Steam and Carbon Dioxide Co-Activated Silicon Carbide-Derived Carbons for High Power Density Electrical Double Layer Capacitors

Different mico-mesoporous silicon carbide-derived carbons (SiC-CDC) were synthesized via gas phase chlorination at 1100 ◦ C and thereafter activated at 900 ◦ C and 1000 ◦ C with H 2 O steam using Ar and CO 2 as the carrier gases. The physical characterization data show that these materials are mainly amorphous, the structure does not change remarkably during the activation process and the surface chemistry of the differently activated and treated materials remains the same and there are no functional groups at the SiC-CDC surface. N 2 , Ar and CO 2 sorption measurements indicate an increase in the speciﬁc surface area and pore size distribution with increasing the activation temperature, whereas the inﬂuence of the carrier gas during synthesis is minimal. Although the speciﬁc surface areas and pore size distributions differed, the electrochemical parameters in 1 M (C 2 H 5 ) 3 CH 3 NBF 4 acetonitrile solution for all SiC-CDC materials were similar - speciﬁc gravimetric capacitances 130 ± 18 F g − 1 and volumetric capacitance 67 ± 14 F cm − 3 were calculated. Absolute phase angle values from − 85 ◦ to − 88 ◦ at low frequencies and very high energy and power densities 22 Wh kg − 1 at 20 kW kg − 1 and 12 Wh dm − 3 at 10 kW dm − 3 have been achieved.

Activated carbon materials are widely studied for several energy storage applications. In electrical-double layer capacitors (EDLC) different carbon materials are used for preparing the hierarchically micromesoporous electrodes. [1][2][3][4][5][6] The use of activated carbon electrodes in EDLC-s is taking advantages of its micro-mesoporous structure to reduce the electrode's electrical resistance and improve energy storage performance. 1 Organic and waste materials are quite often being used to prepare cost-effective carbon materials, but carbide-derived carbons (CDC) have an unique nanoporous structure with a narrow pore size distribution, a possibility to fine-tune the pore size and volume and also they have high electric conductivity. 1,[6][7][8][9] These properties make the CDC materials especially promising for energy storage/power generation applications. 2,4,8 To improve the carbon materials specific surface area and average pore size they are often activated by gas phase or liquid phase activation. Since gas phase activation enables cleaner production it is more favorable than liquid phase chemical activation. 9,10 Carbon dioxide and H 2 O steam are the most widespread activating agents due to the cost-efficiency and endothermic nature of their reactions which allows better process control. 1 Previously we have shown that gas phase activation with CO 2 of silicon carbide-derived carbon (SiC-CDC) resulted in doubling of the Brunauer-Emmett-Teller (BET) specific surface area and noticeable increase in pore size. 12 Due to that the electrochemical parameters where significantly enhanced. 12,13 Román et al. have reported that while CO 2 produces narrow micropores on the carbons and widens them as activation time is increased, steam yields pores of all the sizes from the early stages of the process and results in wider pore size distribution and more obvious development of macropores in the carbon materials. 9,11 Steam activation has been widely used before for different activated carbons, but it is not much studied on CDC materials. 2 Due to the above mentioned results the steam activation method seems promising also for our SiC-CDC materials.
In the present work SiC-CDC materials were activated with steam as well as by combination of steam+CO 2 to enhance the specific surface area and pore size of the materials to in turn also improve the electrochemical parameters of EDLCs using these for electrode materials. The relationship between activation and treatment parameters and the obtained activated carbon materials properties were studied.

Experimental
Preparation and activation of SiC-CDC materials.-The carbon materials used in the EDLC electrodes were made from different SiC-CDC materials that were synthesized from SiC powders (particle size 2 μm) via gas phase chlorination at 1100 • C. The synthesis procedures have been discussed in more detail earlier. 12 Thereafter the carbon materials were activated with steam at fixed temperatures, T, 900 • C and 1000 • C, for 2 hours. The saturated steam content in the carrier gas (Ar or CO 2 ) was 47% and the carrier gas flow rate was 50 ml min −1 . For the first materials group Ar and for the second materials CO 2 were used as the carrier gases. After activation half of each material was treated with H 2 at 900 • C for 1 hour to remove oxygencontaining functional groups from the surface of the porous carbon under study. The other half of the carbon powder was not treated further. This enables also the comparison of H 2 treated and non-H 2 treated materials. All synthesized materials with their abbreviations are given in Table I.

XRD, Raman, IR spectroscopy and XPS analysis.-XRD and
Raman spectroscopy measurements were carried out to investigate if any structural changes occurred during activation step. The XRD measurements were performed using CuKα radiation (45 kV, 35 mA, l = 0.154056 nm) with a step size 0.02 • of glancing angle θ and with the holding time of 2 s at fixed θ on Bruker D8 Advance diffractometer (Bruker Corporation). The diffraction spectra for the activated carbon materials were recorded at 25 • C and treated by the Topaz 4.0 software.
The Raman spectra for the activated CDCs were recorded using a Renishaw inVia micro-Raman spectrometer using Ar laser excitation (λ L = 514 nm).
The infrared spectroscopic measurements were performed using a PerkinElmer Spectrum GX FTIR equipped with a liquid nitrogencooled mid-range MCT detector. Measurements were performed in attenuated total reflection (ATR) mode using Harrick Meridian accessory with 3 mm diameter silicon ATR hemisphere, where infrared beam was directed at 45 degrees of incidence.
For detailed surface chemistry analysis the X-ray photoelectron spectra (XPS) were conducted in ultra-high vacuum condition using a surface station equipped with an electron energy analyzer (SCIENTA SES 100) and a non-monochromatic twin anode X-ray tube (Thermo XR3E2).
Surface area and pore size distribution measurements.-The porous structure of the activated carbon materials has been characterized using the low-temperature nitrogen sorption method. The N 2 In addition to nitrogen, sorption measurements were also performed with argon and carbon dioxide in order to see if there are any functional groups on the surface interacting differently with different gases (N 2 or Ar) and to get a better estimation of smaller pores with CO 2 testing gas. The sorption measurements with Ar (at −186.15 • C) and CO 2 (at 9.85 • C) were performed using the ASAP 2020 system (Micromeritics). Ar and N 2 are mostly suitable in case of micro-mesoporous materials with pore width d > 0.7 nm, providing reliable information on the surface available for charge storage in case of typical porous carbons. For materials with high proportion of ultramicropores, CO 2 adsorption has to be implemented for better estimation due to the fact that CO 2 can more easily access ultramicropores. 5,6,14 The specific surface area, S BET , and other parameters for the carbon materials were calculated according to Brunauer-Emmett-Teller (BET) theory (Table II). 15,16 The total volume of pores (V tot ) was obtained at the conditions near to saturation pressure p/p 0 = 0.97. The micropore area (S micro ) and micropore volume (V micro ) were calculated using the t-plot method. The pore size distributions have been calculated applying non-local density functional theory (NLDFT) to nitrogen adsorption isotherms within the relative pressure range from 10 −7 to 0.95 with new program Solution of Adsorption Integral Equation Using Splines (SAIEUS, Micromeritics). 16,17 Compared to previously introduced modifications of NLDFT, the current application provides smoother and more realistic shape of pore size distribution for different amorphous micro-mesoporous carbon materials successfully taking into account the surface heterogeneity of CDC carbons. 18 Electrochemical measurements.-The EDLC electrodes were composed of an aluminum current collector and a mixture of the activated carbon materials with 4% binder (PTFE, 60% dispersion in H 2 O). This mixture was laminated and roll-pressed (HS-160N, Hohsen Corporation, Japan) together to form a flexible layer of the active electrode material with thickness 100 ± 3 μm. After drying under vacuum, the pure Al layer (2 μm) was deposited onto one side of the CDC by the magnetron sputtering method. [19][20][21] The electrolyte used was prepared from pure acetonitrile (AN, H 2 O <20 ppm), and from dry (C 2 H 5 ) 3 CH 3 NBF 4 (Stella Chemifa). The two-electrode standard Al test cell (HS Test Cell, Hohsen Corporation) with two identical electrodes (geometric area of about 2.0 cm 2 ) was completed inside a glove box (Labmaster sp, MBraun; O 2 and H 2 O concentrations lower than 0.1 ppm) and all electrochemical experiments were carried out at temperature T = 20 • C. The 25 μm thick TF4425 (Nippon Kodoshi) separator sheet was used for mechanical separation of the working electrodes. Electrochemical characteristics of the EDLCs containing 1 M (C 2 H 5 ) 3 CH 3 NBF 4 acetonitrile solution have been studied by the cyclic voltammetry (CV), constant current charge/discharge (CC/CD) and the electrochemical impedance spectroscopy methods (EIS), using a SI1287 Solartron potentiostat and 1252 A frequency response analyzer over ac frequency (f) range from 1 mHz to 300 kHz at 5 mV modulation. The constant power (CP) method (using a BT2000 testing system Arbin Instruments, USA) has been used for obtaining the experimental Ragone plots.

Results and Discussion
Physical characterization data.-The XRD patterns (Fig. 1a) of the activated carbon materials show low intensity reflections, specifically (002) at 2θ∼26 • and (100)/(101) at 43 • , which refer to small graphitic areas but no significant long-range ordering in these materials. The (002) diffraction peak characterizes the existence of the parallel graphene layers, while the (100)/(101) peak is associated with the 2D in-plane symmetry along the graphite layers. 22 The XRD patterns for the different materials under study in this paper are almost identical, which indicates that no structural changes take place during the H 2 O steam activation process.
The Raman spectra for the synthesized materials ( Fig. 1b) are very similar to the CO 2 activated materials spectra demonstrated in previous work. 12 The spectra are characteristic of disordered amorphous carbon, demonstrating mainly two peaks in first order excitation region: the so-called graphite (G) peak at ∼1577 cm −1 and the disorderinduced (D) peak at ∼1388 cm −1 . The G-peak corresponds to the graphite in-plane bond-stretching motion of the pairs of C atoms in sp 2 configuration with E 2g symmetry. Thus, this mode does not require the presence of sixfold aromatic C rings, and it occurs at all sp 2 sites, not only for those atoms located in the hexagonal aromatic carbon structure. The D-peak is a breathing mode with A 1g symmetry, which is forbidden in perfect graphite and only becomes active in the presence of disorder in the graphite structure. 23 The spectrum also shows the second-order peak of D-band (2D) at ∼2670 cm −1 . The increase of the 2D peak intensities can be related with the crystallographic ordering of the graphitic structure. 24 The XRD and Raman data for 900 • C and 1000 • C SiC-CDC confirm that no significant structural changes take place and the difference between the materials activated at 900 • C Table II. Results of the sorption measurements for the synthesized materials.
St   Table I. and 1000 • C is very minimal, showing at 1000 • C only a slightly more ordered structure.
IR spectroscopy spectra are shown in Fig. 2a. It can be seen that all the materials have the same IR peaks and there is only a slight difference in the peak intensities. The same tendencies can also be seen on the XPS spectra (Fig. 2b). Therefore the results of both methods confirm that the surface chemistry of the differently activated (H 2 O or H 2 O+CO 2 ) and treated materials remains the same. It can be explained by the activation conditions -after activation the materials have been cooled down in an Ar atmosphere starting from high temperatures, where the oxygen containing functional groups are removed and the surface stays mostly free of them. Otherwise, there should be observed a different surface chemistry between the H 2 treated versus non-H 2 treated materials.
According to the data in the Table II and Fig. 3a the materials have mainly microporous structure. The H 2 treated materials are not separately shown in the figures, because they were almost identical to the non-H 2 treated versions, i.e. the H 2 treatment has a very minimal influence on the specific surface area and pore size distribution. The specific surface area and pore volumes are similar or slightly higher compared to the previously studied CO 2 activated materials. 12 The pore size distributions show that the materials are mainly microporous, but a small increase in mesoporous area with increasing the activation temperature from 900 • C to 1000 • C, can be observed. The comparison of sorption measurements data using N 2 , Ar and CO 2 is shown in Fig. 3b. As expected, the data for CO 2 measurement shows a bigger peak below 1 nm, i.e. for the smallest pores, whereas Ar and N 2 show more peaks within a bit wider range (Fig. 3b). These tendencies were same for all the material studied in this work. Also there is small dependence of the peak position on the gas used, which indicates that there might be specific adsorption interactions between the used gas and the carbon surface. However, the pore size distribution plots shape depends on the model used for fitting the adsorption data. Nevertheless, the comparison of N 2 , Ar and CO 2 adsorption confirms that the difference between H 2 treated and H 2 -untreated materials is negligible.
Cyclic voltammetry data.-The cyclic voltammetry (CV) curves expressed as specific gravimetric capacitance (C m ) and volumetric capacitance (C v ) vs. cell potential ( E) are presented in Figs. 4a-4f. The CVs were measured within various electrode potential regions at potential scan rates (ν) from 1 to 500 mv s −1 . In addition to the activated materials prepared and tested in this work, data for the initial un-activated carbon, denoted as SiC-CDC 1100 and the CO 2 activated material (with the best electrochemical properties) denoted as SiC-CDC 1100 A3, from our previous work, are shown for comparison. 12 The current density, j (obtained using the flat-cross section geometric surface area), measured at fixed scan rate has been used for  Table I.   Table I) and (b) differential pore size distribution vs. pore width plots obtained using the SAIEUS method 16,17 for different gases (N 2 , Ar, CO 2 ). calculation of the medium capacitance values according to Eq. 1: Eq. 1 is exactly correct if the capacitance, C, is constant (C = f( E)) and the series resistance R s → 0, if j → 0. Thus, Eq. 1 can be used to obtain the capacitance values only within the region of slow v if the values of current are small, as the ohmic potential drop, i.e. IR-drop, is negligible only under these conditions and the current response is essentially equal to that of a pure capacitor. 25,26 In a symmetrical two-electrode system the specific gravimetric capacitance C m (F g −1 ) for one activated ν carbon electrode can be obtained as follows: where m is the weight in g cm −2 per one activated carbon electrode, assuming that the positively and negatively charged electrodes have the same capacitance at fixed E. The volume of electrodes has been used for calculation of the volumetric capacitance values (Figs. 4c and 4f). All the activated carbon materials have nearly mirror image symmetry of the current responses about the zero current line, obtained at potential scan rates ν ≤ 500 mV s −1 and E ≤ 3.0 V (Figs. 4a-4f). Measuring up to 3.4 V cell potential the materials continue showing quite good capacitive behavior (Fig. 4e). Although the materials differ in their activation and treatment parameters and because of that also in their specific surface areas, their CV curves are very similar and for all carbon electrodes nearly ideal capacitive behavior can be observed. So, from the cyclic voltammetry data it can be seen, that the pore size distribution and specific surface area is having only moderate influence on the electrochemical parameters of the EDLCs. Compared to CO 2 activated material, the steam activated materials have slightly better capacitive behavior.
Specific gravimetric capacitance C m;CV and volumetric capacitance C v;CV values calculated from the linear region of the CV curves after changing the direction of the potential scan rate at cell potential 3.0 V are shown in Table III. The highest specific capacitance values C m;CV = 142 F g −1 and C v;CV = 81 F cm −3 were obtained for St-1000 material, but the difference between the values was not over 18 F g −1 or 14 F cm −3 .

Constant current charge/discharge data.-The
EDLCs were tested at constant current charge/discharge (CC/CD) regimes (at current densities from 0.05 to 1 A g −1 ) at the cell potentials from 0 to 3.0 V (Figs. 5a-5b). The charge-discharge curves for the activated materials are very slightly non-linear and non-symmetrical at moder-ate and high current densities (Figs. 5a-5b). Also it can be seen that the IR-drop is very negligible.
The discharge and charge capacitances, C cc , were calculated from the data of the third cycle. Due to the slight non-linearity of CC/CD plots the integration of CC and CD has been conducted and integral capacitance values have been calculated according to Eq. 3: where Q is the integrated charge density and E is the cell potential applied. Therefore, the medium integral capacitance C cc values calculated from CC/CD data are very slightly different than the C m;CV or C v;CV values calculated from CVs (Table III). Slightly lower capacitance values obtained by CC/CD method can be explained by physical differences in methods applied for charging/discharging of electrodes as well as by not totally established adsorption equilibrium of ions in meso-micropores. The cycling efficiency, i.e., the so-called round-trip efficiency (RTE) has been calculated as a ratio of charge (coulombic) and as a ratio of energy (energetic) released and accumulated during discharging and charging of the EDLCs. 27 The calculated coulombic efficiency values for all the systems remained within the range of 99.2 to 99.9% (charge at 1 A g −1 ) and energy efficiency values from 93 to 98% (charge from 1 A g −1 to 0.1 A g −1 respectively), showing that the steam activated SiC-CDC materials are nearly ideal materials for various energy/power storage applications.
Impedance spectroscopy data.-The impedance complex plane plots (i.e. imaginary part Z(Im) vs. real part Z(Re) dependences, also called as Nyquist plots) 6,28-30 for EDLCs based on the steam activated carbon electrodes in 1 M (C 2 H 5 ) 3 CH 3 NBF 4 acetonitrile solution have been measured within the range of ac frequency from 1 mHz to 300 kHz and at fixed cell potentials from 0 to 3.4 V (Fig. 6a). The shape of the Z'', Z' plots is similar for all the steam activated carbons, showing no significant influence of the different activation and treatment parameters on the capacitance values that have been observed. The Nyquist plots for the steam activated materials consist mainly of two parts: (1) of the so-called "porous" region with a slope of α nearly −45 • in ac frequency region f > 1 Hz, characteristic of the mass transfer limited process (with absorption boundary conditions) in the micro/mesoporous carbon electrode matrix of an electrode and (2) of the low frequency double-layer capacitance region with a slope of α ≈ −90 • ("knee" at f ≤ 1 Hz), obtained by the finite length absorption effect. 19,21,31,32 As can be seen, the shape for H 2 O activated SiC-CDC is slightly different from what was obtained for the CO 2 activated materials, which showed also a third region, i.e.  Table I) at potential scan rate 5 mV s −1 (a), at potential scan rate 50 mV s −1 (b), (volumetric capacitance) at potential scan rate 5 mV s −1 (c), at potential scan rate 500 mV s −1 (d) and for St-1000 at different potential scan rates specific gravimetric capacitance (e) and volumetric capacitance (f). a small semicircle at higher ac frequencies (f > 300 Hz). Based on systematical analysis of EDLC data, it is known that the semicircle shape depends on the adsorption kinetics of ions at the porous carbon electrode and on the series resistance of a material and mass transfer resistance inside a macroporous carbon structure. In addition, the very high frequency semi-circle can also be caused by the artifacts of the measurement system as well as by formation of a passive layer, covering the (positively charged) aluminum current collector, which introduces so-called electrostatic capacitor in series with the two seriesconnected double layer capacitors. [28][29][30][31][32] In the initial SiC-CDC 1100 (untreated) EDLCs system, with very high microporosity, but without mesopores in the carbon particles, the mass transfer limited processes prevail at f = 1 mHz and there is no ideal capacitive behavior even at f = 1 mHz. 12 The specific series capacitance values, C m , calculated from the Z'',Z' plots at ac frequency f = 1 mHz are shown in Table III and Fig. 6b. The obtained C m values are in the range from 123 F g −1 to 141 F g −1 , which is slightly higher than the values obtained for the CO 2 activated materials. 12 The C m values are in a good agreement with the values obtained using CV and CC methods.
Data in Fig. 6c show the phase angel θ vs ac frequency dependencies. The phase angle values for the steam activated materials are Table III. Calculated specific capacitance (gravimetric and volumetric) values from cyclic voltammetry ( E = 3.0 V; v = 5 mV s −1 ), constant current ( E = 3.0 V; j = 1 A g −1 ) and impedance spectroscopy measurements ( E = 3.0 V; f = 1 mHz).
) unless CC License in place (see abstract   Table I), respectively, for the EDLCs completed using CDC electrodes in 1 M (C 2 H 5 ) 3 CH 3 NBF 4 acetonitrile solution, obtained from constant power tests within the cell potential range from 1.5 V to 3.0 V (a, c) as well as from 1.7 V to 3.4 V (b, d).
within the range from −85 • to −87 • (f > 5 mHz). Small drop in the absolute phase angle value at low frequencies (to −80 • at f = 1 mHz) can be caused by some water residues and other electrochemically active components in the system, as well as by the non-equilibrium adsorption of ions into (ultra)micropores. Nevertheless the obtained values are again somewhat higher from these established for the CO 2 activated systems results.
Ragone plots.-The gravimetric (Figs. 7a, 7b) and volumetric (Figs. 7c, 7d) Ragone plots (specific energy, E, vs. specific power, P, dependencies) 28,33 calculated to the total material weight (except the mass of cases) or volume of two electrodes have been obtained from constant power test within the cell potential range from 1.5 V to 3.0 V (Figs. 7a, 7c) as well as from 1.7 V to 3.4 V (Figs. 7b, 7d). All the activated materials show again quite similar results and behavior. The Ragone plots were compared also with the CO 2 activated materials previously studied. Confirming the results from other methods here again, very high energy densities at high power densities have been observed, however independent of activation temperature and method applied. It is interesting to mention, that there is no difference between H 2 treated and H 2 non-treated materials. Thus, steam activation and Ar treatment simplifies the synthesis of SiC-CDC materials noticeably.

Conclusions
Different mico-mesoporous carbide-derived carbons (CDC) were synthesized from SiC powders via gas phase chlorination at 1100 • C and thereafter activated at 900 • C and 1000 • C with steam using Ar and CO 2 as the carrier gases. The physical characterization data from XRD and Raman spectroscopy show that these materials are mainly amorphous, but containing small graphitic crystallites. However, the crystallographic structure does not change remarkably during the steam activation process at 900 • C and 1000 • C. IR spectroscopy and XPS data confirm that the surface chemistry of the differently activated and treated materials remains the same and there are no functional groups at the SiC-CDC surface, due to the treatment at high temperature in Ar atmosphere. The low-temperature N 2 , Ar and CO 2 sorption measurements indicate an increase in the specific surface area and pore size distribution with increasing the activation temperature from 900 • C to 1000 • C, whereas the influence of the carrier gas during synthesis (Ar vs. CO 2 ) is minimal.
The electrochemical parameters were obtained using cyclic voltammetry, constant current charge/discharge, electrochemical impedance spectroscopy and constant power discharge methods. It has been suggested earlier that a large specific surface area of carbons is the most important parameter leading to a high capacitance values. However, the results of this paper clearly demonstrated that although the specific surface area increases significantly and also the pore size distribution expands, it has a very weak influence on the electrochemical parameters, which are quite similar for all the steam activated materials -specific gravimetric capacitances 130 ± 18 F g −1 or volumetric capacitances 67 ± 14 F cm −3 . Very high absolute phase angle values from −85 • to −88 • at low frequencies have been calculated indicating nearly ideally polarizable systems.
Very high energy and power densities (22 Wh kg −1 at 20 kW kg −1 and 12 Wh dm −3 at 10 kW dm −3 ) have been calculated for SiC-CDC steam activated materials based EDLCs.
However the comparison of current data with CO 2 activated materials based EDLCs from previous work shows that the steam activation is more effective, giving slightly better results with a shorter activation time. This can be explained by the higher reactivity of steam at high temperatures. The small differences between the results of steam activation and the steam+CO 2 activation shows that at higher temperatures both CO 2 and H 2 O compete for the same carbon reaction (active) sites.