Phase transformation mechanism in lithium manganese nickel oxide revealed by single-crystal hard X-ray microscopy

Understanding the reaction pathway and kinetics of solid-state phase transformation is critical in designing advanced electrode materials with better performance and stability. Despite the first-order phase transition with a large lattice mismatch between the involved phases, spinel LiMn1.5Ni0.5O4 is capable of fast rate even at large particle size, presenting an enigma yet to be understood. The present study uses advanced two-dimensional and three-dimensional nano-tomography on a series of well-formed LixMn1.5Ni0.5O4 (0≤x≤1) crystals to visualize the mesoscale phase distribution, as a function of Li content at the sub-particle level. Inhomogeneity along with the coexistence of Li-rich and Li-poor phases are broadly observed on partially delithiated crystals, providing direct evidence for a concurrent nucleation and growth process instead of a shrinking-core or a particle-by-particle process. Superior kinetics of (100) facets at the vertices of truncated octahedral particles promote preferential delithiation, whereas the observation of strain-induced cracking suggests mechanical degradation in the material.

about cycle life of the prepared LMNO cathode. 2. In Figure 6, the authors show fractured surface of MNO crystals. How much the MNO was cycled? Just one cycle? 3. In order to discuss mechanical pulverization of the MNO cathode combined with phase transformation, the authors should show 2D and 3D chemical mapping data as function of cycles together with SEM and TEM data of the corresponding MNO samples. Please refer to a recent work (Adv Energy mater. DOI: 10.1002/aenm.201601417) studying degradation mechanism after longterm cycling of layered NMC cathode and discuss mechanical pulverization mechanism. 4. Reviewer doesn't agree the authors' conclusion that reducing primary particle size to below 1 µm is important strategies in improving cycle life because nano-sized electrode (cathode or anode) is significantly decreased volumetric energy density due to lower packing density and cycle life due to parasitic reaction between cathode and electrolyte.
I am mostly satisfied with the authors' response to my previous questions. I recommend this manuscript to be published after the authors consider some very minor revisions.
1) The 3D surface renderings in Figure SI-4, which shows cracks on the LNMO particle surface, look very nice and interesting. They could be better shown in together with the SEM images in Figure 6 of the main text. 2) Some revisions need to be done to the paragraph added on page 13-14. e.g. "The presence of such a large overpotential is likely to promote a high Li flux during the extraction, which may lead to significant variations in local current density and contribute to the particle-level heterogeneity observed in this study." There is no "current" in chemical delithiation. The delithiation process under a large overpotential is likely limited by bulk diffusion (of Li+) (perhaps similar to what has been observed in LiFePO4). Therefore, in one single particle, it is possible that the 2nd delithiation process (L0.5NMO to NMO) at the outer surface starts before the 1st delithiation process (LNMO to L0.5NMO) fully finish in the entire particle due to Li-ion transport limitation. This could be one explanation for the particle-level heterogeneity observed in this study. "It is possible that under small constant-current charging without this voltage driving force (small Crate)" It is not correct to say "without this voltage driving force". It should be "under a small overpotential". The electrochemical reaction (delithiation of LNMO here) will not proceed if without an overpotential. "in operando techniques such as those recently demonstrated on LiFePO4" (ref. 30, 31) Ref. 31 is an ex situ work, not operando (also it is operando not in operando).
Reviewer #2 (Remarks to the Author): All the comments raised by me have been well addressed. I recommend the paper to be published in Nat Commun. ** See Nature Research's author and referees' website at www.nature.com/authors for information about policies, services and author benefits

Reviewer #1 (Remarks to the Author):
The manuscript by Kuppan et al reports the investigation of phase transformation mechanism in ordered LiMn1.5Ni0.5O4 at single-particle level during chemical delithiation using 2D and 3D hard X-ray spectroscopic imaging. The manuscript is well-written and the results are quite interesting and insightful for Li-ion cathode materials. I recommend this manuscript for publication in Nature Communications after some revisions as discussed below: Major comments: 1) The authors proposes a concurrent 3-phase transformation mechanism for the LNMO cathode during chemical delithiation. While this claim is indeed convincingly supported by the data, I am not sure this will be the (only) phase transformation mechanism that can be observed during electrochemical delithiation (constant-current charge). The phase transformation of LNMO is very likely to be controlled (or at least strongly influenced) by the overpotential, similar to what has been reported for another phase-separating electrode material, LiFePO4. The NO2+/NO2 redox couple has a standard potential of ~5.1 V vs Li+/Li, which is ~400 mV higher than the plateau potential (quasi-equilibrium potential) of the LNMO cathode (~4.7 V). Therefore, this may be analogous to a phase transformation process under a large overpotential so that concurrent 3-phase mechanism is observed. However, if the overpotential is very small, i.e. electrochemically charging the LNMO cathode at a small C-rate such as 1/50 or 1/100 C, two-phase mechanism (at single-particle level) and particle-by-particle mechanism (at multiple-particle level) may be observed. I believe that the authors need to point out the important role of overpotential in electrochemically-drive phase transformations and discuss the possible difference between chemical and electrochemical delithiation.
(Authors' response): we thank the reviewer for pointing out the possible mechanistic differences between chemical and electrochemical delithiation and the role of overpotential in phase transformation mechanism. We fully agree that the Li extraction conditions used in this study is similar to electrochemical delithiation under a large overpotential of ~400 mV rather than under a small constant current (slow-rate charging).
A large body of work has already been carried out in order to understand the phase transition mechanism in LiFePO 4 , another lithium-ion battery cathode material operating through the firstorder transition. The phase transformation mechanism was found to be sensitive to a number of parameters, particularly charging/discharging rate, particle size and temperature. The role of overpotential is indeed important as it is likely to promote a high Li flux during the extraction/insertion, which may lead to significant variations in local current density and contribute to the particle-level heterogeneity observed in this study. It is possible that under small constant-current charging without the voltage driving force, other phase transformation mechanism can dominate or coexist. Systematic studies using different experimental conditions, especially the use of in operando techniques such as those recently demonstrated on LiFePO 4 by Li el al [Nat. Mater. 13, 1149-1156(2014] and Lim et al [Science 353, 6299, 566-571 (2016)], should provide further information on phase transformation mechanism(s) under electrochemical charge/discharge conditions. We would like to emphasize that the main advantage of chemical delithiation used in this study is the uniformity in delithiation as all LMNO particles were immersed in the same oxidizing NO 2 BF 4 solution. In electrochemical charge/discharge, Li extraction is greatly influenced by ionic and electronic connectivity which introduces additional variations contributing to the observed delithiation mechanism at the particle-level.
The following paragraph was added on page 13-14 and highlighted in red: "We would like to point out that the chemical delithiation condition used in this study is similar to that of constant-voltage electrochemical charging under an overpotential. The standard oxidation potential of the NO 2 + /NO 2 redox couple is ~5.1 V vs. Li + /Li, which is ~400 mV higher than that of the plateau potential of the LMNO cathode at ~4.7 V. The presence of such a large overpotential is likely to promote a high Li flux during the extraction, which may lead to significant variations in local current density and contribute to the particle-level heterogeneity observed in this study. It is possible that under small constant-current charging without this voltage driving force (small C-rate), other phase transformation mechanism could dominate or coexist. Systematic studies using different experimental conditions, especially the use of in operando techniques such as those recently demonstrated on LiFePO 4 , 30, 31 should provide further information on phase transformation mechanism(s) under electrochemical charge/discharge conditions. One caution is that in electrochemical charge/discharge, Li extraction is greatly influenced by ionic and electronic connectivity which introduces additional factors controlling the observed delithiation mechanism at the particle-level. Whereas in chemical delithiation, all LMNO particles are immersed in the oxidizing NO 2 BF 4 solution with the same concentration and therefore, delithiation uniformity at both particle-level and bulk-level can be achieved." 2) I am still confused about how the standard reference spectrum of L0.5NMO is numerically extrapolated. Did the authors assume a certain crystal structure and site occupancy of L0.5NMO and simulate the XANES spectrum of L0.5NMO or use any other methods? Can the authors provide more details in supporting information? (Authors' response): we sincerely apologize for the confusion. The extrapolation of the phasepure L 0.5 NMO XANES spectrum (S . ) is essentially a mathematical optimization process of obtaining a solution that minimizes the deviation among the results obtained from multiple independent experimental measurements while satisfies the constraints set by a given equation.
More detailed explanation on the method of extracting the XANES spectrum of L 0.5 NMO is included in the revised manuscript. The following sentences were added on page 7-8: "Specifically, the spectra of phase-pure LMNO and MNO (S 1 and S 0 ) were acquired directly. All of the experimentally measured spectra of partially dilithiated L x MNO (S x , x=0.06, 0.25, 0.51, 0.71, 0.82, 0.9) were composed of varying contributions from the three primary phases (LMNO, L 0.5 NMO and MNO), with the relative weight fraction of each phase at a given x value (R , R . and R ) obtainable from the Rietveld refinement of the full XRD patterns. This gives us six redundant linear equations that can be expressed as S = (S × R ) + (S . × R . ) + (S × R ). The only unknown parameter in these equations is S . , which is the signature spectrum of the phase-pure L 0.5 NMO. S . was then solved by minimizing the deviation among the results derived from multiple independent experimental measurements while satisfying the constraints set by the equation described above. The calculation was implemented using Matlab's optimization tool box." 3) The authors mention that "due to the spatial resolution limitation of FFTXM imaging, strain and fracturing survey at the LixNMO phase boundaries was not feasible". However, the cracks in Figure 6b seem to be large enough (~100 nm?) for FFTXM to see, which has a spatial resolution around 30 nm (in 2D). Can the authors comment on this? How large does the crack need to be if one wants to use FFTXM technique to study crack formation?
(Authors' response): we greatly appreciate the reviewer for catching this misleading statement in the paper.
It is not feasible to study the strain in the particle using FF-TXM because the contrast mechanism of this technique is not sensitive to the local crystal structure distortion. The FF-TXM detects the difference in the X-ray absorption and, by extension, the absorption coefficient's variation as a function of the X-ray energy, which is sensitive to the local oxidation state. For the survey of the crack in the particle, FF-TXM can indeed provide useful insight in 3D. For most part of our study, we operated the FF-TXM around the Ni K-edge for the chemical sensitivity. However, we can optimize the microscope at lower X-ray energy in order to study particle morphology, such as the formation of cracks. This is because better absorption contrast and higher spatial resolution could be achieved at lower energy. Below we show the three-dimensional surface rendering of two selected MNO particles (one at about 4 microns and the other at about 2 microns) which nicely complement the observation in our SEM studies, as the formation of cracks is also clearly observed on the surface. Figure SI-4. 3D surface renderings of two selected MNO crystals prepared by chemical delithiation (the left one at about 4 µm and the right one at 2 µm) show the formation of irregularly shaped cracks on the particle surface.
We have modified the paragraphs to incorporate these changes. The following sentences were included on page 14-15, along with the changes made in the reference section and the addition of Figure SI-4 in the Supporting Information.
"A recent study using the in situ coherent X-ray diffractive imaging (CXDI) technique showed that the strain energy is relatively low at the early stages of LMNO delithiation but it increases more than ten-fold during the topotactic phase transformation process. 16,47 Striking inhomogeneity in the strain distribution throughout the particle was also reported. Further evidence was provided by in situ multi-beam optical stress sensor (MOSS) on LMNO thin-film electrodes, which showed that the stress in LMNO increases with lithium removal and it can reach up to 126 MPa at fully delithiated state (MNO). 48 Strain survey on Li x MNO crystals was not feasible with FF-TXM as the contrast mechanism of the technique is not sensitive to the local crystal structure distortion. However, morphology survey on the particles, such as the formation of cracks, can be carried out especially if the microscope is operated at lower X-ray energy than that of the Ni K-edge. This is because better absorption contrast and higher spatial resolution are achievable at lower energy. Figure SI-4 shows the three-dimensional surface rendering of two selected fully delithiated MNO crystals (4 and 2 µm in size) prepared by chemical oxidation. The formation of irregularly shaped cracks on the particle surface is clearly shown.
Further evidence on delithiation-induced morphological changes was provided by SEM studies, as shown by the images obtained on pristine LMNO and chemically delithiated MNO in Figure  6." The following references were modified in the reference section: