Figure 1 depicts the schematic of the preparation of Sb/C–SiOC materials. Sb-MOFs were synthesized using the previously reported procedure. By replacing terephthalic acid (TPA), which is the backbone of MOFs, with the amino-terephthalic acid, Sb-MOFs containing a –NH2 group were prepared. SbCl3 exists as individual ions in methanol solvent when an aqueous solution containing amino-TPA is added, three amino-TPAs bind to one Sb ion. On the opposite end of these amino TPAs, successive Sb ions bind to produce MOFs [54, 55]. The Sb metal lies in the center of the synthesized Sb-MOFs, and because the NH2 group exists as a functional group, it exhibits hydrogen bonding, allowing it to operate as a driving force for the Sb-MOFs to be well dispersed in silicone oil. Therefore, when the prepared Sb-MOFs were put into silicone oil, the hydrogen bonds between the NH2 groups of Sb-MOFs and oxygen present in the siloxane structure of silicone oil contribute the uniform dispersion of Sb-MOFs in silicone oil. Additionally, the self-dispersion characteristics of the hydrocarbon terephthalic acid can also contribute to the dispersion. When this suspension was subjected to heat treatment, the silicone oil underwent a cross-linking reaction at 400°C to form a pre-ceramic [53]. It then underwent heat treatment at 900°C to complete ceramization [32]. Simultaneously, the MOFs carbonized to porous carbon above 650°C, and finally, an Sb/C–SiOC complex wherein the Sb/C composites are uniformly inserted into the SiOC matrix was obtained [51].
The sample crystalline structures were examined and confirmed using Powder X-ray diffraction (XRD). All samples exhibited peaks consistent with hexagonal Sb (space group R-3m, JCPDS # 00-005-0562). This confirms that no impurities other than amorphous C and SiOC existed in the final material and proved that these amorphous materials do not affect the Sb crystal structure during final material synthesis. In the XRD pattern of the Sb–SiOC material, a weak Sb2O3-related peak appeared around 27.6° (shown in inserted graphs). This suggests that in the Sb–SiOC material produced by simply dispersing pure Sb in silicone oil without any driving force then heat treatment, not all Sb is well covered by SiOC and it is exposed, causing some Sb particle surfaces to oxidize to Sb2O3 [57]. Also, in pure Sb, peaks associated with Sb2O3 were pronounced, demonstrating that surface oxidation occurs during storage when no surface protective matrix exists [32, 56]. Sb2O3 material is also employed as a conversion-alloying type anode, but for Sb to express its capacity, Na must first react with O and be desorbed into NaO2, which continuously consumes Na [58, 59]. Additionally, its intrinsically low electronic conductivity and substantial volume expansion are disadvantageous in terms of rate performance and electrochemical stability [57]. However, the oxidized Sb2O3-related peak does not appear in the Sb/C–SiOC composites owing to the presence of the SiOC matrix, which is an amorphous protective layer on the surface [58]. The SiOC and C materials surrounding the Sb particles in the Sb/C–SiOC sample were amorphous; therefore, no related peaks were observed. The X-ray diffraction patterns of pure SiOC material exhibited one broad peak near 23°, which was consistent with those reported in previous studies [52, 53, 60]. Therefore, this demonstrates that the amorphous SiOC was effectively produced in the material synthesized using this method.
To determine the Sb composition ratio in the Sb/C–SiOC composites, the TGA of Sb-based materials was performed between 25–900°C with a heating rate of 5°C·min− 1 in air atmosphere. Pure Sb was oxidized to Sb2O3 up to 600°C, and its mass increased. Then, at temperatures above 600°C, evaporation and oxidation from Sb2O3 to Sb2O4, or thermal decomposition of Sb2O3 to SbO2, decreased its mass [57, 61, 62]. The weight gain of approximately 10.29% at 300–400°C is attributed to the oxidation of Sb to Sb2O4, which is consistent with prior research on Sb and Sb2O3 [62]. Assuming that the Sb material in each sample exhibits the same behavior as pure Sb during TGA analysis, the Sb content of each sample was calculated using Equations S1 and S2 (In supplementary material) [53]. In the Sb/C sample, all carbon evaporated to CO2 gas, while the evaporation of Sb2O3 and Sb2O4 left behind only 48.72% of the mass [63, 64]. Using Equation S1 and the TGA results of the Sb/C–SiOC sample, the Sb content of the Sb/C–SiOC composites were calculated to be 52% (Supplementary material).
The Raman spectra of the Sb/C–SiOC materials in Fig. 2c exhibit the characteristic Eg and A1g peaks of Sb at 110 and 147 cm− 1, respectively [34]. The two broad peaks near 1337.8 and 1590.11 are attributed to the disorder-induced D band and graphitic G band of carbon, respectively [65]. The D band and G band positions of the Sb/C–SiOC sample and the pure SiOC sample are identical, showing that SiOC and free carbon domains were successfully produced in the Sb/C–SiOC sample [26]. ID/IG value, which indicates the graphitization degree, is an essential factor in determining the electrical conductivity of carbon-based materials [57]. It has been reported that as graphitization increases, electrical conductivity decreases [66]. Due to the low SiOC contents, the ID/IG value of the Sb/C–SiOC composites were slightly lower than that of the pure SiOC material, but the high ID/IG value (0.98) was close to 1, indicating that the electrical conductivity is greater than that of pure Sb [53]. This is expected to result in improved electrode kinetics and high rate capability [65].
To confirm the morphological significance of the process using Sb-MOFs to obtain the final material in which Sb was uniformly inserted into the SiOC matrix. Figure 3 shows the SEM images of the Sb/C–SiOC material and the Sb–SiOC material obtained by dispersing Sb-MOFs and pure Sb particles separately in silicone oil and then heat-treating them. The Sb/C–SiOC composites were observed to be uniformly distributed and devoid of large agglomerations of several hundred nanometers-sized particles, similar to pure Sb shown in Fig. S1a [32, 46]. In contrast, the surface morphology of the Sb–SiOC sample resembled that of the pure SiOC material shown in Fig. S1b, and it was confirmed that the Sb particles that were unevenly embedded in the SiOC matrix were agglomerated on the surface [53]. Additionally, SiOC and Sb materials were agglomerated separately or combinedly, demonstrating that the size of Sb-SiOC particles grew substantially [47, 51]. Therefore, the Sb–SiOC sample particle sizes were non-uniform, with small particles measuring several hundred nanometers, and large particles ranging between 10 to 20 µm. This confirms that using Sb-MOFs is effective for embedding pure Sb into the SiOC matrix so that the final product does not agglomerate as a whole and consists of small, homogeneous particles [60].
HR-TEM and EDS-mapping analyses were performed to investigate the morphology and microstructure in detail. The HR-TEM image of the Sb/C–SiOC composites revealed that the Sb nanoparticles were uniformly embedded in the SiOC matrix, and the Sb/C–SiOC particles measured several hundred nanometers in diameter (about 500 nm in Fig. 4a). Additionally, pure Sb nanoparticles were confirmed to be uniformly distributed throughout the SiOC matrix without dark agglomerated portions. In the enlarged images of the internal Sb particles (Fig. 4c–4d), a lattice pattern corresponding to the (0 1 2) plane of hexagonal Sb was observed, and the interplanar distance was measured to be around 0.31 nm, which is consistent with the results of previous studies [33, 34, 61]. This indicates that Sb particles were not oxidized during embedding and that pure Sb was stably embedded in the SiOC matrix [57]. Around pure Sb, an amorphous SiOC matrix exists; however its amorphous nature prevented the observation of lattice patterns [26]. Combined with the XRD results, it is demonstrated that the material resulting from the thermal decomposition of silicone oil is pure SiOC [47]. Figure 4f–4i depict HR-TEM images of Sb–SiOC materials synthesized without Sb-MOFs. As validated by the FE-SEM image, the particle size of the Sb–SiOC material employed for TEM analysis is approximately 1–2 µm, and SiOC and Sb particles were agglomerated separately or together. In contrast to the Sb/C–SiOC material, the Sb particles in the SiOC matrix of the Sb–SiOC material were agglomerated and not uniformly dispersed throughout the SiOC matrix. The EDS mapping image revealed whether the constituent elements of the material and the Sb material embedded in the SiOC matrix were uniformly distributed throughout the particles. The Sb/C–SiOC material seen in Fig. 4e demonstrates that the Sb particles are evenly distributed throughout the SiOC matrix. However, the EDS-mapping image of the Sb–SiOC material in Fig. 4j reveals that the Sb particles within the SiOC matrix are randomly clustered, indicating a dark region in the HR-TEM image. These results indicate that pure Sb materials do not interact with silicone oil owing to the absence of a driving force, as seen by the non-uniform distribution of Sb particles in the SiOC matrix of Sb–SiOC samples [34, 56]. However, in the Sb/C–SiOC composites, the NH2 functional group in the Sb-MOFs established hydrogen bonds with the siloxane group of silicone oil and maintained these bonds until heat treatment [54, 55]. Therefore, the Sb nanoparticles were uniformly inserted into the SiOC matrix.
XPS measurements were performed to confirm the chemical composition and valence state of Sb-based anode materials. The Sb 3d spectrum overlaps that of O 1s and binding energy. The peaks at 540.1 and 530.7 eV are attributable to the oxidation state of Sb3+ (Antimony oxide) in Sb2O5 and Sb2O3, whereas the peaks at 537.9 and 528.5 eV correspond to the oxidation state of Sb0 in pure Sb [34]. Antimony oxide-related peaks such as Sb2O5 and Sb2O3, including pure Sb-related peaks, were evident in all Sb samples [67]. The Si–O–Si related peak near 533 eV existed only in the pure SiOC sample and the Sb/C–SiOC composites containing the SiOC matrix, proving that only the SiOC material was present in both samples [68]. Notable in the Sb spectrum is the difference in peak intensity associated with antimony oxide. The antimony oxide-related peaks of the pure Sb and Sb/C samples showed stronger intensity than the Sb-related peaks, whereas the antimony oxide-related peaks of the Sb/C–SiOC composites showed similar intensity to the Sb-related peaks [69]. These results indicate that the SiOC matrix, which exists solely in Sb/C–SiOC samples, effectively protects the Sb material within the SiOC matrix from antimony oxide formation [26]. Additionally, the low intensity of the Sb2O5 and Sb2O3 related peaks in the Sb/C–SiOC composites indicate that only the sample surface is slightly oxidized [56]. Only the SiOC sample and the Sb/C–SiOC sample exhibited O-Si and Si-O-Si peaks in the Si 2p spectra, indicating that SiOC was well-formed only in these materials [4, 26]. The Si 2p spectrum of Sb–SiOC and Sb/C–SiOC materials exhibits three major peaks. The SiOC material exhibited the largest peak corresponding to SiOx at 103.3 eV, followed by peaks corresponding to Si–C and SiO2 bonding at 101.6 and 103.5 eV, respectively [32, 53]. This agrees well with previous reports on SiOC-containing anode materials, demonstrating that the SiOC material was well formed [47, 70].
Figure 6a depicts the cyclic voltammograms (CV) of Sb/C–SiOC at a scan rate of 0.1 mV·s− 1 in the 0.001–2.0 V voltage range. During the initial cathodic scan, the strong peak at approximately 0.22 V corresponds to the alloying process of Sb and Na [71]. In contrast to the first cathodic scan, the peaks related to the multi-step conversion of crystalline Sb to hexagonal Na3Sb by Na+ insertion occurred at 0.35, 0.53, and 0.68 V in the second scan, indicating a stable sodiation process [72, 73]. In addition to the peak associated with the Sb redox reaction, a broad peak near 0 V is observed, which is commonly attributed to the electrochemical reaction between carbon black (conducting materials) and Na ions [74]. Subsequently, the anodic scan revealed a peak at approximately 0.86 V corresponding to the dealloying process of Na3Sb in both the first and second cycles [32, 52]. Figure 6b depicts the first and second voltage profiles of the Sb/C–SiOC material at a current density of 0.01 A·g− 1. The specific capacity was calculated using the weight of Sb/C–SiOC composites. The Sb/C–SiOC electrode exhibits a long plateau corresponding to the sodiation reaction at 0.5 V (vs. Na/Na+) [74]. At approximately 0.8 and 0.9 V, two plateaus corresponding to desodiation were observed, showed with a sodiation capacity of approximately 672.58 mA·h·g− 1 and a desodiation capacity of 464.99 mA·h·g− 1, and a charge/discharge efficiency of 69.14% [56]. In contrast to the single plateau observed in the first cycle, multiple plateaus were observed in the second cycle between 0.7 and 0.5 V, with a sodiation capacity of approximately 551.60 mA·h·g− 1, desodiation capacity of 468.09 mA·h·g− 1, and coulombic efficiency of 85% [52]. These multiple plateau phenomenon results from a multi-step transformation wherein amorphous Sb produced during the first cycle (desodiation) is continuously transformed into amorphous Na3Sb, Na3Sbhex/Na3Sbcub, and hexagonal Na3Sb by Na ion and sodiation reaction [33]. The CV curve and voltage profile of the Sb/C–SiOC composites are identical to those of the pure Sb material examined previously [34, 62, 64, 75], demonstrating that no material other than Sb significantly contributes to the capacity of Sb/C–SiOC composites. Additionally, CV curves and voltage profiles for Sb nanoparticles synthesized by SbCl3 reduction are shown in the supplementary material (Fig. S2) [34, 57, 61].
The results of comparing the cycle performance of various Sb-based composites are depicted in Fig. 7a. After three formation cycles at a current density of 0.1 A·g− 1, all electrodes were charged/discharged 200 times at a current density of 0.5 A·g− 1. Table S1 lists the specific capacity and capacity retention of each electrode after the first, 100th, and 200th cycles. In the Sb/C–SiOC sample, the initial capacity decreased slightly as the Sb proportion in the total active material decreased owing to the presence of the MOFs-derived porous carbon material that did not contribute to the capacity [41, 61]. However, at a current density of 0.5 A·g− 1, the Sb/C–SiOC composites demonstrated a relatively stable cycle performance with no capacity loss even after 200 cycles [48, 51, 60]. Due to the rigid and elastic nature of the SiOC matrix, which accommodates the volume change of Sb during the cycle, Sb/C–SiOC composites were stable despite their phase transformation during multiple charge/discharge cycles [47, 48]. In contrast, the Pure Sb electrode exhibited a rapid capacity loss, and after 40 cycles, and the cycle performance could no longer be evaluated [32]. After 200 cycles, the Sb/C material obtained by carbonization of Sb-MOFs exhibited unstable cycle performance and rapidly decreasing capacity, with a capacity retention rate of 12.6%. Although the Sb–SiOC electrode accommodates a certain volume change owing to the presence of SiOC, the Sb particles are not completely embedded in the SiOC; consequently, the capacity tends to decrease slowly [52]. Additionally, the pure SiOC material reacted with sodium to provide a very low capacity of 50–60 mA·h·g− 1 (Fig. S3), whereas the Sb/C–SiOC composites revealed a 25% SiOC content based on TGA results [26]. Therefore, SiOC contributes negligibly to the total reversible capacity [26, 32, 60]. Due to the excellent incorporation of the Sb material into the SiOC matric, the synthesized Sb/C–SiOC exhibits the most stable cycling characteristics among various Sb-based materials [32].
EIS measurements were performed before and after 200 cycles, and the resulting Nyquist plots are depicted in Fig. 7b–7c. Table S2 lists the electrolyte resistance (Rs), SEI layer-related resistance (RSEI), and charge transfer resistance (Rct) values obtained using the simplified equivalent circuit [76]. Comparing each R-value before and after the cycle indicates that the Rs value increased by 1.145 Ω and 0.467 Ω in the Sb–SiOC and Sb/C–SiOC samples, respectively, so that the Rs value in the Sb–SiOC sample increased significantly. Additionally, the RSEI value of Sb/C–SiOC appears identical to, or, slightly decreased than its previous value, which can be attributed to the formation and stabilization of a stable SEI film on the electrode during cycling [62, 77]. As shown in Table S2, the Rct of the Sb/C–SiOC electrode increased marginally from before to after 200th cycle, whereas the Rct of the Sb–SiOC electrode more than doubled from 112.9 to 283.7 Ω. These results indicate that embedding Sb particles in the SiOC matrix aids in the formation of a stable SEI layer and improves the structural stability of the material, which may account for the excellent cycling stability of the Sb/C–SiOC composite electrodes [34, 56]. Therefore, the Sb/C–SiOC sample exhibited improved interfacial stability between the active material and the electrolyte and a faster charge transfer rate, suggesting that the Sb/C–SiOC composites reacted without capacity deterioration over prolonged cycling [46, 57].
Figure 8a depicts the desodiation capacity when charging/discharging once for each rate to observe only the rate characteristics, excluding factors related to the cycle performance of the electrode. Sb/C–SiOC sample exhibited desodiation capacities of 437.58, 429.62, 420.58, 406.71, 395.45, 384.32, and 366.83 mA·h·g− 1 at current densities of 0.05, 0.1, 0.2, 0.5, 1, 2, and 5 A·g− 1, respectively. The capacity of Sb/C–SiOC electrode was recovered to 424.58 mA·h·g− 1 with a reduced current density of 0.1 A·g− 1 [52]. The Sb/C–SiOC composites demonstrated an excellent rate performance with a high desodiation capacity of 366.83 mA·h·g− 1 even at a very fast current density of 5 A·g− 1. This indicates that the Sb/C–SiOC electrode has excellent reversibility [32]. As the current density increased, pure Sb material exhibited a sharp decrease in capacity and unstable rate performance [71]. At a current density of 5 A·g− 1, the Sb–SiOC sample demonstrated 29% of its capacity at a current density of 0.05 A·g− 1. At a current density of 5 A·g− 1, the Sb/C sample demonstrated 34% of its capacity at a current density of 0.05 A·g− 1. This suggests that the presence of carbon improves the electrochemical kinetics than SiOC alone [48]. Therefore, the rate capability test revealed that the presence of carbon can kinetically aid the rapid sodiation/desodiation reaction of the electrode [4, 47].
Further investigation of the electrochemical reaction kinetics of the Sb/C–SiOC electrode was conducted by measuring a set of CV curves at scan rates ranging from 0.2 to 2.0 mV·s− 1. The relationship between the peak current (ip) and the scan rate (ν) follows the power law equation, where the parameters a and b can be obtained from the slope of the log (i) – log (ν) plot (Eq. 1) [78, 79].
$$i =a{v}^{b} \left(1\right)$$
The sodium ion storage mechanism of the Sb material can be divided into diffusion-controlled and capacitive behaviors, according to the calculated b-values [61, 80]. Generally, b-value = 0.5 indicates a diffusion-controlled process, whereas b-value = 1 indicates a surface capacitive-controlled process [79, 81]. In most cases, the b-value ranges between 0.5 and 1, indicating a mixed contribution from the two processes [78]. Figure 8c depicts the log (i) – log (ν) plot for the Sb/C–SiOC electrode [80]. The b-values of peaks C1 and C2 were 0.79 and 0.92, respectively, suggesting that the capacitance reaction dominated the reaction kinetics of the Sb/C–SiOC electrode [81–83]. The ratio of the capacitive contribution in the Sb/C–SiOC electrode can be quantitatively estimated using Eq. 2 [82].
$$i\left(V\right)= {k}_{1}v + {k}_{2}{v}^{1/2} \left(2\right)$$
Here, k1ν and k2ν1/2 represent currents generated in capacitive and diffusion-controlled processes, respectively, and are constants under a specific voltage [78, 82]. Figure 8d depicts the calculated capacitance contribution of the Sb/C–SiOC electrode. At a sweep speed of 1 mV·s− 1, the capacitance distribution of the Sb/C–SiOC electrode gradually increases and attains a value of 93.2% as the scan speed increases [33]. This suggests that Sb/C–SiOC electrodes have extremely rapid electrochemical kinetics, which may account for their excellent rate performance [81, 83, 84]. Therefore, the significant capacitive contribution of the Sb/C–SiOC composites can be attributed to the synergistic effect of the presence of MOFs-derived porous carbon (the relevant BET data is depicted in Fig. S4), based on the pseudo-capacitive behavior of Sb as reported in previous studies [26, 33, 85].
Figure 9 depicts the morphology of the electrodes of Sb/C–SiOC and Sb–SiOC before and after the 100th cycle, as determined by FE-SEM. In Fig. 9a–9b, the Sb–SiOC and Sb/C–SiOC electrodes before the cycle exhibited extremely comparable morphologies without cracks. However, after 200 cycles, the Sb–SiOC sample developed significant cracks, which may have been caused by crushing owing to the large volume expansion of the Sb particles that were not embedded in the SiOC matrix. However, some macrocracks in the Sb/C–SiOC electrode were observed after the 200th cycle, but they were significantly reduced compared to the Sb–SiOC electrode. Consistent with the previous cycle performance test, these results indicate that the NH2 group of Sb-MOFs exerted an effective driving force to attract silicone oil and that Sb-MOFs were effectively incorporated into the SiOC matrix [54, 55]. Additionally, it is demonstrated that the SiOC buffer layer effectively accommodated the volume change of the Sb particles during cycling, hence mitigating the cracks in the electrode [52, 53, 60].