Bioresource Technology 251 (2018) 143–150

Contents lists available at ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Improved levulinic acid production from agri-residue biomass in biphasic T

solvent system through synergistic catalytic effect of acid and products
Sandeep Kumara,b, Vivek Ahluwaliaa, Pranati Kundua,b, Rajender S. Sangwana, Sushil K. Kansalb,
⁎
Troy M. Rungec, Sasikumar Elumalaia,
a
Chemical Engineering Division, Center of Innovative and Applied Bioprocessing (CIAB), Sector 81 (Knowledge City), Mohali, Punjab 140306, India
b
Dr. S. S. Bhatnagar University Institute of Chemical Engineering & Technology, Panjab University, Chandigarh 160014, India
c
Wisconsin Energy Institute (WEI), University of Wisconsin-Madison, 1552 University Ave, Madison, WI 53726, USA

G RA P H I C A L AB S T R A C T

A R T I C L E I N F O A B S T R A C T

Keywords: In this study, levulinic acid (LA) was produced from rice straw biomass in co-solvent biphasic reactor system
Rice straw consisting of hydrochloric acid and dichloromethane organic solvent. The modified protocol achieved a 15% wt
Levulinic acid LA yield through the synergistic effect of acid and acidic products (auto-catalysis) and the designed system
Co-solvent system allowed facile recovery of LA to the organic phase. Further purification of the resulting extractant was achieved
Activated charcoal
through traditional column chromatography, which yielded a high purity LA product while recovering ∼85%
Furfural
wt. Upon charcoal treatment of the resultant fraction generated an industrial grade target molecule of ∼99%
purity with ∼95% wt recovery. The system allows the solvent to be easily recovered, in excess of 90%, which
was shown to be able to be recycled up to 5 runs without significant loss of final product concentrations. Overall,
this system points to a method to significantly reduce manufacturing cost during large-scale LA preparation.

1. Introduction perform many catalytic transformations towards synthesis of in-

dustrially important green solvents (γ-valerolactone, GVL) and fuel
Of the platform molecules identified by US DOE, levulinic acid (LA) chemicals (butene, 5-nonanone, 2-methyl-tetrahydrofuran (MTHF))
or 4-oxopentanoic is of considerable economic interest due to its bi- (Caes et al., 2015; Pileidis and Titirici, 2016). This five-carbon molecule
functional nature, i.e., keto- and carboxyl-groups, which allow it to is commonly produced from glucose in the presence of acid catalyst.

⁎
Corresponding author.
E-mail address: sasikumar@ciab.res.in (S. Elumalai).

https://doi.org/10.1016/j.biortech.2017.12.033
Received 24 October 2017; Received in revised form 11 December 2017; Accepted 12 December 2017
Available online 13 December 2017
0960-8524/ © 2017 Elsevier Ltd. All rights reserved.
S. Kumar et al.
After glucose isomerization, the process first delivers 5-hydro-
xymethylfurfural (HMF) as principal intermediate product via dehy-
dration reaction (−3H2O) and further, it is rehydrated (+2H2O) to
produce LA and formic acid (FA) at equal proportions along with H2O
(Chen et al., 2017; Weiqi and Shubin, 2017). An undesirable insoluble
by-product, humins, are also formed during the sequential transfor-
mation reactions, reducing the final LA yield to 2/3 or lower than the
theoretical yield. Therefore, the extent of the degradation reactions
generally determines the final LA concentration under the modest re-
action conditions (Rackemann and Doherty, 2011). To date, several
studies have demonstrated LA production through both homogeneous
and heterogeneous catalyzed reactions using nearly all possible forms of
carbohydrate substrates (e.g., lignocellulose, cellulose, cellobiose, glu-
cose, fructose, inulin, cassava etc.)(Elumalai et al., 2016b; Girisuta
et al., 2013; Mukherjee et al., 2015; Runge and Zhang, 2012; Zhao
et al., 2017). The most widely investigated method is homogeneous
reaction using inorganic mineral acids, mainly H2SO4, HCl, H3PO4
owing to its high efficiency and easy operation. Despite its highly cor-
rosive nature, HCl exhibits promise due to its high dissociation constant
(Shen and Wyman, 2012; Victor et al., 2014). However, separation of
final product compounds from aqueous acidic solutions is considered
one of the main challenges during LA preparations. Due to the presence
of a carboxyl group, LA has the strongest interaction with water
forming hydrogen bonds with H2O molecules. Also, LA dissociated ions
can be solvated by H2O (Szabolcs et al., 2013). Furthermore, in typical
reaction, many of these acidic reagents lead to the production of un-
wanted by-products, which negatively impact LA yield. In order to
avoid these limitations, later studies have introduced co-solvent reactor
setup consisting of either biphasic or monophasic system. The goal of
this setup is improving overall conversion efficiency through con-
tinuous extraction of furans, preferentially HMF to the organic phase,
thereby reducing the residence time of HMF in aqueous medium. This
approach helps protect HMF from further degradation to humic species,
and allows easy product separation (Alonso et al., 2013).
In this regard, an ideal solvent is one which is suited for substrate
conversion, increases the product yield and selectivity, and poses no, or
minimal, adverse ecological impacts. Various studies have employed a
range of polar protic and aprotic solvents with different polarities, in-
cluding methyl isobutyl ketone (MIBK), MTHF, GVL, dichloromethane
(DCM), 2-butanol, tetrahydrofuran (THF), dimethylsulfoxide, 2-sec-
butylphenol etc. forming monophasic or biphasic systems in combina-
tion with mineral acid reagents (aqueous), and showed furans extrac-
tion to be dependent on the polarity of the solvents (Byun and Han,
2016; Chheda et al., 2007; Elumalai et al., 2016a; Gürbüz et al., 2012;
Morone et al., 2015). THF (polarity index, P = 4.2) exhibited excellent
performance during carbohydrates conversion to HMF (partition coef-
ficient, RHMF = 3.1), ascribed to its greater affinity for HMF due to the
similarity in chemical structure (Cai et al., 2013; Yang et al., 2012).
Several studies have also demonstrated the use of DCM and proposed
that it could be a suitable extraction agent for HMF due to its attractive
affinity (Chheda et al., 2007), despite having the comparatively low P
value of 3.1. Therefore, the polarity is not the only factor affecting
furans distribution to the organic phase. In fact, distribution of HMF to
the organic phase is relatively poor due to its high polarity and there-
fore, it tends to be retained in the aqueous phase (Gallo et al., 2013).
Later studies have attempted to improve the HMF distribution through
the addition of salts, commonly NaCl, KCl, and Na2SO4, into the co-
solvent systems but the movement of salts into the extractant medium
could not be easily controlled (Mukherjee et al., 2015). In theory, these
approaches could potentially prevent further rehydration of 5-HMF to
LA due to the low availability of H2O molecules. Due to these con-
straints, only a few studies have been focused on LA synthesis in co-
solvent systems using carbohydrate sources. For instance, Dumesic and
co-workers (Alonso et al., 2013; Wettstein et al., 2012) have extensively
investigated the deconstruction of both cellulose and lignocellulose in
monophasic or biphasic systems consisting of GVL, as an extraction
144
Bioresource Technology 251 (2018) 143–150
solvent, and mineral acids, particularly H2SO4 and HCl (with the ad-
dition of salts). The main interest of these studies is the production of
GVL through continuous extraction of LA into organic medium. Inter-
estingly, maximum LA conversion (between 18 and 66% depending on
the acid and substrate concentrations) and recovery to the organic
medium were achieved under modest reaction conditions, where
complete transformation to GVL took place. The lack of residual con-
centration of HMF in GVL medium might be attributed to its instability
in GVL and therefore, achieving high LA conversion in the aqueous
phase is mainly due to its complete water miscibility property.
The present study employed a biphasic system consisting of dilute
HCl (aqueous) and DCM (organic extraction solvent) for the decon-
struction of secondary agriculture residue (rice straw) to LA with the
objective of establishing a cost-effective processing method by
achieving maximum product recovery to the organic phase. Rice straw
is considered one of the abundantly available biomass around the world
(∼900 million tons produced per year globally), is relatively in-
expensive and is currently underutilized due to its recalcitrant nature,
low proteinaceous matter and high silica content (Sarkar et al., 2012).
This approach would enable facile product separations, and solvent
recovery for reuse due to its highly volatile and low boiling properties
(boiling point 39.6 °C).
2. Experimental methods
2.1. Chemicals
All reagents and chemicals (including cellulose, glucose, di-
chloromethane (DCM), methanol, levulinic acid (LA), formic acid (FA),
5-hydroxymethylfurfural (HMF), furfural (FF), activated carbon, and
carbohydrate standards) were purchased from Sigma Aldrich India as
analytical grade and used as received without any modifications.
2.2. Feedstock preparation
Native rice straw sample was collected from the agricultural field
near Mohali, Punjab State, India. It was air dried under sunlight to a
final moisture content of < 6% wt, ground to a final particle size ran-
ging between 5 and 4 mm using a commercial grinder machine
(Kinematica PX-MFC 90 D) and followed by, sieved to obtain a uniform
particle size of 0.5 mm (30 mesh).
2.3. Fractional composition determination
The fractional compositional analysis was performed on the solvent
pre-extracted rice straw using standard National Renewable Energy
Laboratory (NREL) protocol in order to determine its primary con-
stituents, such as carbohydrate sugars, lignin, and ash (Elumalai et al.,
2016a). Solvent pre-extraction was done on native rice straw using a
solvent mixture of ethanol/benzene at 80:20 ratio in a Soxhlet extrac-
tion apparatus for overnight (or at least 8 siphons). The solids collected
were then oven dried for overnight at 105 °C.
2.4. Single-pot reaction of rice straw fiber
Powdered rice straw (30 mesh) or cellulose was soaked using dilute
HCl solution (aqueous) and fresh organic solvent (DCM) in 100 mL
stainless steel autoclave vessel containing Teflon liner. A magnetic
stirring bead was added to the vessel and sealed tightly to avoid any
solvent escape. The vessel was heated to the desired temperature (up to
200 °C) using a magnetic stirring hot plate (Thermofisher). After com-
pletion of reaction (3h), the post-reaction medium was collected and
transferred to a separating funnel. Approximately 2 volumes of fresh
DCM was added externally into the dark brown colored medium in
order to achieve a clear separation of organic and aqueous phases.
Further, the separated organic phase was evaporated under vacuum
S. Kumar et al.
condition using rotary evaporator at 30 °C. The recovered solvent was
stored separately (R1) for reuse purpose. The extractant (E1) was diluted
using methanol and analyzed on high-performance liquid chromato-
graphy (HPLC, Agilent Technologies 1200 infinity series) equipped
with Agilent Hi-Plex H column (300 mm length; 8 µm porosity) and RI
detector operated at 55 °C. The column conditions maintained were
60 °C, 0.7 mL/min of 5 mM H2SO4 (mobile eluent). Before injection, the
samples were filtered using 0.2 µm nylon filter. LA and other residuals
concentration, preferentially HMF, FA, furfural were calculated using
their respective calibration curves prepared using known amount. The
percent yield of the product chemicals was calculated to the initial
biomass (rice straw) weight. LA conversion yield was calculated based
on the potential glucose sugar available in native rice straw fiber to
product formation (dry wt basis). All experiments were performed in
duplicate and the data reported is the average of duplicate runs.
The protocol was repeated to produce LA from native rice straw but
the fresh solvent was replaced by R1 (recovered solvent having neutral
pH) in order to test for its recyclable efficiency.
2.5. Glucose and HMF conversion to LA
Thermochemical conversion of commercial glucose (∼1.0 gm) or
HMF (∼10 mg) to LA reaction was performed in a thick-walled glass
reactor (38 mL) consisted of DCM and acidic reagents either LA or HCl
or formic acid at desired concentrations depending on its strength.
Aqueous to organic ratio maintained were between 1:1 and 1:9 (v/v).
The reactor was sealed tightly to avoid any escape of solvent after the
addition of magnetic stirring bead. The reaction was conducted in an oil
bath at 180 °C for 3 h using a magnetic stirring hot plate
(Thermofisher). After completion of the reaction, the organic phase was
collected and evaporated under vacuum using rotary evaporator at
30 °C. The resultant products were diluted using methanol and analyzed
on HPLC by adapting the same method, as described earlier. LA and
residual HMF concentrations were calculated using their respective
calibration curves prepared using known amount. All experiments were
performed in triplicates.
2.6. LA purification on column chromatography
The collected E1 before dilution using methanol was dried in an
oven at 50 °C for 24 h in order to remove residual water content, and
the resultant black residue was transferred to the column packed with
60–120 mesh silica gel (Sigma Aldrich India). The eluent system used
was 30% ethyl acetate in pure hexane. Fractions consisting pure LA was
collected together, based on the thin layer chromatography (TLC) re-
sults ran using the same solvent system as mobile eluent. Then the
organic solvents were evaporated using rotary vacuum evaporated at
30 °C to obtain purified target product. The final concentration of LA
was quantified on HPLC after dilution using distilled water (C1) by
adopting the same method, as mentioned earlier. Percent recovery of
LA was calculated as the amount differences between initial supply and
after column purification.
2.7. Charcoal treatment
The chromatographically purified product chemical was further
treated with activated carbon to remove the colored impurities. Briefly,
5 mL of C1 (aqueous) was taken in 20 mL thick-walled glass tube added
with 5% activated carbon (100 mesh particle size). The mixture was
continuously stirred for 1 h at room temperature. After thorough
mixing, it was then filtered using Whatman No. 1 filter paper to obtain a
clear LA product (C2). Final purity of the product compound was ver-
ified on HPLC by adopting the same method, as mentioned earlier.
Percent recovery of LA was calculated as the amount differences be-
tween initial supply and after treatment.
145
Bioresource Technology 251 (2018) 143–150
2.8. Analytical instrumentation
Scanning electron microscopy: Post-reaction solid residues of different
reactions conducted with and without solvent combinations were col-
lected, washed repeatedly using distilled water and freeze-dried (Christ,
Delta 2-24 LSC plus). The completely dried native rice straw and post-
reaction solid residues were coated with gold in a smart coater (25 mm
stub size) and scanned by scanning electron microscope (JEOL JCM
6000, Nikon Corporation, Japan) at an acceleration voltage of 10.0 kV.
NMR spectroscopy: 1H and 13C NMR spectrum were acquired for the
post-charcoal treatment LA sample C2 on a Bruker Advance III 600 MHz
spectrometer. The sample was diluted in D2O solvent (Victor et al.,
2014).
Elemental analysis: The composition of C, H and N in the pre- and
post-treatment solid samples (activated charcoal) were determined
using an elemental analyzer (Thermo Scientific Flash 2000 HT). Before
analysis, the samples were dried overnight in the oven at 50 °C in order
to remove the residual water content.
2.9. Statistical data analysis
Statistical regression and ANOVA analyses were performed to the
products recovery data using SPSS software v23. The quality of re-
gression model fit was expressed by the squared correlation coefficient,
R2 value. Model terms were evaluated based on the probability (p)
value with 95% confidence. Multiple (pairwise) comparisons among
products recovery means during the biphasic reaction were carried out
using Tukey’s honestly significance difference (HSD) test at α levels of
0.05.
3. Results and discussion
3.1. Influence of reaction medium for LA formation
Considering the advantages of DCM’s good extraction ability of
furans (HMF), the idea was to allow further degradation to LA in or-
ganic medium under limited acidic environment, since water solubility
of DCM is < 10% wt, thereby possibly controlling the unwanted for-
mation of humins which requires water molecules under elevated
conditions (Liu et al., 2014). The activation energy of humin reaction is
relatively higher than LA conversion (33% at 200 °C using cellulose)
(Shen and Wyman, 2012). In practice, the reaction requirements for
H2O molecules during LA conversion can be fulfilled by mass transfer
during the continuous countercurrent circulation phenomenon of DCM
(density, ρ = 1.33 g/cc) and water (ρ = 1.0 g/cc) after reaching its sa-
turation equilibrium, based on the density differences. Moreover, mo-
lecules with different existing forms behave differently when dis-
tributed between aqueous and organic phases (McFearin and
Richmond, 2009). Nevertheless, DCM can undergo hydrolysis reactions
in the presence of H2O molecules at elevated temperatures, around
250 °C, generating HCl with possible formation of formaldehyde
(O]CH2), depending on the degradation reaction mechanism
(Wettstein et al., 2012). By taking account of all the above character-
istic features of DCM, it can be reasonably expected that DCM could
favor improved LA synthesis in the presence of acidic aqueous solu-
tions. To test this hypothesis, separate glucose and HMF reactions were
conducted in different reaction systems under elevated conditions
(180 °C for 3 h, since LA is more stable at 180 °C up to 24 h) (Yan et al.,
2008). Comparative preliminary glucose conversion result showed
(Fig. 1) that DCM substantially increased the final LA yield (relatively
55% mol. higher yield than the control experiment using only water)
with increased solvent volume resulting in improved yield (∼21% mol.
increase with 19% selectivity) when the acid dosage was maintained
same. These results highlight the impact of product distribution equi-
librium between aqueous and organic phases. Furthermore, little to no
residual HMF was detected in the post-reaction mediums, indicating the
S. Kumar et al. Bioresource Technology 251 (2018) 143–150

Fig. 1. Comparative result of glucose conversion in dif-

ferent monophasic and biphasic systems at 180 °C for 3 h.
Error bars represent standard error. Asterisks (*) and (**)
indicate significance and not significance at 95% confidence
level, respectively. Yx and Sx are the conversion yield and
selectivity data of the respective compounds, respectively,
where x can be LA, HMF, FF. N.D-not detected.

Table 1 lignocellulosic substrate to LA and further purification. Rice straw was

Concentration result of residuals left during glucose conversion reaction in monophasic selected as the substrate and was measured to indicate a composition of
and biphasic systems.
40.4% glucose, 18.1% xylose, 26% total of acid soluble and insoluble
Entry Medium HCl load on Conditions Residual Residual
lignins, and 11% ash contents (dry wt basis). In a typical lignocellulose
system (v/v) glucose (wt HMF furfural reaction, the influence of the raw material upon LA formation is quite
per wt) (% wt) (% wt) important. The presence of lignin changes the optimum reaction con-
ditions as it requires more severe conditions to open up the tightly
1 Only water 0.4:1 180 °C for 3 h 2.65 0.37
(W)
packed structure compared to cellulose (Szabolcs et al., 2013). Indeed,
2 W:DCM 0.4:1 180 °C for 3 h 0.08 0.27 the effectiveness of catalyst depends on concentration, strength of their
ratio 1:1 primary dissociation constant, type and concentration of the employed
3 W:DCM 0.4:1 180 °C for 3 h 0.30 0.49 raw material, and severity of the reaction, in terms of temperature and/
ratio 1:9
or time (Chang et al., 2007). Therefore, maximum conversion efficiency
4 Only DCM 0.4:1 180 °C for 3 h 0.00 0.07
can only be achieved through process optimization. Consequently, the
effect of acid concentration on LA preparation was studied in the range
biphasic system enabled maximum conversion of HMF through con- between 0.5 and 2.5 gm HCl per gm solid by maintaining aqueous to
tinuous extraction strategy, as data presented in Table 1. However, the DCM ratio 1:5 at 180 °C for 3 h, with the results observed for changes in
setup could achieve only 58% LA of the theoretical yield, presumably the products formation, including LA, FA and FF with respect to the
due to insufficient supply of acid to promote sugar degradation (Li incremented reaction severity. Slightly improved LA conversion was
et al., 2017). recorded (14.7% wt with 2% relative difference) in response to the
Meanwhile, HMF reactions were conducted in the same fashion immediate incremented acid dose from the initial level (exactly 2-times
using HCl, LA, and FA at varying concentrations depending on the more load). Further increase of the acid loads negatively affected the LA
strength, in order to verify its effectiveness on LA formation. yields i.e., with up to 2.5-fold lower concentrations (Table 2 entries
Interestingly, carboxylic acids LA (pKa = 4.59) and FA (pKa = 3.75) 1–4). The effect was not limited to LA; a somewhat similar trend was
facilitated LA formation regardless of their acidic strength. For instance, observed to the residual precursors conversion i.e., relative differences
the former accounted for 18% mol. yield with 98% selectivity, whereas of ≥2-fold lower HMF and FF yields were obtained beyond the op-
the later promoted 33% mol. conversion with 93% selectivity (Fig. 2). timum acid dose levels (1.2 gm HCl per gm solid). These results could
However, no sign of progression of HMF reaction using FA in the bi- be attributed to the easy decomposition of HMF under higher acid
phasic medium was observed, mainly due to its greater affinity to concentrations to humins in the presence of H2O either in vapor phase
aqueous medium. Contrary to expectations, the influence of DCM on or condensed state below the liquid medium, thereby reducing LA yield.
HMF conversion was quite low (< 1.0% mol. LA) under the modest FF is generally unstable under highly acidic conditions at elevated
reaction conditions (180 °C for 3 h), which might have reduced the temperatures and, thus, contributed to the loss of side reactions (pre-
chance of self-degradation to produce HCl. Overall, preliminary results ferentially humins generation) (Bernal et al., 2014; Gürbüz et al.,
demonstrated that the maximum carbohydrate conversion to LA in 2012). Furthermore, higher acid dosages significantly affected the LA
plain biphasic system with the aid of mineral acid and products (auto- recovery to organic medium (up to 14-fold lower relative differences of
catalysis i.e., products formed acts as catalyst). RLA which is the ratio of concentration of LA in organic phase to aqu-
eous phase) assumed to be partitioning of the mineral acid into the
organic layer, thereby initiating its attraction to H2O molecules.
3.2. Lignocellulose to LA in biphasic solvent system Nevertheless, increased release of glucose molecules (relatively up to
25% higher recovery) was found at the higher acid loadings due to the
The approach was further validated on deconstruction of enhanced cleavage of 1,4-β-glycosidic linkages of biomass cellulose

146
S. Kumar et al. Bioresource Technology 251 (2018) 143–150

Fig. 2. Comparative result of HMF to LA conversion in

different monophasic and biphasic systems at 180 °C for
3 h. Error bars represent standard error. Asterisks (*) and
(**) indicate significance and not significance at 95%
confidence level, respectively. N.D-not detected.

(Kumar et al., 2009). The biphasic system achieved the highest con-
version to LA (∼52% wt theoretical) with 3.8% wt HMF as residual
concentration under optimum conditions (Table 2 entry 2). This sug-
gests an influence of equilibrium product and other residuals distribu-
tion between aqueous and organic mediums under relatively harsh
conditions along with solvent effect (extraction capability) (Cai et al.,
2013). Further influence of solvent volume on LA formation was stu-
died in the range between 1:1 and 1:5 (aqueous to organic ratio) and
keeping all other parameters fixed (1.2 gm HCl dose at 180 °C for 3 h).
Results showed that the introduction of organic solvent greatly accel-
erates the recovery of product compounds and proportionately in-
creased with the additional volumes (Table 2 entries 5–9). For instance,
∼4-fold higher LA concentration was achieved by introducing equal
volume of DCM into the reaction medium and further added DCM
volumes steadily improved the yields (significance at p < .05) aver-
aging 8% relative increments at each increased DCM additions.
Therefore, equilibrium products distribution greatly influenced the
rate of LA formation, based on the residuals conversion data i.e., glu-
cose utilization rate was decreased at R2 = 0.96 (significance at
p < .05), whereas HMF showed a contrary response of R2 = 0.87. In
support, the RLA was maintained averagely 1.0 up to 4 added volumes
of DCM and further increased organic content improved the ratio.
Nearly same production strategy was observed with FF formation.
Overall, the results imply that large volumes of solvent are required to
achieve maximum LA conversion, which permits DCM to go pre-
ferentially to the vapor phase during the reaction at high temperatures
with the desired effect of extracting synthesis compounds from the
aqueous phase. However, use of large volumes of solvent would not

Table 2
Response yield data of levulinic acid and other associated products with respect to the change in conditions during single-pot reaction of rice straw.

Entry HCl load on rice straw Temp Water to DCM ratio Time LA yield FA Residual precursors yield
(gm per gm) (°C) (min) (% wt) (% wt) (% wt)

Glucose HMF FF

1 0.5 180 1:5 180 14.4(1.5) 2.4(0.1) 4.6 2.5(4.0) 11.7(1.8)

2 1.2 180 1:5 180 14.7(4.3) 2.3(0.3) 2.8 2.9(13.5) 12.5(3.6)
3 1.8 180 1:5 180 11.7(0.7) 1.5(0.3) 3.1 3.8(11.7) 7.8(2.9)
4 2.5 180 1:5 180 5.9(0.4) 2.6(0.1) 3.5 1.9(18.0) 6.1(0.1)
5 1.2 180 Only water 180 2.5(1.8) 1.8(0.2) 16.8 1.5(14.0) 7.8(1.1)
6 1.2 180 1:1 180 9.0(0.7) 1.3(0.2) 7.8 1.2(5.0) 1.3(3.3)
7 1.2 180 1:2 180 10.5(0.9) 1.7(0.1) 7.1 1.7(7.5) 3.5(0.8)
8 1.2 180 1:3 180 12.7(1.1) 1.6(0.1) 4.2 0.7(6.0) 10.2(0.1)
9 1.2 180 1:4 180 13.8(1.7) 3.7(0.1) 3.7 3.6(17.0) 10.8(0.5)
10 1.2 140 1:5 180 4.4(1.4) 1.6(0.3) 4.5 6.4(63.0) 11.0(26.5)
11 1.2 160 1:5 180 10.9(1.3) 1.9(0.3) 4.2 4.3(20.5) 10.1(4.1)
12 1.2 200 1:5 180 16.6(0.7) 2.0(0.1) 2.6 0.6(5.0) 12.6(0.8)
13 1.2 180 1:5 (W:THF) 180 3.4(33.0) 1.1(0.1) 9.4 6.5(64.0) 15.4(15.1)
14 1.2 on cellulose 180 1:5 180 38.3(2.4) 14.5(0.1) 3.5 1.6(15.0) 3.7(6.4)

Tukey’s HSD 6.9* 2.0** 2.0** 4.6* 4.6*

Number in parentheses represent partition coefficient of the respective compound calculated based on the concentration present in organic phase to aqueous phase. Asterisks (*) and (**)
indicate significance at p < .05 and significance at p < .1, respectively. Data show the reaction products yield based on solid dry wt W-water; LA-levulinic acid; FA-formic acid; HMF-5-
hydroxymethylfurfural; FF-furfural.

147
S. Kumar et al. Bioresource Technology 251 (2018) 143–150

likely be a cost-effective processing method. Table 3

In a typical reaction, temperature plays a critical role in determining Comparative carbohydrate analysis data of post-reaction solids obtained from different
monophasic and biphasic reaction systems at 1.2 gm HCl per gm biomass solids.
the extent of the reaction. During LA formation, higher temperatures
generally accelerate the rate of chemical reactions along with undesired Entry Medium Conditions Solids Residual Total
side reactions, leading to the production of humins, due to the higher system recovery carbohydrate carbohydrate
equilibrium concentration of protonated pyranose forms of the sugar (v/v) (wt%) content yield
molecules (Alipour and Omidvarborna, 2016; Chang et al., 2007). The (wt%)a (wt%)

influence of temperature on LA formation using rice straw was studied 1 W:DCM 140 °C for 3 h 52.5 40.6 52.0
in the range between 140 °C and 200 °C and other parameters were ratio 1:5
maintained constant i.e., 1.2 gm HCl dose and aqueous to organic ratio 2 W:DCM 160 °C for 3 h 51.1 15.0 18.7
of 1:5 for 3 h (Table 2 entries 10–12). There is a risk of LA degradation ratio 1:5
3 W:DCM 180 °C for 3 h 48.6 14.3 17.0
to unsaturated lactones through dehydration reaction in the presence of
ratio 1:5
acid catalyst at elevated temperatures (∼230 °C) (Yan et al., 2008). 4 W:DCM 200 °C for 3 h 45.8 13.5 15.1
Results implied that temperature had a significant effect on LA forma- ratio 1:5
tion. For instance, more than a 2-fold incremented yield was achieved R2 0.99** 0.85** 0.86**
as the temperature was increased from 140 °C to 160 °C within 6 min. 5 Only W 180 °C for 3 h 54.5 58.7 78.0
Upon further temperature increases, LA yields increased by 50–80%. 6 W:THF 180 °C for 3 h 34.0 35.6 29.5
However, no such trend was observed with FF production, suggesting ratio 1:5

that the presence of DCM increases the stability of FF at elevated con- a

ditions and therefore, reaching its maximum productivity of ∼70% of
the potential xylose possibly through the complete dissolution of bio-
mass hemicelluloses under the optimum conditions (Bernal et al.,
2014). Overall, acid concentration and temperature had a most sig-
nificant impact on LA and other products formation when compared to
other reaction parameters, based on statistical analysis (see E-supple-
ment data).
Although higher temperatures yielded maximum product con-
centration, the reduced partition coefficient achieved might be attrib-
uted to the increased population of unwanted products, possibly hu-
mins that expelled LA to the aqueous phase due to having a naturally
higher affinity to H2O molecules (Liu et al., 2014). Therefore, the
modified system achieved up to ∼15% wt total LA concentration from
native rice straw substrate (∼57% of theoretical) with maximum re-
covery of LA to the organic phase (RLA = 4.3) under optimum reaction
conditions (15% wt HCl conc. and 180 °C for 180 min) along with other
associated products (including 2.3% FA, 2.9% HMF, and 12.5% wt FF).
However, it is ∼13% point lower LA yield than the literature reports
using lignocelluloses, where harsher conditions were employed (6–32%
wt mineral acid concentration and 190–220 °C) (Galletti et al., 2011;
Mukherjee et al., 2015). In addition to several root cause for the lower
yield achievement, frustratingly, the equally high yield of furfural from
lignocellulose derived xylose negatively affects the final LA yield due to
the polymerization reaction with saccharides leading to the insoluble
humins formation (Elumalai et al., 2016b). Whereas, the impact of re-
sidual lignin on LA formation is negligible.
To further test the hypothesis, an analogous trial run involving THF
by replacing DCM during rice straw conversion (monophasic) resulted
in relatively ∼4-fold lower yield with considerably enriched residuals
concentration (i.e., approximately 3-fold glucose, 2-fold HMF and 6-
fold FF) than DCM biphasic system (Table 2 entry 13). However, the
biphasic system yielded comparatively lower conversion during direct
cellulose (∼16 percentage points of LA) than the literature studies,
where the medium was modified by the addition of salts or sugars to
achieve maximum partition coefficient (Wettstein et al., 2012).
Total carbohydrate includes mainly glucose and xylose. Asterisk (**) indicate not
significance at 95% confidence level. W-water.
significant effect on the release of more carbohydrate sugars via hy-
drolysis reaction for the active participation in the furans synthesis.
However, in all the cases negligible levels of xylose content (≤1% wt)
were recovered ascribed to the enhanced hydrolysis of hemicellulose
sugars than cellulose with the aid of HCl. Comparative THF reaction
yielded much lower solids recovery with residual carbohydrate content,
indicating progression of more hemicellulose degradation to make them
available for furfural production. The statement was in good agreement
with the furfural yield data. It can be noted that addition of organic
solvent (DCM) greatly helped for the carbohydrate conversion through
preventing equilibrium product saturation.
Evident morphological analysis result of the post-reaction solids
visualized for the corresponding products yield increment at higher
temperatures due to heavy surface destruction, thereby increasing ac-
cessibility of carbohydrates to chemical reactions (see E-supplement
data). Comparative SEM images genuinely revealed that solids en-
countered severe surface destruction under relatively increasing reac-
tion severities. It is also visible from the imaging results that DCM
greatly helps for the solid deconstruction i.e., a sponge-like surface,
implying more supply of carbohydrates for furans production due to
continuous extraction to the organic phase, thereby preventing the
undesired polymerization reactions. In complementary to the above
statement, condensed droplets were spotted on the solids underwent
reaction in only water medium, caused due to repolymerization reac-
tions of polysaccharides degradation products and/or polymerization
with lignin (so-called pseudo lignin), which typically takes place during
dilute acid pretreatment of lignocellulose under higher severity condi-
tions (> 120 °C and ≤4% acid conc.). Further analysis of the solids
revealed that it consisted of 14.3% carbohydrate (majorly glucose),
73.2% acid insoluble content and 11.4% ash as residual constituents
and therefore, in the context of biomass biorefinery, this carbohydrate
free material can be used for further preparation of silica- and lignin-
derived products (Khaleghian et al., 2017; Minu et al., 2012).

3.3. Effectiveness of biphasic system on lignocellulose transformation

Investigating post-reaction solids residual carbohydrate content
(Table 3), it was found that higher temperatures facilitate preferentially
sugar degradation reactions, since LA is less sensitive to higher tem-
peratures (≤230°) (Shen and Wyman, 2012). While incremented tem-
peratures remarkably reduced the solids recovery along with residual
carbohydrate content (mainly glucose and xylose) in biphasic system.
For instance, linearly decreasing fashion of yield recoveries were
achieved throughout the rising temperatures (R2 = 0.99 for solids and
R2 = 0.86 for total carbohydrate), indicating temperature had a
3.4. LA separation and purification
During downstream processing, the typically employed methods are
vacuum distillation, solvent-extraction, precipitation, over-liming, and
adsorption with ion-exchange resin or basic polymers. However, these
methods are considered unattractive due to the high cost, competitive
product adsorption, surface deactivation and environmental toxicity
(Brouwer et al., 2017; Mussatto and Roberto, 2004). Alternatively,
HPLC separation protocol is also followed but this method employs
aqueous sulfuric acid as mobile eluent and encounters difficulties in

148
S. Kumar et al.
Fig. 3. Overall mass balance of the process method of rice straw deconstruction to furan and derived compounds.
separation of LA and acetic acid (typically formed during acidic de-
gradation of hemicelluloses) due to co-elution (Xie et al., 2011). In
order to achieve effective separation, a traditional column chromato-
graphy technique was employed for LA purification from associated
product compounds (Frade et al., 2014), including HMF, furfural, and
traces of formic acid. Requisite sample preparations were followed,
despite their drawbacks such as time requirements and involvement of
large volumes of expensive solvent. The result of thin layer chromato-
graphy (TLC) and HPLC analyses displayed that high purity LA se-
paration with ∼85% wt recovery was achieved (see E-supplement
data). However, the resultant product appeared dark brown in color,
attributable to the soluble contaminants derived from the acidic de-
gradation reactions of lignin and humin (Rodrigues et al., 2001).
Therefore, activated charcoal treatment was performed to yield a col-
orless liquid having ∼99% LA purity with about 95% wt recovery (see
E-supplement data), based on the HPLC and NMR (1H, 13C) analyses
(see E-supplement data) (Ahlkvist et al., 2013; Victor et al., 2014).
Further elemental analysis of the post-treatment solids (activated
charcoal) data found consistent with the LA yield recovery i.e., ∼5%
relative differences in carbon content (the determined C content of pre-
and post-treatment solid samples were 70.90 and 74.67% dry wt basis,
respectively). Based on the overall material balance, the biphasic
system was capable of producing ∼10% wt net industrial grade LA
from native rice straw biomass (Fig. 3). Although the process demon-
strated for facile lignocellulose processing to value addition, the pro-
cessing cost, energy-intensive product preparation, and prolonged
purification through column chromatography, might be the challenges
during large-scale processing.
3.5. Solvent recycle efficiency
The designed system allowed achieving up to 90% DCM recovery
from the high severity reactions. The recovered solvent was further
tested for its recyclability performances on concurrent LA production
runs. Evident results demonstrated that it can be used for up to 5 re-
cycle runs on effective LA production (significance at 95% confidence
level) at moderate reaction conditions (see E-supplement data). These
results imply that employed condition was insufficient to encourage
149
Bioresource Technology 251 (2018) 143–150
DCM self-degradation, thereby enabling cost-effective process method
development and potentially reducing the discharge of environmentally
toxic halogenated material as waste.
4. Conclusions
The present study demonstrates a facile approach to the preparation
of LA from rice straw. The method involves, employing a co-solvent
system, consisting of HCl and DCM, as extraction solvent to improve the
final product yield through continuous extraction of furans. Further
isolating and purifying of the target compound through conventional
separation techniques. The addition of DCM greatly improves product
enrichment during the reaction along with the aid of HCl. It also per-
mits maximum recovery of LA, along with HMF and FF as residuals, to
the organic phase. Nearly complete recovery and reuse are also pos-
sible.
Acknowledgements
The authors (SK, PK, RSS, and SE) gratefully thank Department of
Biotechnology, New Delhi, India for their consistent financial support.
VA thanks DST-SERB, India for financial support. SE acknowledge
Department of Science and Technology (DST-SERB), New Delhi for the
support through Grant No. YSS/2014/000031(2015-18). The authors
thank NMR facility at IISER Mohali for recording the spectra.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.biortech.2017.12.033.
References
Ahlkvist, J., Ajaikumar, S., Larsson, W., Mikkola, J.-P., 2013. One-pot catalytic conver-
sion of Nordic pulp media into green platform chemicals. Appl. Catal. A 454, 21–29.
Alipour, S., Omidvarborna, H., 2016. High concentration levulinic acid production from
corn stover. RSC Adv. 6 (112), 111616–111621.
Alonso, D.M., Wettstein, S.G., Mellmer, M.A., Gurbuz, E.I., Dumesic, J.A., 2013.
Integrated conversion of hemicellulose and cellulose from lignocellulosic biomass.
Energy Environ. Sci. 6 (1), 76–80.
S. Kumar et al.
Bernal, H.G., Bernazzani, L., Galletti, A.M.R., 2014. Furfural from corn stover hemi-
celluloses. A mineral acid-free approach. Green Chem. 16 (8), 3734–3740.
Brouwer, T., Blahusiak, M., Babic, K., Schuur, B., 2017. Reactive extraction and recovery
of levulinic acid, formic acid and furfural from aqueous solutions containing sul-
phuric acid. Separ. Purif. Technol.
Byun, J., Han, J., 2016. Process synthesis and analysis for catalytic conversion of lig-
nocellulosic biomass to fuels: separate conversion of cellulose and hemicellulose
using 2-sec-butylphenol (SBP) solvent. Appl. Energy 171 (Supplement C), 483–490.
Caes, B.R., Teixeira, R.E., Knapp, K.G., Raines, R.T., 2015. Biomass to furanics: renewable
routes to chemicals and fuels. ACS Sustain. Chem. Eng. 3 (11), 2591–2605.
Cai, C.M., Zhang, T., Kumar, R., Wyman, C.E., 2013. THF co-solvent enhances hydro-
carbon fuel precursor yields from lignocellulosic biomass. Green Chem. 15 (11),
3140–3145.
Chang, C., Cen, P., Ma, X., 2007. Levulinic acid production from wheat straw. Bioresour.
Technol. 98 (7), 1448–1453.
Chen, S.S., Yu, I.K.M., Tsang, D.C.W., Yip, A.C.K., Khan, E., Wang, L., Ok, Y.S., Poon, C.S.,
2017. Valorization of cellulosic food waste into levulinic acid catalyzed by hetero-
geneous Brønsted acids: temperature and solvent effects. Chem. Eng. J. 327
(Supplement C), 328–335.
Chheda, J.N., Román-Leshkov, Y., Dumesic, J.A., 2007. Production of 5-hydro-
xymethylfurfural and furfural by dehydration of biomass-derived mono-and poly-
saccharides. Green Chem. 9 (4), 342–350.
Elumalai, S., Agarwal, B., Runge, T.M., Sangwan, R.S., 2016a. Integrated two-stage
chemically processing of rice straw cellulose to butyl levulinate. Carbohydr. Polym.
150, 286–298.
Elumalai, S., Agarwal, B., Sangwan, R.S., 2016b. Thermo-chemical pretreatment of rice
straw for further processing for levulinic acid production. Bioresour. Technol. 218,
232–246.
Frade, R.F., Coelho, J., Simeonov, S., Afonso, C.A., 2014. An emerging platform from
renewable resources: selection guidelines for human exposure of furfural-related
compounds. Toxicol. Res. 3 (5), 311–314.
Galletti, A.M.R., Antonetti, C., De Luise, V., Valentini, G., 2011. Conversion of biomass to
levulinic acid, a new feedstock for the chemical industry. Chim. Ind. 93, 112–117.
Gallo, J.M.R., Alonso, D.M., Mellmer, M.A., Dumesic, J.A., 2013. Production and up-
grading of 5-hydroxymethylfurfural using heterogeneous catalysts and biomass-de-
rived solvents. Green Chem. 15 (1), 85–90.
Girisuta, B., Dussan, K., Haverty, D., Leahy, J.J., Hayes, M.H.B., 2013. A kinetic study of
acid catalysed hydrolysis of sugar cane bagasse to levulinic acid. Chem. Eng. J. 217
(Supplement C), 61–70.
Gürbüz, E.I., Wettstein, S.G., Dumesic, J.A., 2012. Conversion of hemicellulose to furfural
and levulinic acid using biphasic reactors with alkylphenol solvents. ChemSusChem 5
(2), 383–387.
Khaleghian, H., Molaverdi, M., Karimi, K., 2017. Silica removal from rice straw to im-
prove its hydrolysis and ethanol production. Ind. Eng. Chem. Res. 56 (35),
9793–9798.
Kumar, P., Barrett, D.M., Delwiche, M.J., Stroeve, P., 2009. Methods for pretreatment of
lignocellulosic biomass for efficient hydrolysis and biofuel production. Ind. Eng.
Chem. Res. 48 (8), 3713–3729.
Li, M., Li, W., Lu, Y., Jameel, H., Chang, H.-M., Ma, L., 2017. High conversion of glucose
to 5-hydroxymethylfurfural using hydrochloric acid as a catalyst and sodium chloride
as a promoter in a water/γ-valerolactone system. RSC Adv. 7 (24), 14330–14336.
Liu, F., Sivoththaman, S., Tan, Z., 2014. Solvent extraction of 5-HMF from simulated
hydrothermal conversion product. Sustain. Environ. Res. 24 (2).
McFearin, C.L., Richmond, G.L., 2009. The role of interfacial molecular structure in the
150
Bioresource Technology 251 (2018) 143–150
adsorption of ions at the liquid− liquid interface. J. Phys. Chem. C 113 (50),
21162–21168.
Minu, K., Jiby, K.K., Kishore, V.V.N., 2012. Isolation and purification of lignin and silica
from the black liquor generated during the production of bioethanol from rice straw.
Biomass Bioenergy 39 (Supplement C), 210–217.
Morone, A., Apte, M., Pandey, R., 2015. Levulinic acid production from renewable waste
resources: Bottlenecks, potential remedies, advancements and applications. Renew.
Sustain. Energy Rev. 51, 548–565.
Mukherjee, A., Dumont, M.-J., Raghavan, V., 2015. Sustainable production of hydro-
xymethylfurfural and levulinic acid: challenges and opportunities. Biomass Bioenergy
72, 143–183.
Mussatto, S.I., Roberto, I.C., 2004. Alternatives for detoxification of diluted-acid lig-
nocellulosic hydrolyzates for use in fermentative processes: a review. Bioresour.
Technol. 93 (1), 1–10.
Pileidis, F.D., Titirici, M.M., 2016. Levulinic acid biorefineries: new challenges for effi-
cient utilization of biomass. ChemSusChem 9 (6), 562–582.
Rackemann, D.W., Doherty, W.O., 2011. The conversion of lignocellulosics to levulinic
acid. Biofuels Bioprod. Biorefin. 5 (2), 198–214.
Rodrigues, R., Felipe, M., Silva, J., Vitolo, M., Gómez, P., 2001. The influence of pH,
temperature and hydrolyzate concentration on the removal of volatile and non-
volatile compounds from sugarcane bagasse hemicellulosic hydrolyzate treated with
activated charcoal before or after vacuum evaporation. Braz. J. Chem. Eng. 18 (3),
299–311.
Runge, T., Zhang, C., 2012. Two-stage acid-catalyzed conversion of carbohydrates into
levulinic acid. Ind. Eng. Chem. Res. 51 (8), 3265–3270.
Sarkar, N., Ghosh, S.K., Bannerjee, S., Aikat, K., 2012. Bioethanol production from
agricultural wastes: an overview. Renew. Energy 37 (1), 19–27.
Shen, J., Wyman, C.E., 2012. Hydrochloric acid-catalyzed levulinic acid formation from
cellulose: data and kinetic model to maximize yields. AIChE J. 58 (1), 236–246.
Szabolcs, Á., Molnár, M., Dibó, G., Mika, L.T., 2013. Microwave-assisted conversion of
carbohydrates to levulinic acid: an essential step in biomass conversion. Green Chem.
15 (2), 439–445.
Victor, A., Pulidindi, I.N., Gedanken, A., 2014. Levulinic acid production from Cicer ar-
ietinum, cotton, Pinus radiata and sugarcane bagasse. RSC Adv. 4 (84),
44706–44711.
Weiqi, W., Shubin, W., 2017. Experimental and kinetic study of glucose conversion to
levulinic acid catalyzed by synergy of Lewis and Brønsted acids. Chem. Eng. J. 307
(Supplement C), 389–398.
Wettstein, S.G., Alonso, D.M., Chong, Y., Dumesic, J.A., 2012. Production of levulinic acid
and gamma-valerolactone (GVL) from cellulose using GVL as a solvent in biphasic
systems. Energy Environ. Sci. 5 (8), 8199–8203.
Xie, R., Tu, M., Wu, Y., Adhikari, S., 2011. Improvement in HPLC separation of acetic acid
and levulinic acid in the profiling of biomass hydrolysate. Bioresour. Technol. 102
(7), 4938–4942.
Yan, L., Yang, N., Pang, H., Liao, B., 2008. Production of levulinic acid from bagasse and
paddy straw by liquefaction in the presence of hydrochloride acid. CLEAN–Soil Air
Water 36 (2), 158–163.
Yang, Y., Hu, C.-W., Abu-Omar, M.M., 2012. Conversion of carbohydrates and lig-
nocellulosic biomass into 5-hydroxymethylfurfural using AlCl 3· 6H 2 O catalyst in a
biphasic solvent system. Green Chem. 14 (2), 509–513.
Zhao, W., Li, Y., Song, C., Liu, S., Li, X., Long, J., 2017. Intensified levulinic acid/ester
production from cassava by one-pot cascade prehydrolysis and delignification. Appl.
Energy.