Journal of Power Sources 189 (2009) 179184

Contents lists available at ScienceDirect

Journal of Power Sources

journa l homepage: www.e lsev ier .com/

Nanosi 0.05electro

U. Lafont gasc

E.M. Kelda NanoStructurb Fundamental dsc Faculty of Phyd Functional M ted Ki

a r t i c l

Article history:Received 25 JuReceived in re25 SeptemberAccepted 26 September 2008Available online 19 October 2008

Keywords:High voltage spinelIn situ XANESLi-ion batteryNanomaterialsLiNi0.5Mn1.5O4

nthehis m

terisa(TEM,neutrondiffraction, galvanostaticmeasurements, and impedance spectroscopy). Besides, in situXAShas been performed to monitor the evolution of Ni and Mn oxidation state during Li intercalation. Thematerial presents an ordered cubic spinel structure, good capacity retention upon cycling (131mAhg1

at C/10 and 117mAhg1 at 1C) and good electronic conductivity (106 S cm1 at RT). The simultaneouspresence of Mn3+/Mn4+ in the structure has been investigated and explained by inclusion of disorderednanodomains in the structure.

1. Introduc

Nowadaresearch, cothey are seesition to noMost of theelectronics,negative eleelectrode. Fbased matetrolytes indemands fotional Meetbatteries forespect, alth(LTO) matermercial higexcellent re

CorresponE-mail add

0378-7753/$ doi:10.1016/j.j 2008 Elsevier B.V. All rights reserved.

tion

ys, many political decisions, supported by scienticncern the implementation of Li-ion batteries, becausen as key electrical storage media for a sustainable tran-n-polluting automotive vehicles (HEV, EV and PHEV).commercial Li-ion batteries are still used for portableand use typically a carbonaceous based material asctrode and a 4-V transition metal oxide as positiveor high power application, the use of carbonaceousrial and their derivatives with the PC-containing elec-these systems are not safe enough to full safetyr commercial applications. Indeed, during the Interna-ing on Li-ion Battery 2008, the safety issue of Li-ionr automotive application was a hot topic. In thisough it is known for several years, the Li4Ti5O12 spinelial present most of the requirements needed for com-h power applications, i.e. safe, cheap, non-toxic, hasversibility and structural stability (zero strain during

ding author. Tel.: +31 152785536; fax: +31 152784945.ress: u.lafont@tudelft.nl (U. Lafont).

Li insertion and removal). The main disadvantage of this materialis the working voltage of 1.5V vs. Li/Li+. However, the 1.5V plateaucan avoid the carbonate based electrolyte solvent to decomposeinto gaseous products. Using LTO material as negative electrodewill give a cell of 2.5V using a state-of-the-art 4V cathodematerial.In order to obtain a sufciently higher operating voltage, so calledhigh voltage cathodematerials are needed. Since a decade, the sub-stitution of Mn by other transition metal ions, such as by Me=Ni,Co, Fe, Cu, Cr in LiMn2O4 has been extensively studied [112] lead-ing to a high voltage spinel LiMe0.5Mn1.5O4 and LiMeMnO4. In thisrespect, LiNi0.5Mn1.5O4 and LiCrMnO4 material are the only onespresenting a sufciently well dene high voltage plateau [13].

On theonehand it is clear that highvoltagematerials areneededand the LiNi0.5Mn1.5O4 type material being the most promisingsystem has been extensively studied. On the other hand, besideprogress made on electrode engineering, nanosized material canbe of great help due to an enhanced charge transfer surface areaper weight and improved rate capability [1416] due to a reduceddiffusion/migration-length compared to state-of-the-art cathodematerials sizes.

In this study we will focus on the structural and electrochem-ical characterisation of Mg and Ni substituted nanosized spinelmaterial LiMn2O4 [14,15,1719], LiMg0.05Ni0.45Mn1.5O4, exhibit-

see front matter 2008 Elsevier B.V. All rights reserved.powsour.2008.09.121zed high voltage cathode material LiMgchemical and in situ investigationa,, C. Locati a, W.J.H. Borgholsb, A. asinskac, J. Dyera

ed Materials, DCT/TUDelft, Julianalaan 136, 2628BL Delft, The NetherlandsAspects of Materials and Energy, R3/TUDelft, Mekelweg 15, 2629JB Delft, The Netherlansics, Warsaw University of Technology, Koszykowa 75, 00-662 Warszawa, Polandaterials Group, School of Physical Sciences, University of Kent, Canterbury CT2 7NR, Uni

e i n f o

ly 2008vised form2008

a b s t r a c t

In this study a modied solid state sytype material LiMg0.05Ni0.45Mn1.5O4. TStructural and electrochemical characlocate / jpowsour

Ni0.45Mn1.5O4: Structural,

, A.V. Chadwickd,

ngdom

sis (auto-ignition method) is used to form nanosized spinelaterial presents a high voltage plateau at 4.75V vs. Li/Li+.

tions have been performed using a wide range of techniques

180 U. Lafont et al. / Journal of Power Sources 189 (2009) 179184

ing 148.10mAhg1 of theoretical capacity. The Mg atoms arenot directly involved in redox reactions during lithium intercala-tion/deintercalation. As it has been demonstrated in previous work[17,18], Mgoverall inseandwill notMg substitusystem [20]

2. Experim

2.1. Synthes

The synignition mreactions thdizer to matemperatursized by miNi(NO3)26Htain amoumolar com0.05:0.0025are thermalThe resultinThe as-synt

2.2. Structu

The comtively CoupThesurfacemethod usi6B. Transma Philips CMtrons sourcewere perfotime-of-igsource (RutXANES meaout on statiwere perforchambers.metal foil perence for tsynchrotronthe experimtaking as thNi foils (833coffee bagbled. The cduring the m

2.3. Electro

Theelecin galvanostmaterial, p99.9%,fa) andgraphite (0.1mm;Fluka) (80:10:3:7wt.%)were1-methyl-2-pyrrolydon (NMP, Merck-Schuchardt). Theste is ball-milled for 15min and then deposited on anstrip with a 200-m thick doctor-blade and immedi-at 80 C for a couple of hours. The aluminium strip isated until 70m thickness. The electrochemical tests

Fig. 1. Tmethod.

were delectroLiPF6 1mm Licoin-ceH2O

U. Lafont et al. / Journal of Power Sources 189 (2009) 179184 181

Table 1Rened structural parameters for LiMg0.05Ni0.45Mn1.5O4.

Phase Atom Site x/a y/b z/c Fraction [0,1]

P4332 cubic space group 212, a=8.1748,wRp=0.049, Rp=0.048

Li 8c 0.001 0 0 0.81Li 4a 1/8 1/8 1/8 0.38Mg 4b 5/8 5/8 5/8 0.10Ni 4b 5/8 5/8 5/8 0.65Mn 4b 5/8 5/8 5/8 0.23Ni 12d 1/8 0.4173 0.8327 0.08Mn 12d 1/8 0.4007 0.8493 0.92O 8c 0.3884 0.3884 0.3884 1.00O 24e 0.1069 0.1214 0.3902 1.00

3. Results and discussion

3.1. Structure characterisation

The synthesized material by the auto-ignition method presentsa nal composition of Li0.98Mg0.049Ni0.44Mn1.55O4 measured withICP. The surface area deduced from the nitrogen adsorptionisotherm using the BET theory is 4.5m2 g1. The material doesnot present any structural porosity. Indeed, the TEM micrograph(Fig. 1) exhibits soft agglomerates of 2080nm particles. The par-ticles reect clearly the cubic structure of the spinel phase thematerial exhibits.

Fig. 2 diswith the reHere the recurve is pldue to anresolved by(space grouextra phasea full concontaminatNeverthelesther from Rto be 51 (eters andthe data wthe LiMg0.0ture.

Fig. 3. Temper

3.2. Electrochemical characterisation

Electrochemical impedance spectroscopy results of the pressedpellets are shown in Fig. 3. Impedance measurement are recordedas a function of temperature T, ranging from 60 to 200 C. In orderto calculate the Li-ion conductivity (T), the recorded spectra notshown here are tted to a state-of-the-art equivalent circuit. Typ-ically, the material presents a conductivity of 106 S cm1 at 25 C.The activation energy Ea deduced from the data is calculated usingan Arrhenius type dependency, i.e. (T) = A exp[Ea/kT], with A apre-exponeEa values ar

perat0 Cgalv

4. Thrrent5C),eastspecbehaThisandrgeisapp5C,fromon oe hig

rge cFig

anddplays the observed neutron diffraction pattern togetherned spectrum calculated with the Rietveld method.sidual error is about 5% and the corresponding errorotted as a grey line in Fig. 2. This error is mainlyextra phase in the material, which reections areassuming a contamination of about 2% of LiMn2O4p Fd3m) in the sample. The main reection of thisis present at 2.33. It is, however, stressed that,

rmation cannot be given as the 2% calculated asion is within the accuracy limits of the renements.s, the peak near 2.33 is clearly visible by eye. Fur-ietveld renement, the particle size was calculated4) nm. The renement data as the unit cell param-

the atomic position are gathered in Table 1. Sinceere best rened according to the P4332 space group,5Ni0.45Mn1.5O4 adopts an ordered cubic spinel struc-

in tem10014

Thein Fig.sive cuC/10 tofor at l

Irrecharge4.77V.to Ni3+

the chaever, d2C andLi-ionsoxidatition) thdischa

Fromchargeature dependency of the conductivity of LiMg0.05Ni0.45Mn1.5O4. Fig. 4. Chargential factor and k the Boltzmann constant. The obtainede 0.37 eV in the temperature range 40 to 30 C, 0.40 eVure range 30100 C and 0.41 eV in temperature range.anostatic chargedischarge measurements are plottede experiments were performed using different succes-rates, ranging from 15 to 750mAg1 (giving a C-rate ofand nally back to 30mAg1 (C/5). Data were recordedve chargedischarge cycles.tive of the current used, the curves present the sameviour showing a high voltage plateau between 4.71 andplateau is attributed to the successive oxidation of Ni2+

Ni4+. For currents from 15 to 150mAg1 (C/10 to C) allcurves present a plateau near 4.1V. This plateau, how-ears at higher charge current (300 and 750mAg1, i.e.respectively). The de-intercalation of Li, i.e. taking outthe spinel structure, at this voltage is attributed to the

f Mn3+ to Mn4+. For the discharge prole (Li intercala-h voltage plateau is shifted to lower voltages when the

urrent is increased.. 5 we see that increasing the current rate during theischarge lead to signicant loss of capacity. It is remark-discharge prole in galvanostatic mode at various C rate vs. Li/Li+.

182 U. Lafont et al. / Journal of Power Sources 189 (2009) 179184

Fig. 5. Dischar

able that iretention urate. The caered in Tabl

After thesystem wasThe obtainecomplete recell does no

Fromthethe capacityspecies at 4tribution cospecies neaage of the toNi species (and discharcentage of tcurrent fromprocess, thicurrent. Thienergy barrwhen the mso as to mining the chaand overcomon Mn3+/Mmaterial ma

3.3. In situ

Fig. 7 shoedge (6538charge andfor the Ni K

Table 2Capacity obtai

Current rate (m

153075

150300750

ercentage of the total capacity attributed to the high voltage (HV) plateaution of the current rate.

charge) whereas the curves for the Mn K-edge are recordedthe rst cycle for the charge and during the second cycle forcharge. The Ni K-edge is shifted to higher energy during theindicating a change in the oxidation state. Thepristinemate-sents an average oxidation state of +2.6. Some Ni speciesge capacity as a function of cycle number and C rate in mAg1 (top).

rrespective of the charge current used, the capacitypon cycling is really stable within that specic currentpacity measures at the various current rates are gath-e 2.series of charging and discharging at various rates, thecharged and discharged at a C/5 rate, i.e. 30mAg1.d capacity at the end stage is 128mAhg1, showingcovering of the initial capacity. Further cycling of thet show anymore degradation.galvanostaticmeasurements it is clear that theoriginofis not only due to the oxidation/reduction of Ni2+/Ni4+

.75V. The total capacity presents a non-negligible con-ming from the oxidation/reduction of the Mn3+/Mn4+

r 4.1V. In this respect, we plotted in Fig. 6, the percent-tal capacity attributed to the oxidation/reduction of thethe high-voltage plateau at 4.75V) during the chargege at different current rates. During the charge, the per-his part increased from 84 to 98% upon increasing the

15 to 750mAg1, respectively. During the discharges value is decreasing from 81 to 60% with increasing thes phenomenon can be explained in terms of an allegedier. According to Ceder et al. [22], Li diffusion is fasteretals of the host structure present a low oxidation stateimise the lithium-metal repulsion. At high C-rate, dur-rge, Li+ transport will encounter much less repulsione a lower energy barrier on Ni2+/Ni3+ transition than

n4+. Besides, the distribution of the Mn in the overally play a role as well, as will be explained later on.

Fig. 6. Pas a func

and disduringthe dischargerial prestructure analysis

ws the curves of the Ni K-edge (8332eV) and theMnK-eV) XANES spectra obtained at various voltages duringdischarge recorded between 3.5 and 4.9V. The curves-edge have been recorded during the rst cycle (charge

ned as a function of the current rate.

A g1) C-rate Capacity (mAhg1)

C/10 131C/5 128C/2 122C 1172C 1085C 90

Fig. 7. In situat C/5 (a) Ni K-XANES spectra at different voltage during chargedischarge vs. Li/Li+

edge and (b) Mn K-edge.

U. Lafont et al. / Journal of Power Sources 189 (2009) 179184 183

present an initial oxidation state higher than the expected oxida-tion state of +2. At 4.9V (end of charge) all Ni species present a+4 oxidation state. During the discharge, the Ni is reduced and at4.5V (end oaverage oxibehaviour oof Ni2+/Ni4+

The shistate, is leLiMg0.05Ni0around +3.5maximal vacycle, the avchange in oxatoms can b

The LiMignition mepristine maand not allpure orderesurementspresent a cMn3+ ionsupon cyclinthe electrorecent studLiMn1.5+dNidescribe thecubic phasephase exhibdoes not. Thstructure, ware presentsites. Fromformed so fMn3+ cationductivity isdisorderedductivity [2i.e. 107 S cmcan be attrinanosized pthe relationmaterial iselectrochempresence ofanorderedcmentioneddiffraction.tron diffracexplain comand reductistand this dwe propose

Mn3+ spethe surfaLiMg0.05N

Mnions eso as to cr

The rstfor ions andbefore thetive of the

eutrod Fd

learlondnodoes. Insordeoneffrac8.iousf anuishof thty answithiffere

clus

h theLiM

g anal prmAh

nosizcondy up/reduction ofMn andMn is evident and can be explainedrich nanodomainswithin the nanosized particles. Therefore,h power applications, the material is an interesting positivede in Li-ion batteries using Li4Ti5O12 as an anode.

wledgments

authors would like to thank the European Commission forg the European Network of Excellence ALISTORE, and in par-M. Morcrette for helping in the coffee bag cells preparation,klocks for helping inXASdata treatments andM.Wagemakertful discussions.

nces

igala, D. Guyomard, A. Verbaere, Y. Piffard, M. Tournoux, Solid State Ionics(1995) 167170.mine, H. Tukamoto, H. Yasuda, Y. Fujita, J. Electrochem. Soc. 143 (1996)71613.f the high-voltage plateau), all the Ni atoms present andation state of +2.7. It is stressed that the high voltagef thismaterial is due to a reversible oxidation/reduction.ft, induced by a change in the Mn oxidationss pronounced. Obviously, the pristine material of.45Mn1.5O4 shows an initial Mn oxidation state of7. During the rst charge the Mn atoms quickly adopt alue of +4 at 4.78V. During the discharge in the seconderage oxidation sate is stable at +3.8 until 4V. From thisidation state, it is concluded that around 20% of theMne reversibly oxidized and reduced.g0.05Ni0.45Mn1.5O4 material, synthesized by the auto-thod, exhibits a multi cation valency. Indeed, in theterial not all the Ni atoms show a +2 oxidation statethe Mn atoms have a +4 oxidation state as expected ford LiMg0.05Ni0.45Mn1.5O4. From electrochemical mea-and XANES investigations it is clear that the materialertain amount of Mn3+ ions. The presence of thesein the spinel structure implies a Jan-Teller distortiong, which is usually responsible of the degradation ofchemical properties of the material. However, someies focussing on LiNi0.5Mn1.5O4, LiNi0.4Mn1.6O4 [23],0.5dO4 [16] and LiMn2xNixO4 [24] spinel materialsspinel in a disordered (Fd3m) and an ordered (P4332). During electrochemical tests vs. Li/Li+ the disorderedits a plateau around 44.1Vwhereas the ordered phaseerefore, the disordered phase present Mn3+ ions in thehereas in the ordered phase only Mn4+ and Ni2+ ions

, and which they are ordered in 3:1 ratio on octahedralthe electrochemical tests on this type of material per-ar [1416,2325], the positive effect of the presence ofs for high rate application is clear. The electronic con-also inuenced by cation ordering in the structure: thephase can exhibit a 2.5 orders of magnitude higher con-4] compared to the conductivity of the ordered phase,

1. Ourmaterial had a conductivity of 106 S cm1 thatbuted to a benecial effect of the Mg ions [17] and thearticles as it has been reported earlier [14,15]. Howeverbetween the presence ofMn3+ and the pure disorderednot clear yet. Obviously, our material presents all theical characteristics of a disordered cubic phase, i.e. theaMn3+/Mn4+ oxidation reduction at 44.1V, but revealsubic structure (P4332) fromneutrondiffractiondata.Asearlier, around 2% of LiMn2O4 was detected by neutronThis low amount of extra phase observed with neu-tion, partly contribute to the Mn3+ presence but cannotpletely the signicant amount of Mn3+/Mn4+ oxidationon seen during electrochemical experiments. To under-isagreement and nd out where the Mn3+ arises from,two models:

cies are located at the outer part of the particles ce and are responsible for the relaxation of thei0.45Mn1.5O4 structure as formed during the synthesis.xist in clusterswithin LiMg0.05Ni0.45Mn1.5O4 crystalliteseate Mn rich and Mn poor domains in the particles.

model implies a serial insertion/extraction mechanismelectrons in which the surface Mn ions have to react

core Ni ions are being reached. In this case, irrespec-C rate, the Mn3+/Mn4+ transition should always been

Fig. 8. NP4332 an

seen. CThe secrichnaparticlin a diwithintron diin Fig.

Obvtions odistingttingintensibationlocal d

4. Con

Witto formshowinmateriand 90the naLi-ionstabilitdationbyMnfor higelectro

Ackno

ThefundinticularA. Blacfor frui

Refere

[1] C. S81

[2] K. A160ndiffraction of LiMg0.05Ni0.45Mn1.5O4. The reections belonging to the3m space groups are labelled as P and F, respectively.

y, this does not happen in our system at high C rate.model is based on a parallel system in which some Mnmains arepresent in theordered LiMg0.05Ni0.45Mn1.5O4deed, this Mn rich domain will present Mn3+ ions likered cubic structure. The small size of these domainsparticle of ordered spinel cannot be detected by neu-tion but their presence can be suggested as it is shown

ly, the neutron diffraction data show the main reec-ordered cubic phase (P4332) that can clearly be

from the disordered one (Fd3m) [19]. However thee reections at 1.745 and 1.785 show a difference ind peak width, which can be attributed to either pertur-in the ordering or a shorter range ordering due to somence in cation ordering, hence Mn clusters.

ion

synthesis process used in this research we were ableg0.05Ni0.45Mn1.5O4 nano-particles of 50nm in size,ordered cubic spinel phase (P4332). This high voltageesent good capacity retention of 131mAhg1 at C/10g1 at 5C. The Mg doping eventually supported by

e of the particles allow this material to exhibit a gooductivity at room temperature (106 S cm1) and goodon cycling even at high C-rate. The presence of the oxi-

3+ 4+

184 U. Lafont et al. / Journal of Power Sources 189 (2009) 179184

[3] Y. Gao, K.Myrtle,M. Zhang, J.N. Reimers, J.R. Dahn, Phys. Rev. B 54 (1996) 16670.[4] K. Amine,H. Tukamoto,H. Yasuda, Y. Fujita, J. Power Sources 68 (1997) 604608.[5] Q.M. Zhong, A. Bonakdarpour, M.J. Zhang, Y. Gao, J.R. Dahn, J. Electrochem. Soc.

144 (1997) 205213.[6] Y. Ein-Eli, J.W.F. Howard, S.H. Lu, S. Mukerjee, J. McBreen, J.T. Vaughey, M.M.

Thackeray, J. Electrochem. Soc. 145 (1998) 12381244.[7] Y. Ein-Eli, S.H. Lu, M.A. Rzeznik, S. Mukerjee, X.Q. Yang, J. McBreen, J. Elec-

trochem. Soc. 145 (1998) 33833386.[8] H.Kawai,M.Nagata,M.Tabuchi,H. Tukamoto,A.R.West, Chem.Mater. 10 (1998)

32663268.[9] H. Kawai, M. Nagata, H. Tukamoto, A.R. West, Electrochem. Solid State Lett. 1

(1998) 212214.[10] H. Kawai, M. Nagata, H. Tukamoto, A.R. West, J. Mater. Chem. 8 (1998) 837839.[11] Y. Ein-Eli, J.T. Vaughey, M.M. Thackeray, S. Mukerjee, X.Q. Yang, J. McBreen, J.

Electrochem. Soc. 146 (1999) 908913.[12] H. Kawai, M. Nagata, H. Kageyama, H. Tukamoto, A.R. West, Electrochim. Acta

45 (1999) 315327.[13] H. Kawai, M. Nagata, H. Tukamoto, A.R. West, J. Power Sources 81 (1999) 6772.[14] U. Lafont, C. Locati, E.M. Kelder, Solid State Ionics 177 (2006) 30233029.

[15] C. Locati, U. Lafont, L. Simonin, F. Ooms, E.M. Kelder, J. Power Sources 174 (2007)847851.

[16] M. Kunduraci, G.G. Amatucci, Electrochim. Acta 53 (2008) 41934199.[17] F.G.B. Ooms, E.M. Kelder, J. Schoonman,M.Wagemaker, F.M.Mulder, Solid State

Ionics 152153 (2002) 143153.[18] F.G.B. Ooms,M.Wagemaker, A.A.v.Well, F.M.Mulder, E.M. Kelder, J. Schoonman,

Appl. Phys. A: Mater. Sci. Process. A74 (2002) S1089S1091.[19] M.Wagemaker, F.G.B.Ooms,E.M.Kelder, J. Schoonman,G.J. Kearley, F.M.Mulder,

J. Am. Chem. Soc. 126 (2004) 1352613533.[20] C.H. Chen, J.T. Vaughey, A.N. Jansen, D.W. Dees, A.J. Kahaian, T. Goacher, M.M.

Thackeray, J. Electrochem. Soc. 148 (2001) A102A104.[21] J.-M. Tarascon, A.S. Gozdz, C. Schmutz, F. Shokoohi, P.C. Warren, Solid State

Ionics 8688 (1996) 4954.[22] K. Kang, Y.S. Meng, J. Brger, C.P. Grey, G. Ceder, Science 311 (2006) 977980.[23] S. Patoux, L. Sannier, H. Lignier, Y. Reynier, C. Bourbon, S. Jouanneau, F. Le Cras,

S. Martinet, Electrochim. Acta 53 (2008) 41374145.[24] M.Kunduraci, J.F. Al-Sharab,G.G. Amatucci, Chem.Mater. 18 (2006) 35853592.[25] H.F. Xiang, Q.Y. Jin, R. Wang, C.H. Chen, X.W. Ge, J. Power Sources 179 (2008)

351356.

Nanosized high voltage cathode material LiMg0.05Ni0.45Mn1.5O4: Structural, electrochemical and in situ investigationIntroductionExperimentalSynthesisStructure characterisationElectrochemical characterisation

Results and discussionStructure characterisationElectrochemical characterisationIn situ structure analysis

ConclusionAcknowledgmentsReferences