**Fractionation of Cellulose-Rich Products from an Empty Fruit Bunch (EFB) by Means of Steam Explosion Followed by Organosolv Treatment**

#### **Jae Hoon Lee <sup>1</sup> , Muhammad Ajaz Ahmed <sup>2</sup> , In-Gyu Choi <sup>1</sup> and Joon Weon Choi 2,\***


Received: 12 December 2019; Accepted: 16 January 2020; Published: 24 January 2020

**Abstract:** In this study an empty fruit bunch (EFB) was subjected to a two-step pretreatment to defragment cellulose-rich fractions as well as lignin polymers from its cell walls. First pretreatment: acid-catalyzed steam explosion (ACSE) pretreatment of EFB was conducted under the temperature range of 180–220 ◦C and residence time of 5–20 min. The ACSE-treated EFB was further placed into the reactor containing 50% aq. ethanol and NaOH as a catalyst and heated at a temperature of 160 ◦C for 120 min for the second pretreatment: alkali-catalyzed organosolv treatment (ACO). The mass balance and properties of treated EFB were affected by the residence time. The lowest yield of a solid fraction was obtained when the residence time was kept at 15 min. Xylose drastically decreased, especially under the ACSE pretreatment. However, the crystallinity of cellulose increased by increasing the severity factor of the pretreatment and was 47.8% and 57% udner the most severe conditions. The organosolv lignin fractions also showed the presence of 14 major peaks via their pyrolysis-GC analysis. From here, it can be suggested that this kind of pretreatment can indeed be one potential option for lignocellulosic pretreatment.

**Keywords:** Biomass; two-step pretreatment; steam explosion; organosolv treatment; empty fruit bunch

### **1. Introduction**

Over the recent decades, the utilization of empty fruit bunch (EFB) to produce renewable energy source has been studied extensively. This, EFB typically, is a leftover of the fresh fruit bunches after their fruit harvesting. Moreover, it is the major downstream residue from the Southeast Asian palm oil industry. This industry, recently, has produced 70 million metric tons of crude palm oil (CPO) just within one production year 2017–2018 amounting their production even up to 35% of the total production of edible vegetable oils [1]. Theoretically, palm oil plantations can produce more than 100 million metric tons of EFB since 1.42–1.88 tons of EFB are generated for every one ton of CPO produced [2]. However, unfortunately, most of these EFB is dumped and pose serious environmental problems. Therefore, it is highly desirable to utilize these biomasses for sustainable energy production.

Biochemical conversion of lignocelluloses involving chemical pretreatment followed by enzymatic hydrolysis and finally anaerobic fermentation is a well-established production line for lignocellulosic bioethanol production. In this classical approach, due to an inherited complex cell wall structure, effective fractionation of major lignocellulosic biomass components (cellulose, hemicellulose, and lignin) is crucial for effective subsequent processing. Undoubtedly, there have been several pretreatment studies involving EFB to deconstruct its native web into its constituents. In this regard, steam

explosion, which uses high-pressure steam to breakdown biomass structure, has been widely adapted strategy [3]. The effects of several factors like biomass particle size, moisture, pressure, residence time, reaction temperature and catalysts on the steam explosion were consistently investigated for several decades [4]. Previously, Choi et al. used alkaline reagent (3% NaOH at 160 ◦C) to catalyze steam explosion pretreatment and dissolved out C5 sugar-based components as well as lignin components from biomass [5]. Their yields, after enzymatic hydrolysis, for glucose and xylose heightened up to 93% and 78% respectively. In another study, the steam explosion of rose gum tree (Eucalyptus grandis) was carried out under the temperature range of 200–210 ◦C and residence time of 2–5 min after impregnation with 0.087% and 0.175% (*w*/*w*) H2SO<sup>4</sup> [6]. Their results showed that 0.175% (*w*/*w*) H2SO4-impregnated chips at 210 ◦C for 2 min was the optimal condition for hemicellulose extraction. Today steam explosion is seen as one of the most widely employed and cost-effective process for lignocellulose pretreatment, yet the process still needs to be improved to realize industrialization [7]. In addition to this steam explosion strategy, another effective fractionation scheme for lignocelluloses is organosolv treatment, so-called organosolvation. This process, in principle, extracts lignin from lignocellulose with organic solvents, such as alcohols, acetone and organic peracids [8]. Previous studies have shown considerable effects of organosolv treatment (diethylene glycol) of rice straw and EFB as compared to soda pulping by Gonzalez et al. [9] keeping in view the heterogeneous yet complex cell wall structure of EFB, it is highly likely to investigate an appropriate pretreatment strategy for commercial production of biofuels [10]. Therefore, it has to be investigated the proper pretreatment process of EFB. In this study we adapted a two-step pretreatment strategy to fractionate EFB biomass into its sugar and non-sugar components. For this two-step pretreatments, EFB was first subjected to acid-catalyzed steam explosion (ACSE) followed by alkali-catalyzed organosolv treatment (ACO). Pretreatment efficacy was evaluated in terms of yields, crystallinity, and chemical composition of each solid fraction.

#### **2. Materials and Methods**

#### *2.1. Feedstock Analysis*

As raw biomass for this study, EFB (≤6% moisture) was provided by the Korea Institute of Science and Technology, in a setting of 0.5–2 mm fiber. Holocellulose, lignin, and ash content in the EFB sample were analyzed according to the National Renewable Energy Laboratory (NREL) standard procedures [11–13]. Carbon, hydrogen, and nitrogen contents were measured by using a CHNS-932 analyzer (LECO Corp., St. Joseph, MI, USA) and inorganic elements were analyzed by inductively coupled plasma emission spectroscopy (iCAP7400 Duo, Thermo Fisher Scientific, Waltham, MA, USA) [14]. Thermogravimetric analysis was performed under an inert atmosphere (50 mL/min N<sup>2</sup> flow) at a heating rate of 10 ◦C/min up to 800 ◦C with a TGA/DSC 3+ (METTLER TOLEDO, Columbus, OH, USA) [14]. All the results with the corresponding values are listed in Table 1.


**Table 1.** Chemical and thermal characteristics of empty fruit bunch.


**Table 1.** *Cont.*

<sup>1</sup> By difference.

#### *2.2. ACSE-First Pretreatment*

Initially, we screened out experimental conditions based on our initial trials and proceeded accordingly as described below for the pretreatment. The complete fractionation processes are delineated in Figure 1. Briefly, for ACSE treatment, EFB was first washed to remove inorganic elements using a method suggested by Moon et al. [15] Later on it was impregnated in distilled water for 120 min with slow stirring, and finally the feedstock was dried in an oven at 75 ◦C. It was then impregnated in dilute sulfuric acid (0.5, 1, 2 wt%) for 4 h to make it ready for steam explosion pretreatment. The effect of dilute acid concentration on impregnation was examined by the component analysis (holocellulose, lignin, and sugar) [12].

For ACSE experiments, autoclave reactors with 2 L high pressure steam were loaded first with 150 g of raw EFB biomass samples on dry basis as a control. The final temperature in the reactor was applied to reach in the interval of 180–220 ◦C and the residence time was 5–15 min. After the intended reaction time, a valve was opened to release the pressure immediately. Solid fractions (E-S) from the steam explosion was washed with distilled water, and the resulting aqueous fraction (E-L) was also collected. Then, acid impregnated EFB, with the same solids loading, was steam-treated under the temperature of 220 ◦C and 20 min of residence time, the most severe conditions that could be stably maintained with the reactor. After the reaction and cooling, solid fraction (EA-S), and aqueous fraction (EA-L) were also collected.

**Figure 1.** Schematic diagram of the two-step fractionation process of empty fruit bunches (EFB) by (acid-catalyzed) steam treatment followed by alkali-catalyzed organosolv treatment.

#### *2.3. ACO Second Pretreatment*

Organosolv treatment experiments, as a 2nd treatment for both the E-S & EA-S solid fractions obtained from 1st treatment, were performed as follows: both the solid fractions were placed, one by one, in a solvent-circulating reactor with a solvent of a 50:50% (*w*/*w*) ethanol:water mixture maintaining a (10:1 (*w*/*w*) solvent:solid ratio) along with a 2 wt% of NaOH as a catalyst. The reaction was carried out at a temperature of 160 ◦C and 120 min residence time. After the desired reaction time, cellulose-rich solid fractions were collected and labelled as E-S-CS and EA-S-CS and liquid fractions were labelled as E-S-OL and EA-S-OL corresponding to E-S and EA-S biomass samples, respectively. The organosolv lignin was precipitated from E-S-OL and EA-S-OL by acidification with a 2 M of hydrochloric acid maintaining pH = 2. The precipitated organosolv lignin was freeze-dried and collected as a powder and saved for further experiments.

#### *2.4. Analysis*

Yields of each solid fraction were calculated using the following equation:

Yield (%) = weight of a solid fraction (g)/weight of dry feedstock (g) × 100

Holocellulose, lignin, and sugar content in each solid fraction were analyzed by the same process mentioned in Section 2.1. X-ray powder diffraction patterns (XRD) were analyzed by a Bruker D8 Advance with DAVINCI using Cu Kα radiation (λ = 1.5418 Å), operated at 40 kV and 40 mA with a scan speed of 0.5 s/min in a range of 2–50 degrees (2 thetas). Chemical compounds in liquid fraction were quantified and qualified using gas chromatography-mass spectrometry systems (5975C Series GC/MSD System, Agilent Technologies, Santa Clara, CA, USA) [14]. Organosolv lignin was analyzed by a coil-type CDS Pyroprobe 5000 (CDS Analytical Inc., Oxford, PA, USA) [16].

#### **3. Results and Discussion**

#### *3.1. Mass Balance*

3.1.1. Effect of Residence Time and Temperature on Fractionation

The effects of residence time and temperature on the fractionation process were studied at various intervals (residence time 5, 10, 15 min and temperature 180, 200, 220 ◦C). Figure 2 shows the mass distribution of solid fractions obtained from EFB at various temperatures and residence time conditions.

**Figure 2.** Yields of the solid fractions from steam explosion and organosolv treatment of EFB under different temperatures and residence times.

Mass composition of the fraction was affected by residence time. At all temperature conditions, E-S from the steam explosion, and its subsequent, E-S-CS from organosolv treatment decreased with increasing residence time. For example, at 200 ◦C and in the range of 5–15 min, yields of E-S and E-S-CS decreased from 79.0 wt% and 73.8 wt% to 68.3 wt% and 61.2 wt%, respectively. The lowest yield was obtained when the residence time was kept for 15 min, regardless of reaction temperatures. The effect of temperature was not much clear as that of residence time. Under the same residence time condition, the yield of E-S and E-S-CS decreased only 3–8 wt%. The combined effect of temperature (T) and residence time (t) can be expressed as severity factor (log<sup>10</sup> (*R*0)) [*R*<sup>0</sup> = t × exp((T−100)/14.75)) [17]. The severity factor obtained ranged from 3.1 to 4.7. There was a scant negative correlation between the severity factor and the solid fraction yield, showing that increasing temperature and residence time results in the removal of compounds from solid fractions.

In general, both mechanical and chemical effects take place when an acetyl group derived from hemicellulose acts as an acid at high temperature during the steam explosion [18]. These effects play important roles in the hydrothermal degradation of holocellulose. Hemicellulose and lignin are isolated by the high temperature in the steam condition. Organosolv treatment, however, on the other hand, solubilize lignin and provide holocellulose by using organic solvent mixture and inorganic catalysts [7]. Therefore, holocellulose and lignin compositions of solid fractions were also analyzed to examine the fractionation efficiency of the process.

Holocellulose and lignin contents of E-S and E-S-CS were calculated for each experiment (Table 2). Percentage of holocellulose composition decreased with increasing severity of steam explosion. A maximum of 24.1 wt% of holocellulose (220-15) was dissolved in E-L. On the contrary, the percentage of lignin composition increased with increasing severity factor. It can be anticipated that lignin melts under a high temperature and pressure, then condensed with other compounds like extractives or ash while cooling. Solid fractions which have undergone organosolvation showed an unremarkable change

of holocellulose contents. In addition, a considerable amount of lignin content still remained in the solids. Maximum 7.5% of lignin (220-15) was solubilized in 50% ethanol mixture. Lignin fractionation efficiency of organosolv treatment from EFB was falling short of our expectation because of its low lignin removal. These results showed EFB presents a high level of recalcitrance to the fractionation process. Thus, dilute acid pretreatment to level down the EFB recalcitrance had considerable effects and the results are discussed in Section 3.1.2.


**Table 2.** Component analysis of pretreated EFB solid fractions under different temperatures and residence times.

E-S: solid fraction obtained from steam explosion treatment of EF. E-S-CS: cellulose rich solid fraction obtained from alkali catalyzed organosolv treatment of E-S fraction.

#### 3.1.2. Effect of Dilute Acid Impregnation on Fractionation

Acid impregnation has been considered to play a vital role for steam explosion treatment [4,19]. Keeping this in view, the effects of dilute acid on the fractionation process was studied with the feedstock impregnation in dilute sulfuric acid (0.5, 1, 2 wt%) for 4 h. Steam explosion process was further operated at residence time of 20 min and temperature of 220 ◦C, which showed the highest severity score so that the effect of dilute acid concentration can be compared clearly. Figure 3 shows the mass distribution of solid fractions obtained from EFB at various acid concentrations.

Mass composition of EA-S did not affect by the acid concentration clearly until the concentration reaches 2.0%. Moreover, at 0.5 and 1.0% dilute acid concentrations, EA-S from ACSE also showed no remarkable changes in the mass distribution. In the range of 0%–1.0% dilute acid concentration, yields of EA-S decreased from 73.0 to 70.8 wt%. However, EA-S-CS from ACO decreased with an increase in the acid concentration. In the same range of 0%–1.0% dilute acid concentration, EA-S-CS decreased from 59.0 to 45.9 wt%. At 2.0% dilute acid concentration, EA-S and EA-S-CS decreased with an increase in the acidity.

**Figure 3.** Yields of the solid fractions from acid catalyzed steam explosion (ACSE) and alkali-catalyzed organosolv treatment (ACO) of EFB.

Table 3 shows the holocellulose and lignin contents of EA-S and EA-S-CS calculated for each experimental run. The total amount of holocellulose harshly decreased with increasing dilute acid concentration. At 0.5% dilute acid concentration, 41.3 and 26.8 wt% of holocellulose still remained in EA-S and EA-S-CS, respectively. When the concentration increased to 2.0%, however, only 28.1 and 16.0 wt% of holocellulose remained in EA-S and EA-S-CS, respectively. Lignin exhibited somewhat different behavior in mass decrease. The amount of lignin did change slightly by the steam treatment. Hwoever, after ACO, 4.3%–6.7% of lignin was removed from the solid fractions. The removal efficiency of lignin between 0.5% and 2.0% acid concentration respectively. Similar studies were reported for the dilute acid treatment of rice straw [20] and corn stover [21]. However, the proper acid concentration or removal efficiencies of hemicellulose and lignin were different by feedstock properties. Since the purpose of this experiments is to lower only hemicellulose and lignin contents in EFB, so a high acid concentration was not adapted to avoid the loss of holocellulose contents.


**Table 3.** Component analysis of acid pretreated EFB solid fractions.

EA-S: solid fraction obtained from acid catalyzed steam explosion treatment of acid impregnated EFB. EA-S-CS: cellulose rich solid fraction obtained from alkali catalyzed organosolv treatment of EA-S fraction.

#### 3.1.3. Sugars Quantification

Tables 4 and 5 show the total sugar contents of each product fractions (EA-S, EA-S-CS, EA-L, and EA-S-OL). Glucose content in the solid fractions showed no significant change while lignin content increased compared to the raw material. It means that the glucose in the EFB took limited damage by the steam explosion process and the organosolv treatment. It is possible that long reaction times increased the accumulation of degradation by-products and mass losses by volatilization [22]. As the acid concentration increased, however, the amount of glucose in the solid steadily decreased. When the acid concentration increased to 2.0%, the glucose amount decreased from 415 to 380 mg/g fraction. In contrast, concentration of glucose in EA-L dropped with 0.5 wt% acid pretreatment but increased by increasing acid concentration. Xylosyl groups were the most affected hemicellulose components during acid impregnation and the pretreatment process. The amount of xylose in the solid drastically decreased with increasing dilute acid concentration. As a result, concentration of xylose in the liquid fraction increased to twice of the control (10.1–21.9 mg/mL liquid). At 2.0% dilute acid concentration, more than 70% of the xylan in the EFB was hydrolyzed during two-step pretreatments. Arabinosyl, galactosyl, and mannosyl residues were removed from the solids by steam explosion with more than 0.5% acid concentration (Table 5).


**Table 4.** Sugar content of pretreated EFB solid fractions under different temperatures and residence times.

**Table 5.** Sugar content of acid pretreated EFB liquid fractions.


EA-L: liquid fraction obtained from acid catalyzed steam explosion treatment of acid impregnated EFB. EA-S-OL: organosolv liquor obtained from alkali catalyzed organosolv treatment of EA-S fraction.

#### *3.2. Cellulose Crystallinity*

The change of cellulose crystallinity as a result of the steam explosion and organosolv treatment was determined by analyzing the x-ray diffraction patterns (Figure 4 and Table 6). By the XRD peak height method, crystallinity index was calculated from the ratio of the height of the 002 peaks (22.7◦ ) and the height of the minimum (18.3◦ ) between the 002 and the 101 peaks [23]. The crystallinity of EFB fiber decreased due to the steam explosion and organosolv treatment. The lowest crystallinity was shown in E-S-CS under the condition of 180-10 (35.9%). The crystallinity of EFB increased by increasing the severity factor. This might be caused by the removal of the amorphous structure of cellulose or components like hemicellulose.

Cellulose crystallinity after the dilute acid impregnation followed by two-step pretreatment was determined by analyzing the X-ray diffraction patterns. The crystallinity of EFB fiber also decreased by both ACSE process and ACO. There was little change of crystallinity, however, between the raw and organosolv treatment.

**Figure 4.** X-ray diffraction of (**A**) E-S-CS and (**B**) EA-S and follow-up EA-S-CS.


**Table 6.** X-ray diffraction of (**A**) E-S-CS and (**B**) EA-S and follow-up EA-S-CS.

#### *3.3. Chemical Properties of Organosolv Lignin*

Qualitative and quantitative analysis of obtained organosolv lignins from E-S-OL were performed via pyrolysis-GC-MS analysis, and the results are listed in Table 7. Organosolv lignin obtained under steam explosion conditions of 200 ◦C and 5–10 min exhibited more than 14 peaks. Phenolic compounds, such as toluene (1), phenol (2), guaiacol (5), syringol (8), and isoeugenol (10), were the major compounds. As shown in Table 7, the number of chemical compounds and the overall peak size in the organosolv lignin decreased by increasing residence time. This might be explained by the fact that the lignin compounds can be condensed in harsher conditions and less decomposed by heat.

**Table 7.** Qualitative and quantitative analysis of chemical compounds in the organosolv lignin from E-S-OL.


### **4. Conclusions**

Two-step pretreatment, ACSE followed by ACO of EFB, was performed under a temperature range of 180–220 ◦C and residence time of 5–20 min after impregnation with 0.5%–2.0% dilute sulfuric acid. The mass balance and the chemical composition of pretreated EFB fractions were noticeably affected by the reaction temperature, residence time, and acid concentration during the steam explosion process. Moreover, glucose and xylose contents decreased under the acid-catalyzed pretreatment conditions. All the analyses together with the XRD peaks showed that the crystallinity of cellulose increased by increasing severity factor. In this experiment, the effect of two-step pretreatment was in no way remarkable but showed potential to advance it. Further study is needed to sort out the optimal conditions for this two-step EFB pretreatment to conclusively suggest its potential for EFB fractionation.

**Author Contributions:** Conceptualization, J.W.C.; data curation, J.H.L.; investigation, J.H.L.; methodology, J.H.L. and J.W.C.; writing—original draft preparation, J.H.L.; writing—review and editing, M.A.A., I.-G.C. and J.W.C.; supervision, J.W.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Basic Science Research Program (NRF-2019R1A2C2086328) and the Technology Development Program to Solve Climate Changes (2017M1A2A2087627) of the National Research Foundation funded by the Ministry of Science and ICT.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **A Simultaneous Conversion and Extraction of Furfural from Pentose in Dilute Acid Hydrolysate of** *Quercus mongolica* **Using an Aqueous Biphasic System**

**Jong-Hwa Kim <sup>1</sup> , Seong-Min Cho <sup>1</sup> , June-Ho Choi <sup>1</sup> , Hanseob Jeong <sup>2</sup> , Soo Min Lee <sup>2</sup> , Bonwook Koo 3,\* and In-Gyu Choi 4,5,\***


**Abstract:** This study optimizes furfural production from pentose released in the liquid hydrolysate of hardwood using an aqueous biphasic system. Dilute acid pretreatment with 4% sulfuric acid was conducted to extract pentose from liquid *Quercus mongolica* hydrolysate. To produce furfural from xylose, a xylose standard solution with the same acid concentration of the liquid hydrolysate and extracting solvent (tetrahydrofuran) were applied to the aqueous biphasic system. A response surface methodology was adopted to optimize furfural production in the aqueous biphasic system. A maximum furfural yield of 72.39% was achieved at optimal conditions as per the RSM; a reaction temperature of 170 ◦C, reaction time of 120 min, and a xylose concentration of 10 g/L. Tetrahydrofuran, toluene, and dimethyl sulfoxide were evaluated to understand the effects of the solvent on furfural production. Tetrahydrofuran generated the highest furfural yield, while DMSO gave the lowest yield. A furfural yield of 68.20% from pentose was achieved in the liquid hydrolysate of *Quercus mongolica* under optimal conditions using tetrahydrofuran as the extracting solvent. The aqueous and tetrahydrofuran fractions were separated from the aqueous biphasic solvent by salting out using sodium chloride, and 94.63% of the furfural produced was drawn out through two extractions using tetrahydrofuran.

**Keywords:** aqueous biphasic system; dilute acid hydrolysate; furfural production; solvent extraction; response surface methodology

### **1. Introduction**

Lignocellulosic biomass is considered as an alternative energy resource that can mitigate the climate change associated with the excessive use of fossil fuels [1]. Owing to its abundance and non-edibility, cellulose in particular, the key component of biomass, is a rich source of carbohydrates and has been applied to value-add to chemicals or biofuels [2]. Hemicellulose and lignin, the other main components of biomass, combine complex and dense forms of cellulose [3]. This physical barrier makes lignocellulosic biomass chemically and microbiologically resistant [4]. This recalcitrance of biomass requires a pretreatment process to ensure that lignocellulosic biomass is utilized properly and efficiently. The purpose of the pretreatment process is to cleave lignin and hemicellulose and obtain cellulose to improve accessibility for chemicals or enzymes [5]. Among several pretreatment methods, dilute acid

**Citation:** Kim, J.-H.; Cho, S.-M.; Choi, J.-H.; Jeong, H.; Lee, S.M.; Koo, B.; Choi, I.-G. A Simultaneous Conversion and Extraction of Furfural from Pentose in Dilute Acid Hydrolysate of *Quercus mongolica* Using an Aqueous Biphasic System. *Appl. Sci.* **2021**, *11*, 163. https:// dx.doi.org/10.3390/app11010163

Received: 29 October 2020 Accepted: 23 December 2020 Published: 26 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/).

pretreatment is considered to be a leading pretreatment technology. This is because it is able to enhance the total sugar yield of the process by solubilizing and converting hemicellulose to pentose [6]. As the pretreatment of lignocellulosic biomass is highly demanding in terms of energy and additional processes compared to edible biomass, its economic feasibility is reduced [7]. As such, utilization of all the main components of biomass to generate valuable products is essential to make this resource more economically feasible.

Hemicellulose is a heteropolysaccharide composed of various monosaccharides such as xylose, arabinose, and mannose [8]. Traditionally, it is considered a by-product in the pulping industry, whereby most hemicellulose is dissolved into black liquor and used as a heating source [9]. However, pentose in hemicellulose may also be converted to valuable chemicals in the same manner as cellulose. This will subsequently improve the economic feasibility of the biorefinery industry through the utilization of lignocellulosic biomass.

Furfural is a building block chemical applied to various fields such as fuel, chemicals, polymers, and pharmaceuticals. Most furfural is produced by acid-catalyzed dehydration of pentose derived from hemicellulose [10]. It may also be produced from lignocellulosic biomass directly via acid catalyst treatment. Although this is a simple and mature process, it is characterized by several disadvantages including low furfural yield, generation of undesirable by-products, and difficulties in utilizing the remaining biomass, such as cellulose or lignin [11]. For this reason, many studies have proposed a two-step process in which hemicellulose is hydrolyzed into pentose or pentose-derived oligomers, and then the pentose or the oligomer is dehydrated into furfural in different reaction systems. This separated system offers several advantages including being able to produce a high furfural yield via the optimization of the reaction system for furfural production (catalyst, solvent, reaction condition, etc.) [12]. The separation of furfural from reaction media continues to be a challenge due to several complicated processes [13].

Furfural may be degraded and condensed in the presence of an acidic catalyst and water [14]. In an aqueous furfural production system that uses water as a solvent, the degradation and self-condensation of furfural limits high yields of furfural. This is despite the fact that water is a commonly used solvent for furfural production because it is inexpensive and eco-friendly. To solve this problem, furfural must be separated from the aqueous solvent system immediately after conversion from pentose.

An aqueous biphasic system may be used to separate furfural from water in the system. The aqueous phase contains water, and the acid catalyst converts pentose to furfural, while the organic phase is composed of the organic solvent that absorbs furfural converted into the organic phase. Dichloromethane (DCM) [15], methylisobuthylketone (MIBK) [16], and tetrahydrofuran (THF) [17] have been used as organic solvents for the extraction. These organic solvents are advantageous in terms of conducting a simultaneous process, including the conversion of pentose to furfural and the extraction of furfural. Furfural is produced by an acid catalyst in the aqueous phase and is extracted immediately by the organic phase [18]. Organic solvents may minimize furfural degradation, improving furfural yield [19].

This study aims to optimize furfural production from the dilute acid hydrolysate of *Quercus mongolica*. A xylose standard solution with the same acid concentration of dilute acid hydrolysate and extracting solvent (THF) were applied to the aqueous biphasic system to determine the optimal conditions for furfural production. A response surface methodology (RSM) in which the independent variables were reaction temperature, time, and xylose concentration was adopted to optimize the aqueous biphasic system. The organic solvent was also evaluated to select the solvent that prevented the degradation and self-condensation of furfural.

#### **2. Materials and Methods**

#### *2.1. Materials*

The xylose standard was purchased from Sigma-Aldrich Korea Co. (Yongin, Republic of Korea). *Quercus mongolica*, a xylan-rich hardwood was used as the raw material for pentose and it was supplied by the National Institute of Forest Science (NIFoS, Seoul, Republic of Korea). The particle size of the raw material was reduced to less than 0.5 mm through grinding and sieving using a sawdust producer and air classifier mill, respectively. The moisture content was less than 5%, and the chemical composition was determined using the Laboratory Analytical Procedure of National Renewable Energy Laboratory (NREL, Golden, CO, USA) [20].

#### *2.2. Dilute Acid Pretreatment for Pentose Production from Lignocellulosic Biomass*

Dilute acid pretreatment of *Quercus mongolica* for pentose production was conducted following the methodology described in previous research [21]. Briefly, raw material was mixed with sulfuric acid solution (4%, *w*/*w*) in an Erlenmeyer flask; the solid to liquid ratio was 1 to 7. Then, the flask was placed in an autoclave (MLS-3020, Sanyo, Osaka, Japan) at 121 ◦C for 102.3 min; these are the optimal conditions for pentose production from *Quercus mongolica* as described in the previous research [21]. Following the dilute acid pretreatment, the flask was immediately cooled to room temperature in the ice chamber to stop the reaction. Then, the pentose-rich hydrolysate was separated from the solid residue using a Büchner funnel equipped with filter paper (No. 52, Hyundai Micro Co., Seoul, Korea). The chemical composition of the hydrolysate is shown in Table 1.

**Table 1.** Chemical composition of dilute acid hydrolysate of *Quercus mongolica*.


#### *2.3. Response Surface Methodology for Optimization of Furfural Production from Xylose Standard*

To maximize furfural production from the hydrolysate, the optimum conditions needed to be determined; optimization was conducted using a xylose standard solution. The xylose standard solution was adopted to investigate the relationship between reaction conditions and furfural yield, excluding the effect of impurities. Briefly, 10 mL of xylose standard solution containing a certain concentration of xylose was mixed with 20 mL of organic solvent in a Teflon-lined reactor. The sulfuric acid concentration of the xylose standard solution was adjusted to 4% (*w*/*w*); this was the same as the dilute acid hydrolysate used to obtain the optimum reaction conditions for the hydrolysate. The reactor was then sealed and soaked in an oil bath that had been pre-heated to a target temperature. The mixed hydrolysate was stirred during the reaction using a magnetic stirrer, and the temperature was maintained for a certain reaction time at the target temperature. After the reaction, the reactor was removed from the oil bath and immediately stored in the ice chamber to cool to room temperature and prevent undesirable reactions.

An RSM was adopted to optimize furfural production using a xylose standard solution. The analysis was conducted based on a 2<sup>3</sup> factorial central design (CCD) using Design Expert 11.1.0.1 software (Stat-Ease, Inc., Minneapolis, MN, USA). Table 2 presents 17 sets of reaction conditions composed of six axial points and a duplication of the central point. The reaction temperature (X1, ◦C), reaction time (X2, min), and xylose concentration (X3, g/L) were designated as independent variables, while furfural yield (Y1, %) was the dependent variable.


**Table 2.** 2 3 factorial experimental design varying on 3 factors and results of furfural yield.

The coded level of the CCD from each run was applied to real independent variables as follows:

Variable = value of central point/variation of coded level per one point

Reaction temperature (◦C) = 160/20, reaction time (min) = 120/60, xylose concentration (g/L) = 20/10

Following the optimization of reaction conditions based on the RSM, the extraction solvent was altered from THF to dimethyl sulfoxide (DMSO) or toluene in the optimum conditions to analyze the effects of the extraction solvent.

#### *2.4. Furfural Production from Pentose Derived from Quercus Mongolica*

Simultaneous reactions occurred involving the release of pentose from xylose and the conversion of pentose to furfural in the same reactor without a separation process; this places it in a good stead for industrial application. The pentose derived from xylose was converted to furfural in the same reactor, used for pentose release. The dilute acid hydrolysate of *Quercus mongolica* was mixed with 4% sulfuric acid to adjust xylose concentration to optimal conditions in the Teflon-lined reactor; the total volume was adjusted to 10 mL. Then, 20 mL of THF (i.e., the organic solvent), was added into reactor for furfural extraction. The reaction temperature and time were set to the optimal values as per the results from the RSM analysis. After the reaction, the reactor was removed from the oil bath and immediately stored in the ice chamber to cool to room temperature and prevent undesirable reactions. The organic phase of THF, containing furfural, was separated from the aqueous phase through the addition of NaCl. This was done to investigate the separation efficiency of furfural from the final solution, including furfural and other products.

#### *2.5. Analysis of Furfural and Other Products in Mixed Hydrolysate*

The content of furfural and other products such as the remaining sugars was determined using high performance liquid chromatography (Ultimate-3000, Thermo Dionex, Waltham, MA, USA) with an Aminex 87H column (eluent: 0.01 N sulfuric acid, oven temperature: 40 ◦C, flow rate: 0.5 mL/min). Peaks were identified by comparing the retention time of each peak. The concentration of peaks was identified by comparing the standard calibration curve of each chemical. The furfural yield (Equation (1)) and pentose conversion (Equation (2)) were calculated as follows:

Furfural yield (%) = furfural after reaction (mol)/pentose before reaction (mol) × 100 (1)

Pentose conversion (%) = (pentose before − after reaction (mol))/pentose before reaction (mol) × 100 (2)

#### **3. Results and Discussion**

#### *3.1. Pentose Production during Dilute Acid Pretreatment*

*Quercus mongolica* is chemically composed of 46.67% of glucose, 19.14% of xylose, mannose and galactose (XMG), 0.77% of arabinose, 22.56% of acid insoluble lignin (AIL), 3.19% of acid soluble lignin (ASL), 2.06% of extractives, and 0.05% of ash. It is known that the glycosidic bond between hemicellulose-cellulose is cleaved, and the separated hemicellulose is converted to pentose-like XMG dilute acid pretreatment [22]. As shown in Table 1, the main component in dilute acid hydrolysate was XMG, and the main pentose was xylose [21]. Glucose and arabinose derived from arabinoxylan or glucuronoxylan were also detected; however, they were present in very small amounts compared with pentose. In sugar derivatives, the main product was acetic acid derived from the O-acetyl group in hemicellulose. Although furanic compounds such as 5-hydroxymethylfurfural (5-HMF) and furfural were produced by the acidic dehydration of released sugar, there was only a small amount of these compounds owing to the low severity of the dilute acid pretreatment. This pretreatment had not boosted the dehydration of sugar to the furanic [23].

#### *3.2. RSM for Furfural Production from Xylose Standard Solution with Extracting Solvent*

Furfural was produced from the xylose standard by acid-catalyzed dehydration, as listed in Table 2. In run #2 with a reaction temperature of 180 ◦C, over a 60 min duration using a xylose concentration of 10 g/L, the maximum furfural yield was 69.87% with 96.02% xylose conversion. Although the maximum xylose conversion was 99.32% at run #12, the furfural yield under these conditions was 68.12%; this is lower than the yield in run #2. These results indicate that the more severe experimental conditions of run #12 as a result of a longer reaction time induced the degradation of the furfural that had been produced. The lower furfural yield due to furfural degradation under severe experimental conditions was also observed in runs #4, 8, and 10; these runs were also characterized by long reaction times at certain temperatures. It is inferred that these reaction conditions caused furfural degradation despite the presence of the extraction solvent to prevent furfural degradation. However, a previous study has reported that the amount of furfural degradation due to severe experimental conditions when using an extracting solvent is marginal compared to the amount of furfural degradation in conditions with no extracting solvent [24]. The condensation of furfural was suppressed through the addition of the THF; as such, there were no insoluble precipitates detected from furfural condensation in all experimental conditions.

To evaluate the effect of each variable on furfural yield, regression analysis was undertaken using a 2<sup>3</sup> factorial design matrix with corresponding furfural yield (%). The following quadratic equation was generated (Equation (3)), based on the outcomes of the regression analysis:

Furfural yield (%) = −1122.2746 + 11.8539 X<sup>1</sup> + 2.4079 X<sup>2</sup> + 1.0355 X<sup>3</sup> − 0.0091 X1X<sup>2</sup> − 0.0082 X1X<sup>3</sup> + 0.0015 X2X<sup>3</sup> − 0.0302 X<sup>1</sup> <sup>2</sup> − 0.0032 X<sup>2</sup> <sup>2</sup> + 0.0024 X<sup>3</sup> 2 (3)

In the equation, X1, X2, and X<sup>3</sup> represent the actual reaction temperature, reaction time, and xylose concentration, respectively. The model had a high regression coefficient (R<sup>2</sup> = 0.95), indicating 95% variability in the response, while the *p*-value was extremely low (0.001), indicating that this regression model was significant. The coefficient of variation (CV) was 18.74%, which indicates the high precision and reliability of the experiments [25].

A three-dimensional (3D) plot and detailed contour of the RSM for furfural yield was established using Equation (3), by varying the three variables within the experimental range (Figure 1). As shown in Figure 1a, the furfural yield increased with reaction temperature and time to approximately 185 ◦C and 180 min, respectively. Once the temperature and time exceeded these points, there was a decrease in furfural yield due to its degradation under severe experimental conditions [26].

**2021**, , x FOR PEER REVIEW 6 of 12

− − − − −

**Figure 1.** Three-dimensional plot and detailed contour of response surface methodology of furfural yield from xylose standard solution with extraction solvent (THF): (**a**) fixed xylose concentration (10 g/L); (**b**) fixed reaction time (120 min); (**c**) fixed reaction temperature (160 ◦C); (**d**) contour of furfural yield depending on reaction temperature and time.

This phenomenon was clearly observed in Figure 1b,c. Figure 1b depicts the furfural yield based on the reaction temperature and xylose concentration with a fixed reaction time of 120 min. The furfural yield increased with temperature until approximately 165 ◦C, and then remained stable at temperatures of 165–185 ◦C, which was where the maximum yield was observed. Then, yield decreased when the temperature increased above 185 ◦C regardless of the xylose concentration. These results are similar to the findings from previous research where furfural production from biomass occurred using sulfuric acid as a catalyst [27,28]. Figure 1c depicts the relationship between the furfural yield, the reaction time, and xylose concentration with a fixed reaction temperature of 160 ◦C. The furfural yield increased with reaction time up to 180 min. Beyond this time, yield decreased due to furfural decomposition [29] and self-condensation [30]. The xylose concentration was considered a less influential variable when compared to reaction temperature and time, as shown in Figure 1b,c. This inference was also supported by the p-value of the variables (Table 3).

The sources related to xylose concentration (X3, X1X3, X2X3, X<sup>3</sup> 2 ) had high *p*-values, indicating that xylose concentration was not as significant a factor for furfural yield. This result differs from previous research, which has demonstrated that furfural yield generally has an inverse relationship with xylose concentration. Higher xylose concentrations are able to produce more furfural at once, leading to a higher collision possibility that causes

> − −

− −

condensation of the furfural reaction [31]. It was assumed that the 10–30 g/L xylose concentration range used in this study was too narrow of a range to affect furfural yield. This was unlike a study by Yang [32] where they produced furfural at high xylose concentrations. Yang [32] varied the xylose concentration from 40 to 120 g/L and found that furfural yield had been relatively stable when the xylose concentration was from 40 to 70 g/L. Based on the xylose concentration range used in this study (10–30 g/L), the effect on furfural yield was negligible compared to reaction time and temperature. With a fixed xylose concentration there was greater clarity in terms of the optimal conditions to maximize furfural yield. Additionally, the reaction temperature and time were expanded compared to the experimental range (Figure 1d).


**Table 3.** Analysis of variance (ANOVA) for furfural yield in dehydration of xylose model compound and coefficients for quadratic equation.

The optimal conditions to maximize furfural yield was calculated based on Equation (3). The maximum furfural yield in the predicted reaction conditions was 75.1%, where the reaction temperature was 170 ◦C, reaction time was 120 min, and xylose concentration was 10 g/L. To verify the model, actual furfural production was carried out using these predicted optimal conditions, rendering a furfural yield of 72.39%, similar to the predicted yield.

#### *3.3. Effect of Organic Solvent for Furfural Production and Extraction*

Three kinds of organic solvents, THF, toluene, and DMSO, were evaluated to understand the effects of organic solvents on furfural production and extraction. Toluene is considered an effective solvent for furfural extraction [33], whereby it does not require additional salt for phase separation because of its immiscibility with water. DMSO has been used to improve the selectivity of 5-HMF from glucose by increasing the fructofuranose isomer and stabilizing 5-HMF by hydrogen bonding [34]. Under a similar mechanism, it was anticipated that DMSO could also improve furfural yield from xylose. Reaction conditions were set to the optimal values predicted from analysis of variance (ANOVA); this was a reaction temperature of 170 ◦C, reaction time of 120 min, xylose concentration of 10 g/L of, and the use of a 4% of sulfuric acid solution.

Table 4 shows the furfural yield from the xylose standard solution depending on the organic solvent. THF had the highest furfural yield from the xylose standard, while DMSO had the lowest among the three solvents. It was assumed that DMSO may not effectively protect the generated furfural from acid or water as furfural has no hydroxyl group compared with 5-HMF. In addition, DMSO has a reduced interaction with furfural compared to the other organic solvents such as toluene or THF, as its polarity is higher than that of furfural. THF had a higher furfural yield than toluene, even though the polarity of toluene was close to furfural. To explain this phenomenon, the partition coefficient of furfural in THF/water and toluene/water was calculated by dividing furfural

concentration in the organic solvent by the concentration of furfural in aqueous water (Table 4) to compare the solubility of furfural in a two-liquid mixture (water and organic solvent). To calculate the partition coefficient of furfural to THF/water, NaCl was added to separate THF from the water phase (salting out). THF had a higher partition coefficient than toluene, indicating that the former was more effective in extracting furfural than toluene, and yielding higher amounts of furfural.

**Table 4.** Properties of organic solvent and furfural yield from xylose standard based on the type of organic solvent utilized.


\* Partition coefficient = [Furfural]org/[Furfural]aq, \*\* N/D: Not detected.

#### *3.4. Production of Furfural from Pentose in Dilute Acid Hydrolysate*

The optimal reaction conditions as predicted using the ANOVA (i.e., 170 ◦C, 120 min, and 10 g/L of xylose) were adopted to maximize furfural production from pentose in dilute acid hydrolysate. The xylose concentration was adjusted by mixing the hydrolysate, which had already contained 4% (w/w) sulfuric acid, with 4% sulfuric acid solution. The total volume of the aqueous phase was adjusted to 10 mL, similar to the xylose standard solution, and 20 mL of THF was added to extract furfural from the aqueous phase.

Table 5 describes the pentose conversion and furfural yield from the xylose standard and dilute acid hydrolysate. The furfural yield and pentose conversion of hydrolysate were slightly lower, compared with the xylose standard. The presence of impurities such as hexoses, organic acids, and acid soluble lignin, may have impacted on the extraction efficiency of furfural from the hydrolysate [39].

**Table 5.** Pentose conversion and furfural yield from xylose standard and dilute acid hydrolysate.


The difference in furfural yield between the xylose standard and the hydrolysate was not as considerable as had been expected. This indicates that THF effectively prevents furfural loss by inhibiting the ring opening of furfural and condensation between furfural and acid soluble lignin to form insoluble precipitate [40].

To investigate the distribution of impurities in the organic and aqueous phases, NaCl was added to separate the hydrolysate into the organic and aqueous phases. Table 6 presents the change in the distribution of furfural and other chemicals in furfural that produced hydrolysate by phase separation. Most furfural produced was extracted in the organic phase with a partition coefficient of 8.43; this is slightly lower than that of the xylose standard solution (9.05). Sugars, such as glucose, XMG, and arabinose favor the aqueous phase owing to their hydroxyl group, while other chemicals such as furfural, organic acids, and ASL tend to be extracted by the organic phase of THF. It is known that THF is able to effectively dissolve lignin as it has a high affinity to phenolic compounds [41], thus, most ASL was extracted to THF. Approximately three-quarters of organic acids were extracted to the organic phase, and the THF was effective in extracting various organic acids; however, the detailed extraction mechanism of THF to organic acid continues to be unclear [42,43].


**Table 6.** Concentration of chemicals in furfural produced liquor prior to and after phase separation (organic phase, aqueous phase) through the addition of NaCl.

> THF, the separated organic phase, was removed, and 20 mL of fresh THF was added to 10 mL of furfural extracted aqueous phase to enhance the extraction rate of furfural to the organic phase. After each additional extraction, the amount of extracted furfural and other compounds were analyzed; the extraction rate was calculated as (Equation (4)):

Extraction rate (%) = Amount of products extracted in THF phase (g)/Amount of products existed in furfural produced liquor before phase separation (g) (4)

Figure 2 shows the increase in the furfural extraction rate to THF with the number of extractions. Following the second additional extraction, the extraction rate increased from 86.03% to 94.63%. However, impurities such as organic acids and ASL had also been further extracted as the number of extractions increased. In particular, organic acids such as formic acid and acetic acid were completely extracted to THF following the second extraction. This means that additional impurity separation is required to acquire furfural with higher purity. It was reported that organic acids and ASL may be separated from THF by the ion exchange resin [44,45], and absorbents such as activated carbon, respectively [46]. However, lignin separation by an absorbent must occur prior to furfural production, as the absorbent absorbs furfural and ASL through a π-π interaction [47]. **2021**, , x FOR PEER REVIEW 10 of 12 π π

**Figure 2.** Extraction rate of products in furfural generated hydrolysate by extracting solvent (THF).

#### **4. Conclusions**

This study aimed to optimize furfural production from pentose in the dilute acid hydrolysate of *Quercus mongolica*. The main component of the acid hydrolysate was XMG, which was dominated by xylose. The optimization of furfural production was conducted using RSM with a xylose standard solution and an extracting solvent THF, to enhance furfural yield. The optimal conditions included a reaction temperature of 170 ◦C at a reaction time of 120 min with a xylose concentration of 10 g/L; the predicted furfural yield


was 75.1% under these conditions. An experimental furfural yield of 72.39% was obtained under the optimal experimental conditions, similar to the predicted yield. Extracted solvents such as THF, toluene, and DMSO were evaluated to understand the effect of the solvent on furfural yield. THF achieved the highest furfural yield, while DMSO had the lowest yield. Based on this result, furfural was produced from dilute acid hydrolysate under optimized reaction conditions using THF as the extracting solvent. A furfural yield of 68.20% was obtained based on pentose in the hydrolysate, similar to that of the xylose standard solution (72.39%). Following phase separation through the addition of NaCl, 86.03% of the furfural produced was in the organic phase. The THF, and two additional extractions using fresh THF enhanced the furfural extraction rate to 94.63%.

**Author Contributions:** Conceptualization, J.-H.K. and B.K.; methodology, S.-M.C. and J.-H.K.; formal analysis, J.-H.K. and J.-H.C.; investigation, J.-H.K.; resources, H.J. and S.M.L.; data curation, J.-H.K. and J.-H.C.; writing-original draft preparation, J.-H.K.; writing-review and editing, J.-H.K.; visualization, J.-H.K.; supervision, I.-G.C. and B.K.; project administration, B.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was carried out with the support of the R & D program for Forest Science Technology (Project No. 2020226C10-2022-AC01) provided by the Korea Forest Service (Korea Forestry Promotion Institute) and the Research Program (FP0900-2019-01) of the National Institute of Forest Science (NIFoS) (Seoul, Republic of Korea).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Environmentally Friendly Approach for the Production of Glucose and High-Purity Xylooligosaccharides from Edible Biomass Byproducts**

#### **Soo-Kyeong Jang <sup>1</sup> , Chan-Duck Jung <sup>2</sup> , Ju-Hyun Yu <sup>2</sup> and Hoyong Kim 2,\***


Received: 26 October 2020; Accepted: 13 November 2020; Published: 16 November 2020 -

**Abstract:** Xylooligosaccharides (XOS) production from sweet sorghum bagasse (SSB) has been barely studied using other edible biomasses. Therefore, we evaluated the XOS content as well as its purity by comparing the content of total sugars from SSB. An environmentally friendly approach involving autohydrolysis was employed, and the reaction temperature and time had variations in order to search for the conditions that would yield high-purity XOS. After autohydrolysis, the remaining solid residues, the glucan-rich fraction, were used as substrates to be enzymatically hydrolyzed for glucose conversion. The highest XOS was observed for total sugars (68.7%) at 190 ◦C for 5 min among the autohydrolysis conditions. However, we also suggested two alternative conditions, 180 ◦C for 20 min and 190 ◦C for 15 min, because the former condition might have the XOS at a low degree of polymerization with a high XOS ratio (67.6%), while the latter condition presented a high glucose to total sugar ratio (91.4%) with a moderate level XOS ratio (64.4%). Although it was challenging to conclude on the autohydrolysis conditions required to obtain the best result of XOS content and purity and glucose yield, this study presented approaches that could maximize the desired product from SSB, and additional processes to reduce these differences in conditions may warrant further research.

**Keywords:** xylooligosaccharides; autohydrolysis; enzymatic hydrolysis; sweet sorghum bagasse

#### **1. Introduction**

Due to an emerging interest in a healthy lifestyle, probiotics have attracted attention [1]. Probiotics have several beneficial effects on the human body, especially as it reduces the risk of colon-related diseases and dysfunctions [2]. Prebiotics have also attracted great interest because they facilitate the growth of probiotics and inhibit the growth of pathogenic microorganisms in the human intestine [3].

Xylooligosaccharides (XOS) are one of the fascinating products of biomass that have unique properties as prebiotics [4]. XOS do not only reduce the risk of colon cancer, but also improve beneficial biological activities, including reducing dental caries, boosting the immune system, and restraining the growth of pathogens [5]. Food and food additives have been recognized as conventional methods for XOS ingestion, but the application area of XOS is expanding the pharmaceutical, chemical, and nutraceutical industries [6]. The global market size of XOS is expected to grow by 4.1% annually and reach 135.7 million dollars by 2026 [7].

XOS is typically defined as two to ten xylose combined by β-1,4 linkages with arabinose, acetyl groups, and uronic acid substitution [8]. Chemical (acid pretreatment and autohydrolysis) and biological (enzymatic hydrolysis by xylanase) approaches have been developed for hydrolyzing and extracting the xylan chain in a biomass [9]. Concentrated acid pretreatment using inorganic acids has been reported to result in fast isolation of the xylan fraction, but this method is not environmentally friendly and corrodes the equipment [10]. Meanwhile, the biological method can control the generation of byproducts, as shown in the results of acid pretreatment, but it has drawbacks such as long reaction time, relatively low XOS yield, and high cost of xylanase [11,12]. Autohydrolysis is considered an eco-friendly pretreatment method because it does not use any chemicals for the reaction [13]. When a biomass is heated in water, the uronic and acetyl groups from the hemicellulose fraction can be released in the form of an acid followed by acid-hydrolysis of the carbohydrate [14]. This acidic circumstance of autohydrolysis from the side groups of hemicellulose does not lead to an extremely low pH condition compared to the mineral acids of typical acid pretreatments [15]. Thus, autohydrolysis can be a less corrosive and low-cost approach for XOS production than other chemical methods [13].

Several types of xylan-rich biomasses are used for producing XOS, such as corn stover, wheat straw, sugarcane straw, and sugarcane bagasse [16–19]. However, the straws from agricultural residues usually contain a high amount of minerals, including amorphous silica, which can cause chemical reactions to occur in an undesired manner [20]. The production of numerous XOS has been focused on the utilization of sugarcane bagasse, while interest for sweet sorghum is still lower than that of other biomasses [21]. Sweet sorghum can be cultivated in most regions of the world, from temperate to tropical [22]. It typically grows over 3 m tall, ensuring a high energy density per cultivating area up to 20.2 tons per ha [23]. Additionally, sweet sorghum bagasse (SSB) contains an abundant amount of xylan, which is an advantage for XOS production [24].

The physiological activities of XOS have been considered to be strongly related to the linkage-type, substituted side groups, and the degree of polymerization (DP) [25]. Regarding the DP, XOS has a low DP from xylobiose (DP = 2) to xylotetraose (DP = 4), and performs better as a prebiotic than XOS with a high DP [26,27]. On the other hand, xylose and furfural, which are dehydrated products from xylose, can be excessively produced when high temperature or/and low pH conditions are induced to focus on the cleavage of the xylan structure [28]. Therefore, suitable autohydrolysis conditions for improving the XOS yield and purity should be investigated. In addition, when the high-valued XOS containing a dominant amount of DP 2-4 XOS produced with additional xylanase treatment as a follow-up step, the optimum conditions can reduce the purification and production costs. In this study, SSB, a promising biomass, was subjected to the autohydrolysis process to produce XOS. The conditions were evaluated for a high ratio of XOS to total sugars. After autohydrolysis, the remaining solid residues, cellulose-rich fraction, were used for enzymatic hydrolysis to produce a glucose solution.

#### **2. Materials and Methods**

#### *2.1. Feedstock*

Sweet sorghum (*Sorghum bicolor* (L.)) bagasse was generously supplied by Good Farmer Co., Ltd., Yecheon-gun, Gyeongsangbuk-do, Korea. The feedstock was oven dried and milled to <0.5 mm using a knife mill. The chemical composition of sweet sorghum is shown in Table 1.


**Table 1.** Chemical composition of sweet sorghum (g/100 g oven dry weight).


**Table 1.** *Cont.*

<sup>1</sup> All the numbers are based on the initial (OD) weight of the analyzed sample. <sup>2</sup> Sum of xylan, galactan, and mannan.

#### *2.2. Autohydrolysis*

Autohydrolysis was conducted to identify a good point to maximize the XOS amount in the liquid hydrolysate. It was carried out using a bomb-type mini-reactor (30 mm ID × 140 mm L, wall thickness 5 mm, total volume 100 mL, KRICT, Daejeon, Korea), equipped with a K-type thermocouple for monitoring the actual reaction temperature. The reaction temperature was set as a variable from 160 to 200 ◦C at 10 ◦C intervals. One gram of the oven-dried weight of sweet sorghum bagasse was placed in a custom-made reactor with 20 mL of water. When the reactor reached the designated reaction temperature in an oil bath, it was held for 20 min (reaction time). The reaction time was also changed for 180 and 190 ◦C conditions from 5 to 20 min with 5 min intervals. After autohydrolysis, slurries were recovered from the reactor into a 50 mL conical tube using an additional 20 mL of water and separated into a solid residue and liquid hydrolysate by centrifugation.

#### *2.3. Enzymatic Hydrolysis*

Enzymatic hydrolysis was performed using the solid residue from autohydrolysis. The solid residue obtained from each autohydrolysis condition was directly used as a substrate for the commercial cellulase complex Cellic CTec3 (Novozyme Korea, Seoul, Korea). The dosage of the cellulase complex was 10 FPU/g glucan in the substrate. The citrate buffer solution was adjusted to pH 5.5, and the mixture was incubated in a shaker at 50 ◦C for 72 h at 150 rpm.

The glucose to total sugar ratio was calculated as follow:

$$\text{Glucose to total sugar ratio (\%)} = \frac{\text{Glucose in the solution after enzymatic hydroxide (g)}}{\text{Total sugars in the solution after enzymatic sulphlysis}} \times 100 \tag{1}$$

#### *2.4. Determination of the Chemical Composition*

The chemical composition of the liquid hydrolysate, enzymatic hydrolysis solution, and the raw material (SSB) were determined using the Laboratory Analytical Procedure of NREL [29,30]. Furthermore, additional acid hydrolysis using 4% sulfuric acid solution was conducted to convert all kinds of carbohydrates in the liquid hydrolysate to monomeric sugars. The amount of monomeric sugars was quantified by high-performance liquid chromatography (HPLC) (Agilent 1200 Infinity, Agilent Technologies Korea Inc., Seoul, Korea). HPLC was equipped with an HPX-87H (300 × 7.8 mm) column (Bio-Rad, Hercules, CA, USA) and a refractive index detector (RID-6A, Bio-Rad, Hercules, CA, USA). Twenty microliters of sample was injected for the HPLC analysis, and the oven temperature was set at 50 ◦C. The mobile phase was 5 mM sulfuric acid solution and eluted at a 0.6 mL/min flow rate. For quantification, the calibration curves were made using standard materials (glucose, xylose, and arabinose), which were purchased from Sigma-Aldrich Korea Co. (Yongin, Korea). The nitrogen content of SSB was determined with an elemental analyzer (Flash EA 2000, Thermo Electron Corporation, Waltham, MA, USA), and crude protein content was calculated using nitrogen content and conversion factor (6.25) [31].

The XOS to total sugar ratio was calculated by following formula:

$$\text{XOS to total sugar ratio } (\%) = \frac{\text{XOS in the liquid hydrolyte after naturallys is } (\text{g})}{\text{Total sugars in the liquid hydrolyte after naturallys is } (\text{g})} \times 100\tag{2}$$

#### **3. Results**

#### *3.1. Xylooligosaccharides Production Depending on the Changes of the Reaction Temperature*

Hemicellulose fractions in SSB were released into the liquid hydrolysate in the form of both monomeric and oligomeric sugars after autohydrolysis (Table 2). Hemicellulosic sugars could not be fully hydrolyzed due to the severity of autohydrolysis, and a large amount of oligomeric sugars might be dissolved in the liquid hydrolysate together with monomeric sugars.


**Table 2.** Chemical composition of the liquid hydrolysate by reaction temperature variables.

<sup>1</sup> Sum of xylose, galactose, and mannose; <sup>2</sup> Hydroxymethylfurfural; <sup>3</sup> Not detected.

The release trends of the three kinds of sugars were different depending on the changes in the reaction temperature. The XGM content in the liquid hydrolysate increased by increasing the reaction temperature to 18.4% based on the initial biomass, while the XGM content sharply decreased at temperatures above 200 ◦C. Based on the initial arabinan content as monomers, 84.6% of the arabinose could be released at 170 ◦C. The hemicellulosic sugars have good solubility in water after autohydrolysis, but the maximum release point was slightly different between XGM and arabinose. The glucose content showed a constant value of approximately 4% in the liquid hydrolysate. Compared to the initial glucan content in SSB, a small portion of glucose was released from SSB regardless of the reaction temperature changes. Thus, the autohydrolysis could selectively extract XGM in SSB, and the liquid hydrolysate generally had a large amount of XGM and a small amount of glucose and arabinose.

As the reaction temperature increased, the content of monomeric XGM dramatically increased due to the excess hydrolysis of hemicellulose (Figure 1). Meanwhile, XOS content was maximized at 180 ◦C (15.5%). Similar to the XOS content, the highest XOS to total sugar ratio was achieved at the same reaction temperature (180 ◦C). Therefore, an appropriate reaction temperature for improving the XOS ratio in the liquid hydrolysate could be found for SSB.

#### *3.2. Xylooligosaccharides Production Depending on the Changes of the Reaction Time*

The autohydrolysis conditions for XOS production were investigated by changing the reaction time because it has been considered as one of the critical hydrothermal treatment parameters for the control of XOS DP. The reaction time was specified in 5-min intervals within 20 min to confirm the result in a shorter reaction time than that of the previous sections. The reaction temperature was set at 180 and 190 ◦C, which has a good performance for the high XOS ratio in the liquid hydrolysates.

The XGM content in the liquid hydrolysate was changed by an increase in the reaction time (Table 3). Both reaction temperature conditions presented an increasing trend of XGM content with longer reaction times, except for 20 min at 190 ◦C. The maximum XGM content was obtained at 180 ◦C for 20 min (16.9%) and 190 ◦C for 15 min (18.3%), respectively.


**Table 3.** Monomeric sugar composition of the liquid hydrolysate by reaction temperature and time variables.

<sup>1</sup> Sum of xylose, galactose, and mannose; <sup>2</sup> Hydroxymethylfurfural; <sup>3</sup> Not detected.

Although a good maximum XGM content was obtained at 190 ◦C than at 180 ◦C, the results of the XOS to total sugar ratio are quite different between the two conditions (Figure 2). In the cases of 180 ◦C, the XOS to total sugar ratio was constant even though the reaction time was changed (Figure 2a). Meanwhile, the ratio showed a sharp decrease as the reaction time was extended to 20 min at 190 ◦C (Figure 2b). Although 180 ◦C for 20 min showed the highest XOS content (15.4%), this condition could not guarantee the highest level of XOS to total sugar ratio (67.7%). Therefore, 190 ◦C for 5 min might

be considered as the best point for XOS production from SSB due to the high content of XOS (14.7%) and high purity (XOS to total sugar ratio: 68.7%) than other autohydrolysis conditions.

**Figure 2.** XGM content and XOS to total sugar ratio with reaction time change: (**a**) 180 ◦C; (**b**) 190 ◦C.

#### *3.3. Glucose Conversion Depending on the Reaction Temparature Changes*

The cellulose fraction in the solid residues after autohydrolysis was hydrolyzed to glucose by an enzyme cocktail (Figure 3). According to the sugar composition in the liquid hydrolysate, a small amount of glucose was released regardless of the changes in the reaction temperature (Table 1). However, the glucose yield after enzymatic hydrolysis showed an increasing trend with increasing reaction temperature. The glucose to total sugar ratio presented a similar trend with the glucose yield corresponding to the reaction temperature increase. Thus, the maximum glucose yield (36.6%) and glucose to total sugar ratio (94.2%) were obtained at 200 ◦C.

**Figure 3.** Glucose yield and glucose to total sugar ratio after enzymatic hydrolysis.

#### *3.4. Glucose Conversion Depending on the Changes of the Reaction Time*

The glucose yield was improved by extending the reaction time under both reaction temperature conditions (Figure 4). In addition, the glucose to total sugar ratio was maximized at 20 min, but the results were slightly different between the 180 ◦C (86.2%) and 190 ◦C (93.9%) conditions. However, it was revealed that a condition with a high glucose to total sugar ratio does not ensure a high XOS to total sugar ratio by comparing the results of the previous section.

**Figure 4.** Glucose content and conversion ratio after enzymatic hydrolysis: (**a**) 180 ◦C; (**b**) 190 ◦C.

This trade-off relationship between XOS and glucose is obvious in the mass balance profile of this study (Figure 5). This figure suggests three autohydrolysis conditions: the first and second conditions represent the highest XOS to total sugar ratio at 180 and 190 ◦C, respectively. At 190 ◦C for 5 min, the XOS to total sugar ratio was higher (68.7%) than that at 180 ◦C for 20 min (67.6%). Meanwhile, the 180 ◦C condition has a superb result for the glucose to total sugars ratio (86.2%) than the 180 ◦C result (83.5%). Meanwhile, the 190 ◦C for 15 min condition presented a higher number of glucose to total sugar ratios (91.4%) compared to the two previous conditions, even though this condition had a moderate level of XOS to total sugar ratio (64.4%). Therefore, the autohydrolysis condition should be properly considered for optimizing the production of major products from SSB.

**Figure 5.** Mass balance of the XOS and glucose production from sweet sorghum bagasse.

#### **4. Discussion**

Numerous studies have attempted to improve the release performance of XOS from biomasses that directly connected the productivity of XOS [16–19]. However, control of impurities in the liquid fraction after treatment is considered a major factor affecting the market price of XOS as well as its yield [2,32]. Due to the fact that the purification process of XOS is costly, it can dramatically increase if a high amount of impurities are simultaneously mixed with XOS [28]. Another reason for the importance of controlling the impurities is related to the XOS activity [26]. In other words, the concept of impurity can be expanded to monomeric sugars, such as xylose [33]. XOS has several beneficial activities for the human body as a prebiotic, while xylose does not have any function like XOS [34]. Thus, preventing the excess generation of xylose is an essential factor for increasing the productivity of XOS.

The chemical composition of SSB showed good potential for XOS production (Table 1). It had a high amount of xylan, and a low amount of ash and other hemicellulosic sugars. In particular, a very tiny amount of galactan and mannan in SSB has been reported in previous studies [35,36]. Therefore, the XGM monomer in the liquid hydrolysate can be considered as xylose alone, and the XGM oligomer implied XOS in this study.

The acid-hydrolysis pathway of xylan is shown by the variation of the XGM content with an increase in the reaction temperature (Table 2). The xylan seemed to be fully degraded and dissolved in liquid hydrolysate at 190 ◦C. Then, the XGM content sharply dropped at 200 ◦C because it converted to HMF and furfural, which are degradation products. The pH level gradually decreased by providing more heat energy due to the cleavage of the acetyl group from the hemicellulose chain, which resulted in more acidic circumstances at higher reaction temperature conditions [37]. This means that xylan degradation might be accelerated by positive feedback from heat energy and acids during autohydrolysis. This phenomenon was revealed by the variation in the XGM monomer content, which increased more in the higher reaction temperature conditions (Figure 1). Thus, autohydrolysis proved that the xylan structure in SSB can be sufficiently decomposed without the addition of chemical catalysts.

As mentioned above, the quality of XOS might be controlled by reducing the impurities, including the monomeric sugars. In this term, the ratio between the XOS and the sum of glucose, arabinose, and XGM monomer in the liquid hydrolysate was calculated to evaluate the quality of XOS, and it was named XOS to total sugar ratio in this study. In other words, the XOS to total sugar ratio implies the proportion of XOS compared to the total sugars in the liquid hydrolysate. When the autohydrolysis conditions had a designated reaction time (20 min), the 170 ◦C condition showed a better XOS to total sugar ratio (64.2%) than that at 190 ◦C (57.5%) (Figure 1). The liquid hydrolysate at 170 ◦C might have fewer impurities than that at 190 ◦C, but the quantity of XOS and monomers at 190 ◦C (13.9% and 4.5%) was higher than that at 170 ◦C (11.6% and 0.5%). Considering the acid-hydrolysis pathway of xylan during autohydrolysis, the high amount of XGM monomer implies that the liquid hydrolysate at 190 ◦C could have a higher amount of low DP of XOS than that at 170 ◦C [38]. According to previous research, the low DP of XOS presented performed better as a prebiotic than high DP (above 4) of XOS [39].

180 ◦C, which had the highest XOS to total sugar ratio, and 190 ◦C conditions were selected for evaluating XOS production from SSB with several variations of the reaction time (Figure 3). The highest XOS content (15.4%) in this study was obtained at 180 ◦C for 20 min and 190 ◦C for 15 min. This XOS content can be converted to conversion ratio based on the initial amount of xylan in SSB (64.1%), which may be considered as a noteworthy value compared with the results of previous autohydrolysis studies: 60.9% from miscanthus [32], 61% from the brewery's spent grain [40], 51.7% from rice husks, and 51.8% from corn cob [41].

Glucan was presented as a major carbohydrate in SSB, even though it contained twice as much as the XGM in this study. Glucose has been recognized as a platform chemical from biomass and can be easily produced by enzymatic hydrolysis if the cell wall structure is appropriately degraded by the pretreatment [42]. The glucose to total sugar ratio, which was calculated in the same manner as the case of XOS, gradually increased with an increase in the reaction temperature and time (Figures 2 and 4). Meanwhile, less than 4% of glucose was released into the liquid hydrolysate even at the highest reaction temperature (Tables 2 and 3). This means that the amount of remaining glucan had not changed much by changing the autohydrolysis conditions, but the cell wall structure might be

degraded toward a suitable form for cellulase activity at high reaction temperatures and long reaction times. The highest glucose to total sugar ratio (94.2%) was observed at 200 ◦C for 20 min, and an indication of glucan degradation could not be found under the autohydrolysis conditions in this study.

The ideal autohydrolysis condition for XOS production could be suggested by the variations in reaction temperature and time (Figure 5). A XOS to total sugar ratio of 68.7% was obtained at 190 ◦C for 5 min, and a short reaction time can be an advantage for large-scale plants [43]. However, the result of this condition is concerned that the XOS might consist of a significant amount of high DP of XOS. According to this assumption, the liquid hydrolysate under this condition may require further processing to reduce the DP by xylanase. Additionally, this condition showed a relatively low glucose to total sugar ratio (83.5%), another main product of autohydrolysis, than other conditions. In this term, 180 ◦C for 20 min can be an alternative to 190 ◦C for 5 min because the XOS to total sugar ratio in this condition (67.6%) is slightly lower than that at 190 ◦C for 5 min. Meanwhile, the higher XGM monomer content (1.4%) in this condition means that the average DP might be lower than that at 190 ◦C for 5 min. If xylanase treatment followed by autohydrolysis induces the formation of XOS with a narrow DP from 2 to 4, the XOS DP variation in the liquid hydrolysate should be considered as an important factor in determining the amount of xylanase consumed. In this regard, 180 ◦C for 20 min might be more favorable for producing high-value XOS than at 190 ◦C for 5 min. On the other hand, 190 ◦C for 15 min can be chosen to obtain a high purity of glucose, 91.4% of glucose to total sugar ratio, even though this condition showed a moderate level of XOS to total sugar ratio (64.4%). This trade-off relationship between XOS and glucose might be controlled by inducing an additional process between the autohydrolysis and enzymatic hydrolysis steps. For instance, mechanical refining can be employed for solid residues, and it improves glucose conversion by collapsing the cell wall structure, even though the solid residue is produced under low reaction temperature conditions [44].

#### **5. Conclusions**

The xylan-rich edible biomass, SSB, was utilized for XOS and glucose production using the autohydrolysis process. The purity of XOS was improved by controlling the autohydrolysis conditions, and XOS to total sugar ratio was maximized by up to 68.7%, which is expected to reduce production costs for the purification process. However, both XOS purity and content should be considered to determine the optimum conditions for XOS production. On the other hand, the remaining solid residue after autohydrolysis was successfully converted to glucose by enzymatic hydrolysis, and the highest glucose to total sugar ratio was 94.2%. Although there was a discrepancy between the ideal conditions for XOS and glucose production, the results of this study can be useful in selecting the suitable autohydrolysis condition for improving the production of the desired material from SSB.

**Author Contributions:** Conceptualization, H.K.; methodology, C.-D.J. and H.K.; validation, H.K.; formal analysis, C.-D.J.; investigation, C.-D.J. and H.K.; resources J.-H.Y. and H.K.; data curation, S.-K.J. and H.K.; writing—original draft preparation, S.-K.J.; writing—review and editing, S.-K.J. and H.K.; visualization, S.-K.J.; supervision, H.K.; project administration, H.K.; funding acquisition, H.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Korea Research Institute of Chemical Technology (KRICT, Korea) project (SS2042-10) and R&D Program (20008416) of the Ministry of Trade, Industry & Energy (MOTIE/KEIT, Korea).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
