Na - American Chemical Society

Article pubs.acs.org/cm

Na3V2(PO4)2F3 Revisited: A High-Resolution Diffraction Study M. Bianchini,†,‡,§,# N. Brisset,‡,# F. Fauth,∇ F. Weill,‡,# E. Elkaim,⊥ E. Suard,§ C. Masquelier,†,#,∥ and L. Croguennec*,‡,#,∥ †

Laboratoire de Réactivité et Chimie des Solides, CNRS-UMR#7314, Université de Picardie Jules Verne, F-80039 Amiens Cedex 1, France ‡ CNRS, Univ. Bordeaux, ICMCB, UPR 9048, F-33600 Pessac, France § Institut Laue-Langevin, 71 Avenue des Martyrs, F-38000 Grenoble, France ∇ CELLSALBA Synchrotron, E-08290 Cerdanyola del Vallès, Barcelona, Spain ⊥ SOLEIL Synchrotron, F-91192 Gif-sur-Yvette, France # RS2E, Réseau Français sur le Stockage Electrochimique de l’Energie, FR CNRS#3459, F-80039 Amiens Cedex 1, France ∥ ALISTORE-ERI, FR CNRS#3104, F-80039 Amiens Cedex 1, France ABSTRACT: Na3V2(PO4)2F3 is a material that has been attracting great interest as a potential positive electrode for Na-ion batteries. Its crystal structure was determined from single-crystal X-ray diffraction in 1999 by Le Meins et al. in the tetragonal space group P42/mnm at 298 K. In this work, we show how the use of very high angular resolution synchrotron radiation diffraction reveals a subtle orthorhombic distortion with unit-cell parameters of a = 9.02847(3) Å, b = 9.04444(3) Å, c = 10.74666(6) Å in the Amam space group. Although this only slightly impacts the structural framework of the material, it reveals a significantly modified distribution of Na ions. Furthermore, the crystal structure of the high-temperature form of Na3V2(PO4)2F3 (at 400 K) was determined for the first time. This allowed comparing the totally disordered distribution of Na ions in this case with the partially ordered one of the room-temperature phase. We report here on an original structure and on an original electrochemical signature for stoichiometric Na3V2(PO4)2F3, and we propose that fluctuations in the O/F ratio are at the origin of discrepancies found in the literature.

1. INTRODUCTION In the last 30 years, the world of portable electronics has been powered by Ni-Cd, Ni-MH, and then Li-ion batteries, which rapidly became the technology of choice for such applications. More recently, research and industry have been converging on the common goal of extending the use of high-energy-density technologies to massive transport and storage applications, even though important challenges must be overcome. Concerns have been raised about the future availability of lithium resources, which are located in remote and politically sensitive areas1,2 and about the extent to which Li-ion technology is economically sound.3 For this reason, much research has been addressed to lithium’s alternatives, such as Na-based batteries4 and, more specifically, to the Na-ion technology.5 Sodium has been the obvious material to replace lithium since it is abundant and inexpensive, while possessing the same external electronic structure (valence shell). Nevertheless, a few drawbacks are present: (i) a standard reduction potential of E° = −2.7 V vs. SHE for the Na+aq/Na redox couple is observed, which is less negative than that of Li+aq/Li (−3.04 V vs. SHE); (ii) a larger ionic radius (1.20 Å instead of 0.7 Å); and © XXXX American Chemical Society

(iii) a bigger molecular weight, implying a lower energy density. Yet, materials for Na-ion batteries could develop rapidly because of the essential similarity between Li+ and Na+ intercalation chemistry,6 which made it possible to build a sodium counterpart for many positive-electrode materials of interest for Li-ion batteries (NaCoO2,7,8 NaFePO4,9−11 NaVPO4F,12−14 Na3V2(PO4)3,15−17) while candidates for negative electrodes are more scarce.18−20 A significant fraction of recently proposed electrodes are phosphate-based polyanionic materials,21 which possess stable structural frameworks and whose potential can be tailored thanks to the inductive effect. Among them, vanadium-based fluorophosphates22 are of particular interest, because they exploit the PO43− polyanion and vanadium’s chemical versatility as, for example, in the Na3V2O2x(PO4)2F3−2x (0 ≤ x ≤ 1) family of compositions.23−27 Depending on x, the oxidation state of vanadium varies between 3+ and 4+ with a concomitant modification of the physical properties and of the electrochemical signature of the material. The two extreme members Received: May 6, 2014 Revised: June 18, 2014

A

dx.doi.org/10.1021/cm501644g | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials

Article

Table 1. Space Group and Cell Parameters of Na3V2(PO4)2F3 (and Other Na3V2O2x(PO4)2F3−2x Compositions) As Gathered from the Literature reference 36

Hao et al. Gover et al.32 Chihara et al.34 Liu et al.37 Shakoor et al.35 XRD ND DFT Le Meins et al.33 Song et al.38 Barker et al.31 Serras et al.23 Tsirlin et al.29 this work XRD ND Synchro this work XRD NDa Synchroa a

space group

a (Å)

b (Å)

c (Å)

volume (Å3)

announced composition

P42/mnm P42/mnm P42/mnm P42/mnm

9.15 9.0378(3) 9.04 9.04

9.15 9.0378(3) 9.04 9.04

10.86 10.7482(4) 10.74 10.735

909.22 877.94(6) 877.69 877.5

Na3V2(PO4)2F3 Na3V2(PO4)2F3 Na3V2(PO4)2F3 Na3V2(PO4)2F3

P42/mnm P42/mnm P42/mnm P42/mnm P42/mnm P42/mnm P42/mnm P42/mnm

9.04(5) 9.034(3) 9.034 9.047(2) 9.05 9.0304(5) 9.042 58 9.03051(2)

9.04(5) 9.034(3) 9.034 9.047(2) 9.05 9.0304(5) 9.042 58 9.03051(2)

10.73(9) 10.740(8) 10.679 10.705(2) 10.679 10.6891(9) 10.627 38 10.62002(3)

876.9(4) 876.7(3) 871.5 876.2(3) 874.64 871.68 869.0 866.06

Na3V2(PO4)2F3

P42/mnm P42/mnm

9.0358(2) 9.0358(2)

9.0358(2) 9.0358(2)

10.7403(4) 10.7412(4)

876.90(4) 876.98(5)

Na3V2(PO4)2F3

Amam Amam Amam

9.0288(6) 9.02847(3) 9.02847(3)

9.0426(6) 9.04444(3) 9.04444(3)

10.7402(5) 10.74666(6) 10.74666(6)

876.88(9) 877.544(6) 877.544(6)

Na3V2(PO4)2F3

Na3V2(PO4)2F3 Na3V2(PO4)2F3 Na3V2(PO4)2F3 Na3VIII0.4(VIVO)1.6(PO4)2F3 Na3(VIVO)2(PO4)2F

Combined refinement.

are Na3(VIVO) 2(PO4) 2F (for x = 1, often written as Na1.5VOPO4F0.5) and Na3VIII2(PO4)2F3 (x = 0). • Na1.5VIVOPO4F0.5 crystallizes in the P42/mnm space group and it can deliver a theoretical capacity of 156 mAh/g through two voltage-composition plateaus, at 3.6 and 4 V vs Na+/ Na.28−30 Tsirlin et al.29 proposed the existence of three different polymorphs with different space groups, depending on the short-, medium-, or long-range ordering of Na ions within the framework. The high-temperature Na-disordered form (αphase) is described in the highly symmetric space group I4/ mmm. At lower temperatures, Na becomes partially ordered in the β-phase (298 K, space group P42/mnm) and completely ordered (γ-phase) below 235 K (P42/mbc space group). • Na3VIII2(PO4)2F3 was first proposed as a positive electrode material by Barker et al.,31,32 while its crystal structure had been already described in 1999 by Le Meins et al.,33 also in the P42/ mnm space group. A maximal theoretical capacity of 192.4 mAh/g can be calculated for the extraction of three Na ions, although, until now, only the reversible extraction of two of them could be achieved in a Na-ion cell.27,34 Despite recent published studies,34,35 some of the properties of Na3VIII2(PO4)2F3 have not been observed in sufficient details yet: for instance, a close look at the literature reveals very strong discrepancies36 between the structural unit-cell parameters of “Na3V2(PO4)2F3”, as gathered in Table 1. The reason for this is the versatility of vanadium toward oxidation/ reduction and the concomitant existence of possible F/O mixing and substitution in the Na3V2O2x(PO4)2F3−2x framework. Moreover, these stoichiometry issues may have a profound impact on the electrochemical signature of these compositions, when used as positive electrodes in Na-ion batteries, involving the V3+/V4+ and/or the V4+/V5+ redox couples. Contrary to the two (ΔV = 0.5 V) or even three voltage domains previously reported upon Na+ extraction,34,35,37,38 we found indeed that Na+ extraction from stoichiometric Na3VIII2(PO4)2F3 toward Na1VIV2(PO4)2F3

proceeds reversibly through four oxidation potential steps, as displayed in Figure 1 and as already reported by some of us in ref 27. The relationship between these features and the crystal structure of Na3V2(PO4)2F3 will be discussed in the following. However, the precise mechanism associated with this complex electrochemistry is currently being investigated through in situ/ operando Synchrotron X-ray diffraction and is beyond the scope of this paper. We focus here on how the use of high-resolution techniques both in the field of electrochemistry (slow galvanostatic measurements) and diffraction (synchrotron X-rays, e-microscopy) reveals, for the first time, new important details about the behavior of stoichiometric Na3V2(PO4)2F3, which we consider as the material of choice for the future development of the Naion technology. The most striking finding concerns its crystal structure, being different from the one considered for the last 15 years.

2. EXPERIMENTAL SECTION The static molar magnetic susceptibility of Na3V2(PO4)2F3, (χ(T) = M(T)/H (H = 1 T), where H is the magnetic field and M is the magnetization) was measured at temperatures between 5 K and 300 K using a SQUID magnetometer (Quantum Design). The zero-fieldcooled χ values were obtained by cooling the sample in zero field down to 5 K and then heating it under the measuring field. The diamagnetic contributions were corrected using the atomic values from Bain and Berry,39 yielding the final paramagnetic susceptibility (χM). High-resolution scanning electron microscopy (SEM) analysis of metallized samples (Pd-deposited) was performed using a Hitachi Model S-4500 microscope. Diffraction data were collected using several instruments: laboratory X-ray diffraction (XRD) was performed on a PANalytical Empyrean diffractometer using Cu Kα1,2 radiation, with the powder being sealed in a 0.3-mm capillary used in Debye−Scherrer geometry. Synchrotron radiation data at room temperature were collected at the CRISTAL diffractometer of the SOLEIL synchrotron facility, at 0.6687 Å in the 2θ angular range of 5°−50°. The measurement was performed in Debye−Scherrer geometry, with the powder sealed in a 0.7-mm B



Article

mixture then was cast on an aluminum foil with a casting thickness of 25 μm and dried at 85 °C for 24 h. A 1 M solution of NaPF6 in PC (Aldrich) was used as electrolyte. Glass microfiber filters (Whatman) were used as separators, and metallic sodium was used as the negative electrode. During galvanostatic cycling, the electrochemical cell was charged and discharged at a current rate equivalent to C/50, meaning in the case of Na3V2(PO4)2F3 the extraction/insertion of each Na+ ion within 50 h.

3. RESULTS AND DISCUSSION (a). Synthesis of Stoichiometric Na3V2(PO4)2F3. The synthesis of Na3V2(PO4)2F3 was performed in two steps, following a procedure similar to the one patented by Barker et al.,42 where VPO4 was used as an intermediate phase. 1 V2O5 + (NH4)2 HPO4 + Csp 2 3 → H 2O + CO + c‐VPO4 + 2NH3 2

(1)

2c‐VPO4 + 3NaF → Na3V2(PO4 )2 F3

(2)

The presence of Csp in reaction 1 (in the form of highly divided soot) is necessary to allow for the carbothermal reduction of vanadium from 5+ to 3+. For reaction 1, the precursors V2O5, (NH4)2HPO4, and Csp were mixed in a planetary grinder Fritsch Mono Mill for 12 h at 400 rpm. The mixture was then pressed in a pellet and heated for 5 h at 300 °C in an alumina crucible to evaporate possible ammonia residues. For homogenization, the obtained product was milled again for 4 h at the same speed and a pellet was remade. This was finally heated at 800 °C for 5 h to allow for carbothermal reduction and the formation of carbon-coated VPO4 (c-VPO4). For reaction 2, c-VPO4 and NaF were mixed in a planetary grinder and then pressed in a pellet. This was placed in a gold tube (10 mm diameter) under argon. The tube was subsequently sealed and heated for 1 h at 800 °C. Note that the oven was systematically prewarmed and the chemicals were exposed to a steep temperature gradient. Finally, the tube was cooled at ambient temperature within ∼15 min. The phase was obtained almost pure, with small amounts of Na3VF6 and Na5P3O10 as impurities, subsequently eliminated by washing in water while stirring. The resulting Na3V2(PO4)2F3 powder is of high purity and composed of large aggregates (4−5 μm in size), formed by the agglomeration of smaller particles (100−300 nm in size) (see Figure 2). The precise chemical composition of Na3V2(PO4)2F3 was confirmed by means of an inductively coupled plasma− optical emission spectroscopy (ICP-OES) spectrometer system (Varian Model 720-ES) after complete dissolution of the powder into a hydrochloric acid (HCl) solution. Ratios of Na/ V = 1.55, Na/P = 1.60, and V/P = 1.03 were found, which agree rather well with the expected stoichiometry (1.50, 1.50, and 1.00, respectively). To further confirm the value found for the vanadium oxidation state, we performed magnetic susceptibility measurements. Figure 2 reports the resulting 1/ χM versus temperature plot, showing a typical behavior of the Curie−Weiss law for paramagnetic materials. A linear fit of the data in the 100−300 K region leads to a value for the Curie constant of C = 0.944(9), in remarkable agreement with the previous report of Le Meins et al.33 and with the approximate theoretical value of C = 1, obtained as χT = ((1/8)n(n + 2)), where n is the number of unpaired electrons (i.e., n = 2 for V3+,

Figure 1. Galvanostatic cycling of a Na3V2(PO4)2F3 // Na battery at C/50 rate per exchanged ion. (Top) First cycle and (bottom) inverse derivative curve (dV/dx)−1 for the first three cycles, showing the presence of four distinct electrochemical features. capillary and the acquisition done with a multianalyzer (21 Si(111) crystals). High-temperature measurements were performed at the MSPD beamline of the ALBA synchrotron facility.40 Data were collected in Debye−Scherrer geometry at 0.6200 Å in the 2θ angular range of 2°−50°, using the high-angular mode of the station (13channel Si(111) multianalyzer setup). The sample was enclosed in a 1mm-diameter borosilicate capillary and heated using a Cryostream Plus nitrogen blower from Oxford Cryosytems. Neutron diffraction data were acquired at the D2B high-resolution powder diffractometer of the Institut Laue-Langevin (Grenoble), at a wavelength of 1.594 Å, calibrated with a Na2Ca3Al2F14 reference. The powder was put in a vanadium cylindrical sample holder of the diameter of 6.5 mm and data collected in the 2θ angular range of 10°−160°. Diffraction data treatment and Rietveld refinement were performed using the FullProf Suite.41 Prior to the observation in electron microscopy, a suspension was obtained by grinding the material in hexane, and then depositing a droplet of this suspension on a Formvar carbon film supported on a copper grid. Electron diffraction experiments were carried out on many crystallites and very reproducible results were observed. The particles studied were chosen as isolated as possible. Electron diffraction patterns were obtained using a JEOL 2100 microscope system. The electrochemical signature of the Na3V2(PO4)2F3 active material was measured in Swagelok cells. The composite electrodes were prepared using a mixture composed of 75 wt % active material, 12.5 wt % Carbon Super P, and 12.5 wt % PVDF binder. The slurry of this C



Article

Figure 2. SQUID measurement of Na3V2(PO4)2F3 between 5 K and 300 K. Inset shows a SEM image of the metallized sample.

which has 4s03d2 electronic configuration). The antiferromagnetic behavior of the material is also described by a Néel temperature (θN) of 74 K. (b). Crystal Chemistry of Na3V2(PO4)2F3 at 298 K. The crystal structure of Na3V2(PO4)2F3 has been regarded as being well-established since its description in 1999 by Le Meins et al.,33 who completed a combined determination from singlecrystal and powder XRD. However, close inspection of the published atomic coordinates points toward a very high (meaningless) value of the thermal motion factor Beq on the Na(2) site (8.3 Å2), occupied at 50%. That work assigned the tetragonal space group P42/mnm to Na3V2(PO4)2F3, used until now to describe the structure (referenced in the following discussion as the standard structure) and displayed in Figure 3.

Figure 4. Top image shows Rietveld refinement of Na3V2(PO4)2F3 in the tetragonal space group P42/mnm from X-ray diffraction (XRD) on a laboratory diffractometer; bottom image gives Rietveld refinement of neutron diffraction data in the same space group, showing a good fit.

and Na(2) sites (both in 8i Wyckoff positions), respectively, and confirmed, accordingly, an abnormally high thermal motion parameter value for Na ions on the Na(2) site: 6.15 Å2. One notes that this site cannot be fully occupied, because two equivalent positions generated by the mirror m perpendicular to [110] are too close (1.8 Å). To gain further insights into the crystal structure of Na3V2(PO4)2F3 (in particular, in the Na ions distribution within the available crystallographic sites), we performed high angular resolution synchrotron radiation diffraction. Interestingly, we found that the indexation of the crystal structure of our Na3V2(PO4)2F3 powder sample in the tetragonal space group P42/mnm is not correct. Indeed, as Figure 5 shows, inhouse XRD generates diffraction peaks with rather large full width at half maximum values (FWHM > 0.1°) from which (h00)tetragonal and (0k0)tetragonal reflections, for example, seem to be merged at the same 2θ angle. For a very subtle orthorhombic distortion, which appears to be the case here, high angular resolution (FWHM ≈ 0.005° in our case for 2θ in the 0°−25° angular range) synchrotron radiation diffraction was essential for allowing us to refine the unit-cell parameters of Na3V2(PO4)2F3. We chose the low-symmetry Pmmm space group to begin with, finding from Le Bail fit cell parameters values of a = 9.02837(3) Å, b = 9.04438(3) Å, c = 10.74643(3) Å, i.e., a b/a ratio of ∼1.002 (see Figure 6). Such distortion had never been observed before because of a 2-fold reason. On one side, the synthesis conditions may surely play a role, even if it appears to be a “general” property for the series of samples we have studied and obtained through the two-step solid-state reaction described above. For instance, whatever the cooling rate (quenching in air, cooling at a

Figure 3. Crystal structure of Na3V2(PO4)2F3 in the tetragonal space group P42/mnm (ICSD No. 88808). Bioctahedral V2O8F3 (green) and tetrahedral PO4 (yellow) building blocks are showed.

In Na3V2(PO4)2F3, vanadium is placed in the center of VO4F2 octahedra, forming V2O8F3 bioctahedral units, which are alternately bridged by PO4 tetrahedra. This results in a stable three-dimensional (3D) framework with large tunnels in the [110] and [11̅0] directions, which grant Na ions significant mobility. This property confers to Na3V2(PO4)2F3 the ability to intercalate/deintercalate sodium, making it a positive electrode material of choice for Na-ion batteries.27 The structure refinements of both neutrons and laboratory XRD data, using the published structure of Na3V2(PO4)2F3 (space group P42/mnm33) as a starting model are plotted in Figure 4. In the light of this model, we fixed the occupancy factors at values of 1 and 1/2 on the two crystallographic Na(1) D



Article

(such as an image plate detector, for example) might not result in sufficient angular resolution. Having found the cell parameters, a suitable space group was sought, since it was easily observed that Pmmm does not take into account many of the systematic extinctions. First, we observed that the obtained cell parameters are the only possible choice, i.e., none of the other possibilities reported in the literature for similar compounds could fit the data. Indeed, cell parameters that were bigger by a factor of √2 would shift the (hh0) reflections, which are not split in the orthorhombic group, in the angular position of the (h00)−(0k0), where a splitting is needed to fit the data. Also, a cell smaller by a factor √2 would be incorrect, since the first peak in angular range (at 5.54°) could not be taken into account. Therefore, we chose a combined approach using indexing softwares (Dicvol, MacMaille, Fox), electron diffraction, and crystallographic considerations to select suitable space groups. We focused on nonprimitive cells, since electron diffraction showed the existence of systematic extinctions in the entire reciprocal space. In particular, as shown in Figure 7a, with our choice of cell parameters (and, therefore, orthorhombic setting) the only possible centering found from indexation is A centering. It can be observed that general (hkl) peaks respected k + l = 2n. Moreover, an examination of systematic extinctions in the synchrotron radiation pattern revealed the following conditions for reflections to exist: for (h00), (0k0), and (00l), h = 2n, k = 2n, and l = 2n; for (0kl), k + l = 2n; for (h0l) and (hk0), conditions were less clear, because of the vicinity of the peaks, but we finally converged on h,l = 2n and k = 2n, respectively. Thanks to these considerations, the structure was subsequently refined for the best groups, namely, Ama2 (No. 40), Amam (No. 63), and A21am (No. 36). The first is in its

Figure 5. Comparison between laboratory (blue) and synchrotron radiation (red) XRD data of Na3V2(PO4)2F3, showing how the high resolution of synchrotron data allows one to resolve the orthorhombic splitting.

controlled temperature rate (

Na - American Chemical Society

Recommend Documents