C as cathode materials for lithium-ion batteries

C as cathode materials for lithium-ion batteries

Solid State Ionics 317 (2018) 149–155 Contents lists available at ScienceDirect Solid State Ionics journal homepage: www.elsevier.com/locate/ssi Li...

3MB Sizes 0 Downloads 21 Views

Solid State Ionics 317 (2018) 149–155

Contents lists available at ScienceDirect

Solid State Ionics journal homepage: www.elsevier.com/locate/ssi

LiFe1-XMgXPO4/C as cathode materials for lithium-ion batteries a

b,⁎

a

T a

Sergey Yaroslavtsev , Svetlana Novikova , Vyacheslav Rusakov , Nikita Vostrov , Tatyana Kulovac, Alexander Skundinc, Andrey Yaroslavtsevb a b c

Lomonosov Moscow State University, 119991 Leninsky gory 1, Moscow, Russia Kurnakov Institute of General and Inorganic Chemistry, Russian Academy of Sciences, 119991, Leninsky pr. 31, Moscow, Russia Frumkin Institute of Physical Chemistry and Electrochemistry, Russian Academy of Sciences, Leninsky pr. 31-4, 119071, Moscow, Russia

A R T I C L E I N F O

A B S T R A C T

Keywords: LIBs LiFePO4 Olivine Mg-doping Mössbauer spectroscopy

A cathode nanomaterial LiFe0.8Mg0.2PO4 with olivine structure was synthesized by the sol-gel method and studied using X-ray diffraction analysis, scanning electron microscopy, Mössbauer spectroscopy and electrochemical testing under the operating conditions of a lithium-ion battery. It is demonstrated that the iron substitution with magnesium occurs in the studied material. Discharge capacity of LiFe0.8Mg0.2PO4/С is 127 mAh g−1 at a current of 20 mA g−1 and is close to the theoretical value for the considered composition. It is determined utilizing Mössbauer spectroscopy that at the early stage of the LiFe0.8Mg0.2PO4 charging process nanoscale regions are formed, having a FePO4 structure and an enhanced solubility of divalent iron ions.

1. Introduction Olivine-type LiFePO4 is a promising cathode material for lithiumion batteries (LIBs) due to its low cost, safety, low toxicity, high cycling stability and comparatively high theoretical capacity (170 mAh g−1) [1–4]. However, this material has certain drawbacks, such as low conductivity, which hinders its possible application in devices requiring a high peak power [5–7]. Among the efficient methods of LiFePO4 modification, which make it possible to overcome the limitations associated with its low electrical conductivity, the formation of composites with carbon [8–12], preparation of nanoscale materials [3,9,13–17], as well as partial substitution of iron sites with other ions [18–25] should be outlined. A combination of the above approaches to LiFePO4 modification makes it possible to achieve the discharge capacity close to the theoretical value (170 mAh g−1) at a low charge/discharge rate and to markedly enhance its capacity at high charge/discharge rates. LiFePO4 doping is performed mostly to increase the bulk conductivity of the material [26–29]. In some cases, such an approach allows to increase charge and discharge current densities, while retaining a comparatively high capacity [19, 21,22,25,30–36]. The best-studied and most efficient option is doping with divalent cations. The majority of studies are aimed at investigating the materials based on LiFePO4 doped with transition metal cations [19,22,30,32–34,36]. For example, the substitution of 10% of iron ions with manganese [22] or nickel [19] was shown to enhance the discharge capacity values, especially at high



charge/discharge rates. Investigation of lithium iron phosphate-based materials doped with magnesium demonstrated that upon magnesium insertion conductivity of LiFePO4 [27] and its electrochemical performance [31,35,37–41] can be improved. Since Mg2+ is not electrochemically active, high capacity can be achieved only when its content is low. For example, a study of LiFe1-XMgXPO4/C (x = 0.00, 0.02, 0.04, 0.06, 0.08) samples proved the composition with 4% Mg to be the optimal one [35]. At the same time, literature data on sites in the LiFePO4 structure that are occupied by magnesium remains contradictory. According to some published data [28,35,37,38], magnesium ions substitutes for iron affording LiFe1-XMgXPO4 materials, while the other data [27,39,40] suggests that the insertion of magnesium introduction occurs into lithium sites resulting in Li1-XMgXFePO4 formation. Formation of both materials where either iron sites or lithium sites are occupied by magnesium, has also been reported and proved by Rietveld method [31,41]. It was hypothesized that mix-site doping in LiFePO4 is unfavorable [41]. According to [31,41], a more pronounced improvement of conductivity and electrochemical performance is attained when magnesium occupies iron sites. On the other hand, other authors [19] assumed that, even at stoichiometric loading of precursors, magnesium occupies both sites, owing to relatively equal ionic radii. Further information required to understand the mechanisms of the processes associated with lithium deintercalation in compounds based on LiFePO4, including the doped ones, can be obtained from Mössbauer spectroscopy data. An dependence of the temperature of mutual

Corresponding author. E-mail address: [email protected] (S. Novikova).

https://doi.org/10.1016/j.ssi.2018.01.011 Received 13 September 2017; Received in revised form 13 December 2017; Accepted 9 January 2018 0167-2738/ © 2018 Elsevier B.V. All rights reserved.

Solid State Ionics 317 (2018) 149–155

S. Yaroslavtsev et al.

using α-Fe foil as a reference sample. Low temperature spectra have been recorded by cooling the sample with the help of the helium cryostat SHI-850-5 from JANIS RESEARCH. Processing of Mössbauer spectra was carried out within the framework of the Hamilton model, allowing for combined magnetic dipole and electric quadrupole interactions. The model fitting of spectra was used to process the Fe2+ contribution, and the reconstruction of hyperfine parameter distribution to process the Fe3+ contribution. Alongside this, a subspectrum corresponding to Fe3+ ions, that have four other Fe3+ ions in the nearest environment, was searched for using the parameters are known from the earlier investigations of non-doped lithium iron phosphate FePO4 [19,45].

dissolution of LiFePO4 and FePO4 phases on lithiation degree of the sample has been determined [42]. Mössbauer studies at low temperatures are more sensitive to local structural features. Indeed, investigation of LiFePO4 based materials doped with Co2+, Ni2+, Mn2+ cations [19,22,43–45] revealed the nature of dopant cation ordering in the iron sites [19,22,45]. It was concluded that uniform dopant distribution in LixFe1-yCoyPO4 [19] or LixFe1-yMnyPO4 [22] samples takes place when iron ion has no more than one dopant cation in the nearest environment. In the case of LiFePO4 doping with nickel the ordering effect is less pronounced [19]. In the meantime, such studies have not been reported for lithium iron phosphate doped with magnesium. The aim of this work was to obtain new data on the influence of LiFePO4 doping with magnesium on the charge and discharge processes in the studied material. The choice of the sample with 20% iron substituted with magnesium is based on a desire to have a sufficiently strong impact on the lattice and, at the same time, to collect highquality Mössbauer spectra by retaining a high iron concentration.

3. Results and discussion 3.1. Structure and morphology X-ray diffraction patterns of LFP and LFP-Mg are indexed in the orthorhombic modification of LiFePO4 (space group Pnma, Card № 811173 PDF-2), which indicates that the obtained material has olivinetype structure. For the LFP-Mg sample a shift of diffraction peaks to the high-angle region (Fig. 1(1,2)) and shrinkage of lattice parameters, compared to the non-doped samples, are observed (Table 1), which is in good agreement with the previously reported data [35,38–40]. This indicates that magnesium cations, having a smaller radius than iron or lithium (0.78 Å for Fe2+, 0.76 Å for Li+ and 0.72 Å for Mg2+) [46], are introduced into the LiFePO4 structure. By comparing diffraction patterns of materials with different degree of charge it can be concluded that two phases are present in a material with an intermediate degree of charge, corresponding to LiFePO4 and FePO4 (Fig. 2). Rietveld refinement resulted in low R-factor values (χ2 = 0.422, Rp = 6.56, Rwp = 8.43) only if we assume that magnesium is introduced into the iron sites, affording LiFe0.8Mg0.2PO4/C. A refinement based on the assumption that magnesium is partially or fully doped into lithium sites resulted in a sharp increase of R-factors and a poorer concurrence of experimental and calculated data. Consequently, it can be concluded that magnesium occupies iron sites in the obtained material. According to scanning electron microscopy data, the particle size distribution of LFP is rather narrow; the mean diameter is ~50 nm [9]. The LFP-Mg sample is characterized by a wider size distribution (from 50 to 300 nm, (Fig. 3)). The mean size of the particles (~100 nm) is

2. Experimental Synthesis of the materials based on LiFePO4 (LFP) and LFP doped with magnesium LiFe0.8Mg0.2PO4/C (LFP-Mg) was performed by sol–gel method according to a previously reported procedure [7]. Doping with Mg2+ ions was carried out by adding a calculated amount of magnesium nitrate to the reaction mixture in place of the same amount of iron nitrate. Carbon content in the obtained materials was determined by measuring the mass of the residue after annealing at 700 °С for 3 h. X-Ray diffraction (XRD) analysis of the samples was carried out using Rigaku D/MAX 2200 diffractometer with CuKα radiation. The spectra were processed using Rigaku Application Data Processing and FullProf software. Microstructure analysis was performed using a scanning electron microscope Carl Zeiss NVision 40. Electrode paste was prepared by thoroughly mixing 85% LFP or LFP-Mg as an active material, 10% of conductive carbon black (Timcal, Belgium) and 5% of the polyvinylidene fluoride binder (Aldrich) dissolved in anhydrous N-methyl-2-pyrrolidinone. Electrode paste was deposited on a stainless steel grid, serving as the current collector, as a 10 mg cm−2 layer. The resulting electrode was kept under 1000 kg cm−2 pressure at 120 °С for 8 h. Electrochemical testing was performed in hermetically sealed threeelectrode cells with lithium as auxiliary and reference electrodes. The surface of the working electrode was 2.25 cm2 and the surface of the auxiliary lithium electrode was 5 cm2. The cell was assembled in a glove box under argon with humidity level < 10 ppm. All electrodes were separated by polypropylene separator (NPO “Ufim”, Moscow). A 1 М LiPF6 solution in an ethylcarbonate-diethylcarbonate-dimethylcarbonate mixture (Novolyte, US) was used as an electrolyte. Electrochemical testing of the cells was performed in a potential window from 2.5 to 4.1 V using a ZRU 50 mA-10 V charge–discharge system (Buster, Russia). Testing was carried out in a galvanostatic mode at current densities 20–800 mA g−1. The measurement error of capacity determination is up to ± 3 mAh g−1. To perform the X-ray diffraction analysis and the Mössbauer study the samples with a different degree of charge or discharge were prepared using the same electrochemical cells. Degree of Fe2+↔Fe3+ conversion was controlled by means of the electrical charge passed through the cell. Afterwards, the cathode material was dried in a vacuum chamber and carefully scraped off the current collector. Mössbauer spectra have been recorded by using a conventional spectrometer MS-1101E of electromechanical-type with the constant acceleration mode and absorption geometry in the temperature range 4.8–300 K. The time dependence of Doppler rate has a “triangular” shape. The spectra corresponding to the ascending and descending areas of the “triangle” have been registered in the 1024 channels. A 57 Co (Rh) γ-quantum source has been used. Calibration has been done

Fig. 1. Portions of X-ray diffraction patterns of the synthesized samples: LFP (1), LFP-Mg (2), LFP-Mg charged to ~50% (3) and a fully charged LFP-Mg (4); bar diagrams for LiFePO4 (the reference card of LiFePO4, orthorhombic modification Card 81–1173 ICDD PDF-2) (5), and FePO4 (orthorhombic modification Card 70–6685 ICDD PDF-2) (6).

150

Solid State Ionics 317 (2018) 149–155

S. Yaroslavtsev et al.

Table 1 Unit cell parameters and coherent scattering regions (CSR) of LFP and LFP-Mg samples. Sample

a, Å

b, Å

c, Å

CSR hkl (311), nm

LFP LFP-Mg

10.3248(7) 10.270(4)

6.0047(5) 5.977(3)

4.6906(3) 4.677(2)

42 46

Fig. 4. Charge–discharge curves of LFP(1) and LFP-Mg(2) samples (20 mA g−1). Fig. 2. Rietveld refinement of XRD profiles of LiFe0.8Mg0.2PO4/C. (Space group Pnma, Reliability factors: χ2 = 0.422, Rp = 6.56, Rwp = 8.43).

Fig. 3. SEM image of LFP-Mg.

Fig. 5. Discharge capacities of LFP (1) и LFP-Mg (2) for cycling at different current densities. Current densities (mA g-1) are shown in the figure.

markedly larger than the coherent scattering regions, calculated form X-ray diffraction data (Table 1), which allows for a conclusion that LFPMg particles exist as agglomerates of single crystallites. According to thermogravimetry data, the obtained samples contain approximately 5 wt% of carbon.

reason is a higher degree of iron substitution by electrochemically inactive magnesium (20%). The optimal electrochemical performance was achieved usually at low divalent dopants contents (2–5%) [31,35,37–41]. With increasing the current density to 400 mA g−1, the LFP-Mg capacity is lowered by approximately 40%. After returning to low current densities the capacity is fully restored, demonstrating that the capacity drop at high current densities is not associated with degradation of the material.

3.2. Charge/discharge behavior Magnesium ions have only one oxidation state 2+ and do not participate in the charge process. Charge–discharge curves for both the pristine and the magnesium-doped samples feature a plateau at the potential range of ~3.4 V (Fig. 4-(1)), corresponding to the Fe2+↔Fe3+ transition, which proceeds through the two-phase interface movement between LiFePO4 and FePO4. However, the difference between charge and discharged potentials, as well as the slopes of the curves, is higher for LFP-Mg, compared to the non-doped sample (Fig. 4-(2)). In this respect, the situation appears to differ from the previously studied nickel- and cobalt-doped samples [19,33]. Most likely, the reason for the observed discrepancies is a higher resistance of LFP-Mg. The results of cycling of the LFP and LFP-Mg samples at different current densities are shown in Fig. 5. At the current density of 20 mA g−1 the discharge capacity was 127 mAh g−1. It should be noted that this value approaches the theoretical value considering a lower content of electrochemically active iron and a carbon content of ~5 wt %. At the same time, it is lower compared to the capacities of magnesium-doped LiFePO4 samples, reported elsewhere [31,35,37–41]. The

3.3. Mössbauer spectroscopy A spectrum of the initial sample LixFe0.8Mg0.2PO4 (x = 1), collected at 5.5 K (Fig. 6), features only minor differences from the spectrum of the non-doped lithiated (non-charged) sample [19,45]. No significant contribution of Fe3+ ions to the Mössbauer spectrum was observed. Nonetheless, there are four sites available in the nearest cationic environment of Fe2+ ions, part of which (from 0 to 4) can be occupied by magnesium. Fitting of the spectrum was carried out on the assumption of binomial distribution of magnesium atoms in the iron sites, in the frame of which the most probable nearest neighbors in iron sites are: (4Fe2+) ~41%, (3Fe2+;1Mg2+) ~41%, (2Fe2+;2Mg2+) ~15% (Table 2). Parameters of the subspectrum corresponding to Fe2+ ions with the (4Fe2+) environment are almost identical to that of the spectrum of the non-doped sample [19,45]. It was determined that the 151

Solid State Ionics 317 (2018) 149–155

S. Yaroslavtsev et al.

Fig. 6. Mössbauer spectrum, collected at 5.5 K, of the pristine LFP-Mg, its fitting and positions of the resonance lines of subspectra.

substitution of one iron atom with magnesium a significant decrease of the hyperfine magnetic field (~9 kOe) at Fe2+ nuclei occurs. Since the variations of quadrupole coupling constants are insignificant (~ 0.03 mm s−1), it may be assumed that the symmetry of distribution of charges, which create an electric field gradient at the 57Fe nucleus, remains virtually unchanged. This fact confirms that Mg2+ ions are located in the sites of Fe2+ ions, which have the same charge state. If significant quantities of magnesium also occupied the sites of singly charged lithium ions, it would change the symmetry of charge distribution. With an estimation utilizing the localized charge approximation a value of ~ 0.3 mm s−1 is predicted, which is ten times more than observed. A spectrum of the fully charged sample LixFe0.8Mg0.2PO4 (x = 0.265), collected at 5.5 K, features six well-resolved resonance lines (Fig. 7), although, unlike the spectrum of charged non-doped sample [19,45], an asymmetry coupled with non-uniform broadening of resonance lines is observed. A relative contribution to the integral intensity of the subspectrum, corresponding to Fe3+ ions having only Fe3+ ions in the nearest environment, Fe3+ (4Fe3+), is ~25%. It is in good correlation with the probability of their detection assuming a random distribution of Fe2+ and Mg2+ ions in iron sites (~27%) (Table 2). Spectral parameters are very close to those of the non-doped FePO4 [19, 45]. A fraction of Fe3+ ions having a nearest environment different from (4Fe3+) is ~67% (marked as “Fe3+ average”). Based on the reconstruction of hyperfine parameters distribution of the mentioned Fe3+ ions (Fig. 8) it was

Fig. 7. Mössbauer spectrum of the charged LFP-Mg sample at 5.5 K, its fitting and positions of the resonance lines of subspectra.

determined that the mean value of hyperfine magnetic field experiences only a slight change with altering of the nearest environment compared to (4Fe3+). Nonetheless, noticeable change of the quadrupole coupling constant e2qQ was observed (Fig. 8, Table 3). In addition, a low intensity (~8%) contribution of unoxidized Fe2+ ions, associated with substantially lower values of the hyperfine magnetic field, was observed in the mentioned spectrum. Since the X-ray diffraction pattern processing demonstrated that the fully charged sample (Fig. 1-(4)) has the FePO4 structure, it may be concluded that Fe2+ ions as well as Mg2+

Table 2 Relative intensities of subspectra corresponding to Fe2+ and Fe3+ with different nearest neighbors in iron sites in LixFe0.8Mg0.2PO4, where “Fe3+ average” correspond to all Fe3+ ions with exception of Fe3+(4Fe3+). The calculated probabilities in binomial distribution are relative to the intensity of the whole Mössbauer spectrum. Ion and nearest neighbors in iron sites

x=1

x = 0.63

experiment 2+

Fe Fe2+ Fe2+ Fe2+ Fe3+ Fe3+

2+

(4Fe ) (3Fe2+;1Mg2+) (2Fe2+;2Mg2+) in FePO4 structure (4Fe3+) average

binom

experiment

binom

experiment

binom

– – –

19.5 ± 0.7% 10.3 ± 1.1% 8.2 ± 0.8% 16.3 ± 0.7% 5.2 ± 0.7% 40.4 ± 0.3%

– – – – – –

0 0 0 8.1 ± 0.3% 25.2 ± 1.0% 66.7 ± 1.1%

– – – – 26.8% 65.1%

42.1% 42.1% 15.8% 0 0 0

x = 0.265

152

Solid State Ionics 317 (2018) 149–155

S. Yaroslavtsev et al.

Fig. 8. Hyperfine magnetic field p(Hn) (a) and quadrupole coupling constant p(e2qQ) (b) distributions, obtained for Fe3+ in the fully charged (1) and partially charged (2) samples of LFPMg; the solid line indicates the values, corresponding to the Fe3+ ions having four Fe3+ ions in the nearest environment.

solubility of Fe2+ ions in FePO4, compared to the reported data [7], is in this case due to the presence of Mg2+ ions in the sample. Unfortunately, it is not possible to determine the solubility of the Fe3+ ions on the basis of Mossbauer studies, because the subspectrum corresponding to Fe3+ cations in the LiFePO4 structure is not observed for the pristine sample, and in the spectra of charged samples it cannot be distinguished, since its contribution to the Mossbauer spectrum is small and the parameters are unknown. However, the solubility of the Fe3+ ions can be estimated based on the analysis of the shape of the charge curves, as it was done in [7]. This allows us to conclude that for the Mg-doped sample the solubility of Fe3+ ions in LiFePO4 structure is much lower than that of Fe2+ in FePO4 structure. At the same time this value is somewhat higher than the solubility of Fe3+ in LiFePO4 in undoped sample. For partially charged sample, probabilities of realization of different cationic environments cannot been calculated, because it is unknown how many Fe2+ and Mg2+ ions are randomly distributed in FePO4 structure and Mg2+ ions in LiFePO4 structure. A relative intensity of the subspectrum corresponding to Fe3+ ions having four Fe3+ ions in the nearest environment is only ~5% (Fig. 9, Table 2). In the case of other Fe3+ ions a large increase of the quadrupole coupling constant e2qQ is observed (Table 3). Mean values of the hyperfine magnetic field at the nuclei of Fe3+ ions and quadrupole coupling constants are significantly higher, compared to the fully charged sample, which is evident from the hyperfine parameter distribution obtained from the spectra processing (Fig. 8). This means that the nearest environment of virtually all Fe3+ ions contain several Fe2+ and/or Mg2+ ions. No such change was observed during deintercalation of lithium iron phosphates doped with transition metals [22]. Doping of lithium iron phosphate leads to a decrease of Neel temperature, due to breakdown of superexchange interaction between iron ions. However, it is also important to mention that if the regions of Fe3+ ion localization are sufficiently small, a superparamagnetic relaxation effect may be observed. Thus the spectrum can appear as a doublet even below the Neel temperature. At lower temperatures the spectrum acquires a typical relaxation profile, featuring an evident broadening and asymmetry of resonance lines. Further temperature lowering to 5.5 K leads to the disappearance of relaxation effects, and the spectrum becomes an octet. In the spectra of partially and fully charged samples, collected at 82 K, two quadrupole doublets can be seen, corresponding to Fe2+ и Fe3+ ions, respectively. At the same time, a subspectrum of Fe3+ ions, appearing as a five-step pedestal, was also discovered (Fig. 10). Such a profile is typical for the case of slow relaxation [47]. The observed relaxation effects mentioned above are due to the existence of nanoscale magnetic regions, containing Fe3+ ions. Characteristic linear

Table 3 Average values of hyperfine parameters for Fe3+ in LixFe0.8Mg0.2PO4 at 5 K, where δ is the isomer shift, e2qQ is the quadrupole coupling constant and Hn is the magnetic hyperfine field. Hyperfine parameter

Hn, kOe e2qQ, mm/s δ, mm/s

Fe3+ (4Fe3+)

497 −2.5 0.55

Fe3+ average⁎ x = 0.63

x = 0.265

515.2 ± 0.5 −1.00 ± 0.04 0.55 ± 0.01

500.4 ± 0.3 −2.2 ± 0.2 0.55 ± 0.01

⁎ Parameters were obtained under the assumption that the direction of the hyperfine magnetic field in the principal axes of GEF tensor is identical for Fe3+ in partially (x = 0.63) and fully (x = 0.265) charged samples.

ions are present in the structure of the this phase. A Mössbauer spectrum of the partially charged sample LixFe0.8Mg0.2PO4 (x = 0.63) at 5.5 K appears as a combination of spectra of the pristine and fully charged samples (Fig. 9). Contribution to the spectrum intensity from all the Fe2+ ions is ~54%, and from Fe3+ it is ~46%. Analysis of X-ray diffraction patterns shows the contributions of the phases with the LiFePO4 and FePO4 structures to be 48% and 52%, respectively. Besides, the contribution of the subspectrum corresponding to Fe2+ ions in the FePO4 structure is ~17% (Table 2). Its parameters are similar to those of the Fe2+ ions in the FePO4 structure of the fully charged sample. Yet, a considerable fraction of these ions can be located at interfaces. Nonetheless, relying on all the obtained data, one can conclude that the solubility of LiFePO4 in FePO4 is noticeably higher, than that of FePO4 in LiFePO4. An enhanced

Fig. 9. Mössbauer spectrum, collected at 5.5 K, of a partially charged LFP-Mg sample, its fitting and positions of the resonance lines of subspectra.

153

Solid State Ionics 317 (2018) 149–155

S. Yaroslavtsev et al.

corresponding to the oxidation and reduction of iron. The portion of the charge corresponding to the solid solution mechanism increases with an increase in the manganese content in LiFe1-xMnxPO4 (x = 0–0.5) [36]. Whereas, Bramnik et al. [52] reported that lithium extraction from LiMn0.6Fe0.4PO4 proceeds through a two-phase mechanism similar to the lithium extraction from isostructural LiFePO4. The results obtained in our article indicates significant phase separation for partially charged LFP-Mg sample from the diffraction patterns that is confirmed by the Mössbauer spectroscopy data. This seems to be different from the ones of Omenya et al. [53] who observed a shift and broadness of diffraction peaks during charge/discharge and the “single-phase-like” reaction during lithiation/delithiation processes, leading to the existence of particles with varying lithium composition upon LiMg0.2Fe0.8PO4 charge/discharge. However, Omenya et al. had found this effect for the LiMg0.2Fe0.8PO4 sample with small sizes of particles (40–50 nm) and narrow particle size distribution. The reason for this could be a high magnesium content, which increases the mutual solubility of the phases. In addition, as it was noted for LiFePO4 in [54] and for Li4Ti5O12 in [55], mutual solubility of coexisting phases depends on the particle size, so the smaller particles exhibit higher solubility limits compared with larger ones. At the same time Omenya et al. [53] observed the effect of “suppression of phase separation” in a much less pronounced form for Li1xFe0.8Mg0.2PO4 sample with larger particles with wide distribution of particle sizes (50–150 nm). Most likely, namely the particles with uniform small size (50 nm) in the polydisperse material caused the presence of intermediate phases fixed upon lithiation of Li1-xFe0.8Mg0.2PO4 (x = 0.28, 0.48). At the same time, the authors expected that in the case of partially lithiated/delithiated large particles at equilibrium, the separation to occur within the same particles. The LFP-Mg sample investigated in this paper is characterized by a wider size distribution (from 50 to 300 nm) with the mean size of the particles of ~100 nm. So, the significant phase separation for partially charged LFP-Mg sample fixed in our work does not contradict with the ones of Omenya et al. [53] for materials with wide particle size distribution and accentuates the importance of particle size for the charge/ discharge processes mechanism.

Fig. 10. Spectra of partially (a) and fully (b) charged LFP-Mg samples, collected at 82 K, their fitting and positions of the lines of quadrupole doublets.

sizes allowing for such spectra are usually < 10 nm. The existence of nanoscale regions is in good agreement with the fact that for an intermediate charge only a small amount of Fe3+ ions with four Fe3+ ions in the nearest environment is observed. In the case of small-sized regions, the fraction of Fe3+ ions located at interfaces, and consequently having Fe2+ or Mg2+ ions in the nearest environment, increases. It is important to discuss the nature of the origin of the mentioned regions. Mg2+ ions cannot be oxidized to the trivalent state during electrochemical cycling, and a low mutual solubility of FePO4 and LiFePO4 phases emphasizes that formation of trivalent ions near the divalent ones is unpreferable. Considering this, it may be assumed that during the charge process the removal of lithium atoms from the regions with a high Mg2+ concentration is hindered. In the extreme case, Fe2+ ions with no Mg2+ ions in the nearest environment should be oxidized first, but this would mean that lithium deintercalation proceeds with a solid solution formation; moreover, a fraction of lithium ions would have to be extracted from the inside of the particles, which is infeasible until a non-charged (lithiated) phase is present on the surface. For this reason, small regions having a Mg2+ concentration below the mean value (20%) and the majority of iron ions in the Fe3+ state are formed. In the region where Mg2+ concentration is above the mean value Fe2+ oxidation is hindered and proceeds at later stages of charging, i.e. the process takes place with a deviation from the shrinking sphere model [1,14]. It results in the formation of nanoscale regions based on the FePO4 phase, having a lower Mg2+ concentration, as well as the regions based on the LiFePO4 phase, having a higher Mg2+ concentration. There are contradictory results regarding the charge/discharge mechanism of LFP and doped samples [1,14,48,49]. The most of authors believe that this mechanisms is two-phase [1,14]. At the same time domino-cascade [48], and nonequilibrium single-phase transformation mechanisms [49] have been proposed also for LFP lithiation–delithiation processes. A number of authors note the change the two phase Li+ deintercalation/intercalation mechanism to the solid solution formation mechanism upon a partial replacement of iron in LFP with another dopant atom (manganese) [22, 36, 50, 51]. But even in this case, the process proceeds according to the two-phase mechanism at the initial stages of charge and discharge at a potential

4. Conclusion Composite nanomaterials based on LiFePO4 and LiFe0.8Mg0.2PO4, which crystallize in the orthorhombic modification of LiFePO4 (space group Pnma), and carbon have been obtained. It is shown using the Rietveld refinement and Mössbauer spectroscopy that magnesium occupies predominantly iron sites. Discharge capacity of LFP-Mg is 127 mAh g−1 at a current density of 20 mA g−1, which is close to the theoretical value for the sample with such content of electrochemically active iron. Mössbauer spectroscopy was utilized to investigate the influence of magnesium doping on the charge/discharge processes. It is demonstrated that at the early stage of the charge process nanoscale regions with FePO4 structure and a reduced Mg2+ concentration are formed. Meanwhile, doping with magnesium leads to an enhancement of Fe2+ solubility in such regions. At the maximum charge state all Mg2+ ions and the residual Fe2+ ions are dissolved in the FePO4 structure. Acknowledgments This work was financially supported by the Russian Foundation for Basic Research (project №16-29-05241). This work was performed using the equipment of the Joint Research Centre of IGIC RAS. References [1] A.K. Padhi, K.S. Nanjundaswamy, J.B. Goodenough, Phospho-olivines as positiveelectrode materials for rechargeable lithium batteries, J. Electrochem. Soc. 144

154

Solid State Ionics 317 (2018) 149–155

S. Yaroslavtsev et al.

dx.doi.org/10.1021/cm801781k. [30] T. Nakamura, Y. Miwa, M. Tabuchi, Y. Yamada, Structural and surface modifications of LiFePO4 olivine particles and their electrochemical properties, J. Electrochem. Soc. 153 (2006) A1108–A1114, http://dx.doi.org/10.1149/1. 2192732. [31] M.R. Roberts, G. Vitins, J.R. Owen, High-throughput studies of Li1−xMgx/2FePO4 and LiFe1−yMgyPO4 and the effect of carbon coating, J. Power Sources 179 (2008) 754–762, http://dx.doi.org/10.1016/j.jpowsour.2008.01.034. [32] Y. Ge, X. Yan, J. Liu, X. Zhang, J. Wang, X. He, R. Wang, H. Xie, An optimized Ni doped LiFePO4/C nanocomposite with excellent rate performance, Electrochim. Acta 55 (2010) 5886–5890, http://dx.doi.org/10.1016/j.electacta.2010.05.040. [33] D.V. Safronov, S.A. Novikova, T.L. Kulova, A.M. Skundin, A.B. Yaroslavtsev, Lithium diffusion in materials based on LiFePO4 doped with cobalt and magnesium, Inorg. Mater. 48 (2012) 513–519, http://dx.doi.org/10.1134/ S0020168512050159. [34] H. Gao, L. Jiao, J. Yang, Z. Qi, Y. Wang, H. Yuan, High rate capability of Co-doped LiFePO4/C, Electrochim. Acta 97 (2013) 143–149, http://dx.doi.org/10.1016/j. electacta.2013.02.132. [35] A. Örnek, O. Efe, Doping qualifications of LiFe1-xMgxPO4/C Nano-scale composite cathode materials, Electrochim. Acta 166 (2015) 338–349, http://dx.doi.org/10. 1016/j.electacta.2015.03.010. [36] O.A. Drozhzhin, V.D. Sumanov, O.M. Karakulina, A.M. Abakumov, J. Hadermann, A.N. Baranov, K.J. Stevenson, E.V. Antipov, Electrochim. Acta 191 (2016) 149–157, http://dx.doi.org/10.1016/j.electacta.2016.01.018. [37] B. Wang, B. Xu, T. Liu, P. Liu, Ch. Guo, Sh. Wang, Q. Wang, Zh. Xiong, D. Wang, X.S. Zhao, Mesoporous carbon-coated LiFePO4 nanocrystals co-modified with graphene and Mg2+ doping as superior cathode materials for lithium ion batteries, Nano 6 (2014) 986–995, http://dx.doi.org/10.1039/c3nr04611g. [38] H. Liu, J. Xie, Synthesis and characterization of LiFe0.9Mg0.1PO4/nano-carbon webs composite cathode, Int. J. Mater. Prod. Technol. 209 (2009) 477–481, http://dx. doi.org/10.1016/j.jmatprotec.2008.02.014. [39] X. Ou, G. Liang, L. Wang, Sh. Xu, X. Zhao, Effects of magnesium doping on electronic conductivity and electrochemical properties of LiFePO4 prepared via hydrothermal route, J. Power Sources 184 (2008) 543–547, http://dx.doi.org/10. 1016/j.jpowsour.2008.02.077. [40] Z. Yang, J. Xia, L. Zhi, W. Zhang, B. Pei, An improved solid-state reaction route to Mg2+-doped LiFePO4/C cathode material for Li-ion battery, Ionics 20 (2014) 169–174, http://dx.doi.org/10.1007/s11581-013-0974-2. [41] J. Ni, Y. Zhao, J. Chen, L. Gao, L. Lu, Site-dependent electrochemical performance of Mg doped LiFePO4, Electrochem. Commun. 44 (2014) 4–7, http://dx.doi.org/10. 1016/j.elecom.2014.04.004. [42] Ch. Delacourt, Ph. Poizot, J.-M. Tarascon, Ch. Masquelier, The existence of a temperature-driven solid solution in LixFePO4 for 0 ≤ x ≤ 1, Nat. Mater. 4 (2005) 254–260, http://dx.doi.org/10.1038/nmat1335. [43] I.K. Lee, Ch.M. Kim, S.J. Kim, Ch.S. Kim, Structural and magnetic phase transition of mixed olivines LixFe1-yNiyPO4 by lithium deintercalation, J. Appl. Phys. 111 (2012) (07d722-1-07d722-3), https://doi.org/10.1063/1.3678468. [44] Ch.H. Rhee, S.J. Kim, Ch.S. Kim, Mossbauer studies of spin-orbit coupling in LiCo0.9957Fe0.01PO4, IEEE T. Magn. 47 (2011) 2697–2700. [45] V. Rusakov, S. Yaroslavtsev, M. Matsnev, T. Kulova, A. Skundin, S. Novikova, A. Yaroslavtsev, Mossbauer study of LixFe1-yCoyPO4 as cathode materials for Li-ion batteries, Hyperfine Interact. 226 (2014) 791–796, http://dx.doi.org/10.1007/ s10751-013-0979-2. [46] R.D. Shannon, Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides, Acta Cryst A32 (1976) 751–767, http://dx. doi.org/10.1107/S0567739476001551. [47] M.A. Chuev, V.M. Cherepanov, M.A. Polikarpov, On the shape of the gamma resonance spectra of slowly relaxing nanoparticles in a magnetic field, JETP Lett. 92 (2010) 21–27, http://dx.doi.org/10.1134/S0021364010130047. [48] G. Brunetti, D. Robert, P. Bayle-Guillemaud, J.L. Rouviere, E.F. Rauch, J.F. Martin, J.F. Colin, F. Bertin, C. Cayron, Confirmation of the domino-Cascade model by LiFePO4/FePO4 precession electron diffraction, Chem. Mater. 23 (2011) 4515–4524, http://dx.doi.org/10.1021/cm201783z. [49] R. Malik, F. Zhou, G. Ceder, Kinetics of non-equilibrium lithium incorporation in LiFePO4, Nat. Mater. 10 (2011) 587–590, http://dx.doi.org/10.1038/nmat3065. [50] J. Molenda, W. Ojczyk, J. Marzec, Electrical conductivity and reaction with lithium of LiFe1−yMnyPO4 olivine-type cathode materials, J. Power Sources 174 (2007) 689–694, http://dx.doi.org/10.1016/j.jpowsour.2007.06.238. [51] D.B. Ravnsbaek, K. Xiang, W. Xing, O.J. Borkiewicz, K.M. Wiaderek, P. Gionet, K.W. Chapman, P.J. Chupas, Y._.M. Chiang, Extended solid solutions and coherent transformations in nanoscale olivine cathodes, Nano Lett. 14 (2014) 1484–1491, http://dx.doi.org/10.1021/nl404679t. [52] N.N. Bramnik, K.G. Bramnik, K. Nikolowski, M. Hinterstein, C. Baehtz, H. Ehrenberg, Synchrotron Diffraction Study of Lithium Extraction from LiMn0.6Fe0.4PO4, Electrochem. Solid-State Lett. 8 (2005) A379–A381, http://dx.doi. org/10.1149/1.1940487. [53] F. Omenya, B. Wen, J. Fang, R. Zhang, Q. Wang, N.A. Chernova, J. SchneiderHaefner, F. Cosandey, M.S. Whittingham, Adv. Energy Mater. 5 (2015) 1401204, http://dx.doi.org/10.1002/aenm.201401204. [54] G. Kobayashi, S.-I. Nishimura, M.-S. Park, R. Kanno, M. Yashima, T. Ida, A. Yamada, Isolation of solid solution phases in size-controlled LixFePO4 at room temperature, Adv. Funct. Mater. 19 (2009) 395–403, http://dx.doi.org/10.1002/adfm. 200801522. [55] I.A. Stenina, T.L. Kulova, A.M. Skundin, A.B. Yaroslavtsev, High grain boundary density Li4Ti5O12/anatase-TiO2 nanocomposites as anode material for Li-ion batteries, Mater. Res. Bull. 75 (2016) 178–184, http://dx.doi.org/10.1016/j. materresbull.2015.11.050.

(1997) 1188–1194, http://dx.doi.org/10.1149/1.1837571. [2] Zh. Yang, Y. Dai, Sh. Wang, J. Yu, How to make lithium iron phosphate better: a review exploring classical modification approaches in-depth and proposing future optimization methods, J. Mater. Chem. A 4 (2016) 18210–18222, http://dx.doi. org/10.1039/C6TA05048D. [3] A. Eftekhari, LiFePO4/C nanocomposites for lithium-ion batteries, J. Power Sources 343 (2017) 395–411, http://dx.doi.org/10.1016/j.jpowsour.2017.01.080. [4] S.A. Novikova, A.B. Yaroslavtsev, Cathode materials based on olivine lithium iron phosphates for lithium-ion batteries, Rev. Adv. Mater. Sci. 49 (2017) 129–139. [5] C. Benoit, S. Franger, Chemistry and electrochemistry of lithium iron phosphate, J. Solid State Electrochem. 12 (2008) 987–993, http://dx.doi.org/10.1007/s10008007-0443-9. [6] R. Amin, J. Maier, P. Balaya, D.P. Chen, C.T. Lin, Ionic and electronic transport in single crystalline LiFePO4 grown by optical floating zone technique, Solid State Ionics 179 (2008) 1683–1687, http://dx.doi.org/10.1016/j.ssi.2008.01.079. [7] D.V. Safronov, S.A. Novikova, A.M. Skundin, A.B. Yaroslavtsev, Lithium intercalation and deintercalation processes in Li4Ti5O12 and LiFePO4, Inorg. Mater. 48 (2012) 57–61, http://dx.doi.org/10.1134/S0020168512010141. [8] Z.-Y. Chen, H.-L. Zhu, S. Ji, R. Fakir, V. Linkov, Influence of carbon sources on electrochemical performances of LiFePO4/C composites, Solid State Ionics 179 (2008) 1810–1815, http://dx.doi.org/10.1016/j.ssi.2008.04.018. [9] D.V. Safronov, I.Y. Pinus, I.A. Profatilova, V.A. Tarnopol'skii, A.M. Skundin, A.B. Yaroslavtsev, Kinetics of lithium deintercalation from LiFePO4, Inorg. Mater. 47 (2011) 303–307, http://dx.doi.org/10.1134/S0020168511030198. [10] H. Li, H. Zhou, Enhancing the performances of Li-ion batteries by carbon-coating:present and future, Chem. Commun. 48 (2012) 1201–1217, http://dx.doi.org/ 10.1039/C1CC14764A. [11] D. Gryzlov, S. Novikova, T. Kulova, A. Skundin, A. Yaroslavtsev, Behavior of LiFePO4/CPVDF/Ag-based cathode materials obtained using polyvinylidene fluoride as the carbon source, Mater. Des. 104 (2016) 95–101, http://dx.doi.org/10.1016/j. matdes.2016.05.040. [12] X. Tu, Y. Zhou, Y. Song, Freeze-drying synthesis of three-dimensional porous LiFePO4 modified with well-dispersed nitrogen-doped carbon nanotubes for highperformance lithium-ion batteries, Appl. Surf. Sci. 400 (2017) 329–338, http://dx. doi.org/10.1016/j.apsusc.2016.12.220. [13] K. Zaghib, A. Guerfi, P. Hovington, A. Vijh, M. Trudeau, A. Mauger, J.B. Goodenough, C.M. Julien, Review and analysis of nanostructured olivine-based lithium rechargeable batteries: status and trends, J. Power Sources 232 (2013) 357–369, http://dx.doi.org/10.1016/j.jpowsour.2012.12.095. [14] A.B. Yaroslavtsev, T.L. Kulova, A.M. Skundin, Electrode nanomaterials for lithiumion batteries, Russ. Chem. Rev. 84 (2015) 826–852, http://dx.doi.org/10.1070/ RCR4497. [15] M. Gao, N. Liu, Zh. Li, W. Wang, Y. Huang, A gelatin-based sol–gel procedure to synthesize the LiFePO4/C nanocomposite for lithium ion batteries, Solid State Ionics 258 (2014) 8–12, http://dx.doi.org/10.1016/j.ssi.2014.01.041. [16] H. Guo, Y. Liu, Y. Xi, Ch. Xu, Q. L, Investigation on high performance LiFePO4 nanoplates with the {010} face prominent for lithium battery cathode materials, Solid State Ionics 298 (2016) 44–50, http://dx.doi.org/10.1016/j.ssi.2016.10.015. [17] R.R. Kapaev, S.A. Novikova, T.L. Kulova, A.M. Skundin, A.B. Yaroslavtsev, Nanotechnol. Russ. 11-12 (2016) 757–760, http://dx.doi.org/10.1134/ S1995078016060136. [18] A. Örnek, E. Bulut, M. Can, M. Özacar, Characteristics of nanosized LiNixFe1−xPO4/ C (x = 0.00–0.20) composite material prepared via sol–gel-assisted carbothermal reduction method, J. Solid State Electrochem. 17 (2013) 3101–3107, http://dx.doi. org/10.1007/s10008-013-2201-5. [19] S. Novikova, S. Yaroslavtsev, V. Rusakov, T. Kulova, A. Skundin, A. Yaroslavtsev, LiFe1−xMIIxPO4/C (MII = Co, Ni, Mg) as cathode materials for lithium-ion batteries, Electrochim. Acta 122 (2014) 180–186, http://dx.doi.org/10.1016/j.electacta. 2013.08.118. [20] W. Liu, Q. Huang, G. Hu, A novel preparation route for multi-doped LiFePO4/C from spent electroless nickel plating solution, J. Alloys Compd. 632 (2015) 185–189, http://dx.doi.org/10.1016/j.jallcom.2015.01.110. [21] A. Kulka, A. Braun, Tz.-W. Huang, A. Wolska, J. Molenda, Evidence for Al doping in lithium sublattice of LiFePO4, Solid State Ionics 270 (2015) 33–38, http://dx.doi. org/10.1016/j.ssi.2014.12.004. [22] S. Novikova, S. Yaroslavtsev, V. Rusakov, A. Chekannikov, T. Kulova, A. Skundin, A. Yaroslavtsev, Behavior of LiFe1-yMnyPO4/C cathode materials upon electrochemical lithium intercalation/deintercalation, J. Power Sources 300 (2015) 444–452, http://dx.doi.org/10.1016/j.jpowsour.2015.09.092. [23] Q. Liu, W. Liu, D. Li, Z. Wang, Synthesis and characterization of grape-like LiFe0.97M0.03PO4/C (M = Ni, Co, Mn) composites, Mater. Lett. 162 (2016) 87–90, http://dx.doi.org/10.1016/j.matlet.2015.09.129. [24] A. Kulka, K. Świerczek, K. Walczak, A. Braun, J. Molenda, Correlation between transport properties and lithium extraction/insertion mechanism in Fe-site substituted phosphoolivine, Solid State Ionics 288 (2016) 184–192, http://dx.doi.org/ 10.1016/j.ssi.2016.01.034. [25] L. Gao, Zh. Xu, Sh. Zhang, J. Xu, K. Tang, Enhanced electrochemical properties of LiFePO4 cathode materials by co and Zr multi-doping, Solid State Ionics 305 (2017) 52–56, http://dx.doi.org/10.1016/j.ssi.2017.04.021. [26] S.A. Rise, Diffusion-Limited Reactions, Elsevier, 1985. [27] S.-Y. Chung, J.T. Bloking, Y.-M. Chiang, Electronically conductive phospho-olivines as lithium storage electrodes, Nat. Mater. 1 (2002) 123–128, http://dx.doi.org/10. 1038/nmat732. [28] D. Wang, H. Li, S. Shi, X. Huang, L. Chen, Improving the rate performance of LiFePO4 by Fe-site doping, Electrochim. Acta 50 (2005) 2955–2958, http://dx.doi. org/10.1016/j.electacta.2004.11.045. [29] M. Wagemaker, B.L. Ellis, D. Lutzenkirchen-Hecht, F.M. Mulder, L.F. Nazar, Proof of Supervalent doping in olivine LiFePO4, Chem. Mater. 20 (2008) 6313–6315, http://

155