Accepted Manuscript

Title: Production of levulinic acid from corn cob residue in a

fed-batch acid hydrolysis process

Authors: Chen Liang, Yangdong Hu, Yan Wang, Lianying Wu,

Weitao Zhang

PII: S1359-5113(18)30572-5
DOI: https://doi.org/10.1016/j.procbio.2018.08.002
Reference: PRBI 11413

To appear in: Process Biochemistry

Received date: 16-4-2018

Revised date: 13-7-2018
Accepted date: 1-8-2018

Please cite this article as: Liang C, Yangdong H, Wang Y, Lianying W, Zhang W,
Production of levulinic acid from corn cob residue in a fed-batch acid hydrolysis process,
Process Biochemistry (2018), https://doi.org/10.1016/j.procbio.2018.08.002

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Production of levulinic acid from corn cob residue in a fed-batch

acid hydrolysis process

Chen Lianga, Yangdong Hua*, Yan Wangb, Lianying Wua, and Weitao Zhanga

a.
College of Chemistry and Chemical Engineering, Ocean University of China, 238

T
Songling Road, Qingdao , PR China

IP
b.
College of Applied Technology, Qingdao University, 93 Songling Road, Qingdao,

R
PR China

SC
*Correspondence to: Yangdong Hu, Ocean University of China, 238 Songling Road,
Qingdao, 266100 PR China

Tel: +86 0532-66782141 U

N
E-mail: ydhuhd@ouc.edu.cn
A
Graphical Abstract
M
ED
PT

Highlights

 A fed-batch process for producing high concentration of levulinic acid

 The hydrolysis process could effectively reduce the consumption of sulfuric acid
E

 The soluble humin analogues affect the yield of levulinic acid

CC

 The yield of levulinic acid was not influenced by lignocellulose, salts and ash

ABSTRACT:
A

Levulinic acid (LA) is an important platform chemical, the production of which by

using biomass resources such as corncob is of great significance to the sustainable

development. Traditional hydrolysis processes yield low concentrations of levulinic

1
acid with large amounts of acid being consumed. In this paper, a new fed-batch

process for hydrolyzing corncob residues with sulfuric acid being utilized as the

catalysis to produce levulinic acid at high concentrations is proposed. The mass

concentration of the levulinic acid increased with growing the times of feeding and a

107.93 g/L of the levulinic acid can be reached at the 7th hydrolysis. However, the

T
yield of levulinic acid reduces gradually during the fed-batch process. To explore the

IP
phenomenon, the reaction mechanism of cellulose acid hydrolysis was experimentally

R
studied. It can be deduced that the reduction of the levulinic acid yield was caused by

SC
the polymerization of the soluble humin analogues and the 5-hydroxymethyl furfural

(5-HMF) as well as the glucose.

U
N
Keywords: Fed-batch acid hydrolysis; Levulinic acid; Corncob; Mechanism of
A
M

humins formation.

1. INTRODUCTION
ED

Biomass resources are known as low-cost, renewable, extensive and

PT

environment-friendly, and the utilization of which to produce correlative chemicals is

of great significance to the sustainable development [1]. For example, an important

E
CC

platform chemical, the levulinic acid, which is predominantly adopted to produce

various bio-chemicals such as solvents, resins, plasticisers, polymers, herbicides,

A

pharmaceuticals, flavoring agents and biofuels via esterification, substitution and

redox etc., can be obtained from cellulose-rich biomass resources [2-4].

2
 At the end of the last century, Biofine company used a two-step hydrolysis

method with dilute acids and carbohydrate-containing materials, e.g., the slurry of

paper material, to produce the levulinic acid [5]. With over 70 % yield of the levulinic

acid, they started a small-scale industrial production. In addition to paper slurry,

levulinic acid can be obtained by the hydrolysis of other biomass materials such as

T
sorghum [6], water hyacinth [7], bagasse [8], rice husks [9], wheat straw [10] and

IP
corn stalk [11], etc. Among various bioresource, the corncob, with an annual output of

R
approx. 50 million tons in China [12], has been widely used in producing furfural [13,

SC
14], xylose [15] and xylitol [16]. The corncob residue contains large amounts of

U
cellulose, which can be recycled to acquire downstream chemicals, such as the
N
levulinic acid.
A
M

In the process of producing the levulinic acid from biomass resources, the acid

catalyst is required during the hydrolysis reaction [17]. Acid catalysts, such as solid
ED

acid [18], inorganic acid [19, 20], organic acid [21], and acidic functionalized ionic
PT

liquids [22], are usually considered. Among them, inorganic acids are the most

commonly used catalyst in producing the levulinic acid since they are cost effective.
E

However, if large amounts of acid were utilized in hydrolysis processes, the recycling
CC

of acid residuals after reaction would be difficult, resulting in significant waste of

A

resources and environmental pollution. In addition, the levulinic acid, after being

produced, has to be concentrated for the subsequent separation and purification

processes. The more water contents exist in the hydrolysate, the more energy has to be

consumed in the concentration process. It is not conducive with respect to energy

3
saving. Aiming at this problem, Guo and co-workers [23] proposed an improved

hydrolysis technology, in which the hemicellulose in corncob was hydrolyzed to

produce high concentrations of the xylose production. However, technology refers to

the hydrolysis of the cellulose to produce high concentrations of the levulinic acid has

rarely been reported. Therefore, it is of great significance to develop new cellulose

T
hydrolysis processes by consuming small quantity of acid to produce high

IP
concentrations of the levulinic acid.

R
SC
As an important topic in the hydrolysis reaction, the mechanism of acid

hydrolysis of the cellulose in producing the levulinic acid has been extensively

U
studied. In the hydrolysis reaction, the cellulose is firstly depolymerized to the
N
glucose, then the glucose is converted to the 5-HMF. The 5-HMF can be further
A
M

transformed into equimolar amounts of the levulinic acid and the formic acid [19, 24].

However, for the by-product humins, the exact transformation mechanism has not
ED

been fully understood. Different opinions on the mechanism of humins transformation

PT

during the cellulose hydrolysis are still existing. Girusuta and co-workers [19]

believed that humins can be transformed from cellulose, glucose, and 5-HMF in their
E

kinetic study, respectively. Besides, majority of previous kinetics [17, 20, 25]
CC

assumed that humins may originate from self-polymerization of the glucose and self-
A

polymerization of the 5-HMF, respectively. Kupiainen and coworkers [26] proposed

that glucose would first be converted into an intermediate and followed by the 5-

HMF, and the humins mainly came from the intermediate and the 5-HMF. Girisuta et

al. and Chang et al. assumed that humins were produced directly by the conversion of
4
the glucose [8, 27]. Yan et al. [28] believed that humins were derived from the 5-

HMF, while the glucose would produce other undesired products instead of humins.

In our previous work [29], a kinetic study was conducted for the hydrolysis of the

corncob, in which we assumed that the humins were mainly derived from the 5-HMF,

and a good fitting effect has been achieved. However, direct evidences are still needed

T
to support these assumptions with respect to humins transformation in all above

IP
kinetics. In addition to the afore-mentioned kinetic studies, a few research groups

R
have also developed mechanisms for the formation of humins by-products. A typical

SC
mechanism was proposed by Sumerskii et al. [30], which involves a polycondensation

U
pathway of humins, leading to a network of furan rings that is linked by ether or
N
acetal bonds. Patil et al.[31, 32] put forward a pathway for the humins formation from
A

the 5-HMF and the cellulose. They proposed that humins were mainly derived from
M

the aldol addition/condensation of the 2,5-dioxo-6-hydroxyhexanal, which was

ED

converted from the 5-HMF with aldehydes and ketones. However, direct conversion

of the cellulose and the glucose to humins was not obvious. Van and co-workers [33]
PT

revealed that humins were formed by a dehydration pathway, in which the furanic
E

structure reacted with alcohol, acid, ketone, and aldehyde functional groups. As a
CC

result, they believe that humins were mainly derived from the 5-HMF. Mechanisms

proposed by the afore-mentioned studies were predominately based on

A

characterization analysis of the humins structure via instruments such as infrared,

SEM, NMR spectroscopy and pyrolysis/GC-MS etc., rather than direct experimental

analysis which is based on the concentration variation of each component throughout

5
the experiment and/or the experimental phenomena in different hydrolysis

experiments. The latter is considered to be a supplementary evidence which helps to

better understand the hydrolysis reaction mechanism from a new perspective.

For the comprehensive utilization of the corncob, in the present work, the

corncob, in which the hemicellulose have been removed, is adopted as the raw

T
IP
material to study the proposed new fed-batch process in producing the levulinic acid

R
at high concentration. In addition, the effects of other impurities in the corncob on the

SC
levulinic acid yield are investigated. Furthermore, reaction mechanisms of the overall

hydrolysis process are studied via direct experimental analysis. Combined with the

U
mechanisms, the reasons that affect the yield of levulinic acid in the proposed
N
technology are discussed.
A
M

2. MATERIAL AND METHODS

ED

2.1 Materials

The following materials were selected for this study: corncobs (30-mesh, Local
PT

farm, Qingdao, China), glucose (Sinopharm chemical, Shanghai, China), levulinic

E

acid (Aladdin, Shanghai, China), sulfuric acid (98 wt.%, Sinopharm chemical,
CC

Shanghai, China), deionized water, microcrystalline cellulose (Macklin, Shanghai,

A

China), and 5-HMF (Aladdin, Shanghai, China).

2.2 Experimental procedure

Corncob pretreatment: The corncobs were processed by a 3 wt. % sulfuric acid

solution at a temperature of 373.15 K and a solid-to-liquid ratio of 1:10 for 3 h. The

6
purpose of this pretreatment procedure was to simulate the operating conditions of the

xylose industries, and to remove the hemicellulose from corncobs. Later, the

hydrolysate was filtered, and the corncob residue was washed for three times with

sufficient deionized water. Finally, the corncob residue was dried in a vacuum oven

(DZF-6050, Hecheng Instrument Manufacturing Inc., Shanghai, China).

T
IP
The fed-batch acid hydrolysis experiment: The experiment was carried out in

R
a 500 ml reaction vessel (as shown in Fig. 1). The inner wall of the reactor which

SC
contacted with raw materials was made by Hastelloy (C276) for preventing corrosion

under acidic conditions. Corncob residue of 30g (or the microcrystalline cellulose that

U
containing an identical amount of the cellulose in corncob residue) and 300ml
N
0.5mol/L sulfuric acid solution were put into the reaction vessel. The reaction was
A
M

maintained at 453.15 K for 50 min and the kettle impeller speed was set to 400 rpm.

Reaction under above conditions could ensure the cellulose to be fully converted, as
ED

shown in Fig. 2. The yield of glucose was nearly zero when the reaction went to 45
PT

minutes, indicating that the cellulose could not be converted to the glucose any more.

After the reaction, the filter residue was filtered off and the volume of the filtrate was
E

measured. The same volume of 0.5 mol/L sulfuric acid solution as the hydrolysate
CC

adsorbed by the residue was added for washing. Later, the washing solution was
A

added into the filtrate to recover the reaction solution to 300 ml for the next

hydrolysis reaction. Continue to hydrolyze new corncob residue (or microcrystalline

cellulose) using this hydrolysate while maintaining the solid-liquid ratio at 1:10.

Reaction conditions were kept the same as those for the first time to ensure the
7
cellulose to be completely converted. This process was referred to as the second batch

hydrolysis. Based on this method, the hydrolysate was continuous used to hydrolyze

the corncob residue. The concentration of the levulinic acid after each hydrolysis was

determined by chromatographic systems.

To evaluate the mechanism of humins formation in the fed-batch hydrolysis

T
IP
process, a number of verification experiments were performed.

R
Verification experiment 1: Hydrolysate filtered by a 0.45 μm filter (Jinteng,

SC
Tianjin, China) of 500 ml containing 63.18 g/L levulinic acid and a corresponding

amount of soluble humin analogues (soluble humin analogues are complex mixtures,
U
N
the contents of which are difficult to measure and quantify. In this paper, the amount
A
of soluble humin analogues in the hydrolysate was determined corresponding to the
M

amount of levulinic acid and the times of hydrolysis.) was divided into two equal

parts: Solution 1 was utilized for the next reaction immediately without treatment;
ED

Solution 2 was placed at room temperature for 15 days and filtered by a 0.45 μm filter
PT

thereafter. Two copies of 15 g celluloses were added into two solutions for the

hydrolysis reaction, respectively. The reaction was maintained at 453.15 K for 50

E
CC

minutes and the speed of kettle impeller was set to 400 rpm. After the reaction, the

hydrolysates were filtered again, and the components in the hydrolysate were
A

determined by chromatographic systems. Took another 500 ml hydrolysate containing

61.58 g/L levulinic acid and a corresponding amount of soluble humin analogues and

divided the solution into two parts, i.e., solution 3 and solution 4, and repeated the

above experimental step.

8
 Verification experiment 2: Two hydrolysates containing 38.76 g/L (hydrolysate

1) and 127.43 g/L (hydrolysate 2) of the levulinic acid, and the corresponding

amounts of soluble humin analogues were heated up to 453.15 K, respectively. This

temperature was maintained for 100 min and the mixing speed was kept at 400 rpm.

The concentrations of the levulinic acid in hydrolysate 1 and 2 were measured again

T
after the heating reaction.

R IP
Verification experiment 3: A 400 ml 0.5 mol/L sulfuric acid solution, and a 400

SC
ml hydrolysates solution with 73 g/L levulinic acid together with the corresponding

amounts of soluble humin analogues were chosen for the hydrolysis reaction. Two

U
copies of 24 g microcrystalline cellulose were added into above two solutions,
N
respectively. The temperature of the reactor was rapidly raised to 423.15 K, allowing
A
M

the reaction to be processed. The time dependent concentration of the glucose can be

observed by sampling at each time point. Samples at each reaction time point were
ED

taken out through a sampling tube. To clean the sampling tube before and after
PT

sampling, high-pressure nitrogen gas was injected into the reactor through the

sampling tube. Then, samples were filtered through the 0.45 μm syringe filter. After
E

being diluted, the components in the samples were determined by chromatographic

CC

systems.
A

Verification experiment 4: A 450 ml 0.5 mol/L pure sulfuric acid solution, a

450 ml levulinic acid solution (containing 50 g/L LA and 0.5 mol/L sulfuric acid) and

a 450 ml hydrolysate solution (containing 47 g/L LA, 0.5 mol/L sulfuric acid and a

corresponding amount of soluble humin analogues) were chosen, whilst three copies
9
of 36 g dextrose monohydrate were added into above three solutions, respectively.

The temperature of the reactor was rapidly raised to 423.15 K to ensure the reaction to

be processed. The concentration of the component in each sample at each time point

was measured afterwards.

Conditions of four verification experiments were summarized, as shown in Table

T
IP
1.

R
2.3 Analytical method

SC
The glucose, the 5-HMF, and the LA contents in solution were analyzed by

U
HPLC (L-2000, Hitachi Limited, Tokyo, Japan) with a refractive index detector and a
N
Aminex HPX-87H column (300 mm × 7.8 mm, Bio-Rad, Hercules, USA). The
A
mobile phase was aqueous solution of the sulfuric acid (0.005 mol/L), and the flow
M

rate was 0.55 ml/min. The temperature of the column was set at 333.15 K, and the
ED

temperature of the refractive index detector was set at 311.15 K. A standard curve was

plotted by using the chromatographic pure chemical, and the concentration of each
PT

component in the samples was measured thereafter by standard curves.

E

3. RESULT AND DISCUSSION

CC

3.1 Component of the corncob residue

A

The analysis of the corncob residue was based on the Laboratory Analytical

Procedure developed by the National Renewable Energy Laboratory [34-36]. The

components in corncob residue (after pretreatment) are given in Table 2.

10
3.2 Study of the LA production in the fed-batch acid hydrolysis process

Sulfuric acid aqueous solution of 0.5 mol/L was used to hydrolyze the corncob

residue at 453.15 K. The concentration of the levulinic acid in solution after the

proposed fed-batch hydrolysis is listed in Table 3. The concentration of the levulinic

acid increased with the times of the corncob residue hydrolysate usage. Meanwhile,

T
IP
total amount of the levulinic acid increased correspondingly, resulting in

R
accumulations of the levulinic acid in the hydrolysate. After the first hydrolysis

SC
reaction, total amount of the levulinic acid in solution was 7.09 g. The yield of the

levulinic acid Y (%) during the fed-batch hydrolysis can be calculated by using Eq.

(1): U
N
A
Y = (𝑚𝑖 ⁄𝑛) × 100 % (1)
M

where mi denotes the increment of the levulinic acid at the ith feeding, with i the times
ED

of feeding; and n is the theoretical yield of the levulinic acid (Per molar of cellulose

monomer could transform into equimolar amounts of levulinic acid and formic acid.
PT

Assuming that there is no side reaction in this process, all cellulose was converted to
E

levulinic acid and formic acid, and no humins was formed), with n = 30 g × 0.607 ×
CC

(116 / 162) = 13.04 g in this case.

A

As can be seen from Table 3 and Fig. 3, the concentration of the levulinic acid

reached to 107.93 g/L after the 7th feeding, which effectively reduces the amount of

acid used in the reaction. As a result, it can effectively reduce the energy consumption

for the subsequent separation and purification processes in industrial production. In

11
addition, the remaining lignin and solid humins residue can be adopted for yielding

high-value products such as activated carbon, thus achieving comprehensive

utilization of the corncob.

A comparison of LA concentration in this work was made with other references,

as shown in Table 4. Clearly, the proposed fed-batch process in this work exhibits

T
IP
much higher concentration of the levulinic acid than others. However, as the times of

R
reactions increased, the yield of the levulinic acid presented a decreasing tendency. As

SC
shown in Table 3, the yield of the levulinic acid at the 5th hydrolysis was 26.84%,

nearly half of the levulinic acid yield comparing to the 1st hydrolysis. When the

U
reaction progressed to the 7th, the yield was only 19.71%. More times of the
N
hydrolysis reaction were considered to be meaningless since the accumulation of the
A
M

levulinic acid was not significant at the 7th feeding. Therefore, the reaction was

stopped at the 7th feeding and we expected this as an adequate treatment cycle for the
ED

fed-batch acid hydrolysis process. The yield at the 4th feeding was 70.68% of the 1st,

while the yield at the 5th feeding was only 49.36% of the 1st. The yield decreased
PT

significantly between 4th and 5th feeding process. With respect to the yield of levulinic
E

acid, a four times feeding is considered to be reasonable, as it not only obtains

CC

relatively high concentrations of the levulinic acid, but also effectively reduces the
A

operating costs.

3.3 Effect of other impurities in corncob residue on the yield of levulinic acid

12
 The corncob residue after producing the xylose or the furfural contains much

lignin, a small amount of inorganic salts and some ashes. The microcrystalline

cellulose containing an equal amount of cellulose in the corncob residue was used to

explore the effect of other impurities in the corncob residue on the yield of the

levulinic acid in the hydrolysis reaction. As shown in Fig. 4, by using the

T
microcrystalline cellulose and the corncob residue as raw materials respectively, the

IP
increment of the levulinic acid concentration was roughly consistent during the fed-

R
batch hydrolysis process. It indicates that lignin, trace inorganic salts and ashes in

SC
corncob residue have almost no effect on the yield of the levulinic acid. Also, it

U
implies that high molecular cellulose was depolymerized into low molecular
N
cellulose, and low molecular cellulose was disaggregated into monosaccharides in the
A

process of hydrolysis. At a sufficiently low solid-liquid ratio, the chain length of

M

cellulose may have little effect on the yield of the levulinic acid.
ED

3.4 Mechanisms of the humins formation and the yield of LA in the fed-batch
acid hydrolysis process
PT

After hydrolysis of cellulose in corncob residue, a red-brown solution and a

hydrolyzed residue were obtained. The sum of the solid humin yield and products
E
CC

detected by HPLC analysis did not follow the mass balance in each feeding reaction,

indicating that water-soluble oligomers were generated. The red-brown solution

A

mainly contains the levulinic acid, the formic acid and water-soluble oligomers. The

water-soluble oligomers were named as soluble humin analogues in this paper. As

increasing the times of the hydrolysis reaction, the amount of soluble humin

analogues in the hydrolysate increased gradually. As can be seen from Table 3, the
13
hydrolysis residue increases successively with times of the hydrolysis reaction,

indicating that the amount of solid humins has increased gradually in hydrolysis

residue and more ingredients involve in the production of the solid humins. During

the verification experiment 1, it was found that when hydrolysate was filtered through

a 0.45μm filter and was placed at room temperature for 15 days, the clear solution

T
became turbid and a small amount of precipitation was formed. Later, the

IP
precipitation was filtered and detected by infrared spectroscopy. As shown in Fig. 5.

R
the characteristic peaks of the infrared spectrum lines of the precipitation and the solid

SC
humins residue are nearly identical, implying that the precipitation and solid humins

U
are the same substance. However, the concentrations of the levulinic acid and the
N
formic acid were unchanged. The soluble humin analogues may self-polymerize or
A

flocculate to precipitate a small amount of solid humins. Besides, the soluble humin
M

analogues may contain some reactive sites which participate in self-polymerization or

ED

other reactions. As a result, the reduced yield of the levulinic acid may also be related

to the existing soluble humin analogues in solution.

PT

A series of experiments were designed to verify the above assumptions

E

(verification experiment 1). From Table 5, the increase of the levulinic acid
CC

concentration in solutions 1, 2, 3 and 4 were 14.36 g/L, 16.41 g/L, 14.98 g/L and
A

17.03 g/L, respectively. After treatment, the soluble humin analogues were self-

polymerized or flocculated so that small amounts of solid humins were formed, and

the polymerization of soluble humin analogues with other substances in solution was

14
reduced. As a result, the yields of the levulinic acid after treatment are higher than

that of the levulinic acid without treatment.

The above experimental results showed that the soluble humin analogues reacted

with other substances in the solution, thus affecting the yield of the levulinic acid.

There are two possibilities for this situation: one is that soluble humin analogues

T
IP
directly reacted with levulinic acid, causing an reduction of the levulinic acid yield

R
when the hydrolysate was reused. Relevant experiments (Verification experiment 2)

SC
were designed to verify the above hypothesis, as shown in Table 6. After the heating

reaction, the concentrations of the levulinic acid in two hydrolysates were 37.87 g/L

U
and 126.66 g/L, respectively, only a bit lower than that of the levulinic acid (38.76
N
g/L and 127.43 g/L) before heating. The decrease of levulinic acid in two
A
M

hydrolysates was not obvious. Therefore, it can be concluded that the main reason for

the decrease of the levulinic acid yield in the fed-batch process was not caused by the
ED

reaction between levulinic acid and soluble humin analogues.

PT

Another possibility is that the soluble humin analogues were further polymerized

with the glucose and the intermediate 5-HMF to generate new humins, thus reducing
E
CC

the yield of levulinic acid when the hydrolyzate was reused. Levulinic acid was

obtained by the 5-HMF, while 5-HMF was derived from the glucose. If the glucose
A

and the 5-HMF polymerize with soluble humin analogues to form new humins in the

reaction, the yield of the levulinic acid will eventually be affected.

15
 In previous studies [30-33], the furan ring structure was detected in humins that

formed from cellulose, indicating that the 5-HMF, which is characterized by a furan

ring, participated in the humins forming reaction. However, for the glucose, there is

no direct evidence that glucose could be involved in humins production. Hence, the

reaction of soluble humin analogues with the glucose will be further studied in this

T
paper. Experimental process is shown in verification experiment 3 and the result is

IP
shown in Fig. 6. At the same time point, the concentration of the glucose in sulfuric

R
acid solution was higher than that of hydrolysates, indicating that the glucose in

SC
hydrolysates was consumed faster than the pure sulfuric acid solution (The catalysis

U
of organic acids in hydrolysates is very limited at high concentration of the sulfuric
N
acid solution, thus the organic acids catalytic action can be ignored). Meanwhile, as
A

shown in Fig. 7, the increment of levulinic acid in sulfuric acid solution was higher
M

than that of hydrolysate solution at the same time point. Combining Figs. 6 and 7, it
ED

indicates that glucose was involved in other reaction apart from the reaction of

glucose degradation into 5-HMF. An adequate explanation is that soluble humin

PT

analogues would polymerize with some glucose in solution, the higher the
E

concentration of soluble humin analogues in solution, the easier the soluble humin
CC

analogues polymerized with the glucose. With increasing the times of the hydrolysis

reaction, the concentration of soluble humin analogues in solution increased

A

continuously, enhancing the possibility of the polymerization of soluble humin

analogues with the glucose. However, the selectivity of the conversion of glucose to

5-hydroxymethylfurfural decreased, giving rise to a reduced yield of the levulinic

16
acid. Our previous kinetic study [29] suggested that humins were only derived from

the 5-hydroxymethylfurfural, while good fitting results have been achieved. The

reason may be that when the solid-liquid ratio was low enough, the concentration of

soluble humin analogues in solution can be neglected. Humins mainly came from

mutual polymerization of the 5-hydroxymethylfurfural, and a very small amount of

T
humins were less likely to react with the glucose. Therefore, based on our previous

IP
model, good fitting effect can be obtained. However, when the concentration of

R
soluble humin analogues in solution increased with feeding times, the model was no

SC
longer applicable. Since all previous kinetic studies cannot explain this situation, a

new model is therefore required.

U
N
The above experiments showed that soluble humin analogues did not directly
A
M

react with the levulinic acid, whereas it would polymerize with glucose and

intermediate 5-HMF to generate new humins. However, the problem that whether
ED

some humins were derived from the polymerization between levulinic acid and
PT

glucose has not been known yet. For this a further study was performed, the specific

experimental process was mentioned in verification experiment 4. As shown in Fig. 8,

E

real-time concentrations of the glucose in pure sulfuric acid solution and levulinic
CC

acid solution were almost the same, and the degradation of glucose was not
A

accelerated in solution with levulinic acid. It indicates that the levulinic acid did not

directly react with the glucose. However, the concentration of the glucose in

hydrolysates at each time point was slightly lower than that of the prior, implying that

the glucose was involved in other reactions aside from being converted to the 5-HMF.
17
As can be seen from Fig. 9, the increase of both the concentration at each time point

and the final yield of levulinic acid in former two solutions were nearly identical. As

for the hydrolysate, the situations were lower than that of the former two solutions,

which further showed that part of the glucose and the 5-HMF were involved in other

reactions.

T
IP
Based on above validation experiments and, it (shown in Fig. 10) can be

R
concluded that the yield of levulinic acid in the fed-batch process reduced due to the

SC
accumulation of soluble humin analogues with increasing times of the hydrolysis.

Soluble humin analogues can easily polymerize with glucose and 5-HMF, resulting in

U
decreased selectivity of glucose to 5-HMF and 5-HMF to levulinic acid, and thus the
N
reduced yield of the levulinic acid. Soluble humin analogues did not polymerize
A
M

directly with the levulinic acid, while glucose did not directly react with the levulinic

acid. It can be deduced from the reaction mechanism of the cellulose acid hydrolysis
ED

that by reducing the amount of soluble humin analogues in solution before each
PT

feeding, the yield of levulinic acid is expected to be improved.

4. CONCLUSIONS
E

In this work, a new fed-batch process for producing levulinic acid by reusing
CC

hydrolyzate to hydrolyze corncob residue was proposed. After the first hydrolysis, the
A

mass concentration of the levulinic acid in solution was 23.65 g/L. After feeding to 2,

3, 4, 5, 6, 7 times, the mass concentration of the levulinic acid reached 42.82, 59.90,

76.59, 88.26, 99.38, 107.93 g/L, respectively. The new process could effectively

18
reduce the amount of acid used in the reaction, allowing a reduction of the energy

consumption for the subsequent separation and purification processes.

The reaction mechanism of cellulose acid hydrolysis to produce levulinic acid

and by-product humins was discussed. It was found that the decrease of levulinic acid

yield during the fed-batch process was caused by the accumulation of soluble humin

T
IP
analogues with increasing times of the hydrolysis. The accumulated soluble humin

R
analogues in solution can easily polymerize with glucose and 5-HMF, resulting in a

SC
decrease of the selectivity to the levulinic acid.

ACKNOWLEDGEMENT

U
This work was financially supported by the Nature Science Foundation of China
N
A
21376231 and the Nature Science Foundation of China 21176264.
M

REFERENCES
ED

[1] D. W. Rackemann, W. O. Doherty, The conversion of lignocellulosics to levulinic

acid. Biofpr. 5 (2015) 198-214.

PT

[2] J. Harris, Acid hydrolysis and dehydration reaction for utilizing plant
E

carbohydrates. Appl. Polym. Symp. 28 (1975) 131-144.

CC

[3] J. J. Bozell, L. Moens, D. C. Elliott, Y. Wang, G. G. Neuenscwander, S. W.

A

Fitzpatrick, R. J. Bilski, J. L. Jarnefeld, Production of levulinic acid and use as a

platform chemical for derived products. Resour. Conserv. Recy. 28 (2000) 227-239.

19
[4] D. J. Hayes, F. Steve, M. H. B. Hayes, J. R. H. Ross, The Biofine Process-

Production of Levulinic Acid, Furfural, and Formic Acid from Lignocellulosic

Feedstocks. Wiley-VCH Verlag GmbH, 2008, pp. 139-164.

[5] S. W. Fitzpatrick, Production of levulinic acid from carbohydrate-containing

materials. US Patent, 1997, 5608105.

T
IP
[6] Q. Fang, M. A. Hanna, Experimental studies for levulinic acid production from

R
whole kernel grain sorghum. Bioresource Technol. 81 (2002) 187-92.

SC
[7] B. Girisuta, B. Danon, R. Manurung, L. P. Janssen, H. J. Heeres, Experimental

U
and kinetic modelling studies on the acid-catalysed hydrolysis of the water hyacinth
N
plant to levulinic acid. Bioresource Technol. 99 (2008) 8367-75.
A

[8] B. Girisuta, K. Dussan, D. Haverty, J. J. Leahy, M. H. B. Hayes, A kinetic study

M

of acid catalysed hydrolysis of sugar cane bagasse to levulinic acid. Chem. Eng. J.
ED

217 (2013) 61-70.

PT

[9] O. P. Fleig, E. S. Lopes, E. C. Rivera, R. M. Filho, L. P. Tovar, Concept of rice

husk biorefining for levulinic acid production integrating three steps: Multi-response
E

optimization, new perceptions and limitations. Process Biochem. 65 (2018) 146-156.

CC

[10] C. Chang, P. Cen, X. Ma, Levulinic acid production from wheat straw.
A

Bioresource Technol. 98 (2007) 1448-1453.

20
[11] Z. Zhi, N. Li, Y. Qiao, X. Zheng, H. Wang, X. Lu, Kinetic study of levulinic acid

production from corn stalk at relatively high temperature using FeCl3 as catalyst: A

simplified model evaluated. Ind. Crop. Prod. 76 (2015) 672-680.

[12] H. Wang, X. Zhang, D. Wang, C. Cui, C. Gao, L. Wang, Y. Wang, Y. Bi,

Estimation and utilization of corn cob resources in china. Chinese Journal of

T
IP
Agricultural Resources and Regional Planning. 37 (2016) 1-8.

R
[13] C. D. Blasi, C. Branca, A. Galgano, Biomass Screening for the Production of

SC
Furfural via Thermal Decomposition. Ind. Eng. Chem. Res. 49 (2010) 2658-2671.

U
[14] C. Branca, C. D. Blasi, A. Galgano, Catalyst Screening for the Production of
N
Furfural from Corncob Pyrolysis. Energ. Fuel. 26 (2012) 1520-1530.
A

[15] H. J. Zhang, X. G. Fan, X. L. Qiu, Q. X. Zhang, W. Y. Wang, S. X. Li, L. H.

M

Deng, M. A. Koffas, D. S. Wei, Q. P. Yuan, A novel cleaning process for industrial

ED

production of xylose in pilot scale from corncob by using screw-steam-explosive

extruder. Bioproc. Biosyst. Eng. 37 (2014) 2425-36.

PT

[16] T. L. Albuquerque, I. J. Silva Jr., J. R. Macedo, M. V. P. Rocha,

E

Biotechnological production of xylitol from lignocellulosic wastes: A review. Process

CC

Biochem. 49 (2014) 1779-1789.

A

[17] C. Chang, X. Ma, P. Cen, Kinetic Studies on Wheat Straw Hydrolysis to

Levulinic Acid. Chinese J. Chem. Eng. 17 (2009) 835-839.

21
[18] Y. Zuo, Y. Zhang, Y. Fu, Catalytic Conversion of Cellulose into Levulinic Acid

by a Sulfonated Chloromethyl Polystyrene Solid Acid Catalyst. Chemcatchem. 6

(2014) 753-757.

[19] B. Girisuta, L. P. B. M. Janssen, H. J. Heeres, Kinetic Study on the Acid-

Catalyzed Hydrolysis of Cellulose to Levulinic Acid. Ind. Eng. Chem. Res. 46 (2007)

T
IP
1696-1708.

R
[20] J. Shen, C. E. Wyman, Hydrochloric acid-catalyzed levulinic acid formation

SC
from cellulose: data and kinetic model to maximize yields. Aiche J. 58 (2011) 236-

246.
U
N
[21] T. Retsina, V. Pylkkanen, K. Nelson, R. O'Connor, PROCESSES AND
A
APPARATUS FOR PRODUCING FURFURAL, LEVULINIC ACID, AND OTHER
M

SUGAR-DERIVED PRODUCTS FROM BIOMASS. US Patent, 2014, CA2933847.

ED

[22] N. A. S. Ramli, N. A. S. Amin, A new functionalized ionic liquid for efficient

glucose conversion to 5-hydroxymethyl furfural and levulinic acid. J. Mol. Catal. A-

PT

Chem. 407 (2015) 113-121.

E

[23] L. Guo, Y. Hu, L. Wu, C. Liang, W. Zhang, The green hydrolysis technology of
CC

hemicellulose in corncob by the repeated use of hydrolysate. Chinese J. Chem. Eng.

A

16 (2017) 191-195.

[24] B. Girisuta, L. P. B. M. Janssen, H. J. Heeres, Green Chemicals: A Kinetic Study

on the Conversion of Glucose to Levulinic Acid. Chem. Eng. Res. Des. 84 (2006)

339-349.
22
[25] N. A. S. Ramli, N. A. S. Amin, Kinetic study of glucose conversion to levulinic

acid over Fe/HY zeolite catalyst. Chem. Eng. J. 283 (2016) 150-159.

[26] L. Kupiainen, J. Ahola, J. Tanskanen, Kinetics of glucose decomposition in

formic acid. Chem. Eng. Res. Des. 89 (2011) 2706-2713.

T
[27] C. Chang, X. Ma, P. Cen, Kinetics of Levulinic Acid Formation from Glucose

IP
Decomposition at High Temperature. Chinese J. Chem. Eng. 14 (2006) 708-712.

R
[28] L. Yan, A. A. Greenwood, A. Hossain, B. Yang, A comprehensive mechanistic

SC
kinetic model for dilute acid hydrolysis of switchgrass cellulose to glucose, 5-HMF

and levulinic acid. RSC Adv. 4 (2014) 23492-23504.

U
N
[29] C. Liang, Y. Hu, L. Guo, L. Wu, W. Zhang, Kinetic Study of Acid Hydrolysis of
A

Corncobs to Levulinic Acid. Bioresources. 12 (2017) 4049-4061.

M

[30] I. V. Sumerskii, S. M. Krutov, M. Y. Zarubin, Humin-like substances formed

ED

under the conditions of industrial hydrolysis of wood. Russ. J. Appl. Chem. 83 (2010)
PT

320-327.

[31] S. K. R. Patil, C. R. F. Lund, Formation and Growth of Humins via Aldol

E
CC

Addition and Condensation during Acid-Catalyzed Conversion of 5-

Hydroxymethylfurfural. Energ. Fuel. 25 (2011) 4745-4755.

A

[32] S. K. R. Patil, J. Heltzel, C. R. F. Lund, Comparison of Structural Features of

Humins Formed Catalytically from Glucose, Fructose, and 5-

Hydroxymethylfurfuraldehyde. Energ. Fuel. 26 (2012) 5281-5293.

23
[33] Z. I. Van, Y. Wang, C. B. Rasrendra, E. R. Van Eck, P. C. Bruijnincx, H. J.

Heeres, B. M. Weckhuysen, Formation, molecular structure, and morphology of

humins in biomass conversion: influence of feedstock and processing conditions.

Chemsuschem. 6 (2013) 1745-1758.

[34] A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, D. Templeton, D. Crocker,

T
IP
Determination of Structural Carbohydrates and Lignin in Biomass, National

R
Renewable Energy Laboratory, NREL/TP-510-42618, 2010.

SC
[35] A. Sluiter, R. Ruiz, C. Scarlata, J. Sluiter, D. Templeton, Determination of

Extractives in Biomass, National Renewable Energy Laboratory, NREL/TP-510-

U
N
42619, 2008.
A
[36] A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, and D. Templeton,
M

Determination of Ash in Biomass, National Renewable Energy Laboratory,

ED

NREL/TP-510-42622, 2008.

TABLES:
PT

Table 1. Comparison of the verification experiments.

E

Verification Solution Treatment method Purpose

experiment
CC

Filtered
1 hydrolysate 1 No treatment and add
cellulose for hydrolysis To verify whether the
Filtered
decrease of yield is related to
A

hydrolysate 3
soluble humin analogues
Filtered Placed for 15 days and
hydrolysate 2 filtered,
Filtered then add cellulose for
hydrolysate 4 hydrolysis
Hydrolysate 1 To verify whether the
2 Hydrolysate 2 Heat at 453.15 k for levulinic acid could react with
100 min soluble humin analogues
24
 Sulfuric acid Verify that whether the
3 solution Add cellulose for glucose and 5-HMF could
Hydrolysates hydrolysis polymerize with soluble
solution humin analogues
Sulfuric acid
solution Add dextrose To verify whether the glucose
4 Levulinic acid monohydrate for and 5-HMF could polymerize
solution hydrolysis with levulinic to produce
Hydrolysate some humins.
solution

T
IP
Table 2. Analyses of the corncobs residue.

Compound Value (wt.%)

R
Cellulose 60.70
Lignin 31.40

SC
Hemicellulose 2.70
Salt and ash 3.00
Others 2.20

U
N
Table 3. Concentration of the levulinic acid in solution after fed-batch hydrolysis.
A

Times of the Filtrate + Volume lost Levulinic acid Total Yield The
M

hydrolysate washing due to concentration amount of (%) amount

usage solution adsorption (g/L) levulinic of
(ml) (ml) acid residue
ED

(g) (g)
1 285+15 15 23.65 7.09 54.37 12.49
2 285+17 17 42.82 12.85 44.18 13.15
3 281+19 19 59.90 17.97 39.27 13.63
PT

4 280+20 20 76.59 22.98 38.43 13.85

5 278+22 22 88.26 26.48 26.84 14.69
6 273+27 27 99.38 29.81 25.54 14.97
E

7 275+25 25 107.93 32.38 19.71 15.13

CC

Table 4. Comparison of the LA concentrations.

Biomass Cellulose Catalyst Solid LA LA References

A

content and to concentration concentratio

(%) concentratio liquid after 1st n after
n ratio reaction feeding 7th
(mol/L) (g/L) time
(g/L)
Sugar 43.30 H2SO4 0.55 1:9 19.50 - Girisuta et
cane al. [8]
bagasse
25
 Wheat 40.40 H2SO4 0.30 1:16 10.40 - Chang et al.
straw [10]
Corn stalk 41.50 FeCl3 0.50 1:9 16.00 - Zhi et al.
[11]
Corncob 60.70 H2SO4 0.50 1:10 23.70 107.90 This work
residue

Table 5. The influence of soluble humin analogues on LA concentration in the

T
hydrolysis reaction.

IP
Solution LA Processing The amount LA LA

R
concentration method of concentration concentration
before reaction precipitation after reaction increase

SC
(g/L) produced (g/L) (g/L)
(g/L)
1 63.18 No 0 77.54 14.36
treatment
2 63.18 Placed for
15 days
U
0.08g 79.59 16.41
N
3 61.58 No 0 76.57 14.99
treatment
A
4 61.58 Placed for 0.07g 78.61 17.03
15 days
M

Table 6. Variation of the levulinic acid concentration before and after heating.
ED

Hydrolysate Heating Heating time LA concentration LA concentration

temperature (min) before heating after heating
(K) (g/L) (g/L)
PT

1 453.15 100 38.76 37.87

2 453.15 100 127.43 126.66
E
CC
A

26
 A
CC

FIGURES：
EPT
ED

27
M
A
N
U
SC
RIP
T
 T
R IP
SC
Figure 1. Schematic diagram of the reactor.

U
N
A
M
ED
E PT
CC

Figure 2. Variation of each component concentration vs. time at 453.15 K and H2SO4
= 0.5 mol/L.
A

28
 T
IP
Figure 3. Variation of the levulinic acid concentration and the yield with feeding

R
times of the hydrolysis.

SC
U
N
A
M
ED
E PT

Figure 4. Comparison of the LA concentration by corncob residue and the

CC

microcrystalline cellulose in the hydrolysis reaction.

A

29
 T
R IP
SC
Figure 5. IR spectra of humin solid residue and precipitated solids in solution.

U
N
A
M
ED
E PT
CC

Figure 6. Variation of the glucose concentration vs. reaction time in different

solutions.
A

30
 T
R IP
SC
Figure 7. Increment of the levulinic acid vs. reaction time in sulfuric acid solution and
hydrolysate solution, respectively.

U
N
A
M
ED
E PT
CC

Figure 8. Variation of glucose concentration vs. reaction time in three different

A

solutions.

31
 T
R IP
SC
Figure 9. Variation of the levulinic acid concentration vs. reaction time in three
different solutions

U
N
A
M
ED
E PT
CC

Figure 10. Path way of glucose decomposition to LA and humins

A

32