2023 Volume 91 Issue 7 Pages 077008
Current efforts to improve sodium-ion batteries are heavily focused on developing high performance carbon materials for the negative electrode. With significant research, hard carbons have come to show massive storage capacities and fast discharge rates. On the other hand, soft carbons have received very little attention, though they likewise encompass a wide variety of materials with structures highly dependent on the starting material and preparation temperature. In our contribution, we systematically evaluate the electrochemical performance of soft carbon electrodes made from mesophase-pitch carbon fibers (MCF). By using felt electrodes, we evaluate the cyclic voltammetry of MCFs prepared at 600–1300 °C and show the best performance with MCF prepared between 700–950 °C. In addition, using a surface modification step with silver showed significantly improved voltammetry for all the materials. Electrochemical impedance measurements further indicated that the surface modification step could decrease both of charge transfer resistances and film resistances attributed to the solid electrolyte interphase. Upon comparing lithium- and sodium-cell, it was revealed that sodium-cell demonstrated more significant increase in current density and decrease in resistance through surface treatment. We further verified our results with measurements on single-fiber electrodes; an increase in currents and a decrease in impedance were also observed by the surface modification, as with the felt electrodes. Overall, we speculate our surface modification removes inhibitors, such as functional groups or impurities, on the MCF surface to prevent sluggish ion transfer or trapping during sodium insertion/extraction.
Recently commercialized sodium-ion batteries (SIBs) are gaining significant interest as an alternative energy storage to lithium-ion batteries. SIBs are being considered for large-scale energy storage due to their low cost and abundance of raw materials for electrolytes and electrode materials.1,2 Among negative electrode materials for SIBs, hard carbons are regarded as the most promising with high discharge capacities, low working potentials and high capacity-retention during long cycling.2–4 However, hard carbon tend to show large irreversible capacities during initial cycling5 and poor sodium insertion at high currents.6,7 On the other hand, soft carbons are another electrode material that has received much less attention for SIBs, though they can show improved coulombic efficiencies, charging rates and can be prepared as stand-alone electrodes that do not require a binder or conductive additives.8–11 In addition, the performance of hard carbon electrodes has been improved through modification with soft carbons.5,12 Like hard carbons, the synthesis procedure greatly impacts the structure of the prepared material, but few studies have systematically evaluated the relationship between electrochemical performance and the preparation conditions for soft carbons.8–10
During synthesis, the soft carbon structure undergoes several changes including loss of functional groups (e.g. hydrogen elimination) and an increase in crystallinity with temperature. As such, the preparation temperature strongly impacts the final structure of the material.13 At temperatures from 600–1000 °C, small domains of stacked graphene begin to develop. These domains increase in size, and the stacking layers increase between 1000–2000 °C, eventually becoming graphitic at even higher temperatures.13–16 In the recent pasts (∼1990–2000), soft carbons were widely studied for energy storage.13,17–21 For one type of soft carbon, mesophase-pitch carbon fibers (MCF), high reversible capacities for Li insertion/extraction have been observed (up to 500 mAh g−1).22 MCF are derived from petroleum and can be prepared as felt electrodes without binder or conductive additives. Despite variations in preparation temperature, the raw materials, manufacturing process, and shape remain consistent, with no discrepancies attributed to manufacturing methods. As the preparation temperature of MCF increases, the reversible capacities tend to decrease down to 170 mAh g−1 (at 2000 °C) before increasing again toward 372 mAh g−1 (theoretical capacity of graphite).13,22,23 In addition, the surface of the soft carbon also impacts performance, which has led to several reports focused on surface modifications, including oxidation and silver deposition.24,25 Omae et al. reported that a combination of silver vacuum-deposition and vacuum-heating dramatically enhanced lithium insertion/extraction on soft carbon electrodes.24
For application in SIBs, we previously reported sodium insertion/extraction on soft carbon electrode10 using the same experimental conditions as Omae et al.24 Cyclic voltammetry (CV) was utilized to evaluate the sodium extraction rate at MCFs prepared at temperatures between 600–950 °C. Using potential step chronoamperometry, MCF prepared at 700 °C and 800 °C exhibited diffusion coefficient on the order of 10−9 cm2 s−1 during charging reaction, while 600 °C and 950 °C showed slower diffusion. They also showed that surface modification, via silver deposition and vacuum heating, led to further improvements in the diffusion coefficient, up to 10−8 cm2 s−1 and an improved discharge capacity. Despite these improvements, this surface modification with silver has only been applied for one type of MCF prepared at 800 °C. Moreover, the relationship between preparation temperature and surface resistance have not been studied systematically. In this study, the impacts of preparation temperature and surface modification for the low temperature prepared-soft carbon fibers have been identified by combination of cyclic voltammetry and electrochemical impedance spectroscopy (EIS). Building on previous research, we conducted a systematic evaluation to understand the electrochemical performance of soft carbon electrodes fabricated from MCF at preparation temperatures ranging from 600–1300 °C. Utilizing both CV and EIS, we assessed felt and single fiber electrodes in a 1 mol dm−3 NaClO4 electrolyte in propylene carbonate (PC). Additionally, we evaluated the impact of the surface modification process, which entailed vacuum-deposition of silver followed by vacuum-heating, on various samples and discussed the influence of interfacial resistance on sodium insertion/extraction. Consistent with prior findings, our measurements indicate that temperatures of 700–800 °C yielded the best overall performance, and surface modification enhanced the performance of all the MCF samples.
MCF (Melblon®) manufactured at 600–1300 °C, 2000 °C and 3100 °C were purchased from Petoca Co., Ltd. Before use, MCF was put into a vacuum cylinder and heated at 300 °C to remove adsorbed water and then stored inside a glovebox. Additional samples of MCF were modified using silver vacuum-deposition and vacuum-heating, before transfer to the glovebox. Silver deposition was performed in a vacuum chamber on MCF felt, which had been peeled until the opposite side was visible. Silver film was deposited on MCF until the film thickness reached 90 nm. The thickness of deposited film was monitored with a quartz crystal oscillation type deposition controller. Subsequently, the silver covered MCF was heated at 350 °C for 1 hour in vacuum. The silver film melted at Tammann temperature and formed tiny particles during heating process. After cooling down to room temperature, the sample was removed from the heating glass cylinder, and the MCF sheets were compressed into felt.
The morphology of MCF was obtained using a scanning electron microscope (JCM-7000, JEOL). The samples were also characterized using X-ray diffraction with a laboratory XRD diffractometer (MiniFlex600, Rigaku) equipped with a 2D hybrid pixel array detector (HyPix-400 MF, Rigaku) by setting MCF felts on the specific sample stage and using Ni-filtered CuKα radiation.
2.2 Electrochemical characterizationAll electrochemical measurements were conducted using a 3-electrode cell (Fig. S1) and a potentiostat (HA-151B, HOKUTO DENKO) equipped with a function generator (HB-305 or HB-111A, HOKUTO DENKO) and an X-Y pen recorder (WX1000, GRAFTECH). Other potentiostats (SP-150, BioLogic, HSV-110, HOKUTO DENKO) were also used. A sandwich-type MCF felt electrode (1 × 1 cm) between two pieces of nickel expanded metal was used as the working electrode, and metallic sodium was used as the counter and reference electrodes. For the electrolyte, we used 1 mol dm−3 NaClO4 dissolved in PC (Tomiyama chemical) or co-solvent of ethylene carbonate and dimethyl carbonate (1 : 1 v/v). Additional experiments were conducted in 1 mol dm−3 LiClO4 dissolved in PC (Tomiyama chemical) with Li metal for the reference and counter electrodes. MCF single fiber experiments were conducted in a 2-electrode configuration as previously reported.10 In the case of a single-fiber electrode, we found that the current flowing through the reference electrode (Luggin capillary) introduced noise. Our experience indicated that such conditions would impede accurate measurements, so we opted against using a three-electrode cell. In brief, one MCF fiber was attached to a nickel wire with carbon paste, and 5 mm was immersed in the electrolyte. Cyclic voltammetry measurements for both sandwich-type and single wire electrodes were performed in the potential range of 0.005 V–2.0 V vs. Na+/Na using a scan rate of 0.1 mV s−1. After three cycles, electrochemical impedance measurements were collected at various potentials using a 10 mV amplitude in the frequency range of 100 kHz to 100 mHz. All measurements were performed in a grove box filled with Ar at room temperature.
The as-purchased MCF single fibers showed an approximate diameter of 10 um (Fig. 1a) for all the preparation temperatures. Likewise, the MCF felt electrodes were composed of multiple cylindrical fibers pressed together. XRD of the MCF felts (Fig. 1b) showed a broad peak at ∼25°, suggesting low crystallinity for the carbon structure. As the preparation temperature for each sample, this peak became sharper and showed a shift to higher angles. For MCFs manufactured at 2000 °C and 3100 °C, 002 and 004 diffraction peaks were observed indicative of graphite (See Fig. S2), but these were not observed for our samples treated at lower temperature. We further modified the MCF felt electrodes using vacuum deposition to deposit a thin layer (∼90 nm) of silver on the carbon surface, as previously reported.10 After deposition, we observed a uniform coating of silver across the MCF fibers under SEM (Fig. S3). Thereafter, the samples were further heated to 350 °C for 1 hour under vacuum producing aggregates of silver across the MCF surface (Fig. 1c).
Characterization of MCFs obtained by SEM and XRD. (a) The morphology and (b) XRD patterns of MCFs. (c) Aggregates of silver across the MCF surface after surface modification.
To evaluate the sodium insertion/extraction process on our MCF electrodes, we used CV in 3-electrode configuration at a scan-rate of 0.1 mV s−1. As seen in Fig. 2a, MCF700, MCF800 and MCF950 showed two peaks at 0.75 V, Na+ adsorption at defects/intercalation into interlayer space, and at 0.1 V vs. Na+/Na, filling of nano-size pores.11,18,26 The results of the MCFs are collected in Table S1, showing capacities of 180 mAh g−1 for MCF700, MCF800 and MCF950 comparable to other reports of soft carbons.8,9 For MCF600, we only observed broad peaks during cycling and the capacities were significantly smaller. In general, carbonization of organic compounds progresses from 600 °C, so the MCF600 structure likely contains few sodium insertion sites and a large amount functional groups that cause electrolyte decomposition and Na+ trapping.9,11 MCF1300 also showed some differences in its CV, where Na insertion was observed at a lower potential of 0.3 V vs. Na+/Na, and the discharge capacities were the lowest among the soft carbons. The higher synthesis temperature of MCF1300 would lead to increased crystallinity and a higher number of layer stacking, while defects and interlayer space are decreased. The cyclic voltammograms of MCF2000 and MCF3100 are displayed in Fig. S4. In the PC-based electrolyte, MCF2000 and MCF3100 experienced significant electrolyte decomposition due to the co-intercalation of PC solvent into the turbostratic and graphitized carbon.27 Conversely, in the ethylene carbonate (EC) and dimethyl carbonate (DMC) co-solvent electrolyte, these carbons exhibited ten-fold lower currents than MCF600–1300, yet they show reversible Na insertion/extraction. For the surface modified soft carbons (Fig. 2b), we observed higher current densities, higher extraction capacities and a negative shift of the anodic peak suggesting a reduction of overvoltage28 for sodium extraction. The extraction capacities for MCF700, MCF800 and MCF950 were all higher than the best unmodified sample, MCF700. Silver appears to play a significant role in the improvement as MCF samples prepared by heating at 350 °C for 1 hour in a vacuum demonstrated a slight increase in current density, although not as dramatic as the surface-modified MCFs (Fig. S5). The silver particles on the MCF do not electrochemically alloy with sodium, as evidenced by the absence of a discernible difference before and after cyclic voltammetry (Fig. S6). The capacities obtained from a 0.01 mV s−1 scan, as shown in Table S2, exceed those in Table S1 for both pristine and surface-modified samples. Notably, the increased capacities obtained from pristine MCFs closely matched the capacities of surface-modified MCFs ascertained from a 0.1 mV s−1 scan (Fig. 2c). In addition, the surface-treated MCFs demonstrated smaller increases when the scan rate was reduced, suggesting that the capacity gleaned from CVs is approaching the maximum performance achievable by the MCFs. In other words, compared to pristine MCF, the surface modified MCF exhibits a faster reaction rate, enabling it to achieve its full capacity even at a scan rate of 0.1 mV s−1. Similarly, surface modification improved the performance of the MCF in a Li+ electrolyte (Fig. 2d). There is speculation that impurities (e.g. metal compounds) as well as functional groups can inhibit the Na insertion/extraction process, and by using a surface modification step, these inhibitive structures can be minimized.10,25 Our results indicate surface modification as a viable strategy for improving performance without significantly impacting the CV profile.
Cyclic voltammograms of sodium insertion/extraction on (a) pristine and (b) surface modified MCF felts electrode in 3-electrode cells. (c) Sodium extraction capacities of the pristine and surface modified MCFs estimated from CVs at the scan rate of 0.1 and 0.01 mV s−1. (d) Cyclic voltammograms of sodium or lithium insertion/extraction on pristine or surface modified MCF700 felt electrodes.
To further characterize the MCFs, we conducted EIS after CV cycling. As seen in the Nyquist plot in Fig. 3a, all the samples showed a single semicircle initiating at ∼10 Ω. We observed a decreased impedance as the preparation temperature was increased up to 950 °C, and the impedance increased again for MCF1300. The observed decrease in impedance is likely due to more substantial carbonization and changes to the surface structure. For MCF1300, the increased graphitization gave the smallest impedance on Li insertion compared with other MCFs prepared at lower temperature though negative impact on Na insertion (Fig. S7a). All of our surface modified samples (Fig. 3b) showed lower impedances compared with the untreated samples with MFC700, MFC800, MCF950 and MFC1300 showing improved charge transfer over the best untreated sample, MFC950. Figure 3c shows results of impedance analysis. The Nyquist plot obtained at 0.3 V of the pristine MCF shows a single depressed semicircle, while the surface-modified MCF clearly shows two semicircles. Fitting results showed that the peak-top frequencies of the low-frequency side were almost identical to those of the pristine and surface-treated MCFs (fct, Pristine and fct, Surface modified). The resistance on the low-frequency side had a potential dependence, suggesting that it is a charge-transfer resistance24–26,29,30 (Rct, Pristine and Rct, Surface modified). The Rct, Pristine was 570 Ω at MCF600, decreased to 50 Ω at MCF700 and 12 Ω at MCF950, and increased again to 44 Ω at MCF1300. Surface treatment resulted in a decrease in charge transfer resistance at all preparation temperatures, with MCF700-MCF1300 showing Rct, Surface modified below 15 Ω. MCF600 still exhibited 109 Ω, however, the decrease derived from surface modification was 460 Ω, which is more pronounced than those for other MCFs. We expect that the high frequency resistance could be associated with the solid electrolyte interphase (SEI).24,25,30 Rf, Pristine varied between 14 and 60 Ω, demonstrating no significant dependency on preparation temperature, unlike Rct, Pristine. This variability can be linked to the amount of functional groups and residual impurities on the MCF surface. Peak-top frequency of the surface modified samples (ff, Surface modified) shifted to higher values compared to that of the pristine samples (ff, Pristine), in the range ∼160–330 Hz and ∼50–140 Hz, respectively. This observation suggests that the structure and composition of the SEI formed on surface modified MCF is different from those of the pristine MCF due to the coexistence of carbon and silver, although the detailed mechanisms underlying these observations remain unclear. As the influence of preparation temperature on the resistance at the electrode interface has virtually disappeared, the disparity due to the surface condition of the MCF electrode has been eliminated, suggesting that the intrinsic difference in sodium diffusion inside the MCF can likely be evaluated in CVs. As with the CV results, the reduced electrochemical impedance likely derives from a decrease in the inter-fiber contact resistance due to silver deposition, and also may further be related to elimination of surface impurities. Same tends were obtained with impedance of surface modified MCFs on Li insertion (Fig. S7b). To compare EIS between Li and Na cell with MCF700 felt electrode, measurements were carried out at 0.03 V vs. Na, 0.03 V and 0.3 V vs. Li based on the current peak in CVs and the difference in standard electrode potentials between Na and Li (Fig. 3d). The total resistance for sodium insertion/extraction was initially greater than that of lithium in unmodified MCF700: 62 Ω at 0.03 V vs. Na+/Na, 23 Ω and 52 Ω at 0.03 V and 0.3 V vs. Li+/Li, respectively. However, after the surface modification, the total resistance for sodium insertion/extraction became lower than that of lithium: 9 Ω at 0.03 V vs. Na+/Na, 19 Ω and 29 Ω at 0.03 V and 0.3 V vs. Li+/Li, respectively. These observations indicate that, given an appropriately treated soft carbon surface, sodium could be more readily inserted/extracted than lithium. Figure S8 displays the electric double layer of MCF as measured by CV. The surface modification increased the electric double layer capacitance roughly fourfold. This observation indicates that the electrochemically active surface area on the MCF felt was increased. We believe that this enhancement is due not only to the increase in geometric surface area, but also to the removal of impurities, which may activate the electrochemically inactive surfaces of MCFs.
Nyquist plot obtained from (a) pristine and (b) surface modified MCF felts electrode in 3-electrode cells. (c) Resistances (left axis) and time constants (right axis) of pristine and surface modified MCFs. Pink dots show charge transfer resistances of pristine (○) and surface modified (□) MCFs. Black dots (△) indicate resistances obtained in the high-frequency region of the surface-modified MCFs obtained at 0.3 V vs. Na+/Na. (d) Nyquist plots of pristine or surface modified MCF700 felt electrodes in sodium/lithium cell.
While we have not yet fully deciphered the mechanism of this novel surface modification, we propose several potential effects. Firstly, the presence of silver nanoparticles is hypothesized to influence: 1) the enhancement of electrical conductivity by silver particles; 2) the catalytic activity of silver particles on Na absorption/desorption reactions in MCF; 3) the differences in composition and structure between the SEI formed on the MCF surface and the SEI formed on the silver surface. Further effects are inferred based on an increase in the current seen in the cyclic voltammogram, the increase in electric double-layer capacitance, and changes in morphology: 4) an increase in surface area due to the presence of silver particles; 5) an increase in surface area attributable to the heat-treatment of carbon surface with silver; 6) the removal of functional groups present on the MCF surface by melting and aggregation of silver; 7) the removal of residual impurities (Fe, etc.) on the MCF surface due to the melting and agglomeration of silver. To further evaluate surface modification on MCF, we conducted single fiber measurements which eliminates many of the complications due to using large composite electrodes.10,31 For these measurements, a single MCF700 fiber was used as the working electrode in a 2-electrode configuration (Fig. 4a). CV measurements at pristine and surface modified single fibers (Fig. 4b) showed a shift in the reduction/oxidation peak near ∼10 mV, in line with our measurements on the felt electrodes. Again, the surface modification led to a higher observed current than the untreated sample. Figures 4c and 4d show the Nyquist and Bode plots for the single fibers, respectively, with an observed reduction in the impedance after surface modification. Unlike our felt electrodes, single fiber measurements do not involve fiber-to-fiber contact resistances.10 So, this suggests the improved performance for the surface modified samples is not only due to an improvement in the fiber-to-fiber contact, but likely due to a decrease in the inhibitory factors at the electrodes surface (e.g. charge trapping functional groups). The XRD pattern of the surface modified MFC felts shows that the diffraction peak of the Ag particles on the CSM is broadened and shifted to a higher angle than the diffraction peak of the Ag bulk (Fig. S9). These observations imply that elements with smaller atomic radii than Ag, such as C and Fe, might be dissolved in Ag particles during surface modification. Further studies may show that surface treatment with other metals32,33 or functionalities21,34 can be highly effective for further improving the performance of soft carbons.
Electrochemical performance of pristine and surface modified MCF700 single fiber electrode. (a) 2-electrodes cell for electrochemical measurements. (b) Cyclic voltammograms, (c) Nyquist plots and (d) Bord plots obtained from MCF700 single fiber electrode.
This study investigated the use of MCF prepared at different temperatures and after surface modification for sodium energy storage. Overall, increasing the preparation temperature improved sodium insertion/extraction up 950 °C. However, the performance was further enhanced through surface-modification using a silver deposition and vacuum heating. The surface modification led to a decrease in both the charge transfer resistances and the film resistances attributed to SEI. Surface-modified MCF700 showed the largest current density, highest discharge capacity, and the lower impedance. Likewise, surface-modified samples produced at 800 °C and 950 °C also showed good performance and low impedances. This study clearly demonstrates that both the heat-treatment temperature and surface treatment significantly impact electrochemical performance. Further measurements comparing electrochemical profiles in lithium- and sodium-cell revealed a reduction in resistance for surface-modified samples, with sodium showing the lowest impedance (∼10 Ω). Interestingly, the pristine samples showed higher resistance in sodium-cell, suggesting that the surface modification was essential for improving the sodium insertion/extraction. We hypothesize that the surface modification process could eliminate functional groups or impurities on the MCF surface, which are potentially impeding rapid insertion/extraction reactions. These findings provide fundamental insights into the use of soft carbons for SIB applications.
This work was partially supported by the Asahi Glass Foundation. We greatly appreciate the contribution of Jun-ichi Saito, the principal researcher at Japan Atomic Energy Agency, who gave us proper advice on handling sodium. We greatly appreciate Zachary T. Gossage and Ryoichi Tatara at Tokyo University of Science for fruitful discussion.
The data that support the findings of this study are openly available under the terms of the designated Creative Commons License in J-STAGE Data at https://doi.org/10.50892/data.electrochemistry.23500659.
Yuki Fujii: Conceptualization (Lead), Data curation (Lead), Formal analysis (Lead), Investigation (Lead), Methodology (Equal), Writing – original draft (Lead)
Keisuke Sugata: Data curation (Equal), Methodology (Lead)
Yukikazu Omura: Data curation (Supporting), Methodology (Lead)
Narumi Kubota: Methodology (Lead)
Kento Kisa: Methodology (Equal)
Hiroaki Sofuji: Methodology (Equal)
Junji Suzuki: Conceptualization (Equal), Funding acquisition (Lead), Project administration (Lead), Resources (Lead), Supervision (Lead), Writing – review & editing (Equal)
The authors declare no conflict of interest in the manuscript.
Y. Fujii: ECSJ Student Member
J. Suzuki: ECSJ Active Member
