Evaluation of highly filled epoxy composites modified with walnut shell waste filler

Epidian 624 epoxy resin, based on bisphenol A (BPA, epoxy number 0.48–0.51 mol/100 g, viscosity 600–800 mPa s, density 1.11 g/cm³), and isophorone diamine curing agent (IDA, amine number 250–350 mg KOH/g, viscosity 150–300 mPa s, density 1.02 g/cm³), both produced by CIECH Sarzyna S.A.(Poland), were used as polymer matrixes. Composition ingredients were proportioned by weight: 50 wt% of curing agent for 100 wt% of epoxy resin. Pot life of the composition was ca. 50 min. Walnut husk (Juglans regia Species, WS) obtained from AGRO Jarosław Seroczyński (Poland) was used as a lingo-cellulosic post-agricultural waste filler.

Preparation involved grinding of the filler was carried out on a laboratory mill sieve type MUKF-10 (Młynpol, Chwaszczyno, Poland). During the milling process, sieve mesh with a size of 0.2 mm was used. Next, the drying process was performed using Plus II Incubator (Gallenkamp, Loughborough, Great Britain). The drying time at a temperature of 40 ± 2 °C was 7 days. Additionally, a drying process at a temperature of 60 ± 2 °C and time 24 h was carried out before the production of the composites.

Mixing of the natural components with the resin was carried out using a high-speed mechanical stirrer with a water jacket (CILP-NRI, Warsaw, Poland). Observations conducted during the mixing showed that the solution prevented excessive temperature rise and the maximum temperatures during mixings not exceeded 43 °C. The natural filler of 20–50% by weight was introduced into the epoxy resin and stirring was performed using three rotational speeds: 7000, 10,000, and 17,000 rpm; which were applied for 3, 1.5, and 0.5 min, respectively. Subsequently, the mixture was degassed at a temperature of about 25 °C for 30 min, using a laboratory vacuum dryer Goldbrunn 1450 (GOLDBRUNN THERM, Zielona Góra, Poland). Then, IDA curing agent was added, the components were stirred again for 2 min at a speed of 7000 rpm and degassed in a vacuum dryer for 10 min. Normalized specimens were formed by casting into heated forms coated with a resolving agent. The cast samples were conditioned after curing at room temperature for 7 days.

In Fig. 1, the grain size distribution of ground walnut shell corresponds to the amount of material which remained on a sieve with a specified mesh size. Based on the analysis, it was found that the prepared filler was characterized by a relatively homogeneous size distribution. The analysis of the results allowed to determine that the size of the majority of the constituted particles (95%) was in the range of 32–125 μm.

SEM photographs presented in Fig. 2 showed grains of the natural filler applied in this study after the grinding and drying processes. The particles of ground walnut shell differed in size and geometry (Fig. 2a). A considerable number of grains, despite different roughness of the surface (rough or smooth, Fig. 2c), were characterized by similar spherical shape (Fig. 2b). This filler type may be characterized as particle shaped, moreover, most of WS particles were characterized with developed surface (Fig. 2d).

FT-IR spectra of ground walnut shell, the unmodified epoxy resin sample, and the composite containing 50 wt% of the filler, were presented in Fig. 3. At WS spectra, the broad absorption band at 3500 cm⁻¹ indicates bonded hydroxyl –OH groups existing in the organic filler. Absorption peaks at wavenumbers of around 2900, 1744, and 1053 cm⁻¹ are assigned to –CH, C=O and C–O stretchings, respectively [23]. Moreover, the observed peak at 1623 cm⁻¹ corresponds to C=C stretching and confirms the presence of lignin and its aromatic group in the natural filler [24, 25]. At epoxy resin and composite sample spectra, vibration band at 915 cm⁻¹, resulting from C–O deformation of oxirane group, was recorded. The decrease of this band reflects the development of epoxy groups (a reaction of epoxide rings to form ether linkages). Despite high amount of the filler in the composition, intensities of the unmodified epoxy resin absorbance and of the composite with the highest WS amount were comparable. The increasing peak denoted near 1200 cm⁻¹, corresponding to C–O–C stretching of aliphatic linear ester, was connected with homopolymerization of epoxy groups. The sample containing 50 wt% of WS showed a strong decrease in absorbance at this intensity, therefore, it can be stated that the curing process was modified by the presence of the organic filler. During epoxy cast curing, epoxide ring opening process is usually accompanied by the occurrence of an increasing peak of the –OH stretching, at 3500 cm⁻¹. In case of the 50% WS sample, a higher intensity of the absorbance peak (at 3500 cm⁻¹) may be observed than for unmodified epoxy resin [26, 27]. This fact may be attributed both to the presence of the organic filler in the composite material, as well as the effect catalytic reaction of the curing process caused by the presence of additional –OH groups originating from the filler [28, 29]. Detailed description of all observed at FT-IR spectra peaks was presented in Table 1.

Moisture testing provides information on the amount of water contained in the material. The presence of water in the composites suggests that, despite the drying process performed, certain amount of water was introduced to the NFC. Unfortunately, to realize complete elimination of the water from the natural fillers, it is necessary to use high temperature drying processes (400 °C), which increases production costs and may cause thermal and hydrolytic decomposition of the hemicellulose material [10]. Preliminary drying of the filler and elevated temperature during curing of the epoxy casts should provide to complete removal of moisture from organic particles, therefore, the absorption of moisture from the air during conditioning of the materials cannot be excluded. This is confirmed by the fact that, under real conditions of use, the moisture content of such composites depends on humidity and is subject to seasonal fluctuations (higher values in the summer months) [12]. In Fig. 4 results of moisture content of pure epoxy and composite materials, determined as the loss of its mass between the time of sampling and state after drying to constant mass at 103 °C, were presented. Similar values were measured for all composite samples, whereas the highest moisture content was observed for EP. The high values of investigated parameter denoted for unmodified epoxy resin may be attributed to the release of low molecular weight degradation products of the thermoset polymeric matrix. The thermal stability of epoxy resin and epoxy composites was comprehensively described in the part of the article concerning thermogravimetric analysis. Surprisingly, evaluated parameter was almost at the same level for all composite materials. The stabilization effect of the parameter, measured in accordance to ISO 16979 standard, observed for all composite material is a result of simultaneously decreasing amount of low molecular weight products of epoxy resin degradation, as well as increasing amount of hydrophilic natural waste filler.

Tensile strength and Young’s modulus were measured by means of static tensile experiment to assess the influence of an organic filler addition on epoxy composites’ properties. Figure 5a, b showed variations of those two mechanical properties describing samples’ behavior as a function of filler amount. The organic filler incorporation into polymeric matrix caused a decrease of tensile strength in comparison to the unmodified polymer. Simultaneously, the values of Young’s modulus measured for the samples made of composite materials were significantly higher in comparison to unmodified epoxy resin. Gradual improvement of the sample stiffness was observed for samples containing 20 and 30 wt%. For composite samples modified with 30 wt% of the filler, the Young’s modulus value, mainly 2.0 GPa, was almost twice higher in comparison to the reference-unmodified epoxy sample, which reached 1.2 GPa (p < 0.00001). However, in case of higher amount of the filler, i.e., above 30 wt%, the elasticity modulus-measured values were lower in comparison to the sample containing 20 wt% of the filler. Description of NFC for which a fall in the Young’s modulus was recorded after exceeding the optimum filler content 50 wt% can be found in the literature [35]. It can be stated that the improved elasticity modulus is a result of epoxy resin curing process modification. It is well known that any incorporation of a filler, with a surface rich in hydroxyl –OH groups, may cause an additional catalytic effect resulting in the creation of additional segments of curing network [28]. Therefore, the improved stiffness of the composites containing lower amount of ground walnut shell is probably caused simultaneously by stiff filler structures’ presence, as well as increased cross-linking density of epoxy matrix. The reason for lowering Young’s modulus values for composites containing 40 and 50 wt% of fine ground walnut shell, as compared to the samples which reveal the highest elasticity modulus values, i.e., 30 wt% WS, could be a random arrangement of filler particles and a hindered degassing of highly viscous compositions. Moreover, it should be mentioned that volumetric share of organic waste filler in polymer matrix was significantly higher then amount calculated by weight. It may be stated that for composites containing more than 30 wt% of the filler some particles were in a mutual contact. This phenomenon provides to local variations of stress transfer, which caused lowered values of whole composite material elasticity modulus. Lower values of mechanical properties obtained for the composites with the highest share of the filler suggest that the incorporation of high amount of fine ground walnut shell may be reasonable only in case of fabrication of the products that will not be highly loaded during their exploitation. It cannot be excluded that natural rigidity and ability to transfer mechanical loads, resulting from the walnut shell structure, were lost because of the significant level of its fragmentation. The rigidity of polymer composites depends on the shape of the filler. If value of aspect ratio of the filler is close to one, as in the case of a applied ground walnut shell, the amount and elasticity of incorporated filler as well as elasticity of polymeric matrix gain in significance [35, 36].

In Fig. 5c the results of the impact strength of epoxy resin and composite materials were depicted. Natural filler incorporation caused a decrease in the impact strength of the composites in comparison to the unmodified polymeric matrix. The lowest impact strength, equal to 1.66 kJ/m², was observed for the sample containing 20 wt% of the WS, while for unmodified epoxy resin, the value was 4.6 kJ/m² (p < 0.001). This phenomenon is typical for polymeric composites modified by a particle-shaped filler and its insufficient dispersion in polymer matrix [37]. Moreover, it cannot be excluded that residual moisture present in the filler and incorporated to composite material provided micropores that acted as notch during abrupt load in the course of Charpy experiment. It should be mentioned that the measured values of impact strength were not corresponding to higher amount of the filler in composite material.

Hardness measurement results for epoxy resin and composite materials were presented in Fig. 5d. It can be observed that the incorporation of ground walnut shell led to an improvement in the composites’ hardness in comparison to the unmodified thermoset polymer. The highest value of HB hardness was denoted for the composite containing 50 wt% of the filler, mainly 86.6 N/mm², while for unmodified epoxy resin, the value was 76.6 N/mm² (p < 0.01). However, the lack of a distinct relationship between the filler amount and the composites’ hardness may be attributed both to the low hardness of walnut shell itself (2.5–3.5 in Mosh scale), as well as insufficient dispersion in epoxy matrix. According to our previously published results, the hardness of the natural fiber composites with ground sunflower husks and pistachio shell depends on the size of the particles and decreased with increasing fillers content [38]. Moreover, the lack of correlation between the amount of incorporated into polymeric matrix natural filler and mechanical properties of the composites was observed also in the case of polyethylene filled with hazelnut shell [39].

In Fig. 6, storage modulus (G′) and loss factor (tanδ) curves were presented as a function of temperature (T). It can be clearly observed that the natural waste filler incorporation caused an increase in G′ in the considered temperature range. The denoted improvement in G′ value after adding 50% of WS was equal to 100 and 403% measured at 30 and 80 °C, respectively, in comparison to unmodified epoxy resin. This means that composite materials reveal a higher ability to maintain mechanical loading with recoverable viscoelastic deformation than unmodified epoxy resin, also in elevated temperature [26, 40]. Changes observed for resin modified with walnut shell particles on the G′ vs. T curves are in good agreement with the static tensile test results and may be assigned to the improved stiffness of composite materials. However, it should be mentioned that for samples containing more than 20 wt% of the organic filler, the differences between measured storage modulus values were negligible. In Fig. 6b, loss factor changes were depicted. The unmodified epoxy sample revealed the highest value of tanδ. With the increasing amount of natural filler particles, the maximum measured loss factor was slightly reduced. This phenomenon is comprehensible because of the incorporation of filler stiff domains which acted as a steric hindrance reducing material attenuation. The glass transition temperature (T _g) measured as a peak of tanδ vs. T curve was lowered with a higher content of the filler. It is probable that residual fat in ground walnut shell plasticized composite materials [41]. However, this effect was not accompanied by G′ decrease, therefore, it may be stated that the used filler did not modify the curing process in the entire polymer material but only led to a modification of the cross-link density near the organic particles’ surface.

Thermal stability of highly filled epoxy composites was evaluated by thermogravimetric analysis to complement the characteristics and to define a potential temperature usability range of final products. The results of TG measurements realized in inert atmosphere are presented in Fig. 7. The additional data on the degradation process, including: 5, 10, and 50% mass loss temperatures, residual mass, and information on maximum intensity of degradation read from DTG curves for all measured samples, are summarized in Table 2. Based on the change in mass, ranging from ambient temperature to 130 °C, it was found that despite the drying process the content of water in the shell reached 1.58%. It is possible that in filler-remained water, which was constantly linked with hydrophilic fiber components by chemical bonds. Natural filler decomposition occurred in the course of a two-step process, while the unmodified epoxy resin took place in a single-step way. The first mass loss of the TG curve observed for walnut shell sample was up to 130 °C and corresponded to the evaporation of the moisture which was present in the organic material. The next mass loss observed at the natural filler TG and DTG curves began at about 190 °C, rose to the maximum decomposition rate at 345 °C, and finished at about 390 °C. The courses of the WS thermogravimetric curves are similar to those presented in the literature and correspond to decomposition of the three main components of walnut shell: hemicelulose (210–325 °C), α-cellulose (310–400 °C), and lignin (160–500 °C) [42‐45]. Passive pyrolysis of the natural filler occurs above 470 °C. All composite materials containing walnut shell undergo a two-step decomposition with an increasing intensity during the first step, adequate to the increasing content of ground walnut shell. Because of the possibility of ligno-cellulose fiber decomposition, due to the temperature increase during the manufacture of the composites, the degradation temperature of the filler (T _dmax) was determined. T _dmax, corresponding to a weight loss of 1% at temperatures above 150 °C, in presented study reached 215 °C and was at a similar level to the literature data [18]. A comparison of the values shown in Table 2 allows to ascertain that the natural filler has a significantly higher thermal stability in the range of 10% mass loss than unmodified epoxy resin. Therefore, it is comprehensible that composite materials containing walnut shell particles have higher thermal stability in a temperature range up to 355 °C. The sample containing 30 wt% of ground walnut shell revealed the highest thermal stability based on 5 and 10% mass loss. The difference between composite materials with various amounts of WS results from higher natural filler amount in comparison to the 20% WS sample, and better dispersion in comparison to samples filled with 40 and 50 wt% of ground walnut shell. It should be mentioned that thermal degradation of all composite materials was significantly suppressed. The lowest improvement in thermal stability, measured as temperature of 5% mass loss, was 15 °C, while the highest was 27 °C for samples containing 20 and 30 wt% of the filler, respectively. At a measuring point recorded at 50% mass loss, a different tendency may be observed. The highest temperature was denoted for epoxy resin, whereas the increasing content of WS caused a gradual decrease of 50% mass loss temperature. Measured at 900 °C, the sample residual mass grew with the increasing content of the organic filler. DTG data indicate that composite decomposition rate was lower than for unmodified epoxy resin, however, the maximum intensity temperature of composite thermal degradation decreased as a function of an increasing organic filler content.

The analysis of epoxy composites’ structures was carried out on the basis of the scanning electron microscope photographs. Images of sample fractures taken at 400× magnification were summarized in Fig. 8. The microstructure of composites containing various amounts of the filler allowed to observe a good saturation of organic particles’ surface by polymeric matrix. The application of thermoset material characterized by a relatively low molecular weight favorably influenced the wettability of the filler particles’ surface and increased interfacial adhesion between polymer and reinforcement. This phenomenon may be confirmed by the observation of fractured fragments of walnut shell particles instead of entire ones in polymeric matrix. The fractures are caused by a pull-out effect occurring when there are limited interactions in the interface. It should also be noted that the amount of voids resulting from the contact between the resin and the filler was small. The voids were distributed mainly in the polymeric matrix which may be an effect of a hindered degassing of the highly viscous compositions. SEM photographs allowed to observe the filler particles with size of several dozens to more than 100 μm in the case of the material containing the highest amount of the WS. It cannot be unequivocally stated that large grain sizes predominate in the case of high-filled composites, since a fragment of the shell of approximately 100 μm can be found in the photographs of the WS 30%. The occurrence of varied sizes and shapes of grains is adequate to the SEM photographs of the filler before it is introduced into the polymer. It can be presumed that the method of mixing the components did not cause further shredding of the shells. SEM analysis allowed to ascertain that the previously described negative effect of the natural filler incorporation on the mechanical properties of the composites may be attributed to a random arrangement of filler particles.