Vous êtes sur la page 1sur 10

Small polaron hopping conduction mechanism in LiFePO4 glass and crystal

Azeem Banday, and Sevi Murugavel

Citation: Journal of Applied Physics 121, 045111 (2017);


View online: https://doi.org/10.1063/1.4974948
View Table of Contents: http://aip.scitation.org/toc/jap/121/4
Published by the American Institute of Physics

Articles you may be interested in


Small polaron hopping conduction mechanism in Fe doped LaMnO3
The Journal of Chemical Physics 135, 054501 (2011); 10.1063/1.3615720

Nature of small-polaron hopping conduction and the effect of Cr doping on the transport properties of rare-earth
manganite
The Journal of Chemical Physics 115, 1550 (2001); 10.1063/1.1378018

Structural refinement of Pbnm-type perovskite films from analysis of half-order diffraction peaks
Journal of Applied Physics 121, 045303 (2017); 10.1063/1.4974362

Determination of geometric and electronic structures of organic crystals from first principles: Role of the
molecular configuration on the electronic structure
Journal of Applied Physics 121, 045501 (2017); 10.1063/1.4974844

Nanoscale η-NiSi formation via ion irradiation of Si/Ni/Si


Journal of Applied Physics 121, 045302 (2017); 10.1063/1.4974456

Spatial luminescence imaging of dopant incorporation in CdTe Films


Journal of Applied Physics 121, 045304 (2017); 10.1063/1.4974459
JOURNAL OF APPLIED PHYSICS 121, 045111 (2017)

Small polaron hopping conduction mechanism in LiFePO4 glass and crystal


Azeem Banday and Sevi Murugavela)
Department of Physics & Astrophysics, University of Delhi, Delhi 110007, India
(Received 17 October 2016; accepted 13 January 2017; published online 31 January 2017)
The optimization of a cathode material is the most important criterion of lithium ion battery
technology, which decides the power density. In order to improve the rate capability, a cathode
material must possess high electronic and ionic conductivities. Therefore, it is important to
understand the charge transport mechanism in such an advanced cathode material in its intrinsic
state before modifying it by various means. In this work, we report the thermal, structural, and
electrical conductivity studies on lithium iron phosphate, LiFePO4, both in its polycrystalline
(LFPC) and glassy (LFPG) counterpart states. The vibrational spectroscopic measurements reveal
the characteristic vibrational modes, which are the intrinsic part of LFPC, whereas in LFPG, the
phonon modes become broader and overlap with each other due to the lattice disorder. The
electrical conductivity measurements reveal that LFPG exhibits a higher polaronic conductivity of
1.6 orders than the LFPC sample. The temperature dependent dc conductivity has been analyzed
with the Mott model of polarons and reveals the origin of enhanced polaronic conductivity in
LFPG. Based on the analysis, the enhanced polaronic conductivity in LFPG has been attributed to
the combined effect of reduced hopping length, decreased activation energy, and enhanced polaron
concentration. Published by AIP Publishing. [http://dx.doi.org/10.1063/1.4974948]

I. INTRODUCTION material has prompted many reports and debates on its physi-
cal properties including the nature of ionic and electronic
Lithium iron phosphate (LFP) is considered to be one
conductivities.
of the most promising cathode materials for next generation
More commonly, an electron transport in transition
rechargeable lithium ion batteries (LiBs).1 LFP has a high
metal oxides is described by the small polaron model in
theoretical specific capacity (170 mAh/g) and operating
which the excess charge carriers are coupled with distortions
potential (3.5 vs Liþ/Li) in addition to various other advan-
of nearby ions. In LFP, the active redox center is the transi-
tages including excellent electrochemical properties, low
tion metal ion with the Fe2þ state that becomes the Fe3þ state
cost, eco-friendliness, and extreme thermal and structural
upon removal of a lithium ion and an associated electron.
stability due to the strong PO4 covalent bonding.2 However,
Hence, the excess charge carriers are localized holes, and the
its usefulness as a potential cathode material in current lith-
subsequent deformation induced in the surrounding oxygen
ium ion batteries is limited by its intrinsic poor electronic
ions can be regarded as a pseudo particle called small hole-
and ionic conductivities. Several strategies have been made
polaron. Further, the formed small polarons are coupled with
to overcome these inherent limitations in LFP such as car-
lattice ions, and consequently, they are linked to phonons.
bon coating, doping with supervalent cations, and reduction
in particle size.3–7 Although carbon coating and/or doping Therefore, small polaron transport in solids is associated
with aliovalent cations lead to an adequate enhancement in with the strength of localization and coupled nature of pho-
the electronic conductivity at the expense of its energy den- nons. The transfer of these small polarons from an occupied
sity depending on the nature of electrochemically inactive to a neighboring unoccupied site can be thermally activated
additives.7 by a hopping or tunneling process depending on the tempera-
The chemical structure of polycrystalline LFP is classi- ture region considered.
fied into olivine family (also known as triphylite) having a Significant progress has been made in recent decades
distorted hcp framework with lithium and iron ions coexist- with respect to microscopic level understanding of the
ing in octahedral sites and phosphorous in tetrahedral sites.8 polaron mobility in LFP both theoretically and experimen-
The layers of FeO6 octahedra are corner-shared in the bc tally. Ab initio methods on the olivine phosphate structure
plane, and linear chains of LiO6 octahedra are edge-shared in reveal the migration barrier associated with small hole-
a direction parallel to the b-axis. These chains are bridged by polarons and its link with lithium ion vacancy, i.e., coupled
edge shared and corner shared phosphate tetrahedra, creating nature of movement.9 Subsequently, the first-principle den-
a stable three-dimensional structure. Importantly, the alkali sity functional theory reveals the polaron formation and its
cation arrangement in LFP differs from the layered and spi- migration energy within the framework of the GGA þ U
nel structures, where the alkali ions are considered to follow method in different olivine phosphate compounds. It predicts
one dimensional paths. The ability of LFP as an electrode that a lithium vacancy and an analogous hole-polaron form
the complex owing to lattice distortion and Coulomb interac-
a)
Author to whom correspondence should be addressed. Electronic mail: tion between them.10 Recently, the influence of the polaron
murug@physics.du.ac.in concentration on charge transport by comparing lithiated and

0021-8979/2017/121(4)/045111/9/$30.00 121, 045111-1 Published by AIP Publishing.


045111-2 A. Banday and S. Murugavel J. Appl. Phys. 121, 045111 (2017)

de-lithiated phases of the olivine phosphate structure has all the precursors is thoroughly mixed and ground in an agate
been shown experimenatlly.11 In the case of the lithiated mortar and pestle. The mixture was transferred into a platinum
phase, the concentration of polarons is determined by lithium crucible, kept in an electric furnace, and heated up to 1553 K.
vacancies, which have low formation energy, whereas in the In order to obtain the homogeneous glass, the melt was held at
de-lithiated phase, the polaron concentration is determined this temperature for half an hour and then rapidly poured out
by oxygen vacancies. In comparison with the lithiated phase, onto a stainless steel plate held under atmospheric conditions
the formation energy of oxygen vacancy is significantly and immediately pressed by another stainless steel plate. The
larger than that of lithium ion vacancy. Based on these ideas average thickness of the samples thus obtained was 1 mm.
and energy consideration, one would expect more number of Furthermore, in order to avoid the lithium loss, we have lim-
lithium vacancies in the disordered potential landscape, i.e., ited the melt duration (dwell time) held at high temperatures.
in the glassy state of LFP than the ordered landscape (poly- Additionally, we have carried out ICP-AES analysis on the
crystalline LFP). In this work, we report the detailed thermal, LFPG sample and estimated the content of individual oxides,
structural, and electrical conductivity measurements on stoi- and it has been observed that the loss of lithium content is less
chiometric polycrystalline LFP and its corresponding glassy than 0.5%.
state. The thermal study on the as-prepared glass sample was
We confirmed the absence of any impurity phase in as performed by using a high temperature Differential Scanning
synthesized polycrystalline LFP by high resolution XRD and Calorimeter (DSC) (SDT Q600 simultaneous TGA/DSC, TA
spectroscopic investigations. The obtained LFP glass has instruments). The DSC measurements were carried on the
been found to exhibit the characteristic glass transition and grounded glass sample under a nitrogen atmosphere with a
crystallization temperature. The spectroscopic investigations heating rate of 10 K/ min and from room temperature to
reveal the characteristic vibrational modes in the polycrystal- 923 K. X-ray diffraction (XRD) analysis on both the samples
line state, whereas in its glassy state, the modes become fea- was performed by using a Rigaku MINIFLEX II diffractom-
tureless with broad bands due to the localized nature of eter with a Cu Ka1 radiation source (k ¼ 1.5406 Å) and with
vibrational phonons. The electrical conductivity study a step size of 0.02 and at a scan rate of 1 min–1. Rietveld
reveals that the polaronic conductivity in glassy LFP is refinement analysis was carried out on the LFPC sample
higher than the polycrystalline counterpart. The temperature using TOPAS software, and the refinement was done by tak-
dependent electrical conductivity clearly shows that the ing into account the initial cell parameters and atomic posi-
small polaronic behavior is responsible in intermediate and tions reported in the literature.12 Furthermore, the various
high temperature regions. The analysis of temperature structural species present in both LFPC and LFPG were
dependent conductivity by using the Mott model of polaronic obtained with the help of Raman and Fourier Transform
conduction suggests the origin of enhanced polaronic con- Infra-Red Spectroscopy (FTIR) spectroscopic studies. FTIR
ductivity in glassy LFP. absorption spectroscopy (Perkin Elmer FTIR spectrometer,
Spectrum RXI-Mid IR) was carried out on powdered sam-
II. EXPERIMENTAL ples vacuum pressed to translucent discs with a spectral reso-
lution of 4 cm–1. The pellets were made by mixing the
Stoichiometric polycrystalline LiFePO4 (hereafter referred sample and KBr with the ratio of 1:100.
as LFPC) was synthesized by a single step solid state route The room temperature Raman spectra of both LFPC and
using precursors lithium carbonate (Li2CO3) (Sigma Aldrich, LFPG samples were recorded in the 180 back scattering geom-
99.997%), iron (II), oxalate dehydrate (FeC2O4.2H2O) (Sigma etry, using a 514 nm excitation of an air-cooled Argon Ion laser
Aldrich, >99.99%), and ammonium di-hydrogen phosphate (RenishawInViaReflex Micro Raman Spectrometer). The spec-
(NH4H2PO4) (Sigma Aldrich, 99.999%). The precursors in trometer was fitted with a single monochromator with a Peltier-
appropriate amounts were weighed, vigorously mixed, and cooled CCD (RenishawInVia Reflex). The spectral resolution
ground in an agate mortar and pestle for 4 h. The mixture was of the instrument was better than 1 cm1, and the incident
transferred into an alumina boat and then kept in a furnace. The power was adjusted to 3 mW for both the samples. The electri-
temperature of the furnace was raised slowly at the rate of 5 K/ cal conductivity measurements were done by using a
min up to 1023 K and maintained at this temperature for 6 h Novocontrol a-S high-resolution broad band dielectric analyzer
under reducing conditions (N2 with 10% H2). Subsequently, the in a frequency range from 10 mHz to 1 MHz and at different
furnace was then cooled down to room temperature, thereby temperatures between 153 and 673 K. The electrical conductiv-
giving the final product with average particle size in the micron ity measurements were carried out on the pellets of 13 mm
range. As obtained LFPC sample was analyzed by various tech- diameter and 1 mm thickness in the case of LFPC and for
niques to confirm the phase purity. Based on the high resolution LFPG case on samples with average diameter 8 mm and aver-
XRD and vibrational spectroscopic investigations, we ascertain age thickness 0.8 mm. On both sides, the specimens were sput-
the absence of any parasitic phases. tered with silver electrodes prior to conductivity measurements.
Glassy lithium iron phosphate (LFPG) was prepared by
the standard melt quenching technique. An appropriate molar III. RESULTS AND DISCUSSION
ratio of precursors lithium carbonate (Li2CO3) (Sigma
A. Thermal analysis
Aldrich, 99.997%), Iron (III) oxide (Fe2O3) (Sigma Aldrich,
 99%), and ammonium di-hydrogen phosphate (NH4H2PO4) In Figure 1, we illustrate the DSC scan for the LFPG
(Sigma Aldrich, 99.999%) was used. The weighed amount of sample, which exhibits the characteristic endothermic and
045111-3 A. Banday and S. Murugavel J. Appl. Phys. 121, 045111 (2017)

analysis on the LFPC diffraction pattern using TOPAS software,


and it is represented in Figure 2(b). The extracted cell parame-
ters are found to be a ¼ 10.3109 6 0.08 Å, b ¼ 5.9973 6 0.06 Å,
and c ¼ 4.6929 6 0.08 Å, and cell volume is 291.8 6 0.1Å3.
The estimated cell parameters are in agreement with the earlier
reports, and it corresponds to the orthorhombic structure.12,14 In
addition to the cell parameters, we have estimated the lithium

vacancy ðVLi Þ and anti-site defect ðFe•Li Þ concentrations on the
LFP crystal, and they were found to be 2.21% and 3.08%,
respectively.

C. FTIR studies
The vibrational characteristic properties of LFPC and
LFPG have been studied by FTIR spectroscopy as shown in
FIG. 1. The DSC scan of the LFPG sample at a scanning rate of 10 K/min. Figure 3, which is an effective tool for probing the various
Glass transition (Tg ¼ 758 K) and crystallization temperature (Tc ¼ 813 K) structural units present. Additionally, it also complements the
are marked in the figure. XRD pattern in terms of the existence of any parasitic phases
in the amorphous form. In LFPC, we observe distinct absorp-
exothermic transitions at 758 and 813 K, respectively. The tion bands, which are characteristic to the nature of vibrational
observed transitions are related to the glass transition tem- modes belonging to the olivine phosphate structure. The
perature (Tg) and crystallization temperature (Tc), and they assignments of various vibrational modes to respective struc-
are in agreement with earlier reported results.13 The differ- tural species have been reported and discussed13,15–17 recently.
ence between the Tg and Tc reflects the thermal stability of We briefly discuss the important vibrational bands and more
the LFPG sample. Furthermore, the observed single glass specifically those related to transition metal ion and oxygen
transition temperature in LFPG indicates the homogeneity of ions.
the glass sample without any secondary phases. In LFP, the layers of iron octahedra are corner-shared in
the bc plane, h110i, and the linear chains of LiO6 octahedra
B. XRD studies are edge-shared in a direction parallel to the b-axis, h010i. A
stable three dimensional network structure has been formed
The XRD patterns of both LFPC and LFPG samples are by the linear chains, where they are bridged by edge and
shown in Figure 2(a). The LFPG sample shows the characteris- corner-shared phosphate tetrahedral units. Hence, the vibra-
tic broad hump at 2h angle between 15 and 35 in the diffrac- tional spectrum is mainly dominated by the fundamental
tion pattern. On the other hand, the diffraction pattern on the vibrations of PO43 anions, which are assumed to separate
LFPC sample shows the characteristic peaks, and they are into many other components due to the correlation effect
assigned to the stoichiometric LFP crystal with an orthorhombic induced by the coupling with Fe-O units in the structure.
structure and space group Pnma (JCPDS-NO. 83–2092). The Typically, the vibrational bands originate from the internal
XRD pattern clearly indicates that the obtained LFP polycrystal- vibrations ( 1- 4) of PO43 units, which concern the dis-
line sample is free from any impurity phases, and all the Bragg placement of oxygen atoms of the tetrahedral and lie at high
reflections are indexed to the orthorhombic olivine-type struc- frequencies (above 800 cm–1). The bands in the range of 450
ture of LFPC. Further, we have carried out Rietveld-refinement to 600 cm1 are due to the bending modes ( 2 and  4), which

FIG. 2. (a) Room temperature XRD pattern of LFPG and LFPC samples, showing, respectively, the amorphous and crystalline nature of as prepared samples.
The peaks are assigned to the LiFePO4 crystal with an orthorhombic structure. (JCPDS file No.: 83-2092) (b) Experimental X-ray powder diffraction pattern
(red open circles) of the LiFePO4 sample compared to the Rietveld refine profile (continuous blue lines) and the difference curve (bottom curve) taken at room
temperature. The vertical markers (blue) below the diffraction pattern indicate positions of possible Bragg reflections.
045111-4 A. Banday and S. Murugavel J. Appl. Phys. 121, 045111 (2017)

FIG. 3. (a) Fourier Transform Infra-Red


(FTIR) absorption spectra of LFPG and
LFPC samples recorded at room tem-
perature, showing good resemblance
between the bands of both the samples.
(b) The deconvoluted FTIR spectra of
LFPG (400–750 cm1) depicting bands,
which are envelops of those observable
in LFPC.

involve mainly O-P-O symmetric and anti-symmetric bend- with a formula unit cell. Li, Fe, and P atoms are distributed
ing. The band at around 548 cm1 is due to the translational on 4a, 4c, and 4c positions (Wyckoff notations), respec-
vibration of Li within the PO43 cage. The bands at tively. Since the structure of the phospho-olivines is built
638 cm1 and 645 cm1 are corresponding to the contribu- upon LiO6 and MO6 octahedral linked to PO43 tetrahedral
tions of FeO6. On the other hand, the bands in the range anions, the local cation environment can be analyzed with
900–1200 cm1 are due to the stretching modes ( 1 and  3) the factor group and the molecular vibrational modes.18
involving symmetric and anti-symmetric modes of the P-O The Raman spectroscopic studies of LFPC have already
bonds. It is important to mention that we do not observe any been reported and discussed in the literature.16,19–24 In the
other bands corresponding to other phosphate ion complexes. present study, the LFPC sample exhibits a high intensity
By comparing the FTIR spectrum of LFPC with LFPG, we peak located at 950 cm1 and is ascribed to symmetric P-O
find that the characteristic bands become featureless but stretching vibrational modes. Two weak intensity bands at
coincident and lie somewhat at similar positions. However, it 998 cm1 and 1070 cm1 are ascribed to anti-symmetric
can be seen that the peak intensities are not very prominent vibrational modes. The vibrational bands at around 570,
and visible in vitreous samples, and these broad peaks are 590, and 630 cm1 are attributed to Fe-O stretching modes,
quite obvious in glassy structures. In contrast to LFPC, the and bands at 400 and 450 cm1 are belonging to Fe-O-P
FTIR spectra of LFPG exhibit two broad bands centered in bending modes. The vibrational bands corresponding to
high and low frequency regions, and these bands consist of these iron octahedral units are playing an important role in
several other vibrational modes, which are localized in the polaron transport phenomenon. Below 400 cm1, the
nature due to the lattice disorder. In Figure 3(b), we illustrate external optical modes are observable, which are due to
the de-convoluted FTIR spectra of LFPG, which shows a vibrations of the lattice as a whole, and they are mainly due
good resemblance between the FTIR spectra of glassy and to translational motion of PO43 and Fe2þ ions.21 The band
polycrystalline LFP samples and confirms that the character- around 300 cm1 corresponds to the motion of lithium ions,
istic vibrational modes are retained. but it gives very weak contributions to these vibrations. In
order to obtain a quantitative estimation of various struc-
D. Raman studies
tural species in the LFPG, it is necessary to use the decon-
In Figure 4(a), we depict the Raman spectra of both volution procedure with the aim of separating individual
LFPC and LFPG samples. In olivine LFPC, most intense vibrational modes from the composite line envelope. In
Raman bands are observable in the region 900–1200 cm1, Figure 4(b), we illustrate the de-convoluted Raman spectra
while many weak peaks can be seen in the low frequency of LFPG, which clearly depicts characteristic bands that are
region. The olivine structure belongs to the spectroscopic observable in the LFPC with a good concurrence between
group D16
2h with the primitive cell being centro-symmetric the spectral frequencies.

FIG. 4. (a) Raman spectra of LFPC


and LFPG samples recorded at room
temperature, showing a visible differ-
ence in the peak intensities. (b) The
deconvoluted Raman Spectra of LFPG
(540–800 cm1) depicting bands, which
are envelops of those observable in
LFPC.
045111-5 A. Banday and S. Murugavel J. Appl. Phys. 121, 045111 (2017)

E. Broad band electrical conductivity spectroscopic from few others.4,34 This discrepancy could be due to various
study reasons such as carbon content, metallic impurity phases,
The frequency dependent electrical conductivity spectra synthesis procedure, and lithium vacancy related contribu-
of both LFPC and LFPG samples at different temperatures tions. However, it should be noted that in our samples, the
are shown in Figure 5. The observed frequency dependent amount of carbon content is far below the detection limit of
conductivity behavior is typical of most polaron conducting Raman spectroscopy or it may be less than the ppm level.
crystals and glasses.25–28 At low frequencies and relatively Therefore, the measured dc conductivity exclusively arises
higher temperatures, the measured conductivity becomes from the contributions of small polarons, and it is an intrinsic
independent of frequency and is called dc conductivity. On property of the LFPC and LFPG samples. Remarkably, the
the other hand, at higher frequencies and comparatively measured dc conductivity of the LFPG sample is roughly
lower temperature, the conductivity becomes frequency 1.64 times higher than the corresponding polycrystalline
dependent above the characteristic or cross over frequency counterpart. The obtained dc conductivity values of LFPG
( *), defined by r0 ( *) ¼ 2rdc, which increases with an are in agreement with the literature report.13 Typically,
increase in temperature.29,30 It can be said that the dc con- polaron conducting glasses are found to exhibit lower electri-
ductivity and the characteristic frequency are thermally acti- cal conductivity than the corresponding polycrystalline
vated processes with the same activation energy. It is worth state.35 However, the situation in the case of LFPG is differ-
to mention that the measured polaronic conductivity in ent, and its origin would be discussed in detail by analyzing
LFPC and LFPG samples are exclusively due to the bulk the temperature dependent dc conductivity data using the
part, thereby neglecting the contribution (if any) of grain Mott model of polaronic conduction. In Figure 6, we illus-
boundaries as evidenced by a single semi-circle in the trate the measured dc conductivity of both LFPG and LFPC
Nyquist plot. Due to very high impedance at low tempera- over a wide range of temperatures. Clearly, the temperature
tures, electrical conductivity measurements were limited at dependent dc conductivity data of LFPC show the typical
273 K and 153 K, respectively, for LFPC and LFPG samples. polaronic type behavior with different activation energies
We ascertain that the measured conductivity at low frequen- depending on the temperature region. We separate the dis-
cies is exclusively due to polarons, which has been con- tinct regions where the slope gets changed, and it is fitted
firmed by independent ionic conductivity measurement on with the Mott model of polaronic conduction at temperatures
these samples. We have also carried out lithium ion conduc- above room temperature. In the case of LFPG, we could
tivity measurements using the electron blocking electrode extend the temperature range down to 153 K (due to low
geometry, i.e., Cu/LiI/LFP/LiI/Cu, and compared with the impedance), where the observed behavior resembles the vari-
Cu/Ag/LFP/Ag/Cu electrode geometry. Interestingly, the able range hopping (VRH) mechanism. In the following, we
measured ionic conductivity is much higher than (order of discuss each region in detail on the underlying polaronic
107 S/cm) the polaronic conductivity at 303 K. Therefore, conduction mechanism.
we suggest that the presented dc conductivity results in this
1. High and intermediate temperature regimes
work are exclusively due to the polarons, and the ionic con-
tributions can be neglected. Mott theoretically devised a model for hopping conduc-
The measured dc conductivity values of LFPC are in tion in non-crystalline solids in terms of phonon assisted
accordance with the earlier reports31–33 while they differ hopping of small polarons between the localized states.36

FIG. 5. The real part of the frequency dependent conductivity spectra of (a) LFPC and (b) LFPG measured at various temperatures.
045111-6 A. Banday and S. Murugavel J. Appl. Phys. 121, 045111 (2017)

obtained Debye temperature for the LFPC sample is in


agreement with the earlier simulation results37 while the hD
value for LFPG is also close to the reported literature
value.38 The hD is a fundamental quantity, which character-
izes the vibrational spectrum of a polycrystalline material,
and attempts have been made to use these concepts in ana-
lyzing the behavior of amorphous materials like glasses. The
obtained Debye temperature for LFPG is significantly
smaller than the LFPC, which indicates that there is a
decrease in the vibrational energy spectrum for the LFP
glass. The experimental dc conductivity data have been fitted
by the Mott equation in high and intermediate temperature
regions and give a very good fitting with the best parameter
fit, where R2 (R is the fitting approximation) lies in the range
of 0.9962–0.9989. The estimated values of the Debye tem-
perature along with the various other physical parameters
FIG. 6. Arrhenius plots of the dc conductivity of LFPC and LFPG samples
measured over a wide range of temperatures. The symbols represent experi-
derived from the Mott equation are given in the Table I.
mental data points, while the solid line corresponds to the fitting of tempera- In LFPC, the polaron hopping takes place within the bc
ture dependent dc conductivity with Mott’s Equation (1) at high and plane because the inter-planer hopping distance (5.5 Å) is
intermediate temperatures. The best parameter fit, R2, lies in the range larger than the intra-planer distance (3.8 Å).39,40 Hence, the
0.9983–0.9989, and the error factor of least square fit (Chi2) lies in the range
4.3681  1013–9.4109  1014 in the case of high temperature. Similarly, for intra-planer distance corresponds to the typical hopping
the intermediate temperatures, R2 and Chi2 lie in the range 0.9962–0.9971 and length, i.e., distance between two neighboring Fe sites. We
1.5212  1013–2.3068  1014, respectively. found from our analysis that the polaron hopping length is
3.78 Å for LFPC, and the average hopping length is 2.65 Å
The dc conductivity of nearest neighbor hopping in the non- for LFPG due to the lattice disorder. However, it is interest-
adiabatic regime or at high temperature (T > hD/2) is given ing to note that the intermediate temperature activation
by energy values are comparable with the reported results of
 2    iron containing crystals and glasses.33,41 Furthermore, it is
e c ð1  c Þ Ea worthwhile to mention that the estimated values are in agree-
rdc ¼ tph  expð2aRÞexp  ; (1)
kB TR kB T ment with some of the reports in the literature and differ
from some others. The discrepancy between our and earlier
where  ph is the longitudinal optical phonon frequency, R is
works could be due to various factors like purity of the sam-
the distance between two neighboring Fe ions, a is the inverse
ple and the temperature window used for the conductivity
localization length of the s-like wave function that is assumed
measurements. In the present work, we have covered a wide
to describe the localized state at each transition metal ion, hD
temperature range between high and low temperatures and
is the Debye temperature, and “c” is the ratio of the concentra-
extending to the sub-hertz region of the frequency window,
tion of polaronic sites (Fe3þ) over the total amount of Fe ions
where the estimated dc conductivity shows well defined pla-
in the matrix. Ea is the activation energy for the hopping con-
teau, and it removes any uncertainties in the estimated val-
duction. In Figure 6, we illustrate experimental data fitted
ues. Therefore, we infer that the assessed values are intrinsic
with Equation (1). Clearly, we find distinct slopes with tem-
properties of our LFPC and LFPG samples. It is interesting
perature exhibiting the positive curvature. We are able to iden-
to note that the activation energies of LFPG are significantly
tify three different regions namely (i) high temperature (Ht),
lower than that of LFPC, which emphasizes that the energy
(ii) intermediate temperature (It), and (iii) low temperature
barrier involved for polaron hopping in the glass network is
regions (Lt). Further, we identify half of the Debye tempera-
smaller than the crystal.
ture, hD, at which the crossover takes place from intermediate
Remarkably, we find that the polaron concentration esti-
temperature to high temperature by defining
mated from the Mott fitting analysis substantiates the differ-
htph ¼ khD ; (2) ence between the dc conductivities of LFPG and LFPC
samples. It is interesting to note that the concentration of
where  ph is the longitudinal optical phonon frequency, h polaronic sites in LFPG is nearly an order of magnitude
is Planck’s constant, and k is Boltzmann’s constant. The higher than the LFPC. More commonly, in the case of LFPC,
TABLE I. Summary of Mott’s parameters for LFPC and LFPG samples extracted by using Equation (1) along with obtained dc conductivity data at 313 K. hD is
the Debye temperature,  ph is the optical phonon frequency, R is the polaron hopping length, EaLt is the low temperature activation energy, EaIt is the intermediate
temperature activation energy, EaHt is the high temperature activation energy, C is the fraction of concentration of polarons, and rdc is the dc polaronic conductivity.

Sample hD (K)  p (Hz) R (Å) EaLt (eV) EaIt (eV) EaHt (eV) C rdc at 313 K (S/cm)
2
LFPC 579 1.20  10 13
3.78 … 0.62 0.78 4  10 2.3  1010
LFPG 396 0.82  1013 2.65 0.28 0.46 0.67 1  101 8.7  109
045111-7 A. Banday and S. Murugavel J. Appl. Phys. 121, 045111 (2017)

the active redox center is the transition metal ion with the where the constants A and B are defined as
Fe2þ state, which becomes Fe3þ upon removal of lithium ion
and an associated electron. Thus, the excess charge carriers A ¼ vph e2 NðEF ÞR2 ; (4)
are localized holes, and the corresponding contraction of sur- " #14
rounding oxygen ions is together known as small hole polar- a3
B ¼ B0 ; (5)
ons. Hence, the source of Fe3þ centers is considered to be kN ðEF Þ
lithium vacancies generated during the synthesis and by the
surface energy kinetics. The recent theoretical studies reveal where Bo is a constant and has different values for different
that the formation energy of lithium vacancy is 0.22 eV, treatments,45 a is the inverse localization length, N(EF) is the
which is much lower than the formation energy of other density of states at the Fermi level, and R is the hopping dis-
types of defect present in the LFP.11,42 Due to the low forma- tance given by the following relation:
tion energy, the occurrence of lithium vacancy in LFPG  14
becomes more probable than the corresponding LFPC. 3
R¼ : (6)
Among the various factors contributing to polaronic conduc- 2paN ðEF ÞkT
tion in transition metal phosphates, the polaron concentration
is more relevant and is proportional to lithium vacancies in Using the above equations, we get
LFP. The estimated concentration of polarons, c, is slightly hBi

larger than the earlier reported values on LFPC.10,33 To the 1
rdc T 2 ¼ C:e
1
T4
; (7)
best of our knowledge, the c value for LFPG has not been
reported earlier in the literature. The disparity between the where C is a constant given by
different values of c could be due to differences in prepara-

1
tion conditions used during the synthesis process and temper- 3N ðEF Þ 2
2
ature range of measured values. Furthermore, it is C ¼ ph e : (8)
2pak
worthwhile to mention that the polaron transport in solids is
closely related to a certain type of phonon mode with spe- In Figure 7, we represent a plot of log rdc T1/2 vs T1/4 where
cific phonon energy. These phonons are typically longitudi- the present experimental data have been fitted with Mott and
nal optically (LO) active phonons with the frequency range Greaves’s model [Equation (7)]. It is found that the rdc T1/2
of 1013–1014 Hz. In LFPC, the observed vibrational mode at vs T1/4 plot gives a linear fit in the low temperature region,
570, 590, and 630 cm1corresponds to the Fe-O stretching which is suggestive of the presence of variable range hopping.
mode, which contributes to the polaronic conduction, where The values of a and R are in agreement with previously
the two iron sites involved in the polaron transport have a estimated values at the intermediate temperature regime, and
common bridging oxygen atom. It is interesting to note that the value of N(EF) obtained as such is equal to 6.97
the deconvoluted Raman bands on LFPG result in similar  1021 eV1cm3 and is quite reasonable with those obtained
peak positions (with chi square value of one) and corre- from polaron conducting glasses.46
spondingly similar phonon frequency ( ph) values in LFPC The real part of frequency dependent conductivity has
and LFPG. been analyzed by using various formalisms in the literature
and in different classes of materials, which provide important
2. Low temperature regime
Low temperature conductivity measurements were con-
fined to the LFPG sample because of the high impedance in
the LFPC sample. Mott and Davis43 suggested that when aR
 1, the hopping of charge carriers is described by the near-
est neighbor hopping and when aR 1 or at sufficiently low
temperatures, hopping may preferentially occur beyond the
nearest neighbor by the variable range hopping (VRH) pro-
cess. It is clear from Figure 6 that the estimated dc conduc-
tivity values below 223 K have a distinct slope from the
high/intermediate temperature regions. We ascribe the low
temperature behavior to be the signature of VRH and discuss
it in detail in terms of the theoretical model developed by
Mott and Greaves. The theoretical models proposed by
Mott43 and Greaves44 are generally used to explain the 3D
VRH in bulk disordered semiconductors. Based on Mott’s
VRH model, the temperature dependent dc conductivity is
FIG. 7. A plot of log (rdc .T1/2) vs T1/4 for LFPG at low temperatures where
given by
the symbols represent the present experimental data points, and the solid
  line corresponds to the fitting in the low temperature region, fitted with Mott
B and Greaves’ model given by Equation (7). The best parameter fit (R2) is
rdc ¼ A exp  1 ; (3)
T4 0.9987, and the error factor of least square fit (Chi2) is 3.6425  1013.
045111-8 A. Banday and S. Murugavel J. Appl. Phys. 121, 045111 (2017)

information on the charge transport in various length charge carrier in the structure. The magnitude of the expo-
scales.47 In this context, we used the empirical Jonscher rela- nent directly implies the strength of the localized motions,
tion to describe the observed frequency dependent conduc- which is found to be larger at low temperatures and at high
tivity behavior and is given by48 frequencies. In addition, it has been suggested that the mag-
"  s # nitude of the exponent indicates effective dimensionality of
t the charge carrier performing a hopping event in homoge-
rðtÞ ¼ rdc 1 þ
; (9)
t neous potential landscape.47 On the other hand, the relatively
smaller exponent value of LFPG indicates that the charge
where the exponent s is independent over a range of tempera- carrier motion becomes less localized in character. The mini-
tures and t* is the characteristic frequency at which the dis- mal localized nature of polarons can be understood by the
persive conductivity changes into the dc conductivity. It is a smaller energy barrier between the active redox sites. In the
general trend that in different ion and polaron conducting case of LFPC, the large value of the exponent could be
glasses and crystals, the exponent appears to increase with attributed to the strong Coulomb interaction between the
decreasing temperatures, and it approaches the value of unity polarons influenced by the lithium vacancy.
at sufficiently low temperatures.48 In Figure 8, we illustrate
the real part of the frequency dependent conductivity spectra IV. CONCLUSIONS
at 313 K fitted with the Jonscher relation, and the exponent
A comparison between olivine structured LFPC and
value is found to be 0.94 6 0.01 in the LFPC sample and
their amorphous analogs is drawn for the first time.
0.65 6 0.005 in the LFPG sample, respectively. Similarly,
Characterization by various analytical techniques confirmed
we have fitted all the experimental conductivity isotherms of
the phase purity of the as prepared samples, which is neces-
both LFPC and LFPG samples and found that the exponent
sary for understanding the microscopic charge transport
does not vary with temperature at high and intermediate tem-
mechanism. The polaronic conductivity of glassy samples
peratures. On the other hand, in LFPG, we found that there is
(LFPG) is found to be 1.64 orders of magnitude higher than
a significant increase (approaching unity) in the exponent
values at very low temperatures, which could be ascribed to that of LFPC. The conductivity enhancement in LFPG has
nearly constant loss behavior.49 been attributed to the increased concentration of polarons
More commonly, the Jonscher exponent (0 < s < 1) is due to the large concentration of lithium vacancies.
often used to describe the dispersive region of conductivity Furthermore, the temperature dependent dc conductivity has
spectra and is the result of localized hopping motions of been explained by using the Mott model of polaronic con-
mobile charge carriers. The ac conductivity measurement, duction in the intermediate and high temperature regions. In
rðtÞ, generally shows a frequency dispersion region, i.e., a the case of LFPG, we find variable range hopping conduc-
dependence on the angular frequency, t. Exploring this dis- tion, and it has been explained on the basis of the model pro-
persive behavior provides an opportunity to gain insight into posed by Mott and Greeves. The extracted Mott parameters
the details of charge carrier migratory processes, in particu- support the origin of enhanced polaronic conductivity in the
lar, the interaction of the migrating polarons with other LFPG sample. The estimated Jonscher exponent value for
LFPC and LFPG indicates the dimensionality and nature of
polaron transport.

ACKNOWLEDGMENTS
A.B. thanks UGC for financial support. We also thank
the Department of Science and Technology (DST), India,
and R&D Scheme of Delhi University for the financial
support. We also acknowledge M.Tech Nano Science lab
and USIC, University of Delhi, for providing various
characterization facilities during the course this work.
1
A. K. Padhi, K. S. Nanjundaswami, and J. B. Goodenough,
J. Electrochem. Soc. 144, 1188 (1997).
2
M. Takahashi, S.-I. Tobishima, K. Takei, and Y. Sakurai, Solid State
Ionics 148, 283 (2002).
3
S. Y. Chung, J. T. Bloking, and Y. M. Chiang, Nat. Mater. 1, 123 (2002).
4
P. S. Herle, B. Ellis, N. Coombs, and L. F. Nazar, Nat. Mater. 3, 147
(2004).
5
FIG. 8. The real part of the frequency dependent conductivity spectra of H. Huang, S. C. Yin, and L. F. Nazar, Electrochem. Solid-State Lett. 4,
LFPC and LFPG samples measured at 313 K. The symbols represent experi- A170 (2001).
6
mental data points, while the solid line corresponds to the fitting done with T. Drezen, N. H. Kwon, P. Bowen, I. Teerlinck, M. Isono, and I. Exnar,
the Jonscher relation given in Equation (9). The best parameter fit, R2, lies J. Power Sources 174, 949 (2007).
7
in the range 0.99957–0.99972, and the error factor of least square fit (Chi2) S. Ferrari, R. L. Lavall, D. Capsoni, E. Quartarone, A. Magistris, P.
lies in the range 7.2085  1012–3.1086  1013 for LFPC and the corre- Mustarelli, and P. Canton, J. Phys. Chem. C 114, 12598 (2010).
sponding values of 0.99974–0.99983 and 8.8107  1013–8.4171  1014 8
A. A. Salah, A. Mauger, C. M. Julien, and F. Gendron, J. Mater. Sci. Eng.
for LFPG, respectively. B 129, 232 (2006).
045111-9 A. Banday and S. Murugavel J. Appl. Phys. 121, 045111 (2017)

9 27
B. Ellis, L. K. Perry, D. H. Ryan, and L. F. Nazar, J. Am. Chem. Soc. 128, B. Roling, and K. Funke. J. Non-Cryst. Solids 212, 1 (1997).
28
11416 (2006). S. K. Deshpande, S. N. Achary, R. Mani, J. Gopalakrishnan, and A. K.
10
Y. Asari, Y. Suwa, and T. Hamada, Phys. Rev. B 84, 134113 (2011). Tyagi, Phys. Rev. B 84, 064301 (2011).
11 29
S. Murugavel, M. Sharma, and R. Shahid, J. App. Phys. 119, 045103 J. L. Barton, Verres Refract. 20, 328 (1966).
30
(2016). H. Namikawa, J. Non-Cryst. Solids. 18, 173 (1975).
12 31
N. Meethong, H. Y. S. Huang, S. A. Speakman, W. C. Carter, and Y. M. C. Delacourt, L. Laffont, R. Bouchet, C. Wurm, J. B. Leriche, M.
Chiang, Adv. Funct. Mater. 17, 1115 (2007). Morcrette, J. M. Tarascon, and C. Masquelier, J. Electrochem. Soc. 152,
13
P. Jozwiak, J. E. Garbarczyk, M. Wasiucionek, I. Gorzkowska, F. A913 (2005).
32
Gendron, A. Mauger, and C. M. Julien, Solid State Ionics 179, 46 (2008). K. Zaghib, A. Mauger, J. B. Goodenough, F. Gendron, and C. M. Julien,
14
P. Gibot, M. C. Cabanas, L. Laffont, S. Levasseur, P. Carlach, S. Hamelet, Chem. Mater. 19, 3740 (2007).
33
J. M. Tarascon, and C. Masquelier, Nat. Mater. 7, 741 (2008). R. Shahid and S. Murugavel, Phys. Chem. Chem. Phys. 15, 18809 (2013).
15 34
M. T. Faques-Ledent and P. Tarte, Spectrochim. Acta, Part A 30, 673 R. Amin, P. Balaya, and J. Mair, Electrochem. Solid-State Lett. 10, A13
(1974). (2007).
16 35
C. M. Burba and R. Frech, J. Electrochem. Soc. 151, A1032 (2004). H. Hirashima, M. Ide, and T. Yoshida, J. Non-Cryst. Solids 86, 327
17
A. A. Salah, P. Jozwiak, K. Zaghib, J. E. Garbarczyk, F. Gendron, A. (1986).
36
Mauger, and C. M. Julien, Spectrochim. Acta, Part A 65, 1007 (2006). I. G. Austin and N. F. Mott, Adv. Phys. 50, 757 (2001).
18 37
T. Nakamura, Y. Mima, M. Tabuchi, and Y. Yamada, J. Electrochem. Soc. T. Maxisch and G. Ceder, Phys. Rev. B 73, 174112 (2006).
38
153, A1108 (2006). J. M. Farley and G. A. Saunders, Phys. Status Solidi A 28, 199 (1975).
19 39
M. Bini, M. C. Mozzati, P. Galinetto, D. Capsoni, S. Ferrari, M. S. Grandi, K. Hoang and M. Johannes, Chem. Mater. 23, 3003 (2011).
40
and V. Massarotti, J. Solid State Chem. 182, 1972 (2009). J. Lee, S. J. Pennycook, and S. T. Pantelides, Appl. Phys. Lett. 101,
20
D. Arumugam, G. P. Kalaignan, and P. Manisankar, J. Solid State 033901 (2012).
41
Electrochem. 13, 301 (2009). L. Murawski and O. Gzowski, Phys. Status Solidi A 19, K125 (1973).
21 42
A. A. Salah, A. Mauger, K. Zaghib, J. B. Goodenough, N. Ravet, M. F. Zhou, C. A. Marianetti, M. Cococcioni, D. Morgan, and G. Ceder, Phys.
Gauthier, F. Gendron, and C. M. Julien, J. Electrochem. Soc. 153, A1692 Rev. B 69, 201101 (2004).
43
(2006). N. F. Mott and E. A. Davis, Electronic Processes in Non-Crystalline
22
C. M. Burba, J. M. Palmer, and B. S. Holinsworth, J. Raman Spectrosc. Materials (Oxford University Press, Oxford, 1979).
44
40, 225 (2009). G. N. Greaves, J. Non-Cryst. Solids 11, 427 (1973).
23 45
D. X. Gouveia, V. Lemos, J. A. C. Paiva, A. G. S. Filho, and J. M. Filho, C. H. Seager and G. E. Pike, Phys. Rev. B 10, 1435 (1973).
46
Phys. Rev. B 72, 024105 (2005). R. Punia, R. S. Kundu, S. Murugavel, and N. Kishore, J. App. Phys. 112,
24
W. Paraguassu, P. T. C. Freire, V. Lemos, S. M. Lala, L. A. Montoro, and 113716 (2012).
47
J. M. Rosolen, J. Raman Spectrosc. 36, 213 (2005). D. L. Sidebottom, Rev. Mod. Phys. 81, 999 (2009).
25 48
N. F. Mott, J. Non-Cryst. Solids 1, 1 (1968). A. K. Jonscher, Nature 267, 673 (1977).
26 49
A. P. Schmid, J. App. Phys. 40, 4128 (1969). W. K. Lee, J. F. Liu, and A. S. Nowick, Phys. Rev. Lett. 67, 1559 (1991).

Vous aimerez peut-être aussi