Revisiting the layered Na3Fe3(PO4)4 phosphate sodium insertion compound: structure, magnetic and electrochemical study

Layered sodium iron phosphate phase [Na3Fe3(PO4)4] was synthesized by solution combustion synthesis method, marking the first attempt of solvothermal synthesis of this phase. Its crystal structure was verified by synchrotron and neutron powder diffraction. Rietveld analyses proved the phase purity and formation of monoclinic framework with C2/c symmetry. It undergoes an antiferromagnetic ordering ∼27 K. This combustion prepared nanoscale Na3Fe3(PO4)4 compound was found to be electrochemically active with a stepwise voltage profile involving an Fe3+/Fe2+ redox activity centred at 2.43 V vs. Na/Na+. Despite various cathode optimization, only 1.8 Na+ per formula unit could be reversibly inserted into the Na3Fe3(PO4)4 framework leading to capacity close to 50 mAh g−1. This limited electrochemical activity can be rooted to (i) relatively large diffusion barrier (ca. 0.28 eV) as per Bond valence site energy (BVSE) calculations and (ii) possible structural instability during (de)sodiation reaction.


Introduction
With industrial and technological evolution, global energy consumption is increasing at a very high rate. In this scenario, energy generation with minimal CO 2 emission is pivotal. Also, efficient energy storage and delivery is a key sector propelling myriads of consumer electronics, (hybrid) electric vehicles and stationary micro-to-mega grid storage in the 21st century. While several technologies do exist, electrochemical energy storage in general and rechargeable batteries in particular offer the most pragmatic approach with feasible large-scale dissemination. In the last three decades, rechargeable Li-ion batteries have seen unprecedented commercialization ushering a wireless revolution and vying for zero-emission transportation. Batteries with good combination and energy/power density, reversibility, safe operation and materials/process economy are crucial to meet the demand of the energy-hungry world [1,2]. From application point of view, batteries can be divided into two categories: volume/weight restricted portable batteries and volume/weight independent stationary batteries. While the former is solely dominated by Li-ion chemistry, the latter category can be catered by alternate chemistry like lead-acid, Ni-MH, Li-ion as well as Na-ion batteries.
While the Li-ion and Na-ion batteries work on similar operating principles, sodium-ion batteries are touted as economic alternatives to Li-ion counterparts owing to the elemental abundance and uniform geographic distribution of sodium-based precursors. Similar to the story of Li-ion batteries, successful implementation of Na-ion batteries relies a lot on robust positive insertion (cathode) materials. Over the last decade, Na-ion batteries have attracted significant research effort to develop new cathode materials, where layered transition metal oxides (Na x T M O 2 , T M = Co, Fe, Mn, Cr etc) have ruled the field ever since their discovery in 1980s [3][4][5][6][7][8]. However, these layered have issues with structural and thermal stability. To circumvent these issues, many a polyanionic insertion materials have been unravelled. While the theoretical capacity (Q Th ) is lower owing to high Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. molecular weight, using inductive effect principle, superior operating voltage can be realized leading to competitive energy density [9,10]. Among these polyanionic materials, while the SO 4 2− -based compounds offer the highest redox voltage [11,12], the PO 4 3− -based ones render easy synthesis and materials stability [13]. Exploring the phosphate chemistry, various polyanionic cathodes have been reported such as NaFePO 4 , NaVPO 4 F, Na 3 V 2 (PO 4 ) 2 F 3 , Na 2 FePO 4 F, Na 2 FeP 2 O 7 , Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 ) etc [6]. One such PO 4 -based cathode system is Na 3 Fe 3 (PO 4 ) 4 , a layer structured sodium iron phosphate [14,15]. It offers suitable channels for Na + ion insertion between the layers built from corner-sharing FeO 6 octahedra and PO 4 tetrahedra [16], with a theoretical insertion ability of 3 Na per formula unit. However, the previous report realized the reversible intercalation of only 1.8 Na + ions per formula unit of Na 3 Fe 3 (PO 4 ) 4 framework, that too in very low current rage of C/100 (1 Na + in 100 h) [14]. The layered structure and high theoretical capacity based on 2 electron transfer motivated us to re-examine and optimize this compound to achieve better electrochemical performances.
In this current work, we have synthesized Na 3 Fe 3 (PO 4 ) 4 framework via wet chemistry route for the first time. Its crystal structure was analysed combining synchrotron and neutron powder diffraction. Na 3 Fe 3 (PO 4 ) 4 exhibits layered structure assuming a monoclinic system with C2/c symmetry. Solution route leads to smaller particles with improved electrochemical reversible capacity of 50 mAh g −1 at a faster rate of C/20 involving a 2.45 V Fe 3+ /Fe 2+ activity. Despite particle size optimization and carbon coating, it was impossible to obtain near theoretical capacity. Here, we report the wet chemical synthesis, structure, magnetic and electrochemical properties of as-synthesized Na 3 Fe 3 (PO 4 ) 4 . The Na + migration and diffusional analysis has been calculated to probe the reason behind electrochemical limitations.

Materials synthesis
The target compound Na 3 Fe 3 (PO 4 ) 4 was synthesized via wet chemistry route for the first time. The wet chemical synthesis was carried out with or without using fuels such as glycine (C 2 H 5 NO 2 , 99%, Merck), urea (CH 4 N 2 O, 99%, SDFCL) and ascorbic acid (C 6 H 8 O 6 , 99%, Sigma Aldrich). Stoichiometric quantities of NaH 2 (PO 4 ).H 2 O (99%, EMPARTA.ACS), Fe(NO 3 ) 3 .9H 2 O (98%, Fisher Scientific) and (NH 4 ) 2 HPO 4 (98%, Merck) were dissolved in de-ionised water separately and then mixed together to have a final precursor solution having a pH value of 2. This solution was placed on a hot plate (maintained at 120°C) with steady magnetic stirring to remove excess water. After complete drying, the solid residue was ground by an agate mortar and pestle was pressed into pellets using hydraulic press. These pellets were calcined in a tubular furnace at 400°C for 6 h, 600°C for 6 h and finally at 750°C for 48 h to get the final product. This annealing condition was used for precursor mixture without using any fuel. In contrast, when fuels were mixed to precursor solution, the desired product was obtained by one-step calcination at 750°C for 6 h. A fire yellow coloured powder was finally obtained.

Structure and morphological analysis
The powder diffraction patterns of Na 3 Fe 3 (PO 4 ) 4 samples were acquired by a PANalytical X'Pert Pro diffractometer equipped with a Cu-Kα source (λ 1 =1.5405 Å, λ 2 = 1.5443 Å) operating at 40 kV/30 mA. Typical XRD patterns were collected using Bragg-Brentano geometry in the 2θ window of 5°-90°with a step size of 0.0267°(count time = 110 s per step). Synchrotron diffraction pattern was collected at the BL-18B Indian beamline (High Energy Accelerator Research Organization, KEK-Photon Factory, Tsukuba, Japan) using a synchrotron beamline of wavelength λ = 0.7861(2) Å and energy E = 15.77 keV calibrated with Si (640b NIST) standard. Neutron powder diffraction (NPD) pattern was acquired with Echidna high-resolution diffractometer at OPAL facility (Lucas Heights, Australia) using neutrons of wavelength 2.4395 Å. Rietveld analysis of the powder diffraction patterns was conducted using GSAS [17,18] or FullProf [19] suite of programs. The background was refined using a shifted Chybeshev polynomial function and the diffraction profile was fitted by a pseudo-Voigt function. Brunauer-Emmett-Teller (BET) surface area measurements were calculated with a Micromeritics ASAP 2020 instrument using surface adsorption of N 2 (at 77 K). Prior to BET test, the powder samples were evacuated at 373 K under vacuum for 2 h. The morphology of as-synthesized product was examined using an FEI Inspect F50 scanning electron microscope (10 kV) as well as an FEI Tecnai F30 STwin transmission electron microscope (200 kV). For TEM analysis, few drops of powder sample, soaked in acetone, were deposited on a copper grid.

Magnetic characterization
Temperature-dependent magnetic susceptibility data over a 4-300 K range were collected for Na 3 Fe 3 (PO 4 ) 4 under field-cooled (FC) and zero field-cooled (ZFC) conditions, in an applied magnetic field of 1000 Oe. Fielddependent data were collected at 4 K over the range ±5000 Oe. Measurements were made using a Quantum Design Physical Property Measurement System (PPMS) equipped with a Vibrating Sample Magnetometer (VSM).

Electrochemical characterization
The electrochemical properties of Na 3 Fe 3 (PO 4 ) 4 were tested (at 25°C) in sodium half-cell architectured CR2032 coin type cells. Composite positive electrode (slurry) was prepared by thoroughly mixing the active material (85 wt%) with carbon black (10 wt%) and polyvinylidene fluoride binder (5 wt%) in N-methyl-pyrrolidone. The slurry was coated uniformly with brush onto thin circular Al-foil (∅ = 16 mm) and was dried overnight in vacuum oven at 90°C. The coin cells were assembled in an argon-filled glove box (MBraun Inc.) using Na metal  foils as counter electrode and a sheet of Whatman filter paper as separator soaked with 1.0 mol.l −1 NaClO 4 in propylene carbonate as electrolyte. These cells were galvanostatically cycled between 2-2.55 V (versus Na/Na + ) at a rate of C/50 (i.e. 1 Na in 50 h at 25°C) using a VMP3 Biologic battery cycler.

Results and discussion
3.1. Synthesis and structure Phosphate based compounds are important among the polyanionic insertion materials as they are durable over high temperature and provides more chemical stability along with good electrochemical properties. From elemental point of view, Fe-based cathodes are projected as economic candidates due to the abundance of Febased precursors. Nonetheless, the processing cost plays the spoiler as most Fe-based cathode materials are based on Fe 2+ species warranting careful processing and packaging in inert (N 2 /Ar) atmosphere. Switching towards Fe 3+ based cathode materials [like layered Na 3 Fe 3+ 3 (PO 4 ) 4 ] can reduce the processing cost ($/watthour) of material as whole process can be carried in air instead of inert (Argon) atmosphere. Also, conventional solid-state synthesis involves high temperature and prolonged calcination time which augment the processing cost of material. In search of alternative novel synthesis, wet-synthesis and solution combustion synthesis involving short calcination duration can be employed for scalable production of variety of cathode materials. Here, (fuel-assisted) solvothermal route was employed for energy-savvy synthesis of layered Na 3 Fe 3 (PO 4 ) 4 .
Without any fuel, wet synthesis was conducted involving multi-step calcination at 400°C and 600°C for 6 h followed by final heat treatment at 750°C for 48 h. The phase evolution with various calcination steps is depicted  in figure 1. While the intermediate product (after dehydration) was amorphous in nature, progressive phase evolution occurred with high temperature annealing. Usage of wet synthesis favouring intimate mixing of precursors led to reduction of final annealing duration from (previously reported) 72 h to 48 h [14,15]. Faster reaction can be realized by triggering exothermic combustion reaction in presence of fuels that leads to intimate precursor mixing/metal-nitrate complexation, exothermic reaction involving high local heating, rapid product formation with porous (nanoscale) morphology along with one-step carbon coating [20,21]. When combustion synthesis was employed, facilitated by complexation and partial reaction completion during low temperature step, the final calcination duration (at 750°C) was drastically reduced to 6 h independent of the fuels used (figure 2). Structural and phase purity was analysed combining lab/synchrotron XRD and neutron powder diffraction. Rietveld refinement confirmed the formation of desired target product as shown in  4(a)-(c)). This aggressive grain growth and large particle size results from prolonged thermal treatment. BET analysis revealed the overall surface area to be in the range of 2-3 m 2 g −1 with   (table 2). High-resolution TEM study further confirmed the formation of inhomogeneous micrometric Na 3 Fe 3 (PO 4 ) 4 particles in the size range of 1-2 μm (figures 4(d)-(e)). SAED pattern could be indexed to monoclinic structure (figure 4(f)). The crystallinity of final product was further confirmed by HRTEM study showing lattice planes corresponding to miller indices (202) with an interatomic d-spacing of 4.5 Å (figure 4(g)).

Magnetic properties
The presence of Fe imparts magnetic characteristic to Na 3 Fe 3 (PO 4 ) 4 , similar to many Fe-based cathode materials with antiferromagnetic ordering at low temperature. Temperature-dependent magnetic susceptibility data for Na 3 Fe 3 (PO 4 ) 4 under field-cooled (FC) and zero field-cooled (ZFC) conditions are presented in figure 5. An antiferromagnetic ordering transition is clearly observed at T N = 27 K. There was no significant offset between the FC and ZFC data, consistent with the absence of any opening in the field-dependent data at 4 K (inset to figure 5). A Curie-Weiss fit to the FC data over the range 100-300 K (red line in figure 5) yields an effective magnetic moment μ eff = 5.95 μ B /Fe, consistent with the theoretical spin-only value for high-spin Fe 3+ , μ s.o. = 5.92 μ B /Fe. A detailed analysis of long range magnetic ordering and elucidation of antiferromagnetic structure of Na 3 Fe 3 (PO 4 ) 4 from low-temperature neutron diffraction data will be reported shortly.

Electrochemical performance of Na 3 Fe 3 (PO 4 ) 4
Electrochemical performance of wet chemistry prepared Na 3 Fe 3 (PO 4 ) 4 was analysed using standard Na half-cell assembly. Being an Fe 3+ insertion compound, Na 3 Fe 3 (PO 4 ) 4 was first discharged to uptake further Na atoms to realize Fe 3+ /Fe 2+ redox activity. Ideally, it will be great to utilize Fe 3+ /Fe 4+ redox activity by charging (desodiating) the starting compound. However, the reversible conversion of Fe 3+ -Fe 4+ involving FeO 6 octahedral environment is structurally unstable and this redox can occur beyond safe operation limit of commonly used electrolytes. Galvanostatic voltage-capacity profiles of layered Na 3 Fe 3 (PO 4 ) 4 is shown in figure 6(a). When tested at a rate of C/50 (1 Na + in 50 h), the first discharge capacity of 66 mAh g −1 was observed (theoretical capacity ∼130 mAh g −1 ). In subsequent cycles, ∼1.6 Na + ions could be reversibly inserted into the structure delivering a discharge capacity ∼48 mAh g −1 . Similar to many polyanionic cathodes (like Li 2 FeP 2 O 7 ), the capacity decreased from 1st to 2nd cycle hinting at irreversible structural rearrangement during the first discharge. Afterwards, the cell performed with steady capacity (ca. 48 mAh g −1 ) with 98% Coulombic efficiency. A closer look revealed sloping step-wise profiles with several distinct pseudo-plateaus. The major Fe 3+ /Fe 2+ redox peak appears ∼2.5 V followed by several minor redox peaks ∼2.42 V and 2.14 V (vs. Na/Na + ) ( figure 6(b)). This sloping profile is indicative of solid-solution (single-phase) redox mechanism during (de) sodiation reaction. While the original report involving prolonged solid-state synthesis showed poor electrochemical activity even at a slow rate of C/100 [14,15], the usage of solvothermal synthesis (delivering smaller particles) led to similar electrochemical performance at improved rate of C/50. Nonetheless, layered Na 3 Fe 3 (PO 4 ) 4 is found to have poor Na + diffusion despite all attempts on processing/material optimization.
To understand the reason behind poor electrochemical activity in Na 3 Fe 3 (PO 4 ) 4 , the Na + diffusional kinetics and pathways were explored using bond valence site energy (BVSE) calculations, which is a simple yet reliable approach to speculate ion migration pathways in structural models. The detail methodology of BVSE analysis has been reported elsewhere [22][23][24]. In this calculations, the bond valences s Na−X = exp[(R 0,Na−X −R Na−X )/b Na−X ] and the BV sum mismatch |ΔV| can be correlated to an absolute energy scale by expressing the bond valence site energy E BVSE of a Na + cation coordinated by X − anions as a Morse-type interaction energy: The required bond valence parameters were taken from softBV database [25]. Migration pathways for Na + are then investigated as regions of low bond valence site energy E BVSE (Na) in grids spanning the structure model with a resolution of ca. (0.1 Å) 3 . Using Rietveld refinement of neutron diffraction of Na 3 Fe 3 (PO 4 ) 4 sample, the initial structure model was obtained for BVSE calculations. Based upon structure model, energy barrier for Na + migration as hollow spheres connecting the Na and interstitial sites in Na 3 Fe 3 (PO 4 ) 4 was obtained and represented as superimposed projection of the crystal structure along ac plane in figure 7. Na 3 Fe 3 (PO 4 ) 4 has two sites for Na (Na1 and Na2), where Na2 has lowest energy barrier of 0.15 eV for hopping to other Na2 via the interstitial sites i1-i1 with hopping energy barrier of 0.17 eV as illustrated in figure 7 (top). Hopping of Na2 also takes place from i2 interstitial site with high energy barrier of 0.28 eV along one dimensional pathway to Na2. Three-dimensional percolation of Na2 is possible involving Na1 with highest migration energy barrier of 0.30 eV from the interstitial sites i3 site. Using VETSA, one-dimensional as well as three-dimensional percolation pathways are superimposed over crystal structure of Na 3 Fe 3 (PO 4 ) 4 along ac plane as shown in figure 7 (bottom). Migration of Na in Na 3 Fe 3 (PO 4 ) 4 follows one-dimensional pathway along Na2-i1-i1-Na2 (figure 7 black arrows) which are more feasible than three-dimensional percolation pathway along Na2-i3-Na1 due less energy barrier (figure 7 red arrows). Overall, it offers feasible one-dimensional Na + diffusivity with high energy barrier, which is the root cause behind the poor rate kinetics in this system.

Conclusions
This work presents the first wet chemistry synthesis and layered sodium iron phosphate Na 3 Fe 3 (PO 4 ) 4 insertion system. Using this solvothermal synthesis, the target compound was prepared by less aggressive heat treatment resulting subdued grain growth/Ostwald ripening vis-à-vis solid-state synthesis. Rietveld analysis confirmed the formation of monoclinic layered system. Microstructural analysis showed large agglomerated platelets of 5-7 μm range with primary anisotropic platelet shaped particles in the size range of 1-2 μm. It exhibited a longrange antiferromagnetic ordering below (T N =) 27 K. Wet synthesis improved the Na + diffusion kinetics in Na 3 Fe 3 (PO 4 ) 4 when tested in sodium cells. Involving a nominal Fe 3+ /Fe 2+ redox potential centred around 2.43 V, it delivered a reversible capacity of 48 mAh g −1 . The poor Na + insertion behaviour in Na 3 Fe 3 (PO 4 ) 4 can be rooted to structural disordering during initial cycle and occurrence of one-dimensional Na + diffusion pathways along [101] direction with high energy barrier of 0.28 eV. While far from any practical application, Na 3 Fe 3 (PO 4 ) 4 forms an economic Fe 3+ -based sodium insertion material.