Designing cellulose nanofiber surface for high density polyethylene reinforcement

The chemical structures and their abbreviations of these six aliphatic CNF-esters are shown in Fig. 1, i.e. three linear acyclic: acetyl (C2), myristoyl(C14) and stearoyl(C18), two cyclic: cyclohexanecarboxyl(C7) and t-butylcyclohexanecarboxyl(C11) without and with branched structures, respectively, one branched esters: pivaloyl (C5).

The esterification was carried out under the general reaction conditions shown in Table 1 to afford the products with DS of around 0.4, although the introduction of pivaloyl group took longer reaction time and higher temperature because its steric hindrance. Here, we determined to be the expected DS around 0.4, because the esterification giving the DS less than 0.55 did not affected to the crystallinity of the original CNF as described in our previous paper (Sato et al. 2016). That is, the esterification is limited in only on the surface of the CNFs.

The melt compounding of the esterified CNFs and HDPE was performed at 140 °C using a twin screw extruder followed by injection molding at 160 °C. In these processes, the CNFs are subjected to high shear stress. Thus, thermal deterioration of the CNFs during the melt compounding and injection molding process is a great concern because damage to the CNFs would affect their reinforcing ability. Especially, elimination of the incorporated ester groups from the esterified CNFs under such high temperature may be serious (Agustin et al. 2016a). Even if it results in 1% weight loss, the evolved gas could affect the micro and nanostructures of the molded samples, resulting in poor mechanical properties. Hence, the temperatures resulting in 1 and 5% weight loss of the esterified CNFs were determined as summarized in Table 3 along with the peak temperatures of the differential curves of TGA.

As shown in the table, the 1% weight loss temperatures for all six esterified CNFs are higher than that of the unmodified CNF, and the temperatures are in the range 270–280 °C. In particular, the CNF(Cyc) and CNF(Piv) exhibit high 1% weight loss temperature of 282 and 285 °C, respectively. These temperatures are 27 and 24 °C higher than that of the unmodified CNF.

As reported by Agustin et al. (2016b) from the pyrolysis products, it can be deduced that removal of esters groups on the surface of CNFs can take place in two ways: the direct cleavage of the ester bonds from the cellulose chain to release the equivalent carboxylic acids; and the possible condensation reaction between adjacent two esters on the surface of CNFs to produce symmetrical ketones. It seems that straight chain esters were removed mainly by condensation reaction of the chain esters. On the other hand, bulky esters of CNF (Cyc) and CNF (Piv) were removed only by direct ester bond cleavage by steric hindrance effect of bulky esters. The lower onset of thermal degradation displayed by CNFs esters with straight-chain carbon esters than those of bulky esters is attributed to the possible lower energy to deprotect by condensation reaction than direct ester bond cleavage.

The 5% weight loss temperatures of the esterified CNFs are all higher than that of the unmodified CNF, and they are all above 300 °C. Thus, the esterification used in this study is great useful for the protection against the thermal deterioration and disengagement during melt compounding and injection molding.

The dispersibility of esterified CNFs in HDPE and the interaction between CNFs and HDPE at the melted condition were evaluated by X-ray CT (resolution: 700 nm) and by the G′ value of the CNFs/HDPE composite at the melted condition obtained by dynamic mechanical analysis, respectively.

The X-ray CT images of esterified CNFs-reinforced HDPE are shown in Fig. 3 along with the images of unmodified CNF-reinforced HDPE and neat HDPE. The X-ray CT images displays density distribution detected by X-ray intensity attenuation. The resolution of 700 nm in this work is insufficient to allow a detailed evaluation of the nanocomposite materials, but the 3D images do offer useful information about agglomerations of cellulose fibers in the low density polymers such as HDPE (Sato et al. 2016; Sakakibara et al. 2016) and acrylic resin (Yano et al. 2014). In this study, the cellulose with a density of 1.6 g/cm³ and HDPE with a density of 0.92 g/cm³ is represented by white bright color and blue color, respectively. The bright points observed in the CNF-reinforced HDPE indicate the CNF aggregates with the size over 700 nm.

The agglomeration to produce the aggregates seems to mostly have occurred during melt compounding and injection molding, because NBKP fibrillated to nanofibers as shown in Fig. 2 was carefully esterified to avoid CNF agglomeration. The agglomeration can be attributed to the difference in the interaction between the functional groups incorporated on the surface of the CNF and HDPE.

The size and frequency of the CNF aggregates differ among the composites. As shown in Fig. 3, many bright points were observed especially in unmodified-CNF-and also CNF(Ac, C2)-reinforced HDPE, but did not in CNF(Myr, C14) and CNF(Ste, C18)-reinforced HDPE, indicating that the incorporation of acyclic long alkyl chain on the surface of CNF enable the CNFs to well disperse in HDPE. This is because that the long alkyl chains resembling the repeating unit of PE thickly cover the surface of CNFs.

The CNF(Cyc, C7) with the different carbon skeleton from the long alkyl chain is suggested not to disperse well in HDPE. Actually, the CNF(Cyc, C7)-HDPE composite gave a many bright spots, but interestingly the spots markedly disappeared in the case of the CNF(t-BuCyc, C11), indicating that the additional incorporation of bulky t-butyl group enables to improve its dispersibility.

Finally, the CNF (Piv, C5)-HDPE composite which is reinforced with CNF having only the simplest bulky t-butyl group gave comparable X-ray CT image with that of the CNF(t-BuCyc, C11). Thus, we found that terminal bulky t-butyl group play an important role in improving the dispersity in HDPE, in spite of the completely different carbon skeleton for HDPE.

Temperature dependence of the shear modulus (G′) of the esterified CNFs/HDPE composites are shown in Fig. 4 along with the unmodified CNF/HDPE composite and neat HDPE. The results are summarized in Table 4.

The CTE values were determined as the mean values in the range 50–100 °C in the second run. The numbers in parentheses indicate the standard deviations.

As shown in Fig. 4, the G′ value of HDPE decreases at around 120 °C and becomes 0.0036 MPa at 140 °C owing to melting. When untreated CNFs were added, the G′ value at 140 °C is about 20 times higher than that of HDPE and maintains an almost constant value up to 160 °C. This constant G′ value can be explained by the formation of the mechanical network of CNFs whose Young’s modulus does not change until 200 °C (Nishino et al. 2007; Nakagaito et al. 2009).

The G′ values at the melted condition are significantly different among the esterified CNFs. The combination of unmodified CNF and HDPE gives the lowest G′ value at 140 °C, followed by CNF(Ac)/HDPE. Unmodified CNF and CNF(Ac) result in many aggregates in the composite, as shown in X-ray CT, indicating a poor affinity with HDPE. On the other hand, CNF(Cyc) which also exhibit pooor dispersibility comparable with that of CNF(Ac) (Fig. 3), exhibit a similar G′ value to those of CNF(Myr) and CNF(Ste) at the melting condition, which show good dispersibility in HDPE. Judging from G′ at the melted condition and X-ray CT analysis the rigid ring structure incorporated on the CNFs seems to restrain movement of PE molecule chains more effectively than the flexible long alkyl chain. Introduction of a tertiary butyl group to cyclohexane ring further increases the G′. This can be explained by the improved dispersibility of the CNFs, as described above.

An interesting result is observed for CNF(Piv)-HDPE composite. The CNF(Piv) has poor dispersibility in HDPE compare to CNF(Myr) and CNF(Ste) with long alkyl chain as shown in Fig. 3, however, the G′ value at 140 °C is highest in all esterified CNFs-reinforced HDPE composites. Although the mechanism should be further investigated, it is worth noting that the interaction between esterified CNFs and HDPE cannot be explained by the only the dispersibilituy of the CNFs in the matrix HDPE.

The Young’s modulus, and tensile strength of the injection molded samples are compared in Table 4. The typical stress–strain curves in tension and the thermal expansion curves are shown in Fig. 5. The densities of the samples are in the range 0.96–1.02 g/cm³. The cellulose content was fixed at 10 wt% to compare the effect of surface modification of the CNFs.

Addition of acetylated CNFs increases the Young’s modulus of HDPE from 1.20 to 2.25 GPa, and the tensile strength from 23.4 to 39.6 MPa. The effect of acetylation is not significant when it is compared with untreated-CNF-reinforced HDPE. As shown in Figs. 3 and 4, CNF(Ac) does not disperse well in HDPE and the G′ value at the melting condition is low. This poor affinity of CNF(Ac) for HDPE is the main reason for the low reinforcing efficiency. Increasing the length of the alkyl chain improves the mechanical reinforcement efficiency of CNFs. The CNF(Myr) attains a Young’s modulus of 2.60 GPa, which is approximately 2 times that of neat HDPE, and a tensile strength of 50.4 MPa. The increase of the affinity with the increase of the length of the alkyl chain results in good dispersion of CNFs and thus increases the interphase area between the CNFs and HDPE, contributing to the high reinforcing efficiency of HDPE. However, further increasing the alkyl chain length, such as CNF(Ste), decreases the Young’s modulus of the composite, although stearoylation achieves the best dispersibility of CNFs. It is possible that the longer chain length plasticizes the HDPE because of the flexible and bulky functional groups.

Introduction of rigid functional groups was then evaluated using cyclic fatty acids to compare with linear fatty acids. CNF(Cyc) increases the Young’s modulus of HDPE to 2.78 GPa, and interestingly higher than that of CNF(Myr)-reinforced HDPE. As shown in Fig. 3, the CNF(Cyc) shows poor dispersibility in HDPE, although they have a similar G′ value to CNF(Myr) at the melting condition. Thus, rigid cyclic fatty acid structures effectively inhibit deformation of HDPE molecules and achieve higher reinforcing efficiency than linear fatty acid structures irrespective of their poor dispersibility. The effect of addition of a tertiary butyl group to the cyclic fatty acid structure was then investigated. Although introduction of a tertiary butyl group to cyclohexane improves the dispersibility of CNFs in HDPE, the degree of the improvement in the mechanical properties is relatively small.

The CTEs calculated between 50 and 100 °C are shown in Table 4. In general, there is a relationship between the degree of increase of the rigidity and the degree of decrease of thermal expansion in CNF composites (Nakagaito and Yano 2008). The decrease in the CTE is pronounced in the composites reinforced with CNFs incorporating long alkyl chains, such as myristoyl and stearoyl groups. In addition, despite its poor dispersibility, CNF(Cyc) exhibited a similar reinforcing efficiency to CNF(Myr). As observed in the Young’s modulus, the highest reinforcing efficiency in the CTE is attained with CNF(Piv). The CTE is 72.5 ppm/K, which is less than one-third that of the neat resin.

The highest reinforcing efficiency in the Young’s modulus and CTE attained with the CNF(Piv) indicates that a chemical structure that strongly restrains HDPE molecules at the melting condition gives a high reinforcing efficiency after injection molding. Analysis of the mechanism of the high reinforcing efficiency would give insight into design of chemical structures incorporated on CNFs for polymer reinforcement. Hence, TEM observation of injection-molded CNF-reinforced HDPE was performed (Fig. 6). The TEM image of CNF(Piv)-reinforced HDPE reveals an interesting polymer structure, that is, a CNF “shish”/HDPE “kebab” structure and a HDPE “shish”/HDPE “kebab” structure. The white lines indicated by green arrows are HDPE “shish” structures. The flow-induced HDPE shish structure is a semicrystalline extended polymer chain (Mitsuhashi 1964; Pennings and Kiel 1965; kimata et al. 2007), and the crystal of HDPE is known to exhibit incredibly high elastic modulus such as 240GPa (Sakurada et al. 1966). This HDPE “shish”/HDPE “kebab” structure is slightly observed in CNF(Ac)-reinforced HDPE and is not clearly observed in CNF(Cyc)-reinforced HDPE and CNF(t-BuCyc)-reinforced HDPE. Furthermore, for the CNF(Myr), irregular growth of HDPE kebab structures is observed. The high reinforcing efficiency of CNF(Piv) can be partially attributed to generation of extended HDPE chain structures as well as a well-ordered folded HDPE crystal structure. The kebab structures disappeared when the composite was melted and cooled. This indicates that chemically modified CNFs do not act as a nucleating agent. Because CNF(Piv)-reinforced HDPE exhibits a high G′ value at the melted condition, it seems that the shear flow of the melted polymer in the injection molding die is trapped by the CNF network and polymer chains are subjected to high shear stress because of the surface chemical structure of the CNFs, which generate a CNF shish/HDPE kebab structure as well as a HDPE shish/HDPE kebab structure. It is worthy to note that surface modification of CNFs controls the crystal structure of HDPE during injection molding.