Structural insights into the formation and voltage degradation of lithium- and manganese-rich layered oxides

One major challenge in the field of lithium-ion batteries is to understand the degradation mechanism of high-energy lithium- and manganese-rich layered cathode materials. Although they can deliver 30 % excess capacity compared with today’s commercially- used cathodes, the so-called voltage decay has been restricting their practical application. In order to unravel the nature of this phenomenon, we have investigated systematically the structural and compositional dependence of manganese-rich lithium insertion compounds on the lithium content provided during synthesis. Structural, electronic and electrochemical characterizations of LixNi0.2Mn0.6Oy with a wide range of lithium contents (0.00 ≤ x ≤ 1.52, 1.07 ≤ y < 2.4) and an analysis of the complexity in the synthesis pathways of monoclinic-layered Li[Li0.2Ni0.2Mn0.6]O2 oxide provide insight into the underlying processes that cause voltage fading in these cathode materials, i.e. transformation of the lithium-rich layered phase to a lithium-poor spinel phase via an intermediate lithium-containing rock-salt phase with release of lithium/oxygen.

(Comments) 1. How are Li-and Mn-rich layered oxides formed during synthesis? This reviewer finds that this part (especially the title) is written a bit misleadingly although data itself can be useful. Here, the authors used terms such as "growth" or "evolution" and tried to relate their results to interpret how Li-rich and Mn-rich materials form during synthesis.
Nevertheless, essentially what the authors showed in this section is a portion of a Li-Ni-Mn-O (Grand potential) phase diagram. The appearance of different phases as a result of a varied Liprecursor amount (or any other factors) is thermodynamically dictated, and strictly speaking there is no need to bring any "kinetics"when seeing a thermodynamic phase diagram. So, terms like "growth" or "evolution" "during" synthesis are confusing. In fact, this reviewer rather expected to see data like in Fig. 4 when someone says about the "formation of a compound during synthesis". Thermodynamic stability information is, of course, valuable, but this reviewer finds it is not relevant in this context in explaining "how the Li-rich cathodes form during synthesis". This reviewer would not care to know what the intermediate phases are when Li content is small at a high temperature synthesis, if Li-rich cathodes were to be made because one would immediately use sufficient amount of Li precursor in the beginning. Also, many of their conclusions are already known or predictive: LMLO phase stability as a function of Li content and O content (O chemical potential), spinel-rocksalt appearance at a low Li content, etc.
2. Monoclinic layered Li[Li0.2Ni0.2Mn0.6]O2 cathode with ultra-long cycling life First of all, the authors claimed that their findings provided the prerequisites for the synthesis of lithium insertion compounds with high performance. How would the provided data from the section "How are Li-and Mn-rich …" help to control the synthesis process precisely? Basically, it is well known in the Li-ion/Li-rich cathode field that one needs a certain amount of excess Li precursors when synthesizing Li-containing cathodes to account for Li evaporation at a high temperature. Also, the information in the previous section only contains a limited phase diagram along the Licontent axis and fixed Ni-Mn ratio, and also the temperature and other conditions (e.g., partial oxygen pressure in the furnace) are fixed. So, how do the findings help to precisely control the synthesis conditions to make "ultra-long cycling life" Li-rich cathodes? Also, making a pure phase does not translate into making ultra-long-cycle-life and high-performance Li-rich cathodes, as they have intrinsic limitations to address. The authors' claims can be justified only when making a pure phase has been the limiting factor for achieving high-performance cathodes, which obviously is not. Also, this reviewer does not understand how the authors can argue "ultra-long cycling life" when the capacity and voltage fading is so obvious during their lifetime (Fig. 3). Cycling for a long time is not long-cycle-life for a cathode. If the authors want to argue that their material is indeed a long cycle life material, they at least need to show a comparison of data which shows poorer performance for the poor selection of synthesis conditions. Finally, the authors claimed "reversible oxygen redox" during the initial ~300 cycles, and claim that further cycles involve spinel/rocksalt phase formation. There is a fast capacity fading from ~240 mAh/g to ~210 mAh/g during the initial 50 cycles. Also, the voltage fading is already very This manuscript reports very interesting results using well-designed in situ characterizations on synthesis/heating experiments to understand the thermodynamic driving force of the degradation process of Li-and Mn-rich layered oxides cathode materials. The conclusions are well supported by the experimental results and the data interpretation. The layered (ordered) rock salt to spinel transition/degradation has been long known in the field of Li-ion batteries. However, this process was not fully revealed in great details before this work. Using careful in situ characterizations to investigate the reactions between transition metal oxide and Li2CO3 is a unique approach to understand the thermodynamics and kinetics of layered oxides with respect to the Li and O contents. The publication of this work can bring new insights of the layered oxide cathodes to the field and inspire the materials design and battery performance improvement. I'd suggest this work to be published in Nature Communications. Some minor questions and comments are provided below. But they are optional and will not change my recommendation.   . Since the most important phase transformations/evolutions take place during the heating in the low temperature range in (a) (e.g. 0-100 minutes). It will be helpful to provide a zoomed-in image of the low temperature range in the supplementary information that can better present the spinel to disordered rocksalt transition.
It's a bit confusing in line 417-420 "For the fast thermal treatment process, the Li-free spinel phase rapidly converts to the intermediate phases, i.e. layered Li2TMO2 phase (P3m1), spinel phases (Fd3m and I41/amd) and rock-salt phase (Fm3m), in the early stage of the reaction, see  Table S4" The P3m1 phase was not labeled in Figure 4. Was it included in the phase fraction calculation or not? Line 463-464 "i.e. a kind of partially reversible formation of Li-rich phase (lithium extraction 464 accompanying oxygen evolution), see Figure 5." This statement is probably right in terms of thermodynamics. But it might be worth it to note that the kinetics of the degradation reaction, particularly the rate-limiting steps may be different from what is observed in the synthesis reactions. I.e. the synthesis seems to be a surface-bulk-limited reaction, while it is not clear what the kinetics of the degradation reaction would be like.
Reviewer #3 (Remarks to the Author): The manuscript discussed: i) structure evolution of Li-Ni-Mn-O system with changing Li composition from the high temperature synthesis. ii) structure evolution of layered Li1.2Ni0.2Mn0.6O2 after long electrochemical cycling. iii) structure evolution of Li-free spinel plus Li precursor with increasing synthesis temperature up to high temperatures. iv) Some understandings by connecting the above points i), ii), iii).
The results are in general interesting. Points i) and iii) ( Fig. 1 and Fig. 4) are straightforward and solid. However, point ii) (Fig. 2, 3) on the electrochemical structure evolution is not convincing. I am also skeptical about the methodology that the authors tried to make the connections among i) ii) iii). Thus I cannot recommend the manuscript to be published on Nature Communications.
Although it is nice to show the long cycling battery test (Fig. 2c), it is not clear how representative it is, i.e. do they always drop the capacity quickly in the first ~50 cycles, then stabilize to ~300 cycles, beyond which drop more quickly following a fixed slope? I hope that the authors had a few batteries running simultaneously at the same or different rates. If such trend is general, it will be valuable to show additional ex situ characterizations after each stage, i.e., after 50 cycles, 300 cycles, etc. That will help understand why the capacity plateau ends at around 300 cycles, as I feel it lacks an articulation about why the gradual oxygen loss beyond a certain point (such as beyond 300 cycles) can lead to an abrupt slope change in Fig. 2c. What is the critical structural change that ends the capacity plateau at ~300 cycles? Furthermore, although it is interesting to make the comparison between phase evolution at high temperature synthesis conditions and that at room temperature electrochemical cycling conditions, I am not sure the authors made it clear the fundamental connection in the very different temperature scales. At high temperature, many kinetic barriers can be easily overcome, such as the oxygen vacancy diffusion barrier or transition metal migration barrier, while at room temperature, these barriers could largely limit or even forbidden a thermodynamic phase transformation. Thus, it is not clear why the high temperature phase evolution can give insights to the room temperature one here. In my opinion, it will be more valuable to focus on the differences between different temperature scales, and discuss more about the unique electrochemical structural evolution at room temperature.

Point-by-point response (in blue) to the reviewers' comments
We thank the reviewers for their constructive ideas, which have helped us to substantially improve the quality of our manuscript. The revised manuscript includes a series of new ex situ synchrotron radiation diffraction and X-ray absorption spectroscopy results of the Li-rich electrode after different cycle numbers and stateof-the-art Bragg coherent diffraction imaging (BCDI) of a single particle obtained after one year of cycling. Below we list detailed responses to the original reviewers' comments.

Reviewer #1:
This manuscript probes the formation, evolution and voltage degradation of high- Unfortunately, the majority of the claims in this paper are known in the field, and there are unsupported or irrelevant claims about their use of data. Therefore, this reviewer believes that this paper is not suited for publication in Nature Communications.
We thank the reviewer for her/his assessment and suggestions. Her/his constructive remarks helped us improve the clarity of our manuscript, as detailed below.
Comment #1: How are Li-and Mn-rich layered oxides formed during synthesis? This reviewer finds that this part (especially the title) is written a bit misleadingly although data itself can be useful. Here, the authors used terms such as "growth" or "evolution" and tried to relate their results to interpret how Li-rich and Mn-rich materials form 2 during synthesis. Nevertheless, essentially what the authors showed in this section is a portion of a Li-Ni-Mn-O (Grand potential) phase diagram. The appearance of different phases as a result of a varied Li-precursor amount (or any other factors) is thermodynamically dictated, and strictly speaking there is no need to bring any "kinetics" when seeing a thermodynamic phase diagram. So, terms like "growth" or "evolution" "during" synthesis are confusing. In fact, this reviewer rather expected to see data like in Fig. 4 when someone says about the "formation of a compound during synthesis". Thermodynamic stability information is, of course, valuable, but this reviewer finds it is not relevant in this context in explaining "how the Li-rich cathodes All the terms such as "growth" or "evolution" have been modified in our updated manuscript.
We do not fully agree with the reviewer's statement that it is not relevant in this context in explaining how the Li-rich cathodes form during synthesis, because for the formation of the final products the system has to overcome the activation energy of We also do not agree with the reviewers' statement that many of our conclusions are already known or predictive. The appearance of spinel/rock-salt phases at a low Li content was already reported in the literature. 1 However, the kinetics of spinel/rocksalt-to-layered transition is still poorly understood, especially under nonequilibrium conditions, please see the reaction pathways for the formation of Li-rich materials in How would the provided data from the section "How are Li-and Mn-rich …" help to control the synthesis process precisely? Basically, it is well known in the Li-ion/Lirich cathode field that one needs a certain amount of excess Li precursors when synthesizing Li-containing cathodes to account for Li evaporation at a high temperature. Also, the information in the previous section only contains a limited phase diagram along the Li-content axis and fixed Ni-Mn ratio, and also the temperature and other conditions (e.g., partial oxygen pressure in the furnace) are fixed. So, how do the findings help to precisely control the synthesis conditions to make "ultra-long cycling life" Li-rich cathodes? Also, making a pure phase does not translate into making ultra-long-cycle-life and high-performance Li-rich cathodes, as they have intrinsic limitations to address. The authors' claims can be justified only when making a pure phase has been the limiting factor for achieving high- 5 performance cathodes, which obviously is not. Also, this reviewer does not understand how the authors can argue "ultra-long cycling life" when the capacity and voltage fading is so obvious during their lifetime (Fig. 3). Cycling for a long time is not long-cycle-life for a cathode. If the authors want to argue that their material is indeed a long cycle life material, they at least need to show a comparison of data which shows poorer performance for the poor selection of synthesis conditions. Finally, the authors claimed "reversible oxygen redox" during the initial ~300 cycles, and claim that further cycles involve spinel/rocksalt phase formation. There is a fast capacity fading from ~240 mAh/g to ~210 mAh/g during the initial 50 cycles. Also, the voltage fading is already very severe after 50 cycles. Therefore, it appears very confusing to say "reversible oxygen redox during early cycling" when there are such obvious changes in the performance, which should be due to the formation of spinel/rocksalt domains already. In addition, the capacity loss can come from many factors. This reviewer suspects that the capacity decay past 200 cycles is due to the accumulation of HF in the electrolyte during extended high voltage cycling and corrosion of the cathode, which is well known to occur in the Li-ion battery field, not due to the sudden appearance of spinel/rocksalt-phases "after 200 cycles". If the authors want to prove that their analysis is correct, they need to characterize their material upon (i) early, (ii) mid, and (iii) later cycling and show that during the early stage of cycling they do not observe spinel or rocksalts. corresponding to the remaining capacity of ~ 100 mA h g 1 .
The related discussion has been added to our revised manuscript, please also see below.
Main text: After 50 and 200 cycles, the reflections in the SRD patterns of the cycled L1.28 electrodes can be indexed to a layered structure with the space symmetry of C2/m and/or R m, see Figure S9. Compared to the SRD pattern of L1.28 electrode before cycling, all reflections in the SRD pattern of L1.28 cathode after 200 cycles move to lower 2-theta angles (lattice expansion), the split reflections of -133 m /33-1 m or 018 h /110 h tend to merge into a single reflection (less distortion from a cubic phase).
The oxidation state of TM in the cycled electrode does not change substantially and can still be assigned to Ni 2+ and Mn 4+ state (Figure S9(c) Figure 5. Additionally, the TM occupancy on the Li sites in the layered structure of L1.28 electrode has increased from ~ 3(2) % before cycling to ~ 6(2) % after 200 cycles (Rietveld refinement results in Figure S9 and Table S4)       Considering that the structural evolution of Li-rich layered oxides during electrochemical cycling is far away from thermal equilibrium (see Figure 3), the fast thermal treatment process was then utilized to gain new insights into the kinetics of non-equilibrium lithiation reaction. These data unambiguously demonstrate that the nonequilibrium phase transition from Li-poor spinel to Li-rich layered structure is rate-limited by the formation of an often ignored Li-rich rock-salt-type intermediate.
These results will lead to new insights into the voltage degradation of Li-rich layered cathode materials. The related discussion has been added to this section in our revised manuscript (Page 19-23).
A comprehensive and nuanced analysis of the influence of treatment (microwave heating) and precursor-type (hydroxide) on the structure and the performance of Lirich materials was reported in our previous work 2 . The changes in the treatment or precursor-type could affect the chemical reaction pathways, and therefore the 14 chemical-and phase-composition of final product. However, the formation of Li-rich cathode materials during high-temperature lithiation reaction is thermodynamically favored. When a mixture of different precursors with an appropriate amount of Li source is heated at high-temperature (e.g., 850 o C) for a long time (e.g., 12 h), the mixture would always convert to the Li-rich layered oxides.
As temperature decreases, the time needed to reach equilibrium increases exponentially, this is one reason why the thermodynamically driven formation of spinel (AB 2 O 4 , Fd-3m) phase during degradation of Li-rich oxides is so difficult to observe in the cycled electrode within a limit cycling at room temperature. The low temperature or fast heating process will take the system far away from thermal Communications. Some minor questions and comments are provided below. But they are optional and will not change my recommendation.
We thank the reviewer for recognizing the value of our work and for the insightful comments and questions. Comment #1: Figure 3. It should be labeled in the figure or stated in the caption the cycle number of the "cycled electrode" What is the shoulder on the left side of the 131 peak in Figure 3 around 11.6 degree 2-theta?
Answer #1: We thank the reviewer for this observation. The cycle number of the "cycled electrode" has been now added in the caption of Figure 3 (page 16). We cannot find the shoulder on the left side of the 131 reflection in Figure 3 in the raw materials, so it is not related to our active material.
Comment #2: Figure 4. Since the most important phase transformations/evolutions take place during the heating in the low temperature range in (a) (e.g. 0-100 minutes).
It will be helpful to provide a zoomed-in image of the low temperature range in the supplementary information that can better present the spinel to disordered rocksalt transition.
Answer #2: A zoomed-in image of the low temperature range has been now added in our new supplementary document ( Figure S22 in Page 23), please also see below. The results are in general interesting. Points i) and iii) ( Fig. 1 and Fig. 4) are straightforward and solid. However, point ii) (Fig. 2, 3) on the electrochemical structure evolution is not convincing. I am also skeptical about the methodology that the authors tried to make the connections among i) ii) iii). Thus I cannot recommend the manuscript to be published on Nature Communications.
We thank the reviewer for valuable comments for improving the quality of our manuscript.
Comment #1: Although it is nice to show the long cycling battery test (Fig. 2c), it is not clear how representative it is, i.e. do they always drop the capacity quickly in the first ~50 cycles, then stabilize to ~300 cycles, beyond which drop more quickly following a fixed slope? I hope that the authors had a few batteries running simultaneously at the same or different rates. If such trend is general, it will be valuable to show additional ex situ characterizations after each stage, i.e., after 50 cycles, 300 cycles, etc. That will help understand why the capacity plateau ends at around 300 cycles, as I feel it lacks an articulation about why the gradual oxygen loss beyond a certain point (such as beyond 300 cycles) can lead to an abrupt slope change in Fig. 2c. What is the critical structural change that ends the capacity plateau at ~300 cycles?
Answer #1: We note that the sudden capacity loss of about 30 mA h g 1 is not always observed in the first 50 cycles (Figure S7) The related discussion has been added to our revised manuscript, please also see below.
Main text: After 50 and 200 cycles, the reflections in the SRD patterns of the cycled L1.28 electrodes can be indexed to a layered structure with the space symmetry of C2/m and/or R m, see Figure S9. Compared to the SRD pattern of L1.28 electrode before cycling, all reflections in the SRD pattern of L1.28 cathode after 200 cycles move to lower 2-theta angles (lattice expansion), the split reflections of -133 m /33-1 m or 018 h /110 h tend to merge into a single reflection (less distortion from a cubic phase).
The oxidation state of TM in the cycled electrode does not change substantially and can still be assigned to Ni 2+ and Mn 4+ state (Figure S9(c) Figure 5. Additionally, the TM occupancy on the Li sites in the layered structure of L1.28 electrode has increased from ~ 3(2) % before cycling to ~ 6(2) % 22 after 200 cycles (Rietveld refinement results in Figure S9 and  Figure 1). Therefore, the commonly called 'spinel-like' phase formed within a to be located at the surface (see Figure 4). Unfortunately, the accumulation of layered & rock-salt-type & spinel domains with the large lattice mismatch induces a much higher diffusion barrier for Li ions, compared to the stage II, and thereby a considerably capacity loss at stage III in Figure 2(c). In addition, the capacity fading is probably also a consequence of the slow dissolution of manganese (III) ions in the spinel phase (2Mn 3+ Mn 4+ + Mn 2+ ), the lithium dendrite and SEI growth in the coin cell, and electrolyte evaporation. 5 barrier or transition metal migration barrier, while at room temperature, these barriers could largely limit or even forbidden a thermodynamic phase transformation. Thus, it is not clear why the high temperature phase evolution can give insights to the room temperature one here. In my opinion, it will be more valuable to focus on the differences between different temperature scales, and discuss more about the unique electrochemical structural evolution at room temperature.
Answer #2: We thank the reviewer for the insightful comments, we agree with these comments. Our experimental results demonstrate that the high-temperature reaction of Thermodynamically, different temperature scales are strongly correlated to various time scales. As temperature decreases, the time needed to reach equilibrium increases exponentially, this is why the thermodynamically driven formation of spinel (AB 2 O 4 , Fd-3m) phase during degradation of Li-rich oxides is so difficult to observe in the cycled electrode within a limited cycling time at room temperature. The low temperature or fast heating process will take the system far away from equilibrium. spinel phase (Fd m), but this transition is limited by the generation of the Licontaining rock-salt phase (Fm m). Such an inversed relationship between formation pathway and degradation process of LMLOs is not a coincidence but intrinsic due to