Micromachined Planar Supercapacitor with Interdigital Buckypaper Electrodes

In this work, a flexible micro-supercapacitor with interdigital planar buckypaper electrodes is presented. A simple fabrication process involving vacuum filtration method and SU-8 molding techniques is proposed to fabricate in-plane interdigital buckypaper electrodes on a membrane filter substrate. The proposed process exhibits excellent flexibility for future integration of the micro-supercapacitors (micro-SC) with other electronic components. The device’s maximum specific capacitance measured using cyclic voltammetry was 107.27 mF/cm2 at a scan rate of 20 mV/s. The electrochemical stability was investigated by measuring the performance of charge-discharge at different discharge rates. Devices with different buckypaper electrode thicknesses were also fabricated and measured. The specific capacitance of the proposed device increased linearly with the buckypaper electrode thickness. The measured leakage current was approximately 9.95 µA after 3600 s. The device exhibited high cycle stability, with 96.59% specific capacitance retention after 1000 cycles. A Nyquist plot of the micro-SC was also obtained by measuring the impedances with frequencies from 1 Hz to 50 kHz; it indicated that the equivalent series resistance value was approximately 18 Ω.


Introduction
Supercapacitors (SCs) are energy storage devices that provide a higher density of energy than conventional dielectric capacitors, and higher density of power than batteries. SCs are frequently employed to power portable electronic devices because of their efficient charging/discharging performance and capability to reliably operate for millions of cycles. In the past decade, SCs with carbon nanotubes (CNTs) electrodes have drawn global attention because of the unique properties of CNTs, such as high specific surface area, electrical conductivity, and chemical stability [1][2][3][4]; these qualities make CNTs high-quality electrode materials for SCs [5][6][7][8].
Thin films formed with aggregates of CNTs are frequently employed as both current collectors and active materials for flexible SCs [9]. Various techniques for realizing CNT films have been reported. Najafabadi et al. proposed high-power SC electrodes with composites of carbon nanohorns and CNTs [10]; the meso-macro pore structure engineered by employing single-walled CNTs as scaffolding for single-walled carbon nanohorns improved the electrode's power density. Do et al. proposed a method of vanadium oxide deposition on multi-walled CNT buckypaper that served as SC electrodes [11]. A novel supercritical fluid process was used for the deposition of vanadium oxide onto the buckypapers. In [12], a new technique of synthesizing CNT and nanofiber ensembles using a template method was proposed. Synthesis of carbon by using chemical vapor deposition within Microscopy Sciences Corporation, Hatfield, PA, USA) of 200 nm was deposited on the top surface of the patterned buckypaper and served as the current collector. Also, the trenches between the electrodes were filled with gel electrolyte.
Micromachines 2018, 9, x FOR PEER REVIEW 3 of 11 surface of the patterned buckypaper and served as the current collector. Also, the trenches between the electrodes were filled with gel electrolyte.
(a) (b) As shown in Figure 1b, during the micro-SC's charging process, electrons moved from the positive to the negative electrode via external power sources. Additionally, positive and negative ions in the electrolyte separated and moved to the electrode surfaces, resulting in the formation of electric double layers [31]. The device stored energy because ions of opposite charge accumulated on the double layers of electrochemically stable electrodes with high specific surface area. This study used patterned interdigital buckypaper to serve as the electrode material, which provides high electrochemical stability during the charge-discharge process. Moreover, the high surface-to-volume ratio of the porous buckypaper electrodes caused the micro-SC to have high energy and power density [25]. Also, using a filtration paper as the substrate allowed the proposed in-plane device to be flexible and relatively thin, thus easily integrated with wearable devices.
Notably, the capacity of the proposed in-plane interdigital micro-SC can be increased by increasing the thickness of the buckypaper electrodes; its charge-discharge rates are barely affected because there is almost no increase in the ion migration distance. For SCs with planar sandwiched structures, however, as the thickness of the buckypaper increases, the ion migration distance also increases, which in turn deteriorates the charge-discharge performance [25,32].

Fabrication
The fabrication process of the proposed micro-SC is shown in Figure 2. First, a layer of 170-µm SU-8 thick-film photoresist (SU-8 2050, MicroChem Corporation, Westborough, MA, USA) was spin-coated (20 s at 500 rpm and 50 s at 1000 rpm) onto a silicon handling wafer, as shown in Figure 2a. The model of the spin-coater was SP-01 (APISC Corporation, Taoyuan, Taiwan). Figure 2b shows how a nylon membrane filter, which served as the micro-SC's flexible substrate, was then placed on the top of the SU-8 photoresist. Then, a second layer of 100-µm SU-8 photoresist was spincoated (20 s at 500 rpm and 50 s at 1400 rpm) on top of the membrane filter ( Figure 2c). The SU-8 layer was patterned ( Figure 2d) and developed using a standard photolithography process, forming an SU-8 mold for the interdigital electrodes pair (Figure 2e). Then, CNT solution dispersed with MWCNTs (0.01 wt % concentration) was filtrated through the membrane filter with the SU-8 mold by using the vacuum filtration technique [33]; MWCNTs filled into the SU-8 mold, as shown in Figure 2f. The solution dispersed with MWCNTs was subjected to ultrasonic agitation for 120 min to reduce the CNTs' tendency to bundle. After vacuum filtration, the MWCNT film was kept at room temperature for 120 min to evaporate residual solvent thoroughly. Then, a 200-nm Au layer, which served as the current collector, was deposited on top of the MWCNT film ( Figure 2g). After removing the SU-8 mold with solvent stripper (Remover PG, MicroChem Corporation, Westborough, MA, USA) [24], the patterned interdigital buckypaper electrodes were fabricated (Figure 2h). The removal As shown in Figure 1b, during the micro-SC's charging process, electrons moved from the positive to the negative electrode via external power sources. Additionally, positive and negative ions in the electrolyte separated and moved to the electrode surfaces, resulting in the formation of electric double layers [31]. The device stored energy because ions of opposite charge accumulated on the double layers of electrochemically stable electrodes with high specific surface area. This study used patterned interdigital buckypaper to serve as the electrode material, which provides high electrochemical stability during the charge-discharge process. Moreover, the high surface-to-volume ratio of the porous buckypaper electrodes caused the micro-SC to have high energy and power density [25]. Also, using a filtration paper as the substrate allowed the proposed in-plane device to be flexible and relatively thin, thus easily integrated with wearable devices.
Notably, the capacity of the proposed in-plane interdigital micro-SC can be increased by increasing the thickness of the buckypaper electrodes; its charge-discharge rates are barely affected because there is almost no increase in the ion migration distance. For SCs with planar sandwiched structures, however, as the thickness of the buckypaper increases, the ion migration distance also increases, which in turn deteriorates the charge-discharge performance [25,32].

Fabrication
The fabrication process of the proposed micro-SC is shown in Figure 2. First, a layer of 170-µm SU-8 thick-film photoresist (SU-8 2050, MicroChem Corporation, Westborough, MA, USA) was spin-coated (20 s at 500 rpm and 50 s at 1000 rpm) onto a silicon handling wafer, as shown in Figure 2a. The model of the spin-coater was SP-01 (APISC Corporation, Taoyuan, Taiwan). Figure 2b shows how a nylon membrane filter, which served as the micro-SC's flexible substrate, was then placed on the top of the SU-8 photoresist. Then, a second layer of 100-µm SU-8 photoresist was spin-coated (20 s at 500 rpm and 50 s at 1400 rpm) on top of the membrane filter ( Figure 2c). The SU-8 layer was patterned ( Figure 2d) and developed using a standard photolithography process, forming an SU-8 mold for the interdigital electrodes pair (Figure 2e). Then, CNT solution dispersed with MWCNTs (0.01 wt % concentration) was filtrated through the membrane filter with the SU-8 mold by using the vacuum filtration technique [33]; MWCNTs filled into the SU-8 mold, as shown in Figure 2f. The solution dispersed with MWCNTs was subjected to ultrasonic agitation for 120 min to reduce the CNTs' tendency to bundle. After vacuum filtration, the MWCNT film was kept at room temperature for 120 min to evaporate residual solvent thoroughly. Then, a 200-nm Au layer, which served as the current collector, was deposited on top of the MWCNT film ( Figure 2g). After removing the SU-8 mold with solvent stripper (Remover PG, MicroChem Corporation, Westborough, MA, USA) [24], the patterned interdigital buckypaper electrodes were fabricated ( Figure 2h). The removal process was facilitated by the nylon membrane filter's permeability since the solvent stripper solution easily penetrated the membrane and reached the contact interface between the SU-8 structure and the membrane.
Micromachines 2018, 9, x FOR PEER REVIEW 4 of 11 process was facilitated by the nylon membrane filter's permeability since the solvent stripper solution easily penetrated the membrane and reached the contact interface between the SU-8 structure and the membrane. Before packaging the device, PVA-KOH gel electrolyte was added to the buckypaper electrodes using a syringe. Then, the trenches of interdigital porous buckypaper electrodes were filled and soaked with gel electrolyte (Figure 2i). The electrolyte was synthesized by mixing 2.8 g of potassium hydroxide (KOH) and 5 g of polyvinyl alcohol (PVA) with 50 mL of deionized water at 85 °C. Then, the mixture was stirred for 120 min until the solution became clear. Finally, the device was sealed by polymer films (Surlyn ® , DuPont Corp, Wilmington, DE, USA) ( Figure 2j). The measured relationships between the thickness of the fabricated buckypaper and the consumption of MWCNT solution is shown in Figure 3. The relationships are quite linear. This figure also compares the results of SCs fabricated using filter membranes patterned with SU-8 structures with those using filter membranes without SU-8 structures. The two curves are almost identical, which indicates the SU-8 structure does not affect the filtration's efficiency. Note that the thickness Before packaging the device, PVA-KOH gel electrolyte was added to the buckypaper electrodes using a syringe. Then, the trenches of interdigital porous buckypaper electrodes were filled and soaked with gel electrolyte (Figure 2i). The electrolyte was synthesized by mixing 2.8 g of potassium hydroxide (KOH) and 5 g of polyvinyl alcohol (PVA) with 50 mL of deionized water at 85 • C. Then, the mixture was stirred for 120 min until the solution became clear. Finally, the device was sealed by polymer films (Surlyn ® , DuPont Corp, Wilmington, DE, USA) (Figure 2j).
The measured relationships between the thickness of the fabricated buckypaper and the consumption of MWCNT solution is shown in Figure 3. The relationships are quite linear. This figure also compares the results of SCs fabricated using filter membranes patterned with SU-8 structures with those using filter membranes without SU-8 structures. The two curves are almost identical, which indicates the SU-8 structure does not affect the filtration's efficiency. Note that the thickness of the SU-8 frame should be at least 20 µm greater than the buckypaper thickness to ensure successful SU-8 removal after Au film deposition.

Measurement and Discussion
The CV curves of the capacitor at different scanning rates, which were measured using an electrochemical station (CHI 627D, CH Instruments, Austin, TX, USA), are shown in Figure 6a. The proposed device exhibited typical capacitive behavior with quasi-rectangular CV curves. In addition, specific capacitances can be evaluated by using Equation (1) [33]: where AREACV is the integral area of a CV curve obtained by integrating the forward and backward sweeps in the cyclic voltammogram, A is the total active area of the buckypaper electrodes, s is the potential scanning rate, and ΔV is the range of the potential sweep. The calculated specific capacitances at different scan rates are listed in Table 1.
Obviously, at very low scan rates, the capacitance values are higher because the ions have sufficient time to penetrate and reside in all the available pores on electrodes, thereby forming electric double layers, which are essential to yield larger capacitance. Using Equation (1), a maximum specific capacitance of 107.27 mF/cm 2 was obtained at a scan rate of 20 mV/s. Devices with different electrode thicknesses were also fabricated and measured. Table 2 shows the results for each electrode configuration at a scan rate of 20 mV/s. The results indicate that the specific capacitance of the devices increased linearly with the thickness of the buckypaper electrodes. Table 3 shows the comparison of specific capacitances among SC with interdigitated electrodes published in recent works. The proposed micro-SC of this work exhibits excellent performance. Figure 6b shows the capacitance retention ratio versus the number of repeating CV cycles at a scan rate of 1 V/s. The specific capacitance retained 96.59% of its initial value after 1000 cycles. The figure inset shows the CV curves of the 1st, 250th, 750th, and 1000th cycles. These results indicate that the proposed micro-SC has satisfactory cycle stability.

Measurement and Discussion
The CV curves of the capacitor at different scanning rates, which were measured using an electrochemical station (CHI 627D, CH Instruments, Austin, TX, USA), are shown in Figure 6a. The proposed device exhibited typical capacitive behavior with quasi-rectangular CV curves. In addition, specific capacitances can be evaluated by using Equation (1) [33]: where AREA CV is the integral area of a CV curve obtained by integrating the forward and backward sweeps in the cyclic voltammogram, A is the total active area of the buckypaper electrodes, s is the potential scanning rate, and ∆V is the range of the potential sweep. The calculated specific capacitances at different scan rates are listed in Table 1.
Obviously, at very low scan rates, the capacitance values are higher because the ions have sufficient time to penetrate and reside in all the available pores on electrodes, thereby forming electric double layers, which are essential to yield larger capacitance. Using Equation (1), a maximum specific capacitance of 107.27 mF/cm 2 was obtained at a scan rate of 20 mV/s. Devices with different electrode thicknesses were also fabricated and measured. Table 2 shows the results for each electrode configuration at a scan rate of 20 mV/s. The results indicate that the specific capacitance of the devices increased linearly with the thickness of the buckypaper electrodes. Table 3 shows the comparison of specific capacitances among SC with interdigitated electrodes published in recent works. The proposed micro-SC of this work exhibits excellent performance. Figure 6b shows the capacitance retention ratio versus the number of repeating CV cycles at a scan rate of 1 V/s. The specific capacitance retained 96.59% of its initial value after 1000 cycles. The figure inset shows the CV curves of the 1st, 250th, 750th, and 1000th cycles. These results indicate that the proposed micro-SC has satisfactory cycle stability.
The proposed device was also tested by galvanostatic charge-discharge cycling at various current densities, as shown in Figure 7a. The corresponding current densities of these curves are 1, 2, 5, and 10 mA/cm 2 . The linear galvanostatic discharge shows that the proposed SC exhibits excellent capacitive behaviors. Note that a small voltage drop at the start of the discharge curve for each galvanostatic charge-discharge curve indicates the existence of internal resistance.      Figure 7b shows the leakage current curves of the device, which was charged at 2 mA from 0.0 to 0.8 V, and then maintained at 0.8 V for 3600 s. At the onset of the charging, the leakage current dropped significantly (from 0.534 mA to 19.8 µA after 10 s). The leakage current then decreased gradually and reached a steady value of approximately 9.95 µA after 3600 s.
A Nyquist plot of the micro-SC is shown in Figure 8. The impedances were measured with frequencies from 1 Hz to 50 kHz. At high frequencies, the micro-SC behaved as a resistor, whereas at low frequencies, it behaved as a capacitor. The measured resistance is a combination of various contributions, including the electronic resistance of the patterned buckypaper, the contact resistance between the buckypaper and current collector, and the electrolytic resistance of the buckypaper's porous structure. The equivalent series resistance value was approximately 18 Ω.  The galvanostatic charge-discharge curves can also be used to evaluate the specific capacitance of the SC by the following equation: where i is the discharge current, ∆V is the potential drop during discharge, and A is the total active area of electrodes. The calculated maximum specific capacitance using Equation (2) was 76.5 mF/cm 2 at a constant current of 1 mA/cm 2 . Table 4 shows the calculated specific capacitance for each current density.  Figure 7b shows the leakage current curves of the device, which was charged at 2 mA from 0.0 to 0.8 V, and then maintained at 0.8 V for 3600 s. At the onset of the charging, the leakage current dropped significantly (from 0.534 mA to 19.8 µA after 10 s). The leakage current then decreased gradually and reached a steady value of approximately 9.95 µA after 3600 s.
A Nyquist plot of the micro-SC is shown in Figure 8. The impedances were measured with frequencies from 1 Hz to 50 kHz. At high frequencies, the micro-SC behaved as a resistor, whereas at low frequencies, it behaved as a capacitor. The measured resistance is a combination of various contributions, including the electronic resistance of the patterned buckypaper, the contact resistance between the buckypaper and current collector, and the electrolytic resistance of the buckypaper's porous structure. The equivalent series resistance value was approximately 18 Ω.

Conclusions
This paper presents a flexible micro-supercapacitor with interdigital buckypaper electrodes realized by a fabrication process including vacuum filtration and lithography techniques. An SU-8 photoresist layer, which served as the filtration mask, was deposited on a nylon membrane filter and patterned as the mold for an interdigital buckypaper electrode. The device's electrochemical stability was confirmed by the CV and charge-discharge experiments. The measured maximum specific capacitance was 107.27 mF/cm 2 at a scan rate of 20 mV/s. Devices with different electrode thicknesses were also fabricated and measured to study the relationship between specific capacitance and buckypaper electrode thickness. In addition, a Nyquist plot of the micro-SC obtained by measuring the impedances showed a resistance value of approximately 18 Ω at high frequency. A small leakage current of 9.95 µA was observed at 3600 s after charging to 0.8 V. The specific capacitance of the device retained 96.59% of its initial value after 1000 cycles.

Conclusions
This paper presents a flexible micro-supercapacitor with interdigital buckypaper electrodes realized by a fabrication process including vacuum filtration and lithography techniques. An SU-8 photoresist layer, which served as the filtration mask, was deposited on a nylon membrane filter and patterned as the mold for an interdigital buckypaper electrode. The device's electrochemical stability was confirmed by the CV and charge-discharge experiments. The measured maximum specific capacitance was 107.27 mF/cm 2 at a scan rate of 20 mV/s. Devices with different electrode thicknesses were also fabricated and measured to study the relationship between specific capacitance and buckypaper electrode thickness. In addition, a Nyquist plot of the micro-SC obtained by measuring the impedances showed a resistance value of approximately 18 Ω at high frequency. A small leakage current of 9.95 µA was observed at 3600 s after charging to 0.8 V. The specific capacitance of the device retained 96.59% of its initial value after 1000 cycles.