Robust cement-graphite thick electrode enables high-performance molten salt aluminum batteries

Abstract

Molten-salt aluminum batteries are promising for grid-scale energy storage, leveraging the abundance and recyclability of aluminum along with the intrinsic safety, non-flammability, and high ionic conductivity of chloroaluminate electrolytes. However, conventional polymeric binders lack stability in high-temperature, corrosive molten-salt environments, leading to electrode structural degradation and premature battery failure. Here we propose thick cement-graphite electrodes for molten salt aluminum batteries that feature a self-supporting structure, high mechanical strength, high thermal tolerance, and resistance to chemical corrosion. In this design, graphite provides continuous electron-conducting pathways, while hydrated cement phases encapsulate and interconnect graphite, forming an inorganic skeleton with measurable mechanical properties, chemically inert, and hierarchically porous. Operating at 150 °C, the Al| |cement-graphite cells achieve a long cycle life of over 11,000 cycles at 16 mg cm^–2 and 10 A g^–1, rate capability (retaining nearly 80 mAh g^–1 at 15 A g^–1) and stable cycling performance under high mass loadings up to 100 mg cm^–2. After extended cycling, the cement-graphite electrode retains its structural integrity. Multi-cell assemblies demonstrate scalability, while cost analysis shows cost savings associated with cement binders for molten-salt batteries.

Molten salt aluminum batteries face electrode degradation from traditional polymer binders in high-temperature, corrosive environments. Here, authors develop robust self-supporting cement-graphite thick electrodes as a low-cost inorganic alternative. The electrodes deliver over 11,000 stable cycles, good rate performance and reliable operation at high material loadings, with scalable production and lower overall battery costs.

Introduction

Large-scale electrochemical energy storage systems are playing an increasingly important role in enabling the reliable integration of renewable electricity and mitigating the challenges posed by the intermittency of power supply¹. While lithium-ion batteries have emerged as the dominant technology, owing to their high energy density and extended cycle life, their widespread adoption is hampered by high costs, limited resource availability, and safety issues². These limitations have accelerated the exploration of alternative battery chemistries^3,4. Al metal, being the most abundant metallic element in the Earth’s crust, presents an option for rechargeable batteries due to its low cost, high capacity (2980 mAh g^–1 and 8046 mAh cm^–3), and well-established recycling networks^5,6.

To advance the development of Al batteries, researchers have recently focused on developing advanced electrolyte systems^7,8. Compared to conventional ionic liquid electrolytes (1-ethyl-3-methylimidazolium chloride with AlCl₃, EMIC-AlCl₃), the AlCl₃-NaCl-KCl ternary molten salt electrolyte contains a higher proportion of composite Al_nCl_3n+1^– clusters^9,10. This composition reduces the energy barrier for Al³⁺ desolvation, enhancing electrode reaction kinetics and ionic conductivity (approximately 190 mS cm^–1)^11,12. Consequently, it enables rapid charge-discharge cycles, uniform aluminum deposition, and stable cycling performance^13,14. Nevertheless, the commercialization of molten salt Al batteries remains hindered by the absence of durable electrode frameworks capable of withstanding demanding operating conditions, such as high temperatures and highly corrosive electrolytes.

Electrodes typically employ polymer binders such as Polyvinylidene Fluoride (PVDF), Polytetrafluoroethylene (PTFE), or Carboxymethyl Cellulose (CMC) to maintain the conductive network and structural integrity, enhancing electron transport and cycling stability¹⁵. However, PVDF and CMC exhibit limited chemical and thermal stability in high-temperature and corrosive molten salt electrolytes^16,17. The C-H bonds in PVDF (~414 kJ/mol) are susceptible to attack by Lewis acids, while CMC’s glycosidic bonds and functional groups are prone to cleavage or degradation in strongly acidic conditions, resulting in diminished performance^18,19. At elevated temperatures, PVDF and CMC tend to soften, swell, or decompose, disrupting electron conduction pathways and triggering parasitic side reactions at the electrode-electrolyte interface. PTFE has a fully fluorinated structure with C-F bonds (~485 kJ/mol) and shows resistance to degradation in high-temperature and corrosive molten salt electrolytes^20,21. Although PTFE demonstrates stability as a binder in molten salt electrolytes, its processing involves considerable difficulties. Through mechanical kneading, mixing, and shearing, PTFE forms a micron-scale fibrillar network that encapsulates and binds active materials and conductive additives, producing robust composite electrodes²². However, the extent of PTFE fibrillation is dependent on operator skill, which complicates control through standardized parameters and thus impedes large-scale continuous manufacturing. Moreover, PTFE’s intrinsic elasticity leads to electrode rebound after rolling, reducing thickness control accuracy. Additionally, its high cost-approximately 25 USD/Kg-undermines economic feasibility. Consequently, there is a need to develop stable, cost-effective binders compatible with standardized manufacturing processes to overcome the technical and economic limitations of PTFE in electrode production.

Here we propose a departure from polymer-based architecture by reimagining cement—a material historically used for structural infrastructure—as a functional, inorganic binder for electrochemical systems^23–25. Building on this concept, we studied a binder-free, self-supporting electrode using a cement-graphite (C-G) framework. In this design, graphite establishes a continuous electron-conducting pathway, while hydrated cement encapsulates the conductive network to form a mechanically robust and chemically stable framework (Fig. 1a). This structure exhibits stability under high-temperature and corrosive molten salt environments (Fig. 1b). Its hierarchical porous architecture facilitates efficient electrolyte infiltration, enabling rapid ion transport and performance under high active material loadings (Fig. 1c). As a result, the C-G framework enables molten salt Al batteries to operate reliably between 120 and 180 °C, demonstrating a cycle life of 11,000 cycles at a mass loading of 16 mg cm^–2 and a specific current of 10 A g^–1, while maintaining stable capacity at mass loadings up to 100 mg cm^–2. The abundance and low cost of cement, combined with straightforward water-based processing, enable scalable and sustainable electrode manufacturing. These findings show that cement, traditionally a structural material, can be used as a functional electrochemical binder. By establishing a durable, low-cost, and scalable inorganic self-supporting framework, this work paves the way for developing robust molten salt aluminum batteries for large-scale energy storage.

Fig. 1. Mechanism and robustness of the C-G frameworks in molten-salt Al batteries.

a Schematic of the C-G electrode and the reversible AlCl₄^– intercalation/de-intercalation in graphite. b Flame-resistance comparison of PVDF, PTFE and cement-based electrodes, highlighting the high-temperature stability of the cement framework (>500 °C). c Photograph of a fre-standing C-G electrode showing mechanical integrity. d Radar chart benchmarking thermal stability, mechanical modulus, cost and sustainability, underscoring the advantage of cement-graphite over polymer-bound electrodes. The corresponding data for this radar chart are provided in Supplementary Table 1. Source data are provided as a Source Data file.

Results

Fabrication and structural characterizations of C-G electrodes

Cement shows performance across several critical metrics that is competitive with that of some polymer binders such as PVDF and PTFE (Fig. 1d and Supplementary Table 1). With a mechanical modulus ranging from 20 to 40 GPa, cement provides structural rigidity to electrodes^26–28. Furthermore, cement exhibits high thermal stability, enabling reliable operation in high-temperature molten salt battery systems^29–31. Its low raw material cost, coupled with its abundance, structural stability, and compatibility with high-temperature battery environments, positions cement as a cost-effective binder material for scalable battery applications^32,33. We utilized ultrafine cement with effective bonding properties to fabricate the C-G electrodes. The primary features of ultrafine cement include its small particle size and high specific surface area, which enhance its adhesion to substrate surfaces. Furthermore, the rapid hydration reaction of ultrafine cement results in increased bonding strength³⁴. X-ray diffraction (XRD) analysis of raw cement identified characteristic crystalline phases including tricalcium silicate (C₃S), dicalcium silicate (C₂S), tricalcium aluminate (C₃A), tetracalcium aluminoferrite (C₄AF), calcite and quartz (Supplementary Fig. 1a). Scanning electron microscopy (SEM) images showed cement particles with well-defined boundaries and smooth surfaces (Supplementary Fig. 1b, c), features that enable efficient hydration reactions³⁵.

The fabrication process of the C-G electrodes, illustrated in Fig. 2a, is fundamentally distinct from conventional electrode manufacturing. Instead of using polymer binders and organic solvents such as NMP, our method relies solely on water as the processing medium, making the approach environmentally benign and scalable³⁶. Graphite and cement powders are intimately blended with water to form a highly viscous and cohesive paste. Owing to the hydration-driven setting behavior of cement, this paste can be readily cast into molds of various geometries and cured under ambient conditions without the need for additional thermal or chemical treatments. As demonstrated in Fig. 2b, c, the resulting electrodes can be shaped into self-supporting cubic blocks or large, flat plates, confirming both the process flexibility and intrinsic mechanical integrity. This versatility can be useful for practical battery assembly, where diverse electrode sizes and geometries may be required. This method allows precise and straightforward fabrication of electrodes in multiple forms, making it suitable for large-scale production.

Fig. 2. Fabrication and structural characterizations of the C-G electrodes.

a–c Fabrication process of free-standing electrodes by mixing, molding and stacking. d Illustration of hydration-driven C-S-H gels formation interconnecting graphite. e SEM image, scale bar: 1 μm. f Pore-size distribution of the hierarchical pore network derived from Micro-CT analysis. scale bar: 50 μm. g TEM image, scale bar: 200 nm. h Typical results of compressive strength, flexural strength, and electrical conductivity. i XRD patterns of cement before and after molten salt electrolyte immersion. Source data are provided as a Source Data file.

The hydration mechanism of cement within the C-G electrodes is illustrated in Fig. 2d. Upon water addition, the cement rapidly dissolves, elevating the pore solution pH above 12 and releasing substantial amounts of Ca²⁺ and silicate ions. This highly alkaline environment induces supersaturation; however, rather than undergoing homogeneous precipitation, graphite flakes-with their flat surfaces and edge defects-act as preferential sites for heterogeneous nucleation. On these surfaces, nanoscale clusters of calcium silicate hydrate (C-S-H) initially form, reducing the energy barrier for hydrate precipitation. As hydration proceeds, the C-S-H nuclei develop into fibrous and foil-like networks that span the gaps between neighboring graphite sheets. These bridging structures effectively interlock the conductive flakes while forming mineral necks that densify the matrix without fully occluding porosity. Concurrently, portlandite [Ca(OH)₂] crystals precipitate within the pore space, reinforcing the inorganic framework, while aluminates react with sulfate to form ettringite needles that fill micro-voids at early stages. Together, these hydration products immobilize the graphite network into a rigid inorganic skeleton capable of withstanding thermal and mechanical stresses, while maintaining accessible pathways for molten salt infiltration^37–39. This mechanism accounts for the incorporation and subsequent release of hydration water below 100 °C, even at reduced cement contents, as confirmed by thermogravimetric analysis. The proportional mass losses (1.0%, 0.81% and 0.41% for 100, 50 and 25 wt% cement, respectively) validate that graphite facilitates early hydrate nucleation, enabling a stable scaffold to form even when cement is not the dominant phase (Supplementary Fig. 2a–c).

XRD pattern of the pristine graphite powder exhibits a sharp diffraction peak at approximately 26° corresponding to the (002) crystal plane of graphite (Supplementary Fig. 3a). Raman spectroscopy reveals distinct characteristic peaks at approximately 1350 cm^–1 and 1580 cm^–1, attributed to the D and G bands of graphite, respectively (Supplementary Fig. 3b). SEM images reveal the characteristic layered morphology of graphite (Supplementary Fig. 3c). In the composite electrode, XRD analysis reveals minor diffraction peaks for silicate and calcite at 2θ angles of approximately 18°, 29°, and 34°, indicating the presence of typical hydration and carbonation products within the cement matrix⁴⁰. A broad background signal points to the amorphous nature of the C-S-H gels. The coexistence of these cement hydration phases with the graphite (002) diffraction peak, alongside the consistent I_D/I_G ratio in Raman spectroscopy (Supplementary Fig. 4a, b), confirms that the cement hydration process does not compromise the structural integrity of graphite⁴¹. Furthermore, scanning transmission electron microscopy coupled with electron energy loss spectroscopy (STEM-EELS) analysis reveals that the C K-edge spectrum of the graphite within the composite electrode exhibits a sharp π* resonance peak at approximately 285 eV and the onset of the σ* edge near 292 eV (Supplementary Fig. 5a–c). The position and line shape of these features are nearly identical to those of pristine graphite, indicating that the carbon atoms retain their intrinsic sp² hybridization state and that the chemical configuration of graphite remains unaltered⁴².

SEM images reveal that the C-G composite electrode features a three-dimensional network formed by interwoven cement hydration products. Graphite particles are homogeneously distributed and tightly embedded within this matrix, forming a mechanically interlocked structure that ensures good interfacial contact (Fig. 2e). The electrode contains numerous well-connected micropores that facilitate efficient electrolyte infiltration (Supplementary Fig. 6a–c). The hierarchical porosity of the framework was further demonstrated by Micro-Computed Tomography (Micro-CT) 3D reconstructions and pore-size analysis. The C-G electrode forms a continuous solid structure (gray), with uniformly dispersed irregular pores (blue) throughout (Supplementary Fig. 7). These pores constitute a continuous network from the surface to the interior, with a total porosity of 20.5%, providing ample space for electrolyte access. The pore size distribution is dominated by small to medium pores (~10 μm), accompanied by a limited fraction of larger pores. This hierarchical pore structure ensures electrolyte penetration while balancing infiltration and structural stability (Fig. 2f). Elemental mapping shows the homogeneous distribution of carbon (from graphite) alongside Ca, Si, Al, Fe, Mg, and O (from cement) throughout the electrode (Supplementary Fig. 8). Further Transmission electron microscopy (TEM) analysis reveals that the three-dimensional network framework consists largely of rod-like C-S-H gel structures-typical cement hydration products-with lengths of 50–300 nm and diameters of 10–30 nm. These nanoscale rods are arranged in a dense, disordered manner, forming a continuous framework. Their elongated morphology increases the contact area with graphite surfaces, enhancing interfacial adhesion (Fig. 2g). Nanoscale elemental mapping within C-S-H regions shows uniform dispersion of Ca, O and Si, underscoring the critical role of C-S-H in binding graphite lamellae and reinforcing the composite structure (Supplementary Fig. 9). To quantitatively evaluate the mechanical uniformity of this composite at the microscopic scale, systematic nanoindentation tests were performed on a representative 90 μm × 90 μm region using a grid-based scheme with 10 μm spacing, yielding 100 valid data points (Supplementary Fig. 10a). The results reveal that 75% of the measured points exhibit indentation modulus values in the range of 3.69–7.60 GPa and hardness values in the range of 0.07–0.20 GPa, with small variations (Supplementary Fig. 10d–f). As shown in the hardness and modulus distribution maps (Supplementary Fig. 10b, c), the mechanical properties exhibit a continuous transition across microscopic regions, with no evidence of localized low-performance weak zones at the graphite-cement interfaces. This microscale mechanical homogeneity indicates that the graphite particles and cement matrix are integrated primarily through mechanical interlocking, forming a cohesive and structurally uniform electrode material. This mechanical interlocking mechanism provides sufficient structural integrity for the electrode under the demanding operating conditions of molten salt batteries.

Electronic conduction in the C-G electrodes is governed by a graphite-dominated percolation network, while the cement matrix provides mechanical integrity. To resolve the trade-off between these competing functions, we systematically evaluated the electrical conductivity and mechanical properties of composites with cement mass fractions ranging from 20 to 60 wt%. Increasing the cement content progressively enhanced mechanical strength: compressive strength increased from 1.47 MPa at 20 wt% to 3.28, 6.96, 11.61 and 15.1 MPa at 30, 40, 50 and 60 wt%, respectively; flexural strength followed a similar trend, rising from 0.695 MPa at 20 wt% to 2.293, 3.81, 5.218 and 5.894 MPa at 30–60 wt% (Supplementary Fig. 11). However, this mechanical reinforcement came at the expense of diluting the conductive graphite network, with electrical conductivity decreasing from 23.8 S/cm at 20 wt% to 12.8, 5.21, 1.26 and 0.99 S/cm at 30–60 wt% (Fig. 2h). Notably, a sharp decline in conductivity was observed above 40 wt% cement, as the increasing proportion of insulating hydration products disrupted the continuity of the graphite network. Below 40 wt%, despite higher conductivity, the electrodes lacked sufficient mechanical robustness and were prone to brittleness or disintegration during handling and electrochemical cycling. Therefore, a cement content of 40 wt% represents the optimal formulation, delivering adequate mechanical strength (compressive: 6.96 MPa; flexural: 3.81 MPa) for self-supporting electrode fabrication while maintaining an effective graphite percolation network (conductivity: 5.21 S/cm) to ensure efficient charge transport under the demanding conditions of molten salt batteries.

The chemical and thermal stability of the binder play a critical role in determining the electrochemical performance of the electrode. Previous studies have shown that conventional organic binders, such as PVDF, styrene-butadiene rubber (SBR), CMC, and polyethylene oxide (PEO), undergo chemical degradation in AlCl₃-NaCl-KCl ternary molten salt electrolytes¹⁵. To evaluate the chemical stability of cement as a binder, samples of cement, PVDF, and PTFE were exposed to the ternary molten salt electrolyte and heated from 25 ± 5 °C to 250 °C at a rate of 5 °C/min. As the electrolyte melted, PVDF exhibited pronounced color changes, whereas cement and PTFE maintained structural integrity (Supplementary Fig. 12). To further confirm the stability of cement in the molten salt environment, XRD and Fourier‑Transform Infrared Spectroscopy (FTIR) spectroscopy were performed on pristine cement and cement samples after immersion in the electrolyte (Supplementary Figs. 13 and 14). XRD patterns show that the characteristic diffraction peaks of the major crystalline phases—including calcite, C₃S, C₂S and C₄AF—show no significant changes in both position and relative intensity after electrolyte exposure, with no detectable phase transformations or additional diffraction peaks (Fig. 2i). Quantitative phase analysis using Rietveld refinement further confirms that the mass fractions of all constituents, including the amorphous calcium silicate hydrate (C-S-H) gel, remain essentially unchanged. FTIR spectra reveal that the characteristic absorption bands of C-S-H gel—Si-O stretching vibrations (950–1100 cm^–1), Q² tetrahedral Si-O stretching (870–880 cm^–1), and Si-O-Si/O-Si-O bending vibrations (500–520 cm^–1)—are preserved after electrolyte immersion, with no evidence of bond cleavage or silicate depolymerization^43,44. These results demonstrate that cement exhibits structural and chemical stability in the molten salt electrolyte environment, making it a relevant binder for high-temperature battery applications. The C-G electrodes with an inorganic framework also exhibit high thermal stability. Under flame-induced thermal shock, graphite electrodes fabricated with PVDF or PTFE binders suffered structural failure, whereas cement-graphite electrodes maintained structural integrity (Supplementary Fig. 15). These findings demonstrate that the C-G electrode successfully integrates the high crystallinity and conductivity of graphite with the structural stability and porous architecture of hydrated cement. The combination of scalable processing, stable hydration products, and a uniform composite structure creates a robust conductive framework, offering a reliable solution for high-temperature electrochemical applications.

Electrochemical performance

To evaluate the electrochemical performance, an Al | |C-G battery was assembled using the C-G composite (40% cement by mass) as positive electrode, Al foil as negative electrode and a ternary AlCl₃-NaCl-KCl molten salt electrolyte (Supplementary Fig. 16). Prior to battery assembly, the cured composite electrode was dried at 200 °C for 3 h to remove crystalline and adsorbed water (Supplementary Fig. 17). Cyclic voltammetry (CV) curves, recorded at scan rates of 1–3 mV s^–1, exhibit multiple redox peaks, corresponding to the stepwise intercalation and deintercalation of AlCl₄^– ions within the graphite structure (Supplementary Fig. 18). At a low specific current of 80 mA g^–1, the battery retains a reversible capacity of 78.89 mAh g^–1 after 190 cycles with an average Coulombic efficiency of 98.21%. When the specific current is increased to 200 mA g^–1, the capacity remains stable at 78.56 mAh g^–1 over 450 cycles, maintaining an average Coulombic efficiency of 98% (Supplementary Fig. 19). This stable performance at slow rates confirms that the C-G electrode is not reliant on kinetic suppression of side reactions for its high efficiency, underscoring its intrinsic material stability. At a specific current of 1 A g^–1 and a mass loading of 15 mg cm^–2, the Al | |C-G battery demonstrates a discharge specific capacity of 78.3 mAh g^–1 after over 1100 cycles, with minimal capacity change (Fig. 3a). This stability is reflected in the reproducible charge-discharge profiles over extended cycling (Fig. 3b). The voltage plateaus align with the redox potentials observed in cyclic voltammetry, consistent with reversible intercalation/deintercalation behavior. When tested at 5 A g^–1 with a loading of 16 mg cm^–2, the cell maintains 78.9 mAh g^–1 for 5000 cycles with an average Coulombic efficiency of 99.1% (Supplementary Fig. 21). Notably, even at a higher specific current of 10 A g^–1, the electrode sustains a specific capacity of approximately 80 mAh g^–1 after over 11,000 cycles (Fig. 3e). These results collectively demonstrate the structural and electrochemical stability of the C-G electrode at high current densities and during long-term cycling, providing a basis for its practical application in high-performance batteries.

Fig. 3. Electrochemical performances of the C-G electrodes at 150 °C.

a Cycling performance and b Voltage profiles of the Al | |C-G battery at a specific current of 1 A g^–1. c Rate performance and d corresponding voltage profiles of the C-G composite electrode at current densities of 1–20 A g^–1. e Cycling performance of the Al | |C-G battery at an active material loading of 16 mg cm^–2 and a high specific current of 10 A g^–1. f The corresponding voltage profiles of the C-G composite electrode under active material loadings of 35, 50 and 80 mg cm^–2, respectively. g The impact of active material loading (15–100 mg cm^–2) on the areal capacity of the C-G electrodes. h Comparison of the electrochemical performance presented in our work and previous reports. The data are selected based on the loading of active material mass, specific capacity of the last cycle, and specific current. The source of the literature data shown in this figure can be found in Supplementary Table 2. Source data are provided as a Source Data file.

The observed rate capability of aluminum-graphite batteries stems from the pseudocapacitive intercalation behavior of AlCl₄^– anions into the graphite structure. Unlike diffusion-controlled intercalation in conventional battery materials, the insertion and de-insertion of AlCl₄^– between graphite layers exhibits charge-transfer kinetics with a low diffusion barrier (approximately 0.01 eV)¹⁵, enabling electrochemical reactions with low polarization. This intrinsic kinetic advantage is further amplified by the architectural design of the C-G electrode. The cement binder, through its C-S-H gel, forms a continuous three-dimensional porous structure that enhances the microstructure of the C-G electrodes, achieving a porosity of 20.5%. This structure improves electrolyte penetration and increases the contact area between the electrolyte and active materials, facilitating efficient ion transport. As a result, the electrode can operate under high active material loading and elevated current densities. The electrode delivers discharge capacities of 78.4, 78.5, 79.5, 79.4, 79.6, 79.4, 79.2, and 60.9 mAh g^–1 at current densities of 1, 2, 3, 4, 5, 10, 15, and 20 A g^–1, respectively (Fig. 3c). When the specific current is returned to 1.0 A g^–1, the specific discharge capacity recovers to 78.9 mAh g^–1. Across these varying current densities, the charge-discharge curves exhibit a stable voltage plateau, indicating low polarization characteristics and rapid ion diffusion kinetics (Fig. 3d). As shown in Supplementary Fig. 22, the areal capacity of the C-G composite electrode with an active material loading of 13 mg cm^–2 remains stable across various current densities. Within a current density range of 13–200 mA cm^–2, the electrode maintains a stable areal capacity of 1.03 to 1.04 mAh cm^–2. This consistent capacity shows the electrode’s electrochemical performance under high loading and elevated current densities. These properties position the C-G composite electrode as a promising candidate for applications that demand both high energy and high power densities. At a mass loading of 35 mg cm^–2, the electrode sustains a specific capacity of approximately 80 mAh g^–1 after 750 cycles at a specific current of 1 A g^–1. Similarly, under higher loadings ranging from 50 to 80 mg cm^–2 at the same specific current, the Al | |C-G batteries deliver a stable discharge specific capacity around 80 mAh g^–1 for over 100 cycles (Supplementary Fig. 23). As shown in Fig. 3f, the charge-discharge curves for electrodes with mass loadings of 35, 50, and 80 mg cm^–2 at 1 A g^–1 reveal specific capacities of 78.7, 79.2, and 79 mAh g^–1, respectively. Even at a high loading of 100 mg cm^–2 and a specific current of 0.5 A g^–1, the electrode delivers a discharge capacity greater than 70 mAh g^–1, with stable performance over 240 cycles (Supplementary Fig. 24). Increasing areal capacity effectively reduces the relative costs of current collectors and separators, which is a crucial factor for improving both the energy and power densities of batteries while simultaneously lowering production expenses⁴⁵. Consequently, areal capacity stands as a pivotal metric for evaluating the performance of electrochemical energy storage systems. Figure 3g illustrates the relationship between the mass loading of active battery material (mg cm^–2) and the resulting areal capacity (mAh cm^–2) at a specific current of 1 A g^–1. As mass loading increases from 15 to 80 mg cm^–2, the areal capacity exhibits a clear linear increase without a clear plateau or decay. This suggests that, across the tested range, the electrode structure avoids significant limitations in ion diffusion or charge transfer despite the increased thickness.

The specific discharge capacity of the Al | |C-G battery in this study is competitive with some recent reports on Al-graphite battery systems, despite employing a higher mass loading of active material^9,15. This result relates to addressing the typical limitations of high-mass-loading electrodes^46–49, which often suffer from capacity loss caused by restricted ion diffusion or charge transfer bottlenecks (Fig. 3h). Overall, the C-G electrodes exhibit stable cycling, rate capability, and operation under high mass loadings. Together with a Coulombic efficiency above 95%, limited capacity fade, and mechanical resilience, these properties are relevant to developing durable molten-salt Al batteries.

The charge-discharge plateaus observed in the Al | |C-G molten salt battery correspond to the intercalation reactions of graphite at different stages. To elucidate the underlying differences in reaction kinetics, electrochemical impedance spectroscopy (EIS) measurements were performed at different voltages during the charging and discharging processes (Fig. 4a, b). Supplementary Fig. 25 presents the equivalent circuit used for quantitative analysis, where R_s represents the total resistance arising from the electrolyte and electrode materials, and R_ct denotes the resistance associated with the charge transfer process at the electrolyte-electrode interface. Notably, R_s remained largely consistent across multiple tests. Since R_s is primarily associated with the resistive properties of the electrolyte and electrode materials, it remains stable under constant temperature conditions. The decrease in R_ct at the charge-discharge voltage plateaus indicates that the C-G electrode exhibits stability throughout the intercalation reaction process (Fig. 4c).

Fig. 4. Kinetic analysis and practical demonstrations of the Al | |C-G cells.

a–c In situ EIS characterization. a The marked data points on the charge-discharge voltage profile indicate the specific potentials at which EIS measurements were performed. b Nyquist plots of the EIS data. c Corresponding fitted impedance parameters derived from EIS spectra, and the data are presented in Supplementary Table 3. d Freezing-thawing test. The cell was terminated and cooled down to 25 °C for 12 or 24 h during charge and discharge processes, respectively. The specific current was 2 A g^–1. e Cycling performance of the Al | |C-G battery during multiple freeze-thaw cycles. Practical application demonstrations: f Open-circuit voltage of four Al | |C-G cells connected in series. g Photograph of the four cells powering an electronic display. h–j SEM image and elemental mapping of the C-G electrode after cycling. The electrode was disassembled at 150 °C in the fully discharged state after 200 cycles at 2 A g⁻¹, where the cell delivered a specific capacity of approximately 80 mAh g⁻¹. Scale bar: 10 μm. Source data are provided as a Source Data file.

To further evaluate the practical application potential of the Al | |C-G molten salt battery, freeze-thaw experiments were conducted. The molten salt electrolyte, which remains conductive in the liquid state at elevated temperatures, allows normal charge and discharge operations. Upon cooling and solidification, the electrolyte effectively preserves the stored chemical energy. After 25 charge-discharge cycles at 2 A g^–1, the battery was cooled to 25 ± 5 °C for 12 or 24 h by switching off the oven. Upon reheating to 150 °C, the Al | |C-G molten salt battery resumed stable operation. Repeated freeze-thaw tests revealed consistent cycling profiles (Fig. 4d), along with stable discharge specific capacity and Coulombic efficiency (Fig. 4e). At 150 °C, the high mobility of AlCl₄^– anions in the molten salt electrolyte promotes gradual charge redistribution between the graphite positive electrode and aluminum negative electrode under open-circuit conditions, resulting in significant self-discharge. After charging the cell to 80 mAh g^–1 and storing it under open-circuit conditions for 4, 12, 24, and 48 h, the cell retained 95.11%, 91.64%, 88.09%, and 85.79% of its initial discharge capacity, respectively (Supplementary Fig. 26). The elevated-temperature self-discharge in our system is attributed to faster ion mobility in the molten salt electrolyte^50,51. However, this drawback is offset by the freeze–thaw capability of the molten salt electrolyte: upon cooling below its melting point, solidification imparts ion-insulating behavior that suppresses self-discharge—as shown by no significant change in the time-voltage profile after storing the battery at 25 °C for 24 h (Supplementary Fig. 27). Such performance may be useful for renewable energy storage systems-such as those coupled with solar or wind power-that are subject to seasonal or diurnal intermittency. In a practical demonstration, four cells connected in series delivered a stable open-circuit voltage of 8.02 V (Fig. 4i), sufficient to power a 6.5 V LED display (Fig. 4j).

Electrode stability analysis

To elucidate the long-term structural and chemical stability of the self-supporting C-G electrode in molten salt Al batteries, comprehensive morphological and elemental analyses were conducted on electrodes after 200 cycles. The electrode exhibits no obvious macroscopic morphological changes before and after cycling (Supplementary Fig. 28a, b). SEM images and corresponding elemental distribution analyses reveal no significant structural deterioration-such as cracking, delamination, or active material loss-after extended cycling. Graphite particles retain a continuous layered morphology, while the cement-based binding is uniformly distributed between graphite, forming a stable three-dimensional network that preserves the mechanical integrity and microstructural stability of the composite electrode (Fig. 4h–j and Supplementary Fig. 28c, d). TEM images further confirm that the C-S-H gel within cement hydration products remains stable after cycling, with no detectable phase transformation or decomposition (Supplementary Fig. 29a).

To further substantiate the phase composition and chemical stability, we conducted post-mortem analysis using XRD coupled with Rietveld refinement, FTIR spectroscopy, and X-ray Photoelectron Spectroscopy (XPS) on both pristine and cycled electrodes. XRD patterns, analyzed via Rietveld refinement, demonstrate that the principal crystalline phases in the cement matrix—calcite, C₃S, C₂S, and C₄AF—exhibit no detectable shifts in peak positions or changes in relative intensities after 200 charge-discharge cycles (Supplementary Fig. 30a). High-resolution O 1 s XPS spectra exhibit a prominent peak at approximately 532 eV, attributed to oxygen in Si-O-M bonds (where M = Si or Ca) within the C-S-H gel. This peak remains sharp and unchanged following cycling, indicating that the local chemical environment of the silicate framework is fully preserved, with no evidence of oxidation or bond cleavage (Supplementary Fig. 31a, b). FTIR spectroscopy provides further evidence of the structural integrity of the C-S-H gel. The characteristic absorption bands associated with Si-O stretching (950–1100 cm^–1), Q² tetrahedral Si-O stretching (870–880 cm^–1), and Si-O-Si/O-Si-O bending modes (500–520 cm^–1) show no measurable shifts in position, broadening, or intensity loss after prolonged cycling. These observations indicate that the silicate backbone of the C-S-H gel remains intact, with no detectable bond breaking or structural degradation of the network (Supplementary Fig. 31c).

Scalability and economic prospects

The scalability of the C-G electrode intrinsically is linked to its fabrication process. As illustrated in Fig. 5a, these electrodes can be manufactured through slurry mixing and molding-a method closely aligned with conventional practices in the cement and concrete industries. This industrial compatibility allows for large-scale manufacturing using existing infrastructure, while eliminating the need for metallic current collectors or polymeric binders. The capability to form robust, self-supporting sheets through such widely adopted methods provides a characteristic relevant to practical implementation. This technology is relevant for stationary storage applications. Conceptual layouts of grid-level installations (Fig. 5b) illustrate how C-G-based molten-saltAl batteries could be coupled with renewable energy sources such as wind and solar power. Together with low cost and long service life, these properties are relevant for grid balancing, industrial backup power, and remote energy storage applications. To validate the feasibility of the C-G electrode for practical application, we fabricated and assessed a single-layer laminated Al | |C-G pouch cell. Galvanostatic charge-discharge tests were conducted at a constant current of 80 mA. The prototype cell exhibits stable voltage plateaus and electrochemical reversibility, delivering a discharge capacity of 116 mAh (Fig. 5c).

Fig. 5. Scalable fabrication and economic prospects of the C-G molten-salt Al batteries.

a Schematic of the scalable preparation process of the C-G electrodes using slurry casting. b Conceptual design of large-scale stationary storage modules integrated with renewable energy. c Galvanostatic charge–discharge profiles of Al | |C–G pouch cell. Detailed parameters are provided in Supplementary Table 10. d Structural model of a single molten-salt Al battery cell employing the C-G electrodes. e Cost distribution analysis of different battery components, highlighting the dominant contribution of the positive electrode. f Comparative cost analysis between the C-G molten-salt Al batteries and conventional systems. Source data are provided as a Source Data file.

The basic configuration of a single molten-salt Al cell is shown in Fig. 5d. The positive electrode consists primarily of a C-G composite, while both the Al negative electrode and the molten-salt electrolyte are made from low-cost, earth-abundant materials. This design does not use scarce transition metals or fluorinated polymers, which is relevant to environmental and economic considerations. Component-level cost analysis indicates that nearly 90% of the total cost originates from the electrode, with separators, electrolytes, and packaging contributing the remaining portion (Fig. 5e). In conventional lithium-ion batteries, electrolytes, separators, and current collectors constitute major cost components⁵². A component-level cost analysis indicates that the Al | |C-G batteries have an energy cost of 66.7 USD/kWh. For reference, conventional graphite-LiFePO₄ systems have an energy cost of 94.4 USD/kWh (Fig. 5f and Supplementary Table 4–5). When accounting for thermal management hardware and associated energy losses, the system-level cost of the Al | |C-G battery increases to an estimated 85–115 USD/kWh (Supplementary Table 6–8). However, this thermal overhead is partially offset by the high-temperature performance of the system, including its kinetics and cycle life. Moreover, in deployments—such as stationary storage integrated with industrial waste heat or applications using internal Joule heating from high-rate cycling—external heating demands can be reduced.

Discussion

In conclusion, we report a self-supporting C-G composite electrode for molten-salt Al batteries, which can serve as an alternative to electrodes fabricated with polymer binders in high-temperature, corrosive electrolytes. The electrode design employs graphite to establish a continuous electron-conducting network, while the hydrated cement phase acts as a mechanically robust and chemically inert inorganic framework with a hierarchical porous structure. This structure exhibits thermal stability and facilitates electrolyte infiltration through its hierarchical porosity, thereby supporting reaction kinetics. With a mass loading of 16 mg cm^–2 at a specific current of 10 A g^–1, the electrode achieves a lifespan over 11,000 cycles. It also retains a specific capacity of about 80 mAh g^–1 at 15 A g^–1 and exhibits stable cycling even under high mass loadings of up to 100 mg cm^–2. Beyond its economic benefits, this work discusses the use of cement in electrochemical systems. Here, cement serves as a scaffold that is stable under high-temperature molten-salt conditions. By integrating the electrical conductivity of graphite with the stability and scalability of hydrated cement, we describe a structural-functional material that connects construction materials with energy storage technologies. Overall, this study discusses C-G based molten-salt Al batteries for large-scale energy storage. The combination of structural integrity, electrochemical durability, and safety is relevant to terawatt-hour-scale deployment. More broadly, this work describes how common structural materials can be used in energy systems.

Methods

Electrolyte preparation

The preparation of the inorganic chloroaluminate salt AlCl₃-NaCl-KCl adheres to the eutectic ratio indicated by the phase diagram. NaCl (99.99%, Aladdin) and KCl (99.99%, Aladdin) were dried in a muffle furnace at 300 °C for 24 h prior to use and subsequently transferred to a glove box. AlCl₃ (98 + %, Thermo Fisher) was used as received. All the operations below were performed inside the argon-filled glovebox (Condition: O₂ < 0.01 ppm, H₂O < 0.01 ppm). The precursor powders were mixed in a specific molar ratio of NaCl: KCl: AlCl₃ (26:13:61), placed in a sealed weighing bottle, and heated from 20 ± 15 °C to 160 °C at a rate of 1 °C/min, followed by a 24-h hold. After cooling to 20 ± 15 °C, the solidified melt was crushed and ground into a powder for use in battery assembly.

Synthesis of the cement-graphite composite electrode

High-rate graphite (Canrd) and ultrafine cement (fineness: 800 mesh, vendor: Henan Tusheng Highway Engineering Co., Ltd., certified under ISO9001:2015) were mixed at specific mass ratios according to the experimental design and manually stirred initially to ensure uniformity. Subsequently, deionized water was added at a water-to-solid mass ratio of 0.5:1, along with a polycarboxylate ether superplasticizer (PCA®-Ⅰ series, Sobute; solid content ~40 wt%) at a dosage of 0.2–0.4 wt% relative to the cement mass. The cement-graphite slurry was initially stirred slowly, with the stirring speed gradually increased to 600 rpm as the fluidity improved, followed by high-speed stirring for 3 min to obtain a cement-graphite composite slurry with flowability and moldability. The slurry was poured into molds of specified shapes. The molds were first vibrated on a vibration table for 60 cycles (1 s per cycle), followed by scraping the surface to achieve a smooth finish. The formed cement-graphite composite electrode was cured for 28 days under ambient conditions at a temperature of 20 ± 1 °C and a constant relative humidity of 60% to complete the cement hydration process, yielding a self-supporting composite electrode with adequate mechanical strength. Finally, the electrode was dried at 200 °C for 3 h to remove crystal water.

Fabrication procedure for the graphite electrode with a PTFE binder

A positive electrode slurry was prepared by mixing high-rate graphite, carbon black (Super P), and PTFE powder (8:1:1 by mass, all from Canrd) in isopropyl alcohol (Sinopharm, ≥99.7%). After uniform grinding, the slurry was calendered using a roller press (Shenzhen Kejing) to a dense, uniform sheet of approximately 200 μm thickness. The resulting electrode was vacuum-dried at 60 °C for 12 h to remove residual solvent.

Fabrication procedure for the graphite electrode with a PVDF binder

A binder solution was prepared by stirring PVDF powder (Canrd) in N-methylpyrrolidone (≥99.5%, Aladdin) at a 1:15 mass ratio until homogeneous. The positive electrode slurry was prepared by dispersing high-rate graphite (Canrd) and conductive carbon black (Canrd) into the binder solution at a graphite: carbon black: PVDF mass ratio of 8:1:1, followed by stirring until homogeneous. The resulting slurry was coated onto a 16-μm-thick aluminum foil current collector (Canrd) using a doctor blade (Deliao Technology Development Co., Ltd.), with a controlled coating thickness of 100 μm. The coated electrode was vacuum-dried at 60 °C for 12 h to remove residual solvent.

Materials characterizations

XRD patterns were recorded using a Bruker D8 Discover X-ray diffractometer equipped with a Cu-Kα radiation source to analyze the crystal structure of the samples. Rietveld refinements were conducted using the HighScore Plus software, with initial structural models obtained from the ICSD database. Diffraction patterns with a high signal-to-noise ratio were selected for refinement. A stepwise refinement strategy was employed, consisting of background fitting, peak-profile refinement, and subsequent refinement of structural parameters. Structural parameters, including atomic coordinates, were refined only for major phases with phase contents exceeding 3 wt.%, whereas minor phases were retained as the initial structural models without further refinement. The quality of the refinement was evaluated using the difference curve and the reliability factors (Rwp, Rp, and χ²). XPS tests were carried out using an AXIS SUPRA spectrometer (Shimadzu, Japan). FTIR spectra of the samples were recorded using a Thermo Fisher Nicolet iS50 Fourier-transform infrared spectrometer. Sample preparation for XRD, XPS, and FTIR analysis of samples exposed to molten salt electrolyte shall be carried out in a glovebox to exclude moisture and avoid hydrolysis of the electrolyte. SEM images were obtained using a JEOL JSM-7100F field emission scanning electron microscope to observe the microscopic morphology and perform elemental distribution analysis. Raman spectra were collected using a HORIBA LabRAM HR Evolution Raman spectrometer (integrated with SmartSPM functionality) to study the molecular vibration modes of the materials. STEM was conducted using a Thermo Fisher Talos F200S transmission electron microscope to characterize the nanostructure, while EDS was employed for elemental mapping. In addition, EELS was performed on the same instrument to investigate the local electronic structure. Non-destructive three-dimensional imaging was carried out using an Xradia 510 Versa high-resolution micro-CT system to obtain the overall morphology and pore network of the electrodes. The instrument resolution was 0.7 μm, and the image reconstruction software used was Dragonfly. The bending and compressive properties of the composite electrodes were evaluated using an Instron 5967 mechanical testing system with a load capacity of 30 kN. Nanoindentation tests were conducted using a Bruker Hysitron TI 980 nanomechanical testing system. After vacuum resin impregnation and curing, the cement sample was sequentially ground using silicon carbide papers from 240 to 4000 grit, followed by polishing with diamond suspensions of decreasing particle sizes. The sample was thoroughly cleaned before each change of abrasive. The polishing duration was extended as needed to achieve the surface quality required for nanoindentation testing.

Electrochemical characterizations

The Al | |C-G molten salt battery was assembled using a commercial aluminum foil (thickness: 0.2 mm, diameter: 12 mm) as the negative electrode, a single layer of GF/D glass fiber (Canrd, thickness: 0.62 ± 0.01 mm, diameter: 12 mm, porosity: >90%, pore size: 2.7 μm) as the separator, and the quaternary chloroaluminate melt as the electrolyte, with the C-G composite electrode (5 mm × 8 mm) serving as the positive electrode. The assembly was conducted in a custom-designed Swagelok® cell equipped with a protective sheath, with the electrolyte mass precisely controlled at 200 mg per cell. The assembled cells were placed in a constant-temperature drying oven (WHL-25AB, Tianjin Taisite Instrument Co., Ltd.) at 150 °C for 8 h to facilitate thorough electrolyte infiltration, after which all measurements were performed at the same temperature. Constant current charge-discharge tests were carried out using a NEWARE multichannel battery testing system. CV and in situ EIS measurements were conducted using a CHI760E workstation. The EIS measurements were performed at 150 °C during galvanostatic charge and discharge of the battery. Once the system reached a preset potential, impedance data were acquired immediately. A potentiostatic mode was employed with an AC signal amplitude of 5 mV and a frequency range from 0.01 Hz to 10⁵Hz. Each spectrum consisted of 84 data points, and a stabilization period of 2 s was applied before each measurement. All scientific graphing was conducted using Origin software. Raw galvanostatic charge–discharge data were processed with the Neware Battery Test Data Analysis Software to extract key electrochemical parameters and perform a comparative analysis.

Supplementary information

Supplementary Information ^{(27.4MB, pdf)}

Transparent Peer Review file ^{(3.5MB, pdf)}

Source data

Source Data ^{(11.2MB, xlsx)}

Author contributions

J.M., K.L., X.Q. and F.W conceived the study. K.L., J.M., and C.H. designed the experiments. K.L., Y.Q., M.W., and J.W. prepared the cement-graphite composite electrodes and performed structural characterizations. K.L., X.Y., and M.Z. carried out electrochemical measurements, while K.L. and F.L. conducted XRD, SEM, and TEM analyses. K.L. and Y.Q. provided resources and assisted with CT analysis and schematic illustrations. K.L., J.M., and X.Q. wrote the manuscript with input from all authors. J.M., S.H., X.W., and C.S. Poon supervised the project.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by Fundamental and Interdisciplinary Disciplines Breakthrough Plan of the Ministry of Education of China (JYB2025XDXM408); the National Key Research and Development Program of China (2024YFA1211800); the Natural Science Foundation of Hubei Province (2025AFA105); National Natural Science Foundation (Grant No. 52472033, 52130208).

Data availability

The data that support the plots within this paper and other findings of this study are available from the corresponding author upon reasonable request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76400-y.