Attribution-NonCommercial-NoDerivatives 4.0 International license Effects of fluorination on the structure, magnetic and electrochemical properties of the P2- type NaxCoO2 powder

The main goal of this research has been to investigate for the first time the effects of fluorination on the crystal structure, magnetic, and electrochemical properties of the P2-type NaxCoO2 powder. Sodium cobalt oxide with a P2-type structure is synthesized by a modified solid-state reaction consisting of alternating processes of rapid heating up to 750 oC and rapid cooling to the room temperature. The obtained powder is fluorinated using a gas-solid reaction with NH4HF2 as fluorinating agent. Fluorination causes a decrease of sodium content in the parent phase with the concurrent formation of the minor phases of Na2CO3 and NaF. The structure of NaxCoO2 in both powders is refined in P63/mmc space group. The results of the Rietveld refinement combined with the findings from the XPS measurements confirm the Na0.76CoO2 and Na0.44CoO1.96F0.04 stoichiometries for the pristine and fluorinated powders, respectively, which indicates that 4 at.% of fluorine ions per formula unit are incorporated in the structure. Preliminary electrochemical investigations have revealed an improved charge/discharge performance. The influence of fluorination on morphology and magnetic properties has also been examined.


Introduction
The rising demands for energy storage systems increase the interest in sodium-ion batteries as an alternative to lithium-ion batteries [1]. Layered sodium transition metal oxides (Na x T M O 2 , T M has one or more transition metal cations) have been extensively studied as cathode materials for sodium-ion batteries [2]. Na x CoO 2 is the most attractive among them due to its characteristics: an unusually large thermopower [3,4], interesting magnetic properties [5], and its hydrated compound shows superconductivity below 5 K [6]. Na x CoO 2 can crystallize in several layered structures depending on sodium content, oxygen partial pressure and temperature [7]. In each of them, the edge-sharing CoO 6 octahedra form (CoO 2 ) n sheets, between which sodium ions are inserted with octahedral or prismatic surroundings [8]. The packing also differs in the number of sheets (2 or 3 sheets) within the unit cell, i.e. AB CA BC, AB BA, and AB BC CA oxygen packing, revealing three possible structural types: O3, P2, and P3, respectively [9]. The P2 type of structure is considered to be the most suitable structure for the electrochemical application, as an energy storage material. Within the P2 structure, the Na + ions are intercalated in a trigonal prismatic environment between adjacent (CoO 2 ) n sheets, taking one of two possible positions denoted as Na (1) and Na(2) (Figure 1).
Cobalt ions lie above and below Na(1) sites. Na (1) shares faces only with two CoO 6 octahedra of the adjacent sheets, whereas Na(2) shares edges with the six surrounding CoO 6 octahedra [10].
Since the distance between the adjacent Na(1) and Na(2) sites is two times smaller than the ionic radius of Na, simultaneous occupancy of the nearest-neighboring sites is not possible. From an electrostatic point of view, a Na(1) site is expected to be less stable than a Na(2) site.
Stoichiometric reasons do not require the presence of sodium ions at a Na(1) site; however, the simultaneous occupation of both sites leads to stable configurations through the minimization of the in-plane Na + -Na + electrostatic repulsion [11]. The magnetic properties of Na x CoO 2 phases strongly depend on the sodium content (x). The composition with x~ 2/3 follows the Curie-Weiss law; with the further decrease in x, Na x CoO 2, changes from a "Curie-Weiss metal" to a "paramagnetic metal" through an insulating charge-ordered state about x=0.5 [12]. An unusual low-temperature (22 K) magnetic phase transition is observed in the composition with x = 0.75 [13].
The galvanostatic electrochemical curve of the P2 type Na x CoO 2 has a complex profile: it consists of several voltage plateaus, which imply the existence of biphasic domains, potential drops that are related to stable single-phase domains, and the sloping parts that are typical of solid-solution behaviour [10,14]. This is a consequence of the Na-vacancy ordering and the resulting structural response of the CoO 2 layers to changes in the electron count and Na ion distribution [15].
Numerous studies on cation doping seek to improve the performance of sodium transition metal oxides with a P2-type structure [16][17][18][19][20][21]. Fluorine doping has been demonstrated to enhance electrochemical properties of various cathode materials for lithium-ion batteries [22][23][24][25][26][27]. On the other hand, to the best of our knowledge, there are no reports on anion doping of the P2-type Na x CoO 2 . The main goal of this research has been to investigate for the first time the possibility of fluorine doping of the P2-type Na x CoO 2 powder and to examine the effects that fluorination has on its crystal structure, magnetic, and electrochemical properties.
2. Experimental 2.1. Synthesis of a pristine Na x CoO 2 A solid state reaction was used for the preparation of the pristine Na x CoO 2 powder. Na 2 CO 3 and Co 3 O 4 were mixed in the molar ratio 1.7 : 1 (the targeted stoichiometry was 0.75 sodium per cobalt, but due to sodium volatility, Na 2 CO 3 was added in excess) and thoroughly grounded in a mortar. The precursor powder was not pelletized; it was placed in the powder form in a platinum crucible and then subject to a high-temperature treatment. As opposed to the conventional solidstate method, which involves long-term heating, our research relied on the method of sequential cycles of rapid heating and cooling, with an intermediate grinding stage between two cycles. In one cycle, the precursor powder was rapidly heated to 750 ºC in air with a dwell time of 15 minutes and then air-quenched to room temperature. This was done to reduce both the preferential orientation during the particle growth and sodium volatilization. Fourier transform infrared (FTIR) spectra and XRD patterns were collected after each cycle to monitor both the stage of the solid-state reaction and the potential presence of impurities. Namely, after the initial grinding of the reactants, the long-range order in Na 2 CO 3 broke and it became undetectable by XRD. However, a short-range order was preserved, due to which Na 2 CO 3 could easily be detected by FTIR spectroscopy. The final powder that was subsequently fluorinated, denoted as NCO, was obtained after 12 cycles, which corresponded to 3 hours of the high-temperature treatment. The guiding criterion was to obtain the powder containing no or a minimal amount of impurity phases (Co 3 O 4 and Na 2 CO 3 ). Interestingly, an XRD analysis of the powder obtained after seven 15-minute intervals of thermal treatment showed phase purity, but FTIR revealed the remains of the Na 2 CO 3 phase. Due to this, the alternating processes of heating and cooling had been prolonged until FTIR confirmed the absence of the Na 2 CO 3 phase. The evolution of FTIR spectra during synthesis is presented in Supporting Information, Figure S1.

Synthesis of F-doped Na x CoO 2
Several approaches were used in the synthesis of F-doped Na x CoO 2 . The first unsuccessful attempt was a solid-state reaction of stoichiometric amounts of Na 2 CO 3 , Co 3 O 4 , and NaF at 900 ºC for 20 hours in air. Beside the Na x CoO 2 phase, NaF and Co 3 O 4 were still present. By rising the temperature to 950 ºC, Na x CoO 2 started to decompose and at 1000 ºC both Na x CoO 2 and NaF phases vanished, while CoO emerged (Supporting Information, Figure S2). Similar results were obtained by varying the precursors. A solid-state reaction between NH 4 CoF 3 and Na 2 CO 3 at 800 ºC in air resulted in the formation of NaF, Co 3 O 4 and CoO phases.
Another approach was also tested: the fluorination of the previously synthesized Na x CoO 2 with NH 4 HF 2 as a fluorination agent -a method standardly used for the fluorination of Sr 2 CuO 3 [28] and rare-earth oxides [29,30]. Na x CoO 2 and NH 4 HF 2 were thoroughly mixed and then introduced to different temperatures (from 450 -800 ºC), but NaF was always allocated as a separate phase.
However, when the fluorination was conducted at 200 °C, in a vacuum evacuated atmosphere, a routine XRD check revealed the absence of the NaF phase, as well as a significant shift of diffraction peaks. Therefore, this method was adopted for the fluorination process. The requred amounts of Na x CoO 2 and NH 4 HF 2 were placed in two separated teflon-lined containers, so as to avoid contact of the reagents either in the solid or in the liquid form, and to enable only the vapor-solid reaction. After vacuuming the system, the temperature was raised to 200 °C with a dwell time of two hours, and before spontaneous cooling, it was vacuumed again to prevent the contamination of the powder with the remains of NH 4 HF 2 and NH 4 F vapors. The obtained powder will hereinafter be referred to as the NCOF sample.

Materials characterization
X-ray powder diffraction measurements were used for both the phase purity check and the structural analysis of the synthesized samples. The diffraction data were collected on a Philips PW 1050 diffractometer with Cu-Kα 1,2 radiation (Ni filter) at room temperature. The measurements were done in a 2θ range of 10-70° with a scanning step width of 0.05° and 3 s time per step, for routine phase checking; for the crystal structure refinement, the 2θ range was 10-110° with a scanning step width of 0.02° and 14 s time per step. The crystal structure refinement was based on the Rietveld full profile method [31] using the Koalariet computing program, based on the fundamental parameters convolution approach in diffraction-line-profile fitting [32].
The FTIR spectra of the samples were recorded in ambient conditions in the mid-IR region (400- The electrochemical measurements were carried out in a closed, argon-filled two-electrode cell at room temperature, with metallic sodium as a counter electrode. A 1M-solution of NaClO 4 in propylene carbonate (PC, Aldrich) was used as an electrolyte. NaClO 4 was obtained by the vacuum drying of NaClO 4 ·H 2 O (p.a., J. T. Baker) at 170 ºC. The working electrodes were prepared by mixing the synthesized material, carbon black and polyvinylidene fluoride (PVdF, Aldrich) in the weight ratio 75:15:10, respectively, with N-methyl-2-pyrrolidone to form a slurry that was afterward deposited on platinum foils. Galvanostatic charge/discharge tests were performed within the potential range of 2 -3.8 V at C/10 current rates by using Vertex.One (Ivium Technologies) potentiostat.

Morphology study
The morphologies of the synthesized powders, determined by FESEM, were basically similar, consisted of strongly agglomerated polyhedral particles (Figures 2a and 3a). The particles of the fluorinated powder were smaller in size than the particles of the pristine powder. In some zones of both samples, larger particles with a laminar structure could be observed, as well (Figures 2b   and 3b). Also, in the fluorinated powder rod-like particles could be observed ( Figure 3) which probably originated from the Na 2 CO 3 impurity phase revealed by FTIR.

XRD analysis
Routine X-ray powder diffraction measurements for phase identification confirmed only the presence of a P2-type Na x CoO 2 phase in both powders. However, there were notable differences in X-ray diffractograms: the shift of the diffraction maxima, the change of the relative peak intensity ratio, and the peak broadening observed for the fluorinated sample. These could be ascribed to the sodium deficiency and/or to the incorporation of fluorine ions. The fluorinated powder also contained the Na 2 CO 3 impurity phase, revealed only by the FTIR analysis. The results of the Rietveld refinement are given in Figure 4 and Tables 1 and 2. The obtained lattice parameters for pristine Na x CoO 2 were in good agreement with the literature data [33].
Fluorination caused a small decrease of the lattice parameter a and an increase of the lattice parameter c (Table 1). This trend of changes in lattice parameters is common during sodium deintercalation [10,15] as decreased sodium concentration increases the spacing between adjacent oxygen layers. In addition, the NCOF powder had smaller crystallites, with an increased NaO 2 interslab distance and a decreased CoO 2 slab thickness (Table 1) Table 2) correspond to the compression of the CoO 2 layers along the c-axis which usually gives wider bands, smaller effective mass and enhanced electron conductivity [15].
The refinement of the occupancies of both sodium sites revealed that the Na/Co ratio was of 0.76 and 0.44 for the pristine and fluorinated powders, respectively. Apparently, during fluorination, a part of sodium ions migrated from the lattice and formed both a distinct NaF phase (in the amount of 2.8 wt.%) and a Na 2 CO 3 phase (revealed only by FTIR). The obtained lattice parameters for the NCOF sample deviated from the literature data for sodium concentrations around 0.44 [15,34]. In addition, the interslab distance, taken as a measure of the sodium content, also implied a higher sodium content [35]. As the Na(1) site shared faces with the octahedron around Co cations, the Na(1) site had a larger site energy than the Na(2) site due to the direct repulsion between Na and the adjacent Co ions [33]. The smaller site energy of the Na(2) site causes the preferred occupancy of the Na(2) site over the Na(1) site, as observed in this investigation. This difference in site energies was expressed in the Na(1)/Na(2) ratio and it was a strong function of total Na content. The Na(1)/Na(2) ratio for the NCO sample (0.48) was in accord with the literature data for the given sodium concentration [15], while its value for the NCOF sample (0.63) corresponded to a sodium content higher than the refined value [36], which was equivalent to a smaller average oxidation state of cobalt. Apart from the extraction of sodium ions from the lattice, this further implies that fluorination additionally modified the average oxidation state of cobalt through tiny fluorine incorporation in the lattice, as revealed by structural refinement (Table 1).

XPS analysis
More information on the cobalt oxidation state was provided by XPS experiments. The survey photoelectron spectra of both powders show the presence of the Co 2p, O 1s, and Na 1s peaks, as well as the presence of an F 1s peak in the fluorinated powder ( Figure 5). High-resolution scans of constituent elements (Co, O, and Na) are compared in Figure 6. The Co 2p spectra of both powders consist of two major lines, Co 2p 3/2 and Co 2p 1/2 from spin-orbital splitting, and the satellite shake-up lines ( Table 3). The main doublet of the pristine powder (779.8 -794.9 eV) is assigned to Co 3+ ions and it is well-documented in the literature [37,38]. The binding energy of the second doublet (781.1 -796.4 eV) is ascribed to Co 4+ ions, which is less commonly presented in the literature [39,40]. The presence of Co 2+ can be excluded [41]. The deconvoluted XPS spectrum of the fluorinated powder contains components with almost the same binding energies as the pristine powder (Table 3) (Table 1): taking into account the multiplicity of the oxygen site, the fluorine content per formula unit obtained by the refinement is also 0.04. Additional broad lines in the Co 2p spectra are shake-up lines that come from the photoelectrons of excited ions. They show that both ions (Co 3+ and Co 4+ ) are in a paramagnetic state [38]. The binding energies of the O 1s signals are listed in Table 3. Three main components come from oxygen ions and the remaining lines of larger widths are sodium Auger lines [42] and O 1s shake-up satellite line [43]. The component at 529.4 eV is attributed to the O 1s lattice oxide, while the most intense component at 531.7 eV can be assigned to a defective oxide or hydrated oxide [44]. The third component at around 533.4 eV is assigned to the physisorbed water [45]. As we have already mentioned, the powders are hygroscopic and sensitive to moisture from air. The analyses of the Na 1s peaks of the pristine and the fluorinated powders indicate the existence of three and five components, respectively, but a reliable interpretation and assignments to different sodium chemical states is not straightforward. However, significant peak broadening observed for the fluorinated powder inticates a more disordered and nonuniform structure.

Magnetic properties
Generally speaking, the total susceptibility χ(T) of the investigated system can be expressed as: (1) where χ P , Χ dia , Χ VV represent Pauli paramagnetic, diamagnetic, and Van Vleck temperature independent susceptibilities, respectively. Term χ d (T) denotes temperature-dependent Curie-Weiss paramagnetism that originates from the localized d electrons. Therefore, we have fitted our experimental data in the high temperature region 150-300 K to the following expression: (2) where C represents the Curie constant, θ is the Curie-Weiss paramagnetic temperature, and includes all temperature-independent contributions. Figure 7 shows the temperature dependence of the inverse susceptibility 1/( -, while the fitting parameters (C, θ, and ), as well as the calculated effective magnetic moments μ eff are given in Table 4. This dependence of the inverse susceptibility is linear at higher temperatures, and bellow approximately 100 K it starts to decrease more rapidly, while the magnetic phase transition temperature is observed around 25 K.
The negative Curie-Weiss paramagnetic temperatures indicate antiferromagnetically interacting spins, while the downturn of the inverse susceptibility suggests the magnetic ordering transition of a ferromagnetic type. Namely, a ferromagnetic component can be observed due to small deviations of the spins from the completely antiparallel alignment [46]. The obtained effective moments per cobalt ion are 1.13 μ B and 1.04 μ B, for the NCO and the NCOF samples, respectively. These values are in good agreement with previous findings for the Na x CoO 2 (0.7 ≤ x ≤ 0.8) compounds [47][48][49]. Since Co 3+ ions are in a low spin state (S=0), the obtained values of effective magnetic moments can be considered to be a contribution of Co 4+ spins only, and the recalculated values of effective magnetic moments expressed per Co 4+ ion are given in Table 4.
It can be observed that the value for the NCO sample is above the theoretical value for low spin Co 4+ ion (1.73 μ B ), while for the NCOF sample it is below this value. Similar behaviour has been observed in three-layered Li x CoO 2 [50], Na x CoO 2 [49,50], and Li/Na alternate-layered Li x Na y CoO 2 [46], and is attributed to the changed localization of the unpaired spins [50].
Namely, by decreasing the sodium content from the Na x CoO 2 phase (by electrochemical and chemical deintercalation or, as in our case, by fluorination), the spin system is modified from the one that can be described by localized spins to the one that can be described by delocalized spins, which leads to a decreased magnetic moment per cobalt ion [49].

Electrochemical measurements
Preliminary electrochemical measurements were done to determine the differences between two powders. The results should be taken with some reservation, only as an indication of the general behaviour, since our electrochemical test cell was not fully optimized. Figure 8 presents the galvanostatic charge-discharge curves of the second cycle at C/10 current rate in the potential range from 2 -3.8 V. It is obvious that the fluorination of the pristine powder leads to the improvement of its electrochemical performance in terms of a higher specific capacity and lower voltage hysteresis. This is certainly the consequence of smaller crystallite size and the CoO 2 layer compression (Table 1), the latter causing weaker Na-O bonding ( Table 2) and enhanced electronic conductivity. These preliminary electrochemical measurements have confirmed the predictions based on the first-principle calculations that sodium conductivity in the P2 structure increases monotonically as sodium content x decreases [33]. The described method of fluorination can be used to synthesize a series of Na x CoO 2 powders with different sodium contents and hence different ionic conductivity. This is significant, since P2 phases with a smaller x can be obtained only through electrochemical or chemical deintercalation [51], which are more demanding and time-consuming methods.

Conclusion
Phase-pure Na x CoO 2 with a P2-type structure was synthesized by a modified solid-state reaction.
Fluorination was performed via a gas-solid reaction with NH 4 HF 2 as a fluorinating agent. The fluorinated powder retained the P2 structure, though with a lower content of sodium.