Introduction

Sodium-ion (Na-ion) batteries have been introduced as a possible alternative to lithium-based ion batteries due to several reasons, including abundant supply, low cost and less toxicity1,2. However, sodium cannot be simply swapped with lithium as sodium has a larger ion size (1.02 Å for Na+ as compared to 0.76 Å for Li+) and slightly different chemistry, resulting in sluggish reaction kinetics that usually causes low capacity, poor rate capability and poor cyclability1,3,4,5. Therefore, challenges still exist to find suitable anode materials for the development of Na-ion batteries and scientists are searching for the best material among a vast number of materials using the trial-and-error approach.

Among the various types of materials, metal oxides have been explored extensively for lithium-ion (Li-ion) batteries. Similar to Li-ion batteries, metal oxides can potentially be used as large capacity anodes for Na-ion batteries because of their high theoretical capacities resulting from conversion reaction in most cases. For example, one-step conversion reaction of metal oxide with Na-ions can deliver high theoretical specific capacities of >600 mAh g−1 according to the reaction described in Eq. (1) 6,7.

$${{\rm{M}}{\rm{O}}}_{x}+2x{\rm{N}}{\rm{a}}+2x{{\rm{e}}}^{-}\leftrightarrow x{{\rm{N}}{\rm{a}}}_{2}{\rm{O}}+{\rm{M}}({\rm{M}}={\rm{C}}{\rm{o}},\,{\rm{F}}{\rm{e}},\,{\rm{M}}{\rm{n}},\,{\rm{C}}{\rm{u}},\,{\rm{e}}{\rm{t}}{\rm{c}}.)$$
(1)

Among different conversion-type transition metal oxides for anodes, manganese oxides exhibit advantages of high capacity, natural richness, low cost and environmental benignity. Even though manganese oxides including mono, Mn3O4, Mn2O3, MnO2 and their carbon-based composites with different nanostructures have been found technologically important in Li-ion batteries8, the use of manganese oxides in Na-ion batteries is rarely reported. In 2014, Jiang et al.9 synthesised Mn3O4 and investigated its reactivity as anode towards sodium for the first time. Subsequently, Weng et al.10 prepared MnO2 using a SiO2-templated hydrothermal approach, which was used as conversion-type anode for Na-ion batteries. However, rapid irreversible fading of capacities following the cycling process is a common problem with MnO2 and Mn3O4 phases due to volume expansion and aggregation as well as low electronic conductivity. To tackle these problems with transition metal oxide anodes, nanostructure engineering is widely adopted, where structural parameters including particle size, crystal size and morphology act as critical factors in achieving maximum electrochemical performance. By adopting nanostructure engineering, much improvement of sodium storage properties was realised with MnO2 anode by the development of new structures such as MnO2 nanorods and nanoflowers11,12.

In this study, Mn2O3 is synthesised by combining a hydrothermal and a thermal decomposition method and used as Na-ion battery anode for the first time. We developed a cubic structure of Mn2O3 by simple thermal decomposition of manganese carbonate (MnCO3) precursor through controlled calcination temperatures. The crucial feature of this structure is that cubic particles of Mn2O3 are approximately 1–2 µm in size and are composed of numerous nanoparticles (sub-units) of 40–50 nm grown on the surface, leading to more accessible sites for electrolyte penetration into the bulk of the electrode, which facilitates ion transportation thereby promoting the insertion/deinsertion of Na-ions13. The obtained Mn2O3 anode prepared at 600 °C demonstrates an impressive capacity of 130 mAh g−1 at 100 mA g−1 after 200 cycles with a remarkable rate capability of 120 mAh g−1 at a very high current of 1000 mA g−1 even without any cation doping or carbon coating.

Results and Discussion

To verify the nature of decomposition and the formation temperature of Mn2O3, TGA analysis of both MnCO3 precursors (synthesized MnCO3 (denoted as MnCO3 (S) and commercially available MnCO3 (MnCO3 (C)) was performed. It is clearly observed that the nature of decomposition of MnCO3 (S) and MnCO3 (C) precursors within the same temperature range is different, as shown in Fig. 1. One-step decomposition is realised with MnCO3 (S), whereas MnCO3 (C) precursor shows the two-step decomposition. Generally, the initial undulation appeared with initial weight loss of ~3 wt.% caused by the loss of hydrated water from MnCO3. The weight loss appeared between 250 and 350 °C for MnCO3 (C), presumably due to the formation of MnO2 which is further confirmed by XRD analysis (Fig. S1). The commercial MnCO3 (C) was heated up to temperature 300 °C in air for 2 h and the obtained product was analysed by XRD. The XRD pattern shows the presence of MnO2 and MnCO3 (C) in the product, suggesting partial decomposition of MnCO3 (C) and formation of MnO214. It is obvious that the weight loss between 400 and 600 °C is attributed to the formation of Mn2O3 and the release of O214,15,16. Since only one-step decomposition slope is observed with MnCO3 (S), the possible reaction during the decomposition of MnCO3 (S) can be summarised as Eq. (2) 17,18:

$$2{{\rm{MnCO}}}_{3}+{{\rm{O}}}_{2}\to {{\rm{Mn}}}_{2}{{\rm{O}}}_{3}+2{{\rm{CO}}}_{2}$$
(2)
Figure 1
figure 1

TGA curves of MnCO3 (C) and MnCO3 (S) at a heating rate of 10 °C min−1 in air.

Full size image

Reducing the decomposition step of MnCO3 (S) indicates that the conversion process is favourable since the kinetic process is shortened to obtain clear facet of Mn2O3.

Figure 2 shows the XRD patterns along with the Rietveld refinement profiles for MnCO3 and Mn2O3. Figure 2(a) shows the refinement patterns for MnCO3 (C) and MnCO3 (S) precursors. All the diffraction peaks could be indexed to the rhodochrosite phase of MnCO3 (JCPDS card No: 044-1472) with R-3c space group (rhombohedral unit cell). The refined lattice parameters for MnCO3 (C) were a = 4.8045 (2) Å, c = 15.6892 (8) Å, and the unit cell volume was 313.636 Å3 (Table S1). For MnCO3 (S), the lattice parameters were smaller than MnCO3 (C) with a = 4.8019 (3) Å, c = 15.6798 (1) Å and a unit cell volume of 313.103 Å3 that are consistent with the previous report19. However, a small diffraction peak appeared at 2θ value of 28.5° for MnCO3 (S), which is attributed to impurity that could have formed due to incomplete utilisation of manganese precursor during the hydrothermal process20. Furthermore, the average crystallite sizes (L) of MnCO3 were calculated according to the Scherrer’s equation as shown in Eq. (3):

$$L=\frac{k}{\beta \,\cos \,\theta }$$
(3)

where, k is the constant (0.9394), λ is the X-ray wavelength of Cu-Kα (1.5148 Å), β is the FWHM of the XRD peak in radian and θ is the angle of diffraction. The calculated average crystallite sizes of the MnCO3 (C) and MnCO3 (S) samples were ~36 nm and ~35 nm, respectively, which is in good agreement with reported result21. Figure 2(b) shows quantitative analysis of the Rietveld refinement fit profiles along with the observed XRD patterns of Mn2O3 (C600), Mn2O3 (S500), Mn2O3 (S600) and Mn2O3 (S700) samples. All the diffraction peaks matched well and could be indexed to a cubic Mn2O3 with the space group of la-3 (JCPDS card no.041-1442). No other diffraction peaks of any impurities have been detected. As the calcination temperature increases, the intensity of the diffraction peaks of Mn2O3 increased, indicating improved crystallinity22. No changes were observed in the crystallite size of Mn2O3 (~23 nm) at the calcination temperature of 500 and 600 °C, whereas it was ~31 nm at 700 °C based on the (222) peak. Nevertheless, crystallite size decreases during transformation of MnCO3 to Mn2O3. The details of lattice parameter, goodness of fit and other related fitting parameters of Mn2O3 obtained from crystal structure refinement are consistent with other reports23,24. Clearly, the lattice parameter of Mn2O3 increases as the calcination temperature increases due to the expansion during crystal growth.

Figure 2
figure 2

Rietveld refinement fits of the XRD data: (a) Commercial and synthesized MnCO3; and (b) Mn2O3 (C600), Mn2O3 (S500), Mn2O3 (S600) and Mn2O3 (S700) powders.

Full size image

Figure 3(a) shows FTIR spectra of D-glucose, MnCO3 and Mn2O3 samples. The FTIR spectrum of D-glucose shows the existence of a strong and broad absorption peak at 3391 cm−1 indicating the presence of v(OH) group stretching vibration. A small peak at 2920 cm−1 was attributed to the absorption peak of v(CH2) group, and the bands at 1475 cm−1 and 1328 cm−1 were assigned to the bending vibration of v(CH). The v(C–O) and v(C–C) stretching bands were observed at 1132 and 1007 cm−1, respectively25,26. During hydrothermal process, the D-glucose peaks (O–H bond at 3391 cm−1 and C–H bond at 1475 cm−1) were completely vanished due to the formation of MnCO3 (S) with the presence of C–O bending vibration of carbonate peaks at 1384, 860 and 721 cm−1 27. When MnCO3 was heated at high temperature, the carbonate peaks diminished. The presence of three absorption peaks located at 487, 559 and 655 cm−1 for Mn2O3 may be attributed to Mn–O and Mn–O–Mn, confirming the formation of Mn2O325,27,28, which is consistent with the Raman spectroscopy results as shown in Fig. 3(b). Raman active bands located at 170–1000 cm−1 may be due to the Mn–O vibration modes of manganese oxides29,30. The Raman bands at 310, 366 and 655 cm−1 are corresponding to the bending modes of Mn2O3, the asymmetric stretch of Mn–O–Mn and an asymmetric stretch of Mn2O3 corresponding to Mn (III)–O mode vibration group, respectively25.

Figure 3
figure 3

(a) FTIR spectra of D-glucose, MnCO3 (S), Mn2O3 (S500), Mn2O3 (S600) and Mn2O3 (S700); and (b) Raman spectra of Mn2O3 (C600), Mn2O3 (S500), Mn2O3 (S600) and Mn2O3 (S700).

Full size image

Surface morphologies of MnCO3 (C) and MnCO3 (S) precursors were examined using SEM analysis, as depicted in Fig. S2. SEM analysis reveals that particles are agglomerated clusters and non-uniformly distributed with a size of 0.5–1.5 µm. It is clearly observed that Mn2O3 (C600) obtained from MnCO3 (C) precursor possesses irregular shape with an approximate particle size between 0.5 and 1.3 µm (Fig. 4(a)). Conversely, well-defined cubic particles of Mn2O3 were formed through the calcination of MnCO3 (S) precursor (Fig. 4(b–d)). Mn2O3 (S500) consists of inhomogeneous cubic particles in the range of 1.0−1.2 µm in sizes (Fig. 4(b)). Mn2O3 (S600) exhibits larger cubic particles (0.9−1.7 µm sizes) (Fig. 4(c)) with numerous nanoparticles (sub-units) observed on the surface. Moreover, cubic Mn2O3 are organised by nanosized sub-units (40−50 nm) with distinct voids between the sub-units (Fig. S3). The d-spacing of 0.38 nm of the sample corresponded well with (211) lattice plane of Mn2O3 (Fig. S4). Clearly, well-defined cubic particles of Mn2O3 were observed with an average size of approximately 1.1−1.7 µm, when the precursor was calcined at a high temperature of 700 °C (Fig. 4(d)). Moreover, a porous-like structure is clearly visible at the surface of the cubic at 700 °C. This type of structure is often formed if metal carbonate is used as a precursor because it releases O2 and CO2 from the interior of the metal carbonate, which possibly leads to a finer or porous structure13,14,16. The obtained electron microscopy results demonstrate that one-step decomposition of the synthesised MnCO3 precursor produces a clear facet with well-defined cubic Mn2O3 structures compared to the two-step decomposition of the commercial MnCO3. The hysteresis loops (Fig. S5) reveal that these materials exhibit type IV isotherm, indicating a disordered mesoporous structure with average pore diameter between 5 and 60 nm. The calculated BET specific surface area for the Mn2O3 powders are tabulated in Table S2. It is well known that nanostructures play a crucial role in electrochemical processes due to their capability to enhance mass diffusion and transportation such as electrolyte penetration or ion transport31. Moreover, porous-like nanostructures can allow an electrolyte to diffuse smoothly within the lattice fringes of the crystals, providing more active sites and shortened ion route. Such structures are also beneficial because they relieve the stress and buffer the volume changes caused by pulverisation and aggregation process during redox reaction32,33. Therefore, the structure of Mn2O3 obtained from MnCO3 (S) precursor is likely to enhance Na-ion storage performance.

Figure 4
figure 4

SEM images of (a) Mn2O3 (C600), (b) Mn2O3 (S500), (c) Mn2O3 (S600) and (d) Mn2O3 (S700).

Full size image

The galvanostatic charge/discharge measurements of the Mn2O3 electrodes at a current density of 100 mA g−1 within the potential range of 0.01−3.00 V (vs. Na/Na+) are shown in Fig. 5. At the first cycle, irreversible capacities observed in all electrodes may be attributed to the undesirable growth of a surface passivation layer of solid electrolyte interphase (SEI). The discharge/charge potential profile of the Mn2O3 (S600) electrode was further supported by CV analysis as demonstrated in Fig. S6. However, the SEI formation plateau for the Na/Na+ system is not as sharp/long as compared to the Li-ion cell34,35. Conversely, no distinct plateau observed in the charge/discharge curves after the first cycle, typically seen and consistent with other metal oxides anode in the Na/Na+ system10,12,36,37,38. During cathodic process, Mn and Na2O were observed from the ex-situ XRD patterns (Fig. S7), whereas, re-formation of Mn2O3 was observed during anodic process. It is important to note that the evidence of Mn phase formation is hardly found in the XRD patterns because of significant overlap of this Mn peak with very strong peak from copper (Cu) current collector. Therefore, the formation of Mn and Na2O and the re-formation of Mn2O3 can be expressed by the following electrochemical reversible conversion reaction in Eq. (4).

$${{\rm{Mn}}}_{2}{{\rm{O}}}_{3}+6{{\rm{Na}}}^{+}+6{{\rm{e}}}^{-}\leftrightarrow 2{\rm{Mn}}+3{{\rm{Na}}}_{2}{\rm{O}}$$
(4)
Figure 5
figure 5

Galvanostatic charge/discharge profiles of (a) Mn2O3 (C600), (b) Mn2O3 (S500), (c) Mn2O3 (S600), and (d) Mn2O3 (S700).

Full size image

Figure 6 shows the cycling performance of the Mn2O3 electrodes. Figure 6(a) compares cycling stability and Coulombic efficiency among the electrodes measured at a current density of 100 mA g−1 up to 200 cycles. All electrodes exhibit high initial discharge capacity of 544 mAh g−1 for Mn2O3 (S600), 429 mAh g−1 for Mn2O3 (C600), 331 mAh g−1 for Mn2O3 (S700) and 163 mAh g−1 for Mn2O3 (S500). High initial discharge capacity could be related to the formation of SEI layer and the electrolyte decomposition itself 39. For Mn2O3 (S600) electrode, the discharge capacity increased after the 2nd cycle and then started to decrease after ~15 cycles. Similar trends were observed for Mn2O3 (S700) electrode, where the discharge capacity increased gradually and then started to decrease after 140 cycles, which was possibly due to the activation and stabilisation processes within the electrode40,41,42,43,44. Nevertheless, the capacity depletion behaviour was noticed in Mn2O3 (S500) and Mn2O3 (C600) electrodes. At 2nd cycle, the Mn2O3 (S600) electrode exhibited the highest discharge capacity of 294 mAh g−1 and gradually decreased to 130 mAh g−1 after 200 cycles. For the Mn2O3 (S700) electrode, the discharge capacity was 137 mAh g−1 at 2nd cycle and reached 116 mAh g−1 after 200 cycles. In the case of Mn2O3 (S500) and Mn2O3 (C600) electrodes, the discharge capacity at the 2nd cycle was 243 and 162 mAh g−1, respectively. After 200 cycles, both electrodes had the lowest discharge capacity, i.e. 66 mAh g−1 and 89 mAh g−1. After the initial cycle, all electrodes showed very high Coulombic efficiency of approximately 100% throughout the cycles.

Figure 6
figure 6

(a) Cycling performances and the Coulombic efficiencies up to 200 cycles at 100 mA g−1 and (b) rate capability for the Mn2O3 electrodes.

Full size image

Rate capability of the Mn2O3 electrodes were also measured at different charge/discharge current densities and sustained for 11 cycles for each current density (Fig. 6(b)). At the initial 11th cycle, the Mn2O3 (S600) electrode delivered a high discharge capacity of 207 mAh g−1 at 200 mA g−1. When the current density increased, the electrode exhibited high retention of 166 mAh g−1 at 400 mA g−1, 143 mAh g−1 at 600 mA g−1, 126 mAh g−1 at 800 mA g−1 and 115 mAh g−1 at 1000 mA g−1. Moreover, consecutive cycling performances of Mn2O3 (S500), Mn2O3 (S700) and Mn2O3 (C600) electrodes were not as good as Mn2O3 (S600). Returning to 200 mA g−1 after it had been exposed to different discharge rates, the Mn2O3 (S600) electrode was able to restore the discharge capacity of 197 mAh g−1, which represents above 90% capacity recovery.

All the above results show that cubic-like Mn2O3 demonstrates a possible insertion/deinsertion of Na-ion with a reasonable capacity and cycling stability. The key factor that contributes to the improved performances may be offered by the special morphology of Mn2O3 itself. It is well known that the surface area is proportional to the insertion sites for ions movements45 and this could be ameliorated by downsizing the materials or porous architectures. The Mn2O3 synthesised here is cubic-like particles with nanoparticles (sub-units) embedded on their surfaces which in turn improve the electrochemical performances of the battery. Such Mn2O3 structure can be obtained by thermal decomposition of high quality starting metal source. Thermal decomposition of commercial MnCO3 produced irregular shapes of Mn2O3 whereas, hydrothermally synthesised MnCO3 resulted in cubic-like Mn2O3 particles. It is important to highlight that the use of glucose as reducing agent gives advantages to the precursor for growth in a required direction, thus developing well-crystallised MnCO3 particles via the simple route with good reproducibility. Without using any scarifying template to form porous-like structure, this method is practical for scaling-up production in the industry. Clearly, the electrochemical characteristics of Mn2O3 in Na-ion storage is promising and needs to be further explored. The cubic-like Mn2O3 with nanoparticles on its surface provides more accessible sites for electrolyte penetration into inner Mn2O3 and exposes a large area for Na-ions transportation. Meanwhile, a short ion diffusion path could facilitate the charge-transfer and greatly improve the rate capability of the Na-ions. Additionally, porous-like structure of Mn2O3 could suppress the stress created by volume changes during insertion/deinsertion process. Overall, the electrochemical activity, i.e. synthesis process, discharge capacity and rate capability, demonstrated by Mn2O3 in this study is quite impressive. The performance of Mn2O3 anode can be further improved by controlled synthesis of highly porous nanostructured with high surface area. Such a porous nanostructured needs to be integrated with conductive matrix such as surface carbon coating or hybrid formation with graphite or graphene or carbon nanotubes46,47,48. These carbon materials will not only enhance electrical conductivities of the Mn2O3 electrodes, but also will prevent agglomeration of nanostructured Mn2O3 active materials during repeated cycling, leading to much improved electrochemical performance in terms of capacity, stability, and rate capability. The findings obtained from this research create opportunities for other researchers to explore this material as an anode for Na-ion batteries.

Materials and Methods

Synthesis of MnCO3 and Mn2O3

Scheme 1 shows the synthesis strategy of Mn2O3 from the starting materials. To prepare MnCO3, 0.3 mmol D-(+)-glucose (C6H12O6, Merck Millipore) and 0.3 mmol KMnO4 (Sigma-Aldrich, 97%) were dissolved in 60 ml deionised water at room temperature and stirred to form a homogeneous solution. Then, the homogeneous mixture was transferred into a 125 ml stainless steel autoclave, sealed and heated at 150 °C for 10 h. During the hydrothermal reaction, MnO4− was reduced by glucose and Mn2+ ions generated, leading to the formation of MnCO3 according to the Eq. (5) below49,50.

$$24{{\rm{KMnO}}}_{4}+5{{\rm{C}}}_{6}{{\rm{H}}}_{12}{{\rm{O}}}_{6}=24{{\rm{MnCO}}}_{3}+6{{\rm{K}}}_{2}{{\rm{CO}}}_{3}+12{\rm{KOH}}+24{{\rm{H}}}_{2}{\rm{O}}$$
(5)
Scheme 1
scheme 1

A schematic presentation for the formation of Mn2O3.

Full size image

The precipitates were collected, washed several times with absolute ethanol and deionised water and dried overnight under vacuum. The dried sample is marked as (MnCO3 (S)). To obtain Mn2O3, MnCO3 (S) was calcined at 500, 600 and 700 °C in air for 2 h and denoted as Mn2O3 (S500), Mn2O3 (S600) and Mn2O3 (S700), respectively. For comparison purpose, as-received MnCO3 (Sigma-Aldrich, 98%) marked as MnCO3 (C) was also used in this study. MnCO3 (C) was calcined at 600 °C in air for 2 h and later denoted as Mn2O3 (C600).

Materials characterization

The phase purity and structure of MnCO3 and Mn2O3 samples were determined by X-ray diffraction (XRD, Rigaku Miniflex II) with monochromatic CuKα radiation at a wavelength (λ) of 1.5406 Å. The morphology of the samples was observed through scanning electron microscopy (SEM, JOEL JSM-6360L) and transmission electron microscopy (TEM, TECNAI G2 F20) with an accelerating voltage of 200 kV. The thermogravimetric analysis (TGA) was conducted on Mettler-Toledo thermogravimetric analysis/differential scanning calorimetry (TGA/DSC 1) Stare System at a heating rate of 10 °C min−1 in air. The Fourier transform infra-red (FTIR) spectroscopy was recorded on an IR Tracer-100. Raman spectra were collected on Raman spectroscopy (Renishaw, 532 nm radiation) extended with 0.1 power laser measurement.

Electrochemical measurements

To investigate the electrochemical performances of Mn2O3 samples, the active materials, carbon black (Sigma-Aldrich, >99.95%) and poly(vinylidene fluoride) (PVDF, Sigma-Aldrich), in a weight ratio of 75:20:5 were dissolved in an N-methylpyrrolidone (NMP). The slurry was pasted onto a copper (Cu) foil with an approximate active material loading of ~ 2 mg cm−2. The electrodes were then dried at 100 °C overnight under vacuum. Subsequently, the electrode was cut to 1 cm × 1 cm size. Coin-type cell (CR 2032) was assembled in an Argon-filled glove box (Mbraun, Unilab, Germany) using sodium metal (Sigma-Aldrich, 99.9% trace metals basis) as the counter electrode. A Whatman glass fibre (GF/D) was used as a separator, and the electrolyte 1 M NaClO4 (Sigma-Aldrich, 98%), was dissolved in propylene carbonate (PC) (Sigma-Aldrich, anhydrous, 99.7%) with the addition of 5 wt.% of fluoroethylene carbonate (FEC) (Sigma-Aldrich, 99%). The cycling performance of the electrodes was conducted by Neware battery tester at room temperature.

Conclusion

The cubic-like Mn2O3 was successfully obtained through thermal decomposition of the hydrothermally synthesised MnCO3. For comparison, Mn2O3 obtained through thermal conversion of commercial MnCO3 was also investigated. The synthesis method employed in this study offers a simple and practical approach to industrial production. A microstructure of cubic-like Mn2O3 with nanoparticles (sub-units) embedded on its surface was obtained. The electrochemical results indicate that the Mn2O3 electrode can deliver a promising discharge capacity, cyclability and rate capability during the insertion/deinsertion of Na-ions. The Mn2O3 electrode exhibited high initial discharge capacity of 544 mAh g−1 at 100 mA g−1 and reached 130 mAh g−1 after 200 cycles. The obtained Mn2O3 structure promotes electrolyte penetration into the interior of Mn2O3, provides large sites to facilitate fast ion transportation and thus, expedites the charge-transfer within the electrode. Therefore, the results demonstrate strong evidence for its application in Na-ion batteries.