Main text

Recent studies on wearable bioelectrodes with real-time graphical displays that enable routine and long-term recording of bioelectrical signals (i.e., electromyography (EMG), electrocardiography, and electrocorticography) have attracted considerable attention in the fields of sports and health1,2,3,4,5,6. To achieve long-term monitoring, a skin contact electrode must have a low elasticity and allow for the transmission of perspiration-induced water vapor during exercise.

The human epidermis comprises sweat gland holes, body hairs, sulci cutis, and cristae cutis (i.e., an uneven surface); in addition, the epidermis stretches to a degree of ~75%7,8,9, and it has a Young’s modulus on the order of several hundred kPa10. Commercially available bioelectrodes are generally not stretchable or permeable to humidity because of their thick construction and hard components. Therefore, these materials are not suitable for long-term application. Self-supporting electrodes that are stretchable, permeable to humidity, and conformable to skin surface bumps are required to allow for the natural deformation of skin without restricting body movements. Moreover, conventional stretchable electrodes for monitoring bioelectrical signals are often composed of metals11,12, conductive polymers13,14,15, or hydrogels16,17,18. Metals have large Young’s moduli (several hundred GPa) and low elongations at break (e.g., Au: ~30%). Thus, patterning is necessary to deposit a thin electrode layer with a structure that can minimize stress concentration (e.g., serpentine structure) on a stretchable substrate and provide a degree of stretchability19,20. Conductive polymers, such as poly(3,4-ethylenedioxithiophene) doped with poly(4-styrenesulfonate acid) (i.e., PEDOT:PSS), have sufficient electrical conductivities (several thousand S cm−1) and low elongations at break (~30%). Although hydrogel composites exhibit lower Young’s moduli and electrode–skin impedances than Au films, increasing the signal-to-noise ratios (SNRs) of the EMG signals16 and the evaporation of water vapor from the hydrogel limits the lifetimes of the electrodes. In addition, conventional electrodes tend to be composed of thick membranes (on the order of μm and mm), which have low humidity permeabilities and induce inflammation in the skin.

To balance stretchability, humidity permeability, and skin conformability, mesh-like structures composed of nanofibers have been proposed for the conductive layers of electrodes21,22,23,24. Vacuum deposition of Au has been used to impart electrical conductivity to the mesh. However, stretching readily induces cracks in the Au layer, greatly increasing the resistance. The application of fibrous network structures composed of conductive nanomaterials, such as carbon nanotubes (CNTs) and metal nanowires (NWs), is essential for increasing strain tolerance25,26,27,28,29,30,31. An ultrathin layer of single-walled carbon nanotubes (SWCNTs) shows no cracks during stretching to a strain of ~170% on a thick elastomer sheet (thickness: 0.2–0.5 mm), and the resistance increases with increasing strain, but it only increases by 1.5 times at 50% strain29. A computational simulation has supported these results and revealed that the relative resistance change with strain exhibits less dependency when the CNT density exceeds the percolation threshold32. Layered textile electrodes employing Ag-NWs with thicknesses of ~15 μm exhibit changes in resistance of 1.5% after 4200 bending cycles22, demonstrating high compatibility between conductors comprising fibrous network structures and highly stretchable structures or electrodes. However, these nanomaterials are not self-supporting as ultrathin structures (i.e., on the order of nanometers), and relatively thick substrates have been used. Moreover, reducing the flexural rigidity, which is proportional to the cube of the electrode thickness, is important for achieving effective skin conformability. Therefore, nanometer thicknesses are considered beneficial for skin-conformable electrodes, especially if chemical adhesives can be avoided.

Polymer ultrathin films (“nanosheets”) have thicknesses of tens to hundreds of nanometers and areas of several square centimeters (i.e., aspect ratios greater than or equal to 106)33,34. We previously reported that free-standing ultrathin films (i.e., nanosheets) consisting of polysaccharides exhibit incremental adhesiveness in terms of the critical load during deposition onto SiO2 substrates for films with thicknesses below 200 nm, according to microscratching tests35. In addition, we have fabricated a conductive nanosheet composed of poly(styrene-b-butadiene-b-styrene) (SBS) coated with a thin layer of PEDOT:PSS (total thickness: 340 nm) using a role-to-role gravure coating method36. Because of their ultraconformability, conductive nanosheets enable surface EMG (sEMG) measurements of palm muscles during baseball pitching motions36 and have been applied as flexible strain sensors37. However, there are issues regarding the application of PEDOT:PSS in long-term biosignal measurements due to their swelling caused by sweat, leading to unexpected detachment from the skin, and low fracture strains (e.g., ~5%37). Thus, we have focused on coating SBS nanosheets with SWCNTs, which naturally form network structures and have high stretchabilities29,38.

In this study, we fabricated conductive nanosheets by combining fibrous networks of SWCNT bundles with SBS elastomer nanosheets (i.e., SWCNT-SBS nanosheets) using SWCNT layers of different thicknesses and densities. Nanosheets were fabricated by coating SBS nanosheets with PEDOT:PSS containing 5 wt.% 1,4-butanediol (BG) (i.e., PH1000/BG5-SBS nanosheets) in a similar manner to previous reports for comparison. The electrical and mechanical properties of the conductive nanosheets were investigated, and we examined the correlation between the SNR and the electrochemical impedance at the skin/electrode interface measured on a forearm by comparing the results obtained for the SWCNT-SBS nanosheets, PH1000/BG5-SBS nanosheets, and Ag/AgCl gel electrodes.

Scanning electron microscopy (SEM) images of the nanosheets and the effects of thickness and fibrous density on the sheet resistance values of the SWCNT-SBS nanosheets

By repeatedly applying SWCNT aqueous dispersion to the SBS nanosheets, we obtained conductive nanosheets composed of multiple layers. Figure 1 shows SEM images of the SBS, SWCNT-SBS, and PH1000/BG5-SBS nanosheet surfaces. The SBS nanosheets exhibited a continuous surface, whereas fibrous networks consisting of SWCNT bundles were observed on the SWCNT-SBS nanosheets (Fig. 1a–d). The nanosheet with one SWCNT coating, namely, SWCNT 1st-SBS, contained voids in the SWCNT fiber network. The SWCNT 3rd-SBS and SWCNT 5th-SBS nanosheets showed denser accumulation of fibers than SWCNT 1st-SBS nanosheets. In contrast, the PH1000/BG5-SBS nanosheet showed a continuous film surface formed by the aggregation of colloidal PEDOT:PSS particles39 (Fig. 1e, f).

Fig. 1: Scanning electron microscopy (SEM) images of the nanosheets.
figure 1

SEM images of the a SBS nanosheets, b SWCNT 1st-SBS nanosheets, c SWCNT 3rd-SBS nanosheets, d SWCNT 5th-SBS nanosheets, and e PH1000/BG5-SBS nanosheets. The images were taken at ×40,000 magnification, except for (d) (×45,000). f Magnified image of a PH1000/BG5-SBS nanosheet.

From the UV‒Vis spectra in the wavelength range of 260–1000 nm (Fig. 2a), the absorbance characteristics of the SWCNT-SBS nanosheets increased as the number of coatings increased from one to five, even though all the SBS layers had similar thicknesses (356 nm). This result implied that the increase in the number of coatings increased the densities and thicknesses of the SWCNT bundles, which was consistent with the SEM images. The thicknesses of the pristine SBS and PH1000/BG5-SBS nanosheets were 356 and 503 nm, respectively (Fig. 2b). The thicknesses of the SWCNT-SBS nanosheets increased from 400 nm for SWCNT 1st-SBS to 430 nm for SWCNT 3rd-SBS. This value was similar to the initial value for SWCNT 5th-SBS (431 nm). This finding suggested that the SWCNT fibers accumulated to increase the densities of the fibers after the 3rd coating, not to increase the thickness of the SWCNT layer on the SBS nanosheet. Consequently, the sheet resistance of the SWCNT-SBS nanosheets decreased from 3.039 kΩ sq−1 for the 1st coating to 0.296 kΩ sq−1 for the 5th coating (Fig. 2c). The PH1000/BG5-SBS nanosheet showed a sheet resistance of 0.527 kΩ sq−1, which was comparable to that of the SWCNT 3rd-SBS nanosheet (0.579 kΩ sq−1).

Fig. 2: Effects of the densities and thicknesses of conductive nanosheets on the sheet resistance.
figure 2

a UV‒Vis absorbance of the SBS and SWCNT-SBS nanosheets deposited on quartz substrates, b thicknesses of the nanosheets measured using a surface profiler, and c sheet resistance values of the conductive nanosheets measured using a four-probe method.

Mechanical properties of the nanosheets

We performed tensile tests on the fabricated nanosheets to investigate the effects of the SWCNT densities and thicknesses on the mechanical properties of the nanosheets. Figure 3a, b shows the S‒S curves of the nanosheets. We observed explicit yield stresses in the S‒S curves of the SWCNT-SBS nanosheets, which was not found in the curves of the pristine SBS nanosheets. The regions of plastic deformation for the SBS, SWCNT 1st-SBS, and SWCNT 3rd-SBS nanosheets were ~30–200% of the strain. These results suggested that the fibrous networks of the SWCNTs were deformed during stretching, and the SWCNT fibers were pulled apart, buckled, and slid past each other, changing the associated areas between the SWCNT fibers. The SWCNT 5th-SBS showed plastic deformation, where the strain range was ~30–100% and the nanosheet broke at 107% strain, as shown in Fig. 3a. Increasing the number of SWCNT coatings decreased the elongation at break (Fig. 3c). The drastic decrease in elongation between SWCNT 3rd-SBS and SWCNT 5th-SBS was potentially caused by increases in the densities of the SWCNT fibers. Figure 3d shows the dependences of the elastic moduli calculated from the slopes of the S‒S curves in the elastic regions (e.g., for strains less than 2.0%) on the thicknesses of the conductive layers. The elastic moduli of the SWCNT-SBS nanosheets increased with the number of coatings from 60.8 (1st) to 104.2 MPa (5th) compared with that of the pristine SBS nanosheets (48.5 MPa), which could have originated from the increased thicknesses and densities of the fibers. Conversely, each PH1000/BG5-SBS nanosheet exhibited a low elongation at break (22%) and high elastic modulus (298 MPa) because of the continuous thin layer of PH1000, which originally had a high Young’s modulus (2.4 GPa) and low elongation at break (<10%). The fibrous networks of SWCNTs showed potential in applications for stretchable electronics that were considered conformable to the mechanical properties of human skin.

Fig. 3: Results of the tensile test for the nanosheets.
figure 3

Stress‒strain curves for the different nanosheets in the ranges of a 0–600% and b 0–5% of tensile strain. Dependences of the (c) elongations at break on different thicknesses of the conductive layer and of the (d) elastic moduli of the nanosheets on the conductive layer at different thicknesses. The elastic modulus was estimated from the slope of a line fitted to the curve at a strain, The slope was steepest in the elastic deformation region.

Effects of thickness and fibrous density on the nanosheet humidity permeability

The densities of hydrophobic SWCNT fibers could affect the humidity permeability (i.e., water vapor) when applying SWCNT-SBS nanosheets as bioelectrodes on human skin. The dish method (JIS Z0208 “Testing methods for the determination of the water vapor transmission rates of moisture-proof packaging materials (dish method)”) is a method for measuring the WVTR of a flat membrane. The WVTRs of the filter papers used as porous substrates and the SWCNT-SBS nanosheets deposited on the filter papers are shown in Fig. 4. Since nanosheets have high WVTRs and CaCl2 tends to deliquesce in a high-humidity atmosphere, CaCl2 deliquesced in 3 h in the chamber at 90% RH, decreasing the weight change per hour and the WVTRs after 3 h of retention. Therefore, we calculated the WVTR of the nanosheet/filter paper at 2 h, which was almost the same as that at 1 h. The 210-nm-thick SBS nanosheets showed similar or lower WVTRs (6198 g m−2 (2 h)−1) than did the filter papers (6345 g m−2 (2 h)−1). In addition, the WVTRs of the SBS nanosheets linearly decreased with increasing thickness, from 6198 to 5545 g m−2 (2 h)−1. Furthermore, an increase in the number of SWCNT coatings for a constant SBS layer thickness of 356 nm significantly reduced the WVTRs from 5632 to 4465 g m−2 (2 h)−1. This phenomenon could be derived from the increased thickness of the SWCNT layer and the increased SWCNT density. A previous report showed that the WVTRs of electrospun polyurethane webs decreased with decreasing web pore size40. In our systems, we considered that the increase in fiber density decreased the pore size of the fibrous network, thereby decreasing the WVTRs.

Fig. 4: Permeability of the nanosheets to humidity.
figure 4

WVTRs of the SBS and SWCNT-SBS nanosheets measured using the dish method (JIS Z0208) at 40 °C and 90% RH for 2 h. WVTRs were obtained for nanosheets on filter paper substrates. The thickness represents the total thickness of the nanosheet.

The nanosheets were laminated with filter papers as the porous supporting substrate because the nanosheets could be decomposed by the metal fixture of the dish during paraffin sealing. We compared the WVTRs of nanosheets without filter paper to those of conventional stretchable bioelectrodes (Table 1) by applying the following equation (ISO21760-1):

$${W}_{N}=\left({W}_{{NS}}\times {W}_{S}\right)/\left({W}_{S}-{W}_{{NS}}\right)$$
(1)

where WN, WS, and WNS are the WVTRs of the nanosheet, filter paper substrate, and laminate of the nanosheet on the filter paper substrate, respectively. The WN of the SWCNT 3rd-SBS was 28,316 g m−2 (2 h)−1. This WVTR was approximately two orders of magnitude greater than that of normal skin, as shown in Table 1 (204 ± 12 g m−2 (24 h)−1)41, indicating that the SWCNT 3rd-SBS nanosheet was suitable for use as a skin-conformable electrode, avoiding sweat accumulation at the skin/electrode interface. The WVRTs were the values normalized by the retention time in a chamber (i.e., 2 h and 24 h). Hence, we compared the WVTRs to those of previous papers despite the occurrence of different retention times.

Table 1 Comparison of the WVTRs obtained in this study and those in previous reports.

Recently, Song et al. reported an epidermal sensor patch that showed greater performance than conventional gel electrodes, although the WVTR was low (9.65 g m−2 h−1) because of the large thickness of the patch (~125 μm)42. In contrast, fibrous network structures generally exhibited higher WVTRs than continuous films. Then, Someya et al. reported an electrospun polyvinyl alcohol (PVA) nanomesh covered with Au (thickness: 1.8 µm), which showed a WVTR of 130 g m−2 h−1 based on Eq. (1), based on the WVTRs of bare skin (WS: ~7.5 g m−2 h−1) and the nanomesh on bare skin (i.e., WNS: ~7.0 g m−2 h−1)21. Moreover, Zhang et al. reported a Janus multilayered electrospun nonwoven membrane (thickness: 15 µm) with a WVTR of 1748 g m−2 (24 h)−1 (see ref. 22). In contrast, our bioelectrodes with nm-thick continuous films and fibrous SWCNT network structures (total thickness: 430 nm) showed equivalent or greater WVTRs than conventional stretchable electrodes. Although the measuring conditions (i.e., temperature and humidity) affected the WVTR, our present findings suggested that combining elastomeric nanosheets with SWCNT fibers could lead to higher WVTRs than those of the conventional electrospun membrane22, because of the minimal thicknesses of the nanosheets and the permeable structures of SWCNT fibers.

In general, the water vapor molecules adsorbed on the membrane surface dissolved into the polymer membrane, diffused between polymer networks, and permeated to the other side of the membrane due to the differences in water vapor pressure throughout the polymer membrane and in the micro-Brownian motion characteristics of the polymer chains on the membrane surface. Water vapor permeation was observed even in thick films (several tens of μm) of hydrophobic elastomeric materials43,44. In addition, the amount of water vapor permeated per unit of time increased with decreasing film thickness, while the area of the film remained unchanged. The SBS consisted of a soft segment composed of polybutadiene in a rubber state and a hard segment composed of polystyrene (PSt) in a glassy state at room temperature, forming a solid film with elastomeric elasticity. We assumed that the rubbery soft segment composed of polybutadiene blocks dissolved and diffused water vapor and gas molecules more easily than the hard segment due to the micro-Brownian motion characteristics of the molecular chains of polybutadiene. We performed a WVTR test on PSt and SBS nanosheets and observed that the WVTR of SBS with a thickness of 787 nm was 1.5 times greater than that of PSt with a thickness of 790 nm, despite having almost the same membrane thickness, as shown in Supplementary Fig. S1b. These results suggested that even hydrophobic polymers exhibited water vapor permeability. Furthermore, the WVTR increased due to the presence of rubbery domains in the polymer membrane.

Adhesion of the nanosheets

The reliable adhesion of SWCNT-SBS nanosheets was considered essential for skin-conformable bioelectrodes. Hence, the adhesion forces of the conductive nanosheets were evaluated using a tack separation test45 (Fig. 5a). Figure 5b shows the curves of the adhesion force as a function of peeling displacement. The curves of the SWCNT-SBS nanosheets exhibited different behaviors from those of the PH1000/BG5-SBS nanosheets. SWCNT-SBS showed a second peak or shoulder after the maximum force peak, whereas PH1000/BG5-SBS only showed one peak. The low elastic moduli of the SWCNT-SBS nanosheets, which decreased the flexural rigidity, enhanced the adhesion to the wrinkled model skin surface, thereby leading to an increasingly gradual peeling process. (i) The SWCNT-SBS nanosheets started to detach from the edge of the model skin while remaining adhered to the inner area, and (ii) an additional peeling force was required to detach the nanosheet from the inner area. Conversely, PH1000/BG5-SBS simultaneously detached from the edge and inner area of the model skin due to its relatively high yield stress and elastic modulus.

Fig. 5: Adhesiveness of the conductive nanosheets.
figure 5

a Schematic of the tack separation test used to measure the adhesion of the conductive nanosheets. b Adhesion force curves of the conductive nanosheets as a function of peeling displacement. c Adhesion energies of the conductive nanosheets.

In addition, the adhesive energy was calculated by integrating the adhesion force as a function of displacement (Fig. 5c). Although there was no significant difference among the groups, the SWCNT group had a greater adhesion energy than did the PH1000/BG5 group, suggesting that decreases in the elastic moduli of the conductive nanosheets could contribute to the increase in adhesion energy.

Durability concerning the electrical resistance values of conductive nanosheets during bending and immersion tests in artificial sweat

In practical applications, changes in the resistance values of electrodes should be small during mechanical deformation and under humid conditions, such as when in contact with sweat. Thus, we performed bending and sweat immersion tests on conductive nanosheets using a model skin. Figure 6a, b shows the changes in the resistance (R/R0) values of the conductive nanosheets during the first cycle of bending and during multiple cycles of bending, respectively. The bending angles were estimated via image analysis by calculating the angles of the black markers on the sides of the specimens, as shown in Supplementary Fig. S1a, b. The R/R0 value of SWCNT 3rd-SBS increased by 1.1 times (based on the average value of three samples) at a maximum bending angle of 47° and slightly increased during recovery to 0°. The cyclical bending increased the R/R0 of the SWCNT 3rd-SBS to ~1.3, and the R/R0 was maintained at ~1.3 for 300 cycles. These resistance changes were attributed to a rearrangement of the fibrous SWCNT network caused by sliding and buckling between SWCNT bundles29. After the rearrangement, little to no structural changes in the fibrous network should occur within the same deformation range, considering that the R/R0 of the SWCNT 3rd-SBS was almost unchanged after reaching 1.3 in the second bending cycle (Fig. 6b). The R/R0 of PH1000/BG5-SBS increased by approximately 3.2 times at 47° and slightly decreased during recovery to 0°. Several tens of bending cycles significantly increased the R/R0 by 14 to 437 times, as shown clearly in Supplementary Fig. S2. Microcracks in the PH1000/BG5 layer formed between the model skin and the Au/PI collecting electrode. These cracks formed because the nanosheet and model skin were bent due to the differences in the tensile strains between the rigid Au/PI electrode and model skin. Model skin had a stretchability comparable to that of human skin. Before reaching 30 bending cycles, PH1000/BG5-SBS ruptured at the interface between the model skin and the Au/PI electrode, as shown in Supplementary Fig. S3d–f, whereas the R/R0 of SWCNT 3rd-SBS was relatively unchanged after 300 cycles. These results suggested that the SWCNT-based fibrous network structure was considered a promising candidate for use in stretchable bioelectrodes.

Fig. 6: Durability of the electrical resistance of conductive nanosheets to bending deformation and immersion in artificial sweat.
figure 6

Dependences of the changes in the resistance values of conductive nanosheets (a) on the bending angle at the first cycle in the bending test and b on the number of bending cycles. Changes in the average resistance values of the conductive nanosheets (c) in alkaline artificial sweat (pH 8.0) and d in acidic artificial sweat (pH 5.5) at room temperature for ~25 h. e Photographs of conductive nanosheets (left: PH1000/BG5-SBS, right: SWCNT 3rd-SBS) on a skin surface before rubbing and after rubbing with water-moistened pulp paper ten times.

The dependences of R/R0 on the immersion time in alkaline and acidic artificial sweat are shown in Fig. 6c, d, respectively. By averaging the R/R0 values of the three specimens (Supplementary Fig. S4), we found that the R/R0 values of the SWCNT 3rd-SBS specimens increased to approximately 1.1 in the alkaline and acidic environments at 1500 min. This trend was comparable to that of PH1000/BG5-SBS. The gradual infiltration of water into the conductive layer could induce monotonic increases in the R/R0 values of the conductive nanosheets. The SWCNT 3rd-SBS and PH1000/BG5-SBS nanosheets were not broken or damaged after immersion in artificial sweat for 1 day at room temperature.

In addition, we investigated the mechanical tolerance characteristics of the nanosheets on the skin against rubbing with water-moistened pulp paper because the nanosheet electrode on the skin could rub against clothes with sweat in practical applications (Fig. 6e, f and Supplementary Movie 1). PH1000/BG5-SBS nanosheets with conductive layers facing the skin surface were detached from the skin after being rubbed ~5 times because of the swelling of the PH1000/BG5 layer. In contrast, the SWCNT 3rd-SBS nanosheets showed little to no exfoliation after being rubbed ten times. The balance between the hydrophilicity and hydrophobicity of the conductive layer was considered important for determining the tolerance to mechanical friction under humidity. The bending test, sweat immersion test, and rubbing test results indicated that the SWCNT 3rd-SBS nanosheet could be an effective skin contact electrode.

Surface electromyography (sEMG) measurements and electrochemical impedance results of the conductive nanosheets

Finally, to evaluate the bioelectrode performance characteristics of the conductive nanosheets, we performed sEMG measurements using SWCNT 3rd-SBS, PH1000/BG5-SBS, and Ag/AgCl gel electrodes on the flexor digitorum of the forearm. A pair of bioelectrodes was attached to the skin and connected to a commercially available wireless sEMG measurement unit, as shown in Fig. 7a. The sEMG signals were recorded under gripping motion using a 20-kg-load gripper, which resulted in 24.6 ± 2.3 dB and 21.7 ± 2.2 dB lower SNRs for the SWCNT 3rd-SBS and PH1000/BG5-SBS electrodes, respectively, than that of the Ag/AgCl gel electrode (33.3 ± 3.5 dB).

Fig. 7: Results of surface electromyogram (sEMG) measurements and electrochemical impedance spectroscopy using conductive nanosheets.
figure 7

a Photograph indicating the layouts of the electrodes, the wires between the electrodes and the wireless sEMG measuring unit. The sEMG signals measured using b Ag/AgCl gel electrodes, c SWCNT 3rd-SBS nanosheets, and d PH1000/BG5-SBS nanosheets when gripping and releasing a 20-kg-load gripper. e Calculated SNRs of the three electrode types. f Electrochemical impedance spectra of the three electrodes (: SWCNT 3rd-SBS, ▲: PH1000/BG5-SBS, ■: Ag/AgCl gel electrode) on the skin obtained using a sinusoidal potential wave with an amplitude of 10 mV applied in the range of 10–1000 Hz.

The electrode–skin impedance could affect the SNR of each electrode. Therefore, we performed EIS measurements of the electrodes on a skin surface. Lopes et al. considered that electrodes with ionic conductivity placed in the human body (i.e., hydrogels and commercially available Ag/AgCl gel electrodes) exhibited low impedances due to their ability to facilitate the passage of biological potential from the human body to the measuring instrument46. Notably, SWCNT 3rd-SBS and PH1000/BG5-SBS had similar |Z| values, as shown in Fig. 7d. However, the total impedance |Z| of the Ag/AgCl gel electrode was the lowest among the three electrodes within the range of the sEMG signals (i.e., 10–500 Hz). The Ag/AgCl gel electrodes had ionic conductivities, resulting in low electrode–skin impedance. These impedance values were in good agreement with the SNR values (Fig. 7e, f).

As listed in Table 2, our electrodes (i.e., SWCNT 3rd-SBS) had thicknesses on the nanoscale, which were significantly thinner than those previously reported, and no adhesives were required for attachment to the skin surface. The SNRs of our electrodes were similar (24.6 ± 2.3 dB) to those of the electrospun Janus nonwoven membrane and commercially available Ag/AgCl gel electrodes (26 and 33.3 ± 3.5 dB, respectively). In addition, our electrodes had the highest WVTRs (Table 1), which was essential for bioelectrode applications, despite having a continuous membrane support layer (i.e., SBS nanosheet layer). In summary, we obtained skin-conformable bioelectrodes with high water vapor permeabilities, which showed comparable performance in sEMG measurements to those of conventional electrodes (e.g., Ag/AgCl gel electrodes).

Table 2 Comparison of the physical and electrochemical properties and SNRs obtained in this study and previous reports.

Conclusions

By combining the conductive fibrous networks of SWCNTs and SBS elastomer nanosheets, we obtained water-vapor permeable and stretchable SWCNT-SBS nanosheets. The densities of SWCNT fibrous bundles on the SBS nanosheets increased with an increase in the number of SWCNT coatings, in addition to an increase in the thickness of the SWCNT layer. This finding was supported by the results from SEM images, profile measurements, and UV‒Vis spectra. With increasing density and thickness of the SWCNT bundles, the elastic modulus slightly increased, and the elongation at break decreased. Furthermore, the SWCNT 3rd-SBS showed a comparable sheet resistance to PH1000/BG5-SBS but a considerably reduced elastic modulus and increased elongation. A structural change in the SWCNT networks affected the water vapor permeability owing to the change in the number of coatings. The WVTR of the 210-nm-thick SBS nanosheet was 268,172 g m−2 (2 h)−1 (i.e., membrane WVTR), which was 42 times greater than that of the filter paper used as a porous substrate (6345 g m−2 (2 h)−1). Moreover, the negative slope as a function of thickness was small. Although the SWCNT-SBS nanosheets showed significant decreases in their WVTRs with increasing SWCNT densities and thicknesses, the membrane WVTR of SWCNT 3rd-SBS was 28,317 g m−2 (2 h)−1, which was one and two orders of magnitude greater than that of a previously reported textile electrode (e.g., 1748 g m−2 (24 h)−1)22 and human skin (204 g m−2 (24 h)−1)41, respectively). These results demonstrated the effective permeability and stretchability characteristics of the SWCNT-SBS nanosheets for application in skin-conformable bioelectrodes.

Furthermore, bending tests demonstrated that the fibrous networks helped suppress the rupture between the nanosheets and collecting electrodes (PI/Au). The SWCNT 3rd-SBS on model skin showed several differences in the changes in R/R0 (<1.3) compared with PH1000/BG5. In addition, the R/R0 of the SWCNT 3rd-SBS exhibited a similar behavior to that of PH1000/BG5-SBS during immersion in artificial sweat, increasing to 1.1 after 25 h. The tolerance of SWCNT 3rd-SBS nanosheets on skin to rubbing with water-moistened paper was greater than that of a PH1000/BG5-SBS nanosheet because of the hydrophobicity of the SWCNT layer. The sEMG results revealed that the SNRs of the SWCNT 3rd-SBS and PH1000/BG5-SBS nanosheets were lower than that of the Ag/AgCl gel electrode which had the lowest impedance on skin among the three electrodes measured by EIS. These results suggested that reducing the impedance at the skin/electrode interface could increase the SNR. Decreasing the thickness and flexural rigidity of conductive nanosheets could enhance their adhesiveness and conformability to human skin and reduce electrode–skin impedance. Overall, our electrodes enabled the long-term continuous detection of bioelectrical potentials without discomfort caused by sweat accumulation.

Methods

Materials

Polyvinyl alcohol (PVA) (Mw 22,000, 86.5–89% hydrolyzed) was purchased from Kanto Chemical Japan. SBS (Mw 140,000) was purchased from Sigma‒Aldrich Corporation. A SWCNT aqueous dispersion (solid content, 0.2 wt.%) was purchased from Meijo Nano Carbon Co., Ltd. A PEDOT:PSS aqueous dispersion, Clevios PH1000, was purchased from Heraeus Deutschland GmbH & Co. (Leverkusen, Germany). Tetrahydrofuran (THF) with a stabilizer and BG (98.0 + %, Wako Special Grade) were purchased from Fujifilm Wako Pure Chemical Corporation. The Capstone FS-31 fluorosurfactant was purchased from Chemours.

Fabrication of the conductive nanosheets

A 5 wt.% aqueous PVA solution was coated on a roll of polyethylene terephthalate (PET) film using a roll-to-roll gravure coating system (Mini-Labo, Yasui Seiki Company, Ltd.). A solution of 5 wt.% SBS in THF 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 containing 0.01 wt.% FS-31 and the PH1000 dispersion containing 5 wt.% BG were stirred separately using a magnetic stirrer for over 18 h at room temperature. Subsequently, the SWCNT or PH1000 dispersion was coated on the PET/PVA/SBS sheet and dried at 80 °C. The PET/PVA/SBS sheet was coated with multiple SWCNT layers by repeating the coating process. The structural schematics are shown in Supplementary Fig. S5a. The PVA sacrificial layer on one side of the fabricated nanosheets was rinsed using deionized water, followed by peeling the nanosheet/PVA film with an adhesive tape frame from PET, as shown in Supplementary Fig. S5b. After depositing the nanosheets on a nylon mesh, the nanosheets were attached to an adherent (e.g., human skin, substrates and model skin). A clean cloth dampened with deionized water was pressed against the back side (i.e., opposite side of the nanosheet) of the nylon mesh to detach the nanosheet from the mesh and transfer the nanosheet to the adherent.

Thickness, sheet resistance, UV‒Vis spectra and mechanical property measurements and SEM imaging of the nanosheets

The resulting nanosheets supported by the water-soluble PVA layers were peeled using adhesive tape frames and attached to a Si substrate after the PVA layer was washed with deionized water. The thicknesses of the conductive nanosheets were determined using a surface profiler (DektakXT, Bruker). The resistance values of the SWCNT-SBS nanosheets were measured using a four-point probe method (Loresta-AX MCP-T370, Nittoseiko Analytech Co., Ltd.). Tensile testing was performed using a tensile tester (EZ Test EZ-SX, SHIMADZU Co., Ltd.) equipped with a 5 N load cell. The conductive nanosheets were first 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 the stress‒strain (S‒S) curve was steepest in the elastic deformation region (i.e., below the yield stress). Then, the elastic modulus was estimated from the slope of a line fitted to the curve within 0.25% of the determined strain. UV‒Vis spectra of the fabricated nanosheets attached to a quartz substrate were measured in the wavelength range of 220–1100 nm. The surface morphologies of the nanosheets attached to Si wafers were observed by field-emission scanning electron microscopy (FE-SEM; S-5500, Hitachi High-Tech Corporation) using an acceleration voltage of 10 kV. For imaging, the surface of the pristine SBS nanosheet on the Si wafer was coated with Pt by sputtering at a discharge current of 10 mA for 30 s.

Permeability measurements of the nanosheets to humidity

The water vapor transmission ratios (WVTRs) of conductive nanosheets deposited on a paper filter (No. 5 B, Kiriyama Glass Works Co.) were measured using the dish method, which was compliant with JIS Z0208. The area of permeating humidity each specimen was φ60 mm. After the introduction of CaCl2 (7.0 g) as a drying agent into an inner dish, the specimen was placed in the dish and sealed with paraffin. The changes in the weights of the dish samples in an atmosphere at a temperature of 40 °C and an RH of 90% for a duration of 2 h were measured using a thermohygrostat to calculate the WVTR according to the following equation:

$${\rm{WVTR}}=\frac{240\times ({W}_{2h}-{W}_{0h})}{T\times A}$$
(2)

where W2h and W0h are the weights at 2 and 0 h, respectively, and T and A are the retention times in the thermohygrostat and the water vapor transmission area of the specimen, respectively.

The WVTRs of the SBS and PSt nanosheets adhered to the spouts of the bottles were calculated by measuring the changes in the weights of the deionized water at each retention time in a thermostatic chamber. The SBS and PSt nanosheets were attached to the spout of a bottle after the introduction of 30.0 g of deionized water into the bottle. Furthermore, the bottle neck (i.e., the boundary between the neck and the nanosheet) was sealed with hot-melt adhesive, as shown in Supplementary Fig. S1a. The humidity of the nanosheet permeating area was φ20 mm. With a commercially available thermohygrometer (temperature range, −10–50 °C; humidity range, 20–50% RH), the temperature in the chamber was ~35 °C, and the humidity was below the detection limit.

Estimation of the adhesion energies of the conductive nanosheets

Tack separation tests45 using a tensile tester (EZ-S-5 N, Shimadzu Co., Ltd.) were performed to evaluate the adhesion characteristics of the conductive nanosheets to the surface of an artificial model skin (BIOSKIN Plate P002-001#10, Beaulax Co., Ltd.) consisting of polyurethane. The conductive nanosheet cut into 30-mm squares was applied to a fixture (Fig. 5a), and the conductive side of the nanosheet was applied to a 10-mm square model skin surface and allowed to rest for 15 min with an applied force of 0.1 N. The fixture was pulled upward at a constant rate (10 mm/min) to obtain a curve of the tack separation force (i.e., adhesion force) versus the separation stroke.

Bending and sweat immersion tests for the conductive nanosheets

To elucidate the dependence levels of the electrical properties on the mechanical deformation, the changes in the resistance values of the conductive nanosheets attached to the surface of the model skin were measured during bending. Two Au-sputtered polyimide (Au/PI) electrodes were attached to both ends of the model skin. The distance between the two Au/PI films was 25 mm. The conductive nanosheets (SWCNT 3rd-SBS or PH1000/BG5-SBS, 5 mm × 40 mm) were attached across the Au/PI films at both ends of the model skin. The conductive nanosheets on the model skin were repeatedly bent 300 times using a three-point bending apparatus on a mechanical tester (UniVert, CellScalle, Canada).

The physicochemical stabilities of conductive nanosheets (SWCNT 3rd-SBS and PH1000/BG5-SBS) under sweat were investigated by immersing the nanosheets in artificial sweat (JIS L0848, Hayashi Pure Chemical Ind., Ltd.) at pH values of 5.5 and 8.0 for ~24 h at room temperature. Two Au/PI electrodes were attached to both ends of the model skin. The distance between the two Au/PI films was 20 mm. The conductive nanosheets (7.5 mm × 40 mm) were attached across the Au/PI films at both ends of the model skin, with the conductive side facing the atmosphere. The conductive nanosheets were sealed using PI tape, except for the φ7 mm area at the center of the nanosheets. A PVC tube (inner diameter: φ9 mm) was placed over the centers of the nanosheets and secured using a glue gun to define and contain the region of exposure. Artificial sweat was introduced to the exposed sample area, and the changes in the resistance values of the nanosheets were measured using LCR meters (U1733C, U5481B, U1780A, Keysight Technologies).

Surface electromyography (sEMG) and electrochemical impedance spectroscopy (EIS) measurements of the conductive nanosheets on human skin

The sEMG signal was measured using an sEMG measuring system (SS-EMGW-HMAG, SPORTS SENSING Co., Ltd.) at a sampling rate of 1000 Hz. The conductive nanosheet and Ag/AgCl gel electrode (Vitrode f150m, Nikon Kohden Corporation) were placed on the flexor digitorum of the dominant arm 5 cm from the elbow with an interelectrode distance of 3 cm. Au/PI electrodes were used as collecting electrodes for measuring the conductive nanosheets. A signal was induced by grabbing and releasing a 20-kg-load grip every 5 s. This process was repeated five times. The SNR (in dB) was calculated using the following equation36,47:

$${{\rm{SNR}}=20\log }_{10}\frac{{A}_{s}}{{A}_{n}}$$
(3)

where As is the highest peak of an sEMG signal during the gripping period and An is the standard deviation of the background noise during the releasing period.

Electrochemical impedance spectroscopy (EIS) measurements of the conductive nanosheets and Ag/AgCl gel electrodes on human skin were performed using a potentiogalvanostat with an FRA circuit (SP-300, BioLogic, Grenoble, France) at room temperature. A sinusoidal signal with an amplitude of 10 mV in the frequency range of 10 Hz–0.1 MHz was applied. The same electrodes were placed on the flexor digitorum on the dominant arm 5 cm from the elbow, with an interelectrode distance of 3 cm. The Au/PI electrodes were used as collecting electrodes to measure the conductive nanosheets. The reference and counter electrodes were Ag/AgCl gel electrodes, and the working electrode was the measured electrode (i.e., conductive nanosheets and Ag/AgCl gel).