We evaluate the utilities of fluorescence-yield (FY) modes in soft X-ray absorption spectroscopy (XAS) of several cathode materials for Li-ion batteries. In the case of total-FY (TFY) XAS for LiNi0.5Mn1.5O4, the line shape of the Mn L3-edge XAS was largely distorted by the self-absorption and saturation effects, while the distortions were less pronounced at the Ni L3 edge. The distortions were suppressed for the inverse-partial-FY (IPFY) spectra. We found that, in the cathode materials, the IPFY XAS is highly effective for the Cr, Mn, and Fe L edges and the TFY and PFY modes are useful enough for the Ni L edge which is far from the O K edge.
Electronic-structure analyses for the electrode materials of Li-ion batteries (LIBs) are essential to understand the charge-discharge properties, which is capable of leading to improvements of the battery performance. The 3d transition metals (TMs) are the major elements in the cathode materials for LIBs, such as TM oxides and phosphates. The TMs in the LIBs cathode materials experience the oxidation and reduction during charging/discharging process (i.e. Li-ion extraction/insertion). The redox chemistry determines the electrochemical properties of the electrode materials. Thus, element-selective electronic-structure analyses are indispensable for the application of LIBs.
In LIB research, TM K-edge X-ray absorption spectroscopy in hard X-ray region is widely used to investigate the electronic structures of cathode materials.1,2 The redox reactions of TMs can be clarified by the edge/peak shift in the TM K-edge absorption spectra. However, the TM K-edge X-ray absorption spectroscopy is less sensitive to the 3d electronic states of TMs than the TM L2,3-edge (i.e. the 2p-3d transition) X-ray absorption spectroscopy, because the 1s-4p dipole transition is dominant at the TM K edges.
To obtain the detailed information of the 3d orbitals, i.e. electronic-structure parameters such as charge-transfer energy, soft X-ray absorption spectroscopy (XAS) covering the TM L2,3 edges (i.e. the 2p-3d transition) has been widely used for various fields including solid state physics.3,4 Total-electron yield (TEY) is the most conventional detection mode for the XAS experiments in the soft X-ray region, while the probing depth is approximately 5 nm at most. The surface sensitivity of the TEY mode is frequently inappropriate for investigating the redox reaction in the cathode materials. In contrast, total-fluorescence-yield (total FY: TFY) XAS can probe the bulk electronic structures as deep as ∼100 nm from the sample surface. However, the TM L3-edge TFY XAS spectra tend to be distorted by self-absorption and saturation effects.5 These effects are dependent on the density, particle size (for powder) and thickness (for thin film) of the sample and the measurement geometry such as normal/grazing incidence. In addition, the distortion in the line shapes of TM L3-edge TFY XAS is enhanced in case where the target sample does not contain heavy elements such as rare earth.6 Thus, powdered Li-TM oxides and related compounds used for cathode materials of LIB are frequently affected by these effects.7 The distortion is significant when the O Kα fluorescent emission is strongly dependent on the incoming photon energies at the TM L2,3 regions. To reduce the distortion in spectral line shapes, partial-fluorescence-yield (PFY) and inverse partial-fluorescence-yield (IPFY) modes have been utilized in previous works,6–14 although the mechanisms of PFY and IPFY are still under debate.6,8,9,15–19 The PFY detection mode using a silicon-drift detector (SDD) can select a specific emission, e.g. TM Lα,β, suppressing the influence of the O Kα emission. The distortions in the line shape of L3-edge XAS can be further reduced by use of IPFY detection that is the inversed intensity of O Kα emission.6–12
In this study, we select LiFePO4 and LiNi0.5Mn1.5O4 as model examples to investigate those effects in TFY mode and utilities of PFY and IPFY modes. LiFePO4 is a promising cathode material because of the low-cost constituent element and the highly stable charge-discharge properties.20 In the olivine-type crystal structure, the Fe atoms are located in the FeO6 octahedrons and should be divalent. Spinel-type LiNi0.5Mn1.5O4 is an attractive cathode material in terms of the high voltage of ∼4.8 V vs. Li/Li+.21 The Ni and Mn atoms also form the (Ni,Mn)O6 octahedrons with Oh symmetry. The formal valences of Ni and Mn for the initial state are considered as 2+ and 4+, respectively. In addition, we performed the Cr L-edge XAS for Li0.35Mn4+0.65Cr3+0.35O222 to judge the utility of FY modes at Cr L edges.
We demonstrate TFY, PFY and IPFY XAS for those as-synthesized typical cathode materials in order to discuss appropriate use of the FY modes for each material/element. In this study, we focus on trends in the TMs of the cathode materials as observed in the FY XAS spectra.
LiFePO4 was synthesized with a solid-state reaction method.23 LiNi0.5Mn1.5O4 was synthesized by an electrospinning method.24 We confirmed the successful syntheses of the cathode materials using powder X-ray diffraction and electrochemical experiments. The fabrication process of Li0.35Mn0.65Cr0.35O2 has been described elsewhere.22 We also prepared a charged sample of LiNi0.5Mn1.5O4. The electrochemical experiments were performed by the same procedures described in Ref. 24.
For LiFePO4 and LiNi0.5Mn1.5O4, TEY, PFY, and IPFY XAS measurements were performed at BL-7A of the Photon Factory. The energy resolution of incident photon energies for the XAS measurements was E/ΔE ∼ 1500. An SDD was used for the PFY and IPFY detection. The energy width of SDD profile for fixed incident energies is approximately 100 eV. TFY XAS measurements were carried out at BL7.0.1 and BL8.0.1 of the Advanced Light Source with an energy resolution of E/ΔE ∼ 2000 for the incident photon energies. A photodiode was used for the detection of TFY system.
For Li0.35Mn0.65Cr0.35O2, TEY, TFY PFY, and IPFY XAS experiments were performed at 11ID-1 Spherical Grating Monochromator (SGM) of the Canadian Light Source. The TFY spectra were measured with a channel plate detector. The IPFY spectra were measured using an SDD with an energy width of ∼120 eV. The energy resolution of incident photon energies was set to 5000. All the XAS measurements were performed at room temperature. Configuration-interaction full-multiplet (CIFM) calculations were adopted for the analysis of the L2,3-edge XAS spectra.25–28 Ligand-to-metal charge transfer (LMCT) was taken into account in the calculation.
Figure 1(a) shows the Fe L2,3-edge TEY, TFY, PFY, and IPFY XAS spectra for LiFePO4. The main peak at 708.5 eV in the TEY spectrum is assigned to Fe2+ high-spin (HS) state from the multiplet calculation while the peak at 710.5 eV is slightly larger than the calculated Fe2+ HS spectrum. Most likely, a small fraction of Fe3+ HS should coexist according to previous reports.29,30 The TFY spectrum exhibits a dip-like structure at 706.5 eV and less pronounced L3 peaks at 708.4 and 710.5 eV. The L3 peak intensity is similar to that in the L2 edge. In the PFY spectrum corresponding to the Fe Lα,β emission, the dip-like structure at 706.5 eV was not observed although the L3/L2 intensity ratio is similar to that in the TFY spectrum and smaller than that in TEY. On the other hand, the IPFY spectrum which is the inverse of the O Kα emission shows the TEY-like intensity ratio of L3/L2. Furthermore, the profile of the IPFY spectrum is almost the same as that of TEY, suggesting the bulk and surface electronic structures are identical. These results indicate that the saturation and self-absorption effects are emphasized at the Fe L edge of this material. IPFY can be the most appropriate method to detect the bulk Fe 3d states. The results are consistent with the data collected by Yang et al.7
Figure 1(b) shows the emitted-photon-energy (hνout) profiles taken at three incident photon energies (hνin). Each profile was measured for 1 min. Although the energy width of the SDD profile is about 100 eV, the O Kα and Fe Lα,β emissions can be distinguished. The O Kα and Fe Lα,β emissions were not observed for hνin = 520 eV because 520 eV is lower than the O K edge absorption threshold. For hνin = 699 eV (below the Fe L3 edge), only the O Kα emission is detected. The Fe Lα,β emission (for 600 < hνout < 800 eV) is enhanced for hνin = 708.5 eV corresponding to the Fe L3 edge, while O Kα emission (for 420 < hνout < 600 eV) is considerably reduced. The ∼40% reduction of O Kα emission causes the apparent reduction of Fe L3 peak in the TFY spectrum. In addition, compared with the profile for hνin = 708.5 eV, the Fe Lα,β and O Kα emission intensities were reduced and unchanged for hνin = 734 eV, respectively. This is consistent with the Fe L2 TFY and PFY spectra which show high intensities.
As Fig. 2(a) shows, the Mn L2,3-edge TEY spectrum is attributed to a Mn4+ state. However, the TFY spectrum exhibits a considerable deformation at the Mn L3 edge, suggesting that the self-absorption and saturation effects strongly affect the Mn L3-edge TFY spectrum. The IPFY spectrum displays a clear Mn L3-edge structure although the L3/L2 intensity ratio is slightly smaller than that in TEY. The IPFY spectrum is also assigned to the Mn4+ state. The similarity between TEY and IPFY spectra indicates that the surface and bulk electronic structures are almost identical.
The Ni L2,3-edge TEY spectrum exhibits the Ni2+ HS multiplet structure (Fig. 2(b)). In contrast to the Mn L edge, the distortions of the Ni L3 edge TFY and PFY spectra are suppressed. Although the L3/L2 intensity ratios of TFY and PFY are slightly smaller than that for TEY, the TFY and PFY spectra are available for the analysis using the multiplet calculations. The IPFY spectrum is also similar to the TFY and PFY spectra. However, the signal-to-noise (S/N) ratio of IPFY spectrum is lower than those of TFY and PFY spectra. Thus, the IPFY detection mode is less advantageous for the analysis of Ni L2,3 edges of this material.
Figure 2(c) shows the hνout profiles of emissions taken at several hνin. The O Kα, Mn Lα,β and Ni Lα,β emissions can be observed. The Mn Lα,β emissions (for 600 < hνout < 720 eV) are enhanced at hνin = 644 eV. However, the degree of enhancement is lower than the degree of reduction of the O Kα emission, leading to the deformation at the Mn L3 edge in the TFY spectrum. On the other hand, at the Ni L3 edge, the degree of reduction of the O Kα emission becomes small and the Ni Lα,β emission (for 750 < hνout < 970 eV) increases considerably, such that the deformations at the Ni L3 edge in the TFY and PFY spectra are less pronounced.
The Mn L3-edge TFY XAS spectrum for LiNi0.5Mn1.5O4 exhibits the large distortions in XAS line shapes, suggesting that the XAS line shapes of Mn L3 edge in Li-TM oxides (and phosphates) should be very sensitive to the self-absorption and saturation effects, while these effects depend on the density and composition of sample and experimental geometries. This is consistent with the Mn L3 edge/Lα,β emission (∼640 to ∼660 eV) being very near O K edge/Kα emission (∼525 to ∼560 eV) in photon energy. The Fe L3 edge XAS is also sensitive, although the influence of self-absorption and saturation effects should be smaller than that in the MnL3 edge. Considering the present results, the IPFY detection mode can become powerful tools for the Mn and Fe L2,3-edge measurements. In the case of Ni L2,3 edges, the TFY and PFY detection modes should be more appropriate than IPFY mode in terms of the S/N ratio. Most likely, Co L2,3 edges for Li-TM oxides (and phosphates) including Co would be positioned as the intermediate character between Fe and Ni L2,3 edges.
As Fig. 2(c) indicates, the hνin dependence of O Kα emission weakens as hνin increases. To further verify this tendency, we also performed the Cr L-edge XAS for Li0.35Mn0.65Cr0.35O2 (Fig. 3(a)). The TEY spectrum is attributed to the Cr3+ state. On the other hand, not only the L3 edge, but also the L2 edge is reversed with negative sign in the TFY spectrum, which is more problematic than the case of Mn L-edges (Fig. 3(b)). The Cr L edges (∼575 to ∼595 eV) are so close to the O K edge (∼525 to ∼560 eV) that the Cr Lα,β emission cannot be separated from the O Kα emission within the energy width of SDD (∼100 eV) (Fig. 3(c)). The Cr L-edge IPFY spectrum in Fig. 3(a) is composed by extracting the structure below 570 eV in Fig. 3(c) as the OKα emission. However, the S/N ratio and L3/L2 intensity ratio are not ideal. If the energy resolution of the detector is improved, for example by using a high-resolution soft X-ray emission spectrometer,31,32 then the separation of Cr Lα,β emissions from O Kα emission should be facilitated.
Figure 4 summarizes the utilities of TFY, PFY and IPFY for each TM L edge of Li-TM oxides/phosphates. Note that the tendency is limited to powdered Li-TM oxides/phosphates with particle sizes between 10 nm and 10 μm generally used as cathode materials of LIB. The IPFY mode can be useful for Cr, Mn, and Fe L edges for those materials. For L edges with higher energies, TFY and PFY are more suitable than IPFY. In addition, TFY and PFY should be available for Ti and V without any problems because these early TMs’ L edges are located below the O K edge. Appropriate choice of the FY modes is inevitable to study the bulk electronic structures of Li-TM oxides and related materials. As Fig. 5 shows, the bulk electronic structure is often different from the surface electronic structure in particular for charged (or discharged) samples. In the case of LiNi0.5Mn1.5O4 (Fig. 5), the degree of oxidation of Ni due to the charge reaction is different between the surface and bulk. The Ni L-edge TFY XAS spectrum for the charged state shows enhanced peaks at 856 and 873 eV, indicating partial oxidation reaction of Ni2+ → Ni3+33 in the bulk region. On the other hand, in the surface-sensitive TEY mode, the line shape attributed to the Ni2+ state is almost maintained for the charged state, suggesting existence of redox-inactive Ni and/or some side reactions such as NiO formation near the surface. Thus, a combined use of TEY and FY is essential for the electronic-structure analyses of the electrode materials.
In conclusion, we demonstrated TEY, TFY, PFY and IPFY XAS studies for powdered several Li-TM oxides and a phosphate which are cathode materials for LIBs. The utilities of FY modes strongly depend on the element (edge) and material (composition). The FY modes should be carefully chosen for each material. The distortion of Ni L3-edge TFY XAS line shapes is not large whereas Cr, Mn and Fe L3-edge TFY XAS are greatly affected. We conclude that, in the FY XAS studies of the cathode materials, the IPFY XAS is most appropriate for the Cr, Mn, and Fe L edges and the TFY and PFY XAS modes are available enough for the Ni L edge which is far from the O K edge. This consideration will contribute to TM L-edge FY XAS studies of charged/discharged samples for novel electrode materials.
This work was performed under the approvals of the Photon Factory Program Advisory Committee (Proposal Nos. 2012G107, 2013G169 and 2014G100) and the Advanced Light Source (Proposal No. 04907R). The Advanced Light Source is supported by the Director, Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. Research described in this paper was partly performed at the Canadian Light Source, which is supported by CFI, NSERC, the University of Saskatchewan, the Government of Saskatchewan, WD, NRC, and CIHR. The synthesis work at NREL was supported by the NREL LDRD Program. This work was also conducted based on the Japan-U.S. cooperation project for research and standardization of Clean Energy Technologies.