–130
129
crystallinity of the sample, however it did not markedly decrease
the total acidity of the zeolite.
in Table 3 are quite large. This variance is primarily due to rela-
tively good fit over a range of values; that is similar forward rate
constants can be calculated with a range of pre-exponential and
activation energies. The initial disappearance of triose sugars was
less than 70% while the initial yield was less than 50%, hence the
model used incorporated the changing concentrations of the re-
agents through the reactor. This reaction model is not differential
and hence the confidence intervals are quite large. Despite these
large intervals, several important conclusions can be made by cal-
culating the forward rate constants at the best fit values for each
step in Scheme 2. It is observed that the forward rate constant
for isomerization is the largest forward rate constant at all temper-
atures of this study. The ratio of k2/k1 increases as the temperature
is increased from a value of 1.1 at 115 °C to 1.7 at 190 °C. This high-
er value for k2 implies that the isomerization step occurs more
readily than the dehydration step and that the isomerization is rel-
atively more favorable as the temperature is increased. A second
trend can be seen by comparing the degradation and isomerization
constants, k2/k3. At 115 °C this value is 14.6, heavily favoring the
isomerization. As the temperature is raised however to 190 °C, this
ratio drops swiftly to 3.6 indicating that the degradation reactions
become much more favorable at higher temperatures.
Carefully repeating the experiment by using a 0.3 M solution of
lactic acid and a shorter treatment period caused a more pro-
nounced destruction of the zeolite than pure water. Entries 11–
14 show the systematic decrease in surface areas, and micropore
volume corresponding to treatment times of 2, 8, and 24 h in
0.3 M lactic acid. At 24 h, the surface areas number of acid sites
and crystallinity of the sample subjected to 0.3 M lactic acid (entry
13) were all lower than the corresponding values in pure water
(entry 9) at 24 h. Further increasing the concentration of lactic acid
to 1.0 M (entry 14) caused a complete destruction of the zeolite
with no recognizable diffraction peaks present in the XPRD diffrac-
togram, i.e. the crystallinity of this sample was ꢃ0%. Results from
NH3-TPD and N2-sorption analysis support the complete disman-
tling of the zeolite. Importantly, a concentration of approximately
0.3 M of lactic acid, which is the concentration of final product in
either the batch or flow experiments, is sufficient to destroy the
zeolite suggesting that it is necessary to work with dilute systems
if water is used as the solvent. In entry 8 we used anhydrous meth-
anol for the treatment. It is however clear from the results that the
zeolite is extremely stable in pure methanol as the zeolite charac-
teristics are practically identical to the fresh zeolite.
The model predicts the initial rates for a variety of concentra-
tions, temperatures, and feeds relatively accurate. The differences
between the calculated and observed initial rate of disappearance
of triose species and pyruvic aldehyde are less than 20% for con-
centrations ranging from 0.31 M to 3.44 M and temperatures from
116 to 190 °C indicating the system is well described by the
assumptions of first order kinetics and simple activation energies.
The estimated production rate of lactic acid is also accurately mod-
eled for the tests at low concentration. Deviations between the
model and experimental data are largest at high concentrations
with the largest differences between the predicted and observed
rates occurring for the production of LA in entries 10 (131 °C
1.25 M) and 13 (159 °C 3.44 M). Since the reaction pathway is be-
lieved to be a series of reactions, and since the model does not in-
clude surface coverage terms, one possible explanation is that the
catalytic sites are covered by either initial species DHA or water,
and are blocked from reaction of DHA. A second explanation could
include a diffusion limitation at high concentration. The surface
blocking explanation is better supported, however, by the close
fit between experimental and calculated rates when the intermedi-
ate species, PA is used. At higher concentrations of PA (entry 12)
the rate of lactic acid production is relatively close to the rate of
PA disappearance, while the same is not true of DHA at the same
conditions (entry 11). If the system was diffusion limited, one
would expect large differences for both cases, however if the initial
reactivity of DHA is limited, large deviations would only be ex-
pected for the case starting with DHA, as observed.
As highlighted from the characterization above it is important
to distinguish between the reversible deactivation occurring
through coking of the zeolite in contrast to the irreversible deacti-
vation through structural damage of the zeolite. The results indi-
cate that coking originates from decomposition of PA and not
from the end products. The conditions that lead to the most rapid
loss in activity can be summarized as follows. Higher temperatures
and higher concentration of lactic acid (either directly or through
the formation from PA) are detrimental to the zeolite structure.
Using water compared to methanol as a solvent causes a higher le-
vel of irreversible damage. This difference arising from the solvent
is further increased as lactic acid also causes far more destruction
than does methyl lactate. Minimal deactivation through destruc-
tion of the zeolite can therefore be expected when operating at
low concentrations, low temperature, and when using an alcohol
as the solvent.
3.7. Modeling the reaction path
The initial rate of reaction was modeled as a plug flow reactor
with a series of simple first order reactions as shown in Scheme
2 for all runs involving DHA and PA with water as a solvent. The
forward rate constants were assumed to be Arrhenius expressions
E
RT
Ai
of the form, ki ¼ Ai exp . The pre-exponential factor A and activa-
tion energy of each step were then estimated by integrating over
the mass of catalyst in Matlab using the lsqnonlin function to min-
imize the differences in calculated and experimental rates for the
disappearance of trioses, appearance of pyruvaldehyde, and the
appearance of lactic acid. The fit parameters for each step are found
in Table 3, while the calculated rates are included in Table 2.
Despite fairly good agreement over the large temperature and
concentration range of this study, the confidence intervals seen
Previous authors have modeled the degradation of triose sugars
under isothermal conditions without a catalyst [26,27]. These stud-
ies looked at the isomerization and degradation of GLA and DHA
into PA. As in the current case, first-order kinetics were found to
appropriately model the disappearances of triose sugars and PA.
As expected, the heterogeneously catalyzed activation energy
OH
O
O
2
1
OH
O
HO
OH
3
O
Scheme 2. Modeled reaction scheme for the conversion of DHA into LA.

130
R.M. West et al. / Journal of Catalysis 269 (2010) 122–130
determined here is much lower than previous reports for the dehy-
dration of triose sugars (53 13 vs. 75–92 kJ/mol, respectively).
The degradation/isomerization of PA determined in this report
are comparable to previous values, with the isomerization having
a lower activation energy (61 15 kJ/mol) and the degradation
reaction having similar value (89 kJ/mol) to the other reports
(77–94 kJ/mol). It is important to note that due to the aqueous
environment of this and previous studies, PA is present as the hy-
drated and dehydrated forms. This implies that the degradation
reactions of PA (as the mono- and di-hydrate) would proceed even
in the absence of heterogeneous catalyst and further suggests
avoidance by operation at lower temperatures.
From the modeling study, we see that the optimal reactions
conditions for the production of a lactic acid derivative are low
concentrations in an alcohol solvent such as methanol at low tem-
peratures. The experimental data for the isomerization reactions
can despite a relative large confidence interval be described by a
series of first-order reaction and the model can be used to gain use-
ful information of the involved rate constants. Specifically, the
higher temperatures favor isomerization, k2 over dehydration k1,
but also favor degradation k3 over isomerization k2. To limit degra-
dation to other products, it is therefore suggested that the reaction
be run at lower temperatures.
Acknowledgments
Supported by the Danish National Research Foundation, and the
National Science Foundation, PIRE Program (Award # 0730277).
We also thank Dr. J. Rass-Hansen and R. Johansson for technical
assistance.
References
[1] R. Dhatta, M. Henry, J. Chem. Technol. Biotechnol. 81 (2006) 1119.
[2] R.A. Sheldon, Green Chem. 7 (2005) 267.
[3] Y.-J. Wee, J.-N. Kim, H.-W. Ryu, Food Technol. Biotechnol. 44 (2) (2006) 163.
[4] R. Leaversuch, Plast. Technol. 48 (3) (2002) 50.
[5] S. Varadarajan, D.J. Miller, Biotechnol. Prog. 15 (1999) 845.
[6] H.F. Shi, Y.C. Hu, Y. Wang, H. Huang, Chin. Chem. Lett. 18 (2007) 476.
[7] R.D. Cortright, M. Sanchez-Castillo, J.A. Dumesic, Appl. Catal. B: Environ. 39
(2002) 353.
[8] J.C. Serrano-Ruiz, J.A. Dumesic, Chem. Sus. Chem. 2 (6) (2009) 581.
[9] D. Garlotta, J. Polym. Environ. 9 (2) (2001) 63.
[10] R.P. John, K.M. Nampoothiri, A. Pandey, Appl. Microbiol. Biotechnol. 74 (2007)
524.
[11] H. Kimura, K. Tsuto, Appl. Catal. A: Gen. 96 (1993) 217.
[12] H. Kimura, Appl. Catal. A: Gen. 105 (1993) 147.
[13] R. Garcia, M. Besson, P. Gallezot, Appl. Catal. A: Gen. 127 (1995) 165.
[14] E. Farnetti, J. Kaspar, C. Crotti, Green Chem. 11 (2009) 704.
[15] D. Hekmat, R. Bauer, V. Neff, Process Biochem. 42 (2007) 71.
[16] S. Yamada, K. Nabe, N. Izuo, M. Wada, I. Chibata, J. Ferment. Technol. 57 (3)
(1979) 215.
[17] K. Nabe, N. Izuo, S. Yamada, I. Chibata, Appl. Environ. Microbiol. 38 (6) (1979)
1056.
[18] M.C. Flickinger, D. Perlman, Appl. Environ. Microbiol. 33 (3) (1977) 706.
[19] M.J. Antal Jr., W.S.L. Mok, G.N. Richards, Carbohydr. Res. 199 (1990) 111.
[20] H. Kishida, F. Jin, X. Yan, T. Moriya, H. Enomoto, Carbohydr. Res. 341 (2006)
2619.
[21] M. Bicker, S. Endres, L. Ott, H. Vogel, J. Mol. Catal. A: Chem. 239 (2005) 151.
[22] Y. Hayashi, Y. Sasaki, Chem. Commun. (2005) 2716.
[23] K.P.F. Janssen, J.S. Paul, B.F. Sels, P.A. Jacobs, Stud. Surf. Sci. Catal. 170B (2007)
1222.
[24] E. Taarning, S. Saravanamurugan, M. Spangsberg Holm, J. Xiong, R.W. West,
C.H. Christensen, Chem. Sus. Chem. 2 (7) (2009) 625.
[25] S. Sombatchaisak, P. Praserthdam, C. Chaisuk, J. Panpranot, Ind. Eng. Chem. Res.
43 (2004) 4066.
[26] G. Bonn, M. Rinderer, O. Bobleter, J. Carbohydr. Chem. 4 (1) (1985) 67.
[27] B.M. Kabyemela, T. Adschiri, R. Malaluan, K. Arai, Ind. Eng. Chem. Res. 36
(1997) 2025.
[28] G. Lookhart, M.S. Feather, Carbohydr. Res. 60 (1978) 259.
[29] A.C. McLellan, P.J. Thornalley, Anal. Chim. Acta 263 (1992) 137.
[30] C. Rae, S.J. Berners-Price, B.T. Bulliman, P.W. Kuchel, Eur. J. Biochem. 193
(1990) 83.
[31] F. Wakabayashi, J.N. Kondo, K. Domen, C. Hirose, Microporous Mater. 8 (1997)
29.
[32] O. Cairon, T. Chevreau, J.-C. Lavalley, J. Chem., Soc., Faraday Trans. 94 (1998)
3039.
4. Conclusion
It has been demonstrated that the Lewis acidic H-USY-6 zeolite
is highly effective for the isomerization of triose sugars to lactic
acid and methyl lactate. Indeed, methyl lactate can be formed in al-
most quantitative yields directly from DHA and GLA, using this
inexpensive zeolite. The transformation can be modeled via a ser-
ies of first order reactions in water. In methanol, the rate of reac-
tion is as much as an order of magnitude higher than in water.
The catalyst deactivates both by carbon deposition and by frame-
work degradation. However, deactivation by carbon deposition as
well as structural damage can be minimized by using optimized
reaction conditions of low concentration and temperature and
with an alcohol such as methanol as the solvent. The presented
work adds to the pool of information regarding this very interest-
ing process wherein it is possible to produce racemic lactic acid or
the corresponding methyl ester from biomass-derived substrates.
Furthermore, since methyl lactate is formed directly in this pro-
cess, purification to form high-grade lactic acid could be facilitated
by distillation and hydrolysis, thereby avoiding the cumbersome
purification process associated with the current production of lac-
tic acid.
[33] S. Kotrel, J.H. Lunsford, H. Knözinger, J. Phys. Chem. B 105 (2001) 3917.
[34] T. Barzetti, E. Selli, D. Moscotti, L. Forni, J. Chem. Soc. Faraday Trans. 92 (8)
(1996) 1401.
[35] C.A. Emeis, J. Catal. 141 (1993) 347.