MRBs are attracting the attention to the scientific community as a complement technology to the well-known Li-ion batteries [1,2]. Several cathodic compounds including Chevrel phase (Mo₆S₈) [3], transition metal oxides (V₂O₅, MnO₂, Mn₃O₄) [4–6], layered chalcogenides (MoS₂, TiS₂ and WSe₂) [7–10], spinels (MgMn₂O₄ and Mg₂MnO₄) [11–13], V-based sulfides [14], silicates (MgMnSiO₄) [15,16], phosphates (Na₃V₂(PO₄)₃) [17,18], nanorod-encapsulated CuS quantum dots [19] among others, are under study targeting to find the most suitable material for this new approach. On the electrolyte part, the major challenge lies on developing safe electrolytes, which demonstrate appropriate electrochemical voltage window and compatibility with metallic Mg [20]. Undoubtedly, an interesting material for this type of emerging technology is the NASICON-type compound because among its virtues one can found it has stable framework and facile ion transportation [17,18,21].

Herein, to obtain multi-electron reaction cathode materials for Mg batteries, a two-metal doped vanadium-based compound with Na₃V_2-x-yA_xB_y(PO₄)₃ (A = Cr(Ⅲ) and B = Fe(Ⅲ) with x = 0.5 and y = 0.5) stoichiometry is proposed. It is worthy to note the previous study with Na₃VCr(PO₄)₃ in Mg cell exhibiting sodium and/or magnesium co-intercalation reaction in the Mg cell where the energy density is increased due to the activation of the V⁴⁺/V⁵⁺ redox couple [21]. In the field of Na ion batteries such kind of structure received lot of attention in the last decade [22–25], but in the case of Mg batteries are still in the early stage [26].

The Na₃VCr_0.5Fe_0.5(PO₄)₃ is synthesized via a simple sol-gel method, followed by thermal annealing at 750 ℃ under argon atmosphere. The structure exhibits a lantern-like three-dimensional open framework that consists of corner-sharing MO₆ octahedra (with M = V/Cr/Fe) and PO₄ tetrahedra (Fig. 1). Six lanterns' units build up the primitive cell where two different oxygen environment interstitial sites exist. There are two types of Na sites: Na1 (6b) and Na2 (18e) having 6-fold and 8-fold coordination with oxygen atoms. XRD patterns show that the diffraction peaks can be indexed in a rhombohedral system with an R-3c space group (Fig. 2), and around 16° the peak has been assigned to a superstructure [27,28]. The detailed atomic positions are listed in Table S1 (Supporting information). The lattice parameters (Table 1) are similar to that reported by Li et al. [25], smaller than that of Na₃V₂(PO₄)₃ and bigger than those of Na₃VCr(PO₄)₃, as expected due to the crystal radius of V³⁺ (0.78 Å), Fe³⁺ (0.785 Å), and Cr³⁺ (0.755 Å) [21,29].

SEM image of the pristine sample shows that the particles surface is not rough, has rounded edges and an average diameter 0.5–1 µm (Fig. 3). The EDS images show that the elements (Na, V, Cr, Fe, P) are distributed uniformly on the particles and confirmed the atomic percentage of the compound (Fig. 2, inset). Moreover, the ICP results confirmed that the Na: V: Cr: Fe atomic ratio for the pristine electrode is 2.986:1.010:0.494:0.495.

To investigate the textural properties, nitrogen adsorption-desorption measurements were performed (Fig. 3). The specific surface area, pore volume and pore diameter were 3.94 m²/g, 0.024 cm³/g, and 3.82 nm, respectively. In addition, Raman spectrum unveiled vibrational modes of the phosphate groups (~1050 cm⁻¹) belonging to Na₃VCr_0.5Fe_0.5(PO₄)₃ phase, and revealed that the carbon residua has a highly amorphous character. This carbon phase is characterized by two main bands at ca. 1345 ± 4 cm⁻¹ and ca. 1582 ± 4 cm⁻¹ known as D and G bands and they are assigned to the lack of long range translation symmetry in disordered domains and 'in plane' displacement of carbon atoms in the crystalline domains of graphene sheets with E_2g symmetry, respectively [30,31]. The elemental CHNS analysis determined 5.6% of carbon content. Therefore, the addition of propanol during the formation of the gel yields, after further calcination, a carbonaceous residua which is helpful to improve the intrinsic electronic conductivity of Na₃VCr_0.5Fe_0.5(PO₄)₃ particles.

Figs. 4A and B show the galvanostatic and derivative curves obtained for Mg/Na₃VCr_0.5Fe_0.5(PO₄)₃ cell. Upon charging, it exhibits two pseudo-plateaus at 2.25 and 3.15 V, whose derivative curves revealed a broad peak between 1.75-2.6 V and 3-3.3 V, respectively. It is worthy to note that the theoretical capacity for one electron reaction is 58.4 mAh/g (ca. extraction or insertion of one sodium ion). The first charge the capacity is 105.2 mAh/g (extraction of 1.8 Na⁺ ion per unit formula). Once that the Na⁺ ions are extracted, the electrode is washed to avoid the presence of sodium in the electrolyte during the following cycles (see experimental section in Supporting information). However, the rest of the mobile sodium ions (ca. 0.2 Na⁺ per unit formula) remained in the structure and may affect the cycling properties of the magnesium battery. In any case, the sodium content is low compared to the concentration of the magnesium electrolyte and can be treated as trace contents. On discharge, several reduction peaks are observed at around 3.05, 2.15, 1.6 and 0.23 V. The charge and discharge capacities from first to fifth cycles are around 106-116 and 107-101 mAh/g, respectively. These capacities match with the consumption of one Mg²⁺ per unit formula (theoretical capacity: 116.8 mAh/g). The cyclic voltammetry curves revealed the reversibility of the reaction in the range between 1.25 V and 2.5 V (Fig. S1 in Supporting information), but the peak at 0.23 V disappears in the the following cycles suggesting the irreversibility of magnesium ions at low potential which is in agreement with the galvanostatic and derivatives curves. A comparison of the different electrochemical properties of NASICON-type structures used in magnesium cells is shown in Table 2. Depending on the magnesium cell configuration (electrolyte, cathode and anode) the electrochemical behaviour showed different performances [17,21,26,32–35].

Regarding the cyclability and capability, the Na₃VCr_0.5Fe_0.5(PO₄)₃ can be cycled at 30 ℃ with 65% capacity retention after 100 cycles, and afford good rate performance of 104.3, 97.5, 90.7, 77.7 and 60.5 mAh/g at 10, 15, 20, 30 and 40 mA/g, respectively (Figs. 4C and D). For the sake of comparison, the electrochemical performance of Na₃VCr(PO₄)₃ with the same cell configuration has been studied (Fig. S2 in Supporting information). The first charge and discharge capacity is 112.5 and 97.5 mAh/g, respectively. Differently to the Na₃VCr_0.5Fe_0.5(PO₄)₃ sample, the charge and discharge capacities of Na₃VCr(PO₄)₃ decreased during the first 5 cycles to 73.7 and 70.5 mAh/g. This loss of the capacity at room temperature is ascribed to the loss of the high-voltage capacity due to the migration of V atoms [28,36]. Therefore, the presence of both (iron and chromium) in the vanadium sites provides structural stability when cycling the magnesium cell at 30 ℃ (Fig. 4A).

The changes in the structure, ⁵¹V local environment and chemical state are studied (Fig. 5). The Bragg peaks of the partially and fully charged electrodes shifted toward higher angles and some peaks disappeared indicating that the structure has changed (Fig. 3). A contraction of cell parameters is observed (Table 1) evidencing efficient extraction of sodium ions from 18e sites [21,28]. As the mobile sodium ions have been extracted from the structure and afterwards removed from the electrolyte, the magnesium ions are inserted during the cell discharge. It showed additional peaks (ca. 29° and 53°) due to the presence of a two-phase system with magnesium rich and poor phase. At the fully discharge state the values of the lattice parameters almost recovered, and hence the reversibility of the process. As Mg²⁺ (0.72 Å) is about 30% smaller than Na⁺ (1.02 Å) the former moves easier than sodium implying that the effect of ionic radii is important as recently found in olivine-type materials [16]. Then, a slight cell volume change of 4.2% during a complete cycle is observed. The ex-situ SEM image was recorded after five cycles and revealed that the average particle size is preserved, but the edges have changed. The particles tend to agglomerate with with less rounded edges. Therefore, no apparent volume variation is observed by SEM, in agreement with the ex-situ XRD.

The above phenomena occurred together with oxidation/reduction of the transition metal in Na₃V_2-x-yA_xB_y(PO₄)₃ (with A = Cr(Ⅲ) and B = Fe(Ⅲ)). The XPS spectra at the Cr 2p, Fe 2p and V 2p showed initial oxidation state of Cr^3+, Fe³⁺ and V³⁺ whose main components of the 2p_3/2 core level are at 577.9, 712.5 and 516.5 eV, respectively [37–39]. This confirms the trivalent state of the transition metals in the initial structure (Fig. 6). After the charge at 3.3 V, the Cr 2p and Fe 2p core levels remained unchanged. Contrarily, the V 2p spectrum showed two new asymmetric peaks at 517.4 and 518 eV which could be ascribed to vanadium oxidation to tetravalent and pentavalent states, respectively [21].

The ⁵¹V-NMR spectra of pristine, charged and charged/discharged electrodes are registered (Fig. 5B). It is worthy to note that the ⁵¹V-NMR signal of pentavalent vanadium compounds without localized d electron are detectable [21,28,40]. In consequence, the ⁵¹V signal of Na₃V^ⅢCr_0.5Fe_0.5(PO₄)₃ is not observed, but for the fully charged electrode at 3.3 V the signal of V⁵⁺ in NaV^VCr_0.5Fe_0.5(PO₄)₃ is detected at -689.1 ppm with their corresponding spinning sidebands (labeled as "+"). After a full discharge, the NMR signal disappeared entailing the presence of V³⁺ in NaMgVCr_0.5Fe_0.5(PO₄)₃ which should be the main phase present at this stage. It is obviously detected the presence of V⁵⁺ in the charged state of the electrode, implying that V⁵⁺ should be considered in the redox process. Therefore, while the redox couples such as V³⁺/V⁴⁺/V⁵⁺ participate in the electrochemical reaction, the other non-active redox couples such as Cr³⁺/Cr⁴⁺ and Fe²⁺/Fe³⁺ could serve as stabilizer to effectively buffer the volume change. These two electrons reaction (ca. by one Mg²⁺) induced 4.2% volume change which is smaller to the volume variation found in other NASICON-type cathodes such as Na₃MnTi(PO₄)₃ or Na₃V_1.5Cr_0.5(PO₄)₃ with ca. 8.1% and 7.8%, respectively [41,42].

The DSC results of Na₃VCr_0.5Fe_0.5(PO₄)₃ electrode charged to 3.3 V (104.6 mAh/g) revealed an exothermic process with the main peak appearing at 400–412 ℃ and a thermal effect of 72.3 J/g (Fig. 7). After the second charge to 3.3 V (116 mAh/g) the exothermic peak appeared at similar position, but the thermal effect is a little higher (ca. 105.4 J/g). The introduction of iron and chromium in the structure lead to an increase of the exothermic processes of ca. 32, 90 and 109 ℃ as compared to Na₃V₂(PO₄)₃, Na₃V₂(PO₄)₂F₃ and Na₄VMn(PO₄)₃ electrodes, respectively [43,44]. In view of these results, the choice of doping with iron and chromium could be a good option for batteries where high volumetric/gravimetric energy density and safety are important.

In conclusion, a Mg-coin-cell is fabricated using NASICON-type Na₃VCr_0.5Fe_0.5(PO₄)₃ cathode exhibiting two-electrons reaction reaching 170 Wh/kg. The V³⁺/V⁴⁺/V⁵⁺ redox couple participated in the reaction while the non-active Fe²⁺/Fe³⁺ and Cr³⁺/Cr⁴⁺ provided structural stability minimizing strains as confirmed by ex-situ ⁵¹V-NMR, XRD and XPS. The cell showed a capacity retention of 65% over 100 cycles and good rate performance at 30 ℃. The dual doping provides different functions to develop advanced electrochemical behaviour.

There are no conflicts to declare.

This work was supported by the Scientific Research Funds of Huaqiao University and Xiamen University Foreign Young Talents Program (No. G2022149004L). Sincere thanks to Prof. Z. Wei and Prof. Y. Yang from the College of Materials Science and Engineering (Huaqiao University) and College of Chemistry and Chemical Engineering (Xiamen University) for supporting the research activities. The author thanks Peixuan Lu for the kind discussions.

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2023.109391.