Self-supporting carbon-rich SiOC ceramic electrodes for lithium-ion batteries and aqueous supercapacitors

Fabrication of precursor-derived ceramic fibers as electrodes for energy storage applications remains largely unexplored. Within this work, three little known polymer-derived ceramic (PDC)-based fibers are being studied systemically as potential high-capacity electrode materials for electrochemical energy devices. We report fabrication of precursor-derived SiOC fibermats via one-step spinning from various compositions of siloxane oligomers followed by stabilization and pyrolysis at 800 °C. Electron microscopy, Raman, FTIR, XPS, and NMR spectroscopies reveal transformation from polymer to ceramic stages of the various SiOC ceramic fibers. The ceramic samples are a few microns in diameter with a free carbon phase embedded in the amorphous Si–O–C structure. The free carbon phase improves the electronic conductivity and provides major sites for ion storage, whereas the Si–O–C structure contributes to high efficiency. The self-standing electrodes in lithium-ion battery half-cells deliver a charge capacity of 866 mA h gelectrode−1 with a high initial coulombic efficiency of 72%. As supercapacitor electrode, SiOC fibers maintain 100% capacitance over 5000 cycles at a current density of 3 A g−1.


Introduction
Despite the exponential rise in research activity in the design and development of micro-/nano-structured electrode materials for electrochemical energy storage devices, 1-4 graphite or carbon-coated metal foil remain the electrode of choice for most capacitors and Li-ion battery (LIB) technologies. Materials with higher charge storage than traditional carbons such as silicon [almost 10Â lithium (Li) storage capacity ($4.2 A h g À1 )] emerged as potential replacements for graphite owing to silicon's large abundance in the earth's crust, low toxicity, and wellestablished manufacturing technology for its large scale production. [5][6][7] However, drastic volume variation and pulverization during charge/discharge, excessive formation of solid electrolyte interfaces (SEIs), and stress-induced cracks in Si electrodes has rendered bulk silicon unusable as an electrode for LIBs. 6,8 Efforts related to the development and use of nanostructured silicon-based electrodes (utilizing the size effect to achieve fast Li-ion transport, and prevent fracture propagation in electrodes), in the form of nanoparticles,; 9-12 nanotubes and nanowires, [13][14][15][16][17] and composite nanoparticles with carbon [18][19][20][21][22][23] are now well-documented. Nevertheless, capacity decay over a long period of charge/discharge, high rst cycle loss, high cost of material synthesis, and challenges associated with scalability limit their applications as electrodes for commercial applications.
To circumvent the issues associated with silicon electrodes, silicon oxycarbide (SiOC) derived from pyrolysis of ceramic precursor polymers has renewed attention in recent years from electrochemical storage point of view. [24][25][26][27][28][29][30][31][32][33] Potential of polymerderived ceramic (PDC) (e.g.; SiOC, SiCN) materials in energy storage, initially proposed by Dahn et al. in 1990s, largely depends on the chemical structures of the material. 34,35 SiOC is an amorphous ceramic mainly consists of a Si-O-C glass phase with a free carbon region, where the mixed bonds of Si with O and C enable high Li storage. 36,37 According to Graczyk-Zajac et al., mixed bonds in SiOC allows more disordered carbon phase for higher Li capacity (325 mA h g À1 ) as well as reversible storage sites at the interface of the carbon phase and amorphous network. 38 Kasper et al. showed that the electrical conductivity of SiOC samples improved (0.07 to 2.2 S m À1 ) with the increasing amount of free carbon phase, in addition to enabling more active sites for Li. 39 The edges of the free carbon phase and the interstitial spaces are major electrochemically active sites for Li ions, whereas micro-/nano-voids and amorphous Si-O-C network contribute to minor Li storage. 25 from PSS-octakis(dimethylsilyloxy) silsesquioxane (POSS) provided efficient ion pathway and withstood structural degradation by buffering during lithiation/ delithiation. 33 As a result, the authors were able to achieve a capacity of 412 mA h g À1 at a high current density of 3600 mA g À1 . Furthermore, pyrolysis parameters are always crucial in controlling the microstructure (e.g.; free/disordered carbon phase and nanovoids) and thus maintaining the electrochemical properties of SiOC. For example, pyrolysis temperature above 900 C results in lower capacity and unstable cycling behavior of the SiOC ceramics compared to lower temperature. 29,42 With increasing temperature disordered carbon phase arranges into ordered structure and amorphous Si-O-C phase decomposes to electrochemically inactive SiC crystals. In addition, metal foil-based electrode, conventionally prepared using a doctor blade, carries the inactive weight of polymeric binders, conducting agents (usually carbon black), and the foil itself, which do not contribute towards electrode capacity. 31 Since the development of electrospinning in 1930s by Formhlas, 43,44 the technique has been explored to fabricate a variety of articial bers from homogenous polymer solution using high potential electric eld. Nanobers produced by electrospinning can provide large surface area, exceptional tolerance against mechanical deformation with improved electrical conductivity and enhanced electrochemical performance. For example, electrospinning of PAN nanobers with Si nanoparticles delivered a exible electrode aer carbonization that showed a high capacity of 1600 mA h g À1 in LIBs. 45 Xiao et al. reported a stretchable PVDF nanober membrane which was coated with Ni and Si to form a Si/Ni/PVDF core-shell structure. 46 This so silicon electrode showed 56.9% capacity retention aer 1000 cycles with 20% stretching capability.
Herein, we report on the fabrication of freestanding, binderfree SiOC electrodes by an electrospinning method using liquid precursor polymer, followed by thermal treatment and pyrolysis. Three new types of SiOC precursors, that are readily available, and easy to draw into low-cost, high-yield ceramic bers compared to commercial ceramic bers, 47,48 are utilized in this study both for LIBs and supercapacitors. To the best of our knowledge such siloxanes have not been reported before for the fabrication of SiOC ceramic bermats, that are investigated in electrochemical devices. Scanning-and transmissionelectron microscopy reveals uniform, defect-free amorphous structure of the SiOC bers. Fourier transform infrared spectroscopy (FTIR) and nuclear magnetic resonance (NMR) are utilized to study the polymeric precursor to ceramic transformation of the siloxanes, whereas Raman spectroscopy and Xray photoelectron spectroscopy (XPS) conrms the presence of "free carbon" content. Amorphous Si-O-C phase provides chemical stability and host sites for Li storage, and free carbon phase contributes to conductive network as well as the high Li capacity of the bermat. As a result, as-prepared SiOC electrodes deliver specic capacity as high as 866 mA h g electrode À1 (considering total weight of the electrode) with high rst cycle coulombic efficiency of 72% in LIBs. Specic capacity of 800 mA h g À1 at a current density of 50 mA g À1 is achieved with a 100% capacity retention aer 50 cycles for a particular SiOC electrode. Further, as supercapacitor electrode SiOC bers have delivered a high gravimetric capacitance of 55 F g À1 at 100 mV s À1 and maintained 100% capacitance over 5000 cycles at 3 A g À1 .
To accomplish electrospinning of the preceramic oligomers 10 wt% of polyvinylpyrrolidone (PVP) with M w z 1 300 000 g mol À1 (Sigma Aldrich, Missouri, USA) was used as a spinning agent. Dicumyl peroxide (Sigma Aldrich, Missouri, USA) was used as a crosslinker for the oligomers and iso-propanol (Fisher Scientic, Massachusetts, USA) was used as a solvent. Ultra-high purity argon (Ar) gas was used for inert environment during pyrolysis which was supplied by Matheson (Kansas, USA).

Electrospinning and synthesis of bermat
Various SiOC bermat samples were fabricated via electrospinning of polymer solutions, followed by stabilization and pyrolysis at elevated temperatures. The electrospinning setup that was designed and built in the lab consisted of four major parts: 47 (1) a syringe which is controlled with a stepper motor works as a feeder for polymer solution; (2) a voltage source (up to 30 kV) that controls the amount of electrical charge applied to the solution; (3) an aluminum roller that works as a bermat collector; and (4) a microcontroller made of Arduino® that controls the motors' speed and the sensors. The major parts of the setup are illustrated in Fig. 2.
The electrospinning solution was prepared by dissolving 10 wt% of PVP in 7500 mg of iso-propanol; aer that preceramic oligomer was added into the solution with a weight ratio of 3 : 1 to PVP. DCP as a crosslinking catalyst was already dissolved into the preceramic oligomer, which was 1 wt% of the siloxane oligomer. The solution was then stirred for about 2 h to get a homogenous mixture of PVP and preceramic oligomer. The solution was then poured into the feeder syringe. During electrospinning, the feed rate was set at 5 mL h À1 while the voltage supply was maintained at 15 kV. The positive output was connected to the syringe nozzle and the collector was grounded. The nozzle to collector distance was set to 25 cm. The collector was wrapped with an Al foil and a bermat of 15 Â 15 cm 2 was collected onto the foil aer the electrospinning. The electrospun raw or as-spun bermat is shown in Fig. 2b.

Crosslinking and pyrolysis of the bermat
The electrospun bermats of the three siloxanes were dried at room temperature overnight and then crosslinked at $300 C in an oven for 8 h in the presence of air. The crosslinked bermats (Fig. 2c) were then pyrolyzed in an alumina tube furnace at approx. 800 C for 30 min in owing Ar gas. The heating rate was approx. 2 C min À1 and Ar gas ow rate was 5 mL min À1 in the furnace. The crosslinked bermats were cut into smaller rectangular shape (50 Â 25 mm 2 ) to t in the aluminum ceramic boat that was used to hold the samples in the furnace during pyrolysis. The bermat polymer-to-ceramic conversion was complete at 800 C; a representative of SiOC ceramic bermat is shown in Fig. 2d. The lateral shrinkage during polymer-to-ceramic conversion for DDTS-derived SiOC ceramic bermat was 40%, whereas for DTDS-and TPTS-derived SiOC ceramic bermats were 60% and 55% respectively.

Characterization techniques
Several material characterization techniques were employed to determine the morphological, compositional, and chemical conversions at various stages of processing of the bers-asspun, crosslinked, and pyrolyzed bermats. The images of surface features, shapes, and diameters of the bers in the SiOC ceramic bermats were investigated using a FEI Nova NanoSEM 450 scanning electron microscope (SEM). An FEI Tecnai Osiris 200 kV (scanning) transmission electron microscope (TEM) was used to obtain morphological and crystallographic information as well as energy dispersive X-ray (EDX) mapping analysis of the ceramic bermats.
To determine the molecular structure and chemical interactions of the ceramic bermats Raman spectroscopy and Fourier-transform infrared spectroscopy (FTIR) were used. Raman analysis in the range of 800-2000 cm À1 was performed to determine mainly the carbon vibrational modes using a confocal micro-Raman microscope (Horiba Jobin Yvon Lab-Ram ARAMIS) equipped with a HeNe laser source (632.8 nm). The presence and evolution of various chemical functional groups of the siloxane precursors in the bermats were investigated using a PerkinElmer Spectrum 400 FTIR spectrometer in the range of 500-3500 cm À1 .
To further investigate the chemical bonds in each sample, nuclear magnetic resonance (NMR) spectroscopy was used. A Bruker AVANCE 300 spectrometer was used to record solid-state 13 C CP MAS and 29 Si MAS NMR spectra using following MHz, y 0 ( 29 Si) ¼ 59.66 MHz), and a 4 mm Bruker probe and spinning frequency of 10 kHz. 13 C CP MAS experiments were recorded with ramped amplitude cross-polarization in the 1 H channel to transfer magnetization from 1 H to 13 C (recycle delay ¼ 3 s, CP contact time ¼ 1 ms, optimized 1 H spinal-64 decoupling). Single-pulse 29 Si NMR MAS spectra were recorded with recycle delays of 60 s. Liquid-state NMR spectra were recorded on a Bruker AVANCE 300 spectrometer Chemical shi values were referenced to tetramethyl silane for 13 C and 29 Si.
X-ray photoelectron spectroscopy (XPS) was carried out to analyze the surface composition using a Thermo Scientic Al Ka + ion beam (beam energy ¼ 1486.6 eV and spot size ¼ 400 mm) XPS. To remove the surface contamination of the ceramic ber mats, initial sputtering of the surface with Ar + at 3.0 keV for 2 min was performed.

Electrode preparation and electrochemical measurement
Ceramic bermats were tested electrochemically both as organic electrolyte-based lithium-ion batteries (LIBs) and aqueous supercapacitors electrodes. Electrospun SiOC ceramic bermats were used as freestanding electrodes in LIBs halfcells. A disk electrode was punched out from the pyrolyzed bermat ( Fig. 2e) with diameter of about 6.35 mm (1/4 inch), which was used as the working electrode. A glass ber membrane (4 z 19 mm, t z 25 mm) (GE, USA) as separator, pure Li metal (4 z 14.3 mm, t z 75 mm) (Alfa Aesar, USA) as the counter electrode, and approximately 6 drops of 1 M lithium hexauorophosphate (LiPF 6 ) in (1 : 1 v/v) dimethyl carbonate (DMC) : ethylene carbonate (EC) (Sigma Aldrich, USA) as the electrolyte were used. The cells were assembled in LIR 2032 coin cells in a glove box, and the assembled cells were tested using a multichannel BT2000 Arbin test system (Texas, USA) between 10 mV to 2.5 V vs. Li/Li + . The cells were subjected to symmetric cycling at current densities of 50, 100, 200, 400, 800, 400, 200, 100, 50 mA g À1 for 5 cycles each.
For supercapacitor testing, a three-electrode setup was used. Ceramic bermats (as an active material) were mixed with 10 wt% carbon black (as a conducting agent) (Alfa Aesar, Massachusetts, USA) and 5 wt% polyvinylidene uoride (PVDF) (as a binder) (Sigma Aldrich, Missouri, USA) thoroughly; approximately 4-6 drops of N-methyl 2 pyrrolidone (NMP) (Alfa Aesar, Massachusetts, USA) was also added to form a slurry of uniform consistency. The slurry was then pasted on stainless steel (SS) mesh (1 Â 1 cm 2 ) using a at paint brush, followed by drying at 80 C overnight in an oven. The ceramic coated SS was then used as working electrode in the three-electrode setup, where Pt wire and Ag/AgCl were used as the counter and reference electrode, respectively. 1 M NaCl was used as an electrolyte. A CHI 660 electrochemical workstation (CH Instruments, Inc., Texas, USA) was used to perform cyclic voltammetry (CV) and galvanostatic charge-discharge (GCD) of the electrodes. CV and GCD were performed in the potential window of 0-1 V at various scan rates and current densities, respectively. Electrochemical impedance spectroscopy (EIS) was also carried out from 0.01 Hz to 100 kHz at an amplitude of 5 mV. For CV and GCD, the specic capacitance values were calculated using eqn (1) and (2), respectively. 49 where, m is the active mass of electrode, v is the scan rate, V i and V f are the initial and nal voltage, I(V) is response current density, I is charge/discharge current, and DV is the potential window.

Results and discussions
3.1 Structure and morphology of bermats SEM and TEM images of the pyrolyzed samples revealed the surface and internal features of the SiOC bers. Fig. 3 shows that SiOC bermats fabricated from the three siloxane oligomer samples were largely uniform in diameter along the ber length. For DDTS-derived SiOC bers the average ber diameter was between 1-3 mm, whereas for DTDS-and TPTS-derived SiOC bers the average dimeter stayed between 0.2-1.5 mm. SiOC bers derived from the siloxane oligomer with high molecular weight resulted in large diameter bers compared to low molecular weight precursors. Furthermore, high molecular weight precursor produced thick and dense bers as can be seen from the inset of the high-resolution SEM images (Fig. 3a). The thinner and hollow bers derived from the DTDS and TPTS may have resulted from the steric hindrance, and molecular weight difference with the PVP (spinning agent) during the spinning process. 47 As a result, in our case of single-nozzle electrospinning, the PVP molecules moved to the inner layers and preceramic polymer stayed in the outer layers, and the separation likely occurred during the crosslinking and pyrolysis processes. 50 The core-shell structure of the hollow TPTS-derived SiOC bers was also conrmed by TEM. In addition, SEM and TEM revealed a smoother surface for DDTS-and DTDS-SiOC compared to TPTS-SiOC. The hollow bers with bead-onstring structure of the TPTS-SiOC was attributed to the low molecular weight as well as the concentration and viscosity of the spinning solution. [51][52][53] High-resolution TEM images of the pyrolyzed bermats, illustrated in Fig. S1, ‡ suggested the amorphous or featureless structure of the three ber types. The results of the EDX elemental analyses, also shown in Fig. S1, ‡ conrmed the homogenous distribution of the Si, O, and C elements in the ceramic bers.

Compositional analysis
Raman spectra (Fig. 4) show the presence of carbon domains which allows to evaluate corresponding microstructures in the ceramic materials. The D and G vibrational bands at 1342 and $1600 cm À1 for the DDTS-and DTDS-derived SiOC, suggested the presence of "free carbon" structures i.e., carbon bonded to carbon and not any other elements in the ceramic. The intense D bands suggested the existence of disordered aromatic rings, whereas G bands were associated with the sp 2 -hybridized carbon atoms. 54 Additionally, under the tted curves the T (1424 cm À1 ) and D 00 ($1490 cm À1 ) bands indicated the likely presence of disordered sp 2 -sp 3 bonds and amorphous carbons in the SiOC bers, respectively. However, for TPTS-derived SiOC sample, no pronounced D and G peaks were observed as a strong uorescence background was produced under the visible laser source (HeNe 632.8 nm). The presence of carbon domains in the TPTS-SiOC bers were later conrmed using XPS analysis. FTIR spectra, plotted in Fig. 5, shows the characteristic absorption bands of the PVP and polymer-to-ceramic conversion of preceramic Si precursors from the as-spun to pyrolyzed stages. For the PVP, as shown in Fig. 5a, as-spun and crosslinked samples exhibited the main peaks at 1660, 1427, 1291, and 570 cm À1 , that were assigned to the stretching of C-O, C-H, C-H 2 , and N-C]O bonds, respectively. 55 The presence of these peaks in both as-spun and crosslinked PVP bers suggested no noticeable crosslinking happened during heat treatment at $300 C, while aer pyrolysis at $800 C these peaks disappeared. DDTS bers electrospun with PVP (as spinning agent) showed (Fig. 5b) Si-CH 3 ($1260 cm À1 ), and Si-O-Si ($1060 cm À1 ) stretching bonds in the as-spun bers indicating the obvious presence of Si precursors in these samples. 47,56 Si-CH]CH 2 ($1660 cm À1 ) are superimposed with the C-O vibration. Reduction of Si-CH 3 and Si-CH]CH 2 peak intensities in the crosslinked samples suggested crosslinking reactions occurred at approx. 300 C. Aer pyrolysis, the two obvious peaks at $1060 and $800 cm À1 conrmed the presence of Si-O and Si-C bonds, respectively, in the pyrolyzed SiOC bers. 57 DTDS-and TPTS-derived SiOC bers also showed similar characteristic peaks from as-spun to pyrolyzed stages, which are presented in Fig. 5c and d. However, Si-O bonds were more intense than Si-C bonds, especially in DTDS-derived SiOC bers (Fig. 5c), indicating the presence of more Si-O bonds as compared to Si-C bonds in the ceramic samples. Note the peaks around 3000-3100 cm À1 for DDTS, which can be attributed to ] C-H stretching of the aromatic phenyl group. 58 XPS survey spectra of the electrospun SiOC bers, presented in Fig. S2, ‡ showed similar Si 2p, C 1s, and O 1s peaks, irrespective of the source of the preceramic polymer. The surface composition of the elements was determined by integrating the area under the respective peaks and presented in Table 1. As expected, pyrolyzed PVP bers mostly showed the presence of carbon (93.06 at%) aer heat treatment. DDTS-and TPTSderived ceramic bers shows increased level of oxygen in the pyrolyzed samples indicating the inclusion of oxygen during crosslinking of the bers. All specimens showed a signicant amount of carbon, mainly free carbon, also proposed by Raman analysis. However, pyrolyzed samples presented a little amount of nitrogen, which was suspected to have come from PVP. Highresolution XPS spectra, plotted in Fig. 6, shows the bonding of the pyrolyzed SiOC samples derived from various siloxane precursors. Curve ttings were done to the Si 2p, C 1s, and O 1s peaks. The spectra under Si 2p band indicated the presence of SiCO 3 (102.1 eV) and SiO 4 (103.5 eV) peaks. 31 Low intensity of SiCO 3 compared to SiO 4 for pyrolyzed-DDTS ceramic bers were due to the low carbon content in the elemental composition. In addition, C-Si (284.0 eV), C-C (284.9 eV), and C]O (287.1 eV) were observed in C 1s band. 47 The higher amount of carbon in the pyrolyzed-DTDS and pyrolyzed-TPTS bers lead to higher intensity of C-C and C-Si peaks. The intense C-C peaks also pointed to increased possibility of free carbon phase in the ceramic bers. O 1s band were tted with 3 peaks at 532.4, 532.9, and 534.4 eV corresponding to SiO 2 , O-Si, and C-O, respectively.
The initial composition of the electrospinning solution was investigated by liquid state NMR. 29 Si spectra (Fig. 7a) show for all systems a signal around À3 ppm corresponding to ViMe 2 SiO environments. In addition, signals are observed at À46.8 ppm and À34.0 ppm in DDTS and TPTS based solutions that can respectively be assigned to Ph 2 SiO 2 and ViMeSiO 2 groups. This assignment was conrmed by 2D experiments. For example, 1 H- 29 Si HMBC map of DDTS solution (Fig. S3a ‡) shows that the 29 Si signal at À47 ppm shows cross-peaks with 1 H phenyl signals between 7.5 and 8 ppm while the 29 Si signal at À34 ppm shows cross-peaks with 1 H vinyl signals between 6 and 5.5 ppm. The evolution of the various preceramic siloxane precursor structures during heat treatment and the resulting SiOC samples were also investigated using solid-state NMR. Spectra were recorded on bulk powders instead of bers to improve the signal/noise ratio. Fig. 7b shows 29 Si MAS NMR spectra for crosslinked (at $300 C) polymers and compares the key features of the DDTS, DTDS, and TPTS polymers. Appearance of -CH x -Me 2 SiO ($10 ppm) in the polymers indicated effective crosslinking of the vinyl groups occurred in presence of DCP. 59 On the other hand, a signal appeared at À21 ppm which was characteristic of Me 2 SiO 2 units in polydimethylsiloxane (PDMS) chains. 60,61 This would suggest that part of the Si-CH]CH 2 groups were replaced by Si-O-Si bonds during heat-treatment as conrmed by the total disappearance of ViMeSiO 2 units at À3 ppm in DTDS and TPTS crosslinked systems. Moreover, in the case of TPTS, ViMeSiO 2 units observed at À34 ppm in the initial solution disappeared in the cross-linked system while new signals were present around À70 ppm, a chemical shi range characteristic of MeSiO 3   (Fig. 5b). The signal corresponding to Ph 2 SiO 2 units (À48 ppm) was still intense aer crosslinking suggesting that Si-Ph bonds were preserved. The 29 Si MAS spectra of the samples pyrolyzed at 800 C showed SiO 4 bonds (À110.0 ppm) (Fig. 7d), while no Si-C bonds were found. 62 This might be due to the fact mostly free C phase had formed during pyrolysis, and very low Si-C bonds were present in the ceramic samples. 47 Nevertheless, high-res XPS of the SiOC bers indicated the presence of Si-C and C-C bonds in the ceramic bers. 13 C cross polarization (CP) MAS NMR spectra in Fig. 7c show strong peaks of Si-CH x indicating the crosslinking of the Si precursor polymers. In the DTDS polymer, sp 2 carbons were observed around 130 ppm that corresponded mainly to the phenyl groups. Indeed, from 1 H-13 C HSQC map of DDTS solution (Fig. S3b ‡) cross-peaks with 1 H phenyl signals (between 7.5 and 8 ppm) were observed for carbon signals at 127.6, 129.7 and 134.4 ppm, while the cross-peaks with 1 H vinyl signals (between 6 and 5.5 ppm) were observed with carbon signals at 132.1 and 139.0 ppm. The absence of a clear 13 C peak at 139 ppm in the DTDS cross-linked sample suggested that the proportion of vinyl groups was small compared to phenyl groups. No evidence of crosslinking behavior between PVP molecules and siloxane was observed in all the crosslinked samples when compared with PVP polymer. 13 C CP MAS NMR spectrum of the TPTSderived SiOC conrmed the presence of typical free carbon as shown in Fig. S4. ‡ 13 C spectra of the remaining SiOC samples were not acquired as it was obvious from the other characterization techniques that all the systems were full of graphite.

Electrochemical performance
The electrochemical energy storage capability of the precursor derived SiOC bers were analyzed in Li half-cells. The asprepared ceramic bermats were used as electrodes. Fig. 8a-c shows the potential vs. capacity plots for various SiOC ber electrodes. A closer look at the charge-discharge proles of the samples at a current density of 50 mA g À1 showed that rst cycle had experienced irreversible capacity decay. Among the SiOC electrodes, the DDTS-derived SiOC delivered initial discharge capacity of 1188 and charge capacities of 866 mA h g À1 , corresponding to a high coulombic efficiency of over 72%. DTDSderived SiOC electrode delivered 1332 and 800 mA h g À1 , respectively for discharge and charge capacities in the rst cycle with 59% efficiency. TPTS-derived SiOC provided 1150 and 636 mA h g À1 for discharge and charge capacities, respectively, with 52% coulombic efficiency. DDTS-and DTDS-SiOC electrodes showed stable response with coulombic efficiency of $100% in the second and third cycles, whereas TPTS-SiOC displayed poor reversibility with only $83% efficiency. Differential capacity plots, presented in Fig. S5, ‡ showed two distinct regions for the SiOC electrodes: a sharp peak around 0.1 V, which corresponded to the Li insertion into the carbons; and a broad region in between 0.2-0.6 V, that was related to Li and Si-O-C phase. 27 Accordingly, when tested for cycling stability TPTS-SiOC electrode showed poor performance and the charge capacity dropped sharply to 100 mA h g À1 aer 50 cycles at 50 mA g À1 . For DDTS-and DTDS-SiOC electrodes, the capacities aer 50 cycles were 580 and 800 mA h g À1 , showing 100% capacity retention for the DTDS-derived SiOC ber electrode. The distinct ber morphology and the "free C" content in the DTDS-SiOC particles provided additional Li-ion storing sights and improved the electrochemical properties of SiOC bers. 63 The rate capability performance of the various SiOC electrodes is presented in Fig. S5. ‡ DTDS-SiOC delivered a high capacity of 450 mA h g À1 at a high current density of 800 mA g À1 . However, thin diameter and hollow structure of the TPTS-SiOC bers might have been destroyed during Li-ion insertion/extraction process, which resulted in poor cycleability and rate capability. 64 The SEM and TEM images of the SiOC electrodes aer cycling in LIBs are shown in Fig. S7, ‡ where the broken bers of the TPTS-SiOC can be seen post-cycling. Whereas poor rate capability performance of the DDTS-SiOC can be attributed to the low C content as well as dense structure of the bers which limits the ion diffusion. To show the consistent performance of the cells, three Li half-cells of each type were assembled and tested and the performance of all the cells are presented in Fig. S5. ‡ Fig. S5(a-c) ‡ shows the differential capacity curves of the SiOC electrodes for the initial 3 cycles. The sharp cathodic peak at 0.005 V and the broad anodic peak can be attributed to the lithiation and delithiation of Li + ions in the amorphous SiOC structure. 54,65 The cathodic peaks in the range of 0.05-0.16 V can be attributed to the irreversible reaction in hard carbon present in the SiOC structure. 25,65 However, cathodic peaks due to formation of SEI (typically at $0.5 V) are less pronounced in the dQ/dV plots. A comparison of the electrochemical performance of the polymer-derived SiOC anodes with reported SiOC anodes for LIBs is presented in Table S1. ‡ The electrochemical storage capabilities of SiOC bers in supercapacitors were also investigated using three electrode system and 1 M H 2 SO 4 as electrolyte. CV plots obtained for TPTS-derived SiOC electrode at various scan rates of 2 to 500 mV s À1 from 0 to 1 V are shown in Fig. 9a. The almost rectangular shape of the voltammogram proles indicated mostly the electrochemical double layer capacitive behavior (Type A) of the SiOC electrode during charge storage. 66 However, at a low scan rate of 2 mV s À1 , a broad redox peak was observed for TPTS-SiOC electrode in between 0.2 and 0.4 V (Fig. S6d ‡). The charge storage mechanism at lower scan rate can be contributed to the combination of double-layer capacitance of C in SiOC and pseudocapacitance from Si-O and Si-Si components. The quasirectangular shape of the CV plot retained even at higher scan rate of 500 mV s À1 , indicating higher ionic diffusivity and charge transfer of the TPTS-SiOC electrode. As a result, the electrode delivered high specic capacitances of 78, 69, 55, 47, and 38 F g À1 at the scan rates of 2, 10, 100, 200, and 500 mV s À1 . The areal capacitance of the TPTS-SiOC electrode was also calculated and a high capacitance of 474 mF cm À2 was achieved at 2 mV s À1 . The cyclic voltammograms of DTDS-and DDTSderived SiOC supercapacitor electrodes are presented in Fig. S6. ‡ DTDS-SiOC delivered 107 F g À1 at a scan rate of 2 mV s À1 and decreased to 31 F g À1 at 100 mV s À1 , where DDTS-SiOC delivered 25 and 3 F g À1 at 2 and 100 mV s À1 , respectively. The highest performance of TPTS-SiOC among the three SiOC electrodes was correlated to the lowest diameter and hollow core of the bers, contributing to a higher electrochemically active area for double-layer capacitor. The BET analysis and avg. pore diameter of the SiOC samples are presented in ESI Section 7. ‡ As-anticipated the TPTS-SiOC had a high specic surface area of 235 m 2 g À1 . In addition, the presence of "free C" contributed to the enhanced electronic conductivity. Fig. 9b demonstrates the charge/discharge proles of the best performing TPTS-SiOC electrodes at various current densities from 0 to 1 V. The quasi-triangular shape of the GCD curves without any obvious plateau suggested the dominating double-layer capacitive behavior of the electrode, which was in accord with earlier results. 67 The triangular shape of the GCD is a typical behavior of highly reversible supercapacitor with constant charge/discharge. 68 As a result, TPTS-SiOC delivered 30 F g À1 even at a high current density of 10 A g À1 . When tested for cycling stability, the electrode demonstrated $100% capacitance retention over 5000 cycles at 3 A g À1 . DDTS-and DTDSderived SiOC electrodes also showed stable cycling ability over 5000 cycles as shown in Fig. S6. ‡ EIS was done for the various SiOC electrodes to further understand the reaction kinetics of the supercapacitor samples. In Fig. 9d, the ohmic resistance, R s between the aqueous electrolyte and the electrode was measured in the high frequency region where the curves intercept the Z 0 axis. The semicircle in medium frequency region was attributed to the charge transfer resistance (R CT ) of the electrolyte-electrode interface, and the inclined lines in the low frequency region corresponded to the ion diffusion into the SiOC electrode materials. 54,69 An equivalent circuit was obtained for the electrodes, as shown in Fig. S6, ‡ from which the calculated R CT for TPTS-SiOC electrode (6 U) was much lower than the DDTS-SiOC (15 U) and DTDS-SiOC (50 U) electrodes. In the low frequency region, the near vertical line to the real axis of the TPTS-SiOC electrode corresponded to the ideal capacitive behavior. 70 Furthermore, the Bode plot (Fig. 9e) showed that TPTS-SiOC had the nearest phase angle (80.5 ) to 90 , indicating the best capacitive behavior among the SiOC electrodes. 71 At phase angle of 45 , the relaxation time constant, s o (s o ¼ 1/f o ) were measured to be 2.61, 6.81, and 56.18 ms for the TPTS-SiOC, DDTS-SiOC, and DTDS-SiOC electrodes, respectively. The signicantly lower time constant of the TPTS-SiOC electrode conrmed the fast ion diffusion and transport characteristic as a supercapacitor electrode.

Conclusion
In summary, we have presented a method to fabricate scalable, self-supporting electrodes from precursor-based SiOC bers utilizing three unique siloxane precursors. The electrospinning of the short-chain siloxane oligomers was not achievable without the addition of a spinning agent (such as PVP). In addition, the precursor-to-ceramic yield of the bers was also observed to be lower than some of the previously reported polymer-derived ceramic materials. This phenomenon suggested limited thermal crosslinking of the bers in the presence of 1 wt% DCP. The electron microscopy of the SiOC bers featured rigid surface structures and small diameters (0.2-3 mm). Raman, XPS, FTIR, and NMR characterization techniques outlined the polymer to ceramic conversion stages. 29 Si MAS NMR spectra of the SiOC bers showed only SiO 4 bonds, indicating mostly free C phase had formed during pyrolysis with a low amount of Si-C bonds in the ceramic samples. The amorphous SiOC structures, comprised of free carbon phase and Si-O-C mixed bonds of DTDS-derived SiOC contributed to high reversibility of Li storage. The free carbon phase served as electron conductor as well as the major electrochemically active sites for Li ions, while amorphous Si-O-C network contributed to minor Li storage. In terms of electrochemical properties, the SiOC electrodes displayed excellent capacity with a high coulombic efficiency in LIBs. As supercapacitor electrodes, superior cycleability of 100% capacitance retention over 5000 cycles was achieved. The as-prepared SiOC ber mats can provide highly efficient, high-energy, and high-power electrodes and reduce the total weight of the electrochemical energy storage devices.

Author contributions
S. M. carried out the preparation of all samples, perform characterization, data analysis and draed the manuscript; F. R. and C. G. contributed to perform NMR characterizations, data analysis and draing of the manuscript; G. S. conceived the idea, designed the project and helped with draing the manuscript.

Conflicts of interest
There are no conicts of interest.