Enabling the Electrochemical Performance of Maricite-NaMnPO 4 and Maricite-NaFePO 4 Cathode Materials in Sodium-Ion Batteries

,


Introduction
In the past few years, sodium-ion batteries (SIBs) appear as an alternative due to their low cost and abundance of the raw materials, which make SIBs extremely suitable for large-scale stationary applications [1,2] by having a similar working mechanism as known from lithium-ion batteries [3]. Despite the advantages, the redox potential and ionic radius of sodium ions make intercalation/deintercalation more difcult than for lithium ions [4,5]. Tese factors result in lower energy density, lower electrochemical stability, and shorter battery life. Among the transition metals, Fe and Mn are mostly used as the transition metal element for the polyanions because of their environmental friendliness and their low costs [6]. In the olivine structure of NaFePO 4 , there are three possible ways for the sodium ion migration and migration energies of sodium ions along diferent paths are shown in literature [7] and only a 1D channel for the sodium ion migration exists, which makes NaFePO 4 intrinsically low ionic conductive. In comparison with Schottky and Frenkel defects, the formation of antisite defects [8] is much more energy-efcient, which means the antisite defects are more favorable for migration. One method to improve the electrochemical properties is to reduce the particle size, which reduces the difusion path length and increases the contact area with the carbon black. Drezen et al. [6] have systematically studied the efect of the particle size on the olivine structure for LiMnPO 4 . However, Kim et al. [9] reported an unexpected fnding in maricite NaFePO 4 in 2015. According to their study, maricite NaFePO 4 exhibits more than 90% of its theoretical capacity and outstanding cyclability. Various polyanion cathode materials for SIBs have been reported, such as NaFePO 4 , Na 2 FePO 4 F, Na 2 FeP 2 O 7 , and Na 4 Fe(PO 4 ) 2 P 2 O 7 [10][11][12]. However, NaMnPO 4 is seldom reported as the active material for battery application, due to its low achievable capacity. Nevertheless [13], the theoretical capacity of NaMnPO 4 and its crystal structure is similar to NaFePO 4 (154 mAh.g −1 ) [14]; therefore, it is still worth investigating NaMnPO 4 and NaFePO 4 materials.
Te focus of this work is to enable m-NaMnPO 4 and m-NaFePO4 cathode materials and improve their electrochemical properties by means of reducing particle size accompanied by carbon coating as reported in the literature [15,16] for other cathode materials. Terefore, m-NaMnPO 4 and m-NaFePO4 powders were milled to fne particle sizes followed by carbon coating by utilizing diferent carbon sources using a thermal/mechanical approach to enhance the electrochemical performance.

Experimental Methods
In order to determine the elemental composition of the active cathode materials, the ICP-OES (OPTIMA 4300 DV Perkin Elmer) technique was used, and the oxygen content was analyzed with the method of carrier gas hot extraction (CGHE) by the oxygen/nitrogen analyzer TC600 (LECO).
Te laser scattering (Horiba Partica LA-950) method was used to determine the particle size distribution of commercial m-NaMnPO 4 and m-NaFePO 4 powder materials from the NEI corporation.
Tese materials were also studied by powder X-ray difraction through a SEIFERT X-ray difractometer, and Cu Kα (λ �1.54056 Ǻ) was used as the X-ray source. Te powder materials were fxed on the ω-axis. Te scanning angle was from 10°to 70°, with a step size increment of 0.01°.
Te m-NaMnPO 4 and m-NaFePO 4 powder materials were reduced to smaller particle sizes by a ball mill (Pulverisette 7 premium, Fritsch) at 1000-1200 rpm for 12 h with an interval of 30 min with a rest time of 2 h. To minimize the contamination during the ball milling process, a nylon jar and high purity yttrium stabilized ZrO 2 balls (1 mm diameter) were used to grind the pristine m-NaMnPO 4 and m-NaFePO 4 (active material : balls � 1 : 80 by weight). In this work, the wet (isopropanol) grinding method was chosen to achieve a better grinding efect. On the one hand, the liquid can reduce the surface tension of small particles, especially nanoparticles, and on the other hand, the liquid can slow down the temperature rise and provide safety. Te ground materials were subsequently vacuum-dried (10 −3 bar) at 70°C for 12 h. Glucose (10 wt.%) was dissolved in distilled water, and then, active material m-NaFePO 4 was added to this solution and further mixed uniformly with a speed mixer (DC 150) followed by calcination at 600°C under an Ar atmosphere.
To coat the active material milled NaMnPO 4 with carbon black, a mechanochemical process was employed. At frst, the carbon black (20 wt.%) was mixed with active material 2 : 8 by weight and milled for 3 hours at 1200 rpm with an interval of 30 min accompanied by drying at 400°C with a holding time of 4 hours. Finally, the carbon-coated m-NaMnPO 4 and m-NaFePO 4 materials were analyzed by scanning electron microscopy (SEM) (Supra 55, Zeiss) to confrm that the coating process was successful.
Since it was necessary to coat the cathode materials with carbon up to a certain temperature under Ar atmosphere, the determination of the phase transformation during the heating of the pristine materials was essential. A diferential scanning calorimeter (NETZSCH DSC 404) was utilized in this investigation using an Al 2 O 3 crucible under an Ar gas fow of 100 ml/min. In the case of NaFePO 4 , a heating rate of 5 K/min temperature was used from room temperature to 600°C, and for NaMnPO 4 , the heating rate of 10 K/min was applied up to 900°C.
A NETZSCH TG 449 F1 Jupiter thermal nanobalance coupled with the NETZSCH QMS 403C mass spectrometer (MS) was used to investigate the thermal stability of the cathode materials with respect to the phase transformation as indicated in the diferential scanning calorimeter (DSC) measurement. Te fused silica transfer lines (75 μm diameter) leading to the MS were heated to 200°C to ensure that all the samples entered the MS in a gaseous state, i.e., to avoid condensation losses. An alumina sample holder was used for the measurements.
Te respective slurries were prepared by dissolving 10 wt.%. polyvinylidene fuoride (PVDF) binder in a N-methyl-2-pyrrolidone (NMP) organic solvent using 80 wt.% active cathode materials (including carbon coating) and 10 wt.% carbon black (C-NERGY Super C65). Te slurries were coated on an Al foil (19 μm) and were dried at 70°C for 1 day under vacuum (10 −3 bar). Dried cathode sheets were calendered by 10-15% using a calendering machine (Hot Rolling Press with Variable Speed, MTI) followed by subsequent drying at 120°C under vacuum (10 −3 bar) for 1 day. Calendered sheets were punched into 13 mm diameter discs. In order to study the electrochemical properties of pristine and milled carbon-coated m-NaMnPO 4 and m-NaFePO 4 materials, coin cells (CR-2032) were assembled against pure Na metal as the anode material with 1 M NaClO 4 (EC : DMC 1 : 1 vol % + 5 vol % FEC) electrolyte of 100 μl and glass fber separator (GF/A from Whatman) in an Ar-flled glovebox (H 2 O < 0.1 ppm, O 2 < 0.1 ppm). Te electrochemical data were recorded with a BioLogic instrument (MPG2). Te coin cells were charged and discharged at a C/10 rate during a constant current (CC) phase in the voltage range of 1.0-4.5 V (m-NaMnPO 4 ) and 1.5-4.5 V (m-NaFePO 4 ).

Results and Discussion
With respect to the elemental analysis results, the stoichiometric ratios of Na, Mn, Fe, P, and O are shown in Table 1, and commercial cathode materials (pristine) stoichiometry is confrmed as m-NaMnPO 4 and m-NaFePO 4 .
Termal carbon coating of m-NaFePO 4 is performed under an argon atmosphere. However, it is observed that the color of the sample powder changes after 12 h of milling before and after calcination, i.e., a chemical reaction occurred during the high-temperature heating process of the sample in the argon atmosphere. In addition, the calcination process further increases the oxygen content, and the stoichiometric ratio of oxygen content increases to 4.57 (compared to the milled m-NaFePO 4 powder oxygen content of 4.29). It is suggested that oxidation occurs during the high-temperature heat treatment, as the carbon source (glucose) is dissolved in water to have a homogenous mixture before calcination. Such an increase in oxygen content occurs and relates to the oxidation of iron (II) to iron (III) beyond 520°C. By comparison with the XRD difractogram, it is evident that calcination causes recrystallization and the newly formed crystals contain a presumably olivine/oxidized-NaFePO 4 @C structure (see Figure 1). Terefore, by means of elemental chemical analysis, the substances before and after calcination are measured again to verify the composition of the product.

NaMnPO 4 Material.
Te morphology of pristine m-NaMnPO 4 is made up of rod-like aggregates and the particle size is from several micrometers to one hundred micrometers, which makes the pristine m-NaMnPO 4 not suitable for battery applications. Figure 2 shows the particle size distribution of pristine m-NaMnPO 4 (Figure 2(A)) and milled NaMnPO 4 (d 50 � 400 nm) (Figure 2(C)) in accordance with SEM images, respectively, in Figures 2(B) and (D) (milled + coated/covered with carbon). Te particle size of NaMnPO 4 is reduced to d 50 � 400 nm during ball milling and the result shows the bimodal distribution, indicating (Figure 2(C)) that any further extension of the milling time has little impact on the particle size. Additionally, Figure 2(D) shows the SEM images of milled NaMnPO 4 @C partially covered carbon particles and the change in the morphology of particles into a sponge-like structure. It is most probable that the sponge-like structure is due to the C coating. At the same time, it can also be seen that ball milling with 1 mm ZrO 2 grinding beads cannot reduce the particle size to the level of 50-100 nm, which is the required size reported in the literature [17]. Te diferent grinding balls (2 mm and 1 mm) were used to reduce the particle size, and particle distribution is shown in the supplementary material (see Figure S1). With smaller grinding balls of 1 mm ZrO 2, the 400 nm (d 50 ) particles were achieved.
All peaks that are observed in the XRD pattern of the pristine m-NaMnPO 4 (see Figure 3)match with the maricite structure as reported for m-NaMnPO 4 [18]. After milling and carbon covering, the intensity of the XRD peaks increases and indicates improved crystallization and crystallite growth. Teoretically, the m-NaMnPO 4 could transfer into natrophilite NaMnPO 4 and amorphous NaMnPO 4 [9]. Terefore, the newly formed peaks are compared with the standard natrophilite NaMnPO4 XRD pattern and are found to be in accordance with it. Tese peaks are highlighted by the red line. Tese peaks clearly demonstrate the increase of the natrophilite NaMnPO 4 content [9]. Te SEM image of milled 40 nm NaMnPO 4 without carbon coating for better comparison is listed in the supplementary material (see Figure S2).
In Figure 4, the thermal analysis of m-NaMnPO 4 is presented. According to this result, two endothermic peaks during the heating and one exothermic peak during the cooling steps are observed. Te occurrence of the frst endothermic peak is due to the decomposition/transformation of m-NaMnPO 4 into new phases as impurity [19]. Te second endothermic peak shows that m-NaMnPO 4 melts at 839.7°C, which is reversible, and an exothermic peak is clearly observed during the cooling phase at 810°C (crystallization/solidifcation). According to the result, there is no change in the material up to 700°C, which demonstrates that there is no phase transition/decomposition. Hence, the thermal procedure with carbon black at 400°C is a safe choice. Furthermore, the TGA-MS result of m-NaMnPO 4 (see Figure S3 (Supplementary Material)) shows material thermal stability up to 900°C (at 727°C, it is nonvolatile).
In Figure 5, the comparison of the achievable capacities of m-NaMnPO 4 with pristine and fne carbon-coated particles is shown. For NaMnPO 4 @C, the maximum capacity is 47 mAh.g −1 in the third cycle, which is signifcantly higher than that of the pristine material. Tis indicates that particle size and carbon coating play an important role in enhancing capacity. Te cyclability of pristine and milled/coated at diferent particle sizes is shown in the supplementary material (see Figure S4) and a comparison was made. However, it is worth mentioning that the achievable capacity of NaMnPO 4 @C (400 nm) could be expected to be even higher. A reasonable explanation for not having a higher specifc capacity of this fne partially coated/covered material could be that the nanoparticles exhibit high surface energy and tend to form agglomerations. However, the shear force of the speed mixer is not strong enough to break these agglomerations. Terefore, the active material could not be dispersed homogeneously in the slurry. EDX analysis in supplementary Figure S7 confrms the elemental distribution of NaMnPO 4 with carbon contents. During charging, the Mn 2+ intercalates with the sodium atoms and Mn 2+ oxidizes to Mn 3+ , and upon discharge, the sodium intercalates, reducing Mn 3+ to Mn 2+ . Figure 6, the particle size distribution and SEM images of the pristine (Figure 6(A)/(B)), 12 h milled (Figure 6(C)/(D)), and milled-coated NaFePO 4 ( Figure 6(E)/(F)) are shown. During ball milling, the particle size is reduced to d 50 � 1.57 μm and is evenly distributed. By coating with glucose and after calcination, it is observed that the particle size becomes larger (about d 50 � 3.5 μm). Tis is because of agglomeration during the mixing of active   In the difractogram of pristine m-NaFePO 4 , shown in Figure 1, all the refections are in agreement with the maricite structure from JCPDS fle #04-012-9665 [20]. After grinding and dissolution in a glucose solution, the m-NaFePO 4 powder is calcined and recrystallized. It is obvious that new refections appear (highlighted in red). Tese refections can be partially assigned to the olivine structure. It can be assumed that calcination up to 600°C causes a phase transformation from m-NaFePO 4 to olivine NaFePO 4 , which can be observed in DSC thermal analysis as well, as discussed later in the article. To verify whether the ball milling process causes amorphization or the subsequent high-temperature heat treatment causes recrystallization, additional comparative tests are performed. Te 12 h ground powder was taken out and heat-treated separately (without glucose), i.e., without carbon coating. It is found that the results are consistent with those in the carbon-coated condition, and this fnding is also consistent with the results reported by Hwang et al. [14]. Tey indicated that ball-milled m-NaFePO 4 partly contains an amorphous phase and longer ball-milling (24 h) did not result in complete amorphization. Terefore, it is concluded that high-temperature heat treatment up to 600°C can recrystallize the amorphous NaFePO 4 . Afterward, the possible phase changes between room temperature and 600°C were investigated, because the coating is thermally processed at 600°C under an argon atmosphere. No phase change was found in pristine m-NaFePO 4 (Figure 7). Surprisingly, a large exothermic peak is observed at 390°C in the case of milled m-NaFePO 4 (without calcination), which means that the ball-milled powder undergoes a signifcant phase change at 390°C. Te newly formed phase is presumably olivine NaFePO 4 / oxidized-NaFePO 4 phases, and subsequently, carbon coating is made by the method used for LiFePO 4 [21]. After the temperature treatment at 600°C, the milled powder partly changed its color from black to orange. Comparing the X-ray difractograms in Figure 1, it can be concluded that the ball milling process amorphized m-NaFePO 4, and then, the calcination process recrystallized it, but the crystal structure is no longer pure maricite afterward. On the one hand, this newly formed substance could be oxidized-   International Journal of Electrochemistry NaFePO 4 /FePO 4 (as oxygen content increased, see Table 1), as transforming the maricite phase into electrochemically active amorphous FePO 4 is reported in literature [22,23] and this phase exhibits a reversible capacity of 142 mAh.g −1 at room temperature [9]; on the other hand, it could consist of a mixture of olivine NaFePO 4 and impurity compounds with Na, Fe, P, O (such as Xenophyllite Na 4 Fe 7 (PO 4 ) 6 , or NASICON-type Na 3 Fe 2 (PO 4 ) 3 ) [24]. Regarding the phase change of milled m-NaFePO4 at 390°C, unfortunately, no information is reported in the literature. Terefore, this needs to be further investigated in future studies. Subsequently, no noticeable material loss/outgassing (nonvolatile transformation) is observed until 600°C (see Figure S6, supplementary material). EDX analysis in supplementary Figure S8 confrms the elemental distribution of NaFePO 4 with carbon contents. Figure 8 shows the galvanostatic measurement results on coin cells made with pristine m-NaFePO 4 , 12 h milled (amorphous phase not calcinated), and 12 h milled + calcinated at 600°C (mixed with glucose) active cathode materials. A capacity of nearly 48 mAh.g −1 is achieved in the case of milled-coated material. Nevertheless, 12 h milled powder without calcination shows no change in electrochemical performance. Tis result confrms the hypothesis of Ellis et al. [25] in part and the result of Zaghib et al. [26] regarding the difculty to remove sodium from maricite materials. Tis underlines that extended grinding transforms the material into an amorphous phase, which is consistent with the XRD analysis ( Figure 1). It is interesting to note that a distinct plateau region is observed in the charge and discharge curves at 2.6 V, which is also reported for Li 2 MnSiO 4 @C (glucose as a source for carbon) cathode by Devaraj et al. [27] and recently by Boyadzhieva et al. for m-NaFePO 4 [28]. Te appearance of this plateau indicates the characteristics of the phase transformation in the crystal structure [29]. Tis plateau expresses an increase of capacity of around 25 mAh.g −1 (50% of achieved capacity). Tis result further proves that 12 h ground NaFePO 4 with the calcination (with coating) process produces partly a mixture of olivine structure/oxidized-NaFePO 4 @C, which is comparable to electrochemical data in reference [13]. Te maricite NaMnPO 4 is in the Fe 2+ oxidation state, in which the sodium intercalates during charging and Fe 2+ oxidizes to the Fe 3+ state. While discharging, the sodium deintercalates and reduces from Fe 3+ to Fe 2+ oxidation state.

Conclusions
To enable and improve the electrochemical performance of the maricite electrode materials, milling and carbon coating were implemented and the electrochemical performance of the respective half-coin cells was investigated and compared. Te pristine m-NaMnPO 4 and m-NaFePO 4 showed a very low achievable capacity (electrochemically inactive) and milling combined with carbon coating facilitates the electron transport, which demonstrates better cell performances with discharge capacities of ∼50 mAh.g −1 in both materials. However, the long milling time (12 h) caused m-NaFePO4 to undergo an amorphous phase change, yielding no signifcant change in the capacity without calcination instead. Tis study also suggests that many other inactive materials can be converted into active materials by a combination of nanosizing and carbon coating procedures. Te nanosized particles reduce the difusion pathway in process of intercalation and increase the active surface area, which facilitates enhancement of the electrochemical performance. However, due to poor electronic conductivity, additives like carbon or fuorine are used to increase the electronic conductivity.