**1. Introduction**

Over the past two decades, much research attention focused on developing synthetic nanofillers such as carbon nanotubes (CNT), nanoclays, TiO2, SiO2, CaCO3, and Al2O3, etc., for polymeric matrices. These nanofillers were reinforced in a range of polymeric matrices such as thermosets, thermoplastics, elastomers and their blends to develop polymer nanocomposites with nanofillers as reinforcement material. Dramatic improvements in thermal, mechanical, chemical, and physical properties were observed in nanoparticle filled in polymeric matrix composites [1–5]. However, in recent years because of the environmental pollution, rapid depletion in natural resources, recycling, and greenhouse carbon gas emission effects, researchers started looking for alternative green nanofillers for polymeric materials [6–8].

Some of the green nanofillers such as eggshell-based nanoparticles [9,10], plant-based materials [11–14] such as nanocellulose, lignin and hemicellulose, polylactic acid (PLA) [15,16], chitin/chitosan [17–19], etc., were the focus in recent past. Considerable breakthrough had been obtained with comparable thermo-mechanical properties using these green nanofillers to that of synthetic nanofillers.

In particular, plant-based cellulosic particles have gained a significant interest in the research communities because of their abundant availability and cost effective processing methods. Cellulose is abundantly available worldwide from plant source. Their availability is higher than that of commonly produced synthetic materials such as polyethylene, polypropylene, PET, glass fibers, and carbon fibers. One observation suggests that cellulose fibril has a comparable tensile property to that of plain carbon steel material. Cellulose is an important component in the plant source which induces structural properties such as strength, sti ffness, thermal, and physical stability. In general, cellulose content varies from 20 wt.% to 90 wt.% in plant source depending on the type of the plant [20–25].

Plants have three major chemical constituents, namely, cellulose, hemicellulose, lignin and other chemicals at minor level. Successful attempts have been made in the recent past regarding extraction of cellulosic material from plant source by chemical and physical methods [26–28]. Cellulose is available in the form of microfibril, nanocrystal, or nanofibers depending upon the obtainable size and processing methods [29,30].

These cellulosic particles were filled in polymer matrix to develop micro or nanocomposites, depending upon the size of the cellulosic particle fillers. In general, most of the literature focused on the addition of cellulosic particles in thermoplastic polymeric matrices, such as PP (polypropylene), LDPE (low density polyethylene), PLA (Polylactic acid), etc., [31–34]. The literature on cellulosic particles filled in thermoset polymers are relatively scarce or negligibly available [35]. It is important to understand the e ffect of these green nanoparticles on thermoset polymers such as epoxy, polyester, and urethanes. Thermoset epoxy and epoxy composites are widely used from bicycle industry to aerospace industry, because of their high thermal, mechanical, and barrier properties than that of polymeric materials [36,37]. A possible replacement of synthetic fillers with cellulosic nanofiber (CNF) fillers in thermosets will have huge impact on economic and social benefits.

Therefore, understanding the e ffect of CNF on thermal, mechanical, and barrier properties of the epoxy polymer matrix is vital as the study has significant industrial application. This paper focuses on the e ffect of CNF in epoxy polymer matrix on curing characteristics, structure, thermal, and mechanical properties. The sub-micron sized CNF were extracted from the stem section of banana fibers using acid hydrolysis method and the fibers were reinforced in the epoxy polymer to form epoxy-CNF nanocomposites.

#### **2. Results and Discussion**

Figure 1a shows the photograph image of raw banana fiber and CNF extracted from banana fiber. The raw banana fibers appear as a brown colored long fiber extracted from stem section of the banana plant. The CNF appears as a bright white colored particle, with above 95% cellulose content as obtained from our earlier study, with the CNF yield of ~28% from initial banana fiber mass content [38]. Figure 1b shows SEM image of cellulose nanofibers (CNF). The diameter of CNF varies from 50 nm to 100 nm and length 100 nm to 1000 nm. Figure 2 shows the TEM image of 2 wt.% CNF-filled epoxy nanocomposite (Figure 2a) and 5 wt.% CNF-filled epoxy nanocomposite (Figure 2b) observed under bright field mode. In this mode the CNF phase is represented by dark discontinuous needle like phase and the matrix phase is represented by bright continuous phase. Dispersion of CNF particles on the matrix was uniform at lower (2 wt.%) and higher concentration (5 wt.%). However, CNF particles tend to form agglomeration at higher concentration (5 wt.%).

**Figure 1.** Fiber and cellulose nanofiber (CNF) images shows (**a**) photography image of raw banana fiber and extracted CNF; (**b**) SEM image of CNF fibrils.

**Figure 2.** TEM image of epoxy with (**a**) 2 wt.% CNF; (**b**) 5 wt.% CNF-filled nanocomposites.

Figure 3 shows the measurement of heat release during gelation of epoxy and epoxy-CNF series during gelation. Unfilled neat epoxy resin shows the maximum heat release of 76 ◦C at around 96 min of gelation, whereas, epoxy with CNF up to 1 wt.% shows maximum heat release at relatively lower time around 90 min. The reduction in gelation time sugges<sup>t</sup> the catalytic e ffect of CNF particles during epoxy curing. CNF with above 1 wt.% shows the gelation peak similar to that of neat epoxy resin, however, with reduced peak width as function of CNF concentration. This suggests the higher concentration may not act as a catalytic agent, however, it may aid faster curing. This curing characteristics of the of nanocomposite series was further examined using the FTIR study. Figure 4 shows the FTIR spectrum of CNF, unfilled neat cured epoxy polymer, 0.5 wt.% and 3 wt.% CNF-filled epoxy nanocomposites. CNF shows a broad band around 3400 cm<sup>−</sup><sup>1</sup> suggesting presence of –OH group. Presence of a shoulder at around 1660 cm<sup>−</sup><sup>1</sup> suggests oxidation of carbohydrate during CNF extraction. A peak around 1102 cm<sup>−</sup><sup>1</sup> is due to C–H stretching of the cellulosic group. A peak at 1102 cm<sup>−</sup><sup>1</sup> sugges<sup>t</sup> changes induced in hydrogen bonds indicating transition from cellulose I to cellulose II structure during chemical treatment. A band at around 1034 cm<sup>−</sup><sup>1</sup> suggests a fraction of xyloglucans associated with non-hydrolyzed hemicellulose which is strongly bound within cellulosic fibrils. A sharp peak at around 888 cm<sup>−</sup><sup>1</sup> suggests a typical cellulosic structure [39,40]. Unfilled neat epoxy polymer shows a characteristics peak of a typical epoxy group (3500 cm<sup>−</sup><sup>1</sup> OH stretching vibration; shoulder at 2950 cm<sup>−</sup>1–2775 cm<sup>−</sup><sup>1</sup> of CH stretching of CH2 and CH aromatic or aliphatic vibration; 1608 cm<sup>−</sup><sup>1</sup> of aromatic C=C stretching; 1504 cm<sup>−</sup><sup>1</sup> of C-C stretching; 1031 cm<sup>−</sup><sup>1</sup> C–O–C of ether stretching; 830 cm<sup>−</sup><sup>1</sup> of oxirane group) [41,42]. The nanocomposites show almost all characteristics peaks of epoxy and CNF and sugges<sup>t</sup> the presence of both these phases. However, in nanocomposites the intensity of oxirane peak (830 cm<sup>−</sup>1) was reduced with the disappearance of carbohydrate oxidation peak of CNF (1660 cm<sup>−</sup>1) in 0.5 wt.% CNF and 3 wt.% CNF-filled epoxy nanocomposites. This suggests that CNF might induce a catalytic curing of oxirane rings of epoxy polymer. The reduced oxirane peak shows the higher level of cross-link density of the epoxy polymer in nanocomposites due to CNF particles, possibly the oxidized cellulose in presence of amine curing agen<sup>t</sup> induced amide bonding. This "amidated" cellulose molecules might have incorporated covalent bonding with epoxide structure, showing the presence of amide group's band at around 1660 cm<sup>−</sup><sup>1</sup> in nanocomposites. This e ffect suggests catalytic e ffect of CNF in epoxy polymer and may also a ffect the properties of composite.

**Figure 3.** Time–temperature gelation of unfilled and CNF-filled epoxy nanocomposite series.

**Figure 4.** FTIR spectrum of CNF, epoxy and epoxy filled with 0.5 wt.% CNF and 3 wt.% CNF-filled nanocomposites.

Tensile properties of nanocomposites were examined and the tensile stress–strain curves are shown in Figure 5 with their values in Table 1. In general, tensile properties are positively affected because of the addition of CNF particles. Maximum improvement of 86% increased modulus at 5 wt.% CNF, 11% increased strength at 5 wt.% CNF, and an 26% increased elongation at 3 wt.% CNF was obtained than that of unfilled neat epoxy polymer. Although the general trend shows increased tensile properties of nanocomposites, Table 1 shows the tensile values are fluctuating as the function of CNF content. This could be due to the few factors, namely, inconsistent particle size, cellulose concentration within and among the particles, and other non-cellulosic phases presenting in the particles. Since the material was extracted from natural source such level inconsistence properties are expected. Similar results were also obtained elsewhere [8,31,33].

TGA of nanocomposites were examined and their plot is shown in Figure 6 with their values in Table 2. The thermal decomposition of CNF was lower than that of neat banana fiber. This could be due to the larger heat energy exposed by CNF due to its high surface area when compared with banana fiber [43]. Whereas, the thermal stability of the nanocomposite is higher than that of neat epoxy polymer, possible that the nanolevel dispersion of CNF particles acted as a thermal barrier to the epoxy polymer matrix and resulted in improved thermal stability. Similar improved thermal properties were also obtained in CNF-filled thermoplastic polymer nanocomposites [32,44]. Relatively lesser improvement at thermal property at higher content CNF (>3 wt.%) could be due to the higher amount of surface area exposure of particles. Much of the improvement in thermal stability is obtained at 2 to 3 wt.% CNF-filled epoxy nanocomposites.

**Figure 5.** Tensile stress–strain curves of unfilled and CNF-filled epoxy nanocomposite series.



**Figure 6.** TGA curves of unfilled and CNF-filled epoxy nanocomposite series.


**Table 2.** TGA properties of unfilled epoxy and CNF-filled epoxy nanocomposites.

Figure 7 shows the DMA properties of CNF-filled epoxy nanocomposite series with their properties in Table 3. The CNF addition increases the storage modulus of epoxy at room and elevated temperatures (Figure 7a). This improvement could be due to the hard and sti ff properties of the CNF particles when compared with epoxy polymer matrix. Figure 7b shows the Tanδ versus temperature values of nanocomposite series. The temperature at which maximum Tanδ value obtained is called glass transition temperature ( *T*g). *T*g is a temperature at which molecular relaxation occurs in the amorphous phase of the polymer and changes toward rubbery phase. Marginal increase in *T*g value is obtained in nanocomposites due to the addition of CNF particles. The CNF particles possibly resisted the molecular relaxation movement during heating and increased the *T*g temperature. The Tan δ peak value was reduced in composites (>1 wt.% CNF) than that of neat epoxy polymer. Similar results on the reduced Tan δ peak values were observed elsewhere, suggesting e ffective reinforcement of particles in the polymer matrix [35,45,46]. Tan δ peak value relates to damping and impact characteristics of the material. The reduction in Tan δ peak value in composites (>1 wt.% CNF) shows improved damping/impact characteristics due to increased packing e fficiency of the CNF particles in the matrix [45,46]. DMA results show optimized improvement was observed at 2–3 wt.% nanocomposite.

**Figure 7.** Dynamic mechanical analysis (DMA) of (**a**) storage modulus; (**b**) Tan δ of unfilled and CNF-filled nanocomposite series.


Water uptake properties of nanocomposites is shown in Figure 8 with their values in Table 4. When compared with neat epoxy polymer, water uptake is reduced in CNF-filled epoxy nanocomposites. The reduction in water uptake is proportional to CNF content in epoxy polymer matrix. Maximum reduction of ~47% decreased water uptake was observed in 5 wt.% CNF-filled epoxy nanocomposite than that of neat epoxy polymer. This reduction may be due to the hydrophobic characteristics of CNF particles than that of banana fiber [26–30] and possibly this e ffect would have resulted in improved water barrier properties of nanocomposites. Moreover, large surface of CNF particles might have exposed a larger area of water medium and served as a barrier medium and protected the matrix polymer against water uptake.

**Figure 8.** Water uptake of unfilled and CNF-filled epoxy nanocomposite series.

**Table 4.** Water uptake properties of unfilled epoxy and CNF-filled epoxy nanocomposites.


#### **3. Materials and Methods**

## *3.1. Raw Materials*

The matrix material is a cured thermoset epoxy polymer made up of resin and hardener.

Epoxy resin (Diglycidyl ether of bisphenol-A) and hardener (unmodified cyclic aliphatic amine) supplied under trade name LR-20 and LH-281 respectively were obtained from AMT Composites (Mumbai, South Africa). Banana fibers were obtained from Richmond traders, Mumbai, India. The fibers were mechanically extracted from the stem part of the banana plant. All other chemicals used for the extraction of cellulose nanofibers were obtained from Merck Chemicals, Durban, South Africa.

#### *3.2. Extraction of Cellulose Nanofiber (CNF)*

CNF were extracted from banana fibers using acid hydrolysis process. Initially the banana fibers were cut into lengths of 3 cm. CNF extraction process involved three steps, namely, alkaline treatment, bleaching, and acid hydrolysis.

In alkaline treatment, the chopped banana fibers were soaked in 5 wt.% NaOH solution at the weight ratio of 1:10 of fiber to solution content. The fibers were soaked for 24 h after which the fibers were extracted and dried in an oven at 60 ◦C for 4 h.

In the bleaching process, alkaline-treated fibers were soaked and rinsed in concentrated sodium hypochlorite solution for 1 h. 3 wt.% fraction of fibers were soaked in a solution and bleached in running tap water. The fibers were removed and then placed in an oven at 60 ◦C for 4 h.

In the acid hydrolysis process, 10% of diluted sulfuric acid was prepared using distilled water in a beaker. Thereafter, bleached fibers were then soaked into the acid solution at 7 wt.% fiber weight fraction. The fiber-solution was stirred using a mechanical stirrer for 30 min. Thereafter, the excess liquid solution was extracted by a vacuum process. The acid-hydrolyzed fibers were then washed with distilled water and dried in an oven at 60 ◦C for 4 h.

In the above processes, cellulosic phase in the banana fiber was extracted and all other non-cellulosic phases such as lignin, hemicellulose, and other non-cellulosic phases were eliminated.

#### *3.3. Processing of CNF-Filled Epoxy Nanocomposites*

Synthesis of nanocomposites were carried out in two steps. First step involved mixing of resin with desired concentration of CNF and the second step involved casting of resin-hardener-CNF mixture in a glass mold. An electric shear mixer (Heidolph MR Hei: Standard, Labotec, South Africa) was used to mix resin and CNF particles. Initially, 100 g epoxy resin was heated in a glass beaker at 80 ◦C and CNF were added followed by shear mixing at 500 rpm for 0.5 h at 80 ◦C. The resin and CNF particle mixture were then cooled to room temperature (RT) and 30% weight fraction of hardener to that of epoxy resin (as per supplier's manual) was added into the mixture for curing purpose. The resin, hardener, and CNF together were gently stirred using a glass rod for ~3 min and then cast into glass molds. The casting process involved pouring of resin-hardener-CNF mixture in between two glass mold plates (30 cm × 30 cm × 3 cm) separated by a rubber gasket running along the three sides of the plates. The pouring of the resin mixture was assisted by a runner attached to the top side of the glass plate molds. To facilitate easier removal of the cast product, wax was used as a mold release agen<sup>t</sup> and applied at the inner face of the glass plates and rubber gasket before the pouring process. The nanocomposite cast sample (with dimension of ~27 cm × 27 cm × 3 mm) was obtained 24 h after it was poured inside the mold cavity. The testing and characterization was carried out after seven days of curing (i.e., fully cured state of the epoxy polymer under experimental condition, though 100% curing cannot be achieved). The cured sheets were further cut and sized as per the standard dimension required for conducting testing.
