M.A. Hossain et al.
Applied Catalysis A, General 611 (2021) 117979
BAS (Fig. 4B). These results indicated that BAS was responsible for the
formation of coke.
To determine the stability of lactic acid, we heated lactic acid at 140
◦C for 6 h without catalyst (blank) and with catalysts, Z-500 and Z-900.
Lactic acid conversion was <10 % in all cases (Fig. S6); thus, lactic acid
was thermally stable under this experimental condition. Moreover, these
results suggested that the catalysts did not cause side reactions with
lactic acid.
4. Discussion
Fig. 5. Proposed chemical pathway for the dihydroxyacetone conversion to
lactic acid by thermally treated ZSM-5. LAS = Lewis acid sites, BAS = Brønsted
acid sites, Δ = heat.
We investigated a heat treatment strategy to generate active Lewis
acid sites (LAS) within ZSM-5. These LAS were active sites for dihy-
droxyacetone isomerization in water. The dihydroxyacetone isomeri-
zation reaction is a cascade of (1) dihydroxyacetone dehydration to
pyruvaldehyde, followed by (2) pyruvaldehyde rehydration to lactic
acid. Specifically, our results demonstrated that the dihydroxyacetone
dehydration step was driven by reaction temperature, and the LAS were
needed for pyruvaldehyde rehydration to lactic acid. Previously, little
was known about the activities of the different acid sites in the reaction.
One of our most significant findings was that the high temperature
treatment drove the modified ZSM-5 toward a high lactic acid selec-
tivity. The heat treated ZSM-5 had a high LAS density. The dihydroxy-
acetone dehydration to pyruvaldehyde was a thermal reaction and
pyruvaldehyde was formed readily at 140 ◦C, whereas LAS were needed
to convert the resulting pyruvaldehyde to lactic acid. Moreover, the
amount of LAS was consistent with the order of the catalytic perfor-
mance, which suggested that the LAS were the active sites for pyr-
uvaldehyde rehydration. Dapsens et al. [35] showed that an increase in
LAS density by desilication of MFI zeolites enhanced the lactic acid
selectivity of dihydroxyacetone conversion at 140 ◦C for 6 h. After acid
washing, the LAS density of the desilicated MFI decreased, resulting in a
decrease in the lactic acid. Takagaki et al. [54] used supported chro-
mium and titanium oxide catalysts with varying compositions to
generate catalysts with various LAS/BAS ratios for dihydroxyacetone
conversion. They found that catalysts with a high LAS were more se-
lective toward LA than those with low LAS. Our results agree with the
findings of Dapsens et al. [35] and Takagaki et al. [54]
selectivity was in the following order: H-USY (Si/Al = 6) > H-β (Si/Al =
12.5) > H-MOR (Si/Al = 10) > H-ZSM-5 (Si/Al = 11.5). Moreover, the
zeolite with a low Si/Al ratio was more selective to lactic acid, and H-
USY (Si/Al = 6) was the most selective to lactic acid (71 % lactic acid
selectivity) (Table S2). Although these zeolites were active for dihy-
droxyacetone isomerization, their catalytic performance was still infe-
rior to the Sn-based catalysts (Sn-containing β-zeolites [61,62] and
Sn-containing silica [63]) with a high selectivity (>90 %) to lactic
acid at a full conversion (Table S3). Two major limitations of using
Sn-based catalysts are (1) a long and complicated synthesis [35], and (2)
a scarcity of tin [64].
Clear advantages of this heat treatment approach are (1) the appli-
cability to commercially available ZSM-5, and (2) the ability to control
the LAS and BAS densities of ZSM-5 is a superior property compared
with the active Sn-containing catalysts for dihydroxyacetone isomeri-
zation. Moreover, this strategy can be used for other acid-catalyzed re-
actions, such as dehydration [65], esterification [66], isomerization
[67], etherification [68], and cascade reactions in which both LAS and
BAS are needed, such as hydroxymethylfurfural production from cellu-
lose [69,70]. The proximity of EFAL and Brønsted acid sites can lead to
enhance catalytic activity of alkane cracking [71,72]. The enhancement
of catalytic activity depends on the EFAL properties, such as proximity of
EFAL concentration, speciation, location in the framework, distribution,
and proximity of Bronsted acid sites [73]. Our work could be extended
by identifying the EFAL features using 29Si and 27Al magic angle spin-
ning nuclear magnetic resonance (MAS-NMR) spectroscopy in combi-
nation with density functional theory calculations [74,75] and
correlating the results of MAS-NMR and density functional theory with
catalytic activity. This information will be important for the develop-
ment of a cost-effective and sustainable catalytic process for lactic acid
production from biomass. In addition, the recyclability and change in
mechanical property after catalyst recycling should be assessed to
ensure long catalyst lifetime.
Another significant finding was that BAS did not have any activity in
dihydroxyacetone isomerization. Moreover, the presence of BAS caused
unwanted coke formation from pyruvaldehyde decomposition under our
experimental condition (140 ◦C). The spent catalyst with high BAS (Z-
◦
500) had more coke compared to catalysts treated at 900 C (Z-900).
Takagaki et al. [54] showed that the presence of BAS in catalysts low-
ered the LA selectivity, results that corroborate our findings. Similarly,
Nakajima et al. [58] used a Brønsted acid catalyst, H2SO4, for dihy-
droxyacetone isomerization and pyruvaldehyde rehydration. They did
not observe any lactic acid yield at 100 ◦C, a further confirmation of our
findings.
5. Conclusion
In comparing blank controls of dihydroxyacetone isomerization and
pyruvaldehyde rehydration, we found that the blank control of dihy-
droxyacetone isomerization produced pyruvaldehyde as the only prod-
uct, whereas the blank control of pyruvaldehyde rehydration did not
produce any observable products. These results suggested that dihy-
droxyacetone dehydration to pyruvaldehyde was a thermal conversion
and LAS was needed to convert pyruvaldehyde to lactic acid. Consid-
ering together the effects of reaction temperature, LAS, and BAS, we
propose the chemical pathway of the modified ZSM-5 for dihydroxyac-
etone isomerization shown in Fig. 5. Dihydroxyacetone dehydration to
pyruvaldehyde proceeded with reaction temperature (140 ◦C). The LAS
generated within modified ZSM-5 was responsible for the selective
dihydroxyacetone isomerization to lactic acid.
We investigated dihydroxyacetone isomerization in water using heat
treated ZSM-5. The treatment at elevated temperature increased the
Lewis acid sites density and decreased Brønsted acid site density of
modified ZSM-5, which promoted high lactic acid selectivity in water.
Dihydroxyacetone isomerization to lactic acid is a cascade of dehydra-
tion to pyruvaldehyde, followed by pyruvaldehyde rehydration to lactic
acid. We demonstrated that dihydroxyacetone dehydration to pyr-
uvaldehyde readily occurred at 140 ◦C and reached 50 % lactic acid
yield after 6 h using heat treated ZSM-5 at 900 ◦C. The high LAS density
of heat treated ZSM-5 was responsible for the pyruvaldehyde rehydra-
tion. This heat treatment strategy offers a new basis to tune LAS density
for biomass processing reactions, including isomerization, dehydration,
and esterification.
We summarized the performance of selected catalysts for dihy-
droxyacetone isomerization. West et al. compared the catalytic perfor-
mance of different zeolites with different Si/Al ratios. They found that
all zeolites were active in dihydroxyacetone isomerization, and the LA
5