Con fi guration modi fi cation of a submerged membrane reactor for enzymatic hydrolysis of cellulose

Contents lists available atScienceDirect

Biocatalysis and Agricultural Biotechnology

journal homepage:www.elsevier.com/locate/bab

Con fi guration modi fi cation of a submerged membrane reactor for enzymatic hydrolysis of cellulose

Thaothy Nguyenhuynh

^a

, Rajesh Nithyanandam

^a,⁎

, Chien Hwa Chong

^a

, Duduku Krishnaiah

^b

aSchool of Engineering, Taylor's University, Taylor's Lakeside Campus, No. 1 Jalan Taylor's, 47500 Subang Jaya, Selangor DE, Malaysia

bFaculty of Engineering, Universiti Malaysia Sabah, Kota Kinabalu, Sabah 88400, Malaysia

A R T I C L E I N F O

Keywords:

Submerged membrane reactors Enzymatic hydrolysis Intermittent product removal Reusing enzyme

Ultrafiltration

A B S T R A C T

In this study, a modified submerged membrane reactor was investigated as a new configuration for efficient enzymatic cellulose hydrolysis. From the results, effectiveness of ultrafiltration was increased due to the com- plete glucose permeation and enzyme retention up to 80%. Intermittent product removal at 50% volume re- placement was better because it was able to regain the glucose concentration after ultrafiltration and minimised the over-dilution as occurred with the continuous product removal. Results obtained from response surface design showed the quadratic cellulose concentration (A²), the enzyme to substrate ratio (B) and interaction (AC) of cellulose concentration with intermittent product removal were identified to be the significant factors to maximise the glucose concentration. The maximised glucose concentration (7.6 g/L) was obtained at the optimal conditions (10% cellulose concentration, 6% E/S ratio and 50% intermittent product removal). The retained enzyme could be reused to extend hydrolysis beyond 8 h as the glucose concentration was maintained at 7 g/L with an insignificant reduction. Thus, the modified submerged membrane reactor in this new configuration could be an alternative for the conventional batch reactor used in the enzymatic hydrolysis.

1. Introduction

Research and development in bioprocessing of agricultural residues available in abundance such as rice, wheat straws, sugar bagasse and corn stover with 30–50% of cellulose content are continuously im- proved and getting matured. Efficiency of the bioethanol fuel produc- tion mainly relies on the conversion of cellulose to glucose through enzymatic hydrolysis, followed by fermentation and purification of bioethanol product (Abels et al., 2013; Andric et al., 2010a).

The enzymatic hydrolysis of cellulose is a complex reaction system, which involves an interfacial heterogeneity of solid cellulose substrate, cellulase enzyme adsorption, inhibition of glucose and cellobiose on enzyme. Using batch reactors for hydrolysis has disadvantages of having severe product inhibition of the enzyme by glucose and cello- biose and high enzyme cost. These are known as the major challenges making the process far from being technologically and economically feasible. During hydrolysis, the released glucose is accumulated in the batch reactor and becomes an inhibitor of the enzyme (Andric et al., 2010a). Additionally, consumption of cellulase enzyme contributes significantly 20% of the total cost of the entire ethanol process and 50%

of the hydrolysis step, but the enzyme is actually able to be recycled and reused for a longer period of time (Knutsen and Davis, 2002; Tu et al.,

2007). To overcome these challenges, research related to the enzymatic hydrolysis has been progressed in improvement for the pretreatment step, simultaneous saccharification and fermentation (SSF) and devel- oping new enzymes with higher tolerance for glucose inhibition. In- tegrating a membrane separation with the hydrolysis reactor is a better approach. The membrane integration approach with two configurations of membrane reactors, i.e. external loop and submerged membranes, has effectively solved two main issues (Malmali et al., 2015). First, ultrafiltration used in the membrane reactors allows an effective re- moval of glucose from the reactor to the permeate, hence minimises product inhibition. Secondly, enzyme is rejected by the membrane and retained inside the reactor for continuing hydrolysis of cellulose.

However, these two configurations have some limitations such as a low glucose concentration in the permeate, occurrence of membrane fouling, operation at low cellulose concentrations from 2 to 5 w/v %, and enzyme recovery at liquid phase (Andric et al., 2010a;

NguyenHuynh et al., 2017). Notably, research focused on solving these current limitations of the membrane reactor is rather scarce.

By redesigning the submerged membrane reactor, a new config- uration was created and named as a modified submerged membrane reactor to carry out an efficient enzymatic cellulose hydrolysis (Nguyenhuynh and Nithyanandam, 2016). The arrangement of a sealed

http://dx.doi.org/10.1016/j.bcab.2017.08.013

Received 21 March 2017; Received in revised form 28 August 2017; Accepted 28 August 2017

⁎Corresponding author.

E-mail address:chemrajesh1982@gmail.com(R. Nithyanandam).

Available online 01 September 2017

1878-8181/ © 2017 Elsevier Ltd. All rights reserved.

MARK

bag/infuser and the ultrafiltration device (Fig. 1) both immersing in the hydrolysis reactor is novel. In some studies where a complete mixing of solid cellulose, enzyme and buffer was applied for the external loop and submerged membrane reactors, ultrafiltration was interrupted due to the deposit of cellulose on the surface of the membrane, known as membrane fouling, and consequently hydrolysis was affected (Gan et al., 2002; Lozano et al., 2014; Malmali et al., 2015). In addition, the factor of complete mixing for the content of hydrolysis reaction was reported to be insignificant in a wide range from 50 to 1200 rpm on the production of glucose (Gan et al., 2002; Liu et al., 2011). In a study of using dialysis, cellulose and enzyme were contained within the dialysis membrane, submerged in the liquid buffer while the agitator was at the other side of the membrane, the glucose recovery was 94%, glucose formation rate at 64 mg/L h, which were higher than that of 59% and 28 mg/L h compared to the use of ultrafiltration membrane reactor (Andric et al., 2010b). Hence, it proved that the mass transfer was not limited in this case. In the modified submerged membrane reactor as a new configuration in this study, the immersed infuser, which contains cellulose inside, would help to prevent the membrane fouling without affecting the contact and binding of enzyme with cellulose in hydrolysis to produce glucose. Thus, the cellulose-free hydrolysate was easilyfil- trated to remove glucose into the permeate, whereas retained enzyme inside the reactor by the ultrafiltration device.

It was hypothesised that the modified submerged membrane reactor not only can carry out hydrolysis of cellulose, but also increase effec- tiveness of ultrafiltration for product removal and enzyme recovery.

Therefore, the objective of this research was to further investigate the technical feasibility of the modified submerged membrane reactor for efficient enzymatic cellulose hydrolysis. Experiments were conducted for glucose permeation and enzyme retention of ultrafiltration with 10 kDa cutoff. Reactor operation was studied at 2 modes of product removal (intermittent and continuous) together with a variation in cellulose concentrations at 10 w/v % and 20 w/v %. Response surface methodology was applied to study the influence of three factors namely cellulose concentration, enzyme to substrate ratio and volume of pro- duct removal on the enzymatic hydrolysis. With the retained enzyme in the submerged membrane reactor, potential of reusing enzyme was conducted to extend hydrolysis beyond 8 h.

2. Materials and methods

2.1. Materials

Cellic CTec2 enzyme, supplied by Novozymes (China) as free

samples, was a complex blending of aggressive cellulases and high level of β-glucosidases (Novozymes, 2010). Sodium hydroxide (99.0%

purity) was purchased from R & M Chemicals (Essex, UK). Citric acid (99.0% purity) was purchased from Bendosen Laboratory Chemicals (Selangor, Malaysia). Sodium citrate (99.0–100.5% purity) and anhy- drous glucose were purchased from Systerm (Shah Alam, Malaysia).

Cellobiose (99.9% purity) and microcrystalline cellulose (code:

435236) were obtained from Sigma Aldrich (St. Louis, MO, USA).

Polyethersulfone (PES) membrane type 146 was used as afilter for ultrafiltration with 10 kDa molecular weight cutoff, 47 mm diatmeter (Satorius Stedim Biotech GmbH, Germany). Each membranefilter was cut into four equal pieces with a smaller diameter of 23 mm tofit into the ultrafiltration device (Fig. 1).

2.2. Pretreatment of cellulose

Cellulose was weighted using an electronic balance (model:

TX423L, Shimadzu, Japan). The weight of cellulose to be pretreated in this step was based on the cellulose concentration (weight/volume percentage concentration–w/v %, as shown in Eq.(1)), which was applied for the latter enzymatic hydrolysis.

= ×

w v weightof cellulose g volume of reaction mixture mL

/ % ( )

( ) 100

(1) The modified submerged membrane reactor could hold up at afixed volume of 400 mL for the reaction mixture. The cellulose concentra- tions studied in the hydrolysis were at 10 and 20 w/v %. Therefore, the required weights of cellulose were 40 and 80 g respectively. Calculation were as followed.

= ×

=

= ×

=

Weight of cellulose at w v g

Weight of cellulose at w v g

10 / % 400 10

100 40

20 / % 400 20

100 80

The weighted amounts (40 g and 80 g) of cellulose were mixed with 0.5 M aqueous NaOH solution and made up the volume to 400 mL and 800 mL respectively in 1-l beakers. These beakers were then put in a Daniel electronic water bath at 100 °C with regular stirring for every 30 min. During the pretreatment, colour of the mixture changed from yellow to dark brown. After 3 h, the mixtures were cooled to room temperature (25 °C) and washed three times with distilled water to drain out the NaOH residue. The pretreated cellulose was preserved in distilled water at room temperature to be used for hydrolysis (Nguyenhuynh and Nithyanandam, 2016).

Fig. 1.A Schematic of the modified submerged membrane re- actor.

2.3. Design of the modified submerged membrane reactor

The modified submerged membrane reactor as a new configuration was applied for efficient enzymatic hydrolysis of cellulose carried out simultaneously with glucose removal and enzyme recovery. In this novel configuration, the reactor setup (seeFig. 1) consisted of a 500-mL glass Pyrex beaker as a reactor vessel placed on a hot plate (MR-heitec, Heidolph, USA) to control the reactor temperature and stirring speed.

Inside the reactor vessel, pretreated cellulose was contained in the sealed bag (seeFig. 1), which was immersed in the liquid of enzyme and buffer. The sealed bag (size: 9.5 cm wide × 12 cm long) was made of Nylon Monofilament Mesh (NMO) with a porosity of 50 µm/micro- meter and was able to withstand a working temperature up to 148 °C (Shanghai Filterworkshop, China). Glucose as the product of hydrolysis was released from cellulose and infused into the liquid hydrolysate. An ultrafiltration device (Fig. 1) was a PES membrane (diameter 23 mm) attached to a nozzle secured by a ring. A tubing connection between the ultrafiltration device to a container for glucose storage was made using a peristaltic pump (Master Flex L/S, Cole Parmer, USA), which can adjust the flow rate. Citrate buffer was supplemented using another peristaltic pump (Master Flex L/S, Cole Parmer, USA) to the reactor vessel. All the experiments were performed in triplicate.

The system was maintained under magnetic stirring at 200 rpm, 50 °C, 0.1 M citrate buffer pH 5.0 (Novozymes, 2010). At the desired temperature, hydrolysis was started by adding enzyme into the buffer.

After 2–4 h hydrolysis without product removal, the cellulose-free hy- drolysate was ultrafiltrated to remove glucose from the reactor vessel (retentate) to the glucose container (permeate). The total volume in the reactor was maintained at 400 mL by the replenishment of citrate buffer supplement. Although the duration of the enzymatic hydrolysis in the modified submerged membrane reactor was from 8 to 16 h, it was sufficient to observe the changes in the glucose concentrations in the profile and able to evaluate the effects of different factors (product removal modes, volumes of product removal, cellulose concentrations, and enzyme to substrate ratios) on the hydrolysis.

In the enzymatic hydrolysis of cellulose, yield is the amount of glucose liberated per amount of cellulose hydrolysed by enzyme (Garcia-Aparicio et al., 2007; Van Dyk and Pletschke, 2012). Yield of glucose was calculated by using Eq.(2).

= ×

×

Yield G

(%) C0.9

100 (2)

Where, G: mass of glucose (g); C: mass of polysaccharide substrate or cellulose (g).

Mass of glucose (G) was the total amount of the released glucose after the hydrolysis, in the reactor/retentate and the permeate. It was determined by multiplying the final glucose concentrations with the final volumes in the retentate and the permeate. Mass of cellulose (C) was the initial amount of cellulose used in the pretreatment, which was either 40 g or 80 g. Enzyme efficiency/productivity was calculated by the amount of glucose released per amount of enzyme used in the hy- drolysis (Van Dyk and Pletschke, 2012).

2.4. Effectiveness of the ultrafiltration

In the modified submerged membrane reactor as a new configura- tion, the ultrafiltration (10 kDa cutoff) was initially examined for its effectiveness to permeate glucose and retain enzyme at 5 and 10 mL/

minflow rates through the membrane without the involvement of the enzymatic cellulose hydrolysis. The permeation and retention tests were carried out separately by preparing two different solutions (glu- cose solution and enzyme solution).

For glucose permeation, 500 mL of glucose solution was prepared at 22 g/L concentration and then ultrafiltrated for 2 h using the modified submerged membrane reactor. Samples collected in the permeate and retentate were then quantified for glucose concentrations using High

Performance Liquid Chromatography (HPLC).

For enzyme retention, 250 mL of the diluted enzyme solution was prepared at 500 µg/mL concentration by dissolving 1.7 mL of the Cellic CTec2 enzyme with the original protein concentration of 73,600 µg/mL into citrate buffer pH 5.0. The 500 µg/mL enzyme solution was then ultrafiltrated. After 2 h, samples obtained from the permeate and re- tentate were tested using Bradford Assay for protein quantification (Bradford, 1976). Dilution of the original enzyme was calculated fol- lowed Eq.(3).

=

C V_{1 1} C V_{2 2} (3)

Where, C1= 73600 µg/mL and C2= 500 µg/mL were the original, and diluted concentrations of the enzyme respectively. V2= 250 mL was volume of the diluted enzyme solution. Therefore, volume of the ori- ginal enzyme required to be diluted was calculated to be V1= 1.7 mL.

2.5. Operation of the modified submerged membrane reactor

The modified submerged membrane reactor was then used to con- duct the enzymatic hydrolysis of cellulose after the effectiveness of ultrafiltration in this configuration was examined inSection 2.4. Op- eration was studied with respect to different modes of product removal and variation in the concentrations of cellulose. Since the enzymatic hydrolysis of cellulose was carried out in this experiment, the yield of glucose and enzyme productivity were determined.

Two modes were applied for the product removal. First, the reactor vessel was loaded with 10% (w/v) pretreated cellulose (40 g) contained in the sealed bag and immersed in 0.1 M citrate buffer pH 5.0 under magnetic stirring at 50 °C. At 3% (w/w) enzyme to substrate ratio, 1.2 g Cellic CTec2 enzyme was added into the buffer to start the reaction. A 10 mL/min flow rate through ultrafiltration was applied. For con- tinuous product removal, ultrafiltration started after 2-h hydrolysis in batch mode. For intermittent product removal, ultrafiltration started after every 4-h hydrolysis in batch mode by removing intermittently 50% (200 mL) or 100% (400 mL) of the total reaction volume (400 mL).

The reactor vessel was then replenished back to 400 mL with citrate buffer pH 5.0 and continued hydrolysis in batch mode for the next 4 h before the intermittent product removal started again. Samples of 2 mL were collected from the reactor vessel (retentate) and the glucose sto- rage container (permeate) at every hour of hydrolysis and kept in the chiller at 4 °C to stop the reaction. Samples were then quantified for glucose concentration using HPLC.

The effect of variation in the cellulose concentrations on the pro- duction of glucose was studied. The same conditions for the enzymatic hydrolysis were applied (0.1 M citrate buffer pH 5.0 under magnetic stirring at 50 °C, 3% enzyme to substrate (cellulose) ratio in 400 mL total volume). Cellulose concentrations were varied at 10 w/v % (40 g) and 20 w/v % (80 g). The amounts of enzymes required to add into the hydrolysis were 1.2 g and 2.4 g respectively. Ultrafiltration was de- termined based on the results obtained from the study of continuous and intermittent product removal.

2.6. Response surface design

Relationship between factors (independent variables) and the re- sponses (dependent variables) was evaluated using response surface methodology (RSM) as a statistical technique in Design-Expert (Version 7.0.0, Stat-Ease, Inc., USA). With 95% confidence level, determination on the significance of the model and the significant factors was based on the obtainedp-values < 0.05. The goodness offit was evaluated by the coefficientR².

A design summary of the RSM-Box Behnken Design with 3 centre points is presented inTable 1. There are three factors and one response for the optimisation study, i.e. cellulose concentration, enzyme to substrate (E/S) ratio, and volume of product removal and glucose concentration.

2.7. Potential of reusing enzyme

The enzyme being retained within the system by ultrafiltration was further examined for its reusing potential. At the optimal conditions, the enzymatic hydrolysis was carried for 8 h. The retained enzyme in the liquid hydrolysate was then continued to hydrolyse another batch of pretreated cellulose in the following 8 h. At every hour, 2 mL samples taken from retentate and permeate were kept in the chiller at 4 °C to stop the reaction. Samples were then quantified for the glucose con- centration using HPLC.

2.8. Protein quantification using Bradford assay

Protein concentration was determined using the Bradford assay (Bradford, 1976). The protein present in the enzyme solution reacts with the dye Coomassie brilliant blue G-250 to form a protein-dye complex giving a colour indication. The colour mixture of protein-dye complex was measured for its absorbance at 595 nm wavelength using a Thermo Scientific spectrophotometer (GENESYS 10S UV–Vis, USA). A calibration was prepared using the Bovine Serum Albumin (BSA) as protein standard (R² = 0.9962) to determine the concentration of protein in the samples.

2.9. Quantification of glucose and cellobiose using HPLC analysis HPLC-High performance liquid chromatography (serial no.

DACAH00387, 1220 Infinity II LC System, Agilent Technologies, USA) equipped with a reactive index detector RID (1260 Infinity, Agilent Technology, USA) was used to quantify the glucose and cellobiose concentrations independently. A Hi-Plex H column (serial no.

0006169287-188, Agilent Technologies, USA) was used as the sta- tionary phase at 60 °C, 4.6 MPa. The mobile phase was 0.005 M sul- phuric acid in isocratic gradient (0.7 mL/minflow rate). Injection vo- lume was 20 µL. Method set for the HPLC and identification of peaks were based on a similar work in the sugar analysis done byLozano et al.

(2014). The peaks for glucose (8.34 min retention time) and cellobiose

(7.05 min retention time) were identified. The corresponding calibra- tion straight lines were constructed for quantifying the concentrations of glucose and cellobiose with theR²values of 0.99819 and 0.99905 respectively.

2.10. Fourier Transform Infrared Spectroscopy (FTIR) analysis

Sugar products (glucose and/or cellobiose) from the retentate and permeate were detected based on the characteristic infrared adsorption of its functional groups using FTIR-Fourier Transform Infrared Spectroscopy (Spectrum 100 FT-IR, Perkin Elmer, UK). For sugar, two functional groups attached to the carbon chain are hydroxyl (alcohol) group -OH and aldehydes (CH=O). The hydroxyl group absorbance ranged from 3600 to 3100 cm⁻¹, while the aldehyde (C=O) stretch absorbance ranged from 1750 to 1625 (Gurram and Menkhaus, 2014).

2.11. Statistical analysis

One-way analysis of variance (ANOVA) and independent samplet- test were performed to further evaluate the results whether there was a significant difference between groups using IBM SPSS Statistics (Release 20.0.0, USA). At a confident level of 95%,p-value < 0.05 in- dicates a significant difference between groups.

3. Results and discussion

3.1. Effectiveness of the ultrafiltration device

A membrane was integrated into the enzymatic hydrolysis with the purpose of removing glucose and retaining enzyme for minimizing product inhibition and reusing the enzyme respectively. Therefore, in this study the effectiveness of the ultrafiltration for glucose permeation and enzyme retention was tested using the modified submerged mem- brane reactor depicted inFig. 1with a variation in theflow rates at 5 and 10 mL/min. Enzymatic hydrolysis was not involved in this ex- periment. Result of the glucose permeation test is shown inFig. 2a and result of the enzyme retention test is shown inFig. 2b.

As can be seen inFig. 2a, glucose was able to completely permeate through the ultrafiltration since equal concentrations of glucose at ap- proximately 22.5 g/L were determined in the retentate and permeate for both cases (5 and 10 mL/minflow rates through ultrafiltration). In the test for enzyme retention, the ultrafiltration was able to retain from 60% to 80% of protein inside the system (in the retentate) as shown in Fig. 2b. Using independent samplet-test, the percentage of protein in the retentate was significantly higher than that in the permeate (p= 0.000 < 0.05, t= 11.937 at the flow rate of 10 mL/min and p = 0.000 < 0.05,t= 27.620 at theflow rate of 5 mL/min).

The suitability of 10 kDa cutoffultrafiltration used in the membrane reactor for the enzymatic hydrolysis was also confirmed in other studies Table 1

Design summary of RSM.

Factor name Unit Low

actual

High actual

Mean Std. Dev.

A-Cellulose concentration % 10 20 15 3.651

B-Enzyme to substrate ratio (E/S)

% 3 6 4.5 1.095

C-Product removal % 50 100 75 18.257

Response: Glucose concentration (g/L).

Number of runs: 15.

Analysis: Polynomial.

20 21 22 23 24

5 ml/min 10 ml/min

Glucose concentration (g/L)

Flow rate through ultrafiltration Retentate Permeate a)

0 20 40 60 80 100

5 ml/min 10 ml/min

Percentage of protein (%)

Flow rate throught ultrafiltration Retentate Permeate

b) ^{Fig. 2.}Effectiveness of ultrafiltration. a) Glucose

permeation test; b) Enzyme retention test.

(Alfani et al., 1983; Belafi-Bako et al., 2006; Lozano et al., 2014). Based on the molecular weight, sugar products like glucose (180 Da) and cellobiose (350 Da) are able to easily pass through ultrafiltration (10,000 Da or 10 kDa cutoff). However, enzyme components such as cellulases andβ-glucosidases, which have large molecular weight from 35,000 to 65,000 Da, are retained within the system (Cantarel et al., 2009; Lehmann et al., 2012). Hence, the equal glucose concentrations in the retentate and permeate was expected, whereas the percentage of protein in the retentate was higher than in the permeate (Andric et al., 2010a; Lozano et al., 2014).

Another finding in this work is that an increase in the flowrate through ultrafiltration resulted in a negative effect on the enzyme re- tention and perhaps caused an enzyme loss to the permeate. As can be seen inFig. 2b, the percentage of protein retained in the retentate de- creased from 83.04% to 58.86%, corresponding to an increase in the percentage of protein in the permeate from 16.96% to 41.14% when the flow rate was doubled from 5 to 10 mL/min. In a study of Mameri et al., ultrafiltration membrane showed a complete rejection of enzymes since only 1.5% of enzyme activity was found in the permeate. However, a significant drop of 30% in the enzyme activity was found in the re- tentate due to the blocking of enzyme molecules at the membrane surface and resulting concentration polarisation at all tangential velo- city via the membrane (Mameri et al., 2000). In addition, afinding in another similar study is that collection efficiency of enzyme was in- directly proportional to thefiltrationflow rate since increasing theflow rate from 25 to 100 µL/min resulted in a decrease in the enzyme col- lection efficiency (Nguyen et al., 2015).

Thus, the 10 kDa cutoff ultrafiltration used in the modified sub- merged membrane reactor in this study was effective to remove glucose and retain enzyme. And the flow rate through ultrafiltration for the following experiments on the enzymatic hydrolysis was limited at 10 mL/min to minimise enzyme loss into the permeate.

3.2. Operation of the modified submerged membrane reactor 3.2.1. Different modes of product removal

Although ultrafiltration in the hydrolysis is effective to minimise the inhibition for enzyme by removing glucose, the low output concentra- tion of glucose is realised as a big limitation of the membrane reactor (Andric et al., 2010a; Lozano et al., 2014; Malmali et al., 2015). No- tably, operation of the membrane reactor under continuous product removal mode results in a significant reduction in the concentration of glucose. For example, from a study in the cellulase aggregates trapped inside membrane, the concentration of glucose continuously decreased from 0.95 to 0.61 g/L in the continuous hydrolysis (Nguyen et al., 2015). However, hydrolysis in the membrane reactor under inter- mittent product removal mode is able to regain the glucose con- centration as well as the reaction rate after ultrafiltration (Gan et al., 2002; Gavlighi et al., 2013). Therefore, it was hypothesised that modes of product removal in operation of the membrane reactor can affect the concentration of glucose in the enzymatic hydrolysis. In the present study, two product removal modes, i.e. continuous and intermittent (100% and 50% volume replacement) were examined using the mod- ified submerged membrane reactor.

The obtained concentration profiles of glucose in the retentate and permeate during the enzymatic hydrolysis are shown inFig. 3. It can be seen that the membrane reactor operation under intermittent product removal resulted in better profiles of glucose concentration than that under continuous product removal. A shift from hydrolysis in batch mode for thefirst 2 h to hydrolysis under continuous product removal caused a continuous reduction in the concentration of glucose from 3.55 to 0.88 g/L in the retentate and permeate. The decreasing trend in the glucose concentration approaching near to zero resulted from op- erating in continuous hydrolysis using membrane reactors at various flow rates through the membrane was also reported (Nguyen et al., 2015). Considering the case of continuous product removal in this

study, the rate of glucose formation was calculated at only 1.05 g/L h obtained from thefirst 2-h hydrolysis in batch mode. But 10 mL/min (600 mL/h)flow rates were applied for the continuous product removal accompanied with buffer replenishment to maintain the 400-mL reactor volume without interruption. Because of these, the small amount of glucose released from hydrolysis at the rate 1.05 g/L h was over-diluted by the continuous replenishment of buffer and the concentration of glucose kept decreasing instead of regaining like in the intermittent product removal.

Under intermittent product removal, the increasing trend in the glucose concentration was able to recover and reach a peak at ap- proximately 6 g/L in the retentate despite a sharp decrease during the ultrafiltration at the fourth hour. After 6-h hydrolysis, the concentration of glucose obtained under intermittent mode of product removal was around 4.90 g/L which was 5 times higher than that under continuous mode at just 0.89 g/L. Research related to the intermittent product removal in hydrolysis was found in other study in which the conversion of cellulose increased immediately when shifting from batch mode. And thefluctuating graph for the total reducing sugar concentration results from the intermittent product removal (Gan et al., 2002; Gavlighi et al., 2013).

Using the modified submerged membrane reactor with intermittent product removal 50% volume replacement, the required time for the concentration of glucose to reach 7.66 g/L was 8 h. Whereas in other study, it took 40 h for the glucose concentration to reach 10 g/L in the same process using the submerged membrane reactor in which the membrane was arranged in dead-endflow (Gan et al., 2002). Enzyme productivity or specific cellulase efficiency (g glucose/g enzyme) in this experiment for the intermittent 50% and 100% volume were 3.640 and 3.856 g glucose/g enzyme. Comparing with the results obtained from a study of the intermittent product removal byGavlighi et al. (2013)the enzyme productivity is lower at 2.37 g glucose/ g enzyme.

Using ANOVA one-way test, results show that for the product re- moval, the concentrations of glucose under intermittent mode was significantly higher than that under continuous mode (p =

0 2 4 6 8 10

0 1 2 3 4 5 6 7 8 9

Glucose concentration (g/L)

Time (h)

Continuous Intermittent 100% volume Intermittent 50% volume a)

0 2 4 6 8

0 1 2 3 4 5 6 7 8 9

Glucose concentration (g/L)

Time (h)

Continuous Intermittent 100% volume Intermittent 50% volume b)

Fig. 3.Reactor operation in different modes of product removal. a) Glucose concentration profile in the retentate; b) Glucose concentration profile in the permeate.

0.006 < 0.05,F= 6.372). The mean concentration of glucose under continuous mode was only 1.91 g/L, whereas those under intermittent modes in both cases of 100% and 50% replacement volumes) were 4.24 and 4.67 g/L respectively. In further comparison using independentt- test between the intermittent product removal at 100% and 50% vo- lume replacement, the resultant glucose concentrations were insignif- icantly different (p= 0.558 > 0.05,t=−0.550), plus the mean values of concentrations in both cases were very close to each other.

Therefore, operation of the modified membrane reactor under in- termittent product removal (50% volume replacement) was efficient for the enzymatic hydrolysis of cellulose.

3.2.2. Variation in concentrations of cellulose

Cellulose concentration was varied at 10% and 20% (w/v) to study its effect on the enzymatic hydrolysis and production of glucose using the modified submerged membrane reactor under intermittent product removal (50% volume replacement). Fig. 4 shows the glucose con- centration profile in the permeate. The starting time for ultrafiltration in the hydrolysis at 10% and 20% cellulose concentrations was after 4 and 2 h, respectively. In a minor part of this study, it was observed from hydrolysis in batch reactors that at 10% cellulose concentration, rate of glucose formation started to decrease after 4 h, whereas at 20% cellu- lose concentration, rate of glucose formation started to decrease after only 2 h. Therefore, ultrafiltration needed to start earlier for the hy- drolysis at 20% cellulose concentration than that at 10% cellulose concentration.

As can be seen inFig. 4, the concentrations of glucose were quite closed to each other at about 6.5 g/L and 4.5 g/L respectively for the both cases at 10% and 20% cellulose concentrations although ultra- filtration started at different times. Since the pretreated cellulose was contained in the sealed bag/infuser, which was immersed in the liquid buffer, there was no observation in the deposit of cellulose on the membrane surface and no interruption during the hydrolysis and ul- trafiltration separation. Therefore, using the sealed bag/infuser was an effective way to avoid the occurrence of membrane fouling in this modified submerged membrane reactor as a new configuration. In some other studies, the accumulation of lignocellulose containing not only cellulose, but hemicellulose and lignin, these solid particles gradually accumulated on the membrane surface and fouling occurred (Gan et al., 2002; Lozano et al., 2014; Malmali et al., 2015).

Using ANOVA one-way test, the effect of varying cellulose con- centration on the production of glucose (in term of concentrations in the permeate) was not significant (p = 0.253 > 0.05, F = 1.358).

Furthermore, if comparing cellulose concentration in this study (10%) and other study (2.5%), the yields of glucose from hydrolysis of cellu- lose in the membrane reactor are respectively 0.11 and 0.19 g glucose/

g cellulose, which are not significantly different (Gavlighi et al., 2013).

Therefore, cellulose concentration may not be a significant factor in operation of the modified submerged membrane reactor under

intermittent product removal. The effect of the cellulose concentration on the hydrolysis was further proven in the nextSection 3.3 for the response surface study.

3.3. Response surface design

Using response surface methodology to study the influence of dif- ferent factors on the hydrolysis process conducted in the modified submerged membrane reactor, the ANOVA result inTable 2shows that the reduced quadratic model is significant (p = 0.012 < 0.05, F = 8.07) to represent the relationship between its operating factors (A- cellulose concentration, B-enzyme to substrate (E/S) ratio and C-pro- duct removal) and the glucose concentration. The collected data is well- fitted into the model with an insignificant lack offit (p= 0.935 > 0.05, F= 0.18). The studied model is represented by Eq.(4)and the model graphs are shown inFig. 5.

= − + − + +

G 29.175 2.783A 0.665B 0.20C 0.012AC 0.06A² (4) Where, G: glucose concentration (g/L); A: cellulose concentration (w/v

%); B: enzyme to substrate (cellulose) ratio (w/w %); C: product re- moval (% volume).

TheR²(R-squared) is a statistical measure of the goodness offit for the model with the data. About 87% of the variation of the response data (the glucose concentration) was able to explain by the model (R²

= 0.8706). The difference between theR²and adjustedR²= 0.7628 is only 0.1078. With the predictedR²= 0.6292, this model can be able to make predictions on the glucose concentration in the enzymatic hy- drolysis using the modified membrane reactor.

Regarding each individual factor, factor A (p= 0.09 > 0.05) and factor C (p= 0.134 > 0.05) were non-significant factors affecting the production of glucose in the hydrolysis. Previously, the study in the operation of the modified submerged membrane reactor (Section 3.2) showed a variation in the cellulose concentrations from 10% to 20% did not affect the glucose concentration. Two studies in the optimisation of the enzymatic hydrolysis with the range of substrate concentration 2–12.5% and 5–10% (dry biomass/volume) gave the samefinding that substrate concentration, which is cellulose in this case, is not a sig- nificant factor (Fang et al., 2010; Ferreira et al., 2009). The reason for this is due to insufficient mass transfer when biomass concentration increases (Al-Zuhair et al., 2013; Ferreira et al., 2009). Other reason is that an increase in substrate cellulose concentration would result in end-product inhibition caused by the increase concentration of glucose or cellobiose (Fang et al., 2010; Kristensen et al., 2009; Kumar and Wyman, 2008).

Because of the significant quadratic term A²(p= 0.0118) with a positive sign for cellulose concentration in the empirical model re- presented in Eq. (4), the graph is parabolic and have concave upwards as can be seen inFig. 5a. Therefore, at two extreme points where the cellulose concentration is low at 10% and high at 20%, the glucose -1

0 1 2 3 4 5 6 7 8

Glucose concentration (g/L)

Time (h)

10% 20%

0 2 4 6 8 10 12 14

Fig. 4.Effect of varying cellulose concentrations on the glucose concentration in the permeate.

Table 2

ANOVA results from modelling.

Source Sum of

Squares

df Mean F p-value

Square Value Prob >F

Model 25.64 5 5.13 8.07 0.0122 significant

A-Cellulose concentration

2.55 1 2.55 4.02 0.0918

B-E/S ratio 5.96 1 5.96 9.39 0.0221 significant

C-Volume removal 1.9 1 1.9 2.99 0.1344

AC 8.81 1 8.81 13.88 0.0098 significant

A² 8.08 1 8.08 12.73 0.0118 significant

Residual 3.81 6 0.64

Lack of Fit 1.81 5 0.36 0.18 0.9348 insignificant

Pure error 2 1 2

Total 29.45 11

concentrations were maximised in the hydrolysis. In addition, the vo- lume removal (factor C) is not a significant factor (p= 0.1344), but it is inversely proportional to the response glucose concentration due to the negative sign (referring to Eq. (4)). Therefore, the graph inFig. 5b is linear and goes downwards. It implies that an increase in the volume removal minimised the concentration of glucose as a result. When combining both factor A and factor C to see its interaction effect (AC) on the response glucose concentration as shown inFig. 5c (contour view) andFig. 5d (3D surface), the glucose concentration was max- imised to be above 7 g/L at those two points where the cellulose con- centration is low at 10% and high at 20%, given the volume removal was at 50% and 10% respectively. The interaction of factor A with factor C has a significant effect on the glucose concentration (p = 0.0098 < 0.05). An increased cellulose concentration would corre- spondingly require an increase in the volume of the glucose product to be removed intermittently or vice versa.

Factor B-enzyme to substrate ratio is another significant factor (p= 0.022 < 0.05, F = 9.39) affecting the glucose concentration in the hydrolysis. InFig. 6, glucose concentration increases linearly as the E/S ratio increases. In other studies of optimisation of hydrolysis by re- sponse surface method, the same finding was reported about the Fig. 5.Response surface graphs in the enzymatic hydrolysis using the modified submerged membrane reactor: a) Effect of cellulose concentration; b) Effect of volume removal;

Interaction of cellulose concentration and volume removal in c) Contour view and d) 3D surface.

Fig. 6.3D surface showing the effect of varying E/S ratio on glucose concentration.

quantity of cellulase enzyme present in hydrolysis of cellulose as the most significant factor, in term of cellulase dosage (Fang et al., 2010), or cellulase concentration (Ferreira et al., 2009). When increasing the amount of cellulase enzyme available in the hydrolysis reaction, there would be more cellulose being hydrolysed to liberate glucose.

In the optimisation part, the glucose concentration was maximised.

The optimal conditions for these three factors to obtain the corre- sponding highest concentrations of glucose are shown in Table 3. In general, the highest glucose concentration is at approximately 7.5–7.7 g/L at the conditions of cellulose concentration at 10%, E/S ratio at 6% and volume removal at 50%. InTable 3andFig. 7, it can be seen that the predicted and experimental values of glucose concentra- tions obtained from hydrolysis are closed to each other with theR²= 0.8881.

3.4. Potential of reusing enzyme

Since the ultrafiltration in the modified submerged membrane re- actor was effective to retain enzyme, hydrolysis was extended by re- using the retained enzyme for another 8 h. The glucose and cellobiose concentration profiles in the retentate and permeate are shown in Fig. 8. The glucose concentration after 8-h hydrolysis was 6.67 g/L.

In the extended hydrolysis from 8 h to 16 h, the concentrations of glucose in the retentate in overall are lower than that in thefirst 8 h (Fig. 8a). The conversion of cellulose to glucose decreased when en- zyme was reused to hydrolyse fresh pretreated cellulose. However, in the permeate of the ultrafiltration (Fig. 8b) glucose concentrations during 16 h hydrolysis was able to maintain slightly below 7 g/L with a slight reduction. In a similar study, the concentration of glucose in the permeate was unchanged at approximately 115 mM during 9 operation cycles of hydrolysis when the enzyme was retained in the membrane reactor (Lozano et al., 2014). Therefore, the enzyme which was re- tained in the system by ultrafiltration was potential to be reused for extending hydrolysis.

Cellobiose which is an intermediate product of the conversion from cellulose to glucose, has very low concentrations of less than 0.5 g/L and nearly approaches to zero during the entire 16 h hydrolysis (Fig. 8).

According toLozano et al. (2014)the enzyme components (cellulases andβ-glucosidases) work synergistically to quickly convert cellobiose to glucose because the cellobiose concentration was only observed in the beginning of the reaction. And dosage of enzyme Cellic CTec2 used for the hydrolysis reactor system was suggested to be in the range of

1.5–30% (w/w) (Novozymes, 2010). Hence, the use of enzyme Cellic CTec2 with its dosage of 6% used for the hydrolysis was suitable and sufficient for a quick conversion of cellobiose to thefinal glucose pro- duct.

3.5. FTIR analysis of sugars

Hydrolysis in the permeate of ultrafiltration after the enzymatic hydrolysis was further analysed using the Fourier Transform Infrared Spectroscopy (FTIR) to determine its prominent functional groups of the compounds. Samples contained mainly glucose as analysis of cel- lobiose–the intermediate sugar by HPLC is very low at less than g 0.5 g/

L. Glucose is a simple type of sugar and have a chemical formula C6H12O6, which consists offive -OH (hydroxyl) groups and one CH=O (aldehyde) group attached to the carbon chain. The FTIR absorption spectra of the hydrolysate is shown inFig. 9.

Identification of the functional groups with peaks present in the product was referenced to the other study (Gurram and Menkhaus, 2014). InFig. 9, detection of -OH (hydroxyl) group and CH=O (alde- hyde) group present in glucose is at the two peaks with the wave number of 3398.60 cm⁻¹and 1641.97 cm⁻¹respectively. The highest peak identified for the hydroxyl group shows that glucose is the main component in the product stream. This is validated by studying of the glucose concentration in the permeate.

4. Conclusions

The modified submerged membrane reactor in a new configuration was proved to be technically feasible to conduct the enzymatic hydro- lysis of cellulose. The ultrafiltration in this configuration has a good permeation of glucose, and high retention of enzyme up to 80% in the retentate. For operation of the membrane reactor, the intermittent product removal (50% volume replacement) was better because the glucose concentration was able to regain after ultrafiltration. Whereas the continuous product removal caused over-dilution of glucose in the Table 3

Optimal conditions for hydrolysis in the membrane reactor to maximise glucose con- centration.

Conditions of factors Solution 1 Solution 2 Solution 3 A-cellulose concentration (% w/v) 10.1 10.46 10.64

B-E/S ratio (% w/w) 5.67 5.87 5.99

C-volume removal (%) 51.32 50.16 50.96

Predicted glucose concentration (g/L) 7.755 7.698 7.581 Experimental glucose concentration (g/

L)

7.623 7.638 6.307

y = 8.1069x -55.056 R² = 0.8881 5

5.5 6 6.5 7 7.5 8

7.55 7.6 7.65 7.7 7.75 7.8

Experimental values

Predicted values

Fig. 7.Experimental vs. predicted values of glucose concentration in optimisation.

0 2 4 6 8

0 5 10 15 20

Glucose concentration (g/L)

Time (h)

First 8 hours of hydrolysis (glucose) Extending hydrolysis to reuse enzymes

a)

0 2 4 6 8

0 2 4 6 8 10 12 14 16 18

Glucose concentration (g/L)L)

Time (h)

Glucose Cellobiose b)

Fig. 8.Reusing enzyme in the hydrolysis-Glucose concentration profile: a) in the re- tentate and b) in the permeate.

permeate. The results from response surface show that the quadratic cellulose concentration (A²), enzyme to substrate rato (B) and interac- tion of cellulose concentration with volume removal (AC) are the sig- nificant factors in the hydrolysis to maximise the glucose concentration.

At the optimal conditions (10% cellulose concentration, 6% E/S ratio and 50% volume removal), the maximised glucose concentration was obtained at 7.6 g/L, which was closed to the predicted value. With a high retention of enzyme, reusing enzyme was possible to extend the enzymatic hydrolysis beyond 8 h with a new batch of pretreated cel- lulose. Hence, as an alternative solution for the batch reactor, the modified submerged membrane reactor could be highly recommended as a new configuration for efficient enzymatic hydrolysis of cellulose.

Acknowledgements

Novozymes (China) Investment Co. Ltd was acknowledged for generously giving us free samples of Cellic CTec2 enzyme as a strong support for our research work.

References

Abels, C., Carstensen, F., Wessling, M., 2013. Membrane processes in biorefinery appli- cations. J. Memb. Sci. 444, 285–317.http://dx.doi.org/10.1016/j.memsci.2013.05.

030.

Al-Zuhair, S., Al-Hosany, M., Zooba, Y., Al-Hammadi, A., Al-Kaabi, S., 2013. Development of a membrane bioreactor for enzymatic hydrolysis of cellulose. Renew. Energy 56, 85–89.http://dx.doi.org/10.1016/j.renene.2012.09.044.

Alfani, F., Cantarella, M., Scardi, V., 1983. Use of a membrane reactor for studying en- zymatic hydrolysis of cellulose. J. Memb. Sci. 16, 407–416.

Andric, P., Meyer, A.S., Jensen, P.A., Dam-Johansen, K., 2010a. Reactor design for minimizing product inhibition during enzymatic lignocellulose hydrolysis. II.

Quantification of inhibition and suitability of membrane reactors. Biotechnol. Adv.

28, 407–425.http://dx.doi.org/10.1016/j.biotechadv.2010.02.005.

Andric, P., Meyer, A.S., Jensen, P.A., Dam-Johansen, K., 2010b. Effect and modeling of glucose inhibition and in situ glucose removal during enzymatic hydrolysis of pre- treated wheat straw. Appl. Biochem. Biotechnol. 160, 280–297.http://dx.doi.org/10.

1007/s12010-008-8512-9.

Belafi-Bako, K., Koutinas, A., Nemestthy, N., Gubicza, L., Webb, C., 2006. Continuous enzymatic cellulose hydrolysis in a tubular membrane bioreactor. Enzyme Microb.

Technol. 38, 155–161.http://dx.doi.org/10.1016/j.enzmictec.2005.05.012.

Bradford, M.M., 1976. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248–254.

Cantarel, B.L., Coutinho, P.M., Rancurel, C., Bernard, T., Lombard, V., Henrissat, B., 2009.

The carbohydrate-active enzymes database (CAZy): an expert resource for glycoge- nomics. Nucleic Acids Resour. 37, 233–238.

Fang, H., Zhao, C., Song, X.Y., 2010. Optimization of enzymatic hydrolysis of steam- exploded corn stover by two approaches: response surface methodology or using cellulase from mixed cultures of Trichoderma reesei RUT-C30 and Aspergillus niger NL02. Bioresour. Technol. 101, 4111–4119.http://dx.doi.org/10.1016/j.biortech.

2010.01.078.

Ferreira, S., Duarte, A.P., Ribeiro, M.H.L., Queiroz, J.A., Domingues, F.C., 2009. Response surface optimization of enzymatic hydrolysis ofCistus ladanifer and Cytisus striatusfor bioethanol production. Biochem. Eng. J. 45, 192–200.http://dx.doi.org/10.1016/j.

bej.2009.03.012.

Gan, Q., Allen, S.J., Taylor, G., 2002. Design and operation of an integrated membrane reactor for enzymatic cellulose hydrolysis. Biochem. Eng. J. 12, 223–229.http://dx.

doi.org/10.1016/S1369-703X(02)00072-4.

Garcia-Aparicio, M.P., Ballesteros, M., Manzanares, P., Ballesteros, I., Gonzalez, A., Negro, M.J., 2007. Xylanase conriution to the efficiency of cellulose enzymatic hy- drolysis of barley straw. Appl. Biochem. Biotechnol. 136–140, 353–365.

Gavlighi, H.A., Meyer, A.S., Mikkelsen, J.D., 2013. Enhanced enzymatic cellulose de- gradation by cellobiohydrolases via product removal. Biotechnol. Lett. 35, 205–212.

http://dx.doi.org/10.1007/s10529-012-1067-4.

Gurram, R.N., Menkhaus, T.J., 2014. Continuous enzymatic hydrolysis of lignocellulosic biomass with simultaneous detoxification and enzyme recovery. Appl. Biochem.

Biotechnol. 173, 1319–1335.http://dx.doi.org/10.1007/s12010-014-0873-7.

Knutsen, J., Davis, R., 2002. Combined sedimentation andfiltration process for cellulase recovery during hydrolysis of lignocellulosic biomass. Appl. Biochem. Biotechnol. 98, 1161–1172.

Kristensen, J.B., Felby, C., Jørgensen, H., 2009. Yield-determining factors in high-solids enzymatic hydrolysis of lignocellulose. Biotechnol. Biofuels 2, 11.http://dx.doi.org/

10.1186/1754-6834-2-11.

Kumar, R., Wyman, C.E., 2008. An improved method to directly estimate cellulase ad- sorption on biomass solids. Enzyme Microb. Technol. 42, 426–433.http://dx.doi.

org/10.1016/j.enzmictec.2007.12.005.

Lehmann, C., Sibilla, F., Maugeri, Z., Streit, R.W., Dominguaze de Maria, P., Martinez, R., Schanaberg, U., 2012. Reengineering CelA2 cellulase for hydrolysis in aqueous so- lutions of deep eutectic solvents and concentrated seawater. Green Chem. 14, 2719–2726.http://dx.doi.org/10.1039/C2GC35790A.

Liu, J., Lu, J., Cui, Z., 2011. Enzymatic hydrolysis of cellulose in a membrane bioreactor:

assessment of operating conditions. Bioprocess Biosyst. Eng. 34, 525–532.http://dx.

doi.org/10.1007/s00449-010-0501-z.

Lozano, P., Bernal, B., Jara, A.G., Belleville, M.P., 2014. Enzymatic membrane reactor for full saccharification of ionic liquid-pretreated microcrystalline cellulose. Bioresour.

Technol. 151, 159–165.http://dx.doi.org/10.1016/j.biortech.2013.10.067.

Malmali, M., Stickel, J., Wickramasinghe, S.R., 2015. Investigation of a submerged membrane reactor for continuous biomass hydrolysis. Food Bioprod. Process. 96, 189–197.http://dx.doi.org/10.1016/j.fbp.2015.07.001.

Mameri, N., Hamdache, F., Abdi, N., Belhocine, D., Grib, H., Lounici, H., Piron, D.L., 2000. Enzymatic saccharification of olive mill solid residue in a membrane reactor. J.

Memb. Sci. 178, 121–130.http://dx.doi.org/10.1016/S0376-7388(00)00489-0.

Nguyen, L.T., Neo, K.R.S., Yang, K.L., 2015. Continuous hydrolysis of carboxymethyl cellulose with cellulase aggregates trapped inside membranes. Enzyme Microb.

Technol. 78, 34–39.http://dx.doi.org/10.1016/j.enzmictec.2015.06.005.

Nguyenhuynh, T., Nithyanandam, R., 2016. Fractionation of hydrolysed microcrystalline cellulose by ultrafiltration membrane. J. Eng. Sci. Technol. 11, 136–148.

NguyenHuynh, T., Nithyanandam, R., Chong, C.H., Krishnaiah, D., 2017. A review on using membrane reactors in enzymatic hydrolysis of cellulose. J. Eng. Sci. Technol.

12, 1129–1152.

Novozymes, 2010. Cellic®CTec2 and HTec2 - Enzymes for hydrolysis of lignocellulosic materials 1–9.〈http://dx.doi.org/2010-01668-01〉.

Tu, M., Chandra, R., Saddler, J.N., 2007. Evaluating the distribution of cellulases and the recycling of free cellulases during the ydrolysis of lignocellulosic substrate.

Biotechnol. Prog. 23, 398–406.

Van Dyk, J.S., Pletschke, B.I., 2012. A review of lignocellulose bioconversion using en- zymatic hydrolysis and synergistic cooperation between enzymes-Factors affecting enzymes, conversion and synergy. Biotechnol. Adv. 30, 1458–1480.http://dx.doi.

org/10.1016/j.biotechadv.2012.03.002.

Fig. 9.FTIR absorption spectra of the hydrolysate.