rev jappan

7/25/2019 Rev Jappan

1/30

This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 129.100.175.145

This content was downloaded on 01/02/2016 at 20:20

Please note that terms and conditions apply.

Recent research progress on iron- and manganese-based positive electrode materials for

rechargeable sodium batteries

View the table of contents for this issue, or go to thejournal homepagefor more

2014 Sci. Technol. Adv. Mater. 15 043501

(http://iopscience.iop.org/1468-6996/15/4/043501)

ome Search Collections Journals About Contact us My IOPscience
http://localhost/var/www/apps/conversion/tmp/scratch_1/iopscience.iop.org/page/termshttp://iopscience.iop.org/1468-6996/15/4http://iopscience.iop.org/1468-6996http://iopscience.iop.org/http://iopscience.iop.org/searchhttp://iopscience.iop.org/collectionshttp://iopscience.iop.org/journalshttp://iopscience.iop.org/page/aboutioppublishinghttp://iopscience.iop.org/contacthttp://iopscience.iop.org/myiopsciencehttp://iopscience.iop.org/myiopsciencehttp://iopscience.iop.org/contacthttp://iopscience.iop.org/page/aboutioppublishinghttp://iopscience.iop.org/journalshttp://iopscience.iop.org/collectionshttp://iopscience.iop.org/searchhttp://iopscience.iop.org/http://iopscience.iop.org/1468-6996http://iopscience.iop.org/1468-6996/15/4http://localhost/var/www/apps/conversion/tmp/scratch_1/iopscience.iop.org/page/terms


2/30

Review

Recent research progress on iron- andmanganese-based positive electrode

materials for rechargeable sodium batteries

Naoaki Yabuuchi1,2,3

and Shinichi Komaba1,2

1 Department of Applied Chemistry, Tokyo University of Science, 1-3 Kagurazaka, Shinjuku, Tokyo,162-8061, Japan2 Elements Strategy Initiative for Catalysts and Batteries (ESICB), Kyoto University, Katsura, Kyoto,

615-8520, Japan

E-mail:[email protected] [email protected]

Received 4 February 2014Accepted for publication 7 July 2014Published 30 July 2014

Abstract

Large-scale high-energy batteries with electrode materials made from the Earth-abundantelements are needed to achieve sustainable energy development. On the basis of materialabundance, rechargeable sodium batteries with iron- and manganese-based positive electrodematerials are the ideal candidates for large-scale batteries. In this review, iron- and manganese-

based electrode materials, oxides, phosphates, uorides, etc, as positive electrodes forrechargeable sodium batteries are reviewed. Iron and manganese compounds with sodium ionsprovide high structural exibility. Two layered polymorphs, O3- and P2-type layered structures,show different electrode performance in Na cells related to the different phase transitionand sodium migration processes on sodium extraction/insertion. Similar to layered oxides,iron/manganese phosphates and pyrophosphates also provide the different framework structures,which are used as sodium insertion host materials. Electrode performance and reactionmechanisms of the iron- and manganese-based electrode materials in Na cells are described andthe similarities and differences with lithium counterparts are also discussed. Together with theseresults, the possibility of the high-energy battery system with electrode materials made from onlyEarth-abundant elements is reviewed.

Keywords: sodium batteries, insertion materials, abundant elements

1. Introduction

The demand for advanced energy storage technology israpidly increasing throughout the world. A large-scale energystorage system for the grid is undoubtedly necessary for theefcient use of electrical energy and for peak shift operation[1]. Such energy storage devices may also be utilized to store

electricity generated from solar cells and wind turbines asgreen energy resources. Over the last two centuries, a numberof energy storage devices based on electrochemicalreactionsas the driving force have been commercialized. Among them,the lead-acid battery is the oldest (invented in 1860 [ 2]), but itis still widely used in our daily life, especially for starting,lighting, and ignition batteries in automobiles. Additionally,lead-acid batteries are used in part for electrical energy sto-rage (EES) [1]. Although lead-acid batteries are durable andreliable energy storage devices, their volumetric/gravimetricenergy density is relatively low among currently commer-

cialized rechargeable batteries. Moreover, some lead com-pounds are known to be extremely toxic to humans, and

| National Institute for Materials Science Science and Technology of Advanced Materials

Sci. Technol. Adv. Mater. 15 (2014) 043501 (29pp) doi:10.1088/1468-6996/15/4/043501

3 Current address: Department of Green and Sustainable Chemistry, TokyoDenki University, 5 Senju Asahi-Cho, Adachi, Tokyo 120-8551, Japan.

Content from this work may be used under the terms of theCreative Commons Attribution-NonCommercial-ShareAlike

3.0 licence. Anyfurther distribution of this work must maintain attribution tothe author(s) and the title of the work, journal citation and DOI.

1468-6996/14/043501+29$33.00 2014 National Institute for Materials Science1
mailto:[email protected]:[email protected]://dx.doi.org/10.1088/1468-6996/15/4/043501http://creativecommons.org/licenses/by-nc-sa/3.0http://creativecommons.org/licenses/by-nc-sa/3.0http://creativecommons.org/licenses/by-nc-sa/3.0http://creativecommons.org/licenses/by-nc-sa/3.0http://creativecommons.org/licenses/by-nc-sa/3.0http://creativecommons.org/licenses/by-nc-sa/3.0http://dx.doi.org/10.1088/1468-6996/15/4/043501mailto:[email protected]:[email protected]


3/30

therefore, lead-acid batteries may present environmentalhazards without a well-established battery recycling systemsuch as those used in developed countries.

Rechargeable lithium batteries (often called Li-ion bat-teries, or LIBs) have become the most successful andsophisticated energy storage devices since their commercia-lization in 1991. LIBs were originally developed as a powersource with high energy density for portable electronicdevices. LIBs are now used as an alternative power source forelectric motors in motor vehicles instead of combustionengines with a fuel tank. Electric vehicles equipped withlarge-scale lithium batteries as power sources have beenintroduced into the automotive market, which may reduce theenergy dependence of future transportation systems on fossilfuels. Although the available energy density is inevitablylower when compared with that of cobalt/nickel-based

layered materials used for small-scale batteries, spinel-typeLiMn2O4[36] is often used in current large-scale batteries.Manganese-based materials, and preferably iron-based mate-rials, are used because of the abundance of Mn in the Earthscrust (gure1). Although lithium batteries with manganese/iron-based materials potentially provide a solution to thetough challenge of achieving sustainable energy develop-ment, we must reconsider the feasibility of lithium [7], whichis the element inherent in lithium batteries. The relativeabundance of lithium in the Earths crust is only 20 ppm, asshown in gure1and table1[8]. Indeed, the cost of materialsis increasing following the commercialization of lithium

batteries [9]. Moreover, lithium resources are unevenly dis-tributed (mainly in South America), and therefore, the

production of LIBs depends on the import of lithium fromSouth America. In contrast, sodium resources are unlimitedeverywhere, as sodium is one of the most abundant elementsin the Earths crust. Vast sodium resources are also found inthe ocean. Additionally, sodium is the second-lightest andsecond-smallest alkali metal next to lithium. On the basis ofmaterial abundance and standard electrode potential,rechargeable sodium batteries (i.e.,Na-ion batteries, or NIBs)are the ideal alternative to LIBs [911].

NIBs are operable at ambient temperature without metallicsodium, which is different from commercialized high-tem-perature sodium-based technology, e.g., Na/S [12] and Na/NiCl2[13] batteries. These batteries utilize alumina-based solid(ceramic) electrolyte and therefore require high-temperatureoperations (300 C) to increase the conductivity of sodiumions in the solid electrolyte. Because molten sodium and sulfurare used as active materials at high temperatures, safety issuesregarding these batteries are not fully resolved for consumerappliances. In contrast, NIBs consist of two different sodiuminsertion materials as positive and negative electrodes withaprotic solvent as electrolyte and therefore are free frommetallic sodium unless unfavorable reactions (e.g., overcharge)cause battery failure. Structures, components, systems, and

reaction mechanisms are essentially the same except thatlithium ions are replaced with sodium ions [10].

Historically, studies of Li+/Na+ ions as charge carriers forelectrochemical energy storage at ambient temperature werebegun before 1980. Electrochemical lithium insertion intoTiS2and its application for energy storage devices were rstproposed in the 1970s [14,15]. Soon after that report, elec-trochemical and highly reversible sodium insertion into TiS2at room temperature was demonstrated in 1980 [16]. Elec-trode performance of lithium cobalt oxide, LiCoO2, which is alithium-containing layered oxide and is still widely used as ahigh-energy positive electrode material in LIBs, was rst

reported in 1980 [17]. Similarly, electrochemical properties ofsodium-containing layered oxides, NaxCoO2, were also

Figure 1.Elemental abundance in the Earths crust [8].

Table 1.Comparison of atomic weight, abundance in the Earthscrust, and ionic radii for several key elements used as electrodematerials for rechargeable batteries.

ion standard atomicweight elemental abun-dance (ppm) Shannon

s ionicradii ()

Li+ 6.94 20 0.76Na+ 23.00 23 600 1.02Fe3+ 55.85 56 300 0.645 (high-spin)Mn3+ 55.94 950 0.645 (high-spin)Co3+ 58.93 25 0.545 (low-spin)

Figure 2.Charge/discharge curves of Li/LiCoO2[17] and Na/NaCoO2[18] cells. A schematic illustration of LiCoO2and NaCoO2is also shown.

2

Sci. Technol. Adv. Mater. 15 (2014) 043501 Review


4/30

reported [18]. The early history of sodium insertion materialswas reviewed in 1982 [19]. Nevertheless, in the last threedecades, signicant research efforts have been undertakenonly for LIBs, and studies of sodium insertion materials forenergy storage became almost nonexistent at one point.

This difference may originate from the fact that availableenergy density was much lower for the Na system comparedwith the Li system. Figure 2 compares typical charge/dis-charge curves of Li/LiCoO2 and Na/NaCoO2 cells. Because

the ionic radius of cobalt is relatively small compared withthose of both lithium and sodium ions, as shown in table1,layered materials with both Li and Na are easily prepared.Although both samples have the same crystal lattice built upby sheets of edge-sharing CoO6 octahedra, the operatingvoltage of LiCoO2for the region at end of discharge (start ofcharge) is >1.0 V higher than that of NaCoO2. When bothcells are charged to >100 mAh g1, the difference in voltagedecreases to approximately 0.4 V, which is similar to thedifference in standard electrochemical potential for Li(3.04 V) and Na (2.71 V). The reduction of voltagebecomes signicant with the increase in the Na content. As a

result, the available energy density of NaxCoO2 as positiveelectrodes is approximately 30% lower than that of LixCoO2

(gure 2). This fact also suggests that different chemistrycompared with the Li system is needed to overcome theproblem of low energy density for NIBs.

In this article, iron- and manganese-based materialsoxides, phosphates, uorides, etcare reviewed as positiveelectrodes for NIBs. Iron and manganese compounds con-taining reversibly extractable sodium ions provide highstructural exibility. Therefore, many materials with differentcrystal structures are easily prepared under thermodynamic

equilibrium conditions, as shown in this article. Moreover,interesting electrode chemistry is found in the compoundscontaining trivalent iron ions in Na cells. For instance, O3-type layered LiFeO2is electrochemically inactive in Li cells,whereas NaFeO2, which possesses the same crystal structure,is surprisingly active in Na cells, as rst reported by Okadaand co-workers [20, 21]. The abundance of iron is animportant factor for NIB electrode materials. A combinationof Na and Fe is attractive as a battery system, especially forEES, and could become competitive with the LiCoO2/gra-phite system, which is the system most widely used in LIBswith high energy density, when innovation in materials (e.g.,

high-voltage materials) for iron-based sodium insertion isachieved.

Figure 3.The classi

cation of Na

Me

O layered materials with the sheets of edge-sharing MeO6octahedra and phase transition processesinduced by sodium extraction. See the text for more details.

3



5/30

2. Layered oxides as Na insertion host materials

2.1. Classification of layered structures

Na-containing 3d-transition metal layered oxides are widelystudied as positive electrode materials for NIBs. The mostcommon layered structures are built up from a sheet of edge-sharing MeO6octahedra. Polymorphisms appear when thesesheets are stacked in different ways along thec-axis. Sodium-based layered materials can be categorized into two maingroups using the classication proposed by Delmas et al[22]:O3 type or P2 type, in which the sodium ions are accom-modated at octahedral and prismatic sites, respectively, asshown in gure3. Schematic illustrations of crystal structuresused in this article were drawn using the VESTA program[23]. O3-type NaMeO2consists of a cubic close-packed (ccp)oxygen array, in which sodium and 3d-transition metal ionsare accommodated at distinct octahedral sites because theionic radius of sodium ions (1.02 ) is much larger than those

of 3d-transition-metal ions in the trivalent state (


6/30

extensively studied with a view to developing iron-basedpositive electrode materials for LIBs [39]. Electrochemicalreversibility of sodium extraction/insertion processes for asingle phase of O3-type NaFeO2 is shown in gure 4(a).

Electrochemical Na extraction from NaFeO2 using a Li/NaFeO2 hybrid-ion cell was rst demonstrated by Takeda

and co-workers [40]. Reversible charge and discharge pro-cesses for a Na/NaFeO2 cell were rst reported by Okadaand co-workers as previously mentioned [20, 21]. Reversi-bility in electrode materials is signicantly inuenced by

cutoff conditions upon charge (sodium extraction) as shownin gure 4. Although charging capacity, corresponding to

Figure 4.(a) Galvanostatic charge/discharge curves of Na/NaFeO2cells [41]. (b) Changes in XANES spectra of Na1 xFeO2at the Fe K-edgeand (c) the radial distribution functions obtained from EXAFS spectra. The area of pre-edge peaks of the Fe K-edge is enlarged in the inset in(b). (d) A proposed mechanism of the iron migration process during sodium extraction.

5



7/30

amounts of sodium ions extracted from the crystal lattice,increases as a function of cutoff voltage, reversible capacityobviously decreases when the material is charged beyond3.5 V. Excellent reversibility with small polarization isobserved with the cutoff voltage of 3.4 V. The observedreversible capacity reaches 80 mAh g1, indicating thatapproximately 0.3 mole of Na is reversibly extracted fromNaFeO2 and inserted into the Na0.7FeO2 host structures.Such deterioration of the electrode properties beyond 3.5 V

originates from the irreversible phase transition as suggestedby ex situx-ray diffraction (XRD) [41] and x-ray absorption

spectroscopy (XAS) in gures4(b), (c). Intensity of the pre-edge peak, which is observed at 7114 eV in x-ray absorptionnear the edge structure (XANES), increases with theincrease in the sodium extraction amount. The intensity ofthe pre-edge peak for Fe3+ is often intensied when ironions are located at tetrahedral sites [42]. In addition, clearchanges in the local environment of Fe are noted from theextended x-ray absorption ne structure (EXAFS). Decrea-ses in length for the rst coordination shell (FeO bond) and

in intensity for the second coordination shell (Fe

Fe bond)are also found. These facts suggest that some of the Fe3+

Figure 5.Comparison of galvanostatic charge/discharge curves of layered NaxMeO2samples with different structures. The morphology of

particles for each sample is shown in the right-hand column.

6



8/30

ions migrate into the tetrahedral sites. Note that because theXAS spectra were collected after discharge to 2.0 V in Nacells, this Fe3+ migration process is irreversible. A proposeddegradation mechanism is summarized in gure4(d). Whensodium ions are extracted from the crystal lattice, vacanciesare created at tetrahedral sites that are face-shared with FeO6octahedra. Because trivalent iron ions are energetically sta-bilized at tetrahedral sites, iron ions easily migrate to theface-shared sites, similar to LiCoxFe1xO2 [43]. Solid-statediffusion of sodium ions is easily disturbed by iron at tet-

rahedral sites, leading to the degradation of electrode prop-erties. As a result, the reversible range of NaxFeO2is limited

to be narrow. This reversible process in the Na cell isexpressed as follows:

+ ++ NaFeO Na FeO 0.3Na 0.3e (1)2reduction

oxidation

0.3 0.7 2

where the open square denotes the vacant sites created in the

structure. The energy density available in terms of electrodematerial is also limited (300 mWh g1 vs Na). Although thereversible range is narrow, the rather small polarizationbetween charge/discharge processes with sub-micrometer-sizeparticles is attractive in terms of electrode material, as shown ingure5. The rate capability of NaxFeO2in Na cells is shown ingure6(a). Such cells can deliver >50% of discharge capacityat 1 C rate. (The C rate is dened as the current that delivers anominal capacity in 1 h, and the nominal capacity here isdened as one-electron redox of iron, i.e., 241 mA g1).Another drawback of NaFeO2as an electrode material is theNa+/H+ ion exchange when NaFeO2is in contact with water

[44]. NaFeO2changes into FeOOH and NaOH (Na2CO3and/or NaHCO3by uptake of CO2). Such Na+/H+ ion exchange is

generally observed in O3-type NaMeO2.Recently some iron-based layered materials have also

been reported with a wider reversible range. In these mate-rials, Ni, Mn, and Co ions are partially substituted for Fe ions[45,47]. The metal substitution effectively changes the phasetransition behavior for NaFeO2. For the pure iron system, thecrystal structure of Na0.5FeO2 was reported as an O3-typephase with monoclinic lattice distortion [48]. O3P3 phasetransition is not evidenced for NaxFeO2, and therefore, suchiron migration to adjacent face-shared tetrahedral sites may beunavoidable. The O3P3 phase transition in Na cells, how-

ever, occurs for the metal-substituted NaFe1xMexO2 sam-ples. Because iron migration to the large prismatic sites isexpected to be unlikely, the formation of the P3 phase caneffectively extend reversible ranges for the electrode materialsin Na cells. Stabilization of the P3 phase has also been the-oretically predicted for Na0.5MeO2 (Me=Ni, Mn, and Co)[49]. Among layered materials with 3d-transition metal ions,the P3-type phase is signicantly stabilized for NaxCoO2[49],and excellent electrode performance has been reported forNaFe1/2Co1/2O2 [46] and NaNi1/3Co1/3Fe1/3O2 [50]. Theseexperimental and theoretical approaches are consistent witheach other, and the substitution of metal ions for iron ions is

an important strategy to further improve the reversibility ofiron-based O3-type layered materials.

2.3. O3-type NaMnO2and P2-type NaxMnO2

O3-type LiMnO2 is also known as a metastable phase[5153], similar to O3-type LiFeO2. The orthorhombic phase(zigzag-type layered phase) crystallizes as a thermo-dynamically stable phase. Although relatively large reversiblecapacity is obtained using O3-type and zigzag-type layeredphases, lowering operating voltage (and changing the voltageprole) during continuous electrochemical cycling is knownto be a disadvantage for electrode materials [5457]. The

voltage prole closely resembles that of spinel-typeLixMn2O4 after electrochemical cycle tests. These LiMn

Figure 6.Rate capability of (a) Na/NaFeO2and (b) Na/Na2/3[Fe1/2Mn1/2]O2cells. The Na/NaFeO2cell was charged (oxidized) to3.4 V at a rate of 12.1 mA g1 and then discharged (reduced) to2.5 V at 1/20 C (12.1 mA g1) to 2 C (484 mA g1) rates. The sampleloading of NaFeO2electrode used was 1.3 mg cm

2 [45]. The Na/Na2/3[Fe1/2Mn1/2]O2cell was charged to 4.2 V at a rate of 13 mA g

1

and then discharged to 1.5 V at different rates of 1/20 C (13 mA g1)to 4 C (1040 mA g1). The sample loading on Al foil was8.4 mg cm2 [67], which is much heavier than that of the NaFeO2electrode.

7



9/30

oxides are composed of the common ccp oxygen lattice.Therefore, two polymorphs of LiMnO2easily transform intospinel as the energetically favorable phase during electro-chemical cycles.

O3-type NaMnO2with a monoclinic lattice (SG C2/m)[26] and P2-type NaxMnO2with an orthorhombic lattice (SGCmcm)[27] are prepared as thermodynamically stable poly-morphs of layered NaMn oxides. Non-distorted P2-NaxMnO2 with a hexagonal lattice (SG P63/mmc) is alsoobtained by controlling synthesis conditions [27]. Here onlylayered NaMn oxides are described, and the structuralcomplexity of NaMn oxides will be further discussed in alater section. Electrochemical reversibility of Na extraction/insertion from/into O3-type NaMnO2 was rst reported in1985 [26]. A very narrow reversible range (x< 0.2 inNa1xMnO2) was, however, reported in this original pub-lication. Electrochemical properties of O3-type NaMnO2were revisited in 2011 [58], and it was found that O3-type

Na1xMnO2 shows a wide reversible range (x


10/30

Na1xMnO2. Although in-plane conduction of electrons/holes in MnO2layers must be the same for both polymorphs,the difference is expected to originate from the manner of in-plane Na-ion conduction between MnO2 layers. Schematicillustrations of in-plane Na-ion conduction in different phasesare compared in gure 7. For the O3-type layered system,

because direct hopping from one octahedral site to an adjacentoctahedral site requires high activation energy, sodium ionsmigrate through interstitial tetrahedral sites (similar to O3-type LixCoO2 [61]), which are face-shared sites with MeO6octahedra in MeO2 layers. According to the results of rst-principles calculations, the diffusion barrier of sodium ions(vacancies) in the O3-type layered framework structure isrelatively small (180 meV for O3-type NaxCoO2) [62]. Thediffusion barrier of sodium ions is expected to be slightlysmaller than that of Li in O3-type LixCoO2(205 meV) [62]. Incontrast with the O3-type layered system, because the P2-typelayered framework has an open path for Na ions, a lowerdiffusion barrier than that of the O3-type phase is expected.Sodium ions migrate from one prismatic site to adjacent sitesthrough open rectangular bottlenecks surrounded by fouroxide ions (and thus smaller repulsive interaction from oxideions is anticipated [63]) because of no interstitial tetrahedralsites dissimilar to the O3 structure. Indeed, higher ionic con-ductivity of P2-type layered materials than of O3-type layeredmaterials is observed for the samples containing similar che-mical compositions for sodium/vacancy [63]. Note that O3P3and P2O2 phase transition may alter the mechanism of in-plane Na-ion diffusion. For instance, according to the resultsofrst-principles calculation, the activation energy of Na-iondiffusion in P2-type Nax[Ni1/3Mn2/3]O2 (2/3>x>1/3) sig-

nicantly increases (>100 meV) in O2-type Nax[Ni1/3Mn2/3]O2(1/3 >x> 0) [64] even though the P2O2 phase transition isunavoidable for Nax[Ni1/3Mn2/3]O2in Na cells. On the otherhand, O3-type Nax[Fe1/2Co1/2]O2 shows excellent rate cap-ability in Na cells, which may originate from the fast ionconduction for the P3 phase formed by the O3P3 phasetransition in the desodiation process [46]. Note that the localenvironment of Na ions at prismatic sites and the diffusionpathway in the P2 and P3 phases are different, as shown ingure7. It is expected, therefore, that the mobility of Na ionsin both phases may be different. Although an NMR study onP2 and P3 NaxCoO2suggested that the mobility of Na ions inboth phases clearly differs on the NMR time scale [65], thedifference in Naion mobility in terms of electrode materials isnot known so far. Further systematic studies of the diffusionprocess of Na ions in these different phases are necessary.

2.4. O3-type Na[Fe1/2Mn1/2]O2and P2-type Nax[Fe1/2Mn1/2]O2

The electrode performance of P2-NaxMnO2 is promisingbased on reversible capacity as shown in the precedingsection. The average operating voltage based on the Mn3+/Mn4+ redox couple is, however, limited to less than 3 Vversus metallic Na. The average operating voltage of O3-NaFeO2 with an Fe

3+/Fe4+ redox couple is much higher(3.23.3 V) than that of P2 Na

x

MnO2, even though the

Figure 8.XAS analysis of the P2-type Nax[Fe1/2Mn1/2]O2samplescharged to 3.8 and 4.2 V: iron K-edge XANES (a) and radialdistribution of the iron ion (b). (c) Comparison of the Mssbauerspectra of the as-prepared and the fully charged samples. Observedspectra are shown as black dots and deconvoluted spectra as coloredlines. Here s and d represent sextet and doublet, respectively. Thepurple curve corresponds to the sextet component. This minor sextetcomponent can be assigned to neither Fe3+ nor Fe4+. Instead, thiscomponent could be assigned to Fe5+ because of a relatively lowisomer shift. Fe5+ probably originates from the charge dispropor-tionation of Fe4+ (Fe4+Fe5+ + Fe3+) as reported in the literature[158,159].

9



11/30

reversible range associated with irreversible iron migration isnarrower. From these pieces of experimental evidence, P2-type NaxFeO2 is predictably considered to be a possiblecandidate for high-capacity, high-energy electrode material.All of our trials, however, failed because Fe4+ cannot bestabilized in the oxide-ion framework under ambient condi-

tions [66]. Therefore, manganese ions were partially sub-stituted for iron to stabilize the P2-type phase, and we nallysucceeded in the synthesis of P2-type Fe-based layered oxide,Na2/3[Fe1/2Mn1/2]O2 [67]. P2-Na2/3[Fe1/2Mn1/2]O2 can beeasily prepared via solid-state reaction of Na2O2(or Na2CO3),Fe2O3, and Mn2O3 at 900C for 12h [67]. The electrodeperformance of P2-type Na2/3[Fe1/2Mn1/2]O2 in the voltagerange of 1.54.2 V is also shown in gure 5. P2-Na2/3[Fe1/2Mn1/2]O2 delivers large reversible capacity, similar toP2-NaxMnO2, in the Na cell. The reversible capacity (rever-sible range) of P2-Na2/3[Fe1/2Mn1/2]O2 is much higher(wider) than that of O3-NaFeO2. The rate capability of P2-Na2/3[Fe1/2Mn1/2]O2is also better than that of O3-NaFeO2ascompared in gure6. Additionally, the operating voltage isalso increased compared with P2-NaxMnO2; this will bediscussed in a later section.

O3-type Na[Fe1/2Mn1/2]O2 can be also prepared bychanging the ratio of sodium/(iron and manganese) [67]. Theelectrode performance of O3-Na[Fe1/2Mn1/2]O2in the Na cellis shown in gure 5. Large polarization (>1 V) betweenoxidation (charge) and reduction (discharge) processes isobserved when the Na cell is cycled in the voltage range1.54.2 V. In comparison with the O3 phase, the P2 phaseclearly shows larger reversible capacity. Because the P2 andO3 phases have similar particle morphology (gure 5), the

kinetic limitation related to the diffusion length is notresponsible for the difference in performance. Researchinterest in the NaFeMn system as high-capacity electrodematerial is rapidly increasing, and now the electrode perfor-mance for P2- and O3-Nax[FeyMn1y]O2with different che-mical compositions has been determined [6870].

Charge compensation mechanisms in the sodiumextraction process were further analyzed by XAS. XANESspectra of P2-Nax[Fe1/2Mn1/2]O2 at the iron K-edge areshown in gure 8(a). When Nax[Fe1/2Mn1/2]O2 is oxidizedfrom 3.8 V to 4.2 V, a shift in the iron K-edge absorptionspectrum is found in the higher-energy region. The shift maybe attributable to complicated situations, including changes inthe local structuresfor example, sodium extraction from theiron face-shared sites and phase transitions [67]. Therefore,EXAFS spectra during the sodium extraction were furtheranalyzed. The change in the radial distribution around ironduring oxidation to 4.2 V is shown in gure8(b). The radialdistribution around iron is not affected by oxidation to 3.8 V,and the interatomic distance of FeO remains unchanged(2.00 for the as-prepared and 1.99 for the 3.8 V chargedstate). The Fe-O local environment is markedly changed afteroxidation to 4.2 V. The intensity for both rst and secondcoordination shells, FeO and FeFe(Mn), is reduced, indi-cating distortion around Fe. This distortion also inuences the

local structure of neighboring manganese ions [67]. Theinteratomic distance of FeO is clearly shortened after

oxidation to 4.2 V (1.90 (1) ). A large DebyeWaller factorindicates distortion by a non-cooperative JahnTeller effect ofhigh-spin Fe4+ (t2g

3 eg1). Mssbauer spectroscopy further sup-

ports the oxidation of Fe3+ to Fe4+, as shown in gure8(c).The oxidation of Fe3+ was also reported for O3-NaFeO2 byMssbauer spectroscopy [20, 21]. The results indicate that

Fe3+

in P2-Nax

[Fe1/2Mn1/2]O2 is electrochemically active inthe sodium system based on the Fe3+/Fe4+ redox. Note thatO3-LiFeO2 is never electrochemically active on the basis ofthe Fe3+/Fe4+ redox [38], or the reaction seems to be domi-nated by the oxygen removal process at the solid/electrolyteinterface. Indeed, nanometer-sized O3-type LiFeO2 is elec-trochemically active in association with the Fe2+/Fe3+ redox[71]. As the Fe3+ 3d-orbital is strongly hybridized with theoxygen 2p orbital in the Li system, oxygen removal isfavorable rather than oxidation to Fe4+, similar to theLi2MnO3 system. In general, the sodium system shows alower redox potential than that of the lithium system, asshown in gure 2. Sodium ions are strongly ionized whencompared with lithium ions, resulting in lower covalency withoxygen. Iron and oxygen gain more (net) electrons whencompared with the lithium system. As a result, the electro-chemical potential of the Fe3+/Fe4+ redox is relatively low,approaching that of Na/Na+. Thus, the Fe3+/Fe4+ redox in P2-Nax[Fe1/2Mn1/2]O2 may be accessible without oxygen lossand may be used as electrode material with relatively highoperating voltage in the Na system. Note that tetravalent ironprovides the strong hybridization character with the oxygen2p orbital. Indeed, charge transfer from oxygen to iron hasbeen proposed for SrFeO3 [72]. The possibility of such acharge transfer process (namely, the formation of a hole in the

oxygen 2porbital instead of the formation of Fe4+

) cannot benegligible for P2-Nax[Fe1/2Mn1/2]O2. Further systematicstudy is needed to understand the charge compensation pro-cess in greater detail.

As described previously, the available reversible capacityof P2-Nax[Fe1/2Mn1/2]O2reaches 190 mAh g

1 with an aver-age voltage of 2.75 V versus sodium metal. The energydensity is estimated to be 520 mWh g1 vs Na, which iscomparable to that of LiFePO4 (about 530 mWh g

1 versusLi) and slightly higher than that of LiMn2O4 (about450 mWh g1). The density of P2-Nax[Fe1/2Mn1/2]O2is esti-mated from x-ray diffraction to be 4.1 g cm3, which is higherthan that of LiFePO4 (3.6 g cm

3), and the submicrometer-sized primary particles are electrochemically active withoutcarbon coating. In addition, the many voltage plateausobserved for P2-/O3-NaxMnO2 are not observed for P2-Nax[Fe1/2Mn1/2]O2 (gure 5). The appearance of voltageplateaus is closely correlated with in-plane sodium/vacancyordering (presumably coupled with charge ordering in MnO2layers). Stabilization of the Na/vacancy ordering also resultsin bi-phasic reactions (two phases with, generally, differentunit cell volume coexist in a single particle) in the region ofvoltage plateaus. Rate capability in terms of electrode materialis expected to be restricted because of the phase boundarymovement for two-phase regions [46]. The partial Fe sub-

stitution for Mn reduces the tendency of in-plane sodium/vacancy ordering. These characteristics are benecial for

10



12/30

increasing the energy and power density of the electrodematerials in sodium batteries.

Although the P2-type NaFeMn system is a promisingcandidate as a high-capacity positive electrode material forNIBs, three major drawbacks are known: (1) large volumechange (11.3% shrinkage after charge to 4.2 V) during

electrochemical cycles, (2) the system

s hygroscopic char-acter, which restricts sample handling in moist air (as-pre-pared P2-Na2/3[Fe1/2Mn1/2]O2 is somewhat oxidized bywater, forming P2-Na1/2[Fe1/2Mn1/2]O2 and NaOH), and (3)sodium deciency in the as-prepared sample. In the rstdischarge process, the sodium ions are in part insertedbeyond the starting composition of x=2/3 in Nax[Fe1/2Mn1/2]O2. The excess amount of sodium is estimated to be 0.2(50 mAh g1), forming P2-Na0.86[Fe1/2Mn1/2]O2 in thefully discharged state [67]. It is difcult to directly preparethis sodium-rich phase because this phase is metastable.The O3 and/or O3 phases are thermodynamically stable.Compensation for the decient sodium ions is, therefore,needed to design the NIBs and to use the full capacity ofP2-Nax[Fe1/2Mn1/2]O2. One idea to compensate for thesodium ions is the addition of sacricial salts. For instance,NaN3electrochemically decomposes into sodium ion and N2gas, and this reaction is irreversible. Mixing P2-Na2/3[Fe1/2Mn1/2]O2 with NaN3, therefore, increases the initial chargecapacity and thus effectively compensates for the sodiumdeciency for the P2 phase in the initial state [73]. Furtheroptimization is clearly needed to overcome these dis-advantages in terms of electrode material and to develophigh-energy NIBs on a large scale in the future.

3. Other oxides as Na insertion host

Two layered polymorphs, O3- and P2-type layered oxides,are summarized in the preceding section. In addition to theselayered materials, many different non-layered oxides havebeen studied as electrode materials. Table 2 summarizes theFe/Mn-based oxides that have been studied as positive elec-trode materials for rechargeable Na batteries, and the struc-tural data and electrode performance of Li counterparts arealso compared. In this section, non-layered oxides as thesodium insertion host are reviewed.

3.1. Iron-based oxides

NaFeO2 containing trivalent iron ions crystallizes into twodifferent polymorphs. The electrode performance of the low-temperature phase, -type (O3-type) NaFeO2, is described inan earlier section (gure4). Synthesis at higher temperatures(>760 C) easily results in phase transition into -typeNaFeO2 as the high-temperature polymorph. The crystalstructure of-NaFeO2is related to the mineral wurtzite. Thepacking of oxide ions is a hexagonal close-packed array, andboth sodium and iron ions are located at tetrahedral sites(gure 9(a)) [74]. The electrode performance of -NaFeO2

was examined in Na cells. Charge/discharge curves of a Na/-NaFeO2cell are shown in gure10(a). Beta-NaFeO2seems to

be electrochemically inactive as electrode material in thevoltage range 2.04.0 V versus Na metal. From these results,it is speculated that Fe3+ at tetrahedral sites is difcult tooxidize to the Fe4+ state, which is energetically stable only atoctahedral sites.

Inverse spinel-type Fe3O4 (and -type Fe2O3) and cor-

undum-type (-type) Fe2O3 were also examined as hostmaterials for sodium insertion. Reduction in particle size tothe nanoscale often activates the electrode reversibility oftransition metal oxides, which had been thought to be inactivein Li cells [75,76]. Similar to the Li system, nanosized Fe3O4and Fe2O3are reported to be electrochemically active in Nacells [77, 78], even though micrometer-sized Fe3O4 (andother oxides) are electrochemically inactive. The galvano-static discharge/charge curves of the Fe3O4 samples withdifferent particle sizes are compared in gures 11(a)(c).Electrochemical reversibility in 1.0 mol dm3 LiClO4 dis-solved in propylene carbonate as electrolyte is highlyimproved as the particle size of Fe3O4decreases. The rever-sible capacity is less than 10 mAh g1 for the submicrometer-sized Fe3O4, whereas the rechargeable capacity reaches190 mAh g1, which corresponds to the 1.6 mole of Li inFe3O4, for the nano-sized Fe3O4 particles prepared by aprecipitation method (gure11(i)). Similar to the Li system,the nanocrystalline Fe3O4powder is electrochemically activein the NaClO4electrolyte, as shown in gures11(d)(f). The400 nm Fe3O4sample is electrochemically inactive in the Nasystem, similar to the Li system. For the nanocrystallineFe3O4 powder, the reversible capacity increases to170 mAh g1. Although the reversible capacity of the 10 nmsample is slightly smaller than that of the LiClO4electrolyte,

acceptable capacity retention for the 30-cycle test wasobserved in the NaClO4 electrolyte. The electrochemicalreactivity of the Fe3O4powder is activated by controlling thesize of particles for both Li and Na systems. Recently it wasdemonstrated that a binder-free nanocomposite with carbonnanotube and nano iron oxides shows excellent electrodeperformance as electrode material in Na cells [79].

3.2. Manganese-based oxides

Early studies of NaMnO ternary compounds as the sodiuminsertion host were published in the 1980s [26]. The structuralchemistry of NaMnO ternary compounds is complicated,and many different phases, including polymorphs, areobtained as the atomic ratio of Na to Mn changes [80]. Twodifferent NaMn layered phases, O3-type NaMnO2and P2-type NaxMnO2(x= ca. 0.7), are described in an earlier section.Two polymorphs are known for NaMnO2: -type (O3-type)NaMnO2, the low-temperature phase; and -type NaMnO2,the high-temperature phase [80]. Beta-NaMnO2is prepared at>900 C, and the crystal structure is different from-NaFeO2,with its wurtzite-related structure. The packing of oxide ionsfor -NaMnO2 is the same as for -NaFeO2 with the ccplattice, and distribution of Na and Fe at octahedral sites in theccp lattice is different. The crystal structure of -NaMnO2

is the same as orthorhombic LiMnO2, the so-called zigzag-type layered phase (gure9(c)). The electrode performance

11



13/30

Table 2.Overview of Fe/Mn-based oxides studied as positive electrode materials for rechargeable Li/N

Li

system materialspacegroup electrode performance remark material

spacegroup

Fe(III) -LiFeO2 Fm-3m inactive -NaFeO2 R-3m

-LiFeO2 C2/c inactive -NaFeO2 Pna21 -LiFeO2 I41/amd inactiveMn(III) LiMnO2 Pmmn 130200 mAh g

1

ca. 3 Vphase transition into spinel on electro-

chemical cycle-NaMnO2 C2/m -NaMnO2 Pmmn

Mn(III)+Mn(IV)

LiMn2O4 Fd-3m 110120 mAh g1

4.1 VNa2/3MnO2 P63/mmc Na0.44MnO2 PbamNa0.4MnO2 C2/m

12


14/30

of -NaMnO2 in a Na cell was also reported [26]. Thepotential difference between charge/discharge processes in aNa cell appears to be relatively small, similar to -NaMnO2(gure10(b)).

When the fraction of sodium to manganese ions isreduced, two different phases, Na0.44MnO2 and Na0.4MnO2,are obtained. The Na insertion properties of Na0.44MnO2(Na4Mn9O18) were rst examined in the 1990s at 85 C withsolid-state polymer electrolyte [81,82] and revisited in 2007to be examined at room temperature with aprotic solvent [83].A crystal structure of Na0.44MnO2is also shown in gure9. Aframework structure of Na0.44MnO2 (SG Pbam) consists offour MnO6octahedra and one MnO5square pyramid. Triva-

lent manganese ions are energetically stabilized at the square-pyramidal sites. These octahedral and square-pyramidal sites

are connected to each other by either edge or corner sharing(vertex sharing), forming the complicated framework struc-ture with two different tunnels along the c-axis direction.Sodium ions are accommodated at three different sites in theframework structure, and the sodium ions in both large andsmall tunnels are highly mobile [84]. However, trivalentmanganese ions at the square-pyramidal sites cannot be oxi-dized to the tetravalent state [84], and thus 20% of the sodiumions cannot be extracted from the framework structure. Dur-ing the discharge (reduction) process, NaxMnO2is reversiblyuptaking sodium ions to form Na0.67MnO2, and approxi-mately 120 mAh g1 of reversible capacity is obtained aselectrode material. Complicated phase transitions occur in the

insertion/extraction of sodium ions, as expected from thegalvanostatic charge/discharge curves in gure 10(c) [83].

Figure 9.Crystal structures of miscellaneous iron/manganese-based oxides: (a) -NaFeO2, (b) Fe3O4/-Fe2O3, (c) -NaMnO2,(d) Na0.44MnO2, (e) -MnO2, and (f) Na0.4MnO2.

13



15/30

Figure 10.Electrode performance of miscellaneous iron/manganese-based oxides: (a) -NaFeO2, (b) -NaMnO2[26], and (c) Na0.44MnO2[83]. (d), (e) Nanowires of Na0.44MnO2are highly active as electrode material in the Na cell [85]. (f) Charge/discharge curves of Na0.4MnO2in the Na cell [24]. (b), (f) Reproduced with permission from [26]. Copyright 1983 Elsevier. (c) Reprinted with permission from [83].

Copyright 2007 American Chemical Society. (d), (e) Reproduced with permission from [85]. Copyright 2011 John Wiley and Sons.

14



16/30

Detailed congurations of sodium ion ordering in NaxMnO2in electrochemical cycles have been studied by rst-principlescalculations [84]. Recently well-crystallized, uniform nano-wires (diameter 50 nm and growth orientation along [001])of Na0.44MnO2 were prepared (gure10(d)) by a polymer-pyrolysis method [85]. The Na0.44MnO2 nanowires demon-strate relatively good rate capability and excellent capacityretention, as shown in gure 10(e). Approximately100 mAh g1 of reversible capacity is obtained in the Na cells,even after 100 cycles.

Other manganese (di)oxides studied for lithium insertionmaterials, such as-MnO2and-MnO2, have also been testedas sodium insertion host materials [86]. Framework structuresof both-MnO2and -MnO2consist of MnO6octahedra that

are connected by both edge and corner sharing. The structure of-MnO2is classied as rutile-type with relatively small 1 1tunnels along thec-axis direction, which is built up with chainsof edge-shared MnO6 octahedra. Alkali ions, even smalllithium ions, cannot be inserted into thesmall tunnels of bulk-MnO2[87]. However, nanosized -MnO2with ordered meso-pores is electrochemically active in Li cells [87]. Large rever-sible capacity of >250 mAh g1 is obtained by using nanosized-MnO2. Although the crystalline phase is lost (changes into anamorphous phase) following electrochemical cycles, the elec-trode performance of -MnO2 is activatedby using such nano-engineered samples. Therefore, it is

also expected that this methodology can be used for Na inser-tion materials, similar to iron oxides, as described in an earlier

section. Indeed, recently it was reported that-MnO2nanorodsprepared by a hydrothermal method with a diameter of 100 nm,which preferably grow along [001], deliver large reversiblecapacity (>200 mAh g1) in a Na cell. Because the radius ofsodium ions is much larger than that of lithium ions, phasetransitions from the rutile structure are expected to take place inNa cells.Althoughlarge reversiblecapacity is obtained with the-MnO2nanorods, note that the difference in voltage betweenthe charge/discharge in Na cells is much larger compared withNa insertion reactions in highly crystallized particles based onthe topotactic reaction as shown in gure5.

Alpha-type MnO2, of the group of minerals called hol-landite, has been studied in a Na cell. The structure of -MnO2shown in gure9(e) is built up with edge- and corner-

shared MnO6 octahedra, similar to -MnO2. Large 22tunnels are formed with double chains of edge-shared MnO6octahedra. The open tunnels in -MnO2 are stabilized byincorporation of large cations, such as K+, and sodiummigration in the open diffusion path is expected to be ef-cient. Although -MnO2 nanorods show large reversiblecapacity of >200 mAh g1 in a Na cell, polarization betweencharge/discharge curves appears to be large, similar to -MnO2nanorods [86]. Results of the density functional theory(DFT) calculation suggest that the migration barrier of Naions in-MnO2is comparable to or slightly smaller than thatof Li ions [88].

Manganese oxides with even more open paths for Namigration, 2 3 tunnels consisting of double and triple MnO6

Figure 11.Galvanostatic discharge and charge curves of Fe3O4composite electrodes, which consisted of particles of (a), (d) 400 nm, (b), (e)100 nm, and (c), (f) 10 nm in diameter, at a rate of 20 mA g1 [77]. Electrolytes used were (a)(c) 1.0 mol dm3 LiClO4and (d)(f) NaClO4dissolved in a mixture of ethylene carbonate:dimethyl carbonate (1:1 by volume). For the characterization of Na insertion into iron oxides, athree-electrode cell with lithium metal as a reference electrode was used to compare the electrochemical potential directly with the lithiumsystem. Particle morphology observed via scanning electron microscopy is also shown in (g)(i).

15



17/30

chains, are found in a mineral called psilomelane, as shown ingure 9(f) [89]. Large Ba2+ ions and water molecules areincorporated as a template to stabilize such large tunnels inpsilomelane. A Na+ sample substituted for Ba2+ and H2O inpsilomelane, Na0.4MnO2, is also prepared by a simple solid-state method [26,80]. The sample shows rather small polar-ization in a Na cell, even though the mechanism of sodiuminsertion into Na0.4MnO2 with 23 tunnels is not fullyunderstood. At least 0.3 mole of sodium ions are reversiblyinserted/extracted into/from the framework structure of psi-lomelane with the large tunnels [26].

3.3. Spinel-type manganese oxides

Spinel-type LiMn2O4 is the one of the most widely studiedmaterials in terms of positive electrodes for LIBs [36].Layered O3-type LiMnO2which is also prepared by a Na/Liion-exchange method from O3-NaMnO2, is isostructuralwith O3-NaMnO2 [51, 52]. Although the electrode perfor-mance of O3-LiMnO2in Li cells has also been reported, itsmajor problem as an electrode material is the gradual phase

transition to spinel with the common ccp oxygen lattice.Trivalent manganese ions should be highly mobile in the ccp

lattice, and therefore, LiMnO2easily transforms into spinel asthe energetically favorable phase during electrochemicalcycles in Li cells.

In contrast, a Na counterpart, O3-NaMnO2, can be usedas electrode material without such a phase transition intospinel. The difference probably originates from a large gap insize between Na and Mn ions. A three-dimensional frame-work structure such as that of spinel is stable for the Li systemand not for the Na system. The question arises here whetherspinel-type manganese oxides are used as electrode materialsin Na cells without phase transition [90, 91]. Figure 12compares charge/discharge curves of stoichiometric LiMn2O4cycled in 1.0 mol dm3 lithium or sodium perchlorate dis-solved in propylene carbonate. LiMn2O4 shows two voltageplateaus at approximately 3.92 and 4.16 V in a Li cell in a rstcharge process. Although the almost identical voltage proleis observed in a Na cell, the voltage of the Na cell is 0.34 Vlower than that of the Li cell. This observation directly relatesto the difference in the standard electrode potential, as simi-larly shown in gure2. As shown in gure12(a), a voltageprole upon the rst charge in the Na cell is superimposed onthat in the Li cell when voltage is shifted higher by 0.34 V.These results clearly suggest that the same reaction proceeds

in both Na and Li systems, i.e., lithium extraction fromLiMn2O4, forming LixMn2O4(x= ca. 0.1). A clear difference

Figure 12.Charge and discharge curves of LiMn2O4in lithium andsodium cells with propylene carbonate electrolyte solution contain-ing 1.0 mol dm3 LiClO4and NaClO4, respectively: (a) rst and (b)tenth charge/discharge curves.

Figure 13.X-ray diffraction (XRD) patterns of LiMn2O4electrodescycled in Na cells: (a) as-prepared (a = 8.247 (1) ) and after (b) rstcycle, (c) fth cycle, and (d) tenth cycle [91]. An XRD pattern ofNaMnO2(a = 5.66 (1),b = 2.858 (5),c = 5.80 (1) , and=113.1)is also shown in (e). Asterisks denote diffraction lines from astainless steel mesh used as a current collector. Dotted lines areshown as a visual guide. Schematic illustrations of the phasetransition process are also shown at bottom.

16



18/30

between the two cells is found via a discharge (reduction)process. In the Li cell, Li ions are reversibly inserted into thespinel framework structure with similar voltage proles oncharge. In the Na cell, voltage rapidly decreases to 3.2 Vversus Na/Na+, at which time a long plateau is observed(approximately 100 mAh g1 of discharge capacity), and then

the voltage gradually decreases to 2.0 V. First-dischargecapacity reaches approximately 200 mAh g1 in this experi-mental condition. Although lithium ions are extracted in Nacells, the extracted lithium ions are not reversibly inserted intoNa cells, and instead Na ions seem to be inserted intoLixMn2O4(x= ca. 0.1). Figure12(b) shows tenth charge/dis-charge curves of LiMn2O4 electrodes cycled in Li and Nacells. Voltage proles change during the electrochemicalcycling in the Na cell. For the tenth cycle, the voltage prolesare completely different for both cells, suggesting a phasetransition from the spinel phase to another phase in the Nacell. The changes in crystal structures of stoichiometric

LiMn2O4 cycled in Na cells were examined by XRD.Figure13 shows the XRD patterns of LiMn2O4 electrodesbefore and after the electrochemical cycling in Na cells. Whenthe electrodes are cycled in Na cells, the XRD patterns ofelectrodes drastically change. Peak proles of all the Bragglines from LiMn2O4 are signicantly broadened, which isindicative of the phase transition induced by sodium insertioninto the spinel phase. After 5 cycles in the voltage range2.04.0 V versus Na/Na+, new peaks appear at 17 degrees (intwo theta) with the appearance of some additional peaks.After 10 cycles in the range 2.04.0 V, these new peaksbecome more visible. These new peaks observed after 10cycles resemble those of O3-type NaMnO2. From the XRDpatterns, newly formed O3-type NayMnO2 seems to coexistwith the peak broadened spinel-type phase. Because somelithium-ions still exist in the spinel phase by electrochemicaloxidation to 4.0 V in Na cells, the phase transition to thelayered phase may be partially suppressed. Such spinel-to-layered phase transition in common oxygen lattices (sche-matic illustrations are shown in gure13) is different com-pared with the Li system, and the results suggest that thelayered sodium manganese phase seems to be energeticallystable, as recently supported by the results of rst-principlescalculation [49]. Stabilization of layered manganese systemsis another advantage of the Na system for battery applications.

4. Polyanionic compounds as Na insertion host

Similar to layered and non-layered oxides, the high structuralexibility of sodium-containing polyanionic compounds is anattractive feature when it comes to electrode materials for abattery system. Therefore, many different polyanionic com-pounds have been revisited or newly found in the last fewyears, as summarized in table3. In this section, Fe/Mn-basedpolyanionic compounds as the Na insertion host are reviewed.

4.1. Phosphates with Fe(II) and Mn(II)

Because layered lithium iron(III) oxide LiFeO2 showsunimpressive battery performance based on the Fe2+/Fe3+

redox couple after oxygen loss in the Li cells as previouslydiscussed, polyanionic compounds of iron have been exten-sively studied based on the use of the Fe2+/Fe3+ redox couple[92]. Lithium iron(II) phosphate, LiFePO4, is the most widelystudied polyanionic compound in terms of positive electrodematerials for LIBs because of the interest with regard topractical applications [93]. A crystal structure of LiFePO4isclassied as the triphylite-type structure (SG Pnma) with adistorted hcp oxygen lattice. The structure is closely related tothe olivine-type structure, and two cations (lithium and ironions in the case of LiFePO4) are located in two distinctoctahedral sites in the common framework structure com-posed of XO4(X = Si, P, Mo, etc) tetrahedra. Phosphate ions,(PO4)

3 tetrahedra, share one edge and four corners withFeO6 octahedra, forming the framework structure of triphy-

lite-type FePO4, as shown in gure14(a). Formation of edge-shared sites between FeO6 octahedral and PO4 tetrahedralunits relatively stabilizes the energy of the Fe2+/Fe3+ redoxcouple through the inductive effect, resulting in high voltage(3.45 V versus Li metal) for iron-based compounds based onthe Fe2+/Fe3+ redox [39, 93]. Although the FeO6 octahedrashare corners in the triphylite-type structure (gure 15(a)),electrical conductivity is still too low for electrode materials.The electrode performance of LiFePO4 is signicantlyimproved by carbon coating [94] and shortening of a Li dif-fusion path with nanosized particles. Such engineered nano-sized particles deliver large reversible capacity in Li cells,

which now almost reaches its theoretical limit (170 mAh g1

).Conduction mechanisms of Li ions in triphylite-type LiFePO4have been extensively studied [95, 96]. LiO6 octahedra areconnected to one another by edge sharing along, [010]forming 1D chains, as shown in gure15(c). Li ions migratewith a low diffusion barrier (150 meV [62]) along the 1Dchains throughout particles if the formation of antisite defectsbetween Li and Fe sites is negligible [96].

A sodium counterpart, NaFePO4, is also studied as anelectrode material for NIBs [97]. NaFePO4crystallizes into amaricite-type structure as the thermodynamically stablephase. Its crystal structure is also closely related to the oli-vine-type structure with a distorted hcp oxygen lattice. Alarge gap in the size of ionic radii between Na and Fe resultsin signicant distortion in the hcp oxygen lattice(gure15(b)) in comparison to triphylite-type LiFePO4. Ironions are located at octahedral sites, and FeO6octahedra shareedges with one another, forming 1D chains as shown ingure15(d). The structure of the 1D chains of FeO6 in themaricite phase is essentially the same as that of the 1D LiO 6chains in the triphylite phase. Sodium ions are located at largetetrahedral sites (note that such a site is also regarded as anirregular site coordinated by 10 oxygen within 3 [98]),which share corners with PO4 tetrahedra (gure 15(b)).Because sodium sites are isolated in the structure, as shown in

gure15(b), a large barrier to sodium migration is expected inthe maricite-type NaFePO4. Indeed, the reversibility of Na

17



19/30

Table 3.Overview of phosphate-related and uoride compounds containing Fe(II), Fe(III), and Mn(II), studied as positive electrode m

Li

system material space group electrode performance remark material spac

LiFePO4 Pnma 160 mAh g1 3.4 V triphylite

typeNaFePO4 Pnm

Fe(II) Na2FeP2O7 P-1Li2FeP2O7 P21/c 130 mAh g

1 3.3 V Na4Fe3(PO4)2(P2O7) PnaNaFePO4F Pbc

Li3Fe

2(PO

4)3

P21

100 mAh g1 2.9 V Na3Fe

2(PO

4)3

C2/Fe(III) Na3Fe3(PO4)4

phosphates-related

LiFeP2O7 P21 60 mAh g1 2.9 V Na3Fe2(PO4)2F3 Pmn

NaFeP2O7 P21Fe(III) +

Mn(II)unknown NaMn2Fe(PO4)3 C2/c

LiMnPO4 Pnma 140 mAh g1 4.0 V triphylite

typeNaMnPO4 Pnm

Mn(II) Li2MnP2O7 P21/c inactive Na2MnP2O7 P-1NaMnPO4F P21

uoride Fe(II) unknown NaFeF3 Pnm

18


20/30

Figure 14.Crystal structures of phosphate-based compounds containing Fe(II) ions: (a) Na(Li)FePO4 (triphylite-type), (b) NaFePO4(maricite-type), (c) Na2Fe(P2O7), and (d) Na4Fe3(PO4)2(P2O7).

Figure 15.Crystal structures of Na(Li)FePO4(triphylite-type) and NaFePO4 (maricite-type): the arrangement of (a) FeO6-PO4 in triphyliteand (b) NaO4-PO4in maricite, and 1D chains of (c) LiO6octahedra in triphylite and (d) FeO6octahedra in maricite.

19



21/30

extraction/insertion for maricite-type NaFePO4 seems to beunacceptable in terms of electrode material [97,99].

In contrast, triphylite-type NaFePO4, which is a meta-stable polymorph of NaFePO4, is electrochemically active[97, 100]. Triphylite-type NaFePO4 can be prepared by anion-exchange method from LiFePO4. Li ions are chemically

(and electrochemically) extracted from triphylite-LiFePO4without the destruction of its core structure, forming hetero-site-type FePO4. Chemical (and electrochemical) Na insertioninto heterosite-type FePO4, which possesses the same fra-mework structure as triphylite-LiFePO4, results in the for-mation of triphylite-type NaFePO4. Triphylite-type NaFePO4is stable below 480 C in inert atmosphere and transforms intomaricite-NaFePO4 by further heating above 480 C[100,101]. Charge/discharge (oxidation/reduction) curves oftriphylite-NaFePO4 in a Na cell [100] are shown ingure 16(a). Nearly one mole of Na ions are reversiblyinserted/extracted into/from triphylite-NaFePO4. Because theionic radius of the sodium ion is much larger than that of thelithium ion, the unit cell volume of triphylite-NaFePO4 isapproximately 10% larger than that of triphylite-LiFePO4.Two voltage plateaus are observed on discharge(gure 16(a)). Open-circuit voltage on the plateaus isobserved to be 2.87 V and 2.97 V [100], which is slightlylower than that of 3.1 V versus Na as expected fromLiFePO4 (3.45 V vs Li). The voltage difference between Liand Na cells is, however, less signicant than that observedfor the layered oxides (gure 2). The appearance of twovoltage plateaus originates from the formation of an inter-mediate phase (a sodium-ion ordered phase) as Na0.4FePO4(point B in gure16(a)). Such an intermediate phase is not

generally found in LixFePO4 under equilibrium conditions.Larger repulsive interaction for Na ions, as compared with Liions, may result in the formation of the intermediate phase[100,102], similar to layered oxides. According to the DFTcalculation, the barrier of sodium migration in narrow 1Dchains is much larger (270 meV) compared with Li ions(150 meV) [62]. Diffusion barriers correlate highly with dif-fusion paths in host structures. Indeed, experimentally mea-sured sodium insertion kinetics in triphylite-NaxFePO4appearto be much slower compared with Li ions [103].

The thermodynamically stable phase of NaMnPO4is alsomaricite type [104]. Similar to triphylite-NaFePO4, it is also

expected that lithiophilite-type NaMnPO4, which is iso-structural with triphylite-type Li(Na)FePO4, is obtained byion-exchange reaction and/or related soft chemical methods.It has been demonstrated that lithiophilite-type NaMnPO4 isprepared by topochemical reaction below 100 C fromNH4MnPO4H2O as the precursor [104]. Electrode perfor-mance of triphylite-NaMn1/2Fe1/2PO4 prepared by topo-chemical reaction has also been reported [104]. Electrodereversibility of lithiophilite-NaMnPO4may be limited withoutthe preparation of nano-engineered particles, similar toLiMnPO4, because of the formation of antisite defects[105,106] and/or the slow nucleation rate for the delithiated(desodiated) phase [107].

Figure 16.Electrode performance of phosphate-based compoundscontaining Fe(II) ions in Na cells: (a) NaFePO4(triphylite-type),(b) Na2Fe(P2O7), and (c) Na4Fe3(PO4)2(P2O7). (a) Reprinted withpermission from [100]. Copyright 2010 American Chemical Society.(b) Reproduced with permission from [111]. Copyright 2012Elsevier. (c) Reprinted with permission from [117]. Copyright 2012American Chemical Society.

20



22/30

4.2. Pyrophosphates with Fe(II) and Mn(II)

Following reports touting lithium iron(II) pyrophosphate,Li2FeP2O7, as a positive electrode material for LIBs[108, 109], a great deal of literature has been publishedregarding the pyrophosphate system. The use of pyropho-sphate, instead of phosphate, as the framework structure hasbeen extended to the Na system, i.e., Na2FeP2O7[110112].Na2FeP2O7 is isostructural with one of the polymorphs ofNa2CoP2O7, with an SG ofP1[113]. The crystal structure ofNa2FeP2O7 contains corner-shared FeO6 octahedra (Fe2O11units), which are connected by P2O7 units, forming the 3Dframework structure with open sodium diffusion paths(gure14(c)). Na ions are located at large distorted square-pyramidal sites. One Na ion is reversibly extracted fromNa2FeP2O7based on the Fe

2+/Fe3+ redox couple, and rever-sible capacity reaches 90 mAh g1 (theoretical capacity:97 mAh g1) as shown in gure16(b). Although the availableenergy density is lower than that of triphylite-NaFePO4, direct

synthesis of Na2FeP2O7 is possible by conventional solid-state methods. Additionally, rate capability seems to be muchbetter for Na2FeP2O7, even though FeO6octahedra and FeO5square pyramids are isolated by the pyrophosphate ions in theframework structure. This result is probably because of thepresence of open diffusion paths formed by pyrophosphateions. Na diffusion barriers in NaxFeP2O7 have been alsocalculated by rst-principles calculation [112]. Sodium ionscan diffuse through 1D paths along [011] with a relativelylow migration barrier (480 meV), and all Na sites areinterconnected by 1D/2D paths with a diffusion barrier below540 meV.

Recently it has been reported that one of the Na2MnP2O7polymorphs, -Na2MnP2O7, which is isostructural withNa2FeP2O7, also shows good electrode performance in Nacells [114]. Although the polarization seems to be large,Na2MnP2O7 is potentially usable as a 3.6 Vclass electrodematerial. Off-stoichiometric synthesis for sodium metal pyr-ophosphates induces the formation of Na-rich phases [115].Thus the prepared sample shows better electrode performancethan does the stoichiometric sample without nanosizing andcarbon coating. Pyrophosphates that include other transitionmetal elements as electrode materials have been reviewed inthe literature [116].

An interesting characteristic regarding the design ofelectrode materials for NIBs is the high structuralexibility ofsodium-based compounds, e.g., the P2-type layered structureas mentioned in an earlier section. It is impossible to prepare aP2-type layered structure for the Li system. Similarly, aunique framework structure, with mixed polyanion groups ofphosphate (PO4)

3 and prylophosphate (P2O7)4 ions, is

found in the Na system (and is not known in the Li system).Recently Na4Fe3(PO4)2(P2O7) was synthesized as the rstmixed polyanion compound containing Fe(II) [117]. Thestructure is isostructural with Na4Me3(PO4)2(P2O7) (Me = Co,Mn, and Ni), with an SG ofPn21a[118]. FeO6octahedra areconnected to on another by edge and corner shares, and PO4

tetrahedra also share one edge and two corners with the FeO6octahedra, forming a layer-like unit along the b-c direction.

These layer units, consisting of FeO6 and PO4, are furtherconnected by P2O7 pyrophosphate ions, forming the crystalstructure of Na4Fe3(PO4)2(P2O7), as shown in gure14(d).Na4Fe3(PO4)2(P2O7) delivers approximately 100 mAh g

1 ofreversible capacity, which is higher than that of Na2FeP2O7(gure 16(c)). Three Na ions are reversibly extracted from

Na4Fe3(PO4)2(P2O7) based on the Fe2+

/Fe3+

redox. Thesample shows good capacity retention as electrode material inNa cells [117].

4.3. Fluorophosphates, carbophosphates, and fluorosulfate

with Fe(II) and Mn (II)

The mixed-anions system containing uoride and phosphateions is also used as electrode material for NIBs. The layereduorinated iron phosphate Na2FePO4F has been widelyexamined as a positive electrode material for rechargeablebatteries [119, 120]. The crystal structure of Na2FePO4F isisostructural with Na2CoPO4F-type structures [121] (orNa2FePO4(OH)-type structures) as two-dimensional layereduorophosphates (gure17(a)). Na ions are accommodatedbetween FePO4F layers, in which FeO4F2 octahedra shareedge and corners. According to the results of rst-principlescalculation, sodium ions migrate between FePO4F layersthrough a two-dimensional path [122]. The crystal structure ofNa2MnPO4F is different from that of Na2FePO4F[123]. AllMnO4F2octahedra share each corner and form 1D Mn2F2O8chains. The chains are connected by PO4tetrahedra via cornersharing, thus forming a less dense 3D framework structure(gure 17(b)). Na2[FexMn1x]PO4F ( 0


23/30

show good reversibility as electrode material in Na

cells [125].Recently synthesis and electrode performance have beenreported for sodium-based carbonophosphates,Na3MePO4CO3, which are the new series of mixed-anioncompounds used as electrode materials, similar toNa4Fe3(PO4)2(P2O7) [126,127]. Na3MePO4CO3contains twodifferent anions of carbonate (CO3)

2 and phosphate (PO4)3

ions, and its Mn compound, Na3MnPO4CO3, is found in thenatural mineral sidorenkite [126]. The crystal structure ofNa3MnPO4CO3is shown in gure17(c). FeO6octahedra andPO4tetrahedra share each corner and CO3triangles share oneedge with the FeO6octahedra, resulting in a large distortionof the FeO6 octahedra. Na3MnPO4CO3 can be prepared byhydrothermal reaction at 120 C [116]. Two Na ions areextracted from Na3MnPO4CO3based on the Mn

2+/Mn4+ two-electron redox reaction, resulting in large initial chargecapacity (gure18(c)). Although the initial charge capacity islarge (200 mAh g1), it has been reported that the sampleshows relatively large irreversible capacity (75 mAh g1) inthe initial cycle, with acceptable capacity retention in sub-sequent continuous cycles [126].

Metal uorosulfates are also used as the frameworkstructure for the Na insertion host. Recently sodium ironuorosulfate, NaFeSO4F, was successfully prepared and itselectrode performance was examined in Na cells. The crystal

structure of NaFeSO4F is isostructural with the mineraltavorite, LiFe(PO4)OH [128]. Although the results of rst-

principles calculation suggest diffusion of sodium ions in the

1D path along [101] [129], reversibility of NaFeSO4F interms of electrode material is limited to a narrow range [128].

4.4. Phosphates with Fe(III) and Mn(II)

Because trivalent iron Fe(III) is, in general, stable underambient conditions, many Fe(III)-containing phosphates arefound in natural minerals. Some Fe(III)-containing phos-phates are known to be electrochemically active as sodiuminsertion hosts, based on the Fe2+/Fe3+ redox. The simplestcompounds are FePO4with polymorphs, including an amor-phous phase. A thermodynamically stable phase is iso-

structural with the mineral berlinite, AlPO4 [130], in whichFeO4 and PO4 tetrahedra share corners with each other,forming relatively open 3D channels for the insertion of alkaliions (gure19(a)). The electrode performance of amorphousand crystalline FePO4 (berlinite-type) is reported by Okadaand co-workers [131]. Figure20(a) shows the electrode per-formance of amorphous FePO4in Li and Na cells. Approxi-mately 80 mAh g1 of reversible capacity is obtained usingamorphous FePO4in the Na cell. Crystalline FePO4obtainedby heating the amorphous phase is also used as an electrodematerial [131]. FePO4polymorphs can be prepared by topo-chemical reaction from hydrates, and one of them consists of

corner-shared FeO6 octahedra and PO4 tetrahedra [132].These polymorphs are electrochemically active in Li cells

Figure 17.Schematic illustrations of the crystal structures of (a) Na2FePO4F, (b) Na2[Fe1/2Mn1/2]PO4F, (c) Na3MnPO4CO3, and(d) NaFeSO4F.

22



24/30

[132], but electrode performance in Na cells has not beenreported so far.

According to a study on a Na3PO4FePO4binary phasediagram [133], at least three different phases exist: Na3Fe(PO4)2, Na3Fe2(PO4)3, and Na3Fe3(PO4)3. A structural type ofthe most Na-rich phase, Na3Fe(PO4)2, was reported to be

related to the mineral glaserite, K3Na(SO4)2 [134]. FeO6octahedra and PO4 tetrahedra share corners, forming 2DFe2(PO4)2 layers. Sodium ions are accommodated betweenthe Fe2(PO4)2layers. Similar to Na3Fe(PO4)2, Na3Fe2(PO4)3also consists of corner-shared FeO6 octahedra and PO4 tet-rahedra, but Na3Fe2(PO4)3 has a 3D framework structure.Such a structure of Na3Fe2(PO4)3is known to be related to theNa+-ion super ionic conductors (NASICON) type of structure(gure 19(b)), and the polymorphic characteristics ofNa3Fe2(PO4)3 have been extensively studied [135]. Ion-exchanged Li3Fe2(PO4)3 from Na3Fe2(PO4)3is known to beelectrochemically active in Li cells [39]. Approximately0.4 mole of Na is reversibly inserted/extracted into/fromNa3Fe2(PO4)3 in Na cells [136], even though ionic con-ductivity of Na3Fe2(PO4)3 is reported to be very low(1.2107 S cm1 at 50 C) [137].

Na3Fe3(PO4)4 is also classied as a layered-type phos-phate [133]. FeO6 octahedra and PO4 in Na3Fe3(PO4)4 tet-rahedra share both edge and corners, forming 2D Fe3(PO4)4layers (gure19(c)). This type of layered structure is found inK3Fe3(PO4)4 H2O [138]. The electrode performance ofNa3Fe3(PO4)4in Li and Na cells has been reported [139,140].Approximately 2 moles of Li and Na ions are reversiblyinserted into Na3Fe3(PO4)4. Polarization in the Na cell ismuch smaller than in the Li cell, as shown in gure20(b).

Although highly crystallized and microsized particles showgood electrode performance, the operating voltage is rela-tively low2.5 V versus Na, based on the Fe2+/Fe3+ redox.

Binary phosphates containing Fe(III) and Mn(II) are alsofound in natural minerals. One example is the mineralalluaudite [141]. The crystal structure of alluaudite contains1D chains made from edge-shared Fe(Mn)O6octahedra withdistortion. These 1D chains are connected with PO4tetrahedraby corner sharing, forming the 1D tunnel sites for Na ions(gure19(d)). Na ions are located at distorted octahedral sitesand large eight oxygen coordinated sites. Electrode perfor-mance of alluaudite-type NaMnFe2(PO4)3has been tested inNa cells [140]. Na ions (ca. 0.5 mole) can be partiallyextracted from NaMnFe2(PO4)3 prepared using a sol-gelmethod, and the reversible capacity reaches 60 mAh g1 in aNa cell (gure20(c)). Much better reversibility was found inNa3Fe3(PO4)4without Mn(II) ions.

The electrode performance of the uorophosphate con-taining Fe(III), Na3Fe2(PO4)2F3, was recently reported [142].The crystal structure of Na3Fe2(PO4)2F3is isostructural withNa3Fe2(PO4)2(OH)2F [143]. The framework structure ofNa3Fe2(PO4)2F3 consists of FeO4F2 octahedra and PO4 tet-rahedra, all of which share corners, forming Fe2(PO4)2F3layers. A two-dimensional migration path for sodium ionsalong the ab plane is formed by corner sharing for the

FeO4F2octahedra in the Fe2(PO4)2F3layers. Na3Fe2(PO4)2F3delivers 40 mAh g1 of reversible capacity in a Na cell [142].

Figure 18.Charge/discharge curves of (a) the Na/Na2FePO4F cell,(b) Na/Na2[Fe1/2Mn1/2]PO4F cell, and (c) Na/Na3MnPO4CO3cell.Copyright 2012 The Electrochemical Society of Japan. (c) Reprintedwith permission from [126]. Copyright 2013 American ChemicalSociety.

23



25/30

5. Miscellaneous Na insertion materials

Iron triuoride, FeF3, has been studied as a potential candidateto achieve high energy density as an iron-based electrode

material in Li cells. The electrode performance of FeF3in Licells was rst examined by Arai and co-workers [144]. FeFbonds in FeF3 are strongly polarized because of the highelectronegativity of uorine in comparison with oxygen.Electrons are therefore localized in FeF3, resulting in insuf-cient electrical conductivity as an electrode material. FeF6octahedra share all corners, forming a framework structurewith open channels (gure21(a)). The structure of FeF3 isisostructural with distorted ReO3. (It is also classied as an A-site decient perovskite structure). Although FeF3has an openstructure with three-dimensional pathways for lithium migra-tion, the electrode performance of micrometer-sized FeF3 is

limited by its insulating character. The electrode performanceof FeF3in Li cells is signicantly improved by using nano-composite materials consisting of FeF3and conductive carbon[145]. This concept has been extended to the Na system, and itwas demonstrated that FeF3/C nanocomposite delivers largereversible capacity (150 mAh g1) in Na cells (gure22(a))[146]. LiFeF3 is formed as a metastable phase by electro-chemical (or chemical) reduction of FeF3 in Li cells. It isimpossible to prepare LiFeF3directly by a simple solid-statemethod because the vacant 6a site in FeF3(A-site in ABO3-type perovskite) is too large for Li ions (the interatomic dis-tance reaches 2.67 on average for the distorted 12-coordi-

nation site). Such a lithiated phase is preferable as a positiveelectrode material for assembling complete cells (LIBs) in

combination with carbonaceous materials as negative electro-des. In contrast with LiFeF3, NaFeF3is easily prepared as athermodynamically stable phase because the large Na ions areenergetically stabilized at A-sites of the perovskite structure, asshown ingure21(a). Nanosized NaFeF3is electrochemicallyactive in Na cells and is used as positive electrodes, eventhough relatively large polarization for charge and dischargecycles has been observed [147,148].

Na ions are apparently coordinated by fouruoride ionsat bottleneck sites when the Na ions migrate across the per-ovskite-type framework structure. Relatively large activationenergy for Na migration is expected in the charge/dischargeprocess for NaxFeF3. Therefore, the use of cyanide ions ratherthan uoride ions has been proposed to reduce the activationenergy with respect to the Na migration process [149].Moreover, the cyanide ions as strong-eld ligands drastically

inuence crystal eld splitting for the 3dorbital of iron ions.Recently Prussian blue analogues, which have a frameworkstructure similar to that of perovskite, as shown ingure21(b), have been studied as the Na insertion host. Theiron and manganese system with potassium ions, KFe2(CN)6and KMnFe(CN)6 [150], and the iron (III) system withoutalkaline metal ions, Fe2(CN)6[151], shows 80120 mAh g

1

of reversible capacity as positive electrode materials in Nacells. Polarization of Prussian blue analogues in Na cells ismuch smaller than that of the perovskite system with uorideions. Goodenoughs research group has reported the Na-enriched iron manganese system, Na1.72MnFe(CN)6(an ideal

crystal structure without sodium vacancies is illustrated ingure 21(b)), as a potential positive electrode material for

Figure 19.Crystal structures for phosphate-based compounds containing Fe(III) and Mn(II): (a) FePO4 (berlinite-type), (b) Na3Fe2(PO4)3(NASICON-type), (c) Na3Fe3(PO4)4(Na3Cr3(PO4)4-type), and (d) NaMnFe2(PO4)3(alluaudite-type).

24



26/30

NIBs [149]. Magnetic measurements of Na1.72MnFe(CN)6suggest the presence of low-spin Fe(II) coordinated by car-bon, and the low-spin Fe(II)/Fe(III) is redox active in Na cells[149]. The reversible capacity reaches 130 mAh g1, with anaverage operating voltage of 3.2 V. The sample also showsexcellent rate capability in the Na cell, as shown in

gure22(b). The electrode performance of a similar sodiummanganese ferrocyanide has also been recently reported[152]. The activation energy of the charge transfer reactionfor Na insertion into Prussian blue analogues greatly dependson the electrolyte solution used, and extremely small activa-tion energy (5 kJ mol1) in water-based electrolyte solutionhas been reported [153]. Additionally, sodium ferrocyanide,Na4Fe(CN)6, is reported to be electrochemically active in Nacells. The studied sample delivers 90 mAh g1 of reversiblecapacity, with a at voltage plateau at 3.4 V [154]. Thisreversible capacity nearly corresponds to the capacity denedby the one-electron redox reaction of iron for Na4Fe(CN)6.

Pyrite, FeS2, which is the most common of the suldeminerals, is also used as an electrode material for Na cells,even though its reaction mechanism is not classied as aninsertion reaction [155]. FeS2 consists of Fe

2+ and S22 (per-

sulde ions), and clear evidence of the presence of the SSbond is found in the structure, as shown in gure21(c). In thereduction process in Na cells, Fe2+ is reduced to metallic Fe,and S22 is potentially reduced to form Na2S. If this reaction(so-called conversion reaction [75]) is a reversible process,the theoretical capacity exceeds 950 mAh g1. Although largeinitial discharge capacity was obtained (gure 22(c)), thereversibility is insufcient for an electrode material. The useof the conversion reaction may be a strategy to design a

battery system with high energy density.

6. Conclusions and outlook

Research interest in the use of sodium ions as ion carriers forenergy storage at ambient conditions almost died away com-pletely at one point after innovations in materials for the Lisystem (thending of LiCoO2 and graphite as lithium insertionhosts) and the commercialization of LIBs were achieved in the1990s. However, there has been a complete turnaround inresearch interest regarding sodium ions as a result of increasing

worldwide demand for EES. In considering the design of betterbatteries for EES, of primary importance is the elementalabundance (cost) together with acceptable battery performance(energy density, cycle life, safety, etc). In this regard, the mostimportant discovery as a turning point in the research of NIBsis that NaFeO2is found to be electrochemically active in Nacells based on the Fe3+/Fe4+ redox couple [20]. This discoverymay be on a par with thending of LiCoO2, which is still usedas the primary electrode material for high-energy LIBs, in theLi system in 1980 [17]. The Fe3+/Fe4+ redox chemistry isunique to the Na system and has never been reported as activein the Li system. Additionally, the chemistry of the Mn3+/Mn4+

redox is quite different between the Li and Na systems, as

Figure 20.Electrode performance of phosphate-based compoundscontaining Fe(III) and Mn(II) ions: (a) FePO4(berlinite-type),(b) Na3Fe3(PO4)4(Na3Cr3(PO4)4-type), and (c) NaMnFe2(PO4)3(alluaudite-type). (a) Reproduced with permission from [131].Copyright 2006 Elsevier. (b) Reproduced with permission from[139]. Copyright 2010 The Electrochemical Society. (c) Reprintedwith permission from [160]. Copyright 2010 American ChemicalSociety.

25



27/30

Figure 21.Crystal structures for (a) NaFeF3, (b) NaxMnFe(CN)6, and(c) FeS2(pyrite).

Figure 22.Electrode performance of (a) FeF3, (b) NaxMnFe(CN)6,and (c) FeS2(pyrite) in Na cells. (a) Reproduced with permissionfrom [147]. Copyright 2009 Elsevier. (b) Reprinted with permissionfrom [149]. Copyright 2013 John Wiley and Sons. (c) Reproducedwith permission from [155]. Copyright 2007 Elsevier.

26



28/30

summarized in this reviewarticle. The spinel and layered phasetransition problems are completely different from each other.Aside from elemental abundance, materials synthesis andprocessing methods also inuence the cost of materials.Recently water-based electrolyte was also proposed as a cost-effective system, and the aqueous battery system with carbon/

managanese oxide was commercialized for EES application byAquion Energy [156]. However, its energy density could besignicantly reduced compared with that of the non-aqueous(aprotic) system because of the lower operating voltage ofaqueous cells. Cycle life is also another important factor inreducing total cost. Because the limitation of space for storingbatteries may be a less signicant issue for EES than forelectronic devices and transportation systems, Wh versus costof batteries is of primary importance. Therefore, the strategy ofbattery design (including materials) for EES is needed to bal-ance the increase in energy density and the decrease in the totalcost of batteries. Iron and manganese are believed to be keyelements toward achieving high-energy and cost-effectiveNIBs for EES. In this article, some important points were notprovided in terms of developing NIBs. Safety (especially forcommonly used batteries) is another issue that needs to bediscussed. Electrode performance is often dominated by theinterfacial structure between electrodes and electrolytes.Electrolytes, electrolyte additives, binders, conductive mate-rials, separators, current collectors, and all other cell compo-nents inuence battery performance (and cost). Moreover,negative electrode materials for NIBs were not discussed inthis article. The progress of recent research into negativeelectrodes for NIBs is discussed in another reviewarticle [157].Intense research efforts all over the world will further accel-

erate the development of NIBs for EES in the future, eventhough many difcult challenges may lie ahead.

Acknowledgments

The authors would like to acknowledge the many studentsand collaborators who have contributed to this research seriesover the last several years.

References

[1] Dunn B, Kamath H and Tarascon J-M 2011Science 334 928[2] Retschi P 1961J. Electrochem. Soc. 108297[3] Thackeray M M, Johnson P J, Depicciotto L A,

Bruce P G and Goodenough J B 1984 Mater. Res. Bull.19179

[4] Ohzuku T, Kitagawa M and Hirai T 1990 J. Electrochem.Soc. 137769

[5] Tarascon J M, Wang E, Shokoohi F K, McKinnon W R andColson S 1991 J. Electrochem. Soc. 1382859

[6] Dahn J R, Fuller E W, Obrovac M and Vonsacken U 1994Solid State Ion. 69265

[7] Tarascon J-M 2010 Nat. Chem. 2510[8] Carmichael R S 1989Practical Handbook of Physcial

Properties of Rocks and Minerals(Boca Raton: CRC Press)

[9] Slater M D, Kim D, Lee E and Johnson C S 2013 AdvancedFunctional Materials 23947

[10] Komaba S, Murata W, Ishikawa T, Yabuuchi N, Ozeki T,Nakayama T, Ogata A, Gotoh K and Fujiwara K 2011Advanced Functional Materials 213859

[11] Kim S W, Seo D H, Ma X H, Ceder G and Kang K 2012Adv.Energy Mater. 2710

[12] Oshima T, Kajita M and Okuno A 2004 Int. J. Appl. CeramicTechnol. 1269

[13] Bones R J, Teagle D A, Brooker S D and Cullen F L 1989J. Electrochem. Soc. 1361274[14] Whittingham M S 1978Prog. Solid State Chem. 1241[15] Whittingham M S 1976Science 1921126[16] Newman G H and Klemann L P 1980 J. Electrochem. Soc.

1272097[17] Mizushima K, Jones P C, Wiseman P J and Goodenough J B

1980 Mater. Res. Bull. 15783[18] Delmas C, Braconnier J-J, Fouassier C and Hagenmuller P

1981 Solid State Ion. 34165[19] Abraham K M 1982Solid State Ion. 7199[20] Okada S, Takahashi Y, Kiyabu T, Doi T, Yamaki J-I and

Nishida T 2006 210th ECS Meeting AbstractsMA2006-02 p 201 (available athttp://ma.ecsdl.org/content/MA2006-02/4/201.abstract)

[21] Zhao J, Zhao L, Dimov N, Okada S and Nishida T 2013J. Electrochem. Soc. 160A3077[22] Delmas C, Fouassier C and Hagenmuller P 1980Physica B &

C9981[23] Momma K and Izumi F 2008J. Appl. Crystallogr. 41653[24] Shannon R D 1976 Acta CrystallographicaA 32751[25] Mather G C, Dussarrat C, Etourneau J and West A R 2000

J. Mater. Chem. 102219[26] Mendiboure A, Delmas C and Hagenmuller P 1985 J. Solid

State Chem. 57323[27] Stoyanova R, Carlier D, Sendova-Vassileva M, Yoncheva M,

Zhecheva E, Nihtianova D and Delmas C 2010 J. Solid StateChem. 1831372

[28] Paulsen J M and Dahn J R 2000J. Electrochem. Soc. 1472478[29] Lu Z and Dahn J R 2001 J. Electrochem. Soc. 148A710

[30] Lu Z and Dahn J R 2001J. Electrochem. Soc.148

A1225

9[31] Delmas C, Braconnier J-J and Hagenmuller P 1982Mater.Res. Bull. 17117

[32] Paulsen J M, Donaberger R A and Dahn J R 2000Chem.Mater. 122257

[33] Carlier D, Saadoune I, Croguennec L, Mntrier M,Suard E and Delmas C 2001 Solid State Ion. 144263

[34] Lu Z H and Dahn J R 2001 Chem. Mater. 132078[35] Tournadre F, Croguennec L, Saadoune I, Carlier D,

Shao-Horn Y, Willmann P and Delmas C 2004 J. Solid StateChem. 1772790

[36] Yabuuchi N, Kawamoto Y, Hara R, Ishigaki T, Hoshikawa A,Yonemura M, Kamiyama T and Komaba S 2013 Inorg.Chem. 529131

[37] Takeda Y, Akagi J, Edagawa A, Inagaki M and Naka S 1980

Mater. Res. Bull. 151167[38] Ado K, Tabuchi M, Kobayashi H, Kageyama H,Nakamura O, Inaba Y, Kanno R, Takagi M and Takeda Y1997 J. Electrochem. Soc. 144L177

[39] Padhi A K, Nanjundaswamy K S, Masquelier C, Okada S andGoodenough J B 1997 J. Electrochem. Soc. 1441609

[40] Takeda Y, Nakahara K, Nishijima M, Imanishi N,Yamamoto O, Takano M and Kanno R 1994 Mater. Res.Bull. 29659

[41] Yabuuchi N, Yoshida H and Komaba S 2012Electrochemistry 80716

[42] Yamamoto T 2008 X-Ray Spectrom. 37572[43] McLaren V L, West A R, Tabuchi M, Nakashima A,

Takahara H, Kobayashi H, Sakaebe H, Kageyama H,Hirano A and Takeda Y 2004J. Electrochem. Soc. 151A672

[44] Monyoncho E and Bissessur R 2013 Mater. Res. Bull.482678

27

http://dx.doi.org/10.1126/science.1212741http://dx.doi.org/10.1126/science.1212741http://dx.doi.org/10.1149/1.2428069http://dx.doi.org/10.1149/1.2428069http://dx.doi.org/10.1016/0025-5408(84)90088-6http://dx.doi.org/10.1016/0025-5408(84)90088-6http://dx.doi.org/10.1149/1.2086552http://dx.doi.org/10.1149/1.2086552http://dx.doi.org/10.1149/1.2085330http://dx.doi.org/10.1149/1.2085330http://dx.doi.org/10.1016/0167-2738(94)90415-4http://dx.doi.org/10.1016/0167-2738(94)90415-4http://dx.doi.org/10.1038/nchem.680http://dx.doi.org/10.1038/nchem.680http://dx.doi.org/10.1002/adfm.v23.8http://dx.doi.org/10.1002/adfm.v23.8http://dx.doi.org/10.1002/adfm.v21.20http://dx.doi.org/10.1002/adfm.v21.20http://dx.doi.org/10.1002/aenm.201200026http://dx.doi.org/10.1002/aenm.201200026http://dx.doi.org/10.1111/ijac.2004.1.issue-3http://dx.doi.org/10.1111/ijac.2004.1.issue-3http://dx.doi.org/10.1149/1.2096905http://dx.doi.org/10.1149/1.2096905http://dx.doi.org/10.1016/0079-6786(78)90003-1http://dx.doi.org/10.1016/0079-6786(78)90003-1http://dx.doi.org/10.1126/science.192.4244.1126http://dx.doi.org/10.1126/science.192.4244.1126http://dx.doi.org/10.1149/1.2129353http://dx.doi.org/10.1149/1.2129353http://dx.doi.org/1

rev jappan

Documents