Catalytic conversion of dihydroxyacetone to lactic acid using metal salts in water

Article abstract of DOI:10.1002/cssc.201000457

We herein present a study on the application of homogeneous catalysts in the form of metal salts on the conversion of trioses, such as dihydroxyacetone (DHA), and glyceraldehyde (GLY) to lactic acid (LA) in water. A wide range of metal salts (26 in total) were examined. Al^III salts were identified as the most promising and essentially quantitative LA yields (>90 mol %) were obtained at 140 °C and a reaction time of 90 min. A reaction pathway is proposed and a kinetic model using the power law approach was developed for the conversion of DHA to LA with pyruvaldehyde (PRV) as the intermediate. Good agreement between experimental data and the model was obtained. Model predictions, supported by experiments, indicate that a high yield of LA is favoured in dilute solutions of DHA (0.1 M) at elevated temperatures (180 °C) and reaction times less than 10. Copyright

Full text of DOI:10.1002/cssc.201000457

DOI: 10.1002/cssc.201000457
Catalytic Conversion of Dihydroxyacetone to Lactic Acid
Using Metal Salts in Water
Carolus B. Rasrendra,[a, b] Boy A. Fachri,[a] I. Gusti B. N. Makertihartha,[b]
Sanggono Adisasmito,[b] and Hero J. Heeres*[a]
We herein present a study on the application of homogeneous
catalysts in the form of metal salts on the conversion of trioses,
such as dihydroxyacetone (DHA), and glyceraldehyde (GLY) to
lactic acid (LA) in water. A wide range of metal salts (26 in
total) were examined. AlIII salts were identified as the most
promising and essentially quantitative LA yields (>90 mol%)
were obtained at 1408C and a reaction time of 90 min. A reac-
tion pathway is proposed and a kinetic model using the power
law approach was developed for the conversion of DHA to LA
with pyruvaldehyde (PRV) as the intermediate. Good agree-
ment between experimental data and the model was obtained.
Model predictions, supported by experiments, indicate that a
high yield of LA is favoured in dilute solutions of DHA (0.1m)
at elevated temperatures (1808C) and reaction times less than
10 min.
Introduction
High oil prices and environmental concerns have stimulated re-
search on biofuels and biobased chemicals from lignocellulosic
biomass. Attractive catalytic conversions to platform chemicals
have been identified[1] and activities are ongoing to convert
these findings to economically feasible processes on a large
scale.[2] Examples of interesting platform chemicals are organic
acids, such as levulinic acid, propionic acid, acrylic acid and, in
particular, lactic acid (LA).
are typically performed in subcritical water (T>2508C) in the
presence of homogenous catalysts, such as bases (NaOH and
Ca(OH)2)[7b,e] or metal salts.[7a,c,g,8a] However, the reported yields
of LA are relatively low (<45 mol%). Interestingly, these stud-
ies imply that trioses, such as DHA, GLY, as well as pyruvalde-
hyde (PRV), are intermediates when converting C6 sugars into
LA (Scheme 1).
LA is a commodity chemical with an estimated demand of
300000 tonnes per year (2010).[3] Traditionally, LA is used in
food and food-related applications, such as bakery and meat
products and confectionery.[3] In these applications, LA serves
as a pH regulator, as a preservative or as an emulsifying agent.
LA is also used for non-food applications and examples are
found in cosmetic and pharmaceutical formulations.[3] In the
last decade, the commercialisation of polylactic acid (PLA),
which is a biopolymer with interesting applications, has boost-
ed the production of LA considerably. Market projections indi-
cate that the PLA market may reach up to 3 million tonnes in
2020.[4] Furthermore, LA also has potential as a platform chemi-
cal for the production of bulk chemicals, such as propylene
glycol and acrylic acid.[5]
Scheme 1. Conversion of C6 sugars to LA.
These findings stimulated research activities on the conver-
sion of trioses, such as DHA and GLY, into LA. Homo- and het-
erogeneous catalysts have been explored, mainly in protic sol-
vents, such as water and alcohols (methanol and ethanol), and
an overview is given in Table 1. When using dilute sulfuric acid
as the catalyst in subcritical water (2508C, 24.5 MPa), Antal
et al. observed the formation of LA in low yields (15 mol%;
Table 1, entry 1).[8a] Bicker et al. reported catalytic effects of var-
LA is currently produced by fermentation processes using
carbohydrate sources. However, volumetric production rates
are relatively low and product workup leads to the formation
of large amounts of salts.[3,6] Therefore, novel methods to pro-
duce LA from lignocellulosic biomass through non-fermenta-
tive catalytic chemical transformations have attracted great in-
terest in both academia and industry.
[a] C. B. Rasrendra, B. A. Fachri, Prof. Dr. H. J. Heeres
Chemical Engineering Department, University of Groningen
Nijenborgh 4, 9747 AG Groningen (The Netherlands)
Fax: (+31)50 3634479
A number of studies have been reported in the literature
and involve the use of C6 sugars (d-glucose and d-fructose),[7]
C6 sugar precursors, such as cellulose,[7c] and trioses,[7a,g,8] such
as dihydroxyacetone (DHA) and glyceraldehyde (GLY), as the
feedstock. The reactions with C6 sugars or C6 sugar precursors
E-mail: h.j.heeres@rug.nl
[b] C. B. Rasrendra, Dr. I. G. B. N. Makertihartha, Dr. S. Adisasmito
Chemical Engineering Department, Institut Teknologi Bandung
Ganesha 10 Bandung 40132 (Indonesia)
768
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2011, 4, 768 – 777

Conversion of Dihydroxyacetone to Lactic Acid
Table 1. Overview of catalytic conversions of DHA to LA or alkyl lactates.
Entry Solvent [DHA]
Catalyst
Metal/DHA ratio [mol%] Conditions
Conversion Yield [mol%] Ref.
1
2
3
4
5
6
7
8
H2O
H2O
H2O
H2O
H2O
H2O
H2O
H2O
MeOH
MeOH
MeOH
MeOH
EtOH
MeOH
0.1m
1 wt%
0.1m
40 wt% Cr resin
40 wt% Sn resin
40 wt% Al resin
0.25m
0.31m
0.625m SnCl2 (62.5mm)
0.625m CrCl3·6H2O (62.5 mm)
0.625m AlCl3·6H2O (62.5 mm)
0.25m
0.4m
0.25m
H2SO4 (1 mm)
ZnSO4 (400 ppm)
AlCl3·6H2O (5 mm)
–
2.2
5
0.2–5[d]
0.2–5[d]
0.2–5[d]
–
–
10
10
10
–
2508C, 24.5 MPa, 33s
75
–
15[a]
86[b]
90[a]
86[a]
68[a]
42[a]
90[a]
71[a]
89[c]
50[c]
62[c]
99[c]
59[c]
96[c]
[8a]
[7a]
[7g]
3008C, 25 MPa, 30s
1408C, 90 min
1008C, 5h
1008C, 5h
1008C, 5h
1258C, 24h
1258C, 24h
908C, 3h
908C, 3h
908C, 3h
808C, 24h
908C, 6h
1058C, 24h
100
100
100
100
100
99
–
–
–
100
77
[8b]
Sn-beta zeolite (80 mg, Si/Sn=125)
H-USY zeolite (80 mg, Si/Al=6)
[8f]
[8h]
9
10
11
12
13
14
[8e]
Sn-beta zeolite (80 mg, Si/Sn=125)
USY-CBV 600 zeolite (200 mg, SI/Al=2.6)
H-USY zeolite (80 mg, Si/Al=6)
[8f]
[8g]
[8h]
–
–
99
[a] Yield of LA. [b] Selectivity of LA. [c] Yield of alkyl lactate. [d] Expressed as the ratio of the amount of transition metal and intake of reactant (mol/mol).
ious metal ions (CoII, NiII, CuII and ZnII) on the conversion of
DHA in subcritical water to LA, and ZnII gave the best results
regarding LA selectivity (86 mol%, conversion data not given;
Table 1, entry 2).[7a] We recently discovered that AlIII salts gave a
high yield of LA (90 mol%, 90 min reaction time) in water
under much milder conditions (1408C) (Table 1, entry 3).[7g]
The use of heterogeneous catalysts in water has also been
explored. Kelly evaluated the application of ion-exchange
resins as catalysts for the conversion of DHA to LA under mild
conditions (1008C).[8b] Among the resins tested, Cr-containing
resins gave the highest LA yield (86 mol%; Table 1, entry 4).
While performing this study, the application of a Sn-beta zeoli-
te[8f] and an ultrastable-Y zeolite (H-USY)[8h] on the conversion
of DHA to LA was reported. High yields of LA (up to 90 mol%)
were achieved at 1258C for both types of zeolites, though
long reaction times up to 24 h were required (Table 1, entries 7
and 8).
alyse the reactions of trioses to LA. However, exploratory
screening studies aimed at the identification of the most
promising metals and a better understanding of the reaction
pathways are lacking. We herein report our studies on the cat-
alytic effect of a wide range of homogeneous catalysts in the
form of metal salts on the conversion of DHA and GLY to LA in
aqueous solutions. On the basis of this study, the most promis-
ing metal salts were identified and the yields were optimised
by variation of process conditions. In addition, a reaction path-
way is postulated and supported by kinetic studies.
Results and Discussion
Conversion of DHA to LA using homogeneous catalysts
A wide range of homogeneous catalysts in the form of metal
salts were tested for the conversion of DHA to LA in water.
The reactions were performed in glass ampoules at a tempera-
ture of 1408C using 0.1m solutions of DHA containing 5 mm
metal salt as the catalyst. All reactions were carried out at least
in duplicate and show good reproducibility. The conversions of
DHA after 90 min reaction time are presented in Figure 1. The
selectivity to LA (SLA) and PRV (SPRV) for all reactions is given in
Figure 2. Herein, the selectivities are defined by Equations (1)
and (2):
Several authors reported the use of alcohols as solvents in
combination with homo- and heterogeneous catalysts for the
synthesis of alkyl lactates from trioses. Hayashi and Sasaki
demonstrated that tin chloride was a highly selective catalyst
for the reaction and methyl lactate was obtained at 89 mol%
yield when reacting DHA (0.625m) in methanol using 10 mol%
SnCl2 (3 h, 908C; Table 1, entry 9).[8e] Taarning and co-workers
reported that heterogeneous Sn-based catalysts (Sn-beta zeo-
lite) were also suitable catalysts for the conversion of hexoses
or trioses to methyl lactate in methanol.[7f,8f] A high yield of
methyl lactate (99 mol%) was obtained when using DHA, al-
though relatively long reaction times of up to 24 h were re-
quired (Table 1, entry 12).[8f] The application of commercial zeo-
lites for the conversion of trioses to alkyl lactate has also been
reported.[8g,h] For instance, Pescarmona et al.[8g] and West
et al.[8h] studied the use of ultrastable Y (USY) zeolites for the
conversion of trioses to alkyl lactates (Table 1, entries 13 and
14). Both studies highlight the importance of Lewis acid sites
on the zeolite to achieve a high yield of either LA or alkyl lac-
tates.
½LAꢀ
ð1Þ
ð2Þ
SLA
¼
ꢂ 100 %
ꢂ 100 %
½DHAꢀ0ꢁ½DHAꢀ
½PRVꢀ
SPRV
¼
½DHAꢀ0ꢁ½DHAꢀ
in which [LA], [PRV] and [DHA] are the concentrations of LA,
PRV and DHA after a certain reaction time, respectively, and
[DHA]0 denotes the initial concentration of DHA.
For comparison, the reactions were initially carried out in
the absence of a catalyst (thermal) and using Brønsted acids,
such as HCl and H2SO4. The conversion of DHA was 55 mol%
for the thermal reaction and >90 mol% when using Brønsted
The metal-catalysed studies discussed above indicate that
metal cations, either in solution or in heterogeneous form, cat-
ChemSusChem 2011, 4, 768 – 777
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
769

H. J. Heeres et al.
Of all 26 metal salts tested, Cr, Al and Sn salts were the most
active and gave quantitative conversion of DHA under the con-
ditions employed (Figure 1). However, remarkable differences
for the three metals were observed regarding chemoselectivity.
When using Al and Cr salts, the selectivity for LA was very high
and between 80–92% (Figure 2). The highest selectivity of LA
was found for AlCl3. The formation of brown insoluble solids
was not observed. On the contrary, the selectivity for Sn salts
was always below 25%, making this a less attractive catalyst.
In this case, considerable amounts of dark brown insoluble
products were observed.
When using metal salts, Brønsted acid catalysis may also
play a role because the pH values of aqueous solutions of the
best catalysts, CrCl3 and AlCl3 salts, were 2.62 and 3.24, respec-
tively, at room temperature. Thus, one could argue that the
good yields of LA for these salts were a result of Brønsted acid
catalysis. However, the activity of Brønsted acids such as HCl
and H2SO4 is much lower than that of the Cr and Al solutions
(Figure 1); a clear indication that metal cations indeed show
catalytic activity. In addition, the chemoselectivity is altered
dramatically when using metal salts (Figure 2), from mainly
PRV for Brønsted acids to solely LA for Al and Cr salts. Thus,
the Al and Cr cations both have a positive effect on activity
and chemoselectivity and play a key role in the reaction path-
ways leading to LA.
Figure 1. Conversion of DHA for various catalysts ([DHA]0 =0.1m;
[salt]=5 mm; T=1408C; t=90 min).
When comparing the chemoselectivity for all metal salts, it is
evident that Al and Cr are special and produce selectively LA.
One possible explanation is that these metal ions catalyse the
conversion of the intermediate PRV to LA very efficiently
(Scheme 1). To test this hypothesis, the performance of a range
of metal salts for the conversion of PRV to LA was evaluated.
The reactions were carried out by using a 0.1m PRV solution
containing 5 mm of catalyst at 1408C for a reaction time of
30 min. The conversion of PRV and the yield of LA are provided
in Figure 3. Cr and Al salts are very active and quantitative con-
version to LA was observed. For all other salts, the conversion
and the yield of LA were much lower (<30 mol%) and brown
insoluble products were formed as well. The results indicate
Figure 2. Selectivity of PRV (SPRV) and LA (SLA) for various catalysts
([DHA]0 =0.1m; [salt]=5 mm; T=1408C; t=90 min).
acids. These results imply that H+ ions have a positive effect
on the conversion of DHA. In the absence of catalysts, PRV is
the sole product, whereas mixtures of PRV and LA were ob-
served for the Brønsted acids (Figure 2). The major product for
Brønsted acids was PRV, and LA was only obtained in low
yields (<4 mol%). In addition, brown soluble and insoluble
products were observed for both the thermal and Brønsted
acid catalysed reactions under these conditions. These results
indicate that dehydration of DHA to PRV is the dominant path-
way for the thermal and Brønsted acid catalysed reaction
(Scheme 1).
Figure 3. Conversion of PRV versus yield of LA for various catalysts
([PRV]0 =0.1m; [salt]=5 mm; T=1408C; t=30 min).
770
www.chemsuschem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2011, 4, 768 – 777

Conversion of Dihydroxyacetone to Lactic Acid
that the Al and Cr salts effectively catalyse the conversion of
the intermediate PRV to LA in aqueous solutions and this is
likely to be the explanation for the special role of Al3+ and Cr3+
for the conversion of trioses to LA.
A series of experiments were carried out in a temperature
window of 120–1808C, initial concentrations of DHA between
0.1–1m, and AlCl3 concentrations ranging from 5 to 10 mm.
The product composition at various reaction times was deter-
mined. The effect of AlCl3 concentration on the conversion of
DHA and yield of LA is presented in Figure 4. Evidently, higher
catalyst concentrations lead to higher rates. At the same DHA
conversion, the yield of LA is slightly higher when using higher
catalyst concentrations.
The main product for the less active metals was a brown
solid. It is likely that the source of these insoluble products is
PRV, which is known to polymerise at elevated temperatures in
aqueous solutions and forms brown insoluble products.[8a] This
was tested separately by performing an experiment with PRV
in the absence of a catalyst (blank in Figure 3). The PRV con-
version was about 52 mol%, whereas brown insoluble prod-
ucts were the sole product, confirming that PRV was a source
for the formation of insoluble, polymeric materials.
The starting triose (DHA) may also be involved in the forma-
tion of brown insoluble products. This is evident when per-
forming reactions of DHA in the absence of catalysts or when
using Brønsted acids; this gives significant amounts of solids.
The possible involvement of DHA on the formation of insolu-
ble materials was supported by kinetic modelling results (see
below).
Anion effects on the conversion of DHA to LA were explored
by performing experiments with Al salts containing different
anions. The experiments were performed by using a 0.1m solu-
tion of DHA containing 5 mm of Al salts, and the conversion of
DHA and yield of LA after 15 min were determined (Table 2).
Table 2. Effect of various anions for Al salts for the reaction of DHA in
aqueous solutions ([DHA]0 =0.1m, [salt]=5 mm, T=1408C, t=15 min).
Entry
Catalyst
DHA conversion
[mol%]
Yield [mol%]
PRV
GLY
LA
1
2
3
4
5
–
AlCl3
Al2(SO4)3
Al(NO3)3
Al(OTf)3
9
44
44
44
43
–
3
3
2
2
8
12
17
12
12
–
28
21
27
26
Relatively short reaction times were applied to avoid quantita-
tive conversion of DHA, allowing a proper comparison of cata-
lytic effects. Under these conditions, both LA and the inter-
mediate PRV were present after the reaction. In addition, glyc-
eraldehyde (GLY) was also present in the reaction mixture. The
latter is a known isomerisation product of DHA (see below). No
major differences were observed for both catalyst activity and
product selectivity, suggesting that anion effects were less im-
portant in the catalytic pathways.
&
*
Figure 4. Effect of AlCl3 concentration (5 ( ) and 10mm ( )) on the conver-
sion of DHA and yield of LA ([DHA]0 =0.1m, T=1408C).
The conversion of DHA is a strong function of reaction tem-
perature (Figure 5). Essential quantitative conversions of DHA
were achieved within 10 min at the highest temperature in the
range (1808C). At lower temperatures, the rate was reduced
considerably and, for instance, a reaction time of 2 h was re-
quired for 90% DHA conversion at 1208C. The yield of LA
shows small, though significant, temperature dependence (Fig-
ure 5b). Highest LA yields (>90 mol%) were observed at
1808C.
The effect of initial DHA concentration on the DHA conver-
sion is presented in Figure 6. The DHA conversion is independ-
ent of the initial concentration of DHA; an indication that reac-
tion is first order in DHA. However, the yield of LA is a function
of the initial concentration of DHA and was significantly higher
when using a low initial concentration of DHA (94 vs.
Effect of process conditions for the Al-catalysed reaction of
DHA to LA
In the screening studies, Al and Cr salts were identified as
most promising for the conversion of DHA to LA in aqueous
solutions. Due to the fact that Al is less expensive and more
non-toxic than Cr, Al salts were selected for further focused
studies on the conversion of DHA to LA.
ChemSusChem 2011, 4, 768 – 777
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
771

H. J. Heeres et al.
~
&
*
Figure 5. Effect of reaction temperature (120 ( ), 140 ( ) and 1808C ( )) on
the conversion of DHA (a) and yield of LA (b) ([DHA]0 =0.5m,
[AlCl3]=10 mm).
~
&
*
Figure 6. Effect of [DHA]0 (0.1 ( ), 0.5 ( ) and 1m ( )) on the conversion of
DHA and the yield of LA ([AlCl3]=10 mm, T=1408C)
Based on these observations, a reaction sequence for the
conversion of trioses (DHA and GLY) to LA is proposed and
given in Scheme 2. In the case of GLY as a starting material,
GLY initially isomerises to form DHA.[8h,9] The subsequent reac-
tion of DHA to LA involves two steps with PRV as the inter-
mediate. The first step in the sequence (DHA to PRV) is likely
to involve a keto–enol tautomerisation followed by dehydra-
tion to the enol form of PRV. Subsequent tautomerisation then
leads to PRV. The conversion of PRV to LA may proceed by a
hydration–rearrangement step. The source of the insoluble by-
products are PRV and DHA (see below), most likely by oligo-
and polymerisation reactions.
Catalytic effects of Al3+ (and Cr3+) cations were observed for
both the overall reaction of DHA to LA (Figure 1) and the con-
version of PRV to LA (Figure 3). For the reaction of DHA to PRV,
it is likely that the cations act as Lewis acid catalysts and accel-
erate the keto–enol tautomerisation and subsequent dehydra-
tion by coordination to the carbonyl and hydroxyl groups. This
mechanism was also proposed for the tin-catalysed conversion
of DHA in alcohols to alkyl lactate.[8e] Besides catalytic effects
of the metal ion, the conversion of DHA to PRV is also cata-
lysed by H+, as shown for the reaction of DHA using mineral
acids (Figure 1).
80 mol%). The formation of brown soluble materials was ob-
served when using high initial concentrations of DHA; an indi-
cation that the side reactions leading to insoluble products
have a higher reaction order than the preferred reactions to
form LA.
Determination of reaction pathways
To gain insights in the reaction pathway from DHA to LA, addi-
tional experiments were performed with AlCl3 as the catalyst
using DHA and GLY as starting materials at an intermediate
temperature (1408C) and variable reaction times. Reactions
with GLY were also performed because it is a known isomerisa-
tion product of DHA and may play a role in the reaction se-
quence.[9] A number of representative concentration profiles
are presented in Figure 7.
When starting with DHA (Figure 7a), two products were ob-
served, PRV and LA. PRV is an intermediate product and shows
a clear optimum, in line with a consecutive reaction pathway.
When starting with GLY (Figure 7b), a rapid drop in the con-
centration of GLY is observed with the concomitant formation
of DHA and PRV. At longer reaction times, DHA and PRV react
further to give LA.
The reaction of the intermediate PRV to LA was shown to be
catalysed by Al and Cr ions only and as such these two metals
772
www.chemsuschem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2011, 4, 768 – 777

Conversion of Dihydroxyacetone to Lactic Acid
unique for the reaction and why other metals tested are far
less active and selective.
Kinetic modelling: Model development
A kinetic model for the reaction of DHA to LA in water was de-
veloped by using AlCl3 as the catalyst. It involves the main re-
actions as well as possible side reactions to insoluble products.
It is based on the following considerations:
The reaction rate equations are quantified by using the
power law approach instead of a pseudo-first-order approach.
The first step in the reaction is the conversion of DHA to
PRV (see Scheme 3). The rate of this reaction is expressed by
Equations (3) and (4), in which Tr is the reference temperature
(1408C):
Scheme 3. Proposed kinetic scheme for the conversion of DHA to LA.
a
RA1 ¼ kR ½DHAꢀ
ð3Þ
ð4Þ
A1
&
^
)
Figure 7. Concentration profiles for the reaction of DHA ( ) and GLY (
ꢀ
ꢁ
ꢂꢃ
~
*
using AlCl3 as the catalyst to form PRV ( ) and LA ( ) ([AlCl3]=5 mm,
ER
R
1
T
1
Tr
A1
kR ¼ kR exp ꢁ
ꢁ
T=1408C)
A1
1
The involvement of DHA in the formation of insoluble by-
products is described by Equations (5) and (6):
are unique. The reaction is likely to involve hydration followed
by a 1,2-hydride shift.[8e] The metal ions act as Lewis acid cata-
lysts by coordination and activation of the carbonyl group.
Without further experimentation (e.g., solution studies on the
coordination chemistry of Al3+ with PRV, X-ray diffraction stud-
ies on isolated compounds and theoretical studies), it is not
possible to answer the question why Al3+ (and Cr3+) are
b
ð5Þ
ð6Þ
RA2 ¼ kR ½DHAꢀ
A2
ꢀ
ꢁ
ꢂꢃ
ERA2
R
1
T
1
Tr
kR ¼ kR exp ꢁ
ꢁ
A2
2
Scheme 2. Possible reaction pathway for the conversion of trioses to LA in aqueous solutions by using AlIII salts as the catalyst.
ChemSusChem 2011, 4, 768 – 777
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
773

H. J. Heeres et al.
The reaction of PRV to LA (Scheme 3) is expressed by Equa-
tions (7) and (8):
g
RB1 ¼ kR ½PRVꢀ
ð7Þ
ð8Þ
B1
ꢀ
ꢁ
ꢂꢃ
ER
R
1
T
1
Tr
B1
kR ¼ kR exp ꢁ
ꢁ
B1
3
The conversion of PRV to insoluble byproducts is quantified
by using Equations (9) and (10):
d
ð9Þ
RB2 ¼ kR ½PRVꢀ
B2
ꢀ
ꢁ
ꢂꢃ
ERB2
R
1
T
1
Tr
kR ¼ kR exp ꢁ
ꢁ
ð10Þ
B2
4
The isomerisation of DHA to GLY was not taken into ac-
count. When using DHA as the starting material, GLY was ob-
served only in very minor amounts in the initial stages of the
reaction, which made it difficult to determine its concentration
accurately.
The experiments were carried out at a fixed AlCl3 concentra-
tion of 10 mm and as such the catalyst concentration was not
taken into account in the modelling study.
To investigate the possible autocatalytic effect of LA, addi-
tional experiments were performed in the presence of LA
(0.2m DHA, 10 mm AlCl3, 0.2m LA). The results were compared
with a series of experiments without LA addition at the start of
the reaction (blank). The reactions were performed at 1408C
and the results are depicted in Figure 8. Significant effects of
the addition of LA on the conversion of DHA and yield of LA
(Figure 8) are absent, even though the acidity of the initial re-
action mixture with 0.2m LA present is higher (pH 2.27) than
the blank solution (pH 3.48). We may thus conclude that LA is
by far less active than Al species and that autocatalytic effects
due to formation of LA are absent under the prevailing condi-
tions.
Figure 8. DHA conversion and LA yield as a function of time for experiments
&
*
in the presence ( ) and absence of LA ( ) at the start of the reaction
([DHA]0 =0.2m, [AlCl3]=10 mm, T=1408C).
The rate expressions [Eqs. (3)–(10)] in combination with the
mass balances [Eqs. (11)–(13)] and temperature profile in the
initial stage of the reaction were applied to model the experi-
mental batch data.
For a batch reactor setup, the concentrations of the individ-
ual species as a function of time using the proposed kinetic
model in Scheme 3 may be represented by the ordinary differ-
ential Equations (11)–(13):
The kinetic modelling activities were performed in two dis-
tinct stages. Initially experiments were carried out and mod-
elled by using PRV as the starting material to gain insights in
the reaction rates of PRV to LA and insoluble products (reac-
tions RB1 and RB2). In the second stage, reactions were carried
out by using DHA as the starting material. The experiments in
the second stage were modelled by using kinetic input from
the first stage to obtain the kinetic data for the reaction of
DHA to PRV (reactions RA1 and RA2).
d½DHAꢀ
ð11Þ
ð12Þ
ð13Þ
¼ ꢁRA1ꢁRA2
dt
d½PRVꢀ
dt
¼ RA1ꢁRB1ꢁRB2
A total of nine batch experiments were performed for stage
one of the kinetic study involving the reaction of PRV to LA
and solid byproducts. Each experiment provided concentration
profiles of reactants and soluble products as a function of
time. A temperature window of 120–1808C, initial PRV concen-
trations between 0.1–0.5m and a fixed AlCl3 intake of 10 mm
were applied. A total of 11 batch experiments were performed
for stage 2, using initial concentrations of DHA between 0.1–
1m and a similar temperature window and catalyst intake as
those used for stage 1.
d½LAꢀ
¼ RB1
dt
The kinetic experiments were carried out in glass ampoules
placed in an oven and rapidly heated up to reaction tempera-
ture. To compensate for non-isothermal behaviour at the start
of the reaction, the temperature profile during heating up was
determined experimentally by performing experiments with
pure water. The resulting equation for the heating up profile
was also included in the kinetic model to compensate for non-
isothermal behaviour at the initial phase of the reaction.[10]
774
www.chemsuschem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2011, 4, 768 – 777

Conversion of Dihydroxyacetone to Lactic Acid
The kinetic parameters were estimated by minimising the
errors between experimental data and the kinetic model. The
best estimates of the kinetic parameters and their standard de-
viation were determined by using a MATLAB optimisation rou-
tine (lsqnonlin), and the results are presented in Table 3.
Table 3. Kinetic parameters for the conversion of DHA and PRV to LA
using 10 mm AlCl3 as the catalysts.
Parameter
Value[a]
RA1 and RA2
a
0.97ꢃ0.04
1.08ꢃ0.18
0.075ꢃ0.006
0.008ꢃ0.002
93ꢃ3
b
kR [m1ꢁa minꢁ1
]
[a]
1
kR [m1ꢁb minꢁ1
]
[a]
A2
ER [kJmolꢁ1
]
]
A1
Figure 10. Parity plot for experimental data and model predictions (stage 1:
ER [kJmolꢁ1
58ꢃ14
A2
~
~
&
*
: conversion of DHA, : yield of PRV, : yield of LA; stage 2: : conversion
RB1 and RB2
*
of PRV, : yield of LA).
g
0.73ꢃ0.02
1.30ꢃ0.18
0.105ꢃ0.005
0.022ꢃ0.005
68ꢃ3
d
kR [m1ꢁg minꢁ1
]
]
[a]
[a]
3
kR [m1ꢁd minꢁ1
4
ER [kJmolꢁ1
]
]
Table 4. Model comparison for the conversion of DHA.
B1
ER [kJmolꢁ1
72ꢃ15
B2
Main reaction (RA1
)
Side reaction (RA2
)
R2
AIC
[a] The values were determined at Tr =1408C.
included
included
excluded
included
0.975
0.982
ꢁ906
ꢁ982
A good fit between experimental data and model was ob-
tained, as shown for a number of representative experiments
in Figure 9. A parity plot in Figure 10 also illustrates the good-
ness of fit between the experimental data and model.
yields to high R2 and a low AIC value. It is evident that model
including RA2 has a higher value of R2 and a lower value of AIC
(Table 4), supporting the involvement of DHA in the formation
of insoluble products.
In the proposed kinetic scheme, DHA reacts to form PRV
(RA1) and oligo-/polymeric products (RA2). The data were also
modelled by using a simplified reaction scheme without the
involvement of RA2, and as such following a reaction scheme
proposed by West et al.[8h] for the conversion of DHA to LA by
using USY zeolites as catalysts. To compare the quality of the
models, both the coefficient of determination (R2) and Akaike’s
information criterion (AIC) were used (Table 4). The best model
Model implications
Experimentally, it was observed that the conversion of DHA
was not a function of the initial concentration of DHA (see
Figure 7 for details), rationalised by assuming that the reaction
was first order in DHA. This was
also confirmed by the kinetic
modelling activities and the
order of the reaction for DHA (a
in Table 3) was found to be close
to one.
The exploratory studies also
revealed that the yield of LA was
a function of the initial DHA con-
centration, with a low initial con-
centration of DHA leading to
higher LA yields (Figure 6b). This
may be rationalised by compar-
ing the order of the undesired
reactions that form insoluble by-
products (RA2 and RB2) with that
of the desired reactions. The
order of the side reactions and,
particularly for RB2, is far higher
~
&
*
Figure 9. Comparison between experimental data ( : DHA, : PRV, : LA) and the model (c).
(1.30) than the order of the
ChemSusChem 2011, 4, 768 – 777
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
775

H. J. Heeres et al.
others (Table 3). This indicates that the side reaction to give in-
soluble products may be suppressed by working at lower con-
centration levels, in agreement with experimental findings.
Figure 11 shows the modelled LA yield as a function of the
initial DHA concentration and temperature. It is evident that a
high yield of LA is attainable at low initial DHA concentrations
and high temperatures, in agreement with the experimental
findings.
Experimental Section
Materials and methods: DHA dimer (97%), GLY (ꢄ98%), PRV
(40 wt%) and standardised 1 molLꢁ1 LA were purchased from
Sigma Aldrich (Schnelldorf, German). The halide and sulfate salts
were purchased from Merck KGAa (Darmstadt, Germany), apart
from AlCl3·6H2O, CrCl3·6H2O and Fe2(SO4)3·xH2O, which were pur-
chased from Sigma Aldrich (Schnelldorf, German), and anhydrous
CrCl2 and LaCl3, which were obtained from Acros Organics (Geel,
Belgium).
The reactions were carried out in glass ampoules with an inside di-
ameter of 3 mm, wall thickness of 1.5 mm and length of 150 mm.
The glass ampoules containing a predetermined amount of salt in
aqueous solution of triose were sealed by using a torch and
placed in a constant oven temperature. Typical experiments were
conducted according to experimental procedures described in a
previous study.[7g]
The amounts of DHA, GLY, PRV and LA were quantified by using a
HPLC instrument with an RI detector and a Biorad Aminex HPX-
87H column. A capillary electrophoresis (CE) system from Agilent
Technologies was used to determine the amount of LA in the
aqueous phases. Detailed descriptions of the HPLC and CE meth-
ods are described elsewhere.[7g]
An Inolab pH 730 instrument equipped with a Sentix 81 probe was
used to measure the pH of the reaction mixtures at room tempera-
ture.
Acknowledgements
Figure 11. Effect of [DHA]0 and temperature on the yield of LA
([AlCl3]=10 mm, 100% conversion of DHA).
C.B.R. thanks the University of Groningen for financial support to
perform his Ph.D. study with a Bernoulli Scholarship.
Conclusions
Keywords: kinetics
sustainable chemistry · trioses
· lactic acid · renewable resources ·
The potential of metal salts as catalysts for the conversion of
the trioses DHA and GLY into LA in aqueous solutions has
been explored. Al3+ and Cr3+ were found to be the most prom-
ising cations and LA yields of up to 93 mol% were obtained. A
reaction pathway was proposed that involved PRV as the inter-
mediate. A kinetic model was developed for the conversion of
DHA to LA by using Al salts. The model implies that high LA
yields were attainable by performing the reaction at low initial
concentrations of DHA at the highest temperature in the
range studied (1808C).
The findings of this study are also relevant for the develop-
ment of improved heterogeneous catalysts for the conversion
of DHA into LA in aqueous solutions. Based on this study, solid
catalysts with Lewis acidic Al3+ or Cr3+ sites may be very prom-
ising. In addition, our study also shows that homogeneous sys-
tems are useful model systems for heterogeneous counterparts
to gain insights into mechanistic pathways.
[1] a) A. J. Ragauskas, C. K. Williams, B. H. Davison, G. Britovsek, J. Cairney,
C. A. Eckert, W. J. Frederick, Jr., J. P. Hallett, D. J. Leak, C. L. Liotta, J. R.
Mielenz, R. Murphy, R. Templer, T. Tschaplinski, Science 2006, 311, 484;
b) A. Corma, S. Iborra, A. Velty, Chem. Rev. 2007, 107, 2411.
[2] a) T. Werpy, G. Petersen, Technical Report No. DOE/GO-102004-1992, Na-
tional Renewable Energy Lab, Golden, CO, 2004; b) J. J. Bozell, G. R. Pe-
tersen, Green Chem. 2010, 12, 539.
[3] a) R. Datta, M. Henry, J. Chem. Technol. Biotechnol. 2006, 81, 1119; b) Y. J.
Wee, J. N. Kim, H. W. Ryu, Food Technol. Biotechnol. 2006, 44, 163; c) M.
Malveda, M. Blagoev, T. Kumamoto, http://www.sriconsulting.com/CEH/
Public/Reports/670.5000/, 20-01-2011, 2011.
[4] J. Byrne, http://www.foodnavigator-usa.com/Financial-Industry/CSM-
buys-up-Cargill-s-shares-in-US-lactic-acid-plant, 15-08-2010, 2010.
[5] a) R. D. Cortright, M. Sanchez-Castillo, J. A. Dumesic, Appl. Catal. B 2002,
39, 353; b) J. C. Serrano-Ruiz, J. A. Dumesic, ChemSusChem 2009, 2, 581.
[6] a) K. L. Wasewar, A. A. Yawalkar, J. A. Moulijn, V. G. Pangarkar, Ind. Eng.
Chem. Res. 2004, 43, 5969; b) K. L. Wasewar, Chem. Biochem. Eng. Q.
2005, 19, 159.
The conversion of trioses to LA may be envisaged as an al-
ternative for the currently applied fermentation route to LA,
providing that the trioses are readily available. Potential sour-
ces for the trioses are cheap hexoses, such as d-glucose,[7a,b,g]
pyrolysis oil fractions which may contain up to 4 wt% of DHA
and PRV[11] or glycerol,[12] which is a main byproduct of biodie-
sel processes.
[7] a) M. Bicker, S. Endres, L. Ott, H. Vogel, J. Mol. Catal. A 2005, 239, 151;
b) X. Yan, F. Jin, K. Tohji, T. Moriya, H. Enomoto, J. Mater. Sci. 2007, 42,
9995; c) L. Kong, G. Li, H. Wang, W. He, F. Ling, J. Chem. Technol. Biotech-
nol. 2008, 83, 383; d) A. Onda, T. Ochi, K. Kajiyoshi, K. Yanagisawa, Appl.
Catal. A 2008, 343, 49; e) A. Onda, T. Ochi, K. Kajiyoshi, K. Yanagisawa,
Catal. Commun. 2008, 9, 1050; f) M. S. Holm, S. Saravanamurugan, E.
Taarning, Science 2010, 328, 602; g) C. B. Rasrendra, I. G. B. N. Makerti-
hartha, S. Adisasmito, H. J. Heeres, Top. Catal. 2010, 53, 1241.
[8] a) M. J. Antal, Jr., W. S. L. Mok, G. N. Richards, Carbohydr. Res. 1990, 199,
111; b) R. L. Kelly, EP541330A2, BP Chemicals, Saltend, Hedon Hull, 1992;
c) E. Bang, J. Eriksen, L. Mønsted, O. Mønsted, Acta Chem. Scand. 1994,
48, 12; d) J. Eriksen, O. Mønsted, L. Mønsted, Trans. Met. Chem. 1998, 23,
776
www.chemsuschem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2011, 4, 768 – 777

Conversion of Dihydroxyacetone to Lactic Acid
783; e) Y. Hayashi, Y. Sasaki, Chem. Commun. 2005, 2716; f) E. Taarning,
S. Saravanamurugan, M. S. Holm, J. M. Xiong, R. M. West, C. H. Christen-
sen, ChemSusChem 2009, 2, 625; g) P. P. Pescarmona, K. P. F. Janssen, C.
Delaet, C. Stroobants, K. Houthoofd, A. Philippaerts, C. D. Jonghe, J. S.
Paul, P. A. Jacobs, B. F. Sels, Green Chem. 2010, 12, 1083; h) R. M. West,
M. S. Holm, S. Saravanamurugan, J. Xiong, Z. Beversdorf, E. Taarning,
C. H. Christensen, J. Catal. 2010, 269, 122.
[11] a) D. Mohan, C. U. Pittman, Jr., P. H. Steele, Energy Fuels 2006, 20, 848;
b) D. K. Shen, S. Gu, Bioresour. Technol. 2009, 100, 6496.
[12] a) K. Nabe, N. Izuo, S. Yamada, I. Chibata, Appl. Environ. Microbiol. 1979,
38, 1056; b) R. Garcia, M. Besson, P. Gallezot, Appl. Catal. A 1995, 127,
165; c) D. Hekmat, R. Bauer, V. Neff, Proc. Biochem. 2007, 42, 71; d) E.
Farnetti, J. Kapar, C. Crotti, Green Chem. 2009, 11, 704.
[9] G. Bonn, M. Rinderer, O. Bobleter, J. Carbohydr. Chem. 1985, 4, 67
[10] B. Girisuta, L. Janssen, H. J. Heeres, Green Chem. 2006, 8, 701.
Received: December 29, 2010
Published online on May 20, 2011
ChemSusChem 2011, 4, 768 – 777
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
777