On the use of guanidine hydrochloride soft template in the synthesis of Na 2/3 Ni 1/3 Mn 2/3 O 2 cathodes for sodium-ion batteries

cell will demonstrate the reliability of this material as an electrode for Na-ion cells


Introduction
Despite of the great success of the commercialization of Li-ion batteries in the electronic consumer market, the transference of this technology to electric vehicles and energy storage is finding significant issues. Lithium is a geochemically scarce element, and hence is found in low concentrations, thus being difficult to extract. This fact leads to several restrictions such as the uneven distribution of lithium deposits, low impact of recycling on supplies and increasing price of Li 2 CO 3 [1e5]. For this reason, researchers are paying their attention to alternative technologies, which may find their own niche in the future energy storage market.
Among them, sodium-ion batteries are being envisaged as promising devices for large-scale grid storage linked to renewable energy and smart grids [6]. The high abundance of sodium and its low health impact are interesting advantages, which may allow the production of batteries cheaper than the Li analogues [7].
Notwithstanding, the less negative reduction potential, as compared to lithium, limits the potential of Na-ion batteries to values ca. 0.3 V lower than those of lithium-ion batteries. In addition, the large atomic weight and ionic radius of Na þ could be considered as factors reducing the ion diffusivity, though certain frameworks, as P2 in layered oxides, provides fast kinetics for sodium migration [8]. Even if these factors lead to a slightly lower performance of the Na cells, these devices may find particular applicability for which these drawbacks have a less impact. Otherwise, the fundamental concept of Na-ion batteries is also based on the intercalation chemistry of the alkaline elements, which is leading to a fast implementation of this research [9e12].
Cathode materials for Na-ion batteries cover a wide variety of compositions including Prussian blue compounds [13], transition metal sulfides [14], organic materials [7], metaleorganic frameworks [15] and mainly transition metal phosphates and oxides [16e18]. Sodium containing transition metal oxides have the Na x TMO 2 nominal stoichiometry, TM being a transition metal element as Mn, Fe, Ni, etc. Numerous Mn containing oxides have been reported as performing cathode for Na-ion batteries because of their high potentials vs. Na þ /Na redox couple, good cyclability and volumetric capacity [19]. The environmental friendship and low cost of this transition metal, and the easy preparation of its oxides are significant advantages.
P2-type Na 2/3 Ni 1/3 Mn 2/3 O 2 is commonly reported as a promising cathode material, whose phase stability provides an excellent cycling stability [20,21]. Among the most recent strategies to improve the electrochemical performance, doping with inactive elements has been efficient to stabilize the P2 structure [22,23]. Also, cobalt substitution allowed to increase significantly the cell capacity [24e26]. The preparation of cathode materials with controlled morphology is crucial to optimize a good electrochemical behavior. Reports on the effect of specific morphologies as plates [27], microflakes [28], or nanosheets [29] have been published. Recently, guanidine has been successfully used as precipitant to prepare transition metal oxide with applicability as electrodes. In fact, the influence of the precipitant to yield a material with optimized morphology was demonstrated by Xin et al. [30]. The weak alkali guanidine hydrochloride slowly releases OH À ions favoring a slow rate of the precipitation reaction, leading to the separation of nucleation and growth steps. Yao et al. have successfully reported the use of guanidine hydrochloride during the synthesis of ribbon-like NiO [31], and hollow Co 3 O 4 nanotubes [32] oxides for supercapacitor electrode applications.
In this work, we propose the synthesis of P2-Na 2/3 Ni 1/3 Mn 2/3 O 2 with an anisotropic growth of particle agglomerates using guanidine as a soft template. The effect of the amount of guanidine on the structural and morphological properties will be characterized by Xray diffraction and electron microscopy. Furthermore, sodium half cells will be tested by the galvanostatic method to examine the electrochemical performance.

Experimental section
Four different samples with Na 2/3 Ni 1/3 Mn 2/3 O 2 general stoichiometry were prepared. The typical synthesis procedure is described as follows: 1.0913 g of sodium nitrate (NaNO 3 98% purity; Synth), 1.8669 g of nickel (II) nitrate hexahydrate (Ni(NO 3 ) 2 $6H 2 O 98% purity; Alfa Aesar), and 3.2231 g of manganese (II) nitrate hexahydrate (Mn(NO 3 ) 2 $6H 2 O 98% purity; Alfa Aesar) were dissolved in 80 mL of deionized water and stirring during 10 min. The amounts corresponding to 2%, 4% and 8% of guanidine hydrochloride (CH 5 N 3 $HCl 98% purity; Alfa Aesar) (in weight proportion referred to the precursor mass) was added. The resulting solution was exposed to high-intensity ultrasonic irradiation (Branson Digital Sonifier) in a continuous mode with 65% amplitude for 20 min. Afterwards, water was evaporated from the solution in an oven during 24 h at 100 C until a green gel was achieved.
The gel was dried at 150 C (1 C/min) for 30 min, and further heated at 350 C (5 C/min) for 30 min to allow the decomposition of nitrate compounds present in the gel, and then at 900 C (10 C/ min) for 6 h in order to obtain the Na 2/3 Ni 1/3 Mn 2/3 O 2 powders. These samples will be named as NaNMO-X%, being X the relative amount of guanidine used as a soft template.
The purity and crystallinity of the obtained samples was characterized by X-ray diffractometry (XRD). The patterns were scanned between 10 and 120 (2q) on a Shimadzu diffractometer (XRD-6000) equipment provided with Cu Ka radiation and a graphite monochromator. A scan rate of a 1 /min was applied. The unit cell   parameters, crystallite size and microstrains were calculated with TOPAS software. The size of the coherently diffracting domains (volume-weighted mean column height) and microstrain content (e0) were determined from the Gaussian and Lorentzian components of the integral breadths by assuming modified Voigt functions. Low resolution transmission electron micrographs (TEM) were acquired in a JEOL 1400 microscope. Field Emission Gun-Scanning Electron Microscopy (FEG-SEM) was performed with a Zeiss Auriga scanning electron micro analyzer with an accelerating voltage of 15 kV. High resolution TEM (HRTEM) images were obtained with a JEOL JEM 2010 microscope working at 200 kV and provided with a high-resolution side-entry objective lens.
Electrochemical cyclability was tested in Swagelok™ type twoelectrode cells by the galvanostatic method. Sodium half cells were assembled in an argon filled MBraun glove box under controlled O 2 and H 2 O traces. Working electrodes, with a mass loading of about 3 mg, were prepared by dissolving PVDF (polyvinylidene fluoride) (10%) in N-methyl-2-pyrrolidone (Emplura, 99.5%). Then, the active material (80%) and carbon black (10%) were dispersed to ensure a homogenous paste. The mixture was spread onto a 9 mm aluminum disk and vacuum dried at 120 C for several hours. The counter electrode was a 9 mm sodium disk (Panreac, 99.8%). Both electrodes were separated by glass fiber disks (GF/A-Whatman) soaked in an electrolyte solution consisting of 1 M NaClO 4 (Strem, >99.%) (PC:propylene carbonate) (Sigma-Aldrich, 99.7%) (2% wt. FEC:fluoroethylene carbonate) (Sigma-aldrich, 99%). A VMP multichannel system was used to monitor the charge and discharge half cycles. In order to maximize both capacity and reversibility, the potential windows were set at 2.0e4.3 V and 2.0e4.0 V. The cells were cycled at several C rates ranging from C/20 to 4 C. Cyclic voltammograms were performed at 0.1 mV s À1 . Electrochemical Impedance Spectra (EIS) were acquired to resolve the cell internal impedance. Swagelok™ type three-electrode cells were assembled to discriminate the effect on the working electrode. Sodium disks were used as counter and reference electrodes. After subjecting the cells to one cycle, they were allowed to relax to reach a quasi-equilibrium state. Then, the impedance spectra were measured on an SP-150 Biologic by perturbing the open circuit voltage with an AC signal of 5 mV from 100 kHz to 2 mHz.
The electrochemical behavior of a full sodium-ion cell was performed by using a selected Na 2/3 Ni 1/3 Mn 2/3 O 2 sample as a positive electrode. A sucrose derived hard carbon, carbonized at 1300 C for 5 h, was chosen as a counter electrode. These electrodes were assembled in a three electrode Swagelok™-type cell with metallic sodium as reference electrode. An excess of the carbonaceous anode was weighed to ensure that the cell potential was controlled by the behavior of the cathode. Capacity values were referred to the active mass of the positive electrode. Initial irreversibility during the first half cycle were circumvented by sequentially charging the positive and discharging the negative electrodes in half cell configuration before starting the full cell experiment. Fig. 1 shows the XRD patterns of P2-Na 2/3 Ni 1/3 Mn 2/3 O 2 samples prepared in the presence of different amounts of guanidine template. These patterns reveal a set of narrow reflections, which were mostly indexed in the P6 3 /mmc space group of the hexagonal system (JCPDS 27-0751 e Na 0.7 MnO 2.05 ). This structural scaffold is similar to that of related lithium oxides, resulting from the hexacoordination of the transition metal to surrounding oxygens (TMO 6 ). These polyhedra are assembled as two-dimensional slabs, parallelly stacked and accommodating mobile sodium ions at the interlayer spacing. Structural varieties as O3/, P2-, and P3-phases arise from the distinct stacking of slabs, which ultimately create either octahedral or prismatic sites for sodium at the interslab space. Particularly, the P2-type structure studied here refers to sodium located at prismatic sites and 2 slabs to define the unit cell [33]. Peaks at 37.2, 43.3 and 62.9 2-theta were ascribed to the presence of minor impurities of NiO (JCPDS 04-0835) which grow with the content of guanidine. Eventually, a peak located at ca. 45 was ascribed to the presence of and P3-Na x Ni 0.5 Mn 0.5 O 2 phase (R3m space group). The lattice parameters of the P2 phase were close to those reported elsewhere (Table 1) [34,35]. Crystallite size and strain values were calculated from the line broadening analysis of reflections by fitting to pseudo-Voigt functions. The crystallite size is referred as the integral breadth-based volume-weighted column height. On increasing the percentage of guanidine, minimum values of both parameters were determined for NaNMO-2%. Low crystallite size and strain values may favor the accessibility of Na þ ions coming from the electrolyte to the inner core of particles.

Results and discussion
FEG-SEM images of non-treated NaNMO-0% sample showed particle agglomerates with heterogeneous size (Fig. 2). Otherwise, the use of guanidine as a soft template during the synthesis led to agglomerates in which the primary particles appeared enchained in a single direction. This fact is mainly observed for NaNMO-2% and NaNMO-4%. Finally, NaNMO-8% is again characterized by large particles with undefined shape. These results seem to indicate that the effect of guanidine as a template on particle morphology is limited to a specific percentage. Thus, the use of a large amount of guanidine, larger than 4%, exerts an opposed effect to the expected one. Fig. 3a shows a low resolution TEM top view of the plate like particles of the mixed oxide and variable diameter of several hundreds of nanometers. The crystalline nature of the particles is evidenced by the 〈301〉 zone axis in the selected area electron diffraction (SAED) pattern in Fig. 3b. A side view of the plate like particles and its corresponding fast Fourier transform (FFT). Fig. 3c and d, respectively, reveal 002 lattice fringes of ca. 0.56 nm and plate thickness of ca. 5 nm. Fig. S1 shows the TEM micrographs of Na 2/3 Ni 1/3 Mn 2/3 O 2 powders. For the NaNMO-0% powder, the Na 2/ 3 Ni 1/3 Mn 2/3 O 2 secondary particles are constituted by aggregates of several primary particles, forming irregular morphologies. The increase of guanidine hydrochloride (NaNMO-2%) favored the anisotropic growth of aggregates consisting of plate-like primary particles. On increasing the guanidine hydrochloride concentration, mainly for NaNMO-8%, an enhanced compaction of the plates was found. X-ray energy dispersive spectroscopy (EDS) analysis, performed at different particles evidenced that all samples were composed by Na, Ni, Mn and O elements with a uniform distribution (Fig. S2).
Cyclic voltammograms for NaNiMO-0% and NaNMO-2% were performed to unveil three redox couples at 3.30/3.17, 3.69/3.6, and 4.25/4.00 V (Fig. 4). They belong to the successive redox reactions from Ni 2þ to Ni 4þ , as responsible for the overall capacity of this electrode within this potential window [36]. The splitting of peaks has been ascribed to the different Na-vacancy rearrangement in the sodium layer [37]. Fig. 5 shows the galvanostatic charge and discharge curves of sodium half-cells assembled with P2 type Na x Ni 1/3 Mn 2/3 O 2 samples, recorded at several rates in the 2.0e4.3 V range. The first charge for NaNMO-2% and 4% respectively reached values of 162 and 150 mA h g À1 , at the lowest rate. These values are close to the theoretical value of 173 mA h g /1 for the full sodium extraction at x ¼ 0.66. Notwithstanding, the coulombic efficiencies achieved after cell discharge were only 77.6% and 83.1%. These low values can be explained in terms of the P2-O2 phase transition at the 4.2 V plateau, as will be further discussed. As expected, the increase in C rate involves a decrease of capacity and an increase of the charge-discharge hysteresis because of the imposed kinetic. An analysis of the hysteresis will allow to infer conclusions about the kinetic behavior, as discussed below. Fig. 6 shows the ex-situ XRD patterns recorded on charged-discharged NaNMO-2% electrodes (JCPDS 27-0751 e Na 0.7 MnO 2.05 ). The presence of small plateaus below 3.9 V arise from either single-phase or two-phase regions with very similar cell parameters in which the P2 structure is preserved. Most likely, due to the ordered arrangements of intercalant. Also, the small plateaus below 3.0 V for guanidine treated samples, suggest that additional capacity comes from the contribution of the Mn 3þ /Mn 4þ couple [38].
In contrast, the plateau at 4.2 V corresponds to a two-phase system in which the P2 modification coexists with new a Na free O2-type structure with stacking faults [39,40]. Mayor cyclability issues are commonly ascribed to this phase transition and hence large efforts are made to alleviate this effect. For instance, good results are obtained by metal doping [41]. XRD patterns in Fig. 6 b and c, recorded along that plateau, show the appearance of a small peak at 20.2 ascribable to the (002) reflection of the O2 phase [39]. Otherwise, the reflections of the P2 phase are attenuated but not disappeared. This partial suppression of the P2-O2 phase transition is commonly considered as a favorable fact for cyclability improvement [41]. Because of the lack of stability of the O2 phase, a hydrate phase would be produced in the presence of unavoidable air traces, leading to the occurrence of two peaks at ca. 12.5 and 25 (2q) [41,42], marked as # in Fig. 6.
From the observation of the galvanostatic profiles in Fig. 5, we can detect how the charge-discharge hysteresis depends on the applied current intensity. This effect is due to the thermodynamics of the insertion/extraction reaction and the transport hindrance through the electrode interfaces [43]. Fig. 7a shows a linear relationship between cell polarization and current density. As can be seen, there is a correlation between capacity fading and the linear slope. Thus, NaNMO-2% and NaNMO-4%, featuring the best capacity retention, also showed low slope values. In fact, the latter value is the so-called direct current resistance imposed to ion migration during cell operation. These values have been plotted in Fig. 7b and show minimal values for NaNMO-2% and NaNMO-4%. Surprisingly, the linear extrapolation of these plots at zero current reveals a finite cell polarization, which has been interpreted in terms of a sequential particle-by-particle mechanism [43]. These values have been also plotted in Fig. 7b, revealing again low values for NaNMO-2% and NaNMO-4%. Fig. 8 shows the rate capabilities of the electrode materials when the current rate is increased from C/20 to 4 C for some cycles in the potential window between 2.0 and 4.3 V. Although this procedure allows to maximize the cell capacity to values close to the theoretical one, its well known that the plateau at 4.2 V, belonging the P2-O2 two phase system induces some capacity fading on cycling. This fact was more pronounced at the lowest rate for the NaNMO-0% samples non-treated with guanidine, evidencing the positive effects of the synthesis procedure. Thus, all samples showed first discharge capacity values close to 130 mA h g À1 , which are close to those previously reported [40,44]. Then, this value decreased to 111 mA h g À1 after the fifth cycle. Otherwise, guanidine treated samples preserved values of ca.125 mA h g À1 . This effect was even more marked when the current density was increased to 4C, leading to values of only 16 mA h g À1 at the 35th cycle. It can be also observed that an excess of guanidine led to an enhanced capacity fading for NaNMO-8%. Despite this fact, all samples evidenced a good capacity recovery when retrieving the lowest C rate after 35 cycles. Further cycling at C/20 was prolonged beyond the 45th cycle. At this point, guanidine treated samples showed a better capacity retention than NaNMO-0%.
It is a common strategy to limit the upper cut-off voltage to 4.0 V to avoid the detrimental presence of the two-phase system at the end of charge. It is well known that these conditions limit the capacity values but notoriously improve the capacity retention. Fig. 9a displays the rate capabilities of the studied samples cycled in the range from 2.0 to 4.0 V. After the first cycle at C/20, samples treated with guanidine showed higher capacity (ca. 77 mA h g À1 ) than NaNMO-0% (64 mA h g À1 ). This difference was even more marked at the highest rate for which the latter sample performed only 35 mA h g À1 . On increasing the C rate, NaNMO-4% retained the highest capacity values after 2C (61 mA h g À1 ), but then a marked capacity decrease was detected at 4C. Similarly, the sample treated with the largest amount of guanidine (NaNMO-8%) showed a poor behavior at high C rates. Although all samples showed a good capacity recovery when returning back to C/20 after the 35th cycle, NaNMO-2% showed the best capacity retention. An extended cycling experiment was performed at C/5 (Fig. 9b). Coulombic efficiency values close to 100% were recorded for NaNiMO-2% leading to an excellent capacity retention of 97% of the initial capacity after 200 cycles. This value decreased to 87% for NaNMO-0% in the same conditions.
Expectedly, the differences in particle morphology would affect the activation barriers exerted to sodium migration through the electrode-electrolyte interphase at the working electrode. To unveil this question, impedance spectroscopy is a helpful technique that allows the determination of the internal resistance of the electrochemical cell. Fig. 10a shows the typical Nyquist plots recorded for the studied samples after one cycle. These profiles feature two semicircles at high and medium frequencies, respectively, arising from capacitive behavior at the surface passivating layer and charge-transfer reaction at the interphase. In order to calculate the internal resistances at the surface layer (R sl ) and charge-transfer reaction (R ct ) at the interphase of the working electrodes, the impedance spectra were fitted to the equivalent circuit, shown as an inset in this figure (Table 2). Another component of this circuit is the ohmic drop at the electrolyte (R el ), which value was negligible as compared to the overall resistance. In addition, the constant phase element (CPE) is attributed to the capacitive behavior in polycrystalline samples and the Warburg element (W) refers to ion diffusion in the bulk of the particles. Charge transfer resistance values are commonly higher than R sl ones. Particularly, NaNMO-0% features the largest R ct value so that the semicircles appear strongly overlapped with the straight line at low frequencies. On increasing the amount of guanidine, the R ct value abruptly decreases until 0.74 Ohm g for NaNMO-2%, evidencing the beneficial effect of the use of guanidine to diminish the resistance of the charge transfer reaction. This result confirms the superior electrochemical performance of the latter sample on galvanostatic cycling. Also, the analysis of the linear plot of Z 0 versus the reciprocal square root of the lower angular frequencies allows to calculate the apparent diffusion coefficients of sodium (Fig. 10b). For this purpose, the following equation was used: where R is the gas constant, T is the absolute temperature, A is the geometrical electrode area, F is the Faraday's constant, and C is the molar concentration of Na þ ions in the formula unit. The Warburg coefficients (s w ) can be inferred from the slope of the straight lines [45]. Table 2 reveals that the apparent diffusion coefficients determined for the guanidine treated samples were two orders of magnitude higher than that of NaNMO-0%, involving that the morphological modifications induced by guanidine favor the accessibility of sodium to the host framework. The reliability of these electrode materials to act as a positive electrode in a sodium-ion cell was demonstrated by assembling a full cell using NaNMO-2% as a cathode. This sample was selected in view of its excellent rate capability and low internal resistance. As a negative electrode, a hard carbon was employed. A large excess of hard carbon was added to discriminate the influence of the anode material. In order to improve the cell cyclability, cell charging proceeded at a slower rate (C/25) than discharge (C/15). For this reason, charge curves appear longer than discharge ones in the capacity axis. The expected electrochemical reaction during discharging the cell is described as follows: The galvanostatic profiles show a capacity deliverance of ca. 70 mA h g À1 after 20 cycles with an excellent capacity retention (Fig. 11). The average cell discharge is 3 V, resulting an energy density value of 210 W h kg À1 . This superior performance allows regarding the use of guanidine as an optimized synthesis procedure to promising electrode materials for sodium-ion cells.

Conclusions
The effect of guanidine as a soft template in the synthesis of Na 2/ 3 Ni 1/3 Mn 2/3 O 2 has been evaluated. XRD patterns revealed highly crystalline samples, with minor NiO impurities, and the decrease of both crystallite size and strain for guanidine treated samples, which may favor sodium accessibility at the electrode/electrolyte interface. Guanidine promoted the aggregation of primary particles as elongated agglomerates with heterogeneous size, though this effect was limited to intermediate amounts of added guanidine. Galvanostatic tests between 2.0 and 4.3 V showed multi-stage charge and discharge curves. NaNMO-2% and 4% samples evidenced better capacity retention and low charge-discharge hysteresis when subjected at rates from C/20 to 4C. The hysteresis was quantified by measuring the cell polarization and further calculation of the zero-current polarization and direct current resistance. These parameters confirmed the benefits of using 2 and 4% of guanidine during the synthesis of these samples. A partial suppression of the P2-O2 phase transition could be responsible for the electrochemical improvement.
Notwithstanding, the two-phase transition existing at the plateau at 4.2 V prevented an acceptable cycling stability after only 45 cycles. To solve this issue, the voltage window was limited at 2-4 V in further experiments. These tests showed that the use of 2% of guanidine provides excellent rate capability at the highest applied current (4C) and capacity retention after a long cycling at C/5. It was corroborated by determining lower internal resistance values at the electrode-electrolyte interphase and higher apparent diffusion coefficients for this sample. A full sodium-ion cell was assembled with NaNMO-2% as a positive electrode to demonstrate the reliability of this material as an electrode for Na-ion cells. Fig. 10. a) Nyquist plots of P2-Na 2/3 Ni 1/3 Mn 2/3 O 2 samples recorded after the first cycle at C/20. Inset: Equivalent circuits used for the spectra fitting; b) Linear segment of the real impedance versus the reciprocal frequency for the calculation of diffusion coefficients. Table 2 Resistance values, Warburg impedance coefficients (s W ) and diffusion coefficients (D) calculated from the impedance spectra of P2-Na 2/3 Ni 1/3 Mn 2/3 O 2 electrodes after one cycle at C/20. R el /Ohm g R sl /Ohm g R ct /Ohm g s W /Ohm s À1/2 D/cm 2 Fig. 11. Galvanostatic charge and discharge curves of a full hard carbon//NaNMO-2% sodium-ion cell recorded at C/25 charge and C/15 discharge rates. Inset: plot of discharge capacity versus number of cycles.