Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions

G Model

ARTICLE IN PRESS

INDCRO-9268; No. of Pages 11

Industrial Crops and Products xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Industrial Crops and Products journal homepage: www.elsevier.com/locate/indcrop

Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions Heloisa Tibolla a,∗ , Franciele M. Pelissari b , Maria I. Rodrigues a , Florencia C. Menegalli a a b

Department of Food Engineering, School of Food Engineering, University of Campinas, Campinas, SP, CEP 13083-862, Brazil Institute of Science and Technology, Food Engineering, University of Jequitinhonha and Mucuri, Diamantina, MG, CEP 39100-000, Brazil

a r t i c l e

i n f o

Article history: Received 5 August 2016 Received in revised form 6 October 2016 Accepted 17 November 2016 Available online xxx Keywords: Banana peel waste Delignification Xylanase Enzymatic hydrolysis Cellulose nanofibers

a b s t r a c t Cellulose nanofibers (CNFs) were isolated from banana peel bran via alkaline treatment followed by enzymatic treatment with xylanase. The influence of process conditions such as pH, temperature, and concentrations of the enzyme and substrate on the properties of the CNFs was evaluated with a 24−1 fractional factorial design with three central points. Enzyme at 70 U/g of bran, substrate at 15%, pH 6.0, and temperature between 35 and 55 ◦ C favored enzymatic hydrolysis. Transmission electron microscopy (TEM) images confirmed that treatment with xylanase effectively isolated cellulose fibers at the nanometer scale. Fourier transform infrared spectroscopy (FTIR) showed that a fraction of amorphous compounds was removed. X-ray diffraction revealed that the CNFs presented high crystallinity index (66.2%). The CNFs had a diameter of 3.7 nm, their aspect ratio was in the range of long nanofibers, and their suspension was stable (−29.1 mV). These features make the CNFs potentially applicable as reinforcing agents in composites. The results evidenced that enzymatic hydrolysis with xylanase successfully afforded CNFs from banana peel, a residue that constitutes a potential source of biodegradable materials of commercial interest. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Cellulose, the main component of the cell walls of plant fibers, has been extensively explored because it resembles synthetic polymers with the advantage that it originates from natural, renewable, and biodegradable resources. Cellulose is an ideal material to produce nanoparticles for use as reinforcing agent in composite materials. It presents good mechanical strength and stiffness, interesting thermal and electrical properties, and high degree of crystallinity (Bhattacharya et al., 2008; Cherian et al., 2008; Deepa et al., 2011; Siqueira et al., 2010a). Recently, interest in obtaining nanometric cellulose fibers from natural sources has increased. Bananas are a popular fruit that grows in tropical and subtropical regions (Pelissari et al., 2014). Cultivation and industrialization of banana fruit generates a considerable amount of waste with high lignocellulosic content. One example of such waste is banana peel, a byproduct of banana processing during food production (Elanthikkal et al., 2010). Banana peel is a source of cellulose. Banana peel processing not only adds value to this byproduct, but

∗ Corresponding author. E-mail addresses: [email protected], [email protected] (H. Tibolla).

it also helps to reduce the environmental impact of this waste (Rosa et al., 2010). Molina (2013) has suggested the integral use of banana: its peel could be used to produce nanofibers that could be introduced as reinforcing agents in films produced from the banana pulp. A series of processes are necessary to isolate cellulose nanofibers (CNFs). There are many ways to extract CNFs, all of which lead to different types of fibrillar material with characteristics that will depend on the raw material (cellulose), pretreatment, and disintegration process (Chen et al., 2011). In general, plant materials are lignocellulosic, which makes them resistant to bioconversion and requires pretreatment to increase their digestibility and render cellulose more accessible for hydrolysis. Chemical treatment can remove the amorphous fractions (hemicellulose and lignin) present in the structure of a plant fiber. Alkali treatment causes the structure to swell, modifying the physical features of the fiber wall and consequently increasing the surface area that is exposed to hydrolysis in the cellulose fibers (Andrade-Mahecha et al., 2015; Castro and Pereira Jr, 2010). Different techniques afford cellulose nanoparticles from plant sources. CNFs are commonly prepared by chemical treatment, but new techniques to isolate CNFs are currently being developed. Enzymatic hydrolysis can help to isolate cellulose fibers from plant cell walls. Because enzymatic hydrolysis dismisses the

http://dx.doi.org/10.1016/j.indcrop.2016.11.035 0926-6690/© 2016 Elsevier B.V. All rights reserved.

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035

G Model INDCRO-9268; No. of Pages 11

ARTICLE IN PRESS

2

H. Tibolla et al. / Industrial Crops and Products xxx (2016) xxx–xxx

need for solvents and chemicals, the mild conditions of this process make it economically attractive and environmentally friendly (Meyabadi and Dadashian, 2012; Siqueira et al., 2010a; Yu et al., 2008). Xylanases are usually employed in enzymatic hydrolysis. These enzymes initially promote catalytic hydrolysis of the hemicellulose fractions present in the plant fiber. Then, they attack the glycosidic bonds ␤-1,4 located between the glucose units comprising cellulose, which culminates in hydrolytic cleavage. Hydrolysis usually produces CNFs in colloidal suspensions (Hubbe et al., 2008; Pääkko et al., 2007). To develop a new technique to isolate CNFs with sustainable characteristics, Tibolla et al. (2014) studied the production of CNFs from banana peel bran by enzymatic hydrolysis at fixed conditions of pH (5.5), temperature (45 ◦ C), and concentrations of the substrate (25%, w/v) and enzyme (50 U/g of bran). The authors compared their results with results obtained by acid hydrolysis. Enzymatic hydrolysis proved to be a very promising technique to prepare CNFs: the resulting nanofibers were longer, and they had smaller diameter and greater aspect ratio. In addition, the CNFs presented higher negative surface charge, which is important to prevent the nanofibers from agglomerating. Previous studies have shown that CNFs obtained by enzymatic hydrolysis have potential application as reinforcing agents in composites. However, the efficiency of enzymatic hydrolysis depends on factors such as the hydrolysis time (h), the concentrations of substrate (%) and enzyme (U/g of bran), pH, and temperature. These factors often interact with one another, so it is important to optimize the hydrolysis process to improve its yield (Meyabadi and Dadashian, 2012). Here, experiments were performed with a fractional factorial design 24−1 with three central points. This study aimed to analyze how process conditions (pH, temperature, and concentrations of enzyme and substrate) employed during the enzymatic treatment of unripe banana peels of the variety “Terra” (Musa paradisiaca) influenced the properties of CNFs produced via hydrolysis by xylanase.

2.3. Production of cellulose nanofibers (CNFs) Producing cellulose nanofibers is complex and depends on numerous factors, so a statistical study that considered different process conditions was conducted herein. On the basis of the paper by Rodrigues and Iemma (2014), the experiments were performed by employing a 24−1 fractional factorial design with three central points as represented by the experiment matrix shown in Table 1. The independent variables were temperature (T), pH, and concentrations of the enzyme [E] and substrate [S]; the analyzed response variables were length, diameter, aspect ratio, zeta potential, yield, and crystallinity. The ranges of the variables had been defined in preliminary tests. Analysis of the effects of the fractional factorial design on all the evaluated responses allowed to establish conditions for the validation tests that would provide CNFs with maximum length, minimum diameter, and higher aspect ratio, zeta potential, yield, and crystallinity. Enzymatic hydrolysis was conducted according to the method adapted from Tibolla et al. (2014). Erlenmeyer flasks containing the substrate (i.e., banana peel bran at concentrations of 15, 25, or 35%) and 0.1 M acetate buffer (pH of 4.0, 5.0, or 6.0) were placed in the thermostatic shaker (temperature of 35, 45, or 55 ◦ C) for 10 min, for the medium to adapt. Then, the enzyme xylanase (concentration of 30, 50, or 70 U/g of bran) was added to the mixture and left at the desired temperature for 24 h, under agitation (150 rpm). The suspensions were placed in a thermostatic bath at 80 ◦ C for 30 min, to denature the enzyme. Next, the residual pulp was washed with deionized water, and the solid was separated by centrifugation (10,000 rpm, 5 ◦ C, 15 min) and suspended in deionized water. At the end of these procedures, a colloidal suspension of CNFs was obtained and stored at 4 ◦ C in a sealed container. The effects of the independent variables were evaluated at 10% significance (Rodrigues and Costa, 2014) by using the software Protimiza Experimental Design Statistical. In the case of fractional factorial design and biological processes, which are complex procedures, it is better to accept p < 0.1 than leave out an important factor. In this case, a validation test was required.

2. Materials and methods 2.1. Materials The banana peel bran was prepared from unripe banana peels (mature green) of the variety “Terra” (Musa paradisiaca), according to the methodology described by Pelissari et al. (2012). The fruit was obtained from the southeastern region of Brazil; the crop was harvested in March 2013, but it was not subjected to any postharvest treatment. All the chemicals used in this work were reagent grade. Xylanase enzyme, kindly provided by Novozymes (Araucária – PR, Brazil), was used to produce CNFs by enzymatic hydrolysis.

2.2. Pretreatment of the banana peel bran Plant materials contain a large amount of amorphous compounds, so it was necessary to delignify the bran. An alkaline treatment was conducted according to the method described by Tibolla et al. (2014). This process was performed in 5% w/v KOH alkaline solution at a bran/solution ratio of 1:20. Vigorous stirring and room temperature were employed for 14 h. Then, the substrate was subjected to successive washings with deionized water and centrifuged after each washing (10,000 rpm, 5 ◦ C, 15 min). This process removed hemicellulose and lignin, to improve the next step of enzymatic hydrolysis. Fig. 1a depicts the sequence of steps used to obtain the unripe banana peel bran. Fig. 1b shows the bran delignification process (alkaline treatment) and the resulting residue applied during enzymatic hydrolysis.

2.4. Characterization For the XRD and FTIR analyses, an amount of each suspension (suspension from alkaline treatment (ATS) and suspension of CNFs) was dried in a freeze-dryer (Equipamentos Terroni, model LS 3000, São Paulo, Brazil). The freeze-dried samples were stored at 4 ◦ C in sealed containers.

2.4.1. Physicochemical and size particle analysis of the bran The chemical composition of the bran was determined in terms of the ash, total extractives (polysaccharides including hemicelluloses), lignin, and cellulose contents. The ash content was obtained by using AOAC (AOAC, 2005). The total extractives content was obtained by digestion with water and alcohol according to NREL/TP 520-42619 (Sluiter et al., 2008b). The lignin content was determined by digestion with sulfuric acid (72% w/w) in combination with high pressure at 121 ◦ C according to NREL/TP 520-42618 (Sluiter et al., 2008a). The cellulose content was obtained by digestion with acetic acid (80% w/w) and nitric acid (70% w/w) (Sun et al., 2004). A laser diffraction analyzer (Laser Scattering Spectrometer Mastersizer S, model MAM 5005–Malvern Instruments Ltd., Surrey, England) was used to determine the particle size of the banana peel bran; ethanol was used as solvent. An ultrasound device was coupled to the equipment to increase the dispersion of the sample. Measurements were performed at 25 ◦ C, in triplicate.

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035

G Model

ARTICLE IN PRESS

INDCRO-9268; No. of Pages 11

H. Tibolla et al. / Industrial Crops and Products xxx (2016) xxx–xxx

3

Fig. 1. Photographs of the steps followed to (a) obtain the bran and (b) to delignify the bran via alkaline treatment.

2.4.2. Scanning electron microscopy (SEM) The microstructure of the untreated banana peel bran (UB) and of the bran submitted to alkaline treatment with 5% KOH solution (ATB) was analyzed by SEM. The samples were fixed on aluminum stubs with double-sided tape and coated with a gold layer (Sputter Coater POLARON, model SCD050), to improve conductivity. The coated samples were viewed under a scanning electron microscope

(JEOL, model JSM-5800LV, Tokyo, Japan) operating at an acceleration voltage of 10 kV. 2.4.3. Transmission electron (TEM) and atomic force microscopy (AFM) The structure and morphology of the CNFs were analyzed by microscopy methods. To perform the TEM analysis, a suspension of

Table 1 Experiments matrix of 24−1 fractional factorial design with three center points and the respective results from CNFs characterization. Experiment

(X1 )a

(X2 )b

(X3 )c

(X4 )d

Length (L) (nm)

Diameter (d) (nm)

Aspect ratio (L/d)

Zeta potential (mV)

Yield (%)

Crystallinity index (%)

1 2 3 4 5 6 7 8 9 10 11

−1 (35) +1 (55) −1 (35) +1 (55) −1 (35) +1 (55) −1 (35) +1 (55) 0 (45) 0 (45) 0 (45)

−1 (4.0) −1 (4.0) +1 (6.0) +1 (6.0) −1 (4.0) −1 (4.0) +1 (6.0) +1 (6.0) 0 (5.0) 0 (5.0) 0 (5.0)

−1 (30) −1 (30) −1 (30) −1 (30) +1 (70) +1 (70) +1 (70) +1 (70) 0 (50) 0 (50) 0 (50)

−1 (15) +1 (35) +1 (35) −1 (15) +1 (35) −1 (15) −1 (15) +1 (35) 0 (25) 0 (25) 0 (25)

3091 1620 905 615 1124 1838 1820 3633 2192 2244 2689

11.2 15.8 9.4 7.4 13.9 6.5 5.2 5.2 7.6 7.6 7.6

276.6 102.8 96.6 82.7 80.9 282.5 353.2 694.3 288.4 295.2 353.9

−29.5 −21.2 −27.2 −25.5 −26.8 −26.5 −27.0 −22.8 −28.4 −26.7 −25.7

60.0 94.0 93.0 92.0 67.0 83.0 83.0 91.0 85.0 97.0 90.0

48.5 58.3 55.4 60.0 59.7 61.0 58.5 54.3 50.7 53.2 49.2

a b c d

X1: Temperature (◦ C). X2: pH. X3: [E] (U/g), where U is defined as the mL of xylose released per minute per mL of enzyme. X4: [S] (%); (L/d): length/diameter.

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035

G Model INDCRO-9268; No. of Pages 11

ARTICLE IN PRESS

4

H. Tibolla et al. / Industrial Crops and Products xxx (2016) xxx–xxx

CNFs was treated with ultrasound for 5 min, to separate agglomerated fibers. A drop of the suspension was placed on a carbon and parlodion (300 mesh) microgrid and left to rest for 60 s. Next, the microgrid was washed twice with a drop of deionized water, and a drop of uranyl acetate (2%) was deposited on the microgrid. The TEM images were obtained by using a transmission electron microscope TEM-MSC (JEOL 2100) equipped with a LaB6 electron gun, at an accelerating voltage of 200 kV. The morphology and diameter of the CNFs were determined by analyzing the images with the ImageJ software. To perform AFM analysis, 1.5 ␮L of the suspension of CNFs was placed on a grid with mica surface and dried at room temperature. The images were obtained with a microscope (Park systems, model Nx-10, Suwon, Korea) equipped with a camera, under controlled parameters (relative humidity = 10% UR and temperature = 25 ◦ C). TEM and AFM analyses were performed at the Laboratory of Electron Microscopy (LME) and at the Laboratory for Surface Science (LCS), respectively, of the National Nanotechnology Laboratory (LNNano) (Campinas, Brazil).

2.4.4. Dynamic light scattering (DLS) To determine the surface charge and to estimate the length of the CNFs in aqueous suspension, DLS was measured on a Zetasizer (Malvern Instruments Ltd., Zetasizer Nano Series—model Nano ZS, UK) at room temperature (25 ◦ C). Six length and zeta potential values were obtained, and the results are presented as average values.

2.4.5. Yield of CNFs prepared by enzymatic hydrolysis The yield of CNFs obtained by enzymatic hydrolysis was determined in triplicate. The suspensions of CNFs were previously homogenized by mechanical stirring for 30 min. Then, 2 g of each sample was dried at 105 ◦ C for 24 h. The yield was calculated from the difference between the initial mass and the final mass (on dry basis) of the bran sample treated with alkaline solution. The yield was calculated according to Eq. (1):

Yield =

g nanofiber (d.b) g bran pre − treated (d.b)

(1)

2.4.6. X-Ray diffraction (XRD) The crystallinity of the freeze-dried sample was determined by XRD according to Van Soest et al. (1996). An X-ray diffractometer (Siemens, model D5005, Baden-Wurttemberg, Germany) operating at a voltage of 40 kV and a current of 30 mA was employed; the target was Cu. The crystallinity index (ICr %) of the bran and CNFs was calculated by using Eq. (2), following the method proposed by Segal et al. (1959). In this method, ICr was calculated as the ratio of heights between the maximum intensity of the crystalline peak close to 2␪ = 22◦ (I200 ) and the intensity of the non-crystalline material diffraction peak close to 2␪ = 18◦ (Inon-cr ). Icr =

I200 − Inon−cr X100 I200

(2)

2.4.7. Fourier-transform infrared spectroscopy (FTIR) The functional groups were analyzed by absorption spectroscopy in the infrared region (4000–650 cm−1 ) with a resolution of 4 cm−1 and 16 scans (Vicentini et al., 2005). An infrared Fouriertransform spectrometer (Perkin Elmer, model Spectrum One, Ohio, USA) equipped with a UATR (universal attenuator total reflectance) accessory was used.

3. Results and discussion 3.1. Characterization 3.1.1. Chemical composition of the bran The chemical composition (dry basis) of the unripe banana peel bran was 9.6% of total extractives, 0.01% of ash, 74.9% of polysaccharides including hemicellulose, 7.9% of total lignin, and 7.5% of cellulose. The pretreatment should remove the hemicellulose and lignin fractions, to facilitate the enzymatic attack at native cellulose and the isolation of nanoparticles. Alkaline treatment of the bran afforded 16.5% yield of the treated sample relative to the initial amount of banana bran (dry basis). This result revealed that the bran contained a high amount of non-cellulosic components. 3.1.2. Influence of process conditions on the properties of CNFs The enzymatic treatment of CNFs isolated from banana peel bran previously treated with alkaline solution gave a yield of about 75%. Table 1 shows the length and diameter size (nm), aspect ratio, zeta potential (mV), yield (gCNFs /gtreatedbran ), and crystallinity index of the CNFs obtained after each experiment. In general, all the parameters determined for the CNFs originating from enzymatic hydrolysis showed that the fibers could be applied as reinforcing agent in composites. The data resembled the results achieved for other CNFs arising from enzymatic hydrolysis of raw materials from different sources, such as soft wood pulp, bagasse pulp, and cotton (Hassan et al., 2010; Henriksson et al., 2007; Pääkko et al., 2007; Satyamurthy et al., 2011). The following CNF dimensions were characterized: length (L), diameter (d), and aspect ratio (L\d). When nanofibers are produced, the main goal is to obtain nanoparticles measuring less than 100 nm because smaller diameter, larger length, and higher zeta potential are better for the application of nanofibers, especially in composites. In general, enzymatic hydrolysis gave CNFs with average diameter of 8.8 nm and length between 615.0 and 3633.3 nm. Therefore, the environmentally friendly treatment of materials from plant sources with xylanase enzymes afforded CNFs with smaller size than the CNFs obtained by chemical treatment of banana peel bran (10.9 nm) (Tibolla et al., 2014), bagasse pulp (9–23 nm) (Hassan et al., 2011), and rice straw (10–65 nm) (Lu and Hsieh, 2012). Table 2 summarizes the estimated effects of the independent variables on the length, diameter, aspect ratio, zeta potential, yield, and crystallinity index of CNFs. A level of significance of 10% (p < 0.1) and MSE (mean squared error) were considered for all the analyses. Rodrigues and Iemma (2014) have recommended that, if the central points are evaluated, the designs planned to select variables should always analyze the effects of the variables by considering the analysis of curvature. In the present study, the curvature was considered because this tool minimizes the probability of making an incorrect decision in further stages of the project. The length of the cellulosic filaments obtained after alkaline treatment of the banana peel bran ranged from 5000 to 7000 nm, with average length of 6588 nm. The results evidenced that enzymatic hydrolysis of these cellulosic filaments reduced the length of the fibers at least to half the initial value. Independent variables in the levels studied during the fractional factorial design did not influence the response variable length significantly for a significance level of 10% (p < 0.1). However, all the process conditions at the higher levels of study provided the highest fiber length value (experiment number 8, fiber length = 3633 nm). Statistical analysis revealed that the variables pH, [S], and [E] affected particle size significantly (p < 0.1). The pH of the process influenced this dependent variable the most, with a linear negative effect. When the pH was increased from level −1 (4.0) to +1 (6.0), particle size diminished the most, probably because xylanase was

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035

G Model

ARTICLE IN PRESS

INDCRO-9268; No. of Pages 11

H. Tibolla et al. / Industrial Crops and Products xxx (2016) xxx–xxx

5

Table 2 Estimated effects and p-values of the variables pH, temperature (T), enzyme concentration ([E]), and substrate concentration ([S]) on the response variables length, diameter, aspect ratio, zeta potential, yield, and crystallinity index. Factor

Mean Curvature T pH [E] [S]

Length (nm)

Diameter (nm)

Aspect ratio

Zeta Potential (mV)

Yield (%)

Effect

p-value

Effect

p-value

Effect

p-value

Effect

p-value

Effect

p-value

Effect

p-value

1944 1088 191 −175 546 −20

0.001 0.531 0.825 0.839 0.534 0.981

9.3 −3.4 −1.2 −5.1 −3.3 3.5

0.000 0.169 0.333 0.006 0.033 0.026

246.2 132.6 88.7 121.0 213.0 −5.1

0.014 0.626 0.535 0.406 0.171 0.971

−25.8 −2.2 3.6 0.4 0.1 2.6

0.000 0.337 0.022 0.748 0.948 0.063

88.0 15.5 14.2 13.7 −3.7 6.7

0.000 0.209 0.059 0.066 0.564 0.314

56.9 −11.8 2.8 0.2 2.8 −0.1

0.000 0.099 0.391 0.956 0.398 0.981

Crystallinity index

The highlighted letters indicate a statistically significant difference (p < 0.1).

more active in medium with pH close to neutrality. Several authors have reported maximal xylanase activity over a large pH range (4.5–10.0) (Heck et al., 2005a,b, 2006). Nevertheless, the experimental design performed by Heck et al. (2005a) suggested that xylanase was more active in neutral pH. Kansoh and Nagieb (2004) described that xylanase presented optimal activity at pH 6.5. For pH at lower levels, another issue was that hydrogen ions neutralized the charges of the hemicellulose associated with the CNFs, to reduce electrostatic repulsion and promote higher interfibrillar interaction and more entangled CNFs networks. In other words, lower pH (4.0) values culminated in larger fibril aggregates. On the other hand, at higher pH (6.0), the number of charges increased, to raise electrostatic repulsion and decrease interaction between the particles, thereby yielding particles of smaller size (Pääkko et al., 2007). [S] and [E] had linear positive and negative effect on particle size, respectively. A rise in [E] from level −1 (30 U/g) to +1 (70 U/g) and a reduction in [S] from level +1 (35%) to −1 (15%) diminished particle diameter. Meyabadi and Dadashian (2012) reported similar findings: higher [S] (35%) and lower [E] (30 U/g) increased particle size probably because the substrate or product inhibited the enzyme (Chowdary et al., 2000). According to several authors, the aspect ratio of CNFs is an important characteristic when it comes to avoiding the agglomeration of CNFs incorporated into polymeric matrixes. In general, the aspect ratio ranged from 80.9 to 698.7. In this study, the highest aspect ratio value was 698.7 (experiment 8), a consequence of the higher fiber length and of the lower fiber diameter obtained in the process conditions. High aspect ratio may favor tension transfer in the nanofiber-matrix interface (Andrade-Mahecha et al., 2015; Chen et al., 2011; George et al., 2011). According to Hongming et al. (2003), to achieve significant composite reinforcement, CNFs should be ideally oriented and present aspect ratio greater than 100 nm. Overall, the results showed that the independent variables in the levels of study of the fractional factorial design did not influence the response variable aspect ratio significantly for a significance level of 10% (p < 0.1). In neutral water, all the suspensions of CNFs exhibited high, negative zeta potential values ranging between −21.2 and −29.5 mV. The variables T and [S] significantly affected this response variable in the studied levels (p = 0.017 and p = 0.057, respectively). T was the variable that impacted the zeta potential the most; the effect was linear and positive (3.6). When the value of the response variable was negative, the transition from level −1 (35 ◦ C) to +1 (50 ◦ C) reduced the negative surface charge of the CNFs. Thus, to obtain a more stable nanofiber suspension, T should be kept at the lower level. The variable [S] had a linear positive effect (2.6). An increase in [S] from level −1 (15%) to +1 (35%) decreased the negative surface charge of the CNFs, and the best condition to obtain this response was to maintain [S] at the lower level. According to Tholstrup Sejersen et al. (2007), higher negative zeta potential values should confer stability to the CNFs; i.e., the colloidal suspension should resist aggregation and raise the degree of dispersion in the composite.

The factors T and pH significantly affected the yield of CNFs at the studied levels (p = 0.059 and p = 0.066, respectively). T had the highest positive linear effect (14.2); that is, it was the factor that influenced the yield of CNFs the most. Transition from level −1 (35 ◦ C) to level +1 (55 ◦ C) increased the production of CNFs. The pH exerted a positive linear effect (13.7); i.e., transition from level −1 (4.0) to +1 (6.0) increased the yield of CNFs. The upper level conditions of T and pH favored enzymatic activity the most, giving higher yield of CNFs. Experiment 6 afforded the highest crystallinity index (61.0%) because the largest amount of amorphous compounds was removed in these test conditions. However, the independent variables in the levels of study of the fractional factorial design did not affect the crystallinity index significantly. To evaluate process repeatability, at least three repetitions had to be conducted under the conditions of the central points. Definition of the lower (−1) and upper (+1) limits may not be the most suitable strategy to carry out a factorial design, and the central points can aid selection of these levels. Therefore, the central points are extremely important to analyze the effects of the study variables in a reliable way (Rodrigues and Iemma, 2014). In the present factorial design study, repeatability of the values obtained for all the response variables of the three central points was quite good. 3.1.3. Structural and morphological properties of CNFs The effect of pretreatment and enzymatic hydrolysis on the structure of cellulose was studied by microscopy analysis. Fig. 2 contains the SEM micrographs of the untreated banana peel bran structure (UB) and of the banana peel bran structure after alkaline treatment (ATB). The SEM images of UB (Fig. 2a) revealed an irregular particulate structure, which resulted from the grinding procedure. The arrows in Fig. 2a pointed to the presence of starch granules from the banana pulp in contact with the peel. The laser diffraction equipment gave an average bran particle size of 22.4 ␮m. The SEM image of ATB (Fig. 2b) showed that the process changed the structure completely. The modified fiber morphology evidenced that this treatment partially removed the amorphous compounds hemicellulose and lignin as well as other compounds present in the fiber, such as pectin, polysaccharides, saponins, gums, waxes, and fats, among others. Alkaline treatment of the bran afforded 16.5% yield of the treated sample relative to the initial amount of bran (dry basis). According to Liao et al. (2005) amorphous compounds can inhibit enzymes physically or chemically. Removal of the hemicellulose and lignin fractions augmented the accessible surface area in cellulose. Other works reported that partial removal of hemicellulose and lignin also improved the yield of enzymatic cellulose hydrolysis in the case of other lignocellulosic materials; e.g., cotton cellulose (Pirani and Hashaikeh, 2013), brewer’s spent grain (Mussatto et al., 2008) and sugarcane bagasse pulp (Hassan et al., 2010). Fig. 3 summarizes the micrographs of the CNF samples. TEM and AFM images confirmed the presence of CNFs after enzymatic

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035

G Model INDCRO-9268; No. of Pages 11

ARTICLE IN PRESS

6

H. Tibolla et al. / Industrial Crops and Products xxx (2016) xxx–xxx

Fig. 2. SEM images of the (a) untreated banana peel bran (UB, 1400×, scale bar = 20 ␮m) and (b) banana peel bran after alkaline treatment (ATB, 500×, scale bar = 30 ␮m).

treatment in all the experiments. According to the images, the structures of CNFs consisted of long and entangled cellulosic filaments, which formed a network. After enzymatic treatment, the diameter of the CNFs ranged from 5.2 to 15.8 nm, which attested that particle size decreased as compared with UB (22.4 ␮m). The CNFs obtained in this work were also smaller than the CNFs achieved after treatment of bagasse pulp with xylanase (9–25 nm) described by Hassan et al. (2010). According to these authors, enzymatic treatment of the bagasse pulp with xylanase partially removed hemicellulose and increased the cellulose content in the isolated CNFs. Fig. 4 displays the X-ray diffraction patterns of UB, ATB, and CNFs produced from the 24−1 fractional factorial design. According to Pirani and Hashaikeh (2013), wide regions in the diffractograms represent amorphous portions, whereas thin and sharp peaks correspond to the crystalline region. The X-ray diffraction pattern (Fig. 4) revealed that UB contained a large amorphous portion. The typical B-type pattern peak at 2␪ = 17◦ evidenced the presence of starch. The diffraction patterns of the CNF samples were similar–they exhibited two broad peaks. These sharper diffraction peaks at 2␪ = 16◦ and 2␪ = 22◦ are typical of cellulose I, indicating the higher crystallinity of the CNFs [7]. The UB presented a less pronounced peak at 2␪ = 22◦ because this sample had low crystallinity index (15%) due to the presence of amorphous lignocellulosic residues (banana peel). After alkaline treatment, the crystallinity of the bran sample increased by 200%. The bleaching technique proved to be efficient—alkali (5% KOH) removed lignin and disrupted hydrogen

bonding in the cellulose structure, to increase the crystal surface area and to reduce the amount of certain waxes and oils that covered the outer surface of the fiber cell (Alemdar and Sain, 2008). The diffractogram of ATB showed that peaks representing the amorphous region became less intense, whereas the standard peak at 2␪ = 22◦ intensified (Fig. 4). The latter peak is characteristic of cellulose I and indicates a rise in crystallinity, which was up to 46.8% after lignin was removed. All the CNFs produced by enzymatic hydrolysis presented higher crystallinity (from 48.5% to 61.0%, Table 1) as compared to UB and ATB. The final CNFs had higher crystallinity as compared to the starting material (UB) and the pretreated bran (ATB). According to Mussatto et al. (2008), removal of hemicellulose and lignin significantly modifies the bran structure: the crystal surface area increases, and the cellulose fiber becomes more accessible to the enzyme. The values of crystallinity obtained here were similar to the values obtained with CNFs submitted to more aggressive treatments like CNFs from banana peel subjected to acid hydrolysis (58.6% to 64.9%) (Pelissari et al., 2014) and CNFs from achira (57.5% and 69.8%) (Andrade-Mahecha et al., 2015). The FTIR spectra agreed with the results described above. Fig. 5 depicts the FTIR spectra of UB, ATB, and all the CNFs obtained by enzymatic hydrolysis. The spectra of all the samples presented a broad band in the region of 3311 cm−1 and 3364 cm−1 , due to OH elongation. This band is specific to cellulose material (Siqueira et al., 2010b; Yang et al., 2007). The spectrum of UB displayed a small band in the region of 2925 cm−1 , which intensified in the spectra of ATB and CNFs. This band originated from stretching vibrations of

Fig. 3. (a) TEM images (1000×, scale bar = 100 nm) and (b) AFM 3D images (scanning area 1.0 ␮m × 1.0 ␮m, scale bar = 200 nm) of the samples of cellulose nanofibers (CNFs).

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035

G Model INDCRO-9268; No. of Pages 11

ARTICLE IN PRESS H. Tibolla et al. / Industrial Crops and Products xxx (2016) xxx–xxx

7

Fig. 4. X-ray diffractograms of untreated banana peel bran (UB), banana peel bran submitted to alkaline treatment (ATB), and cellulose nanofiber (CNF) samples obtained by 24−1 fractional factorial design.

C H bonds, characteristic of hemicellulose and cellulose (Cherian et al., 2008). Note that the bands due to cellulose (3311 cm−1 , 3364 cm−1 , 2925 cm−1 , and 1350 cm−1 ) intensified after the treatments because the area of crystalline cellulose increased. The intense bands in the region of 1350 cm−1 in the spectra of ATB and CNFs were attributed to C H deformation in cellulose. The delignification step efficiently removed lignin and hemicellulose from the bran: the band at 1253 cm−1 (C O stretching of

the guaiacyl ring) in the spectrum of UB became less intense in the spectrum of ATB and disappeared in the spectrum of CNFs. This band is typical of ester, ether, or phenol compounds bearing CO groups (Siqueira et al., 2010b). The spectrum of ATB still had a band at 1609 cm−1 , assigned to aromatic rings and conjugated carbonyl groups present in the polyphenolic groups of the lignin structure. This same band did not occur in the spectrum of CNFs, attesting to delignification. However, this band still appeared in the spectrum of

Fig. 5. FTIR spectra of UB, ATB, and CNFs obtained by 24−1 fractional factorial design.

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035

G Model INDCRO-9268; No. of Pages 11

ARTICLE IN PRESS

8

H. Tibolla et al. / Industrial Crops and Products xxx (2016) xxx–xxx

Table 3 Validation tests (VTs) and respective results obtained from the characterization of cellulose nanofibers (CNFs). Exp

(X1 )*

(X2 )**

(X3 )***

(X4 )****

Length (L) (nm)

Diameter (d) (nm)

Aspect ratio (L\d)

Zeta potential (mV)

Crystallinity index (%)

Process yield (%)

VT1 VT2 VT3 VT4

35 35 55 55

6.0 6.0 6.0 6.0

70 70 70 70

35 15 35 15

1551.7 ± 73.8a 1490.0 ± 107.3a 1544.5 ± 40.6a 1941.3 ± 29.1b

7.4 ± 0.9a 3.7 ± 0.4b 8.8 ± 0.7a 8.5 ± 1.2a

214.6 ± 14.6a 404.5 ± 63.9b 170.2 ± 14.7a 245.1 ± 26.5a

−30.5 ± 1.7a −29.1 ± 0.7a −31.5 ± 2.9a −30.8 ± 0.3a

57.2 ± 0.7a 61.5 ± 1.1ab 66.2 ± 4.1b 67.0 ± 2.0b

83.2 ± 7.8a 68.0 ± 9.0a 68.1 ± 5.6a 65.2 ± 12.2a

a,b

Different letter superscripts in the same column indicate a statistically significant difference (p < 0.05). X1: T (◦ C). X2: pH. *** X3: [E] (U/g), where U is defined as the mL of xylose released per minute per mL of enzyme. **** X4: [S] (%). *

**

ATB, indicating that some lignin remained after the alkaline treatment (Hassan et al., 2010). According to Cherian et al. (2008), bands in this region (1253 cm−1 ) are related to the hemicellulose fraction. This helped to confirm that enzymatic hydrolysis with xylanase not only isolated CNFs, but also removed the hemicellulose fractions efficiently. Moreover, according to Cherian et al. (2008), the bands in the region of 1731 cm−1 in the spectra of ATB and CNFs originated from dissolution of the hemicellulose component of the banana fibers during the washing process. The band at 1102 cm−1 in the spectrum of CNFs may be associated with changes in hydrogen bonds and possibly indicates the transition from cellulose I to cellulose II (Zuluaga et al., 2007). The band in the region of 1034 cm−1 , which emerged in all the spectra, referred to fractions of xyloglucans associated with hemicellulose that were strongly bound inside the cellulose microfibrils and were not hydrolyzed (Viikari et al., 1994; Zuluaga et al., 2009). As described by Alemdar and Sain (2008), the bands in the region of 896 cm−1 in the spectra of CNFs and ATB corresponded to typical structures of cellulose. This band did not appear in the spectrum of UB; however, the latter spectrum exhibited a band in the region of 840 cm−1 , which is characteristic of aromatic ring CH groups in lignin (Marcovich et al., 1996). The results described above evidenced that the process that isolated CNFs (alkaline treatment with 5% KOH followed by enzymatic hydrolysis with xylanase) sufficed to remove amorphous compounds from the material. Hence, this process may be considered environmentally friendly to treat CNFs. 3.2. Experimental validation tests of enzymatic hydrolysis The experimental validation tests (VTs) considered the best results obtained from the fractional factorial design (Table 1) and took the statistically significant independent variables into account during production of the CNFs (Table 2). The choice of conditions aimed to verify the tendencies and to establish the optimal values for the independent variables, to improve the properties of the CNFs. Treatments were conducted in triplicate for each selected condition. Table 3 shows the process conditions that were used during the VTs. The pH was fixed at 6.0 because this independent variable significantly influenced particle diameter. [S] was kept at the higher levels, which increased the production of CNFs and gave particles with smaller diameter. [E] was fixed at 70 U/g of bran because this value affected particle diameter significantly. In addition, maintaining this variable at the higher levels afforded particles with smaller diameter. T significantly impacted the zeta potential and yield; however, higher levels increased the yield of CNFs but reduced the negative zeta potential, so the temperature was tested at two levels—35 and 55 ◦ C. Table 3 lists the results obtained from the characterization of CNFs for the treatments carried out for the VTs. [S] was statistically significant with respect to the response variables diameter and zeta potential. Lower [S] (15%) generated fibers with smaller

diameter and higher negative potential. However, for experiment test 8 (Table 1), [S] at 35% provided fibers with the best properties, which led us to assess [S] at 15% and 35% during the VTs. TEM analysis confirmed that CNFs were present in the four aqueous suspensions used in the VTs (Fig. 6a–d), attesting that all the treatments effectively isolated cellulose fibers from the banana peel bran at the nanometric scale. TEM images also showed that CNFs had a network structure with long entangled cellulosic fibers. However, compared to Fig. 3, the VT samples were not agglomerated, and the fibers were well separated from each other. This is an important feature for incorporation of these fibers into polymeric matrixes. Optimization of the enzymatic hydrolysis process conditions enabled production of dispersed and typical CNFs. The diameters of the CNFs obtained in these tests ranged from 3.7 to 8.8 nm and the length lay between 1941.3 and 1490.0 nm. Sample VT2 had the smallest diameter (3.7 nm), which was statistically different from the diameter of the other samples (p > 0.05). Because VT2 presented fibers with the smallest diameter and highest length, it had the largest aspect ratio (404.5), which was significantly different (p > 0.05) from the aspect ratio of the other samples. Zeta potential (ZP) measurements confirmed the presence of negative charges on the surface of the CNFs (−29.1 to −31.5 mV). Sample VT3 afforded the highest ZP value, without statistical difference (p > 0.05) as compared to the other samples. Because here the pH was fixed (6.0), all the experiments provided similar ZP values. According to Cho et al. (2012), electrolytic properties (pH level) largely influence the surface charge of a material. Moreover, the higher ZP values obtained for the CNFs produced under VT conditions ensured more stable suspensions (higher ZP) than the CNFs generated during the fractionated factorial design. This fact attested to the optimization of the independent variable pH to produce CNFs by enzymatic hydrolysis. Sample VT1 afforded the highest yield of CNFs (83.2%), without statistical significance (p < 0.05). In this case, the high yield obtained for VT1 implied that a smaller amount of amorphous components was removed, as confirmed by the lower crystallinity index. Fig. 7 illustrates the X-ray diffraction pattern and the FTIR spectra of the CNFs obtained by the validation tests (VTs). Note that all the samples (Fig. 7a) exhibited two main reflection peaks, at 2␪ = 16◦ and 2␪ = 22◦ , related to the crystalline structure of cellulose (AndradeMahecha et al., 2015). Sample VT4 gave the highest crystallinity index, as expected—for this sample, the test involved lower [S] and higher [E]. In other words, the use of more enzyme per gram of substrate allowed a larger amount of hemicellulose to be removed and increased the crystallinity of the resulting CNFs. Larger removal of hemicellulose resulted in lower yield of product. However, the value obtained for VT4 (67%) was not statistically different (p < 0.1) from the values achieved for VT2 (66.2%) and VT3 (61.5%) (Table 3). The FTIR spectra (Fig. 7b) of all the samples produced during the VTs proved that all the treatments efficiently removed fractions of amorphous compounds from the substrate, to expose the crystalline area. The FTIR spectra displayed small bands at 1736–1730 cm−1 , assigned to carbonyl stretching in hemicellulose.

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035

G Model INDCRO-9268; No. of Pages 11

ARTICLE IN PRESS H. Tibolla et al. / Industrial Crops and Products xxx (2016) xxx–xxx

9

Fig. 6. TEM images of the validation test (VT) samples: (a) VT1 (T = 35 ◦ C, pH = 6.0, [E] = 70 U/g, [S] = 35%) (20,000×, scale bar = 200 nm), (b) VT2 (T = 35 ◦ C, pH = 6.0, [E] = 70 U/g, [S] = 15%) (50,000×, scale bar = 100 nm), (c) VT3 (T = 55 ◦ C, pH = 6.0, [E] = 70 U/g, [S] = 35%) (25,000×, scale bar = 200 nm), and (d) VT4 (T = 55 ◦ C, pH = 6.0, [E] = 70 U/g e [S] = 15%) (25,000×, scale bar = 200 nm).

Fig. 7. (a) X-ray diffraction pattern and (b) FTIR spectra of the CNFs obtained by the validation tests (VTs).

The band at 1609 cm−1 , due to aromatic rings and typical of lignin, did not appear in the spectra after treatment. Analysis of the results from the VTs revealed the best conditions to produce CNFs by enzymatic hydrolysis. In general, combination of different process parameters during enzymatic treatment successfully led to CNFs. CNFs produced during VT2 displayed ideal features for future use in biodegradable composites; i.e., the lowest diameter and the highest aspect ratio (3.7 nm and 404.5, respectively), as well as high yield of CNFs (68%), good stability

(−29.1 mV), and high crystallinity index (61.5%). However, the yield of CNFs and the crystallinity index were lower than the values obtained under the conditions of VT2. Moreover, VT4, conducted at 55 ◦ C, also gave better zeta potential (-30.8 mV) and crystallinity index (67.0%) than VT2. On the basis of all the tested variables, pH, [E], and [S] should be fixed at 6.0, 70 U/g, and 15%, respectively. The optimal temperature ranged from 35 to 55 ◦ C. The VT2 experiment was advantageous over the other VT experiments: it employed lower temperature

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035

G Model INDCRO-9268; No. of Pages 11

ARTICLE IN PRESS

10

H. Tibolla et al. / Industrial Crops and Products xxx (2016) xxx–xxx

(35 ◦ C) to give considerably high yield of CNFs, making the process more economically viable. 4. Conclusion The studies reported herein have demonstrated that an agricultural residue (banana peel) can be potentially used to produce value-added products. The cellulose nanofibers produced from this raw material by enzymatic hydrolysis display ideal characteristics for use as reinforcing agents in polymeric matrixes. The green banana peel bran (Musa paradisiaca) presents cellulose content of 7.5%, which makes it an extremely strong structural material with axial stiffness, a desirable feature for a reinforcing fiber. The use of fibers as reinforcement in composite materials with surface irregularities provides better anchoring of the fibers to the polymeric matrix. Given the need to develop environmentally friendly methods to obtain cellulose nanofibers, compared to acid hydrolysis, enzymatic hydrolysis constitutes a mild process that dismisses the use of chemicals. Although the enzymatic treatment requires the use of an alkaline treatment step for delignification, the KOH concentration (5%) is low and does not pose any improper environmental risks. Furthermore, 5% KOH effectively removes large part of the amorphous fraction present in the fiber, yielding a more crystalline material, from 15% (bran non-treated) to about 65% (CNFs). The fractional factorial design is a helpful tool to assess how factors impact response variables and to define the optimum conditions to produce CNFs. Fixed values of pH (6.0), fixed concentrations of the enzyme (70 U/g of substrate) and substrate (15%), and temperatures ranging between 35 and 55 ◦ C afford the best materials. Enzymatic hydrolysis with xylanase is a relatively new method and is a promising treatment to isolate CNFs from banana peel. In conclusion, this study has shown the potential application of unripe banana peel as raw material to isolate CNFs. Acknowledgments The authors would like to acknowledge the financial support provided by Coordenac¸ão de Aperfeic¸oamento de Pessoal de Nível Superior (CAPES) (2952/2011) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) (477842/2011-9). The authors would also like to acknowledge the Brazilian Nanotechnology National Laboratory (LNNano) for allocation of the TEM and AFM apparatus. References AOAC, 2005. Official Methods of Analysis, 18th ed. Association of Official Analytical Chemists, Washington, USA. Alemdar, A., Sain, M., 2008. Biocomposites from wheat straw nanofibers: morphology, thermal and mechanical properties. Compos. Sci. Technol. 68, 557–565. Andrade-Mahecha, M.M., Pelissari, F.M., Tapia-Blácido, D.R., Menegalli, F.C., 2015. Achira as a source of biodegradable materials: isolation and characterization of nanofibers. Carbohydr. Polym. 123, 406–415. Bhattacharya, D., Germinario, L.T., Winter, W.T., 2008. Isolation, preparation and characterization of cellulose microfibers obtained from bagasse. Carbohydr. Polym. 73, 371–377. Castro, A.M.d., Pereira Jr, N., 2010. Produc¸ão, propriedades e aplicac¸ão de celulases na hidrólise de resíduos agroindustriais. Quím. Nova 33, 181–188. Chen, W., Yu, H., Liu, Y., Hai, Y., Zhang, M., Chen, P., 2011. Isolation and characterization of cellulose nanofibers from four plant cellulose fibers using a chemical-ultrasonic process. Cellulose 18, 433–442. Cherian, B.M., Pothan, L.A., Nguyen-Chung, T., Mennig, G., Kottaisamy, M., Thomas, S., 2008. A novel method for the synthesis of cellulose nanofibril whiskers from banana fibers and characterization. J. Agric. Food Chem. 56, 5617–5627. Cho, D., Lee, S., Frey, M.W., 2012. Characterizing zeta potential of functional nanofibers in a microfluidic device. Colloid Interface Sci. 372, 252–260. Chowdary, G.V., Hari Krishna, S., Hanumantha Rao, G., 2000. Optimization of enzymatic hydrolysis of mango kernel starch by response surface methodology. Bioprocess Biosyst. Eng. 23, 681–685. Deepa, B., Abraham, E., Cherian, B.M., Bismarck, A., Blaker, J.J., Pothan, L.A., Leao, A.L., de Souza, S.F., Kottaisamy, M., 2011. Structure, morphology and thermal

characteristics of banana nano fibers obtained by steam explosion. Bioresour. Technol. 102, 1988–1997. Elanthikkal, S., Gopalakrishnapanicker, U., Varghese, S., Guthrie, J.T., 2010. Cellulose microfibres produced from banana plant wastes: isolation and characterization. Carbohydr. Polym. 80, 852–859. George, J., Ramana, K.V., Bawa, A.S., Siddaramaiah, 2011. Bacterial cellulose nanocrystals exhibiting high thermal stability and their polymer nanocomposites. Int. J. Biol. Macromol. 48, 50–57. Hassan, M., Mathew, A., Hassan, E., Oksman, K., 2010. Effect of pretreatment of bagasse pulp on properties of isolated nanofibers and nanopaper sheets. Wood Fiber Sci. 42, 362–376. Hassan, M., Hassan, E., Oksman, K., 2011. Effect of pretreatment of bagasse fibers on the properties of chitosan/microfibrillated cellulose nanocomposites. J. Mater. Sci. 46, 1732–1740. Heck, J.X., de Barros Soares, L.H., Ayub, M.A.Z., 2005a. Optimization of xylanase and mannanase production by Bacillus circulans strain BL53 on solid-state cultivation. Enzyme Microb. Technol. 37, 417–423. Heck, J.X., Flôres, S.H., Hertz, P.F., Ayub, M.A.Z., 2005b. Optimization of cellulase-free xylanase activity produced by Bacillus coagulans BL69 in solid-state cultivation. Process Biochem. 40, 107–112. Heck, J.X., Flôres, S.H., Hertz, P.F., Ayub, M.A.Z., 2006. Statistical optimization of thermo-tolerant xylanase activity from Amazon isolated Bacillus circulans on solid-state cultivation. Bioresour. Technol. 97, 1902–1906. Henriksson, M., Henriksson, G., Berglund, L.A., Lindström, T., 2007. An environmentally friendly method for enzyme-assisted preparation of microfibrillated cellulose (MFC) nanofibers. Eur. Polym. J. 43, 3434–3441. Hongming, M., Zeng, J., Realff, M.L., Kumar, S., Schiraldi, D.A., 2003. Processing structure, and properties of fibers from polyester/carbon nanofiber composites. Compos. Sci. Technol. 63, 1617–1628. Hubbe, M.A., Rojas, O.J., Lucia, L.A., Sain, M., 2008. Cellulosic nanocomposites: a review. Bioresources 3, 929–980. Kansoh, A.L., Nagieb, Z.A., 2004. Xylanase and mannanase enzymes from Streptomyces galbus NR and their use in biobleaching of softwood kraft pulp. Antonie Van Leeuwenhoek 85, 103–114. Liao, W., Wen, Z., Hurley, S., Liu, Y., Liu, C., Chen, S., 2005. Effects of hemicellulose and lignin on enzymatic hydrolysis of cellulose from dairy manure. Appl. Biochem. Biotechnol. 121–124, 1017–1030. Lu, P., Hsieh, Y.L., 2012. Preparation and characterization of cellulose nanocrystals from rice straw. Carbohydr. Polym. 87, 564–573. Marcovich, N.E., Reboredo, M.M., Aranguren, M.I., 1996. FTIR spectroscopy applied to woodflour. Compos. Interfaces 4, 119–132. Meyabadi, T.F., Dadashian, F., 2012. Optimization of enzymatic hydrolysis of waste cotton fibers for nanoparticles production using response surface methodology. Fibers Polym. 13, 313–321. Molina, F.M.P., 2013. Production and Characharacterization of Biodegradable Films of Banana Starch and Flour Reinforced with Cellulose Nanofibers. University of Campinas, Thesis. Mussatto, S.I., Fernandes, M., Milagres, A.M.F., Roberto, I.C., 2008. Effect of hemicellulose and lignin on enzymatic hydrolysis of cellulose from brewer’s spent grain. Enzyme Microb. Technol. 43, 124–129. Pääkko, M., Ankerfors, M., Kosonen, H., Nykänen, A., Ahola, S., Österberg, M., Ruokolainen, J., Laine, J., Larsson, P.T., Ikkala, O., Lindström, T., 2007. Enzymatic hydrolysis combined with mechanical shearing and high-pressure homogenization for nanoscale cellulose fibrils and strong gels. Biomacromolecules 8, 1934–1941. Pelissari, F.M., Andrade-Mahecha, M.M., Sobral, P.J.d.A., Menegalli, F.C., 2012. Isolation and characterization of the flour and starch of plantain bananas (Musa paradisiaca). Starch 64, 382–391. Pelissari, F.M., Sobral, P.J.A., Menegalli, F.C., 2014. Isolation and characterization of cellulose nanofibers from banana peels. Cellulose 21, 417–432. Pirani, S., Hashaikeh, R., 2013. Nanocrystalline cellulose extraction process and utilization of the byproduct for biofuels production. Carbohydr. Polym. 93, 357–363. Rodrigues, M.I., Costa, P.R.A., 2014. Protimiza Experimental Design, Available in: http://experimental-design.protimiza.com.br/. Accessed in: 2015. Rodrigues, M.I., Iemma, A.F., 2014. Experimental strategy for fractional factorial and central composite rotational designs. In: CRC Press TFG (Ed.), Experimental Design and Process Optimization. CRC Press, Florida. Rosa, M.F., Medeiros, E.S., Malmonge, J.A., Gregorski, K.S., Wood, D.F., Mattoso, L.H.C., Glenn, G., Orts, W.J., Imam, S.H., 2010. Cellulose nanowhiskers from coconut husk fibers: effect of preparation conditions on their thermal and morphological behavior. Carbohydr. Polym. 81, 83–92. Satyamurthy, P., Jain, P., Balasubramanya, R.H., Vigneshwaran, N., 2011. Preparation and characterization of cellulose nanowhiskers from cotton fibres by controlled microbial hydrolysis. Carbohydr. Polym. 83, 122–129. Segal, L., Creely, J.J., Martin, A.E., Conrad, C.M., 1959. An empirical method for estimating the degree of crystallinity of native cellulose using the X-ray diffractometer. Text. Res. J. 29, 786–794. Siqueira, G., Bras, J., Dufresne, A., 2010a. Cellulosic bionanocomposites: a review of preparation, properties and applications. Polymers 2, 728–765. Siqueira, G., Bras, J., Dufresne, A., 2010b. Luffa cylindrica as a lignocellulosic source of fiber, microfibrillated cellulose, and cellulose nanocrystals. Bioresources 5, 727–740. Sluiter, A., Hames, B., Ruiz, R., Sacarlata, C., Sluiter, J., Templeton, D., 2008a. Determination os Structural Carbohydrates and Lignin in Biomass. NREL. National renewable Energy Laboratory of Analytical procedure (LAP).

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035

G Model INDCRO-9268; No. of Pages 11

ARTICLE IN PRESS H. Tibolla et al. / Industrial Crops and Products xxx (2016) xxx–xxx

Sluiter, A., Ruiz, R., Sacarlata, C., Sluiter, J., Templeton, D., 2008b. Determination of Extractives in Biomass. NREL. National renewable Energy Laboratory of Analytical procedure (LAP). Sun, J.X., Sun, X.F., Zhao, H., Sun, R.C., 2004. Isolation and characterization of cellulose from sugarcane bagasse. Polym. Degrad. Stab. 84, 331–339. Tholstrup Sejersen, M., Salomonsen, T., Ipsen, R., Clark, R., Rolin, C., Balling Engelsen, S., 2007. Zeta potential of pectin-stabilised casein aggregates in acidified milk drinks. Int. Dairy J. 17, 302–307. Tibolla, H., Pelissari, F.M., Menegalli, F.C., 2014. Cellulose nanofibers produced from banana peel by chemical and enzymatic treatment. LWT – Food Sci. Technol. 59, 1311–1318. Van Soest, J.J.G., Hulleman, S.H.D., de Wit, D., Vliegenthart, J.F.G., 1996. Changes in the mechanical properties of thermoplastic potato starch in relation with changes in B-type crystallinity. Carbohydr. Polym. 29, 225–232. Vicentini, N.M., Dupuy, N., Leitzelman, M., Cereda, M.P., Sobral, P.J.A., 2005. Prediction of cassava starch edible film properties by chemometric analysis of infrared spectra. Spectrosc. Lett. 38, 749–767.

11

Viikari, L., Kantelinen, A., Sundquist, J., Linko, M., 1994. Xylanases in bleaching: from an idea to the industry. FEMS Microbiol. Rev. 13, 335–350. Yang, H., Yan, R., Chen, H., Lee, D.H., Zheng, C., 2007. Characteristics of hemicellulose, cellulose and lignin pyrolysis. Fuel 86, 1781–1788. Yu, Y., Lou, X., Wu, H., 2008. Some recent advances in hydrolysis of biomass in hot-compressed water and its comparisons with other hydrolysis methods. Energy Fuels 22, 46–60. ˜ Zuluaga, R., Putaux, J.-L., Restrepo, A., Mondragon, I., Ganán, P., 2007. Cellulose microfibrils from banana farming residues: isolation and characterization. Cellulose 14, 585–592. ˜ Zuluaga, R., Putaux, J.L., Cruz, J., Vélez, J., Mondragon, I., Ganán, P., 2009. Cellulose microfibrils from banana rachis: effect of alkaline treatments on structural and morphological features. Carbohydr. Polym. 76, 51–59.

Please cite this article in press as: Tibolla, H., et al., Cellulose nanofibers produced from banana peel by enzymatic treatment: Study of process conditions. Ind. Crops Prod. (2016), http://dx.doi.org/10.1016/j.indcrop.2016.11.035