Partially graphitic structure-assisted hard carbon derived from lignin for sodium-ion battery anodes

Biomass and carbonization

Larch (Larix kaempferi) biomass was pretreated using a DES to obtain CF and LF. The DES used in this study was synthesized by mixing choline chloride and lactic acid in a weight ratio of 1:10. It was synthesized by mixing choline chloride and lactic acid in a weight ratio of 1:10. The DES solution (200 g) was combined with 5 g biomass and pretreated in an oil bath at 130 °C for 9 h. The pretreated biomass was then separated into solid and liquid phases through filtration. The solid CF was sequentially washed with acetone and distilled water, and subsequently dried at 45 °C. The liquid LF was diluted ten-fold with distilled water and allowed to stand for 12 h. The diluted liquid fraction was then centrifuged at 4,000 rpm and 4 °C, and the resulting precipitate was collected. The precipitate was further washed with 10% ethanol to eliminate any residual DES and subsequently freeze-dried. Prior to carbonization, all samples were dried at 120 °C for 24 h to remove moisture. After drying, 30 g of the dried samples were carbonized at 1,300 °C for 2 h under a N₂ atmosphere in a horizontal furnace.

Characterization of materials

The biomass composition was determined using the National Renewable Energy Laboratory method. The ash content was determined using the Technical Association of the Pulp and Paper Industry T211 om-02 method at a temperature of 775 ± 25 °C over a 6-h period. Inductively coupled plasma mass spectrometry (ICP-MS) using a Nexion 300X instrument (PerkinElmer, Waltham, MA, USA) was employed to quantify the inorganic components in the biomass. The chemical composition and structure of LF and CF were analyzed using Attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectroscopy (Spectrum 400, PerkinElmer Ltd., Buckinghamshire, UK). The analysis was conducted with a spectral resolution of 4 cm^-1, a scanning range of 4,000-500 cm^-1, and four scans per minute. The elements were identified using X-ray photoelectron spectroscopy (XPS, PHI 5000 VersaProbe, ULVAC-PHI Inc., Osaka, Japan). Sugars and degradation products in the hydrolysate were analyzed using high-performance liquid chromatography (HPLC, Waters e2695, Alliance, USA) equipped with an Aminex HPX-87H column (300 × 7.8 mm², Bio-Rad, USA) and a refractive index detector (Waters 2414 System, USA). The analysis was performed with a 5 mM sulfuric acid solution as the mobile phase at a flow rate of 0.6 mL/min for 55 min. Thermogravimetric analysis (TGA; NETZSCH, Germany) was conducted in an air atmosphere from 25 to 1,000 °C at a heating rate of 5 °C/min. Field-emission scanning electron microscopy (FE-SEM; JSM-6700F, JEOL, Japan) was used to examine the structural morphology of the samples. High-resolution transmission electron microscopy (HR-TEM; JEM-2200FS, JEOL, Japan) was utilized to obtain TEM images and high-angle annular dark-field images. X-ray diffraction (XRD; Rigaku D/Max 2500/PC, Japan) analyses were conducted using Cu Kα radiation (λ = 0.154 nm) at 40 kV and 200 mA. Cross-sectional images of the electrodes were obtained using focused ion beam scanning electron microscopy (FIB-SEM; Helios G4 UC, Thermo Fisher Scientific, USA). Raman spectroscopy (LabRAM HR, HORIBA Jobin Yvon SAS, Japan) was performed using an Ar-ion laser.

Electrochemical measurements

The anodes were fabricated at room temperature and subsequently dried in a vacuum oven at 80 °C for 10 h. After drying, Half-cell tests were conducted using CR2032 coin cells, which were assembled in a dry room under controlled conditions. Anodes were prepared using a slurry consisting of 80 wt.% active material, 10 wt.% Super P (TIMCAL Graphite & Carbon Co., Belgium), and 10 wt.% poly(acrylic acid) (PAA, 2 wt.% solution in distilled water; Sigma-Aldrich). PAA was selected over the commonly used a poly(vinylidene fluoride) (PVDF) binder due to its superior adhesion, ability to accommodate volume changes, and enhanced chemical stability^[23]. The slurry was uniformly applied on Al foil using the doctor-blade technique, followed by drying in a vacuum oven at 80 °C for 10 h. The electrolyte solution was prepared by dissolving 1 M NaClO₄ in a 1:1 (v/v) mixture of ethylene carbonate (EC) and diethyl carbonate (DEC) with 5 vol% fluoroethylene carbonate (FEC) as an additive, following established protocols for SIB electrolytes^[24]. The separator used was a polypropylene membrane (Celgard 2400). For evaluating Na-ion storage performance of the anodes in SIBs, half-cells were constructed, where the working electrode was a hard carbon electrode, and the counter electrode was sodium metal. Electrochemical charge-discharge cycling tests of the half-cells were conducted between 0.01 and 2.0 V vs. Na/Na⁺ using a potentiostat/galvanostat (WPG 100, WonATech) at room temperature. Charging (sodiation) was performed at 0.2 C to 0.01 V, followed by a constant voltage step at 0.01 V until the current dropped to 0.02 C, and discharging (desodiation) was carried out at 0.2 C. The LF and CF-based hard carbon anode was prepared with a mass loading of ~0.45 mg/cm², ensuring accurate control of the weight of the active material. Rate performance was tested at 0.2 C, 0.5 C, 1 C, and 2 C with 5 cycles per rate, followed by long-term cycling at 0.5 C for 100 cycles to evaluate stability and capacity retention. The current densities and specific capacities were determined based on the mass of the active material. Electrochemical impedance spectroscopy (EIS) measurements were conducted using a BioLogic SAS VSP over a frequency range of 0.1 Hz to 100 kHz.

Characterization of materials

Table 1 lists the chemical composition of the biomass. The raw materials (RMs) consisted of 28.09% lignin and 44.88% glucan, with galactan being the main hemicellulose component. DESs are employed to decompose lignin and hemicellulose by decomposing the ester and ether bonds in the cell walls of biomass^[25]. Therefore, lignin was effectively decomposed to obtain CF, which is consistent with previously reported results^[26]. The CF was primarily composed of glucan (78.71%) and contained a small amount of lignin, whereas the LF was composed of 86.12% lignin and a small amount of glucan. The ether and ester bonds were not completely decomposed during the DES pretreatment; therefore, the LF contained a small amount of glucan^[25]. Figure 1 shows the crystallinity of RM, CF, and LF. RM and CF exhibit peaks at 16.4°, 22.6°, and 34.4°, which are typical peaks corresponding to the crystalline region, indicating that the biomass has a crystalline structure^[26,27]. The crystallinity of CF increased from 41.34% to 60.81%, as compared with that of RM, which is attributed to the relative glucan content increase caused by the hemicellulose and lignin decomposition^[26]. The DES pretreatment did not significantly alter the crystal structure. In contrast, LF exhibited a broad peak at 22.6° corresponding to the amorphous structure of lignin^[28], which is consistent with the chemical composition results [Table 1].

Figure 1. XRD patterns of RM, CF, and LF.

The structural properties of each fraction were investigated using FTIR spectroscopy [Figure 2]. RM exhibits peaks corresponding to cellulose at 3,700-3,000, 1,376, 1,112, 1,061, 1,035, and 898 cm^-1. These peaks are related to the -OH stretching, bending vibration of the C-H group, skeletal vibration of C-OH, ester bond of C-O-C, C-O stretching, and C-O-C stretching in the β-1,4-glycosidic linkage in cellulose, respectively^[29-31]. Peaks corresponding to hemicellulose are apparent at 1,388, 1,251, 1,163, 1,080, 1,043, and 993 cm^-1. The peaks at 1,388, 1,043, and 1,251 cm^-1 correspond to the C-O stretching and acetyl groups^[31,32]. The peaks appearing at 1,163 and 993 cm^-1 correspond to arabinogalactan, confirming the hemicellulose structure of the larch biomass^[31]. The peaks at 1,508, 1,420, 1,225, and 1,034 cm^-1 are related to lignin and correspond to the C=C stretching and C-H and C-O stretching of the methoxyl group^[33]. The CF FTIR spectrum exhibits peaks corresponding to cellulose, similar to that of RM, with a wide peak at 1,035 cm^-1 being clearly visible after the DES pretreatment, which is attributed to the increase in the cellulose content after the DES pretreatment [Table 1]. In contrast, the DES pretreatment reduced the intensities of the peaks corresponding to lignin and hemicellulose. The intensities of the peaks corresponding to cellulose and hemicellulose for LF decreased, whereas those corresponding to lignin increased. Peaks corresponding to the stretching vibration related to the aromatic ring skeleton of lignin are clearly visible at 1,600, 1,510, and 1,455 cm^-1, which indicate that lignin was well-separated and that its aromatic ring structure did not change significantly^[25], which is consistent with the chemical composition results [Table 1]. Table 2 lists the elemental analysis results of RM, CF, and LF. The C content in CF was lower than that in RM because of the delignification of the biomass. The DES pretreatment reduced the lignin content and increased the cellulose ratio, which is consistent with the chemical composition analysis results [Table 1]. However, the C and O contents in LF, as compared with those in RM, increased to 62.17% and decreased to 32.28%, respectively. This was attributed to the lignin condensation reaction caused by the organic acids contained in the DES solution^[34]. The O/C ratio of the CF increased to 0.86, indicating an increase in the number of O-containing functional groups^[35], suggesting that the C-O bond in the lignin structure was broken^[34].

Figure 2. ATR-FTIR spectra of RM, CF, and LF.

The CF and LF samples carbonized at 1,300 °C were denoted as CF-1300C and LF-1300C, respectively. The surface morphology of the samples was analyzed using FE-SEM as shown in Figure 3. Figure 3A and B shows images of CF and LF before carbonization, respectively, indicating that the two samples exhibit different colors and particle sizes. Figure 3C and D shows FE-SEM images of CF and LF, respectively, before carbonization, and Figure 3E and F presents FE-SEM images of the CF and LF, respectively, after carbonization at 1,300 °C. The surface morphologies of the carbonized samples are not significantly different. The CF sample exhibits holes of a specific size, which are commonly observed in biomass. Figure 4A and B shows HR-TEM images of the two samples. CF-1300C exhibits a relatively amorphous structure compared to LF-1300C, as evidenced by the diffuse patterns observed in the fast Fourier transform (FFT). In contrast, LF-1300C exhibits a mixed partially graphitic and amorphous structure. The FFT analysis confirmed the presence of a crystalline structure with an interlayer spacing of approximately 0.43 nm. The interlayer spacing of LF-1300C is conducive to the smooth intercalation of Na ions, rendering it a suitable anode for SIBs.

Figure 3. Photographic images of (A) CF and (B) LF. FE-SEM images of (C) CF, (D) LF, (E) CF-1300C, and (F) LF-1300C.

Figure 4. HR-TEM images of (A) CF-1300C and (B) LF-1300C. The inset images show the diffraction patterns obtained via FFT. (C) XRD patterns and (D) Raman spectra of CF-1300C and LF-1300C. TGA results for (E) CF-1300C and (F) LF-1300C.

In addition to HR-TEM, XRD and Raman spectroscopy were conducted to examine the graphitized structures. Figure 4C shows the XRD patterns of CF-1300C and LF-1300C, which exhibit peaks corresponding to the (002) reflection and an ordered graphitic domain reflection. These reflections originate from the inter-graphene layer stacking, whereas the (100) reflections correspond to ordered graphitic domains. The intensities of the (002) and (100) peaks of LF-1300C were higher than those of CF-1300C, suggesting that LF-1300C has a more stable crystallographic structure than amorphous-like C. These findings were substantiated by the Raman spectroscopy results, which show that CF-1300C and LF-1300C exhibit the D- and G-bands at approximately 1,350 and 1,590 cm^-1, respectively [Figure 4D]. The D- and G-bands are critical indicators of the structural characteristics of C materials. The D-band, observed at approximately 1,350 cm^-1, is attributed to structural defects, disorder, or edge effects in sp² carbon materials. This band is typically absent in highly crystalline graphite but becomes pronounced with the introduction of structural irregularities. The D-band is attributed to the A1g breathing mode of six-membered aromatic rings, which becomes active in the presence of disorder within the C lattice and serves as a valuable indicator of the defect density in the material. The G-band at approximately 1,580 cm^-1 results from the in-plane bond-stretching motion of sp²-bonded C atoms. The G-band corresponds to the E_2g vibrational mode of the graphitic lattice and serves as an indicator of the quality and integrity of the graphene planes. The I_D/I_G ratio, derived from the intensities of the G- and D-bands, serves as an indicator of defect density in carbon-based materials. For CF-1300C and LF-1300C, the I_D/I_G ratios were found to be 1.08 and 0.99, respectively. The higher I_D/I_G ratio of CF-1300C than that of LF-1300C suggests that LF-1300C exhibits a higher degree of crystallinity. All samples contained numerous sp² C atoms, with both samples exhibiting comparable results. Figure 4E and F show the TGA curves of CF-1300C and LF-1300C, respectively. Neither sample exhibits weight loss up to 500 °C; however, the weight significant decreases at approximately 550 °C. Carbon typically oxidizes in air at temperatures exceeding 500 °C, leading to weight loss. However, C materials with crystalline structures, such as graphite, exhibit higher oxidation temperatures^[36]. Figure 4E and F indicate that LF-1300C exhibits a higher weight-loss temperature than CF-1300C, suggesting its stronger crystalline C structure.

XPS analyses were conducted to confirm the binding properties of the samples and elucidate their elemental composition and chemical state. The XPS spectra of CF-1300C and LF-1300C were subjected to optimized curve-fitting procedures within the binding energy range of 280.0-292.0 eV, as shown in Figure 5. Table 3 lists the atomic and calculated non-functional and functional C atom contents for both samples. Both samples exhibited C and O contents of > 96% and < 4%, respectively, on their surfaces. The deconvoluted C1s XPS spectra of CF-1300C and LF-1300C, presented in Figure 5, demonstrate that both samples are primarily composed of graphitic carbon (sp2), which constitutes over 68% of the total carbon species.

Figure 5. C1s peaks in the XPS spectra of (A) CF-1300C and (B) LF-1300C.

Electrochemical characterization of biomass

The electrochemical properties of CF-1300C and LF-1300C as SIB anode materials were thoroughly investigated [Figure 6]. While the precise mechanism of Na storage in hard carbon materials remains a subject of debate, two distinct potential regions are generally recognized, namely, a sloping region from 2.0 to 0.15 V associated with the storage of Na⁺ ions in various sites within the C structure, including defects, edges, surfaces, and pores, and a plateau region below 0.1 V primarily attributed to the storage of Na⁺ ions within the graphitic structures of the C materials^[37].

Figure 6. Electrochemical performance of CF-1300C and LF-1300C: (A and B) first GCD curve at 0.2 C in the potential window of 0.01-2.0 V, (C) rate performance at various current densities, and (D) cycling performance at 0.5 C.

Figure 6A and B shows typical galvanostatic charge-discharge (GCD) curves for CF-1300C and LF-1300C in the Na half-cell at a rate of 0.2 C. The initial charge/discharge capacities of CF-1300C and LF-1300C are 232/331 and 261/350 mAh g^-1, respectively, with corresponding ICEs of 71% and 74%. LF-1300C exhibits slightly higher ICE values and specific capacities than CF-1300C. As described in "Section Characterization of materials", LF-1300C exhibits partially graphitic structures, which results in a longer flat region below 0.1 V in the discharge graph, as compared with that of CF-1300C. This suggests that Na+ ion intercalation plays a more significant role in LF-1300C, leading to enhanced electrochemical performance.

LF-1300C also demonstrates lower irreversibility than CF-1300C, which is attributed to the lower fraction of amorphous carbon. Amorphous carbon regions tend to immobilize Na+ ions within disordered structures, contributing to irreversible capacity loss. In contrast, the partially graphitized regions in LF-1300C provide more ordered pathways for Na+ intercalation and extraction, reducing irreversible capacity loss and enhancing long-term cycling stability. Additionally, the balance between graphitic domains and amorphous regions in LF-1300C plays a crucial role in achieving stable electrochemical performance.

The rate capabilities of CF-1300C and LF-1300C were evaluated through GCD tests at various current densities ranging from 0.2 to 2 C. Following these tests, a recovery cycle at 0.5 C was performed to assess capacity recovery after high-rate cycling [Figure 6C]. The specific capacities of CF-1300C at current densities of 0.2, 0.5, 1, and 2 C are 246, 177, 110, and 68 mAh g^-1, respectively, whereas LF-1300C exhibits higher capacities of 263, 228, 182, and 127 mAh g^-1 at the same current densities. Both samples exhibit capacity decay with increasing current density but recover close to their initial values at 0.5 C, indicating excellent structural stability.

The superior rate performance of LF-1300C is attributed to its partially graphitized structure, which enhances Na+ intercalation efficiency and electrical conductivity compared to the amorphous CF-1300C. The graphitic domains in LF-1300C facilitate faster charge transfer and improved Na+ diffusion kinetics, leading to enhanced performance under high-rate conditions. To quantify this conductivity, an EIS analysis was conducted. As shown in the Nyquist plot in Supplementary Figure 1, LF-1300C exhibits a smaller semicircle diameter in the high-frequency region, indicating lower charge-transfer resistance. These results highlight the critical role of partial crystallinity in enhancing rate performance and overall electrochemical properties.

The capacity retention rates for CF-1300C and LF-1300C from 0.2 to 2 C were 27% and 48%, respectively. Specifically, CF-1300C and LF-1300C outperformed phenolic resin-derived hard carbon (311 mAh g^-1, 63% ICE) and sorghum stalk waste-derived hard carbon (259 mAh g^-1, 59% ICE) in initial discharge capacity and ICE, as shown in Table 4, while LF-1300C’s rate capability (127 mAh g^-1 at 2 C) and 48% retention from 0.2 C to 2 C aligned closely with high-performing waste tea-derived hard carbon (282 mAh g^-1, 69% ICE), as shown in Table 4^[38-40]. Additionally, LF-1300C’s 84% capacity retention improvement over CF-1300C after 100 cycles at 0.5 C exceeds that of polyvinyl chloride-derived hard carbon (389 mAh g^-1, 70% ICE) in stability^[41]. Notably, LF-1300C’s initial capacity and ICE are comparable to or higher than those of biomass-derived hard carbons reported in recent reviews and studies^[42-44], suggesting that its partial crystallinity enhances electrical conductivity and thereby contributes to superior electrochemical performance.