Introduction

In recent years, developing materials from natural polymers to substitute conventional synthetic materials has attracted the attention of the scientific community, owing to the considerable impact caused by the inappropriate manner in which such polymers are discarded in the environment (Mohanty et al. 2000; Yu et al. 2006; Russo et al. 2005). Conventional plastic films, such as polypropylene (PP), low density polyethylene (LDPE) and polyvinyl chloride (PVC), which are obtained from synthetic raw materials, have been largely used in agricultural sectors, in plastic packaging for foods (Zhu et al. 2014; Kirwan and Strawbridge 2003; Ishihara 2002; Marques et al. 2002), greenhouses, bags for seedlings, plastic cover for soil or much-films (Kijchavengkul et al. 2010), bagging and covering of fruits (Murray et al. 2005; Moldão-Martins et al. 2003), covering of seeds (Almeida et al. 2005), and system for controlled release of nutrients (Liang et al. 2009). It is estimated that approximately 10.6 million tons of plastic was discarded in 2015 (Plastics Europe 2015). From 2017 to 2018, it is estimated that there will be an increase of about 4.0% in plastic consumption by the sector (Plastics Europe 2017). In 2015, approximately 322 million of tons of synthetic polymers was produced worldwide (Plastics Europe 2016); the agricultural and the plastic packaging sector were responsible for about 3.3 and 39.9% of this global production, respectively. In Brazil, about 6.3 million of tons of plastic was produced in 2014 (ABIPLAST 2016), whose 53% could be attributed to the sectors of packaging and discardables. According to Plastics Europe, the most produced resins in the world during 2015, with a total of 49 million tones, were PP (~ 9.2 million tons), linear low-density polyethylene—LLDPE and LDPE (~ 8.1 million tons), and PVC (~ 4.9 million tons) the most consumed synthetic plastics.

Biodegradable polymers such as polylactic acid (PLA), polycaprolactone (PCL), polyhydroxybutyrate (PHB) and some commercial brands like Bioplast, Ecoflex and Mater-Bi are of interest for use in the agricultural sector and in the plastic packaging; however, their applicability as a substitute to petroleum-based polymers is hampered by their high costs. Polymers such as PCL, PLA and Ecoflex may be fourfold the cost of synthetic polymers like LLDPE, PP and PVC, which cost US$2.00 per Kg on average (MSR&PC 2014). Natural fibers and biopolymers, such as starch, can be considered as a cost-effective alternative for development of new biodegradable composites; however, for composite film applications, some limitations still have to be overcome, such as the poor mechanical properties, high solubility in water, and high permeability to water steam (Vieira et al. 2011; Khan et al. 2014).

The use of polymeric blends for the production of new sustainable materials with improved properties is a technological option with high potential for industrial utilization (Olabarrieta 2005). Some aspects such as low cost and the possibility of combining polymers with additional properties have caught the attention of many researchers (Guimarães Junior et al. 2015a; Utracki 1989; Ishiaku et al. 2002). In this sense, poly(vinyl alcohol) (PVA) is an interesting alternative since it presents a good capacity to form films, is soluble in water, is biodegradable, and has polar character as most polysaccharides (Khan et al. 2006; Cinelli et al. 2006; Follain et al. 2005; Siddaramaiah et al. 2004; Zhai et al. 2003).

Cellulose nanofibrils or nanofibers (CNFs) are fibrillar units resulting from the linear combination of cellulose chains in amorphous and crystalline fractions (Tonoli et al. 2016; Bufalino et al. 2015a; Elazzouzzi-Hafraoui et al. 2008; Podsiadlo et al. 2005; Helbert et al. 2004). The large surface area in conjunction with the high aspect ratio and high capacity to form nanostructured entangled networks are some characteristics that make nanofibrils excellent low cost and non-toxic reinforcing agents (Mirmehdi et al. 2017; Arantes et al. 2017; Fonseca et al. 2016; Guimarães Junior et al. 2015b; Chang et al. 2010; Petersson et al. 2007) in comparison to their counterparts in macro scale. Higher values of optical transmittance, possibility of adjustment in optical parameters, high flexibility, high mechanical performance, low density (Scatolino et al. 2017a; Zhu et al. 2013; Bufalino et al. 2015b) and low coefficient of thermal expansion (Habibi et al. 2010), are some particularities observed in nanocomposites obtained from the insertion of cellulosic nanofibrils in thermoplastic polymers. The great advantage related to the feedstock for production of this nano-reinforcement is that cellulose is the most abundant, renewable, and biodegradable polymer existing in nature (Mathew et al. 2006), with approximately 1 trillion tons of biomass available in the world (Eichhorn et al. 2010; Ioelovich and Leykin 2008). Nevertheless, despite there is available information on obtention of nanofibrils from lignocellulosic materials in literature, their efficient production regarding to energy consumption during defibrillation remains a challenge. The complex multilayered structure of the plant fibers and interfibrillar hydrogen bonds are the reason for this high-energy consumption for their deconstruction (Abe et al. 2007).

Therefore, the present work aimed to evaluate the influence of adding bamboo cellulose nanofibrils (with 5 or 30 passes through the mechanical defibrillator) on the performance (thermal, structural, mechanical and physical properties) of poly(vinyl alcohol)—PVA and modified cassava starch—FMM blend nanocomposites.

Experimental procedures

Materials

Commercially available refined bamboo pulp (Bambusa Vulgaris Schrad around 2 years old, CEPASA S/A, Brazil) was received in wet form and used as raw material in the preparation of nanofibrils; the average length and diameter of the fibers were around 2098 and 12 µm respectively (as reported by Guimarães Junior 2011). The pulps were produced using the soda-anthraquinone process (pH between 12 and 13) and refined in a disk mill (Schopper Riegler between 25 and 30). Modified commercial cassava starch—FMM (Cargill, type A, with crystallinity index and amylopectin content of around 45 and 85%, respectively, as reported by Guimarães Junior et al. 2015c) and PVA (obtained from Sigma-Aldrich, degree of hydrolysis of 99% and molecular weight of 130.000 g/mol) were also used for the production of the polymeric matrix.

Isolation of commercial bamboo cellulose pulp

Before obtaining the bamboo nanofibrils, with the aim of decrease the energy consumption during mechanical nano-fibrillation, commercial cellulose from refined bamboo pulp was isolated. Refined pulp was obtained by the soda-anthraquinone process using about 18% sodium hydroxide and 0.03 g anthraquinone per 100 g of solid mass included in the solution with pH between 12 and 13. The pressure inside the industrial biodigestor was about 7 bars, at an average temperature of 170 °C for 45 min. The refining process was performed industrially in a disk refiner until reach a Schopper Riegler index (that is the rate of drainability of a dilute suspension of pulp) between 25 and 30. After that, the refined pulp was subjected to an alkaline treatment (Corradini et al. 2006) followed by bleaching (Pereira et al. 2010), both performed twice, in laboratory scale, with some modifications. However, the pH was maintained near to 11 in order to prevent the whiteness reversion, according to Halpern (1975). The cellulosic pulp was washed in distillated water and immersed in NaOH 5% (w/w) aqueous solution, in the proportion of 1:20 (solid mass of pulp:NaOH solution), at 70 °C for 60 min under mechanical stirring at 200 rpm. After this procedure, the sample was washed in distilled water and dried at room temperature (~ 25 °C). The bleaching treatment was performed in solution containing NaOH 4% (w/w) e H2O2 30% (v/v) in the ratio 1:1, under mechanical stirring at 2500 rpm at 65 °C for 120 min. The ratio of pulp mass by bleaching solution was 1:20. Samples were washed with tap water until the pH became neutral, and then the samples were dried in an air-circulated oven at 60 °C for 24 h for subsequent chemical characterization. Figure 1 shows the industrial process for obtaining refined bamboo pulp (Fig. 1a) and the purification procedures performed in laboratory (Fig. 1b).

Fig. 1
figure 1

Treatments of the bamboo fiber from the native bamboo stem to obtain bleached cellulosic pulps: a industrial processes; and b treatments performed in laboratory

Full size image

Production of nanofibrils

The cellulose pulp chemically treated was diluted to 1.2 wt% in a water suspension, which was allowed to stand for 48 h to hydrate the cellulose and guarantee the swelling of the fiber cell wall (Alila et al. 2013; Missou et al. 2013); this suspension was then subjected to defibrillation (Supermasscolloider MKCA6-3, Masuko Sangyo, Japan). The rotational speed was set at 1500 rpm (Ifuku et al. 2010) and the operation was repeated 5 and 30 times (passes) according to Guimarães Junior et al. (2015b). The number of passes (or cycles) adopted for this study was selected to demonstrate the influence of the minimum and maximum number of passes of the nanofibrils suspensions through the defibrillator in the deconstruction of cell wall of the fibers and interactions with the polymers of the matrix. The gap between the two disks was reduced to 100 µm, from the zero position whose the disks begin to rub (Nakagaito and Yano 2004). After this defibrillation process, the obtained suspension was sonicated for 30 min using a Brandon sonicator with a 13-mm tip at 450 W and 20–20 kHz. The suspensions will hereafter be referred to as NRB5x and NRB30x. Figure 2 shows the steps involved in the production of the bamboo nanofibrils.

Fig. 2
figure 2

Production of bamboo nanofibrils from the bleached pulp with the obtention of a gel after 5 and 30 passes through the mechanical defibrillator

Full size image

Preparation of blend films

The production of blend films was based on a solution casting and evaporation method, using water as a solvent, according to the methodology proposed by Guimarães Junior et al. (2015c). Different concentrations of FMM and PVA were prepared and analyzed for the purpose of finding the best film-forming composition, and this optimal composition was used for this work (i.e. 3 wt% for FMM film and 4 wt% for PVA film). The glycerol content for the FMM and PVA films was 12 and 25 wt%, respectively.

Solutions A (FMM) and B (PVA) were mixed together and homogenized for 5 min at room temperature using a digital homogenizer (IKAT-25) at 15,000 rpm, and then the mixture was subjected to an ultrasonic bath to facilitate the removal of air bubbles. Different proportions of FMM were added to the PVA solution to obtain film blends (100/0; 80/20; 60/40; 50/50; 40/60; 20/80 and 0/100). The blends containing 20 wt% of FMM and 80 wt% of PVA (80P20A) was chosen, according to Guimarães Junior et al. (2015c).

After defoaming in an ultrasonic bath, the solutions were poured into acrylic plates (40 g) of 150 mm diameter. The thickness was controlled through the mass of the samples, that were conditioned for drying in an acclimatized room (20 ± 3 °C and 60% relative humidity—RH) for 7 days. The samples were then inserted into a desiccator at around 43% RH for 3 days before being tested. The steps used for the preparation of the films and the blends are depicted in Fig. 3.

Fig. 3
figure 3

Production of films of poly(vinyl alcohol) (PVA) (a, d, e, h and i); modified commercial cassava starch (FMM) (b, d, e, f and g) and PVA and FMM 80P20A blend (g, h, j, k, l and m) after choosing the correct concentration of plasticizer and filmogenic solutions

Full size image

Preparation of the nanocomposites

The choice of the 80P20A blend was made based on the results of mechanical, physical, thermal, micro structural and morphological properties, in accordance with studies conducted by Guimarães Junior et al. (2015c). The suspension of cellulose nanofibrils was previously dispersed in distilled water at a concentration of 0.15 wt%. The suspension was mixed in a high-speed homogenizer set at 15,000 rpm for 5 min followed by sonication (450 W) for about 30 min using a 13-mm tip at 25% amplitude. The suspension of cellulose nanofibrils was added to the FMM/PVA blend, which was continuously stirred (at 750 rpm) in a magnetic stirrer and heated at about 80 °C. After the suspension became homogeneous and viscous (about 15 min), it was poured into acrylic plates (40 g) with a diameter of 150 mm. The acrylic plates with suspensions were conditioned in an acclimation room (20 ± 3 °C and 60% RH) for 7 days of drying. Nanocomposites with thickness of approximately 90 µm were obtained and conditioned at 43% RH using a saturated solution of K2CO3 at 25 °C for 10 days before testing. The solution cast of the FMM/PVA blend with cellulose nanofibrils was made with 6.5% nanofibrils (by mass) obtained with 5 and 30 passes through the mechanical defibrillator; which were named 6.5%5x and 6.5%30x, respectively. This content was chosen according to preliminary results obtained during SEM and XRD analyses followed by the values obtained of tensile strength (TST) and elongation at break (EAB), both related to tensile properties. The results show that contents of nanofibrils lower than 6.5% (0.5, 1.5 and 4.5%), obtained both for 5 and 30 passes through the mechanical nano-fibrillator promoted the presence of fissures, delaminations and some erupted bubbles due the formation of agglomerates of nanofibrils. The XRD analyses show a progressive reduction in the values of crystallinity indexes of the nanocomposites with nanofibril contents lower than 6.5% for 5 and 30 passes, in relation to the control blend. It is known that lower crystallinity of the film may lead to lower tensile and perforation strength, higher opacity and higher susceptibility to gas permeability. For this reason, the content of 6.5% was chosen for the present work. Figure 4 shows the steps used for the preparation of the FMM/PVA blend reinforced with cellulose nanofibrils.

Fig. 4
figure 4

Production of PVA and FMM 80P20A blends (a, b) reinforced with bamboo cellulose nanofibrils obtained from mechanical defibrillation, with concentration of 6.5% after 5 and 30 passes through the defibrillator (d, e, f) for obtention of the nanocomposites (c, g, h, i, j). 80P20A: blend containing 80% of PVA and 20% FMM; 6.5%5x: nanocomposites containing blend with 6.5% of nanofibrils obtained after 5 passes through the defibrillator; and 6.5%30x: nanocomposites containing biodegradable blend with 6.5% of nanofibrils obtained after 30 passes through the mechanical defibrillator (grinder)

Full size image

Characterization

Chemical characterization of the bamboo pulps

The chemical composition of the refined pulps, with and without treatment, was determined. Sample preparation for chemical analysis was carried out according to the procedures described in TAPPI T 257 cm-97 (1997a) and TAPPI T 264 cm-97 (1997b)   standards. The holocellulose fraction was determined according to the guidelines outlined in TAPPI T 9 m-94 (1994) standard. Alpha-cellulose and total lignin contents were delimited according to TAPPI T 203 cm-99 (1999) and TAPPI T 222 cm-98 (1998a), respectively. The ash and extractive contents were defined according to TAPPI T 211 cm-93 (1993) and TAPPI T 204 cm-97 (1997c), respectively. The content of hemicelluloses was calculated from the difference in the values of holocellulose and alpha-cellulose. Determinations were performed in triplicate. The native fiber and refined pulp without bleaching were called NAFI and PORT, respectively, and the pulp subjected to the alkaline treatment followed by bleaching (applied twice) was called PRMD and PRMDBD respectively.

Characterization of cellulose nanofibrils

Atomic force microscopy (AFM) was carried out using an Agilent 5500 N9410S microscope. A drop (~ 5 µL) of the dilute suspension (approximately 0.05 g/L) was deposited on the mica substrate and dried at 60 °C for 12 h. Areas of 5 µm × 5 µm were scanned in dynamic mode at room temperature at a scanning rate equal to 1 Hz, using silicone tips mounted on a silicon cantilever with a nominal radius of about 10 nm with a spring constant of 42 N/m. The results were analyzed using the 64-bit Gwyddion software (Kaboorani et al. 2012). Around 100 nanofibrils were randomly chosen for each condition, and two height measurements were used for determination of the average diameter (Silvério et al. 2013). The average surface roughness was determined using the parameter Ra in Eq. (1). The roughness (Ra) was evaluated in order to estimate the dispersion of the nanofibrils over 5 and 30 passes through the defibrillator.

$$Ra = \frac{1}{n}\sum\limits_{i = 1}^{n} {Zi - Zm}$$
(1)
$$Zm = \frac{1}{n}\sum\limits_{i = 1}^{n} {Zi}$$
(2)

where Ra (average roughness) is the arithmetic average of the absolute deviation values from the surface height. Zm (Eq. 2) is the average height of the sample points, and Zi is the height of each sample point. The number of points within each area is represented by n.

Crystallinity index of the cellulose nanofibrils (nanof.30x) were estimated by X-ray diffraction (XRD) using a LabX XRD-6000 Shimadzu diffractometer at room temperature, step of 0.05° and an integration time of 1 s in a 2θ range from 3° to 45°. The radiation used was CuKα (1.5406°), which was generated by 40 kV and an incident current of 30 mA in Bragg-Brentano geometry. The aqueous solutions of cellulose nanofibrils (20 mL of a 1.15% w/v suspension) were lyophilized under vacuum of 20 to 40 µHg, for approximately 48 h at around − 25 ± 5 °C (Guimarães Junior et al. 2015b). A press was used to obtain 13 mm diameter pellets (the same dimension as the diffractometer sample holder). The mass of each lyophilized samples was approximately 70 mg, and they were pressed for 1 min at approximately 150 MPa, as used in Guimarães Junior et al. (2015b). Deconvolution of the XRD peaks was applied for determination of their crystallinity indexes, using the Origin Pro 9.3® software to separate the non-crystalline and crystalline contributions of the diffraction spectrum. For the aqueous suspension of cellulose nanofibrils obtained through the mechanical defibrillation process (30x), the fitting method was performed based on the monoclinic unit cell investigated and explained by Nishiyama et al. (2002). It was done in accordance with the modern nomenclature (c as the fiber axis) (French 2014), with reflection peaks corresponding to (1-10), (110), (200) and (004) planes as reported in Flauzino Neto et al. (2016); Kafle et al. (2014); Xiang et al. (2016); Driemeier et al. (2015) and Panthapulakkal and Sain (2012). The chosen shape function for fitting the X-ray diffractograms was the Pseudo-Voigt 2 function using least squares fitting as performed in Garve et al. (2005). The crystallinity index (CI) was calculated from the ratio between the area below all the crystalline peaks and the total area below the whole curve, determined after deconvolution (including non-crystalline fraction) following the Eq. (3):

$$CI\left( \% \right) = \frac{{\sum A_{Crystalline\,Peaks} }}{{\sum A_{Crystalline\,Peaks} + A_{amorphous\,halo} }} \times 100$$
(3)

Characterization of the films/nanocomposites

A DSC-60 Shimadzu equipment was used to measure the crystallization and melting temperatures and variations in the enthalpies of melting and crystallization of the film samples (blends and nanocomposites). The samples were analyzed in an aluminum pan (not sealed) using approximately 3 mg of material by heating from 30 to 240 °C; this temperature was maintained for 3 min, then the sample was cooled to 50 °C and again heated to 240 °C at heating and cooling rates of 10 and − 10 °C/min, respectively, under flow of around 50 mL/min. In order to eliminate the thermal history of the samples, the thermal properties were obtained during the second heating cycle.

The infrared spectra of the film samples were analyzed using Fourier transform infrared spectroscopy (FT-IR) in attenuated total reflectance mode (FT-IR/ATR-301); they were used to observe the evolution of the chemical modifications of the samples. The experiments were carried out in the range of 400 to 4000 cm−1 with 32 scans and a resolution of 2 cm−1. The test was conducted at ∼ 20 ± 3 °C.

Tensile strength (TST), elongation at break (EAB) and tensile modulus (TMO) were evaluated according to ASTM D882-00 (2000) in an INSTRON model 5966–E2 instrument with a 1 kN load cell, deflection rate of 50 mm/min, and an initial distance of 50 mm between the clamping jaws. Test film samples with 15 mm width and 100 mm length were used, and the TMO was calculated from the tangent of the initial linear function of the stress-strain curve. The test was conducted at 23 ± 2 °C and 5 repetitions were carried out for each sample type, resulting in fifteen repetitions. Before testing, the film samples were conditioned at 50% RH and 25 °C for 48 h. The thickness of each specimen was measured at 5 points along its length with a digital micrometer.

Crystallinity indexes of the PVA film, 80P20A blend and nanocomposites (5x and 30x) were estimated by X-ray diffraction (XRD) using a Shimadzu LabX XRD-6000 diffractometer at room temperature, step of 0.05° and an integration time of 1 s in a 2θ range from 3° to 60°. The radiation used was CuKα, which was generated by 40 kV and an incident current of 30 mA in Bragg-Brentano geometry. Flat pieces of the film samples were cut and fixed on the glass sample holder. Deconvolution of the XRD peaks was applied for determination of their crystallinity indexes, using the Origin Pro (2017) software to separate the non-crystalline and crystalline contributions of the diffraction spectrum.

For the pure PVA film, 80P20A blend and nanocomposites with 6.5% nanofibrils, the diffractogram was performed using the monoclinic unit cell with atactic structure (the polymer chains are described as found along the b-axis of the unit cell), with peaks of reflections assigned to (100), (10-1), (200) and (202) planes, according to Bunn (1948); Takahashi (1997) and Ricciardi et al. (2004).

The chosen shape function for fitting the X-ray diffractograms was the Pseudo-Voigt 2 function using least squares fitting as performed in Garve et al. (2005). The crystallinity index (CI) was calculated from the ratio between the area below all the crystalline peaks and the total area below the whole curve, determined after deconvolution (including non-crystalline fraction) following the same Eq. (3) reported in the previous section.

A Shimadzu SSX-550 scanning electron microscope (SEM) operating in a high vacuum chamber at an acceleration voltage of 15 kV was used to analyze the surface morphology and also the fracture surface of the blends and nanocomposites. A secondary electron detector was used to capture the images. A nanolayer of gold particles was deposited on the film surfaces for 2 min at 330 V and 20 mA in a physical vapor depositor (PVD).

Film thickness (in µm) was measured with a digital micrometer fast-forward (0–30 mm) with 0.001 mm resolution (Mitutoyo Manufacturing Co. Ltd., Japan). It was conducted 20 random measures for each sample (Choi and Han 2001).

Apparent density of the films (in g/m3) was calculated by dividing the film dry mass by the film volume (Park and Zhao 2004). It was used seven samples (30 × 30 mm2), conditioned at 0% (RH) for drying for 2 weeks.

Grammage (in g/m2) of the films was determined as suggested by TAPPI T 410 om-98 (1998b) standard. The prepared samples were placed in a conditioned environment at 23 ± 2 °C and 50 ± 2% RH (TAPPI T 402 om-98, 1998c) before weight and area determinations by triplicate using samples of nominal dimensions of 100 mm x 100 mm.

Water absorption (%) of the films was determined according to guidelines specified in ASTM D570-98 (2000). Circular samples with 50 mm diameter were dried for 24 h at 50 ± 3 °C. After drying, the initial weight was determined (M0), and the samples were immersed in a vessel containing about 300 mL of distilled water for 24 h at 23 °C. The samples were dried, and the final weight was determined (MT). The water uptake of the samples was calculated by dividing the gain in weight (MT − M0) by the initial weight (M0). Five replicates were tested.

Statistical analysis

Sisvar 5.0 was used for carrying out statistical analyses of the data through comparison of the averages. Fisher’s least significant difference and a Scott-Knott test were used at the 95% confidence level for mechanical and physicals tests, respectively.

Results and discussion

Bamboo pulps

Chemical characterization

The chemical composition of the present samples is similar to the values reported by Yoshizawa et al. (1991) and Chen et al. (1998) for the same studied species and similar to the reported values for other bamboo species for the same age range (Wahab et al. 2013; Li et al. 2007). The refined bamboo pulp that was chemically pre-treated (PRMDBD) presented the lowest percentage of hemicelluloses, total lignin and extractives and the highest percentage of holocellulose and alpha-cellulose, as compared to the native fiber (NAFI). The holocellulose and alpha-cellulose relative contents of the refined bamboo pulp chemically pre-treated (PRMDBD) increased of around 42 and 112%, respectively, as compared to the native fiber (NAFI). Similarly, the hemicelluloses, total lignin and extractives contents decreased around 61; 95 and 84%, respectively, as indicated by Table 1.

Table 1 Average values of chemical composition of the native fiber and of the bamboo cellulose pulps in the different conditions
Full size table

The increased alpha-cellulose content after chemical pre-treatment confirms the removal of the main amorphous components of the fiber (lignin, pectins, and hemicelluloses), demonstrating the effectiveness of the pre-treatments used. It is expected that the pre-treatments lead to: (1) hydrolysis of hemicelluloses and cleavage of lignin-hemicelluloses bonds (Suess 2010), (2) partial solvation of hemicelluloses (Lee et al. 2014; Wang et al. 2007), (3) breakdown of linkages between carbohydrates and lignin by the saponification of intermolecular ester bonds (Lee et al. 2014; Soccol et al. 2011), (4) disrupting of the lignin structure by breaking glycosidic β-ether bonds (Lee et al. 2014), and (5) bleaching of the pulp (Correia et al. 2013). These statements are confirmed by FT-IR analyses.

Alkali extraction was carefully performed to avoid undesirable degradation of the cellulose and to ensure the occurrence of hydrolysis only at the fiber surface (Siró and Plackett 2010; Wang and Sain 2007), to maintain the intangibility of the obtained nanofibrils. It was also possible that the pre-treatments work synergistically to reduce the consumption of electric energy during the production of the nanofibrils (Zhu et al. 2014; Isogai et al. 2011).

According to literature (Ankerfors and Lindstrom 2007), the interactions between these pre-treatments can reduce the energy consumption for nanofibrils production (Siró and Plackett 2010). Eriksen et al. (2008) reported values up to 70 MWh/T, depending on the origin of the fiber. This reduction of energy consumption favored by the pre-treatments, occurred due to the breakdown of interfibrillar hydrogen bonding networks in the fiber cell wall, with subsequent release of the nanometric fibrils. Results also suggest that the removal of lignin, without the solubilization of hemicelluloses, contributed greatly to reduction of energy consumption. According to literature (Chaker et al. 2013, 2014; Iwamoto et al. 2008), the higher the hemicelluloses content in the fibers, the easier the nano-fibrillation, since this short chain polysaccharide can act as an inhibitor of the microfibrils coalescence (Iwamoto et al. 2008). The greater removal of lignin in detriment of hemicelluloses, can also be attributed to the presence of anthraquinone in the pulp, which is an organic additive with selective delignification capacity (Gierer 1980; Silva et al. 2002; Oksanen et al. 2009). It stabilizes the polysaccharides and promotes the intramolecular cleavage of aryl ether linkage in the lignin structure (Suckling 1988).

The reduction in the amount of extractives of the cellulosic pulps can have favored the tensile mechanical properties and the thermal stability of the nanocomposites in relation to the reference matrix 80P20A, since, at higher temperatures, the extractives can negatively affect the synergism between the elements of the compound (80P20A blend/cellulose nanofibrils), hindering the good adhesion between them. Due to its lower molecular weight, as comparison to cellulose, the highest content of these components can also promote the reduction of the thermal stability of the nanocomposite accelerating its degradation process (Poletto et al. 2012; Shebani et al. 2008).

Bamboo nanofibrils

Sedimentation time

The evaluation of the changes on sedimentation time of the nanofibrils suspensions after mechanical defibrillation followed by sonication was performed through visual inspection after 72 h. The concentration of the suspensions was adjusted to 0.25%wt and they were then transferred to test tubes to be photographed.

Morphological characterization

Figure 5a, c exhibit AFM images of the nanofibrils obtained after 5 and 30 passes, respectively. A great difference is observed after 30 passes (Fig. 5c); the pulp has a high degree of defibrillation compared to the nanofibril suspension obtained after 5 passes (Fig. 5a) through the mechanical defibrillator. Notably, in Fig. 5a, the nanofibrils are not homogeneous, including bundles of nanofibrils and microfibrils (Kaboorani et al. 2012). From the images in Fig. 5a, c, e, f, it is observed the changes in the morphology of the nanofibrils after 30 passes. The result shown by Fig. 5e, f clearly indicates that effective nano-fibrillation has occurred for fibers of the bleached pulps. The high mechanical shearing action produced through the increasing number of passes from 5 to 30 promoted a reduction in the average fibers diameter of about 90%, which increases the stability of the nanofibril suspensions. The results suggest that the interfibrillar hydrogen bonds were broken-up with greater effectiveness after 30 passes, facilitating the deconstruction of the cell walls and releasing of the nanofibrils (Chaker et al. 2014). Figure 5f confirms the fiber deconstruction, once about 67% of the measured nanofibrils presented average diameters between 5 and 15 nm. The opposite occurred with the nanofibrils after 5 passes. A considerable amount of large pieces of fiber cell wall still remain in the suspensions. The average diameter of the cellulose nanofibrils was of about 80 nm. Furthermore, it is observed that approximately 25% of the measured nanofibrils possess average diameters between 40 and 60 nm (Fig. 5e). The morphological results of the nanofibrils are consistent with literature (Karimi et al. 2014; Li et al. 2014; Tonoli et al. 2012). The average values of Ra (surface roughness) found for the mica surface covered with the nanofibril suspensions with 30 passes were quite low (approximately 1 nm) as shown in Fig. 5f; which demonstrates the high quality of the nanofibril dispersion (Kaboorani et al. 2012).

Fig. 5
figure 5

Typical atomic force microscopy (AFM) topography images of the cellulose nanofibrils suspensions: a 5 passes and c 30 passes through the defibrillator; b, d sedimentation of aqueous suspensions formed by bleached bamboo cellulose nanofibrils (NBR), chemically and mechanically pre-treated after 5 and 30 passes (NRB5x and NRB30x) through a mechanical defibrillator; e, f diameter distribution (determined from measurements in the AFM images) of the bamboo nanofibrils after 5 and 30 passes through the defibrillator, respectively. The average diameter of the nanofibrils and roughness (Ra) of the mica surface covered with nanofibrils are presented. Scanning area: 25 µm2. The sedimentation time of the nanofibril suspension in (b) and (d) was 72 h

Full size image

X-ray diffraction

The X-ray diffractogram (Fig. 6a) revealed a relatively ordered structure with a narrow peak at 2θ ~ 22.3° corresponding to the (200) plane and a diffuse peak between 14.5° and 16.2° corresponding to the (1-10) and (110) planes of cellulose. The pattern also showed (Fig. 6a) the characteristic peak of cellulose at 2θ ~ 34.6° due to reflection from (004) lattice plane (Nan et al. 2016; Tan et al. 2015; Sassi and Chanzy 1995). The sharper diffraction peak at 2θ ~ 22.3° for (200) plane indicates it is the strongest peak arising from the crystals of cellulose. The findings are in agreement with observations reported by earlier researchers (Chen et al. 2011; Quddiani et al. 2011). The present diffraction peaks are representative of typical cellulose I (allomorph Iβ) structure with absence of cellulose II, which arises from the fact that there is no doublet in the intensity of the main peak (Adel et al. 2011). The overlapping of (1-10) and (110) peaks also suggest that the structure of the nanofibrils corresponds to cellulose I (Besbes et al. 2011; Wada et al. 2004). According to French and Cintrón (2013), the overlapping of these (1-10) and (110) planes result in a single broad peak, due to the large full width at half maximum (FWHM). This behavior may have an inverse relation with the crystallite size that form the cellulose nanofibrils (Nan et al. 2016; Tonoli et al. 2016; Oliveira and Driemeier 2013) as defined in Scherrer equation (Langford and Wilson 1978).

Fig. 6
figure 6

(a) Typical X-ray diffraction patterns of nanofibrils 30x; (b) circular test piece of 13 mm diameter from cellulose nanofibrils lyophilized after pressing of approximately 150 MPa (left) and aqueous solution of cellulose nanofibrils lyophilized (right)

Full size image

The crystallinity index (CI) of the cellulose nanofibrils suspension obtained after 30 passes through the mechanical defibrillator was about 58% (Figs. 6,8b). The increase in the number of passes (from 5x to 30x) resulted in higher rates of defibrillation and caused a decrease of the CI of the nanofibrils of approximately 13% (Guimarães Junior et al. 2015b). It is attributed to the disruption of the crystalline regions or a reduction in crystal size during defibrillation (Iwamoto et al. 2007, 2008). It indicates that there was an increase in the interlamellar distances, consistent with a less ordered crystallite structure indicating a certain disorder in the lattice and lower amount of crystalline phase, probably due to the process of defibrillation, which affected the crystalline domains more than the amorphous fraction. According to Kim et al. (2010); Moon et al. (2011) and Poletto et al. (2014), this result indicates a relative change of crystallite size caused by the action of hydrodynamic forces associated with high shearing forces during the isolation of the nanofibrils.

The small peak at 2θ ~ 34.6° refers to the (004) plane for cellulose I (Fig. 6). Tonoli et al. (2016), investigating micro/nanofibrils obtained from eucalyptus pulp fiber treated with anaerobic digestate and high shear mixing, demonstrated that this peak (2θ = 34.6°) became less pronounced after hydrolysis and more pronounced after mechanical defibrillation. The CI value calculated here for bamboo nanofibrils obtained after 30 passes through the mechanical defibrillator (30x) is probably somewhat lower than the true value. This is because the presence of mostly small peaks calculated from the coordinates in the crystal structure in cellulose I diffraction pattern (French 2014) were not considered in the fittings presented in Fig. 6a. Also, it did not predict the peaks overlap, which may have been caused by more advanced FWMH values, thus increasing the minimum intensity in the region related to amorphous fraction (2θ between 18° and 22°). Those effects were not compensated in the present work, although the diffraction patterns were processed as suggested in recent studies (French and Cintrón 2013; Nan et al. 2016; Correia et al. 2016; Ju et al. 2015) for reliable determinations of the cellulose polymorphs in perfect crystals.

Blends and nanocomposites

Differential scanning calorimetry

Figure 7 presents the DSC curves for the nanocomposites 6.5%5x and 6.5%30x. There was a considerable increase of melting temperature and enthalpy for the nanocomposite 6.5%30x in relation to the plain blend and nanocomposite with nanofibrils obtained with 5 passes. The melting temperature increased by approximately 6 °C, while the melting enthalpy increased 23% (from 35 to 43 J/g) in relation to the control blend. Such behavior presumes the formation of more perfect crystals with greater dimensions, containing lamellae with similar thickness (Liu and Donovan 1995). According to Sin et al. (2010), this performance resulted from a good dispersion of nanofibrils along the matrix and an efficient interfacial adhesion between the nano-reinforcement and matrix. Such synergy between the nanofibrils and the blend, according to Mandelkern (2004), led the nanofibrils to act as nucleating agents in the PVA/FMM matrix. The same behavior occurred with the crystallization temperature and enthalpy. The crystallization temperature increased by approximately 2 °C, while the enthalpy increased by 55%. The temperature increases strongly, which indicates that the disperse phase (nanofibrils) acted as a nucleating agent (Averous and Le Digabel 2006), while the increase of crystallization enthalpy suggests an increase of nanocomposite crystallinity as a function of the nanofibril concentration.

Fig. 7
figure 7

Typical DSC curves obtained for the control blend (80P20A) and nanocomposites with 6.5% of nanofibrils with 5 (6.5%5x) and 30 (6.5%30x) passes: a after the second heating cycle; and b after the second cooling cycle. TM: melting temperature (°C); TC: crystallization temperature (°C); ΔHM: melting enthalpy (J/g); ΔHC: crystallization enthalpy (J/g)

Full size image

The higher crystallinity is an important criterion for improving mechanical properties of nanocomposites, and according to Vieille et al. (2013) has a direct correlation with rigidity and strength of the composites, which might explain the high tensile properties of nanocomposites developed in this work. Similar results were found by Coleman et al. (2004), who applied nanotubes as reinforcing agents in a PVOH matrix; Srithep et al. (2012), who used nanofibrillar cellulose to reinforce PVOH matrix; and Mathew et al. (2008), who applied tunicin nanowhiskers to a plasticized starch matrix. According to Savadekar and Mhaske (2012), the increase of crystallinity is influenced by the nanofibrils reinforcement that eases the formation and arrangement of crystals in the crystalline phase. Under the conditions of this study, it seems that the nanofibrillated material obtained after 30 passes through the defibrillator acted as a nucleating agent in the 80P20A blend, leading to the higher crystallinity index of the 6.5%30x nanocomposite (Fig. 8c). Such behavior was consistent with the experiments of Luz et al. (2008), who noticed that the addition of lignocellulosic nucleating agents alters the crystallization of the matrix around the fiber.

Fig. 8
figure 8

Typical X-ray diffraction patterns of a nanocomposite 6.5%30x, showing the peak deconvolution method; b starch film, nanofibril with 30 passes (30x), nanocomposite 6.5%5x, 80P20A blend, PVA film and nanocomposite 6.5%30x. Miller indexes of the nanocomposite 6.5%30x are (100), (10-1), (200) and (202). Miller indexes of the nanofibrils with 30 passes are (1-10), (110), (200) and (004). CI (%): crystallinity index; (30x): 30 passes through the defibrillator

Full size image

Results in the present work corroborate with the evidence of a more stable, compact and homogeneous structure, whose nanofibrils acted as nucleating agents, as observed in the cooling DSC scans in Fig. 7b, in the FT-IR absorption spectrum in Fig. 9, and in the micrographs presented in Fig. 12c, f. In this sense, Martin and Averous (2001) and Matzinos et al. (2002) pointed out similar conclusions. The nanocomposite obtained with the same concentration of nanofibrils, but with 5 passes through the defibrillator, exhibited a reduction in both parameters (melting temperature and melting enthalpy), as observed in Fig. 7a. The presence of large pieces of micro/nanofibril agglomerates (obtained after 5 passes) in the control matrix, as those observed in Fig. 5a, may have caused effects of steric hindrance, which is a type of repulsion that occurs when atoms are forced to stay together, thus blocking or precluding the occurrence of new reactions that may have restricted the growth of the matrix crystalline regions. This may have reduced the sample crystallinity, culminating in lower melting and crystallization temperatures. Similar behavior was observed by Jonjankiat et al. (2011) studying the properties of stickers made with PVOH and cellulose microfibrils of sugar cane bagasse. The crystallization temperature of the nanocomposite 6.5%5x also decreased, while the crystallization enthalpy increased by about 3 J/g as presented in Fig. 7b.

Fig. 9
figure 9

Typical FT-IR spectra of the control blend (80P20A) and nanocomposites with the main changes in transmittance bands of the 6.5%5x and 6.5%30x samples in the following intervals: (a) 4000 to 400 cm−1, (b) 4000 to 2750 cm−1, (c) 1808 to 707 cm−1 and (d) 3394 to 3258 cm−1

Full size image

A slight dependence of the onset temperature of crystallization and temperature corresponding to the maximum of the crystallization peak was observed as the nano-reinforcement was added to the control matrix. Such an increase in crystallization temperature reveals that both the nucleating and growth steps were affected, indicating a higher nucleating rate and higher growth rate in the crystalline phase (Luz et al. 2008); these agree with the results obtained by Guimarães Junior et al. (2015c) for the crystallinity indices of the referred nanocomposites. Similar results were found in literature (Sanchez et al. 2000; Gopakumar et al. 2002).

X-ray diffraction

The PVA diffraction pattern (Fig. 8a, b) presented four major peaks at 2θ ~ 11.2°, 19.4°, 22.4° and 40.6°, and under these peaks, ranging from approximately 10° to 70°, the broad amorphous fraction was evidenced. The average crystallinity of pure PVA film was estimated at 39%. These results are in agreement with previous investigation of similar polymeric system, whose sharp crystalline reflections, with a strong maximum reflection at 2θ ~ 19.4° and a shoulder at 2θ ~ 22°, typical of crystalline atactic PVA (Ricciardi et al. 2004; Guirguis and Moselhey 2012; Wang et al. 2008; Das et al. 2010; Jayasekara et al. 2004). The strongest crystalline reflections at 2θ ~ 19.4° correspond to the overlapping of the equatorial (10-1) and (101) reflections, respectively (Bunn 1948; Sakurada et al. 1950). Crystalline domains in semi-crystalline PVA are characterized by chains in a trans-planar conformation, packed in a monoclinic unit cell with a = 7.81 Å, b = 2.52 Å (chain axis), c = 5.51 Å and beta = 91.42° (Bunn 1948; Takahashi 1997). The peak at 19.4° shows that the crystallinity of the nanocomposites is mainly due to the insertion of cellulose nanofibrils, suggesting a molecular reorganization of their crystalline structures. This peak for the 6.5%30x nanocomposite was more prominent than that of the PVA/FMM blend, as the presence of nanofibrils increased the crystallinity of the nanocomposite by around 11% over the control 80P20A blend (Fig. 8b). With the introduction of 6.5 wt% of nanofibrils, the high crystallinity of the cellulose nanofibrils further increased the number of nucleating agents and bundled together a number of small crystallites in the composite (Liu et al. 2013), as previously demonstrated by the DSC results.

However, it was found that the crystallinity of the PVA/FMM blend with the presence of 6.5 wt% nanofibrils after 5 passes through the defibrillator was lower than the neat PVA film (Fig. 8b). The original crystallinity of the PVA matrix was not observed in the PVA/FMM blend, since the crystalline component and the non-crystalline component in a composite exhibit good compatibility, the resulting crystallinity is lower than the individual components (Li and Xie 2004). According to Chen et al. (2008) this behavior is due to the disruption of crystalline structures of the starch granules (Fig. 8b) after the gelatinization process, during blending with PVA. Thus, one can say that the semi-crystalline structure of pure PVA was partially kept upon mixing with thermoplastic starch and cellulose nanofibrils.

FT-IR spectra

The FT-IR spectra are presented in Fig. 9, and show the interactions between the matrix and bamboo cellulosic nanofibrils after 5 and 30 passes through the mechanical defibrillator. In addition, the spectra for the matrix (blend 80P20A) and components of the matrix (PVA and FMM) are also shown.

The insertion of 6.5% of nanofibrils affected the intensity of the O-H stretching peak at 3257 cm−1 (Fig. 8a, b). There was a reduction of the intensity of this band for the nanocomposite formed by nanofibrils with 30 passes, thus suggesting the occurrence of more interactions with the matrix. It reveals that this decrease may be explained by consumption of the available hydroxyl groups during reaction of the 80P/20A blend with the cellulosic nanofibrils. The higher surface area caused by the nano-fibrillation degree may have led to the formation of such interactions. In this context, the range between 1200 and 900 cm−1 presented significant changes in relation to the pure polymers (Fig. 9c) with decrease of the peak intensity. The weakening of the deformations of C–O bonds, caused by the increase of the strength of O–H interactions, may have contributed to this reduction.

PVA with high degree of hydrolysis led to the formation of strong intra and intermolecular hydrogen bonds, thus reducing the number of free hydroxyls, which caused a sharp reduction in the intensity of the vibration band at 3257 cm−1 (black) in relation to the PVA with low degree of hydrolysis, shown in Fig. 10a, b. The intensity of this band increased during blend formation and also after insertion of the nanofibrils, since stronger hydrogen bonds were formed between the components of the nanocomposites (Fig. 10). The absence of peaks at 1733 and 1713 cm−1 demonstrates the high degree of hydrolysis in this polymer, since these bands are attributed to the residual acetate group (Jayasekara et al. 2004).

Fig. 10
figure 10

Mechanism of interaction proposed for molecules of a modified cassava starch (FMM) (acetyl group incorporated), PVA and bamboo cellulose nanofibrils. The macromolecule of starch is represented by b amylopectin and amylose

Full size image

The increase of the intensity and width of bands at 1085 and 1026 cm−1 suggests partial miscibility, and under certain conditions depending on the proportion of polymers applied to the mixture and the reinforcement concentration, the compatibility of the mixture with the reinforcement is favored. No phase separation was observed between the polymers (FMM and PVA), which indicate good interaction between them. Several bands overlapped in the region between 1150 and 400 cm−1 for both PVA and FMM, mainly due to angular deformations of the C–H and O–H bonds. In this range, a significant interaction between the molecules of the blend and the nanofibrils was observed in the function of the C–OH bond stretch, thus indicating good compatibility between the polymers of the blend film. A small shift in the position of the bands may indicate the existence of interactions between functional groups of the same nature. The intensity and number of these intermolecular interactions between such polymers is a positive factor for compatibilization of this system (Barban et al. 2005; Srinivasa et al. 2003).

The small reduction in the band intensity at 1642 cm−1 (angular deformation of water) after the insertion of nanofibrils (Fig. 9c) suggests a decrease of the amount of water molecules absorbed by the nanocomposites (Fang et al. 2002) due to the strong hydrogen bonds promoted by the insertion of these nanofibrils. Regarding the use of starch, its chemical modification was confirmed by the appearance of bands at 1730 cm−1 (vibration of the carbonyl C=O group) and 1240 cm−1 (vibration of the C–O–C bond). According to Fringant et al. (1998), the presence of these two bands is characteristic of the esterification reaction of native starch.

The unchanged peaks at approximately 918 and 837 cm−1, attributed respectively to the stretching of the α-1,4 glycosidic bond, C–O–C and of the glucopyranose ring concerning the starch molecules (Musca et al. 2012), confirm the successful blending between PVA and FMM. According to Limpan et al. (2012), the presence of the band next to 847 cm−1 in both the nanocomposites (6.5%5x and 6.5%30x), attributed to the C–H vibration in the PVA polymer, may be a strong evidence of the good interaction between PVA and FMM. In contrast, their shift to larger values suggests the existence of specific interactions between the polysaccharide chains with the bamboo nanofibrils. In this sense, it is quite plausible that in nanocomposites prepared with cellulosic bamboo nanofibrils, the nano-reinforcement have been strongly adhered through hydrogen bonds to the matrix, especially those produced with nanofibrils with 30 passes. Similar results were found in literature (Mandal and Chakrabarty 2014).

Mansur et al. (2008) studied PVA hydrogels with varying degrees of hydrolysis crosslinked with glutaraldehyde, and found a strong and broad band between 3650 and 3584 cm−1, concerning to the vibration of free hydroxyl group. When the hydrogen bonds became predominant, the bands appeared at lower wavenumbers typically between 3550 and 3200 cm−1, thus endorsing the results found in the present work. Figure 9 presents the mechanisms of interaction between FMM, PVA and cellulosic bamboo nanofibrils in the nanocomposites.

Mechanical properties

Figure 11a, b show the typical stress vs. strain curves of the control blend and 6.5%5x and 6.5%30x nanocomposites during the tensile test. The most important mechanical properties for flexible blends are tensile strength (TST), elongation at break (EAB) and tensile modulus (TMO), which are directly related to the nature of the material used and with the cohesion of the formed polymer matrix (Cuq et al.1998). The high value of TST indicates an ideal performance for these films, whereas the necessity of EAB values depends on their applications. Figure 11b shows the TST, EAB, and TMO values obtained from the stress vs. strain curves (Fig. 11a). A significant reduction (p ≤ 0.05) of approximately 39% is observed for the TMO value of nanocomposites with nanofibrils from 30 passes through the defibrillator (6.5%30×), while the TST and EAB values increased significantly (p ≤ 0.05) by approximately 23 and 50%, respectively, compared to those of the 80P20A blend (Fig. 11a, b).

Fig. 11
figure 11

(a) Typical stress vs. strain curves for the blends (80P20A); nanocomposite 6.5%5x; nanocomposite 6.5%30x; (b) average and standard deviation values of tensile strength (TST); elongation at break (EAB) and tensile modulus (TMO) of the 80P20A blend and 6.5%5x and 6.5%30x nanocomposites. Lower case letters (b, c, d and e) represent statistical comparisons using Fisher’s test, while different letters indicate significant (p ≤ 0.05) differences between the samples

Full size image

The nanocomposites containing nanofibrils with higher degree of nano-fibrillation (30 passes through the defibrillator) became more flexible, presenting higher values of EAB. This higher performance of the 6.5%30x nanocomposite can be attributed to the reduction of empty spaces in the interfaces between the matrix and nanofibrils, promoted by the increase of surface area of the highly individualized nanofibrils (Xu et al. 2013). The reduction of this free volume suggests the formation of three-dimensional networks from strong hydrogen bonds between the nano-reinforcement and the matrix, which effectively promote the stress transfer from the matrix to the nanofibrils (Iwamoto et al. 2011; Bilbao-Sainz et al. 2011). The mechanical properties do not relate directly to the resistance of the nanofibrils, but to the hydrogen bonds formed between their structures. In this way, the increase of the number of passes permits a greater individualization of the nanofibrils that favors the availability of higher amounts of free hydroxyl groups. It is believed that this behavior has promoted more uniform distribution of nanofibrils into the 80P20A matrix, since the results found for the physical, thermal, structural, morphological and mechanical analyzes suggest the formation of a strong intra and intermolecular network. Moreover, this behavior indicates that the nanofibrils network is more closed and denser than the less defibrillated (5 passes) nanofibrils, presenting a reduced amount of free-volumes (Labuschagne et al. 2008; Aulin et al. 2010). This may explain the performance of the nanocomposites analyzed in this work. The results suggest that more nano-fibrillated suspensions create nanocomposites with higher tensile strength and also more flexible.

The results obtained for the tensile test are consistent with the findings obtained by visual/sedimentation inspection tests (Fig. 5b, d), AFM (Fig. 5a, c), DSC (Fig. 7), density (Fig. 13a), and FT-IR (Fig. 9) analyses of the nanocomposites, which indicated that nanofibrils obtained with higher number of passes can lead to better performance to the nanocomposites. Similar results for the tensile strength, elongation at break and tensile modulus have been reported elsewhere (Guimarães et al. 2016; Petersson et al. 2007; Lu et al. 2008).

On the other hand, the lower performance of the 6.5%5x nanocomposite is attributed to the presence of agglomerations of nanofibrils and microfibrils, which can cause fissures and cracks in their microstructure (Bilbao-Sainz et al. 2011). Probably, the lower amount of hydrogen bonds to produce three-dimensional networks may have contributed to the lower values of TST and EAB of this nanocomposites in relation to its counterpart with more nano-fibrillated suspension (30 passes). This hypothesis was also confirmed through the physical, thermal, structural and morphological tests.

The values of TST achieved for the 6.5%30x nanocomposite are analogous to those of HDPE (19–39 MPa) and PP (21–37 MPa) (Frank and Biederbick 1984), and higher than those of LDPE (24 MPa) and LLDPE (37 MPa) (Coutinho et al. 2003) which are commonly used as raw material for applications in the flexible plastic film industries, especially in the packaging and agricultural sectors (Doak 1986; Averous 2004; Auras et al. 2004). When compared to commercial biopolymers, the produced nanocomposite shows higher performance regarding TST and EAB. Table 2 shows the mechanical properties of some commercial biopolymers.

Table 2 Main mechanical properties (tensile strength: TST and elongation at break: EAB) for some commercial biopolymers
Full size table

All the biopolymers exhibit TST and EAB values lower than those presented by the 6.5%30x nanocomposite. Furthermore, these biopolymers present relatively high cost compared to petroleum-based polymers, which is undoubtedly one of the main causes for the low up-scale and utilization of these products in the polymer market. According to Bugnicourt et al. (2014), these costs can range from 7 to 10 Euro/Kg, depending on the type of biopolymer.

Scanning electron microscopy (SEM)

Figure 12a–c present SEM micrographs of the surface of the 80P20A blend and of the 6.5%5x and 6.5%30x nanocomposites, respectively. Figure 12d–f show the fracture surfaces of the specimens.

Fig. 12
figure 12

Typical scanning electron microscopy (SEM) micrographs of the film surface of: a 80P20A blend; b 6.5%5x nanocomposite; and c nanocomposite 6.5%30x. Typical SEM images of the fracture surface of: d 80P20A blend; e 6.5%5x nanocomposite; and f 6.5%30x nanocomposite

Full size image

Figure 12c shows the surface of the nanocomposite containing 6.5% of nano-reinforcement (after 30 passes) inserted into the 80P20A blend. The micrograph shows uniform, homogeneous, and dense surface without evidence of phase separation, irregularities, cracks, or nanofibril agglomerates (Kavoosi et al. 2014) as promoted by the good dispersion of nanofibrils in the 80P20A blend, as well as by the good compatibility between them (Mali et al. 2004). The increase of the specific surface area of the nanofibrils as well as the exposure of high amount of free hydroxyl groups on their surfaces are the main reasons for this improved dispersion. This fact may enhance the density of entanglements of the nanofibrils/80P20A blend (Lundahl et al. 2016; Ding et al. 2016). The cross-sections of these samples were investigated and the fibrous network clearly became closer and denser for the nanocomposite reinforced with nanofibrils after 30 passes. These results suggest that better interfacial adhesion is achieved with the adding of nanofibrils having smaller average diameters due the greater number of passes through the defibrillator. These findings were supported by the mechanical properties of this nanocomposite (Fig. 11a, b).

The 6.5%5x nanocomposite (Fig. 12e) presents the fracture surface without phase separation, but presenting small cracks, caused by large agglomerates of nanofibrils that were not completely defibrillated with the lower number of passes (5 passes) through the defibrillator. This was previously discussed with the AFM results (Fig. 5a–f), which show, among other features, that 5 passes is insufficient to convert the fibers into well-dispersed nanofibrils.

According to Jayasekara et al. (2004), the surface wrinkling related to the presence of nanofibrils and agglomerates, promoted by the drying process (during casting) in controlled environment, may have favored the appearance of cracks in the nanocomposite. This is due to the faster water evaporation on the surface of the sample, which causes the development of a stiffer surface layer in relation to the formed gel (Cerda and Mahadevan 2003; Cerda et al. 2002). According to Rizzieri et al. (2006), this layer contracted isotropically due to loss of water, while the hydrated gel contracts laterally producing internal stress that lead to cracks.

Physical properties

Average and standard deviation values of water absorption, density, thickness and grammage of the control blend and nanocomposites are presented in Fig. 13.

Fig. 13
figure 13

a Average and standard deviation values of water absorption and density; and b thickness and grammage of the control blend (80P20A) and nanocomposites reinforced with 6.5% (dry mass) of nanofibrils after 5 and 30 passes through the mechanical defibrillator. Lower cases (a–c) represent statistical comparisons using the Scott-Knott test, while different letter indicates (p ≤ 0.05) significant differences between the values

Full size image

Figure 13b shows that nanofibrils obtained with 5 passes through the mechanical defibrillator led to a significant increase (of about 40%; p ≤ 0.05) of the thickness of the nanocomposite (6.5%5x), with values ranging from 84.6 to 118.6 µm. The same increasing trend was observed when the same concentration of nanofibrils obtained after 30 passes was added to the blend. The trend of increase of about 6% in the thickness was not significant at 95% of confidence (p ≥ 0.05), and therefore there was no significant variation in the thickness compared to the control blend (80P20A). Such behavior may be explained by the low nano-fibrillation degree of the reinforcement after 5 passes through the defibrillator, since the inadequate dispersion of nanofibrils in the blend promotes the formation of agglomerates of nanofibrils with larger diameters, which contribute to the increase of thickness. Difficulties in controlling the mass/area ratio, problems in controlling the relative humidity of air drying, and variations in the viscosity of the filmogenic solution may also have influenced the thickness of the films (Scatolino et al. 2017b; Sobral 2000). Results indicate that additional passes (from 5 to 30) contribute to the increase of surface area and density of hydrogen bonds, as well as to the formation of tridimensional networks between the bamboo nanofibrils and the blend components, which renders the new structure more continuous and homogeneous (Eichhorn et al. 2010; Klemm et al. 2006).

Figure 13a corroborates the explanation above, since the reinforcement of the composites with nanofibrils with 5 passes (6.5%5x) promoted a decrease of approximately 24% of their apparent density in relation to the plain matrix (80P20A). The irregular and heterogeneous structure of this nanocomposite formed by aggregations of nanofibrils/microfibrils may have formed a network with many empty spaces occupied by air, favoring the decrease of density. Figure 13a also shows that the value of apparent density increased by approximately 11% when the nanofibrils added to the blend were produced with 30 passes through the mechanical defibrillator. The improvement of the dispersion of the nano-reinforcement in the blend matrix contributed to filling of empty spaces, thus increasing the density values.

According to Sterberg et al. (2013), nanocomposites with denser structures and lower porosity are important for applications that require high barrier properties to gases. Sanyang et al. (2016) found apparent densities of 1.27 g/cm3 and 1.46 g/cm3 for PLA films and sugar palm starch, respectively, while Doak (1986) found values of 0.92, 0.92–0.94 and 0.94–0.97 g/cm3 for LDPE, LLDPE and HDPE, respectively.

The grammage presented a distinct behavior for both nanocomposites with 6.5% of nanofibrils (Fig. 13). There was no significant increase (p ≥ 0.05) of the film grammage when including nanofibrils with 5 passes through the defibrillator. However, the apparent density reduced significantly due to the reduction of the sample thickness values caused by the low nano-fibrillation degree of those nanofibrils. The explanation for such behavior may be related to the heterogeneity in the production of nanofibrils with large diameters and the existence of residual parts of the starting fibers in the nanofibril suspension used in the 6.5%5x nanocomposite (Fig. 5a, b). This may have culminated in the formation of great amounts of empty spaces, thus promoting the lower density and consequently the reduction of tensile strength of the nanocomposites, as presented by Fig. 11a, b.

In the second case, as the nanofibrils produced after 30 passes were added, there was a significant increase in the film grammage (p ≤ 0.05) compared to the control blend. Contrarily to what happened in the first case, with nanofibrils obtained after 5 passes, there was an increase in density (about 11%) followed by a reduction of approximately 24% of the nanocomposite thickness. The higher nano-fibrillation degree culminated in more homogeneous nanofibrils with reduced diameters. Such high defibrillation led to high exposition of O-H groups, which were adhered by inter and intramolecular hydrogen bonds with the matrix polymers of the control blend, thus resulting in less empty spaces, followed by an increase of the apparent density. This better placement of nanofibrils allowed the reduction of nanocomposite thickness, which led to improved tensile properties compared to the control blend (Fig. 11a, b) (Sarantópoulos et al. 2002).

The nano-fibrillation degree, as well as the geometry and morphology of the cellulose nanofibrils influenced apparent density and grammage of the films, and altering their thickness. Similar results using the same methodology were found in literature (Sim and Youn 2016; Almeida et al. 2013; Szabo et al. 2013).

Figure 13a shows a significant increase (p ≤ 0.05) of approximately 32% in the value of water absorption for the nanocomposite with nanofibrils after 5 passes, compared to the control blend. This blend also presented a significant decrease (p ≤ 0.05) of water absorption of approximately 22% after the addition of the same amount of nanofibrils with 30 passes through the defibrillator. This means that the obtained nanofibrils influenced the kinetics of water absorption of the nanocomposite, which absorbed approximately 25% less humidity than the control matrix. This behavior is expected from most of the biopolymers whose applications prioritize mechanical and barrier properties, since they are strongly affected by moisture. The use of totally hydrolyzed PVA blended with modified starch (acetylation) contributed to the lower water absorption of the matrix. This occurs because the hydrolyzed PVA has the capacity to form strong inter and intramolecular hydrogen bonds by means of the O-H groups. Probably this behavior reduced the interaction between PVA and the water molecules, thus resulting in lower water absorption (Carvalho et al. 2009; Skeist 1990; Maria et al. 2008). The good compatibility between PVA and starch and the introduction of the ester group in the starch chains may have corroborated to the reduction of water absorption, since it may favor the formation of strong hydrogen bonds between the reinforcement and the matrix (Mao et al. 2000; Tang and Alavi 2012).

The 6.5%5x nanocomposite absorbed more water due to the large amounts of empty spaces produced during the formation phase. The high nanofibril diameters caused by the low nano-fibrillation degree using 5 passes of bamboo fibers may have precluded the formation of more homogeneous bonds between the reinforcement agents and the matrix, thus easing the water absorption. The opposite occurred with the 6.5%30x nanocomposite. The formation of strong hydrogen bonds and Van der Walls interactions between the O-H groups of nanofibrils and matrix components may have led to more compact and denser packaging. These results are in accordance with mechanical, morphological, thermal and FT-IR analysis previously presented.

After investigating the water absorption in nanocomposites using clay nanoparticles and starch in PVA/Polysaccharides blends, Shi et al. (2008) and Mahdavi et al. (2013) concluded that the use of these materials in hydrophilic matrices leads to a reduction of water absorption of the composites, increasing their potential of use. Results of water absorption of PVA/starch blends reinforced with cellulose nanofibrils are scarce in literature and require further investigation for better understanding.

Conclusions

The AFM investigation indicated that bamboo nanofibrils were efficiently isolated from commercial refined bamboo pulp followed by chemical pre-treatments and mechanical defibrillation combined with sonication. A reduction of approximately 88% of the average diameter was observed for nanofibrils obtained with 30 passes through the defibrillator, in relation to nanofibrils from 5 passes. SEM, DSC and FT-IR analysis revealed the compatibility of the control blend (80P20A), as well as confirmed the interaction among the polymers and cellulose nanofibrils with 30 passes. For 30 passes, DSC analysis revealed a considerable increase of the values of melting temperature (6%), melting enthalpy (23%) and crystallization enthalpy (55%) of the reinforced nanocomposites compared to the control blend, therefore, acting as a nucleating agent in the matrix. XRD confirmed the increase of the composite crystallinity index (CI) resultant of the improved nucleation of the polymers when using nanofibrils with 30 passes. 24 and 51% increase were observed for the values of tensile strength (TST) and elongation at break (EAB), respectively, when compared to the control blend. A significant reduction of up to 40% in the tensile modulus (TMO) was observed for the 6.5%30x nanocomposite in relation to the control blend. This nanocomposite is stronger and more flexible when compared to the 80P20A blend and to the nanocomposite containing 6.5% of less nanofibrilated solutions (5 passes). The insertion of 6.5% of nanofibrils also promoted the increase of the crystallinity index of the nanocomposite in around 10%, and led to an increase of approximately 11 and 25% on the grammage and density values, respectively, while the water absorption suffered a 22% reduction compared to the control blend. This work showed the potential of producing cellulose-based nanocomposites to meet the demands of the agricultural and plastic packaging sectors, making use of renewable materials for partial substitution of the petroleum-based ones.