2.1. Dispersing CNTs
Our initial attempts to hydrate clays involving simultaneous heating (80 °C), stirring and sonicating for up to two days were unsuccessful. The resulting clay suspension was unstable and it was not possible to obtain a stable CNT dispersion. This suggested that this treatment does not fully delaminate the clay's platelet layers. Full delamination is only achieved through vigorous application of mechanical force or with the assistance of surfactants, as shown previously [
14]. Therefore, all our clay suspensions were hydrated using a homogenizer. Light microscopy images (
Figure 1a and b) show that the presence of aggregates is significantly reduced after hydration.
Single-walled carbon nanotubes (SWNT) were easily dispersed in these properly hydrated clay suspensions (1.12% w/v, pH = 7.9), and were stable for months (
Figure 1c). Typical UV-visible spectra (
Figure 1d) show broad CNT absorption features due to the presence of nanotube aggregates. The absorbance of the dispersions at 747 nm was plotted as a function of concentration (inset in
Figure 1d). This particular wavelength was selected as it corresponds to the maxima of an absorption band arising from the van Hove singularities for SWNT [
15,
16].
Figure 1d shows that the absorption intensity increases linearly with increasing carbon nanotube concentration, indicating an excellent degree of disperse-ability (in the concentration range studied). This allowed us to determine the extinction coefficient (ε) of CNTs in the clay suspension, yielding ε= 0.864 mL mg
−1cm
−1.
2.2. Rheological Studies
Rheological studies were undertaken to examine the flow and time-dependent behavior of the clay-CNT dispersions as well as the effect of incorporating chitosan. Both types of dispersions and the chitosan solution display shear thinning behavior,
i.e., viscosity (η) decreases with increasing shear rate (data not shown). Combining chitosan with clay–CNT into a clay–CNT–chitosan dispersion results in a two and three orders of magnitude decrease in the apparent viscosity compared to that of the chitosan solution and clay-CNT dispersion, respectively. For example, at a shear rate of 0.01 s
−1 the viscosity values of typical chitosan solutions, and clay–CNT (1000 mg/L) and clay–CNT–chitosan dispersions are 15.4 Pa.s, 370 Pa.s, and 0.266 Pa.s, respectively. The apparent viscosity of the ternary dispersion is lower than the oppositely charged solutions used to form the dispersion,
i.e., the anionic clay–CNT dispersion and the cationic chitosan solution.
Figure 2a shows that the clay-CNT dispersion exhibits a yield point,
i.e., the sample starts to flow only when a certain amount of force is applied. This point can be determined using the Bingham model [
17],
where τ
B and τ
B indicate the Bingham yield point and Bingham flow coefficient, respectively. Although, the values obtained using the Bingham model are dependent on the shear rate range it provides a good approximation for the determination of yield points. The model shows that the yield point of clay–CNT dispersion decreases by 2-orders of magnitude upon addition of chitosan (
Table 1). Similar differences are observed for the Bingham flow coefficient.
These results indicate that the electrostatic interaction between the negatively charged clay and positively charged chitosan decreases the resistance against flow. Similar observations have been reported previously for the addition of other types of clay (montmorillonite) to chitosan [
18]. This study showed that a decrease in the electrostatic potential of chitosan upon addition of clay was coupled with a decrease in flow resistance [
18].
Thixotropic behavior testing (
Figure 2b) revealed that clay–CNT materials exhibit the expected time-dependent rheology characteristics consistent with a “house of cards” structure [
2]. As evident from the 20% decrease in viscosity during the reference and high-shear intervals applying a constant shear, results in disruption of this structure. During the regeneration interval, clay–CNT dispersions exhibit a rapid increase in viscosity, which is indicative of the re-building of the colloidal structure. Eventually, the viscosity will start to decrease again due to effect of applying a constant shear rate (
Table 1). In contrast, chitosan does not show any of these characteristics,
i.e., the viscosity does not exhibit any significant time-dependent behavior in any of the three intervals. Whereas, combining chitosan with clay–CNT results in a dispersion which has retained the time-dependent characteristics of clay–CNT dispersions. The difference in the magnitude of these viscosity effects is evident from the inset in
Figure 2b,
i.e., a binary dispersion can be easily inverted without flowing, whereas the ternary composite will still flow.
Oscillatory amplitude sweeps confirmed that the ternary (clay–CNT–chitosan) dispersion has more in common with the binary (clay–CNT) dispersion than the chitosan solution (
Figure 2c–d). Both dispersions display distinctive linear viscoelastic (LVE) regions, although the length of LVE region and maximum shear stress is lower for the ternary dispersion due to the presence of chitosan (
Table 1). The magnitude of the storage (G′) and loss (G″) moduli of the ternary dispersion (in the LVE region) is lower than those of the binary dispersion. This difference is also reflected in the shear modulus obtained using G* = ((G′)
2 + (G″)
2)
1/2, resulting in values of 80.6 ± 1.9 Pa and 1.40 ± 0.23 Pa for the binary and ternary dispersions, respectively. The corresponding value for chitosan is 4.81 ± 0.08 Pa. The value for the clay-CNT dispersion is similar to that of typical dispersions such as lotions and creams, whereas that of chitosan and clay–CNT–chitosan is comparable to that of salad dressings [
19].
Figure 2c–d shows that for the dispersions, the storage modulus (G′) is larger than the loss modulus (G″) in the LVE region, indicating that the elastic behavior dominates over the viscous behavior. In contrast, chitosan solutions exhibit the opposite trend,
i.e., viscous behavior is dominating (G′ < G″). Above the maximum shear strain, a cross-over from elastic to viscous behavior (tan δ > 1,
Figure 2d) takes place for both dispersions indicates a disruption of the “house-of-card” structure. Furthermore, the strain at which the cross-over takes place is larger in the binary dispersion than that in the ternary dispersion. These results clearly indicate lower resistance to flow behavior due to addition of chitosan.
2.3. Clay–CNT Films
Free-standing clay–CNT films (
Figure 1a) were prepared by evaporative casting of clay–CNT dispersions. The current–voltage (
I–
V) characteristics were investigated under controlled ambient conditions (21 °C, 45% relative humidity, RH). All films exhibited linear
I–
V characteristics, which indicate Ohmic behavior. The conductivity (
σ) can then be evaluated by making resistance measurements as a function of sample length (
l) [
20]. The total resistance was found to scale linearly according to:
where
Ac is the film's cross-sectional area. The straight line fit for a typical film with nanotube mass fraction 0.067 is shown in
Figure 3b. The slope is used to calculate the so-called two-probe dc conductivity, yielding 0.14 ± 0.04 S/cm under controlled ambient conditions.
Previously, we have demonstrated that (dried) CNT composite materials prepared using water soluble dispersants change their electrical behavior upon hydration. For example, exposure to a humid atmosphere resulted in an increase in electrical resistance for water soluble polyaniline and polypeptide-CNT composite materials [
21,
22], while gellan gum–CNT composite materials decrease their resistance [
20,
23,
24]. It was demonstrated that resistance decreased due to an increased cation mobility upon exposure to humid atmosphere [
20].
Figure 3c shows that exposing our clay–CNT film to humid atmosphere for 15 hours results in a decrease in the current compared to that observed under ambient conditions. This decrease in current corresponds to an increase in electrical resistance, from 9.7 ± 2.0 kΩ (R
B, before exposure) to 36 ± 4 kΩ(R
A, after exposure). Exposure to the humid atmosphere results in hydration of the clay–CNT film,
i.e., osmotic forces drive water in between the smectite platelet galleries. This leads to a swelling-induced disruption of conductive pathways resulting in an increase in resistance.
Figure 3d shows that the current response to a square wave potential is different before and after exposure to humid conditions. Under ambient conditions (before exposure) the magnitude of the current response to a square wave potential is linear, while after exposure to humid atmosphere the current displays non-linear behavior.
This behavior can be explained through the mobility and charge collection of the counter-ions. Under an applied positive potential the counter-ion (cations) will migrate towards the negative electrode (1) leading to a buildup of positive charge. Upon reversal of the potential, the cations will be repelled from the now positive electrode 1 causing a non-linear current flow due to migration of the ionic charge carriers (indicated in the circled area in
Figure 3d). The cations migrate towards the negative electrode (2) leading to a charge collection at this electrode. This effect manifests itself as the non-linear current response, until all mobile ions have migrated and the current becomes linear again.
Thus, the resistance of our composite material consists of an electrical contribution from electron transport through the carbon nanotube network and an ionic contribution due to the cations. The latter is small or negligible under ambient conditions. Under humid conditions we would expect a decrease in resistance due to an increased ionic contribution, similar to that observed in our previous work on composites consisting of the anionic polysaccharide gellan gum and SWNT [
20,
24]. However, the swelling-induced disruption of conductive pathways results in a more significant reduction in the electrical contribution (−70%, estimated from
Figure 3d). As such the resistance of a hydrated film is higher compared to that of a dry film.
Clay–CNT dispersions were used to fabricate buckypapers via vacuum filtration. The two-probe dc conductivity of a typical buckypaper yielded 0.9 ± 0.2 S/cm under controlled ambient conditions (
Figure 3b). As expected, the buckypaper conductivity is higher compared to the conductivity (0.14 ± 0.04 S/cm) of the evaporative cast film. Exposure of buckypapers to humid atmosphere resulted in a swelling-induced decrease in the current (increase in resistance), but we did not observe any non-linear current behavior in response to a square wave potential. This indicates that most of the counter-ions were removed during the washing procedure in the buckypaper preparation method.
2.4. Clay–CNT–Chitosan Films
The clay–CNT films produced by evaporative casting and vacuum filtration were too brittle to allow a detailed analysis of their mechanical properties, i.e., the films could not be subjected to any significant strain without breaking. Polyelectrolyte complexation of the negatively charged, hydrated clay platelets with the positively charged biopolymer chitosan was utilized to improve the mechanical robustness of these materials, i.e., the materials could be subjected to strain.
Free-standing ternary clay–CNT–chitosan composite films (
Figure 4a) were prepared by evaporative casting of clay–CNT–chitosan dispersions with CNT mass fraction of 0.028. The resulting materials were more mechanically robust compared to clay–CNT films, allowing for an assessment of their mechanical properties (see
Figure 4b). Combining clay–CNT with chitosan results in an improvement in Young's modulus (E), coupled with a decrease in tensile strength and strain at break values compared to chitosan (
Table 2). More significant increases in E as well as an increase in TS have been observed for composites prepared using functionalized multi-walled carbon nanotubes (FMWNT, see also
Table 2) [
7,
9]. This larger increase can be attributed to the presence of carboxy and hydroxyl functional groups on the nanotube surface, which facilitates an improved interfacial adhesion between clay and chitosan through electrostatic interactions and hydrogen bonding, compared to the non-functionalized SWNT used in our composites. Larger increases in modulus were also observed for composites prepared using other matrix materials (epoxy and latex) in combination with carbon black and FMWNT (
Table 2) [
8,
10–
12].
The increased robustness of the ternary (clay–CNT–chitosan) composite materials is coupled with a decrease in conductivity by 3-orders of magnitude (from 0.14 S/cm to 1.0 × 10
−4 S/cm) compared to the binary (clay–CNT) composites, see
Table 2. These observations suggest that chitosan may act as “glue” or “binder” between the clay–CNT domains thereby improving the mechanical properties, as suggested previously [
25]. However, the significant reduction in conductivity suggests that the number of electrical (CNT–CNT) pathways has decreased and the number of ionic-electrical pathways has increased compared to clay-SWNT films,
i.e., pathways dominated by chitosan and clay–chitosan. This is evident from the difference in surface morphology between the two types of films. The CNT pathways are clearly visible in the clay–CNT film (
Figure 3a), but almost entirely covered by the biopolymer in the clay–CNT–chitosan film (
Figure 4a). We were unable to compare our conductivity values with that of the other clay–CNT–chitosan materials shown in
Table 2, due to lack of available data (at least to our knowledge). However, our conductivity value is in the same order of magnitude as clay–CNT–epoxy materials, with higher values (8.6 mS/cm) reported for carbon black (CB) containing materials (
Table 2).
Under ambient conditions (in the absence of water vapor), chitosan and clay act as tunneling barriers in these junctions thereby blocking transport. We have already seen that exposure to humid atmosphere of clay–CNT materials results in an additional contribution to the current. As chitosan is a cationic polyelectrolyte, exposure to humid atmosphere increases the counter-ion mobility allowing these anionic charge carriers to transport the current along the polymer component of the chitosan-dominated junctions. This may enable transport through these pathways leading to an additional contribution to the current.
Despite the increase in current (as a result of increased ion-mobility), the resistance of the clay–CNT films increased upon exposure to humid atmosphere due to a swelling effect.
Figure 4c shows that the clay–CNT–chitosan films exhibit different behavior. The current magnitude increases with increasing time of exposure to humid atmosphere. After 140 min of exposure the resistance has decreased by one order of magnitude from R
B = 2.8 ± 0.6 MΩ to R
A = 0.27 ± 0.08 MΩ (see also
Table 2). It is likely that interactions between the oppositely charged clay and chitosan materials limits expansion (swelling) of the clay. As such swelling-induced disruption of conductive pathways (resulting in an increase in resistance) is not significant in these composites. The decrease in resistance can then be attributed to enhanced ion-mobility of the clay and chitosan counter-ions.
These ternary composite materials showed another interesting and somewhat unexpected response to humidity. Exposing one face of the film to a higher humidity than the other face, results in rapid curling (
Figure 4c). This response was found to be reversible,
i.e., the film uncurled upon removal of the humidity gradient. This may suggests that water is adsorbed into the inter-layer spacing on only one side of the film; expansion of that side relative to the other (dryer) side results in the curling actuator response. The actuator response (the level of reversible curling) was better for dry films compared to hydrated films. The latter do not exhibit the same degree of actuation as transport of water in and out of the film becomes more uniform and with it, the amount of expansion.
2.5. Clay–CNT–Chitosan Fibers
In our previous work we prepared fibers by facilitating polyelectrolyte complexation through injection of a SWNT-biopolymer dispersion into a coagulation bath containing a biopolymer of opposite charge [
23]. Initial attempts to produce fibers via this approach,
i.e., injection of a clay–CNT dispersion into a chitosan coagulation bath, were unsuccessful. The resulting fibers were not mechanically robust enough to be recovered after passing through the coagulation bath. We suspect that this may be a result of the high yield strength (5.87 Pa) and apparent viscosity (370 Pa.s at 0.01 s
−1) of the clay–CNT dispersion which may inhibit the diffusion of chitosan and subsequent coagulation of chitosan with the clay platelets. In other words, during the continuous spinning approach the clay–CNT dispersion is passed too quickly through the chitosan coagulation bath to facility polyelectrolyte complexation.
We devised an alternative spinning method whereby the chitosan coagulation bath is replaced by a long coagulation channel into which a stream of a clay–CNT dispersion is injected, which remains in the channel for three hours. This is followed by removing the fiber from the channel to a supporting frame and drying under controlled ambient conditions. We refer to this modification of the continuous spinning approach as “stop-and-go wet spinning”. During the “stop stage”, the additional three hours in the coagulation channel, chitosan diffuses into the clay thereby facilitating the polyelectrolyte complexation. The gradual inclusion of the chitosan between the smectite platelets, causes a reduction in the thickness of the charged double layer responsible for face-face electrostatic repulsion of adjacent clays platelets. The observed shrinkage of the fibers is in support of this suggestion.
The stop-and-go spinning method allowed us to easily spin clay–CNT–chitosan fibers (
Figure 5a). These fibers (diameter 210 ± 40 μm) showed an interesting surface morphology as evident from the scanning electron microscopy micrographs (
Figures 5b and 5c). These ternary composite materials appear to be composed of numerous smaller fibers (diameter 23 ± 9 μm), producing a yarn like appearance. Similar surface features have been reported for other types of polyelectrolyte complexed fibers using gellan gum and chitosan solutions [
26].
Figure 4b and
Table 2 clearly show that a typical ternary composite fiber exhibits significantly higher E, similar TS and lower strain at break values compared to a typical ternary composite film. The electrical resistance of typical dry fibers (R
B = 300 ± 14 MΩ) is two order of magnitude higher compared to typical dry films of similar length, but due to the difference in the cross-sectional area of fiber and film samples the difference in conductivity is only 1 order of magnitude (
Table 2).
The fiber's electrical response to humid atmosphere is similar to that observed for clay–CNT–chitosan films. After 250 min of exposure the resistance has decreased by almost one order of magnitude from RB = 300 ± 14 MΩ to RA = 68 ± 4 MΩ. Swelling of the fiber in response to exposure to humid atmosphere was apparent through elongation of the fiber (+20%) within its constrained position in the environmental chamber. This swelling behavior was found to be reversible. Similar to the ternary film composites we do not consider the swelling-induced disruption of conductive pathways (resulting in an increase in resistance) to be significant in the fibers. As such the decrease in resistance is attributed to enhanced ion-mobility of the clay and chitosan counter-ions.