xLi2MnO3-(1-x)LiNi13Co13Mn13O2 (x=0.5) layered complex cathode showing voltage hysteresis

Electrochimica Acta 146(2014)79–88Contents lists available at ScienceDirectElectrochimicaActaj o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /e l e c t a c taElectrochemical study on x Li 2MnO 3-(1-x )LiNi 1/3Co 1/3Mn 1/3O 2(x =0.5)layered complex cathode showing voltage hysteresisMasahiro Kasai a ,∗,Shin Nishimura a ,Akira Gunji b ,Hiroaki Konishi b ,Xiaoliang Feng b ,Sho Furutsuki b ,Shin Takahashi ba International Research Center for Hydrogen Energy,Kyushu University,744Motooka,Nishi-ku,Fukuoka 819-0395,Japan bHitachi Research Laboratory,Hitachi,Ltd.,7-1-1Omika,Hitachi,Ibaraki 319-1292,Japana r t i c l e i n f o Article history:Received 1May 2014Received in revised form 13August 2014Accepted 13August 2014Available online 15September 2014Keywords:Lithium ion battery Cathode materials Layered cathodeExcess lithium compound OCP hysteresisa b s t r a c tThe layered complex cathode has attracted much interest recently due to its high capacity,which exceeds 250mAhg −1.On the other hand,it also shows large voltage hysteresis which causes energy loss when applied as a practical battery material.It has been reported that the hysteresis is an intrinsic bulk property,and the phenomenological model due to the migration of transition metal has been also proposed.Then,it is considered that to study the hysteresis for various compositions of transition metal is valuable to clar-ify the phenomenon.In this work,the layered complex cathode of 0.5Li 2MnO 3-0.5LiNi 1/3Co 1/3Mn 1/3O 2was investigated.The results are discussed in comparison with the previous reports for 0.5Li 2MnO 3-0.5LiMn 0.5Ni 0.5O 2.Investigating d Q /d V profilesrevealed that redox peaks at around 3.75V showed small dependency on composition of transition metal,while it seemed that a reduction peak at 3.26V was inde-pendent on the composition.Precise measurement of open circuit potential vs.capacity characteristics demonstrated that the energy dissipation during a cycle was abut 26kJmol −1.©2014Elsevier Ltd.All rights reserved.1.IntroductionRecently,there has been intensive investigation on the x Li 2MnO 3-(1−x )LiMO 2(M =Ni,Co,Mn)layered complex cath-ode [1,2].This material is very attractive as a high-energy-density cathode material for lithium ion batteries because it has a high capacity that exceeds 250mAhg −1at a charge voltage above 4.5V [3].However,the layered complex cathode shows a large hysteresis in the potential versus capacity curves.Cell potentials for the cor-responding capacity show different values between the charging and discharging process.Such differences in the potential decrease the energy efficiency of the battery because the discharge poten-tial is lower than the charge potential.It is considered that there are two possible causes for the hysteresis.One is the intrinsic electrochemical properties of the bulk material and the other is a large overpotential as a surface property.In general,every cath-ode material has a certain magnitude of overpotential,as shown in Fig.1(a).Therefore,the small deviation from the equilibrium potential is observed in the charge and discharge curves.When the overpotential is rather large,as shown in Fig.1(b),or the∗Corresponding author.Tel.:+81928023270.E-mail address:kasai.masahiro.987@m.kyushu-u.ac.jp (M.Kasai).capacity dependency of the equilibrium potential itself has a hysteretic characteristic as shown in Fig.1(c),hysteresis will appear in the charge/discharge curves.If the hysteresis is caused by the intrinsic bulk properties,then the strategy for improving the material will be optimization of the chem-ical composition or cation substitution;if it is caused by the overpotential,the optimization will include surface modifica-tion or changes in the particle structure.Therefore it is very important to clarify the origin of the hysteresis.The earliest work referring to the hysteresis in charge/discharge curves was reported by Ohzuku et al.[4]for x Li 2MnO 3-(1−x )LiMn 0.5Ni 0.5O 2(that is Li[Li 1/5Ni 1/5Mn 3/5]O 2in Dahn’s notation)cathode.They considered that the hysteresis could not be explained by kinetic effect because the hysteresis loop observed at 55◦C is larger than that at room temperature.However,the origin of the hysteresis was still unclear then.Croy et al.also investigated the hysteresis for the same cathode as Ohzuku and some other layered complex cathodes [5]They examined OCP vs.capacity characteristics vary-ing the lower and upper limit voltages,and investigated structural change during a charge/discharge procedure using an in-situ and ex-situ X-ray diffraction method.Croy et al concluded that the hysteresis was an intrinsic bulk property of the layered complex cathode,which accompanied by crystallographic change,and was caused by a small amount of lithium which experienced a hystere-/10.1016/j.electacta.2014.08.0730013-4686/©2014Elsevier Ltd.All rights reserved.80M.Kasai et al./Electrochimica Acta 146(2014)79–88SmallOverpotential ηEquilibrium PotentialP o t e n t i a lCapacity LargeOverpotential ηEquilibrium PotentialP o t e n t i a lCapacity HystereticEquilibrium PotentialP o t e n t i a lCapacity(a)(b)(c)Fig.1.Schematic illustrations of the relationships for the OCP and overpotential.(a)A case showing a small overpotential,(b)a case showing a large overpotential and (c)hysteretic OCP.sis about 1V in the site energy.They considered that the difference in the site energy was result from reversible migration of tran-sition metal ions,which could affect that of neighboring Li ions,between a metastable tetrahedral site in the lithium layer and the octahedral site in the transition metal layer.Gallagher et al.extended the idea to propose the phenomenological model cor-relating the reversible hysteresis and the irreversible voltage fade [6][7].According to the model,the migrated transition metal ions could return to the original octahedral sites when sufficient driv-ing force is applied.The reduction potential which acted as the driving force was observed as a 1eV hysteresis in a d Q /d V curve.It is considered that the elementary processes of the migration,when the external potential is applied to the electrode,are com-posed of simultaneous occurrence of lithium insertion and diffusion in the cathode material,valence change of the transition metal ions and subsequent structural change.All these process will show dependency of the transition metal composition to some extent [8].Therefore,it is expected that to study the hysteretic behavior for cathode materials having various composition of transition metals will give us important information about the phenomenon.The hysteresis of redox peaks in d Q /d V curves indicates the dif-ference of the site energy between the desertion and reinsertion state.On the other hand,the hysteresis width in the OCP versus capacity characteristics gives total dissipation of the Gibbs free energy during a cycle,which is consumed due to heat loss accompa-nied by an electrochemical reaction or irreversible structural decay etc.Generally speaking,the equilibrium potential of a lithium insertion material is determined by the difference of the Gibbs free energy between the reactant and the product.For exam-ple,in the case of the cathode material LiMO 2(M:transition metal),the equilibrium potential (in this paper,OCP is used to denote the equilibrium potential as they are equivalent)is given by E =−[G (LiMO 2)−G (x Li ++xe −+Li 1−x MO 2)]/nF .Here,G ()indicates the Gibbs free energy of the reactant or product of the corresponding lithium composition,given by x ,n is the number of electrons,and F is the Faraday constant.The OCP versus capacity characteristics are schematically shown in Fig.2(a)for the case in which the OCP is determined by composition x ,as described above.The OCP shows no hysteresis in the charge/discharge process,and no rapid change is observed in the OCP.A rapid decay of OCP occurs by elastic deformation,as shown in Fig.2(b).Hirai reported such behavior for Li 2/5Sn alloys [9].The potential drop was estimated by the change in the Gibbs free energy due to elastic strain as G /nF .It is considered that the potential change is reversible because the origin is elastic deformation.An OCP versus capacity characteristic accompanied by an energy loss during a cycle is also schematically shown in Fig.2(c).In such cases,hysteresis in the OCP versus capac-ity characteristics appears and the change in Gibbs free energy is given by G ABCD =−nF (E AB −E CD )during a cycle.The hysteresis observed for the layered complex cathode corresponds to the case shown in Fig.2(c).Zhu and Wang,and Dreyer et al.reported the hysteretic behavior of the OCP for a LiFePO 4cathode [10,11].They estimated the accommodation energy accompanied by phase sep-aration to be on the order of 10mV by applying the elastic/plastic deformation model for Ni–MH alloys [12].Croy et al.already reported properties for x Li 2MnO 3-(1-x )LiMn 0.5Ni 0.5O 2,and revealed that hysteresis width increased as the contents x of Li 2MnO 3component increased [13].However,dependency of the hysteresis width on the transition metal com-position is not still clear,as well as dependency of the redox peaks.There are various cathode materials in a ternary layered oxide system LiMO 2(M:Ni,Co,Mn),some of which are commercially available,for example LiNi 1/3Co 1/3Mn 1/3O 2,LiNi 0.5Co 0.2Mn 0.3O 2or LiNi 0.4Co 0.2Mn 0.4O 2.The LiNi 1/3Co 1/3Mn 1/3O 2cathode is the most popular and well-studied material among the family.In this work,we will start a series of systematic studies for dependency on the transition metal composition from the 0.5Li 2MnO 3-0.5LiNi 1/3Co 1/3Mn 1/3O 2cathode.Obtained electro-chemical characteristics are discussed in comparison with theCapacityCapacity CapacityO C PO C PO C PABCDΔG/nF(a)(b)(c)Fig.2.Schematic illustrations of three kinds of OCP vs.characteristics.(a)A conventional lithium insertion material,(b)OCP change by elastic lattice strain and (c)Hysteretic OCP.M.Kasai et al./Electrochimica Acta 146(2014)79–8881results of the previously reported 0.5Li 2MnO 3-0.5LiMn 0.5Ni 0.5O 2cathode.The characteristics of the LiNi 1/3Co 1/3Mn 1/3O 2cathode are also briefly described as a typical conventional insertion cathode.2.ExperimentalsThe x Li 2MnO 3-(1−x )LiNi 1/3Co 1/3Mn 1/3O 2(x =0.5)cathode powder was synthesized with spray-dried acetate precursors of the transition metals.The precursor powder was lithiated with LiOH at 850◦C for 12h in air.Electrodes were fabricated by mixing the powder with a solution of polyvinylidene difluoride (PVDF)in N-methyl-2-pyrrolidone,using carbon powder as a conductive agent in a cathode:carbon:PVDF ratio of 85:10:5wt%.The obtained slurry was coated on aluminum foil and pressed after vacuum drying.The typical electrode density was 2.2gcm −3and the thickness was approximately 50 m .Three-terminal electro-chemical cells were fabricated and measured in an Ar-circulated glovebox.Lithium metal foil was used as the counter and reference electrodes.Ethylene carbonate (EC)/ethyl methyl carbonate (EMC)(1/2,v/v)with 1M LiPF 6was used as the electrolyte.Four cells were fabricated for this study and all cells were initialized by charging at a current density of 0.11mA cm −2using the constant current–constant voltage (CCCV)method,followed by a CC dis-charge procedure at the same current density.The CV was set at 4.3,4.4,4.5,and 4.6V for each cell and the cut-off current density was 0.011mA cm −2.The cut-off voltage for the discharge process for all cells was 2.5V.The current density of 0.11mA cm −2corresponds to the 0.05C rate for the discharge capacity of the 4.6-V-charged cell.The GITT was used to investigate the OCP versus capacity characteristics after the initialization process.A battery tester (Hokuto SM8)with a voltage resolution of 0.3mV was used for the GITT measurements.A GITT pulse was applied to a cell for 20min followed by a 10h rest time.The OCP was determined as the potential 10h after the GITT pulse was terminated.Redox peaks were examined by using d Q /d V curves,which were obtained by numerically differentiating the constant-current region in the charge/discharge curves of the 2nd -mercially purchased LiNi 1/3Co 1/3Mn 1/3O 2cathode powder was examined in the same way as a reference material.3.Results and DiscussionA charge/discharge initialization procedure was conducted to stabilize the cathode material and confirm the initial capacity for the four upper-limit voltages before starting the GITT measure-ments.A LiNi 1/3Co 1/3Mn 1/3O 2cathode was tested at the same time for demonstrating a typical behavior of a conventional insertion cath-ode.Potential versus capacity curves for the first and second cycles are shown in Fig.3and 4respectively for various upper-limit volta-ges.As shown in Fig.3,there is no rapid change in the curve profile according to the upper-limit voltages for the LiNi 1/3Co 1/3Mn 1/3O 2cathode.Although potential hysteresis is observed for each 1st cycle due to the irreversible capacity,there is little hysteresis shown in the each 2nd cycle.The behavior is exactly corresponding to the Fig.1(a)described above.On the other hand,in the case of the layered complex cathode of 0.5Li 2MnO 3-0.5LiNi 1/3Co 1/3Mn 1/3O 2drastic change in the curve profile was observed.For upper-limit voltages of 4.3and 4.4V,the capacity is not very high,approximately 100mAhg −1.However,a rapid increase of the capacity is observed as the upper-limit voltage exceeds 4.5V,with the simultaneous appearance of a plateau region in the charge curve.The deviation between the first and second cycle curves quickly increases,which indicates an increase in the irreversible capacity for the first cycle.Here,it is considered that the observed irreversible capacity includes the energy loss due to a certain kind of conver-sion reaction of the cathode for the upper-limit voltages of 4.5and 4.6V.Therefore,the “coulombic efficiency”described in this work is a “nominal”value,which is obtained by simply dividing the dis-charge capacity by the charge capacity in the 1st cycle.However,we will use the notation for comparing the characteristics of the layered complex cathode with the conventional insertion cathode.It will be helpful to understand the difference between two kinds of cathodes.Hysteresis in the potential versus capacity curves in the second cycle is clearly observed especially for the upper-limit voltages of 4.5and 4.6V.Results of he GITT measurement will be described later to confirm the hysteresis by excluding the contri-bution of the overpotential.2.53.03.54.04.55.02.53.03.54.04.55.02.53.03.54.04.55.02.53.03.54.04.55.0P o t e n t i a l (V v s . L i /L i +)Specific capacity (mAhg -1)Fig.3.Charge and discharge potential curves of LiNi 1/3Co 1/3Mn 1/3O 2as a conventional insertion cathode,for various upper limit voltages (CV voltage)of 4.3,4.4,4.5,and 4.6V.Cells were measured at current density of 0.2mA cmˆ−2.82M.Kasai et al./Electrochimica Acta 146(2014)79–88P o t e n t i a l (V v s . L i /L i +)Specific capacity (mAhg -1)Fig.4.Charge and discharge potential curves of 0.5Li 2MnO 3-0.5LiNi 1/3Co 1/3Mn 1/3O 2layered complex cathode,for various upper limit voltages (CV voltage)of 4.3,4.4,4.5,and 4.6V.Cells were measured at a current density of 0.1mAcm −2.The limit voltage dependency of the capacity and coulombic effi-ciency for the 1st and 2nd cycles is shown in Fig.5by black squares.For the 1st cycle,the charge capacity increased from 107.4mAhg −1(at 4.4V)to 249.2mAhg −1(at 4.5V),finally reaching 308.1mAhg −1(at 4.6V)as the upper-limit voltage increased [Fig.5].The increase of the charge capacity was accompanied by a rapid decrease of the coulombic efficiency from 90.6(at 4.3V)and 88.0%(at 4.4V)to 75.9(at 4.5V)and 71.1%(at 4.6V),as shown in Fig.5(c).For the 2nd cycle,the charge and discharge capacity were very close to each other,as shown in Fig.5(d)and (e),respectively.Therefore,the coulombic efficiency exceeded 90%for all the upper-limit voltages shown in Fig.5(f).The results for LiNi 1/3Co 1/3Mn 1/3O 2cathode are also shown in the Fig.5by white triangles.A remarkable change in the coulombic efficiency is not observed accordingFig.5.Charge-discharge capacities and coulombic efficiency of the 1st and 2nd cycles for various upper-limit voltages.Results of the 0.5Li 2MnO 3-0.5LiNi 1/3Co 1/3Mn 1/3cathode are shown by black squares,and results of the LiNi 1/3Co 1/3Mn 1/3cathode are shown by white triangles.The observed irreversible capacity includes the energy loss due to an activation reaction of the cathode for the upper-limit voltages of 4.5and 4.6V.The “coulombic efficiency”described in this work was obtained by simply dividing the 1st discharge capacity by the 1st charge capacity.The notation was used for comparing the characteristics of the layered complex cathode with the conventional insertion cathode.M.Kasai et al./Electrochimica Acta146(2014)79–8883Fig.6.Obtained d Q/d V profiles of LiNi1/3Co1/3Mn1/3O2examined at different upper-limit voltages for various upper-limit voltages of4.3,4.4,4.5,and4.6V.The curves was obtained by differentiating the2nd charge/discharge curves.to the upper-limit voltages,both for the1st and the2nd cycle respectively.The efficiencies are about88%for the1st cycle and 99%for the2nd cycle.The result shows that the rapid decrease of the coulombic efficiency in the layered complex cathode is not due to a side reaction like decomposition of the electrolyte but due to a certain kind of material conversion reaction as called“activation”, which occurs when the charge voltage crosses4.5V.Figure6shows d Q/d V plots of the2nd charge/discharge cycle for the LiNi1/3Co1/3Mn1/3O2cathode.The upper-limit voltages were set at4.3,4.4,4.5and4.6V,while the lower voltage wasfixed at 2.5V,as same as the layered complex cathode.An oxidation peak and a reduction peak were observed at3.77and3.73V respectively for the upper-limit voltage of4.3V,as shown in Fig.6(a).If the midpoint of them indicates the equilibrium potential,the value is 3.75V and the overpotentialÁfor the redox reaction is20mV.There are no significant change observed when the voltage is increased up to4.6V,although small increase of the overpotential is shown for the upper limit voltages of4.5and4.6V[Fig.6(c),(d)].The obtained results indicate the typical behavior of the d Q/d V charac-teristics in a conventional insertion cathode.Measurement results of d Q/d V curves for the layered complex cathode of0.5Li2MnO3-0.5LiNi1/3Co1/3Mn1/3O2are shown in Fig.7(a)and(b)for upper limit voltages of4.3and4.4V.An oxidation peak at4.02V and a reduction peak at3.95V are observed as indicated by arrows.The equilibrium potential and overpotential are estimated as3.985V and35mV respectively.The redox peak profiles have a character of the conventional insertion cathode,while the equilibrium potential is about250mV higher than the results of the LiNi1/3Co1/3Mn1/3O2 cathode.As shown in Fig.7(c),once the cell is charged up to4.5V the oxidation peak shifts to3.87V and the reduction peak also shifts to 3.77V,while a broad reduction peak is formed at around3.36V.For the upper-limit voltage of4.6V shown in Fig.7(d),the whole shape of the peak profile is unchanged from the result for the voltage of 4.5V.It seemed that a pair of redox peaks at3.84and3.71V still show a character of the conventional insertion cathode.The oxi-dation peak is70mV higher than that of LiNi1/3Co1/3Mn1/3O2,and the reduction peak appears at the nearly same potential of3.71V. On the other hand,there is also a broad reduction peak observed at3.26V with simultaneous generation of a small oxidation peak at4.42V,having a character of a layered complex cathode.Ohzuku also reported d Q/d V characteristics,showing the peaks with two kinds of characters,of0.5Li2MnO3-0.5LiMn0.5Ni0.5O2cathode for the upper-limit4.7V[4](cf.Fig.5in Ohzuku et al.).The oxida-tion peak with a character of a conventional insertion cathode appeared at3.93V,which was160mV higher than the oxidation peak of LiMn0.5Ni0.5O2shown in the samefigure.The difference in the oxidation potential is two times larger than our work.The cor-responding reduction peak appeared at the nearly same potential of3.74V as the above-mentioned conventional insertion cathode. This behavior is similar with our result,although the correspond-ing peak positions are different.It is difficult to ascribe each peak to an appropriate electrochemical reaction right now.However, it is possible to ascribe the redox peaks at3.84and3.71V to the redox reaction Ni2+/Ni3+/Ni4+due to insertion/desertion of lithium ions,because the reaction occurs at3.75V for LiNi1/3Co1/3Mn1/3O2 cathode.The oxidation peak at4.42V was not clearly observed in this work as reported by Croy[5],rather in similar broad shape with Ohzuku.Recent study on charge compensation mechanisms for the layered complex cathode reported that all the transition metal ions are oxidized except Mn ions to a tetravalent state(i.e. Ni4+and Co4+)after the charging process to4.5V[16].Therefore, it is considered that the oxidation peak is possible to be ascribed to the oxidation process of Ni and Co ions.However,it seemed that the peak is sensitive not only to the chemical composition but also synthesis condition,because Ohzuku and Croy examined the cathode material with the same chemical composition.On the contrary,the reduction peak at3.26V is common to a cathode mate-rial in this work and the cathodes in the previous works.For the 0.5Li2MnO3-0.5LiMn0.5Ni0.5O2cathode,the peak was observed at 3.28V in Ohzuku’s work and at3.25V in the Croy’s work(cf.Fig.5in Croy et al.).Furthermore,a cathode material with a different com-position of0.5Li2MnO3-0.5LiNi0.375Co0.15Mn0.375O2cathode also84M.Kasai et al./Electrochimica Acta 146(2014)79–88d Q /d Vd Q /d VCell Potential (V vs. Li/Li +)Cell Potential (V vs. Li/Li +)Fig.7.Obtained d Q /d V profiles of 0.5Li 2MnO 3-0.5LiNi 1/3Co 1/3Mn 1/3O 2examined at different upper-limit voltages for various upper-limit voltages of 4.3,4.4,4.5,and 4.6V.The curves was obtained by differentiating the 2nd charge/discharge curves.showed the reduction peak at 3.25V (we can read it off with the Fig.7(b)in Croy et al.).The reduction peak was assigned to Mn 4+/Mn 3+by Croy et al [18].The corresponding oxidation occurs at about 4.3V showing a potential difference about 1V,which is caused by a structural change accompanying the transition metal migration [5].We are considering such potential shift will show a certain extent of dependency on the transition metal composition,although it was not observed for the cathode materials discussed in this work.Fur-ther study is necessary for wider range of chemical composition of the layered complex cathode.In this section,we presented the electrochemical char-acteristics for the layered complex cathode of 0.5Li 2MnO 3-0.5LiNi 1/3Co 1/3Mn 1/3O 2.The obtained redox curves showed two kinds of peaks.One pair is the oxidation peak at 3.84V and the cor-responding reduction peak at 3.71V,which has a character of the conventional insertion cathode,and another pair is the oxidation peak at 4.42V and the reduction peak at 3.26V showing a char-acter of the layered complex cathode.It seemed that the former three peaks showed small dependency on the composition of tran-sition meal,however the latter reduction peak at 3.26V appeared at almost exactly same potential.We are considering that to investi-gate the behavior of the redox reaction for more wide range of metal composition is necessary to reach the clarification of the hysteresis.In general,hysteresis is accompanied by energy dissipation.Therefore,precise estimation of the dissipation is also important to clarify the hysteresis.GITT measurement was examined to obtain OCP versus characteristic,which gives the dissipation.Figure 8shows complete GITT curves for upper-limit voltages of 4.3,4.4,4.5,and 4.6V.The 20min galvanostatic pulse is contin-ued until the charge voltage reaches the set upper-limit voltage.When the cell becomes unable to maintain a 20min galvanostatic current,the charge process is terminated and the discharge pro-cess is started in the same way until the cell potential reaches the lower-limit voltage of 2.5V.In the present test system,it was not possible to stop the GITT test automatically when the galvanostatic pulse was shorter than 20min.Therefore,several pulses shorter than 20min were incorporated in the original raw data for exper-imental convenience.The galvanostatic pulses that did not satisfy the defined measurement condition of 20min duration were elim-inated from Fig.8and the other results reported here.The relaxation procedure starts as soon as the galvanostatic pulse charge or discharge ends.Relaxation curves,which were extracted from the GITT test results for the 4.6V upper-limit,are shown according to several capacities in Fig.9(a)for the charge pro-cess and in Fig.9(b)for the discharge process.The point where the 20min galvanostatic current was cut off is indicated by an arrow.After the pulse duration,the relaxation procedure continued for 10h to the OCP.There is a large difference in OCP between charge and discharge capacities that are close to each other.For example,the top curves in Figs.9(a)and (b)show capacities of 215.9mAhg −1for the charge process and 215.4mAhg −1for the discharge process,with corresponding OCPs of 4.441and 4.097V,respectively,which is a difference of approximately 340mV.This result indicates that the dependency of the OCP on the capacity has a large width of hys-teresis.The relaxation rate was evaluated according to the potential change between 9and 10h.All the obtained values were below 2mVh −1,which is stable enough to determine the OCP character-istics.The dependency of the OCP on capacity is shown in Fig.10for the various upper-limit voltages.For the upper-limit voltages of 4.3and 4.4V [Fig.10(a)],the difference in the OCP for the same capacity was not significantly large.For example,at a capacity of ca.50mAhg −1,the difference in the OCP was ca.60mV for 4.3V and 100mV for 4.4V.However,for the upper-limit voltages of 4.5andM.Kasai et al./Electrochimica Acta 146(2014)79–88852.02.53.03.54.04.52.02.53.03.54.04.52.02.53.03.54.04.502004006008001000120014002.02.53.03.54.04.5Upper Limit 4.6 VUpper Limit 4.5 VP o t e n t i a l v s . (L i /L i +)/ VUpper Limit 4.4 Vt /hUpper Limit 4.3 VFig.8.GITT curves measured for upper limit voltages of 4.3,4.4.,4.5and 4.6V.The GITT measurements were performed after the 1st charge and discharge cycle for initialization up to the corresponding upper limit voltages.4.6V [Fig.10(b)],the difference in the OCP at around 135mAhg −1was 430mV for 4.5V and 450mV for 4.6V.The dependencies of the overpotential Áwere also obtained by the GITT curves,as shown in Fig.11.For example,point A,shown in the upper panel of Fig.11(a),indicates the OCP (E 0)using the notation in the inset,and points B and C indicate E t and E 1,respectively.In the discharge pulse of the GITT test,the measurement started from point A and then pro-ceeded to point B after a discharge pulse for 20min.The relaxation process of 10h is indicated as line BC,i.e.,the potential difference of B and C corresponds to the overpotential Áat the capacity.At around the capacity of 135mAhg −1,the overpotentials were 20mV for the charge process and 100mV for the discharge process in the case of the upper-limit voltage of 4.6V.It is obvious that these results also support the claim that the OCP hysteresis exceeding 400mV can-not be explained by the large overpotential of the layered complex cathode.We also investigated OCP versus capacity characteristics for LiNi 1/3Co 1/3Mn 1/3O 2as a conventional insertion cathode.The GITT pulse length was 20min,the same as the layered complex cathode test,and the upper-limit voltages were 4.3,4.4,4.5and 4.6V.However,the GITT current was set at 0.16mA cm −2and the relaxation interval was 5h because the relaxation rate of LiNi 1/3Co 1/3Mn 1/3O 2was faster than that of the layered complexcathode.The measurement was performed after an initializa-tion procedure consisting of a charge/discharge cycle up to the set upper-limit voltages,respectively.The results are shown in Fig.12(a)for the upper-limit voltages of 4.3and 4.4V,and in Fig.12(b)for the upper-limit voltages of 4.5and 4.6V.It is clearly seen that there is no hysteretic behavior observed for 4.3V.The result corresponds to the case described in Fig.2(a),and the OCP is decided only by the lithium composition,which is equivalent to the capacity.On the other hand,small hysteresis,showing the hysteresis width of abut 50mV,was observed for the upper-limit voltages of 4.4,4.5and 4.6V.Such small hysteresis was also observed for the layered complex cathode in this work for the voltage of 4.3V.Structural studies of a cathode material for various charged states have already been reported.For example,Nam et al.reported results for LiNi 1/3Co 1/3Mn 1/3O 2by in situ X-ray measurements [14].They showed that there is a lattice volume change from 3to 5%for capacities exceeding about 160mAhg −1.It is possible that such lattice strain causes the change in the OCP.For example,Hirai’s results showed an OCP drop of 0.528V caused by an elastic volume strain of 22%accompanied by the discharge process from ˇ-Sn to LiSn 2/5[9].The volume change of the layered complex cathode during the charge/discharge process,which wasFig.9.Relaxation curves obtained from the GITT measurements for (a)charge and (b)discharge processes.Each galvanostatic pulse was maintained for 20min and the rest time was 10h.The results are shown for the upper limit voltage of 4.6V in accordance with the capacity.。