Silicon oxycarbide-derived carbons from a polyphenylsilsequioxane precursor for supercapacitor applications

Microporous and Mesoporous Materials 188 (2014) 140–148

Contents lists available at ScienceDirect

Microporous and Mesoporous Materials journal homepage: www.elsevier.com/locate/micromeso

Silicon oxycarbide-derived carbons from a polyphenylsilsequioxane precursor for supercapacitor applications Andreas Meier a, Manuel Weinberger b, Katja Pinkert c,d, Martin Oschatz a, Silvia Paasch e, Lars Giebeler c,d, Holger Althues f, Eike Brunner e, Jürgen Eckert c,d, Stefan Kaskel a,f,⇑ a

Department of Inorganic Chemistry, Dresden University of Technology, D-01062 Dresden, Germany Electrochemical Materials Research, Center for Solar Energy and Hydrogen Research (ZSW), D-89073 Ulm, Germany Leibniz Institute for Solid State and Materials Research (IFW), D-01171 Dresden, Germany d Institute of Materials Science, Dresden University of Technology, D-01069 Dresden, Germany e Department of Bioanalytical Chemistry, Dresden University of Technology, D-01062 Dresden, Germany f Fraunhofer Institution for Material and Beam Technology (IWS), D-01277 Dresden, Germany b c

a r t i c l e

i n f o

Article history: Received 15 August 2013 Received in revised form 15 December 2013 Accepted 17 December 2013 Available online 27 December 2013 Keywords: Xerogel Silicon oxycarbide-derived carbon Supercapacitor Organic electrolyte

a b s t r a c t In this study, we report on the preparation of new silicon oxycarbide-derived carbons (SiOCDC) obtained by pyrolysis and chlorination of a polyphenylsilsesquioxane pre-ceramic precursor. Wet-chemical conversion of phenyltrimethoxysilane (PhTMS) to the organosilica material was conducted using a two-step acid/base sol–gel process in aqueous medium. The resulting material was subsequently pyrolysed at 700, 1000 and 1300 °C to obtain a non-porous silicon oxycarbide ceramic. Chlorination at 700 and 1000 °C led to carbons having large surface areas exceeding 2000 m2 g1 as well as large micro-/mesopore volumes up to 1.4 cm3 g1. The temperature of the thermal treatment significantly influences the carbon and final pore structure. Pyrolysis at 700 °C and subsequent chlorination at 700 °C led to a mainly microporous material, whereas pyrolysis at 1300 °C and subsequent chlorination at 1000 °C generated a hierarchically porous SiOCDC with micro- and mesopores, respectively. All SiOCDC materials were prepared as supercapacitor electrodes using an aqueous slurry containing polytetrafluoroethylene (PTFE) and sodium carboxymethyl cellulose (CMC) as binder. With an organic electrolyte (1 M TEABF4 in acetonitrile) capacitances of up to 110 F g1 and good long term stabilities could be observed. Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction Electrochemical energy storage devices become progressively more important for mobile and stationary applications. Up to now, batteries and supercapacitors complement each other. On the one hand, batteries show large energy densities (for example 120–200 Wh kg1 for modern lithium ion batteries [1]) but still suffer from slow charge/discharge characteristics and poor cycle life. In contrast, porous carbon based supercapacitors show at best 10% of the energy density of lithium ion batteries. Supercapacitors may roughly be distinguished into three groups [2,3]: True electrical double layer capacitors (EDLCs), pseudocapacitors, and lithium ⇑ Corresponding author at: Department of Inorganic Chemistry, Dresden University of Technology, Bergstrasse 66, 01062 Dresden, Germany. Tel.: +49 35146333632. E-mail addresses: [email protected] (A. Meier), Manuel. [email protected] (M. Weinberger), [email protected] (K. Pinkert), [email protected] (M. Oschatz), Silvia.Paasch@chemie. tu-dresden.de (S. Paasch), [email protected] (L. Giebeler), Eike.Brunner@ chemie.tu-dresden.de (E. Brunner), [email protected] (J. Eckert), Stefan. [email protected] (S. Kaskel). 1387-1811/$ - see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.micromeso.2013.12.022

ion hybrid capacitors. In EDLCs, the electrodes consist of porous carbons. Immersed into an electrolyte, ions will be adsorbed on the carbon surface to form an electrical double layer if a potential is applied. They may be charged and discharged in seconds for more than 1 million cycles. Many efforts have been made in recent years to develop novel carbon materials for energy storage applications [2–8]. Among them, carbide-derived carbons (CDCs) are interesting candidates. Because CDCs are a special class of highly microporous carbons with surface areas exceeding 2000 m2 g1 and thus they are suited well for several potential applications such as gas [9–11] and energy storage [6,12,13] as well as catalysis [14]. CDCs are obtained by high temperature chlorination of carbides. The metals or metalloids are extracted as gaseous chlorides leaving behind the highly porous carbon. Due to the layer-by-layer metal extraction a precise control of the porosity of the final carbon material is possible. Gogotsi and co-workers were the first to demonstrate the possibility to tune the micropore size for a Ti3SiC2 material [15]. By variation of the chlorination temperature between 200 and 1200 °C the pore size could be tailored from 0.3 to 0.8 nm, with almost a linear dependence between pore size and chlorination temperature. The

A. Meier et al. / Microporous and Mesoporous Materials 188 (2014) 140–148

possibility to tune the micropore size also allowed for a detailed investigation of the influence of pore size on double layer capacitance. By reducing the pore size to values below 1 nm an enormous increase in capacitance was obtained. Specific capacitance may exceed the barrier for activated charcoal of 100 F g1 to values well above 150 F g1 (with an organic electrolyte). The enormous increase in capacitance is best explained by a distortion of the electrolyte shell within small micropores (<1 nm) and thus a stronger adsorption of the ion centre to the pore surface [16]. According to the facile chlorination process it is possible to transform carbides in various shapes and nanostructures by maintaining the structural features. For instance, Chmiola et al. recently converted monolithic TiC plates with 300 lm thickness into the corresponding monolithic CDC. From this material a micro-supercapacitor device was fabricated which showed volumetric capacitances exceeding those of conventional CDCs by a factor of 2 [17]. Yushin and co-workers have recently demonstrated the conversion of a periodically mesostructured silicon carbide to give a hierarchically porous CDC where the initial pore walls of the SiC now contain a large micropore volume [18]. This material is promising as electrode material for supercapacitors since the mesopores allow fast ion transport throughout the particles. Thus these carbons demonstrate outstanding capacitance values at large currents. However, the synthesis of this material involves a nanocasting step with a rather expensive polycarbosilane. Therefore much cheaper routes have to be found for industrial applicability. Sol–gel processing is a facile, cheap and environmentallyfriendly tool for the preparation of various pre-ceramic materials in various shapes such as powders, films or monoliths [19]. Especially in the case of organosilanes, reaction kinetics are quite slow and allow precise control of hydrolysis and condensation reactions and so the final molecular structure, porosity and morphology. Pyrolysis of organosilica materials at temperatures up to 1000 °C leads to a rearrangement of silicon, carbon and oxygen bonds to form a complex network, where silicon is bound to oxygen and carbidic carbon and a free carbon phase is formed as a by-product. The final composition of the silicon oxycarbide strongly depends on factors such as the organosilane precursor itself, e.g. type and length of side chains, or the microstructure of the polymer [20–22]. Very recently, silicon oxycarbides were shown to be valuable CDC precursors. Gogotsi et al. pyrolysed and chlorinated precondensed polyorganosilsesquioxanes [23]. The resulting silicon oxycarbide-derived carbon (SiOCDC) is characterised by a hierarchical pore structure due to the presence of mesopores, which are probably obtained due to larger silica or carbon domains formed during pyrolysis. In the same work, excellent gas adsorption properties for hydrogen and methane were demonstrated. However, there was no report on the electrochemical properties. In this manuscript, the chlorination of sol–gel derived silicon oxycarbide materials and the influence of the preparation parameters on the performance of SiOCDC as supercapacitor electrodes are investigated in detail. In Fig. 1, the synthesis procedure is schematically demonstrated. Phenyltrimethoxysilane (PhTMS) is used as sol–gel precursor for the formation of SiOC materials via pyrolysis in inert atmosphere. PhTMS is one of the cheapest silanes available

141

and is hence suitable with respect to cost reduction for a scale-up. Different pyrolysis and chlorination parameters were investigated with regard to structural build up and pore formation as well as their influence on the electrochemical properties. It is also demonstrated that quite good long term stability can be reached for the resulting carbons when they are processed as EDLC electrode materials using an aqueous slurry containing PTFE and sodium carboxymethyl cellulose (CMC) as a novel binder. The mixture of both binding agents shows the best mechanical stability in comparison with the pure binders. Thus, a reduction of the amount of halogenated polymer in the electrode by partial substitution of polytetrafluoroethylene with carboxymethyl cellulose is possible. 2. Experimental 2.1. Materials The silicon oxycarbide derived carbons (SiOCDCs) were synthesised via pyrolysis and chlorination of a xerogel precursor (Fig. 1). The precursor was formed using a sol–gel process reported by Liu et al. [24]. The gels were produced in a two-step process at room temperature. In a typical synthesis approach, 99.15 g (0.5 mol) Phenyltrimethoxysilane (PhTMS; 94%+, Sigma Aldrich) were mixed with 72.06 g (4 mol) of water and then the pH value was adjusted to 2 by adding 1 M HCl (Sigma Aldrich). After 2 h of hydrolysis, a 2 M solution of NaOH (Grüssing GmbH) was added until the pH value reached 7. The gels formed were heated for 3 d at 80 °C in order to evaporate the solvent. The xerogels were pyrolysed under Ar atmosphere at the respective temperatures (700, 1000 and 1300 °C) forming silicon oxycarbides (SiOCs). In the heat treatment a ramp of 150 °C h1 is used. The final temperature during pyrolysis was held for 3 h in the tube furnace and then cooled to room temperature under Ar atmosphere. Subsequently, the oxycarbides were etched at temperatures of 700 or 1000 °C in a stream of 80 ml min1 Cl2 and 70 ml min1 argon followed by a treatment with H2 at 600 °C for 1 h to remove remaining chlorine in the samples forming SiOCDCs. 2.2. Electrode preparation In a typical electrode preparation, 90 mg of the active material and 5 mg of each binding agent were used. The as-made material was mixed with sodium carboxymethylcellulose (NaCMC; technical grade, VWR) and polytetrafluoroethylene (PTFE; 60 wt.% dispersion in water, Sigma Aldrich) in an aqueous solution resulting in a 5 wt.% concentration for each binding agent in the final electrodes. The mixture was homogenised for 1 h in a ball mill with a power of 30 W. The resulting slurry was cast on a carbon-coated aluminium foil from Exopack Advanced Coatings (UK) with a coating speed of 50 mm s1 using an electrical doctor blade (MTV Messtechnik OHG). The thickness of the wet film was adjusted to 120 lm. Round shaped electrodes with 10 mm diameter were blanked from the dried films and activated in a vacuum oven at 120 °C for more than 12 h.

Fig. 1. Scheme of SiOCDC synthesis.

142

A. Meier et al. / Microporous and Mesoporous Materials 188 (2014) 140–148

As a reference the standard activated carbon material YP-50F (Sanwa Components) was used to produce reference electrodes. 2.3. Characterisation The surface area and the pore structure of the synthesised porous materials were determined by nitrogen adsorption measurements at 196 °C with a Quantachrome Quadrasorb SI. The samples were degassed at 150 °C for more than 24 h. The surface areas were calculated using the BET (Brunauer–Emmett–Teller) equation in the relative p/p0 range from 0.07 to 0.3. Low pressure N2 physisorption with a Quantachrome Autosorb 1 was used to analyse the micropore system. Pore size distributions were plotted using the QS-DFT method in equilibrium state (Quenched Solid Density Functional Theory) under the assumption of slit and cylindrical pores. To determine the degree of graphitisation in the synthesised oxycarbide derived carbons Raman measurements (Renishaw 3000, Renishaw UK) were performed at an excitation wavelength of 514.5 nm. In the recorded spectra, the signals for the disordered D (1330–1345 cm1) and graphitic carbon G bands (1598– 1600 cm1) were each fitted with Lorentzian functions. The areas of the fitted peaks were used to calculate the ID/IG ratio. Thermal analysis was performed with the thermal analysis system Netzsch STA409 Luxx heating the samples from room temperature to 1000 °C under synthetic air. Elemental analysis was performed by cracking the sample via microwave treatment (Mars, Fa. CEM, 800 W, 20 min to 190 °C, 20 min at 190 °C) in HCl(conc.)/ HNO3(conc.)/HF(conc.): 3/1/1 vol.%. The silicon content was analysed with ICP-OES (iCAP 6500 Duo (Thermo Fisher Scientific), wavelength {diffraction order}: 251,611 {134} and 288,158 {117}). Elemental analysis by the carrier gas hot extraction methods was performed with respect to the oxygen concentration (TC436DR, Leco) and the carbon content (EMIA820V, Horiba). The morphology of the samples was characterised by scanning electron microscopy (SEM) and the composition was analysed by energy-dispersive X-ray spectroscopy (EDX) (Zeiss type DSM 982 GEMINI with a heat-able field emission tungsten cathode). The polyphenylsilsequioxane xerogel was sputtered first with a gold film to increase conductivity of the sample in SEM. The structure of the samples was analysed using a transmission electron microscope (TEM) TECNAI TEM type F 30 with Schottky field emission gun and SuperTwin lenses operated at an acceleration voltage of 300 kV. For the experiments, the samples were suspended in ethanol. The suspension was treated with ultrasound for one second and a drop was subsequently placed at a copper grid coated with a lacey carbon film. Solid-state NMR experiments were recorded on a Bruker Avance 300 spectrometer operating at 59.63 MHz for 29Si and 75.47 MHz for 13C using a commercial 4 mm MAS NMR probe. Ramped 1 H–13C cross-polarization (CP) [25,26] was applied (contact time: 4 ms). SPINAL 1H-decoupling [27] was used during signal acquisition. The MAS frequency was 14 kHz for all samples. The chemical shifts were referenced with adamantan for 13C and Q8M8 for 29Si. Electrochemical testing was performed with an IviumStat standard-version electrochemical interface and impedance analyzer (Ivium Technologies, NL). The electrodes were measured in a symmetric setup separated by a polypropylene membrane (CelgardÒ 2500, Celgard) in a SwagelokÒ cell. A volume of 150 ll of 1 M tetraethylammonium tetrafluoroborate (TEABF4; 99%, Alfa Aesar) in acetonitrile (anhydrous 99.8%, Sigma Aldrich) was used as electrolyte. The testing cells were prepared in a glovebox under Ar atmosphere. Specific capacitances of the systems were calculated from the discharge curves using Eq. (1):

C spec ¼

2I ðdV=dtÞ  m

ð1Þ

where I is the current used to discharge the system, while dV/dt is the slope of the discharge curve and m is the (arithmetic) average of both carbon electrode masses in grams. The supercaps were charged and discharged with constant currents at different current per mass ratios in a potential range of 0–2 V. The energy density of the electrode material was estimated under the assumption of Eq. (2):

E¼

C spec  V 2 1  1000ðg=kgÞ  ðW  h=JÞ 3600 2

ð2Þ

where Cspec is the gravimetric capacitance of the carbon material and V the operating voltage of the EDLC. The power density of the electrode was calculated by dividing the energy density by the discharge time at certain current densities. 3. Results and discussion Sol–gel processing of trimethoxyphenylsilane under aqueous conditions was performed according to a procedure described previously [24,28,29]. The precursor is first hydrolysed under acidic conditions at a pH of 2 in order to produce a clear sol (phenyltrimethoxysilane itself is not soluble under aqueous conditions) and to transform most of the methoxy groups into more reactive silanol groups. Condensation is greatly suppressed at pH = 2 since it is the point of zero charge for (organo)silica. Basification of the sol by addition of NaOH solution subsequently induces condensation and the formation of larger aggregates by consumption of the silanol groups. Polymerisation under basic conditions usually leads to the formation of larger and denser particles. Adsorption analysis of the purified and dried powder materials showed no significant surface area (<5 m2 g1) confirming the formation of a dense material on the nanoscale. Scanning electron microscopy (Fig. 2) revealed the formation of spherical particles with diameters up to 20 lm as well as very large and dense aggregates with dimensions of several hundred lm. This morphology may be explained by coalescence of the spherical particles during condensation and ageing [30–32]. More precisely, if the silane concentration is high, as in this case, the density of nuclei produced upon basification becomes high. For that reason, rapid coagulation may occur between nuclei, leading to more polydisperse spherical particles or even larger aggregates. In addition, polyphenylsilsesquioxanes show unique thermal behaviour due to a low glass transition temperature and thus tend to densify at higher temperatures. Ageing under basic conditions and at higher temperatures might lead to denser objects, which are frozen in at some point by condensation reactions. The high temperature conversion of polyphenylsilsesquioxanes was already investigated in literature [24,28]. Polyphenylsilsesquioxane is a well known and excellent pre-ceramic precursor for Si(O)C materials. Liu et al. have conducted studies concerning the temperature dependence of the pyrolysis of similar polyphenylsilsesquioxane precursors. They found that upon pyrolysis at 1000 °C a non-porous silicon oxycarbide was obtained. Interestingly, when the temperature was raised above 1000 °C, they observed the formation of porosity. Surface areas e.g. above 400 m2 g1 were achieved at 1500 °C [24]. The new approach presented here is found in the application of these oxycarbides as precursors for chlorination processes in order to generate hierarchically porous SiOCDC materials. Pyrolysis under argon was conducted either at 700, 1000 or 1300 °C (for sample notation see Table 1). However, with nitrogen adsorption experiments no significant surface areas (<6 m2 g1) were observed. As expected for pyrolysis temperatures at 700 and 1000 °C, samples were X-ray amorphous excluding the formation of larger amounts of silicon carbide or graphite crystallites. Also the samples prepared at

A. Meier et al. / Microporous and Mesoporous Materials 188 (2014) 140–148

143

Fig. 2. Scanning electron micrographs of (a) polyphenylsilsesquioxane, (b) sample SiOC-2, (c) sample SiOCDC-2 and (d) top view of SiOCDC-2 electrode with 5 wt.% CMC and 5 wt.% PTFE binder.

Table 1 Sample codes according to pyrolysis and chlorination temperature. Sample

Pyrolysis temperature [°C]

Chlorination temperature [°C]

SiOC-1 SiOC-2 SiOC-3 SiOCDC-1 SiOCDC-2 SiOCDC-3

700 1000 1300 700 1000 1300

n.a. n.a. n.a. 700 1000 1000

n.a. – Not applied.

1300 °C showed no signals in XRD measurements. The distribution and connection of C, Si and O bonds of silicon oxycarbides is very sensitive to the pyrolysis temperature. In fact, at temperatures above 1000 °C it is more likely that larger nanoscopic domains are formed due to crystallisation. All organosilica materials turned black during pyrolysis, indicating the formation of a free carbon phase, which is a typical byproduct [33,34]. SEM images of the pyrolysed samples clearly show, that the initial shape of the organosilica material may be maintained. The images still show spherical particles after pyrolysis as well as condensed material (Fig. 2).

Elemental analysis of the oxycarbides proves a nearly constant Si content of the ceramics formed at different temperatures (Table. 2). This content is in good agreement with the residual masses of the thermal analysis of the SiOCs assuming the formation of pure silica at an oxidation under synthetic air. The detectable carbon ratio in the ceramics increases with higher pyrolysis temperature of the oxycarbides. This might be according to a higher phase separation of the ceramics into silica and free carbon. Raman spectroscopy was performed in order to identify the carbon structures of the pyrolysed samples (Fig. 3). Sample SiOC-1 gave no reasonable scattering pattern due to immense background scattering. The huge luminescence background is typical for silicon oxycarbide materials which are synthesised at lower temperature and can be attributed to dangling bonds of the carbon atoms from the free carbon phase [35]. Pantano et al. already discussed the microstructure of polyphenylsilsesquioxane-derived SiOCs in detail. They state and confirm by solid state NMR that the molecular framework of samples pyrolysed below 1000 °C consists most

Table 2 Summary of elemental and thermal analysis data of silicon oxycarbides and SiOCDCs samples investigated. Sample

SiOC-1 SiOC-2 SiOC-3 SiOCDC-1 SiOCDC-2 SiOCDC-3

Elemental analysis

Thermogravimetry

Si [wt.%]

C [wt.%]

O [wt.%]

Weight loss [%]

27.3 27.1 28.4 4.2 0.2 0.2

27.8 35.9 43.6 85.4 94.0 98.5

21.0 22.2 24.5 6.6 0.6 0.5

38.70 35.85 32.73 95.96 100.00 100.00

Fig. 3. Raman spectra of pyrolysed PhTMS xerogels prepared at temperatures of 1000 °C (SiOC-2) and 1300 °C (SiOC-3) at an excitation wavelength of 514.5 nm.

144

A. Meier et al. / Microporous and Mesoporous Materials 188 (2014) 140–148

likely of a silica backbone where Si atoms on the surface are bond to terminal CH3 groups or polyaromatic species with sizes below 2.5 nm [33]. Pyrolysis at 1000 °C and above, however, will lead to a larger degree of graphitisation, observed by D (1330– 1345 cm1) and G bands (1598–1600 cm1) in the Raman spectrum (Fig. 3). The ID/IG ratios are 4.1 for sample SiOC-2 and 3.0 for SiOC-3. The peaks show similar FWHM values indicating minor differences of the electronic structure between the two carbon materials. The occurrence of such an intense disorder-induced D band together with a graphite G band which is shifted to values of 1600 cm1 is direct evidence for a more disordered distribution of nanocrystalline graphite domains [36]. For further discussions on the SiOCDCs it is thus necessary to keep in mind that there are graphite domains already present. These graphitic domains can also be observed by using transmission electron microscopy like in Fig. 9a shown for sample SiOC-3. The pyrolysis process has also been observed by solid-state 13C and 29Si NMR. The change in signals of the 29Si MAS NMR of the PhTMS xerogel as well as the pyrolysed samples is shown in Fig. 4. In the precursor xerogel the only peak present is the SiRO3/2 signal at – 80 ppm showing the formation of a silica backbone with phenyl substitution during the condensation reaction [34]. At 700 °C, the predominant species in the spectra is SiCO3/2, with a peak at – 77 ppm. According to redistribution reactions in the network, smaller signals of the species SiO4/2 at – 105 ppm and SiC2O2/2 at – 39 ppm are also evident. By increasing the pyrolysis temperature, the relative intensities of the peaks are shifted. Thus at 1000 °C the SiO4/2 peak is predominant, followed by SiCO3/2 and SiC2O2/2. At the final temperature of 1300 °C the decomposition of the oxycarbide species is almost complete, because the spectra mainly shows the signal of the silicon–oxygen tetrahedra SiO4 [36,37]. The 13C MAS spectrum of the PhTMS xerogel clearly shows the aromatic signal of the phenyl substitution with a peak at 131 ppm (Fig. 5). The spectra of the pyrolysed materials reveal the presence of free carbon in the form of polyaromatics. As the pyrolysis temperature is increased, a signal broadening appears according to the formation of disordered graphitic carbon [33,36]. For sample SiOC-3 pyrolysed at 1300 °C no 13C CP signal can be observed due to an extreme signal broadening and a decreased quantity of protons. For subsequent chlorination, we have chosen two temperatures, 700 and 1000 °C. The material pyrolysed at 700 °C was chlorinated at 700 °C. The samples pyrolysed at 1000 and 1300 °C were chlorinated at 1000 °C (for sample notation see Table 1). Fig. 6 shows nitrogen adsorption–desorption isotherms and corresponding QSDFT pore size distributions. Adsorption isotherms for samples SiOCDC-1 and SiOCDC-2 show the typical steep increase for adsorbed volume at very low relative pressure and a Type I isotherm (IUPAC classification) indicating the formation of highly microporous materials with specific surface areas of 1730 m2 g1 for SiOCDC-1 and 2480 m2 g1 for SiOCDC-2, respectively. Gogotsi et al. report on the effect that lower chlorination temperatures favour the formation of smaller pores [15]. In this case, pores with average diameters of about 1.1 nm for SiOCDC-1 and pores with diameters as large as 1.2 nm for sample SiOCDC-2 were obtained (see Table 3). A completely different picture is obtained for sample SiOCDC-3.

Fig. 4. 29Si NMR spectra of PhTMS xerogel and xerogels pyrolysed at 700 °C (SiOC1), 1000 °C (SiOC-2) and 1300 °C (SiOC-3).

Fig. 5. 13C CP NMR spectra of the PhTMS xerogel and the silicon oxycarbides synthesised via pyrolysis at 700 °C (SiOC-1) and 1000 °C (SiOC-2).

The higher pyrolysis temperature seems to have a significant effect on the microstructure of the silicon oxycarbide. Here the adsorption isotherm indicates the formation of additional mesopores according to a hysteresis at a relative pressure p/p0 of 0.4–0.5. The corresponding pore size distribution is bimodal, with small micropores (1.0 nm) and larger mesopores (3.7 nm). The examination of the carbon pore systems with low pressure nitrogen physisorption shows a micropore regime ranging down to pore sizes less than 0.5 nm (Fig. 7). Chlorination of silicon oxycarbides is a complex procedure due to the presence of many different species, namely the silicon oxycarbide phase itself, free carbon, silica domains as well as silicon carbide nanocrystals. In their recent work, Gogotsi et al. discuss this matter in a similar study [23]. From their point of view, the hierarchical pore formation during chlorination is due to SiC nanocrystals and the silicon oxycarbide domains. Micropores are thus generated by SiC, whereas pores in the range of 1–5 nm are achieved by the removal of the Si and O atoms from the silicon oxycarbide. Since the SiOC is an amorphous phase, the pore size distributions for all materials are rather broad.

Table 3 Calculated SSA, pore volume and average pore size of prepared SiOCDC samples.

a

Sample

BET [m2/g]

Pore volume [cm3/g] (<2 nm)

Total pore volume [cm3/g]

Average pore sizea [nm]

SiOCDC-1 SiOCDC-2 SiOCDC-3

1730 2480 2298

0.77 1.16 0.94

0.93 1.29 1.36

1.1 1.2 3.3

Average pore size is calculated with quenched solid density function theory.

A. Meier et al. / Microporous and Mesoporous Materials 188 (2014) 140–148

145

Fig. 6. (a) N2 physisorption isotherms recorded at 196 °C and (b) pore size distributions of the activated SiOCDC samples. (Filled symbols represent the adsorption branch, empty symbols the corresponding desorption).

Fig. 7. Estimated pore size distributions of low pressure N2 physisorption at 196 °C of SiOCDC samples prepared by different pyrolysis and chlorination temperatures.

Raman spectra of all three SiOCDC materials indicate significant differences in the microstructures of the samples (Fig. 8). Chlorination of the silicon oxycarbide at 700 °C leads to a Raman spectrum with a very intense G band together with a smaller D band (ID/IG ratio is 3.42). A similar scattering pattern was also observed by Nishihara et al. for zeolite templated carbons (ZTC) [38]. In their case, the small micropores of the zeolite template do not allow the formation of larger graphite domains during carbonisation. Instead, highly curved graphene sheets are formed. For the 700 °C sample a significant background scattering is observed which can still be assigned to luminescence and thus might be an indication that the initial carbon structure of the silicon oxycarbide matrix is less affected by the chlorination due to the relatively low temperature. The other

Fig. 8. Raman spectra of the SiOCDC samples synthesised at different temperature regimes at 514.5 nm (laser excitation wavelength).

samples are specified by a smaller ID/IG ratio as their silicon oxycarbides after chlorine etching indicating an increase in ordering. The chlorination at 1000 °C for sample SiOCDC-2 with an ID/IG ratio of 2.44 most probably leads to a higher degree of graphitisation and thus the sample loses a significant amount of the graphene-like structures. In contrast, sample SiOCDC-3 shows the lowest ID/IG ratio with 2.38. Here a formation of larger graphite nanodomains within the silicon oxycarbide matrix is expected. Transmission electron microscopy demonstrates that the SiOCDCs mostly consist of disordered carbon, while the samples pyrolysed and chlorinated at 1000 °C and above also exhibit graphitic domains (Fig. 9). These observations can also be confirmed by thermal analysis of the materials (Fig. 10). Combustion of SiOCDC-2 and SiOCDC-3 under atmospheric conditions up to 1000 °C starts at 550 °C and is residue-free suggesting the formation of pure carbons. In the differential thermal analysis of these materials two exothermic steps are observed indicating the presence of different forms of carbon. In the range of 480–760 °C amorphous carbon is oxidised, while at temperatures above graphite is combusted under atmospheric conditions. These observations are in good agreement with the thermal analysis of carbonaceous products by Soares et al. [39]. The thermal analysis of the SiOCDC synthesised at 700 °C shows a residual content of 4 wt.% after combustion. This residuum can be attributed to some silica left in the sample after chlorination as can be seen in elemental analysis (Table. 2). The beginning combustion of the sample at 480 °C can be explained with a higher degree of disordered carbon resulting in a higher reactivity under atmospheric conditions. This behaviour is supported by the differential thermal analysis data of SiOCDC-1 where no signal for the decomposition of graphitic structures has been detected. All SiOCDC samples are integrated into electrodes using an aqueous slurry containing a PTFE/CMC binder mixture. The use of solely PTFE dispersion results in coatings with a very weak adhesion to the current collector. Active material can easily be scratched off. If only CMC is used, coatings with good adhesion are obtained. These coatings are quite brittle and thus bending of the electrode sheets leads to the formation of cracks. The combination of both binders gives optimal results. All constituents are homogenised by ball milling and are subsequently deposited onto carbon-coated aluminium foil via doctor blade technique. The thickness of the active layers of the dried coatings is in the range of 60 lm. Fig. 2d shows an SEM image for sample SiOCDC-2 with 5 wt.% CMC and PTFE, respectively. Although the starting material consists of rather large particles, the coating process clearly demonstrates a quite homogenous layer. The ball milling conditions produce smaller particles with average particle dimensions of about 1–2 lm.

146

A. Meier et al. / Microporous and Mesoporous Materials 188 (2014) 140–148

Fig. 9. Transmission electron microscopy of (a) SiOC-3, (b) SiOCDC-1, (c) SiOCDC-2 and (d) SiOCDC-3. (Arrows indicate graphitic domains).

Fig. 10. (a) Thermogravimetric and (b) differential thermal analysis of the investigated oxycarbide derived carbons.

For electrochemical experiments, a binder composition of 5 wt.% for each component (PTFE and CMC) is used. Fig. 11a shows results from galvanostatic charge/discharge tests vs. discharge current. For low current densities (0.6 A g1) specific capacitances of up to 109 F g1 for sample SiOCDC-2 are achieved, being 10% higher than the value of 98 F g1 of the reference electrode. With increasing specific current the specific capacitance decreases significantly (Fig. 11a). From 0 to 30 A g1, sample SiOCDC-1 and sample SiOCDC-2 drops by 10% and 21%, respectively, whereas sample SiOCDC-3 looses 30%. The idea of fast ion transport through a hierarchical pore system and thus an improved rate capability seems not to work for this sample. Actually, sample SiOCDC-3 should demonstrate the best stability over the measured current range. However, in our recent study sample SiOCDC-1, with a microporous pore system, shows better capacitance retention at higher

current densities. This phenomenon seems to have another origin which is discussed within the dependency of rate capability from the overall electrical conductivity in electrochemical storage materials. As already discussed in the Raman section we conclude from our investigations that sample SiOCDC-1 consists of a larger amount of sp2-hybridised carbon atoms that should lead to superior conductivity compared to the other samples. Excellent conductivity of similar nanostructures in zeolite-templated carbons was already reported by Nishihara et al. [40] They have recently investigated the electrochemical performance of zeolite-templated carbons with uniform pores of 1.2 nm in diameter. The group compares the rate capability of these carbons with standard activated charcoals and show that the zeolite-templated materials show superior performance at large currents, as explained by the special pore geometry, consisting of three-dimensional and

A. Meier et al. / Microporous and Mesoporous Materials 188 (2014) 140–148

147

Fig. 11. Electrochemical results of silicon oxycarbide-derived carbons prepared via doctorblade technique and measured in sandwich setup, (a) specific capacitance vs. current density (each point has an error bar between 1 and 2 F g1) and (b) cyclic voltammograms at a sweep rate of 100 mV s1.

mutually connected nanopores. A facilitated ion transport throughout their material is discussed and they state that a diameter of 1.2 nm could be optimal for supercapacitor application. In contrast to the conventional carbons, the molecular structure mainly consists of graphene-like structures and less graphitic fractions significantly improving the conductivity of the material. Despite the highly ordered pore geometry in ZTCs, the material SiOCDC-1 may consist of similar molecular structural motifs, a potential explanation for the slightly better performance compared to the other SiOCDC samples. The fact that the purely microporous material SiOCDC-1 shows the lowest specific capacitance of the measured test cells is explained by a more in depth PSD analysis of the sample. The major part of the SiOCDC-1 pore sizes lies in the region of the micro- and ultramicropores. These small pores are inaccessible for the electrolyte ions during the charging process of the capacitor leaving electrolyte-inaccessible pores which can not contribute to the capacitance of the system [41]. While the carbon synthesised via pyrolysis and chlorination of the ceramic PhTMS-precursor at 1000 °C has the best overall capacitance of all samples. The PSD of SiOCDC-2 shifts to slightly larger pore sizes than SiOCDC-1 and seems to perfectly fit to the sizes of the ion pair of the electrolyte (tetraethylammonium cation and tetrafluoroborate anion) [16]. According to Eq. (3):

C¼

eA d

ð3Þ

the capacitance C of a porous system with constant permittivity e is proportional to the electrode surface A that is accessible to electrolyte ions and the separation d between electrolyte ions and charged carbon surface. The SiOCDC prepared at temperatures of 1000 °C and above demonstrates a good retention of the specific capacitance resulting in good energy densities (33.3 Wh kg1) at very high power densities up to 310 kW kg1. Fig. 12 shows the Ragone plot of energy density versus power density of the presented carbon materials. Especially, samples SiOCDC-3 and SiOCDC-2 are specified by higher energy and power characteristics than the reference activated carbon material YP-50F. Also a clear trend is observed in cyclic voltammetry measurements. In Fig. 11b, cyclic voltammograms for all three samples measured at a scan rate of 100 mV s1 are illustrated. All three samples show rectangular voltammograms and no peaks due to faradaic redox reactions are visible. This curve is typical for an ideal capacitor and shows the reversibility of the energy storage process. Concerning the slope at the inversion potentials there is a significant difference between the samples. The curve of sample SiOCDC-1 is described by the steepest slope at ±2 V, followed by the sample SiOCDC-2 and sample SiOCDC-3 with the smallest slope. The behaviour at the inversion potentials is directly related to

Fig. 12. Ragone plot of energy density vs. power density of the prepared SiOCDC samples.

the inner resistance of the electrode and thus determines the rate of supercapacitor discharge. Conway et al. show by comparing theoretical calculations and real measurements that the shapes of the CV curves are directly related to internal resistances in the system [42]. Also the cyclic voltammograms of the oxycarbide derived carbons SiOCDC-1 and SiOCDC-2 are characterised by a butterfly shape, which is a common form for EDLCs in organic electrolytes. Salitra et al. attribute this observation to the potential dependency of the ion penetration into the pores on the nanoscale [43]. Cycling stability tests at a charge/discharge current of 5 A g1 for the different electrodes show a relation between the stability of the electrode and the temperature regimes used for the synthesis of the SiOCDCs (Fig. 13). With upwarding pyrolysis

Fig. 13. Development of the specific discharge capacity over 10,000 charge/ discharge cycles.

148

A. Meier et al. / Microporous and Mesoporous Materials 188 (2014) 140–148

temperatures from 700 to 1300 °C the material increases its electrochemical stability. At preparation temperatures of 1300 °C the electron scattering defects are reduced to a minimum and more graphitic domains are formed resulting in an improved stability during charging and discharging. The proof of the previous assumption is given by the sample SiOCDC-3. Synthesised at 1300 °C pyrolysis temperature and 1000 °C chlorination temperature, it shows the best stability over 10,000 charging/discharging cycles with retention for the specific capacitance of 95%. The capacitance of the carbon materials SiOCDC-2 and SiOCDC-1 drop of about 12% and 21%, respectively. They show an equal or lower stability compared to the reference with a decrease of 10%. In conclusion, silicon oxycarbide glasses were prepared and converted into carbon materials via pyrolysis and chlorination of phenyltrimethoxysilane xerogels at different temperatures. The route successfully demonstrates the potential of organosilica materials as pre-ceramic precursors for the preparation of SiOCDC materials. Sol–gel processing of organosilanes allows facile control of various material properties such as porosity or morphology. Since these features are preserved upon pyrolysis to obtain the SiOC materials, similar is found for the chlorination step. Sol–gel derived precursors can serve as a great opportunity to prepare SiOCDCs with complex nanostructures and morphologies such as mesostructured CDCs, porous thin films or hierarchically porous monoliths. Phenyltrimethoxysilane itself is a cheap precursor and therefore the facile sol–gel process has higher potential for scaleup than other hard-templated carbon materials. Moreover, it was possible to prepare large-scale electrodes with doctor blade technique using an aqueous slurry of the active material in combination with the binding agents. The SiOCDC-2 electrode shows a specific capacitance of 86 F g1 at a current density of 30 A g1, which is higher than the activated carbon YP-50F with 74 F g1, which is a standard carbon material for supercap application. Also it was possible to show that the carbon structure and therefore the electronic behaviour of the oxycarbide derived carbon can be controlled with the pyrolysis and chlorination temperature. The supercap electrodes cast with SiOCDC are characterised by high power densities up to 310 kW kg1 and a good long-term stability with a capacitance retention of up to 95%. Thus they are promising materials for application in supercapacitors. Acknowledgements The authors would like to thank Ronny Buckan, Andrea Voss, Heike Bußkamp, Cornelia Geringswald and Wolfgang Gruner for the chemical bulk analysis and Beate Leupolt for the Raman measurements. This work has been financially supported by the European Union (ERDF) and the Free State of Saxony (SAB) in the framework of the European Centre for Emerging Materials and Processes (ECEMP), Grant-Nr. 100111670. References [1] A. Vezzini, Phys. Unserer Zeit 41 (2010) 36–42. [2] G. Wang, L. Zhang, J. Zhang, Chem. Soc. Rev. 41 (2012) 797–828.

[3] P.J. Hall, M. Mirzaeian, S.I. Fletcher, F.B. Sillars, A.J.R. Rennie, G.O. Shitta-Bey, G. Wilson, A. Cruden, R. Carter, Energy Environ. Sci. 3 (2010) 1238–1251. [4] E. Frackowiak, F. Béguin, Carbon 39 (2001) 937–950. [5] J. Gamby, P.L. Taberna, P. Simon, J.F. Fauvarque, M. Chesneau, J. Power Sources 101 (2001) 109–116. [6] P. Simon, Y. Gogotsi, Nat. Mater. 7 (2008) 845–854. [7] A.S. Aricò, P. Bruce, B. Scrosati, J.-M. Tarascon, W. van Schalkwijk, Nat. Mater. 4 (2005) 366–377. [8] A. Thomas, P. Kuhn, J. Weber, M.-M. Titirici, M. Antonietti, Macromol. Rapid Commun. 30 (2009) 221–236. [9] G. Yushin, R. Dash, J. Jagiello, J.E. Fischer, Y. Gogotsi, Adv. Funct. Mater. 16 (2006) 2288–2293. [10] M. Oschatz, E. Kockrick, M. Rose, L. Borchardt, N. Klein, I. Senkovska, T. Freudenberg, Y. Korenblit, G. Yushin, S. Kaskel, Carbon 48 (2010) 3987–3992. [11] V. Presser, J. McDonough, S.-H. Yeon, Y. Gogotsi, Energy Environ. Sci. 4 (2011) 3059–3066. [12] V. Presser, M. Heon, Y. Gogotsi, Adv. Funct. Mater. 21 (2011) 810–833. [13] J. Chmiola, G. Yushin, R. Dash, Y. Gogotsi, J. Power Sources 158 (2006) 765– 772. [14] P. Krawiec, E. Kockrick, L. Borchardt, D. Geiger, A. Corma, S. Kaskel, J. Phys. Chem. C 113 (2009) 7755–7761. [15] Y. Gogotsi, A. Nikitin, H. Ye, W. Zhou, J.E. Fischer, B. Yi, H.C. Foley, M.W. Barsoum, Nat. Mater. 2 (2003) 591–594. [16] J. Chmiola, G. Yushin, Y. Gogotsi, C. Portet, P. Simon, P.L. Taberna, Science 313 (2006) 1760–1763. [17] J. Chmiola, C. Largeot, P.-L. Taberna, P. Simon, Y. Gogotsi, Science 328 (2010) 480–483. [18] Y. Korenblit, M. Rose, E. Kockrick, L. Borchardt, A. Kvit, S. Kaskel, G. Yushin, ACS Nano 4 (2010) 1337–1344. [19] C.J. Brinker, G.W. Scherer, Sol–Gel Science: The Physics and Chemistry of Sol– Gel Processing: The Physics and Chemistry of Sol–Gel Processing, Academic Press, 1990. [20] F. Babonneau, L. Bois, C.Y. Yang, L.V. Interrante, Chem. Mater. 6 (1994) 51–57. [21] G.D. Sorarù, Q. Liu, L.V. Interrante, T. Apple, Chem. Mater. 10 (1998) 4047– 4054. [22] M. Weinberger, S. Puchegger, T. Fröschl, F. Babonneau, H. Peterlik, N. Hüsing, Chem. Mater. 22 (2010) 1509–1520. [23] C. Vakifahmetoglu, V. Presser, S.-H. Yeon, P. Colombo, Y. Gogotsi, Microporous Mesoporous Mater. 144 (2011) 105–112. [24] C. Liu, H.Z. Chen, S. Komarneni, C.G. Pantano, J. Porous Mater. 2 (1995) 245– 252. [25] G. Metz, X.L. Wu, S.O. Smith, J. Magn. Reson. 110 (1994) 219–227. [26] A. Pines, M.G. Gibby, J.S. Waugh, J. Chem. Phys. 59 (1973) 569–590. [27] B.M. Fung, A.K. Khitrin, K. Ermolaev, J. Magn. Reson. 142 (2000) 97–101. [28] P. Gupta, W. Wang, L.-S. Fan, Ind. Eng. Chem. Res. 43 (2004) 4732–4739. [29] K. Takahashi, K. Tadanaga, A. Matsuda, A. Hayashi, M. Tatsumisago, J. Ceram. Soc. Jpn. 115 (2007) 131–135. [30] L.A.S.D.A. Prado, E. Radovanovic, H.O. Pastore, I.V. Yoshida, I.V.P. Torriani, J. Polym. Sci. A Polym. Chem. 38 (2000) 1580–1589. [31] I.-S. Seog, C.H. Kim, J. Mater. Sci. 28 (1993) 3277–3282. [32] K. Takahashi, K. Tadanaga, A. Matsuda, A. Hayashi, M. Tatsumisago, J. Sol–Gel Sci. Technol. 41 (2007) 217–222. [33] C.G. Pantano, A.K. Singh, H. Zhang, J. Sol–Gel Sci. Technol. 14 (1999) 7–25. [34] H. Zhang, C.G. Pantano, J. Am. Ceram. Soc. 73 (1990) 958–963. [35] G.D. Sorarù, G. D’Andrea, R. Campostrini, F. Babonneau, G. Mariotto, J. Am. Ceram. Soc. 78 (1995) 379–387. [36] F.I. Hurwitz, P. Heimann, S.C. Farmer, D.M. Hembree, J. Mater. Sci. 28 (1993) 6622–6630. [37] R.J.P. Corriu, D. Leclercq, P.H. Mutin, A. Vioux, J. Sol–Gel Sci. Technol. 8 (1997) 327–330. [38] H. Nishihara, Q.-H. Yang, P.-X. Hou, M. Unno, S. Yamauchi, R. Saito, J.I. Paredes, A. Martínez-Alonso, J.M.D. Tascón, Y. Sato, M. Terauchi, T. Kyotani, Carbon 47 (2009) 1220–1230. [39] R.W. Soares, V.J. Menezes, M.V.A. Fonseca, J. Dweck, J. Therm. Anal. Calorim. 49 (1997) 657–661. [40] H. Itoi, H. Nishihara, T. Kogure, T. Kyotani, J. Am. Chem. Soc. 133 (2011) 1165– 1167. [41] H. Shi, Electrochim. Acta 41 (1996) 1633–1639. [42] W.G. Pell, B.E. Conway, J. Power Sources 96 (2001) 57–67. [43] G. Salitra, A. Soffer, L. Eliad, Y. Cohen, D. Aurbach, J. Electrochem. Soc. 147 (2000) 2486–2493.