Artificial Hybrid Solid Electrolyte-Mediated Ca Metal for Ultradurable Room Temperature 5 V Calcium Batteries

2.1 Artificial Hybrid Solid Electrolyte Layer

The surface solid electrolyte at Ca, that is, AHSEL, was prepared by a solid–liquid–solid-guided reaction between Na, Ca, and NMP. SEM characterization proved that the surface topology of the pristine Ca plate (Figure S1) was modulated after the formation of AHSEL (Figure 1a). Large up-and-down fluctuations are replaced by a uniform roughness configuration of the solid electrolyte, appearing as extra details of the coating layer (Figure 1a,b). TOF-SIMS 2D mapping analysis of C⁺, N⁺, Na⁺, and Ca⁺ modes (Figure 1c) indicated the uniform existence of C, N, and Na elements in the surface coating. A distinct delaminated microstructure can be observed in the low-magnification cross-sectional SEM image of AHSEL-Ca (Figure 1d), indicating possible different compositions between AHSEL and the matrix Ca layer. The EDS element mapping depth profile (Figure 1e) of the high-magnification cross-sectional SEM image further shows the gradient distribution of Na and Ca across the layer microstructure. The quantitative line-scan EDS of the sample scraped from the surface of AHSEL-Ca was performed using high-angle annular dark-field scanning transmission microscopy (HAADF-STEM; Figure S2a,b), verifying the existence of these elements in AHSEL. The sudden increase in intensity of the Ca peak at 180 nm in the line-scan EDS is probably due to exfoliated Ca from the matrix. Moreover, the higher carbon content in the outer surface indicates that the organic carbon encapsulates the inorganic phase configuration of the AHSEL. TOF-SIMS depth profiling and 3D mapping analysis (Figure 1f) showed the C-, N-, Na-, and Ca-containing AHSEL surface and their gradient distribution configurations.

In addition to the major phase of cubic Ca (PDF No. 23-0430) from matrix Ca, the crystalline phases in the as-prepared AHSEL-Ca contain the common carbonates of sodium and calcium, that is, orthorhombic CaCO₃ (PDF No. 17-0763) and monoclinic Na₂CO₃ (PDF No. 19-1130), as well as an unusual Ca compound, that is, cubic Ca₂NH (PDF No. 71-0051), which were all verified by XRD (Figure S3a) characterization. Moreover, a wide amorphous package is also discernable at 2θ ≈ 26°, which is quite different from the XRD pattern for pristine Ca. Surface elemental analysis by XPS (Figure S3b) confirmed the presence of C, N, O, Na, and Ca in the outmost thin layer. Chemical bonding of C=O from carbonates (Na₂CO₃ and CaCO₃) at 289.2 eV and Ca–NH from Ca₂NH at 346.3 and 349.8 eV was also substantiated in the deconvoluted C 1s, N 1s, Na 1s, and Ca 2p XPS spectra (Figure S3c), which is well consistent with the XRD results, except for the extra bonding of C–C (284.6 eV), C–N (285.3 eV), and –C=O (288.3 eV), indicating the presence of organic carbon in the AHSEL.^{[24, 25]} Interestingly, although a high sodium content was confirmed at the interface by the EDS (Figure S3d,e) spectra, no N element was detectable, implying that the phase of Ca₂NH may be deeply embedded in organic carbon, which contains a very small amount of N according to the TOF-SIMS depth profiling analysis. HRTEM images (Figure 1g,h) of the sample scraped from AHSEL show that crystalline particles were fully encapsulated by the amorphous phase, possibly organic carbon, consistent with the results of XPS and TOF-SIMS analyses.

Nanocrystals smaller than 10 nm display well-defined lattice fringes, of which the diffraction pattern by fast Fourier transform (FFT), lattice spacings, and angles of crystal planes are consistent with those of sodium/calcium carbonates and Ca₂NH, as observed in the XRD pattern. Obviously, we have realized the protection of Ca by the surface AHSEL, which can avoid the direct Ca electrolyte reaction. In addition to a full protection, good ion conductivity is essential for AHSEL-Ca to be effective in Ca metal batteries. The electrochemical impedance spectrum (EIS) of an all-solid Ca//AHSEL-Ca cell showed that the artificial solid electrolyte exhibits a good ion conductivity of 0.01 mS cm⁻¹ (Figure S4).

2.2 AHSEL-Ca Symmetric Batteries

The Ca plating–stripping performance was tested in symmetric batteries to verify the protection efficiency of the AHSEL. Ca plating–stripping of pristine Ca in 1 m Ca(PF₆)₂ or Ca(BF₄)₂ EC/DEC (1:1, v/v) electrolyte exhibited very poor stability due to severe anion corrosion and polarization.^{[21, 22]} At current densities of 0.025 and 0.1 mA cm⁻², the Ca//Ca symmetric batteries can only last for less than 52 and 12 h, respectively. After that, the deposition potentials quickly shift to −5.0 V due to the extremely large internal resistance from mass formation of thick CaF₂ insulative layer.^{[21, 26]} In contrast, the plating–stripping of pristine Ca in a 1 m KPF₆ EC/DMC/EMC (1:1:1, v/v/v) electrolyte can be prolonged to more than 800 and 250 h under the same test conditions (Figure 2). This result is comparable to pristine Ca in the NaPF₆ electrolyte,^[21] because similar K/Ca hybrid SEIs characteristics of KCa(PO₃)₃ (PDF No. 39-1408), Ca(H₂PO₄)₂ (PDF No.70-1381), and CaF₂ (PDF No.77-2093) nanocrystals smaller than 20 nm evolved above the Ca deposition layer, which, to some extent, alleviated anion corrosion and hindered out-of-control fluoridation, as revealed by the SEM, XRD, EDS, TOF-SIMS, and HRTEM characterization of the cycled Ca electrode (Figure S5). However, fluoridation of fresh Ca deposit occurred slowly, leading to failure of Ca//Ca batteries in 258 h with a capacity of 0.6 mAh cm⁻² at 0.1 mA cm⁻² (Figure 2b). The result also suggests that the microstructure of K/Ca hybrid SEIs is still not compact enough and may lack some organic carbon species (as revealed by TOF-SIMS; Figure S5e) to fully avoid mild etching and passivation of freshly formed Ca.

In contrast, polarization and anion corrosion were radically suppressed using AHSEL-Ca. At the current rate of 0.1 mA cm⁻², the overpotentials for Ca deposition/dissolution have reduced by 0.31 and 0.90 V (from 0.31 to 0 V and from 0.98 to 0.08 V; Figure 2a). At 0.025 mA cm⁻², the potential barrier for Ca deposition and dissolution was also slightly lowered by 0.17 and 0.30 V, respectively (Figure 2c). Although a steady Ca deposition potential of −1.35 V versus Ca/Ca²⁺ is still much more negative than in ether-based electrolytes or with precious metal current collectors owing to the pseudo-reference of Ca,^{[12, 13, 26]} the result is still significant considering the use of cost-effective commercial Ca as the current collector. More importantly, the fluoridation of fresh Ca deposits fundamentally changes, enabling more than 1400 h stable plating–stripping cycles with a very small polarization shift of <0.4 mV/h at a capacity of 0.15 mAh cm⁻² (Figure 2d).

The interface microstructure characterization reveals that the surface of the cycled AHSEL-Ca appears as a uniform and compact coating, confirmed as ternary Na/K/Ca hybrid SEIs, which cover the mixture deposition layer of K/Ca, and then the matrix Ca below (Figure 3a–f). The major crystalline phases in the hybrid SEIs were verified as Ca₂NH from AHSEL and more from AHSEL–electrolyte interaction, while minor phases of hexagonal KCa(PO₃)₃, orthorhombic Na₃(PO₃)₃·H₂O (PDF No. 15-0740), and cubic KCaF₃ (PDF No.03-0567) were also observed in the XRD pattern (Figure S6a). The uniform distribution of N, Na, and K elements revealed by the 2D TOF-SIMS N⁺, Na⁺, K⁺, and Ca⁺ analyses, and their uniform distribution in the EDS mapping of the corresponding cross-sectional SEM and HAADF-STEM images (Figures S6b and S7, and Figure 3c–e) of the hybrid SEIs, demonstrates that the gradient distribution of Ca/Na elements in the pre-engineered interface of AHSEL has suffered considerable rearrangement through the interaction with the K-based electrolyte. Thus, the well-defined amorphous layer in AHSEL disappears. Instead, small amorphous zones remain interspersed among the gaps of nanocrystals, preventing their further growth, which accounts for the non-coarsening crystals of <10 nm, similar to that in the AHSEL (Figure S8 and Figure 3g). The HRTEM image (Figure 3h) shows that Ca₂NH nanocrystals have well-defined lattice fringes with plane spacings of 0.25 and 0.29 nm, consistent with the interplanar spacings of (222) and (400) reflections in the XRD pattern. The Ca₂NH-rich inorganic–organic Na/K/Ca hybrid SEIs were also confirmed by the surface N-/C-rich element distribution displayed by the TOF-SIMS depth profiling and 3D mapping analysis (Figure 3f), which is quite different from AHSEL with amorphous layer encapsulation N-rich Ca₂NH and that of K/Ca SEIs with few organic species. The C–N and C=N stretching vibrations, and the C–H stretching vibration of N–CH₃ moiety, are characterized by distinct absorbance peaks at 1026–1306, 1650, and 2855 cm⁻¹ in the FT-IR spectra (Figure S9) of the Na/K/Ca hybrid SEIs and precursor AHSEL, which both substantiate their unique organic species, quite different from that of K/Ca hybrid SEIs.^[24] XPS of the cycled AHSEL-Ca and after 120 s of plasma etching (Figure S10) also confirm the composition and variation of the SEIs, consistent with the above analysis. CN species and Ca₂NH nanocrystals in hybrid SEIs have at least two benefits. One is the enhancement of the charge-transfer conductivity for the SEIs (Figure S11), and the other is the direct inhibition of CaF₂ formation, as clearly revealed by the XRD patterns (Figures S5b and S6a), in sharp contrast to K/Ca or Na/Ca SEIs with CaF₂ surviving as the major component of the interphases.^{[21, 22]} Overall, the compact packing of crystalline–amorphous hybrid SEIs, especially the introduction of Ca₂NH nanocrystals, guarantees a selective rapid permeability of Ca²⁺, rather than ${\text{PF}}_6^{-}$, and thus fundamentally solves the problems of passivation and corrosion of Ca metal by forming CaF₂.

2.3 High-Voltage and Ultradurable AHSEL-Ca Batteries

FGCFM was chosen as the cathode for full Ca metal batteries and AHSEL-Ca as the anode. The membrane electrode, which has good ion/electron conductivity, can avoid the adverse effects of the binder, and simultaneously afford many active sites for energy storage, benefiting from the few-layered graphitic nanostructure, many heteroatom-doping defects, and a large surface and porous structure, as revealed by the SEM, TEM, EDX, XRD, XPS, and N₂ adsorption–desorption characterization (Figures S12 and S13).^[27-29] To test the safe reliability and anodic stability of AHSEL-Ca under deep plating–stripping conditions, a high window voltage of 2.0–5.0 V is preset. The Ca battery delivers an initial discharge capacity of ≈86 mAh g⁻¹ at 200 mA g⁻¹, and a high capacity of ≈79 mAh g⁻¹ can be retained after 50 cycles, with the average discharge voltages ranging from 3.32 to 3.47 V (Figure 4a). The voltage profiles are characteristics of two plateaus at about 3.8–4.9 V and 2.0–2.9 V, with a steep slope between them. The two plateaus correspond to the cathodic peaks at 2.41 and 4.40 V in the differential capacity curves (inset view of Figure 4a), similar to those of previous Ca metal batteries,^[21] to be discussed in the following section. After 200 cycles (Figure 4b), a capacity of ≈60 mAh g⁻¹ was retained, which is equivalent to a superior capacity retention of ≈70%. To facilitate comparison, a cycle with a coulombic efficiency of 90% was chosen for the rate analysis. At various rates ranging from 200 mA g⁻¹ to 1 A g⁻¹ (Figure 4c), the average discharge voltage holds around high voltage of 3.13–3.30 V. In the meantime, ≈66% of the reversible capacity at 200 mA g⁻¹ can be output at 1 A g⁻¹ (50 mAh g⁻¹ vs 76 mAh g⁻¹). The Ragone plots (Figure 4d) show that an impressive energy density of ≈250 Wh kg⁻¹ can be achieved at a power density of 657 W kg⁻¹, while at a large power density of 3.19 kW kg⁻¹, the Ca battery can still deliver a high energy density of ≈161 Wh kg⁻¹, comparable to advanced dual-ion batteries, Al-ion batteries, and Ca metal batteries reported previously.^{[17, 21, 22, 30]}

Except for good energy/power performance, Ca metal batteries also exhibit good long-term cycling stability. At 300 mA g⁻¹, the Ca battery ran stably for 1000 cycles with a very small capacity decay of ≈0.05% per cycle. More impressively, the coulombic efficiencies are roughly above 95%, except for the initial aging cycles, needed to evolve a stable interphase for the cathode.^{[21, 22]} Even at a very high current rate of 1 A g⁻¹, the Ca battery can safely run for nearly 5000 cycles, with coulombic efficiencies being roughly larger than 90% for the initial 3000 cycles and slowly declining to above 80% subsequently. Notably, the capacity decay always holds at a very low level of ≈0.01%, except the initial aging cycles. Evidently, though Ca metal batteries were tested under high voltage up to 5 V, high rate, and long duration, it is verified that the AHSEL-Ca metal is suitable for safe and deep Ca plating–stripping battery operation conditions. This indirectly proves the feasibility of AHSEL in decoupling metal electrodes and liquid electrolytes, and points to a possible direction for realizing safe metal batteries with high energy density.

2.4 Electrochemical Energy Storage Mechanism

Because the charge-transfer mechanism for the anodes of all metal batteries follows the same equation as M ↔ Mⁿ⁺ + ne⁻, Ca plating–stripping of AHSEL-Ca has also been investigated in symmetric batteries (Section 2.2). Hence, we mainly analyzed the mechanism of the cathode. The ex situ XRD patterns for the FGCFM cathode at various stages of charge–discharge (SOC) are presented in Figure 5a. During the first charge process from the open circuit voltage (OCV) to 5 V, the left shift of the (002) reflection and weakened intensity of the (101) reflection of the graphite-2H structure indicates that the insertion reaction of ${\text{PF}}_6^{-}$ anions occurs in the graphitic carbon due to intercalation or adsorption reactions according to Bragg law (d_(hkl) = nλ/2sinθ, where d_(hkl) is the interplanar spacing of the (hkl) crystal plane, θ is the diffraction angle, and λ reflects the wavelength of the incident radiation, for example, X-ray or neutron), similar to those of dual-ion and anion-cation relay batteries.^{[17, 31]} The discharge process features two plateaus, as mentioned above, the upper voltage corresponding to the extraction reaction of ${\text{PF}}_6^{-}$ anions, which leads to the recovery of characteristic reflections for the graphite-2H structure, and the lower voltage one assignable to the insertion reaction of cations, K⁺ and Ca²⁺, as confirmed by the slight shift of the (002) reflection.

Considering the lack of an obvious shift, it is reasonably speculated that the cation insertion reaction mainly arises from chemical adsorption of defects, pores, and some surface/interface organic groups in the graphitic carbon.^[32] This also accounts for almost no shift of the (002) reflection in the early charge stage before 4 V for the subsequent charge process, implying the desorption of the absorbed cations. When further charging to 5 V, an obvious shift of the (002) reflection to the left appears again as a result of the reinsertion of anions into the cathode. The reflection shift repeatedly recovers after a subsequent discharge to 3 V and then reappears because of cation adsorption for discharge to 2 V afterward. Interestingly, the relevant adsorption–desorption reaction of cations results in a very small reflection shift for (002), but leads to a significant weakening of the (101) reflections. These variations in the XRD profiles are clearly reflected in the corresponding contour color map. Overall, an anion–cation relay storage mechanism can be rationally deduced for the cathode (C_n + x ${\text{PF}}_6^{-}$ ↔ C_n(PF₆)_x + xe⁻, C_n + yK⁺ + zCa²⁺ ↔ C_n(K_y/Ca_z) + (y + 2z)e⁻, where C_n represents the graphitic carbon). The variation of insertion–extraction species is also reflected in the EDS results. The normalized intensity of EDS spectra (Figure 5b) for cathode at various SOCs indicates a clear rule that peak intensity of K and Ca elements (relative content) is strengthened at the cation adsorption stage of 2.0 V, with weakened peak intensity for P, while peak intensity of P increases with decreasing peak intensity of K and Ca elements at anion-intercalated stage of 5.0 V, confirming the anion–cation relay storage mechanism. The contour color map vividly reflects the intensity variation of the normalized EDS spectra. The oxidation and reduction states of graphitic carbon in the relay insertion/extraction process can be analyzed using the deconvoluted C 1s, F 1s, and P 2p XPS spectra (Figure 5c). From reduction state to oxidation state, the doping O species, that is, C=O bonding at the lower energy state, is oxidized to C–O–PF_n (n < 5) species, that is, C–O–P bonding at higher energy state, in the charge, as revealed by the weakening peak intensity of C=O bonding at 288.7 eV for C 1s, and new emerging P–F bonding at 687.4 eV and P–O bonding at 134.5 and 136.5 eV of spin split peaks of P 2p from the C–O–PF_n species.^[33]

Furthermore, the EDS element mapping in HAADF-STEM mode and the HRTEM images for the cathodes at different discharge depths (Figure 6) indicate the same mechanism, in coincidence with the XRD and XPS results. At the discharge depth of 0, that is, anion-intercalated stage, element mapping and HRTEM images of the cathode (Figure 6a–c) are characteristics of very expanded lattice spacings of 0.52 nm, even 0.68 nm for graphitic layers, and finer color for P element due to inclusion and aggregation of ${\text{PF}}_6^{-}$ ions. The lattice spacing recovered to about 0.38 nm after a discharge depth of 50%, indicating the extraction of anions (Figure 6d–f). As the discharge process progresses to 100% discharge depth, a slight increase in the graphitic lattice spacing to 0.42 nm is attributed to the insertion of K-/Ca-based cations due to adsorption or intercalation reactions, in addition to the brighter colors for the corresponding element mapping images (Figure 6g–i). A more reliable quantitative element analysis (Figures S14–S22) similarly indicates an anion–cation relay insertion–extraction mechanism accompanying a K⁺/Ca⁺ competitive adsorption process, implying that the total number of active adsorption sites for is constant. Almost equally divided parts of anion insertion/extraction and cation adsorption/desorption in the discharge depth profile also illustrate their nearly equivalent capacity contributions.