Ec

LMa

b

Hc

d

e

ARRAA

KLCLCE

1

mi(sec[emie

mno

h

0h

Electrochimica Acta 108 (2013) 182– 190

Contents lists available at SciVerse ScienceDirect

Electrochimica Acta

jou rn al hom ep age: www.elsev ier .com/ locate /e lec tac ta

volution of electrochemical performance in Li3V2(PO4)3/Composites caused by cation incorporation�

u-Lu Zhanga,b,c, Gan Liangc, Gang Penga, Yan Jiangb, Hui Fangc, Yun-Hui Huangb,∗,ark C. Croftd,e, Alexander Ignatovd

College of Mechanical and Material Engineering, Three Gorges University, 8 Daxue Road, Yichang Hubei 443002, PR ChinaKey Laboratory for Advanced Battery Materials and System, Ministry of Education, School of Materials Science and Engineering,uazhong University of Science and Technology, 1037 Luoyu Road, Wuhan, Hubei 430074, PR ChinaDepartment of Physics, Sam Houston State University, Huntsville, TX 77341, USADepartment of Physics and Astronomy, Rutgers University, Piscataway, NJ 08854, USANSLS, Brookhaven National Laboratory, Upton, NY 11973, USA

a r t i c l e i n f o

rticle history:eceived 7 March 2013eceived in revised form 26 April 2013ccepted 19 June 2013vailable online xxx

a b s t r a c t

Li3V2(PO4)3/C (LVP/C) composites incorporated by a series of electrochemically active cations (Fe, Co,Ni, Mn) have been successfully prepared by a conventional solid-state reaction. M-incorporation (M = Fe,Co, Ni, Mn) in Li3V2(PO4)3 does not change the monoclinic structure. Analyzed with X-ray photoelec-tron spectroscopy, X-ray absorption spectroscopy and high-resolution transmission electron microscopy,we find that the valence is between +2.67 and +3 for Fe, and is +2 for Co, Ni and Mn. M-doped LVP

eywords:ithium ion batteryathode materialsithium vanadium phosphateation incorporationlectrochemical performance

and LiMPO4 phases coexist in the incorporated LVP/C composites. Compared with pristine LVP/C, Fe-incorporated LVP/C shows the best electrochemical performance with the highest initial dischargecapacity of 131.4 mAh g−1 at 0.1 C between 2.5 and 4.3 V. The Fe-incorporated LVP/C sample also exhibitsexcellent rate capability with an average capacity of 122.4 mAh g−1 at 1 C and 93.5 mAh g−1 at 5 C, resultingfrom the reduced particle size, the improved electronic conductivity, the high Li-ion diffusion coefficient,and the contribution of LiFePO4 to the capacity.

. Introduction

Monoclinic Li3V2(PO4)3 (LVP) has been proposed as one of theost promising cathode materials for lithium-ion batteries due to

ts large theoretical capacity (197 mAh g−1), high operating voltageup to 4.0 V), and fast lithium-ion migration [1–7]. However,imilar to other lithium transition metal phosphates, poor intrinsiclectronic transport of LVP limits its application [1,2]. Carbonoating [1,6–19], oxide modification [5,20], and cation doping3,21–33] have been proved to be effective to enhance the intrinsiclectronic conductivity. Compared to other modification methods,

etallic ion doping has been paid more attention. Hitherto, there

s still some controversy about the mechanism and the dopingffect. Actually, cation doping is an efficient way to improve the

� This is an open-access article distributed under the terms of the Creative Com-ons Attribution-NonCommercial-No Derivative Works License, which permits

on-commercial use, distribution, and reproduction in any medium, provided theriginal author and source are credited.∗ Corresponding author. Tel.: +86 27 87558237; fax: +86 27 87558241.

E-mail addresses: luluzhang924@gmail.com (L.-L. Zhang),uangyh@mail.hust.edu.cn, yunhuihuang@yahoo.com (Y.-H. Huang).

013-4686/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.electacta.2013.06.071

© 2013 Elsevier Ltd. All rights reserved.

intrinsic electronic conductivity and chemical diffusion coefficientof lithium ions within the crystals. Due to the electrochemicalactivity, Fe, Co, Ni and Mn are commonly used as doping elementsin Li3V2(PO4)3, LiFePO4 and other cathode materials [21,33–41].For Li3V2(PO4)3, Ren et al. [21] synthesized Li3FexV2−x(PO4)3(0 ≤ x ≤ 0.06) by solid-state reaction, and found that LVP withoptimal Fe-doping content (x = 0.02 − 0.04) could achieve highdischarge capacity and good cyclic stability. They also reportedthat the oxidation state of Fe in Li3Fe0.02V1.98(PO4)3 was +3, whichis somewhat different from our previous work [42]. Kuang et al.[24] prepared Co-doped Li3V2−xCox(PO4)3/C (0 ≤ x ≤ 0.15) viasolid-state reaction, and found that V4+ and V3+ ions coexistedin the Co2+-doped LVP samples; Li3V1.85Co0.15(PO4)3/C exhibitedthe best electrochemical performance due to enhanced structuralstability and alleviated volume change (expansion/contraction).Bini et al. [30] investigated the Mn substitution in LVP on aseries of samples obtained by different synthesis methods, andfound that the sol-gel-derived samples showed slightly enhancedcapacity due to the Mn substitution within a solubility limit

(x = 0.124), whereas the solid-state synthesized samples showedsignificantly enhanced capacity due to the microstructure of thecrystallites and the phase formation of LiMnPO4. Zhang et al. [33]prepared a series of Ni-doped LVP samples by microwave assisted

imica Acta 108 (2013) 182– 190 183

siNbr

eotaafsapMtpXr(a

2

2

pmMecintSwt(Lsp

2

dPwH(cim(itCtL(ewm

L.-L. Zhang et al. / Electroch

ol–gel method, and found that Ni doping level had an importantmpact on lithium storage capacity. The LVP with 1 wt.% dopedi exhibited a high discharge capacity and good rate capabilityecause of large exchange current density and low charge transferesistance.

It is generally known that the physicochemical property andlectrochemical performance of LVP significantly depend not onlyn doping element as well as doping level, but also on syn-hesis method. In the previous reports, there is no comparisonmong the doping effects of some important electrochemicallyctive cations (Fe, Co, Ni, and Mn) on the electrochemical per-ormance of LVP. Moreover, except for the Rietveld refinement,o far there is no other evidence to indicate that Fe, Co, Ni,nd Mn ions can be doped into the LVP lattice structure. In theresent work, we synthesized 5 at.% M-doped (M = Fe, Co, Ni, andn) Li3V2(PO4)3/C composites via a conventional solid-state reac-

ion and investigated their doping effects on the physicochemicalroperties and electrochemical performance of Li3V2(PO4)3 by-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), X-ay absorption spectroscopy (XAS), scanning electron microscopySEM), high-resolution transmission electron microscopy (HRTEM),nd electrochemical measurements.

. Experimental

.1. Synthesis of Li3V1.9M0.1(PO4)3/C (M = Fe, Co, Ni, Mn)

To prepare Li3V1.9M0.1(PO4)3/C (M = Fe, Co, Ni, and Mn) com-osites, stoichiometric Li2CO3, NH4VO3 and NH4H2PO4 wereixed with FeC2O4, Co(CH3COO)2·4H2O, Ni(CH3COO)2·4H2O andn(CH3COO)2·4H2O, respectively, and then ball-milled for 10 h in

thanol. The mixture was dried in air at 60 ◦C for 12 h, followed byalcining at 350 ◦C for 6 h in nitrogen atmosphere and then cool-ng down to room temperature. Glucose (15 wt.%), which acts asot only carbon source but also reductive agent, was added intohe resulting precursor and further ball-milled for 6 h in ethanol.ubsequently, the mixture was dried and sintered at 700 ◦C for 10 hith a heating rate of 3 ◦C min−1 in nitrogen atmosphere to achieve

he final products. Here, the M-incorporated Li3V1.9M0.1(PO4)3/CM = Fe, Co, Ni, and Mn) composites are denoted as LVFeP/C,VCoP/C, LVNiP/C, and LVMnP/C, respectively. For compari-on, pristine LVP/C composite was also prepared via the samerocess.

.2. Sample characterization

X-ray diffraction patterns were obtained using a X’Pert Proiffractometer with Cu-k� radiation (� = 1.5406 A) (XRD, X’Pertro, PANalytical B.V.). Morphology and structure of the samplesere observed with scanning electron microscope (SEM, Sirion 200,olland) and high-resolution transmission electron microscope

HRTEM, JEM-2100, JEOL). Carbon content was determined byarbon-sulfur analyzer (CS600, LECO, US), and electrical conductiv-ty was measured with standard four-probe method on resistivity

easurement system (RTS-8, China). Oxidation state of V and MM = Fe, Co, Ni, Mn) in pristine LVP/C and LVP/C-Fe samples wasnvestigated by X-ray photoelectron spectroscopy (XPS, PHI Quan-eraII, Japan) and X-ray absorption spectroscopy (XAS). The V, Fe,o, Ni, and Mn K-edge XAS data were taken in fluorescence andransmission mode at beam line X–19A of the National Synchrotronight Source (NSLS) at Brookhaven National Laboratory. A standard

Sr3V2O8, FeO, CoO, NiO, and MnO for V, Fe, Co, Ni, and Mn K-dge, respectively) was run in transmission mode simultaneouslyith the samples for energy calibration. A double-crystal Si(1 1 1)onochromator was used for energy selection and was detuned

Fig. 1. XRD patterns of pristine LVP/C and LVMP/C (M = Fe, Co, Ni, Mn) samples.

by reducing the incident photon flux 20% from its maximum valueto suppress contamination from harmonics. The energy resolution(�E/E) of the X–19A beam line was 2 × 10−4, corresponding toabout 1.1 eV at V K–edge, 1.4 eV at Fe K–edge, 1.5 eV at Co K–edge,1.7 eV at Ni K–edge, and 1.3 eV at Mn K–edge. The XAS spectrapresented in this paper were background subtracted and normal-ized to unity in the continuum region about 100 eV above theedge.

2.3. Electrochemical measurements

The working electrodes were prepared by roll milling a mix-ture of 75 wt.% active materials, 20 wt.% carbon black and 5 wt.%PTFE with N-methyl pyrrolidinone. Electrodes were punched intodiscs (�8 mm) and transferred into argon-filled glove box (Super1220/750, Mikrouna) shortly after drying at 80 ◦C for 24 h in vac-uum. 2032 coin cells were assembled with Celgard 2300 as theseparator, lithium foil as counter and reference electrodes, and1 mol L−1 LiPF6 in EC: DMC (LB-3711, China) as electrolyte. Toavoid the influence of electrolyte decomposition under high volt-age on the electrochemical behavior of LVP, a low charge voltageof 4.3 V was chosen here. The specific capacities of samples weremeasured by constant current charge/discharge testing between2.5 and 4.3 V (vs. Li+/Li) (LAND CT2001A, China). Cyclic voltam-metry (CV) measurements were performed on an electrochemicalworkstation (PARSTAT 2273, Princeton Applied Research, US) atprogressive scanning rates from 0.05 to 0.3 mV s−1 within a voltagerange of 3.3 − 4.3 V.

3. Results and discussion

Fig. 1 shows the XRD patterns of pristine LVP/C and LVMP/C(M = Fe, Co, Ni, Mn) samples. For pristine LVP/C and the M = Fe,Co, and Ni samples, the XRD patterns can be indexed as mono-clinic structure of LVP with a space group of P21/n and no anyimpurity phases were detected. For the M = Mn sample, some weaklines of LiMnPO4 impurity phase are observed in the XRD pattern.These results indicate that a small incorporation level (i.e., about5 at.% doping level) of alien electrochemically-active metal ionsdoes not change the monoclinic structure of LVP. The enlarged pat-terns shown on the right side of Fig. 1 reveal that all patterns forthe LVMP/C (M = Fe, Co, Ni, Mn) samples display a slight shift ofthe (−1 2 1), (0 1 2), (2 0 1) and (−2 1 1) reflection lines to higher 2�

degree, suggesting that the lattice parameters of the LVP lattice arereduced by the doping of the M atoms into the lattice structure[43–45]. To confirm this conclusion, a full Rietveld refinement wascarried out on all these five samples and the results are listed in

184 L.-L. Zhang et al. / Electrochimica

Table 1The lattice parameters of pristine LVP/C and LVMP/C (Fe, Co, Ni, Mn) samples.

Sample Lattice constant (Å) Latticevolume (Å3)

Rwp (%)

a b c

LVP/C 8.588(8) 12.010(9) 8.607(5) 887.72 8.20LVFeP/C 8.546(9) 12.039(8) 8.559(6) 880.61 7.23LVCoP/C 8.561(5) 12.017(9) 8.558(9) 880.47 7.31

TgaptmmMbrplmcf

MVLdoa∼VVNttoitMcdobi

titfSvhdeaoea5fl

LVNiP/C 8.540(7) 11.971(7) 8.576(6) 876.77 5.18LVMnP/C 8.525(9) 11.968(9) 8.569(6) 874.27 8.39

able 1. The best refinement model was chosen from a P21/n spaceroup. It can be clearly seen that the unit cell volume decreasesfter incorporation of M (M = Fe, Co, Ni, Mn), indicating at leastart of these transition metal ions are doped into the lattice struc-ure of LVP. The excessive part beyond solubility limit may form

inor impurity phases that are difficult to be detected by XRDeasurement. It is noticed that the cell volume change with the

incorporation cannot be fully explained by the size differenceetween the V ion and doped M ions due to the lack of informationegarding vacancies in the lattice and whether the doped M ionsrimarily occupy the M1 or M2 sites in the lattice structure. The

ack of the diffraction lines from the XRD patterns shown in Fig. 1ay indicate that the carbon concentration from pyrolysis of glu-

ose was too low to detect or the produced carbon is in amorphousorm.

To investigate the oxidation state of V and M (M = Fe, Co, Ni,n) and to find whether M atoms are doped into the LVP lattice,

2p and M 2p XPS data were taken for the pristine LVP/C andVMP/C samples at both the surface and in the interior (at ∼60 nmepth). The XPS spectra are shown in Fig. 2. The binding energy (BE)btained from the XPS data analysis was calibrated using 284.5 eVs the BE of C 1 s line. Figs. 2a2–e2 show a V 2p3/2 main peak at517 eV, which is the value reported by Ren et al. [21] for the BE of3+ in LVP. Thus, our XPS results indicate that the oxidation state of in all the five samples is close to +3, and the incorporation of Fe, Co,i, and Mn in LVP does not change the valence state of the V ions in

he M-doped LVMP/C samples. In addition, as shown in Fig. 2b3–e3,he Fe 2p3/2, Co 2p3/2, Ni 2p3/2, and Mn 2p3/2 main peaks are clearlybserved not only at the surface but also in the interior (∼60 nmn depth) of the LVMP/C particles. This result clearly demonstrateshat M atoms are doped into the bulk of the LVP particles. Since the

2p3/2 XPS spectra shown in Fig. 2 are quite noisy due to the lowoncentration level of the doped M atoms, it is difficult to draw aefinite conclusion regarding the valence state (or oxidation state)f the M ions in the M-doped LVP/C samples and thus XAS wille used below for determining the valence state of these doped M

ons.In Fig. 3a, the V K–edge XAS spectra of the pristine LVP/C and

he M-doped LVMP/C (M = Fe, Co, Ni, and Mn) samples are shownn comparison with the spectra of some reference compounds. Forhe reference compounds, the formal valence values for V are: +3or LaVO3 and Eu2VO4; +4 for La2NiVO6 and Sr3V2O7; and +5 forr3V2O8. It is well known that the primary indicator of the valenceariation is the chemical shift in the main portion of the edge toigher energy with increasing valence. The edge energy here isefined as the energy at inflection point of the rising part of thedge. It can be seen from Fig. 3a that the edge energy increases frombout 5480 eV to 5482 eV and then to 5484 eV with the increasef valence of V from +3 to +4 and then to +5. The values of thedge energy for the pristine LVP/C and LVMP/C with M = Fe, Co, Ni,

nd Mn samples are 5480.2 eV, 5480.3 eV, 5480.3 eV, 5479.8 eV, and479.8 eV, respectively. These values are almost the same as thoseor the trivalent V reference compounds (see the largest rectangu-ar box in the Fig. 3a). This result indicates that the valence of V in

Acta 108 (2013) 182– 190

LVP/C and LVMP/C samples should be very close to +3 and it showsthat the M-incorporation into LVP/C does not change the valencestate of the V ions.

To find the valence values of the doped M (M = Fe, Co, Ni, andMn) in LVMP/C, we show in Fig. 3b–e the M K-edge spectra forthe LVMP/C samples and some corresponding reference samples.In Fig. 3b, the Fe K–edge spectra for the LVFeP/C sample are shownalong with the spectra of five reference compounds. The formalvalence values for Fe are: +2 for FeO and LiFePO4; +2.67 for Fe3O4;and +3 for �-Fe2O3 and FePO4·H2O. Fig. 3b shows that the edgeenergies are about 7120 eV for the two Fe2+ compounds, but areshifted about 6 eV to the higher energy side for the Fe3+ compounds�-Fe2O3 and Fe3+PO4·H2O. The edge energy for the LVFeP/C sampleis between that for Fe3O4 and the two Fe3+ compounds �-Fe2O3and Fe3+PO4·H2O, indicating that the valence of Fe in this LVFeP/Csample is between +2.67 and +3. This result means that there areboth Fe2+ and Fe3+ in the LVFeP/C sample. This result is in consistentwith our previous study [42] on the Fe-doped LVP cathode materi-als for which Fe in the interior of the sample particles is found toexist mainly in the lower valence state (i.e., Li1+xV2−yFe2+

y(PO4)3)while Fe at the surface exists largely in the higher valence state (i.e.,FePO4).

In Fig. 3c, the Co K–edge spectra for the LVCoP/C sample andsome Co reference compounds: Co2+O, La2Co2+O4, Sr2Co2+MoO6,and LaCo3+O3. It can be seen from Fig. 3c that with increasing theCo formal valence from +2 to +3, there is a big chemical shift tohigher energy from the edge energies of the three divalent Co ref-erence compounds to that of the trivalent Co compound LaCo3+O3.The main edge portion of the spectrum for the LVCoP/C sample issituated between that for the divalent Co standard Co2+O and theother two divalent compounds La2Co2+O4 and Sr2Co2+MoO6, indi-cating clearly the Co valence in the LVCoP/C sample is +2, which isin consistent with the result reported by Kuang et al. [24]. Sim-ilar to the Co K-edge analysis, the main edge portion of the NiK-edge spectrum for LVNiP/C displayed in Fig. 3d is positionedbetween the spectra for the divalent Ni reference compounds Ni2+Oand Sr2Ni2+MO6 but much lower in energy than that for the triva-lent Ni compound LaNi3+O3, indicating that the valence of Ni inthe LVNiP/C sample is +2. In Fig. 3e, it can be seen that the edgeenergy of the Mn K-edge for LVMnP/C sample is very close to thatfor the Mn standard MnO but far below the edge energies of the Mnreference compounds Mn2

3+O3, Mn4+O2, and KMn7+O4, which hashigher Mn formal valence +3, +4, and +7, respectively. This resultdemonstrates that the valence of Mn in LVMNP/C sample is also+2.

Fig. 4 shows the SEM images of pristine LVP/C and LVMP/C(M = Fe, Co, Ni, Mn) powders. All the five samples exhibit irregulargranular shapes, but M incorporation slightly affects particle sizeand morphology. Especially, the LVFeP/C particles present lessagglomeration with a relatively narrow size distribution rangingfrom ∼100 to 500 nm (Fig. 4b), which is advantageous to shortenthe diffusion distance for lithium-ion transport during the interca-lation/deintercalation reaction.

The structure and morphology of LVMP/C (M = Fe, Co, Ni, Mn)samples were further studied by HRTEM. It is clearly seen in Fig. 5that amorphous carbon is either connecting the LVMP particles orcoated on the surface of the LVMP particles, which is favorablefor improving conductivity and alleviating vanadium dissolution inthe electrolyte [46]. In addition, the HRTEM and the correspondingfast Fourier transform (FFT) patterns confirm the formation of crys-talline LiMPO4 associated with the corresponding LVMP/C samples,which agrees well with the valence of M (+2) obtained from XPS

and XAS analysis. However, Fe3+ should also exist in the LVFeP/Csample though no Fe3+-containing composites are detected. It isspeculated that the content level of any FePO4 is too low or isamorphous [42].

L.-L. Zhang et al. / Electrochimica Acta 108 (2013) 182– 190 185

Fig. 2. XPS spectra of (a) pristine LVP/C, (b) LVFeP/C, (c) LVCoP/C, (d) LVNiP/C, and (e) LVMnP/C.

186 L.-L. Zhang et al. / Electrochimica Acta 108 (2013) 182– 190

Fig. 3. XAS K-edge spectra for pristine LVP/C, LVMP/C (M = Fe, Co, Ni, and Mn), and some reference compounds: (a) V K-edge, (b) Fe K-edge, (c) Co K-edge, (d) Ni K-edge, and(e) Mn K-edge.

mMtatLriaficcc

Fig. 6 shows the initial charge/discharge profiles, cycle perfor-ance and rate capability of pristine and LVMP/C (M = Fe, Co, Ni,n) cathode samples. In the first cycle, as shown in Fig. 6a, there are

hree charge/discharge plateaus around 3.60/3.57 V, 3.69/3.66 Vnd 4.09/4.06 V for all the five electrode samples, correspondingo a sequence of phase transitions: Li3V2(PO4)3 ↔ Li2.5V2(PO4)3,i2.5V2(PO4)3 ↔ Li2V2(PO4)3, and Li2V2(PO4)3 ↔ LiV2(PO4)3,espectively, associated with V3+/V4+ redox couple [4]. Surpris-ngly, for LVFeP/C electrode, another charge/discharge plateauround 3.49/3.40 V appears in the initial charge/discharge pro-

les (Fig. 6a, marked in orange circles), which is exactly theharacteristics of the electrochemical reaction of LiFePO4, asso-iated with Fe2+/Fe3+ redox couple. However, no other extraharge/discharge plateaus related to Co, Ni and Mn-containg

composites are detected, which may attribute to the low chargevoltage (4.3 V). It is also found in Fig. 6 that the electrochemicallyactive metal ions (Fe, Co, Ni, Mn) do not have the same effect onthe electrochemical performance of LVP. LVFeP/C, LVCoP/C, andLVMnP/C electrodes show better electrochemical performancethan LVNiP/C electrode. Particularly, the LVFeP/C delivers thehighest initial discharge capacity of 131.4 mAh g−1 at 0.1 C, whichis close to the theoretical capacity (132 mAh g−1). In addition,LVFeP/C exhibits a significantly enhanced rate capability with anaverage capacity of 122.4 mAh g−1 at 1 C and 93.5 mAh g−1 at 5 C

(Fig. 6c). Whereas, LVNiP/C delivers the lowest initial dischargecapacity of 117.1 mAh g−1 at 0.1 C (Fig. 6a and b) and also showsthe decreased rate capability with 102.3 mAh g−1 at 1 C and35.4 mAh g−1 at 5 C (Fig. 6c).

L.-L. Zhang et al. / Electrochimica Acta 108 (2013) 182– 190 187

Fig. 4. SEM images of the as-prepared samples: (a) pristine LVP/C, (b) LVFeP/C, (c) LVCoP/C, (d) LVNiP/C, and (e) LVMnP/C.

Fig. 5. HRTEM images of samples: (a) LVFeP/C, (b) LVCoP/C, (c) LVNiP/C, and (d) LVMnP/C.

188 L.-L. Zhang et al. / Electrochimica Acta 108 (2013) 182– 190

Fig. 6. (a, b) Initial charge/discharge profiles and cyclic performances of pristineLVP/C and M-modified LVP/C electrodes at 0.1 C, respectively; and (c) Cyclic per-fr

Mtstftoahi

Table 2Carbon content, electronic conductivity and lithium ion coefficient of pristine LVP/Cand LVMP/C (Fe, Co, Ni, Mn) samples.

Sample Carbon content(wt.%)

Electronicconductivity(S cm−1)

Lithium-ioncoefficient(cm2 s−1)

LVP/C 3.66 3.17 × 10−2 2.63 × 10−9

LVFeP/C 3.00 1.75 × 10−2 8.59 × 10−9

LVCoP/C 2.94 6.38 × 10−3 4.48 × 10−9

LVNiP/C 3.01 4.52 × 10−3 1.79 × 10−9

LVMnP/C 3.13 1.49 × 10−2 4.96 × 10−9

ormances of pristine LVP/C and M-modified LVP/C electrodes at different C-rate,anging from 0.5 to 5.0 C for ten cycles at each rate.

To investigate the effect of the doped M atoms (M = Fe, Co, Ni,n) on the electrochemical performance of LVP/C, carbon con-

ent, electronic conductivity and CV profiles for all as-obtainedamples were measured. Carbon content and electronic conduc-ivity of all samples are listed in Table 2. The carbon contentsor the four M-incorporated LVP/C samples are all lower thanhat for pristine LVP/C. This may result from the consumption

f pyrolytic carbon to maintain the reductive state of Fe, Cond Ni during the synthesis process. Pristine LVP/C shows theighest electronic conductivity value of 3.17 × 10−2 S cm−1 due to

ts highest carbon content. The electronic conductivity is in the

following order from higher to lower values for the LVMP/C sam-ples: LVFeP/C > LVMnP/C > LVCoP/C > LVNiP/C. The LVNiP/C showsthe lowest electronic conductivity with 4.52 × 10−3 S cm−1, whichcould be mainly responsible for the observed poorest electro-chemical performance. However, in spite of lower electronicconductivity caused by lower-level carbon, the electrochemicalperformance of LVFeP/C is actually better than that of pristine LVP/C(see Fig. 6).

Fig. 7 displays the CV curves for the pristine and M-dopedLVP/C samples. All the electrodes present three pairs of oxida-tion and reduction peaks, which are labeled as A1/C1, A2/C2, andA3/C3, corresponding to a sequence of phase transition processesfor LixV2(PO4)3 at x = 3.0, 2.5, 2.0 and 1.0 [4]. This CV result is in goodagreement with the positions of the charge/discharge plateausobserved in Fig. 6a. For LVFeP/C, another pair of redox potentialpeaks around 3.5/3.4 V are also seen in Fig. 7b (marked by redarrows). These peaks are associated with Fe2+/Fe3+ redox couplefor LiFePO4 and are consistent with the observed charge/dischargeplateaus in Fig. 6a. However, such extra redox peaks related to Co, Niand Mn are not observed and this may due to the low charge volt-age (4.3 V). Thus, combined with our results from XRD, TEM andcharge/discharge profiles, it is reasonable to believe that LiMPO4(M = Fe, Co, Ni and Mn) coexists as an impurity phase with theprimary M-doped LVP phase in LVMP/C samples though it is notdetected in the XRD patterns (see Fig. 1). For LVFeP/C, the theo-retical capacity of LiFePO4 (170 mAh g−1) impurity is higher thanthat of LVP (132 mAh g−1) when charged to 4.3 V. So, the existenceof LiFePO4 in the LVFeP/C sample can make additional contribu-tion to the total capacity, whereas LiCoPO4, LiNiPO4 and LiMnPO4are electrochemically inactive between 2.5 and 4.3 V. Furthermore,the CV peak current (Ip) increases with scanning rate, indicativeof a diffusion control. The relationship between the peak currentand square root of the scan rate can be expressed by the classicalRandles Sevcik equation [47]:

Ip = 2.69 × 105 n3/2AD1/2v1/2C, where Ip is the CV peak current(A), n is the number of electrons involved in the redox process, A isthe electrode area (cm2), D is the Li+ diffusion coefficient (cm2 s−1),v is the potential scan rate (V s−1), and C is the concentration oflithium ions in the cathode (mol cm−3). Therefore, the lower CVpeak current, the lower the Li ion diffusion coefficient is. In theinsets of Figs. 7a–d, the CV peak current Ip for the peak A3 is plot-ted against the potential scan rate v. Using the equation above andthe values of the slope of Ip vs. v1/2 plots shown in the inserts ofFig. 7a–e, the values of the lithium ion diffusion coefficient D arecalculated to be 2.63 × 10−9, 8.59 × 10−9, 4.48 × 10−9, 1.79 × 10−9,and 4.96 × 10−9 cm2 s−1, respectively (Table 2), for LVP/C, LVFeP/C,LVCoP/C, LVNiP/C, and LVMnP/C samples. Obviously, comparedwith LVP/C, LVFe/C has the fastest lithium ion diffusion and LVNiP/Chas the slowest lithium ion diffusion. This result agrees well with

the conclusion made in Fig. 6 that LVFeP/C and LVNiP/C show thebest and worst electrochemical performance, respectively, amongthe four LMVP/C cathode samples.

L.-L. Zhang et al. / Electrochimica Acta 108 (2013) 182– 190 189

LVP/C

4

aicsceLfdicw

Fig. 7. CV curves for the as-obtained samples: (a) pristine

. Conclusions

Li3V2(PO4)3/C was incorporated by four electrochemicallyctive metal atoms (Fe, Co, Ni, Mn) by a two-step solid-state sinter-ng method. The experiment shows that M-incorporation does nothange the monoclinic structure of Li3V2(PO4)3, but forms someolid solutions. Minor LiMPO4 (M = Fe, Co, Ni, Mn) impurity phaseould be formed in the LVMP/C samples. Moreover, FePO4 alsoxists as impurity in the LVFeP/C sample. Compared with pristineVP/C, LVNiP/C electrode exhibits the lowest capacity, resultingrom the decreased electronic conductivity and the lowest Li-ion

iffusion coefficient, whereas LVFeP/C shows the best electrochem-

cal performance which is indicated by the highest initial dischargeapacity of 131.4 mAh g−1 at 0.1 C and an excellent rate capabilityith average capacity of 122.4 mAh g−1 at 1 C and 93.5 mAh g−1 at

, (b) LVFeP/C, (c) LVCoP/C, (d) LVNiP/C, and (e) LVMnP/C.

5 C. The enhancement in electrochemical performance of LVFeP/Cis attributed to the reduced particle size, the improved electronicconductivity, the highest Li-ion diffusion coefficient, and the con-tribution of LiFePO4 to the capacity. Our results have specified theinfluence of electrochemically active metal ions (Fe, Co, Ni, Mn) onthe electrochemical performance of Li3V2(PO4)3, which is helpfulto understand the modified mechanism of some cathode materialsfor lithium-ion batteries.

Acknowledgements

This work was supported by the NSFC (No. 21175050,51272128) and the MOST of China (Nos. 2011AA11290,2011DFB70020); the key project of Hubei Provincial Departmentof Education (No. D20131303); US National Science Foundation

1 imica

(SUToNL

R

[[

[

[

[

[

[

[[

[

[

[

[[[

[

[

[[

[

[[[

[

[[

[[[

[

[

[[

[

[

[

90 L.-L. Zhang et al. / Electroch

No. CHE-0718482), an award from Research Corporation forcience Advancement, and an FRG grant from Sam Houston Stateniversity. In addition, the authors thank the Analytical andesting Center of Huazhong University of Science and Technol-gy for providing XRD and SEM measurements, and thank theational Synchrotron Light Source (NSLS) at Brookhaven Nationalaboratoty for providing facilities for XAS measurement.

eferences

[1] H. Huang, S.C. Yin, T. Kerr, N. Taylor, L.F. Nazar, Advanced Materials 14 (2002)1525–1528.

[2] S.C. Yin, H. Grondey, P. Strobel, M. Anne, L.F. Nazar, Journal of the AmericanChemical Society 125 (2003) 10402.

[3] W. Yuan, J. Yan, Z.Y. Tang, O. Sha, J.M. Wang, W.F. Mao, L. Ma, ElectrochimicaActa 72 (2012) 138.

[4] M.Y. Saïdi, J. Barker, H. Huang, J.L. Swoyer, G. Adamson, Journal of Power Sources119–121 (2003) 266.

[5] L.L. Zhang, X. Zhang, Y.M. Sun, W. Luo, X.L. Hu, X.J. Wu, Y.H. Huang, Journal ofthe Electrochemical Society 158 (2011) A924.

[6] M.M. Ren, Z. Zhou, X.P. Gao, W.X. Peng, J.P. Wei, Journal of Physical ChemistryC 112 (2008) 5689.

[7] Y.Q. Qiao, J.P. Tu, X.L. Wang, C.D. Gu, Journal of Power Sources 199 (2012) 287.[8] P. Fu, Y. Zhao, Y. Dong, X. An, G. Shen, Journal of Power Sources 162 (2006) 651.[9] X.H. Rui, Y. Jin, X.Y. Feng, L.C. Zhang, C.H. Chen, Journal of Power Sources 196

(2011) 2109.10] Y. Li, Z. Zhou, M. Ren, X. Gao, J. Yan, Electrochimica Acta 51 (2006) 6498.11] C. Chang, J. Xiang, X. Shi, X. Han, L. Yuan, J. Sun, Electrochimica Acta 54 (2008)

623.12] Y.Q. Qiao, J.P. Tu, J.Y. Xiang, X.L. Wang, Y.J. Mai, D. Zhang, W.L. Liu, Electrochimica

Acta 56 (2011) 4139.13] A. Pan, J. Liu, J.G. Zhang, W. Xu, G. Cao, Z. Nie, B.W. Arey, S. Liang, Electrochem-

istry Communications 12 (2010) 1674.14] L.L. Zhang, Y. Li, G. Peng, Z.H. Wang, J. Ma, W.X. Zhang, X.L. Hu, Y.H. Huang,

Journal of Alloys and Compounds 513 (2012) 414.15] F. Teng, Z.H. Hu, X.H. Ma, L.C. Zhang, C.X. Ding, Y. Yu, C.H. Chen, Electrochimica

Acta 91 (2013) 43.16] Y.Q. Qiao, J.P. Tu, Y.J. Mai, L.J. Cheng, X.L. Wang, C.D. Gu, Journal of Alloys and

Compounds 509 (2011) 7181.

17] H. Liu, P. Gao, J. Fang, G. Yang, Chemical Communications 47 (2011) 9110.18] L. Zhang, X.L. Wang, J.Y. Xiang, Y. Zhou, S.J. Shi, J.P. Tu, Journal of Power Sources

195 (2010) 5057.19] Y.Q. Qiao, X.L. Wang, J.Y. Xiang, D. Zhang, W.L. Liu, J.P. Tu, Electrochimica Acta

56 (2011) 2269.

[

[

Acta 108 (2013) 182– 190

20] J. Zhai, M. Zhao, D. Wang, Y. Qiao, Journal of Alloys and Compounds 502 (2010)401.

21] M.M. Ren, Z. Zhou, Y. Li, X.P. Gao, J. Yan, Journal of Power Sources 162 (2006)1357.

22] Y. Chen, Y. Zhao, X.J. An, Y. Dong, L. Chen, Electrochimica Acta 54 (2009) 5844.23] C. Dai, Z. Chen, H. Jin, X. Hu, Journal of Power Sources 195 (2010) 5775.24] Q. Kuang, Y. Zhao, X. An, J. Liu, Y. Dong, L. Chen, Electrochimica Acta 55 (2010)

1575.25] J.S. Huang, L. Yang, K.Y. Liu, Y.F. Tang, Journal of Power Sources 195 (2010)

5013.26] Y.G. Mateyshina, N.F. Uvarov, Journal of Power Sources 196 (2011)

1494.27] B. Jiang, S. Hu, M. Wang, X. Ouyang, Z. Gong, Rare Metals 30 (2011) 115.28] D. Ai, K. Liu, Z. Lu, M. Zou, D. Zeng, J. Ma, Electrochimica Acta 56 (2011)

2823.29] C. Deng, S. Zhang, S.Y. Yang, Y. Gao, B. Wu, L. Ma, B.L. Fu, Q. Wu, F.L. Liu, Journal

of Physical Chemistry C 115 (2011) 15048.30] M. Bini, S. Ferrari, D. Capsoni, V. Massarotti, Electrochimica Acta 56 (2011) 2648.31] Q. Kuang, Y. Zhao, Z. Liang, Journal of Power Sources 196 (2011) 10169.32] Y. Xia, W. Zhang, H. Huang, Y. Gan, C. Li, X. Tao, Materials Science and Engineer-

ing: B 176 (2011) 633.33] Y. Zhang, Q.Y. Huo, Y. Lv, L.Z. Wang, A.Q. Zhang, Y.H. Song, G.Y. Li, H.L. Gao, T.C.

Xia, H.C. Dong, Journal of Alloys and Compounds 2012 (2012) 187.34] S.Y. Chung, J.T. Bloking, Y.M. Chiang, Nature Materials 1 (2002) 123.35] J. Yang, Y. Bai, C. Qing, W. Zhang, Journal of Alloys and Compounds 509 (2011)

9010.36] D. Wang, H. Li, S. Shi, X. Huang, L. Chen, Electrochimica Acta 50 (2005) 2955.37] K.T. Lee, K.S. B. Lee, Journal of Power Sources 189 (2009) 435.38] D.W. Han, Y.M. Kang, R.Z. Yin, M.S. Song, H.S. Kwon, Electrochemical Commu-

nications 11 (2009) 137.39] H. Guo, X. Cao, X. Li, L. Li, X. Li, Z. Wang, W. Peng, Q. Li, Electrochimica Acta 55

(2010) 8036.40] C. Deng, S. Zhang, S.Y. Yang, B.L. Fu, L. Ma, Journal of Power Sources 196 (2011)

386.41] C. Deng, S. Zhang, S.Y. Yang, Journal of Alloys and Compounds 487 (2009) L18.42] L.L. Zhang, G. Liang, G. Peng, Y.H. Huang, L. Wang, L. Qie, M.C. Croft, A. Ignatov,

J.B. Goodenough, Journal of the Electrochemical Society 159 (2012) A1573.43] G. Yang, H. Ni, H. Liu, P. Gao, H. Ji, S. Roy, J. Pinto, X. Jiang, Journal of Power

Sources 196 (2011) 4747.44] T. Muraliganth, A. Manthiram, Journal of Physical Chemistry C 114 (2010)

15530.45] M. Wagemaker, B.L. Ellis, D.L. Hecht, F.M. Mulder, L.F. Nazar, Chemistry of

Materials 20 (2008) 6313.46] S. Patoux, C. Wurm, M. Morcrette, G. Rousse, C. Masquelier, Journal of Power

Sources 119–121 (2003) 278.47] J.R. Dahn, J. Jiang, L.M. Moshurchak, M.D. Fleischauer, C. Buhrmester, L.J. Krause,

Journal of the Electrochemical Society 152 (2005) A1283.