Production of methyl levulinate from cellulose: Selectivity and mechanism study

Article abstract of DOI:10.1039/c5gc00440c

The alcoholysis of cellulose into methyl levulinate (ML) in methanol media was investigated in the presence of several kinds of acid catalyst. One of the synthesized solid niobium-based phosphate catalysts was found to be highly efficient for the generation of ML, reaching an ML yield as high as 56%, higher than the LA yield (52%) in aqueous solution with the same reaction conditions as those used in our previous study (Green Chem., 2014, 16, 3846-3853). More interestingly, in water, very strong Lewis acid promoted the formation of LA; but in methanol, Br?nsted acid enhanced the formation of ML. In-depth investigation showed that the mechanism and type of intermediates of cellulose alcoholysis in methanol were different from those in water and a high Br?nsted/Lewis acid ratio (known as B/L acid ratio) of solid catalysts is needed to prevent the generation of by-products, namely, methyl lactate and 1,1,2-trimethoxyethane. This new-proposed reaction mechanism affected by the B/L acid ratio was very helpful for the design of efficient catalysts.

Full text of DOI:10.1039/c5gc00440c

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Green Chemistry
View Article Online
View Journal
PAPER
Production of methyl levulinate from cellulose:
selectivity and mechanism study†
Cite this: DOI: 10.1039/c5gc00440c
Daqian Ding, Jinxu Xi, Jianjian Wang, Xiaohui Liu,* Guanzhong Lu and Yanqin Wang*
The alcoholysis of cellulose into methyl levulinate (ML) in methanol media was investigated in the pres-
ence of several kinds of acid catalyst. One of the synthesized solid niobium-based phosphate catalysts
was found to be highly efficient for the generation of ML, reaching an ML yield as high as 56%, higher
than the LA yield (52%) in aqueous solution with the same reaction conditions as those used in our pre-
vious study (Green Chem., 2014, 16, 3846–3853). More interestingly, in water, very strong Lewis acid pro-
moted the formation of LA; but in methanol, Brönsted acid enhanced the formation of ML. In-depth
investigation showed that the mechanism and type of intermediates of cellulose alcoholysis in methanol
were different from those in water and a high Brönsted/Lewis acid ratio (known as B/L acid ratio) of solid
catalysts is needed to prevent the generation of by-products, namely, methyl lactate and 1,1,2-trimethoxy-
ethane. This new-proposed reaction mechanism affected by the B/L acid ratio was very helpful for the
design of efficient catalysts.
Received 26th February 2015,
Accepted 1st June 2015
DOI: 10.1039/c5gc00440c
www.rsc.org/greenchem
such as high lubricity, flashpoint stability, non-toxicity and
better flow properties under cold conditions.5
Introduction
Recently, abundant and renewable biomass has been regarded
Technically, there are three ways to achieve high selectivity
as a promising alternative to non-renewable resources for sus- for alkyl levulinate,6,7 namely the esterification of LA,8,9 the
tainable biofuels and biochemicals due to the deterioration of alcoholysis of 5-(chloromethyl)furfural10 that was synthesized
the environment and inevitable depletion of fossil resources. from biomass in high concentration hydrochloric acid, and
More and more research is focused on the production of the direct conversion of biomass into alcohol. In industry, the
liquid fuels and high-quality chemicals from the conversion of production of levulinate esters has been mainly achieved
biomass.1,2 In all these explorations, levulinic acid (LA) is one directly from the esterification of LA with alcohols,8,9 catalyzed
of the most popular derivatives3 converted from carbohydrates, by liquid acid. The alcoholysis of 5-(chloromethyl)furfural10
which could be used widely in agriculture, foods, medicines, can also give high selectivity for alkyl levulinate (ca. 90%).
cosmetics, spice industries and biofuels in the future. However, both esterification of LA with alcohols and alcoholy-
However, due to its high acidity, high viscosity and high sis of 5-(chloromethyl)furfural need a high concentration of
boiling point (around 246 °C), the production of LA is limited liquid acid, which is not environmentally friendly. Therefore,
by the cost of the reactor and the separation equipment. As the direct production of levulinate esters from biomass or
one of the alternatives to levulinic acid, methyl levulinate (ML) biomass-based platforms through alcoholysis is a viable
gives better performance in production and separation option worth studying.
because it is almost non-corrosive to the reactor and has a
The generally accepted pathways for the direct conversion
lower boiling point (around 190 °C), indicating lower energy of cellulose to LA in water and levulinate esters in alcohol are
consumption for its separation. ML is as useful as LA in many drawn in Scheme 1. The structures of intermediates in alcohol
fields, such as medicines, solvents, organic chemistry,4 and are different from those generated in water. For example, in
fragrance, and furthermore it can be directly used as an addi- water it follows a cellulose–glucose–HMF–LA pathway, while in
tive for gasoline and diesel due to its excellent performance, methanol, methyl glucoside and 5-methoxymethyfurfural (5-
MMF) are formed in sequence, then converted into methyl
levulinate.10,11
In previous reports, sulfuric acid has been used as a very
Key Lab for Advanced Materials, Research Institute of Industrial Catalysis East China
efficient catalyst to achieve high selectivity of methyl levulinate
University of Science and Technology # 130, Meilong Road, Shanghai, 200237,
directly from cellulose alcoholysis. For example, Wu et al.14
P. R. China. E-mail: xhliu@ecust.edu.cn, wangyanqin@ecust.edu.cn
†Electronic supplementary information (ESI) available. See DOI: 10.1039/
achieved a 55% yield of methyl levulinate from cellulose, cata-
lyzed by H2SO4. Other liquid acids, including p-toluenesulfonic
c5gc00440c
This journal is © The Royal Society of Chemistry 2015
Green Chem.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Article Online
Paper
Green Chemistry
both achieved about 20% yields of methyl levulinate, with
large amounts of methyl glucoside as the main product. The
large amount of methyl glucoside remaining in solution
means that these catalysts were not active enough to catalyze
the further hydration or transformation.
There are some other interesting works that have investi-
gated the selectivity for methyl levulinate in cellulose conver-
sion. Tominaga et al.18 investigated the influence of the
Brönsted/Lewis acid ratio on the alcoholysis of cellulose by
using a series of mixed Brönsted/Lewis liquid acids. They
found that a mixed Brönsted/Lewis acid system with a B/L
ratio of 5/1 had the highest selectivity for methyl levulinate
(70%), compared to pure Brönsted (20%) or pure Lewis acid
(52%), which was never reported before. As far as we know,
many kinds of solid acid have both Brönsted and Lewis acid
sites, such as SO42−/ZrO2, metal phosphate and zeolites,
except for sulfonic acid resin, so this result might be useful for
the design of solid acids for biomass conversion.
Although Tominaga et al.18 realized that the B/L acid ratio
would affect the methyl levulinate selectivity, a reasonable
mechanism was not given to explain the phenomenon,
especially how Brönsted and Lewis acids work on the trans-
formation of the intermediate, such as methyl glucoside and
5-methoxymethyfurfural (5-MMF), to methyl levulinate. Learn-
ing the details of the transformation mechanism on Brönsted
and Lewis acid sites will be helpful in enhancing the pro-
duction yield of methyl levulinate.
On the other hand, lactic acid,19 methyl lactate20 and 1,1,2-tri-
methoxyethane21 could also be generated from the breaking of a
C–C bond, catalyzed by the Lewis acid sites on solid catalysts. In
the production of methyl levulinate, the generation of by-pro-
ducts should always be considered, for it will consume carbon
atoms and compete with the selectivity for methyl levulinate.
In summary, there are several problems to be solved in the
conversion of cellulose into methyl levulinate. Firstly, the pre-
vention of methanol etherification; secondly, improving the
cellulose conversion rate and the methyl levulinate selectivity;
thirdly, the influence of Brönsted and Lewis acid sites on the
intermediate and product selectivity through analysing the
transformation pathway and mechanism. To solve these pro-
blems, a series of solid acid catalysts, including mesoporous
niobium phosphate catalysts,22,23 were used in these studies,
and the highest yield of 56% methyl levulinate was achieved
over mesoporous niobium phosphate prepared at pH 1. More-
over, this catalyst has excellent reusability and can be used 5
times with only a slight loss of activity and with little etherifica-
tion of methanol. The detailed reaction pathway, the product
distributions (including intermediates) and the role of the
Lewis acid in alcoholysis were also investigated in the main text.
Scheme 1 Reaction pathway for the acid-catalyzed conversion of cel-
lulose to levulinate esters in alcohol,11,12 compared with that to levulinic
acid in water.13
acid, H3PO4, HCOOH and CH3COOH, were also tested in the
reaction. Unfortunately, H3PO4, HCOOH and CH3COOH gave
out very low yields of methyl levulinate. p-Toluenesulfonic acid
and H2SO4 had another serious problem because they were
excellent catalysts for methanol etherification, and sharply
decreased the amount of solvent. For example, Peng et al.11,15
reported that the –SO3H containing solid acid was a very good
catalyst for the conversion of methanol to dimethyl ether,
which would consume solvent very fast, so they used a series
of solid catalysts including SO42−/TiO2, SO42−/ZrO2, Zr3(PO4)4
and Si–Al zeolites to generate methyl levulinate from biomass,
and found that SO42−/TiO2 was a potential catalyst.11 When
using glucose as a substrate, a 35% yield of methyl levulinate
was achieved over SO42−/TiO2. However, when cellulose was
used as the substrate, the yield of methyl levulinate was only
2−
about 10%, which means that the performance of the SO4
/
TiO2 catalyst was poor for the production of methyl levulinate
from cellulose. A large amount of methyl glucoside and a
small amount of 5-methoxymethyfurfural (5-MMF) were still
present in the solution. Some similar situations were found
when using heteropolyacids and their salts as catalysts. Rata-
boul16 and Deng17 used Cs2.5H0.5PW12O40 and H4SiW20O40,
respectively, for the alcoholysis of cellulose in methanol, and
Experimental section
Chemicals
All chemicals used were purchased from Sinopharm Chemical
Reagent Co. Ltd, except the cellulose, which was bought from
Green Chem.
This journal is © The Royal Society of Chemistry 2015

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Article Online
Green Chemistry
Paper
Fluka Analytical Co. Ltd. All bought chemicals were of analyti- 0.5 g cellulose, 0.5 g water and 9.5 g methanol was loaded into
cal grade and used without further purification.
a Teflon-lined stainless steel autoclave and placed into a temp-
erature-controlled electric heating mantle with a thermocouple
probe detector and magnetic stirring. N2 was used as a protec-
Catalysts preparation
Mesoporous niobium-based phosphate NbP–pH2 was syn- tive gas and kept at a pressure of about 0.8 MPa. The zero time
thesized according to our previous reports.13,22 Typically, was taken when the Teflon-lined stainless steel autoclave was
0.01 mol of diammonium hydrogen phosphate was dissolved placed into the heating mantle under magnetic stirring, which
in 20 mL water and adjusted to pH 2 using phosphoric acid, had already been heated to the set temperature. After reaction,
then 20 mL of 0.5 M niobium tartrate (pH = 2) was added to the solid catalyst was collected by centrifugation and washed
the above solution with stirring. The mixed solution was with water and ethanol for several times, then dried at 100 °C
slowly dropped into a previously prepared aqueous solution of for 24 hours for the next run.
cetyltrimethyl ammonium bromide (CTAB), which contained
Analysis of the reaction mixture was carried out by HPLC
1.0 g of CTAB and 13 ml of water, and the final pH was around (Agilent 1200 Series) and GC7890-MSD. The former was
2. A large amount of niobium–phosphorus precipitation equipped with an Aminex HPX-87H(Bio-Rad) ion-exclusion
appeared immediately from the solution and the mixture was column, eluting with an aqueous solution of sulfuric acid as
stirred for an additional 60 min at 35 °C. Then it was trans- the mobile phase. The products and intermediates were ana-
ferred into a Teflon-lined autoclave and aged at 160 °C for lyzed with a refractive index detector (Agilent G1362A) in the
24 h. The deposit was filtered after cooling down, washed with HPLC to complete the quantitative analysis. All the intermedi-
distilled water and dried at 100 °C overnight. Finally, the cata- ates were also separated by GC7890 using high purity helium
lyst was obtained by calcination at 500 °C for 5 h in air to as the carrier gas, and the molecular structures were analyzed
remove organics.
qualitatively by the MSD connected with the GC.
The NbP–pH1 was synthesized using identical steps to
Cellulose remaining in the solid mixture was analyzed by
those for catalyst NbP–pH2, expect the solution of diammo- weighing, for it could be dissolved quickly and completely in
nium hydrogen phosphate was adjusted to pH 1 by using poly- hot sulfuric acid solution. Meanwhile, the catalysts and
phosphoric acid. The synthesis of aluminum–modified porous humins were 100% insoluble, so after the reaction, the solid
niobium phosphate was similar to the above procedure except mixture was dried and weighed for the first round, then
for the addition of a calculated amount of aluminium precur- soaked into 0.5 mol L−1 sulfuric acid solution and heated to
sor, which was prepared by dissolving aluminum hydroxide in 160 °C for 6 hours, to make sure the cellulose was completely
10 mL of 1.5 M oxalic acid solution and then added into the dissolved. The rest of the solid mixture was washed, dried and
mixed solution with stirring.
weighed for a second round. By comparing the data from the
two weighing rounds, the cellulose conversion could be calcu-
lated with high precision. The humins dissolved in methanol
Characterizations
The pyridine adsorption infrared (IR) spectra were recorded on solution were also weighed and quantitated25,26 after methanol
a Nicolet NEXUS 670 FT-IR spectrometer, with 32 scans at an and small molecules was removed at 100 °C for 12 hours.
effective resolution of 4 cm−1. Around 50 mg of the catalyst
was pressed into a self-supporting disk and placed in an IR
cell attached to a closed glass-circulation system. The catalyst
disk was dehydrated by heating at 400 °C under vacuum in
order to remove physisorbed mixtures. The IR spectrum back-
ground was recorded at room temperature when the cell had
Results and discussion
Performance and product distribution of various catalysts in
the alcoholysis of cellulose
cooled down. Pyridine vapour was then introduced into the Normally, in the alcoholysis of cellulose to levulinate, a
cell at room temperature until equilibrium was reached. Sub- Brönsted acid was used. Here, mesoporous niobium phos-
sequent evacuations were performed at 100 °C for 60 min fol- phate synthesized at pH 1 was used in the alcoholysis of cellu-
lowed by spectral acquisitions at room temperature using the lose, along with other catalysts used in the conversion of
background recorded before. The acid sites amounts of cata- cellulose to LA in water.13 Table 1 summarizes their physical
lysts were calculated from pyridine adsorption integral curve and acidic properties, and the N2 adsorption–desorption iso-
recorded.
therm and Py-FTIR spectrum are provided in the ESI.† For
The BET properties of porous catalysts were calculated from comparison, hydrochloric acid, trifluoromethanesulfonic acid
nitrogen adsorption–desorption isotherms measured at −196 °C (a strong Brönsted acid), H-beta (Si/Al = 50) and (TfO)3Yb
on a NOVA 4200e analyser (Quantachrome Co. Ltd). All samples (Lewis acid) were also used by controlling the total acid
were outgassed at 180 °C for 12 h under vacuum to remove amount of 0.0004 mol. The yields of methyl levulinate as well
moisture and volatile impurities before the measurements.
as other products are presented in Fig. 1. Except for (TfO)3Yb,
a Lewis acid, the cellulose conversion over all other catalysts
was higher than 95%, even though the product distribution
Catalytic reactions
A batch reactor was used in the conversion of cellulose to levu- was different. In these catalysts, the performance of NbP–pH1
linate ester. Typically, a mixture of 0.0004 mol acid catalyst, prepared with the precursor mixture of pH1 was much better
This journal is © The Royal Society of Chemistry 2015
Green Chem.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Article Online
Paper
Green Chemistry
1 nm), which will lead to mass transfer limitations and coke
formation, in accordance with the results reported before.24
The interesting thing was that Al–NbP–pH2, an excellent cata-
lyst in the production of LA from cellulose in water,22 was very
poor for the formation of methyl levulinate. The yield of
methyl levulinate is only 25%, and more carbon was in the by-
products, such as 1,1,2-trimethoxythane and 5-methoxymethy-
furfural (5-MMF), indicating the different pathways in water
and alcohol.
Table 1 Physical and acidic properties (Py-FTIR at 373 K) of various
solid acids
SBET
/
Pore
Brönsted
Lewis acid/ B/L acid
Catalyst
m2 g size/nm acid/μmol g−1 μmol g−1
ratio
H-beta
501.1 0.50
152.0 3.9
233.0 3.8
50.0 9.6
1125.0
850.8
641.5
492.6
645.6
359.2
345.1
469.3
1.74/1
2.15/1
1.86/1
1.05/1
NbP–pH1
NbP–pH2
Al–NbP–pH2
Element
Nb/at%
P/at%
Al/at%
O/at%
During the analysis of GC-MS and HPLC data, we found
that except for the intermediates methyl glucoside and
5-methoxymethylfurfural, drawn in Scheme 1 and shown in
Fig. 1a, there were other two by-products from the C–C bond
break of methyl glucoside, known as methyl lactate and 1,1,2-
trimethoxyethane, shown in Fig. 1b over various solid acid
catalysts. These catalysts all have large amounts of Lewis acid
sites. As we know that methyl glucoside was always the inter-
mediate during the alcoholysis of cellulose,21 it can be trans-
formed through different ways: (i) direct dehydration to 5-
MMF, following by hydration to methyl levulinate; (ii) C–C
cleavage through retro-aldol condensation catalyzed by Lewis
acid directly; or (iii) through methyl fructoside. The possible
pathways over Lewis acid are shown in Scheme 2. In the pro-
duction of lactic acid and alkyl lactate from biomass,19,20 the
breakage of the C–C bond will always occur when enough
Lewis acid exists in solution, thus making the levulinate
species and lactate species co-exist here. It was found from
Fig. 1 and Table 1 that the catalyst containing higher Lewis/
Brönsted acid ratios would give higher yields of methyl lactate/
1,1,2-trimethoxyethane and lower methyl levulinate yields. To
further understand the generation of these small molecules
and the influence of Lewis acidic sites, a deeper investigation
was done.
NbP–pH1
NbP–pH2
Al–NbP–pH2
15.1
20.9
17.4
20.7
21.0
18.6
—
—
2.49
64.2
58.1
61.5
Generation of other small molecules and their relationships
with the acid sites of catalysts
As methyl glucoside was always the intermediate during the
alcoholysis of cellulose, and then converted into different
species over different kinds of catalysts, so the product distri-
bution during methyl glucoside conversion in methanol over
different acid catalysts was assessed. Among the three species
Fig. 1 (a) Catalytic performance of various catalysts in the conversion
of cellulose; (b) the detailed distribution of other small molecules, green
part in (a). Reaction conditions: 0.5 g of cellulose, 0.0004 mol of acid
catalysts, 10 g of 95% methanol, 180 °C, 24 h.
than any other solid acids reported here and before,11,14–17
and almost as excellent as soluble acid catalyst (hydrochloric
acid), giving a 56% yield of methyl levulinate. This provides
the possibility of producing methyl levulinate from cellulose
over a solid acid catalyst, which can prevent the neutralization
of the liquid acid. The yield of methyl levulinate (14.7%) over
H-beta was also reasonable for its small pore size (less than
Scheme 2 Three kinds of reaction pathways of methyl glucoside in
methanol.
Green Chem.
This journal is © The Royal Society of Chemistry 2015