Introduction

Cellulose nanofibers (CNFs), also known as nanofibrillated cellulose (NFC), microfibrillated cellulose (MFC), or nanocellulose, are the nano-sized cellulose fibers extracted from the cell wall of a plant. CNFs have high strength, stiffness (modulus ~ 13 GPa) (Nogi et al. 2009), and high specific surface area (~ 50 m2/g) (Andrady 2008). CNFs usually have a diameter of 5–50 nm or 20–100 nm, and a length of several or several tens of micrometers depending on the processing methods and starting materials (Freywyssling and Muhlethaler 1963; Moon et al. 2011; Nogi et al. 2009). CNFs can be used in a variety of high value applications such as reinforcing agents in plastic composites and polymer films, membranes, paper coating, food additives, medicines and cosmetic products (Eichhorn et al. 2009). More importantly, CNFs are made from renewable biomass and they are biodegradable, which are beneficial for the environment and sustainable development. In recent years, more and more studies have reported using CNFs in more advanced applications such as our previous studies in the area of flexible battery (Fu et al. 2016a; Yang et al. 2018; Zhang et al. 2016), UV light blocking (Zhang et al. 2017a) and anti-counterfeiting security materials (Zhang et al. 2017b), and others in printed flexible energy devices (Fu et al. 2016b; Iwamoto et al. 2008), and nano-based devices and materials (Huang et al. 2013; Wu et al. 2015).

Typically, CNFs can be produced by three different methods: ultrafine grinding, homogenization and microfluidization (Nechyporchuk et al. 2016). Among them, ultrafine grinding can be operated in a continuous mode and can be potentially scaled up to industrial production. In an ultrafine grinding process, generally, plant cells/fibers are disintegrated into nano-sized fibrils under high mechanical shearing actions or friction forces which break the interfibrillar hydrogen-bond networks within cell walls, thus separating the fibrils from one another. Therefore, the process demands an extensive amount of mechanical energy, and this becomes the major challenge for wide-range applications of CNFs.

A number of studies have been conducted for reducing the mechanical energy consumption by using pretreatments such as enzymatic (Paakko et al. 2007; Wang et al. 2014) and chemical (Bian et al. 2018b; Tejado et al. 2012) to facilitate the fibrillation process. Chemical pretreatment with 2,2,6,6-tetramethylpiperidinyl-1-oxyl (TEMPO) oxidation is one of the most popular ones used in main lab research, with which the oxidation reaction converts primary hydroxyls on cellulose into carboxylate, thus producing interfibrillar repulsive forces between fibrils that facilitate fibrillation (Saito et al. 2007). However, it has been recognized that TEMPO process could have significant environmental issue as the chemicals used are not easily recovered, not to mention it is expensive as well (Isogai et al. 2011; Mao et al. 2010; Serra et al. 2017; Testa et al. 2004). On the other hand, enzyme pretreatment has been explored by several studies as enzymes have high selectivity for targeted substrates and are environmentally-friendly (Filson et al. 2009; Long et al. 2017; Sacui et al. 2014; Wang et al. 2014, 2015).

Raw materials could be another important factor that determines the energy consumption needed to produce CNFs. In most cases, CNFs are produced from the bleached softwood/hardwood pulp fibers (Stelte and Sanadi 2009). More recently, non-wood and agricultural residues have also attracted interests as alternative sources for CNF production due to their abundance, low-cost, and being renewable (Alemdar and Sain 2008; Chaker et al. 2014; Zimmermann et al. 2010). Especially, bagasse is a solid waste or byproduct generated in the sugar manufacturing process and sugar canes have a shorter growing cycle and lower lignin content (Khristova et al. 2006; Pandey et al. 2000), meaning that delignification and purification processes of bagasse are easier. In theory, all plant cells have similar chemistry, i.e., mainly cellulose, hemicellulose and lignin, and similar cell wall architecture, i.e., layered structure of cellulose fibrils. Although longer fibers from wood are good for physical strength when making paper, it is probably making no difference in terms of producing CNFs as the nano-sized fibrils in wood and in non-wood plant cells are not much different (Missoum et al. 2013; Rajinipriya et al. 2018). Most importantly, it is found that bagasse fibers are shorter, have more hemicellulose, and the fibrils are less tightly bonded than fibrils in wood cell walls, which are expected to make the mechanical disintegration easier in CNF manufacturing (Iwamoto et al. 2008). CNFs have been isolated from bagasse by cellulase, mono-component endoglucanase and xylanase pretreatment (Saelee et al. 2016a) together with mechanical process.

Several studies have been reported with regard to producing CNFs from bagasse pulp fibers using enzyme pretreatment together with a mechanical process (de Campos et al. 2013; Feng et al. 2018; Saelee et al. 2016b), very few has elucidated the energy consumption and process efficiency in the production of CNFs from bagasse pulp.

This study aimed to investigate how bagasse fibers respond to enzyme pretreatment and mechanical grinding in a CNF production process. To further understand the fundamentals behind, both bleached bagasse and softwood kraft pulp, denoted BBK and BSK, respectively, were processed for producing CNFs using the same process, i.e., enzyme pretreatment followed by mechanical grinding. The energy consumption, morphology of the produced CNFs and the physical and optical properties of CNF films were studied.

Materials and methods

Materials

A commercial bleached bagasse and softwood kraft pulp (namely BBK and BSK, respectively) were supplied by a pulp mill in Guangxi, China. Both fibers were received in dry form. The pulp fibers were first soaked in deionized water overnight and then disintegrated in a lab disintegrator at 10,000 revolutions. Mono-component endoglucanase (FbierCare R) supplied by Novozymes Denmark was used. FiberCare R was produced from a genetically modified Aspergillus species. Its cellulolytic activity of 2036 U/mL was determined as sodium carboxymethyl cellulose (CMC-Na) activity (Wood and Bhat 1988).

Enzyme pretreatment

The BBK and BSK fibers were pretreated with enzyme prior to mechanical grinding under following conditions: 50 °C, 5% pulp consistency, pH 7 (adjusted with phosphate buffer), enzyme dosage of 3% based on the weight of dried pulp, 12 h, at 200 rpm for magnetic stirring in a reactor. After treatment, enzyme was deactivated after reaction by increasing temperature to 100 °C for 15 min. Subsequently, the samples were washed with deionized water. The washed solids were used for analysis and subsequent mechanical fibrillation.

Materials characterization

The untreated BBK and BSK, as well as their enzymatically treated fibers were characterized in terms of chemical composition, fiber length, fines content, degree of polymerization (DP) and crystallinity index (CI).

The chemical components were determined according to TAPPI standard methods: cellulose (TAPPI T17 wd-70) and pentosan (TAPPI T223 cm-01). Fiber length and fines content were determined using Kajaani fiber analyzer (FS300, Metso automation, Finland) according to ISO 16065-1.

Viscosity of resulting fibril suspension was measured according to TAPPI standard method T230 om-99. DP was calculated by the following formula:

$$ {\text{DP}}^{ 0. 9 0 5} = 0 . 7 5 \times {\text{(954log}}\left( {\text{X}} \right) - 3 2 5 ) $$

where X is viscosity.Cellulose crystallinity of fibers was measured with powder X-ray diffraction (SmartLab 3 kW, Rigaku, Japan) using CuKα radiation generated at 40 kV and 35 mA. The dried neat fiber films were individually packed in a special holder through which X-ray beam passed. Scans were obtained in a 2θ range from an angle of 5° to 50° at a scanning rate of 5°/min. The crystallinity index was calculated from the XRD patterns following the equation according to Segal’s method (Segal et al. 1959).

$$ {\text{CI }}(\% ) = \frac{{{\text{I}}_{ 2 0 0} - {\text{I}}_{\text{am}} }}{{{\text{I}}_{ 2 0 0} }} \times 100 $$

where I200 was the maximum intensity of the crystalline peak at 2θ between 22° and 23°, Iam was the minimum intensity of the amorphous cellulose at 2θ between 18° and 19° (French 2014).

Mechanical fibrillation by ultrafine grinder

The untreated BBK and BSK, as well as their enzymatically treated fibers were immersed in deionized water for 2 h at 2% (w/v) solid consistency, respectively. Afterwards, the suspensions were mechanically fibrillated using a MKZA 10–15 J ultrafine grinder (Masuko Sangyo Co. Ltd, Saitama, Japan) at 1500 rpm as described previously (Rambabu et al. 2016). The gap between the grinding stones was adjusted to − 100 μm from motion zero position with the presence of fiber suspension. The motion zero position was defined at the contact position between the two grinding parts before loading fiber suspension. The total volume of the suspensions was 8 L. The fibrillated pulp was sampled at 0.5, 1.0, 1.5 and 2.0 h for analysis, respectively.

Yield of CNFs

A diluted suspension at 0.1–0.2 wt % was centrifuged at 4500 rpm for 20 min to separate CNFs (in supernatant) from the non-fibrillated and partially fibrillated residue (in sediment). Then the supernatant was removed, and the sediment was dried to a constant weight at 100 °C in an oven. The yield of CNFs was calculated as follows:

$$ {\text{CNF yield }}\left( {{\%}} \right) =\frac{{{\text{Wc}} - {\text{Ws}}}}{\text{Wc}} \times 100 $$

where Wc is a weight of dried cellulose in the suspension before centrifugation and Ws is a weight of dried sediment after centrifugation.

Energy consumption

Energy consumption for mechanical grinding was determined with an energy meter which can measure total input energy, instant power and grinding time (Qingzhi, 8904F, China). In this work, energy consumption was reported in terms of total energy consumption (the consumed total input energy) and net energy consumption (used for the real grinding work for disintegration fibers into nanofibers). The idle power of the grinder (operated without feeding materials) was determined to be a constant value of 1 kW. The energy consumption was reported in kWh per kg of feeding fibers on a dry weight basis (0.16 kg), according to the following equation:

$$ \begin{aligned} {\text{Total energy consumption }}\left( {{\text{kWh}}/{\text{kg}}} \right) = \frac{\text{total input energy (kWh)}}{{ 0. 1 6 ( {\text{kg)}}}} \hfill \\ {\text{Net energy consumption }}\left( {{\text{kWh}}/{\text{kg}}} \right) = {\text{total energy consumption }}\left( {{\text{kWh}}/{\text{kg}}} \right)- \frac{{ 1 \left( {\text{kW}} \right) \times {\text{grinding time (h)}}}}{{ 0. 1 6 ( {\text{kg)}}}} \hfill \\ \end{aligned} $$

Electron microscopy

The morphology of CNFs was observed by Field Emission Scanning Electron Microscopy (FE-SEM) and Transmission Electron Microscopy (TEM).

For FE-SEM imaging, the fibrillated fibril suspensions (without any centrifugation) was dispersed in deionized water (with solid consistency of about 0.1%) and stirred for 2 h to ensure well dispersion. The droplet was deposited on a silicon wafer and dried by a Labconco freeze-drying system (Kansas City, MO, USA). In order to improve conductivity, the samples were sputtered with thin gold layer. The images were observed and recorded using a Hitachi SU8220 (Japan), and the acceleration voltage was set to 15 kV with the resolution of 0.8 nm. TEM images were obtained with a Hitachi HT-7700 (Japan) at an acceleration voltage 100 kV. A drop of a diluted supernatant of CNFs (0.005–0.01 wt %) obtained by centrifugation was deposited on a carbon-coated grid (400 mesh) and allowed to dry at room temperature for 12 h. Subsequently, they were negatively stained with 2% uranyl acetate for 20 min in dark place followed by blotting and were used for imaging. The diameters for different CNFs were determined and calculated on the basis of TEM images using an image analyzer program software, Nano Measurer. A minimum of 150 measurements were performed for evaluating the CNF diameters.

Preparation of the films

The films were prepared by filtering fibrillated fibril suspensions, followed by air- and then oven drying of the wet films. Prior to filtration, the fibrillated fibril suspension was diluted to 0.1 wt % and stirred for 12 h to ensure uniform dispersion. 240 g of suspension was vacuum filtered with a glass filter (G3-90, Shang Hai aladdin, Ltd.) using a hydrophilic polytetrafluoroethylene membrane (pore size: 0.22 μm). Filter membranes were supported by filter paper. The wet films were peeled from the filter membrane and stacked first between two blotting paper sheets. This assembly was placed in between two glass plates, and was allowed to dry at room temperature for 24 h and then at 60 °C for 8 h under a load of approximately 250 N. The films were then conditioned at 23 °C and 50% relative humidity chamber before testing.

Characterization of the films

UV–vis light transmittance

The light transmittance for different films was measured at a wavelength range from 400 to 800 nm using an UV spectrometer (U-4100, Hitachi High-Tech. Corp.). Rectangular specimens with size of 40 mm × 9 mm were placed in a quartz cuvette and measured by placing it 25 cm from the entrance port of the integrating sphere. Three tests were performed for each sample.

Cross-section morphologies

FE-SEM was carried out using a Hitachi SU8220 (Japan) to obtain cross-section morphologies of the films. For this, film samples were quickly frozen using liquid nitrogen to achieve a brittle manner and thus avoid structural deformations of the materials. Afterwards, the resultant samples were mounted on aluminum stubs and coated with gold to enhance the conductivity before imaging.

Density and mechanical properties

The thickness and basic weight of nanofiber films were determined according to TAPPI standard T411 and T410, respectively. The density of the films was then calculated by dividing sample basis weight by thickness. Tensile tests were performed using an Instron 3365 universal material testing machine (Instron Engineering Corporation, MA, USA) with a 500 N load cell. Specimens were cut into a size of 20 mm × 3 mm and were subsequently conditioned at 50% RH and 23 °C for at least 1 week prior to testing. Rectangular specimens were pre-loaded with a force of 5 N to remove slack and the tests were conducted at a crosshead speed of 1 mm/min with a specimen gauge length of 10 mm. The specific tensile strength and specific Young’s modulus were calculated by dividing the tensile stress and Young’s modulus by density, respectively (Bian et al. 2018a).

Thermogravimetric analysis (TGA)

Thermogravimetric analysis (TGA) of the freeze-dried cellulose fibrils were employed based on a TGA–DSC/DTA analyzer (STA 449 F5, NETZSCH-Gerätebau GmbH, Germany). The dried samples were heated from 30 to 600 °C with a rate of 10 °C/min in a nitrogen environment.

Results and discussion

Energy consumption and cellulose nanofiber (CNF) yield

Bleached bagasse and softwood kraft pulp fibers (namely BBK and BSK, respectively) were subjected to enzyme pretreatment prior to grinding to reduce the energy consumption in grinding. As shown in Fig. 1, for all the samples, specific total energy consumption for mechanical fibrillation increases with the extending of grinding time. CNF production is very energy intensive when grinding BBK and BSK directly. In comparison, the specific total energy consumption is significantly lower for BBK compared to that of BSK at the same grinding time. For example, at grinding time of 2 h, the specific total energy consumption is 33.55 kWh/kg for BBK and 35.21 kWh/kg for BSK, respectively. This suggests that BBK cell wall is more susceptible to mechanical separation in grinding than the BSK cell wall. In addition, employing enzyme pretreatment reduced the specific total energy consumption for both BBK and BSK. In case of BBK, the specific total energy consumption at grinding time of 2 h was reduced from 33.55 to 29.58 kWh/kg, by 11.83% reduction.

Fig. 1
figure 1

Specific total energy consumption of mechanical grinding as a function of grinding time. The BBK and BSK denote bleached bagasse and softwood kraft pulp fibers, respectively. The BBK-E and BSK-E denote the BBK and BSK with enzyme pretreatment, respectively

Full size image

In fact, the total energy consumption in grinding consists of idle energy, which is mainly for driving the grinder, and net grinding energy, which contributes to the real grinding work for disintegration of fibers into nanofibers. The idle power measured in this process was 1 kW. Therefore, the idle energy consumption accounted for approximately 35–43% of specific total energy consumption for all the samples, which means only 21.05 kWh/kg was consumed directly in CNF production in case of BBK at grinding time of 2 h. As the idle energy depends mainly on the equipment and employed process, it is possible to reduce it through process optimization and improving the design of the equipment.

On the other hand, the net energy consumption can be more accurately evaluated considering the yield of CNFs, not the grinding time alone. As indicated in Fig. 2, grinding of BBK yields more CNFs than BSK, which means for the same energy consumption BBK produces more CNFs, or in other words, producing the same amounts of CNFs from BBK requires less net energy consumption than from BSK. For example, at 50% CNF yield, BBK consumed 21.05 kWh/kg, which was 7.31% lower than the energy consumption of BSK for the same yield. Applying enzyme pretreatment further improved the yield of CNFs for both BBK and BSK, i.e., reducing further the net energy consumption in CNF processing; however, the reduction of the net energy consumption with enzyme pretreatment was greater for BBK than for BSK. At about 50% yield, the specific net energy consumption reduced from 21.05 to 8.48 kWh/kg for BBK, by 59.71%, in contrast, it was from 22.71 to 12.95 kWh/kg for BSK, by 42.98%. These results indicate that enzyme pretreatment had greater impact on the fiber cell wall structure of BBK than that of BSK, which helped loosen up more the cell wall integrity. In this study, the maximum CNF yield for BBK was 70.56%, with grinding time of 2 h.

Fig. 2
figure 2

CNF yield as a function of specific net energy consumption of mechanical grinding

Full size image

Chemical composition and fiber characteristics

In order to understand the effects of raw materials and enzyme pretreatment on the energy consumption of mechanical fibrillation and CNF yield, the chemical composition, fiber length, fines content, degree of polymerization (DP) and cellulose crystallinity index (CI) of BBK and BSK before and after enzyme pretreatment were analyzed as listed in Table 1.

Table 1 Chemical composition and fiber characteristics of BBK, BSK, and their enzymatically treated samples
Full size table

The results indicate that BBK contains a significantly larger fraction of pentosan (hemicellulose) and less cellulose in comparison with BSK. It has been reported that high content of hemicellulose facilitates the fibrillation of the pulp fibers and improved CNF yield (Petroudy et al. 2014). In general, hemicellulose exists between cellulose microfibrils, and its presence may help keep the cellulose microfibrils apart from each other or not being strongly bound by hydrogen bonding. The hemicellulose is also related to fiber flexibility and swelling, which impacts the fibrillation process (Chaker et al. 2013). In addition, hemicellulose, especially its carboxylic groups may prevent the re-aggregation of cellulose fibrils through electrostatic repulsion during mechanical fibrillation. The results from this study support the hypothesis that higher content of hemicellulose in BBK benefits fiber disintegration in mechanical grinding, leading to less energy demand and higher yield of CNFs.

On the other hand, although the chemical composition of either BBK or BSK was not significantly changed by enzyme pretreatment, the fiber morphology was. The fiber length dramatically decreased by 41.46% and 74.77% for BBK and BSK, respectively, in comparison to their original fiber length. The fines content increased by 23.78% and 19.36%, respectively. These results indicate that enzyme cut the fibers thus generating shorter fibers and fines. Fiber cutting by enzymes was also observed in our previous studies of enzymatic hydrolysis for bioethanol production (Clarke et al. 2011; Li et al. 2012). Fiber cutting not only makes fibers smaller in size but also opens more loose ends on the fiber which is expected to make fiber disintegration easier. This may explain the lower energy consumption in the grinding process after enzyme pretreatment as observed foregoing for both BBK and BSK. Of course, enzyme can also cleave cellulose chains in the fiber wall to make the entire fiber structure weaker leading to less energy demand in grinding. This can be seen from the drastic reduction in cellulose DP by about 50% for both BBK and BSK. It was also observed that the CI of fibers actually increased after enzyme pretreatment. Specifically, the CI of BBK changed from 45.95% (BBK) to 58.29% (BBK-E), and the CI of BSK increased from 53.66% (BSK) to 61.85% (BSK-E). This indicated that some of the amorphous region in the cellulose might be reduced by enzyme action so the relative portion of crystalline region was increased.

Morphology of the produced CNFs

To evaluate the effects of raw materials and enzyme pretreatment on the morphology of CNFs, Field Emission Scanning Electron Microscopy (FE-SEM) and Transmission Electron Microscopy (TEM) observations were performed on the resultant CNFs from BBK and BSK without and with enzyme pretreatment, and the results are shown in Figs. 3 and  4.

Fig. 3
figure 3

The morphology of fibrils produced from (a) BBK after grinding for 2 h. b BBK with enzyme pretreatment followed by grinding for 2 h. c BSK after grinding for 2 h and (d) BSK with enzyme pretreatment followed by grinding for 2 h

Full size image
Fig. 4
figure 4

TEM images and diameter distributions of the CNFs from (a) BBK after grinding for 2 h. b BBK with enzyme pretreatment followed by grinding for 2 h. c BSK after grinding for 2 h and (d) BSK with enzyme pretreatment followed by grinding for 2 h

Full size image

In Fig. 3, the fibrils prepared by directly mechanical grinding of BBK and BSK appear web-like structures consisting of randomly entangled sub-micron fibrils mostly. This indicates that mechanical grinding of BBK/BSK for 2 h was not sufficient to completely disintegrate the fibers into nano-size fibrils. Many micro-scale fibers were present in BBK and BSK, which is in agreement with the CNF yield of BBK (54.04%) and BSK (50.26%). However, enzyme pretreatment can significantly improve the degree of fibrillation. The fibers were further defibrillated and much fewer micrometer-scale fibrils are observed in Fig. 3b, d which is consistent with their higher CNF yields (70.56% for BBK and 63.68% for BSK, respectively) as determined previously. It was worth noting that the BBK fibrils showed more entanglement linking between the adjacent nanofibrils. This may stem from the higher hemicellulose content in BBK fibrils, which resulted in differential extent of defibrillation between BBK and BSK during mechanical grinding.

The FE-SEM analysis mainly focused on examining the overall view of the fibrillated cellulose samples. Thus, small or nanoscale cellulose structures existing in unfractionated samples are sometimes undetectable by FE-SEM due to its limited resolution. Herein, TEM imaging was employed to provide closer look of the ultrafine structure of CNFs and obtain a quantitative analysis of the size distribution of CNFs. In this work, TEM observations were carried out on the supernatant fraction obtained by centrifuging the fibril suspension after mechanical grinding. The results are shown in Fig. 4.

The results showed that the CNFs from BBK and BSK fibers without enzyme pretreatment had non-uniform diameters ranging from several to 45 nm. Specifically, the average diameter of BBK and BSK was 23.58 ± 8.63 nm and 25.59 ± 8.58 nm, respectively. With enzyme pretreatment, the diameters of the CNFs were greatly reduced, and became more uniform for both BBK and BSK. For example, for BBK, the average diameter of CNFs was 8.7 ± 3.03 nm and the CNFs with the diameters in the range of 9–12 nm accounted for about 50% of the total produced CNFs.

The length of the CNFs prepared by directly mechanical grinding of BBK and BSK was difficult to be measured from the TEM images due to their highly entangled characteristics. However, the length of the resultant CNFs from fibers subjected to enzyme pretreatment prior to grinding was significantly reduced. For example, for BBK, the length of CNFs varied from 200 to 1000 nm. On the other hand, these CNFs showed rod-like and more individualized morphologies, which were completely different from those prepared by only mechanical grinding.

Characteristics of the films prepared by CNFs

In order to evaluate the film forming properties of the CNFs, nanopaper/film was prepared and both optical property and physical strength of the films were determined. As shown in Fig. 5, the visible light transmittance from 400 to 800 nm is higher for films from BBK than those from BSK. Moreover, enzyme pretreatment improved the transmittance of CNF films from both BBK and BSK. Generally, the dimensions of partially-fibrillated fibers were much larger than the wavelength of visible light, which can result in light scattering, and in turn reduce the light transmittance of the films (Besbes et al. 2011). In addition, when the fibrils were densely packed in the films, the interstices between the fibrils would be small enough to avoid light scattering, and consequently the films will have a greater transparency (Yang et al. 2017). Therefore, the visible light transmittance of the film also reflects the sizes of the nanofibers in them.

Fig. 5
figure 5

UV–vis light transmittance of the films prepared by (a) BBK after grinding for 2 h. b BBK with enzyme pretreatment followed by grinding for 2 h. c BSK after grinding for 2 h and (d) BSK with enzyme pretreatment followed by grinding for 2 h

Full size image

Figure 6 showed the digital photographs and cross-section morphologies of the films. The most transparent film with the highest light transmittance, i.e. 60.02% at 600 nm, was made from BBK after enzyme pretreatment and mechanical grinding in Fig. 5b and it exhibited the most tightly packed cross-section structure in Fig. 6b. The results further confirmed that finer and more uniform nanofibers was produced from BBK compared with BSK.

Fig. 6
figure 6

Digital photographs and cross-section morphologies of the films prepared by (a) BBK after grinding for 2 h. b BBK with enzyme pretreatment followed by grinding for 2 h. c BSK after grinding for 2 h and (d) BSK with enzyme pretreatment followed by grinding for 2 h

Full size image

Figure 7 shows typical stress–strain curves of the films prepared by BBK and BSK fibrils, and the associated data are presented in Table 2.

Fig. 7
figure 7

Stress–strain curves of the films prepared by (a) BBK after grinding for 2 h, b BBK with enzyme pretreatment followed by grinding for 2 h, c BSK after grinding for 2 h and (d) BSK with enzyme pretreatment followed by grinding for 2 h

Full size image
Table 2 The physical and mechanical properties of the films prepared by (a) BBK after grinding for 2 h, (b) BBK with enzyme pretreatment followed by grinding for 2 h, (c) BSK after grinding for 2 h and (d) BSK with enzyme pretreatment followed by grinding for 2 h
Full size table

In general, the tensile strength and specific tensile strength of films produced by BBK and BSK fibrils are not significantly different, which supports the idea that although conventional paper made from bagasse fibers are significantly weaker in physical strengths than that made from wood fibers, this was not necessarily true for nanofibers made from them. In paper, wood cells or fibers are the basic building units of the paper material structure, hence, the sizes and morphologies of bagasse fibers and wood fibers are the critical factors in determining the paper strength. Bagasse fibers are much smaller than softwood fibers, that is the reason that paper made from bagasse fibers has significantly weaker strength than that made from wood fibers. However, in case of CNFs, the fiber sizes produced from both bagasse and wood were nearly the same. This explains the similar tensile strength of CNF films as observed in this study. In addition, employing enzyme pretreatment reduced the tensile strength and elongation at break for films produced by both BBK and BSK fibrils. In case of BBK fibrils, the specific tensile strength of the films was reduced from 160.40 to 113.61 MPa, by 29.17% reduction with enzyme pretreatment. It is because that fibril length is decreased after enzyme pretreatment as revealed by the TEM analysis in Fig. 3, which led to less interconnected network of fibrils in their films and thus lower tensile strength and elongation at break (Chen et al. 2017).

Young’s modulus is a mechanical property that measures the stiffness of a solid material. It defines the relationship between stress and strain in a material in the linear elasticity regime of a uniaxial deformation (Bian et al. 2018a). The Young’s modulus and specific Young’s modulus of the film prepared by BSK fibrils was slightly higher than that of film prepared by BBK fibrils, and this is believed to be the result of higher CI for BSK fibrils than BBK fibrils (Henriksson et al. 2008). Enzyme pretreatment further increases the Young’s modulus of the films, due to the increased CI of fibrils and to the density of the films with enzyme pretreatment (Henriksson et al. 2008; Qing et al. 2013).

Table 2 also showed the physical properties of the films. Films from BBK fibrils generally showed lower thickness and higher density than those from BSK fibrils. This is mainly due to that the smaller and more uniform BBK fibrils contributed to a tighter structure of the film (Fig. 6). Moreover, the enzyme pretreatment can significantly increase the densities of films from BBK and BSK fibrils.

The thermal stability of cellulose nanofibers is also a key property for their potential application as reinforcing agent in bio-composites. Thermal stability of both BBK and BSK fibrils was investigated by the thermogravimetric method (Chancelier et al. 2014; Nair and Yan 2015). Figure 8 shows the thermogravimetric analysis (TGA) and the derivative thermogravimetric curves (DTG) of BBK and BSK fibrils. The details of more apparent degradation temperatures (Tonset) and maximum degradation rate temperatures (Tmax) are listed in Table 3.

Fig. 8
figure 8

TGA and DTG curves (inset graphs) for the fibrils prepared by (a) BBK after grinding for 2 h, b BBK with enzyme pretreatment followed by grinding for 2 h, c BSK after grinding for 2 h and (d) BSK with enzyme pretreatment followed by grinding for 2 h

Full size image
Table 3 Tonset and Tmax of the fibrils prepared by (a) BBK after grinding for 2 h, (b) BBK with enzyme pretreatment followed by grinding for 2 h, (c) BSK after grinding for 2 h and (d) BSK with enzyme pretreatment followed by grinding for 2 h
Full size table

The Tonset and Tmax respectively appeared at around 314.6–327.3 °C and 350.4–361.2 °C for all samples. It suggested that the thermal stability of BBK and BSK fibrils is not significantly different. Also, enzyme pretreatment had no negative effects on the thermal stability of cellulose nanofibers.

Conclusions

Bagasse fibers can be a promising alternative for nanofiber production. Bagasse fibers have smaller sizes and higher content of hemicellulose which facilitate mechanical disintegration of fiber wall, and thus requires less energy in nanofiber production than softwood fibers (21.05 kWh/kg vs 22.71 kWh/kg). Enzyme pretreatment further reduced the energy consumption for mechanical defibrillation by 59.71%. Due probably to the same reason, the nanofibers produced from bagasse had more uniform sizes, which led to a tighter structure of film made from them with higher transparency. The mechanical strength of nanofiber film from bagasse was even higher than that from softwood fibers. Also, nanofiber film from bagasse fibers has a similar thermal stability with that from softwood fibers.