Na₂Fe(C₂O₄)(HPO₄): a promising new oxalate-phosphate based mixed polyanionic cathode for Li/Na ion batteries

Simple single-step hydrothermal synthesis was performed to obtain single crystals of the title compound. In this procedure, as obtained iron (II) chloride hexahydrate (99%, Alfa Aesar), disodium hydrogen phosphate (⩾99.0%, Merck), and oxalic acid dihydrate (98%, Alfa Aesar) were mixed in the ratio 1:3:2 (1 corresponding to 10 mmol) in a 23 ml Teflon-lined autoclave with 3 ml of distilled water, sealed and placed in an oven at 160 °C for 3 d. The autoclave was then extracted from the oven and cooled to room temperature. The contents were decanted onto filter paper and washed several times with deionised water and acetone. The orange single-crystalline product was then dried in an oven at 60 °C for 4 h.

A fine orange single crystal was selected and mounted on nylon loops in inert oil. The SXRD data were acquired using Mo Kα radiation (λ = 0.71073 Å) at 173 K, on a Rigaku SCX Mini diffractometer. The data processing was carried out using Rigaku CrystalClear 2.0. SHELX-2014 incorporated in the WinGX program was used to solve the crystal structure by direct methods followed by further refinement [22, 23]. Absorption corrections were performed semi-empirically from equivalent reflections using multi scans data. All non-hydrogen atoms were refined anisotropically. The atomic positions of the material are shown in table S1 (available online at stacks.iop.org/JPMATER/4/024004/mmedia).

Powder x-ray diffraction (PXRD) patterns were recorded using a Stoe STADI/P diffractometer operating in either transmission mode or Debye–Scherrer mode with Cu Kα₁ radiation (λ = 1.5406 Å) in the 2θ range 5–100°. In order to verify phase purity of the sample the datasets were refined by conventional Rietveld methods, using the GSAS package with the EXPGUI interface [23]. For this purpose, a fixed structural model, taken from the SXRD study, was used, and only lattice parameters and profile parameters were refined, including the modelling of some preferred orientation using the spherical harmonics model [24]. A JEOL JSM-6700F scanning electron microscope (SEM) was used to record the morphological features and elemental mapping of the sample. The instrument was also equipped with a field emission gun electron source. Secondary electron images were recorded with a tungsten filament electron source using an accelerating voltage of 5 kV for the hand-ground pristine sample, and 15 kV for ball-milled and composite samples. For atomic number contrast imaging a retractable backscattered electron detector was used. FTIR spectroscopy was recorded using a Shimadzu IR affinity-1 FTIR spectrophotometer in the range of 400–4000 cm⁻¹. The cells were opened inside the glovebox after 15 charge/discharge cycles and the samples recovered and washed with dry dimethyl carbonate (anhydrous, 99.8% Merck). Then capillaries were filled with dried sample inside the glovebox and PXRD were recorded using Mo Kα1 radiation (λ = 0.7107 Å) in the 2θ range 4–40° for both NIB and LIB. Thermogravimetric analysis (TGA) was carried out on a Stanton Redcroft STA-780 simultaneous TG-DTA, under an air atmosphere at a heating ramp rate of 10 °C min⁻¹.

A 44 mg of powdered material was packed into a polycarbonate capsule and suspended inside a polypropylene straw. This was then attached to a sample rod and lowered into a Quantum Design MPMS^® XL for SQUID Magnetometry measurements. The powder sample at 300 K was subjected to a magnetic field of 100 Oe (0.1 T) and the sample was cooled to 2 K while the magnetic susceptibility was recorded.

The crystalline material was the first ball milled for 1 h to form a fine powder using a Fritsch Pulverisette 8 mill. Then 0.6 g of powdered material was mixed with 0.3 g Super C65 conductive carbon black (Imerys) via ball milling for a further 30 min. The composite powder was then ground with 0.1 g polytetrafluoroethylene (PTFE) binder until a homogeneous mixture was obtained. Pellets were pressed using hand hydraulic press at 3 tonnes cm⁻². The self-standing 13 mm pellets (∼0.08 mm thickness) were dried in a vacuum oven overnight and directly transferred to glovebox to use as working electrode. Typically, cathode pellet electrodes containing ∼7–8 mg active material were tested in half cells using CR2325 (NRC Canada) coin cells with Na metal as the anode, 1 M NaClO₄ in propylene carbonate with 3% fluoroethylene carbonate as the electrolyte for NIB; and Li metal as the anode, LP30 (1 M LiPF₆ in ethylene carbonate: dimethyl carbonate = 1:1) as the electrolyte for LIB. The half-cells were cycled galvanostatically in the potential window of 1.7–4.3 V for NIB and 1.9–4.5 V for LIB at a rate of 10 mA g⁻¹ using a Biologic Macpile II system. The rate capability was tested by cycling at progressively increasing rates up to 100 mA g⁻¹ before returning to 10 mA g⁻¹.

The single-crystal XRD data are shown in table 1. The structure viewed along the a-axis is shown in figure 1(a). This complex structure consists of distorted Fe²⁺O₆ octahedra (figure 1(a)) which are linked into triangular chains by oxalate ligands along the c-axis. These chains, which may be regarded as derived from the triangular layers seen in KFe(C₂O₄)F and related materials [25] are linked into layers in the bc-plane by surrounding hydrogen phosphate groups. These, in turn, link stacked octahedra directly along the a-axis and corrugate the chains through the b-axis (figure 1(b)). Interspersed through the structure are Na⁺ ions at four distinct crystallographic positions. These may be regarded as four or five-coordinate to oxygen ligands (assuming Na–O bond lengths to 2.6 Å), from both phosphate and oxalate moieties. Half of the Na⁺ sites are arranged along the a-axis, with their coordination polyhedra alternately sharing faces and edges (labelled Na1 in figure 1(a)). This channel is the prime candidate for any ion migration, with the small possibility of a secondary channel along the b-axis.

Zoom In Zoom Out Reset image size

Rietveld refinement of the PXRD pattern from a hand-ground sample (figure 2(a)) confirmed that the sample is isostructural with the low temperature (−100 °C) single-crystal XRD data. We estimate only a small amount (≪5%) of an unidentified impurity (figure S1).

Zoom In Zoom Out Reset image size

SEM images reveal that the hand-ground sample shows an average particle size of 12–20 μm (figures 3(a) and (b)). Elemental mapping confirms that the elements Na, Fe, C, O, and P are homogeneously distributed throughout the sample (figure S2). As the particle size obtained from hand-grinding is too large for electrochemical measurements the sample was ball milled for 30 min and the particle size checked by SEM analysis.

Zoom In Zoom Out Reset image size

This revealed that the particle size was reduced to 2–4 μm (figure 3(c)). These materials typically exhibit poor electronic conductivity hence 30 wt. % of conductive carbon C65 was added and the mixture ball-milled for a further 30 min. SEM images of the composite show that there is a homogeneous mixture of sample and carbon C65 (figure 3(c)) with reduced particle size.

The pristine Na₂Fe(C₂O₄)(HPO₄) was further characterized by FTIR spectroscopy to determine coordination of atoms and to confirm the absence of H₂O or OH⁻ from the structure. Figure 4 confirms the absence of absorption peaks of H₂O or OH⁻, in the range 3000–4000 cm⁻¹. The absorption peaks at 438, 505, 778, 885 and 1095 cm⁻¹ correspond to Fe–PO₄ antisymmetric stretching, Fe–O symmetric stretching, δ(OCO) bending modes, ν(C–C) stretching vibration and asymmetric stretching modes of oxalate, respectively. The other absorption peaks at 1313 and 1606 cm⁻¹ represent ν_s(COO) involving dangling oxygen atoms. Another two main peaks at 1471 and 1662 cm⁻¹ correspond to ν_s(COO) vibrations of oxalate. In addition, TGA of the pristine material in air was performed to check the stability of the sample (figure S3). The plot shows that the compound is stable up to 320 °C.

Zoom In Zoom Out Reset image size

The magnetic susceptibility data are shown in figure 5. The value of the µ_eff at its plateau around 100 K (figure 5(a)) confirms the presence of solely Fe²⁺ ions in the structure (i.e. compatible with high spin d⁶, S = 2, µ_SO = 4.92 µ_B), with no evidence of significant impurity. The inverse susceptibility plot (figure 5(b)) suggests strong ferromagnetic (FM) interactions at high T, however a marked change of slope on cooling suggests a change to dominant antiferromagnetic (AFM) interactions below ∼150 K, and a long-range ordered AFM state below ∼10 K (figure 5(a)). It is possible that due to a variation of exchange pathways between magnetic ions within the crystal structure, FM clusters are initially formed which at lower temperature interact AFM. This AFM ordering seems rather weak, with a Weiss constant of −1.36 K calculated from the linear fit in figure 5(b), but is nonetheless sufficient to lead to a zero net moment below the transition temperature as the moments become fully compensated. A similar competition between FM and AFM correlations has previously been observed in La₂NiMnO₆ [26], although in that case the phenomenon was related to cation anti-site defects. Further studies are necessary to determine the precise mechanism in this case, which may be due to the enhanced structural complexity.

Zoom In Zoom Out Reset image size

3.3.1. Sodium-ion battery

In order to evaluate the electrochemical performance of Na₂Fe(C₂O₄)(HPO₄), pellet electrodes were fabricated using a 60:30:10 weight % ratio of active material: carbon C65: PTFE binder with a typical active mass loading of ∼7–8 mg. The typical dQ/dV plot of a stabilized half-cell cycled in the range of 1.7–4.3 V vs Na⁺/Na at a rate of 10 mA g⁻¹ for the tenth cycle, is displayed in figure 6(a). It reveals two distinct peaks at 2.91 V and 3.32 V during charge with corresponding peaks at 2.81 V and 3.12 V during discharge.

Zoom In Zoom Out Reset image size

Figure 6(b) shows the voltage profile for the tenth cycle, with the corresponding two pairs of plateaus.

The long term cycling performance of Na₂Fe(C₂O₄)(HPO₄) was evaluated at a rate of 10 mA g⁻¹ in the potential window 1.7–4.3 V vs Na⁺/Na, as shown in figure 6(c). The first cycle discharge capacity is ∼104 mAh g⁻¹, which gradually decreases for the first eight cycles before increasing and stabilizing at around ∼105 mAh g⁻¹ after 300 cycles. The initial cell polarization is relatively high for the first few cycles before progressively decreasing, associated with the increase in reversible capacity. The Coulombic efficiency is low for the first few cycles before improving in subsequent cycles and reaches ∼95%. These observations are presumably associated with an electrochemical grinding effect on extended cycling which leads to improved contact between the grains and conducting carbon additive. The rate capability of Na₂Fe(C₂O₄)(HPO₄) was evaluated at different current densities (10–100 mA g⁻¹) as shown in figure 6(d). At a current density of 10 mA g⁻¹, the capacity approached 96 mAh g⁻¹. Upon increasing the cycling rate, the reversible capacity progressively decreases. The average discharge capacities are ∼71, 67, 55, 46, 31 and 21 mAh g⁻¹ at rates of 20, 30, 40, 50, 75 and 100 mA g⁻¹ respectively. Once the rate was restored to the initial current density of 10 mA g⁻¹, the discharge capacity reached ∼94 mAh g⁻¹, which indicates 98% restoration of the initial capacity.

3.3.2. Lithium-ion battery

Na₂Fe(C₂O₄)(HPO₄) was also tested as a positive electrode for LIBs. Typically, galvanostatic charge–discharge experiments were performed in the potential window 2.0–4.5 V vs Li⁺/Li at a rate of 10 mA g⁻¹. Figure 7(a) shows the dQ/dV plots for the 15th and 100th cycles. The lithiation/ delithiation of Na₂Fe(C₂O₄)(HPO₄) shows two processes analogous with those observed when cycled vs sodium. The galvanostatic voltage profile for the 100th cycle corresponds to two plateaus with an average potential of ∼3.2 V (figure 7(b)). The two plateaus on charge/discharge are situated at 3.0/2.9 V and 3.56/3.32 V respectively, with an average voltage hysteresis of 0.15 V. This confirms a reversible electrochemical Li extraction/insertion process [27].

Zoom In Zoom Out Reset image size

The performance of Na₂Fe(C₂O₄)(HPO₄) on long-term cycling vs Li⁺/Li was tested by constant galvanostatic charge/ discharge from 2.0 to 4.5 V at 10 mA g⁻¹, as shown in figure 7(c). The first cycle discharge capacity was ∼71 mAh g⁻¹, corresponding to ∼76% of its theoretical capacity. The coulombic efficiency was low for the first few cycles, before progressively increasing to ∼96%. Analagous to the behaviour observed in NIBs the cell polarisation decreases and Coulombic efficiency increases with cycle number. The material exhibits a very stable reversible capacity without fading up to 300 cycles. Note, we believe that further synthesis optimization, i.e. reducing the particle size, can significantly improve the electrochemical properties by obviating the electrochemical grinding process.

The PXRD patterns after 15 charge and discharge cycles are shown in figure S4. Although poorly crystalline it is apparent that there were no significant crystallographic changes during charge and discharge.