Relithiation of LiCoPO4: A Two-Step

Jul 9, 2014 - D. Liu , W. Zhu , C. Kim , M. Cho , A. Guerfi , S.A. Delp , J.L. Allen , T.R. Jow , K. Zaghib ... Monica Kosa , Doron Aurbach , Dan Thom...
0 downloads 0 Views 6MB Size
Download PDF
Article pubs.acs.org/JPCC

Electrochemical Delithiation/Relithiation of LiCoPO4: A Two-Step Reaction Mechanism Investigated by in Situ X‑ray Diffraction, in Situ X‑ray Absorption Spectroscopy, and ex Situ 7Li/31P NMR Spectroscopy Maximilian Kaus,† Ibrahim Issac,† Ralf Heinzmann,†,‡ Stephen Doyle,§ Stefan Mangold,§ Horst Hahn,†,# Venkata Sai Kiran Chakravadhanula,# Christian Kübel,†,# Helmut Ehrenberg,#,∥ and Sylvio Indris*,†,#,∥ †

Institute of Nanotechnology, Karlsruhe Institute of Technology, P.O. Box 3640, 76021 Karlsruhe, Germany Bruker Biospin GmbH, Silberstreifen 4, 76287 Rheinstetten, Germany § ANKA Synchrotron Radiation Facility, Karlsruhe Institute of Technology, P.O. Box 3640, 76021Karlsruhe, Germany # Helmholtz Institute Ulm for Electrochemical Energy Storage, P.O. Box 3640, 76021 Karlsruhe, Germany ∥ Institute for Applied Materials − Energy Storage Systems, Karlsruhe Institute of Technology, P.O. Box 3640, 76021 Karlsruhe, Germany ‡

ABSTRACT: LiCoPO4 was synthesized by a solid-state and a supercritical solvo-thermal method. In situ X-ray absorption near-edge structure (XANES) experiments were evaluated on the basis of full-cycle experiments confirming the predominantly reversible character of the redox reaction. In situ X-ray diffraction (XRD) measurements were performed to follow structural changes during cycling indicating the existence of an intermediate phase upon cycling. The local phosphorus and lithium environments were studied by ex situ 31P and 7Li NMR at different states of charge proving the existence of an intermediate phase of the composition Li2/3CoPO4. On the basis of these findings, a model of the (de)lithiation process of LiCoPO4 is developed and discussed.

surface films on the cathode side.31 Though LiCoPO4 has been extensively studied in recent years, very little is known about the nature of the intermediate phase LixCoPO4. A first detailed analysis of this intermediate phase was published by Bramnik et al.32 suggesting a composition of Li0.7CoPO4. These results were reconfirmed by Ehrenberg et al.33 who employed neutron and X-ray powder diffraction simultaneously to determine the Li content of the intermediate phase LixCoPO4 to be x = 0.60(10). These findings, however, are contradicted by a recent work by Ju et al. 3 4 suggesting a composition of Li0.20−0.45CoPO4. In the present work, the reversible character of the (de)lithiation process is proven by in situ X-ray absorption spectroscopy (XAS) and a model of the same is developed on the basis of ex situ 31P/7Li nuclear magnetic resonance (NMR) spectroscopy. The utilization of in situ techniques is of key importance especially when dealing with metastable phases such as the investigated CoPO4 phase which undergoes amorphization when exposed to air.32 All the ex situ samples must be dealt with great care in order to prevent any exposure to atmospheric conditions.

1. INTRODUCTION LiCoPO4 has attracted a lot of interest in recent years as a cathode material for Li-ion batteries due to its high redox potential (4.8 V vs Li/Li+), large theoretical specific capacity (167 mAh/g), and appealing energy density of about 802 Wh kg−1.1−5 The two major challenges for this material have been the intrinsically low electronic conductivity6,7 and the pronounced capacity fading upon cycling.2,8 Several different strategies have been employed in the past to enhance the cyclic stability and rate performance. Most notably, doping with metal ions,9−12 application of different synthesis procedures,13−16 and the processing of electrically conductive additives17−20 all proved to be effective methods for further optimizing the material. As reported by Markevich et al.,21 the electrolyte used for this high-voltage cathode material is of key importance and numerous studies were dealing with the modification of the same.22−29 More recently, major advancements have been made identifying the nucleophilic attack of F− anions on the P atoms in LiPF6 containing electrolytes as the underlying cause for the strong degradation of this material.21 This nucleophilic attack causes the P−O bonds of the LiCoPO4 to break, eventually leading to rapid capacity fading. Subsequent studies tried to alleviate this problem by introducing HF scavengers30 or the utilization of electrolyte additives forming protective © 2014 American Chemical Society

Received: April 3, 2014 Revised: June 16, 2014 Published: July 9, 2014 17279

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

2. EXPERIMENTAL SECTION 2.1. Synthesis. LiCoPO4 was synthesized by (a) a solidstate reaction and (b) a supercritical solvo-thermal method. The sample obtained via the solid-state reaction (labeled in the following as ss-LiCoPO4) was prepared by mixing stoichiometric amounts of (NH4)Co(PO4)·H2O and Li2CO3 which were then ground in a mortar. The powder was calcined in air for 16 h at 600 °C, reground, and again calcined in air for 8 h at 600 °C. The precursors for the supercritical solvo-thermal synthesis (sc-LiCoPO4) were Co(CH3CO2)2·4H2O, LiOOCCH3, and phosphoric acid. LiCoPO4 was prepared by mixing stoichiometric amounts of Co(CH3CO2)2·4H2O and phosphoric acid in 40 mL of ethylene glycol after which lithium acetate was added to the solution and left stirring for 30 min. The solution was then transferred to a 100 mL Hastelloy-C276 autoclave (Parr Instrument Co., IL, USA, Model 4793), which was subsequently purged with argon. The solution was then heated to 400 °C and 30 MPa and was kept under supercritical conditions for 10 min followed by quenching the autoclave. The product was obtained by filtration and repeated washing with distilled water and ethanol. Finally, the powder was calcined at 650 °C for 4 h in an argon atmosphere. 2.2. Materials Characterization. Electrochemical measurements were performed using Swagelok type cells comprising a Li metal (Goodfellow, 99.9%, 200 mm thickness) counter electrode, 2 sheets of Whatman GF/B separator, LP-30 electrolyte (BASF, EC/DMC 1:1 w/w, 1 M LiPF6), and a working electrode based on 80% LiCoPO4, 20% carbon black (Super P, Timcal). Tris(hexafluoro-iso-propyl)phosphate (HFiP) was purchased from Suzhou Fluolyte and used as an electrolyte additive in a selected case (see Figure 5). Galvanostatic cycling experiments were carried out on a VMP3 multipotentiostat (Biologic Science Instruments, France). Ex situ crystal structure analysis was performed by X-ray diffraction (XRD) analysis using a Bruker D8 advance diffractometer with Cu Kα1,2 radiation (40 kV, 40 mA) in the 2θ range 10−80° (Bragg−Brentano geometry). Structural refinements of the different materials were performed by Rietveld analysis using the FULLPROF program. In situ XAS measurements were performed using a custombuilt transmission cell composed of two aluminum plates, each with a central aperture of 14 × 2 mm. Both aluminum plates were isolated using a Kapton foil of 25 μm thickness. The cathode film was prepared by mixing the active material, carbon black, and poly(vinylidene fluoride) (PVDF, Arkema Inc.) in weight ratios of 80:10:10 and using N-methylpyrrolidone (NMP, Sigma-Aldrich) as a solvent. The slurry was then doctor-bladed at a wet thickness of 200 μm onto an Al foil and kept in the glovebox for drying. Li foil was used as the counter electrode, which was pressed onto a ring-shaped Cu current collector. LP-30 was used as the electrolyte, and the electrodes were separated by a microporous trilayer membrane (Celgard 2325). XAS measurements were performed at the XAS beamline (2.5 GeV, injection current 80−150 mA) of the ANKA Synchrotron Light Source, Karlsruhe (Germany), both in transmission and fluorescence modes. The transmitted radiation was measured by three consecutive ionization chambers, which were filled with an appropriate partial pressure of N2, He, and Ar to reach an absorption of 15% in the first ionization chamber and 40% and 60% in the second and third

chamber, respectively. Fluorescence radiation was measured by a 5-element germanium detector. The cell was located between the first and the second ionization chamber, and the energy calibration was realized by a Co reference foil located between the second and third ionization chamber. XAS scans of the Co K-edge were taken every 5 min during cell cycling at a rate of C/10, i.e., charging or discharging within 10 h. In situ XRD was performed at the PDIFF beamline at the ANKA Synchrotron Light Source, Karlsruhe, (Germany), at an energy of 16 keV (λ = 0.77490 Å) using the same transmission cell as described above. XRD powder patterns were collected within 5 min during galvanostatic cycling at a current density of 16.7 mA g−1 (i.e., C/10 with respect to the theoretical capacity of 167 mA h g−1) using a Princeton 165 mm diameter 2D-CCD detector. 7 Li and 31P magic-angle spinning (MAS) NMR was performed on a Bruker Avance 200 MHz spectrometer (B0 = 4.7 T) using 1.3 mm zirconia rotors in a dry nitrogen atmosphere. An aqueous 1 M LiCl solution was used as the reference for the chemical shift of 7Li (0 ppm). 85% H3PO4 was used as the reference for the chemical shift of 31P (0 ppm). Typical values for the recycle delay and the π/2 pulse length were 5 s and 2 μs, respectively. The experiments were performed at 298 K and a spinning speed of 60 kHz with a rotor synchronized Hahn-echo sequence (π/2-τ-π-τ-acquisition). The carbon content of the samples was determined by a CHNS elemental analyzer (Elementar Vario MICRO CUBE). The morphology of the samples was analyzed by scanning electron microscopy (SEM, LEO 1530). Transmission electron microscopy (TEM) characterization was carried out on an image-corrected FEI Titan 80-300 microscope operated at 80 kV and equipped with a Gatan Tridiem energy filter, providing an information limit of 0.08 nm in TEM mode. TEM samples were prepared by dispersing the powdered cathode material on holey carbon coated copper grids (Quantifoil GmbH).

3. RESULTS AND DISCUSSION 3.1. Physicochemical Properties. Structural refinements based on the ex situ XRD data were performed on both samples, and the results are presented in Figure 1. Both samples contain small amounts of impurities. The refinement results of ss-LiCoPO4 exhibit a phase impurity of Co2P2O7 (4.0 wt %). In contrast to that, sc-LiCoPO4 contains 9.1 wt % of Co2P. The results of the structural refinement of the LiCoPO4 phase together with other physicochemical properties are detailed in Table 1 for both materials. Figure 2 displays the particle size distribution for both materials confirming a much broader size distribution as well as a distinct increased d50 value in the case of sc-LiCoPO4. TEM brightfield images (inset: selected area electron diffraction, SAED pattern) and HRTEM micrograph (inset: fast Fourier transformation, FFT) of ss-LiCoPO4 are shown in Figure 3a,b, respectively. Both the SAED pattern and the FFT reveal reflections corresponding to the (110) LiCoPO 4 , (120)LiCoPO4, and (11̅0)Co2P2O7 in good agreement with Rietveld refinement results of Figure 1. 3.2. Electrochemical Characterization. The voltage profiles obtained during the first charge−discharge cycle of ss-LiCoPO4 and sc-LiCoPO4 are depicted in Figure 4. 17280

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

Figure 1. Rietveld refinement results of (a) ss-LiCoPO4 and (b) scLiCoPO4.

Figure 2. Particle size distribution (obtained from SEM image analysis) for (a) ss-LiCoPO4 and (b) sc-LiCoPO4.

Table 1. Physicochemical Properties of the LiCoPO4 Phase for ss-LiCoPO4 and sc-LiCoPO4

explained by the Co2P impurity phase present in sc-LiCoPO4 (see Figure 1). A similar effect was already reported by Wolfenstine et al.35 They observed that the Co2P impurity phase enhances the electronic conductivity and in smaller concentrations of up to 4% leads to an increase in capacity. However, for higher concentrations of Co2P, a drastic decline in capacity was observed caused by the electrochemically inert Co2P phase, which on the one hand improves the electronic conductivity but on the other hand tends to obstruct the Li+ insertion/ extraction. Similar observations were also recently made by Xu et al.36 The reason for the distinct content of Co2P in the case of sc-LiCoPO4 is the fact that carbon containing precursors and an organic solvent were chosen for the synthesis, which led to the formation of the Co2P phase on the surface of the particles by carbothermal reduction of LiCoPO4 during heat treatment. One important characteristic of the charging curve of ssLiCoPO4 is the distinct occurrence of two plateaus. These two plateaus however can no longer be clearly distinguished during the discharge process. In general, the two step characteristics can also be observed for the discharge cycle at low C-rates (C/ 10).16 At higher C-rates, this feature becomes less obvious and eventually the two plateaus are indistinguishable. The severe capacity fade of the solid-state sample in the subsequent cycles is commonly observed for this cathode material. Markevich et

space group a (Å) b (Å) c (Å) V (Å3) RBragg RF χ2 d50a (μm) carbon content (wt %)

solid state

supercritical

Pnma 10.2027(1) 5.9219(1) 4.6998(1) 283.960(7) 7.32 7.12 5.54 0.577 n/a

Pnma 10.2074(1) 5.9229(1) 4.7008(1) 284.209(8) 7.99 12.31 0.97 0.859 1.08

a

d50 is the median of the particle size distribution as obtained from SEM analysis.

Cycling was performed at C/20 at room temperature in the potential range of 2−5 V (vs Li+/Li). The initial specific charge capacities for ss-LiCoPO4 and sc-LiCoPO4 are 119 and 152 mAh g−1, respectively (with respect to the mass of the active material). For the subsequent discharge of ss-LiCoPO4, a specific discharge capacity of about 101 mAh g−1 is observed while sc-LiCoPO4 shows only low electrochemical activity (14 mAh g−1). This electrochemical inactivity can most likely be 17281

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

Figure 3. (a) TEM image of ss-LiCoPO4 with corresponding SAED pattern; (b) HRTEM micrograph.

al. discovered that LiCoPO4 in its charged state is prone to nucleophilic attack of F− anions (originating from the presence of HF in LiPF6 based electrolytes) on the P atoms causing a breakdown of the P−O bonds leading to the structural degradation of this material.21,37 One possible way to alleviate this problem can be realized by electrolyte additives like HFiP which form a protective surface film on the cathode side resulting in an enhanced cyclability.38,39 The improvement of the cyclability due to the addition of 1 wt % HFiP to the standard LP30 electrolyte is shown in Figure 5. 3.3. In Situ XAS. First Co K-edge X-ray absorption nearedge structure (XANES) measurements on LiCoPO4 were reported by Okada et al.40 These measurements, though only performed ex situ on three (initial, charged, and discharged) samples, already hinted qualitatively at a partially reversible redox reaction of Co. For the first time, the present work sheds further light on the reversibility of the electrochemical reaction processes occurring in the course of delithiation and relithiation by collecting in situ XAS spectra of the Co K-edge during the initial full cycle. As charging of LiCoPO4 commences, one would expect an oxidation of the Co2+ to Co3+ which would cause a shift of the XANES spectra to higher energies. The evolution of the XANES spectra for the ss-LiCoPO4 and scLiCoPO4 cell is displayed in Figure 6. Qualitatively, several observations can already be drawn from the measurements of ss-LiCoPO4. (i) Upon charging, a distinct shift of the edge position to higher energies is observable, confirming the partial oxidation of Co from 2+ to 3+. (ii) There is a notable asymmetry of the charging−discharging voltage

Figure 4. Charge−discharge profiles of LiCoPO4 synthesized by (a) solid state and (b) supercritical route.

Figure 5. Impact of 1 wt % HFiP as an electrolyte additive to LP30 on the cyclability of LiCoPO4.

profile most likely caused by the decomposition reaction of the electrolyte during the charging process. The issue of the electrolyte instability at higher voltages is worsened by the increased overpotential of the in situ cell and the concomitant 17282

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

Figure 6. (a) Voltage profile of the first full cycle combined with a PCA of the Co K-edge for ss-LiCoPO4. Corresponding XAS absorption and fluorescence spectra of the solid-state sample during (b) charge and (c) discharge. Respective plots for sc-LiCoPO4 (d), (e), and (f). The numbered markers in the voltage profiles correspond to the XAS scan numbers shown below.

generation of HF in LiPF6 based organic carbonate electrolytes.21,41 The presence of a pre-edge feature in the spectra hints at the asymmetry of the oxygen octahedron around the Co ions which is also confirmed by the XRD measurements. The XANES spectra of the sc-LiCoPO4 cell vary significantly from the ss-LiCoPO4 cell. Several points are apparent: (i) Almost no features of a XANES shift to higher energies throughout the charging process were observed most likely caused by the distinct presence of the electrochemically inert Co2P impurity phase inhibiting the Li extraction.35 (ii) Therefore, the occurrence of the plateau-like profile of the

need for a higher cutoff voltage. (iii) The pronounced overpotential in combination with the onset of the electrolyte decomposition at higher voltages causes the apparent disappearance of the two charging plateaus observed for the Swagelok based cells (Figure 4a). (iv) The initial and final (at the end of discharge) XANES spectra are not completely congruent. This is a clear indication for incomplete reversibility of the delithiation/lithiation process. Arguably, one of the most dominating factors for this capacity fade upon cycling is the structural degradation of the cathode material caused by the decomposition of the electrolyte at elevated potentials and the 17283

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

Figure 7. XRD diffraction patterns acquired in situ during the first full cycle illustrating the appearance/disappearance of the three phases “A”, “B” and “C”.

Figure 9. Phase evolution of the phases “A”, “B”, and “C” as a function of scan numbers.

The results of the PCA (Figure 6d) confirm the highly irreversible process of the charging/discharging-cycle of scLiCoPO4. This is in coincidence with the completely different change in the spectra during charging. While the absorption edge in the spectra of ss-LiCoPO4 shifts to higher energy, scLiCoPO4 shows an increase in the first oscillation of the spectrum (known as white line). 3.4. In Situ XRD. The structural changes occurring during the first full cycle were investigated by in situ XRD on ssLiCoPO4. These measurements reveal the occurrence of two two-phase regions upon charge/discharge first reported by Bramnik et al.32 A qualitative overview of the occurring phase changes during the de(lithiation) process is depicted in Figure 7. Two apparent features can be seen in the charge/discharge voltage profile of Figure 7. (i) Just like in the case of the XANES measurements, there is a distinct asymmetry between the charge and the discharge profile mainly caused by the decomposition of the electrolyte which superimposes the charging profile. (ii) The two well-defined plateaus seen in the charging profile of Figure 4a can no longer be recognized due to the increased overpotential of the in situ cell effectively masking the two charging plateaus. As charging (delithiation) of LiCoPO4 (indicated as “A”) commences, the formation of another phase “B” can be clearly identified. At this point, one can only make the statement that this intermediate phase “B” most likely represents a partly delithiated form of lithium cobalt phosphate. At the later stages of the charging step, the reflections of this partly delithiated phase LixCoPO4 are reduced in intensity while a third phase “C” is formed. During the discharge step, these phase transitions are qualitatively reversed. At the end of the

Figure 8. Residual value RBragg and χ2 as a measure for the quality of the Rietveld refinement as a function of the Li occupation in the intermediate phase.33

supercritical sample has to be attributed to the electrolyte decomposition. (iii) The lack of Co oxidation during the charging of sc-LiCoPO4 brings about the absence of any reduction reaction during the discharge process. To gain a more quantitative measure for the redox-reaction occurring upon cycling, principle component analysis (PCA)42 was performed on all XANES spectra which provides deeper knowledge about the degree of reversibility. In order to perform a PCA, a background subtraction, interpolation, and a normalization of the measured spectra have to be done with extremely high reproducibility. The code is highly efficient and can therefore also be used for full field spectroscopic imaging.43 The asymmetric charging profile of ss-LiCoPO4 is even more obvious in the 80% recovery of the first component of the PCA (Figure 6a). The results based on the fluorescence and absorption data (blue and green) are nearly identical, which excludes any depth dependent reaction activities. An additional component of the PCA contains only changes within ±50 eV around the absorption edge and demonstrates an overall strongly nonreversible behavior but also two different regimes during charging. The latter effect might hint at the two reaction steps also observed by XRD and NMR (see below). 17284

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

Figure 11. Changes of unit cell volumes during (de)lithiation.

The nature of this intermediate phase is discussed controversially in the literature. Bramnik et al.32 reported the existence of Li0.70CoPO4 as an intermediate phase based on in situ XRD while Ju et al. 34 stated a composition of Li0.20−0.45CoPO4 as the intermediate phase. In the latter case, the lithium content was solely derived from electrochemical data, which can be very misleading since a non-negligible amount of charge is consumed by side reactions (see Figure 6). Therefore, it is not surprising that in this case the intermediate phase is believed to have a relatively low lithium content of 0.2−0.45. In another study, Ehrenberg et al. employed neutron and X-ray powder diffraction to uncover a lithiation state of x = 0.60(10) for the intermediate phase LixCoPO433 from the optimum agreement between observed and calculated profiles; see Figure 8. The phase evolution of the phases “A”,“B”, and “C” was evaluated by Rietveld refinement with the program FULLPROF. Thompson-Cox-Hastings Pseudo-Voigt was used as the peak shape function. Scale factors, lattice parameters (a, b, c), atomic positions (Co, P, O), overall isotropic displacement factor, and Lorentzian/Gaussian size and strain parameters were free parameters. The results are depicted in Figure 9. It is important to point out that the statistics of the diffraction patterns worsen toward the refilling of the synchrotron ring (“beam dump”), which is the cause for the increase in the standard deviation. At the initial stage of the charging step, no phase transformation can be observed until scan 14. This apparent inactivity can most likely be explained by the initial onset of parasitic side reactions under the highly oxidizing conditions. Subsequently, the formation of phase “B” commences. The phase “C” can be evidenced at scan number 50 which continues to gain in intensity until scan 69. A clear indication for a phase transformation of the nature “A” ↔ “B” ↔ “C” can be drawn from the discharge region. At the beginning of the discharge step, phase “A” stays almost constant while a decrease of phase “C” is accompanied by an increase of phase “B” which hints at an exclusive phase transition between “B” and “C”. After the beam dump, phase “C” can no longer be evidenced and the phase transition between “B” and “A” is progressing. It should be pointed out that the observed coexistence of three phases at the end of the charge step and the beginning of the discharge step is in violation of Gibbs’

Figure 10. Exemplary Rietveld refinements at different states of charge: (a) initial, (b) end of charge, and (c) end of discharge.

discharge step, however, the intermediate phase LixCoPO4 is still present in a small amount. The most likely explanation for this is the incomplete relithiation of the sample due to the slow kinetics. In the course of the first full cycle, no signs of an amorphization of the phases can be found. This was confirmed by comparing the integrated intensities of the initial/final patterns after correction for beam intensity. 17285

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

Table 2. Unit Cell Parameters of the Three Investigated Phases phase “A” phase “B” phase “C”

space group

a (Å)

b (Å)

c (Å)

V (Å3)

Pnma Pnma Pnma

10.2005(3) 10.0860(14) 9.637(5)

5.9213(2) 5.8585(6) 5.813(3)

4.6995(2) 4.7187(9) 4.738 (2)

283.85(2) 278.82(7) 265.5(3)

cally44−47 but vary severely among themselves (261 Å3 45 compared to 275.8254 Å3 46). Phase “C” is identified to be CoPO4 which has a unit cell volume similar to the 264.50 Å3 previously reported.33 On the basis of the assumption of a linear correlation between the lithium content and the unit cell volume of the olivine structure, a rough estimate of the lithium content of phase “B” can be drawn from Figure 12. On the basis of Figure 12, the lithiation state of the intermediate phase can be estimated to be in the range between 0.68 and 0.82, in agreement with results from neutron powder diffraction.33 In accordance with the pronounced irreversible and limited (discharge) capacity of the in situ cell, the CoPO4 phase is only detectable in a small amount (17.7%) in the fully charged state (Figure 9). A recently published theoretical work by Osnis et al.45 suggests that the stability of LiCoPO4 upon delithiation is by far lower compared to that of LiFePO4. The reaction enthalpies of the two reactions differ by −128.79 kJ/ mol.45

Figure 12. Graphical estimation of the lithium content of phase “B” determined from the value of the unit cell volume. The range of the unit cell volumes corresponds to the scattering of all values obtained from the Rietveld refinement.

Li + FePO4 → LiFePO4

(1)

Li + CoPO4 → LiCoPO4

(2)

The energetic unfavorable formation of the CoPO4 phase could in turn affect the formation/existence of the investigated intermediate phase. The thermal stability of LiCoPO4 at different states of charge were analyzed by Bramnik et al.48 confirming a surprisingly low stability of the partly and fully delithiated compounds. As charging progresses, the thermal stability decreases further which indicates an even lower thermal stability of CoPO4 compared to the intermediate phase LixCoPO4. 3.5. Ex Situ 7Li and 31P NMR Spectroscopy. Both 7Li and 31 P NMR spectroscopy was performed on ss-LiCoPO4 with different lithiation states. The results are shown in Figure 13. The 7Li NMR spectrum of as prepared ss-LiCoPO4 is characterized by a broad peak at about −105 ppm (labeled with “A”). During charging, this peak successively disappears and two new peaks at about 80 ppm (“B”) and −135 ppm (“C”) appear. The ratio of the intensities of these two peaks is 1:1 (including the spinning sidebands that are not shown in this picture). During further charging, the peaks “B” and “C” disappear again and the spectrum obtained after full charge is characterized by a structureless, flat line confirming the removal of most of the lithium. Thus, these spectra reveal an intermediate phase with two nonequivalent Li environments. During discharging, the phase evolution is reversed. In the beginning, the peaks “B” and “C” are reformed, and at later stages, these peaks disappear again and are replaced by peak “A”. The spectra of the completely discharged sample and of the as-prepared sample show a high degree of congruence. A very similar behavior is observed in the 31P NMR spectra. The initial spectrum is characterized by a single peak “A” at about 3000 ppm. In the course of the charging step, an intermediate phase becomes apparent as can be seen by the formation of two new peaks at about 2620 ppm (“B”) and 2230 ppm (“C”). In this case, the peak ratio is 2:1 representing two

phase rule. Therefore, the coexistence of these three phases has to be ascribed to kinetic restraints. The reported weight percent concentrations for the individual phases “A”, “B”, and “C” in Figure 9 must be treated with some caution since different absorption coefficients and/or preferred orientation of the individual phases were not included in the model. Comparing the electrochemical data of the in situ cell from Figure 9 with the electrochemical data obtained from the Swagelok cell (Figure 4a), it becomes apparent that the discharge capacities show strong differences. The discharge capacity of the Swagelok cell reaches 101.3 mAh g−1 which is strikingly larger than the 46.8 mAh g−1 obtained in the case of the in situ cell. The charge capacities on the other hand are somewhat comparable: 119.7 mAh g−1 (Swagelok) and 96.2 mAh g−1 (in situ). In this context, it is important to recall that these values are obtained by employing different boundary conditions. The Swagelok based cell is cycled at C/20 with potential limitations of 5 and 2.5 V while the in situ cell is set up to cycle at C/10 in the voltage range of 5.2 to 2 V (accounting for the increased overpotential of the cell). The pronounced irreversible capacity of the in situ cell can be explained by the extended voltage window which promotes parasitic side reactions. Selected refinement results and the changes in the unit cell volumes of the individual phases during the cycling can be found in Figures 10 and 11. Toward the end of the discharge step, the unit cell volume of phase “B” markedly increases (Fig 11). This is a trend which was also reported by Bramnik et al.32 and was discussed in terms of the possible existence of a solid solution region. Phase “A” represents the fully lithiated compound LiCoPO4, and the lattice parameters are in good agreement with the values obtained from the Bruker D8 advance diffractometer (Tables 1,2). Values for the unit cell volume of CoPO4 have been determined both experimentally32,33 and theoreti17286

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

Figure 13. 7Li and 31P NMR spectra of ss-LiCoPO4 at different states of charge. The sidebands of peaks “A”, “B”, and “D” in the 31P NMR spectra are marked by ∗, +, and ×, respectively.

nonequivalent P environments with a corresponding ratio of occupancies. At the end of the charge, the 31P NMR spectrum is dominated by a new peak at 3210 ppm (labeled “D”). The large 31P NMR shift in the charged state hints at a high-spin state of Co3+ as it was also observed earlier by magnetic measurements.33 Just like in the case of the 7Li NMR spectra, the intermediate phase is then being reformed upon discharge and vanishes again for the completely discharged sample. Both the 7Li and the 31P NMR spectra show a good agreement between the spectra of the as-prepard sample and the sample obtained after one full cycle. In particular, no peak broadening is visible. This hints at a preservation of the crystal structure, and in accordance with our XRD findings, an amorphization during cycling can be ruled out. On the basis of the 31P NMR findings, the phase evolution during the (de)lithiation process is displayed in Figure 14. It should be pointed out that the “x” in Figure 14 should not be taken literally since it merely indicates the progress of the (dis-)charging step. A comparison of the phase evolution based on NMR and in situ XRD (Figure 9) is worthwhile which are consistent in many points:

Figure 14. Phase evolution based on 31P NMR results. 17287

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

Figure 15. Proposed charging/discharging mechanism of LiCoPO4.

(i) For the first two 31P NMR spectra, phase “B” cannot be evidenced. This is in agreement with the observation that phase “B” was only detectable after the 14th scan of the in situ XRD experiment. (ii) The fully delithiated phase is only formed in the final stages of the delithiation process at “x” = 0.25. This trend is also reflected by the in situ XRD measurements. (iii) Finally, the phase transformation of the nature “A” ↔ “B” ↔ “C” suggested on the basis of the XRD measurements is also supported by the NMR results: In the first two NMR spectra of the discharge step, the LiCoPO4 phase concentration remains almost constant while a decrease in the CoPO4 phase concentration is accompanied by a rise in the phase “B” concentration. This is exactly the same behavior as observed in the in situ XRD measurements. The fact that we observe two nonequivalent Li environments with a peak ratio of 1:1 and two nonequivalent P environments with a peak ratio of 2:1 for the intermediate phase suggests a lithiation state of 2/3. This value is in agreement with the estimated lithiation state based on our XRD measurements (Figure 12). According to these findings and the onedimensional character of lithium diffusion in this material class,49 we propose the following (de)lithiation mechanism for LiCoPO4 (Figure 15): The structure of the as-prepared material LiCoPO4 contains a single Li site and a single P site. Therefore, only one peak was observed in both the 7Li and the 31 P NMR spectrum. As charging commences, LiCoPO4 is locally selectively delithiated forming the intermediate phase Li2/3CoPO4 where one-third of the Li ions is removed from the structure and correspondingly one-third of the Co2+ is oxidized to Co3+ (dark green and light green octahedra). Different arrangements of the Li vacancies are possible. In Figure 15, we depicted an arrangement where complete channels along the b axis are emptied in a way that alternating layers along the c direction are formed with full or empty Li sites in a 2:1 stacking sequence. Since the Co is partially oxidized, now two different Li environments and two different P environments are present, although the Li and P ions are still located on the same sites. There are P ions with direct Co3+ neighbors (light purple tetrahedra) and P ions without Co3+ neighbors (dark purple tetrahedra), and the relative occupancy is 2:1. For the Li, there are also environments with and without Co3+ neighbors (light blue and dark blue spheres), but since here part of the Li is removed, the relative occupancy of these two environments is 1:1. These ratios fit exactly to the relative intensities in the 7Li and 31P NMR spectra. At the later stages of the charging process, this intermediate phase then starts to be further delithiated forming finally CoPO4. Again, this compound contains only a single P site (and no Li). Since the number

of Co3+ neighbors is different from the sites in the LiCoPO4 and Li2/3CoPO4, the position of the 31P NMR peak is also different for CoPO4 in comparison to the NMR peaks of the other phases, although the P is always located on the same site (neglecting the small changes in lattice constants and thus bond lengths/angles). These outlined phase transformations differ fundamentally from the established “domino-cascade” model in the case of LiFePO4 in which only the end members are stable and partly lithiated compounds are highly metastable at room temperature.50−54

4. CONCLUSION In situ XRD, ex situ 7Li/31P NMR spectroscopy, and to some extent in situ XAS, all reveal the existence of two two-phase redox reaction steps upon the (de)lithiation process of LiCoPO4. For the first time, the lithiation state of the intermediate phase is determined by ex situ 7Li and 31P NMR spectroscopy to be Li2/3CoPO4 which is in good agreement with our and previous XRD and neutron diffraction measurements. On the basis of these findings, a model for the charge/ discharge process is proposed which differs greatly from the well-established “domino-cascade” model for LiFePO4. The existence of this intermediate phase Li2/3CoPO4 might be explained by recent theoretical findings which suggest that the formation of CoPO4 is energetically unfavorable compared to the system of LiFePO4.

■

AUTHOR INFORMATION

Corresponding Author

*Telephone: +49 721 608-28508. Fax: +49 721 608-28521. Email: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We are grateful to the German Federal Ministry of Education and Research for financial support. Sample characterization was partly carried out with the support of the Karlsruhe Nano Micro Facility, a Helmholtz Research Infrastructure at Karlsruhe Institute of Technology.

■

REFERENCES

(1) Amine, K.; Yasuda, H.; Yamachi, M. Olivine LiCoPO4 as 4.8 V Electrode Material for Lithium Batteries. Electrochem. Solid-State Lett. 2000, 3, 178−179. (2) Bramnik, N.; Bramnik, K.; Buhrmester, T.; Baehtz, C.; Ehrenberg, H.; Fuess, H. Electrochemical and Structural Study of LiCoPO4-Based Electrodes. J. Solid State Electrochem. 2004, 8, 558−564. 17288

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

(3) Kumar, P. R.; Venkateswarlu, M.; Misra, M.; Mohanty, A. K.; Satyanayana, N. Synthesis, Characterization and Electrical Properties of Carbon Coated LiCoPO4 Nanoparticles. J. Nanosci. Nanotechnol. 2011, 11, 3314−3322. (4) Huang, X.; Ma, J.; Wu, P.; Hu, Y.; Dai, J.; Zhu, Z.; Chen, H.; Wang, H. Hydrothermal Synthesis of LiCoPO4 Cathode Materials for Rechargeable Lithium Ion Batteries. Mater. Lett. 2005, 59, 578−582. (5) Doan, T. N. L.; Taniguchi, I. Preparation of LiCoPO4/C Nanocomposite Cathode of Lithium Batteries with High Rate Performance. J. Power Sources 2011, 196, 5679−5684. (6) Wolfenstine, J.; Lee, U.; Poese, B.; Allen, J. L. Effect of Oxygen Partial Pressure on the Discharge Capacity of LiCoPO4. J. Power Sources 2005, 144, 226−230. (7) Wolfenstine, J. Electrical Conductivity of Doped LiCoPO4. J. Power Sources 2006, 158, 1431−1435. (8) Rabanal, M. E.; Gutierrez, M. C.; Garcia-Alvarado, F.; Gonzalo, E. C.; Arroyo-de Dompablo, M. E. Improved Electrode Characteristics of Olivine−LiCoPO4 Processed by High Energy Milling. J. Power Sources 2006, 160, 523−528. (9) Muraliganth, T.; Manthiram, A. Understanding the Shifts in the Redox Potentials of Olivine LiM1−yMyPO4 (M = Fe, Mn, Co, and Mg) Solid Solution Cathodes. J. Phys. Chem. C 2010, 4, 15530−15540. (10) Wang, F.; Yang, J.; NuLi, Y.; Wang, J. Highly Promoted Electrochemical Performance of 5V LiCoPO4 Cathode Material by Addition of Vanadium. J. Power Sources 2010, 195, 6884−6887. (11) Kishore, M. V. V. M. S.; Varadaraju, U. V. Influence of Isovalent Ion Substitution on the Electrochemical Performance of LiCoPO4. Mater. Res. Bull. 2005, 40, 1705−1712. (12) Han, D.-W.; Kang, Y.-M.; Yin, R.-Z.; Song, M.-S.; Kwon, H.-S. Effects of Fe Doping on the Electrochemical Performance of LiCoPO4/C Composites for High Power-Density Cathode Materials. Electrochem. Commun. 2009, 11, 137−140. (13) Lloris, J. M.; Pérez Vicente, C.; Tirado, J. L. Improvement of the Electrochemical Performance of LiCoPO4 5 V Material Using a Novel Synthesis Procedure. Electrochem. Solid-State Lett. 2002, 5, A234− A237. (14) Koleva, V.; Zhecheva, E.; Stoyanova, R. Ordered Olivine-Type Lithium-Cobalt and Lithium-Nickel Phosphates Prepared by a New Precursor Method. Eur. J. Inorg. Chem. 2010, 2010, 4091−4099. (15) Wang, F.; Yang, J.; NuLi, Y.; Wang, J. Novel Hedgehog-like 5V LiCoPO4 Positive Electrode Material for Rechargeable Lithium Battery. J. Power Sources 2011, 196, 4806−4810. (16) Liu, J.; Conry, T. E.; Song, X.; Yang, L.; Doeff, M. M.; Richardson, T. J. Spherical Nanoporous LiCoPO4/C Composites as High Performance Cathode Materials for Rechargeable Lithium-Ion Batteries. J. Mater. Chem. 2011, 21, 9984−9987. (17) Eftekhari, A. Surface Modification of Thin-Film Based LiCoPO4 5 V Cathode with Metal Oxide. J. Electrochem. Soc. 2004, 151, A1456− A1460. (18) Sun, Q.; Luo, J.-Y.; Fu, Z.-W. Facile Synthesis and Electrochemical Properties of Carbon-Coated LiCoPO4 Submicron Particles as Positive Materials for Lithium Ion Batteries. Electrochem. Solid-State Lett. 2011, 14, A151−A153. (19) Li, H. H.; Jin, J.; Wei, J. P.; Zhou, Z.; Yan, J. Fast Synthesis of Core-Shell LiCoPO4/C Nanocomposite via Microwave Heating and Its Electrochemical Li Intercalation Performances. Electrochem. Commun. 2009, 11, 95−98. (20) Jin, B.; Gu, H.-B.; Kim, K.-W. Effect of Different Conductive Additives on Charge/Discharge Properties of LiCoPO4/Li Batteries. J. Solid State Electrochem. 2007, 12, 105−111. (21) Markevich, E.; Sharabi, R.; Gottlieb, H.; Borgel, V.; Fridman, K.; Salitra, G.; Aurbach, D.; Semrau, G.; Schmidt, M. A.; Schall, N.; et al. Reasons for Capacity Fading of LiCoPO 4 Cathodes in LiPF6 Containing Electrolyte Solutions. Electrochem. Commun. 2012, 15, 22−25. (22) Abe, K.; Ushigoe, Y.; Yoshitake, H.; Yoshio, M. Functional Electrolytes: Novel Type Additives for Cathode Materials, Providing High Cycleability Performance. J. Power Sources 2006, 153, 328−335.

(23) Nagahama, M.; Hasegawa, N.; Okada, S. High Voltage Performances of Li2NiPO4F Cathode with Dinitrile-Based Electrolytes. J. Electrochem. Soc. 2010, 157, A748−A752. (24) Tarascon, J. M.; Guyomard, D. New Electrolyte Compositions Stable over the 0 to 5 V Voltage Range and Compatible with the Lil+xMn2O4/Carbon Li-Ion Cells. Solid State Ionics 1994, 69, 293−305. (25) Abu-Lebdeh, Y.; Davidson, I. High-Voltage Electrolytes Based on Adiponitrile for Li-Ion Batteries. J. Electrochem. Soc. 2009, 156, A60−A65. (26) Yang, L.; Markmaitree, T.; Lucht, B. L. Inorganic Additives for Passivation of High Voltage Cathode Materials. J. Power Sources 2011, 196, 2251−2254. (27) Kim, K.; Cho, Y.-H.; Kam, D.; Kim, H.-S.; Lee, J.-W. Effects of Organic Acids as Reducing Agents in the Synthesis of LiFePO4. J. Alloys Compd. 2010, 504, 166−170. (28) Srinivasan, V.; Newman, J. Existence of Path-Dependence in the LiFePO4 Electrode. Electrochem. Solid-State Lett. 2006, 9, A110−A114. (29) Choi, J.-A.; Eo, S.-M.; MacFarlane, D. R.; Forsyth, M.; Cha, E.; Kim, D.-W. Effect of Organic Additives on the Cycling Performances of Lithium Metal Polymer Cells. J. Power Sources 2008, 178, 832−836. (30) Sharabi, R.; Markevich, E.; Borgel, V.; Salitra, G.; Aurbach, D.; Semrau, G.; Schmidt, M. A.; Schall, N.; Stinner, C. Significantly Improved Cycling Performance of LiCoPO4 Cathodes. Electrochem. Commun. 2011, 13, 800−802. (31) Sharabi, R.; Markevich, E.; Fridman, K.; Gershinsky, G.; Salitra, G.; Aurbach, D.; Semrau, G.; Schmidt, M. A.; Schall, N.; Bruenig, C. Electrolyte Solution for the Improved Cycling Performance of LiCoPO4/C Composite Cathodes. Electrochem. Commun. 2013, 28, 20−23. (32) Bramnik, N. N.; Nikolowski, K.; Baehtz, C.; Bramnik, K. G.; Ehrenberg, H. Phase Transitions Occurring upon Lithium InsertionExtraction of LiCoPO4. Chem. Mater. 2007, 19, 908−915. (33) Ehrenberg, H.; Bramnik, N. N.; Senyshyn, A.; Fuess, H. Crystal and Magnetic Structures of Electrochemically Delithiated Li1−xCoPO4 Phases. Solid State Sci. 2009, 11, 18−23. (34) Ju, H.; Wu, J.; Xu, Y. Lithium Ion Intercalation Mechanism for LiCoPO4 Electrode. Int. J. Energy Environ. Eng. 2013, 4, 2−7. (35) Wolfenstine, J.; Read, J.; Allen, J. L. Effect of Carbon on the Electronic Conductivity and Discharge Capacity LiCoPO4. J. Power Sources 2007, 163, 1070−1073. (36) Ou, X.; Xu, S.; Liang, G.; Wang, L.; Zhao, X. Effect of Fe(III) Impurity on the Electrochemical Performance of LiFePO4 Prepared by Hydrothermal Process. Sci. China Ser. E Technol. Sci. 2009, 52, 264− 268. (37) Sharabi, R.; Markevich, E.; Borgel, V.; Salitra, G.; Gershinsky, G.; Aurbach, D.; Semrau, G.; Schmidt, M. A.; Schall, N.; Stinner, C. Raman Study of Structural Stability of LiCoPO4 Cathodes in LiPF6 Containing Electrolytes. J. Power Sources 2012, 203, 109−114. (38) Von Cresce, A.; Xu, K. Electrolyte Additive in Support of 5 V Li Ion Chemistry. J. Electrochem. Soc. 2011, 158, A337−A342. (39) Allen, J. L.; Jow, T. R.; Wolfenstine, J. Improved Cycle Life of Fe-Substituted LiCoPO4. J. Power Sources 2011, 196, 8656−8661. (40) Okada, S.; Sawa, S.-I.; Uebou, Y.; Egashira, M.; Yamaki, J.-I.; Tabuchi, M.; Kobayashi, H.; Fukumi, K.; Kageyama, H. ChargeDischarge Mechanism of LiCoPO4 Cathode for Rechargeable Lithium Batteries. Electrochemistry 2003, 71, 1136−1138. (41) Lux, S. F.; Lucas, I. T.; Pollak, E.; Passerini, S.; Winter, M.; Kostecki, R. The Mechanism of HF Formation in LiPF6 Based Organic Carbonate Electrolytes. Electrochem. Commun. 2012, 14, 47−50. (42) Malinowski, E. R. Factor Analysis in Chemistry, 3rd ed.; Wiley: New York, 2002. (43) Mangold, S.; Steininger, R.; Rolo, T. D. S.; Göttlicher, J. Full Field Spectroscopic Imaging at the ANKA-XAS- and -SUL-XBeamlines. J. Phys. Conf. Ser. 2013, 430, 1−5. (44) Ling, C.; Banerjee, D.; Song, W.; Zhang, M.; Matsui, M. FirstPrinciples Study of the Magnesiation of Olivines: Redox Reaction Mechanism, Electrochemical and Thermodynamic Properties. J. Mater. Chem. 2012, 22, 13517−13523. 17289

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290

The Journal of Physical Chemistry C

Article

(45) Osnis, A.; Kosa, M.; Aurbach, D.; Major, D. T. Systematic FirstPrinciples Investigation of Mixed Transition Metal Olivine Phosphates. J. Phys. Chem. C 2013, 117, 17919−17926. (46) Osorioguillen, J. A Theoretical Study of Olivine LiMPO4 Cathodes. Solid State Ionics 2004, 167, 221−227. (47) Bacq, O.; Pasturel, A.; Bengone, O. Impact on Electronic Correlations on the Structural Stability, Magnetism, and Voltage of LiCoPO4 Battery. Phys. Rev. B 2004, 69, 245107. (48) Bramnik, N. N.; Nikolowski, K.; Trots, D. M.; Ehrenberg, H. Thermal Stability of LiCoPO4 Cathodes. Electrochem. Solid-State Lett. 2008, 11, A89−A93. (49) Morgan, D.; Van der Ven, A.; Ceder, G. Li Conductivity in LixMPO4 (M = Mn, Fe, Co, Ni) Olivine Materials. Electrochem. SolidState Lett. 2004, 7, A30−A32. (50) Orikasa, Y.; Maeda, T.; Koyama, Y.; Murayama, H.; Fukuda, K.; Tanida, H.; Arai, H.; Matsubara, E.; Uchimoto, Y.; Ogumi, Z. Direct Observation of a Metastable Crystal Phase of LixFePO4 under Electrochemical Phase Transition. J. Am. Chem. Soc. 2013, 135, 5497−5500. (51) Delmas, C.; Maccario, M.; Croguennec, L.; Le Cras, F.; Weill, F. Lithium Deintercalation in LiFePO4 Nanoparticles via a DominoCascade Model. Nat. Mater. 2008, 7, 665−671. (52) Sugar, J. D.; El Gabaly, F.; Chueh, W. C.; Fenton, K. R.; Tyliszczak, T.; Kotula, P. G.; Bartelt, N. C. High-Resolution Chemical Analysis on Cycled LiFePO4 Battery Electrodes Using Energy-Filtered Transmission Electron Microscopy. J. Power Sources 2014, 246, 512− 521. (53) Chueh, W. C.; El Gabaly, F.; Sugar, J. D.; Bartelt, N. C.; McDaniel, A. H.; Fenton, K. R.; Zavadil, K. R.; Tyliszczak, T.; Lai, W.; McCarty, K. F. Intercalation Pathway in Many-Particle LiFePO4 Electrode Revealed by Nanoscale State-of-Charge Mapping. Nano Lett. 2013, 13, 866−872. (54) Brunetti, G.; Robert, D.; Rouvi, J. L.; Rauch, E. F.; Martin, J. F.; Colin, J. F. Confirmation of the Domino-Cascade Model by LiFePO4/ FePO4 Precession Electron Diffraction. Chem. Mater. 2011, 23, 4515− 4524.

17290

dx.doi.org/10.1021/jp503306v | J. Phys. Chem. C 2014, 118, 17279−17290