Kwangjin Park,
Dongwook Han,
Hyunjin Kim*,
Won-seok Chang,
Byungjin Choi,
Benayad Anass and
Seoksoo Lee*
Samsung Advanced Institute of Technology (SAIT), Samsung Electronics, Mt. 14-1, Nongseo-dong, Giheung-gu, Yongin-si, Gyeonggi-do 446-712, Republic of Korea. E-mail: ydmj79.park@samsung.com; seoksoo.lee@samsung.com; hyunjin1.kim@samusng.com; Fax: +82-31-8061-1319; Tel: +82-31-8061-1250
First published on 1st April 2014
A chelating-agent-assisted Na2/3Fe1/2Mn1/2O2 material is studied as a cathode for sodium-ion batteries. The addition of NH4OH results in the formation of a homogeneous powder with a lower resistance and larger carbon content on the cathode surface. The formation of a thin and stable solid-electrolyte interface layer leads to its enhanced electrochemical performance.
Among known NaxMO2 materials, a P2-type Na2/3Fe1/2Mn1/2O2 (NFMO) has recently emerged as a superior material with the best capacity (190 mA h g−1) between 4.2–1.5 V in a 1 M NaClO4 electrolyte.1 However, this fascinating layered oxide cathode must still overcome several major drawbacks for its successful commercialization. The first issue to be addressed is its high initial irreversible capacity, which is thought to arise from side reactions occurring with the electrolyte.8 The second issue is its low rate capability, possibly originating from the formation of a thick solid-electrolyte interfacial (SEI) layer on the cathode surface during cycling or from its low electronic conductivity. This phenomenon may be significantly related to the preparation method. Therefore, effecting precise control over the preparation conditions or synthetic modifications is a promising strategy to overcome these drawbacks.8,9 Many research groups have reported that the addition of a chelating agent during synthesis enhances the electrochemical performance by increasing the uniformity of powder size and crystallinity.8–11
In this study, the NFMO material was synthesized by a co-precipitation method using NH4OH as the chelating agent. A 15% increase in the initial capacity as well as a capacity retention of 94% after 30 cycles were obtained utilizing this chelating-agent-assisted Na2/3Fe1/2Mn1/2O2 (C-NFMO) material. Furthermore, we also investigated the effects of the chelating agent on the NFMO structure, morphology, and electrochemical performance.
Determination of the sample’s structural characteristics was performed by X-ray diffractometry (XRD) using Cu-Kα radiation and a scan speed of 0.02° min−1 between 2θ = 10–65° at an applied potential of 40 kV and current of 40 mA. X-ray photoelectron spectroscopic (XPS) analyses were performed with a Physical Electronics (PHI) (Quantum 2000 Scanning ESCA Microprobe) spectrometer using a focused monochromatized Al-Kα radiation source (1486.6 eV). To avoid any external contamination, all samples were transferred under an Ar environment from the glove box to the spectrometer using a transfer chamber. Morphological changes on the NFMO surface were investigated using scanning electron microscopy (SEM, Hitachi S-4700N).
Composite positive electrodes comprising 60 wt% active materials, 20 wt% DENKA BLACK®, and 20 wt% polyvinylidene difluoride (PVDF) were pasted onto Al foil as a current collector. The electrodes were dried at 120 °C under vacuum and pressed. Metallic Na was used as the anode electrode. The electrolyte solution used was 1.0 M NaPF6 dissolved in propylene carbonate (PC). Glass fiber was used as the separator. R2032-type coin cells were assembled in an Ar-filled glove box. Cyclic voltammetry (Model VMP3, NanoQuébec) was performed at a 10 μV s−1 scan rate. For galvanostatic experiments, the cells were discharged–charged galvanostatically at a constant current.
Cathode electrochemical activities were characterized by electrochemical impedance spectroscopy (EIS) using a Solartron 1260 frequency-response analyzer with an applied frequency from 10 mHz to 1 MHz. EIS measurements were performed on the charged and discharged states during the cycle. Impedance spectra were analyzed with an equivalent circuit using the ZView® (Windows) software.
During synthesis, precipitation occurs when the aqueous acid and alkali solutions are mixed together. Upon the addition of a chelating agent, homogeneous [Mn1/2Fe1/2(NH3)n2+] is formed with a stoichiometric ratio of Mn and Fe (eqn (1)), which facilitates the formation of a homogeneous MnFe(OH)2 phase in the precipitation process (eqn (2)):
![]() | (1) |
![]() | (2) |
Therefore, the formation of by-products as a result of phase separation – Mn(OH)2 and Fe(OH)2 – is prevented, and a well-defined metal hydroxide is acquired using a chelating agent, resulting in the synthesis of a homogeneous, highly crystalline NFMO powder.
Fig. 2(a) compares the cycling performance of the cells cycled at 25 °C with a constant rate of 0.1 C. At a cutoff voltage of 1.5 V, the cells delivered reversible capacities of ∼165 mA h g−1 and ∼195 mA h g−1 for NFMO and C-NFMO, respectively. When the cutoff voltage was increased to 2.0 V, NFMO and C-NFMO displayed capacities of ∼145 mA h g−1 and ∼166 mA h g−1, respectively. The initial capacity of C-NFMO corresponded to an increase of about 14% compared to NFMO, regardless of cutoff voltage. The capacity retention of C-NFMO was found to be similar to that of NFMO under the 1.5 V voltage cutoff conditions. This result might be due to the phase transition as the result of a stacking fault, in which the octahedral and prismatic layers are piled up along the c-axis direction, which effectively decreases repulsion between Na ions.1,12 On the other hand, the capacity fading of C-NFMO was much slower than that of NFMO, with a capacity retention of 94% at 0.1 C at 30 cycles at a cutoff voltage of 2.0 V; this corresponds to an improvement in capacity retention of greater than 20%, compared to that for NFMO. The cyclic performance is clearly improved with the incorporation of a chelating agent during preparation.
![]() | ||
Fig. 2 (a) Cycling performance comparison of NFMO and C-FNMO. (b) Comparison of rate capabilities of NFMO and C-FNMO. |
NFMO rate capabilities were next evaluated by cycling at various rates from 0.05–5 C, and the results are presented in Fig. 2(b). As the applied current was increased, the capacities of all samples decreased; this observation can be ascribed to an increase in over-potential resulting from the lack of reaction time for mobile cation intercalation into the crystal lattice, as only the surface of the active materials participates in the reaction.7 After C-rate tests, the original capacities of both samples were restored.
The electrochemical reactions of NFMO and C-NFMO were confirmed by cyclic voltammetry analyses, and the resulting cyclic voltammograms are shown in Fig. 3(a) and (b), respectively. For NFMO, two anodic peaks at 2.78 V and 3.95 V, and two cathodic peaks at 1.88 V and 3.25 V were observed, which correspond to the intercalation and extraction processes of mobile Na+ ions, respectively. On the other hand, for C-NFMO, the anodic peaks were shifted to a lower voltage (2.68 V and 3.88 V), while the cathodic peaks experienced a similar shift to higher voltage (1.89 V and 3.37 V), indicative of the improved kinetic properties. The lower cathodic peak is related to the Mn3+/Mn4+ redox process, while the higher cathodic peak is concerned with the Fe3+/Fe4+ redox. The sharp increase in the anodic curve at around 4.2 V is related to a quasi-reversible P2-type-to-O2-type phase transition occurring at low Na content in the material.1,5,7
![]() | ||
Fig. 3 Cyclic voltammograms: (a) NFMO, (b) C-NFMO. Typical charge–discharge characteristics: (c) NFMO, (d) C-NFMO. |
Fig. 3(c) and (d) show the potential versus capacity profiles of NFMO and C-NFMO during the 1st, 2nd, 5th, 10th, 20th and 30th cycles at 0.1 C, with profiles for both materials showing a similar slope as well as other features. The location of the potential plateau was consistent with the results of the cyclic voltammetry experiments; the profile for C-NFMO exhibited a wider plateau than bare NFMO, indicating the existence of more reaction sites. The capacity fading for the two samples can be mostly attributed to the decrease in the high potential plateaus, especially those at around 3.4 V. However, the capacity fading of C-NFMO after 30 cycles is much smaller than that of NFMO.
During the intercalation process, Na+ ions can be inserted into two different types of trigonal prismatic sites: Naf and Nae, where Naf shares faces with two TmO6 octahedra of adjacent slabs, while Nae shares edges with six surrounding TmO6 octahedra. This phenomenon of partial interchange in the occupancy of Na+ and transition-metal ions among the sites is referred to as “cation mixing”.7,11 When cation mixing occurs, the capacity fading gradually increases due to the lower diffusivity of the mobile ion in conjunction with an increase in electronic resistance.12 Unfortunately, Mn and Fe materials possess higher probabilities of cation mixing in the layered structure compared to other transition metals such as Co, Ni, V and Cr.12 The probability of cation mixing can be confirmed by examining the integrated ratio of the (003) to (104) peak in the XRD pattern, where the (003) peak appears as a result of diffraction of the layered rock-salt structure (Rm), whereas the (104) peak appears as a result of diffractions of layered and cubic rock-salt structures.13 A value of I(003)/I(104) < 1.2 is indicative of a high probability of cation mixing.11–14
The I(003)/I(104) peak ratios of NFMO and C-NFMO were calculated to be 1.3 and 4.8, respectively, indicating a low probability of cation mixing for both materials. In particular, the value for C-NFMO was much larger than that of bare NFMO, suggesting that the chance for cation mixing in C-NFMO is much smaller than that of NFMO. Thus, this high crystallinity attributed to C-NFMO may contribute to the increased capacity and improved cycle performance by reducing the probability of cation mixing and formation of impurity phases.
SEM images of NFMO and C-NFMO are shown in Fig. 4(a) and (b), respectively, in which a very distinct difference in the particle size distribution could be observed between the two materials. NFMO appeared moderately aggregated, with particle sizes of ca. 2–4 μm in diameter; C-NFMO, in contrast, was fairly homogeneous, with particle sizes of less than 1 μm in diameter. The surface areas of NFMO and C-NFMO as measured by the Brunauer–Emmett–Teller (BET) method were 4.158 m2 g−1 and 6.761 m2 g−1, respectively. The uniform and small particle size for C-NFMO may be ascribed to the co-precipitation procedure, which is known to result from the preparation of a homogeneous precursor solution by use of chelating agent.14,15
EDX analyses were performed to investigate the particle surface compositions, the results of which are shown in Table 1. Surface elemental compositions of the two samples were almost identical, with the exception of carbon; the carbon content on the C-NFMO surface was twice the amount found on the NFMO surface. The electrical conductivities of NFMO and C-NFMO, as measured by the four-point probe method, were 1.978 × 10−6 S cm−1 and 2.296 × 10−5 S cm−1, respectively. The electronic conductivity of NFMO with a smaller powder size is 2.65 × 10−6 S cm−1, which is slightly higher than NFMO, but still one order of magnitude smaller than C-NFMO. The higher conductivity of C-NFMO is attributed to a larger amount of carbon on the powder surface as well as the close packing of the C-MFMO particles resulting from the small particle size.
Element (at%) | |||||
---|---|---|---|---|---|
Na | Mn | Fe | O | C | |
NFMO | 14.85 | 28.55 | 30.08 | 22.33 | 4.18 |
C-NFMO | 16.52 | 25.73 | 23.77 | 25.81 | 8.16 |
To obtain further information on the SEI layer formed on the cathode surface, XPS analyses were performed after the 1st and 30th charge–discharge cycles. All samples were prepared after discharging until 2.0 V. The Fe 2p peaks and C 1s peaks are shown in Fig. 5(a) and (b), respectively. Considering that the X-ray-excited photoelectrons escaped only from the uppermost surface layer (∼10 nm), a smaller number of Fe 2p photoelectrons observed after 30 cycles suggested that the NFMO electrode was sufficiently buried beneath the SEI layer. In contrast, XPS results for C-NFMO saw the appearance of Fe 2p peaks in the same position although in weaker intensities. In both samples, the Fe 2p peak region showed an overlap with the F KLL Auger transition resulting from the deposition of NaF-like species during the salt decomposition process; this may be due to the formation of a thicker SEI layer on the cathode surface of NFMO.18
![]() | ||
Fig. 5 XPS core peaks of NFMO and C-FNMO at the 1st and 30th cycles: (a) Fe 2p and (b) C 1s. AC impedance spectra: (c) NFMO (d) C-NFMO. |
The core C 1s XPS peak (284.8 eV) was assigned to graphite in the electrode. Other components were located at higher binding energies: peaks at ∼289.1 eV and ∼290.2 eV were assigned to CO3 and –CF2 environments, respectively. After the first charge–discharge cycle, the two samples showed almost graphite peak and –CF2 peak. However, the main C 1s peak intensity at ∼285.7 eV for NFMO increased due to the deposition of polymeric materials associated with PC decomposition to form Na2CO3 and aliphatic carbon or alkyl –OH species after testing over 30 cycles. The formed Na2CO3 creates resistance so that the capacity fading for the sample using NFMO is faster compared to that using C-NFMO.
EIS tests were next conducted to investigate differences in electrochemical polarization effects between the pristine and chelating-agent-assisted materials. The impedance of a two-electrode coin cell was measured before and after cycle tests. The circuit and EIS plots are depicted in Fig. 5(c) and (d), respectively. The measured EIS can be divided into three parts composed of frequency regions similar to a Li-ion battery system, where each part is related to the charge–discharge process. The DC resistance (Rs in the equivalent circuit) mainly arises from the resistance of the electrolyte. Impedance in the intermediate-frequency region, i.e., charge transfer resistance (Rct), is related to the charge transfer in the cathode–electrolyte interface. Impedance in the low-frequency region, i.e., diffusion resistance (Rdiff), is attributed to both the migration of Na+ ions through the SEI film and the diffusion of Na+ in the electrolyte, a phenomenon referred to as Warburg impedance.11,13–16
The charge-transfer resistance (Rct) is shown in the first semicircle in the middle-frequency regions in Fig. 5(c) and (d). The resistances of both samples were similar since the cathode materials have the same material composition, regardless of the assistance of a chelating agent during preparation. After the 30th charge–discharge test, the radii of the first semicircles for both samples decreased, as shown in Fig. 5(c) and (d); this is attributed to the activation of NFMO during cycling, since the electrolyte diffuses into the cathode material. The lower resistance comes from the enhanced contact area between the electrolyte and cathode electrode and the stabilized layer of cathode material.
On the other hand, the diffusion resistance (Rdiff), represented by the second semicircles in the low-frequency region, showed a significant difference. After 30 cycle tests, the semicircle radius of the bare sample was much larger than that of C-NFMO. The reason for this result is that a thicker SEI layer or insulating layer is formed on the bare NFMO when compared to C-NFMO, which hinders the Na+ migration, in agreement with the XPS results. A stable SEI layer can be achieved through surface treatment of the active material, e.g., carbon coating.17 The higher amount of carbon on the C-MFMO surface might play a role as a carbon coating, which leads to the prevention of direct contact between the cathode material and electrolyte, as confirmed by XPS and impedance analysis.
This journal is © The Royal Society of Chemistry 2014 |