Journal ofMaterials Chemistry A

PAPER

Publ

ishe

d on

22

July

201

3. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 23

/09/

2013

06:

12:1

7.

View Article OnlineView Journal | View Issue

aWestfalische Wilhelms-Universitat, Insti

Corrensstraße 28/30, 48149 Munster,

uni-muenster.debWestfalische Wilhelms-Universitat, Institut

Corrensstraße 30, 48149 Munster, GermanycWestfalische Wilhelms-Universitat, Institu

Straße 10, 48149 Munster, GermanydHelmholtz-Zentrum Berlin fur Materialien u

14109 Berlin, Germany

Cite this: J. Mater. Chem. A, 2013, 1,11229

Received 8th May 2013Accepted 18th July 2013

DOI: 10.1039/c3ta11821e

www.rsc.org/MaterialsA

This journal is ª The Royal Society of

An investigation of the electrochemical delithiationprocess of carbon coated a-Fe2O3 nanoparticles

Adrian Brandt,a Florian Winter,b Sebastian Klamor,a Frank Berkemeier,c

Jatinkumar Rana,d Rainer Pottgenb and Andrea Balducci*a

The electrochemical lithiation–delithiation of iron oxide is a rather complex process, which is still not

fully understood. In this study we investigated the electrochemical lithiation–delithiation mechanism of

hematite by means of X-ray diffraction (XRD), 57Fe Mossbauer spectroscopy, high-resolution transmission

electron microscopy (HRTEM) and X-ray absorption spectroscopy (XAS). Since the delithiation process

has been so far less investigated, particular attention was dedicated to the characterization of the

chemical species that are formed during this process. The results of this investigation indicated that at

the end of the delithiation process lithium iron oxide a-LiFeO2 is formed. The formation of this

compound may be the explanation for the irreversible capacity loss in the first cycle, which is usually

assigned to the formation of an organic gel-like layer. Based on these results a new charge–discharge

mechanism of hematite in lithium-ion batteries (LIBs) is proposed and discussed.

1 Introduction

For several years metal oxides (MxOy, M ¼ Cu, Fe, Co, Ni, etc.)have been considered as attractive active materials for the real-ization of advanced electrochemical storage devices.1 MxOy

belongs to the so-called conversion materials, and during elec-trochemical cycling a reduction ofMxOy tometal (M) and lithiumoxide (Li2O) takes place, followed by oxidation of M to MxOy.

Among the various MxOy oxides considered so far, hematiteiron oxide (Fe2O3) appears to be one of the most promising. As amatter of fact, Fe2O3 displays a high theoretical capacity (1007mA h g�1) and is cheap, safe and environmentally friendly.Because of these advantageous features, in the past decade theuse of Fe2O3 as an active material in lithium-ion batteries (LIBs)and supercapacitors (SCs) has been widely investigated.2–10 Sofar, however, the introduction of Fe2O3 in real devices has beenhindered by its poor reversibility (compared to the state-of-the-art graphite). In order to overcome this limitation several strat-egies have been proposed, and the use of carbon-coated nano-particles is presently considered as the most promising.11,12

It is interesting to note that even if many studies havebeen dedicated to Fe2O3, the (reversible) lithiation–delithiation

tut fur Physikalische Chemie-MEET,

Germany. E-mail: andrea.balducci@

fur Anorganische und Analytische Chemie,

t fur Materialphysik, Wilhelm-Klemm-

nd Energie GmbH, Hahn-Meitner-Platz 1,

Chemistry 2013

mechanism of this material is still not completely understood.Larcher et al. showed that the lithiation process leads to theformation of very ne iron clusters (10–15 A) dispersed in alithium oxide matrix.13 Both iron and lithium oxides are in turnembedded in a gel-like organic layer, which is formed upondecomposition of the electrolyte. Moreover, they also showedthat the particle size of Fe2O3 (and more in general of MxOy) hasan inuence on the lithiation process since it might lead to theformation of different intermediate phases, e.g. Li0.6Fe2O3 andLi2Fe2O3.13,14 Concerning the delithiation process, the infor-mation available in the literature is not that detailed. It isaccepted that a conversion of Fe to Fe2O3 (or other iron oxidephases) takes place, but it is not completely clear which phasesmight be formed during this oxidation process. Taking thesepoints into account, it is evident that more accurate informa-tion about the delithiation process would be very benecial forthe understanding of the charge–discharge mechanism ofFe2O3 (and in general of MxOy). This information could be inturn of importance for the improvement of the cycle life of thismaterial.

Herein we report a study of the lithiation–delithiationmechanism of carbon-coated Fe2O3 nanoparticles obtainedfrom ferrocene.11 As shown in a previous work, these nano-particles feature high purity and high homogeneity, and theyare therefore a suitable material for the investigation of thelithiation–delithiation mechanism. The lithiation–delithiationprocess of these nanoparticles was investigated using electro-chemical techniques, powder XRD, 57Fe Mossbauer spectros-copy, XAS and HRTEM. Considering the limited informationabout the delithiation process, much attention was dedicated tothis process.

J. Mater. Chem. A, 2013, 1, 11229–11236 | 11229

Journal of Materials Chemistry A Paper

Publ

ishe

d on

22

July

201

3. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 23

/09/

2013

06:

12:1

7.

View Article Online

2 Experimental section2.1 Electrochemical characterization

a-Fe2O3 nano-particles were synthesized using the proceduredescribed in ref. 11. The produced nanoparticles were used forthe realization of composite electrodes containing 85% activematerial (Fe2O3), 10 wt% conducting carbon (Super C65, Tim-cal) and 5 wt% sodium carboxymethyl-cellulose (CMC, WalocelCRT 2000 PPA 12 from Dow Wolff Cellulosics). The electrodeactive material mass loading was ca. 1.3 mg cm�2, and theelectrode area was either 1.13 cm2 or 22 cm2 (pouch cell). Theelectrochemical investigation on the prepared electrodes wascarried out using Swagelok� and pouch-type cells. In bothcases, metallic lithium foil was used as the counter electrodeand reference electrode, respectively. A solution of 1 M lithiumhexauorophosphate (LiPF6) in propylene carbonate (PC, UBE,Japan) or 0.3 M lithium bis(triuoromethanesulfonyl)imide(LiTFSI) in the ionic liquid N-butyl-N-methylpyrrolidiniumbis(triuoromethanesulfonyl)imide (PYR14TFSI) was used asthe electrolyte. In all experiments a Whatman GF/D glassmicrober lter was used as the separator. The separator wasdrenched with either 100 mL (Swagelok� cells) or 1.8 mL (pouchcell) of the used electrolyte. Cyclic voltammetries (CVs) wereperformed on a VMP multichannel potentiostatic–galvanostaticsystem (Biologic Science Instrument, France), whereas galva-nostatic cycling (GC) was carried out using a Maccor series4300 battery tester. All potentials reported in this work refer tothe Li/Li+ couple.

The samples for the powder XRD, 57Fe Mossbauer spectros-copy and XAS investigations were recovered from cycled cellsstopped at the fully lithiated or delithiated state. The cells weredisassembled inside an argon-lled glove box and washedcarefully with PC in order to remove electrolyte residues. Aerthe washing step the electrodes were dried under vacuum at180 �C overnight. Either the recovered electrodes or the powdermixture, which was scraped from the current collectors, wasused for characterization experiments.

2.2 Powder X-ray diffraction and 57Fe Mossbauerspectroscopy

To analyze structural changes during cycling, the morphologyand phase composition of the materials were examined bypowder X-ray diffraction (XRD). The XRD measurements werecarried out ex situ directly with electrodes or bare powder usinga Bruker D8 Advance (Cu-Ka-radiation, l ¼ 0.154 nm) diffrac-tometer in a 2q range from 10 to 90�. Cell parameters weredetermined by using the STOE WinXPOW program package.

A 57Co/Rh source was available for 57Fe-Mossbauer spectro-scopic investigation. The measurements were conducted ex situwith the pure active material and the electrode powder. Bothsamples were rst ground with SiO2 and placed in a thin-walled PMMA container which was sealed with an epoxy resin.The isomer shi refers to a-iron at room temperature. Themeasurements were performed in the usual transmissiongeometry at room temperature. Fitting of the spectra was per-formed using the NORMOS-90 program system.15

11230 | J. Mater. Chem. A, 2013, 1, 11229–11236

2.3 High-resolution transmission electron microscopy

High-resolution transmission electron microscopy (HRTEM)with electrode powders was performed on a Zeiss LIBRA 200FE(Germany) instrument with an acceleration voltage of 200 kV. Ahigh angle angular dark eld (HAADF) detector was used forenergy dispersive X-ray (EDX) mapping. For the measurements,powder particles were distributed over a conventional, carboncoated Cu grid.

2.4 X-ray absorption spectroscopy

The recovered electrode powder was ground using a mortar andpestle. The nely ground powder was applied to adhesive-coated Kapton tape using a brush. The powder-coated Kaptontape was cut into several pieces which were stacked together toform a thin, uniform transmission X-ray absorption spectros-copy (XAS) sample. XAS measurements were carried out in thetransmission mode at the Fe K-edge of various electrodesamples. The samples were measured at beamline A1 of theHamburger Synchrotron Radiation Laboratory (HASYLAB),Hamburg, Germany. The beamline is equipped with a Si(111)double crystal monochromator. Higher harmonics were rejec-ted by detuning the monochromator such that the intensity ofthe beam on the sample was 65% of the maximum possibleintensity. The intensities of the incident and transmitted beamswere recorded by gas-lled ionization chambers. Data pre-pro-cessing operations were carried out using the soware Athenaof the package IFEFFIT.16 The model function was generated byab initio calculations using the code FEFF8.2.17 The modelfunction was least-square tted to the data using the sowareArtemis of the package IFEFFIT which uses the algorithmFEFFIT.16 The tting parameters involved a single amplitudereduction factor S0

2 and an overall energy parameter DE0. Inaddition, each single scattering path was assigned a fractionalchange in the bond length a and a mean squared relativedisplacement parameter s2 depending on the type of backscatterer and its distance from the central absorbing atom. Theparameters for multiple scattering paths were deduced basedon those of their corresponding single scattering paths.

3 Results and discussion

Fig. 1 shows the voltammetric proles of a Fe2O3-based elec-trode as obtained during tests carried out in the voltage rangefrom 3.0 to 0.6 V vs. Li/Li+.11 As shown, during the rst lithiationa reduction of Fe2O3 (from Fe3+ to Fe0) occurred at a potential of0.8 V vs. Li/Li+ as indicated by the rst large cathodic peakvisible in the CV.3 During the rst delithiation process, ananodic peak at 1.60 V vs. Li/Li+ and a “shoulder” at 1.76 V vs.Li/Li+ were recorded. These two peaks can be assigned tothe oxidation from Fe0 to Fe2+ (1.60 V) and from Fe2+ to Fe3+

(1.76 V).18,19 Starting from the second cycle, both cathodic andanodic peaks shied towards a positive potential, and theintensity of both peaks changed. These changes in the CVs aretypical for conversion electrodes and it is commonly acceptedthat they are caused by the conversion reaction, which takesplace during the lithiation–delithiation process.

This journal is ª The Royal Society of Chemistry 2013

Fig. 1 Cyclic voltammetry of a carbon-coated Fe2O3 based electrode cycledbetween 3.0 and 0.6 V vs. Li/Li+. Inset: comparison of the pictures of a pristineelectrode with an electrode after the 1st delithiation.

Fig. 2 Carbon-coated Fe2O3 based electrode cycled between 3.0 and 0.6 V vs.Li/Li+ in the electrolyte 0.3 M LiTFSI in PYR14TFSI: (a) voltage profiles, (b) dischargecapacity and coulombic efficiency.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

22

July

201

3. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 23

/09/

2013

06:

12:1

7.

View Article Online

It is well known that the coulombic efficiency of the rstlithiation–delithiation process of Fe2O3 (as generally observedin conversion materials) is rather low. It is commonly acceptedthat such low efficiency originates from the decomposition ofthe electrolyte during the rst cycle, which in turn leads tothe formation of a gel-like, organic layer (also called SEIlayer).2,13,14,20 Starting from the second cycle, the efficiency of thelithiation–delithiation process normally increases.3,7,8,21–23

Several factors, e.g. size, morphology and purity of the Fe2O3

particles, might inuence such an increase. Moreover, also theelectrolyte composition might affect signicantly the efficiencyof the lithiation–delithiation process.

The electrolyte used for the test described above was 1 MLiPF6 in PC. This electrolyte features interesting properties (e.g.a relatively high ashpoint), but it exhibits poorer SEI formingproperties compared to other electrolytes which are normallyused for iron oxide based cells.24 When this electrolyte was used,the efficiency of the rst lithiation–delithiation process was70% � 2% (corresponding to an irreversible capacity of 28%from the lithiation capacity). Nevertheless, it is interesting tonote that starting from the second cycle the efficiency of thecharge–discharge was already 99%.11 These efficiencies werecomparable with those observed in electrolytes with good lm-forming ability, e.g. 1 M LiPF6 in EC/DMC 1 : 1.6–8,10 Thus, thisresult seems to indicate that the lm-forming ability of theelectrolyte might not have a strong inuence on the efficiency ofthe charge–discharge of Fe2O3-based electrodes.

With the aim to further investigate this point, we decided totest the Fe2O3-based electrodes in combination with the elec-trolyte 0.3 M LiTFSI in PYR14TFSI. The ionic liquid PYR14TFSIhas an overall electrochemical stability of more than 5 V (with acathodic limit close to 0 V vs. Li/Li+) and previous studiesshowed that this IL does not exhibit lm-forming ability.25–28 Forthese reasons, the electrolyte 0.3 M LiTFSI in PYR14TFSI cannotbe used in combination with graphite electrodes since anexfoliation process of this anode, caused by the co-intercalationof the cation of the ionic liquid, takes place. Considering theseproperties, the efficiency of the lithiation–delithiation process

This journal is ª The Royal Society of Chemistry 2013

of Fe2O3-based electrodes in 0.3 M LiTFSI in PYR14TFSI shouldbe therefore rather poor, since no source for the necessary SEIlayer is present in the system. Fig. 2 shows the voltage proles,the capacity and the efficiency of Fe2O3-based electrodes in 0.3M LiTFSI in PYR14TFSI. The measurements were carried out at60 �C in order to reduce the viscosity of the electrolyte andmakethe system more comparable with that working in a conven-tional electrolyte. It is very important to take into account that at60 �C the electrolyte 0.3 M LiTFSI in PYR14TFSI is thermallystable, and no decomposition processes occur.29,30 In this IL-based electrolyte the capacity (ca. 750 mA h g�1) and the effi-ciency of the lithiation–delithiation process (76% � 2% duringthe rst cycle, 99% starting from the second cycle) shown byFe2O3-based electrodes were both comparable with thoseobserved in 1 M LiPF6 in PC.11

These results clearly show that 0.3 M LiTFSI in PYR14TFSIcan be successfully used in combination with conversion elec-trodes. Moreover, they also supply very interesting informationabout the inuence of the electrolytes on the efficiency of thelithiation–delithiation process. As mentioned above, theformation of the SEI layer originated from the electrolytedecomposition is supposed to be one of the main reasons forthe loss of capacity occurring in conversion electrodes. Minordecomposition processes in the electrolyte 0.3 M LiTFSI inPYR14TFSI cannot be completely excluded. Nevertheless, theseprocesses are certainly signicantly smaller than those occur-ring in conventional electrolytes. Therefore, taking into accountthe results obtained with 0.3 M LiTFSI in PYR14TFSI, it is

J. Mater. Chem. A, 2013, 1, 11229–11236 | 11231

Journal of Materials Chemistry A Paper

Publ

ishe

d on

22

July

201

3. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 23

/09/

2013

06:

12:1

7.

View Article Online

reasonable to believe that the electrolyte decomposition (andconsequently the formation of the SEI-layer) cannot always bethe reason for the loss of capacity during the rst cycle.

As mentioned in the introduction, while the rst reductionprocess (rst lithiation) has been deeply investigated13 the rstoxidation process (rst delithiation) is still not fully understood.Some researchers indicated that a full backformation of ironoxide occurs during this process.2 Nevertheless, the phases thatmight be formed during this oxidation have not been identied.Fig. 3 compares the XRD patterns of a pristine, a lithiated and adelithiated Fe2O3-based electrode as recorded aer galvano-static charge–discharge carried out at C/8 (1C corresponds to1 mA per 1 mg active material) in the voltage range from 3.0 to0.6 V vs. Li/Li+. The XRD pattern of the lithiated electrode wasrecorded at 0.6 V vs. Li/Li+ aer 10 lithiation cycles, while that ofthe delithiated electrode was recorded at 3.0 V vs. Li/Li+ aer 10delithiation cycles. To make the comparison easier, the XRDpattern of the Al current collector is also included in the gure.It is important to mention that identical XRD patterns wereobtained during cycling, indicating a good reversibility of thecharge–discharge process. As shown, aer the rst lithiation(Fig. 3c) the XRD pattern of the crystalline hematite changescompletely and the presence of cubic bcc iron (e.g. broad peak at2q value of 45�) became evident. The presence of iron impliesthe formation of lithium oxide, which seems to be completelyamorphous and cannot be detected by XRD measurements.This pattern conrms (once more) the mechanism proposed forthe lithiation of Fe2O3. Nevertheless, it is interesting to note thatfrom the XRD pattern of the fully delithiated electrode (Fig. 3d)ve peaks centered at 2q values of 38.1, 43.1, 62.7, 75.0 and79.0�, respectively, could be identied. These peaks match interms of both 2q position and relative intensities, with thosetypical for the cubic lithium-iron oxide LiFeO2.31 The different

Fig. 3 (a) Voltage profiles of carbon-coated Fe2O3 based electrode cycledbetween 3.0 and 0.6 V vs. Li/Li+ in the electrolyte 1 M LiPF6 in PC; (b) comparisonof powder XRD patterns of the pristine (-), fully lithiated (+) and fully delithiated(A) electrodes after 10 lithiation and 10 delithiation cycles; (c) zoom in about thelithiated electrode; (d) zoom in about the delithiated electrode. For comparisonthe pattern of the bare aluminium current collector is also included. Calculatedpatterns for bcc Fe and a-LiFeO2 are given for better comparison.

11232 | J. Mater. Chem. A, 2013, 1, 11229–11236

modications of LiFeO2 are well known in the literature andhave repeatedly been studied with respect to their electro-chemical properties.32 Taking into account these measure-ments, it is therefore evident that the delithiation process is notcausing the back-formation of Fe2O3. Instead, such oxidationprocess seems to lead to the formation of a lithium iron oxide,and consequently the compound obtained aer the conversionprocess does not seem to be fully delithiated. It is interesting tonote that aer the rst conversion process the color of theelectrode changes and the red colored hematite turns into ablack compound, both at the lithiated as well as the delithiatedstate (see inset of Fig. 1).

In order to collect further information about the speciesformed aer the rst delithiation process, a more detailedpowder XRD investigation as well as 57Fe Mossbauer spectros-copy, XAS and HRTEM measurements were carried out. In thiscase the XRD pattern was acquired using powder recoveredfrom Fe2O3-based electrodes, which were subjected to lithiationand delithiation processes (for details see Experimentalsection). As shown in Fig. 4 the results of the detailed XRDinvestigation conrm that LiFeO2 was effectively formed aerthe rst delithiation process. LiFeO2 crystallizes with a rocksalt-like cubic structure, with space group Fm�3m. Indexing of thepowder pattern and a least-squares renement led to a latticeparameter of 417.6(7) pm. The lithium and iron atoms arerandomly distributed on the 4a site (0 0 0) and the oxygen atomsll Wyckoff position 4b (1/2 0 0). The rened cell parameter is ingood agreement with literature data.33,34 Additional peaks(superstructure reections) that would point to a possible Li/Feordering like in b-LiFeO2 (R�3m, NaFeO2 type)35 have not beenobserved in the powder XRD pattern. From the Scherrerequation the average particle size of a-LiFeO2 results in theorder of 7–11 nm.

57Fe Mossbauer spectra (298 K data) were recorded for thepure hematite as well as for the recovered powder. The latter

Fig. 4 Powder XRD pattern of a fully delithiated carbon-coated Fe2O3 basedelectrode after the first cycle. The pattern matches with the cubic phase a-LiFeO2.

This journal is ª The Royal Society of Chemistry 2013

Fig. 5 Experimental and simulated 57Fe Mossbauer spectra of a-Fe2O3 andcycled electrode (a-LiFeO2) at 298 K.

Fig. 6 EXAFS fit (a-LiFeO2) to the data of the delithiated carbon-coated Fe2O3

sample. The dotted lines indicate the fitting range.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

22

July

201

3. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 23

/09/

2013

06:

12:1

7.

View Article Online

also contained conductive carbon and binder. The 57FeMossbauer spectra of both materials are presented in Fig. 5together with transmission integral ts. The spectrum of pure a-Fe2O3 shows a huge magnetic hyperne splitting with Bhf ¼51.3(1) T. Based on the isomer shi of d ¼ 0.25(1) mm s�1 anoxidation state of Fe3+ can clearly be attributed.36 The least-squares t of the experimental data further revealed theparameters for the electric quadrupole splitting DEQ ¼ �0.20(1)mm s�1 and the line width of G¼ 0.25(2) mm s�1. The electrodesample aer the cycling procedure no longer shows magnetichyperne eld splitting in its 57Fe Mossbauer spectrum. Itshows a single Fe3+ signal at an isomer shi of d ¼ 0.29(1) mms�1, which is subjected to electric quadrupole splitting of DEQ ¼0.59(2) mm s�1. The experimental line width parameter is G ¼0.88(3) mm s�1. The spectrum matches with the published datafor a-LiFeO2.35,37

XAS measurements were also carried out on a delithiatedstate of Fe2O3. As shown in Fig. 6, the EXAFS data of the deli-thiated sample were explained by the cubic structure of a-LiFeO2. A good agreement between the data and theory can beseen up to around 5 A. The recorded XAS spectrum was in verygood agreement with that of the cubic structure of a-LiFeO2

already reported in the literature,38 conrming the formation ofnanodomains of this compound aer the rst delithiationprocess. The average distances of the 1st shell of O and the 2ndshell of Fe atoms from central Fe atoms were rened to be2.029(6) A and 3.033(6) A, respectively. The 3rd and 4th shellshave mixed O and Fe atoms. The average distances of O and Featoms in the 3rd shell were rened to be 3.608(24) A and4.462(96) A, respectively. Similarly, O and Fe atoms in the 4th

This journal is ª The Royal Society of Chemistry 2013

shell were found to be located at average distances of 4.583(46)A and 5.158(28) A, respectively. From these measurements, thepresence of Fe nanoparticles could also be seen. Due to the highcorrelation among tting parameters, both the fraction ofelemental Fe and the number of nearest neighbours inelemental Fe cannot be rened at the same time. Thus, thenumber of nearest neighbours in elemental Fe was constrainedto 8 and the fraction of Fe was rened in the t. It must be notedthat due to the superimposition of the elemental iron EXAFSsignal with that of the delithiated sample it is difficult toprecisely quantify the amount of elemental iron present inthe sample.38–40

In order to collect further information about the presence ofelemental iron in the delithiated state HRTEM measurementscoupled with EDXmapping were performed. As shown in Fig. 7aand b, all particles are randomly distributed and form large,non-transparent clusters. It is not possible to distinguishbetween specic particles in shape and size. Furthermore, theparticles appear amorphous in the HRTEM mode. The Fouriertransform of the images gives no indication of crystalline pha-ses and no lattice planes are visible. These observations are ingood agreement with the results obtained by XRD (see Fig. 3 and4), conrming the amorphous character of all particles. InFig. 7c and d, the EDX mapping of two different spots is pre-sented. As shown, oxygen and iron are randomly distributedover the sample spot (Fig. 7e–h). Nevertheless, it is important tonote that in some regions the iron concentration is signicantlyhigher compared to the oxygen concentration, suggesting thepresence of pure amorphous iron clusters on the delithiatedsample (Fig. 7d, f and h).

Keeping the results reported above inmind, the “traditional“charge–discharge mechanism proposed for iron oxide does notappear to be fully supported by experimental evidence. This isparticularly true when the delithiation process and loss ofcapacity during the rst cycle are considered.

Taking this point into account, an “alternative” mechanismseems therefore necessary to explain the behavior of Fe2O3-based electrodes. Fig. 8 shows the scheme of an alternative

J. Mater. Chem. A, 2013, 1, 11229–11236 | 11233

Fig. 7 HRTEM images (a and b) of a fully delithiated carbon-coated Fe2O3 basedelectrode after the first cycle and EDX mapping of different spots (c and d)showing the concentration of the elements Fe (e and f) and O (g and h).

Fig. 8 Comparison between the “traditional” and “new” mechanism of lithiation–

11234 | J. Mater. Chem. A, 2013, 1, 11229–11236

Journal of Materials Chemistry A Paper

Publ

ishe

d on

22

July

201

3. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 23

/09/

2013

06:

12:1

7.

View Article Online

charge–discharge mechanism, which is based on the ndingsreported above. As shown, during the rst lithiation process theformation of iron and Li2O takes place, in agreement with the“traditional” mechanism. Such reduction process involves theconsumption of 6 Li atoms (Fe3+ to Fe0 reaction) and corre-sponds, for hematite, to a theoretical capacity of 1007 mA h g�1.

As indicated above (Fig. 1 and 2), during the rst delithiationprocess about 30% of the capacity is lost. In the traditional viewthis oxidation process leads to the formation of amorphous ironoxide, and the loss of capacity is attributed to the formation ofan SEI layer. However, we could show that this loss of capacitytakes place also when no or negligible SEI is formed. Moreover,we also showed that during the delithiation process thecompound a-LiFeO2 is formed. Therefore, it is reasonable tosuppose that the rst delithiation leads to the formation of thiscompound (see reaction Fig. 8) and not of any binary iron oxide.It is important to take into account that the formation of suchan oxide implies a reduction of lithium available for the charge–discharge process, since one lithium equivalent is necessary forthe formation of one equivalent of a-LiFeO2. For this reason, theformation of this oxide would be the origin for the loss ofcapacity during the rst cycle. Considering the proposedmechanism, the coulombic efficiency in the rst cycle should be75%, which is in very good agreement with the results reportedabove for both electrolytes. In addition, since no material candisappear from the system, it is necessary to consider the factthat at the end of the rst delithiation also iron must benecessarily present in the electrode. Consequently, it isreasonable to suppose that during the delithiation process onlythree-quarters of Fe undergo an electron reaction (oxidation) toLiFeO2. As shown above, starting from the second cycle theefficiency of the charge–discharge process is 99%. Taking thisvalue into account, the lithiation–delithiation of LiFeO2 and Feappears therefore as a reversible process. Such iron nano-particles are hardly detectable by 57Fe Mossbauer spectroscopyparallel with a-LiFeO2, since the isomer shis are almost

delithiation of carbon-coated Fe2O3 based electrodes.

This journal is ª The Royal Society of Chemistry 2013

Paper Journal of Materials Chemistry A

Publ

ishe

d on

22

July

201

3. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 23

/09/

2013

06:

12:1

7.

View Article Online

similar.41,42 The experimental spectrum is thus an overlap ofboth phases. However, HRTEM measurements coupled withEDX mapping could give indications about the presence ofamorphous iron particles, which seem to be embedded inamorphous clusters of LiFeO2 in the delithiated state. Assumingthat 4.5 lithium equivalents are involved in the charge–discharge process of 1.5 equivalents of LiFeO2, a theoreticalcapacity of 848 mA h g�1 can be calculated for this compound.Nevertheless, since iron (which is not contributing to thecapacity) is also present in the delithiated state, it is clear thatsuch a capacity cannot be delivered by the electrodes. Takingthis point into account, and assuming that under the investi-gated conditions the electrode might deliver about 90–95%of the capacity of the electro-active species, a value of ca. 700–750 mA h g�1 can be estimated as the practical capacity whichcan be delivered by the electrode during the lithiation of LiFeO2

and Fe to Li2O and metallic Fe. In the investigated voltage range(from 3.0 to 0.6 V vs. Li/Li+) all electrode components arestable,11 and the contribution to the capacity of the carbona-ceous electrode components is only minor.11 It can beconcluded that the values of capacity delivered by the Fe2O3-based electrodes, both in conventional and ionic liquid basedelectrolytes seem to be in very good agreement with the capacityestimated above.

4 Conclusions

The delithiation process of a-Fe2O3 nanoparticles is a complexprocess, which is still not completely fully understood. In thisstudy we showed that the poor efficiency observed during therst lithiation–delithiation process of carbon coated a-Fe2O3

nanoparticles is not necessarily due to the formation of a SEIlayer, as usually accepted. Instead, this poor efficiency seems tobe due to the formation of cubic a-LiFeO2 during the rst lith-iation–delithiation process. The formation of this compound,which has been conrmed by powder X-ray diffraction 57FeMossbauer spectroscopy and X-ray absorption spectroscopy,reduced the number of lithium-ions available for the followingcharge–discharge process and consequently reduces thecapacity that can be delivered by the electrodes. Once thiscompound is formed, the reversibility of the lithiation–deli-thiation process becomes extremely high (99%), and the Fe2O3-based electrodes can deliver a stable capacity. In the future, itwould be interesting to investigate whether a similar mecha-nism occurs in other types of conversion materials also.

Acknowledgements

The authors wish to thank Westfalische Wilhelms-UniversitatMunster and theMinisterium fur Innovation, Wissenscha undForschung des Landes Nordrhein-Westfalen (MIWFT) for thenancial support.

Notes and references

1 P. Poizot, S. Laruelle, S. Grugeon, L. Dupont andJ. M. Tarascon, J. Power Sources, 2001, 97–98, 235–239.

This journal is ª The Royal Society of Chemistry 2013

2 J. Hassoun, F. Croce, I. Hong and B. Scrosati, Electrochem.Commun., 2011, 13, 228–231.

3 S. Morzilli and B. Scrosati, Electrochim. Acta, 1985, 30, 1271–1276.

4 M. M. Thackeray, W. I. F. David and J. B. Goodenough,J. Solid State Chem., 1984, 55, 280–286.

5 K. M. Abraham, D. M. Pasquariello and E. B. Willstaedt,J. Electrochem. Soc., 1990, 137, 743–749.

6 E. Garcıa-Tamayo, M. Valvo, U. Lafont, C. Locati, D. Munaoand E. M. Kelder, J. Power Sources, 2011, 196, 6425–6432.

7 H. Liu, D. Wexler and G. Wang, J. Alloys Compd., 2009, 487,L24–L27.

8 Y. NuLi, P. Zhang, Z. Guo, P. Munroe and H. Liu, Electrochim.Acta, 2008, 53, 4213–4218.

9 H. S. Ryu, J. S. Kim, Z. Guo, H. Liu, K. W. Kim, J. H. Ahn andH. J. Ahn, Phys. Scr., 2010, 014066.

10 P. C. Wang, H. P. Ding, T. Bark and C. H. Chen, Electrochim.Acta, 2007, 52, 6650–6655.

11 A. Brandt and A. Balducci, J. Power Sources, 2013, 230, 44–49.12 J. Maier, Nat. Mater., 2005, 4, 805–815.13 D. Larcher, D. Bonnin, R. Cortes, I. Rivals, L. Personnaz and

J. M. Tarascon, J. Electrochem. Soc., 2003, 150, A1643–A1650.14 D. Larcher, C. Masquelier, D. Bonnin, Y. Chabre, V. Masson,

J. B. Leriche and J. M. Tarascon, J. Electrochem. Soc., 2003,150, A133–A139.

15 R. A. Brand, Normos Mossbauer tting program, UniversitatDuisburg, Germany, 2002.

16 B. Ravel and M. Newville, J. Synchrotron Radiat., 2005, 12,537–541.

17 A. L. Ankudinov, B. Ravel, J. J. Rehr and S. D. Conradson,Phys. Rev. B: Condens. Matter Mater. Phys., 1998, 58, 7565–7576.

18 J. Morales, L. Sanchez, F. Martın, F. Berry and X. Ren,J. Electrochem. Soc., 2005, 152, A1748–A1754.

19 M. M. Rahman, J.-Z. Wang, M. F. Hassan, Z. Chen andH.-K. Liu, J. Alloys Compd., 2011, 509, 5408–5413.

20 J.-M. Tarascon, S. Grugeon, M. Morcrette, S. Laruelle,P. Rozier and P. Poizot, C. R. Chim., 2005, 8, 9–15.

21 M. F. Hassan, M. M. Rahman, Z. P. Guo, Z. X. Chen andH. K. Liu, Electrochim. Acta, 2010, 55, 5006–5013.

22 H. Morimoto, S.-i. Tobishima and Y. Iizuka, J. Power Sources,2005, 146, 315–318.

23 M. V. Reddy, T. Yu, C. H. Sow, Z. X. Shen, C. T. Lim,G. V. Subba Rao and B. V. R. Chowdari, Adv. Funct. Mater.,2007, 17, 2792–2799.

24 H. Zheng, G. Liu and V. Battaglia, J. Phys. Chem. C, 2010, 114,6182–6189.

25 D. R. MacFarlane, P. Meakin, J. Sun, N. Amini andM. Forsyth, J. Phys. Chem. B, 1999, 103, 4164–4170.

26 A. Lewandowski and A. Swiderska-Mocek, J. Power Sources,2009, 194, 601–609.

27 G. B. Appetecchi, M. Montanino, A. Balducci, S. F. Lux,M. Winterb and S. Passerini, J. Power Sources, 2009, 192,599–605.

28 S. F. Lux, M. Schmuck, G. B. Appetecchi, S. Passerini,M. Winter and A. Balducci, J. Power Sources, 2009, 192,606–611.

J. Mater. Chem. A, 2013, 1, 11229–11236 | 11235

Journal of Materials Chemistry A Paper

Publ

ishe

d on

22

July

201

3. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 23

/09/

2013

06:

12:1

7.

View Article Online

29 G. B. Appetecchi, M. Montanino, D. Zane, M. Carewska,F. Alessandrini and S. Passerini, Electrochim. Acta, 2009,54, 1325–1332.

30 R. S. Kuhnel, N. Bockenfeld, S. Passerini, M. Winter andA. Balducci, Electrochim. Acta, 2011, 56, 4092–4099.

31 M. Vucinic-Vasic, B. Antic, J. Blanusa, S. Rakic,A. Kremenovic, A. S. Nikolic and A. Kapor, Appl. Phys. A,2006, 82, 49–54.

32 J. Li, J. Li, J. Luo, L. Wang and X. He, Int. J. Electrochem. Sci.,2011, 6, 1550–1561.

33 J. C. Anderson and M. Schieber, J. Phys. Chem. Solids, 1964,25, 961–968.

34 M. R. Collongues and G. Chaundron, C. R. Hebd. SeancesAcad. Sci., 1950, 231, 143–145.

35 T. Shirane, R. Kanno, Y. Kawamoto, Y. Takeda, M. Takano,T. Kamiyama and F. Izumi, Solid State Ionics, 1995, 79,227–233.

11236 | J. Mater. Chem. A, 2013, 1, 11229–11236

36 E. Kuzmann, S. Nagy and A. Vertes, Pure Appl. Chem., 2003,75, 801–858.

37 O. Knop, C. Ayasse, J. S. Carlow, W. W. Barker,F. W. D. Woodhams, R. E. Meads and W. G. Parker, J. SolidState Chem., 1978, 25, 329–346.

38 G. Waychunas, G. Brown Jr and M. Apted, Phys. Chem.Miner., 1986, 13, 31–47.

39 L. Kuen-Song, C. Ni-Bin and C. Tien-Deng, Sci. Technol. Adv.Mater., 2008, 9, 025015.

40 G. Waychunas, M. Apted and G. Brown Jr, Phys. Chem.Miner., 1983, 10, 1–9.

41 K. Klacanova, P. Fondran, P. Simon, P. Rapta, R. Boca,V. Jorık, M. Miglierini, E. Kolek and L. Caplovic, J. Chem.,2013, 2013, 10.

42 R. Nazir, M. Mazhar, T. Wakeel, M. Akhtar, M. Siddique,M. Nadeem, N. Khan and M. Shah, J. Therm. Anal.Calorim., 2012, 110, 707–713.

This journal is ª The Royal Society of Chemistry 2013