Manipulating Interphase Chemistry by Endogenous Doping Toward High-Performance Hard Carbon Anodes for Sodium-Ion Batteries

Highlights

• Based on the Maillard reaction principle, an endogenous doping strategy was developed to induce the formation of a rich-inorganic solid–electrolyte interphase (SEI) layer on a hard carbon anode.

• The hard carbon anode with inorganic-enriched SEI layer delivers enhanced rate, high initial coulombic efficiency and stable cycling performance.

• The assembled full cell with a Na₃V₂(PO₄)₃ cathode exhibits excellent cycling stability over 700 cycles, achieving a capacity retention of 89.2% at 1 C with an N/P ratio of 1.12.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40820-026-02124-9.

Abstract

The practical of hard carbon (HC) anodes in sodium-ion batteries is primarily limited by their unsatisfactory initial coulombic efficiency (ICE), cycling stability and rate performance, which are closely related to their interphase chemistry and microstructure. Herein, a unique manipulating interphase chemistry strategy by endogenous N/S doping is proposed to simultaneously achieve the both issues. Specifically, a series of reducing sugars and amino acid have been proven to trigger the Maillard reaction, thereby enabling endogenous N/S doping and microstructural design for HC anodes. Endogenous doping facilitates the formation of an inorganic-enriched solid–electrolyte interface (SEI) layer on cycled HC, which can effectively accelerate ion transport kinetics and reduce side effects for enhanced rate, ICE, cycling performance and reversible capacity. Meanwhile, the increase in the number of closed pores boosts both the platform capacity and cycling stability of HC. Consequently, the features HC anodes demonstrate a splendid reversible capacity (363 mAh g⁻¹ at 0.05 A g⁻¹), superior cycling performance (over 2500 cycles with 79% retention at 5.0 A g⁻¹) and adequate ICE (89%). The assembled full cell with Na₃V₂(PO₄)₃ cathode exhibits splendid cycling stability over 700 cycles with capacity retention of 89.2% at 1 C. Surprisingly, the pouch cell with high cathode mass loading of 20.7 mg cm⁻¹ maintains 98.1% capacity retention after 175 cycles at 1 C. This strategy provides new ideas and insights for the design and screening of high-performance HC anodes.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s40820-026-02124-9.

Introduction

Sodium-ion batteries (SIBs) have emerged as promising candidates for large-scale stationary energy storage due to their abundant sodium resources, cost-effectiveness and analogous working principles to lithium-ion batteries (LIBs), hinting them dramatically magnetic for exploring energy technologies [1, 2]. As a pivotal component in SIBs, the anode critically influences electrochemical performance [3, 4]. Among many candidates, hard carbon (HC) has emerged as a preferred anode material for SIBs due to its disordered graphitic domains, expanded interlayer spacing (0.37–0.42 nm), high reversible capacity, tunable porous architecture and wide precursor sources [5–7]. Despite this superior potential, the compromises in critical electrochemical performances, including initial coulombic efficiency (ICE), reversible capacity, rate capability and cycle life, collectively limit the commercial development of HC in SIBs [8].

It is tough and pivotal to achieving a balance between ICE, reversible capacity, rate capability and cycling life of HC [9, 10]. Specifically, a high density of surface defects and porous electrode design are in favor of Na⁺ diffusion, thus enhancing the rate capability of HC [11, 12]. Nevertheless, the abundant surface defects and porous electrode design also accelerate the formation of undesirable solid–electrolyte interphase (SEI) and promote irreversible side reactions, resulting in loss of irreversible capacity, ICE and cycle life [13, 14]. To balance both high ICE and capacity of HC, the key lies in prioritizing the creation of adequate sodium storage spaces over electrochemically active surfaces. Recent studies indicate that refining electrode structure design, such as constructing closed or semi-closed pores [15, 16], can simultaneously increase sodium storage space without adding specific surface area, thereby achieving high ICE and capacity for HC. Unfortunately, the simultaneous improvement of the other two metrics (cycling stability and rate capability) remains a challenge. As commonly acknowledged, the properties of solid–electrolyte interphase (SEI) layer are closely related to interfacial stability and Na⁺ transfer kinetics, thereby influencing the cycling stability and rate capability of HC [17]. Therefore, establishing a robust and satisfactory SEI layer is vital for enhancing the cycling stability and rate capability of HC anodes. For example, through atomic-level cobalt doping to modulate the structure of HC, and concurrently promoting the formation of an inorganic-rich SEI layer, rapid sodium-ion conduction and sustained interface stability are successfully achieved [18]. Furthermore, esterification treatment of cellulose successfully achieved catalytic dehydration and activation, leading to an optimized microporous structure distribution in the hard carbon material. This structural modification facilitated the pre-desolvation of Na⁺ and thereby effectively suppressed electrolyte decomposition [19]. The uniform and inorganic-rich SEI established from C = O (carbonyl) on HC surface could efficiently enhance the cycling stability of HC anodes [20]. Despite significant progress, it remains challenging to precise doping modulates HC structure while maintaining excellent HC/electrolyte interphase. There is an urgent need for a convenient strategy that can simultaneously tailor the microstructure and interphase chemistry of HC, thereby boosting its ICE, reversible capacity, rate capability and cycling life.

Fortunately, in this work, we developed a manipulating interphase chemistry strategy via endogenous N/S doping and microstructural designing to comprehensively enhance the overall performance of HC anodes (Fig. 1a). Specifically, a series of reducing sugars (carbonyl compounds) and amino acids (amino compounds) trigger the Maillard reaction and form various intermediate compounds containing N/S elements, thereby promoting carbon sites nucleation (Fig. 1b). Endogenous N/S doping facilitates the formation of an inorganic-enriched SEI layer on HC, which can effectively accelerate ion transport kinetics and reduce side effects for enhanced rate, ICE and cycling performance. Meanwhile, the increase in the number of closed pores simultaneously boosts the platform capacity and cycle stability of HC. Consequently, the features HC anodes demonstrate a high reversible capacity of 363 mAh g⁻¹ at 0.05 A g⁻¹, superior cycling performance over 2500 cycles with 79% retention at 5.0 A g⁻¹ and adequate ICE of 89%. The assembled full cell with Na₃V₂(PO₄)₃ cathode exhibits splendid cycling stability over 700 cycles with capacity retention of 89.2% at 1 C. Full cells paired with Na₃V₂(PO₄)₃ cathodes deliver an excellent cycling performance over 700 cycles with 89.2% retention at 1 C. The pouch cell with high cathode mass loading of 20.7 mg cm⁻¹ maintains 98.1% capacity retention after 175 cycles at 1 C.

Fig. 1.

a Schematic highlighting the synthesis procedures of 1300-F and 1300-FCA. b Schematic diagram of N/S intrinsic doped carbon spheres synthesized based on the Maillard reaction

Experimental Section

Material Synthesis

Preparation of Materials

Fructose (C₆H₁₂O₆), L-cysteine (C₃H₇NO₂S) and ammonium persulfate (H₈N₂O₈S₂) were purchased from Macklin. All reagents were of analytical grade. Separator, acetylene black, Al foil, Cu foil, electrolyte, polyvinylidene difluoride (PVDF), carboxymethylcellulose sodium, 1-methyl-2-pyrrolidinone (NMP) were bought from Guangdong Canrd New Energy Technology Co., Ltd.

Material Synthesis

1.8 mmol of L-cysteine was dissolved in 70 mL of deionized water under continuous stirring for 1 h. Subsequently, 28 mmol of fructose and 0.2 mmol of ammonium persulfate were added to the resulting solution, followed by stirring for an additional 3 h. The mixture was then transferred into a Teflon-lined autoclave and subjected to hydrothermal treatment at 190 °C for 12 h. After cooling to room temperature, the obtained suspension was centrifuged and repeatedly washed with deionized water and ethanol. The final product was dried in a vacuum oven at 80 °C for 24 h to yield the precursor powder. The precursor powder was subsequently heated to target temperatures of 1100, 1300 and 1500 °C at a heating rate of 5 °C min⁻¹ under an argon atmosphere for 3 h, yielding the materials 1100-FCA, 1300-FCA and 1500-FCA, respectively. For comparative purposes, hard carbon was also prepared by directly carbonizing fructose under the same conditions at 1300 °C without the addition of L-cysteine and ammonium persulfate. All chemical reagents were used as received without further purification, and deionized water was used throughout the process.

Material Characterization

Morphologies of hard carbon were observed by scanning electron microscopy (SEM, TESCAN MIRA3), transmission electron microscope (TEM, TECNAI G2F20) with additional energy-dispersive spectroscope (EDS), high-resolution transmission electron microscope (HRTEM, TECNAI G2F20). Structures of hard carbon were characterized by X-ray diffraction (XRD, Rigaku D/Max-2500) with Cu Kα radiation (λ = 0.15406 nm). X-ray photoelectron spectroscopy (XPS, Thermo Scientific K-Alpha) was used to investigate chemical composition and SEI component after cycling. Fourier-transform infrared spectroscopy (FTIR, Thermo Scientific iN10) spectra were characterized in the range of 500–4000 cm⁻¹. Raman spectra were obtained using the LabRam HR Evolution Raman spectrometer (Thermo Fischer DXR). The specific surface area, pore size distribution and adsorption/desorption isotherms were analyzed under a nitrogen atmosphere using a Micromeritics ASAP 2460 surface area and porosity analyzer. Small-angle X-ray scattering (SAXS) was tested using the Xeuss 3.0 (France). In situ Raman testing was recorded during the initial cycle at 0.05 A g⁻¹. The carbon content was confirmed by thermogravimetric analysis (TG, Netzsch STA449C) under an air atmosphere from 30 to 800 °C. TOF–SIMS measurements were conducted with an IONTOF M6 spectrometer. 30 keV bunched Bi³⁺ ion beam was used as the primary ion beam (Ion current 0.4 pA). DRT analysis from the EIS data was calculated by MATLAB R2022 with a toolbox of DRT-TOOLS developed by the research group of Professor Francesco Ciucci.

Electrochemical Measurement

The hard carbon material was blended with a suitable quantity of deionized water to form a uniform slurry containing 80 wt% active substance, 10 wt% binder (carboxymethylcellulose, CMC) and 10 wt% acetylene black. This mixture was then applied onto a copper (Cu) foil support and dried at 60 °C for 24 h. Following this, the coated foil was processed into disk-shaped electrodes with a diameter of 12 mm, each exhibiting an active material loading of approximately ~ 1.5 mg cm⁻². Electrochemical characterizations were carried out through the assembly of CR2032 coin cells inside a glovebox filled with argon gas, where the concentrations of oxygen and moisture were controlled below 0.1 ppm. The cell configuration included metallic sodium as the reference electrode, a glass fiber membrane as the separator and a 1 M solution of sodium hexafluorophosphate (NaPF₆) in diglyme as the electrolyte. Galvanostatic charge–discharge tests on the half-cells were executed using a NEWARE Battery Test System (CT-4008 T-5V10mA-164, Shenzhen, China), operating within a potential range of 0.01 to 3.0 V relative to the Na⁺/Na reference at standard room conditions. GITT investigations were implemented with the LAND Battery Test System (CT2001-A, Wuhan, China). Cyclic voltammetry (CV) assessments were conducted across a voltage interval of 0.01–3.0 V at varying scan rates between 0.1 and 1 mV s⁻¹ using a CHI660 electrochemical instrument. Electrochemical impedance spectroscopy (EIS), including in situ EIS, was analyzed using an Admiral Squidstat electrochemical measurement device.

For further exploring the electrochemical performance of 1300-FCA, NVP//1300-FCA full cells were performed with 1300-FCA as anodes, and Na₃V₂(PO₃)₄ (NVP) as cathodes. The NVP electrodes were prepared by mixing 80 wt% NVP, 10 wt% acetylene black and 10 wt% PVDF to form a slurry mixture in an NMP solution. The slurry was coated on Al foil and dried at 80 °C in a vacuum oven for 12 h. After that, the Al foil was punched into disks with a diameter of 10 mm and the mass loading of NVP cathode was approximately ~ 2.5 mg cm⁻². Then, the activated 1300-FCA anode and NVP cathode were constructed for full cells where the N/P ratio was approximately 1.10–1.20. To test the best-performing pouch cells, the activated 1300-FCA anode and NVP cathode were punched into rectangular piece. The mass loadings of anode and cathode were approximately 8.8 and 20.7 mg cm⁻², respectively. For cycle testing, the pouch cells were charged and discharged at 1 C (1 C = 117 mA g⁻¹, based on the active mass of the cathode side) between 2.0 and 3.8 V.

Theoretical Calculations

Density functional theory (DFT) simulations were conducted to examine the structural properties of both unaltered and modified hard carbon systems under the generalized gradient approximation (GGA), utilizing the Perdew–Burke–Ernzerhof (PBE) exchange correlation functional within the Vienna Ab initio Simulation Package (VASP). The ionic cores were represented using the projected augmented wave (PAW) approach, while valence electrons were treated using a plane-wave basis set with an energy cutoff of 450 eV. To account for long-range dispersion forces, the DFT-D3 correction scheme was applied. Structural relaxation was performed under a force convergence threshold of 0.02 eV Å⁻¹ or lower. Furthermore, Brillouin zone integration was carried out using a single k-point grid based on the Gamma-centered scheme. All the atoms are relaxed in all the calculations.

All adsorption energies (E_ads) used in this paper were defined as:

$$E_{\mathit{ads}}=E_{\mathit{total}}-\left(E_{\mathit{bare}},+,E_{\mathit{ref}}\right)$$ 1

where E_total is the total energy of the adsorbate interacting system and E_bare and E_ref are the energies of the Gra and the isolated Na, respectively.

A Porod method proposed by Stevens and Dahn was employed on small-angle X-ray scattering (SAXS) data to probe the pore size of the carbon sample:

$$I\left(Q\right)=\frac{A}{q^a}+\frac{B_1{a}_1^4}{{\left(1,+,a_1^2,Q^2\right)}^2}+D$$ 2

where I(Q) is the scattered intensity as a function of Q, A is a scale factor for the surface scattering at low Q, and B₁ is a scale factor for the pore scattering which is proportional to the total surface area of pores. D is a constant background, and a₁ is the characteristic length and associated with the radius of a spherical pore, by $R=a_1·\sqrt{10}$.

The galvanostatic intermittent titration technique (GITT) measurements were performed using a pulse current of 0.05 A g⁻¹ for 20 min with a rest interval of 40 min. This estimation is based on Fick’s second law:

$$D=\frac{4}{\pi \tau }{\left(\frac{m_B{V}_M}{M_{B}S}\right)}^2{\left(\frac{{\mathrm{\Delta }E}_s}{{\mathrm{\Delta }E}_{\tau }}\right)}^2$$ 3

where D represents Na⁺ diffusion coefficient (cm² s⁻¹), τ (s) represents pulse duration, m_B (g) represents the active mass of the electrode, V_M (cm³ mol⁻¹) represents molar volume, M_B (g mol⁻¹) represents molecular weight, S (m² g⁻¹) represents the active surface area of the electrode, and ∆E_S and ∆E_τ are obtained from the GITT curves.

CV curves at various scan rates of all electrodes were carried out to analyze the kinetic behavior of sodium-ion diffusion. The b value can be estimated based on the formula:

$$i=a{v}^b$$ 4

$$\text{log}\left(i\right)=b·\text{log}\left(v\right)+\text{log}(a)$$ 5

where v represents the sweep rate, i indicates the response current, and a and b are adjustable parameters, representing different sodium storage forms. b value close to 0.5 means diffusion behavior, whereas b value close to 1 indicates pseudo-capacitance behavior.

Results and Discussion

Analysis of the Microstructure and Pore

In this study, various HC precursors were synthesized through hydrothermal reactions between fructose and a series of amino acids. Subsequently, fructose and cysteine were selected as the primary reactants for further investigation. Following precursor synthesis, HC materials of endogenous N/S doping were obtained via high-temperature pyrolysis at 1100, 1300 and 1500 °C. They were designated as 1100-FCA, 1300-FCA and 1500-FCA, respectively. The HC materials derived from fructose without amino acid participation were labeled as 1300-F. In this synthesis process, fructose as a reducing sugar can generate the Maillard reaction with amino acid during hydrothermal treatment [21, 22]. Fig. 1a illustrates the synthesis of hard carbon on the basis of Maillard reaction via hydrothermal reaction and high-temperature pyrolysis as compared to unreacted method. The Maillard reaction is a non-enzymatic browning reaction, which is a series of oxidation, cyclization, dehydration and condensation reactions between carbonyl and amino compounds (Fig. 1b) [23]. In this process, ammonium persulfate (APS) serves as a catalyst and initiator, the formation of melanoidins, higher unsaturated aldehydes, 5-hydroxymethylfurfural and a variety of derived N/S-containing elemental compounds provide a rich source of carbon nuclei to form [24]. This leads to the formation of carbon spheres in which the doping of N/S elements favors intrinsic doping over marginal doping so that the defect concentration is stabilized [25, 26].

As shown in Fig. S1, excluding the influence of pyrolysis temperature, the carbon yield of 1300-FCA (25.27%) is significantly higher than that of 1300-F (6.19%). Moreover, related thermogravimetric analysis (TGA) and differential scanning calorimeter (DSC) are shown in Fig. S2. The thermal stability of 1300-FCA is superior to that of 1300-F, indicating the formation of molecular cross-linking after Maillard reaction [27]. The morphology and microstructure of hard carbon were characterized via scanning electron microscopy (SEM) and high-resolution transmission electron microscopy (HRTEM) analyses. As shown in Fig. S3, HC precursor prepared by hydrothermal treatment displays sphere morphology with smooth and dense structure. And 1300-F, 1100-FCA, 1300-FCA and 1500-FCA still maintain the dense carbon sphere structure after high-temperature carbonization, which is conducive to reducing the contact area between the HC anodes and the electrolyte, thereby influencing the interfacial behavior (Fig. S4). But the clustering of 1300-F is more severe, resulting in limited carbonation. HRTEM was further performed to explore internal carbon microstructure. 1300-F mainly exhibits highly disordered and marginal long-range graphitic crystallites with a narrow carbon interlayer spacing of 0.370 nm (Figs. 2a and S5). For 1100-FCA, 1300-FCA and 1500-FCA, the interlayer spacing gradually decreases with increasing carbonization temperature, indicating a significant influence of temperature on the HC structure [28, 29]. Furthermore, 1300-FCA emerges an abundant turbulent structural composed by randomly and closely stacked carbon layers, forming closed pores as marked through the red circles. For 1500-FCA, the higher pyrolysis temperature brings about visible graphite-like structures and narrower interlayer spacing (0.368 nm), generating an increase in the number of closed pores. What is more importantly, the carbon interlayer spacing of 1300-FCA is extremely expanded to 0.388 nm, which is conducive to facilitate the intercalation of Na⁺ into carbon layer. The disordered and clear carbon lattice stripes are attributed to doping treatment of Maillard reaction, which introduces N/S element to expand the interlayer spacing. The appropriate interlayer spacing and existence of closed pores are beneficial to provide more Na⁺ storage sites.

Fig. 2.

a SEM images of 1300-F and 1300-FCA, HRTEM images of 1300-F, 1100-FCA, 1300-FCA and 1500-FCA. b FTIR spectra. c XRD patterns. d Physical parameters of samples. e Raman spectra. f N₂ sorption isotherms. g SAXS patterns. h SAXS 2D spectral structure of 1300-F, 1100-FCA, 1300-FCA and 1500-FCA

Fourier-transform infrared spectroscopy (FTIR) of all samples exhibits identical functional groups. Compared to 1300-F, the C–C and C = C bonds of 1100-FCA, 1300-FCA and 1500-FCA display slight variations, indicating that the presence of endogenous doping has altered the sp²/sp³-carbon hybridization structure (Fig. 2b). Ulteriorly, X-ray photoelectron spectroscopy (XPS) was performed to investigate the elemental composition of the samples, and the N and S are detected on the sample surface on account of the Maillard reaction (Figs. S6 and S7). The higher pyrolysis temperatures promote the graphitization of hard carbon, reduce structural defects and facilitate the transformation of disordered sp³-carbon domains into more ordered sp²-carbon layers [30]. As the carbon framework becomes increasingly graphitic and structurally uniform, the number of active sites capable of stabilizing heteroatoms diminishes, thereby resulting in the release of small molecules in gaseous form. Moreover, the content of C = O group has significantly increased for 1300-FCA. Compared with 1300-F, 1100-FCA and 1500-FCA, the presence of oxidized N implies that electron cloud distribution near C = O is stabilized, which is conducive to C = O as active sites that catalyze the preferential reduction of salt and form a uniform SEI layer [18]. X-ray diffraction (XRD) and Raman spectroscopies measurements were performed to further grasp the microcrystalline structure and defect characteristics of HC. The XRD patterns display that all samples appear two amorphous peaks at around 23° and 43°, corresponding to the (002) and (100) crystal planes of HC, respectively (Fig. 2c). The interlayer spacings of 1300-F, 1100-FCA, 1300-FCA and 1500-FCA were determined to be 0.370, 0.443, 0.388 and 0.368 nm, respectively, based on XRD pattern analysis of hard carbon, which is consistent with HRTEM results (Table S1). Moreover, the XRD patterns are performed by profiles fitting method for analyzing the micro-crystallinity of HC (Fig. S8). The (002) diffraction peak is fitted three types, which is highly disordered domain with interlayer spacing of above 0.4 nm, pseudo-graphitic domain of 0.37–0.40 nm and graphite-like domain of below 0.37 nm [31]. With the increase in carbonization temperature, the stacking thickness L_c and lateral width L_a display an incremental tendency (Fig. 2d). This indicates that the structure of HC tends to graphite-like, resulting in a smaller interlayer spacing. Hence, applicable carbonization temperature is crucial adjective. Compared with 1300-F (24.7%), the pseudo-graphitic of 1300-FCA is approximately 43.6%, indicating that an appropriate pyrolysis temperature combined with endogenous doping can effectively fine-tune the pseudo-graphitic domain, increase the interlayer spacing and reduce surface defects. Specially, the I_D1/I_G ratio of 1300-FCA (1.40) is close to 1300-F (1.36), suggesting no significant increase in the number of defects by this endogenous doping strategy (Fig. 2e, S9) [32]. The stability of defect concentration not only effectively reduces interfacial side reactions at the hard carbon anode, but also directly determines the ICE. The I_D1/I_G ratio of 1100-FCA (1.82) and 1500-FCA (1.35) demonstrates that pyrolysis temperature significantly influences the defect density of HC materials [33, 34]. Accordingly, determining the appropriate pyrolysis temperature is crucial for optimizing performance of HC anode. The analysis of porous microstructure for the three samples was investigated via the nitrogen adsorption/desorption isothermal curves as shown in Fig. 2f. The specific surface areas of 1300-F, 1100-FCA, 1300-FCA and 1500-FCA are extremely low at 0.6526, 2.7574, 0.3213 and 0.1595 m² g⁻¹, respectively (Table S1). This verifies that the intrinsic doping constructed by the Maillard reaction can maintain low specific surface area, resulting in more reversibly defects and turbostratic domain for Na⁺ storage [35]. Meanwhile, the pore size distribution curves (Fig. S10) reveal that all samples are mainly micropores (around 1.85 nm). The low specific surface areas and abundant micropores are conductive to suppress excessive decomposition of electrolyte and overgrowth of SEI layer [36]. Ulteriorly, the small-angle X-ray scattering (SAXS) was employed for investigating the characteristics of closed pores that cannot be detected. As displayed in Fig. 2g, the SAXS patterns of all samples reveal shoulder-shaped peaks at around 0.1–1 Å, which is attributed to scattering from the closed nanopores in HC matrix [37]. Moreover, combined with the 2D SAXS profiles (Fig. 2h), 1300-FCA and 1500-FCA also exhibit stronger scattering intensities than 1300-F and 1100-FCA in center region, indicating the increase in closed pores. Fitting the SAXS curves is conductive to further investigate the average closed pores aperture of HC (Figs. S11 and S12). With the carbonization temperature increasing, the closed pore aperture obviously enlarges, corresponding to 1.78 nm of 1100-FCA, 2.18 nm of 1300-FCA and 2.30 nm of 1500-FCA. This result demonstrates that high temperature can promote short-range ordered graphite microcrystals of HC materials, resulting in forming the closed nanopores via the stacking of ordered carbon layers. However, the excessively high carbonization temperature can reduce defects and increase graphite-like extent, leading to fusion of vast adjacent closed pore carbon layers. The situation leads to larger closed pore aperture, as well as gradually decreased closed pores number. The appropriate carbonization temperature can effectively facilitate that the open pores and curved graphite lattice change into closed pores. Interestingly, at the same carbonization temperature, the scattering intensity and closed pore aperture of 1300-FCA (2.18 nm) is larger than 1300-F (1.94 nm), implying the existence of more and larger closed pores. It is in favor of boosting the plateau capacity and providing more space for Na⁺ storage and sodium quasi-metallic clusters.

Electrochemical Performances of HC Anodes

The above studies investigate the microscopic structural characteristics of endogenously doped HC in order to further elucidate its impact on the electrochemical performance of SIBs. A variety of electrochemical characterizations were employed to evaluate the electrochemical performance of HC anodes. As shown in Figs. 3a and S13, the cycling voltammetry (CV) measurements were performed in initial three cycles. In the first cycle, the CV curve exhibited irreversible features associated with SEI formation, specifically a significant decomposition of the electrolyte and consumption of active sodium ions at the interface of the hard carbon anode. A weak reduction peak at 0.6 V in the first cycle of 1300-FCA further indicated the formation of the SEI layer. The high degree of overlap between the second and third cycles suggests that the hard carbon electrode rapidly achieved a highly reversible state, indicating that the SEI layer possesses good ionic conductivity, facilitates ion transport and effectively suppresses further electrolyte decomposition. The galvanostatic discharge–charge (GCD) curves at the current density of 0.05, 0.1 and 0.2 A g⁻¹ also overlap after the second cycle in Fig. 3b-d and Table S2. The GCD curves display two evident regions, which correspond to the sloping region (above 0.1 V) and the plateau region (below 0.1 V). The reversible capacity and ICE of 1300-FCA are 363 mAh g⁻¹ and 89% at 0.05 A g⁻¹, which are much higher than those of 1300-F (209 mAh g⁻¹ and 79%), 1100-FCA (295 mAh g⁻¹ and 84%) and 1500-FCA (312 mAh g⁻¹ and 88%). Notably, with increasing current density, the capacity and ICE of comparative samples are extremely reduced (Fig. S14), which is owing to electrochemical polarization effect and interfacial kinetic limitations [38]. The severe polarization at high rates impedes the diffusion and storage of Na⁺ within closed pores and pseudo-graphitic lattices so that the capacity damages in the plateau region [39]. In addition, the accelerated rate of Na⁺ embedding/de-embedding gives rise to electrolyte decomposition and SEI overgrowth at the HC/electrolyte interface, which further reduces ICE. Surprisingly, 1300-FCA still maintains excellent capacity and ICE at the higher current density, revealing a splendid interfacial effect. Furthermore, by separating the capacity contributed by sloping/plateau region, 1300-FCA reveals a conspicuous augment in both sloping and plateau capacities (Fig. 3e-g). This demonstrates that 1300-FCA is constructed with expanded carbon interlayer spacing, plentiful closed pores and supererogatory Na⁺ storage sites due to the endogenous doping formed through the Maillard reaction.

Fig. 3.

a CV curves of 1300-FCA. b-d GCD curves of the first three cycles at 0.05, 0.1 and 0.2 A g⁻¹. e–g Specific capacities of sample contributed from the slope and plateau region at different current densities. h Rate performance. i Charge–discharge profiles of 1300-FCA at different current densities. j 1300-FCA anode compared with previously reported hard carbon anodes. k Cycling performance at 0.1 A g⁻¹. l Cycling performance at 0.3 A g⁻¹. m Cycling performance at 5.0 A g⁻¹. n 1300-FCA anode compared with previously reported hard carbon anodes for various electrochemical performance. o 1300-FCA anode compared with diverse amino acid

Rate performance is a significant metric for HC anodes in the field of SIBs. All the samples were evaluated at the various current densities (Figs. 3h and S15). The reversible capacities of 1300-FCA are 363, 352, 333, 319, 303, 293, 259, 219 and 136 mAh g⁻¹ at 0.05, 0.1, 0.2, 0.3, 0.5, 0.8, 1.0, 2.0 and 5.0 A g⁻¹, respectively. When the current density turns back to 0.05 A g⁻¹, a reversible capacity of 363 mAh g⁻¹ can be revitalized, further indicating the interphase chemistry stability and reversibility. The corresponding GCD curves of various rate cycles are illustrated Figs. 3i and S16. The outstanding voltage stability of 1300-FCA is demonstrated during various high-rate discharge–charge processes, which is conducted to rapid Na⁺ transfer kinetics. These imply that the HC anodes have excellent SEI layer. The 1300-FCA anode constructed on the principle of the Maillard reaction shows an excellent Na⁺ storage performance, standing out among the previous reported in the literature, as compared in Fig. 3j. Moreover, the long-term cycling performance was further performed at the different current densities. As depicted in Fig. 3k, 1300-FCA demonstrates excellent cycling stability with nearly 100% capacity retention (344 mAh g⁻¹) after 150 cycles at 0.1 A g⁻¹, which is superior to 1300-F, 1100-FCA and 1500-FCA. When the current densities increase, the cycling stabilities of HC anodes display more significant differences. 1300-F delivers short circuit after 38 cycles at 0.3 A g⁻¹ (Fig. 3l), corresponding to polarization effect and interfacial kinetic limitations mentioned above. In contrast, 1300-FCA can still preserve outstanding cycling stabilities at the higher current density. As displayed in Figs. S17 and S18, 1300-FCA exhibits the capacity retention of 95% after 300 cycles at 0.5 A g⁻¹ and 94% after 700 cycles at 1.0 A g⁻¹, respectively. In addition, at the ultrahigh current density of 5.0 A g⁻¹, 1300-FCA delivers a magnificent capacity retention of 79% after 2500 cycles, corresponding to a capacity decay as low as 0.0084% per cycles (Fig. 3m). The unique endogenous doping and microstructure of 1300-FCA are conducted to the admirable electrochemical properties, including reversible capacity, ICE, rate performance and cycling stability, which outperforms the most previously reported HC anodes (Fig. 3n). Specially, the Na⁺ storage performance of 1300-FCA verifies that the Maillard reaction of fructose with cysteine establishes endogenous doping to realize the regulation of the microstructure and interface of HC. Furthermore, we derive inferences through analogical reasoning based on the underlying mechanism of this strategy. The more HC precursors were prepared by reacting fructose with various kinds of amino acids. There is no excessive graphitization in the series of HC (Fig. S19). Meanwhile, this series of HC demonstrates excellent reversible capacity, ICE and rate performance, especially containing the amino acid of N/S element (Figs. 3o and S20). In addition to fructose, reducing sugars such as glucose, xylose and lactose can undergo the Maillard reaction with amino acids, implying that a very large variety of HC can be prepared. Based on the principle of the Maillard reaction, the strategy can be further investigated until the optimal reducing sugars and amino acids are selected as a reaction matrix.

Analysis of Sodium Storage Mechanism

To elucidate the Na⁺ storage kinetic of HC anodes during the charge–discharge process, CV curves of 1300-F and 1300-FCA were conducted at various scan rates (from 0.1 to 1.0 mV s⁻¹), revealing characteristic oxidation–reduction peaks (Fig. S21). The samples consistently sustain sharp and evident redox peaks at elevated scan rates, reflecting the similar electrochemical behavior of the both. Furthermore, their reaction kinetics can be investigated by b values according to equation i = av^b [40]. The b values of 1300-F and 1300-FCA are adjacent, but 1300-FCA exhibits faster kinetics attributed to slightly higher b values. To further quantify the diffusion and capacitive contribution, the response currents at a fixed potential are isolating into diffusion-controlled and capacitive-controlled according to equation i = k₁v + k₂v ^1/2 (Fig. S22) [41]. Compared to 1300-F, 1300-FCA demonstrates higher diffusion-controlled contribution at different scan rates, indicating the excellent diffusion kinetics. In order to acquire further insight into the Na⁺ storage kinetics of HC anodes, the galvanostatic intermittent titration technique (GITT) is performed to estimate Na⁺ diffusion coefficients ($D_{{\mathit{Na}}^{+}}$) during sodiation and desodiation (Figs. 4a and S23-S24). It can be distinctly observed that 1300-FCA displays a higher $D_{{\mathit{Na}}^{+}}$ values than 1300-F, especially at low voltage, implying a prominent kinetics of Na⁺ transfer [42]. The $D_{{\mathit{Na}}^{+}}$ values vary insignificantly in the sloping region, corresponding to Na⁺ adsorption on the surface defect. In contrast, the $D_{{\mathit{Na}}^{+}}$ values exhibit a dramatic decrease in the plateau region, revealing slower diffusion due to the intercalation adsorption and intercalation of Na⁺ through carbon layer. With a large number of Na⁺ embedded in the short-range ordered graphite microcrystals, the charge repulsion of the Na⁺ at the restricted interlayer spacing increases significantly, leading to impeded diffusion. The $D_{{\mathit{Na}}^{+}}$ values emerge a rebound condition at approximate 0.05 V, demonstrating that the Na⁺ ions fill the nanopores and accumulate in the closed pores [16]. Meanwhile, the cumulative Na⁺ ions gradually configurate sodium cluster in the closed pores, resulting in generating from quasi-metallic sodium [43]. To further comprehend the Na⁺ storage mechanism of 1300-FCA, the existence of the above quasi-metallic sodium is authenticated via a phenolphthalein solution color reaction. Figure 4b illustrates that the ethanol–phenolphthalein solution of immersing the electrode pole pieces at various potentials demonstrates color change. Specially, the obvious pink color arises only at 0.05 and 0.01 V throughout the charging and discharging process. The dissolution of quasi-metallic sodium in ethanol makes the solution alkaline, resulting in turning the phenolphthalein red. Meanwhile, the solution exhibits no apparent color change at other potentials, corresponding to adsorption and intercalation of the Na⁺. Through comprehensive analysis of the above characterization, the Na⁺ storage mechanism of HC can be divided into three stages, including adsorption of Na⁺ on the surface and defects, interlayer adsorption and intercalation, closed pores filling and sodium cluster (Fig. 4c) [44].

Fig. 4.

a Na⁺ ions diffusion coefficients of 1300-F and 1300-FCA. b Typical GCD curves of 1300-FCA and the color changes of 1300-FCA at different potentials after immersion in an ethanol solution containing phenolphthalein. c Schematic diagram of various Na⁺ storage stages for 1300-FCA anode. d GCD curves of 1300-FCA. e In situ Raman spectra of the 1300-FCA electrode during the first charge–discharge cycle. f-g In situ Raman spectra of 1300-FCA during the discharge and charge process. h-j In situ EIS at different potentials during the first discharge and charge process and the corresponding DRT results of 1300-F. k-m In situ EIS at different potentials during the first discharge and charge process and the corresponding DRT results of 1300-FCA

To further elucidate the Na⁺ storage mechanism, the in situ Raman was conducted by real-time monitoring during the charging and discharging process. As depicted in Fig. 4d, e, the Raman spectra of 1300-FCA show the two apparent characteristic peaks at around 1350 and 1580 cm⁻¹, corresponding to the D band and G band [45]. Concretely, during the sodiation process, the G band exhibits a significant redshift due to the Na⁺ intercalation graphene layer (Fig. 4f). Notably, the intercalation graphene layer of Na⁺ incorporates the possibility of filling closed pores. Moreover, the strength of D bond gradually decreases during the sodiation process, indicating that the Na⁺ storage transforms from defects adsorption to intercalation and pores filling. Meanwhile, the breathing vibrations of carbon rings are suppressed on account of Na⁺ clusters covering the defect sites, resulting in the widens or fades in the plateau region [15]. In contrast, during the desodiation process, the strength of G band regains the initial states (Fig. 4g), implying that the Na⁺ in the closed pores interacts with carbon atoms in sp² configurations so that the E_2g stretching mode of the G band is impacted [46]. These manifest the splendid reversibility for Na⁺ storage of 1300-FCA anode. In addition, the excellent reversibility of HC anodes is connected with interfacial electrochemistry. Electrochemical impedance spectroscopy (EIS) was employed for investigating the interfacial electrochemical behavior (Fig. S25). Generally, the Nyquist plots show two parts, which consist of a semicircle in the low-frequency region and a sloping plot in the high-frequency region, corresponding to the charge transfer resistance (R_ct) and the diffusion of Na⁺ [47]. 1300-FCA demonstrates a lower charge transfer resistance than other samples, indicating fast surface and internal diffusion kinetics within HC anodes. To further explore SEI evolution and interface dynamics of HC anodes, in situ EIS of 1300-F and 1300-FCA was performed (Fig. 4h, k). All electrochemical impedance spectra were fitted by the equivalent circuits. During the discharging process, the low-frequency semicircle gradually minishes, thus relating to electrolyte decomposition and SEI formation. Apparently, the R_SEI of 1300-FCA is smaller than that of 1300-F (Fig. 4i, l). During the charging process, the R_SEI of 1300-F increases significantly due to the desolvation and transport of Na⁺ across the SEI layer. On the contrary, the R_SEI of 1300-FCA demonstrates stabilized smaller value during the desodiation process, signifying that the homogeneous and inorganic-rich SEI is conducive to interface diffusion kinetics. To vindicate the above statement, the distribution of relaxation times (DRT) was performed (Figs. 4j, m and S26), verifying that 1300-FCA exhibits lower cell polarizations than 1300-F [48]. Therefore, 1300-FCA prepared based on endogenous doping obtains a fast interfacial transportation kinetics.

Interphase Chemical Analysis of HC Anodes

The excellent interface electrochemical behavior is recognized to hinge on the SEI layer, and the remaining argumentum demands to reveal the relationship between the SEI composition and Na⁺ transfer kinetics on the HC/electrolyte interphase. 1300-F and 1300-FCA electrodes after 20 cycles were inquired through XPS testing with various etching depths, thus attaining element distribution and chemical bonding at the depths of 0, 10, 20 and 30 nm (Figs. 5a, and S27). The C 1s peaks at C–C (284.8 eV), C-O (285.4 eV) and C = O (286.6 eV), O 1s peaks at C-O (533.5 eV) and C = O (532.5 eV) are detected at various etching depths, which are acknowledged for the organic ROCO₂Na/(CH₂OCO₂Na)₂ due to the decomposition of solvents [49]. Notably, the CO₃ (289.8 eV) of C 1s peak and Na–O (531.7 eV) of O 1s peak correspond to the inorganic components Na₂CO₃ and Na₂O, which are formed via Na⁺ and solvent-decomposition products [50]. Furthermore, the Na-F (685.1 eV) of F 1s peak is spotted, deriving from the decomposition of NaPF₆ [51]. As the etching depth increases, 1300-FCA exhibits significantly more inorganic components than 1300-F (Fig. S28), which can restrain the dissolution and unremitting reactions of SEI layer, especially for NaF component. This is consistent with the splendid cycling performance of HC anode. In addition, HRTEM analysis directly manifests the existence of abundant inorganic (Figs. 5b and S29). 1300-FCA displays a consecutive and homogenous SEI layer (3.20 nm), and the whole SEI layer demonstrates evident lattice fringes, corresponding to interplanar crystal spacing of Na₂CO₃ (310/0.257 nm), Na₂O (220/0.206 nm) and NaF (200/0.232 nm) [52]. In contrast, 1300-F exhibits a gradually increasing SEI layer (7.32 nm), implying a heterogeneous evolution of SEI layer during formation. Meanwhile, the SEI layer shows a less lattice belonging and plentiful pronounced amorphous region, attributing to organic components. On the contrary, the SEI layer of 1300-F exhibits a continuous and asymmetrical growth, which might lead to the rupture of SEI layer. This is the reason for the poor cycling stability. Inversely, the homogenous and thin SEI layer with more NaF of 1300-FCA can provide a higher Na⁺ conductivity, which distinctly enhances the rate performance. To more directly investigate the distribution and type of SEI layer, time-of-flight secondary-ion mass spectrometry (TOF–SIMS) was utilized for supplying detailed insights via 3D visualization. As shown in Figs. 5c and S30, 1300-F intuitively demonstrates a SEI layer with luxuriant organic components throughout the SEI yet fewer inorganic components in the outer SEI surface. Meanwhile, as the sputtering time increases, the intensity of inorganic components continuously declines, especially Na₂F₃⁻ and NaF₂⁻ (Fig. 5d). Compared to 1300-F, the 3D TOF–SIMS visual maps of 1300-FCA display an apparently higher concentration of inorganic components, which is far more than the organic components (Fig. 5e). Furthermore, with increasing the sputtering times, the intensity of inorganic components maintains consistently high (Fig. 5f).

Fig. 5.

a XPS etching spectra of 1300-F and 1300-FCA. b HRTEM images of 1300-F and 1300-FCA after cycling. c-f 3D TOF–SIMS visual maps and depth profiling showing the distribution of inorganic and organic species in the SEI of 1300-F and 1300-FCA. g-h Charge density difference of 1300-F and 1300-FCA. i Optimized structures and corresponding adsorption energies of Na salt adsorbed on 1300-F and 1300-FCA

Based on the above observations and analysis, the endogenously doped 1300-FCA via the principle of the Maillard reaction possesses a stable HC/electrolyte interphase, prompting further investigation to which a series of strategies enhance electrochemical behaviors. Accordingly, the density functional theory (DFT) calculations are employed. The optimized structural models for 1300-F and 1300-FCA are demonstrated in Fig. S31, which are utilized to ascertain the surface adsorption energy of Na⁺. The Na⁺ adsorption energy on 1300-FCA (1.555 eV) is obviously higher than 1300-F (0.185 eV) (Fig. S32), indicating that the intrinsically doped N/S atoms can availably improve the adsorption energy of Na⁺ on the HC [53]. The higher adsorption energy of the 1300-FCA anode toward Na⁺ effectively weakens the interaction between Na⁺ and solvent molecules within the solvation shell, thereby promoting the preferential reduction and deposition of Na⁺. This selective interaction minimizes the parasitic consumption of sodium ions in defect-rich regions and at carbon edges, thus suppressing interfacial side reactions at the electrode/electrolyte interface [54]. Furthermore, the N/S active sites with higher adsorption energy for Na⁺ weaken the interaction between Na⁺ and anions (PF₆⁻), thereby significantly lowering the energy barrier for anion reduction and decomposition, and promoting the catalytic formation of inorganic SEI components such as NaF, Na₂O and Na₂CO₃ [55]. The preferential formation of an inorganic-rich SEI layer predominantly composed of NaF effectively suppresses continuous electrolyte decomposition by minimizing the generation of organic by-products [52]. Thereby, the overall capacity of 1300-FCA anodes is enhanced. Notably, for the charge density difference analysis, 1300-FCA exhibits a more pronounced charge transfer around the N/S atoms (Fig. 5g, h), thus reinforcing the electrochemical activity and electronegativity [56]. In additional, the curved graphene layers imply that the endogenous doping significantly enlarges interlayer spacing, thus facilitating Na⁺ intercalation behavior. To further investigate the influence of endogenous doping for the HC/electrolyte interface, the deposition process of inorganic salt components on the surface of HC is simulated. As depicted in Figs. 5i and S33, the adsorption energies of 1300-FCA anode (Na₂CO₃, Na₂O, NaF) are 1.170, 1.822 and 1.350 eV, respectively, which is higher than 1300-F. This suggests that intrinsically doped N/S atoms might change the electronic structure of the carbon matrix, inducing the formation of inorganic-rich SEI layer on the electrode [57]. Ulteriorly, the solvated Na⁺ attains desolvation and diffusion across the SEI layer, which is largely influenced by electrolyte and SEI composition [58]. The inorganic-rich SEI layer of 1300-FCA can steady the solid/liquid interface, thus promoting the rapid diffusion of Na⁺. These results visibly reveal that the endogenous doping constructed by the principle of the Maillard reaction realizes the synergistic regulation of microstructure designing and interface engineering, thus obtaining the splendid reversible capacity, rate performance and cycling stability simultaneously.

Electrochemical Performances of Full Cell

Due to the excellent Na⁺ storage performance of the 1300-FCA half cells, the full cells are assembled with Na₃V₂(PO₄)₃ as cathode (Fig. S34) and 1300-FCA as anode to further evaluate practical application (Fig. 6a). The capacity ratio of the negative electrode to the positive electrode (N/P) is maintained at the range of 1.1–1.2. As presented in Fig. S35, the luminous diode with CSU can be lit by the full cell, verifying that the feasible strategy shows energy storage applications. Figure 6b is illustrated for the GCD curves of NVP and 1300-FCA half-cell. The NVP//1300-FCA full cell demonstrates a high ICE of 94.0% at the voltage window of 2.0–3.8 V. Notably, the NVP//1300-FCA full cell exhibits reversible capacities of 115, 113, 111, 108 and 97 mAh g⁻¹ at the current density of 0.1, 0.2, 0.5, 1 and 2 C, respectively (1 C = 117.6 mA g⁻¹) (Fig. 6c, d). Moreover, when the current density returns to 0.1 C, the reversible capacity can restore the initial capacity value, demonstrating the excellent rate performance. Specially, the full cell exhibits splendid cycling stability with capacity retention of 89.2% after 700 cycles at 1 C (Fig. 6e, f) and capacity retention of 94.2% after 300 cycles at 0.5 C (Fig. S36), delivering better cycling performance than that of most previous reports (Fig. 6g and Table S3). To further evaluate scale-up potential, the pouch cell is fabricated (Fig. 6h), approaching to commercial requirements (high areal mass loading of 20.7 mg cm⁻² and low N/P ratio of 1.20). Furthermore, NVP//1300-FCA pouch cell demonstrates a remarkable cycling stability with a capacity retention of 98.1% after 175 cycles at 1 C, thus substantiating that the strategy based on Maillard reaction is generally applicable.

Fig. 6.

a Schematic of SIBs with 1300-FCA anode and NVP cathode. b Capacity matching of 1300-FCA anode and NVP cathode. c Rate performance of full cell. d GCD curves of NVP//1300-FCA full cell at various current densities. e Cycling performance of NVP//1300-FCA full cell. f Charge and discharge curves corresponding to the full cell cycle performance of NVP//1300-FCA. g NVP//1300-FCA full cell compared with previously reported full cell. h Schematic diagram of a Na-ion pouch cell in a configuration of NVP//1300-FCA. i Long-term cycling test of NVP//1300-FCA pouch cells

Conclusion

In summary, inspired by the principle of the Maillard reaction, we proposed a unique manipulating interphase chemistry strategy by endogenous N/S doping to achieve a balance between ICE, reversible capacity, rate capability and cycle life of HC. Endogenous N/S doping facilitates an inorganic-enriched SEI layer on HC, which effectively accelerates ion transport kinetics and reduces side effects, thereby enhancing rate capacity, ICE and cycling performance. Additionally, the increased closed pores simultaneously boost the platform capacity and cycle stability of HC. Consequently, 1300-FCA anodes demonstrate a high ICE of 89% and superior cycling stability (a capacity retention of 79% after 2500 cycles at 5.0 A g⁻¹). Impressively, the assembled NVP//1300-FCA full cells also exhibit a capacity retention of 89.2% after 700 cycles at 1 C, and the NVP//1300-FCA pouch cell with mass loading of 20.7 mg cm⁻² delivers a capacity retention of 98.1% after 175 cycles at 1 C. This approach offers innovative perspectives for developing and optimizing advanced HC anode.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

H. L. designed detailed experimental schemes. A.P. and S. Z. conceptualized and supervised this work. H. L. and Z. W. conducted electrochemical tests for coin cells and pouch cells. Y. Y and Y. Z. contributed in situ Raman data. Y. Z contributed ToF–SIMs test. Q. Z. and S. L. provided guidance and advice on this work. H. L. wrote the paper, and S. Z. and A. P. assisted in the revision and improvement of the paper. All authors discussed the results and commented on the manuscript.

Declarations

Conflict of interest

The authors declare no interest conflict. They have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.