8626
J. Phys. Chem. C 2010, 114, 8626–8632
Structural and Magnetic Nature for Fully Delithiated LixNiO2: Comparative Study between Chemically and Electrochemically Prepared Samples Kazuhiko Mukai,*,† Jun Sugiyama,† Yutaka Ikedo,†,| Yoshifumi Aoki,† Daniel Andreica,‡,§ and Alex Amato‡ Toyota Central Research and DeVelopment Laboratories, Inc., 41-1 Yokomichi, Nagakute, Aichi 480-1192, Japan, Laboratory for Muon-Spin Spectroscopy, Paul Scherrer Institut, Villigen CH-5232, Switzerland, and Faculty of Physics, Babes-Bolyai UniVersity, 400084 Cluj-Napoca, Romania ReceiVed: January 26, 2010; ReVised Manuscript ReceiVed: March 7, 2010
Structural and magnetic properties of two LixNiO2 samples with x e 0.1 have been studied by a powder X-ray diffraction, magnetic susceptibility (χ), and muon-spin rotation/relaxation (µSR) measurements. One of the two samples was prepared by an electrochemical (EC) reaction in a nonaqueous lithium cell, whereas the other sample by a chemical (C) reaction in a HNO3 aqueous solution. Although the crystal structure of both C-Li0.01NiO2 and EC-Li0.10NiO2 samples were assigned as a mixture of a cubic close-packed phase and a hexagonal close-packed phase, the effective magnetic moment (µeff) of Ni ions for C-Li0.01NiO2 was estimated as µeff ) 1.43 µB and was very close to that for EC-Li0.5NiO2 (µeff ) 1.39 µB). This implies that the vacant tetrahedral or octahedral sites in C-Li0.01NiO2 are partially occupied by H+ ions. Actually, the zero-field µSR time spectrum in the paramagnetic state for C-Li0.01NiO2 exhibited a large relaxation compared to those for EC-Li0.5NiO2 and EC-Li0.10NiO2. Furthermore, a pryrolysis gas chromatography/mass spectroscopy analysis confirmed the existence of H+ ions in the C-Li0.01NiO2 crystalline lattice. The actual composition of C-Li0.01NiO2 is, thus, determined to be H∼0.5Li0.01NiO2. Introduction The lithium insertion materials, for which Li+ ions are inserted into (or extracted from) a rigid matrix without destruction of a framework structure (so-called topotactic), have been heavily investigated by electrochemists and battery researchers, because of their practical application to Li ion batteries (LIB).1 Although a nonaqueous electrolyte is currently used in the commercial LIB, it is widely recognized that the acid treating with an aqueous solution provides a fully delithiated phase. Hunter reported2 that a fully delithiated LixMn2O4 (λ-MnO2) is produced by digesting LiMn2O4 in an acid solution with the following disproportionation reaction
4LiMn2O4 + 8H+ f 6λ-MnO2 + 2Mn2+ + 4Li+ + 4H2O (1) This is because the acid treating in a sufficient H+ solution corresponds to the voltage of 1.23 V vs the standard hydrogen electrode (SHE) at pH ) 03 and eventually is equivalent to the electrochemical charging up to ∼4.2 V vs Li+/Li in a nonaqueous solution. The fully delithiated LixNiO2 is also obtained by a chemical reaction in an aqueous solution of acid, as in the case for LixMn2O4.2 Actually, Arai et al.4,5 prepared LixNiO2 with x e 0.1 using the reaction of LiNiO2 with a sulfuric acid and reported that the electrochemical discharge curve (lithiated * To whom correspondence should be addressed. E-mail: e1089@ mosk.tytlabs.co.jp. Phone: +81-561-71-7698. Fax: +81-561-63-6137. † Toyota Central Research and Development Laboratories, Inc. ‡ Paul Scherrer Institut. § Babes-Bolyai University. | Present address: Muon Science Laboratory, Institute of Materials Structure Science, KEK, 1-1 Oho, Tsukuba, Ibaraki 305-0801, Japan.
process) for the chemically delithiated LixNiO2 compound with x ) 0.1 almost trace that for the pristine LiNiO2. This indicates that the chemically delithiated process for LixNiO2 is almost the same to the electrochemically process, although the disproportionation reaction described in eq 1 always involves a dissolution from a particle, and consequently, the solid-liquid interface is renewed during the reaction. In our previous paper,6 the magnetic nature for the fully delithiated LixMn2O4 and LixNiO2 samples, which are prepared by both electrochemical (EC) and chemical (C) reactions, was examined by a magnetic susceptibility (χ) measurement. Although the magnetism for C-Li0.07Mn2O4 is almost identical to that for EC-Li0.03Mn2O4, the magnetism for C-Li0.01NiO2 is very different from that for EC-Li0.05NiO2. That is, the magnitude of χ for C-Li0.01NiO2 is considerably large, if we assume that Ni4+ ions are nonmagnetic (t62g), as in the case for EC-Li0.05NiO2. Furthermore, the effective magnetic moment (µeff) for C-Li0.01NiO2 () 1.43 µB) is twice the size of µeff for ECLi0.05NiO2 () 0.71 µB). This means that the average valence of Ni ions (Vave,Ni) is not simply determined by the Li/Ni ratio for C-LixNiO2.4 According to past structural analyses on LixMn2O4,2 the oxygen stacking sequence for LixMn2O4 is maintained in a cubic close-packed (CCP) structure down to x ≈ 0. On the other hand, that for both C-LixNiO24,5 and EC-LixNiO27-9 partially changes from the CCP structure to a hexagonal close-packed (HCP) structure below x ≈ 0.1. Here, the proton insertion materials such as NiOOH and CoOOH have the HCP structure with a ABAB stacking sequence, while the lithium insertion materials such as LiNiO2 (NiOOLi) and LiCoO2 (CoOOLi) have the CCP structure with a ABCABC stacking sequence. Therefore, such a structural variation from CCP to HCP leads to the question whether H+ ions are inserted into the C-Li0.01NiO2 sample. In other words, there is a possibility that the inserted H+ ions
10.1021/jp1007818 2010 American Chemical Society Published on Web 04/16/2010
Structure and Magnetism for LixNiO2 increase µeff for C-Li0.01NiO2, although Arai et al.5 proposed a “proton-free model” for C-LixNiO2 by an X-ray photoelectron spectroscopy (XPS) analysis. The existence of H+ ions is usually detected by a solid state 1H NMR. However, albeit not impossible, it is very difficult to determine the content of H+ ions, because the 1H nucleus (I ) 1/2) gives a broad NMR power pattern in hertz due to a dipolar coupling to the paramegnets.10 Therefore, we have investigated the structural and magnetic nature for both EC-LixNiO2 and C-Li0.01NiO2 samples by X-ray diffraction (XRD), χ, thermogravimetry (TG), pyrolysis gas chromatography/mass spectrometry (Py-GC/MS) measurements to clarify whether H+ ions are inserted into the C-Li0.01NiO2 sample. Furthermore, we have performed muon-spin rotation/ relaxation (µSR) measurements because µSR is a powerful technique to detect local magnetic fields caused by nuclear and electronic origin and corresponds to the volume fraction of the magnetic phases in the sample.11 Indeed, Ariza et al.12 reported that zero-field µSR spectrum for C-Li[Li1/3Mn5/3]O4 (H+-MnO2) shows a larger relaxation rate than that for C-LiMn2O4 at 100 K and speculated the presence of H+ ions in the C-Li[Li1/3Mn5/3]O4 sample. Experimental Methods Two different samples of LiNiO2 (Lot A and B) were synthesized by a conventional solid-state reaction technique using reagent grade LiNO3 (Li(OH) · H2O for Lot B) and NiCO3 (NiO for Lot B) powders. The reaction mixture was pressed into a pellet of 23 mm diameter and ∼5 mm thickness and then heated at 650 °C in oxygen flow for 12 h. The obtained powder was crushed, pressed into a pellet again, and finally fired at 750 °C in an oxygen flow for 12 h. The products were characterized by a powder XRD (RINT-2200, Rigaku Co. Ltd., Japan) analysis and an electrochemical charge/discharge test. The LixNiO2 samples with x < 1 were prepared by two different methods; an electrochemical reaction in a nonaqueous lithium cell and a chemical reaction, as previously reported.2,4,5 For the electrochemical reaction, the pressed LiNiO2 powder (Lot A and B) and a Li-metal sheet were used as a working and counter electrode, respectively. The electrolyte was 1 M LiPF6 dissolved in ethylene carbonate (EC)/diethyl carbonate (DMC) (3:7 volume ratio) solution. To avoid the signals from conducting additives and a binder, the electrodes were entirely made from LiNiO2 powder. The Li/LiNiO2 cells were operated with a rate of 0.057 mA · cm-2 (constant current mode) at 298 K (25 °C). For the chemical reaction, 2 g of powder of LiNiO2 (Lot A) was immersed in 100 mL of 1 M HNO3 and then stirred at room temperature for 24 h. The initial molar ratio of H+/ LiNiO2 was ∼4.9. Such large ratio was selected so as to complete the following disproportionation reaction
2LiNi3+O2 + 4H+ f Ni4+O2 + Ni2+ + 2Li+ + 2H2O (2) The product formed after the reaction was filtered and dried at 40 °C in an air-oven in order to avoid the decomposition of the sample. The Li/Ni ratio was determined by an inductively coupled plasma (ICP) atomic emission spectral (AES) analysis (CIROS 120, Rigaku Co. Ltd., Japan) and was found to be 0.01. TG was performed using a thermal analyzer (TGA-50, Shimadzu Co. Ltd., Japan). Py-GC/MS (Py-2010D, Frontier Lab Co. Ltd., Japan) was used to confirm the existence of H+ ions. Direct current χ measurements were carried out using a superconducting quantum interference device magnetometer
J. Phys. Chem. C, Vol. 114, No. 18, 2010 8627 (MPMS, Quantum Design) in the temperature (T) range between 5 and 350 K under the magnetic field H e 10 kOe. Electron spin resonance (ESR) spectra were recorded by a ESP300E (Bruker) spectrometer in the temperature range between 100 and 300 K. The gyromagnetic ratio of Ni ions (g) was determined with respect to a MnO/MgO standard. µSR experiments were performed at the Paul Scherrer Institut, Switzerland. Here, we wish to describe the features of µSR briefly. Muon is a spin 1/2 particle with a gyromagnetic ratio γµ/2π ) 13.554 kHz/Oe. When polarized muons are implanted into a material, the muon-spin processes by the local magnetic field of the material. The unstable muons soon decay into positrons (the muon lifetime is 2.2 µs). The decay positron is emitted preferentially along the muons pin direction. By collecting several million positrons as a function of the evolution time, one can construct the time dependence of the muon-spin polarization [A0P(t)], which reflects the magnitude of the magnetic field at the muon site. Zero-field (ZF) µSR is a very sensitive method for detecting weak internal magnetism that arises due to ordered magnetic moments or random fields that are static or fluctuating with time. Transverse-field (TF) µSR involves the application of an external magnetic field perpendicular to the initial direction of the muon-spin polarization. More details about µSR technique are found in elsewhere.11 For the µSR measurements, the powder LixNiO2 samples were pressed into a disk of about 15 mm diameter and 1 mm thickness and subsequently placed into a fork-type low background sample holder. The LixNiO2 powders were removed from the cells in a He-filled glovebox just before the µSR measurements. The above procedure is essentially the same to that of our recent µSR works on LixNiO213,14 and LixCoO2.15,16 Results Electrochemical and Structural Properties. The XRD analysis shows that both LiNiO2 samples (Lot A and B) have a layered structure with space group of R3jm, in which Li+ and Ni3+ ions are located at the 3b and 3a sites, respectively. The lattice parameters in hexagonal setting are calculated as ah ) 2.8749(1) Å, ch ) 14.2033(5) Å for Lot A and ah ) 2.8785(1) Å, ch ) 14.1940(1) Å for Lot B. It is widely accepted that a disordered rocksalt-type LiNiO2 (Fm3jm) domain, which is electrochemically inactive, is easily formed in the LiNiO2 (R3jm) phase.17 This is because the LiNiO2 (Fm3jm) domain is thermodynamically stable at high temperatures above 750 °C. The rocksalt-type LiNiO2 (Fm3jm) domain is also known to prevent a structural change from the CCP to HCP structure.5,9 According to a Rietveld analysis using RIETAN-2000,18 the amount of Ni ions at the 3b site (z in Li1-zNi1+zO2) is estimated as z e 0.03 for both samples. Figure 1 shows the charge curves of Li/LiNiO2 cells for the χ and µSR measurements. The charge curves for all the samples exhibit three plateaus around 3.67, 4.02, and 4.20 V, indicating that the crystal structure of LixNiO2 varies as a function of x; as x decreases from 1, a rhombohedral (R3jm) phase is stable down to x ) 0.75, whereas a monoclinic (C2/m) phase in the x range between 0.75 and ∼0.45. Then, a rhombohedral phase appears again with 0.45 g x > 0.25, and finally two rhombohedral phases coexist with 0.25 g x > ∼ 0.1.17 The electrochemical properties for the present LiNiO2 (Lots A and B) are essentially the same to the result for the nearly stoichiometric LiNiO2.17 Here, the initial charge curve of the cell using the electrode mix, which consists of 88 mass % LiNiO2, 6 mass % acetylene black, and 6 mass % PVdF dispersed in N-methyl2-pyrrolidone, is also shown for comparison. In spite of the
8628
J. Phys. Chem. C, Vol. 114, No. 18, 2010
Mukai et al. TABLE 1: Lattice Parametersa of ah and ch Axes for the EC-Li0.10NiO2 and C-Li0.01NiO2 Samples lattice parameter/Å sample this work
composition
method
Li0.10NiO2 nonaqueous Li0.01NiO2 aqueous
Croguennec Ni1.02O2 et al.8,9 Arai et al.5
Figure 1. The charge curves of Li/LiNiO2 cells operated at 298 K (25 °C) for the χ and µSR measurements. To avoid the signals from other components, the electrodes for χ and µSR measurements were entirely made from LiNiO2 powder. The charge curve of the cell using an electrode mix, which consists of 88 mass % LiNiO2, 6 mass % conductive carbon, and 6 mass % binder, is also shown for comparison. The applied current density of the cells for χ (or µSR) and electrode mix was 0.057 and 0.17 mA · cm-2, respectively. Note that the Li/Ni ratio determined by ICP-AES analysis is in good agreement with the result of an electrochemical reaction (EC).
Figure 2. XRD patterns of the fully delithiated LixNiO2 samples: (a) EC-Li0.10NiO2 and (b) C-Li0.01NiO2. The enlarged XRD patterns in the 2θ range between 16 and 22° are also shown in the left side. Four phases for (a) are denoted as E1, E2, E3, and E4, while those for (b) are denoted as C1, C2, C3, and C4, respectively. The open circle, closed circle, and closed triangle indicate the E1, E4, and C4 phases, respectively.
absence of conducting additives and binder in the positive electrode, the charge curves for all the samples are almost the same to the curve for the electrode mix. The Li-extracted reaction is, therefore, successfully achieved for all the LixNiO2 samples. The Li/Ni ratios were examined by the ICP-AES analysis after the χ and µSR measurements. In this paper, we use the Li/Ni ratios determined by the ICP-AES analysis. Figure 2a shows the XRD pattern for the highest delithiated sample by an electrochemical reaction (EC-Li0.10NiO2), and Figure 2b does that by a chemical reaction (C-Li0.01NiO2). The Li/Ni ratio for the electrochemically prepared sample was estimated as 0.10 by the ICP-AES analysis. As reported previously,5,9 a single-phase was not obtained for both samples, and consequently, a detailed structural analysis by a Rietveld refinement cannot be performed. As seen in the 2θ range between 16 and 22°, there are at least four phases in both samples, that is, E1, E2, E3, and E4 phases for the EC-Li0.10NiO2 sample, and C1, C2, C3, and C4 phases for the C-Li0.01NiO2 sample. The d-value for each phase is calculated as 4.84, 4.70, 4.48, and 4.38 Å for the E1, E2, E3, and E4 phases and 7.36,
nonaqueous
Li0.04NiO2 aqueous
phase
ah
chb
packing
E1 E3 E4 C3 C4 R3
2.833 2.824 2.825 2.817 2.818 2.815
14.407 13.352 13.051 13.274 13.042 13.363
CCP CCP HCP CCP HCP CCP
R4 B1 B2 B3
2.815 2.817 2.819 2.818
13.039 13.820 13.318 13.101
HCP CCP CCP HCP
a The lattice parameters for the E3 and C3 phases were calculated by a least-squares method by using more than 11 diffraction lines, while those for the E1, E4, and C4 phases by 5 diffraction lines. b The length of the ch axis for the E4, C4, H4, and B3 phases was multiplied by 3 for comparison.
4.68, 4.43, 4.36 Å for the C1, C2, C3, and C4 phases, respectively, by using the relation d ) λ/2 sin θ. The lattice parameters of ah and ch axes for the E3 and C3 phases were calculated by a least-squares method by using more than 11 diffraction lines, while those for the E1, E4, and C4 phases by 5 diffraction lines. Here, the crystal structure of the fully delithiated LixNiO2 was studied by Croguennec et al. using ECLixNiO2 samples,7-9 and Arai et al. using C-LixNiO2 samples4,5 (see Table 1). Although the lattice parameters for the E4 and C4 phases are ambiguous due to the lack of multiple diffraction lines, on the basis of the ch axis length, the E3 (E4) phase is almost identical to the R3 (H4) phase,7-9 while the C3 (C4) phase corresponds to their B2 (B3) phase.5 Therefore, it is considered that the E3 and C3 phases belong to the CCP structure with a ABCABC stacking sequence, whereas the E4 and C4 phases the HCP structure with a ABAB stacking sequence. Since the XRD peaks for the E3 (C3) phase are most intense among those of four phases, the majority of LixNiO2 is assigned to be a CCP structure for both samples. It is difficult to determine the crystal structure and the lattice parameters of ah and ch axes for the E2, C1, and C2 phases, because the diffraction lines above 2θ ) 20° were not seen for these samples. To our knowledge, the C1 phase, having the largest d-value with 7.36 Å (2θ ) 12.01°), has never been reported in LixNiO2. The large d-value implies that a small amount of H2O molecule would be inserted into the NiO2 interlayer only for the C-LixNiO2 sample, as in the case for γ-NiOOH (d-value ≈ 7 Å).19 Macroscopic Magnetism by χ Measurements. Figure 3 shows the T dependence of (a) χ and (b) χ-1 for the EC-LixNiO2 (open circles) and C-Li0.01NiO2 (closed circle) samples. χ was measured in field-cooling (FC) mode with H ) 10 kOe. The Ni3+ ions in LiNiO2 are known to be in a low-spin state with t62ge1g (S ) 1/2).20 Actually, the χ(T) curve for the x ) 1 sample exhibits a rapid increase below ∼100 K, indicating the presence of localized moments of the Ni3+ ions. For the EC-LixNiO2 samples with x < 1, the magnitude of χ at low T decreases with decreasing x. This means that the amount of nonmagnetic Ni4+ ions with t62g (S ) 0) increases with decreasing x. On the other hand, the χ(T) curve for C-Li0.01NiO2 locates between those for EC-Li0.88NiO2 and EC-Li0.75NiO2, in contrast to the fact that the Li+ ions are almost fully delithiated. To speculate the Vave,Ni, we attempted to fit the χ(T) curve with a Curie-Weiss formula in the T range between 200 and 350 K
Structure and Magnetism for LixNiO2 2 Nµeff χ) + χ0 3kB(T - Θp)
J. Phys. Chem. C, Vol. 114, No. 18, 2010 8629
(3)
where kB is the Boltzmann’s constant, T is the absolute T, Θp is the Weiss temperature, N is the number density of Ni ions, µeff is the effective magnetic moment of Ni ions, and χ0 is the T-independent susceptibility. Figure 4 shows the x dependences of (a) Θp, (b) µeff, and (c) χ0 for the EC-LixNiO2 and C-Li0.01NiO2 samples. The Θp(x) curve for the EC-LixNiO2 sample exhibits a broad minimum around x ) 0.5 (Θp ) 12 K). Since Θp ) 45 K for the C-Li0.01NiO2 sample, the predominant interaction between the Ni moments is still ferromagnetic (FM). As seen in Figure 4b, µeff for EC-LixNiO2 decreases monotonously with decreasing x. The solid line in Figure 4b represents the predicted µeffpre for LixNiO2, assuming that Ni3+ ions are in a low-spin state with t62ge1g (S ) 1/2), Ni4+ ions with t62g (S ) 0), and g ) 2.13. Here, the g factor was estimated from the ESR spectra for LiNiO2 using the following relation
hVr ) gµBHr
(4)
where h is the Planck constant, νr is the resonance frequency, µB is the Bohr magneton, and Hr is the resonance field. Although the obtained µeff is slightly larger than µeffpre, as in the case for LiNixCo1-xO2,21 the x dependence of µeff for EC-LixNiO2 is quantitatively explained by the decrease in the number density of magnetic Ni3+ ions. In other words, the electrochemically delithiated process for LixNiO2 is roughly understood by the oxidation reaction of Ni3+ (S ) 1/2) f Ni4+ (S ) 0) + e-. As reported previously,6 µeff () 1.43 µB) for the C-Li0.01NiO2 sample is considerably large compared with µeff for ECLi0.10NiO2 and is comparable to µeff for EC-Li0.50NiO2. This indicates the presence of larger number of the Ni3+ ions in the C-Li0.01NiO2 sample than the prediction from the Li/Ni ratio. To clarify the magnetic nature of the C-Li0.01NiO2 sample, a µSR experiment was performed. Microscopic Magnetism by µSR Measurements. Figure 5 shows the T dependence of the normalized weak TF asymmetry
Figure 4. Variation of (a) Weiss temperature (Θp), (b) effective magnetic moment (µeff), and (c) temperature (T)-independent susceptibility (χ0) for the EC-LixNiO2 (open circles) and C-Li0.01NiO2 (closed circle) samples. µeff and Θp were estimated by fitting the χ(T) curves in the T range between 200 and 350 K using eq 3. χ0 was used for fitting due to a convex χ(T) curve particularly at x e 0.75. The solid line in (b) is the predicted µeffpre for LixNiO2 using the assumption that Ni3+ are in the low-spin state with S ) 1/2, Ni4+ ions also in the lowspin state with S ) 0 and g ) 2.13.
Figure 5. Temperature dependence of the normalized wTF asymmetry (NATF) for the (a) EC-LixNiO2 with x ) 0.60, 050, and 0.10 and (b) C-Li0.01NiO2 samples. The applied magnetic field was HwTF ) 30 Oe. The solid lines are a guide to the eye. Tm is the transition temperature at which NATF ) 0.5. Mi symbolizes the magnetic phase, and PM the paramagnetic phase.
Figure 3. Magnetic susceptibility (a) χ and (b) χ-1 for the EC-LixNiO2 (open circles) and C-Li0.01NiO2 (closed circle) samples. χ was measured in FC mode with H ) 10 kOe.
(NATF) for (a) EC-LixNiO2 with x ) 0.60, 050, and 0.10 and (b) C-Li0.01NiO2 samples obtained in an applied magnetic field (HwTF )) 30 Oe. Here, NATF is defined by NATF ) ATF/ATF, max ) ATF/ ∼0.25 and is roughly proportional to the volume fraction of paramagnetic (PM) phases in the sample. In other words, when NATF ) 1, the whole sample is in a PM state, but, when NATF ) 0, the whole sample is in a magnetic phase, such as FM, antiferromagnetic, ferrimagnetic, or spin-glass-like phase. For
8630
J. Phys. Chem. C, Vol. 114, No. 18, 2010
Mukai et al.
Figure 6. ZF µSR time spectra for the (a) EC-LixNiO2 samples with x ) 0.50 and 0.10, and (b) C-Li0.01NiO2 sample. The solid lines are fitting results using eq 5.
the EC-Li0.10NiO2 sample, as T decreases from 50 K, NATF ∼ 1 down to ∼20 K, then slightly decreases by ∼0.1 with further lowering T. This is consistent with the fact that the majority of the Ni ions is in a nonmagnetic 4+ state with S ) 0. On the other hand, as T decreases from 30 K the NATF (T) curve for the EC-LixNiO2 samples with x ) 0.60 (0.50) exhibits a steplike decrease from 1 to 0 at 10 K (4 K), demonstrating the existence of a sharp magnetic transition. The magnetic transition T (Tm) was determined as the T, at which NATF ) 0.5. Since the NATF(T) curve reaches almost 0 below the vicinity of Tm, the whole sample enters into the magnetic phase below Tm. For the C-Li0.01NiO2 sample, as T decreases from 50 K, the NATF(T) curve shows two step-like drops at ∼15 K and below 10 K, although NATF ) 0.32 even at 1.8 K. This indicates that; (i) the magnetic Ni ions still exist in the sample as expected from Figure 4(b), (ii) there are at least two different magnetic phases denoted as M1 and M2 in the sample, and (iii) 1/3 of the sample is still PM even at 1.8 K. Figure 6 shows the ZF-µSR time spectra for the (a) ECLixNiO2 samples with x ) 0.5 and 0.10 and (b) C-Li0.01NiO2 sample. Note that all the samples are in the PM state (see parts a and b of Figure 5). In the PM state, muon-spins should be depolarized by the randomly oriented nuclear magnetism, not by the electronic magnetism. The ZF-µSR spectrum is, thus, fitted by a dynamic Kubo-Toyabe function GDGKT(t, ∆, ν)22
A0PZF(t) ) AKTGDGKT(t, ∆, V) + AMexp(-λMt)
(5)
where A0 is the empirical maximum muon decay asymmetry, AKT and AM are the asymmetries of Kubo-Toyabe (paramagnetic) and residual magnetic phases, ∆ is the static width of the local frequencies at the disordered sites, ν is the field distribution rate, and λM is the relaxation rate. When ν ) 0, GDGKT(t, ∆, ν) is the static Gaussian Kubo-Toyabe function GKTzz (t, ∆) given by KT Gzz (t, ∆) ) 1/3 + 2/3(1 - ∆2t2)exp(∆2t2/2)
(6)
Figure 7. (a) Thermogravimetric curve and (b) Py-GC/MS spectrum for the C-Li0.01NiO2 sample. Heating rates are 10 K · min-1 for (a) and 20 K · min-1 for (b).
The ZF-µSR spectrum for EC-Li0.10NiO2 shows a slow relaxation, indicating that muon-spins are mainly depolarized by the residual nuclear magnetism of 6Li and 7Li. On the other hand, the ZF-µSR spectrum for the C-Li0.01NiO2 sample shows a large relaxation. The ∆, which corresponds to the internal magnetic field, is calculated as 0.075(2) µs-1 at 12.5 K for the EC-Li0.10NiO2 sample, while that is calculated as 0.355(3) µs-1 at 15 K for the C-Li0.01NiO2 sample. Here, the C-Li0.01NiO2 sample is prepared by digesting the LiNiO2 into a HNO3 solution and is almost fully delithiated. This implies that, even if we consider the effect of the Ni moment of C-Li0.01NiO2 (µeff for C-Li0.01NiO2 is ∼1.4 µB, which is comparable to µeff for ECLi0.50NiO2), there is the other origin for the large relaxation rate only for the C-Li0.01NiO2 sample. Since the nuclear magnetic moment caused by both O and Ni nuclei is negligibly small, the large relaxation rate for the C-Li0.01NiO2 sample is most likely due to the nuclear magnetic moment of the 1H nucleus. In other words, µSR result suggests the presence of H+ ions in the C-Li0.01NiO2 sample, although the past work for C-LixNiO2 proposed the proton-free model.4,5 Here, we wish to emphasize that µSR is a bulk probe, i.e., µSR signals roughly correspond to the volume fraction of the magnetic phases in the sample. Thermal Analysis. To confirm the absence/existence of oxygen deficiency and H+ ions, TG and Py-GC/MS analyses were carried out. Figure 7a shows the TG curve for the C-Li0.01NiO2 sample. As T increases from ambient T, the TG curve shows a slight decrease above ∼380 K, then exhibits a drastic decrease around 473 K, and finally keeps nearly constant value until 873 K. Here, after the acid reaction, the C-Li0.01NiO2 sample was dried only at 40 °C in an air oven in order to avoid the decomposition of the sample. Thus, the decrease around 380 K is most likely to the absorbed and/or adsorbed water of the sample. The weight loss at 1273 K [m(1273 K) - m(300 K)] is ∼82.3%, which well agrees with the expected change (17.6%) by the following oxygen evolution reaction
NiO2 f NiO + 1/2O2
(7)
This suggests the absence of the oxygen deficiency in the C-Li0.01NiO2 sample, as reported previously.4,5 Note that it is
Structure and Magnetism for LixNiO2
J. Phys. Chem. C, Vol. 114, No. 18, 2010 8631
difficult to distinguish the absence/existence of H+ ions in the C-Li0.01NiO2 sample by the TG curve, since the weight loss is calculated as ∼81.5% even for the HNiO2 (NiOOH) composition. On the other hand, the two spectra, which are identified as the O2+ (mass/charge ) 32) and H2O+ (mass/charge ) 18) ions by the mass spectrum at 522 K are clearly observed in the PyGC/MS spectrum (see Figure 7b). This unambiguously demonstrates the presence of H+ ions in the C-Li0.01NiO2 sample, being consistent with the result by µSR measurements. Here, it should be noted that the trend for the H2O+ spectrum is almost the similar to that for the O2+ spectrum; i.e., both spectra show two peaks at ∼490 and 535 K. This indicates that the structurally bonded and/or intercalated water exists in the C-Li0.01NiO2 crystalline lattice. Indeed, the TG curve for the Ni(OH)2 compound showed two weight loss regions in the T ranges of 323-363 K and 453-723 K, which are attributed to the dehydration reactions of absorbed/adsorbed water and structurally bonded/intercalated water, respectively.23 Therefore, the decomposition reaction for the C-Li0.01NiO2 sample is represented by
380-470 K . HxLi0.01NiO2 yH2O f HxLi0.01NiO2 + yH2O
(8)
470-870 K HxLi0.01NiO2 f Li0.01NiO + x/2H2O + (1 - x/4)O2
(9) Discussion According to the χ measurements, µeff for the C-Li0.01NiO2 sample was estimated as µeff ) 1.43 µB and was almost the same to that for the EC-Li0.5NiO2 sample (µeff ) 1.39 µB), indicating the decrease in Vave,Ni from the tetravalent state (Ni4+O2). Furthermore, µSR measurements and Py-GC/MS analysis clearly demonstrated the existence of H+ ions in the C-Li0.01NiO2 crystalline lattice. Note that we prepared the C-Li0.07Mn2O4 sample by drying at 60 °C in an air-oven after the acid reaction.6 However, χ measurements indicated that µeff () 3.85 µB) for the C-Li0.07Mn2O4 sample is very close to that (3.80 µB) for the EC-Li0.03Mn2O4 sample.6 Therefore, it is most likely that not the absorbed and/or adsorbed water but the structurally bonded and/or intercalated water (H+ ions) increases the µeff for the C-Li0.01NiO2 sample. As seen in Figure 4b, µeff for the EC-LixNiO2 samples decreases with decreasing x as expected. The total amount of H+ ions in the C-Li0.01NiO2 sample is, thus, evaluated as ∼0.5, i.e., H∼0.5Li0.01NiO2, by using the linear relationship between µeff and x in the EC-LixNiO2 sample. In other words, the vacant tetrahedral or octahedral sites in C-Li0.01NiO2 are partially occupied by H+ ions. In this calculation, we assume that both Ni3+ and Ni4+ ions are in the low-spin state, since Ni ions occupy the octahedral site in both CCP and HCP structure. Actually, the g-factor (2.10) for C-Li0.01NiO2 is almost temperature-independent above 100 K, as for LiNiO2 (g ) 2.13). Moreover, Demorgues et al.24 showed by extended X-ray absorption fine structure that the Ni3+O6 octahedra in both β-NiOOH and γ-NiOOH (specifically H0.20Na0.10K0.20Ni0.70Co0.30O2 · 0.5H2O) exhibit a Jahn-Teller distortion. This supports that the Ni3+ ions in the C-Li0.01NiO2 sample are also in a low-spin state with t62ge1g (S ) 1/2). The results by wTF-µSR measurements also support the above consideration. That is, as seen in Figure 5b, three different magnetic phases of M1, M2, and PM are found from the T dependence of NATF, indicating the inhomogeneity of the sample.
The volume fraction of the M1, M2, and PM phases is approximately 40, 30, and 30%, respectively. Since Tm of the M1 (M2) phase is almost the same to that for the EC-Li0.60NiO2 (EC-Li0.50NiO2) sample, the Vave,Ni for the C-Li0.01NiO2 sample is roughly estimated as ∼3.6 (4 - 0.6 × 0.4 - 0.5 × 0.3). This is very consistent with the result of the χ measurements (Vave, Ni ∼ 3.5). As described in the introduction, Arai et al.4,5 proposed the proton-free model for the chemically delithiated LixNiO2 compound with x e 0.1. Although the origin of the difference between the present and past results is currently unknown, the amount of Ni ions in the Li layer (z) would participate in such a difference, since the Li1-zNi1+zO2 compound with z ) 0.07 was reported to keep CCP structure even for the fully delithiated state.7 Moreover, they employed the X-ray photoelectron spectroscopy (XPS) analysis for studying the oxidation state of Ni ions and concluded the proton-free.5 However, it is thought to be difficult to detect the H+ ions in a crystalline lattice by the surface probe such as XPS and Fourier transform infrared spectroscopy (FT-IR) analyses. XRD measurements indicated that the C-Li0.01NiO2 sample has at least four phases with different d-values (see Figure 2b). Although the total amount of H+ ions in the C-Li0.01NiO2 sample is estimated as ∼0.5, the location of H+ ions in the phase and lattice is currently unclear. We expect that further neutron diffraction measurement provides crucial information on the location of the H+ ions. It should be noted that the major phase of C-Li0.01NiO2 is the CCP structure (C3 phase) and not the HCP structure (C4 phase). Croguennec et al. reported that the EC-LixNiO2 phase in the HCP structure (H4) is thermodynamically unstable and slowly converts into a new LixNiO2 phase in the CCP structure (R3′) with decreasing the distance of NiO2 interlayer.7-9 Since the interlayer distance of the C3 phase ranges at the middle of the C4 (or E4) and E3 phases, the C4 phase in the HCP structure is thought to transform into the C3 phase in the CCP structure. Finally, we wish to comment on the possibility of the existence of H+ ions in the EC-Li0.10NiO2 sample, because an electrochemical charging at high voltages above 4.2 V vs Li+/ Li sometimes generates H+ ions resulting from the decomposition of a nonaqueous electrolyte. Indeed, Robertson and Bruce reported25 that the abnormal capacity around 4.5 V of a nonaqueous Li/Li2MnO3 cell is attributed to an exchange reaction between Li+ and H+ ions. For the EC-LixNiO2 samples described here, the values of µeff decreases with decreasing x, as expected (see Figure 4b). Also, the temperature dependence of NATF shows that ∼90% of the sample is in the PM state even at 1.8 K (see Figure 5a). The proton insertion reaction, hence, does not occur in the present EC-Li0.10NiO2 samples. Conclusion The magnetic and thermal analyses clearly showed that H+ ions exist in the C-Li0.01NiO2 crystalline lattice. That is, the expected composition of C-Li0.01NiO2 is represented as H∼0.5Li0.01NiO2. Although the mechanism of proton insertion for LixNiO2 is still unclear, we expect that further studies on LixNiO2 and NiOOH provide crucial information on the differences and similarities of electrochemical reaction between nonaqueous and aqueous solutions. We also emphasize that the detection of H+ ions in the actual LIB is significantly important for optimizing the cycle and storage performances at elevated temperatures above 55 °C. The magnetic susceptibility and muon-spin rotation/relaxation measurements are found to be a powerful technique for studying such subjects.
8632
J. Phys. Chem. C, Vol. 114, No. 18, 2010
Acknowledgment. This work was performed at the Swiss Muon Source, Paul Scherrer Institut (PSI), Villigen, Switzerland. We thank the staff of PSI for help with the µSR experiments. We also appreciate T. Ohzuku, K. Ariyoshi, and S. Kohno of Osaka City University for preparation of LiNiO2 and discussion through this work. We wish to thank Y. Kondo of TCRDL for ICP-AES analysis and M. Yamamoto of TCRDL for Py-GC/ MS analysis. K.M., Y.I., and J.S. are partially supported by the KEK-MSL Inter-University Program for Overseas Muon Facilities. This work is supported by Grant-in-Aid for Scientific Research (B), 1934107, MEXT, Japan. Supporting Information Available: XRD patterns in log scale for the EC-Li0.10NiO2 and C-Li0.01NiO2 samples. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Ohzuku, T.; Ueda, A. Solid State Ionics. 1994, 69, 201–211. (2) Hunter, J. C. J. Solid State Chem. 1981, 39, 142–147. (3) Pourbaix, M. Atlas of Electrochemical Equilibria in Aqueous Solutions; Pergamon: New York, 1966. (4) Arai, H.; Sakurai, Y. J. Power Sources 1999, 81-82, 401–405. (5) Arai, H.; Tsuda, M.; Saito, K.; Hayashi, M.; Takei, K.; Sakurai, Y. J. Solid State Chem. 2002, 163, 340–349. (6) Mukai, K.; Sugiyama, J. Chem. Lett. 2009, 38, 944–945. (7) Croguennec, L.; Pouillerie, C.; Delmas, C. J. Electrochem. Soc. 2000, 147, 1314–1321. (8) Croguennec, L.; Pouillerie, C.; Delmas, C. Solid State Ionics 2000, 135, 259–266. (9) Croguennec, L.; Pouillerie, C.; Mansour, A. N.; Delmas, C. J. Mater. Chem. 2001, 11, 131–141.
Mukai et al. (10) Paik, Y.; Bowden, W.; Richards, T.; Grey, C. P. J. Electrochem. Soc. 2005, 152, A1539–A1547. (11) Schenck, A. Muon Spin Rotation Spectroscopy Principles and Applications in Solid State Physics; Adam Hilger: Bristol and Boston, 1985, and references cited therein. (12) Ariza, M. J.; Jones, D. J.; Rozie`re, J.; Lord, J. S.; Ravot, D. J. Phys. Chem. B 2003, 107, 6003–6011. (13) Sugiyama, J.; Mukai, K.; Ikedo, Y.; Russo, P. L.; Nozaki, H.; Andreica, D.; Amato, A.; Ariyoshi, K.; Ohzuku, T. Phys. ReV. B 2008, 78, 144412. (14) Mukai, K.; Sugiyama, J.; Ikedo, Y.; Russo, P. L.; Andreica, D.; Amato, A.; Ariyoshi, K.; Ohzuku, T. J. Power Sources 2009, 189, 665– 668. (15) Mukai, K.; Sugiyama, J.; Ikedo, Y.; Nozaki, H.; Simomura, K.; Nishiyama, K.; Ariyoshi, K.; Ohzuku, T. J. Power Sources 2007, 174, 711– 715. (16) Mukai, K.; Ikedo, Y.; Nozaki, H.; Sugiyama, J.; Nishiyama, K.; Andreica, D.; Amato, A.; Russo, P. L.; Ansaldo, E. J.; Brewer, J. H.; Chow, K. H.; Ariyoshi, K.; Ohzuku, T. Phys. ReV. Lett. 2007, 99, 087601. (17) Ohzuku, T.; Ueda, A.; Nagayama, M. J. Electrochem. Soc. 1993, 140, 1862–1870. (18) Izumi, F; Ikeda, T. Mater. Sci. Forum 2000, 198, 321–324. (19) Sac-Epe´e, N.; Palacı´n, M. R.; Beaudoin, B.; Delahaye-Vidal, A.; Jamin, T.; Chabre, Y.; Tarascon, J.-M. J. Electrohcem. Soc. 1997, 144, 3896–3907. (20) Goodenough, J. B.; Wickham, D. G.; Croft, W. J. J. Phys. Chem. Solids 1958, 5, 107–116. (21) Mukai, K.; Sugiyama, J.; Ikedo, Y.; Brewer, J. H.; Ansaldo, E. J.; Morris, G. D.; Ariyoshi, K.; Ohzuku, T. J. Power Sources 2007, 174, 843– 846. (22) Hayano, R. S.; Uemura, Y. J.; Imazato, J.; Nishida, N.; Yamazaki, T.; Kubo, R. Phys. ReV. B 1979, 20, 850–859. (23) Mani, B.; de Neufville, J. P. J. Electrochem. Soc. 1988, 135, 800– 803. (24) Demourgues, A.; Gautier, L.; Chadwick, A. V.; Delmas, C. Nucl. Instr. Meth. B 1997, 133, 39–44. (25) Robertson, A. D.; Bruce, P. G. Chem. Mater. 2003, 15, 1984–1992.
JP1007818
