3.1. Macrodispersion of MWCNTs
To adjust suitable melt mixing parameters for optimized dispersion and distribution of the MWCNTs within the block copolymer melt, the processing temperature and screw speed were varied for composites containing 1 wt% of MWCNTs. The macrodispersion of the MWCNTs was evaluated on the ultrathin slices from the extruded strands for different mixing conditions using TLM as displayed in Figure 2
. The letters A, B, C, and D in the sample designations refer to the specific processing conditions, as given in Table 2
. The macrodispersion of the MWCNTs improves with temperature and screw speed, as indicated by the significantly decreased agglomerate area ratio (A
) from A
= 1.22 % (for processing condition A) to A
= 0.40 % (for processing condition D). When comparing the evaluated A
-values of processing condition B (A
= 0.68 %) and C (A
= 0.94 %), it can be assumed that the increase of screw speed from 100 rpm to 200 rpm has a more significant influence on the improved MWCNT dispersion than the increase of temperature from 180 °C to 200 °C. This assumption is also supported by the standard deviation values of A
(displayed in Figure 10, which will be discussed in Section 3.3
), which are distinctly lower for processing conditions B and D (higher scew speed) than for processing conditions A and C (lower screw speed). This implies the presences of larger remaining MWCNT agglomerates when applying lower screw speed, i.e., lower shear forces during melt mixing. Furthermore, on the basis of these results, composites with 0.1–0.5 wt% MWCNTs and the neat LN3 matrix (as reference) were processed under the condition D. The macrodispersion of MWCNTs in these composites was observed to be very good with values of A
≤ 0.07% (see Figure 10) and thereby indicating a good distribution and dispersion of the primary CNT agglomerates within the BCP matrix. It has to be considered that only agglomerates with diameter > 5 µm are included in the analysis of the agglomerate area ratio.
3.2. Block Copolymer Nanostructure and Phase Behaviour
The linear S-SB-S triblock copolymer used in this work exhibits a lamellar morphology. Its phase behavior and nanostructure have been investigated in former studies [5
]. The outer PS-blocks act as glassy domains comprising ~50 wt% of the overall block copolymer composition, and the SB middle block acts as rubbery phase. This hard and soft phase composition ratio of 50:50 causes the formation of a lamellar morphology instead of a hexagonal morphology, which may theoretically be expected for linear SBS triblock copolymers having the same overall PS content of 75 wt% [1
]. The long period of the lamellar structure and the domain sizes of the PS-rich and PB-rich phases have been found to be significantly dependent on the processing conditions. Average values of lamellar sizes are given in Table 3
. It has to be noted that all samples show a broad distribution of the measured lamellar sizes, as shown by their standard deviations (Table 3
). For example, long periods in the range of ~14 nm to ~46 nm for LN3_D were observed, as revealed from image analysis. Shear forces during the extrusion process result in considerable orientation effects of the molecular chains, indicated by stretched or widened lamellae. Such orientation effects are mainly responsible for the variations in long period and domain sizes of the PB-rich and PS-rich phases within the lamellar block copolymer morphology. Therefore, a detailed discussion about the size variations among the various samples should be treated with caution, and hence the influence of processing temperature and/or screw speed on the lamellae size cannot be evaluated.
displays the lamellar morphology of sample LN3_A. The addition of 1 wt% MWCNTs does not change the lamellar matrix morphology, as shown in the TEM image in Figure 3
b. Figure 3
c represents the local dispersion of MWCNTs within the LN3_A matrix on a non-contrasted sample. Such an observation may lead to the understanding that the incorporation of MWCNT (1 wt%) in the given processing conditions (A) does not alter the thermodynamic stability of the block copolymer nanomorphology.
However, the average lamellar long period thicknesses of the composites were measured to be significantly reduced, in contrast to their block copolymer matrix LN3 processed under identical conditions (processing conditions C and D). Such an observation is particularly striking in case of samples LN3_C and LN3_1_C, as revealed by their corresponding long periods (Table 3
) as shown in the morphology images given in Figure 4
shows the morphologies of LN3_D and its corresponding nanocomposites. All samples show lamellar structures, whereby the orientation of the lamellae and the long-range order of the nanostructures are differently pronounced. For nanocomposites with MWCNT contents of 0.5 wt% and less, the MWCNTs are hardly visible in the images. In the TEM image of sample LN3_1_D, well-distributed MWCNTs penetrating the nanomorphology can be identified (Figure 5
e). It is difficult to judge if or to what extent the CNTs are preferentially located in one of the BCP phases or at the interface. The TEM image in Figure 5
f presents a longitudinal view into the extruded strand of LN3_1_D (see also Figure 1
), showing significant orientation of the lamellae along the melt flow direction. The MWCNTs visible in the image do not show a preferred orientation direction. MWCNTs agglomerate with diameters smaller than 1 µm are distributed in the matrix and interfere with the morphology formation.
shows two enlarged TEM images of the morphology of sample LN3_1_D. Due to the size ratio between the BCP nanodomains and the MWCNTs, which is illustrated by Figure 6
a, it is hardly possible to localize CNTs in one of the BCP phases, especially the longer MWCNTs. Figure 6
b shows a magnified area of the BCP nanostructure in which single MWCNTs seem to be preferentially anchored in the PB-rich phase. However, the MWCNTs are located mainly across the BCP domains, as shown in Figure 5
e, demonstrating that the MWCNTs, despite being shortened in a ball mill, are mostly too long to be localized in the PS or PB lamellas.
The SAXS analysis was performed for the pure LN3 samples and the nanocomposites with 1 wt% MWCNT. Primary information of the scattering behavior is illustrated in Figure 7
as SAXS patterns. From the radial scattering curves in Figure 8
, the formation of a lamellar morphology was confirmed for all samples exhibiting long periods between ~30 nm and ~32 nm. In contrast to the TEM images, no significant reduction of the long period could be observed with the addition of MWCNTs. The reason could be the different measuring ranges used for the calculation of the long period. While the TEM image analysis is performed in a range of a few µm2
area with a sample thickness of only 80 nm, X-rays with a beam diameter of about 0.5 mm irradiate through a sample thickness of about 2 mm (see Figure 1
). This means that the SAXS analysis provides mean values of the long periods over the entire strand thickness, while the TEM images show the morphology of very local, high-resolution areas. For the pure LN3 samples, a significant orientation of the lamellar structure in the direction of the axis of the extrusion strand is detectable, as shown for LN3_A in Figure 7
a. The degree of orientation reduces by the addition of the CNTs (Figure 7
b). As shown by the radial intensity curves in Figure 8
, the phase separation is enhanced by the addition of MWCNTs. Due to the presence of amorphous MWCNTs, an additional scattering maximum in the d-range of 4 nm to 10 nm occurs that is similar to non-phase separated behavior. Due to the particle scattering of the MWCNTs, the total scattering intensity of the composites is increased compared to the pure LN3 samples.
From DMA measurements, the glass transition temperatures of the PB-rich and PS-rich phases of the LN3 block copolymers were evaluated. For the analysis, specimens were taken in horizontal and vertical directions from the pressed plates from each composition in order to take into account possible orientation effects of the lamellar structure as observed by TEM and SAXS. This means that for each composition, the curves of two specimens prepared perpendicular to each other are shown. To get a better overview, only the tan δ
curves for neat LN3_D and composites melt processed under the same conditions (D) are plotted in Figure 9
. The glass transition temperature of the PB-rich phase (mixed phase of styrene and butadiene) of all investigated samples remained lower than the theoretical values estimated using Fox equation [52
] for a corresponding random SB copolymer (Tg-SB
= −37 °C). Such an observation indicated a partial miscibility of the SB middle block with the PS outer blocks. A broad peak-maximum rather than a sharp peak was observed for Tg
of the PB-rich phase, with maximum values ranging from -24 °C to −12 °C. Obviously, these variations are caused by the alignment of the polymer chains during the extrusion process. Comparing the two curves of one composition, which refer in each case to the specimens prepared perpendicular to each other, the tan δ
curves show a clear dependence of the peak intensity of the PB-rich phase on the structural orientation. This is particularly pronounced in the case of sample LN3_D (black curves in Figure 9
), where the peak intensities of both curves differ significantly from each other. The reasons behind such effects are not fully understood. However, the possibility of CNTs being aligned during the shear flow of the melt, which in turn may lead to higher peak intensity, should be considered.
No significant influence of CNT addition and processing parameters on Tg
could be determined. The Tg
of the PS-rich phase for the neat LN3 samples and composites remains unaffected in the form of a sharp peak at ~92 °C irrespective of processing types. Such an observation remained well in accordance with the phase behavior of previously investigated starblock copolymer/MWCNT-based composites [31
3.3. Structure–Property Correlations (Electrical and Mechanical Properties)
The electrical properties of the BCP/MWCNT composites are strongly dependent on the state of dispersion and electrical percolation of the nanofiller in the BCP matrix. As shown in Figure 10
, composites with a filler content up to 0.5 wt% MWCNTs with very low agglomerate area ratios are non-conductive and thus indicate a well dispersed morphology with isolated MWCNTs distributed within the BCP matrix. Composites containing 1 wt% of MWCNTs are electrically conductive, having a very low volume resistivity of ~8.1 × 103
Ω·cm when processed under condition A and ~1.8 × 103
Ω·cm when processed under condition B. Sample LN3_1D exhibits the lowest agglomerate area of the conductive samples due to the optimized melt mixing parameters, such as a temperature of 200 °C and a screw speed of 200 rpm. The electrical volume resistivity of LN3_D was measured to be ~3.0 × 103
Ω·cm. Thus, the percolation threshold for LN3/MWCNT composites remains between 0.5 wt% and 1 wt% of MWCNT content. Such a percolation threshold range is lower than in recently investigated melt-mixed MWCNT filled composites based on a styrene-butadiene starblock copolymer (3G55, Styrolution Group) such as the matrix with an overall PS content of 75 wt% and a complex wormlike morphology due to the presence of short, long, and highly curved lamellae [31
]. In that system, another grade of MWCNTs named NC3150 from Nanocyl was used as a filler, which has higher carbon purity than NC7000 (~95 %) and with an average length of <1 µm. It is apparent from these observations that to reach low resistivity values like that in LN3/MWCNT composites, ~2 wt% of NC3150 may be necessary to add into the star block copolymer matrix.
To study the influence of the carbon nanofiller on the mechanical properties of the triblock copolymer, tensile tests were performed on specimens cut from compression-molded plates, which were processed using granules as supplied by the manufacturer. The mechanical characteristics of the so-called LN3_neat sample, compared to the LN3 types and their respective LN3/MWCNT composites containing 1 wt% MWCNTs processed under conditions A, C, and D, are presented in Figure 11
. To provide further information about the deformation behavior of each sample, the stress-strain diagrams of all samples are available in the Appendix A
in Figure A1
. Each diagram contains the curves of five tested specimens. LN3_neat exhibits very ductile behavior, with large strain at break of ~ 415 %, but also large average value of Young’s modulus of ~915 MPa (Figure A1
a), indicating high resistance against elastic deformation, i.e., high stiffness. Depending on the orientation of the polymer chains and hence on the alignment of the lamellae within the tested specimens of sample LN3_neat, the Young’s modulus and yield point exhibited significant variations. This is also evident from the large standard deviations of these mechanical values, as shown in Figure 11
a,b. As discussed by Allan et al. [53
], samples with orientation of the lamellae parallel to the deformation direction exhibit the highest Young’s modulus, followed by samples with perpendicular direction of the lamellae. Lowest moduli occur if the lamellas are aligned at an angle of 45 ° to the deformation direction. Structural orientations of these samples were mainly caused by the extrusion process within the granules of the extruded strands, followed by the subsequent compression molding process. Thus, depending on the orientation of the lamellae within the deformed region of the specimens, the nature of the material ranges from thermoplastic with distinct yield point and high ductility to thermoplastic elastomeric behavior indicated by the absence of a yield point and large strain prior to failure. Detailed research on the influence of ram extrusion with different shearing rates on structure and mechanical properties of such triblock copolymer was published by Mahmood et al. [54
] but is not the focus of this study.
Similar mechanical behavior like that of LN3_neat could be observed for the melt-extruded samples of LN3_A and LN3_C (Figure A1
b,d). These samples exhibited large strain at break and tensile strength values similar to the neat LN3 (Figure 11
c,d), but lower stiffness, due to larger influence of processing-induced orientations of the lamellar morphology. However, in sample LN3_D the ductility has been observed to reduce significantly, and large variations in the magnitudes of the mechanical properties implied defect areas such as those originating from voids and melt inhomogeneities in the pressed plates.
Interestingly, the mechanical property values of the corresponding LN3/MWCNT composites with 1 wt% nanofiller content are stabilized compared to the pure LN3 samples. By adding MWCNTs, structural orientation effects decreased and there were no more large fluctuations of the Young’s modulus within a sample series (Figure 11
a). The composites showed a stable yield point (Figure 11
b) accompanied with the unaltered tensile strength (Figure 11
c) and high elongation at break of the pure BCPs (Figure 11
Furthermore, a slight influence of processing conditions on the mechanical properties could be observed from the data shown in Figure 11
. With increased screw speed (processing condition B and D), an enhancement in Young’s modulus and yield stress of the composites LN3_1_B and LN3_1_D compared to LN3_1_A and LN3_1_C, respectively, could be observed. This can be attributed to the better dispersion and more homogeneous distribution of the CNTs due to the higher shear forces applied during melt mixing, which was indicated by the lower values of the agglomerate area ratio and their variations (Figure 10
). Similar observations were made by Arrigo et al. [55
] in composites of SBS and modified MWCNTS containing COOH groups, which were melt mixed in a Brabender mixer at 50 rpm and 100 rpm. The study also reported a significant influence of processing temperature on the mechanical properties, an observation that could not be validated from the results obtained in the context of the present work.