LiM1−yMyPO4 (M = Fe, Mn, Co, and Mg) Solid Solution Cathodes

Understanding the Shifts in the Redox Potentials of Olivine LiM1-yMyPO4 (M ) Fe, Mn, Co, and Mg) Solid Solution Cathodes T. Muraliganth and A. Manthiram* Electrochemical Energy Laboratory & Materials Science and Engineering Program, The UniVersity of Texas at Austin, Austin, Texas 78712 ReceiVed: June 15, 2010; ReVised Manuscript ReceiVed: August 8, 2010 A facile, high-energy mechanical milling (HEMM) approach has been developed to synthesize carbon-coated olivine LiM1-yMyPO4 (M ) Fe, Mn, Co, and Mg) solid solution nanoparticles. A systematic structural and electrochemical characterization of the solid solution series has been carried out by X-ray diffraction (XRD), scanning electron microscopy (SEM), charge-discharge measurements, and galvanostatic intermittent titration technique (GITT). The discharge capacity, voltage profile, and cycle performance of the LiM1-yMyPO4 solid solution cathodes are found to be dependent on the different redox couples involved in the reaction. Equilibrium potentials obtained from GITT and dQ/dV plot reveal a systematic shift in the redox potential of Fe2+/3+, Mn2+/3+, and Co2+/3+ couples in the LiM1-yMyPO4 solid solution compared to their pristine end members (LiMPO4). The shifts in the redox potential are explained on the basis of the changes in the covalency of the M-O bond and M-O-M interaction, and the consequent change in the position of the M2+/3+ redox energy. The self-discharge phenomenon of the Co2+/3+ couple in LiM1-yCoyPO4 has also been investigated by electrochemical impedance spectroscopy. Introduction Among the numerous energy storage systems investigated, lithium-ion batteries have been the dominant battery chemistry because of their high energy density, wide range of operating temperatures, and long cycle life. Recently, there has been considerable interest in developing olivine LiMPO4 (M ) Mn, Fe, Co, and Ni) as a cathode material for high-power, large- scale applications such as electric vehicles.1–13 Particularly, LiFePO4 has drawn much attention because of its low cost, nontoxicity, and high thermal stability. Also, the Fe2+/Fe3+ couple operates around 3.45 V with a theoretical capacity of 170 mAh/g, providing energy density comparable to those of layered LiCoO2 and spinel LiMn2O4 cathodes. However, the key issue with LiFePO4 is its one-dimensional lithium-ion diffusion and poor electronic conductivity (∼10-9 S/cm).1–4 To overcome these difficulties, many processing methods to decrease the particle size and realize conductive coatings have been pursued in recent years.5–13 Replacing the transition-metal ion Fe2+ by Mn2+, Co2+, and Ni2+ increases the redox potential significantly from 3.45 V in LiFePO4 to 4.1, 4.8, and 5.1 V, respectively, in LiMnPO4, LiCoPO4, and LiNiPO4 because of the changes in the positions of the various redox couples.1,14–16 LiMnPO4 is of particular interest because of the environmentally benign manganese and the favorable position of the Mn2+/3+ redox couple at 4.1 V vs Li/Li+, which is compatible with most of the electrolytes. However, it has been shown to offer very low capacity and poor rate capability due to the large lattice distortions induced by Jahn-Teller active Mn3+ ions and the much lower electronic conductivity (∼10-14 S/cm) of LiMnPO4 compared to LiFePO4 (∼10-9 S/cm).17,18 On the other hand, the flat voltage profile at 4.8 V of LiCoPO4 is attractive to increase the energy density. However, its full theoretical capacity (∼167 mAh/g) has not been realized in practical cells so far, and the cycle life of LiCoPO4 is also poor due to the instability of the commonly used LiPF6 in EC:DEC electrolyte at these high operating voltages.19,20 The redox potential of LiNiPO4 is even higher at 5.1 V, which makes it hard to test with the currently available electrolytes. Recently, the mixed transition-metal systems have attracted considerable interest.1,21–24 Among them, the LiFe1-yMnyPO4 solid solution system has drawn much attention, as it exhibits higher energy density and improved redox kinetics due to improved electronic conductivity compared to their end members. On the other hand, the LiMn1-yCoyPO4 and LiFe1-yCoyPO4 solid solutions are appealing due to their high operating voltage arising from the Co2+/3+ redox couple. The reaction behavior of olivine phosphates has been studied in detail recently, showcasing several interesting observations. Although the initial work revealed a two-phase reaction mech- anism with LiFePO4 and FePO4 as end members,1 subsequent investigations by Yamada et al.8 have indicated that the reaction involves two narrow monophase end members LiRFePO4 and Li1-�FePO4, where the values of R and � have been found to be dependent on temperature,25 particle size,26,27 and doping. Studies on the reaction behavior of the LiMn1-yFeyPO4 solid solution have also revealed several interesting and contrasting observations. Earlier, a phase diagram for LiMn1-yFeyPO4 was proposed in which the reaction behavior of the Fe2+/Fe3+ couple has been reported to change from two-phase behavior in LiFePO4 to single-phase behavior in LiMn1-yFeyPO4 (0.2 > y < 0.6), with the Mn3+/Mn2+ couple always showing a two-phase behavior.28,29 However, recent in situ synchrotron X-ray dif- fraction studies have contradicted the above results, revealing a two-phase reaction mechanism for both the Fe2+/Fe3+ and Mn2+/Mn3+ redox couples in LiMn0.6Fe0.4PO4 with a solid solution domain occurring in the intermediate region.30 Recently, Kobayashi et al.31 have reported a systematic shift in the redox potential of the Fe2+/3+ and Mn2+/3+ couples in the LiMn1-yFeyPO4 solid solution compared to their pristine end * To whom correspondence should be addressed. Phone: 512-471-1791. Fax: 512-471-7681. E-mail: rmanth@mail.utexas.edu. J. Phys. Chem. C XXXX, xxx, 000 A 10.1021/jp1055107  XXXX American Chemical Society members (LiFePO4 and LiMnPO4). However, measurement of the redox potentials and investigation of the reaction behavior are not available for other solid solutions such as LiMn1-y- CoyPO4, LiFe1-yCoyPO4, and LiFe1/3Mn1/3Co1/3PO4. Understand- ing the reaction behavior of such solid solutions can help to improve the performance of olivine cathodes and the search for new cathode materials. With an aim to develop a broader understanding of the factors that influence the redox potentials, we present here a systematic investigation of the structure, electrochemical performance, redox potential, reaction behavior, and self-discharge phenom- enon of carbon-coated LiM1-yMyPO4 (M ) Mn, Fe, Co, or Mg) solid solution cathodes (Figure 1) synthesized by an efficient high-energy mechanical milling (HEMM) method. The solid solution samples are characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), electrochemical charge- discharge measurements, galvanostatic intermittent titration technique (GITT), and electrochemical impedance spectroscopy (EIS). Experimental Section Carbon-coated LiFe1-yMnyPO4, LiFe1-yCoyPO4, LiMn1-yCoy- PO4, LiFe1-yMgyPO4, and LiMn1-yMgyPO4 with y ) 0, 0.25, 0.50, 0.75, and 1.0 as well as LiFe1/3Mn1/3Co1/3PO4 were prepared by a high-energy mechanical milling (HEMM) method (Figure 1), followed by heating as described below. Required amounts of lithium carbonate (Fisher), diammonium hydrogen phosphate (ACROS-Organics), and the divalent oxalate dihy- drates of M (Alfa Aesar, M ) Mn, Fe, Co, or Mg) were mixed with conductive super-p carbon to get a LiM1-yMyPO4:C weight ratio of 80:20 in the final product. The precursors were loaded into a hardened stainless steel vial, with a ball to weight ratio of 1:20, and sealed in an argon-filled glovebox. The steel vial was then subjected to high-energy ball milling for 10 h at a speed of 500 rpm in FRITSCH equipment. Crystalline Figure 1. Phase diagram of the LiM1-yMyPO4 (M ) Fe, Mn, or Co) solid solution system. The compositions studied are indicated by closed circles. Figure 2. (a) XRD patterns of the LiFe1-yMnyPO4 solid solution, with the expanded 2θ region on the right showing the continuous shift in the positions of the reflections to lower angles with the substitution of larger Mn2+ for Fe2+ in LiFe1-yMnyPO4 and (b) variations of the unit cell parameters of LiFe1-yMnyPO4 with y. B J. Phys. Chem. C, Vol. xxx, No. xx, XXXX Muraliganth and Manthiram LiM1-yMyPO4 samples were finally formed by heating the milled precursors at 550 °C for 6 h in an Ar atmosphere in a tube furnace. XRD characterizations of the samples were carried out with a Philips X’Pert diffractometer and filtered Cu KR radiation. SEM analyses were carried out with a JEOL-JSM5610 SEM instrument. Electrochemical performances were evaluated with CR2032 coin cells with an Arbin battery cycler. The coin cells were fabricated with the LiM1-yMyPO4 cathode, metallic lithium anode, 1 M LiPF6 in 1:1 diethyl carbonate/ethylene carbonate electrolyte, and Celgard polypropylene separator. The cathodes were prepared by mixing 75 wt % active material (carbon-coated LiMyM1-yPO4) with 20 wt % conductive carbon and 5 wt % polytetrafluoroethylene (PTFE) binder, rolling the mixture into thin sheets, and cutting them into circular electrodes of 0.64 cm2 area. The electrodes typically had an active material mass of ∼8 mg and were dried in a vacuum at 100 °C for more than 3 h before assembling the cells in an argon-filled glovebox. Results and Discussion Structural and Morphological Analysis. XRD patterns of the LiFe1-yMnyPO4, LiFe1-yCoyPO4, and LiMn1-yCoyPO4 solid solution systems with y ) 0, 0.25, 0.50, 0.75, and 1.0 are shown in Figures 2a, 3a, and 4a. As seen, the samples are single phase and all the reflections could be indexed on the basis of olivine structure with the space group pnma. The enlarged patterns shown on the right side of Figures 2a, 3a, and 4a reveal a continuous shift in the positions of the (301), (311), and (121) reflections, confirming the formation of solid solutions. Figures 2b, 3b, and 4b show the variations of lattice parameters and unit cell volume for the three solid solution systems.32 The systematic variations in lattice parameters and unit cell volume can be understood on the basis of the ionic radii of the various M2+ cations (Mn2+, 0.83 Å; Fe2+, 0.78 Å; Co2+, 0.75 Å).33 The linear variations in lattice parameters again confirm the forma- tion of homogeneous solid solutions. The XRD pattern of Li- Fe1/3Mn1/3Co1/3PO4 (Supporting Information Figure S1) also confirms the formation of single-phase solid solution with lattice parameters of a ) 10.326, b ) 6.009, and c ) 4.712 Å, which match closely with the literature values.34 An analysis of the XRD data by Scherrer’s formula revealed an average crystallite size of 30-50 nm for all the solid solutions. Also, the XRD patterns did not show any peaks corresponding to impurity phases like Fe2P or Co2P that are usually formed on heating at temperatures higher than 700 °C for longer periods of time in reducing gas atmospheres.3,35 Morphological characterization of the synthesized samples was carried out by ultrahigh-resolution field emission scanning electron microscopy (FE-SEM) and high-resolution transmission electron microscopy (HR-TEM). Figure 5a and b shows the low- and high-magnification FE-SEM images of the LiFe1/3Mn1/3- Co1/3PO4/C nanocomposite. The SEM image indicates the average particle size to be <50 nm with some large agglomer- ates. The TEM image in Figure 5c confirms the formation of LiFe1/3Mn1/3Co1/3PO4 nanocrystals of ∼50 nm diameter. The HR-TEM image in Figure 5d shows a uniform coating of carbon (light amorphous region) on the LiMn1/3Fe1/3Co1/3PO4 nanopar- ticles (dark crystalline fringe region). The HEMM approach used is able to produce nanometer size LiM1-yMyPO4 due to the low firing temperature used (550 °C) and the inhibition of the particle Figure 3. (a) XRD patterns of the LiFe1-yCoyPO4 solid solution, with the expanded 2θ region on the right showing the continuous shift in the positions of the reflections to higher angles with the substitution of smaller Co2+ for Fe2+ in LiFe1-yCoyPO4 and (b) variations of the unit cell parameters of LiFe1-yCoyPO4 with y. Figure 4. (a) XRD patterns of the LiMn1-yCoyPO4 solid solution, with the expanded 2θ region on the right showing the continuous shift in the positions of the reflections to higher angles with the substitution of smaller Co2+ for Mn2+ in LiMn1-yCoyPO4 and (b) variations of the unit cell parameters of LiMn1-yCoyPO4 with y. Olivine LiM1-yMyPO4 Solid Solution Cathodes J. Phys. Chem. C, Vol. xxx, No. xx, XXXX C growth by the carbon present during the synthesis process.9 The smaller particle size is beneficial to enhance the electrochemical performance due to the shortening of both the lithium-ion and electron diffusion lengths within the particle, while the uniform carbon network formed during the mechanical milling of conductive carbon with the precursors improves the electronic conductivity of the electrode. Energy dispersive spectroscopic (EDS) analysis (Supporting Information Figure S2) carried out in SEM on the LiFe1/3Mn1/3Co1/3PO4 nanoparticles indicates a uniform distribution of Mn, Fe, and Co in the sample. Electrochemical Characterization. Figure 6a shows the first discharge profiles of the LiFe1-yMnyPO4 solid solutions at C/20 rate. LiFePO4 delivers a high capacity of 162 mAh/g with a flat voltage of ∼3.45 V (close to its theoretical capacity of 170 mAhg-1), while LiMnPO4 delivers a low capacity of 91 mAh/g at ∼4.1 V. The discharge profiles of the LiFe1-yMnyPO4 solid solution system exhibit voltage plateaus corresponding to both the Fe2+/3+ (3.45 V) and Mn2+/3+ (4.1 V) redox couples. The discharge capacity increases significantly from 91 mAh/g in LiMnPO4 to 142 mAh/g in LiFe0.25Mn0.75PO4, and both LiFe0.50Mn0.50PO4 and LiFe0.75Mn0.25PO4 exhibit discharge ca- pacity values close to 155 mAh/g. The increase in electrochemi- cal activity on substituting Fe for Mn can be attributed to the improved electronic conductivity and lithium-ion conductivity compared to LiMnPO4. Figure 6b shows the cyclability of the LiFe1-yMnyPO4 solid solutions. Although LiMnPO4 exhibits a slight capacity fade, all the LiFe1-yMnyPO4 solid solutions exhibit stable cycle performances. Figure 7a shows the discharge profiles of the LiFe1-yCoyPO4 solid solutions. LiCoPO4 delivers a discharge capacity of only ∼93 mAh/g, which is much lower than its theoretical capacity of 167 mAh/g. The discharge capacity improves with iron doping, and LiFe1-yCoyPO4 delivers capacities of 160, 129, and 121 mAh/g, respectively, at y ) 0.25, 0.50, and 0.75. Figure 7b compares the cycle life of the LiFe1-yCoyPO4 solid solution cathodes. LiCoPO4 shows dramatic capacity fade during cycling, which is consistent with the previous reports,36 possibly due to the instability of the electrolyte at 4.8 V. However, both LiFe0.50Co0.50PO4 and LiFe0.75Co0.25PO4 exhibit excellent capac- ity retention because of the large contribution of the Fe2+/3+ redox couple to the capacity and the decrease in the operating voltage of the Co2+/3+ redox couple in the solid solution compared to that in pure LiCoPO4 (see the next section). Figure 8 shows the first discharge profile and cyclability of the Figure 5. (a) Low- and (b) high-magnification FE-SEM images of LiFe1/3Mn1/3Co1/3PO4/C and (c) TEM and (d) HRTEM images of LiFe1/3- Mn1/3Co1/3PO4/C. Figure 6. (a) Discharge profiles (recorded at C/20 rate) and (b) cycle life of the LiFe1-yMnyPO4 solid solution cathodes. D J. Phys. Chem. C, Vol. xxx, No. xx, XXXX Muraliganth and Manthiram LiMn1-yCoyPO4 solid solutions. The LiMn1-yCoyPO4 solid solution cathodes deliver low discharge capacities of around 120 mAh/g with poor capacity retention during cycling. The poor electrochemical performances of the LiMn1-yCoyPO4 solid solutions are in accordance with the poor electrochemical performances of their end members LiMnPO4 and LiCoPO4. Figure 9 shows the first discharge profile and cyclability of the LiFe1/3Mn1/3Co1/3PO4 solid solution cathode. LiFe1/3Mn1/3- Co1/3PO4 delivers a discharge capacity of 141 mAh/g with three plateaus corresponding to the Fe2+/3+, Mn2+/3+, and Co2+/3+ couples present in the sample. However, LiFe1/3Mn1/3Co1/3PO4 exhibits some capacity fade during cycling. A comparison of the theoretical and experimental energy densities (at C/20 rate) of the LiFe1-yMnyPO4, LiFe1-yMnyPO4, LiFe1-yMnyPO4, and LiFe1/3Mn1/3Co1/3PO4 solid solutions is given in Supporting Information Figure S3. The data show that, even though LiMnPO4, LiCoPO4, and their solid solutions have a higher theoretical energy density than LiFePO4 as a result of higher operating voltages, their experimental energy densities are lower or equal to that of LiFePO4 due to their lower capacity values and high polarization loss. Further improvements in high- voltage electrolytes could help to efficiently use the high-voltage Co2+/3+ redox couples and enhance the energy density. Equilibrium Potential Measurements. The true redox potentials of the electrodes cannot be obtained precisely from the voltages measured under a current load, as the measured voltages involve overpotentials due to various polarization losses. With an objective to fully understand the electrochemical reactions occurring during charge-discharge and obtain the equilibrium redox potentials of the LiM1-yMyPO4 solid solution cathodes at different stages of lithium deintercalation, galvano- static intermittent titrations (GITT) were carried out. The GITT measurements were carried out by applying a galvanic step (constant current at C/50 rate) corresponding to the removal of 0.05 Li (state of charge) and measuring the open-circuit voltage (OCV) of the cell for 48 h (relaxation time). Figure 10a shows the GITT measurements of the LiFe1-y- MnyPO4 solid solution series, illustrating the voltage dependence on lithium content under load and rest. Here, the nearly flat region indicates the voltage measured during charging (load) Figure 7. (a) Discharge profiles (recorded at C/20 rate) and (b) cycle life of the LiFe1-yCoyPO4 solid solution cathodes. Figure 8. (a) Discharge profiles (recorded at C/20 rate) and (b) cycle life of the LiMn1-yCoyPO4 solid solution cathodes. Figure 9. (a) Discharge profiles (recorded at C/20 rate) and (b) cycle life of the LiFe1/3Mn1/3Co1/3PO4 solid solution cathode. Olivine LiM1-yMyPO4 Solid Solution Cathodes J. Phys. Chem. C, Vol. xxx, No. xx, XXXX E and the relaxation spikes at a given state of charge or lithium content (1 - x) indicate the change in voltage during relaxation or equilibration. A short spike during relaxation characterizes a small polarization and fast equilibration, while a long spike indicates a large polarization and slow equilibration. The equilibrium OCVs for the Fe2+/3+ and Mn2+/3+ couples in LiFePO4 and LiMnPO4 reveal nearly flat potentials of, respec- tively, 3.43 and 4.13 V, indicating that the lithium extraction follows a two-phase reaction mechanism. LiMnPO4 exhibits higher overpotential (long spikes) than LiFePO4 due to its lower electronic conductivity. In addition, the sluggish kinetics of LiMnPO4 could be partly related to a larger volume change (8.9%) occurring between the two phases LiMnPO4 and MnPO4 compared to the 7% volume change occurring between the two phases LiFePO4 and FePO4.20 However, the OCVs for the LiFe1-yMnyPO4 solid solutions show a sloping region on going from the Fe2+/3+ to Mn2+/3+ redox couple, which may imply a single-phase region over a small lithium-content range. More interestingly, there is a notable and continuous upward shift in the OCV of the Fe2+/3+ couple with increasing substitution of Mn2+ for Fe2+ and a downshift in the OCV of the Mn2+/3 couple with increasing substitution of Fe2+ for Mn2+ in LiFe1-yMnyPO4, as seen in Figure 10b, which will be discussed further later. Figure 11a shows the GITT profiles of LiFe1-yCoyPO4. As seen, the Fe2+/3+ coup