Ultra‐Thin and Conformable Electrodes Composed of Single‐Walled Carbon Nanotube Networks for Skin‐Contact Dielectric Elastomer Actuators

Skin‐contact dielectric elastomer actuators (DEAs) consisting of skin‐conformable, stretchable electrodes are fabricated using a roll‐to‐roll‐based gravure coating method. In this method, single‐walled carbon nanotubes (SWCNTs) are continuously applied on a free‐standing ultra‐thin film (nanosheet) of poly(styrene‐b‐butadiene‐b‐styrene) (SBS) to produce an SWCNT‐SBS nanosheet of 101‐nm thickness. After the first SWCNT coating, the SWCNT‐SBS nanosheet shows a Young's modulus (i.e., 80.9 MPa) comparable to that of the SBS film and a sheet resistance of 4.6 kΩ sq−1. Using the free‐standing SWCNT‐SBS nanosheets as electrodes, a ten‐layered DEA is fabricated without glue or dielectric elastomer precursors on three substrates with different stiffness, namely glass, Ecoflex 00–30, and a urethane elastomer model skin. The low flexural rigidity of the ten‐layered DEA (105 nN m) ensures conformability to the shape of an index finger. Application of an actuation voltage of 2100 V produces a two‐fold larger displacement of the DEA on the Ecoflex 00–30 substrate compared with that on the glass substrate. The ability of the DEA to conform to the surface of skin will enable its application in skin‐contact haptic devices.


Introduction
Dielectric elastomer actuators (DEAs) have attracted much attention as potential actuation systems in wearable devices such as artificial muscles [1][2][3][4][5][6] of motion assisting suits and haptics. [7][8][9][10][11][12] They have high flexibility, low weight, large actuation strain, and high energy density compared with conventional actuators such as electromagnetic motors and hydraulic cylinders. The fundamental structure of DEAs is similar to that of a single-layered capacitor, consisting of a dielectric elastomer (DE) layer sandwiched between two compliant electrodes. This layer contracts in the thickness direction and expands in the inplane direction owing to the attractive electrostatic force (i.e., (PEDOT:PSS) coated on a poly(styrene-b-butadiene-b-styrene) (SBS) thin layer with a total thickness of ≈340 nm (referred to as a "conductive nanosheet" or "nanosheet electrode"). [31] The conductive nanosheet was fabricated using a gravure coating method and used as a skin-conformable electrode to precisely record the surface electromyogram. The gravure coating method is advantageous in the mass fabrication of nanosheet electrodes because the thickness can be simply controlled by changing the concentrations of the constituent polymer solutions and the gravure roll speed ( Figure S1, Supporting Information).
According to Equations (1) and (2), a DE layer with a small Young's modulus and a thin layer thickness results in a large deformation of the DEA. Furthermore, reducing the thickness of free-standing nanosheet electrodes by two orders of magnitude from that of conventional electrodes (several hundred nanometers) would allow to use lower Young's modulus of DE layers than that of conventional DE materials, such as polydimethylsiloxane (PDMS) [32] and acrylic elastomer [33] (≈1-100 MPa), by the same order of magnitude. Therefore, multi-layered DEAs composed of DE layers with a small Young's modulus (i.e., 1 MPa) alternately stacked with nanosheet electrodes should show a large displacement and excellent flexibility. Previously, we reported the low-voltage oscillations of DEAs; a DE layer with a thickness of 600 nm consisting of PDMS and a pair of nanosheet electrodes with a thickness of 200 nm consisting of PEDOT:PSS and SBS. [34] While the DEAs showed oscillatory response with sine wave signals of 50 V in the range of 1-30 kHz, the deformation of the DEAs in the thickness direction would be small due to the rigid PEDOT:PSS layer with a Young's modulus of several GPa (e.g., 2.4 GPa [35] ) on the SBS nanosheet. On the other hand, the presently reported singlewalled carbon nanotubes (SWCNTs) layer does not impair the elasticity of the SBS layer, resulting in the improvement of the deformation in the thickness direction.
In this study, we used the gravure coating method to fabricate a conductive nanosheet composed of SWCNTs coated on an SBS film (SWCNT-SBS nanosheet) and experimentally clarified the correlation between the apparent Young's modulus and thickness of the SWCNT layer by varying the number of SWCNT coatings. Then, SWCNT-SBS nanosheets with a Young's modulus close to that of the SBS nanosheet were directly attached to Ecoflex 00-30 sheets as DE layers with a small Young's modulus to construct a multi-layered DEA. Furthermore, we examined the mechanical effect of substrates such as glass, Ecoflex 00-30, and an artificial skin model (urethane elastomer) on the contractile displacement and strain of the DEA. The present study will advance the structural and mechanical design of thin-film DEAs for wearable device applications.

Fabrication of SBS and SWCNT-SBS Nanosheets and Ten-Layered DEAs
A polyethylene terephthalate (PET) substrate was coated with a thin sacrificial layer of polyvinyl alcohol (PVA), followed by the application of SBS elastomer solutions prepared in tetrahydrofuran. Then, an aqueous dispersion of SWCNTs was repeatedly coated using gravure coater to achieve a variable number of SWCNT coatings ( Figure S1 and Table S1, Supporting Information). The free-standing nanosheets were obtained by rinsing the PVA layer out with deionized water after peeling them from the PET substrate with the aid of an adhesive tape frame, as we previously reported. [36] Photographs of the free-standing SBS nanosheet and SWCNT-SBS nanosheet are shown in Figure 1a,b, respectively.
In comparison with electrodes formed by patterning solvated conductive materials on a DE layer, the process of alternately stacking electrodes and DE layers is a simple way to fabricate multi-layered DEAs because the wettability of the solvated material on the DE need not to be considered. In this study, an SBS film with a single SWCNT coating was used to fabricate the electrodes of a ten-layered DEA. An SBS nanosheet (81 nm thickness) was prepared in advance (Table S1, Supporting Information) and covered with a SWCNT layer (20 nm thickness), yielding a sheet resistance of 3.9 kΩ sq −1 . The resulting SWCNT-SBS nanosheet (101 nm thickness) supported with a tape frame was directly attached to a sheet of Ecoflex 00-30 (95 µm thickness) without glue, followed by attaching the Ecoflex 00-30 sheet to the electrode. The stacking process was repeated to fabricate a ten-layered DEA (Figure 1c-e). The surface energy of human skin makes free-standing nanosheets inherently adhesive, therefore no glue is required to prevent positional shifts in response to external forces. Indeed, the ten-layered DEA showed a good affinity for the surface of an index finger (Figure 1f). The flexural rigidity, D, is given by where Y is Young's modulus, d is the thickness, and ν is Poisson's ratio. [37] Young's modulus of Ecoflex 00-30 was estimated from the results of a tensile test ( Figure S2, Supporting Information). Using this value, and assuming ν ≈ 0.5 for elastomeric materials, the flexural rigidity of the ten-layered DEA is estimated to be 105 nN m. Height images (Figure 2a,b) obtained using atomic force microscopy (AFM) showed that the SBS nanosheet had a homogeneous surface with a roughness of 0.6 nm, a fibril network of SWCNT bundles was formed on the SBS layer (Figure 2b), and the roughness of the SWCNT-SBS nanosheet was 2.5 nm. The gradations of brightness of AFM phase images indicate that the SBS structure was phase-separated into polystyrene and polybutadiene microdomains (Figure 2c). The phase image of the SWCNT-SBS nanosheet shows bright domains dispersed among dark domains, reflecting the dispersion of hard regions (presumed to contain SWCNT bundles) among soft regions ( Figure 2d).

Effect of SWCNT Layer Thickness on Resistance of SWCNT-SBS Nanosheets
AFM height profiles of the SWCNT-SBS nanosheets fabricated with different numbers of SWCNT coatings are shown in Figure 3a. The fibril structure was only partially visible after 1st coating, whereas the fibers were notably thicker and more numerous after 3rd coating. After 5th or more coatings, the boundary of the fibril structure was broadened and a continuous surface with large recesses was formed. An SBS-like phase-separated structure was observed after 1st coating, and the aggregates of SWCNT bundles observed after 3rd or more coatings ( Figure S4, Supporting Information; high-intensity areas) were assumed to result from the intermolecular attractive forces between the applied SWCNT dispersion and the previous layer.
The UV-vis spectra of the SWCNT-SBS nanosheets on a quartz substrate showed absorbance from 380 to 780 nm that increased with the number of SWCNT coatings on the SBS layer, whose relationship was approximately in direct proportion ( Figure 3b). This result was visually confirmed by the progressively darker color of the nanosheet ( Figure S3, Supporting Information). The SBS nanosheet absorbed light strongly at wavelengths below 350 nm, which can be attributed to the π-π* transition of styrene. [38] In addition, absorbance at 275 nm was observed for SWCNT-SBS nanosheets and increased in proportion to the number of coatings (Figure 3b, inset). An absorption peak between 270 and 278 nm is a signature of the π-plasmon excitation of the SWCNT surface. [39][40][41] The increase in the absorbance at 275 nm therefore suggests a proportionate increase in the thickness of the SWCNT layer, which should result in a reduced sheet resistance of the SWCNT-SBS nanosheets. In addition, it was assumed that the density of SWCNT aggregates increased but not the thickness with increment in the number of coatings from 5th to 7th from the results that i) the SWCNT layer thickness of 7th approximately equals to that of 5th ( Figure 3c) and ii) the absorbance of the nanosheets at the wavelength of 275 nm proportionally increased with increasing the number of coatings (Figure 3b, inset). Therefore, progressing an aggregation of SWCNT resulted in broadening the boundary in the height images of 5th and 7th.
As shown in Figure 3c-e, the thickness and average profile roughness (Ra) of the SWCNT layer increased from 26 nm and 2.4 nm, respectively, after 1st coating to 206 and 5.9 nm, respectively, after 7th coating. The sheet resistance, measured using a four-point probe method with an LCR meter, decreased from 4.6 kΩ sq −1 after 1st coating to 0.4 kΩ sq −1 after 7th coating, consistent with the increase in the thickness of SWCNT layer.

Tensile Testing
The tensile test results of the SBS and SWCNT-SBS nanosheets are shown in Figure 4a-c. The values of Young's modulus, yield stress, tensile strength, and elongation at break of the nanosheets were estimated from the stress-strain curves (Figure 4a,b, Table S2, Supporting Information). Young's modulus, calculated from the slope of the elastic region of the stress-strain curves before the yield points, increased in proportion to the SWCNT layer thickness from 70.5 MPa at 0 nm to 181.7 MPa at 206 nm. The yield stress showed the same tendency, while the sheet resistance, which was 4.6 kΩ sq −1 at 26 nm (1st), decreased commensurately with the SWCNT layer thickness (Figure 4c). Carbon nanotubes are generally chosen as a filler material to reinforce the composite structure because they have a large Young's modulus (≈TPa) and form bundle structures mediated by van der Waals forces in solutions and on substrates. [42,43] AFM images (Figure 3a; Figure S4, Supporting Information) show that three or more coatings can increase the size of the aggregates and continuous network of SWCNTs (which will increase the number of electrical conduction pathways) to produce a large Young's modulus and low sheet resistance. Notably, Young's modulus after 1st coating was 80.9 MPa, which is close to the corresponding value for SBS (70.5 MPa). Furthermore, we successfully supplied power to a blue LED by applying 6 V DC across wires connected to the 1st coating of the SWCNT-SBS nanosheet in the bent state of the index finger (Figure 4d).
In comparison to SWCNTs, using PEDOT:PSS as a conducting polymer (Young's modulus of 2.4 GPa [35] ) would impair the elasticity of the SBS nanosheet. We previously reported that Young's modulus of an ultra-thin strain sensor composed of PEDOT:PSS printed on an SBS nanosheet was 227 ± 32 MPa, which is ≈3 times larger than that measured for SBS. [44] In addition, stretching up to a strain of 50% was reported to generate cracks in the PEDOT:PSS/Zonyl layer coated on a PDMS substrate, resulting in a 42-fold increase in electrical resistance at a strain of 188%. [45] On the other hand, an SWCNT thin layer prepared on a PDMS substrate showed no cracks during stretching of the substrate up to a strain of ≈170%, with the SWCNT bundles able to retain the carbon nanotube network because of sliding and buckling of nanotubes during stretching and releasing. [46] At this maximum strain, a fivefold increase in resistance was also measured, much less than that observed for the PEDOT:PSS system. We suggest that the ultra-thin layer of SWCNTs, which can form a nanotube network but not a continuous film (Figure 3a), plays an important role in the conductivity and flexibility of a single-layer SWCNT-SBS nanosheet. The small Young's modulus of the SWCNT-SBS nanosheet, therefore, makes it a promising electrode to realize large deformations in multi-layered DEAs.

Ten-Layered DEAs Using SWCNT-SBS Nanosheets as Electrodes
The SWCNT-SBS nanosheet with an SWCNT thickness of 20 nm had a sheet resistance of 3.9 kΩ sq −1 , which is relatively large compared with that of metal materials. An important consideration is whether this resistance makes the SWCNT-SBS nanosheet suitable for use as a DEA electrode. Ji et al. described the resistance of the SWCNT thin electrode layer in terms of the electrical cut-off frequency (f c ) of a DEA [30] where R is the resistance of the electrodes, C is the capacitance of the DEA, d is the thickness of the dielectric elastomer layer, ε 0 is the electric constant, ε r is the relative permittivity of the dielectric elastomer, and A is the active area of the DEA (i.e., the overlapped area of the electrodes). They used a circular capacitor composed of a PDMS DE layer with a diameter of 3 mm, a thickness of 6 µm, and an estimated capacitance of 30 pF to increase the output force of the DEA at a driving voltage less than 500 V. According to the Equation (4), the electrode resistance should be less than 5 MΩ to achieve a driving frequency greater than 1 kHz. Therefore, Ji et al. fabricated the SWCNT electrode with 0.2 MΩ sq −1 using the Langmuir-Schaefer method. [30] In contrast, we calculated that a capacitance of 7.5 pF for a 5 mm × 5 mm × 95 µm Ecoflex 00-30 DE layer (ε r = 3.2, measured using an LCR meter) and f c = 24.8 MHz for an electrode resistance of 0.9 kΩ (i.e., 3.9 kΩ sq −1 ). Therefore, the SWCNT-SBS nanosheet shows a sufficiently low sheet resistance to function as a DEA electrode. Then, we focused on the static displacement of the fabricated DEAs in various electric fields as a function of electrode thickness and substrate rigidity. We prepared three types of SWCNT-SBS nanosheets, each composed of an initial SWCNT coating (1st) on SBS layers of various thicknesses, as electrodes of ten-layered DEAs with a 95 µm DE layer thickness. The displacement and contractile strain measured during repeated 20 s on-off cycles of the actuation voltage are shown for different SBS layer thicknesses in Figure 5 and Figure S5, Supporting Information. Figure 5 shows the contractile strain at 2000 V was 0.2%, 0.6%, and 1.9% for layer thicknesses of 10.5 µm, 566 nm, and 94 nm, respectively. The DE actuation of 10.5 µm showed similar trend to that of 566 nm against the applied voltage. Interestingly, the actuating performance of 94 nm was superior to that of 10.5 µm and 566 nm. This suggests the rigidity of the electrode (i.e., Y e t e ) dramatically decreases with the thickness of SBS layer, resulting in a larger deformation of the DEA in an electric field.
To further investigate the influence of substrate rigidity on the DEA performance, contraction measurements were performed using a laser displacement meter. Photographs of the fabricated ten-layered DEA on the glass, Ecoflex 00-30, and urethane elastomer model skin substrates are shown in Figure 6a and their time-dependent responses to switching of the actuation voltage are shown in Figure 6b and Figure S6, Supporting Information. The dependence of the displacement and contractile strain on the applied voltage are shown in Figure 6c,d, respectively. Figure 6b shows that stable DEA displacement responses were observed after switching the voltage two or more times. The displacement and contractile strain of the DEAs increased commensurately with the applied voltage. Notably, the displacement and the contractile strain of the DEA on glass at 2100 V were 19 µm and 1.8%, respectively, while the corresponding values on Ecoflex 00-30 were 47 µm and 4.9%, more than twofold larger. The glass substrate may constrain the in-plane stretching of the layer in the electric field at the DEA-glass interface, while the Ecoflex 00-30 substrate does not affect the stretching of each layer near the substrate because the DE layers are made from the same material. The ten-layered DEA was also fabricated on the urethane elastomer, which has mechanical properties similar to human skin. On this substrate, the displacement and contractile strain at 2100 V were as large as 24 µm and 2.2%, respectively, slightly larger than the corresponding values obtained on glass. Given the different response of the DEA on each substrate to the voltage switching, we further measured the DEA displacement in response to an actuation voltage of Figure 5. Voltage dependence of a) displacement and b) contractile strain of the ten-layered DEA using SWCNT-SBS nanosheets with different SBS layer thicknesses as electrodes. Data were measured using a laser displacement meter.  2100 V applied for 1 min ( Figure S7, Supporting Information). The Ecoflex 00-30 substrate resulted in the largest maximum displacement (58 µm), followed by the model skin (i.e., 29 µm) and glass (22 µm). The response time of the DEA, calculated as the difference between the times at 30% and 70% of the maximum DEA displacement, was 0.23, 0.34, and 1.81 s on glass, Ecoflex 00-30, and model skin substrates, respectively. Despite the long response time, the DEA on the model skin showed a larger displacement than the DEA on glass, suggesting that the flexible urethane elastomer substrate can improve the static performance of multi-layered DEAs. The frequency dependence of DEA displacement should be investigated in future work because a dynamic property of DEAs, which inertial and dissipative effects would limit, is not expected to coincide with an electrical cut-off frequency.

Conclusion
We fabricated conductive nanosheets composed of SWCNT and SBS using the gravure coating method. Using SWCNT-SBS nanosheets with a thickness of 101 nm (20 nm SWCNT layer) as the compliant electrodes and Ecoflex 00-30 sheets with a thickness of 95 µm as the DE layers, ten-layered DEAs were fabricated by direct attachment of each layer. Resistance measurements and tensile testing showed that the DEA electrode, formed after the first SWCNT coating, showed a low sheet resistance (i.e., 3.9 kΩ sq −1 ) and a Young's modulus (i.e., 80.9 MPa) similar to that of the SBS nanosheet (i.e., 70.5 MPa). This result suggested that the SWCNT layer did not impair the elasticity of the SBS nanosheet. Previously, we reported that an ultrathin strain sensor composed of PEDOT:PSS printed on the SBS nanosheet showed high Young's modulus (i.e., 227 ± 32MPa). [44] Therefore, the SWCNT-based conductive nanosheet in this report have the potential for highly flexible and stretchable electrode materials. Notably, the flexible Ecoflex 00-30 and urethane elastomer model skin substrates yielded a twofold larger DEA displacement and contractile strain compared with the corresponding values on a glass substrate. The ten-layered DEA was attached to the model skin substrate and the skin surface of an index finger. These results demonstrate the potential to integrate low-voltage, highly deformable DEAs with skinattachable devices for rehabilitation and haptics without prestretching of DE layers.

Experimental Section
Fabrication and Characterization of SWCNT-SBS Nanosheets: PVA and SBS powders and the aqueous SWCNT dispersion (1.0 mg mL −1 ) were purchased from Sigma-Aldrich. A 5 wt% aqueous PVA solution was coated on a PET sheet roll using a roll-to-roll gravure coating system (Mini-Labo, Yasui Seiki Company, Ltd.) under the conditions listed in Table S1, Supporting Information. A 1 wt% solution of SBS in tetrahydrofuran was applied to the PVA-coated PET and the resulting PET/PVA/SBS sheet was dried at 80 °C using heaters in the coating system. The SWCNT dispersion was applied to the PET/PVA/SBS sheet and dried at 80 °C. The SWCNT-coated PET/PVA/SBS sheet with different numbers of SWCNT coating layers was fabricated by repeated application of the SWCNT dispersion. The resulting SWCNT-SBS nanosheets were peeled off using adhesive tape frames and attached to an Si substrate after washing the PVA layer away with deionized water. The thickness and average surface roughness of the SWCNT-SBS nanosheets were determined using AFM (Innova, Bruker). The sheet resistance of the SWCNT-SBS nanosheet was measured using a fourpoint probe method (1 mm probe distance) and an LCR meter (IM 3533-01, HIOKI E.E. CORPORATION). Tensile testing was carried out using a tensile tester (EZ Test EZ-SX, SHIMADZU Co., Ltd.) equipped with a 5 N load cell. The nanosheets were cut to a size of 40 mm × 20 mm using a tape frame and then stretched at a strain rate of 10 mm min −1 at 25 °C. A strain value was determined where an S-S curve should be steepest in the elastic deformation region (i.e., bellow the yield stress), and then a Young's modulus was estimated from the slope of a line fitted to the curve within ±0.25% of the determined strain. UV-vis spectra of nanosheets attached to a quartz substrate were acquired at 23.9 °C and 56% relative humidity using a spectrophotometer (DU730, Beckman Coulter). The absorbance of the quartz substrate was subtracted to yield the spectrum of the nanosheet.
Fabrication and Displacement Measurement of the Ten-Layered DEAs: Equal weights of Ecoflex 00-30 liquids A and B (Smooth-On, Inc.) were mixed, stirred, and defoamed using a mixer (AR-100, THINKY CORPORATION). The obtained Ecoflex 00-30 precursor solution was spin-coated onto a polystyrene substrate at 1000 rpm for 20 s (Opticoat, MS-B150, MIKASA) and heated on a hot plate at 70 °C for more than 1 h to obtain cured Ecoflex 00-30 sheets. After washing the PVA layer with water, the rectangular SWCNT-SBS nanosheets (5 mm × 25 mm) were directly attached to the substrate or Ecoflex 00-30 DE layer. By alternately laminating SWCNT-SBS nanosheets and Ecoflex dielectric layers, the ten-layered DEAs were fabricated on glass, Ecoflex 00-30, and a urethane elastomer model skin (BIOSKIN Plate P002-001#10, Beaulax Co., Ltd), respectively.
Anodes and cathodes of the ten-layered DEA were individually connected to two sheets of copper tapes that acted as collecting electrodes on a substrate, ensuring no contact between the anode and cathode collecting copper tapes. The anode and cathode collecting tapes were connected to the anode and cathode terminals, respectively, of a high voltage power supply (M10-HV5000A, MCP Japan), and the applied voltage was modulated between 500 and 2100 V. The displacement in the film thickness direction was measured at each voltage as a position profile using a laser displacement meter (ZX-LD40, ZX-SF11, ZX-LDA11-N, OMRON Corporation).

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.