Mechanical, Thermal Properties and Void Characteristics of Bamboo Fiber-Reinforced Epoxy Resin Composites Prepared by Vacuum-Assisted Resin Transfer Molding Process

ABSTRACT This study aimed to examine the effect of epoxy resin (EP) systems on the impregnation of bamboo fibers (BFs) by EP and the properties of BF/EP composites prepared by vacuum-assisted resin transfer molding (VARTM) process. BF preforms obtained by solution suspension method were used as reinforcement, and epoxy/anhydride system (EP-1), epoxy/amine system (EP-2) and two-component epoxy system (EP-3) were employed as matrix; on this basis, BF/EP-1, BF/EP-2 and BF/EP-3 composites were prepared by VARTM. The physical, mechanical and thermal properties and void characteristics of BF/EP composites were characterized. The results showed that EP-1 with a viscosity of 265 mPa·s at 23°C and an operating time of room temperature without curing was beneficial for the EP-1 to impregnate BFs adequately and uniformly. The incorporation of BFs into EP-1 significantly improved the flexural, shear and impact properties, thermal properties and interface properties of EP-1. Moreover, the properties of BF/EP-1 composites were better than those of BF/EP-2 and BF/EP-3 composites. Three-dimensional X-ray microscopy scans revealed that the volume and distribution of voids in the BF/EP composites differed significantly depending on the EP systems, and the percentage of void volume in the BF/EP-1 composites was only 0.14% in 3D reconstruction models.


Introduction
With the worsening global energy crisis and ecological environment, widely sourced, recyclable and degradable plant fibers of high performance are replacing nonrenewable and hard-todegrade artificial fibers (carbon fibers, glass fibers, aramid fibers, basalt fibers, etc.) as the reinforcement of fiber-reinforced polymer composites (FRPs) (Chokshi et al. 2020;Furtos et al. 2012). Jute fibers, hemp fibers, flax fibers, wood fibers and bamboo fibers (BFs) have been used as the common reinforcement of plant fiber-reinforced polymer composites (PFRPs) (Furtos et al. 2021(Furtos et al. , 2022. BFs, known as "natural glass fiber," have become a research hotspot of PFRPs owing to their low density, high strength and high rigidity (Hu et al. 2019). BF-reinforced polymer composites (BFRPs) are used in green packaging materials, automotive molded linings, machine product housings and home decoration (Muhammad et al. 2018). Presently, BFRPs are mainly prepared by compression molding, extrusion molding, filament winding molding and liquid composite molding. BFRPs with thermoset resins as matrix are usually prepared by liquid composite molding, including autoclave molding, resin transfer molding (RTM) and vacuumassisted resin infiltration molding (VARI) (Chang and Chang 2022). Vacuum-assisted resin transfer molding (VARTM) was developed from RTM, which used vacuum pressure to drive resins to impregnate BF preforms and then cured the whole material at room temperature or under heated conditions so as to prepare BFRPs with a certain BF content . BFRPs prepared by VARTM had fewer voids and greater properties than that by other molding techniques (Xia et al. 2015).
The current research related to BFRP preparation by VARTM mainly focused on the properties of BFs and the process parameters of VARTM. Studies on BFs were related to BF content (Yang et al. 2019), BF morphology (Rocky and Thompson 2018), BF surface modification (Chen et al. 2021;Kim et al. 2013;Wang et al. 2020) and BF preform preparation (Chiu and Young 2020). BF preforms were often prepared by hand lay-up, needle-punched process, textile process and twisting process. Among them, hand lay-up was convenient and efficient but might lead to unevenly dispersed BFs because of the unsatisfying skill and proficiency of operators. In contrast, the needle-punched process, textile process and twisting process could be used to prepare uniformly dispersed BF preforms but might easily cause secondary damage to BFs. Therefore, it is necessary to explore a suitable method of BF preform preparation for VARTM. The properties of BFRPs could be affected by numerous process parameters, such as vacuum level, injection method, injection time, injection temperature, the existence of diversion network, permeability and pressure. These parameters were interrelated and interplayed, jointly determining the properties of final BFRPs (Kedari, Farah, and Hsiao 2011;Li, Xie, and Ma 2015).
BFRP preparation by VARTM was highly dependent on vacuum-driven resin flow to impregnate BFs compared to other molding processes, making resin systems a critical part. Different viscosity, operating times, and curing processes of various resin systems had a significant impact on BF impregnation by resins and properties of BFRPs (Yang et al. 2011). BF impregnation by resins could be reflected by the void content in BFRPs and the properties of BFRPs. The voids were mainly flow-induced and gas-induced bubbles in BFRPs, as well as gaps at the BF-resin interface primarily determined by the physical and chemical properties of the two materials and the VARTM process (Sul, Youn, and Song 2012). However, at present, resin systems for BFRP preparation by VARTM tend to borrow those used for the preparation of other PFRPs or artificial FRPs by VARTM. For BFs with unique structures and properties, the effect of different resin systems on the properties of VARTM-prepared BFRPs is worth further examination.
In this study, BFs were prepared into BF preforms by solution suspension method to achieve nondestructive lay-up and uniformity. BFs were used as reinforcement. Epoxy/anhydride system (EP-1), epoxy/amine system (EP-2) and two-component epoxy system (EP-3) were used as the matrix. BF/ EP-1, BF/EP-2 and BF/EP-3 composites were prepared by VARTM. The physical, mechanical and thermal properties and void characteristics of BF/EP composites were characterized to investigate the effect of EP systems on BF impregnation by EP and the properties of BF/EP composites. The research and optimization of VARTM process parameters are of great significance to the practical application of BFRP products.

Preparation of bamboo fiber preforms and composites
BF preforms were prepared by the solution suspension method ( Figure 1a). First, BFs were selected to remove impurities and undispersed and lumpy BFs from raw material. Then BFs were evenly scattered in a self-made sieve with the same shape as the mold. The sieve was placed into the suspension device and vibrated for even suspension and dispersal of BF in water. Finally, BF preforms were taken out and dried. The moisture content of BF preforms was controlled to be below 2%.
The process flow of preparing BF/EP composites with different EP systems by VARTM is shown in Figure 1b. First, after the mold was cleaned, valves and hoses were installed on its left and right sides, and three layers of release agent were brushed inside it. Then, BF preforms, peel ply and diversion network were laid in turn. Finally, the sealing tape was applied to the surface of the female mold, and the vacuum pump was turned on to check the sealing of the mold after closing the mold. According to the ratio of each of EP systems, raw materials were weighed to prepare the matrix. EP injection was carried out by boundary injection and boundary output with a vacuum level of 0.080-0.085 MPa. When EP was sucked out of output ports, it continued to be injected for 20 min so that the air inside the mold was pumped out as much as possible. The valves on both sides of the mold were closed after EP injection, and BF/EP composites were cured according to the curing process in Table 1.

Physical property tests
The viscosity of the EP systems was tested according to (GB/T 22314-2008) by a rotary viscometer (NDJ-1, Shanghai Changji Geological Instruments Co., Ltd, China). The density of samples was measured based on the geometric method in GB/T 1463-2005. The water absorption of samples was measured according to GB/T 1462-2005: Samples were immersed in boiling water for 30 min and cooled in distilled water at room temperature for 15 min. Five samples of each type of composites were selected for each physical property test.

Mechanical property tests
The flexural, shear and impact properties of the samples were investigated according to (GB/T 1449, (ASTM D2344/D2344M-16, 2016) and (GB/T 1451-2005), respectively. Both flexural and shear tests were performed with a mechanical testing machine (5582, Instron, USA). The impact test was conducted with an impact experimental machine (XJJ-5, Chengde Kesheng Testing Machine Co., Ltd., China). Five samples of each type of composites were selected for each mechanical property test.

Microtopography analysis
The section images and fracture morphology of samples after the flexural test were observed with a scanning electron microscope (SEM, Quanta 200, FEI, USA).

Thermal property analysis
The thermal stability of samples was analyzed by a thermogravimetric analyzer (TGA, STA409PC, Netzsch, Germany). All measurements were conducted under a nitrogen atmosphere and heated up from 25°C to 800°C at a constant heating rate of 20°C/min. The dynamic mechanical properties of samples were measured in the dual-cantilever mode by a dynamic mechanical analyzer (DMA, Q800, TA instruments Inc., USA). Tests were conducted at temperatures ranging from 30°C to 120°C at a constant heating rate of 3°C/min with an amplitude of 15 µm and a frequency of 1 Hz.

Void characterization
The samples were scanned by a three-dimensional (3D) X-ray microscope (Micro-CT, nanoVoxel 3000, Tianjin Sanying Precision Instrument Co., Ltd., China) to observe the impregnation effect of different EP systems on BFs and the void characteristics of samples with a tube voltage of 80 kV, a tube current of 50 μA, a scanning time of 2 h, and a resolution of 6.60 μm. A threshold segmentation method was used to extract voids in samples to reconstruct 3D models.

Physical properties
The viscosities of different EP systems EP-1, EP-2 and EP-3 were 265, 275 and 450 MPa·s at 23°C, respectively (Figure 2a). During BF/EP composite preparation with consistent VARTM process parameters, the higher the viscosity of EP systems, the more challenging the EP injection, making no contributions to pumping out air from the mold. As can be seen in Figure 2b, the densities of BF/ EP-1 and BF/EP-2 composites were higher than those of EP-1 and EP-2, while the density of BF/EP-3 composites was lower than that of EP-3. During BF/EP composite preparation by VARTM, the density of BF/EP composites was related to BF and EP system density and was determined by the impregnation effect of EP on BFs. When the BF content of BF/EP composites was identical, EP-1 and EP-2 with a low viscosity were beneficial to EP flow to every corner in the mold during molding, significantly reducing voids in BF/EP composites, thus increasing the density of BF/EP-1 and BF/EP-2 composites. The section images of BF/EP composites are presented in Figure 2d. Compared with BF/EP-1 and BF/ EP-2 composites, BF/EP-3 composites had significant voids, which were mainly due to the difficulty of flowing EP-3 with high viscosity during molding. BF/EP composites could absorb water through three primary pathways: BF, EP and BF/EP interface (Jena, Pradhan, and Pandit 2014). As shown in Figure 2c, EP-2 presented higher water absorption than the other neat EP samples, mostly attributed to the component composition of EP systems. When water entered the interior of BF/EP composites through EP, it would fill BF/EP composites along BF/ EP interface by capillary action. BFs could absorb water through physical and chemical ways. Physical water absorption was mainly because of the certain water retention capacity of the BF lumen, while chemical water absorption was mainly due to a large number of hydroxyl groups on the surface of cellulose in BFs (Depuydt et al. 2019). These groups were extremely easy to contact water molecules and form hydrogen bonds, resulting in strong water absorption of BFs. The water absorption of different EP systems greatly increased by incorporating BFs. In this study. BF/EP-1 composites showed the lowest water absorption among the three BF/EP composites, which was because EP-1 with low viscosity and long operation time impregnated BFs adequately and effectively, BF/EP-1 composites had fewer voids, and EP-1 could wrap around BFs to form a barrier. The shorter operation time of EP- 2 and the higher viscosity of EP-3 tended to lead to inadequate and poorer impregnation of BFs with EP, resulting in higher and different water absorption of BF/EP-2 and BF/EP-3 composites.

Mechanical properties
As shown in Figure 3a, the incorporated BFs into three EP systems enhanced the flexural strength of EP-1 but slightly decreased that of EP-2 and EP-3. This was mainly because when the BF content of BF/EP composites was consistent, the BF impregnation by EP had a greater influence on the flexural strength of BF/EP composites. EP-1 with low viscosity and long operation time was beneficial to the void reduction of BF/EP composites and improvement of the interfacial bonding between BFs and EP. In contrast, the BF impregnation of EP-2 and EP-3 was inferior to that of EP-1. BF incorporation into EP showed a significant increase in the flexural modulus of different EP systems, with the largest enhancement in BF/EP-1 composites by 98% compared to that of EP-1. The fracture morphology of BFs and the interfacial bonding between BFs and EP after the flexural property test of BF/EP composites are shown in Figure 3d and e. As can be seen, basically no obvious gaps existed between BFs and EP-1, and the interfacial bonding between BFs and EP-1 was well. BFs with flat ends were not pulled out when damaged by an external load, and they could play an obvious strengthening role. BFs in BF/EP-2 and BF/EP-3 composites had obvious gaps with EP. BFs were pulled out when damaged by an external load, with the ends torn. The interfacial properties of BFs and EP were weak.
The shear strength of BF/EP composites with different EP systems is shown in Figure 3b, presenting a similar trend to that of flexural strength. Only the shear strength of BF-incorporated EP-1 was improved substantially, which was mainly because shear strength could be used as a mechanical index to measure the interfacial properties of BF/EP composites. With consistent BF content and VARTM process, BF/EP-1 composites presented the highest shear strength due to the low viscosity and long operation time of EP-1 and medium temperature curing of BF/EP-1 composites, and the favorable BF impregnation of EP-1 during molding to establish a physical and chemical bonding between EP-1 and BFs.
The BF incorporation enhanced the impact toughness of EP-1, EP-2 and EP-3 (Figure 3c). BFs were prepared through the solution suspension method to produce BF preforms. BFs were interspersed with each other, and the mechanical interlocking between BFs was enhanced, which could absorb more impact energy, thus effectively enhancing the impact toughness of the three neat EP samples. The impact toughness of BF/EP-1, BF/EP-2 and BF/EP-3 composites increased by 56%, 25% and 49%, respectively, compared to that of EP-1, EP-2 and EP-3. The impact toughness of BF/EP composites was determined by the impact properties of EP systems and interfacial bonding of BFs and EP. BFimpregnated EP-1 showed the optimum effect, and thus BF/EP-1 composites represented the largest improvement in energy absorption during the impact among all BF/EP composites.

Thermal stability
The thermal stability of three EP systems was higher than that of BFs (Figure 4a,b). The main pyrolysis stage of BFs was 220-410°C, in which numerous hemicellulose and cellulose in BFs were pyrolyzed   Yang et al. 2019). Among the three EP systems, EP-1 showed the optimum thermal stability and was mainly pyrolyzed at 300-500°C. EP-2 presented a large mass loss at 25-300°C and a main pyrolysis stage at 300-500°C. EP-3 pyrolyzed slightly at 25-250°C and significantly at 250-500°C. As shown in Figure 4c, when the thermal stability of BFs was lower than that of the three EP systems, the latter decreased after BF incorporation. BF/EP composites presented a main pyrolysis stage at 250-500°C, among which BF/EP-1 composites showed a relatively high thermal stability. BF/EP-2 composites showed a significant mass loss at 25-250°C, about 13.76% at 250°C. In particular, the maximum pyrolysis rates of three BF/EP composites were lower than that of the corresponding neat EP, indicating that numerous BF incorporation could effectively reduce the pyrolysis rates of BF/EP composites in the main pyrolysis stages (Figure 4b,d).

Dynamic mechanical properties
The maximum values of storage modulus for the neat EP samples followed the order of EP-1 > EP-3 > EP-2 ( Figure 5a). BF incorporation into EP could enhance the rigidity of EP. However, the maximum values of storage modulus for BF/EP composites showed a trend of BF/EP-1 > BF/EP-2 > BF/EP-3, which was mainly because the dynamic mechanical properties of BF/EP composites were associated with the properties of BFs, EP systems and the interfacial bonding between BFs and EP. Although the storage modulus of EP-3 was higher than that of EP-2, the storage modulus of BF/EP-3 composites was lower than that of BF/EP-2 composites due to EP-3 with a higher viscosity, more voids in BF/EP-3 composites and poor interfacial bonding between BFs and EP-3. In this paper, the temperature at the maximum loss modulus was used as the glass transition temperature (T g ) of BF/EP composites. BF incorporation into EP-1 decreased T g of EP-1 (Figure 5b), while BF incorporation into EP-2 and EP-3 increased T g of EP-2 and EP-3, which was mainly related to T g of EP systems. Previous TGA indicated that EP-2 and EP-3 were pyrolyzed at low temperatures.

Void characteristics
BF/EP composites prepared from different EP systems were scanned by Micro-CT, and the 3D models of BF/EP composites were reconstructed (Figure 6a). The internal slices of BF/EP composites are shown in Figure 6b, where the white-gray is BFs, the gray is EP, and the black is voids. Voids in BF/EP composites were mainly bubbles in BF/EP composites and gaps at the interface between BFs and EP. BF/EP-1 composites were basically free of obvious voids in the slices, while BF/EP-2 and BF/EP-3 composites had more voids, particularly the latter. In the 3D models (Figure 6c), the white was BFs and EP, and the blue was the voids in BF/EP composites. The percentages of void volume in BF/EP-1, BF/ EP-2 and BF/EP-3 composites were 0.14%, 5.18% and 11.82%, respectively. BF/EP-1 composites had relatively few and evenly distributed voids, mainly a small number of interfacial bonding gaps generated at the interface bonding of BFs and EP. More bubbles occurred in BF/EP-2 composites. Although EP-1 and EP-2 had similar viscosity, the shorter operation time of EP-2 was not conducive to the adequate BF impregnation by EP-2, and BFs inside BF preforms were more difficult to be impregnated by EP-2 for a short time due to the high BF content of BF/EP-2 composites. Bubbles and gaps were more distributed in the interior of BF/EP-2 composites than on the surface layer. They were most abundant in BF/EP-3 composites in both interior and surface layers. High viscosity and short operation time significantly increased voids in BF/EP composites.
As shown in Figure 7, BF/EP composites comprised BFs, EP and BF/EP interface. EP systems exerted a significant impact on the preparation of BF/EP composites by VARTM. The properties of EP systems laid the basis and had a direct effect on those of BF/EP composites. Moreover, When the VARTM process was identical, the viscosity of EP systems determined the impregnation effect of EP on BFs. The lower the viscosity of EP systems, the less difficult the EP impregnation into BFs. EP could fully impregnate BFs and penetrate into the internal BF bundles. Bubbles in the BF/EP composites became less, and the interfacial gaps between BFs and EP got smaller. The strengthening of physical and chemical bonding of the BF/EP interface was beneficial for EP reinforcement by BFs. EP systems with the same viscosity and shorter operation time would significantly lessen EP filling time, and EP failed to impregnate BFs adequately and effectively. Moreover, more bubbles would appear in BF/EP composites. The medium temperature curing of BF/EP composites often needs a shorter time than room temperature curing, conducive to promoting VARTM preparation of BF/EP composites. Before the epoxy system gelation in BF/EP composites cured at medium temperature, the temperature would promote the movement of epoxy molecular chains to reduce the viscosity of EP systems and enhance the impregnation effect of EP on BFs (Francucci et al. 2012). Moreover, in the preparation of BF/EP composites with high BF content, pressurization of BF preforms is of significance, and the combined effect of temperature and pressure was more favorable to the formation of hydrogen bonds and van der Waals forces between functional groups in the EP systems and free hydroxyl groups on the BF surface (Liu et al. 2012), which enhanced the BF/EP interfacial bonding and thus improved the properties of BF/EP composites. Taken together, VARTM preparation of BF/EP composites required EP systems with low viscosity, long operation time and medium temperature curing.

Conclusions
Vacuum-assisted resin transfer molding (VARTM) process could be a suitable method to maximize the advantages of bamboo fiber reinforced polymer composites and improve their suitability for industrial manufacturingThe preparation of high-performance BF/EP composites using VARTM required EP systems with low viscosity, long operation time and medium temperature curing. Epoxy/anhydride system (EP-1, 265 MPa·s at 23°C, no curing at room temperature and medium temperature curing) as the matrix for the BF/EP composite preparation by VARTM had relatively high effectiveness. The flexural strength, flexural modulus, shear strength and impact toughness of BF/EP composites prepared by EP-1 increased by 9%, 98%, 73% and 56%, respectively, compared with those of the neat EP-1 samples. The pyrolysis rate of BF/EP composites could be effectively reduced during the main pyrolysis stage and the BF/EP composites presented more excellent dynamic thermomechanical properties when prepared with EP-1. When the BF content was high, EP-1 was beneficial to EP impregnation to BFs, reducing the voids in BF/EP composites and improving the BF/EP interfacial bonding. The percentage of void volume was only 0.14% in 3D reconstruction models of BF/EP-1 composites.