lable at ScienceDirect

Electrochimica Acta 276 (2018) 142e152

Contents lists avai

Electrochimica Acta

journal homepage: www.elsevier .com/locate/e lectacta

Pseudocapacitive charge storage properties of Na2/3Co2/3Mn2/9Ni1/9O2in Na-ion batteries

M. Valvo a, *, S. Doubaji b, I. Saadoune b, c, K. Edstr€om a

a Department of Chemistry e Ångstr€om Laboratory, Uppsala University, Box 538, 75121 Uppsala, Swedenb LCME, FST Marrakech, University Cadi Ayyad, Av. A. Khattabi, BP 549, 40000 Marrakech, Moroccoc Materials Science and Nanoengineering Department, Mohamed VI Polytechnic University, Lot 660 e Hay Moulay Rachid, 43150 Benguerir, Morocco

a r t i c l e i n f o

Article history:Received 14 November 2017Received in revised form20 April 2018Accepted 21 April 2018Available online 23 April 2018

Keywords:Na-ion batteriesNa2/3Co2/3Mn2/9Ni1/9O2

Layered oxidesPositive electrodesPseudocapacitance

* Corresponding author.E-mail address: mario.valvo@kemi.uu.se (M. Valvo

https://doi.org/10.1016/j.electacta.2018.04.1500013-4686/© 2018 The Authors. Published by Elsevier

a b s t r a c t

The behaviour of Na2/3Co2/3Mn2/9Ni1/9O2 in composite electrodes is studied via Na half-cells utilizing adedicated cyclic voltammetry approach. The application of increasing sweep rates enabled a detailedanalysis of the red-ox peaks of this material. All peak currents due to cathodic/anodic processesdemonstrated clear capacitive properties. This finding widens the picture of classical Naþ insertion/extraction in this complex oxide, as purely diffusive processes of Naþ through its layers do not fullyexplain the pseudocapacitance displayed by its red-ox peaks, which are typically linked to concomitantoxidation state changes for its transition metal atoms and/or phase transitions. No phase transition wasobserved during in operando X-Ray diffraction upon charge to 4.2 V vs. Naþ/Na, proving good structuralstability for P2-NaxCo2/3Mn2/9Ni1/9O2 upon Naþ removal. The origin of such pseudocapacitive propertiesis likely nested in strong electrostatic interactions among the metal oxide slabs and a tendency to releaseNaþ from its crystallites, e.g. to form surface by-products upon air exposure. Such a reactivity inducesdefects (e.g. vacancies) in its lattice and charge compensation mechanisms are required to maintain anoverall charge neutrality, thus ultimately influencing the electrochemical properties. Possible limitingfactors for the performances of this compound in composite coatings are also discussed.© 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Rechargeable Na-ion batteries (NIBs) are progressivelybecoming more attractive for large-scale applications aiming atstationary storage of electrical energy [1] thanks to their vantagepoints of relatively contained costs [2] and high abundance ofsodium-rich raw materials in both the earth crust and the oceans[3]. Although basic similarities exist between Na-ion batteriesand Li-ion batteries (LIBs), in terms of both materials andfundamental electrochemical processes, the development of NIBsand LIBs has not proceeded in a parallel way for these twotechnologies so far.

The noticeable differences in characteristic radii, masses andstandard electrochemical potentials of Liþ and Naþ ions [4], as wellas their typical diffusivities in solid state compounds, create a widescenario for the discovery of unusual properties in NIB electrodes.The latter cannot directly be predicted on the sheer basis of their

).

Ltd. This is an open access article u

LIB counterparts, which have been object of more extensive in-vestigations. Many compounds and related electrochemical re-actions in NIBs do not find a close behaviour in LIBs, yielding in afew cases surprising results [5] and favourable characteristics interms of structural changes [6] and overall performances [7e9].These are, however, interesting exceptions, since in most of thecases the reactions involving Naþ ions and their resulting electro-chemical features are not as beneficial as those obtainable by Liþ

ions under analogous conditions. Electrochemical sodiation ofgraphite [10,11] or silicon to form Na-Si alloys [12], as well as fullconversion of iron oxide [13,14] via Naþ are typical examples ofhurdles for the development of negative electrodes in NIBs. On theother hand, positive electrodes based on sodium-containing tran-sition metal compounds [15e22] have experienced a higher degreeof success in NIBs, revealing intriguing electrochemical mecha-nisms that do not necessarily mimic those taking place in similarLIB electrodes [23]. One possible advantage of employing Na-basedcompounds is, for example, avoiding cationic intermixing [24] (i.e.transition metal/sodium) in the structure, thanks to the larger sizeof Naþ compared to that of Liþ, which can result in some cases in adistinctive electrochemical activity and thermal stability, such as

nder the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

M. Valvo et al. / Electrochimica Acta 276 (2018) 142e152 143

for NaFeO2 [25], contrarily to its metastable LiFeO2 counterpart[26].

Among the large varieties of sodium-containing positive elec-trodes, layered transition metal oxides (e.g. NaMO2, where M¼ Fe,Co, Mn, Ni, Cr, V, etc.) are a topic of intense investigation[23,27e35], as the combination of two or more of these transitionmetals in associated crystalline compounds offers the possibility oftuning crucial properties (e.g. red-ox potentials, capacities, struc-tural changes, cycleability, etc.) for convenient operation in NIBs.The simultaneous presence of Co, Mn and Ni in this family oflayered oxides proved beneficial for achieving suitable electro-chemical performances [36e40]. Finding an optimal stoichiometry,while controlling the atomic composition of the layers, remains achallenge, especially if one aims at the best combination of mate-rials stability, electrochemical properties and simultaneouslywishes to lower the amount of Co, due to its high costs and poorsustainability [3].

Nonetheless, we have synthesized and accurately characterizedin previous studies [41,42] a P2-type compound with the nominalcomposition Na2/3Co2/3Mn2/9Ni1/9O2, which demonstrates that asimultaneous achievement of good specific capacity, limited po-larization, reasonable Coulombic efficiency and cycle life is indeedpossible, without sacrificing too much the average operation po-tential upon charge/discharge. Moreover, all the transition metalsin this P2-type compound appeared to be electrochemically activeupon cycling and earlier results from our Hard X-Ray PhotoelectronSpectroscopy (HAXPES) study [42] indicate that the plateaus foundin these galvanostatic measurements do not tie up with specificredox couples, as it was previously suggested for this type ofcomplex layered oxides [38]. This intriguing feature of activityseems to be general for this particular family of materials, in whichseveral electronic phenomena influence the resulting electro-chemical properties [43].

With these ideas in mind, we investigate here in depth theelectrochemical behaviour of Na2/3Co2/3Mn2/9Ni1/9O2 through adetailed voltammetric study to unravel the nature of the capacityarising from its characteristic redox peaks upon cycling in Na half-cells. Possible structural changes, typically accountable for varia-tions in the electrochemical properties of these materials, weremonitored by means of in operando X-ray diffraction coupled togalvanostaticmeasurements. The overall analytical approach of thisinvestigation represents also a valid tool to pinpoint relevantlimiting factors existing in these composite coatings based on Na2/3Co2/3Mn2/9Ni1/9O2 powders, thereby providing additional infor-mation on this complex electrode system and possible strategies toimprove its overall electrochemical performance as advanced NIBcathode.

2. Experimental

2.1. Electrode and electrolyte preparation

The active compound Na2/3Co2/3Mn2/9Ni1/9O2 was synthesizedvia a sol-gel route, according to an earlier procedure [41]. Theelectrode formulation consisted of 75wt% of activematerial, 10wt%of polyvinylidene fluoride (PVdF) binder (Kynar Flex, 2801) and15wt% of carbon black (CB e Super P). The PVdF binder and theother components were mixed in a ball-mill jar together with N-methyl-pyrrolidone (NMP) and ball-milled for 1 h. The slurry wascast on an aluminium foil by a coating apparatus (KR e K ControlCoater) and dried for 3 h in a convection oven at 60 �C. Coatedelectrodes having a diameter of 20mm were cut by means of aprecision perforator (Hohsen) and subsequently dried in a vacuumoven (Büchi) at 120 �C for 12 h before battery preparation. Theexposure of the electrodes to air was minimized throughout the

various preparation stages, due to the sensitivity of this compoundand its tendency of reaction with other species to form unwantedby-products (e.g. Na2CO3). The thickness of the composite coatingswas measured by a digital calliper (Mitutoyo), yielding an averagevalue of approximately 45 mm. The electrode coatings had a typicalmass loading per footprint area of roughly 1.96mg cm�2 for theactive material.

The electrolyte was prepared in an Ar-filled glove-box (M-Braun) by first drying in vacuum a NaPF6 salt at 110 �C for 12 h in atubular oven and then dissolving it in propylene carbonate (PC) toobtain a 0.5M electrolytic solution. No additives were included inthis electrolyte preparation.

2.2. Cell assembly and electrochemical measurements

Polymer-laminated (i.e. ‘coffee-bag’) cells were prepared in anAr-filled glove box (M-Braun) having oxygen and moisture levelsbelow 1 ppm. The batteries enclosed the composite Na2/3Co2/3Mn2/

9Ni1/9O2/PVdF/CB coating as working electrode, sodium metal assimultaneous reference- and counter-electrode and a 0.5M NaPF6electrolyte having PC as solvent. A thin membrane (Solupor) wasdrenched with a few electrolyte droplets and used as separator inthese Na half-cells, which had aluminium tabs as electricalcontacts.

The electrochemical measurements were performed at roomtemperature via a VMP2 (Bio-Logic) equipment. Cyclic voltamme-try (CV) was conducted on the Na half-cells using a series of pro-gressively increasing sweep rates ranging from 0.01 to 0.64mVs-1

between 2.00 V and 4.20 V vs. Naþ/Na to study the evolution of thered-ox peak currents and the variations of their associated peakpotentials. Higher sweep rates were not applied here because thesize of the active particles was large and the Naþ transfer for thesecompounds is typically not as fast as that of Liþ. A thorough analysiswas performed then to assess the electrochemical behaviour ofthese electrodes and their main limitations upon operation at suchlow and moderate rates.

2.3. Materials characterization and in operando XRDmeasurements

The morphology of the active powders was investigated byscanning electron microscopy (SEM) using a high resolutionscanning electron microscope (Zeiss Gemini 1550) with a fieldemission gun (FEG) electron source and a dedicated InLens de-tector for the detection of secondary electrons. Energy dispersiveX-ray (EDX) spectroscopy was carried out on the same specimensby means of a built-in EDX probe (Oxford) operated via an AZtec(INCA energy) software for X-ray mapping and simultaneouselemental analysis. FT-IR measurements of the powders wereperformed between 750 and 4000 cm�1 with a PerkinElmer(Spectrum One) spectrometer using an attenuated total reflec-tance (ATR) mode. The spectra were recorded through 20 cumu-lative scans to improve the signal-to-noise (S/N) ratio ofcorrespondent spectral features. Nitrogen gas adsorption mea-surements were conducted using a Micromeritics (Flowsorb III)equipment to determine the surface area of the active powders.The specific surface area of the powders was calculated byBrunauer-Emmet-Teller (BET) method. Powder X-Ray diffraction(XRD) of the pristine compound was run on a wide angular range(10� � 2q� 88�) in a Bragg-Brentano geometry via a Bruker (D8Advance) diffractometer equipped with a Cu Ka radiation.

In operando XRD measurements were carried out in reflectionmode using a dedicated electrochemical cell. The cell consisted ofa main window made of beryllium (i.e. transparent to X-rays)located in the upper part and an aluminum current collector,

M. Valvo et al. / Electrochimica Acta 276 (2018) 142e152144

where the active material was placed together with the separatorand the metallic sodium. In operando XRD measurements wereperformed with a Cu Ka radiation source and a movable detectoraround the fixed sample. The material was charged in a galvano-static mode up to 4.2 V vs. Naþ/Na with a current rate of C/35 (i.e.7.21mAg�1).

3. Results and discussion

3.1. Morphology and composition analysis

The morphology of the pristine Na2/3Co2/3Mn2/9Ni1/9O2 powderis presented in Fig. 1a together with a series of associated chemicalmaps for its constitutive elements (see also overall EDX spectrum inFig. S1 of Supporting Information e SI).

The shape of the particles in Fig. 1a exhibits hexagonal- andprismatic-like features, as expected for the characteristic P2-typestructure of this compound [41]. The particle size is large andvaries on a typical scale of a few microns, indicating that thesynthesis procedure led to the formation of bulky crystallites.These active material particles possess an overall structurereflecting the order of their atomic constituents arranged in a 2Dlayered pattern, where the sodium atoms are found between twosubsequent metal oxide slabs (i.e. MO2 where M¼ Co, Mn, Ni) [41].The particles in Fig. 1a display some additional surface charac-teristics, i.e. a few irregularly shaped regions that generate aslightly different contrast in the same topographic SEM image (seeblack arrows) and extend on a typical scale length of �1 mm. Thesame regions appear brighter in the chemical map of Fig. 1b (seewhite arrows), where the respective distribution of Na is shown,whereas slightly darker in correspondence of the other elementalmaps of Co, Ni and Mn (see Fig. 1def and white arrows). This candirectly be related with the presence of sodium carbonate on thesurface of this compound [36,37], which can be due to both itssynthesis and subsequent air exposure (see Fig. S2 in SI), thusinfluencing both morphology and interface reactivity of theresulting material. Contact with CO2 during air exposure of thepowder surface was clearly demonstrated to lead to Na2CO3 for-mation by earlier FT-IR [41] and HAXPES [42] measurements. Thesensitivity of the Na in this compound has been ascertained and,although the exposed surface area of the powders in Fig. 1a is byno means large (i.e. 3.6m2g-1 according to BET analysis), its ten-dency of being removed from the crystallites and reacting withelectronegative species and/or groups in the surrounding envi-ronment was confirmed by previous studies, showing alsopossible additional formation of sodium fluoride by-products incomposite electrodes with PVdF [42]. These preliminary observa-tions confirm the high mobility of Na within the bulk of this P2-type compound [44] and the ease of sodium extraction from thelattice structure, especially if an external (e.g. chemical) drivingforce exists. This could lead in principle also to Hþ/Naþ exchangeunder particular conditions (i.e. direct contact with acid species).However, this circumstance is ruled out here, due to carefulhandling and storage of the as-prepared compound under Argon.Possible presence of NiO by-products was also checked, as thisoxide is a common crystalline impurity found in sodium layeredoxides containing Ni and since a small local inhomogeneity of theNi signal in its chemical map (Fig. 1e) was seen as well. XRD of thepristine Na2/3Co2/3Mn2/9Ni1/9O2 powders was carried out over awide angular range to ascertain hypothetical existence of such NiOimpurities that might arise during the synthesis process. However,no trace of NiO was found (see Fig. S3b in SI) and the purity of theactive powders was confirmed, except from their surface Na2CO3deposits.

3.2. Analysis of possible structural changes upon sodium extractionby in operando XRD

The structural stability of NaxCo2/3Mn2/9Ni1/9O2 subjected tosodium extraction during charge from 2.00 V to 4.20 V vs. Naþ/Nawas probed by in operando X-ray diffraction coupled to galvano-static measurements. Fig. 2 shows a series of diffractograms ofNaxCo2/3Mn2/9Ni1/9O2 (with 0.69� x� 0.39) recorded during cellcharge up to 4.2 V vs. Naþ/Na at a low rate of C/35 (i.e.z7.2mAg�1)using a dedicated Na half-cell setup.

NaxCo2/3Mn2/9Ni1/9O2 possesses a P2-type structure with a P63/mmc space group, in which Co, Mn and Ni are located in octahedralsites, while sodium ions are positioned in two different prismaticsites (i.e. Nae, where Na shares only edges with MO6 and Naf, whereNa shares faces with MO6). This is schematically illustrated inFig. 2a and a detailed Rietveld refinement of this structure can befound in an earlier study [41] and in Fig. S3a in SI. The compoundunderwent the extraction of 0.33 Naþ during the charge leading toa final composition of Na0.36Co2/3Mn2/9Ni1/9O2 at 4.2 V vs. Naþ/Nacorresponding to z82mAhg�1 in Fig. 2c. Its characteristic diffrac-tion peaks (002), (004) and (100)maintained their distinctive sharpshape and related widths (see Table S1 in SI) during cycling in allthe diffractograms showed in Fig. 2b. No additional peaks wereobserved, thus indicating that the material preserved both itscrystallinity and distinctive P2-type structure in this range of po-tentials, although the presence of stacking faults cannot beexcluded. A very weak broadening around 17� was observed inFig. 2a for a Na content �0.54. However, the origin of this minorfeature remains unclear, also because no apparent peak or evidentbroadening can be detected upon careful inspection of the samediffractograms (see later a magnified portion of the same angularrange in Fig. 3a). Two major inflections were noticed in the voltagecurve of Fig. 2c, one approximately around 3.4 V vs. Naþ/Na andanother atz4.0 V vs. Naþ/Na corresponding to a sodium content ofz0.53 andz0.39, respectively. These characteristic features in thevoltage profiles of this type of compounds are often associated withsodium ordering [35]. A more pronounced inflection of the po-tential is noticed in typical charge curves of NaxCoO2 at x¼ 0.5corresponding to the formation of an ordered phase [35,45].However, the presence of several other inflections and steps, whichcharacterize the voltage profiles of NaxCoO2, is not observed hereand this fact is corroborated by the consistency of the diffractionpatterns in Fig. 2b. These results are also in good agreement withseveral other studies devoted to similar P2-sodium layered mate-rials [15,28,32], where the P2 structure was found stable withrespect to the compositional range investigated here.

The main detectable change in the XRD patterns was a shift ofthe characteristic peaks (see Fig. 3), which is representative of avariation in lattice parameters during the sodium extraction.

The peaks indexed (002) and (004) represent the latticeparameter “c” and a progressive shift towards lower 2q values wasobserved for them during the initial stages of the Naþ extraction.This trend was followed by another shift directed to higher 2qvalues, while reaching a composition of x¼ 0.39. The other peak(100) corresponds here to the lattice parameter “a”, which under-went a shift toward higher 2q values in a compositional range of0.69� x� 0.54 before showing a stabilization at lower sodiumcontents. The evolution of the lattice parameters as respectivefunctions of the sodium amount in the phase P2-NaxCo2/3Mn2/9Ni1/9O2 during charge is presented in Fig. 3d.

The lattice parameter “a” represents the Metal-Metal distanceand is mainly affected by the oxidation/reduction of the transitionmetals. A clear increase for it was noticed between the startingcomposition x¼ 0.69, in correspondence of the open circuit voltage(OCV), and the sodium content x¼ 0.54 (i.e. 3.5 V vs. Naþ/Na). Its

Fig. 1. (a) SEM image showing the surface morphology of the pristine Na2/3Co2/3Mn2/9Ni1/9O2 powders. (bef) Elemental maps of Na, O, Co, Ni and Mn. The respective electronicedges are indicated in each panel.

M. Valvo et al. / Electrochimica Acta 276 (2018) 142e152 145

decrease is mainly caused by an increase in the oxidation state ofthe transition metals to compensate for the charge lost during theNaþ extraction. Indeed, it was found via HAXPES that the oxidationof both Co3þ and Ni3þ ions occurred in this range of potentials [42].However, this decrease for “a” is mostly attributed to the oxidationof Ni3þ (0.060 nm) to Ni4þ (0.048 nm), as the ionic radii of Co4þ

(0.0530 nm) and Co3þ (0.0545 nm) are very close to each other andthis hardly affects the M-M distance. This is confirmed by the sta-bility of the “a” values when x� 0.54 (i.e. �3.5 V vs. Naþ/Na), for

which the totality of the Ni3þ ions (i.e. 1/9) is expected to beoxidized and a further contribution by the oxidation of the Co3þ

ions should not cause major variations for this lattice parameter. Itshould be mentioned that Mn4þ ions are partially reduced to Mn3þ

during the first discharge below the initial OCV value and that thismechanism is totally reversible, thus accounting for an overallcontribution of all the transition metals in the electrochemicalprocesses [42]. Hence, the inflections and plateaus characterizingthe galvanostatic cycling curves of NaxCo2/3Mn2/9Ni1/9O2 cannot

Fig. 2. (a) Illustration of P2-NaxCo2/3Mn2/9Ni1/9O2 structure. (b) In operando XRD patterns of the NaxCo2/3Mn2/9Ni1/9O2 electrode material recorded during complete charge in adedicated Na half-cell. (c) Corresponding cycling curve for the galvanostatic charge with rotated ‘voltage’ and ‘capacity’ axes. Note that the cell capacity increases as the value of “x”in NaxCo2/3Mn2/9Ni1/9O2 decreases from 0.69 at the OCV to 0.39 in correspondence of 4.0 V vs. Naþ/Na. The colors of these points in (c) refer to the respective diffraction patterns in(b), which were collected at such voltage values. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3. Simultaneous evolution of each diffraction peak (a) (002), (b) (004) and (c) (100) upon charge. (d) Related evolution of lattice parameters “a” and “c” as respective functionsof sodium content in NaxCo2/3Mn2/9Ni1/9O2.

M. Valvo et al. / Electrochimica Acta 276 (2018) 142e152146

univocally be related to specific red-ox couples. On the other hand,the lattice parameter “c”, which is equal to twice the interlayerdistance, shows an increase with the decrease of the sodium con-tent in the material during the charge. As the amount of sodiumions sandwiched between the MO2 sheets decreases, the

electrostatic repulsion between the (MO2)n layers increases on bothsides causing their separation and, consequently, an overall in-crease of “c”. Possible effects of Na-ordering in the interlayerspacing between (MO2)n sheets due to the Metal-Metal (M-M)distance in this P2-type structure might also play a role, although

Fig. 4. Cyclic voltammograms of a composite Na2/3Co2/3Mn2/9Ni1/9O2 electrode cycledin a Na half-cell between 2.00 and 4.20 V vs. Naþ/Na at increasing scan rates from 0.01to 0.64mVs-1. The peaks labelled ‘I’, ‘II’, ‘III’ correspond to oxidation processes, whereas‘IV’, ‘V’, ‘VI’ are their reductive counterparts. Note that all the cycles presented in thegraph were obtained after an initial cycle.

M. Valvo et al. / Electrochimica Acta 276 (2018) 142e152 147

the M-M distance is mainly controlled by the oxidation state of thetransition metals [42] in this compound. In O3-type oxides, thepresence of Naþ in the interlayers affects the M-M distance and, assuch, different Liþ and Naþ cations show distinctive M-M distancesin LiCoO2 and NaCoO2, for which the influence of NaþeNaþ repul-sive interaction was reported earlier [59].

Interestingly, the lattice parameter “c” appeared stable after thecomposition x¼ 0.49 was reached and its trend upon subsequentde-sodiation followed that of the M-M distance (i.e. “a”) in aspecular way. Its evolution is mainly influenced by two compo-nents: i) a steric interaction that induces its decrease upon de-sodiation and ii) an electrostatic factor that causes an increase ofthe interlayer distance as a result of the O-O repulsion upon Naþ

removal. Therefore, the electrostatic contribution appeared to playa possible major role for a sodium content x> 0.49, whereas forx< 0.49 it is reasonable to expect that the steric factor dominated.Regardless of the variations observed for the lattice parameters, theP2-type structure of this compound remained clearly consistentupon charge without undergoing any major rearrangement.

3.3. Electrochemical analysis via cyclic voltammetry

The electrochemical response of a composite Na2/3Co2/3Mn2/

9Ni1/9O2 electrode subjected to cyclic voltammetry (CV) withincreasing sweep rates from 0.01 to 0.64mVs-1 between 2.00 and4.20 V vs. Naþ/Na is presented in Fig. 4.

In the series of cycles carried out at various scan rates, theevolution of three distinctive red-ox couples can be observed. Thesered-ox peaks, which are labelled here with roman numbers, arecharacteristic of this compound and correspond to various stagesupon electrochemical de-sodiation and sodiation. The possibleorigin and attribution of such features was previously discussed inliterature [41], though their fundamental nature still remainsmatter of debate [42,43].

The curves in Fig. 4 display typical red-ox components (i.e.current peaks at potentials >3.00 V) combined with characteristiccapacitive features (i.e. flat lozenge-like curves at lower voltages).Both properties are also present in the initial CV cycles of the

electrode (see Fig. S4 in SI) and significantly increased in Fig. 4 uponraising the sweep rate. The shape of the voltammograms in cor-respondence of the different sweep rates was progressively modi-fied in proximity of the upper and lower cut-off voltages. As thesweep rate was increased, the ending part of the curves around4.20 V became progressively steeper, as if the oxidation processeshad more difficulty in being completed. This could be inferred fromthe lack of a short tail for the peak ‘III’ at scan rates higher than0.16mVs-1 and from the absence of a flat ending. This feature isevident for the cycles at higher scan rates, where also a simulta-neous shift in potential is observed for the red-ox peaks. The latteris consistent with the fact that both oxidative and reductive sig-natures in the correspondent half cycles shifted respectively to-wards higher (i.e. for the anodic peaks) and lower (i.e. for thecathodic ones) voltages with increasing scan rates. This type ofbehaviour is usually related to processes limited by the rate ofcharge transfer and/or ohmic drop [46,47]. In particular, if a sub-stantial ohmic drop is present in the system, a linear relationshipbetween the peak potential (Ep) and the scan rate (n) can be ex-pected on the basis of Ohm's law [47]. Indeed, resistive aspectsneed to be considered in Na half-cells, as an additional polarizationderives from the presence of metallic sodium and its unstableinterface upon electrochemical reactions [60,61].

Nevertheless, it can be seen that in all the graphs of Fig. 5 theevolution of the peak potentials, Ep, did not exhibit a clear lineartrend with the scan rate, while the opposite occurred for the peakcurrents, Ip. The peak currents displayed an evident linear rela-tionship with the scan rates (see also the r-values for jIpj in Table 1),suggesting a significant capacitive response for this electrode undersuch conditions, as explained more in detail in the following.

The various contributions associated with the electrochemicalprocesses taking place in an electrode subjected to CV at varioussweep rates can be analysed by the general power law [48e50]:

I ¼ avb (1)

where I is the experimental value of the measured current, n is theapplied scan rate, while a and b are varying parameters that can bedetermined. By rewriting eq. (1) as (I/a)¼ n b and by taking thelogarithms of both members, it is possible to calculate the b-valuein such expression. The latter can be obtained by plotting the log-arithm of themeasured current vs. the logarithm of the scan rate foreach set of data points and by extrapolating the resulting slope. Inthis way, it is possible to point out two distinctive ideal regimes forthe electrochemical processes. The first one is characterized byb¼ 0.5, which corresponds to a regime where the processes areentirely diffusion-limited (i.e. semi-infinite diffusion), whereas theother one (i.e. b¼ 1.0) refers to the other extreme case in which thereactions follow a complete capacitor-like kinetics. Under such acondition the current is given by Refs. [48,49]:

I ¼ CdAv (2)

where Cd is the electrode capacitance and A represents the surfacearea of the electrode material. Accordingly, the contribution to thecurrent in eq. (2) is purely capacitive and results proportional to n.

Vice versa, when b¼ 0.5 the current follows another character-istic equation [50]:

I ¼ nFC*AD1=2ðanF=RTÞ1=2p1=2cðbtÞv1=2 (3)

where n is the number of electrons exchanged in the electro-chemical reaction, F is the Faraday constant, C* is the surface con-centration of the electrode material, D is the chemical diffusioncoefficient, a is the transfer coefficient, R is themolar gas constant, T

Fig. 5. Composite graphs showing the trends of the peak currents (triangles) and peak potentials (squares) as a function of the applied scan rates. Each graph from (a) to (f) refers toa specific red-ox feature, whose values have been extracted from the corresponding peaks in Fig. 4. Note in all the cases the linear trend of the peak currents vs. the scan rates, with asimultaneous non-linear behaviour for the related peak potentials.

Table 1Synoptic table of the various parameters obtained from different linear fits regarding the red-ox peak features presented in Fig. 4. Note that the reported values refer to thefeatures found in the series of graphs shown in Figs. 5 and 6.

Peak feature r-value (jIpj fit) Slope jIpj fit r-value (jlnjIpjj fit) b r-value (jIpj/n1/2 fit) k1 k2 k2/k1

I 0.99947 1.01373 0.99960 0.90550 0.99929 0.95600 0.06337 0.0663II 0.99945 1.16039 0.99996 0.90934 0.99914 1.09362 0.07559 0.0691III 0.99945 1.16039 0.99935 0.87647 0.99904 1.18834 0.11017 0.0927IV 0.99611 0.90226 0.99965 0.86455 0.98933 0.81374 0.11765 0.1446V 0.99833 1.03540 0.99980 0.93556 0.99582 1.00084 0.05985 0.0598VI 0.99831 0.89434 0.99931 0.90307 0.99654 0.84629 0.06399 0.0756

M. Valvo et al. / Electrochimica Acta 276 (2018) 142e152148

Fig. 6. Graphs for determining the experimental values employed in the analysis of theelectrochemical behaviour of the electrode. (a) Logarithmic plot to determine the b-values linked to the rate-limiting factors of the electrochemical processes. (b) Plotreferring to the square root of the scan rate used to evaluate the k1 and k2 parametersrespectively related to capacitive and intercalation contributions for each red-oxfeature.

M. Valvo et al. / Electrochimica Acta 276 (2018) 142e152 149

the temperature and c(bt) represents a function that correspondsto the normalized current for a totally irreversible system, asindicated by the CV response. Hence, the contribution to the cur-rent in eq. (3) is merely diffusive due to faradaic intercalationprocesses, being the latter proportional to n 1/2.

Here the different contributions arising from capacitive anddiffusive mechanisms have been analysed on the basis of an earlierapproach [49] where the current response at a fixed voltage, I(V),can be expressed as a combination of the characteristic terms n andn 1/2 [49,51]:

IðVÞ ¼ k1vþ k2v1=2 (4)

where k1n represents the capacitive element, whereas k2n1/2 ac-

counts for the diffusive intercalation component. The quantities k1and k2 are associated parameters that can be further determined viaan analysis of the experimental data. The above equation canconveniently be rewritten as:

IðVÞ.v1=2 ¼ k1v

1=2 þ k2 (5)

where the related slope, k1, and y-intercept, k2, can be obtained byfitting the respective data points. The resulting values enable aquantitative evaluation of the different contributions and their ra-tio at a fixed voltage.

From Fig. 6a it is observed that for all the red-ox peaks verysimilar slopes were obtained, independently of their state ofreduction/oxidation. Typical b-values, yielded by the linear fits ofthe logarithm of the currents for the various peaks, were aroundz0.9 (see Table 1).

The data points in Fig. 6a closely followed a linear trend (see r-values for jlnjIpjj fit in Table 1) and their b-values indicate that for allthe peak currents a clear part of the charge storage is derived fromcapacitive-like contributions. It is also seen that for the red-oxpeaks at the highest voltage (i.e. ‘III’ and ‘IV’) the b-values re-ported in Table 1 are slightly below 0.9, thus suggesting that minorfaradaic contributions due to intercalation played a role as well.These considerations are further strengthened by the results inFig. 6b, where the values for k1 and k2 were determined by thelinear fits of the various sets of data points.

From the data reported in Table 1 for all the couples of peaks it isclear that k1>> k2, being the values of k2 slightly higher for thefeatures ‘III’ and ‘IV’, in line with the precedent finding. From thispreliminary analysis, it can be argued that the electrochemicalbehaviour of this electrode material is mainly controlled, underthese conditions, by capacitive-like processes and that suchmechanisms dominate the charge storage even in correspondenceof the various peak currents. This is quite unexpected, as the latterare typically ascribed to faradaic insertion/de-insertion of Naþ andsimultaneous changes in oxidation state for the transition metals inthis type of layered compounds [38]. Nevertheless, the fact that, onaverage, the contribution to the currents through these peaksoriginates only for z20% from diffusive faradaic processes (i.e.b¼ 0.5 ideally corresponds to 0% capacitive behaviour and 100%diffusion characteristics, whereas b¼ 1 represents its comple-mentary case) cannot be neglected (see Table 1). This finding isinteresting, as capacitor-like features are often associated withlarge interface areas and clearly manifest when the electrode ma-terials are subjected to quick variations of the applied voltage (e.g.in supercapacitors), while none of these conditions are fulfilledhere, since low and moderate scan rates were utilized to favour thecompletion of the reactions in these micron-sized active particles.

Conversely, a sheer capacitive mechanism due to electric doublelayer (EDL) charging is unlikely here for a series of reasons. First, the

active material has an evident bulk nature (see Fig. 1a) and itscharacteristic values of specific surface area are low (i.e.<10m2 g�1). This means that, on average, a pure EDL componentcould play only a minor role, since a high surface area is normallyrequired to enhance this type of capacitive charge storage process.Secondly, the carbon black in the electrode formulation can giverise to a larger electrochemically active area (i.e. characteristicspecific surface areas for Super P carbon lie in the range of60e65m2 g�1) and thereby could account for more significant EDLcontributions. However, the carbon content in the compositecoating is limited to 20% of that of Na2/3Co2/3Mn2/9Ni1/9O2, while itcannot sensibly contribute to the observed peaks above 3.0 V, sinceits characteristic electrochemical activity vs. Naþ/Na is limited tolower voltages (e.g.�1.2 V vs. Naþ/Na) [52]. Sparse surface traces ofsodium carbonates and sodium fluoride [42] could play only anegligible role in these electrochemical reactions, since most oftheir dissolution (i.e. oxidation) was observed only at higher volt-ages (i.e. 4.5 V vs. Naþ/Na). The use of such high potentials is known

M. Valvo et al. / Electrochimica Acta 276 (2018) 142e152150

to cause possible electrolyte degradation and, for this reason, it wasavoided in this study. In any case, approximate theoretical values ofpurely capacitive currents, IEDL, in correspondence of the differentsweep rates were estimated for this composite electrode via eq. (2),as described more in detail in the separate section S.1 in the SIsection. The series of calculated values for IEDL at each n are listedand compared in Table S2 (see SI) with the actual experimentalvalues, Ip, which were recorded for each peak during the CV mea-surements. It is clear from Table S2 that the values of Ip are muchlarger than those estimated for IEDL for each peak in correspondenceof all the scan rates. Hence, this leaves open only one main route toexplain the high specific capacitance per unit area observed for thepeak currents (see Fig. 7 later) in this composite electrode: pseu-docapacitance. This intriguing phenomenon has attracted moreand more attention in the last years, not only for supercapacitors[53], but also for Li- and Na-ion batteries [54e57], especially for anumber of different oxides [49,58] undergoing insertion mecha-nisms of these ions.

Fig. 7 presents the evolution of estimated (i.e. approximate)specific areal capacitances for each CV peak as a respective function

Fig. 7. Evaluation of approximate specific capacitance per unit area of the compositeNa2/3Co2/3Mn2/9Ni1/9O2 electrode for each peak as a function of (a) the scan rate and (b)the respective reaction potentials. Open symbols refer to oxidative processes, whereassolid ones are related to their reductive counterparts. Note that the capacitance valuesare normalized with respect to a slightly overestimated total specific surface area ofthe active material and the carbon black additive.

of the scan rate (a) and the voltage recorded for each individual red-ox feature in correspondence of the various sweep rates (b).

It is seen that the specific areal capacitance is not constant foreach peak, but tends to level off with increasing sweep rates afteran initial drop in correspondence of the lowest scan rates. Most ofthe values estimated for the specific areal capacitance lie in theapproximate range of 0.7e1.2mFcm�2, which is order of magni-tudes larger than 10e50 mFcm�2 [53]. This interval of tens ofmFcm�2 is often documented for conductive surfaces in contactwith aqueous electrolytes (i.e. with organic solvents such valuesshould be further reduced) [53]. Therefore, Na2/3Co2/3Mn2/9Ni1/9O2appears to possess an intrinsic ability of storing charge in adistinctive pseudo-capacitive way. This scenario can still fit withthe overall picture of characteristic Naþ insertion/extraction in thelayered structure of this compound, if one recalls that a strongrepulsive electrostatic interaction exists between two neighbour-ing metal oxide slabs [41] and that charged defects can easily begenerated in its lattice, too. The latter observation agrees well alsowith the fact that a partial removal of Naþ from its pristine struc-ture readily occurs in contact with air yielding clear surface Na2CO3deposits (see Fig. S2 in SI), which can also cause sliding of the metaloxide slabs [36]. These processes, in turn, imply a necessary chargecompensation mechanism to hold overall electroneutrality acrossthe material. Indeed, a variation of the sodium concentration in P2-NaxCoO2 generates various single-phase domains with distinctNaþ/vacancy patterns [35].

From Fig. 7b it can be noticed that also some subtle differencesexist in the pseudo-capacitive behaviour of the three couples ofpeaks (i.e. I-VI, II-V, III-IV). The asymmetry for the distribution ofassociated data points (solid vs. open) appears to be enhanced withincreasing peak potential vs. Naþ/Na, together with the earliermentioned shift of the peak positions with rising scan rates. Theseparation in potential for each couple of peaks, DEp, grows withincreasing sweep rates. Possible limiting factors influencing theperformance of this type of electrode can be better understood bychecking the dependence of DEp on the currents involved in theseprocesses.

Fig. 8 shows the difference in potential existing for each coupleof peaks as a function of the average peak current recorded for therespective cathodic and anodic processes.

Fig. 8. Comparative graph of the difference in potential existing between each coupleof redox peaks as a function of the average (i.e. red-ox) peak current obtained atincreasing sweep rates. Note the linear trend of DEp for all the couples of related peaks.

Table 2Pearson's values for the linear fits presented in Fig. 8 and estimated resistances fromthe slopes obtained from the same fit for each couple of peaks.

Peak Couple r-value (DEp fit) Slope [kU]

I-VI 0.998 0.310II-V 0.994 0.317III-IV 0.994 0.326

M. Valvo et al. / Electrochimica Acta 276 (2018) 142e152 151

The trend of the data points in the three panels of Fig. 8 suggestsan overall linear-like relationship between DEp and the average ofthe absolute values of the cathodic/anodic currents for each peakcouple, thus highlighting Ohmic losses as main contributors to theobserved voltage separation. Although a neat linear relationship iseffectively obtained only for the peak couple I-VI, it can be noticedthat in all the cases the extrapolated values of the slopes indicate asignificant resistance for all the peak pairs (see Table 2) with typicalvalues exceeding 300U.

Considering the active mass loading in these electrodes, theirthickness and the fact that they were not calendered, it is notsurprising to observe such resistances for the electrochemicalprocesses. This can be further justified by the micron-sized natureof the oxide particles and their electrically insulating surface by-products (e.g. NaF, Na2CO3) in these composite coatings withPVdF, together with unfavourable resistive properties of the metalsodium interface [60,61] upon reaction.

Bearing in mind the pseudo-capacitive contribution to thecharge storage in this layered oxide compound, a possible reduc-tion in size for its crystalline particles can be regarded as certainlyconvenient. A larger electrochemically active interface can signifi-cantly boost the pseudocapacitance andwell-developed crystallitesof this oxide with average sizes below 1 mm should be most likelyaccessible via tailored syntheses. This approach, combined withadditional electrode calendering and local electronic bridging ofthe active particles by carbon black, should contribute to an effec-tive lowering of the overall electrode resistance. In this way,improved electrochemical performances can be enabled withoutsacrificing too much the energy density of the resulting electrodes.

4. Conclusions

To sum up, the electrochemical behaviour of Na2/3Co2/3Mn2/

9Ni1/9O2 was carefully investigated in Na half-cells by means of adedicated cyclic voltammetry study and its structural stability uponsodium extraction was checked through in operando XRD coupledwith galvanostatic measurements. The analysis of the various cur-rents and potentials characterizing the peaks in correspondence ofthe different sweep rates suggests that the picture of classicalinsertion/extraction of Naþ ions into/from this layered oxide doesnot fully capture the intrinsic electrochemical properties displayedby this compound in composite electrodes. The examination of theresults of the various linear fits, to determine whether the currentsfor the peakswere purely faradaic (i.e. due to insertion/de-insertionof Naþ ions diffusing into/from the lattice structure), or intrinsicallycapacitive, indicates that significant pseudo-capacitive features areat play in the electrochemical reactions of this complex oxide. Thecontribution of pure electric double layer charging as capacitivestorage mechanismwas evaluated in this context and it was shownthat this alone could not justify the high values of specific arealcapacitance obtained in correspondence of all the peaks. Pseudo-capacitance, which is most likely associated with electrostatic in-teractions present in the layered lattice structure, as well as withthe generation of possible charge compensation mechanisms for apartial Naþ removal from the crystallites to form surface by-products (i.e. Na2CO3, NaF), represents here the most logical

mechanism for rationalizing the charge storage displayed by thiselectrode system.

The rather stable cycle performances demonstrated in ourearlier study [41] for this complex oxide can be related then to twomain factors: i. a good stability of its P2-type structure for thosereactions occurring between 2.0 and 4.2 V vs. Naþ/Na, as proved bythe previous in operando XRD measurements and ii. highlyreversible pseudocapacitive processes brought about by the metaloxide slab framework, as well as by the defects (e.g. vacancies)generated in its lattice and related charge compensation mecha-nisms arising from charge neutrality requirements. Most likely,such phenomena should apply also to other Na-containing metaloxide compounds relying on similar layered structures, as the basicphysical and chemical interactions should not sensibly depart fromthose considered so far.

Finally, possible limiting factors existing for this type of com-posite electrodeswere taken into account as well. It was shown thatreducing the overall resistance experienced by the electrode duringthe electrochemical processes is of paramount importance to ach-ieve improved performances. Reducing the electrical resistance ofthese composite electrodes by a careful selection of smaller crys-tallite sizes (e.g. few hundreds of nanometers), as well as promotingtheir effective connection in the resulting coatings by well-adhering conductive additives are crucial aspects that need to beaddressed to enhance the performances of this type of positiveelectrode for NIBs. In this way, quicker charges/discharges can besupported by better exploiting the pseudocapacitive featureswithout the penalty of a high electrode resistance.

Acknowledgements

The authors are grateful to H. Eriksson for valuable technicalsupport. B. Philippe and A. Liivat are acknowledged for useful dis-cussions and comments. M. Valvo gratefully acknowledges thefinancial support of the Swedish Research Council for Environment,Agricultural Sciences and Spatial Planning (FORMAS) via the per-sonal grant no. 245-2014-668. STandUp for Energy is alsoacknowledged. We would like to thank Siham Difi and Pierre-Emmanuel Lippens for in operando XRD measurements.

Appendix A. Supplementary data

Supplementary data related to this article can be found athttps://doi.org/10.1016/j.electacta.2018.04.150.

Conflicts of interest

The authors declare that there are no conflicts of interest.

References

[1] V. Palomares, P. Serras, I. Villaluenga, K.B. Hueso, J. Carretero-Gonz�alez,T. Rojo, Na-ion batteries, recent advances and present challenges to becomelow cost energy storage systems, Energy Environ. Sci. 5 (2012) 5884e5901.

[2] B.L. Ellis, L.F. Nazar, Sodium and sodium-ion energy storage batteries, Curr.Opin. Solid State Mater. Sci. 16 (2012) 168e177.

[3] N. Yabuuchi, K. Kubota, M. Dahbi, S. Komaba, Research development onsodium-ion batteries, Chem. Rev. 114 (2014) 11636e11682.

[4] M. Dahbi, N. Yabuuchi, K. Kubota, K. Tokiwa, S. Komaba, Negative electrodesfor Na-ion batteries, Phys. Chem. Chem. Phys. 16 (2014) 15007e15028.

[5] A. Darwiche, C. Marino, M.T. Sougrati, B. Fraisse, L. Stievano, L. Monconduit,Better cycling performances of bulk Sb in Na-ion batteries compared to Li-ionsystems: an unexpected electrochemical mechanism, J. Am. Chem. Soc. 134(2012) 20805e20811.

[6] P.K. Allan, J.M. Griffin, A. Darwiche, O.J. Borkiewicz, K.M. Wiaderek,K.W. Chapman, A.J. Morris, P.J. Chupas, L. Monconduit, C.P. Grey, Trackingsodium-antimonide phase transformations in sodium-ion anodes: insightsfrom operando pair distribution function analysis and solid-state NMR spec-troscopy, J. Am. Chem. Soc. 138 (2016) 2352e2365.

M. Valvo et al. / Electrochimica Acta 276 (2018) 142e152152

[7] L. Bodenes, A. Darwiche, L. Monconduit, H. Martinez, The Solid ElectrolyteInterphase a key parameter of the high performance of Sb in sodium-ionbatteries: comparative X-ray Photoelectron Spectroscopy study of Sb/Na-ionand Sb/Li-ion batteries, J. Power Sources 273 (2014) 14e24.

[8] A. Darwiche, M. Toiron, M.T. Sougrati, B. Fraisse, L. Stievano, L. Monconduit,Performance and mechanism of FeSb2 as negative electrode for Na-ion bat-teries, J. Power Sources 280 (2015) 588e592.

[9] A. Darwiche, M.T. Sougrati, B. Fraisse, L. Stievano, L. Monconduit, Facile syn-thesis and long cycle life of SnSb as negative electrode material for Na-ionbatteries, Electrochem. Commun. 32 (2013) 18e21.

[10] P. Ge, M. Fouletier, Electrochemical intercalation of sodium in graphite, SolidState Ionics 28e30 (1988) 1172e1175.

[11] D.A. Stevens, J.R. Dahn, The mechanisms of lithium and sodium insertion incarbon materials, J. Electrochem. Soc. 148 (2001) A803eA811.

[12] V.L. Chevrier, G. Ceder, Challenges for Na-ion negative electrodes,J. Electrochem. Soc. 158 (2011) A1011eA1014.

[13] B. Huang, K. Tai, M. Zhang, Y. Xiao, S.J. Dillon, Comparative study of li and naelectrochemical reactions with iron oxide nanowires, Electrochim. Acta 118(2014) 143e149.

[14] M. Valvo, B. Philippe, F. Lindgren, C.W. Tai, K. Edstr€om, Insight into the pro-cesses controlling the electrochemical reactions of nanostructured iron oxideelectrodes in Li- and Na-half cells, Electrochim. Acta 194 (2016) 74e83.

[15] N. Yabuuchi, M. Kajiyama, J. Iwatate, H. Nishikawa, S. Hitomi, R. Okuyama,R. Usui, Y. Yamada, S. Komaba, P2-type Nax[Fe1/2Mn1/2O2] made from earth-abundant elements for rechargeable Na batteries, Nat. Mater. 11 (2012)512e517.

[16] H. Yoshida, N. Yabuuchi, K. Kubota, I. Ikeuchi, A. Garsuch, M. Schulz-Dobrick,S. Komaba, P2-type Na2/3Ni1/3Mn2/3-xTixO2 as a new positive electrode forhigher energy Na-ion batteries, Chem. Commun. 50 (2014) 3677e3680.

[17] E. Talaie, V. Duffort, H.L. Smith, B. Fultz, L.F. Nazar, Structure of the highvoltage phase of layered P2-Na2/3-z[Mn1/2Fe1/2]O2 and the positive effect of Nisubstitution on its stability, Energy Environ. Sci. 8 (2015) 2512e2523.

[18] W.K. Pang, S. Kalluri, V.K. Peterson, N. Sharma, J. Kimpton, B. Johannessen,H.K. Liu, S.X. Dou, Z.P. Guo, Interplay between electrochemistry and phaseevolution of the P2-type Nax(Fe1/2Mn1/2)O2 cathode for use in sodium-ionbatteries, Chem. Mater. 27 (2015) 3150e3158.

[19] B.L. Ellis, W.R.M. Makahnouk, Y. Makimura, K. Toghill, L.F. Nazar,A multifunctional 3.5V iron-based phosphate cathode for rechargeable bat-teries, Nat. Mater. 6 (2007) 749e753.

[20] Y. Lu, L. Wang, J. Cheng, J.B. Goodenough, Prussian blue: a new framework ofelectrode materials for sodium batteries, Chem. Commun. 48 (2012)6544e6546.

[21] R.J. Cl�ement, J. Billaud, A. Robert Armstrong, G. Singh, T. Rojo, P.G. Bruce,C.P. Grey, Structurally stable Mg-doped P2-Na2/3Mn1-yMgyO2 sodium-ionbattery cathodes with high rate performance: insights from electrochemical,NMR and diffraction studies, Energy Environ. Sci. 9 (2016) 3240e3251.

[22] Y. Bai, L. Zhao, C. Wu, H. Li, Y. Li, F. Wu, Enhanced sodium ion storage behaviorof P2-Type Na2/3Fe1/2Mn1/2O2 synthesized via a chelating agent assisted route,ACS Appl. Mater. Interfaces 8 (2016) 2857e2865.

[23] S. Komaba, C. Takei, T. Nakayama, A. Ogata, N. Yabuuchi, Electrochemicalintercalation activity of layered NaCrO2 vs. LiCrO2, Electrochem. Commun. 12(2010) 355e358.

[24] C. Delmas, M. M�en�etrier, L. Croguennec, I. Saadoune, A. Rougier, C. Pouillerie,G. Prado, M. Grüne, L. Fourn�es, An overview of the Li(Ni,M)O2 systems: syn-theses, structures and properties, Electrochim. Acta 45 (1999) 243e253.

[25] J. Zhao, L. Zhao, N. Dimov, S. Okada, T. Nishida, Electrochemical and thermalproperties of a-NaFeO2 cathode for Na-ion batteries, J. Electrochem. Soc. 160(2013) A3077eA3081.

[26] S. Okada, Y. Takahashi, T. Kiyabu, T. Doi, J.-I. Yamaki, T. Nishida, Layeredtransition metal oxides as cathodes for sodium secondary battery, MeetingAbstracts, The Electrochemical Society (2006), pp. 201e201.

[27] C. Didier, M. Guignard, C. Denage, O. Szajwaj, S. Ito, I. Saadoune, J. Darriet,C. Delmas, Electrochemical Na-deintercalation from NaVO2, Electrochem.Solid State Lett. 14 (2011) A75eA78.

[28] B.M. de Boisse, D. Carlier, M. Guignard, L. Bourgeois, C. Delmas, P-2-NaxMn1/

2Fe1/2O2 phase used as positive electrode in Na batteries: structural changesinduced by the electrochemical (De)intercalation process, Inorg. Chem. 53(2014) 11197e11205.

[29] D. Hamani, M. Ati, J.M. Tarascon, P. Rozier, NaxVO2 as possible electrode forNa-ion batteries, Electrochem. Commun. 13 (2011) 938e941.

[30] D. Kim, E. Lee, M. Slater, W. Lu, S. Rood, C.S. Johnson, Layered Na[Ni1/3Fe 1/

3Mn1/3]O2 cathodes for Na-ion battery application, Electrochem. Commun. 18(2012) 66e69.

[31] M. D'Arienzo, R. Ruffo, R. Scotti, F. Morazzoni, C.M. Mari, S. Polizzi, LayeredNa0.71CoO2: a powerful candidate for viable and high performance Na-bat-teries, Phys. Chem. Chem. Phys. 14 (2012) 5945e5952.

[32] Z. Lu, J.R. Dahn, In situ x-ray diffraction study of P2-Na2/3[Ni1/3Mn2/3]O2,J. Electrochem. Soc. 148 (2001) A1225eA1229.

[33] X. Chen, X. Zhou, M. Hu, J. Liang, D. Wu, J. Wei, Z. Zhou, Stable layered P3/P2Na0.66Co0.5Mn0.5O2 cathode materials for sodium-ion batteries, J. Mater.Chem. 3 (2015) 20708e20714.

[34] C. Delmas, J.J. Braconnier, C. Fouassier, P. Hagenmuller, Electrochemicalintercalation of sodium in NaxCoO2 bronzes, Solid State Ionics 3e4 (1981)

165e169.[35] R. Berthelot, D. Carlier, C. Delmas, Electrochemical investigation of the P2-

NaxCoO2 phase diagram, Nat. Mater. 10 (2011) 74e80.[36] M. Sathiya, K. Hemalatha, K. Ramesha, J.M. Tarascon, A.S. Prakash, Synthesis,

structure, and electrochemical properties of the layered sodium insertioncathode material: NaNi1/3Mn1/3Co1/3O2, Chem. Mater. 24 (2012) 1846e1853.

[37] D. Buchholz, A. Moretti, R. Kloepsch, S. Nowak, V. Siozios, M. Winter,S. Passerini, Toward Na-ion batteries - synthesis and characterization of anovel high capacity Na ion intercalation material, Chem. Mater. 25 (2013)142e148.

[38] D. Yuan, W. He, F. Pei, F. Wu, Y. Wu, J. Qian, Y. Cao, X. Ai, H. Yang, Synthesisand electrochemical behaviors of layered Na0.67[Mn0.65Co0.2Ni0.15]O2 micro-flakes as a stable cathode material for sodium-ion batteries, J. Mater. Chem. 1(2013) 3895e3899.

[39] D. Buchholz, L.G. Chagas, M. Winter, S. Passerini, P2-type layeredNa0.45Ni0.22Co0.11Mn0.66O2 as intercalation host material for lithium and so-dium batteries, Electrochim. Acta 110 (2013) 208e213.

[40] D. Buchholz, L.G. Chagas, C. Vaalma, L. Wu, S. Passerini, Water sensitivity oflayered P2/P3-NaxNi0.22Co0.11Mn0.66O2 cathode material, J. Mater. Chem. 2(2014) 13415e13421.

[41] S. Doubaji, M. Valvo, I. Saadoune, M. Dahbi, K. Edstr€om, Synthesis and char-acterization of a new layered cathode material for sodium ion batteries,J. Power Sources 266 (2014) 275e281.

[42] S. Doubaji, B. Philippe, I. Saadoune, M. Gorgoi, T. Gustafsson, A. Solhy,M. Valvo, K. Edstr€om, Passivation layer and cathodic redox reactions insodium-ion batteries probed by HAXPES, ChemSusChem 9 (2016) 97e108.

[43] N. Sabi, S. Doubaji, K. Hashimoto, S. Komaba, K. Amine, A. Solhy, B. Manoun,E. Bilal, I. Saadoune, Layered P2-Na2/3Co1/2Ti1/2O2 as a high-performancecathode material for sodium-ion batteries, J. Power Sources 342 (2017)998e1005.

[44] Y. Mo, S.P. Ong, G. Ceder, Insights into diffusion mechanisms in P2 layeredoxide materials by first-principles calculations, Chem. Mater. 26 (2014)5208e5214.

[45] Q. Huang, M.L. Foo, J.W. Lynn, H.W. Zandbergen, G. Lawes, Y. Wang, B.H. Toby,A.P. Ramirez, N.P. Ong, R.J. Cava, Low temperature phase transitions andcrystal structure of Na0.5CoO2, J. Phys. Condens. Matter 16 (2004) 5803e5814.

[46] A.K. Hjelm, G. Lindbergh, A. Lundqvist, Investigation of LiMn2O4 cathodes foruse in rechargeable lithium batteries by linear sweep voltammetry - Part I.Theoretical study, J. Electroanal. Chem. 506 (2001) 82e91.

[47] S. Nordlinder, L. Nyholm, T. Gustafsson, K. Edstr€om, Lithium insertion intovanadium oxide nanotubes: electrochemical and structural aspects, Chem.Mater. 18 (2006) 495e503.

[48] H. Lindstr€om, S. S€odergren, A. Solbrand, H. Rensmo, J. Hjelm, A. Hagfeldt,S.E. Lindquist, Liþ ion insertion in TiO2 (anatase). 2. Voltammetry on nano-porous films, J. Phys. Chem. B 101 (1997) 7717e7722.

[49] J. Wang, J. Polleux, J. Lim, B. Dunn, Pseudocapacitive contributions to elec-trochemical energy storage in TiO2 (anatase) nanoparticles, J. Phys. Chem. C111 (2007) 14925e14931.

[50] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals and Appli-cations, Electrochemical Methods: Fundamentals and Applications, JohnWiley & Sons, NY, 1980.

[51] T.C. Liu, W.G. Pell, B.E. Conway, S.L. Roberson, Behavior of molybdenum ni-trides as materials for electrochemical capacitors comparison with rutheniumoxide, J. Electrochem. Soc. 145 (1998) 1882e1888.

[52] R. Alc�antara, J.M. Jim�enez-Mateos, P. Lavela, J.L. Tirado, Carbon black: apromising electrode material for sodium-ion batteries, Electrochem. Com-mun. 3 (2001) 639e642.

[53] B.E. Conway, V. Birss, J. Wojtowicz, The role and utilization of pseudocapaci-tance for energy storage by supercapacitors, J. Power Sources 66 (1997) 1e14.

[54] H.S. Kim, J.B. Cook, H. Lin, J.S. Ko, S.H. Tolbert, V. Ozolins, B. Dunn, Oxygenvacancies enhance pseudocapacitive charge storage properties of MoO3�x,Nat. Mater. 16 (2017) 454e460.

[55] C. Chen, Y. Wen, X. Hu, X. Ji, M. Yan, L. Mai, P. Hu, B. Shan, Y. Huang, Naþ

intercalation pseudocapacitance in graphene-coupled titanium oxideenabling ultra-fast sodium storage and long-term cycling, Nat. Commun. 6(2015).

[56] N. Louvain, A. Henry, L. Daenens, B. Boury, L. Stievano, L. Monconduit, On theelectrochemical encounter between sodium and mesoporous anatase TiO2 asa Na-ion electrode, CrystEngComm 18 (2016) 4431e4437.

[57] M. Jayakumar, K. Hemalatha, K. Ramesha, A.S. Prakash, Framework structuredNa4Mn4Ti5O18 as an electrode for Na-ion storage hybrid devices, Phys. Chem.Chem. Phys. 17 (2015) 20733e20740.

[58] T. Brezesinski, J. Wang, S.H. Tolbert, B. Dunn, Ordered mesoporous a-MoO3with iso-oriented nanocrystalline walls for thin-film pseudocapacitors, Nat.Mater. 9 (2010) 146e151.

[59] J.B. Goodenough, K. Mizushima, T. Takeda, Solid-solution oxides for storage-battery electrodes, Jpn. J. Appl. Phys. 19 (1980) 305e313.

[60] Z.W. Seh, J. Sun, Y. Sun, Y. Cui, A highly reversible room-temperature sodiummetal anode, ACS Cent. Sci. 1 (2015) 449e455.

[61] D.I. Iermakova, R. Dugas, M.R. Palacín, A. Ponrouch, On the comparative sta-bility of Li and Na metal anode interfaces in conventional alkyl carbonateelectrolytes, J. Electrochem. Soc. 162 (2015) A7060eA7066.