Magnetic measurements as a viable tool to assess the relative degrees of cation ordering and Mn3+...

Journal ofMaterials Chemistry A

PAPER

Publ

ishe

d on

12

July

201

3. D

ownl

oade

d by

Dal

hous

ie U

nive

rsity

on

23/0

8/20

13 1

5:21

:25.

View Article OnlineView Journal | View Issue

Electrochemical Energy Laboratory & Materi

University of Texas at Austin, Austin, Texas

Cite this: J. Mater. Chem. A, 2013, 1,10745

Received 23rd May 2013Accepted 12th July 2013

DOI: 10.1039/c3ta12021j

www.rsc.org/MaterialsA

This journal is ª The Royal Society of

Magnetic measurements as a viable tool to assess therelative degrees of cation ordering and Mn3+ content indoped LiMn1.5Ni0.5O4 spinel cathodes

Zachary Moorhead-Rosenberg, Katharine R. Chemelewski, John B. Goodenoughand Arumugam Manthiram*

Themagnetic properties of the doped high-voltage spinels LiMn1.5Ni0.42M0.08O4 (M¼ Cr, Fe, Co, Cu, Al, and

Ga) as well as the Mn-rich spinels LiMn1.5+xNi0.5�xO4 (x ¼ 0, 0.05, 0.1) synthesized by a tank-reactor

coprecipitation method have been investigated. It is found that the Curie temperatures correlate well

with the degree of cation ordering determined by measuring the capacity value at �2.7 V, which

corresponds to the insertion of additional lithium into the empty 16c octahedral sites of the spinel

lattice. Additionally, it is found that the saturation magnetization near 0 K provides a reliable means of

calculating the Mn3+ content in each sample based on the predicted magnetic exchange interactions of

the spinel structure. The magnetic results match well with the capacity values obtained in the 4 V region

of the samples when assembled as cathodes in lithium-ion cells. The Mn-rich spinels offer insight into

the relationship between Mn3+ content and degree of cation ordering and how they affect the

magnitude of the capacity in the �2.7 V region. The effects of Mn3+ content and cation ordering were

isolated by examining the relationship among lattice parameter, Mn3+ content, capacity contribution at

�2.7 V, and Curie temperature. Thus, it is posited that the Curie temperature is a reliable qualitative

means of measuring the degree of cation order in a variety of doped or Mn-rich spinel oxides.

Introduction

The high-voltage (�4.7 V) spinel LiMn1.5Ni0.5O4 has become anattractive candidate for next generation lithium-ion batteriesdue to its high power and fast-recharge capability.1–3 Over thepast decade, intensive research has been devoted to under-standing its electrochemical behavior in an effort to elucidateits suitability as a commercial cathode material. One of themost attractive characteristics of LiMn1.5Ni0.5O4 is the highoperating voltage of �4.7 V vs. Li/Li+, ensuring a higher energydensity compared to the 4 V spinel LiMn2O4, which is currentlyemployed as the cathode in production electric vehicles pow-ered by Li-ion batteries.3 However, several challenges persist,whichmust be further examined in order to develop an in-depthunderstanding of the properties of the high-voltage spinelcathodes: role of cation order/disorder on the cyclability andrate capability,4–6 surface segregation of dopant ions,6

morphological considerations,7 problems inherent with full-cellconstruction,8 and stability with the electrolyte and solid-elec-trolyte interphase (SEI) layer formation.8

The degree of cation ordering is particularly interesting,since no consensus exists on why the ordered P4332 phase is

als Science and Engineering Program, The

78712, USA. E-mail: [email protected]

Chemistry 2013

innately inferior to the higher-symmetry Fd�3m (disordered)phase in terms of electrochemical behavior. The P4332 phaseconsists of a unit cell in which the Ni2+ and Mn4+ ions have a1 : 3 ratio and they preferentially order, respectively, in the 4band 12d octahedral sites. There is no such regularity in thedisordered structure wherein the Mn4+ and Ni2+ ions are, torelative degrees, randomly distributed among the 16d octahe-dral sites.9 Doping to obtain compositions such as LiMn1.5-Ni0.5�xMxO4 (M ¼ Cr, Mn, Fe, Co, Cu, Al, and Ga) has beenemployed as a strategy to suppress the cation ordering behaviorand enhance the safety and electrochemical performance.10,11

This improvement has been attributed to factors such assegregation of the dopant ions to the surface, enhanced elec-tronic conductivity, elimination of the rock-salt phase impurity,and effective disruption of Mn4+ and Ni2+ cation ordering.12,13

One of the most effective ways to synthesize a disorderedspinel phase is to re the precursor materials to 900 �C and coolthe product to room temperature. An impurity Ni-rich rock-saltphase, identied herein as LixNi1�xO, typically forms at thistemperature; its contribution to the nal product can bereduced by cooling the sample at a slow rate.14 As a result of theformation of the rock-salt phase, some amount of Ni is robbedfrom the electrochemically active spinel, resulting in a Mn : Niratio >3 in the spinel phase. To maintain electrical chargeneutrality, some amount of Mn4+ is reduced to Mn3+, the

J. Mater. Chem. A, 2013, 1, 10745–10752 | 10745

http://dx.doi.org/10.1039/c3ta12021j

http://pubs.rsc.org/en/journals/journal/TA

http://pubs.rsc.org/en/journals/journal/TA?issueid=TA001036

Journal of Materials Chemistry A Paper

Publ

ishe

d on

12

July

201

3. D

ownl

oade

d by

Dal

hous

ie U

nive

rsity

on

23/0

8/20

13 1

5:21

:25.

View Article Online

presence of which is clearly identied in the charge/dischargecurves of the Li-ion cells as a voltage plateau at �4 V corre-sponding to the Mn3+/4+ redox couple. This plateau is absent ornegligible in a well-ordered spinel, indicating a 3 : 1 ratio of Mnto Ni in the spinel phase with an elimination of the rock-saltimpurity. The high-voltage spinel is typically transformed to theordered phase by annealing the product at 700 �C for anextended amount of time with a slow cooling rate.14 Dependingon the synthesis technique, sample preparation, Mn : Ni stoi-chiometry, annealing time, and dopant species concentration,it is thought that the relative degree of cation ordering canchange drastically, as can the electrochemical performance ofthe spinel as a cathode material.4 Thus, it is critical to developmethods by which the relative degree of cation ordering amongdifferent samples can be readily determined.

To this end, magnetic susceptibility and magnetizationmeasurements are emerging as a valuable tool to examine thechemistry and physics of the ordered and disordered phases,allowing for the differentiation between the two.15–20 Accord-ingly, this study is focused on correlating the magnetic prop-erties of doped LiMn1.5Ni0.42M0.08O4 (M¼ Cr, Fe, Co, Cu, Al, andGa) and Mn-rich LiMn1.5+xNi0.5�xO4 (0 # x # 0.1) to the Mn3+

content and relative degree of cation order. The results obtainedare compared with the electrochemical charge/discharge curvesof lithium-ion cells.

Experimental

All samples were synthesized with a continuously-stirred tankreactor (CSTR) with two feedstock solutions. The rst solution wasprepared at 2M concentration with the appropriate stoichiometricamount of metal salts necessary to produce the hydroxideprecursor. The metal salts used in the synthesis were:MnSO4$H2O, NiSO4$6H2O, Cr(NO3)3$9H2O, FeSO4$7H2O,CoSO4$7H2O, Cu(NO3)2$H2O, Al(NO3)3$9H2O, and Ga(NO3)3$H2O.Another solution with a 2 M concentration of NaOH and 0.05 Mconcentration of NH4OH was continuously added to the mixturewith the purpose of maintaining a solution pH of 10. The mixturewas stirred at 1000 rpm in a sealed nitrogen atmosphere at atemperature of 60 �C for 12 h. The coprecipitate formed wasltered and washed with deionized water and subsequently driedovernight in air at 110 �C. The water content of the metalhydroxide precursor was determined by thermogravimetric anal-ysis (TGA), aer which it was groundwith a stoichiometric amountof LiOH$H2O. This mixture was then red at 900 �C for 15 h andcooled at a rate of 1 �C min�1 to produce the nal product.A portion of thismaterial was also annealed at 700 �C for 48 h witha cooling rate of 1 �C min�1 in an effort to produce a cation-ordered spinel.

X-ray diffraction (XRD) patterns were collected with a Rigakupowder diffractometer with a 0.02 2q step size and a threesecond dwell time. Magnetic measurements were carried outwith a Quantum Design SQUID magnetometer on powdersamples of the raw cathode material (no carbon or binder)encased in glycerin capsules. Field-cooled (FC) measurementswere performed by cooling the sample in a eld of 0.1 T down to5 K and then measuring the DC susceptibility as the

10746 | J. Mater. Chem. A, 2013, 1, 10745–10752

temperature was increased to 300 K. Saturation magnetizationmeasurements were conducted at eld strengths of 1–4 T at 5 K.

Electrochemical measurements were carried out with Li-ionhalf-cells. The cathodes were prepared by a PVDF castingmethod. A slurry was prepared with 80 wt% active material, 10 wt% conductive carbon, and 10 wt% polyvinylidene uoride (PVDF)in a N-methyl-2-pyrrolidone (NMP) solvent. The slurry was stirredovernight with a magnetic stir bar in a glass vial and was thencast onto an aluminum-foil current collector and dried undervacuum at 100 �C for 12 h. Cathode disks loaded with approxi-mately 4 mg active material were punched from the sheets andassembled in CR2032 coin cells with a lithium metal anode and1 M LiPF6 in ethylene carbonate (EC)/diethyl carbonate (DEC)(1 : 2 vol.) as the electrolyte. Deep discharge (below 2 V) tests wereconducted at a current density of 10 mA g�1.

Results and discussionCrystal chemistry

The doped LiMn1.5Ni0.42M0.08O4 (M¼ Cr, Fe, Co, Cu, Al, and Ga)samples will be hereaer referred to by their dopant ion andsynthesis temperature (900 �C for the as-prepared samples or700 �C for the post-annealed samples), e.g., Cr900, Cr700, etc.The XRD patterns of the doped and Mn-rich samples are shownin Fig. 1. All samples are primarily composed of the cubic spinelphase; a trace amount of the rock-salt impurity LixNi1�xO ispresent in the doped Cu900 and undoped LiMn1.5Ni0.5O4

samples.

Magnetic interactions

Atomically ordered LiMn1.5Ni0.5O4 adopts a ferrimagneticallyordered spin structure below the Curie temperature in whichthe Mn4+ spins align antiparallel to the Ni2+ spins, yielding atheoretical net magnetization of 3.5 mB for each formula unit.19

Because Li+ is not magnetic, but Ni2+ and Mn4+ are, anymagnetic interactions between the 8a tetrahedral sites (A-sites)and the 16d octahedral sites (B-sites) are ignored and the B–Binteractions are solely of interest here.

The geometry of the cubic spinel structure creates a compe-tition between interatomic metal–metal interactions throughdirect overlap of the t2g orbitals and oxygen-mediated super-exchange through 90� t2g–O:2pps–eg pathways. High-spin Ni2+

holds two unpaired electrons in the eg orbitals, while Mn4+ hasthree unpaired electrons in the t2g orbitals. One might expect theinteraction between Mn4+ ions to favor antiferromagnetic (AF)order due to the direct t2g–t2g interaction between two unpairedelectrons, but in fact it appears that the Mn4+:t2g

3–O:2pps–Mn4+:eg

0 ferromagnetic (F) interaction dominates, presumablybecause the overlap integral between the t orbitals is too small tocompete with the AF Ni2+–Mn4+ interactions.19 Likewise, theNi2+:eg

2–O:2pps–Mn4+:t2g3 virtual electron transfer from the Ni2+

to the Mn4+ dictates that the interaction between the two isantiferromagnetic. The t2g–t2g overlap between the full Ni2+:t2g

6

orbitals and half-lled Mn4+:t2g3 orbitals are ferromagnetic, but

too weak to dominate the stronger antiferromagnetic interaction.Thus, if we assume a structure where Ni2+ is primarily

This journal is ª The Royal Society of Chemistry 2013


Fig. 1 XRD Patterns of (a) Mn-rich LiMn1.5+xNi0.5-xO4 (x = 0, 0.05, and 0.1) as wellas LiMn1.5Ni0.42M0.08O4 (M = Cr, Fe, Co, Cu, Al, and Ga) (b) synthesized at 900 �Cand (c) annealed at 700 �C. Insets show the most prominent rock-salt-phase peakat 43.5� in the undoped LiMn1.5Ni0.5O4 and Cu-doped LiMn1.5Ni0.42Cu0.08O4

samples synthesized at 900 �C.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

12

July

201

3. D

ownl

oade

d by

Dal

hous

ie U

nive

rsity

on

23/0

8/20

13 1

5:21

:25.

View Article Online

coordinated with Mn4+ ions, as in the ordered P4332 LiMn1.5-Ni0.5O4 spinel, we arrive at ferrimagnetic magnetization with anoverall magnetic moment of 3.5 mB per formula unit (or per Li+).

If there exists an excess Mn (>1.5) in the spinel structure or anon-magnetic trivalent cation is substituted for the divalentNi2+, some Mn4+ will be reduced to Mn3+ to maintain chargeneutrality. High-spin Mn3+ contains a single unpaired electronin the eg orbital, so the p–s 90� pathway between Mn4+ andMn3+ ions becomes antiferromagnetic, and the antiferromag-netic direct t2g–t2g overlap is retained. As a result, the spins onthe Mn3+ ions will align antiparallel to the Mn4+ spins andparallel to the Ni2+ spins. The resulting modication to theoverall magnetic moment can be computed as follows:


MmB/fu ¼ gSMn4+NMn4+ � gSNi2+NNi2+ � gSMn3+NMn3+ (1)

where g is the electron gyromagnetic factor (�2), S is the totalspin of the d-electrons on the ion indicated by the subscript andNion is the number of such ions in the formula unit (fu).

In this manner, the Mn3+ content of each sample not con-taining a magnetic dopant ion can be calculated from thesaturation magnetization.19 Certain dopant ions are non-magnetic (Co3+, Al3+, and Ga3+); therefore, their magneticinteractions with Ni2+, Mn3+, and Mn4+ are non-existent andmay be neglected. Cr3+, Fe3+, and Cu2+, on the other hand,contain at least one unpaired electron and so contribute to theoverall magnetization in the ferrimagnetic phase.

Considering the relatively low concentration of the dopantions (�4%), it is reasonable to ignore the dopant–dopantmagnetic interactions and to concentrate primarily on theinteraction between the dopant and Mn4+ ions since Mn4+ is themost common ion in the high-voltage spinel. Additionally,the dopant ions are likely to repel each other at high tempera-ture and coordinate primarily with Mn4+ ions to lower theelectrostatic energy of the lattice. If the assumption is made thatthe vast majority of the dopant ions will coordinate in thismanner in the material, it can also be assumed that all dopantion spins will be aligned in the same direction below TC. Amodied version of eqn (1), which includes the contribution ofthe magnetic dopants, can then be given as,

MmB/fu ¼ gSMn4+NMn4+ � gSNi2+NNi2+

� gSMn3+NMn3+ � gSMn+NMn+ (2)

where Mn+ is the dopant ion and n is the oxidation state of thedopant ion.

The octahedral Cu2+ ion having a t2g6eg

3 conguration can beexpected to substitute for Ni2+, and with one unpaired electronin the eg orbital, it can be assumed that Cu2+ will behave similarto Ni2+ and its S ¼ 1/2 spin aligns antiparallel to the spin of theMn4+ ions owing to a p–s 90� superexchange. On the basis ofboth direct t2g–t2g and t2g–O:2pps–eg overlaps, Fe

3+ should alignantiparallel to Mn4+ because the virtual charge transfer fromFe3+ to Mn4+ dominates that from Mn4+ to Fe3+.

The Cr3+ ion has the same d3-electron conguration as Mn4+,so one might expect the Cr3+ spin to align in the same directionas that of Mn. However, the calculated Mn3+ content under thisassumption is 0.23, which is larger than the value obtainedfrom the electrochemical data (0.18). A much better t to theelectrochemical data is obtained when the spin on the Cr3+ ionsis assumed to be antiparallel to that of the Mn4+ ions. Anantiferromagnetic exchange between Cr3+ and Mn4+ is reason-able and can be explained as below. Because Cr has a smallernuclear charge than Mn, the electron orbitals are extendedfurther (alternatively, Cr3+ is larger in size than Mn4+). Theimplication is that the direct t2g–t2g interaction between a Cr3+

and Mn4+ would have a much larger magnitude than theinteraction between two Mn4+ ions, large enough to make theantiferromagnetic exchange between the two half-lled t2gorbitals dominant over the t2g–O:2pps–eg superexchange, as isthe case of l-MnO2.21 In this situation the 3/2 spin on the Cr3+

J. Mater. Chem. A, 2013, 1, 10745–10752 | 10747



Publ

ishe

d on

12

July

201

3. D

ownl

oade

d by

Dal

hous

ie U

nive

rsity

on

23/0

8/20

13 1

5:21

:25.

View Article Online

would align antiparallel to the majority spin direction of theMn4+ ions. Nevertheless, it is entirely possible that some Cr3+

spin may be aligned parallel to the Mn4+ spin depending on thenumber of coordinated Ni ions (the Cr:t2g–O:2pps–Ni:egexchange is antiferromagnetic). This uncertainty is recognizedand noted in Table 1.

Table 1 shows the calculated Mn3+ contents of all the dopedand Mn-rich samples by eqn (2) (where Mn3+ content is thestoichiometric amount of Mn3+ ions per formula unit). Thevalues of Nion for the doped samples were such that the totalamount of Mn was 1.5 per fu. For theMn-rich samples, theMn4+

and Mn3+ contents would sum to equal 1.5, 1.55, and 1.6. Theresults from the magnetization study are compared with theelectrochemical data of Li-ion cells. Fig. 2 displays the rstdischarge of the doped samples down to 1.8 V. Because theMn3+/4+ couple in the high-voltage spinel is centered around 4 Vand no other Faradaic process occurs at that potential (exceptfor the Cu2+/3+ couple at �4.2 V), the Mn3+ content can beestimated by comparing the capacity in the 4 V region to theoverall capacity. It can be seen in Fig. 3 that the Mn3+ valuesmatch very closely, validating that saturation magnetization is auseful metric for estimating the amount of Mn3+ present in awide variety of doped or undoped high-voltage spinel samples.

Electrochemical voltage proles

Another useful quantity to be derived from the electrochemicaldischarge curves is the capacity in the �2.7 V region when thecells are subjected to deep discharge. This capacity correspondsto the insertion of an additional lithium ion into the 16c octa-hedral sites of the spinel lattice, which are normally empty inthe Li[Mn1.5Ni0.5]O4 spinel.22 As lithium is inserted into the 16coctahedral sites, a two-phase reaction proceeds in whichMn4+ is

Table 1 Magnetization, Mn3+ content, lattice parameter, 2.7 V capacity, and CuLiMn1.5+xNi0.5�xO4 (0 # x # 0.1) samples prepared at 900 �C and after post-anntemperature dependent magnetic susceptibility curve

SampleMagnetization(mB per fu)

Mn3+ content(magnetic)

Mn3+ co(electroc

Cr900 2.26 (3) 0.17 (6) 0.18Co900 2.26 (4) 0.20 (1) 0.18Fe900 2.46 (2) 0.11 (1) 0.11Cu900 2.79 (2) 0.11 (1) 0.09Al900 2.06 (4) 0.23 (1) 0.21Ga900 2.11 (3) 0.22 (1) 0.24Mn1.5 2.85 (3) 0.15 (1) 0.15Mn1.55 2.72 (3) 0.17 (1) 0.19Mn1.6 1.92 (5) 0.35 (1) 0.32Cr700 2.28 (5) 0.16 (6) 0.18Co700 2.57 (3) 0.16 (1) 0.16Fe700 2.58 (2) 0.10 (1) 0.11Cu700 3.01 (3) 0.08 (1) 0.05Al700 2.19 (6) 0.21 (1) 0.24Ga700 2.10 (5) 0.22 (1) 0.25Mn1.5–700 3.45 (1) 0.01 (1) 0.06Mn1.55–700 3.00 (2) 0.11 (1) 0.15Mn1.6–700 2.22 (5) 0.28 (1) 0.29

10748 | J. Mater. Chem. A, 2013, 1, 10745–10752

reduced to Mn3+, fostering a Jahn–Teller induced tetragonaldistortion of the cubic spinel structure. Recently it was shownthat a second tetragonal spinel phase forms while under load asa result of an additional distortion associated with Li inser-tion.23 When this second tetragonal phase develops, thedischarge voltage is decreased from 2.7 V to �2 V vs. Li/Li+. It sohappens that the ordered spinel structure with the P4332symmetry possesses larger 16c octahedral sites than the Fd�3mphase owing to the preferential ordering of the Mn4+ and Ni2+

ions. Because the ordered-spinel regions accommodate Li ionspreferentially in their larger 16c sites, isolated ordered regionshave the Mn3+ concentration needed for a cooperative room-temperature Jahn–Teller distortion before the disorderedmatrix in which they are embedded. Tetragonal regions in acubic matrix create a stress on the tetragonal phase that limitsthe magnitude of its c/a ratio. However, at a certain onset Liconcentration, the disordered regions also have a high enoughMn3+ concentration to participate in the cooperative Jahn–Teller distortion, thus relieving the stress to allow the largertetragonal c/a > 1 of the Li-rich phase. This onset can be visu-alized as a sudden change in voltage from 2.7 to 2.0 V in thedeep-discharge curves. Therefore, the highly ordered spinelsamples with a higher volume of atomically ordered regionsexhibit a larger capacity in the 2.7 V region than the disorderedsamples.23 This technique has already been used to determinethe degree of cation order of a few doped high-voltage spinelsamples synthesized in a different manner.24 Thus, the 2.7 Vplateau serves as a means of semi-quantitatively assessing thedegree of cation order in high-voltage spinels with similarcompositions.

The deep-discharge curves of the series LiMn1.5+xNi0.5�xO4

with x ¼ 0, 0.05, 0.1, synthesized at 900 �C and also annealed at700 �C (to promote the ordered phase) are shown in Fig. 2(b).

rie temperature of LiMn1.5Ni0.42M0.08O4 (M ¼ Cr, Fe, Co, Cu, Al, and Ga) andealing at 700 �C. The Curie temperature is taken as the inflection point of the

ntenthemical)

Lattice constant(A)

2.7 V capacity(mA h g�1) Tc (K)

8.188 (1) 69 94 (2)8.178 (2) 51 95 (2)8.183 (1) 82 107 (1)8.176 (2) 83 111 (4)8.185 (2) 54 91 (4)8.197 (2) 54 91 (4)8.175 (1) 54 116 (2)8.178 (1) 62 119 (3)8.186 (2) 71 106 (4)8.190 (1) 69 95 (2)8.177 (1) 55 101 (2)8.183 (1) 82 110 (1)8.175 (2) 100 120 (3)8.183 (1) 49 93 (2)8.194 (1) 60 90 (4)8.174 (1) 67 124 (1)8.177 (1) 84 122 (1)8.187 (2) 87 120 (2)



Fig. 2 First discharge curves of Mn-rich LiMn1.5+xNi0.5�xO4 (x ¼ 0, 0.05, 0.1) (a) synthesized at 900 �C and (b) annealed at 700 �C and doped LiMn1.5Ni0.42M0.08O4

(M ¼ Cr, Fe, Co, Cu, Al, and Ga) (c) synthesized at 900 �C and (d) annealed at 700 �C.

Fig. 3 Relationship between Mn3+ content determined by saturation magneti-zation (from magnetic data) and Mn3+ content determined by the 4 V capacity(from electrochemical data).


Publ

ishe

d on

12

July

201

3. D

ownl

oade

d by

Dal

hous

ie U

nive

rsity

on

23/0

8/20

13 1

5:21

:25.

View Article Online

From here on, the Mn-rich samples will be denoted by theatomic symbol for manganese followed by the manganesecontent of the sample. If the sample has been annealed at700 �C, these annotations will be followed by the number 700,e.g., Mn1.6, Mn1.6–700, etc. Two notable trends are visible fromthese data concerning the 2.7 V capacity: it increases signi-cantly as the Mn content is increased or aer the material hasbeen annealed at 700 �C. Another noticeable piece of informa-tion is that the Mn1.6 sample has a larger 2.7 V capacity thanthe Mn1.5–700 sample. In this case, distinguishing the degreeof ordering via the 2.7 Vmethod is difficult due to the differencein Mn3+ content between the two samples. It is reasonable toassume that this difficulty also applies to the doped samples,


which display a wide-ranging variation of Mn3+ content, from0.08 to 0.23.

There are several explanations as to why an increase in theMn3+ content is associated with an increase in the 2.7 V plateauin the Mn-rich samples. The most basic explanation concernsthe unit cell volume and the size of the empty octahedral sites.Because the Mn3+ ion is much larger than Mn4+ ions owing tothe extra electron in the eg orbital, samples with a higher Mn3+

content tend to have a larger lattice parameter. This is evident inthe data presented in Table 1. Amore expanded unit cell volumewill result in larger octahedral sites, resulting in less stress onthe structure during Li insertion and thus delaying the onset ofthe second tetragonal phase associated with the 2.0 V plateau. Itis also possible that mobile Mn3+ ions are attracted to theordered phase upon insertion of lithium into the octahedralsites in an effort to foster cooperative Jahn–Teller distortion.The ordered phase, with the larger 16c sites, will ll with Libefore the disordered phase. The elastic strain energy of atetragonal phase enclosed by a cubic phase can be reduced ifJahn–Teller ions surround the tetragonal phase and participatein the distortion. In this manner, the strain energy cold berelieved and the 2.7 V plateau extended. Thus, the two param-eters that appear to govern the 2.7 V plateau length, Mn3+

content and the degree of order, both extend the 2.7 V plateau.However, the contribution of the degree of order to the 2.7 Vcapacity appears to be more signicant than the contribution bythe expanded cell volume due to higher Mn3+ content. This isevidenced by Fig. 4, which shows that a reliable trend existsamong the doped samples in which the 2.7 V plateau increaseswith decreasing Mn3+ content, indicating that the degree oforder may be the dominant factor in determining the 2.7 Vcapacity.

J. Mater. Chem. A, 2013, 1, 10745–10752 | 10749


Fig. 4 Relationship between 2.7 V capacity and Mn3+ content for LiMn1.5-Ni0.42M0.08O4 (M ¼ Cr, Fe, Co, Cu, Al, and Ga).


Publ

ishe

d on

12

July

201

3. D

ownl

oade

d by

Dal

hous

ie U

nive

rsity

on

23/0

8/20

13 1

5:21

:25.

View Article Online

TC as an indicator of atomic order

It has been suggested that the DC susceptibility provides areliable means of investigating magnetic clustering of the high-voltage spinels with the implication that magnetic clustering isdirectly related to the degree of long-range order.15,17 Whileprevious work has approached the subject, an in-depth quali-tative study of multiple samples comparing the DC suscepti-bility curve and the degree of cation ordering has not beenestablished. The eld-cooled DC susceptibility curves for thedoped compositions can be seen in Fig. 5. The ferrimagneticordering temperature (TC) can be identied easily by the sharpchange in susceptibility around 90–125 K. Such a large variationin TC reects the existence of structural and compositionaldifferences among the samples. The magnetic orderingtemperature of a variety of materials can commonly be affectedby a range of parameters, including dopant concentration andheat treatment as is the case with Si-doped cobalt ferrite.25 It hasalready been suggested that the TC of LiMn1.5Ni0.5O4 is higher

Fig. 5 Variation of DC magnetic susceptibility with temperature of LiMn1.5-Ni0.42M0.08O4 (M ¼ Cr, Fe, Co, Cu, Al, and Ga).

10750 | J. Mater. Chem. A, 2013, 1, 10745–10752

when the material possesses a high degree of cation order.15,17,19

It follows that a more disordered structure, possessing a widevariation in local composition, will have a more gradualmagnetic transition with a lower Curie temperature. It has alsobeen suggested that different dopant ions will induce a differentamount of cation disorder in high-voltage spinels.24 Thus,measuring the TC can be practically applied as a way to deter-mine the relative degree of cation order among a group ofmaterials with similar compositions.

The Curie temperatures of the Mn-rich samples are listed inTable 1. All of the annealed samples possess a higher TC thantheir unannealed counterparts, consistent with the under-standing that annealing promotes cation ordering and thatordered materials maintain higher Curie temperatures. Themore uniform distribution of Mn and Ni in the ordered struc-ture prevents excessive Mn4+–Mn3+ and Mn3+–Mn3+ interac-tions, which can cause frustration as in the case of LiMn2O4.26

By virtue of the fact that TC is lowered by magnetic frustrationand that a more ordered high-voltage spinel is typically lessfrustrated than the disordered form, a qualitative connectioncan be made between the degree of cation order in high-voltagespinels and the measured value of the Curie temperature.

Fig. 6 shows the qualitative relationship between TC and the2.7 V capacities of the doped samples. This relationship can beidentied based on the correlation between Mn3+ and the 2.7 Vplateau as previously discussed (Fig. 4). At rst glance, it wouldseem that the TC is strictly dependent on the Mn3+ content ofeach sample based on the strong correlation seen in Fig. 4.However, since Mn3+ content and the degree of cation orderingtypically go hand-in-hand, sometimes it is difficult to isolate theeffects of the two. Fig. 7 compares the Mn3+ content and TC withthe 2.7 V capacity for the Mn-rich samples, but no clear corre-lation is seen. This is to be expected considering the widevariation of Mn content in each sample based on the argumentspresented earlier. However, it is clear that the annealing processincreases the TC and the 2.7 V capacity while decreasing theMn3+ content in every case. It is important to note that theMn1.6 sample, while expected to have more disorder than

Fig. 6 Relationship between 2.7 V capacity and the Curie temperature forLiMn1.5Ni0.42M0.08O4 (M ¼ Cr, Fe, Co, Cu, Al, and Ga).



Fig. 7 Correlation of 2.7 V capacity with (a) TC and (b) Mn3+ content for the Mn-rich LiMn1.5+xNi0.5�xO4 (x ¼ 0, 0.05, 0.1) samples. Solid and open symbols refer,respectively, to 900 and 700 �C samples. In (b), circles and triangles refer,respectively, to data from electrochemical and magnetic measurements.


Publ

ishe

d on

12

July

201

3. D

ownl

oade

d by

Dal

hous

ie U

nive

rsity

on

23/0

8/20

13 1

5:21

:25.

View Article Online

the Mn1.5–700 sample, has a larger 2.7 V capacity. In this case,the TC provides a much better idea of the relative degree ofcation ordering between the two samples.

As noted earlier among the doped samples, however, a cleartrend exists in which the 2.7 V plateau increases as the Mn3+

content decreases, a trend which is exactly opposite to that inthe Mn-rich samples. In this case it must be assumed that thedegree of ordering, which depends heavily on the nature of thedopant ion, has a more pronounced effect on the 2.7 V capacitythan the Mn3+ content.

Among the doped samples, it is easy to discern that the mostordered sample of the group is Cu-700, having both the highestTC and largest 2.7 V capacity as well as the smallest amount ofMn3+. The Cu doped compound is in fact the only doped sampleto display a change in TC and 2.7 V capacity upon annealing to acomparable extent as the undoped Mn-rich materials. Thisresult is not surprising, considering the similarity between thevalence and ionic radii of Cu2+ and Ni2+ ions, allowing Cu tobehave in much the samemanner as Ni. Therefore, the ability ofthe Cu-doped samples to order atomically at 700 �C is similar tothat of the undoped samples.

On the opposite end of the spectrum, the Al and Ga samplesare characterized by both the lowest Curie temperatures and thesmallest 2.7 V capacities. The fact that neither the 2.7 V capacitynor the TC change by a signicant amount aer annealingimplies that the Al and Ga dopants provide an effective meansof inhibiting cation ordering, with the consequence that thesematerials are highly disordered.


As in the case of Al and Ga, replacing some Ni with Crappears to prevent the spinel from ordering; the Cr-900 andCr-700 samples possess the same TC, 2.7 V capacity, and Mn3+

content. Based on the Curie temperature, one would expect theCr-doped samples to be less ordered than the Co-doped coun-terparts. However, the 2.7 V capacities imply just the opposite.Cr900 and Cr700 are characterized by larger lattice parameters(8.188 A and 8.190 A, respectively) compared to the Co900(8.178 A) and Co700 (8.177 A) samples, which may explain thediscrepancy. The 2.7 V capacities of the Cr-doped samples maybe larger than those of Co900 and Co700 not because of a higherdegree of cation order, but because the larger lattice volume ofthe Cr-doped samples delays the onset of the strain-inducedtetragonal distortion. Based on the TC, it would appear asthough the Cr-doped samples are similarly ordered as theCo900 sample.

Among the trivalent doped samples (Cr, Fe, Co, Al, and Ga),the Fe-doped sample stands out as the most ordered, possess-ing both the highest Curie temperature and 2.7 V plateaulength. It is found that upon annealing, the TC increases slightlyimplying that the Fe-700 sample is only marginally moreordered than the Fe-900 sample. The saturation magnetizationshows the Mn3+ concentration also decreases slightly, whichmay explain why the 2.7 V capacity is nearly identical for the twosamples.

Interpretation of the data for the Co-doped sample is madecomplicated by the following: (i) The Co3+/Co2+ redox energy isonly a little below the Mn4+/Mn3+ energy so that the Co2+ +Mn4+ ¼ Co3+ + Mn3+ equilibrium is biased more strongly to theright with increasing temperature. (ii) The low-spin (LS) Co3+

state transforms increasingly to an intermediate spin (IS) statewith increasing temperature, and this transition is biased in thepresence of a cooperative Jahn–Teller distortion on the Mn3+

ions because IS Co3+ is also a strong Jahn–Teller ion. (iii) If theCo2+ ion is more mobile than Ni2+ and more easily displacedinto a rock-salt phase on loss of oxygen above 700 �C, and a Co-rich rock-salt phase may not be reincorporated into the spinelphase at 700 �C as is the Ni-rich rock-salt phase.

Conclusions

With an aim to develop a better understanding of the complexand oen confusing roles of cation ordering and Mn3+ contenton the electrochemical properties of the high-voltage LiMn1.5-Ni0.5O4 spinel cathodes, magnetic measurements have beenpursued as a viable tool. The saturation magnetization providesa reliable means of calculating the Mn3+ content in the dopedsamples, as conrmed by the electrochemical capacity data. It isfound that among the doped samples, the TC increases withincreasing 2.7 V plateau length and decreasing Mn3+ content,both of which are indicative of higher degrees of cation order.Thus, the Curie temperature of doped or Mn-rich high-voltagespinel samples may be a useful metric for qualitatively deter-mining the degree of cation order among samples with verydifferent Mn3+ contents. This is important because it has beenshown in the literature to be a factor in controlling the perfor-mance of the samples as cathodes in Li-ion batteries.

J. Mater. Chem. A, 2013, 1, 10745–10752 | 10751



Publ

ishe

d on

12

July

201

3. D

ownl

oade

d by

Dal

hous

ie U

nive

rsity

on

23/0

8/20

13 1

5:21

:25.

View Article Online

Notes and references

1 A. Manthiram, J. Phys. Chem. Lett., 2011, 2, 176.2 J. B. Goodenough and Y. Kim, Chem. Mater., 2010, 22, 587.3 M. Thackeray, C. Wolverton and E. D. Isaacs, Energy Environ.Sci., 2012, 5, 7854.

4 J. Zheng, J. Xiao, X. Yu, L. Kovarik, M. Gu, F. Omenya,X. Chen, X.-Q. Yang, J. Liu, G. L. Graff, M. S. Whittinghamand J.-G. Zhang, Phys. Chem. Chem. Phys., 2012, 14, 13515.

5 N. Amdouni, K. Zaghib, F. Gendron, A. Mauger andC. M. Julien, Ionics, 2006, 12, 117.

6 D.-W. Shin, C. A. Bridges, A. Huq, M. P. Paranthaman andA. Manthiram, Chem. Mater., 2012, 24, 3720.

7 K. R. Chemelewski, D.-W. Shin, W. Li and A. Manthiram,J. Mater. Chem. A, 2013, 1, 3347.

8 S. Patoux, L. Sannier, H. Lignier, Y. Reynier, C. Bourbon,S. Jouanneau, F. Le Cras and S. Martinet, Electrochim. Acta,2008, 53, 4137.

9 L. Wang, H. Li, X. Huang and E. Baudrin, Solid State Ionics,2011, 193, 32.

10 G. B. Zhong, Y. Y. Wang, X. J. Zhao, Q. S. Wang, Y. Yu andC. H. Chen, J. Power Sources, 2012, 215, 368.

11 J. Molenda, Solid State Ionics, 2004, 175, 297.12 J. Liu and A. Manthiram, J. Phys. Chem. C, 2009, 113, 15073.13 Y. Kadoma, S. Sato, K. Ui and N. Kumagai, Electrochemistry,

2010, 8, 658.14 J. Song, D.-W. Shin, Y. Lu, C. D. Amos, A. Manthiram and

J. B. Goodenough, Chem. Mater., 2012, 24, 3103.

10752 | J. Mater. Chem. A, 2013, 1, 10745–10752

15 K. Mukai and J. Sugiyama, J. Electrochem. Soc., 2010, 157,A672.

16 N. Amdouni, K. Zaghib, F. Gendron, A. Mauger andC. M. Julien, J. Magn. Magn. Mater., 2007, 309, 100.

17 J. Cabana, F. O. Omenya, N. A. Chernova, D. Zeng,M. S. Whittingham and C. P. Grey, Chem. Mater., 2012, 24,2952.

18 N. Kawai, T. Nakamura, Y. Yamada and M. Tabuchi, J. PowerSources, 2011, 196, 6969.

19 Z. Moorhead-Rosenberg, D.-W. Shin, K. R. Chemelewski,J. B. Goodenough and A. Manthiram, Appl. Phys. Lett.,2012, 100, 213909.

20 T. Nakamura, Y. Yamada and M. Tabuchi, J. Appl. Phys.,2005, 98, 093905.

21 J. E. Greedan, N. P. Raju, A. S. Wills, C. Morin andS. M. Shaw, Chem. Mater., 1998, 10, 3058.

22 K. Ariyoshi, Y. Iwakoshi, N. Nakayama and T. Ohzuku,J. Electrochem. Soc., 2004, 151, A296.

23 E.-S. Lee, K. Nam, E. Hu and A. Manthiram, Chem. Mater.,2012, 24, 3610.

24 E.-S. Lee and A. Manthiram, J. Mater. Chem. A, 2013, 1,3118.

25 J. A. Paulsen, C. C. H. Lo, J. E. Snyder, S. Member,A. P. Ring, L. L. Jones and D. C. Jiles, IEEE Trans. Magn.,2003, 39, 3316.

26 C. R. Wiebe, P. L. Russo, a. T. Savici, Y. J. Uemura,G. J. MacDougall, G. M. Luke, S. Kuchta and J. E. Greedan,J. Phys.: Condens. Matter, 2005, 17, 6469.



Magnetic measurements as a viable tool to assess the relative degrees of cation ordering and Mn3+...

Documents

Transcript of Magnetic measurements as a viable tool to assess the relative degrees of cation ordering and Mn3+...