Effect of Ionic Polymer Membrane with Multiwalled Carbon Nanotubes on the Mechanical Performance of Ionic Electroactive Polymer Actuators

Ionic electroactive polymer (IEAP) actuators have received interest because of their advantageous properties, including their large displacement, high energy density, light weight, and low power consumption under a low electric field. However, they have a low blocking force under driving, and it is difficult to control the thickness of the ionic polymer membrane. In this study, an IEAP actuator is fabricated using a Nafion membrane with added multiwalled carbon nanotubes to increase the blocking force. A heat press two-step process is also developed to produce a constant and uniform membrane. The fabricated Nafion membrane with 0.2 wt% multiwalled carbon nanotubes has the largest displacement and highest blocking force. As a result, the developed heat press two-step method can be used in various polymer-casting fields, and the fabricated carbon nanotube-based IEAP actuators can serve as useful references in fields such as flexible robotics and artificial muscles.


Introduction
The advantageous properties of ionic electroactive polymer (IEAP) actuators include their large displacement, high energy density, light weight, and low power consumption under low electric fields [1][2][3][4]. Typically, IEAP actuators are composed of an ionic polymer membrane sandwiched between two noble metal electrodes [5,6], and the diffusion of hydrated cations within the membrane under an applied voltage and the associated electrostatic interactions induce bending.
Nafion membranes have excellent ionic conductivity and chemical and thermal stability, and they are suitable up to a thickness of 180 µm for application as ionic membranes in IEAP actuators [7]. Commonly, IEAP actuators are limited by the commercially available Nafion thicknesses (e.g., N115, N117, and N1110), and thus various studies on controlling their thickness using a Nafion solution have been reported [6,[8][9][10][11][12][13][14][15][16][17]. Furthermore, since the thickness of the membrane significantly affects its properties and experimental behavior, it is highly important to fabricate consistent and uniform membranes like those that are commercially available membranes in order to compare and verify experimental results. For these reasons, N,N -dimethylformamide (DMF) was used to cast the Nafion membranes in this study. DMF is highly compatible with the Nafion tetrafluoroethylene backbone, and thus enables the formation of a membrane in which the Nafion molecules do not aggregate [17]. It is also used as a solution for carbon nanotube (CNT) dispersions such as for surfactants [18][19][20]. However, in the Nafion membrane casting process, it proved to be difficult to acquire a uniform

Fabrication of N-MWCNT Membranes
The developed HPTS method employs a hot press and metal plate in two steps. Figure 1a shows step 1, in which the approximate film thickness was determined by the weight of the dispersion. Nafion solutions containing DMF and MWCNTs were mixed using a probe sonicator, and the mixture was poured into a preheated casting mold (9.0 × 9.0 cm 2 ). The filled mold was heated at 40 • C for over 12 h to vaporize the volatile components of the Nafion solution (deionized water and acetone) and thus obtain the casted N-MWCNT membrane. Next, in step 2 (Figure 1b), the casted N-MWCNT membrane from step 1 and support metal plates of the desired thickness were placed between substrates with high heat resistance and strength and were heat-pressed at 140 • C and 10 MPa. Herein, the metal plate that was used was not deformed by heat and pressure, had a clean surface, and was thinner than the membrane cast in step 1. Conditions of the process such as time, temperature and pressure were optimized through repeated experiments. Finally, we obtained N-MWCNT membranes with uniform thicknesses by the HPTS method.
(ionic liquid) purchased from Merck KGaA (Darmstadt, Hesse, Germany) were used for actuator ion exchange.

Fabrication of N-MWCNT Membranes
The developed HPTS method employs a hot press and metal plate in two steps. Figure 1a shows step 1, in which the approximate film thickness was determined by the weight of the dispersion. Nafion solutions containing DMF and MWCNTs were mixed using a probe sonicator, and the mixture was poured into a preheated casting mold (9.0 × 9.0 cm 2 ). The filled mold was heated at 40 °C for over 12 h to vaporize the volatile components of the Nafion solution (deionized water and acetone) and thus obtain the casted N-MWCNT membrane. Next, in step 2 (Figure 1b), the casted N-MWCNT membrane from step 1 and support metal plates of the desired thickness were placed between substrates with high heat resistance and strength and were heat-pressed at 140 °C and 10 MPa. Herein, the metal plate that was used was not deformed by heat and pressure, had a clean surface, and was thinner than the membrane cast in step 1. Conditions of the process such as time, temperature and pressure were optimized through repeated experiments. Finally, we obtained N-MWCNT membranes with uniform thicknesses by the HPTS method.

Fabrication of N-MWCNT-Based Actuators
Actuators were manufactured using the N-MWCNT membranes and GO-Ag NWs electrodes. We used the GO-Ag NWs (weight ratio 1:2.5) paper electrodes that Yoo, S. studied [40], which were made using a vacuum filtration system. Each N-MWCNT membrane fabricated by the HPTS method was placed between the GO-Ag NWs paper electrodes and attached using the heat press to yield the N-MWCNT-membrane-based actuator. The size of fabricated actuator was 4.0 × 0.5 cm 2 .

Analytical Techniques
We created N-MWCNT membranes with various MWCNT weight ratios for IEAP actuators and observed surface morphologies and thicknesses using field emission scanning electron microscopy (FE-SEM; S-4300, Hitachi, Tokyo, Japan) and a Micrometer analyzer. The thermal stabilities of the fabricated membranes were assessed by thermo-gravimetric analysis (TGA), and the mechanical performances of the N-MWCNT-based IEAP actuators were analyzed using a driving characteristics analyzer and tensile strength machine. The actuation performances of the actuators were measured using a laser displacement sensor (ZS-LD80, OMRON Korea, Seocho, Seoul, South Korea) and load cell sensor. The displacement was analyzed to the changing distance of the laser on the surface of the

Fabrication of N-MWCNT-Based Actuators
Actuators were manufactured using the N-MWCNT membranes and GO-Ag NWs electrodes. We used the GO-Ag NWs (weight ratio 1:2.5) paper electrodes that Yoo, S. studied [40], which were made using a vacuum filtration system. Each N-MWCNT membrane fabricated by the HPTS method was placed between the GO-Ag NWs paper electrodes and attached using the heat press to yield the N-MWCNT-membrane-based actuator. The size of fabricated actuator was 4.0 × 0.5 cm 2 .

Analytical Techniques
We created N-MWCNT membranes with various MWCNT weight ratios for IEAP actuators and observed surface morphologies and thicknesses using field emission scanning electron microscopy (FE-SEM; S-4300, Hitachi, Tokyo, Japan) and a Micrometer analyzer. The thermal stabilities of the fabricated membranes were assessed by thermo-gravimetric analysis (TGA), and the mechanical performances of the N-MWCNT-based IEAP actuators were analyzed using a driving characteristics analyzer and tensile strength machine. The actuation performances of the actuators were measured using a laser displacement sensor (ZS-LD80, OMRON Korea, Seocho, Seoul, South Korea) and load Polymers 2020, 12, 396 4 of 10 cell sensor. The displacement was analyzed to the changing distance of the laser on the surface of the actuator when AC voltage was applied to the actuator. The blocking force was analyzed to the contacting load cell sensor with the surface of the actuator when DC voltage was applied to the driver.

Results
Nafion membranes for use as ionic polymer membranes in actuators were made using evaporation and HPTS methods, and were compared. As shown in Figure 2a, the surface of the Nafion membrane made by evaporation method was not flat or uniform, and it is clear that the image behind the Nafion membrane is distorted. In contrast, the surface of the Nafion membrane made by HPTS method was smooth; the image is clear, well transmitted, and not distorted ( Figure 2b).
Polymers 2020, 12, 396 4 of 10 actuator when AC voltage was applied to the actuator. The blocking force was analyzed to the contacting load cell sensor with the surface of the actuator when DC voltage was applied to the driver.

Results
Nafion membranes for use as ionic polymer membranes in actuators were made using evaporation and HPTS methods, and were compared. As shown in Figure 2a, the surface of the Nafion membrane made by evaporation method was not flat or uniform, and it is clear that the image behind the Nafion membrane is distorted. In contrast, the surface of the Nafion membrane made by HPTS method was smooth; the image is clear, well transmitted, and not distorted ( Figure 2b).  Figure 3 shows the SEM cross-sectional images of the Nafion membrane fabricated with 15 g of the Nafion solution measured at three different areas. Figure 4a shows the average thicknesses and standard deviations of the membranes fabricated by evaporation method according the amount of Nafion solution (10, 12, and 15 g). The membranes fabricated by evaporation did not have uniform thicknesses and also showed large thickness variations between samples with the same weights of added Nafion. Therefore, it was not possible to obtain the desired membrane thickness. Herein, because the required Nafion membrane thickness was 200 μm, we manufactured the membrane using a 15 g Nafion solution having the sufficient thickness to which the pressing process can be applied. Figure 3a-c shows SEM cross-sectional images of the evaporation-fabricated Nafion membrane, and Figure 3d-f shows those of the HPTS-fabricated membrane. It is clear from the images that the thickness of the evaporation-fabricated membrane was nonuniform, whereas the HPTS-fabricated membrane exhibited a uniform thickness. Figure 3g-i shows SEM cross-sectional images of the HPTS Nafion membrane fabricated with MWCNTs at 1.0 wt%, which also had a uniform thickness. Figure 4b,c shows the thicknesses of the Nafion membranes fabricated using a 15 g Nafion solution and N-MWCNT 1.0 wt% by both evaporation and HPTS methods. The thickness deviations derived from nine experiments of the evaporation-fabricated Nafion and N-MWCNT 1.0 wt% membranes were 18.2 and 87 times larger, respectively, than those of the corresponding HPTSfabricated membranes. Despite the addition of MWCNTs, the casted HPTS Nafion membranes had uniform thicknesses of 200 μm. These results show that our HPTS method is more suitable than the evaporation method for producing uniform Nafion membranes and is extensively applicable to various other efforts, such as controlling the thickness of polymers.   Figure 4a shows the average thicknesses and standard deviations of the membranes fabricated by evaporation method according the amount of Nafion solution (10, 12, and 15 g). The membranes fabricated by evaporation did not have uniform thicknesses and also showed large thickness variations between samples with the same weights of added Nafion. Therefore, it was not possible to obtain the desired membrane thickness. Herein, because the required Nafion membrane thickness was 200 µm, we manufactured the membrane using a 15 g Nafion solution having the sufficient thickness to which the pressing process can be applied. Figure 3a-c shows SEM cross-sectional images of the evaporation-fabricated Nafion membrane, and Figure 3d-f shows those of the HPTS-fabricated membrane. It is clear from the images that the thickness of the evaporation-fabricated membrane was nonuniform, whereas the HPTS-fabricated membrane exhibited a uniform thickness. Figure 3g-i shows SEM cross-sectional images of the HPTS Nafion membrane fabricated with MWCNTs at 1.0 wt%, which also had a uniform thickness. Figure 4b,c shows the thicknesses of the Nafion membranes fabricated using a 15 g Nafion solution and N-MWCNT 1.0 wt% by both evaporation and HPTS methods. The thickness deviations derived from nine experiments of the evaporation-fabricated Nafion and N-MWCNT 1.0 wt% membranes were 18.2 and 87 times larger, respectively, than those of the corresponding HPTS-fabricated membranes. Despite the addition of MWCNTs, the casted HPTS Nafion membranes had uniform thicknesses of 200 µm. These results show that our HPTS method is more suitable than the evaporation method for producing uniform Nafion membranes and is extensively applicable to various other efforts, such as controlling the thickness of polymers.          (Figure 5f), and thus we fabricated N-MWCNT membranes with different MWCNT weight ratios from 0.0 to 1.0 wt% and compared their properties. The SEM images show that MWCNT and Nafion were evenly mixed in general, but a greater MWCNT weight ratio led to a partially aggregated surface (the white circles in the figure). The Nafion membrane made with MWCNT at 0.2 wt% (N-MWCNT 0.2 wt%) had the smoothest surface and showed no aggregation. This result implies that 0.2 wt% MWCNTs is the most suitable amount for achieving a good dispersion in the Nafion solution.
Polymers 2020, 12, 396 6 of 10 properties. The SEM images show that MWCNT and Nafion were evenly mixed in general, but a greater MWCNT weight ratio led to a partially aggregated surface (the white circles in the figure). The Nafion membrane made with MWCNT at 0.2 wt% (N-MWCNT 0.2 wt%) had the smoothest surface and showed no aggregation. This result implies that 0.2 wt% MWCNTs is the most suitable amount for achieving a good dispersion in the Nafion solution. The tensile strengths of the Nafion and N-MWCNT membranes (Figure 6a) were also analyzed. The tensile strength of the N-MWCNT 0.2 wt% membrane was 22.3 MPa, which was 1.19 times higher than that of the Nafion membrane without MWCNTs. The N-MWCNT 0.2 wt% membrane also had the highest tensile strength of all the N-MWCNT membranes, owing to its nonaggregated surface.
The obtained TGA results indicate the thermal stability of the Nafion and N-MWCNT membranes (Figure 6b). Weight loss occurred in all the Nafion membranes between approximately 370 and 450 °C. The initial weight loss above 300 °C could be due to the decomposition of sulfonic acid groups, and the remaining majority of the weight loss is due to decomposition of carbon-fluorine bonds [41]. The N-MWCNT 0.2 wt% membrane had a higher thermal stability than both the Nafion membrane without MWCNTs and the other N-MWCNT membranes (inset of Figure 6b). These results show that the thermal stability of the N-MWCNT 0.2 wt% membrane was improved because the MWCNTs were well dispersed in the Nafion membrane.  properties. The SEM images show that MWCNT and Nafion were evenly mixed in general, but a greater MWCNT weight ratio led to a partially aggregated surface (the white circles in the figure). The Nafion membrane made with MWCNT at 0.2 wt% (N-MWCNT 0.2 wt%) had the smoothest surface and showed no aggregation. This result implies that 0.2 wt% MWCNTs is the most suitable amount for achieving a good dispersion in the Nafion solution. The tensile strengths of the Nafion and N-MWCNT membranes (Figure 6a) were also analyzed. The tensile strength of the N-MWCNT 0.2 wt% membrane was 22.3 MPa, which was 1.19 times higher than that of the Nafion membrane without MWCNTs. The N-MWCNT 0.2 wt% membrane also had the highest tensile strength of all the N-MWCNT membranes, owing to its nonaggregated surface.
The obtained TGA results indicate the thermal stability of the Nafion and N-MWCNT membranes (Figure 6b). Weight loss occurred in all the Nafion membranes between approximately 370 and 450 °C. The initial weight loss above 300 °C could be due to the decomposition of sulfonic acid groups, and the remaining majority of the weight loss is due to decomposition of carbon-fluorine bonds [41]. The N-MWCNT 0.2 wt% membrane had a higher thermal stability than both the Nafion membrane without MWCNTs and the other N-MWCNT membranes (inset of Figure 6b). These results show that the thermal stability of the N-MWCNT 0.2 wt% membrane was improved because the MWCNTs were well dispersed in the Nafion membrane.  The obtained TGA results indicate the thermal stability of the Nafion and N-MWCNT membranes (Figure 6b). Weight loss occurred in all the Nafion membranes between approximately 370 and 450 • C. The initial weight loss above 300 • C could be due to the decomposition of sulfonic acid groups, and the remaining majority of the weight loss is due to decomposition of carbon-fluorine bonds [41]. The N-MWCNT 0.2 wt% membrane had a higher thermal stability than both the Nafion membrane without MWCNTs and the other N-MWCNT membranes (inset of Figure 6b). These results show that the thermal stability of the N-MWCNT 0.2 wt% membrane was improved because the MWCNTs were well dispersed in the Nafion membrane.
The GO-Ag NWs paper electrode that we used had an electrical conductivity (σ) of about 9615 S/cm, a sheet resistance of about 250 mΩ/sq., and a thickness of about 14 µm. The actuator was fabricated with the N-MWCNT membrane. Figure 7 shows the actuation performances of the Nafion-based actuator without MWCNTs and the N-MWCNT-based actuators. The N-MWCNT 0.2 wt%-based actuator featured a displacement of 0.845 mm, which was 2.34 times larger than that of the Nafion-based actuator without MWCNTs (Figure 7a). We also observed that the maximum displacement decreased as the MWCNT weight ratio increased. Figure 7b shows the actuation performances of the N-MWCNT 0.2 wt%-based actuator under various input voltages (1 and 2 V AC ) at 0.2 Hz. The displacement of the N-MWCNT 0.2 wt%-based actuator decreased but still stably actuated under 1 V AC . Figure 7c shows the bending curvatures and blocking forces of the casted Nafion-based actuator and N-MWCNT-based actuators with different MWCNT weight ratios under 2 V DC . The bending curvatures were calculated using Equation (1): where R, δ, and l are the radius of curvature, tip displacement, and actuator free length, respectively. The free length in the actuation measurement is the length from the position of the electrode in contact for voltage application to the laser point measuring the displacement with actuator surface standard. The N-MWCNT 0.2 wt%-based actuator had the highest blocking force of 0.086 mN, which was 1.6 times larger than that of the Nafion-based actuator without MWCNTs (Figure 7c). These results show that a high actuator performance resulted from the addition of MWCNTs, yielding a Nafion membrane with high strength and thermal stability. Herein, we observed that the N-MWCNT 0.2 wt%-based actuator had the best actuation performance compared with the Nafion-based actuator and other N-MWCNT-based actuators.
Polymers 2020, 12, 396 7 of 10 The GO-Ag NWs paper electrode that we used had an electrical conductivity (σ) of about 9615 S/cm, a sheet resistance of about 250 mΩ/sq., and a thickness of about 14 μm. The actuator was fabricated with the N-MWCNT membrane. Figure 7 shows the actuation performances of the Nafionbased actuator without MWCNTs and the N-MWCNT-based actuators. The N-MWCNT 0.2 wt%based actuator featured a displacement of 0.845 mm, which was 2.34 times larger than that of the Nafion-based actuator without MWCNTs (Figure 7a). We also observed that the maximum displacement decreased as the MWCNT weight ratio increased. Figure 7b shows the actuation performances of the N-MWCNT 0.2 wt%-based actuator under various input voltages (1 and 2 VAC) at 0.2 Hz. The displacement of the N-MWCNT 0.2 wt%-based actuator decreased but still stably actuated under 1 VAC. Figure 7c shows the bending curvatures and blocking forces of the casted Nafion-based actuator and N-MWCNT-based actuators with different MWCNT weight ratios under 2 VDC. The bending curvatures were calculated using Equation (1): where R, δ, and l are the radius of curvature, tip displacement, and actuator free length, respectively. The free length in the actuation measurement is the length from the position of the electrode in contact for voltage application to the laser point measuring the displacement with actuator surface standard. The N-MWCNT 0.2 wt%-based actuator had the highest blocking force of 0.086 mN, which was 1.6 times larger than that of the Nafion-based actuator without MWCNTs (Figure 7c). These results show that a high actuator performance resulted from the addition of MWCNTs, yielding a Nafion membrane with high strength and thermal stability. Herein, we observed that the N-MWCNT 0.2 wt%-based actuator had the best actuation performance compared with the Nafion-based actuator and other N-MWCNT-based actuators.

Conclusions
In this study, we developed an HPTS method to produce Nafion membranes with consistent and uniform thicknesses, and we fabricated N-MWCNT membranes using different MWCNT weight ratios to increase the actuator blocking force. The deviations in thickness of the HPTS-fabricated Nafion and N-MWCNT membranes were 18.2 and 87 times lower, respectively, than those of the corresponding membranes fabricated by evaporation method. Therefore, the Nafion and N-MWCNT membranes fabricated by the HPTS method had highly uniform thicknesses and surfaces. We determined the suitable MWCNT weight ratio to be approximately 0.2 wt% for application to IEAP actuators. The N-MWCNT 0.2 wt%-based actuator had a tensile strength of approximately 22.3 MPa, which was 1.19 times higher than that of the Nafion-based actuator without MWCNTs and was the largest among the tensile strengths of all the N-MWCNT-based actuators tested. It also had the highest thermal stability among all the actuators, both with and without MWCNT addition. In addition, the blocking force and displacement of the N-MWCNT 0.2 wt%-based actuator were 0.086 mN and 0.0845 mm, which are 1.6 and 2.34 times higher, respectively, than those of the actuator without MWCNTs. As a result, the addition of MWCNTs afforded the actuator with good mechanical and chemical properties, including high strength and thermal stability of the Nafion membrane. However, if the MWCNT weight ratio exceeds the suitable amount, sufficient dispersion is not achieved, which can decrease the membrane mechanical and chemical properties. Finally, the developed HPTS method can effectively be used in research on polymer materials that requires controlled thicknesses, and the N-MWCNT 0.2 wt%-based actuator can be applied to flexible devices, artificial robots, and flexible actuator technologies.