Carbon Dioxide Tornado-Type Atmospheric-Pressure-Plasma-Jet-Processed rGO-SnO2 Nanocomposites for Symmetric Supercapacitors

Pastes containing reduced graphene oxide (rGO) and SnCl2 solution were screen printed on carbon cloth and then calcined using a CO2 tornado-type atmospheric-pressure plasma jet (APPJ). The tornado circulation of the plasma gas enhances the mixing of the reactive plasma species and thus ensures better reaction uniformity. Scanning electron microscopy (SEM), energy-dispersive spectroscopy (EDS), and X-ray photoelectron spectroscopy (XPS) were performed to characterize the synthesized rGO-SnO2 nanocomposites on carbon cloth. After CO2 tornado-type APPJ treatment, the pastes were converted into rGO-SnO2 nanocomposites for use as the active electrode materials of polyvinyl alcohol (PVA)-H2SO4 gel-electrolyte flexible supercapacitors (SCs). Various APPJ scanning times were tested to obtain SCs with optimized performance. With seven APPJ scans, the SC achieved the best areal capacitance of 37.17 mF/cm2 in Galvanostatic charging/discharging (GCD) and a capacitance retention rate of 84.2% after 10,000-cycle cyclic voltammetry (CV) tests. The capacitance contribution ratio, calculated as pseudocapacitance/electrical double layer capacitance (PC/EDLC), is ~50/50 as analyzed by the Trasatti method. GCD data were also analyzed to obtain Ragone plots; these indicated an energy density comparable to those of SCs processed using a fixed-point nitrogen APPJ in our previous study.


Introduction
A supercapacitor (SC) is a passive energy storage device that has a higher energy density than conventional capacitors and a higher power density than batteries [1]. Many aspects of SCs have been investigated, including their low-cost manufacturing process [2,3], flexibility [4], thermal stability [5], electrochemical stability [6], and high power density [7]. The selection and preparation of electrode active materials play critical roles in determining the SC performance [8]. SCs have two main types of charge storage mechanisms: (1) electric double-layer capacitance (EDLC) [9], in which charges are stored on the electrodeelectrolyte interface, and (2) pseudocapacitance (PC) [10], which is based on the Faraday redox reaction. Typically, in a combination of carbon-based materials, metal oxides, and conductive polymers, carbon-based materials contribute the EDLC, and the multivalent metal oxides and conductive polymers contribute the PC [11,12].

Fabrication of rGO-SnO 2 Electrodes
Figure 1a-c shows the fabrication process of the rGO-SnO 2 electrode. First, rGO-SnCl 2 pastes were screen printed on a carbon cloth current collector three times. The printed area was 1.5 × 2 cm 2 , as shown in Figure 2b. Next, the rGO-SnCl 2 pastes printed on the carbon cloth were calcined at 100 • C for 10 min to remove excess solvent. Then, the screen-printed electrodes were scanned one, three, five, seven, and nine times using CO 2 tornado-type APPJ (AC-PG-E-02, Click Sunshine Co., Ltd., New Taipei City, Taiwan). Figure 2a shows the schematic diagram of the APPJ. The APPJ parameters are as follows: CO 2 flow rate is 35 slm, the distance between the plasma jet and the substrate is 4 mm, the power is 700 W, and the frequency of voltage source is 33 kHz. Figure 2c shows the APPJ scanning routine. Owing to the use of a rotating jet (i.e., tornado-type APPJ), the plasma was formed at the circumference of the jet rather than at the center of the jet. To ensure a homogenous process, the scanning area was much larger than the screen-printed area. The time required for each scan was~55 s. The distance between each horizontal scan was 2 mm, and the distance between the exit of the plasma jet and the substrate was 4 mm. ials 2021, 14, x FOR PEER REVIEW 3 of 16 a binder, and the ethanol and terpineol were used as solvents. The details are described elsewhere [49].

Fabrication of rGO-SnO2 Electrodes
Figure 1a-c shows the fabrication process of the rGO-SnO2 electrode. First, rGO-SnCl2 pastes were screen printed on a carbon cloth current collector three times. The printed area was 1.5 × 2 cm 2 , as shown in Figure 2b. Next, the rGO-SnCl2 pastes printed on the carbon cloth were calcined at 100 °C for 10 min to remove excess solvent. Then, the screenprinted electrodes were scanned one, three, five, seven, and nine times using CO2 tornadotype APPJ (AC-PG-E-02, Click Sunshine Co., Ltd., New Taipei City, Taiwan). Figure 2a shows the schematic diagram of the APPJ. The APPJ parameters are as follows: CO2 flow rate is 35 slm, the distance between the plasma jet and the substrate is 4 mm, the power is 700 W, and the frequency of voltage source is 33 kHz. Figure 2c shows the APPJ scanning routine. Owing to the use of a rotating jet (i.e., tornado-type APPJ), the plasma was formed at the circumference of the jet rather than at the center of the jet. To ensure a homogenous process, the scanning area was much larger than the screen-printed area. The time required for each scan was ~55 s. The distance between each horizontal scan was 2 mm, and the distance between the exit of the plasma jet and the substrate was 4 mm.    a binder, and the ethanol and terpineol were used as solvents. The details are described elsewhere [49].

Fabrication of rGO-SnO2 Electrodes
Figure 1a-c shows the fabrication process of the rGO-SnO2 electrode. First, rGO-SnCl2 pastes were screen printed on a carbon cloth current collector three times. The printed area was 1.5 × 2 cm 2 , as shown in Figure 2b. Next, the rGO-SnCl2 pastes printed on the carbon cloth were calcined at 100 °C for 10 min to remove excess solvent. Then, the screenprinted electrodes were scanned one, three, five, seven, and nine times using CO2 tornadotype APPJ (AC-PG-E-02, Click Sunshine Co., Ltd., New Taipei City, Taiwan). Figure 2a shows the schematic diagram of the APPJ. The APPJ parameters are as follows: CO2 flow rate is 35 slm, the distance between the plasma jet and the substrate is 4 mm, the power is 700 W, and the frequency of voltage source is 33 kHz. Figure 2c shows the APPJ scanning routine. Owing to the use of a rotating jet (i.e., tornado-type APPJ), the plasma was formed at the circumference of the jet rather than at the center of the jet. To ensure a homogenous process, the scanning area was much larger than the screen-printed area. The time required for each scan was ~55 s. The distance between each horizontal scan was 2 mm, and the distance between the exit of the plasma jet and the substrate was 4 mm.

Preparation of Gel-Electrolyte and Assembly of rGO-SnO 2 Supercapacitor
The polyvinyl alcohol (PVA)-H 2 SO 4 gel-electrolyte was prepared by mixing 1.7 g of PVA (99+% hydrolyzed) and 1 M of H 2 SO 4 (purity: 95-97%, AUECC, Kaohsiung, Taiwan) Figure 1d−f shows the assembly procedure and the structure of SCs. First, the rGO-SnO 2 electrode was attached to a PVC substrate. Then, 0.5 mL PVA-H 2 SO 4 electrolyte was dropped uniformly on the rGO-SnO 2 electrode and dried for one day. This step was repeated three times. Finally, two pieces of PVA-H 2 SO 4 electrolyte-coated electrodes were sandwiched on the gel-electrolyte sides to form a sandwich-structured SC.

Characterization of Materials and SCs
A k-type thermocouple and module (NI-9211, National Instruments, Austin, TX, USA) were used to measure the sample temperature during APPJ processing. A spectrometer (Mars HS2000+, GIE Optics, Taipei, Taiwan) was used to detect the optical emission spectra (OES). The integration time for the OES detection was 16 ms. The water contact angle of the electrodes was measured using an optical goniometer (Model 100SB, Sindetake, Taipei, Taiwan). The morphology of the rGO-SnO 2 nanocomposites was observed by scanning electron microscopy (SEM, JSM-7800 Prime, JEOL, Tokyo, Japan) with energy-dispersive spectroscopy (EDS) at magnifications of 3000×, 30,000×, 100,000×. The bonding configuration was characterized using X-ray photoelectron spectroscopy (XPS, Sigma Probe, Thermo VG Scientific, Waltham, MA, USA) with an Al-Kα source and EDS. Electrochemical measurements were performed using cyclic voltammetry (CV) and galvanostatic charging/discharging (GCD) with a two-electrode configuration. The CV measurement was performed using an electrochemical workstation (Zennium, Zahner-Elektrik, Kronach, Germany) with potential scan rates of 200, 20, and 2 mV/s and a potential window of 0-0.8 V. The GCD measurement was performed using another electrochemical workstation (PGSTAT204, Metrohm Autolab, Utrecht, The Netherlands) with a potential window of 0-0.8 V and charging/discharging currents of 4, 2, 1, 0.5, and 0.25 mA.

Results and Discussions
3.1. Basic Information about CO 2 Tornado-Type APPJ Figure 3a shows the substrate temperature during CO 2 APPJ processing (one scan). The temperature is slightly higher than 50 • C before APPJ scanning because of the residual heat left by the warm-up scan. The temperature quickly increases to hundreds of degrees Celsius. Therefore, the APPJ is suitable for rapid thermal processing. The plasma directly bombards the substrate surface, and the instantaneous maximum temperature becomes 350 • C. Temperature oscillations could be caused by the rotation of the jet (i.e., tornadotype APPJ). Figure 3b shows the OES of the CO 2 APPJ. The peaks generated at wavelengths of 282.8, 297.9, and 519.7 nm are mainly caused by CO species [50,51]. The peak observed at 359.9 nm is caused by the transition (v 3 -v 16 ) of CO 2 . The peak at 406.0 nm is caused by the excitation of O 2 . Further, atomic O lines can be found at 776.4 and 843.9 nm [50]. The emission peaks at 386.5 and 458.1 nm are assigned to the CN violet system; the emission peaks at 391.0, 425.5 nm are assigned to the tail bands of CN; and those at 463.7 and 469.0 nm, to Le Blanc's system [52]. Balmer's series of H δ , H γ , and H α can be detected at 410.8, 433.5, and 657.0 nm, respectively [53]. Further, another emission peak of H 2 can be found at 722.5 nm [54]. The reaction of the CO 2 APPJ is quite complicated and involves reactions with environmental gases. Many characteristic peaks related to oxygen can be detected, indicating that the plasma system has a strong oxidizing property. CN species were generated owing to the use of a carbon source in CO 2 , which would react with nitrogen in the surroundings. In addition, the generation of hydrogen and some oxygen species was attributed to the splitting of water vapor in the atmospheric environments by Equation (1) [55,56].
This phenomenon could have further increased the number of oxygen species. As a result, the oxygen species were produced by the splitting of both CO 2 and water vapor.
This phenomenon could have further increased the number of oxygen species. As a result, the oxygen species were produced by the splitting of both CO2 and water vapor.  Figure 4 shows the water contact angle of the rGO-SnO2 electrode. The as-deposited rGO-SnCl2 has a high-water contact angle of 120.9°. After CO2 APPJ processing, the water droplet penetrated into the carbon cloth instantly for all APPJ scanning times. The increased hydrophilicity is mainly due to the oxygen functional groups implanted by the APPJ. This is partly due to heat and partly due to the oxidizing nature of our CO2 APPJ. This shows that the CO2 APPJ treatment can also effectively improve hydrophilicity. The increased hydrophilicity can improve the contact between the electrode and the gel-electrolyte, thereby improving the capacitance values.  Figure S1 shows SEM images of the rGO-SnO2 electrode at a magnification of 3000×. The surface of the as-deposited electrode is relatively smooth, and the rGOs cannot be identified clearly because too much ethyl cellulose is attached to their surface. This could degrade the EDLC provided by rGOs, resulting in poor SC performance. Upon increasing  Figure 4 shows the water contact angle of the rGO-SnO 2 electrode. The as-deposited rGO-SnCl 2 has a high-water contact angle of 120.9 • . After CO 2 APPJ processing, the water droplet penetrated into the carbon cloth instantly for all APPJ scanning times. The increased hydrophilicity is mainly due to the oxygen functional groups implanted by the APPJ. This is partly due to heat and partly due to the oxidizing nature of our CO 2 APPJ. This shows that the CO 2 APPJ treatment can also effectively improve hydrophilicity. The increased hydrophilicity can improve the contact between the electrode and the gel-electrolyte, thereby improving the capacitance values.

Water Contact Angles of rGO-SnO 2
This phenomenon could have further increased the number of oxygen species. As a result, the oxygen species were produced by the splitting of both CO2 and water vapor.  Figure 4 shows the water contact angle of the rGO-SnO2 electrode. The as-deposited rGO-SnCl2 has a high-water contact angle of 120.9°. After CO2 APPJ processing, the water droplet penetrated into the carbon cloth instantly for all APPJ scanning times. The increased hydrophilicity is mainly due to the oxygen functional groups implanted by the APPJ. This is partly due to heat and partly due to the oxidizing nature of our CO2 APPJ. This shows that the CO2 APPJ treatment can also effectively improve hydrophilicity. The increased hydrophilicity can improve the contact between the electrode and the gel-electrolyte, thereby improving the capacitance values.  Figure S1 shows SEM images of the rGO-SnO2 electrode at a magnification of 3000×. The surface of the as-deposited electrode is relatively smooth, and the rGOs cannot be identified clearly because too much ethyl cellulose is attached to their surface. This could degrade the EDLC provided by rGOs, resulting in poor SC performance. Upon increasing  Figure S1 shows SEM images of the rGO-SnO 2 electrode at a magnification of 3000×. The surface of the as-deposited electrode is relatively smooth, and the rGOs cannot be identified clearly because too much ethyl cellulose is attached to their surface. This could degrade the EDLC provided by rGOs, resulting in poor SC performance. Upon increasing the number of APPJ scans, the graphene sheets or stacks become more obvious, and most of the ethyl cellulose is burned off. Figure S2 shows SEM images of the rGO-SnO 2 electrodes at a magnification of 30,000×. A small number of SnO 2 particles is seen in the sample with one APPJ scan. After more than five APPJ scans, these particles are uniformly distributed on the rGO sheets. In our previous study of rGO SCs, no nanoparticle could be seen [57]. This indicates the successful  Figure 5 shows the SEM images of rGO-SnO 2 electrodes at a magnification of 100,000×. SnO 2 can be seen clearly; further, this SnO 2 contributes PC to SCs. the number of APPJ scans, the graphene sheets or stacks become more obvious, and most of the ethyl cellulose is burned off. Figure S2 shows SEM images of the rGO-SnO2 electrodes at a magnification of 30,000×. A small number of SnO2 particles is seen in the sample with one APPJ scan. After more than five APPJ scans, these particles are uniformly distributed on the rGO sheets. In our previous study of rGO SCs, no nanoparticle could be seen [57]. This indicates the successful synthesis of metal oxides in the study. Figure 5 shows the SEM images of rGO-SnO2 electrodes at a magnification of 100,000×. SnO2 can be seen clearly; further, this SnO2 contributes PC to SCs.  Figure 6 shows an XPS survey scan of the screen-printed electrode. Elemental identification was conducted based on the reference book [58]. The as-deposited electrode exhibits a little Si pollution. This Si may come from the screen-printing process. In addition, only C, O, Sn, and a little Cl can be detected. Table 1 shows the elemental ratio of C, O, Sn, and Cl. Higher C and lower Sn atomic contents are seen because a large amount of ethyl cellulose exists in the printed electrode. After CO2 APPJ processing, the C content apparently decreased, mainly owing to the removal of excess ethyl cellulose. Furthermore, the large increase in the O content indicates that the CO2 APPJ is fairly suitable for oxidizing carbon-based materials. The detectability of Sn increases after APPJ processing. Further, the Cl content decreases after APPJ processing. This demonstrates that the CO2 tornadotype APPJ can quickly decompose SnCl2 and convert it into SnO2.  Figure 6 shows an XPS survey scan of the screen-printed electrode. Elemental identification was conducted based on the reference book [58]. The as-deposited electrode exhibits a little Si pollution. This Si may come from the screen-printing process. In addition, only C, O, Sn, and a little Cl can be detected. Table 1 shows the elemental ratio of C, O, Sn, and Cl. Higher C and lower Sn atomic contents are seen because a large amount of ethyl cellulose exists in the printed electrode. After CO 2 APPJ processing, the C content apparently decreased, mainly owing to the removal of excess ethyl cellulose. Furthermore, the large increase in the O content indicates that the CO 2 APPJ is fairly suitable for oxidizing carbon-based materials. The detectability of Sn increases after APPJ processing. Further, the Cl content decreases after APPJ processing. This demonstrates that the CO 2 tornado-type APPJ can quickly decompose SnCl 2 and convert it into SnO 2 . Figure S3 shows the EDS analysis of the as-deposited and APPJ-processed electrode (seven scans). In contrast to the detection depth of XPS, which is less than 10 nm, EDS can be used to obtain chemical information from great depths in samples [59]. Figure S3a,b shows the EDS analysis of the as-deposited electrodes. The C content is 75.5% owing to the combined contribution from carbon cloth, rGOs, ethyl cellulose, and little organic solvent. The O content is contributed to by the ethyl cellulose, rGOs, and SnO 2 . The Sn and Cl signals arise from SnCl 2 . Figure S3c,d shows the XPS results obtained when the electrode was scanned seven times using the APPJ. The differences in C and O contents are small between the as-deposited and the seven-times APPJ-scanned samples. After seven APPJ scans, the C content increased from 75.5% to 77.9%, whereas the O content decreased from 20.0% to 17.1%. The O content may have decreased owing to the burnout of ethyl cellulose. Further, the Cl content decreased from 2.2% to 0.5%, clearly indicating the decomposition of SnCl 2 .  Table 1. Elemental ratios analyzed based on XPS survey scan spectra in Figure 6.  Figure S3 shows the EDS analysis of the as-deposited and APPJ-processed electrode (seven scans). In contrast to the detection depth of XPS, which is less than 10 nm, EDS can be used to obtain chemical information from great depths in samples [59]. Figure S3a-b shows the EDS analysis of the as-deposited electrodes. The C content is 75.5% owing to the combined contribution from carbon cloth, rGOs, ethyl cellulose, and little organic solvent. The O content is contributed to by the ethyl cellulose, rGOs, and SnO2. The Sn and Cl signals arise from SnCl2. Figure S3c-d shows the XPS results obtained when the electrode was scanned seven times using the APPJ. The differences in C and O contents are small between the as-deposited and the seven-times APPJ-scanned samples. After seven APPJ scans, the C content increased from 75.5% to 77.9%, whereas the O content decreased from 20.0% to 17.1%. The O content may have decreased owing to the burnout of ethyl cellulose. Further, the Cl content decreased from 2.2% to 0.5%, clearly indicating the decomposition of SnCl2.

CV Measurements
To confirm that the increased capacitance mainly arises from the deposited rGO-SnO2 and not from the carbon fibers of carbon cloth, bare carbon cloth SCs without rGO-SnO2  Table 1. Elemental ratios analyzed based on XPS survey scan spectra in Figure 6.

CV Measurements
To confirm that the increased capacitance mainly arises from the deposited rGO-SnO 2 and not from the carbon fibers of carbon cloth, bare carbon cloth SCs without rGO-SnO 2 coatings were fabricated. The CO 2 APPJ was also used for processing the carbon cloth before gel-electrolyte was spread on it. The areal capacitance is calculated using Equation (2) [23] as follows: where C A is the areal capacitance (mF/cm 2 ); S, the integral area of the total cyclic voltammetry loop; ∆V, the potential window (V); v, the potential scan rate (mV/s); and A, the area of each electrode. Figure S4 shows the CV measurement of the bare carbon cloth SC processed using the CO 2 APPJ. The cyclic curves are triangular owing to the EDLC of an ideal carbon SC [60]. Table S1 shows the relationship between the APPJ scan time and the areal capacitance value. The best performance can be obtained with seven APPJ scans. The improved areal capacitance mainly arises from the increased hydrophilicity. However, the capacitance is 4.48 mF/cm 2 , as evaluated under a potential scan rate of 2 mV/s; this value is much lower than that of rGO-SnO 2 processed using the CO 2 APPJ, as described below. Figure 7 shows the CV measurement of the rGO-SnO 2 SCs processed using the CO 2 APPJ. Compared with the CV curves of the bare carbon cloth SC ( Figure S4) and rGO SCs processed using the APPJ [61], the curve shapes are no longer squarish. This is attributable to the PC caused by the deposition of SnO 2 . A slight hump is detected at~0.3 V in the forward scan (0 to 0.8 V) and~0.2 V in the reverse scan (0.8 to 0 V); this small hump becomes clearer in the sample scanned seven times. Similar characteristics were noted in related studies [26,62]. These could be redox peaks caused by SnO 2 . The addition of SnO 2 apparently provided PC with EDLC, thereby further increasing the capacitance value. With a slower potential scan rate, a larger areal capacitance can be obtained. This is because ions can have more reaction time to intercalate the surface materials, thereby providing PC. As the potential scan rate increases, the charges may not have sufficient time to undergo redox Faradaic reactions. Therefore, the PC effect becomes less obvious, leading to a smaller capacitance value [63]. Table 2 shows the relationship between the areal capacitance value and the number of APPJ scans. The rGO-SnO 2 SC that was scanned seven times shows the optimal areal capacitance of 33.59 mF/cm 2 under a potential scan rate of 2 mV/s. Based on our previous study with Raman spectra, an overlong plasma process will damage the structure of graphene. It will reduce the capacitance value [8]. As a result, the areal capacitance decreased with the electrode APPJ scanned nine times.
In addition, the current change with the scan rate in the CV curve follows the powerlaw in Equation (3) [5].
where i is the response current at 0.3 V; v is potential scan rate; a and b are variable parameters. The b-values are calculated from log(i) vs. log(v). The electrodes with b = 1 are classified as EDLCs; the electrodes with 0.8 < b < 1 are corresponded to PCs; the electrode with 0.5 < b < 0.8 is attributed to battery-type behavior [64]. Figure S4d and Figure 7d show the power-law relationship with carbon cloth and rGO-SnO 2 SCs. In carbon cloth SCs, the b-values were 0.93~0.96, which is close to ideal EDLCs. In rGO-SnO 2 SCs, the b-values were 0.86~0.9, reflecting the behavior of PCs.

Trasatti Analysis
The capacitance contribution ratio, calculated as EDLC/PC, can be easily determined by the Trasatti method [65]. The Trasatti method divides the charge storage mechanisms of the surface charge (C out ), which is associated with EDLC, and the diffusion-controlled charge (C in ), which is proportional to v −1/2 (v is the potential scan rate) and is associated with PC. The two mechanisms have different responses to different potential scan rates [66]. When the potential scan rate approaches zero, both C in and C out readily respond to applied electric fields. When the potential scan rate approaches infinity, only C out responds to applied electric fields [67,68]. Figure S5 shows the Trasatti plots and capacitive contribution of carbon cloth SCs. Figure S5a,b shows the relationship between 1/C A (subscript A denotes the areal capacitance) and v 1/2 , which can be used to extrapolate the vertical axis (v 1/2 = 0) to find the C total = C in + C out . Figure S5c shows the relationship between C A and v −1/2 , which can be used to extrapolate to the vertical axis (v −1/2 = 0) to find C out . C in can then be calculated by subtracting C out from C total . Table S2 lists the calculated capacitive contribution, indicating that most of the electrochemical reaction of the carbon cloth SCs is dominated by EDLC. This is reasonable because carbon fibers contribute capacitance to the EDLC mechanism. The capacitive contributions of rGO-SnO 2 SCs are also calculated. Figure 8 shows the Trasatti plots of rGO-SnO 2 SCs. Table 3 shows C total , C in , and C out , as well as the capacitive contribution of rGO-SnO 2 SCs based on Trasatti analysis. The PC/EDLC ratio is~50% for all APPJ scan times. This confirms that the rGO-SnO 2 composites effectively improved the areal capacitance and provided both PC and EDLC. The improved PC mainly arises from SnO 2 [25]. EDLC mechanism. The capacitive contributions of rGO-SnO2 SCs are also calculated. Figure 8 shows the Trasatti plots of rGO-SnO2 SCs. Table 3 shows Ctotal, Cin, and Cout, as well as the capacitive contribution of rGO-SnO2 SCs based on Trasatti analysis. The PC/EDLC ratio is ~50% for all APPJ scan times. This confirms that the rGO-SnO2 composites effectively improved the areal capacitance and provided both PC and EDLC. The improved PC mainly arises from SnO2 [25].

GCD Measurements
The GCD measurement was used to evaluate the areal capacitance. The areal capacitance is calculated using Equation (4) [23] as follows: where C A is the areal capacitance (mF/cm 2 ); ∆V, the potential window (V); I, the constant discharging current (mA); T, the discharging time (s); and A, the area of each electrode. Figure S6 and Table S3 show the GCD results of carbon cloth SCs processed using the CO 2 APPJ. As the I-R drop would be large when the GCD current is too large, the GCD current was varied as 1, 0.5, 0.25, and 0.1 mA in this experiment. The charging/discharging curves are shaped like an isosceles triangle; this is characteristic of EDLC and confirmed the results of the CV results. The capacitance of the carbon cloth SC is much lower than that of the rGO-SnO 2 SC, as described below. This suggests that the rGO-SnO 2 coating processed using CO 2 tornado-type APPJ significantly improves the capacitance value. Figure 9 and Table 4 show the GCD results of rGO-SnO 2 SCs. The GCD results show an optimal capacitance of 37.17 mF/cm 2 with seven APPJ scans. The charging and discharging times are similar, indicating excellent Coulombic efficiency [69]. Finally, Figure S6e and Figure 9f show the areal capacitance of the carbon cloth and rGO-SnO 2 SCs under different charging/discharging currents. A smaller GCD charging/discharging current results in a larger areal capacitance owing to the response of the PC effect.

Ragone Plots
Ragone plots are used to evaluate the energy and power density of the SCs. There are, respectively, calculated with Equations (5) and (6) [70] as follows: where E A is the energy density (per unit area) (µWh/cm 2 ); C A , the areal capacitance calculated by the GCD method (mF/cm 2 ); ∆V, the potential window (V); P A , power density (per unit area) (mW/cm 2 ); and T, the discharging time (s). Figure 10 shows the Ragone plots of the SCs. With screen-printed rGO-SnO 2 and optimal CO 2 APPJ processing, the rGO-SnO 2 SC exhibits the best energy density of 3.304 µWh/cm 2 under a discharging current of 0.25 mA and the best power density of 1.067 mW/cm 2 under a discharging current of 4 mA. Compared with bare carbon cloth SCs, rGO-SnO 2 apparently promotes energy density. In comparison to the rGO-SnO 2 SCs processed using fixed-point nitrogen APPJ [49], the energy density of SCs processed by CO 2 tornado-type APPJ is nearly comparable. The slightly lower energy density value could be partly due to the temperature difference. Previously, the working temperature of our nitrogen APPJ was stably maintained at~500-600 • C; in contrast, the working temperature of the CO 2 tornadotype APPJ was 200-350 • C. We did not increase the operating temperature of the CO 2 tornado-type APPJ because the rotation and scanning modes of the plasma jet would make heat dissipate more easily forming the surface and thereby reducing the temperature. °C . We did not increase the operating temperature of the CO2 tornado-type APPJ because the rotation and scanning modes of the plasma jet would make heat dissipate more easily forming the surface and thereby reducing the temperature. SCs (our previous study [49]).

Electrochemical and Mechanical Stability of SCs
The 10,000-cycle CV stability was tested under a potential scan rate of 200 mV/s. Figure 11a shows the results. The rGO-SnO2 SC has an areal capacitance of 18.84 mF/cm 2 in the first cycle. After 10,000 cycles, its areal capacitance decreased to 15.86 mF/cm 2 . Therefore, the capacitance retention was 84.2%. Furthermore, the rGO-SnO2 SC was tested un- Figure 10. Ragone plots of rGO-SnO 2 . Comparison of results of CO 2 tornado-type APPJ-processed rGO-SnO 2 and bare carbon cloth SCs to those of fixed-point nitrogen APPJ-processed rGO-SnO 2 SCs (our previous study [49]).

Electrochemical and Mechanical Stability of SCs
The 10,000-cycle CV stability was tested under a potential scan rate of 200 mV/s. Figure 11a shows the results. The rGO-SnO 2 SC has an areal capacitance of 18.84 mF/cm 2 in the first cycle. After 10,000 cycles, its areal capacitance decreased to 15.86 mF/cm 2 . Therefore, the capacitance retention was 84.2%. Furthermore, the rGO-SnO 2 SC was tested under repeated bending with a bending radius of 7.5 mm (1000 times). Figure 11b shows the results of the bending test; the areal capacitance increased by 7.7% after 1000 bending cycles. This is because a small bending mechanical stress mainly facilitates improved contact between the electrolyte and the rGO-SnO 2 composites. Figure 10. Ragone plots of rGO-SnO2. Comparison of results of CO2 tornado-type APPJ-processed rGO-SnO2 and bare carbon cloth SCs to those of fixed-point nitrogen APPJ-processed rGO-SnO2 SCs (our previous study [49]).

Electrochemical and Mechanical Stability of SCs
The 10,000-cycle CV stability was tested under a potential scan rate of 200 mV/s. Figure 11a shows the results. The rGO-SnO2 SC has an areal capacitance of 18.84 mF/cm 2 in the first cycle. After 10,000 cycles, its areal capacitance decreased to 15.86 mF/cm 2 . Therefore, the capacitance retention was 84.2%. Furthermore, the rGO-SnO2 SC was tested under repeated bending with a bending radius of 7.5 mm (1000 times). Figure 11b shows the results of the bending test; the areal capacitance increased by 7.7% after 1000 bending cycles. This is because a small bending mechanical stress mainly facilitates improved contact between the electrolyte and the rGO-SnO2 composites.

Illumination of LED by Charged SCs
To verify whether the rGO-SnO2 SCs fabricated using CO2 tornado-type APPJ can actually be used as a power source for external electronic components, we use charged SCs to power up an LED. Toward this end, two rGO-SnO2 SCs were connected serially and charged by a 4.5 V DC power source for only 10 s. Then, the charged SCs were

Illumination of LED by Charged SCs
To verify whether the rGO-SnO 2 SCs fabricated using CO 2 tornado-type APPJ can actually be used as a power source for external electronic components, we use charged SCs to power up an LED. Toward this end, two rGO-SnO 2 SCs were connected serially and charged by a 4.5 V DC power source for only 10 s. Then, the charged SCs were connected to the LED. The LED can be lighted at least for 30 s by the charged SCs, as shown in Figure 11c and Video S1 (Supplementary Materials).

Conclusions
A CO 2 tornado-type APPJ was successfully used for converting rGO-SnCl 2 pastes into rGO-SnO 2 composites that were then used as the electrodes of flexible SCs. CO 2 tornadotype APPJ treatment could also improve the hydrophilicity of the electrode materials and facilitate contact between the gel-electrolyte and the electrodes. XPS and EDS validated the conversion of SnCl 2 into SnO 2 . The best-performing SC, processed using seven CO 2 tornado-type APPJ scans, exhibited an areal capacitance of 37.17 mF/cm 2 in the GCD test. Comparing with the results of bare carbon cloth SCs confirms that the contribution of the capacitance was mainly from CO 2 APPJ-processed rGO-SnO 2 . Trasatti analysis shows that the capacitance contribution ratio PC/EDLC is~50/50 with PC enhanced by the inclusion of SnO 2 . The SC shows a capacitance retention rate of 84.2% after a 10,000-cycle CV stability test. Further, no degradation was observed after 1000 bending cycles with a bending radius of 7.5 mm.  Figure S4: CV of bare carbon cloth SCs (without rGO-SnO 2 ) under potential scan rates of (a) 200 mV, (b) 20 mV, and (c) 2 mV. (d) Logarithm of currents and scan rates. Figure  S5: Trasatti plots of bare carbon cloth electrode SCs: (a,b) 1/C A vs. v 1/2 , (c) C A vs. v −1/2 , and (d) capacitive contribution, Figure S6: GCD of carbon cloth SC under constant current of (a) 0.10 mA, (b) 0.25 mA, (c) 0.50 mA, and (d) 1 mA. (e) Areal capacitances calculated based on GCD results, Table S1: Areal capacitance of bare carbon cloth SCs (without rGO-SnO 2 ), Table S2: Capacitive contribution of carbon cloth SCs, Table S3: Areal capacitance of carbon cloth SCs calculated by GCD. Video S1: LED lit using two serially connected rGO-SnO 2 SCs.