Isolation of cellulose nanocrystals from pseudostems of banana plants

Abstract

On account of their excellent mechanical properties, cellulose nanocrystals (CNCs) are attracting significant interest as a naturally sourced, renewable and inexpensive component of a broad range of nanomaterials. CNCs can be extracted from virtually any natural cellulosic material, but characteristic properties such as maximum aspect ratio, crystal structure, and crystallinity vary considerably between sources. In this work, the isolation of CNCs from the pseudostems of banana plants was explored. The dried stems from the species musa sapientum linn were first cleaned by Soxhlet extraction, alkali treatment and bleaching and subsequently hydrolyzed to CNCs using sulfuric acid. The hydrolysis time was systematically varied, with the objective to maximize the length (L = 375 ± 100 nm) and aspect ratio (A = 28) of the resulting CNCs. The surface charge density of the CNCs thus isolated was 168 mmol kg⁻¹, the predominant crystal structure was that of cellulose I, and the crystallinity was 74%. In order to elucidate the reinforcing capability of the new CNCs, nanocomposites with an ethylene oxide–epichlorohydrin copolymer were prepared and their mechanical properties were investigated by dynamic mechanical analysis (DMA). A comparison with reference nanocomposites made with CNCs isolated from cotton shows that the new CNCs display a higher reinforcing capability.

---

Introduction

Cellulose, the most abundant polymer on earth, occurs in plant cell walls where it is organized in a hierarchical architecture.¹ At the lowest hierarchical level, the cellulose macromolecules are uniaxially oriented and form highly crystalline domains, which are interrupted by amorphous material.¹ If natural cellulosic materials are hydrolyzed in a controlled manner with mineral acids such as sulfuric acid,² hydrochloric acid,³ or phosphoric acid,⁴ the amorphous parts are dissolved and cellulose nanocrystals (CNCs) of rather well-defined shape can be obtained. CNCs are highly crystalline nanoparticles with a rod-like shape^5–7 having a typical diameter of 5–50 nm and a length of 100–3000 nm.^8–11 CNCs offer intriguing mechanical properties,^9,10,12 i.e. high stiffness and strength, exhibit a high aspect ratio (10–85),^11,13–15 are naturally sourced, renewable, inexpensive and the health risks associated with their use may be lower than in case of other nanofillers.^16,17 Due to these characteristics and the ease of surface modification, CNCs have become an attractive component for a broad variety of nanomaterials,¹ such as mechanically adaptive materials,^18–25 aerogels,^26–28 optically²⁹ or electrically^30–32 active materials, as well as a large family of mechanically reinforced polymer nanocomposites.^33–36

The most widely investigated sources for CNCs are cotton (c-CNCs),³⁷ tunicates (t-CNCs),^33,38 and wood pulp,⁶ but CNCs can also be extracted from many other sources.³⁹ Characteristic properties such as maximum aspect ratio, crystal structure, and crystallinity vary considerably between sources. For example, average length (L) and aspect ratio (A) increase from ∼225 nm and ∼10 in the case of c-CNCs to ∼1160 nm and ∼85 in the case of t-CNCs.^{8,10,11,25,33} Bras et al.³⁹ showed that the stiffness of sheets produced from different CNCs scales with the CNC's aspect ratio A. A comparison of our group's data for the stiffness of nanocomposites of PVAc and c-CNCs and t-CNCs respectively, reveals that a higher aspect ratio not only reduces the onset for percolation, but also increases the level of reinforcement.^18–20,25 Therefore, sources and protocols that permit the isolation of high-aspect ratio CNCs are a promising target to investigate. The highest reinforcing capability has been reported for t-CNCs isolated from sessile sea creatures known as tunicates,^14,19,25 but this source may be difficult to exploit industrially. Recent reports on the extraction of high(er)-aspect ratio CNCs from plants such as luffa cylindrica (A = 46.8),⁴⁰ rice husk (A = 10–15),⁴¹ kenaf bast fibers (A = 12),⁴² and stipa tenacissima (A = 20),⁴³ are therefore of great interest.^44–48

Here, the isolation of CNCs from fibrous portions of banana plants was explored. The banana plant is one of the most important fruit crops in the world,^49–51 but the eatable fruit constitutes only 12% w/w of the plant.⁴⁹ Currently, the pseudostem, which makes up a large part of the plant, is either discarded as agricultural waste,^49,52 used for the production of cellulose,⁵³ the fabrication of paper,⁵⁴ the manufacture of handicrafts,⁵⁵ as a substrate for the production of mushrooms,⁵⁶ and several other applications.^53–62 This setting motivates our goal to develop processes to convert the fibrous fraction of banana farming waste, in particular the large pseudostem, into high-aspect-ratio CNCs.

The extraction of cellulose microfibrils from the rachis (the small portion of the plant that carries the fruits) of banana plants^49,50 have previously been reported and fibers isolated from banana plants were successfully incorporated into several polymer matrices.^63,64 However, the extraction of well-defined CNCs from the pseudostems of banana plants has, to our best knowledge, not been documented. Bolio-López et al.⁵¹ converted a mixture of banana rachis and banana pseudostem into cellulose pulp by acid hydrolysis, sodium perchlorate bleaching, and alkaline extraction using sodium hydroxide followed by a second perchlorate bleaching and then drying and milling. From this material, the authors extracted nanocellulose by hydrolysis with 4 M HCl, purification by centrifugation, hydrolysis of the supernatant using 64% H₂SO₄ and ultrasonication. The dimensions of the resulting nanofibers were investigated using AFM and width of 7–70 nm and a length of 200–1300 nm were reported. However, the published electron microscopy images show no indication of CNCs.⁵¹ Cherian et al. extracted cellulose nanofibrils by steam explosion of banana fibers.⁶⁵ Alkali treatment with 2% NaOH was done under high pressure, bleaching was done at ambient pressure and acid hydrolysis using up to 11% oxalic acid was again done under high pressure. The resulting nanofibers displayed a width of 2–5 nm, and a length of 200–250 nm. However, the AFM images reveal ill-defined cylindrical objects, instead of well-defined CNCs.

We here report on the isolation of high-aspect ratio CNCs from the pseudostems of banana plants of the species musa sapientum linn. Our process combines cleaning and bleaching protocols that were used for the extraction of cellulose microfibrils from banana farming residues^49,50 with hydrolysis conditions that we successfully employed to isolate c-CNCs.² The hydrolysis conditions were systematically varied with the objective to maximize the length and aspect ratio. In order to elucidate the reinforcing capability of the new b-CNCs, nanocomposites with an ethylene oxide–epichlorohydrin copolymer were prepared and their mechanical properties were investigated by thermal dynamic mechanical analysis and compared to those of reference nanocomposites made with c-CNCs isolated from cotton.

Experimental

Materials and general methods

Pseudostems from the species musa sapientum linn were harvested on private properties in Bangkok, Thailand. Sodium hydroxide (reagent grade pellets, anhydrous, ≥98%); acetic acid (Reagent Plus®, ≥99%); dimethyl sulfoxide (puriss p.a., ACS reagent, ≥99.99%) and hydrogen peroxide solution (35% w/w in H₂O) were purchased from Sigma Aldrich. Dichloromethane (ACS/HPLC, stab.); ethyl alcohol (reagent grade, ACS/HPLC certified); toluene (B&J Brand, for liquid chromatography) and sulfuric acid (95–97%, for analysis) were purchased from Honeywell Chemicals. Dimethylformamide (assay, ≥99.9%) was purchased from ROMIL. All chemicals were used without any further purification. The ethylene oxide–epichlorohydrin (EO–EPI) copolymer was purchased from Daiso Co., Ltd., Japan epichlomer, with a co-monomer ratio of: 1 : 1, and a density of 1.39 g cm⁻³. c-CNCs, which were used as a reference, were extracted from Whatman no. 1 filter paper according to an already existing protocol,¹⁴ which was slightly modified from the procedure reported by Dong et al.² Typical dimensions for length and width of c-CNCs were 230 ± 71 nm and 17 ± 7 nm, respectively, resulting in an aspect ratio of 13; the sulfate content was determined to be 92.7 mmol kg⁻¹ by conductometric titration. TEM-micrographs as well as the titration plot of c-CNCs are shown in the ESI (Fig. S1 and S2†). For all experiments requiring H₂O, MilliQ H₂O produced by a Sartorius arium 611VF water purifying system was used. Sonication was done in a BANDELIN SONOREX TECHNIK RL 70 UH sonicator at a power of 40 kHz. Lyophilization was done with a VirTis benchtop K lyophilizer.

Extraction and bleaching of pseudostems

The outer parts of the pseudostems were cut into pieces measuring ∼1 × 1 × 1 cm and dried. This material (20 g) was purified by Soxhlet extraction for 24 h, using a 1 : 2 v/v ethyl alcohol : toluene mixture as solvent and a solid : solvent ratio of 11.25 mg mL⁻¹. The soxhlet-purified stem pieces were dried for at least 24 h at ambient temperature and pressure, subsequently transferred into an aqueous NaOH solution (2 L, 2% mol mol⁻¹, solid : solvent ratio maximal 10 mg mL⁻¹) and stirred during 2 h at 80 °C to remove hemicelluloses and ash. After 2 h the solids were filtered off and washed with H₂O (3 × 500 mL). This alkali treatment/washing cycle was repeated twice, until the NaOH solution remained colorless. After three cycles, the solids were washed with H₂O until the pH was ∼7, and stored for 12 h in a refrigerator. The material was subsequently bleached in an aqueous solution containing H₂O₂ (1.3% w/w) and acetic acid (0.1% v/v) (this solution was prepared by combining 74 mL 35% w/w H₂O₂ in H₂O with 2 L H₂O and adding 2 mL of glacial acetic acid) by stirring at 70 °C during 2 h. The solids were filtered off and washed with H₂O (3 × 500 mL). This bleaching/washing cycle was repeated once. After two cycles, the solids were washed with H₂O until the pH was neutral, and dried for 24 h at ambient temperature and pressure. The solids were finally subjected to another Soxhlet extraction for 24 h, using a 1 : 2 v/v ethyl alcohol : toluene mixture. The resulting material was dried for at least 24 h at ambient temperature and then directly used for the next step without any further characterization.

Hydrolysis of cellulose pulp

The extracted and bleached solids (3 g) were soaked in H₂O (250 mL) for 1 h and disintegrated in a Rotel Promix 474 blender. After blending, the pulp was transferred into a beaker, which was placed in an ice-bath, and the pulp was stirred with a magnetic stirrer until the temperature had reached 4 °C. Concentrated sulfuric acid (150 mL) was slowly added while the mixture was vigorously stirred; the addition was controlled so that the temperature was kept below 20 °C. After the addition of sulfuric acid was complete, the mixture was heated to 50 °C and stirred at this temperature. The hydrolysis time was systematically varied between 30 and 240 min in order to establish the optimum time (90 min, vide infra). After the hydrolysis was complete, the solid was separated by centrifugation at 3600 rpm for 15 min. The supernatant was decanted, replaced by an equal amount of fresh H₂O, and the mixture was centrifuged again. This procedure was repeated at least three times, until the pH of the supernatant was neutral (∼6–7). The solid remaining after the last centrifugation was dialyzed against H₂O for 7 days; during this process, the H₂O was exchanged twice per day. After the dialysis was complete, the cellulose dispersion was sonicated for 12 h and H₂O was removed by lyophilization during 48 h. b-CNCs were thus obtained as a white fluffy solid. For b-CNCs hydrolyzed for 90 minutes, the yield was 10% based on the weight of the dried banana stems.

Preparation of EO–EPI/CNC nanocomposites

A master solution of the ethylene oxide–epichlorohydrin (EO–EPI) copolymer (25 g) in DMSO (500 mL) was prepared by stirring the mixture for 96 h at ambient temperature, atmosphere and pressure.

Neat EO–EPI films were prepared by sonicating an aliquot (25 mL) of the above master solution for 2 h, and casting the solution into round Teflon® Petri-dishes with a diameter of 80 mm. Most of the solvent was evaporated by heating at ambient pressure for 96 h to 70 °C and followed by heating at 70 °C under a vacuum of 150 mbar during 48 h. The transparent, colorless films thus produced had a thickness of typically 20 μm.

EO–EPI/CNC nanocomposites films were made by first individually combining the lyophilized b-CNCs or c-CNCs with DMSO (CNC concentration = 12 mg mL⁻¹) and sonicating 10 mL of the resulting dispersions for 2 h (c-CNC) or 12 h (b-CNC). The CNC dispersions thus produced were each combined with the EO–EPI master solution (17.6 mL, containing 880 mg of EO–EPI), which had previously been sonicated for 60 min. The EO–EPI/CNC/DMSO mixtures were stirred for 60 min at ambient conditions, before they were cast into round Teflon® Petri-dishes with a diameter of 80 mm. The drying procedure followed the one for the neat EO–EPI films (vide supra) and the films thus produced had a thickness of typically 20 μm.

Determination of chemical composition

The chemical composition of the b-CNCs produced was established by FTIR spectroscopy (spectra were recorded on a Perkin Elmer Spectrum 65 FTIR spectrometer between 4000 and 600 cm⁻¹ at a resolution of 4 cm⁻¹ with a total scan number of 15, ESI Fig. S4†) and Technical Aspects in Pulp and Paper Industry (TAPPI) protocols as detailed below.

The content of solvent extractives was determined following TAPPI-protocol 204 cm-97 doing a quantitative Soxhlet extraction using 1 : 3 ethyl alcohol v/v benzene mixture. The content of acid-insoluble lignin was determined following TAPPI-protocol T 222 om-11 where samples were dissolved in concentrated H₂SO₄ and the acid-insoluble part, which is supposed to be lignin, was quantified.

Transmission electron microscopy

Transmission electron microscopy (TEM) was carried out to investigate the structure, the aspect ratio and dispersibility of CNCs in different solvents. To investigate the structure, CNCs were suspended in H₂O at a concentration of 0.1 mg mL⁻¹ and the suspension was sonicated for 10 h. One drop of the suspension was then placed onto a carbon film on a 200 square mesh copper grid (Electron Microscopy Science, Hatfield, USA) and the grid was dried at ambient for 2 h. To investigate the dispersibility of the CNCs in different solvents, CNCs were dispersed in dichloromethane, toluene, DMF, DMSO, H₂O, or ethyl alcohol at a concentration of 0.5 mg mL⁻¹ and the mixtures were sonicated for 10 h. Two drops of each suspension were transferred onto a carbon film on a 200 square mesh copper grid and dried on top of a tissue paper at ambient for at least 24 h. All samples were imaged using a Philips CM 100 microscope operating with an acceleration voltage of 80 kV.

Aspect ratios of b-CNCs were determined by measuring the dimensions of b-CNCs in the TEM figures using the measure IT-program provided by Olympus and dividing the measured length of b-CNC by the measured width. For each hydrolysis time at least 50 different CNCs were measured to obtain the values listed in Table 2.

Conductometric titration

Charge densities of b-CNCs and c-CNCs were determined by conductometric titration. 50 mg of CNCs were suspended in 10 mL 0.01 M aqueous HCl by 4 h of sonication. Once the CNCs were suspended, the suspension was titrated with 0.01 M NaOH. To determine the charge density, the titration graph (ESI Fig. S5†) was divided into three regions, namely titration of the excessive HCl, titration of the sulfate-groups on the surface of CNCs and addition of excess NaOH at the end of the titration. After fitting a line to each of those three regions, the amount of NaOH used to titrate the sulfate groups was determined by the intercept points between the linear fits. From the volume of NaOH used to titrate the sulfate groups, the charge density of CNCs could be determined:
where c_NaOH is the concentration of NaOH (0.01 M), V_NaOH is the volume of NaOH required to titrate all the HCl and m_CNC is the mass of CNCs used for the measurement.

Thermogravimetric analysis

The thermal stability of the CNCs after the various processing steps was examined by thermogravimetric analysis (TGA) using a TGA/DSC1 from Mettler Toledo. The mass of the samples was ∼5–9 mg and samples were heated in air from 40–600 °C with a heating rate of 10 °C min⁻¹ (ESI Fig. S3†); the residual mass at 600 °C was used as an estimation of the ash content.

Crystal structure and degree of crystallinity

The crystal structure and degree of crystallinity of b-CNCs and c-CNCs (ESI Fig. S6†) prepared under different hydrolysis conditions were established by X-ray diffraction on a theta–theta type XRD-Rigaku Ultima IV using Cu-radiation generated at a voltage of 40 kV and a current of 40 mA and a D/texUltra detector. The scan range detected was 2θ = 10°–50°. The crystallinity index of b-CNCs was determined by the X-ray diffraction peak height method,^{40–43,45,46,48} wherein the crystallinity index is determined by the equation:
where I₀₀₂ is the maximum intensity of the peak corresponding to plane having the Miller indices 002 and I_am is the minimal intensity of diffraction of the amorphous phase at 2θ = 18°.

Dispersibility

The dispersibility of the b-CNCs in different solvents was determined by TEM (vide supra) as well as visual inspection. For the latter, b-CNCs were transferred into glass vials and dispersed in dichloromethane, toluene, DMF, DMSO, ethyl alcohol, and H₂O at a b-CNC-concentration of 2 mg mL⁻¹ by 10 h of sonication. Photographs of the suspensions thus produced were taken immediately after the sonication was over, as well as 5 min, 2 h, 12 h, 24 h, and 1 week after sonication.

Mechanical properties

The mechanical properties of neat EO–EPI films as well as of EO–EPI/b-CNC and EO–EPI/c-CNC nanocomposite films were determined by dynamic mechanical analysis (DMA) as a function of temperature on a TA Q800 instrument using rectangular samples (6 mm × 25 mm × 20 μm) that were cut from the films. The shear storage modulus was measured in the temperature range from -70–100 °C using a heating rate of 3 °C min⁻¹. Testing was done in the temperature ramp/frequency sweep mode using tensile clamps, and frequency and strain amplitude were kept constant at 1 Hz and 15 μm, respectively. Each composition was measured at least 5 times; representative curves are shown and storage moduli at −60 °C and 25 °C are reported as averages of all measurements.

Results and discussion

Extraction and bleaching

The raw material utilized in this study was taken from the outer portions of the pseudostem of the species musa sapientum linn, mechanically disintegrated and dried. This material was subjected to a multi-step process, which combines extraction and bleaching protocols that were previously employed for the isolation of cellulose microfibrils from banana farming residues^49,50 with hydrolysis conditions that we previously used to isolate CNCs from cotton (Fig. 1).² To document the removal of non-cellulosic matter and highlight the importance of the various extraction and bleaching steps, the contents of organic solvent extractives and lignin in the material were determined after each step, using Technical Aspects in Pulp and Paper Industry (TAPPI) protocols; the results are summarized in Table 1. For the purpose of comparison, c-CNCs, which were isolated from cotton according to a literature protocol by hydrolysis with H₂SO₄,² were also analyzed.

Fig. 1 Photographs of (a) a banana plant; (b) dried and cut pseudostem pieces; (c) dried pulp after organic extraction, treatment with base, and bleaching; and (d) lyophilized cellulose nanocrystals (b-CNC) obtained by hydrolysis of the pulp with 65% H₂SO₄.

Table 1 Chemical composition of the dried pseudostem, the cellulosic material at different stages of extraction, and of the b-CNCs isolated at the end of the process. For reference purposes, the values of c-CNCs (isolated according to literature procedures from cotton) are also included. All values are w/w%. *Determined by TAPPI-protocol 204 cm-97, **determined by TAPPI-protocol T 222 om-11 ***determined using the mass residue in TGA at 600 °C

Stage of treatment	Solvent extractives*	Lignin**	Ash***
Dried pseudostem	9.9 ± 0.8%	9.9 ± 0.5%	34.1%
After Soxhlet extraction	1.1 ± 0.7%	9.9 ± 0.5%	36.0%
After NaOH treatment	0.6%	6.1%	8.3%
After H2O2 treatment	1.2%	2.8%	15.0%
Isolated b-CNCs	1.3 ± 0.1%	3.8 ± 1.2%	9.7%
Isolated c-CNCs	0.6 ± 0.01%	0.5%	8.4%

---

Compared to other cellulose sources,^40,43,45,46 which in their dried, but otherwise original form have a content of organic solvent extractives between 0.7 and 3%, the pseudostems used here have an extractable content (likely low-molecular-weight carbohydrates or polyphenols) of ca. 10%, which was readily removed by Soxhlet extraction with a 1 : 2 ethyl alcohol : toluene mixture. Table 1 shows that the extraction step did not affect the lignin content. The subsequent extraction with aqueous NaOH was primarily carried out to remove hemicellulose and ashes. TGA graphs (ESI Fig. S3† and Table 1) show that during this step, a significant amount of non-cellulosic compounds has been removed. According to literature, these compounds can be assigned to either silica ash or hemicellulose.⁴⁵ Finally, the material was bleached with acidified H₂O₂. This step served to reduce the lignin content from 6% after basic treatment to 3%. Comparing TGA traces between 200 and 250 °C before and after Soxhlet extraction, a degradation peak corresponding to the solvent extractives detected by TAPPI-protocol 204 cm-97 is visible.

Hydrolysis of CNCs

The extracted and bleached solid was pulpified and the resulting pulp was hydrolyzed at 50 °C in 11 M aqueous H₂SO₄ for 30–240 min, before the b-CNCs were isolated by centrifugation, dialysis, and lyophilization (see Experimental section for details). To identify the hydrolysis time at which b-CNCs of maximum length and aspect ratio were obtained, the b-CNCs obtained under the various conditions were analyzed by means of transmission electron microscopy (TEM), X-ray diffraction (XRD), and surface charge titration; the results are shown in Fig. 2 and Table 2.

Fig. 2 TEM micrographs (13 500× magnification) of b-CNCs, isolated after hydrolysis with H₂SO₄ (65%) for (a) 30 min; (b) 60 min; (c) 90 min (optimal hydrolysis time, inset: magnified b-CNCs (scalebar = 400 nm)); (d) 120 min; and (e) 240 min.

Table 2 Characteristics of b-CNCs isolated after different hydrolysis times

Hydrolysis time [min]	Length [nm]	Width [nm]	Aspect ratio	Charge density [mmol kg^−1]	Apparent crystallinity [%]
0	—	—	—	—	71
30	466 ± 159	19 ± 6	24	161	74
60	441 ± 116	17 ± 5	26	178	74
90	375 ± 100	13 ± 4	28	168	74
120	361 ± 61	17 ± 4	22	135	70
180	378 ± 66	17 ± 4	22	143	69
240	319 ± 68	15 ± 4	21	131	70

---

TEM images of the materials isolated after hydrolysis for 30 or 60 min (Fig. 2a and b) show a considerable amount of aggregates and cellulose fibrils, which implies that hydrolysis partially occurred, but is incomplete. However, in both cases, b-CNCs can be seen, whose average dimensions are 466 ± 159 × 19 ± 6 and 441 ± 1.16 × 17 ± 5 nm, respectively, resulting in an aspect ratio of 24 and 26. After 90 min of hydrolysis, most of the aggregates and fibrils disappeared and the morphology of the b-CNCs thus produced appears as rather homogeneous, with average dimensions of 375 ± 100 × 13 ± 4 nm and an average aspect ratio, A, of 28 (Fig. 2c). TEM images of b-CNCs produced with hydrolysis times longer than 90 min show a further reduction in length and aspect ratio, and impurities resulting from over-hydrolyzation and aggregation are visible (Fig. 2d and e). At first glance, b-CNCs hydrolyzed for 90 min seem to have the highest A. Regarding deviations in the size of b-CNCs extracted for 30, 60 and 90 minutes, b-CNCs hydrolyzed for 90 min are not statistically different from the ones hydrolyzed for a less amount of time, and small amounts of aggregates are still present in all samples. Still, on the TEM-micrograph there are more aggregates not yet dissolved for CNCs hydrolyzed for 30 or 60 min compared to b-CNCs hydrolyzed for 90 min. These impurities have proven difficult to remove, therefore b-CNCs hydrolyzed for 90 minutes were studied in more detail.

The chemical composition of b-CNCs was studied by IR spectroscopy and compared to c-CNCs (ESI Fig. S4†). As there is no significant difference between the two spectra, one can conclude that the composition of c-CNCs and b-CNCs is identical.

It is well known that the hydrolysis with H₂SO₄ leads to the formation of negatively charged sulphate ester moieties on the surface of the CNCs and therefore the charge density of b-CNCs was determined by conductometric titration (see Table 2 and ESI Fig. S5†). Interestingly, the values obtained in this study appear to decrease with hydrolysis time. The value determined for the material isolated after hydrolysis for 90 min (168 mmol kg⁻¹) is higher than that measured for CNCs extracted from other sources such as tunicates (charge density: 135 mmol kg⁻¹) or cotton (charge density: 92 mmol kg⁻¹) isolated using a similar protocol; the reason for this is, however, not immediately clear.

Crystal structure and degree of crystallinity

The crystal structure and degree of crystallinity of the b-CNCs prepared under different hydrolysis conditions were established by X-ray diffraction. The diffraction patterns show peaks at 2θ = 15.7°, 22.6°, and 34.6° (ESI Fig. S6†). The reflections are consistent with the form of cellulose I, which is known to feature peaks at 14.7°, 16.6°, 20.6°, 22.5° and 34.7°.⁶⁶ Due to non-Gaussian distributions, the peaks at 14.7° and 16.6° and those at 20.6° and 22.5° are not well resolved and appear as broad signals centered around 15.7° and 22.6°, respectively.

During the first 30 minutes of hydrolysis, the crystallinity is increased from 70% to 74% and then remains constant until 90 min of hydrolysis time. This increase in crystallinity can be assigned to preferential hydrolysis of amorphous cellulose while the crystalline regions of cellulose remain intact.³⁸ After 90 min of hydrolysis, the crystallinity decreases with on-going hydrolysis as the crystalline regions of cellulose are also hydrolyzed. This effect has already been observed by H. Kargarzadeh et al.⁴²

Crystallinity was determined by the peak height method; for b-CNCs hydrolyzed for 90 min the crystallinity was found to be 74%. The degree of crystallinity of the present b-CNCs is slightly lower than that of the reference c-CNCs (85%). The degree of crystallinity for b-CNCs was compared with CNCs extracted from other sources.^{40–43,45,46,48} CNCs extracted from other natural sources show crystallinities between 59% for CNCs extracted from rice husk⁴¹ and 96% for CNCs extracted from luffa cylindrica⁴⁰ which implies that the relative crystallinity of b-CNCs is comparable to values obtained in other works.

Dispersibility

The ability of CNCs to form stable colloidal suspensions of adequate concentration in useful solvents^67,68 is essential for their processing, for example to create useful CNC-reinforced polymers. The dispersibility of CNCs depends on several factors, which include the CNC's aspect ratio and surface functionality^8,69 as well as the solvent's ability to disrupt hydrogen bonding between the surface OH-groups of the CNCs.⁷⁰ The dispersibility of b-CNCs was tested in six different solvents. Dichloromethane (DCM, dielectric permittivity, ε = 9.1) and toluene (ε = 2.4) were chosen as non-polar solvents, dimethylformamide (DMF, ε = 38) and DMSO (ε = 46) were chosen as polar aprotic solvents, and H₂O (ε = 80) and ethyl alcohol (ε = 24.6) were chosen as polar protic solvents. Macroscopically, the dispersibility was determined by ultrasonication of b-CNCs in these solvents, and visually inspecting the suspensions thus produced after certain time intervals.

Photographs of suspensions taken immediately after sonication, as well as 12 h and 7 days after sonication are shown in Fig. 3. Immediately after sonication, b-CNCs appeared to be dispersed in all solvents except in ethyl alcohol, in which sedimentation was observed immediately after the end of sonication. After 2 h, most of the b-CNCs had settled in ethyl alcohol, DCM and toluene, and sedimentation had also started for b-CNCs dispersed in DMF. After 7 days, b-CNCs remained dispersed in H₂O and DMSO, which are well-known solvents for CNCs.⁶⁹ Rather surprisingly, the dispersion of b-CNCs in DMF phase-separated, even though other CNCs isolated by hydrolysis with H₂SO₄ usually disperse well in DMF. This behavior may be related to the high surface charge density of the present b-CNCs, which promotes better dispersibility in solvents with high dielectric permittivity (DMSO and H₂O), but not DMF, which has a slightly lower ε.

Fig. 3 Photographs of suspensions of b-CNCs in (a) H₂O; (b) ethyl alcohol; (c) dimethyl sulfoxide; (d) dimethylformamide; (e) dichloromethane; and (f) toluene. In all cases, the b-CNC concentration was 2 mg mL⁻¹, and the suspensions were made by ultrasonication for 10 h. Pictures were taken immediately after preparation, 12 h after preparation, and 7 days after preparation.

TEM images of dilute b-CNC dispersions that had been dried on a TEM grid immediately after sonication (Fig. 4) confirm the macroscopic observations. b-CNCs deposited from H₂O and DMSO are well-individualized and homogeneously distributed on the TEM-grids, while b-CNCs that had been dispersed in other solvents are aggregated. b-CNCs deposited from DMF and ethyl alcohol can be recognized as individual nanocrystals, whereas b-CNCs processed in DCM and toluene show large-scale aggregates. Thus, H₂O and DMSO appear to be the best dispersants for the b-CNCs made here

Fig. 4 TEM micrographs of b-CNCs deposited from suspensions in different solvents. In all cases, the b-CNC concentration was 0.5 mg mL⁻¹, and the suspensions were made by ultrasonication for 10 h. Solvents: (a) EtOH, (b) DMF, (c) DCM, (d) H₂O, (e) DMSO and (f) toluene.

Fabrication of polymer nanocomposites and mechanical characterization

To investigate the usefulness of b-CNCs as a reinforcing filler for polymers, nanocomposites with a rubbery ethylene oxide–epichlorohydrin copolymer (EO–EPI, statistical copolymer, monomer ratio = 1 : 1) were prepared; EO–EPI was chosen as the matrix, because reference data on nanocomposites of EO–EPI with c-CNCs and t-CNCs is already available.^4,14,18,20 Thus, films of nanocomposites comprising 12% w/w (corresponding to 11.5% v/v) b-CNCs or c-CNCs, and also films of the neat EO–EPI were prepared by solution-casting from DMSO and subsequent hot-pressing. All films are transparent, which indicates that the CNCs are well dispersed (Fig. 5, inset). The EO–EPI/b-CNC nanocomposites film is slightly darker than the neat EO–EPI and the EO–EPI/c-CNC nanocomposites. Since TGA experiments reveal no difference in the thermal behavior of b-CNCs and c-CNCs, we relate this coloration to a minor amount of highly colored decomposition products, for example residual lignin. The mechanical properties of the neat EO–EPI and the nanocomposites with b-CNCs and c-CNCs were probed by dynamic mechanical analysis (DMA), which results the shear storage modulus E ′as a function of temperature; representative graphs are shown in Fig. 5.

Fig. 5 DMA curves of a neat EO–EPI and of EO–EPI/b-CNC and EO-EPI/c-CNC nanocomposite films. The nanocomposites contain either 12% w/w b-CNC or 12% w/w c-CNCs. The inset shows a picture of the EO–EPI film (top), the c-CNC/EO–EPI film (middle) and the b-CNC/EO–EPI film (bottom).

The graphs reveal that at all temperatures the nanocomposites exhibit a higher E′ than the neat EO–EPI polymer. Additionally, the reinforcing effect of the b-CNCs is indeed significantly larger than that of the c-CNCs (Table 3). For example at 25 °C, the neat EO–EPI exhibits an E′ of 3.2 ± 0.9 MPa, whereas the nanocomposites bearing 12% w/w of b-CNCs and c-CNCs feature values of 35 ± 4 and 20 ± 1 MPa, respectively. The values observed for neat EO–EPI and EO–EPI/c-CNC nanocomposites are in close agreement with results previously reported, where at 25 °C E′ for neat EO–EPI was found to be 3.7 MPa and 21 MPa for EO–EPI/c-CNC nanocomposite bearing 15% v/v.¹⁴ Importantly, the new b-CNCs display a higher reinforcing capability than their lower-aspect ratio counterparts isolated from cotton. The higher stiffness observed for the composites with b-CNCs is consistent with findings of Bras et al. as well as results of other authors, who reported that the reinforcing effect of CNCs increases with the aspect ratio.^{18–20,25,39}

Table 3 Storage moduli of neat EO–EPI and of EO–EPI/b-CNC and EO–EPI/c-CNC nanocomposite below (−60 °C) and above T_g (25 °C)

Composition	E′ [MPa] at −60 °C	E′ [MPa] at 25 °C
EO–EPI	3056 ± 429	3.2 ± 0.9
EO–EPI + 12% w/w c-CNC	4700 ± 197	20 ± 1
EO–EPI + 12% w/w b-CNC	6169 ± 389	35 ± 4

---

Conclusions

We have reported the extraction procedure for CNCs from the pseudostem of banana plants and characterized the resulting b-CNCs. Acid hydrolysis and purification indicates b-CNCs can be isolated with an aspect ratio of 28 and an apparent crystallinity of 74%. Due to high surface charge density, b-CNCs disperse well in polar solvents such as DMSO and H₂O, which allows the production of nanocomposites containing c-CNCs and b-CNCs in EO–EPI. As suspected, b-CNCs have a higher reinforcing effect than the shorter c-CNCs. Given the bio-renewable, high availability of b-CNCs from the waste of banana production, they represent a significant improvement of aspect ratio over similarly scalable CNC sources, such as cotton and rice husks.

Acknowledgements

We gratefully acknowledge funding from the Swiss National Science Foundation (National Research Programme 64, Project #406440_131264/1) and the Adolphe Merkle Foundation. We thank Christoph Neururer for conducting XRD-experiments. We are indebted to Pomthong Malakul, Manit Nithitanakul, Ratana Rujiravanit and their students at The Petroleum and Petrochemical College and The Center for Petroleum, Petrochemicals, and Advanced Materials, Chulalongkorn University for harvesting the banana plants.

Notes and references

1. S. J. Eichhorn, A. Dufresne, M. Aranguren, N. E. Marcovich, J. R. Capadona, S. J. Rowan, C. Weder, W. Thielemans, M. Roman, S. Renneckar, W. Gindl, S. Veigel, J. Keckes, H. Yano, K. Abe, M. Nogi, A. N. Nakagaito, A. Mangalam, J. Simonsen, A. S. Benight, A. Bismarck, L. A. Berglund and T. Peijs, J. Mater. Sci., 2010, 45, 1–33.
2. X. Dong, J.-F. Revol and D. Gray, Cellulose, 1998, 5, 19–32.
3. J. Araki, M. Wada, S. Kuga and T. Okano, Colloids Surf., A, 1998, 142, 75–82.
4. S. Camarero Espinosa, T. Kuhnt, E. J. Foster and C. Weder, Biomacromolecules, 2013, 14, 1223–1230.
5. D. Klemm, B. Heublein, H.-P. Fink and A. Bohn, Angew. Chem. Int. Ed., 2005, 44, 3358–3393.
6. J. F. Revol, H. Bradford, J. Giasson, R. H. Marchessault and D. G. Gray, Int. J. Biol. Macromol., 1992, 14, 170–172.
7. D. Klemm, F. Kramer, S. Moritz, T. Lindström, M. Ankerfors, D. Gray and A. Dorris, Angew. Chem. Int. Ed., 2011, 50, 5438–5466.
8. J. Araki, M. Wada and S. Kuga, Langmuir, 2000, 17, 21–27.
9. A. Šturcová, G. R. Davies and S. J. Eichhorn, Biomacromolecules, 2005, 6, 1055–1061.
10. R. Rusli and S. J. Eichhorn, Appl. Phys. Lett., 2008, 93, 033111.
11. M. M. De Souza Lima, J. T. Wong, M. Paillet, R. Borsali and R. Pecora, Langmuir, 2002, 19, 24–29.
12. W. J. Lyons, J. Appl. Phys., 1959, 30, 796–797.
13. S. J. Eichhorn, Soft Matter, 2011, 7, 303–315.
14. J. R. Capadona, O. Van Den Berg, L. A. Capadona, M. Schroeter, S. J. Rowan, D. J. Tyler and C. Weder, Nat. Nanotechnol., 2007, 2, 765–769.
15. M. A. S. Azizi Samir, F. Alloin and A. Dufresne, Biomacromolecules, 2005, 6, 612–626.
16. M. J. D. Clift, E. J. Foster, D. Vanhecke, D. Studer, P. Wick, P. Gehr, B. Rothen-Rutishauser and C. Weder, Biomacromolecules, 2011, 12, 3666–3673.
17. S. M. C. Endes, O. Schmid, D. Vanhecke, E. J. Foster, A. Petri-Fink, B. Rothen-Rutishauser, C. Weder and M. J. D. Clift, J. Phys.: Conf. Ser., 2013, 429, 012008.
18. J. R. Capadona, K. Shanmuganathan, D. J. Tyler, S. J. Rowan and C. Weder, Science, 2008, 319, 1370–1374.
19. K. Shanmuganathan, J. R. Capadona, S. J. Rowan and C. Weder, J. Mater. Chem., 2010, 20, 180–186.
20. K. Shanmuganathan, J. R. Capadona, S. J. Rowan and C. Weder, Prog. Polym. Sci., 2010, 35, 212–222.
21. R. Rusli, K. Shanmuganathan, S. J. Rowan, C. Weder and S. J. Eichhorn, Biomacromolecules, 2010, 11, 762–768.
22. J. Mendez, P. K. Annamalai, S. J. Eichhorn, R. Rusli, S. J. Rowan, E. J. Foster and C. Weder, Macromolecules, 2011, 44, 6827–6835.
23. K. L. Dagnon, K. Shanmuganathan, C. Weder and S. J. Rowan, Macromolecules, 2012, 45, 4707–4715.
24. A. E. Way, L. Hsu, K. Shanmuganathan, C. Weder and S. J. Rowan, ACS Macro Lett., 2012, 1, 1001–1006.
25. M. Jorfi, M. N. Roberts, E. J. Foster and C. Weder, ACS Appl. Mater. Interfaces, 2013, 5, 1517–1526.
26. M. D. Gawryla, O. van den Berg, C. Weder and D. A. Schiraldi, J. Mater. Chem., 2009, 19, 2118–2124.
27. L. Heath and W. Thielemans, Green Chem., 2010, 12, 1448–1453.
28. J. T. Korhonen, M. Kettunen, R. H. A. Ras and O. Ikkala, ACS Appl. Mater. Interfaces, 2011, 3, 1813–1816.
29. M. V. Biyani, E. J. Foster and C. Weder, ACS Macro Lett., 2013, 2, 236–240.
30. O. van den Berg, M. Schroeter, J. R. Capadona and C. Weder, J. Mater. Chem., 2007, 17, 2746–2753.
31. J. D. Mendez and C. Weder, Polym. Chem., 2010, 1, 1237–1244.
32. S. Padalkar, J. R. Capadona, S. J. Rowan, C. Weder, Y.-H. Won, L. A. Stanciu and R. J. Moon, Langmuir, 2010, 26, 8497–8502.
33. V. Favier, H. Chanzy and J. Y. Cavaille, Macromolecules, 1995, 28, 6365–6367.
34. S. Kumar, M. Hofmann, B. Steinmann, E. J. Foster and C. Weder, ACS Appl. Mater. Interfaces, 2012, 4, 5399–5407.
35. L. Tang and C. Weder, ACS Appl. Mater. Interfaces, 2010, 2, 1073–1080.
36. M. Schroers, A. Kokil and C. Weder, J. Appl. Polym. Sci., 2004, 93, 2883–2888.
37. J.-F. Revol, L. Godbout, X.-M. Dong, D. G. Gray, H. Chanzy and G. Maret, Liq. Cryst., 1994, 16, 127–134.
38. M. M. de Souza Lima and R. Borsali, Macromol. Rapid Commun., 2004, 25, 771–787.
39. J. Bras, D. Viet, C. Bruzzese and A. Dufresne, Carbohydr. Polym., 2011, 84, 211–215.
40. G. Siquiera, J. Bras and A. Dufresne, BioResources, 2010, 5, 14.
41. N. Johar, I. Ahmad and A. Dufresne, Ind. Crops Prod., 2012, 37, 93–99.
42. H. Kargarzadeh, I. Ahmad, I. Abdullah, A. Dufresne, S. Zainudin and R. Sheltami, Cellulose, 2012, 19, 855–866.
43. A. Mabrouk, H. Kaddami, S. Boufi, F. Erchiqui and A. Dufresne, Cellulose, 2012, 19, 843–853.
44. P. Lu and Y.-L. Hsieh, Carbohydr. Polym., 2012, 87, 2546–2553.
45. R. M. Sheltami, I. Abdullah, I. Ahmad, A. Dufresne and H. Kargarzadeh, Carbohydr. Polym., 2012, 88, 772–779.
46. G. Siqueira, H. Abdillahi, J. Bras and A. Dufresne, Cellulose, 2010, 17, 289–298.
47. P. M. Visakh, S. Thomas, K. Oksman and A. P. Mathew, Composites, Part A, 2012, 43, 735–741.
48. W. Chen, H. Yu, Y. Liu, Y. Hai, M. Zhang and P. Chen, Cellulose, 2011, 18, 433–442.
49. R. Zuluaga, J.-L. Putaux, A. Restrepo, I. Mondragon and P. Gañán, Cellulose, 2007, 14, 585–592.
50. R. Zuluaga, J. L. Putaux, J. Cruz, J. Vélez, I. Mondragon and P. Gañán, Carbohydr. Polym., 2009, 76, 51–59.
51. G. I. Bolio-López, A. Valadez-Conzález, L. Veleva and A. Andereeva, Rev. Mex. Ing. Quim., 2011, 10, 10.
52. D. Feriotti and A. Iguti, Proposal for use of pseudostem from banana tree (Musa cavendish), International Congress of Engineering and Food, Athens, Greece, 2012, FEW893, http://www.icef11.org/main.php?fullpaper&categ=FEW.
53. M. Begum, C. Savitha and P. J. Jaison, Res. J. Pharm., Biol. Chem. Sci., 2012, 3, 10.
54. A. Rosal, A. Rodríguez, Z. González and L. Jiménez, Int. J. Phys. Sci., 2012, 7, 8.
55. D. Mohapatra, S. Mishra and N. Sutar, J. Sci. Ind. Res., 2010, 69, 7.
56. P. Rani, N. Kalyani and K. Prathiba, Appl. Biochem. Biotechnol., 2008, 151, 151–159.
57. G. Canché-Escamilla, J. M. De los Santos-Hernández, S. Andrade-Canto and R. Gómez-Cruz, Inf. Tecnol., 2005, 16, 83–88.
58. N. Cordeiro, M. N. Belgacem, I. C. Torres and J. C. V. P. Moura, Ind. Crops Prod., 2004, 19, 147–154.
59. A. Hossain, S. A. Ahmed, A. M. Alshammari, F. M. A. Adnan, M. S. M. Annuar, H. Mustafa and N. Hammad, Afr. J. Microbiol. Res., 2011, 5, 586–598.
60. J. S. Jayaprabha, M. Brahmakumar and V. B. Manilal, J. Nat. Fibers, 2011, 8, 149–160.
61. J.-M. Kuo, A. Hwang, D.-B. Yeh, M.-H. Pan, M.-L. Tsai and B. S. Pan, J. Agric. Food Chem., 2006, 54, 3151–3156.
62. J. Y. Tock, C. L. Lai, K. T. Lee, K. T. Tan and S. Bhatia, Renewable Sustainable Energy Rev., 2010, 14, 798–805.
63. H. Faria, N. Cordeiro, M. N. Belgacem and A. Dufresne, Macromol. Mater. Eng., 2006, 291, 16–26.
64. P. Gañán, J. Cruz, S. Garbizu, A. Arbelaiz and I. Mondragon, J. Appl. Polym. Sci., 2004, 94, 1489–1495.
65. B. M. Cherian, L. A. Pothan, T. Nguyen-Chung, G. n. Mennig, M. Kottaisamy and S. Thomas, J. Agric. Food Chem., 2008, 56, 5617–5627.
66. E. N. J. Ford, S. K. Mendon, S. F. Thames and J. W. Rawlins, J. Eng. Fibers Fabr., 2010, 5, 11.
67. B. G. Ranby, Discuss. Faraday Soc., 1951, 11, 158–164.
68. R. H. Marchessault, F. F. Morehead and N. M. Walter, Nature, 1959, 184, 632–633.
69. O. van den Berg, J. R. Capadona and C. Weder, Biomacromolecules, 2007, 8, 1353–1357.
70. R. P. Swatloski, S. K. Spear, J. D. Holbrey and R. D. Rogers, J. Am. Chem. Soc., 2002, 124, 4974–4975.

---

Footnote

† Electronic supplementary information (ESI) available: TEM-figure of CNCs derived from cotton, TGA-graphs of all the extraction intermediates, IR-spectra of b-CNCs, titration curves from charge titrations of b-CNCs and c-CNCs and XRD-patterns taken for b-CNCs and c-CNCs. See DOI: 10.1039/c3ra46390g

---