Understanding of the sodium storage mechanism in hard carbon anodes

Abstract

Hard carbon has been regarded as the most promising anode material for sodium‐ion batteries (SIBs) due to its low cost, high reversible capacity, and low working potential. However, the uncertain sodium storage mechanism hinders the rational design and synthesis of high‐performance hard carbon anode materials for practical SIBs. During the past decades, tremendous efforts have been put to stimulate the development of hard carbon materials. In this review, we discuss the recent progress of the study on the sodium storage mechanism of hard carbon anodes, and the effective strategies to improve their sodium storage performance have been summarized. It is anticipated that hard carbon anodes with high electrochemical properties will be inspired and fabricated for large‐scale energy storage applications.

INTRODUCTION

With the growing concern of global warming, the energy community is being forced to innovate by replacing traditional fossil energy with renewable energy resources. Thus, there is an urgent demand to build large-scale electrical energy storage systems (EESs) to store wind power, solar power, and other intermittent renewable energy resources.^1,2 In the past several decades, lithium-ion batteries (LIBs) have been considered as the most efficient secondary batteries, due to their outstanding advantages of high energy densities, long cycle life, and no memory effect.^3,4 However, because of the low reserve abundance, uneven distribution, and ever-increasing price of lithium, LIBs can hardly meet the demands for large-scale EESs. Given this, sodium-ion batteries (SIBs) have been regarded as the most promising candidate for EESs, owing to the low cost of sodium resources, a wide abundance of sodium sources, and similar physiochemical properties to lithium.^5–10 A lot of cathode materials can be employed for sodium storage, while the alternative of anode materials is very limited.^11–17 For the anode materials of SIBs, compared to other materials, such as alloys (e.g., Sn, Sb, Se, Ge, and P),^18–22 Ti-based materials (e.g., TiO₂ and Na₂Ti₃O₇),^23–25 metal oxides/sulfides/phosphides (e.g., Fe₂O₃, SnO₂, Sb₂O₃, MoS₂, and FeP₄),^26–29 the carbon-based materials show more fascinating properties, such as high reversible capacity, low working voltage, long cycling stability, high initial Coulombic efficiency (ICE), small volume changes during electrochemical reactions, high electronic conductivity, low cost, and affordable resources. Commercial LIBs usually adopt graphite as the anode material. However, it has been reported that graphite delivers a very limited capacity when used as anode for SIBs, contrary to the behavior in LIBs and K-ion batteries.^30,31 Based on the first-principles calculation results, the low sodium storage capacity in graphite resulted from the energetic instability of Na-graphite intercalation compounds (Na-GICs).^32,33 And Liu et al.³⁰ further demonstrated that among all alkali metals, Na has the weakest binding with graphite substrate, probably because of the competition between the ionization of metal atoms and the ion–substrate coupling. Although it has been demonstrated that expanded graphite and Na-solvent cointercalation in ether-based electrolytes can realize the goal of sodium storage in graphite, the high intercalation voltage and low sodium storage capacity limit their practical applications.^34,35

Experimental and theoretical simulation results indicated that Na⁺ can easily insert into carbon materials with a larger interlayer distance and appropriated defects.^36–38 As a result, amorphous carbon materials have been studied extensively. According to whether they can be transformed into graphite at an appropriate temperature, amorphous carbons can be further divided into soft carbon and hard carbon. Hard carbon is used to describe the amorphous carbon that cannot be transformed into graphite even at temperatures higher than 3000°C, while soft carbon can be readily converted to graphite by heating to 3000°C. Generally, soft carbon exhibits sloping charge/discharge voltage curves. Along with the sloping voltage region, hard carbon also exhibits a low voltage plateau below 0.1 V that leads to a large capacity and a low average redox potential, resulting in a high energy density.³⁹ Furthermore, the ICE of hard carbon is higher than that of soft carbon in general.^40,41 Hence, hard carbon with both superior electrochemical properties and low cost is considered as a state-of-the-art anode material for SIBs. It has been reported that the reversible capacities of hard carbon materials vary between 250 and 480 mAh g⁻¹, which are closely associated with their micro/nanostructures.^42–48 Particularly, the micro/nano structures of hard carbon materials are tunable, and versatile hard carbon materials can be obtained by adjusting the carbon precursors and pyrolysis processes. However, due to the complexity of the hard carbon structures, there is still a lack of consensus on their sodium storage mechanism, which hinders the structural design and electrochemical performance optimization of hard carbon electrodes.⁴⁹ Hence, a comprehensive understanding of the structure–electrochemical property relationships is very important, which has extraordinary significance for developing high-performance anode materials for SIBs.

In this review, we overview the recent progress in hard carbon anodes, with emphasis on sodium storage mechanism studies by various techniques and computational methods. Guided by these results, we try to provide some ideas on how to rationally design and fabricate hard carbon structures with high electrochemical performance. It is believed that this article will help researchers to better understand the development of carbon anode materials and offer some guidance for future materials engineering, and finally, facilitating the practical applications of high energy/power density hard carbon anodes.

MICROSTRUCTURE AND SODIUM STORAGE SITES OF HARD CARBON

Unlike graphite, hard carbon lacks a long-range ordered structure in the plane and stacking directions. The microstructure of hard carbon can be described as a combination of rumpled and twisted graphene sheet fragments consisting of sp² hybridized carbon in a hexagonal network disrupted by defects, such as vacancies, pentagons, heptagons, and heteroatoms.^50–52 Because of the existence of van der Waals forces between carbon layers, part of curved graphene sheets locally stack forming randomly oriented graphitic-like nanodomains with multiple layers, while some other rather twisted and randomly oriented graphene sheets form pores with various sizes in the bulk of materials. It is worth mentioning that the texture of graphitic-like nanodomains (the average width, stacked layer, and interlayer spacing, and so on) and pores (the pore volume, distribution, and average pore size) varies from sample to sample, which is determined by the preparation process and the property of the precursor.^53,54 Despite the complex and versatile structure of hard carbon materials, the main active sites with the ability to uptake sodium ions are listed below (Figure 1):

• 1)

  Adsorption of Na⁺ at the surface of open pores is influenced by the specific surface area, which contributes to the sloping capacity at the beginning of the sloping region.^55,56 It is a surface-induced capacitive process and inevitably causes the decomposition of electrolytes, resulting in irreversible sodium storage capacity.⁵⁷

• 2)

  Adsorption of Na⁺ at defect sites of graphene sheets, including the edges, heteroatoms, vacancies, and so forth, is influenced by the defect degrees of hard carbon materials. This process is related to the sloping capacity.⁵⁸

• 3)

  Insertion of Na⁺ between graphene layers is affected by the texture of graphitic-like nanodomains. This can be further separated into two types: procedural intercalation between graphene layers with phase transition to form Na-GICs, which corresponds to the plateau capacity³¹ and random insertion into graphene layers with a wide energy distribution, which is related to the sloping capacity.⁵⁹

• 4)

  Pore filling with the formation of quasi metallic clusters is influenced by the property of pores, including numbers and sizes of pores. This behavior corresponds to a plateau capacity near the potential of metallic sodium.^60,61

Figure 1. Schematic representation of the microstructure of the hard carbon and the main active sites with the ability to uptake of sodium ions

SODIUM STORAGE MECHANISM IN HARD CARBON MATERIALS Depicting the sodium storage behavior

The sodium storage behavior in hard carbon has long been the research topic, and some typical reaction models have been proposed by different researchers as illustrated in the timeline (Figure 2). Up to now, the sodium storage mechanism of hard carbon materials is still controversial and there are four prevailing models (Figure 3), including the “insertion–adsorption” model,^59,62–65 “adsorption–intercalation” model,^37,66–69 “three-stage” model,^70–72 and “adsorption–filling” model.^73–76 Different models were based on different experimental results on various hard carbon samples; we will comprehensively discuss them as follows.

Figure 2. Timeline of key developments of sodium storage mechanism in hard carbon

Figure 3. Summary of the proposed sodium storage models

In 2000, Stevens and Dahn⁶⁵ first proposed the “insertion–adsorption” mechanism and “house of cards” model to interpret the electrochemical data of glucose-derived hard carbon anode materials (Figure 4A). This model is composed of aromatic fragments randomly stacked like a house of cards, forming parallel graphitic-like nanodomains and regions of nanoscale porosity. They suggested a similar charge storage mechanism of Na⁺ and Li⁺ in hard carbon, where the high-potential sloping region and low-potential plateau region are related to the insertion of alkali metal between carbon layers and adsorption of alkali metal into nanopores, respectively; whereafter, they confirmed this mechanism by in situ wide-angle X-ray diffraction (XRD) and small-angle X-ray scattering studies (SAXS) technologies (Figure 4B,C).⁵⁹ In 2010, Komaba et al.⁶⁴ further proved this mechanism by ex situ XRD, SAXS, and Raman spectroscopy studies.

Figure 4. (A) “House of cards” model for sodium-filled hard carbon. Reproduced with permission: Copyright 2000, The Electrochemical Society, Inc.65 (B) Selected in situ X-ray scattering and (C) in situ SAXS scans for sodium insertion into hard carbon. Reproduced with permission: Copyright 2001, The Electrochemical Society, Inc.59 SAXS, small-angle X-ray scattering studies

However, there are still some experimental phenomena that conflicted with the “insertion–adsorption” mechanism. For example, as the pyrolysis temperature increases, the surface area and total pore volume of the hard carbon material decrease while the plateau capacity appears as a “volcano”-shaped tendency with a maximum plateau capacity at about 1400°C.^68,77,78 Whereafter, some different opinions were proposed to interpret these disparities. Considering that Na⁺ storage in hard carbon is similar to that of Li⁺ in graphite but very different from Li⁺ storage in hard carbon, Cao et al.³⁷ proposed an “adsorption–intercalation” mechanism in 2012. They indicated that the high-potential sloping capacity mainly comes from the adsorption of Na⁺ on the surface and defective sites, while the low-potential plateau capacity corresponds to the intercalation behavior of Na⁺ between graphite layer forming NaC_x. Subsequently, they further verified this mechanism through systematic structural and electrochemical characterizations of cellulose-derived hard carbon prepared at different temperatures using in situ XRD, ex situ nuclear magnetic resonance (NMR), electron paramagnetic resonance techniques, and so forth.^{68,69,79–81} Specifically, from the in situ XRD (Figure 5A), the (002) peak shifts to the low angle, which can be ascribed that graphite sheets with relatively larger spacing are favored for Na⁺ intercalation causing the expansion of interlayer spacing. While another part of graphite sheets with narrow spacing does not allow Na⁺ intercalation and the corresponding peak keeps unchanged.⁶⁸ In 2013, Ding et al.⁶⁶ confirmed this mechanism by systematically studying the relationship between microstructure and electrochemical sodium storage behavior of peat moss tissue-derived hard carbon materials prepared at different pyrolysis temperatures. The ex situ XRD results verified that intercalation occurs below 0.2 V versus Na/Na⁺ in hard carbon materials (Figure 5B,C).

Figure 5. (A) The in situ X-ray diffraction (XRD) mapping of hard carbon electrode during the first discharge–charge at 200 mA g−1 at a temperature of 293 K. Reproduced with permission: Copyright 2017, Wiley-VCH.68 (B) XRD spectra for hard carbon and activated carbon electrodes and (C) dependence of the mean interlayer spacing on discharge voltage. The electrodes discharged to (Ⅰ) 0.2, (Ⅱ) 0.1, (Ⅲ) 0.05, and (Ⅳ) 0.001 V versus Na/Na+. Reproduced with permission: Copyright 2014, American Chemical Society.66 (D) Galvanostatic intermittent titration profile and diffusivity as a function of states of charge (inset) and (E) dQ/dV plot from 0.12 to 0.01 V with corresponding diffusivity values. Reproduced with permission: Copyright 2015, American Chemical Society72

Nevertheless, Bommier et al.⁷² put forward a three-stage model in 2015. They agreed with the “adsorption–intercalation” model and described the increase in the diffusion coefficient at the end of the plateau region as a kind of Na⁺ adsorption on the pore surface based on galvanostatic intermittent titration measurements (Figure 5D,E).

Significantly different from the above mechanisms, Zhang et al.⁷⁴ demonstrated that the insertion of Na⁺ between carbon layers did not happen during the whole sodiation process. They synthesized hard carbon nanofibers with tailored microstructure through pyrolysis in a wide temperature range between 650 and 2800°C and then systematically investigated the correlation between the electrochemical performance and microstructure. They demonstrated that the sloping capacity could be attributed to the absorption of Na⁺ at defect sites (>1 V) and disordered isolated graphene sheets (1–0.1 V), and the plateau capacity was due to the mesopore filling (<0.1 V) (Figure 6A). And they observed that the (002) peak did not shift even the cell was discharged to 0 V (Figure 6B)⁷⁴; thereafter, Li et al.⁷⁵ further supported this model. They reported that the sodiated cotton-derived microtubes showed similar interlayer spacing to the pristine samples by using ex situ transmission electron microscopy (TEM) (Figure 6C). And ex situ X-ray photoelectron spectroscopy (XPS) spectra showed that as the sodiation proceeded in the plateau region, the intensities and values of the binding energy for the Na 1s spectra increased and approached the signal of metallic Na (Figure 6D). They further carried out in situ XRD analysis (Figure 6E) for the anthracite-derived electrode and observed that there was no shift of the (002) peak when the cell was discharged to 0 V versus Na/Na⁺, indicating that Na insertion did not happen during whole discharge process.⁷³

Figure 6. (A) Voltage profiles (top) and schematic representation of Na ion storage mechanisms (bottom) in stages I–III. (B) X-ray diffraction (XRD) pattern recorded in situ on the first discharge of hard carbon electrode. Reproduced with permission: Copyright 2016, Wiley-VCH.74 (C) Ex situ transmission electron microscopy images of pristine hollow carbon microtubes-1300 electrodes and after discharging to 0 V. (D) XPS Na 1s spectrum profiles of hollow carbon microtubes-1300 electrodes and Na metal after etching 60 nm. Reproduced with permission: Copyright 2016, Wiley-VCH.75 (E) In situ XRD patterns collected during the first discharge/charge of the anthracite-derived electrode prepared at 1200°C cycled between 0 and 2 V under a current rate of 0.05 C. Reproduced with permission: Copyright 2016 Elsevier.73 XPS, X-ray photoelectron spectroscopy

Sodium storage mechanism in the low-potential plateau

The discrepancies in sodium storage mechanism are owing to the complex structures of the hard carbon, which bring difficulties to the characterization of hard carbon microstructure and the interpretation of the data obtained by different in situ and ex situ analyses technics, such as XRD, Raman spectroscopy, and NMR. After two decades of research, the structure–electrochemical behaviors of hard carbon materials are still controversial. Of note, recent work reaches a consensus on the sloping region, which is related to adsorption of Na⁺ at the open surface, defect sites, and between isolated larger interlayer spacing (d > 0.4 nm), while hitherto the understanding of origination of low-potential plateau capacity has been disputed about whether it is caused by intercalation between graphitic layers or sodium filling into pores.

Interlayer intercalation mechanism Depicting of interlayer intercalation mechanism

The experimental identification of Na⁺ insertion into graphitic sheets during sodiation is quite difficult due to the intrinsic highly disordered structure of hard carbon. XRD analysis is a vital tool to detect the structural evolution caused by the insertion of Na⁺ into graphitic sheets. The inserted Na⁺ can not only expand the interlayer spacing accompanying the slight shift of (002) diffraction peaks to lower angles but also act as scattering species with the decrease of scattering intensity. However, as mentioned above, according to XRD analyses, many researchers inferred contradictory sodium storage mechanisms in different types of hard carbons.^63,82–86 It may be due to the disordered structure of hard carbon, the reflection is broad and weak, and thus it is difficult to detect its changes in intensity and position. Furthermore, the signal of (002) peak will inevitably be interfered with by the background, causing a shift of the apparent peak position to lower angles. Therefore, the XRD analysis should be properly applied to detect structure evolution during the sodiaiton and desodiation processes, especially for ex situ XRD analysis. In addition, high-resolution transmission electron microscopy (HRTEM) can also be employed to show the changes of interlayer spacing, for example, Wang et al.⁸⁷ observed an obvious volume expansion of sweet-gums-derived hard carbon near the end of sodiation process (Figure 7A–E), indicating the intercalation of Na⁺ in the plateau region. But due to the inhomogeneous structure at the subnanometric, ex situ HRTEM can hardly obtain convincing results.^75,89

Figure 7. Sodiation of hard carbon studied by in situ transmission electron microscopy (TEM). The (A) TEM image and (B) electron diffraction pattern of the pristine hard carbon. The (C) TEM image and (D) electron diffraction pattern of the sodiated hard carbon. (E) Plot of the volumetric change of the hard carbon in sodiation and illustration of Na-absorption and intercalation. Reproduced with permission: Copyright 2019, The Royal Society of Chemistry.87 (F) Theoretical energy cost for Na (red curve) and Li (blue curve) ions insertion into carbon as a function of carbon interlayer distance. The inset illustrates the mechanism of Na and Li-ions insertion into carbon. Reproduced with permission: Copyright 2012, American Chemical Society.37 (G) Schematic illustration of the evolution of the microstructure, sodium storage mechanism, and behavior with the pyrolysis temperature of hard carbons. Reproduced with permission: Copyright 2019, Wiley-VCH88

Theoretical calculations revealed that the intercalation energy depends on the interlayer spacing of hard carbon materials.^90,91 For example, Cao et al.³⁴ proposed that Na⁺ can overcome the energy barrier and intercalate into graphitic sheets with an interlayer spacing larger than 0.37 nm (Figure 7F). An ab initio study revealed that the larger interlayer spacing and monovacancy, divacancy, and Stone-Wales defects would facilitate the insertion of Na⁺ into graphitic layers.³⁶ Sun et al.⁸⁸ proposed the extended “adsorption–intercalation” model (Figure 7G) and supplemented that as the interlayer spacing increased to 0.40 nm, Na⁺ could freely access the carbon layers with a “pseudo-adsorption” behavior, contributing to sloping capacity above 0.1 V. Based on theoretical calculation of the formation energy, they further revealed that NaC₈ was the most likely Na-GIC structure for hard carbon, corresponding to a theoretical capacity of 279 mAh g⁻¹. The extended “adsorption–intercalation” model can interpret well the “volcano”-shaped tendency of plateau capacity with the increase of pyrolysis temperature. When the carbon precursor is pyrolyzed at low temperature, the carbon matrix generates a highly disordered structure with an interlayer distance larger than 0.40 nm, exhibiting a “pseudo-adsorption” behavior and a sloping volta. As the pyrolysis temperature increases, the graphitization of hard carbon is enhanced with a proper interlayer spacing between 0.36 and 0.40 nm, exhibiting an intercalation behavior and a plateau capacity. Too much high heat treatment temperature will further reduce the interlayer spacing to less than 0.36 nm, causing the hardly insertion of Na⁺ into carbon layers and a reduced plateau capacity.

Boosting the plateau capacity from intercalation mechanism

According to the intercalation mechanism, the plateau capacity can be extended by expanding interlayer spacing and improving the growth of graphitic-like nanodomains. Heteroatom-doping is an effective method to tailor the microstructure of hard carbon. Due to the large atomic radius, S and P can serve as a pillar to expand the interlayer spacing.^92–96 By introducing phosphoric acid or sulfuric acid into sucrose precursor and pyrolysis at 1100°C, Li et al.⁹⁷ extended the interlayer spacing of sucrose-derived hard carbon from 0.377 nm to 0.395 and 0.383 nm (Figure 8A,B). Accordingly, the reversible plateau capacity was improved from 149 mAh g⁻¹ to 202 and 177 mAh g⁻¹, respectively. Alvin et al.⁹⁸ reported the P-doped hard carbon obtained by adding phosphoric acid into lignin and pyrolysis at 1300°C. The P-doping amount was increased from 0 to 1.10 at% by adjusting the ratio of phosphoric acid to lignin, and the interlayer spacings of as-obtained hard carbon were extended from 0.375 to 0.387 nm, along with the plateau capacities improved from 183 to 223 mAh g⁻¹ (Figure 8C–E). Yuan et al.¹⁰⁰ enlarged the interlayer spacing of flexible freestanding multichannel carbon nanofibers through a vacuum heat treatment method, resulting in an increase in reversible capacity from 160 to 233 mAh g⁻¹. Yuan et al.⁹⁹ directly pyrolyzed natural potassium-doped coconut endocarp at 1100°C (Figure 8F). The as-obtained hard carbon featured a dilated interlayer spacing (0.4 nm) and delivered an initial reversible capacity of 314 mAh g⁻¹ at 50 mA g⁻¹. Once it was pickled to remove the potassium ions, the interlayer spacing and reversible capacity decreased simultaneously. There are few reports on the expansion of the size of graphitic-like nanodomains to increase the plateau capacity. Both high-temperature heat treatment and metal ion-catalysis can enhance the growth of graphite nanocrystals, but excessively high temperature or ion content will also reduce interlayer spacing to less than 0.36 nm, resulting in insufficient Na⁺ insertion between carbon layers. Therefore, only by ensuring sufficient interlayer spacing (>0.37 nm) while extending the graphite-like nanodomains, the sodium storage capacity of hard carbon can be effectively increased.

Figure 8. (A) X-ray diffraction (XRD) patterns and (B) initial galvanostatic sodiation/desodiation potential profiles of doped and undoped sucrose-derived hard carbons. Reproduced with permission: Copyright 2017, Wiley-VCH.97 (C) XRD patterns, (D) interlayer spacing, and (E) discharge–charge profiles of the undoped and P-doped hard carbons. Reproduced with permission: Copyright 2020, Elsevier.98 (F) Schematic diagram of the synthetic route for natural K-doped hard carbon anode. Reproduced with permission: Copyright 2018, American Chemical Society99

Micropores filling mechanism Depicting of pores filling mechanism

Many works have confirmed the pores filling mechanism by in situ or ex situ SAXS analysis. The scattering intensity of shoulder peaks corresponding to the nanopore decreases at the plateau region, which indicates a decrease in the electron density contrast between the carbon matrix and nanopores.^{59,63,64,101,102} Through in operando ²³Na solid-state NMR analysis, Stratford et al.¹⁰³ observed a clear shift of the signal to positive frequencies during the sodiation steps in the plateau region (Figure 9A). They attributed it to the extended sodium clusters with more metallic in nature, resulting in increased contribution from Knight shift. Meanwhile, the calculation results of pair distribution function (PDF) data indicated the formed Na clusters with coherence lengths of >10 Å.¹⁰³ It is worth noting that the pores in hard carbon can be further divided into external pores (i.e., open pores) and internal pores (i.e., closed pores). The gas adsorption–desorption technique is only sensitive to the former, while true density analysis can detect the closed pore volume and SAXS is accessible to all pores. Thus, it is essential to utilize integrated characterization methods to ascribe the pore texture of hard carbon. It has been reported that several hard carbons with low-open porosity deliver a considerable plateau capacity.^69,105,106 In addition, Qiu et al.⁶⁸ indicated that the theoretical capacity of cellulose-derived hard carbon calculated based on the saturated Na filling into micropores (established by N₂ adsorption technique) was lower than the experimental plateau capacity. Therefore, it is agreed that open porosity is not the main contribution to the plateau capacity. What's worse, larger surface porosity accessible to the electrolyte will cause the decomposition of electrolyte, leading to the low ICE and damage of the cycle performance of the batteries. In 2018, Zhang et al.⁴⁵ proposed that Na deposition in the closed pores of lotus stems derived hard carbon, given that both the reversible capacity and the closed pore ratio reached the maximum value at 1400°C. Thereafter, Li et al.¹⁰² synthesized waste cork-derived hard carbon between 800°C and 1400°C and investigated their pore information by true density, SAXS, and nitrogen adsorption/desorption tests. It demonstrated that the plateau capacity suffered a weak correlation with open-pore volume, but exhibited an obvious positive relationship with the closed-pore diameter or volume. Au et al.⁶⁰ employed ex situ ²³Na solid-state NMR and total scattering studies on a series of hard carbons prepared at different temperatures. As shown in Figure 9B–D, they observed the quasi-metallic peak of the complete sodiated carbon materials grew in intensity and shifted to the higher frequency, and the calculated pore size based on the small-angle neutron scattering data also increased as the pyrolyzed temperature increased. Thus, they concluded that the increased metallic character of sodium clusters was due to the increased pore size with temperature, which leads to the growth of sodium clusters within the pores. Luo et al.¹⁰⁴ observed different ²³Na solid-state NMR shifts in two commercial relevant carbons with different average pore diameters obtained on SAXS data (Figure 9E). However, the differential PDF analysis of total scattering data indicated similar sodium cluster sizes (~13–15 Å) in these two kinds of carbon (Figure 9F). They attributed the different NMR shifts to the different electronic structures of the materials, which generated a difference Na partial density of states at the Fermi level. And they concluded that a larger average pore diameter does not form a larger sodium cluster and the plateau capacity depends on the number of pores suitable for sodium cluster formation, not the number of pores that allow the formation of larger sodium clusters. Of note, interlayer spacing also plays an important role in the micropores filling mechanism; it acts as a diffusion channel that allows Na⁺ diffusion into the pores. The limited interlayer spacing will restrict the passage of Na⁺ into the pores, resulting in a reduced plateau capacity even with sufficient Na storage sites in micropores. Therefore, even for the pores filling mechanism, the structures of graphite-like crystallites also significantly affect the low-potential plateau capacity, which further contributes to the difficulty to clarify the sodium storage mechanism in hard carbon materials.

Figure 9. (A) Operando 23Na NMR (nuclear magnetic resonance) spectra for an electrochemical cell with sodium metal and hard carbon electrodes, and NaPF6 electrolyte. Reproduced with permission: Copyright 2016, Royal Society of Chemistry.103 (B) Ex situ 23Na magic-angle spinning (MAS) NMR of glucose-derived hard carbon prepared in a range of 1000–1900°C after discharging to 5 mV. (C) The relationship between the pore diameter, quasi-metallic Na peak shift, and sample pyrolysis temperature, and (D) schematic illustration of pore size dependence on basal plane lateral size. Reproduced with permission: Copyright 2020, The Royal Society of Chemistry. (E) Operando 23Na NMR spectra for electrochemical cells with sodium metal, and an NaPF6 electrolyte and carbon A or carbon B electrodes. (F) The top is differential pair distribution functions (PDFs) for carbon A (red) and carbon B (blue) for the end of slope process, red/blue lines offset below represent the deconvolution into sharp (based on an expansion of the carbon matrix) and broad components. The green line represents the calculated PDF for sodium intercalated between two expanded curved graphene fragments; the bottom is PDFs for the end of the plateau process. The green line shows the calculated PDF for sodium metal (with Uiso = 0.35; spherical particle diameter = 13 Å). Reproduced with permission: Copyright 2021, American Chemical Society104

Boosting the plateau capacity by using pore-forming agents

It has been reported that engineering close pores in hard carbon materials is an effective way to extend the plateau capacity. Bin et al.⁷⁶ reported precise structural engineering of hard carbon into multishelled hollow carbon nanospheres (Figure 10A). As the shell number increased, the sloping capacity was almost unchanged while the plateau capacity continuously increased and the four-shelled hollow carbon nanospheres delivered a reversible capacity of 360 mAh g⁻¹ at 30 mA g⁻¹. Adopting pore-forming agents in carbon precursors is a feasible method to produce abundant closed pores. Kamiyama et al.¹⁰⁷ utilized an organic polymer as a pore-forming additive to synthesize macroporous phenolic resin block from liquid solution-type resin (Figure 10B). After further carbonization at 1500°C, the hard carbon anode delivered a larger reversible capacity of 386 mAh g⁻¹ and a plateau capacity of ~300 mAh g⁻¹ at a current density of 10 mA g⁻¹. Meng et al.¹⁰⁶ introduced ethanol as the pore-forming agent into the phenol-formaldehyde precursor, and through the solvothermal process and further carbonization, hard carbon with sufficient closed pores was obtained (Figure 10C). It is found that ethanol acts as a steam generator leading to steam gas to introduce pore cavities among the cross-linked carbon matrixes. The obtained hard carbon materials exhibited a low specific surface area of 1.44 m² g⁻¹ but a low true density of about 1.4 g cm⁻¹ and delivered a high sodium storage capacity of 410 mAh g⁻¹ with a plateau capacity of 284 mAh g⁻¹. Kamiyama et al.¹⁰⁸ also successfully prepared hard carbon with an extremely high sodium storage capacity (478 mAh g⁻¹) by a MgO-template method (Figure 10D). They freeze-dried the mixture of magnesium gluconate and glucose and then the obtained mixture was pyrolyzed in two steps. Nano-size MgO particles are generated in the pre-pyrolysis process and removed by acid pickling after the following pyrolysis process, leaving abundant pores in the carbon matrix. By adjusting the ratio of magnesium gluconate and glucose and the carbonization temperatures, the optimal hard carbon anode delivered an ultrahigh plateau capacity of 400 mAh g⁻¹. Deng et al.¹¹⁰ introduced closed micropores and carbonyl groups into pine-derived hard carbon by the sulfuric acid pretreatment and two-step carbonization process. The hard carbon anode exhibited a reversible capacity of 354.6 mAh g⁻¹ at 30 mA g⁻¹ and a high ICE of 88.7%.¹¹⁰ Converting open pores to closed pores can also effectively increase the plateau capacity. Li et al.¹⁰⁹ coated the preheat-treated commercial activated carbon with a pitch by the “solvent evaporation” method and then recarbonized at 1400°C (Figure 10E). The as-prepared sample exhibited a low surface area of 24 m² g⁻¹ and high internal porosity, delivering a high reversible capacity of 391 mAh g⁻¹ at a current density of 25 mA g⁻¹ with a high ICE of 80%. It has been reported that preoxidation of carbon precursor could introduce oxygen functional groups to facilitate the crosslinking and interrupt the graphitizing of carbon mixture, resulting in more disordered microstructures, and hence closed pores could be generated in the carbon framework.¹¹¹ Even the soft carbon without plateau capacity can show considerable plateau capacity after preoxidation treatment.^112–114 Lu et al.¹¹⁴ preoxidated pitch in the air at 300°C for 3 h and then thermal-treated it at 1400°C. It is found that the preoxidation process induced a more turbostratic structure, generating close voids constructed by random piled curved carbon layers, and the reversible sodium storage capacity and ICE of the preoxidation-tuned carbon increased to 300.6 mAh g⁻¹ and 88.6%, respectively.

Figure 10. (A) Illustration of the synthesis route for multishelled hollow hard carbon nanospheres. Reproduced with permission: Copyright 2018, Wiley-VCH.76 (B) Scanning electron microscopy images and initial discharge/charge curves of microporous phenolic resin derived hard carbon samples prepared at different temperatures. Reproduced with permission: Copyright 2020, American Chemical Society.107 (C) Illustration of the typical synthesis process of hard carbon using liquid phenol-formaldehyde resin as the precursor and ethanol as the pore-forming agent. Reproduced with permission: Copyright 2019, American Chemical Society.106 (D) Schematic illustration for the two mixing procedures for preparation of the mixtures of Mg Glu and Glc. Reproduced with permission: Copyright 2021, Wiley-VCH.108 (E) Schematic illustration for the tailored strategy to synthesize the pitch-coated hard carbon. Reproduced with permission: Copyright 2018, Elsevier109

Deficiencies of the interlayer intercalation and pores-filling mechanisms

Despite both interlayer intercalation and pore-filling mechanism can interpret the experimental phenomenon to a certain extent, and the plateau capacity can be improved by reconstruction of the graphitic-like domains or pore structures, both the two mechanisms cannot fully explain all experimental observations. A vital matter is that regardless of intercalation or pore-filling mechanism, the experimental plateau capacity in some literature exceeds the corresponding theoretical specific capacity. The theoretical specific capacity for intercalation mechanism is 279 mAh g⁻¹ corresponding to form NaC₈. And the maximum possible specific capacity for the nanopore mechanism can be calculated using the theoretical specific capacity of metallic sodium (1128 mAh cm⁻³) and total pores volume, including internal and external pores. As discussed above, hard carbon anode materials with ultrahigh plateau capacity (>279 mAh g⁻¹) can be obtained by using pore-forming agents. On the contrary, for example, Meng et al.¹⁰⁶ used ethanol as the pore-forming agent and formaldehyde resin as carbon sources via a solvothermal method to prepare hard carbon with rich closed pores, which exhibited a high Na storage capacity of 410 mAh g⁻¹. Based on the true density (~1.43 cm³ g⁻¹) measured by He gas and d₀₀₂ (0.391 nm) of the sample, we roughly calculated that the maximum possible capacity derived from the saturated pore filling is about 209 mAh g⁻¹, which is much lower than the experimental plateau capacity of 284 mAh g⁻¹. Kutoba et al.⁵³ have prepared sucrose-derived hard carbon at different heat treatment temperatures and systematically characterized their structures. The experimental plateau capacity at 900°C, 1100°C, 1300°C, 1500°C, and 1600°C are 139, 152, 231, 240, and 239 mAh g⁻¹, respectively, larger than the theoretical specific capacity of 88, 39, 35, 218, and 225 mAh g⁻¹ calculated based on the helium pycnometer density.⁵³ Hence, a more comprehensive and in-depth sodium storage insight is urgent to be put forward. Some recent literature reported that the sodium storage mechanism is different under different circumstances.^63,115 Morikawa et al.⁶³ conducted systematically ex situ SAXS and wide-angle X-ray scattering study of a series of hard carbons prepared at temperatures from 1000°C to 2000°C. They proposed that for the hard carbon prepared at low temperature (1000°C), sodium adsorption at defects and intercalation into graphitic domains are larger due to a larger number of defects and the larger graphene–graphene interlayer distance for sodium storage, while the capacity of hard carbon prepared at 1900°C ascribed to the nanopores filling. However, it generates another fundamental issue that whether these two mechanisms act as an individual or a few overlapped processes, or they occurred as concurrent processes. But what is assured is that pore-filling needs to first undergo diffusion of Na⁺ between graphitic sheets, due to closed pores being inaccessible to the electrolyte.

SUMMARY AND OUTLOOK

There has been a tremendous effort to develop advanced hard carbon anodes and understand their sodium storage mechanism during the past decade. In this review, we provide a comprehensive overview of the recent progress in sodium storage mechanisms and strategies in optimizing the electrochemical properties of hard carbons. Different models and reaction paradigms have been proposed as confirmed by different techniques (such as XRD, Raman spectroscopy, and NMR) and computational methods. The discrepancies in the sodium storage mechanism are due to complex structures of the hard carbon, which bring difficulties in the characterization of microstructure. It seems that the sodium storage mechanism encounters the bottleneck and controversies lie in the assignment of the plateau capacity (Figure 11).

Figure 11. Schematic illustration of the sodium storage mechanism in hard carbons and future research directions

Up to now, debates remain in the orientation of sodium storage mechanism in the plateau region, including interlayer intercalation and pore-filling mechanism. Both mechanisms can explain the corresponding experimental results. Although these disputes remain, it is assuring that the plateau capacity of hard carbon anode can be extended by expanding the interlayer spacing between graphitic sheets or generating abundant closed pores in the carbon framework. Specifically, hard carbon with ultrahigh plateau capacity can be achieved by the pore-forming strategy. However, neither separated interlayer intercalation nor pore-filling mechanism cannot interpret that the experimental plateau capacities of some carbons exceed the corresponding theoretical capacity. Hence, further insight on this issue is urgently desired.

Due to the complex nature of the microstructure of hard carbon, further research is highly needed to precisely describe the microstructure. There is a massive misinterpretation of experimental data obtained by advanced structural characterization techniques due to a lack of in-depth understanding of hard carbon structure. For instance, because of the existence of defects, the local structure of graphitic sheets in hard carbon suffers significant curvature, and part of curved graphene sheets locally stack forming randomly oriented graphitic-like nanodomains, while the other part of highly curved and randomly oriented graphene sheets form pores in the bulk of materials. However, when considering sodium storage mechanism, such curvature has rarely been introduced in the hard carbon structure model. Hence, the structural models should be refined to introduce the curvature of graphitic-like domains and closed pores, and then evaluate their impact on sodium storage behaviors. Besides, due to the curved nature of graphene sheets, the concepts of interlayer and pores need to be clearly defined.

Moreover, more powerful techniques are desirable for the structural characterization of pristine and sodiated hard carbon, and operando techniques are essential due to the unstable characteristics of carbons with intercalated and filled sodium ions. For the pore-filling mechanism, it should be clarified whether there is a suitable pore size range conducive to the formation of quasi-metallic sodium. And more efforts are needed to clarify the characteristics of Na-GICs and quasi-metallic sodium. In addition, new strategies are urgently demanded to precisely control the graphitic-like domains and pore structures in the carbon framework. Despite controversy remains in the sodium storage mechanism, the optimum interlayer spacing and pore structure will lead to improved plateau capacity. Besides, the ICE is also an important property of the carbon anodes. Although engineering closed pores can effectively extend the plateau capacity, it is worth noting that open pores inescapably increase the specific surface area of the hard carbon materials, which leads to severe decomposition of electrolyte on the surface of anode materials, resulting in a large irreversible capacity and a relatively low ICE. Thus, the pore-forming strategy might be carried out precisely to engineer closed pores in bulk carbon and prevent open pores on the surface of hard carbon materials. On the contrary, decreasing defect concentration and regulating the degree of graphitization of hard carbon materials are also effective methods to alleviate the irreversible sodium ions trapped at defect sites and achieve a hard carbon with high ICE.

To facilitate the practical application of hard carbon, future research of hard carbon anodes should be focused on the development of advanced hard carbon materials with high plateau capacity and high ICE. The external surface area and defect concentration should be decreased to prevent the irreversible side reaction and improve the ICE, and closed pores should be engineered to improve the internal surface and pore volume. Furthermore, a larger interlayer spacing is especially crucial for interlayer intercalation and diffusion of Na⁺ in the bulk materials. There are some reports on hard carbon materials for pouch cells. Zhao et al.¹¹⁶ assembled a batch of pouch cells composed of Na₄Fe_2.91(PO₄)₂P₂O₇ materials as the cathode and hard carbon as the anode materials. The as-assembled pouch cells delivered a discharging capacity of 0.35 Ah with an average discharge voltage of 2.9 V and a capacity retention ratio of 87.4% after 1000 cycles. Tang et al.¹¹⁷ evaluated the electrochemical performance of starch packing peanuts derived hard carbon in a full-pouch cell, which exhibited a 90% capacity retention after 200 cycles and 84% retention after 300 cycles. These reports of hard carbon in pouch cells further illustrate the credibility of commercialization, boosting confidence in practical applications of hard carbon for SIBs.

Despite great challenges ahead, a lot of achievements on hard carbon materials have been made by many researchers. With the increasing interest from both industrial and academic communities, it is optimistic that more convincing sodium storage mechanisms will be proposed and hard carbon anode materials with high electrochemical properties will be constructed and commercialized for grid-scale applications soon.

ACKNOWLEDGMENTS

We thank the financial support from the National Nature Science Foundation of China (Nos. U20A20249 and 21972108) and the Key Research Program of Hubei Province (2020BAA030).

CONFLICTS OF INTEREST

The authors declare no conflicts of interest.