Kruti K. Halankara,
B. P. Mandal*ae,
Manoj K. Jangidb,
A. Mukhopadhyayb,
Sher Singh Meenac,
R. Acharyade and
A. K. Tyagi*ae
aChemistry Division, Bhabha Atomic Research Centre, Mumbai-400085, India. E-mail: bpmandal@barc.gov.in; aktyagi@barc.gov.in; Fax: +91-22-25505151; Tel: +91-22-25592274
bHigh Temperature and Energy Materials Laboratory, Department of Metallurgical Engineering and Materials Science, IIT Bombay, Mumbai-400076, India
cSolid State Physics Division, Bhabha Atomic Research Centre, Mumbai-400085, India
dRadiochemistry Division, Bhabha Atomic Research Centre, Mumbai-400085, India
eHomi Bhabha National Institute, Mumbai-400085, India
First published on 3rd January 2018
Carbon coated LixFePO4 samples with systematically varying Li-content (x = 1, 1.02, 1.05, 1.10) have been synthesized via a sol–gel route. The Li:Fe ratios for the as-synthesized samples is found to vary from ∼0.96:1 to 1.16:1 as determined by the proton induced gamma emission (PIGE) technique (for Li) and ICP-OES (for Fe). According to Mössbauer spectroscopy, sample Li1.05FePO4 has the highest content (i.e., ∼91.5%) of the actual electroactive phase (viz., crystalline LiFePO4), followed by samples Li1.02FePO4, Li1.1FePO4 and LiFePO4; with the remaining content being primarily Fe-containing impurities, including a conducting FeP phase in samples Li1.02FePO4 and Li1.05FePO4. Electrodes based on sample Li1.05FePO4 show the best electrochemical performance in all aspects, retaining ∼150 mA h g−1 after 100 charge/discharge cycles at C/2, followed by sample Li1.02FePO4 (∼140 mA h g−1), LiFePO4 (∼120 mA h g−1) and Li1.10FePO4 (∼115 mA h g−1). Furthermore, the electrodes based on sample Li1.05FePO4 retain ∼107 mA h g−1 even at a high current density of 5C. Impedance spectra indicate that electrodes based on sample Li1.05FePO4 possess the least charge transfer resistance, plausibly having influence from the compositional aspects. This low charge transfer resistance is partially responsible for the superior electrochemical behavior of that specific composition.
In order to obtain LiFePO4 with smaller particle size (preferably, in the nanosized regime, along with a conducting carbon coating), carbon sources like sucrose, glucose, fatty acids, polyaniline etc. have been used during synthesis by wet chemical methods.14–21 However, it has been frequently observed that during synthesis, in presence of above mentioned carbon sources, several impurity phases also form. Some of the impurities, like Fe2P and FeP, are believed to be useful, while others, such as Fe4P6O21 and Fe2P2O7 are detrimental towards the electrochemical performance of LiFePO4. The formation of these types of Li-deficient compounds is not surprising since energetically they have been reported to be more favourable.22 Formation of the Li-deficient compounds takes place due to volatility of lithium above 600 °C.19 Since Li-loss due to volatilization is unavoidable, researchers attempt using Li precursors in excess of those corresponding to the stoichiometric compositions. However, these are usually done fairly arbitrarily, with the excess lithium often precipitating as Li3PO4 which is electrochemically inactive and just add up to the dead weight. Also, these impurities, whether formed due to Li-deficit or Li-excess are often resistive in nature and thus highly undesirable.23
Accordingly, it is very important to determine very precisely the excess quantities of Li-precursors that need to be added to result in the optimum composition/stoichiometry, which need not be 100% phase pure LiFePO4, but which may lead to the best possible combination of electrochemical performances. For example, Hu et al.24 attempted to prepare non-stoichiomtric LiFePO4 by using precursors with Li:Fe in 1:2 ratio, which resulted in lower content of detrimental Li3PO4 and higher content of the desired Fe2P as the impurity phases. However, the content of LiFePO4 was low which led to lower specific capacity of the sample. There are few more reports on the effects of lithium non-stoichiometry on the electrochemical performances of LiFePO4, which also include discussions on how non-stoichiometry influences the particle size and lattice defects in LiFePO4.25–27
However, extensive information, especially based on systematic studies, regarding the Li:Fe stoichiometry attained, the concomitant impurity contents and their impact on the various electrochemical performances is scarcely reported in the literature. The challenge lies in precisely determining the lithium and iron concentrations in the as-synthesized samples. In most of the reported work, either atomic absorption spectroscopy (AAS) or inductively coupled plasma-optical emission spectroscopy (ICP-OES) has been used to determine the concentrations of Fe, as well as of Li.26,28 AAS and ICP-OES are not sensitive enough for determining the lithium concentration very precisely.
Accordingly, in the present work proton induced gamma emission (PIGE) technique has been used to determine the lithium concentration in the as-synthesized LixFePO4-based samples. PIGE is an isotope specific nuclear analytical technique capable of determining elements with very low Z (such as Li, B, F, N, Si, Al) using low energy proton beam (2–5 MeV). It can determine concentrations of elements non-destructively in complex materials like ceramics, glass and carbides, which are otherwise difficult to be analysed using conventional wet-chemical methods. PIGE is a particularly sensitive method for Li, which involves measurement of prompt gamma-rays at 478 keV from 7Li (p, p′γ)7Li. On the other hand 57Fe Mössbauer spectroscopy is very sensitive to determine the Fe-based impurities and accordingly has been used for the same in present work.
PIGE and Mössbauer spectroscopy has been used simultaneously to determine the final Li:Fe atomic ratios including impurity phases. Their influences on the electrochemical performances have been investigated for a set of LixFePO4-based samples synthesized with systematic variation of the starting Li contents (i.e., precursor amounts). The impacts of Li-deficit, as well as Li-excess, on the phase/impurity contents and the concomitant electrochemical properties like charge transfer resistances, charge/discharge capacities, rate capabilities and cyclic stabilities have been discussed. Accordingly, the correlations between composition/stoichiometry, phase assemblage, impurity contents and electrochemical behavior as obtained for the first time in such systematically conducted study not only highlights the importance of precise control of the stoichiometry while synthesizing LixFePO4-based electrode materials, but also shed light into the desired Li:Fe atomic ratio for obtaining the best possible electrochemical performances.
The concentration of iron was measured using Inductively Coupled Plasma-Optical Emission Spectroscopy (ICP-OES). Li-content was precisely determined using proton induced gamma emission (PIGE) technique. An in situ current normalized PIGE method has been developed by us, which was used earlier for non-destructive determination of Li-content in Li-doped neodymium di-titanate and lithium titanate ceramics29,30 and boron in boron-based compounds, including B4C.31 Accordingly, the optimized PIGE method has now been used for determination of Li concentrations in the four LiFePO4 samples as developed here with varied starting Li-contents (i.e., LiFePO4, Li1.02FePO4, Li1.05FePO4, Li1.10FePO4). As mentioned earlier, this is important because during synthesis Li has a tendency to sublime and also form compounds with Fe (i.e., impurity phases). The four samples with varied Li concentrations were analysed by the PIGE method with fluoride (in the form CaF2) as in situ current normalizer. The concerned samples and lithium phosphate standards (75 mg each) were pelletized in cellulose as major matrix, with constant amount of fluoride (in the form of CaF2). The method was also validated by analysing lithium acetate and lithium carbonate samples. The samples and standard pellets were irradiated (in vacuum at 10−6 torr) using 4 MeV proton beam from Folded Tandem Ion Accelerator (FOTIA) at 15 nA current. Radioactive assay of prompt gamma-rays at 478 keV from 7Li (p, p′γ)7Li and 197 keV of 19F (p, p′γ)19F was carried out using a 30% HPGe detector. In situ current normalized count rate was used for concentration calculation by relative method. Details regarding the calculations can be found in ref. 29 and 30. As will also be reported later, the concentrations of lithium in the four samples were found to be in the range of 3.7–4.7 wt%, with the associated uncertainties being in the range of 0.4–0.6% in the form of standard deviations of triplicate sample analyses.
Fig. 1 Representative XRD patterns recorded with all the four LFP-based samples (i.e., A, B, C and D; or starting Li-contents corresponding to x = 1.0, 1.02, 1.05 and 1.10 in LixFePO4). |
Samples | a (Å) | b (Å) | c (Å) |
---|---|---|---|
Sample A | 10.3224 (4) | 6.0025 (3) | 4.6857 (2) |
Sample B | 10.3233 (3) | 6.0031 (2) | 4.6881 (2) |
Sample C | 10.3232 (5) | 6.0032 (3) | 4.8951 (3) |
Sample D | 10.3243 (5) | 6.0038 (3) | 4.8889 (3) |
SEM images of the samples have been presented in ESI Fig. SI 1† which shows that the particles are of irregular shape. Mapping of Fe in all the samples have been performed to investigate the homogeneous distribution of Fe throughout the samples (see Fig. SI 2†). Raman spectra of all the samples show the presence of broad bands at 1358 (i.e., D-band from A1g vibration; partly representative of disorderness) and 1590 cm−1 (i.e., G-band from graphitic E2g vibration) (see Fig. 2). Not much variation in the ID/IG ratios (i.e., ∼1.02, ∼0.97, ∼0.97, ∼0.99, for samples A, B, C and D, respectively) could be seen across the sample sets, suggesting not much difference in the character (including graphitic nature and electronic conductivity) of the carbon-coatings.
Fig. 3 Typical gamma-ray spectrum of sample A, as obtained with proton induced gamma emission (PIGE) technique. |
Nominal composition | Li:Fe (atomic ratio) | |
---|---|---|
Sample A | LiFePO4 | 0.96:1 |
Sample B | Li1.02FePO4 | 0.97:1 |
Sample C | Li1.05FePO4 | 1.02:1 |
Sample D | Li1.1FePO4 | 1.16:1 |
Mössbauer spectra and parameters of all the samples have been presented in Fig. 4 and Table 3, respectively. The Mössbauer spectra of sample A could be well fitted into two doublets. The isomeric shift of the dominant doublet is found to be at 1.232 mm s−1 with quadrupole splitting (QS) as 2.964 mm s−1. This doublet corresponds to octahedral Fe2+ in ionic LiFePO4 and the relatively large QS is due to high spin configuration of d electron of Fe2+ (t42ge2g) and asymmetric electronic arrangement. The another doublet with IS and QS as 0.345 mm s−1 and 0.606 mm s−1, respectively, could be assigned to Fe3+ at octahedral site with high spin state. Interestingly, these Mössbauer parameters do not match well with either FePO4 or Fe2P, which has otherwise been reported to form during the synthesis of LiFePO4.19,33 In the second sample (i.e., sample B), three doublets have been observed which could be fitted well using previously reported data. In this sample, doublets corresponding to LiFePO4, FePO4 and FeP could be observed. More importantly, LiFePO4 concentration (88.2%) is found to be higher than that in the first sample (84.4%). Another striking observation is that the impurity FeP, which is known to be conducting, could be detected in this sample. Several authors have described its positive impact on electrochemical behaviour of LiFePO4.33,34 In the next sample (i.e., sample C), it has been observed that the concentration of LiFePO4 further increases to 91.5%, but with Fe3+-based impurity still being present. The IS and QS of the Fe3+-based compound matches with that of Fe4P6O21. The conducting FeP phase could be observed in this sample also. In the fourth sample (i.e., sample D) the concentration of LiFePO4 was found to be lower (86.9%) than the samples B and C.
Sample code | Fe sites | Quadrupole splitting (Δ EQ) mm s−1 | Isomer shift (δ) mm s−1 | Line width (Γ) mm s−1 | Relative area, RA (%) | Goodness of fit (χ2) |
---|---|---|---|---|---|---|
A | Doublet A (Fe2+) LiFePO4 | 2.964 ± 0.002 | 1.232 ± 0.001 | 0.287 ± 0.003 | 84.4 | 1.17213 |
Doublet B (Fe3+) | 0.606 ± 0.015 | 0.345 ± 0.009 | 0.373 ± 0.017 | 15.6 | ||
B | Doublet A (Fe2+) LiFePO4 | 2.967 ± 0.003 | 1.209 ± 0.001 | 0.29 ± 0.001 | 88.2 | 1.1305 |
Doublet B (Fe3+) FePO4 | 0.151 ± 0.06 | 0.238 ± 0.03 | 0.368 ± 0.1 | 7.4 | ||
Doublet C (FeP) | 1.477 ± 0.08 | 0.457 ± 0.08 | 0.24 ± 0.1 | 4.4 | ||
C | Doublet A (Fe2+) LiFePO4 | 2.97 ± 0.001 | 1.225 ± 0.003 | 0.266 ± 0.003 | 91.5 | 1.27856 |
Doublet B (Fe3+) Fe4P6O21 | 0.769 ± 0.07 | 0.504 ± 0.03 | 0.451 ± 0.07 | 5.9 | ||
Doublet C (FeP) | 1.171 ± 0.05 | 0.643 ± 0.02 | 0.23 ± 0.04 | 2.6 | ||
D | Doublet A (Fe2+) LiFePO4 | 2.973 ± 0.001 | 1.227 ± 0.001 | 0.276 ± 0.002 | 86.9 | 1.0252 |
Doublet B (Fe3+) Fe4P6O21 | 0.794 ± 0.07 | 0.516 ± 0.03 | 0.574 ± 0.1 | 8.0 | ||
Doublet C (Fe2P) | 0.147 ± 0.05 | 0.458 ± 0.06 | 0.416 ± 0.09 | 5.1 |
Fig. 5 Galvanostatic charge–discharge behaviour of LFP-based samples A, B, C and D at current density equivalent to C/2 rate. |
With respect to cyclic stability, the discharge capacity retentions after 100 galvanostatic cycles at C/2 for samples, A, B, C and D are ∼94%, ∼97%, ∼98% and ∼94%, respectively, of the corresponding first cycle capacities (Fig. 6). The slightly superior behaviour of samples B and C, as compared to A and D can be noted. In the context of rate capability, all the samples show the expected systematic decrease in specific capacity with increase in current density, with all the samples showing stable capacity retention in terms of recovering of the discharge capacity at C/2 rate after cycling through higher current densities (Fig. 7). More importantly, at a considerably high rate of 5C, the specific discharge capacities of samples B and C get retained at ∼99 and ∼107 mA h g−1, much better than those of samples A and D (∼85 mA h g−1). Overall, the results concerning electrochemical performances indicate that LFP-based samples B and C, especially C, are superior compared to samples A and D in all the aspects, viz., specific capacity, cyclic stability and rate capability. It has been reported that formation of two major defects, namely and , (anti-site defect) are energetically favourable in off-stoichiometric LiFePO4 which also degrade the electrochemical properties severely.37,38 It is not unlikely that these might also have a role in our LFP-based samples with relatively inferior performances. The CV curves of all the samples have been presented in ESI Fig. SI 3†.
Fig. 6 Specific (discharge) capacities recorded with the LFP-based samples A, B, C and D as functions of cycle numbers when galvanostatically cycled for up to 100 cycles at C/2 rate. |
Fig. 7 Specific (discharge) capacities recorded with the LFP-based samples A, B, C and D as at different current densities (C-rates). |
io = RT/nFRct | (1) |
Fig. 8 (Left panel) Nyquist plots obtained during EIS experiments with the LFP-based samples A, B, C and D (at discharged state). (Right panel) Circuit considered for fitting. |
Samples | Rs (Ohm) | Rct (Ohm) | Rw (Ohm) | io (mA) |
---|---|---|---|---|
Sample A | 10 | 76 | 11.0 | 0.335 |
Sample B | 9 | 68 | 55.3 | 0.374 |
Sample C | 9 | 57 | 54.9 | 0.446 |
Sample D | 10 | 77 | 63.5 | 0.331 |
Accordingly, the EIS data indicate that part of the reasons behind the superior overall electrochemical performances (including rate capability) of the electrodes prepared from sample C is related to lower charge transfer resistance and accordingly the greater exchange current density (implying greater charge transfer kinetics). It may also be recalled here that the stoichiometry of sample C also led to the formation of a conducting impurity phase, viz., FeP (as mentioned in Section 3.3). Such observations tend to indicate that the structural/interfacial defects/features and presence/absence of impurity phases caused due to off-stoichiometry (when deviate from the optimum Li concentration; as in samples A and D) negatively affected charge transfer resistance at the electrode/electrolyte interface (as opposed to the bulk transport), which in turn had the dominant effect on all the electrochemical performances.
With respect to the electrochemical performances, the electrodes prepared from sample C show the best performance in all aspects (viz., specific capacity, cyclic stability and rate capability). Furthermore, the electrode based on sample C could retain ∼107 mA h g−1 even when cycled at very high current density equivalent to 5C. Analysis of EIS data indicated that the electrode based on sample C possess the least charge transfer resistance, which in all probability partly accounted for the superior electrochemical behavior for the same. Overall, in more practical terms, the correlations between composition/stoichiometry, phase assemblage, impurity contents and electrochemical behavior, as obtained in the presently conducted systematic study, not only highlights the importance of precise control of the stoichiometry while synthesizing LixFePO4-based electrode materials, but also throws insights into the desired Li:Fe atomic ratio for obtaining the best possible electrochemical performances.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7ra10112k |
This journal is © The Royal Society of Chemistry 2018 |